initial_commit

149 files changed, 98982 insertions, 0 deletions

diff --git a/fs/xfs/Kconfig b/fs/xfs/Kconfig
new file mode 100644
index 00000000..6100ec0f
--- /dev/null
+++ b/fs/xfs/Kconfig
@@ -0,0 +1,82 @@
+config XFS_FS
+	tristate "XFS filesystem support"
+	depends on BLOCK
+	select EXPORTFS
+	help
+	  XFS is a high performance journaling filesystem which originated
+	  on the SGI IRIX platform.  It is completely multi-threaded, can
+	  support large files and large filesystems, extended attributes,
+	  variable block sizes, is extent based, and makes extensive use of
+	  Btrees (directories, extents, free space) to aid both performance
+	  and scalability.
+
+	  Refer to the documentation at <http://oss.sgi.com/projects/xfs/>
+	  for complete details.  This implementation is on-disk compatible
+	  with the IRIX version of XFS.
+
+	  To compile this file system support as a module, choose M here: the
+	  module will be called xfs.  Be aware, however, that if the file
+	  system of your root partition is compiled as a module, you'll need
+	  to use an initial ramdisk (initrd) to boot.
+
+config XFS_QUOTA
+	bool "XFS Quota support"
+	depends on XFS_FS
+	select QUOTACTL
+	help
+	  If you say Y here, you will be able to set limits for disk usage on
+	  a per user and/or a per group basis under XFS.  XFS considers quota
+	  information as filesystem metadata and uses journaling to provide a
+	  higher level guarantee of consistency.  The on-disk data format for
+	  quota is also compatible with the IRIX version of XFS, allowing a
+	  filesystem to be migrated between Linux and IRIX without any need
+	  for conversion.
+
+	  If unsure, say N.  More comprehensive documentation can be found in
+	  README.quota in the xfsprogs package.  XFS quota can be used either
+	  with or without the generic quota support enabled (CONFIG_QUOTA) -
+	  they are completely independent subsystems.
+
+config XFS_POSIX_ACL
+	bool "XFS POSIX ACL support"
+	depends on XFS_FS
+	select FS_POSIX_ACL
+	help
+	  POSIX Access Control Lists (ACLs) support permissions for users and
+	  groups beyond the owner/group/world scheme.
+
+	  To learn more about Access Control Lists, visit the POSIX ACLs for
+	  Linux website <http://acl.bestbits.at/>.
+
+	  If you don't know what Access Control Lists are, say N.
+
+config XFS_RT
+	bool "XFS Realtime subvolume support"
+	depends on XFS_FS
+	help
+	  If you say Y here you will be able to mount and use XFS filesystems
+	  which contain a realtime subvolume.  The realtime subvolume is a
+	  separate area of disk space where only file data is stored.  It was
+	  originally designed to provide deterministic data rates suitable
+	  for media streaming applications, but is also useful as a generic
+	  mechanism for ensuring data and metadata/log I/Os are completely
+	  separated.  Regular file I/Os are isolated to a separate device
+	  from all other requests, and this can be done quite transparently
+	  to applications via the inherit-realtime directory inode flag.
+
+	  See the xfs man page in section 5 for additional information.
+
+	  If unsure, say N.
+
+config XFS_DEBUG
+	bool "XFS Debugging support (EXPERIMENTAL)"
+	depends on XFS_FS && EXPERIMENTAL
+	help
+	  Say Y here to get an XFS build with many debugging features,
+	  including ASSERT checks, function wrappers around macros,
+	  and extra sanity-checking functions in various code paths.
+
+	  Note that the resulting code will be HUGE and SLOW, and probably
+	  not useful unless you are debugging a particular problem.
+
+	  Say N unless you are an XFS developer, or you play one on TV.
diff --git a/fs/xfs/Makefile b/fs/xfs/Makefile
new file mode 100644
index 00000000..284a7c89
--- /dev/null
+++ b/fs/xfs/Makefile
@@ -0,0 +1,111 @@
+#
+# Copyright (c) 2000-2005 Silicon Graphics, Inc.
+# All Rights Reserved.
+#
+# This program is free software; you can redistribute it and/or
+# modify it under the terms of the GNU General Public License as
+# published by the Free Software Foundation.
+#
+# This program is distributed in the hope that it would be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, write the Free Software Foundation,
+# Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+#
+
+ccflags-y := -I$(src) -I$(src)/linux-2.6
+ccflags-$(CONFIG_XFS_DEBUG) += -g
+
+XFS_LINUX := linux-2.6
+
+obj-$(CONFIG_XFS_FS)		+= xfs.o
+
+xfs-y				+= linux-2.6/xfs_trace.o
+
+xfs-$(CONFIG_XFS_QUOTA)		+= $(addprefix quota/, \
+				   xfs_dquot.o \
+				   xfs_dquot_item.o \
+				   xfs_trans_dquot.o \
+				   xfs_qm_syscalls.o \
+				   xfs_qm_bhv.o \
+				   xfs_qm.o)
+xfs-$(CONFIG_XFS_QUOTA)		+= linux-2.6/xfs_quotaops.o
+
+ifeq ($(CONFIG_XFS_QUOTA),y)
+xfs-$(CONFIG_PROC_FS)		+= quota/xfs_qm_stats.o
+endif
+
+xfs-$(CONFIG_XFS_RT)		+= xfs_rtalloc.o
+xfs-$(CONFIG_XFS_POSIX_ACL)	+= $(XFS_LINUX)/xfs_acl.o
+xfs-$(CONFIG_PROC_FS)		+= $(XFS_LINUX)/xfs_stats.o
+xfs-$(CONFIG_SYSCTL)		+= $(XFS_LINUX)/xfs_sysctl.o
+xfs-$(CONFIG_COMPAT)		+= $(XFS_LINUX)/xfs_ioctl32.o
+
+
+xfs-y				+= xfs_alloc.o \
+				   xfs_alloc_btree.o \
+				   xfs_attr.o \
+				   xfs_attr_leaf.o \
+				   xfs_bit.o \
+				   xfs_bmap.o \
+				   xfs_bmap_btree.o \
+				   xfs_btree.o \
+				   xfs_buf_item.o \
+				   xfs_da_btree.o \
+				   xfs_dir2.o \
+				   xfs_dir2_block.o \
+				   xfs_dir2_data.o \
+				   xfs_dir2_leaf.o \
+				   xfs_dir2_node.o \
+				   xfs_dir2_sf.o \
+				   xfs_error.o \
+				   xfs_extfree_item.o \
+				   xfs_filestream.o \
+				   xfs_fsops.o \
+				   xfs_ialloc.o \
+				   xfs_ialloc_btree.o \
+				   xfs_iget.o \
+				   xfs_inode.o \
+				   xfs_inode_item.o \
+				   xfs_iomap.o \
+				   xfs_itable.o \
+				   xfs_dfrag.o \
+				   xfs_log.o \
+				   xfs_log_cil.o \
+				   xfs_log_recover.o \
+				   xfs_mount.o \
+				   xfs_mru_cache.o \
+				   xfs_rename.o \
+				   xfs_trans.o \
+				   xfs_trans_ail.o \
+				   xfs_trans_buf.o \
+				   xfs_trans_extfree.o \
+				   xfs_trans_inode.o \
+				   xfs_utils.o \
+				   xfs_vnodeops.o \
+				   xfs_rw.o
+
+xfs-$(CONFIG_XFS_TRACE)		+= xfs_btree_trace.o
+
+# Objects in linux/
+xfs-y				+= $(addprefix $(XFS_LINUX)/, \
+				   kmem.o \
+				   xfs_aops.o \
+				   xfs_buf.o \
+				   xfs_discard.o \
+				   xfs_export.o \
+				   xfs_file.o \
+				   xfs_fs_subr.o \
+				   xfs_globals.o \
+				   xfs_ioctl.o \
+				   xfs_iops.o \
+				   xfs_message.o \
+				   xfs_super.o \
+				   xfs_sync.o \
+				   xfs_xattr.o)
+
+# Objects in support/
+xfs-y				+= support/uuid.o
diff --git a/fs/xfs/linux-2.6/kmem.c b/fs/xfs/linux-2.6/kmem.c
new file mode 100644
index 00000000..a907de56
--- /dev/null
+++ b/fs/xfs/linux-2.6/kmem.c
@@ -0,0 +1,132 @@
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include <linux/mm.h>
+#include <linux/highmem.h>
+#include <linux/slab.h>
+#include <linux/swap.h>
+#include <linux/blkdev.h>
+#include <linux/backing-dev.h>
+#include "time.h"
+#include "kmem.h"
+#include "xfs_message.h"
+
+/*
+ * Greedy allocation.  May fail and may return vmalloced memory.
+ *
+ * Must be freed using kmem_free_large.
+ */
+void *
+kmem_zalloc_greedy(size_t *size, size_t minsize, size_t maxsize)
+{
+	void		*ptr;
+	size_t		kmsize = maxsize;
+
+	while (!(ptr = kmem_zalloc_large(kmsize))) {
+		if ((kmsize >>= 1) <= minsize)
+			kmsize = minsize;
+	}
+	if (ptr)
+		*size = kmsize;
+	return ptr;
+}
+
+void *
+kmem_alloc(size_t size, unsigned int __nocast flags)
+{
+	int	retries = 0;
+	gfp_t	lflags = kmem_flags_convert(flags);
+	void	*ptr;
+
+	do {
+		ptr = kmalloc(size, lflags);
+		if (ptr || (flags & (KM_MAYFAIL|KM_NOSLEEP)))
+			return ptr;
+		if (!(++retries % 100))
+			xfs_err(NULL,
+		"possible memory allocation deadlock in %s (mode:0x%x)",
+					__func__, lflags);
+		congestion_wait(BLK_RW_ASYNC, HZ/50);
+	} while (1);
+}
+
+void *
+kmem_zalloc(size_t size, unsigned int __nocast flags)
+{
+	void	*ptr;
+
+	ptr = kmem_alloc(size, flags);
+	if (ptr)
+		memset((char *)ptr, 0, (int)size);
+	return ptr;
+}
+
+void
+kmem_free(const void *ptr)
+{
+	if (!is_vmalloc_addr(ptr)) {
+		kfree(ptr);
+	} else {
+		vfree(ptr);
+	}
+}
+
+void *
+kmem_realloc(const void *ptr, size_t newsize, size_t oldsize,
+	     unsigned int __nocast flags)
+{
+	void	*new;
+
+	new = kmem_alloc(newsize, flags);
+	if (ptr) {
+		if (new)
+			memcpy(new, ptr,
+				((oldsize < newsize) ? oldsize : newsize));
+		kmem_free(ptr);
+	}
+	return new;
+}
+
+void *
+kmem_zone_alloc(kmem_zone_t *zone, unsigned int __nocast flags)
+{
+	int	retries = 0;
+	gfp_t	lflags = kmem_flags_convert(flags);
+	void	*ptr;
+
+	do {
+		ptr = kmem_cache_alloc(zone, lflags);
+		if (ptr || (flags & (KM_MAYFAIL|KM_NOSLEEP)))
+			return ptr;
+		if (!(++retries % 100))
+			xfs_err(NULL,
+		"possible memory allocation deadlock in %s (mode:0x%x)",
+					__func__, lflags);
+		congestion_wait(BLK_RW_ASYNC, HZ/50);
+	} while (1);
+}
+
+void *
+kmem_zone_zalloc(kmem_zone_t *zone, unsigned int __nocast flags)
+{
+	void	*ptr;
+
+	ptr = kmem_zone_alloc(zone, flags);
+	if (ptr)
+		memset((char *)ptr, 0, kmem_cache_size(zone));
+	return ptr;
+}
diff --git a/fs/xfs/linux-2.6/kmem.h b/fs/xfs/linux-2.6/kmem.h
new file mode 100644
index 00000000..f7c8f7a9
--- /dev/null
+++ b/fs/xfs/linux-2.6/kmem.h
@@ -0,0 +1,124 @@
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_SUPPORT_KMEM_H__
+#define __XFS_SUPPORT_KMEM_H__
+
+#include <linux/slab.h>
+#include <linux/sched.h>
+#include <linux/mm.h>
+#include <linux/vmalloc.h>
+
+/*
+ * General memory allocation interfaces
+ */
+
+#define KM_SLEEP	0x0001u
+#define KM_NOSLEEP	0x0002u
+#define KM_NOFS		0x0004u
+#define KM_MAYFAIL	0x0008u
+
+/*
+ * We use a special process flag to avoid recursive callbacks into
+ * the filesystem during transactions.  We will also issue our own
+ * warnings, so we explicitly skip any generic ones (silly of us).
+ */
+static inline gfp_t
+kmem_flags_convert(unsigned int __nocast flags)
+{
+	gfp_t	lflags;
+
+	BUG_ON(flags & ~(KM_SLEEP|KM_NOSLEEP|KM_NOFS|KM_MAYFAIL));
+
+	if (flags & KM_NOSLEEP) {
+		lflags = GFP_ATOMIC | __GFP_NOWARN;
+	} else {
+		lflags = GFP_KERNEL | __GFP_NOWARN;
+		if ((current->flags & PF_FSTRANS) || (flags & KM_NOFS))
+			lflags &= ~__GFP_FS;
+	}
+	return lflags;
+}
+
+extern void *kmem_alloc(size_t, unsigned int __nocast);
+extern void *kmem_zalloc(size_t, unsigned int __nocast);
+extern void *kmem_realloc(const void *, size_t, size_t, unsigned int __nocast);
+extern void  kmem_free(const void *);
+
+static inline void *kmem_zalloc_large(size_t size)
+{
+	void *ptr;
+
+	ptr = vmalloc(size);
+	if (ptr)
+		memset(ptr, 0, size);
+	return ptr;
+}
+static inline void kmem_free_large(void *ptr)
+{
+	vfree(ptr);
+}
+
+extern void *kmem_zalloc_greedy(size_t *, size_t, size_t);
+
+/*
+ * Zone interfaces
+ */
+
+#define KM_ZONE_HWALIGN	SLAB_HWCACHE_ALIGN
+#define KM_ZONE_RECLAIM	SLAB_RECLAIM_ACCOUNT
+#define KM_ZONE_SPREAD	SLAB_MEM_SPREAD
+
+#define kmem_zone	kmem_cache
+#define kmem_zone_t	struct kmem_cache
+
+static inline kmem_zone_t *
+kmem_zone_init(int size, char *zone_name)
+{
+	return kmem_cache_create(zone_name, size, 0, 0, NULL);
+}
+
+static inline kmem_zone_t *
+kmem_zone_init_flags(int size, char *zone_name, unsigned long flags,
+		     void (*construct)(void *))
+{
+	return kmem_cache_create(zone_name, size, 0, flags, construct);
+}
+
+static inline void
+kmem_zone_free(kmem_zone_t *zone, void *ptr)
+{
+	kmem_cache_free(zone, ptr);
+}
+
+static inline void
+kmem_zone_destroy(kmem_zone_t *zone)
+{
+	if (zone)
+		kmem_cache_destroy(zone);
+}
+
+extern void *kmem_zone_alloc(kmem_zone_t *, unsigned int __nocast);
+extern void *kmem_zone_zalloc(kmem_zone_t *, unsigned int __nocast);
+
+static inline int
+kmem_shake_allow(gfp_t gfp_mask)
+{
+	return ((gfp_mask & __GFP_WAIT) && (gfp_mask & __GFP_FS));
+}
+
+#endif /* __XFS_SUPPORT_KMEM_H__ */
diff --git a/fs/xfs/linux-2.6/mrlock.h b/fs/xfs/linux-2.6/mrlock.h
new file mode 100644
index 00000000..ff6a1987
--- /dev/null
+++ b/fs/xfs/linux-2.6/mrlock.h
@@ -0,0 +1,90 @@
+/*
+ * Copyright (c) 2000-2006 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_SUPPORT_MRLOCK_H__
+#define __XFS_SUPPORT_MRLOCK_H__
+
+#include <linux/rwsem.h>
+
+typedef struct {
+	struct rw_semaphore	mr_lock;
+#ifdef DEBUG
+	int			mr_writer;
+#endif
+} mrlock_t;
+
+#ifdef DEBUG
+#define mrinit(mrp, name)	\
+	do { (mrp)->mr_writer = 0; init_rwsem(&(mrp)->mr_lock); } while (0)
+#else
+#define mrinit(mrp, name)	\
+	do { init_rwsem(&(mrp)->mr_lock); } while (0)
+#endif
+
+#define mrlock_init(mrp, t,n,s)	mrinit(mrp, n)
+#define mrfree(mrp)		do { } while (0)
+
+static inline void mraccess_nested(mrlock_t *mrp, int subclass)
+{
+	down_read_nested(&mrp->mr_lock, subclass);
+}
+
+static inline void mrupdate_nested(mrlock_t *mrp, int subclass)
+{
+	down_write_nested(&mrp->mr_lock, subclass);
+#ifdef DEBUG
+	mrp->mr_writer = 1;
+#endif
+}
+
+static inline int mrtryaccess(mrlock_t *mrp)
+{
+	return down_read_trylock(&mrp->mr_lock);
+}
+
+static inline int mrtryupdate(mrlock_t *mrp)
+{
+	if (!down_write_trylock(&mrp->mr_lock))
+		return 0;
+#ifdef DEBUG
+	mrp->mr_writer = 1;
+#endif
+	return 1;
+}
+
+static inline void mrunlock_excl(mrlock_t *mrp)
+{
+#ifdef DEBUG
+	mrp->mr_writer = 0;
+#endif
+	up_write(&mrp->mr_lock);
+}
+
+static inline void mrunlock_shared(mrlock_t *mrp)
+{
+	up_read(&mrp->mr_lock);
+}
+
+static inline void mrdemote(mrlock_t *mrp)
+{
+#ifdef DEBUG
+	mrp->mr_writer = 0;
+#endif
+	downgrade_write(&mrp->mr_lock);
+}
+
+#endif /* __XFS_SUPPORT_MRLOCK_H__ */
diff --git a/fs/xfs/linux-2.6/time.h b/fs/xfs/linux-2.6/time.h
new file mode 100644
index 00000000..387e695a
--- /dev/null
+++ b/fs/xfs/linux-2.6/time.h
@@ -0,0 +1,36 @@
+/*
+ * Copyright (c) 2000-2003,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_SUPPORT_TIME_H__
+#define __XFS_SUPPORT_TIME_H__
+
+#include <linux/sched.h>
+#include <linux/time.h>
+
+typedef struct timespec timespec_t;
+
+static inline void delay(long ticks)
+{
+	schedule_timeout_uninterruptible(ticks);
+}
+
+static inline void nanotime(struct timespec *tvp)
+{
+	*tvp = CURRENT_TIME;
+}
+
+#endif /* __XFS_SUPPORT_TIME_H__ */
diff --git a/fs/xfs/linux-2.6/xfs_acl.c b/fs/xfs/linux-2.6/xfs_acl.c
new file mode 100644
index 00000000..f86e0348
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_acl.c
@@ -0,0 +1,464 @@
+/*
+ * Copyright (c) 2008, Christoph Hellwig
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_acl.h"
+#include "xfs_attr.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_inode.h"
+#include "xfs_vnodeops.h"
+#include "xfs_trace.h"
+#include <linux/slab.h>
+#include <linux/xattr.h>
+#include <linux/posix_acl_xattr.h>
+
+
+/*
+ * Locking scheme:
+ *  - all ACL updates are protected by inode->i_mutex, which is taken before
+ *    calling into this file.
+ */
+
+STATIC struct posix_acl *
+xfs_acl_from_disk(struct xfs_acl *aclp)
+{
+	struct posix_acl_entry *acl_e;
+	struct posix_acl *acl;
+	struct xfs_acl_entry *ace;
+	unsigned int count, i;
+
+	count = be32_to_cpu(aclp->acl_cnt);
+	if (count > XFS_ACL_MAX_ENTRIES)
+		return ERR_PTR(-EFSCORRUPTED);
+
+	acl = posix_acl_alloc(count, GFP_KERNEL);
+	if (!acl)
+		return ERR_PTR(-ENOMEM);
+
+	for (i = 0; i < count; i++) {
+		acl_e = &acl->a_entries[i];
+		ace = &aclp->acl_entry[i];
+
+		/*
+		 * The tag is 32 bits on disk and 16 bits in core.
+		 *
+		 * Because every access to it goes through the core
+		 * format first this is not a problem.
+		 */
+		acl_e->e_tag = be32_to_cpu(ace->ae_tag);
+		acl_e->e_perm = be16_to_cpu(ace->ae_perm);
+
+		switch (acl_e->e_tag) {
+		case ACL_USER:
+		case ACL_GROUP:
+			acl_e->e_id = be32_to_cpu(ace->ae_id);
+			break;
+		case ACL_USER_OBJ:
+		case ACL_GROUP_OBJ:
+		case ACL_MASK:
+		case ACL_OTHER:
+			acl_e->e_id = ACL_UNDEFINED_ID;
+			break;
+		default:
+			goto fail;
+		}
+	}
+	return acl;
+
+fail:
+	posix_acl_release(acl);
+	return ERR_PTR(-EINVAL);
+}
+
+STATIC void
+xfs_acl_to_disk(struct xfs_acl *aclp, const struct posix_acl *acl)
+{
+	const struct posix_acl_entry *acl_e;
+	struct xfs_acl_entry *ace;
+	int i;
+
+	aclp->acl_cnt = cpu_to_be32(acl->a_count);
+	for (i = 0; i < acl->a_count; i++) {
+		ace = &aclp->acl_entry[i];
+		acl_e = &acl->a_entries[i];
+
+		ace->ae_tag = cpu_to_be32(acl_e->e_tag);
+		ace->ae_id = cpu_to_be32(acl_e->e_id);
+		ace->ae_perm = cpu_to_be16(acl_e->e_perm);
+	}
+}
+
+struct posix_acl *
+xfs_get_acl(struct inode *inode, int type)
+{
+	struct xfs_inode *ip = XFS_I(inode);
+	struct posix_acl *acl;
+	struct xfs_acl *xfs_acl;
+	int len = sizeof(struct xfs_acl);
+	unsigned char *ea_name;
+	int error;
+
+	acl = get_cached_acl(inode, type);
+	if (acl != ACL_NOT_CACHED)
+		return acl;
+
+	switch (type) {
+	case ACL_TYPE_ACCESS:
+		ea_name = SGI_ACL_FILE;
+		break;
+	case ACL_TYPE_DEFAULT:
+		ea_name = SGI_ACL_DEFAULT;
+		break;
+	default:
+		BUG();
+	}
+
+	/*
+	 * If we have a cached ACLs value just return it, not need to
+	 * go out to the disk.
+	 */
+
+	xfs_acl = kzalloc(sizeof(struct xfs_acl), GFP_KERNEL);
+	if (!xfs_acl)
+		return ERR_PTR(-ENOMEM);
+
+	error = -xfs_attr_get(ip, ea_name, (unsigned char *)xfs_acl,
+							&len, ATTR_ROOT);
+	if (error) {
+		/*
+		 * If the attribute doesn't exist make sure we have a negative
+		 * cache entry, for any other error assume it is transient and
+		 * leave the cache entry as ACL_NOT_CACHED.
+		 */
+		if (error == -ENOATTR) {
+			acl = NULL;
+			goto out_update_cache;
+		}
+		goto out;
+	}
+
+	acl = xfs_acl_from_disk(xfs_acl);
+	if (IS_ERR(acl))
+		goto out;
+
+ out_update_cache:
+	set_cached_acl(inode, type, acl);
+ out:
+	kfree(xfs_acl);
+	return acl;
+}
+
+STATIC int
+xfs_set_acl(struct inode *inode, int type, struct posix_acl *acl)
+{
+	struct xfs_inode *ip = XFS_I(inode);
+	unsigned char *ea_name;
+	int error;
+
+	if (S_ISLNK(inode->i_mode))
+		return -EOPNOTSUPP;
+
+	switch (type) {
+	case ACL_TYPE_ACCESS:
+		ea_name = SGI_ACL_FILE;
+		break;
+	case ACL_TYPE_DEFAULT:
+		if (!S_ISDIR(inode->i_mode))
+			return acl ? -EACCES : 0;
+		ea_name = SGI_ACL_DEFAULT;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	if (acl) {
+		struct xfs_acl *xfs_acl;
+		int len;
+
+		xfs_acl = kzalloc(sizeof(struct xfs_acl), GFP_KERNEL);
+		if (!xfs_acl)
+			return -ENOMEM;
+
+		xfs_acl_to_disk(xfs_acl, acl);
+		len = sizeof(struct xfs_acl) -
+			(sizeof(struct xfs_acl_entry) *
+			 (XFS_ACL_MAX_ENTRIES - acl->a_count));
+
+		error = -xfs_attr_set(ip, ea_name, (unsigned char *)xfs_acl,
+				len, ATTR_ROOT);
+
+		kfree(xfs_acl);
+	} else {
+		/*
+		 * A NULL ACL argument means we want to remove the ACL.
+		 */
+		error = -xfs_attr_remove(ip, ea_name, ATTR_ROOT);
+
+		/*
+		 * If the attribute didn't exist to start with that's fine.
+		 */
+		if (error == -ENOATTR)
+			error = 0;
+	}
+
+	if (!error)
+		set_cached_acl(inode, type, acl);
+	return error;
+}
+
+int
+xfs_check_acl(struct inode *inode, int mask, unsigned int flags)
+{
+	struct xfs_inode *ip;
+	struct posix_acl *acl;
+	int error = -EAGAIN;
+
+	ip = XFS_I(inode);
+	trace_xfs_check_acl(ip);
+
+	/*
+	 * If there is no attribute fork no ACL exists on this inode and
+	 * we can skip the whole exercise.
+	 */
+	if (!XFS_IFORK_Q(ip))
+		return -EAGAIN;
+
+	if (flags & IPERM_FLAG_RCU) {
+		if (!negative_cached_acl(inode, ACL_TYPE_ACCESS))
+			return -ECHILD;
+		return -EAGAIN;
+	}
+
+	acl = xfs_get_acl(inode, ACL_TYPE_ACCESS);
+	if (IS_ERR(acl))
+		return PTR_ERR(acl);
+	if (acl) {
+		error = posix_acl_permission(inode, acl, mask);
+		posix_acl_release(acl);
+	}
+
+	return error;
+}
+
+static int
+xfs_set_mode(struct inode *inode, mode_t mode)
+{
+	int error = 0;
+
+	if (mode != inode->i_mode) {
+		struct iattr iattr;
+
+		iattr.ia_valid = ATTR_MODE | ATTR_CTIME;
+		iattr.ia_mode = mode;
+		iattr.ia_ctime = current_fs_time(inode->i_sb);
+
+		error = -xfs_setattr(XFS_I(inode), &iattr, XFS_ATTR_NOACL);
+	}
+
+	return error;
+}
+
+static int
+xfs_acl_exists(struct inode *inode, unsigned char *name)
+{
+	int len = sizeof(struct xfs_acl);
+
+	return (xfs_attr_get(XFS_I(inode), name, NULL, &len,
+			    ATTR_ROOT|ATTR_KERNOVAL) == 0);
+}
+
+int
+posix_acl_access_exists(struct inode *inode)
+{
+	return xfs_acl_exists(inode, SGI_ACL_FILE);
+}
+
+int
+posix_acl_default_exists(struct inode *inode)
+{
+	if (!S_ISDIR(inode->i_mode))
+		return 0;
+	return xfs_acl_exists(inode, SGI_ACL_DEFAULT);
+}
+
+/*
+ * No need for i_mutex because the inode is not yet exposed to the VFS.
+ */
+int
+xfs_inherit_acl(struct inode *inode, struct posix_acl *default_acl)
+{
+	struct posix_acl *clone;
+	mode_t mode;
+	int error = 0, inherit = 0;
+
+	if (S_ISDIR(inode->i_mode)) {
+		error = xfs_set_acl(inode, ACL_TYPE_DEFAULT, default_acl);
+		if (error)
+			return error;
+	}
+
+	clone = posix_acl_clone(default_acl, GFP_KERNEL);
+	if (!clone)
+		return -ENOMEM;
+
+	mode = inode->i_mode;
+	error = posix_acl_create_masq(clone, &mode);
+	if (error < 0)
+		goto out_release_clone;
+
+	/*
+	 * If posix_acl_create_masq returns a positive value we need to
+	 * inherit a permission that can't be represented using the Unix
+	 * mode bits and we actually need to set an ACL.
+	 */
+	if (error > 0)
+		inherit = 1;
+
+	error = xfs_set_mode(inode, mode);
+	if (error)
+		goto out_release_clone;
+
+	if (inherit)
+		error = xfs_set_acl(inode, ACL_TYPE_ACCESS, clone);
+
+ out_release_clone:
+	posix_acl_release(clone);
+	return error;
+}
+
+int
+xfs_acl_chmod(struct inode *inode)
+{
+	struct posix_acl *acl, *clone;
+	int error;
+
+	if (S_ISLNK(inode->i_mode))
+		return -EOPNOTSUPP;
+
+	acl = xfs_get_acl(inode, ACL_TYPE_ACCESS);
+	if (IS_ERR(acl) || !acl)
+		return PTR_ERR(acl);
+
+	clone = posix_acl_clone(acl, GFP_KERNEL);
+	posix_acl_release(acl);
+	if (!clone)
+		return -ENOMEM;
+
+	error = posix_acl_chmod_masq(clone, inode->i_mode);
+	if (!error)
+		error = xfs_set_acl(inode, ACL_TYPE_ACCESS, clone);
+
+	posix_acl_release(clone);
+	return error;
+}
+
+static int
+xfs_xattr_acl_get(struct dentry *dentry, const char *name,
+		void *value, size_t size, int type)
+{
+	struct posix_acl *acl;
+	int error;
+
+	acl = xfs_get_acl(dentry->d_inode, type);
+	if (IS_ERR(acl))
+		return PTR_ERR(acl);
+	if (acl == NULL)
+		return -ENODATA;
+
+	error = posix_acl_to_xattr(acl, value, size);
+	posix_acl_release(acl);
+
+	return error;
+}
+
+static int
+xfs_xattr_acl_set(struct dentry *dentry, const char *name,
+		const void *value, size_t size, int flags, int type)
+{
+	struct inode *inode = dentry->d_inode;
+	struct posix_acl *acl = NULL;
+	int error = 0;
+
+	if (flags & XATTR_CREATE)
+		return -EINVAL;
+	if (type == ACL_TYPE_DEFAULT && !S_ISDIR(inode->i_mode))
+		return value ? -EACCES : 0;
+	if ((current_fsuid() != inode->i_uid) && !capable(CAP_FOWNER))
+		return -EPERM;
+
+	if (!value)
+		goto set_acl;
+
+	acl = posix_acl_from_xattr(value, size);
+	if (!acl) {
+		/*
+		 * acl_set_file(3) may request that we set default ACLs with
+		 * zero length -- defend (gracefully) against that here.
+		 */
+		goto out;
+	}
+	if (IS_ERR(acl)) {
+		error = PTR_ERR(acl);
+		goto out;
+	}
+
+	error = posix_acl_valid(acl);
+	if (error)
+		goto out_release;
+
+	error = -EINVAL;
+	if (acl->a_count > XFS_ACL_MAX_ENTRIES)
+		goto out_release;
+
+	if (type == ACL_TYPE_ACCESS) {
+		mode_t mode = inode->i_mode;
+		error = posix_acl_equiv_mode(acl, &mode);
+
+		if (error <= 0) {
+			posix_acl_release(acl);
+			acl = NULL;
+
+			if (error < 0)
+				return error;
+		}
+
+		error = xfs_set_mode(inode, mode);
+		if (error)
+			goto out_release;
+	}
+
+ set_acl:
+	error = xfs_set_acl(inode, type, acl);
+ out_release:
+	posix_acl_release(acl);
+ out:
+	return error;
+}
+
+const struct xattr_handler xfs_xattr_acl_access_handler = {
+	.prefix	= POSIX_ACL_XATTR_ACCESS,
+	.flags	= ACL_TYPE_ACCESS,
+	.get	= xfs_xattr_acl_get,
+	.set	= xfs_xattr_acl_set,
+};
+
+const struct xattr_handler xfs_xattr_acl_default_handler = {
+	.prefix	= POSIX_ACL_XATTR_DEFAULT,
+	.flags	= ACL_TYPE_DEFAULT,
+	.get	= xfs_xattr_acl_get,
+	.set	= xfs_xattr_acl_set,
+};
diff --git a/fs/xfs/linux-2.6/xfs_aops.c b/fs/xfs/linux-2.6/xfs_aops.c
new file mode 100644
index 00000000..79ce38be
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_aops.c
@@ -0,0 +1,1506 @@
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_trans.h"
+#include "xfs_mount.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_alloc.h"
+#include "xfs_error.h"
+#include "xfs_rw.h"
+#include "xfs_iomap.h"
+#include "xfs_vnodeops.h"
+#include "xfs_trace.h"
+#include "xfs_bmap.h"
+#include <linux/gfp.h>
+#include <linux/mpage.h>
+#include <linux/pagevec.h>
+#include <linux/writeback.h>
+
+
+/*
+ * Prime number of hash buckets since address is used as the key.
+ */
+#define NVSYNC		37
+#define to_ioend_wq(v)	(&xfs_ioend_wq[((unsigned long)v) % NVSYNC])
+static wait_queue_head_t xfs_ioend_wq[NVSYNC];
+
+void __init
+xfs_ioend_init(void)
+{
+	int i;
+
+	for (i = 0; i < NVSYNC; i++)
+		init_waitqueue_head(&xfs_ioend_wq[i]);
+}
+
+void
+xfs_ioend_wait(
+	xfs_inode_t	*ip)
+{
+	wait_queue_head_t *wq = to_ioend_wq(ip);
+
+	wait_event(*wq, (atomic_read(&ip->i_iocount) == 0));
+}
+
+STATIC void
+xfs_ioend_wake(
+	xfs_inode_t	*ip)
+{
+	if (atomic_dec_and_test(&ip->i_iocount))
+		wake_up(to_ioend_wq(ip));
+}
+
+void
+xfs_count_page_state(
+	struct page		*page,
+	int			*delalloc,
+	int			*unwritten)
+{
+	struct buffer_head	*bh, *head;
+
+	*delalloc = *unwritten = 0;
+
+	bh = head = page_buffers(page);
+	do {
+		if (buffer_unwritten(bh))
+			(*unwritten) = 1;
+		else if (buffer_delay(bh))
+			(*delalloc) = 1;
+	} while ((bh = bh->b_this_page) != head);
+}
+
+STATIC struct block_device *
+xfs_find_bdev_for_inode(
+	struct inode		*inode)
+{
+	struct xfs_inode	*ip = XFS_I(inode);
+	struct xfs_mount	*mp = ip->i_mount;
+
+	if (XFS_IS_REALTIME_INODE(ip))
+		return mp->m_rtdev_targp->bt_bdev;
+	else
+		return mp->m_ddev_targp->bt_bdev;
+}
+
+/*
+ * We're now finished for good with this ioend structure.
+ * Update the page state via the associated buffer_heads,
+ * release holds on the inode and bio, and finally free
+ * up memory.  Do not use the ioend after this.
+ */
+STATIC void
+xfs_destroy_ioend(
+	xfs_ioend_t		*ioend)
+{
+	struct buffer_head	*bh, *next;
+	struct xfs_inode	*ip = XFS_I(ioend->io_inode);
+
+	for (bh = ioend->io_buffer_head; bh; bh = next) {
+		next = bh->b_private;
+		bh->b_end_io(bh, !ioend->io_error);
+	}
+
+	/*
+	 * Volume managers supporting multiple paths can send back ENODEV
+	 * when the final path disappears.  In this case continuing to fill
+	 * the page cache with dirty data which cannot be written out is
+	 * evil, so prevent that.
+	 */
+	if (unlikely(ioend->io_error == -ENODEV)) {
+		xfs_do_force_shutdown(ip->i_mount, SHUTDOWN_DEVICE_REQ,
+				      __FILE__, __LINE__);
+	}
+
+	xfs_ioend_wake(ip);
+	mempool_free(ioend, xfs_ioend_pool);
+}
+
+/*
+ * If the end of the current ioend is beyond the current EOF,
+ * return the new EOF value, otherwise zero.
+ */
+STATIC xfs_fsize_t
+xfs_ioend_new_eof(
+	xfs_ioend_t		*ioend)
+{
+	xfs_inode_t		*ip = XFS_I(ioend->io_inode);
+	xfs_fsize_t		isize;
+	xfs_fsize_t		bsize;
+
+	bsize = ioend->io_offset + ioend->io_size;
+	isize = MAX(ip->i_size, ip->i_new_size);
+	isize = MIN(isize, bsize);
+	return isize > ip->i_d.di_size ? isize : 0;
+}
+
+/*
+ * Update on-disk file size now that data has been written to disk.  The
+ * current in-memory file size is i_size.  If a write is beyond eof i_new_size
+ * will be the intended file size until i_size is updated.  If this write does
+ * not extend all the way to the valid file size then restrict this update to
+ * the end of the write.
+ *
+ * This function does not block as blocking on the inode lock in IO completion
+ * can lead to IO completion order dependency deadlocks.. If it can't get the
+ * inode ilock it will return EAGAIN. Callers must handle this.
+ */
+STATIC int
+xfs_setfilesize(
+	xfs_ioend_t		*ioend)
+{
+	xfs_inode_t		*ip = XFS_I(ioend->io_inode);
+	xfs_fsize_t		isize;
+
+	if (unlikely(ioend->io_error))
+		return 0;
+
+	if (!xfs_ilock_nowait(ip, XFS_ILOCK_EXCL))
+		return EAGAIN;
+
+	isize = xfs_ioend_new_eof(ioend);
+	if (isize) {
+		ip->i_d.di_size = isize;
+		xfs_mark_inode_dirty(ip);
+	}
+
+	xfs_iunlock(ip, XFS_ILOCK_EXCL);
+	return 0;
+}
+
+/*
+ * Schedule IO completion handling on the final put of an ioend.
+ */
+STATIC void
+xfs_finish_ioend(
+	struct xfs_ioend	*ioend)
+{
+	if (atomic_dec_and_test(&ioend->io_remaining)) {
+		if (ioend->io_type == IO_UNWRITTEN)
+			queue_work(xfsconvertd_workqueue, &ioend->io_work);
+		else
+			queue_work(xfsdatad_workqueue, &ioend->io_work);
+	}
+}
+
+/*
+ * IO write completion.
+ */
+STATIC void
+xfs_end_io(
+	struct work_struct *work)
+{
+	xfs_ioend_t	*ioend = container_of(work, xfs_ioend_t, io_work);
+	struct xfs_inode *ip = XFS_I(ioend->io_inode);
+	int		error = 0;
+
+	/*
+	 * For unwritten extents we need to issue transactions to convert a
+	 * range to normal written extens after the data I/O has finished.
+	 */
+	if (ioend->io_type == IO_UNWRITTEN &&
+	    likely(!ioend->io_error && !XFS_FORCED_SHUTDOWN(ip->i_mount))) {
+
+		error = xfs_iomap_write_unwritten(ip, ioend->io_offset,
+						 ioend->io_size);
+		if (error)
+			ioend->io_error = error;
+	}
+
+	/*
+	 * We might have to update the on-disk file size after extending
+	 * writes.
+	 */
+	error = xfs_setfilesize(ioend);
+	ASSERT(!error || error == EAGAIN);
+
+	/*
+	 * If we didn't complete processing of the ioend, requeue it to the
+	 * tail of the workqueue for another attempt later. Otherwise destroy
+	 * it.
+	 */
+	if (error == EAGAIN) {
+		atomic_inc(&ioend->io_remaining);
+		xfs_finish_ioend(ioend);
+		/* ensure we don't spin on blocked ioends */
+		delay(1);
+	} else {
+		if (ioend->io_iocb)
+			aio_complete(ioend->io_iocb, ioend->io_result, 0);
+		xfs_destroy_ioend(ioend);
+	}
+}
+
+/*
+ * Call IO completion handling in caller context on the final put of an ioend.
+ */
+STATIC void
+xfs_finish_ioend_sync(
+	struct xfs_ioend	*ioend)
+{
+	if (atomic_dec_and_test(&ioend->io_remaining))
+		xfs_end_io(&ioend->io_work);
+}
+
+/*
+ * Allocate and initialise an IO completion structure.
+ * We need to track unwritten extent write completion here initially.
+ * We'll need to extend this for updating the ondisk inode size later
+ * (vs. incore size).
+ */
+STATIC xfs_ioend_t *
+xfs_alloc_ioend(
+	struct inode		*inode,
+	unsigned int		type)
+{
+	xfs_ioend_t		*ioend;
+
+	ioend = mempool_alloc(xfs_ioend_pool, GFP_NOFS);
+
+	/*
+	 * Set the count to 1 initially, which will prevent an I/O
+	 * completion callback from happening before we have started
+	 * all the I/O from calling the completion routine too early.
+	 */
+	atomic_set(&ioend->io_remaining, 1);
+	ioend->io_error = 0;
+	ioend->io_list = NULL;
+	ioend->io_type = type;
+	ioend->io_inode = inode;
+	ioend->io_buffer_head = NULL;
+	ioend->io_buffer_tail = NULL;
+	atomic_inc(&XFS_I(ioend->io_inode)->i_iocount);
+	ioend->io_offset = 0;
+	ioend->io_size = 0;
+	ioend->io_iocb = NULL;
+	ioend->io_result = 0;
+
+	INIT_WORK(&ioend->io_work, xfs_end_io);
+	return ioend;
+}
+
+STATIC int
+xfs_map_blocks(
+	struct inode		*inode,
+	loff_t			offset,
+	struct xfs_bmbt_irec	*imap,
+	int			type,
+	int			nonblocking)
+{
+	struct xfs_inode	*ip = XFS_I(inode);
+	struct xfs_mount	*mp = ip->i_mount;
+	ssize_t			count = 1 << inode->i_blkbits;
+	xfs_fileoff_t		offset_fsb, end_fsb;
+	int			error = 0;
+	int			bmapi_flags = XFS_BMAPI_ENTIRE;
+	int			nimaps = 1;
+
+	if (XFS_FORCED_SHUTDOWN(mp))
+		return -XFS_ERROR(EIO);
+
+	if (type == IO_UNWRITTEN)
+		bmapi_flags |= XFS_BMAPI_IGSTATE;
+
+	if (!xfs_ilock_nowait(ip, XFS_ILOCK_SHARED)) {
+		if (nonblocking)
+			return -XFS_ERROR(EAGAIN);
+		xfs_ilock(ip, XFS_ILOCK_SHARED);
+	}
+
+	ASSERT(ip->i_d.di_format != XFS_DINODE_FMT_BTREE ||
+	       (ip->i_df.if_flags & XFS_IFEXTENTS));
+	ASSERT(offset <= mp->m_maxioffset);
+
+	if (offset + count > mp->m_maxioffset)
+		count = mp->m_maxioffset - offset;
+	end_fsb = XFS_B_TO_FSB(mp, (xfs_ufsize_t)offset + count);
+	offset_fsb = XFS_B_TO_FSBT(mp, offset);
+	error = xfs_bmapi(NULL, ip, offset_fsb, end_fsb - offset_fsb,
+			  bmapi_flags,  NULL, 0, imap, &nimaps, NULL);
+	xfs_iunlock(ip, XFS_ILOCK_SHARED);
+
+	if (error)
+		return -XFS_ERROR(error);
+
+	if (type == IO_DELALLOC &&
+	    (!nimaps || isnullstartblock(imap->br_startblock))) {
+		error = xfs_iomap_write_allocate(ip, offset, count, imap);
+		if (!error)
+			trace_xfs_map_blocks_alloc(ip, offset, count, type, imap);
+		return -XFS_ERROR(error);
+	}
+
+#ifdef DEBUG
+	if (type == IO_UNWRITTEN) {
+		ASSERT(nimaps);
+		ASSERT(imap->br_startblock != HOLESTARTBLOCK);
+		ASSERT(imap->br_startblock != DELAYSTARTBLOCK);
+	}
+#endif
+	if (nimaps)
+		trace_xfs_map_blocks_found(ip, offset, count, type, imap);
+	return 0;
+}
+
+STATIC int
+xfs_imap_valid(
+	struct inode		*inode,
+	struct xfs_bmbt_irec	*imap,
+	xfs_off_t		offset)
+{
+	offset >>= inode->i_blkbits;
+
+	return offset >= imap->br_startoff &&
+		offset < imap->br_startoff + imap->br_blockcount;
+}
+
+/*
+ * BIO completion handler for buffered IO.
+ */
+STATIC void
+xfs_end_bio(
+	struct bio		*bio,
+	int			error)
+{
+	xfs_ioend_t		*ioend = bio->bi_private;
+
+	ASSERT(atomic_read(&bio->bi_cnt) >= 1);
+	ioend->io_error = test_bit(BIO_UPTODATE, &bio->bi_flags) ? 0 : error;
+
+	/* Toss bio and pass work off to an xfsdatad thread */
+	bio->bi_private = NULL;
+	bio->bi_end_io = NULL;
+	bio_put(bio);
+
+	xfs_finish_ioend(ioend);
+}
+
+STATIC void
+xfs_submit_ioend_bio(
+	struct writeback_control *wbc,
+	xfs_ioend_t		*ioend,
+	struct bio		*bio)
+{
+	atomic_inc(&ioend->io_remaining);
+	bio->bi_private = ioend;
+	bio->bi_end_io = xfs_end_bio;
+
+	/*
+	 * If the I/O is beyond EOF we mark the inode dirty immediately
+	 * but don't update the inode size until I/O completion.
+	 */
+	if (xfs_ioend_new_eof(ioend))
+		xfs_mark_inode_dirty(XFS_I(ioend->io_inode));
+
+	submit_bio(wbc->sync_mode == WB_SYNC_ALL ? WRITE_SYNC : WRITE, bio);
+}
+
+STATIC struct bio *
+xfs_alloc_ioend_bio(
+	struct buffer_head	*bh)
+{
+	int			nvecs = bio_get_nr_vecs(bh->b_bdev);
+	struct bio		*bio = bio_alloc(GFP_NOIO, nvecs);
+
+	ASSERT(bio->bi_private == NULL);
+	bio->bi_sector = bh->b_blocknr * (bh->b_size >> 9);
+	bio->bi_bdev = bh->b_bdev;
+	return bio;
+}
+
+STATIC void
+xfs_start_buffer_writeback(
+	struct buffer_head	*bh)
+{
+	ASSERT(buffer_mapped(bh));
+	ASSERT(buffer_locked(bh));
+	ASSERT(!buffer_delay(bh));
+	ASSERT(!buffer_unwritten(bh));
+
+	mark_buffer_async_write(bh);
+	set_buffer_uptodate(bh);
+	clear_buffer_dirty(bh);
+}
+
+STATIC void
+xfs_start_page_writeback(
+	struct page		*page,
+	int			clear_dirty,
+	int			buffers)
+{
+	ASSERT(PageLocked(page));
+	ASSERT(!PageWriteback(page));
+	if (clear_dirty)
+		clear_page_dirty_for_io(page);
+	set_page_writeback(page);
+	unlock_page(page);
+	/* If no buffers on the page are to be written, finish it here */
+	if (!buffers)
+		end_page_writeback(page);
+}
+
+static inline int bio_add_buffer(struct bio *bio, struct buffer_head *bh)
+{
+	return bio_add_page(bio, bh->b_page, bh->b_size, bh_offset(bh));
+}
+
+/*
+ * Submit all of the bios for all of the ioends we have saved up, covering the
+ * initial writepage page and also any probed pages.
+ *
+ * Because we may have multiple ioends spanning a page, we need to start
+ * writeback on all the buffers before we submit them for I/O. If we mark the
+ * buffers as we got, then we can end up with a page that only has buffers
+ * marked async write and I/O complete on can occur before we mark the other
+ * buffers async write.
+ *
+ * The end result of this is that we trip a bug in end_page_writeback() because
+ * we call it twice for the one page as the code in end_buffer_async_write()
+ * assumes that all buffers on the page are started at the same time.
+ *
+ * The fix is two passes across the ioend list - one to start writeback on the
+ * buffer_heads, and then submit them for I/O on the second pass.
+ */
+STATIC void
+xfs_submit_ioend(
+	struct writeback_control *wbc,
+	xfs_ioend_t		*ioend)
+{
+	xfs_ioend_t		*head = ioend;
+	xfs_ioend_t		*next;
+	struct buffer_head	*bh;
+	struct bio		*bio;
+	sector_t		lastblock = 0;
+
+	/* Pass 1 - start writeback */
+	do {
+		next = ioend->io_list;
+		for (bh = ioend->io_buffer_head; bh; bh = bh->b_private)
+			xfs_start_buffer_writeback(bh);
+	} while ((ioend = next) != NULL);
+
+	/* Pass 2 - submit I/O */
+	ioend = head;
+	do {
+		next = ioend->io_list;
+		bio = NULL;
+
+		for (bh = ioend->io_buffer_head; bh; bh = bh->b_private) {
+
+			if (!bio) {
+ retry:
+				bio = xfs_alloc_ioend_bio(bh);
+			} else if (bh->b_blocknr != lastblock + 1) {
+				xfs_submit_ioend_bio(wbc, ioend, bio);
+				goto retry;
+			}
+
+			if (bio_add_buffer(bio, bh) != bh->b_size) {
+				xfs_submit_ioend_bio(wbc, ioend, bio);
+				goto retry;
+			}
+
+			lastblock = bh->b_blocknr;
+		}
+		if (bio)
+			xfs_submit_ioend_bio(wbc, ioend, bio);
+		xfs_finish_ioend(ioend);
+	} while ((ioend = next) != NULL);
+}
+
+/*
+ * Cancel submission of all buffer_heads so far in this endio.
+ * Toss the endio too.  Only ever called for the initial page
+ * in a writepage request, so only ever one page.
+ */
+STATIC void
+xfs_cancel_ioend(
+	xfs_ioend_t		*ioend)
+{
+	xfs_ioend_t		*next;
+	struct buffer_head	*bh, *next_bh;
+
+	do {
+		next = ioend->io_list;
+		bh = ioend->io_buffer_head;
+		do {
+			next_bh = bh->b_private;
+			clear_buffer_async_write(bh);
+			unlock_buffer(bh);
+		} while ((bh = next_bh) != NULL);
+
+		xfs_ioend_wake(XFS_I(ioend->io_inode));
+		mempool_free(ioend, xfs_ioend_pool);
+	} while ((ioend = next) != NULL);
+}
+
+/*
+ * Test to see if we've been building up a completion structure for
+ * earlier buffers -- if so, we try to append to this ioend if we
+ * can, otherwise we finish off any current ioend and start another.
+ * Return true if we've finished the given ioend.
+ */
+STATIC void
+xfs_add_to_ioend(
+	struct inode		*inode,
+	struct buffer_head	*bh,
+	xfs_off_t		offset,
+	unsigned int		type,
+	xfs_ioend_t		**result,
+	int			need_ioend)
+{
+	xfs_ioend_t		*ioend = *result;
+
+	if (!ioend || need_ioend || type != ioend->io_type) {
+		xfs_ioend_t	*previous = *result;
+
+		ioend = xfs_alloc_ioend(inode, type);
+		ioend->io_offset = offset;
+		ioend->io_buffer_head = bh;
+		ioend->io_buffer_tail = bh;
+		if (previous)
+			previous->io_list = ioend;
+		*result = ioend;
+	} else {
+		ioend->io_buffer_tail->b_private = bh;
+		ioend->io_buffer_tail = bh;
+	}
+
+	bh->b_private = NULL;
+	ioend->io_size += bh->b_size;
+}
+
+STATIC void
+xfs_map_buffer(
+	struct inode		*inode,
+	struct buffer_head	*bh,
+	struct xfs_bmbt_irec	*imap,
+	xfs_off_t		offset)
+{
+	sector_t		bn;
+	struct xfs_mount	*m = XFS_I(inode)->i_mount;
+	xfs_off_t		iomap_offset = XFS_FSB_TO_B(m, imap->br_startoff);
+	xfs_daddr_t		iomap_bn = xfs_fsb_to_db(XFS_I(inode), imap->br_startblock);
+
+	ASSERT(imap->br_startblock != HOLESTARTBLOCK);
+	ASSERT(imap->br_startblock != DELAYSTARTBLOCK);
+
+	bn = (iomap_bn >> (inode->i_blkbits - BBSHIFT)) +
+	      ((offset - iomap_offset) >> inode->i_blkbits);
+
+	ASSERT(bn || XFS_IS_REALTIME_INODE(XFS_I(inode)));
+
+	bh->b_blocknr = bn;
+	set_buffer_mapped(bh);
+}
+
+STATIC void
+xfs_map_at_offset(
+	struct inode		*inode,
+	struct buffer_head	*bh,
+	struct xfs_bmbt_irec	*imap,
+	xfs_off_t		offset)
+{
+	ASSERT(imap->br_startblock != HOLESTARTBLOCK);
+	ASSERT(imap->br_startblock != DELAYSTARTBLOCK);
+
+	xfs_map_buffer(inode, bh, imap, offset);
+	set_buffer_mapped(bh);
+	clear_buffer_delay(bh);
+	clear_buffer_unwritten(bh);
+}
+
+/*
+ * Test if a given page is suitable for writing as part of an unwritten
+ * or delayed allocate extent.
+ */
+STATIC int
+xfs_is_delayed_page(
+	struct page		*page,
+	unsigned int		type)
+{
+	if (PageWriteback(page))
+		return 0;
+
+	if (page->mapping && page_has_buffers(page)) {
+		struct buffer_head	*bh, *head;
+		int			acceptable = 0;
+
+		bh = head = page_buffers(page);
+		do {
+			if (buffer_unwritten(bh))
+				acceptable = (type == IO_UNWRITTEN);
+			else if (buffer_delay(bh))
+				acceptable = (type == IO_DELALLOC);
+			else if (buffer_dirty(bh) && buffer_mapped(bh))
+				acceptable = (type == IO_OVERWRITE);
+			else
+				break;
+		} while ((bh = bh->b_this_page) != head);
+
+		if (acceptable)
+			return 1;
+	}
+
+	return 0;
+}
+
+/*
+ * Allocate & map buffers for page given the extent map. Write it out.
+ * except for the original page of a writepage, this is called on
+ * delalloc/unwritten pages only, for the original page it is possible
+ * that the page has no mapping at all.
+ */
+STATIC int
+xfs_convert_page(
+	struct inode		*inode,
+	struct page		*page,
+	loff_t			tindex,
+	struct xfs_bmbt_irec	*imap,
+	xfs_ioend_t		**ioendp,
+	struct writeback_control *wbc)
+{
+	struct buffer_head	*bh, *head;
+	xfs_off_t		end_offset;
+	unsigned long		p_offset;
+	unsigned int		type;
+	int			len, page_dirty;
+	int			count = 0, done = 0, uptodate = 1;
+ 	xfs_off_t		offset = page_offset(page);
+
+	if (page->index != tindex)
+		goto fail;
+	if (!trylock_page(page))
+		goto fail;
+	if (PageWriteback(page))
+		goto fail_unlock_page;
+	if (page->mapping != inode->i_mapping)
+		goto fail_unlock_page;
+	if (!xfs_is_delayed_page(page, (*ioendp)->io_type))
+		goto fail_unlock_page;
+
+	/*
+	 * page_dirty is initially a count of buffers on the page before
+	 * EOF and is decremented as we move each into a cleanable state.
+	 *
+	 * Derivation:
+	 *
+	 * End offset is the highest offset that this page should represent.
+	 * If we are on the last page, (end_offset & (PAGE_CACHE_SIZE - 1))
+	 * will evaluate non-zero and be less than PAGE_CACHE_SIZE and
+	 * hence give us the correct page_dirty count. On any other page,
+	 * it will be zero and in that case we need page_dirty to be the
+	 * count of buffers on the page.
+	 */
+	end_offset = min_t(unsigned long long,
+			(xfs_off_t)(page->index + 1) << PAGE_CACHE_SHIFT,
+			i_size_read(inode));
+
+	len = 1 << inode->i_blkbits;
+	p_offset = min_t(unsigned long, end_offset & (PAGE_CACHE_SIZE - 1),
+					PAGE_CACHE_SIZE);
+	p_offset = p_offset ? roundup(p_offset, len) : PAGE_CACHE_SIZE;
+	page_dirty = p_offset / len;
+
+	bh = head = page_buffers(page);
+	do {
+		if (offset >= end_offset)
+			break;
+		if (!buffer_uptodate(bh))
+			uptodate = 0;
+		if (!(PageUptodate(page) || buffer_uptodate(bh))) {
+			done = 1;
+			continue;
+		}
+
+		if (buffer_unwritten(bh) || buffer_delay(bh) ||
+		    buffer_mapped(bh)) {
+			if (buffer_unwritten(bh))
+				type = IO_UNWRITTEN;
+			else if (buffer_delay(bh))
+				type = IO_DELALLOC;
+			else
+				type = IO_OVERWRITE;
+
+			if (!xfs_imap_valid(inode, imap, offset)) {
+				done = 1;
+				continue;
+			}
+
+			lock_buffer(bh);
+			if (type != IO_OVERWRITE)
+				xfs_map_at_offset(inode, bh, imap, offset);
+			xfs_add_to_ioend(inode, bh, offset, type,
+					 ioendp, done);
+
+			page_dirty--;
+			count++;
+		} else {
+			done = 1;
+		}
+	} while (offset += len, (bh = bh->b_this_page) != head);
+
+	if (uptodate && bh == head)
+		SetPageUptodate(page);
+
+	if (count) {
+		if (--wbc->nr_to_write <= 0 &&
+		    wbc->sync_mode == WB_SYNC_NONE)
+			done = 1;
+	}
+	xfs_start_page_writeback(page, !page_dirty, count);
+
+	return done;
+ fail_unlock_page:
+	unlock_page(page);
+ fail:
+	return 1;
+}
+
+/*
+ * Convert & write out a cluster of pages in the same extent as defined
+ * by mp and following the start page.
+ */
+STATIC void
+xfs_cluster_write(
+	struct inode		*inode,
+	pgoff_t			tindex,
+	struct xfs_bmbt_irec	*imap,
+	xfs_ioend_t		**ioendp,
+	struct writeback_control *wbc,
+	pgoff_t			tlast)
+{
+	struct pagevec		pvec;
+	int			done = 0, i;
+
+	pagevec_init(&pvec, 0);
+	while (!done && tindex <= tlast) {
+		unsigned len = min_t(pgoff_t, PAGEVEC_SIZE, tlast - tindex + 1);
+
+		if (!pagevec_lookup(&pvec, inode->i_mapping, tindex, len))
+			break;
+
+		for (i = 0; i < pagevec_count(&pvec); i++) {
+			done = xfs_convert_page(inode, pvec.pages[i], tindex++,
+					imap, ioendp, wbc);
+			if (done)
+				break;
+		}
+
+		pagevec_release(&pvec);
+		cond_resched();
+	}
+}
+
+STATIC void
+xfs_vm_invalidatepage(
+	struct page		*page,
+	unsigned long		offset)
+{
+	trace_xfs_invalidatepage(page->mapping->host, page, offset);
+	block_invalidatepage(page, offset);
+}
+
+/*
+ * If the page has delalloc buffers on it, we need to punch them out before we
+ * invalidate the page. If we don't, we leave a stale delalloc mapping on the
+ * inode that can trip a BUG() in xfs_get_blocks() later on if a direct IO read
+ * is done on that same region - the delalloc extent is returned when none is
+ * supposed to be there.
+ *
+ * We prevent this by truncating away the delalloc regions on the page before
+ * invalidating it. Because they are delalloc, we can do this without needing a
+ * transaction. Indeed - if we get ENOSPC errors, we have to be able to do this
+ * truncation without a transaction as there is no space left for block
+ * reservation (typically why we see a ENOSPC in writeback).
+ *
+ * This is not a performance critical path, so for now just do the punching a
+ * buffer head at a time.
+ */
+STATIC void
+xfs_aops_discard_page(
+	struct page		*page)
+{
+	struct inode		*inode = page->mapping->host;
+	struct xfs_inode	*ip = XFS_I(inode);
+	struct buffer_head	*bh, *head;
+	loff_t			offset = page_offset(page);
+
+	if (!xfs_is_delayed_page(page, IO_DELALLOC))
+		goto out_invalidate;
+
+	if (XFS_FORCED_SHUTDOWN(ip->i_mount))
+		goto out_invalidate;
+
+	xfs_alert(ip->i_mount,
+		"page discard on page %p, inode 0x%llx, offset %llu.",
+			page, ip->i_ino, offset);
+
+	xfs_ilock(ip, XFS_ILOCK_EXCL);
+	bh = head = page_buffers(page);
+	do {
+		int		error;
+		xfs_fileoff_t	start_fsb;
+
+		if (!buffer_delay(bh))
+			goto next_buffer;
+
+		start_fsb = XFS_B_TO_FSBT(ip->i_mount, offset);
+		error = xfs_bmap_punch_delalloc_range(ip, start_fsb, 1);
+		if (error) {
+			/* something screwed, just bail */
+			if (!XFS_FORCED_SHUTDOWN(ip->i_mount)) {
+				xfs_alert(ip->i_mount,
+			"page discard unable to remove delalloc mapping.");
+			}
+			break;
+		}
+next_buffer:
+		offset += 1 << inode->i_blkbits;
+
+	} while ((bh = bh->b_this_page) != head);
+
+	xfs_iunlock(ip, XFS_ILOCK_EXCL);
+out_invalidate:
+	xfs_vm_invalidatepage(page, 0);
+	return;
+}
+
+/*
+ * Write out a dirty page.
+ *
+ * For delalloc space on the page we need to allocate space and flush it.
+ * For unwritten space on the page we need to start the conversion to
+ * regular allocated space.
+ * For any other dirty buffer heads on the page we should flush them.
+ *
+ * If we detect that a transaction would be required to flush the page, we
+ * have to check the process flags first, if we are already in a transaction
+ * or disk I/O during allocations is off, we need to fail the writepage and
+ * redirty the page.
+ */
+STATIC int
+xfs_vm_writepage(
+	struct page		*page,
+	struct writeback_control *wbc)
+{
+	struct inode		*inode = page->mapping->host;
+	int			delalloc, unwritten;
+	struct buffer_head	*bh, *head;
+	struct xfs_bmbt_irec	imap;
+	xfs_ioend_t		*ioend = NULL, *iohead = NULL;
+	loff_t			offset;
+	unsigned int		type;
+	__uint64_t              end_offset;
+	pgoff_t                 end_index, last_index;
+	ssize_t			len;
+	int			err, imap_valid = 0, uptodate = 1;
+	int			count = 0;
+	int			nonblocking = 0;
+
+	trace_xfs_writepage(inode, page, 0);
+
+	ASSERT(page_has_buffers(page));
+
+	/*
+	 * Refuse to write the page out if we are called from reclaim context.
+	 *
+	 * This avoids stack overflows when called from deeply used stacks in
+	 * random callers for direct reclaim or memcg reclaim.  We explicitly
+	 * allow reclaim from kswapd as the stack usage there is relatively low.
+	 *
+	 * This should really be done by the core VM, but until that happens
+	 * filesystems like XFS, btrfs and ext4 have to take care of this
+	 * by themselves.
+	 */
+	if ((current->flags & (PF_MEMALLOC|PF_KSWAPD)) == PF_MEMALLOC)
+		goto redirty;
+
+	/*
+	 * We need a transaction if there are delalloc or unwritten buffers
+	 * on the page.
+	 *
+	 * If we need a transaction and the process flags say we are already
+	 * in a transaction, or no IO is allowed then mark the page dirty
+	 * again and leave the page as is.
+	 */
+	xfs_count_page_state(page, &delalloc, &unwritten);
+	if ((current->flags & PF_FSTRANS) && (delalloc || unwritten))
+		goto redirty;
+
+	/* Is this page beyond the end of the file? */
+	offset = i_size_read(inode);
+	end_index = offset >> PAGE_CACHE_SHIFT;
+	last_index = (offset - 1) >> PAGE_CACHE_SHIFT;
+	if (page->index >= end_index) {
+		if ((page->index >= end_index + 1) ||
+		    !(i_size_read(inode) & (PAGE_CACHE_SIZE - 1))) {
+			unlock_page(page);
+			return 0;
+		}
+	}
+
+	end_offset = min_t(unsigned long long,
+			(xfs_off_t)(page->index + 1) << PAGE_CACHE_SHIFT,
+			offset);
+	len = 1 << inode->i_blkbits;
+
+	bh = head = page_buffers(page);
+	offset = page_offset(page);
+	type = IO_OVERWRITE;
+
+	if (wbc->sync_mode == WB_SYNC_NONE && wbc->nonblocking)
+		nonblocking = 1;
+
+	do {
+		int new_ioend = 0;
+
+		if (offset >= end_offset)
+			break;
+		if (!buffer_uptodate(bh))
+			uptodate = 0;
+
+		/*
+		 * set_page_dirty dirties all buffers in a page, independent
+		 * of their state.  The dirty state however is entirely
+		 * meaningless for holes (!mapped && uptodate), so skip
+		 * buffers covering holes here.
+		 */
+		if (!buffer_mapped(bh) && buffer_uptodate(bh)) {
+			imap_valid = 0;
+			continue;
+		}
+
+		if (buffer_unwritten(bh)) {
+			if (type != IO_UNWRITTEN) {
+				type = IO_UNWRITTEN;
+				imap_valid = 0;
+			}
+		} else if (buffer_delay(bh)) {
+			if (type != IO_DELALLOC) {
+				type = IO_DELALLOC;
+				imap_valid = 0;
+			}
+		} else if (buffer_uptodate(bh)) {
+			if (type != IO_OVERWRITE) {
+				type = IO_OVERWRITE;
+				imap_valid = 0;
+			}
+		} else {
+			if (PageUptodate(page)) {
+				ASSERT(buffer_mapped(bh));
+				imap_valid = 0;
+			}
+			continue;
+		}
+
+		if (imap_valid)
+			imap_valid = xfs_imap_valid(inode, &imap, offset);
+		if (!imap_valid) {
+			/*
+			 * If we didn't have a valid mapping then we need to
+			 * put the new mapping into a separate ioend structure.
+			 * This ensures non-contiguous extents always have
+			 * separate ioends, which is particularly important
+			 * for unwritten extent conversion at I/O completion
+			 * time.
+			 */
+			new_ioend = 1;
+			err = xfs_map_blocks(inode, offset, &imap, type,
+					     nonblocking);
+			if (err)
+				goto error;
+			imap_valid = xfs_imap_valid(inode, &imap, offset);
+		}
+		if (imap_valid) {
+			lock_buffer(bh);
+			if (type != IO_OVERWRITE)
+				xfs_map_at_offset(inode, bh, &imap, offset);
+			xfs_add_to_ioend(inode, bh, offset, type, &ioend,
+					 new_ioend);
+			count++;
+		}
+
+		if (!iohead)
+			iohead = ioend;
+
+	} while (offset += len, ((bh = bh->b_this_page) != head));
+
+	if (uptodate && bh == head)
+		SetPageUptodate(page);
+
+	xfs_start_page_writeback(page, 1, count);
+
+	if (ioend && imap_valid) {
+		xfs_off_t		end_index;
+
+		end_index = imap.br_startoff + imap.br_blockcount;
+
+		/* to bytes */
+		end_index <<= inode->i_blkbits;
+
+		/* to pages */
+		end_index = (end_index - 1) >> PAGE_CACHE_SHIFT;
+
+		/* check against file size */
+		if (end_index > last_index)
+			end_index = last_index;
+
+		xfs_cluster_write(inode, page->index + 1, &imap, &ioend,
+				  wbc, end_index);
+	}
+
+	if (iohead)
+		xfs_submit_ioend(wbc, iohead);
+
+	return 0;
+
+error:
+	if (iohead)
+		xfs_cancel_ioend(iohead);
+
+	if (err == -EAGAIN)
+		goto redirty;
+
+	xfs_aops_discard_page(page);
+	ClearPageUptodate(page);
+	unlock_page(page);
+	return err;
+
+redirty:
+	redirty_page_for_writepage(wbc, page);
+	unlock_page(page);
+	return 0;
+}
+
+STATIC int
+xfs_vm_writepages(
+	struct address_space	*mapping,
+	struct writeback_control *wbc)
+{
+	xfs_iflags_clear(XFS_I(mapping->host), XFS_ITRUNCATED);
+	return generic_writepages(mapping, wbc);
+}
+
+/*
+ * Called to move a page into cleanable state - and from there
+ * to be released. The page should already be clean. We always
+ * have buffer heads in this call.
+ *
+ * Returns 1 if the page is ok to release, 0 otherwise.
+ */
+STATIC int
+xfs_vm_releasepage(
+	struct page		*page,
+	gfp_t			gfp_mask)
+{
+	int			delalloc, unwritten;
+
+	trace_xfs_releasepage(page->mapping->host, page, 0);
+
+	xfs_count_page_state(page, &delalloc, &unwritten);
+
+	if (WARN_ON(delalloc))
+		return 0;
+	if (WARN_ON(unwritten))
+		return 0;
+
+	return try_to_free_buffers(page);
+}
+
+STATIC int
+__xfs_get_blocks(
+	struct inode		*inode,
+	sector_t		iblock,
+	struct buffer_head	*bh_result,
+	int			create,
+	int			direct)
+{
+	struct xfs_inode	*ip = XFS_I(inode);
+	struct xfs_mount	*mp = ip->i_mount;
+	xfs_fileoff_t		offset_fsb, end_fsb;
+	int			error = 0;
+	int			lockmode = 0;
+	struct xfs_bmbt_irec	imap;
+	int			nimaps = 1;
+	xfs_off_t		offset;
+	ssize_t			size;
+	int			new = 0;
+
+	if (XFS_FORCED_SHUTDOWN(mp))
+		return -XFS_ERROR(EIO);
+
+	offset = (xfs_off_t)iblock << inode->i_blkbits;
+	ASSERT(bh_result->b_size >= (1 << inode->i_blkbits));
+	size = bh_result->b_size;
+
+	if (!create && direct && offset >= i_size_read(inode))
+		return 0;
+
+	if (create) {
+		lockmode = XFS_ILOCK_EXCL;
+		xfs_ilock(ip, lockmode);
+	} else {
+		lockmode = xfs_ilock_map_shared(ip);
+	}
+
+	ASSERT(offset <= mp->m_maxioffset);
+	if (offset + size > mp->m_maxioffset)
+		size = mp->m_maxioffset - offset;
+	end_fsb = XFS_B_TO_FSB(mp, (xfs_ufsize_t)offset + size);
+	offset_fsb = XFS_B_TO_FSBT(mp, offset);
+
+	error = xfs_bmapi(NULL, ip, offset_fsb, end_fsb - offset_fsb,
+			  XFS_BMAPI_ENTIRE,  NULL, 0, &imap, &nimaps, NULL);
+	if (error)
+		goto out_unlock;
+
+	if (create &&
+	    (!nimaps ||
+	     (imap.br_startblock == HOLESTARTBLOCK ||
+	      imap.br_startblock == DELAYSTARTBLOCK))) {
+		if (direct) {
+			error = xfs_iomap_write_direct(ip, offset, size,
+						       &imap, nimaps);
+		} else {
+			error = xfs_iomap_write_delay(ip, offset, size, &imap);
+		}
+		if (error)
+			goto out_unlock;
+
+		trace_xfs_get_blocks_alloc(ip, offset, size, 0, &imap);
+	} else if (nimaps) {
+		trace_xfs_get_blocks_found(ip, offset, size, 0, &imap);
+	} else {
+		trace_xfs_get_blocks_notfound(ip, offset, size);
+		goto out_unlock;
+	}
+	xfs_iunlock(ip, lockmode);
+
+	if (imap.br_startblock != HOLESTARTBLOCK &&
+	    imap.br_startblock != DELAYSTARTBLOCK) {
+		/*
+		 * For unwritten extents do not report a disk address on
+		 * the read case (treat as if we're reading into a hole).
+		 */
+		if (create || !ISUNWRITTEN(&imap))
+			xfs_map_buffer(inode, bh_result, &imap, offset);
+		if (create && ISUNWRITTEN(&imap)) {
+			if (direct)
+				bh_result->b_private = inode;
+			set_buffer_unwritten(bh_result);
+		}
+	}
+
+	/*
+	 * If this is a realtime file, data may be on a different device.
+	 * to that pointed to from the buffer_head b_bdev currently.
+	 */
+	bh_result->b_bdev = xfs_find_bdev_for_inode(inode);
+
+	/*
+	 * If we previously allocated a block out beyond eof and we are now
+	 * coming back to use it then we will need to flag it as new even if it
+	 * has a disk address.
+	 *
+	 * With sub-block writes into unwritten extents we also need to mark
+	 * the buffer as new so that the unwritten parts of the buffer gets
+	 * correctly zeroed.
+	 */
+	if (create &&
+	    ((!buffer_mapped(bh_result) && !buffer_uptodate(bh_result)) ||
+	     (offset >= i_size_read(inode)) ||
+	     (new || ISUNWRITTEN(&imap))))
+		set_buffer_new(bh_result);
+
+	if (imap.br_startblock == DELAYSTARTBLOCK) {
+		BUG_ON(direct);
+		if (create) {
+			set_buffer_uptodate(bh_result);
+			set_buffer_mapped(bh_result);
+			set_buffer_delay(bh_result);
+		}
+	}
+
+	/*
+	 * If this is O_DIRECT or the mpage code calling tell them how large
+	 * the mapping is, so that we can avoid repeated get_blocks calls.
+	 */
+	if (direct || size > (1 << inode->i_blkbits)) {
+		xfs_off_t		mapping_size;
+
+		mapping_size = imap.br_startoff + imap.br_blockcount - iblock;
+		mapping_size <<= inode->i_blkbits;
+
+		ASSERT(mapping_size > 0);
+		if (mapping_size > size)
+			mapping_size = size;
+		if (mapping_size > LONG_MAX)
+			mapping_size = LONG_MAX;
+
+		bh_result->b_size = mapping_size;
+	}
+
+	return 0;
+
+out_unlock:
+	xfs_iunlock(ip, lockmode);
+	return -error;
+}
+
+int
+xfs_get_blocks(
+	struct inode		*inode,
+	sector_t		iblock,
+	struct buffer_head	*bh_result,
+	int			create)
+{
+	return __xfs_get_blocks(inode, iblock, bh_result, create, 0);
+}
+
+STATIC int
+xfs_get_blocks_direct(
+	struct inode		*inode,
+	sector_t		iblock,
+	struct buffer_head	*bh_result,
+	int			create)
+{
+	return __xfs_get_blocks(inode, iblock, bh_result, create, 1);
+}
+
+/*
+ * Complete a direct I/O write request.
+ *
+ * If the private argument is non-NULL __xfs_get_blocks signals us that we
+ * need to issue a transaction to convert the range from unwritten to written
+ * extents.  In case this is regular synchronous I/O we just call xfs_end_io
+ * to do this and we are done.  But in case this was a successful AIO
+ * request this handler is called from interrupt context, from which we
+ * can't start transactions.  In that case offload the I/O completion to
+ * the workqueues we also use for buffered I/O completion.
+ */
+STATIC void
+xfs_end_io_direct_write(
+	struct kiocb		*iocb,
+	loff_t			offset,
+	ssize_t			size,
+	void			*private,
+	int			ret,
+	bool			is_async)
+{
+	struct xfs_ioend	*ioend = iocb->private;
+
+	/*
+	 * blockdev_direct_IO can return an error even after the I/O
+	 * completion handler was called.  Thus we need to protect
+	 * against double-freeing.
+	 */
+	iocb->private = NULL;
+
+	ioend->io_offset = offset;
+	ioend->io_size = size;
+	if (private && size > 0)
+		ioend->io_type = IO_UNWRITTEN;
+
+	if (is_async) {
+		/*
+		 * If we are converting an unwritten extent we need to delay
+		 * the AIO completion until after the unwrittent extent
+		 * conversion has completed, otherwise do it ASAP.
+		 */
+		if (ioend->io_type == IO_UNWRITTEN) {
+			ioend->io_iocb = iocb;
+			ioend->io_result = ret;
+		} else {
+			aio_complete(iocb, ret, 0);
+		}
+		xfs_finish_ioend(ioend);
+	} else {
+		xfs_finish_ioend_sync(ioend);
+	}
+}
+
+STATIC ssize_t
+xfs_vm_direct_IO(
+	int			rw,
+	struct kiocb		*iocb,
+	const struct iovec	*iov,
+	loff_t			offset,
+	unsigned long		nr_segs)
+{
+	struct inode		*inode = iocb->ki_filp->f_mapping->host;
+	struct block_device	*bdev = xfs_find_bdev_for_inode(inode);
+	ssize_t			ret;
+
+	if (rw & WRITE) {
+		iocb->private = xfs_alloc_ioend(inode, IO_DIRECT);
+
+		ret = __blockdev_direct_IO(rw, iocb, inode, bdev, iov,
+					    offset, nr_segs,
+					    xfs_get_blocks_direct,
+					    xfs_end_io_direct_write, NULL, 0);
+		if (ret != -EIOCBQUEUED && iocb->private)
+			xfs_destroy_ioend(iocb->private);
+	} else {
+		ret = __blockdev_direct_IO(rw, iocb, inode, bdev, iov,
+					    offset, nr_segs,
+					    xfs_get_blocks_direct,
+					    NULL, NULL, 0);
+	}
+
+	return ret;
+}
+
+STATIC void
+xfs_vm_write_failed(
+	struct address_space	*mapping,
+	loff_t			to)
+{
+	struct inode		*inode = mapping->host;
+
+	if (to > inode->i_size) {
+		/*
+		 * punch out the delalloc blocks we have already allocated. We
+		 * don't call xfs_setattr() to do this as we may be in the
+		 * middle of a multi-iovec write and so the vfs inode->i_size
+		 * will not match the xfs ip->i_size and so it will zero too
+		 * much. Hence we jus truncate the page cache to zero what is
+		 * necessary and punch the delalloc blocks directly.
+		 */
+		struct xfs_inode	*ip = XFS_I(inode);
+		xfs_fileoff_t		start_fsb;
+		xfs_fileoff_t		end_fsb;
+		int			error;
+
+		truncate_pagecache(inode, to, inode->i_size);
+
+		/*
+		 * Check if there are any blocks that are outside of i_size
+		 * that need to be trimmed back.
+		 */
+		start_fsb = XFS_B_TO_FSB(ip->i_mount, inode->i_size) + 1;
+		end_fsb = XFS_B_TO_FSB(ip->i_mount, to);
+		if (end_fsb <= start_fsb)
+			return;
+
+		xfs_ilock(ip, XFS_ILOCK_EXCL);
+		error = xfs_bmap_punch_delalloc_range(ip, start_fsb,
+							end_fsb - start_fsb);
+		if (error) {
+			/* something screwed, just bail */
+			if (!XFS_FORCED_SHUTDOWN(ip->i_mount)) {
+				xfs_alert(ip->i_mount,
+			"xfs_vm_write_failed: unable to clean up ino %lld",
+						ip->i_ino);
+			}
+		}
+		xfs_iunlock(ip, XFS_ILOCK_EXCL);
+	}
+}
+
+STATIC int
+xfs_vm_write_begin(
+	struct file		*file,
+	struct address_space	*mapping,
+	loff_t			pos,
+	unsigned		len,
+	unsigned		flags,
+	struct page		**pagep,
+	void			**fsdata)
+{
+	int			ret;
+
+	ret = block_write_begin(mapping, pos, len, flags | AOP_FLAG_NOFS,
+				pagep, xfs_get_blocks);
+	if (unlikely(ret))
+		xfs_vm_write_failed(mapping, pos + len);
+	return ret;
+}
+
+STATIC int
+xfs_vm_write_end(
+	struct file		*file,
+	struct address_space	*mapping,
+	loff_t			pos,
+	unsigned		len,
+	unsigned		copied,
+	struct page		*page,
+	void			*fsdata)
+{
+	int			ret;
+
+	ret = generic_write_end(file, mapping, pos, len, copied, page, fsdata);
+	if (unlikely(ret < len))
+		xfs_vm_write_failed(mapping, pos + len);
+	return ret;
+}
+
+STATIC sector_t
+xfs_vm_bmap(
+	struct address_space	*mapping,
+	sector_t		block)
+{
+	struct inode		*inode = (struct inode *)mapping->host;
+	struct xfs_inode	*ip = XFS_I(inode);
+
+	trace_xfs_vm_bmap(XFS_I(inode));
+	xfs_ilock(ip, XFS_IOLOCK_SHARED);
+	xfs_flush_pages(ip, (xfs_off_t)0, -1, 0, FI_REMAPF);
+	xfs_iunlock(ip, XFS_IOLOCK_SHARED);
+	return generic_block_bmap(mapping, block, xfs_get_blocks);
+}
+
+STATIC int
+xfs_vm_readpage(
+	struct file		*unused,
+	struct page		*page)
+{
+	return mpage_readpage(page, xfs_get_blocks);
+}
+
+STATIC int
+xfs_vm_readpages(
+	struct file		*unused,
+	struct address_space	*mapping,
+	struct list_head	*pages,
+	unsigned		nr_pages)
+{
+	return mpage_readpages(mapping, pages, nr_pages, xfs_get_blocks);
+}
+
+const struct address_space_operations xfs_address_space_operations = {
+	.readpage		= xfs_vm_readpage,
+	.readpages		= xfs_vm_readpages,
+	.writepage		= xfs_vm_writepage,
+	.writepages		= xfs_vm_writepages,
+	.releasepage		= xfs_vm_releasepage,
+	.invalidatepage		= xfs_vm_invalidatepage,
+	.write_begin		= xfs_vm_write_begin,
+	.write_end		= xfs_vm_write_end,
+	.bmap			= xfs_vm_bmap,
+	.direct_IO		= xfs_vm_direct_IO,
+	.migratepage		= buffer_migrate_page,
+	.is_partially_uptodate  = block_is_partially_uptodate,
+	.error_remove_page	= generic_error_remove_page,
+};
diff --git a/fs/xfs/linux-2.6/xfs_aops.h b/fs/xfs/linux-2.6/xfs_aops.h
new file mode 100644
index 00000000..71f721e1
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_aops.h
@@ -0,0 +1,68 @@
+/*
+ * Copyright (c) 2005-2006 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_AOPS_H__
+#define __XFS_AOPS_H__
+
+extern struct workqueue_struct *xfsdatad_workqueue;
+extern struct workqueue_struct *xfsconvertd_workqueue;
+extern mempool_t *xfs_ioend_pool;
+
+/*
+ * Types of I/O for bmap clustering and I/O completion tracking.
+ */
+enum {
+	IO_DIRECT = 0,	/* special case for direct I/O ioends */
+	IO_DELALLOC,	/* mapping covers delalloc region */
+	IO_UNWRITTEN,	/* mapping covers allocated but uninitialized data */
+	IO_OVERWRITE,	/* mapping covers already allocated extent */
+};
+
+#define XFS_IO_TYPES \
+	{ 0,			"" }, \
+	{ IO_DELALLOC,		"delalloc" }, \
+	{ IO_UNWRITTEN,		"unwritten" }, \
+	{ IO_OVERWRITE,		"overwrite" }
+
+/*
+ * xfs_ioend struct manages large extent writes for XFS.
+ * It can manage several multi-page bio's at once.
+ */
+typedef struct xfs_ioend {
+	struct xfs_ioend	*io_list;	/* next ioend in chain */
+	unsigned int		io_type;	/* delalloc / unwritten */
+	int			io_error;	/* I/O error code */
+	atomic_t		io_remaining;	/* hold count */
+	struct inode		*io_inode;	/* file being written to */
+	struct buffer_head	*io_buffer_head;/* buffer linked list head */
+	struct buffer_head	*io_buffer_tail;/* buffer linked list tail */
+	size_t			io_size;	/* size of the extent */
+	xfs_off_t		io_offset;	/* offset in the file */
+	struct work_struct	io_work;	/* xfsdatad work queue */
+	struct kiocb		*io_iocb;
+	int			io_result;
+} xfs_ioend_t;
+
+extern const struct address_space_operations xfs_address_space_operations;
+extern int xfs_get_blocks(struct inode *, sector_t, struct buffer_head *, int);
+
+extern void xfs_ioend_init(void);
+extern void xfs_ioend_wait(struct xfs_inode *);
+
+extern void xfs_count_page_state(struct page *, int *, int *);
+
+#endif /* __XFS_AOPS_H__ */
diff --git a/fs/xfs/linux-2.6/xfs_buf.c b/fs/xfs/linux-2.6/xfs_buf.c
new file mode 100644
index 00000000..5e68099d
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_buf.c
@@ -0,0 +1,1899 @@
+/*
+ * Copyright (c) 2000-2006 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include <linux/stddef.h>
+#include <linux/errno.h>
+#include <linux/gfp.h>
+#include <linux/pagemap.h>
+#include <linux/init.h>
+#include <linux/vmalloc.h>
+#include <linux/bio.h>
+#include <linux/sysctl.h>
+#include <linux/proc_fs.h>
+#include <linux/workqueue.h>
+#include <linux/percpu.h>
+#include <linux/blkdev.h>
+#include <linux/hash.h>
+#include <linux/kthread.h>
+#include <linux/migrate.h>
+#include <linux/backing-dev.h>
+#include <linux/freezer.h>
+
+#include "xfs_sb.h"
+#include "xfs_inum.h"
+#include "xfs_log.h"
+#include "xfs_ag.h"
+#include "xfs_mount.h"
+#include "xfs_trace.h"
+
+static kmem_zone_t *xfs_buf_zone;
+STATIC int xfsbufd(void *);
+STATIC void xfs_buf_delwri_queue(xfs_buf_t *, int);
+
+static struct workqueue_struct *xfslogd_workqueue;
+struct workqueue_struct *xfsdatad_workqueue;
+struct workqueue_struct *xfsconvertd_workqueue;
+
+#ifdef XFS_BUF_LOCK_TRACKING
+# define XB_SET_OWNER(bp)	((bp)->b_last_holder = current->pid)
+# define XB_CLEAR_OWNER(bp)	((bp)->b_last_holder = -1)
+# define XB_GET_OWNER(bp)	((bp)->b_last_holder)
+#else
+# define XB_SET_OWNER(bp)	do { } while (0)
+# define XB_CLEAR_OWNER(bp)	do { } while (0)
+# define XB_GET_OWNER(bp)	do { } while (0)
+#endif
+
+#define xb_to_gfp(flags) \
+	((((flags) & XBF_READ_AHEAD) ? __GFP_NORETRY : \
+	  ((flags) & XBF_DONT_BLOCK) ? GFP_NOFS : GFP_KERNEL) | __GFP_NOWARN)
+
+#define xb_to_km(flags) \
+	 (((flags) & XBF_DONT_BLOCK) ? KM_NOFS : KM_SLEEP)
+
+#define xfs_buf_allocate(flags) \
+	kmem_zone_alloc(xfs_buf_zone, xb_to_km(flags))
+#define xfs_buf_deallocate(bp) \
+	kmem_zone_free(xfs_buf_zone, (bp));
+
+static inline int
+xfs_buf_is_vmapped(
+	struct xfs_buf	*bp)
+{
+	/*
+	 * Return true if the buffer is vmapped.
+	 *
+	 * The XBF_MAPPED flag is set if the buffer should be mapped, but the
+	 * code is clever enough to know it doesn't have to map a single page,
+	 * so the check has to be both for XBF_MAPPED and bp->b_page_count > 1.
+	 */
+	return (bp->b_flags & XBF_MAPPED) && bp->b_page_count > 1;
+}
+
+static inline int
+xfs_buf_vmap_len(
+	struct xfs_buf	*bp)
+{
+	return (bp->b_page_count * PAGE_SIZE) - bp->b_offset;
+}
+
+/*
+ * xfs_buf_lru_add - add a buffer to the LRU.
+ *
+ * The LRU takes a new reference to the buffer so that it will only be freed
+ * once the shrinker takes the buffer off the LRU.
+ */
+STATIC void
+xfs_buf_lru_add(
+	struct xfs_buf	*bp)
+{
+	struct xfs_buftarg *btp = bp->b_target;
+
+	spin_lock(&btp->bt_lru_lock);
+	if (list_empty(&bp->b_lru)) {
+		atomic_inc(&bp->b_hold);
+		list_add_tail(&bp->b_lru, &btp->bt_lru);
+		btp->bt_lru_nr++;
+	}
+	spin_unlock(&btp->bt_lru_lock);
+}
+
+/*
+ * xfs_buf_lru_del - remove a buffer from the LRU
+ *
+ * The unlocked check is safe here because it only occurs when there are not
+ * b_lru_ref counts left on the inode under the pag->pag_buf_lock. it is there
+ * to optimise the shrinker removing the buffer from the LRU and calling
+ * xfs_buf_free(). i.e. it removes an unnecessary round trip on the
+ * bt_lru_lock.
+ */
+STATIC void
+xfs_buf_lru_del(
+	struct xfs_buf	*bp)
+{
+	struct xfs_buftarg *btp = bp->b_target;
+
+	if (list_empty(&bp->b_lru))
+		return;
+
+	spin_lock(&btp->bt_lru_lock);
+	if (!list_empty(&bp->b_lru)) {
+		list_del_init(&bp->b_lru);
+		btp->bt_lru_nr--;
+	}
+	spin_unlock(&btp->bt_lru_lock);
+}
+
+/*
+ * When we mark a buffer stale, we remove the buffer from the LRU and clear the
+ * b_lru_ref count so that the buffer is freed immediately when the buffer
+ * reference count falls to zero. If the buffer is already on the LRU, we need
+ * to remove the reference that LRU holds on the buffer.
+ *
+ * This prevents build-up of stale buffers on the LRU.
+ */
+void
+xfs_buf_stale(
+	struct xfs_buf	*bp)
+{
+	bp->b_flags |= XBF_STALE;
+	atomic_set(&(bp)->b_lru_ref, 0);
+	if (!list_empty(&bp->b_lru)) {
+		struct xfs_buftarg *btp = bp->b_target;
+
+		spin_lock(&btp->bt_lru_lock);
+		if (!list_empty(&bp->b_lru)) {
+			list_del_init(&bp->b_lru);
+			btp->bt_lru_nr--;
+			atomic_dec(&bp->b_hold);
+		}
+		spin_unlock(&btp->bt_lru_lock);
+	}
+	ASSERT(atomic_read(&bp->b_hold) >= 1);
+}
+
+STATIC void
+_xfs_buf_initialize(
+	xfs_buf_t		*bp,
+	xfs_buftarg_t		*target,
+	xfs_off_t		range_base,
+	size_t			range_length,
+	xfs_buf_flags_t		flags)
+{
+	/*
+	 * We don't want certain flags to appear in b_flags.
+	 */
+	flags &= ~(XBF_LOCK|XBF_MAPPED|XBF_DONT_BLOCK|XBF_READ_AHEAD);
+
+	memset(bp, 0, sizeof(xfs_buf_t));
+	atomic_set(&bp->b_hold, 1);
+	atomic_set(&bp->b_lru_ref, 1);
+	init_completion(&bp->b_iowait);
+	INIT_LIST_HEAD(&bp->b_lru);
+	INIT_LIST_HEAD(&bp->b_list);
+	RB_CLEAR_NODE(&bp->b_rbnode);
+	sema_init(&bp->b_sema, 0); /* held, no waiters */
+	XB_SET_OWNER(bp);
+	bp->b_target = target;
+	bp->b_file_offset = range_base;
+	/*
+	 * Set buffer_length and count_desired to the same value initially.
+	 * I/O routines should use count_desired, which will be the same in
+	 * most cases but may be reset (e.g. XFS recovery).
+	 */
+	bp->b_buffer_length = bp->b_count_desired = range_length;
+	bp->b_flags = flags;
+	bp->b_bn = XFS_BUF_DADDR_NULL;
+	atomic_set(&bp->b_pin_count, 0);
+	init_waitqueue_head(&bp->b_waiters);
+
+	XFS_STATS_INC(xb_create);
+
+	trace_xfs_buf_init(bp, _RET_IP_);
+}
+
+/*
+ *	Allocate a page array capable of holding a specified number
+ *	of pages, and point the page buf at it.
+ */
+STATIC int
+_xfs_buf_get_pages(
+	xfs_buf_t		*bp,
+	int			page_count,
+	xfs_buf_flags_t		flags)
+{
+	/* Make sure that we have a page list */
+	if (bp->b_pages == NULL) {
+		bp->b_offset = xfs_buf_poff(bp->b_file_offset);
+		bp->b_page_count = page_count;
+		if (page_count <= XB_PAGES) {
+			bp->b_pages = bp->b_page_array;
+		} else {
+			bp->b_pages = kmem_alloc(sizeof(struct page *) *
+					page_count, xb_to_km(flags));
+			if (bp->b_pages == NULL)
+				return -ENOMEM;
+		}
+		memset(bp->b_pages, 0, sizeof(struct page *) * page_count);
+	}
+	return 0;
+}
+
+/*
+ *	Frees b_pages if it was allocated.
+ */
+STATIC void
+_xfs_buf_free_pages(
+	xfs_buf_t	*bp)
+{
+	if (bp->b_pages != bp->b_page_array) {
+		kmem_free(bp->b_pages);
+		bp->b_pages = NULL;
+	}
+}
+
+/*
+ *	Releases the specified buffer.
+ *
+ * 	The modification state of any associated pages is left unchanged.
+ * 	The buffer most not be on any hash - use xfs_buf_rele instead for
+ * 	hashed and refcounted buffers
+ */
+void
+xfs_buf_free(
+	xfs_buf_t		*bp)
+{
+	trace_xfs_buf_free(bp, _RET_IP_);
+
+	ASSERT(list_empty(&bp->b_lru));
+
+	if (bp->b_flags & _XBF_PAGES) {
+		uint		i;
+
+		if (xfs_buf_is_vmapped(bp))
+			vm_unmap_ram(bp->b_addr - bp->b_offset,
+					bp->b_page_count);
+
+		for (i = 0; i < bp->b_page_count; i++) {
+			struct page	*page = bp->b_pages[i];
+
+			__free_page(page);
+		}
+	} else if (bp->b_flags & _XBF_KMEM)
+		kmem_free(bp->b_addr);
+	_xfs_buf_free_pages(bp);
+	xfs_buf_deallocate(bp);
+}
+
+/*
+ * Allocates all the pages for buffer in question and builds it's page list.
+ */
+STATIC int
+xfs_buf_allocate_memory(
+	xfs_buf_t		*bp,
+	uint			flags)
+{
+	size_t			size = bp->b_count_desired;
+	size_t			nbytes, offset;
+	gfp_t			gfp_mask = xb_to_gfp(flags);
+	unsigned short		page_count, i;
+	xfs_off_t		end;
+	int			error;
+
+	/*
+	 * for buffers that are contained within a single page, just allocate
+	 * the memory from the heap - there's no need for the complexity of
+	 * page arrays to keep allocation down to order 0.
+	 */
+	if (bp->b_buffer_length < PAGE_SIZE) {
+		bp->b_addr = kmem_alloc(bp->b_buffer_length, xb_to_km(flags));
+		if (!bp->b_addr) {
+			/* low memory - use alloc_page loop instead */
+			goto use_alloc_page;
+		}
+
+		if (((unsigned long)(bp->b_addr + bp->b_buffer_length - 1) &
+								PAGE_MASK) !=
+		    ((unsigned long)bp->b_addr & PAGE_MASK)) {
+			/* b_addr spans two pages - use alloc_page instead */
+			kmem_free(bp->b_addr);
+			bp->b_addr = NULL;
+			goto use_alloc_page;
+		}
+		bp->b_offset = offset_in_page(bp->b_addr);
+		bp->b_pages = bp->b_page_array;
+		bp->b_pages[0] = virt_to_page(bp->b_addr);
+		bp->b_page_count = 1;
+		bp->b_flags |= XBF_MAPPED | _XBF_KMEM;
+		return 0;
+	}
+
+use_alloc_page:
+	end = bp->b_file_offset + bp->b_buffer_length;
+	page_count = xfs_buf_btoc(end) - xfs_buf_btoct(bp->b_file_offset);
+	error = _xfs_buf_get_pages(bp, page_count, flags);
+	if (unlikely(error))
+		return error;
+
+	offset = bp->b_offset;
+	bp->b_flags |= _XBF_PAGES;
+
+	for (i = 0; i < bp->b_page_count; i++) {
+		struct page	*page;
+		uint		retries = 0;
+retry:
+		page = alloc_page(gfp_mask);
+		if (unlikely(page == NULL)) {
+			if (flags & XBF_READ_AHEAD) {
+				bp->b_page_count = i;
+				error = ENOMEM;
+				goto out_free_pages;
+			}
+
+			/*
+			 * This could deadlock.
+			 *
+			 * But until all the XFS lowlevel code is revamped to
+			 * handle buffer allocation failures we can't do much.
+			 */
+			if (!(++retries % 100))
+				xfs_err(NULL,
+		"possible memory allocation deadlock in %s (mode:0x%x)",
+					__func__, gfp_mask);
+
+			XFS_STATS_INC(xb_page_retries);
+			congestion_wait(BLK_RW_ASYNC, HZ/50);
+			goto retry;
+		}
+
+		XFS_STATS_INC(xb_page_found);
+
+		nbytes = min_t(size_t, size, PAGE_SIZE - offset);
+		size -= nbytes;
+		bp->b_pages[i] = page;
+		offset = 0;
+	}
+	return 0;
+
+out_free_pages:
+	for (i = 0; i < bp->b_page_count; i++)
+		__free_page(bp->b_pages[i]);
+	return error;
+}
+
+/*
+ *	Map buffer into kernel address-space if necessary.
+ */
+STATIC int
+_xfs_buf_map_pages(
+	xfs_buf_t		*bp,
+	uint			flags)
+{
+	ASSERT(bp->b_flags & _XBF_PAGES);
+	if (bp->b_page_count == 1) {
+		/* A single page buffer is always mappable */
+		bp->b_addr = page_address(bp->b_pages[0]) + bp->b_offset;
+		bp->b_flags |= XBF_MAPPED;
+	} else if (flags & XBF_MAPPED) {
+		int retried = 0;
+
+		do {
+			bp->b_addr = vm_map_ram(bp->b_pages, bp->b_page_count,
+						-1, PAGE_KERNEL);
+			if (bp->b_addr)
+				break;
+			vm_unmap_aliases();
+		} while (retried++ <= 1);
+
+		if (!bp->b_addr)
+			return -ENOMEM;
+		bp->b_addr += bp->b_offset;
+		bp->b_flags |= XBF_MAPPED;
+	}
+
+	return 0;
+}
+
+/*
+ *	Finding and Reading Buffers
+ */
+
+/*
+ *	Look up, and creates if absent, a lockable buffer for
+ *	a given range of an inode.  The buffer is returned
+ *	locked.	 If other overlapping buffers exist, they are
+ *	released before the new buffer is created and locked,
+ *	which may imply that this call will block until those buffers
+ *	are unlocked.  No I/O is implied by this call.
+ */
+xfs_buf_t *
+_xfs_buf_find(
+	xfs_buftarg_t		*btp,	/* block device target		*/
+	xfs_off_t		ioff,	/* starting offset of range	*/
+	size_t			isize,	/* length of range		*/
+	xfs_buf_flags_t		flags,
+	xfs_buf_t		*new_bp)
+{
+	xfs_off_t		range_base;
+	size_t			range_length;
+	struct xfs_perag	*pag;
+	struct rb_node		**rbp;
+	struct rb_node		*parent;
+	xfs_buf_t		*bp;
+
+	range_base = (ioff << BBSHIFT);
+	range_length = (isize << BBSHIFT);
+
+	/* Check for IOs smaller than the sector size / not sector aligned */
+	ASSERT(!(range_length < (1 << btp->bt_sshift)));
+	ASSERT(!(range_base & (xfs_off_t)btp->bt_smask));
+
+	/* get tree root */
+	pag = xfs_perag_get(btp->bt_mount,
+				xfs_daddr_to_agno(btp->bt_mount, ioff));
+
+	/* walk tree */
+	spin_lock(&pag->pag_buf_lock);
+	rbp = &pag->pag_buf_tree.rb_node;
+	parent = NULL;
+	bp = NULL;
+	while (*rbp) {
+		parent = *rbp;
+		bp = rb_entry(parent, struct xfs_buf, b_rbnode);
+
+		if (range_base < bp->b_file_offset)
+			rbp = &(*rbp)->rb_left;
+		else if (range_base > bp->b_file_offset)
+			rbp = &(*rbp)->rb_right;
+		else {
+			/*
+			 * found a block offset match. If the range doesn't
+			 * match, the only way this is allowed is if the buffer
+			 * in the cache is stale and the transaction that made
+			 * it stale has not yet committed. i.e. we are
+			 * reallocating a busy extent. Skip this buffer and
+			 * continue searching to the right for an exact match.
+			 */
+			if (bp->b_buffer_length != range_length) {
+				ASSERT(bp->b_flags & XBF_STALE);
+				rbp = &(*rbp)->rb_right;
+				continue;
+			}
+			atomic_inc(&bp->b_hold);
+			goto found;
+		}
+	}
+
+	/* No match found */
+	if (new_bp) {
+		_xfs_buf_initialize(new_bp, btp, range_base,
+				range_length, flags);
+		rb_link_node(&new_bp->b_rbnode, parent, rbp);
+		rb_insert_color(&new_bp->b_rbnode, &pag->pag_buf_tree);
+		/* the buffer keeps the perag reference until it is freed */
+		new_bp->b_pag = pag;
+		spin_unlock(&pag->pag_buf_lock);
+	} else {
+		XFS_STATS_INC(xb_miss_locked);
+		spin_unlock(&pag->pag_buf_lock);
+		xfs_perag_put(pag);
+	}
+	return new_bp;
+
+found:
+	spin_unlock(&pag->pag_buf_lock);
+	xfs_perag_put(pag);
+
+	if (xfs_buf_cond_lock(bp)) {
+		/* failed, so wait for the lock if requested. */
+		if (!(flags & XBF_TRYLOCK)) {
+			xfs_buf_lock(bp);
+			XFS_STATS_INC(xb_get_locked_waited);
+		} else {
+			xfs_buf_rele(bp);
+			XFS_STATS_INC(xb_busy_locked);
+			return NULL;
+		}
+	}
+
+	/*
+	 * if the buffer is stale, clear all the external state associated with
+	 * it. We need to keep flags such as how we allocated the buffer memory
+	 * intact here.
+	 */
+	if (bp->b_flags & XBF_STALE) {
+		ASSERT((bp->b_flags & _XBF_DELWRI_Q) == 0);
+		bp->b_flags &= XBF_MAPPED | _XBF_KMEM | _XBF_PAGES;
+	}
+
+	trace_xfs_buf_find(bp, flags, _RET_IP_);
+	XFS_STATS_INC(xb_get_locked);
+	return bp;
+}
+
+/*
+ *	Assembles a buffer covering the specified range.
+ *	Storage in memory for all portions of the buffer will be allocated,
+ *	although backing storage may not be.
+ */
+xfs_buf_t *
+xfs_buf_get(
+	xfs_buftarg_t		*target,/* target for buffer		*/
+	xfs_off_t		ioff,	/* starting offset of range	*/
+	size_t			isize,	/* length of range		*/
+	xfs_buf_flags_t		flags)
+{
+	xfs_buf_t		*bp, *new_bp;
+	int			error = 0;
+
+	new_bp = xfs_buf_allocate(flags);
+	if (unlikely(!new_bp))
+		return NULL;
+
+	bp = _xfs_buf_find(target, ioff, isize, flags, new_bp);
+	if (bp == new_bp) {
+		error = xfs_buf_allocate_memory(bp, flags);
+		if (error)
+			goto no_buffer;
+	} else {
+		xfs_buf_deallocate(new_bp);
+		if (unlikely(bp == NULL))
+			return NULL;
+	}
+
+	if (!(bp->b_flags & XBF_MAPPED)) {
+		error = _xfs_buf_map_pages(bp, flags);
+		if (unlikely(error)) {
+			xfs_warn(target->bt_mount,
+				"%s: failed to map pages\n", __func__);
+			goto no_buffer;
+		}
+	}
+
+	XFS_STATS_INC(xb_get);
+
+	/*
+	 * Always fill in the block number now, the mapped cases can do
+	 * their own overlay of this later.
+	 */
+	bp->b_bn = ioff;
+	bp->b_count_desired = bp->b_buffer_length;
+
+	trace_xfs_buf_get(bp, flags, _RET_IP_);
+	return bp;
+
+ no_buffer:
+	if (flags & (XBF_LOCK | XBF_TRYLOCK))
+		xfs_buf_unlock(bp);
+	xfs_buf_rele(bp);
+	return NULL;
+}
+
+STATIC int
+_xfs_buf_read(
+	xfs_buf_t		*bp,
+	xfs_buf_flags_t		flags)
+{
+	int			status;
+
+	ASSERT(!(flags & (XBF_DELWRI|XBF_WRITE)));
+	ASSERT(bp->b_bn != XFS_BUF_DADDR_NULL);
+
+	bp->b_flags &= ~(XBF_WRITE | XBF_ASYNC | XBF_DELWRI | \
+			XBF_READ_AHEAD | _XBF_RUN_QUEUES);
+	bp->b_flags |= flags & (XBF_READ | XBF_ASYNC | \
+			XBF_READ_AHEAD | _XBF_RUN_QUEUES);
+
+	status = xfs_buf_iorequest(bp);
+	if (status || XFS_BUF_ISERROR(bp) || (flags & XBF_ASYNC))
+		return status;
+	return xfs_buf_iowait(bp);
+}
+
+xfs_buf_t *
+xfs_buf_read(
+	xfs_buftarg_t		*target,
+	xfs_off_t		ioff,
+	size_t			isize,
+	xfs_buf_flags_t		flags)
+{
+	xfs_buf_t		*bp;
+
+	flags |= XBF_READ;
+
+	bp = xfs_buf_get(target, ioff, isize, flags);
+	if (bp) {
+		trace_xfs_buf_read(bp, flags, _RET_IP_);
+
+		if (!XFS_BUF_ISDONE(bp)) {
+			XFS_STATS_INC(xb_get_read);
+			_xfs_buf_read(bp, flags);
+		} else if (flags & XBF_ASYNC) {
+			/*
+			 * Read ahead call which is already satisfied,
+			 * drop the buffer
+			 */
+			goto no_buffer;
+		} else {
+			/* We do not want read in the flags */
+			bp->b_flags &= ~XBF_READ;
+		}
+	}
+
+	return bp;
+
+ no_buffer:
+	if (flags & (XBF_LOCK | XBF_TRYLOCK))
+		xfs_buf_unlock(bp);
+	xfs_buf_rele(bp);
+	return NULL;
+}
+
+/*
+ *	If we are not low on memory then do the readahead in a deadlock
+ *	safe manner.
+ */
+void
+xfs_buf_readahead(
+	xfs_buftarg_t		*target,
+	xfs_off_t		ioff,
+	size_t			isize)
+{
+	if (bdi_read_congested(target->bt_bdi))
+		return;
+
+	xfs_buf_read(target, ioff, isize,
+		     XBF_TRYLOCK|XBF_ASYNC|XBF_READ_AHEAD|XBF_DONT_BLOCK);
+}
+
+/*
+ * Read an uncached buffer from disk. Allocates and returns a locked
+ * buffer containing the disk contents or nothing.
+ */
+struct xfs_buf *
+xfs_buf_read_uncached(
+	struct xfs_mount	*mp,
+	struct xfs_buftarg	*target,
+	xfs_daddr_t		daddr,
+	size_t			length,
+	int			flags)
+{
+	xfs_buf_t		*bp;
+	int			error;
+
+	bp = xfs_buf_get_uncached(target, length, flags);
+	if (!bp)
+		return NULL;
+
+	/* set up the buffer for a read IO */
+	xfs_buf_lock(bp);
+	XFS_BUF_SET_ADDR(bp, daddr);
+	XFS_BUF_READ(bp);
+	XFS_BUF_BUSY(bp);
+
+	xfsbdstrat(mp, bp);
+	error = xfs_buf_iowait(bp);
+	if (error || bp->b_error) {
+		xfs_buf_relse(bp);
+		return NULL;
+	}
+	return bp;
+}
+
+xfs_buf_t *
+xfs_buf_get_empty(
+	size_t			len,
+	xfs_buftarg_t		*target)
+{
+	xfs_buf_t		*bp;
+
+	bp = xfs_buf_allocate(0);
+	if (bp)
+		_xfs_buf_initialize(bp, target, 0, len, 0);
+	return bp;
+}
+
+/*
+ * Return a buffer allocated as an empty buffer and associated to external
+ * memory via xfs_buf_associate_memory() back to it's empty state.
+ */
+void
+xfs_buf_set_empty(
+	struct xfs_buf		*bp,
+	size_t			len)
+{
+	if (bp->b_pages)
+		_xfs_buf_free_pages(bp);
+
+	bp->b_pages = NULL;
+	bp->b_page_count = 0;
+	bp->b_addr = NULL;
+	bp->b_file_offset = 0;
+	bp->b_buffer_length = bp->b_count_desired = len;
+	bp->b_bn = XFS_BUF_DADDR_NULL;
+	bp->b_flags &= ~XBF_MAPPED;
+}
+
+static inline struct page *
+mem_to_page(
+	void			*addr)
+{
+	if ((!is_vmalloc_addr(addr))) {
+		return virt_to_page(addr);
+	} else {
+		return vmalloc_to_page(addr);
+	}
+}
+
+int
+xfs_buf_associate_memory(
+	xfs_buf_t		*bp,
+	void			*mem,
+	size_t			len)
+{
+	int			rval;
+	int			i = 0;
+	unsigned long		pageaddr;
+	unsigned long		offset;
+	size_t			buflen;
+	int			page_count;
+
+	pageaddr = (unsigned long)mem & PAGE_MASK;
+	offset = (unsigned long)mem - pageaddr;
+	buflen = PAGE_ALIGN(len + offset);
+	page_count = buflen >> PAGE_SHIFT;
+
+	/* Free any previous set of page pointers */
+	if (bp->b_pages)
+		_xfs_buf_free_pages(bp);
+
+	bp->b_pages = NULL;
+	bp->b_addr = mem;
+
+	rval = _xfs_buf_get_pages(bp, page_count, XBF_DONT_BLOCK);
+	if (rval)
+		return rval;
+
+	bp->b_offset = offset;
+
+	for (i = 0; i < bp->b_page_count; i++) {
+		bp->b_pages[i] = mem_to_page((void *)pageaddr);
+		pageaddr += PAGE_SIZE;
+	}
+
+	bp->b_count_desired = len;
+	bp->b_buffer_length = buflen;
+	bp->b_flags |= XBF_MAPPED;
+
+	return 0;
+}
+
+xfs_buf_t *
+xfs_buf_get_uncached(
+	struct xfs_buftarg	*target,
+	size_t			len,
+	int			flags)
+{
+	unsigned long		page_count = PAGE_ALIGN(len) >> PAGE_SHIFT;
+	int			error, i;
+	xfs_buf_t		*bp;
+
+	bp = xfs_buf_allocate(0);
+	if (unlikely(bp == NULL))
+		goto fail;
+	_xfs_buf_initialize(bp, target, 0, len, 0);
+
+	error = _xfs_buf_get_pages(bp, page_count, 0);
+	if (error)
+		goto fail_free_buf;
+
+	for (i = 0; i < page_count; i++) {
+		bp->b_pages[i] = alloc_page(xb_to_gfp(flags));
+		if (!bp->b_pages[i])
+			goto fail_free_mem;
+	}
+	bp->b_flags |= _XBF_PAGES;
+
+	error = _xfs_buf_map_pages(bp, XBF_MAPPED);
+	if (unlikely(error)) {
+		xfs_warn(target->bt_mount,
+			"%s: failed to map pages\n", __func__);
+		goto fail_free_mem;
+	}
+
+	xfs_buf_unlock(bp);
+
+	trace_xfs_buf_get_uncached(bp, _RET_IP_);
+	return bp;
+
+ fail_free_mem:
+	while (--i >= 0)
+		__free_page(bp->b_pages[i]);
+	_xfs_buf_free_pages(bp);
+ fail_free_buf:
+	xfs_buf_deallocate(bp);
+ fail:
+	return NULL;
+}
+
+/*
+ *	Increment reference count on buffer, to hold the buffer concurrently
+ *	with another thread which may release (free) the buffer asynchronously.
+ *	Must hold the buffer already to call this function.
+ */
+void
+xfs_buf_hold(
+	xfs_buf_t		*bp)
+{
+	trace_xfs_buf_hold(bp, _RET_IP_);
+	atomic_inc(&bp->b_hold);
+}
+
+/*
+ *	Releases a hold on the specified buffer.  If the
+ *	the hold count is 1, calls xfs_buf_free.
+ */
+void
+xfs_buf_rele(
+	xfs_buf_t		*bp)
+{
+	struct xfs_perag	*pag = bp->b_pag;
+
+	trace_xfs_buf_rele(bp, _RET_IP_);
+
+	if (!pag) {
+		ASSERT(list_empty(&bp->b_lru));
+		ASSERT(RB_EMPTY_NODE(&bp->b_rbnode));
+		if (atomic_dec_and_test(&bp->b_hold))
+			xfs_buf_free(bp);
+		return;
+	}
+
+	ASSERT(!RB_EMPTY_NODE(&bp->b_rbnode));
+
+	ASSERT(atomic_read(&bp->b_hold) > 0);
+	if (atomic_dec_and_lock(&bp->b_hold, &pag->pag_buf_lock)) {
+		if (!(bp->b_flags & XBF_STALE) &&
+			   atomic_read(&bp->b_lru_ref)) {
+			xfs_buf_lru_add(bp);
+			spin_unlock(&pag->pag_buf_lock);
+		} else {
+			xfs_buf_lru_del(bp);
+			ASSERT(!(bp->b_flags & (XBF_DELWRI|_XBF_DELWRI_Q)));
+			rb_erase(&bp->b_rbnode, &pag->pag_buf_tree);
+			spin_unlock(&pag->pag_buf_lock);
+			xfs_perag_put(pag);
+			xfs_buf_free(bp);
+		}
+	}
+}
+
+
+/*
+ *	Lock a buffer object, if it is not already locked.
+ *
+ *	If we come across a stale, pinned, locked buffer, we know that we are
+ *	being asked to lock a buffer that has been reallocated. Because it is
+ *	pinned, we know that the log has not been pushed to disk and hence it
+ *	will still be locked.  Rather than continuing to have trylock attempts
+ *	fail until someone else pushes the log, push it ourselves before
+ *	returning.  This means that the xfsaild will not get stuck trying
+ *	to push on stale inode buffers.
+ */
+int
+xfs_buf_cond_lock(
+	xfs_buf_t		*bp)
+{
+	int			locked;
+
+	locked = down_trylock(&bp->b_sema) == 0;
+	if (locked)
+		XB_SET_OWNER(bp);
+	else if (atomic_read(&bp->b_pin_count) && (bp->b_flags & XBF_STALE))
+		xfs_log_force(bp->b_target->bt_mount, 0);
+
+	trace_xfs_buf_cond_lock(bp, _RET_IP_);
+	return locked ? 0 : -EBUSY;
+}
+
+int
+xfs_buf_lock_value(
+	xfs_buf_t		*bp)
+{
+	return bp->b_sema.count;
+}
+
+/*
+ *	Lock a buffer object.
+ *
+ *	If we come across a stale, pinned, locked buffer, we know that we
+ *	are being asked to lock a buffer that has been reallocated. Because
+ *	it is pinned, we know that the log has not been pushed to disk and
+ *	hence it will still be locked. Rather than sleeping until someone
+ *	else pushes the log, push it ourselves before trying to get the lock.
+ */
+void
+xfs_buf_lock(
+	xfs_buf_t		*bp)
+{
+	trace_xfs_buf_lock(bp, _RET_IP_);
+
+	if (atomic_read(&bp->b_pin_count) && (bp->b_flags & XBF_STALE))
+		xfs_log_force(bp->b_target->bt_mount, 0);
+	down(&bp->b_sema);
+	XB_SET_OWNER(bp);
+
+	trace_xfs_buf_lock_done(bp, _RET_IP_);
+}
+
+/*
+ *	Releases the lock on the buffer object.
+ *	If the buffer is marked delwri but is not queued, do so before we
+ *	unlock the buffer as we need to set flags correctly.  We also need to
+ *	take a reference for the delwri queue because the unlocker is going to
+ *	drop their's and they don't know we just queued it.
+ */
+void
+xfs_buf_unlock(
+	xfs_buf_t		*bp)
+{
+	if ((bp->b_flags & (XBF_DELWRI|_XBF_DELWRI_Q)) == XBF_DELWRI) {
+		atomic_inc(&bp->b_hold);
+		bp->b_flags |= XBF_ASYNC;
+		xfs_buf_delwri_queue(bp, 0);
+	}
+
+	XB_CLEAR_OWNER(bp);
+	up(&bp->b_sema);
+
+	trace_xfs_buf_unlock(bp, _RET_IP_);
+}
+
+STATIC void
+xfs_buf_wait_unpin(
+	xfs_buf_t		*bp)
+{
+	DECLARE_WAITQUEUE	(wait, current);
+
+	if (atomic_read(&bp->b_pin_count) == 0)
+		return;
+
+	add_wait_queue(&bp->b_waiters, &wait);
+	for (;;) {
+		set_current_state(TASK_UNINTERRUPTIBLE);
+		if (atomic_read(&bp->b_pin_count) == 0)
+			break;
+		io_schedule();
+	}
+	remove_wait_queue(&bp->b_waiters, &wait);
+	set_current_state(TASK_RUNNING);
+}
+
+/*
+ *	Buffer Utility Routines
+ */
+
+STATIC void
+xfs_buf_iodone_work(
+	struct work_struct	*work)
+{
+	xfs_buf_t		*bp =
+		container_of(work, xfs_buf_t, b_iodone_work);
+
+	if (bp->b_iodone)
+		(*(bp->b_iodone))(bp);
+	else if (bp->b_flags & XBF_ASYNC)
+		xfs_buf_relse(bp);
+}
+
+void
+xfs_buf_ioend(
+	xfs_buf_t		*bp,
+	int			schedule)
+{
+	trace_xfs_buf_iodone(bp, _RET_IP_);
+
+	bp->b_flags &= ~(XBF_READ | XBF_WRITE | XBF_READ_AHEAD);
+	if (bp->b_error == 0)
+		bp->b_flags |= XBF_DONE;
+
+	if ((bp->b_iodone) || (bp->b_flags & XBF_ASYNC)) {
+		if (schedule) {
+			INIT_WORK(&bp->b_iodone_work, xfs_buf_iodone_work);
+			queue_work(xfslogd_workqueue, &bp->b_iodone_work);
+		} else {
+			xfs_buf_iodone_work(&bp->b_iodone_work);
+		}
+	} else {
+		complete(&bp->b_iowait);
+	}
+}
+
+void
+xfs_buf_ioerror(
+	xfs_buf_t		*bp,
+	int			error)
+{
+	ASSERT(error >= 0 && error <= 0xffff);
+	bp->b_error = (unsigned short)error;
+	trace_xfs_buf_ioerror(bp, error, _RET_IP_);
+}
+
+int
+xfs_bwrite(
+	struct xfs_mount	*mp,
+	struct xfs_buf		*bp)
+{
+	int			error;
+
+	bp->b_flags |= XBF_WRITE;
+	bp->b_flags &= ~(XBF_ASYNC | XBF_READ);
+
+	xfs_buf_delwri_dequeue(bp);
+	xfs_bdstrat_cb(bp);
+
+	error = xfs_buf_iowait(bp);
+	if (error)
+		xfs_force_shutdown(mp, SHUTDOWN_META_IO_ERROR);
+	xfs_buf_relse(bp);
+	return error;
+}
+
+void
+xfs_bdwrite(
+	void			*mp,
+	struct xfs_buf		*bp)
+{
+	trace_xfs_buf_bdwrite(bp, _RET_IP_);
+
+	bp->b_flags &= ~XBF_READ;
+	bp->b_flags |= (XBF_DELWRI | XBF_ASYNC);
+
+	xfs_buf_delwri_queue(bp, 1);
+}
+
+/*
+ * Called when we want to stop a buffer from getting written or read.
+ * We attach the EIO error, muck with its flags, and call xfs_buf_ioend
+ * so that the proper iodone callbacks get called.
+ */
+STATIC int
+xfs_bioerror(
+	xfs_buf_t *bp)
+{
+#ifdef XFSERRORDEBUG
+	ASSERT(XFS_BUF_ISREAD(bp) || bp->b_iodone);
+#endif
+
+	/*
+	 * No need to wait until the buffer is unpinned, we aren't flushing it.
+	 */
+	XFS_BUF_ERROR(bp, EIO);
+
+	/*
+	 * We're calling xfs_buf_ioend, so delete XBF_DONE flag.
+	 */
+	XFS_BUF_UNREAD(bp);
+	XFS_BUF_UNDELAYWRITE(bp);
+	XFS_BUF_UNDONE(bp);
+	XFS_BUF_STALE(bp);
+
+	xfs_buf_ioend(bp, 0);
+
+	return EIO;
+}
+
+/*
+ * Same as xfs_bioerror, except that we are releasing the buffer
+ * here ourselves, and avoiding the xfs_buf_ioend call.
+ * This is meant for userdata errors; metadata bufs come with
+ * iodone functions attached, so that we can track down errors.
+ */
+STATIC int
+xfs_bioerror_relse(
+	struct xfs_buf	*bp)
+{
+	int64_t		fl = XFS_BUF_BFLAGS(bp);
+	/*
+	 * No need to wait until the buffer is unpinned.
+	 * We aren't flushing it.
+	 *
+	 * chunkhold expects B_DONE to be set, whether
+	 * we actually finish the I/O or not. We don't want to
+	 * change that interface.
+	 */
+	XFS_BUF_UNREAD(bp);
+	XFS_BUF_UNDELAYWRITE(bp);
+	XFS_BUF_DONE(bp);
+	XFS_BUF_STALE(bp);
+	XFS_BUF_CLR_IODONE_FUNC(bp);
+	if (!(fl & XBF_ASYNC)) {
+		/*
+		 * Mark b_error and B_ERROR _both_.
+		 * Lot's of chunkcache code assumes that.
+		 * There's no reason to mark error for
+		 * ASYNC buffers.
+		 */
+		XFS_BUF_ERROR(bp, EIO);
+		XFS_BUF_FINISH_IOWAIT(bp);
+	} else {
+		xfs_buf_relse(bp);
+	}
+
+	return EIO;
+}
+
+
+/*
+ * All xfs metadata buffers except log state machine buffers
+ * get this attached as their b_bdstrat callback function.
+ * This is so that we can catch a buffer
+ * after prematurely unpinning it to forcibly shutdown the filesystem.
+ */
+int
+xfs_bdstrat_cb(
+	struct xfs_buf	*bp)
+{
+	if (XFS_FORCED_SHUTDOWN(bp->b_target->bt_mount)) {
+		trace_xfs_bdstrat_shut(bp, _RET_IP_);
+		/*
+		 * Metadata write that didn't get logged but
+		 * written delayed anyway. These aren't associated
+		 * with a transaction, and can be ignored.
+		 */
+		if (!bp->b_iodone && !XFS_BUF_ISREAD(bp))
+			return xfs_bioerror_relse(bp);
+		else
+			return xfs_bioerror(bp);
+	}
+
+	xfs_buf_iorequest(bp);
+	return 0;
+}
+
+/*
+ * Wrapper around bdstrat so that we can stop data from going to disk in case
+ * we are shutting down the filesystem.  Typically user data goes thru this
+ * path; one of the exceptions is the superblock.
+ */
+void
+xfsbdstrat(
+	struct xfs_mount	*mp,
+	struct xfs_buf		*bp)
+{
+	if (XFS_FORCED_SHUTDOWN(mp)) {
+		trace_xfs_bdstrat_shut(bp, _RET_IP_);
+		xfs_bioerror_relse(bp);
+		return;
+	}
+
+	xfs_buf_iorequest(bp);
+}
+
+STATIC void
+_xfs_buf_ioend(
+	xfs_buf_t		*bp,
+	int			schedule)
+{
+	if (atomic_dec_and_test(&bp->b_io_remaining) == 1)
+		xfs_buf_ioend(bp, schedule);
+}
+
+STATIC void
+xfs_buf_bio_end_io(
+	struct bio		*bio,
+	int			error)
+{
+	xfs_buf_t		*bp = (xfs_buf_t *)bio->bi_private;
+
+	xfs_buf_ioerror(bp, -error);
+
+	if (!error && xfs_buf_is_vmapped(bp) && (bp->b_flags & XBF_READ))
+		invalidate_kernel_vmap_range(bp->b_addr, xfs_buf_vmap_len(bp));
+
+	_xfs_buf_ioend(bp, 1);
+	bio_put(bio);
+}
+
+STATIC void
+_xfs_buf_ioapply(
+	xfs_buf_t		*bp)
+{
+	int			rw, map_i, total_nr_pages, nr_pages;
+	struct bio		*bio;
+	int			offset = bp->b_offset;
+	int			size = bp->b_count_desired;
+	sector_t		sector = bp->b_bn;
+
+	total_nr_pages = bp->b_page_count;
+	map_i = 0;
+
+	if (bp->b_flags & XBF_ORDERED) {
+		ASSERT(!(bp->b_flags & XBF_READ));
+		rw = WRITE_FLUSH_FUA;
+	} else if (bp->b_flags & XBF_LOG_BUFFER) {
+		ASSERT(!(bp->b_flags & XBF_READ_AHEAD));
+		bp->b_flags &= ~_XBF_RUN_QUEUES;
+		rw = (bp->b_flags & XBF_WRITE) ? WRITE_SYNC : READ_SYNC;
+	} else if (bp->b_flags & _XBF_RUN_QUEUES) {
+		ASSERT(!(bp->b_flags & XBF_READ_AHEAD));
+		bp->b_flags &= ~_XBF_RUN_QUEUES;
+		rw = (bp->b_flags & XBF_WRITE) ? WRITE_META : READ_META;
+	} else {
+		rw = (bp->b_flags & XBF_WRITE) ? WRITE :
+		     (bp->b_flags & XBF_READ_AHEAD) ? READA : READ;
+	}
+
+
+next_chunk:
+	atomic_inc(&bp->b_io_remaining);
+	nr_pages = BIO_MAX_SECTORS >> (PAGE_SHIFT - BBSHIFT);
+	if (nr_pages > total_nr_pages)
+		nr_pages = total_nr_pages;
+
+	bio = bio_alloc(GFP_NOIO, nr_pages);
+	bio->bi_bdev = bp->b_target->bt_bdev;
+	bio->bi_sector = sector;
+	bio->bi_end_io = xfs_buf_bio_end_io;
+	bio->bi_private = bp;
+
+
+	for (; size && nr_pages; nr_pages--, map_i++) {
+		int	rbytes, nbytes = PAGE_SIZE - offset;
+
+		if (nbytes > size)
+			nbytes = size;
+
+		rbytes = bio_add_page(bio, bp->b_pages[map_i], nbytes, offset);
+		if (rbytes < nbytes)
+			break;
+
+		offset = 0;
+		sector += nbytes >> BBSHIFT;
+		size -= nbytes;
+		total_nr_pages--;
+	}
+
+	if (likely(bio->bi_size)) {
+		if (xfs_buf_is_vmapped(bp)) {
+			flush_kernel_vmap_range(bp->b_addr,
+						xfs_buf_vmap_len(bp));
+		}
+		submit_bio(rw, bio);
+		if (size)
+			goto next_chunk;
+	} else {
+		xfs_buf_ioerror(bp, EIO);
+		bio_put(bio);
+	}
+}
+
+int
+xfs_buf_iorequest(
+	xfs_buf_t		*bp)
+{
+	trace_xfs_buf_iorequest(bp, _RET_IP_);
+
+	if (bp->b_flags & XBF_DELWRI) {
+		xfs_buf_delwri_queue(bp, 1);
+		return 0;
+	}
+
+	if (bp->b_flags & XBF_WRITE) {
+		xfs_buf_wait_unpin(bp);
+	}
+
+	xfs_buf_hold(bp);
+
+	/* Set the count to 1 initially, this will stop an I/O
+	 * completion callout which happens before we have started
+	 * all the I/O from calling xfs_buf_ioend too early.
+	 */
+	atomic_set(&bp->b_io_remaining, 1);
+	_xfs_buf_ioapply(bp);
+	_xfs_buf_ioend(bp, 0);
+
+	xfs_buf_rele(bp);
+	return 0;
+}
+
+/*
+ *	Waits for I/O to complete on the buffer supplied.
+ *	It returns immediately if no I/O is pending.
+ *	It returns the I/O error code, if any, or 0 if there was no error.
+ */
+int
+xfs_buf_iowait(
+	xfs_buf_t		*bp)
+{
+	trace_xfs_buf_iowait(bp, _RET_IP_);
+
+	wait_for_completion(&bp->b_iowait);
+
+	trace_xfs_buf_iowait_done(bp, _RET_IP_);
+	return bp->b_error;
+}
+
+xfs_caddr_t
+xfs_buf_offset(
+	xfs_buf_t		*bp,
+	size_t			offset)
+{
+	struct page		*page;
+
+	if (bp->b_flags & XBF_MAPPED)
+		return XFS_BUF_PTR(bp) + offset;
+
+	offset += bp->b_offset;
+	page = bp->b_pages[offset >> PAGE_SHIFT];
+	return (xfs_caddr_t)page_address(page) + (offset & (PAGE_SIZE-1));
+}
+
+/*
+ *	Move data into or out of a buffer.
+ */
+void
+xfs_buf_iomove(
+	xfs_buf_t		*bp,	/* buffer to process		*/
+	size_t			boff,	/* starting buffer offset	*/
+	size_t			bsize,	/* length to copy		*/
+	void			*data,	/* data address			*/
+	xfs_buf_rw_t		mode)	/* read/write/zero flag		*/
+{
+	size_t			bend, cpoff, csize;
+	struct page		*page;
+
+	bend = boff + bsize;
+	while (boff < bend) {
+		page = bp->b_pages[xfs_buf_btoct(boff + bp->b_offset)];
+		cpoff = xfs_buf_poff(boff + bp->b_offset);
+		csize = min_t(size_t,
+			      PAGE_SIZE-cpoff, bp->b_count_desired-boff);
+
+		ASSERT(((csize + cpoff) <= PAGE_SIZE));
+
+		switch (mode) {
+		case XBRW_ZERO:
+			memset(page_address(page) + cpoff, 0, csize);
+			break;
+		case XBRW_READ:
+			memcpy(data, page_address(page) + cpoff, csize);
+			break;
+		case XBRW_WRITE:
+			memcpy(page_address(page) + cpoff, data, csize);
+		}
+
+		boff += csize;
+		data += csize;
+	}
+}
+
+/*
+ *	Handling of buffer targets (buftargs).
+ */
+
+/*
+ * Wait for any bufs with callbacks that have been submitted but have not yet
+ * returned. These buffers will have an elevated hold count, so wait on those
+ * while freeing all the buffers only held by the LRU.
+ */
+void
+xfs_wait_buftarg(
+	struct xfs_buftarg	*btp)
+{
+	struct xfs_buf		*bp;
+
+restart:
+	spin_lock(&btp->bt_lru_lock);
+	while (!list_empty(&btp->bt_lru)) {
+		bp = list_first_entry(&btp->bt_lru, struct xfs_buf, b_lru);
+		if (atomic_read(&bp->b_hold) > 1) {
+			spin_unlock(&btp->bt_lru_lock);
+			delay(100);
+			goto restart;
+		}
+		/*
+		 * clear the LRU reference count so the bufer doesn't get
+		 * ignored in xfs_buf_rele().
+		 */
+		atomic_set(&bp->b_lru_ref, 0);
+		spin_unlock(&btp->bt_lru_lock);
+		xfs_buf_rele(bp);
+		spin_lock(&btp->bt_lru_lock);
+	}
+	spin_unlock(&btp->bt_lru_lock);
+}
+
+int
+xfs_buftarg_shrink(
+	struct shrinker		*shrink,
+	struct shrink_control	*sc)
+{
+	struct xfs_buftarg	*btp = container_of(shrink,
+					struct xfs_buftarg, bt_shrinker);
+	struct xfs_buf		*bp;
+	int nr_to_scan = sc->nr_to_scan;
+	LIST_HEAD(dispose);
+
+	if (!nr_to_scan)
+		return btp->bt_lru_nr;
+
+	spin_lock(&btp->bt_lru_lock);
+	while (!list_empty(&btp->bt_lru)) {
+		if (nr_to_scan-- <= 0)
+			break;
+
+		bp = list_first_entry(&btp->bt_lru, struct xfs_buf, b_lru);
+
+		/*
+		 * Decrement the b_lru_ref count unless the value is already
+		 * zero. If the value is already zero, we need to reclaim the
+		 * buffer, otherwise it gets another trip through the LRU.
+		 */
+		if (!atomic_add_unless(&bp->b_lru_ref, -1, 0)) {
+			list_move_tail(&bp->b_lru, &btp->bt_lru);
+			continue;
+		}
+
+		/*
+		 * remove the buffer from the LRU now to avoid needing another
+		 * lock round trip inside xfs_buf_rele().
+		 */
+		list_move(&bp->b_lru, &dispose);
+		btp->bt_lru_nr--;
+	}
+	spin_unlock(&btp->bt_lru_lock);
+
+	while (!list_empty(&dispose)) {
+		bp = list_first_entry(&dispose, struct xfs_buf, b_lru);
+		list_del_init(&bp->b_lru);
+		xfs_buf_rele(bp);
+	}
+
+	return btp->bt_lru_nr;
+}
+
+void
+xfs_free_buftarg(
+	struct xfs_mount	*mp,
+	struct xfs_buftarg	*btp)
+{
+	unregister_shrinker(&btp->bt_shrinker);
+
+	xfs_flush_buftarg(btp, 1);
+	if (mp->m_flags & XFS_MOUNT_BARRIER)
+		xfs_blkdev_issue_flush(btp);
+
+	kthread_stop(btp->bt_task);
+	kmem_free(btp);
+}
+
+STATIC int
+xfs_setsize_buftarg_flags(
+	xfs_buftarg_t		*btp,
+	unsigned int		blocksize,
+	unsigned int		sectorsize,
+	int			verbose)
+{
+	btp->bt_bsize = blocksize;
+	btp->bt_sshift = ffs(sectorsize) - 1;
+	btp->bt_smask = sectorsize - 1;
+
+	if (set_blocksize(btp->bt_bdev, sectorsize)) {
+		xfs_warn(btp->bt_mount,
+			"Cannot set_blocksize to %u on device %s\n",
+			sectorsize, XFS_BUFTARG_NAME(btp));
+		return EINVAL;
+	}
+
+	return 0;
+}
+
+/*
+ *	When allocating the initial buffer target we have not yet
+ *	read in the superblock, so don't know what sized sectors
+ *	are being used is at this early stage.  Play safe.
+ */
+STATIC int
+xfs_setsize_buftarg_early(
+	xfs_buftarg_t		*btp,
+	struct block_device	*bdev)
+{
+	return xfs_setsize_buftarg_flags(btp,
+			PAGE_SIZE, bdev_logical_block_size(bdev), 0);
+}
+
+int
+xfs_setsize_buftarg(
+	xfs_buftarg_t		*btp,
+	unsigned int		blocksize,
+	unsigned int		sectorsize)
+{
+	return xfs_setsize_buftarg_flags(btp, blocksize, sectorsize, 1);
+}
+
+STATIC int
+xfs_alloc_delwrite_queue(
+	xfs_buftarg_t		*btp,
+	const char		*fsname)
+{
+	INIT_LIST_HEAD(&btp->bt_delwrite_queue);
+	spin_lock_init(&btp->bt_delwrite_lock);
+	btp->bt_flags = 0;
+	btp->bt_task = kthread_run(xfsbufd, btp, "xfsbufd/%s", fsname);
+	if (IS_ERR(btp->bt_task))
+		return PTR_ERR(btp->bt_task);
+	return 0;
+}
+
+xfs_buftarg_t *
+xfs_alloc_buftarg(
+	struct xfs_mount	*mp,
+	struct block_device	*bdev,
+	int			external,
+	const char		*fsname)
+{
+	xfs_buftarg_t		*btp;
+
+	btp = kmem_zalloc(sizeof(*btp), KM_SLEEP);
+
+	btp->bt_mount = mp;
+	btp->bt_dev =  bdev->bd_dev;
+	btp->bt_bdev = bdev;
+	btp->bt_bdi = blk_get_backing_dev_info(bdev);
+	if (!btp->bt_bdi)
+		goto error;
+
+	INIT_LIST_HEAD(&btp->bt_lru);
+	spin_lock_init(&btp->bt_lru_lock);
+	if (xfs_setsize_buftarg_early(btp, bdev))
+		goto error;
+	if (xfs_alloc_delwrite_queue(btp, fsname))
+		goto error;
+	btp->bt_shrinker.shrink = xfs_buftarg_shrink;
+	btp->bt_shrinker.seeks = DEFAULT_SEEKS;
+	register_shrinker(&btp->bt_shrinker);
+	return btp;
+
+error:
+	kmem_free(btp);
+	return NULL;
+}
+
+
+/*
+ *	Delayed write buffer handling
+ */
+STATIC void
+xfs_buf_delwri_queue(
+	xfs_buf_t		*bp,
+	int			unlock)
+{
+	struct list_head	*dwq = &bp->b_target->bt_delwrite_queue;
+	spinlock_t		*dwlk = &bp->b_target->bt_delwrite_lock;
+
+	trace_xfs_buf_delwri_queue(bp, _RET_IP_);
+
+	ASSERT((bp->b_flags&(XBF_DELWRI|XBF_ASYNC)) == (XBF_DELWRI|XBF_ASYNC));
+
+	spin_lock(dwlk);
+	/* If already in the queue, dequeue and place at tail */
+	if (!list_empty(&bp->b_list)) {
+		ASSERT(bp->b_flags & _XBF_DELWRI_Q);
+		if (unlock)
+			atomic_dec(&bp->b_hold);
+		list_del(&bp->b_list);
+	}
+
+	if (list_empty(dwq)) {
+		/* start xfsbufd as it is about to have something to do */
+		wake_up_process(bp->b_target->bt_task);
+	}
+
+	bp->b_flags |= _XBF_DELWRI_Q;
+	list_add_tail(&bp->b_list, dwq);
+	bp->b_queuetime = jiffies;
+	spin_unlock(dwlk);
+
+	if (unlock)
+		xfs_buf_unlock(bp);
+}
+
+void
+xfs_buf_delwri_dequeue(
+	xfs_buf_t		*bp)
+{
+	spinlock_t		*dwlk = &bp->b_target->bt_delwrite_lock;
+	int			dequeued = 0;
+
+	spin_lock(dwlk);
+	if ((bp->b_flags & XBF_DELWRI) && !list_empty(&bp->b_list)) {
+		ASSERT(bp->b_flags & _XBF_DELWRI_Q);
+		list_del_init(&bp->b_list);
+		dequeued = 1;
+	}
+	bp->b_flags &= ~(XBF_DELWRI|_XBF_DELWRI_Q);
+	spin_unlock(dwlk);
+
+	if (dequeued)
+		xfs_buf_rele(bp);
+
+	trace_xfs_buf_delwri_dequeue(bp, _RET_IP_);
+}
+
+/*
+ * If a delwri buffer needs to be pushed before it has aged out, then promote
+ * it to the head of the delwri queue so that it will be flushed on the next
+ * xfsbufd run. We do this by resetting the queuetime of the buffer to be older
+ * than the age currently needed to flush the buffer. Hence the next time the
+ * xfsbufd sees it is guaranteed to be considered old enough to flush.
+ */
+void
+xfs_buf_delwri_promote(
+	struct xfs_buf	*bp)
+{
+	struct xfs_buftarg *btp = bp->b_target;
+	long		age = xfs_buf_age_centisecs * msecs_to_jiffies(10) + 1;
+
+	ASSERT(bp->b_flags & XBF_DELWRI);
+	ASSERT(bp->b_flags & _XBF_DELWRI_Q);
+
+	/*
+	 * Check the buffer age before locking the delayed write queue as we
+	 * don't need to promote buffers that are already past the flush age.
+	 */
+	if (bp->b_queuetime < jiffies - age)
+		return;
+	bp->b_queuetime = jiffies - age;
+	spin_lock(&btp->bt_delwrite_lock);
+	list_move(&bp->b_list, &btp->bt_delwrite_queue);
+	spin_unlock(&btp->bt_delwrite_lock);
+}
+
+STATIC void
+xfs_buf_runall_queues(
+	struct workqueue_struct	*queue)
+{
+	flush_workqueue(queue);
+}
+
+/*
+ * Move as many buffers as specified to the supplied list
+ * idicating if we skipped any buffers to prevent deadlocks.
+ */
+STATIC int
+xfs_buf_delwri_split(
+	xfs_buftarg_t	*target,
+	struct list_head *list,
+	unsigned long	age)
+{
+	xfs_buf_t	*bp, *n;
+	struct list_head *dwq = &target->bt_delwrite_queue;
+	spinlock_t	*dwlk = &target->bt_delwrite_lock;
+	int		skipped = 0;
+	int		force;
+
+	force = test_and_clear_bit(XBT_FORCE_FLUSH, &target->bt_flags);
+	INIT_LIST_HEAD(list);
+	spin_lock(dwlk);
+	list_for_each_entry_safe(bp, n, dwq, b_list) {
+		ASSERT(bp->b_flags & XBF_DELWRI);
+
+		if (!XFS_BUF_ISPINNED(bp) && !xfs_buf_cond_lock(bp)) {
+			if (!force &&
+			    time_before(jiffies, bp->b_queuetime + age)) {
+				xfs_buf_unlock(bp);
+				break;
+			}
+
+			bp->b_flags &= ~(XBF_DELWRI|_XBF_DELWRI_Q|
+					 _XBF_RUN_QUEUES);
+			bp->b_flags |= XBF_WRITE;
+			list_move_tail(&bp->b_list, list);
+			trace_xfs_buf_delwri_split(bp, _RET_IP_);
+		} else
+			skipped++;
+	}
+	spin_unlock(dwlk);
+
+	return skipped;
+
+}
+
+/*
+ * Compare function is more complex than it needs to be because
+ * the return value is only 32 bits and we are doing comparisons
+ * on 64 bit values
+ */
+static int
+xfs_buf_cmp(
+	void		*priv,
+	struct list_head *a,
+	struct list_head *b)
+{
+	struct xfs_buf	*ap = container_of(a, struct xfs_buf, b_list);
+	struct xfs_buf	*bp = container_of(b, struct xfs_buf, b_list);
+	xfs_daddr_t		diff;
+
+	diff = ap->b_bn - bp->b_bn;
+	if (diff < 0)
+		return -1;
+	if (diff > 0)
+		return 1;
+	return 0;
+}
+
+void
+xfs_buf_delwri_sort(
+	xfs_buftarg_t	*target,
+	struct list_head *list)
+{
+	list_sort(NULL, list, xfs_buf_cmp);
+}
+
+STATIC int
+xfsbufd(
+	void		*data)
+{
+	xfs_buftarg_t   *target = (xfs_buftarg_t *)data;
+
+	current->flags |= PF_MEMALLOC;
+
+	set_freezable();
+
+	do {
+		long	age = xfs_buf_age_centisecs * msecs_to_jiffies(10);
+		long	tout = xfs_buf_timer_centisecs * msecs_to_jiffies(10);
+		struct list_head tmp;
+		struct blk_plug plug;
+
+		if (unlikely(freezing(current))) {
+			set_bit(XBT_FORCE_SLEEP, &target->bt_flags);
+			refrigerator();
+		} else {
+			clear_bit(XBT_FORCE_SLEEP, &target->bt_flags);
+		}
+
+		/* sleep for a long time if there is nothing to do. */
+		if (list_empty(&target->bt_delwrite_queue))
+			tout = MAX_SCHEDULE_TIMEOUT;
+		schedule_timeout_interruptible(tout);
+
+		xfs_buf_delwri_split(target, &tmp, age);
+		list_sort(NULL, &tmp, xfs_buf_cmp);
+
+		blk_start_plug(&plug);
+		while (!list_empty(&tmp)) {
+			struct xfs_buf *bp;
+			bp = list_first_entry(&tmp, struct xfs_buf, b_list);
+			list_del_init(&bp->b_list);
+			xfs_bdstrat_cb(bp);
+		}
+		blk_finish_plug(&plug);
+	} while (!kthread_should_stop());
+
+	return 0;
+}
+
+/*
+ *	Go through all incore buffers, and release buffers if they belong to
+ *	the given device. This is used in filesystem error handling to
+ *	preserve the consistency of its metadata.
+ */
+int
+xfs_flush_buftarg(
+	xfs_buftarg_t	*target,
+	int		wait)
+{
+	xfs_buf_t	*bp;
+	int		pincount = 0;
+	LIST_HEAD(tmp_list);
+	LIST_HEAD(wait_list);
+	struct blk_plug plug;
+
+	xfs_buf_runall_queues(xfsconvertd_workqueue);
+	xfs_buf_runall_queues(xfsdatad_workqueue);
+	xfs_buf_runall_queues(xfslogd_workqueue);
+
+	set_bit(XBT_FORCE_FLUSH, &target->bt_flags);
+	pincount = xfs_buf_delwri_split(target, &tmp_list, 0);
+
+	/*
+	 * Dropped the delayed write list lock, now walk the temporary list.
+	 * All I/O is issued async and then if we need to wait for completion
+	 * we do that after issuing all the IO.
+	 */
+	list_sort(NULL, &tmp_list, xfs_buf_cmp);
+
+	blk_start_plug(&plug);
+	while (!list_empty(&tmp_list)) {
+		bp = list_first_entry(&tmp_list, struct xfs_buf, b_list);
+		ASSERT(target == bp->b_target);
+		list_del_init(&bp->b_list);
+		if (wait) {
+			bp->b_flags &= ~XBF_ASYNC;
+			list_add(&bp->b_list, &wait_list);
+		}
+		xfs_bdstrat_cb(bp);
+	}
+	blk_finish_plug(&plug);
+
+	if (wait) {
+		/* Wait for IO to complete. */
+		while (!list_empty(&wait_list)) {
+			bp = list_first_entry(&wait_list, struct xfs_buf, b_list);
+
+			list_del_init(&bp->b_list);
+			xfs_buf_iowait(bp);
+			xfs_buf_relse(bp);
+		}
+	}
+
+	return pincount;
+}
+
+int __init
+xfs_buf_init(void)
+{
+	xfs_buf_zone = kmem_zone_init_flags(sizeof(xfs_buf_t), "xfs_buf",
+						KM_ZONE_HWALIGN, NULL);
+	if (!xfs_buf_zone)
+		goto out;
+
+	xfslogd_workqueue = alloc_workqueue("xfslogd",
+					WQ_MEM_RECLAIM | WQ_HIGHPRI, 1);
+	if (!xfslogd_workqueue)
+		goto out_free_buf_zone;
+
+	xfsdatad_workqueue = alloc_workqueue("xfsdatad", WQ_MEM_RECLAIM, 1);
+	if (!xfsdatad_workqueue)
+		goto out_destroy_xfslogd_workqueue;
+
+	xfsconvertd_workqueue = alloc_workqueue("xfsconvertd",
+						WQ_MEM_RECLAIM, 1);
+	if (!xfsconvertd_workqueue)
+		goto out_destroy_xfsdatad_workqueue;
+
+	return 0;
+
+ out_destroy_xfsdatad_workqueue:
+	destroy_workqueue(xfsdatad_workqueue);
+ out_destroy_xfslogd_workqueue:
+	destroy_workqueue(xfslogd_workqueue);
+ out_free_buf_zone:
+	kmem_zone_destroy(xfs_buf_zone);
+ out:
+	return -ENOMEM;
+}
+
+void
+xfs_buf_terminate(void)
+{
+	destroy_workqueue(xfsconvertd_workqueue);
+	destroy_workqueue(xfsdatad_workqueue);
+	destroy_workqueue(xfslogd_workqueue);
+	kmem_zone_destroy(xfs_buf_zone);
+}
+
+#ifdef CONFIG_KDB_MODULES
+struct list_head *
+xfs_get_buftarg_list(void)
+{
+	return &xfs_buftarg_list;
+}
+#endif
diff --git a/fs/xfs/linux-2.6/xfs_buf.h b/fs/xfs/linux-2.6/xfs_buf.h
new file mode 100644
index 00000000..36d6ee44
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_buf.h
@@ -0,0 +1,351 @@
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_BUF_H__
+#define __XFS_BUF_H__
+
+#include <linux/list.h>
+#include <linux/types.h>
+#include <linux/spinlock.h>
+#include <asm/system.h>
+#include <linux/mm.h>
+#include <linux/fs.h>
+#include <linux/buffer_head.h>
+#include <linux/uio.h>
+
+/*
+ *	Base types
+ */
+
+#define XFS_BUF_DADDR_NULL	((xfs_daddr_t) (-1LL))
+
+#define xfs_buf_ctob(pp)	((pp) * PAGE_CACHE_SIZE)
+#define xfs_buf_btoc(dd)	(((dd) + PAGE_CACHE_SIZE-1) >> PAGE_CACHE_SHIFT)
+#define xfs_buf_btoct(dd)	((dd) >> PAGE_CACHE_SHIFT)
+#define xfs_buf_poff(aa)	((aa) & ~PAGE_CACHE_MASK)
+
+typedef enum {
+	XBRW_READ = 1,			/* transfer into target memory */
+	XBRW_WRITE = 2,			/* transfer from target memory */
+	XBRW_ZERO = 3,			/* Zero target memory */
+} xfs_buf_rw_t;
+
+#define XBF_READ	(1 << 0) /* buffer intended for reading from device */
+#define XBF_WRITE	(1 << 1) /* buffer intended for writing to device */
+#define XBF_MAPPED	(1 << 2) /* buffer mapped (b_addr valid) */
+#define XBF_ASYNC	(1 << 4) /* initiator will not wait for completion */
+#define XBF_DONE	(1 << 5) /* all pages in the buffer uptodate */
+#define XBF_DELWRI	(1 << 6) /* buffer has dirty pages */
+#define XBF_STALE	(1 << 7) /* buffer has been staled, do not find it */
+#define XBF_ORDERED	(1 << 11)/* use ordered writes */
+#define XBF_READ_AHEAD	(1 << 12)/* asynchronous read-ahead */
+#define XBF_LOG_BUFFER	(1 << 13)/* this is a buffer used for the log */
+
+/* flags used only as arguments to access routines */
+#define XBF_LOCK	(1 << 14)/* lock requested */
+#define XBF_TRYLOCK	(1 << 15)/* lock requested, but do not wait */
+#define XBF_DONT_BLOCK	(1 << 16)/* do not block in current thread */
+
+/* flags used only internally */
+#define _XBF_PAGES	(1 << 18)/* backed by refcounted pages */
+#define	_XBF_RUN_QUEUES	(1 << 19)/* run block device task queue	*/
+#define	_XBF_KMEM	(1 << 20)/* backed by heap memory */
+#define _XBF_DELWRI_Q	(1 << 21)/* buffer on delwri queue */
+
+typedef unsigned int xfs_buf_flags_t;
+
+#define XFS_BUF_FLAGS \
+	{ XBF_READ,		"READ" }, \
+	{ XBF_WRITE,		"WRITE" }, \
+	{ XBF_MAPPED,		"MAPPED" }, \
+	{ XBF_ASYNC,		"ASYNC" }, \
+	{ XBF_DONE,		"DONE" }, \
+	{ XBF_DELWRI,		"DELWRI" }, \
+	{ XBF_STALE,		"STALE" }, \
+	{ XBF_ORDERED,		"ORDERED" }, \
+	{ XBF_READ_AHEAD,	"READ_AHEAD" }, \
+	{ XBF_LOCK,		"LOCK" },  	/* should never be set */\
+	{ XBF_TRYLOCK,		"TRYLOCK" }, 	/* ditto */\
+	{ XBF_DONT_BLOCK,	"DONT_BLOCK" },	/* ditto */\
+	{ _XBF_PAGES,		"PAGES" }, \
+	{ _XBF_RUN_QUEUES,	"RUN_QUEUES" }, \
+	{ _XBF_KMEM,		"KMEM" }, \
+	{ _XBF_DELWRI_Q,	"DELWRI_Q" }
+
+typedef enum {
+	XBT_FORCE_SLEEP = 0,
+	XBT_FORCE_FLUSH = 1,
+} xfs_buftarg_flags_t;
+
+typedef struct xfs_bufhash {
+	struct list_head	bh_list;
+	spinlock_t		bh_lock;
+} xfs_bufhash_t;
+
+typedef struct xfs_buftarg {
+	dev_t			bt_dev;
+	struct block_device	*bt_bdev;
+	struct backing_dev_info	*bt_bdi;
+	struct xfs_mount	*bt_mount;
+	unsigned int		bt_bsize;
+	unsigned int		bt_sshift;
+	size_t			bt_smask;
+
+	/* per device delwri queue */
+	struct task_struct	*bt_task;
+	struct list_head	bt_delwrite_queue;
+	spinlock_t		bt_delwrite_lock;
+	unsigned long		bt_flags;
+
+	/* LRU control structures */
+	struct shrinker		bt_shrinker;
+	struct list_head	bt_lru;
+	spinlock_t		bt_lru_lock;
+	unsigned int		bt_lru_nr;
+} xfs_buftarg_t;
+
+struct xfs_buf;
+typedef void (*xfs_buf_iodone_t)(struct xfs_buf *);
+
+#define XB_PAGES	2
+
+typedef struct xfs_buf {
+	/*
+	 * first cacheline holds all the fields needed for an uncontended cache
+	 * hit to be fully processed. The semaphore straddles the cacheline
+	 * boundary, but the counter and lock sits on the first cacheline,
+	 * which is the only bit that is touched if we hit the semaphore
+	 * fast-path on locking.
+	 */
+	struct rb_node		b_rbnode;	/* rbtree node */
+	xfs_off_t		b_file_offset;	/* offset in file */
+	size_t			b_buffer_length;/* size of buffer in bytes */
+	atomic_t		b_hold;		/* reference count */
+	atomic_t		b_lru_ref;	/* lru reclaim ref count */
+	xfs_buf_flags_t		b_flags;	/* status flags */
+	struct semaphore	b_sema;		/* semaphore for lockables */
+
+	struct list_head	b_lru;		/* lru list */
+	wait_queue_head_t	b_waiters;	/* unpin waiters */
+	struct list_head	b_list;
+	struct xfs_perag	*b_pag;		/* contains rbtree root */
+	xfs_buftarg_t		*b_target;	/* buffer target (device) */
+	xfs_daddr_t		b_bn;		/* block number for I/O */
+	size_t			b_count_desired;/* desired transfer size */
+	void			*b_addr;	/* virtual address of buffer */
+	struct work_struct	b_iodone_work;
+	xfs_buf_iodone_t	b_iodone;	/* I/O completion function */
+	struct completion	b_iowait;	/* queue for I/O waiters */
+	void			*b_fspriv;
+	void			*b_fspriv2;
+	struct page		**b_pages;	/* array of page pointers */
+	struct page		*b_page_array[XB_PAGES]; /* inline pages */
+	unsigned long		b_queuetime;	/* time buffer was queued */
+	atomic_t		b_pin_count;	/* pin count */
+	atomic_t		b_io_remaining;	/* #outstanding I/O requests */
+	unsigned int		b_page_count;	/* size of page array */
+	unsigned int		b_offset;	/* page offset in first page */
+	unsigned short		b_error;	/* error code on I/O */
+#ifdef XFS_BUF_LOCK_TRACKING
+	int			b_last_holder;
+#endif
+} xfs_buf_t;
+
+
+/* Finding and Reading Buffers */
+extern xfs_buf_t *_xfs_buf_find(xfs_buftarg_t *, xfs_off_t, size_t,
+				xfs_buf_flags_t, xfs_buf_t *);
+#define xfs_incore(buftarg,blkno,len,lockit) \
+	_xfs_buf_find(buftarg, blkno ,len, lockit, NULL)
+
+extern xfs_buf_t *xfs_buf_get(xfs_buftarg_t *, xfs_off_t, size_t,
+				xfs_buf_flags_t);
+extern xfs_buf_t *xfs_buf_read(xfs_buftarg_t *, xfs_off_t, size_t,
+				xfs_buf_flags_t);
+
+extern xfs_buf_t *xfs_buf_get_empty(size_t, xfs_buftarg_t *);
+extern void xfs_buf_set_empty(struct xfs_buf *bp, size_t len);
+extern xfs_buf_t *xfs_buf_get_uncached(struct xfs_buftarg *, size_t, int);
+extern int xfs_buf_associate_memory(xfs_buf_t *, void *, size_t);
+extern void xfs_buf_hold(xfs_buf_t *);
+extern void xfs_buf_readahead(xfs_buftarg_t *, xfs_off_t, size_t);
+struct xfs_buf *xfs_buf_read_uncached(struct xfs_mount *mp,
+				struct xfs_buftarg *target,
+				xfs_daddr_t daddr, size_t length, int flags);
+
+/* Releasing Buffers */
+extern void xfs_buf_free(xfs_buf_t *);
+extern void xfs_buf_rele(xfs_buf_t *);
+
+/* Locking and Unlocking Buffers */
+extern int xfs_buf_cond_lock(xfs_buf_t *);
+extern int xfs_buf_lock_value(xfs_buf_t *);
+extern void xfs_buf_lock(xfs_buf_t *);
+extern void xfs_buf_unlock(xfs_buf_t *);
+
+/* Buffer Read and Write Routines */
+extern int xfs_bwrite(struct xfs_mount *mp, struct xfs_buf *bp);
+extern void xfs_bdwrite(void *mp, xfs_buf_t *bp);
+
+extern void xfsbdstrat(struct xfs_mount *, struct xfs_buf *);
+extern int xfs_bdstrat_cb(struct xfs_buf *);
+
+extern void xfs_buf_ioend(xfs_buf_t *,	int);
+extern void xfs_buf_ioerror(xfs_buf_t *, int);
+extern int xfs_buf_iorequest(xfs_buf_t *);
+extern int xfs_buf_iowait(xfs_buf_t *);
+extern void xfs_buf_iomove(xfs_buf_t *, size_t, size_t, void *,
+				xfs_buf_rw_t);
+#define xfs_buf_zero(bp, off, len) \
+	    xfs_buf_iomove((bp), (off), (len), NULL, XBRW_ZERO)
+
+static inline int xfs_buf_geterror(xfs_buf_t *bp)
+{
+	return bp ? bp->b_error : ENOMEM;
+}
+
+/* Buffer Utility Routines */
+extern xfs_caddr_t xfs_buf_offset(xfs_buf_t *, size_t);
+
+/* Delayed Write Buffer Routines */
+extern void xfs_buf_delwri_dequeue(xfs_buf_t *);
+extern void xfs_buf_delwri_promote(xfs_buf_t *);
+
+/* Buffer Daemon Setup Routines */
+extern int xfs_buf_init(void);
+extern void xfs_buf_terminate(void);
+
+#define xfs_buf_target_name(target)	\
+	({ char __b[BDEVNAME_SIZE]; bdevname((target)->bt_bdev, __b); __b; })
+
+
+#define XFS_BUF_BFLAGS(bp)	((bp)->b_flags)
+#define XFS_BUF_ZEROFLAGS(bp)	((bp)->b_flags &= \
+		~(XBF_READ|XBF_WRITE|XBF_ASYNC|XBF_DELWRI|XBF_ORDERED))
+
+void xfs_buf_stale(struct xfs_buf *bp);
+#define XFS_BUF_STALE(bp)	xfs_buf_stale(bp);
+#define XFS_BUF_UNSTALE(bp)	((bp)->b_flags &= ~XBF_STALE)
+#define XFS_BUF_ISSTALE(bp)	((bp)->b_flags & XBF_STALE)
+#define XFS_BUF_SUPER_STALE(bp)	do {				\
+					XFS_BUF_STALE(bp);	\
+					xfs_buf_delwri_dequeue(bp);	\
+					XFS_BUF_DONE(bp);	\
+				} while (0)
+
+#define XFS_BUF_DELAYWRITE(bp)		((bp)->b_flags |= XBF_DELWRI)
+#define XFS_BUF_UNDELAYWRITE(bp)	xfs_buf_delwri_dequeue(bp)
+#define XFS_BUF_ISDELAYWRITE(bp)	((bp)->b_flags & XBF_DELWRI)
+
+#define XFS_BUF_ERROR(bp,no)	xfs_buf_ioerror(bp,no)
+#define XFS_BUF_GETERROR(bp)	xfs_buf_geterror(bp)
+#define XFS_BUF_ISERROR(bp)	(xfs_buf_geterror(bp) ? 1 : 0)
+
+#define XFS_BUF_DONE(bp)	((bp)->b_flags |= XBF_DONE)
+#define XFS_BUF_UNDONE(bp)	((bp)->b_flags &= ~XBF_DONE)
+#define XFS_BUF_ISDONE(bp)	((bp)->b_flags & XBF_DONE)
+
+#define XFS_BUF_BUSY(bp)	do { } while (0)
+#define XFS_BUF_UNBUSY(bp)	do { } while (0)
+#define XFS_BUF_ISBUSY(bp)	(1)
+
+#define XFS_BUF_ASYNC(bp)	((bp)->b_flags |= XBF_ASYNC)
+#define XFS_BUF_UNASYNC(bp)	((bp)->b_flags &= ~XBF_ASYNC)
+#define XFS_BUF_ISASYNC(bp)	((bp)->b_flags & XBF_ASYNC)
+
+#define XFS_BUF_ORDERED(bp)	((bp)->b_flags |= XBF_ORDERED)
+#define XFS_BUF_UNORDERED(bp)	((bp)->b_flags &= ~XBF_ORDERED)
+#define XFS_BUF_ISORDERED(bp)	((bp)->b_flags & XBF_ORDERED)
+
+#define XFS_BUF_HOLD(bp)	xfs_buf_hold(bp)
+#define XFS_BUF_READ(bp)	((bp)->b_flags |= XBF_READ)
+#define XFS_BUF_UNREAD(bp)	((bp)->b_flags &= ~XBF_READ)
+#define XFS_BUF_ISREAD(bp)	((bp)->b_flags & XBF_READ)
+
+#define XFS_BUF_WRITE(bp)	((bp)->b_flags |= XBF_WRITE)
+#define XFS_BUF_UNWRITE(bp)	((bp)->b_flags &= ~XBF_WRITE)
+#define XFS_BUF_ISWRITE(bp)	((bp)->b_flags & XBF_WRITE)
+
+#define XFS_BUF_IODONE_FUNC(bp)			((bp)->b_iodone)
+#define XFS_BUF_SET_IODONE_FUNC(bp, func)	((bp)->b_iodone = (func))
+#define XFS_BUF_CLR_IODONE_FUNC(bp)		((bp)->b_iodone = NULL)
+
+#define XFS_BUF_FSPRIVATE(bp, type)		((type)(bp)->b_fspriv)
+#define XFS_BUF_SET_FSPRIVATE(bp, val)		((bp)->b_fspriv = (void*)(val))
+#define XFS_BUF_FSPRIVATE2(bp, type)		((type)(bp)->b_fspriv2)
+#define XFS_BUF_SET_FSPRIVATE2(bp, val)		((bp)->b_fspriv2 = (void*)(val))
+#define XFS_BUF_SET_START(bp)			do { } while (0)
+
+#define XFS_BUF_PTR(bp)			(xfs_caddr_t)((bp)->b_addr)
+#define XFS_BUF_SET_PTR(bp, val, cnt)	xfs_buf_associate_memory(bp, val, cnt)
+#define XFS_BUF_ADDR(bp)		((bp)->b_bn)
+#define XFS_BUF_SET_ADDR(bp, bno)	((bp)->b_bn = (xfs_daddr_t)(bno))
+#define XFS_BUF_OFFSET(bp)		((bp)->b_file_offset)
+#define XFS_BUF_SET_OFFSET(bp, off)	((bp)->b_file_offset = (off))
+#define XFS_BUF_COUNT(bp)		((bp)->b_count_desired)
+#define XFS_BUF_SET_COUNT(bp, cnt)	((bp)->b_count_desired = (cnt))
+#define XFS_BUF_SIZE(bp)		((bp)->b_buffer_length)
+#define XFS_BUF_SET_SIZE(bp, cnt)	((bp)->b_buffer_length = (cnt))
+
+static inline void
+xfs_buf_set_ref(
+	struct xfs_buf	*bp,
+	int		lru_ref)
+{
+	atomic_set(&bp->b_lru_ref, lru_ref);
+}
+#define XFS_BUF_SET_VTYPE_REF(bp, type, ref)	xfs_buf_set_ref(bp, ref)
+#define XFS_BUF_SET_VTYPE(bp, type)		do { } while (0)
+
+#define XFS_BUF_ISPINNED(bp)	atomic_read(&((bp)->b_pin_count))
+
+#define XFS_BUF_VALUSEMA(bp)	xfs_buf_lock_value(bp)
+#define XFS_BUF_CPSEMA(bp)	(xfs_buf_cond_lock(bp) == 0)
+#define XFS_BUF_VSEMA(bp)	xfs_buf_unlock(bp)
+#define XFS_BUF_PSEMA(bp,x)	xfs_buf_lock(bp)
+#define XFS_BUF_FINISH_IOWAIT(bp)	complete(&bp->b_iowait);
+
+#define XFS_BUF_SET_TARGET(bp, target)	((bp)->b_target = (target))
+#define XFS_BUF_TARGET(bp)		((bp)->b_target)
+#define XFS_BUFTARG_NAME(target)	xfs_buf_target_name(target)
+
+static inline void xfs_buf_relse(xfs_buf_t *bp)
+{
+	xfs_buf_unlock(bp);
+	xfs_buf_rele(bp);
+}
+
+/*
+ *	Handling of buftargs.
+ */
+extern xfs_buftarg_t *xfs_alloc_buftarg(struct xfs_mount *,
+			struct block_device *, int, const char *);
+extern void xfs_free_buftarg(struct xfs_mount *, struct xfs_buftarg *);
+extern void xfs_wait_buftarg(xfs_buftarg_t *);
+extern int xfs_setsize_buftarg(xfs_buftarg_t *, unsigned int, unsigned int);
+extern int xfs_flush_buftarg(xfs_buftarg_t *, int);
+
+#ifdef CONFIG_KDB_MODULES
+extern struct list_head *xfs_get_buftarg_list(void);
+#endif
+
+#define xfs_getsize_buftarg(buftarg)	block_size((buftarg)->bt_bdev)
+#define xfs_readonly_buftarg(buftarg)	bdev_read_only((buftarg)->bt_bdev)
+
+#define XFS_bflush(buftarg)		xfs_flush_buftarg(buftarg, 1)
+
+#endif	/* __XFS_BUF_H__ */
diff --git a/fs/xfs/linux-2.6/xfs_discard.c b/fs/xfs/linux-2.6/xfs_discard.c
new file mode 100644
index 00000000..572494fa
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_discard.c
@@ -0,0 +1,222 @@
+/*
+ * Copyright (C) 2010 Red Hat, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_sb.h"
+#include "xfs_inum.h"
+#include "xfs_log.h"
+#include "xfs_ag.h"
+#include "xfs_mount.h"
+#include "xfs_quota.h"
+#include "xfs_trans.h"
+#include "xfs_alloc_btree.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_ialloc_btree.h"
+#include "xfs_btree.h"
+#include "xfs_inode.h"
+#include "xfs_alloc.h"
+#include "xfs_error.h"
+#include "xfs_discard.h"
+#include "xfs_trace.h"
+
+STATIC int
+xfs_trim_extents(
+	struct xfs_mount	*mp,
+	xfs_agnumber_t		agno,
+	xfs_fsblock_t		start,
+	xfs_fsblock_t		len,
+	xfs_fsblock_t		minlen,
+	__uint64_t		*blocks_trimmed)
+{
+	struct block_device	*bdev = mp->m_ddev_targp->bt_bdev;
+	struct xfs_btree_cur	*cur;
+	struct xfs_buf		*agbp;
+	struct xfs_perag	*pag;
+	int			error;
+	int			i;
+
+	pag = xfs_perag_get(mp, agno);
+
+	error = xfs_alloc_read_agf(mp, NULL, agno, 0, &agbp);
+	if (error || !agbp)
+		goto out_put_perag;
+
+	cur = xfs_allocbt_init_cursor(mp, NULL, agbp, agno, XFS_BTNUM_CNT);
+
+	/*
+	 * Force out the log.  This means any transactions that might have freed
+	 * space before we took the AGF buffer lock are now on disk, and the
+	 * volatile disk cache is flushed.
+	 */
+	xfs_log_force(mp, XFS_LOG_SYNC);
+
+	/*
+	 * Look up the longest btree in the AGF and start with it.
+	 */
+	error = xfs_alloc_lookup_le(cur, 0,
+			    be32_to_cpu(XFS_BUF_TO_AGF(agbp)->agf_longest), &i);
+	if (error)
+		goto out_del_cursor;
+
+	/*
+	 * Loop until we are done with all extents that are large
+	 * enough to be worth discarding.
+	 */
+	while (i) {
+		xfs_agblock_t fbno;
+		xfs_extlen_t flen;
+
+		error = xfs_alloc_get_rec(cur, &fbno, &flen, &i);
+		if (error)
+			goto out_del_cursor;
+		XFS_WANT_CORRUPTED_GOTO(i == 1, out_del_cursor);
+		ASSERT(flen <= be32_to_cpu(XFS_BUF_TO_AGF(agbp)->agf_longest));
+
+		/*
+		 * Too small?  Give up.
+		 */
+		if (flen < minlen) {
+			trace_xfs_discard_toosmall(mp, agno, fbno, flen);
+			goto out_del_cursor;
+		}
+
+		/*
+		 * If the extent is entirely outside of the range we are
+		 * supposed to discard skip it.  Do not bother to trim
+		 * down partially overlapping ranges for now.
+		 */
+		if (XFS_AGB_TO_FSB(mp, agno, fbno) + flen < start ||
+		    XFS_AGB_TO_FSB(mp, agno, fbno) >= start + len) {
+			trace_xfs_discard_exclude(mp, agno, fbno, flen);
+			goto next_extent;
+		}
+
+		/*
+		 * If any blocks in the range are still busy, skip the
+		 * discard and try again the next time.
+		 */
+		if (xfs_alloc_busy_search(mp, agno, fbno, flen)) {
+			trace_xfs_discard_busy(mp, agno, fbno, flen);
+			goto next_extent;
+		}
+
+		trace_xfs_discard_extent(mp, agno, fbno, flen);
+		error = -blkdev_issue_discard(bdev,
+				XFS_AGB_TO_DADDR(mp, agno, fbno),
+				XFS_FSB_TO_BB(mp, flen),
+				GFP_NOFS, 0);
+		if (error)
+			goto out_del_cursor;
+		*blocks_trimmed += flen;
+
+next_extent:
+		error = xfs_btree_decrement(cur, 0, &i);
+		if (error)
+			goto out_del_cursor;
+	}
+
+out_del_cursor:
+	xfs_btree_del_cursor(cur, error ? XFS_BTREE_ERROR : XFS_BTREE_NOERROR);
+	xfs_buf_relse(agbp);
+out_put_perag:
+	xfs_perag_put(pag);
+	return error;
+}
+
+int
+xfs_ioc_trim(
+	struct xfs_mount		*mp,
+	struct fstrim_range __user	*urange)
+{
+	struct request_queue	*q = mp->m_ddev_targp->bt_bdev->bd_disk->queue;
+	unsigned int		granularity = q->limits.discard_granularity;
+	struct fstrim_range	range;
+	xfs_fsblock_t		start, len, minlen;
+	xfs_agnumber_t		start_agno, end_agno, agno;
+	__uint64_t		blocks_trimmed = 0;
+	int			error, last_error = 0;
+
+	if (!capable(CAP_SYS_ADMIN))
+		return -XFS_ERROR(EPERM);
+	if (!blk_queue_discard(q))
+		return -XFS_ERROR(EOPNOTSUPP);
+	if (copy_from_user(&range, urange, sizeof(range)))
+		return -XFS_ERROR(EFAULT);
+
+	/*
+	 * Truncating down the len isn't actually quite correct, but using
+	 * XFS_B_TO_FSB would mean we trivially get overflows for values
+	 * of ULLONG_MAX or slightly lower.  And ULLONG_MAX is the default
+	 * used by the fstrim application.  In the end it really doesn't
+	 * matter as trimming blocks is an advisory interface.
+	 */
+	start = XFS_B_TO_FSBT(mp, range.start);
+	len = XFS_B_TO_FSBT(mp, range.len);
+	minlen = XFS_B_TO_FSB(mp, max_t(u64, granularity, range.minlen));
+
+	start_agno = XFS_FSB_TO_AGNO(mp, start);
+	if (start_agno >= mp->m_sb.sb_agcount)
+		return -XFS_ERROR(EINVAL);
+
+	end_agno = XFS_FSB_TO_AGNO(mp, start + len);
+	if (end_agno >= mp->m_sb.sb_agcount)
+		end_agno = mp->m_sb.sb_agcount - 1;
+
+	for (agno = start_agno; agno <= end_agno; agno++) {
+		error = -xfs_trim_extents(mp, agno, start, len, minlen,
+					  &blocks_trimmed);
+		if (error)
+			last_error = error;
+	}
+
+	if (last_error)
+		return last_error;
+
+	range.len = XFS_FSB_TO_B(mp, blocks_trimmed);
+	if (copy_to_user(urange, &range, sizeof(range)))
+		return -XFS_ERROR(EFAULT);
+	return 0;
+}
+
+int
+xfs_discard_extents(
+	struct xfs_mount	*mp,
+	struct list_head	*list)
+{
+	struct xfs_busy_extent	*busyp;
+	int			error = 0;
+
+	list_for_each_entry(busyp, list, list) {
+		trace_xfs_discard_extent(mp, busyp->agno, busyp->bno,
+					 busyp->length);
+
+		error = -blkdev_issue_discard(mp->m_ddev_targp->bt_bdev,
+				XFS_AGB_TO_DADDR(mp, busyp->agno, busyp->bno),
+				XFS_FSB_TO_BB(mp, busyp->length),
+				GFP_NOFS, 0);
+		if (error && error != EOPNOTSUPP) {
+			xfs_info(mp,
+	 "discard failed for extent [0x%llu,%u], error %d",
+				 (unsigned long long)busyp->bno,
+				 busyp->length,
+				 error);
+			return error;
+		}
+	}
+
+	return 0;
+}
diff --git a/fs/xfs/linux-2.6/xfs_discard.h b/fs/xfs/linux-2.6/xfs_discard.h
new file mode 100644
index 00000000..344879ae
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_discard.h
@@ -0,0 +1,10 @@
+#ifndef XFS_DISCARD_H
+#define XFS_DISCARD_H 1
+
+struct fstrim_range;
+struct list_head;
+
+extern int	xfs_ioc_trim(struct xfs_mount *, struct fstrim_range __user *);
+extern int	xfs_discard_extents(struct xfs_mount *, struct list_head *);
+
+#endif /* XFS_DISCARD_H */
diff --git a/fs/xfs/linux-2.6/xfs_export.c b/fs/xfs/linux-2.6/xfs_export.c
new file mode 100644
index 00000000..fed3f3c8
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_export.c
@@ -0,0 +1,250 @@
+/*
+ * Copyright (c) 2004-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_types.h"
+#include "xfs_inum.h"
+#include "xfs_log.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_dir2.h"
+#include "xfs_mount.h"
+#include "xfs_export.h"
+#include "xfs_vnodeops.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_inode.h"
+#include "xfs_inode_item.h"
+#include "xfs_trace.h"
+
+/*
+ * Note that we only accept fileids which are long enough rather than allow
+ * the parent generation number to default to zero.  XFS considers zero a
+ * valid generation number not an invalid/wildcard value.
+ */
+static int xfs_fileid_length(int fileid_type)
+{
+	switch (fileid_type) {
+	case FILEID_INO32_GEN:
+		return 2;
+	case FILEID_INO32_GEN_PARENT:
+		return 4;
+	case FILEID_INO32_GEN | XFS_FILEID_TYPE_64FLAG:
+		return 3;
+	case FILEID_INO32_GEN_PARENT | XFS_FILEID_TYPE_64FLAG:
+		return 6;
+	}
+	return 255; /* invalid */
+}
+
+STATIC int
+xfs_fs_encode_fh(
+	struct dentry		*dentry,
+	__u32			*fh,
+	int			*max_len,
+	int			connectable)
+{
+	struct fid		*fid = (struct fid *)fh;
+	struct xfs_fid64	*fid64 = (struct xfs_fid64 *)fh;
+	struct inode		*inode = dentry->d_inode;
+	int			fileid_type;
+	int			len;
+
+	/* Directories don't need their parent encoded, they have ".." */
+	if (S_ISDIR(inode->i_mode) || !connectable)
+		fileid_type = FILEID_INO32_GEN;
+	else
+		fileid_type = FILEID_INO32_GEN_PARENT;
+
+	/*
+	 * If the the filesystem may contain 64bit inode numbers, we need
+	 * to use larger file handles that can represent them.
+	 *
+	 * While we only allocate inodes that do not fit into 32 bits any
+	 * large enough filesystem may contain them, thus the slightly
+	 * confusing looking conditional below.
+	 */
+	if (!(XFS_M(inode->i_sb)->m_flags & XFS_MOUNT_SMALL_INUMS) ||
+	    (XFS_M(inode->i_sb)->m_flags & XFS_MOUNT_32BITINODES))
+		fileid_type |= XFS_FILEID_TYPE_64FLAG;
+
+	/*
+	 * Only encode if there is enough space given.  In practice
+	 * this means we can't export a filesystem with 64bit inodes
+	 * over NFSv2 with the subtree_check export option; the other
+	 * seven combinations work.  The real answer is "don't use v2".
+	 */
+	len = xfs_fileid_length(fileid_type);
+	if (*max_len < len) {
+		*max_len = len;
+		return 255;
+	}
+	*max_len = len;
+
+	switch (fileid_type) {
+	case FILEID_INO32_GEN_PARENT:
+		spin_lock(&dentry->d_lock);
+		fid->i32.parent_ino = XFS_I(dentry->d_parent->d_inode)->i_ino;
+		fid->i32.parent_gen = dentry->d_parent->d_inode->i_generation;
+		spin_unlock(&dentry->d_lock);
+		/*FALLTHRU*/
+	case FILEID_INO32_GEN:
+		fid->i32.ino = XFS_I(inode)->i_ino;
+		fid->i32.gen = inode->i_generation;
+		break;
+	case FILEID_INO32_GEN_PARENT | XFS_FILEID_TYPE_64FLAG:
+		spin_lock(&dentry->d_lock);
+		fid64->parent_ino = XFS_I(dentry->d_parent->d_inode)->i_ino;
+		fid64->parent_gen = dentry->d_parent->d_inode->i_generation;
+		spin_unlock(&dentry->d_lock);
+		/*FALLTHRU*/
+	case FILEID_INO32_GEN | XFS_FILEID_TYPE_64FLAG:
+		fid64->ino = XFS_I(inode)->i_ino;
+		fid64->gen = inode->i_generation;
+		break;
+	}
+
+	return fileid_type;
+}
+
+STATIC struct inode *
+xfs_nfs_get_inode(
+	struct super_block	*sb,
+	u64			ino,
+	u32			generation)
+ {
+ 	xfs_mount_t		*mp = XFS_M(sb);
+	xfs_inode_t		*ip;
+	int			error;
+
+	/*
+	 * NFS can sometimes send requests for ino 0.  Fail them gracefully.
+	 */
+	if (ino == 0)
+		return ERR_PTR(-ESTALE);
+
+	/*
+	 * The XFS_IGET_UNTRUSTED means that an invalid inode number is just
+	 * fine and not an indication of a corrupted filesystem as clients can
+	 * send invalid file handles and we have to handle it gracefully..
+	 */
+	error = xfs_iget(mp, NULL, ino, XFS_IGET_UNTRUSTED, 0, &ip);
+	if (error) {
+		/*
+		 * EINVAL means the inode cluster doesn't exist anymore.
+		 * This implies the filehandle is stale, so we should
+		 * translate it here.
+		 * We don't use ESTALE directly down the chain to not
+		 * confuse applications using bulkstat that expect EINVAL.
+		 */
+		if (error == EINVAL)
+			error = ESTALE;
+		return ERR_PTR(-error);
+	}
+
+	if (ip->i_d.di_gen != generation) {
+		IRELE(ip);
+		return ERR_PTR(-ENOENT);
+	}
+
+	return VFS_I(ip);
+}
+
+STATIC struct dentry *
+xfs_fs_fh_to_dentry(struct super_block *sb, struct fid *fid,
+		 int fh_len, int fileid_type)
+{
+	struct xfs_fid64	*fid64 = (struct xfs_fid64 *)fid;
+	struct inode		*inode = NULL;
+
+	if (fh_len < xfs_fileid_length(fileid_type))
+		return NULL;
+
+	switch (fileid_type) {
+	case FILEID_INO32_GEN_PARENT:
+	case FILEID_INO32_GEN:
+		inode = xfs_nfs_get_inode(sb, fid->i32.ino, fid->i32.gen);
+		break;
+	case FILEID_INO32_GEN_PARENT | XFS_FILEID_TYPE_64FLAG:
+	case FILEID_INO32_GEN | XFS_FILEID_TYPE_64FLAG:
+		inode = xfs_nfs_get_inode(sb, fid64->ino, fid64->gen);
+		break;
+	}
+
+	return d_obtain_alias(inode);
+}
+
+STATIC struct dentry *
+xfs_fs_fh_to_parent(struct super_block *sb, struct fid *fid,
+		 int fh_len, int fileid_type)
+{
+	struct xfs_fid64	*fid64 = (struct xfs_fid64 *)fid;
+	struct inode		*inode = NULL;
+
+	switch (fileid_type) {
+	case FILEID_INO32_GEN_PARENT:
+		inode = xfs_nfs_get_inode(sb, fid->i32.parent_ino,
+					      fid->i32.parent_gen);
+		break;
+	case FILEID_INO32_GEN_PARENT | XFS_FILEID_TYPE_64FLAG:
+		inode = xfs_nfs_get_inode(sb, fid64->parent_ino,
+					      fid64->parent_gen);
+		break;
+	}
+
+	return d_obtain_alias(inode);
+}
+
+STATIC struct dentry *
+xfs_fs_get_parent(
+	struct dentry		*child)
+{
+	int			error;
+	struct xfs_inode	*cip;
+
+	error = xfs_lookup(XFS_I(child->d_inode), &xfs_name_dotdot, &cip, NULL);
+	if (unlikely(error))
+		return ERR_PTR(-error);
+
+	return d_obtain_alias(VFS_I(cip));
+}
+
+STATIC int
+xfs_fs_nfs_commit_metadata(
+	struct inode		*inode)
+{
+	struct xfs_inode	*ip = XFS_I(inode);
+	struct xfs_mount	*mp = ip->i_mount;
+	int			error = 0;
+
+	xfs_ilock(ip, XFS_ILOCK_SHARED);
+	if (xfs_ipincount(ip)) {
+		error = _xfs_log_force_lsn(mp, ip->i_itemp->ili_last_lsn,
+				XFS_LOG_SYNC, NULL);
+	}
+	xfs_iunlock(ip, XFS_ILOCK_SHARED);
+
+	return error;
+}
+
+const struct export_operations xfs_export_operations = {
+	.encode_fh		= xfs_fs_encode_fh,
+	.fh_to_dentry		= xfs_fs_fh_to_dentry,
+	.fh_to_parent		= xfs_fs_fh_to_parent,
+	.get_parent		= xfs_fs_get_parent,
+	.commit_metadata	= xfs_fs_nfs_commit_metadata,
+};
diff --git a/fs/xfs/linux-2.6/xfs_export.h b/fs/xfs/linux-2.6/xfs_export.h
new file mode 100644
index 00000000..3272b6ae
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_export.h
@@ -0,0 +1,72 @@
+/*
+ * Copyright (c) 2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_EXPORT_H__
+#define __XFS_EXPORT_H__
+
+/*
+ * Common defines for code related to exporting XFS filesystems over NFS.
+ *
+ * The NFS fileid goes out on the wire as an array of
+ * 32bit unsigned ints in host order.  There are 5 possible
+ * formats.
+ *
+ * (1)	fileid_type=0x00
+ *	(no fileid data; handled by the generic code)
+ *
+ * (2)	fileid_type=0x01
+ *	inode-num
+ *	generation
+ *
+ * (3)	fileid_type=0x02
+ *	inode-num
+ *	generation
+ *	parent-inode-num
+ *	parent-generation
+ *
+ * (4)	fileid_type=0x81
+ *	inode-num-lo32
+ *	inode-num-hi32
+ *	generation
+ *
+ * (5)	fileid_type=0x82
+ *	inode-num-lo32
+ *	inode-num-hi32
+ *	generation
+ *	parent-inode-num-lo32
+ *	parent-inode-num-hi32
+ *	parent-generation
+ *
+ * Note, the NFS filehandle also includes an fsid portion which
+ * may have an inode number in it.  That number is hardcoded to
+ * 32bits and there is no way for XFS to intercept it.  In
+ * practice this means when exporting an XFS filesystem with 64bit
+ * inodes you should either export the mountpoint (rather than
+ * a subdirectory) or use the "fsid" export option.
+ */
+
+struct xfs_fid64 {
+	u64 ino;
+	u32 gen;
+	u64 parent_ino;
+	u32 parent_gen;
+} __attribute__((packed));
+
+/* This flag goes on the wire.  Don't play with it. */
+#define XFS_FILEID_TYPE_64FLAG	0x80	/* NFS fileid has 64bit inodes */
+
+#endif	/* __XFS_EXPORT_H__ */
diff --git a/fs/xfs/linux-2.6/xfs_file.c b/fs/xfs/linux-2.6/xfs_file.c
new file mode 100644
index 00000000..b679198d
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_file.c
@@ -0,0 +1,1114 @@
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_trans.h"
+#include "xfs_mount.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_alloc.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_inode_item.h"
+#include "xfs_bmap.h"
+#include "xfs_error.h"
+#include "xfs_vnodeops.h"
+#include "xfs_da_btree.h"
+#include "xfs_ioctl.h"
+#include "xfs_trace.h"
+
+#include <linux/dcache.h>
+#include <linux/falloc.h>
+
+static const struct vm_operations_struct xfs_file_vm_ops;
+
+/*
+ * Locking primitives for read and write IO paths to ensure we consistently use
+ * and order the inode->i_mutex, ip->i_lock and ip->i_iolock.
+ */
+static inline void
+xfs_rw_ilock(
+	struct xfs_inode	*ip,
+	int			type)
+{
+	if (type & XFS_IOLOCK_EXCL)
+		mutex_lock(&VFS_I(ip)->i_mutex);
+	xfs_ilock(ip, type);
+}
+
+static inline void
+xfs_rw_iunlock(
+	struct xfs_inode	*ip,
+	int			type)
+{
+	xfs_iunlock(ip, type);
+	if (type & XFS_IOLOCK_EXCL)
+		mutex_unlock(&VFS_I(ip)->i_mutex);
+}
+
+static inline void
+xfs_rw_ilock_demote(
+	struct xfs_inode	*ip,
+	int			type)
+{
+	xfs_ilock_demote(ip, type);
+	if (type & XFS_IOLOCK_EXCL)
+		mutex_unlock(&VFS_I(ip)->i_mutex);
+}
+
+/*
+ *	xfs_iozero
+ *
+ *	xfs_iozero clears the specified range of buffer supplied,
+ *	and marks all the affected blocks as valid and modified.  If
+ *	an affected block is not allocated, it will be allocated.  If
+ *	an affected block is not completely overwritten, and is not
+ *	valid before the operation, it will be read from disk before
+ *	being partially zeroed.
+ */
+STATIC int
+xfs_iozero(
+	struct xfs_inode	*ip,	/* inode			*/
+	loff_t			pos,	/* offset in file		*/
+	size_t			count)	/* size of data to zero		*/
+{
+	struct page		*page;
+	struct address_space	*mapping;
+	int			status;
+
+	mapping = VFS_I(ip)->i_mapping;
+	do {
+		unsigned offset, bytes;
+		void *fsdata;
+
+		offset = (pos & (PAGE_CACHE_SIZE -1)); /* Within page */
+		bytes = PAGE_CACHE_SIZE - offset;
+		if (bytes > count)
+			bytes = count;
+
+		status = pagecache_write_begin(NULL, mapping, pos, bytes,
+					AOP_FLAG_UNINTERRUPTIBLE,
+					&page, &fsdata);
+		if (status)
+			break;
+
+		zero_user(page, offset, bytes);
+
+		status = pagecache_write_end(NULL, mapping, pos, bytes, bytes,
+					page, fsdata);
+		WARN_ON(status <= 0); /* can't return less than zero! */
+		pos += bytes;
+		count -= bytes;
+		status = 0;
+	} while (count);
+
+	return (-status);
+}
+
+STATIC int
+xfs_file_fsync(
+	struct file		*file,
+	int			datasync)
+{
+	struct inode		*inode = file->f_mapping->host;
+	struct xfs_inode	*ip = XFS_I(inode);
+	struct xfs_mount	*mp = ip->i_mount;
+	struct xfs_trans	*tp;
+	int			error = 0;
+	int			log_flushed = 0;
+
+	trace_xfs_file_fsync(ip);
+
+	if (XFS_FORCED_SHUTDOWN(mp))
+		return -XFS_ERROR(EIO);
+
+	xfs_iflags_clear(ip, XFS_ITRUNCATED);
+
+	xfs_ioend_wait(ip);
+
+	if (mp->m_flags & XFS_MOUNT_BARRIER) {
+		/*
+		 * If we have an RT and/or log subvolume we need to make sure
+		 * to flush the write cache the device used for file data
+		 * first.  This is to ensure newly written file data make
+		 * it to disk before logging the new inode size in case of
+		 * an extending write.
+		 */
+		if (XFS_IS_REALTIME_INODE(ip))
+			xfs_blkdev_issue_flush(mp->m_rtdev_targp);
+		else if (mp->m_logdev_targp != mp->m_ddev_targp)
+			xfs_blkdev_issue_flush(mp->m_ddev_targp);
+	}
+
+	/*
+	 * We always need to make sure that the required inode state is safe on
+	 * disk.  The inode might be clean but we still might need to force the
+	 * log because of committed transactions that haven't hit the disk yet.
+	 * Likewise, there could be unflushed non-transactional changes to the
+	 * inode core that have to go to disk and this requires us to issue
+	 * a synchronous transaction to capture these changes correctly.
+	 *
+	 * This code relies on the assumption that if the i_update_core field
+	 * of the inode is clear and the inode is unpinned then it is clean
+	 * and no action is required.
+	 */
+	xfs_ilock(ip, XFS_ILOCK_SHARED);
+
+	/*
+	 * First check if the VFS inode is marked dirty.  All the dirtying
+	 * of non-transactional updates no goes through mark_inode_dirty*,
+	 * which allows us to distinguish beteeen pure timestamp updates
+	 * and i_size updates which need to be caught for fdatasync.
+	 * After that also theck for the dirty state in the XFS inode, which
+	 * might gets cleared when the inode gets written out via the AIL
+	 * or xfs_iflush_cluster.
+	 */
+	if (((inode->i_state & I_DIRTY_DATASYNC) ||
+	    ((inode->i_state & I_DIRTY_SYNC) && !datasync)) &&
+	    ip->i_update_core) {
+		/*
+		 * Kick off a transaction to log the inode core to get the
+		 * updates.  The sync transaction will also force the log.
+		 */
+		xfs_iunlock(ip, XFS_ILOCK_SHARED);
+		tp = xfs_trans_alloc(mp, XFS_TRANS_FSYNC_TS);
+		error = xfs_trans_reserve(tp, 0,
+				XFS_FSYNC_TS_LOG_RES(mp), 0, 0, 0);
+		if (error) {
+			xfs_trans_cancel(tp, 0);
+			return -error;
+		}
+		xfs_ilock(ip, XFS_ILOCK_EXCL);
+
+		/*
+		 * Note - it's possible that we might have pushed ourselves out
+		 * of the way during trans_reserve which would flush the inode.
+		 * But there's no guarantee that the inode buffer has actually
+		 * gone out yet (it's delwri).	Plus the buffer could be pinned
+		 * anyway if it's part of an inode in another recent
+		 * transaction.	 So we play it safe and fire off the
+		 * transaction anyway.
+		 */
+		xfs_trans_ijoin(tp, ip);
+		xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
+		xfs_trans_set_sync(tp);
+		error = _xfs_trans_commit(tp, 0, &log_flushed);
+
+		xfs_iunlock(ip, XFS_ILOCK_EXCL);
+	} else {
+		/*
+		 * Timestamps/size haven't changed since last inode flush or
+		 * inode transaction commit.  That means either nothing got
+		 * written or a transaction committed which caught the updates.
+		 * If the latter happened and the transaction hasn't hit the
+		 * disk yet, the inode will be still be pinned.  If it is,
+		 * force the log.
+		 */
+		if (xfs_ipincount(ip)) {
+			error = _xfs_log_force_lsn(mp,
+					ip->i_itemp->ili_last_lsn,
+					XFS_LOG_SYNC, &log_flushed);
+		}
+		xfs_iunlock(ip, XFS_ILOCK_SHARED);
+	}
+
+	/*
+	 * If we only have a single device, and the log force about was
+	 * a no-op we might have to flush the data device cache here.
+	 * This can only happen for fdatasync/O_DSYNC if we were overwriting
+	 * an already allocated file and thus do not have any metadata to
+	 * commit.
+	 */
+	if ((mp->m_flags & XFS_MOUNT_BARRIER) &&
+	    mp->m_logdev_targp == mp->m_ddev_targp &&
+	    !XFS_IS_REALTIME_INODE(ip) &&
+	    !log_flushed)
+		xfs_blkdev_issue_flush(mp->m_ddev_targp);
+
+	return -error;
+}
+
+STATIC ssize_t
+xfs_file_aio_read(
+	struct kiocb		*iocb,
+	const struct iovec	*iovp,
+	unsigned long		nr_segs,
+	loff_t			pos)
+{
+	struct file		*file = iocb->ki_filp;
+	struct inode		*inode = file->f_mapping->host;
+	struct xfs_inode	*ip = XFS_I(inode);
+	struct xfs_mount	*mp = ip->i_mount;
+	size_t			size = 0;
+	ssize_t			ret = 0;
+	int			ioflags = 0;
+	xfs_fsize_t		n;
+	unsigned long		seg;
+
+	XFS_STATS_INC(xs_read_calls);
+
+	BUG_ON(iocb->ki_pos != pos);
+
+	if (unlikely(file->f_flags & O_DIRECT))
+		ioflags |= IO_ISDIRECT;
+	if (file->f_mode & FMODE_NOCMTIME)
+		ioflags |= IO_INVIS;
+
+	/* START copy & waste from filemap.c */
+	for (seg = 0; seg < nr_segs; seg++) {
+		const struct iovec *iv = &iovp[seg];
+
+		/*
+		 * If any segment has a negative length, or the cumulative
+		 * length ever wraps negative then return -EINVAL.
+		 */
+		size += iv->iov_len;
+		if (unlikely((ssize_t)(size|iv->iov_len) < 0))
+			return XFS_ERROR(-EINVAL);
+	}
+	/* END copy & waste from filemap.c */
+
+	if (unlikely(ioflags & IO_ISDIRECT)) {
+		xfs_buftarg_t	*target =
+			XFS_IS_REALTIME_INODE(ip) ?
+				mp->m_rtdev_targp : mp->m_ddev_targp;
+		if ((iocb->ki_pos & target->bt_smask) ||
+		    (size & target->bt_smask)) {
+			if (iocb->ki_pos == ip->i_size)
+				return 0;
+			return -XFS_ERROR(EINVAL);
+		}
+	}
+
+	n = XFS_MAXIOFFSET(mp) - iocb->ki_pos;
+	if (n <= 0 || size == 0)
+		return 0;
+
+	if (n < size)
+		size = n;
+
+	if (XFS_FORCED_SHUTDOWN(mp))
+		return -EIO;
+
+	/*
+	 * Locking is a bit tricky here. If we take an exclusive lock
+	 * for direct IO, we effectively serialise all new concurrent
+	 * read IO to this file and block it behind IO that is currently in
+	 * progress because IO in progress holds the IO lock shared. We only
+	 * need to hold the lock exclusive to blow away the page cache, so
+	 * only take lock exclusively if the page cache needs invalidation.
+	 * This allows the normal direct IO case of no page cache pages to
+	 * proceeed concurrently without serialisation.
+	 */
+	xfs_rw_ilock(ip, XFS_IOLOCK_SHARED);
+	if ((ioflags & IO_ISDIRECT) && inode->i_mapping->nrpages) {
+		xfs_rw_iunlock(ip, XFS_IOLOCK_SHARED);
+		xfs_rw_ilock(ip, XFS_IOLOCK_EXCL);
+
+		if (inode->i_mapping->nrpages) {
+			ret = -xfs_flushinval_pages(ip,
+					(iocb->ki_pos & PAGE_CACHE_MASK),
+					-1, FI_REMAPF_LOCKED);
+			if (ret) {
+				xfs_rw_iunlock(ip, XFS_IOLOCK_EXCL);
+				return ret;
+			}
+		}
+		xfs_rw_ilock_demote(ip, XFS_IOLOCK_EXCL);
+	}
+
+	trace_xfs_file_read(ip, size, iocb->ki_pos, ioflags);
+
+	ret = generic_file_aio_read(iocb, iovp, nr_segs, iocb->ki_pos);
+	if (ret > 0)
+		XFS_STATS_ADD(xs_read_bytes, ret);
+
+	xfs_rw_iunlock(ip, XFS_IOLOCK_SHARED);
+	return ret;
+}
+
+STATIC ssize_t
+xfs_file_splice_read(
+	struct file		*infilp,
+	loff_t			*ppos,
+	struct pipe_inode_info	*pipe,
+	size_t			count,
+	unsigned int		flags)
+{
+	struct xfs_inode	*ip = XFS_I(infilp->f_mapping->host);
+	int			ioflags = 0;
+	ssize_t			ret;
+
+	XFS_STATS_INC(xs_read_calls);
+
+	if (infilp->f_mode & FMODE_NOCMTIME)
+		ioflags |= IO_INVIS;
+
+	if (XFS_FORCED_SHUTDOWN(ip->i_mount))
+		return -EIO;
+
+	xfs_rw_ilock(ip, XFS_IOLOCK_SHARED);
+
+	trace_xfs_file_splice_read(ip, count, *ppos, ioflags);
+
+	ret = generic_file_splice_read(infilp, ppos, pipe, count, flags);
+	if (ret > 0)
+		XFS_STATS_ADD(xs_read_bytes, ret);
+
+	xfs_rw_iunlock(ip, XFS_IOLOCK_SHARED);
+	return ret;
+}
+
+STATIC void
+xfs_aio_write_isize_update(
+	struct inode	*inode,
+	loff_t		*ppos,
+	ssize_t		bytes_written)
+{
+	struct xfs_inode	*ip = XFS_I(inode);
+	xfs_fsize_t		isize = i_size_read(inode);
+
+	if (bytes_written > 0)
+		XFS_STATS_ADD(xs_write_bytes, bytes_written);
+
+	if (unlikely(bytes_written < 0 && bytes_written != -EFAULT &&
+					*ppos > isize))
+		*ppos = isize;
+
+	if (*ppos > ip->i_size) {
+		xfs_rw_ilock(ip, XFS_ILOCK_EXCL);
+		if (*ppos > ip->i_size)
+			ip->i_size = *ppos;
+		xfs_rw_iunlock(ip, XFS_ILOCK_EXCL);
+	}
+}
+
+/*
+ * If this was a direct or synchronous I/O that failed (such as ENOSPC) then
+ * part of the I/O may have been written to disk before the error occurred.  In
+ * this case the on-disk file size may have been adjusted beyond the in-memory
+ * file size and now needs to be truncated back.
+ */
+STATIC void
+xfs_aio_write_newsize_update(
+	struct xfs_inode	*ip)
+{
+	if (ip->i_new_size) {
+		xfs_rw_ilock(ip, XFS_ILOCK_EXCL);
+		ip->i_new_size = 0;
+		if (ip->i_d.di_size > ip->i_size)
+			ip->i_d.di_size = ip->i_size;
+		xfs_rw_iunlock(ip, XFS_ILOCK_EXCL);
+	}
+}
+
+/*
+ * xfs_file_splice_write() does not use xfs_rw_ilock() because
+ * generic_file_splice_write() takes the i_mutex itself. This, in theory,
+ * couuld cause lock inversions between the aio_write path and the splice path
+ * if someone is doing concurrent splice(2) based writes and write(2) based
+ * writes to the same inode. The only real way to fix this is to re-implement
+ * the generic code here with correct locking orders.
+ */
+STATIC ssize_t
+xfs_file_splice_write(
+	struct pipe_inode_info	*pipe,
+	struct file		*outfilp,
+	loff_t			*ppos,
+	size_t			count,
+	unsigned int		flags)
+{
+	struct inode		*inode = outfilp->f_mapping->host;
+	struct xfs_inode	*ip = XFS_I(inode);
+	xfs_fsize_t		new_size;
+	int			ioflags = 0;
+	ssize_t			ret;
+
+	XFS_STATS_INC(xs_write_calls);
+
+	if (outfilp->f_mode & FMODE_NOCMTIME)
+		ioflags |= IO_INVIS;
+
+	if (XFS_FORCED_SHUTDOWN(ip->i_mount))
+		return -EIO;
+
+	xfs_ilock(ip, XFS_IOLOCK_EXCL);
+
+	new_size = *ppos + count;
+
+	xfs_ilock(ip, XFS_ILOCK_EXCL);
+	if (new_size > ip->i_size)
+		ip->i_new_size = new_size;
+	xfs_iunlock(ip, XFS_ILOCK_EXCL);
+
+	trace_xfs_file_splice_write(ip, count, *ppos, ioflags);
+
+	ret = generic_file_splice_write(pipe, outfilp, ppos, count, flags);
+
+	xfs_aio_write_isize_update(inode, ppos, ret);
+	xfs_aio_write_newsize_update(ip);
+	xfs_iunlock(ip, XFS_IOLOCK_EXCL);
+	return ret;
+}
+
+/*
+ * This routine is called to handle zeroing any space in the last
+ * block of the file that is beyond the EOF.  We do this since the
+ * size is being increased without writing anything to that block
+ * and we don't want anyone to read the garbage on the disk.
+ */
+STATIC int				/* error (positive) */
+xfs_zero_last_block(
+	xfs_inode_t	*ip,
+	xfs_fsize_t	offset,
+	xfs_fsize_t	isize)
+{
+	xfs_fileoff_t	last_fsb;
+	xfs_mount_t	*mp = ip->i_mount;
+	int		nimaps;
+	int		zero_offset;
+	int		zero_len;
+	int		error = 0;
+	xfs_bmbt_irec_t	imap;
+
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
+
+	zero_offset = XFS_B_FSB_OFFSET(mp, isize);
+	if (zero_offset == 0) {
+		/*
+		 * There are no extra bytes in the last block on disk to
+		 * zero, so return.
+		 */
+		return 0;
+	}
+
+	last_fsb = XFS_B_TO_FSBT(mp, isize);
+	nimaps = 1;
+	error = xfs_bmapi(NULL, ip, last_fsb, 1, 0, NULL, 0, &imap,
+			  &nimaps, NULL);
+	if (error) {
+		return error;
+	}
+	ASSERT(nimaps > 0);
+	/*
+	 * If the block underlying isize is just a hole, then there
+	 * is nothing to zero.
+	 */
+	if (imap.br_startblock == HOLESTARTBLOCK) {
+		return 0;
+	}
+	/*
+	 * Zero the part of the last block beyond the EOF, and write it
+	 * out sync.  We need to drop the ilock while we do this so we
+	 * don't deadlock when the buffer cache calls back to us.
+	 */
+	xfs_iunlock(ip, XFS_ILOCK_EXCL);
+
+	zero_len = mp->m_sb.sb_blocksize - zero_offset;
+	if (isize + zero_len > offset)
+		zero_len = offset - isize;
+	error = xfs_iozero(ip, isize, zero_len);
+
+	xfs_ilock(ip, XFS_ILOCK_EXCL);
+	ASSERT(error >= 0);
+	return error;
+}
+
+/*
+ * Zero any on disk space between the current EOF and the new,
+ * larger EOF.  This handles the normal case of zeroing the remainder
+ * of the last block in the file and the unusual case of zeroing blocks
+ * out beyond the size of the file.  This second case only happens
+ * with fixed size extents and when the system crashes before the inode
+ * size was updated but after blocks were allocated.  If fill is set,
+ * then any holes in the range are filled and zeroed.  If not, the holes
+ * are left alone as holes.
+ */
+
+int					/* error (positive) */
+xfs_zero_eof(
+	xfs_inode_t	*ip,
+	xfs_off_t	offset,		/* starting I/O offset */
+	xfs_fsize_t	isize)		/* current inode size */
+{
+	xfs_mount_t	*mp = ip->i_mount;
+	xfs_fileoff_t	start_zero_fsb;
+	xfs_fileoff_t	end_zero_fsb;
+	xfs_fileoff_t	zero_count_fsb;
+	xfs_fileoff_t	last_fsb;
+	xfs_fileoff_t	zero_off;
+	xfs_fsize_t	zero_len;
+	int		nimaps;
+	int		error = 0;
+	xfs_bmbt_irec_t	imap;
+
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL|XFS_IOLOCK_EXCL));
+	ASSERT(offset > isize);
+
+	/*
+	 * First handle zeroing the block on which isize resides.
+	 * We only zero a part of that block so it is handled specially.
+	 */
+	error = xfs_zero_last_block(ip, offset, isize);
+	if (error) {
+		ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL|XFS_IOLOCK_EXCL));
+		return error;
+	}
+
+	/*
+	 * Calculate the range between the new size and the old
+	 * where blocks needing to be zeroed may exist.  To get the
+	 * block where the last byte in the file currently resides,
+	 * we need to subtract one from the size and truncate back
+	 * to a block boundary.  We subtract 1 in case the size is
+	 * exactly on a block boundary.
+	 */
+	last_fsb = isize ? XFS_B_TO_FSBT(mp, isize - 1) : (xfs_fileoff_t)-1;
+	start_zero_fsb = XFS_B_TO_FSB(mp, (xfs_ufsize_t)isize);
+	end_zero_fsb = XFS_B_TO_FSBT(mp, offset - 1);
+	ASSERT((xfs_sfiloff_t)last_fsb < (xfs_sfiloff_t)start_zero_fsb);
+	if (last_fsb == end_zero_fsb) {
+		/*
+		 * The size was only incremented on its last block.
+		 * We took care of that above, so just return.
+		 */
+		return 0;
+	}
+
+	ASSERT(start_zero_fsb <= end_zero_fsb);
+	while (start_zero_fsb <= end_zero_fsb) {
+		nimaps = 1;
+		zero_count_fsb = end_zero_fsb - start_zero_fsb + 1;
+		error = xfs_bmapi(NULL, ip, start_zero_fsb, zero_count_fsb,
+				  0, NULL, 0, &imap, &nimaps, NULL);
+		if (error) {
+			ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL|XFS_IOLOCK_EXCL));
+			return error;
+		}
+		ASSERT(nimaps > 0);
+
+		if (imap.br_state == XFS_EXT_UNWRITTEN ||
+		    imap.br_startblock == HOLESTARTBLOCK) {
+			/*
+			 * This loop handles initializing pages that were
+			 * partially initialized by the code below this
+			 * loop. It basically zeroes the part of the page
+			 * that sits on a hole and sets the page as P_HOLE
+			 * and calls remapf if it is a mapped file.
+			 */
+			start_zero_fsb = imap.br_startoff + imap.br_blockcount;
+			ASSERT(start_zero_fsb <= (end_zero_fsb + 1));
+			continue;
+		}
+
+		/*
+		 * There are blocks we need to zero.
+		 * Drop the inode lock while we're doing the I/O.
+		 * We'll still have the iolock to protect us.
+		 */
+		xfs_iunlock(ip, XFS_ILOCK_EXCL);
+
+		zero_off = XFS_FSB_TO_B(mp, start_zero_fsb);
+		zero_len = XFS_FSB_TO_B(mp, imap.br_blockcount);
+
+		if ((zero_off + zero_len) > offset)
+			zero_len = offset - zero_off;
+
+		error = xfs_iozero(ip, zero_off, zero_len);
+		if (error) {
+			goto out_lock;
+		}
+
+		start_zero_fsb = imap.br_startoff + imap.br_blockcount;
+		ASSERT(start_zero_fsb <= (end_zero_fsb + 1));
+
+		xfs_ilock(ip, XFS_ILOCK_EXCL);
+	}
+
+	return 0;
+
+out_lock:
+	xfs_ilock(ip, XFS_ILOCK_EXCL);
+	ASSERT(error >= 0);
+	return error;
+}
+
+/*
+ * Common pre-write limit and setup checks.
+ *
+ * Returns with iolock held according to @iolock.
+ */
+STATIC ssize_t
+xfs_file_aio_write_checks(
+	struct file		*file,
+	loff_t			*pos,
+	size_t			*count,
+	int			*iolock)
+{
+	struct inode		*inode = file->f_mapping->host;
+	struct xfs_inode	*ip = XFS_I(inode);
+	xfs_fsize_t		new_size;
+	int			error = 0;
+
+	xfs_rw_ilock(ip, XFS_ILOCK_EXCL);
+	error = generic_write_checks(file, pos, count, S_ISBLK(inode->i_mode));
+	if (error) {
+		xfs_rw_iunlock(ip, XFS_ILOCK_EXCL | *iolock);
+		*iolock = 0;
+		return error;
+	}
+
+	new_size = *pos + *count;
+	if (new_size > ip->i_size)
+		ip->i_new_size = new_size;
+
+	if (likely(!(file->f_mode & FMODE_NOCMTIME)))
+		file_update_time(file);
+
+	/*
+	 * If the offset is beyond the size of the file, we need to zero any
+	 * blocks that fall between the existing EOF and the start of this
+	 * write.
+	 */
+	if (*pos > ip->i_size)
+		error = -xfs_zero_eof(ip, *pos, ip->i_size);
+
+	xfs_rw_iunlock(ip, XFS_ILOCK_EXCL);
+	if (error)
+		return error;
+
+	/*
+	 * If we're writing the file then make sure to clear the setuid and
+	 * setgid bits if the process is not being run by root.  This keeps
+	 * people from modifying setuid and setgid binaries.
+	 */
+	return file_remove_suid(file);
+
+}
+
+/*
+ * xfs_file_dio_aio_write - handle direct IO writes
+ *
+ * Lock the inode appropriately to prepare for and issue a direct IO write.
+ * By separating it from the buffered write path we remove all the tricky to
+ * follow locking changes and looping.
+ *
+ * If there are cached pages or we're extending the file, we need IOLOCK_EXCL
+ * until we're sure the bytes at the new EOF have been zeroed and/or the cached
+ * pages are flushed out.
+ *
+ * In most cases the direct IO writes will be done holding IOLOCK_SHARED
+ * allowing them to be done in parallel with reads and other direct IO writes.
+ * However, if the IO is not aligned to filesystem blocks, the direct IO layer
+ * needs to do sub-block zeroing and that requires serialisation against other
+ * direct IOs to the same block. In this case we need to serialise the
+ * submission of the unaligned IOs so that we don't get racing block zeroing in
+ * the dio layer.  To avoid the problem with aio, we also need to wait for
+ * outstanding IOs to complete so that unwritten extent conversion is completed
+ * before we try to map the overlapping block. This is currently implemented by
+ * hitting it with a big hammer (i.e. xfs_ioend_wait()).
+ *
+ * Returns with locks held indicated by @iolock and errors indicated by
+ * negative return values.
+ */
+STATIC ssize_t
+xfs_file_dio_aio_write(
+	struct kiocb		*iocb,
+	const struct iovec	*iovp,
+	unsigned long		nr_segs,
+	loff_t			pos,
+	size_t			ocount,
+	int			*iolock)
+{
+	struct file		*file = iocb->ki_filp;
+	struct address_space	*mapping = file->f_mapping;
+	struct inode		*inode = mapping->host;
+	struct xfs_inode	*ip = XFS_I(inode);
+	struct xfs_mount	*mp = ip->i_mount;
+	ssize_t			ret = 0;
+	size_t			count = ocount;
+	int			unaligned_io = 0;
+	struct xfs_buftarg	*target = XFS_IS_REALTIME_INODE(ip) ?
+					mp->m_rtdev_targp : mp->m_ddev_targp;
+
+	*iolock = 0;
+	if ((pos & target->bt_smask) || (count & target->bt_smask))
+		return -XFS_ERROR(EINVAL);
+
+	if ((pos & mp->m_blockmask) || ((pos + count) & mp->m_blockmask))
+		unaligned_io = 1;
+
+	if (unaligned_io || mapping->nrpages || pos > ip->i_size)
+		*iolock = XFS_IOLOCK_EXCL;
+	else
+		*iolock = XFS_IOLOCK_SHARED;
+	xfs_rw_ilock(ip, *iolock);
+
+	ret = xfs_file_aio_write_checks(file, &pos, &count, iolock);
+	if (ret)
+		return ret;
+
+	/*
+	 * Recheck if there are cached pages that need invalidate after we got
+	 * the iolock to protect against other threads adding new pages while
+	 * we were waiting for the iolock.
+	 */
+	if (mapping->nrpages && *iolock == XFS_IOLOCK_SHARED) {
+		xfs_rw_iunlock(ip, *iolock);
+		*iolock = XFS_IOLOCK_EXCL;
+		xfs_rw_ilock(ip, *iolock);
+	}
+
+	if (mapping->nrpages) {
+		ret = -xfs_flushinval_pages(ip, (pos & PAGE_CACHE_MASK), -1,
+							FI_REMAPF_LOCKED);
+		if (ret)
+			return ret;
+	}
+
+	/*
+	 * If we are doing unaligned IO, wait for all other IO to drain,
+	 * otherwise demote the lock if we had to flush cached pages
+	 */
+	if (unaligned_io)
+		xfs_ioend_wait(ip);
+	else if (*iolock == XFS_IOLOCK_EXCL) {
+		xfs_rw_ilock_demote(ip, XFS_IOLOCK_EXCL);
+		*iolock = XFS_IOLOCK_SHARED;
+	}
+
+	trace_xfs_file_direct_write(ip, count, iocb->ki_pos, 0);
+	ret = generic_file_direct_write(iocb, iovp,
+			&nr_segs, pos, &iocb->ki_pos, count, ocount);
+
+	/* No fallback to buffered IO on errors for XFS. */
+	ASSERT(ret < 0 || ret == count);
+	return ret;
+}
+
+STATIC ssize_t
+xfs_file_buffered_aio_write(
+	struct kiocb		*iocb,
+	const struct iovec	*iovp,
+	unsigned long		nr_segs,
+	loff_t			pos,
+	size_t			ocount,
+	int			*iolock)
+{
+	struct file		*file = iocb->ki_filp;
+	struct address_space	*mapping = file->f_mapping;
+	struct inode		*inode = mapping->host;
+	struct xfs_inode	*ip = XFS_I(inode);
+	ssize_t			ret;
+	int			enospc = 0;
+	size_t			count = ocount;
+
+	*iolock = XFS_IOLOCK_EXCL;
+	xfs_rw_ilock(ip, *iolock);
+
+	ret = xfs_file_aio_write_checks(file, &pos, &count, iolock);
+	if (ret)
+		return ret;
+
+	/* We can write back this queue in page reclaim */
+	current->backing_dev_info = mapping->backing_dev_info;
+
+write_retry:
+	trace_xfs_file_buffered_write(ip, count, iocb->ki_pos, 0);
+	ret = generic_file_buffered_write(iocb, iovp, nr_segs,
+			pos, &iocb->ki_pos, count, ret);
+	/*
+	 * if we just got an ENOSPC, flush the inode now we aren't holding any
+	 * page locks and retry *once*
+	 */
+	if (ret == -ENOSPC && !enospc) {
+		ret = -xfs_flush_pages(ip, 0, -1, 0, FI_NONE);
+		if (ret)
+			return ret;
+		enospc = 1;
+		goto write_retry;
+	}
+	current->backing_dev_info = NULL;
+	return ret;
+}
+
+STATIC ssize_t
+xfs_file_aio_write(
+	struct kiocb		*iocb,
+	const struct iovec	*iovp,
+	unsigned long		nr_segs,
+	loff_t			pos)
+{
+	struct file		*file = iocb->ki_filp;
+	struct address_space	*mapping = file->f_mapping;
+	struct inode		*inode = mapping->host;
+	struct xfs_inode	*ip = XFS_I(inode);
+	ssize_t			ret;
+	int			iolock;
+	size_t			ocount = 0;
+
+	XFS_STATS_INC(xs_write_calls);
+
+	BUG_ON(iocb->ki_pos != pos);
+
+	ret = generic_segment_checks(iovp, &nr_segs, &ocount, VERIFY_READ);
+	if (ret)
+		return ret;
+
+	if (ocount == 0)
+		return 0;
+
+	xfs_wait_for_freeze(ip->i_mount, SB_FREEZE_WRITE);
+
+	if (XFS_FORCED_SHUTDOWN(ip->i_mount))
+		return -EIO;
+
+	if (unlikely(file->f_flags & O_DIRECT))
+		ret = xfs_file_dio_aio_write(iocb, iovp, nr_segs, pos,
+						ocount, &iolock);
+	else
+		ret = xfs_file_buffered_aio_write(iocb, iovp, nr_segs, pos,
+						ocount, &iolock);
+
+	xfs_aio_write_isize_update(inode, &iocb->ki_pos, ret);
+
+	if (ret <= 0)
+		goto out_unlock;
+
+	/* Handle various SYNC-type writes */
+	if ((file->f_flags & O_DSYNC) || IS_SYNC(inode)) {
+		loff_t end = pos + ret - 1;
+		int error, error2;
+
+		xfs_rw_iunlock(ip, iolock);
+		error = filemap_write_and_wait_range(mapping, pos, end);
+		xfs_rw_ilock(ip, iolock);
+
+		error2 = -xfs_file_fsync(file,
+					 (file->f_flags & __O_SYNC) ? 0 : 1);
+		if (error)
+			ret = error;
+		else if (error2)
+			ret = error2;
+	}
+
+out_unlock:
+	xfs_aio_write_newsize_update(ip);
+	xfs_rw_iunlock(ip, iolock);
+	return ret;
+}
+
+STATIC long
+xfs_file_fallocate(
+	struct file	*file,
+	int		mode,
+	loff_t		offset,
+	loff_t		len)
+{
+	struct inode	*inode = file->f_path.dentry->d_inode;
+	long		error;
+	loff_t		new_size = 0;
+	xfs_flock64_t	bf;
+	xfs_inode_t	*ip = XFS_I(inode);
+	int		cmd = XFS_IOC_RESVSP;
+	int		attr_flags = XFS_ATTR_NOLOCK;
+
+	if (mode & ~(FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE))
+		return -EOPNOTSUPP;
+
+	bf.l_whence = 0;
+	bf.l_start = offset;
+	bf.l_len = len;
+
+	xfs_ilock(ip, XFS_IOLOCK_EXCL);
+
+	if (mode & FALLOC_FL_PUNCH_HOLE)
+		cmd = XFS_IOC_UNRESVSP;
+
+	/* check the new inode size is valid before allocating */
+	if (!(mode & FALLOC_FL_KEEP_SIZE) &&
+	    offset + len > i_size_read(inode)) {
+		new_size = offset + len;
+		error = inode_newsize_ok(inode, new_size);
+		if (error)
+			goto out_unlock;
+	}
+
+	if (file->f_flags & O_DSYNC)
+		attr_flags |= XFS_ATTR_SYNC;
+
+	error = -xfs_change_file_space(ip, cmd, &bf, 0, attr_flags);
+	if (error)
+		goto out_unlock;
+
+	/* Change file size if needed */
+	if (new_size) {
+		struct iattr iattr;
+
+		iattr.ia_valid = ATTR_SIZE;
+		iattr.ia_size = new_size;
+		error = -xfs_setattr(ip, &iattr, XFS_ATTR_NOLOCK);
+	}
+
+out_unlock:
+	xfs_iunlock(ip, XFS_IOLOCK_EXCL);
+	return error;
+}
+
+
+STATIC int
+xfs_file_open(
+	struct inode	*inode,
+	struct file	*file)
+{
+	if (!(file->f_flags & O_LARGEFILE) && i_size_read(inode) > MAX_NON_LFS)
+		return -EFBIG;
+	if (XFS_FORCED_SHUTDOWN(XFS_M(inode->i_sb)))
+		return -EIO;
+	return 0;
+}
+
+STATIC int
+xfs_dir_open(
+	struct inode	*inode,
+	struct file	*file)
+{
+	struct xfs_inode *ip = XFS_I(inode);
+	int		mode;
+	int		error;
+
+	error = xfs_file_open(inode, file);
+	if (error)
+		return error;
+
+	/*
+	 * If there are any blocks, read-ahead block 0 as we're almost
+	 * certain to have the next operation be a read there.
+	 */
+	mode = xfs_ilock_map_shared(ip);
+	if (ip->i_d.di_nextents > 0)
+		xfs_da_reada_buf(NULL, ip, 0, XFS_DATA_FORK);
+	xfs_iunlock(ip, mode);
+	return 0;
+}
+
+STATIC int
+xfs_file_release(
+	struct inode	*inode,
+	struct file	*filp)
+{
+	return -xfs_release(XFS_I(inode));
+}
+
+STATIC int
+xfs_file_readdir(
+	struct file	*filp,
+	void		*dirent,
+	filldir_t	filldir)
+{
+	struct inode	*inode = filp->f_path.dentry->d_inode;
+	xfs_inode_t	*ip = XFS_I(inode);
+	int		error;
+	size_t		bufsize;
+
+	/*
+	 * The Linux API doesn't pass down the total size of the buffer
+	 * we read into down to the filesystem.  With the filldir concept
+	 * it's not needed for correct information, but the XFS dir2 leaf
+	 * code wants an estimate of the buffer size to calculate it's
+	 * readahead window and size the buffers used for mapping to
+	 * physical blocks.
+	 *
+	 * Try to give it an estimate that's good enough, maybe at some
+	 * point we can change the ->readdir prototype to include the
+	 * buffer size.  For now we use the current glibc buffer size.
+	 */
+	bufsize = (size_t)min_t(loff_t, 32768, ip->i_d.di_size);
+
+	error = xfs_readdir(ip, dirent, bufsize,
+				(xfs_off_t *)&filp->f_pos, filldir);
+	if (error)
+		return -error;
+	return 0;
+}
+
+STATIC int
+xfs_file_mmap(
+	struct file	*filp,
+	struct vm_area_struct *vma)
+{
+	vma->vm_ops = &xfs_file_vm_ops;
+	vma->vm_flags |= VM_CAN_NONLINEAR;
+
+	file_accessed(filp);
+	return 0;
+}
+
+/*
+ * mmap()d file has taken write protection fault and is being made
+ * writable. We can set the page state up correctly for a writable
+ * page, which means we can do correct delalloc accounting (ENOSPC
+ * checking!) and unwritten extent mapping.
+ */
+STATIC int
+xfs_vm_page_mkwrite(
+	struct vm_area_struct	*vma,
+	struct vm_fault		*vmf)
+{
+	return block_page_mkwrite(vma, vmf, xfs_get_blocks);
+}
+
+const struct file_operations xfs_file_operations = {
+	.llseek		= generic_file_llseek,
+	.read		= do_sync_read,
+	.write		= do_sync_write,
+	.aio_read	= xfs_file_aio_read,
+	.aio_write	= xfs_file_aio_write,
+	.splice_read	= xfs_file_splice_read,
+	.splice_write	= xfs_file_splice_write,
+	.unlocked_ioctl	= xfs_file_ioctl,
+#ifdef CONFIG_COMPAT
+	.compat_ioctl	= xfs_file_compat_ioctl,
+#endif
+	.mmap		= xfs_file_mmap,
+	.open		= xfs_file_open,
+	.release	= xfs_file_release,
+	.fsync		= xfs_file_fsync,
+	.fallocate	= xfs_file_fallocate,
+};
+
+const struct file_operations xfs_dir_file_operations = {
+	.open		= xfs_dir_open,
+	.read		= generic_read_dir,
+	.readdir	= xfs_file_readdir,
+	.llseek		= generic_file_llseek,
+	.unlocked_ioctl	= xfs_file_ioctl,
+#ifdef CONFIG_COMPAT
+	.compat_ioctl	= xfs_file_compat_ioctl,
+#endif
+	.fsync		= xfs_file_fsync,
+};
+
+static const struct vm_operations_struct xfs_file_vm_ops = {
+	.fault		= filemap_fault,
+	.page_mkwrite	= xfs_vm_page_mkwrite,
+};
diff --git a/fs/xfs/linux-2.6/xfs_fs_subr.c b/fs/xfs/linux-2.6/xfs_fs_subr.c
new file mode 100644
index 00000000..ed88ed16
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_fs_subr.c
@@ -0,0 +1,96 @@
+/*
+ * Copyright (c) 2000-2002,2005-2006 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_vnodeops.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_inode.h"
+#include "xfs_trace.h"
+
+/*
+ * note: all filemap functions return negative error codes. These
+ * need to be inverted before returning to the xfs core functions.
+ */
+void
+xfs_tosspages(
+	xfs_inode_t	*ip,
+	xfs_off_t	first,
+	xfs_off_t	last,
+	int		fiopt)
+{
+	/* can't toss partial tail pages, so mask them out */
+	last &= ~(PAGE_SIZE - 1);
+	truncate_inode_pages_range(VFS_I(ip)->i_mapping, first, last - 1);
+}
+
+int
+xfs_flushinval_pages(
+	xfs_inode_t	*ip,
+	xfs_off_t	first,
+	xfs_off_t	last,
+	int		fiopt)
+{
+	struct address_space *mapping = VFS_I(ip)->i_mapping;
+	int		ret = 0;
+
+	trace_xfs_pagecache_inval(ip, first, last);
+
+	xfs_iflags_clear(ip, XFS_ITRUNCATED);
+	ret = filemap_write_and_wait_range(mapping, first,
+				last == -1 ? LLONG_MAX : last);
+	if (!ret)
+		truncate_inode_pages_range(mapping, first, last);
+	return -ret;
+}
+
+int
+xfs_flush_pages(
+	xfs_inode_t	*ip,
+	xfs_off_t	first,
+	xfs_off_t	last,
+	uint64_t	flags,
+	int		fiopt)
+{
+	struct address_space *mapping = VFS_I(ip)->i_mapping;
+	int		ret = 0;
+	int		ret2;
+
+	xfs_iflags_clear(ip, XFS_ITRUNCATED);
+	ret = -filemap_fdatawrite_range(mapping, first,
+				last == -1 ? LLONG_MAX : last);
+	if (flags & XBF_ASYNC)
+		return ret;
+	ret2 = xfs_wait_on_pages(ip, first, last);
+	if (!ret)
+		ret = ret2;
+	return ret;
+}
+
+int
+xfs_wait_on_pages(
+	xfs_inode_t	*ip,
+	xfs_off_t	first,
+	xfs_off_t	last)
+{
+	struct address_space *mapping = VFS_I(ip)->i_mapping;
+
+	if (mapping_tagged(mapping, PAGECACHE_TAG_WRITEBACK)) {
+		return -filemap_fdatawait_range(mapping, first,
+					last == -1 ? ip->i_size - 1 : last);
+	}
+	return 0;
+}
diff --git a/fs/xfs/linux-2.6/xfs_globals.c b/fs/xfs/linux-2.6/xfs_globals.c
new file mode 100644
index 00000000..76e81cff
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_globals.c
@@ -0,0 +1,43 @@
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_sysctl.h"
+
+/*
+ * Tunable XFS parameters.  xfs_params is required even when CONFIG_SYSCTL=n,
+ * other XFS code uses these values.  Times are measured in centisecs (i.e.
+ * 100ths of a second).
+ */
+xfs_param_t xfs_params = {
+			  /*	MIN		DFLT		MAX	*/
+	.sgid_inherit	= {	0,		0,		1	},
+	.symlink_mode	= {	0,		0,		1	},
+	.panic_mask	= {	0,		0,		255	},
+	.error_level	= {	0,		3,		11	},
+	.syncd_timer	= {	1*100,		30*100,		7200*100},
+	.stats_clear	= {	0,		0,		1	},
+	.inherit_sync	= {	0,		1,		1	},
+	.inherit_nodump	= {	0,		1,		1	},
+	.inherit_noatim = {	0,		1,		1	},
+	.xfs_buf_timer	= {	100/2,		1*100,		30*100	},
+	.xfs_buf_age	= {	1*100,		15*100,		7200*100},
+	.inherit_nosym	= {	0,		0,		1	},
+	.rotorstep	= {	1,		1,		255	},
+	.inherit_nodfrg	= {	0,		1,		1	},
+	.fstrm_timer	= {	1,		30*100,		3600*100},
+};
diff --git a/fs/xfs/linux-2.6/xfs_ioctl.c b/fs/xfs/linux-2.6/xfs_ioctl.c
new file mode 100644
index 00000000..acca2c5c
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_ioctl.c
@@ -0,0 +1,1556 @@
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_alloc.h"
+#include "xfs_mount.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_ioctl.h"
+#include "xfs_rtalloc.h"
+#include "xfs_itable.h"
+#include "xfs_error.h"
+#include "xfs_attr.h"
+#include "xfs_bmap.h"
+#include "xfs_buf_item.h"
+#include "xfs_utils.h"
+#include "xfs_dfrag.h"
+#include "xfs_fsops.h"
+#include "xfs_vnodeops.h"
+#include "xfs_discard.h"
+#include "xfs_quota.h"
+#include "xfs_inode_item.h"
+#include "xfs_export.h"
+#include "xfs_trace.h"
+
+#include <linux/capability.h>
+#include <linux/dcache.h>
+#include <linux/mount.h>
+#include <linux/namei.h>
+#include <linux/pagemap.h>
+#include <linux/slab.h>
+#include <linux/exportfs.h>
+
+/*
+ * xfs_find_handle maps from userspace xfs_fsop_handlereq structure to
+ * a file or fs handle.
+ *
+ * XFS_IOC_PATH_TO_FSHANDLE
+ *    returns fs handle for a mount point or path within that mount point
+ * XFS_IOC_FD_TO_HANDLE
+ *    returns full handle for a FD opened in user space
+ * XFS_IOC_PATH_TO_HANDLE
+ *    returns full handle for a path
+ */
+int
+xfs_find_handle(
+	unsigned int		cmd,
+	xfs_fsop_handlereq_t	*hreq)
+{
+	int			hsize;
+	xfs_handle_t		handle;
+	struct inode		*inode;
+	struct file		*file = NULL;
+	struct path		path;
+	int			error;
+	struct xfs_inode	*ip;
+
+	if (cmd == XFS_IOC_FD_TO_HANDLE) {
+		file = fget(hreq->fd);
+		if (!file)
+			return -EBADF;
+		inode = file->f_path.dentry->d_inode;
+	} else {
+		error = user_lpath((const char __user *)hreq->path, &path);
+		if (error)
+			return error;
+		inode = path.dentry->d_inode;
+	}
+	ip = XFS_I(inode);
+
+	/*
+	 * We can only generate handles for inodes residing on a XFS filesystem,
+	 * and only for regular files, directories or symbolic links.
+	 */
+	error = -EINVAL;
+	if (inode->i_sb->s_magic != XFS_SB_MAGIC)
+		goto out_put;
+
+	error = -EBADF;
+	if (!S_ISREG(inode->i_mode) &&
+	    !S_ISDIR(inode->i_mode) &&
+	    !S_ISLNK(inode->i_mode))
+		goto out_put;
+
+
+	memcpy(&handle.ha_fsid, ip->i_mount->m_fixedfsid, sizeof(xfs_fsid_t));
+
+	if (cmd == XFS_IOC_PATH_TO_FSHANDLE) {
+		/*
+		 * This handle only contains an fsid, zero the rest.
+		 */
+		memset(&handle.ha_fid, 0, sizeof(handle.ha_fid));
+		hsize = sizeof(xfs_fsid_t);
+	} else {
+		int		lock_mode;
+
+		lock_mode = xfs_ilock_map_shared(ip);
+		handle.ha_fid.fid_len = sizeof(xfs_fid_t) -
+					sizeof(handle.ha_fid.fid_len);
+		handle.ha_fid.fid_pad = 0;
+		handle.ha_fid.fid_gen = ip->i_d.di_gen;
+		handle.ha_fid.fid_ino = ip->i_ino;
+		xfs_iunlock_map_shared(ip, lock_mode);
+
+		hsize = XFS_HSIZE(handle);
+	}
+
+	error = -EFAULT;
+	if (copy_to_user(hreq->ohandle, &handle, hsize) ||
+	    copy_to_user(hreq->ohandlen, &hsize, sizeof(__s32)))
+		goto out_put;
+
+	error = 0;
+
+ out_put:
+	if (cmd == XFS_IOC_FD_TO_HANDLE)
+		fput(file);
+	else
+		path_put(&path);
+	return error;
+}
+
+/*
+ * No need to do permission checks on the various pathname components
+ * as the handle operations are privileged.
+ */
+STATIC int
+xfs_handle_acceptable(
+	void			*context,
+	struct dentry		*dentry)
+{
+	return 1;
+}
+
+/*
+ * Convert userspace handle data into a dentry.
+ */
+struct dentry *
+xfs_handle_to_dentry(
+	struct file		*parfilp,
+	void __user		*uhandle,
+	u32			hlen)
+{
+	xfs_handle_t		handle;
+	struct xfs_fid64	fid;
+
+	/*
+	 * Only allow handle opens under a directory.
+	 */
+	if (!S_ISDIR(parfilp->f_path.dentry->d_inode->i_mode))
+		return ERR_PTR(-ENOTDIR);
+
+	if (hlen != sizeof(xfs_handle_t))
+		return ERR_PTR(-EINVAL);
+	if (copy_from_user(&handle, uhandle, hlen))
+		return ERR_PTR(-EFAULT);
+	if (handle.ha_fid.fid_len !=
+	    sizeof(handle.ha_fid) - sizeof(handle.ha_fid.fid_len))
+		return ERR_PTR(-EINVAL);
+
+	memset(&fid, 0, sizeof(struct fid));
+	fid.ino = handle.ha_fid.fid_ino;
+	fid.gen = handle.ha_fid.fid_gen;
+
+	return exportfs_decode_fh(parfilp->f_path.mnt, (struct fid *)&fid, 3,
+			FILEID_INO32_GEN | XFS_FILEID_TYPE_64FLAG,
+			xfs_handle_acceptable, NULL);
+}
+
+STATIC struct dentry *
+xfs_handlereq_to_dentry(
+	struct file		*parfilp,
+	xfs_fsop_handlereq_t	*hreq)
+{
+	return xfs_handle_to_dentry(parfilp, hreq->ihandle, hreq->ihandlen);
+}
+
+int
+xfs_open_by_handle(
+	struct file		*parfilp,
+	xfs_fsop_handlereq_t	*hreq)
+{
+	const struct cred	*cred = current_cred();
+	int			error;
+	int			fd;
+	int			permflag;
+	struct file		*filp;
+	struct inode		*inode;
+	struct dentry		*dentry;
+
+	if (!capable(CAP_SYS_ADMIN))
+		return -XFS_ERROR(EPERM);
+
+	dentry = xfs_handlereq_to_dentry(parfilp, hreq);
+	if (IS_ERR(dentry))
+		return PTR_ERR(dentry);
+	inode = dentry->d_inode;
+
+	/* Restrict xfs_open_by_handle to directories & regular files. */
+	if (!(S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode))) {
+		error = -XFS_ERROR(EPERM);
+		goto out_dput;
+	}
+
+#if BITS_PER_LONG != 32
+	hreq->oflags |= O_LARGEFILE;
+#endif
+
+	/* Put open permission in namei format. */
+	permflag = hreq->oflags;
+	if ((permflag+1) & O_ACCMODE)
+		permflag++;
+	if (permflag & O_TRUNC)
+		permflag |= 2;
+
+	if ((!(permflag & O_APPEND) || (permflag & O_TRUNC)) &&
+	    (permflag & FMODE_WRITE) && IS_APPEND(inode)) {
+		error = -XFS_ERROR(EPERM);
+		goto out_dput;
+	}
+
+	if ((permflag & FMODE_WRITE) && IS_IMMUTABLE(inode)) {
+		error = -XFS_ERROR(EACCES);
+		goto out_dput;
+	}
+
+	/* Can't write directories. */
+	if (S_ISDIR(inode->i_mode) && (permflag & FMODE_WRITE)) {
+		error = -XFS_ERROR(EISDIR);
+		goto out_dput;
+	}
+
+	fd = get_unused_fd();
+	if (fd < 0) {
+		error = fd;
+		goto out_dput;
+	}
+
+	filp = dentry_open(dentry, mntget(parfilp->f_path.mnt),
+			   hreq->oflags, cred);
+	if (IS_ERR(filp)) {
+		put_unused_fd(fd);
+		return PTR_ERR(filp);
+	}
+
+	if (inode->i_mode & S_IFREG) {
+		filp->f_flags |= O_NOATIME;
+		filp->f_mode |= FMODE_NOCMTIME;
+	}
+
+	fd_install(fd, filp);
+	return fd;
+
+ out_dput:
+	dput(dentry);
+	return error;
+}
+
+/*
+ * This is a copy from fs/namei.c:vfs_readlink(), except for removing it's
+ * unused first argument.
+ */
+STATIC int
+do_readlink(
+	char __user		*buffer,
+	int			buflen,
+	const char		*link)
+{
+        int len;
+
+	len = PTR_ERR(link);
+	if (IS_ERR(link))
+		goto out;
+
+	len = strlen(link);
+	if (len > (unsigned) buflen)
+		len = buflen;
+	if (copy_to_user(buffer, link, len))
+		len = -EFAULT;
+ out:
+	return len;
+}
+
+
+int
+xfs_readlink_by_handle(
+	struct file		*parfilp,
+	xfs_fsop_handlereq_t	*hreq)
+{
+	struct dentry		*dentry;
+	__u32			olen;
+	void			*link;
+	int			error;
+
+	if (!capable(CAP_SYS_ADMIN))
+		return -XFS_ERROR(EPERM);
+
+	dentry = xfs_handlereq_to_dentry(parfilp, hreq);
+	if (IS_ERR(dentry))
+		return PTR_ERR(dentry);
+
+	/* Restrict this handle operation to symlinks only. */
+	if (!S_ISLNK(dentry->d_inode->i_mode)) {
+		error = -XFS_ERROR(EINVAL);
+		goto out_dput;
+	}
+
+	if (copy_from_user(&olen, hreq->ohandlen, sizeof(__u32))) {
+		error = -XFS_ERROR(EFAULT);
+		goto out_dput;
+	}
+
+	link = kmalloc(MAXPATHLEN+1, GFP_KERNEL);
+	if (!link) {
+		error = -XFS_ERROR(ENOMEM);
+		goto out_dput;
+	}
+
+	error = -xfs_readlink(XFS_I(dentry->d_inode), link);
+	if (error)
+		goto out_kfree;
+	error = do_readlink(hreq->ohandle, olen, link);
+	if (error)
+		goto out_kfree;
+
+ out_kfree:
+	kfree(link);
+ out_dput:
+	dput(dentry);
+	return error;
+}
+
+STATIC int
+xfs_fssetdm_by_handle(
+	struct file		*parfilp,
+	void			__user *arg)
+{
+	int			error;
+	struct fsdmidata	fsd;
+	xfs_fsop_setdm_handlereq_t dmhreq;
+	struct dentry		*dentry;
+
+	if (!capable(CAP_MKNOD))
+		return -XFS_ERROR(EPERM);
+	if (copy_from_user(&dmhreq, arg, sizeof(xfs_fsop_setdm_handlereq_t)))
+		return -XFS_ERROR(EFAULT);
+
+	dentry = xfs_handlereq_to_dentry(parfilp, &dmhreq.hreq);
+	if (IS_ERR(dentry))
+		return PTR_ERR(dentry);
+
+	if (IS_IMMUTABLE(dentry->d_inode) || IS_APPEND(dentry->d_inode)) {
+		error = -XFS_ERROR(EPERM);
+		goto out;
+	}
+
+	if (copy_from_user(&fsd, dmhreq.data, sizeof(fsd))) {
+		error = -XFS_ERROR(EFAULT);
+		goto out;
+	}
+
+	error = -xfs_set_dmattrs(XFS_I(dentry->d_inode), fsd.fsd_dmevmask,
+				 fsd.fsd_dmstate);
+
+ out:
+	dput(dentry);
+	return error;
+}
+
+STATIC int
+xfs_attrlist_by_handle(
+	struct file		*parfilp,
+	void			__user *arg)
+{
+	int			error = -ENOMEM;
+	attrlist_cursor_kern_t	*cursor;
+	xfs_fsop_attrlist_handlereq_t al_hreq;
+	struct dentry		*dentry;
+	char			*kbuf;
+
+	if (!capable(CAP_SYS_ADMIN))
+		return -XFS_ERROR(EPERM);
+	if (copy_from_user(&al_hreq, arg, sizeof(xfs_fsop_attrlist_handlereq_t)))
+		return -XFS_ERROR(EFAULT);
+	if (al_hreq.buflen > XATTR_LIST_MAX)
+		return -XFS_ERROR(EINVAL);
+
+	/*
+	 * Reject flags, only allow namespaces.
+	 */
+	if (al_hreq.flags & ~(ATTR_ROOT | ATTR_SECURE))
+		return -XFS_ERROR(EINVAL);
+
+	dentry = xfs_handlereq_to_dentry(parfilp, &al_hreq.hreq);
+	if (IS_ERR(dentry))
+		return PTR_ERR(dentry);
+
+	kbuf = kzalloc(al_hreq.buflen, GFP_KERNEL);
+	if (!kbuf)
+		goto out_dput;
+
+	cursor = (attrlist_cursor_kern_t *)&al_hreq.pos;
+	error = -xfs_attr_list(XFS_I(dentry->d_inode), kbuf, al_hreq.buflen,
+					al_hreq.flags, cursor);
+	if (error)
+		goto out_kfree;
+
+	if (copy_to_user(al_hreq.buffer, kbuf, al_hreq.buflen))
+		error = -EFAULT;
+
+ out_kfree:
+	kfree(kbuf);
+ out_dput:
+	dput(dentry);
+	return error;
+}
+
+int
+xfs_attrmulti_attr_get(
+	struct inode		*inode,
+	unsigned char		*name,
+	unsigned char		__user *ubuf,
+	__uint32_t		*len,
+	__uint32_t		flags)
+{
+	unsigned char		*kbuf;
+	int			error = EFAULT;
+
+	if (*len > XATTR_SIZE_MAX)
+		return EINVAL;
+	kbuf = kmalloc(*len, GFP_KERNEL);
+	if (!kbuf)
+		return ENOMEM;
+
+	error = xfs_attr_get(XFS_I(inode), name, kbuf, (int *)len, flags);
+	if (error)
+		goto out_kfree;
+
+	if (copy_to_user(ubuf, kbuf, *len))
+		error = EFAULT;
+
+ out_kfree:
+	kfree(kbuf);
+	return error;
+}
+
+int
+xfs_attrmulti_attr_set(
+	struct inode		*inode,
+	unsigned char		*name,
+	const unsigned char	__user *ubuf,
+	__uint32_t		len,
+	__uint32_t		flags)
+{
+	unsigned char		*kbuf;
+	int			error = EFAULT;
+
+	if (IS_IMMUTABLE(inode) || IS_APPEND(inode))
+		return EPERM;
+	if (len > XATTR_SIZE_MAX)
+		return EINVAL;
+
+	kbuf = memdup_user(ubuf, len);
+	if (IS_ERR(kbuf))
+		return PTR_ERR(kbuf);
+
+	error = xfs_attr_set(XFS_I(inode), name, kbuf, len, flags);
+
+	return error;
+}
+
+int
+xfs_attrmulti_attr_remove(
+	struct inode		*inode,
+	unsigned char		*name,
+	__uint32_t		flags)
+{
+	if (IS_IMMUTABLE(inode) || IS_APPEND(inode))
+		return EPERM;
+	return xfs_attr_remove(XFS_I(inode), name, flags);
+}
+
+STATIC int
+xfs_attrmulti_by_handle(
+	struct file		*parfilp,
+	void			__user *arg)
+{
+	int			error;
+	xfs_attr_multiop_t	*ops;
+	xfs_fsop_attrmulti_handlereq_t am_hreq;
+	struct dentry		*dentry;
+	unsigned int		i, size;
+	unsigned char		*attr_name;
+
+	if (!capable(CAP_SYS_ADMIN))
+		return -XFS_ERROR(EPERM);
+	if (copy_from_user(&am_hreq, arg, sizeof(xfs_fsop_attrmulti_handlereq_t)))
+		return -XFS_ERROR(EFAULT);
+
+	/* overflow check */
+	if (am_hreq.opcount >= INT_MAX / sizeof(xfs_attr_multiop_t))
+		return -E2BIG;
+
+	dentry = xfs_handlereq_to_dentry(parfilp, &am_hreq.hreq);
+	if (IS_ERR(dentry))
+		return PTR_ERR(dentry);
+
+	error = E2BIG;
+	size = am_hreq.opcount * sizeof(xfs_attr_multiop_t);
+	if (!size || size > 16 * PAGE_SIZE)
+		goto out_dput;
+
+	ops = memdup_user(am_hreq.ops, size);
+	if (IS_ERR(ops)) {
+		error = PTR_ERR(ops);
+		goto out_dput;
+	}
+
+	attr_name = kmalloc(MAXNAMELEN, GFP_KERNEL);
+	if (!attr_name)
+		goto out_kfree_ops;
+
+	error = 0;
+	for (i = 0; i < am_hreq.opcount; i++) {
+		ops[i].am_error = strncpy_from_user((char *)attr_name,
+				ops[i].am_attrname, MAXNAMELEN);
+		if (ops[i].am_error == 0 || ops[i].am_error == MAXNAMELEN)
+			error = -ERANGE;
+		if (ops[i].am_error < 0)
+			break;
+
+		switch (ops[i].am_opcode) {
+		case ATTR_OP_GET:
+			ops[i].am_error = xfs_attrmulti_attr_get(
+					dentry->d_inode, attr_name,
+					ops[i].am_attrvalue, &ops[i].am_length,
+					ops[i].am_flags);
+			break;
+		case ATTR_OP_SET:
+			ops[i].am_error = mnt_want_write(parfilp->f_path.mnt);
+			if (ops[i].am_error)
+				break;
+			ops[i].am_error = xfs_attrmulti_attr_set(
+					dentry->d_inode, attr_name,
+					ops[i].am_attrvalue, ops[i].am_length,
+					ops[i].am_flags);
+			mnt_drop_write(parfilp->f_path.mnt);
+			break;
+		case ATTR_OP_REMOVE:
+			ops[i].am_error = mnt_want_write(parfilp->f_path.mnt);
+			if (ops[i].am_error)
+				break;
+			ops[i].am_error = xfs_attrmulti_attr_remove(
+					dentry->d_inode, attr_name,
+					ops[i].am_flags);
+			mnt_drop_write(parfilp->f_path.mnt);
+			break;
+		default:
+			ops[i].am_error = EINVAL;
+		}
+	}
+
+	if (copy_to_user(am_hreq.ops, ops, size))
+		error = XFS_ERROR(EFAULT);
+
+	kfree(attr_name);
+ out_kfree_ops:
+	kfree(ops);
+ out_dput:
+	dput(dentry);
+	return -error;
+}
+
+int
+xfs_ioc_space(
+	struct xfs_inode	*ip,
+	struct inode		*inode,
+	struct file		*filp,
+	int			ioflags,
+	unsigned int		cmd,
+	xfs_flock64_t		*bf)
+{
+	int			attr_flags = 0;
+	int			error;
+
+	/*
+	 * Only allow the sys admin to reserve space unless
+	 * unwritten extents are enabled.
+	 */
+	if (!xfs_sb_version_hasextflgbit(&ip->i_mount->m_sb) &&
+	    !capable(CAP_SYS_ADMIN))
+		return -XFS_ERROR(EPERM);
+
+	if (inode->i_flags & (S_IMMUTABLE|S_APPEND))
+		return -XFS_ERROR(EPERM);
+
+	if (!(filp->f_mode & FMODE_WRITE))
+		return -XFS_ERROR(EBADF);
+
+	if (!S_ISREG(inode->i_mode))
+		return -XFS_ERROR(EINVAL);
+
+	if (filp->f_flags & (O_NDELAY|O_NONBLOCK))
+		attr_flags |= XFS_ATTR_NONBLOCK;
+
+	if (filp->f_flags & O_DSYNC)
+		attr_flags |= XFS_ATTR_SYNC;
+
+	if (ioflags & IO_INVIS)
+		attr_flags |= XFS_ATTR_DMI;
+
+	error = xfs_change_file_space(ip, cmd, bf, filp->f_pos, attr_flags);
+	return -error;
+}
+
+STATIC int
+xfs_ioc_bulkstat(
+	xfs_mount_t		*mp,
+	unsigned int		cmd,
+	void			__user *arg)
+{
+	xfs_fsop_bulkreq_t	bulkreq;
+	int			count;	/* # of records returned */
+	xfs_ino_t		inlast;	/* last inode number */
+	int			done;
+	int			error;
+
+	/* done = 1 if there are more stats to get and if bulkstat */
+	/* should be called again (unused here, but used in dmapi) */
+
+	if (!capable(CAP_SYS_ADMIN))
+		return -EPERM;
+
+	if (XFS_FORCED_SHUTDOWN(mp))
+		return -XFS_ERROR(EIO);
+
+	if (copy_from_user(&bulkreq, arg, sizeof(xfs_fsop_bulkreq_t)))
+		return -XFS_ERROR(EFAULT);
+
+	if (copy_from_user(&inlast, bulkreq.lastip, sizeof(__s64)))
+		return -XFS_ERROR(EFAULT);
+
+	if ((count = bulkreq.icount) <= 0)
+		return -XFS_ERROR(EINVAL);
+
+	if (bulkreq.ubuffer == NULL)
+		return -XFS_ERROR(EINVAL);
+
+	if (cmd == XFS_IOC_FSINUMBERS)
+		error = xfs_inumbers(mp, &inlast, &count,
+					bulkreq.ubuffer, xfs_inumbers_fmt);
+	else if (cmd == XFS_IOC_FSBULKSTAT_SINGLE)
+		error = xfs_bulkstat_single(mp, &inlast,
+						bulkreq.ubuffer, &done);
+	else	/* XFS_IOC_FSBULKSTAT */
+		error = xfs_bulkstat(mp, &inlast, &count, xfs_bulkstat_one,
+				     sizeof(xfs_bstat_t), bulkreq.ubuffer,
+				     &done);
+
+	if (error)
+		return -error;
+
+	if (bulkreq.ocount != NULL) {
+		if (copy_to_user(bulkreq.lastip, &inlast,
+						sizeof(xfs_ino_t)))
+			return -XFS_ERROR(EFAULT);
+
+		if (copy_to_user(bulkreq.ocount, &count, sizeof(count)))
+			return -XFS_ERROR(EFAULT);
+	}
+
+	return 0;
+}
+
+STATIC int
+xfs_ioc_fsgeometry_v1(
+	xfs_mount_t		*mp,
+	void			__user *arg)
+{
+	xfs_fsop_geom_t         fsgeo;
+	int			error;
+
+	error = xfs_fs_geometry(mp, &fsgeo, 3);
+	if (error)
+		return -error;
+
+	/*
+	 * Caller should have passed an argument of type
+	 * xfs_fsop_geom_v1_t.  This is a proper subset of the
+	 * xfs_fsop_geom_t that xfs_fs_geometry() fills in.
+	 */
+	if (copy_to_user(arg, &fsgeo, sizeof(xfs_fsop_geom_v1_t)))
+		return -XFS_ERROR(EFAULT);
+	return 0;
+}
+
+STATIC int
+xfs_ioc_fsgeometry(
+	xfs_mount_t		*mp,
+	void			__user *arg)
+{
+	xfs_fsop_geom_t		fsgeo;
+	int			error;
+
+	error = xfs_fs_geometry(mp, &fsgeo, 4);
+	if (error)
+		return -error;
+
+	if (copy_to_user(arg, &fsgeo, sizeof(fsgeo)))
+		return -XFS_ERROR(EFAULT);
+	return 0;
+}
+
+/*
+ * Linux extended inode flags interface.
+ */
+
+STATIC unsigned int
+xfs_merge_ioc_xflags(
+	unsigned int	flags,
+	unsigned int	start)
+{
+	unsigned int	xflags = start;
+
+	if (flags & FS_IMMUTABLE_FL)
+		xflags |= XFS_XFLAG_IMMUTABLE;
+	else
+		xflags &= ~XFS_XFLAG_IMMUTABLE;
+	if (flags & FS_APPEND_FL)
+		xflags |= XFS_XFLAG_APPEND;
+	else
+		xflags &= ~XFS_XFLAG_APPEND;
+	if (flags & FS_SYNC_FL)
+		xflags |= XFS_XFLAG_SYNC;
+	else
+		xflags &= ~XFS_XFLAG_SYNC;
+	if (flags & FS_NOATIME_FL)
+		xflags |= XFS_XFLAG_NOATIME;
+	else
+		xflags &= ~XFS_XFLAG_NOATIME;
+	if (flags & FS_NODUMP_FL)
+		xflags |= XFS_XFLAG_NODUMP;
+	else
+		xflags &= ~XFS_XFLAG_NODUMP;
+
+	return xflags;
+}
+
+STATIC unsigned int
+xfs_di2lxflags(
+	__uint16_t	di_flags)
+{
+	unsigned int	flags = 0;
+
+	if (di_flags & XFS_DIFLAG_IMMUTABLE)
+		flags |= FS_IMMUTABLE_FL;
+	if (di_flags & XFS_DIFLAG_APPEND)
+		flags |= FS_APPEND_FL;
+	if (di_flags & XFS_DIFLAG_SYNC)
+		flags |= FS_SYNC_FL;
+	if (di_flags & XFS_DIFLAG_NOATIME)
+		flags |= FS_NOATIME_FL;
+	if (di_flags & XFS_DIFLAG_NODUMP)
+		flags |= FS_NODUMP_FL;
+	return flags;
+}
+
+STATIC int
+xfs_ioc_fsgetxattr(
+	xfs_inode_t		*ip,
+	int			attr,
+	void			__user *arg)
+{
+	struct fsxattr		fa;
+
+	memset(&fa, 0, sizeof(struct fsxattr));
+
+	xfs_ilock(ip, XFS_ILOCK_SHARED);
+	fa.fsx_xflags = xfs_ip2xflags(ip);
+	fa.fsx_extsize = ip->i_d.di_extsize << ip->i_mount->m_sb.sb_blocklog;
+	fa.fsx_projid = xfs_get_projid(ip);
+
+	if (attr) {
+		if (ip->i_afp) {
+			if (ip->i_afp->if_flags & XFS_IFEXTENTS)
+				fa.fsx_nextents = ip->i_afp->if_bytes /
+							sizeof(xfs_bmbt_rec_t);
+			else
+				fa.fsx_nextents = ip->i_d.di_anextents;
+		} else
+			fa.fsx_nextents = 0;
+	} else {
+		if (ip->i_df.if_flags & XFS_IFEXTENTS)
+			fa.fsx_nextents = ip->i_df.if_bytes /
+						sizeof(xfs_bmbt_rec_t);
+		else
+			fa.fsx_nextents = ip->i_d.di_nextents;
+	}
+	xfs_iunlock(ip, XFS_ILOCK_SHARED);
+
+	if (copy_to_user(arg, &fa, sizeof(fa)))
+		return -EFAULT;
+	return 0;
+}
+
+STATIC void
+xfs_set_diflags(
+	struct xfs_inode	*ip,
+	unsigned int		xflags)
+{
+	unsigned int		di_flags;
+
+	/* can't set PREALLOC this way, just preserve it */
+	di_flags = (ip->i_d.di_flags & XFS_DIFLAG_PREALLOC);
+	if (xflags & XFS_XFLAG_IMMUTABLE)
+		di_flags |= XFS_DIFLAG_IMMUTABLE;
+	if (xflags & XFS_XFLAG_APPEND)
+		di_flags |= XFS_DIFLAG_APPEND;
+	if (xflags & XFS_XFLAG_SYNC)
+		di_flags |= XFS_DIFLAG_SYNC;
+	if (xflags & XFS_XFLAG_NOATIME)
+		di_flags |= XFS_DIFLAG_NOATIME;
+	if (xflags & XFS_XFLAG_NODUMP)
+		di_flags |= XFS_DIFLAG_NODUMP;
+	if (xflags & XFS_XFLAG_PROJINHERIT)
+		di_flags |= XFS_DIFLAG_PROJINHERIT;
+	if (xflags & XFS_XFLAG_NODEFRAG)
+		di_flags |= XFS_DIFLAG_NODEFRAG;
+	if (xflags & XFS_XFLAG_FILESTREAM)
+		di_flags |= XFS_DIFLAG_FILESTREAM;
+	if ((ip->i_d.di_mode & S_IFMT) == S_IFDIR) {
+		if (xflags & XFS_XFLAG_RTINHERIT)
+			di_flags |= XFS_DIFLAG_RTINHERIT;
+		if (xflags & XFS_XFLAG_NOSYMLINKS)
+			di_flags |= XFS_DIFLAG_NOSYMLINKS;
+		if (xflags & XFS_XFLAG_EXTSZINHERIT)
+			di_flags |= XFS_DIFLAG_EXTSZINHERIT;
+	} else if ((ip->i_d.di_mode & S_IFMT) == S_IFREG) {
+		if (xflags & XFS_XFLAG_REALTIME)
+			di_flags |= XFS_DIFLAG_REALTIME;
+		if (xflags & XFS_XFLAG_EXTSIZE)
+			di_flags |= XFS_DIFLAG_EXTSIZE;
+	}
+
+	ip->i_d.di_flags = di_flags;
+}
+
+STATIC void
+xfs_diflags_to_linux(
+	struct xfs_inode	*ip)
+{
+	struct inode		*inode = VFS_I(ip);
+	unsigned int		xflags = xfs_ip2xflags(ip);
+
+	if (xflags & XFS_XFLAG_IMMUTABLE)
+		inode->i_flags |= S_IMMUTABLE;
+	else
+		inode->i_flags &= ~S_IMMUTABLE;
+	if (xflags & XFS_XFLAG_APPEND)
+		inode->i_flags |= S_APPEND;
+	else
+		inode->i_flags &= ~S_APPEND;
+	if (xflags & XFS_XFLAG_SYNC)
+		inode->i_flags |= S_SYNC;
+	else
+		inode->i_flags &= ~S_SYNC;
+	if (xflags & XFS_XFLAG_NOATIME)
+		inode->i_flags |= S_NOATIME;
+	else
+		inode->i_flags &= ~S_NOATIME;
+}
+
+#define FSX_PROJID	1
+#define FSX_EXTSIZE	2
+#define FSX_XFLAGS	4
+#define FSX_NONBLOCK	8
+
+STATIC int
+xfs_ioctl_setattr(
+	xfs_inode_t		*ip,
+	struct fsxattr		*fa,
+	int			mask)
+{
+	struct xfs_mount	*mp = ip->i_mount;
+	struct xfs_trans	*tp;
+	unsigned int		lock_flags = 0;
+	struct xfs_dquot	*udqp = NULL;
+	struct xfs_dquot	*gdqp = NULL;
+	struct xfs_dquot	*olddquot = NULL;
+	int			code;
+
+	trace_xfs_ioctl_setattr(ip);
+
+	if (mp->m_flags & XFS_MOUNT_RDONLY)
+		return XFS_ERROR(EROFS);
+	if (XFS_FORCED_SHUTDOWN(mp))
+		return XFS_ERROR(EIO);
+
+	/*
+	 * Disallow 32bit project ids when projid32bit feature is not enabled.
+	 */
+	if ((mask & FSX_PROJID) && (fa->fsx_projid > (__uint16_t)-1) &&
+			!xfs_sb_version_hasprojid32bit(&ip->i_mount->m_sb))
+		return XFS_ERROR(EINVAL);
+
+	/*
+	 * If disk quotas is on, we make sure that the dquots do exist on disk,
+	 * before we start any other transactions. Trying to do this later
+	 * is messy. We don't care to take a readlock to look at the ids
+	 * in inode here, because we can't hold it across the trans_reserve.
+	 * If the IDs do change before we take the ilock, we're covered
+	 * because the i_*dquot fields will get updated anyway.
+	 */
+	if (XFS_IS_QUOTA_ON(mp) && (mask & FSX_PROJID)) {
+		code = xfs_qm_vop_dqalloc(ip, ip->i_d.di_uid,
+					 ip->i_d.di_gid, fa->fsx_projid,
+					 XFS_QMOPT_PQUOTA, &udqp, &gdqp);
+		if (code)
+			return code;
+	}
+
+	/*
+	 * For the other attributes, we acquire the inode lock and
+	 * first do an error checking pass.
+	 */
+	tp = xfs_trans_alloc(mp, XFS_TRANS_SETATTR_NOT_SIZE);
+	code = xfs_trans_reserve(tp, 0, XFS_ICHANGE_LOG_RES(mp), 0, 0, 0);
+	if (code)
+		goto error_return;
+
+	lock_flags = XFS_ILOCK_EXCL;
+	xfs_ilock(ip, lock_flags);
+
+	/*
+	 * CAP_FOWNER overrides the following restrictions:
+	 *
+	 * The user ID of the calling process must be equal
+	 * to the file owner ID, except in cases where the
+	 * CAP_FSETID capability is applicable.
+	 */
+	if (current_fsuid() != ip->i_d.di_uid && !capable(CAP_FOWNER)) {
+		code = XFS_ERROR(EPERM);
+		goto error_return;
+	}
+
+	/*
+	 * Do a quota reservation only if projid is actually going to change.
+	 */
+	if (mask & FSX_PROJID) {
+		if (XFS_IS_QUOTA_RUNNING(mp) &&
+		    XFS_IS_PQUOTA_ON(mp) &&
+		    xfs_get_projid(ip) != fa->fsx_projid) {
+			ASSERT(tp);
+			code = xfs_qm_vop_chown_reserve(tp, ip, udqp, gdqp,
+						capable(CAP_FOWNER) ?
+						XFS_QMOPT_FORCE_RES : 0);
+			if (code)	/* out of quota */
+				goto error_return;
+		}
+	}
+
+	if (mask & FSX_EXTSIZE) {
+		/*
+		 * Can't change extent size if any extents are allocated.
+		 */
+		if (ip->i_d.di_nextents &&
+		    ((ip->i_d.di_extsize << mp->m_sb.sb_blocklog) !=
+		     fa->fsx_extsize)) {
+			code = XFS_ERROR(EINVAL);	/* EFBIG? */
+			goto error_return;
+		}
+
+		/*
+		 * Extent size must be a multiple of the appropriate block
+		 * size, if set at all. It must also be smaller than the
+		 * maximum extent size supported by the filesystem.
+		 *
+		 * Also, for non-realtime files, limit the extent size hint to
+		 * half the size of the AGs in the filesystem so alignment
+		 * doesn't result in extents larger than an AG.
+		 */
+		if (fa->fsx_extsize != 0) {
+			xfs_extlen_t    size;
+			xfs_fsblock_t   extsize_fsb;
+
+			extsize_fsb = XFS_B_TO_FSB(mp, fa->fsx_extsize);
+			if (extsize_fsb > MAXEXTLEN) {
+				code = XFS_ERROR(EINVAL);
+				goto error_return;
+			}
+
+			if (XFS_IS_REALTIME_INODE(ip) ||
+			    ((mask & FSX_XFLAGS) &&
+			    (fa->fsx_xflags & XFS_XFLAG_REALTIME))) {
+				size = mp->m_sb.sb_rextsize <<
+				       mp->m_sb.sb_blocklog;
+			} else {
+				size = mp->m_sb.sb_blocksize;
+				if (extsize_fsb > mp->m_sb.sb_agblocks / 2) {
+					code = XFS_ERROR(EINVAL);
+					goto error_return;
+				}
+			}
+
+			if (fa->fsx_extsize % size) {
+				code = XFS_ERROR(EINVAL);
+				goto error_return;
+			}
+		}
+	}
+
+
+	if (mask & FSX_XFLAGS) {
+		/*
+		 * Can't change realtime flag if any extents are allocated.
+		 */
+		if ((ip->i_d.di_nextents || ip->i_delayed_blks) &&
+		    (XFS_IS_REALTIME_INODE(ip)) !=
+		    (fa->fsx_xflags & XFS_XFLAG_REALTIME)) {
+			code = XFS_ERROR(EINVAL);	/* EFBIG? */
+			goto error_return;
+		}
+
+		/*
+		 * If realtime flag is set then must have realtime data.
+		 */
+		if ((fa->fsx_xflags & XFS_XFLAG_REALTIME)) {
+			if ((mp->m_sb.sb_rblocks == 0) ||
+			    (mp->m_sb.sb_rextsize == 0) ||
+			    (ip->i_d.di_extsize % mp->m_sb.sb_rextsize)) {
+				code = XFS_ERROR(EINVAL);
+				goto error_return;
+			}
+		}
+
+		/*
+		 * Can't modify an immutable/append-only file unless
+		 * we have appropriate permission.
+		 */
+		if ((ip->i_d.di_flags &
+				(XFS_DIFLAG_IMMUTABLE|XFS_DIFLAG_APPEND) ||
+		     (fa->fsx_xflags &
+				(XFS_XFLAG_IMMUTABLE | XFS_XFLAG_APPEND))) &&
+		    !capable(CAP_LINUX_IMMUTABLE)) {
+			code = XFS_ERROR(EPERM);
+			goto error_return;
+		}
+	}
+
+	xfs_trans_ijoin(tp, ip);
+
+	/*
+	 * Change file ownership.  Must be the owner or privileged.
+	 */
+	if (mask & FSX_PROJID) {
+		/*
+		 * CAP_FSETID overrides the following restrictions:
+		 *
+		 * The set-user-ID and set-group-ID bits of a file will be
+		 * cleared upon successful return from chown()
+		 */
+		if ((ip->i_d.di_mode & (S_ISUID|S_ISGID)) &&
+		    !capable(CAP_FSETID))
+			ip->i_d.di_mode &= ~(S_ISUID|S_ISGID);
+
+		/*
+		 * Change the ownerships and register quota modifications
+		 * in the transaction.
+		 */
+		if (xfs_get_projid(ip) != fa->fsx_projid) {
+			if (XFS_IS_QUOTA_RUNNING(mp) && XFS_IS_PQUOTA_ON(mp)) {
+				olddquot = xfs_qm_vop_chown(tp, ip,
+							&ip->i_gdquot, gdqp);
+			}
+			xfs_set_projid(ip, fa->fsx_projid);
+
+			/*
+			 * We may have to rev the inode as well as
+			 * the superblock version number since projids didn't
+			 * exist before DINODE_VERSION_2 and SB_VERSION_NLINK.
+			 */
+			if (ip->i_d.di_version == 1)
+				xfs_bump_ino_vers2(tp, ip);
+		}
+
+	}
+
+	if (mask & FSX_EXTSIZE)
+		ip->i_d.di_extsize = fa->fsx_extsize >> mp->m_sb.sb_blocklog;
+	if (mask & FSX_XFLAGS) {
+		xfs_set_diflags(ip, fa->fsx_xflags);
+		xfs_diflags_to_linux(ip);
+	}
+
+	xfs_trans_ichgtime(tp, ip, XFS_ICHGTIME_CHG);
+	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
+
+	XFS_STATS_INC(xs_ig_attrchg);
+
+	/*
+	 * If this is a synchronous mount, make sure that the
+	 * transaction goes to disk before returning to the user.
+	 * This is slightly sub-optimal in that truncates require
+	 * two sync transactions instead of one for wsync filesystems.
+	 * One for the truncate and one for the timestamps since we
+	 * don't want to change the timestamps unless we're sure the
+	 * truncate worked.  Truncates are less than 1% of the laddis
+	 * mix so this probably isn't worth the trouble to optimize.
+	 */
+	if (mp->m_flags & XFS_MOUNT_WSYNC)
+		xfs_trans_set_sync(tp);
+	code = xfs_trans_commit(tp, 0);
+	xfs_iunlock(ip, lock_flags);
+
+	/*
+	 * Release any dquot(s) the inode had kept before chown.
+	 */
+	xfs_qm_dqrele(olddquot);
+	xfs_qm_dqrele(udqp);
+	xfs_qm_dqrele(gdqp);
+
+	return code;
+
+ error_return:
+	xfs_qm_dqrele(udqp);
+	xfs_qm_dqrele(gdqp);
+	xfs_trans_cancel(tp, 0);
+	if (lock_flags)
+		xfs_iunlock(ip, lock_flags);
+	return code;
+}
+
+STATIC int
+xfs_ioc_fssetxattr(
+	xfs_inode_t		*ip,
+	struct file		*filp,
+	void			__user *arg)
+{
+	struct fsxattr		fa;
+	unsigned int		mask;
+
+	if (copy_from_user(&fa, arg, sizeof(fa)))
+		return -EFAULT;
+
+	mask = FSX_XFLAGS | FSX_EXTSIZE | FSX_PROJID;
+	if (filp->f_flags & (O_NDELAY|O_NONBLOCK))
+		mask |= FSX_NONBLOCK;
+
+	return -xfs_ioctl_setattr(ip, &fa, mask);
+}
+
+STATIC int
+xfs_ioc_getxflags(
+	xfs_inode_t		*ip,
+	void			__user *arg)
+{
+	unsigned int		flags;
+
+	flags = xfs_di2lxflags(ip->i_d.di_flags);
+	if (copy_to_user(arg, &flags, sizeof(flags)))
+		return -EFAULT;
+	return 0;
+}
+
+STATIC int
+xfs_ioc_setxflags(
+	xfs_inode_t		*ip,
+	struct file		*filp,
+	void			__user *arg)
+{
+	struct fsxattr		fa;
+	unsigned int		flags;
+	unsigned int		mask;
+
+	if (copy_from_user(&flags, arg, sizeof(flags)))
+		return -EFAULT;
+
+	if (flags & ~(FS_IMMUTABLE_FL | FS_APPEND_FL | \
+		      FS_NOATIME_FL | FS_NODUMP_FL | \
+		      FS_SYNC_FL))
+		return -EOPNOTSUPP;
+
+	mask = FSX_XFLAGS;
+	if (filp->f_flags & (O_NDELAY|O_NONBLOCK))
+		mask |= FSX_NONBLOCK;
+	fa.fsx_xflags = xfs_merge_ioc_xflags(flags, xfs_ip2xflags(ip));
+
+	return -xfs_ioctl_setattr(ip, &fa, mask);
+}
+
+STATIC int
+xfs_getbmap_format(void **ap, struct getbmapx *bmv, int *full)
+{
+	struct getbmap __user	*base = *ap;
+
+	/* copy only getbmap portion (not getbmapx) */
+	if (copy_to_user(base, bmv, sizeof(struct getbmap)))
+		return XFS_ERROR(EFAULT);
+
+	*ap += sizeof(struct getbmap);
+	return 0;
+}
+
+STATIC int
+xfs_ioc_getbmap(
+	struct xfs_inode	*ip,
+	int			ioflags,
+	unsigned int		cmd,
+	void			__user *arg)
+{
+	struct getbmapx		bmx;
+	int			error;
+
+	if (copy_from_user(&bmx, arg, sizeof(struct getbmapx)))
+		return -XFS_ERROR(EFAULT);
+
+	if (bmx.bmv_count < 2)
+		return -XFS_ERROR(EINVAL);
+
+	bmx.bmv_iflags = (cmd == XFS_IOC_GETBMAPA ? BMV_IF_ATTRFORK : 0);
+	if (ioflags & IO_INVIS)
+		bmx.bmv_iflags |= BMV_IF_NO_DMAPI_READ;
+
+	error = xfs_getbmap(ip, &bmx, xfs_getbmap_format,
+			    (struct getbmap *)arg+1);
+	if (error)
+		return -error;
+
+	/* copy back header - only size of getbmap */
+	if (copy_to_user(arg, &bmx, sizeof(struct getbmap)))
+		return -XFS_ERROR(EFAULT);
+	return 0;
+}
+
+STATIC int
+xfs_getbmapx_format(void **ap, struct getbmapx *bmv, int *full)
+{
+	struct getbmapx __user	*base = *ap;
+
+	if (copy_to_user(base, bmv, sizeof(struct getbmapx)))
+		return XFS_ERROR(EFAULT);
+
+	*ap += sizeof(struct getbmapx);
+	return 0;
+}
+
+STATIC int
+xfs_ioc_getbmapx(
+	struct xfs_inode	*ip,
+	void			__user *arg)
+{
+	struct getbmapx		bmx;
+	int			error;
+
+	if (copy_from_user(&bmx, arg, sizeof(bmx)))
+		return -XFS_ERROR(EFAULT);
+
+	if (bmx.bmv_count < 2)
+		return -XFS_ERROR(EINVAL);
+
+	if (bmx.bmv_iflags & (~BMV_IF_VALID))
+		return -XFS_ERROR(EINVAL);
+
+	error = xfs_getbmap(ip, &bmx, xfs_getbmapx_format,
+			    (struct getbmapx *)arg+1);
+	if (error)
+		return -error;
+
+	/* copy back header */
+	if (copy_to_user(arg, &bmx, sizeof(struct getbmapx)))
+		return -XFS_ERROR(EFAULT);
+
+	return 0;
+}
+
+/*
+ * Note: some of the ioctl's return positive numbers as a
+ * byte count indicating success, such as readlink_by_handle.
+ * So we don't "sign flip" like most other routines.  This means
+ * true errors need to be returned as a negative value.
+ */
+long
+xfs_file_ioctl(
+	struct file		*filp,
+	unsigned int		cmd,
+	unsigned long		p)
+{
+	struct inode		*inode = filp->f_path.dentry->d_inode;
+	struct xfs_inode	*ip = XFS_I(inode);
+	struct xfs_mount	*mp = ip->i_mount;
+	void			__user *arg = (void __user *)p;
+	int			ioflags = 0;
+	int			error;
+
+	if (filp->f_mode & FMODE_NOCMTIME)
+		ioflags |= IO_INVIS;
+
+	trace_xfs_file_ioctl(ip);
+
+	switch (cmd) {
+	case FITRIM:
+		return xfs_ioc_trim(mp, arg);
+	case XFS_IOC_ALLOCSP:
+	case XFS_IOC_FREESP:
+	case XFS_IOC_RESVSP:
+	case XFS_IOC_UNRESVSP:
+	case XFS_IOC_ALLOCSP64:
+	case XFS_IOC_FREESP64:
+	case XFS_IOC_RESVSP64:
+	case XFS_IOC_UNRESVSP64:
+	case XFS_IOC_ZERO_RANGE: {
+		xfs_flock64_t		bf;
+
+		if (copy_from_user(&bf, arg, sizeof(bf)))
+			return -XFS_ERROR(EFAULT);
+		return xfs_ioc_space(ip, inode, filp, ioflags, cmd, &bf);
+	}
+	case XFS_IOC_DIOINFO: {
+		struct dioattr	da;
+		xfs_buftarg_t	*target =
+			XFS_IS_REALTIME_INODE(ip) ?
+			mp->m_rtdev_targp : mp->m_ddev_targp;
+
+		da.d_mem = da.d_miniosz = 1 << target->bt_sshift;
+		da.d_maxiosz = INT_MAX & ~(da.d_miniosz - 1);
+
+		if (copy_to_user(arg, &da, sizeof(da)))
+			return -XFS_ERROR(EFAULT);
+		return 0;
+	}
+
+	case XFS_IOC_FSBULKSTAT_SINGLE:
+	case XFS_IOC_FSBULKSTAT:
+	case XFS_IOC_FSINUMBERS:
+		return xfs_ioc_bulkstat(mp, cmd, arg);
+
+	case XFS_IOC_FSGEOMETRY_V1:
+		return xfs_ioc_fsgeometry_v1(mp, arg);
+
+	case XFS_IOC_FSGEOMETRY:
+		return xfs_ioc_fsgeometry(mp, arg);
+
+	case XFS_IOC_GETVERSION:
+		return put_user(inode->i_generation, (int __user *)arg);
+
+	case XFS_IOC_FSGETXATTR:
+		return xfs_ioc_fsgetxattr(ip, 0, arg);
+	case XFS_IOC_FSGETXATTRA:
+		return xfs_ioc_fsgetxattr(ip, 1, arg);
+	case XFS_IOC_FSSETXATTR:
+		return xfs_ioc_fssetxattr(ip, filp, arg);
+	case XFS_IOC_GETXFLAGS:
+		return xfs_ioc_getxflags(ip, arg);
+	case XFS_IOC_SETXFLAGS:
+		return xfs_ioc_setxflags(ip, filp, arg);
+
+	case XFS_IOC_FSSETDM: {
+		struct fsdmidata	dmi;
+
+		if (copy_from_user(&dmi, arg, sizeof(dmi)))
+			return -XFS_ERROR(EFAULT);
+
+		error = xfs_set_dmattrs(ip, dmi.fsd_dmevmask,
+				dmi.fsd_dmstate);
+		return -error;
+	}
+
+	case XFS_IOC_GETBMAP:
+	case XFS_IOC_GETBMAPA:
+		return xfs_ioc_getbmap(ip, ioflags, cmd, arg);
+
+	case XFS_IOC_GETBMAPX:
+		return xfs_ioc_getbmapx(ip, arg);
+
+	case XFS_IOC_FD_TO_HANDLE:
+	case XFS_IOC_PATH_TO_HANDLE:
+	case XFS_IOC_PATH_TO_FSHANDLE: {
+		xfs_fsop_handlereq_t	hreq;
+
+		if (copy_from_user(&hreq, arg, sizeof(hreq)))
+			return -XFS_ERROR(EFAULT);
+		return xfs_find_handle(cmd, &hreq);
+	}
+	case XFS_IOC_OPEN_BY_HANDLE: {
+		xfs_fsop_handlereq_t	hreq;
+
+		if (copy_from_user(&hreq, arg, sizeof(xfs_fsop_handlereq_t)))
+			return -XFS_ERROR(EFAULT);
+		return xfs_open_by_handle(filp, &hreq);
+	}
+	case XFS_IOC_FSSETDM_BY_HANDLE:
+		return xfs_fssetdm_by_handle(filp, arg);
+
+	case XFS_IOC_READLINK_BY_HANDLE: {
+		xfs_fsop_handlereq_t	hreq;
+
+		if (copy_from_user(&hreq, arg, sizeof(xfs_fsop_handlereq_t)))
+			return -XFS_ERROR(EFAULT);
+		return xfs_readlink_by_handle(filp, &hreq);
+	}
+	case XFS_IOC_ATTRLIST_BY_HANDLE:
+		return xfs_attrlist_by_handle(filp, arg);
+
+	case XFS_IOC_ATTRMULTI_BY_HANDLE:
+		return xfs_attrmulti_by_handle(filp, arg);
+
+	case XFS_IOC_SWAPEXT: {
+		struct xfs_swapext	sxp;
+
+		if (copy_from_user(&sxp, arg, sizeof(xfs_swapext_t)))
+			return -XFS_ERROR(EFAULT);
+		error = xfs_swapext(&sxp);
+		return -error;
+	}
+
+	case XFS_IOC_FSCOUNTS: {
+		xfs_fsop_counts_t out;
+
+		error = xfs_fs_counts(mp, &out);
+		if (error)
+			return -error;
+
+		if (copy_to_user(arg, &out, sizeof(out)))
+			return -XFS_ERROR(EFAULT);
+		return 0;
+	}
+
+	case XFS_IOC_SET_RESBLKS: {
+		xfs_fsop_resblks_t inout;
+		__uint64_t	   in;
+
+		if (!capable(CAP_SYS_ADMIN))
+			return -EPERM;
+
+		if (mp->m_flags & XFS_MOUNT_RDONLY)
+			return -XFS_ERROR(EROFS);
+
+		if (copy_from_user(&inout, arg, sizeof(inout)))
+			return -XFS_ERROR(EFAULT);
+
+		/* input parameter is passed in resblks field of structure */
+		in = inout.resblks;
+		error = xfs_reserve_blocks(mp, &in, &inout);
+		if (error)
+			return -error;
+
+		if (copy_to_user(arg, &inout, sizeof(inout)))
+			return -XFS_ERROR(EFAULT);
+		return 0;
+	}
+
+	case XFS_IOC_GET_RESBLKS: {
+		xfs_fsop_resblks_t out;
+
+		if (!capable(CAP_SYS_ADMIN))
+			return -EPERM;
+
+		error = xfs_reserve_blocks(mp, NULL, &out);
+		if (error)
+			return -error;
+
+		if (copy_to_user(arg, &out, sizeof(out)))
+			return -XFS_ERROR(EFAULT);
+
+		return 0;
+	}
+
+	case XFS_IOC_FSGROWFSDATA: {
+		xfs_growfs_data_t in;
+
+		if (copy_from_user(&in, arg, sizeof(in)))
+			return -XFS_ERROR(EFAULT);
+
+		error = xfs_growfs_data(mp, &in);
+		return -error;
+	}
+
+	case XFS_IOC_FSGROWFSLOG: {
+		xfs_growfs_log_t in;
+
+		if (copy_from_user(&in, arg, sizeof(in)))
+			return -XFS_ERROR(EFAULT);
+
+		error = xfs_growfs_log(mp, &in);
+		return -error;
+	}
+
+	case XFS_IOC_FSGROWFSRT: {
+		xfs_growfs_rt_t in;
+
+		if (copy_from_user(&in, arg, sizeof(in)))
+			return -XFS_ERROR(EFAULT);
+
+		error = xfs_growfs_rt(mp, &in);
+		return -error;
+	}
+
+	case XFS_IOC_GOINGDOWN: {
+		__uint32_t in;
+
+		if (!capable(CAP_SYS_ADMIN))
+			return -EPERM;
+
+		if (get_user(in, (__uint32_t __user *)arg))
+			return -XFS_ERROR(EFAULT);
+
+		error = xfs_fs_goingdown(mp, in);
+		return -error;
+	}
+
+	case XFS_IOC_ERROR_INJECTION: {
+		xfs_error_injection_t in;
+
+		if (!capable(CAP_SYS_ADMIN))
+			return -EPERM;
+
+		if (copy_from_user(&in, arg, sizeof(in)))
+			return -XFS_ERROR(EFAULT);
+
+		error = xfs_errortag_add(in.errtag, mp);
+		return -error;
+	}
+
+	case XFS_IOC_ERROR_CLEARALL:
+		if (!capable(CAP_SYS_ADMIN))
+			return -EPERM;
+
+		error = xfs_errortag_clearall(mp, 1);
+		return -error;
+
+	default:
+		return -ENOTTY;
+	}
+}
diff --git a/fs/xfs/linux-2.6/xfs_ioctl.h b/fs/xfs/linux-2.6/xfs_ioctl.h
new file mode 100644
index 00000000..d56173b3
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_ioctl.h
@@ -0,0 +1,85 @@
+/*
+ * Copyright (c) 2008 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_IOCTL_H__
+#define __XFS_IOCTL_H__
+
+extern int
+xfs_ioc_space(
+	struct xfs_inode	*ip,
+	struct inode		*inode,
+	struct file		*filp,
+	int			ioflags,
+	unsigned int		cmd,
+	xfs_flock64_t		*bf);
+
+extern int
+xfs_find_handle(
+	unsigned int		cmd,
+	xfs_fsop_handlereq_t	*hreq);
+
+extern int
+xfs_open_by_handle(
+	struct file		*parfilp,
+	xfs_fsop_handlereq_t	*hreq);
+
+extern int
+xfs_readlink_by_handle(
+	struct file		*parfilp,
+	xfs_fsop_handlereq_t	*hreq);
+
+extern int
+xfs_attrmulti_attr_get(
+	struct inode		*inode,
+	unsigned char		*name,
+	unsigned char		__user *ubuf,
+	__uint32_t		*len,
+	__uint32_t		flags);
+
+extern int
+xfs_attrmulti_attr_set(
+	struct inode		*inode,
+	unsigned char		*name,
+	const unsigned char	__user *ubuf,
+	__uint32_t		len,
+	__uint32_t		flags);
+
+extern int
+xfs_attrmulti_attr_remove(
+	struct inode		*inode,
+	unsigned char		*name,
+	__uint32_t		flags);
+
+extern struct dentry *
+xfs_handle_to_dentry(
+	struct file		*parfilp,
+	void __user		*uhandle,
+	u32			hlen);
+
+extern long
+xfs_file_ioctl(
+	struct file		*filp,
+	unsigned int		cmd,
+	unsigned long		p);
+
+extern long
+xfs_file_compat_ioctl(
+	struct file		*file,
+	unsigned int		cmd,
+	unsigned long		arg);
+
+#endif
diff --git a/fs/xfs/linux-2.6/xfs_ioctl32.c b/fs/xfs/linux-2.6/xfs_ioctl32.c
new file mode 100644
index 00000000..54e623bf
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_ioctl32.c
@@ -0,0 +1,672 @@
+/*
+ * Copyright (c) 2004-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include <linux/compat.h>
+#include <linux/ioctl.h>
+#include <linux/mount.h>
+#include <linux/slab.h>
+#include <asm/uaccess.h>
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_mount.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_vnode.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_itable.h"
+#include "xfs_error.h"
+#include "xfs_dfrag.h"
+#include "xfs_vnodeops.h"
+#include "xfs_fsops.h"
+#include "xfs_alloc.h"
+#include "xfs_rtalloc.h"
+#include "xfs_attr.h"
+#include "xfs_ioctl.h"
+#include "xfs_ioctl32.h"
+#include "xfs_trace.h"
+
+#define  _NATIVE_IOC(cmd, type) \
+	  _IOC(_IOC_DIR(cmd), _IOC_TYPE(cmd), _IOC_NR(cmd), sizeof(type))
+
+#ifdef BROKEN_X86_ALIGNMENT
+STATIC int
+xfs_compat_flock64_copyin(
+	xfs_flock64_t		*bf,
+	compat_xfs_flock64_t	__user *arg32)
+{
+	if (get_user(bf->l_type,	&arg32->l_type) ||
+	    get_user(bf->l_whence,	&arg32->l_whence) ||
+	    get_user(bf->l_start,	&arg32->l_start) ||
+	    get_user(bf->l_len,		&arg32->l_len) ||
+	    get_user(bf->l_sysid,	&arg32->l_sysid) ||
+	    get_user(bf->l_pid,		&arg32->l_pid) ||
+	    copy_from_user(bf->l_pad,	&arg32->l_pad,	4*sizeof(u32)))
+		return -XFS_ERROR(EFAULT);
+	return 0;
+}
+
+STATIC int
+xfs_compat_ioc_fsgeometry_v1(
+	struct xfs_mount	  *mp,
+	compat_xfs_fsop_geom_v1_t __user *arg32)
+{
+	xfs_fsop_geom_t		  fsgeo;
+	int			  error;
+
+	error = xfs_fs_geometry(mp, &fsgeo, 3);
+	if (error)
+		return -error;
+	/* The 32-bit variant simply has some padding at the end */
+	if (copy_to_user(arg32, &fsgeo, sizeof(struct compat_xfs_fsop_geom_v1)))
+		return -XFS_ERROR(EFAULT);
+	return 0;
+}
+
+STATIC int
+xfs_compat_growfs_data_copyin(
+	struct xfs_growfs_data	 *in,
+	compat_xfs_growfs_data_t __user *arg32)
+{
+	if (get_user(in->newblocks, &arg32->newblocks) ||
+	    get_user(in->imaxpct,   &arg32->imaxpct))
+		return -XFS_ERROR(EFAULT);
+	return 0;
+}
+
+STATIC int
+xfs_compat_growfs_rt_copyin(
+	struct xfs_growfs_rt	 *in,
+	compat_xfs_growfs_rt_t	__user *arg32)
+{
+	if (get_user(in->newblocks, &arg32->newblocks) ||
+	    get_user(in->extsize,   &arg32->extsize))
+		return -XFS_ERROR(EFAULT);
+	return 0;
+}
+
+STATIC int
+xfs_inumbers_fmt_compat(
+	void			__user *ubuffer,
+	const xfs_inogrp_t	*buffer,
+	long			count,
+	long			*written)
+{
+	compat_xfs_inogrp_t	__user *p32 = ubuffer;
+	long			i;
+
+	for (i = 0; i < count; i++) {
+		if (put_user(buffer[i].xi_startino,   &p32[i].xi_startino) ||
+		    put_user(buffer[i].xi_alloccount, &p32[i].xi_alloccount) ||
+		    put_user(buffer[i].xi_allocmask,  &p32[i].xi_allocmask))
+			return -XFS_ERROR(EFAULT);
+	}
+	*written = count * sizeof(*p32);
+	return 0;
+}
+
+#else
+#define xfs_inumbers_fmt_compat xfs_inumbers_fmt
+#endif	/* BROKEN_X86_ALIGNMENT */
+
+STATIC int
+xfs_ioctl32_bstime_copyin(
+	xfs_bstime_t		*bstime,
+	compat_xfs_bstime_t	__user *bstime32)
+{
+	compat_time_t		sec32;	/* tv_sec differs on 64 vs. 32 */
+
+	if (get_user(sec32,		&bstime32->tv_sec)	||
+	    get_user(bstime->tv_nsec,	&bstime32->tv_nsec))
+		return -XFS_ERROR(EFAULT);
+	bstime->tv_sec = sec32;
+	return 0;
+}
+
+/* xfs_bstat_t has differing alignment on intel, & bstime_t sizes everywhere */
+STATIC int
+xfs_ioctl32_bstat_copyin(
+	xfs_bstat_t		*bstat,
+	compat_xfs_bstat_t	__user *bstat32)
+{
+	if (get_user(bstat->bs_ino,	&bstat32->bs_ino)	||
+	    get_user(bstat->bs_mode,	&bstat32->bs_mode)	||
+	    get_user(bstat->bs_nlink,	&bstat32->bs_nlink)	||
+	    get_user(bstat->bs_uid,	&bstat32->bs_uid)	||
+	    get_user(bstat->bs_gid,	&bstat32->bs_gid)	||
+	    get_user(bstat->bs_rdev,	&bstat32->bs_rdev)	||
+	    get_user(bstat->bs_blksize,	&bstat32->bs_blksize)	||
+	    get_user(bstat->bs_size,	&bstat32->bs_size)	||
+	    xfs_ioctl32_bstime_copyin(&bstat->bs_atime, &bstat32->bs_atime) ||
+	    xfs_ioctl32_bstime_copyin(&bstat->bs_mtime, &bstat32->bs_mtime) ||
+	    xfs_ioctl32_bstime_copyin(&bstat->bs_ctime, &bstat32->bs_ctime) ||
+	    get_user(bstat->bs_blocks,	&bstat32->bs_size)	||
+	    get_user(bstat->bs_xflags,	&bstat32->bs_size)	||
+	    get_user(bstat->bs_extsize,	&bstat32->bs_extsize)	||
+	    get_user(bstat->bs_extents,	&bstat32->bs_extents)	||
+	    get_user(bstat->bs_gen,	&bstat32->bs_gen)	||
+	    get_user(bstat->bs_projid_lo, &bstat32->bs_projid_lo) ||
+	    get_user(bstat->bs_projid_hi, &bstat32->bs_projid_hi) ||
+	    get_user(bstat->bs_dmevmask, &bstat32->bs_dmevmask)	||
+	    get_user(bstat->bs_dmstate,	&bstat32->bs_dmstate)	||
+	    get_user(bstat->bs_aextents, &bstat32->bs_aextents))
+		return -XFS_ERROR(EFAULT);
+	return 0;
+}
+
+/* XFS_IOC_FSBULKSTAT and friends */
+
+STATIC int
+xfs_bstime_store_compat(
+	compat_xfs_bstime_t	__user *p32,
+	const xfs_bstime_t	*p)
+{
+	__s32			sec32;
+
+	sec32 = p->tv_sec;
+	if (put_user(sec32, &p32->tv_sec) ||
+	    put_user(p->tv_nsec, &p32->tv_nsec))
+		return -XFS_ERROR(EFAULT);
+	return 0;
+}
+
+/* Return 0 on success or positive error (to xfs_bulkstat()) */
+STATIC int
+xfs_bulkstat_one_fmt_compat(
+	void			__user *ubuffer,
+	int			ubsize,
+	int			*ubused,
+	const xfs_bstat_t	*buffer)
+{
+	compat_xfs_bstat_t	__user *p32 = ubuffer;
+
+	if (ubsize < sizeof(*p32))
+		return XFS_ERROR(ENOMEM);
+
+	if (put_user(buffer->bs_ino,	  &p32->bs_ino)		||
+	    put_user(buffer->bs_mode,	  &p32->bs_mode)	||
+	    put_user(buffer->bs_nlink,	  &p32->bs_nlink)	||
+	    put_user(buffer->bs_uid,	  &p32->bs_uid)		||
+	    put_user(buffer->bs_gid,	  &p32->bs_gid)		||
+	    put_user(buffer->bs_rdev,	  &p32->bs_rdev)	||
+	    put_user(buffer->bs_blksize,  &p32->bs_blksize)	||
+	    put_user(buffer->bs_size,	  &p32->bs_size)	||
+	    xfs_bstime_store_compat(&p32->bs_atime, &buffer->bs_atime) ||
+	    xfs_bstime_store_compat(&p32->bs_mtime, &buffer->bs_mtime) ||
+	    xfs_bstime_store_compat(&p32->bs_ctime, &buffer->bs_ctime) ||
+	    put_user(buffer->bs_blocks,	  &p32->bs_blocks)	||
+	    put_user(buffer->bs_xflags,	  &p32->bs_xflags)	||
+	    put_user(buffer->bs_extsize,  &p32->bs_extsize)	||
+	    put_user(buffer->bs_extents,  &p32->bs_extents)	||
+	    put_user(buffer->bs_gen,	  &p32->bs_gen)		||
+	    put_user(buffer->bs_projid,	  &p32->bs_projid)	||
+	    put_user(buffer->bs_projid_hi,	&p32->bs_projid_hi)	||
+	    put_user(buffer->bs_dmevmask, &p32->bs_dmevmask)	||
+	    put_user(buffer->bs_dmstate,  &p32->bs_dmstate)	||
+	    put_user(buffer->bs_aextents, &p32->bs_aextents))
+		return XFS_ERROR(EFAULT);
+	if (ubused)
+		*ubused = sizeof(*p32);
+	return 0;
+}
+
+STATIC int
+xfs_bulkstat_one_compat(
+	xfs_mount_t	*mp,		/* mount point for filesystem */
+	xfs_ino_t	ino,		/* inode number to get data for */
+	void		__user *buffer,	/* buffer to place output in */
+	int		ubsize,		/* size of buffer */
+	int		*ubused,	/* bytes used by me */
+	int		*stat)		/* BULKSTAT_RV_... */
+{
+	return xfs_bulkstat_one_int(mp, ino, buffer, ubsize,
+				    xfs_bulkstat_one_fmt_compat,
+				    ubused, stat);
+}
+
+/* copied from xfs_ioctl.c */
+STATIC int
+xfs_compat_ioc_bulkstat(
+	xfs_mount_t		  *mp,
+	unsigned int		  cmd,
+	compat_xfs_fsop_bulkreq_t __user *p32)
+{
+	u32			addr;
+	xfs_fsop_bulkreq_t	bulkreq;
+	int			count;	/* # of records returned */
+	xfs_ino_t		inlast;	/* last inode number */
+	int			done;
+	int			error;
+
+	/* done = 1 if there are more stats to get and if bulkstat */
+	/* should be called again (unused here, but used in dmapi) */
+
+	if (!capable(CAP_SYS_ADMIN))
+		return -XFS_ERROR(EPERM);
+
+	if (XFS_FORCED_SHUTDOWN(mp))
+		return -XFS_ERROR(EIO);
+
+	if (get_user(addr, &p32->lastip))
+		return -XFS_ERROR(EFAULT);
+	bulkreq.lastip = compat_ptr(addr);
+	if (get_user(bulkreq.icount, &p32->icount) ||
+	    get_user(addr, &p32->ubuffer))
+		return -XFS_ERROR(EFAULT);
+	bulkreq.ubuffer = compat_ptr(addr);
+	if (get_user(addr, &p32->ocount))
+		return -XFS_ERROR(EFAULT);
+	bulkreq.ocount = compat_ptr(addr);
+
+	if (copy_from_user(&inlast, bulkreq.lastip, sizeof(__s64)))
+		return -XFS_ERROR(EFAULT);
+
+	if ((count = bulkreq.icount) <= 0)
+		return -XFS_ERROR(EINVAL);
+
+	if (bulkreq.ubuffer == NULL)
+		return -XFS_ERROR(EINVAL);
+
+	if (cmd == XFS_IOC_FSINUMBERS_32) {
+		error = xfs_inumbers(mp, &inlast, &count,
+				bulkreq.ubuffer, xfs_inumbers_fmt_compat);
+	} else if (cmd == XFS_IOC_FSBULKSTAT_SINGLE_32) {
+		int res;
+
+		error = xfs_bulkstat_one_compat(mp, inlast, bulkreq.ubuffer,
+				sizeof(compat_xfs_bstat_t), 0, &res);
+	} else if (cmd == XFS_IOC_FSBULKSTAT_32) {
+		error = xfs_bulkstat(mp, &inlast, &count,
+			xfs_bulkstat_one_compat, sizeof(compat_xfs_bstat_t),
+			bulkreq.ubuffer, &done);
+	} else
+		error = XFS_ERROR(EINVAL);
+	if (error)
+		return -error;
+
+	if (bulkreq.ocount != NULL) {
+		if (copy_to_user(bulkreq.lastip, &inlast,
+						sizeof(xfs_ino_t)))
+			return -XFS_ERROR(EFAULT);
+
+		if (copy_to_user(bulkreq.ocount, &count, sizeof(count)))
+			return -XFS_ERROR(EFAULT);
+	}
+
+	return 0;
+}
+
+STATIC int
+xfs_compat_handlereq_copyin(
+	xfs_fsop_handlereq_t		*hreq,
+	compat_xfs_fsop_handlereq_t	__user *arg32)
+{
+	compat_xfs_fsop_handlereq_t	hreq32;
+
+	if (copy_from_user(&hreq32, arg32, sizeof(compat_xfs_fsop_handlereq_t)))
+		return -XFS_ERROR(EFAULT);
+
+	hreq->fd = hreq32.fd;
+	hreq->path = compat_ptr(hreq32.path);
+	hreq->oflags = hreq32.oflags;
+	hreq->ihandle = compat_ptr(hreq32.ihandle);
+	hreq->ihandlen = hreq32.ihandlen;
+	hreq->ohandle = compat_ptr(hreq32.ohandle);
+	hreq->ohandlen = compat_ptr(hreq32.ohandlen);
+
+	return 0;
+}
+
+STATIC struct dentry *
+xfs_compat_handlereq_to_dentry(
+	struct file		*parfilp,
+	compat_xfs_fsop_handlereq_t *hreq)
+{
+	return xfs_handle_to_dentry(parfilp,
+			compat_ptr(hreq->ihandle), hreq->ihandlen);
+}
+
+STATIC int
+xfs_compat_attrlist_by_handle(
+	struct file		*parfilp,
+	void			__user *arg)
+{
+	int			error;
+	attrlist_cursor_kern_t	*cursor;
+	compat_xfs_fsop_attrlist_handlereq_t al_hreq;
+	struct dentry		*dentry;
+	char			*kbuf;
+
+	if (!capable(CAP_SYS_ADMIN))
+		return -XFS_ERROR(EPERM);
+	if (copy_from_user(&al_hreq, arg,
+			   sizeof(compat_xfs_fsop_attrlist_handlereq_t)))
+		return -XFS_ERROR(EFAULT);
+	if (al_hreq.buflen > XATTR_LIST_MAX)
+		return -XFS_ERROR(EINVAL);
+
+	/*
+	 * Reject flags, only allow namespaces.
+	 */
+	if (al_hreq.flags & ~(ATTR_ROOT | ATTR_SECURE))
+		return -XFS_ERROR(EINVAL);
+
+	dentry = xfs_compat_handlereq_to_dentry(parfilp, &al_hreq.hreq);
+	if (IS_ERR(dentry))
+		return PTR_ERR(dentry);
+
+	error = -ENOMEM;
+	kbuf = kmalloc(al_hreq.buflen, GFP_KERNEL);
+	if (!kbuf)
+		goto out_dput;
+
+	cursor = (attrlist_cursor_kern_t *)&al_hreq.pos;
+	error = -xfs_attr_list(XFS_I(dentry->d_inode), kbuf, al_hreq.buflen,
+					al_hreq.flags, cursor);
+	if (error)
+		goto out_kfree;
+
+	if (copy_to_user(compat_ptr(al_hreq.buffer), kbuf, al_hreq.buflen))
+		error = -EFAULT;
+
+ out_kfree:
+	kfree(kbuf);
+ out_dput:
+	dput(dentry);
+	return error;
+}
+
+STATIC int
+xfs_compat_attrmulti_by_handle(
+	struct file				*parfilp,
+	void					__user *arg)
+{
+	int					error;
+	compat_xfs_attr_multiop_t		*ops;
+	compat_xfs_fsop_attrmulti_handlereq_t	am_hreq;
+	struct dentry				*dentry;
+	unsigned int				i, size;
+	unsigned char				*attr_name;
+
+	if (!capable(CAP_SYS_ADMIN))
+		return -XFS_ERROR(EPERM);
+	if (copy_from_user(&am_hreq, arg,
+			   sizeof(compat_xfs_fsop_attrmulti_handlereq_t)))
+		return -XFS_ERROR(EFAULT);
+
+	/* overflow check */
+	if (am_hreq.opcount >= INT_MAX / sizeof(compat_xfs_attr_multiop_t))
+		return -E2BIG;
+
+	dentry = xfs_compat_handlereq_to_dentry(parfilp, &am_hreq.hreq);
+	if (IS_ERR(dentry))
+		return PTR_ERR(dentry);
+
+	error = E2BIG;
+	size = am_hreq.opcount * sizeof(compat_xfs_attr_multiop_t);
+	if (!size || size > 16 * PAGE_SIZE)
+		goto out_dput;
+
+	ops = memdup_user(compat_ptr(am_hreq.ops), size);
+	if (IS_ERR(ops)) {
+		error = PTR_ERR(ops);
+		goto out_dput;
+	}
+
+	attr_name = kmalloc(MAXNAMELEN, GFP_KERNEL);
+	if (!attr_name)
+		goto out_kfree_ops;
+
+	error = 0;
+	for (i = 0; i < am_hreq.opcount; i++) {
+		ops[i].am_error = strncpy_from_user((char *)attr_name,
+				compat_ptr(ops[i].am_attrname),
+				MAXNAMELEN);
+		if (ops[i].am_error == 0 || ops[i].am_error == MAXNAMELEN)
+			error = -ERANGE;
+		if (ops[i].am_error < 0)
+			break;
+
+		switch (ops[i].am_opcode) {
+		case ATTR_OP_GET:
+			ops[i].am_error = xfs_attrmulti_attr_get(
+					dentry->d_inode, attr_name,
+					compat_ptr(ops[i].am_attrvalue),
+					&ops[i].am_length, ops[i].am_flags);
+			break;
+		case ATTR_OP_SET:
+			ops[i].am_error = mnt_want_write(parfilp->f_path.mnt);
+			if (ops[i].am_error)
+				break;
+			ops[i].am_error = xfs_attrmulti_attr_set(
+					dentry->d_inode, attr_name,
+					compat_ptr(ops[i].am_attrvalue),
+					ops[i].am_length, ops[i].am_flags);
+			mnt_drop_write(parfilp->f_path.mnt);
+			break;
+		case ATTR_OP_REMOVE:
+			ops[i].am_error = mnt_want_write(parfilp->f_path.mnt);
+			if (ops[i].am_error)
+				break;
+			ops[i].am_error = xfs_attrmulti_attr_remove(
+					dentry->d_inode, attr_name,
+					ops[i].am_flags);
+			mnt_drop_write(parfilp->f_path.mnt);
+			break;
+		default:
+			ops[i].am_error = EINVAL;
+		}
+	}
+
+	if (copy_to_user(compat_ptr(am_hreq.ops), ops, size))
+		error = XFS_ERROR(EFAULT);
+
+	kfree(attr_name);
+ out_kfree_ops:
+	kfree(ops);
+ out_dput:
+	dput(dentry);
+	return -error;
+}
+
+STATIC int
+xfs_compat_fssetdm_by_handle(
+	struct file		*parfilp,
+	void			__user *arg)
+{
+	int			error;
+	struct fsdmidata	fsd;
+	compat_xfs_fsop_setdm_handlereq_t dmhreq;
+	struct dentry		*dentry;
+
+	if (!capable(CAP_MKNOD))
+		return -XFS_ERROR(EPERM);
+	if (copy_from_user(&dmhreq, arg,
+			   sizeof(compat_xfs_fsop_setdm_handlereq_t)))
+		return -XFS_ERROR(EFAULT);
+
+	dentry = xfs_compat_handlereq_to_dentry(parfilp, &dmhreq.hreq);
+	if (IS_ERR(dentry))
+		return PTR_ERR(dentry);
+
+	if (IS_IMMUTABLE(dentry->d_inode) || IS_APPEND(dentry->d_inode)) {
+		error = -XFS_ERROR(EPERM);
+		goto out;
+	}
+
+	if (copy_from_user(&fsd, compat_ptr(dmhreq.data), sizeof(fsd))) {
+		error = -XFS_ERROR(EFAULT);
+		goto out;
+	}
+
+	error = -xfs_set_dmattrs(XFS_I(dentry->d_inode), fsd.fsd_dmevmask,
+				 fsd.fsd_dmstate);
+
+out:
+	dput(dentry);
+	return error;
+}
+
+long
+xfs_file_compat_ioctl(
+	struct file		*filp,
+	unsigned		cmd,
+	unsigned long		p)
+{
+	struct inode		*inode = filp->f_path.dentry->d_inode;
+	struct xfs_inode	*ip = XFS_I(inode);
+	struct xfs_mount	*mp = ip->i_mount;
+	void			__user *arg = (void __user *)p;
+	int			ioflags = 0;
+	int			error;
+
+	if (filp->f_mode & FMODE_NOCMTIME)
+		ioflags |= IO_INVIS;
+
+	trace_xfs_file_compat_ioctl(ip);
+
+	switch (cmd) {
+	/* No size or alignment issues on any arch */
+	case XFS_IOC_DIOINFO:
+	case XFS_IOC_FSGEOMETRY:
+	case XFS_IOC_FSGETXATTR:
+	case XFS_IOC_FSSETXATTR:
+	case XFS_IOC_FSGETXATTRA:
+	case XFS_IOC_FSSETDM:
+	case XFS_IOC_GETBMAP:
+	case XFS_IOC_GETBMAPA:
+	case XFS_IOC_GETBMAPX:
+	case XFS_IOC_FSCOUNTS:
+	case XFS_IOC_SET_RESBLKS:
+	case XFS_IOC_GET_RESBLKS:
+	case XFS_IOC_FSGROWFSLOG:
+	case XFS_IOC_GOINGDOWN:
+	case XFS_IOC_ERROR_INJECTION:
+	case XFS_IOC_ERROR_CLEARALL:
+		return xfs_file_ioctl(filp, cmd, p);
+#ifndef BROKEN_X86_ALIGNMENT
+	/* These are handled fine if no alignment issues */
+	case XFS_IOC_ALLOCSP:
+	case XFS_IOC_FREESP:
+	case XFS_IOC_RESVSP:
+	case XFS_IOC_UNRESVSP:
+	case XFS_IOC_ALLOCSP64:
+	case XFS_IOC_FREESP64:
+	case XFS_IOC_RESVSP64:
+	case XFS_IOC_UNRESVSP64:
+	case XFS_IOC_FSGEOMETRY_V1:
+	case XFS_IOC_FSGROWFSDATA:
+	case XFS_IOC_FSGROWFSRT:
+	case XFS_IOC_ZERO_RANGE:
+		return xfs_file_ioctl(filp, cmd, p);
+#else
+	case XFS_IOC_ALLOCSP_32:
+	case XFS_IOC_FREESP_32:
+	case XFS_IOC_ALLOCSP64_32:
+	case XFS_IOC_FREESP64_32:
+	case XFS_IOC_RESVSP_32:
+	case XFS_IOC_UNRESVSP_32:
+	case XFS_IOC_RESVSP64_32:
+	case XFS_IOC_UNRESVSP64_32:
+	case XFS_IOC_ZERO_RANGE_32: {
+		struct xfs_flock64	bf;
+
+		if (xfs_compat_flock64_copyin(&bf, arg))
+			return -XFS_ERROR(EFAULT);
+		cmd = _NATIVE_IOC(cmd, struct xfs_flock64);
+		return xfs_ioc_space(ip, inode, filp, ioflags, cmd, &bf);
+	}
+	case XFS_IOC_FSGEOMETRY_V1_32:
+		return xfs_compat_ioc_fsgeometry_v1(mp, arg);
+	case XFS_IOC_FSGROWFSDATA_32: {
+		struct xfs_growfs_data	in;
+
+		if (xfs_compat_growfs_data_copyin(&in, arg))
+			return -XFS_ERROR(EFAULT);
+		error = xfs_growfs_data(mp, &in);
+		return -error;
+	}
+	case XFS_IOC_FSGROWFSRT_32: {
+		struct xfs_growfs_rt	in;
+
+		if (xfs_compat_growfs_rt_copyin(&in, arg))
+			return -XFS_ERROR(EFAULT);
+		error = xfs_growfs_rt(mp, &in);
+		return -error;
+	}
+#endif
+	/* long changes size, but xfs only copiese out 32 bits */
+	case XFS_IOC_GETXFLAGS_32:
+	case XFS_IOC_SETXFLAGS_32:
+	case XFS_IOC_GETVERSION_32:
+		cmd = _NATIVE_IOC(cmd, long);
+		return xfs_file_ioctl(filp, cmd, p);
+	case XFS_IOC_SWAPEXT_32: {
+		struct xfs_swapext	  sxp;
+		struct compat_xfs_swapext __user *sxu = arg;
+
+		/* Bulk copy in up to the sx_stat field, then copy bstat */
+		if (copy_from_user(&sxp, sxu,
+				   offsetof(struct xfs_swapext, sx_stat)) ||
+		    xfs_ioctl32_bstat_copyin(&sxp.sx_stat, &sxu->sx_stat))
+			return -XFS_ERROR(EFAULT);
+		error = xfs_swapext(&sxp);
+		return -error;
+	}
+	case XFS_IOC_FSBULKSTAT_32:
+	case XFS_IOC_FSBULKSTAT_SINGLE_32:
+	case XFS_IOC_FSINUMBERS_32:
+		return xfs_compat_ioc_bulkstat(mp, cmd, arg);
+	case XFS_IOC_FD_TO_HANDLE_32:
+	case XFS_IOC_PATH_TO_HANDLE_32:
+	case XFS_IOC_PATH_TO_FSHANDLE_32: {
+		struct xfs_fsop_handlereq	hreq;
+
+		if (xfs_compat_handlereq_copyin(&hreq, arg))
+			return -XFS_ERROR(EFAULT);
+		cmd = _NATIVE_IOC(cmd, struct xfs_fsop_handlereq);
+		return xfs_find_handle(cmd, &hreq);
+	}
+	case XFS_IOC_OPEN_BY_HANDLE_32: {
+		struct xfs_fsop_handlereq	hreq;
+
+		if (xfs_compat_handlereq_copyin(&hreq, arg))
+			return -XFS_ERROR(EFAULT);
+		return xfs_open_by_handle(filp, &hreq);
+	}
+	case XFS_IOC_READLINK_BY_HANDLE_32: {
+		struct xfs_fsop_handlereq	hreq;
+
+		if (xfs_compat_handlereq_copyin(&hreq, arg))
+			return -XFS_ERROR(EFAULT);
+		return xfs_readlink_by_handle(filp, &hreq);
+	}
+	case XFS_IOC_ATTRLIST_BY_HANDLE_32:
+		return xfs_compat_attrlist_by_handle(filp, arg);
+	case XFS_IOC_ATTRMULTI_BY_HANDLE_32:
+		return xfs_compat_attrmulti_by_handle(filp, arg);
+	case XFS_IOC_FSSETDM_BY_HANDLE_32:
+		return xfs_compat_fssetdm_by_handle(filp, arg);
+	default:
+		return -XFS_ERROR(ENOIOCTLCMD);
+	}
+}
diff --git a/fs/xfs/linux-2.6/xfs_ioctl32.h b/fs/xfs/linux-2.6/xfs_ioctl32.h
new file mode 100644
index 00000000..80f4060e
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_ioctl32.h
@@ -0,0 +1,237 @@
+/*
+ * Copyright (c) 2004-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_IOCTL32_H__
+#define __XFS_IOCTL32_H__
+
+#include <linux/compat.h>
+
+/*
+ * on 32-bit arches, ioctl argument structures may have different sizes
+ * and/or alignment.  We define compat structures which match the
+ * 32-bit sizes/alignments here, and their associated ioctl numbers.
+ *
+ * xfs_ioctl32.c contains routines to copy these structures in and out.
+ */
+
+/* stock kernel-level ioctls we support */
+#define XFS_IOC_GETXFLAGS_32	FS_IOC32_GETFLAGS
+#define XFS_IOC_SETXFLAGS_32	FS_IOC32_SETFLAGS
+#define XFS_IOC_GETVERSION_32	FS_IOC32_GETVERSION
+
+/*
+ * On intel, even if sizes match, alignment and/or padding may differ.
+ */
+#if defined(CONFIG_IA64) || defined(CONFIG_X86_64)
+#define BROKEN_X86_ALIGNMENT
+#define __compat_packed __attribute__((packed))
+#else
+#define __compat_packed
+#endif
+
+typedef struct compat_xfs_bstime {
+	compat_time_t	tv_sec;		/* seconds		*/
+	__s32		tv_nsec;	/* and nanoseconds	*/
+} compat_xfs_bstime_t;
+
+typedef struct compat_xfs_bstat {
+	__u64		bs_ino;		/* inode number			*/
+	__u16		bs_mode;	/* type and mode		*/
+	__u16		bs_nlink;	/* number of links		*/
+	__u32		bs_uid;		/* user id			*/
+	__u32		bs_gid;		/* group id			*/
+	__u32		bs_rdev;	/* device value			*/
+	__s32		bs_blksize;	/* block size			*/
+	__s64		bs_size;	/* file size			*/
+	compat_xfs_bstime_t bs_atime;	/* access time			*/
+	compat_xfs_bstime_t bs_mtime;	/* modify time			*/
+	compat_xfs_bstime_t bs_ctime;	/* inode change time		*/
+	int64_t		bs_blocks;	/* number of blocks		*/
+	__u32		bs_xflags;	/* extended flags		*/
+	__s32		bs_extsize;	/* extent size			*/
+	__s32		bs_extents;	/* number of extents		*/
+	__u32		bs_gen;		/* generation count		*/
+	__u16		bs_projid_lo;	/* lower part of project id	*/
+#define	bs_projid	bs_projid_lo	/* (previously just bs_projid)	*/
+	__u16		bs_projid_hi;	/* high part of project id	*/
+	unsigned char	bs_pad[12];	/* pad space, unused		*/
+	__u32		bs_dmevmask;	/* DMIG event mask		*/
+	__u16		bs_dmstate;	/* DMIG state info		*/
+	__u16		bs_aextents;	/* attribute number of extents	*/
+} __compat_packed compat_xfs_bstat_t;
+
+typedef struct compat_xfs_fsop_bulkreq {
+	compat_uptr_t	lastip;		/* last inode # pointer		*/
+	__s32		icount;		/* count of entries in buffer	*/
+	compat_uptr_t	ubuffer;	/* user buffer for inode desc.	*/
+	compat_uptr_t	ocount;		/* output count pointer		*/
+} compat_xfs_fsop_bulkreq_t;
+
+#define XFS_IOC_FSBULKSTAT_32 \
+	_IOWR('X', 101, struct compat_xfs_fsop_bulkreq)
+#define XFS_IOC_FSBULKSTAT_SINGLE_32 \
+	_IOWR('X', 102, struct compat_xfs_fsop_bulkreq)
+#define XFS_IOC_FSINUMBERS_32 \
+	_IOWR('X', 103, struct compat_xfs_fsop_bulkreq)
+
+typedef struct compat_xfs_fsop_handlereq {
+	__u32		fd;		/* fd for FD_TO_HANDLE		*/
+	compat_uptr_t	path;		/* user pathname		*/
+	__u32		oflags;		/* open flags			*/
+	compat_uptr_t	ihandle;	/* user supplied handle		*/
+	__u32		ihandlen;	/* user supplied length		*/
+	compat_uptr_t	ohandle;	/* user buffer for handle	*/
+	compat_uptr_t	ohandlen;	/* user buffer length		*/
+} compat_xfs_fsop_handlereq_t;
+
+#define XFS_IOC_PATH_TO_FSHANDLE_32 \
+	_IOWR('X', 104, struct compat_xfs_fsop_handlereq)
+#define XFS_IOC_PATH_TO_HANDLE_32 \
+	_IOWR('X', 105, struct compat_xfs_fsop_handlereq)
+#define XFS_IOC_FD_TO_HANDLE_32 \
+	_IOWR('X', 106, struct compat_xfs_fsop_handlereq)
+#define XFS_IOC_OPEN_BY_HANDLE_32 \
+	_IOWR('X', 107, struct compat_xfs_fsop_handlereq)
+#define XFS_IOC_READLINK_BY_HANDLE_32 \
+	_IOWR('X', 108, struct compat_xfs_fsop_handlereq)
+
+/* The bstat field in the swapext struct needs translation */
+typedef struct compat_xfs_swapext {
+	__int64_t		sx_version;	/* version */
+	__int64_t		sx_fdtarget;	/* fd of target file */
+	__int64_t		sx_fdtmp;	/* fd of tmp file */
+	xfs_off_t		sx_offset;	/* offset into file */
+	xfs_off_t		sx_length;	/* leng from offset */
+	char			sx_pad[16];	/* pad space, unused */
+	compat_xfs_bstat_t	sx_stat;	/* stat of target b4 copy */
+} __compat_packed compat_xfs_swapext_t;
+
+#define XFS_IOC_SWAPEXT_32	_IOWR('X', 109, struct compat_xfs_swapext)
+
+typedef struct compat_xfs_fsop_attrlist_handlereq {
+	struct compat_xfs_fsop_handlereq hreq; /* handle interface structure */
+	struct xfs_attrlist_cursor	pos; /* opaque cookie, list offset */
+	__u32				flags;	/* which namespace to use */
+	__u32				buflen;	/* length of buffer supplied */
+	compat_uptr_t			buffer;	/* returned names */
+} __compat_packed compat_xfs_fsop_attrlist_handlereq_t;
+
+/* Note: actually this is read/write */
+#define XFS_IOC_ATTRLIST_BY_HANDLE_32 \
+	_IOW('X', 122, struct compat_xfs_fsop_attrlist_handlereq)
+
+/* am_opcodes defined in xfs_fs.h */
+typedef struct compat_xfs_attr_multiop {
+	__u32		am_opcode;
+	__s32		am_error;
+	compat_uptr_t	am_attrname;
+	compat_uptr_t	am_attrvalue;
+	__u32		am_length;
+	__u32		am_flags;
+} compat_xfs_attr_multiop_t;
+
+typedef struct compat_xfs_fsop_attrmulti_handlereq {
+	struct compat_xfs_fsop_handlereq hreq; /* handle interface structure */
+	__u32				opcount;/* count of following multiop */
+	/* ptr to compat_xfs_attr_multiop */
+	compat_uptr_t			ops; /* attr_multi data */
+} compat_xfs_fsop_attrmulti_handlereq_t;
+
+#define XFS_IOC_ATTRMULTI_BY_HANDLE_32 \
+	_IOW('X', 123, struct compat_xfs_fsop_attrmulti_handlereq)
+
+typedef struct compat_xfs_fsop_setdm_handlereq {
+	struct compat_xfs_fsop_handlereq hreq;	/* handle information   */
+	/* ptr to struct fsdmidata */
+	compat_uptr_t			data;	/* DMAPI data   */
+} compat_xfs_fsop_setdm_handlereq_t;
+
+#define XFS_IOC_FSSETDM_BY_HANDLE_32 \
+	_IOW('X', 121, struct compat_xfs_fsop_setdm_handlereq)
+
+#ifdef BROKEN_X86_ALIGNMENT
+/* on ia32 l_start is on a 32-bit boundary */
+typedef struct compat_xfs_flock64 {
+	__s16		l_type;
+	__s16		l_whence;
+	__s64		l_start	__attribute__((packed));
+			/* len == 0 means until end of file */
+	__s64		l_len __attribute__((packed));
+	__s32		l_sysid;
+	__u32		l_pid;
+	__s32		l_pad[4];	/* reserve area */
+} compat_xfs_flock64_t;
+
+#define XFS_IOC_ALLOCSP_32	_IOW('X', 10, struct compat_xfs_flock64)
+#define XFS_IOC_FREESP_32	_IOW('X', 11, struct compat_xfs_flock64)
+#define XFS_IOC_ALLOCSP64_32	_IOW('X', 36, struct compat_xfs_flock64)
+#define XFS_IOC_FREESP64_32	_IOW('X', 37, struct compat_xfs_flock64)
+#define XFS_IOC_RESVSP_32	_IOW('X', 40, struct compat_xfs_flock64)
+#define XFS_IOC_UNRESVSP_32	_IOW('X', 41, struct compat_xfs_flock64)
+#define XFS_IOC_RESVSP64_32	_IOW('X', 42, struct compat_xfs_flock64)
+#define XFS_IOC_UNRESVSP64_32	_IOW('X', 43, struct compat_xfs_flock64)
+#define XFS_IOC_ZERO_RANGE_32	_IOW('X', 57, struct compat_xfs_flock64)
+
+typedef struct compat_xfs_fsop_geom_v1 {
+	__u32		blocksize;	/* filesystem (data) block size */
+	__u32		rtextsize;	/* realtime extent size		*/
+	__u32		agblocks;	/* fsblocks in an AG		*/
+	__u32		agcount;	/* number of allocation groups	*/
+	__u32		logblocks;	/* fsblocks in the log		*/
+	__u32		sectsize;	/* (data) sector size, bytes	*/
+	__u32		inodesize;	/* inode size in bytes		*/
+	__u32		imaxpct;	/* max allowed inode space(%)	*/
+	__u64		datablocks;	/* fsblocks in data subvolume	*/
+	__u64		rtblocks;	/* fsblocks in realtime subvol	*/
+	__u64		rtextents;	/* rt extents in realtime subvol*/
+	__u64		logstart;	/* starting fsblock of the log	*/
+	unsigned char	uuid[16];	/* unique id of the filesystem	*/
+	__u32		sunit;		/* stripe unit, fsblocks	*/
+	__u32		swidth;		/* stripe width, fsblocks	*/
+	__s32		version;	/* structure version		*/
+	__u32		flags;		/* superblock version flags	*/
+	__u32		logsectsize;	/* log sector size, bytes	*/
+	__u32		rtsectsize;	/* realtime sector size, bytes	*/
+	__u32		dirblocksize;	/* directory block size, bytes	*/
+} __attribute__((packed)) compat_xfs_fsop_geom_v1_t;
+
+#define XFS_IOC_FSGEOMETRY_V1_32  \
+	_IOR('X', 100, struct compat_xfs_fsop_geom_v1)
+
+typedef struct compat_xfs_inogrp {
+	__u64		xi_startino;	/* starting inode number	*/
+	__s32		xi_alloccount;	/* # bits set in allocmask	*/
+	__u64		xi_allocmask;	/* mask of allocated inodes	*/
+} __attribute__((packed)) compat_xfs_inogrp_t;
+
+/* These growfs input structures have padding on the end, so must translate */
+typedef struct compat_xfs_growfs_data {
+	__u64		newblocks;	/* new data subvol size, fsblocks */
+	__u32		imaxpct;	/* new inode space percentage limit */
+} __attribute__((packed)) compat_xfs_growfs_data_t;
+
+typedef struct compat_xfs_growfs_rt {
+	__u64		newblocks;	/* new realtime size, fsblocks */
+	__u32		extsize;	/* new realtime extent size, fsblocks */
+} __attribute__((packed)) compat_xfs_growfs_rt_t;
+
+#define XFS_IOC_FSGROWFSDATA_32 _IOW('X', 110, struct compat_xfs_growfs_data)
+#define XFS_IOC_FSGROWFSRT_32   _IOW('X', 112, struct compat_xfs_growfs_rt)
+
+#endif /* BROKEN_X86_ALIGNMENT */
+
+#endif /* __XFS_IOCTL32_H__ */
diff --git a/fs/xfs/linux-2.6/xfs_iops.c b/fs/xfs/linux-2.6/xfs_iops.c
new file mode 100644
index 00000000..f5b697bf
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_iops.c
@@ -0,0 +1,778 @@
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_acl.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_alloc.h"
+#include "xfs_quota.h"
+#include "xfs_mount.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_bmap.h"
+#include "xfs_rtalloc.h"
+#include "xfs_error.h"
+#include "xfs_itable.h"
+#include "xfs_rw.h"
+#include "xfs_attr.h"
+#include "xfs_buf_item.h"
+#include "xfs_utils.h"
+#include "xfs_vnodeops.h"
+#include "xfs_trace.h"
+
+#include <linux/capability.h>
+#include <linux/xattr.h>
+#include <linux/namei.h>
+#include <linux/posix_acl.h>
+#include <linux/security.h>
+#include <linux/fiemap.h>
+#include <linux/slab.h>
+
+/*
+ * Bring the timestamps in the XFS inode uptodate.
+ *
+ * Used before writing the inode to disk.
+ */
+void
+xfs_synchronize_times(
+	xfs_inode_t	*ip)
+{
+	struct inode	*inode = VFS_I(ip);
+
+	ip->i_d.di_atime.t_sec = (__int32_t)inode->i_atime.tv_sec;
+	ip->i_d.di_atime.t_nsec = (__int32_t)inode->i_atime.tv_nsec;
+	ip->i_d.di_ctime.t_sec = (__int32_t)inode->i_ctime.tv_sec;
+	ip->i_d.di_ctime.t_nsec = (__int32_t)inode->i_ctime.tv_nsec;
+	ip->i_d.di_mtime.t_sec = (__int32_t)inode->i_mtime.tv_sec;
+	ip->i_d.di_mtime.t_nsec = (__int32_t)inode->i_mtime.tv_nsec;
+}
+
+/*
+ * If the linux inode is valid, mark it dirty, else mark the dirty state
+ * in the XFS inode to make sure we pick it up when reclaiming the inode.
+ */
+void
+xfs_mark_inode_dirty_sync(
+	xfs_inode_t	*ip)
+{
+	struct inode	*inode = VFS_I(ip);
+
+	if (!(inode->i_state & (I_WILL_FREE|I_FREEING)))
+		mark_inode_dirty_sync(inode);
+	else {
+		barrier();
+		ip->i_update_core = 1;
+	}
+}
+
+void
+xfs_mark_inode_dirty(
+	xfs_inode_t	*ip)
+{
+	struct inode	*inode = VFS_I(ip);
+
+	if (!(inode->i_state & (I_WILL_FREE|I_FREEING)))
+		mark_inode_dirty(inode);
+	else {
+		barrier();
+		ip->i_update_core = 1;
+	}
+
+}
+
+/*
+ * Hook in SELinux.  This is not quite correct yet, what we really need
+ * here (as we do for default ACLs) is a mechanism by which creation of
+ * these attrs can be journalled at inode creation time (along with the
+ * inode, of course, such that log replay can't cause these to be lost).
+ */
+STATIC int
+xfs_init_security(
+	struct inode	*inode,
+	struct inode	*dir,
+	const struct qstr *qstr)
+{
+	struct xfs_inode *ip = XFS_I(inode);
+	size_t		length;
+	void		*value;
+	unsigned char	*name;
+	int		error;
+
+	error = security_inode_init_security(inode, dir, qstr, (char **)&name,
+					     &value, &length);
+	if (error) {
+		if (error == -EOPNOTSUPP)
+			return 0;
+		return -error;
+	}
+
+	error = xfs_attr_set(ip, name, value, length, ATTR_SECURE);
+
+	kfree(name);
+	kfree(value);
+	return error;
+}
+
+static void
+xfs_dentry_to_name(
+	struct xfs_name	*namep,
+	struct dentry	*dentry)
+{
+	namep->name = dentry->d_name.name;
+	namep->len = dentry->d_name.len;
+}
+
+STATIC void
+xfs_cleanup_inode(
+	struct inode	*dir,
+	struct inode	*inode,
+	struct dentry	*dentry)
+{
+	struct xfs_name	teardown;
+
+	/* Oh, the horror.
+	 * If we can't add the ACL or we fail in
+	 * xfs_init_security we must back out.
+	 * ENOSPC can hit here, among other things.
+	 */
+	xfs_dentry_to_name(&teardown, dentry);
+
+	xfs_remove(XFS_I(dir), &teardown, XFS_I(inode));
+	iput(inode);
+}
+
+STATIC int
+xfs_vn_mknod(
+	struct inode	*dir,
+	struct dentry	*dentry,
+	int		mode,
+	dev_t		rdev)
+{
+	struct inode	*inode;
+	struct xfs_inode *ip = NULL;
+	struct posix_acl *default_acl = NULL;
+	struct xfs_name	name;
+	int		error;
+
+	/*
+	 * Irix uses Missed'em'V split, but doesn't want to see
+	 * the upper 5 bits of (14bit) major.
+	 */
+	if (S_ISCHR(mode) || S_ISBLK(mode)) {
+		if (unlikely(!sysv_valid_dev(rdev) || MAJOR(rdev) & ~0x1ff))
+			return -EINVAL;
+		rdev = sysv_encode_dev(rdev);
+	} else {
+		rdev = 0;
+	}
+
+	if (IS_POSIXACL(dir)) {
+		default_acl = xfs_get_acl(dir, ACL_TYPE_DEFAULT);
+		if (IS_ERR(default_acl))
+			return PTR_ERR(default_acl);
+
+		if (!default_acl)
+			mode &= ~current_umask();
+	}
+
+	xfs_dentry_to_name(&name, dentry);
+	error = xfs_create(XFS_I(dir), &name, mode, rdev, &ip);
+	if (unlikely(error))
+		goto out_free_acl;
+
+	inode = VFS_I(ip);
+
+	error = xfs_init_security(inode, dir, &dentry->d_name);
+	if (unlikely(error))
+		goto out_cleanup_inode;
+
+	if (default_acl) {
+		error = -xfs_inherit_acl(inode, default_acl);
+		if (unlikely(error))
+			goto out_cleanup_inode;
+		posix_acl_release(default_acl);
+	}
+
+
+	d_instantiate(dentry, inode);
+	return -error;
+
+ out_cleanup_inode:
+	xfs_cleanup_inode(dir, inode, dentry);
+ out_free_acl:
+	posix_acl_release(default_acl);
+	return -error;
+}
+
+STATIC int
+xfs_vn_create(
+	struct inode	*dir,
+	struct dentry	*dentry,
+	int		mode,
+	struct nameidata *nd)
+{
+	return xfs_vn_mknod(dir, dentry, mode, 0);
+}
+
+STATIC int
+xfs_vn_mkdir(
+	struct inode	*dir,
+	struct dentry	*dentry,
+	int		mode)
+{
+	return xfs_vn_mknod(dir, dentry, mode|S_IFDIR, 0);
+}
+
+STATIC struct dentry *
+xfs_vn_lookup(
+	struct inode	*dir,
+	struct dentry	*dentry,
+	struct nameidata *nd)
+{
+	struct xfs_inode *cip;
+	struct xfs_name	name;
+	int		error;
+
+	if (dentry->d_name.len >= MAXNAMELEN)
+		return ERR_PTR(-ENAMETOOLONG);
+
+	xfs_dentry_to_name(&name, dentry);
+	error = xfs_lookup(XFS_I(dir), &name, &cip, NULL);
+	if (unlikely(error)) {
+		if (unlikely(error != ENOENT))
+			return ERR_PTR(-error);
+		d_add(dentry, NULL);
+		return NULL;
+	}
+
+	return d_splice_alias(VFS_I(cip), dentry);
+}
+
+STATIC struct dentry *
+xfs_vn_ci_lookup(
+	struct inode	*dir,
+	struct dentry	*dentry,
+	struct nameidata *nd)
+{
+	struct xfs_inode *ip;
+	struct xfs_name	xname;
+	struct xfs_name ci_name;
+	struct qstr	dname;
+	int		error;
+
+	if (dentry->d_name.len >= MAXNAMELEN)
+		return ERR_PTR(-ENAMETOOLONG);
+
+	xfs_dentry_to_name(&xname, dentry);
+	error = xfs_lookup(XFS_I(dir), &xname, &ip, &ci_name);
+	if (unlikely(error)) {
+		if (unlikely(error != ENOENT))
+			return ERR_PTR(-error);
+		/*
+		 * call d_add(dentry, NULL) here when d_drop_negative_children
+		 * is called in xfs_vn_mknod (ie. allow negative dentries
+		 * with CI filesystems).
+		 */
+		return NULL;
+	}
+
+	/* if exact match, just splice and exit */
+	if (!ci_name.name)
+		return d_splice_alias(VFS_I(ip), dentry);
+
+	/* else case-insensitive match... */
+	dname.name = ci_name.name;
+	dname.len = ci_name.len;
+	dentry = d_add_ci(dentry, VFS_I(ip), &dname);
+	kmem_free(ci_name.name);
+	return dentry;
+}
+
+STATIC int
+xfs_vn_link(
+	struct dentry	*old_dentry,
+	struct inode	*dir,
+	struct dentry	*dentry)
+{
+	struct inode	*inode = old_dentry->d_inode;
+	struct xfs_name	name;
+	int		error;
+
+	xfs_dentry_to_name(&name, dentry);
+
+	error = xfs_link(XFS_I(dir), XFS_I(inode), &name);
+	if (unlikely(error))
+		return -error;
+
+	ihold(inode);
+	d_instantiate(dentry, inode);
+	return 0;
+}
+
+STATIC int
+xfs_vn_unlink(
+	struct inode	*dir,
+	struct dentry	*dentry)
+{
+	struct xfs_name	name;
+	int		error;
+
+	xfs_dentry_to_name(&name, dentry);
+
+	error = -xfs_remove(XFS_I(dir), &name, XFS_I(dentry->d_inode));
+	if (error)
+		return error;
+
+	/*
+	 * With unlink, the VFS makes the dentry "negative": no inode,
+	 * but still hashed. This is incompatible with case-insensitive
+	 * mode, so invalidate (unhash) the dentry in CI-mode.
+	 */
+	if (xfs_sb_version_hasasciici(&XFS_M(dir->i_sb)->m_sb))
+		d_invalidate(dentry);
+	return 0;
+}
+
+STATIC int
+xfs_vn_symlink(
+	struct inode	*dir,
+	struct dentry	*dentry,
+	const char	*symname)
+{
+	struct inode	*inode;
+	struct xfs_inode *cip = NULL;
+	struct xfs_name	name;
+	int		error;
+	mode_t		mode;
+
+	mode = S_IFLNK |
+		(irix_symlink_mode ? 0777 & ~current_umask() : S_IRWXUGO);
+	xfs_dentry_to_name(&name, dentry);
+
+	error = xfs_symlink(XFS_I(dir), &name, symname, mode, &cip);
+	if (unlikely(error))
+		goto out;
+
+	inode = VFS_I(cip);
+
+	error = xfs_init_security(inode, dir, &dentry->d_name);
+	if (unlikely(error))
+		goto out_cleanup_inode;
+
+	d_instantiate(dentry, inode);
+	return 0;
+
+ out_cleanup_inode:
+	xfs_cleanup_inode(dir, inode, dentry);
+ out:
+	return -error;
+}
+
+STATIC int
+xfs_vn_rename(
+	struct inode	*odir,
+	struct dentry	*odentry,
+	struct inode	*ndir,
+	struct dentry	*ndentry)
+{
+	struct inode	*new_inode = ndentry->d_inode;
+	struct xfs_name	oname;
+	struct xfs_name	nname;
+
+	xfs_dentry_to_name(&oname, odentry);
+	xfs_dentry_to_name(&nname, ndentry);
+
+	return -xfs_rename(XFS_I(odir), &oname, XFS_I(odentry->d_inode),
+			   XFS_I(ndir), &nname, new_inode ?
+			   			XFS_I(new_inode) : NULL);
+}
+
+/*
+ * careful here - this function can get called recursively, so
+ * we need to be very careful about how much stack we use.
+ * uio is kmalloced for this reason...
+ */
+STATIC void *
+xfs_vn_follow_link(
+	struct dentry		*dentry,
+	struct nameidata	*nd)
+{
+	char			*link;
+	int			error = -ENOMEM;
+
+	link = kmalloc(MAXPATHLEN+1, GFP_KERNEL);
+	if (!link)
+		goto out_err;
+
+	error = -xfs_readlink(XFS_I(dentry->d_inode), link);
+	if (unlikely(error))
+		goto out_kfree;
+
+	nd_set_link(nd, link);
+	return NULL;
+
+ out_kfree:
+	kfree(link);
+ out_err:
+	nd_set_link(nd, ERR_PTR(error));
+	return NULL;
+}
+
+STATIC void
+xfs_vn_put_link(
+	struct dentry	*dentry,
+	struct nameidata *nd,
+	void		*p)
+{
+	char		*s = nd_get_link(nd);
+
+	if (!IS_ERR(s))
+		kfree(s);
+}
+
+STATIC int
+xfs_vn_getattr(
+	struct vfsmount		*mnt,
+	struct dentry		*dentry,
+	struct kstat		*stat)
+{
+	struct inode		*inode = dentry->d_inode;
+	struct xfs_inode	*ip = XFS_I(inode);
+	struct xfs_mount	*mp = ip->i_mount;
+
+	trace_xfs_getattr(ip);
+
+	if (XFS_FORCED_SHUTDOWN(mp))
+		return -XFS_ERROR(EIO);
+
+	stat->size = XFS_ISIZE(ip);
+	stat->dev = inode->i_sb->s_dev;
+	stat->mode = ip->i_d.di_mode;
+	stat->nlink = ip->i_d.di_nlink;
+	stat->uid = ip->i_d.di_uid;
+	stat->gid = ip->i_d.di_gid;
+	stat->ino = ip->i_ino;
+	stat->atime = inode->i_atime;
+	stat->mtime = inode->i_mtime;
+	stat->ctime = inode->i_ctime;
+	stat->blocks =
+		XFS_FSB_TO_BB(mp, ip->i_d.di_nblocks + ip->i_delayed_blks);
+
+
+	switch (inode->i_mode & S_IFMT) {
+	case S_IFBLK:
+	case S_IFCHR:
+		stat->blksize = BLKDEV_IOSIZE;
+		stat->rdev = MKDEV(sysv_major(ip->i_df.if_u2.if_rdev) & 0x1ff,
+				   sysv_minor(ip->i_df.if_u2.if_rdev));
+		break;
+	default:
+		if (XFS_IS_REALTIME_INODE(ip)) {
+			/*
+			 * If the file blocks are being allocated from a
+			 * realtime volume, then return the inode's realtime
+			 * extent size or the realtime volume's extent size.
+			 */
+			stat->blksize =
+				xfs_get_extsz_hint(ip) << mp->m_sb.sb_blocklog;
+		} else
+			stat->blksize = xfs_preferred_iosize(mp);
+		stat->rdev = 0;
+		break;
+	}
+
+	return 0;
+}
+
+STATIC int
+xfs_vn_setattr(
+	struct dentry	*dentry,
+	struct iattr	*iattr)
+{
+	return -xfs_setattr(XFS_I(dentry->d_inode), iattr, 0);
+}
+
+#define XFS_FIEMAP_FLAGS	(FIEMAP_FLAG_SYNC|FIEMAP_FLAG_XATTR)
+
+/*
+ * Call fiemap helper to fill in user data.
+ * Returns positive errors to xfs_getbmap.
+ */
+STATIC int
+xfs_fiemap_format(
+	void			**arg,
+	struct getbmapx		*bmv,
+	int			*full)
+{
+	int			error;
+	struct fiemap_extent_info *fieinfo = *arg;
+	u32			fiemap_flags = 0;
+	u64			logical, physical, length;
+
+	/* Do nothing for a hole */
+	if (bmv->bmv_block == -1LL)
+		return 0;
+
+	logical = BBTOB(bmv->bmv_offset);
+	physical = BBTOB(bmv->bmv_block);
+	length = BBTOB(bmv->bmv_length);
+
+	if (bmv->bmv_oflags & BMV_OF_PREALLOC)
+		fiemap_flags |= FIEMAP_EXTENT_UNWRITTEN;
+	else if (bmv->bmv_oflags & BMV_OF_DELALLOC) {
+		fiemap_flags |= FIEMAP_EXTENT_DELALLOC;
+		physical = 0;   /* no block yet */
+	}
+	if (bmv->bmv_oflags & BMV_OF_LAST)
+		fiemap_flags |= FIEMAP_EXTENT_LAST;
+
+	error = fiemap_fill_next_extent(fieinfo, logical, physical,
+					length, fiemap_flags);
+	if (error > 0) {
+		error = 0;
+		*full = 1;	/* user array now full */
+	}
+
+	return -error;
+}
+
+STATIC int
+xfs_vn_fiemap(
+	struct inode		*inode,
+	struct fiemap_extent_info *fieinfo,
+	u64			start,
+	u64			length)
+{
+	xfs_inode_t		*ip = XFS_I(inode);
+	struct getbmapx		bm;
+	int			error;
+
+	error = fiemap_check_flags(fieinfo, XFS_FIEMAP_FLAGS);
+	if (error)
+		return error;
+
+	/* Set up bmap header for xfs internal routine */
+	bm.bmv_offset = BTOBB(start);
+	/* Special case for whole file */
+	if (length == FIEMAP_MAX_OFFSET)
+		bm.bmv_length = -1LL;
+	else
+		bm.bmv_length = BTOBB(length);
+
+	/* We add one because in getbmap world count includes the header */
+	bm.bmv_count = !fieinfo->fi_extents_max ? MAXEXTNUM :
+					fieinfo->fi_extents_max + 1;
+	bm.bmv_count = min_t(__s32, bm.bmv_count,
+			     (PAGE_SIZE * 16 / sizeof(struct getbmapx)));
+	bm.bmv_iflags = BMV_IF_PREALLOC | BMV_IF_NO_HOLES;
+	if (fieinfo->fi_flags & FIEMAP_FLAG_XATTR)
+		bm.bmv_iflags |= BMV_IF_ATTRFORK;
+	if (!(fieinfo->fi_flags & FIEMAP_FLAG_SYNC))
+		bm.bmv_iflags |= BMV_IF_DELALLOC;
+
+	error = xfs_getbmap(ip, &bm, xfs_fiemap_format, fieinfo);
+	if (error)
+		return -error;
+
+	return 0;
+}
+
+static const struct inode_operations xfs_inode_operations = {
+	.check_acl		= xfs_check_acl,
+	.getattr		= xfs_vn_getattr,
+	.setattr		= xfs_vn_setattr,
+	.setxattr		= generic_setxattr,
+	.getxattr		= generic_getxattr,
+	.removexattr		= generic_removexattr,
+	.listxattr		= xfs_vn_listxattr,
+	.fiemap			= xfs_vn_fiemap,
+};
+
+static const struct inode_operations xfs_dir_inode_operations = {
+	.create			= xfs_vn_create,
+	.lookup			= xfs_vn_lookup,
+	.link			= xfs_vn_link,
+	.unlink			= xfs_vn_unlink,
+	.symlink		= xfs_vn_symlink,
+	.mkdir			= xfs_vn_mkdir,
+	/*
+	 * Yes, XFS uses the same method for rmdir and unlink.
+	 *
+	 * There are some subtile differences deeper in the code,
+	 * but we use S_ISDIR to check for those.
+	 */
+	.rmdir			= xfs_vn_unlink,
+	.mknod			= xfs_vn_mknod,
+	.rename			= xfs_vn_rename,
+	.check_acl		= xfs_check_acl,
+	.getattr		= xfs_vn_getattr,
+	.setattr		= xfs_vn_setattr,
+	.setxattr		= generic_setxattr,
+	.getxattr		= generic_getxattr,
+	.removexattr		= generic_removexattr,
+	.listxattr		= xfs_vn_listxattr,
+};
+
+static const struct inode_operations xfs_dir_ci_inode_operations = {
+	.create			= xfs_vn_create,
+	.lookup			= xfs_vn_ci_lookup,
+	.link			= xfs_vn_link,
+	.unlink			= xfs_vn_unlink,
+	.symlink		= xfs_vn_symlink,
+	.mkdir			= xfs_vn_mkdir,
+	/*
+	 * Yes, XFS uses the same method for rmdir and unlink.
+	 *
+	 * There are some subtile differences deeper in the code,
+	 * but we use S_ISDIR to check for those.
+	 */
+	.rmdir			= xfs_vn_unlink,
+	.mknod			= xfs_vn_mknod,
+	.rename			= xfs_vn_rename,
+	.check_acl		= xfs_check_acl,
+	.getattr		= xfs_vn_getattr,
+	.setattr		= xfs_vn_setattr,
+	.setxattr		= generic_setxattr,
+	.getxattr		= generic_getxattr,
+	.removexattr		= generic_removexattr,
+	.listxattr		= xfs_vn_listxattr,
+};
+
+static const struct inode_operations xfs_symlink_inode_operations = {
+	.readlink		= generic_readlink,
+	.follow_link		= xfs_vn_follow_link,
+	.put_link		= xfs_vn_put_link,
+	.check_acl		= xfs_check_acl,
+	.getattr		= xfs_vn_getattr,
+	.setattr		= xfs_vn_setattr,
+	.setxattr		= generic_setxattr,
+	.getxattr		= generic_getxattr,
+	.removexattr		= generic_removexattr,
+	.listxattr		= xfs_vn_listxattr,
+};
+
+STATIC void
+xfs_diflags_to_iflags(
+	struct inode		*inode,
+	struct xfs_inode	*ip)
+{
+	if (ip->i_d.di_flags & XFS_DIFLAG_IMMUTABLE)
+		inode->i_flags |= S_IMMUTABLE;
+	else
+		inode->i_flags &= ~S_IMMUTABLE;
+	if (ip->i_d.di_flags & XFS_DIFLAG_APPEND)
+		inode->i_flags |= S_APPEND;
+	else
+		inode->i_flags &= ~S_APPEND;
+	if (ip->i_d.di_flags & XFS_DIFLAG_SYNC)
+		inode->i_flags |= S_SYNC;
+	else
+		inode->i_flags &= ~S_SYNC;
+	if (ip->i_d.di_flags & XFS_DIFLAG_NOATIME)
+		inode->i_flags |= S_NOATIME;
+	else
+		inode->i_flags &= ~S_NOATIME;
+}
+
+/*
+ * Initialize the Linux inode, set up the operation vectors and
+ * unlock the inode.
+ *
+ * When reading existing inodes from disk this is called directly
+ * from xfs_iget, when creating a new inode it is called from
+ * xfs_ialloc after setting up the inode.
+ *
+ * We are always called with an uninitialised linux inode here.
+ * We need to initialise the necessary fields and take a reference
+ * on it.
+ */
+void
+xfs_setup_inode(
+	struct xfs_inode	*ip)
+{
+	struct inode		*inode = &ip->i_vnode;
+
+	inode->i_ino = ip->i_ino;
+	inode->i_state = I_NEW;
+
+	inode_sb_list_add(inode);
+	/* make the inode look hashed for the writeback code */
+	hlist_add_fake(&inode->i_hash);
+
+	inode->i_mode	= ip->i_d.di_mode;
+	inode->i_nlink	= ip->i_d.di_nlink;
+	inode->i_uid	= ip->i_d.di_uid;
+	inode->i_gid	= ip->i_d.di_gid;
+
+	switch (inode->i_mode & S_IFMT) {
+	case S_IFBLK:
+	case S_IFCHR:
+		inode->i_rdev =
+			MKDEV(sysv_major(ip->i_df.if_u2.if_rdev) & 0x1ff,
+			      sysv_minor(ip->i_df.if_u2.if_rdev));
+		break;
+	default:
+		inode->i_rdev = 0;
+		break;
+	}
+
+	inode->i_generation = ip->i_d.di_gen;
+	i_size_write(inode, ip->i_d.di_size);
+	inode->i_atime.tv_sec	= ip->i_d.di_atime.t_sec;
+	inode->i_atime.tv_nsec	= ip->i_d.di_atime.t_nsec;
+	inode->i_mtime.tv_sec	= ip->i_d.di_mtime.t_sec;
+	inode->i_mtime.tv_nsec	= ip->i_d.di_mtime.t_nsec;
+	inode->i_ctime.tv_sec	= ip->i_d.di_ctime.t_sec;
+	inode->i_ctime.tv_nsec	= ip->i_d.di_ctime.t_nsec;
+	xfs_diflags_to_iflags(inode, ip);
+
+	switch (inode->i_mode & S_IFMT) {
+	case S_IFREG:
+		inode->i_op = &xfs_inode_operations;
+		inode->i_fop = &xfs_file_operations;
+		inode->i_mapping->a_ops = &xfs_address_space_operations;
+		break;
+	case S_IFDIR:
+		if (xfs_sb_version_hasasciici(&XFS_M(inode->i_sb)->m_sb))
+			inode->i_op = &xfs_dir_ci_inode_operations;
+		else
+			inode->i_op = &xfs_dir_inode_operations;
+		inode->i_fop = &xfs_dir_file_operations;
+		break;
+	case S_IFLNK:
+		inode->i_op = &xfs_symlink_inode_operations;
+		if (!(ip->i_df.if_flags & XFS_IFINLINE))
+			inode->i_mapping->a_ops = &xfs_address_space_operations;
+		break;
+	default:
+		inode->i_op = &xfs_inode_operations;
+		init_special_inode(inode, inode->i_mode, inode->i_rdev);
+		break;
+	}
+
+	xfs_iflags_clear(ip, XFS_INEW);
+	barrier();
+
+	unlock_new_inode(inode);
+}
diff --git a/fs/xfs/linux-2.6/xfs_iops.h b/fs/xfs/linux-2.6/xfs_iops.h
new file mode 100644
index 00000000..ef41c92c
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_iops.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright (c) 2000-2003,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_IOPS_H__
+#define __XFS_IOPS_H__
+
+struct xfs_inode;
+
+extern const struct file_operations xfs_file_operations;
+extern const struct file_operations xfs_dir_file_operations;
+
+extern ssize_t xfs_vn_listxattr(struct dentry *, char *data, size_t size);
+
+extern void xfs_setup_inode(struct xfs_inode *);
+
+#endif /* __XFS_IOPS_H__ */
diff --git a/fs/xfs/linux-2.6/xfs_linux.h b/fs/xfs/linux-2.6/xfs_linux.h
new file mode 100644
index 00000000..87315168
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_linux.h
@@ -0,0 +1,307 @@
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_LINUX__
+#define __XFS_LINUX__
+
+#include <linux/types.h>
+
+/*
+ * XFS_BIG_BLKNOS needs block layer disk addresses to be 64 bits.
+ * XFS_BIG_INUMS requires XFS_BIG_BLKNOS to be set.
+ */
+#if defined(CONFIG_LBDAF) || (BITS_PER_LONG == 64)
+# define XFS_BIG_BLKNOS	1
+# define XFS_BIG_INUMS	1
+#else
+# define XFS_BIG_BLKNOS	0
+# define XFS_BIG_INUMS	0
+#endif
+
+#include <xfs_types.h>
+#include <xfs_arch.h>
+
+#include <kmem.h>
+#include <mrlock.h>
+#include <time.h>
+
+#include <support/uuid.h>
+
+#include <linux/semaphore.h>
+#include <linux/mm.h>
+#include <linux/kernel.h>
+#include <linux/blkdev.h>
+#include <linux/slab.h>
+#include <linux/module.h>
+#include <linux/mutex.h>
+#include <linux/file.h>
+#include <linux/swap.h>
+#include <linux/errno.h>
+#include <linux/sched.h>
+#include <linux/bitops.h>
+#include <linux/major.h>
+#include <linux/pagemap.h>
+#include <linux/vfs.h>
+#include <linux/seq_file.h>
+#include <linux/init.h>
+#include <linux/list.h>
+#include <linux/proc_fs.h>
+#include <linux/sort.h>
+#include <linux/cpu.h>
+#include <linux/notifier.h>
+#include <linux/delay.h>
+#include <linux/log2.h>
+#include <linux/spinlock.h>
+#include <linux/random.h>
+#include <linux/ctype.h>
+#include <linux/writeback.h>
+#include <linux/capability.h>
+#include <linux/kthread.h>
+#include <linux/freezer.h>
+#include <linux/list_sort.h>
+
+#include <asm/page.h>
+#include <asm/div64.h>
+#include <asm/param.h>
+#include <asm/uaccess.h>
+#include <asm/byteorder.h>
+#include <asm/unaligned.h>
+
+#include <xfs_vnode.h>
+#include <xfs_stats.h>
+#include <xfs_sysctl.h>
+#include <xfs_iops.h>
+#include <xfs_aops.h>
+#include <xfs_super.h>
+#include <xfs_buf.h>
+#include <xfs_message.h>
+
+/*
+ * Feature macros (disable/enable)
+ */
+#ifdef CONFIG_SMP
+#define HAVE_PERCPU_SB	/* per cpu superblock counters are a 2.6 feature */
+#else
+#undef  HAVE_PERCPU_SB	/* per cpu superblock counters are a 2.6 feature */
+#endif
+
+#define irix_sgid_inherit	xfs_params.sgid_inherit.val
+#define irix_symlink_mode	xfs_params.symlink_mode.val
+#define xfs_panic_mask		xfs_params.panic_mask.val
+#define xfs_error_level		xfs_params.error_level.val
+#define xfs_syncd_centisecs	xfs_params.syncd_timer.val
+#define xfs_stats_clear		xfs_params.stats_clear.val
+#define xfs_inherit_sync	xfs_params.inherit_sync.val
+#define xfs_inherit_nodump	xfs_params.inherit_nodump.val
+#define xfs_inherit_noatime	xfs_params.inherit_noatim.val
+#define xfs_buf_timer_centisecs	xfs_params.xfs_buf_timer.val
+#define xfs_buf_age_centisecs	xfs_params.xfs_buf_age.val
+#define xfs_inherit_nosymlinks	xfs_params.inherit_nosym.val
+#define xfs_rotorstep		xfs_params.rotorstep.val
+#define xfs_inherit_nodefrag	xfs_params.inherit_nodfrg.val
+#define xfs_fstrm_centisecs	xfs_params.fstrm_timer.val
+
+#define current_cpu()		(raw_smp_processor_id())
+#define current_pid()		(current->pid)
+#define current_test_flags(f)	(current->flags & (f))
+#define current_set_flags_nested(sp, f)		\
+		(*(sp) = current->flags, current->flags |= (f))
+#define current_clear_flags_nested(sp, f)	\
+		(*(sp) = current->flags, current->flags &= ~(f))
+#define current_restore_flags_nested(sp, f)	\
+		(current->flags = ((current->flags & ~(f)) | (*(sp) & (f))))
+
+#define spinlock_destroy(lock)
+
+#define NBBY		8		/* number of bits per byte */
+
+/*
+ * Size of block device i/o is parameterized here.
+ * Currently the system supports page-sized i/o.
+ */
+#define	BLKDEV_IOSHIFT		PAGE_CACHE_SHIFT
+#define	BLKDEV_IOSIZE		(1<<BLKDEV_IOSHIFT)
+/* number of BB's per block device block */
+#define	BLKDEV_BB		BTOBB(BLKDEV_IOSIZE)
+
+#define ENOATTR		ENODATA		/* Attribute not found */
+#define EWRONGFS	EINVAL		/* Mount with wrong filesystem type */
+#define EFSCORRUPTED	EUCLEAN		/* Filesystem is corrupted */
+
+#define SYNCHRONIZE()	barrier()
+#define __return_address __builtin_return_address(0)
+
+#define XFS_PROJID_DEFAULT	0
+#define MAXPATHLEN	1024
+
+#define MIN(a,b)	(min(a,b))
+#define MAX(a,b)	(max(a,b))
+#define howmany(x, y)	(((x)+((y)-1))/(y))
+
+/*
+ * Various platform dependent calls that don't fit anywhere else
+ */
+#define xfs_sort(a,n,s,fn)	sort(a,n,s,fn,NULL)
+#define xfs_stack_trace()	dump_stack()
+
+
+/* Move the kernel do_div definition off to one side */
+
+#if defined __i386__
+/* For ia32 we need to pull some tricks to get past various versions
+ * of the compiler which do not like us using do_div in the middle
+ * of large functions.
+ */
+static inline __u32 xfs_do_div(void *a, __u32 b, int n)
+{
+	__u32	mod;
+
+	switch (n) {
+		case 4:
+			mod = *(__u32 *)a % b;
+			*(__u32 *)a = *(__u32 *)a / b;
+			return mod;
+		case 8:
+			{
+			unsigned long __upper, __low, __high, __mod;
+			__u64	c = *(__u64 *)a;
+			__upper = __high = c >> 32;
+			__low = c;
+			if (__high) {
+				__upper = __high % (b);
+				__high = __high / (b);
+			}
+			asm("divl %2":"=a" (__low), "=d" (__mod):"rm" (b), "0" (__low), "1" (__upper));
+			asm("":"=A" (c):"a" (__low),"d" (__high));
+			*(__u64 *)a = c;
+			return __mod;
+			}
+	}
+
+	/* NOTREACHED */
+	return 0;
+}
+
+/* Side effect free 64 bit mod operation */
+static inline __u32 xfs_do_mod(void *a, __u32 b, int n)
+{
+	switch (n) {
+		case 4:
+			return *(__u32 *)a % b;
+		case 8:
+			{
+			unsigned long __upper, __low, __high, __mod;
+			__u64	c = *(__u64 *)a;
+			__upper = __high = c >> 32;
+			__low = c;
+			if (__high) {
+				__upper = __high % (b);
+				__high = __high / (b);
+			}
+			asm("divl %2":"=a" (__low), "=d" (__mod):"rm" (b), "0" (__low), "1" (__upper));
+			asm("":"=A" (c):"a" (__low),"d" (__high));
+			return __mod;
+			}
+	}
+
+	/* NOTREACHED */
+	return 0;
+}
+#else
+static inline __u32 xfs_do_div(void *a, __u32 b, int n)
+{
+	__u32	mod;
+
+	switch (n) {
+		case 4:
+			mod = *(__u32 *)a % b;
+			*(__u32 *)a = *(__u32 *)a / b;
+			return mod;
+		case 8:
+			mod = do_div(*(__u64 *)a, b);
+			return mod;
+	}
+
+	/* NOTREACHED */
+	return 0;
+}
+
+/* Side effect free 64 bit mod operation */
+static inline __u32 xfs_do_mod(void *a, __u32 b, int n)
+{
+	switch (n) {
+		case 4:
+			return *(__u32 *)a % b;
+		case 8:
+			{
+			__u64	c = *(__u64 *)a;
+			return do_div(c, b);
+			}
+	}
+
+	/* NOTREACHED */
+	return 0;
+}
+#endif
+
+#undef do_div
+#define do_div(a, b)	xfs_do_div(&(a), (b), sizeof(a))
+#define do_mod(a, b)	xfs_do_mod(&(a), (b), sizeof(a))
+
+static inline __uint64_t roundup_64(__uint64_t x, __uint32_t y)
+{
+	x += y - 1;
+	do_div(x, y);
+	return(x * y);
+}
+
+static inline __uint64_t howmany_64(__uint64_t x, __uint32_t y)
+{
+	x += y - 1;
+	do_div(x, y);
+	return x;
+}
+
+/* ARM old ABI has some weird alignment/padding */
+#if defined(__arm__) && !defined(__ARM_EABI__)
+#define __arch_pack __attribute__((packed))
+#else
+#define __arch_pack
+#endif
+
+#define ASSERT_ALWAYS(expr)	\
+	(unlikely(expr) ? (void)0 : assfail(#expr, __FILE__, __LINE__))
+
+#ifndef DEBUG
+#define ASSERT(expr)	((void)0)
+
+#ifndef STATIC
+# define STATIC static noinline
+#endif
+
+#else /* DEBUG */
+
+#define ASSERT(expr)	\
+	(unlikely(expr) ? (void)0 : assfail(#expr, __FILE__, __LINE__))
+
+#ifndef STATIC
+# define STATIC noinline
+#endif
+
+#endif /* DEBUG */
+
+#endif /* __XFS_LINUX__ */
diff --git a/fs/xfs/linux-2.6/xfs_message.c b/fs/xfs/linux-2.6/xfs_message.c
new file mode 100644
index 00000000..bd672def
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_message.c
@@ -0,0 +1,108 @@
+/*
+ * Copyright (c) 2011 Red Hat, Inc.  All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_mount.h"
+
+/*
+ * XFS logging functions
+ */
+static void
+__xfs_printk(
+	const char		*level,
+	const struct xfs_mount	*mp,
+	struct va_format	*vaf)
+{
+	if (mp && mp->m_fsname) {
+		printk("%sXFS (%s): %pV\n", level, mp->m_fsname, vaf);
+		return;
+	}
+	printk("%sXFS: %pV\n", level, vaf);
+}
+
+#define define_xfs_printk_level(func, kern_level)		\
+void func(const struct xfs_mount *mp, const char *fmt, ...)	\
+{								\
+	struct va_format	vaf;				\
+	va_list			args;				\
+								\
+	va_start(args, fmt);					\
+								\
+	vaf.fmt = fmt;						\
+	vaf.va = &args;						\
+								\
+	__xfs_printk(kern_level, mp, &vaf);			\
+	va_end(args);						\
+}								\
+
+define_xfs_printk_level(xfs_emerg, KERN_EMERG);
+define_xfs_printk_level(xfs_alert, KERN_ALERT);
+define_xfs_printk_level(xfs_crit, KERN_CRIT);
+define_xfs_printk_level(xfs_err, KERN_ERR);
+define_xfs_printk_level(xfs_warn, KERN_WARNING);
+define_xfs_printk_level(xfs_notice, KERN_NOTICE);
+define_xfs_printk_level(xfs_info, KERN_INFO);
+#ifdef DEBUG
+define_xfs_printk_level(xfs_debug, KERN_DEBUG);
+#endif
+
+void
+xfs_alert_tag(
+	const struct xfs_mount	*mp,
+	int			panic_tag,
+	const char		*fmt, ...)
+{
+	struct va_format	vaf;
+	va_list			args;
+	int			do_panic = 0;
+
+	if (xfs_panic_mask && (xfs_panic_mask & panic_tag)) {
+		xfs_alert(mp, "Transforming an alert into a BUG.");
+		do_panic = 1;
+	}
+
+	va_start(args, fmt);
+
+	vaf.fmt = fmt;
+	vaf.va = &args;
+
+	__xfs_printk(KERN_ALERT, mp, &vaf);
+	va_end(args);
+
+	BUG_ON(do_panic);
+}
+
+void
+assfail(char *expr, char *file, int line)
+{
+	xfs_emerg(NULL, "Assertion failed: %s, file: %s, line: %d",
+		expr, file, line);
+	BUG();
+}
+
+void
+xfs_hex_dump(void *p, int length)
+{
+	print_hex_dump(KERN_ALERT, "", DUMP_PREFIX_ADDRESS, 16, 1, p, length, 1);
+}
diff --git a/fs/xfs/linux-2.6/xfs_message.h b/fs/xfs/linux-2.6/xfs_message.h
new file mode 100644
index 00000000..7fb7ea00
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_message.h
@@ -0,0 +1,39 @@
+#ifndef __XFS_MESSAGE_H
+#define __XFS_MESSAGE_H 1
+
+struct xfs_mount;
+
+extern void xfs_emerg(const struct xfs_mount *mp, const char *fmt, ...)
+        __attribute__ ((format (printf, 2, 3)));
+extern void xfs_alert(const struct xfs_mount *mp, const char *fmt, ...)
+        __attribute__ ((format (printf, 2, 3)));
+extern void xfs_alert_tag(const struct xfs_mount *mp, int tag,
+			 const char *fmt, ...)
+        __attribute__ ((format (printf, 3, 4)));
+extern void xfs_crit(const struct xfs_mount *mp, const char *fmt, ...)
+        __attribute__ ((format (printf, 2, 3)));
+extern void xfs_err(const struct xfs_mount *mp, const char *fmt, ...)
+        __attribute__ ((format (printf, 2, 3)));
+extern void xfs_warn(const struct xfs_mount *mp, const char *fmt, ...)
+        __attribute__ ((format (printf, 2, 3)));
+extern void xfs_notice(const struct xfs_mount *mp, const char *fmt, ...)
+        __attribute__ ((format (printf, 2, 3)));
+extern void xfs_info(const struct xfs_mount *mp, const char *fmt, ...)
+        __attribute__ ((format (printf, 2, 3)));
+
+#ifdef DEBUG
+extern void xfs_debug(const struct xfs_mount *mp, const char *fmt, ...)
+        __attribute__ ((format (printf, 2, 3)));
+#else
+static inline void
+__attribute__ ((format (printf, 2, 3)))
+xfs_debug(const struct xfs_mount *mp, const char *fmt, ...)
+{
+}
+#endif
+
+extern void assfail(char *expr, char *f, int l);
+
+extern void xfs_hex_dump(void *p, int length);
+
+#endif	/* __XFS_MESSAGE_H */
diff --git a/fs/xfs/linux-2.6/xfs_quotaops.c b/fs/xfs/linux-2.6/xfs_quotaops.c
new file mode 100644
index 00000000..29b9d642
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_quotaops.c
@@ -0,0 +1,139 @@
+/*
+ * Copyright (c) 2008, Christoph Hellwig
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_sb.h"
+#include "xfs_inum.h"
+#include "xfs_log.h"
+#include "xfs_ag.h"
+#include "xfs_mount.h"
+#include "xfs_quota.h"
+#include "xfs_trans.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_inode.h"
+#include "quota/xfs_qm.h"
+#include <linux/quota.h>
+
+
+STATIC int
+xfs_quota_type(int type)
+{
+	switch (type) {
+	case USRQUOTA:
+		return XFS_DQ_USER;
+	case GRPQUOTA:
+		return XFS_DQ_GROUP;
+	default:
+		return XFS_DQ_PROJ;
+	}
+}
+
+STATIC int
+xfs_fs_get_xstate(
+	struct super_block	*sb,
+	struct fs_quota_stat	*fqs)
+{
+	struct xfs_mount	*mp = XFS_M(sb);
+
+	if (!XFS_IS_QUOTA_RUNNING(mp))
+		return -ENOSYS;
+	return -xfs_qm_scall_getqstat(mp, fqs);
+}
+
+STATIC int
+xfs_fs_set_xstate(
+	struct super_block	*sb,
+	unsigned int		uflags,
+	int			op)
+{
+	struct xfs_mount	*mp = XFS_M(sb);
+	unsigned int		flags = 0;
+
+	if (sb->s_flags & MS_RDONLY)
+		return -EROFS;
+	if (op != Q_XQUOTARM && !XFS_IS_QUOTA_RUNNING(mp))
+		return -ENOSYS;
+
+	if (uflags & FS_QUOTA_UDQ_ACCT)
+		flags |= XFS_UQUOTA_ACCT;
+	if (uflags & FS_QUOTA_PDQ_ACCT)
+		flags |= XFS_PQUOTA_ACCT;
+	if (uflags & FS_QUOTA_GDQ_ACCT)
+		flags |= XFS_GQUOTA_ACCT;
+	if (uflags & FS_QUOTA_UDQ_ENFD)
+		flags |= XFS_UQUOTA_ENFD;
+	if (uflags & (FS_QUOTA_PDQ_ENFD|FS_QUOTA_GDQ_ENFD))
+		flags |= XFS_OQUOTA_ENFD;
+
+	switch (op) {
+	case Q_XQUOTAON:
+		return -xfs_qm_scall_quotaon(mp, flags);
+	case Q_XQUOTAOFF:
+		if (!XFS_IS_QUOTA_ON(mp))
+			return -EINVAL;
+		return -xfs_qm_scall_quotaoff(mp, flags);
+	case Q_XQUOTARM:
+		if (XFS_IS_QUOTA_ON(mp))
+			return -EINVAL;
+		return -xfs_qm_scall_trunc_qfiles(mp, flags);
+	}
+
+	return -EINVAL;
+}
+
+STATIC int
+xfs_fs_get_dqblk(
+	struct super_block	*sb,
+	int			type,
+	qid_t			id,
+	struct fs_disk_quota	*fdq)
+{
+	struct xfs_mount	*mp = XFS_M(sb);
+
+	if (!XFS_IS_QUOTA_RUNNING(mp))
+		return -ENOSYS;
+	if (!XFS_IS_QUOTA_ON(mp))
+		return -ESRCH;
+
+	return -xfs_qm_scall_getquota(mp, id, xfs_quota_type(type), fdq);
+}
+
+STATIC int
+xfs_fs_set_dqblk(
+	struct super_block	*sb,
+	int			type,
+	qid_t			id,
+	struct fs_disk_quota	*fdq)
+{
+	struct xfs_mount	*mp = XFS_M(sb);
+
+	if (sb->s_flags & MS_RDONLY)
+		return -EROFS;
+	if (!XFS_IS_QUOTA_RUNNING(mp))
+		return -ENOSYS;
+	if (!XFS_IS_QUOTA_ON(mp))
+		return -ESRCH;
+
+	return -xfs_qm_scall_setqlim(mp, id, xfs_quota_type(type), fdq);
+}
+
+const struct quotactl_ops xfs_quotactl_operations = {
+	.get_xstate		= xfs_fs_get_xstate,
+	.set_xstate		= xfs_fs_set_xstate,
+	.get_dqblk		= xfs_fs_get_dqblk,
+	.set_dqblk		= xfs_fs_set_dqblk,
+};
diff --git a/fs/xfs/linux-2.6/xfs_stats.c b/fs/xfs/linux-2.6/xfs_stats.c
new file mode 100644
index 00000000..76fdc586
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_stats.c
@@ -0,0 +1,122 @@
+/*
+ * Copyright (c) 2000-2003,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include <linux/proc_fs.h>
+
+DEFINE_PER_CPU(struct xfsstats, xfsstats);
+
+static int xfs_stat_proc_show(struct seq_file *m, void *v)
+{
+	int		c, i, j, val;
+	__uint64_t	xs_xstrat_bytes = 0;
+	__uint64_t	xs_write_bytes = 0;
+	__uint64_t	xs_read_bytes = 0;
+
+	static const struct xstats_entry {
+		char	*desc;
+		int	endpoint;
+	} xstats[] = {
+		{ "extent_alloc",	XFSSTAT_END_EXTENT_ALLOC	},
+		{ "abt",		XFSSTAT_END_ALLOC_BTREE		},
+		{ "blk_map",		XFSSTAT_END_BLOCK_MAPPING	},
+		{ "bmbt",		XFSSTAT_END_BLOCK_MAP_BTREE	},
+		{ "dir",		XFSSTAT_END_DIRECTORY_OPS	},
+		{ "trans",		XFSSTAT_END_TRANSACTIONS	},
+		{ "ig",			XFSSTAT_END_INODE_OPS		},
+		{ "log",		XFSSTAT_END_LOG_OPS		},
+		{ "push_ail",		XFSSTAT_END_TAIL_PUSHING	},
+		{ "xstrat",		XFSSTAT_END_WRITE_CONVERT	},
+		{ "rw",			XFSSTAT_END_READ_WRITE_OPS	},
+		{ "attr",		XFSSTAT_END_ATTRIBUTE_OPS	},
+		{ "icluster",		XFSSTAT_END_INODE_CLUSTER	},
+		{ "vnodes",		XFSSTAT_END_VNODE_OPS		},
+		{ "buf",		XFSSTAT_END_BUF			},
+		{ "abtb2",		XFSSTAT_END_ABTB_V2		},
+		{ "abtc2",		XFSSTAT_END_ABTC_V2		},
+		{ "bmbt2",		XFSSTAT_END_BMBT_V2		},
+		{ "ibt2",		XFSSTAT_END_IBT_V2		},
+	};
+
+	/* Loop over all stats groups */
+	for (i=j = 0; i < ARRAY_SIZE(xstats); i++) {
+		seq_printf(m, "%s", xstats[i].desc);
+		/* inner loop does each group */
+		while (j < xstats[i].endpoint) {
+			val = 0;
+			/* sum over all cpus */
+			for_each_possible_cpu(c)
+				val += *(((__u32*)&per_cpu(xfsstats, c) + j));
+			seq_printf(m, " %u", val);
+			j++;
+		}
+		seq_putc(m, '\n');
+	}
+	/* extra precision counters */
+	for_each_possible_cpu(i) {
+		xs_xstrat_bytes += per_cpu(xfsstats, i).xs_xstrat_bytes;
+		xs_write_bytes += per_cpu(xfsstats, i).xs_write_bytes;
+		xs_read_bytes += per_cpu(xfsstats, i).xs_read_bytes;
+	}
+
+	seq_printf(m, "xpc %Lu %Lu %Lu\n",
+			xs_xstrat_bytes, xs_write_bytes, xs_read_bytes);
+	seq_printf(m, "debug %u\n",
+#if defined(DEBUG)
+		1);
+#else
+		0);
+#endif
+	return 0;
+}
+
+static int xfs_stat_proc_open(struct inode *inode, struct file *file)
+{
+	return single_open(file, xfs_stat_proc_show, NULL);
+}
+
+static const struct file_operations xfs_stat_proc_fops = {
+	.owner		= THIS_MODULE,
+	.open		= xfs_stat_proc_open,
+	.read		= seq_read,
+	.llseek		= seq_lseek,
+	.release	= single_release,
+};
+
+int
+xfs_init_procfs(void)
+{
+	if (!proc_mkdir("fs/xfs", NULL))
+		goto out;
+
+	if (!proc_create("fs/xfs/stat", 0, NULL,
+			 &xfs_stat_proc_fops))
+		goto out_remove_entry;
+	return 0;
+
+ out_remove_entry:
+	remove_proc_entry("fs/xfs", NULL);
+ out:
+	return -ENOMEM;
+}
+
+void
+xfs_cleanup_procfs(void)
+{
+	remove_proc_entry("fs/xfs/stat", NULL);
+	remove_proc_entry("fs/xfs", NULL);
+}
diff --git a/fs/xfs/linux-2.6/xfs_stats.h b/fs/xfs/linux-2.6/xfs_stats.h
new file mode 100644
index 00000000..736854b1
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_stats.h
@@ -0,0 +1,223 @@
+/*
+ * Copyright (c) 2000,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_STATS_H__
+#define __XFS_STATS_H__
+
+
+#if defined(CONFIG_PROC_FS) && !defined(XFS_STATS_OFF)
+
+#include <linux/percpu.h>
+
+/*
+ * XFS global statistics
+ */
+struct xfsstats {
+# define XFSSTAT_END_EXTENT_ALLOC	4
+	__uint32_t		xs_allocx;
+	__uint32_t		xs_allocb;
+	__uint32_t		xs_freex;
+	__uint32_t		xs_freeb;
+# define XFSSTAT_END_ALLOC_BTREE	(XFSSTAT_END_EXTENT_ALLOC+4)
+	__uint32_t		xs_abt_lookup;
+	__uint32_t		xs_abt_compare;
+	__uint32_t		xs_abt_insrec;
+	__uint32_t		xs_abt_delrec;
+# define XFSSTAT_END_BLOCK_MAPPING	(XFSSTAT_END_ALLOC_BTREE+7)
+	__uint32_t		xs_blk_mapr;
+	__uint32_t		xs_blk_mapw;
+	__uint32_t		xs_blk_unmap;
+	__uint32_t		xs_add_exlist;
+	__uint32_t		xs_del_exlist;
+	__uint32_t		xs_look_exlist;
+	__uint32_t		xs_cmp_exlist;
+# define XFSSTAT_END_BLOCK_MAP_BTREE	(XFSSTAT_END_BLOCK_MAPPING+4)
+	__uint32_t		xs_bmbt_lookup;
+	__uint32_t		xs_bmbt_compare;
+	__uint32_t		xs_bmbt_insrec;
+	__uint32_t		xs_bmbt_delrec;
+# define XFSSTAT_END_DIRECTORY_OPS	(XFSSTAT_END_BLOCK_MAP_BTREE+4)
+	__uint32_t		xs_dir_lookup;
+	__uint32_t		xs_dir_create;
+	__uint32_t		xs_dir_remove;
+	__uint32_t		xs_dir_getdents;
+# define XFSSTAT_END_TRANSACTIONS	(XFSSTAT_END_DIRECTORY_OPS+3)
+	__uint32_t		xs_trans_sync;
+	__uint32_t		xs_trans_async;
+	__uint32_t		xs_trans_empty;
+# define XFSSTAT_END_INODE_OPS		(XFSSTAT_END_TRANSACTIONS+7)
+	__uint32_t		xs_ig_attempts;
+	__uint32_t		xs_ig_found;
+	__uint32_t		xs_ig_frecycle;
+	__uint32_t		xs_ig_missed;
+	__uint32_t		xs_ig_dup;
+	__uint32_t		xs_ig_reclaims;
+	__uint32_t		xs_ig_attrchg;
+# define XFSSTAT_END_LOG_OPS		(XFSSTAT_END_INODE_OPS+5)
+	__uint32_t		xs_log_writes;
+	__uint32_t		xs_log_blocks;
+	__uint32_t		xs_log_noiclogs;
+	__uint32_t		xs_log_force;
+	__uint32_t		xs_log_force_sleep;
+# define XFSSTAT_END_TAIL_PUSHING	(XFSSTAT_END_LOG_OPS+10)
+	__uint32_t		xs_try_logspace;
+	__uint32_t		xs_sleep_logspace;
+	__uint32_t		xs_push_ail;
+	__uint32_t		xs_push_ail_success;
+	__uint32_t		xs_push_ail_pushbuf;
+	__uint32_t		xs_push_ail_pinned;
+	__uint32_t		xs_push_ail_locked;
+	__uint32_t		xs_push_ail_flushing;
+	__uint32_t		xs_push_ail_restarts;
+	__uint32_t		xs_push_ail_flush;
+# define XFSSTAT_END_WRITE_CONVERT	(XFSSTAT_END_TAIL_PUSHING+2)
+	__uint32_t		xs_xstrat_quick;
+	__uint32_t		xs_xstrat_split;
+# define XFSSTAT_END_READ_WRITE_OPS	(XFSSTAT_END_WRITE_CONVERT+2)
+	__uint32_t		xs_write_calls;
+	__uint32_t		xs_read_calls;
+# define XFSSTAT_END_ATTRIBUTE_OPS	(XFSSTAT_END_READ_WRITE_OPS+4)
+	__uint32_t		xs_attr_get;
+	__uint32_t		xs_attr_set;
+	__uint32_t		xs_attr_remove;
+	__uint32_t		xs_attr_list;
+# define XFSSTAT_END_INODE_CLUSTER	(XFSSTAT_END_ATTRIBUTE_OPS+3)
+	__uint32_t		xs_iflush_count;
+	__uint32_t		xs_icluster_flushcnt;
+	__uint32_t		xs_icluster_flushinode;
+# define XFSSTAT_END_VNODE_OPS		(XFSSTAT_END_INODE_CLUSTER+8)
+	__uint32_t		vn_active;	/* # vnodes not on free lists */
+	__uint32_t		vn_alloc;	/* # times vn_alloc called */
+	__uint32_t		vn_get;		/* # times vn_get called */
+	__uint32_t		vn_hold;	/* # times vn_hold called */
+	__uint32_t		vn_rele;	/* # times vn_rele called */
+	__uint32_t		vn_reclaim;	/* # times vn_reclaim called */
+	__uint32_t		vn_remove;	/* # times vn_remove called */
+	__uint32_t		vn_free;	/* # times vn_free called */
+#define XFSSTAT_END_BUF			(XFSSTAT_END_VNODE_OPS+9)
+	__uint32_t		xb_get;
+	__uint32_t		xb_create;
+	__uint32_t		xb_get_locked;
+	__uint32_t		xb_get_locked_waited;
+	__uint32_t		xb_busy_locked;
+	__uint32_t		xb_miss_locked;
+	__uint32_t		xb_page_retries;
+	__uint32_t		xb_page_found;
+	__uint32_t		xb_get_read;
+/* Version 2 btree counters */
+#define XFSSTAT_END_ABTB_V2		(XFSSTAT_END_BUF+15)
+	__uint32_t		xs_abtb_2_lookup;
+	__uint32_t		xs_abtb_2_compare;
+	__uint32_t		xs_abtb_2_insrec;
+	__uint32_t		xs_abtb_2_delrec;
+	__uint32_t		xs_abtb_2_newroot;
+	__uint32_t		xs_abtb_2_killroot;
+	__uint32_t		xs_abtb_2_increment;
+	__uint32_t		xs_abtb_2_decrement;
+	__uint32_t		xs_abtb_2_lshift;
+	__uint32_t		xs_abtb_2_rshift;
+	__uint32_t		xs_abtb_2_split;
+	__uint32_t		xs_abtb_2_join;
+	__uint32_t		xs_abtb_2_alloc;
+	__uint32_t		xs_abtb_2_free;
+	__uint32_t		xs_abtb_2_moves;
+#define XFSSTAT_END_ABTC_V2		(XFSSTAT_END_ABTB_V2+15)
+	__uint32_t		xs_abtc_2_lookup;
+	__uint32_t		xs_abtc_2_compare;
+	__uint32_t		xs_abtc_2_insrec;
+	__uint32_t		xs_abtc_2_delrec;
+	__uint32_t		xs_abtc_2_newroot;
+	__uint32_t		xs_abtc_2_killroot;
+	__uint32_t		xs_abtc_2_increment;
+	__uint32_t		xs_abtc_2_decrement;
+	__uint32_t		xs_abtc_2_lshift;
+	__uint32_t		xs_abtc_2_rshift;
+	__uint32_t		xs_abtc_2_split;
+	__uint32_t		xs_abtc_2_join;
+	__uint32_t		xs_abtc_2_alloc;
+	__uint32_t		xs_abtc_2_free;
+	__uint32_t		xs_abtc_2_moves;
+#define XFSSTAT_END_BMBT_V2		(XFSSTAT_END_ABTC_V2+15)
+	__uint32_t		xs_bmbt_2_lookup;
+	__uint32_t		xs_bmbt_2_compare;
+	__uint32_t		xs_bmbt_2_insrec;
+	__uint32_t		xs_bmbt_2_delrec;
+	__uint32_t		xs_bmbt_2_newroot;
+	__uint32_t		xs_bmbt_2_killroot;
+	__uint32_t		xs_bmbt_2_increment;
+	__uint32_t		xs_bmbt_2_decrement;
+	__uint32_t		xs_bmbt_2_lshift;
+	__uint32_t		xs_bmbt_2_rshift;
+	__uint32_t		xs_bmbt_2_split;
+	__uint32_t		xs_bmbt_2_join;
+	__uint32_t		xs_bmbt_2_alloc;
+	__uint32_t		xs_bmbt_2_free;
+	__uint32_t		xs_bmbt_2_moves;
+#define XFSSTAT_END_IBT_V2		(XFSSTAT_END_BMBT_V2+15)
+	__uint32_t		xs_ibt_2_lookup;
+	__uint32_t		xs_ibt_2_compare;
+	__uint32_t		xs_ibt_2_insrec;
+	__uint32_t		xs_ibt_2_delrec;
+	__uint32_t		xs_ibt_2_newroot;
+	__uint32_t		xs_ibt_2_killroot;
+	__uint32_t		xs_ibt_2_increment;
+	__uint32_t		xs_ibt_2_decrement;
+	__uint32_t		xs_ibt_2_lshift;
+	__uint32_t		xs_ibt_2_rshift;
+	__uint32_t		xs_ibt_2_split;
+	__uint32_t		xs_ibt_2_join;
+	__uint32_t		xs_ibt_2_alloc;
+	__uint32_t		xs_ibt_2_free;
+	__uint32_t		xs_ibt_2_moves;
+/* Extra precision counters */
+	__uint64_t		xs_xstrat_bytes;
+	__uint64_t		xs_write_bytes;
+	__uint64_t		xs_read_bytes;
+};
+
+DECLARE_PER_CPU(struct xfsstats, xfsstats);
+
+/*
+ * We don't disable preempt, not too worried about poking the
+ * wrong CPU's stat for now (also aggregated before reporting).
+ */
+#define XFS_STATS_INC(v)	(per_cpu(xfsstats, current_cpu()).v++)
+#define XFS_STATS_DEC(v)	(per_cpu(xfsstats, current_cpu()).v--)
+#define XFS_STATS_ADD(v, inc)	(per_cpu(xfsstats, current_cpu()).v += (inc))
+
+extern int xfs_init_procfs(void);
+extern void xfs_cleanup_procfs(void);
+
+
+#else	/* !CONFIG_PROC_FS */
+
+# define XFS_STATS_INC(count)
+# define XFS_STATS_DEC(count)
+# define XFS_STATS_ADD(count, inc)
+
+static inline int xfs_init_procfs(void)
+{
+	return 0;
+}
+
+static inline void xfs_cleanup_procfs(void)
+{
+}
+
+#endif	/* !CONFIG_PROC_FS */
+
+#endif /* __XFS_STATS_H__ */
diff --git a/fs/xfs/linux-2.6/xfs_super.c b/fs/xfs/linux-2.6/xfs_super.c
new file mode 100644
index 00000000..e6ac98c1
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_super.c
@@ -0,0 +1,1720 @@
+/*
+ * Copyright (c) 2000-2006 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+
+#include "xfs.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_dir2.h"
+#include "xfs_alloc.h"
+#include "xfs_quota.h"
+#include "xfs_mount.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_alloc_btree.h"
+#include "xfs_ialloc_btree.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_btree.h"
+#include "xfs_btree_trace.h"
+#include "xfs_ialloc.h"
+#include "xfs_bmap.h"
+#include "xfs_rtalloc.h"
+#include "xfs_error.h"
+#include "xfs_itable.h"
+#include "xfs_fsops.h"
+#include "xfs_attr.h"
+#include "xfs_buf_item.h"
+#include "xfs_utils.h"
+#include "xfs_vnodeops.h"
+#include "xfs_log_priv.h"
+#include "xfs_trans_priv.h"
+#include "xfs_filestream.h"
+#include "xfs_da_btree.h"
+#include "xfs_extfree_item.h"
+#include "xfs_mru_cache.h"
+#include "xfs_inode_item.h"
+#include "xfs_sync.h"
+#include "xfs_trace.h"
+
+#include <linux/namei.h>
+#include <linux/init.h>
+#include <linux/slab.h>
+#include <linux/mount.h>
+#include <linux/mempool.h>
+#include <linux/writeback.h>
+#include <linux/kthread.h>
+#include <linux/freezer.h>
+#include <linux/parser.h>
+
+static const struct super_operations xfs_super_operations;
+static kmem_zone_t *xfs_ioend_zone;
+mempool_t *xfs_ioend_pool;
+
+#define MNTOPT_LOGBUFS	"logbufs"	/* number of XFS log buffers */
+#define MNTOPT_LOGBSIZE	"logbsize"	/* size of XFS log buffers */
+#define MNTOPT_LOGDEV	"logdev"	/* log device */
+#define MNTOPT_RTDEV	"rtdev"		/* realtime I/O device */
+#define MNTOPT_BIOSIZE	"biosize"	/* log2 of preferred buffered io size */
+#define MNTOPT_WSYNC	"wsync"		/* safe-mode nfs compatible mount */
+#define MNTOPT_NOALIGN	"noalign"	/* turn off stripe alignment */
+#define MNTOPT_SWALLOC	"swalloc"	/* turn on stripe width allocation */
+#define MNTOPT_SUNIT	"sunit"		/* data volume stripe unit */
+#define MNTOPT_SWIDTH	"swidth"	/* data volume stripe width */
+#define MNTOPT_NOUUID	"nouuid"	/* ignore filesystem UUID */
+#define MNTOPT_MTPT	"mtpt"		/* filesystem mount point */
+#define MNTOPT_GRPID	"grpid"		/* group-ID from parent directory */
+#define MNTOPT_NOGRPID	"nogrpid"	/* group-ID from current process */
+#define MNTOPT_BSDGROUPS    "bsdgroups"    /* group-ID from parent directory */
+#define MNTOPT_SYSVGROUPS   "sysvgroups"   /* group-ID from current process */
+#define MNTOPT_ALLOCSIZE    "allocsize"    /* preferred allocation size */
+#define MNTOPT_NORECOVERY   "norecovery"   /* don't run XFS recovery */
+#define MNTOPT_BARRIER	"barrier"	/* use writer barriers for log write and
+					 * unwritten extent conversion */
+#define MNTOPT_NOBARRIER "nobarrier"	/* .. disable */
+#define MNTOPT_64BITINODE   "inode64"	/* inodes can be allocated anywhere */
+#define MNTOPT_IKEEP	"ikeep"		/* do not free empty inode clusters */
+#define MNTOPT_NOIKEEP	"noikeep"	/* free empty inode clusters */
+#define MNTOPT_LARGEIO	   "largeio"	/* report large I/O sizes in stat() */
+#define MNTOPT_NOLARGEIO   "nolargeio"	/* do not report large I/O sizes
+					 * in stat(). */
+#define MNTOPT_ATTR2	"attr2"		/* do use attr2 attribute format */
+#define MNTOPT_NOATTR2	"noattr2"	/* do not use attr2 attribute format */
+#define MNTOPT_FILESTREAM  "filestreams" /* use filestreams allocator */
+#define MNTOPT_QUOTA	"quota"		/* disk quotas (user) */
+#define MNTOPT_NOQUOTA	"noquota"	/* no quotas */
+#define MNTOPT_USRQUOTA	"usrquota"	/* user quota enabled */
+#define MNTOPT_GRPQUOTA	"grpquota"	/* group quota enabled */
+#define MNTOPT_PRJQUOTA	"prjquota"	/* project quota enabled */
+#define MNTOPT_UQUOTA	"uquota"	/* user quota (IRIX variant) */
+#define MNTOPT_GQUOTA	"gquota"	/* group quota (IRIX variant) */
+#define MNTOPT_PQUOTA	"pquota"	/* project quota (IRIX variant) */
+#define MNTOPT_UQUOTANOENF "uqnoenforce"/* user quota limit enforcement */
+#define MNTOPT_GQUOTANOENF "gqnoenforce"/* group quota limit enforcement */
+#define MNTOPT_PQUOTANOENF "pqnoenforce"/* project quota limit enforcement */
+#define MNTOPT_QUOTANOENF  "qnoenforce"	/* same as uqnoenforce */
+#define MNTOPT_DELAYLOG    "delaylog"	/* Delayed logging enabled */
+#define MNTOPT_NODELAYLOG  "nodelaylog"	/* Delayed logging disabled */
+#define MNTOPT_DISCARD	   "discard"	/* Discard unused blocks */
+#define MNTOPT_NODISCARD   "nodiscard"	/* Do not discard unused blocks */
+
+/*
+ * Table driven mount option parser.
+ *
+ * Currently only used for remount, but it will be used for mount
+ * in the future, too.
+ */
+enum {
+	Opt_barrier, Opt_nobarrier, Opt_err
+};
+
+static const match_table_t tokens = {
+	{Opt_barrier, "barrier"},
+	{Opt_nobarrier, "nobarrier"},
+	{Opt_err, NULL}
+};
+
+
+STATIC unsigned long
+suffix_strtoul(char *s, char **endp, unsigned int base)
+{
+	int	last, shift_left_factor = 0;
+	char	*value = s;
+
+	last = strlen(value) - 1;
+	if (value[last] == 'K' || value[last] == 'k') {
+		shift_left_factor = 10;
+		value[last] = '\0';
+	}
+	if (value[last] == 'M' || value[last] == 'm') {
+		shift_left_factor = 20;
+		value[last] = '\0';
+	}
+	if (value[last] == 'G' || value[last] == 'g') {
+		shift_left_factor = 30;
+		value[last] = '\0';
+	}
+
+	return simple_strtoul((const char *)s, endp, base) << shift_left_factor;
+}
+
+/*
+ * This function fills in xfs_mount_t fields based on mount args.
+ * Note: the superblock has _not_ yet been read in.
+ *
+ * Note that this function leaks the various device name allocations on
+ * failure.  The caller takes care of them.
+ */
+STATIC int
+xfs_parseargs(
+	struct xfs_mount	*mp,
+	char			*options)
+{
+	struct super_block	*sb = mp->m_super;
+	char			*this_char, *value, *eov;
+	int			dsunit = 0;
+	int			dswidth = 0;
+	int			iosize = 0;
+	__uint8_t		iosizelog = 0;
+
+	/*
+	 * set up the mount name first so all the errors will refer to the
+	 * correct device.
+	 */
+	mp->m_fsname = kstrndup(sb->s_id, MAXNAMELEN, GFP_KERNEL);
+	if (!mp->m_fsname)
+		return ENOMEM;
+	mp->m_fsname_len = strlen(mp->m_fsname) + 1;
+
+	/*
+	 * Copy binary VFS mount flags we are interested in.
+	 */
+	if (sb->s_flags & MS_RDONLY)
+		mp->m_flags |= XFS_MOUNT_RDONLY;
+	if (sb->s_flags & MS_DIRSYNC)
+		mp->m_flags |= XFS_MOUNT_DIRSYNC;
+	if (sb->s_flags & MS_SYNCHRONOUS)
+		mp->m_flags |= XFS_MOUNT_WSYNC;
+
+	/*
+	 * Set some default flags that could be cleared by the mount option
+	 * parsing.
+	 */
+	mp->m_flags |= XFS_MOUNT_BARRIER;
+	mp->m_flags |= XFS_MOUNT_COMPAT_IOSIZE;
+	mp->m_flags |= XFS_MOUNT_SMALL_INUMS;
+	mp->m_flags |= XFS_MOUNT_DELAYLOG;
+
+	/*
+	 * These can be overridden by the mount option parsing.
+	 */
+	mp->m_logbufs = -1;
+	mp->m_logbsize = -1;
+
+	if (!options)
+		goto done;
+
+	while ((this_char = strsep(&options, ",")) != NULL) {
+		if (!*this_char)
+			continue;
+		if ((value = strchr(this_char, '=')) != NULL)
+			*value++ = 0;
+
+		if (!strcmp(this_char, MNTOPT_LOGBUFS)) {
+			if (!value || !*value) {
+				xfs_warn(mp, "%s option requires an argument",
+					this_char);
+				return EINVAL;
+			}
+			mp->m_logbufs = simple_strtoul(value, &eov, 10);
+		} else if (!strcmp(this_char, MNTOPT_LOGBSIZE)) {
+			if (!value || !*value) {
+				xfs_warn(mp, "%s option requires an argument",
+					this_char);
+				return EINVAL;
+			}
+			mp->m_logbsize = suffix_strtoul(value, &eov, 10);
+		} else if (!strcmp(this_char, MNTOPT_LOGDEV)) {
+			if (!value || !*value) {
+				xfs_warn(mp, "%s option requires an argument",
+					this_char);
+				return EINVAL;
+			}
+			mp->m_logname = kstrndup(value, MAXNAMELEN, GFP_KERNEL);
+			if (!mp->m_logname)
+				return ENOMEM;
+		} else if (!strcmp(this_char, MNTOPT_MTPT)) {
+			xfs_warn(mp, "%s option not allowed on this system",
+				this_char);
+			return EINVAL;
+		} else if (!strcmp(this_char, MNTOPT_RTDEV)) {
+			if (!value || !*value) {
+				xfs_warn(mp, "%s option requires an argument",
+					this_char);
+				return EINVAL;
+			}
+			mp->m_rtname = kstrndup(value, MAXNAMELEN, GFP_KERNEL);
+			if (!mp->m_rtname)
+				return ENOMEM;
+		} else if (!strcmp(this_char, MNTOPT_BIOSIZE)) {
+			if (!value || !*value) {
+				xfs_warn(mp, "%s option requires an argument",
+					this_char);
+				return EINVAL;
+			}
+			iosize = simple_strtoul(value, &eov, 10);
+			iosizelog = ffs(iosize) - 1;
+		} else if (!strcmp(this_char, MNTOPT_ALLOCSIZE)) {
+			if (!value || !*value) {
+				xfs_warn(mp, "%s option requires an argument",
+					this_char);
+				return EINVAL;
+			}
+			iosize = suffix_strtoul(value, &eov, 10);
+			iosizelog = ffs(iosize) - 1;
+		} else if (!strcmp(this_char, MNTOPT_GRPID) ||
+			   !strcmp(this_char, MNTOPT_BSDGROUPS)) {
+			mp->m_flags |= XFS_MOUNT_GRPID;
+		} else if (!strcmp(this_char, MNTOPT_NOGRPID) ||
+			   !strcmp(this_char, MNTOPT_SYSVGROUPS)) {
+			mp->m_flags &= ~XFS_MOUNT_GRPID;
+		} else if (!strcmp(this_char, MNTOPT_WSYNC)) {
+			mp->m_flags |= XFS_MOUNT_WSYNC;
+		} else if (!strcmp(this_char, MNTOPT_NORECOVERY)) {
+			mp->m_flags |= XFS_MOUNT_NORECOVERY;
+		} else if (!strcmp(this_char, MNTOPT_NOALIGN)) {
+			mp->m_flags |= XFS_MOUNT_NOALIGN;
+		} else if (!strcmp(this_char, MNTOPT_SWALLOC)) {
+			mp->m_flags |= XFS_MOUNT_SWALLOC;
+		} else if (!strcmp(this_char, MNTOPT_SUNIT)) {
+			if (!value || !*value) {
+				xfs_warn(mp, "%s option requires an argument",
+					this_char);
+				return EINVAL;
+			}
+			dsunit = simple_strtoul(value, &eov, 10);
+		} else if (!strcmp(this_char, MNTOPT_SWIDTH)) {
+			if (!value || !*value) {
+				xfs_warn(mp, "%s option requires an argument",
+					this_char);
+				return EINVAL;
+			}
+			dswidth = simple_strtoul(value, &eov, 10);
+		} else if (!strcmp(this_char, MNTOPT_64BITINODE)) {
+			mp->m_flags &= ~XFS_MOUNT_SMALL_INUMS;
+#if !XFS_BIG_INUMS
+			xfs_warn(mp, "%s option not allowed on this system",
+				this_char);
+			return EINVAL;
+#endif
+		} else if (!strcmp(this_char, MNTOPT_NOUUID)) {
+			mp->m_flags |= XFS_MOUNT_NOUUID;
+		} else if (!strcmp(this_char, MNTOPT_BARRIER)) {
+			mp->m_flags |= XFS_MOUNT_BARRIER;
+		} else if (!strcmp(this_char, MNTOPT_NOBARRIER)) {
+			mp->m_flags &= ~XFS_MOUNT_BARRIER;
+		} else if (!strcmp(this_char, MNTOPT_IKEEP)) {
+			mp->m_flags |= XFS_MOUNT_IKEEP;
+		} else if (!strcmp(this_char, MNTOPT_NOIKEEP)) {
+			mp->m_flags &= ~XFS_MOUNT_IKEEP;
+		} else if (!strcmp(this_char, MNTOPT_LARGEIO)) {
+			mp->m_flags &= ~XFS_MOUNT_COMPAT_IOSIZE;
+		} else if (!strcmp(this_char, MNTOPT_NOLARGEIO)) {
+			mp->m_flags |= XFS_MOUNT_COMPAT_IOSIZE;
+		} else if (!strcmp(this_char, MNTOPT_ATTR2)) {
+			mp->m_flags |= XFS_MOUNT_ATTR2;
+		} else if (!strcmp(this_char, MNTOPT_NOATTR2)) {
+			mp->m_flags &= ~XFS_MOUNT_ATTR2;
+			mp->m_flags |= XFS_MOUNT_NOATTR2;
+		} else if (!strcmp(this_char, MNTOPT_FILESTREAM)) {
+			mp->m_flags |= XFS_MOUNT_FILESTREAMS;
+		} else if (!strcmp(this_char, MNTOPT_NOQUOTA)) {
+			mp->m_qflags &= ~(XFS_UQUOTA_ACCT | XFS_UQUOTA_ACTIVE |
+					  XFS_GQUOTA_ACCT | XFS_GQUOTA_ACTIVE |
+					  XFS_PQUOTA_ACCT | XFS_PQUOTA_ACTIVE |
+					  XFS_UQUOTA_ENFD | XFS_OQUOTA_ENFD);
+		} else if (!strcmp(this_char, MNTOPT_QUOTA) ||
+			   !strcmp(this_char, MNTOPT_UQUOTA) ||
+			   !strcmp(this_char, MNTOPT_USRQUOTA)) {
+			mp->m_qflags |= (XFS_UQUOTA_ACCT | XFS_UQUOTA_ACTIVE |
+					 XFS_UQUOTA_ENFD);
+		} else if (!strcmp(this_char, MNTOPT_QUOTANOENF) ||
+			   !strcmp(this_char, MNTOPT_UQUOTANOENF)) {
+			mp->m_qflags |= (XFS_UQUOTA_ACCT | XFS_UQUOTA_ACTIVE);
+			mp->m_qflags &= ~XFS_UQUOTA_ENFD;
+		} else if (!strcmp(this_char, MNTOPT_PQUOTA) ||
+			   !strcmp(this_char, MNTOPT_PRJQUOTA)) {
+			mp->m_qflags |= (XFS_PQUOTA_ACCT | XFS_PQUOTA_ACTIVE |
+					 XFS_OQUOTA_ENFD);
+		} else if (!strcmp(this_char, MNTOPT_PQUOTANOENF)) {
+			mp->m_qflags |= (XFS_PQUOTA_ACCT | XFS_PQUOTA_ACTIVE);
+			mp->m_qflags &= ~XFS_OQUOTA_ENFD;
+		} else if (!strcmp(this_char, MNTOPT_GQUOTA) ||
+			   !strcmp(this_char, MNTOPT_GRPQUOTA)) {
+			mp->m_qflags |= (XFS_GQUOTA_ACCT | XFS_GQUOTA_ACTIVE |
+					 XFS_OQUOTA_ENFD);
+		} else if (!strcmp(this_char, MNTOPT_GQUOTANOENF)) {
+			mp->m_qflags |= (XFS_GQUOTA_ACCT | XFS_GQUOTA_ACTIVE);
+			mp->m_qflags &= ~XFS_OQUOTA_ENFD;
+		} else if (!strcmp(this_char, MNTOPT_DELAYLOG)) {
+			mp->m_flags |= XFS_MOUNT_DELAYLOG;
+		} else if (!strcmp(this_char, MNTOPT_NODELAYLOG)) {
+			mp->m_flags &= ~XFS_MOUNT_DELAYLOG;
+		} else if (!strcmp(this_char, MNTOPT_DISCARD)) {
+			mp->m_flags |= XFS_MOUNT_DISCARD;
+		} else if (!strcmp(this_char, MNTOPT_NODISCARD)) {
+			mp->m_flags &= ~XFS_MOUNT_DISCARD;
+		} else if (!strcmp(this_char, "ihashsize")) {
+			xfs_warn(mp,
+	"ihashsize no longer used, option is deprecated.");
+		} else if (!strcmp(this_char, "osyncisdsync")) {
+			xfs_warn(mp,
+	"osyncisdsync has no effect, option is deprecated.");
+		} else if (!strcmp(this_char, "osyncisosync")) {
+			xfs_warn(mp,
+	"osyncisosync has no effect, option is deprecated.");
+		} else if (!strcmp(this_char, "irixsgid")) {
+			xfs_warn(mp,
+	"irixsgid is now a sysctl(2) variable, option is deprecated.");
+		} else {
+			xfs_warn(mp, "unknown mount option [%s].", this_char);
+			return EINVAL;
+		}
+	}
+
+	/*
+	 * no recovery flag requires a read-only mount
+	 */
+	if ((mp->m_flags & XFS_MOUNT_NORECOVERY) &&
+	    !(mp->m_flags & XFS_MOUNT_RDONLY)) {
+		xfs_warn(mp, "no-recovery mounts must be read-only.");
+		return EINVAL;
+	}
+
+	if ((mp->m_flags & XFS_MOUNT_NOALIGN) && (dsunit || dswidth)) {
+		xfs_warn(mp,
+	"sunit and swidth options incompatible with the noalign option");
+		return EINVAL;
+	}
+
+	if ((mp->m_flags & XFS_MOUNT_DISCARD) &&
+	    !(mp->m_flags & XFS_MOUNT_DELAYLOG)) {
+		xfs_warn(mp,
+	"the discard option is incompatible with the nodelaylog option");
+		return EINVAL;
+	}
+
+#ifndef CONFIG_XFS_QUOTA
+	if (XFS_IS_QUOTA_RUNNING(mp)) {
+		xfs_warn(mp, "quota support not available in this kernel.");
+		return EINVAL;
+	}
+#endif
+
+	if ((mp->m_qflags & (XFS_GQUOTA_ACCT | XFS_GQUOTA_ACTIVE)) &&
+	    (mp->m_qflags & (XFS_PQUOTA_ACCT | XFS_PQUOTA_ACTIVE))) {
+		xfs_warn(mp, "cannot mount with both project and group quota");
+		return EINVAL;
+	}
+
+	if ((dsunit && !dswidth) || (!dsunit && dswidth)) {
+		xfs_warn(mp, "sunit and swidth must be specified together");
+		return EINVAL;
+	}
+
+	if (dsunit && (dswidth % dsunit != 0)) {
+		xfs_warn(mp,
+	"stripe width (%d) must be a multiple of the stripe unit (%d)",
+			dswidth, dsunit);
+		return EINVAL;
+	}
+
+done:
+	if (!(mp->m_flags & XFS_MOUNT_NOALIGN)) {
+		/*
+		 * At this point the superblock has not been read
+		 * in, therefore we do not know the block size.
+		 * Before the mount call ends we will convert
+		 * these to FSBs.
+		 */
+		if (dsunit) {
+			mp->m_dalign = dsunit;
+			mp->m_flags |= XFS_MOUNT_RETERR;
+		}
+
+		if (dswidth)
+			mp->m_swidth = dswidth;
+	}
+
+	if (mp->m_logbufs != -1 &&
+	    mp->m_logbufs != 0 &&
+	    (mp->m_logbufs < XLOG_MIN_ICLOGS ||
+	     mp->m_logbufs > XLOG_MAX_ICLOGS)) {
+		xfs_warn(mp, "invalid logbufs value: %d [not %d-%d]",
+			mp->m_logbufs, XLOG_MIN_ICLOGS, XLOG_MAX_ICLOGS);
+		return XFS_ERROR(EINVAL);
+	}
+	if (mp->m_logbsize != -1 &&
+	    mp->m_logbsize !=  0 &&
+	    (mp->m_logbsize < XLOG_MIN_RECORD_BSIZE ||
+	     mp->m_logbsize > XLOG_MAX_RECORD_BSIZE ||
+	     !is_power_of_2(mp->m_logbsize))) {
+		xfs_warn(mp,
+			"invalid logbufsize: %d [not 16k,32k,64k,128k or 256k]",
+			mp->m_logbsize);
+		return XFS_ERROR(EINVAL);
+	}
+
+	if (iosizelog) {
+		if (iosizelog > XFS_MAX_IO_LOG ||
+		    iosizelog < XFS_MIN_IO_LOG) {
+			xfs_warn(mp, "invalid log iosize: %d [not %d-%d]",
+				iosizelog, XFS_MIN_IO_LOG,
+				XFS_MAX_IO_LOG);
+			return XFS_ERROR(EINVAL);
+		}
+
+		mp->m_flags |= XFS_MOUNT_DFLT_IOSIZE;
+		mp->m_readio_log = iosizelog;
+		mp->m_writeio_log = iosizelog;
+	}
+
+	return 0;
+}
+
+struct proc_xfs_info {
+	int	flag;
+	char	*str;
+};
+
+STATIC int
+xfs_showargs(
+	struct xfs_mount	*mp,
+	struct seq_file		*m)
+{
+	static struct proc_xfs_info xfs_info_set[] = {
+		/* the few simple ones we can get from the mount struct */
+		{ XFS_MOUNT_IKEEP,		"," MNTOPT_IKEEP },
+		{ XFS_MOUNT_WSYNC,		"," MNTOPT_WSYNC },
+		{ XFS_MOUNT_NOALIGN,		"," MNTOPT_NOALIGN },
+		{ XFS_MOUNT_SWALLOC,		"," MNTOPT_SWALLOC },
+		{ XFS_MOUNT_NOUUID,		"," MNTOPT_NOUUID },
+		{ XFS_MOUNT_NORECOVERY,		"," MNTOPT_NORECOVERY },
+		{ XFS_MOUNT_ATTR2,		"," MNTOPT_ATTR2 },
+		{ XFS_MOUNT_FILESTREAMS,	"," MNTOPT_FILESTREAM },
+		{ XFS_MOUNT_GRPID,		"," MNTOPT_GRPID },
+		{ XFS_MOUNT_DELAYLOG,		"," MNTOPT_DELAYLOG },
+		{ XFS_MOUNT_DISCARD,		"," MNTOPT_DISCARD },
+		{ 0, NULL }
+	};
+	static struct proc_xfs_info xfs_info_unset[] = {
+		/* the few simple ones we can get from the mount struct */
+		{ XFS_MOUNT_COMPAT_IOSIZE,	"," MNTOPT_LARGEIO },
+		{ XFS_MOUNT_BARRIER,		"," MNTOPT_NOBARRIER },
+		{ XFS_MOUNT_SMALL_INUMS,	"," MNTOPT_64BITINODE },
+		{ 0, NULL }
+	};
+	struct proc_xfs_info	*xfs_infop;
+
+	for (xfs_infop = xfs_info_set; xfs_infop->flag; xfs_infop++) {
+		if (mp->m_flags & xfs_infop->flag)
+			seq_puts(m, xfs_infop->str);
+	}
+	for (xfs_infop = xfs_info_unset; xfs_infop->flag; xfs_infop++) {
+		if (!(mp->m_flags & xfs_infop->flag))
+			seq_puts(m, xfs_infop->str);
+	}
+
+	if (mp->m_flags & XFS_MOUNT_DFLT_IOSIZE)
+		seq_printf(m, "," MNTOPT_ALLOCSIZE "=%dk",
+				(int)(1 << mp->m_writeio_log) >> 10);
+
+	if (mp->m_logbufs > 0)
+		seq_printf(m, "," MNTOPT_LOGBUFS "=%d", mp->m_logbufs);
+	if (mp->m_logbsize > 0)
+		seq_printf(m, "," MNTOPT_LOGBSIZE "=%dk", mp->m_logbsize >> 10);
+
+	if (mp->m_logname)
+		seq_printf(m, "," MNTOPT_LOGDEV "=%s", mp->m_logname);
+	if (mp->m_rtname)
+		seq_printf(m, "," MNTOPT_RTDEV "=%s", mp->m_rtname);
+
+	if (mp->m_dalign > 0)
+		seq_printf(m, "," MNTOPT_SUNIT "=%d",
+				(int)XFS_FSB_TO_BB(mp, mp->m_dalign));
+	if (mp->m_swidth > 0)
+		seq_printf(m, "," MNTOPT_SWIDTH "=%d",
+				(int)XFS_FSB_TO_BB(mp, mp->m_swidth));
+
+	if (mp->m_qflags & (XFS_UQUOTA_ACCT|XFS_UQUOTA_ENFD))
+		seq_puts(m, "," MNTOPT_USRQUOTA);
+	else if (mp->m_qflags & XFS_UQUOTA_ACCT)
+		seq_puts(m, "," MNTOPT_UQUOTANOENF);
+
+	/* Either project or group quotas can be active, not both */
+
+	if (mp->m_qflags & XFS_PQUOTA_ACCT) {
+		if (mp->m_qflags & XFS_OQUOTA_ENFD)
+			seq_puts(m, "," MNTOPT_PRJQUOTA);
+		else
+			seq_puts(m, "," MNTOPT_PQUOTANOENF);
+	} else if (mp->m_qflags & XFS_GQUOTA_ACCT) {
+		if (mp->m_qflags & XFS_OQUOTA_ENFD)
+			seq_puts(m, "," MNTOPT_GRPQUOTA);
+		else
+			seq_puts(m, "," MNTOPT_GQUOTANOENF);
+	}
+
+	if (!(mp->m_qflags & XFS_ALL_QUOTA_ACCT))
+		seq_puts(m, "," MNTOPT_NOQUOTA);
+
+	return 0;
+}
+__uint64_t
+xfs_max_file_offset(
+	unsigned int		blockshift)
+{
+	unsigned int		pagefactor = 1;
+	unsigned int		bitshift = BITS_PER_LONG - 1;
+
+	/* Figure out maximum filesize, on Linux this can depend on
+	 * the filesystem blocksize (on 32 bit platforms).
+	 * __block_write_begin does this in an [unsigned] long...
+	 *      page->index << (PAGE_CACHE_SHIFT - bbits)
+	 * So, for page sized blocks (4K on 32 bit platforms),
+	 * this wraps at around 8Tb (hence MAX_LFS_FILESIZE which is
+	 *      (((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
+	 * but for smaller blocksizes it is less (bbits = log2 bsize).
+	 * Note1: get_block_t takes a long (implicit cast from above)
+	 * Note2: The Large Block Device (LBD and HAVE_SECTOR_T) patch
+	 * can optionally convert the [unsigned] long from above into
+	 * an [unsigned] long long.
+	 */
+
+#if BITS_PER_LONG == 32
+# if defined(CONFIG_LBDAF)
+	ASSERT(sizeof(sector_t) == 8);
+	pagefactor = PAGE_CACHE_SIZE;
+	bitshift = BITS_PER_LONG;
+# else
+	pagefactor = PAGE_CACHE_SIZE >> (PAGE_CACHE_SHIFT - blockshift);
+# endif
+#endif
+
+	return (((__uint64_t)pagefactor) << bitshift) - 1;
+}
+
+STATIC int
+xfs_blkdev_get(
+	xfs_mount_t		*mp,
+	const char		*name,
+	struct block_device	**bdevp)
+{
+	int			error = 0;
+
+	*bdevp = blkdev_get_by_path(name, FMODE_READ|FMODE_WRITE|FMODE_EXCL,
+				    mp);
+	if (IS_ERR(*bdevp)) {
+		error = PTR_ERR(*bdevp);
+		xfs_warn(mp, "Invalid device [%s], error=%d\n", name, error);
+	}
+
+	return -error;
+}
+
+STATIC void
+xfs_blkdev_put(
+	struct block_device	*bdev)
+{
+	if (bdev)
+		blkdev_put(bdev, FMODE_READ|FMODE_WRITE|FMODE_EXCL);
+}
+
+void
+xfs_blkdev_issue_flush(
+	xfs_buftarg_t		*buftarg)
+{
+	blkdev_issue_flush(buftarg->bt_bdev, GFP_KERNEL, NULL);
+}
+
+STATIC void
+xfs_close_devices(
+	struct xfs_mount	*mp)
+{
+	if (mp->m_logdev_targp && mp->m_logdev_targp != mp->m_ddev_targp) {
+		struct block_device *logdev = mp->m_logdev_targp->bt_bdev;
+		xfs_free_buftarg(mp, mp->m_logdev_targp);
+		xfs_blkdev_put(logdev);
+	}
+	if (mp->m_rtdev_targp) {
+		struct block_device *rtdev = mp->m_rtdev_targp->bt_bdev;
+		xfs_free_buftarg(mp, mp->m_rtdev_targp);
+		xfs_blkdev_put(rtdev);
+	}
+	xfs_free_buftarg(mp, mp->m_ddev_targp);
+}
+
+/*
+ * The file system configurations are:
+ *	(1) device (partition) with data and internal log
+ *	(2) logical volume with data and log subvolumes.
+ *	(3) logical volume with data, log, and realtime subvolumes.
+ *
+ * We only have to handle opening the log and realtime volumes here if
+ * they are present.  The data subvolume has already been opened by
+ * get_sb_bdev() and is stored in sb->s_bdev.
+ */
+STATIC int
+xfs_open_devices(
+	struct xfs_mount	*mp)
+{
+	struct block_device	*ddev = mp->m_super->s_bdev;
+	struct block_device	*logdev = NULL, *rtdev = NULL;
+	int			error;
+
+	/*
+	 * Open real time and log devices - order is important.
+	 */
+	if (mp->m_logname) {
+		error = xfs_blkdev_get(mp, mp->m_logname, &logdev);
+		if (error)
+			goto out;
+	}
+
+	if (mp->m_rtname) {
+		error = xfs_blkdev_get(mp, mp->m_rtname, &rtdev);
+		if (error)
+			goto out_close_logdev;
+
+		if (rtdev == ddev || rtdev == logdev) {
+			xfs_warn(mp,
+	"Cannot mount filesystem with identical rtdev and ddev/logdev.");
+			error = EINVAL;
+			goto out_close_rtdev;
+		}
+	}
+
+	/*
+	 * Setup xfs_mount buffer target pointers
+	 */
+	error = ENOMEM;
+	mp->m_ddev_targp = xfs_alloc_buftarg(mp, ddev, 0, mp->m_fsname);
+	if (!mp->m_ddev_targp)
+		goto out_close_rtdev;
+
+	if (rtdev) {
+		mp->m_rtdev_targp = xfs_alloc_buftarg(mp, rtdev, 1,
+							mp->m_fsname);
+		if (!mp->m_rtdev_targp)
+			goto out_free_ddev_targ;
+	}
+
+	if (logdev && logdev != ddev) {
+		mp->m_logdev_targp = xfs_alloc_buftarg(mp, logdev, 1,
+							mp->m_fsname);
+		if (!mp->m_logdev_targp)
+			goto out_free_rtdev_targ;
+	} else {
+		mp->m_logdev_targp = mp->m_ddev_targp;
+	}
+
+	return 0;
+
+ out_free_rtdev_targ:
+	if (mp->m_rtdev_targp)
+		xfs_free_buftarg(mp, mp->m_rtdev_targp);
+ out_free_ddev_targ:
+	xfs_free_buftarg(mp, mp->m_ddev_targp);
+ out_close_rtdev:
+	if (rtdev)
+		xfs_blkdev_put(rtdev);
+ out_close_logdev:
+	if (logdev && logdev != ddev)
+		xfs_blkdev_put(logdev);
+ out:
+	return error;
+}
+
+/*
+ * Setup xfs_mount buffer target pointers based on superblock
+ */
+STATIC int
+xfs_setup_devices(
+	struct xfs_mount	*mp)
+{
+	int			error;
+
+	error = xfs_setsize_buftarg(mp->m_ddev_targp, mp->m_sb.sb_blocksize,
+				    mp->m_sb.sb_sectsize);
+	if (error)
+		return error;
+
+	if (mp->m_logdev_targp && mp->m_logdev_targp != mp->m_ddev_targp) {
+		unsigned int	log_sector_size = BBSIZE;
+
+		if (xfs_sb_version_hassector(&mp->m_sb))
+			log_sector_size = mp->m_sb.sb_logsectsize;
+		error = xfs_setsize_buftarg(mp->m_logdev_targp,
+					    mp->m_sb.sb_blocksize,
+					    log_sector_size);
+		if (error)
+			return error;
+	}
+	if (mp->m_rtdev_targp) {
+		error = xfs_setsize_buftarg(mp->m_rtdev_targp,
+					    mp->m_sb.sb_blocksize,
+					    mp->m_sb.sb_sectsize);
+		if (error)
+			return error;
+	}
+
+	return 0;
+}
+
+/* Catch misguided souls that try to use this interface on XFS */
+STATIC struct inode *
+xfs_fs_alloc_inode(
+	struct super_block	*sb)
+{
+	BUG();
+	return NULL;
+}
+
+/*
+ * Now that the generic code is guaranteed not to be accessing
+ * the linux inode, we can reclaim the inode.
+ */
+STATIC void
+xfs_fs_destroy_inode(
+	struct inode		*inode)
+{
+	struct xfs_inode	*ip = XFS_I(inode);
+
+	trace_xfs_destroy_inode(ip);
+
+	XFS_STATS_INC(vn_reclaim);
+
+	/* bad inode, get out here ASAP */
+	if (is_bad_inode(inode))
+		goto out_reclaim;
+
+	xfs_ioend_wait(ip);
+
+	ASSERT(XFS_FORCED_SHUTDOWN(ip->i_mount) || ip->i_delayed_blks == 0);
+
+	/*
+	 * We should never get here with one of the reclaim flags already set.
+	 */
+	ASSERT_ALWAYS(!xfs_iflags_test(ip, XFS_IRECLAIMABLE));
+	ASSERT_ALWAYS(!xfs_iflags_test(ip, XFS_IRECLAIM));
+
+	/*
+	 * We always use background reclaim here because even if the
+	 * inode is clean, it still may be under IO and hence we have
+	 * to take the flush lock. The background reclaim path handles
+	 * this more efficiently than we can here, so simply let background
+	 * reclaim tear down all inodes.
+	 */
+out_reclaim:
+	xfs_inode_set_reclaim_tag(ip);
+}
+
+/*
+ * Slab object creation initialisation for the XFS inode.
+ * This covers only the idempotent fields in the XFS inode;
+ * all other fields need to be initialised on allocation
+ * from the slab. This avoids the need to repeatedly initialise
+ * fields in the xfs inode that left in the initialise state
+ * when freeing the inode.
+ */
+STATIC void
+xfs_fs_inode_init_once(
+	void			*inode)
+{
+	struct xfs_inode	*ip = inode;
+
+	memset(ip, 0, sizeof(struct xfs_inode));
+
+	/* vfs inode */
+	inode_init_once(VFS_I(ip));
+
+	/* xfs inode */
+	atomic_set(&ip->i_iocount, 0);
+	atomic_set(&ip->i_pincount, 0);
+	spin_lock_init(&ip->i_flags_lock);
+	init_waitqueue_head(&ip->i_ipin_wait);
+	/*
+	 * Because we want to use a counting completion, complete
+	 * the flush completion once to allow a single access to
+	 * the flush completion without blocking.
+	 */
+	init_completion(&ip->i_flush);
+	complete(&ip->i_flush);
+
+	mrlock_init(&ip->i_lock, MRLOCK_ALLOW_EQUAL_PRI|MRLOCK_BARRIER,
+		     "xfsino", ip->i_ino);
+}
+
+/*
+ * Dirty the XFS inode when mark_inode_dirty_sync() is called so that
+ * we catch unlogged VFS level updates to the inode.
+ *
+ * We need the barrier() to maintain correct ordering between unlogged
+ * updates and the transaction commit code that clears the i_update_core
+ * field. This requires all updates to be completed before marking the
+ * inode dirty.
+ */
+STATIC void
+xfs_fs_dirty_inode(
+	struct inode	*inode,
+	int		flags)
+{
+	barrier();
+	XFS_I(inode)->i_update_core = 1;
+}
+
+STATIC int
+xfs_fs_write_inode(
+	struct inode		*inode,
+	struct writeback_control *wbc)
+{
+	struct xfs_inode	*ip = XFS_I(inode);
+	struct xfs_mount	*mp = ip->i_mount;
+	int			error = EAGAIN;
+
+	trace_xfs_write_inode(ip);
+
+	if (XFS_FORCED_SHUTDOWN(mp))
+		return -XFS_ERROR(EIO);
+
+	if (wbc->sync_mode == WB_SYNC_ALL || wbc->for_kupdate) {
+		/*
+		 * Make sure the inode has made it it into the log.  Instead
+		 * of forcing it all the way to stable storage using a
+		 * synchronous transaction we let the log force inside the
+		 * ->sync_fs call do that for thus, which reduces the number
+		 * of synchronous log foces dramatically.
+		 */
+		xfs_ioend_wait(ip);
+		error = xfs_log_dirty_inode(ip, NULL, 0);
+		if (error)
+			goto out;
+		return 0;
+	} else {
+		if (!ip->i_update_core)
+			return 0;
+
+		/*
+		 * We make this non-blocking if the inode is contended, return
+		 * EAGAIN to indicate to the caller that they did not succeed.
+		 * This prevents the flush path from blocking on inodes inside
+		 * another operation right now, they get caught later by
+		 * xfs_sync.
+		 */
+		if (!xfs_ilock_nowait(ip, XFS_ILOCK_SHARED))
+			goto out;
+
+		if (xfs_ipincount(ip) || !xfs_iflock_nowait(ip))
+			goto out_unlock;
+
+		/*
+		 * Now we have the flush lock and the inode is not pinned, we
+		 * can check if the inode is really clean as we know that
+		 * there are no pending transaction completions, it is not
+		 * waiting on the delayed write queue and there is no IO in
+		 * progress.
+		 */
+		if (xfs_inode_clean(ip)) {
+			xfs_ifunlock(ip);
+			error = 0;
+			goto out_unlock;
+		}
+		error = xfs_iflush(ip, SYNC_TRYLOCK);
+	}
+
+ out_unlock:
+	xfs_iunlock(ip, XFS_ILOCK_SHARED);
+ out:
+	/*
+	 * if we failed to write out the inode then mark
+	 * it dirty again so we'll try again later.
+	 */
+	if (error)
+		xfs_mark_inode_dirty_sync(ip);
+	return -error;
+}
+
+STATIC void
+xfs_fs_evict_inode(
+	struct inode		*inode)
+{
+	xfs_inode_t		*ip = XFS_I(inode);
+
+	trace_xfs_evict_inode(ip);
+
+	truncate_inode_pages(&inode->i_data, 0);
+	end_writeback(inode);
+	XFS_STATS_INC(vn_rele);
+	XFS_STATS_INC(vn_remove);
+	XFS_STATS_DEC(vn_active);
+
+	/*
+	 * The iolock is used by the file system to coordinate reads,
+	 * writes, and block truncates.  Up to this point the lock
+	 * protected concurrent accesses by users of the inode.  But
+	 * from here forward we're doing some final processing of the
+	 * inode because we're done with it, and although we reuse the
+	 * iolock for protection it is really a distinct lock class
+	 * (in the lockdep sense) from before.  To keep lockdep happy
+	 * (and basically indicate what we are doing), we explicitly
+	 * re-init the iolock here.
+	 */
+	ASSERT(!rwsem_is_locked(&ip->i_iolock.mr_lock));
+	mrlock_init(&ip->i_iolock, MRLOCK_BARRIER, "xfsio", ip->i_ino);
+	lockdep_set_class_and_name(&ip->i_iolock.mr_lock,
+			&xfs_iolock_reclaimable, "xfs_iolock_reclaimable");
+
+	xfs_inactive(ip);
+}
+
+STATIC void
+xfs_free_fsname(
+	struct xfs_mount	*mp)
+{
+	kfree(mp->m_fsname);
+	kfree(mp->m_rtname);
+	kfree(mp->m_logname);
+}
+
+STATIC void
+xfs_fs_put_super(
+	struct super_block	*sb)
+{
+	struct xfs_mount	*mp = XFS_M(sb);
+
+	/*
+	 * Unregister the memory shrinker before we tear down the mount
+	 * structure so we don't have memory reclaim racing with us here.
+	 */
+	xfs_inode_shrinker_unregister(mp);
+	xfs_syncd_stop(mp);
+
+	/*
+	 * Blow away any referenced inode in the filestreams cache.
+	 * This can and will cause log traffic as inodes go inactive
+	 * here.
+	 */
+	xfs_filestream_unmount(mp);
+
+	XFS_bflush(mp->m_ddev_targp);
+
+	xfs_unmountfs(mp);
+	xfs_freesb(mp);
+	xfs_icsb_destroy_counters(mp);
+	xfs_close_devices(mp);
+	xfs_free_fsname(mp);
+	kfree(mp);
+}
+
+STATIC int
+xfs_fs_sync_fs(
+	struct super_block	*sb,
+	int			wait)
+{
+	struct xfs_mount	*mp = XFS_M(sb);
+	int			error;
+
+	/*
+	 * Not much we can do for the first async pass.  Writing out the
+	 * superblock would be counter-productive as we are going to redirty
+	 * when writing out other data and metadata (and writing out a single
+	 * block is quite fast anyway).
+	 *
+	 * Try to asynchronously kick off quota syncing at least.
+	 */
+	if (!wait) {
+		xfs_qm_sync(mp, SYNC_TRYLOCK);
+		return 0;
+	}
+
+	error = xfs_quiesce_data(mp);
+	if (error)
+		return -error;
+
+	if (laptop_mode) {
+		/*
+		 * The disk must be active because we're syncing.
+		 * We schedule xfssyncd now (now that the disk is
+		 * active) instead of later (when it might not be).
+		 */
+		flush_delayed_work_sync(&mp->m_sync_work);
+	}
+
+	return 0;
+}
+
+STATIC int
+xfs_fs_statfs(
+	struct dentry		*dentry,
+	struct kstatfs		*statp)
+{
+	struct xfs_mount	*mp = XFS_M(dentry->d_sb);
+	xfs_sb_t		*sbp = &mp->m_sb;
+	struct xfs_inode	*ip = XFS_I(dentry->d_inode);
+	__uint64_t		fakeinos, id;
+	xfs_extlen_t		lsize;
+	__int64_t		ffree;
+
+	statp->f_type = XFS_SB_MAGIC;
+	statp->f_namelen = MAXNAMELEN - 1;
+
+	id = huge_encode_dev(mp->m_ddev_targp->bt_dev);
+	statp->f_fsid.val[0] = (u32)id;
+	statp->f_fsid.val[1] = (u32)(id >> 32);
+
+	xfs_icsb_sync_counters(mp, XFS_ICSB_LAZY_COUNT);
+
+	spin_lock(&mp->m_sb_lock);
+	statp->f_bsize = sbp->sb_blocksize;
+	lsize = sbp->sb_logstart ? sbp->sb_logblocks : 0;
+	statp->f_blocks = sbp->sb_dblocks - lsize;
+	statp->f_bfree = statp->f_bavail =
+				sbp->sb_fdblocks - XFS_ALLOC_SET_ASIDE(mp);
+	fakeinos = statp->f_bfree << sbp->sb_inopblog;
+	statp->f_files =
+	    MIN(sbp->sb_icount + fakeinos, (__uint64_t)XFS_MAXINUMBER);
+	if (mp->m_maxicount)
+		statp->f_files = min_t(typeof(statp->f_files),
+					statp->f_files,
+					mp->m_maxicount);
+
+	/* make sure statp->f_ffree does not underflow */
+	ffree = statp->f_files - (sbp->sb_icount - sbp->sb_ifree);
+	statp->f_ffree = max_t(__int64_t, ffree, 0);
+
+	spin_unlock(&mp->m_sb_lock);
+
+	if ((ip->i_d.di_flags & XFS_DIFLAG_PROJINHERIT) ||
+	    ((mp->m_qflags & (XFS_PQUOTA_ACCT|XFS_OQUOTA_ENFD))) ==
+			      (XFS_PQUOTA_ACCT|XFS_OQUOTA_ENFD))
+		xfs_qm_statvfs(ip, statp);
+	return 0;
+}
+
+STATIC void
+xfs_save_resvblks(struct xfs_mount *mp)
+{
+	__uint64_t resblks = 0;
+
+	mp->m_resblks_save = mp->m_resblks;
+	xfs_reserve_blocks(mp, &resblks, NULL);
+}
+
+STATIC void
+xfs_restore_resvblks(struct xfs_mount *mp)
+{
+	__uint64_t resblks;
+
+	if (mp->m_resblks_save) {
+		resblks = mp->m_resblks_save;
+		mp->m_resblks_save = 0;
+	} else
+		resblks = xfs_default_resblks(mp);
+
+	xfs_reserve_blocks(mp, &resblks, NULL);
+}
+
+STATIC int
+xfs_fs_remount(
+	struct super_block	*sb,
+	int			*flags,
+	char			*options)
+{
+	struct xfs_mount	*mp = XFS_M(sb);
+	substring_t		args[MAX_OPT_ARGS];
+	char			*p;
+	int			error;
+
+	while ((p = strsep(&options, ",")) != NULL) {
+		int token;
+
+		if (!*p)
+			continue;
+
+		token = match_token(p, tokens, args);
+		switch (token) {
+		case Opt_barrier:
+			mp->m_flags |= XFS_MOUNT_BARRIER;
+			break;
+		case Opt_nobarrier:
+			mp->m_flags &= ~XFS_MOUNT_BARRIER;
+			break;
+		default:
+			/*
+			 * Logically we would return an error here to prevent
+			 * users from believing they might have changed
+			 * mount options using remount which can't be changed.
+			 *
+			 * But unfortunately mount(8) adds all options from
+			 * mtab and fstab to the mount arguments in some cases
+			 * so we can't blindly reject options, but have to
+			 * check for each specified option if it actually
+			 * differs from the currently set option and only
+			 * reject it if that's the case.
+			 *
+			 * Until that is implemented we return success for
+			 * every remount request, and silently ignore all
+			 * options that we can't actually change.
+			 */
+#if 0
+			xfs_info(mp,
+		"mount option \"%s\" not supported for remount\n", p);
+			return -EINVAL;
+#else
+			break;
+#endif
+		}
+	}
+
+	/* ro -> rw */
+	if ((mp->m_flags & XFS_MOUNT_RDONLY) && !(*flags & MS_RDONLY)) {
+		mp->m_flags &= ~XFS_MOUNT_RDONLY;
+
+		/*
+		 * If this is the first remount to writeable state we
+		 * might have some superblock changes to update.
+		 */
+		if (mp->m_update_flags) {
+			error = xfs_mount_log_sb(mp, mp->m_update_flags);
+			if (error) {
+				xfs_warn(mp, "failed to write sb changes");
+				return error;
+			}
+			mp->m_update_flags = 0;
+		}
+
+		/*
+		 * Fill out the reserve pool if it is empty. Use the stashed
+		 * value if it is non-zero, otherwise go with the default.
+		 */
+		xfs_restore_resvblks(mp);
+	}
+
+	/* rw -> ro */
+	if (!(mp->m_flags & XFS_MOUNT_RDONLY) && (*flags & MS_RDONLY)) {
+		/*
+		 * After we have synced the data but before we sync the
+		 * metadata, we need to free up the reserve block pool so that
+		 * the used block count in the superblock on disk is correct at
+		 * the end of the remount. Stash the current reserve pool size
+		 * so that if we get remounted rw, we can return it to the same
+		 * size.
+		 */
+
+		xfs_quiesce_data(mp);
+		xfs_save_resvblks(mp);
+		xfs_quiesce_attr(mp);
+		mp->m_flags |= XFS_MOUNT_RDONLY;
+	}
+
+	return 0;
+}
+
+/*
+ * Second stage of a freeze. The data is already frozen so we only
+ * need to take care of the metadata. Once that's done write a dummy
+ * record to dirty the log in case of a crash while frozen.
+ */
+STATIC int
+xfs_fs_freeze(
+	struct super_block	*sb)
+{
+	struct xfs_mount	*mp = XFS_M(sb);
+
+	xfs_save_resvblks(mp);
+	xfs_quiesce_attr(mp);
+	return -xfs_fs_log_dummy(mp);
+}
+
+STATIC int
+xfs_fs_unfreeze(
+	struct super_block	*sb)
+{
+	struct xfs_mount	*mp = XFS_M(sb);
+
+	xfs_restore_resvblks(mp);
+	return 0;
+}
+
+STATIC int
+xfs_fs_show_options(
+	struct seq_file		*m,
+	struct vfsmount		*mnt)
+{
+	return -xfs_showargs(XFS_M(mnt->mnt_sb), m);
+}
+
+/*
+ * This function fills in xfs_mount_t fields based on mount args.
+ * Note: the superblock _has_ now been read in.
+ */
+STATIC int
+xfs_finish_flags(
+	struct xfs_mount	*mp)
+{
+	int			ronly = (mp->m_flags & XFS_MOUNT_RDONLY);
+
+	/* Fail a mount where the logbuf is smaller than the log stripe */
+	if (xfs_sb_version_haslogv2(&mp->m_sb)) {
+		if (mp->m_logbsize <= 0 &&
+		    mp->m_sb.sb_logsunit > XLOG_BIG_RECORD_BSIZE) {
+			mp->m_logbsize = mp->m_sb.sb_logsunit;
+		} else if (mp->m_logbsize > 0 &&
+			   mp->m_logbsize < mp->m_sb.sb_logsunit) {
+			xfs_warn(mp,
+		"logbuf size must be greater than or equal to log stripe size");
+			return XFS_ERROR(EINVAL);
+		}
+	} else {
+		/* Fail a mount if the logbuf is larger than 32K */
+		if (mp->m_logbsize > XLOG_BIG_RECORD_BSIZE) {
+			xfs_warn(mp,
+		"logbuf size for version 1 logs must be 16K or 32K");
+			return XFS_ERROR(EINVAL);
+		}
+	}
+
+	/*
+	 * mkfs'ed attr2 will turn on attr2 mount unless explicitly
+	 * told by noattr2 to turn it off
+	 */
+	if (xfs_sb_version_hasattr2(&mp->m_sb) &&
+	    !(mp->m_flags & XFS_MOUNT_NOATTR2))
+		mp->m_flags |= XFS_MOUNT_ATTR2;
+
+	/*
+	 * prohibit r/w mounts of read-only filesystems
+	 */
+	if ((mp->m_sb.sb_flags & XFS_SBF_READONLY) && !ronly) {
+		xfs_warn(mp,
+			"cannot mount a read-only filesystem as read-write");
+		return XFS_ERROR(EROFS);
+	}
+
+	return 0;
+}
+
+STATIC int
+xfs_fs_fill_super(
+	struct super_block	*sb,
+	void			*data,
+	int			silent)
+{
+	struct inode		*root;
+	struct xfs_mount	*mp = NULL;
+	int			flags = 0, error = ENOMEM;
+
+	mp = kzalloc(sizeof(struct xfs_mount), GFP_KERNEL);
+	if (!mp)
+		goto out;
+
+	spin_lock_init(&mp->m_sb_lock);
+	mutex_init(&mp->m_growlock);
+	atomic_set(&mp->m_active_trans, 0);
+
+	mp->m_super = sb;
+	sb->s_fs_info = mp;
+
+	error = xfs_parseargs(mp, (char *)data);
+	if (error)
+		goto out_free_fsname;
+
+	sb_min_blocksize(sb, BBSIZE);
+	sb->s_xattr = xfs_xattr_handlers;
+	sb->s_export_op = &xfs_export_operations;
+#ifdef CONFIG_XFS_QUOTA
+	sb->s_qcop = &xfs_quotactl_operations;
+#endif
+	sb->s_op = &xfs_super_operations;
+
+	if (silent)
+		flags |= XFS_MFSI_QUIET;
+
+	error = xfs_open_devices(mp);
+	if (error)
+		goto out_free_fsname;
+
+	error = xfs_icsb_init_counters(mp);
+	if (error)
+		goto out_close_devices;
+
+	error = xfs_readsb(mp, flags);
+	if (error)
+		goto out_destroy_counters;
+
+	error = xfs_finish_flags(mp);
+	if (error)
+		goto out_free_sb;
+
+	error = xfs_setup_devices(mp);
+	if (error)
+		goto out_free_sb;
+
+	error = xfs_filestream_mount(mp);
+	if (error)
+		goto out_free_sb;
+
+	/*
+	 * we must configure the block size in the superblock before we run the
+	 * full mount process as the mount process can lookup and cache inodes.
+	 * For the same reason we must also initialise the syncd and register
+	 * the inode cache shrinker so that inodes can be reclaimed during
+	 * operations like a quotacheck that iterate all inodes in the
+	 * filesystem.
+	 */
+	sb->s_magic = XFS_SB_MAGIC;
+	sb->s_blocksize = mp->m_sb.sb_blocksize;
+	sb->s_blocksize_bits = ffs(sb->s_blocksize) - 1;
+	sb->s_maxbytes = xfs_max_file_offset(sb->s_blocksize_bits);
+	sb->s_time_gran = 1;
+	set_posix_acl_flag(sb);
+
+	xfs_inode_shrinker_register(mp);
+
+	error = xfs_mountfs(mp);
+	if (error)
+		goto out_filestream_unmount;
+
+	error = xfs_syncd_init(mp);
+	if (error)
+		goto out_unmount;
+
+	root = igrab(VFS_I(mp->m_rootip));
+	if (!root) {
+		error = ENOENT;
+		goto out_syncd_stop;
+	}
+	if (is_bad_inode(root)) {
+		error = EINVAL;
+		goto out_syncd_stop;
+	}
+	sb->s_root = d_alloc_root(root);
+	if (!sb->s_root) {
+		error = ENOMEM;
+		goto out_iput;
+	}
+
+	return 0;
+
+ out_filestream_unmount:
+	xfs_inode_shrinker_unregister(mp);
+	xfs_filestream_unmount(mp);
+ out_free_sb:
+	xfs_freesb(mp);
+ out_destroy_counters:
+	xfs_icsb_destroy_counters(mp);
+ out_close_devices:
+	xfs_close_devices(mp);
+ out_free_fsname:
+	xfs_free_fsname(mp);
+	kfree(mp);
+ out:
+	return -error;
+
+ out_iput:
+	iput(root);
+ out_syncd_stop:
+	xfs_syncd_stop(mp);
+ out_unmount:
+	xfs_inode_shrinker_unregister(mp);
+
+	/*
+	 * Blow away any referenced inode in the filestreams cache.
+	 * This can and will cause log traffic as inodes go inactive
+	 * here.
+	 */
+	xfs_filestream_unmount(mp);
+
+	XFS_bflush(mp->m_ddev_targp);
+
+	xfs_unmountfs(mp);
+	goto out_free_sb;
+}
+
+STATIC struct dentry *
+xfs_fs_mount(
+	struct file_system_type	*fs_type,
+	int			flags,
+	const char		*dev_name,
+	void			*data)
+{
+	return mount_bdev(fs_type, flags, dev_name, data, xfs_fs_fill_super);
+}
+
+static const struct super_operations xfs_super_operations = {
+	.alloc_inode		= xfs_fs_alloc_inode,
+	.destroy_inode		= xfs_fs_destroy_inode,
+	.dirty_inode		= xfs_fs_dirty_inode,
+	.write_inode		= xfs_fs_write_inode,
+	.evict_inode		= xfs_fs_evict_inode,
+	.put_super		= xfs_fs_put_super,
+	.sync_fs		= xfs_fs_sync_fs,
+	.freeze_fs		= xfs_fs_freeze,
+	.unfreeze_fs		= xfs_fs_unfreeze,
+	.statfs			= xfs_fs_statfs,
+	.remount_fs		= xfs_fs_remount,
+	.show_options		= xfs_fs_show_options,
+};
+
+static struct file_system_type xfs_fs_type = {
+	.owner			= THIS_MODULE,
+	.name			= "xfs",
+	.mount			= xfs_fs_mount,
+	.kill_sb		= kill_block_super,
+	.fs_flags		= FS_REQUIRES_DEV,
+};
+
+STATIC int __init
+xfs_init_zones(void)
+{
+
+	xfs_ioend_zone = kmem_zone_init(sizeof(xfs_ioend_t), "xfs_ioend");
+	if (!xfs_ioend_zone)
+		goto out;
+
+	xfs_ioend_pool = mempool_create_slab_pool(4 * MAX_BUF_PER_PAGE,
+						  xfs_ioend_zone);
+	if (!xfs_ioend_pool)
+		goto out_destroy_ioend_zone;
+
+	xfs_log_ticket_zone = kmem_zone_init(sizeof(xlog_ticket_t),
+						"xfs_log_ticket");
+	if (!xfs_log_ticket_zone)
+		goto out_destroy_ioend_pool;
+
+	xfs_bmap_free_item_zone = kmem_zone_init(sizeof(xfs_bmap_free_item_t),
+						"xfs_bmap_free_item");
+	if (!xfs_bmap_free_item_zone)
+		goto out_destroy_log_ticket_zone;
+
+	xfs_btree_cur_zone = kmem_zone_init(sizeof(xfs_btree_cur_t),
+						"xfs_btree_cur");
+	if (!xfs_btree_cur_zone)
+		goto out_destroy_bmap_free_item_zone;
+
+	xfs_da_state_zone = kmem_zone_init(sizeof(xfs_da_state_t),
+						"xfs_da_state");
+	if (!xfs_da_state_zone)
+		goto out_destroy_btree_cur_zone;
+
+	xfs_dabuf_zone = kmem_zone_init(sizeof(xfs_dabuf_t), "xfs_dabuf");
+	if (!xfs_dabuf_zone)
+		goto out_destroy_da_state_zone;
+
+	xfs_ifork_zone = kmem_zone_init(sizeof(xfs_ifork_t), "xfs_ifork");
+	if (!xfs_ifork_zone)
+		goto out_destroy_dabuf_zone;
+
+	xfs_trans_zone = kmem_zone_init(sizeof(xfs_trans_t), "xfs_trans");
+	if (!xfs_trans_zone)
+		goto out_destroy_ifork_zone;
+
+	xfs_log_item_desc_zone =
+		kmem_zone_init(sizeof(struct xfs_log_item_desc),
+			       "xfs_log_item_desc");
+	if (!xfs_log_item_desc_zone)
+		goto out_destroy_trans_zone;
+
+	/*
+	 * The size of the zone allocated buf log item is the maximum
+	 * size possible under XFS.  This wastes a little bit of memory,
+	 * but it is much faster.
+	 */
+	xfs_buf_item_zone = kmem_zone_init((sizeof(xfs_buf_log_item_t) +
+				(((XFS_MAX_BLOCKSIZE / XFS_BLF_CHUNK) /
+				  NBWORD) * sizeof(int))), "xfs_buf_item");
+	if (!xfs_buf_item_zone)
+		goto out_destroy_log_item_desc_zone;
+
+	xfs_efd_zone = kmem_zone_init((sizeof(xfs_efd_log_item_t) +
+			((XFS_EFD_MAX_FAST_EXTENTS - 1) *
+				 sizeof(xfs_extent_t))), "xfs_efd_item");
+	if (!xfs_efd_zone)
+		goto out_destroy_buf_item_zone;
+
+	xfs_efi_zone = kmem_zone_init((sizeof(xfs_efi_log_item_t) +
+			((XFS_EFI_MAX_FAST_EXTENTS - 1) *
+				sizeof(xfs_extent_t))), "xfs_efi_item");
+	if (!xfs_efi_zone)
+		goto out_destroy_efd_zone;
+
+	xfs_inode_zone =
+		kmem_zone_init_flags(sizeof(xfs_inode_t), "xfs_inode",
+			KM_ZONE_HWALIGN | KM_ZONE_RECLAIM | KM_ZONE_SPREAD,
+			xfs_fs_inode_init_once);
+	if (!xfs_inode_zone)
+		goto out_destroy_efi_zone;
+
+	xfs_ili_zone =
+		kmem_zone_init_flags(sizeof(xfs_inode_log_item_t), "xfs_ili",
+					KM_ZONE_SPREAD, NULL);
+	if (!xfs_ili_zone)
+		goto out_destroy_inode_zone;
+
+	return 0;
+
+ out_destroy_inode_zone:
+	kmem_zone_destroy(xfs_inode_zone);
+ out_destroy_efi_zone:
+	kmem_zone_destroy(xfs_efi_zone);
+ out_destroy_efd_zone:
+	kmem_zone_destroy(xfs_efd_zone);
+ out_destroy_buf_item_zone:
+	kmem_zone_destroy(xfs_buf_item_zone);
+ out_destroy_log_item_desc_zone:
+	kmem_zone_destroy(xfs_log_item_desc_zone);
+ out_destroy_trans_zone:
+	kmem_zone_destroy(xfs_trans_zone);
+ out_destroy_ifork_zone:
+	kmem_zone_destroy(xfs_ifork_zone);
+ out_destroy_dabuf_zone:
+	kmem_zone_destroy(xfs_dabuf_zone);
+ out_destroy_da_state_zone:
+	kmem_zone_destroy(xfs_da_state_zone);
+ out_destroy_btree_cur_zone:
+	kmem_zone_destroy(xfs_btree_cur_zone);
+ out_destroy_bmap_free_item_zone:
+	kmem_zone_destroy(xfs_bmap_free_item_zone);
+ out_destroy_log_ticket_zone:
+	kmem_zone_destroy(xfs_log_ticket_zone);
+ out_destroy_ioend_pool:
+	mempool_destroy(xfs_ioend_pool);
+ out_destroy_ioend_zone:
+	kmem_zone_destroy(xfs_ioend_zone);
+ out:
+	return -ENOMEM;
+}
+
+STATIC void
+xfs_destroy_zones(void)
+{
+	kmem_zone_destroy(xfs_ili_zone);
+	kmem_zone_destroy(xfs_inode_zone);
+	kmem_zone_destroy(xfs_efi_zone);
+	kmem_zone_destroy(xfs_efd_zone);
+	kmem_zone_destroy(xfs_buf_item_zone);
+	kmem_zone_destroy(xfs_log_item_desc_zone);
+	kmem_zone_destroy(xfs_trans_zone);
+	kmem_zone_destroy(xfs_ifork_zone);
+	kmem_zone_destroy(xfs_dabuf_zone);
+	kmem_zone_destroy(xfs_da_state_zone);
+	kmem_zone_destroy(xfs_btree_cur_zone);
+	kmem_zone_destroy(xfs_bmap_free_item_zone);
+	kmem_zone_destroy(xfs_log_ticket_zone);
+	mempool_destroy(xfs_ioend_pool);
+	kmem_zone_destroy(xfs_ioend_zone);
+
+}
+
+STATIC int __init
+xfs_init_workqueues(void)
+{
+	/*
+	 * max_active is set to 8 to give enough concurency to allow
+	 * multiple work operations on each CPU to run. This allows multiple
+	 * filesystems to be running sync work concurrently, and scales with
+	 * the number of CPUs in the system.
+	 */
+	xfs_syncd_wq = alloc_workqueue("xfssyncd", WQ_CPU_INTENSIVE, 8);
+	if (!xfs_syncd_wq)
+		return -ENOMEM;
+	return 0;
+}
+
+STATIC void
+xfs_destroy_workqueues(void)
+{
+	destroy_workqueue(xfs_syncd_wq);
+}
+
+STATIC int __init
+init_xfs_fs(void)
+{
+	int			error;
+
+	printk(KERN_INFO XFS_VERSION_STRING " with "
+			 XFS_BUILD_OPTIONS " enabled\n");
+
+	xfs_ioend_init();
+	xfs_dir_startup();
+
+	error = xfs_init_zones();
+	if (error)
+		goto out;
+
+	error = xfs_init_workqueues();
+	if (error)
+		goto out_destroy_zones;
+
+	error = xfs_mru_cache_init();
+	if (error)
+		goto out_destroy_wq;
+
+	error = xfs_filestream_init();
+	if (error)
+		goto out_mru_cache_uninit;
+
+	error = xfs_buf_init();
+	if (error)
+		goto out_filestream_uninit;
+
+	error = xfs_init_procfs();
+	if (error)
+		goto out_buf_terminate;
+
+	error = xfs_sysctl_register();
+	if (error)
+		goto out_cleanup_procfs;
+
+	vfs_initquota();
+
+	error = register_filesystem(&xfs_fs_type);
+	if (error)
+		goto out_sysctl_unregister;
+	return 0;
+
+ out_sysctl_unregister:
+	xfs_sysctl_unregister();
+ out_cleanup_procfs:
+	xfs_cleanup_procfs();
+ out_buf_terminate:
+	xfs_buf_terminate();
+ out_filestream_uninit:
+	xfs_filestream_uninit();
+ out_mru_cache_uninit:
+	xfs_mru_cache_uninit();
+ out_destroy_wq:
+	xfs_destroy_workqueues();
+ out_destroy_zones:
+	xfs_destroy_zones();
+ out:
+	return error;
+}
+
+STATIC void __exit
+exit_xfs_fs(void)
+{
+	vfs_exitquota();
+	unregister_filesystem(&xfs_fs_type);
+	xfs_sysctl_unregister();
+	xfs_cleanup_procfs();
+	xfs_buf_terminate();
+	xfs_filestream_uninit();
+	xfs_mru_cache_uninit();
+	xfs_destroy_workqueues();
+	xfs_destroy_zones();
+}
+
+module_init(init_xfs_fs);
+module_exit(exit_xfs_fs);
+
+MODULE_AUTHOR("Silicon Graphics, Inc.");
+MODULE_DESCRIPTION(XFS_VERSION_STRING " with " XFS_BUILD_OPTIONS " enabled");
+MODULE_LICENSE("GPL");
diff --git a/fs/xfs/linux-2.6/xfs_super.h b/fs/xfs/linux-2.6/xfs_super.h
new file mode 100644
index 00000000..50a3266c
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_super.h
@@ -0,0 +1,87 @@
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_SUPER_H__
+#define __XFS_SUPER_H__
+
+#include <linux/exportfs.h>
+
+#ifdef CONFIG_XFS_QUOTA
+extern void xfs_qm_init(void);
+extern void xfs_qm_exit(void);
+# define vfs_initquota()	xfs_qm_init()
+# define vfs_exitquota()	xfs_qm_exit()
+#else
+# define vfs_initquota()	do { } while (0)
+# define vfs_exitquota()	do { } while (0)
+#endif
+
+#ifdef CONFIG_XFS_POSIX_ACL
+# define XFS_ACL_STRING		"ACLs, "
+# define set_posix_acl_flag(sb)	((sb)->s_flags |= MS_POSIXACL)
+#else
+# define XFS_ACL_STRING
+# define set_posix_acl_flag(sb)	do { } while (0)
+#endif
+
+#define XFS_SECURITY_STRING	"security attributes, "
+
+#ifdef CONFIG_XFS_RT
+# define XFS_REALTIME_STRING	"realtime, "
+#else
+# define XFS_REALTIME_STRING
+#endif
+
+#if XFS_BIG_BLKNOS
+# if XFS_BIG_INUMS
+#  define XFS_BIGFS_STRING	"large block/inode numbers, "
+# else
+#  define XFS_BIGFS_STRING	"large block numbers, "
+# endif
+#else
+# define XFS_BIGFS_STRING
+#endif
+
+#ifdef DEBUG
+# define XFS_DBG_STRING		"debug"
+#else
+# define XFS_DBG_STRING		"no debug"
+#endif
+
+#define XFS_VERSION_STRING	"SGI XFS"
+#define XFS_BUILD_OPTIONS	XFS_ACL_STRING \
+				XFS_SECURITY_STRING \
+				XFS_REALTIME_STRING \
+				XFS_BIGFS_STRING \
+				XFS_DBG_STRING /* DBG must be last */
+
+struct xfs_inode;
+struct xfs_mount;
+struct xfs_buftarg;
+struct block_device;
+
+extern __uint64_t xfs_max_file_offset(unsigned int);
+
+extern void xfs_blkdev_issue_flush(struct xfs_buftarg *);
+
+extern const struct export_operations xfs_export_operations;
+extern const struct xattr_handler *xfs_xattr_handlers[];
+extern const struct quotactl_ops xfs_quotactl_operations;
+
+#define XFS_M(sb)		((struct xfs_mount *)((sb)->s_fs_info))
+
+#endif	/* __XFS_SUPER_H__ */
diff --git a/fs/xfs/linux-2.6/xfs_sync.c b/fs/xfs/linux-2.6/xfs_sync.c
new file mode 100644
index 00000000..2f277a04
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_sync.c
@@ -0,0 +1,1132 @@
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_trans_priv.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_mount.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_inode.h"
+#include "xfs_dinode.h"
+#include "xfs_error.h"
+#include "xfs_filestream.h"
+#include "xfs_vnodeops.h"
+#include "xfs_inode_item.h"
+#include "xfs_quota.h"
+#include "xfs_trace.h"
+#include "xfs_fsops.h"
+
+#include <linux/kthread.h>
+#include <linux/freezer.h>
+
+struct workqueue_struct	*xfs_syncd_wq;	/* sync workqueue */
+
+/*
+ * The inode lookup is done in batches to keep the amount of lock traffic and
+ * radix tree lookups to a minimum. The batch size is a trade off between
+ * lookup reduction and stack usage. This is in the reclaim path, so we can't
+ * be too greedy.
+ */
+#define XFS_LOOKUP_BATCH	32
+
+STATIC int
+xfs_inode_ag_walk_grab(
+	struct xfs_inode	*ip)
+{
+	struct inode		*inode = VFS_I(ip);
+
+	ASSERT(rcu_read_lock_held());
+
+	/*
+	 * check for stale RCU freed inode
+	 *
+	 * If the inode has been reallocated, it doesn't matter if it's not in
+	 * the AG we are walking - we are walking for writeback, so if it
+	 * passes all the "valid inode" checks and is dirty, then we'll write
+	 * it back anyway.  If it has been reallocated and still being
+	 * initialised, the XFS_INEW check below will catch it.
+	 */
+	spin_lock(&ip->i_flags_lock);
+	if (!ip->i_ino)
+		goto out_unlock_noent;
+
+	/* avoid new or reclaimable inodes. Leave for reclaim code to flush */
+	if (__xfs_iflags_test(ip, XFS_INEW | XFS_IRECLAIMABLE | XFS_IRECLAIM))
+		goto out_unlock_noent;
+	spin_unlock(&ip->i_flags_lock);
+
+	/* nothing to sync during shutdown */
+	if (XFS_FORCED_SHUTDOWN(ip->i_mount))
+		return EFSCORRUPTED;
+
+	/* If we can't grab the inode, it must on it's way to reclaim. */
+	if (!igrab(inode))
+		return ENOENT;
+
+	if (is_bad_inode(inode)) {
+		IRELE(ip);
+		return ENOENT;
+	}
+
+	/* inode is valid */
+	return 0;
+
+out_unlock_noent:
+	spin_unlock(&ip->i_flags_lock);
+	return ENOENT;
+}
+
+STATIC int
+xfs_inode_ag_walk(
+	struct xfs_mount	*mp,
+	struct xfs_perag	*pag,
+	int			(*execute)(struct xfs_inode *ip,
+					   struct xfs_perag *pag, int flags),
+	int			flags)
+{
+	uint32_t		first_index;
+	int			last_error = 0;
+	int			skipped;
+	int			done;
+	int			nr_found;
+
+restart:
+	done = 0;
+	skipped = 0;
+	first_index = 0;
+	nr_found = 0;
+	do {
+		struct xfs_inode *batch[XFS_LOOKUP_BATCH];
+		int		error = 0;
+		int		i;
+
+		rcu_read_lock();
+		nr_found = radix_tree_gang_lookup(&pag->pag_ici_root,
+					(void **)batch, first_index,
+					XFS_LOOKUP_BATCH);
+		if (!nr_found) {
+			rcu_read_unlock();
+			break;
+		}
+
+		/*
+		 * Grab the inodes before we drop the lock. if we found
+		 * nothing, nr == 0 and the loop will be skipped.
+		 */
+		for (i = 0; i < nr_found; i++) {
+			struct xfs_inode *ip = batch[i];
+
+			if (done || xfs_inode_ag_walk_grab(ip))
+				batch[i] = NULL;
+
+			/*
+			 * Update the index for the next lookup. Catch
+			 * overflows into the next AG range which can occur if
+			 * we have inodes in the last block of the AG and we
+			 * are currently pointing to the last inode.
+			 *
+			 * Because we may see inodes that are from the wrong AG
+			 * due to RCU freeing and reallocation, only update the
+			 * index if it lies in this AG. It was a race that lead
+			 * us to see this inode, so another lookup from the
+			 * same index will not find it again.
+			 */
+			if (XFS_INO_TO_AGNO(mp, ip->i_ino) != pag->pag_agno)
+				continue;
+			first_index = XFS_INO_TO_AGINO(mp, ip->i_ino + 1);
+			if (first_index < XFS_INO_TO_AGINO(mp, ip->i_ino))
+				done = 1;
+		}
+
+		/* unlock now we've grabbed the inodes. */
+		rcu_read_unlock();
+
+		for (i = 0; i < nr_found; i++) {
+			if (!batch[i])
+				continue;
+			error = execute(batch[i], pag, flags);
+			IRELE(batch[i]);
+			if (error == EAGAIN) {
+				skipped++;
+				continue;
+			}
+			if (error && last_error != EFSCORRUPTED)
+				last_error = error;
+		}
+
+		/* bail out if the filesystem is corrupted.  */
+		if (error == EFSCORRUPTED)
+			break;
+
+	} while (nr_found && !done);
+
+	if (skipped) {
+		delay(1);
+		goto restart;
+	}
+	return last_error;
+}
+
+int
+xfs_inode_ag_iterator(
+	struct xfs_mount	*mp,
+	int			(*execute)(struct xfs_inode *ip,
+					   struct xfs_perag *pag, int flags),
+	int			flags)
+{
+	struct xfs_perag	*pag;
+	int			error = 0;
+	int			last_error = 0;
+	xfs_agnumber_t		ag;
+
+	ag = 0;
+	while ((pag = xfs_perag_get(mp, ag))) {
+		ag = pag->pag_agno + 1;
+		error = xfs_inode_ag_walk(mp, pag, execute, flags);
+		xfs_perag_put(pag);
+		if (error) {
+			last_error = error;
+			if (error == EFSCORRUPTED)
+				break;
+		}
+	}
+	return XFS_ERROR(last_error);
+}
+
+STATIC int
+xfs_sync_inode_data(
+	struct xfs_inode	*ip,
+	struct xfs_perag	*pag,
+	int			flags)
+{
+	struct inode		*inode = VFS_I(ip);
+	struct address_space *mapping = inode->i_mapping;
+	int			error = 0;
+
+	if (!mapping_tagged(mapping, PAGECACHE_TAG_DIRTY))
+		goto out_wait;
+
+	if (!xfs_ilock_nowait(ip, XFS_IOLOCK_SHARED)) {
+		if (flags & SYNC_TRYLOCK)
+			goto out_wait;
+		xfs_ilock(ip, XFS_IOLOCK_SHARED);
+	}
+
+	error = xfs_flush_pages(ip, 0, -1, (flags & SYNC_WAIT) ?
+				0 : XBF_ASYNC, FI_NONE);
+	xfs_iunlock(ip, XFS_IOLOCK_SHARED);
+
+ out_wait:
+	if (flags & SYNC_WAIT)
+		xfs_ioend_wait(ip);
+	return error;
+}
+
+STATIC int
+xfs_sync_inode_attr(
+	struct xfs_inode	*ip,
+	struct xfs_perag	*pag,
+	int			flags)
+{
+	int			error = 0;
+
+	xfs_ilock(ip, XFS_ILOCK_SHARED);
+	if (xfs_inode_clean(ip))
+		goto out_unlock;
+	if (!xfs_iflock_nowait(ip)) {
+		if (!(flags & SYNC_WAIT))
+			goto out_unlock;
+		xfs_iflock(ip);
+	}
+
+	if (xfs_inode_clean(ip)) {
+		xfs_ifunlock(ip);
+		goto out_unlock;
+	}
+
+	error = xfs_iflush(ip, flags);
+
+	/*
+	 * We don't want to try again on non-blocking flushes that can't run
+	 * again immediately. If an inode really must be written, then that's
+	 * what the SYNC_WAIT flag is for.
+	 */
+	if (error == EAGAIN) {
+		ASSERT(!(flags & SYNC_WAIT));
+		error = 0;
+	}
+
+ out_unlock:
+	xfs_iunlock(ip, XFS_ILOCK_SHARED);
+	return error;
+}
+
+/*
+ * Write out pagecache data for the whole filesystem.
+ */
+STATIC int
+xfs_sync_data(
+	struct xfs_mount	*mp,
+	int			flags)
+{
+	int			error;
+
+	ASSERT((flags & ~(SYNC_TRYLOCK|SYNC_WAIT)) == 0);
+
+	error = xfs_inode_ag_iterator(mp, xfs_sync_inode_data, flags);
+	if (error)
+		return XFS_ERROR(error);
+
+	xfs_log_force(mp, (flags & SYNC_WAIT) ? XFS_LOG_SYNC : 0);
+	return 0;
+}
+
+/*
+ * Write out inode metadata (attributes) for the whole filesystem.
+ */
+STATIC int
+xfs_sync_attr(
+	struct xfs_mount	*mp,
+	int			flags)
+{
+	ASSERT((flags & ~SYNC_WAIT) == 0);
+
+	return xfs_inode_ag_iterator(mp, xfs_sync_inode_attr, flags);
+}
+
+STATIC int
+xfs_sync_fsdata(
+	struct xfs_mount	*mp)
+{
+	struct xfs_buf		*bp;
+
+	/*
+	 * If the buffer is pinned then push on the log so we won't get stuck
+	 * waiting in the write for someone, maybe ourselves, to flush the log.
+	 *
+	 * Even though we just pushed the log above, we did not have the
+	 * superblock buffer locked at that point so it can become pinned in
+	 * between there and here.
+	 */
+	bp = xfs_getsb(mp, 0);
+	if (XFS_BUF_ISPINNED(bp))
+		xfs_log_force(mp, 0);
+
+	return xfs_bwrite(mp, bp);
+}
+
+int
+xfs_log_dirty_inode(
+	struct xfs_inode	*ip,
+	struct xfs_perag	*pag,
+	int			flags)
+{
+	struct xfs_mount	*mp = ip->i_mount;
+	struct xfs_trans	*tp;
+	int			error;
+
+	if (!ip->i_update_core)
+		return 0;
+
+	tp = xfs_trans_alloc(mp, XFS_TRANS_FSYNC_TS);
+	error = xfs_trans_reserve(tp, 0, XFS_FSYNC_TS_LOG_RES(mp), 0, 0, 0);
+	if (error) {
+		xfs_trans_cancel(tp, 0);
+		return error;
+	}
+
+	xfs_ilock(ip, XFS_ILOCK_EXCL);
+	xfs_trans_ijoin_ref(tp, ip, XFS_ILOCK_EXCL);
+	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
+	return xfs_trans_commit(tp, 0);
+}
+
+/*
+ * When remounting a filesystem read-only or freezing the filesystem, we have
+ * two phases to execute. This first phase is syncing the data before we
+ * quiesce the filesystem, and the second is flushing all the inodes out after
+ * we've waited for all the transactions created by the first phase to
+ * complete. The second phase ensures that the inodes are written to their
+ * location on disk rather than just existing in transactions in the log. This
+ * means after a quiesce there is no log replay required to write the inodes to
+ * disk (this is the main difference between a sync and a quiesce).
+ */
+/*
+ * First stage of freeze - no writers will make progress now we are here,
+ * so we flush delwri and delalloc buffers here, then wait for all I/O to
+ * complete.  Data is frozen at that point. Metadata is not frozen,
+ * transactions can still occur here so don't bother flushing the buftarg
+ * because it'll just get dirty again.
+ */
+int
+xfs_quiesce_data(
+	struct xfs_mount	*mp)
+{
+	int			error, error2 = 0;
+
+	/* push non-blocking */
+	xfs_sync_data(mp, 0);
+	xfs_qm_sync(mp, SYNC_TRYLOCK);
+
+	/* push and block till complete */
+	xfs_sync_data(mp, SYNC_WAIT);
+
+	/*
+	 * Log all pending size and timestamp updates.  The vfs writeback
+	 * code is supposed to do this, but due to its overagressive
+	 * livelock detection it will skip inodes where appending writes
+	 * were written out in the first non-blocking sync phase if their
+	 * completion took long enough that it happened after taking the
+	 * timestamp for the cut-off in the blocking phase.
+	 */
+	xfs_inode_ag_iterator(mp, xfs_log_dirty_inode, 0);
+
+	xfs_qm_sync(mp, SYNC_WAIT);
+
+	/* write superblock and hoover up shutdown errors */
+	error = xfs_sync_fsdata(mp);
+
+	/* make sure all delwri buffers are written out */
+	xfs_flush_buftarg(mp->m_ddev_targp, 1);
+
+	/* mark the log as covered if needed */
+	if (xfs_log_need_covered(mp))
+		error2 = xfs_fs_log_dummy(mp);
+
+	/* flush data-only devices */
+	if (mp->m_rtdev_targp)
+		XFS_bflush(mp->m_rtdev_targp);
+
+	return error ? error : error2;
+}
+
+STATIC void
+xfs_quiesce_fs(
+	struct xfs_mount	*mp)
+{
+	int	count = 0, pincount;
+
+	xfs_reclaim_inodes(mp, 0);
+	xfs_flush_buftarg(mp->m_ddev_targp, 0);
+
+	/*
+	 * This loop must run at least twice.  The first instance of the loop
+	 * will flush most meta data but that will generate more meta data
+	 * (typically directory updates).  Which then must be flushed and
+	 * logged before we can write the unmount record. We also so sync
+	 * reclaim of inodes to catch any that the above delwri flush skipped.
+	 */
+	do {
+		xfs_reclaim_inodes(mp, SYNC_WAIT);
+		xfs_sync_attr(mp, SYNC_WAIT);
+		pincount = xfs_flush_buftarg(mp->m_ddev_targp, 1);
+		if (!pincount) {
+			delay(50);
+			count++;
+		}
+	} while (count < 2);
+}
+
+/*
+ * Second stage of a quiesce. The data is already synced, now we have to take
+ * care of the metadata. New transactions are already blocked, so we need to
+ * wait for any remaining transactions to drain out before proceeding.
+ */
+void
+xfs_quiesce_attr(
+	struct xfs_mount	*mp)
+{
+	int	error = 0;
+
+	/* wait for all modifications to complete */
+	while (atomic_read(&mp->m_active_trans) > 0)
+		delay(100);
+
+	/* flush inodes and push all remaining buffers out to disk */
+	xfs_quiesce_fs(mp);
+
+	/*
+	 * Just warn here till VFS can correctly support
+	 * read-only remount without racing.
+	 */
+	WARN_ON(atomic_read(&mp->m_active_trans) != 0);
+
+	/* Push the superblock and write an unmount record */
+	error = xfs_log_sbcount(mp, 1);
+	if (error)
+		xfs_warn(mp, "xfs_attr_quiesce: failed to log sb changes. "
+				"Frozen image may not be consistent.");
+	xfs_log_unmount_write(mp);
+	xfs_unmountfs_writesb(mp);
+}
+
+static void
+xfs_syncd_queue_sync(
+	struct xfs_mount        *mp)
+{
+	queue_delayed_work(xfs_syncd_wq, &mp->m_sync_work,
+				msecs_to_jiffies(xfs_syncd_centisecs * 10));
+}
+
+/*
+ * Every sync period we need to unpin all items, reclaim inodes and sync
+ * disk quotas.  We might need to cover the log to indicate that the
+ * filesystem is idle and not frozen.
+ */
+STATIC void
+xfs_sync_worker(
+	struct work_struct *work)
+{
+	struct xfs_mount *mp = container_of(to_delayed_work(work),
+					struct xfs_mount, m_sync_work);
+	int		error;
+
+	if (!(mp->m_flags & XFS_MOUNT_RDONLY)) {
+		/* dgc: errors ignored here */
+		if (mp->m_super->s_frozen == SB_UNFROZEN &&
+		    xfs_log_need_covered(mp))
+			error = xfs_fs_log_dummy(mp);
+		else
+			xfs_log_force(mp, 0);
+		error = xfs_qm_sync(mp, SYNC_TRYLOCK);
+
+		/* start pushing all the metadata that is currently dirty */
+		xfs_ail_push_all(mp->m_ail);
+	}
+
+	/* queue us up again */
+	xfs_syncd_queue_sync(mp);
+}
+
+/*
+ * Queue a new inode reclaim pass if there are reclaimable inodes and there
+ * isn't a reclaim pass already in progress. By default it runs every 5s based
+ * on the xfs syncd work default of 30s. Perhaps this should have it's own
+ * tunable, but that can be done if this method proves to be ineffective or too
+ * aggressive.
+ */
+static void
+xfs_syncd_queue_reclaim(
+	struct xfs_mount        *mp)
+{
+
+	/*
+	 * We can have inodes enter reclaim after we've shut down the syncd
+	 * workqueue during unmount, so don't allow reclaim work to be queued
+	 * during unmount.
+	 */
+	if (!(mp->m_super->s_flags & MS_ACTIVE))
+		return;
+
+	rcu_read_lock();
+	if (radix_tree_tagged(&mp->m_perag_tree, XFS_ICI_RECLAIM_TAG)) {
+		queue_delayed_work(xfs_syncd_wq, &mp->m_reclaim_work,
+			msecs_to_jiffies(xfs_syncd_centisecs / 6 * 10));
+	}
+	rcu_read_unlock();
+}
+
+/*
+ * This is a fast pass over the inode cache to try to get reclaim moving on as
+ * many inodes as possible in a short period of time. It kicks itself every few
+ * seconds, as well as being kicked by the inode cache shrinker when memory
+ * goes low. It scans as quickly as possible avoiding locked inodes or those
+ * already being flushed, and once done schedules a future pass.
+ */
+STATIC void
+xfs_reclaim_worker(
+	struct work_struct *work)
+{
+	struct xfs_mount *mp = container_of(to_delayed_work(work),
+					struct xfs_mount, m_reclaim_work);
+
+	xfs_reclaim_inodes(mp, SYNC_TRYLOCK);
+	xfs_syncd_queue_reclaim(mp);
+}
+
+/*
+ * Flush delayed allocate data, attempting to free up reserved space
+ * from existing allocations.  At this point a new allocation attempt
+ * has failed with ENOSPC and we are in the process of scratching our
+ * heads, looking about for more room.
+ *
+ * Queue a new data flush if there isn't one already in progress and
+ * wait for completion of the flush. This means that we only ever have one
+ * inode flush in progress no matter how many ENOSPC events are occurring and
+ * so will prevent the system from bogging down due to every concurrent
+ * ENOSPC event scanning all the active inodes in the system for writeback.
+ */
+void
+xfs_flush_inodes(
+	struct xfs_inode	*ip)
+{
+	struct xfs_mount	*mp = ip->i_mount;
+
+	queue_work(xfs_syncd_wq, &mp->m_flush_work);
+	flush_work_sync(&mp->m_flush_work);
+}
+
+STATIC void
+xfs_flush_worker(
+	struct work_struct *work)
+{
+	struct xfs_mount *mp = container_of(work,
+					struct xfs_mount, m_flush_work);
+
+	xfs_sync_data(mp, SYNC_TRYLOCK);
+	xfs_sync_data(mp, SYNC_TRYLOCK | SYNC_WAIT);
+}
+
+int
+xfs_syncd_init(
+	struct xfs_mount	*mp)
+{
+	INIT_WORK(&mp->m_flush_work, xfs_flush_worker);
+	INIT_DELAYED_WORK(&mp->m_sync_work, xfs_sync_worker);
+	INIT_DELAYED_WORK(&mp->m_reclaim_work, xfs_reclaim_worker);
+
+	xfs_syncd_queue_sync(mp);
+	xfs_syncd_queue_reclaim(mp);
+
+	return 0;
+}
+
+void
+xfs_syncd_stop(
+	struct xfs_mount	*mp)
+{
+	cancel_delayed_work_sync(&mp->m_sync_work);
+	cancel_delayed_work_sync(&mp->m_reclaim_work);
+	cancel_work_sync(&mp->m_flush_work);
+}
+
+void
+__xfs_inode_set_reclaim_tag(
+	struct xfs_perag	*pag,
+	struct xfs_inode	*ip)
+{
+	radix_tree_tag_set(&pag->pag_ici_root,
+			   XFS_INO_TO_AGINO(ip->i_mount, ip->i_ino),
+			   XFS_ICI_RECLAIM_TAG);
+
+	if (!pag->pag_ici_reclaimable) {
+		/* propagate the reclaim tag up into the perag radix tree */
+		spin_lock(&ip->i_mount->m_perag_lock);
+		radix_tree_tag_set(&ip->i_mount->m_perag_tree,
+				XFS_INO_TO_AGNO(ip->i_mount, ip->i_ino),
+				XFS_ICI_RECLAIM_TAG);
+		spin_unlock(&ip->i_mount->m_perag_lock);
+
+		/* schedule periodic background inode reclaim */
+		xfs_syncd_queue_reclaim(ip->i_mount);
+
+		trace_xfs_perag_set_reclaim(ip->i_mount, pag->pag_agno,
+							-1, _RET_IP_);
+	}
+	pag->pag_ici_reclaimable++;
+}
+
+/*
+ * We set the inode flag atomically with the radix tree tag.
+ * Once we get tag lookups on the radix tree, this inode flag
+ * can go away.
+ */
+void
+xfs_inode_set_reclaim_tag(
+	xfs_inode_t	*ip)
+{
+	struct xfs_mount *mp = ip->i_mount;
+	struct xfs_perag *pag;
+
+	pag = xfs_perag_get(mp, XFS_INO_TO_AGNO(mp, ip->i_ino));
+	spin_lock(&pag->pag_ici_lock);
+	spin_lock(&ip->i_flags_lock);
+	__xfs_inode_set_reclaim_tag(pag, ip);
+	__xfs_iflags_set(ip, XFS_IRECLAIMABLE);
+	spin_unlock(&ip->i_flags_lock);
+	spin_unlock(&pag->pag_ici_lock);
+	xfs_perag_put(pag);
+}
+
+STATIC void
+__xfs_inode_clear_reclaim(
+	xfs_perag_t	*pag,
+	xfs_inode_t	*ip)
+{
+	pag->pag_ici_reclaimable--;
+	if (!pag->pag_ici_reclaimable) {
+		/* clear the reclaim tag from the perag radix tree */
+		spin_lock(&ip->i_mount->m_perag_lock);
+		radix_tree_tag_clear(&ip->i_mount->m_perag_tree,
+				XFS_INO_TO_AGNO(ip->i_mount, ip->i_ino),
+				XFS_ICI_RECLAIM_TAG);
+		spin_unlock(&ip->i_mount->m_perag_lock);
+		trace_xfs_perag_clear_reclaim(ip->i_mount, pag->pag_agno,
+							-1, _RET_IP_);
+	}
+}
+
+void
+__xfs_inode_clear_reclaim_tag(
+	xfs_mount_t	*mp,
+	xfs_perag_t	*pag,
+	xfs_inode_t	*ip)
+{
+	radix_tree_tag_clear(&pag->pag_ici_root,
+			XFS_INO_TO_AGINO(mp, ip->i_ino), XFS_ICI_RECLAIM_TAG);
+	__xfs_inode_clear_reclaim(pag, ip);
+}
+
+/*
+ * Grab the inode for reclaim exclusively.
+ * Return 0 if we grabbed it, non-zero otherwise.
+ */
+STATIC int
+xfs_reclaim_inode_grab(
+	struct xfs_inode	*ip,
+	int			flags)
+{
+	ASSERT(rcu_read_lock_held());
+
+	/* quick check for stale RCU freed inode */
+	if (!ip->i_ino)
+		return 1;
+
+	/*
+	 * do some unlocked checks first to avoid unnecessary lock traffic.
+	 * The first is a flush lock check, the second is a already in reclaim
+	 * check. Only do these checks if we are not going to block on locks.
+	 */
+	if ((flags & SYNC_TRYLOCK) &&
+	    (!ip->i_flush.done || __xfs_iflags_test(ip, XFS_IRECLAIM))) {
+		return 1;
+	}
+
+	/*
+	 * The radix tree lock here protects a thread in xfs_iget from racing
+	 * with us starting reclaim on the inode.  Once we have the
+	 * XFS_IRECLAIM flag set it will not touch us.
+	 *
+	 * Due to RCU lookup, we may find inodes that have been freed and only
+	 * have XFS_IRECLAIM set.  Indeed, we may see reallocated inodes that
+	 * aren't candidates for reclaim at all, so we must check the
+	 * XFS_IRECLAIMABLE is set first before proceeding to reclaim.
+	 */
+	spin_lock(&ip->i_flags_lock);
+	if (!__xfs_iflags_test(ip, XFS_IRECLAIMABLE) ||
+	    __xfs_iflags_test(ip, XFS_IRECLAIM)) {
+		/* not a reclaim candidate. */
+		spin_unlock(&ip->i_flags_lock);
+		return 1;
+	}
+	__xfs_iflags_set(ip, XFS_IRECLAIM);
+	spin_unlock(&ip->i_flags_lock);
+	return 0;
+}
+
+/*
+ * Inodes in different states need to be treated differently, and the return
+ * value of xfs_iflush is not sufficient to get this right. The following table
+ * lists the inode states and the reclaim actions necessary for non-blocking
+ * reclaim:
+ *
+ *
+ *	inode state	     iflush ret		required action
+ *      ---------------      ----------         ---------------
+ *	bad			-		reclaim
+ *	shutdown		EIO		unpin and reclaim
+ *	clean, unpinned		0		reclaim
+ *	stale, unpinned		0		reclaim
+ *	clean, pinned(*)	0		requeue
+ *	stale, pinned		EAGAIN		requeue
+ *	dirty, delwri ok	0		requeue
+ *	dirty, delwri blocked	EAGAIN		requeue
+ *	dirty, sync flush	0		reclaim
+ *
+ * (*) dgc: I don't think the clean, pinned state is possible but it gets
+ * handled anyway given the order of checks implemented.
+ *
+ * As can be seen from the table, the return value of xfs_iflush() is not
+ * sufficient to correctly decide the reclaim action here. The checks in
+ * xfs_iflush() might look like duplicates, but they are not.
+ *
+ * Also, because we get the flush lock first, we know that any inode that has
+ * been flushed delwri has had the flush completed by the time we check that
+ * the inode is clean. The clean inode check needs to be done before flushing
+ * the inode delwri otherwise we would loop forever requeuing clean inodes as
+ * we cannot tell apart a successful delwri flush and a clean inode from the
+ * return value of xfs_iflush().
+ *
+ * Note that because the inode is flushed delayed write by background
+ * writeback, the flush lock may already be held here and waiting on it can
+ * result in very long latencies. Hence for sync reclaims, where we wait on the
+ * flush lock, the caller should push out delayed write inodes first before
+ * trying to reclaim them to minimise the amount of time spent waiting. For
+ * background relaim, we just requeue the inode for the next pass.
+ *
+ * Hence the order of actions after gaining the locks should be:
+ *	bad		=> reclaim
+ *	shutdown	=> unpin and reclaim
+ *	pinned, delwri	=> requeue
+ *	pinned, sync	=> unpin
+ *	stale		=> reclaim
+ *	clean		=> reclaim
+ *	dirty, delwri	=> flush and requeue
+ *	dirty, sync	=> flush, wait and reclaim
+ */
+STATIC int
+xfs_reclaim_inode(
+	struct xfs_inode	*ip,
+	struct xfs_perag	*pag,
+	int			sync_mode)
+{
+	int	error;
+
+restart:
+	error = 0;
+	xfs_ilock(ip, XFS_ILOCK_EXCL);
+	if (!xfs_iflock_nowait(ip)) {
+		if (!(sync_mode & SYNC_WAIT))
+			goto out;
+
+		/*
+		 * If we only have a single dirty inode in a cluster there is
+		 * a fair chance that the AIL push may have pushed it into
+		 * the buffer, but xfsbufd won't touch it until 30 seconds
+		 * from now, and thus we will lock up here.
+		 *
+		 * Promote the inode buffer to the front of the delwri list
+		 * and wake up xfsbufd now.
+		 */
+		xfs_promote_inode(ip);
+		xfs_iflock(ip);
+	}
+
+	if (is_bad_inode(VFS_I(ip)))
+		goto reclaim;
+	if (XFS_FORCED_SHUTDOWN(ip->i_mount)) {
+		xfs_iunpin_wait(ip);
+		goto reclaim;
+	}
+	if (xfs_ipincount(ip)) {
+		if (!(sync_mode & SYNC_WAIT)) {
+			xfs_ifunlock(ip);
+			goto out;
+		}
+		xfs_iunpin_wait(ip);
+	}
+	if (xfs_iflags_test(ip, XFS_ISTALE))
+		goto reclaim;
+	if (xfs_inode_clean(ip))
+		goto reclaim;
+
+	/*
+	 * Now we have an inode that needs flushing.
+	 *
+	 * We do a nonblocking flush here even if we are doing a SYNC_WAIT
+	 * reclaim as we can deadlock with inode cluster removal.
+	 * xfs_ifree_cluster() can lock the inode buffer before it locks the
+	 * ip->i_lock, and we are doing the exact opposite here. As a result,
+	 * doing a blocking xfs_itobp() to get the cluster buffer will result
+	 * in an ABBA deadlock with xfs_ifree_cluster().
+	 *
+	 * As xfs_ifree_cluser() must gather all inodes that are active in the
+	 * cache to mark them stale, if we hit this case we don't actually want
+	 * to do IO here - we want the inode marked stale so we can simply
+	 * reclaim it. Hence if we get an EAGAIN error on a SYNC_WAIT flush,
+	 * just unlock the inode, back off and try again. Hopefully the next
+	 * pass through will see the stale flag set on the inode.
+	 */
+	error = xfs_iflush(ip, SYNC_TRYLOCK | sync_mode);
+	if (sync_mode & SYNC_WAIT) {
+		if (error == EAGAIN) {
+			xfs_iunlock(ip, XFS_ILOCK_EXCL);
+			/* backoff longer than in xfs_ifree_cluster */
+			delay(2);
+			goto restart;
+		}
+		xfs_iflock(ip);
+		goto reclaim;
+	}
+
+	/*
+	 * When we have to flush an inode but don't have SYNC_WAIT set, we
+	 * flush the inode out using a delwri buffer and wait for the next
+	 * call into reclaim to find it in a clean state instead of waiting for
+	 * it now. We also don't return errors here - if the error is transient
+	 * then the next reclaim pass will flush the inode, and if the error
+	 * is permanent then the next sync reclaim will reclaim the inode and
+	 * pass on the error.
+	 */
+	if (error && error != EAGAIN && !XFS_FORCED_SHUTDOWN(ip->i_mount)) {
+		xfs_warn(ip->i_mount,
+			"inode 0x%llx background reclaim flush failed with %d",
+			(long long)ip->i_ino, error);
+	}
+out:
+	xfs_iflags_clear(ip, XFS_IRECLAIM);
+	xfs_iunlock(ip, XFS_ILOCK_EXCL);
+	/*
+	 * We could return EAGAIN here to make reclaim rescan the inode tree in
+	 * a short while. However, this just burns CPU time scanning the tree
+	 * waiting for IO to complete and xfssyncd never goes back to the idle
+	 * state. Instead, return 0 to let the next scheduled background reclaim
+	 * attempt to reclaim the inode again.
+	 */
+	return 0;
+
+reclaim:
+	xfs_ifunlock(ip);
+	xfs_iunlock(ip, XFS_ILOCK_EXCL);
+
+	XFS_STATS_INC(xs_ig_reclaims);
+	/*
+	 * Remove the inode from the per-AG radix tree.
+	 *
+	 * Because radix_tree_delete won't complain even if the item was never
+	 * added to the tree assert that it's been there before to catch
+	 * problems with the inode life time early on.
+	 */
+	spin_lock(&pag->pag_ici_lock);
+	if (!radix_tree_delete(&pag->pag_ici_root,
+				XFS_INO_TO_AGINO(ip->i_mount, ip->i_ino)))
+		ASSERT(0);
+	__xfs_inode_clear_reclaim(pag, ip);
+	spin_unlock(&pag->pag_ici_lock);
+
+	/*
+	 * Here we do an (almost) spurious inode lock in order to coordinate
+	 * with inode cache radix tree lookups.  This is because the lookup
+	 * can reference the inodes in the cache without taking references.
+	 *
+	 * We make that OK here by ensuring that we wait until the inode is
+	 * unlocked after the lookup before we go ahead and free it.  We get
+	 * both the ilock and the iolock because the code may need to drop the
+	 * ilock one but will still hold the iolock.
+	 */
+	xfs_ilock(ip, XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL);
+	xfs_qm_dqdetach(ip);
+	xfs_iunlock(ip, XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL);
+
+	xfs_inode_free(ip);
+	return error;
+
+}
+
+/*
+ * Walk the AGs and reclaim the inodes in them. Even if the filesystem is
+ * corrupted, we still want to try to reclaim all the inodes. If we don't,
+ * then a shut down during filesystem unmount reclaim walk leak all the
+ * unreclaimed inodes.
+ */
+int
+xfs_reclaim_inodes_ag(
+	struct xfs_mount	*mp,
+	int			flags,
+	int			*nr_to_scan)
+{
+	struct xfs_perag	*pag;
+	int			error = 0;
+	int			last_error = 0;
+	xfs_agnumber_t		ag;
+	int			trylock = flags & SYNC_TRYLOCK;
+	int			skipped;
+
+restart:
+	ag = 0;
+	skipped = 0;
+	while ((pag = xfs_perag_get_tag(mp, ag, XFS_ICI_RECLAIM_TAG))) {
+		unsigned long	first_index = 0;
+		int		done = 0;
+		int		nr_found = 0;
+
+		ag = pag->pag_agno + 1;
+
+		if (trylock) {
+			if (!mutex_trylock(&pag->pag_ici_reclaim_lock)) {
+				skipped++;
+				xfs_perag_put(pag);
+				continue;
+			}
+			first_index = pag->pag_ici_reclaim_cursor;
+		} else
+			mutex_lock(&pag->pag_ici_reclaim_lock);
+
+		do {
+			struct xfs_inode *batch[XFS_LOOKUP_BATCH];
+			int	i;
+
+			rcu_read_lock();
+			nr_found = radix_tree_gang_lookup_tag(
+					&pag->pag_ici_root,
+					(void **)batch, first_index,
+					XFS_LOOKUP_BATCH,
+					XFS_ICI_RECLAIM_TAG);
+			if (!nr_found) {
+				done = 1;
+				rcu_read_unlock();
+				break;
+			}
+
+			/*
+			 * Grab the inodes before we drop the lock. if we found
+			 * nothing, nr == 0 and the loop will be skipped.
+			 */
+			for (i = 0; i < nr_found; i++) {
+				struct xfs_inode *ip = batch[i];
+
+				if (done || xfs_reclaim_inode_grab(ip, flags))
+					batch[i] = NULL;
+
+				/*
+				 * Update the index for the next lookup. Catch
+				 * overflows into the next AG range which can
+				 * occur if we have inodes in the last block of
+				 * the AG and we are currently pointing to the
+				 * last inode.
+				 *
+				 * Because we may see inodes that are from the
+				 * wrong AG due to RCU freeing and
+				 * reallocation, only update the index if it
+				 * lies in this AG. It was a race that lead us
+				 * to see this inode, so another lookup from
+				 * the same index will not find it again.
+				 */
+				if (XFS_INO_TO_AGNO(mp, ip->i_ino) !=
+								pag->pag_agno)
+					continue;
+				first_index = XFS_INO_TO_AGINO(mp, ip->i_ino + 1);
+				if (first_index < XFS_INO_TO_AGINO(mp, ip->i_ino))
+					done = 1;
+			}
+
+			/* unlock now we've grabbed the inodes. */
+			rcu_read_unlock();
+
+			for (i = 0; i < nr_found; i++) {
+				if (!batch[i])
+					continue;
+				error = xfs_reclaim_inode(batch[i], pag, flags);
+				if (error && last_error != EFSCORRUPTED)
+					last_error = error;
+			}
+
+			*nr_to_scan -= XFS_LOOKUP_BATCH;
+
+		} while (nr_found && !done && *nr_to_scan > 0);
+
+		if (trylock && !done)
+			pag->pag_ici_reclaim_cursor = first_index;
+		else
+			pag->pag_ici_reclaim_cursor = 0;
+		mutex_unlock(&pag->pag_ici_reclaim_lock);
+		xfs_perag_put(pag);
+	}
+
+	/*
+	 * if we skipped any AG, and we still have scan count remaining, do
+	 * another pass this time using blocking reclaim semantics (i.e
+	 * waiting on the reclaim locks and ignoring the reclaim cursors). This
+	 * ensure that when we get more reclaimers than AGs we block rather
+	 * than spin trying to execute reclaim.
+	 */
+	if (trylock && skipped && *nr_to_scan > 0) {
+		trylock = 0;
+		goto restart;
+	}
+	return XFS_ERROR(last_error);
+}
+
+int
+xfs_reclaim_inodes(
+	xfs_mount_t	*mp,
+	int		mode)
+{
+	int		nr_to_scan = INT_MAX;
+
+	return xfs_reclaim_inodes_ag(mp, mode, &nr_to_scan);
+}
+
+/*
+ * Inode cache shrinker.
+ *
+ * When called we make sure that there is a background (fast) inode reclaim in
+ * progress, while we will throttle the speed of reclaim via doiing synchronous
+ * reclaim of inodes. That means if we come across dirty inodes, we wait for
+ * them to be cleaned, which we hope will not be very long due to the
+ * background walker having already kicked the IO off on those dirty inodes.
+ */
+static int
+xfs_reclaim_inode_shrink(
+	struct shrinker	*shrink,
+	struct shrink_control *sc)
+{
+	struct xfs_mount *mp;
+	struct xfs_perag *pag;
+	xfs_agnumber_t	ag;
+	int		reclaimable;
+	int nr_to_scan = sc->nr_to_scan;
+	gfp_t gfp_mask = sc->gfp_mask;
+
+	mp = container_of(shrink, struct xfs_mount, m_inode_shrink);
+	if (nr_to_scan) {
+		/* kick background reclaimer and push the AIL */
+		xfs_syncd_queue_reclaim(mp);
+		xfs_ail_push_all(mp->m_ail);
+
+		if (!(gfp_mask & __GFP_FS))
+			return -1;
+
+		xfs_reclaim_inodes_ag(mp, SYNC_TRYLOCK | SYNC_WAIT,
+					&nr_to_scan);
+		/* terminate if we don't exhaust the scan */
+		if (nr_to_scan > 0)
+			return -1;
+       }
+
+	reclaimable = 0;
+	ag = 0;
+	while ((pag = xfs_perag_get_tag(mp, ag, XFS_ICI_RECLAIM_TAG))) {
+		ag = pag->pag_agno + 1;
+		reclaimable += pag->pag_ici_reclaimable;
+		xfs_perag_put(pag);
+	}
+	return reclaimable;
+}
+
+void
+xfs_inode_shrinker_register(
+	struct xfs_mount	*mp)
+{
+	mp->m_inode_shrink.shrink = xfs_reclaim_inode_shrink;
+	mp->m_inode_shrink.seeks = DEFAULT_SEEKS;
+	register_shrinker(&mp->m_inode_shrink);
+}
+
+void
+xfs_inode_shrinker_unregister(
+	struct xfs_mount	*mp)
+{
+	unregister_shrinker(&mp->m_inode_shrink);
+}
diff --git a/fs/xfs/linux-2.6/xfs_sync.h b/fs/xfs/linux-2.6/xfs_sync.h
new file mode 100644
index 00000000..ef5b2ce4
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_sync.h
@@ -0,0 +1,62 @@
+/*
+ * Copyright (c) 2000-2006 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef XFS_SYNC_H
+#define XFS_SYNC_H 1
+
+struct xfs_mount;
+struct xfs_perag;
+
+typedef struct xfs_sync_work {
+	struct list_head	w_list;
+	struct xfs_mount	*w_mount;
+	void			*w_data;	/* syncer routine argument */
+	void			(*w_syncer)(struct xfs_mount *, void *);
+	struct completion	*w_completion;
+} xfs_sync_work_t;
+
+#define SYNC_WAIT		0x0001	/* wait for i/o to complete */
+#define SYNC_TRYLOCK		0x0002  /* only try to lock inodes */
+
+extern struct workqueue_struct	*xfs_syncd_wq;	/* sync workqueue */
+
+int xfs_syncd_init(struct xfs_mount *mp);
+void xfs_syncd_stop(struct xfs_mount *mp);
+
+int xfs_quiesce_data(struct xfs_mount *mp);
+void xfs_quiesce_attr(struct xfs_mount *mp);
+
+void xfs_flush_inodes(struct xfs_inode *ip);
+
+int xfs_log_dirty_inode(struct xfs_inode *ip, struct xfs_perag *pag, int flags);
+
+int xfs_reclaim_inodes(struct xfs_mount *mp, int mode);
+
+void xfs_inode_set_reclaim_tag(struct xfs_inode *ip);
+void __xfs_inode_set_reclaim_tag(struct xfs_perag *pag, struct xfs_inode *ip);
+void __xfs_inode_clear_reclaim_tag(struct xfs_mount *mp, struct xfs_perag *pag,
+				struct xfs_inode *ip);
+
+int xfs_sync_inode_grab(struct xfs_inode *ip);
+int xfs_inode_ag_iterator(struct xfs_mount *mp,
+	int (*execute)(struct xfs_inode *ip, struct xfs_perag *pag, int flags),
+	int flags);
+
+void xfs_inode_shrinker_register(struct xfs_mount *mp);
+void xfs_inode_shrinker_unregister(struct xfs_mount *mp);
+
+#endif
diff --git a/fs/xfs/linux-2.6/xfs_sysctl.c b/fs/xfs/linux-2.6/xfs_sysctl.c
new file mode 100644
index 00000000..ee2d2ada
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_sysctl.c
@@ -0,0 +1,252 @@
+/*
+ * Copyright (c) 2001-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include <linux/sysctl.h>
+#include <linux/proc_fs.h>
+#include "xfs_error.h"
+
+static struct ctl_table_header *xfs_table_header;
+
+#ifdef CONFIG_PROC_FS
+STATIC int
+xfs_stats_clear_proc_handler(
+	ctl_table	*ctl,
+	int		write,
+	void		__user *buffer,
+	size_t		*lenp,
+	loff_t		*ppos)
+{
+	int		c, ret, *valp = ctl->data;
+	__uint32_t	vn_active;
+
+	ret = proc_dointvec_minmax(ctl, write, buffer, lenp, ppos);
+
+	if (!ret && write && *valp) {
+		xfs_notice(NULL, "Clearing xfsstats");
+		for_each_possible_cpu(c) {
+			preempt_disable();
+			/* save vn_active, it's a universal truth! */
+			vn_active = per_cpu(xfsstats, c).vn_active;
+			memset(&per_cpu(xfsstats, c), 0,
+			       sizeof(struct xfsstats));
+			per_cpu(xfsstats, c).vn_active = vn_active;
+			preempt_enable();
+		}
+		xfs_stats_clear = 0;
+	}
+
+	return ret;
+}
+
+STATIC int
+xfs_panic_mask_proc_handler(
+	ctl_table	*ctl,
+	int		write,
+	void		__user *buffer,
+	size_t		*lenp,
+	loff_t		*ppos)
+{
+	int		ret, *valp = ctl->data;
+
+	ret = proc_dointvec_minmax(ctl, write, buffer, lenp, ppos);
+	if (!ret && write) {
+		xfs_panic_mask = *valp;
+#ifdef DEBUG
+		xfs_panic_mask |= (XFS_PTAG_SHUTDOWN_CORRUPT | XFS_PTAG_LOGRES);
+#endif
+	}
+	return ret;
+}
+#endif /* CONFIG_PROC_FS */
+
+static ctl_table xfs_table[] = {
+	{
+		.procname	= "irix_sgid_inherit",
+		.data		= &xfs_params.sgid_inherit.val,
+		.maxlen		= sizeof(int),
+		.mode		= 0644,
+		.proc_handler	= proc_dointvec_minmax,
+		.extra1		= &xfs_params.sgid_inherit.min,
+		.extra2		= &xfs_params.sgid_inherit.max
+	},
+	{
+		.procname	= "irix_symlink_mode",
+		.data		= &xfs_params.symlink_mode.val,
+		.maxlen		= sizeof(int),
+		.mode		= 0644,
+		.proc_handler	= proc_dointvec_minmax,
+		.extra1		= &xfs_params.symlink_mode.min,
+		.extra2		= &xfs_params.symlink_mode.max
+	},
+	{
+		.procname	= "panic_mask",
+		.data		= &xfs_params.panic_mask.val,
+		.maxlen		= sizeof(int),
+		.mode		= 0644,
+		.proc_handler	= xfs_panic_mask_proc_handler,
+		.extra1		= &xfs_params.panic_mask.min,
+		.extra2		= &xfs_params.panic_mask.max
+	},
+
+	{
+		.procname	= "error_level",
+		.data		= &xfs_params.error_level.val,
+		.maxlen		= sizeof(int),
+		.mode		= 0644,
+		.proc_handler	= proc_dointvec_minmax,
+		.extra1		= &xfs_params.error_level.min,
+		.extra2		= &xfs_params.error_level.max
+	},
+	{
+		.procname	= "xfssyncd_centisecs",
+		.data		= &xfs_params.syncd_timer.val,
+		.maxlen		= sizeof(int),
+		.mode		= 0644,
+		.proc_handler	= proc_dointvec_minmax,
+		.extra1		= &xfs_params.syncd_timer.min,
+		.extra2		= &xfs_params.syncd_timer.max
+	},
+	{
+		.procname	= "inherit_sync",
+		.data		= &xfs_params.inherit_sync.val,
+		.maxlen		= sizeof(int),
+		.mode		= 0644,
+		.proc_handler	= proc_dointvec_minmax,
+		.extra1		= &xfs_params.inherit_sync.min,
+		.extra2		= &xfs_params.inherit_sync.max
+	},
+	{
+		.procname	= "inherit_nodump",
+		.data		= &xfs_params.inherit_nodump.val,
+		.maxlen		= sizeof(int),
+		.mode		= 0644,
+		.proc_handler	= proc_dointvec_minmax,
+		.extra1		= &xfs_params.inherit_nodump.min,
+		.extra2		= &xfs_params.inherit_nodump.max
+	},
+	{
+		.procname	= "inherit_noatime",
+		.data		= &xfs_params.inherit_noatim.val,
+		.maxlen		= sizeof(int),
+		.mode		= 0644,
+		.proc_handler	= proc_dointvec_minmax,
+		.extra1		= &xfs_params.inherit_noatim.min,
+		.extra2		= &xfs_params.inherit_noatim.max
+	},
+	{
+		.procname	= "xfsbufd_centisecs",
+		.data		= &xfs_params.xfs_buf_timer.val,
+		.maxlen		= sizeof(int),
+		.mode		= 0644,
+		.proc_handler	= proc_dointvec_minmax,
+		.extra1		= &xfs_params.xfs_buf_timer.min,
+		.extra2		= &xfs_params.xfs_buf_timer.max
+	},
+	{
+		.procname	= "age_buffer_centisecs",
+		.data		= &xfs_params.xfs_buf_age.val,
+		.maxlen		= sizeof(int),
+		.mode		= 0644,
+		.proc_handler	= proc_dointvec_minmax,
+		.extra1		= &xfs_params.xfs_buf_age.min,
+		.extra2		= &xfs_params.xfs_buf_age.max
+	},
+	{
+		.procname	= "inherit_nosymlinks",
+		.data		= &xfs_params.inherit_nosym.val,
+		.maxlen		= sizeof(int),
+		.mode		= 0644,
+		.proc_handler	= proc_dointvec_minmax,
+		.extra1		= &xfs_params.inherit_nosym.min,
+		.extra2		= &xfs_params.inherit_nosym.max
+	},
+	{
+		.procname	= "rotorstep",
+		.data		= &xfs_params.rotorstep.val,
+		.maxlen		= sizeof(int),
+		.mode		= 0644,
+		.proc_handler	= proc_dointvec_minmax,
+		.extra1		= &xfs_params.rotorstep.min,
+		.extra2		= &xfs_params.rotorstep.max
+	},
+	{
+		.procname	= "inherit_nodefrag",
+		.data		= &xfs_params.inherit_nodfrg.val,
+		.maxlen		= sizeof(int),
+		.mode		= 0644,
+		.proc_handler	= proc_dointvec_minmax,
+		.extra1		= &xfs_params.inherit_nodfrg.min,
+		.extra2		= &xfs_params.inherit_nodfrg.max
+	},
+	{
+		.procname	= "filestream_centisecs",
+		.data		= &xfs_params.fstrm_timer.val,
+		.maxlen		= sizeof(int),
+		.mode		= 0644,
+		.proc_handler	= proc_dointvec_minmax,
+		.extra1		= &xfs_params.fstrm_timer.min,
+		.extra2		= &xfs_params.fstrm_timer.max,
+	},
+	/* please keep this the last entry */
+#ifdef CONFIG_PROC_FS
+	{
+		.procname	= "stats_clear",
+		.data		= &xfs_params.stats_clear.val,
+		.maxlen		= sizeof(int),
+		.mode		= 0644,
+		.proc_handler	= xfs_stats_clear_proc_handler,
+		.extra1		= &xfs_params.stats_clear.min,
+		.extra2		= &xfs_params.stats_clear.max
+	},
+#endif /* CONFIG_PROC_FS */
+
+	{}
+};
+
+static ctl_table xfs_dir_table[] = {
+	{
+		.procname	= "xfs",
+		.mode		= 0555,
+		.child		= xfs_table
+	},
+	{}
+};
+
+static ctl_table xfs_root_table[] = {
+	{
+		.procname	= "fs",
+		.mode		= 0555,
+		.child		= xfs_dir_table
+	},
+	{}
+};
+
+int
+xfs_sysctl_register(void)
+{
+	xfs_table_header = register_sysctl_table(xfs_root_table);
+	if (!xfs_table_header)
+		return -ENOMEM;
+	return 0;
+}
+
+void
+xfs_sysctl_unregister(void)
+{
+	unregister_sysctl_table(xfs_table_header);
+}
diff --git a/fs/xfs/linux-2.6/xfs_sysctl.h b/fs/xfs/linux-2.6/xfs_sysctl.h
new file mode 100644
index 00000000..b9937d45
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_sysctl.h
@@ -0,0 +1,102 @@
+/*
+ * Copyright (c) 2001-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_SYSCTL_H__
+#define __XFS_SYSCTL_H__
+
+#include <linux/sysctl.h>
+
+/*
+ * Tunable xfs parameters
+ */
+
+typedef struct xfs_sysctl_val {
+	int min;
+	int val;
+	int max;
+} xfs_sysctl_val_t;
+
+typedef struct xfs_param {
+	xfs_sysctl_val_t sgid_inherit;	/* Inherit S_ISGID if process' GID is
+					 * not a member of parent dir GID. */
+	xfs_sysctl_val_t symlink_mode;	/* Link creat mode affected by umask */
+	xfs_sysctl_val_t panic_mask;	/* bitmask to cause panic on errors. */
+	xfs_sysctl_val_t error_level;	/* Degree of reporting for problems  */
+	xfs_sysctl_val_t syncd_timer;	/* Interval between xfssyncd wakeups */
+	xfs_sysctl_val_t stats_clear;	/* Reset all XFS statistics to zero. */
+	xfs_sysctl_val_t inherit_sync;	/* Inherit the "sync" inode flag. */
+	xfs_sysctl_val_t inherit_nodump;/* Inherit the "nodump" inode flag. */
+	xfs_sysctl_val_t inherit_noatim;/* Inherit the "noatime" inode flag. */
+	xfs_sysctl_val_t xfs_buf_timer;	/* Interval between xfsbufd wakeups. */
+	xfs_sysctl_val_t xfs_buf_age;	/* Metadata buffer age before flush. */
+	xfs_sysctl_val_t inherit_nosym;	/* Inherit the "nosymlinks" flag. */
+	xfs_sysctl_val_t rotorstep;	/* inode32 AG rotoring control knob */
+	xfs_sysctl_val_t inherit_nodfrg;/* Inherit the "nodefrag" inode flag. */
+	xfs_sysctl_val_t fstrm_timer;	/* Filestream dir-AG assoc'n timeout. */
+} xfs_param_t;
+
+/*
+ * xfs_error_level:
+ *
+ * How much error reporting will be done when internal problems are
+ * encountered.  These problems normally return an EFSCORRUPTED to their
+ * caller, with no other information reported.
+ *
+ * 0	No error reports
+ * 1	Report EFSCORRUPTED errors that will cause a filesystem shutdown
+ * 5	Report all EFSCORRUPTED errors (all of the above errors, plus any
+ *	additional errors that are known to not cause shutdowns)
+ *
+ * xfs_panic_mask bit 0x8 turns the error reports into panics
+ */
+
+enum {
+	/* XFS_REFCACHE_SIZE = 1 */
+	/* XFS_REFCACHE_PURGE = 2 */
+	/* XFS_RESTRICT_CHOWN = 3 */
+	XFS_SGID_INHERIT = 4,
+	XFS_SYMLINK_MODE = 5,
+	XFS_PANIC_MASK = 6,
+	XFS_ERRLEVEL = 7,
+	XFS_SYNCD_TIMER = 8,
+	/* XFS_PROBE_DMAPI = 9 */
+	/* XFS_PROBE_IOOPS = 10 */
+	/* XFS_PROBE_QUOTA = 11 */
+	XFS_STATS_CLEAR = 12,
+	XFS_INHERIT_SYNC = 13,
+	XFS_INHERIT_NODUMP = 14,
+	XFS_INHERIT_NOATIME = 15,
+	XFS_BUF_TIMER = 16,
+	XFS_BUF_AGE = 17,
+	/* XFS_IO_BYPASS = 18 */
+	XFS_INHERIT_NOSYM = 19,
+	XFS_ROTORSTEP = 20,
+	XFS_INHERIT_NODFRG = 21,
+	XFS_FILESTREAM_TIMER = 22,
+};
+
+extern xfs_param_t	xfs_params;
+
+#ifdef CONFIG_SYSCTL
+extern int xfs_sysctl_register(void);
+extern void xfs_sysctl_unregister(void);
+#else
+# define xfs_sysctl_register()		(0)
+# define xfs_sysctl_unregister()	do { } while (0)
+#endif /* CONFIG_SYSCTL */
+
+#endif /* __XFS_SYSCTL_H__ */
diff --git a/fs/xfs/linux-2.6/xfs_trace.c b/fs/xfs/linux-2.6/xfs_trace.c
new file mode 100644
index 00000000..88d25d4a
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_trace.c
@@ -0,0 +1,56 @@
+/*
+ * Copyright (c) 2009, Christoph Hellwig
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_da_btree.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_alloc_btree.h"
+#include "xfs_ialloc_btree.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_btree.h"
+#include "xfs_mount.h"
+#include "xfs_ialloc.h"
+#include "xfs_itable.h"
+#include "xfs_alloc.h"
+#include "xfs_bmap.h"
+#include "xfs_attr.h"
+#include "xfs_attr_leaf.h"
+#include "xfs_log_priv.h"
+#include "xfs_buf_item.h"
+#include "xfs_quota.h"
+#include "xfs_iomap.h"
+#include "xfs_aops.h"
+#include "quota/xfs_dquot_item.h"
+#include "quota/xfs_dquot.h"
+#include "xfs_log_recover.h"
+#include "xfs_inode_item.h"
+
+/*
+ * We include this last to have the helpers above available for the trace
+ * event implementations.
+ */
+#define CREATE_TRACE_POINTS
+#include "xfs_trace.h"
diff --git a/fs/xfs/linux-2.6/xfs_trace.h b/fs/xfs/linux-2.6/xfs_trace.h
new file mode 100644
index 00000000..d48b7a57
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_trace.h
@@ -0,0 +1,1776 @@
+/*
+ * Copyright (c) 2009, Christoph Hellwig
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM xfs
+
+#if !defined(_TRACE_XFS_H) || defined(TRACE_HEADER_MULTI_READ)
+#define _TRACE_XFS_H
+
+#include <linux/tracepoint.h>
+
+struct xfs_agf;
+struct xfs_alloc_arg;
+struct xfs_attr_list_context;
+struct xfs_buf_log_item;
+struct xfs_da_args;
+struct xfs_da_node_entry;
+struct xfs_dquot;
+struct xlog_ticket;
+struct log;
+struct xlog_recover;
+struct xlog_recover_item;
+struct xfs_buf_log_format;
+struct xfs_inode_log_format;
+
+DECLARE_EVENT_CLASS(xfs_attr_list_class,
+	TP_PROTO(struct xfs_attr_list_context *ctx),
+	TP_ARGS(ctx),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_ino_t, ino)
+		__field(u32, hashval)
+		__field(u32, blkno)
+		__field(u32, offset)
+		__field(void *, alist)
+		__field(int, bufsize)
+		__field(int, count)
+		__field(int, firstu)
+		__field(int, dupcnt)
+		__field(int, flags)
+	),
+	TP_fast_assign(
+		__entry->dev = VFS_I(ctx->dp)->i_sb->s_dev;
+		__entry->ino = ctx->dp->i_ino;
+		__entry->hashval = ctx->cursor->hashval;
+		__entry->blkno = ctx->cursor->blkno;
+		__entry->offset = ctx->cursor->offset;
+		__entry->alist = ctx->alist;
+		__entry->bufsize = ctx->bufsize;
+		__entry->count = ctx->count;
+		__entry->firstu = ctx->firstu;
+		__entry->flags = ctx->flags;
+	),
+	TP_printk("dev %d:%d ino 0x%llx cursor h/b/o 0x%x/0x%x/%u dupcnt %u "
+		  "alist 0x%p size %u count %u firstu %u flags %d %s",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		   __entry->ino,
+		   __entry->hashval,
+		   __entry->blkno,
+		   __entry->offset,
+		   __entry->dupcnt,
+		   __entry->alist,
+		   __entry->bufsize,
+		   __entry->count,
+		   __entry->firstu,
+		   __entry->flags,
+		   __print_flags(__entry->flags, "|", XFS_ATTR_FLAGS)
+	)
+)
+
+#define DEFINE_ATTR_LIST_EVENT(name) \
+DEFINE_EVENT(xfs_attr_list_class, name, \
+	TP_PROTO(struct xfs_attr_list_context *ctx), \
+	TP_ARGS(ctx))
+DEFINE_ATTR_LIST_EVENT(xfs_attr_list_sf);
+DEFINE_ATTR_LIST_EVENT(xfs_attr_list_sf_all);
+DEFINE_ATTR_LIST_EVENT(xfs_attr_list_leaf);
+DEFINE_ATTR_LIST_EVENT(xfs_attr_list_leaf_end);
+DEFINE_ATTR_LIST_EVENT(xfs_attr_list_full);
+DEFINE_ATTR_LIST_EVENT(xfs_attr_list_add);
+DEFINE_ATTR_LIST_EVENT(xfs_attr_list_wrong_blk);
+DEFINE_ATTR_LIST_EVENT(xfs_attr_list_notfound);
+
+DECLARE_EVENT_CLASS(xfs_perag_class,
+	TP_PROTO(struct xfs_mount *mp, xfs_agnumber_t agno, int refcount,
+		 unsigned long caller_ip),
+	TP_ARGS(mp, agno, refcount, caller_ip),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_agnumber_t, agno)
+		__field(int, refcount)
+		__field(unsigned long, caller_ip)
+	),
+	TP_fast_assign(
+		__entry->dev = mp->m_super->s_dev;
+		__entry->agno = agno;
+		__entry->refcount = refcount;
+		__entry->caller_ip = caller_ip;
+	),
+	TP_printk("dev %d:%d agno %u refcount %d caller %pf",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->agno,
+		  __entry->refcount,
+		  (char *)__entry->caller_ip)
+);
+
+#define DEFINE_PERAG_REF_EVENT(name)	\
+DEFINE_EVENT(xfs_perag_class, name,	\
+	TP_PROTO(struct xfs_mount *mp, xfs_agnumber_t agno, int refcount,	\
+		 unsigned long caller_ip),					\
+	TP_ARGS(mp, agno, refcount, caller_ip))
+DEFINE_PERAG_REF_EVENT(xfs_perag_get);
+DEFINE_PERAG_REF_EVENT(xfs_perag_get_tag);
+DEFINE_PERAG_REF_EVENT(xfs_perag_put);
+DEFINE_PERAG_REF_EVENT(xfs_perag_set_reclaim);
+DEFINE_PERAG_REF_EVENT(xfs_perag_clear_reclaim);
+
+TRACE_EVENT(xfs_attr_list_node_descend,
+	TP_PROTO(struct xfs_attr_list_context *ctx,
+		 struct xfs_da_node_entry *btree),
+	TP_ARGS(ctx, btree),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_ino_t, ino)
+		__field(u32, hashval)
+		__field(u32, blkno)
+		__field(u32, offset)
+		__field(void *, alist)
+		__field(int, bufsize)
+		__field(int, count)
+		__field(int, firstu)
+		__field(int, dupcnt)
+		__field(int, flags)
+		__field(u32, bt_hashval)
+		__field(u32, bt_before)
+	),
+	TP_fast_assign(
+		__entry->dev = VFS_I(ctx->dp)->i_sb->s_dev;
+		__entry->ino = ctx->dp->i_ino;
+		__entry->hashval = ctx->cursor->hashval;
+		__entry->blkno = ctx->cursor->blkno;
+		__entry->offset = ctx->cursor->offset;
+		__entry->alist = ctx->alist;
+		__entry->bufsize = ctx->bufsize;
+		__entry->count = ctx->count;
+		__entry->firstu = ctx->firstu;
+		__entry->flags = ctx->flags;
+		__entry->bt_hashval = be32_to_cpu(btree->hashval);
+		__entry->bt_before = be32_to_cpu(btree->before);
+	),
+	TP_printk("dev %d:%d ino 0x%llx cursor h/b/o 0x%x/0x%x/%u dupcnt %u "
+		  "alist 0x%p size %u count %u firstu %u flags %d %s "
+		  "node hashval %u, node before %u",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		   __entry->ino,
+		   __entry->hashval,
+		   __entry->blkno,
+		   __entry->offset,
+		   __entry->dupcnt,
+		   __entry->alist,
+		   __entry->bufsize,
+		   __entry->count,
+		   __entry->firstu,
+		   __entry->flags,
+		   __print_flags(__entry->flags, "|", XFS_ATTR_FLAGS),
+		   __entry->bt_hashval,
+		   __entry->bt_before)
+);
+
+TRACE_EVENT(xfs_iext_insert,
+	TP_PROTO(struct xfs_inode *ip, xfs_extnum_t idx,
+		 struct xfs_bmbt_irec *r, int state, unsigned long caller_ip),
+	TP_ARGS(ip, idx, r, state, caller_ip),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_ino_t, ino)
+		__field(xfs_extnum_t, idx)
+		__field(xfs_fileoff_t, startoff)
+		__field(xfs_fsblock_t, startblock)
+		__field(xfs_filblks_t, blockcount)
+		__field(xfs_exntst_t, state)
+		__field(int, bmap_state)
+		__field(unsigned long, caller_ip)
+	),
+	TP_fast_assign(
+		__entry->dev = VFS_I(ip)->i_sb->s_dev;
+		__entry->ino = ip->i_ino;
+		__entry->idx = idx;
+		__entry->startoff = r->br_startoff;
+		__entry->startblock = r->br_startblock;
+		__entry->blockcount = r->br_blockcount;
+		__entry->state = r->br_state;
+		__entry->bmap_state = state;
+		__entry->caller_ip = caller_ip;
+	),
+	TP_printk("dev %d:%d ino 0x%llx state %s idx %ld "
+		  "offset %lld block %lld count %lld flag %d caller %pf",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->ino,
+		  __print_flags(__entry->bmap_state, "|", XFS_BMAP_EXT_FLAGS),
+		  (long)__entry->idx,
+		  __entry->startoff,
+		  (__int64_t)__entry->startblock,
+		  __entry->blockcount,
+		  __entry->state,
+		  (char *)__entry->caller_ip)
+);
+
+DECLARE_EVENT_CLASS(xfs_bmap_class,
+	TP_PROTO(struct xfs_inode *ip, xfs_extnum_t idx, int state,
+		 unsigned long caller_ip),
+	TP_ARGS(ip, idx, state, caller_ip),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_ino_t, ino)
+		__field(xfs_extnum_t, idx)
+		__field(xfs_fileoff_t, startoff)
+		__field(xfs_fsblock_t, startblock)
+		__field(xfs_filblks_t, blockcount)
+		__field(xfs_exntst_t, state)
+		__field(int, bmap_state)
+		__field(unsigned long, caller_ip)
+	),
+	TP_fast_assign(
+		struct xfs_ifork	*ifp = (state & BMAP_ATTRFORK) ?
+						ip->i_afp : &ip->i_df;
+		struct xfs_bmbt_irec	r;
+
+		xfs_bmbt_get_all(xfs_iext_get_ext(ifp, idx), &r);
+		__entry->dev = VFS_I(ip)->i_sb->s_dev;
+		__entry->ino = ip->i_ino;
+		__entry->idx = idx;
+		__entry->startoff = r.br_startoff;
+		__entry->startblock = r.br_startblock;
+		__entry->blockcount = r.br_blockcount;
+		__entry->state = r.br_state;
+		__entry->bmap_state = state;
+		__entry->caller_ip = caller_ip;
+	),
+	TP_printk("dev %d:%d ino 0x%llx state %s idx %ld "
+		  "offset %lld block %lld count %lld flag %d caller %pf",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->ino,
+		  __print_flags(__entry->bmap_state, "|", XFS_BMAP_EXT_FLAGS),
+		  (long)__entry->idx,
+		  __entry->startoff,
+		  (__int64_t)__entry->startblock,
+		  __entry->blockcount,
+		  __entry->state,
+		  (char *)__entry->caller_ip)
+)
+
+#define DEFINE_BMAP_EVENT(name) \
+DEFINE_EVENT(xfs_bmap_class, name, \
+	TP_PROTO(struct xfs_inode *ip, xfs_extnum_t idx, int state, \
+		 unsigned long caller_ip), \
+	TP_ARGS(ip, idx, state, caller_ip))
+DEFINE_BMAP_EVENT(xfs_iext_remove);
+DEFINE_BMAP_EVENT(xfs_bmap_pre_update);
+DEFINE_BMAP_EVENT(xfs_bmap_post_update);
+DEFINE_BMAP_EVENT(xfs_extlist);
+
+DECLARE_EVENT_CLASS(xfs_buf_class,
+	TP_PROTO(struct xfs_buf *bp, unsigned long caller_ip),
+	TP_ARGS(bp, caller_ip),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_daddr_t, bno)
+		__field(size_t, buffer_length)
+		__field(int, hold)
+		__field(int, pincount)
+		__field(unsigned, lockval)
+		__field(unsigned, flags)
+		__field(unsigned long, caller_ip)
+	),
+	TP_fast_assign(
+		__entry->dev = bp->b_target->bt_dev;
+		__entry->bno = bp->b_bn;
+		__entry->buffer_length = bp->b_buffer_length;
+		__entry->hold = atomic_read(&bp->b_hold);
+		__entry->pincount = atomic_read(&bp->b_pin_count);
+		__entry->lockval = xfs_buf_lock_value(bp);
+		__entry->flags = bp->b_flags;
+		__entry->caller_ip = caller_ip;
+	),
+	TP_printk("dev %d:%d bno 0x%llx len 0x%zx hold %d pincount %d "
+		  "lock %d flags %s caller %pf",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  (unsigned long long)__entry->bno,
+		  __entry->buffer_length,
+		  __entry->hold,
+		  __entry->pincount,
+		  __entry->lockval,
+		  __print_flags(__entry->flags, "|", XFS_BUF_FLAGS),
+		  (void *)__entry->caller_ip)
+)
+
+#define DEFINE_BUF_EVENT(name) \
+DEFINE_EVENT(xfs_buf_class, name, \
+	TP_PROTO(struct xfs_buf *bp, unsigned long caller_ip), \
+	TP_ARGS(bp, caller_ip))
+DEFINE_BUF_EVENT(xfs_buf_init);
+DEFINE_BUF_EVENT(xfs_buf_free);
+DEFINE_BUF_EVENT(xfs_buf_hold);
+DEFINE_BUF_EVENT(xfs_buf_rele);
+DEFINE_BUF_EVENT(xfs_buf_iodone);
+DEFINE_BUF_EVENT(xfs_buf_iorequest);
+DEFINE_BUF_EVENT(xfs_buf_bawrite);
+DEFINE_BUF_EVENT(xfs_buf_bdwrite);
+DEFINE_BUF_EVENT(xfs_buf_lock);
+DEFINE_BUF_EVENT(xfs_buf_lock_done);
+DEFINE_BUF_EVENT(xfs_buf_cond_lock);
+DEFINE_BUF_EVENT(xfs_buf_unlock);
+DEFINE_BUF_EVENT(xfs_buf_iowait);
+DEFINE_BUF_EVENT(xfs_buf_iowait_done);
+DEFINE_BUF_EVENT(xfs_buf_delwri_queue);
+DEFINE_BUF_EVENT(xfs_buf_delwri_dequeue);
+DEFINE_BUF_EVENT(xfs_buf_delwri_split);
+DEFINE_BUF_EVENT(xfs_buf_get_uncached);
+DEFINE_BUF_EVENT(xfs_bdstrat_shut);
+DEFINE_BUF_EVENT(xfs_buf_item_relse);
+DEFINE_BUF_EVENT(xfs_buf_item_iodone);
+DEFINE_BUF_EVENT(xfs_buf_item_iodone_async);
+DEFINE_BUF_EVENT(xfs_buf_error_relse);
+DEFINE_BUF_EVENT(xfs_trans_read_buf_io);
+DEFINE_BUF_EVENT(xfs_trans_read_buf_shut);
+
+/* not really buffer traces, but the buf provides useful information */
+DEFINE_BUF_EVENT(xfs_btree_corrupt);
+DEFINE_BUF_EVENT(xfs_da_btree_corrupt);
+DEFINE_BUF_EVENT(xfs_reset_dqcounts);
+DEFINE_BUF_EVENT(xfs_inode_item_push);
+
+/* pass flags explicitly */
+DECLARE_EVENT_CLASS(xfs_buf_flags_class,
+	TP_PROTO(struct xfs_buf *bp, unsigned flags, unsigned long caller_ip),
+	TP_ARGS(bp, flags, caller_ip),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_daddr_t, bno)
+		__field(size_t, buffer_length)
+		__field(int, hold)
+		__field(int, pincount)
+		__field(unsigned, lockval)
+		__field(unsigned, flags)
+		__field(unsigned long, caller_ip)
+	),
+	TP_fast_assign(
+		__entry->dev = bp->b_target->bt_dev;
+		__entry->bno = bp->b_bn;
+		__entry->buffer_length = bp->b_buffer_length;
+		__entry->flags = flags;
+		__entry->hold = atomic_read(&bp->b_hold);
+		__entry->pincount = atomic_read(&bp->b_pin_count);
+		__entry->lockval = xfs_buf_lock_value(bp);
+		__entry->caller_ip = caller_ip;
+	),
+	TP_printk("dev %d:%d bno 0x%llx len 0x%zx hold %d pincount %d "
+		  "lock %d flags %s caller %pf",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  (unsigned long long)__entry->bno,
+		  __entry->buffer_length,
+		  __entry->hold,
+		  __entry->pincount,
+		  __entry->lockval,
+		  __print_flags(__entry->flags, "|", XFS_BUF_FLAGS),
+		  (void *)__entry->caller_ip)
+)
+
+#define DEFINE_BUF_FLAGS_EVENT(name) \
+DEFINE_EVENT(xfs_buf_flags_class, name, \
+	TP_PROTO(struct xfs_buf *bp, unsigned flags, unsigned long caller_ip), \
+	TP_ARGS(bp, flags, caller_ip))
+DEFINE_BUF_FLAGS_EVENT(xfs_buf_find);
+DEFINE_BUF_FLAGS_EVENT(xfs_buf_get);
+DEFINE_BUF_FLAGS_EVENT(xfs_buf_read);
+
+TRACE_EVENT(xfs_buf_ioerror,
+	TP_PROTO(struct xfs_buf *bp, int error, unsigned long caller_ip),
+	TP_ARGS(bp, error, caller_ip),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_daddr_t, bno)
+		__field(size_t, buffer_length)
+		__field(unsigned, flags)
+		__field(int, hold)
+		__field(int, pincount)
+		__field(unsigned, lockval)
+		__field(int, error)
+		__field(unsigned long, caller_ip)
+	),
+	TP_fast_assign(
+		__entry->dev = bp->b_target->bt_dev;
+		__entry->bno = bp->b_bn;
+		__entry->buffer_length = bp->b_buffer_length;
+		__entry->hold = atomic_read(&bp->b_hold);
+		__entry->pincount = atomic_read(&bp->b_pin_count);
+		__entry->lockval = xfs_buf_lock_value(bp);
+		__entry->error = error;
+		__entry->flags = bp->b_flags;
+		__entry->caller_ip = caller_ip;
+	),
+	TP_printk("dev %d:%d bno 0x%llx len 0x%zx hold %d pincount %d "
+		  "lock %d error %d flags %s caller %pf",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  (unsigned long long)__entry->bno,
+		  __entry->buffer_length,
+		  __entry->hold,
+		  __entry->pincount,
+		  __entry->lockval,
+		  __entry->error,
+		  __print_flags(__entry->flags, "|", XFS_BUF_FLAGS),
+		  (void *)__entry->caller_ip)
+);
+
+DECLARE_EVENT_CLASS(xfs_buf_item_class,
+	TP_PROTO(struct xfs_buf_log_item *bip),
+	TP_ARGS(bip),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_daddr_t, buf_bno)
+		__field(size_t, buf_len)
+		__field(int, buf_hold)
+		__field(int, buf_pincount)
+		__field(int, buf_lockval)
+		__field(unsigned, buf_flags)
+		__field(unsigned, bli_recur)
+		__field(int, bli_refcount)
+		__field(unsigned, bli_flags)
+		__field(void *, li_desc)
+		__field(unsigned, li_flags)
+	),
+	TP_fast_assign(
+		__entry->dev = bip->bli_buf->b_target->bt_dev;
+		__entry->bli_flags = bip->bli_flags;
+		__entry->bli_recur = bip->bli_recur;
+		__entry->bli_refcount = atomic_read(&bip->bli_refcount);
+		__entry->buf_bno = bip->bli_buf->b_bn;
+		__entry->buf_len = bip->bli_buf->b_buffer_length;
+		__entry->buf_flags = bip->bli_buf->b_flags;
+		__entry->buf_hold = atomic_read(&bip->bli_buf->b_hold);
+		__entry->buf_pincount = atomic_read(&bip->bli_buf->b_pin_count);
+		__entry->buf_lockval = xfs_buf_lock_value(bip->bli_buf);
+		__entry->li_desc = bip->bli_item.li_desc;
+		__entry->li_flags = bip->bli_item.li_flags;
+	),
+	TP_printk("dev %d:%d bno 0x%llx len 0x%zx hold %d pincount %d "
+		  "lock %d flags %s recur %d refcount %d bliflags %s "
+		  "lidesc 0x%p liflags %s",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  (unsigned long long)__entry->buf_bno,
+		  __entry->buf_len,
+		  __entry->buf_hold,
+		  __entry->buf_pincount,
+		  __entry->buf_lockval,
+		  __print_flags(__entry->buf_flags, "|", XFS_BUF_FLAGS),
+		  __entry->bli_recur,
+		  __entry->bli_refcount,
+		  __print_flags(__entry->bli_flags, "|", XFS_BLI_FLAGS),
+		  __entry->li_desc,
+		  __print_flags(__entry->li_flags, "|", XFS_LI_FLAGS))
+)
+
+#define DEFINE_BUF_ITEM_EVENT(name) \
+DEFINE_EVENT(xfs_buf_item_class, name, \
+	TP_PROTO(struct xfs_buf_log_item *bip), \
+	TP_ARGS(bip))
+DEFINE_BUF_ITEM_EVENT(xfs_buf_item_size);
+DEFINE_BUF_ITEM_EVENT(xfs_buf_item_size_stale);
+DEFINE_BUF_ITEM_EVENT(xfs_buf_item_format);
+DEFINE_BUF_ITEM_EVENT(xfs_buf_item_format_stale);
+DEFINE_BUF_ITEM_EVENT(xfs_buf_item_pin);
+DEFINE_BUF_ITEM_EVENT(xfs_buf_item_unpin);
+DEFINE_BUF_ITEM_EVENT(xfs_buf_item_unpin_stale);
+DEFINE_BUF_ITEM_EVENT(xfs_buf_item_trylock);
+DEFINE_BUF_ITEM_EVENT(xfs_buf_item_unlock);
+DEFINE_BUF_ITEM_EVENT(xfs_buf_item_unlock_stale);
+DEFINE_BUF_ITEM_EVENT(xfs_buf_item_committed);
+DEFINE_BUF_ITEM_EVENT(xfs_buf_item_push);
+DEFINE_BUF_ITEM_EVENT(xfs_buf_item_pushbuf);
+DEFINE_BUF_ITEM_EVENT(xfs_trans_get_buf);
+DEFINE_BUF_ITEM_EVENT(xfs_trans_get_buf_recur);
+DEFINE_BUF_ITEM_EVENT(xfs_trans_getsb);
+DEFINE_BUF_ITEM_EVENT(xfs_trans_getsb_recur);
+DEFINE_BUF_ITEM_EVENT(xfs_trans_read_buf);
+DEFINE_BUF_ITEM_EVENT(xfs_trans_read_buf_recur);
+DEFINE_BUF_ITEM_EVENT(xfs_trans_log_buf);
+DEFINE_BUF_ITEM_EVENT(xfs_trans_brelse);
+DEFINE_BUF_ITEM_EVENT(xfs_trans_bjoin);
+DEFINE_BUF_ITEM_EVENT(xfs_trans_bhold);
+DEFINE_BUF_ITEM_EVENT(xfs_trans_bhold_release);
+DEFINE_BUF_ITEM_EVENT(xfs_trans_binval);
+
+DECLARE_EVENT_CLASS(xfs_lock_class,
+	TP_PROTO(struct xfs_inode *ip, unsigned lock_flags,
+		 unsigned long caller_ip),
+	TP_ARGS(ip,  lock_flags, caller_ip),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_ino_t, ino)
+		__field(int, lock_flags)
+		__field(unsigned long, caller_ip)
+	),
+	TP_fast_assign(
+		__entry->dev = VFS_I(ip)->i_sb->s_dev;
+		__entry->ino = ip->i_ino;
+		__entry->lock_flags = lock_flags;
+		__entry->caller_ip = caller_ip;
+	),
+	TP_printk("dev %d:%d ino 0x%llx flags %s caller %pf",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->ino,
+		  __print_flags(__entry->lock_flags, "|", XFS_LOCK_FLAGS),
+		  (void *)__entry->caller_ip)
+)
+
+#define DEFINE_LOCK_EVENT(name) \
+DEFINE_EVENT(xfs_lock_class, name, \
+	TP_PROTO(struct xfs_inode *ip, unsigned lock_flags, \
+		 unsigned long caller_ip), \
+	TP_ARGS(ip,  lock_flags, caller_ip))
+DEFINE_LOCK_EVENT(xfs_ilock);
+DEFINE_LOCK_EVENT(xfs_ilock_nowait);
+DEFINE_LOCK_EVENT(xfs_ilock_demote);
+DEFINE_LOCK_EVENT(xfs_iunlock);
+
+DECLARE_EVENT_CLASS(xfs_inode_class,
+	TP_PROTO(struct xfs_inode *ip),
+	TP_ARGS(ip),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_ino_t, ino)
+	),
+	TP_fast_assign(
+		__entry->dev = VFS_I(ip)->i_sb->s_dev;
+		__entry->ino = ip->i_ino;
+	),
+	TP_printk("dev %d:%d ino 0x%llx",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->ino)
+)
+
+#define DEFINE_INODE_EVENT(name) \
+DEFINE_EVENT(xfs_inode_class, name, \
+	TP_PROTO(struct xfs_inode *ip), \
+	TP_ARGS(ip))
+DEFINE_INODE_EVENT(xfs_iget_skip);
+DEFINE_INODE_EVENT(xfs_iget_reclaim);
+DEFINE_INODE_EVENT(xfs_iget_reclaim_fail);
+DEFINE_INODE_EVENT(xfs_iget_hit);
+DEFINE_INODE_EVENT(xfs_iget_miss);
+
+DEFINE_INODE_EVENT(xfs_getattr);
+DEFINE_INODE_EVENT(xfs_setattr);
+DEFINE_INODE_EVENT(xfs_readlink);
+DEFINE_INODE_EVENT(xfs_alloc_file_space);
+DEFINE_INODE_EVENT(xfs_free_file_space);
+DEFINE_INODE_EVENT(xfs_readdir);
+#ifdef CONFIG_XFS_POSIX_ACL
+DEFINE_INODE_EVENT(xfs_check_acl);
+#endif
+DEFINE_INODE_EVENT(xfs_vm_bmap);
+DEFINE_INODE_EVENT(xfs_file_ioctl);
+DEFINE_INODE_EVENT(xfs_file_compat_ioctl);
+DEFINE_INODE_EVENT(xfs_ioctl_setattr);
+DEFINE_INODE_EVENT(xfs_file_fsync);
+DEFINE_INODE_EVENT(xfs_destroy_inode);
+DEFINE_INODE_EVENT(xfs_write_inode);
+DEFINE_INODE_EVENT(xfs_evict_inode);
+
+DEFINE_INODE_EVENT(xfs_dquot_dqalloc);
+DEFINE_INODE_EVENT(xfs_dquot_dqdetach);
+
+DECLARE_EVENT_CLASS(xfs_iref_class,
+	TP_PROTO(struct xfs_inode *ip, unsigned long caller_ip),
+	TP_ARGS(ip, caller_ip),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_ino_t, ino)
+		__field(int, count)
+		__field(int, pincount)
+		__field(unsigned long, caller_ip)
+	),
+	TP_fast_assign(
+		__entry->dev = VFS_I(ip)->i_sb->s_dev;
+		__entry->ino = ip->i_ino;
+		__entry->count = atomic_read(&VFS_I(ip)->i_count);
+		__entry->pincount = atomic_read(&ip->i_pincount);
+		__entry->caller_ip = caller_ip;
+	),
+	TP_printk("dev %d:%d ino 0x%llx count %d pincount %d caller %pf",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->ino,
+		  __entry->count,
+		  __entry->pincount,
+		  (char *)__entry->caller_ip)
+)
+
+#define DEFINE_IREF_EVENT(name) \
+DEFINE_EVENT(xfs_iref_class, name, \
+	TP_PROTO(struct xfs_inode *ip, unsigned long caller_ip), \
+	TP_ARGS(ip, caller_ip))
+DEFINE_IREF_EVENT(xfs_ihold);
+DEFINE_IREF_EVENT(xfs_irele);
+DEFINE_IREF_EVENT(xfs_inode_pin);
+DEFINE_IREF_EVENT(xfs_inode_unpin);
+DEFINE_IREF_EVENT(xfs_inode_unpin_nowait);
+
+DECLARE_EVENT_CLASS(xfs_namespace_class,
+	TP_PROTO(struct xfs_inode *dp, struct xfs_name *name),
+	TP_ARGS(dp, name),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_ino_t, dp_ino)
+		__dynamic_array(char, name, name->len)
+	),
+	TP_fast_assign(
+		__entry->dev = VFS_I(dp)->i_sb->s_dev;
+		__entry->dp_ino = dp->i_ino;
+		memcpy(__get_str(name), name->name, name->len);
+	),
+	TP_printk("dev %d:%d dp ino 0x%llx name %s",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->dp_ino,
+		  __get_str(name))
+)
+
+#define DEFINE_NAMESPACE_EVENT(name) \
+DEFINE_EVENT(xfs_namespace_class, name, \
+	TP_PROTO(struct xfs_inode *dp, struct xfs_name *name), \
+	TP_ARGS(dp, name))
+DEFINE_NAMESPACE_EVENT(xfs_remove);
+DEFINE_NAMESPACE_EVENT(xfs_link);
+DEFINE_NAMESPACE_EVENT(xfs_lookup);
+DEFINE_NAMESPACE_EVENT(xfs_create);
+DEFINE_NAMESPACE_EVENT(xfs_symlink);
+
+TRACE_EVENT(xfs_rename,
+	TP_PROTO(struct xfs_inode *src_dp, struct xfs_inode *target_dp,
+		 struct xfs_name *src_name, struct xfs_name *target_name),
+	TP_ARGS(src_dp, target_dp, src_name, target_name),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_ino_t, src_dp_ino)
+		__field(xfs_ino_t, target_dp_ino)
+		__dynamic_array(char, src_name, src_name->len)
+		__dynamic_array(char, target_name, target_name->len)
+	),
+	TP_fast_assign(
+		__entry->dev = VFS_I(src_dp)->i_sb->s_dev;
+		__entry->src_dp_ino = src_dp->i_ino;
+		__entry->target_dp_ino = target_dp->i_ino;
+		memcpy(__get_str(src_name), src_name->name, src_name->len);
+		memcpy(__get_str(target_name), target_name->name, target_name->len);
+	),
+	TP_printk("dev %d:%d src dp ino 0x%llx target dp ino 0x%llx"
+		  " src name %s target name %s",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->src_dp_ino,
+		  __entry->target_dp_ino,
+		  __get_str(src_name),
+		  __get_str(target_name))
+)
+
+DECLARE_EVENT_CLASS(xfs_dquot_class,
+	TP_PROTO(struct xfs_dquot *dqp),
+	TP_ARGS(dqp),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(u32, id)
+		__field(unsigned, flags)
+		__field(unsigned, nrefs)
+		__field(unsigned long long, res_bcount)
+		__field(unsigned long long, bcount)
+		__field(unsigned long long, icount)
+		__field(unsigned long long, blk_hardlimit)
+		__field(unsigned long long, blk_softlimit)
+		__field(unsigned long long, ino_hardlimit)
+		__field(unsigned long long, ino_softlimit)
+	), \
+	TP_fast_assign(
+		__entry->dev = dqp->q_mount->m_super->s_dev;
+		__entry->id = be32_to_cpu(dqp->q_core.d_id);
+		__entry->flags = dqp->dq_flags;
+		__entry->nrefs = dqp->q_nrefs;
+		__entry->res_bcount = dqp->q_res_bcount;
+		__entry->bcount = be64_to_cpu(dqp->q_core.d_bcount);
+		__entry->icount = be64_to_cpu(dqp->q_core.d_icount);
+		__entry->blk_hardlimit =
+			be64_to_cpu(dqp->q_core.d_blk_hardlimit);
+		__entry->blk_softlimit =
+			be64_to_cpu(dqp->q_core.d_blk_softlimit);
+		__entry->ino_hardlimit =
+			be64_to_cpu(dqp->q_core.d_ino_hardlimit);
+		__entry->ino_softlimit =
+			be64_to_cpu(dqp->q_core.d_ino_softlimit);
+	),
+	TP_printk("dev %d:%d id 0x%x flags %s nrefs %u res_bc 0x%llx "
+		  "bcnt 0x%llx bhardlimit 0x%llx bsoftlimit 0x%llx "
+		  "icnt 0x%llx ihardlimit 0x%llx isoftlimit 0x%llx]",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->id,
+		  __print_flags(__entry->flags, "|", XFS_DQ_FLAGS),
+		  __entry->nrefs,
+		  __entry->res_bcount,
+		  __entry->bcount,
+		  __entry->blk_hardlimit,
+		  __entry->blk_softlimit,
+		  __entry->icount,
+		  __entry->ino_hardlimit,
+		  __entry->ino_softlimit)
+)
+
+#define DEFINE_DQUOT_EVENT(name) \
+DEFINE_EVENT(xfs_dquot_class, name, \
+	TP_PROTO(struct xfs_dquot *dqp), \
+	TP_ARGS(dqp))
+DEFINE_DQUOT_EVENT(xfs_dqadjust);
+DEFINE_DQUOT_EVENT(xfs_dqreclaim_want);
+DEFINE_DQUOT_EVENT(xfs_dqreclaim_dirty);
+DEFINE_DQUOT_EVENT(xfs_dqreclaim_unlink);
+DEFINE_DQUOT_EVENT(xfs_dqattach_found);
+DEFINE_DQUOT_EVENT(xfs_dqattach_get);
+DEFINE_DQUOT_EVENT(xfs_dqinit);
+DEFINE_DQUOT_EVENT(xfs_dqreuse);
+DEFINE_DQUOT_EVENT(xfs_dqalloc);
+DEFINE_DQUOT_EVENT(xfs_dqtobp_read);
+DEFINE_DQUOT_EVENT(xfs_dqread);
+DEFINE_DQUOT_EVENT(xfs_dqread_fail);
+DEFINE_DQUOT_EVENT(xfs_dqlookup_found);
+DEFINE_DQUOT_EVENT(xfs_dqlookup_want);
+DEFINE_DQUOT_EVENT(xfs_dqlookup_freelist);
+DEFINE_DQUOT_EVENT(xfs_dqlookup_done);
+DEFINE_DQUOT_EVENT(xfs_dqget_hit);
+DEFINE_DQUOT_EVENT(xfs_dqget_miss);
+DEFINE_DQUOT_EVENT(xfs_dqput);
+DEFINE_DQUOT_EVENT(xfs_dqput_wait);
+DEFINE_DQUOT_EVENT(xfs_dqput_free);
+DEFINE_DQUOT_EVENT(xfs_dqrele);
+DEFINE_DQUOT_EVENT(xfs_dqflush);
+DEFINE_DQUOT_EVENT(xfs_dqflush_force);
+DEFINE_DQUOT_EVENT(xfs_dqflush_done);
+
+DECLARE_EVENT_CLASS(xfs_loggrant_class,
+	TP_PROTO(struct log *log, struct xlog_ticket *tic),
+	TP_ARGS(log, tic),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(unsigned, trans_type)
+		__field(char, ocnt)
+		__field(char, cnt)
+		__field(int, curr_res)
+		__field(int, unit_res)
+		__field(unsigned int, flags)
+		__field(int, reserveq)
+		__field(int, writeq)
+		__field(int, grant_reserve_cycle)
+		__field(int, grant_reserve_bytes)
+		__field(int, grant_write_cycle)
+		__field(int, grant_write_bytes)
+		__field(int, curr_cycle)
+		__field(int, curr_block)
+		__field(xfs_lsn_t, tail_lsn)
+	),
+	TP_fast_assign(
+		__entry->dev = log->l_mp->m_super->s_dev;
+		__entry->trans_type = tic->t_trans_type;
+		__entry->ocnt = tic->t_ocnt;
+		__entry->cnt = tic->t_cnt;
+		__entry->curr_res = tic->t_curr_res;
+		__entry->unit_res = tic->t_unit_res;
+		__entry->flags = tic->t_flags;
+		__entry->reserveq = list_empty(&log->l_reserveq);
+		__entry->writeq = list_empty(&log->l_writeq);
+		xlog_crack_grant_head(&log->l_grant_reserve_head,
+				&__entry->grant_reserve_cycle,
+				&__entry->grant_reserve_bytes);
+		xlog_crack_grant_head(&log->l_grant_write_head,
+				&__entry->grant_write_cycle,
+				&__entry->grant_write_bytes);
+		__entry->curr_cycle = log->l_curr_cycle;
+		__entry->curr_block = log->l_curr_block;
+		__entry->tail_lsn = atomic64_read(&log->l_tail_lsn);
+	),
+	TP_printk("dev %d:%d type %s t_ocnt %u t_cnt %u t_curr_res %u "
+		  "t_unit_res %u t_flags %s reserveq %s "
+		  "writeq %s grant_reserve_cycle %d "
+		  "grant_reserve_bytes %d grant_write_cycle %d "
+		  "grant_write_bytes %d curr_cycle %d curr_block %d "
+		  "tail_cycle %d tail_block %d",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __print_symbolic(__entry->trans_type, XFS_TRANS_TYPES),
+		  __entry->ocnt,
+		  __entry->cnt,
+		  __entry->curr_res,
+		  __entry->unit_res,
+		  __print_flags(__entry->flags, "|", XLOG_TIC_FLAGS),
+		  __entry->reserveq ? "empty" : "active",
+		  __entry->writeq ? "empty" : "active",
+		  __entry->grant_reserve_cycle,
+		  __entry->grant_reserve_bytes,
+		  __entry->grant_write_cycle,
+		  __entry->grant_write_bytes,
+		  __entry->curr_cycle,
+		  __entry->curr_block,
+		  CYCLE_LSN(__entry->tail_lsn),
+		  BLOCK_LSN(__entry->tail_lsn)
+	)
+)
+
+#define DEFINE_LOGGRANT_EVENT(name) \
+DEFINE_EVENT(xfs_loggrant_class, name, \
+	TP_PROTO(struct log *log, struct xlog_ticket *tic), \
+	TP_ARGS(log, tic))
+DEFINE_LOGGRANT_EVENT(xfs_log_done_nonperm);
+DEFINE_LOGGRANT_EVENT(xfs_log_done_perm);
+DEFINE_LOGGRANT_EVENT(xfs_log_reserve);
+DEFINE_LOGGRANT_EVENT(xfs_log_umount_write);
+DEFINE_LOGGRANT_EVENT(xfs_log_grant_enter);
+DEFINE_LOGGRANT_EVENT(xfs_log_grant_exit);
+DEFINE_LOGGRANT_EVENT(xfs_log_grant_error);
+DEFINE_LOGGRANT_EVENT(xfs_log_grant_sleep1);
+DEFINE_LOGGRANT_EVENT(xfs_log_grant_wake1);
+DEFINE_LOGGRANT_EVENT(xfs_log_grant_sleep2);
+DEFINE_LOGGRANT_EVENT(xfs_log_grant_wake2);
+DEFINE_LOGGRANT_EVENT(xfs_log_grant_wake_up);
+DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_enter);
+DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_exit);
+DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_error);
+DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_sleep1);
+DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_wake1);
+DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_sleep2);
+DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_wake2);
+DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_wake_up);
+DEFINE_LOGGRANT_EVENT(xfs_log_regrant_reserve_enter);
+DEFINE_LOGGRANT_EVENT(xfs_log_regrant_reserve_exit);
+DEFINE_LOGGRANT_EVENT(xfs_log_regrant_reserve_sub);
+DEFINE_LOGGRANT_EVENT(xfs_log_ungrant_enter);
+DEFINE_LOGGRANT_EVENT(xfs_log_ungrant_exit);
+DEFINE_LOGGRANT_EVENT(xfs_log_ungrant_sub);
+
+DECLARE_EVENT_CLASS(xfs_file_class,
+	TP_PROTO(struct xfs_inode *ip, size_t count, loff_t offset, int flags),
+	TP_ARGS(ip, count, offset, flags),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_ino_t, ino)
+		__field(xfs_fsize_t, size)
+		__field(xfs_fsize_t, new_size)
+		__field(loff_t, offset)
+		__field(size_t, count)
+		__field(int, flags)
+	),
+	TP_fast_assign(
+		__entry->dev = VFS_I(ip)->i_sb->s_dev;
+		__entry->ino = ip->i_ino;
+		__entry->size = ip->i_d.di_size;
+		__entry->new_size = ip->i_new_size;
+		__entry->offset = offset;
+		__entry->count = count;
+		__entry->flags = flags;
+	),
+	TP_printk("dev %d:%d ino 0x%llx size 0x%llx new_size 0x%llx "
+		  "offset 0x%llx count 0x%zx ioflags %s",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->ino,
+		  __entry->size,
+		  __entry->new_size,
+		  __entry->offset,
+		  __entry->count,
+		  __print_flags(__entry->flags, "|", XFS_IO_FLAGS))
+)
+
+#define DEFINE_RW_EVENT(name)		\
+DEFINE_EVENT(xfs_file_class, name,	\
+	TP_PROTO(struct xfs_inode *ip, size_t count, loff_t offset, int flags),	\
+	TP_ARGS(ip, count, offset, flags))
+DEFINE_RW_EVENT(xfs_file_read);
+DEFINE_RW_EVENT(xfs_file_buffered_write);
+DEFINE_RW_EVENT(xfs_file_direct_write);
+DEFINE_RW_EVENT(xfs_file_splice_read);
+DEFINE_RW_EVENT(xfs_file_splice_write);
+
+DECLARE_EVENT_CLASS(xfs_page_class,
+	TP_PROTO(struct inode *inode, struct page *page, unsigned long off),
+	TP_ARGS(inode, page, off),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_ino_t, ino)
+		__field(pgoff_t, pgoff)
+		__field(loff_t, size)
+		__field(unsigned long, offset)
+		__field(int, delalloc)
+		__field(int, unwritten)
+	),
+	TP_fast_assign(
+		int delalloc = -1, unwritten = -1;
+
+		if (page_has_buffers(page))
+			xfs_count_page_state(page, &delalloc, &unwritten);
+		__entry->dev = inode->i_sb->s_dev;
+		__entry->ino = XFS_I(inode)->i_ino;
+		__entry->pgoff = page_offset(page);
+		__entry->size = i_size_read(inode);
+		__entry->offset = off;
+		__entry->delalloc = delalloc;
+		__entry->unwritten = unwritten;
+	),
+	TP_printk("dev %d:%d ino 0x%llx pgoff 0x%lx size 0x%llx offset %lx "
+		  "delalloc %d unwritten %d",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->ino,
+		  __entry->pgoff,
+		  __entry->size,
+		  __entry->offset,
+		  __entry->delalloc,
+		  __entry->unwritten)
+)
+
+#define DEFINE_PAGE_EVENT(name)		\
+DEFINE_EVENT(xfs_page_class, name,	\
+	TP_PROTO(struct inode *inode, struct page *page, unsigned long off),	\
+	TP_ARGS(inode, page, off))
+DEFINE_PAGE_EVENT(xfs_writepage);
+DEFINE_PAGE_EVENT(xfs_releasepage);
+DEFINE_PAGE_EVENT(xfs_invalidatepage);
+
+DECLARE_EVENT_CLASS(xfs_imap_class,
+	TP_PROTO(struct xfs_inode *ip, xfs_off_t offset, ssize_t count,
+		 int type, struct xfs_bmbt_irec *irec),
+	TP_ARGS(ip, offset, count, type, irec),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_ino_t, ino)
+		__field(loff_t, size)
+		__field(loff_t, new_size)
+		__field(loff_t, offset)
+		__field(size_t, count)
+		__field(int, type)
+		__field(xfs_fileoff_t, startoff)
+		__field(xfs_fsblock_t, startblock)
+		__field(xfs_filblks_t, blockcount)
+	),
+	TP_fast_assign(
+		__entry->dev = VFS_I(ip)->i_sb->s_dev;
+		__entry->ino = ip->i_ino;
+		__entry->size = ip->i_d.di_size;
+		__entry->new_size = ip->i_new_size;
+		__entry->offset = offset;
+		__entry->count = count;
+		__entry->type = type;
+		__entry->startoff = irec ? irec->br_startoff : 0;
+		__entry->startblock = irec ? irec->br_startblock : 0;
+		__entry->blockcount = irec ? irec->br_blockcount : 0;
+	),
+	TP_printk("dev %d:%d ino 0x%llx size 0x%llx new_size 0x%llx "
+		  "offset 0x%llx count %zd type %s "
+		  "startoff 0x%llx startblock %lld blockcount 0x%llx",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->ino,
+		  __entry->size,
+		  __entry->new_size,
+		  __entry->offset,
+		  __entry->count,
+		  __print_symbolic(__entry->type, XFS_IO_TYPES),
+		  __entry->startoff,
+		  (__int64_t)__entry->startblock,
+		  __entry->blockcount)
+)
+
+#define DEFINE_IOMAP_EVENT(name)	\
+DEFINE_EVENT(xfs_imap_class, name,	\
+	TP_PROTO(struct xfs_inode *ip, xfs_off_t offset, ssize_t count,	\
+		 int type, struct xfs_bmbt_irec *irec),		\
+	TP_ARGS(ip, offset, count, type, irec))
+DEFINE_IOMAP_EVENT(xfs_map_blocks_found);
+DEFINE_IOMAP_EVENT(xfs_map_blocks_alloc);
+DEFINE_IOMAP_EVENT(xfs_get_blocks_found);
+DEFINE_IOMAP_EVENT(xfs_get_blocks_alloc);
+
+DECLARE_EVENT_CLASS(xfs_simple_io_class,
+	TP_PROTO(struct xfs_inode *ip, xfs_off_t offset, ssize_t count),
+	TP_ARGS(ip, offset, count),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_ino_t, ino)
+		__field(loff_t, size)
+		__field(loff_t, new_size)
+		__field(loff_t, offset)
+		__field(size_t, count)
+	),
+	TP_fast_assign(
+		__entry->dev = VFS_I(ip)->i_sb->s_dev;
+		__entry->ino = ip->i_ino;
+		__entry->size = ip->i_d.di_size;
+		__entry->new_size = ip->i_new_size;
+		__entry->offset = offset;
+		__entry->count = count;
+	),
+	TP_printk("dev %d:%d ino 0x%llx size 0x%llx new_size 0x%llx "
+		  "offset 0x%llx count %zd",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->ino,
+		  __entry->size,
+		  __entry->new_size,
+		  __entry->offset,
+		  __entry->count)
+);
+
+#define DEFINE_SIMPLE_IO_EVENT(name)	\
+DEFINE_EVENT(xfs_simple_io_class, name,	\
+	TP_PROTO(struct xfs_inode *ip, xfs_off_t offset, ssize_t count),	\
+	TP_ARGS(ip, offset, count))
+DEFINE_SIMPLE_IO_EVENT(xfs_delalloc_enospc);
+DEFINE_SIMPLE_IO_EVENT(xfs_unwritten_convert);
+DEFINE_SIMPLE_IO_EVENT(xfs_get_blocks_notfound);
+
+
+TRACE_EVENT(xfs_itruncate_start,
+	TP_PROTO(struct xfs_inode *ip, xfs_fsize_t new_size, int flag,
+		 xfs_off_t toss_start, xfs_off_t toss_finish),
+	TP_ARGS(ip, new_size, flag, toss_start, toss_finish),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_ino_t, ino)
+		__field(xfs_fsize_t, size)
+		__field(xfs_fsize_t, new_size)
+		__field(xfs_off_t, toss_start)
+		__field(xfs_off_t, toss_finish)
+		__field(int, flag)
+	),
+	TP_fast_assign(
+		__entry->dev = VFS_I(ip)->i_sb->s_dev;
+		__entry->ino = ip->i_ino;
+		__entry->size = ip->i_d.di_size;
+		__entry->new_size = new_size;
+		__entry->toss_start = toss_start;
+		__entry->toss_finish = toss_finish;
+		__entry->flag = flag;
+	),
+	TP_printk("dev %d:%d ino 0x%llx %s size 0x%llx new_size 0x%llx "
+		  "toss start 0x%llx toss finish 0x%llx",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->ino,
+		  __print_flags(__entry->flag, "|", XFS_ITRUNC_FLAGS),
+		  __entry->size,
+		  __entry->new_size,
+		  __entry->toss_start,
+		  __entry->toss_finish)
+);
+
+DECLARE_EVENT_CLASS(xfs_itrunc_class,
+	TP_PROTO(struct xfs_inode *ip, xfs_fsize_t new_size),
+	TP_ARGS(ip, new_size),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_ino_t, ino)
+		__field(xfs_fsize_t, size)
+		__field(xfs_fsize_t, new_size)
+	),
+	TP_fast_assign(
+		__entry->dev = VFS_I(ip)->i_sb->s_dev;
+		__entry->ino = ip->i_ino;
+		__entry->size = ip->i_d.di_size;
+		__entry->new_size = new_size;
+	),
+	TP_printk("dev %d:%d ino 0x%llx size 0x%llx new_size 0x%llx",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->ino,
+		  __entry->size,
+		  __entry->new_size)
+)
+
+#define DEFINE_ITRUNC_EVENT(name) \
+DEFINE_EVENT(xfs_itrunc_class, name, \
+	TP_PROTO(struct xfs_inode *ip, xfs_fsize_t new_size), \
+	TP_ARGS(ip, new_size))
+DEFINE_ITRUNC_EVENT(xfs_itruncate_finish_start);
+DEFINE_ITRUNC_EVENT(xfs_itruncate_finish_end);
+
+TRACE_EVENT(xfs_pagecache_inval,
+	TP_PROTO(struct xfs_inode *ip, xfs_off_t start, xfs_off_t finish),
+	TP_ARGS(ip, start, finish),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_ino_t, ino)
+		__field(xfs_fsize_t, size)
+		__field(xfs_off_t, start)
+		__field(xfs_off_t, finish)
+	),
+	TP_fast_assign(
+		__entry->dev = VFS_I(ip)->i_sb->s_dev;
+		__entry->ino = ip->i_ino;
+		__entry->size = ip->i_d.di_size;
+		__entry->start = start;
+		__entry->finish = finish;
+	),
+	TP_printk("dev %d:%d ino 0x%llx size 0x%llx start 0x%llx finish 0x%llx",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->ino,
+		  __entry->size,
+		  __entry->start,
+		  __entry->finish)
+);
+
+TRACE_EVENT(xfs_bunmap,
+	TP_PROTO(struct xfs_inode *ip, xfs_fileoff_t bno, xfs_filblks_t len,
+		 int flags, unsigned long caller_ip),
+	TP_ARGS(ip, bno, len, flags, caller_ip),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_ino_t, ino)
+		__field(xfs_fsize_t, size)
+		__field(xfs_fileoff_t, bno)
+		__field(xfs_filblks_t, len)
+		__field(unsigned long, caller_ip)
+		__field(int, flags)
+	),
+	TP_fast_assign(
+		__entry->dev = VFS_I(ip)->i_sb->s_dev;
+		__entry->ino = ip->i_ino;
+		__entry->size = ip->i_d.di_size;
+		__entry->bno = bno;
+		__entry->len = len;
+		__entry->caller_ip = caller_ip;
+		__entry->flags = flags;
+	),
+	TP_printk("dev %d:%d ino 0x%llx size 0x%llx bno 0x%llx len 0x%llx"
+		  "flags %s caller %pf",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->ino,
+		  __entry->size,
+		  __entry->bno,
+		  __entry->len,
+		  __print_flags(__entry->flags, "|", XFS_BMAPI_FLAGS),
+		  (void *)__entry->caller_ip)
+
+);
+
+DECLARE_EVENT_CLASS(xfs_busy_class,
+	TP_PROTO(struct xfs_mount *mp, xfs_agnumber_t agno,
+		 xfs_agblock_t agbno, xfs_extlen_t len),
+	TP_ARGS(mp, agno, agbno, len),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_agnumber_t, agno)
+		__field(xfs_agblock_t, agbno)
+		__field(xfs_extlen_t, len)
+	),
+	TP_fast_assign(
+		__entry->dev = mp->m_super->s_dev;
+		__entry->agno = agno;
+		__entry->agbno = agbno;
+		__entry->len = len;
+	),
+	TP_printk("dev %d:%d agno %u agbno %u len %u",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->agno,
+		  __entry->agbno,
+		  __entry->len)
+);
+#define DEFINE_BUSY_EVENT(name) \
+DEFINE_EVENT(xfs_busy_class, name, \
+	TP_PROTO(struct xfs_mount *mp, xfs_agnumber_t agno, \
+		 xfs_agblock_t agbno, xfs_extlen_t len), \
+	TP_ARGS(mp, agno, agbno, len))
+DEFINE_BUSY_EVENT(xfs_alloc_busy);
+DEFINE_BUSY_EVENT(xfs_alloc_busy_enomem);
+DEFINE_BUSY_EVENT(xfs_alloc_busy_force);
+DEFINE_BUSY_EVENT(xfs_alloc_busy_reuse);
+DEFINE_BUSY_EVENT(xfs_alloc_busy_clear);
+
+TRACE_EVENT(xfs_alloc_busy_trim,
+	TP_PROTO(struct xfs_mount *mp, xfs_agnumber_t agno,
+		 xfs_agblock_t agbno, xfs_extlen_t len,
+		 xfs_agblock_t tbno, xfs_extlen_t tlen),
+	TP_ARGS(mp, agno, agbno, len, tbno, tlen),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_agnumber_t, agno)
+		__field(xfs_agblock_t, agbno)
+		__field(xfs_extlen_t, len)
+		__field(xfs_agblock_t, tbno)
+		__field(xfs_extlen_t, tlen)
+	),
+	TP_fast_assign(
+		__entry->dev = mp->m_super->s_dev;
+		__entry->agno = agno;
+		__entry->agbno = agbno;
+		__entry->len = len;
+		__entry->tbno = tbno;
+		__entry->tlen = tlen;
+	),
+	TP_printk("dev %d:%d agno %u agbno %u len %u tbno %u tlen %u",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->agno,
+		  __entry->agbno,
+		  __entry->len,
+		  __entry->tbno,
+		  __entry->tlen)
+);
+
+TRACE_EVENT(xfs_trans_commit_lsn,
+	TP_PROTO(struct xfs_trans *trans),
+	TP_ARGS(trans),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(struct xfs_trans *, tp)
+		__field(xfs_lsn_t, lsn)
+	),
+	TP_fast_assign(
+		__entry->dev = trans->t_mountp->m_super->s_dev;
+		__entry->tp = trans;
+		__entry->lsn = trans->t_commit_lsn;
+	),
+	TP_printk("dev %d:%d trans 0x%p commit_lsn 0x%llx",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->tp,
+		  __entry->lsn)
+);
+
+TRACE_EVENT(xfs_agf,
+	TP_PROTO(struct xfs_mount *mp, struct xfs_agf *agf, int flags,
+		 unsigned long caller_ip),
+	TP_ARGS(mp, agf, flags, caller_ip),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_agnumber_t, agno)
+		__field(int, flags)
+		__field(__u32, length)
+		__field(__u32, bno_root)
+		__field(__u32, cnt_root)
+		__field(__u32, bno_level)
+		__field(__u32, cnt_level)
+		__field(__u32, flfirst)
+		__field(__u32, fllast)
+		__field(__u32, flcount)
+		__field(__u32, freeblks)
+		__field(__u32, longest)
+		__field(unsigned long, caller_ip)
+	),
+	TP_fast_assign(
+		__entry->dev = mp->m_super->s_dev;
+		__entry->agno = be32_to_cpu(agf->agf_seqno),
+		__entry->flags = flags;
+		__entry->length = be32_to_cpu(agf->agf_length),
+		__entry->bno_root = be32_to_cpu(agf->agf_roots[XFS_BTNUM_BNO]),
+		__entry->cnt_root = be32_to_cpu(agf->agf_roots[XFS_BTNUM_CNT]),
+		__entry->bno_level =
+				be32_to_cpu(agf->agf_levels[XFS_BTNUM_BNO]),
+		__entry->cnt_level =
+				be32_to_cpu(agf->agf_levels[XFS_BTNUM_CNT]),
+		__entry->flfirst = be32_to_cpu(agf->agf_flfirst),
+		__entry->fllast = be32_to_cpu(agf->agf_fllast),
+		__entry->flcount = be32_to_cpu(agf->agf_flcount),
+		__entry->freeblks = be32_to_cpu(agf->agf_freeblks),
+		__entry->longest = be32_to_cpu(agf->agf_longest);
+		__entry->caller_ip = caller_ip;
+	),
+	TP_printk("dev %d:%d agno %u flags %s length %u roots b %u c %u "
+		  "levels b %u c %u flfirst %u fllast %u flcount %u "
+		  "freeblks %u longest %u caller %pf",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->agno,
+		  __print_flags(__entry->flags, "|", XFS_AGF_FLAGS),
+		  __entry->length,
+		  __entry->bno_root,
+		  __entry->cnt_root,
+		  __entry->bno_level,
+		  __entry->cnt_level,
+		  __entry->flfirst,
+		  __entry->fllast,
+		  __entry->flcount,
+		  __entry->freeblks,
+		  __entry->longest,
+		  (void *)__entry->caller_ip)
+);
+
+TRACE_EVENT(xfs_free_extent,
+	TP_PROTO(struct xfs_mount *mp, xfs_agnumber_t agno, xfs_agblock_t agbno,
+		 xfs_extlen_t len, bool isfl, int haveleft, int haveright),
+	TP_ARGS(mp, agno, agbno, len, isfl, haveleft, haveright),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_agnumber_t, agno)
+		__field(xfs_agblock_t, agbno)
+		__field(xfs_extlen_t, len)
+		__field(int, isfl)
+		__field(int, haveleft)
+		__field(int, haveright)
+	),
+	TP_fast_assign(
+		__entry->dev = mp->m_super->s_dev;
+		__entry->agno = agno;
+		__entry->agbno = agbno;
+		__entry->len = len;
+		__entry->isfl = isfl;
+		__entry->haveleft = haveleft;
+		__entry->haveright = haveright;
+	),
+	TP_printk("dev %d:%d agno %u agbno %u len %u isfl %d %s",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->agno,
+		  __entry->agbno,
+		  __entry->len,
+		  __entry->isfl,
+		  __entry->haveleft ?
+			(__entry->haveright ? "both" : "left") :
+			(__entry->haveright ? "right" : "none"))
+
+);
+
+DECLARE_EVENT_CLASS(xfs_alloc_class,
+	TP_PROTO(struct xfs_alloc_arg *args),
+	TP_ARGS(args),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_agnumber_t, agno)
+		__field(xfs_agblock_t, agbno)
+		__field(xfs_extlen_t, minlen)
+		__field(xfs_extlen_t, maxlen)
+		__field(xfs_extlen_t, mod)
+		__field(xfs_extlen_t, prod)
+		__field(xfs_extlen_t, minleft)
+		__field(xfs_extlen_t, total)
+		__field(xfs_extlen_t, alignment)
+		__field(xfs_extlen_t, minalignslop)
+		__field(xfs_extlen_t, len)
+		__field(short, type)
+		__field(short, otype)
+		__field(char, wasdel)
+		__field(char, wasfromfl)
+		__field(char, isfl)
+		__field(char, userdata)
+		__field(xfs_fsblock_t, firstblock)
+	),
+	TP_fast_assign(
+		__entry->dev = args->mp->m_super->s_dev;
+		__entry->agno = args->agno;
+		__entry->agbno = args->agbno;
+		__entry->minlen = args->minlen;
+		__entry->maxlen = args->maxlen;
+		__entry->mod = args->mod;
+		__entry->prod = args->prod;
+		__entry->minleft = args->minleft;
+		__entry->total = args->total;
+		__entry->alignment = args->alignment;
+		__entry->minalignslop = args->minalignslop;
+		__entry->len = args->len;
+		__entry->type = args->type;
+		__entry->otype = args->otype;
+		__entry->wasdel = args->wasdel;
+		__entry->wasfromfl = args->wasfromfl;
+		__entry->isfl = args->isfl;
+		__entry->userdata = args->userdata;
+		__entry->firstblock = args->firstblock;
+	),
+	TP_printk("dev %d:%d agno %u agbno %u minlen %u maxlen %u mod %u "
+		  "prod %u minleft %u total %u alignment %u minalignslop %u "
+		  "len %u type %s otype %s wasdel %d wasfromfl %d isfl %d "
+		  "userdata %d firstblock 0x%llx",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->agno,
+		  __entry->agbno,
+		  __entry->minlen,
+		  __entry->maxlen,
+		  __entry->mod,
+		  __entry->prod,
+		  __entry->minleft,
+		  __entry->total,
+		  __entry->alignment,
+		  __entry->minalignslop,
+		  __entry->len,
+		  __print_symbolic(__entry->type, XFS_ALLOC_TYPES),
+		  __print_symbolic(__entry->otype, XFS_ALLOC_TYPES),
+		  __entry->wasdel,
+		  __entry->wasfromfl,
+		  __entry->isfl,
+		  __entry->userdata,
+		  (unsigned long long)__entry->firstblock)
+)
+
+#define DEFINE_ALLOC_EVENT(name) \
+DEFINE_EVENT(xfs_alloc_class, name, \
+	TP_PROTO(struct xfs_alloc_arg *args), \
+	TP_ARGS(args))
+DEFINE_ALLOC_EVENT(xfs_alloc_exact_done);
+DEFINE_ALLOC_EVENT(xfs_alloc_exact_notfound);
+DEFINE_ALLOC_EVENT(xfs_alloc_exact_error);
+DEFINE_ALLOC_EVENT(xfs_alloc_near_nominleft);
+DEFINE_ALLOC_EVENT(xfs_alloc_near_first);
+DEFINE_ALLOC_EVENT(xfs_alloc_near_greater);
+DEFINE_ALLOC_EVENT(xfs_alloc_near_lesser);
+DEFINE_ALLOC_EVENT(xfs_alloc_near_error);
+DEFINE_ALLOC_EVENT(xfs_alloc_near_noentry);
+DEFINE_ALLOC_EVENT(xfs_alloc_near_busy);
+DEFINE_ALLOC_EVENT(xfs_alloc_size_neither);
+DEFINE_ALLOC_EVENT(xfs_alloc_size_noentry);
+DEFINE_ALLOC_EVENT(xfs_alloc_size_nominleft);
+DEFINE_ALLOC_EVENT(xfs_alloc_size_done);
+DEFINE_ALLOC_EVENT(xfs_alloc_size_error);
+DEFINE_ALLOC_EVENT(xfs_alloc_size_busy);
+DEFINE_ALLOC_EVENT(xfs_alloc_small_freelist);
+DEFINE_ALLOC_EVENT(xfs_alloc_small_notenough);
+DEFINE_ALLOC_EVENT(xfs_alloc_small_done);
+DEFINE_ALLOC_EVENT(xfs_alloc_small_error);
+DEFINE_ALLOC_EVENT(xfs_alloc_vextent_badargs);
+DEFINE_ALLOC_EVENT(xfs_alloc_vextent_nofix);
+DEFINE_ALLOC_EVENT(xfs_alloc_vextent_noagbp);
+DEFINE_ALLOC_EVENT(xfs_alloc_vextent_loopfailed);
+DEFINE_ALLOC_EVENT(xfs_alloc_vextent_allfailed);
+
+DECLARE_EVENT_CLASS(xfs_dir2_class,
+	TP_PROTO(struct xfs_da_args *args),
+	TP_ARGS(args),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_ino_t, ino)
+		__dynamic_array(char, name, args->namelen)
+		__field(int, namelen)
+		__field(xfs_dahash_t, hashval)
+		__field(xfs_ino_t, inumber)
+		__field(int, op_flags)
+	),
+	TP_fast_assign(
+		__entry->dev = VFS_I(args->dp)->i_sb->s_dev;
+		__entry->ino = args->dp->i_ino;
+		if (args->namelen)
+			memcpy(__get_str(name), args->name, args->namelen);
+		__entry->namelen = args->namelen;
+		__entry->hashval = args->hashval;
+		__entry->inumber = args->inumber;
+		__entry->op_flags = args->op_flags;
+	),
+	TP_printk("dev %d:%d ino 0x%llx name %.*s namelen %d hashval 0x%x "
+		  "inumber 0x%llx op_flags %s",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->ino,
+		  __entry->namelen,
+		  __entry->namelen ? __get_str(name) : NULL,
+		  __entry->namelen,
+		  __entry->hashval,
+		  __entry->inumber,
+		  __print_flags(__entry->op_flags, "|", XFS_DA_OP_FLAGS))
+)
+
+#define DEFINE_DIR2_EVENT(name) \
+DEFINE_EVENT(xfs_dir2_class, name, \
+	TP_PROTO(struct xfs_da_args *args), \
+	TP_ARGS(args))
+DEFINE_DIR2_EVENT(xfs_dir2_sf_addname);
+DEFINE_DIR2_EVENT(xfs_dir2_sf_create);
+DEFINE_DIR2_EVENT(xfs_dir2_sf_lookup);
+DEFINE_DIR2_EVENT(xfs_dir2_sf_replace);
+DEFINE_DIR2_EVENT(xfs_dir2_sf_removename);
+DEFINE_DIR2_EVENT(xfs_dir2_sf_toino4);
+DEFINE_DIR2_EVENT(xfs_dir2_sf_toino8);
+DEFINE_DIR2_EVENT(xfs_dir2_sf_to_block);
+DEFINE_DIR2_EVENT(xfs_dir2_block_addname);
+DEFINE_DIR2_EVENT(xfs_dir2_block_lookup);
+DEFINE_DIR2_EVENT(xfs_dir2_block_replace);
+DEFINE_DIR2_EVENT(xfs_dir2_block_removename);
+DEFINE_DIR2_EVENT(xfs_dir2_block_to_sf);
+DEFINE_DIR2_EVENT(xfs_dir2_block_to_leaf);
+DEFINE_DIR2_EVENT(xfs_dir2_leaf_addname);
+DEFINE_DIR2_EVENT(xfs_dir2_leaf_lookup);
+DEFINE_DIR2_EVENT(xfs_dir2_leaf_replace);
+DEFINE_DIR2_EVENT(xfs_dir2_leaf_removename);
+DEFINE_DIR2_EVENT(xfs_dir2_leaf_to_block);
+DEFINE_DIR2_EVENT(xfs_dir2_leaf_to_node);
+DEFINE_DIR2_EVENT(xfs_dir2_node_addname);
+DEFINE_DIR2_EVENT(xfs_dir2_node_lookup);
+DEFINE_DIR2_EVENT(xfs_dir2_node_replace);
+DEFINE_DIR2_EVENT(xfs_dir2_node_removename);
+DEFINE_DIR2_EVENT(xfs_dir2_node_to_leaf);
+
+DECLARE_EVENT_CLASS(xfs_dir2_space_class,
+	TP_PROTO(struct xfs_da_args *args, int idx),
+	TP_ARGS(args, idx),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_ino_t, ino)
+		__field(int, op_flags)
+		__field(int, idx)
+	),
+	TP_fast_assign(
+		__entry->dev = VFS_I(args->dp)->i_sb->s_dev;
+		__entry->ino = args->dp->i_ino;
+		__entry->op_flags = args->op_flags;
+		__entry->idx = idx;
+	),
+	TP_printk("dev %d:%d ino 0x%llx op_flags %s index %d",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->ino,
+		  __print_flags(__entry->op_flags, "|", XFS_DA_OP_FLAGS),
+		  __entry->idx)
+)
+
+#define DEFINE_DIR2_SPACE_EVENT(name) \
+DEFINE_EVENT(xfs_dir2_space_class, name, \
+	TP_PROTO(struct xfs_da_args *args, int idx), \
+	TP_ARGS(args, idx))
+DEFINE_DIR2_SPACE_EVENT(xfs_dir2_leafn_add);
+DEFINE_DIR2_SPACE_EVENT(xfs_dir2_leafn_remove);
+DEFINE_DIR2_SPACE_EVENT(xfs_dir2_grow_inode);
+DEFINE_DIR2_SPACE_EVENT(xfs_dir2_shrink_inode);
+
+TRACE_EVENT(xfs_dir2_leafn_moveents,
+	TP_PROTO(struct xfs_da_args *args, int src_idx, int dst_idx, int count),
+	TP_ARGS(args, src_idx, dst_idx, count),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_ino_t, ino)
+		__field(int, op_flags)
+		__field(int, src_idx)
+		__field(int, dst_idx)
+		__field(int, count)
+	),
+	TP_fast_assign(
+		__entry->dev = VFS_I(args->dp)->i_sb->s_dev;
+		__entry->ino = args->dp->i_ino;
+		__entry->op_flags = args->op_flags;
+		__entry->src_idx = src_idx;
+		__entry->dst_idx = dst_idx;
+		__entry->count = count;
+	),
+	TP_printk("dev %d:%d ino 0x%llx op_flags %s "
+		  "src_idx %d dst_idx %d count %d",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->ino,
+		  __print_flags(__entry->op_flags, "|", XFS_DA_OP_FLAGS),
+		  __entry->src_idx,
+		  __entry->dst_idx,
+		  __entry->count)
+);
+
+#define XFS_SWAPEXT_INODES \
+	{ 0,	"target" }, \
+	{ 1,	"temp" }
+
+#define XFS_INODE_FORMAT_STR \
+	{ 0,	"invalid" }, \
+	{ 1,	"local" }, \
+	{ 2,	"extent" }, \
+	{ 3,	"btree" }
+
+DECLARE_EVENT_CLASS(xfs_swap_extent_class,
+	TP_PROTO(struct xfs_inode *ip, int which),
+	TP_ARGS(ip, which),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(int, which)
+		__field(xfs_ino_t, ino)
+		__field(int, format)
+		__field(int, nex)
+		__field(int, max_nex)
+		__field(int, broot_size)
+		__field(int, fork_off)
+	),
+	TP_fast_assign(
+		__entry->dev = VFS_I(ip)->i_sb->s_dev;
+		__entry->which = which;
+		__entry->ino = ip->i_ino;
+		__entry->format = ip->i_d.di_format;
+		__entry->nex = ip->i_d.di_nextents;
+		__entry->max_nex = ip->i_df.if_ext_max;
+		__entry->broot_size = ip->i_df.if_broot_bytes;
+		__entry->fork_off = XFS_IFORK_BOFF(ip);
+	),
+	TP_printk("dev %d:%d ino 0x%llx (%s), %s format, num_extents %d, "
+		  "Max in-fork extents %d, broot size %d, fork offset %d",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->ino,
+		  __print_symbolic(__entry->which, XFS_SWAPEXT_INODES),
+		  __print_symbolic(__entry->format, XFS_INODE_FORMAT_STR),
+		  __entry->nex,
+		  __entry->max_nex,
+		  __entry->broot_size,
+		  __entry->fork_off)
+)
+
+#define DEFINE_SWAPEXT_EVENT(name) \
+DEFINE_EVENT(xfs_swap_extent_class, name, \
+	TP_PROTO(struct xfs_inode *ip, int which), \
+	TP_ARGS(ip, which))
+
+DEFINE_SWAPEXT_EVENT(xfs_swap_extent_before);
+DEFINE_SWAPEXT_EVENT(xfs_swap_extent_after);
+
+DECLARE_EVENT_CLASS(xfs_log_recover_item_class,
+	TP_PROTO(struct log *log, struct xlog_recover *trans,
+		struct xlog_recover_item *item, int pass),
+	TP_ARGS(log, trans, item, pass),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(unsigned long, item)
+		__field(xlog_tid_t, tid)
+		__field(int, type)
+		__field(int, pass)
+		__field(int, count)
+		__field(int, total)
+	),
+	TP_fast_assign(
+		__entry->dev = log->l_mp->m_super->s_dev;
+		__entry->item = (unsigned long)item;
+		__entry->tid = trans->r_log_tid;
+		__entry->type = ITEM_TYPE(item);
+		__entry->pass = pass;
+		__entry->count = item->ri_cnt;
+		__entry->total = item->ri_total;
+	),
+	TP_printk("dev %d:%d trans 0x%x, pass %d, item 0x%p, item type %s "
+		  "item region count/total %d/%d",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->tid,
+		  __entry->pass,
+		  (void *)__entry->item,
+		  __print_symbolic(__entry->type, XFS_LI_TYPE_DESC),
+		  __entry->count,
+		  __entry->total)
+)
+
+#define DEFINE_LOG_RECOVER_ITEM(name) \
+DEFINE_EVENT(xfs_log_recover_item_class, name, \
+	TP_PROTO(struct log *log, struct xlog_recover *trans, \
+		struct xlog_recover_item *item, int pass), \
+	TP_ARGS(log, trans, item, pass))
+
+DEFINE_LOG_RECOVER_ITEM(xfs_log_recover_item_add);
+DEFINE_LOG_RECOVER_ITEM(xfs_log_recover_item_add_cont);
+DEFINE_LOG_RECOVER_ITEM(xfs_log_recover_item_reorder_head);
+DEFINE_LOG_RECOVER_ITEM(xfs_log_recover_item_reorder_tail);
+DEFINE_LOG_RECOVER_ITEM(xfs_log_recover_item_recover);
+
+DECLARE_EVENT_CLASS(xfs_log_recover_buf_item_class,
+	TP_PROTO(struct log *log, struct xfs_buf_log_format *buf_f),
+	TP_ARGS(log, buf_f),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(__int64_t, blkno)
+		__field(unsigned short, len)
+		__field(unsigned short, flags)
+		__field(unsigned short, size)
+		__field(unsigned int, map_size)
+	),
+	TP_fast_assign(
+		__entry->dev = log->l_mp->m_super->s_dev;
+		__entry->blkno = buf_f->blf_blkno;
+		__entry->len = buf_f->blf_len;
+		__entry->flags = buf_f->blf_flags;
+		__entry->size = buf_f->blf_size;
+		__entry->map_size = buf_f->blf_map_size;
+	),
+	TP_printk("dev %d:%d blkno 0x%llx, len %u, flags 0x%x, size %d, "
+			"map_size %d",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->blkno,
+		  __entry->len,
+		  __entry->flags,
+		  __entry->size,
+		  __entry->map_size)
+)
+
+#define DEFINE_LOG_RECOVER_BUF_ITEM(name) \
+DEFINE_EVENT(xfs_log_recover_buf_item_class, name, \
+	TP_PROTO(struct log *log, struct xfs_buf_log_format *buf_f), \
+	TP_ARGS(log, buf_f))
+
+DEFINE_LOG_RECOVER_BUF_ITEM(xfs_log_recover_buf_not_cancel);
+DEFINE_LOG_RECOVER_BUF_ITEM(xfs_log_recover_buf_cancel);
+DEFINE_LOG_RECOVER_BUF_ITEM(xfs_log_recover_buf_cancel_add);
+DEFINE_LOG_RECOVER_BUF_ITEM(xfs_log_recover_buf_cancel_ref_inc);
+DEFINE_LOG_RECOVER_BUF_ITEM(xfs_log_recover_buf_recover);
+DEFINE_LOG_RECOVER_BUF_ITEM(xfs_log_recover_buf_inode_buf);
+DEFINE_LOG_RECOVER_BUF_ITEM(xfs_log_recover_buf_reg_buf);
+DEFINE_LOG_RECOVER_BUF_ITEM(xfs_log_recover_buf_dquot_buf);
+
+DECLARE_EVENT_CLASS(xfs_log_recover_ino_item_class,
+	TP_PROTO(struct log *log, struct xfs_inode_log_format *in_f),
+	TP_ARGS(log, in_f),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_ino_t, ino)
+		__field(unsigned short, size)
+		__field(int, fields)
+		__field(unsigned short, asize)
+		__field(unsigned short, dsize)
+		__field(__int64_t, blkno)
+		__field(int, len)
+		__field(int, boffset)
+	),
+	TP_fast_assign(
+		__entry->dev = log->l_mp->m_super->s_dev;
+		__entry->ino = in_f->ilf_ino;
+		__entry->size = in_f->ilf_size;
+		__entry->fields = in_f->ilf_fields;
+		__entry->asize = in_f->ilf_asize;
+		__entry->dsize = in_f->ilf_dsize;
+		__entry->blkno = in_f->ilf_blkno;
+		__entry->len = in_f->ilf_len;
+		__entry->boffset = in_f->ilf_boffset;
+	),
+	TP_printk("dev %d:%d ino 0x%llx, size %u, fields 0x%x, asize %d, "
+			"dsize %d, blkno 0x%llx, len %d, boffset %d",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->ino,
+		  __entry->size,
+		  __entry->fields,
+		  __entry->asize,
+		  __entry->dsize,
+		  __entry->blkno,
+		  __entry->len,
+		  __entry->boffset)
+)
+#define DEFINE_LOG_RECOVER_INO_ITEM(name) \
+DEFINE_EVENT(xfs_log_recover_ino_item_class, name, \
+	TP_PROTO(struct log *log, struct xfs_inode_log_format *in_f), \
+	TP_ARGS(log, in_f))
+
+DEFINE_LOG_RECOVER_INO_ITEM(xfs_log_recover_inode_recover);
+DEFINE_LOG_RECOVER_INO_ITEM(xfs_log_recover_inode_cancel);
+DEFINE_LOG_RECOVER_INO_ITEM(xfs_log_recover_inode_skip);
+
+DECLARE_EVENT_CLASS(xfs_discard_class,
+	TP_PROTO(struct xfs_mount *mp, xfs_agnumber_t agno,
+		 xfs_agblock_t agbno, xfs_extlen_t len),
+	TP_ARGS(mp, agno, agbno, len),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_agnumber_t, agno)
+		__field(xfs_agblock_t, agbno)
+		__field(xfs_extlen_t, len)
+	),
+	TP_fast_assign(
+		__entry->dev = mp->m_super->s_dev;
+		__entry->agno = agno;
+		__entry->agbno = agbno;
+		__entry->len = len;
+	),
+	TP_printk("dev %d:%d agno %u agbno %u len %u\n",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->agno,
+		  __entry->agbno,
+		  __entry->len)
+)
+
+#define DEFINE_DISCARD_EVENT(name) \
+DEFINE_EVENT(xfs_discard_class, name, \
+	TP_PROTO(struct xfs_mount *mp, xfs_agnumber_t agno, \
+		 xfs_agblock_t agbno, xfs_extlen_t len), \
+	TP_ARGS(mp, agno, agbno, len))
+DEFINE_DISCARD_EVENT(xfs_discard_extent);
+DEFINE_DISCARD_EVENT(xfs_discard_toosmall);
+DEFINE_DISCARD_EVENT(xfs_discard_exclude);
+DEFINE_DISCARD_EVENT(xfs_discard_busy);
+
+#endif /* _TRACE_XFS_H */
+
+#undef TRACE_INCLUDE_PATH
+#define TRACE_INCLUDE_PATH .
+#define TRACE_INCLUDE_FILE xfs_trace
+#include <trace/define_trace.h>
diff --git a/fs/xfs/linux-2.6/xfs_vnode.h b/fs/xfs/linux-2.6/xfs_vnode.h
new file mode 100644
index 00000000..7c220b42
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_vnode.h
@@ -0,0 +1,64 @@
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_VNODE_H__
+#define __XFS_VNODE_H__
+
+#include "xfs_fs.h"
+
+struct file;
+struct xfs_inode;
+struct xfs_iomap;
+struct attrlist_cursor_kern;
+
+/*
+ * Return values for xfs_inactive.  A return value of
+ * VN_INACTIVE_NOCACHE implies that the file system behavior
+ * has disassociated its state and bhv_desc_t from the vnode.
+ */
+#define	VN_INACTIVE_CACHE	0
+#define	VN_INACTIVE_NOCACHE	1
+
+/*
+ * Flags for read/write calls - same values as IRIX
+ */
+#define IO_ISDIRECT	0x00004		/* bypass page cache */
+#define IO_INVIS	0x00020		/* don't update inode timestamps */
+
+#define XFS_IO_FLAGS \
+	{ IO_ISDIRECT,	"DIRECT" }, \
+	{ IO_INVIS,	"INVIS"}
+
+/*
+ * Flush/Invalidate options for vop_toss/flush/flushinval_pages.
+ */
+#define FI_NONE			0	/* none */
+#define FI_REMAPF		1	/* Do a remapf prior to the operation */
+#define FI_REMAPF_LOCKED	2	/* Do a remapf prior to the operation.
+					   Prevent VM access to the pages until
+					   the operation completes. */
+
+/*
+ * Some useful predicates.
+ */
+#define VN_MAPPED(vp)	mapping_mapped(vp->i_mapping)
+#define VN_CACHED(vp)	(vp->i_mapping->nrpages)
+#define VN_DIRTY(vp)	mapping_tagged(vp->i_mapping, \
+					PAGECACHE_TAG_DIRTY)
+
+
+#endif	/* __XFS_VNODE_H__ */
diff --git a/fs/xfs/linux-2.6/xfs_xattr.c b/fs/xfs/linux-2.6/xfs_xattr.c
new file mode 100644
index 00000000..87d3e038
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_xattr.c
@@ -0,0 +1,241 @@
+/*
+ * Copyright (C) 2008 Christoph Hellwig.
+ * Portions Copyright (C) 2000-2008 Silicon Graphics, Inc.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+
+#include "xfs.h"
+#include "xfs_da_btree.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_inode.h"
+#include "xfs_attr.h"
+#include "xfs_attr_leaf.h"
+#include "xfs_acl.h"
+#include "xfs_vnodeops.h"
+
+#include <linux/posix_acl_xattr.h>
+#include <linux/xattr.h>
+
+
+static int
+xfs_xattr_get(struct dentry *dentry, const char *name,
+		void *value, size_t size, int xflags)
+{
+	struct xfs_inode *ip = XFS_I(dentry->d_inode);
+	int error, asize = size;
+
+	if (strcmp(name, "") == 0)
+		return -EINVAL;
+
+	/* Convert Linux syscall to XFS internal ATTR flags */
+	if (!size) {
+		xflags |= ATTR_KERNOVAL;
+		value = NULL;
+	}
+
+	error = -xfs_attr_get(ip, (unsigned char *)name, value, &asize, xflags);
+	if (error)
+		return error;
+	return asize;
+}
+
+static int
+xfs_xattr_set(struct dentry *dentry, const char *name, const void *value,
+		size_t size, int flags, int xflags)
+{
+	struct xfs_inode *ip = XFS_I(dentry->d_inode);
+
+	if (strcmp(name, "") == 0)
+		return -EINVAL;
+
+	/* Convert Linux syscall to XFS internal ATTR flags */
+	if (flags & XATTR_CREATE)
+		xflags |= ATTR_CREATE;
+	if (flags & XATTR_REPLACE)
+		xflags |= ATTR_REPLACE;
+
+	if (!value)
+		return -xfs_attr_remove(ip, (unsigned char *)name, xflags);
+	return -xfs_attr_set(ip, (unsigned char *)name,
+				(void *)value, size, xflags);
+}
+
+static const struct xattr_handler xfs_xattr_user_handler = {
+	.prefix	= XATTR_USER_PREFIX,
+	.flags	= 0, /* no flags implies user namespace */
+	.get	= xfs_xattr_get,
+	.set	= xfs_xattr_set,
+};
+
+static const struct xattr_handler xfs_xattr_trusted_handler = {
+	.prefix	= XATTR_TRUSTED_PREFIX,
+	.flags	= ATTR_ROOT,
+	.get	= xfs_xattr_get,
+	.set	= xfs_xattr_set,
+};
+
+static const struct xattr_handler xfs_xattr_security_handler = {
+	.prefix	= XATTR_SECURITY_PREFIX,
+	.flags	= ATTR_SECURE,
+	.get	= xfs_xattr_get,
+	.set	= xfs_xattr_set,
+};
+
+const struct xattr_handler *xfs_xattr_handlers[] = {
+	&xfs_xattr_user_handler,
+	&xfs_xattr_trusted_handler,
+	&xfs_xattr_security_handler,
+#ifdef CONFIG_XFS_POSIX_ACL
+	&xfs_xattr_acl_access_handler,
+	&xfs_xattr_acl_default_handler,
+#endif
+	NULL
+};
+
+static unsigned int xfs_xattr_prefix_len(int flags)
+{
+	if (flags & XFS_ATTR_SECURE)
+		return sizeof("security");
+	else if (flags & XFS_ATTR_ROOT)
+		return sizeof("trusted");
+	else
+		return sizeof("user");
+}
+
+static const char *xfs_xattr_prefix(int flags)
+{
+	if (flags & XFS_ATTR_SECURE)
+		return xfs_xattr_security_handler.prefix;
+	else if (flags & XFS_ATTR_ROOT)
+		return xfs_xattr_trusted_handler.prefix;
+	else
+		return xfs_xattr_user_handler.prefix;
+}
+
+static int
+xfs_xattr_put_listent(
+	struct xfs_attr_list_context *context,
+	int		flags,
+	unsigned char	*name,
+	int		namelen,
+	int		valuelen,
+	unsigned char	*value)
+{
+	unsigned int prefix_len = xfs_xattr_prefix_len(flags);
+	char *offset;
+	int arraytop;
+
+	ASSERT(context->count >= 0);
+
+	/*
+	 * Only show root namespace entries if we are actually allowed to
+	 * see them.
+	 */
+	if ((flags & XFS_ATTR_ROOT) && !capable(CAP_SYS_ADMIN))
+		return 0;
+
+	arraytop = context->count + prefix_len + namelen + 1;
+	if (arraytop > context->firstu) {
+		context->count = -1;	/* insufficient space */
+		return 1;
+	}
+	offset = (char *)context->alist + context->count;
+	strncpy(offset, xfs_xattr_prefix(flags), prefix_len);
+	offset += prefix_len;
+	strncpy(offset, (char *)name, namelen);			/* real name */
+	offset += namelen;
+	*offset = '\0';
+	context->count += prefix_len + namelen + 1;
+	return 0;
+}
+
+static int
+xfs_xattr_put_listent_sizes(
+	struct xfs_attr_list_context *context,
+	int		flags,
+	unsigned char	*name,
+	int		namelen,
+	int		valuelen,
+	unsigned char	*value)
+{
+	context->count += xfs_xattr_prefix_len(flags) + namelen + 1;
+	return 0;
+}
+
+static int
+list_one_attr(const char *name, const size_t len, void *data,
+		size_t size, ssize_t *result)
+{
+	char *p = data + *result;
+
+	*result += len;
+	if (!size)
+		return 0;
+	if (*result > size)
+		return -ERANGE;
+
+	strcpy(p, name);
+	return 0;
+}
+
+ssize_t
+xfs_vn_listxattr(struct dentry *dentry, char *data, size_t size)
+{
+	struct xfs_attr_list_context context;
+	struct attrlist_cursor_kern cursor = { 0 };
+	struct inode		*inode = dentry->d_inode;
+	int			error;
+
+	/*
+	 * First read the regular on-disk attributes.
+	 */
+	memset(&context, 0, sizeof(context));
+	context.dp = XFS_I(inode);
+	context.cursor = &cursor;
+	context.resynch = 1;
+	context.alist = data;
+	context.bufsize = size;
+	context.firstu = context.bufsize;
+
+	if (size)
+		context.put_listent = xfs_xattr_put_listent;
+	else
+		context.put_listent = xfs_xattr_put_listent_sizes;
+
+	xfs_attr_list_int(&context);
+	if (context.count < 0)
+		return -ERANGE;
+
+	/*
+	 * Then add the two synthetic ACL attributes.
+	 */
+	if (posix_acl_access_exists(inode)) {
+		error = list_one_attr(POSIX_ACL_XATTR_ACCESS,
+				strlen(POSIX_ACL_XATTR_ACCESS) + 1,
+				data, size, &context.count);
+		if (error)
+			return error;
+	}
+
+	if (posix_acl_default_exists(inode)) {
+		error = list_one_attr(POSIX_ACL_XATTR_DEFAULT,
+				strlen(POSIX_ACL_XATTR_DEFAULT) + 1,
+				data, size, &context.count);
+		if (error)
+			return error;
+	}
+
+	return context.count;
+}
diff --git a/fs/xfs/quota/xfs_dquot.c b/fs/xfs/quota/xfs_dquot.c
new file mode 100644
index 00000000..6fa21460
--- /dev/null
+++ b/fs/xfs/quota/xfs_dquot.c
@@ -0,0 +1,1496 @@
+/*
+ * Copyright (c) 2000-2003 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_alloc.h"
+#include "xfs_quota.h"
+#include "xfs_mount.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_inode.h"
+#include "xfs_bmap.h"
+#include "xfs_rtalloc.h"
+#include "xfs_error.h"
+#include "xfs_itable.h"
+#include "xfs_attr.h"
+#include "xfs_buf_item.h"
+#include "xfs_trans_space.h"
+#include "xfs_trans_priv.h"
+#include "xfs_qm.h"
+#include "xfs_trace.h"
+
+
+/*
+   LOCK ORDER
+
+   inode lock		    (ilock)
+   dquot hash-chain lock    (hashlock)
+   xqm dquot freelist lock  (freelistlock
+   mount's dquot list lock  (mplistlock)
+   user dquot lock - lock ordering among dquots is based on the uid or gid
+   group dquot lock - similar to udquots. Between the two dquots, the udquot
+		      has to be locked first.
+   pin lock - the dquot lock must be held to take this lock.
+   flush lock - ditto.
+*/
+
+#ifdef DEBUG
+xfs_buftarg_t *xfs_dqerror_target;
+int xfs_do_dqerror;
+int xfs_dqreq_num;
+int xfs_dqerror_mod = 33;
+#endif
+
+static struct lock_class_key xfs_dquot_other_class;
+
+/*
+ * Allocate and initialize a dquot. We don't always allocate fresh memory;
+ * we try to reclaim a free dquot if the number of incore dquots are above
+ * a threshold.
+ * The only field inside the core that gets initialized at this point
+ * is the d_id field. The idea is to fill in the entire q_core
+ * when we read in the on disk dquot.
+ */
+STATIC xfs_dquot_t *
+xfs_qm_dqinit(
+	xfs_mount_t  *mp,
+	xfs_dqid_t   id,
+	uint	     type)
+{
+	xfs_dquot_t	*dqp;
+	boolean_t	brandnewdquot;
+
+	brandnewdquot = xfs_qm_dqalloc_incore(&dqp);
+	dqp->dq_flags = type;
+	dqp->q_core.d_id = cpu_to_be32(id);
+	dqp->q_mount = mp;
+
+	/*
+	 * No need to re-initialize these if this is a reclaimed dquot.
+	 */
+	if (brandnewdquot) {
+		INIT_LIST_HEAD(&dqp->q_freelist);
+		mutex_init(&dqp->q_qlock);
+		init_waitqueue_head(&dqp->q_pinwait);
+
+		/*
+		 * Because we want to use a counting completion, complete
+		 * the flush completion once to allow a single access to
+		 * the flush completion without blocking.
+		 */
+		init_completion(&dqp->q_flush);
+		complete(&dqp->q_flush);
+
+		trace_xfs_dqinit(dqp);
+	} else {
+		/*
+		 * Only the q_core portion was zeroed in dqreclaim_one().
+		 * So, we need to reset others.
+		 */
+		dqp->q_nrefs = 0;
+		dqp->q_blkno = 0;
+		INIT_LIST_HEAD(&dqp->q_mplist);
+		INIT_LIST_HEAD(&dqp->q_hashlist);
+		dqp->q_bufoffset = 0;
+		dqp->q_fileoffset = 0;
+		dqp->q_transp = NULL;
+		dqp->q_gdquot = NULL;
+		dqp->q_res_bcount = 0;
+		dqp->q_res_icount = 0;
+		dqp->q_res_rtbcount = 0;
+		atomic_set(&dqp->q_pincount, 0);
+		dqp->q_hash = NULL;
+		ASSERT(list_empty(&dqp->q_freelist));
+
+		trace_xfs_dqreuse(dqp);
+	}
+
+	/*
+	 * In either case we need to make sure group quotas have a different
+	 * lock class than user quotas, to make sure lockdep knows we can
+	 * locks of one of each at the same time.
+	 */
+	if (!(type & XFS_DQ_USER))
+		lockdep_set_class(&dqp->q_qlock, &xfs_dquot_other_class);
+
+	/*
+	 * log item gets initialized later
+	 */
+	return (dqp);
+}
+
+/*
+ * This is called to free all the memory associated with a dquot
+ */
+void
+xfs_qm_dqdestroy(
+	xfs_dquot_t	*dqp)
+{
+	ASSERT(list_empty(&dqp->q_freelist));
+
+	mutex_destroy(&dqp->q_qlock);
+	kmem_zone_free(xfs_Gqm->qm_dqzone, dqp);
+
+	atomic_dec(&xfs_Gqm->qm_totaldquots);
+}
+
+/*
+ * This is what a 'fresh' dquot inside a dquot chunk looks like on disk.
+ */
+STATIC void
+xfs_qm_dqinit_core(
+	xfs_dqid_t	id,
+	uint		type,
+	xfs_dqblk_t	*d)
+{
+	/*
+	 * Caller has zero'd the entire dquot 'chunk' already.
+	 */
+	d->dd_diskdq.d_magic = cpu_to_be16(XFS_DQUOT_MAGIC);
+	d->dd_diskdq.d_version = XFS_DQUOT_VERSION;
+	d->dd_diskdq.d_id = cpu_to_be32(id);
+	d->dd_diskdq.d_flags = type;
+}
+
+/*
+ * If default limits are in force, push them into the dquot now.
+ * We overwrite the dquot limits only if they are zero and this
+ * is not the root dquot.
+ */
+void
+xfs_qm_adjust_dqlimits(
+	xfs_mount_t		*mp,
+	xfs_disk_dquot_t	*d)
+{
+	xfs_quotainfo_t		*q = mp->m_quotainfo;
+
+	ASSERT(d->d_id);
+
+	if (q->qi_bsoftlimit && !d->d_blk_softlimit)
+		d->d_blk_softlimit = cpu_to_be64(q->qi_bsoftlimit);
+	if (q->qi_bhardlimit && !d->d_blk_hardlimit)
+		d->d_blk_hardlimit = cpu_to_be64(q->qi_bhardlimit);
+	if (q->qi_isoftlimit && !d->d_ino_softlimit)
+		d->d_ino_softlimit = cpu_to_be64(q->qi_isoftlimit);
+	if (q->qi_ihardlimit && !d->d_ino_hardlimit)
+		d->d_ino_hardlimit = cpu_to_be64(q->qi_ihardlimit);
+	if (q->qi_rtbsoftlimit && !d->d_rtb_softlimit)
+		d->d_rtb_softlimit = cpu_to_be64(q->qi_rtbsoftlimit);
+	if (q->qi_rtbhardlimit && !d->d_rtb_hardlimit)
+		d->d_rtb_hardlimit = cpu_to_be64(q->qi_rtbhardlimit);
+}
+
+/*
+ * Check the limits and timers of a dquot and start or reset timers
+ * if necessary.
+ * This gets called even when quota enforcement is OFF, which makes our
+ * life a little less complicated. (We just don't reject any quota
+ * reservations in that case, when enforcement is off).
+ * We also return 0 as the values of the timers in Q_GETQUOTA calls, when
+ * enforcement's off.
+ * In contrast, warnings are a little different in that they don't
+ * 'automatically' get started when limits get exceeded.  They do
+ * get reset to zero, however, when we find the count to be under
+ * the soft limit (they are only ever set non-zero via userspace).
+ */
+void
+xfs_qm_adjust_dqtimers(
+	xfs_mount_t		*mp,
+	xfs_disk_dquot_t	*d)
+{
+	ASSERT(d->d_id);
+
+#ifdef QUOTADEBUG
+	if (d->d_blk_hardlimit)
+		ASSERT(be64_to_cpu(d->d_blk_softlimit) <=
+		       be64_to_cpu(d->d_blk_hardlimit));
+	if (d->d_ino_hardlimit)
+		ASSERT(be64_to_cpu(d->d_ino_softlimit) <=
+		       be64_to_cpu(d->d_ino_hardlimit));
+	if (d->d_rtb_hardlimit)
+		ASSERT(be64_to_cpu(d->d_rtb_softlimit) <=
+		       be64_to_cpu(d->d_rtb_hardlimit));
+#endif
+	if (!d->d_btimer) {
+		if ((d->d_blk_softlimit &&
+		     (be64_to_cpu(d->d_bcount) >=
+		      be64_to_cpu(d->d_blk_softlimit))) ||
+		    (d->d_blk_hardlimit &&
+		     (be64_to_cpu(d->d_bcount) >=
+		      be64_to_cpu(d->d_blk_hardlimit)))) {
+			d->d_btimer = cpu_to_be32(get_seconds() +
+					mp->m_quotainfo->qi_btimelimit);
+		} else {
+			d->d_bwarns = 0;
+		}
+	} else {
+		if ((!d->d_blk_softlimit ||
+		     (be64_to_cpu(d->d_bcount) <
+		      be64_to_cpu(d->d_blk_softlimit))) &&
+		    (!d->d_blk_hardlimit ||
+		    (be64_to_cpu(d->d_bcount) <
+		     be64_to_cpu(d->d_blk_hardlimit)))) {
+			d->d_btimer = 0;
+		}
+	}
+
+	if (!d->d_itimer) {
+		if ((d->d_ino_softlimit &&
+		     (be64_to_cpu(d->d_icount) >=
+		      be64_to_cpu(d->d_ino_softlimit))) ||
+		    (d->d_ino_hardlimit &&
+		     (be64_to_cpu(d->d_icount) >=
+		      be64_to_cpu(d->d_ino_hardlimit)))) {
+			d->d_itimer = cpu_to_be32(get_seconds() +
+					mp->m_quotainfo->qi_itimelimit);
+		} else {
+			d->d_iwarns = 0;
+		}
+	} else {
+		if ((!d->d_ino_softlimit ||
+		     (be64_to_cpu(d->d_icount) <
+		      be64_to_cpu(d->d_ino_softlimit)))  &&
+		    (!d->d_ino_hardlimit ||
+		     (be64_to_cpu(d->d_icount) <
+		      be64_to_cpu(d->d_ino_hardlimit)))) {
+			d->d_itimer = 0;
+		}
+	}
+
+	if (!d->d_rtbtimer) {
+		if ((d->d_rtb_softlimit &&
+		     (be64_to_cpu(d->d_rtbcount) >=
+		      be64_to_cpu(d->d_rtb_softlimit))) ||
+		    (d->d_rtb_hardlimit &&
+		     (be64_to_cpu(d->d_rtbcount) >=
+		      be64_to_cpu(d->d_rtb_hardlimit)))) {
+			d->d_rtbtimer = cpu_to_be32(get_seconds() +
+					mp->m_quotainfo->qi_rtbtimelimit);
+		} else {
+			d->d_rtbwarns = 0;
+		}
+	} else {
+		if ((!d->d_rtb_softlimit ||
+		     (be64_to_cpu(d->d_rtbcount) <
+		      be64_to_cpu(d->d_rtb_softlimit))) &&
+		    (!d->d_rtb_hardlimit ||
+		     (be64_to_cpu(d->d_rtbcount) <
+		      be64_to_cpu(d->d_rtb_hardlimit)))) {
+			d->d_rtbtimer = 0;
+		}
+	}
+}
+
+/*
+ * initialize a buffer full of dquots and log the whole thing
+ */
+STATIC void
+xfs_qm_init_dquot_blk(
+	xfs_trans_t	*tp,
+	xfs_mount_t	*mp,
+	xfs_dqid_t	id,
+	uint		type,
+	xfs_buf_t	*bp)
+{
+	struct xfs_quotainfo	*q = mp->m_quotainfo;
+	xfs_dqblk_t	*d;
+	int		curid, i;
+
+	ASSERT(tp);
+	ASSERT(XFS_BUF_ISBUSY(bp));
+	ASSERT(XFS_BUF_VALUSEMA(bp) <= 0);
+
+	d = (xfs_dqblk_t *)XFS_BUF_PTR(bp);
+
+	/*
+	 * ID of the first dquot in the block - id's are zero based.
+	 */
+	curid = id - (id % q->qi_dqperchunk);
+	ASSERT(curid >= 0);
+	memset(d, 0, BBTOB(q->qi_dqchunklen));
+	for (i = 0; i < q->qi_dqperchunk; i++, d++, curid++)
+		xfs_qm_dqinit_core(curid, type, d);
+	xfs_trans_dquot_buf(tp, bp,
+			    (type & XFS_DQ_USER ? XFS_BLF_UDQUOT_BUF :
+			    ((type & XFS_DQ_PROJ) ? XFS_BLF_PDQUOT_BUF :
+			     XFS_BLF_GDQUOT_BUF)));
+	xfs_trans_log_buf(tp, bp, 0, BBTOB(q->qi_dqchunklen) - 1);
+}
+
+
+
+/*
+ * Allocate a block and fill it with dquots.
+ * This is called when the bmapi finds a hole.
+ */
+STATIC int
+xfs_qm_dqalloc(
+	xfs_trans_t	**tpp,
+	xfs_mount_t	*mp,
+	xfs_dquot_t	*dqp,
+	xfs_inode_t	*quotip,
+	xfs_fileoff_t	offset_fsb,
+	xfs_buf_t	**O_bpp)
+{
+	xfs_fsblock_t	firstblock;
+	xfs_bmap_free_t flist;
+	xfs_bmbt_irec_t map;
+	int		nmaps, error, committed;
+	xfs_buf_t	*bp;
+	xfs_trans_t	*tp = *tpp;
+
+	ASSERT(tp != NULL);
+
+	trace_xfs_dqalloc(dqp);
+
+	/*
+	 * Initialize the bmap freelist prior to calling bmapi code.
+	 */
+	xfs_bmap_init(&flist, &firstblock);
+	xfs_ilock(quotip, XFS_ILOCK_EXCL);
+	/*
+	 * Return if this type of quotas is turned off while we didn't
+	 * have an inode lock
+	 */
+	if (XFS_IS_THIS_QUOTA_OFF(dqp)) {
+		xfs_iunlock(quotip, XFS_ILOCK_EXCL);
+		return (ESRCH);
+	}
+
+	xfs_trans_ijoin_ref(tp, quotip, XFS_ILOCK_EXCL);
+	nmaps = 1;
+	if ((error = xfs_bmapi(tp, quotip,
+			      offset_fsb, XFS_DQUOT_CLUSTER_SIZE_FSB,
+			      XFS_BMAPI_METADATA | XFS_BMAPI_WRITE,
+			      &firstblock,
+			      XFS_QM_DQALLOC_SPACE_RES(mp),
+			      &map, &nmaps, &flist))) {
+		goto error0;
+	}
+	ASSERT(map.br_blockcount == XFS_DQUOT_CLUSTER_SIZE_FSB);
+	ASSERT(nmaps == 1);
+	ASSERT((map.br_startblock != DELAYSTARTBLOCK) &&
+	       (map.br_startblock != HOLESTARTBLOCK));
+
+	/*
+	 * Keep track of the blkno to save a lookup later
+	 */
+	dqp->q_blkno = XFS_FSB_TO_DADDR(mp, map.br_startblock);
+
+	/* now we can just get the buffer (there's nothing to read yet) */
+	bp = xfs_trans_get_buf(tp, mp->m_ddev_targp,
+			       dqp->q_blkno,
+			       mp->m_quotainfo->qi_dqchunklen,
+			       0);
+	if (!bp || (error = XFS_BUF_GETERROR(bp)))
+		goto error1;
+	/*
+	 * Make a chunk of dquots out of this buffer and log
+	 * the entire thing.
+	 */
+	xfs_qm_init_dquot_blk(tp, mp, be32_to_cpu(dqp->q_core.d_id),
+			      dqp->dq_flags & XFS_DQ_ALLTYPES, bp);
+
+	/*
+	 * xfs_bmap_finish() may commit the current transaction and
+	 * start a second transaction if the freelist is not empty.
+	 *
+	 * Since we still want to modify this buffer, we need to
+	 * ensure that the buffer is not released on commit of
+	 * the first transaction and ensure the buffer is added to the
+	 * second transaction.
+	 *
+	 * If there is only one transaction then don't stop the buffer
+	 * from being released when it commits later on.
+	 */
+
+	xfs_trans_bhold(tp, bp);
+
+	if ((error = xfs_bmap_finish(tpp, &flist, &committed))) {
+		goto error1;
+	}
+
+	if (committed) {
+		tp = *tpp;
+		xfs_trans_bjoin(tp, bp);
+	} else {
+		xfs_trans_bhold_release(tp, bp);
+	}
+
+	*O_bpp = bp;
+	return 0;
+
+      error1:
+	xfs_bmap_cancel(&flist);
+      error0:
+	xfs_iunlock(quotip, XFS_ILOCK_EXCL);
+
+	return (error);
+}
+
+/*
+ * Maps a dquot to the buffer containing its on-disk version.
+ * This returns a ptr to the buffer containing the on-disk dquot
+ * in the bpp param, and a ptr to the on-disk dquot within that buffer
+ */
+STATIC int
+xfs_qm_dqtobp(
+	xfs_trans_t		**tpp,
+	xfs_dquot_t		*dqp,
+	xfs_disk_dquot_t	**O_ddpp,
+	xfs_buf_t		**O_bpp,
+	uint			flags)
+{
+	xfs_bmbt_irec_t map;
+	int		nmaps = 1, error;
+	xfs_buf_t	*bp;
+	xfs_inode_t	*quotip = XFS_DQ_TO_QIP(dqp);
+	xfs_mount_t	*mp = dqp->q_mount;
+	xfs_disk_dquot_t *ddq;
+	xfs_dqid_t	id = be32_to_cpu(dqp->q_core.d_id);
+	xfs_trans_t	*tp = (tpp ? *tpp : NULL);
+
+	dqp->q_fileoffset = (xfs_fileoff_t)id / mp->m_quotainfo->qi_dqperchunk;
+
+	xfs_ilock(quotip, XFS_ILOCK_SHARED);
+	if (XFS_IS_THIS_QUOTA_OFF(dqp)) {
+		/*
+		 * Return if this type of quotas is turned off while we
+		 * didn't have the quota inode lock.
+		 */
+		xfs_iunlock(quotip, XFS_ILOCK_SHARED);
+		return ESRCH;
+	}
+
+	/*
+	 * Find the block map; no allocations yet
+	 */
+	error = xfs_bmapi(NULL, quotip, dqp->q_fileoffset,
+			  XFS_DQUOT_CLUSTER_SIZE_FSB, XFS_BMAPI_METADATA,
+			  NULL, 0, &map, &nmaps, NULL);
+
+	xfs_iunlock(quotip, XFS_ILOCK_SHARED);
+	if (error)
+		return error;
+
+	ASSERT(nmaps == 1);
+	ASSERT(map.br_blockcount == 1);
+
+	/*
+	 * Offset of dquot in the (fixed sized) dquot chunk.
+	 */
+	dqp->q_bufoffset = (id % mp->m_quotainfo->qi_dqperchunk) *
+		sizeof(xfs_dqblk_t);
+
+	ASSERT(map.br_startblock != DELAYSTARTBLOCK);
+	if (map.br_startblock == HOLESTARTBLOCK) {
+		/*
+		 * We don't allocate unless we're asked to
+		 */
+		if (!(flags & XFS_QMOPT_DQALLOC))
+			return ENOENT;
+
+		ASSERT(tp);
+		error = xfs_qm_dqalloc(tpp, mp, dqp, quotip,
+					dqp->q_fileoffset, &bp);
+		if (error)
+			return error;
+		tp = *tpp;
+	} else {
+		trace_xfs_dqtobp_read(dqp);
+
+		/*
+		 * store the blkno etc so that we don't have to do the
+		 * mapping all the time
+		 */
+		dqp->q_blkno = XFS_FSB_TO_DADDR(mp, map.br_startblock);
+
+		error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp,
+					   dqp->q_blkno,
+					   mp->m_quotainfo->qi_dqchunklen,
+					   0, &bp);
+		if (error || !bp)
+			return XFS_ERROR(error);
+	}
+
+	ASSERT(XFS_BUF_ISBUSY(bp));
+	ASSERT(XFS_BUF_VALUSEMA(bp) <= 0);
+
+	/*
+	 * calculate the location of the dquot inside the buffer.
+	 */
+	ddq = (struct xfs_disk_dquot *)(XFS_BUF_PTR(bp) + dqp->q_bufoffset);
+
+	/*
+	 * A simple sanity check in case we got a corrupted dquot...
+	 */
+	error = xfs_qm_dqcheck(mp, ddq, id, dqp->dq_flags & XFS_DQ_ALLTYPES,
+			   flags & (XFS_QMOPT_DQREPAIR|XFS_QMOPT_DOWARN),
+			   "dqtobp");
+	if (error) {
+		if (!(flags & XFS_QMOPT_DQREPAIR)) {
+			xfs_trans_brelse(tp, bp);
+			return XFS_ERROR(EIO);
+		}
+		XFS_BUF_BUSY(bp); /* We dirtied this */
+	}
+
+	*O_bpp = bp;
+	*O_ddpp = ddq;
+
+	return (0);
+}
+
+
+/*
+ * Read in the ondisk dquot using dqtobp() then copy it to an incore version,
+ * and release the buffer immediately.
+ *
+ */
+/* ARGSUSED */
+STATIC int
+xfs_qm_dqread(
+	xfs_trans_t	**tpp,
+	xfs_dqid_t	id,
+	xfs_dquot_t	*dqp,	/* dquot to get filled in */
+	uint		flags)
+{
+	xfs_disk_dquot_t *ddqp;
+	xfs_buf_t	 *bp;
+	int		 error;
+	xfs_trans_t	 *tp;
+
+	ASSERT(tpp);
+
+	trace_xfs_dqread(dqp);
+
+	/*
+	 * get a pointer to the on-disk dquot and the buffer containing it
+	 * dqp already knows its own type (GROUP/USER).
+	 */
+	if ((error = xfs_qm_dqtobp(tpp, dqp, &ddqp, &bp, flags))) {
+		return (error);
+	}
+	tp = *tpp;
+
+	/* copy everything from disk dquot to the incore dquot */
+	memcpy(&dqp->q_core, ddqp, sizeof(xfs_disk_dquot_t));
+	ASSERT(be32_to_cpu(dqp->q_core.d_id) == id);
+	xfs_qm_dquot_logitem_init(dqp);
+
+	/*
+	 * Reservation counters are defined as reservation plus current usage
+	 * to avoid having to add every time.
+	 */
+	dqp->q_res_bcount = be64_to_cpu(ddqp->d_bcount);
+	dqp->q_res_icount = be64_to_cpu(ddqp->d_icount);
+	dqp->q_res_rtbcount = be64_to_cpu(ddqp->d_rtbcount);
+
+	/* Mark the buf so that this will stay incore a little longer */
+	XFS_BUF_SET_VTYPE_REF(bp, B_FS_DQUOT, XFS_DQUOT_REF);
+
+	/*
+	 * We got the buffer with a xfs_trans_read_buf() (in dqtobp())
+	 * So we need to release with xfs_trans_brelse().
+	 * The strategy here is identical to that of inodes; we lock
+	 * the dquot in xfs_qm_dqget() before making it accessible to
+	 * others. This is because dquots, like inodes, need a good level of
+	 * concurrency, and we don't want to take locks on the entire buffers
+	 * for dquot accesses.
+	 * Note also that the dquot buffer may even be dirty at this point, if
+	 * this particular dquot was repaired. We still aren't afraid to
+	 * brelse it because we have the changes incore.
+	 */
+	ASSERT(XFS_BUF_ISBUSY(bp));
+	ASSERT(XFS_BUF_VALUSEMA(bp) <= 0);
+	xfs_trans_brelse(tp, bp);
+
+	return (error);
+}
+
+
+/*
+ * allocate an incore dquot from the kernel heap,
+ * and fill its core with quota information kept on disk.
+ * If XFS_QMOPT_DQALLOC is set, it'll allocate a dquot on disk
+ * if it wasn't already allocated.
+ */
+STATIC int
+xfs_qm_idtodq(
+	xfs_mount_t	*mp,
+	xfs_dqid_t	id,	 /* gid or uid, depending on type */
+	uint		type,	 /* UDQUOT or GDQUOT */
+	uint		flags,	 /* DQALLOC, DQREPAIR */
+	xfs_dquot_t	**O_dqpp)/* OUT : incore dquot, not locked */
+{
+	xfs_dquot_t	*dqp;
+	int		error;
+	xfs_trans_t	*tp;
+	int		cancelflags=0;
+
+	dqp = xfs_qm_dqinit(mp, id, type);
+	tp = NULL;
+	if (flags & XFS_QMOPT_DQALLOC) {
+		tp = xfs_trans_alloc(mp, XFS_TRANS_QM_DQALLOC);
+		error = xfs_trans_reserve(tp, XFS_QM_DQALLOC_SPACE_RES(mp),
+				XFS_WRITE_LOG_RES(mp) +
+				BBTOB(mp->m_quotainfo->qi_dqchunklen) - 1 +
+				128,
+				0,
+				XFS_TRANS_PERM_LOG_RES,
+				XFS_WRITE_LOG_COUNT);
+		if (error) {
+			cancelflags = 0;
+			goto error0;
+		}
+		cancelflags = XFS_TRANS_RELEASE_LOG_RES;
+	}
+
+	/*
+	 * Read it from disk; xfs_dqread() takes care of
+	 * all the necessary initialization of dquot's fields (locks, etc)
+	 */
+	if ((error = xfs_qm_dqread(&tp, id, dqp, flags))) {
+		/*
+		 * This can happen if quotas got turned off (ESRCH),
+		 * or if the dquot didn't exist on disk and we ask to
+		 * allocate (ENOENT).
+		 */
+		trace_xfs_dqread_fail(dqp);
+		cancelflags |= XFS_TRANS_ABORT;
+		goto error0;
+	}
+	if (tp) {
+		if ((error = xfs_trans_commit(tp, XFS_TRANS_RELEASE_LOG_RES)))
+			goto error1;
+	}
+
+	*O_dqpp = dqp;
+	return (0);
+
+ error0:
+	ASSERT(error);
+	if (tp)
+		xfs_trans_cancel(tp, cancelflags);
+ error1:
+	xfs_qm_dqdestroy(dqp);
+	*O_dqpp = NULL;
+	return (error);
+}
+
+/*
+ * Lookup a dquot in the incore dquot hashtable. We keep two separate
+ * hashtables for user and group dquots; and, these are global tables
+ * inside the XQM, not per-filesystem tables.
+ * The hash chain must be locked by caller, and it is left locked
+ * on return. Returning dquot is locked.
+ */
+STATIC int
+xfs_qm_dqlookup(
+	xfs_mount_t		*mp,
+	xfs_dqid_t		id,
+	xfs_dqhash_t		*qh,
+	xfs_dquot_t		**O_dqpp)
+{
+	xfs_dquot_t		*dqp;
+	uint			flist_locked;
+
+	ASSERT(mutex_is_locked(&qh->qh_lock));
+
+	flist_locked = B_FALSE;
+
+	/*
+	 * Traverse the hashchain looking for a match
+	 */
+	list_for_each_entry(dqp, &qh->qh_list, q_hashlist) {
+		/*
+		 * We already have the hashlock. We don't need the
+		 * dqlock to look at the id field of the dquot, since the
+		 * id can't be modified without the hashlock anyway.
+		 */
+		if (be32_to_cpu(dqp->q_core.d_id) == id && dqp->q_mount == mp) {
+			trace_xfs_dqlookup_found(dqp);
+
+			/*
+			 * All in core dquots must be on the dqlist of mp
+			 */
+			ASSERT(!list_empty(&dqp->q_mplist));
+
+			xfs_dqlock(dqp);
+			if (dqp->q_nrefs == 0) {
+				ASSERT(!list_empty(&dqp->q_freelist));
+				if (!mutex_trylock(&xfs_Gqm->qm_dqfrlist_lock)) {
+					trace_xfs_dqlookup_want(dqp);
+
+					/*
+					 * We may have raced with dqreclaim_one()
+					 * (and lost). So, flag that we don't
+					 * want the dquot to be reclaimed.
+					 */
+					dqp->dq_flags |= XFS_DQ_WANT;
+					xfs_dqunlock(dqp);
+					mutex_lock(&xfs_Gqm->qm_dqfrlist_lock);
+					xfs_dqlock(dqp);
+					dqp->dq_flags &= ~(XFS_DQ_WANT);
+				}
+				flist_locked = B_TRUE;
+			}
+
+			/*
+			 * id couldn't have changed; we had the hashlock all
+			 * along
+			 */
+			ASSERT(be32_to_cpu(dqp->q_core.d_id) == id);
+
+			if (flist_locked) {
+				if (dqp->q_nrefs != 0) {
+					mutex_unlock(&xfs_Gqm->qm_dqfrlist_lock);
+					flist_locked = B_FALSE;
+				} else {
+					/* take it off the freelist */
+					trace_xfs_dqlookup_freelist(dqp);
+					list_del_init(&dqp->q_freelist);
+					xfs_Gqm->qm_dqfrlist_cnt--;
+				}
+			}
+
+			XFS_DQHOLD(dqp);
+
+			if (flist_locked)
+				mutex_unlock(&xfs_Gqm->qm_dqfrlist_lock);
+			/*
+			 * move the dquot to the front of the hashchain
+			 */
+			ASSERT(mutex_is_locked(&qh->qh_lock));
+			list_move(&dqp->q_hashlist, &qh->qh_list);
+			trace_xfs_dqlookup_done(dqp);
+			*O_dqpp = dqp;
+			return 0;
+		}
+	}
+
+	*O_dqpp = NULL;
+	ASSERT(mutex_is_locked(&qh->qh_lock));
+	return (1);
+}
+
+/*
+ * Given the file system, inode OR id, and type (UDQUOT/GDQUOT), return a
+ * a locked dquot, doing an allocation (if requested) as needed.
+ * When both an inode and an id are given, the inode's id takes precedence.
+ * That is, if the id changes while we don't hold the ilock inside this
+ * function, the new dquot is returned, not necessarily the one requested
+ * in the id argument.
+ */
+int
+xfs_qm_dqget(
+	xfs_mount_t	*mp,
+	xfs_inode_t	*ip,	  /* locked inode (optional) */
+	xfs_dqid_t	id,	  /* uid/projid/gid depending on type */
+	uint		type,	  /* XFS_DQ_USER/XFS_DQ_PROJ/XFS_DQ_GROUP */
+	uint		flags,	  /* DQALLOC, DQSUSER, DQREPAIR, DOWARN */
+	xfs_dquot_t	**O_dqpp) /* OUT : locked incore dquot */
+{
+	xfs_dquot_t	*dqp;
+	xfs_dqhash_t	*h;
+	uint		version;
+	int		error;
+
+	ASSERT(XFS_IS_QUOTA_RUNNING(mp));
+	if ((! XFS_IS_UQUOTA_ON(mp) && type == XFS_DQ_USER) ||
+	    (! XFS_IS_PQUOTA_ON(mp) && type == XFS_DQ_PROJ) ||
+	    (! XFS_IS_GQUOTA_ON(mp) && type == XFS_DQ_GROUP)) {
+		return (ESRCH);
+	}
+	h = XFS_DQ_HASH(mp, id, type);
+
+#ifdef DEBUG
+	if (xfs_do_dqerror) {
+		if ((xfs_dqerror_target == mp->m_ddev_targp) &&
+		    (xfs_dqreq_num++ % xfs_dqerror_mod) == 0) {
+			xfs_debug(mp, "Returning error in dqget");
+			return (EIO);
+		}
+	}
+#endif
+
+ again:
+
+#ifdef DEBUG
+	ASSERT(type == XFS_DQ_USER ||
+	       type == XFS_DQ_PROJ ||
+	       type == XFS_DQ_GROUP);
+	if (ip) {
+		ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
+		if (type == XFS_DQ_USER)
+			ASSERT(ip->i_udquot == NULL);
+		else
+			ASSERT(ip->i_gdquot == NULL);
+	}
+#endif
+	mutex_lock(&h->qh_lock);
+
+	/*
+	 * Look in the cache (hashtable).
+	 * The chain is kept locked during lookup.
+	 */
+	if (xfs_qm_dqlookup(mp, id, h, O_dqpp) == 0) {
+		XQM_STATS_INC(xqmstats.xs_qm_dqcachehits);
+		/*
+		 * The dquot was found, moved to the front of the chain,
+		 * taken off the freelist if it was on it, and locked
+		 * at this point. Just unlock the hashchain and return.
+		 */
+		ASSERT(*O_dqpp);
+		ASSERT(XFS_DQ_IS_LOCKED(*O_dqpp));
+		mutex_unlock(&h->qh_lock);
+		trace_xfs_dqget_hit(*O_dqpp);
+		return (0);	/* success */
+	}
+	XQM_STATS_INC(xqmstats.xs_qm_dqcachemisses);
+
+	/*
+	 * Dquot cache miss. We don't want to keep the inode lock across
+	 * a (potential) disk read. Also we don't want to deal with the lock
+	 * ordering between quotainode and this inode. OTOH, dropping the inode
+	 * lock here means dealing with a chown that can happen before
+	 * we re-acquire the lock.
+	 */
+	if (ip)
+		xfs_iunlock(ip, XFS_ILOCK_EXCL);
+	/*
+	 * Save the hashchain version stamp, and unlock the chain, so that
+	 * we don't keep the lock across a disk read
+	 */
+	version = h->qh_version;
+	mutex_unlock(&h->qh_lock);
+
+	/*
+	 * Allocate the dquot on the kernel heap, and read the ondisk
+	 * portion off the disk. Also, do all the necessary initialization
+	 * This can return ENOENT if dquot didn't exist on disk and we didn't
+	 * ask it to allocate; ESRCH if quotas got turned off suddenly.
+	 */
+	if ((error = xfs_qm_idtodq(mp, id, type,
+				  flags & (XFS_QMOPT_DQALLOC|XFS_QMOPT_DQREPAIR|
+					   XFS_QMOPT_DOWARN),
+				  &dqp))) {
+		if (ip)
+			xfs_ilock(ip, XFS_ILOCK_EXCL);
+		return (error);
+	}
+
+	/*
+	 * See if this is mount code calling to look at the overall quota limits
+	 * which are stored in the id == 0 user or group's dquot.
+	 * Since we may not have done a quotacheck by this point, just return
+	 * the dquot without attaching it to any hashtables, lists, etc, or even
+	 * taking a reference.
+	 * The caller must dqdestroy this once done.
+	 */
+	if (flags & XFS_QMOPT_DQSUSER) {
+		ASSERT(id == 0);
+		ASSERT(! ip);
+		goto dqret;
+	}
+
+	/*
+	 * Dquot lock comes after hashlock in the lock ordering
+	 */
+	if (ip) {
+		xfs_ilock(ip, XFS_ILOCK_EXCL);
+
+		/*
+		 * A dquot could be attached to this inode by now, since
+		 * we had dropped the ilock.
+		 */
+		if (type == XFS_DQ_USER) {
+			if (!XFS_IS_UQUOTA_ON(mp)) {
+				/* inode stays locked on return */
+				xfs_qm_dqdestroy(dqp);
+				return XFS_ERROR(ESRCH);
+			}
+			if (ip->i_udquot) {
+				xfs_qm_dqdestroy(dqp);
+				dqp = ip->i_udquot;
+				xfs_dqlock(dqp);
+				goto dqret;
+			}
+		} else {
+			if (!XFS_IS_OQUOTA_ON(mp)) {
+				/* inode stays locked on return */
+				xfs_qm_dqdestroy(dqp);
+				return XFS_ERROR(ESRCH);
+			}
+			if (ip->i_gdquot) {
+				xfs_qm_dqdestroy(dqp);
+				dqp = ip->i_gdquot;
+				xfs_dqlock(dqp);
+				goto dqret;
+			}
+		}
+	}
+
+	/*
+	 * Hashlock comes after ilock in lock order
+	 */
+	mutex_lock(&h->qh_lock);
+	if (version != h->qh_version) {
+		xfs_dquot_t *tmpdqp;
+		/*
+		 * Now, see if somebody else put the dquot in the
+		 * hashtable before us. This can happen because we didn't
+		 * keep the hashchain lock. We don't have to worry about
+		 * lock order between the two dquots here since dqp isn't
+		 * on any findable lists yet.
+		 */
+		if (xfs_qm_dqlookup(mp, id, h, &tmpdqp) == 0) {
+			/*
+			 * Duplicate found. Just throw away the new dquot
+			 * and start over.
+			 */
+			xfs_qm_dqput(tmpdqp);
+			mutex_unlock(&h->qh_lock);
+			xfs_qm_dqdestroy(dqp);
+			XQM_STATS_INC(xqmstats.xs_qm_dquot_dups);
+			goto again;
+		}
+	}
+
+	/*
+	 * Put the dquot at the beginning of the hash-chain and mp's list
+	 * LOCK ORDER: hashlock, freelistlock, mplistlock, udqlock, gdqlock ..
+	 */
+	ASSERT(mutex_is_locked(&h->qh_lock));
+	dqp->q_hash = h;
+	list_add(&dqp->q_hashlist, &h->qh_list);
+	h->qh_version++;
+
+	/*
+	 * Attach this dquot to this filesystem's list of all dquots,
+	 * kept inside the mount structure in m_quotainfo field
+	 */
+	mutex_lock(&mp->m_quotainfo->qi_dqlist_lock);
+
+	/*
+	 * We return a locked dquot to the caller, with a reference taken
+	 */
+	xfs_dqlock(dqp);
+	dqp->q_nrefs = 1;
+
+	list_add(&dqp->q_mplist, &mp->m_quotainfo->qi_dqlist);
+	mp->m_quotainfo->qi_dquots++;
+	mutex_unlock(&mp->m_quotainfo->qi_dqlist_lock);
+	mutex_unlock(&h->qh_lock);
+ dqret:
+	ASSERT((ip == NULL) || xfs_isilocked(ip, XFS_ILOCK_EXCL));
+	trace_xfs_dqget_miss(dqp);
+	*O_dqpp = dqp;
+	return (0);
+}
+
+
+/*
+ * Release a reference to the dquot (decrement ref-count)
+ * and unlock it. If there is a group quota attached to this
+ * dquot, carefully release that too without tripping over
+ * deadlocks'n'stuff.
+ */
+void
+xfs_qm_dqput(
+	xfs_dquot_t	*dqp)
+{
+	xfs_dquot_t	*gdqp;
+
+	ASSERT(dqp->q_nrefs > 0);
+	ASSERT(XFS_DQ_IS_LOCKED(dqp));
+
+	trace_xfs_dqput(dqp);
+
+	if (dqp->q_nrefs != 1) {
+		dqp->q_nrefs--;
+		xfs_dqunlock(dqp);
+		return;
+	}
+
+	/*
+	 * drop the dqlock and acquire the freelist and dqlock
+	 * in the right order; but try to get it out-of-order first
+	 */
+	if (!mutex_trylock(&xfs_Gqm->qm_dqfrlist_lock)) {
+		trace_xfs_dqput_wait(dqp);
+		xfs_dqunlock(dqp);
+		mutex_lock(&xfs_Gqm->qm_dqfrlist_lock);
+		xfs_dqlock(dqp);
+	}
+
+	while (1) {
+		gdqp = NULL;
+
+		/* We can't depend on nrefs being == 1 here */
+		if (--dqp->q_nrefs == 0) {
+			trace_xfs_dqput_free(dqp);
+
+			list_add_tail(&dqp->q_freelist, &xfs_Gqm->qm_dqfrlist);
+			xfs_Gqm->qm_dqfrlist_cnt++;
+
+			/*
+			 * If we just added a udquot to the freelist, then
+			 * we want to release the gdquot reference that
+			 * it (probably) has. Otherwise it'll keep the
+			 * gdquot from getting reclaimed.
+			 */
+			if ((gdqp = dqp->q_gdquot)) {
+				/*
+				 * Avoid a recursive dqput call
+				 */
+				xfs_dqlock(gdqp);
+				dqp->q_gdquot = NULL;
+			}
+		}
+		xfs_dqunlock(dqp);
+
+		/*
+		 * If we had a group quota inside the user quota as a hint,
+		 * release it now.
+		 */
+		if (! gdqp)
+			break;
+		dqp = gdqp;
+	}
+	mutex_unlock(&xfs_Gqm->qm_dqfrlist_lock);
+}
+
+/*
+ * Release a dquot. Flush it if dirty, then dqput() it.
+ * dquot must not be locked.
+ */
+void
+xfs_qm_dqrele(
+	xfs_dquot_t	*dqp)
+{
+	if (!dqp)
+		return;
+
+	trace_xfs_dqrele(dqp);
+
+	xfs_dqlock(dqp);
+	/*
+	 * We don't care to flush it if the dquot is dirty here.
+	 * That will create stutters that we want to avoid.
+	 * Instead we do a delayed write when we try to reclaim
+	 * a dirty dquot. Also xfs_sync will take part of the burden...
+	 */
+	xfs_qm_dqput(dqp);
+}
+
+/*
+ * This is the dquot flushing I/O completion routine.  It is called
+ * from interrupt level when the buffer containing the dquot is
+ * flushed to disk.  It is responsible for removing the dquot logitem
+ * from the AIL if it has not been re-logged, and unlocking the dquot's
+ * flush lock. This behavior is very similar to that of inodes..
+ */
+STATIC void
+xfs_qm_dqflush_done(
+	struct xfs_buf		*bp,
+	struct xfs_log_item	*lip)
+{
+	xfs_dq_logitem_t	*qip = (struct xfs_dq_logitem *)lip;
+	xfs_dquot_t		*dqp = qip->qli_dquot;
+	struct xfs_ail		*ailp = lip->li_ailp;
+
+	/*
+	 * We only want to pull the item from the AIL if its
+	 * location in the log has not changed since we started the flush.
+	 * Thus, we only bother if the dquot's lsn has
+	 * not changed. First we check the lsn outside the lock
+	 * since it's cheaper, and then we recheck while
+	 * holding the lock before removing the dquot from the AIL.
+	 */
+	if ((lip->li_flags & XFS_LI_IN_AIL) &&
+	    lip->li_lsn == qip->qli_flush_lsn) {
+
+		/* xfs_trans_ail_delete() drops the AIL lock. */
+		spin_lock(&ailp->xa_lock);
+		if (lip->li_lsn == qip->qli_flush_lsn)
+			xfs_trans_ail_delete(ailp, lip);
+		else
+			spin_unlock(&ailp->xa_lock);
+	}
+
+	/*
+	 * Release the dq's flush lock since we're done with it.
+	 */
+	xfs_dqfunlock(dqp);
+}
+
+/*
+ * Write a modified dquot to disk.
+ * The dquot must be locked and the flush lock too taken by caller.
+ * The flush lock will not be unlocked until the dquot reaches the disk,
+ * but the dquot is free to be unlocked and modified by the caller
+ * in the interim. Dquot is still locked on return. This behavior is
+ * identical to that of inodes.
+ */
+int
+xfs_qm_dqflush(
+	xfs_dquot_t		*dqp,
+	uint			flags)
+{
+	struct xfs_mount	*mp = dqp->q_mount;
+	struct xfs_buf		*bp;
+	struct xfs_disk_dquot	*ddqp;
+	int			error;
+
+	ASSERT(XFS_DQ_IS_LOCKED(dqp));
+	ASSERT(!completion_done(&dqp->q_flush));
+
+	trace_xfs_dqflush(dqp);
+
+	/*
+	 * If not dirty, or it's pinned and we are not supposed to block, nada.
+	 */
+	if (!XFS_DQ_IS_DIRTY(dqp) ||
+	    (!(flags & SYNC_WAIT) && atomic_read(&dqp->q_pincount) > 0)) {
+		xfs_dqfunlock(dqp);
+		return 0;
+	}
+	xfs_qm_dqunpin_wait(dqp);
+
+	/*
+	 * This may have been unpinned because the filesystem is shutting
+	 * down forcibly. If that's the case we must not write this dquot
+	 * to disk, because the log record didn't make it to disk!
+	 */
+	if (XFS_FORCED_SHUTDOWN(mp)) {
+		dqp->dq_flags &= ~XFS_DQ_DIRTY;
+		xfs_dqfunlock(dqp);
+		return XFS_ERROR(EIO);
+	}
+
+	/*
+	 * Get the buffer containing the on-disk dquot
+	 */
+	error = xfs_trans_read_buf(mp, NULL, mp->m_ddev_targp, dqp->q_blkno,
+				   mp->m_quotainfo->qi_dqchunklen, 0, &bp);
+	if (error) {
+		ASSERT(error != ENOENT);
+		xfs_dqfunlock(dqp);
+		return error;
+	}
+
+	/*
+	 * Calculate the location of the dquot inside the buffer.
+	 */
+	ddqp = (struct xfs_disk_dquot *)(XFS_BUF_PTR(bp) + dqp->q_bufoffset);
+
+	/*
+	 * A simple sanity check in case we got a corrupted dquot..
+	 */
+	error = xfs_qm_dqcheck(mp, &dqp->q_core, be32_to_cpu(ddqp->d_id), 0,
+			   XFS_QMOPT_DOWARN, "dqflush (incore copy)");
+	if (error) {
+		xfs_buf_relse(bp);
+		xfs_dqfunlock(dqp);
+		xfs_force_shutdown(mp, SHUTDOWN_CORRUPT_INCORE);
+		return XFS_ERROR(EIO);
+	}
+
+	/* This is the only portion of data that needs to persist */
+	memcpy(ddqp, &dqp->q_core, sizeof(xfs_disk_dquot_t));
+
+	/*
+	 * Clear the dirty field and remember the flush lsn for later use.
+	 */
+	dqp->dq_flags &= ~XFS_DQ_DIRTY;
+
+	xfs_trans_ail_copy_lsn(mp->m_ail, &dqp->q_logitem.qli_flush_lsn,
+					&dqp->q_logitem.qli_item.li_lsn);
+
+	/*
+	 * Attach an iodone routine so that we can remove this dquot from the
+	 * AIL and release the flush lock once the dquot is synced to disk.
+	 */
+	xfs_buf_attach_iodone(bp, xfs_qm_dqflush_done,
+				  &dqp->q_logitem.qli_item);
+
+	/*
+	 * If the buffer is pinned then push on the log so we won't
+	 * get stuck waiting in the write for too long.
+	 */
+	if (XFS_BUF_ISPINNED(bp)) {
+		trace_xfs_dqflush_force(dqp);
+		xfs_log_force(mp, 0);
+	}
+
+	if (flags & SYNC_WAIT)
+		error = xfs_bwrite(mp, bp);
+	else
+		xfs_bdwrite(mp, bp);
+
+	trace_xfs_dqflush_done(dqp);
+
+	/*
+	 * dqp is still locked, but caller is free to unlock it now.
+	 */
+	return error;
+
+}
+
+int
+xfs_qm_dqlock_nowait(
+	xfs_dquot_t *dqp)
+{
+	return mutex_trylock(&dqp->q_qlock);
+}
+
+void
+xfs_dqlock(
+	xfs_dquot_t *dqp)
+{
+	mutex_lock(&dqp->q_qlock);
+}
+
+void
+xfs_dqunlock(
+	xfs_dquot_t *dqp)
+{
+	mutex_unlock(&(dqp->q_qlock));
+	if (dqp->q_logitem.qli_dquot == dqp) {
+		/* Once was dqp->q_mount, but might just have been cleared */
+		xfs_trans_unlocked_item(dqp->q_logitem.qli_item.li_ailp,
+					(xfs_log_item_t*)&(dqp->q_logitem));
+	}
+}
+
+
+void
+xfs_dqunlock_nonotify(
+	xfs_dquot_t *dqp)
+{
+	mutex_unlock(&(dqp->q_qlock));
+}
+
+/*
+ * Lock two xfs_dquot structures.
+ *
+ * To avoid deadlocks we always lock the quota structure with
+ * the lowerd id first.
+ */
+void
+xfs_dqlock2(
+	xfs_dquot_t	*d1,
+	xfs_dquot_t	*d2)
+{
+	if (d1 && d2) {
+		ASSERT(d1 != d2);
+		if (be32_to_cpu(d1->q_core.d_id) >
+		    be32_to_cpu(d2->q_core.d_id)) {
+			mutex_lock(&d2->q_qlock);
+			mutex_lock_nested(&d1->q_qlock, XFS_QLOCK_NESTED);
+		} else {
+			mutex_lock(&d1->q_qlock);
+			mutex_lock_nested(&d2->q_qlock, XFS_QLOCK_NESTED);
+		}
+	} else if (d1) {
+		mutex_lock(&d1->q_qlock);
+	} else if (d2) {
+		mutex_lock(&d2->q_qlock);
+	}
+}
+
+
+/*
+ * Take a dquot out of the mount's dqlist as well as the hashlist.
+ * This is called via unmount as well as quotaoff, and the purge
+ * will always succeed unless there are soft (temp) references
+ * outstanding.
+ *
+ * This returns 0 if it was purged, 1 if it wasn't. It's not an error code
+ * that we're returning! XXXsup - not cool.
+ */
+/* ARGSUSED */
+int
+xfs_qm_dqpurge(
+	xfs_dquot_t	*dqp)
+{
+	xfs_dqhash_t	*qh = dqp->q_hash;
+	xfs_mount_t	*mp = dqp->q_mount;
+
+	ASSERT(mutex_is_locked(&mp->m_quotainfo->qi_dqlist_lock));
+	ASSERT(mutex_is_locked(&dqp->q_hash->qh_lock));
+
+	xfs_dqlock(dqp);
+	/*
+	 * We really can't afford to purge a dquot that is
+	 * referenced, because these are hard refs.
+	 * It shouldn't happen in general because we went thru _all_ inodes in
+	 * dqrele_all_inodes before calling this and didn't let the mountlock go.
+	 * However it is possible that we have dquots with temporary
+	 * references that are not attached to an inode. e.g. see xfs_setattr().
+	 */
+	if (dqp->q_nrefs != 0) {
+		xfs_dqunlock(dqp);
+		mutex_unlock(&dqp->q_hash->qh_lock);
+		return (1);
+	}
+
+	ASSERT(!list_empty(&dqp->q_freelist));
+
+	/*
+	 * If we're turning off quotas, we have to make sure that, for
+	 * example, we don't delete quota disk blocks while dquots are
+	 * in the process of getting written to those disk blocks.
+	 * This dquot might well be on AIL, and we can't leave it there
+	 * if we're turning off quotas. Basically, we need this flush
+	 * lock, and are willing to block on it.
+	 */
+	if (!xfs_dqflock_nowait(dqp)) {
+		/*
+		 * Block on the flush lock after nudging dquot buffer,
+		 * if it is incore.
+		 */
+		xfs_qm_dqflock_pushbuf_wait(dqp);
+	}
+
+	/*
+	 * XXXIf we're turning this type of quotas off, we don't care
+	 * about the dirty metadata sitting in this dquot. OTOH, if
+	 * we're unmounting, we do care, so we flush it and wait.
+	 */
+	if (XFS_DQ_IS_DIRTY(dqp)) {
+		int	error;
+
+		/* dqflush unlocks dqflock */
+		/*
+		 * Given that dqpurge is a very rare occurrence, it is OK
+		 * that we're holding the hashlist and mplist locks
+		 * across the disk write. But, ... XXXsup
+		 *
+		 * We don't care about getting disk errors here. We need
+		 * to purge this dquot anyway, so we go ahead regardless.
+		 */
+		error = xfs_qm_dqflush(dqp, SYNC_WAIT);
+		if (error)
+			xfs_warn(mp, "%s: dquot %p flush failed",
+				__func__, dqp);
+		xfs_dqflock(dqp);
+	}
+	ASSERT(atomic_read(&dqp->q_pincount) == 0);
+	ASSERT(XFS_FORCED_SHUTDOWN(mp) ||
+	       !(dqp->q_logitem.qli_item.li_flags & XFS_LI_IN_AIL));
+
+	list_del_init(&dqp->q_hashlist);
+	qh->qh_version++;
+	list_del_init(&dqp->q_mplist);
+	mp->m_quotainfo->qi_dqreclaims++;
+	mp->m_quotainfo->qi_dquots--;
+	/*
+	 * XXX Move this to the front of the freelist, if we can get the
+	 * freelist lock.
+	 */
+	ASSERT(!list_empty(&dqp->q_freelist));
+
+	dqp->q_mount = NULL;
+	dqp->q_hash = NULL;
+	dqp->dq_flags = XFS_DQ_INACTIVE;
+	memset(&dqp->q_core, 0, sizeof(dqp->q_core));
+	xfs_dqfunlock(dqp);
+	xfs_dqunlock(dqp);
+	mutex_unlock(&qh->qh_lock);
+	return (0);
+}
+
+
+#ifdef QUOTADEBUG
+void
+xfs_qm_dqprint(xfs_dquot_t *dqp)
+{
+	struct xfs_mount	*mp = dqp->q_mount;
+
+	xfs_debug(mp, "-----------KERNEL DQUOT----------------");
+	xfs_debug(mp, "---- dquotID =  %d",
+		(int)be32_to_cpu(dqp->q_core.d_id));
+	xfs_debug(mp, "---- type    =  %s", DQFLAGTO_TYPESTR(dqp));
+	xfs_debug(mp, "---- fs      =  0x%p", dqp->q_mount);
+	xfs_debug(mp, "---- blkno   =  0x%x", (int) dqp->q_blkno);
+	xfs_debug(mp, "---- boffset =  0x%x", (int) dqp->q_bufoffset);
+	xfs_debug(mp, "---- blkhlimit =  %Lu (0x%x)",
+		be64_to_cpu(dqp->q_core.d_blk_hardlimit),
+		(int)be64_to_cpu(dqp->q_core.d_blk_hardlimit));
+	xfs_debug(mp, "---- blkslimit =  %Lu (0x%x)",
+		be64_to_cpu(dqp->q_core.d_blk_softlimit),
+		(int)be64_to_cpu(dqp->q_core.d_blk_softlimit));
+	xfs_debug(mp, "---- inohlimit =  %Lu (0x%x)",
+		be64_to_cpu(dqp->q_core.d_ino_hardlimit),
+		(int)be64_to_cpu(dqp->q_core.d_ino_hardlimit));
+	xfs_debug(mp, "---- inoslimit =  %Lu (0x%x)",
+		be64_to_cpu(dqp->q_core.d_ino_softlimit),
+		(int)be64_to_cpu(dqp->q_core.d_ino_softlimit));
+	xfs_debug(mp, "---- bcount  =  %Lu (0x%x)",
+		be64_to_cpu(dqp->q_core.d_bcount),
+		(int)be64_to_cpu(dqp->q_core.d_bcount));
+	xfs_debug(mp, "---- icount  =  %Lu (0x%x)",
+		be64_to_cpu(dqp->q_core.d_icount),
+		(int)be64_to_cpu(dqp->q_core.d_icount));
+	xfs_debug(mp, "---- btimer  =  %d",
+		(int)be32_to_cpu(dqp->q_core.d_btimer));
+	xfs_debug(mp, "---- itimer  =  %d",
+		(int)be32_to_cpu(dqp->q_core.d_itimer));
+	xfs_debug(mp, "---------------------------");
+}
+#endif
+
+/*
+ * Give the buffer a little push if it is incore and
+ * wait on the flush lock.
+ */
+void
+xfs_qm_dqflock_pushbuf_wait(
+	xfs_dquot_t	*dqp)
+{
+	xfs_mount_t	*mp = dqp->q_mount;
+	xfs_buf_t	*bp;
+
+	/*
+	 * Check to see if the dquot has been flushed delayed
+	 * write.  If so, grab its buffer and send it
+	 * out immediately.  We'll be able to acquire
+	 * the flush lock when the I/O completes.
+	 */
+	bp = xfs_incore(mp->m_ddev_targp, dqp->q_blkno,
+			mp->m_quotainfo->qi_dqchunklen, XBF_TRYLOCK);
+	if (!bp)
+		goto out_lock;
+
+	if (XFS_BUF_ISDELAYWRITE(bp)) {
+		if (XFS_BUF_ISPINNED(bp))
+			xfs_log_force(mp, 0);
+		xfs_buf_delwri_promote(bp);
+		wake_up_process(bp->b_target->bt_task);
+	}
+	xfs_buf_relse(bp);
+out_lock:
+	xfs_dqflock(dqp);
+}
diff --git a/fs/xfs/quota/xfs_dquot.h b/fs/xfs/quota/xfs_dquot.h
new file mode 100644
index 00000000..5da3a23b
--- /dev/null
+++ b/fs/xfs/quota/xfs_dquot.h
@@ -0,0 +1,143 @@
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_DQUOT_H__
+#define __XFS_DQUOT_H__
+
+/*
+ * Dquots are structures that hold quota information about a user or a group,
+ * much like inodes are for files. In fact, dquots share many characteristics
+ * with inodes. However, dquots can also be a centralized resource, relative
+ * to a collection of inodes. In this respect, dquots share some characteristics
+ * of the superblock.
+ * XFS dquots exploit both those in its algorithms. They make every attempt
+ * to not be a bottleneck when quotas are on and have minimal impact, if any,
+ * when quotas are off.
+ */
+
+/*
+ * The hash chain headers (hash buckets)
+ */
+typedef struct xfs_dqhash {
+	struct list_head  qh_list;
+	struct mutex	  qh_lock;
+	uint		  qh_version;	/* ever increasing version */
+	uint		  qh_nelems;	/* number of dquots on the list */
+} xfs_dqhash_t;
+
+struct xfs_mount;
+struct xfs_trans;
+
+/*
+ * The incore dquot structure
+ */
+typedef struct xfs_dquot {
+	uint		 dq_flags;	/* various flags (XFS_DQ_*) */
+	struct list_head q_freelist;	/* global free list of dquots */
+	struct list_head q_mplist;	/* mount's list of dquots */
+	struct list_head q_hashlist;	/* gloabl hash list of dquots */
+	xfs_dqhash_t	*q_hash;	/* the hashchain header */
+	struct xfs_mount*q_mount;	/* filesystem this relates to */
+	struct xfs_trans*q_transp;	/* trans this belongs to currently */
+	uint		 q_nrefs;	/* # active refs from inodes */
+	xfs_daddr_t	 q_blkno;	/* blkno of dquot buffer */
+	int		 q_bufoffset;	/* off of dq in buffer (# dquots) */
+	xfs_fileoff_t	 q_fileoffset;	/* offset in quotas file */
+
+	struct xfs_dquot*q_gdquot;	/* group dquot, hint only */
+	xfs_disk_dquot_t q_core;	/* actual usage & quotas */
+	xfs_dq_logitem_t q_logitem;	/* dquot log item */
+	xfs_qcnt_t	 q_res_bcount;	/* total regular nblks used+reserved */
+	xfs_qcnt_t	 q_res_icount;	/* total inos allocd+reserved */
+	xfs_qcnt_t	 q_res_rtbcount;/* total realtime blks used+reserved */
+	struct mutex	 q_qlock;	/* quota lock */
+	struct completion q_flush;	/* flush completion queue */
+	atomic_t          q_pincount;	/* dquot pin count */
+	wait_queue_head_t q_pinwait;	/* dquot pinning wait queue */
+} xfs_dquot_t;
+
+/*
+ * Lock hierarchy for q_qlock:
+ *	XFS_QLOCK_NORMAL is the implicit default,
+ * 	XFS_QLOCK_NESTED is the dquot with the higher id in xfs_dqlock2
+ */
+enum {
+	XFS_QLOCK_NORMAL = 0,
+	XFS_QLOCK_NESTED,
+};
+
+#define XFS_DQHOLD(dqp)		((dqp)->q_nrefs++)
+
+/*
+ * Manage the q_flush completion queue embedded in the dquot.  This completion
+ * queue synchronizes processes attempting to flush the in-core dquot back to
+ * disk.
+ */
+static inline void xfs_dqflock(xfs_dquot_t *dqp)
+{
+	wait_for_completion(&dqp->q_flush);
+}
+
+static inline int xfs_dqflock_nowait(xfs_dquot_t *dqp)
+{
+	return try_wait_for_completion(&dqp->q_flush);
+}
+
+static inline void xfs_dqfunlock(xfs_dquot_t *dqp)
+{
+	complete(&dqp->q_flush);
+}
+
+#define XFS_DQ_IS_LOCKED(dqp)	(mutex_is_locked(&((dqp)->q_qlock)))
+#define XFS_DQ_IS_DIRTY(dqp)	((dqp)->dq_flags & XFS_DQ_DIRTY)
+#define XFS_QM_ISUDQ(dqp)	((dqp)->dq_flags & XFS_DQ_USER)
+#define XFS_QM_ISPDQ(dqp)	((dqp)->dq_flags & XFS_DQ_PROJ)
+#define XFS_QM_ISGDQ(dqp)	((dqp)->dq_flags & XFS_DQ_GROUP)
+#define XFS_DQ_TO_QINF(dqp)	((dqp)->q_mount->m_quotainfo)
+#define XFS_DQ_TO_QIP(dqp)	(XFS_QM_ISUDQ(dqp) ? \
+				 XFS_DQ_TO_QINF(dqp)->qi_uquotaip : \
+				 XFS_DQ_TO_QINF(dqp)->qi_gquotaip)
+
+#define XFS_IS_THIS_QUOTA_OFF(d) (! (XFS_QM_ISUDQ(d) ? \
+				     (XFS_IS_UQUOTA_ON((d)->q_mount)) : \
+				     (XFS_IS_OQUOTA_ON((d)->q_mount))))
+
+#ifdef QUOTADEBUG
+extern void		xfs_qm_dqprint(xfs_dquot_t *);
+#else
+#define xfs_qm_dqprint(a)
+#endif
+
+extern void		xfs_qm_dqdestroy(xfs_dquot_t *);
+extern int		xfs_qm_dqflush(xfs_dquot_t *, uint);
+extern int		xfs_qm_dqpurge(xfs_dquot_t *);
+extern void		xfs_qm_dqunpin_wait(xfs_dquot_t *);
+extern int		xfs_qm_dqlock_nowait(xfs_dquot_t *);
+extern void		xfs_qm_dqflock_pushbuf_wait(xfs_dquot_t *dqp);
+extern void		xfs_qm_adjust_dqtimers(xfs_mount_t *,
+					xfs_disk_dquot_t *);
+extern void		xfs_qm_adjust_dqlimits(xfs_mount_t *,
+					xfs_disk_dquot_t *);
+extern int		xfs_qm_dqget(xfs_mount_t *, xfs_inode_t *,
+					xfs_dqid_t, uint, uint, xfs_dquot_t **);
+extern void		xfs_qm_dqput(xfs_dquot_t *);
+extern void		xfs_dqlock(xfs_dquot_t *);
+extern void		xfs_dqlock2(xfs_dquot_t *, xfs_dquot_t *);
+extern void		xfs_dqunlock(xfs_dquot_t *);
+extern void		xfs_dqunlock_nonotify(xfs_dquot_t *);
+
+#endif /* __XFS_DQUOT_H__ */
diff --git a/fs/xfs/quota/xfs_dquot_item.c b/fs/xfs/quota/xfs_dquot_item.c
new file mode 100644
index 00000000..8126fc2e
--- /dev/null
+++ b/fs/xfs/quota/xfs_dquot_item.c
@@ -0,0 +1,533 @@
+/*
+ * Copyright (c) 2000-2003 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_alloc.h"
+#include "xfs_quota.h"
+#include "xfs_mount.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_inode.h"
+#include "xfs_bmap.h"
+#include "xfs_rtalloc.h"
+#include "xfs_error.h"
+#include "xfs_itable.h"
+#include "xfs_attr.h"
+#include "xfs_buf_item.h"
+#include "xfs_trans_priv.h"
+#include "xfs_qm.h"
+
+static inline struct xfs_dq_logitem *DQUOT_ITEM(struct xfs_log_item *lip)
+{
+	return container_of(lip, struct xfs_dq_logitem, qli_item);
+}
+
+/*
+ * returns the number of iovecs needed to log the given dquot item.
+ */
+STATIC uint
+xfs_qm_dquot_logitem_size(
+	struct xfs_log_item	*lip)
+{
+	/*
+	 * we need only two iovecs, one for the format, one for the real thing
+	 */
+	return 2;
+}
+
+/*
+ * fills in the vector of log iovecs for the given dquot log item.
+ */
+STATIC void
+xfs_qm_dquot_logitem_format(
+	struct xfs_log_item	*lip,
+	struct xfs_log_iovec	*logvec)
+{
+	struct xfs_dq_logitem	*qlip = DQUOT_ITEM(lip);
+
+	logvec->i_addr = &qlip->qli_format;
+	logvec->i_len  = sizeof(xfs_dq_logformat_t);
+	logvec->i_type = XLOG_REG_TYPE_QFORMAT;
+	logvec++;
+	logvec->i_addr = &qlip->qli_dquot->q_core;
+	logvec->i_len  = sizeof(xfs_disk_dquot_t);
+	logvec->i_type = XLOG_REG_TYPE_DQUOT;
+
+	ASSERT(2 == lip->li_desc->lid_size);
+	qlip->qli_format.qlf_size = 2;
+
+}
+
+/*
+ * Increment the pin count of the given dquot.
+ */
+STATIC void
+xfs_qm_dquot_logitem_pin(
+	struct xfs_log_item	*lip)
+{
+	struct xfs_dquot	*dqp = DQUOT_ITEM(lip)->qli_dquot;
+
+	ASSERT(XFS_DQ_IS_LOCKED(dqp));
+	atomic_inc(&dqp->q_pincount);
+}
+
+/*
+ * Decrement the pin count of the given dquot, and wake up
+ * anyone in xfs_dqwait_unpin() if the count goes to 0.	 The
+ * dquot must have been previously pinned with a call to
+ * xfs_qm_dquot_logitem_pin().
+ */
+STATIC void
+xfs_qm_dquot_logitem_unpin(
+	struct xfs_log_item	*lip,
+	int			remove)
+{
+	struct xfs_dquot	*dqp = DQUOT_ITEM(lip)->qli_dquot;
+
+	ASSERT(atomic_read(&dqp->q_pincount) > 0);
+	if (atomic_dec_and_test(&dqp->q_pincount))
+		wake_up(&dqp->q_pinwait);
+}
+
+/*
+ * Given the logitem, this writes the corresponding dquot entry to disk
+ * asynchronously. This is called with the dquot entry securely locked;
+ * we simply get xfs_qm_dqflush() to do the work, and unlock the dquot
+ * at the end.
+ */
+STATIC void
+xfs_qm_dquot_logitem_push(
+	struct xfs_log_item	*lip)
+{
+	struct xfs_dquot	*dqp = DQUOT_ITEM(lip)->qli_dquot;
+	int			error;
+
+	ASSERT(XFS_DQ_IS_LOCKED(dqp));
+	ASSERT(!completion_done(&dqp->q_flush));
+
+	/*
+	 * Since we were able to lock the dquot's flush lock and
+	 * we found it on the AIL, the dquot must be dirty.  This
+	 * is because the dquot is removed from the AIL while still
+	 * holding the flush lock in xfs_dqflush_done().  Thus, if
+	 * we found it in the AIL and were able to obtain the flush
+	 * lock without sleeping, then there must not have been
+	 * anyone in the process of flushing the dquot.
+	 */
+	error = xfs_qm_dqflush(dqp, 0);
+	if (error)
+		xfs_warn(dqp->q_mount, "%s: push error %d on dqp %p",
+			__func__, error, dqp);
+	xfs_dqunlock(dqp);
+}
+
+STATIC xfs_lsn_t
+xfs_qm_dquot_logitem_committed(
+	struct xfs_log_item	*lip,
+	xfs_lsn_t		lsn)
+{
+	/*
+	 * We always re-log the entire dquot when it becomes dirty,
+	 * so, the latest copy _is_ the only one that matters.
+	 */
+	return lsn;
+}
+
+/*
+ * This is called to wait for the given dquot to be unpinned.
+ * Most of these pin/unpin routines are plagiarized from inode code.
+ */
+void
+xfs_qm_dqunpin_wait(
+	struct xfs_dquot	*dqp)
+{
+	ASSERT(XFS_DQ_IS_LOCKED(dqp));
+	if (atomic_read(&dqp->q_pincount) == 0)
+		return;
+
+	/*
+	 * Give the log a push so we don't wait here too long.
+	 */
+	xfs_log_force(dqp->q_mount, 0);
+	wait_event(dqp->q_pinwait, (atomic_read(&dqp->q_pincount) == 0));
+}
+
+/*
+ * This is called when IOP_TRYLOCK returns XFS_ITEM_PUSHBUF to indicate that
+ * the dquot is locked by us, but the flush lock isn't. So, here we are
+ * going to see if the relevant dquot buffer is incore, waiting on DELWRI.
+ * If so, we want to push it out to help us take this item off the AIL as soon
+ * as possible.
+ *
+ * We must not be holding the AIL lock at this point. Calling incore() to
+ * search the buffer cache can be a time consuming thing, and AIL lock is a
+ * spinlock.
+ */
+STATIC bool
+xfs_qm_dquot_logitem_pushbuf(
+	struct xfs_log_item	*lip)
+{
+	struct xfs_dq_logitem	*qlip = DQUOT_ITEM(lip);
+	struct xfs_dquot	*dqp = qlip->qli_dquot;
+	struct xfs_buf		*bp;
+	bool			ret = true;
+
+	ASSERT(XFS_DQ_IS_LOCKED(dqp));
+
+	/*
+	 * If flushlock isn't locked anymore, chances are that the
+	 * inode flush completed and the inode was taken off the AIL.
+	 * So, just get out.
+	 */
+	if (completion_done(&dqp->q_flush) ||
+	    !(lip->li_flags & XFS_LI_IN_AIL)) {
+		xfs_dqunlock(dqp);
+		return true;
+	}
+
+	bp = xfs_incore(dqp->q_mount->m_ddev_targp, qlip->qli_format.qlf_blkno,
+			dqp->q_mount->m_quotainfo->qi_dqchunklen, XBF_TRYLOCK);
+	xfs_dqunlock(dqp);
+	if (!bp)
+		return true;
+	if (XFS_BUF_ISDELAYWRITE(bp))
+		xfs_buf_delwri_promote(bp);
+	if (XFS_BUF_ISPINNED(bp))
+		ret = false;
+	xfs_buf_relse(bp);
+	return ret;
+}
+
+/*
+ * This is called to attempt to lock the dquot associated with this
+ * dquot log item.  Don't sleep on the dquot lock or the flush lock.
+ * If the flush lock is already held, indicating that the dquot has
+ * been or is in the process of being flushed, then see if we can
+ * find the dquot's buffer in the buffer cache without sleeping.  If
+ * we can and it is marked delayed write, then we want to send it out.
+ * We delay doing so until the push routine, though, to avoid sleeping
+ * in any device strategy routines.
+ */
+STATIC uint
+xfs_qm_dquot_logitem_trylock(
+	struct xfs_log_item	*lip)
+{
+	struct xfs_dquot	*dqp = DQUOT_ITEM(lip)->qli_dquot;
+
+	if (atomic_read(&dqp->q_pincount) > 0)
+		return XFS_ITEM_PINNED;
+
+	if (!xfs_qm_dqlock_nowait(dqp))
+		return XFS_ITEM_LOCKED;
+
+	if (!xfs_dqflock_nowait(dqp)) {
+		/*
+		 * dquot has already been flushed to the backing buffer,
+		 * leave it locked, pushbuf routine will unlock it.
+		 */
+		return XFS_ITEM_PUSHBUF;
+	}
+
+	ASSERT(lip->li_flags & XFS_LI_IN_AIL);
+	return XFS_ITEM_SUCCESS;
+}
+
+/*
+ * Unlock the dquot associated with the log item.
+ * Clear the fields of the dquot and dquot log item that
+ * are specific to the current transaction.  If the
+ * hold flags is set, do not unlock the dquot.
+ */
+STATIC void
+xfs_qm_dquot_logitem_unlock(
+	struct xfs_log_item	*lip)
+{
+	struct xfs_dquot	*dqp = DQUOT_ITEM(lip)->qli_dquot;
+
+	ASSERT(XFS_DQ_IS_LOCKED(dqp));
+
+	/*
+	 * Clear the transaction pointer in the dquot
+	 */
+	dqp->q_transp = NULL;
+
+	/*
+	 * dquots are never 'held' from getting unlocked at the end of
+	 * a transaction.  Their locking and unlocking is hidden inside the
+	 * transaction layer, within trans_commit. Hence, no LI_HOLD flag
+	 * for the logitem.
+	 */
+	xfs_dqunlock(dqp);
+}
+
+/*
+ * this needs to stamp an lsn into the dquot, I think.
+ * rpc's that look at user dquot's would then have to
+ * push on the dependency recorded in the dquot
+ */
+STATIC void
+xfs_qm_dquot_logitem_committing(
+	struct xfs_log_item	*lip,
+	xfs_lsn_t		lsn)
+{
+}
+
+/*
+ * This is the ops vector for dquots
+ */
+static struct xfs_item_ops xfs_dquot_item_ops = {
+	.iop_size	= xfs_qm_dquot_logitem_size,
+	.iop_format	= xfs_qm_dquot_logitem_format,
+	.iop_pin	= xfs_qm_dquot_logitem_pin,
+	.iop_unpin	= xfs_qm_dquot_logitem_unpin,
+	.iop_trylock	= xfs_qm_dquot_logitem_trylock,
+	.iop_unlock	= xfs_qm_dquot_logitem_unlock,
+	.iop_committed	= xfs_qm_dquot_logitem_committed,
+	.iop_push	= xfs_qm_dquot_logitem_push,
+	.iop_pushbuf	= xfs_qm_dquot_logitem_pushbuf,
+	.iop_committing = xfs_qm_dquot_logitem_committing
+};
+
+/*
+ * Initialize the dquot log item for a newly allocated dquot.
+ * The dquot isn't locked at this point, but it isn't on any of the lists
+ * either, so we don't care.
+ */
+void
+xfs_qm_dquot_logitem_init(
+	struct xfs_dquot	*dqp)
+{
+	struct xfs_dq_logitem	*lp = &dqp->q_logitem;
+
+	xfs_log_item_init(dqp->q_mount, &lp->qli_item, XFS_LI_DQUOT,
+					&xfs_dquot_item_ops);
+	lp->qli_dquot = dqp;
+	lp->qli_format.qlf_type = XFS_LI_DQUOT;
+	lp->qli_format.qlf_id = be32_to_cpu(dqp->q_core.d_id);
+	lp->qli_format.qlf_blkno = dqp->q_blkno;
+	lp->qli_format.qlf_len = 1;
+	/*
+	 * This is just the offset of this dquot within its buffer
+	 * (which is currently 1 FSB and probably won't change).
+	 * Hence 32 bits for this offset should be just fine.
+	 * Alternatively, we can store (bufoffset / sizeof(xfs_dqblk_t))
+	 * here, and recompute it at recovery time.
+	 */
+	lp->qli_format.qlf_boffset = (__uint32_t)dqp->q_bufoffset;
+}
+
+/*------------------  QUOTAOFF LOG ITEMS  -------------------*/
+
+static inline struct xfs_qoff_logitem *QOFF_ITEM(struct xfs_log_item *lip)
+{
+	return container_of(lip, struct xfs_qoff_logitem, qql_item);
+}
+
+
+/*
+ * This returns the number of iovecs needed to log the given quotaoff item.
+ * We only need 1 iovec for an quotaoff item.  It just logs the
+ * quotaoff_log_format structure.
+ */
+STATIC uint
+xfs_qm_qoff_logitem_size(
+	struct xfs_log_item	*lip)
+{
+	return 1;
+}
+
+/*
+ * This is called to fill in the vector of log iovecs for the
+ * given quotaoff log item. We use only 1 iovec, and we point that
+ * at the quotaoff_log_format structure embedded in the quotaoff item.
+ * It is at this point that we assert that all of the extent
+ * slots in the quotaoff item have been filled.
+ */
+STATIC void
+xfs_qm_qoff_logitem_format(
+	struct xfs_log_item	*lip,
+	struct xfs_log_iovec	*log_vector)
+{
+	struct xfs_qoff_logitem	*qflip = QOFF_ITEM(lip);
+
+	ASSERT(qflip->qql_format.qf_type == XFS_LI_QUOTAOFF);
+
+	log_vector->i_addr = &qflip->qql_format;
+	log_vector->i_len = sizeof(xfs_qoff_logitem_t);
+	log_vector->i_type = XLOG_REG_TYPE_QUOTAOFF;
+	qflip->qql_format.qf_size = 1;
+}
+
+/*
+ * Pinning has no meaning for an quotaoff item, so just return.
+ */
+STATIC void
+xfs_qm_qoff_logitem_pin(
+	struct xfs_log_item	*lip)
+{
+}
+
+/*
+ * Since pinning has no meaning for an quotaoff item, unpinning does
+ * not either.
+ */
+STATIC void
+xfs_qm_qoff_logitem_unpin(
+	struct xfs_log_item	*lip,
+	int			remove)
+{
+}
+
+/*
+ * Quotaoff items have no locking, so just return success.
+ */
+STATIC uint
+xfs_qm_qoff_logitem_trylock(
+	struct xfs_log_item	*lip)
+{
+	return XFS_ITEM_LOCKED;
+}
+
+/*
+ * Quotaoff items have no locking or pushing, so return failure
+ * so that the caller doesn't bother with us.
+ */
+STATIC void
+xfs_qm_qoff_logitem_unlock(
+	struct xfs_log_item	*lip)
+{
+}
+
+/*
+ * The quotaoff-start-item is logged only once and cannot be moved in the log,
+ * so simply return the lsn at which it's been logged.
+ */
+STATIC xfs_lsn_t
+xfs_qm_qoff_logitem_committed(
+	struct xfs_log_item	*lip,
+	xfs_lsn_t		lsn)
+{
+	return lsn;
+}
+
+/*
+ * There isn't much you can do to push on an quotaoff item.  It is simply
+ * stuck waiting for the log to be flushed to disk.
+ */
+STATIC void
+xfs_qm_qoff_logitem_push(
+	struct xfs_log_item	*lip)
+{
+}
+
+
+STATIC xfs_lsn_t
+xfs_qm_qoffend_logitem_committed(
+	struct xfs_log_item	*lip,
+	xfs_lsn_t		lsn)
+{
+	struct xfs_qoff_logitem	*qfe = QOFF_ITEM(lip);
+	struct xfs_qoff_logitem	*qfs = qfe->qql_start_lip;
+	struct xfs_ail		*ailp = qfs->qql_item.li_ailp;
+
+	/*
+	 * Delete the qoff-start logitem from the AIL.
+	 * xfs_trans_ail_delete() drops the AIL lock.
+	 */
+	spin_lock(&ailp->xa_lock);
+	xfs_trans_ail_delete(ailp, (xfs_log_item_t *)qfs);
+
+	kmem_free(qfs);
+	kmem_free(qfe);
+	return (xfs_lsn_t)-1;
+}
+
+/*
+ * XXX rcc - don't know quite what to do with this.  I think we can
+ * just ignore it.  The only time that isn't the case is if we allow
+ * the client to somehow see that quotas have been turned off in which
+ * we can't allow that to get back until the quotaoff hits the disk.
+ * So how would that happen?  Also, do we need different routines for
+ * quotaoff start and quotaoff end?  I suspect the answer is yes but
+ * to be sure, I need to look at the recovery code and see how quota off
+ * recovery is handled (do we roll forward or back or do something else).
+ * If we roll forwards or backwards, then we need two separate routines,
+ * one that does nothing and one that stamps in the lsn that matters
+ * (truly makes the quotaoff irrevocable).  If we do something else,
+ * then maybe we don't need two.
+ */
+STATIC void
+xfs_qm_qoff_logitem_committing(
+	struct xfs_log_item	*lip,
+	xfs_lsn_t		commit_lsn)
+{
+}
+
+static struct xfs_item_ops xfs_qm_qoffend_logitem_ops = {
+	.iop_size	= xfs_qm_qoff_logitem_size,
+	.iop_format	= xfs_qm_qoff_logitem_format,
+	.iop_pin	= xfs_qm_qoff_logitem_pin,
+	.iop_unpin	= xfs_qm_qoff_logitem_unpin,
+	.iop_trylock	= xfs_qm_qoff_logitem_trylock,
+	.iop_unlock	= xfs_qm_qoff_logitem_unlock,
+	.iop_committed	= xfs_qm_qoffend_logitem_committed,
+	.iop_push	= xfs_qm_qoff_logitem_push,
+	.iop_committing = xfs_qm_qoff_logitem_committing
+};
+
+/*
+ * This is the ops vector shared by all quotaoff-start log items.
+ */
+static struct xfs_item_ops xfs_qm_qoff_logitem_ops = {
+	.iop_size	= xfs_qm_qoff_logitem_size,
+	.iop_format	= xfs_qm_qoff_logitem_format,
+	.iop_pin	= xfs_qm_qoff_logitem_pin,
+	.iop_unpin	= xfs_qm_qoff_logitem_unpin,
+	.iop_trylock	= xfs_qm_qoff_logitem_trylock,
+	.iop_unlock	= xfs_qm_qoff_logitem_unlock,
+	.iop_committed	= xfs_qm_qoff_logitem_committed,
+	.iop_push	= xfs_qm_qoff_logitem_push,
+	.iop_committing = xfs_qm_qoff_logitem_committing
+};
+
+/*
+ * Allocate and initialize an quotaoff item of the correct quota type(s).
+ */
+struct xfs_qoff_logitem *
+xfs_qm_qoff_logitem_init(
+	struct xfs_mount	*mp,
+	struct xfs_qoff_logitem	*start,
+	uint			flags)
+{
+	struct xfs_qoff_logitem	*qf;
+
+	qf = kmem_zalloc(sizeof(struct xfs_qoff_logitem), KM_SLEEP);
+
+	xfs_log_item_init(mp, &qf->qql_item, XFS_LI_QUOTAOFF, start ?
+			&xfs_qm_qoffend_logitem_ops : &xfs_qm_qoff_logitem_ops);
+	qf->qql_item.li_mountp = mp;
+	qf->qql_format.qf_type = XFS_LI_QUOTAOFF;
+	qf->qql_format.qf_flags = flags;
+	qf->qql_start_lip = start;
+	return qf;
+}
diff --git a/fs/xfs/quota/xfs_dquot_item.h b/fs/xfs/quota/xfs_dquot_item.h
new file mode 100644
index 00000000..5acae2ad
--- /dev/null
+++ b/fs/xfs/quota/xfs_dquot_item.h
@@ -0,0 +1,48 @@
+/*
+ * Copyright (c) 2000-2003 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_DQUOT_ITEM_H__
+#define __XFS_DQUOT_ITEM_H__
+
+struct xfs_dquot;
+struct xfs_trans;
+struct xfs_mount;
+struct xfs_qoff_logitem;
+
+typedef struct xfs_dq_logitem {
+	xfs_log_item_t		 qli_item;	   /* common portion */
+	struct xfs_dquot	*qli_dquot;	   /* dquot ptr */
+	xfs_lsn_t		 qli_flush_lsn;	   /* lsn at last flush */
+	xfs_dq_logformat_t	 qli_format;	   /* logged structure */
+} xfs_dq_logitem_t;
+
+typedef struct xfs_qoff_logitem {
+	xfs_log_item_t		 qql_item;	/* common portion */
+	struct xfs_qoff_logitem *qql_start_lip; /* qoff-start logitem, if any */
+	xfs_qoff_logformat_t	 qql_format;	/* logged structure */
+} xfs_qoff_logitem_t;
+
+
+extern void		   xfs_qm_dquot_logitem_init(struct xfs_dquot *);
+extern xfs_qoff_logitem_t *xfs_qm_qoff_logitem_init(struct xfs_mount *,
+					struct xfs_qoff_logitem *, uint);
+extern xfs_qoff_logitem_t *xfs_trans_get_qoff_item(struct xfs_trans *,
+					struct xfs_qoff_logitem *, uint);
+extern void		   xfs_trans_log_quotaoff_item(struct xfs_trans *,
+					struct xfs_qoff_logitem *);
+
+#endif	/* __XFS_DQUOT_ITEM_H__ */
diff --git a/fs/xfs/quota/xfs_qm.c b/fs/xfs/quota/xfs_qm.c
new file mode 100644
index 00000000..e70c7fc9
--- /dev/null
+++ b/fs/xfs/quota/xfs_qm.c
@@ -0,0 +1,2462 @@
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_alloc.h"
+#include "xfs_quota.h"
+#include "xfs_mount.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_ialloc_btree.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_ialloc.h"
+#include "xfs_itable.h"
+#include "xfs_rtalloc.h"
+#include "xfs_error.h"
+#include "xfs_bmap.h"
+#include "xfs_attr.h"
+#include "xfs_buf_item.h"
+#include "xfs_trans_space.h"
+#include "xfs_utils.h"
+#include "xfs_qm.h"
+#include "xfs_trace.h"
+
+/*
+ * The global quota manager. There is only one of these for the entire
+ * system, _not_ one per file system. XQM keeps track of the overall
+ * quota functionality, including maintaining the freelist and hash
+ * tables of dquots.
+ */
+struct mutex	xfs_Gqm_lock;
+struct xfs_qm	*xfs_Gqm;
+uint		ndquot;
+
+kmem_zone_t	*qm_dqzone;
+kmem_zone_t	*qm_dqtrxzone;
+
+STATIC void	xfs_qm_list_init(xfs_dqlist_t *, char *, int);
+STATIC void	xfs_qm_list_destroy(xfs_dqlist_t *);
+
+STATIC int	xfs_qm_init_quotainos(xfs_mount_t *);
+STATIC int	xfs_qm_init_quotainfo(xfs_mount_t *);
+STATIC int	xfs_qm_shake(struct shrinker *, struct shrink_control *);
+
+static struct shrinker xfs_qm_shaker = {
+	.shrink = xfs_qm_shake,
+	.seeks = DEFAULT_SEEKS,
+};
+
+#ifdef DEBUG
+extern struct mutex	qcheck_lock;
+#endif
+
+#ifdef QUOTADEBUG
+static void
+xfs_qm_dquot_list_print(
+	struct xfs_mount *mp)
+{
+	xfs_dquot_t	*dqp;
+	int		i = 0;
+
+	list_for_each_entry(dqp, &mp->m_quotainfo->qi_dqlist_lock, qi_mplist) {
+		xfs_debug(mp, "   %d. \"%d (%s)\"   "
+				  "bcnt = %lld, icnt = %lld, refs = %d",
+			i++, be32_to_cpu(dqp->q_core.d_id),
+			DQFLAGTO_TYPESTR(dqp),
+			(long long)be64_to_cpu(dqp->q_core.d_bcount),
+			(long long)be64_to_cpu(dqp->q_core.d_icount),
+			dqp->q_nrefs);
+	}
+}
+#else
+static void xfs_qm_dquot_list_print(struct xfs_mount *mp) { }
+#endif
+
+/*
+ * Initialize the XQM structure.
+ * Note that there is not one quota manager per file system.
+ */
+STATIC struct xfs_qm *
+xfs_Gqm_init(void)
+{
+	xfs_dqhash_t	*udqhash, *gdqhash;
+	xfs_qm_t	*xqm;
+	size_t		hsize;
+	uint		i;
+
+	/*
+	 * Initialize the dquot hash tables.
+	 */
+	udqhash = kmem_zalloc_greedy(&hsize,
+				     XFS_QM_HASHSIZE_LOW * sizeof(xfs_dqhash_t),
+				     XFS_QM_HASHSIZE_HIGH * sizeof(xfs_dqhash_t));
+	if (!udqhash)
+		goto out;
+
+	gdqhash = kmem_zalloc_large(hsize);
+	if (!gdqhash)
+		goto out_free_udqhash;
+
+	hsize /= sizeof(xfs_dqhash_t);
+	ndquot = hsize << 8;
+
+	xqm = kmem_zalloc(sizeof(xfs_qm_t), KM_SLEEP);
+	xqm->qm_dqhashmask = hsize - 1;
+	xqm->qm_usr_dqhtable = udqhash;
+	xqm->qm_grp_dqhtable = gdqhash;
+	ASSERT(xqm->qm_usr_dqhtable != NULL);
+	ASSERT(xqm->qm_grp_dqhtable != NULL);
+
+	for (i = 0; i < hsize; i++) {
+		xfs_qm_list_init(&(xqm->qm_usr_dqhtable[i]), "uxdqh", i);
+		xfs_qm_list_init(&(xqm->qm_grp_dqhtable[i]), "gxdqh", i);
+	}
+
+	/*
+	 * Freelist of all dquots of all file systems
+	 */
+	INIT_LIST_HEAD(&xqm->qm_dqfrlist);
+	xqm->qm_dqfrlist_cnt = 0;
+	mutex_init(&xqm->qm_dqfrlist_lock);
+
+	/*
+	 * dquot zone. we register our own low-memory callback.
+	 */
+	if (!qm_dqzone) {
+		xqm->qm_dqzone = kmem_zone_init(sizeof(xfs_dquot_t),
+						"xfs_dquots");
+		qm_dqzone = xqm->qm_dqzone;
+	} else
+		xqm->qm_dqzone = qm_dqzone;
+
+	register_shrinker(&xfs_qm_shaker);
+
+	/*
+	 * The t_dqinfo portion of transactions.
+	 */
+	if (!qm_dqtrxzone) {
+		xqm->qm_dqtrxzone = kmem_zone_init(sizeof(xfs_dquot_acct_t),
+						   "xfs_dqtrx");
+		qm_dqtrxzone = xqm->qm_dqtrxzone;
+	} else
+		xqm->qm_dqtrxzone = qm_dqtrxzone;
+
+	atomic_set(&xqm->qm_totaldquots, 0);
+	xqm->qm_dqfree_ratio = XFS_QM_DQFREE_RATIO;
+	xqm->qm_nrefs = 0;
+#ifdef DEBUG
+	mutex_init(&qcheck_lock);
+#endif
+	return xqm;
+
+ out_free_udqhash:
+	kmem_free_large(udqhash);
+ out:
+	return NULL;
+}
+
+/*
+ * Destroy the global quota manager when its reference count goes to zero.
+ */
+STATIC void
+xfs_qm_destroy(
+	struct xfs_qm	*xqm)
+{
+	struct xfs_dquot *dqp, *n;
+	int		hsize, i;
+
+	ASSERT(xqm != NULL);
+	ASSERT(xqm->qm_nrefs == 0);
+	unregister_shrinker(&xfs_qm_shaker);
+	hsize = xqm->qm_dqhashmask + 1;
+	for (i = 0; i < hsize; i++) {
+		xfs_qm_list_destroy(&(xqm->qm_usr_dqhtable[i]));
+		xfs_qm_list_destroy(&(xqm->qm_grp_dqhtable[i]));
+	}
+	kmem_free_large(xqm->qm_usr_dqhtable);
+	kmem_free_large(xqm->qm_grp_dqhtable);
+	xqm->qm_usr_dqhtable = NULL;
+	xqm->qm_grp_dqhtable = NULL;
+	xqm->qm_dqhashmask = 0;
+
+	/* frlist cleanup */
+	mutex_lock(&xqm->qm_dqfrlist_lock);
+	list_for_each_entry_safe(dqp, n, &xqm->qm_dqfrlist, q_freelist) {
+		xfs_dqlock(dqp);
+#ifdef QUOTADEBUG
+		xfs_debug(dqp->q_mount, "FREELIST destroy 0x%p", dqp);
+#endif
+		list_del_init(&dqp->q_freelist);
+		xfs_Gqm->qm_dqfrlist_cnt--;
+		xfs_dqunlock(dqp);
+		xfs_qm_dqdestroy(dqp);
+	}
+	mutex_unlock(&xqm->qm_dqfrlist_lock);
+	mutex_destroy(&xqm->qm_dqfrlist_lock);
+#ifdef DEBUG
+	mutex_destroy(&qcheck_lock);
+#endif
+	kmem_free(xqm);
+}
+
+/*
+ * Called at mount time to let XQM know that another file system is
+ * starting quotas. This isn't crucial information as the individual mount
+ * structures are pretty independent, but it helps the XQM keep a
+ * global view of what's going on.
+ */
+/* ARGSUSED */
+STATIC int
+xfs_qm_hold_quotafs_ref(
+	struct xfs_mount *mp)
+{
+	/*
+	 * Need to lock the xfs_Gqm structure for things like this. For example,
+	 * the structure could disappear between the entry to this routine and
+	 * a HOLD operation if not locked.
+	 */
+	mutex_lock(&xfs_Gqm_lock);
+
+	if (!xfs_Gqm) {
+		xfs_Gqm = xfs_Gqm_init();
+		if (!xfs_Gqm) {
+			mutex_unlock(&xfs_Gqm_lock);
+			return ENOMEM;
+		}
+	}
+
+	/*
+	 * We can keep a list of all filesystems with quotas mounted for
+	 * debugging and statistical purposes, but ...
+	 * Just take a reference and get out.
+	 */
+	xfs_Gqm->qm_nrefs++;
+	mutex_unlock(&xfs_Gqm_lock);
+
+	return 0;
+}
+
+
+/*
+ * Release the reference that a filesystem took at mount time,
+ * so that we know when we need to destroy the entire quota manager.
+ */
+/* ARGSUSED */
+STATIC void
+xfs_qm_rele_quotafs_ref(
+	struct xfs_mount *mp)
+{
+	xfs_dquot_t	*dqp, *n;
+
+	ASSERT(xfs_Gqm);
+	ASSERT(xfs_Gqm->qm_nrefs > 0);
+
+	/*
+	 * Go thru the freelist and destroy all inactive dquots.
+	 */
+	mutex_lock(&xfs_Gqm->qm_dqfrlist_lock);
+
+	list_for_each_entry_safe(dqp, n, &xfs_Gqm->qm_dqfrlist, q_freelist) {
+		xfs_dqlock(dqp);
+		if (dqp->dq_flags & XFS_DQ_INACTIVE) {
+			ASSERT(dqp->q_mount == NULL);
+			ASSERT(! XFS_DQ_IS_DIRTY(dqp));
+			ASSERT(list_empty(&dqp->q_hashlist));
+			ASSERT(list_empty(&dqp->q_mplist));
+			list_del_init(&dqp->q_freelist);
+			xfs_Gqm->qm_dqfrlist_cnt--;
+			xfs_dqunlock(dqp);
+			xfs_qm_dqdestroy(dqp);
+		} else {
+			xfs_dqunlock(dqp);
+		}
+	}
+	mutex_unlock(&xfs_Gqm->qm_dqfrlist_lock);
+
+	/*
+	 * Destroy the entire XQM. If somebody mounts with quotaon, this'll
+	 * be restarted.
+	 */
+	mutex_lock(&xfs_Gqm_lock);
+	if (--xfs_Gqm->qm_nrefs == 0) {
+		xfs_qm_destroy(xfs_Gqm);
+		xfs_Gqm = NULL;
+	}
+	mutex_unlock(&xfs_Gqm_lock);
+}
+
+/*
+ * Just destroy the quotainfo structure.
+ */
+void
+xfs_qm_unmount(
+	struct xfs_mount	*mp)
+{
+	if (mp->m_quotainfo) {
+		xfs_qm_dqpurge_all(mp, XFS_QMOPT_QUOTALL);
+		xfs_qm_destroy_quotainfo(mp);
+	}
+}
+
+
+/*
+ * This is called from xfs_mountfs to start quotas and initialize all
+ * necessary data structures like quotainfo.  This is also responsible for
+ * running a quotacheck as necessary.  We are guaranteed that the superblock
+ * is consistently read in at this point.
+ *
+ * If we fail here, the mount will continue with quota turned off. We don't
+ * need to inidicate success or failure at all.
+ */
+void
+xfs_qm_mount_quotas(
+	xfs_mount_t	*mp)
+{
+	int		error = 0;
+	uint		sbf;
+
+	/*
+	 * If quotas on realtime volumes is not supported, we disable
+	 * quotas immediately.
+	 */
+	if (mp->m_sb.sb_rextents) {
+		xfs_notice(mp, "Cannot turn on quotas for realtime filesystem");
+		mp->m_qflags = 0;
+		goto write_changes;
+	}
+
+	ASSERT(XFS_IS_QUOTA_RUNNING(mp));
+
+	/*
+	 * Allocate the quotainfo structure inside the mount struct, and
+	 * create quotainode(s), and change/rev superblock if necessary.
+	 */
+	error = xfs_qm_init_quotainfo(mp);
+	if (error) {
+		/*
+		 * We must turn off quotas.
+		 */
+		ASSERT(mp->m_quotainfo == NULL);
+		mp->m_qflags = 0;
+		goto write_changes;
+	}
+	/*
+	 * If any of the quotas are not consistent, do a quotacheck.
+	 */
+	if (XFS_QM_NEED_QUOTACHECK(mp)) {
+		error = xfs_qm_quotacheck(mp);
+		if (error) {
+			/* Quotacheck failed and disabled quotas. */
+			return;
+		}
+	}
+	/* 
+	 * If one type of quotas is off, then it will lose its
+	 * quotachecked status, since we won't be doing accounting for
+	 * that type anymore.
+	 */
+	if (!XFS_IS_UQUOTA_ON(mp))
+		mp->m_qflags &= ~XFS_UQUOTA_CHKD;
+	if (!(XFS_IS_GQUOTA_ON(mp) || XFS_IS_PQUOTA_ON(mp)))
+		mp->m_qflags &= ~XFS_OQUOTA_CHKD;
+
+ write_changes:
+	/*
+	 * We actually don't have to acquire the m_sb_lock at all.
+	 * This can only be called from mount, and that's single threaded. XXX
+	 */
+	spin_lock(&mp->m_sb_lock);
+	sbf = mp->m_sb.sb_qflags;
+	mp->m_sb.sb_qflags = mp->m_qflags & XFS_MOUNT_QUOTA_ALL;
+	spin_unlock(&mp->m_sb_lock);
+
+	if (sbf != (mp->m_qflags & XFS_MOUNT_QUOTA_ALL)) {
+		if (xfs_qm_write_sb_changes(mp, XFS_SB_QFLAGS)) {
+			/*
+			 * We could only have been turning quotas off.
+			 * We aren't in very good shape actually because
+			 * the incore structures are convinced that quotas are
+			 * off, but the on disk superblock doesn't know that !
+			 */
+			ASSERT(!(XFS_IS_QUOTA_RUNNING(mp)));
+			xfs_alert(mp, "%s: Superblock update failed!",
+				__func__);
+		}
+	}
+
+	if (error) {
+		xfs_warn(mp, "Failed to initialize disk quotas.");
+		return;
+	}
+
+#ifdef QUOTADEBUG
+	if (XFS_IS_QUOTA_ON(mp))
+		xfs_qm_internalqcheck(mp);
+#endif
+}
+
+/*
+ * Called from the vfsops layer.
+ */
+void
+xfs_qm_unmount_quotas(
+	xfs_mount_t	*mp)
+{
+	/*
+	 * Release the dquots that root inode, et al might be holding,
+	 * before we flush quotas and blow away the quotainfo structure.
+	 */
+	ASSERT(mp->m_rootip);
+	xfs_qm_dqdetach(mp->m_rootip);
+	if (mp->m_rbmip)
+		xfs_qm_dqdetach(mp->m_rbmip);
+	if (mp->m_rsumip)
+		xfs_qm_dqdetach(mp->m_rsumip);
+
+	/*
+	 * Release the quota inodes.
+	 */
+	if (mp->m_quotainfo) {
+		if (mp->m_quotainfo->qi_uquotaip) {
+			IRELE(mp->m_quotainfo->qi_uquotaip);
+			mp->m_quotainfo->qi_uquotaip = NULL;
+		}
+		if (mp->m_quotainfo->qi_gquotaip) {
+			IRELE(mp->m_quotainfo->qi_gquotaip);
+			mp->m_quotainfo->qi_gquotaip = NULL;
+		}
+	}
+}
+
+/*
+ * Flush all dquots of the given file system to disk. The dquots are
+ * _not_ purged from memory here, just their data written to disk.
+ */
+STATIC int
+xfs_qm_dqflush_all(
+	struct xfs_mount	*mp,
+	int			sync_mode)
+{
+	struct xfs_quotainfo	*q = mp->m_quotainfo;
+	int			recl;
+	struct xfs_dquot	*dqp;
+	int			error;
+
+	if (!q)
+		return 0;
+again:
+	mutex_lock(&q->qi_dqlist_lock);
+	list_for_each_entry(dqp, &q->qi_dqlist, q_mplist) {
+		xfs_dqlock(dqp);
+		if (! XFS_DQ_IS_DIRTY(dqp)) {
+			xfs_dqunlock(dqp);
+			continue;
+		}
+
+		/* XXX a sentinel would be better */
+		recl = q->qi_dqreclaims;
+		if (!xfs_dqflock_nowait(dqp)) {
+			/*
+			 * If we can't grab the flush lock then check
+			 * to see if the dquot has been flushed delayed
+			 * write.  If so, grab its buffer and send it
+			 * out immediately.  We'll be able to acquire
+			 * the flush lock when the I/O completes.
+			 */
+			xfs_qm_dqflock_pushbuf_wait(dqp);
+		}
+		/*
+		 * Let go of the mplist lock. We don't want to hold it
+		 * across a disk write.
+		 */
+		mutex_unlock(&q->qi_dqlist_lock);
+		error = xfs_qm_dqflush(dqp, sync_mode);
+		xfs_dqunlock(dqp);
+		if (error)
+			return error;
+
+		mutex_lock(&q->qi_dqlist_lock);
+		if (recl != q->qi_dqreclaims) {
+			mutex_unlock(&q->qi_dqlist_lock);
+			/* XXX restart limit */
+			goto again;
+		}
+	}
+
+	mutex_unlock(&q->qi_dqlist_lock);
+	/* return ! busy */
+	return 0;
+}
+/*
+ * Release the group dquot pointers the user dquots may be
+ * carrying around as a hint. mplist is locked on entry and exit.
+ */
+STATIC void
+xfs_qm_detach_gdquots(
+	struct xfs_mount	*mp)
+{
+	struct xfs_quotainfo	*q = mp->m_quotainfo;
+	struct xfs_dquot	*dqp, *gdqp;
+	int			nrecl;
+
+ again:
+	ASSERT(mutex_is_locked(&q->qi_dqlist_lock));
+	list_for_each_entry(dqp, &q->qi_dqlist, q_mplist) {
+		xfs_dqlock(dqp);
+		if ((gdqp = dqp->q_gdquot)) {
+			xfs_dqlock(gdqp);
+			dqp->q_gdquot = NULL;
+		}
+		xfs_dqunlock(dqp);
+
+		if (gdqp) {
+			/*
+			 * Can't hold the mplist lock across a dqput.
+			 * XXXmust convert to marker based iterations here.
+			 */
+			nrecl = q->qi_dqreclaims;
+			mutex_unlock(&q->qi_dqlist_lock);
+			xfs_qm_dqput(gdqp);
+
+			mutex_lock(&q->qi_dqlist_lock);
+			if (nrecl != q->qi_dqreclaims)
+				goto again;
+		}
+	}
+}
+
+/*
+ * Go through all the incore dquots of this file system and take them
+ * off the mplist and hashlist, if the dquot type matches the dqtype
+ * parameter. This is used when turning off quota accounting for
+ * users and/or groups, as well as when the filesystem is unmounting.
+ */
+STATIC int
+xfs_qm_dqpurge_int(
+	struct xfs_mount	*mp,
+	uint			flags)
+{
+	struct xfs_quotainfo	*q = mp->m_quotainfo;
+	struct xfs_dquot	*dqp, *n;
+	uint			dqtype;
+	int			nrecl;
+	int			nmisses;
+
+	if (!q)
+		return 0;
+
+	dqtype = (flags & XFS_QMOPT_UQUOTA) ? XFS_DQ_USER : 0;
+	dqtype |= (flags & XFS_QMOPT_PQUOTA) ? XFS_DQ_PROJ : 0;
+	dqtype |= (flags & XFS_QMOPT_GQUOTA) ? XFS_DQ_GROUP : 0;
+
+	mutex_lock(&q->qi_dqlist_lock);
+
+	/*
+	 * In the first pass through all incore dquots of this filesystem,
+	 * we release the group dquot pointers the user dquots may be
+	 * carrying around as a hint. We need to do this irrespective of
+	 * what's being turned off.
+	 */
+	xfs_qm_detach_gdquots(mp);
+
+      again:
+	nmisses = 0;
+	ASSERT(mutex_is_locked(&q->qi_dqlist_lock));
+	/*
+	 * Try to get rid of all of the unwanted dquots. The idea is to
+	 * get them off mplist and hashlist, but leave them on freelist.
+	 */
+	list_for_each_entry_safe(dqp, n, &q->qi_dqlist, q_mplist) {
+		/*
+		 * It's OK to look at the type without taking dqlock here.
+		 * We're holding the mplist lock here, and that's needed for
+		 * a dqreclaim.
+		 */
+		if ((dqp->dq_flags & dqtype) == 0)
+			continue;
+
+		if (!mutex_trylock(&dqp->q_hash->qh_lock)) {
+			nrecl = q->qi_dqreclaims;
+			mutex_unlock(&q->qi_dqlist_lock);
+			mutex_lock(&dqp->q_hash->qh_lock);
+			mutex_lock(&q->qi_dqlist_lock);
+
+			/*
+			 * XXXTheoretically, we can get into a very long
+			 * ping pong game here.
+			 * No one can be adding dquots to the mplist at
+			 * this point, but somebody might be taking things off.
+			 */
+			if (nrecl != q->qi_dqreclaims) {
+				mutex_unlock(&dqp->q_hash->qh_lock);
+				goto again;
+			}
+		}
+
+		/*
+		 * Take the dquot off the mplist and hashlist. It may remain on
+		 * freelist in INACTIVE state.
+		 */
+		nmisses += xfs_qm_dqpurge(dqp);
+	}
+	mutex_unlock(&q->qi_dqlist_lock);
+	return nmisses;
+}
+
+int
+xfs_qm_dqpurge_all(
+	xfs_mount_t	*mp,
+	uint		flags)
+{
+	int		ndquots;
+
+	/*
+	 * Purge the dquot cache.
+	 * None of the dquots should really be busy at this point.
+	 */
+	if (mp->m_quotainfo) {
+		while ((ndquots = xfs_qm_dqpurge_int(mp, flags))) {
+			delay(ndquots * 10);
+		}
+	}
+	return 0;
+}
+
+STATIC int
+xfs_qm_dqattach_one(
+	xfs_inode_t	*ip,
+	xfs_dqid_t	id,
+	uint		type,
+	uint		doalloc,
+	xfs_dquot_t	*udqhint, /* hint */
+	xfs_dquot_t	**IO_idqpp)
+{
+	xfs_dquot_t	*dqp;
+	int		error;
+
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
+	error = 0;
+
+	/*
+	 * See if we already have it in the inode itself. IO_idqpp is
+	 * &i_udquot or &i_gdquot. This made the code look weird, but
+	 * made the logic a lot simpler.
+	 */
+	dqp = *IO_idqpp;
+	if (dqp) {
+		trace_xfs_dqattach_found(dqp);
+		return 0;
+	}
+
+	/*
+	 * udqhint is the i_udquot field in inode, and is non-NULL only
+	 * when the type arg is group/project. Its purpose is to save a
+	 * lookup by dqid (xfs_qm_dqget) by caching a group dquot inside
+	 * the user dquot.
+	 */
+	if (udqhint) {
+		ASSERT(type == XFS_DQ_GROUP || type == XFS_DQ_PROJ);
+		xfs_dqlock(udqhint);
+
+		/*
+		 * No need to take dqlock to look at the id.
+		 *
+		 * The ID can't change until it gets reclaimed, and it won't
+		 * be reclaimed as long as we have a ref from inode and we
+		 * hold the ilock.
+		 */
+		dqp = udqhint->q_gdquot;
+		if (dqp && be32_to_cpu(dqp->q_core.d_id) == id) {
+			xfs_dqlock(dqp);
+			XFS_DQHOLD(dqp);
+			ASSERT(*IO_idqpp == NULL);
+			*IO_idqpp = dqp;
+
+			xfs_dqunlock(dqp);
+			xfs_dqunlock(udqhint);
+			return 0;
+		}
+
+		/*
+		 * We can't hold a dquot lock when we call the dqget code.
+		 * We'll deadlock in no time, because of (not conforming to)
+		 * lock ordering - the inodelock comes before any dquot lock,
+		 * and we may drop and reacquire the ilock in xfs_qm_dqget().
+		 */
+		xfs_dqunlock(udqhint);
+	}
+
+	/*
+	 * Find the dquot from somewhere. This bumps the
+	 * reference count of dquot and returns it locked.
+	 * This can return ENOENT if dquot didn't exist on
+	 * disk and we didn't ask it to allocate;
+	 * ESRCH if quotas got turned off suddenly.
+	 */
+	error = xfs_qm_dqget(ip->i_mount, ip, id, type,
+			     doalloc | XFS_QMOPT_DOWARN, &dqp);
+	if (error)
+		return error;
+
+	trace_xfs_dqattach_get(dqp);
+
+	/*
+	 * dqget may have dropped and re-acquired the ilock, but it guarantees
+	 * that the dquot returned is the one that should go in the inode.
+	 */
+	*IO_idqpp = dqp;
+	xfs_dqunlock(dqp);
+	return 0;
+}
+
+
+/*
+ * Given a udquot and gdquot, attach a ptr to the group dquot in the
+ * udquot as a hint for future lookups. The idea sounds simple, but the
+ * execution isn't, because the udquot might have a group dquot attached
+ * already and getting rid of that gets us into lock ordering constraints.
+ * The process is complicated more by the fact that the dquots may or may not
+ * be locked on entry.
+ */
+STATIC void
+xfs_qm_dqattach_grouphint(
+	xfs_dquot_t	*udq,
+	xfs_dquot_t	*gdq)
+{
+	xfs_dquot_t	*tmp;
+
+	xfs_dqlock(udq);
+
+	if ((tmp = udq->q_gdquot)) {
+		if (tmp == gdq) {
+			xfs_dqunlock(udq);
+			return;
+		}
+
+		udq->q_gdquot = NULL;
+		/*
+		 * We can't keep any dqlocks when calling dqrele,
+		 * because the freelist lock comes before dqlocks.
+		 */
+		xfs_dqunlock(udq);
+		/*
+		 * we took a hard reference once upon a time in dqget,
+		 * so give it back when the udquot no longer points at it
+		 * dqput() does the unlocking of the dquot.
+		 */
+		xfs_qm_dqrele(tmp);
+
+		xfs_dqlock(udq);
+		xfs_dqlock(gdq);
+
+	} else {
+		ASSERT(XFS_DQ_IS_LOCKED(udq));
+		xfs_dqlock(gdq);
+	}
+
+	ASSERT(XFS_DQ_IS_LOCKED(udq));
+	ASSERT(XFS_DQ_IS_LOCKED(gdq));
+	/*
+	 * Somebody could have attached a gdquot here,
+	 * when we dropped the uqlock. If so, just do nothing.
+	 */
+	if (udq->q_gdquot == NULL) {
+		XFS_DQHOLD(gdq);
+		udq->q_gdquot = gdq;
+	}
+
+	xfs_dqunlock(gdq);
+	xfs_dqunlock(udq);
+}
+
+
+/*
+ * Given a locked inode, attach dquot(s) to it, taking U/G/P-QUOTAON
+ * into account.
+ * If XFS_QMOPT_DQALLOC, the dquot(s) will be allocated if needed.
+ * Inode may get unlocked and relocked in here, and the caller must deal with
+ * the consequences.
+ */
+int
+xfs_qm_dqattach_locked(
+	xfs_inode_t	*ip,
+	uint		flags)
+{
+	xfs_mount_t	*mp = ip->i_mount;
+	uint		nquotas = 0;
+	int		error = 0;
+
+	if (!XFS_IS_QUOTA_RUNNING(mp) ||
+	    !XFS_IS_QUOTA_ON(mp) ||
+	    !XFS_NOT_DQATTACHED(mp, ip) ||
+	    ip->i_ino == mp->m_sb.sb_uquotino ||
+	    ip->i_ino == mp->m_sb.sb_gquotino)
+		return 0;
+
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
+
+	if (XFS_IS_UQUOTA_ON(mp)) {
+		error = xfs_qm_dqattach_one(ip, ip->i_d.di_uid, XFS_DQ_USER,
+						flags & XFS_QMOPT_DQALLOC,
+						NULL, &ip->i_udquot);
+		if (error)
+			goto done;
+		nquotas++;
+	}
+
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
+	if (XFS_IS_OQUOTA_ON(mp)) {
+		error = XFS_IS_GQUOTA_ON(mp) ?
+			xfs_qm_dqattach_one(ip, ip->i_d.di_gid, XFS_DQ_GROUP,
+						flags & XFS_QMOPT_DQALLOC,
+						ip->i_udquot, &ip->i_gdquot) :
+			xfs_qm_dqattach_one(ip, xfs_get_projid(ip), XFS_DQ_PROJ,
+						flags & XFS_QMOPT_DQALLOC,
+						ip->i_udquot, &ip->i_gdquot);
+		/*
+		 * Don't worry about the udquot that we may have
+		 * attached above. It'll get detached, if not already.
+		 */
+		if (error)
+			goto done;
+		nquotas++;
+	}
+
+	/*
+	 * Attach this group quota to the user quota as a hint.
+	 * This WON'T, in general, result in a thrash.
+	 */
+	if (nquotas == 2) {
+		ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
+		ASSERT(ip->i_udquot);
+		ASSERT(ip->i_gdquot);
+
+		/*
+		 * We may or may not have the i_udquot locked at this point,
+		 * but this check is OK since we don't depend on the i_gdquot to
+		 * be accurate 100% all the time. It is just a hint, and this
+		 * will succeed in general.
+		 */
+		if (ip->i_udquot->q_gdquot == ip->i_gdquot)
+			goto done;
+		/*
+		 * Attach i_gdquot to the gdquot hint inside the i_udquot.
+		 */
+		xfs_qm_dqattach_grouphint(ip->i_udquot, ip->i_gdquot);
+	}
+
+ done:
+#ifdef QUOTADEBUG
+	if (! error) {
+		if (XFS_IS_UQUOTA_ON(mp))
+			ASSERT(ip->i_udquot);
+		if (XFS_IS_OQUOTA_ON(mp))
+			ASSERT(ip->i_gdquot);
+	}
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
+#endif
+	return error;
+}
+
+int
+xfs_qm_dqattach(
+	struct xfs_inode	*ip,
+	uint			flags)
+{
+	int			error;
+
+	xfs_ilock(ip, XFS_ILOCK_EXCL);
+	error = xfs_qm_dqattach_locked(ip, flags);
+	xfs_iunlock(ip, XFS_ILOCK_EXCL);
+
+	return error;
+}
+
+/*
+ * Release dquots (and their references) if any.
+ * The inode should be locked EXCL except when this's called by
+ * xfs_ireclaim.
+ */
+void
+xfs_qm_dqdetach(
+	xfs_inode_t	*ip)
+{
+	if (!(ip->i_udquot || ip->i_gdquot))
+		return;
+
+	trace_xfs_dquot_dqdetach(ip);
+
+	ASSERT(ip->i_ino != ip->i_mount->m_sb.sb_uquotino);
+	ASSERT(ip->i_ino != ip->i_mount->m_sb.sb_gquotino);
+	if (ip->i_udquot) {
+		xfs_qm_dqrele(ip->i_udquot);
+		ip->i_udquot = NULL;
+	}
+	if (ip->i_gdquot) {
+		xfs_qm_dqrele(ip->i_gdquot);
+		ip->i_gdquot = NULL;
+	}
+}
+
+int
+xfs_qm_sync(
+	struct xfs_mount	*mp,
+	int			flags)
+{
+	struct xfs_quotainfo	*q = mp->m_quotainfo;
+	int			recl, restarts;
+	struct xfs_dquot	*dqp;
+	int			error;
+
+	if (!XFS_IS_QUOTA_RUNNING(mp) || !XFS_IS_QUOTA_ON(mp))
+		return 0;
+
+	restarts = 0;
+
+  again:
+	mutex_lock(&q->qi_dqlist_lock);
+	/*
+	 * dqpurge_all() also takes the mplist lock and iterate thru all dquots
+	 * in quotaoff. However, if the QUOTA_ACTIVE bits are not cleared
+	 * when we have the mplist lock, we know that dquots will be consistent
+	 * as long as we have it locked.
+	 */
+	if (!XFS_IS_QUOTA_ON(mp)) {
+		mutex_unlock(&q->qi_dqlist_lock);
+		return 0;
+	}
+	ASSERT(mutex_is_locked(&q->qi_dqlist_lock));
+	list_for_each_entry(dqp, &q->qi_dqlist, q_mplist) {
+		/*
+		 * If this is vfs_sync calling, then skip the dquots that
+		 * don't 'seem' to be dirty. ie. don't acquire dqlock.
+		 * This is very similar to what xfs_sync does with inodes.
+		 */
+		if (flags & SYNC_TRYLOCK) {
+			if (!XFS_DQ_IS_DIRTY(dqp))
+				continue;
+			if (!xfs_qm_dqlock_nowait(dqp))
+				continue;
+		} else {
+			xfs_dqlock(dqp);
+		}
+
+		/*
+		 * Now, find out for sure if this dquot is dirty or not.
+		 */
+		if (! XFS_DQ_IS_DIRTY(dqp)) {
+			xfs_dqunlock(dqp);
+			continue;
+		}
+
+		/* XXX a sentinel would be better */
+		recl = q->qi_dqreclaims;
+		if (!xfs_dqflock_nowait(dqp)) {
+			if (flags & SYNC_TRYLOCK) {
+				xfs_dqunlock(dqp);
+				continue;
+			}
+			/*
+			 * If we can't grab the flush lock then if the caller
+			 * really wanted us to give this our best shot, so
+			 * see if we can give a push to the buffer before we wait
+			 * on the flush lock. At this point, we know that
+			 * even though the dquot is being flushed,
+			 * it has (new) dirty data.
+			 */
+			xfs_qm_dqflock_pushbuf_wait(dqp);
+		}
+		/*
+		 * Let go of the mplist lock. We don't want to hold it
+		 * across a disk write
+		 */
+		mutex_unlock(&q->qi_dqlist_lock);
+		error = xfs_qm_dqflush(dqp, flags);
+		xfs_dqunlock(dqp);
+		if (error && XFS_FORCED_SHUTDOWN(mp))
+			return 0;	/* Need to prevent umount failure */
+		else if (error)
+			return error;
+
+		mutex_lock(&q->qi_dqlist_lock);
+		if (recl != q->qi_dqreclaims) {
+			if (++restarts >= XFS_QM_SYNC_MAX_RESTARTS)
+				break;
+
+			mutex_unlock(&q->qi_dqlist_lock);
+			goto again;
+		}
+	}
+
+	mutex_unlock(&q->qi_dqlist_lock);
+	return 0;
+}
+
+/*
+ * The hash chains and the mplist use the same xfs_dqhash structure as
+ * their list head, but we can take the mplist qh_lock and one of the
+ * hash qh_locks at the same time without any problem as they aren't
+ * related.
+ */
+static struct lock_class_key xfs_quota_mplist_class;
+
+/*
+ * This initializes all the quota information that's kept in the
+ * mount structure
+ */
+STATIC int
+xfs_qm_init_quotainfo(
+	xfs_mount_t	*mp)
+{
+	xfs_quotainfo_t *qinf;
+	int		error;
+	xfs_dquot_t	*dqp;
+
+	ASSERT(XFS_IS_QUOTA_RUNNING(mp));
+
+	/*
+	 * Tell XQM that we exist as soon as possible.
+	 */
+	if ((error = xfs_qm_hold_quotafs_ref(mp))) {
+		return error;
+	}
+
+	qinf = mp->m_quotainfo = kmem_zalloc(sizeof(xfs_quotainfo_t), KM_SLEEP);
+
+	/*
+	 * See if quotainodes are setup, and if not, allocate them,
+	 * and change the superblock accordingly.
+	 */
+	if ((error = xfs_qm_init_quotainos(mp))) {
+		kmem_free(qinf);
+		mp->m_quotainfo = NULL;
+		return error;
+	}
+
+	INIT_LIST_HEAD(&qinf->qi_dqlist);
+	mutex_init(&qinf->qi_dqlist_lock);
+	lockdep_set_class(&qinf->qi_dqlist_lock, &xfs_quota_mplist_class);
+
+	qinf->qi_dqreclaims = 0;
+
+	/* mutex used to serialize quotaoffs */
+	mutex_init(&qinf->qi_quotaofflock);
+
+	/* Precalc some constants */
+	qinf->qi_dqchunklen = XFS_FSB_TO_BB(mp, XFS_DQUOT_CLUSTER_SIZE_FSB);
+	ASSERT(qinf->qi_dqchunklen);
+	qinf->qi_dqperchunk = BBTOB(qinf->qi_dqchunklen);
+	do_div(qinf->qi_dqperchunk, sizeof(xfs_dqblk_t));
+
+	mp->m_qflags |= (mp->m_sb.sb_qflags & XFS_ALL_QUOTA_CHKD);
+
+	/*
+	 * We try to get the limits from the superuser's limits fields.
+	 * This is quite hacky, but it is standard quota practice.
+	 * We look at the USR dquot with id == 0 first, but if user quotas
+	 * are not enabled we goto the GRP dquot with id == 0.
+	 * We don't really care to keep separate default limits for user
+	 * and group quotas, at least not at this point.
+	 */
+	error = xfs_qm_dqget(mp, NULL, (xfs_dqid_t)0,
+			     XFS_IS_UQUOTA_RUNNING(mp) ? XFS_DQ_USER : 
+			     (XFS_IS_GQUOTA_RUNNING(mp) ? XFS_DQ_GROUP :
+				XFS_DQ_PROJ),
+			     XFS_QMOPT_DQSUSER|XFS_QMOPT_DOWARN,
+			     &dqp);
+	if (! error) {
+		xfs_disk_dquot_t	*ddqp = &dqp->q_core;
+
+		/*
+		 * The warnings and timers set the grace period given to
+		 * a user or group before he or she can not perform any
+		 * more writing. If it is zero, a default is used.
+		 */
+		qinf->qi_btimelimit = ddqp->d_btimer ?
+			be32_to_cpu(ddqp->d_btimer) : XFS_QM_BTIMELIMIT;
+		qinf->qi_itimelimit = ddqp->d_itimer ?
+			be32_to_cpu(ddqp->d_itimer) : XFS_QM_ITIMELIMIT;
+		qinf->qi_rtbtimelimit = ddqp->d_rtbtimer ?
+			be32_to_cpu(ddqp->d_rtbtimer) : XFS_QM_RTBTIMELIMIT;
+		qinf->qi_bwarnlimit = ddqp->d_bwarns ?
+			be16_to_cpu(ddqp->d_bwarns) : XFS_QM_BWARNLIMIT;
+		qinf->qi_iwarnlimit = ddqp->d_iwarns ?
+			be16_to_cpu(ddqp->d_iwarns) : XFS_QM_IWARNLIMIT;
+		qinf->qi_rtbwarnlimit = ddqp->d_rtbwarns ?
+			be16_to_cpu(ddqp->d_rtbwarns) : XFS_QM_RTBWARNLIMIT;
+		qinf->qi_bhardlimit = be64_to_cpu(ddqp->d_blk_hardlimit);
+		qinf->qi_bsoftlimit = be64_to_cpu(ddqp->d_blk_softlimit);
+		qinf->qi_ihardlimit = be64_to_cpu(ddqp->d_ino_hardlimit);
+		qinf->qi_isoftlimit = be64_to_cpu(ddqp->d_ino_softlimit);
+		qinf->qi_rtbhardlimit = be64_to_cpu(ddqp->d_rtb_hardlimit);
+		qinf->qi_rtbsoftlimit = be64_to_cpu(ddqp->d_rtb_softlimit);
+ 
+		/*
+		 * We sent the XFS_QMOPT_DQSUSER flag to dqget because
+		 * we don't want this dquot cached. We haven't done a
+		 * quotacheck yet, and quotacheck doesn't like incore dquots.
+		 */
+		xfs_qm_dqdestroy(dqp);
+	} else {
+		qinf->qi_btimelimit = XFS_QM_BTIMELIMIT;
+		qinf->qi_itimelimit = XFS_QM_ITIMELIMIT;
+		qinf->qi_rtbtimelimit = XFS_QM_RTBTIMELIMIT;
+		qinf->qi_bwarnlimit = XFS_QM_BWARNLIMIT;
+		qinf->qi_iwarnlimit = XFS_QM_IWARNLIMIT;
+		qinf->qi_rtbwarnlimit = XFS_QM_RTBWARNLIMIT;
+	}
+
+	return 0;
+}
+
+
+/*
+ * Gets called when unmounting a filesystem or when all quotas get
+ * turned off.
+ * This purges the quota inodes, destroys locks and frees itself.
+ */
+void
+xfs_qm_destroy_quotainfo(
+	xfs_mount_t	*mp)
+{
+	xfs_quotainfo_t *qi;
+
+	qi = mp->m_quotainfo;
+	ASSERT(qi != NULL);
+	ASSERT(xfs_Gqm != NULL);
+
+	/*
+	 * Release the reference that XQM kept, so that we know
+	 * when the XQM structure should be freed. We cannot assume
+	 * that xfs_Gqm is non-null after this point.
+	 */
+	xfs_qm_rele_quotafs_ref(mp);
+
+	ASSERT(list_empty(&qi->qi_dqlist));
+	mutex_destroy(&qi->qi_dqlist_lock);
+
+	if (qi->qi_uquotaip) {
+		IRELE(qi->qi_uquotaip);
+		qi->qi_uquotaip = NULL; /* paranoia */
+	}
+	if (qi->qi_gquotaip) {
+		IRELE(qi->qi_gquotaip);
+		qi->qi_gquotaip = NULL;
+	}
+	mutex_destroy(&qi->qi_quotaofflock);
+	kmem_free(qi);
+	mp->m_quotainfo = NULL;
+}
+
+
+
+/* ------------------- PRIVATE STATIC FUNCTIONS ----------------------- */
+
+/* ARGSUSED */
+STATIC void
+xfs_qm_list_init(
+	xfs_dqlist_t	*list,
+	char		*str,
+	int		n)
+{
+	mutex_init(&list->qh_lock);
+	INIT_LIST_HEAD(&list->qh_list);
+	list->qh_version = 0;
+	list->qh_nelems = 0;
+}
+
+STATIC void
+xfs_qm_list_destroy(
+	xfs_dqlist_t	*list)
+{
+	mutex_destroy(&(list->qh_lock));
+}
+
+/*
+ * Create an inode and return with a reference already taken, but unlocked
+ * This is how we create quota inodes
+ */
+STATIC int
+xfs_qm_qino_alloc(
+	xfs_mount_t	*mp,
+	xfs_inode_t	**ip,
+	__int64_t	sbfields,
+	uint		flags)
+{
+	xfs_trans_t	*tp;
+	int		error;
+	int		committed;
+
+	tp = xfs_trans_alloc(mp, XFS_TRANS_QM_QINOCREATE);
+	if ((error = xfs_trans_reserve(tp,
+				      XFS_QM_QINOCREATE_SPACE_RES(mp),
+				      XFS_CREATE_LOG_RES(mp), 0,
+				      XFS_TRANS_PERM_LOG_RES,
+				      XFS_CREATE_LOG_COUNT))) {
+		xfs_trans_cancel(tp, 0);
+		return error;
+	}
+
+	error = xfs_dir_ialloc(&tp, NULL, S_IFREG, 1, 0, 0, 1, ip, &committed);
+	if (error) {
+		xfs_trans_cancel(tp, XFS_TRANS_RELEASE_LOG_RES |
+				 XFS_TRANS_ABORT);
+		return error;
+	}
+
+	/*
+	 * Make the changes in the superblock, and log those too.
+	 * sbfields arg may contain fields other than *QUOTINO;
+	 * VERSIONNUM for example.
+	 */
+	spin_lock(&mp->m_sb_lock);
+	if (flags & XFS_QMOPT_SBVERSION) {
+		ASSERT(!xfs_sb_version_hasquota(&mp->m_sb));
+		ASSERT((sbfields & (XFS_SB_VERSIONNUM | XFS_SB_UQUOTINO |
+				   XFS_SB_GQUOTINO | XFS_SB_QFLAGS)) ==
+		       (XFS_SB_VERSIONNUM | XFS_SB_UQUOTINO |
+			XFS_SB_GQUOTINO | XFS_SB_QFLAGS));
+
+		xfs_sb_version_addquota(&mp->m_sb);
+		mp->m_sb.sb_uquotino = NULLFSINO;
+		mp->m_sb.sb_gquotino = NULLFSINO;
+
+		/* qflags will get updated _after_ quotacheck */
+		mp->m_sb.sb_qflags = 0;
+	}
+	if (flags & XFS_QMOPT_UQUOTA)
+		mp->m_sb.sb_uquotino = (*ip)->i_ino;
+	else
+		mp->m_sb.sb_gquotino = (*ip)->i_ino;
+	spin_unlock(&mp->m_sb_lock);
+	xfs_mod_sb(tp, sbfields);
+
+	if ((error = xfs_trans_commit(tp, XFS_TRANS_RELEASE_LOG_RES))) {
+		xfs_alert(mp, "%s failed (error %d)!", __func__, error);
+		return error;
+	}
+	return 0;
+}
+
+
+STATIC void
+xfs_qm_reset_dqcounts(
+	xfs_mount_t	*mp,
+	xfs_buf_t	*bp,
+	xfs_dqid_t	id,
+	uint		type)
+{
+	xfs_disk_dquot_t	*ddq;
+	int			j;
+
+	trace_xfs_reset_dqcounts(bp, _RET_IP_);
+
+	/*
+	 * Reset all counters and timers. They'll be
+	 * started afresh by xfs_qm_quotacheck.
+	 */
+#ifdef DEBUG
+	j = XFS_FSB_TO_B(mp, XFS_DQUOT_CLUSTER_SIZE_FSB);
+	do_div(j, sizeof(xfs_dqblk_t));
+	ASSERT(mp->m_quotainfo->qi_dqperchunk == j);
+#endif
+	ddq = (xfs_disk_dquot_t *)XFS_BUF_PTR(bp);
+	for (j = 0; j < mp->m_quotainfo->qi_dqperchunk; j++) {
+		/*
+		 * Do a sanity check, and if needed, repair the dqblk. Don't
+		 * output any warnings because it's perfectly possible to
+		 * find uninitialised dquot blks. See comment in xfs_qm_dqcheck.
+		 */
+		(void) xfs_qm_dqcheck(mp, ddq, id+j, type, XFS_QMOPT_DQREPAIR,
+				      "xfs_quotacheck");
+		ddq->d_bcount = 0;
+		ddq->d_icount = 0;
+		ddq->d_rtbcount = 0;
+		ddq->d_btimer = 0;
+		ddq->d_itimer = 0;
+		ddq->d_rtbtimer = 0;
+		ddq->d_bwarns = 0;
+		ddq->d_iwarns = 0;
+		ddq->d_rtbwarns = 0;
+		ddq = (xfs_disk_dquot_t *) ((xfs_dqblk_t *)ddq + 1);
+	}
+}
+
+STATIC int
+xfs_qm_dqiter_bufs(
+	xfs_mount_t	*mp,
+	xfs_dqid_t	firstid,
+	xfs_fsblock_t	bno,
+	xfs_filblks_t	blkcnt,
+	uint		flags)
+{
+	xfs_buf_t	*bp;
+	int		error;
+	int		type;
+
+	ASSERT(blkcnt > 0);
+	type = flags & XFS_QMOPT_UQUOTA ? XFS_DQ_USER :
+		(flags & XFS_QMOPT_PQUOTA ? XFS_DQ_PROJ : XFS_DQ_GROUP);
+	error = 0;
+
+	/*
+	 * Blkcnt arg can be a very big number, and might even be
+	 * larger than the log itself. So, we have to break it up into
+	 * manageable-sized transactions.
+	 * Note that we don't start a permanent transaction here; we might
+	 * not be able to get a log reservation for the whole thing up front,
+	 * and we don't really care to either, because we just discard
+	 * everything if we were to crash in the middle of this loop.
+	 */
+	while (blkcnt--) {
+		error = xfs_trans_read_buf(mp, NULL, mp->m_ddev_targp,
+			      XFS_FSB_TO_DADDR(mp, bno),
+			      mp->m_quotainfo->qi_dqchunklen, 0, &bp);
+		if (error)
+			break;
+
+		xfs_qm_reset_dqcounts(mp, bp, firstid, type);
+		xfs_bdwrite(mp, bp);
+		/*
+		 * goto the next block.
+		 */
+		bno++;
+		firstid += mp->m_quotainfo->qi_dqperchunk;
+	}
+	return error;
+}
+
+/*
+ * Iterate over all allocated USR/GRP/PRJ dquots in the system, calling a
+ * caller supplied function for every chunk of dquots that we find.
+ */
+STATIC int
+xfs_qm_dqiterate(
+	xfs_mount_t	*mp,
+	xfs_inode_t	*qip,
+	uint		flags)
+{
+	xfs_bmbt_irec_t		*map;
+	int			i, nmaps;	/* number of map entries */
+	int			error;		/* return value */
+	xfs_fileoff_t		lblkno;
+	xfs_filblks_t		maxlblkcnt;
+	xfs_dqid_t		firstid;
+	xfs_fsblock_t		rablkno;
+	xfs_filblks_t		rablkcnt;
+
+	error = 0;
+	/*
+	 * This looks racy, but we can't keep an inode lock across a
+	 * trans_reserve. But, this gets called during quotacheck, and that
+	 * happens only at mount time which is single threaded.
+	 */
+	if (qip->i_d.di_nblocks == 0)
+		return 0;
+
+	map = kmem_alloc(XFS_DQITER_MAP_SIZE * sizeof(*map), KM_SLEEP);
+
+	lblkno = 0;
+	maxlblkcnt = XFS_B_TO_FSB(mp, (xfs_ufsize_t)XFS_MAXIOFFSET(mp));
+	do {
+		nmaps = XFS_DQITER_MAP_SIZE;
+		/*
+		 * We aren't changing the inode itself. Just changing
+		 * some of its data. No new blocks are added here, and
+		 * the inode is never added to the transaction.
+		 */
+		xfs_ilock(qip, XFS_ILOCK_SHARED);
+		error = xfs_bmapi(NULL, qip, lblkno,
+				  maxlblkcnt - lblkno,
+				  XFS_BMAPI_METADATA,
+				  NULL,
+				  0, map, &nmaps, NULL);
+		xfs_iunlock(qip, XFS_ILOCK_SHARED);
+		if (error)
+			break;
+
+		ASSERT(nmaps <= XFS_DQITER_MAP_SIZE);
+		for (i = 0; i < nmaps; i++) {
+			ASSERT(map[i].br_startblock != DELAYSTARTBLOCK);
+			ASSERT(map[i].br_blockcount);
+
+
+			lblkno += map[i].br_blockcount;
+
+			if (map[i].br_startblock == HOLESTARTBLOCK)
+				continue;
+
+			firstid = (xfs_dqid_t) map[i].br_startoff *
+				mp->m_quotainfo->qi_dqperchunk;
+			/*
+			 * Do a read-ahead on the next extent.
+			 */
+			if ((i+1 < nmaps) &&
+			    (map[i+1].br_startblock != HOLESTARTBLOCK)) {
+				rablkcnt =  map[i+1].br_blockcount;
+				rablkno = map[i+1].br_startblock;
+				while (rablkcnt--) {
+					xfs_buf_readahead(mp->m_ddev_targp,
+					       XFS_FSB_TO_DADDR(mp, rablkno),
+					       mp->m_quotainfo->qi_dqchunklen);
+					rablkno++;
+				}
+			}
+			/*
+			 * Iterate thru all the blks in the extent and
+			 * reset the counters of all the dquots inside them.
+			 */
+			if ((error = xfs_qm_dqiter_bufs(mp,
+						       firstid,
+						       map[i].br_startblock,
+						       map[i].br_blockcount,
+						       flags))) {
+				break;
+			}
+		}
+
+		if (error)
+			break;
+	} while (nmaps > 0);
+
+	kmem_free(map);
+
+	return error;
+}
+
+/*
+ * Called by dqusage_adjust in doing a quotacheck.
+ *
+ * Given the inode, and a dquot id this updates both the incore dqout as well
+ * as the buffer copy. This is so that once the quotacheck is done, we can
+ * just log all the buffers, as opposed to logging numerous updates to
+ * individual dquots.
+ */
+STATIC int
+xfs_qm_quotacheck_dqadjust(
+	struct xfs_inode	*ip,
+	xfs_dqid_t		id,
+	uint			type,
+	xfs_qcnt_t		nblks,
+	xfs_qcnt_t		rtblks)
+{
+	struct xfs_mount	*mp = ip->i_mount;
+	struct xfs_dquot	*dqp;
+	int			error;
+
+	error = xfs_qm_dqget(mp, ip, id, type,
+			     XFS_QMOPT_DQALLOC | XFS_QMOPT_DOWARN, &dqp);
+	if (error) {
+		/*
+		 * Shouldn't be able to turn off quotas here.
+		 */
+		ASSERT(error != ESRCH);
+		ASSERT(error != ENOENT);
+		return error;
+	}
+
+	trace_xfs_dqadjust(dqp);
+
+	/*
+	 * Adjust the inode count and the block count to reflect this inode's
+	 * resource usage.
+	 */
+	be64_add_cpu(&dqp->q_core.d_icount, 1);
+	dqp->q_res_icount++;
+	if (nblks) {
+		be64_add_cpu(&dqp->q_core.d_bcount, nblks);
+		dqp->q_res_bcount += nblks;
+	}
+	if (rtblks) {
+		be64_add_cpu(&dqp->q_core.d_rtbcount, rtblks);
+		dqp->q_res_rtbcount += rtblks;
+	}
+
+	/*
+	 * Set default limits, adjust timers (since we changed usages)
+	 *
+	 * There are no timers for the default values set in the root dquot.
+	 */
+	if (dqp->q_core.d_id) {
+		xfs_qm_adjust_dqlimits(mp, &dqp->q_core);
+		xfs_qm_adjust_dqtimers(mp, &dqp->q_core);
+	}
+
+	dqp->dq_flags |= XFS_DQ_DIRTY;
+	xfs_qm_dqput(dqp);
+	return 0;
+}
+
+STATIC int
+xfs_qm_get_rtblks(
+	xfs_inode_t	*ip,
+	xfs_qcnt_t	*O_rtblks)
+{
+	xfs_filblks_t	rtblks;			/* total rt blks */
+	xfs_extnum_t	idx;			/* extent record index */
+	xfs_ifork_t	*ifp;			/* inode fork pointer */
+	xfs_extnum_t	nextents;		/* number of extent entries */
+	int		error;
+
+	ASSERT(XFS_IS_REALTIME_INODE(ip));
+	ifp = XFS_IFORK_PTR(ip, XFS_DATA_FORK);
+	if (!(ifp->if_flags & XFS_IFEXTENTS)) {
+		if ((error = xfs_iread_extents(NULL, ip, XFS_DATA_FORK)))
+			return error;
+	}
+	rtblks = 0;
+	nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
+	for (idx = 0; idx < nextents; idx++)
+		rtblks += xfs_bmbt_get_blockcount(xfs_iext_get_ext(ifp, idx));
+	*O_rtblks = (xfs_qcnt_t)rtblks;
+	return 0;
+}
+
+/*
+ * callback routine supplied to bulkstat(). Given an inumber, find its
+ * dquots and update them to account for resources taken by that inode.
+ */
+/* ARGSUSED */
+STATIC int
+xfs_qm_dqusage_adjust(
+	xfs_mount_t	*mp,		/* mount point for filesystem */
+	xfs_ino_t	ino,		/* inode number to get data for */
+	void		__user *buffer,	/* not used */
+	int		ubsize,		/* not used */
+	int		*ubused,	/* not used */
+	int		*res)		/* result code value */
+{
+	xfs_inode_t	*ip;
+	xfs_qcnt_t	nblks, rtblks = 0;
+	int		error;
+
+	ASSERT(XFS_IS_QUOTA_RUNNING(mp));
+
+	/*
+	 * rootino must have its resources accounted for, not so with the quota
+	 * inodes.
+	 */
+	if (ino == mp->m_sb.sb_uquotino || ino == mp->m_sb.sb_gquotino) {
+		*res = BULKSTAT_RV_NOTHING;
+		return XFS_ERROR(EINVAL);
+	}
+
+	/*
+	 * We don't _need_ to take the ilock EXCL. However, the xfs_qm_dqget
+	 * interface expects the inode to be exclusively locked because that's
+	 * the case in all other instances. It's OK that we do this because
+	 * quotacheck is done only at mount time.
+	 */
+	error = xfs_iget(mp, NULL, ino, 0, XFS_ILOCK_EXCL, &ip);
+	if (error) {
+		*res = BULKSTAT_RV_NOTHING;
+		return error;
+	}
+
+	ASSERT(ip->i_delayed_blks == 0);
+
+	if (XFS_IS_REALTIME_INODE(ip)) {
+		/*
+		 * Walk thru the extent list and count the realtime blocks.
+		 */
+		error = xfs_qm_get_rtblks(ip, &rtblks);
+		if (error)
+			goto error0;
+	}
+
+	nblks = (xfs_qcnt_t)ip->i_d.di_nblocks - rtblks;
+
+	/*
+	 * Add the (disk blocks and inode) resources occupied by this
+	 * inode to its dquots. We do this adjustment in the incore dquot,
+	 * and also copy the changes to its buffer.
+	 * We don't care about putting these changes in a transaction
+	 * envelope because if we crash in the middle of a 'quotacheck'
+	 * we have to start from the beginning anyway.
+	 * Once we're done, we'll log all the dquot bufs.
+	 *
+	 * The *QUOTA_ON checks below may look pretty racy, but quotachecks
+	 * and quotaoffs don't race. (Quotachecks happen at mount time only).
+	 */
+	if (XFS_IS_UQUOTA_ON(mp)) {
+		error = xfs_qm_quotacheck_dqadjust(ip, ip->i_d.di_uid,
+						   XFS_DQ_USER, nblks, rtblks);
+		if (error)
+			goto error0;
+	}
+
+	if (XFS_IS_GQUOTA_ON(mp)) {
+		error = xfs_qm_quotacheck_dqadjust(ip, ip->i_d.di_gid,
+						   XFS_DQ_GROUP, nblks, rtblks);
+		if (error)
+			goto error0;
+	}
+
+	if (XFS_IS_PQUOTA_ON(mp)) {
+		error = xfs_qm_quotacheck_dqadjust(ip, xfs_get_projid(ip),
+						   XFS_DQ_PROJ, nblks, rtblks);
+		if (error)
+			goto error0;
+	}
+
+	xfs_iunlock(ip, XFS_ILOCK_EXCL);
+	IRELE(ip);
+	*res = BULKSTAT_RV_DIDONE;
+	return 0;
+
+error0:
+	xfs_iunlock(ip, XFS_ILOCK_EXCL);
+	IRELE(ip);
+	*res = BULKSTAT_RV_GIVEUP;
+	return error;
+}
+
+/*
+ * Walk thru all the filesystem inodes and construct a consistent view
+ * of the disk quota world. If the quotacheck fails, disable quotas.
+ */
+int
+xfs_qm_quotacheck(
+	xfs_mount_t	*mp)
+{
+	int		done, count, error;
+	xfs_ino_t	lastino;
+	size_t		structsz;
+	xfs_inode_t	*uip, *gip;
+	uint		flags;
+
+	count = INT_MAX;
+	structsz = 1;
+	lastino = 0;
+	flags = 0;
+
+	ASSERT(mp->m_quotainfo->qi_uquotaip || mp->m_quotainfo->qi_gquotaip);
+	ASSERT(XFS_IS_QUOTA_RUNNING(mp));
+
+	/*
+	 * There should be no cached dquots. The (simplistic) quotacheck
+	 * algorithm doesn't like that.
+	 */
+	ASSERT(list_empty(&mp->m_quotainfo->qi_dqlist));
+
+	xfs_notice(mp, "Quotacheck needed: Please wait.");
+
+	/*
+	 * First we go thru all the dquots on disk, USR and GRP/PRJ, and reset
+	 * their counters to zero. We need a clean slate.
+	 * We don't log our changes till later.
+	 */
+	uip = mp->m_quotainfo->qi_uquotaip;
+	if (uip) {
+		error = xfs_qm_dqiterate(mp, uip, XFS_QMOPT_UQUOTA);
+		if (error)
+			goto error_return;
+		flags |= XFS_UQUOTA_CHKD;
+	}
+
+	gip = mp->m_quotainfo->qi_gquotaip;
+	if (gip) {
+		error = xfs_qm_dqiterate(mp, gip, XFS_IS_GQUOTA_ON(mp) ?
+					XFS_QMOPT_GQUOTA : XFS_QMOPT_PQUOTA);
+		if (error)
+			goto error_return;
+		flags |= XFS_OQUOTA_CHKD;
+	}
+
+	do {
+		/*
+		 * Iterate thru all the inodes in the file system,
+		 * adjusting the corresponding dquot counters in core.
+		 */
+		error = xfs_bulkstat(mp, &lastino, &count,
+				     xfs_qm_dqusage_adjust,
+				     structsz, NULL, &done);
+		if (error)
+			break;
+
+	} while (!done);
+
+	/*
+	 * We've made all the changes that we need to make incore.
+	 * Flush them down to disk buffers if everything was updated
+	 * successfully.
+	 */
+	if (!error)
+		error = xfs_qm_dqflush_all(mp, 0);
+
+	/*
+	 * We can get this error if we couldn't do a dquot allocation inside
+	 * xfs_qm_dqusage_adjust (via bulkstat). We don't care about the
+	 * dirty dquots that might be cached, we just want to get rid of them
+	 * and turn quotaoff. The dquots won't be attached to any of the inodes
+	 * at this point (because we intentionally didn't in dqget_noattach).
+	 */
+	if (error) {
+		xfs_qm_dqpurge_all(mp, XFS_QMOPT_QUOTALL);
+		goto error_return;
+	}
+
+	/*
+	 * We didn't log anything, because if we crashed, we'll have to
+	 * start the quotacheck from scratch anyway. However, we must make
+	 * sure that our dquot changes are secure before we put the
+	 * quotacheck'd stamp on the superblock. So, here we do a synchronous
+	 * flush.
+	 */
+	XFS_bflush(mp->m_ddev_targp);
+
+	/*
+	 * If one type of quotas is off, then it will lose its
+	 * quotachecked status, since we won't be doing accounting for
+	 * that type anymore.
+	 */
+	mp->m_qflags &= ~(XFS_OQUOTA_CHKD | XFS_UQUOTA_CHKD);
+	mp->m_qflags |= flags;
+
+	xfs_qm_dquot_list_print(mp);
+
+ error_return:
+	if (error) {
+		xfs_warn(mp,
+	"Quotacheck: Unsuccessful (Error %d): Disabling quotas.",
+			error);
+		/*
+		 * We must turn off quotas.
+		 */
+		ASSERT(mp->m_quotainfo != NULL);
+		ASSERT(xfs_Gqm != NULL);
+		xfs_qm_destroy_quotainfo(mp);
+		if (xfs_mount_reset_sbqflags(mp)) {
+			xfs_warn(mp,
+				"Quotacheck: Failed to reset quota flags.");
+		}
+	} else
+		xfs_notice(mp, "Quotacheck: Done.");
+	return (error);
+}
+
+/*
+ * This is called after the superblock has been read in and we're ready to
+ * iget the quota inodes.
+ */
+STATIC int
+xfs_qm_init_quotainos(
+	xfs_mount_t	*mp)
+{
+	xfs_inode_t	*uip, *gip;
+	int		error;
+	__int64_t	sbflags;
+	uint		flags;
+
+	ASSERT(mp->m_quotainfo);
+	uip = gip = NULL;
+	sbflags = 0;
+	flags = 0;
+
+	/*
+	 * Get the uquota and gquota inodes
+	 */
+	if (xfs_sb_version_hasquota(&mp->m_sb)) {
+		if (XFS_IS_UQUOTA_ON(mp) &&
+		    mp->m_sb.sb_uquotino != NULLFSINO) {
+			ASSERT(mp->m_sb.sb_uquotino > 0);
+			if ((error = xfs_iget(mp, NULL, mp->m_sb.sb_uquotino,
+					     0, 0, &uip)))
+				return XFS_ERROR(error);
+		}
+		if (XFS_IS_OQUOTA_ON(mp) &&
+		    mp->m_sb.sb_gquotino != NULLFSINO) {
+			ASSERT(mp->m_sb.sb_gquotino > 0);
+			if ((error = xfs_iget(mp, NULL, mp->m_sb.sb_gquotino,
+					     0, 0, &gip))) {
+				if (uip)
+					IRELE(uip);
+				return XFS_ERROR(error);
+			}
+		}
+	} else {
+		flags |= XFS_QMOPT_SBVERSION;
+		sbflags |= (XFS_SB_VERSIONNUM | XFS_SB_UQUOTINO |
+			    XFS_SB_GQUOTINO | XFS_SB_QFLAGS);
+	}
+
+	/*
+	 * Create the two inodes, if they don't exist already. The changes
+	 * made above will get added to a transaction and logged in one of
+	 * the qino_alloc calls below.  If the device is readonly,
+	 * temporarily switch to read-write to do this.
+	 */
+	if (XFS_IS_UQUOTA_ON(mp) && uip == NULL) {
+		if ((error = xfs_qm_qino_alloc(mp, &uip,
+					      sbflags | XFS_SB_UQUOTINO,
+					      flags | XFS_QMOPT_UQUOTA)))
+			return XFS_ERROR(error);
+
+		flags &= ~XFS_QMOPT_SBVERSION;
+	}
+	if (XFS_IS_OQUOTA_ON(mp) && gip == NULL) {
+		flags |= (XFS_IS_GQUOTA_ON(mp) ?
+				XFS_QMOPT_GQUOTA : XFS_QMOPT_PQUOTA);
+		error = xfs_qm_qino_alloc(mp, &gip,
+					  sbflags | XFS_SB_GQUOTINO, flags);
+		if (error) {
+			if (uip)
+				IRELE(uip);
+
+			return XFS_ERROR(error);
+		}
+	}
+
+	mp->m_quotainfo->qi_uquotaip = uip;
+	mp->m_quotainfo->qi_gquotaip = gip;
+
+	return 0;
+}
+
+
+
+/*
+ * Just pop the least recently used dquot off the freelist and
+ * recycle it. The returned dquot is locked.
+ */
+STATIC xfs_dquot_t *
+xfs_qm_dqreclaim_one(void)
+{
+	xfs_dquot_t	*dqpout;
+	xfs_dquot_t	*dqp;
+	int		restarts;
+	int		startagain;
+
+	restarts = 0;
+	dqpout = NULL;
+
+	/* lockorder: hashchainlock, freelistlock, mplistlock, dqlock, dqflock */
+again:
+	startagain = 0;
+	mutex_lock(&xfs_Gqm->qm_dqfrlist_lock);
+
+	list_for_each_entry(dqp, &xfs_Gqm->qm_dqfrlist, q_freelist) {
+		struct xfs_mount *mp = dqp->q_mount;
+		xfs_dqlock(dqp);
+
+		/*
+		 * We are racing with dqlookup here. Naturally we don't
+		 * want to reclaim a dquot that lookup wants. We release the
+		 * freelist lock and start over, so that lookup will grab
+		 * both the dquot and the freelistlock.
+		 */
+		if (dqp->dq_flags & XFS_DQ_WANT) {
+			ASSERT(! (dqp->dq_flags & XFS_DQ_INACTIVE));
+
+			trace_xfs_dqreclaim_want(dqp);
+			XQM_STATS_INC(xqmstats.xs_qm_dqwants);
+			restarts++;
+			startagain = 1;
+			goto dqunlock;
+		}
+
+		/*
+		 * If the dquot is inactive, we are assured that it is
+		 * not on the mplist or the hashlist, and that makes our
+		 * life easier.
+		 */
+		if (dqp->dq_flags & XFS_DQ_INACTIVE) {
+			ASSERT(mp == NULL);
+			ASSERT(! XFS_DQ_IS_DIRTY(dqp));
+			ASSERT(list_empty(&dqp->q_hashlist));
+			ASSERT(list_empty(&dqp->q_mplist));
+			list_del_init(&dqp->q_freelist);
+			xfs_Gqm->qm_dqfrlist_cnt--;
+			dqpout = dqp;
+			XQM_STATS_INC(xqmstats.xs_qm_dqinact_reclaims);
+			goto dqunlock;
+		}
+
+		ASSERT(dqp->q_hash);
+		ASSERT(!list_empty(&dqp->q_mplist));
+
+		/*
+		 * Try to grab the flush lock. If this dquot is in the process
+		 * of getting flushed to disk, we don't want to reclaim it.
+		 */
+		if (!xfs_dqflock_nowait(dqp))
+			goto dqunlock;
+
+		/*
+		 * We have the flush lock so we know that this is not in the
+		 * process of being flushed. So, if this is dirty, flush it
+		 * DELWRI so that we don't get a freelist infested with
+		 * dirty dquots.
+		 */
+		if (XFS_DQ_IS_DIRTY(dqp)) {
+			int	error;
+
+			trace_xfs_dqreclaim_dirty(dqp);
+
+			/*
+			 * We flush it delayed write, so don't bother
+			 * releasing the freelist lock.
+			 */
+			error = xfs_qm_dqflush(dqp, 0);
+			if (error) {
+				xfs_warn(mp, "%s: dquot %p flush failed",
+					__func__, dqp);
+			}
+			goto dqunlock;
+		}
+
+		/*
+		 * We're trying to get the hashlock out of order. This races
+		 * with dqlookup; so, we giveup and goto the next dquot if
+		 * we couldn't get the hashlock. This way, we won't starve
+		 * a dqlookup process that holds the hashlock that is
+		 * waiting for the freelist lock.
+		 */
+		if (!mutex_trylock(&dqp->q_hash->qh_lock)) {
+			restarts++;
+			goto dqfunlock;
+		}
+
+		/*
+		 * This races with dquot allocation code as well as dqflush_all
+		 * and reclaim code. So, if we failed to grab the mplist lock,
+		 * giveup everything and start over.
+		 */
+		if (!mutex_trylock(&mp->m_quotainfo->qi_dqlist_lock)) {
+			restarts++;
+			startagain = 1;
+			goto qhunlock;
+		}
+
+		ASSERT(dqp->q_nrefs == 0);
+		list_del_init(&dqp->q_mplist);
+		mp->m_quotainfo->qi_dquots--;
+		mp->m_quotainfo->qi_dqreclaims++;
+		list_del_init(&dqp->q_hashlist);
+		dqp->q_hash->qh_version++;
+		list_del_init(&dqp->q_freelist);
+		xfs_Gqm->qm_dqfrlist_cnt--;
+		dqpout = dqp;
+		mutex_unlock(&mp->m_quotainfo->qi_dqlist_lock);
+qhunlock:
+		mutex_unlock(&dqp->q_hash->qh_lock);
+dqfunlock:
+		xfs_dqfunlock(dqp);
+dqunlock:
+		xfs_dqunlock(dqp);
+		if (dqpout)
+			break;
+		if (restarts >= XFS_QM_RECLAIM_MAX_RESTARTS)
+			break;
+		if (startagain) {
+			mutex_unlock(&xfs_Gqm->qm_dqfrlist_lock);
+			goto again;
+		}
+	}
+	mutex_unlock(&xfs_Gqm->qm_dqfrlist_lock);
+	return dqpout;
+}
+
+/*
+ * Traverse the freelist of dquots and attempt to reclaim a maximum of
+ * 'howmany' dquots. This operation races with dqlookup(), and attempts to
+ * favor the lookup function ...
+ */
+STATIC int
+xfs_qm_shake_freelist(
+	int	howmany)
+{
+	int		nreclaimed = 0;
+	xfs_dquot_t	*dqp;
+
+	if (howmany <= 0)
+		return 0;
+
+	while (nreclaimed < howmany) {
+		dqp = xfs_qm_dqreclaim_one();
+		if (!dqp)
+			return nreclaimed;
+		xfs_qm_dqdestroy(dqp);
+		nreclaimed++;
+	}
+	return nreclaimed;
+}
+
+/*
+ * The kmem_shake interface is invoked when memory is running low.
+ */
+/* ARGSUSED */
+STATIC int
+xfs_qm_shake(
+	struct shrinker	*shrink,
+	struct shrink_control *sc)
+{
+	int	ndqused, nfree, n;
+	gfp_t gfp_mask = sc->gfp_mask;
+
+	if (!kmem_shake_allow(gfp_mask))
+		return 0;
+	if (!xfs_Gqm)
+		return 0;
+
+	nfree = xfs_Gqm->qm_dqfrlist_cnt; /* free dquots */
+	/* incore dquots in all f/s's */
+	ndqused = atomic_read(&xfs_Gqm->qm_totaldquots) - nfree;
+
+	ASSERT(ndqused >= 0);
+
+	if (nfree <= ndqused && nfree < ndquot)
+		return 0;
+
+	ndqused *= xfs_Gqm->qm_dqfree_ratio;	/* target # of free dquots */
+	n = nfree - ndqused - ndquot;		/* # over target */
+
+	return xfs_qm_shake_freelist(MAX(nfree, n));
+}
+
+
+/*------------------------------------------------------------------*/
+
+/*
+ * Return a new incore dquot. Depending on the number of
+ * dquots in the system, we either allocate a new one on the kernel heap,
+ * or reclaim a free one.
+ * Return value is B_TRUE if we allocated a new dquot, B_FALSE if we managed
+ * to reclaim an existing one from the freelist.
+ */
+boolean_t
+xfs_qm_dqalloc_incore(
+	xfs_dquot_t **O_dqpp)
+{
+	xfs_dquot_t	*dqp;
+
+	/*
+	 * Check against high water mark to see if we want to pop
+	 * a nincompoop dquot off the freelist.
+	 */
+	if (atomic_read(&xfs_Gqm->qm_totaldquots) >= ndquot) {
+		/*
+		 * Try to recycle a dquot from the freelist.
+		 */
+		if ((dqp = xfs_qm_dqreclaim_one())) {
+			XQM_STATS_INC(xqmstats.xs_qm_dqreclaims);
+			/*
+			 * Just zero the core here. The rest will get
+			 * reinitialized by caller. XXX we shouldn't even
+			 * do this zero ...
+			 */
+			memset(&dqp->q_core, 0, sizeof(dqp->q_core));
+			*O_dqpp = dqp;
+			return B_FALSE;
+		}
+		XQM_STATS_INC(xqmstats.xs_qm_dqreclaim_misses);
+	}
+
+	/*
+	 * Allocate a brand new dquot on the kernel heap and return it
+	 * to the caller to initialize.
+	 */
+	ASSERT(xfs_Gqm->qm_dqzone != NULL);
+	*O_dqpp = kmem_zone_zalloc(xfs_Gqm->qm_dqzone, KM_SLEEP);
+	atomic_inc(&xfs_Gqm->qm_totaldquots);
+
+	return B_TRUE;
+}
+
+
+/*
+ * Start a transaction and write the incore superblock changes to
+ * disk. flags parameter indicates which fields have changed.
+ */
+int
+xfs_qm_write_sb_changes(
+	xfs_mount_t	*mp,
+	__int64_t	flags)
+{
+	xfs_trans_t	*tp;
+	int		error;
+
+#ifdef QUOTADEBUG
+	xfs_notice(mp, "Writing superblock quota changes");
+#endif
+	tp = xfs_trans_alloc(mp, XFS_TRANS_QM_SBCHANGE);
+	if ((error = xfs_trans_reserve(tp, 0,
+				      mp->m_sb.sb_sectsize + 128, 0,
+				      0,
+				      XFS_DEFAULT_LOG_COUNT))) {
+		xfs_trans_cancel(tp, 0);
+		return error;
+	}
+
+	xfs_mod_sb(tp, flags);
+	error = xfs_trans_commit(tp, 0);
+
+	return error;
+}
+
+
+/* --------------- utility functions for vnodeops ---------------- */
+
+
+/*
+ * Given an inode, a uid, gid and prid make sure that we have
+ * allocated relevant dquot(s) on disk, and that we won't exceed inode
+ * quotas by creating this file.
+ * This also attaches dquot(s) to the given inode after locking it,
+ * and returns the dquots corresponding to the uid and/or gid.
+ *
+ * in	: inode (unlocked)
+ * out	: udquot, gdquot with references taken and unlocked
+ */
+int
+xfs_qm_vop_dqalloc(
+	struct xfs_inode	*ip,
+	uid_t			uid,
+	gid_t			gid,
+	prid_t			prid,
+	uint			flags,
+	struct xfs_dquot	**O_udqpp,
+	struct xfs_dquot	**O_gdqpp)
+{
+	struct xfs_mount	*mp = ip->i_mount;
+	struct xfs_dquot	*uq, *gq;
+	int			error;
+	uint			lockflags;
+
+	if (!XFS_IS_QUOTA_RUNNING(mp) || !XFS_IS_QUOTA_ON(mp))
+		return 0;
+
+	lockflags = XFS_ILOCK_EXCL;
+	xfs_ilock(ip, lockflags);
+
+	if ((flags & XFS_QMOPT_INHERIT) && XFS_INHERIT_GID(ip))
+		gid = ip->i_d.di_gid;
+
+	/*
+	 * Attach the dquot(s) to this inode, doing a dquot allocation
+	 * if necessary. The dquot(s) will not be locked.
+	 */
+	if (XFS_NOT_DQATTACHED(mp, ip)) {
+		error = xfs_qm_dqattach_locked(ip, XFS_QMOPT_DQALLOC);
+		if (error) {
+			xfs_iunlock(ip, lockflags);
+			return error;
+		}
+	}
+
+	uq = gq = NULL;
+	if ((flags & XFS_QMOPT_UQUOTA) && XFS_IS_UQUOTA_ON(mp)) {
+		if (ip->i_d.di_uid != uid) {
+			/*
+			 * What we need is the dquot that has this uid, and
+			 * if we send the inode to dqget, the uid of the inode
+			 * takes priority over what's sent in the uid argument.
+			 * We must unlock inode here before calling dqget if
+			 * we're not sending the inode, because otherwise
+			 * we'll deadlock by doing trans_reserve while
+			 * holding ilock.
+			 */
+			xfs_iunlock(ip, lockflags);
+			if ((error = xfs_qm_dqget(mp, NULL, (xfs_dqid_t) uid,
+						 XFS_DQ_USER,
+						 XFS_QMOPT_DQALLOC |
+						 XFS_QMOPT_DOWARN,
+						 &uq))) {
+				ASSERT(error != ENOENT);
+				return error;
+			}
+			/*
+			 * Get the ilock in the right order.
+			 */
+			xfs_dqunlock(uq);
+			lockflags = XFS_ILOCK_SHARED;
+			xfs_ilock(ip, lockflags);
+		} else {
+			/*
+			 * Take an extra reference, because we'll return
+			 * this to caller
+			 */
+			ASSERT(ip->i_udquot);
+			uq = ip->i_udquot;
+			xfs_dqlock(uq);
+			XFS_DQHOLD(uq);
+			xfs_dqunlock(uq);
+		}
+	}
+	if ((flags & XFS_QMOPT_GQUOTA) && XFS_IS_GQUOTA_ON(mp)) {
+		if (ip->i_d.di_gid != gid) {
+			xfs_iunlock(ip, lockflags);
+			if ((error = xfs_qm_dqget(mp, NULL, (xfs_dqid_t)gid,
+						 XFS_DQ_GROUP,
+						 XFS_QMOPT_DQALLOC |
+						 XFS_QMOPT_DOWARN,
+						 &gq))) {
+				if (uq)
+					xfs_qm_dqrele(uq);
+				ASSERT(error != ENOENT);
+				return error;
+			}
+			xfs_dqunlock(gq);
+			lockflags = XFS_ILOCK_SHARED;
+			xfs_ilock(ip, lockflags);
+		} else {
+			ASSERT(ip->i_gdquot);
+			gq = ip->i_gdquot;
+			xfs_dqlock(gq);
+			XFS_DQHOLD(gq);
+			xfs_dqunlock(gq);
+		}
+	} else if ((flags & XFS_QMOPT_PQUOTA) && XFS_IS_PQUOTA_ON(mp)) {
+		if (xfs_get_projid(ip) != prid) {
+			xfs_iunlock(ip, lockflags);
+			if ((error = xfs_qm_dqget(mp, NULL, (xfs_dqid_t)prid,
+						 XFS_DQ_PROJ,
+						 XFS_QMOPT_DQALLOC |
+						 XFS_QMOPT_DOWARN,
+						 &gq))) {
+				if (uq)
+					xfs_qm_dqrele(uq);
+				ASSERT(error != ENOENT);
+				return (error);
+			}
+			xfs_dqunlock(gq);
+			lockflags = XFS_ILOCK_SHARED;
+			xfs_ilock(ip, lockflags);
+		} else {
+			ASSERT(ip->i_gdquot);
+			gq = ip->i_gdquot;
+			xfs_dqlock(gq);
+			XFS_DQHOLD(gq);
+			xfs_dqunlock(gq);
+		}
+	}
+	if (uq)
+		trace_xfs_dquot_dqalloc(ip);
+
+	xfs_iunlock(ip, lockflags);
+	if (O_udqpp)
+		*O_udqpp = uq;
+	else if (uq)
+		xfs_qm_dqrele(uq);
+	if (O_gdqpp)
+		*O_gdqpp = gq;
+	else if (gq)
+		xfs_qm_dqrele(gq);
+	return 0;
+}
+
+/*
+ * Actually transfer ownership, and do dquot modifications.
+ * These were already reserved.
+ */
+xfs_dquot_t *
+xfs_qm_vop_chown(
+	xfs_trans_t	*tp,
+	xfs_inode_t	*ip,
+	xfs_dquot_t	**IO_olddq,
+	xfs_dquot_t	*newdq)
+{
+	xfs_dquot_t	*prevdq;
+	uint		bfield = XFS_IS_REALTIME_INODE(ip) ?
+				 XFS_TRANS_DQ_RTBCOUNT : XFS_TRANS_DQ_BCOUNT;
+
+
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
+	ASSERT(XFS_IS_QUOTA_RUNNING(ip->i_mount));
+
+	/* old dquot */
+	prevdq = *IO_olddq;
+	ASSERT(prevdq);
+	ASSERT(prevdq != newdq);
+
+	xfs_trans_mod_dquot(tp, prevdq, bfield, -(ip->i_d.di_nblocks));
+	xfs_trans_mod_dquot(tp, prevdq, XFS_TRANS_DQ_ICOUNT, -1);
+
+	/* the sparkling new dquot */
+	xfs_trans_mod_dquot(tp, newdq, bfield, ip->i_d.di_nblocks);
+	xfs_trans_mod_dquot(tp, newdq, XFS_TRANS_DQ_ICOUNT, 1);
+
+	/*
+	 * Take an extra reference, because the inode
+	 * is going to keep this dquot pointer even
+	 * after the trans_commit.
+	 */
+	xfs_dqlock(newdq);
+	XFS_DQHOLD(newdq);
+	xfs_dqunlock(newdq);
+	*IO_olddq = newdq;
+
+	return prevdq;
+}
+
+/*
+ * Quota reservations for setattr(AT_UID|AT_GID|AT_PROJID).
+ */
+int
+xfs_qm_vop_chown_reserve(
+	xfs_trans_t	*tp,
+	xfs_inode_t	*ip,
+	xfs_dquot_t	*udqp,
+	xfs_dquot_t	*gdqp,
+	uint		flags)
+{
+	xfs_mount_t	*mp = ip->i_mount;
+	uint		delblks, blkflags, prjflags = 0;
+	xfs_dquot_t	*unresudq, *unresgdq, *delblksudq, *delblksgdq;
+	int		error;
+
+
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL|XFS_ILOCK_SHARED));
+	ASSERT(XFS_IS_QUOTA_RUNNING(mp));
+
+	delblks = ip->i_delayed_blks;
+	delblksudq = delblksgdq = unresudq = unresgdq = NULL;
+	blkflags = XFS_IS_REALTIME_INODE(ip) ?
+			XFS_QMOPT_RES_RTBLKS : XFS_QMOPT_RES_REGBLKS;
+
+	if (XFS_IS_UQUOTA_ON(mp) && udqp &&
+	    ip->i_d.di_uid != (uid_t)be32_to_cpu(udqp->q_core.d_id)) {
+		delblksudq = udqp;
+		/*
+		 * If there are delayed allocation blocks, then we have to
+		 * unreserve those from the old dquot, and add them to the
+		 * new dquot.
+		 */
+		if (delblks) {
+			ASSERT(ip->i_udquot);
+			unresudq = ip->i_udquot;
+		}
+	}
+	if (XFS_IS_OQUOTA_ON(ip->i_mount) && gdqp) {
+		if (XFS_IS_PQUOTA_ON(ip->i_mount) &&
+		     xfs_get_projid(ip) != be32_to_cpu(gdqp->q_core.d_id))
+			prjflags = XFS_QMOPT_ENOSPC;
+
+		if (prjflags ||
+		    (XFS_IS_GQUOTA_ON(ip->i_mount) &&
+		     ip->i_d.di_gid != be32_to_cpu(gdqp->q_core.d_id))) {
+			delblksgdq = gdqp;
+			if (delblks) {
+				ASSERT(ip->i_gdquot);
+				unresgdq = ip->i_gdquot;
+			}
+		}
+	}
+
+	if ((error = xfs_trans_reserve_quota_bydquots(tp, ip->i_mount,
+				delblksudq, delblksgdq, ip->i_d.di_nblocks, 1,
+				flags | blkflags | prjflags)))
+		return (error);
+
+	/*
+	 * Do the delayed blks reservations/unreservations now. Since, these
+	 * are done without the help of a transaction, if a reservation fails
+	 * its previous reservations won't be automatically undone by trans
+	 * code. So, we have to do it manually here.
+	 */
+	if (delblks) {
+		/*
+		 * Do the reservations first. Unreservation can't fail.
+		 */
+		ASSERT(delblksudq || delblksgdq);
+		ASSERT(unresudq || unresgdq);
+		if ((error = xfs_trans_reserve_quota_bydquots(NULL, ip->i_mount,
+				delblksudq, delblksgdq, (xfs_qcnt_t)delblks, 0,
+				flags | blkflags | prjflags)))
+			return (error);
+		xfs_trans_reserve_quota_bydquots(NULL, ip->i_mount,
+				unresudq, unresgdq, -((xfs_qcnt_t)delblks), 0,
+				blkflags);
+	}
+
+	return (0);
+}
+
+int
+xfs_qm_vop_rename_dqattach(
+	struct xfs_inode	**i_tab)
+{
+	struct xfs_mount	*mp = i_tab[0]->i_mount;
+	int			i;
+
+	if (!XFS_IS_QUOTA_RUNNING(mp) || !XFS_IS_QUOTA_ON(mp))
+		return 0;
+
+	for (i = 0; (i < 4 && i_tab[i]); i++) {
+		struct xfs_inode	*ip = i_tab[i];
+		int			error;
+
+		/*
+		 * Watch out for duplicate entries in the table.
+		 */
+		if (i == 0 || ip != i_tab[i-1]) {
+			if (XFS_NOT_DQATTACHED(mp, ip)) {
+				error = xfs_qm_dqattach(ip, 0);
+				if (error)
+					return error;
+			}
+		}
+	}
+	return 0;
+}
+
+void
+xfs_qm_vop_create_dqattach(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*ip,
+	struct xfs_dquot	*udqp,
+	struct xfs_dquot	*gdqp)
+{
+	struct xfs_mount	*mp = tp->t_mountp;
+
+	if (!XFS_IS_QUOTA_RUNNING(mp) || !XFS_IS_QUOTA_ON(mp))
+		return;
+
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
+	ASSERT(XFS_IS_QUOTA_RUNNING(mp));
+
+	if (udqp) {
+		xfs_dqlock(udqp);
+		XFS_DQHOLD(udqp);
+		xfs_dqunlock(udqp);
+		ASSERT(ip->i_udquot == NULL);
+		ip->i_udquot = udqp;
+		ASSERT(XFS_IS_UQUOTA_ON(mp));
+		ASSERT(ip->i_d.di_uid == be32_to_cpu(udqp->q_core.d_id));
+		xfs_trans_mod_dquot(tp, udqp, XFS_TRANS_DQ_ICOUNT, 1);
+	}
+	if (gdqp) {
+		xfs_dqlock(gdqp);
+		XFS_DQHOLD(gdqp);
+		xfs_dqunlock(gdqp);
+		ASSERT(ip->i_gdquot == NULL);
+		ip->i_gdquot = gdqp;
+		ASSERT(XFS_IS_OQUOTA_ON(mp));
+		ASSERT((XFS_IS_GQUOTA_ON(mp) ?
+			ip->i_d.di_gid : xfs_get_projid(ip)) ==
+				be32_to_cpu(gdqp->q_core.d_id));
+		xfs_trans_mod_dquot(tp, gdqp, XFS_TRANS_DQ_ICOUNT, 1);
+	}
+}
+
diff --git a/fs/xfs/quota/xfs_qm.h b/fs/xfs/quota/xfs_qm.h
new file mode 100644
index 00000000..567b29b9
--- /dev/null
+++ b/fs/xfs/quota/xfs_qm.h
@@ -0,0 +1,172 @@
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_QM_H__
+#define __XFS_QM_H__
+
+#include "xfs_dquot_item.h"
+#include "xfs_dquot.h"
+#include "xfs_quota_priv.h"
+#include "xfs_qm_stats.h"
+
+struct xfs_qm;
+struct xfs_inode;
+
+extern uint		ndquot;
+extern struct mutex	xfs_Gqm_lock;
+extern struct xfs_qm	*xfs_Gqm;
+extern kmem_zone_t	*qm_dqzone;
+extern kmem_zone_t	*qm_dqtrxzone;
+
+/*
+ * Used in xfs_qm_sync called by xfs_sync to count the max times that it can
+ * iterate over the mountpt's dquot list in one call.
+ */
+#define XFS_QM_SYNC_MAX_RESTARTS	7
+
+/*
+ * Ditto, for xfs_qm_dqreclaim_one.
+ */
+#define XFS_QM_RECLAIM_MAX_RESTARTS	4
+
+/*
+ * Ideal ratio of free to in use dquots. Quota manager makes an attempt
+ * to keep this balance.
+ */
+#define XFS_QM_DQFREE_RATIO		2
+
+/*
+ * Dquot hashtable constants/threshold values.
+ */
+#define XFS_QM_HASHSIZE_LOW		(PAGE_SIZE / sizeof(xfs_dqhash_t))
+#define XFS_QM_HASHSIZE_HIGH		((PAGE_SIZE * 4) / sizeof(xfs_dqhash_t))
+
+/*
+ * This defines the unit of allocation of dquots.
+ * Currently, it is just one file system block, and a 4K blk contains 30
+ * (136 * 30 = 4080) dquots. It's probably not worth trying to make
+ * this more dynamic.
+ * XXXsup However, if this number is changed, we have to make sure that we don't
+ * implicitly assume that we do allocations in chunks of a single filesystem
+ * block in the dquot/xqm code.
+ */
+#define XFS_DQUOT_CLUSTER_SIZE_FSB	(xfs_filblks_t)1
+
+typedef xfs_dqhash_t	xfs_dqlist_t;
+
+/*
+ * Quota Manager (global) structure. Lives only in core.
+ */
+typedef struct xfs_qm {
+	xfs_dqlist_t	*qm_usr_dqhtable;/* udquot hash table */
+	xfs_dqlist_t	*qm_grp_dqhtable;/* gdquot hash table */
+	uint		 qm_dqhashmask;	 /* # buckets in dq hashtab - 1 */
+	struct list_head qm_dqfrlist;	 /* freelist of dquots */
+	struct mutex	 qm_dqfrlist_lock;
+	int		 qm_dqfrlist_cnt;
+	atomic_t	 qm_totaldquots; /* total incore dquots */
+	uint		 qm_nrefs;	 /* file systems with quota on */
+	int		 qm_dqfree_ratio;/* ratio of free to inuse dquots */
+	kmem_zone_t	*qm_dqzone;	 /* dquot mem-alloc zone */
+	kmem_zone_t	*qm_dqtrxzone;	 /* t_dqinfo of transactions */
+} xfs_qm_t;
+
+/*
+ * Various quota information for individual filesystems.
+ * The mount structure keeps a pointer to this.
+ */
+typedef struct xfs_quotainfo {
+	xfs_inode_t	*qi_uquotaip;	 /* user quota inode */
+	xfs_inode_t	*qi_gquotaip;	 /* group quota inode */
+	struct list_head qi_dqlist;	 /* all dquots in filesys */
+	struct mutex	 qi_dqlist_lock;
+	int		 qi_dquots;
+	int		 qi_dqreclaims;	 /* a change here indicates
+					    a removal in the dqlist */
+	time_t		 qi_btimelimit;	 /* limit for blks timer */
+	time_t		 qi_itimelimit;	 /* limit for inodes timer */
+	time_t		 qi_rtbtimelimit;/* limit for rt blks timer */
+	xfs_qwarncnt_t	 qi_bwarnlimit;	 /* limit for blks warnings */
+	xfs_qwarncnt_t	 qi_iwarnlimit;	 /* limit for inodes warnings */
+	xfs_qwarncnt_t	 qi_rtbwarnlimit;/* limit for rt blks warnings */
+	struct mutex	 qi_quotaofflock;/* to serialize quotaoff */
+	xfs_filblks_t	 qi_dqchunklen;	 /* # BBs in a chunk of dqs */
+	uint		 qi_dqperchunk;	 /* # ondisk dqs in above chunk */
+	xfs_qcnt_t	 qi_bhardlimit;	 /* default data blk hard limit */
+	xfs_qcnt_t	 qi_bsoftlimit;	 /* default data blk soft limit */
+	xfs_qcnt_t	 qi_ihardlimit;	 /* default inode count hard limit */
+	xfs_qcnt_t	 qi_isoftlimit;	 /* default inode count soft limit */
+	xfs_qcnt_t	 qi_rtbhardlimit;/* default realtime blk hard limit */
+	xfs_qcnt_t	 qi_rtbsoftlimit;/* default realtime blk soft limit */
+} xfs_quotainfo_t;
+
+
+extern void	xfs_trans_mod_dquot(xfs_trans_t *, xfs_dquot_t *, uint, long);
+extern int	xfs_trans_reserve_quota_bydquots(xfs_trans_t *, xfs_mount_t *,
+			xfs_dquot_t *, xfs_dquot_t *, long, long, uint);
+extern void	xfs_trans_dqjoin(xfs_trans_t *, xfs_dquot_t *);
+extern void	xfs_trans_log_dquot(xfs_trans_t *, xfs_dquot_t *);
+
+/*
+ * We keep the usr and grp dquots separately so that locking will be easier
+ * to do at commit time. All transactions that we know of at this point
+ * affect no more than two dquots of one type. Hence, the TRANS_MAXDQS value.
+ */
+#define XFS_QM_TRANS_MAXDQS		2
+typedef struct xfs_dquot_acct {
+	xfs_dqtrx_t	dqa_usrdquots[XFS_QM_TRANS_MAXDQS];
+	xfs_dqtrx_t	dqa_grpdquots[XFS_QM_TRANS_MAXDQS];
+} xfs_dquot_acct_t;
+
+/*
+ * Users are allowed to have a usage exceeding their softlimit for
+ * a period this long.
+ */
+#define XFS_QM_BTIMELIMIT	(7 * 24*60*60)          /* 1 week */
+#define XFS_QM_RTBTIMELIMIT	(7 * 24*60*60)          /* 1 week */
+#define XFS_QM_ITIMELIMIT	(7 * 24*60*60)          /* 1 week */
+
+#define XFS_QM_BWARNLIMIT	5
+#define XFS_QM_IWARNLIMIT	5
+#define XFS_QM_RTBWARNLIMIT	5
+
+extern void		xfs_qm_destroy_quotainfo(xfs_mount_t *);
+extern int		xfs_qm_quotacheck(xfs_mount_t *);
+extern int		xfs_qm_write_sb_changes(xfs_mount_t *, __int64_t);
+
+/* dquot stuff */
+extern boolean_t	xfs_qm_dqalloc_incore(xfs_dquot_t **);
+extern int		xfs_qm_dqpurge_all(xfs_mount_t *, uint);
+extern void		xfs_qm_dqrele_all_inodes(xfs_mount_t *, uint);
+
+/* quota ops */
+extern int		xfs_qm_scall_trunc_qfiles(xfs_mount_t *, uint);
+extern int		xfs_qm_scall_getquota(xfs_mount_t *, xfs_dqid_t, uint,
+					fs_disk_quota_t *);
+extern int		xfs_qm_scall_setqlim(xfs_mount_t *, xfs_dqid_t, uint,
+					fs_disk_quota_t *);
+extern int		xfs_qm_scall_getqstat(xfs_mount_t *, fs_quota_stat_t *);
+extern int		xfs_qm_scall_quotaon(xfs_mount_t *, uint);
+extern int		xfs_qm_scall_quotaoff(xfs_mount_t *, uint);
+
+#ifdef DEBUG
+extern int		xfs_qm_internalqcheck(xfs_mount_t *);
+#else
+#define xfs_qm_internalqcheck(mp)	(0)
+#endif
+
+#endif /* __XFS_QM_H__ */
diff --git a/fs/xfs/quota/xfs_qm_bhv.c b/fs/xfs/quota/xfs_qm_bhv.c
new file mode 100644
index 00000000..a0a829ad
--- /dev/null
+++ b/fs/xfs/quota/xfs_qm_bhv.c
@@ -0,0 +1,176 @@
+/*
+ * Copyright (c) 2000-2006 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_alloc.h"
+#include "xfs_quota.h"
+#include "xfs_mount.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_inode.h"
+#include "xfs_itable.h"
+#include "xfs_bmap.h"
+#include "xfs_rtalloc.h"
+#include "xfs_error.h"
+#include "xfs_attr.h"
+#include "xfs_buf_item.h"
+#include "xfs_qm.h"
+
+
+STATIC void
+xfs_fill_statvfs_from_dquot(
+	struct kstatfs		*statp,
+	xfs_disk_dquot_t	*dp)
+{
+	__uint64_t		limit;
+
+	limit = dp->d_blk_softlimit ?
+		be64_to_cpu(dp->d_blk_softlimit) :
+		be64_to_cpu(dp->d_blk_hardlimit);
+	if (limit && statp->f_blocks > limit) {
+		statp->f_blocks = limit;
+		statp->f_bfree = statp->f_bavail =
+			(statp->f_blocks > be64_to_cpu(dp->d_bcount)) ?
+			 (statp->f_blocks - be64_to_cpu(dp->d_bcount)) : 0;
+	}
+
+	limit = dp->d_ino_softlimit ?
+		be64_to_cpu(dp->d_ino_softlimit) :
+		be64_to_cpu(dp->d_ino_hardlimit);
+	if (limit && statp->f_files > limit) {
+		statp->f_files = limit;
+		statp->f_ffree =
+			(statp->f_files > be64_to_cpu(dp->d_icount)) ?
+			 (statp->f_ffree - be64_to_cpu(dp->d_icount)) : 0;
+	}
+}
+
+
+/*
+ * Directory tree accounting is implemented using project quotas, where
+ * the project identifier is inherited from parent directories.
+ * A statvfs (df, etc.) of a directory that is using project quota should
+ * return a statvfs of the project, not the entire filesystem.
+ * This makes such trees appear as if they are filesystems in themselves.
+ */
+void
+xfs_qm_statvfs(
+	xfs_inode_t		*ip,
+	struct kstatfs		*statp)
+{
+	xfs_mount_t		*mp = ip->i_mount;
+	xfs_dquot_t		*dqp;
+
+	if (!xfs_qm_dqget(mp, NULL, xfs_get_projid(ip), XFS_DQ_PROJ, 0, &dqp)) {
+		xfs_fill_statvfs_from_dquot(statp, &dqp->q_core);
+		xfs_qm_dqput(dqp);
+	}
+}
+
+int
+xfs_qm_newmount(
+	xfs_mount_t	*mp,
+	uint		*needquotamount,
+	uint		*quotaflags)
+{
+	uint		quotaondisk;
+	uint		uquotaondisk = 0, gquotaondisk = 0, pquotaondisk = 0;
+
+	quotaondisk = xfs_sb_version_hasquota(&mp->m_sb) &&
+				(mp->m_sb.sb_qflags & XFS_ALL_QUOTA_ACCT);
+
+	if (quotaondisk) {
+		uquotaondisk = mp->m_sb.sb_qflags & XFS_UQUOTA_ACCT;
+		pquotaondisk = mp->m_sb.sb_qflags & XFS_PQUOTA_ACCT;
+		gquotaondisk = mp->m_sb.sb_qflags & XFS_GQUOTA_ACCT;
+	}
+
+	/*
+	 * If the device itself is read-only, we can't allow
+	 * the user to change the state of quota on the mount -
+	 * this would generate a transaction on the ro device,
+	 * which would lead to an I/O error and shutdown
+	 */
+
+	if (((uquotaondisk && !XFS_IS_UQUOTA_ON(mp)) ||
+	    (!uquotaondisk &&  XFS_IS_UQUOTA_ON(mp)) ||
+	     (pquotaondisk && !XFS_IS_PQUOTA_ON(mp)) ||
+	    (!pquotaondisk &&  XFS_IS_PQUOTA_ON(mp)) ||
+	     (gquotaondisk && !XFS_IS_GQUOTA_ON(mp)) ||
+	    (!gquotaondisk &&  XFS_IS_OQUOTA_ON(mp)))  &&
+	    xfs_dev_is_read_only(mp, "changing quota state")) {
+		xfs_warn(mp, "please mount with%s%s%s%s.",
+			(!quotaondisk ? "out quota" : ""),
+			(uquotaondisk ? " usrquota" : ""),
+			(pquotaondisk ? " prjquota" : ""),
+			(gquotaondisk ? " grpquota" : ""));
+		return XFS_ERROR(EPERM);
+	}
+
+	if (XFS_IS_QUOTA_ON(mp) || quotaondisk) {
+		/*
+		 * Call mount_quotas at this point only if we won't have to do
+		 * a quotacheck.
+		 */
+		if (quotaondisk && !XFS_QM_NEED_QUOTACHECK(mp)) {
+			/*
+			 * If an error occurred, qm_mount_quotas code
+			 * has already disabled quotas. So, just finish
+			 * mounting, and get on with the boring life
+			 * without disk quotas.
+			 */
+			xfs_qm_mount_quotas(mp);
+		} else {
+			/*
+			 * Clear the quota flags, but remember them. This
+			 * is so that the quota code doesn't get invoked
+			 * before we're ready. This can happen when an
+			 * inode goes inactive and wants to free blocks,
+			 * or via xfs_log_mount_finish.
+			 */
+			*needquotamount = B_TRUE;
+			*quotaflags = mp->m_qflags;
+			mp->m_qflags = 0;
+		}
+	}
+
+	return 0;
+}
+
+void __init
+xfs_qm_init(void)
+{
+	printk(KERN_INFO "SGI XFS Quota Management subsystem\n");
+	mutex_init(&xfs_Gqm_lock);
+	xfs_qm_init_procfs();
+}
+
+void __exit
+xfs_qm_exit(void)
+{
+	xfs_qm_cleanup_procfs();
+	if (qm_dqzone)
+		kmem_zone_destroy(qm_dqzone);
+	if (qm_dqtrxzone)
+		kmem_zone_destroy(qm_dqtrxzone);
+}
diff --git a/fs/xfs/quota/xfs_qm_stats.c b/fs/xfs/quota/xfs_qm_stats.c
new file mode 100644
index 00000000..8671a0b3
--- /dev/null
+++ b/fs/xfs/quota/xfs_qm_stats.c
@@ -0,0 +1,105 @@
+/*
+ * Copyright (c) 2000-2003 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_alloc.h"
+#include "xfs_quota.h"
+#include "xfs_mount.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_inode.h"
+#include "xfs_itable.h"
+#include "xfs_bmap.h"
+#include "xfs_rtalloc.h"
+#include "xfs_error.h"
+#include "xfs_attr.h"
+#include "xfs_buf_item.h"
+#include "xfs_qm.h"
+
+struct xqmstats xqmstats;
+
+static int xqm_proc_show(struct seq_file *m, void *v)
+{
+	/* maximum; incore; ratio free to inuse; freelist */
+	seq_printf(m, "%d\t%d\t%d\t%u\n",
+			ndquot,
+			xfs_Gqm? atomic_read(&xfs_Gqm->qm_totaldquots) : 0,
+			xfs_Gqm? xfs_Gqm->qm_dqfree_ratio : 0,
+			xfs_Gqm? xfs_Gqm->qm_dqfrlist_cnt : 0);
+	return 0;
+}
+
+static int xqm_proc_open(struct inode *inode, struct file *file)
+{
+	return single_open(file, xqm_proc_show, NULL);
+}
+
+static const struct file_operations xqm_proc_fops = {
+	.owner		= THIS_MODULE,
+	.open		= xqm_proc_open,
+	.read		= seq_read,
+	.llseek		= seq_lseek,
+	.release	= single_release,
+};
+
+static int xqmstat_proc_show(struct seq_file *m, void *v)
+{
+	/* quota performance statistics */
+	seq_printf(m, "qm %u %u %u %u %u %u %u %u\n",
+			xqmstats.xs_qm_dqreclaims,
+			xqmstats.xs_qm_dqreclaim_misses,
+			xqmstats.xs_qm_dquot_dups,
+			xqmstats.xs_qm_dqcachemisses,
+			xqmstats.xs_qm_dqcachehits,
+			xqmstats.xs_qm_dqwants,
+			xqmstats.xs_qm_dqshake_reclaims,
+			xqmstats.xs_qm_dqinact_reclaims);
+	return 0;
+}
+
+static int xqmstat_proc_open(struct inode *inode, struct file *file)
+{
+	return single_open(file, xqmstat_proc_show, NULL);
+}
+
+static const struct file_operations xqmstat_proc_fops = {
+	.owner		= THIS_MODULE,
+	.open		= xqmstat_proc_open,
+	.read		= seq_read,
+	.llseek		= seq_lseek,
+	.release	= single_release,
+};
+
+void
+xfs_qm_init_procfs(void)
+{
+	proc_create("fs/xfs/xqmstat", 0, NULL, &xqmstat_proc_fops);
+	proc_create("fs/xfs/xqm", 0, NULL, &xqm_proc_fops);
+}
+
+void
+xfs_qm_cleanup_procfs(void)
+{
+	remove_proc_entry("fs/xfs/xqm", NULL);
+	remove_proc_entry("fs/xfs/xqmstat", NULL);
+}
diff --git a/fs/xfs/quota/xfs_qm_stats.h b/fs/xfs/quota/xfs_qm_stats.h
new file mode 100644
index 00000000..5b964fc0
--- /dev/null
+++ b/fs/xfs/quota/xfs_qm_stats.h
@@ -0,0 +1,53 @@
+/*
+ * Copyright (c) 2002 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_QM_STATS_H__
+#define __XFS_QM_STATS_H__
+
+#if defined(CONFIG_PROC_FS) && !defined(XFS_STATS_OFF)
+
+/*
+ * XQM global statistics
+ */
+struct xqmstats {
+	__uint32_t		xs_qm_dqreclaims;
+	__uint32_t		xs_qm_dqreclaim_misses;
+	__uint32_t		xs_qm_dquot_dups;
+	__uint32_t		xs_qm_dqcachemisses;
+	__uint32_t		xs_qm_dqcachehits;
+	__uint32_t		xs_qm_dqwants;
+	__uint32_t		xs_qm_dqshake_reclaims;
+	__uint32_t		xs_qm_dqinact_reclaims;
+};
+
+extern struct xqmstats xqmstats;
+
+# define XQM_STATS_INC(count)	( (count)++ )
+
+extern void xfs_qm_init_procfs(void);
+extern void xfs_qm_cleanup_procfs(void);
+
+#else
+
+# define XQM_STATS_INC(count)	do { } while (0)
+
+static inline void xfs_qm_init_procfs(void) { };
+static inline void xfs_qm_cleanup_procfs(void) { };
+
+#endif
+
+#endif	/* __XFS_QM_STATS_H__ */
diff --git a/fs/xfs/quota/xfs_qm_syscalls.c b/fs/xfs/quota/xfs_qm_syscalls.c
new file mode 100644
index 00000000..2dadb15d
--- /dev/null
+++ b/fs/xfs/quota/xfs_qm_syscalls.c
@@ -0,0 +1,1259 @@
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+
+#include <linux/capability.h>
+
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_alloc.h"
+#include "xfs_quota.h"
+#include "xfs_mount.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_inode.h"
+#include "xfs_itable.h"
+#include "xfs_bmap.h"
+#include "xfs_rtalloc.h"
+#include "xfs_error.h"
+#include "xfs_attr.h"
+#include "xfs_buf_item.h"
+#include "xfs_utils.h"
+#include "xfs_qm.h"
+#include "xfs_trace.h"
+
+STATIC int	xfs_qm_log_quotaoff(xfs_mount_t *, xfs_qoff_logitem_t **, uint);
+STATIC int	xfs_qm_log_quotaoff_end(xfs_mount_t *, xfs_qoff_logitem_t *,
+					uint);
+STATIC uint	xfs_qm_export_flags(uint);
+STATIC uint	xfs_qm_export_qtype_flags(uint);
+STATIC void	xfs_qm_export_dquot(xfs_mount_t *, xfs_disk_dquot_t *,
+					fs_disk_quota_t *);
+
+
+/*
+ * Turn off quota accounting and/or enforcement for all udquots and/or
+ * gdquots. Called only at unmount time.
+ *
+ * This assumes that there are no dquots of this file system cached
+ * incore, and modifies the ondisk dquot directly. Therefore, for example,
+ * it is an error to call this twice, without purging the cache.
+ */
+int
+xfs_qm_scall_quotaoff(
+	xfs_mount_t		*mp,
+	uint			flags)
+{
+	struct xfs_quotainfo	*q = mp->m_quotainfo;
+	uint			dqtype;
+	int			error;
+	uint			inactivate_flags;
+	xfs_qoff_logitem_t	*qoffstart;
+	int			nculprits;
+
+	/*
+	 * No file system can have quotas enabled on disk but not in core.
+	 * Note that quota utilities (like quotaoff) _expect_
+	 * errno == EEXIST here.
+	 */
+	if ((mp->m_qflags & flags) == 0)
+		return XFS_ERROR(EEXIST);
+	error = 0;
+
+	flags &= (XFS_ALL_QUOTA_ACCT | XFS_ALL_QUOTA_ENFD);
+
+	/*
+	 * We don't want to deal with two quotaoffs messing up each other,
+	 * so we're going to serialize it. quotaoff isn't exactly a performance
+	 * critical thing.
+	 * If quotaoff, then we must be dealing with the root filesystem.
+	 */
+	ASSERT(q);
+	mutex_lock(&q->qi_quotaofflock);
+
+	/*
+	 * If we're just turning off quota enforcement, change mp and go.
+	 */
+	if ((flags & XFS_ALL_QUOTA_ACCT) == 0) {
+		mp->m_qflags &= ~(flags);
+
+		spin_lock(&mp->m_sb_lock);
+		mp->m_sb.sb_qflags = mp->m_qflags;
+		spin_unlock(&mp->m_sb_lock);
+		mutex_unlock(&q->qi_quotaofflock);
+
+		/* XXX what to do if error ? Revert back to old vals incore ? */
+		error = xfs_qm_write_sb_changes(mp, XFS_SB_QFLAGS);
+		return (error);
+	}
+
+	dqtype = 0;
+	inactivate_flags = 0;
+	/*
+	 * If accounting is off, we must turn enforcement off, clear the
+	 * quota 'CHKD' certificate to make it known that we have to
+	 * do a quotacheck the next time this quota is turned on.
+	 */
+	if (flags & XFS_UQUOTA_ACCT) {
+		dqtype |= XFS_QMOPT_UQUOTA;
+		flags |= (XFS_UQUOTA_CHKD | XFS_UQUOTA_ENFD);
+		inactivate_flags |= XFS_UQUOTA_ACTIVE;
+	}
+	if (flags & XFS_GQUOTA_ACCT) {
+		dqtype |= XFS_QMOPT_GQUOTA;
+		flags |= (XFS_OQUOTA_CHKD | XFS_OQUOTA_ENFD);
+		inactivate_flags |= XFS_GQUOTA_ACTIVE;
+	} else if (flags & XFS_PQUOTA_ACCT) {
+		dqtype |= XFS_QMOPT_PQUOTA;
+		flags |= (XFS_OQUOTA_CHKD | XFS_OQUOTA_ENFD);
+		inactivate_flags |= XFS_PQUOTA_ACTIVE;
+	}
+
+	/*
+	 * Nothing to do?  Don't complain. This happens when we're just
+	 * turning off quota enforcement.
+	 */
+	if ((mp->m_qflags & flags) == 0)
+		goto out_unlock;
+
+	/*
+	 * Write the LI_QUOTAOFF log record, and do SB changes atomically,
+	 * and synchronously. If we fail to write, we should abort the
+	 * operation as it cannot be recovered safely if we crash.
+	 */
+	error = xfs_qm_log_quotaoff(mp, &qoffstart, flags);
+	if (error)
+		goto out_unlock;
+
+	/*
+	 * Next we clear the XFS_MOUNT_*DQ_ACTIVE bit(s) in the mount struct
+	 * to take care of the race between dqget and quotaoff. We don't take
+	 * any special locks to reset these bits. All processes need to check
+	 * these bits *after* taking inode lock(s) to see if the particular
+	 * quota type is in the process of being turned off. If *ACTIVE, it is
+	 * guaranteed that all dquot structures and all quotainode ptrs will all
+	 * stay valid as long as that inode is kept locked.
+	 *
+	 * There is no turning back after this.
+	 */
+	mp->m_qflags &= ~inactivate_flags;
+
+	/*
+	 * Give back all the dquot reference(s) held by inodes.
+	 * Here we go thru every single incore inode in this file system, and
+	 * do a dqrele on the i_udquot/i_gdquot that it may have.
+	 * Essentially, as long as somebody has an inode locked, this guarantees
+	 * that quotas will not be turned off. This is handy because in a
+	 * transaction once we lock the inode(s) and check for quotaon, we can
+	 * depend on the quota inodes (and other things) being valid as long as
+	 * we keep the lock(s).
+	 */
+	xfs_qm_dqrele_all_inodes(mp, flags);
+
+	/*
+	 * Next we make the changes in the quota flag in the mount struct.
+	 * This isn't protected by a particular lock directly, because we
+	 * don't want to take a mrlock every time we depend on quotas being on.
+	 */
+	mp->m_qflags &= ~(flags);
+
+	/*
+	 * Go through all the dquots of this file system and purge them,
+	 * according to what was turned off. We may not be able to get rid
+	 * of all dquots, because dquots can have temporary references that
+	 * are not attached to inodes. eg. xfs_setattr, xfs_create.
+	 * So, if we couldn't purge all the dquots from the filesystem,
+	 * we can't get rid of the incore data structures.
+	 */
+	while ((nculprits = xfs_qm_dqpurge_all(mp, dqtype)))
+		delay(10 * nculprits);
+
+	/*
+	 * Transactions that had started before ACTIVE state bit was cleared
+	 * could have logged many dquots, so they'd have higher LSNs than
+	 * the first QUOTAOFF log record does. If we happen to crash when
+	 * the tail of the log has gone past the QUOTAOFF record, but
+	 * before the last dquot modification, those dquots __will__
+	 * recover, and that's not good.
+	 *
+	 * So, we have QUOTAOFF start and end logitems; the start
+	 * logitem won't get overwritten until the end logitem appears...
+	 */
+	error = xfs_qm_log_quotaoff_end(mp, qoffstart, flags);
+	if (error) {
+		/* We're screwed now. Shutdown is the only option. */
+		xfs_force_shutdown(mp, SHUTDOWN_CORRUPT_INCORE);
+		goto out_unlock;
+	}
+
+	/*
+	 * If quotas is completely disabled, close shop.
+	 */
+	if (((flags & XFS_MOUNT_QUOTA_ALL) == XFS_MOUNT_QUOTA_SET1) ||
+	    ((flags & XFS_MOUNT_QUOTA_ALL) == XFS_MOUNT_QUOTA_SET2)) {
+		mutex_unlock(&q->qi_quotaofflock);
+		xfs_qm_destroy_quotainfo(mp);
+		return (0);
+	}
+
+	/*
+	 * Release our quotainode references if we don't need them anymore.
+	 */
+	if ((dqtype & XFS_QMOPT_UQUOTA) && q->qi_uquotaip) {
+		IRELE(q->qi_uquotaip);
+		q->qi_uquotaip = NULL;
+	}
+	if ((dqtype & (XFS_QMOPT_GQUOTA|XFS_QMOPT_PQUOTA)) && q->qi_gquotaip) {
+		IRELE(q->qi_gquotaip);
+		q->qi_gquotaip = NULL;
+	}
+
+out_unlock:
+	mutex_unlock(&q->qi_quotaofflock);
+	return error;
+}
+
+STATIC int
+xfs_qm_scall_trunc_qfile(
+	struct xfs_mount	*mp,
+	xfs_ino_t		ino)
+{
+	struct xfs_inode	*ip;
+	struct xfs_trans	*tp;
+	int			error;
+
+	if (ino == NULLFSINO)
+		return 0;
+
+	error = xfs_iget(mp, NULL, ino, 0, 0, &ip);
+	if (error)
+		return error;
+
+	xfs_ilock(ip, XFS_IOLOCK_EXCL);
+
+	tp = xfs_trans_alloc(mp, XFS_TRANS_TRUNCATE_FILE);
+	error = xfs_trans_reserve(tp, 0, XFS_ITRUNCATE_LOG_RES(mp), 0,
+				  XFS_TRANS_PERM_LOG_RES,
+				  XFS_ITRUNCATE_LOG_COUNT);
+	if (error) {
+		xfs_trans_cancel(tp, 0);
+		xfs_iunlock(ip, XFS_IOLOCK_EXCL);
+		goto out_put;
+	}
+
+	xfs_ilock(ip, XFS_ILOCK_EXCL);
+	xfs_trans_ijoin(tp, ip);
+
+	error = xfs_itruncate_finish(&tp, ip, 0, XFS_DATA_FORK, 1);
+	if (error) {
+		xfs_trans_cancel(tp, XFS_TRANS_RELEASE_LOG_RES |
+				     XFS_TRANS_ABORT);
+		goto out_unlock;
+	}
+
+	xfs_trans_ichgtime(tp, ip, XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG);
+	error = xfs_trans_commit(tp, XFS_TRANS_RELEASE_LOG_RES);
+
+out_unlock:
+	xfs_iunlock(ip, XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL);
+out_put:
+	IRELE(ip);
+	return error;
+}
+
+int
+xfs_qm_scall_trunc_qfiles(
+	xfs_mount_t	*mp,
+	uint		flags)
+{
+	int		error = 0, error2 = 0;
+
+	if (!xfs_sb_version_hasquota(&mp->m_sb) || flags == 0) {
+		xfs_debug(mp, "%s: flags=%x m_qflags=%x\n",
+			__func__, flags, mp->m_qflags);
+		return XFS_ERROR(EINVAL);
+	}
+
+	if (flags & XFS_DQ_USER)
+		error = xfs_qm_scall_trunc_qfile(mp, mp->m_sb.sb_uquotino);
+	if (flags & (XFS_DQ_GROUP|XFS_DQ_PROJ))
+		error2 = xfs_qm_scall_trunc_qfile(mp, mp->m_sb.sb_gquotino);
+
+	return error ? error : error2;
+}
+
+/*
+ * Switch on (a given) quota enforcement for a filesystem.  This takes
+ * effect immediately.
+ * (Switching on quota accounting must be done at mount time.)
+ */
+int
+xfs_qm_scall_quotaon(
+	xfs_mount_t	*mp,
+	uint		flags)
+{
+	int		error;
+	uint		qf;
+	__int64_t	sbflags;
+
+	flags &= (XFS_ALL_QUOTA_ACCT | XFS_ALL_QUOTA_ENFD);
+	/*
+	 * Switching on quota accounting must be done at mount time.
+	 */
+	flags &= ~(XFS_ALL_QUOTA_ACCT);
+
+	sbflags = 0;
+
+	if (flags == 0) {
+		xfs_debug(mp, "%s: zero flags, m_qflags=%x\n",
+			__func__, mp->m_qflags);
+		return XFS_ERROR(EINVAL);
+	}
+
+	/* No fs can turn on quotas with a delayed effect */
+	ASSERT((flags & XFS_ALL_QUOTA_ACCT) == 0);
+
+	/*
+	 * Can't enforce without accounting. We check the superblock
+	 * qflags here instead of m_qflags because rootfs can have
+	 * quota acct on ondisk without m_qflags' knowing.
+	 */
+	if (((flags & XFS_UQUOTA_ACCT) == 0 &&
+	    (mp->m_sb.sb_qflags & XFS_UQUOTA_ACCT) == 0 &&
+	    (flags & XFS_UQUOTA_ENFD))
+	    ||
+	    ((flags & XFS_PQUOTA_ACCT) == 0 &&
+	    (mp->m_sb.sb_qflags & XFS_PQUOTA_ACCT) == 0 &&
+	    (flags & XFS_GQUOTA_ACCT) == 0 &&
+	    (mp->m_sb.sb_qflags & XFS_GQUOTA_ACCT) == 0 &&
+	    (flags & XFS_OQUOTA_ENFD))) {
+		xfs_debug(mp,
+			"%s: Can't enforce without acct, flags=%x sbflags=%x\n",
+			__func__, flags, mp->m_sb.sb_qflags);
+		return XFS_ERROR(EINVAL);
+	}
+	/*
+	 * If everything's up to-date incore, then don't waste time.
+	 */
+	if ((mp->m_qflags & flags) == flags)
+		return XFS_ERROR(EEXIST);
+
+	/*
+	 * Change sb_qflags on disk but not incore mp->qflags
+	 * if this is the root filesystem.
+	 */
+	spin_lock(&mp->m_sb_lock);
+	qf = mp->m_sb.sb_qflags;
+	mp->m_sb.sb_qflags = qf | flags;
+	spin_unlock(&mp->m_sb_lock);
+
+	/*
+	 * There's nothing to change if it's the same.
+	 */
+	if ((qf & flags) == flags && sbflags == 0)
+		return XFS_ERROR(EEXIST);
+	sbflags |= XFS_SB_QFLAGS;
+
+	if ((error = xfs_qm_write_sb_changes(mp, sbflags)))
+		return (error);
+	/*
+	 * If we aren't trying to switch on quota enforcement, we are done.
+	 */
+	if  (((mp->m_sb.sb_qflags & XFS_UQUOTA_ACCT) !=
+	     (mp->m_qflags & XFS_UQUOTA_ACCT)) ||
+	     ((mp->m_sb.sb_qflags & XFS_PQUOTA_ACCT) !=
+	     (mp->m_qflags & XFS_PQUOTA_ACCT)) ||
+	     ((mp->m_sb.sb_qflags & XFS_GQUOTA_ACCT) !=
+	     (mp->m_qflags & XFS_GQUOTA_ACCT)) ||
+	    (flags & XFS_ALL_QUOTA_ENFD) == 0)
+		return (0);
+
+	if (! XFS_IS_QUOTA_RUNNING(mp))
+		return XFS_ERROR(ESRCH);
+
+	/*
+	 * Switch on quota enforcement in core.
+	 */
+	mutex_lock(&mp->m_quotainfo->qi_quotaofflock);
+	mp->m_qflags |= (flags & XFS_ALL_QUOTA_ENFD);
+	mutex_unlock(&mp->m_quotainfo->qi_quotaofflock);
+
+	return (0);
+}
+
+
+/*
+ * Return quota status information, such as uquota-off, enforcements, etc.
+ */
+int
+xfs_qm_scall_getqstat(
+	struct xfs_mount	*mp,
+	struct fs_quota_stat	*out)
+{
+	struct xfs_quotainfo	*q = mp->m_quotainfo;
+	struct xfs_inode	*uip, *gip;
+	boolean_t		tempuqip, tempgqip;
+
+	uip = gip = NULL;
+	tempuqip = tempgqip = B_FALSE;
+	memset(out, 0, sizeof(fs_quota_stat_t));
+
+	out->qs_version = FS_QSTAT_VERSION;
+	if (!xfs_sb_version_hasquota(&mp->m_sb)) {
+		out->qs_uquota.qfs_ino = NULLFSINO;
+		out->qs_gquota.qfs_ino = NULLFSINO;
+		return (0);
+	}
+	out->qs_flags = (__uint16_t) xfs_qm_export_flags(mp->m_qflags &
+							(XFS_ALL_QUOTA_ACCT|
+							 XFS_ALL_QUOTA_ENFD));
+	out->qs_pad = 0;
+	out->qs_uquota.qfs_ino = mp->m_sb.sb_uquotino;
+	out->qs_gquota.qfs_ino = mp->m_sb.sb_gquotino;
+
+	if (q) {
+		uip = q->qi_uquotaip;
+		gip = q->qi_gquotaip;
+	}
+	if (!uip && mp->m_sb.sb_uquotino != NULLFSINO) {
+		if (xfs_iget(mp, NULL, mp->m_sb.sb_uquotino,
+					0, 0, &uip) == 0)
+			tempuqip = B_TRUE;
+	}
+	if (!gip && mp->m_sb.sb_gquotino != NULLFSINO) {
+		if (xfs_iget(mp, NULL, mp->m_sb.sb_gquotino,
+					0, 0, &gip) == 0)
+			tempgqip = B_TRUE;
+	}
+	if (uip) {
+		out->qs_uquota.qfs_nblks = uip->i_d.di_nblocks;
+		out->qs_uquota.qfs_nextents = uip->i_d.di_nextents;
+		if (tempuqip)
+			IRELE(uip);
+	}
+	if (gip) {
+		out->qs_gquota.qfs_nblks = gip->i_d.di_nblocks;
+		out->qs_gquota.qfs_nextents = gip->i_d.di_nextents;
+		if (tempgqip)
+			IRELE(gip);
+	}
+	if (q) {
+		out->qs_incoredqs = q->qi_dquots;
+		out->qs_btimelimit = q->qi_btimelimit;
+		out->qs_itimelimit = q->qi_itimelimit;
+		out->qs_rtbtimelimit = q->qi_rtbtimelimit;
+		out->qs_bwarnlimit = q->qi_bwarnlimit;
+		out->qs_iwarnlimit = q->qi_iwarnlimit;
+	}
+	return 0;
+}
+
+#define XFS_DQ_MASK \
+	(FS_DQ_LIMIT_MASK | FS_DQ_TIMER_MASK | FS_DQ_WARNS_MASK)
+
+/*
+ * Adjust quota limits, and start/stop timers accordingly.
+ */
+int
+xfs_qm_scall_setqlim(
+	xfs_mount_t		*mp,
+	xfs_dqid_t		id,
+	uint			type,
+	fs_disk_quota_t		*newlim)
+{
+	struct xfs_quotainfo	*q = mp->m_quotainfo;
+	xfs_disk_dquot_t	*ddq;
+	xfs_dquot_t		*dqp;
+	xfs_trans_t		*tp;
+	int			error;
+	xfs_qcnt_t		hard, soft;
+
+	if (newlim->d_fieldmask & ~XFS_DQ_MASK)
+		return EINVAL;
+	if ((newlim->d_fieldmask & XFS_DQ_MASK) == 0)
+		return 0;
+
+	tp = xfs_trans_alloc(mp, XFS_TRANS_QM_SETQLIM);
+	if ((error = xfs_trans_reserve(tp, 0, sizeof(xfs_disk_dquot_t) + 128,
+				      0, 0, XFS_DEFAULT_LOG_COUNT))) {
+		xfs_trans_cancel(tp, 0);
+		return (error);
+	}
+
+	/*
+	 * We don't want to race with a quotaoff so take the quotaoff lock.
+	 * (We don't hold an inode lock, so there's nothing else to stop
+	 * a quotaoff from happening). (XXXThis doesn't currently happen
+	 * because we take the vfslock before calling xfs_qm_sysent).
+	 */
+	mutex_lock(&q->qi_quotaofflock);
+
+	/*
+	 * Get the dquot (locked), and join it to the transaction.
+	 * Allocate the dquot if this doesn't exist.
+	 */
+	if ((error = xfs_qm_dqget(mp, NULL, id, type, XFS_QMOPT_DQALLOC, &dqp))) {
+		xfs_trans_cancel(tp, XFS_TRANS_ABORT);
+		ASSERT(error != ENOENT);
+		goto out_unlock;
+	}
+	xfs_trans_dqjoin(tp, dqp);
+	ddq = &dqp->q_core;
+
+	/*
+	 * Make sure that hardlimits are >= soft limits before changing.
+	 */
+	hard = (newlim->d_fieldmask & FS_DQ_BHARD) ?
+		(xfs_qcnt_t) XFS_BB_TO_FSB(mp, newlim->d_blk_hardlimit) :
+			be64_to_cpu(ddq->d_blk_hardlimit);
+	soft = (newlim->d_fieldmask & FS_DQ_BSOFT) ?
+		(xfs_qcnt_t) XFS_BB_TO_FSB(mp, newlim->d_blk_softlimit) :
+			be64_to_cpu(ddq->d_blk_softlimit);
+	if (hard == 0 || hard >= soft) {
+		ddq->d_blk_hardlimit = cpu_to_be64(hard);
+		ddq->d_blk_softlimit = cpu_to_be64(soft);
+		if (id == 0) {
+			q->qi_bhardlimit = hard;
+			q->qi_bsoftlimit = soft;
+		}
+	} else {
+		xfs_debug(mp, "blkhard %Ld < blksoft %Ld\n", hard, soft);
+	}
+	hard = (newlim->d_fieldmask & FS_DQ_RTBHARD) ?
+		(xfs_qcnt_t) XFS_BB_TO_FSB(mp, newlim->d_rtb_hardlimit) :
+			be64_to_cpu(ddq->d_rtb_hardlimit);
+	soft = (newlim->d_fieldmask & FS_DQ_RTBSOFT) ?
+		(xfs_qcnt_t) XFS_BB_TO_FSB(mp, newlim->d_rtb_softlimit) :
+			be64_to_cpu(ddq->d_rtb_softlimit);
+	if (hard == 0 || hard >= soft) {
+		ddq->d_rtb_hardlimit = cpu_to_be64(hard);
+		ddq->d_rtb_softlimit = cpu_to_be64(soft);
+		if (id == 0) {
+			q->qi_rtbhardlimit = hard;
+			q->qi_rtbsoftlimit = soft;
+		}
+	} else {
+		xfs_debug(mp, "rtbhard %Ld < rtbsoft %Ld\n", hard, soft);
+	}
+
+	hard = (newlim->d_fieldmask & FS_DQ_IHARD) ?
+		(xfs_qcnt_t) newlim->d_ino_hardlimit :
+			be64_to_cpu(ddq->d_ino_hardlimit);
+	soft = (newlim->d_fieldmask & FS_DQ_ISOFT) ?
+		(xfs_qcnt_t) newlim->d_ino_softlimit :
+			be64_to_cpu(ddq->d_ino_softlimit);
+	if (hard == 0 || hard >= soft) {
+		ddq->d_ino_hardlimit = cpu_to_be64(hard);
+		ddq->d_ino_softlimit = cpu_to_be64(soft);
+		if (id == 0) {
+			q->qi_ihardlimit = hard;
+			q->qi_isoftlimit = soft;
+		}
+	} else {
+		xfs_debug(mp, "ihard %Ld < isoft %Ld\n", hard, soft);
+	}
+
+	/*
+	 * Update warnings counter(s) if requested
+	 */
+	if (newlim->d_fieldmask & FS_DQ_BWARNS)
+		ddq->d_bwarns = cpu_to_be16(newlim->d_bwarns);
+	if (newlim->d_fieldmask & FS_DQ_IWARNS)
+		ddq->d_iwarns = cpu_to_be16(newlim->d_iwarns);
+	if (newlim->d_fieldmask & FS_DQ_RTBWARNS)
+		ddq->d_rtbwarns = cpu_to_be16(newlim->d_rtbwarns);
+
+	if (id == 0) {
+		/*
+		 * Timelimits for the super user set the relative time
+		 * the other users can be over quota for this file system.
+		 * If it is zero a default is used.  Ditto for the default
+		 * soft and hard limit values (already done, above), and
+		 * for warnings.
+		 */
+		if (newlim->d_fieldmask & FS_DQ_BTIMER) {
+			q->qi_btimelimit = newlim->d_btimer;
+			ddq->d_btimer = cpu_to_be32(newlim->d_btimer);
+		}
+		if (newlim->d_fieldmask & FS_DQ_ITIMER) {
+			q->qi_itimelimit = newlim->d_itimer;
+			ddq->d_itimer = cpu_to_be32(newlim->d_itimer);
+		}
+		if (newlim->d_fieldmask & FS_DQ_RTBTIMER) {
+			q->qi_rtbtimelimit = newlim->d_rtbtimer;
+			ddq->d_rtbtimer = cpu_to_be32(newlim->d_rtbtimer);
+		}
+		if (newlim->d_fieldmask & FS_DQ_BWARNS)
+			q->qi_bwarnlimit = newlim->d_bwarns;
+		if (newlim->d_fieldmask & FS_DQ_IWARNS)
+			q->qi_iwarnlimit = newlim->d_iwarns;
+		if (newlim->d_fieldmask & FS_DQ_RTBWARNS)
+			q->qi_rtbwarnlimit = newlim->d_rtbwarns;
+	} else {
+		/*
+		 * If the user is now over quota, start the timelimit.
+		 * The user will not be 'warned'.
+		 * Note that we keep the timers ticking, whether enforcement
+		 * is on or off. We don't really want to bother with iterating
+		 * over all ondisk dquots and turning the timers on/off.
+		 */
+		xfs_qm_adjust_dqtimers(mp, ddq);
+	}
+	dqp->dq_flags |= XFS_DQ_DIRTY;
+	xfs_trans_log_dquot(tp, dqp);
+
+	error = xfs_trans_commit(tp, 0);
+	xfs_qm_dqprint(dqp);
+	xfs_qm_dqrele(dqp);
+
+ out_unlock:
+	mutex_unlock(&q->qi_quotaofflock);
+	return error;
+}
+
+int
+xfs_qm_scall_getquota(
+	xfs_mount_t	*mp,
+	xfs_dqid_t	id,
+	uint		type,
+	fs_disk_quota_t *out)
+{
+	xfs_dquot_t	*dqp;
+	int		error;
+
+	/*
+	 * Try to get the dquot. We don't want it allocated on disk, so
+	 * we aren't passing the XFS_QMOPT_DOALLOC flag. If it doesn't
+	 * exist, we'll get ENOENT back.
+	 */
+	if ((error = xfs_qm_dqget(mp, NULL, id, type, 0, &dqp))) {
+		return (error);
+	}
+
+	/*
+	 * If everything's NULL, this dquot doesn't quite exist as far as
+	 * our utility programs are concerned.
+	 */
+	if (XFS_IS_DQUOT_UNINITIALIZED(dqp)) {
+		xfs_qm_dqput(dqp);
+		return XFS_ERROR(ENOENT);
+	}
+	/* xfs_qm_dqprint(dqp); */
+	/*
+	 * Convert the disk dquot to the exportable format
+	 */
+	xfs_qm_export_dquot(mp, &dqp->q_core, out);
+	xfs_qm_dqput(dqp);
+	return (error ? XFS_ERROR(EFAULT) : 0);
+}
+
+
+STATIC int
+xfs_qm_log_quotaoff_end(
+	xfs_mount_t		*mp,
+	xfs_qoff_logitem_t	*startqoff,
+	uint			flags)
+{
+	xfs_trans_t		*tp;
+	int			error;
+	xfs_qoff_logitem_t	*qoffi;
+
+	tp = xfs_trans_alloc(mp, XFS_TRANS_QM_QUOTAOFF_END);
+
+	if ((error = xfs_trans_reserve(tp, 0, sizeof(xfs_qoff_logitem_t) * 2,
+				      0, 0, XFS_DEFAULT_LOG_COUNT))) {
+		xfs_trans_cancel(tp, 0);
+		return (error);
+	}
+
+	qoffi = xfs_trans_get_qoff_item(tp, startqoff,
+					flags & XFS_ALL_QUOTA_ACCT);
+	xfs_trans_log_quotaoff_item(tp, qoffi);
+
+	/*
+	 * We have to make sure that the transaction is secure on disk before we
+	 * return and actually stop quota accounting. So, make it synchronous.
+	 * We don't care about quotoff's performance.
+	 */
+	xfs_trans_set_sync(tp);
+	error = xfs_trans_commit(tp, 0);
+	return (error);
+}
+
+
+STATIC int
+xfs_qm_log_quotaoff(
+	xfs_mount_t	       *mp,
+	xfs_qoff_logitem_t     **qoffstartp,
+	uint		       flags)
+{
+	xfs_trans_t	       *tp;
+	int			error;
+	xfs_qoff_logitem_t     *qoffi=NULL;
+	uint			oldsbqflag=0;
+
+	tp = xfs_trans_alloc(mp, XFS_TRANS_QM_QUOTAOFF);
+	if ((error = xfs_trans_reserve(tp, 0,
+				      sizeof(xfs_qoff_logitem_t) * 2 +
+				      mp->m_sb.sb_sectsize + 128,
+				      0,
+				      0,
+				      XFS_DEFAULT_LOG_COUNT))) {
+		goto error0;
+	}
+
+	qoffi = xfs_trans_get_qoff_item(tp, NULL, flags & XFS_ALL_QUOTA_ACCT);
+	xfs_trans_log_quotaoff_item(tp, qoffi);
+
+	spin_lock(&mp->m_sb_lock);
+	oldsbqflag = mp->m_sb.sb_qflags;
+	mp->m_sb.sb_qflags = (mp->m_qflags & ~(flags)) & XFS_MOUNT_QUOTA_ALL;
+	spin_unlock(&mp->m_sb_lock);
+
+	xfs_mod_sb(tp, XFS_SB_QFLAGS);
+
+	/*
+	 * We have to make sure that the transaction is secure on disk before we
+	 * return and actually stop quota accounting. So, make it synchronous.
+	 * We don't care about quotoff's performance.
+	 */
+	xfs_trans_set_sync(tp);
+	error = xfs_trans_commit(tp, 0);
+
+error0:
+	if (error) {
+		xfs_trans_cancel(tp, 0);
+		/*
+		 * No one else is modifying sb_qflags, so this is OK.
+		 * We still hold the quotaofflock.
+		 */
+		spin_lock(&mp->m_sb_lock);
+		mp->m_sb.sb_qflags = oldsbqflag;
+		spin_unlock(&mp->m_sb_lock);
+	}
+	*qoffstartp = qoffi;
+	return (error);
+}
+
+
+/*
+ * Translate an internal style on-disk-dquot to the exportable format.
+ * The main differences are that the counters/limits are all in Basic
+ * Blocks (BBs) instead of the internal FSBs, and all on-disk data has
+ * to be converted to the native endianness.
+ */
+STATIC void
+xfs_qm_export_dquot(
+	xfs_mount_t		*mp,
+	xfs_disk_dquot_t	*src,
+	struct fs_disk_quota	*dst)
+{
+	memset(dst, 0, sizeof(*dst));
+	dst->d_version = FS_DQUOT_VERSION;  /* different from src->d_version */
+	dst->d_flags = xfs_qm_export_qtype_flags(src->d_flags);
+	dst->d_id = be32_to_cpu(src->d_id);
+	dst->d_blk_hardlimit =
+		XFS_FSB_TO_BB(mp, be64_to_cpu(src->d_blk_hardlimit));
+	dst->d_blk_softlimit =
+		XFS_FSB_TO_BB(mp, be64_to_cpu(src->d_blk_softlimit));
+	dst->d_ino_hardlimit = be64_to_cpu(src->d_ino_hardlimit);
+	dst->d_ino_softlimit = be64_to_cpu(src->d_ino_softlimit);
+	dst->d_bcount = XFS_FSB_TO_BB(mp, be64_to_cpu(src->d_bcount));
+	dst->d_icount = be64_to_cpu(src->d_icount);
+	dst->d_btimer = be32_to_cpu(src->d_btimer);
+	dst->d_itimer = be32_to_cpu(src->d_itimer);
+	dst->d_iwarns = be16_to_cpu(src->d_iwarns);
+	dst->d_bwarns = be16_to_cpu(src->d_bwarns);
+	dst->d_rtb_hardlimit =
+		XFS_FSB_TO_BB(mp, be64_to_cpu(src->d_rtb_hardlimit));
+	dst->d_rtb_softlimit =
+		XFS_FSB_TO_BB(mp, be64_to_cpu(src->d_rtb_softlimit));
+	dst->d_rtbcount = XFS_FSB_TO_BB(mp, be64_to_cpu(src->d_rtbcount));
+	dst->d_rtbtimer = be32_to_cpu(src->d_rtbtimer);
+	dst->d_rtbwarns = be16_to_cpu(src->d_rtbwarns);
+
+	/*
+	 * Internally, we don't reset all the timers when quota enforcement
+	 * gets turned off. No need to confuse the user level code,
+	 * so return zeroes in that case.
+	 */
+	if ((!XFS_IS_UQUOTA_ENFORCED(mp) && src->d_flags == XFS_DQ_USER) ||
+	    (!XFS_IS_OQUOTA_ENFORCED(mp) &&
+			(src->d_flags & (XFS_DQ_PROJ | XFS_DQ_GROUP)))) {
+		dst->d_btimer = 0;
+		dst->d_itimer = 0;
+		dst->d_rtbtimer = 0;
+	}
+
+#ifdef DEBUG
+	if (((XFS_IS_UQUOTA_ENFORCED(mp) && dst->d_flags == FS_USER_QUOTA) ||
+	     (XFS_IS_OQUOTA_ENFORCED(mp) &&
+			(dst->d_flags & (FS_PROJ_QUOTA | FS_GROUP_QUOTA)))) &&
+	    dst->d_id != 0) {
+		if (((int) dst->d_bcount >= (int) dst->d_blk_softlimit) &&
+		    (dst->d_blk_softlimit > 0)) {
+			ASSERT(dst->d_btimer != 0);
+		}
+		if (((int) dst->d_icount >= (int) dst->d_ino_softlimit) &&
+		    (dst->d_ino_softlimit > 0)) {
+			ASSERT(dst->d_itimer != 0);
+		}
+	}
+#endif
+}
+
+STATIC uint
+xfs_qm_export_qtype_flags(
+	uint flags)
+{
+	/*
+	 * Can't be more than one, or none.
+	 */
+	ASSERT((flags & (FS_PROJ_QUOTA | FS_USER_QUOTA)) !=
+		(FS_PROJ_QUOTA | FS_USER_QUOTA));
+	ASSERT((flags & (FS_PROJ_QUOTA | FS_GROUP_QUOTA)) !=
+		(FS_PROJ_QUOTA | FS_GROUP_QUOTA));
+	ASSERT((flags & (FS_USER_QUOTA | FS_GROUP_QUOTA)) !=
+		(FS_USER_QUOTA | FS_GROUP_QUOTA));
+	ASSERT((flags & (FS_PROJ_QUOTA|FS_USER_QUOTA|FS_GROUP_QUOTA)) != 0);
+
+	return (flags & XFS_DQ_USER) ?
+		FS_USER_QUOTA : (flags & XFS_DQ_PROJ) ?
+			FS_PROJ_QUOTA : FS_GROUP_QUOTA;
+}
+
+STATIC uint
+xfs_qm_export_flags(
+	uint flags)
+{
+	uint uflags;
+
+	uflags = 0;
+	if (flags & XFS_UQUOTA_ACCT)
+		uflags |= FS_QUOTA_UDQ_ACCT;
+	if (flags & XFS_PQUOTA_ACCT)
+		uflags |= FS_QUOTA_PDQ_ACCT;
+	if (flags & XFS_GQUOTA_ACCT)
+		uflags |= FS_QUOTA_GDQ_ACCT;
+	if (flags & XFS_UQUOTA_ENFD)
+		uflags |= FS_QUOTA_UDQ_ENFD;
+	if (flags & (XFS_OQUOTA_ENFD)) {
+		uflags |= (flags & XFS_GQUOTA_ACCT) ?
+			FS_QUOTA_GDQ_ENFD : FS_QUOTA_PDQ_ENFD;
+	}
+	return (uflags);
+}
+
+
+STATIC int
+xfs_dqrele_inode(
+	struct xfs_inode	*ip,
+	struct xfs_perag	*pag,
+	int			flags)
+{
+	/* skip quota inodes */
+	if (ip == ip->i_mount->m_quotainfo->qi_uquotaip ||
+	    ip == ip->i_mount->m_quotainfo->qi_gquotaip) {
+		ASSERT(ip->i_udquot == NULL);
+		ASSERT(ip->i_gdquot == NULL);
+		return 0;
+	}
+
+	xfs_ilock(ip, XFS_ILOCK_EXCL);
+	if ((flags & XFS_UQUOTA_ACCT) && ip->i_udquot) {
+		xfs_qm_dqrele(ip->i_udquot);
+		ip->i_udquot = NULL;
+	}
+	if (flags & (XFS_PQUOTA_ACCT|XFS_GQUOTA_ACCT) && ip->i_gdquot) {
+		xfs_qm_dqrele(ip->i_gdquot);
+		ip->i_gdquot = NULL;
+	}
+	xfs_iunlock(ip, XFS_ILOCK_EXCL);
+	return 0;
+}
+
+
+/*
+ * Go thru all the inodes in the file system, releasing their dquots.
+ *
+ * Note that the mount structure gets modified to indicate that quotas are off
+ * AFTER this, in the case of quotaoff.
+ */
+void
+xfs_qm_dqrele_all_inodes(
+	struct xfs_mount *mp,
+	uint		 flags)
+{
+	ASSERT(mp->m_quotainfo);
+	xfs_inode_ag_iterator(mp, xfs_dqrele_inode, flags);
+}
+
+/*------------------------------------------------------------------------*/
+#ifdef DEBUG
+/*
+ * This contains all the test functions for XFS disk quotas.
+ * Currently it does a quota accounting check. ie. it walks through
+ * all inodes in the file system, calculating the dquot accounting fields,
+ * and prints out any inconsistencies.
+ */
+xfs_dqhash_t *qmtest_udqtab;
+xfs_dqhash_t *qmtest_gdqtab;
+int	      qmtest_hashmask;
+int	      qmtest_nfails;
+struct mutex  qcheck_lock;
+
+#define DQTEST_HASHVAL(mp, id) (((__psunsigned_t)(mp) + \
+				 (__psunsigned_t)(id)) & \
+				(qmtest_hashmask - 1))
+
+#define DQTEST_HASH(mp, id, type)   ((type & XFS_DQ_USER) ? \
+				     (qmtest_udqtab + \
+				      DQTEST_HASHVAL(mp, id)) : \
+				     (qmtest_gdqtab + \
+				      DQTEST_HASHVAL(mp, id)))
+
+#define DQTEST_LIST_PRINT(l, NXT, title) \
+{ \
+	  xfs_dqtest_t	*dqp; int i = 0;\
+	  xfs_debug(NULL, "%s (#%d)", title, (int) (l)->qh_nelems); \
+	  for (dqp = (xfs_dqtest_t *)(l)->qh_next; dqp != NULL; \
+	       dqp = (xfs_dqtest_t *)dqp->NXT) { \
+		xfs_debug(dqp->q_mount,		\
+			"  %d. \"%d (%s)\"  bcnt = %d, icnt = %d", \
+			 ++i, dqp->d_id, DQFLAGTO_TYPESTR(dqp),	     \
+			 dqp->d_bcount, dqp->d_icount); } \
+}
+
+typedef struct dqtest {
+	uint		 dq_flags;	/* various flags (XFS_DQ_*) */
+	struct list_head q_hashlist;
+	xfs_dqhash_t	*q_hash;	/* the hashchain header */
+	xfs_mount_t	*q_mount;	/* filesystem this relates to */
+	xfs_dqid_t	d_id;		/* user id or group id */
+	xfs_qcnt_t	d_bcount;	/* # disk blocks owned by the user */
+	xfs_qcnt_t	d_icount;	/* # inodes owned by the user */
+} xfs_dqtest_t;
+
+STATIC void
+xfs_qm_hashinsert(xfs_dqhash_t *h, xfs_dqtest_t *dqp)
+{
+	list_add(&dqp->q_hashlist, &h->qh_list);
+	h->qh_version++;
+	h->qh_nelems++;
+}
+STATIC void
+xfs_qm_dqtest_print(
+	struct xfs_mount	*mp,
+	struct dqtest		*d)
+{
+	xfs_debug(mp, "-----------DQTEST DQUOT----------------");
+	xfs_debug(mp, "---- dquot ID = %d", d->d_id);
+	xfs_debug(mp, "---- fs       = 0x%p", d->q_mount);
+	xfs_debug(mp, "---- bcount   = %Lu (0x%x)",
+		d->d_bcount, (int)d->d_bcount);
+	xfs_debug(mp, "---- icount   = %Lu (0x%x)",
+		d->d_icount, (int)d->d_icount);
+	xfs_debug(mp, "---------------------------");
+}
+
+STATIC void
+xfs_qm_dqtest_failed(
+	xfs_dqtest_t	*d,
+	xfs_dquot_t	*dqp,
+	char		*reason,
+	xfs_qcnt_t	a,
+	xfs_qcnt_t	b,
+	int		error)
+{
+	qmtest_nfails++;
+	if (error)
+		xfs_debug(dqp->q_mount,
+			"quotacheck failed id=%d, err=%d\nreason: %s",
+			d->d_id, error, reason);
+	else
+		xfs_debug(dqp->q_mount,
+			"quotacheck failed id=%d (%s) [%d != %d]",
+			d->d_id, reason, (int)a, (int)b);
+	xfs_qm_dqtest_print(dqp->q_mount, d);
+	if (dqp)
+		xfs_qm_dqprint(dqp);
+}
+
+STATIC int
+xfs_dqtest_cmp2(
+	xfs_dqtest_t	*d,
+	xfs_dquot_t	*dqp)
+{
+	int err = 0;
+	if (be64_to_cpu(dqp->q_core.d_icount) != d->d_icount) {
+		xfs_qm_dqtest_failed(d, dqp, "icount mismatch",
+			be64_to_cpu(dqp->q_core.d_icount),
+			d->d_icount, 0);
+		err++;
+	}
+	if (be64_to_cpu(dqp->q_core.d_bcount) != d->d_bcount) {
+		xfs_qm_dqtest_failed(d, dqp, "bcount mismatch",
+			be64_to_cpu(dqp->q_core.d_bcount),
+			d->d_bcount, 0);
+		err++;
+	}
+	if (dqp->q_core.d_blk_softlimit &&
+	    be64_to_cpu(dqp->q_core.d_bcount) >=
+	    be64_to_cpu(dqp->q_core.d_blk_softlimit)) {
+		if (!dqp->q_core.d_btimer && dqp->q_core.d_id) {
+			xfs_debug(dqp->q_mount,
+				"%d [%s] BLK TIMER NOT STARTED",
+				d->d_id, DQFLAGTO_TYPESTR(d));
+			err++;
+		}
+	}
+	if (dqp->q_core.d_ino_softlimit &&
+	    be64_to_cpu(dqp->q_core.d_icount) >=
+	    be64_to_cpu(dqp->q_core.d_ino_softlimit)) {
+		if (!dqp->q_core.d_itimer && dqp->q_core.d_id) {
+			xfs_debug(dqp->q_mount,
+				"%d [%s] INO TIMER NOT STARTED",
+				d->d_id, DQFLAGTO_TYPESTR(d));
+			err++;
+		}
+	}
+#ifdef QUOTADEBUG
+	if (!err) {
+		xfs_debug(dqp->q_mount, "%d [%s] qchecked",
+			d->d_id, DQFLAGTO_TYPESTR(d));
+	}
+#endif
+	return (err);
+}
+
+STATIC void
+xfs_dqtest_cmp(
+	xfs_dqtest_t	*d)
+{
+	xfs_dquot_t	*dqp;
+	int		error;
+
+	/* xfs_qm_dqtest_print(d); */
+	if ((error = xfs_qm_dqget(d->q_mount, NULL, d->d_id, d->dq_flags, 0,
+				 &dqp))) {
+		xfs_qm_dqtest_failed(d, NULL, "dqget failed", 0, 0, error);
+		return;
+	}
+	xfs_dqtest_cmp2(d, dqp);
+	xfs_qm_dqput(dqp);
+}
+
+STATIC int
+xfs_qm_internalqcheck_dqget(
+	xfs_mount_t	*mp,
+	xfs_dqid_t	id,
+	uint		type,
+	xfs_dqtest_t	**O_dq)
+{
+	xfs_dqtest_t	*d;
+	xfs_dqhash_t	*h;
+
+	h = DQTEST_HASH(mp, id, type);
+	list_for_each_entry(d, &h->qh_list, q_hashlist) {
+		if (d->d_id == id && mp == d->q_mount) {
+			*O_dq = d;
+			return (0);
+		}
+	}
+	d = kmem_zalloc(sizeof(xfs_dqtest_t), KM_SLEEP);
+	d->dq_flags = type;
+	d->d_id = id;
+	d->q_mount = mp;
+	d->q_hash = h;
+	INIT_LIST_HEAD(&d->q_hashlist);
+	xfs_qm_hashinsert(h, d);
+	*O_dq = d;
+	return (0);
+}
+
+STATIC void
+xfs_qm_internalqcheck_get_dquots(
+	xfs_mount_t	*mp,
+	xfs_dqid_t	uid,
+	xfs_dqid_t	projid,
+	xfs_dqid_t	gid,
+	xfs_dqtest_t	**ud,
+	xfs_dqtest_t	**gd)
+{
+	if (XFS_IS_UQUOTA_ON(mp))
+		xfs_qm_internalqcheck_dqget(mp, uid, XFS_DQ_USER, ud);
+	if (XFS_IS_GQUOTA_ON(mp))
+		xfs_qm_internalqcheck_dqget(mp, gid, XFS_DQ_GROUP, gd);
+	else if (XFS_IS_PQUOTA_ON(mp))
+		xfs_qm_internalqcheck_dqget(mp, projid, XFS_DQ_PROJ, gd);
+}
+
+
+STATIC void
+xfs_qm_internalqcheck_dqadjust(
+	xfs_inode_t		*ip,
+	xfs_dqtest_t		*d)
+{
+	d->d_icount++;
+	d->d_bcount += (xfs_qcnt_t)ip->i_d.di_nblocks;
+}
+
+STATIC int
+xfs_qm_internalqcheck_adjust(
+	xfs_mount_t	*mp,		/* mount point for filesystem */
+	xfs_ino_t	ino,		/* inode number to get data for */
+	void		__user *buffer,	/* not used */
+	int		ubsize,		/* not used */
+	int		*ubused,	/* not used */
+	int		*res)		/* bulkstat result code */
+{
+	xfs_inode_t		*ip;
+	xfs_dqtest_t		*ud, *gd;
+	uint			lock_flags;
+	boolean_t		ipreleased;
+	int			error;
+
+	ASSERT(XFS_IS_QUOTA_RUNNING(mp));
+
+	if (ino == mp->m_sb.sb_uquotino || ino == mp->m_sb.sb_gquotino) {
+		*res = BULKSTAT_RV_NOTHING;
+		xfs_debug(mp, "%s: ino=%llu, uqino=%llu, gqino=%llu\n",
+			__func__, (unsigned long long) ino,
+			(unsigned long long) mp->m_sb.sb_uquotino,
+			(unsigned long long) mp->m_sb.sb_gquotino);
+		return XFS_ERROR(EINVAL);
+	}
+	ipreleased = B_FALSE;
+ again:
+	lock_flags = XFS_ILOCK_SHARED;
+	if ((error = xfs_iget(mp, NULL, ino, 0, lock_flags, &ip))) {
+		*res = BULKSTAT_RV_NOTHING;
+		return (error);
+	}
+
+	/*
+	 * This inode can have blocks after eof which can get released
+	 * when we send it to inactive. Since we don't check the dquot
+	 * until the after all our calculations are done, we must get rid
+	 * of those now.
+	 */
+	if (! ipreleased) {
+		xfs_iunlock(ip, lock_flags);
+		IRELE(ip);
+		ipreleased = B_TRUE;
+		goto again;
+	}
+	xfs_qm_internalqcheck_get_dquots(mp,
+					(xfs_dqid_t) ip->i_d.di_uid,
+					(xfs_dqid_t) xfs_get_projid(ip),
+					(xfs_dqid_t) ip->i_d.di_gid,
+					&ud, &gd);
+	if (XFS_IS_UQUOTA_ON(mp)) {
+		ASSERT(ud);
+		xfs_qm_internalqcheck_dqadjust(ip, ud);
+	}
+	if (XFS_IS_OQUOTA_ON(mp)) {
+		ASSERT(gd);
+		xfs_qm_internalqcheck_dqadjust(ip, gd);
+	}
+	xfs_iunlock(ip, lock_flags);
+	IRELE(ip);
+	*res = BULKSTAT_RV_DIDONE;
+	return (0);
+}
+
+
+/* PRIVATE, debugging */
+int
+xfs_qm_internalqcheck(
+	xfs_mount_t	*mp)
+{
+	xfs_ino_t	lastino;
+	int		done, count;
+	int		i;
+	int		error;
+
+	lastino = 0;
+	qmtest_hashmask = 32;
+	count = 5;
+	done = 0;
+	qmtest_nfails = 0;
+
+	if (! XFS_IS_QUOTA_ON(mp))
+		return XFS_ERROR(ESRCH);
+
+	xfs_log_force(mp, XFS_LOG_SYNC);
+	XFS_bflush(mp->m_ddev_targp);
+	xfs_log_force(mp, XFS_LOG_SYNC);
+	XFS_bflush(mp->m_ddev_targp);
+
+	mutex_lock(&qcheck_lock);
+	/* There should be absolutely no quota activity while this
+	   is going on. */
+	qmtest_udqtab = kmem_zalloc(qmtest_hashmask *
+				    sizeof(xfs_dqhash_t), KM_SLEEP);
+	qmtest_gdqtab = kmem_zalloc(qmtest_hashmask *
+				    sizeof(xfs_dqhash_t), KM_SLEEP);
+	do {
+		/*
+		 * Iterate thru all the inodes in the file system,
+		 * adjusting the corresponding dquot counters
+		 */
+		error = xfs_bulkstat(mp, &lastino, &count,
+				 xfs_qm_internalqcheck_adjust,
+				 0, NULL, &done);
+		if (error) {
+			xfs_debug(mp, "Bulkstat returned error 0x%x", error);
+			break;
+		}
+	} while (!done);
+
+	xfs_debug(mp, "Checking results against system dquots");
+	for (i = 0; i < qmtest_hashmask; i++) {
+		xfs_dqtest_t	*d, *n;
+		xfs_dqhash_t	*h;
+
+		h = &qmtest_udqtab[i];
+		list_for_each_entry_safe(d, n, &h->qh_list, q_hashlist) {
+			xfs_dqtest_cmp(d);
+			kmem_free(d);
+		}
+		h = &qmtest_gdqtab[i];
+		list_for_each_entry_safe(d, n, &h->qh_list, q_hashlist) {
+			xfs_dqtest_cmp(d);
+			kmem_free(d);
+		}
+	}
+
+	if (qmtest_nfails) {
+		xfs_debug(mp, "******** quotacheck failed  ********");
+		xfs_debug(mp, "failures = %d", qmtest_nfails);
+	} else {
+		xfs_debug(mp, "******** quotacheck successful! ********");
+	}
+	kmem_free(qmtest_udqtab);
+	kmem_free(qmtest_gdqtab);
+	mutex_unlock(&qcheck_lock);
+	return (qmtest_nfails);
+}
+
+#endif /* DEBUG */
diff --git a/fs/xfs/quota/xfs_quota_priv.h b/fs/xfs/quota/xfs_quota_priv.h
new file mode 100644
index 00000000..94a3d927
--- /dev/null
+++ b/fs/xfs/quota/xfs_quota_priv.h
@@ -0,0 +1,53 @@
+/*
+ * Copyright (c) 2000-2003 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_QUOTA_PRIV_H__
+#define __XFS_QUOTA_PRIV_H__
+
+/*
+ * Number of bmaps that we ask from bmapi when doing a quotacheck.
+ * We make this restriction to keep the memory usage to a minimum.
+ */
+#define XFS_DQITER_MAP_SIZE	10
+
+/*
+ * Hash into a bucket in the dquot hash table, based on <mp, id>.
+ */
+#define XFS_DQ_HASHVAL(mp, id) (((__psunsigned_t)(mp) + \
+				 (__psunsigned_t)(id)) & \
+				(xfs_Gqm->qm_dqhashmask - 1))
+#define XFS_DQ_HASH(mp, id, type)   (type == XFS_DQ_USER ? \
+				     (xfs_Gqm->qm_usr_dqhtable + \
+				      XFS_DQ_HASHVAL(mp, id)) : \
+				     (xfs_Gqm->qm_grp_dqhtable + \
+				      XFS_DQ_HASHVAL(mp, id)))
+#define XFS_IS_DQUOT_UNINITIALIZED(dqp) ( \
+	!dqp->q_core.d_blk_hardlimit && \
+	!dqp->q_core.d_blk_softlimit && \
+	!dqp->q_core.d_rtb_hardlimit && \
+	!dqp->q_core.d_rtb_softlimit && \
+	!dqp->q_core.d_ino_hardlimit && \
+	!dqp->q_core.d_ino_softlimit && \
+	!dqp->q_core.d_bcount && \
+	!dqp->q_core.d_rtbcount && \
+	!dqp->q_core.d_icount)
+
+#define DQFLAGTO_TYPESTR(d)	(((d)->dq_flags & XFS_DQ_USER) ? "USR" : \
+				 (((d)->dq_flags & XFS_DQ_GROUP) ? "GRP" : \
+				 (((d)->dq_flags & XFS_DQ_PROJ) ? "PRJ":"???")))
+
+#endif	/* __XFS_QUOTA_PRIV_H__ */
diff --git a/fs/xfs/quota/xfs_trans_dquot.c b/fs/xfs/quota/xfs_trans_dquot.c
new file mode 100644
index 00000000..2a364873
--- /dev/null
+++ b/fs/xfs/quota/xfs_trans_dquot.c
@@ -0,0 +1,895 @@
+/*
+ * Copyright (c) 2000-2002 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_alloc.h"
+#include "xfs_quota.h"
+#include "xfs_mount.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_inode.h"
+#include "xfs_itable.h"
+#include "xfs_bmap.h"
+#include "xfs_rtalloc.h"
+#include "xfs_error.h"
+#include "xfs_attr.h"
+#include "xfs_buf_item.h"
+#include "xfs_trans_priv.h"
+#include "xfs_qm.h"
+
+STATIC void	xfs_trans_alloc_dqinfo(xfs_trans_t *);
+
+/*
+ * Add the locked dquot to the transaction.
+ * The dquot must be locked, and it cannot be associated with any
+ * transaction.
+ */
+void
+xfs_trans_dqjoin(
+	xfs_trans_t	*tp,
+	xfs_dquot_t	*dqp)
+{
+	ASSERT(dqp->q_transp != tp);
+	ASSERT(XFS_DQ_IS_LOCKED(dqp));
+	ASSERT(dqp->q_logitem.qli_dquot == dqp);
+
+	/*
+	 * Get a log_item_desc to point at the new item.
+	 */
+	xfs_trans_add_item(tp, &dqp->q_logitem.qli_item);
+
+	/*
+	 * Initialize i_transp so we can later determine if this dquot is
+	 * associated with this transaction.
+	 */
+	dqp->q_transp = tp;
+}
+
+
+/*
+ * This is called to mark the dquot as needing
+ * to be logged when the transaction is committed.  The dquot must
+ * already be associated with the given transaction.
+ * Note that it marks the entire transaction as dirty. In the ordinary
+ * case, this gets called via xfs_trans_commit, after the transaction
+ * is already dirty. However, there's nothing stop this from getting
+ * called directly, as done by xfs_qm_scall_setqlim. Hence, the TRANS_DIRTY
+ * flag.
+ */
+void
+xfs_trans_log_dquot(
+	xfs_trans_t	*tp,
+	xfs_dquot_t	*dqp)
+{
+	ASSERT(dqp->q_transp == tp);
+	ASSERT(XFS_DQ_IS_LOCKED(dqp));
+
+	tp->t_flags |= XFS_TRANS_DIRTY;
+	dqp->q_logitem.qli_item.li_desc->lid_flags |= XFS_LID_DIRTY;
+}
+
+/*
+ * Carry forward whatever is left of the quota blk reservation to
+ * the spanky new transaction
+ */
+void
+xfs_trans_dup_dqinfo(
+	xfs_trans_t	*otp,
+	xfs_trans_t	*ntp)
+{
+	xfs_dqtrx_t	*oq, *nq;
+	int		i,j;
+	xfs_dqtrx_t	*oqa, *nqa;
+
+	if (!otp->t_dqinfo)
+		return;
+
+	xfs_trans_alloc_dqinfo(ntp);
+	oqa = otp->t_dqinfo->dqa_usrdquots;
+	nqa = ntp->t_dqinfo->dqa_usrdquots;
+
+	/*
+	 * Because the quota blk reservation is carried forward,
+	 * it is also necessary to carry forward the DQ_DIRTY flag.
+	 */
+	if(otp->t_flags & XFS_TRANS_DQ_DIRTY)
+		ntp->t_flags |= XFS_TRANS_DQ_DIRTY;
+
+	for (j = 0; j < 2; j++) {
+		for (i = 0; i < XFS_QM_TRANS_MAXDQS; i++) {
+			if (oqa[i].qt_dquot == NULL)
+				break;
+			oq = &oqa[i];
+			nq = &nqa[i];
+
+			nq->qt_dquot = oq->qt_dquot;
+			nq->qt_bcount_delta = nq->qt_icount_delta = 0;
+			nq->qt_rtbcount_delta = 0;
+
+			/*
+			 * Transfer whatever is left of the reservations.
+			 */
+			nq->qt_blk_res = oq->qt_blk_res - oq->qt_blk_res_used;
+			oq->qt_blk_res = oq->qt_blk_res_used;
+
+			nq->qt_rtblk_res = oq->qt_rtblk_res -
+				oq->qt_rtblk_res_used;
+			oq->qt_rtblk_res = oq->qt_rtblk_res_used;
+
+			nq->qt_ino_res = oq->qt_ino_res - oq->qt_ino_res_used;
+			oq->qt_ino_res = oq->qt_ino_res_used;
+
+		}
+		oqa = otp->t_dqinfo->dqa_grpdquots;
+		nqa = ntp->t_dqinfo->dqa_grpdquots;
+	}
+}
+
+/*
+ * Wrap around mod_dquot to account for both user and group quotas.
+ */
+void
+xfs_trans_mod_dquot_byino(
+	xfs_trans_t	*tp,
+	xfs_inode_t	*ip,
+	uint		field,
+	long		delta)
+{
+	xfs_mount_t	*mp = tp->t_mountp;
+
+	if (!XFS_IS_QUOTA_RUNNING(mp) ||
+	    !XFS_IS_QUOTA_ON(mp) ||
+	    ip->i_ino == mp->m_sb.sb_uquotino ||
+	    ip->i_ino == mp->m_sb.sb_gquotino)
+		return;
+
+	if (tp->t_dqinfo == NULL)
+		xfs_trans_alloc_dqinfo(tp);
+
+	if (XFS_IS_UQUOTA_ON(mp) && ip->i_udquot)
+		(void) xfs_trans_mod_dquot(tp, ip->i_udquot, field, delta);
+	if (XFS_IS_OQUOTA_ON(mp) && ip->i_gdquot)
+		(void) xfs_trans_mod_dquot(tp, ip->i_gdquot, field, delta);
+}
+
+STATIC xfs_dqtrx_t *
+xfs_trans_get_dqtrx(
+	xfs_trans_t	*tp,
+	xfs_dquot_t	*dqp)
+{
+	int		i;
+	xfs_dqtrx_t	*qa;
+
+	qa = XFS_QM_ISUDQ(dqp) ?
+		tp->t_dqinfo->dqa_usrdquots : tp->t_dqinfo->dqa_grpdquots;
+
+	for (i = 0; i < XFS_QM_TRANS_MAXDQS; i++) {
+		if (qa[i].qt_dquot == NULL ||
+		    qa[i].qt_dquot == dqp)
+			return &qa[i];
+	}
+
+	return NULL;
+}
+
+/*
+ * Make the changes in the transaction structure.
+ * The moral equivalent to xfs_trans_mod_sb().
+ * We don't touch any fields in the dquot, so we don't care
+ * if it's locked or not (most of the time it won't be).
+ */
+void
+xfs_trans_mod_dquot(
+	xfs_trans_t	*tp,
+	xfs_dquot_t	*dqp,
+	uint		field,
+	long		delta)
+{
+	xfs_dqtrx_t	*qtrx;
+
+	ASSERT(tp);
+	ASSERT(XFS_IS_QUOTA_RUNNING(tp->t_mountp));
+	qtrx = NULL;
+
+	if (tp->t_dqinfo == NULL)
+		xfs_trans_alloc_dqinfo(tp);
+	/*
+	 * Find either the first free slot or the slot that belongs
+	 * to this dquot.
+	 */
+	qtrx = xfs_trans_get_dqtrx(tp, dqp);
+	ASSERT(qtrx);
+	if (qtrx->qt_dquot == NULL)
+		qtrx->qt_dquot = dqp;
+
+	switch (field) {
+
+		/*
+		 * regular disk blk reservation
+		 */
+	      case XFS_TRANS_DQ_RES_BLKS:
+		qtrx->qt_blk_res += (ulong)delta;
+		break;
+
+		/*
+		 * inode reservation
+		 */
+	      case XFS_TRANS_DQ_RES_INOS:
+		qtrx->qt_ino_res += (ulong)delta;
+		break;
+
+		/*
+		 * disk blocks used.
+		 */
+	      case XFS_TRANS_DQ_BCOUNT:
+		if (qtrx->qt_blk_res && delta > 0) {
+			qtrx->qt_blk_res_used += (ulong)delta;
+			ASSERT(qtrx->qt_blk_res >= qtrx->qt_blk_res_used);
+		}
+		qtrx->qt_bcount_delta += delta;
+		break;
+
+	      case XFS_TRANS_DQ_DELBCOUNT:
+		qtrx->qt_delbcnt_delta += delta;
+		break;
+
+		/*
+		 * Inode Count
+		 */
+	      case XFS_TRANS_DQ_ICOUNT:
+		if (qtrx->qt_ino_res && delta > 0) {
+			qtrx->qt_ino_res_used += (ulong)delta;
+			ASSERT(qtrx->qt_ino_res >= qtrx->qt_ino_res_used);
+		}
+		qtrx->qt_icount_delta += delta;
+		break;
+
+		/*
+		 * rtblk reservation
+		 */
+	      case XFS_TRANS_DQ_RES_RTBLKS:
+		qtrx->qt_rtblk_res += (ulong)delta;
+		break;
+
+		/*
+		 * rtblk count
+		 */
+	      case XFS_TRANS_DQ_RTBCOUNT:
+		if (qtrx->qt_rtblk_res && delta > 0) {
+			qtrx->qt_rtblk_res_used += (ulong)delta;
+			ASSERT(qtrx->qt_rtblk_res >= qtrx->qt_rtblk_res_used);
+		}
+		qtrx->qt_rtbcount_delta += delta;
+		break;
+
+	      case XFS_TRANS_DQ_DELRTBCOUNT:
+		qtrx->qt_delrtb_delta += delta;
+		break;
+
+	      default:
+		ASSERT(0);
+	}
+	tp->t_flags |= XFS_TRANS_DQ_DIRTY;
+}
+
+
+/*
+ * Given an array of dqtrx structures, lock all the dquots associated
+ * and join them to the transaction, provided they have been modified.
+ * We know that the highest number of dquots (of one type - usr OR grp),
+ * involved in a transaction is 2 and that both usr and grp combined - 3.
+ * So, we don't attempt to make this very generic.
+ */
+STATIC void
+xfs_trans_dqlockedjoin(
+	xfs_trans_t	*tp,
+	xfs_dqtrx_t	*q)
+{
+	ASSERT(q[0].qt_dquot != NULL);
+	if (q[1].qt_dquot == NULL) {
+		xfs_dqlock(q[0].qt_dquot);
+		xfs_trans_dqjoin(tp, q[0].qt_dquot);
+	} else {
+		ASSERT(XFS_QM_TRANS_MAXDQS == 2);
+		xfs_dqlock2(q[0].qt_dquot, q[1].qt_dquot);
+		xfs_trans_dqjoin(tp, q[0].qt_dquot);
+		xfs_trans_dqjoin(tp, q[1].qt_dquot);
+	}
+}
+
+
+/*
+ * Called by xfs_trans_commit() and similar in spirit to
+ * xfs_trans_apply_sb_deltas().
+ * Go thru all the dquots belonging to this transaction and modify the
+ * INCORE dquot to reflect the actual usages.
+ * Unreserve just the reservations done by this transaction.
+ * dquot is still left locked at exit.
+ */
+void
+xfs_trans_apply_dquot_deltas(
+	xfs_trans_t		*tp)
+{
+	int			i, j;
+	xfs_dquot_t		*dqp;
+	xfs_dqtrx_t		*qtrx, *qa;
+	xfs_disk_dquot_t	*d;
+	long			totalbdelta;
+	long			totalrtbdelta;
+
+	if (!(tp->t_flags & XFS_TRANS_DQ_DIRTY))
+		return;
+
+	ASSERT(tp->t_dqinfo);
+	qa = tp->t_dqinfo->dqa_usrdquots;
+	for (j = 0; j < 2; j++) {
+		if (qa[0].qt_dquot == NULL) {
+			qa = tp->t_dqinfo->dqa_grpdquots;
+			continue;
+		}
+
+		/*
+		 * Lock all of the dquots and join them to the transaction.
+		 */
+		xfs_trans_dqlockedjoin(tp, qa);
+
+		for (i = 0; i < XFS_QM_TRANS_MAXDQS; i++) {
+			qtrx = &qa[i];
+			/*
+			 * The array of dquots is filled
+			 * sequentially, not sparsely.
+			 */
+			if ((dqp = qtrx->qt_dquot) == NULL)
+				break;
+
+			ASSERT(XFS_DQ_IS_LOCKED(dqp));
+			ASSERT(dqp->q_transp == tp);
+
+			/*
+			 * adjust the actual number of blocks used
+			 */
+			d = &dqp->q_core;
+
+			/*
+			 * The issue here is - sometimes we don't make a blkquota
+			 * reservation intentionally to be fair to users
+			 * (when the amount is small). On the other hand,
+			 * delayed allocs do make reservations, but that's
+			 * outside of a transaction, so we have no
+			 * idea how much was really reserved.
+			 * So, here we've accumulated delayed allocation blks and
+			 * non-delay blks. The assumption is that the
+			 * delayed ones are always reserved (outside of a
+			 * transaction), and the others may or may not have
+			 * quota reservations.
+			 */
+			totalbdelta = qtrx->qt_bcount_delta +
+				qtrx->qt_delbcnt_delta;
+			totalrtbdelta = qtrx->qt_rtbcount_delta +
+				qtrx->qt_delrtb_delta;
+#ifdef QUOTADEBUG
+			if (totalbdelta < 0)
+				ASSERT(be64_to_cpu(d->d_bcount) >=
+				       (xfs_qcnt_t) -totalbdelta);
+
+			if (totalrtbdelta < 0)
+				ASSERT(be64_to_cpu(d->d_rtbcount) >=
+				       (xfs_qcnt_t) -totalrtbdelta);
+
+			if (qtrx->qt_icount_delta < 0)
+				ASSERT(be64_to_cpu(d->d_icount) >=
+				       (xfs_qcnt_t) -qtrx->qt_icount_delta);
+#endif
+			if (totalbdelta)
+				be64_add_cpu(&d->d_bcount, (xfs_qcnt_t)totalbdelta);
+
+			if (qtrx->qt_icount_delta)
+				be64_add_cpu(&d->d_icount, (xfs_qcnt_t)qtrx->qt_icount_delta);
+
+			if (totalrtbdelta)
+				be64_add_cpu(&d->d_rtbcount, (xfs_qcnt_t)totalrtbdelta);
+
+			/*
+			 * Get any default limits in use.
+			 * Start/reset the timer(s) if needed.
+			 */
+			if (d->d_id) {
+				xfs_qm_adjust_dqlimits(tp->t_mountp, d);
+				xfs_qm_adjust_dqtimers(tp->t_mountp, d);
+			}
+
+			dqp->dq_flags |= XFS_DQ_DIRTY;
+			/*
+			 * add this to the list of items to get logged
+			 */
+			xfs_trans_log_dquot(tp, dqp);
+			/*
+			 * Take off what's left of the original reservation.
+			 * In case of delayed allocations, there's no
+			 * reservation that a transaction structure knows of.
+			 */
+			if (qtrx->qt_blk_res != 0) {
+				if (qtrx->qt_blk_res != qtrx->qt_blk_res_used) {
+					if (qtrx->qt_blk_res >
+					    qtrx->qt_blk_res_used)
+						dqp->q_res_bcount -= (xfs_qcnt_t)
+							(qtrx->qt_blk_res -
+							 qtrx->qt_blk_res_used);
+					else
+						dqp->q_res_bcount -= (xfs_qcnt_t)
+							(qtrx->qt_blk_res_used -
+							 qtrx->qt_blk_res);
+				}
+			} else {
+				/*
+				 * These blks were never reserved, either inside
+				 * a transaction or outside one (in a delayed
+				 * allocation). Also, this isn't always a
+				 * negative number since we sometimes
+				 * deliberately skip quota reservations.
+				 */
+				if (qtrx->qt_bcount_delta) {
+					dqp->q_res_bcount +=
+					      (xfs_qcnt_t)qtrx->qt_bcount_delta;
+				}
+			}
+			/*
+			 * Adjust the RT reservation.
+			 */
+			if (qtrx->qt_rtblk_res != 0) {
+				if (qtrx->qt_rtblk_res != qtrx->qt_rtblk_res_used) {
+					if (qtrx->qt_rtblk_res >
+					    qtrx->qt_rtblk_res_used)
+					       dqp->q_res_rtbcount -= (xfs_qcnt_t)
+						       (qtrx->qt_rtblk_res -
+							qtrx->qt_rtblk_res_used);
+					else
+					       dqp->q_res_rtbcount -= (xfs_qcnt_t)
+						       (qtrx->qt_rtblk_res_used -
+							qtrx->qt_rtblk_res);
+				}
+			} else {
+				if (qtrx->qt_rtbcount_delta)
+					dqp->q_res_rtbcount +=
+					    (xfs_qcnt_t)qtrx->qt_rtbcount_delta;
+			}
+
+			/*
+			 * Adjust the inode reservation.
+			 */
+			if (qtrx->qt_ino_res != 0) {
+				ASSERT(qtrx->qt_ino_res >=
+				       qtrx->qt_ino_res_used);
+				if (qtrx->qt_ino_res > qtrx->qt_ino_res_used)
+					dqp->q_res_icount -= (xfs_qcnt_t)
+						(qtrx->qt_ino_res -
+						 qtrx->qt_ino_res_used);
+			} else {
+				if (qtrx->qt_icount_delta)
+					dqp->q_res_icount +=
+					    (xfs_qcnt_t)qtrx->qt_icount_delta;
+			}
+
+			ASSERT(dqp->q_res_bcount >=
+				be64_to_cpu(dqp->q_core.d_bcount));
+			ASSERT(dqp->q_res_icount >=
+				be64_to_cpu(dqp->q_core.d_icount));
+			ASSERT(dqp->q_res_rtbcount >=
+				be64_to_cpu(dqp->q_core.d_rtbcount));
+		}
+		/*
+		 * Do the group quotas next
+		 */
+		qa = tp->t_dqinfo->dqa_grpdquots;
+	}
+}
+
+/*
+ * Release the reservations, and adjust the dquots accordingly.
+ * This is called only when the transaction is being aborted. If by
+ * any chance we have done dquot modifications incore (ie. deltas) already,
+ * we simply throw those away, since that's the expected behavior
+ * when a transaction is curtailed without a commit.
+ */
+void
+xfs_trans_unreserve_and_mod_dquots(
+	xfs_trans_t		*tp)
+{
+	int			i, j;
+	xfs_dquot_t		*dqp;
+	xfs_dqtrx_t		*qtrx, *qa;
+	boolean_t		locked;
+
+	if (!tp->t_dqinfo || !(tp->t_flags & XFS_TRANS_DQ_DIRTY))
+		return;
+
+	qa = tp->t_dqinfo->dqa_usrdquots;
+
+	for (j = 0; j < 2; j++) {
+		for (i = 0; i < XFS_QM_TRANS_MAXDQS; i++) {
+			qtrx = &qa[i];
+			/*
+			 * We assume that the array of dquots is filled
+			 * sequentially, not sparsely.
+			 */
+			if ((dqp = qtrx->qt_dquot) == NULL)
+				break;
+			/*
+			 * Unreserve the original reservation. We don't care
+			 * about the number of blocks used field, or deltas.
+			 * Also we don't bother to zero the fields.
+			 */
+			locked = B_FALSE;
+			if (qtrx->qt_blk_res) {
+				xfs_dqlock(dqp);
+				locked = B_TRUE;
+				dqp->q_res_bcount -=
+					(xfs_qcnt_t)qtrx->qt_blk_res;
+			}
+			if (qtrx->qt_ino_res) {
+				if (!locked) {
+					xfs_dqlock(dqp);
+					locked = B_TRUE;
+				}
+				dqp->q_res_icount -=
+					(xfs_qcnt_t)qtrx->qt_ino_res;
+			}
+
+			if (qtrx->qt_rtblk_res) {
+				if (!locked) {
+					xfs_dqlock(dqp);
+					locked = B_TRUE;
+				}
+				dqp->q_res_rtbcount -=
+					(xfs_qcnt_t)qtrx->qt_rtblk_res;
+			}
+			if (locked)
+				xfs_dqunlock(dqp);
+
+		}
+		qa = tp->t_dqinfo->dqa_grpdquots;
+	}
+}
+
+STATIC void
+xfs_quota_warn(
+	struct xfs_mount	*mp,
+	struct xfs_dquot	*dqp,
+	int			type)
+{
+	/* no warnings for project quotas - we just return ENOSPC later */
+	if (dqp->dq_flags & XFS_DQ_PROJ)
+		return;
+	quota_send_warning((dqp->dq_flags & XFS_DQ_USER) ? USRQUOTA : GRPQUOTA,
+			   be32_to_cpu(dqp->q_core.d_id), mp->m_super->s_dev,
+			   type);
+}
+
+/*
+ * This reserves disk blocks and inodes against a dquot.
+ * Flags indicate if the dquot is to be locked here and also
+ * if the blk reservation is for RT or regular blocks.
+ * Sending in XFS_QMOPT_FORCE_RES flag skips the quota check.
+ */
+STATIC int
+xfs_trans_dqresv(
+	xfs_trans_t	*tp,
+	xfs_mount_t	*mp,
+	xfs_dquot_t	*dqp,
+	long		nblks,
+	long		ninos,
+	uint		flags)
+{
+	xfs_qcnt_t	hardlimit;
+	xfs_qcnt_t	softlimit;
+	time_t		timer;
+	xfs_qwarncnt_t	warns;
+	xfs_qwarncnt_t	warnlimit;
+	xfs_qcnt_t	count;
+	xfs_qcnt_t	*resbcountp;
+	xfs_quotainfo_t	*q = mp->m_quotainfo;
+
+
+	xfs_dqlock(dqp);
+
+	if (flags & XFS_TRANS_DQ_RES_BLKS) {
+		hardlimit = be64_to_cpu(dqp->q_core.d_blk_hardlimit);
+		if (!hardlimit)
+			hardlimit = q->qi_bhardlimit;
+		softlimit = be64_to_cpu(dqp->q_core.d_blk_softlimit);
+		if (!softlimit)
+			softlimit = q->qi_bsoftlimit;
+		timer = be32_to_cpu(dqp->q_core.d_btimer);
+		warns = be16_to_cpu(dqp->q_core.d_bwarns);
+		warnlimit = dqp->q_mount->m_quotainfo->qi_bwarnlimit;
+		resbcountp = &dqp->q_res_bcount;
+	} else {
+		ASSERT(flags & XFS_TRANS_DQ_RES_RTBLKS);
+		hardlimit = be64_to_cpu(dqp->q_core.d_rtb_hardlimit);
+		if (!hardlimit)
+			hardlimit = q->qi_rtbhardlimit;
+		softlimit = be64_to_cpu(dqp->q_core.d_rtb_softlimit);
+		if (!softlimit)
+			softlimit = q->qi_rtbsoftlimit;
+		timer = be32_to_cpu(dqp->q_core.d_rtbtimer);
+		warns = be16_to_cpu(dqp->q_core.d_rtbwarns);
+		warnlimit = dqp->q_mount->m_quotainfo->qi_rtbwarnlimit;
+		resbcountp = &dqp->q_res_rtbcount;
+	}
+
+	if ((flags & XFS_QMOPT_FORCE_RES) == 0 &&
+	    dqp->q_core.d_id &&
+	    ((XFS_IS_UQUOTA_ENFORCED(dqp->q_mount) && XFS_QM_ISUDQ(dqp)) ||
+	     (XFS_IS_OQUOTA_ENFORCED(dqp->q_mount) &&
+	      (XFS_QM_ISPDQ(dqp) || XFS_QM_ISGDQ(dqp))))) {
+#ifdef QUOTADEBUG
+		xfs_debug(mp,
+			"BLK Res: nblks=%ld + resbcount=%Ld > hardlimit=%Ld?",
+			nblks, *resbcountp, hardlimit);
+#endif
+		if (nblks > 0) {
+			/*
+			 * dquot is locked already. See if we'd go over the
+			 * hardlimit or exceed the timelimit if we allocate
+			 * nblks.
+			 */
+			if (hardlimit > 0ULL &&
+			    hardlimit <= nblks + *resbcountp) {
+				xfs_quota_warn(mp, dqp, QUOTA_NL_BHARDWARN);
+				goto error_return;
+			}
+			if (softlimit > 0ULL &&
+			    softlimit <= nblks + *resbcountp) {
+				if ((timer != 0 && get_seconds() > timer) ||
+				    (warns != 0 && warns >= warnlimit)) {
+					xfs_quota_warn(mp, dqp,
+						       QUOTA_NL_BSOFTLONGWARN);
+					goto error_return;
+				}
+
+				xfs_quota_warn(mp, dqp, QUOTA_NL_BSOFTWARN);
+			}
+		}
+		if (ninos > 0) {
+			count = be64_to_cpu(dqp->q_core.d_icount);
+			timer = be32_to_cpu(dqp->q_core.d_itimer);
+			warns = be16_to_cpu(dqp->q_core.d_iwarns);
+			warnlimit = dqp->q_mount->m_quotainfo->qi_iwarnlimit;
+			hardlimit = be64_to_cpu(dqp->q_core.d_ino_hardlimit);
+			if (!hardlimit)
+				hardlimit = q->qi_ihardlimit;
+			softlimit = be64_to_cpu(dqp->q_core.d_ino_softlimit);
+			if (!softlimit)
+				softlimit = q->qi_isoftlimit;
+
+			if (hardlimit > 0ULL && count >= hardlimit) {
+				xfs_quota_warn(mp, dqp, QUOTA_NL_IHARDWARN);
+				goto error_return;
+			}
+			if (softlimit > 0ULL && count >= softlimit) {
+				if  ((timer != 0 && get_seconds() > timer) ||
+				     (warns != 0 && warns >= warnlimit)) {
+					xfs_quota_warn(mp, dqp,
+						       QUOTA_NL_ISOFTLONGWARN);
+					goto error_return;
+				}
+				xfs_quota_warn(mp, dqp, QUOTA_NL_ISOFTWARN);
+			}
+		}
+	}
+
+	/*
+	 * Change the reservation, but not the actual usage.
+	 * Note that q_res_bcount = q_core.d_bcount + resv
+	 */
+	(*resbcountp) += (xfs_qcnt_t)nblks;
+	if (ninos != 0)
+		dqp->q_res_icount += (xfs_qcnt_t)ninos;
+
+	/*
+	 * note the reservation amt in the trans struct too,
+	 * so that the transaction knows how much was reserved by
+	 * it against this particular dquot.
+	 * We don't do this when we are reserving for a delayed allocation,
+	 * because we don't have the luxury of a transaction envelope then.
+	 */
+	if (tp) {
+		ASSERT(tp->t_dqinfo);
+		ASSERT(flags & XFS_QMOPT_RESBLK_MASK);
+		if (nblks != 0)
+			xfs_trans_mod_dquot(tp, dqp,
+					    flags & XFS_QMOPT_RESBLK_MASK,
+					    nblks);
+		if (ninos != 0)
+			xfs_trans_mod_dquot(tp, dqp,
+					    XFS_TRANS_DQ_RES_INOS,
+					    ninos);
+	}
+	ASSERT(dqp->q_res_bcount >= be64_to_cpu(dqp->q_core.d_bcount));
+	ASSERT(dqp->q_res_rtbcount >= be64_to_cpu(dqp->q_core.d_rtbcount));
+	ASSERT(dqp->q_res_icount >= be64_to_cpu(dqp->q_core.d_icount));
+
+	xfs_dqunlock(dqp);
+	return 0;
+
+error_return:
+	xfs_dqunlock(dqp);
+	if (flags & XFS_QMOPT_ENOSPC)
+		return ENOSPC;
+	return EDQUOT;
+}
+
+
+/*
+ * Given dquot(s), make disk block and/or inode reservations against them.
+ * The fact that this does the reservation against both the usr and
+ * grp/prj quotas is important, because this follows a both-or-nothing
+ * approach.
+ *
+ * flags = XFS_QMOPT_FORCE_RES evades limit enforcement. Used by chown.
+ *	   XFS_QMOPT_ENOSPC returns ENOSPC not EDQUOT.  Used by pquota.
+ *	   XFS_TRANS_DQ_RES_BLKS reserves regular disk blocks
+ *	   XFS_TRANS_DQ_RES_RTBLKS reserves realtime disk blocks
+ * dquots are unlocked on return, if they were not locked by caller.
+ */
+int
+xfs_trans_reserve_quota_bydquots(
+	xfs_trans_t	*tp,
+	xfs_mount_t	*mp,
+	xfs_dquot_t	*udqp,
+	xfs_dquot_t	*gdqp,
+	long		nblks,
+	long		ninos,
+	uint		flags)
+{
+	int		resvd = 0, error;
+
+	if (!XFS_IS_QUOTA_RUNNING(mp) || !XFS_IS_QUOTA_ON(mp))
+		return 0;
+
+	if (tp && tp->t_dqinfo == NULL)
+		xfs_trans_alloc_dqinfo(tp);
+
+	ASSERT(flags & XFS_QMOPT_RESBLK_MASK);
+
+	if (udqp) {
+		error = xfs_trans_dqresv(tp, mp, udqp, nblks, ninos,
+					(flags & ~XFS_QMOPT_ENOSPC));
+		if (error)
+			return error;
+		resvd = 1;
+	}
+
+	if (gdqp) {
+		error = xfs_trans_dqresv(tp, mp, gdqp, nblks, ninos, flags);
+		if (error) {
+			/*
+			 * can't do it, so backout previous reservation
+			 */
+			if (resvd) {
+				flags |= XFS_QMOPT_FORCE_RES;
+				xfs_trans_dqresv(tp, mp, udqp,
+						 -nblks, -ninos, flags);
+			}
+			return error;
+		}
+	}
+
+	/*
+	 * Didn't change anything critical, so, no need to log
+	 */
+	return 0;
+}
+
+
+/*
+ * Lock the dquot and change the reservation if we can.
+ * This doesn't change the actual usage, just the reservation.
+ * The inode sent in is locked.
+ */
+int
+xfs_trans_reserve_quota_nblks(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*ip,
+	long			nblks,
+	long			ninos,
+	uint			flags)
+{
+	struct xfs_mount	*mp = ip->i_mount;
+
+	if (!XFS_IS_QUOTA_RUNNING(mp) || !XFS_IS_QUOTA_ON(mp))
+		return 0;
+	if (XFS_IS_PQUOTA_ON(mp))
+		flags |= XFS_QMOPT_ENOSPC;
+
+	ASSERT(ip->i_ino != mp->m_sb.sb_uquotino);
+	ASSERT(ip->i_ino != mp->m_sb.sb_gquotino);
+
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
+	ASSERT((flags & ~(XFS_QMOPT_FORCE_RES | XFS_QMOPT_ENOSPC)) ==
+				XFS_TRANS_DQ_RES_RTBLKS ||
+	       (flags & ~(XFS_QMOPT_FORCE_RES | XFS_QMOPT_ENOSPC)) ==
+				XFS_TRANS_DQ_RES_BLKS);
+
+	/*
+	 * Reserve nblks against these dquots, with trans as the mediator.
+	 */
+	return xfs_trans_reserve_quota_bydquots(tp, mp,
+						ip->i_udquot, ip->i_gdquot,
+						nblks, ninos, flags);
+}
+
+/*
+ * This routine is called to allocate a quotaoff log item.
+ */
+xfs_qoff_logitem_t *
+xfs_trans_get_qoff_item(
+	xfs_trans_t		*tp,
+	xfs_qoff_logitem_t	*startqoff,
+	uint			flags)
+{
+	xfs_qoff_logitem_t	*q;
+
+	ASSERT(tp != NULL);
+
+	q = xfs_qm_qoff_logitem_init(tp->t_mountp, startqoff, flags);
+	ASSERT(q != NULL);
+
+	/*
+	 * Get a log_item_desc to point at the new item.
+	 */
+	xfs_trans_add_item(tp, &q->qql_item);
+	return q;
+}
+
+
+/*
+ * This is called to mark the quotaoff logitem as needing
+ * to be logged when the transaction is committed.  The logitem must
+ * already be associated with the given transaction.
+ */
+void
+xfs_trans_log_quotaoff_item(
+	xfs_trans_t		*tp,
+	xfs_qoff_logitem_t	*qlp)
+{
+	tp->t_flags |= XFS_TRANS_DIRTY;
+	qlp->qql_item.li_desc->lid_flags |= XFS_LID_DIRTY;
+}
+
+STATIC void
+xfs_trans_alloc_dqinfo(
+	xfs_trans_t	*tp)
+{
+	tp->t_dqinfo = kmem_zone_zalloc(xfs_Gqm->qm_dqtrxzone, KM_SLEEP);
+}
+
+void
+xfs_trans_free_dqinfo(
+	xfs_trans_t	*tp)
+{
+	if (!tp->t_dqinfo)
+		return;
+	kmem_zone_free(xfs_Gqm->qm_dqtrxzone, tp->t_dqinfo);
+	tp->t_dqinfo = NULL;
+}
diff --git a/fs/xfs/support/uuid.c b/fs/xfs/support/uuid.c
new file mode 100644
index 00000000..b83f76b6
--- /dev/null
+++ b/fs/xfs/support/uuid.c
@@ -0,0 +1,63 @@
+/*
+ * Copyright (c) 2000-2003,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include <xfs.h>
+
+/* IRIX interpretation of an uuid_t */
+typedef struct {
+	__be32	uu_timelow;
+	__be16	uu_timemid;
+	__be16	uu_timehi;
+	__be16	uu_clockseq;
+	__be16	uu_node[3];
+} xfs_uu_t;
+
+/*
+ * uuid_getnodeuniq - obtain the node unique fields of a UUID.
+ *
+ * This is not in any way a standard or condoned UUID function;
+ * it just something that's needed for user-level file handles.
+ */
+void
+uuid_getnodeuniq(uuid_t *uuid, int fsid [2])
+{
+	xfs_uu_t *uup = (xfs_uu_t *)uuid;
+
+	fsid[0] = (be16_to_cpu(uup->uu_clockseq) << 16) |
+		   be16_to_cpu(uup->uu_timemid);
+	fsid[1] = be32_to_cpu(uup->uu_timelow);
+}
+
+int
+uuid_is_nil(uuid_t *uuid)
+{
+	int	i;
+	char	*cp = (char *)uuid;
+
+	if (uuid == NULL)
+		return 0;
+	/* implied check of version number here... */
+	for (i = 0; i < sizeof *uuid; i++)
+		if (*cp++) return 0;	/* not nil */
+	return 1;	/* is nil */
+}
+
+int
+uuid_equal(uuid_t *uuid1, uuid_t *uuid2)
+{
+	return memcmp(uuid1, uuid2, sizeof(uuid_t)) ? 0 : 1;
+}
diff --git a/fs/xfs/support/uuid.h b/fs/xfs/support/uuid.h
new file mode 100644
index 00000000..4732d712
--- /dev/null
+++ b/fs/xfs/support/uuid.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright (c) 2000-2003,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_SUPPORT_UUID_H__
+#define __XFS_SUPPORT_UUID_H__
+
+typedef struct {
+	unsigned char	__u_bits[16];
+} uuid_t;
+
+extern int uuid_is_nil(uuid_t *uuid);
+extern int uuid_equal(uuid_t *uuid1, uuid_t *uuid2);
+extern void uuid_getnodeuniq(uuid_t *uuid, int fsid [2]);
+
+#endif	/* __XFS_SUPPORT_UUID_H__ */
diff --git a/fs/xfs/xfs.h b/fs/xfs/xfs.h
new file mode 100644
index 00000000..5ad8ad3a
--- /dev/null
+++ b/fs/xfs/xfs.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright (c) 2000-2003,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_H__
+#define __XFS_H__
+
+#ifdef CONFIG_XFS_DEBUG
+#define STATIC
+#define DEBUG 1
+#define XFS_BUF_LOCK_TRACKING 1
+/* #define QUOTADEBUG 1 */
+#endif
+
+#include <linux-2.6/xfs_linux.h>
+#endif	/* __XFS_H__ */
diff --git a/fs/xfs/xfs_acl.h b/fs/xfs/xfs_acl.h
new file mode 100644
index 00000000..11dd7207
--- /dev/null
+++ b/fs/xfs/xfs_acl.h
@@ -0,0 +1,62 @@
+/*
+ * Copyright (c) 2001-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_ACL_H__
+#define __XFS_ACL_H__
+
+struct inode;
+struct posix_acl;
+struct xfs_inode;
+
+#define XFS_ACL_MAX_ENTRIES 25
+#define XFS_ACL_NOT_PRESENT (-1)
+
+/* On-disk XFS access control list structure */
+struct xfs_acl {
+	__be32		acl_cnt;
+	struct xfs_acl_entry {
+		__be32	ae_tag;
+		__be32	ae_id;
+		__be16	ae_perm;
+	} acl_entry[XFS_ACL_MAX_ENTRIES];
+};
+
+/* On-disk XFS extended attribute names */
+#define SGI_ACL_FILE		(unsigned char *)"SGI_ACL_FILE"
+#define SGI_ACL_DEFAULT		(unsigned char *)"SGI_ACL_DEFAULT"
+#define SGI_ACL_FILE_SIZE	(sizeof(SGI_ACL_FILE)-1)
+#define SGI_ACL_DEFAULT_SIZE	(sizeof(SGI_ACL_DEFAULT)-1)
+
+#ifdef CONFIG_XFS_POSIX_ACL
+extern int xfs_check_acl(struct inode *inode, int mask, unsigned int flags);
+extern struct posix_acl *xfs_get_acl(struct inode *inode, int type);
+extern int xfs_inherit_acl(struct inode *inode, struct posix_acl *default_acl);
+extern int xfs_acl_chmod(struct inode *inode);
+extern int posix_acl_access_exists(struct inode *inode);
+extern int posix_acl_default_exists(struct inode *inode);
+
+extern const struct xattr_handler xfs_xattr_acl_access_handler;
+extern const struct xattr_handler xfs_xattr_acl_default_handler;
+#else
+# define xfs_check_acl					NULL
+# define xfs_get_acl(inode, type)			NULL
+# define xfs_inherit_acl(inode, default_acl)		0
+# define xfs_acl_chmod(inode)				0
+# define posix_acl_access_exists(inode)			0
+# define posix_acl_default_exists(inode)		0
+#endif /* CONFIG_XFS_POSIX_ACL */
+#endif	/* __XFS_ACL_H__ */
diff --git a/fs/xfs/xfs_ag.h b/fs/xfs/xfs_ag.h
new file mode 100644
index 00000000..6530769a
--- /dev/null
+++ b/fs/xfs/xfs_ag.h
@@ -0,0 +1,289 @@
+/*
+ * Copyright (c) 2000-2003,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_AG_H__
+#define	__XFS_AG_H__
+
+/*
+ * Allocation group header
+ * This is divided into three structures, placed in sequential 512-byte
+ * buffers after a copy of the superblock (also in a 512-byte buffer).
+ */
+
+struct xfs_buf;
+struct xfs_mount;
+struct xfs_trans;
+
+#define	XFS_AGF_MAGIC	0x58414746	/* 'XAGF' */
+#define	XFS_AGI_MAGIC	0x58414749	/* 'XAGI' */
+#define	XFS_AGF_VERSION	1
+#define	XFS_AGI_VERSION	1
+
+#define	XFS_AGF_GOOD_VERSION(v)	((v) == XFS_AGF_VERSION)
+#define	XFS_AGI_GOOD_VERSION(v)	((v) == XFS_AGI_VERSION)
+
+/*
+ * Btree number 0 is bno, 1 is cnt.  This value gives the size of the
+ * arrays below.
+ */
+#define	XFS_BTNUM_AGF	((int)XFS_BTNUM_CNTi + 1)
+
+/*
+ * The second word of agf_levels in the first a.g. overlaps the EFS
+ * superblock's magic number.  Since the magic numbers valid for EFS
+ * are > 64k, our value cannot be confused for an EFS superblock's.
+ */
+
+typedef struct xfs_agf {
+	/*
+	 * Common allocation group header information
+	 */
+	__be32		agf_magicnum;	/* magic number == XFS_AGF_MAGIC */
+	__be32		agf_versionnum;	/* header version == XFS_AGF_VERSION */
+	__be32		agf_seqno;	/* sequence # starting from 0 */
+	__be32		agf_length;	/* size in blocks of a.g. */
+	/*
+	 * Freespace information
+	 */
+	__be32		agf_roots[XFS_BTNUM_AGF];	/* root blocks */
+	__be32		agf_spare0;	/* spare field */
+	__be32		agf_levels[XFS_BTNUM_AGF];	/* btree levels */
+	__be32		agf_spare1;	/* spare field */
+	__be32		agf_flfirst;	/* first freelist block's index */
+	__be32		agf_fllast;	/* last freelist block's index */
+	__be32		agf_flcount;	/* count of blocks in freelist */
+	__be32		agf_freeblks;	/* total free blocks */
+	__be32		agf_longest;	/* longest free space */
+	__be32		agf_btreeblks;	/* # of blocks held in AGF btrees */
+} xfs_agf_t;
+
+#define	XFS_AGF_MAGICNUM	0x00000001
+#define	XFS_AGF_VERSIONNUM	0x00000002
+#define	XFS_AGF_SEQNO		0x00000004
+#define	XFS_AGF_LENGTH		0x00000008
+#define	XFS_AGF_ROOTS		0x00000010
+#define	XFS_AGF_LEVELS		0x00000020
+#define	XFS_AGF_FLFIRST		0x00000040
+#define	XFS_AGF_FLLAST		0x00000080
+#define	XFS_AGF_FLCOUNT		0x00000100
+#define	XFS_AGF_FREEBLKS	0x00000200
+#define	XFS_AGF_LONGEST		0x00000400
+#define	XFS_AGF_BTREEBLKS	0x00000800
+#define	XFS_AGF_NUM_BITS	12
+#define	XFS_AGF_ALL_BITS	((1 << XFS_AGF_NUM_BITS) - 1)
+
+#define XFS_AGF_FLAGS \
+	{ XFS_AGF_MAGICNUM,	"MAGICNUM" }, \
+	{ XFS_AGF_VERSIONNUM,	"VERSIONNUM" }, \
+	{ XFS_AGF_SEQNO,	"SEQNO" }, \
+	{ XFS_AGF_LENGTH,	"LENGTH" }, \
+	{ XFS_AGF_ROOTS,	"ROOTS" }, \
+	{ XFS_AGF_LEVELS,	"LEVELS" }, \
+	{ XFS_AGF_FLFIRST,	"FLFIRST" }, \
+	{ XFS_AGF_FLLAST,	"FLLAST" }, \
+	{ XFS_AGF_FLCOUNT,	"FLCOUNT" }, \
+	{ XFS_AGF_FREEBLKS,	"FREEBLKS" }, \
+	{ XFS_AGF_LONGEST,	"LONGEST" }, \
+	{ XFS_AGF_BTREEBLKS,	"BTREEBLKS" }
+
+/* disk block (xfs_daddr_t) in the AG */
+#define XFS_AGF_DADDR(mp)	((xfs_daddr_t)(1 << (mp)->m_sectbb_log))
+#define	XFS_AGF_BLOCK(mp)	XFS_HDR_BLOCK(mp, XFS_AGF_DADDR(mp))
+#define	XFS_BUF_TO_AGF(bp)	((xfs_agf_t *)XFS_BUF_PTR(bp))
+
+extern int xfs_read_agf(struct xfs_mount *mp, struct xfs_trans *tp,
+			xfs_agnumber_t agno, int flags, struct xfs_buf **bpp);
+
+/*
+ * Size of the unlinked inode hash table in the agi.
+ */
+#define	XFS_AGI_UNLINKED_BUCKETS	64
+
+typedef struct xfs_agi {
+	/*
+	 * Common allocation group header information
+	 */
+	__be32		agi_magicnum;	/* magic number == XFS_AGI_MAGIC */
+	__be32		agi_versionnum;	/* header version == XFS_AGI_VERSION */
+	__be32		agi_seqno;	/* sequence # starting from 0 */
+	__be32		agi_length;	/* size in blocks of a.g. */
+	/*
+	 * Inode information
+	 * Inodes are mapped by interpreting the inode number, so no
+	 * mapping data is needed here.
+	 */
+	__be32		agi_count;	/* count of allocated inodes */
+	__be32		agi_root;	/* root of inode btree */
+	__be32		agi_level;	/* levels in inode btree */
+	__be32		agi_freecount;	/* number of free inodes */
+	__be32		agi_newino;	/* new inode just allocated */
+	__be32		agi_dirino;	/* last directory inode chunk */
+	/*
+	 * Hash table of inodes which have been unlinked but are
+	 * still being referenced.
+	 */
+	__be32		agi_unlinked[XFS_AGI_UNLINKED_BUCKETS];
+} xfs_agi_t;
+
+#define	XFS_AGI_MAGICNUM	0x00000001
+#define	XFS_AGI_VERSIONNUM	0x00000002
+#define	XFS_AGI_SEQNO		0x00000004
+#define	XFS_AGI_LENGTH		0x00000008
+#define	XFS_AGI_COUNT		0x00000010
+#define	XFS_AGI_ROOT		0x00000020
+#define	XFS_AGI_LEVEL		0x00000040
+#define	XFS_AGI_FREECOUNT	0x00000080
+#define	XFS_AGI_NEWINO		0x00000100
+#define	XFS_AGI_DIRINO		0x00000200
+#define	XFS_AGI_UNLINKED	0x00000400
+#define	XFS_AGI_NUM_BITS	11
+#define	XFS_AGI_ALL_BITS	((1 << XFS_AGI_NUM_BITS) - 1)
+
+/* disk block (xfs_daddr_t) in the AG */
+#define XFS_AGI_DADDR(mp)	((xfs_daddr_t)(2 << (mp)->m_sectbb_log))
+#define	XFS_AGI_BLOCK(mp)	XFS_HDR_BLOCK(mp, XFS_AGI_DADDR(mp))
+#define	XFS_BUF_TO_AGI(bp)	((xfs_agi_t *)XFS_BUF_PTR(bp))
+
+extern int xfs_read_agi(struct xfs_mount *mp, struct xfs_trans *tp,
+				xfs_agnumber_t agno, struct xfs_buf **bpp);
+
+/*
+ * The third a.g. block contains the a.g. freelist, an array
+ * of block pointers to blocks owned by the allocation btree code.
+ */
+#define XFS_AGFL_DADDR(mp)	((xfs_daddr_t)(3 << (mp)->m_sectbb_log))
+#define	XFS_AGFL_BLOCK(mp)	XFS_HDR_BLOCK(mp, XFS_AGFL_DADDR(mp))
+#define XFS_AGFL_SIZE(mp)	((mp)->m_sb.sb_sectsize / sizeof(xfs_agblock_t))
+#define	XFS_BUF_TO_AGFL(bp)	((xfs_agfl_t *)XFS_BUF_PTR(bp))
+
+typedef struct xfs_agfl {
+	__be32		agfl_bno[1];	/* actually XFS_AGFL_SIZE(mp) */
+} xfs_agfl_t;
+
+/*
+ * Busy block/extent entry.  Indexed by a rbtree in perag to mark blocks that
+ * have been freed but whose transactions aren't committed to disk yet.
+ *
+ * Note that we use the transaction ID to record the transaction, not the
+ * transaction structure itself. See xfs_alloc_busy_insert() for details.
+ */
+struct xfs_busy_extent {
+	struct rb_node	rb_node;	/* ag by-bno indexed search tree */
+	struct list_head list;		/* transaction busy extent list */
+	xfs_agnumber_t	agno;
+	xfs_agblock_t	bno;
+	xfs_extlen_t	length;
+	unsigned int	flags;
+#define XFS_ALLOC_BUSY_DISCARDED	0x01	/* undergoing a discard op. */
+#define XFS_ALLOC_BUSY_SKIP_DISCARD	0x02	/* do not discard */
+};
+
+/*
+ * Per-ag incore structure, copies of information in agf and agi,
+ * to improve the performance of allocation group selection.
+ */
+#define XFS_PAGB_NUM_SLOTS	128
+
+typedef struct xfs_perag {
+	struct xfs_mount *pag_mount;	/* owner filesystem */
+	xfs_agnumber_t	pag_agno;	/* AG this structure belongs to */
+	atomic_t	pag_ref;	/* perag reference count */
+	char		pagf_init;	/* this agf's entry is initialized */
+	char		pagi_init;	/* this agi's entry is initialized */
+	char		pagf_metadata;	/* the agf is preferred to be metadata */
+	char		pagi_inodeok;	/* The agi is ok for inodes */
+	__uint8_t	pagf_levels[XFS_BTNUM_AGF];
+					/* # of levels in bno & cnt btree */
+	__uint32_t	pagf_flcount;	/* count of blocks in freelist */
+	xfs_extlen_t	pagf_freeblks;	/* total free blocks */
+	xfs_extlen_t	pagf_longest;	/* longest free space */
+	__uint32_t	pagf_btreeblks;	/* # of blocks held in AGF btrees */
+	xfs_agino_t	pagi_freecount;	/* number of free inodes */
+	xfs_agino_t	pagi_count;	/* number of allocated inodes */
+
+	/*
+	 * Inode allocation search lookup optimisation.
+	 * If the pagino matches, the search for new inodes
+	 * doesn't need to search the near ones again straight away
+	 */
+	xfs_agino_t	pagl_pagino;
+	xfs_agino_t	pagl_leftrec;
+	xfs_agino_t	pagl_rightrec;
+#ifdef __KERNEL__
+	spinlock_t	pagb_lock;	/* lock for pagb_tree */
+	struct rb_root	pagb_tree;	/* ordered tree of busy extents */
+
+	atomic_t        pagf_fstrms;    /* # of filestreams active in this AG */
+
+	spinlock_t	pag_ici_lock;	/* incore inode cache lock */
+	struct radix_tree_root pag_ici_root;	/* incore inode cache root */
+	int		pag_ici_reclaimable;	/* reclaimable inodes */
+	struct mutex	pag_ici_reclaim_lock;	/* serialisation point */
+	unsigned long	pag_ici_reclaim_cursor;	/* reclaim restart point */
+
+	/* buffer cache index */
+	spinlock_t	pag_buf_lock;	/* lock for pag_buf_tree */
+	struct rb_root	pag_buf_tree;	/* ordered tree of active buffers */
+
+	/* for rcu-safe freeing */
+	struct rcu_head	rcu_head;
+#endif
+	int		pagb_count;	/* pagb slots in use */
+} xfs_perag_t;
+
+/*
+ * tags for inode radix tree
+ */
+#define XFS_ICI_NO_TAG		(-1)	/* special flag for an untagged lookup
+					   in xfs_inode_ag_iterator */
+#define XFS_ICI_RECLAIM_TAG	0	/* inode is to be reclaimed */
+
+#define	XFS_AG_MAXLEVELS(mp)		((mp)->m_ag_maxlevels)
+#define	XFS_MIN_FREELIST_RAW(bl,cl,mp)	\
+	(MIN(bl + 1, XFS_AG_MAXLEVELS(mp)) + MIN(cl + 1, XFS_AG_MAXLEVELS(mp)))
+#define	XFS_MIN_FREELIST(a,mp)		\
+	(XFS_MIN_FREELIST_RAW(		\
+		be32_to_cpu((a)->agf_levels[XFS_BTNUM_BNOi]), \
+		be32_to_cpu((a)->agf_levels[XFS_BTNUM_CNTi]), mp))
+#define	XFS_MIN_FREELIST_PAG(pag,mp)	\
+	(XFS_MIN_FREELIST_RAW(		\
+		(unsigned int)(pag)->pagf_levels[XFS_BTNUM_BNOi], \
+		(unsigned int)(pag)->pagf_levels[XFS_BTNUM_CNTi], mp))
+
+#define XFS_AGB_TO_FSB(mp,agno,agbno)	\
+	(((xfs_fsblock_t)(agno) << (mp)->m_sb.sb_agblklog) | (agbno))
+#define	XFS_FSB_TO_AGNO(mp,fsbno)	\
+	((xfs_agnumber_t)((fsbno) >> (mp)->m_sb.sb_agblklog))
+#define	XFS_FSB_TO_AGBNO(mp,fsbno)	\
+	((xfs_agblock_t)((fsbno) & xfs_mask32lo((mp)->m_sb.sb_agblklog)))
+#define	XFS_AGB_TO_DADDR(mp,agno,agbno)	\
+	((xfs_daddr_t)XFS_FSB_TO_BB(mp, \
+		(xfs_fsblock_t)(agno) * (mp)->m_sb.sb_agblocks + (agbno)))
+#define	XFS_AG_DADDR(mp,agno,d)		(XFS_AGB_TO_DADDR(mp, agno, 0) + (d))
+
+/*
+ * For checking for bad ranges of xfs_daddr_t's, covering multiple
+ * allocation groups or a single xfs_daddr_t that's a superblock copy.
+ */
+#define	XFS_AG_CHECK_DADDR(mp,d,len)	\
+	((len) == 1 ? \
+	    ASSERT((d) == XFS_SB_DADDR || \
+		   xfs_daddr_to_agbno(mp, d) != XFS_SB_DADDR) : \
+	    ASSERT(xfs_daddr_to_agno(mp, d) == \
+		   xfs_daddr_to_agno(mp, (d) + (len) - 1)))
+
+#endif	/* __XFS_AG_H__ */
diff --git a/fs/xfs/xfs_alloc.c b/fs/xfs/xfs_alloc.c
new file mode 100644
index 00000000..95862bbf
--- /dev/null
+++ b/fs/xfs/xfs_alloc.c
@@ -0,0 +1,3047 @@
+/*
+ * Copyright (c) 2000-2002,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_mount.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_alloc_btree.h"
+#include "xfs_ialloc_btree.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_btree.h"
+#include "xfs_alloc.h"
+#include "xfs_error.h"
+#include "xfs_trace.h"
+
+
+#define XFS_ABSDIFF(a,b)	(((a) <= (b)) ? ((b) - (a)) : ((a) - (b)))
+
+#define	XFSA_FIXUP_BNO_OK	1
+#define	XFSA_FIXUP_CNT_OK	2
+
+STATIC int xfs_alloc_ag_vextent_exact(xfs_alloc_arg_t *);
+STATIC int xfs_alloc_ag_vextent_near(xfs_alloc_arg_t *);
+STATIC int xfs_alloc_ag_vextent_size(xfs_alloc_arg_t *);
+STATIC int xfs_alloc_ag_vextent_small(xfs_alloc_arg_t *,
+		xfs_btree_cur_t *, xfs_agblock_t *, xfs_extlen_t *, int *);
+STATIC void xfs_alloc_busy_trim(struct xfs_alloc_arg *,
+		xfs_agblock_t, xfs_extlen_t, xfs_agblock_t *, xfs_extlen_t *);
+
+/*
+ * Lookup the record equal to [bno, len] in the btree given by cur.
+ */
+STATIC int				/* error */
+xfs_alloc_lookup_eq(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_agblock_t		bno,	/* starting block of extent */
+	xfs_extlen_t		len,	/* length of extent */
+	int			*stat)	/* success/failure */
+{
+	cur->bc_rec.a.ar_startblock = bno;
+	cur->bc_rec.a.ar_blockcount = len;
+	return xfs_btree_lookup(cur, XFS_LOOKUP_EQ, stat);
+}
+
+/*
+ * Lookup the first record greater than or equal to [bno, len]
+ * in the btree given by cur.
+ */
+STATIC int				/* error */
+xfs_alloc_lookup_ge(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_agblock_t		bno,	/* starting block of extent */
+	xfs_extlen_t		len,	/* length of extent */
+	int			*stat)	/* success/failure */
+{
+	cur->bc_rec.a.ar_startblock = bno;
+	cur->bc_rec.a.ar_blockcount = len;
+	return xfs_btree_lookup(cur, XFS_LOOKUP_GE, stat);
+}
+
+/*
+ * Lookup the first record less than or equal to [bno, len]
+ * in the btree given by cur.
+ */
+int					/* error */
+xfs_alloc_lookup_le(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_agblock_t		bno,	/* starting block of extent */
+	xfs_extlen_t		len,	/* length of extent */
+	int			*stat)	/* success/failure */
+{
+	cur->bc_rec.a.ar_startblock = bno;
+	cur->bc_rec.a.ar_blockcount = len;
+	return xfs_btree_lookup(cur, XFS_LOOKUP_LE, stat);
+}
+
+/*
+ * Update the record referred to by cur to the value given
+ * by [bno, len].
+ * This either works (return 0) or gets an EFSCORRUPTED error.
+ */
+STATIC int				/* error */
+xfs_alloc_update(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_agblock_t		bno,	/* starting block of extent */
+	xfs_extlen_t		len)	/* length of extent */
+{
+	union xfs_btree_rec	rec;
+
+	rec.alloc.ar_startblock = cpu_to_be32(bno);
+	rec.alloc.ar_blockcount = cpu_to_be32(len);
+	return xfs_btree_update(cur, &rec);
+}
+
+/*
+ * Get the data from the pointed-to record.
+ */
+int					/* error */
+xfs_alloc_get_rec(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_agblock_t		*bno,	/* output: starting block of extent */
+	xfs_extlen_t		*len,	/* output: length of extent */
+	int			*stat)	/* output: success/failure */
+{
+	union xfs_btree_rec	*rec;
+	int			error;
+
+	error = xfs_btree_get_rec(cur, &rec, stat);
+	if (!error && *stat == 1) {
+		*bno = be32_to_cpu(rec->alloc.ar_startblock);
+		*len = be32_to_cpu(rec->alloc.ar_blockcount);
+	}
+	return error;
+}
+
+/*
+ * Compute aligned version of the found extent.
+ * Takes alignment and min length into account.
+ */
+STATIC void
+xfs_alloc_compute_aligned(
+	xfs_alloc_arg_t	*args,		/* allocation argument structure */
+	xfs_agblock_t	foundbno,	/* starting block in found extent */
+	xfs_extlen_t	foundlen,	/* length in found extent */
+	xfs_agblock_t	*resbno,	/* result block number */
+	xfs_extlen_t	*reslen)	/* result length */
+{
+	xfs_agblock_t	bno;
+	xfs_extlen_t	len;
+
+	/* Trim busy sections out of found extent */
+	xfs_alloc_busy_trim(args, foundbno, foundlen, &bno, &len);
+
+	if (args->alignment > 1 && len >= args->minlen) {
+		xfs_agblock_t	aligned_bno = roundup(bno, args->alignment);
+		xfs_extlen_t	diff = aligned_bno - bno;
+
+		*resbno = aligned_bno;
+		*reslen = diff >= len ? 0 : len - diff;
+	} else {
+		*resbno = bno;
+		*reslen = len;
+	}
+}
+
+/*
+ * Compute best start block and diff for "near" allocations.
+ * freelen >= wantlen already checked by caller.
+ */
+STATIC xfs_extlen_t			/* difference value (absolute) */
+xfs_alloc_compute_diff(
+	xfs_agblock_t	wantbno,	/* target starting block */
+	xfs_extlen_t	wantlen,	/* target length */
+	xfs_extlen_t	alignment,	/* target alignment */
+	xfs_agblock_t	freebno,	/* freespace's starting block */
+	xfs_extlen_t	freelen,	/* freespace's length */
+	xfs_agblock_t	*newbnop)	/* result: best start block from free */
+{
+	xfs_agblock_t	freeend;	/* end of freespace extent */
+	xfs_agblock_t	newbno1;	/* return block number */
+	xfs_agblock_t	newbno2;	/* other new block number */
+	xfs_extlen_t	newlen1=0;	/* length with newbno1 */
+	xfs_extlen_t	newlen2=0;	/* length with newbno2 */
+	xfs_agblock_t	wantend;	/* end of target extent */
+
+	ASSERT(freelen >= wantlen);
+	freeend = freebno + freelen;
+	wantend = wantbno + wantlen;
+	if (freebno >= wantbno) {
+		if ((newbno1 = roundup(freebno, alignment)) >= freeend)
+			newbno1 = NULLAGBLOCK;
+	} else if (freeend >= wantend && alignment > 1) {
+		newbno1 = roundup(wantbno, alignment);
+		newbno2 = newbno1 - alignment;
+		if (newbno1 >= freeend)
+			newbno1 = NULLAGBLOCK;
+		else
+			newlen1 = XFS_EXTLEN_MIN(wantlen, freeend - newbno1);
+		if (newbno2 < freebno)
+			newbno2 = NULLAGBLOCK;
+		else
+			newlen2 = XFS_EXTLEN_MIN(wantlen, freeend - newbno2);
+		if (newbno1 != NULLAGBLOCK && newbno2 != NULLAGBLOCK) {
+			if (newlen1 < newlen2 ||
+			    (newlen1 == newlen2 &&
+			     XFS_ABSDIFF(newbno1, wantbno) >
+			     XFS_ABSDIFF(newbno2, wantbno)))
+				newbno1 = newbno2;
+		} else if (newbno2 != NULLAGBLOCK)
+			newbno1 = newbno2;
+	} else if (freeend >= wantend) {
+		newbno1 = wantbno;
+	} else if (alignment > 1) {
+		newbno1 = roundup(freeend - wantlen, alignment);
+		if (newbno1 > freeend - wantlen &&
+		    newbno1 - alignment >= freebno)
+			newbno1 -= alignment;
+		else if (newbno1 >= freeend)
+			newbno1 = NULLAGBLOCK;
+	} else
+		newbno1 = freeend - wantlen;
+	*newbnop = newbno1;
+	return newbno1 == NULLAGBLOCK ? 0 : XFS_ABSDIFF(newbno1, wantbno);
+}
+
+/*
+ * Fix up the length, based on mod and prod.
+ * len should be k * prod + mod for some k.
+ * If len is too small it is returned unchanged.
+ * If len hits maxlen it is left alone.
+ */
+STATIC void
+xfs_alloc_fix_len(
+	xfs_alloc_arg_t	*args)		/* allocation argument structure */
+{
+	xfs_extlen_t	k;
+	xfs_extlen_t	rlen;
+
+	ASSERT(args->mod < args->prod);
+	rlen = args->len;
+	ASSERT(rlen >= args->minlen);
+	ASSERT(rlen <= args->maxlen);
+	if (args->prod <= 1 || rlen < args->mod || rlen == args->maxlen ||
+	    (args->mod == 0 && rlen < args->prod))
+		return;
+	k = rlen % args->prod;
+	if (k == args->mod)
+		return;
+	if (k > args->mod) {
+		if ((int)(rlen = rlen - k - args->mod) < (int)args->minlen)
+			return;
+	} else {
+		if ((int)(rlen = rlen - args->prod - (args->mod - k)) <
+		    (int)args->minlen)
+			return;
+	}
+	ASSERT(rlen >= args->minlen);
+	ASSERT(rlen <= args->maxlen);
+	args->len = rlen;
+}
+
+/*
+ * Fix up length if there is too little space left in the a.g.
+ * Return 1 if ok, 0 if too little, should give up.
+ */
+STATIC int
+xfs_alloc_fix_minleft(
+	xfs_alloc_arg_t	*args)		/* allocation argument structure */
+{
+	xfs_agf_t	*agf;		/* a.g. freelist header */
+	int		diff;		/* free space difference */
+
+	if (args->minleft == 0)
+		return 1;
+	agf = XFS_BUF_TO_AGF(args->agbp);
+	diff = be32_to_cpu(agf->agf_freeblks)
+		- args->len - args->minleft;
+	if (diff >= 0)
+		return 1;
+	args->len += diff;		/* shrink the allocated space */
+	if (args->len >= args->minlen)
+		return 1;
+	args->agbno = NULLAGBLOCK;
+	return 0;
+}
+
+/*
+ * Update the two btrees, logically removing from freespace the extent
+ * starting at rbno, rlen blocks.  The extent is contained within the
+ * actual (current) free extent fbno for flen blocks.
+ * Flags are passed in indicating whether the cursors are set to the
+ * relevant records.
+ */
+STATIC int				/* error code */
+xfs_alloc_fixup_trees(
+	xfs_btree_cur_t	*cnt_cur,	/* cursor for by-size btree */
+	xfs_btree_cur_t	*bno_cur,	/* cursor for by-block btree */
+	xfs_agblock_t	fbno,		/* starting block of free extent */
+	xfs_extlen_t	flen,		/* length of free extent */
+	xfs_agblock_t	rbno,		/* starting block of returned extent */
+	xfs_extlen_t	rlen,		/* length of returned extent */
+	int		flags)		/* flags, XFSA_FIXUP_... */
+{
+	int		error;		/* error code */
+	int		i;		/* operation results */
+	xfs_agblock_t	nfbno1;		/* first new free startblock */
+	xfs_agblock_t	nfbno2;		/* second new free startblock */
+	xfs_extlen_t	nflen1=0;	/* first new free length */
+	xfs_extlen_t	nflen2=0;	/* second new free length */
+
+	/*
+	 * Look up the record in the by-size tree if necessary.
+	 */
+	if (flags & XFSA_FIXUP_CNT_OK) {
+#ifdef DEBUG
+		if ((error = xfs_alloc_get_rec(cnt_cur, &nfbno1, &nflen1, &i)))
+			return error;
+		XFS_WANT_CORRUPTED_RETURN(
+			i == 1 && nfbno1 == fbno && nflen1 == flen);
+#endif
+	} else {
+		if ((error = xfs_alloc_lookup_eq(cnt_cur, fbno, flen, &i)))
+			return error;
+		XFS_WANT_CORRUPTED_RETURN(i == 1);
+	}
+	/*
+	 * Look up the record in the by-block tree if necessary.
+	 */
+	if (flags & XFSA_FIXUP_BNO_OK) {
+#ifdef DEBUG
+		if ((error = xfs_alloc_get_rec(bno_cur, &nfbno1, &nflen1, &i)))
+			return error;
+		XFS_WANT_CORRUPTED_RETURN(
+			i == 1 && nfbno1 == fbno && nflen1 == flen);
+#endif
+	} else {
+		if ((error = xfs_alloc_lookup_eq(bno_cur, fbno, flen, &i)))
+			return error;
+		XFS_WANT_CORRUPTED_RETURN(i == 1);
+	}
+
+#ifdef DEBUG
+	if (bno_cur->bc_nlevels == 1 && cnt_cur->bc_nlevels == 1) {
+		struct xfs_btree_block	*bnoblock;
+		struct xfs_btree_block	*cntblock;
+
+		bnoblock = XFS_BUF_TO_BLOCK(bno_cur->bc_bufs[0]);
+		cntblock = XFS_BUF_TO_BLOCK(cnt_cur->bc_bufs[0]);
+
+		XFS_WANT_CORRUPTED_RETURN(
+			bnoblock->bb_numrecs == cntblock->bb_numrecs);
+	}
+#endif
+
+	/*
+	 * Deal with all four cases: the allocated record is contained
+	 * within the freespace record, so we can have new freespace
+	 * at either (or both) end, or no freespace remaining.
+	 */
+	if (rbno == fbno && rlen == flen)
+		nfbno1 = nfbno2 = NULLAGBLOCK;
+	else if (rbno == fbno) {
+		nfbno1 = rbno + rlen;
+		nflen1 = flen - rlen;
+		nfbno2 = NULLAGBLOCK;
+	} else if (rbno + rlen == fbno + flen) {
+		nfbno1 = fbno;
+		nflen1 = flen - rlen;
+		nfbno2 = NULLAGBLOCK;
+	} else {
+		nfbno1 = fbno;
+		nflen1 = rbno - fbno;
+		nfbno2 = rbno + rlen;
+		nflen2 = (fbno + flen) - nfbno2;
+	}
+	/*
+	 * Delete the entry from the by-size btree.
+	 */
+	if ((error = xfs_btree_delete(cnt_cur, &i)))
+		return error;
+	XFS_WANT_CORRUPTED_RETURN(i == 1);
+	/*
+	 * Add new by-size btree entry(s).
+	 */
+	if (nfbno1 != NULLAGBLOCK) {
+		if ((error = xfs_alloc_lookup_eq(cnt_cur, nfbno1, nflen1, &i)))
+			return error;
+		XFS_WANT_CORRUPTED_RETURN(i == 0);
+		if ((error = xfs_btree_insert(cnt_cur, &i)))
+			return error;
+		XFS_WANT_CORRUPTED_RETURN(i == 1);
+	}
+	if (nfbno2 != NULLAGBLOCK) {
+		if ((error = xfs_alloc_lookup_eq(cnt_cur, nfbno2, nflen2, &i)))
+			return error;
+		XFS_WANT_CORRUPTED_RETURN(i == 0);
+		if ((error = xfs_btree_insert(cnt_cur, &i)))
+			return error;
+		XFS_WANT_CORRUPTED_RETURN(i == 1);
+	}
+	/*
+	 * Fix up the by-block btree entry(s).
+	 */
+	if (nfbno1 == NULLAGBLOCK) {
+		/*
+		 * No remaining freespace, just delete the by-block tree entry.
+		 */
+		if ((error = xfs_btree_delete(bno_cur, &i)))
+			return error;
+		XFS_WANT_CORRUPTED_RETURN(i == 1);
+	} else {
+		/*
+		 * Update the by-block entry to start later|be shorter.
+		 */
+		if ((error = xfs_alloc_update(bno_cur, nfbno1, nflen1)))
+			return error;
+	}
+	if (nfbno2 != NULLAGBLOCK) {
+		/*
+		 * 2 resulting free entries, need to add one.
+		 */
+		if ((error = xfs_alloc_lookup_eq(bno_cur, nfbno2, nflen2, &i)))
+			return error;
+		XFS_WANT_CORRUPTED_RETURN(i == 0);
+		if ((error = xfs_btree_insert(bno_cur, &i)))
+			return error;
+		XFS_WANT_CORRUPTED_RETURN(i == 1);
+	}
+	return 0;
+}
+
+/*
+ * Read in the allocation group free block array.
+ */
+STATIC int				/* error */
+xfs_alloc_read_agfl(
+	xfs_mount_t	*mp,		/* mount point structure */
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_agnumber_t	agno,		/* allocation group number */
+	xfs_buf_t	**bpp)		/* buffer for the ag free block array */
+{
+	xfs_buf_t	*bp;		/* return value */
+	int		error;
+
+	ASSERT(agno != NULLAGNUMBER);
+	error = xfs_trans_read_buf(
+			mp, tp, mp->m_ddev_targp,
+			XFS_AG_DADDR(mp, agno, XFS_AGFL_DADDR(mp)),
+			XFS_FSS_TO_BB(mp, 1), 0, &bp);
+	if (error)
+		return error;
+	ASSERT(bp);
+	ASSERT(!XFS_BUF_GETERROR(bp));
+	XFS_BUF_SET_VTYPE_REF(bp, B_FS_AGFL, XFS_AGFL_REF);
+	*bpp = bp;
+	return 0;
+}
+
+STATIC int
+xfs_alloc_update_counters(
+	struct xfs_trans	*tp,
+	struct xfs_perag	*pag,
+	struct xfs_buf		*agbp,
+	long			len)
+{
+	struct xfs_agf		*agf = XFS_BUF_TO_AGF(agbp);
+
+	pag->pagf_freeblks += len;
+	be32_add_cpu(&agf->agf_freeblks, len);
+
+	xfs_trans_agblocks_delta(tp, len);
+	if (unlikely(be32_to_cpu(agf->agf_freeblks) >
+		     be32_to_cpu(agf->agf_length)))
+		return EFSCORRUPTED;
+
+	xfs_alloc_log_agf(tp, agbp, XFS_AGF_FREEBLKS);
+	return 0;
+}
+
+/*
+ * Allocation group level functions.
+ */
+
+/*
+ * Allocate a variable extent in the allocation group agno.
+ * Type and bno are used to determine where in the allocation group the
+ * extent will start.
+ * Extent's length (returned in *len) will be between minlen and maxlen,
+ * and of the form k * prod + mod unless there's nothing that large.
+ * Return the starting a.g. block, or NULLAGBLOCK if we can't do it.
+ */
+STATIC int			/* error */
+xfs_alloc_ag_vextent(
+	xfs_alloc_arg_t	*args)	/* argument structure for allocation */
+{
+	int		error=0;
+
+	ASSERT(args->minlen > 0);
+	ASSERT(args->maxlen > 0);
+	ASSERT(args->minlen <= args->maxlen);
+	ASSERT(args->mod < args->prod);
+	ASSERT(args->alignment > 0);
+	/*
+	 * Branch to correct routine based on the type.
+	 */
+	args->wasfromfl = 0;
+	switch (args->type) {
+	case XFS_ALLOCTYPE_THIS_AG:
+		error = xfs_alloc_ag_vextent_size(args);
+		break;
+	case XFS_ALLOCTYPE_NEAR_BNO:
+		error = xfs_alloc_ag_vextent_near(args);
+		break;
+	case XFS_ALLOCTYPE_THIS_BNO:
+		error = xfs_alloc_ag_vextent_exact(args);
+		break;
+	default:
+		ASSERT(0);
+		/* NOTREACHED */
+	}
+
+	if (error || args->agbno == NULLAGBLOCK)
+		return error;
+
+	ASSERT(args->len >= args->minlen);
+	ASSERT(args->len <= args->maxlen);
+	ASSERT(!args->wasfromfl || !args->isfl);
+	ASSERT(args->agbno % args->alignment == 0);
+
+	if (!args->wasfromfl) {
+		error = xfs_alloc_update_counters(args->tp, args->pag,
+						  args->agbp,
+						  -((long)(args->len)));
+		if (error)
+			return error;
+
+		ASSERT(!xfs_alloc_busy_search(args->mp, args->agno,
+					      args->agbno, args->len));
+	}
+
+	if (!args->isfl) {
+		xfs_trans_mod_sb(args->tp, args->wasdel ?
+				 XFS_TRANS_SB_RES_FDBLOCKS :
+				 XFS_TRANS_SB_FDBLOCKS,
+				 -((long)(args->len)));
+	}
+
+	XFS_STATS_INC(xs_allocx);
+	XFS_STATS_ADD(xs_allocb, args->len);
+	return error;
+}
+
+/*
+ * Allocate a variable extent at exactly agno/bno.
+ * Extent's length (returned in *len) will be between minlen and maxlen,
+ * and of the form k * prod + mod unless there's nothing that large.
+ * Return the starting a.g. block (bno), or NULLAGBLOCK if we can't do it.
+ */
+STATIC int			/* error */
+xfs_alloc_ag_vextent_exact(
+	xfs_alloc_arg_t	*args)	/* allocation argument structure */
+{
+	xfs_btree_cur_t	*bno_cur;/* by block-number btree cursor */
+	xfs_btree_cur_t	*cnt_cur;/* by count btree cursor */
+	int		error;
+	xfs_agblock_t	fbno;	/* start block of found extent */
+	xfs_extlen_t	flen;	/* length of found extent */
+	xfs_agblock_t	tbno;	/* start block of trimmed extent */
+	xfs_extlen_t	tlen;	/* length of trimmed extent */
+	xfs_agblock_t	tend;	/* end block of trimmed extent */
+	xfs_agblock_t	end;	/* end of allocated extent */
+	int		i;	/* success/failure of operation */
+	xfs_extlen_t	rlen;	/* length of returned extent */
+
+	ASSERT(args->alignment == 1);
+
+	/*
+	 * Allocate/initialize a cursor for the by-number freespace btree.
+	 */
+	bno_cur = xfs_allocbt_init_cursor(args->mp, args->tp, args->agbp,
+					  args->agno, XFS_BTNUM_BNO);
+
+	/*
+	 * Lookup bno and minlen in the btree (minlen is irrelevant, really).
+	 * Look for the closest free block <= bno, it must contain bno
+	 * if any free block does.
+	 */
+	error = xfs_alloc_lookup_le(bno_cur, args->agbno, args->minlen, &i);
+	if (error)
+		goto error0;
+	if (!i)
+		goto not_found;
+
+	/*
+	 * Grab the freespace record.
+	 */
+	error = xfs_alloc_get_rec(bno_cur, &fbno, &flen, &i);
+	if (error)
+		goto error0;
+	XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+	ASSERT(fbno <= args->agbno);
+
+	/*
+	 * Check for overlapping busy extents.
+	 */
+	xfs_alloc_busy_trim(args, fbno, flen, &tbno, &tlen);
+
+	/*
+	 * Give up if the start of the extent is busy, or the freespace isn't
+	 * long enough for the minimum request.
+	 */
+	if (tbno > args->agbno)
+		goto not_found;
+	if (tlen < args->minlen)
+		goto not_found;
+	tend = tbno + tlen;
+	if (tend < args->agbno + args->minlen)
+		goto not_found;
+
+	/*
+	 * End of extent will be smaller of the freespace end and the
+	 * maximal requested end.
+	 *
+	 * Fix the length according to mod and prod if given.
+	 */
+	end = XFS_AGBLOCK_MIN(tend, args->agbno + args->maxlen);
+	args->len = end - args->agbno;
+	xfs_alloc_fix_len(args);
+	if (!xfs_alloc_fix_minleft(args))
+		goto not_found;
+
+	rlen = args->len;
+	ASSERT(args->agbno + rlen <= tend);
+	end = args->agbno + rlen;
+
+	/*
+	 * We are allocating agbno for rlen [agbno .. end]
+	 * Allocate/initialize a cursor for the by-size btree.
+	 */
+	cnt_cur = xfs_allocbt_init_cursor(args->mp, args->tp, args->agbp,
+		args->agno, XFS_BTNUM_CNT);
+	ASSERT(args->agbno + args->len <=
+		be32_to_cpu(XFS_BUF_TO_AGF(args->agbp)->agf_length));
+	error = xfs_alloc_fixup_trees(cnt_cur, bno_cur, fbno, flen, args->agbno,
+				      args->len, XFSA_FIXUP_BNO_OK);
+	if (error) {
+		xfs_btree_del_cursor(cnt_cur, XFS_BTREE_ERROR);
+		goto error0;
+	}
+
+	xfs_btree_del_cursor(bno_cur, XFS_BTREE_NOERROR);
+	xfs_btree_del_cursor(cnt_cur, XFS_BTREE_NOERROR);
+
+	args->wasfromfl = 0;
+	trace_xfs_alloc_exact_done(args);
+	return 0;
+
+not_found:
+	/* Didn't find it, return null. */
+	xfs_btree_del_cursor(bno_cur, XFS_BTREE_NOERROR);
+	args->agbno = NULLAGBLOCK;
+	trace_xfs_alloc_exact_notfound(args);
+	return 0;
+
+error0:
+	xfs_btree_del_cursor(bno_cur, XFS_BTREE_ERROR);
+	trace_xfs_alloc_exact_error(args);
+	return error;
+}
+
+/*
+ * Search the btree in a given direction via the search cursor and compare
+ * the records found against the good extent we've already found.
+ */
+STATIC int
+xfs_alloc_find_best_extent(
+	struct xfs_alloc_arg	*args,	/* allocation argument structure */
+	struct xfs_btree_cur	**gcur,	/* good cursor */
+	struct xfs_btree_cur	**scur,	/* searching cursor */
+	xfs_agblock_t		gdiff,	/* difference for search comparison */
+	xfs_agblock_t		*sbno,	/* extent found by search */
+	xfs_extlen_t		*slen,	/* extent length */
+	xfs_agblock_t		*sbnoa,	/* aligned extent found by search */
+	xfs_extlen_t		*slena,	/* aligned extent length */
+	int			dir)	/* 0 = search right, 1 = search left */
+{
+	xfs_agblock_t		new;
+	xfs_agblock_t		sdiff;
+	int			error;
+	int			i;
+
+	/* The good extent is perfect, no need to  search. */
+	if (!gdiff)
+		goto out_use_good;
+
+	/*
+	 * Look until we find a better one, run out of space or run off the end.
+	 */
+	do {
+		error = xfs_alloc_get_rec(*scur, sbno, slen, &i);
+		if (error)
+			goto error0;
+		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+		xfs_alloc_compute_aligned(args, *sbno, *slen, sbnoa, slena);
+
+		/*
+		 * The good extent is closer than this one.
+		 */
+		if (!dir) {
+			if (*sbnoa >= args->agbno + gdiff)
+				goto out_use_good;
+		} else {
+			if (*sbnoa <= args->agbno - gdiff)
+				goto out_use_good;
+		}
+
+		/*
+		 * Same distance, compare length and pick the best.
+		 */
+		if (*slena >= args->minlen) {
+			args->len = XFS_EXTLEN_MIN(*slena, args->maxlen);
+			xfs_alloc_fix_len(args);
+
+			sdiff = xfs_alloc_compute_diff(args->agbno, args->len,
+						       args->alignment, *sbnoa,
+						       *slena, &new);
+
+			/*
+			 * Choose closer size and invalidate other cursor.
+			 */
+			if (sdiff < gdiff)
+				goto out_use_search;
+			goto out_use_good;
+		}
+
+		if (!dir)
+			error = xfs_btree_increment(*scur, 0, &i);
+		else
+			error = xfs_btree_decrement(*scur, 0, &i);
+		if (error)
+			goto error0;
+	} while (i);
+
+out_use_good:
+	xfs_btree_del_cursor(*scur, XFS_BTREE_NOERROR);
+	*scur = NULL;
+	return 0;
+
+out_use_search:
+	xfs_btree_del_cursor(*gcur, XFS_BTREE_NOERROR);
+	*gcur = NULL;
+	return 0;
+
+error0:
+	/* caller invalidates cursors */
+	return error;
+}
+
+/*
+ * Allocate a variable extent near bno in the allocation group agno.
+ * Extent's length (returned in len) will be between minlen and maxlen,
+ * and of the form k * prod + mod unless there's nothing that large.
+ * Return the starting a.g. block, or NULLAGBLOCK if we can't do it.
+ */
+STATIC int				/* error */
+xfs_alloc_ag_vextent_near(
+	xfs_alloc_arg_t	*args)		/* allocation argument structure */
+{
+	xfs_btree_cur_t	*bno_cur_gt;	/* cursor for bno btree, right side */
+	xfs_btree_cur_t	*bno_cur_lt;	/* cursor for bno btree, left side */
+	xfs_btree_cur_t	*cnt_cur;	/* cursor for count btree */
+	xfs_agblock_t	gtbno;		/* start bno of right side entry */
+	xfs_agblock_t	gtbnoa;		/* aligned ... */
+	xfs_extlen_t	gtdiff;		/* difference to right side entry */
+	xfs_extlen_t	gtlen;		/* length of right side entry */
+	xfs_extlen_t	gtlena;		/* aligned ... */
+	xfs_agblock_t	gtnew;		/* useful start bno of right side */
+	int		error;		/* error code */
+	int		i;		/* result code, temporary */
+	int		j;		/* result code, temporary */
+	xfs_agblock_t	ltbno;		/* start bno of left side entry */
+	xfs_agblock_t	ltbnoa;		/* aligned ... */
+	xfs_extlen_t	ltdiff;		/* difference to left side entry */
+	xfs_extlen_t	ltlen;		/* length of left side entry */
+	xfs_extlen_t	ltlena;		/* aligned ... */
+	xfs_agblock_t	ltnew;		/* useful start bno of left side */
+	xfs_extlen_t	rlen;		/* length of returned extent */
+	int		forced = 0;
+#if defined(DEBUG) && defined(__KERNEL__)
+	/*
+	 * Randomly don't execute the first algorithm.
+	 */
+	int		dofirst;	/* set to do first algorithm */
+
+	dofirst = random32() & 1;
+#endif
+
+restart:
+	bno_cur_lt = NULL;
+	bno_cur_gt = NULL;
+	ltlen = 0;
+	gtlena = 0;
+	ltlena = 0;
+
+	/*
+	 * Get a cursor for the by-size btree.
+	 */
+	cnt_cur = xfs_allocbt_init_cursor(args->mp, args->tp, args->agbp,
+		args->agno, XFS_BTNUM_CNT);
+
+	/*
+	 * See if there are any free extents as big as maxlen.
+	 */
+	if ((error = xfs_alloc_lookup_ge(cnt_cur, 0, args->maxlen, &i)))
+		goto error0;
+	/*
+	 * If none, then pick up the last entry in the tree unless the
+	 * tree is empty.
+	 */
+	if (!i) {
+		if ((error = xfs_alloc_ag_vextent_small(args, cnt_cur, &ltbno,
+				&ltlen, &i)))
+			goto error0;
+		if (i == 0 || ltlen == 0) {
+			xfs_btree_del_cursor(cnt_cur, XFS_BTREE_NOERROR);
+			trace_xfs_alloc_near_noentry(args);
+			return 0;
+		}
+		ASSERT(i == 1);
+	}
+	args->wasfromfl = 0;
+
+	/*
+	 * First algorithm.
+	 * If the requested extent is large wrt the freespaces available
+	 * in this a.g., then the cursor will be pointing to a btree entry
+	 * near the right edge of the tree.  If it's in the last btree leaf
+	 * block, then we just examine all the entries in that block
+	 * that are big enough, and pick the best one.
+	 * This is written as a while loop so we can break out of it,
+	 * but we never loop back to the top.
+	 */
+	while (xfs_btree_islastblock(cnt_cur, 0)) {
+		xfs_extlen_t	bdiff;
+		int		besti=0;
+		xfs_extlen_t	blen=0;
+		xfs_agblock_t	bnew=0;
+
+#if defined(DEBUG) && defined(__KERNEL__)
+		if (!dofirst)
+			break;
+#endif
+		/*
+		 * Start from the entry that lookup found, sequence through
+		 * all larger free blocks.  If we're actually pointing at a
+		 * record smaller than maxlen, go to the start of this block,
+		 * and skip all those smaller than minlen.
+		 */
+		if (ltlen || args->alignment > 1) {
+			cnt_cur->bc_ptrs[0] = 1;
+			do {
+				if ((error = xfs_alloc_get_rec(cnt_cur, &ltbno,
+						&ltlen, &i)))
+					goto error0;
+				XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+				if (ltlen >= args->minlen)
+					break;
+				if ((error = xfs_btree_increment(cnt_cur, 0, &i)))
+					goto error0;
+			} while (i);
+			ASSERT(ltlen >= args->minlen);
+			if (!i)
+				break;
+		}
+		i = cnt_cur->bc_ptrs[0];
+		for (j = 1, blen = 0, bdiff = 0;
+		     !error && j && (blen < args->maxlen || bdiff > 0);
+		     error = xfs_btree_increment(cnt_cur, 0, &j)) {
+			/*
+			 * For each entry, decide if it's better than
+			 * the previous best entry.
+			 */
+			if ((error = xfs_alloc_get_rec(cnt_cur, &ltbno, &ltlen, &i)))
+				goto error0;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+			xfs_alloc_compute_aligned(args, ltbno, ltlen,
+						  &ltbnoa, &ltlena);
+			if (ltlena < args->minlen)
+				continue;
+			args->len = XFS_EXTLEN_MIN(ltlena, args->maxlen);
+			xfs_alloc_fix_len(args);
+			ASSERT(args->len >= args->minlen);
+			if (args->len < blen)
+				continue;
+			ltdiff = xfs_alloc_compute_diff(args->agbno, args->len,
+				args->alignment, ltbnoa, ltlena, &ltnew);
+			if (ltnew != NULLAGBLOCK &&
+			    (args->len > blen || ltdiff < bdiff)) {
+				bdiff = ltdiff;
+				bnew = ltnew;
+				blen = args->len;
+				besti = cnt_cur->bc_ptrs[0];
+			}
+		}
+		/*
+		 * It didn't work.  We COULD be in a case where
+		 * there's a good record somewhere, so try again.
+		 */
+		if (blen == 0)
+			break;
+		/*
+		 * Point at the best entry, and retrieve it again.
+		 */
+		cnt_cur->bc_ptrs[0] = besti;
+		if ((error = xfs_alloc_get_rec(cnt_cur, &ltbno, &ltlen, &i)))
+			goto error0;
+		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+		ASSERT(ltbno + ltlen <= be32_to_cpu(XFS_BUF_TO_AGF(args->agbp)->agf_length));
+		args->len = blen;
+		if (!xfs_alloc_fix_minleft(args)) {
+			xfs_btree_del_cursor(cnt_cur, XFS_BTREE_NOERROR);
+			trace_xfs_alloc_near_nominleft(args);
+			return 0;
+		}
+		blen = args->len;
+		/*
+		 * We are allocating starting at bnew for blen blocks.
+		 */
+		args->agbno = bnew;
+		ASSERT(bnew >= ltbno);
+		ASSERT(bnew + blen <= ltbno + ltlen);
+		/*
+		 * Set up a cursor for the by-bno tree.
+		 */
+		bno_cur_lt = xfs_allocbt_init_cursor(args->mp, args->tp,
+			args->agbp, args->agno, XFS_BTNUM_BNO);
+		/*
+		 * Fix up the btree entries.
+		 */
+		if ((error = xfs_alloc_fixup_trees(cnt_cur, bno_cur_lt, ltbno,
+				ltlen, bnew, blen, XFSA_FIXUP_CNT_OK)))
+			goto error0;
+		xfs_btree_del_cursor(cnt_cur, XFS_BTREE_NOERROR);
+		xfs_btree_del_cursor(bno_cur_lt, XFS_BTREE_NOERROR);
+
+		trace_xfs_alloc_near_first(args);
+		return 0;
+	}
+	/*
+	 * Second algorithm.
+	 * Search in the by-bno tree to the left and to the right
+	 * simultaneously, until in each case we find a space big enough,
+	 * or run into the edge of the tree.  When we run into the edge,
+	 * we deallocate that cursor.
+	 * If both searches succeed, we compare the two spaces and pick
+	 * the better one.
+	 * With alignment, it's possible for both to fail; the upper
+	 * level algorithm that picks allocation groups for allocations
+	 * is not supposed to do this.
+	 */
+	/*
+	 * Allocate and initialize the cursor for the leftward search.
+	 */
+	bno_cur_lt = xfs_allocbt_init_cursor(args->mp, args->tp, args->agbp,
+		args->agno, XFS_BTNUM_BNO);
+	/*
+	 * Lookup <= bno to find the leftward search's starting point.
+	 */
+	if ((error = xfs_alloc_lookup_le(bno_cur_lt, args->agbno, args->maxlen, &i)))
+		goto error0;
+	if (!i) {
+		/*
+		 * Didn't find anything; use this cursor for the rightward
+		 * search.
+		 */
+		bno_cur_gt = bno_cur_lt;
+		bno_cur_lt = NULL;
+	}
+	/*
+	 * Found something.  Duplicate the cursor for the rightward search.
+	 */
+	else if ((error = xfs_btree_dup_cursor(bno_cur_lt, &bno_cur_gt)))
+		goto error0;
+	/*
+	 * Increment the cursor, so we will point at the entry just right
+	 * of the leftward entry if any, or to the leftmost entry.
+	 */
+	if ((error = xfs_btree_increment(bno_cur_gt, 0, &i)))
+		goto error0;
+	if (!i) {
+		/*
+		 * It failed, there are no rightward entries.
+		 */
+		xfs_btree_del_cursor(bno_cur_gt, XFS_BTREE_NOERROR);
+		bno_cur_gt = NULL;
+	}
+	/*
+	 * Loop going left with the leftward cursor, right with the
+	 * rightward cursor, until either both directions give up or
+	 * we find an entry at least as big as minlen.
+	 */
+	do {
+		if (bno_cur_lt) {
+			if ((error = xfs_alloc_get_rec(bno_cur_lt, &ltbno, &ltlen, &i)))
+				goto error0;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+			xfs_alloc_compute_aligned(args, ltbno, ltlen,
+						  &ltbnoa, &ltlena);
+			if (ltlena >= args->minlen)
+				break;
+			if ((error = xfs_btree_decrement(bno_cur_lt, 0, &i)))
+				goto error0;
+			if (!i) {
+				xfs_btree_del_cursor(bno_cur_lt,
+						     XFS_BTREE_NOERROR);
+				bno_cur_lt = NULL;
+			}
+		}
+		if (bno_cur_gt) {
+			if ((error = xfs_alloc_get_rec(bno_cur_gt, &gtbno, &gtlen, &i)))
+				goto error0;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+			xfs_alloc_compute_aligned(args, gtbno, gtlen,
+						  &gtbnoa, &gtlena);
+			if (gtlena >= args->minlen)
+				break;
+			if ((error = xfs_btree_increment(bno_cur_gt, 0, &i)))
+				goto error0;
+			if (!i) {
+				xfs_btree_del_cursor(bno_cur_gt,
+						     XFS_BTREE_NOERROR);
+				bno_cur_gt = NULL;
+			}
+		}
+	} while (bno_cur_lt || bno_cur_gt);
+
+	/*
+	 * Got both cursors still active, need to find better entry.
+	 */
+	if (bno_cur_lt && bno_cur_gt) {
+		if (ltlena >= args->minlen) {
+			/*
+			 * Left side is good, look for a right side entry.
+			 */
+			args->len = XFS_EXTLEN_MIN(ltlena, args->maxlen);
+			xfs_alloc_fix_len(args);
+			ltdiff = xfs_alloc_compute_diff(args->agbno, args->len,
+				args->alignment, ltbnoa, ltlena, &ltnew);
+
+			error = xfs_alloc_find_best_extent(args,
+						&bno_cur_lt, &bno_cur_gt,
+						ltdiff, &gtbno, &gtlen,
+						&gtbnoa, &gtlena,
+						0 /* search right */);
+		} else {
+			ASSERT(gtlena >= args->minlen);
+
+			/*
+			 * Right side is good, look for a left side entry.
+			 */
+			args->len = XFS_EXTLEN_MIN(gtlena, args->maxlen);
+			xfs_alloc_fix_len(args);
+			gtdiff = xfs_alloc_compute_diff(args->agbno, args->len,
+				args->alignment, gtbnoa, gtlena, &gtnew);
+
+			error = xfs_alloc_find_best_extent(args,
+						&bno_cur_gt, &bno_cur_lt,
+						gtdiff, &ltbno, &ltlen,
+						&ltbnoa, &ltlena,
+						1 /* search left */);
+		}
+
+		if (error)
+			goto error0;
+	}
+
+	/*
+	 * If we couldn't get anything, give up.
+	 */
+	if (bno_cur_lt == NULL && bno_cur_gt == NULL) {
+		if (!forced++) {
+			trace_xfs_alloc_near_busy(args);
+			xfs_log_force(args->mp, XFS_LOG_SYNC);
+			goto restart;
+		}
+
+		trace_xfs_alloc_size_neither(args);
+		args->agbno = NULLAGBLOCK;
+		return 0;
+	}
+
+	/*
+	 * At this point we have selected a freespace entry, either to the
+	 * left or to the right.  If it's on the right, copy all the
+	 * useful variables to the "left" set so we only have one
+	 * copy of this code.
+	 */
+	if (bno_cur_gt) {
+		bno_cur_lt = bno_cur_gt;
+		bno_cur_gt = NULL;
+		ltbno = gtbno;
+		ltbnoa = gtbnoa;
+		ltlen = gtlen;
+		ltlena = gtlena;
+		j = 1;
+	} else
+		j = 0;
+
+	/*
+	 * Fix up the length and compute the useful address.
+	 */
+	args->len = XFS_EXTLEN_MIN(ltlena, args->maxlen);
+	xfs_alloc_fix_len(args);
+	if (!xfs_alloc_fix_minleft(args)) {
+		trace_xfs_alloc_near_nominleft(args);
+		xfs_btree_del_cursor(bno_cur_lt, XFS_BTREE_NOERROR);
+		xfs_btree_del_cursor(cnt_cur, XFS_BTREE_NOERROR);
+		return 0;
+	}
+	rlen = args->len;
+	(void)xfs_alloc_compute_diff(args->agbno, rlen, args->alignment,
+				     ltbnoa, ltlena, &ltnew);
+	ASSERT(ltnew >= ltbno);
+	ASSERT(ltnew + rlen <= ltbnoa + ltlena);
+	ASSERT(ltnew + rlen <= be32_to_cpu(XFS_BUF_TO_AGF(args->agbp)->agf_length));
+	args->agbno = ltnew;
+
+	if ((error = xfs_alloc_fixup_trees(cnt_cur, bno_cur_lt, ltbno, ltlen,
+			ltnew, rlen, XFSA_FIXUP_BNO_OK)))
+		goto error0;
+
+	if (j)
+		trace_xfs_alloc_near_greater(args);
+	else
+		trace_xfs_alloc_near_lesser(args);
+
+	xfs_btree_del_cursor(cnt_cur, XFS_BTREE_NOERROR);
+	xfs_btree_del_cursor(bno_cur_lt, XFS_BTREE_NOERROR);
+	return 0;
+
+ error0:
+	trace_xfs_alloc_near_error(args);
+	if (cnt_cur != NULL)
+		xfs_btree_del_cursor(cnt_cur, XFS_BTREE_ERROR);
+	if (bno_cur_lt != NULL)
+		xfs_btree_del_cursor(bno_cur_lt, XFS_BTREE_ERROR);
+	if (bno_cur_gt != NULL)
+		xfs_btree_del_cursor(bno_cur_gt, XFS_BTREE_ERROR);
+	return error;
+}
+
+/*
+ * Allocate a variable extent anywhere in the allocation group agno.
+ * Extent's length (returned in len) will be between minlen and maxlen,
+ * and of the form k * prod + mod unless there's nothing that large.
+ * Return the starting a.g. block, or NULLAGBLOCK if we can't do it.
+ */
+STATIC int				/* error */
+xfs_alloc_ag_vextent_size(
+	xfs_alloc_arg_t	*args)		/* allocation argument structure */
+{
+	xfs_btree_cur_t	*bno_cur;	/* cursor for bno btree */
+	xfs_btree_cur_t	*cnt_cur;	/* cursor for cnt btree */
+	int		error;		/* error result */
+	xfs_agblock_t	fbno;		/* start of found freespace */
+	xfs_extlen_t	flen;		/* length of found freespace */
+	int		i;		/* temp status variable */
+	xfs_agblock_t	rbno;		/* returned block number */
+	xfs_extlen_t	rlen;		/* length of returned extent */
+	int		forced = 0;
+
+restart:
+	/*
+	 * Allocate and initialize a cursor for the by-size btree.
+	 */
+	cnt_cur = xfs_allocbt_init_cursor(args->mp, args->tp, args->agbp,
+		args->agno, XFS_BTNUM_CNT);
+	bno_cur = NULL;
+
+	/*
+	 * Look for an entry >= maxlen+alignment-1 blocks.
+	 */
+	if ((error = xfs_alloc_lookup_ge(cnt_cur, 0,
+			args->maxlen + args->alignment - 1, &i)))
+		goto error0;
+
+	/*
+	 * If none or we have busy extents that we cannot allocate from, then
+	 * we have to settle for a smaller extent. In the case that there are
+	 * no large extents, this will return the last entry in the tree unless
+	 * the tree is empty. In the case that there are only busy large
+	 * extents, this will return the largest small extent unless there
+	 * are no smaller extents available.
+	 */
+	if (!i || forced > 1) {
+		error = xfs_alloc_ag_vextent_small(args, cnt_cur,
+						   &fbno, &flen, &i);
+		if (error)
+			goto error0;
+		if (i == 0 || flen == 0) {
+			xfs_btree_del_cursor(cnt_cur, XFS_BTREE_NOERROR);
+			trace_xfs_alloc_size_noentry(args);
+			return 0;
+		}
+		ASSERT(i == 1);
+		xfs_alloc_compute_aligned(args, fbno, flen, &rbno, &rlen);
+	} else {
+		/*
+		 * Search for a non-busy extent that is large enough.
+		 * If we are at low space, don't check, or if we fall of
+		 * the end of the btree, turn off the busy check and
+		 * restart.
+		 */
+		for (;;) {
+			error = xfs_alloc_get_rec(cnt_cur, &fbno, &flen, &i);
+			if (error)
+				goto error0;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+
+			xfs_alloc_compute_aligned(args, fbno, flen,
+						  &rbno, &rlen);
+
+			if (rlen >= args->maxlen)
+				break;
+
+			error = xfs_btree_increment(cnt_cur, 0, &i);
+			if (error)
+				goto error0;
+			if (i == 0) {
+				/*
+				 * Our only valid extents must have been busy.
+				 * Make it unbusy by forcing the log out and
+				 * retrying. If we've been here before, forcing
+				 * the log isn't making the extents available,
+				 * which means they have probably been freed in
+				 * this transaction.  In that case, we have to
+				 * give up on them and we'll attempt a minlen
+				 * allocation the next time around.
+				 */
+				xfs_btree_del_cursor(cnt_cur,
+						     XFS_BTREE_NOERROR);
+				trace_xfs_alloc_size_busy(args);
+				if (!forced++)
+					xfs_log_force(args->mp, XFS_LOG_SYNC);
+				goto restart;
+			}
+		}
+	}
+
+	/*
+	 * In the first case above, we got the last entry in the
+	 * by-size btree.  Now we check to see if the space hits maxlen
+	 * once aligned; if not, we search left for something better.
+	 * This can't happen in the second case above.
+	 */
+	rlen = XFS_EXTLEN_MIN(args->maxlen, rlen);
+	XFS_WANT_CORRUPTED_GOTO(rlen == 0 ||
+			(rlen <= flen && rbno + rlen <= fbno + flen), error0);
+	if (rlen < args->maxlen) {
+		xfs_agblock_t	bestfbno;
+		xfs_extlen_t	bestflen;
+		xfs_agblock_t	bestrbno;
+		xfs_extlen_t	bestrlen;
+
+		bestrlen = rlen;
+		bestrbno = rbno;
+		bestflen = flen;
+		bestfbno = fbno;
+		for (;;) {
+			if ((error = xfs_btree_decrement(cnt_cur, 0, &i)))
+				goto error0;
+			if (i == 0)
+				break;
+			if ((error = xfs_alloc_get_rec(cnt_cur, &fbno, &flen,
+					&i)))
+				goto error0;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+			if (flen < bestrlen)
+				break;
+			xfs_alloc_compute_aligned(args, fbno, flen,
+						  &rbno, &rlen);
+			rlen = XFS_EXTLEN_MIN(args->maxlen, rlen);
+			XFS_WANT_CORRUPTED_GOTO(rlen == 0 ||
+				(rlen <= flen && rbno + rlen <= fbno + flen),
+				error0);
+			if (rlen > bestrlen) {
+				bestrlen = rlen;
+				bestrbno = rbno;
+				bestflen = flen;
+				bestfbno = fbno;
+				if (rlen == args->maxlen)
+					break;
+			}
+		}
+		if ((error = xfs_alloc_lookup_eq(cnt_cur, bestfbno, bestflen,
+				&i)))
+			goto error0;
+		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+		rlen = bestrlen;
+		rbno = bestrbno;
+		flen = bestflen;
+		fbno = bestfbno;
+	}
+	args->wasfromfl = 0;
+	/*
+	 * Fix up the length.
+	 */
+	args->len = rlen;
+	if (rlen < args->minlen) {
+		if (!forced++) {
+			xfs_btree_del_cursor(cnt_cur, XFS_BTREE_NOERROR);
+			trace_xfs_alloc_size_busy(args);
+			xfs_log_force(args->mp, XFS_LOG_SYNC);
+			goto restart;
+		}
+		goto out_nominleft;
+	}
+	xfs_alloc_fix_len(args);
+
+	if (!xfs_alloc_fix_minleft(args))
+		goto out_nominleft;
+	rlen = args->len;
+	XFS_WANT_CORRUPTED_GOTO(rlen <= flen, error0);
+	/*
+	 * Allocate and initialize a cursor for the by-block tree.
+	 */
+	bno_cur = xfs_allocbt_init_cursor(args->mp, args->tp, args->agbp,
+		args->agno, XFS_BTNUM_BNO);
+	if ((error = xfs_alloc_fixup_trees(cnt_cur, bno_cur, fbno, flen,
+			rbno, rlen, XFSA_FIXUP_CNT_OK)))
+		goto error0;
+	xfs_btree_del_cursor(cnt_cur, XFS_BTREE_NOERROR);
+	xfs_btree_del_cursor(bno_cur, XFS_BTREE_NOERROR);
+	cnt_cur = bno_cur = NULL;
+	args->len = rlen;
+	args->agbno = rbno;
+	XFS_WANT_CORRUPTED_GOTO(
+		args->agbno + args->len <=
+			be32_to_cpu(XFS_BUF_TO_AGF(args->agbp)->agf_length),
+		error0);
+	trace_xfs_alloc_size_done(args);
+	return 0;
+
+error0:
+	trace_xfs_alloc_size_error(args);
+	if (cnt_cur)
+		xfs_btree_del_cursor(cnt_cur, XFS_BTREE_ERROR);
+	if (bno_cur)
+		xfs_btree_del_cursor(bno_cur, XFS_BTREE_ERROR);
+	return error;
+
+out_nominleft:
+	xfs_btree_del_cursor(cnt_cur, XFS_BTREE_NOERROR);
+	trace_xfs_alloc_size_nominleft(args);
+	args->agbno = NULLAGBLOCK;
+	return 0;
+}
+
+/*
+ * Deal with the case where only small freespaces remain.
+ * Either return the contents of the last freespace record,
+ * or allocate space from the freelist if there is nothing in the tree.
+ */
+STATIC int			/* error */
+xfs_alloc_ag_vextent_small(
+	xfs_alloc_arg_t	*args,	/* allocation argument structure */
+	xfs_btree_cur_t	*ccur,	/* by-size cursor */
+	xfs_agblock_t	*fbnop,	/* result block number */
+	xfs_extlen_t	*flenp,	/* result length */
+	int		*stat)	/* status: 0-freelist, 1-normal/none */
+{
+	int		error;
+	xfs_agblock_t	fbno;
+	xfs_extlen_t	flen;
+	int		i;
+
+	if ((error = xfs_btree_decrement(ccur, 0, &i)))
+		goto error0;
+	if (i) {
+		if ((error = xfs_alloc_get_rec(ccur, &fbno, &flen, &i)))
+			goto error0;
+		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+	}
+	/*
+	 * Nothing in the btree, try the freelist.  Make sure
+	 * to respect minleft even when pulling from the
+	 * freelist.
+	 */
+	else if (args->minlen == 1 && args->alignment == 1 && !args->isfl &&
+		 (be32_to_cpu(XFS_BUF_TO_AGF(args->agbp)->agf_flcount)
+		  > args->minleft)) {
+		error = xfs_alloc_get_freelist(args->tp, args->agbp, &fbno, 0);
+		if (error)
+			goto error0;
+		if (fbno != NULLAGBLOCK) {
+			xfs_alloc_busy_reuse(args->mp, args->agno, fbno, 1,
+					     args->userdata);
+
+			if (args->userdata) {
+				xfs_buf_t	*bp;
+
+				bp = xfs_btree_get_bufs(args->mp, args->tp,
+					args->agno, fbno, 0);
+				xfs_trans_binval(args->tp, bp);
+			}
+			args->len = 1;
+			args->agbno = fbno;
+			XFS_WANT_CORRUPTED_GOTO(
+				args->agbno + args->len <=
+				be32_to_cpu(XFS_BUF_TO_AGF(args->agbp)->agf_length),
+				error0);
+			args->wasfromfl = 1;
+			trace_xfs_alloc_small_freelist(args);
+			*stat = 0;
+			return 0;
+		}
+		/*
+		 * Nothing in the freelist.
+		 */
+		else
+			flen = 0;
+	}
+	/*
+	 * Can't allocate from the freelist for some reason.
+	 */
+	else {
+		fbno = NULLAGBLOCK;
+		flen = 0;
+	}
+	/*
+	 * Can't do the allocation, give up.
+	 */
+	if (flen < args->minlen) {
+		args->agbno = NULLAGBLOCK;
+		trace_xfs_alloc_small_notenough(args);
+		flen = 0;
+	}
+	*fbnop = fbno;
+	*flenp = flen;
+	*stat = 1;
+	trace_xfs_alloc_small_done(args);
+	return 0;
+
+error0:
+	trace_xfs_alloc_small_error(args);
+	return error;
+}
+
+/*
+ * Free the extent starting at agno/bno for length.
+ */
+STATIC int			/* error */
+xfs_free_ag_extent(
+	xfs_trans_t	*tp,	/* transaction pointer */
+	xfs_buf_t	*agbp,	/* buffer for a.g. freelist header */
+	xfs_agnumber_t	agno,	/* allocation group number */
+	xfs_agblock_t	bno,	/* starting block number */
+	xfs_extlen_t	len,	/* length of extent */
+	int		isfl)	/* set if is freelist blocks - no sb acctg */
+{
+	xfs_btree_cur_t	*bno_cur;	/* cursor for by-block btree */
+	xfs_btree_cur_t	*cnt_cur;	/* cursor for by-size btree */
+	int		error;		/* error return value */
+	xfs_agblock_t	gtbno;		/* start of right neighbor block */
+	xfs_extlen_t	gtlen;		/* length of right neighbor block */
+	int		haveleft;	/* have a left neighbor block */
+	int		haveright;	/* have a right neighbor block */
+	int		i;		/* temp, result code */
+	xfs_agblock_t	ltbno;		/* start of left neighbor block */
+	xfs_extlen_t	ltlen;		/* length of left neighbor block */
+	xfs_mount_t	*mp;		/* mount point struct for filesystem */
+	xfs_agblock_t	nbno;		/* new starting block of freespace */
+	xfs_extlen_t	nlen;		/* new length of freespace */
+	xfs_perag_t	*pag;		/* per allocation group data */
+
+	mp = tp->t_mountp;
+	/*
+	 * Allocate and initialize a cursor for the by-block btree.
+	 */
+	bno_cur = xfs_allocbt_init_cursor(mp, tp, agbp, agno, XFS_BTNUM_BNO);
+	cnt_cur = NULL;
+	/*
+	 * Look for a neighboring block on the left (lower block numbers)
+	 * that is contiguous with this space.
+	 */
+	if ((error = xfs_alloc_lookup_le(bno_cur, bno, len, &haveleft)))
+		goto error0;
+	if (haveleft) {
+		/*
+		 * There is a block to our left.
+		 */
+		if ((error = xfs_alloc_get_rec(bno_cur, &ltbno, &ltlen, &i)))
+			goto error0;
+		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+		/*
+		 * It's not contiguous, though.
+		 */
+		if (ltbno + ltlen < bno)
+			haveleft = 0;
+		else {
+			/*
+			 * If this failure happens the request to free this
+			 * space was invalid, it's (partly) already free.
+			 * Very bad.
+			 */
+			XFS_WANT_CORRUPTED_GOTO(ltbno + ltlen <= bno, error0);
+		}
+	}
+	/*
+	 * Look for a neighboring block on the right (higher block numbers)
+	 * that is contiguous with this space.
+	 */
+	if ((error = xfs_btree_increment(bno_cur, 0, &haveright)))
+		goto error0;
+	if (haveright) {
+		/*
+		 * There is a block to our right.
+		 */
+		if ((error = xfs_alloc_get_rec(bno_cur, &gtbno, &gtlen, &i)))
+			goto error0;
+		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+		/*
+		 * It's not contiguous, though.
+		 */
+		if (bno + len < gtbno)
+			haveright = 0;
+		else {
+			/*
+			 * If this failure happens the request to free this
+			 * space was invalid, it's (partly) already free.
+			 * Very bad.
+			 */
+			XFS_WANT_CORRUPTED_GOTO(gtbno >= bno + len, error0);
+		}
+	}
+	/*
+	 * Now allocate and initialize a cursor for the by-size tree.
+	 */
+	cnt_cur = xfs_allocbt_init_cursor(mp, tp, agbp, agno, XFS_BTNUM_CNT);
+	/*
+	 * Have both left and right contiguous neighbors.
+	 * Merge all three into a single free block.
+	 */
+	if (haveleft && haveright) {
+		/*
+		 * Delete the old by-size entry on the left.
+		 */
+		if ((error = xfs_alloc_lookup_eq(cnt_cur, ltbno, ltlen, &i)))
+			goto error0;
+		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+		if ((error = xfs_btree_delete(cnt_cur, &i)))
+			goto error0;
+		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+		/*
+		 * Delete the old by-size entry on the right.
+		 */
+		if ((error = xfs_alloc_lookup_eq(cnt_cur, gtbno, gtlen, &i)))
+			goto error0;
+		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+		if ((error = xfs_btree_delete(cnt_cur, &i)))
+			goto error0;
+		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+		/*
+		 * Delete the old by-block entry for the right block.
+		 */
+		if ((error = xfs_btree_delete(bno_cur, &i)))
+			goto error0;
+		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+		/*
+		 * Move the by-block cursor back to the left neighbor.
+		 */
+		if ((error = xfs_btree_decrement(bno_cur, 0, &i)))
+			goto error0;
+		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+#ifdef DEBUG
+		/*
+		 * Check that this is the right record: delete didn't
+		 * mangle the cursor.
+		 */
+		{
+			xfs_agblock_t	xxbno;
+			xfs_extlen_t	xxlen;
+
+			if ((error = xfs_alloc_get_rec(bno_cur, &xxbno, &xxlen,
+					&i)))
+				goto error0;
+			XFS_WANT_CORRUPTED_GOTO(
+				i == 1 && xxbno == ltbno && xxlen == ltlen,
+				error0);
+		}
+#endif
+		/*
+		 * Update remaining by-block entry to the new, joined block.
+		 */
+		nbno = ltbno;
+		nlen = len + ltlen + gtlen;
+		if ((error = xfs_alloc_update(bno_cur, nbno, nlen)))
+			goto error0;
+	}
+	/*
+	 * Have only a left contiguous neighbor.
+	 * Merge it together with the new freespace.
+	 */
+	else if (haveleft) {
+		/*
+		 * Delete the old by-size entry on the left.
+		 */
+		if ((error = xfs_alloc_lookup_eq(cnt_cur, ltbno, ltlen, &i)))
+			goto error0;
+		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+		if ((error = xfs_btree_delete(cnt_cur, &i)))
+			goto error0;
+		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+		/*
+		 * Back up the by-block cursor to the left neighbor, and
+		 * update its length.
+		 */
+		if ((error = xfs_btree_decrement(bno_cur, 0, &i)))
+			goto error0;
+		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+		nbno = ltbno;
+		nlen = len + ltlen;
+		if ((error = xfs_alloc_update(bno_cur, nbno, nlen)))
+			goto error0;
+	}
+	/*
+	 * Have only a right contiguous neighbor.
+	 * Merge it together with the new freespace.
+	 */
+	else if (haveright) {
+		/*
+		 * Delete the old by-size entry on the right.
+		 */
+		if ((error = xfs_alloc_lookup_eq(cnt_cur, gtbno, gtlen, &i)))
+			goto error0;
+		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+		if ((error = xfs_btree_delete(cnt_cur, &i)))
+			goto error0;
+		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+		/*
+		 * Update the starting block and length of the right
+		 * neighbor in the by-block tree.
+		 */
+		nbno = bno;
+		nlen = len + gtlen;
+		if ((error = xfs_alloc_update(bno_cur, nbno, nlen)))
+			goto error0;
+	}
+	/*
+	 * No contiguous neighbors.
+	 * Insert the new freespace into the by-block tree.
+	 */
+	else {
+		nbno = bno;
+		nlen = len;
+		if ((error = xfs_btree_insert(bno_cur, &i)))
+			goto error0;
+		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+	}
+	xfs_btree_del_cursor(bno_cur, XFS_BTREE_NOERROR);
+	bno_cur = NULL;
+	/*
+	 * In all cases we need to insert the new freespace in the by-size tree.
+	 */
+	if ((error = xfs_alloc_lookup_eq(cnt_cur, nbno, nlen, &i)))
+		goto error0;
+	XFS_WANT_CORRUPTED_GOTO(i == 0, error0);
+	if ((error = xfs_btree_insert(cnt_cur, &i)))
+		goto error0;
+	XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+	xfs_btree_del_cursor(cnt_cur, XFS_BTREE_NOERROR);
+	cnt_cur = NULL;
+
+	/*
+	 * Update the freespace totals in the ag and superblock.
+	 */
+	pag = xfs_perag_get(mp, agno);
+	error = xfs_alloc_update_counters(tp, pag, agbp, len);
+	xfs_perag_put(pag);
+	if (error)
+		goto error0;
+
+	if (!isfl)
+		xfs_trans_mod_sb(tp, XFS_TRANS_SB_FDBLOCKS, (long)len);
+	XFS_STATS_INC(xs_freex);
+	XFS_STATS_ADD(xs_freeb, len);
+
+	trace_xfs_free_extent(mp, agno, bno, len, isfl, haveleft, haveright);
+
+	return 0;
+
+ error0:
+	trace_xfs_free_extent(mp, agno, bno, len, isfl, -1, -1);
+	if (bno_cur)
+		xfs_btree_del_cursor(bno_cur, XFS_BTREE_ERROR);
+	if (cnt_cur)
+		xfs_btree_del_cursor(cnt_cur, XFS_BTREE_ERROR);
+	return error;
+}
+
+/*
+ * Visible (exported) allocation/free functions.
+ * Some of these are used just by xfs_alloc_btree.c and this file.
+ */
+
+/*
+ * Compute and fill in value of m_ag_maxlevels.
+ */
+void
+xfs_alloc_compute_maxlevels(
+	xfs_mount_t	*mp)	/* file system mount structure */
+{
+	int		level;
+	uint		maxblocks;
+	uint		maxleafents;
+	int		minleafrecs;
+	int		minnoderecs;
+
+	maxleafents = (mp->m_sb.sb_agblocks + 1) / 2;
+	minleafrecs = mp->m_alloc_mnr[0];
+	minnoderecs = mp->m_alloc_mnr[1];
+	maxblocks = (maxleafents + minleafrecs - 1) / minleafrecs;
+	for (level = 1; maxblocks > 1; level++)
+		maxblocks = (maxblocks + minnoderecs - 1) / minnoderecs;
+	mp->m_ag_maxlevels = level;
+}
+
+/*
+ * Find the length of the longest extent in an AG.
+ */
+xfs_extlen_t
+xfs_alloc_longest_free_extent(
+	struct xfs_mount	*mp,
+	struct xfs_perag	*pag)
+{
+	xfs_extlen_t		need, delta = 0;
+
+	need = XFS_MIN_FREELIST_PAG(pag, mp);
+	if (need > pag->pagf_flcount)
+		delta = need - pag->pagf_flcount;
+
+	if (pag->pagf_longest > delta)
+		return pag->pagf_longest - delta;
+	return pag->pagf_flcount > 0 || pag->pagf_longest > 0;
+}
+
+/*
+ * Decide whether to use this allocation group for this allocation.
+ * If so, fix up the btree freelist's size.
+ */
+STATIC int			/* error */
+xfs_alloc_fix_freelist(
+	xfs_alloc_arg_t	*args,	/* allocation argument structure */
+	int		flags)	/* XFS_ALLOC_FLAG_... */
+{
+	xfs_buf_t	*agbp;	/* agf buffer pointer */
+	xfs_agf_t	*agf;	/* a.g. freespace structure pointer */
+	xfs_buf_t	*agflbp;/* agfl buffer pointer */
+	xfs_agblock_t	bno;	/* freelist block */
+	xfs_extlen_t	delta;	/* new blocks needed in freelist */
+	int		error;	/* error result code */
+	xfs_extlen_t	longest;/* longest extent in allocation group */
+	xfs_mount_t	*mp;	/* file system mount point structure */
+	xfs_extlen_t	need;	/* total blocks needed in freelist */
+	xfs_perag_t	*pag;	/* per-ag information structure */
+	xfs_alloc_arg_t	targs;	/* local allocation arguments */
+	xfs_trans_t	*tp;	/* transaction pointer */
+
+	mp = args->mp;
+
+	pag = args->pag;
+	tp = args->tp;
+	if (!pag->pagf_init) {
+		if ((error = xfs_alloc_read_agf(mp, tp, args->agno, flags,
+				&agbp)))
+			return error;
+		if (!pag->pagf_init) {
+			ASSERT(flags & XFS_ALLOC_FLAG_TRYLOCK);
+			ASSERT(!(flags & XFS_ALLOC_FLAG_FREEING));
+			args->agbp = NULL;
+			return 0;
+		}
+	} else
+		agbp = NULL;
+
+	/*
+	 * If this is a metadata preferred pag and we are user data
+	 * then try somewhere else if we are not being asked to
+	 * try harder at this point
+	 */
+	if (pag->pagf_metadata && args->userdata &&
+	    (flags & XFS_ALLOC_FLAG_TRYLOCK)) {
+		ASSERT(!(flags & XFS_ALLOC_FLAG_FREEING));
+		args->agbp = NULL;
+		return 0;
+	}
+
+	if (!(flags & XFS_ALLOC_FLAG_FREEING)) {
+		/*
+		 * If it looks like there isn't a long enough extent, or enough
+		 * total blocks, reject it.
+		 */
+		need = XFS_MIN_FREELIST_PAG(pag, mp);
+		longest = xfs_alloc_longest_free_extent(mp, pag);
+		if ((args->minlen + args->alignment + args->minalignslop - 1) >
+				longest ||
+		    ((int)(pag->pagf_freeblks + pag->pagf_flcount -
+			   need - args->total) < (int)args->minleft)) {
+			if (agbp)
+				xfs_trans_brelse(tp, agbp);
+			args->agbp = NULL;
+			return 0;
+		}
+	}
+
+	/*
+	 * Get the a.g. freespace buffer.
+	 * Can fail if we're not blocking on locks, and it's held.
+	 */
+	if (agbp == NULL) {
+		if ((error = xfs_alloc_read_agf(mp, tp, args->agno, flags,
+				&agbp)))
+			return error;
+		if (agbp == NULL) {
+			ASSERT(flags & XFS_ALLOC_FLAG_TRYLOCK);
+			ASSERT(!(flags & XFS_ALLOC_FLAG_FREEING));
+			args->agbp = NULL;
+			return 0;
+		}
+	}
+	/*
+	 * Figure out how many blocks we should have in the freelist.
+	 */
+	agf = XFS_BUF_TO_AGF(agbp);
+	need = XFS_MIN_FREELIST(agf, mp);
+	/*
+	 * If there isn't enough total or single-extent, reject it.
+	 */
+	if (!(flags & XFS_ALLOC_FLAG_FREEING)) {
+		delta = need > be32_to_cpu(agf->agf_flcount) ?
+			(need - be32_to_cpu(agf->agf_flcount)) : 0;
+		longest = be32_to_cpu(agf->agf_longest);
+		longest = (longest > delta) ? (longest - delta) :
+			(be32_to_cpu(agf->agf_flcount) > 0 || longest > 0);
+		if ((args->minlen + args->alignment + args->minalignslop - 1) >
+				longest ||
+		    ((int)(be32_to_cpu(agf->agf_freeblks) +
+		     be32_to_cpu(agf->agf_flcount) - need - args->total) <
+				(int)args->minleft)) {
+			xfs_trans_brelse(tp, agbp);
+			args->agbp = NULL;
+			return 0;
+		}
+	}
+	/*
+	 * Make the freelist shorter if it's too long.
+	 */
+	while (be32_to_cpu(agf->agf_flcount) > need) {
+		xfs_buf_t	*bp;
+
+		error = xfs_alloc_get_freelist(tp, agbp, &bno, 0);
+		if (error)
+			return error;
+		if ((error = xfs_free_ag_extent(tp, agbp, args->agno, bno, 1, 1)))
+			return error;
+		bp = xfs_btree_get_bufs(mp, tp, args->agno, bno, 0);
+		xfs_trans_binval(tp, bp);
+	}
+	/*
+	 * Initialize the args structure.
+	 */
+	targs.tp = tp;
+	targs.mp = mp;
+	targs.agbp = agbp;
+	targs.agno = args->agno;
+	targs.mod = targs.minleft = targs.wasdel = targs.userdata =
+		targs.minalignslop = 0;
+	targs.alignment = targs.minlen = targs.prod = targs.isfl = 1;
+	targs.type = XFS_ALLOCTYPE_THIS_AG;
+	targs.pag = pag;
+	if ((error = xfs_alloc_read_agfl(mp, tp, targs.agno, &agflbp)))
+		return error;
+	/*
+	 * Make the freelist longer if it's too short.
+	 */
+	while (be32_to_cpu(agf->agf_flcount) < need) {
+		targs.agbno = 0;
+		targs.maxlen = need - be32_to_cpu(agf->agf_flcount);
+		/*
+		 * Allocate as many blocks as possible at once.
+		 */
+		if ((error = xfs_alloc_ag_vextent(&targs))) {
+			xfs_trans_brelse(tp, agflbp);
+			return error;
+		}
+		/*
+		 * Stop if we run out.  Won't happen if callers are obeying
+		 * the restrictions correctly.  Can happen for free calls
+		 * on a completely full ag.
+		 */
+		if (targs.agbno == NULLAGBLOCK) {
+			if (flags & XFS_ALLOC_FLAG_FREEING)
+				break;
+			xfs_trans_brelse(tp, agflbp);
+			args->agbp = NULL;
+			return 0;
+		}
+		/*
+		 * Put each allocated block on the list.
+		 */
+		for (bno = targs.agbno; bno < targs.agbno + targs.len; bno++) {
+			error = xfs_alloc_put_freelist(tp, agbp,
+							agflbp, bno, 0);
+			if (error)
+				return error;
+		}
+	}
+	xfs_trans_brelse(tp, agflbp);
+	args->agbp = agbp;
+	return 0;
+}
+
+/*
+ * Get a block from the freelist.
+ * Returns with the buffer for the block gotten.
+ */
+int				/* error */
+xfs_alloc_get_freelist(
+	xfs_trans_t	*tp,	/* transaction pointer */
+	xfs_buf_t	*agbp,	/* buffer containing the agf structure */
+	xfs_agblock_t	*bnop,	/* block address retrieved from freelist */
+	int		btreeblk) /* destination is a AGF btree */
+{
+	xfs_agf_t	*agf;	/* a.g. freespace structure */
+	xfs_agfl_t	*agfl;	/* a.g. freelist structure */
+	xfs_buf_t	*agflbp;/* buffer for a.g. freelist structure */
+	xfs_agblock_t	bno;	/* block number returned */
+	int		error;
+	int		logflags;
+	xfs_mount_t	*mp;	/* mount structure */
+	xfs_perag_t	*pag;	/* per allocation group data */
+
+	agf = XFS_BUF_TO_AGF(agbp);
+	/*
+	 * Freelist is empty, give up.
+	 */
+	if (!agf->agf_flcount) {
+		*bnop = NULLAGBLOCK;
+		return 0;
+	}
+	/*
+	 * Read the array of free blocks.
+	 */
+	mp = tp->t_mountp;
+	if ((error = xfs_alloc_read_agfl(mp, tp,
+			be32_to_cpu(agf->agf_seqno), &agflbp)))
+		return error;
+	agfl = XFS_BUF_TO_AGFL(agflbp);
+	/*
+	 * Get the block number and update the data structures.
+	 */
+	bno = be32_to_cpu(agfl->agfl_bno[be32_to_cpu(agf->agf_flfirst)]);
+	be32_add_cpu(&agf->agf_flfirst, 1);
+	xfs_trans_brelse(tp, agflbp);
+	if (be32_to_cpu(agf->agf_flfirst) == XFS_AGFL_SIZE(mp))
+		agf->agf_flfirst = 0;
+
+	pag = xfs_perag_get(mp, be32_to_cpu(agf->agf_seqno));
+	be32_add_cpu(&agf->agf_flcount, -1);
+	xfs_trans_agflist_delta(tp, -1);
+	pag->pagf_flcount--;
+	xfs_perag_put(pag);
+
+	logflags = XFS_AGF_FLFIRST | XFS_AGF_FLCOUNT;
+	if (btreeblk) {
+		be32_add_cpu(&agf->agf_btreeblks, 1);
+		pag->pagf_btreeblks++;
+		logflags |= XFS_AGF_BTREEBLKS;
+	}
+
+	xfs_alloc_log_agf(tp, agbp, logflags);
+	*bnop = bno;
+
+	return 0;
+}
+
+/*
+ * Log the given fields from the agf structure.
+ */
+void
+xfs_alloc_log_agf(
+	xfs_trans_t	*tp,	/* transaction pointer */
+	xfs_buf_t	*bp,	/* buffer for a.g. freelist header */
+	int		fields)	/* mask of fields to be logged (XFS_AGF_...) */
+{
+	int	first;		/* first byte offset */
+	int	last;		/* last byte offset */
+	static const short	offsets[] = {
+		offsetof(xfs_agf_t, agf_magicnum),
+		offsetof(xfs_agf_t, agf_versionnum),
+		offsetof(xfs_agf_t, agf_seqno),
+		offsetof(xfs_agf_t, agf_length),
+		offsetof(xfs_agf_t, agf_roots[0]),
+		offsetof(xfs_agf_t, agf_levels[0]),
+		offsetof(xfs_agf_t, agf_flfirst),
+		offsetof(xfs_agf_t, agf_fllast),
+		offsetof(xfs_agf_t, agf_flcount),
+		offsetof(xfs_agf_t, agf_freeblks),
+		offsetof(xfs_agf_t, agf_longest),
+		offsetof(xfs_agf_t, agf_btreeblks),
+		sizeof(xfs_agf_t)
+	};
+
+	trace_xfs_agf(tp->t_mountp, XFS_BUF_TO_AGF(bp), fields, _RET_IP_);
+
+	xfs_btree_offsets(fields, offsets, XFS_AGF_NUM_BITS, &first, &last);
+	xfs_trans_log_buf(tp, bp, (uint)first, (uint)last);
+}
+
+/*
+ * Interface for inode allocation to force the pag data to be initialized.
+ */
+int					/* error */
+xfs_alloc_pagf_init(
+	xfs_mount_t		*mp,	/* file system mount structure */
+	xfs_trans_t		*tp,	/* transaction pointer */
+	xfs_agnumber_t		agno,	/* allocation group number */
+	int			flags)	/* XFS_ALLOC_FLAGS_... */
+{
+	xfs_buf_t		*bp;
+	int			error;
+
+	if ((error = xfs_alloc_read_agf(mp, tp, agno, flags, &bp)))
+		return error;
+	if (bp)
+		xfs_trans_brelse(tp, bp);
+	return 0;
+}
+
+/*
+ * Put the block on the freelist for the allocation group.
+ */
+int					/* error */
+xfs_alloc_put_freelist(
+	xfs_trans_t		*tp,	/* transaction pointer */
+	xfs_buf_t		*agbp,	/* buffer for a.g. freelist header */
+	xfs_buf_t		*agflbp,/* buffer for a.g. free block array */
+	xfs_agblock_t		bno,	/* block being freed */
+	int			btreeblk) /* block came from a AGF btree */
+{
+	xfs_agf_t		*agf;	/* a.g. freespace structure */
+	xfs_agfl_t		*agfl;	/* a.g. free block array */
+	__be32			*blockp;/* pointer to array entry */
+	int			error;
+	int			logflags;
+	xfs_mount_t		*mp;	/* mount structure */
+	xfs_perag_t		*pag;	/* per allocation group data */
+
+	agf = XFS_BUF_TO_AGF(agbp);
+	mp = tp->t_mountp;
+
+	if (!agflbp && (error = xfs_alloc_read_agfl(mp, tp,
+			be32_to_cpu(agf->agf_seqno), &agflbp)))
+		return error;
+	agfl = XFS_BUF_TO_AGFL(agflbp);
+	be32_add_cpu(&agf->agf_fllast, 1);
+	if (be32_to_cpu(agf->agf_fllast) == XFS_AGFL_SIZE(mp))
+		agf->agf_fllast = 0;
+
+	pag = xfs_perag_get(mp, be32_to_cpu(agf->agf_seqno));
+	be32_add_cpu(&agf->agf_flcount, 1);
+	xfs_trans_agflist_delta(tp, 1);
+	pag->pagf_flcount++;
+
+	logflags = XFS_AGF_FLLAST | XFS_AGF_FLCOUNT;
+	if (btreeblk) {
+		be32_add_cpu(&agf->agf_btreeblks, -1);
+		pag->pagf_btreeblks--;
+		logflags |= XFS_AGF_BTREEBLKS;
+	}
+	xfs_perag_put(pag);
+
+	xfs_alloc_log_agf(tp, agbp, logflags);
+
+	ASSERT(be32_to_cpu(agf->agf_flcount) <= XFS_AGFL_SIZE(mp));
+	blockp = &agfl->agfl_bno[be32_to_cpu(agf->agf_fllast)];
+	*blockp = cpu_to_be32(bno);
+	xfs_alloc_log_agf(tp, agbp, logflags);
+	xfs_trans_log_buf(tp, agflbp,
+		(int)((xfs_caddr_t)blockp - (xfs_caddr_t)agfl),
+		(int)((xfs_caddr_t)blockp - (xfs_caddr_t)agfl +
+			sizeof(xfs_agblock_t) - 1));
+	return 0;
+}
+
+/*
+ * Read in the allocation group header (free/alloc section).
+ */
+int					/* error */
+xfs_read_agf(
+	struct xfs_mount	*mp,	/* mount point structure */
+	struct xfs_trans	*tp,	/* transaction pointer */
+	xfs_agnumber_t		agno,	/* allocation group number */
+	int			flags,	/* XFS_BUF_ */
+	struct xfs_buf		**bpp)	/* buffer for the ag freelist header */
+{
+	struct xfs_agf	*agf;		/* ag freelist header */
+	int		agf_ok;		/* set if agf is consistent */
+	int		error;
+
+	ASSERT(agno != NULLAGNUMBER);
+	error = xfs_trans_read_buf(
+			mp, tp, mp->m_ddev_targp,
+			XFS_AG_DADDR(mp, agno, XFS_AGF_DADDR(mp)),
+			XFS_FSS_TO_BB(mp, 1), flags, bpp);
+	if (error)
+		return error;
+	if (!*bpp)
+		return 0;
+
+	ASSERT(!XFS_BUF_GETERROR(*bpp));
+	agf = XFS_BUF_TO_AGF(*bpp);
+
+	/*
+	 * Validate the magic number of the agf block.
+	 */
+	agf_ok =
+		be32_to_cpu(agf->agf_magicnum) == XFS_AGF_MAGIC &&
+		XFS_AGF_GOOD_VERSION(be32_to_cpu(agf->agf_versionnum)) &&
+		be32_to_cpu(agf->agf_freeblks) <= be32_to_cpu(agf->agf_length) &&
+		be32_to_cpu(agf->agf_flfirst) < XFS_AGFL_SIZE(mp) &&
+		be32_to_cpu(agf->agf_fllast) < XFS_AGFL_SIZE(mp) &&
+		be32_to_cpu(agf->agf_flcount) <= XFS_AGFL_SIZE(mp) &&
+		be32_to_cpu(agf->agf_seqno) == agno;
+	if (xfs_sb_version_haslazysbcount(&mp->m_sb))
+		agf_ok = agf_ok && be32_to_cpu(agf->agf_btreeblks) <=
+						be32_to_cpu(agf->agf_length);
+	if (unlikely(XFS_TEST_ERROR(!agf_ok, mp, XFS_ERRTAG_ALLOC_READ_AGF,
+			XFS_RANDOM_ALLOC_READ_AGF))) {
+		XFS_CORRUPTION_ERROR("xfs_alloc_read_agf",
+				     XFS_ERRLEVEL_LOW, mp, agf);
+		xfs_trans_brelse(tp, *bpp);
+		return XFS_ERROR(EFSCORRUPTED);
+	}
+	XFS_BUF_SET_VTYPE_REF(*bpp, B_FS_AGF, XFS_AGF_REF);
+	return 0;
+}
+
+/*
+ * Read in the allocation group header (free/alloc section).
+ */
+int					/* error */
+xfs_alloc_read_agf(
+	struct xfs_mount	*mp,	/* mount point structure */
+	struct xfs_trans	*tp,	/* transaction pointer */
+	xfs_agnumber_t		agno,	/* allocation group number */
+	int			flags,	/* XFS_ALLOC_FLAG_... */
+	struct xfs_buf		**bpp)	/* buffer for the ag freelist header */
+{
+	struct xfs_agf		*agf;		/* ag freelist header */
+	struct xfs_perag	*pag;		/* per allocation group data */
+	int			error;
+
+	ASSERT(agno != NULLAGNUMBER);
+
+	error = xfs_read_agf(mp, tp, agno,
+			(flags & XFS_ALLOC_FLAG_TRYLOCK) ? XBF_TRYLOCK : 0,
+			bpp);
+	if (error)
+		return error;
+	if (!*bpp)
+		return 0;
+	ASSERT(!XFS_BUF_GETERROR(*bpp));
+
+	agf = XFS_BUF_TO_AGF(*bpp);
+	pag = xfs_perag_get(mp, agno);
+	if (!pag->pagf_init) {
+		pag->pagf_freeblks = be32_to_cpu(agf->agf_freeblks);
+		pag->pagf_btreeblks = be32_to_cpu(agf->agf_btreeblks);
+		pag->pagf_flcount = be32_to_cpu(agf->agf_flcount);
+		pag->pagf_longest = be32_to_cpu(agf->agf_longest);
+		pag->pagf_levels[XFS_BTNUM_BNOi] =
+			be32_to_cpu(agf->agf_levels[XFS_BTNUM_BNOi]);
+		pag->pagf_levels[XFS_BTNUM_CNTi] =
+			be32_to_cpu(agf->agf_levels[XFS_BTNUM_CNTi]);
+		spin_lock_init(&pag->pagb_lock);
+		pag->pagb_count = 0;
+		pag->pagb_tree = RB_ROOT;
+		pag->pagf_init = 1;
+	}
+#ifdef DEBUG
+	else if (!XFS_FORCED_SHUTDOWN(mp)) {
+		ASSERT(pag->pagf_freeblks == be32_to_cpu(agf->agf_freeblks));
+		ASSERT(pag->pagf_btreeblks == be32_to_cpu(agf->agf_btreeblks));
+		ASSERT(pag->pagf_flcount == be32_to_cpu(agf->agf_flcount));
+		ASSERT(pag->pagf_longest == be32_to_cpu(agf->agf_longest));
+		ASSERT(pag->pagf_levels[XFS_BTNUM_BNOi] ==
+		       be32_to_cpu(agf->agf_levels[XFS_BTNUM_BNOi]));
+		ASSERT(pag->pagf_levels[XFS_BTNUM_CNTi] ==
+		       be32_to_cpu(agf->agf_levels[XFS_BTNUM_CNTi]));
+	}
+#endif
+	xfs_perag_put(pag);
+	return 0;
+}
+
+/*
+ * Allocate an extent (variable-size).
+ * Depending on the allocation type, we either look in a single allocation
+ * group or loop over the allocation groups to find the result.
+ */
+int				/* error */
+xfs_alloc_vextent(
+	xfs_alloc_arg_t	*args)	/* allocation argument structure */
+{
+	xfs_agblock_t	agsize;	/* allocation group size */
+	int		error;
+	int		flags;	/* XFS_ALLOC_FLAG_... locking flags */
+	xfs_extlen_t	minleft;/* minimum left value, temp copy */
+	xfs_mount_t	*mp;	/* mount structure pointer */
+	xfs_agnumber_t	sagno;	/* starting allocation group number */
+	xfs_alloctype_t	type;	/* input allocation type */
+	int		bump_rotor = 0;
+	int		no_min = 0;
+	xfs_agnumber_t	rotorstep = xfs_rotorstep; /* inode32 agf stepper */
+
+	mp = args->mp;
+	type = args->otype = args->type;
+	args->agbno = NULLAGBLOCK;
+	/*
+	 * Just fix this up, for the case where the last a.g. is shorter
+	 * (or there's only one a.g.) and the caller couldn't easily figure
+	 * that out (xfs_bmap_alloc).
+	 */
+	agsize = mp->m_sb.sb_agblocks;
+	if (args->maxlen > agsize)
+		args->maxlen = agsize;
+	if (args->alignment == 0)
+		args->alignment = 1;
+	ASSERT(XFS_FSB_TO_AGNO(mp, args->fsbno) < mp->m_sb.sb_agcount);
+	ASSERT(XFS_FSB_TO_AGBNO(mp, args->fsbno) < agsize);
+	ASSERT(args->minlen <= args->maxlen);
+	ASSERT(args->minlen <= agsize);
+	ASSERT(args->mod < args->prod);
+	if (XFS_FSB_TO_AGNO(mp, args->fsbno) >= mp->m_sb.sb_agcount ||
+	    XFS_FSB_TO_AGBNO(mp, args->fsbno) >= agsize ||
+	    args->minlen > args->maxlen || args->minlen > agsize ||
+	    args->mod >= args->prod) {
+		args->fsbno = NULLFSBLOCK;
+		trace_xfs_alloc_vextent_badargs(args);
+		return 0;
+	}
+	minleft = args->minleft;
+
+	switch (type) {
+	case XFS_ALLOCTYPE_THIS_AG:
+	case XFS_ALLOCTYPE_NEAR_BNO:
+	case XFS_ALLOCTYPE_THIS_BNO:
+		/*
+		 * These three force us into a single a.g.
+		 */
+		args->agno = XFS_FSB_TO_AGNO(mp, args->fsbno);
+		args->pag = xfs_perag_get(mp, args->agno);
+		args->minleft = 0;
+		error = xfs_alloc_fix_freelist(args, 0);
+		args->minleft = minleft;
+		if (error) {
+			trace_xfs_alloc_vextent_nofix(args);
+			goto error0;
+		}
+		if (!args->agbp) {
+			trace_xfs_alloc_vextent_noagbp(args);
+			break;
+		}
+		args->agbno = XFS_FSB_TO_AGBNO(mp, args->fsbno);
+		if ((error = xfs_alloc_ag_vextent(args)))
+			goto error0;
+		break;
+	case XFS_ALLOCTYPE_START_BNO:
+		/*
+		 * Try near allocation first, then anywhere-in-ag after
+		 * the first a.g. fails.
+		 */
+		if ((args->userdata  == XFS_ALLOC_INITIAL_USER_DATA) &&
+		    (mp->m_flags & XFS_MOUNT_32BITINODES)) {
+			args->fsbno = XFS_AGB_TO_FSB(mp,
+					((mp->m_agfrotor / rotorstep) %
+					mp->m_sb.sb_agcount), 0);
+			bump_rotor = 1;
+		}
+		args->agbno = XFS_FSB_TO_AGBNO(mp, args->fsbno);
+		args->type = XFS_ALLOCTYPE_NEAR_BNO;
+		/* FALLTHROUGH */
+	case XFS_ALLOCTYPE_ANY_AG:
+	case XFS_ALLOCTYPE_START_AG:
+	case XFS_ALLOCTYPE_FIRST_AG:
+		/*
+		 * Rotate through the allocation groups looking for a winner.
+		 */
+		if (type == XFS_ALLOCTYPE_ANY_AG) {
+			/*
+			 * Start with the last place we left off.
+			 */
+			args->agno = sagno = (mp->m_agfrotor / rotorstep) %
+					mp->m_sb.sb_agcount;
+			args->type = XFS_ALLOCTYPE_THIS_AG;
+			flags = XFS_ALLOC_FLAG_TRYLOCK;
+		} else if (type == XFS_ALLOCTYPE_FIRST_AG) {
+			/*
+			 * Start with allocation group given by bno.
+			 */
+			args->agno = XFS_FSB_TO_AGNO(mp, args->fsbno);
+			args->type = XFS_ALLOCTYPE_THIS_AG;
+			sagno = 0;
+			flags = 0;
+		} else {
+			if (type == XFS_ALLOCTYPE_START_AG)
+				args->type = XFS_ALLOCTYPE_THIS_AG;
+			/*
+			 * Start with the given allocation group.
+			 */
+			args->agno = sagno = XFS_FSB_TO_AGNO(mp, args->fsbno);
+			flags = XFS_ALLOC_FLAG_TRYLOCK;
+		}
+		/*
+		 * Loop over allocation groups twice; first time with
+		 * trylock set, second time without.
+		 */
+		for (;;) {
+			args->pag = xfs_perag_get(mp, args->agno);
+			if (no_min) args->minleft = 0;
+			error = xfs_alloc_fix_freelist(args, flags);
+			args->minleft = minleft;
+			if (error) {
+				trace_xfs_alloc_vextent_nofix(args);
+				goto error0;
+			}
+			/*
+			 * If we get a buffer back then the allocation will fly.
+			 */
+			if (args->agbp) {
+				if ((error = xfs_alloc_ag_vextent(args)))
+					goto error0;
+				break;
+			}
+
+			trace_xfs_alloc_vextent_loopfailed(args);
+
+			/*
+			 * Didn't work, figure out the next iteration.
+			 */
+			if (args->agno == sagno &&
+			    type == XFS_ALLOCTYPE_START_BNO)
+				args->type = XFS_ALLOCTYPE_THIS_AG;
+			/*
+			* For the first allocation, we can try any AG to get
+			* space.  However, if we already have allocated a
+			* block, we don't want to try AGs whose number is below
+			* sagno. Otherwise, we may end up with out-of-order
+			* locking of AGF, which might cause deadlock.
+			*/
+			if (++(args->agno) == mp->m_sb.sb_agcount) {
+				if (args->firstblock != NULLFSBLOCK)
+					args->agno = sagno;
+				else
+					args->agno = 0;
+			}
+			/*
+			 * Reached the starting a.g., must either be done
+			 * or switch to non-trylock mode.
+			 */
+			if (args->agno == sagno) {
+				if (no_min == 1) {
+					args->agbno = NULLAGBLOCK;
+					trace_xfs_alloc_vextent_allfailed(args);
+					break;
+				}
+				if (flags == 0) {
+					no_min = 1;
+				} else {
+					flags = 0;
+					if (type == XFS_ALLOCTYPE_START_BNO) {
+						args->agbno = XFS_FSB_TO_AGBNO(mp,
+							args->fsbno);
+						args->type = XFS_ALLOCTYPE_NEAR_BNO;
+					}
+				}
+			}
+			xfs_perag_put(args->pag);
+		}
+		if (bump_rotor || (type == XFS_ALLOCTYPE_ANY_AG)) {
+			if (args->agno == sagno)
+				mp->m_agfrotor = (mp->m_agfrotor + 1) %
+					(mp->m_sb.sb_agcount * rotorstep);
+			else
+				mp->m_agfrotor = (args->agno * rotorstep + 1) %
+					(mp->m_sb.sb_agcount * rotorstep);
+		}
+		break;
+	default:
+		ASSERT(0);
+		/* NOTREACHED */
+	}
+	if (args->agbno == NULLAGBLOCK)
+		args->fsbno = NULLFSBLOCK;
+	else {
+		args->fsbno = XFS_AGB_TO_FSB(mp, args->agno, args->agbno);
+#ifdef DEBUG
+		ASSERT(args->len >= args->minlen);
+		ASSERT(args->len <= args->maxlen);
+		ASSERT(args->agbno % args->alignment == 0);
+		XFS_AG_CHECK_DADDR(mp, XFS_FSB_TO_DADDR(mp, args->fsbno),
+			args->len);
+#endif
+	}
+	xfs_perag_put(args->pag);
+	return 0;
+error0:
+	xfs_perag_put(args->pag);
+	return error;
+}
+
+/*
+ * Free an extent.
+ * Just break up the extent address and hand off to xfs_free_ag_extent
+ * after fixing up the freelist.
+ */
+int				/* error */
+xfs_free_extent(
+	xfs_trans_t	*tp,	/* transaction pointer */
+	xfs_fsblock_t	bno,	/* starting block number of extent */
+	xfs_extlen_t	len)	/* length of extent */
+{
+	xfs_alloc_arg_t	args;
+	int		error;
+
+	ASSERT(len != 0);
+	memset(&args, 0, sizeof(xfs_alloc_arg_t));
+	args.tp = tp;
+	args.mp = tp->t_mountp;
+
+	/*
+	 * validate that the block number is legal - the enables us to detect
+	 * and handle a silent filesystem corruption rather than crashing.
+	 */
+	args.agno = XFS_FSB_TO_AGNO(args.mp, bno);
+	if (args.agno >= args.mp->m_sb.sb_agcount)
+		return EFSCORRUPTED;
+
+	args.agbno = XFS_FSB_TO_AGBNO(args.mp, bno);
+	if (args.agbno >= args.mp->m_sb.sb_agblocks)
+		return EFSCORRUPTED;
+
+	args.pag = xfs_perag_get(args.mp, args.agno);
+	ASSERT(args.pag);
+
+	error = xfs_alloc_fix_freelist(&args, XFS_ALLOC_FLAG_FREEING);
+	if (error)
+		goto error0;
+
+	/* validate the extent size is legal now we have the agf locked */
+	if (args.agbno + len >
+			be32_to_cpu(XFS_BUF_TO_AGF(args.agbp)->agf_length)) {
+		error = EFSCORRUPTED;
+		goto error0;
+	}
+
+	error = xfs_free_ag_extent(tp, args.agbp, args.agno, args.agbno, len, 0);
+	if (!error)
+		xfs_alloc_busy_insert(tp, args.agno, args.agbno, len, 0);
+error0:
+	xfs_perag_put(args.pag);
+	return error;
+}
+
+void
+xfs_alloc_busy_insert(
+	struct xfs_trans	*tp,
+	xfs_agnumber_t		agno,
+	xfs_agblock_t		bno,
+	xfs_extlen_t		len,
+	unsigned int		flags)
+{
+	struct xfs_busy_extent	*new;
+	struct xfs_busy_extent	*busyp;
+	struct xfs_perag	*pag;
+	struct rb_node		**rbp;
+	struct rb_node		*parent = NULL;
+
+	new = kmem_zalloc(sizeof(struct xfs_busy_extent), KM_MAYFAIL);
+	if (!new) {
+		/*
+		 * No Memory!  Since it is now not possible to track the free
+		 * block, make this a synchronous transaction to insure that
+		 * the block is not reused before this transaction commits.
+		 */
+		trace_xfs_alloc_busy_enomem(tp->t_mountp, agno, bno, len);
+		xfs_trans_set_sync(tp);
+		return;
+	}
+
+	new->agno = agno;
+	new->bno = bno;
+	new->length = len;
+	INIT_LIST_HEAD(&new->list);
+	new->flags = flags;
+
+	/* trace before insert to be able to see failed inserts */
+	trace_xfs_alloc_busy(tp->t_mountp, agno, bno, len);
+
+	pag = xfs_perag_get(tp->t_mountp, new->agno);
+	spin_lock(&pag->pagb_lock);
+	rbp = &pag->pagb_tree.rb_node;
+	while (*rbp) {
+		parent = *rbp;
+		busyp = rb_entry(parent, struct xfs_busy_extent, rb_node);
+
+		if (new->bno < busyp->bno) {
+			rbp = &(*rbp)->rb_left;
+			ASSERT(new->bno + new->length <= busyp->bno);
+		} else if (new->bno > busyp->bno) {
+			rbp = &(*rbp)->rb_right;
+			ASSERT(bno >= busyp->bno + busyp->length);
+		} else {
+			ASSERT(0);
+		}
+	}
+
+	rb_link_node(&new->rb_node, parent, rbp);
+	rb_insert_color(&new->rb_node, &pag->pagb_tree);
+
+	list_add(&new->list, &tp->t_busy);
+	spin_unlock(&pag->pagb_lock);
+	xfs_perag_put(pag);
+}
+
+/*
+ * Search for a busy extent within the range of the extent we are about to
+ * allocate.  You need to be holding the busy extent tree lock when calling
+ * xfs_alloc_busy_search(). This function returns 0 for no overlapping busy
+ * extent, -1 for an overlapping but not exact busy extent, and 1 for an exact
+ * match. This is done so that a non-zero return indicates an overlap that
+ * will require a synchronous transaction, but it can still be
+ * used to distinguish between a partial or exact match.
+ */
+int
+xfs_alloc_busy_search(
+	struct xfs_mount	*mp,
+	xfs_agnumber_t		agno,
+	xfs_agblock_t		bno,
+	xfs_extlen_t		len)
+{
+	struct xfs_perag	*pag;
+	struct rb_node		*rbp;
+	struct xfs_busy_extent	*busyp;
+	int			match = 0;
+
+	pag = xfs_perag_get(mp, agno);
+	spin_lock(&pag->pagb_lock);
+
+	rbp = pag->pagb_tree.rb_node;
+
+	/* find closest start bno overlap */
+	while (rbp) {
+		busyp = rb_entry(rbp, struct xfs_busy_extent, rb_node);
+		if (bno < busyp->bno) {
+			/* may overlap, but exact start block is lower */
+			if (bno + len > busyp->bno)
+				match = -1;
+			rbp = rbp->rb_left;
+		} else if (bno > busyp->bno) {
+			/* may overlap, but exact start block is higher */
+			if (bno < busyp->bno + busyp->length)
+				match = -1;
+			rbp = rbp->rb_right;
+		} else {
+			/* bno matches busyp, length determines exact match */
+			match = (busyp->length == len) ? 1 : -1;
+			break;
+		}
+	}
+	spin_unlock(&pag->pagb_lock);
+	xfs_perag_put(pag);
+	return match;
+}
+
+/*
+ * The found free extent [fbno, fend] overlaps part or all of the given busy
+ * extent.  If the overlap covers the beginning, the end, or all of the busy
+ * extent, the overlapping portion can be made unbusy and used for the
+ * allocation.  We can't split a busy extent because we can't modify a
+ * transaction/CIL context busy list, but we can update an entries block
+ * number or length.
+ *
+ * Returns true if the extent can safely be reused, or false if the search
+ * needs to be restarted.
+ */
+STATIC bool
+xfs_alloc_busy_update_extent(
+	struct xfs_mount	*mp,
+	struct xfs_perag	*pag,
+	struct xfs_busy_extent	*busyp,
+	xfs_agblock_t		fbno,
+	xfs_extlen_t		flen,
+	bool			userdata)
+{
+	xfs_agblock_t		fend = fbno + flen;
+	xfs_agblock_t		bbno = busyp->bno;
+	xfs_agblock_t		bend = bbno + busyp->length;
+
+	/*
+	 * This extent is currently being discarded.  Give the thread
+	 * performing the discard a chance to mark the extent unbusy
+	 * and retry.
+	 */
+	if (busyp->flags & XFS_ALLOC_BUSY_DISCARDED) {
+		spin_unlock(&pag->pagb_lock);
+		delay(1);
+		spin_lock(&pag->pagb_lock);
+		return false;
+	}
+
+	/*
+	 * If there is a busy extent overlapping a user allocation, we have
+	 * no choice but to force the log and retry the search.
+	 *
+	 * Fortunately this does not happen during normal operation, but
+	 * only if the filesystem is very low on space and has to dip into
+	 * the AGFL for normal allocations.
+	 */
+	if (userdata)
+		goto out_force_log;
+
+	if (bbno < fbno && bend > fend) {
+		/*
+		 * Case 1:
+		 *    bbno           bend
+		 *    +BBBBBBBBBBBBBBBBB+
+		 *        +---------+
+		 *        fbno   fend
+		 */
+
+		/*
+		 * We would have to split the busy extent to be able to track
+		 * it correct, which we cannot do because we would have to
+		 * modify the list of busy extents attached to the transaction
+		 * or CIL context, which is immutable.
+		 *
+		 * Force out the log to clear the busy extent and retry the
+		 * search.
+		 */
+		goto out_force_log;
+	} else if (bbno >= fbno && bend <= fend) {
+		/*
+		 * Case 2:
+		 *    bbno           bend
+		 *    +BBBBBBBBBBBBBBBBB+
+		 *    +-----------------+
+		 *    fbno           fend
+		 *
+		 * Case 3:
+		 *    bbno           bend
+		 *    +BBBBBBBBBBBBBBBBB+
+		 *    +--------------------------+
+		 *    fbno                    fend
+		 *
+		 * Case 4:
+		 *             bbno           bend
+		 *             +BBBBBBBBBBBBBBBBB+
+		 *    +--------------------------+
+		 *    fbno                    fend
+		 *
+		 * Case 5:
+		 *             bbno           bend
+		 *             +BBBBBBBBBBBBBBBBB+
+		 *    +-----------------------------------+
+		 *    fbno                             fend
+		 *
+		 */
+
+		/*
+		 * The busy extent is fully covered by the extent we are
+		 * allocating, and can simply be removed from the rbtree.
+		 * However we cannot remove it from the immutable list
+		 * tracking busy extents in the transaction or CIL context,
+		 * so set the length to zero to mark it invalid.
+		 *
+		 * We also need to restart the busy extent search from the
+		 * tree root, because erasing the node can rearrange the
+		 * tree topology.
+		 */
+		rb_erase(&busyp->rb_node, &pag->pagb_tree);
+		busyp->length = 0;
+		return false;
+	} else if (fend < bend) {
+		/*
+		 * Case 6:
+		 *              bbno           bend
+		 *             +BBBBBBBBBBBBBBBBB+
+		 *             +---------+
+		 *             fbno   fend
+		 *
+		 * Case 7:
+		 *             bbno           bend
+		 *             +BBBBBBBBBBBBBBBBB+
+		 *    +------------------+
+		 *    fbno            fend
+		 *
+		 */
+		busyp->bno = fend;
+	} else if (bbno < fbno) {
+		/*
+		 * Case 8:
+		 *    bbno           bend
+		 *    +BBBBBBBBBBBBBBBBB+
+		 *        +-------------+
+		 *        fbno       fend
+		 *
+		 * Case 9:
+		 *    bbno           bend
+		 *    +BBBBBBBBBBBBBBBBB+
+		 *        +----------------------+
+		 *        fbno                fend
+		 */
+		busyp->length = fbno - busyp->bno;
+	} else {
+		ASSERT(0);
+	}
+
+	trace_xfs_alloc_busy_reuse(mp, pag->pag_agno, fbno, flen);
+	return true;
+
+out_force_log:
+	spin_unlock(&pag->pagb_lock);
+	xfs_log_force(mp, XFS_LOG_SYNC);
+	trace_xfs_alloc_busy_force(mp, pag->pag_agno, fbno, flen);
+	spin_lock(&pag->pagb_lock);
+	return false;
+}
+
+
+/*
+ * For a given extent [fbno, flen], make sure we can reuse it safely.
+ */
+void
+xfs_alloc_busy_reuse(
+	struct xfs_mount	*mp,
+	xfs_agnumber_t		agno,
+	xfs_agblock_t		fbno,
+	xfs_extlen_t		flen,
+	bool			userdata)
+{
+	struct xfs_perag	*pag;
+	struct rb_node		*rbp;
+
+	ASSERT(flen > 0);
+
+	pag = xfs_perag_get(mp, agno);
+	spin_lock(&pag->pagb_lock);
+restart:
+	rbp = pag->pagb_tree.rb_node;
+	while (rbp) {
+		struct xfs_busy_extent *busyp =
+			rb_entry(rbp, struct xfs_busy_extent, rb_node);
+		xfs_agblock_t	bbno = busyp->bno;
+		xfs_agblock_t	bend = bbno + busyp->length;
+
+		if (fbno + flen <= bbno) {
+			rbp = rbp->rb_left;
+			continue;
+		} else if (fbno >= bend) {
+			rbp = rbp->rb_right;
+			continue;
+		}
+
+		if (!xfs_alloc_busy_update_extent(mp, pag, busyp, fbno, flen,
+						  userdata))
+			goto restart;
+	}
+	spin_unlock(&pag->pagb_lock);
+	xfs_perag_put(pag);
+}
+
+/*
+ * For a given extent [fbno, flen], search the busy extent list to find a
+ * subset of the extent that is not busy.  If *rlen is smaller than
+ * args->minlen no suitable extent could be found, and the higher level
+ * code needs to force out the log and retry the allocation.
+ */
+STATIC void
+xfs_alloc_busy_trim(
+	struct xfs_alloc_arg	*args,
+	xfs_agblock_t		bno,
+	xfs_extlen_t		len,
+	xfs_agblock_t		*rbno,
+	xfs_extlen_t		*rlen)
+{
+	xfs_agblock_t		fbno;
+	xfs_extlen_t		flen;
+	struct rb_node		*rbp;
+
+	ASSERT(len > 0);
+
+	spin_lock(&args->pag->pagb_lock);
+restart:
+	fbno = bno;
+	flen = len;
+	rbp = args->pag->pagb_tree.rb_node;
+	while (rbp && flen >= args->minlen) {
+		struct xfs_busy_extent *busyp =
+			rb_entry(rbp, struct xfs_busy_extent, rb_node);
+		xfs_agblock_t	fend = fbno + flen;
+		xfs_agblock_t	bbno = busyp->bno;
+		xfs_agblock_t	bend = bbno + busyp->length;
+
+		if (fend <= bbno) {
+			rbp = rbp->rb_left;
+			continue;
+		} else if (fbno >= bend) {
+			rbp = rbp->rb_right;
+			continue;
+		}
+
+		/*
+		 * If this is a metadata allocation, try to reuse the busy
+		 * extent instead of trimming the allocation.
+		 */
+		if (!args->userdata &&
+		    !(busyp->flags & XFS_ALLOC_BUSY_DISCARDED)) {
+			if (!xfs_alloc_busy_update_extent(args->mp, args->pag,
+							  busyp, fbno, flen,
+							  false))
+				goto restart;
+			continue;
+		}
+
+		if (bbno <= fbno) {
+			/* start overlap */
+
+			/*
+			 * Case 1:
+			 *    bbno           bend
+			 *    +BBBBBBBBBBBBBBBBB+
+			 *        +---------+
+			 *        fbno   fend
+			 *
+			 * Case 2:
+			 *    bbno           bend
+			 *    +BBBBBBBBBBBBBBBBB+
+			 *    +-------------+
+			 *    fbno       fend
+			 *
+			 * Case 3:
+			 *    bbno           bend
+			 *    +BBBBBBBBBBBBBBBBB+
+			 *        +-------------+
+			 *        fbno       fend
+			 *
+			 * Case 4:
+			 *    bbno           bend
+			 *    +BBBBBBBBBBBBBBBBB+
+			 *    +-----------------+
+			 *    fbno           fend
+			 *
+			 * No unbusy region in extent, return failure.
+			 */
+			if (fend <= bend)
+				goto fail;
+
+			/*
+			 * Case 5:
+			 *    bbno           bend
+			 *    +BBBBBBBBBBBBBBBBB+
+			 *        +----------------------+
+			 *        fbno                fend
+			 *
+			 * Case 6:
+			 *    bbno           bend
+			 *    +BBBBBBBBBBBBBBBBB+
+			 *    +--------------------------+
+			 *    fbno                    fend
+			 *
+			 * Needs to be trimmed to:
+			 *                       +-------+
+			 *                       fbno fend
+			 */
+			fbno = bend;
+		} else if (bend >= fend) {
+			/* end overlap */
+
+			/*
+			 * Case 7:
+			 *             bbno           bend
+			 *             +BBBBBBBBBBBBBBBBB+
+			 *    +------------------+
+			 *    fbno            fend
+			 *
+			 * Case 8:
+			 *             bbno           bend
+			 *             +BBBBBBBBBBBBBBBBB+
+			 *    +--------------------------+
+			 *    fbno                    fend
+			 *
+			 * Needs to be trimmed to:
+			 *    +-------+
+			 *    fbno fend
+			 */
+			fend = bbno;
+		} else {
+			/* middle overlap */
+
+			/*
+			 * Case 9:
+			 *             bbno           bend
+			 *             +BBBBBBBBBBBBBBBBB+
+			 *    +-----------------------------------+
+			 *    fbno                             fend
+			 *
+			 * Can be trimmed to:
+			 *    +-------+        OR         +-------+
+			 *    fbno fend                   fbno fend
+			 *
+			 * Backward allocation leads to significant
+			 * fragmentation of directories, which degrades
+			 * directory performance, therefore we always want to
+			 * choose the option that produces forward allocation
+			 * patterns.
+			 * Preferring the lower bno extent will make the next
+			 * request use "fend" as the start of the next
+			 * allocation;  if the segment is no longer busy at
+			 * that point, we'll get a contiguous allocation, but
+			 * even if it is still busy, we will get a forward
+			 * allocation.
+			 * We try to avoid choosing the segment at "bend",
+			 * because that can lead to the next allocation
+			 * taking the segment at "fbno", which would be a
+			 * backward allocation.  We only use the segment at
+			 * "fbno" if it is much larger than the current
+			 * requested size, because in that case there's a
+			 * good chance subsequent allocations will be
+			 * contiguous.
+			 */
+			if (bbno - fbno >= args->maxlen) {
+				/* left candidate fits perfect */
+				fend = bbno;
+			} else if (fend - bend >= args->maxlen * 4) {
+				/* right candidate has enough free space */
+				fbno = bend;
+			} else if (bbno - fbno >= args->minlen) {
+				/* left candidate fits minimum requirement */
+				fend = bbno;
+			} else {
+				goto fail;
+			}
+		}
+
+		flen = fend - fbno;
+	}
+	spin_unlock(&args->pag->pagb_lock);
+
+	if (fbno != bno || flen != len) {
+		trace_xfs_alloc_busy_trim(args->mp, args->agno, bno, len,
+					  fbno, flen);
+	}
+	*rbno = fbno;
+	*rlen = flen;
+	return;
+fail:
+	/*
+	 * Return a zero extent length as failure indications.  All callers
+	 * re-check if the trimmed extent satisfies the minlen requirement.
+	 */
+	spin_unlock(&args->pag->pagb_lock);
+	trace_xfs_alloc_busy_trim(args->mp, args->agno, bno, len, fbno, 0);
+	*rbno = fbno;
+	*rlen = 0;
+}
+
+static void
+xfs_alloc_busy_clear_one(
+	struct xfs_mount	*mp,
+	struct xfs_perag	*pag,
+	struct xfs_busy_extent	*busyp)
+{
+	if (busyp->length) {
+		trace_xfs_alloc_busy_clear(mp, busyp->agno, busyp->bno,
+						busyp->length);
+		rb_erase(&busyp->rb_node, &pag->pagb_tree);
+	}
+
+	list_del_init(&busyp->list);
+	kmem_free(busyp);
+}
+
+/*
+ * Remove all extents on the passed in list from the busy extents tree.
+ * If do_discard is set skip extents that need to be discarded, and mark
+ * these as undergoing a discard operation instead.
+ */
+void
+xfs_alloc_busy_clear(
+	struct xfs_mount	*mp,
+	struct list_head	*list,
+	bool			do_discard)
+{
+	struct xfs_busy_extent	*busyp, *n;
+	struct xfs_perag	*pag = NULL;
+	xfs_agnumber_t		agno = NULLAGNUMBER;
+
+	list_for_each_entry_safe(busyp, n, list, list) {
+		if (busyp->agno != agno) {
+			if (pag) {
+				spin_unlock(&pag->pagb_lock);
+				xfs_perag_put(pag);
+			}
+			pag = xfs_perag_get(mp, busyp->agno);
+			spin_lock(&pag->pagb_lock);
+			agno = busyp->agno;
+		}
+
+		if (do_discard && busyp->length &&
+		    !(busyp->flags & XFS_ALLOC_BUSY_SKIP_DISCARD))
+			busyp->flags = XFS_ALLOC_BUSY_DISCARDED;
+		else
+			xfs_alloc_busy_clear_one(mp, pag, busyp);
+	}
+
+	if (pag) {
+		spin_unlock(&pag->pagb_lock);
+		xfs_perag_put(pag);
+	}
+}
+
+/*
+ * Callback for list_sort to sort busy extents by the AG they reside in.
+ */
+int
+xfs_busy_extent_ag_cmp(
+	void			*priv,
+	struct list_head	*a,
+	struct list_head	*b)
+{
+	return container_of(a, struct xfs_busy_extent, list)->agno -
+		container_of(b, struct xfs_busy_extent, list)->agno;
+}
diff --git a/fs/xfs/xfs_alloc.h b/fs/xfs/xfs_alloc.h
new file mode 100644
index 00000000..2f52b924
--- /dev/null
+++ b/fs/xfs/xfs_alloc.h
@@ -0,0 +1,253 @@
+/*
+ * Copyright (c) 2000-2002,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_ALLOC_H__
+#define	__XFS_ALLOC_H__
+
+struct xfs_buf;
+struct xfs_btree_cur;
+struct xfs_mount;
+struct xfs_perag;
+struct xfs_trans;
+struct xfs_busy_extent;
+
+/*
+ * Freespace allocation types.  Argument to xfs_alloc_[v]extent.
+ */
+#define XFS_ALLOCTYPE_ANY_AG	0x01	/* allocate anywhere, use rotor */
+#define XFS_ALLOCTYPE_FIRST_AG	0x02	/* ... start at ag 0 */
+#define XFS_ALLOCTYPE_START_AG	0x04	/* anywhere, start in this a.g. */
+#define XFS_ALLOCTYPE_THIS_AG	0x08	/* anywhere in this a.g. */
+#define XFS_ALLOCTYPE_START_BNO	0x10	/* near this block else anywhere */
+#define XFS_ALLOCTYPE_NEAR_BNO	0x20	/* in this a.g. and near this block */
+#define XFS_ALLOCTYPE_THIS_BNO	0x40	/* at exactly this block */
+
+/* this should become an enum again when the tracing code is fixed */
+typedef unsigned int xfs_alloctype_t;
+
+#define XFS_ALLOC_TYPES \
+	{ XFS_ALLOCTYPE_ANY_AG,		"ANY_AG" }, \
+	{ XFS_ALLOCTYPE_FIRST_AG,	"FIRST_AG" }, \
+	{ XFS_ALLOCTYPE_START_AG,	"START_AG" }, \
+	{ XFS_ALLOCTYPE_THIS_AG,	"THIS_AG" }, \
+	{ XFS_ALLOCTYPE_START_BNO,	"START_BNO" }, \
+	{ XFS_ALLOCTYPE_NEAR_BNO,	"NEAR_BNO" }, \
+	{ XFS_ALLOCTYPE_THIS_BNO,	"THIS_BNO" }
+
+/*
+ * Flags for xfs_alloc_fix_freelist.
+ */
+#define	XFS_ALLOC_FLAG_TRYLOCK	0x00000001  /* use trylock for buffer locking */
+#define	XFS_ALLOC_FLAG_FREEING	0x00000002  /* indicate caller is freeing extents*/
+
+/*
+ * In order to avoid ENOSPC-related deadlock caused by
+ * out-of-order locking of AGF buffer (PV 947395), we place
+ * constraints on the relationship among actual allocations for
+ * data blocks, freelist blocks, and potential file data bmap
+ * btree blocks. However, these restrictions may result in no
+ * actual space allocated for a delayed extent, for example, a data
+ * block in a certain AG is allocated but there is no additional
+ * block for the additional bmap btree block due to a split of the
+ * bmap btree of the file. The result of this may lead to an
+ * infinite loop in xfssyncd when the file gets flushed to disk and
+ * all delayed extents need to be actually allocated. To get around
+ * this, we explicitly set aside a few blocks which will not be
+ * reserved in delayed allocation. Considering the minimum number of
+ * needed freelist blocks is 4 fsbs _per AG_, a potential split of file's bmap
+ * btree requires 1 fsb, so we set the number of set-aside blocks
+ * to 4 + 4*agcount.
+ */
+#define XFS_ALLOC_SET_ASIDE(mp)  (4 + ((mp)->m_sb.sb_agcount * 4))
+
+/*
+ * When deciding how much space to allocate out of an AG, we limit the
+ * allocation maximum size to the size the AG. However, we cannot use all the
+ * blocks in the AG - some are permanently used by metadata. These
+ * blocks are generally:
+ *	- the AG superblock, AGF, AGI and AGFL
+ *	- the AGF (bno and cnt) and AGI btree root blocks
+ *	- 4 blocks on the AGFL according to XFS_ALLOC_SET_ASIDE() limits
+ *
+ * The AG headers are sector sized, so the amount of space they take up is
+ * dependent on filesystem geometry. The others are all single blocks.
+ */
+#define XFS_ALLOC_AG_MAX_USABLE(mp)	\
+	((mp)->m_sb.sb_agblocks - XFS_BB_TO_FSB(mp, XFS_FSS_TO_BB(mp, 4)) - 7)
+
+
+/*
+ * Argument structure for xfs_alloc routines.
+ * This is turned into a structure to avoid having 20 arguments passed
+ * down several levels of the stack.
+ */
+typedef struct xfs_alloc_arg {
+	struct xfs_trans *tp;		/* transaction pointer */
+	struct xfs_mount *mp;		/* file system mount point */
+	struct xfs_buf	*agbp;		/* buffer for a.g. freelist header */
+	struct xfs_perag *pag;		/* per-ag struct for this agno */
+	xfs_fsblock_t	fsbno;		/* file system block number */
+	xfs_agnumber_t	agno;		/* allocation group number */
+	xfs_agblock_t	agbno;		/* allocation group-relative block # */
+	xfs_extlen_t	minlen;		/* minimum size of extent */
+	xfs_extlen_t	maxlen;		/* maximum size of extent */
+	xfs_extlen_t	mod;		/* mod value for extent size */
+	xfs_extlen_t	prod;		/* prod value for extent size */
+	xfs_extlen_t	minleft;	/* min blocks must be left after us */
+	xfs_extlen_t	total;		/* total blocks needed in xaction */
+	xfs_extlen_t	alignment;	/* align answer to multiple of this */
+	xfs_extlen_t	minalignslop;	/* slop for minlen+alignment calcs */
+	xfs_extlen_t	len;		/* output: actual size of extent */
+	xfs_alloctype_t	type;		/* allocation type XFS_ALLOCTYPE_... */
+	xfs_alloctype_t	otype;		/* original allocation type */
+	char		wasdel;		/* set if allocation was prev delayed */
+	char		wasfromfl;	/* set if allocation is from freelist */
+	char		isfl;		/* set if is freelist blocks - !acctg */
+	char		userdata;	/* set if this is user data */
+	xfs_fsblock_t	firstblock;	/* io first block allocated */
+} xfs_alloc_arg_t;
+
+/*
+ * Defines for userdata
+ */
+#define XFS_ALLOC_USERDATA		1	/* allocation is for user data*/
+#define XFS_ALLOC_INITIAL_USER_DATA	2	/* special case start of file */
+
+/*
+ * Find the length of the longest extent in an AG.
+ */
+xfs_extlen_t
+xfs_alloc_longest_free_extent(struct xfs_mount *mp,
+		struct xfs_perag *pag);
+
+#ifdef __KERNEL__
+void
+xfs_alloc_busy_insert(struct xfs_trans *tp, xfs_agnumber_t agno,
+	xfs_agblock_t bno, xfs_extlen_t len, unsigned int flags);
+
+void
+xfs_alloc_busy_clear(struct xfs_mount *mp, struct list_head *list,
+	bool do_discard);
+
+int
+xfs_alloc_busy_search(struct xfs_mount *mp, xfs_agnumber_t agno,
+	xfs_agblock_t bno, xfs_extlen_t len);
+
+void
+xfs_alloc_busy_reuse(struct xfs_mount *mp, xfs_agnumber_t agno,
+	xfs_agblock_t fbno, xfs_extlen_t flen, bool userdata);
+
+int
+xfs_busy_extent_ag_cmp(void *priv, struct list_head *a, struct list_head *b);
+
+static inline void xfs_alloc_busy_sort(struct list_head *list)
+{
+	list_sort(NULL, list, xfs_busy_extent_ag_cmp);
+}
+
+#endif	/* __KERNEL__ */
+
+/*
+ * Compute and fill in value of m_ag_maxlevels.
+ */
+void
+xfs_alloc_compute_maxlevels(
+	struct xfs_mount	*mp);	/* file system mount structure */
+
+/*
+ * Get a block from the freelist.
+ * Returns with the buffer for the block gotten.
+ */
+int				/* error */
+xfs_alloc_get_freelist(
+	struct xfs_trans *tp,	/* transaction pointer */
+	struct xfs_buf	*agbp,	/* buffer containing the agf structure */
+	xfs_agblock_t	*bnop,	/* block address retrieved from freelist */
+	int		btreeblk); /* destination is a AGF btree */
+
+/*
+ * Log the given fields from the agf structure.
+ */
+void
+xfs_alloc_log_agf(
+	struct xfs_trans *tp,	/* transaction pointer */
+	struct xfs_buf	*bp,	/* buffer for a.g. freelist header */
+	int		fields);/* mask of fields to be logged (XFS_AGF_...) */
+
+/*
+ * Interface for inode allocation to force the pag data to be initialized.
+ */
+int				/* error */
+xfs_alloc_pagf_init(
+	struct xfs_mount *mp,	/* file system mount structure */
+	struct xfs_trans *tp,	/* transaction pointer */
+	xfs_agnumber_t	agno,	/* allocation group number */
+	int		flags);	/* XFS_ALLOC_FLAGS_... */
+
+/*
+ * Put the block on the freelist for the allocation group.
+ */
+int				/* error */
+xfs_alloc_put_freelist(
+	struct xfs_trans *tp,	/* transaction pointer */
+	struct xfs_buf	*agbp,	/* buffer for a.g. freelist header */
+	struct xfs_buf	*agflbp,/* buffer for a.g. free block array */
+	xfs_agblock_t	bno,	/* block being freed */
+	int		btreeblk); /* owner was a AGF btree */
+
+/*
+ * Read in the allocation group header (free/alloc section).
+ */
+int					/* error  */
+xfs_alloc_read_agf(
+	struct xfs_mount *mp,		/* mount point structure */
+	struct xfs_trans *tp,		/* transaction pointer */
+	xfs_agnumber_t	agno,		/* allocation group number */
+	int		flags,		/* XFS_ALLOC_FLAG_... */
+	struct xfs_buf	**bpp);		/* buffer for the ag freelist header */
+
+/*
+ * Allocate an extent (variable-size).
+ */
+int				/* error */
+xfs_alloc_vextent(
+	xfs_alloc_arg_t	*args);	/* allocation argument structure */
+
+/*
+ * Free an extent.
+ */
+int				/* error */
+xfs_free_extent(
+	struct xfs_trans *tp,	/* transaction pointer */
+	xfs_fsblock_t	bno,	/* starting block number of extent */
+	xfs_extlen_t	len);	/* length of extent */
+
+int					/* error */
+xfs_alloc_lookup_le(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_agblock_t		bno,	/* starting block of extent */
+	xfs_extlen_t		len,	/* length of extent */
+	int			*stat);	/* success/failure */
+
+int					/* error */
+xfs_alloc_get_rec(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_agblock_t		*bno,	/* output: starting block of extent */
+	xfs_extlen_t		*len,	/* output: length of extent */
+	int			*stat);	/* output: success/failure */
+
+#endif	/* __XFS_ALLOC_H__ */
diff --git a/fs/xfs/xfs_alloc_btree.c b/fs/xfs/xfs_alloc_btree.c
new file mode 100644
index 00000000..2b351882
--- /dev/null
+++ b/fs/xfs/xfs_alloc_btree.c
@@ -0,0 +1,458 @@
+/*
+ * Copyright (c) 2000-2001,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_mount.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_alloc_btree.h"
+#include "xfs_ialloc_btree.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_btree.h"
+#include "xfs_btree_trace.h"
+#include "xfs_alloc.h"
+#include "xfs_error.h"
+#include "xfs_trace.h"
+
+
+STATIC struct xfs_btree_cur *
+xfs_allocbt_dup_cursor(
+	struct xfs_btree_cur	*cur)
+{
+	return xfs_allocbt_init_cursor(cur->bc_mp, cur->bc_tp,
+			cur->bc_private.a.agbp, cur->bc_private.a.agno,
+			cur->bc_btnum);
+}
+
+STATIC void
+xfs_allocbt_set_root(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*ptr,
+	int			inc)
+{
+	struct xfs_buf		*agbp = cur->bc_private.a.agbp;
+	struct xfs_agf		*agf = XFS_BUF_TO_AGF(agbp);
+	xfs_agnumber_t		seqno = be32_to_cpu(agf->agf_seqno);
+	int			btnum = cur->bc_btnum;
+	struct xfs_perag	*pag = xfs_perag_get(cur->bc_mp, seqno);
+
+	ASSERT(ptr->s != 0);
+
+	agf->agf_roots[btnum] = ptr->s;
+	be32_add_cpu(&agf->agf_levels[btnum], inc);
+	pag->pagf_levels[btnum] += inc;
+	xfs_perag_put(pag);
+
+	xfs_alloc_log_agf(cur->bc_tp, agbp, XFS_AGF_ROOTS | XFS_AGF_LEVELS);
+}
+
+STATIC int
+xfs_allocbt_alloc_block(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*start,
+	union xfs_btree_ptr	*new,
+	int			length,
+	int			*stat)
+{
+	int			error;
+	xfs_agblock_t		bno;
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+
+	/* Allocate the new block from the freelist. If we can't, give up.  */
+	error = xfs_alloc_get_freelist(cur->bc_tp, cur->bc_private.a.agbp,
+				       &bno, 1);
+	if (error) {
+		XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
+		return error;
+	}
+
+	if (bno == NULLAGBLOCK) {
+		XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+		*stat = 0;
+		return 0;
+	}
+
+	xfs_alloc_busy_reuse(cur->bc_mp, cur->bc_private.a.agno, bno, 1, false);
+
+	xfs_trans_agbtree_delta(cur->bc_tp, 1);
+	new->s = cpu_to_be32(bno);
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	*stat = 1;
+	return 0;
+}
+
+STATIC int
+xfs_allocbt_free_block(
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*bp)
+{
+	struct xfs_buf		*agbp = cur->bc_private.a.agbp;
+	struct xfs_agf		*agf = XFS_BUF_TO_AGF(agbp);
+	xfs_agblock_t		bno;
+	int			error;
+
+	bno = xfs_daddr_to_agbno(cur->bc_mp, XFS_BUF_ADDR(bp));
+	error = xfs_alloc_put_freelist(cur->bc_tp, agbp, NULL, bno, 1);
+	if (error)
+		return error;
+
+	xfs_alloc_busy_insert(cur->bc_tp, be32_to_cpu(agf->agf_seqno), bno, 1,
+			      XFS_ALLOC_BUSY_SKIP_DISCARD);
+	xfs_trans_agbtree_delta(cur->bc_tp, -1);
+	return 0;
+}
+
+/*
+ * Update the longest extent in the AGF
+ */
+STATIC void
+xfs_allocbt_update_lastrec(
+	struct xfs_btree_cur	*cur,
+	struct xfs_btree_block	*block,
+	union xfs_btree_rec	*rec,
+	int			ptr,
+	int			reason)
+{
+	struct xfs_agf		*agf = XFS_BUF_TO_AGF(cur->bc_private.a.agbp);
+	xfs_agnumber_t		seqno = be32_to_cpu(agf->agf_seqno);
+	struct xfs_perag	*pag;
+	__be32			len;
+	int			numrecs;
+
+	ASSERT(cur->bc_btnum == XFS_BTNUM_CNT);
+
+	switch (reason) {
+	case LASTREC_UPDATE:
+		/*
+		 * If this is the last leaf block and it's the last record,
+		 * then update the size of the longest extent in the AG.
+		 */
+		if (ptr != xfs_btree_get_numrecs(block))
+			return;
+		len = rec->alloc.ar_blockcount;
+		break;
+	case LASTREC_INSREC:
+		if (be32_to_cpu(rec->alloc.ar_blockcount) <=
+		    be32_to_cpu(agf->agf_longest))
+			return;
+		len = rec->alloc.ar_blockcount;
+		break;
+	case LASTREC_DELREC:
+		numrecs = xfs_btree_get_numrecs(block);
+		if (ptr <= numrecs)
+			return;
+		ASSERT(ptr == numrecs + 1);
+
+		if (numrecs) {
+			xfs_alloc_rec_t *rrp;
+
+			rrp = XFS_ALLOC_REC_ADDR(cur->bc_mp, block, numrecs);
+			len = rrp->ar_blockcount;
+		} else {
+			len = 0;
+		}
+
+		break;
+	default:
+		ASSERT(0);
+		return;
+	}
+
+	agf->agf_longest = len;
+	pag = xfs_perag_get(cur->bc_mp, seqno);
+	pag->pagf_longest = be32_to_cpu(len);
+	xfs_perag_put(pag);
+	xfs_alloc_log_agf(cur->bc_tp, cur->bc_private.a.agbp, XFS_AGF_LONGEST);
+}
+
+STATIC int
+xfs_allocbt_get_minrecs(
+	struct xfs_btree_cur	*cur,
+	int			level)
+{
+	return cur->bc_mp->m_alloc_mnr[level != 0];
+}
+
+STATIC int
+xfs_allocbt_get_maxrecs(
+	struct xfs_btree_cur	*cur,
+	int			level)
+{
+	return cur->bc_mp->m_alloc_mxr[level != 0];
+}
+
+STATIC void
+xfs_allocbt_init_key_from_rec(
+	union xfs_btree_key	*key,
+	union xfs_btree_rec	*rec)
+{
+	ASSERT(rec->alloc.ar_startblock != 0);
+
+	key->alloc.ar_startblock = rec->alloc.ar_startblock;
+	key->alloc.ar_blockcount = rec->alloc.ar_blockcount;
+}
+
+STATIC void
+xfs_allocbt_init_rec_from_key(
+	union xfs_btree_key	*key,
+	union xfs_btree_rec	*rec)
+{
+	ASSERT(key->alloc.ar_startblock != 0);
+
+	rec->alloc.ar_startblock = key->alloc.ar_startblock;
+	rec->alloc.ar_blockcount = key->alloc.ar_blockcount;
+}
+
+STATIC void
+xfs_allocbt_init_rec_from_cur(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_rec	*rec)
+{
+	ASSERT(cur->bc_rec.a.ar_startblock != 0);
+
+	rec->alloc.ar_startblock = cpu_to_be32(cur->bc_rec.a.ar_startblock);
+	rec->alloc.ar_blockcount = cpu_to_be32(cur->bc_rec.a.ar_blockcount);
+}
+
+STATIC void
+xfs_allocbt_init_ptr_from_cur(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*ptr)
+{
+	struct xfs_agf		*agf = XFS_BUF_TO_AGF(cur->bc_private.a.agbp);
+
+	ASSERT(cur->bc_private.a.agno == be32_to_cpu(agf->agf_seqno));
+	ASSERT(agf->agf_roots[cur->bc_btnum] != 0);
+
+	ptr->s = agf->agf_roots[cur->bc_btnum];
+}
+
+STATIC __int64_t
+xfs_allocbt_key_diff(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_key	*key)
+{
+	xfs_alloc_rec_incore_t	*rec = &cur->bc_rec.a;
+	xfs_alloc_key_t		*kp = &key->alloc;
+	__int64_t		diff;
+
+	if (cur->bc_btnum == XFS_BTNUM_BNO) {
+		return (__int64_t)be32_to_cpu(kp->ar_startblock) -
+				rec->ar_startblock;
+	}
+
+	diff = (__int64_t)be32_to_cpu(kp->ar_blockcount) - rec->ar_blockcount;
+	if (diff)
+		return diff;
+
+	return (__int64_t)be32_to_cpu(kp->ar_startblock) - rec->ar_startblock;
+}
+
+#ifdef DEBUG
+STATIC int
+xfs_allocbt_keys_inorder(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_key	*k1,
+	union xfs_btree_key	*k2)
+{
+	if (cur->bc_btnum == XFS_BTNUM_BNO) {
+		return be32_to_cpu(k1->alloc.ar_startblock) <
+		       be32_to_cpu(k2->alloc.ar_startblock);
+	} else {
+		return be32_to_cpu(k1->alloc.ar_blockcount) <
+			be32_to_cpu(k2->alloc.ar_blockcount) ||
+			(k1->alloc.ar_blockcount == k2->alloc.ar_blockcount &&
+			 be32_to_cpu(k1->alloc.ar_startblock) <
+			 be32_to_cpu(k2->alloc.ar_startblock));
+	}
+}
+
+STATIC int
+xfs_allocbt_recs_inorder(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_rec	*r1,
+	union xfs_btree_rec	*r2)
+{
+	if (cur->bc_btnum == XFS_BTNUM_BNO) {
+		return be32_to_cpu(r1->alloc.ar_startblock) +
+			be32_to_cpu(r1->alloc.ar_blockcount) <=
+			be32_to_cpu(r2->alloc.ar_startblock);
+	} else {
+		return be32_to_cpu(r1->alloc.ar_blockcount) <
+			be32_to_cpu(r2->alloc.ar_blockcount) ||
+			(r1->alloc.ar_blockcount == r2->alloc.ar_blockcount &&
+			 be32_to_cpu(r1->alloc.ar_startblock) <
+			 be32_to_cpu(r2->alloc.ar_startblock));
+	}
+}
+#endif	/* DEBUG */
+
+#ifdef XFS_BTREE_TRACE
+ktrace_t	*xfs_allocbt_trace_buf;
+
+STATIC void
+xfs_allocbt_trace_enter(
+	struct xfs_btree_cur	*cur,
+	const char		*func,
+	char			*s,
+	int			type,
+	int			line,
+	__psunsigned_t		a0,
+	__psunsigned_t		a1,
+	__psunsigned_t		a2,
+	__psunsigned_t		a3,
+	__psunsigned_t		a4,
+	__psunsigned_t		a5,
+	__psunsigned_t		a6,
+	__psunsigned_t		a7,
+	__psunsigned_t		a8,
+	__psunsigned_t		a9,
+	__psunsigned_t		a10)
+{
+	ktrace_enter(xfs_allocbt_trace_buf, (void *)(__psint_t)type,
+		(void *)func, (void *)s, NULL, (void *)cur,
+		(void *)a0, (void *)a1, (void *)a2, (void *)a3,
+		(void *)a4, (void *)a5, (void *)a6, (void *)a7,
+		(void *)a8, (void *)a9, (void *)a10);
+}
+
+STATIC void
+xfs_allocbt_trace_cursor(
+	struct xfs_btree_cur	*cur,
+	__uint32_t		*s0,
+	__uint64_t		*l0,
+	__uint64_t		*l1)
+{
+	*s0 = cur->bc_private.a.agno;
+	*l0 = cur->bc_rec.a.ar_startblock;
+	*l1 = cur->bc_rec.a.ar_blockcount;
+}
+
+STATIC void
+xfs_allocbt_trace_key(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_key	*key,
+	__uint64_t		*l0,
+	__uint64_t		*l1)
+{
+	*l0 = be32_to_cpu(key->alloc.ar_startblock);
+	*l1 = be32_to_cpu(key->alloc.ar_blockcount);
+}
+
+STATIC void
+xfs_allocbt_trace_record(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_rec	*rec,
+	__uint64_t		*l0,
+	__uint64_t		*l1,
+	__uint64_t		*l2)
+{
+	*l0 = be32_to_cpu(rec->alloc.ar_startblock);
+	*l1 = be32_to_cpu(rec->alloc.ar_blockcount);
+	*l2 = 0;
+}
+#endif /* XFS_BTREE_TRACE */
+
+static const struct xfs_btree_ops xfs_allocbt_ops = {
+	.rec_len		= sizeof(xfs_alloc_rec_t),
+	.key_len		= sizeof(xfs_alloc_key_t),
+
+	.dup_cursor		= xfs_allocbt_dup_cursor,
+	.set_root		= xfs_allocbt_set_root,
+	.alloc_block		= xfs_allocbt_alloc_block,
+	.free_block		= xfs_allocbt_free_block,
+	.update_lastrec		= xfs_allocbt_update_lastrec,
+	.get_minrecs		= xfs_allocbt_get_minrecs,
+	.get_maxrecs		= xfs_allocbt_get_maxrecs,
+	.init_key_from_rec	= xfs_allocbt_init_key_from_rec,
+	.init_rec_from_key	= xfs_allocbt_init_rec_from_key,
+	.init_rec_from_cur	= xfs_allocbt_init_rec_from_cur,
+	.init_ptr_from_cur	= xfs_allocbt_init_ptr_from_cur,
+	.key_diff		= xfs_allocbt_key_diff,
+
+#ifdef DEBUG
+	.keys_inorder		= xfs_allocbt_keys_inorder,
+	.recs_inorder		= xfs_allocbt_recs_inorder,
+#endif
+
+#ifdef XFS_BTREE_TRACE
+	.trace_enter		= xfs_allocbt_trace_enter,
+	.trace_cursor		= xfs_allocbt_trace_cursor,
+	.trace_key		= xfs_allocbt_trace_key,
+	.trace_record		= xfs_allocbt_trace_record,
+#endif
+};
+
+/*
+ * Allocate a new allocation btree cursor.
+ */
+struct xfs_btree_cur *			/* new alloc btree cursor */
+xfs_allocbt_init_cursor(
+	struct xfs_mount	*mp,		/* file system mount point */
+	struct xfs_trans	*tp,		/* transaction pointer */
+	struct xfs_buf		*agbp,		/* buffer for agf structure */
+	xfs_agnumber_t		agno,		/* allocation group number */
+	xfs_btnum_t		btnum)		/* btree identifier */
+{
+	struct xfs_agf		*agf = XFS_BUF_TO_AGF(agbp);
+	struct xfs_btree_cur	*cur;
+
+	ASSERT(btnum == XFS_BTNUM_BNO || btnum == XFS_BTNUM_CNT);
+
+	cur = kmem_zone_zalloc(xfs_btree_cur_zone, KM_SLEEP);
+
+	cur->bc_tp = tp;
+	cur->bc_mp = mp;
+	cur->bc_nlevels = be32_to_cpu(agf->agf_levels[btnum]);
+	cur->bc_btnum = btnum;
+	cur->bc_blocklog = mp->m_sb.sb_blocklog;
+
+	cur->bc_ops = &xfs_allocbt_ops;
+	if (btnum == XFS_BTNUM_CNT)
+		cur->bc_flags = XFS_BTREE_LASTREC_UPDATE;
+
+	cur->bc_private.a.agbp = agbp;
+	cur->bc_private.a.agno = agno;
+
+	return cur;
+}
+
+/*
+ * Calculate number of records in an alloc btree block.
+ */
+int
+xfs_allocbt_maxrecs(
+	struct xfs_mount	*mp,
+	int			blocklen,
+	int			leaf)
+{
+	blocklen -= XFS_ALLOC_BLOCK_LEN(mp);
+
+	if (leaf)
+		return blocklen / sizeof(xfs_alloc_rec_t);
+	return blocklen / (sizeof(xfs_alloc_key_t) + sizeof(xfs_alloc_ptr_t));
+}
diff --git a/fs/xfs/xfs_alloc_btree.h b/fs/xfs/xfs_alloc_btree.h
new file mode 100644
index 00000000..a6caa002
--- /dev/null
+++ b/fs/xfs/xfs_alloc_btree.h
@@ -0,0 +1,110 @@
+/*
+ * Copyright (c) 2000,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_ALLOC_BTREE_H__
+#define	__XFS_ALLOC_BTREE_H__
+
+/*
+ * Freespace on-disk structures
+ */
+
+struct xfs_buf;
+struct xfs_btree_cur;
+struct xfs_mount;
+
+/*
+ * There are two on-disk btrees, one sorted by blockno and one sorted
+ * by blockcount and blockno.  All blocks look the same to make the code
+ * simpler; if we have time later, we'll make the optimizations.
+ */
+#define	XFS_ABTB_MAGIC	0x41425442	/* 'ABTB' for bno tree */
+#define	XFS_ABTC_MAGIC	0x41425443	/* 'ABTC' for cnt tree */
+
+/*
+ * Data record/key structure
+ */
+typedef struct xfs_alloc_rec {
+	__be32		ar_startblock;	/* starting block number */
+	__be32		ar_blockcount;	/* count of free blocks */
+} xfs_alloc_rec_t, xfs_alloc_key_t;
+
+typedef struct xfs_alloc_rec_incore {
+	xfs_agblock_t	ar_startblock;	/* starting block number */
+	xfs_extlen_t	ar_blockcount;	/* count of free blocks */
+} xfs_alloc_rec_incore_t;
+
+/* btree pointer type */
+typedef __be32 xfs_alloc_ptr_t;
+
+/*
+ * Minimum and maximum blocksize and sectorsize.
+ * The blocksize upper limit is pretty much arbitrary.
+ * The sectorsize upper limit is due to sizeof(sb_sectsize).
+ */
+#define XFS_MIN_BLOCKSIZE_LOG	9	/* i.e. 512 bytes */
+#define XFS_MAX_BLOCKSIZE_LOG	16	/* i.e. 65536 bytes */
+#define XFS_MIN_BLOCKSIZE	(1 << XFS_MIN_BLOCKSIZE_LOG)
+#define XFS_MAX_BLOCKSIZE	(1 << XFS_MAX_BLOCKSIZE_LOG)
+#define XFS_MIN_SECTORSIZE_LOG	9	/* i.e. 512 bytes */
+#define XFS_MAX_SECTORSIZE_LOG	15	/* i.e. 32768 bytes */
+#define XFS_MIN_SECTORSIZE	(1 << XFS_MIN_SECTORSIZE_LOG)
+#define XFS_MAX_SECTORSIZE	(1 << XFS_MAX_SECTORSIZE_LOG)
+
+/*
+ * Block numbers in the AG:
+ * SB is sector 0, AGF is sector 1, AGI is sector 2, AGFL is sector 3.
+ */
+#define	XFS_BNO_BLOCK(mp)	((xfs_agblock_t)(XFS_AGFL_BLOCK(mp) + 1))
+#define	XFS_CNT_BLOCK(mp)	((xfs_agblock_t)(XFS_BNO_BLOCK(mp) + 1))
+
+/*
+ * Btree block header size depends on a superblock flag.
+ *
+ * (not quite yet, but soon)
+ */
+#define XFS_ALLOC_BLOCK_LEN(mp)	XFS_BTREE_SBLOCK_LEN
+
+/*
+ * Record, key, and pointer address macros for btree blocks.
+ *
+ * (note that some of these may appear unused, but they are used in userspace)
+ */
+#define XFS_ALLOC_REC_ADDR(mp, block, index) \
+	((xfs_alloc_rec_t *) \
+		((char *)(block) + \
+		 XFS_ALLOC_BLOCK_LEN(mp) + \
+		 (((index) - 1) * sizeof(xfs_alloc_rec_t))))
+
+#define XFS_ALLOC_KEY_ADDR(mp, block, index) \
+	((xfs_alloc_key_t *) \
+		((char *)(block) + \
+		 XFS_ALLOC_BLOCK_LEN(mp) + \
+		 ((index) - 1) * sizeof(xfs_alloc_key_t)))
+
+#define XFS_ALLOC_PTR_ADDR(mp, block, index, maxrecs) \
+	((xfs_alloc_ptr_t *) \
+		((char *)(block) + \
+		 XFS_ALLOC_BLOCK_LEN(mp) + \
+		 (maxrecs) * sizeof(xfs_alloc_key_t) + \
+		 ((index) - 1) * sizeof(xfs_alloc_ptr_t)))
+
+extern struct xfs_btree_cur *xfs_allocbt_init_cursor(struct xfs_mount *,
+		struct xfs_trans *, struct xfs_buf *,
+		xfs_agnumber_t, xfs_btnum_t);
+extern int xfs_allocbt_maxrecs(struct xfs_mount *, int, int);
+
+#endif	/* __XFS_ALLOC_BTREE_H__ */
diff --git a/fs/xfs/xfs_arch.h b/fs/xfs/xfs_arch.h
new file mode 100644
index 00000000..09022493
--- /dev/null
+++ b/fs/xfs/xfs_arch.h
@@ -0,0 +1,136 @@
+/*
+ * Copyright (c) 2000-2002,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_ARCH_H__
+#define __XFS_ARCH_H__
+
+#ifndef XFS_BIG_INUMS
+# error XFS_BIG_INUMS must be defined true or false
+#endif
+
+#ifdef __KERNEL__
+
+#include <asm/byteorder.h>
+
+#ifdef __BIG_ENDIAN
+#define	XFS_NATIVE_HOST	1
+#else
+#undef XFS_NATIVE_HOST
+#endif
+
+#else /* __KERNEL__ */
+
+#if __BYTE_ORDER == __BIG_ENDIAN
+#define	XFS_NATIVE_HOST	1
+#else
+#undef XFS_NATIVE_HOST
+#endif
+
+#ifdef XFS_NATIVE_HOST
+#define cpu_to_be16(val)	((__force __be16)(__u16)(val))
+#define cpu_to_be32(val)	((__force __be32)(__u32)(val))
+#define cpu_to_be64(val)	((__force __be64)(__u64)(val))
+#define be16_to_cpu(val)	((__force __u16)(__be16)(val))
+#define be32_to_cpu(val)	((__force __u32)(__be32)(val))
+#define be64_to_cpu(val)	((__force __u64)(__be64)(val))
+#else
+#define cpu_to_be16(val)	((__force __be16)__swab16((__u16)(val)))
+#define cpu_to_be32(val)	((__force __be32)__swab32((__u32)(val)))
+#define cpu_to_be64(val)	((__force __be64)__swab64((__u64)(val)))
+#define be16_to_cpu(val)	(__swab16((__force __u16)(__be16)(val)))
+#define be32_to_cpu(val)	(__swab32((__force __u32)(__be32)(val)))
+#define be64_to_cpu(val)	(__swab64((__force __u64)(__be64)(val)))
+#endif
+
+static inline void be16_add_cpu(__be16 *a, __s16 b)
+{
+	*a = cpu_to_be16(be16_to_cpu(*a) + b);
+}
+
+static inline void be32_add_cpu(__be32 *a, __s32 b)
+{
+	*a = cpu_to_be32(be32_to_cpu(*a) + b);
+}
+
+static inline void be64_add_cpu(__be64 *a, __s64 b)
+{
+	*a = cpu_to_be64(be64_to_cpu(*a) + b);
+}
+
+#endif	/* __KERNEL__ */
+
+/*
+ * get and set integers from potentially unaligned locations
+ */
+
+#define INT_GET_UNALIGNED_16_BE(pointer) \
+   ((__u16)((((__u8*)(pointer))[0] << 8) | (((__u8*)(pointer))[1])))
+#define INT_SET_UNALIGNED_16_BE(pointer,value) \
+    { \
+	((__u8*)(pointer))[0] = (((value) >> 8) & 0xff); \
+	((__u8*)(pointer))[1] = (((value)     ) & 0xff); \
+    }
+
+/*
+ * In directories inode numbers are stored as unaligned arrays of unsigned
+ * 8bit integers on disk.
+ *
+ * For v1 directories or v2 directories that contain inode numbers that
+ * do not fit into 32bit the array has eight members, but the first member
+ * is always zero:
+ *
+ *  |unused|48-55|40-47|32-39|24-31|16-23| 8-15| 0- 7|
+ *
+ * For v2 directories that only contain entries with inode numbers that fit
+ * into 32bits a four-member array is used:
+ *
+ *  |24-31|16-23| 8-15| 0- 7|
+ */ 
+
+#define XFS_GET_DIR_INO4(di) \
+	(((__u32)(di).i[0] << 24) | ((di).i[1] << 16) | ((di).i[2] << 8) | ((di).i[3]))
+
+#define XFS_PUT_DIR_INO4(from, di) \
+do { \
+	(di).i[0] = (((from) & 0xff000000ULL) >> 24); \
+	(di).i[1] = (((from) & 0x00ff0000ULL) >> 16); \
+	(di).i[2] = (((from) & 0x0000ff00ULL) >> 8); \
+	(di).i[3] = ((from) & 0x000000ffULL); \
+} while (0)
+
+#define XFS_DI_HI(di) \
+	(((__u32)(di).i[1] << 16) | ((di).i[2] << 8) | ((di).i[3]))
+#define XFS_DI_LO(di) \
+	(((__u32)(di).i[4] << 24) | ((di).i[5] << 16) | ((di).i[6] << 8) | ((di).i[7]))
+
+#define XFS_GET_DIR_INO8(di)        \
+	(((xfs_ino_t)XFS_DI_LO(di) & 0xffffffffULL) | \
+	 ((xfs_ino_t)XFS_DI_HI(di) << 32))
+
+#define XFS_PUT_DIR_INO8(from, di) \
+do { \
+	(di).i[0] = 0; \
+	(di).i[1] = (((from) & 0x00ff000000000000ULL) >> 48); \
+	(di).i[2] = (((from) & 0x0000ff0000000000ULL) >> 40); \
+	(di).i[3] = (((from) & 0x000000ff00000000ULL) >> 32); \
+	(di).i[4] = (((from) & 0x00000000ff000000ULL) >> 24); \
+	(di).i[5] = (((from) & 0x0000000000ff0000ULL) >> 16); \
+	(di).i[6] = (((from) & 0x000000000000ff00ULL) >> 8); \
+	(di).i[7] = ((from) & 0x00000000000000ffULL); \
+} while (0)
+	
+#endif	/* __XFS_ARCH_H__ */
diff --git a/fs/xfs/xfs_attr.c b/fs/xfs/xfs_attr.c
new file mode 100644
index 00000000..99d40116
--- /dev/null
+++ b/fs/xfs/xfs_attr.c
@@ -0,0 +1,2230 @@
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_mount.h"
+#include "xfs_da_btree.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_attr_sf.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_alloc.h"
+#include "xfs_inode_item.h"
+#include "xfs_bmap.h"
+#include "xfs_attr.h"
+#include "xfs_attr_leaf.h"
+#include "xfs_error.h"
+#include "xfs_quota.h"
+#include "xfs_trans_space.h"
+#include "xfs_rw.h"
+#include "xfs_vnodeops.h"
+#include "xfs_trace.h"
+
+/*
+ * xfs_attr.c
+ *
+ * Provide the external interfaces to manage attribute lists.
+ */
+
+/*========================================================================
+ * Function prototypes for the kernel.
+ *========================================================================*/
+
+/*
+ * Internal routines when attribute list fits inside the inode.
+ */
+STATIC int xfs_attr_shortform_addname(xfs_da_args_t *args);
+
+/*
+ * Internal routines when attribute list is one block.
+ */
+STATIC int xfs_attr_leaf_get(xfs_da_args_t *args);
+STATIC int xfs_attr_leaf_addname(xfs_da_args_t *args);
+STATIC int xfs_attr_leaf_removename(xfs_da_args_t *args);
+STATIC int xfs_attr_leaf_list(xfs_attr_list_context_t *context);
+
+/*
+ * Internal routines when attribute list is more than one block.
+ */
+STATIC int xfs_attr_node_get(xfs_da_args_t *args);
+STATIC int xfs_attr_node_addname(xfs_da_args_t *args);
+STATIC int xfs_attr_node_removename(xfs_da_args_t *args);
+STATIC int xfs_attr_node_list(xfs_attr_list_context_t *context);
+STATIC int xfs_attr_fillstate(xfs_da_state_t *state);
+STATIC int xfs_attr_refillstate(xfs_da_state_t *state);
+
+/*
+ * Routines to manipulate out-of-line attribute values.
+ */
+STATIC int xfs_attr_rmtval_set(xfs_da_args_t *args);
+STATIC int xfs_attr_rmtval_remove(xfs_da_args_t *args);
+
+#define ATTR_RMTVALUE_MAPSIZE	1	/* # of map entries at once */
+
+STATIC int
+xfs_attr_name_to_xname(
+	struct xfs_name	*xname,
+	const unsigned char *aname)
+{
+	if (!aname)
+		return EINVAL;
+	xname->name = aname;
+	xname->len = strlen((char *)aname);
+	if (xname->len >= MAXNAMELEN)
+		return EFAULT;		/* match IRIX behaviour */
+
+	return 0;
+}
+
+STATIC int
+xfs_inode_hasattr(
+	struct xfs_inode	*ip)
+{
+	if (!XFS_IFORK_Q(ip) ||
+	    (ip->i_d.di_aformat == XFS_DINODE_FMT_EXTENTS &&
+	     ip->i_d.di_anextents == 0))
+		return 0;
+	return 1;
+}
+
+/*========================================================================
+ * Overall external interface routines.
+ *========================================================================*/
+
+STATIC int
+xfs_attr_get_int(
+	struct xfs_inode	*ip,
+	struct xfs_name		*name,
+	unsigned char		*value,
+	int			*valuelenp,
+	int			flags)
+{
+	xfs_da_args_t   args;
+	int             error;
+
+	if (!xfs_inode_hasattr(ip))
+		return ENOATTR;
+
+	/*
+	 * Fill in the arg structure for this request.
+	 */
+	memset((char *)&args, 0, sizeof(args));
+	args.name = name->name;
+	args.namelen = name->len;
+	args.value = value;
+	args.valuelen = *valuelenp;
+	args.flags = flags;
+	args.hashval = xfs_da_hashname(args.name, args.namelen);
+	args.dp = ip;
+	args.whichfork = XFS_ATTR_FORK;
+
+	/*
+	 * Decide on what work routines to call based on the inode size.
+	 */
+	if (ip->i_d.di_aformat == XFS_DINODE_FMT_LOCAL) {
+		error = xfs_attr_shortform_getvalue(&args);
+	} else if (xfs_bmap_one_block(ip, XFS_ATTR_FORK)) {
+		error = xfs_attr_leaf_get(&args);
+	} else {
+		error = xfs_attr_node_get(&args);
+	}
+
+	/*
+	 * Return the number of bytes in the value to the caller.
+	 */
+	*valuelenp = args.valuelen;
+
+	if (error == EEXIST)
+		error = 0;
+	return(error);
+}
+
+int
+xfs_attr_get(
+	xfs_inode_t	*ip,
+	const unsigned char *name,
+	unsigned char	*value,
+	int		*valuelenp,
+	int		flags)
+{
+	int		error;
+	struct xfs_name	xname;
+
+	XFS_STATS_INC(xs_attr_get);
+
+	if (XFS_FORCED_SHUTDOWN(ip->i_mount))
+		return(EIO);
+
+	error = xfs_attr_name_to_xname(&xname, name);
+	if (error)
+		return error;
+
+	xfs_ilock(ip, XFS_ILOCK_SHARED);
+	error = xfs_attr_get_int(ip, &xname, value, valuelenp, flags);
+	xfs_iunlock(ip, XFS_ILOCK_SHARED);
+	return(error);
+}
+
+/*
+ * Calculate how many blocks we need for the new attribute,
+ */
+STATIC int
+xfs_attr_calc_size(
+	struct xfs_inode 	*ip,
+	int			namelen,
+	int			valuelen,
+	int			*local)
+{
+	struct xfs_mount 	*mp = ip->i_mount;
+	int			size;
+	int			nblks;
+
+	/*
+	 * Determine space new attribute will use, and if it would be
+	 * "local" or "remote" (note: local != inline).
+	 */
+	size = xfs_attr_leaf_newentsize(namelen, valuelen,
+					mp->m_sb.sb_blocksize, local);
+
+	nblks = XFS_DAENTER_SPACE_RES(mp, XFS_ATTR_FORK);
+	if (*local) {
+		if (size > (mp->m_sb.sb_blocksize >> 1)) {
+			/* Double split possible */
+			nblks *= 2;
+		}
+	} else {
+		/*
+		 * Out of line attribute, cannot double split, but
+		 * make room for the attribute value itself.
+		 */
+		uint	dblocks = XFS_B_TO_FSB(mp, valuelen);
+		nblks += dblocks;
+		nblks += XFS_NEXTENTADD_SPACE_RES(mp, dblocks, XFS_ATTR_FORK);
+	}
+
+	return nblks;
+}
+
+STATIC int
+xfs_attr_set_int(
+	struct xfs_inode *dp,
+	struct xfs_name	*name,
+	unsigned char	*value,
+	int		valuelen,
+	int		flags)
+{
+	xfs_da_args_t	args;
+	xfs_fsblock_t	firstblock;
+	xfs_bmap_free_t flist;
+	int		error, err2, committed;
+	xfs_mount_t	*mp = dp->i_mount;
+	int             rsvd = (flags & ATTR_ROOT) != 0;
+	int		local;
+
+	/*
+	 * Attach the dquots to the inode.
+	 */
+	error = xfs_qm_dqattach(dp, 0);
+	if (error)
+		return error;
+
+	/*
+	 * If the inode doesn't have an attribute fork, add one.
+	 * (inode must not be locked when we call this routine)
+	 */
+	if (XFS_IFORK_Q(dp) == 0) {
+		int sf_size = sizeof(xfs_attr_sf_hdr_t) +
+			      XFS_ATTR_SF_ENTSIZE_BYNAME(name->len, valuelen);
+
+		if ((error = xfs_bmap_add_attrfork(dp, sf_size, rsvd)))
+			return(error);
+	}
+
+	/*
+	 * Fill in the arg structure for this request.
+	 */
+	memset((char *)&args, 0, sizeof(args));
+	args.name = name->name;
+	args.namelen = name->len;
+	args.value = value;
+	args.valuelen = valuelen;
+	args.flags = flags;
+	args.hashval = xfs_da_hashname(args.name, args.namelen);
+	args.dp = dp;
+	args.firstblock = &firstblock;
+	args.flist = &flist;
+	args.whichfork = XFS_ATTR_FORK;
+	args.op_flags = XFS_DA_OP_ADDNAME | XFS_DA_OP_OKNOENT;
+
+	/* Size is now blocks for attribute data */
+	args.total = xfs_attr_calc_size(dp, name->len, valuelen, &local);
+
+	/*
+	 * Start our first transaction of the day.
+	 *
+	 * All future transactions during this code must be "chained" off
+	 * this one via the trans_dup() call.  All transactions will contain
+	 * the inode, and the inode will always be marked with trans_ihold().
+	 * Since the inode will be locked in all transactions, we must log
+	 * the inode in every transaction to let it float upward through
+	 * the log.
+	 */
+	args.trans = xfs_trans_alloc(mp, XFS_TRANS_ATTR_SET);
+
+	/*
+	 * Root fork attributes can use reserved data blocks for this
+	 * operation if necessary
+	 */
+
+	if (rsvd)
+		args.trans->t_flags |= XFS_TRANS_RESERVE;
+
+	if ((error = xfs_trans_reserve(args.trans, args.total,
+			XFS_ATTRSET_LOG_RES(mp, args.total), 0,
+			XFS_TRANS_PERM_LOG_RES, XFS_ATTRSET_LOG_COUNT))) {
+		xfs_trans_cancel(args.trans, 0);
+		return(error);
+	}
+	xfs_ilock(dp, XFS_ILOCK_EXCL);
+
+	error = xfs_trans_reserve_quota_nblks(args.trans, dp, args.total, 0,
+				rsvd ? XFS_QMOPT_RES_REGBLKS | XFS_QMOPT_FORCE_RES :
+				       XFS_QMOPT_RES_REGBLKS);
+	if (error) {
+		xfs_iunlock(dp, XFS_ILOCK_EXCL);
+		xfs_trans_cancel(args.trans, XFS_TRANS_RELEASE_LOG_RES);
+		return (error);
+	}
+
+	xfs_trans_ijoin(args.trans, dp);
+
+	/*
+	 * If the attribute list is non-existent or a shortform list,
+	 * upgrade it to a single-leaf-block attribute list.
+	 */
+	if ((dp->i_d.di_aformat == XFS_DINODE_FMT_LOCAL) ||
+	    ((dp->i_d.di_aformat == XFS_DINODE_FMT_EXTENTS) &&
+	     (dp->i_d.di_anextents == 0))) {
+
+		/*
+		 * Build initial attribute list (if required).
+		 */
+		if (dp->i_d.di_aformat == XFS_DINODE_FMT_EXTENTS)
+			xfs_attr_shortform_create(&args);
+
+		/*
+		 * Try to add the attr to the attribute list in
+		 * the inode.
+		 */
+		error = xfs_attr_shortform_addname(&args);
+		if (error != ENOSPC) {
+			/*
+			 * Commit the shortform mods, and we're done.
+			 * NOTE: this is also the error path (EEXIST, etc).
+			 */
+			ASSERT(args.trans != NULL);
+
+			/*
+			 * If this is a synchronous mount, make sure that
+			 * the transaction goes to disk before returning
+			 * to the user.
+			 */
+			if (mp->m_flags & XFS_MOUNT_WSYNC) {
+				xfs_trans_set_sync(args.trans);
+			}
+
+			if (!error && (flags & ATTR_KERNOTIME) == 0) {
+				xfs_trans_ichgtime(args.trans, dp,
+							XFS_ICHGTIME_CHG);
+			}
+			err2 = xfs_trans_commit(args.trans,
+						 XFS_TRANS_RELEASE_LOG_RES);
+			xfs_iunlock(dp, XFS_ILOCK_EXCL);
+
+			return(error == 0 ? err2 : error);
+		}
+
+		/*
+		 * It won't fit in the shortform, transform to a leaf block.
+		 * GROT: another possible req'mt for a double-split btree op.
+		 */
+		xfs_bmap_init(args.flist, args.firstblock);
+		error = xfs_attr_shortform_to_leaf(&args);
+		if (!error) {
+			error = xfs_bmap_finish(&args.trans, args.flist,
+						&committed);
+		}
+		if (error) {
+			ASSERT(committed);
+			args.trans = NULL;
+			xfs_bmap_cancel(&flist);
+			goto out;
+		}
+
+		/*
+		 * bmap_finish() may have committed the last trans and started
+		 * a new one.  We need the inode to be in all transactions.
+		 */
+		if (committed)
+			xfs_trans_ijoin(args.trans, dp);
+
+		/*
+		 * Commit the leaf transformation.  We'll need another (linked)
+		 * transaction to add the new attribute to the leaf.
+		 */
+
+		error = xfs_trans_roll(&args.trans, dp);
+		if (error)
+			goto out;
+
+	}
+
+	if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) {
+		error = xfs_attr_leaf_addname(&args);
+	} else {
+		error = xfs_attr_node_addname(&args);
+	}
+	if (error) {
+		goto out;
+	}
+
+	/*
+	 * If this is a synchronous mount, make sure that the
+	 * transaction goes to disk before returning to the user.
+	 */
+	if (mp->m_flags & XFS_MOUNT_WSYNC) {
+		xfs_trans_set_sync(args.trans);
+	}
+
+	if ((flags & ATTR_KERNOTIME) == 0)
+		xfs_trans_ichgtime(args.trans, dp, XFS_ICHGTIME_CHG);
+
+	/*
+	 * Commit the last in the sequence of transactions.
+	 */
+	xfs_trans_log_inode(args.trans, dp, XFS_ILOG_CORE);
+	error = xfs_trans_commit(args.trans, XFS_TRANS_RELEASE_LOG_RES);
+	xfs_iunlock(dp, XFS_ILOCK_EXCL);
+
+	return(error);
+
+out:
+	if (args.trans)
+		xfs_trans_cancel(args.trans,
+			XFS_TRANS_RELEASE_LOG_RES|XFS_TRANS_ABORT);
+	xfs_iunlock(dp, XFS_ILOCK_EXCL);
+	return(error);
+}
+
+int
+xfs_attr_set(
+	xfs_inode_t	*dp,
+	const unsigned char *name,
+	unsigned char	*value,
+	int		valuelen,
+	int		flags)
+{
+	int             error;
+	struct xfs_name	xname;
+
+	XFS_STATS_INC(xs_attr_set);
+
+	if (XFS_FORCED_SHUTDOWN(dp->i_mount))
+		return (EIO);
+
+	error = xfs_attr_name_to_xname(&xname, name);
+	if (error)
+		return error;
+
+	return xfs_attr_set_int(dp, &xname, value, valuelen, flags);
+}
+
+/*
+ * Generic handler routine to remove a name from an attribute list.
+ * Transitions attribute list from Btree to shortform as necessary.
+ */
+STATIC int
+xfs_attr_remove_int(xfs_inode_t *dp, struct xfs_name *name, int flags)
+{
+	xfs_da_args_t	args;
+	xfs_fsblock_t	firstblock;
+	xfs_bmap_free_t	flist;
+	int		error;
+	xfs_mount_t	*mp = dp->i_mount;
+
+	/*
+	 * Fill in the arg structure for this request.
+	 */
+	memset((char *)&args, 0, sizeof(args));
+	args.name = name->name;
+	args.namelen = name->len;
+	args.flags = flags;
+	args.hashval = xfs_da_hashname(args.name, args.namelen);
+	args.dp = dp;
+	args.firstblock = &firstblock;
+	args.flist = &flist;
+	args.total = 0;
+	args.whichfork = XFS_ATTR_FORK;
+
+	/*
+	 * we have no control over the attribute names that userspace passes us
+	 * to remove, so we have to allow the name lookup prior to attribute
+	 * removal to fail.
+	 */
+	args.op_flags = XFS_DA_OP_OKNOENT;
+
+	/*
+	 * Attach the dquots to the inode.
+	 */
+	error = xfs_qm_dqattach(dp, 0);
+	if (error)
+		return error;
+
+	/*
+	 * Start our first transaction of the day.
+	 *
+	 * All future transactions during this code must be "chained" off
+	 * this one via the trans_dup() call.  All transactions will contain
+	 * the inode, and the inode will always be marked with trans_ihold().
+	 * Since the inode will be locked in all transactions, we must log
+	 * the inode in every transaction to let it float upward through
+	 * the log.
+	 */
+	args.trans = xfs_trans_alloc(mp, XFS_TRANS_ATTR_RM);
+
+	/*
+	 * Root fork attributes can use reserved data blocks for this
+	 * operation if necessary
+	 */
+
+	if (flags & ATTR_ROOT)
+		args.trans->t_flags |= XFS_TRANS_RESERVE;
+
+	if ((error = xfs_trans_reserve(args.trans,
+				      XFS_ATTRRM_SPACE_RES(mp),
+				      XFS_ATTRRM_LOG_RES(mp),
+				      0, XFS_TRANS_PERM_LOG_RES,
+				      XFS_ATTRRM_LOG_COUNT))) {
+		xfs_trans_cancel(args.trans, 0);
+		return(error);
+	}
+
+	xfs_ilock(dp, XFS_ILOCK_EXCL);
+	/*
+	 * No need to make quota reservations here. We expect to release some
+	 * blocks not allocate in the common case.
+	 */
+	xfs_trans_ijoin(args.trans, dp);
+
+	/*
+	 * Decide on what work routines to call based on the inode size.
+	 */
+	if (!xfs_inode_hasattr(dp)) {
+		error = XFS_ERROR(ENOATTR);
+		goto out;
+	}
+	if (dp->i_d.di_aformat == XFS_DINODE_FMT_LOCAL) {
+		ASSERT(dp->i_afp->if_flags & XFS_IFINLINE);
+		error = xfs_attr_shortform_remove(&args);
+		if (error) {
+			goto out;
+		}
+	} else if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) {
+		error = xfs_attr_leaf_removename(&args);
+	} else {
+		error = xfs_attr_node_removename(&args);
+	}
+	if (error) {
+		goto out;
+	}
+
+	/*
+	 * If this is a synchronous mount, make sure that the
+	 * transaction goes to disk before returning to the user.
+	 */
+	if (mp->m_flags & XFS_MOUNT_WSYNC) {
+		xfs_trans_set_sync(args.trans);
+	}
+
+	if ((flags & ATTR_KERNOTIME) == 0)
+		xfs_trans_ichgtime(args.trans, dp, XFS_ICHGTIME_CHG);
+
+	/*
+	 * Commit the last in the sequence of transactions.
+	 */
+	xfs_trans_log_inode(args.trans, dp, XFS_ILOG_CORE);
+	error = xfs_trans_commit(args.trans, XFS_TRANS_RELEASE_LOG_RES);
+	xfs_iunlock(dp, XFS_ILOCK_EXCL);
+
+	return(error);
+
+out:
+	if (args.trans)
+		xfs_trans_cancel(args.trans,
+			XFS_TRANS_RELEASE_LOG_RES|XFS_TRANS_ABORT);
+	xfs_iunlock(dp, XFS_ILOCK_EXCL);
+	return(error);
+}
+
+int
+xfs_attr_remove(
+	xfs_inode_t	*dp,
+	const unsigned char *name,
+	int		flags)
+{
+	int		error;
+	struct xfs_name	xname;
+
+	XFS_STATS_INC(xs_attr_remove);
+
+	if (XFS_FORCED_SHUTDOWN(dp->i_mount))
+		return (EIO);
+
+	error = xfs_attr_name_to_xname(&xname, name);
+	if (error)
+		return error;
+
+	xfs_ilock(dp, XFS_ILOCK_SHARED);
+	if (!xfs_inode_hasattr(dp)) {
+		xfs_iunlock(dp, XFS_ILOCK_SHARED);
+		return XFS_ERROR(ENOATTR);
+	}
+	xfs_iunlock(dp, XFS_ILOCK_SHARED);
+
+	return xfs_attr_remove_int(dp, &xname, flags);
+}
+
+int
+xfs_attr_list_int(xfs_attr_list_context_t *context)
+{
+	int error;
+	xfs_inode_t *dp = context->dp;
+
+	XFS_STATS_INC(xs_attr_list);
+
+	if (XFS_FORCED_SHUTDOWN(dp->i_mount))
+		return EIO;
+
+	xfs_ilock(dp, XFS_ILOCK_SHARED);
+
+	/*
+	 * Decide on what work routines to call based on the inode size.
+	 */
+	if (!xfs_inode_hasattr(dp)) {
+		error = 0;
+	} else if (dp->i_d.di_aformat == XFS_DINODE_FMT_LOCAL) {
+		error = xfs_attr_shortform_list(context);
+	} else if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) {
+		error = xfs_attr_leaf_list(context);
+	} else {
+		error = xfs_attr_node_list(context);
+	}
+
+	xfs_iunlock(dp, XFS_ILOCK_SHARED);
+
+	return error;
+}
+
+#define	ATTR_ENTBASESIZE		/* minimum bytes used by an attr */ \
+	(((struct attrlist_ent *) 0)->a_name - (char *) 0)
+#define	ATTR_ENTSIZE(namelen)		/* actual bytes used by an attr */ \
+	((ATTR_ENTBASESIZE + (namelen) + 1 + sizeof(u_int32_t)-1) \
+	 & ~(sizeof(u_int32_t)-1))
+
+/*
+ * Format an attribute and copy it out to the user's buffer.
+ * Take care to check values and protect against them changing later,
+ * we may be reading them directly out of a user buffer.
+ */
+/*ARGSUSED*/
+STATIC int
+xfs_attr_put_listent(
+	xfs_attr_list_context_t *context,
+	int		flags,
+	unsigned char	*name,
+	int		namelen,
+	int		valuelen,
+	unsigned char	*value)
+{
+	struct attrlist *alist = (struct attrlist *)context->alist;
+	attrlist_ent_t *aep;
+	int arraytop;
+
+	ASSERT(!(context->flags & ATTR_KERNOVAL));
+	ASSERT(context->count >= 0);
+	ASSERT(context->count < (ATTR_MAX_VALUELEN/8));
+	ASSERT(context->firstu >= sizeof(*alist));
+	ASSERT(context->firstu <= context->bufsize);
+
+	/*
+	 * Only list entries in the right namespace.
+	 */
+	if (((context->flags & ATTR_SECURE) == 0) !=
+	    ((flags & XFS_ATTR_SECURE) == 0))
+		return 0;
+	if (((context->flags & ATTR_ROOT) == 0) !=
+	    ((flags & XFS_ATTR_ROOT) == 0))
+		return 0;
+
+	arraytop = sizeof(*alist) +
+			context->count * sizeof(alist->al_offset[0]);
+	context->firstu -= ATTR_ENTSIZE(namelen);
+	if (context->firstu < arraytop) {
+		trace_xfs_attr_list_full(context);
+		alist->al_more = 1;
+		context->seen_enough = 1;
+		return 1;
+	}
+
+	aep = (attrlist_ent_t *)&context->alist[context->firstu];
+	aep->a_valuelen = valuelen;
+	memcpy(aep->a_name, name, namelen);
+	aep->a_name[namelen] = 0;
+	alist->al_offset[context->count++] = context->firstu;
+	alist->al_count = context->count;
+	trace_xfs_attr_list_add(context);
+	return 0;
+}
+
+/*
+ * Generate a list of extended attribute names and optionally
+ * also value lengths.  Positive return value follows the XFS
+ * convention of being an error, zero or negative return code
+ * is the length of the buffer returned (negated), indicating
+ * success.
+ */
+int
+xfs_attr_list(
+	xfs_inode_t	*dp,
+	char		*buffer,
+	int		bufsize,
+	int		flags,
+	attrlist_cursor_kern_t *cursor)
+{
+	xfs_attr_list_context_t context;
+	struct attrlist *alist;
+	int error;
+
+	/*
+	 * Validate the cursor.
+	 */
+	if (cursor->pad1 || cursor->pad2)
+		return(XFS_ERROR(EINVAL));
+	if ((cursor->initted == 0) &&
+	    (cursor->hashval || cursor->blkno || cursor->offset))
+		return XFS_ERROR(EINVAL);
+
+	/*
+	 * Check for a properly aligned buffer.
+	 */
+	if (((long)buffer) & (sizeof(int)-1))
+		return XFS_ERROR(EFAULT);
+	if (flags & ATTR_KERNOVAL)
+		bufsize = 0;
+
+	/*
+	 * Initialize the output buffer.
+	 */
+	memset(&context, 0, sizeof(context));
+	context.dp = dp;
+	context.cursor = cursor;
+	context.resynch = 1;
+	context.flags = flags;
+	context.alist = buffer;
+	context.bufsize = (bufsize & ~(sizeof(int)-1));  /* align */
+	context.firstu = context.bufsize;
+	context.put_listent = xfs_attr_put_listent;
+
+	alist = (struct attrlist *)context.alist;
+	alist->al_count = 0;
+	alist->al_more = 0;
+	alist->al_offset[0] = context.bufsize;
+
+	error = xfs_attr_list_int(&context);
+	ASSERT(error >= 0);
+	return error;
+}
+
+int								/* error */
+xfs_attr_inactive(xfs_inode_t *dp)
+{
+	xfs_trans_t *trans;
+	xfs_mount_t *mp;
+	int error;
+
+	mp = dp->i_mount;
+	ASSERT(! XFS_NOT_DQATTACHED(mp, dp));
+
+	xfs_ilock(dp, XFS_ILOCK_SHARED);
+	if (!xfs_inode_hasattr(dp) ||
+	    dp->i_d.di_aformat == XFS_DINODE_FMT_LOCAL) {
+		xfs_iunlock(dp, XFS_ILOCK_SHARED);
+		return 0;
+	}
+	xfs_iunlock(dp, XFS_ILOCK_SHARED);
+
+	/*
+	 * Start our first transaction of the day.
+	 *
+	 * All future transactions during this code must be "chained" off
+	 * this one via the trans_dup() call.  All transactions will contain
+	 * the inode, and the inode will always be marked with trans_ihold().
+	 * Since the inode will be locked in all transactions, we must log
+	 * the inode in every transaction to let it float upward through
+	 * the log.
+	 */
+	trans = xfs_trans_alloc(mp, XFS_TRANS_ATTRINVAL);
+	if ((error = xfs_trans_reserve(trans, 0, XFS_ATTRINVAL_LOG_RES(mp), 0,
+				      XFS_TRANS_PERM_LOG_RES,
+				      XFS_ATTRINVAL_LOG_COUNT))) {
+		xfs_trans_cancel(trans, 0);
+		return(error);
+	}
+	xfs_ilock(dp, XFS_ILOCK_EXCL);
+
+	/*
+	 * No need to make quota reservations here. We expect to release some
+	 * blocks, not allocate, in the common case.
+	 */
+	xfs_trans_ijoin(trans, dp);
+
+	/*
+	 * Decide on what work routines to call based on the inode size.
+	 */
+	if (!xfs_inode_hasattr(dp) ||
+	    dp->i_d.di_aformat == XFS_DINODE_FMT_LOCAL) {
+		error = 0;
+		goto out;
+	}
+	error = xfs_attr_root_inactive(&trans, dp);
+	if (error)
+		goto out;
+
+	error = xfs_itruncate_finish(&trans, dp, 0LL, XFS_ATTR_FORK, 0);
+	if (error)
+		goto out;
+
+	/*
+	 * Commit the last in the sequence of transactions.
+	 */
+	xfs_trans_log_inode(trans, dp, XFS_ILOG_CORE);
+	error = xfs_trans_commit(trans, XFS_TRANS_RELEASE_LOG_RES);
+	xfs_iunlock(dp, XFS_ILOCK_EXCL);
+
+	return(error);
+
+out:
+	xfs_trans_cancel(trans, XFS_TRANS_RELEASE_LOG_RES|XFS_TRANS_ABORT);
+	xfs_iunlock(dp, XFS_ILOCK_EXCL);
+	return(error);
+}
+
+
+
+/*========================================================================
+ * External routines when attribute list is inside the inode
+ *========================================================================*/
+
+/*
+ * Add a name to the shortform attribute list structure
+ * This is the external routine.
+ */
+STATIC int
+xfs_attr_shortform_addname(xfs_da_args_t *args)
+{
+	int newsize, forkoff, retval;
+
+	retval = xfs_attr_shortform_lookup(args);
+	if ((args->flags & ATTR_REPLACE) && (retval == ENOATTR)) {
+		return(retval);
+	} else if (retval == EEXIST) {
+		if (args->flags & ATTR_CREATE)
+			return(retval);
+		retval = xfs_attr_shortform_remove(args);
+		ASSERT(retval == 0);
+	}
+
+	if (args->namelen >= XFS_ATTR_SF_ENTSIZE_MAX ||
+	    args->valuelen >= XFS_ATTR_SF_ENTSIZE_MAX)
+		return(XFS_ERROR(ENOSPC));
+
+	newsize = XFS_ATTR_SF_TOTSIZE(args->dp);
+	newsize += XFS_ATTR_SF_ENTSIZE_BYNAME(args->namelen, args->valuelen);
+
+	forkoff = xfs_attr_shortform_bytesfit(args->dp, newsize);
+	if (!forkoff)
+		return(XFS_ERROR(ENOSPC));
+
+	xfs_attr_shortform_add(args, forkoff);
+	return(0);
+}
+
+
+/*========================================================================
+ * External routines when attribute list is one block
+ *========================================================================*/
+
+/*
+ * Add a name to the leaf attribute list structure
+ *
+ * This leaf block cannot have a "remote" value, we only call this routine
+ * if bmap_one_block() says there is only one block (ie: no remote blks).
+ */
+STATIC int
+xfs_attr_leaf_addname(xfs_da_args_t *args)
+{
+	xfs_inode_t *dp;
+	xfs_dabuf_t *bp;
+	int retval, error, committed, forkoff;
+
+	/*
+	 * Read the (only) block in the attribute list in.
+	 */
+	dp = args->dp;
+	args->blkno = 0;
+	error = xfs_da_read_buf(args->trans, args->dp, args->blkno, -1, &bp,
+					     XFS_ATTR_FORK);
+	if (error)
+		return(error);
+	ASSERT(bp != NULL);
+
+	/*
+	 * Look up the given attribute in the leaf block.  Figure out if
+	 * the given flags produce an error or call for an atomic rename.
+	 */
+	retval = xfs_attr_leaf_lookup_int(bp, args);
+	if ((args->flags & ATTR_REPLACE) && (retval == ENOATTR)) {
+		xfs_da_brelse(args->trans, bp);
+		return(retval);
+	} else if (retval == EEXIST) {
+		if (args->flags & ATTR_CREATE) {	/* pure create op */
+			xfs_da_brelse(args->trans, bp);
+			return(retval);
+		}
+		args->op_flags |= XFS_DA_OP_RENAME;	/* an atomic rename */
+		args->blkno2 = args->blkno;		/* set 2nd entry info*/
+		args->index2 = args->index;
+		args->rmtblkno2 = args->rmtblkno;
+		args->rmtblkcnt2 = args->rmtblkcnt;
+	}
+
+	/*
+	 * Add the attribute to the leaf block, transitioning to a Btree
+	 * if required.
+	 */
+	retval = xfs_attr_leaf_add(bp, args);
+	xfs_da_buf_done(bp);
+	if (retval == ENOSPC) {
+		/*
+		 * Promote the attribute list to the Btree format, then
+		 * Commit that transaction so that the node_addname() call
+		 * can manage its own transactions.
+		 */
+		xfs_bmap_init(args->flist, args->firstblock);
+		error = xfs_attr_leaf_to_node(args);
+		if (!error) {
+			error = xfs_bmap_finish(&args->trans, args->flist,
+						&committed);
+		}
+		if (error) {
+			ASSERT(committed);
+			args->trans = NULL;
+			xfs_bmap_cancel(args->flist);
+			return(error);
+		}
+
+		/*
+		 * bmap_finish() may have committed the last trans and started
+		 * a new one.  We need the inode to be in all transactions.
+		 */
+		if (committed)
+			xfs_trans_ijoin(args->trans, dp);
+
+		/*
+		 * Commit the current trans (including the inode) and start
+		 * a new one.
+		 */
+		error = xfs_trans_roll(&args->trans, dp);
+		if (error)
+			return (error);
+
+		/*
+		 * Fob the whole rest of the problem off on the Btree code.
+		 */
+		error = xfs_attr_node_addname(args);
+		return(error);
+	}
+
+	/*
+	 * Commit the transaction that added the attr name so that
+	 * later routines can manage their own transactions.
+	 */
+	error = xfs_trans_roll(&args->trans, dp);
+	if (error)
+		return (error);
+
+	/*
+	 * If there was an out-of-line value, allocate the blocks we
+	 * identified for its storage and copy the value.  This is done
+	 * after we create the attribute so that we don't overflow the
+	 * maximum size of a transaction and/or hit a deadlock.
+	 */
+	if (args->rmtblkno > 0) {
+		error = xfs_attr_rmtval_set(args);
+		if (error)
+			return(error);
+	}
+
+	/*
+	 * If this is an atomic rename operation, we must "flip" the
+	 * incomplete flags on the "new" and "old" attribute/value pairs
+	 * so that one disappears and one appears atomically.  Then we
+	 * must remove the "old" attribute/value pair.
+	 */
+	if (args->op_flags & XFS_DA_OP_RENAME) {
+		/*
+		 * In a separate transaction, set the incomplete flag on the
+		 * "old" attr and clear the incomplete flag on the "new" attr.
+		 */
+		error = xfs_attr_leaf_flipflags(args);
+		if (error)
+			return(error);
+
+		/*
+		 * Dismantle the "old" attribute/value pair by removing
+		 * a "remote" value (if it exists).
+		 */
+		args->index = args->index2;
+		args->blkno = args->blkno2;
+		args->rmtblkno = args->rmtblkno2;
+		args->rmtblkcnt = args->rmtblkcnt2;
+		if (args->rmtblkno) {
+			error = xfs_attr_rmtval_remove(args);
+			if (error)
+				return(error);
+		}
+
+		/*
+		 * Read in the block containing the "old" attr, then
+		 * remove the "old" attr from that block (neat, huh!)
+		 */
+		error = xfs_da_read_buf(args->trans, args->dp, args->blkno, -1,
+						     &bp, XFS_ATTR_FORK);
+		if (error)
+			return(error);
+		ASSERT(bp != NULL);
+		(void)xfs_attr_leaf_remove(bp, args);
+
+		/*
+		 * If the result is small enough, shrink it all into the inode.
+		 */
+		if ((forkoff = xfs_attr_shortform_allfit(bp, dp))) {
+			xfs_bmap_init(args->flist, args->firstblock);
+			error = xfs_attr_leaf_to_shortform(bp, args, forkoff);
+			/* bp is gone due to xfs_da_shrink_inode */
+			if (!error) {
+				error = xfs_bmap_finish(&args->trans,
+							args->flist,
+							&committed);
+			}
+			if (error) {
+				ASSERT(committed);
+				args->trans = NULL;
+				xfs_bmap_cancel(args->flist);
+				return(error);
+			}
+
+			/*
+			 * bmap_finish() may have committed the last trans
+			 * and started a new one.  We need the inode to be
+			 * in all transactions.
+			 */
+			if (committed)
+				xfs_trans_ijoin(args->trans, dp);
+		} else
+			xfs_da_buf_done(bp);
+
+		/*
+		 * Commit the remove and start the next trans in series.
+		 */
+		error = xfs_trans_roll(&args->trans, dp);
+
+	} else if (args->rmtblkno > 0) {
+		/*
+		 * Added a "remote" value, just clear the incomplete flag.
+		 */
+		error = xfs_attr_leaf_clearflag(args);
+	}
+	return(error);
+}
+
+/*
+ * Remove a name from the leaf attribute list structure
+ *
+ * This leaf block cannot have a "remote" value, we only call this routine
+ * if bmap_one_block() says there is only one block (ie: no remote blks).
+ */
+STATIC int
+xfs_attr_leaf_removename(xfs_da_args_t *args)
+{
+	xfs_inode_t *dp;
+	xfs_dabuf_t *bp;
+	int error, committed, forkoff;
+
+	/*
+	 * Remove the attribute.
+	 */
+	dp = args->dp;
+	args->blkno = 0;
+	error = xfs_da_read_buf(args->trans, args->dp, args->blkno, -1, &bp,
+					     XFS_ATTR_FORK);
+	if (error) {
+		return(error);
+	}
+
+	ASSERT(bp != NULL);
+	error = xfs_attr_leaf_lookup_int(bp, args);
+	if (error == ENOATTR) {
+		xfs_da_brelse(args->trans, bp);
+		return(error);
+	}
+
+	(void)xfs_attr_leaf_remove(bp, args);
+
+	/*
+	 * If the result is small enough, shrink it all into the inode.
+	 */
+	if ((forkoff = xfs_attr_shortform_allfit(bp, dp))) {
+		xfs_bmap_init(args->flist, args->firstblock);
+		error = xfs_attr_leaf_to_shortform(bp, args, forkoff);
+		/* bp is gone due to xfs_da_shrink_inode */
+		if (!error) {
+			error = xfs_bmap_finish(&args->trans, args->flist,
+						&committed);
+		}
+		if (error) {
+			ASSERT(committed);
+			args->trans = NULL;
+			xfs_bmap_cancel(args->flist);
+			return(error);
+		}
+
+		/*
+		 * bmap_finish() may have committed the last trans and started
+		 * a new one.  We need the inode to be in all transactions.
+		 */
+		if (committed)
+			xfs_trans_ijoin(args->trans, dp);
+	} else
+		xfs_da_buf_done(bp);
+	return(0);
+}
+
+/*
+ * Look up a name in a leaf attribute list structure.
+ *
+ * This leaf block cannot have a "remote" value, we only call this routine
+ * if bmap_one_block() says there is only one block (ie: no remote blks).
+ */
+STATIC int
+xfs_attr_leaf_get(xfs_da_args_t *args)
+{
+	xfs_dabuf_t *bp;
+	int error;
+
+	args->blkno = 0;
+	error = xfs_da_read_buf(args->trans, args->dp, args->blkno, -1, &bp,
+					     XFS_ATTR_FORK);
+	if (error)
+		return(error);
+	ASSERT(bp != NULL);
+
+	error = xfs_attr_leaf_lookup_int(bp, args);
+	if (error != EEXIST)  {
+		xfs_da_brelse(args->trans, bp);
+		return(error);
+	}
+	error = xfs_attr_leaf_getvalue(bp, args);
+	xfs_da_brelse(args->trans, bp);
+	if (!error && (args->rmtblkno > 0) && !(args->flags & ATTR_KERNOVAL)) {
+		error = xfs_attr_rmtval_get(args);
+	}
+	return(error);
+}
+
+/*
+ * Copy out attribute entries for attr_list(), for leaf attribute lists.
+ */
+STATIC int
+xfs_attr_leaf_list(xfs_attr_list_context_t *context)
+{
+	xfs_attr_leafblock_t *leaf;
+	int error;
+	xfs_dabuf_t *bp;
+
+	context->cursor->blkno = 0;
+	error = xfs_da_read_buf(NULL, context->dp, 0, -1, &bp, XFS_ATTR_FORK);
+	if (error)
+		return XFS_ERROR(error);
+	ASSERT(bp != NULL);
+	leaf = bp->data;
+	if (unlikely(be16_to_cpu(leaf->hdr.info.magic) != XFS_ATTR_LEAF_MAGIC)) {
+		XFS_CORRUPTION_ERROR("xfs_attr_leaf_list", XFS_ERRLEVEL_LOW,
+				     context->dp->i_mount, leaf);
+		xfs_da_brelse(NULL, bp);
+		return XFS_ERROR(EFSCORRUPTED);
+	}
+
+	error = xfs_attr_leaf_list_int(bp, context);
+	xfs_da_brelse(NULL, bp);
+	return XFS_ERROR(error);
+}
+
+
+/*========================================================================
+ * External routines when attribute list size > XFS_LBSIZE(mp).
+ *========================================================================*/
+
+/*
+ * Add a name to a Btree-format attribute list.
+ *
+ * This will involve walking down the Btree, and may involve splitting
+ * leaf nodes and even splitting intermediate nodes up to and including
+ * the root node (a special case of an intermediate node).
+ *
+ * "Remote" attribute values confuse the issue and atomic rename operations
+ * add a whole extra layer of confusion on top of that.
+ */
+STATIC int
+xfs_attr_node_addname(xfs_da_args_t *args)
+{
+	xfs_da_state_t *state;
+	xfs_da_state_blk_t *blk;
+	xfs_inode_t *dp;
+	xfs_mount_t *mp;
+	int committed, retval, error;
+
+	/*
+	 * Fill in bucket of arguments/results/context to carry around.
+	 */
+	dp = args->dp;
+	mp = dp->i_mount;
+restart:
+	state = xfs_da_state_alloc();
+	state->args = args;
+	state->mp = mp;
+	state->blocksize = state->mp->m_sb.sb_blocksize;
+	state->node_ents = state->mp->m_attr_node_ents;
+
+	/*
+	 * Search to see if name already exists, and get back a pointer
+	 * to where it should go.
+	 */
+	error = xfs_da_node_lookup_int(state, &retval);
+	if (error)
+		goto out;
+	blk = &state->path.blk[ state->path.active-1 ];
+	ASSERT(blk->magic == XFS_ATTR_LEAF_MAGIC);
+	if ((args->flags & ATTR_REPLACE) && (retval == ENOATTR)) {
+		goto out;
+	} else if (retval == EEXIST) {
+		if (args->flags & ATTR_CREATE)
+			goto out;
+		args->op_flags |= XFS_DA_OP_RENAME;	/* atomic rename op */
+		args->blkno2 = args->blkno;		/* set 2nd entry info*/
+		args->index2 = args->index;
+		args->rmtblkno2 = args->rmtblkno;
+		args->rmtblkcnt2 = args->rmtblkcnt;
+		args->rmtblkno = 0;
+		args->rmtblkcnt = 0;
+	}
+
+	retval = xfs_attr_leaf_add(blk->bp, state->args);
+	if (retval == ENOSPC) {
+		if (state->path.active == 1) {
+			/*
+			 * Its really a single leaf node, but it had
+			 * out-of-line values so it looked like it *might*
+			 * have been a b-tree.
+			 */
+			xfs_da_state_free(state);
+			xfs_bmap_init(args->flist, args->firstblock);
+			error = xfs_attr_leaf_to_node(args);
+			if (!error) {
+				error = xfs_bmap_finish(&args->trans,
+							args->flist,
+							&committed);
+			}
+			if (error) {
+				ASSERT(committed);
+				args->trans = NULL;
+				xfs_bmap_cancel(args->flist);
+				goto out;
+			}
+
+			/*
+			 * bmap_finish() may have committed the last trans
+			 * and started a new one.  We need the inode to be
+			 * in all transactions.
+			 */
+			if (committed)
+				xfs_trans_ijoin(args->trans, dp);
+
+			/*
+			 * Commit the node conversion and start the next
+			 * trans in the chain.
+			 */
+			error = xfs_trans_roll(&args->trans, dp);
+			if (error)
+				goto out;
+
+			goto restart;
+		}
+
+		/*
+		 * Split as many Btree elements as required.
+		 * This code tracks the new and old attr's location
+		 * in the index/blkno/rmtblkno/rmtblkcnt fields and
+		 * in the index2/blkno2/rmtblkno2/rmtblkcnt2 fields.
+		 */
+		xfs_bmap_init(args->flist, args->firstblock);
+		error = xfs_da_split(state);
+		if (!error) {
+			error = xfs_bmap_finish(&args->trans, args->flist,
+						&committed);
+		}
+		if (error) {
+			ASSERT(committed);
+			args->trans = NULL;
+			xfs_bmap_cancel(args->flist);
+			goto out;
+		}
+
+		/*
+		 * bmap_finish() may have committed the last trans and started
+		 * a new one.  We need the inode to be in all transactions.
+		 */
+		if (committed)
+			xfs_trans_ijoin(args->trans, dp);
+	} else {
+		/*
+		 * Addition succeeded, update Btree hashvals.
+		 */
+		xfs_da_fixhashpath(state, &state->path);
+	}
+
+	/*
+	 * Kill the state structure, we're done with it and need to
+	 * allow the buffers to come back later.
+	 */
+	xfs_da_state_free(state);
+	state = NULL;
+
+	/*
+	 * Commit the leaf addition or btree split and start the next
+	 * trans in the chain.
+	 */
+	error = xfs_trans_roll(&args->trans, dp);
+	if (error)
+		goto out;
+
+	/*
+	 * If there was an out-of-line value, allocate the blocks we
+	 * identified for its storage and copy the value.  This is done
+	 * after we create the attribute so that we don't overflow the
+	 * maximum size of a transaction and/or hit a deadlock.
+	 */
+	if (args->rmtblkno > 0) {
+		error = xfs_attr_rmtval_set(args);
+		if (error)
+			return(error);
+	}
+
+	/*
+	 * If this is an atomic rename operation, we must "flip" the
+	 * incomplete flags on the "new" and "old" attribute/value pairs
+	 * so that one disappears and one appears atomically.  Then we
+	 * must remove the "old" attribute/value pair.
+	 */
+	if (args->op_flags & XFS_DA_OP_RENAME) {
+		/*
+		 * In a separate transaction, set the incomplete flag on the
+		 * "old" attr and clear the incomplete flag on the "new" attr.
+		 */
+		error = xfs_attr_leaf_flipflags(args);
+		if (error)
+			goto out;
+
+		/*
+		 * Dismantle the "old" attribute/value pair by removing
+		 * a "remote" value (if it exists).
+		 */
+		args->index = args->index2;
+		args->blkno = args->blkno2;
+		args->rmtblkno = args->rmtblkno2;
+		args->rmtblkcnt = args->rmtblkcnt2;
+		if (args->rmtblkno) {
+			error = xfs_attr_rmtval_remove(args);
+			if (error)
+				return(error);
+		}
+
+		/*
+		 * Re-find the "old" attribute entry after any split ops.
+		 * The INCOMPLETE flag means that we will find the "old"
+		 * attr, not the "new" one.
+		 */
+		args->flags |= XFS_ATTR_INCOMPLETE;
+		state = xfs_da_state_alloc();
+		state->args = args;
+		state->mp = mp;
+		state->blocksize = state->mp->m_sb.sb_blocksize;
+		state->node_ents = state->mp->m_attr_node_ents;
+		state->inleaf = 0;
+		error = xfs_da_node_lookup_int(state, &retval);
+		if (error)
+			goto out;
+
+		/*
+		 * Remove the name and update the hashvals in the tree.
+		 */
+		blk = &state->path.blk[ state->path.active-1 ];
+		ASSERT(blk->magic == XFS_ATTR_LEAF_MAGIC);
+		error = xfs_attr_leaf_remove(blk->bp, args);
+		xfs_da_fixhashpath(state, &state->path);
+
+		/*
+		 * Check to see if the tree needs to be collapsed.
+		 */
+		if (retval && (state->path.active > 1)) {
+			xfs_bmap_init(args->flist, args->firstblock);
+			error = xfs_da_join(state);
+			if (!error) {
+				error = xfs_bmap_finish(&args->trans,
+							args->flist,
+							&committed);
+			}
+			if (error) {
+				ASSERT(committed);
+				args->trans = NULL;
+				xfs_bmap_cancel(args->flist);
+				goto out;
+			}
+
+			/*
+			 * bmap_finish() may have committed the last trans
+			 * and started a new one.  We need the inode to be
+			 * in all transactions.
+			 */
+			if (committed)
+				xfs_trans_ijoin(args->trans, dp);
+		}
+
+		/*
+		 * Commit and start the next trans in the chain.
+		 */
+		error = xfs_trans_roll(&args->trans, dp);
+		if (error)
+			goto out;
+
+	} else if (args->rmtblkno > 0) {
+		/*
+		 * Added a "remote" value, just clear the incomplete flag.
+		 */
+		error = xfs_attr_leaf_clearflag(args);
+		if (error)
+			goto out;
+	}
+	retval = error = 0;
+
+out:
+	if (state)
+		xfs_da_state_free(state);
+	if (error)
+		return(error);
+	return(retval);
+}
+
+/*
+ * Remove a name from a B-tree attribute list.
+ *
+ * This will involve walking down the Btree, and may involve joining
+ * leaf nodes and even joining intermediate nodes up to and including
+ * the root node (a special case of an intermediate node).
+ */
+STATIC int
+xfs_attr_node_removename(xfs_da_args_t *args)
+{
+	xfs_da_state_t *state;
+	xfs_da_state_blk_t *blk;
+	xfs_inode_t *dp;
+	xfs_dabuf_t *bp;
+	int retval, error, committed, forkoff;
+
+	/*
+	 * Tie a string around our finger to remind us where we are.
+	 */
+	dp = args->dp;
+	state = xfs_da_state_alloc();
+	state->args = args;
+	state->mp = dp->i_mount;
+	state->blocksize = state->mp->m_sb.sb_blocksize;
+	state->node_ents = state->mp->m_attr_node_ents;
+
+	/*
+	 * Search to see if name exists, and get back a pointer to it.
+	 */
+	error = xfs_da_node_lookup_int(state, &retval);
+	if (error || (retval != EEXIST)) {
+		if (error == 0)
+			error = retval;
+		goto out;
+	}
+
+	/*
+	 * If there is an out-of-line value, de-allocate the blocks.
+	 * This is done before we remove the attribute so that we don't
+	 * overflow the maximum size of a transaction and/or hit a deadlock.
+	 */
+	blk = &state->path.blk[ state->path.active-1 ];
+	ASSERT(blk->bp != NULL);
+	ASSERT(blk->magic == XFS_ATTR_LEAF_MAGIC);
+	if (args->rmtblkno > 0) {
+		/*
+		 * Fill in disk block numbers in the state structure
+		 * so that we can get the buffers back after we commit
+		 * several transactions in the following calls.
+		 */
+		error = xfs_attr_fillstate(state);
+		if (error)
+			goto out;
+
+		/*
+		 * Mark the attribute as INCOMPLETE, then bunmapi() the
+		 * remote value.
+		 */
+		error = xfs_attr_leaf_setflag(args);
+		if (error)
+			goto out;
+		error = xfs_attr_rmtval_remove(args);
+		if (error)
+			goto out;
+
+		/*
+		 * Refill the state structure with buffers, the prior calls
+		 * released our buffers.
+		 */
+		error = xfs_attr_refillstate(state);
+		if (error)
+			goto out;
+	}
+
+	/*
+	 * Remove the name and update the hashvals in the tree.
+	 */
+	blk = &state->path.blk[ state->path.active-1 ];
+	ASSERT(blk->magic == XFS_ATTR_LEAF_MAGIC);
+	retval = xfs_attr_leaf_remove(blk->bp, args);
+	xfs_da_fixhashpath(state, &state->path);
+
+	/*
+	 * Check to see if the tree needs to be collapsed.
+	 */
+	if (retval && (state->path.active > 1)) {
+		xfs_bmap_init(args->flist, args->firstblock);
+		error = xfs_da_join(state);
+		if (!error) {
+			error = xfs_bmap_finish(&args->trans, args->flist,
+						&committed);
+		}
+		if (error) {
+			ASSERT(committed);
+			args->trans = NULL;
+			xfs_bmap_cancel(args->flist);
+			goto out;
+		}
+
+		/*
+		 * bmap_finish() may have committed the last trans and started
+		 * a new one.  We need the inode to be in all transactions.
+		 */
+		if (committed)
+			xfs_trans_ijoin(args->trans, dp);
+
+		/*
+		 * Commit the Btree join operation and start a new trans.
+		 */
+		error = xfs_trans_roll(&args->trans, dp);
+		if (error)
+			goto out;
+	}
+
+	/*
+	 * If the result is small enough, push it all into the inode.
+	 */
+	if (xfs_bmap_one_block(dp, XFS_ATTR_FORK)) {
+		/*
+		 * Have to get rid of the copy of this dabuf in the state.
+		 */
+		ASSERT(state->path.active == 1);
+		ASSERT(state->path.blk[0].bp);
+		xfs_da_buf_done(state->path.blk[0].bp);
+		state->path.blk[0].bp = NULL;
+
+		error = xfs_da_read_buf(args->trans, args->dp, 0, -1, &bp,
+						     XFS_ATTR_FORK);
+		if (error)
+			goto out;
+		ASSERT(be16_to_cpu(((xfs_attr_leafblock_t *)
+				      bp->data)->hdr.info.magic)
+						       == XFS_ATTR_LEAF_MAGIC);
+
+		if ((forkoff = xfs_attr_shortform_allfit(bp, dp))) {
+			xfs_bmap_init(args->flist, args->firstblock);
+			error = xfs_attr_leaf_to_shortform(bp, args, forkoff);
+			/* bp is gone due to xfs_da_shrink_inode */
+			if (!error) {
+				error = xfs_bmap_finish(&args->trans,
+							args->flist,
+							&committed);
+			}
+			if (error) {
+				ASSERT(committed);
+				args->trans = NULL;
+				xfs_bmap_cancel(args->flist);
+				goto out;
+			}
+
+			/*
+			 * bmap_finish() may have committed the last trans
+			 * and started a new one.  We need the inode to be
+			 * in all transactions.
+			 */
+			if (committed)
+				xfs_trans_ijoin(args->trans, dp);
+		} else
+			xfs_da_brelse(args->trans, bp);
+	}
+	error = 0;
+
+out:
+	xfs_da_state_free(state);
+	return(error);
+}
+
+/*
+ * Fill in the disk block numbers in the state structure for the buffers
+ * that are attached to the state structure.
+ * This is done so that we can quickly reattach ourselves to those buffers
+ * after some set of transaction commits have released these buffers.
+ */
+STATIC int
+xfs_attr_fillstate(xfs_da_state_t *state)
+{
+	xfs_da_state_path_t *path;
+	xfs_da_state_blk_t *blk;
+	int level;
+
+	/*
+	 * Roll down the "path" in the state structure, storing the on-disk
+	 * block number for those buffers in the "path".
+	 */
+	path = &state->path;
+	ASSERT((path->active >= 0) && (path->active < XFS_DA_NODE_MAXDEPTH));
+	for (blk = path->blk, level = 0; level < path->active; blk++, level++) {
+		if (blk->bp) {
+			blk->disk_blkno = xfs_da_blkno(blk->bp);
+			xfs_da_buf_done(blk->bp);
+			blk->bp = NULL;
+		} else {
+			blk->disk_blkno = 0;
+		}
+	}
+
+	/*
+	 * Roll down the "altpath" in the state structure, storing the on-disk
+	 * block number for those buffers in the "altpath".
+	 */
+	path = &state->altpath;
+	ASSERT((path->active >= 0) && (path->active < XFS_DA_NODE_MAXDEPTH));
+	for (blk = path->blk, level = 0; level < path->active; blk++, level++) {
+		if (blk->bp) {
+			blk->disk_blkno = xfs_da_blkno(blk->bp);
+			xfs_da_buf_done(blk->bp);
+			blk->bp = NULL;
+		} else {
+			blk->disk_blkno = 0;
+		}
+	}
+
+	return(0);
+}
+
+/*
+ * Reattach the buffers to the state structure based on the disk block
+ * numbers stored in the state structure.
+ * This is done after some set of transaction commits have released those
+ * buffers from our grip.
+ */
+STATIC int
+xfs_attr_refillstate(xfs_da_state_t *state)
+{
+	xfs_da_state_path_t *path;
+	xfs_da_state_blk_t *blk;
+	int level, error;
+
+	/*
+	 * Roll down the "path" in the state structure, storing the on-disk
+	 * block number for those buffers in the "path".
+	 */
+	path = &state->path;
+	ASSERT((path->active >= 0) && (path->active < XFS_DA_NODE_MAXDEPTH));
+	for (blk = path->blk, level = 0; level < path->active; blk++, level++) {
+		if (blk->disk_blkno) {
+			error = xfs_da_read_buf(state->args->trans,
+						state->args->dp,
+						blk->blkno, blk->disk_blkno,
+						&blk->bp, XFS_ATTR_FORK);
+			if (error)
+				return(error);
+		} else {
+			blk->bp = NULL;
+		}
+	}
+
+	/*
+	 * Roll down the "altpath" in the state structure, storing the on-disk
+	 * block number for those buffers in the "altpath".
+	 */
+	path = &state->altpath;
+	ASSERT((path->active >= 0) && (path->active < XFS_DA_NODE_MAXDEPTH));
+	for (blk = path->blk, level = 0; level < path->active; blk++, level++) {
+		if (blk->disk_blkno) {
+			error = xfs_da_read_buf(state->args->trans,
+						state->args->dp,
+						blk->blkno, blk->disk_blkno,
+						&blk->bp, XFS_ATTR_FORK);
+			if (error)
+				return(error);
+		} else {
+			blk->bp = NULL;
+		}
+	}
+
+	return(0);
+}
+
+/*
+ * Look up a filename in a node attribute list.
+ *
+ * This routine gets called for any attribute fork that has more than one
+ * block, ie: both true Btree attr lists and for single-leaf-blocks with
+ * "remote" values taking up more blocks.
+ */
+STATIC int
+xfs_attr_node_get(xfs_da_args_t *args)
+{
+	xfs_da_state_t *state;
+	xfs_da_state_blk_t *blk;
+	int error, retval;
+	int i;
+
+	state = xfs_da_state_alloc();
+	state->args = args;
+	state->mp = args->dp->i_mount;
+	state->blocksize = state->mp->m_sb.sb_blocksize;
+	state->node_ents = state->mp->m_attr_node_ents;
+
+	/*
+	 * Search to see if name exists, and get back a pointer to it.
+	 */
+	error = xfs_da_node_lookup_int(state, &retval);
+	if (error) {
+		retval = error;
+	} else if (retval == EEXIST) {
+		blk = &state->path.blk[ state->path.active-1 ];
+		ASSERT(blk->bp != NULL);
+		ASSERT(blk->magic == XFS_ATTR_LEAF_MAGIC);
+
+		/*
+		 * Get the value, local or "remote"
+		 */
+		retval = xfs_attr_leaf_getvalue(blk->bp, args);
+		if (!retval && (args->rmtblkno > 0)
+		    && !(args->flags & ATTR_KERNOVAL)) {
+			retval = xfs_attr_rmtval_get(args);
+		}
+	}
+
+	/*
+	 * If not in a transaction, we have to release all the buffers.
+	 */
+	for (i = 0; i < state->path.active; i++) {
+		xfs_da_brelse(args->trans, state->path.blk[i].bp);
+		state->path.blk[i].bp = NULL;
+	}
+
+	xfs_da_state_free(state);
+	return(retval);
+}
+
+STATIC int							/* error */
+xfs_attr_node_list(xfs_attr_list_context_t *context)
+{
+	attrlist_cursor_kern_t *cursor;
+	xfs_attr_leafblock_t *leaf;
+	xfs_da_intnode_t *node;
+	xfs_da_node_entry_t *btree;
+	int error, i;
+	xfs_dabuf_t *bp;
+
+	cursor = context->cursor;
+	cursor->initted = 1;
+
+	/*
+	 * Do all sorts of validation on the passed-in cursor structure.
+	 * If anything is amiss, ignore the cursor and look up the hashval
+	 * starting from the btree root.
+	 */
+	bp = NULL;
+	if (cursor->blkno > 0) {
+		error = xfs_da_read_buf(NULL, context->dp, cursor->blkno, -1,
+					      &bp, XFS_ATTR_FORK);
+		if ((error != 0) && (error != EFSCORRUPTED))
+			return(error);
+		if (bp) {
+			node = bp->data;
+			switch (be16_to_cpu(node->hdr.info.magic)) {
+			case XFS_DA_NODE_MAGIC:
+				trace_xfs_attr_list_wrong_blk(context);
+				xfs_da_brelse(NULL, bp);
+				bp = NULL;
+				break;
+			case XFS_ATTR_LEAF_MAGIC:
+				leaf = bp->data;
+				if (cursor->hashval > be32_to_cpu(leaf->entries[
+				    be16_to_cpu(leaf->hdr.count)-1].hashval)) {
+					trace_xfs_attr_list_wrong_blk(context);
+					xfs_da_brelse(NULL, bp);
+					bp = NULL;
+				} else if (cursor->hashval <=
+					     be32_to_cpu(leaf->entries[0].hashval)) {
+					trace_xfs_attr_list_wrong_blk(context);
+					xfs_da_brelse(NULL, bp);
+					bp = NULL;
+				}
+				break;
+			default:
+				trace_xfs_attr_list_wrong_blk(context);
+				xfs_da_brelse(NULL, bp);
+				bp = NULL;
+			}
+		}
+	}
+
+	/*
+	 * We did not find what we expected given the cursor's contents,
+	 * so we start from the top and work down based on the hash value.
+	 * Note that start of node block is same as start of leaf block.
+	 */
+	if (bp == NULL) {
+		cursor->blkno = 0;
+		for (;;) {
+			error = xfs_da_read_buf(NULL, context->dp,
+						      cursor->blkno, -1, &bp,
+						      XFS_ATTR_FORK);
+			if (error)
+				return(error);
+			if (unlikely(bp == NULL)) {
+				XFS_ERROR_REPORT("xfs_attr_node_list(2)",
+						 XFS_ERRLEVEL_LOW,
+						 context->dp->i_mount);
+				return(XFS_ERROR(EFSCORRUPTED));
+			}
+			node = bp->data;
+			if (be16_to_cpu(node->hdr.info.magic)
+							== XFS_ATTR_LEAF_MAGIC)
+				break;
+			if (unlikely(be16_to_cpu(node->hdr.info.magic)
+							!= XFS_DA_NODE_MAGIC)) {
+				XFS_CORRUPTION_ERROR("xfs_attr_node_list(3)",
+						     XFS_ERRLEVEL_LOW,
+						     context->dp->i_mount,
+						     node);
+				xfs_da_brelse(NULL, bp);
+				return(XFS_ERROR(EFSCORRUPTED));
+			}
+			btree = node->btree;
+			for (i = 0; i < be16_to_cpu(node->hdr.count);
+								btree++, i++) {
+				if (cursor->hashval
+						<= be32_to_cpu(btree->hashval)) {
+					cursor->blkno = be32_to_cpu(btree->before);
+					trace_xfs_attr_list_node_descend(context,
+									 btree);
+					break;
+				}
+			}
+			if (i == be16_to_cpu(node->hdr.count)) {
+				xfs_da_brelse(NULL, bp);
+				return(0);
+			}
+			xfs_da_brelse(NULL, bp);
+		}
+	}
+	ASSERT(bp != NULL);
+
+	/*
+	 * Roll upward through the blocks, processing each leaf block in
+	 * order.  As long as there is space in the result buffer, keep
+	 * adding the information.
+	 */
+	for (;;) {
+		leaf = bp->data;
+		if (unlikely(be16_to_cpu(leaf->hdr.info.magic)
+						!= XFS_ATTR_LEAF_MAGIC)) {
+			XFS_CORRUPTION_ERROR("xfs_attr_node_list(4)",
+					     XFS_ERRLEVEL_LOW,
+					     context->dp->i_mount, leaf);
+			xfs_da_brelse(NULL, bp);
+			return(XFS_ERROR(EFSCORRUPTED));
+		}
+		error = xfs_attr_leaf_list_int(bp, context);
+		if (error) {
+			xfs_da_brelse(NULL, bp);
+			return error;
+		}
+		if (context->seen_enough || leaf->hdr.info.forw == 0)
+			break;
+		cursor->blkno = be32_to_cpu(leaf->hdr.info.forw);
+		xfs_da_brelse(NULL, bp);
+		error = xfs_da_read_buf(NULL, context->dp, cursor->blkno, -1,
+					      &bp, XFS_ATTR_FORK);
+		if (error)
+			return(error);
+		if (unlikely((bp == NULL))) {
+			XFS_ERROR_REPORT("xfs_attr_node_list(5)",
+					 XFS_ERRLEVEL_LOW,
+					 context->dp->i_mount);
+			return(XFS_ERROR(EFSCORRUPTED));
+		}
+	}
+	xfs_da_brelse(NULL, bp);
+	return(0);
+}
+
+
+/*========================================================================
+ * External routines for manipulating out-of-line attribute values.
+ *========================================================================*/
+
+/*
+ * Read the value associated with an attribute from the out-of-line buffer
+ * that we stored it in.
+ */
+int
+xfs_attr_rmtval_get(xfs_da_args_t *args)
+{
+	xfs_bmbt_irec_t map[ATTR_RMTVALUE_MAPSIZE];
+	xfs_mount_t *mp;
+	xfs_daddr_t dblkno;
+	void *dst;
+	xfs_buf_t *bp;
+	int nmap, error, tmp, valuelen, blkcnt, i;
+	xfs_dablk_t lblkno;
+
+	ASSERT(!(args->flags & ATTR_KERNOVAL));
+
+	mp = args->dp->i_mount;
+	dst = args->value;
+	valuelen = args->valuelen;
+	lblkno = args->rmtblkno;
+	while (valuelen > 0) {
+		nmap = ATTR_RMTVALUE_MAPSIZE;
+		error = xfs_bmapi(args->trans, args->dp, (xfs_fileoff_t)lblkno,
+				  args->rmtblkcnt,
+				  XFS_BMAPI_ATTRFORK | XFS_BMAPI_METADATA,
+				  NULL, 0, map, &nmap, NULL);
+		if (error)
+			return(error);
+		ASSERT(nmap >= 1);
+
+		for (i = 0; (i < nmap) && (valuelen > 0); i++) {
+			ASSERT((map[i].br_startblock != DELAYSTARTBLOCK) &&
+			       (map[i].br_startblock != HOLESTARTBLOCK));
+			dblkno = XFS_FSB_TO_DADDR(mp, map[i].br_startblock);
+			blkcnt = XFS_FSB_TO_BB(mp, map[i].br_blockcount);
+			error = xfs_read_buf(mp, mp->m_ddev_targp, dblkno,
+					     blkcnt, XBF_LOCK | XBF_DONT_BLOCK,
+					     &bp);
+			if (error)
+				return(error);
+
+			tmp = (valuelen < XFS_BUF_SIZE(bp))
+				? valuelen : XFS_BUF_SIZE(bp);
+			xfs_buf_iomove(bp, 0, tmp, dst, XBRW_READ);
+			xfs_buf_relse(bp);
+			dst += tmp;
+			valuelen -= tmp;
+
+			lblkno += map[i].br_blockcount;
+		}
+	}
+	ASSERT(valuelen == 0);
+	return(0);
+}
+
+/*
+ * Write the value associated with an attribute into the out-of-line buffer
+ * that we have defined for it.
+ */
+STATIC int
+xfs_attr_rmtval_set(xfs_da_args_t *args)
+{
+	xfs_mount_t *mp;
+	xfs_fileoff_t lfileoff;
+	xfs_inode_t *dp;
+	xfs_bmbt_irec_t map;
+	xfs_daddr_t dblkno;
+	void *src;
+	xfs_buf_t *bp;
+	xfs_dablk_t lblkno;
+	int blkcnt, valuelen, nmap, error, tmp, committed;
+
+	dp = args->dp;
+	mp = dp->i_mount;
+	src = args->value;
+
+	/*
+	 * Find a "hole" in the attribute address space large enough for
+	 * us to drop the new attribute's value into.
+	 */
+	blkcnt = XFS_B_TO_FSB(mp, args->valuelen);
+	lfileoff = 0;
+	error = xfs_bmap_first_unused(args->trans, args->dp, blkcnt, &lfileoff,
+						   XFS_ATTR_FORK);
+	if (error) {
+		return(error);
+	}
+	args->rmtblkno = lblkno = (xfs_dablk_t)lfileoff;
+	args->rmtblkcnt = blkcnt;
+
+	/*
+	 * Roll through the "value", allocating blocks on disk as required.
+	 */
+	while (blkcnt > 0) {
+		/*
+		 * Allocate a single extent, up to the size of the value.
+		 */
+		xfs_bmap_init(args->flist, args->firstblock);
+		nmap = 1;
+		error = xfs_bmapi(args->trans, dp, (xfs_fileoff_t)lblkno,
+				  blkcnt,
+				  XFS_BMAPI_ATTRFORK | XFS_BMAPI_METADATA |
+							XFS_BMAPI_WRITE,
+				  args->firstblock, args->total, &map, &nmap,
+				  args->flist);
+		if (!error) {
+			error = xfs_bmap_finish(&args->trans, args->flist,
+						&committed);
+		}
+		if (error) {
+			ASSERT(committed);
+			args->trans = NULL;
+			xfs_bmap_cancel(args->flist);
+			return(error);
+		}
+
+		/*
+		 * bmap_finish() may have committed the last trans and started
+		 * a new one.  We need the inode to be in all transactions.
+		 */
+		if (committed)
+			xfs_trans_ijoin(args->trans, dp);
+
+		ASSERT(nmap == 1);
+		ASSERT((map.br_startblock != DELAYSTARTBLOCK) &&
+		       (map.br_startblock != HOLESTARTBLOCK));
+		lblkno += map.br_blockcount;
+		blkcnt -= map.br_blockcount;
+
+		/*
+		 * Start the next trans in the chain.
+		 */
+		error = xfs_trans_roll(&args->trans, dp);
+		if (error)
+			return (error);
+	}
+
+	/*
+	 * Roll through the "value", copying the attribute value to the
+	 * already-allocated blocks.  Blocks are written synchronously
+	 * so that we can know they are all on disk before we turn off
+	 * the INCOMPLETE flag.
+	 */
+	lblkno = args->rmtblkno;
+	valuelen = args->valuelen;
+	while (valuelen > 0) {
+		/*
+		 * Try to remember where we decided to put the value.
+		 */
+		xfs_bmap_init(args->flist, args->firstblock);
+		nmap = 1;
+		error = xfs_bmapi(NULL, dp, (xfs_fileoff_t)lblkno,
+				  args->rmtblkcnt,
+				  XFS_BMAPI_ATTRFORK | XFS_BMAPI_METADATA,
+				  args->firstblock, 0, &map, &nmap,
+				  NULL);
+		if (error) {
+			return(error);
+		}
+		ASSERT(nmap == 1);
+		ASSERT((map.br_startblock != DELAYSTARTBLOCK) &&
+		       (map.br_startblock != HOLESTARTBLOCK));
+
+		dblkno = XFS_FSB_TO_DADDR(mp, map.br_startblock),
+		blkcnt = XFS_FSB_TO_BB(mp, map.br_blockcount);
+
+		bp = xfs_buf_get(mp->m_ddev_targp, dblkno, blkcnt,
+				 XBF_LOCK | XBF_DONT_BLOCK);
+		ASSERT(bp);
+		ASSERT(!XFS_BUF_GETERROR(bp));
+
+		tmp = (valuelen < XFS_BUF_SIZE(bp)) ? valuelen :
+							XFS_BUF_SIZE(bp);
+		xfs_buf_iomove(bp, 0, tmp, src, XBRW_WRITE);
+		if (tmp < XFS_BUF_SIZE(bp))
+			xfs_buf_zero(bp, tmp, XFS_BUF_SIZE(bp) - tmp);
+		if ((error = xfs_bwrite(mp, bp))) {/* GROT: NOTE: synchronous write */
+			return (error);
+		}
+		src += tmp;
+		valuelen -= tmp;
+
+		lblkno += map.br_blockcount;
+	}
+	ASSERT(valuelen == 0);
+	return(0);
+}
+
+/*
+ * Remove the value associated with an attribute by deleting the
+ * out-of-line buffer that it is stored on.
+ */
+STATIC int
+xfs_attr_rmtval_remove(xfs_da_args_t *args)
+{
+	xfs_mount_t *mp;
+	xfs_bmbt_irec_t map;
+	xfs_buf_t *bp;
+	xfs_daddr_t dblkno;
+	xfs_dablk_t lblkno;
+	int valuelen, blkcnt, nmap, error, done, committed;
+
+	mp = args->dp->i_mount;
+
+	/*
+	 * Roll through the "value", invalidating the attribute value's
+	 * blocks.
+	 */
+	lblkno = args->rmtblkno;
+	valuelen = args->rmtblkcnt;
+	while (valuelen > 0) {
+		/*
+		 * Try to remember where we decided to put the value.
+		 */
+		xfs_bmap_init(args->flist, args->firstblock);
+		nmap = 1;
+		error = xfs_bmapi(NULL, args->dp, (xfs_fileoff_t)lblkno,
+					args->rmtblkcnt,
+					XFS_BMAPI_ATTRFORK | XFS_BMAPI_METADATA,
+					args->firstblock, 0, &map, &nmap,
+					args->flist);
+		if (error) {
+			return(error);
+		}
+		ASSERT(nmap == 1);
+		ASSERT((map.br_startblock != DELAYSTARTBLOCK) &&
+		       (map.br_startblock != HOLESTARTBLOCK));
+
+		dblkno = XFS_FSB_TO_DADDR(mp, map.br_startblock),
+		blkcnt = XFS_FSB_TO_BB(mp, map.br_blockcount);
+
+		/*
+		 * If the "remote" value is in the cache, remove it.
+		 */
+		bp = xfs_incore(mp->m_ddev_targp, dblkno, blkcnt, XBF_TRYLOCK);
+		if (bp) {
+			XFS_BUF_STALE(bp);
+			XFS_BUF_UNDELAYWRITE(bp);
+			xfs_buf_relse(bp);
+			bp = NULL;
+		}
+
+		valuelen -= map.br_blockcount;
+
+		lblkno += map.br_blockcount;
+	}
+
+	/*
+	 * Keep de-allocating extents until the remote-value region is gone.
+	 */
+	lblkno = args->rmtblkno;
+	blkcnt = args->rmtblkcnt;
+	done = 0;
+	while (!done) {
+		xfs_bmap_init(args->flist, args->firstblock);
+		error = xfs_bunmapi(args->trans, args->dp, lblkno, blkcnt,
+				    XFS_BMAPI_ATTRFORK | XFS_BMAPI_METADATA,
+				    1, args->firstblock, args->flist,
+				    &done);
+		if (!error) {
+			error = xfs_bmap_finish(&args->trans, args->flist,
+						&committed);
+		}
+		if (error) {
+			ASSERT(committed);
+			args->trans = NULL;
+			xfs_bmap_cancel(args->flist);
+			return(error);
+		}
+
+		/*
+		 * bmap_finish() may have committed the last trans and started
+		 * a new one.  We need the inode to be in all transactions.
+		 */
+		if (committed)
+			xfs_trans_ijoin(args->trans, args->dp);
+
+		/*
+		 * Close out trans and start the next one in the chain.
+		 */
+		error = xfs_trans_roll(&args->trans, args->dp);
+		if (error)
+			return (error);
+	}
+	return(0);
+}
diff --git a/fs/xfs/xfs_attr.h b/fs/xfs/xfs_attr.h
new file mode 100644
index 00000000..e920d68e
--- /dev/null
+++ b/fs/xfs/xfs_attr.h
@@ -0,0 +1,146 @@
+/*
+ * Copyright (c) 2000,2002-2003,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_ATTR_H__
+#define	__XFS_ATTR_H__
+
+struct xfs_inode;
+struct xfs_da_args;
+struct xfs_attr_list_context;
+
+/*
+ * Large attribute lists are structured around Btrees where all the data
+ * elements are in the leaf nodes.  Attribute names are hashed into an int,
+ * then that int is used as the index into the Btree.  Since the hashval
+ * of an attribute name may not be unique, we may have duplicate keys.
+ * The internal links in the Btree are logical block offsets into the file.
+ *
+ * Small attribute lists use a different format and are packed as tightly
+ * as possible so as to fit into the literal area of the inode.
+ */
+
+/*========================================================================
+ * External interfaces
+ *========================================================================*/
+
+
+#define ATTR_DONTFOLLOW	0x0001	/* -- unused, from IRIX -- */
+#define ATTR_ROOT	0x0002	/* use attrs in root (trusted) namespace */
+#define ATTR_TRUST	0x0004	/* -- unused, from IRIX -- */
+#define ATTR_SECURE	0x0008	/* use attrs in security namespace */
+#define ATTR_CREATE	0x0010	/* pure create: fail if attr already exists */
+#define ATTR_REPLACE	0x0020	/* pure set: fail if attr does not exist */
+
+#define ATTR_KERNOTIME	0x1000	/* [kernel] don't update inode timestamps */
+#define ATTR_KERNOVAL	0x2000	/* [kernel] get attr size only, not value */
+
+#define XFS_ATTR_FLAGS \
+	{ ATTR_DONTFOLLOW, 	"DONTFOLLOW" }, \
+	{ ATTR_ROOT,		"ROOT" }, \
+	{ ATTR_TRUST,		"TRUST" }, \
+	{ ATTR_SECURE,		"SECURE" }, \
+	{ ATTR_CREATE,		"CREATE" }, \
+	{ ATTR_REPLACE,		"REPLACE" }, \
+	{ ATTR_KERNOTIME,	"KERNOTIME" }, \
+	{ ATTR_KERNOVAL,	"KERNOVAL" }
+
+/*
+ * The maximum size (into the kernel or returned from the kernel) of an
+ * attribute value or the buffer used for an attr_list() call.  Larger
+ * sizes will result in an ERANGE return code.
+ */
+#define	ATTR_MAX_VALUELEN	(64*1024)	/* max length of a value */
+
+/*
+ * Define how lists of attribute names are returned to the user from
+ * the attr_list() call.  A large, 32bit aligned, buffer is passed in
+ * along with its size.  We put an array of offsets at the top that each
+ * reference an attrlist_ent_t and pack the attrlist_ent_t's at the bottom.
+ */
+typedef struct attrlist {
+	__s32	al_count;	/* number of entries in attrlist */
+	__s32	al_more;	/* T/F: more attrs (do call again) */
+	__s32	al_offset[1];	/* byte offsets of attrs [var-sized] */
+} attrlist_t;
+
+/*
+ * Show the interesting info about one attribute.  This is what the
+ * al_offset[i] entry points to.
+ */
+typedef struct attrlist_ent {	/* data from attr_list() */
+	__u32	a_valuelen;	/* number bytes in value of attr */
+	char	a_name[1];	/* attr name (NULL terminated) */
+} attrlist_ent_t;
+
+/*
+ * Given a pointer to the (char*) buffer containing the attr_list() result,
+ * and an index, return a pointer to the indicated attribute in the buffer.
+ */
+#define	ATTR_ENTRY(buffer, index)		\
+	((attrlist_ent_t *)			\
+	 &((char *)buffer)[ ((attrlist_t *)(buffer))->al_offset[index] ])
+
+/*
+ * Kernel-internal version of the attrlist cursor.
+ */
+typedef struct attrlist_cursor_kern {
+	__u32	hashval;	/* hash value of next entry to add */
+	__u32	blkno;		/* block containing entry (suggestion) */
+	__u32	offset;		/* offset in list of equal-hashvals */
+	__u16	pad1;		/* padding to match user-level */
+	__u8	pad2;		/* padding to match user-level */
+	__u8	initted;	/* T/F: cursor has been initialized */
+} attrlist_cursor_kern_t;
+
+
+/*========================================================================
+ * Structure used to pass context around among the routines.
+ *========================================================================*/
+
+
+typedef int (*put_listent_func_t)(struct xfs_attr_list_context *, int,
+			      unsigned char *, int, int, unsigned char *);
+
+typedef struct xfs_attr_list_context {
+	struct xfs_inode		*dp;		/* inode */
+	struct attrlist_cursor_kern	*cursor;	/* position in list */
+	char				*alist;		/* output buffer */
+	int				seen_enough;	/* T/F: seen enough of list? */
+	ssize_t				count;		/* num used entries */
+	int				dupcnt;		/* count dup hashvals seen */
+	int				bufsize;	/* total buffer size */
+	int				firstu;		/* first used byte in buffer */
+	int				flags;		/* from VOP call */
+	int				resynch;	/* T/F: resynch with cursor */
+	int				put_value;	/* T/F: need value for listent */
+	put_listent_func_t		put_listent;	/* list output fmt function */
+	int				index;		/* index into output buffer */
+} xfs_attr_list_context_t;
+
+
+/*========================================================================
+ * Function prototypes for the kernel.
+ *========================================================================*/
+
+/*
+ * Overall external interface routines.
+ */
+int xfs_attr_inactive(struct xfs_inode *dp);
+int xfs_attr_rmtval_get(struct xfs_da_args *args);
+int xfs_attr_list_int(struct xfs_attr_list_context *);
+
+#endif	/* __XFS_ATTR_H__ */
diff --git a/fs/xfs/xfs_attr_leaf.c b/fs/xfs/xfs_attr_leaf.c
new file mode 100644
index 00000000..f49ecf2e
--- /dev/null
+++ b/fs/xfs/xfs_attr_leaf.c
@@ -0,0 +1,2979 @@
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_mount.h"
+#include "xfs_da_btree.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_alloc_btree.h"
+#include "xfs_ialloc_btree.h"
+#include "xfs_alloc.h"
+#include "xfs_btree.h"
+#include "xfs_attr_sf.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_inode_item.h"
+#include "xfs_bmap.h"
+#include "xfs_attr.h"
+#include "xfs_attr_leaf.h"
+#include "xfs_error.h"
+#include "xfs_trace.h"
+
+/*
+ * xfs_attr_leaf.c
+ *
+ * Routines to implement leaf blocks of attributes as Btrees of hashed names.
+ */
+
+/*========================================================================
+ * Function prototypes for the kernel.
+ *========================================================================*/
+
+/*
+ * Routines used for growing the Btree.
+ */
+STATIC int xfs_attr_leaf_create(xfs_da_args_t *args, xfs_dablk_t which_block,
+				    xfs_dabuf_t **bpp);
+STATIC int xfs_attr_leaf_add_work(xfs_dabuf_t *leaf_buffer, xfs_da_args_t *args,
+					      int freemap_index);
+STATIC void xfs_attr_leaf_compact(xfs_trans_t *trans, xfs_dabuf_t *leaf_buffer);
+STATIC void xfs_attr_leaf_rebalance(xfs_da_state_t *state,
+						   xfs_da_state_blk_t *blk1,
+						   xfs_da_state_blk_t *blk2);
+STATIC int xfs_attr_leaf_figure_balance(xfs_da_state_t *state,
+					   xfs_da_state_blk_t *leaf_blk_1,
+					   xfs_da_state_blk_t *leaf_blk_2,
+					   int *number_entries_in_blk1,
+					   int *number_usedbytes_in_blk1);
+
+/*
+ * Routines used for shrinking the Btree.
+ */
+STATIC int xfs_attr_node_inactive(xfs_trans_t **trans, xfs_inode_t *dp,
+				  xfs_dabuf_t *bp, int level);
+STATIC int xfs_attr_leaf_inactive(xfs_trans_t **trans, xfs_inode_t *dp,
+				  xfs_dabuf_t *bp);
+STATIC int xfs_attr_leaf_freextent(xfs_trans_t **trans, xfs_inode_t *dp,
+				   xfs_dablk_t blkno, int blkcnt);
+
+/*
+ * Utility routines.
+ */
+STATIC void xfs_attr_leaf_moveents(xfs_attr_leafblock_t *src_leaf,
+					 int src_start,
+					 xfs_attr_leafblock_t *dst_leaf,
+					 int dst_start, int move_count,
+					 xfs_mount_t *mp);
+STATIC int xfs_attr_leaf_entsize(xfs_attr_leafblock_t *leaf, int index);
+
+/*========================================================================
+ * Namespace helper routines
+ *========================================================================*/
+
+/*
+ * If namespace bits don't match return 0.
+ * If all match then return 1.
+ */
+STATIC int
+xfs_attr_namesp_match(int arg_flags, int ondisk_flags)
+{
+	return XFS_ATTR_NSP_ONDISK(ondisk_flags) == XFS_ATTR_NSP_ARGS_TO_ONDISK(arg_flags);
+}
+
+
+/*========================================================================
+ * External routines when attribute fork size < XFS_LITINO(mp).
+ *========================================================================*/
+
+/*
+ * Query whether the requested number of additional bytes of extended
+ * attribute space will be able to fit inline.
+ *
+ * Returns zero if not, else the di_forkoff fork offset to be used in the
+ * literal area for attribute data once the new bytes have been added.
+ *
+ * di_forkoff must be 8 byte aligned, hence is stored as a >>3 value;
+ * special case for dev/uuid inodes, they have fixed size data forks.
+ */
+int
+xfs_attr_shortform_bytesfit(xfs_inode_t *dp, int bytes)
+{
+	int offset;
+	int minforkoff;	/* lower limit on valid forkoff locations */
+	int maxforkoff;	/* upper limit on valid forkoff locations */
+	int dsize;
+	xfs_mount_t *mp = dp->i_mount;
+
+	offset = (XFS_LITINO(mp) - bytes) >> 3; /* rounded down */
+
+	switch (dp->i_d.di_format) {
+	case XFS_DINODE_FMT_DEV:
+		minforkoff = roundup(sizeof(xfs_dev_t), 8) >> 3;
+		return (offset >= minforkoff) ? minforkoff : 0;
+	case XFS_DINODE_FMT_UUID:
+		minforkoff = roundup(sizeof(uuid_t), 8) >> 3;
+		return (offset >= minforkoff) ? minforkoff : 0;
+	}
+
+	/*
+	 * If the requested numbers of bytes is smaller or equal to the
+	 * current attribute fork size we can always proceed.
+	 *
+	 * Note that if_bytes in the data fork might actually be larger than
+	 * the current data fork size is due to delalloc extents. In that
+	 * case either the extent count will go down when they are converted
+	 * to real extents, or the delalloc conversion will take care of the
+	 * literal area rebalancing.
+	 */
+	if (bytes <= XFS_IFORK_ASIZE(dp))
+		return dp->i_d.di_forkoff;
+
+	/*
+	 * For attr2 we can try to move the forkoff if there is space in the
+	 * literal area, but for the old format we are done if there is no
+	 * space in the fixed attribute fork.
+	 */
+	if (!(mp->m_flags & XFS_MOUNT_ATTR2))
+		return 0;
+
+	dsize = dp->i_df.if_bytes;
+
+	switch (dp->i_d.di_format) {
+	case XFS_DINODE_FMT_EXTENTS:
+		/*
+		 * If there is no attr fork and the data fork is extents, 
+		 * determine if creating the default attr fork will result
+		 * in the extents form migrating to btree. If so, the
+		 * minimum offset only needs to be the space required for
+		 * the btree root.
+		 */
+		if (!dp->i_d.di_forkoff && dp->i_df.if_bytes >
+		    xfs_default_attroffset(dp))
+			dsize = XFS_BMDR_SPACE_CALC(MINDBTPTRS);
+		break;
+	case XFS_DINODE_FMT_BTREE:
+		/*
+		 * If we have a data btree then keep forkoff if we have one,
+		 * otherwise we are adding a new attr, so then we set
+		 * minforkoff to where the btree root can finish so we have
+		 * plenty of room for attrs
+		 */
+		if (dp->i_d.di_forkoff) {
+			if (offset < dp->i_d.di_forkoff)
+				return 0;
+			return dp->i_d.di_forkoff;
+		}
+		dsize = XFS_BMAP_BROOT_SPACE(dp->i_df.if_broot);
+		break;
+	}
+
+	/*
+	 * A data fork btree root must have space for at least
+	 * MINDBTPTRS key/ptr pairs if the data fork is small or empty.
+	 */
+	minforkoff = MAX(dsize, XFS_BMDR_SPACE_CALC(MINDBTPTRS));
+	minforkoff = roundup(minforkoff, 8) >> 3;
+
+	/* attr fork btree root can have at least this many key/ptr pairs */
+	maxforkoff = XFS_LITINO(mp) - XFS_BMDR_SPACE_CALC(MINABTPTRS);
+	maxforkoff = maxforkoff >> 3;	/* rounded down */
+
+	if (offset >= maxforkoff)
+		return maxforkoff;
+	if (offset >= minforkoff)
+		return offset;
+	return 0;
+}
+
+/*
+ * Switch on the ATTR2 superblock bit (implies also FEATURES2)
+ */
+STATIC void
+xfs_sbversion_add_attr2(xfs_mount_t *mp, xfs_trans_t *tp)
+{
+	if ((mp->m_flags & XFS_MOUNT_ATTR2) &&
+	    !(xfs_sb_version_hasattr2(&mp->m_sb))) {
+		spin_lock(&mp->m_sb_lock);
+		if (!xfs_sb_version_hasattr2(&mp->m_sb)) {
+			xfs_sb_version_addattr2(&mp->m_sb);
+			spin_unlock(&mp->m_sb_lock);
+			xfs_mod_sb(tp, XFS_SB_VERSIONNUM | XFS_SB_FEATURES2);
+		} else
+			spin_unlock(&mp->m_sb_lock);
+	}
+}
+
+/*
+ * Create the initial contents of a shortform attribute list.
+ */
+void
+xfs_attr_shortform_create(xfs_da_args_t *args)
+{
+	xfs_attr_sf_hdr_t *hdr;
+	xfs_inode_t *dp;
+	xfs_ifork_t *ifp;
+
+	dp = args->dp;
+	ASSERT(dp != NULL);
+	ifp = dp->i_afp;
+	ASSERT(ifp != NULL);
+	ASSERT(ifp->if_bytes == 0);
+	if (dp->i_d.di_aformat == XFS_DINODE_FMT_EXTENTS) {
+		ifp->if_flags &= ~XFS_IFEXTENTS;	/* just in case */
+		dp->i_d.di_aformat = XFS_DINODE_FMT_LOCAL;
+		ifp->if_flags |= XFS_IFINLINE;
+	} else {
+		ASSERT(ifp->if_flags & XFS_IFINLINE);
+	}
+	xfs_idata_realloc(dp, sizeof(*hdr), XFS_ATTR_FORK);
+	hdr = (xfs_attr_sf_hdr_t *)ifp->if_u1.if_data;
+	hdr->count = 0;
+	hdr->totsize = cpu_to_be16(sizeof(*hdr));
+	xfs_trans_log_inode(args->trans, dp, XFS_ILOG_CORE | XFS_ILOG_ADATA);
+}
+
+/*
+ * Add a name/value pair to the shortform attribute list.
+ * Overflow from the inode has already been checked for.
+ */
+void
+xfs_attr_shortform_add(xfs_da_args_t *args, int forkoff)
+{
+	xfs_attr_shortform_t *sf;
+	xfs_attr_sf_entry_t *sfe;
+	int i, offset, size;
+	xfs_mount_t *mp;
+	xfs_inode_t *dp;
+	xfs_ifork_t *ifp;
+
+	dp = args->dp;
+	mp = dp->i_mount;
+	dp->i_d.di_forkoff = forkoff;
+	dp->i_df.if_ext_max =
+		XFS_IFORK_DSIZE(dp) / (uint)sizeof(xfs_bmbt_rec_t);
+	dp->i_afp->if_ext_max =
+		XFS_IFORK_ASIZE(dp) / (uint)sizeof(xfs_bmbt_rec_t);
+
+	ifp = dp->i_afp;
+	ASSERT(ifp->if_flags & XFS_IFINLINE);
+	sf = (xfs_attr_shortform_t *)ifp->if_u1.if_data;
+	sfe = &sf->list[0];
+	for (i = 0; i < sf->hdr.count; sfe = XFS_ATTR_SF_NEXTENTRY(sfe), i++) {
+#ifdef DEBUG
+		if (sfe->namelen != args->namelen)
+			continue;
+		if (memcmp(args->name, sfe->nameval, args->namelen) != 0)
+			continue;
+		if (!xfs_attr_namesp_match(args->flags, sfe->flags))
+			continue;
+		ASSERT(0);
+#endif
+	}
+
+	offset = (char *)sfe - (char *)sf;
+	size = XFS_ATTR_SF_ENTSIZE_BYNAME(args->namelen, args->valuelen);
+	xfs_idata_realloc(dp, size, XFS_ATTR_FORK);
+	sf = (xfs_attr_shortform_t *)ifp->if_u1.if_data;
+	sfe = (xfs_attr_sf_entry_t *)((char *)sf + offset);
+
+	sfe->namelen = args->namelen;
+	sfe->valuelen = args->valuelen;
+	sfe->flags = XFS_ATTR_NSP_ARGS_TO_ONDISK(args->flags);
+	memcpy(sfe->nameval, args->name, args->namelen);
+	memcpy(&sfe->nameval[args->namelen], args->value, args->valuelen);
+	sf->hdr.count++;
+	be16_add_cpu(&sf->hdr.totsize, size);
+	xfs_trans_log_inode(args->trans, dp, XFS_ILOG_CORE | XFS_ILOG_ADATA);
+
+	xfs_sbversion_add_attr2(mp, args->trans);
+}
+
+/*
+ * After the last attribute is removed revert to original inode format,
+ * making all literal area available to the data fork once more.
+ */
+STATIC void
+xfs_attr_fork_reset(
+	struct xfs_inode	*ip,
+	struct xfs_trans	*tp)
+{
+	xfs_idestroy_fork(ip, XFS_ATTR_FORK);
+	ip->i_d.di_forkoff = 0;
+	ip->i_d.di_aformat = XFS_DINODE_FMT_EXTENTS;
+
+	ASSERT(ip->i_d.di_anextents == 0);
+	ASSERT(ip->i_afp == NULL);
+
+	ip->i_df.if_ext_max = XFS_IFORK_DSIZE(ip) / sizeof(xfs_bmbt_rec_t);
+	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
+}
+
+/*
+ * Remove an attribute from the shortform attribute list structure.
+ */
+int
+xfs_attr_shortform_remove(xfs_da_args_t *args)
+{
+	xfs_attr_shortform_t *sf;
+	xfs_attr_sf_entry_t *sfe;
+	int base, size=0, end, totsize, i;
+	xfs_mount_t *mp;
+	xfs_inode_t *dp;
+
+	dp = args->dp;
+	mp = dp->i_mount;
+	base = sizeof(xfs_attr_sf_hdr_t);
+	sf = (xfs_attr_shortform_t *)dp->i_afp->if_u1.if_data;
+	sfe = &sf->list[0];
+	end = sf->hdr.count;
+	for (i = 0; i < end; sfe = XFS_ATTR_SF_NEXTENTRY(sfe),
+					base += size, i++) {
+		size = XFS_ATTR_SF_ENTSIZE(sfe);
+		if (sfe->namelen != args->namelen)
+			continue;
+		if (memcmp(sfe->nameval, args->name, args->namelen) != 0)
+			continue;
+		if (!xfs_attr_namesp_match(args->flags, sfe->flags))
+			continue;
+		break;
+	}
+	if (i == end)
+		return(XFS_ERROR(ENOATTR));
+
+	/*
+	 * Fix up the attribute fork data, covering the hole
+	 */
+	end = base + size;
+	totsize = be16_to_cpu(sf->hdr.totsize);
+	if (end != totsize)
+		memmove(&((char *)sf)[base], &((char *)sf)[end], totsize - end);
+	sf->hdr.count--;
+	be16_add_cpu(&sf->hdr.totsize, -size);
+
+	/*
+	 * Fix up the start offset of the attribute fork
+	 */
+	totsize -= size;
+	if (totsize == sizeof(xfs_attr_sf_hdr_t) &&
+	    (mp->m_flags & XFS_MOUNT_ATTR2) &&
+	    (dp->i_d.di_format != XFS_DINODE_FMT_BTREE) &&
+	    !(args->op_flags & XFS_DA_OP_ADDNAME)) {
+		xfs_attr_fork_reset(dp, args->trans);
+	} else {
+		xfs_idata_realloc(dp, -size, XFS_ATTR_FORK);
+		dp->i_d.di_forkoff = xfs_attr_shortform_bytesfit(dp, totsize);
+		ASSERT(dp->i_d.di_forkoff);
+		ASSERT(totsize > sizeof(xfs_attr_sf_hdr_t) ||
+				(args->op_flags & XFS_DA_OP_ADDNAME) ||
+				!(mp->m_flags & XFS_MOUNT_ATTR2) ||
+				dp->i_d.di_format == XFS_DINODE_FMT_BTREE);
+		dp->i_afp->if_ext_max =
+			XFS_IFORK_ASIZE(dp) / (uint)sizeof(xfs_bmbt_rec_t);
+		dp->i_df.if_ext_max =
+			XFS_IFORK_DSIZE(dp) / (uint)sizeof(xfs_bmbt_rec_t);
+		xfs_trans_log_inode(args->trans, dp,
+					XFS_ILOG_CORE | XFS_ILOG_ADATA);
+	}
+
+	xfs_sbversion_add_attr2(mp, args->trans);
+
+	return(0);
+}
+
+/*
+ * Look up a name in a shortform attribute list structure.
+ */
+/*ARGSUSED*/
+int
+xfs_attr_shortform_lookup(xfs_da_args_t *args)
+{
+	xfs_attr_shortform_t *sf;
+	xfs_attr_sf_entry_t *sfe;
+	int i;
+	xfs_ifork_t *ifp;
+
+	ifp = args->dp->i_afp;
+	ASSERT(ifp->if_flags & XFS_IFINLINE);
+	sf = (xfs_attr_shortform_t *)ifp->if_u1.if_data;
+	sfe = &sf->list[0];
+	for (i = 0; i < sf->hdr.count;
+				sfe = XFS_ATTR_SF_NEXTENTRY(sfe), i++) {
+		if (sfe->namelen != args->namelen)
+			continue;
+		if (memcmp(args->name, sfe->nameval, args->namelen) != 0)
+			continue;
+		if (!xfs_attr_namesp_match(args->flags, sfe->flags))
+			continue;
+		return(XFS_ERROR(EEXIST));
+	}
+	return(XFS_ERROR(ENOATTR));
+}
+
+/*
+ * Look up a name in a shortform attribute list structure.
+ */
+/*ARGSUSED*/
+int
+xfs_attr_shortform_getvalue(xfs_da_args_t *args)
+{
+	xfs_attr_shortform_t *sf;
+	xfs_attr_sf_entry_t *sfe;
+	int i;
+
+	ASSERT(args->dp->i_d.di_aformat == XFS_IFINLINE);
+	sf = (xfs_attr_shortform_t *)args->dp->i_afp->if_u1.if_data;
+	sfe = &sf->list[0];
+	for (i = 0; i < sf->hdr.count;
+				sfe = XFS_ATTR_SF_NEXTENTRY(sfe), i++) {
+		if (sfe->namelen != args->namelen)
+			continue;
+		if (memcmp(args->name, sfe->nameval, args->namelen) != 0)
+			continue;
+		if (!xfs_attr_namesp_match(args->flags, sfe->flags))
+			continue;
+		if (args->flags & ATTR_KERNOVAL) {
+			args->valuelen = sfe->valuelen;
+			return(XFS_ERROR(EEXIST));
+		}
+		if (args->valuelen < sfe->valuelen) {
+			args->valuelen = sfe->valuelen;
+			return(XFS_ERROR(ERANGE));
+		}
+		args->valuelen = sfe->valuelen;
+		memcpy(args->value, &sfe->nameval[args->namelen],
+						    args->valuelen);
+		return(XFS_ERROR(EEXIST));
+	}
+	return(XFS_ERROR(ENOATTR));
+}
+
+/*
+ * Convert from using the shortform to the leaf.
+ */
+int
+xfs_attr_shortform_to_leaf(xfs_da_args_t *args)
+{
+	xfs_inode_t *dp;
+	xfs_attr_shortform_t *sf;
+	xfs_attr_sf_entry_t *sfe;
+	xfs_da_args_t nargs;
+	char *tmpbuffer;
+	int error, i, size;
+	xfs_dablk_t blkno;
+	xfs_dabuf_t *bp;
+	xfs_ifork_t *ifp;
+
+	dp = args->dp;
+	ifp = dp->i_afp;
+	sf = (xfs_attr_shortform_t *)ifp->if_u1.if_data;
+	size = be16_to_cpu(sf->hdr.totsize);
+	tmpbuffer = kmem_alloc(size, KM_SLEEP);
+	ASSERT(tmpbuffer != NULL);
+	memcpy(tmpbuffer, ifp->if_u1.if_data, size);
+	sf = (xfs_attr_shortform_t *)tmpbuffer;
+
+	xfs_idata_realloc(dp, -size, XFS_ATTR_FORK);
+	bp = NULL;
+	error = xfs_da_grow_inode(args, &blkno);
+	if (error) {
+		/*
+		 * If we hit an IO error middle of the transaction inside
+		 * grow_inode(), we may have inconsistent data. Bail out.
+		 */
+		if (error == EIO)
+			goto out;
+		xfs_idata_realloc(dp, size, XFS_ATTR_FORK);	/* try to put */
+		memcpy(ifp->if_u1.if_data, tmpbuffer, size);	/* it back */
+		goto out;
+	}
+
+	ASSERT(blkno == 0);
+	error = xfs_attr_leaf_create(args, blkno, &bp);
+	if (error) {
+		error = xfs_da_shrink_inode(args, 0, bp);
+		bp = NULL;
+		if (error)
+			goto out;
+		xfs_idata_realloc(dp, size, XFS_ATTR_FORK);	/* try to put */
+		memcpy(ifp->if_u1.if_data, tmpbuffer, size);	/* it back */
+		goto out;
+	}
+
+	memset((char *)&nargs, 0, sizeof(nargs));
+	nargs.dp = dp;
+	nargs.firstblock = args->firstblock;
+	nargs.flist = args->flist;
+	nargs.total = args->total;
+	nargs.whichfork = XFS_ATTR_FORK;
+	nargs.trans = args->trans;
+	nargs.op_flags = XFS_DA_OP_OKNOENT;
+
+	sfe = &sf->list[0];
+	for (i = 0; i < sf->hdr.count; i++) {
+		nargs.name = sfe->nameval;
+		nargs.namelen = sfe->namelen;
+		nargs.value = &sfe->nameval[nargs.namelen];
+		nargs.valuelen = sfe->valuelen;
+		nargs.hashval = xfs_da_hashname(sfe->nameval,
+						sfe->namelen);
+		nargs.flags = XFS_ATTR_NSP_ONDISK_TO_ARGS(sfe->flags);
+		error = xfs_attr_leaf_lookup_int(bp, &nargs); /* set a->index */
+		ASSERT(error == ENOATTR);
+		error = xfs_attr_leaf_add(bp, &nargs);
+		ASSERT(error != ENOSPC);
+		if (error)
+			goto out;
+		sfe = XFS_ATTR_SF_NEXTENTRY(sfe);
+	}
+	error = 0;
+
+out:
+	if(bp)
+		xfs_da_buf_done(bp);
+	kmem_free(tmpbuffer);
+	return(error);
+}
+
+STATIC int
+xfs_attr_shortform_compare(const void *a, const void *b)
+{
+	xfs_attr_sf_sort_t *sa, *sb;
+
+	sa = (xfs_attr_sf_sort_t *)a;
+	sb = (xfs_attr_sf_sort_t *)b;
+	if (sa->hash < sb->hash) {
+		return(-1);
+	} else if (sa->hash > sb->hash) {
+		return(1);
+	} else {
+		return(sa->entno - sb->entno);
+	}
+}
+
+
+#define XFS_ISRESET_CURSOR(cursor) \
+	(!((cursor)->initted) && !((cursor)->hashval) && \
+	 !((cursor)->blkno) && !((cursor)->offset))
+/*
+ * Copy out entries of shortform attribute lists for attr_list().
+ * Shortform attribute lists are not stored in hashval sorted order.
+ * If the output buffer is not large enough to hold them all, then we
+ * we have to calculate each entries' hashvalue and sort them before
+ * we can begin returning them to the user.
+ */
+/*ARGSUSED*/
+int
+xfs_attr_shortform_list(xfs_attr_list_context_t *context)
+{
+	attrlist_cursor_kern_t *cursor;
+	xfs_attr_sf_sort_t *sbuf, *sbp;
+	xfs_attr_shortform_t *sf;
+	xfs_attr_sf_entry_t *sfe;
+	xfs_inode_t *dp;
+	int sbsize, nsbuf, count, i;
+	int error;
+
+	ASSERT(context != NULL);
+	dp = context->dp;
+	ASSERT(dp != NULL);
+	ASSERT(dp->i_afp != NULL);
+	sf = (xfs_attr_shortform_t *)dp->i_afp->if_u1.if_data;
+	ASSERT(sf != NULL);
+	if (!sf->hdr.count)
+		return(0);
+	cursor = context->cursor;
+	ASSERT(cursor != NULL);
+
+	trace_xfs_attr_list_sf(context);
+
+	/*
+	 * If the buffer is large enough and the cursor is at the start,
+	 * do not bother with sorting since we will return everything in
+	 * one buffer and another call using the cursor won't need to be
+	 * made.
+	 * Note the generous fudge factor of 16 overhead bytes per entry.
+	 * If bufsize is zero then put_listent must be a search function
+	 * and can just scan through what we have.
+	 */
+	if (context->bufsize == 0 ||
+	    (XFS_ISRESET_CURSOR(cursor) &&
+             (dp->i_afp->if_bytes + sf->hdr.count * 16) < context->bufsize)) {
+		for (i = 0, sfe = &sf->list[0]; i < sf->hdr.count; i++) {
+			error = context->put_listent(context,
+					   sfe->flags,
+					   sfe->nameval,
+					   (int)sfe->namelen,
+					   (int)sfe->valuelen,
+					   &sfe->nameval[sfe->namelen]);
+
+			/*
+			 * Either search callback finished early or
+			 * didn't fit it all in the buffer after all.
+			 */
+			if (context->seen_enough)
+				break;
+
+			if (error)
+				return error;
+			sfe = XFS_ATTR_SF_NEXTENTRY(sfe);
+		}
+		trace_xfs_attr_list_sf_all(context);
+		return(0);
+	}
+
+	/* do no more for a search callback */
+	if (context->bufsize == 0)
+		return 0;
+
+	/*
+	 * It didn't all fit, so we have to sort everything on hashval.
+	 */
+	sbsize = sf->hdr.count * sizeof(*sbuf);
+	sbp = sbuf = kmem_alloc(sbsize, KM_SLEEP | KM_NOFS);
+
+	/*
+	 * Scan the attribute list for the rest of the entries, storing
+	 * the relevant info from only those that match into a buffer.
+	 */
+	nsbuf = 0;
+	for (i = 0, sfe = &sf->list[0]; i < sf->hdr.count; i++) {
+		if (unlikely(
+		    ((char *)sfe < (char *)sf) ||
+		    ((char *)sfe >= ((char *)sf + dp->i_afp->if_bytes)))) {
+			XFS_CORRUPTION_ERROR("xfs_attr_shortform_list",
+					     XFS_ERRLEVEL_LOW,
+					     context->dp->i_mount, sfe);
+			kmem_free(sbuf);
+			return XFS_ERROR(EFSCORRUPTED);
+		}
+
+		sbp->entno = i;
+		sbp->hash = xfs_da_hashname(sfe->nameval, sfe->namelen);
+		sbp->name = sfe->nameval;
+		sbp->namelen = sfe->namelen;
+		/* These are bytes, and both on-disk, don't endian-flip */
+		sbp->valuelen = sfe->valuelen;
+		sbp->flags = sfe->flags;
+		sfe = XFS_ATTR_SF_NEXTENTRY(sfe);
+		sbp++;
+		nsbuf++;
+	}
+
+	/*
+	 * Sort the entries on hash then entno.
+	 */
+	xfs_sort(sbuf, nsbuf, sizeof(*sbuf), xfs_attr_shortform_compare);
+
+	/*
+	 * Re-find our place IN THE SORTED LIST.
+	 */
+	count = 0;
+	cursor->initted = 1;
+	cursor->blkno = 0;
+	for (sbp = sbuf, i = 0; i < nsbuf; i++, sbp++) {
+		if (sbp->hash == cursor->hashval) {
+			if (cursor->offset == count) {
+				break;
+			}
+			count++;
+		} else if (sbp->hash > cursor->hashval) {
+			break;
+		}
+	}
+	if (i == nsbuf) {
+		kmem_free(sbuf);
+		return(0);
+	}
+
+	/*
+	 * Loop putting entries into the user buffer.
+	 */
+	for ( ; i < nsbuf; i++, sbp++) {
+		if (cursor->hashval != sbp->hash) {
+			cursor->hashval = sbp->hash;
+			cursor->offset = 0;
+		}
+		error = context->put_listent(context,
+					sbp->flags,
+					sbp->name,
+					sbp->namelen,
+					sbp->valuelen,
+					&sbp->name[sbp->namelen]);
+		if (error)
+			return error;
+		if (context->seen_enough)
+			break;
+		cursor->offset++;
+	}
+
+	kmem_free(sbuf);
+	return(0);
+}
+
+/*
+ * Check a leaf attribute block to see if all the entries would fit into
+ * a shortform attribute list.
+ */
+int
+xfs_attr_shortform_allfit(xfs_dabuf_t *bp, xfs_inode_t *dp)
+{
+	xfs_attr_leafblock_t *leaf;
+	xfs_attr_leaf_entry_t *entry;
+	xfs_attr_leaf_name_local_t *name_loc;
+	int bytes, i;
+
+	leaf = bp->data;
+	ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_ATTR_LEAF_MAGIC);
+
+	entry = &leaf->entries[0];
+	bytes = sizeof(struct xfs_attr_sf_hdr);
+	for (i = 0; i < be16_to_cpu(leaf->hdr.count); entry++, i++) {
+		if (entry->flags & XFS_ATTR_INCOMPLETE)
+			continue;		/* don't copy partial entries */
+		if (!(entry->flags & XFS_ATTR_LOCAL))
+			return(0);
+		name_loc = xfs_attr_leaf_name_local(leaf, i);
+		if (name_loc->namelen >= XFS_ATTR_SF_ENTSIZE_MAX)
+			return(0);
+		if (be16_to_cpu(name_loc->valuelen) >= XFS_ATTR_SF_ENTSIZE_MAX)
+			return(0);
+		bytes += sizeof(struct xfs_attr_sf_entry)-1
+				+ name_loc->namelen
+				+ be16_to_cpu(name_loc->valuelen);
+	}
+	if ((dp->i_mount->m_flags & XFS_MOUNT_ATTR2) &&
+	    (dp->i_d.di_format != XFS_DINODE_FMT_BTREE) &&
+	    (bytes == sizeof(struct xfs_attr_sf_hdr)))
+		return(-1);
+	return(xfs_attr_shortform_bytesfit(dp, bytes));
+}
+
+/*
+ * Convert a leaf attribute list to shortform attribute list
+ */
+int
+xfs_attr_leaf_to_shortform(xfs_dabuf_t *bp, xfs_da_args_t *args, int forkoff)
+{
+	xfs_attr_leafblock_t *leaf;
+	xfs_attr_leaf_entry_t *entry;
+	xfs_attr_leaf_name_local_t *name_loc;
+	xfs_da_args_t nargs;
+	xfs_inode_t *dp;
+	char *tmpbuffer;
+	int error, i;
+
+	dp = args->dp;
+	tmpbuffer = kmem_alloc(XFS_LBSIZE(dp->i_mount), KM_SLEEP);
+	ASSERT(tmpbuffer != NULL);
+
+	ASSERT(bp != NULL);
+	memcpy(tmpbuffer, bp->data, XFS_LBSIZE(dp->i_mount));
+	leaf = (xfs_attr_leafblock_t *)tmpbuffer;
+	ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_ATTR_LEAF_MAGIC);
+	memset(bp->data, 0, XFS_LBSIZE(dp->i_mount));
+
+	/*
+	 * Clean out the prior contents of the attribute list.
+	 */
+	error = xfs_da_shrink_inode(args, 0, bp);
+	if (error)
+		goto out;
+
+	if (forkoff == -1) {
+		ASSERT(dp->i_mount->m_flags & XFS_MOUNT_ATTR2);
+		ASSERT(dp->i_d.di_format != XFS_DINODE_FMT_BTREE);
+		xfs_attr_fork_reset(dp, args->trans);
+		goto out;
+	}
+
+	xfs_attr_shortform_create(args);
+
+	/*
+	 * Copy the attributes
+	 */
+	memset((char *)&nargs, 0, sizeof(nargs));
+	nargs.dp = dp;
+	nargs.firstblock = args->firstblock;
+	nargs.flist = args->flist;
+	nargs.total = args->total;
+	nargs.whichfork = XFS_ATTR_FORK;
+	nargs.trans = args->trans;
+	nargs.op_flags = XFS_DA_OP_OKNOENT;
+	entry = &leaf->entries[0];
+	for (i = 0; i < be16_to_cpu(leaf->hdr.count); entry++, i++) {
+		if (entry->flags & XFS_ATTR_INCOMPLETE)
+			continue;	/* don't copy partial entries */
+		if (!entry->nameidx)
+			continue;
+		ASSERT(entry->flags & XFS_ATTR_LOCAL);
+		name_loc = xfs_attr_leaf_name_local(leaf, i);
+		nargs.name = name_loc->nameval;
+		nargs.namelen = name_loc->namelen;
+		nargs.value = &name_loc->nameval[nargs.namelen];
+		nargs.valuelen = be16_to_cpu(name_loc->valuelen);
+		nargs.hashval = be32_to_cpu(entry->hashval);
+		nargs.flags = XFS_ATTR_NSP_ONDISK_TO_ARGS(entry->flags);
+		xfs_attr_shortform_add(&nargs, forkoff);
+	}
+	error = 0;
+
+out:
+	kmem_free(tmpbuffer);
+	return(error);
+}
+
+/*
+ * Convert from using a single leaf to a root node and a leaf.
+ */
+int
+xfs_attr_leaf_to_node(xfs_da_args_t *args)
+{
+	xfs_attr_leafblock_t *leaf;
+	xfs_da_intnode_t *node;
+	xfs_inode_t *dp;
+	xfs_dabuf_t *bp1, *bp2;
+	xfs_dablk_t blkno;
+	int error;
+
+	dp = args->dp;
+	bp1 = bp2 = NULL;
+	error = xfs_da_grow_inode(args, &blkno);
+	if (error)
+		goto out;
+	error = xfs_da_read_buf(args->trans, args->dp, 0, -1, &bp1,
+					     XFS_ATTR_FORK);
+	if (error)
+		goto out;
+	ASSERT(bp1 != NULL);
+	bp2 = NULL;
+	error = xfs_da_get_buf(args->trans, args->dp, blkno, -1, &bp2,
+					    XFS_ATTR_FORK);
+	if (error)
+		goto out;
+	ASSERT(bp2 != NULL);
+	memcpy(bp2->data, bp1->data, XFS_LBSIZE(dp->i_mount));
+	xfs_da_buf_done(bp1);
+	bp1 = NULL;
+	xfs_da_log_buf(args->trans, bp2, 0, XFS_LBSIZE(dp->i_mount) - 1);
+
+	/*
+	 * Set up the new root node.
+	 */
+	error = xfs_da_node_create(args, 0, 1, &bp1, XFS_ATTR_FORK);
+	if (error)
+		goto out;
+	node = bp1->data;
+	leaf = bp2->data;
+	ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_ATTR_LEAF_MAGIC);
+	/* both on-disk, don't endian-flip twice */
+	node->btree[0].hashval =
+		leaf->entries[be16_to_cpu(leaf->hdr.count)-1 ].hashval;
+	node->btree[0].before = cpu_to_be32(blkno);
+	node->hdr.count = cpu_to_be16(1);
+	xfs_da_log_buf(args->trans, bp1, 0, XFS_LBSIZE(dp->i_mount) - 1);
+	error = 0;
+out:
+	if (bp1)
+		xfs_da_buf_done(bp1);
+	if (bp2)
+		xfs_da_buf_done(bp2);
+	return(error);
+}
+
+
+/*========================================================================
+ * Routines used for growing the Btree.
+ *========================================================================*/
+
+/*
+ * Create the initial contents of a leaf attribute list
+ * or a leaf in a node attribute list.
+ */
+STATIC int
+xfs_attr_leaf_create(xfs_da_args_t *args, xfs_dablk_t blkno, xfs_dabuf_t **bpp)
+{
+	xfs_attr_leafblock_t *leaf;
+	xfs_attr_leaf_hdr_t *hdr;
+	xfs_inode_t *dp;
+	xfs_dabuf_t *bp;
+	int error;
+
+	dp = args->dp;
+	ASSERT(dp != NULL);
+	error = xfs_da_get_buf(args->trans, args->dp, blkno, -1, &bp,
+					    XFS_ATTR_FORK);
+	if (error)
+		return(error);
+	ASSERT(bp != NULL);
+	leaf = bp->data;
+	memset((char *)leaf, 0, XFS_LBSIZE(dp->i_mount));
+	hdr = &leaf->hdr;
+	hdr->info.magic = cpu_to_be16(XFS_ATTR_LEAF_MAGIC);
+	hdr->firstused = cpu_to_be16(XFS_LBSIZE(dp->i_mount));
+	if (!hdr->firstused) {
+		hdr->firstused = cpu_to_be16(
+			XFS_LBSIZE(dp->i_mount) - XFS_ATTR_LEAF_NAME_ALIGN);
+	}
+
+	hdr->freemap[0].base = cpu_to_be16(sizeof(xfs_attr_leaf_hdr_t));
+	hdr->freemap[0].size = cpu_to_be16(be16_to_cpu(hdr->firstused) -
+					   sizeof(xfs_attr_leaf_hdr_t));
+
+	xfs_da_log_buf(args->trans, bp, 0, XFS_LBSIZE(dp->i_mount) - 1);
+
+	*bpp = bp;
+	return(0);
+}
+
+/*
+ * Split the leaf node, rebalance, then add the new entry.
+ */
+int
+xfs_attr_leaf_split(xfs_da_state_t *state, xfs_da_state_blk_t *oldblk,
+				   xfs_da_state_blk_t *newblk)
+{
+	xfs_dablk_t blkno;
+	int error;
+
+	/*
+	 * Allocate space for a new leaf node.
+	 */
+	ASSERT(oldblk->magic == XFS_ATTR_LEAF_MAGIC);
+	error = xfs_da_grow_inode(state->args, &blkno);
+	if (error)
+		return(error);
+	error = xfs_attr_leaf_create(state->args, blkno, &newblk->bp);
+	if (error)
+		return(error);
+	newblk->blkno = blkno;
+	newblk->magic = XFS_ATTR_LEAF_MAGIC;
+
+	/*
+	 * Rebalance the entries across the two leaves.
+	 * NOTE: rebalance() currently depends on the 2nd block being empty.
+	 */
+	xfs_attr_leaf_rebalance(state, oldblk, newblk);
+	error = xfs_da_blk_link(state, oldblk, newblk);
+	if (error)
+		return(error);
+
+	/*
+	 * Save info on "old" attribute for "atomic rename" ops, leaf_add()
+	 * modifies the index/blkno/rmtblk/rmtblkcnt fields to show the
+	 * "new" attrs info.  Will need the "old" info to remove it later.
+	 *
+	 * Insert the "new" entry in the correct block.
+	 */
+	if (state->inleaf)
+		error = xfs_attr_leaf_add(oldblk->bp, state->args);
+	else
+		error = xfs_attr_leaf_add(newblk->bp, state->args);
+
+	/*
+	 * Update last hashval in each block since we added the name.
+	 */
+	oldblk->hashval = xfs_attr_leaf_lasthash(oldblk->bp, NULL);
+	newblk->hashval = xfs_attr_leaf_lasthash(newblk->bp, NULL);
+	return(error);
+}
+
+/*
+ * Add a name to the leaf attribute list structure.
+ */
+int
+xfs_attr_leaf_add(xfs_dabuf_t *bp, xfs_da_args_t *args)
+{
+	xfs_attr_leafblock_t *leaf;
+	xfs_attr_leaf_hdr_t *hdr;
+	xfs_attr_leaf_map_t *map;
+	int tablesize, entsize, sum, tmp, i;
+
+	leaf = bp->data;
+	ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_ATTR_LEAF_MAGIC);
+	ASSERT((args->index >= 0)
+		&& (args->index <= be16_to_cpu(leaf->hdr.count)));
+	hdr = &leaf->hdr;
+	entsize = xfs_attr_leaf_newentsize(args->namelen, args->valuelen,
+			   args->trans->t_mountp->m_sb.sb_blocksize, NULL);
+
+	/*
+	 * Search through freemap for first-fit on new name length.
+	 * (may need to figure in size of entry struct too)
+	 */
+	tablesize = (be16_to_cpu(hdr->count) + 1)
+					* sizeof(xfs_attr_leaf_entry_t)
+					+ sizeof(xfs_attr_leaf_hdr_t);
+	map = &hdr->freemap[XFS_ATTR_LEAF_MAPSIZE-1];
+	for (sum = 0, i = XFS_ATTR_LEAF_MAPSIZE-1; i >= 0; map--, i--) {
+		if (tablesize > be16_to_cpu(hdr->firstused)) {
+			sum += be16_to_cpu(map->size);
+			continue;
+		}
+		if (!map->size)
+			continue;	/* no space in this map */
+		tmp = entsize;
+		if (be16_to_cpu(map->base) < be16_to_cpu(hdr->firstused))
+			tmp += sizeof(xfs_attr_leaf_entry_t);
+		if (be16_to_cpu(map->size) >= tmp) {
+			tmp = xfs_attr_leaf_add_work(bp, args, i);
+			return(tmp);
+		}
+		sum += be16_to_cpu(map->size);
+	}
+
+	/*
+	 * If there are no holes in the address space of the block,
+	 * and we don't have enough freespace, then compaction will do us
+	 * no good and we should just give up.
+	 */
+	if (!hdr->holes && (sum < entsize))
+		return(XFS_ERROR(ENOSPC));
+
+	/*
+	 * Compact the entries to coalesce free space.
+	 * This may change the hdr->count via dropping INCOMPLETE entries.
+	 */
+	xfs_attr_leaf_compact(args->trans, bp);
+
+	/*
+	 * After compaction, the block is guaranteed to have only one
+	 * free region, in freemap[0].  If it is not big enough, give up.
+	 */
+	if (be16_to_cpu(hdr->freemap[0].size)
+				< (entsize + sizeof(xfs_attr_leaf_entry_t)))
+		return(XFS_ERROR(ENOSPC));
+
+	return(xfs_attr_leaf_add_work(bp, args, 0));
+}
+
+/*
+ * Add a name to a leaf attribute list structure.
+ */
+STATIC int
+xfs_attr_leaf_add_work(xfs_dabuf_t *bp, xfs_da_args_t *args, int mapindex)
+{
+	xfs_attr_leafblock_t *leaf;
+	xfs_attr_leaf_hdr_t *hdr;
+	xfs_attr_leaf_entry_t *entry;
+	xfs_attr_leaf_name_local_t *name_loc;
+	xfs_attr_leaf_name_remote_t *name_rmt;
+	xfs_attr_leaf_map_t *map;
+	xfs_mount_t *mp;
+	int tmp, i;
+
+	leaf = bp->data;
+	ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_ATTR_LEAF_MAGIC);
+	hdr = &leaf->hdr;
+	ASSERT((mapindex >= 0) && (mapindex < XFS_ATTR_LEAF_MAPSIZE));
+	ASSERT((args->index >= 0) && (args->index <= be16_to_cpu(hdr->count)));
+
+	/*
+	 * Force open some space in the entry array and fill it in.
+	 */
+	entry = &leaf->entries[args->index];
+	if (args->index < be16_to_cpu(hdr->count)) {
+		tmp  = be16_to_cpu(hdr->count) - args->index;
+		tmp *= sizeof(xfs_attr_leaf_entry_t);
+		memmove((char *)(entry+1), (char *)entry, tmp);
+		xfs_da_log_buf(args->trans, bp,
+		    XFS_DA_LOGRANGE(leaf, entry, tmp + sizeof(*entry)));
+	}
+	be16_add_cpu(&hdr->count, 1);
+
+	/*
+	 * Allocate space for the new string (at the end of the run).
+	 */
+	map = &hdr->freemap[mapindex];
+	mp = args->trans->t_mountp;
+	ASSERT(be16_to_cpu(map->base) < XFS_LBSIZE(mp));
+	ASSERT((be16_to_cpu(map->base) & 0x3) == 0);
+	ASSERT(be16_to_cpu(map->size) >=
+		xfs_attr_leaf_newentsize(args->namelen, args->valuelen,
+					 mp->m_sb.sb_blocksize, NULL));
+	ASSERT(be16_to_cpu(map->size) < XFS_LBSIZE(mp));
+	ASSERT((be16_to_cpu(map->size) & 0x3) == 0);
+	be16_add_cpu(&map->size,
+		-xfs_attr_leaf_newentsize(args->namelen, args->valuelen,
+					  mp->m_sb.sb_blocksize, &tmp));
+	entry->nameidx = cpu_to_be16(be16_to_cpu(map->base) +
+				     be16_to_cpu(map->size));
+	entry->hashval = cpu_to_be32(args->hashval);
+	entry->flags = tmp ? XFS_ATTR_LOCAL : 0;
+	entry->flags |= XFS_ATTR_NSP_ARGS_TO_ONDISK(args->flags);
+	if (args->op_flags & XFS_DA_OP_RENAME) {
+		entry->flags |= XFS_ATTR_INCOMPLETE;
+		if ((args->blkno2 == args->blkno) &&
+		    (args->index2 <= args->index)) {
+			args->index2++;
+		}
+	}
+	xfs_da_log_buf(args->trans, bp,
+			  XFS_DA_LOGRANGE(leaf, entry, sizeof(*entry)));
+	ASSERT((args->index == 0) ||
+	       (be32_to_cpu(entry->hashval) >= be32_to_cpu((entry-1)->hashval)));
+	ASSERT((args->index == be16_to_cpu(hdr->count)-1) ||
+	       (be32_to_cpu(entry->hashval) <= be32_to_cpu((entry+1)->hashval)));
+
+	/*
+	 * Copy the attribute name and value into the new space.
+	 *
+	 * For "remote" attribute values, simply note that we need to
+	 * allocate space for the "remote" value.  We can't actually
+	 * allocate the extents in this transaction, and we can't decide
+	 * which blocks they should be as we might allocate more blocks
+	 * as part of this transaction (a split operation for example).
+	 */
+	if (entry->flags & XFS_ATTR_LOCAL) {
+		name_loc = xfs_attr_leaf_name_local(leaf, args->index);
+		name_loc->namelen = args->namelen;
+		name_loc->valuelen = cpu_to_be16(args->valuelen);
+		memcpy((char *)name_loc->nameval, args->name, args->namelen);
+		memcpy((char *)&name_loc->nameval[args->namelen], args->value,
+				   be16_to_cpu(name_loc->valuelen));
+	} else {
+		name_rmt = xfs_attr_leaf_name_remote(leaf, args->index);
+		name_rmt->namelen = args->namelen;
+		memcpy((char *)name_rmt->name, args->name, args->namelen);
+		entry->flags |= XFS_ATTR_INCOMPLETE;
+		/* just in case */
+		name_rmt->valuelen = 0;
+		name_rmt->valueblk = 0;
+		args->rmtblkno = 1;
+		args->rmtblkcnt = XFS_B_TO_FSB(mp, args->valuelen);
+	}
+	xfs_da_log_buf(args->trans, bp,
+	     XFS_DA_LOGRANGE(leaf, xfs_attr_leaf_name(leaf, args->index),
+				   xfs_attr_leaf_entsize(leaf, args->index)));
+
+	/*
+	 * Update the control info for this leaf node
+	 */
+	if (be16_to_cpu(entry->nameidx) < be16_to_cpu(hdr->firstused)) {
+		/* both on-disk, don't endian-flip twice */
+		hdr->firstused = entry->nameidx;
+	}
+	ASSERT(be16_to_cpu(hdr->firstused) >=
+	       ((be16_to_cpu(hdr->count) * sizeof(*entry)) + sizeof(*hdr)));
+	tmp = (be16_to_cpu(hdr->count)-1) * sizeof(xfs_attr_leaf_entry_t)
+					+ sizeof(xfs_attr_leaf_hdr_t);
+	map = &hdr->freemap[0];
+	for (i = 0; i < XFS_ATTR_LEAF_MAPSIZE; map++, i++) {
+		if (be16_to_cpu(map->base) == tmp) {
+			be16_add_cpu(&map->base, sizeof(xfs_attr_leaf_entry_t));
+			be16_add_cpu(&map->size,
+				 -((int)sizeof(xfs_attr_leaf_entry_t)));
+		}
+	}
+	be16_add_cpu(&hdr->usedbytes, xfs_attr_leaf_entsize(leaf, args->index));
+	xfs_da_log_buf(args->trans, bp,
+		XFS_DA_LOGRANGE(leaf, hdr, sizeof(*hdr)));
+	return(0);
+}
+
+/*
+ * Garbage collect a leaf attribute list block by copying it to a new buffer.
+ */
+STATIC void
+xfs_attr_leaf_compact(xfs_trans_t *trans, xfs_dabuf_t *bp)
+{
+	xfs_attr_leafblock_t *leaf_s, *leaf_d;
+	xfs_attr_leaf_hdr_t *hdr_s, *hdr_d;
+	xfs_mount_t *mp;
+	char *tmpbuffer;
+
+	mp = trans->t_mountp;
+	tmpbuffer = kmem_alloc(XFS_LBSIZE(mp), KM_SLEEP);
+	ASSERT(tmpbuffer != NULL);
+	memcpy(tmpbuffer, bp->data, XFS_LBSIZE(mp));
+	memset(bp->data, 0, XFS_LBSIZE(mp));
+
+	/*
+	 * Copy basic information
+	 */
+	leaf_s = (xfs_attr_leafblock_t *)tmpbuffer;
+	leaf_d = bp->data;
+	hdr_s = &leaf_s->hdr;
+	hdr_d = &leaf_d->hdr;
+	hdr_d->info = hdr_s->info;	/* struct copy */
+	hdr_d->firstused = cpu_to_be16(XFS_LBSIZE(mp));
+	/* handle truncation gracefully */
+	if (!hdr_d->firstused) {
+		hdr_d->firstused = cpu_to_be16(
+				XFS_LBSIZE(mp) - XFS_ATTR_LEAF_NAME_ALIGN);
+	}
+	hdr_d->usedbytes = 0;
+	hdr_d->count = 0;
+	hdr_d->holes = 0;
+	hdr_d->freemap[0].base = cpu_to_be16(sizeof(xfs_attr_leaf_hdr_t));
+	hdr_d->freemap[0].size = cpu_to_be16(be16_to_cpu(hdr_d->firstused) -
+					     sizeof(xfs_attr_leaf_hdr_t));
+
+	/*
+	 * Copy all entry's in the same (sorted) order,
+	 * but allocate name/value pairs packed and in sequence.
+	 */
+	xfs_attr_leaf_moveents(leaf_s, 0, leaf_d, 0,
+				be16_to_cpu(hdr_s->count), mp);
+	xfs_da_log_buf(trans, bp, 0, XFS_LBSIZE(mp) - 1);
+
+	kmem_free(tmpbuffer);
+}
+
+/*
+ * Redistribute the attribute list entries between two leaf nodes,
+ * taking into account the size of the new entry.
+ *
+ * NOTE: if new block is empty, then it will get the upper half of the
+ * old block.  At present, all (one) callers pass in an empty second block.
+ *
+ * This code adjusts the args->index/blkno and args->index2/blkno2 fields
+ * to match what it is doing in splitting the attribute leaf block.  Those
+ * values are used in "atomic rename" operations on attributes.  Note that
+ * the "new" and "old" values can end up in different blocks.
+ */
+STATIC void
+xfs_attr_leaf_rebalance(xfs_da_state_t *state, xfs_da_state_blk_t *blk1,
+				       xfs_da_state_blk_t *blk2)
+{
+	xfs_da_args_t *args;
+	xfs_da_state_blk_t *tmp_blk;
+	xfs_attr_leafblock_t *leaf1, *leaf2;
+	xfs_attr_leaf_hdr_t *hdr1, *hdr2;
+	int count, totallen, max, space, swap;
+
+	/*
+	 * Set up environment.
+	 */
+	ASSERT(blk1->magic == XFS_ATTR_LEAF_MAGIC);
+	ASSERT(blk2->magic == XFS_ATTR_LEAF_MAGIC);
+	leaf1 = blk1->bp->data;
+	leaf2 = blk2->bp->data;
+	ASSERT(be16_to_cpu(leaf1->hdr.info.magic) == XFS_ATTR_LEAF_MAGIC);
+	ASSERT(be16_to_cpu(leaf2->hdr.info.magic) == XFS_ATTR_LEAF_MAGIC);
+	args = state->args;
+
+	/*
+	 * Check ordering of blocks, reverse if it makes things simpler.
+	 *
+	 * NOTE: Given that all (current) callers pass in an empty
+	 * second block, this code should never set "swap".
+	 */
+	swap = 0;
+	if (xfs_attr_leaf_order(blk1->bp, blk2->bp)) {
+		tmp_blk = blk1;
+		blk1 = blk2;
+		blk2 = tmp_blk;
+		leaf1 = blk1->bp->data;
+		leaf2 = blk2->bp->data;
+		swap = 1;
+	}
+	hdr1 = &leaf1->hdr;
+	hdr2 = &leaf2->hdr;
+
+	/*
+	 * Examine entries until we reduce the absolute difference in
+	 * byte usage between the two blocks to a minimum.  Then get
+	 * the direction to copy and the number of elements to move.
+	 *
+	 * "inleaf" is true if the new entry should be inserted into blk1.
+	 * If "swap" is also true, then reverse the sense of "inleaf".
+	 */
+	state->inleaf = xfs_attr_leaf_figure_balance(state, blk1, blk2,
+							    &count, &totallen);
+	if (swap)
+		state->inleaf = !state->inleaf;
+
+	/*
+	 * Move any entries required from leaf to leaf:
+	 */
+	if (count < be16_to_cpu(hdr1->count)) {
+		/*
+		 * Figure the total bytes to be added to the destination leaf.
+		 */
+		/* number entries being moved */
+		count = be16_to_cpu(hdr1->count) - count;
+		space  = be16_to_cpu(hdr1->usedbytes) - totallen;
+		space += count * sizeof(xfs_attr_leaf_entry_t);
+
+		/*
+		 * leaf2 is the destination, compact it if it looks tight.
+		 */
+		max  = be16_to_cpu(hdr2->firstused)
+						- sizeof(xfs_attr_leaf_hdr_t);
+		max -= be16_to_cpu(hdr2->count) * sizeof(xfs_attr_leaf_entry_t);
+		if (space > max) {
+			xfs_attr_leaf_compact(args->trans, blk2->bp);
+		}
+
+		/*
+		 * Move high entries from leaf1 to low end of leaf2.
+		 */
+		xfs_attr_leaf_moveents(leaf1, be16_to_cpu(hdr1->count) - count,
+				leaf2, 0, count, state->mp);
+
+		xfs_da_log_buf(args->trans, blk1->bp, 0, state->blocksize-1);
+		xfs_da_log_buf(args->trans, blk2->bp, 0, state->blocksize-1);
+	} else if (count > be16_to_cpu(hdr1->count)) {
+		/*
+		 * I assert that since all callers pass in an empty
+		 * second buffer, this code should never execute.
+		 */
+
+		/*
+		 * Figure the total bytes to be added to the destination leaf.
+		 */
+		/* number entries being moved */
+		count -= be16_to_cpu(hdr1->count);
+		space  = totallen - be16_to_cpu(hdr1->usedbytes);
+		space += count * sizeof(xfs_attr_leaf_entry_t);
+
+		/*
+		 * leaf1 is the destination, compact it if it looks tight.
+		 */
+		max  = be16_to_cpu(hdr1->firstused)
+						- sizeof(xfs_attr_leaf_hdr_t);
+		max -= be16_to_cpu(hdr1->count) * sizeof(xfs_attr_leaf_entry_t);
+		if (space > max) {
+			xfs_attr_leaf_compact(args->trans, blk1->bp);
+		}
+
+		/*
+		 * Move low entries from leaf2 to high end of leaf1.
+		 */
+		xfs_attr_leaf_moveents(leaf2, 0, leaf1,
+				be16_to_cpu(hdr1->count), count, state->mp);
+
+		xfs_da_log_buf(args->trans, blk1->bp, 0, state->blocksize-1);
+		xfs_da_log_buf(args->trans, blk2->bp, 0, state->blocksize-1);
+	}
+
+	/*
+	 * Copy out last hashval in each block for B-tree code.
+	 */
+	blk1->hashval = be32_to_cpu(
+		leaf1->entries[be16_to_cpu(leaf1->hdr.count)-1].hashval);
+	blk2->hashval = be32_to_cpu(
+		leaf2->entries[be16_to_cpu(leaf2->hdr.count)-1].hashval);
+
+	/*
+	 * Adjust the expected index for insertion.
+	 * NOTE: this code depends on the (current) situation that the
+	 * second block was originally empty.
+	 *
+	 * If the insertion point moved to the 2nd block, we must adjust
+	 * the index.  We must also track the entry just following the
+	 * new entry for use in an "atomic rename" operation, that entry
+	 * is always the "old" entry and the "new" entry is what we are
+	 * inserting.  The index/blkno fields refer to the "old" entry,
+	 * while the index2/blkno2 fields refer to the "new" entry.
+	 */
+	if (blk1->index > be16_to_cpu(leaf1->hdr.count)) {
+		ASSERT(state->inleaf == 0);
+		blk2->index = blk1->index - be16_to_cpu(leaf1->hdr.count);
+		args->index = args->index2 = blk2->index;
+		args->blkno = args->blkno2 = blk2->blkno;
+	} else if (blk1->index == be16_to_cpu(leaf1->hdr.count)) {
+		if (state->inleaf) {
+			args->index = blk1->index;
+			args->blkno = blk1->blkno;
+			args->index2 = 0;
+			args->blkno2 = blk2->blkno;
+		} else {
+			blk2->index = blk1->index
+				    - be16_to_cpu(leaf1->hdr.count);
+			args->index = args->index2 = blk2->index;
+			args->blkno = args->blkno2 = blk2->blkno;
+		}
+	} else {
+		ASSERT(state->inleaf == 1);
+		args->index = args->index2 = blk1->index;
+		args->blkno = args->blkno2 = blk1->blkno;
+	}
+}
+
+/*
+ * Examine entries until we reduce the absolute difference in
+ * byte usage between the two blocks to a minimum.
+ * GROT: Is this really necessary?  With other than a 512 byte blocksize,
+ * GROT: there will always be enough room in either block for a new entry.
+ * GROT: Do a double-split for this case?
+ */
+STATIC int
+xfs_attr_leaf_figure_balance(xfs_da_state_t *state,
+				    xfs_da_state_blk_t *blk1,
+				    xfs_da_state_blk_t *blk2,
+				    int *countarg, int *usedbytesarg)
+{
+	xfs_attr_leafblock_t *leaf1, *leaf2;
+	xfs_attr_leaf_hdr_t *hdr1, *hdr2;
+	xfs_attr_leaf_entry_t *entry;
+	int count, max, index, totallen, half;
+	int lastdelta, foundit, tmp;
+
+	/*
+	 * Set up environment.
+	 */
+	leaf1 = blk1->bp->data;
+	leaf2 = blk2->bp->data;
+	hdr1 = &leaf1->hdr;
+	hdr2 = &leaf2->hdr;
+	foundit = 0;
+	totallen = 0;
+
+	/*
+	 * Examine entries until we reduce the absolute difference in
+	 * byte usage between the two blocks to a minimum.
+	 */
+	max = be16_to_cpu(hdr1->count) + be16_to_cpu(hdr2->count);
+	half  = (max+1) * sizeof(*entry);
+	half += be16_to_cpu(hdr1->usedbytes) +
+		be16_to_cpu(hdr2->usedbytes) +
+		xfs_attr_leaf_newentsize(
+				state->args->namelen,
+				state->args->valuelen,
+				state->blocksize, NULL);
+	half /= 2;
+	lastdelta = state->blocksize;
+	entry = &leaf1->entries[0];
+	for (count = index = 0; count < max; entry++, index++, count++) {
+
+#define XFS_ATTR_ABS(A)	(((A) < 0) ? -(A) : (A))
+		/*
+		 * The new entry is in the first block, account for it.
+		 */
+		if (count == blk1->index) {
+			tmp = totallen + sizeof(*entry) +
+				xfs_attr_leaf_newentsize(
+						state->args->namelen,
+						state->args->valuelen,
+						state->blocksize, NULL);
+			if (XFS_ATTR_ABS(half - tmp) > lastdelta)
+				break;
+			lastdelta = XFS_ATTR_ABS(half - tmp);
+			totallen = tmp;
+			foundit = 1;
+		}
+
+		/*
+		 * Wrap around into the second block if necessary.
+		 */
+		if (count == be16_to_cpu(hdr1->count)) {
+			leaf1 = leaf2;
+			entry = &leaf1->entries[0];
+			index = 0;
+		}
+
+		/*
+		 * Figure out if next leaf entry would be too much.
+		 */
+		tmp = totallen + sizeof(*entry) + xfs_attr_leaf_entsize(leaf1,
+									index);
+		if (XFS_ATTR_ABS(half - tmp) > lastdelta)
+			break;
+		lastdelta = XFS_ATTR_ABS(half - tmp);
+		totallen = tmp;
+#undef XFS_ATTR_ABS
+	}
+
+	/*
+	 * Calculate the number of usedbytes that will end up in lower block.
+	 * If new entry not in lower block, fix up the count.
+	 */
+	totallen -= count * sizeof(*entry);
+	if (foundit) {
+		totallen -= sizeof(*entry) +
+				xfs_attr_leaf_newentsize(
+						state->args->namelen,
+						state->args->valuelen,
+						state->blocksize, NULL);
+	}
+
+	*countarg = count;
+	*usedbytesarg = totallen;
+	return(foundit);
+}
+
+/*========================================================================
+ * Routines used for shrinking the Btree.
+ *========================================================================*/
+
+/*
+ * Check a leaf block and its neighbors to see if the block should be
+ * collapsed into one or the other neighbor.  Always keep the block
+ * with the smaller block number.
+ * If the current block is over 50% full, don't try to join it, return 0.
+ * If the block is empty, fill in the state structure and return 2.
+ * If it can be collapsed, fill in the state structure and return 1.
+ * If nothing can be done, return 0.
+ *
+ * GROT: allow for INCOMPLETE entries in calculation.
+ */
+int
+xfs_attr_leaf_toosmall(xfs_da_state_t *state, int *action)
+{
+	xfs_attr_leafblock_t *leaf;
+	xfs_da_state_blk_t *blk;
+	xfs_da_blkinfo_t *info;
+	int count, bytes, forward, error, retval, i;
+	xfs_dablk_t blkno;
+	xfs_dabuf_t *bp;
+
+	/*
+	 * Check for the degenerate case of the block being over 50% full.
+	 * If so, it's not worth even looking to see if we might be able
+	 * to coalesce with a sibling.
+	 */
+	blk = &state->path.blk[ state->path.active-1 ];
+	info = blk->bp->data;
+	ASSERT(be16_to_cpu(info->magic) == XFS_ATTR_LEAF_MAGIC);
+	leaf = (xfs_attr_leafblock_t *)info;
+	count = be16_to_cpu(leaf->hdr.count);
+	bytes = sizeof(xfs_attr_leaf_hdr_t) +
+		count * sizeof(xfs_attr_leaf_entry_t) +
+		be16_to_cpu(leaf->hdr.usedbytes);
+	if (bytes > (state->blocksize >> 1)) {
+		*action = 0;	/* blk over 50%, don't try to join */
+		return(0);
+	}
+
+	/*
+	 * Check for the degenerate case of the block being empty.
+	 * If the block is empty, we'll simply delete it, no need to
+	 * coalesce it with a sibling block.  We choose (arbitrarily)
+	 * to merge with the forward block unless it is NULL.
+	 */
+	if (count == 0) {
+		/*
+		 * Make altpath point to the block we want to keep and
+		 * path point to the block we want to drop (this one).
+		 */
+		forward = (info->forw != 0);
+		memcpy(&state->altpath, &state->path, sizeof(state->path));
+		error = xfs_da_path_shift(state, &state->altpath, forward,
+						 0, &retval);
+		if (error)
+			return(error);
+		if (retval) {
+			*action = 0;
+		} else {
+			*action = 2;
+		}
+		return(0);
+	}
+
+	/*
+	 * Examine each sibling block to see if we can coalesce with
+	 * at least 25% free space to spare.  We need to figure out
+	 * whether to merge with the forward or the backward block.
+	 * We prefer coalescing with the lower numbered sibling so as
+	 * to shrink an attribute list over time.
+	 */
+	/* start with smaller blk num */
+	forward = (be32_to_cpu(info->forw) < be32_to_cpu(info->back));
+	for (i = 0; i < 2; forward = !forward, i++) {
+		if (forward)
+			blkno = be32_to_cpu(info->forw);
+		else
+			blkno = be32_to_cpu(info->back);
+		if (blkno == 0)
+			continue;
+		error = xfs_da_read_buf(state->args->trans, state->args->dp,
+					blkno, -1, &bp, XFS_ATTR_FORK);
+		if (error)
+			return(error);
+		ASSERT(bp != NULL);
+
+		leaf = (xfs_attr_leafblock_t *)info;
+		count  = be16_to_cpu(leaf->hdr.count);
+		bytes  = state->blocksize - (state->blocksize>>2);
+		bytes -= be16_to_cpu(leaf->hdr.usedbytes);
+		leaf = bp->data;
+		ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_ATTR_LEAF_MAGIC);
+		count += be16_to_cpu(leaf->hdr.count);
+		bytes -= be16_to_cpu(leaf->hdr.usedbytes);
+		bytes -= count * sizeof(xfs_attr_leaf_entry_t);
+		bytes -= sizeof(xfs_attr_leaf_hdr_t);
+		xfs_da_brelse(state->args->trans, bp);
+		if (bytes >= 0)
+			break;	/* fits with at least 25% to spare */
+	}
+	if (i >= 2) {
+		*action = 0;
+		return(0);
+	}
+
+	/*
+	 * Make altpath point to the block we want to keep (the lower
+	 * numbered block) and path point to the block we want to drop.
+	 */
+	memcpy(&state->altpath, &state->path, sizeof(state->path));
+	if (blkno < blk->blkno) {
+		error = xfs_da_path_shift(state, &state->altpath, forward,
+						 0, &retval);
+	} else {
+		error = xfs_da_path_shift(state, &state->path, forward,
+						 0, &retval);
+	}
+	if (error)
+		return(error);
+	if (retval) {
+		*action = 0;
+	} else {
+		*action = 1;
+	}
+	return(0);
+}
+
+/*
+ * Remove a name from the leaf attribute list structure.
+ *
+ * Return 1 if leaf is less than 37% full, 0 if >= 37% full.
+ * If two leaves are 37% full, when combined they will leave 25% free.
+ */
+int
+xfs_attr_leaf_remove(xfs_dabuf_t *bp, xfs_da_args_t *args)
+{
+	xfs_attr_leafblock_t *leaf;
+	xfs_attr_leaf_hdr_t *hdr;
+	xfs_attr_leaf_map_t *map;
+	xfs_attr_leaf_entry_t *entry;
+	int before, after, smallest, entsize;
+	int tablesize, tmp, i;
+	xfs_mount_t *mp;
+
+	leaf = bp->data;
+	ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_ATTR_LEAF_MAGIC);
+	hdr = &leaf->hdr;
+	mp = args->trans->t_mountp;
+	ASSERT((be16_to_cpu(hdr->count) > 0)
+		&& (be16_to_cpu(hdr->count) < (XFS_LBSIZE(mp)/8)));
+	ASSERT((args->index >= 0)
+		&& (args->index < be16_to_cpu(hdr->count)));
+	ASSERT(be16_to_cpu(hdr->firstused) >=
+	       ((be16_to_cpu(hdr->count) * sizeof(*entry)) + sizeof(*hdr)));
+	entry = &leaf->entries[args->index];
+	ASSERT(be16_to_cpu(entry->nameidx) >= be16_to_cpu(hdr->firstused));
+	ASSERT(be16_to_cpu(entry->nameidx) < XFS_LBSIZE(mp));
+
+	/*
+	 * Scan through free region table:
+	 *    check for adjacency of free'd entry with an existing one,
+	 *    find smallest free region in case we need to replace it,
+	 *    adjust any map that borders the entry table,
+	 */
+	tablesize = be16_to_cpu(hdr->count) * sizeof(xfs_attr_leaf_entry_t)
+					+ sizeof(xfs_attr_leaf_hdr_t);
+	map = &hdr->freemap[0];
+	tmp = be16_to_cpu(map->size);
+	before = after = -1;
+	smallest = XFS_ATTR_LEAF_MAPSIZE - 1;
+	entsize = xfs_attr_leaf_entsize(leaf, args->index);
+	for (i = 0; i < XFS_ATTR_LEAF_MAPSIZE; map++, i++) {
+		ASSERT(be16_to_cpu(map->base) < XFS_LBSIZE(mp));
+		ASSERT(be16_to_cpu(map->size) < XFS_LBSIZE(mp));
+		if (be16_to_cpu(map->base) == tablesize) {
+			be16_add_cpu(&map->base,
+				 -((int)sizeof(xfs_attr_leaf_entry_t)));
+			be16_add_cpu(&map->size, sizeof(xfs_attr_leaf_entry_t));
+		}
+
+		if ((be16_to_cpu(map->base) + be16_to_cpu(map->size))
+				== be16_to_cpu(entry->nameidx)) {
+			before = i;
+		} else if (be16_to_cpu(map->base)
+			== (be16_to_cpu(entry->nameidx) + entsize)) {
+			after = i;
+		} else if (be16_to_cpu(map->size) < tmp) {
+			tmp = be16_to_cpu(map->size);
+			smallest = i;
+		}
+	}
+
+	/*
+	 * Coalesce adjacent freemap regions,
+	 * or replace the smallest region.
+	 */
+	if ((before >= 0) || (after >= 0)) {
+		if ((before >= 0) && (after >= 0)) {
+			map = &hdr->freemap[before];
+			be16_add_cpu(&map->size, entsize);
+			be16_add_cpu(&map->size,
+				 be16_to_cpu(hdr->freemap[after].size));
+			hdr->freemap[after].base = 0;
+			hdr->freemap[after].size = 0;
+		} else if (before >= 0) {
+			map = &hdr->freemap[before];
+			be16_add_cpu(&map->size, entsize);
+		} else {
+			map = &hdr->freemap[after];
+			/* both on-disk, don't endian flip twice */
+			map->base = entry->nameidx;
+			be16_add_cpu(&map->size, entsize);
+		}
+	} else {
+		/*
+		 * Replace smallest region (if it is smaller than free'd entry)
+		 */
+		map = &hdr->freemap[smallest];
+		if (be16_to_cpu(map->size) < entsize) {
+			map->base = cpu_to_be16(be16_to_cpu(entry->nameidx));
+			map->size = cpu_to_be16(entsize);
+		}
+	}
+
+	/*
+	 * Did we remove the first entry?
+	 */
+	if (be16_to_cpu(entry->nameidx) == be16_to_cpu(hdr->firstused))
+		smallest = 1;
+	else
+		smallest = 0;
+
+	/*
+	 * Compress the remaining entries and zero out the removed stuff.
+	 */
+	memset(xfs_attr_leaf_name(leaf, args->index), 0, entsize);
+	be16_add_cpu(&hdr->usedbytes, -entsize);
+	xfs_da_log_buf(args->trans, bp,
+	     XFS_DA_LOGRANGE(leaf, xfs_attr_leaf_name(leaf, args->index),
+				   entsize));
+
+	tmp = (be16_to_cpu(hdr->count) - args->index)
+					* sizeof(xfs_attr_leaf_entry_t);
+	memmove((char *)entry, (char *)(entry+1), tmp);
+	be16_add_cpu(&hdr->count, -1);
+	xfs_da_log_buf(args->trans, bp,
+	    XFS_DA_LOGRANGE(leaf, entry, tmp + sizeof(*entry)));
+	entry = &leaf->entries[be16_to_cpu(hdr->count)];
+	memset((char *)entry, 0, sizeof(xfs_attr_leaf_entry_t));
+
+	/*
+	 * If we removed the first entry, re-find the first used byte
+	 * in the name area.  Note that if the entry was the "firstused",
+	 * then we don't have a "hole" in our block resulting from
+	 * removing the name.
+	 */
+	if (smallest) {
+		tmp = XFS_LBSIZE(mp);
+		entry = &leaf->entries[0];
+		for (i = be16_to_cpu(hdr->count)-1; i >= 0; entry++, i--) {
+			ASSERT(be16_to_cpu(entry->nameidx) >=
+			       be16_to_cpu(hdr->firstused));
+			ASSERT(be16_to_cpu(entry->nameidx) < XFS_LBSIZE(mp));
+
+			if (be16_to_cpu(entry->nameidx) < tmp)
+				tmp = be16_to_cpu(entry->nameidx);
+		}
+		hdr->firstused = cpu_to_be16(tmp);
+		if (!hdr->firstused) {
+			hdr->firstused = cpu_to_be16(
+					tmp - XFS_ATTR_LEAF_NAME_ALIGN);
+		}
+	} else {
+		hdr->holes = 1;		/* mark as needing compaction */
+	}
+	xfs_da_log_buf(args->trans, bp,
+			  XFS_DA_LOGRANGE(leaf, hdr, sizeof(*hdr)));
+
+	/*
+	 * Check if leaf is less than 50% full, caller may want to
+	 * "join" the leaf with a sibling if so.
+	 */
+	tmp  = sizeof(xfs_attr_leaf_hdr_t);
+	tmp += be16_to_cpu(leaf->hdr.count) * sizeof(xfs_attr_leaf_entry_t);
+	tmp += be16_to_cpu(leaf->hdr.usedbytes);
+	return(tmp < mp->m_attr_magicpct); /* leaf is < 37% full */
+}
+
+/*
+ * Move all the attribute list entries from drop_leaf into save_leaf.
+ */
+void
+xfs_attr_leaf_unbalance(xfs_da_state_t *state, xfs_da_state_blk_t *drop_blk,
+				       xfs_da_state_blk_t *save_blk)
+{
+	xfs_attr_leafblock_t *drop_leaf, *save_leaf, *tmp_leaf;
+	xfs_attr_leaf_hdr_t *drop_hdr, *save_hdr, *tmp_hdr;
+	xfs_mount_t *mp;
+	char *tmpbuffer;
+
+	/*
+	 * Set up environment.
+	 */
+	mp = state->mp;
+	ASSERT(drop_blk->magic == XFS_ATTR_LEAF_MAGIC);
+	ASSERT(save_blk->magic == XFS_ATTR_LEAF_MAGIC);
+	drop_leaf = drop_blk->bp->data;
+	save_leaf = save_blk->bp->data;
+	ASSERT(be16_to_cpu(drop_leaf->hdr.info.magic) == XFS_ATTR_LEAF_MAGIC);
+	ASSERT(be16_to_cpu(save_leaf->hdr.info.magic) == XFS_ATTR_LEAF_MAGIC);
+	drop_hdr = &drop_leaf->hdr;
+	save_hdr = &save_leaf->hdr;
+
+	/*
+	 * Save last hashval from dying block for later Btree fixup.
+	 */
+	drop_blk->hashval = be32_to_cpu(
+		drop_leaf->entries[be16_to_cpu(drop_leaf->hdr.count)-1].hashval);
+
+	/*
+	 * Check if we need a temp buffer, or can we do it in place.
+	 * Note that we don't check "leaf" for holes because we will
+	 * always be dropping it, toosmall() decided that for us already.
+	 */
+	if (save_hdr->holes == 0) {
+		/*
+		 * dest leaf has no holes, so we add there.  May need
+		 * to make some room in the entry array.
+		 */
+		if (xfs_attr_leaf_order(save_blk->bp, drop_blk->bp)) {
+			xfs_attr_leaf_moveents(drop_leaf, 0, save_leaf, 0,
+			     be16_to_cpu(drop_hdr->count), mp);
+		} else {
+			xfs_attr_leaf_moveents(drop_leaf, 0, save_leaf,
+				  be16_to_cpu(save_hdr->count),
+				  be16_to_cpu(drop_hdr->count), mp);
+		}
+	} else {
+		/*
+		 * Destination has holes, so we make a temporary copy
+		 * of the leaf and add them both to that.
+		 */
+		tmpbuffer = kmem_alloc(state->blocksize, KM_SLEEP);
+		ASSERT(tmpbuffer != NULL);
+		memset(tmpbuffer, 0, state->blocksize);
+		tmp_leaf = (xfs_attr_leafblock_t *)tmpbuffer;
+		tmp_hdr = &tmp_leaf->hdr;
+		tmp_hdr->info = save_hdr->info;	/* struct copy */
+		tmp_hdr->count = 0;
+		tmp_hdr->firstused = cpu_to_be16(state->blocksize);
+		if (!tmp_hdr->firstused) {
+			tmp_hdr->firstused = cpu_to_be16(
+				state->blocksize - XFS_ATTR_LEAF_NAME_ALIGN);
+		}
+		tmp_hdr->usedbytes = 0;
+		if (xfs_attr_leaf_order(save_blk->bp, drop_blk->bp)) {
+			xfs_attr_leaf_moveents(drop_leaf, 0, tmp_leaf, 0,
+				be16_to_cpu(drop_hdr->count), mp);
+			xfs_attr_leaf_moveents(save_leaf, 0, tmp_leaf,
+				  be16_to_cpu(tmp_leaf->hdr.count),
+				  be16_to_cpu(save_hdr->count), mp);
+		} else {
+			xfs_attr_leaf_moveents(save_leaf, 0, tmp_leaf, 0,
+				be16_to_cpu(save_hdr->count), mp);
+			xfs_attr_leaf_moveents(drop_leaf, 0, tmp_leaf,
+				be16_to_cpu(tmp_leaf->hdr.count),
+				be16_to_cpu(drop_hdr->count), mp);
+		}
+		memcpy((char *)save_leaf, (char *)tmp_leaf, state->blocksize);
+		kmem_free(tmpbuffer);
+	}
+
+	xfs_da_log_buf(state->args->trans, save_blk->bp, 0,
+					   state->blocksize - 1);
+
+	/*
+	 * Copy out last hashval in each block for B-tree code.
+	 */
+	save_blk->hashval = be32_to_cpu(
+		save_leaf->entries[be16_to_cpu(save_leaf->hdr.count)-1].hashval);
+}
+
+/*========================================================================
+ * Routines used for finding things in the Btree.
+ *========================================================================*/
+
+/*
+ * Look up a name in a leaf attribute list structure.
+ * This is the internal routine, it uses the caller's buffer.
+ *
+ * Note that duplicate keys are allowed, but only check within the
+ * current leaf node.  The Btree code must check in adjacent leaf nodes.
+ *
+ * Return in args->index the index into the entry[] array of either
+ * the found entry, or where the entry should have been (insert before
+ * that entry).
+ *
+ * Don't change the args->value unless we find the attribute.
+ */
+int
+xfs_attr_leaf_lookup_int(xfs_dabuf_t *bp, xfs_da_args_t *args)
+{
+	xfs_attr_leafblock_t *leaf;
+	xfs_attr_leaf_entry_t *entry;
+	xfs_attr_leaf_name_local_t *name_loc;
+	xfs_attr_leaf_name_remote_t *name_rmt;
+	int probe, span;
+	xfs_dahash_t hashval;
+
+	leaf = bp->data;
+	ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_ATTR_LEAF_MAGIC);
+	ASSERT(be16_to_cpu(leaf->hdr.count)
+					< (XFS_LBSIZE(args->dp->i_mount)/8));
+
+	/*
+	 * Binary search.  (note: small blocks will skip this loop)
+	 */
+	hashval = args->hashval;
+	probe = span = be16_to_cpu(leaf->hdr.count) / 2;
+	for (entry = &leaf->entries[probe]; span > 4;
+		   entry = &leaf->entries[probe]) {
+		span /= 2;
+		if (be32_to_cpu(entry->hashval) < hashval)
+			probe += span;
+		else if (be32_to_cpu(entry->hashval) > hashval)
+			probe -= span;
+		else
+			break;
+	}
+	ASSERT((probe >= 0) &&
+	       (!leaf->hdr.count
+	       || (probe < be16_to_cpu(leaf->hdr.count))));
+	ASSERT((span <= 4) || (be32_to_cpu(entry->hashval) == hashval));
+
+	/*
+	 * Since we may have duplicate hashval's, find the first matching
+	 * hashval in the leaf.
+	 */
+	while ((probe > 0) && (be32_to_cpu(entry->hashval) >= hashval)) {
+		entry--;
+		probe--;
+	}
+	while ((probe < be16_to_cpu(leaf->hdr.count)) &&
+	       (be32_to_cpu(entry->hashval) < hashval)) {
+		entry++;
+		probe++;
+	}
+	if ((probe == be16_to_cpu(leaf->hdr.count)) ||
+	    (be32_to_cpu(entry->hashval) != hashval)) {
+		args->index = probe;
+		return(XFS_ERROR(ENOATTR));
+	}
+
+	/*
+	 * Duplicate keys may be present, so search all of them for a match.
+	 */
+	for (  ; (probe < be16_to_cpu(leaf->hdr.count)) &&
+			(be32_to_cpu(entry->hashval) == hashval);
+			entry++, probe++) {
+/*
+ * GROT: Add code to remove incomplete entries.
+ */
+		/*
+		 * If we are looking for INCOMPLETE entries, show only those.
+		 * If we are looking for complete entries, show only those.
+		 */
+		if ((args->flags & XFS_ATTR_INCOMPLETE) !=
+		    (entry->flags & XFS_ATTR_INCOMPLETE)) {
+			continue;
+		}
+		if (entry->flags & XFS_ATTR_LOCAL) {
+			name_loc = xfs_attr_leaf_name_local(leaf, probe);
+			if (name_loc->namelen != args->namelen)
+				continue;
+			if (memcmp(args->name, (char *)name_loc->nameval, args->namelen) != 0)
+				continue;
+			if (!xfs_attr_namesp_match(args->flags, entry->flags))
+				continue;
+			args->index = probe;
+			return(XFS_ERROR(EEXIST));
+		} else {
+			name_rmt = xfs_attr_leaf_name_remote(leaf, probe);
+			if (name_rmt->namelen != args->namelen)
+				continue;
+			if (memcmp(args->name, (char *)name_rmt->name,
+					     args->namelen) != 0)
+				continue;
+			if (!xfs_attr_namesp_match(args->flags, entry->flags))
+				continue;
+			args->index = probe;
+			args->rmtblkno = be32_to_cpu(name_rmt->valueblk);
+			args->rmtblkcnt = XFS_B_TO_FSB(args->dp->i_mount,
+						   be32_to_cpu(name_rmt->valuelen));
+			return(XFS_ERROR(EEXIST));
+		}
+	}
+	args->index = probe;
+	return(XFS_ERROR(ENOATTR));
+}
+
+/*
+ * Get the value associated with an attribute name from a leaf attribute
+ * list structure.
+ */
+int
+xfs_attr_leaf_getvalue(xfs_dabuf_t *bp, xfs_da_args_t *args)
+{
+	int valuelen;
+	xfs_attr_leafblock_t *leaf;
+	xfs_attr_leaf_entry_t *entry;
+	xfs_attr_leaf_name_local_t *name_loc;
+	xfs_attr_leaf_name_remote_t *name_rmt;
+
+	leaf = bp->data;
+	ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_ATTR_LEAF_MAGIC);
+	ASSERT(be16_to_cpu(leaf->hdr.count)
+					< (XFS_LBSIZE(args->dp->i_mount)/8));
+	ASSERT(args->index < be16_to_cpu(leaf->hdr.count));
+
+	entry = &leaf->entries[args->index];
+	if (entry->flags & XFS_ATTR_LOCAL) {
+		name_loc = xfs_attr_leaf_name_local(leaf, args->index);
+		ASSERT(name_loc->namelen == args->namelen);
+		ASSERT(memcmp(args->name, name_loc->nameval, args->namelen) == 0);
+		valuelen = be16_to_cpu(name_loc->valuelen);
+		if (args->flags & ATTR_KERNOVAL) {
+			args->valuelen = valuelen;
+			return(0);
+		}
+		if (args->valuelen < valuelen) {
+			args->valuelen = valuelen;
+			return(XFS_ERROR(ERANGE));
+		}
+		args->valuelen = valuelen;
+		memcpy(args->value, &name_loc->nameval[args->namelen], valuelen);
+	} else {
+		name_rmt = xfs_attr_leaf_name_remote(leaf, args->index);
+		ASSERT(name_rmt->namelen == args->namelen);
+		ASSERT(memcmp(args->name, name_rmt->name, args->namelen) == 0);
+		valuelen = be32_to_cpu(name_rmt->valuelen);
+		args->rmtblkno = be32_to_cpu(name_rmt->valueblk);
+		args->rmtblkcnt = XFS_B_TO_FSB(args->dp->i_mount, valuelen);
+		if (args->flags & ATTR_KERNOVAL) {
+			args->valuelen = valuelen;
+			return(0);
+		}
+		if (args->valuelen < valuelen) {
+			args->valuelen = valuelen;
+			return(XFS_ERROR(ERANGE));
+		}
+		args->valuelen = valuelen;
+	}
+	return(0);
+}
+
+/*========================================================================
+ * Utility routines.
+ *========================================================================*/
+
+/*
+ * Move the indicated entries from one leaf to another.
+ * NOTE: this routine modifies both source and destination leaves.
+ */
+/*ARGSUSED*/
+STATIC void
+xfs_attr_leaf_moveents(xfs_attr_leafblock_t *leaf_s, int start_s,
+			xfs_attr_leafblock_t *leaf_d, int start_d,
+			int count, xfs_mount_t *mp)
+{
+	xfs_attr_leaf_hdr_t *hdr_s, *hdr_d;
+	xfs_attr_leaf_entry_t *entry_s, *entry_d;
+	int desti, tmp, i;
+
+	/*
+	 * Check for nothing to do.
+	 */
+	if (count == 0)
+		return;
+
+	/*
+	 * Set up environment.
+	 */
+	ASSERT(be16_to_cpu(leaf_s->hdr.info.magic) == XFS_ATTR_LEAF_MAGIC);
+	ASSERT(be16_to_cpu(leaf_d->hdr.info.magic) == XFS_ATTR_LEAF_MAGIC);
+	hdr_s = &leaf_s->hdr;
+	hdr_d = &leaf_d->hdr;
+	ASSERT((be16_to_cpu(hdr_s->count) > 0) &&
+	       (be16_to_cpu(hdr_s->count) < (XFS_LBSIZE(mp)/8)));
+	ASSERT(be16_to_cpu(hdr_s->firstused) >=
+		((be16_to_cpu(hdr_s->count)
+					* sizeof(*entry_s))+sizeof(*hdr_s)));
+	ASSERT(be16_to_cpu(hdr_d->count) < (XFS_LBSIZE(mp)/8));
+	ASSERT(be16_to_cpu(hdr_d->firstused) >=
+		((be16_to_cpu(hdr_d->count)
+					* sizeof(*entry_d))+sizeof(*hdr_d)));
+
+	ASSERT(start_s < be16_to_cpu(hdr_s->count));
+	ASSERT(start_d <= be16_to_cpu(hdr_d->count));
+	ASSERT(count <= be16_to_cpu(hdr_s->count));
+
+	/*
+	 * Move the entries in the destination leaf up to make a hole?
+	 */
+	if (start_d < be16_to_cpu(hdr_d->count)) {
+		tmp  = be16_to_cpu(hdr_d->count) - start_d;
+		tmp *= sizeof(xfs_attr_leaf_entry_t);
+		entry_s = &leaf_d->entries[start_d];
+		entry_d = &leaf_d->entries[start_d + count];
+		memmove((char *)entry_d, (char *)entry_s, tmp);
+	}
+
+	/*
+	 * Copy all entry's in the same (sorted) order,
+	 * but allocate attribute info packed and in sequence.
+	 */
+	entry_s = &leaf_s->entries[start_s];
+	entry_d = &leaf_d->entries[start_d];
+	desti = start_d;
+	for (i = 0; i < count; entry_s++, entry_d++, desti++, i++) {
+		ASSERT(be16_to_cpu(entry_s->nameidx)
+				>= be16_to_cpu(hdr_s->firstused));
+		tmp = xfs_attr_leaf_entsize(leaf_s, start_s + i);
+#ifdef GROT
+		/*
+		 * Code to drop INCOMPLETE entries.  Difficult to use as we
+		 * may also need to change the insertion index.  Code turned
+		 * off for 6.2, should be revisited later.
+		 */
+		if (entry_s->flags & XFS_ATTR_INCOMPLETE) { /* skip partials? */
+			memset(xfs_attr_leaf_name(leaf_s, start_s + i), 0, tmp);
+			be16_add_cpu(&hdr_s->usedbytes, -tmp);
+			be16_add_cpu(&hdr_s->count, -1);
+			entry_d--;	/* to compensate for ++ in loop hdr */
+			desti--;
+			if ((start_s + i) < offset)
+				result++;	/* insertion index adjustment */
+		} else {
+#endif /* GROT */
+			be16_add_cpu(&hdr_d->firstused, -tmp);
+			/* both on-disk, don't endian flip twice */
+			entry_d->hashval = entry_s->hashval;
+			/* both on-disk, don't endian flip twice */
+			entry_d->nameidx = hdr_d->firstused;
+			entry_d->flags = entry_s->flags;
+			ASSERT(be16_to_cpu(entry_d->nameidx) + tmp
+							<= XFS_LBSIZE(mp));
+			memmove(xfs_attr_leaf_name(leaf_d, desti),
+				xfs_attr_leaf_name(leaf_s, start_s + i), tmp);
+			ASSERT(be16_to_cpu(entry_s->nameidx) + tmp
+							<= XFS_LBSIZE(mp));
+			memset(xfs_attr_leaf_name(leaf_s, start_s + i), 0, tmp);
+			be16_add_cpu(&hdr_s->usedbytes, -tmp);
+			be16_add_cpu(&hdr_d->usedbytes, tmp);
+			be16_add_cpu(&hdr_s->count, -1);
+			be16_add_cpu(&hdr_d->count, 1);
+			tmp = be16_to_cpu(hdr_d->count)
+						* sizeof(xfs_attr_leaf_entry_t)
+						+ sizeof(xfs_attr_leaf_hdr_t);
+			ASSERT(be16_to_cpu(hdr_d->firstused) >= tmp);
+#ifdef GROT
+		}
+#endif /* GROT */
+	}
+
+	/*
+	 * Zero out the entries we just copied.
+	 */
+	if (start_s == be16_to_cpu(hdr_s->count)) {
+		tmp = count * sizeof(xfs_attr_leaf_entry_t);
+		entry_s = &leaf_s->entries[start_s];
+		ASSERT(((char *)entry_s + tmp) <=
+		       ((char *)leaf_s + XFS_LBSIZE(mp)));
+		memset((char *)entry_s, 0, tmp);
+	} else {
+		/*
+		 * Move the remaining entries down to fill the hole,
+		 * then zero the entries at the top.
+		 */
+		tmp  = be16_to_cpu(hdr_s->count) - count;
+		tmp *= sizeof(xfs_attr_leaf_entry_t);
+		entry_s = &leaf_s->entries[start_s + count];
+		entry_d = &leaf_s->entries[start_s];
+		memmove((char *)entry_d, (char *)entry_s, tmp);
+
+		tmp = count * sizeof(xfs_attr_leaf_entry_t);
+		entry_s = &leaf_s->entries[be16_to_cpu(hdr_s->count)];
+		ASSERT(((char *)entry_s + tmp) <=
+		       ((char *)leaf_s + XFS_LBSIZE(mp)));
+		memset((char *)entry_s, 0, tmp);
+	}
+
+	/*
+	 * Fill in the freemap information
+	 */
+	hdr_d->freemap[0].base = cpu_to_be16(sizeof(xfs_attr_leaf_hdr_t));
+	be16_add_cpu(&hdr_d->freemap[0].base, be16_to_cpu(hdr_d->count) *
+			sizeof(xfs_attr_leaf_entry_t));
+	hdr_d->freemap[0].size = cpu_to_be16(be16_to_cpu(hdr_d->firstused)
+			      - be16_to_cpu(hdr_d->freemap[0].base));
+	hdr_d->freemap[1].base = 0;
+	hdr_d->freemap[2].base = 0;
+	hdr_d->freemap[1].size = 0;
+	hdr_d->freemap[2].size = 0;
+	hdr_s->holes = 1;	/* leaf may not be compact */
+}
+
+/*
+ * Compare two leaf blocks "order".
+ * Return 0 unless leaf2 should go before leaf1.
+ */
+int
+xfs_attr_leaf_order(xfs_dabuf_t *leaf1_bp, xfs_dabuf_t *leaf2_bp)
+{
+	xfs_attr_leafblock_t *leaf1, *leaf2;
+
+	leaf1 = leaf1_bp->data;
+	leaf2 = leaf2_bp->data;
+	ASSERT((be16_to_cpu(leaf1->hdr.info.magic) == XFS_ATTR_LEAF_MAGIC) &&
+	       (be16_to_cpu(leaf2->hdr.info.magic) == XFS_ATTR_LEAF_MAGIC));
+	if ((be16_to_cpu(leaf1->hdr.count) > 0) &&
+	    (be16_to_cpu(leaf2->hdr.count) > 0) &&
+	    ((be32_to_cpu(leaf2->entries[0].hashval) <
+	      be32_to_cpu(leaf1->entries[0].hashval)) ||
+	     (be32_to_cpu(leaf2->entries[
+			be16_to_cpu(leaf2->hdr.count)-1].hashval) <
+	      be32_to_cpu(leaf1->entries[
+			be16_to_cpu(leaf1->hdr.count)-1].hashval)))) {
+		return(1);
+	}
+	return(0);
+}
+
+/*
+ * Pick up the last hashvalue from a leaf block.
+ */
+xfs_dahash_t
+xfs_attr_leaf_lasthash(xfs_dabuf_t *bp, int *count)
+{
+	xfs_attr_leafblock_t *leaf;
+
+	leaf = bp->data;
+	ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_ATTR_LEAF_MAGIC);
+	if (count)
+		*count = be16_to_cpu(leaf->hdr.count);
+	if (!leaf->hdr.count)
+		return(0);
+	return be32_to_cpu(leaf->entries[be16_to_cpu(leaf->hdr.count)-1].hashval);
+}
+
+/*
+ * Calculate the number of bytes used to store the indicated attribute
+ * (whether local or remote only calculate bytes in this block).
+ */
+STATIC int
+xfs_attr_leaf_entsize(xfs_attr_leafblock_t *leaf, int index)
+{
+	xfs_attr_leaf_name_local_t *name_loc;
+	xfs_attr_leaf_name_remote_t *name_rmt;
+	int size;
+
+	ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_ATTR_LEAF_MAGIC);
+	if (leaf->entries[index].flags & XFS_ATTR_LOCAL) {
+		name_loc = xfs_attr_leaf_name_local(leaf, index);
+		size = xfs_attr_leaf_entsize_local(name_loc->namelen,
+						   be16_to_cpu(name_loc->valuelen));
+	} else {
+		name_rmt = xfs_attr_leaf_name_remote(leaf, index);
+		size = xfs_attr_leaf_entsize_remote(name_rmt->namelen);
+	}
+	return(size);
+}
+
+/*
+ * Calculate the number of bytes that would be required to store the new
+ * attribute (whether local or remote only calculate bytes in this block).
+ * This routine decides as a side effect whether the attribute will be
+ * a "local" or a "remote" attribute.
+ */
+int
+xfs_attr_leaf_newentsize(int namelen, int valuelen, int blocksize, int *local)
+{
+	int size;
+
+	size = xfs_attr_leaf_entsize_local(namelen, valuelen);
+	if (size < xfs_attr_leaf_entsize_local_max(blocksize)) {
+		if (local) {
+			*local = 1;
+		}
+	} else {
+		size = xfs_attr_leaf_entsize_remote(namelen);
+		if (local) {
+			*local = 0;
+		}
+	}
+	return(size);
+}
+
+/*
+ * Copy out attribute list entries for attr_list(), for leaf attribute lists.
+ */
+int
+xfs_attr_leaf_list_int(xfs_dabuf_t *bp, xfs_attr_list_context_t *context)
+{
+	attrlist_cursor_kern_t *cursor;
+	xfs_attr_leafblock_t *leaf;
+	xfs_attr_leaf_entry_t *entry;
+	int retval, i;
+
+	ASSERT(bp != NULL);
+	leaf = bp->data;
+	cursor = context->cursor;
+	cursor->initted = 1;
+
+	trace_xfs_attr_list_leaf(context);
+
+	/*
+	 * Re-find our place in the leaf block if this is a new syscall.
+	 */
+	if (context->resynch) {
+		entry = &leaf->entries[0];
+		for (i = 0; i < be16_to_cpu(leaf->hdr.count); entry++, i++) {
+			if (be32_to_cpu(entry->hashval) == cursor->hashval) {
+				if (cursor->offset == context->dupcnt) {
+					context->dupcnt = 0;
+					break;
+				}
+				context->dupcnt++;
+			} else if (be32_to_cpu(entry->hashval) >
+					cursor->hashval) {
+				context->dupcnt = 0;
+				break;
+			}
+		}
+		if (i == be16_to_cpu(leaf->hdr.count)) {
+			trace_xfs_attr_list_notfound(context);
+			return(0);
+		}
+	} else {
+		entry = &leaf->entries[0];
+		i = 0;
+	}
+	context->resynch = 0;
+
+	/*
+	 * We have found our place, start copying out the new attributes.
+	 */
+	retval = 0;
+	for (  ; (i < be16_to_cpu(leaf->hdr.count)); entry++, i++) {
+		if (be32_to_cpu(entry->hashval) != cursor->hashval) {
+			cursor->hashval = be32_to_cpu(entry->hashval);
+			cursor->offset = 0;
+		}
+
+		if (entry->flags & XFS_ATTR_INCOMPLETE)
+			continue;		/* skip incomplete entries */
+
+		if (entry->flags & XFS_ATTR_LOCAL) {
+			xfs_attr_leaf_name_local_t *name_loc =
+				xfs_attr_leaf_name_local(leaf, i);
+
+			retval = context->put_listent(context,
+						entry->flags,
+						name_loc->nameval,
+						(int)name_loc->namelen,
+						be16_to_cpu(name_loc->valuelen),
+						&name_loc->nameval[name_loc->namelen]);
+			if (retval)
+				return retval;
+		} else {
+			xfs_attr_leaf_name_remote_t *name_rmt =
+				xfs_attr_leaf_name_remote(leaf, i);
+
+			int valuelen = be32_to_cpu(name_rmt->valuelen);
+
+			if (context->put_value) {
+				xfs_da_args_t args;
+
+				memset((char *)&args, 0, sizeof(args));
+				args.dp = context->dp;
+				args.whichfork = XFS_ATTR_FORK;
+				args.valuelen = valuelen;
+				args.value = kmem_alloc(valuelen, KM_SLEEP | KM_NOFS);
+				args.rmtblkno = be32_to_cpu(name_rmt->valueblk);
+				args.rmtblkcnt = XFS_B_TO_FSB(args.dp->i_mount, valuelen);
+				retval = xfs_attr_rmtval_get(&args);
+				if (retval)
+					return retval;
+				retval = context->put_listent(context,
+						entry->flags,
+						name_rmt->name,
+						(int)name_rmt->namelen,
+						valuelen,
+						args.value);
+				kmem_free(args.value);
+			} else {
+				retval = context->put_listent(context,
+						entry->flags,
+						name_rmt->name,
+						(int)name_rmt->namelen,
+						valuelen,
+						NULL);
+			}
+			if (retval)
+				return retval;
+		}
+		if (context->seen_enough)
+			break;
+		cursor->offset++;
+	}
+	trace_xfs_attr_list_leaf_end(context);
+	return(retval);
+}
+
+
+/*========================================================================
+ * Manage the INCOMPLETE flag in a leaf entry
+ *========================================================================*/
+
+/*
+ * Clear the INCOMPLETE flag on an entry in a leaf block.
+ */
+int
+xfs_attr_leaf_clearflag(xfs_da_args_t *args)
+{
+	xfs_attr_leafblock_t *leaf;
+	xfs_attr_leaf_entry_t *entry;
+	xfs_attr_leaf_name_remote_t *name_rmt;
+	xfs_dabuf_t *bp;
+	int error;
+#ifdef DEBUG
+	xfs_attr_leaf_name_local_t *name_loc;
+	int namelen;
+	char *name;
+#endif /* DEBUG */
+
+	/*
+	 * Set up the operation.
+	 */
+	error = xfs_da_read_buf(args->trans, args->dp, args->blkno, -1, &bp,
+					     XFS_ATTR_FORK);
+	if (error) {
+		return(error);
+	}
+	ASSERT(bp != NULL);
+
+	leaf = bp->data;
+	ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_ATTR_LEAF_MAGIC);
+	ASSERT(args->index < be16_to_cpu(leaf->hdr.count));
+	ASSERT(args->index >= 0);
+	entry = &leaf->entries[ args->index ];
+	ASSERT(entry->flags & XFS_ATTR_INCOMPLETE);
+
+#ifdef DEBUG
+	if (entry->flags & XFS_ATTR_LOCAL) {
+		name_loc = xfs_attr_leaf_name_local(leaf, args->index);
+		namelen = name_loc->namelen;
+		name = (char *)name_loc->nameval;
+	} else {
+		name_rmt = xfs_attr_leaf_name_remote(leaf, args->index);
+		namelen = name_rmt->namelen;
+		name = (char *)name_rmt->name;
+	}
+	ASSERT(be32_to_cpu(entry->hashval) == args->hashval);
+	ASSERT(namelen == args->namelen);
+	ASSERT(memcmp(name, args->name, namelen) == 0);
+#endif /* DEBUG */
+
+	entry->flags &= ~XFS_ATTR_INCOMPLETE;
+	xfs_da_log_buf(args->trans, bp,
+			 XFS_DA_LOGRANGE(leaf, entry, sizeof(*entry)));
+
+	if (args->rmtblkno) {
+		ASSERT((entry->flags & XFS_ATTR_LOCAL) == 0);
+		name_rmt = xfs_attr_leaf_name_remote(leaf, args->index);
+		name_rmt->valueblk = cpu_to_be32(args->rmtblkno);
+		name_rmt->valuelen = cpu_to_be32(args->valuelen);
+		xfs_da_log_buf(args->trans, bp,
+			 XFS_DA_LOGRANGE(leaf, name_rmt, sizeof(*name_rmt)));
+	}
+	xfs_da_buf_done(bp);
+
+	/*
+	 * Commit the flag value change and start the next trans in series.
+	 */
+	return xfs_trans_roll(&args->trans, args->dp);
+}
+
+/*
+ * Set the INCOMPLETE flag on an entry in a leaf block.
+ */
+int
+xfs_attr_leaf_setflag(xfs_da_args_t *args)
+{
+	xfs_attr_leafblock_t *leaf;
+	xfs_attr_leaf_entry_t *entry;
+	xfs_attr_leaf_name_remote_t *name_rmt;
+	xfs_dabuf_t *bp;
+	int error;
+
+	/*
+	 * Set up the operation.
+	 */
+	error = xfs_da_read_buf(args->trans, args->dp, args->blkno, -1, &bp,
+					     XFS_ATTR_FORK);
+	if (error) {
+		return(error);
+	}
+	ASSERT(bp != NULL);
+
+	leaf = bp->data;
+	ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_ATTR_LEAF_MAGIC);
+	ASSERT(args->index < be16_to_cpu(leaf->hdr.count));
+	ASSERT(args->index >= 0);
+	entry = &leaf->entries[ args->index ];
+
+	ASSERT((entry->flags & XFS_ATTR_INCOMPLETE) == 0);
+	entry->flags |= XFS_ATTR_INCOMPLETE;
+	xfs_da_log_buf(args->trans, bp,
+			XFS_DA_LOGRANGE(leaf, entry, sizeof(*entry)));
+	if ((entry->flags & XFS_ATTR_LOCAL) == 0) {
+		name_rmt = xfs_attr_leaf_name_remote(leaf, args->index);
+		name_rmt->valueblk = 0;
+		name_rmt->valuelen = 0;
+		xfs_da_log_buf(args->trans, bp,
+			 XFS_DA_LOGRANGE(leaf, name_rmt, sizeof(*name_rmt)));
+	}
+	xfs_da_buf_done(bp);
+
+	/*
+	 * Commit the flag value change and start the next trans in series.
+	 */
+	return xfs_trans_roll(&args->trans, args->dp);
+}
+
+/*
+ * In a single transaction, clear the INCOMPLETE flag on the leaf entry
+ * given by args->blkno/index and set the INCOMPLETE flag on the leaf
+ * entry given by args->blkno2/index2.
+ *
+ * Note that they could be in different blocks, or in the same block.
+ */
+int
+xfs_attr_leaf_flipflags(xfs_da_args_t *args)
+{
+	xfs_attr_leafblock_t *leaf1, *leaf2;
+	xfs_attr_leaf_entry_t *entry1, *entry2;
+	xfs_attr_leaf_name_remote_t *name_rmt;
+	xfs_dabuf_t *bp1, *bp2;
+	int error;
+#ifdef DEBUG
+	xfs_attr_leaf_name_local_t *name_loc;
+	int namelen1, namelen2;
+	char *name1, *name2;
+#endif /* DEBUG */
+
+	/*
+	 * Read the block containing the "old" attr
+	 */
+	error = xfs_da_read_buf(args->trans, args->dp, args->blkno, -1, &bp1,
+					     XFS_ATTR_FORK);
+	if (error) {
+		return(error);
+	}
+	ASSERT(bp1 != NULL);
+
+	/*
+	 * Read the block containing the "new" attr, if it is different
+	 */
+	if (args->blkno2 != args->blkno) {
+		error = xfs_da_read_buf(args->trans, args->dp, args->blkno2,
+					-1, &bp2, XFS_ATTR_FORK);
+		if (error) {
+			return(error);
+		}
+		ASSERT(bp2 != NULL);
+	} else {
+		bp2 = bp1;
+	}
+
+	leaf1 = bp1->data;
+	ASSERT(be16_to_cpu(leaf1->hdr.info.magic) == XFS_ATTR_LEAF_MAGIC);
+	ASSERT(args->index < be16_to_cpu(leaf1->hdr.count));
+	ASSERT(args->index >= 0);
+	entry1 = &leaf1->entries[ args->index ];
+
+	leaf2 = bp2->data;
+	ASSERT(be16_to_cpu(leaf2->hdr.info.magic) == XFS_ATTR_LEAF_MAGIC);
+	ASSERT(args->index2 < be16_to_cpu(leaf2->hdr.count));
+	ASSERT(args->index2 >= 0);
+	entry2 = &leaf2->entries[ args->index2 ];
+
+#ifdef DEBUG
+	if (entry1->flags & XFS_ATTR_LOCAL) {
+		name_loc = xfs_attr_leaf_name_local(leaf1, args->index);
+		namelen1 = name_loc->namelen;
+		name1 = (char *)name_loc->nameval;
+	} else {
+		name_rmt = xfs_attr_leaf_name_remote(leaf1, args->index);
+		namelen1 = name_rmt->namelen;
+		name1 = (char *)name_rmt->name;
+	}
+	if (entry2->flags & XFS_ATTR_LOCAL) {
+		name_loc = xfs_attr_leaf_name_local(leaf2, args->index2);
+		namelen2 = name_loc->namelen;
+		name2 = (char *)name_loc->nameval;
+	} else {
+		name_rmt = xfs_attr_leaf_name_remote(leaf2, args->index2);
+		namelen2 = name_rmt->namelen;
+		name2 = (char *)name_rmt->name;
+	}
+	ASSERT(be32_to_cpu(entry1->hashval) == be32_to_cpu(entry2->hashval));
+	ASSERT(namelen1 == namelen2);
+	ASSERT(memcmp(name1, name2, namelen1) == 0);
+#endif /* DEBUG */
+
+	ASSERT(entry1->flags & XFS_ATTR_INCOMPLETE);
+	ASSERT((entry2->flags & XFS_ATTR_INCOMPLETE) == 0);
+
+	entry1->flags &= ~XFS_ATTR_INCOMPLETE;
+	xfs_da_log_buf(args->trans, bp1,
+			  XFS_DA_LOGRANGE(leaf1, entry1, sizeof(*entry1)));
+	if (args->rmtblkno) {
+		ASSERT((entry1->flags & XFS_ATTR_LOCAL) == 0);
+		name_rmt = xfs_attr_leaf_name_remote(leaf1, args->index);
+		name_rmt->valueblk = cpu_to_be32(args->rmtblkno);
+		name_rmt->valuelen = cpu_to_be32(args->valuelen);
+		xfs_da_log_buf(args->trans, bp1,
+			 XFS_DA_LOGRANGE(leaf1, name_rmt, sizeof(*name_rmt)));
+	}
+
+	entry2->flags |= XFS_ATTR_INCOMPLETE;
+	xfs_da_log_buf(args->trans, bp2,
+			  XFS_DA_LOGRANGE(leaf2, entry2, sizeof(*entry2)));
+	if ((entry2->flags & XFS_ATTR_LOCAL) == 0) {
+		name_rmt = xfs_attr_leaf_name_remote(leaf2, args->index2);
+		name_rmt->valueblk = 0;
+		name_rmt->valuelen = 0;
+		xfs_da_log_buf(args->trans, bp2,
+			 XFS_DA_LOGRANGE(leaf2, name_rmt, sizeof(*name_rmt)));
+	}
+	xfs_da_buf_done(bp1);
+	if (bp1 != bp2)
+		xfs_da_buf_done(bp2);
+
+	/*
+	 * Commit the flag value change and start the next trans in series.
+	 */
+	error = xfs_trans_roll(&args->trans, args->dp);
+
+	return(error);
+}
+
+/*========================================================================
+ * Indiscriminately delete the entire attribute fork
+ *========================================================================*/
+
+/*
+ * Recurse (gasp!) through the attribute nodes until we find leaves.
+ * We're doing a depth-first traversal in order to invalidate everything.
+ */
+int
+xfs_attr_root_inactive(xfs_trans_t **trans, xfs_inode_t *dp)
+{
+	xfs_da_blkinfo_t *info;
+	xfs_daddr_t blkno;
+	xfs_dabuf_t *bp;
+	int error;
+
+	/*
+	 * Read block 0 to see what we have to work with.
+	 * We only get here if we have extents, since we remove
+	 * the extents in reverse order the extent containing
+	 * block 0 must still be there.
+	 */
+	error = xfs_da_read_buf(*trans, dp, 0, -1, &bp, XFS_ATTR_FORK);
+	if (error)
+		return(error);
+	blkno = xfs_da_blkno(bp);
+
+	/*
+	 * Invalidate the tree, even if the "tree" is only a single leaf block.
+	 * This is a depth-first traversal!
+	 */
+	info = bp->data;
+	if (be16_to_cpu(info->magic) == XFS_DA_NODE_MAGIC) {
+		error = xfs_attr_node_inactive(trans, dp, bp, 1);
+	} else if (be16_to_cpu(info->magic) == XFS_ATTR_LEAF_MAGIC) {
+		error = xfs_attr_leaf_inactive(trans, dp, bp);
+	} else {
+		error = XFS_ERROR(EIO);
+		xfs_da_brelse(*trans, bp);
+	}
+	if (error)
+		return(error);
+
+	/*
+	 * Invalidate the incore copy of the root block.
+	 */
+	error = xfs_da_get_buf(*trans, dp, 0, blkno, &bp, XFS_ATTR_FORK);
+	if (error)
+		return(error);
+	xfs_da_binval(*trans, bp);	/* remove from cache */
+	/*
+	 * Commit the invalidate and start the next transaction.
+	 */
+	error = xfs_trans_roll(trans, dp);
+
+	return (error);
+}
+
+/*
+ * Recurse (gasp!) through the attribute nodes until we find leaves.
+ * We're doing a depth-first traversal in order to invalidate everything.
+ */
+STATIC int
+xfs_attr_node_inactive(xfs_trans_t **trans, xfs_inode_t *dp, xfs_dabuf_t *bp,
+				   int level)
+{
+	xfs_da_blkinfo_t *info;
+	xfs_da_intnode_t *node;
+	xfs_dablk_t child_fsb;
+	xfs_daddr_t parent_blkno, child_blkno;
+	int error, count, i;
+	xfs_dabuf_t *child_bp;
+
+	/*
+	 * Since this code is recursive (gasp!) we must protect ourselves.
+	 */
+	if (level > XFS_DA_NODE_MAXDEPTH) {
+		xfs_da_brelse(*trans, bp);	/* no locks for later trans */
+		return(XFS_ERROR(EIO));
+	}
+
+	node = bp->data;
+	ASSERT(be16_to_cpu(node->hdr.info.magic) == XFS_DA_NODE_MAGIC);
+	parent_blkno = xfs_da_blkno(bp);	/* save for re-read later */
+	count = be16_to_cpu(node->hdr.count);
+	if (!count) {
+		xfs_da_brelse(*trans, bp);
+		return(0);
+	}
+	child_fsb = be32_to_cpu(node->btree[0].before);
+	xfs_da_brelse(*trans, bp);	/* no locks for later trans */
+
+	/*
+	 * If this is the node level just above the leaves, simply loop
+	 * over the leaves removing all of them.  If this is higher up
+	 * in the tree, recurse downward.
+	 */
+	for (i = 0; i < count; i++) {
+		/*
+		 * Read the subsidiary block to see what we have to work with.
+		 * Don't do this in a transaction.  This is a depth-first
+		 * traversal of the tree so we may deal with many blocks
+		 * before we come back to this one.
+		 */
+		error = xfs_da_read_buf(*trans, dp, child_fsb, -2, &child_bp,
+						XFS_ATTR_FORK);
+		if (error)
+			return(error);
+		if (child_bp) {
+						/* save for re-read later */
+			child_blkno = xfs_da_blkno(child_bp);
+
+			/*
+			 * Invalidate the subtree, however we have to.
+			 */
+			info = child_bp->data;
+			if (be16_to_cpu(info->magic) == XFS_DA_NODE_MAGIC) {
+				error = xfs_attr_node_inactive(trans, dp,
+						child_bp, level+1);
+			} else if (be16_to_cpu(info->magic) == XFS_ATTR_LEAF_MAGIC) {
+				error = xfs_attr_leaf_inactive(trans, dp,
+						child_bp);
+			} else {
+				error = XFS_ERROR(EIO);
+				xfs_da_brelse(*trans, child_bp);
+			}
+			if (error)
+				return(error);
+
+			/*
+			 * Remove the subsidiary block from the cache
+			 * and from the log.
+			 */
+			error = xfs_da_get_buf(*trans, dp, 0, child_blkno,
+				&child_bp, XFS_ATTR_FORK);
+			if (error)
+				return(error);
+			xfs_da_binval(*trans, child_bp);
+		}
+
+		/*
+		 * If we're not done, re-read the parent to get the next
+		 * child block number.
+		 */
+		if ((i+1) < count) {
+			error = xfs_da_read_buf(*trans, dp, 0, parent_blkno,
+				&bp, XFS_ATTR_FORK);
+			if (error)
+				return(error);
+			child_fsb = be32_to_cpu(node->btree[i+1].before);
+			xfs_da_brelse(*trans, bp);
+		}
+		/*
+		 * Atomically commit the whole invalidate stuff.
+		 */
+		error = xfs_trans_roll(trans, dp);
+		if (error)
+			return (error);
+	}
+
+	return(0);
+}
+
+/*
+ * Invalidate all of the "remote" value regions pointed to by a particular
+ * leaf block.
+ * Note that we must release the lock on the buffer so that we are not
+ * caught holding something that the logging code wants to flush to disk.
+ */
+STATIC int
+xfs_attr_leaf_inactive(xfs_trans_t **trans, xfs_inode_t *dp, xfs_dabuf_t *bp)
+{
+	xfs_attr_leafblock_t *leaf;
+	xfs_attr_leaf_entry_t *entry;
+	xfs_attr_leaf_name_remote_t *name_rmt;
+	xfs_attr_inactive_list_t *list, *lp;
+	int error, count, size, tmp, i;
+
+	leaf = bp->data;
+	ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_ATTR_LEAF_MAGIC);
+
+	/*
+	 * Count the number of "remote" value extents.
+	 */
+	count = 0;
+	entry = &leaf->entries[0];
+	for (i = 0; i < be16_to_cpu(leaf->hdr.count); entry++, i++) {
+		if (be16_to_cpu(entry->nameidx) &&
+		    ((entry->flags & XFS_ATTR_LOCAL) == 0)) {
+			name_rmt = xfs_attr_leaf_name_remote(leaf, i);
+			if (name_rmt->valueblk)
+				count++;
+		}
+	}
+
+	/*
+	 * If there are no "remote" values, we're done.
+	 */
+	if (count == 0) {
+		xfs_da_brelse(*trans, bp);
+		return(0);
+	}
+
+	/*
+	 * Allocate storage for a list of all the "remote" value extents.
+	 */
+	size = count * sizeof(xfs_attr_inactive_list_t);
+	list = (xfs_attr_inactive_list_t *)kmem_alloc(size, KM_SLEEP);
+
+	/*
+	 * Identify each of the "remote" value extents.
+	 */
+	lp = list;
+	entry = &leaf->entries[0];
+	for (i = 0; i < be16_to_cpu(leaf->hdr.count); entry++, i++) {
+		if (be16_to_cpu(entry->nameidx) &&
+		    ((entry->flags & XFS_ATTR_LOCAL) == 0)) {
+			name_rmt = xfs_attr_leaf_name_remote(leaf, i);
+			if (name_rmt->valueblk) {
+				lp->valueblk = be32_to_cpu(name_rmt->valueblk);
+				lp->valuelen = XFS_B_TO_FSB(dp->i_mount,
+						    be32_to_cpu(name_rmt->valuelen));
+				lp++;
+			}
+		}
+	}
+	xfs_da_brelse(*trans, bp);	/* unlock for trans. in freextent() */
+
+	/*
+	 * Invalidate each of the "remote" value extents.
+	 */
+	error = 0;
+	for (lp = list, i = 0; i < count; i++, lp++) {
+		tmp = xfs_attr_leaf_freextent(trans, dp,
+				lp->valueblk, lp->valuelen);
+
+		if (error == 0)
+			error = tmp;	/* save only the 1st errno */
+	}
+
+	kmem_free((xfs_caddr_t)list);
+	return(error);
+}
+
+/*
+ * Look at all the extents for this logical region,
+ * invalidate any buffers that are incore/in transactions.
+ */
+STATIC int
+xfs_attr_leaf_freextent(xfs_trans_t **trans, xfs_inode_t *dp,
+				    xfs_dablk_t blkno, int blkcnt)
+{
+	xfs_bmbt_irec_t map;
+	xfs_dablk_t tblkno;
+	int tblkcnt, dblkcnt, nmap, error;
+	xfs_daddr_t dblkno;
+	xfs_buf_t *bp;
+
+	/*
+	 * Roll through the "value", invalidating the attribute value's
+	 * blocks.
+	 */
+	tblkno = blkno;
+	tblkcnt = blkcnt;
+	while (tblkcnt > 0) {
+		/*
+		 * Try to remember where we decided to put the value.
+		 */
+		nmap = 1;
+		error = xfs_bmapi(*trans, dp, (xfs_fileoff_t)tblkno, tblkcnt,
+					XFS_BMAPI_ATTRFORK | XFS_BMAPI_METADATA,
+					NULL, 0, &map, &nmap, NULL);
+		if (error) {
+			return(error);
+		}
+		ASSERT(nmap == 1);
+		ASSERT(map.br_startblock != DELAYSTARTBLOCK);
+
+		/*
+		 * If it's a hole, these are already unmapped
+		 * so there's nothing to invalidate.
+		 */
+		if (map.br_startblock != HOLESTARTBLOCK) {
+
+			dblkno = XFS_FSB_TO_DADDR(dp->i_mount,
+						  map.br_startblock);
+			dblkcnt = XFS_FSB_TO_BB(dp->i_mount,
+						map.br_blockcount);
+			bp = xfs_trans_get_buf(*trans,
+					dp->i_mount->m_ddev_targp,
+					dblkno, dblkcnt, XBF_LOCK);
+			xfs_trans_binval(*trans, bp);
+			/*
+			 * Roll to next transaction.
+			 */
+			error = xfs_trans_roll(trans, dp);
+			if (error)
+				return (error);
+		}
+
+		tblkno += map.br_blockcount;
+		tblkcnt -= map.br_blockcount;
+	}
+
+	return(0);
+}
diff --git a/fs/xfs/xfs_attr_leaf.h b/fs/xfs/xfs_attr_leaf.h
new file mode 100644
index 00000000..9c7d22fd
--- /dev/null
+++ b/fs/xfs/xfs_attr_leaf.h
@@ -0,0 +1,265 @@
+/*
+ * Copyright (c) 2000,2002-2003,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_ATTR_LEAF_H__
+#define	__XFS_ATTR_LEAF_H__
+
+/*
+ * Attribute storage layout, internal structure, access macros, etc.
+ *
+ * Attribute lists are structured around Btrees where all the data
+ * elements are in the leaf nodes.  Attribute names are hashed into an int,
+ * then that int is used as the index into the Btree.  Since the hashval
+ * of an attribute name may not be unique, we may have duplicate keys.  The
+ * internal links in the Btree are logical block offsets into the file.
+ */
+
+struct attrlist;
+struct attrlist_cursor_kern;
+struct xfs_attr_list_context;
+struct xfs_dabuf;
+struct xfs_da_args;
+struct xfs_da_state;
+struct xfs_da_state_blk;
+struct xfs_inode;
+struct xfs_trans;
+
+/*========================================================================
+ * Attribute structure when equal to XFS_LBSIZE(mp) bytes.
+ *========================================================================*/
+
+/*
+ * This is the structure of the leaf nodes in the Btree.
+ *
+ * Struct leaf_entry's are packed from the top.  Name/values grow from the
+ * bottom but are not packed.  The freemap contains run-length-encoded entries
+ * for the free bytes after the leaf_entry's, but only the N largest such,
+ * smaller runs are dropped.  When the freemap doesn't show enough space
+ * for an allocation, we compact the name/value area and try again.  If we
+ * still don't have enough space, then we have to split the block.  The
+ * name/value structs (both local and remote versions) must be 32bit aligned.
+ *
+ * Since we have duplicate hash keys, for each key that matches, compare
+ * the actual name string.  The root and intermediate node search always
+ * takes the first-in-the-block key match found, so we should only have
+ * to work "forw"ard.  If none matches, continue with the "forw"ard leaf
+ * nodes until the hash key changes or the attribute name is found.
+ *
+ * We store the fact that an attribute is a ROOT/USER/SECURE attribute in
+ * the leaf_entry.  The namespaces are independent only because we also look
+ * at the namespace bit when we are looking for a matching attribute name.
+ *
+ * We also store an "incomplete" bit in the leaf_entry.  It shows that an
+ * attribute is in the middle of being created and should not be shown to
+ * the user if we crash during the time that the bit is set.  We clear the
+ * bit when we have finished setting up the attribute.  We do this because
+ * we cannot create some large attributes inside a single transaction, and we
+ * need some indication that we weren't finished if we crash in the middle.
+ */
+#define XFS_ATTR_LEAF_MAPSIZE	3	/* how many freespace slots */
+
+typedef struct xfs_attr_leaf_map {	/* RLE map of free bytes */
+	__be16	base;			  /* base of free region */
+	__be16	size;			  /* length of free region */
+} xfs_attr_leaf_map_t;
+
+typedef struct xfs_attr_leaf_hdr {	/* constant-structure header block */
+	xfs_da_blkinfo_t info;		/* block type, links, etc. */
+	__be16	count;			/* count of active leaf_entry's */
+	__be16	usedbytes;		/* num bytes of names/values stored */
+	__be16	firstused;		/* first used byte in name area */
+	__u8	holes;			/* != 0 if blk needs compaction */
+	__u8	pad1;
+	xfs_attr_leaf_map_t freemap[XFS_ATTR_LEAF_MAPSIZE];
+					/* N largest free regions */
+} xfs_attr_leaf_hdr_t;
+
+typedef struct xfs_attr_leaf_entry {	/* sorted on key, not name */
+	__be32	hashval;		/* hash value of name */
+ 	__be16	nameidx;		/* index into buffer of name/value */
+	__u8	flags;			/* LOCAL/ROOT/SECURE/INCOMPLETE flag */
+	__u8	pad2;			/* unused pad byte */
+} xfs_attr_leaf_entry_t;
+
+typedef struct xfs_attr_leaf_name_local {
+	__be16	valuelen;		/* number of bytes in value */
+	__u8	namelen;		/* length of name bytes */
+	__u8	nameval[1];		/* name/value bytes */
+} xfs_attr_leaf_name_local_t;
+
+typedef struct xfs_attr_leaf_name_remote {
+	__be32	valueblk;		/* block number of value bytes */
+	__be32	valuelen;		/* number of bytes in value */
+	__u8	namelen;		/* length of name bytes */
+	__u8	name[1];		/* name bytes */
+} xfs_attr_leaf_name_remote_t;
+
+typedef struct xfs_attr_leafblock {
+	xfs_attr_leaf_hdr_t	hdr;	/* constant-structure header block */
+	xfs_attr_leaf_entry_t	entries[1];	/* sorted on key, not name */
+	xfs_attr_leaf_name_local_t namelist;	/* grows from bottom of buf */
+	xfs_attr_leaf_name_remote_t valuelist;	/* grows from bottom of buf */
+} xfs_attr_leafblock_t;
+
+/*
+ * Flags used in the leaf_entry[i].flags field.
+ * NOTE: the INCOMPLETE bit must not collide with the flags bits specified
+ * on the system call, they are "or"ed together for various operations.
+ */
+#define	XFS_ATTR_LOCAL_BIT	0	/* attr is stored locally */
+#define	XFS_ATTR_ROOT_BIT	1	/* limit access to trusted attrs */
+#define	XFS_ATTR_SECURE_BIT	2	/* limit access to secure attrs */
+#define	XFS_ATTR_INCOMPLETE_BIT	7	/* attr in middle of create/delete */
+#define XFS_ATTR_LOCAL		(1 << XFS_ATTR_LOCAL_BIT)
+#define XFS_ATTR_ROOT		(1 << XFS_ATTR_ROOT_BIT)
+#define XFS_ATTR_SECURE		(1 << XFS_ATTR_SECURE_BIT)
+#define XFS_ATTR_INCOMPLETE	(1 << XFS_ATTR_INCOMPLETE_BIT)
+
+/*
+ * Conversion macros for converting namespace bits from argument flags
+ * to ondisk flags.
+ */
+#define XFS_ATTR_NSP_ARGS_MASK		(ATTR_ROOT | ATTR_SECURE)
+#define XFS_ATTR_NSP_ONDISK_MASK	(XFS_ATTR_ROOT | XFS_ATTR_SECURE)
+#define XFS_ATTR_NSP_ONDISK(flags)	((flags) & XFS_ATTR_NSP_ONDISK_MASK)
+#define XFS_ATTR_NSP_ARGS(flags)	((flags) & XFS_ATTR_NSP_ARGS_MASK)
+#define XFS_ATTR_NSP_ARGS_TO_ONDISK(x)	(((x) & ATTR_ROOT ? XFS_ATTR_ROOT : 0) |\
+					 ((x) & ATTR_SECURE ? XFS_ATTR_SECURE : 0))
+#define XFS_ATTR_NSP_ONDISK_TO_ARGS(x)	(((x) & XFS_ATTR_ROOT ? ATTR_ROOT : 0) |\
+					 ((x) & XFS_ATTR_SECURE ? ATTR_SECURE : 0))
+
+/*
+ * Alignment for namelist and valuelist entries (since they are mixed
+ * there can be only one alignment value)
+ */
+#define	XFS_ATTR_LEAF_NAME_ALIGN	((uint)sizeof(xfs_dablk_t))
+
+/*
+ * Cast typed pointers for "local" and "remote" name/value structs.
+ */
+static inline xfs_attr_leaf_name_remote_t *
+xfs_attr_leaf_name_remote(xfs_attr_leafblock_t *leafp, int idx)
+{
+	return (xfs_attr_leaf_name_remote_t *)
+		&((char *)leafp)[be16_to_cpu(leafp->entries[idx].nameidx)];
+}
+
+static inline xfs_attr_leaf_name_local_t *
+xfs_attr_leaf_name_local(xfs_attr_leafblock_t *leafp, int idx)
+{
+	return (xfs_attr_leaf_name_local_t *)
+		&((char *)leafp)[be16_to_cpu(leafp->entries[idx].nameidx)];
+}
+
+static inline char *xfs_attr_leaf_name(xfs_attr_leafblock_t *leafp, int idx)
+{
+	return &((char *)leafp)[be16_to_cpu(leafp->entries[idx].nameidx)];
+}
+
+/*
+ * Calculate total bytes used (including trailing pad for alignment) for
+ * a "local" name/value structure, a "remote" name/value structure, and
+ * a pointer which might be either.
+ */
+static inline int xfs_attr_leaf_entsize_remote(int nlen)
+{
+	return ((uint)sizeof(xfs_attr_leaf_name_remote_t) - 1 + (nlen) + \
+		XFS_ATTR_LEAF_NAME_ALIGN - 1) & ~(XFS_ATTR_LEAF_NAME_ALIGN - 1);
+}
+
+static inline int xfs_attr_leaf_entsize_local(int nlen, int vlen)
+{
+	return ((uint)sizeof(xfs_attr_leaf_name_local_t) - 1 + (nlen) + (vlen) +
+		XFS_ATTR_LEAF_NAME_ALIGN - 1) & ~(XFS_ATTR_LEAF_NAME_ALIGN - 1);
+}
+
+static inline int xfs_attr_leaf_entsize_local_max(int bsize)
+{
+	return (((bsize) >> 1) + ((bsize) >> 2));
+}
+
+/*
+ * Used to keep a list of "remote value" extents when unlinking an inode.
+ */
+typedef struct xfs_attr_inactive_list {
+	xfs_dablk_t	valueblk;	/* block number of value bytes */
+	int		valuelen;	/* number of bytes in value */
+} xfs_attr_inactive_list_t;
+
+
+/*========================================================================
+ * Function prototypes for the kernel.
+ *========================================================================*/
+
+/*
+ * Internal routines when attribute fork size < XFS_LITINO(mp).
+ */
+void	xfs_attr_shortform_create(struct xfs_da_args *args);
+void	xfs_attr_shortform_add(struct xfs_da_args *args, int forkoff);
+int	xfs_attr_shortform_lookup(struct xfs_da_args *args);
+int	xfs_attr_shortform_getvalue(struct xfs_da_args *args);
+int	xfs_attr_shortform_to_leaf(struct xfs_da_args *args);
+int	xfs_attr_shortform_remove(struct xfs_da_args *args);
+int	xfs_attr_shortform_list(struct xfs_attr_list_context *context);
+int	xfs_attr_shortform_allfit(struct xfs_dabuf *bp, struct xfs_inode *dp);
+int	xfs_attr_shortform_bytesfit(xfs_inode_t *dp, int bytes);
+
+
+/*
+ * Internal routines when attribute fork size == XFS_LBSIZE(mp).
+ */
+int	xfs_attr_leaf_to_node(struct xfs_da_args *args);
+int	xfs_attr_leaf_to_shortform(struct xfs_dabuf *bp,
+				   struct xfs_da_args *args, int forkoff);
+int	xfs_attr_leaf_clearflag(struct xfs_da_args *args);
+int	xfs_attr_leaf_setflag(struct xfs_da_args *args);
+int	xfs_attr_leaf_flipflags(xfs_da_args_t *args);
+
+/*
+ * Routines used for growing the Btree.
+ */
+int	xfs_attr_leaf_split(struct xfs_da_state *state,
+				   struct xfs_da_state_blk *oldblk,
+				   struct xfs_da_state_blk *newblk);
+int	xfs_attr_leaf_lookup_int(struct xfs_dabuf *leaf,
+					struct xfs_da_args *args);
+int	xfs_attr_leaf_getvalue(struct xfs_dabuf *bp, struct xfs_da_args *args);
+int	xfs_attr_leaf_add(struct xfs_dabuf *leaf_buffer,
+				 struct xfs_da_args *args);
+int	xfs_attr_leaf_remove(struct xfs_dabuf *leaf_buffer,
+				    struct xfs_da_args *args);
+int	xfs_attr_leaf_list_int(struct xfs_dabuf *bp,
+				      struct xfs_attr_list_context *context);
+
+/*
+ * Routines used for shrinking the Btree.
+ */
+int	xfs_attr_leaf_toosmall(struct xfs_da_state *state, int *retval);
+void	xfs_attr_leaf_unbalance(struct xfs_da_state *state,
+				       struct xfs_da_state_blk *drop_blk,
+				       struct xfs_da_state_blk *save_blk);
+int	xfs_attr_root_inactive(struct xfs_trans **trans, struct xfs_inode *dp);
+
+/*
+ * Utility routines.
+ */
+xfs_dahash_t	xfs_attr_leaf_lasthash(struct xfs_dabuf *bp, int *count);
+int	xfs_attr_leaf_order(struct xfs_dabuf *leaf1_bp,
+				   struct xfs_dabuf *leaf2_bp);
+int	xfs_attr_leaf_newentsize(int namelen, int valuelen, int blocksize,
+					int *local);
+#endif	/* __XFS_ATTR_LEAF_H__ */
diff --git a/fs/xfs/xfs_attr_sf.h b/fs/xfs/xfs_attr_sf.h
new file mode 100644
index 00000000..919756e3
--- /dev/null
+++ b/fs/xfs/xfs_attr_sf.h
@@ -0,0 +1,70 @@
+/*
+ * Copyright (c) 2000,2002,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_ATTR_SF_H__
+#define	__XFS_ATTR_SF_H__
+
+/*
+ * Attribute storage when stored inside the inode.
+ *
+ * Small attribute lists are packed as tightly as possible so as
+ * to fit into the literal area of the inode.
+ */
+
+/*
+ * Entries are packed toward the top as tight as possible.
+ */
+typedef struct xfs_attr_shortform {
+	struct xfs_attr_sf_hdr {	/* constant-structure header block */
+		__be16	totsize;	/* total bytes in shortform list */
+		__u8	count;	/* count of active entries */
+	} hdr;
+	struct xfs_attr_sf_entry {
+		__uint8_t namelen;	/* actual length of name (no NULL) */
+		__uint8_t valuelen;	/* actual length of value (no NULL) */
+		__uint8_t flags;	/* flags bits (see xfs_attr_leaf.h) */
+		__uint8_t nameval[1];	/* name & value bytes concatenated */
+	} list[1];			/* variable sized array */
+} xfs_attr_shortform_t;
+typedef struct xfs_attr_sf_hdr xfs_attr_sf_hdr_t;
+typedef struct xfs_attr_sf_entry xfs_attr_sf_entry_t;
+
+/*
+ * We generate this then sort it, attr_list() must return things in hash-order.
+ */
+typedef struct xfs_attr_sf_sort {
+	__uint8_t	entno;		/* entry number in original list */
+	__uint8_t	namelen;	/* length of name value (no null) */
+	__uint8_t	valuelen;	/* length of value */
+	__uint8_t	flags;		/* flags bits (see xfs_attr_leaf.h) */
+	xfs_dahash_t	hash;		/* this entry's hash value */
+	unsigned char	*name;		/* name value, pointer into buffer */
+} xfs_attr_sf_sort_t;
+
+#define XFS_ATTR_SF_ENTSIZE_BYNAME(nlen,vlen)	/* space name/value uses */ \
+	(((int)sizeof(xfs_attr_sf_entry_t)-1 + (nlen)+(vlen)))
+#define XFS_ATTR_SF_ENTSIZE_MAX			/* max space for name&value */ \
+	((1 << (NBBY*(int)sizeof(__uint8_t))) - 1)
+#define XFS_ATTR_SF_ENTSIZE(sfep)		/* space an entry uses */ \
+	((int)sizeof(xfs_attr_sf_entry_t)-1 + (sfep)->namelen+(sfep)->valuelen)
+#define XFS_ATTR_SF_NEXTENTRY(sfep)		/* next entry in struct */ \
+	((xfs_attr_sf_entry_t *)((char *)(sfep) + XFS_ATTR_SF_ENTSIZE(sfep)))
+#define XFS_ATTR_SF_TOTSIZE(dp)			/* total space in use */ \
+	(be16_to_cpu(((xfs_attr_shortform_t *)	\
+		((dp)->i_afp->if_u1.if_data))->hdr.totsize))
+
+#endif	/* __XFS_ATTR_SF_H__ */
diff --git a/fs/xfs/xfs_bit.c b/fs/xfs/xfs_bit.c
new file mode 100644
index 00000000..48228848
--- /dev/null
+++ b/fs/xfs/xfs_bit.c
@@ -0,0 +1,120 @@
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_trans.h"
+#include "xfs_buf_item.h"
+
+/*
+ * XFS bit manipulation routines, used in non-realtime code.
+ */
+
+/*
+ * Return whether bitmap is empty.
+ * Size is number of words in the bitmap, which is padded to word boundary
+ * Returns 1 for empty, 0 for non-empty.
+ */
+int
+xfs_bitmap_empty(uint *map, uint size)
+{
+	uint i;
+	uint ret = 0;
+
+	for (i = 0; i < size; i++) {
+		ret |= map[i];
+	}
+
+	return (ret == 0);
+}
+
+/*
+ * Count the number of contiguous bits set in the bitmap starting with bit
+ * start_bit.  Size is the size of the bitmap in words.
+ */
+int
+xfs_contig_bits(uint *map, uint	size, uint start_bit)
+{
+	uint * p = ((unsigned int *) map) + (start_bit >> BIT_TO_WORD_SHIFT);
+	uint result = 0;
+	uint tmp;
+
+	size <<= BIT_TO_WORD_SHIFT;
+
+	ASSERT(start_bit < size);
+	size -= start_bit & ~(NBWORD - 1);
+	start_bit &= (NBWORD - 1);
+	if (start_bit) {
+		tmp = *p++;
+		/* set to one first offset bits prior to start */
+		tmp |= (~0U >> (NBWORD-start_bit));
+		if (tmp != ~0U)
+			goto found;
+		result += NBWORD;
+		size -= NBWORD;
+	}
+	while (size) {
+		if ((tmp = *p++) != ~0U)
+			goto found;
+		result += NBWORD;
+		size -= NBWORD;
+	}
+	return result - start_bit;
+found:
+	return result + ffz(tmp) - start_bit;
+}
+
+/*
+ * This takes the bit number to start looking from and
+ * returns the next set bit from there.  It returns -1
+ * if there are no more bits set or the start bit is
+ * beyond the end of the bitmap.
+ *
+ * Size is the number of words, not bytes, in the bitmap.
+ */
+int xfs_next_bit(uint *map, uint size, uint start_bit)
+{
+	uint * p = ((unsigned int *) map) + (start_bit >> BIT_TO_WORD_SHIFT);
+	uint result = start_bit & ~(NBWORD - 1);
+	uint tmp;
+
+	size <<= BIT_TO_WORD_SHIFT;
+
+	if (start_bit >= size)
+		return -1;
+	size -= result;
+	start_bit &= (NBWORD - 1);
+	if (start_bit) {
+		tmp = *p++;
+		/* set to zero first offset bits prior to start */
+		tmp &= (~0U << start_bit);
+		if (tmp != 0U)
+			goto found;
+		result += NBWORD;
+		size -= NBWORD;
+	}
+	while (size) {
+		if ((tmp = *p++) != 0U)
+			goto found;
+		result += NBWORD;
+		size -= NBWORD;
+	}
+	return -1;
+found:
+	return result + ffs(tmp) - 1;
+}
diff --git a/fs/xfs/xfs_bit.h b/fs/xfs/xfs_bit.h
new file mode 100644
index 00000000..f1e3c907
--- /dev/null
+++ b/fs/xfs/xfs_bit.h
@@ -0,0 +1,84 @@
+/*
+ * Copyright (c) 2000,2002,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_BIT_H__
+#define	__XFS_BIT_H__
+
+/*
+ * XFS bit manipulation routines.
+ */
+
+/*
+ * masks with n high/low bits set, 64-bit values
+ */
+static inline __uint64_t xfs_mask64hi(int n)
+{
+	return (__uint64_t)-1 << (64 - (n));
+}
+static inline __uint32_t xfs_mask32lo(int n)
+{
+	return ((__uint32_t)1 << (n)) - 1;
+}
+static inline __uint64_t xfs_mask64lo(int n)
+{
+	return ((__uint64_t)1 << (n)) - 1;
+}
+
+/* Get high bit set out of 32-bit argument, -1 if none set */
+static inline int xfs_highbit32(__uint32_t v)
+{
+	return fls(v) - 1;
+}
+
+/* Get high bit set out of 64-bit argument, -1 if none set */
+static inline int xfs_highbit64(__uint64_t v)
+{
+	return fls64(v) - 1;
+}
+
+/* Get low bit set out of 32-bit argument, -1 if none set */
+static inline int xfs_lowbit32(__uint32_t v)
+{
+	return ffs(v) - 1;
+}
+
+/* Get low bit set out of 64-bit argument, -1 if none set */
+static inline int xfs_lowbit64(__uint64_t v)
+{
+	__uint32_t	w = (__uint32_t)v;
+	int		n = 0;
+
+	if (w) {	/* lower bits */
+		n = ffs(w);
+	} else {	/* upper bits */
+		w = (__uint32_t)(v >> 32);
+		if (w && (n = ffs(w)))
+		n += 32;
+	}
+	return n - 1;
+}
+
+/* Return whether bitmap is empty (1 == empty) */
+extern int xfs_bitmap_empty(uint *map, uint size);
+
+/* Count continuous one bits in map starting with start_bit */
+extern int xfs_contig_bits(uint *map, uint size, uint start_bit);
+
+/* Find next set bit in map */
+extern int xfs_next_bit(uint *map, uint size, uint start_bit);
+
+#endif	/* __XFS_BIT_H__ */
diff --git a/fs/xfs/xfs_bmap.c b/fs/xfs/xfs_bmap.c
new file mode 100644
index 00000000..a175933a
--- /dev/null
+++ b/fs/xfs/xfs_bmap.c
@@ -0,0 +1,6139 @@
+/*
+ * Copyright (c) 2000-2006 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_dir2.h"
+#include "xfs_da_btree.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_alloc_btree.h"
+#include "xfs_ialloc_btree.h"
+#include "xfs_dir2_sf.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_btree.h"
+#include "xfs_mount.h"
+#include "xfs_itable.h"
+#include "xfs_dir2_data.h"
+#include "xfs_dir2_leaf.h"
+#include "xfs_dir2_block.h"
+#include "xfs_inode_item.h"
+#include "xfs_extfree_item.h"
+#include "xfs_alloc.h"
+#include "xfs_bmap.h"
+#include "xfs_rtalloc.h"
+#include "xfs_error.h"
+#include "xfs_attr_leaf.h"
+#include "xfs_rw.h"
+#include "xfs_quota.h"
+#include "xfs_trans_space.h"
+#include "xfs_buf_item.h"
+#include "xfs_filestream.h"
+#include "xfs_vnodeops.h"
+#include "xfs_trace.h"
+
+
+#ifdef DEBUG
+STATIC void
+xfs_bmap_check_leaf_extents(xfs_btree_cur_t *cur, xfs_inode_t *ip, int whichfork);
+#endif
+
+kmem_zone_t		*xfs_bmap_free_item_zone;
+
+/*
+ * Prototypes for internal bmap routines.
+ */
+
+
+/*
+ * Called from xfs_bmap_add_attrfork to handle extents format files.
+ */
+STATIC int					/* error */
+xfs_bmap_add_attrfork_extents(
+	xfs_trans_t		*tp,		/* transaction pointer */
+	xfs_inode_t		*ip,		/* incore inode pointer */
+	xfs_fsblock_t		*firstblock,	/* first block allocated */
+	xfs_bmap_free_t		*flist,		/* blocks to free at commit */
+	int			*flags);	/* inode logging flags */
+
+/*
+ * Called from xfs_bmap_add_attrfork to handle local format files.
+ */
+STATIC int					/* error */
+xfs_bmap_add_attrfork_local(
+	xfs_trans_t		*tp,		/* transaction pointer */
+	xfs_inode_t		*ip,		/* incore inode pointer */
+	xfs_fsblock_t		*firstblock,	/* first block allocated */
+	xfs_bmap_free_t		*flist,		/* blocks to free at commit */
+	int			*flags);	/* inode logging flags */
+
+/*
+ * Called by xfs_bmap_add_extent to handle cases converting a delayed
+ * allocation to a real allocation.
+ */
+STATIC int				/* error */
+xfs_bmap_add_extent_delay_real(
+	xfs_inode_t		*ip,	/* incore inode pointer */
+	xfs_extnum_t		*idx,	/* extent number to update/insert */
+	xfs_btree_cur_t		**curp,	/* if *curp is null, not a btree */
+	xfs_bmbt_irec_t		*new,	/* new data to add to file extents */
+	xfs_filblks_t		*dnew,	/* new delayed-alloc indirect blocks */
+	xfs_fsblock_t		*first,	/* pointer to firstblock variable */
+	xfs_bmap_free_t		*flist,	/* list of extents to be freed */
+	int			*logflagsp); /* inode logging flags */
+
+/*
+ * Called by xfs_bmap_add_extent to handle cases converting a hole
+ * to a delayed allocation.
+ */
+STATIC int				/* error */
+xfs_bmap_add_extent_hole_delay(
+	xfs_inode_t		*ip,	/* incore inode pointer */
+	xfs_extnum_t		*idx,	/* extent number to update/insert */
+	xfs_bmbt_irec_t		*new,	/* new data to add to file extents */
+	int			*logflagsp); /* inode logging flags */
+
+/*
+ * Called by xfs_bmap_add_extent to handle cases converting a hole
+ * to a real allocation.
+ */
+STATIC int				/* error */
+xfs_bmap_add_extent_hole_real(
+	xfs_inode_t		*ip,	/* incore inode pointer */
+	xfs_extnum_t		*idx,	/* extent number to update/insert */
+	xfs_btree_cur_t		*cur,	/* if null, not a btree */
+	xfs_bmbt_irec_t		*new,	/* new data to add to file extents */
+	int			*logflagsp, /* inode logging flags */
+	int			whichfork); /* data or attr fork */
+
+/*
+ * Called by xfs_bmap_add_extent to handle cases converting an unwritten
+ * allocation to a real allocation or vice versa.
+ */
+STATIC int				/* error */
+xfs_bmap_add_extent_unwritten_real(
+	xfs_inode_t		*ip,	/* incore inode pointer */
+	xfs_extnum_t		*idx,	/* extent number to update/insert */
+	xfs_btree_cur_t		**curp,	/* if *curp is null, not a btree */
+	xfs_bmbt_irec_t		*new,	/* new data to add to file extents */
+	int			*logflagsp); /* inode logging flags */
+
+/*
+ * xfs_bmap_alloc is called by xfs_bmapi to allocate an extent for a file.
+ * It figures out where to ask the underlying allocator to put the new extent.
+ */
+STATIC int				/* error */
+xfs_bmap_alloc(
+	xfs_bmalloca_t		*ap);	/* bmap alloc argument struct */
+
+/*
+ * Transform a btree format file with only one leaf node, where the
+ * extents list will fit in the inode, into an extents format file.
+ * Since the file extents are already in-core, all we have to do is
+ * give up the space for the btree root and pitch the leaf block.
+ */
+STATIC int				/* error */
+xfs_bmap_btree_to_extents(
+	xfs_trans_t		*tp,	/* transaction pointer */
+	xfs_inode_t		*ip,	/* incore inode pointer */
+	xfs_btree_cur_t		*cur,	/* btree cursor */
+	int			*logflagsp, /* inode logging flags */
+	int			whichfork); /* data or attr fork */
+
+/*
+ * Remove the entry "free" from the free item list.  Prev points to the
+ * previous entry, unless "free" is the head of the list.
+ */
+STATIC void
+xfs_bmap_del_free(
+	xfs_bmap_free_t		*flist,	/* free item list header */
+	xfs_bmap_free_item_t	*prev,	/* previous item on list, if any */
+	xfs_bmap_free_item_t	*free);	/* list item to be freed */
+
+/*
+ * Convert an extents-format file into a btree-format file.
+ * The new file will have a root block (in the inode) and a single child block.
+ */
+STATIC int					/* error */
+xfs_bmap_extents_to_btree(
+	xfs_trans_t		*tp,		/* transaction pointer */
+	xfs_inode_t		*ip,		/* incore inode pointer */
+	xfs_fsblock_t		*firstblock,	/* first-block-allocated */
+	xfs_bmap_free_t		*flist,		/* blocks freed in xaction */
+	xfs_btree_cur_t		**curp,		/* cursor returned to caller */
+	int			wasdel,		/* converting a delayed alloc */
+	int			*logflagsp,	/* inode logging flags */
+	int			whichfork);	/* data or attr fork */
+
+/*
+ * Convert a local file to an extents file.
+ * This code is sort of bogus, since the file data needs to get
+ * logged so it won't be lost.  The bmap-level manipulations are ok, though.
+ */
+STATIC int				/* error */
+xfs_bmap_local_to_extents(
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_inode_t	*ip,		/* incore inode pointer */
+	xfs_fsblock_t	*firstblock,	/* first block allocated in xaction */
+	xfs_extlen_t	total,		/* total blocks needed by transaction */
+	int		*logflagsp,	/* inode logging flags */
+	int		whichfork);	/* data or attr fork */
+
+/*
+ * Search the extents list for the inode, for the extent containing bno.
+ * If bno lies in a hole, point to the next entry.  If bno lies past eof,
+ * *eofp will be set, and *prevp will contain the last entry (null if none).
+ * Else, *lastxp will be set to the index of the found
+ * entry; *gotp will contain the entry.
+ */
+STATIC xfs_bmbt_rec_host_t *		/* pointer to found extent entry */
+xfs_bmap_search_extents(
+	xfs_inode_t	*ip,		/* incore inode pointer */
+	xfs_fileoff_t	bno,		/* block number searched for */
+	int		whichfork,	/* data or attr fork */
+	int		*eofp,		/* out: end of file found */
+	xfs_extnum_t	*lastxp,	/* out: last extent index */
+	xfs_bmbt_irec_t	*gotp,		/* out: extent entry found */
+	xfs_bmbt_irec_t	*prevp);	/* out: previous extent entry found */
+
+/*
+ * Check the last inode extent to determine whether this allocation will result
+ * in blocks being allocated at the end of the file. When we allocate new data
+ * blocks at the end of the file which do not start at the previous data block,
+ * we will try to align the new blocks at stripe unit boundaries.
+ */
+STATIC int				/* error */
+xfs_bmap_isaeof(
+	xfs_inode_t	*ip,		/* incore inode pointer */
+	xfs_fileoff_t   off,		/* file offset in fsblocks */
+	int             whichfork,	/* data or attribute fork */
+	char		*aeof);		/* return value */
+
+/*
+ * Compute the worst-case number of indirect blocks that will be used
+ * for ip's delayed extent of length "len".
+ */
+STATIC xfs_filblks_t
+xfs_bmap_worst_indlen(
+	xfs_inode_t		*ip,	/* incore inode pointer */
+	xfs_filblks_t		len);	/* delayed extent length */
+
+#ifdef DEBUG
+/*
+ * Perform various validation checks on the values being returned
+ * from xfs_bmapi().
+ */
+STATIC void
+xfs_bmap_validate_ret(
+	xfs_fileoff_t		bno,
+	xfs_filblks_t		len,
+	int			flags,
+	xfs_bmbt_irec_t		*mval,
+	int			nmap,
+	int			ret_nmap);
+#else
+#define	xfs_bmap_validate_ret(bno,len,flags,mval,onmap,nmap)
+#endif /* DEBUG */
+
+STATIC int
+xfs_bmap_count_tree(
+	xfs_mount_t     *mp,
+	xfs_trans_t     *tp,
+	xfs_ifork_t	*ifp,
+	xfs_fsblock_t   blockno,
+	int             levelin,
+	int		*count);
+
+STATIC void
+xfs_bmap_count_leaves(
+	xfs_ifork_t		*ifp,
+	xfs_extnum_t		idx,
+	int			numrecs,
+	int			*count);
+
+STATIC void
+xfs_bmap_disk_count_leaves(
+	struct xfs_mount	*mp,
+	struct xfs_btree_block	*block,
+	int			numrecs,
+	int			*count);
+
+/*
+ * Bmap internal routines.
+ */
+
+STATIC int				/* error */
+xfs_bmbt_lookup_eq(
+	struct xfs_btree_cur	*cur,
+	xfs_fileoff_t		off,
+	xfs_fsblock_t		bno,
+	xfs_filblks_t		len,
+	int			*stat)	/* success/failure */
+{
+	cur->bc_rec.b.br_startoff = off;
+	cur->bc_rec.b.br_startblock = bno;
+	cur->bc_rec.b.br_blockcount = len;
+	return xfs_btree_lookup(cur, XFS_LOOKUP_EQ, stat);
+}
+
+STATIC int				/* error */
+xfs_bmbt_lookup_ge(
+	struct xfs_btree_cur	*cur,
+	xfs_fileoff_t		off,
+	xfs_fsblock_t		bno,
+	xfs_filblks_t		len,
+	int			*stat)	/* success/failure */
+{
+	cur->bc_rec.b.br_startoff = off;
+	cur->bc_rec.b.br_startblock = bno;
+	cur->bc_rec.b.br_blockcount = len;
+	return xfs_btree_lookup(cur, XFS_LOOKUP_GE, stat);
+}
+
+/*
+* Update the record referred to by cur to the value given
+ * by [off, bno, len, state].
+ * This either works (return 0) or gets an EFSCORRUPTED error.
+ */
+STATIC int
+xfs_bmbt_update(
+	struct xfs_btree_cur	*cur,
+	xfs_fileoff_t		off,
+	xfs_fsblock_t		bno,
+	xfs_filblks_t		len,
+	xfs_exntst_t		state)
+{
+	union xfs_btree_rec	rec;
+
+	xfs_bmbt_disk_set_allf(&rec.bmbt, off, bno, len, state);
+	return xfs_btree_update(cur, &rec);
+}
+
+/*
+ * Called from xfs_bmap_add_attrfork to handle btree format files.
+ */
+STATIC int					/* error */
+xfs_bmap_add_attrfork_btree(
+	xfs_trans_t		*tp,		/* transaction pointer */
+	xfs_inode_t		*ip,		/* incore inode pointer */
+	xfs_fsblock_t		*firstblock,	/* first block allocated */
+	xfs_bmap_free_t		*flist,		/* blocks to free at commit */
+	int			*flags)		/* inode logging flags */
+{
+	xfs_btree_cur_t		*cur;		/* btree cursor */
+	int			error;		/* error return value */
+	xfs_mount_t		*mp;		/* file system mount struct */
+	int			stat;		/* newroot status */
+
+	mp = ip->i_mount;
+	if (ip->i_df.if_broot_bytes <= XFS_IFORK_DSIZE(ip))
+		*flags |= XFS_ILOG_DBROOT;
+	else {
+		cur = xfs_bmbt_init_cursor(mp, tp, ip, XFS_DATA_FORK);
+		cur->bc_private.b.flist = flist;
+		cur->bc_private.b.firstblock = *firstblock;
+		if ((error = xfs_bmbt_lookup_ge(cur, 0, 0, 0, &stat)))
+			goto error0;
+		/* must be at least one entry */
+		XFS_WANT_CORRUPTED_GOTO(stat == 1, error0);
+		if ((error = xfs_btree_new_iroot(cur, flags, &stat)))
+			goto error0;
+		if (stat == 0) {
+			xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+			return XFS_ERROR(ENOSPC);
+		}
+		*firstblock = cur->bc_private.b.firstblock;
+		cur->bc_private.b.allocated = 0;
+		xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+	}
+	return 0;
+error0:
+	xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
+	return error;
+}
+
+/*
+ * Called from xfs_bmap_add_attrfork to handle extents format files.
+ */
+STATIC int					/* error */
+xfs_bmap_add_attrfork_extents(
+	xfs_trans_t		*tp,		/* transaction pointer */
+	xfs_inode_t		*ip,		/* incore inode pointer */
+	xfs_fsblock_t		*firstblock,	/* first block allocated */
+	xfs_bmap_free_t		*flist,		/* blocks to free at commit */
+	int			*flags)		/* inode logging flags */
+{
+	xfs_btree_cur_t		*cur;		/* bmap btree cursor */
+	int			error;		/* error return value */
+
+	if (ip->i_d.di_nextents * sizeof(xfs_bmbt_rec_t) <= XFS_IFORK_DSIZE(ip))
+		return 0;
+	cur = NULL;
+	error = xfs_bmap_extents_to_btree(tp, ip, firstblock, flist, &cur, 0,
+		flags, XFS_DATA_FORK);
+	if (cur) {
+		cur->bc_private.b.allocated = 0;
+		xfs_btree_del_cursor(cur,
+			error ? XFS_BTREE_ERROR : XFS_BTREE_NOERROR);
+	}
+	return error;
+}
+
+/*
+ * Called from xfs_bmap_add_attrfork to handle local format files.
+ */
+STATIC int					/* error */
+xfs_bmap_add_attrfork_local(
+	xfs_trans_t		*tp,		/* transaction pointer */
+	xfs_inode_t		*ip,		/* incore inode pointer */
+	xfs_fsblock_t		*firstblock,	/* first block allocated */
+	xfs_bmap_free_t		*flist,		/* blocks to free at commit */
+	int			*flags)		/* inode logging flags */
+{
+	xfs_da_args_t		dargs;		/* args for dir/attr code */
+	int			error;		/* error return value */
+	xfs_mount_t		*mp;		/* mount structure pointer */
+
+	if (ip->i_df.if_bytes <= XFS_IFORK_DSIZE(ip))
+		return 0;
+	if ((ip->i_d.di_mode & S_IFMT) == S_IFDIR) {
+		mp = ip->i_mount;
+		memset(&dargs, 0, sizeof(dargs));
+		dargs.dp = ip;
+		dargs.firstblock = firstblock;
+		dargs.flist = flist;
+		dargs.total = mp->m_dirblkfsbs;
+		dargs.whichfork = XFS_DATA_FORK;
+		dargs.trans = tp;
+		error = xfs_dir2_sf_to_block(&dargs);
+	} else
+		error = xfs_bmap_local_to_extents(tp, ip, firstblock, 1, flags,
+			XFS_DATA_FORK);
+	return error;
+}
+
+/*
+ * Called by xfs_bmapi to update file extent records and the btree
+ * after allocating space (or doing a delayed allocation).
+ */
+STATIC int				/* error */
+xfs_bmap_add_extent(
+	xfs_inode_t		*ip,	/* incore inode pointer */
+	xfs_extnum_t		*idx,	/* extent number to update/insert */
+	xfs_btree_cur_t		**curp,	/* if *curp is null, not a btree */
+	xfs_bmbt_irec_t		*new,	/* new data to add to file extents */
+	xfs_fsblock_t		*first,	/* pointer to firstblock variable */
+	xfs_bmap_free_t		*flist,	/* list of extents to be freed */
+	int			*logflagsp, /* inode logging flags */
+	int			whichfork) /* data or attr fork */
+{
+	xfs_btree_cur_t		*cur;	/* btree cursor or null */
+	xfs_filblks_t		da_new; /* new count del alloc blocks used */
+	xfs_filblks_t		da_old; /* old count del alloc blocks used */
+	int			error;	/* error return value */
+	xfs_ifork_t		*ifp;	/* inode fork ptr */
+	int			logflags; /* returned value */
+	xfs_extnum_t		nextents; /* number of extents in file now */
+
+	XFS_STATS_INC(xs_add_exlist);
+
+	cur = *curp;
+	ifp = XFS_IFORK_PTR(ip, whichfork);
+	nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
+	da_old = da_new = 0;
+	error = 0;
+
+	ASSERT(*idx >= 0);
+	ASSERT(*idx <= nextents);
+
+	/*
+	 * This is the first extent added to a new/empty file.
+	 * Special case this one, so other routines get to assume there are
+	 * already extents in the list.
+	 */
+	if (nextents == 0) {
+		xfs_iext_insert(ip, *idx, 1, new,
+				whichfork == XFS_ATTR_FORK ? BMAP_ATTRFORK : 0);
+
+		ASSERT(cur == NULL);
+
+		if (!isnullstartblock(new->br_startblock)) {
+			XFS_IFORK_NEXT_SET(ip, whichfork, 1);
+			logflags = XFS_ILOG_CORE | xfs_ilog_fext(whichfork);
+		} else
+			logflags = 0;
+	}
+	/*
+	 * Any kind of new delayed allocation goes here.
+	 */
+	else if (isnullstartblock(new->br_startblock)) {
+		if (cur)
+			ASSERT((cur->bc_private.b.flags &
+				XFS_BTCUR_BPRV_WASDEL) == 0);
+		error = xfs_bmap_add_extent_hole_delay(ip, idx, new,
+						       &logflags);
+	}
+	/*
+	 * Real allocation off the end of the file.
+	 */
+	else if (*idx == nextents) {
+		if (cur)
+			ASSERT((cur->bc_private.b.flags &
+				XFS_BTCUR_BPRV_WASDEL) == 0);
+		error = xfs_bmap_add_extent_hole_real(ip, idx, cur, new,
+				&logflags, whichfork);
+	} else {
+		xfs_bmbt_irec_t	prev;	/* old extent at offset idx */
+
+		/*
+		 * Get the record referred to by idx.
+		 */
+		xfs_bmbt_get_all(xfs_iext_get_ext(ifp, *idx), &prev);
+		/*
+		 * If it's a real allocation record, and the new allocation ends
+		 * after the start of the referred to record, then we're filling
+		 * in a delayed or unwritten allocation with a real one, or
+		 * converting real back to unwritten.
+		 */
+		if (!isnullstartblock(new->br_startblock) &&
+		    new->br_startoff + new->br_blockcount > prev.br_startoff) {
+			if (prev.br_state != XFS_EXT_UNWRITTEN &&
+			    isnullstartblock(prev.br_startblock)) {
+				da_old = startblockval(prev.br_startblock);
+				if (cur)
+					ASSERT(cur->bc_private.b.flags &
+						XFS_BTCUR_BPRV_WASDEL);
+				error = xfs_bmap_add_extent_delay_real(ip,
+						idx, &cur, new, &da_new,
+						first, flist, &logflags);
+			} else {
+				ASSERT(new->br_state == XFS_EXT_NORM ||
+				       new->br_state == XFS_EXT_UNWRITTEN);
+
+				error = xfs_bmap_add_extent_unwritten_real(ip,
+						idx, &cur, new, &logflags);
+				if (error)
+					goto done;
+			}
+		}
+		/*
+		 * Otherwise we're filling in a hole with an allocation.
+		 */
+		else {
+			if (cur)
+				ASSERT((cur->bc_private.b.flags &
+					XFS_BTCUR_BPRV_WASDEL) == 0);
+			error = xfs_bmap_add_extent_hole_real(ip, idx, cur,
+					new, &logflags, whichfork);
+		}
+	}
+
+	if (error)
+		goto done;
+	ASSERT(*curp == cur || *curp == NULL);
+
+	/*
+	 * Convert to a btree if necessary.
+	 */
+	if (XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_EXTENTS &&
+	    XFS_IFORK_NEXTENTS(ip, whichfork) > ifp->if_ext_max) {
+		int	tmp_logflags;	/* partial log flag return val */
+
+		ASSERT(cur == NULL);
+		error = xfs_bmap_extents_to_btree(ip->i_transp, ip, first,
+			flist, &cur, da_old > 0, &tmp_logflags, whichfork);
+		logflags |= tmp_logflags;
+		if (error)
+			goto done;
+	}
+	/*
+	 * Adjust for changes in reserved delayed indirect blocks.
+	 * Nothing to do for disk quotas here.
+	 */
+	if (da_old || da_new) {
+		xfs_filblks_t	nblks;
+
+		nblks = da_new;
+		if (cur)
+			nblks += cur->bc_private.b.allocated;
+		ASSERT(nblks <= da_old);
+		if (nblks < da_old)
+			xfs_icsb_modify_counters(ip->i_mount, XFS_SBS_FDBLOCKS,
+				(int64_t)(da_old - nblks), 0);
+	}
+	/*
+	 * Clear out the allocated field, done with it now in any case.
+	 */
+	if (cur) {
+		cur->bc_private.b.allocated = 0;
+		*curp = cur;
+	}
+done:
+#ifdef DEBUG
+	if (!error)
+		xfs_bmap_check_leaf_extents(*curp, ip, whichfork);
+#endif
+	*logflagsp = logflags;
+	return error;
+}
+
+/*
+ * Called by xfs_bmap_add_extent to handle cases converting a delayed
+ * allocation to a real allocation.
+ */
+STATIC int				/* error */
+xfs_bmap_add_extent_delay_real(
+	xfs_inode_t		*ip,	/* incore inode pointer */
+	xfs_extnum_t		*idx,	/* extent number to update/insert */
+	xfs_btree_cur_t		**curp,	/* if *curp is null, not a btree */
+	xfs_bmbt_irec_t		*new,	/* new data to add to file extents */
+	xfs_filblks_t		*dnew,	/* new delayed-alloc indirect blocks */
+	xfs_fsblock_t		*first,	/* pointer to firstblock variable */
+	xfs_bmap_free_t		*flist,	/* list of extents to be freed */
+	int			*logflagsp) /* inode logging flags */
+{
+	xfs_btree_cur_t		*cur;	/* btree cursor */
+	int			diff;	/* temp value */
+	xfs_bmbt_rec_host_t	*ep;	/* extent entry for idx */
+	int			error;	/* error return value */
+	int			i;	/* temp state */
+	xfs_ifork_t		*ifp;	/* inode fork pointer */
+	xfs_fileoff_t		new_endoff;	/* end offset of new entry */
+	xfs_bmbt_irec_t		r[3];	/* neighbor extent entries */
+					/* left is 0, right is 1, prev is 2 */
+	int			rval=0;	/* return value (logging flags) */
+	int			state = 0;/* state bits, accessed thru macros */
+	xfs_filblks_t		temp=0;	/* value for dnew calculations */
+	xfs_filblks_t		temp2=0;/* value for dnew calculations */
+	int			tmp_rval;	/* partial logging flags */
+
+#define	LEFT		r[0]
+#define	RIGHT		r[1]
+#define	PREV		r[2]
+
+	/*
+	 * Set up a bunch of variables to make the tests simpler.
+	 */
+	cur = *curp;
+	ifp = XFS_IFORK_PTR(ip, XFS_DATA_FORK);
+	ep = xfs_iext_get_ext(ifp, *idx);
+	xfs_bmbt_get_all(ep, &PREV);
+	new_endoff = new->br_startoff + new->br_blockcount;
+	ASSERT(PREV.br_startoff <= new->br_startoff);
+	ASSERT(PREV.br_startoff + PREV.br_blockcount >= new_endoff);
+
+	/*
+	 * Set flags determining what part of the previous delayed allocation
+	 * extent is being replaced by a real allocation.
+	 */
+	if (PREV.br_startoff == new->br_startoff)
+		state |= BMAP_LEFT_FILLING;
+	if (PREV.br_startoff + PREV.br_blockcount == new_endoff)
+		state |= BMAP_RIGHT_FILLING;
+
+	/*
+	 * Check and set flags if this segment has a left neighbor.
+	 * Don't set contiguous if the combined extent would be too large.
+	 */
+	if (*idx > 0) {
+		state |= BMAP_LEFT_VALID;
+		xfs_bmbt_get_all(xfs_iext_get_ext(ifp, *idx - 1), &LEFT);
+
+		if (isnullstartblock(LEFT.br_startblock))
+			state |= BMAP_LEFT_DELAY;
+	}
+
+	if ((state & BMAP_LEFT_VALID) && !(state & BMAP_LEFT_DELAY) &&
+	    LEFT.br_startoff + LEFT.br_blockcount == new->br_startoff &&
+	    LEFT.br_startblock + LEFT.br_blockcount == new->br_startblock &&
+	    LEFT.br_state == new->br_state &&
+	    LEFT.br_blockcount + new->br_blockcount <= MAXEXTLEN)
+		state |= BMAP_LEFT_CONTIG;
+
+	/*
+	 * Check and set flags if this segment has a right neighbor.
+	 * Don't set contiguous if the combined extent would be too large.
+	 * Also check for all-three-contiguous being too large.
+	 */
+	if (*idx < ip->i_df.if_bytes / (uint)sizeof(xfs_bmbt_rec_t) - 1) {
+		state |= BMAP_RIGHT_VALID;
+		xfs_bmbt_get_all(xfs_iext_get_ext(ifp, *idx + 1), &RIGHT);
+
+		if (isnullstartblock(RIGHT.br_startblock))
+			state |= BMAP_RIGHT_DELAY;
+	}
+
+	if ((state & BMAP_RIGHT_VALID) && !(state & BMAP_RIGHT_DELAY) &&
+	    new_endoff == RIGHT.br_startoff &&
+	    new->br_startblock + new->br_blockcount == RIGHT.br_startblock &&
+	    new->br_state == RIGHT.br_state &&
+	    new->br_blockcount + RIGHT.br_blockcount <= MAXEXTLEN &&
+	    ((state & (BMAP_LEFT_CONTIG | BMAP_LEFT_FILLING |
+		       BMAP_RIGHT_FILLING)) !=
+		      (BMAP_LEFT_CONTIG | BMAP_LEFT_FILLING |
+		       BMAP_RIGHT_FILLING) ||
+	     LEFT.br_blockcount + new->br_blockcount + RIGHT.br_blockcount
+			<= MAXEXTLEN))
+		state |= BMAP_RIGHT_CONTIG;
+
+	error = 0;
+	/*
+	 * Switch out based on the FILLING and CONTIG state bits.
+	 */
+	switch (state & (BMAP_LEFT_FILLING | BMAP_LEFT_CONTIG |
+			 BMAP_RIGHT_FILLING | BMAP_RIGHT_CONTIG)) {
+	case BMAP_LEFT_FILLING | BMAP_LEFT_CONTIG |
+	     BMAP_RIGHT_FILLING | BMAP_RIGHT_CONTIG:
+		/*
+		 * Filling in all of a previously delayed allocation extent.
+		 * The left and right neighbors are both contiguous with new.
+		 */
+		--*idx;
+		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
+		xfs_bmbt_set_blockcount(xfs_iext_get_ext(ifp, *idx),
+			LEFT.br_blockcount + PREV.br_blockcount +
+			RIGHT.br_blockcount);
+		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+
+		xfs_iext_remove(ip, *idx + 1, 2, state);
+		ip->i_d.di_nextents--;
+		if (cur == NULL)
+			rval = XFS_ILOG_CORE | XFS_ILOG_DEXT;
+		else {
+			rval = XFS_ILOG_CORE;
+			if ((error = xfs_bmbt_lookup_eq(cur, RIGHT.br_startoff,
+					RIGHT.br_startblock,
+					RIGHT.br_blockcount, &i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+			if ((error = xfs_btree_delete(cur, &i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+			if ((error = xfs_btree_decrement(cur, 0, &i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+			if ((error = xfs_bmbt_update(cur, LEFT.br_startoff,
+					LEFT.br_startblock,
+					LEFT.br_blockcount +
+					PREV.br_blockcount +
+					RIGHT.br_blockcount, LEFT.br_state)))
+				goto done;
+		}
+		*dnew = 0;
+		break;
+
+	case BMAP_LEFT_FILLING | BMAP_RIGHT_FILLING | BMAP_LEFT_CONTIG:
+		/*
+		 * Filling in all of a previously delayed allocation extent.
+		 * The left neighbor is contiguous, the right is not.
+		 */
+		--*idx;
+
+		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
+		xfs_bmbt_set_blockcount(xfs_iext_get_ext(ifp, *idx),
+			LEFT.br_blockcount + PREV.br_blockcount);
+		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+
+		xfs_iext_remove(ip, *idx + 1, 1, state);
+		if (cur == NULL)
+			rval = XFS_ILOG_DEXT;
+		else {
+			rval = 0;
+			if ((error = xfs_bmbt_lookup_eq(cur, LEFT.br_startoff,
+					LEFT.br_startblock, LEFT.br_blockcount,
+					&i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+			if ((error = xfs_bmbt_update(cur, LEFT.br_startoff,
+					LEFT.br_startblock,
+					LEFT.br_blockcount +
+					PREV.br_blockcount, LEFT.br_state)))
+				goto done;
+		}
+		*dnew = 0;
+		break;
+
+	case BMAP_LEFT_FILLING | BMAP_RIGHT_FILLING | BMAP_RIGHT_CONTIG:
+		/*
+		 * Filling in all of a previously delayed allocation extent.
+		 * The right neighbor is contiguous, the left is not.
+		 */
+		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
+		xfs_bmbt_set_startblock(ep, new->br_startblock);
+		xfs_bmbt_set_blockcount(ep,
+			PREV.br_blockcount + RIGHT.br_blockcount);
+		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+
+		xfs_iext_remove(ip, *idx + 1, 1, state);
+		if (cur == NULL)
+			rval = XFS_ILOG_DEXT;
+		else {
+			rval = 0;
+			if ((error = xfs_bmbt_lookup_eq(cur, RIGHT.br_startoff,
+					RIGHT.br_startblock,
+					RIGHT.br_blockcount, &i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+			if ((error = xfs_bmbt_update(cur, PREV.br_startoff,
+					new->br_startblock,
+					PREV.br_blockcount +
+					RIGHT.br_blockcount, PREV.br_state)))
+				goto done;
+		}
+
+		*dnew = 0;
+		break;
+
+	case BMAP_LEFT_FILLING | BMAP_RIGHT_FILLING:
+		/*
+		 * Filling in all of a previously delayed allocation extent.
+		 * Neither the left nor right neighbors are contiguous with
+		 * the new one.
+		 */
+		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
+		xfs_bmbt_set_startblock(ep, new->br_startblock);
+		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+
+		ip->i_d.di_nextents++;
+		if (cur == NULL)
+			rval = XFS_ILOG_CORE | XFS_ILOG_DEXT;
+		else {
+			rval = XFS_ILOG_CORE;
+			if ((error = xfs_bmbt_lookup_eq(cur, new->br_startoff,
+					new->br_startblock, new->br_blockcount,
+					&i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 0, done);
+			cur->bc_rec.b.br_state = XFS_EXT_NORM;
+			if ((error = xfs_btree_insert(cur, &i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+		}
+
+		*dnew = 0;
+		break;
+
+	case BMAP_LEFT_FILLING | BMAP_LEFT_CONTIG:
+		/*
+		 * Filling in the first part of a previous delayed allocation.
+		 * The left neighbor is contiguous.
+		 */
+		trace_xfs_bmap_pre_update(ip, *idx - 1, state, _THIS_IP_);
+		xfs_bmbt_set_blockcount(xfs_iext_get_ext(ifp, *idx - 1),
+			LEFT.br_blockcount + new->br_blockcount);
+		xfs_bmbt_set_startoff(ep,
+			PREV.br_startoff + new->br_blockcount);
+		trace_xfs_bmap_post_update(ip, *idx - 1, state, _THIS_IP_);
+
+		temp = PREV.br_blockcount - new->br_blockcount;
+		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
+		xfs_bmbt_set_blockcount(ep, temp);
+		if (cur == NULL)
+			rval = XFS_ILOG_DEXT;
+		else {
+			rval = 0;
+			if ((error = xfs_bmbt_lookup_eq(cur, LEFT.br_startoff,
+					LEFT.br_startblock, LEFT.br_blockcount,
+					&i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+			if ((error = xfs_bmbt_update(cur, LEFT.br_startoff,
+					LEFT.br_startblock,
+					LEFT.br_blockcount +
+					new->br_blockcount,
+					LEFT.br_state)))
+				goto done;
+		}
+		temp = XFS_FILBLKS_MIN(xfs_bmap_worst_indlen(ip, temp),
+			startblockval(PREV.br_startblock));
+		xfs_bmbt_set_startblock(ep, nullstartblock((int)temp));
+		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+
+		--*idx;
+		*dnew = temp;
+		break;
+
+	case BMAP_LEFT_FILLING:
+		/*
+		 * Filling in the first part of a previous delayed allocation.
+		 * The left neighbor is not contiguous.
+		 */
+		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
+		xfs_bmbt_set_startoff(ep, new_endoff);
+		temp = PREV.br_blockcount - new->br_blockcount;
+		xfs_bmbt_set_blockcount(ep, temp);
+		xfs_iext_insert(ip, *idx, 1, new, state);
+		ip->i_d.di_nextents++;
+		if (cur == NULL)
+			rval = XFS_ILOG_CORE | XFS_ILOG_DEXT;
+		else {
+			rval = XFS_ILOG_CORE;
+			if ((error = xfs_bmbt_lookup_eq(cur, new->br_startoff,
+					new->br_startblock, new->br_blockcount,
+					&i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 0, done);
+			cur->bc_rec.b.br_state = XFS_EXT_NORM;
+			if ((error = xfs_btree_insert(cur, &i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+		}
+		if (ip->i_d.di_format == XFS_DINODE_FMT_EXTENTS &&
+		    ip->i_d.di_nextents > ip->i_df.if_ext_max) {
+			error = xfs_bmap_extents_to_btree(ip->i_transp, ip,
+					first, flist, &cur, 1, &tmp_rval,
+					XFS_DATA_FORK);
+			rval |= tmp_rval;
+			if (error)
+				goto done;
+		}
+		temp = XFS_FILBLKS_MIN(xfs_bmap_worst_indlen(ip, temp),
+			startblockval(PREV.br_startblock) -
+			(cur ? cur->bc_private.b.allocated : 0));
+		ep = xfs_iext_get_ext(ifp, *idx + 1);
+		xfs_bmbt_set_startblock(ep, nullstartblock((int)temp));
+		trace_xfs_bmap_post_update(ip, *idx + 1, state, _THIS_IP_);
+
+		*dnew = temp;
+		break;
+
+	case BMAP_RIGHT_FILLING | BMAP_RIGHT_CONTIG:
+		/*
+		 * Filling in the last part of a previous delayed allocation.
+		 * The right neighbor is contiguous with the new allocation.
+		 */
+		temp = PREV.br_blockcount - new->br_blockcount;
+		trace_xfs_bmap_pre_update(ip, *idx + 1, state, _THIS_IP_);
+		xfs_bmbt_set_blockcount(ep, temp);
+		xfs_bmbt_set_allf(xfs_iext_get_ext(ifp, *idx + 1),
+			new->br_startoff, new->br_startblock,
+			new->br_blockcount + RIGHT.br_blockcount,
+			RIGHT.br_state);
+		trace_xfs_bmap_post_update(ip, *idx + 1, state, _THIS_IP_);
+		if (cur == NULL)
+			rval = XFS_ILOG_DEXT;
+		else {
+			rval = 0;
+			if ((error = xfs_bmbt_lookup_eq(cur, RIGHT.br_startoff,
+					RIGHT.br_startblock,
+					RIGHT.br_blockcount, &i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+			if ((error = xfs_bmbt_update(cur, new->br_startoff,
+					new->br_startblock,
+					new->br_blockcount +
+					RIGHT.br_blockcount,
+					RIGHT.br_state)))
+				goto done;
+		}
+
+		temp = XFS_FILBLKS_MIN(xfs_bmap_worst_indlen(ip, temp),
+			startblockval(PREV.br_startblock));
+		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
+		xfs_bmbt_set_startblock(ep, nullstartblock((int)temp));
+		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+
+		++*idx;
+		*dnew = temp;
+		break;
+
+	case BMAP_RIGHT_FILLING:
+		/*
+		 * Filling in the last part of a previous delayed allocation.
+		 * The right neighbor is not contiguous.
+		 */
+		temp = PREV.br_blockcount - new->br_blockcount;
+		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
+		xfs_bmbt_set_blockcount(ep, temp);
+		xfs_iext_insert(ip, *idx + 1, 1, new, state);
+		ip->i_d.di_nextents++;
+		if (cur == NULL)
+			rval = XFS_ILOG_CORE | XFS_ILOG_DEXT;
+		else {
+			rval = XFS_ILOG_CORE;
+			if ((error = xfs_bmbt_lookup_eq(cur, new->br_startoff,
+					new->br_startblock, new->br_blockcount,
+					&i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 0, done);
+			cur->bc_rec.b.br_state = XFS_EXT_NORM;
+			if ((error = xfs_btree_insert(cur, &i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+		}
+		if (ip->i_d.di_format == XFS_DINODE_FMT_EXTENTS &&
+		    ip->i_d.di_nextents > ip->i_df.if_ext_max) {
+			error = xfs_bmap_extents_to_btree(ip->i_transp, ip,
+				first, flist, &cur, 1, &tmp_rval,
+				XFS_DATA_FORK);
+			rval |= tmp_rval;
+			if (error)
+				goto done;
+		}
+		temp = XFS_FILBLKS_MIN(xfs_bmap_worst_indlen(ip, temp),
+			startblockval(PREV.br_startblock) -
+			(cur ? cur->bc_private.b.allocated : 0));
+		ep = xfs_iext_get_ext(ifp, *idx);
+		xfs_bmbt_set_startblock(ep, nullstartblock((int)temp));
+		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+
+		++*idx;
+		*dnew = temp;
+		break;
+
+	case 0:
+		/*
+		 * Filling in the middle part of a previous delayed allocation.
+		 * Contiguity is impossible here.
+		 * This case is avoided almost all the time.
+		 *
+		 * We start with a delayed allocation:
+		 *
+		 * +ddddddddddddddddddddddddddddddddddddddddddddddddddddddd+
+		 *  PREV @ idx
+		 *
+	         * and we are allocating:
+		 *                     +rrrrrrrrrrrrrrrrr+
+		 *			      new
+		 *
+		 * and we set it up for insertion as:
+		 * +ddddddddddddddddddd+rrrrrrrrrrrrrrrrr+ddddddddddddddddd+
+		 *                            new
+		 *  PREV @ idx          LEFT              RIGHT
+		 *                      inserted at idx + 1
+		 */
+		temp = new->br_startoff - PREV.br_startoff;
+		temp2 = PREV.br_startoff + PREV.br_blockcount - new_endoff;
+		trace_xfs_bmap_pre_update(ip, *idx, 0, _THIS_IP_);
+		xfs_bmbt_set_blockcount(ep, temp);	/* truncate PREV */
+		LEFT = *new;
+		RIGHT.br_state = PREV.br_state;
+		RIGHT.br_startblock = nullstartblock(
+				(int)xfs_bmap_worst_indlen(ip, temp2));
+		RIGHT.br_startoff = new_endoff;
+		RIGHT.br_blockcount = temp2;
+		/* insert LEFT (r[0]) and RIGHT (r[1]) at the same time */
+		xfs_iext_insert(ip, *idx + 1, 2, &LEFT, state);
+		ip->i_d.di_nextents++;
+		if (cur == NULL)
+			rval = XFS_ILOG_CORE | XFS_ILOG_DEXT;
+		else {
+			rval = XFS_ILOG_CORE;
+			if ((error = xfs_bmbt_lookup_eq(cur, new->br_startoff,
+					new->br_startblock, new->br_blockcount,
+					&i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 0, done);
+			cur->bc_rec.b.br_state = XFS_EXT_NORM;
+			if ((error = xfs_btree_insert(cur, &i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+		}
+		if (ip->i_d.di_format == XFS_DINODE_FMT_EXTENTS &&
+		    ip->i_d.di_nextents > ip->i_df.if_ext_max) {
+			error = xfs_bmap_extents_to_btree(ip->i_transp, ip,
+					first, flist, &cur, 1, &tmp_rval,
+					XFS_DATA_FORK);
+			rval |= tmp_rval;
+			if (error)
+				goto done;
+		}
+		temp = xfs_bmap_worst_indlen(ip, temp);
+		temp2 = xfs_bmap_worst_indlen(ip, temp2);
+		diff = (int)(temp + temp2 - startblockval(PREV.br_startblock) -
+			(cur ? cur->bc_private.b.allocated : 0));
+		if (diff > 0 &&
+		    xfs_icsb_modify_counters(ip->i_mount, XFS_SBS_FDBLOCKS,
+					     -((int64_t)diff), 0)) {
+			/*
+			 * Ick gross gag me with a spoon.
+			 */
+			ASSERT(0);	/* want to see if this ever happens! */
+			while (diff > 0) {
+				if (temp) {
+					temp--;
+					diff--;
+					if (!diff ||
+					    !xfs_icsb_modify_counters(ip->i_mount,
+						    XFS_SBS_FDBLOCKS,
+						    -((int64_t)diff), 0))
+						break;
+				}
+				if (temp2) {
+					temp2--;
+					diff--;
+					if (!diff ||
+					    !xfs_icsb_modify_counters(ip->i_mount,
+						    XFS_SBS_FDBLOCKS,
+						    -((int64_t)diff), 0))
+						break;
+				}
+			}
+		}
+		ep = xfs_iext_get_ext(ifp, *idx);
+		xfs_bmbt_set_startblock(ep, nullstartblock((int)temp));
+		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+		trace_xfs_bmap_pre_update(ip, *idx + 2, state, _THIS_IP_);
+		xfs_bmbt_set_startblock(xfs_iext_get_ext(ifp, *idx + 2),
+			nullstartblock((int)temp2));
+		trace_xfs_bmap_post_update(ip, *idx + 2, state, _THIS_IP_);
+
+		++*idx;
+		*dnew = temp + temp2;
+		break;
+
+	case BMAP_LEFT_FILLING | BMAP_LEFT_CONTIG | BMAP_RIGHT_CONTIG:
+	case BMAP_RIGHT_FILLING | BMAP_LEFT_CONTIG | BMAP_RIGHT_CONTIG:
+	case BMAP_LEFT_FILLING | BMAP_RIGHT_CONTIG:
+	case BMAP_RIGHT_FILLING | BMAP_LEFT_CONTIG:
+	case BMAP_LEFT_CONTIG | BMAP_RIGHT_CONTIG:
+	case BMAP_LEFT_CONTIG:
+	case BMAP_RIGHT_CONTIG:
+		/*
+		 * These cases are all impossible.
+		 */
+		ASSERT(0);
+	}
+	*curp = cur;
+done:
+	*logflagsp = rval;
+	return error;
+#undef	LEFT
+#undef	RIGHT
+#undef	PREV
+}
+
+/*
+ * Called by xfs_bmap_add_extent to handle cases converting an unwritten
+ * allocation to a real allocation or vice versa.
+ */
+STATIC int				/* error */
+xfs_bmap_add_extent_unwritten_real(
+	xfs_inode_t		*ip,	/* incore inode pointer */
+	xfs_extnum_t		*idx,	/* extent number to update/insert */
+	xfs_btree_cur_t		**curp,	/* if *curp is null, not a btree */
+	xfs_bmbt_irec_t		*new,	/* new data to add to file extents */
+	int			*logflagsp) /* inode logging flags */
+{
+	xfs_btree_cur_t		*cur;	/* btree cursor */
+	xfs_bmbt_rec_host_t	*ep;	/* extent entry for idx */
+	int			error;	/* error return value */
+	int			i;	/* temp state */
+	xfs_ifork_t		*ifp;	/* inode fork pointer */
+	xfs_fileoff_t		new_endoff;	/* end offset of new entry */
+	xfs_exntst_t		newext;	/* new extent state */
+	xfs_exntst_t		oldext;	/* old extent state */
+	xfs_bmbt_irec_t		r[3];	/* neighbor extent entries */
+					/* left is 0, right is 1, prev is 2 */
+	int			rval=0;	/* return value (logging flags) */
+	int			state = 0;/* state bits, accessed thru macros */
+
+#define	LEFT		r[0]
+#define	RIGHT		r[1]
+#define	PREV		r[2]
+	/*
+	 * Set up a bunch of variables to make the tests simpler.
+	 */
+	error = 0;
+	cur = *curp;
+	ifp = XFS_IFORK_PTR(ip, XFS_DATA_FORK);
+	ep = xfs_iext_get_ext(ifp, *idx);
+	xfs_bmbt_get_all(ep, &PREV);
+	newext = new->br_state;
+	oldext = (newext == XFS_EXT_UNWRITTEN) ?
+		XFS_EXT_NORM : XFS_EXT_UNWRITTEN;
+	ASSERT(PREV.br_state == oldext);
+	new_endoff = new->br_startoff + new->br_blockcount;
+	ASSERT(PREV.br_startoff <= new->br_startoff);
+	ASSERT(PREV.br_startoff + PREV.br_blockcount >= new_endoff);
+
+	/*
+	 * Set flags determining what part of the previous oldext allocation
+	 * extent is being replaced by a newext allocation.
+	 */
+	if (PREV.br_startoff == new->br_startoff)
+		state |= BMAP_LEFT_FILLING;
+	if (PREV.br_startoff + PREV.br_blockcount == new_endoff)
+		state |= BMAP_RIGHT_FILLING;
+
+	/*
+	 * Check and set flags if this segment has a left neighbor.
+	 * Don't set contiguous if the combined extent would be too large.
+	 */
+	if (*idx > 0) {
+		state |= BMAP_LEFT_VALID;
+		xfs_bmbt_get_all(xfs_iext_get_ext(ifp, *idx - 1), &LEFT);
+
+		if (isnullstartblock(LEFT.br_startblock))
+			state |= BMAP_LEFT_DELAY;
+	}
+
+	if ((state & BMAP_LEFT_VALID) && !(state & BMAP_LEFT_DELAY) &&
+	    LEFT.br_startoff + LEFT.br_blockcount == new->br_startoff &&
+	    LEFT.br_startblock + LEFT.br_blockcount == new->br_startblock &&
+	    LEFT.br_state == newext &&
+	    LEFT.br_blockcount + new->br_blockcount <= MAXEXTLEN)
+		state |= BMAP_LEFT_CONTIG;
+
+	/*
+	 * Check and set flags if this segment has a right neighbor.
+	 * Don't set contiguous if the combined extent would be too large.
+	 * Also check for all-three-contiguous being too large.
+	 */
+	if (*idx < ip->i_df.if_bytes / (uint)sizeof(xfs_bmbt_rec_t) - 1) {
+		state |= BMAP_RIGHT_VALID;
+		xfs_bmbt_get_all(xfs_iext_get_ext(ifp, *idx + 1), &RIGHT);
+		if (isnullstartblock(RIGHT.br_startblock))
+			state |= BMAP_RIGHT_DELAY;
+	}
+
+	if ((state & BMAP_RIGHT_VALID) && !(state & BMAP_RIGHT_DELAY) &&
+	    new_endoff == RIGHT.br_startoff &&
+	    new->br_startblock + new->br_blockcount == RIGHT.br_startblock &&
+	    newext == RIGHT.br_state &&
+	    new->br_blockcount + RIGHT.br_blockcount <= MAXEXTLEN &&
+	    ((state & (BMAP_LEFT_CONTIG | BMAP_LEFT_FILLING |
+		       BMAP_RIGHT_FILLING)) !=
+		      (BMAP_LEFT_CONTIG | BMAP_LEFT_FILLING |
+		       BMAP_RIGHT_FILLING) ||
+	     LEFT.br_blockcount + new->br_blockcount + RIGHT.br_blockcount
+			<= MAXEXTLEN))
+		state |= BMAP_RIGHT_CONTIG;
+
+	/*
+	 * Switch out based on the FILLING and CONTIG state bits.
+	 */
+	switch (state & (BMAP_LEFT_FILLING | BMAP_LEFT_CONTIG |
+			 BMAP_RIGHT_FILLING | BMAP_RIGHT_CONTIG)) {
+	case BMAP_LEFT_FILLING | BMAP_LEFT_CONTIG |
+	     BMAP_RIGHT_FILLING | BMAP_RIGHT_CONTIG:
+		/*
+		 * Setting all of a previous oldext extent to newext.
+		 * The left and right neighbors are both contiguous with new.
+		 */
+		--*idx;
+
+		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
+		xfs_bmbt_set_blockcount(xfs_iext_get_ext(ifp, *idx),
+			LEFT.br_blockcount + PREV.br_blockcount +
+			RIGHT.br_blockcount);
+		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+
+		xfs_iext_remove(ip, *idx + 1, 2, state);
+		ip->i_d.di_nextents -= 2;
+		if (cur == NULL)
+			rval = XFS_ILOG_CORE | XFS_ILOG_DEXT;
+		else {
+			rval = XFS_ILOG_CORE;
+			if ((error = xfs_bmbt_lookup_eq(cur, RIGHT.br_startoff,
+					RIGHT.br_startblock,
+					RIGHT.br_blockcount, &i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+			if ((error = xfs_btree_delete(cur, &i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+			if ((error = xfs_btree_decrement(cur, 0, &i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+			if ((error = xfs_btree_delete(cur, &i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+			if ((error = xfs_btree_decrement(cur, 0, &i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+			if ((error = xfs_bmbt_update(cur, LEFT.br_startoff,
+				LEFT.br_startblock,
+				LEFT.br_blockcount + PREV.br_blockcount +
+				RIGHT.br_blockcount, LEFT.br_state)))
+				goto done;
+		}
+		break;
+
+	case BMAP_LEFT_FILLING | BMAP_RIGHT_FILLING | BMAP_LEFT_CONTIG:
+		/*
+		 * Setting all of a previous oldext extent to newext.
+		 * The left neighbor is contiguous, the right is not.
+		 */
+		--*idx;
+
+		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
+		xfs_bmbt_set_blockcount(xfs_iext_get_ext(ifp, *idx),
+			LEFT.br_blockcount + PREV.br_blockcount);
+		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+
+		xfs_iext_remove(ip, *idx + 1, 1, state);
+		ip->i_d.di_nextents--;
+		if (cur == NULL)
+			rval = XFS_ILOG_CORE | XFS_ILOG_DEXT;
+		else {
+			rval = XFS_ILOG_CORE;
+			if ((error = xfs_bmbt_lookup_eq(cur, PREV.br_startoff,
+					PREV.br_startblock, PREV.br_blockcount,
+					&i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+			if ((error = xfs_btree_delete(cur, &i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+			if ((error = xfs_btree_decrement(cur, 0, &i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+			if ((error = xfs_bmbt_update(cur, LEFT.br_startoff,
+				LEFT.br_startblock,
+				LEFT.br_blockcount + PREV.br_blockcount,
+				LEFT.br_state)))
+				goto done;
+		}
+		break;
+
+	case BMAP_LEFT_FILLING | BMAP_RIGHT_FILLING | BMAP_RIGHT_CONTIG:
+		/*
+		 * Setting all of a previous oldext extent to newext.
+		 * The right neighbor is contiguous, the left is not.
+		 */
+		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
+		xfs_bmbt_set_blockcount(ep,
+			PREV.br_blockcount + RIGHT.br_blockcount);
+		xfs_bmbt_set_state(ep, newext);
+		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+		xfs_iext_remove(ip, *idx + 1, 1, state);
+		ip->i_d.di_nextents--;
+		if (cur == NULL)
+			rval = XFS_ILOG_CORE | XFS_ILOG_DEXT;
+		else {
+			rval = XFS_ILOG_CORE;
+			if ((error = xfs_bmbt_lookup_eq(cur, RIGHT.br_startoff,
+					RIGHT.br_startblock,
+					RIGHT.br_blockcount, &i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+			if ((error = xfs_btree_delete(cur, &i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+			if ((error = xfs_btree_decrement(cur, 0, &i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+			if ((error = xfs_bmbt_update(cur, new->br_startoff,
+				new->br_startblock,
+				new->br_blockcount + RIGHT.br_blockcount,
+				newext)))
+				goto done;
+		}
+		break;
+
+	case BMAP_LEFT_FILLING | BMAP_RIGHT_FILLING:
+		/*
+		 * Setting all of a previous oldext extent to newext.
+		 * Neither the left nor right neighbors are contiguous with
+		 * the new one.
+		 */
+		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
+		xfs_bmbt_set_state(ep, newext);
+		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+
+		if (cur == NULL)
+			rval = XFS_ILOG_DEXT;
+		else {
+			rval = 0;
+			if ((error = xfs_bmbt_lookup_eq(cur, new->br_startoff,
+					new->br_startblock, new->br_blockcount,
+					&i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+			if ((error = xfs_bmbt_update(cur, new->br_startoff,
+				new->br_startblock, new->br_blockcount,
+				newext)))
+				goto done;
+		}
+		break;
+
+	case BMAP_LEFT_FILLING | BMAP_LEFT_CONTIG:
+		/*
+		 * Setting the first part of a previous oldext extent to newext.
+		 * The left neighbor is contiguous.
+		 */
+		trace_xfs_bmap_pre_update(ip, *idx - 1, state, _THIS_IP_);
+		xfs_bmbt_set_blockcount(xfs_iext_get_ext(ifp, *idx - 1),
+			LEFT.br_blockcount + new->br_blockcount);
+		xfs_bmbt_set_startoff(ep,
+			PREV.br_startoff + new->br_blockcount);
+		trace_xfs_bmap_post_update(ip, *idx - 1, state, _THIS_IP_);
+
+		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
+		xfs_bmbt_set_startblock(ep,
+			new->br_startblock + new->br_blockcount);
+		xfs_bmbt_set_blockcount(ep,
+			PREV.br_blockcount - new->br_blockcount);
+		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+
+		--*idx;
+
+		if (cur == NULL)
+			rval = XFS_ILOG_DEXT;
+		else {
+			rval = 0;
+			if ((error = xfs_bmbt_lookup_eq(cur, PREV.br_startoff,
+					PREV.br_startblock, PREV.br_blockcount,
+					&i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+			if ((error = xfs_bmbt_update(cur,
+				PREV.br_startoff + new->br_blockcount,
+				PREV.br_startblock + new->br_blockcount,
+				PREV.br_blockcount - new->br_blockcount,
+				oldext)))
+				goto done;
+			if ((error = xfs_btree_decrement(cur, 0, &i)))
+				goto done;
+			if (xfs_bmbt_update(cur, LEFT.br_startoff,
+				LEFT.br_startblock,
+				LEFT.br_blockcount + new->br_blockcount,
+				LEFT.br_state))
+				goto done;
+		}
+		break;
+
+	case BMAP_LEFT_FILLING:
+		/*
+		 * Setting the first part of a previous oldext extent to newext.
+		 * The left neighbor is not contiguous.
+		 */
+		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
+		ASSERT(ep && xfs_bmbt_get_state(ep) == oldext);
+		xfs_bmbt_set_startoff(ep, new_endoff);
+		xfs_bmbt_set_blockcount(ep,
+			PREV.br_blockcount - new->br_blockcount);
+		xfs_bmbt_set_startblock(ep,
+			new->br_startblock + new->br_blockcount);
+		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+
+		xfs_iext_insert(ip, *idx, 1, new, state);
+		ip->i_d.di_nextents++;
+		if (cur == NULL)
+			rval = XFS_ILOG_CORE | XFS_ILOG_DEXT;
+		else {
+			rval = XFS_ILOG_CORE;
+			if ((error = xfs_bmbt_lookup_eq(cur, PREV.br_startoff,
+					PREV.br_startblock, PREV.br_blockcount,
+					&i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+			if ((error = xfs_bmbt_update(cur,
+				PREV.br_startoff + new->br_blockcount,
+				PREV.br_startblock + new->br_blockcount,
+				PREV.br_blockcount - new->br_blockcount,
+				oldext)))
+				goto done;
+			cur->bc_rec.b = *new;
+			if ((error = xfs_btree_insert(cur, &i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+		}
+		break;
+
+	case BMAP_RIGHT_FILLING | BMAP_RIGHT_CONTIG:
+		/*
+		 * Setting the last part of a previous oldext extent to newext.
+		 * The right neighbor is contiguous with the new allocation.
+		 */
+		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
+		xfs_bmbt_set_blockcount(ep,
+			PREV.br_blockcount - new->br_blockcount);
+		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+
+		++*idx;
+
+		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
+		xfs_bmbt_set_allf(xfs_iext_get_ext(ifp, *idx),
+			new->br_startoff, new->br_startblock,
+			new->br_blockcount + RIGHT.br_blockcount, newext);
+		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+
+		if (cur == NULL)
+			rval = XFS_ILOG_DEXT;
+		else {
+			rval = 0;
+			if ((error = xfs_bmbt_lookup_eq(cur, PREV.br_startoff,
+					PREV.br_startblock,
+					PREV.br_blockcount, &i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+			if ((error = xfs_bmbt_update(cur, PREV.br_startoff,
+				PREV.br_startblock,
+				PREV.br_blockcount - new->br_blockcount,
+				oldext)))
+				goto done;
+			if ((error = xfs_btree_increment(cur, 0, &i)))
+				goto done;
+			if ((error = xfs_bmbt_update(cur, new->br_startoff,
+				new->br_startblock,
+				new->br_blockcount + RIGHT.br_blockcount,
+				newext)))
+				goto done;
+		}
+		break;
+
+	case BMAP_RIGHT_FILLING:
+		/*
+		 * Setting the last part of a previous oldext extent to newext.
+		 * The right neighbor is not contiguous.
+		 */
+		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
+		xfs_bmbt_set_blockcount(ep,
+			PREV.br_blockcount - new->br_blockcount);
+		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+
+		++*idx;
+		xfs_iext_insert(ip, *idx, 1, new, state);
+
+		ip->i_d.di_nextents++;
+		if (cur == NULL)
+			rval = XFS_ILOG_CORE | XFS_ILOG_DEXT;
+		else {
+			rval = XFS_ILOG_CORE;
+			if ((error = xfs_bmbt_lookup_eq(cur, PREV.br_startoff,
+					PREV.br_startblock, PREV.br_blockcount,
+					&i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+			if ((error = xfs_bmbt_update(cur, PREV.br_startoff,
+				PREV.br_startblock,
+				PREV.br_blockcount - new->br_blockcount,
+				oldext)))
+				goto done;
+			if ((error = xfs_bmbt_lookup_eq(cur, new->br_startoff,
+					new->br_startblock, new->br_blockcount,
+					&i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 0, done);
+			cur->bc_rec.b.br_state = XFS_EXT_NORM;
+			if ((error = xfs_btree_insert(cur, &i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+		}
+		break;
+
+	case 0:
+		/*
+		 * Setting the middle part of a previous oldext extent to
+		 * newext.  Contiguity is impossible here.
+		 * One extent becomes three extents.
+		 */
+		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
+		xfs_bmbt_set_blockcount(ep,
+			new->br_startoff - PREV.br_startoff);
+		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+
+		r[0] = *new;
+		r[1].br_startoff = new_endoff;
+		r[1].br_blockcount =
+			PREV.br_startoff + PREV.br_blockcount - new_endoff;
+		r[1].br_startblock = new->br_startblock + new->br_blockcount;
+		r[1].br_state = oldext;
+
+		++*idx;
+		xfs_iext_insert(ip, *idx, 2, &r[0], state);
+
+		ip->i_d.di_nextents += 2;
+		if (cur == NULL)
+			rval = XFS_ILOG_CORE | XFS_ILOG_DEXT;
+		else {
+			rval = XFS_ILOG_CORE;
+			if ((error = xfs_bmbt_lookup_eq(cur, PREV.br_startoff,
+					PREV.br_startblock, PREV.br_blockcount,
+					&i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+			/* new right extent - oldext */
+			if ((error = xfs_bmbt_update(cur, r[1].br_startoff,
+				r[1].br_startblock, r[1].br_blockcount,
+				r[1].br_state)))
+				goto done;
+			/* new left extent - oldext */
+			cur->bc_rec.b = PREV;
+			cur->bc_rec.b.br_blockcount =
+				new->br_startoff - PREV.br_startoff;
+			if ((error = xfs_btree_insert(cur, &i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+			/*
+			 * Reset the cursor to the position of the new extent
+			 * we are about to insert as we can't trust it after
+			 * the previous insert.
+			 */
+			if ((error = xfs_bmbt_lookup_eq(cur, new->br_startoff,
+					new->br_startblock, new->br_blockcount,
+					&i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 0, done);
+			/* new middle extent - newext */
+			cur->bc_rec.b.br_state = new->br_state;
+			if ((error = xfs_btree_insert(cur, &i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+		}
+		break;
+
+	case BMAP_LEFT_FILLING | BMAP_LEFT_CONTIG | BMAP_RIGHT_CONTIG:
+	case BMAP_RIGHT_FILLING | BMAP_LEFT_CONTIG | BMAP_RIGHT_CONTIG:
+	case BMAP_LEFT_FILLING | BMAP_RIGHT_CONTIG:
+	case BMAP_RIGHT_FILLING | BMAP_LEFT_CONTIG:
+	case BMAP_LEFT_CONTIG | BMAP_RIGHT_CONTIG:
+	case BMAP_LEFT_CONTIG:
+	case BMAP_RIGHT_CONTIG:
+		/*
+		 * These cases are all impossible.
+		 */
+		ASSERT(0);
+	}
+	*curp = cur;
+done:
+	*logflagsp = rval;
+	return error;
+#undef	LEFT
+#undef	RIGHT
+#undef	PREV
+}
+
+/*
+ * Called by xfs_bmap_add_extent to handle cases converting a hole
+ * to a delayed allocation.
+ */
+/*ARGSUSED*/
+STATIC int				/* error */
+xfs_bmap_add_extent_hole_delay(
+	xfs_inode_t		*ip,	/* incore inode pointer */
+	xfs_extnum_t		*idx,	/* extent number to update/insert */
+	xfs_bmbt_irec_t		*new,	/* new data to add to file extents */
+	int			*logflagsp) /* inode logging flags */
+{
+	xfs_ifork_t		*ifp;	/* inode fork pointer */
+	xfs_bmbt_irec_t		left;	/* left neighbor extent entry */
+	xfs_filblks_t		newlen=0;	/* new indirect size */
+	xfs_filblks_t		oldlen=0;	/* old indirect size */
+	xfs_bmbt_irec_t		right;	/* right neighbor extent entry */
+	int			state;  /* state bits, accessed thru macros */
+	xfs_filblks_t		temp=0;	/* temp for indirect calculations */
+
+	ifp = XFS_IFORK_PTR(ip, XFS_DATA_FORK);
+	state = 0;
+	ASSERT(isnullstartblock(new->br_startblock));
+
+	/*
+	 * Check and set flags if this segment has a left neighbor
+	 */
+	if (*idx > 0) {
+		state |= BMAP_LEFT_VALID;
+		xfs_bmbt_get_all(xfs_iext_get_ext(ifp, *idx - 1), &left);
+
+		if (isnullstartblock(left.br_startblock))
+			state |= BMAP_LEFT_DELAY;
+	}
+
+	/*
+	 * Check and set flags if the current (right) segment exists.
+	 * If it doesn't exist, we're converting the hole at end-of-file.
+	 */
+	if (*idx < ip->i_df.if_bytes / (uint)sizeof(xfs_bmbt_rec_t)) {
+		state |= BMAP_RIGHT_VALID;
+		xfs_bmbt_get_all(xfs_iext_get_ext(ifp, *idx), &right);
+
+		if (isnullstartblock(right.br_startblock))
+			state |= BMAP_RIGHT_DELAY;
+	}
+
+	/*
+	 * Set contiguity flags on the left and right neighbors.
+	 * Don't let extents get too large, even if the pieces are contiguous.
+	 */
+	if ((state & BMAP_LEFT_VALID) && (state & BMAP_LEFT_DELAY) &&
+	    left.br_startoff + left.br_blockcount == new->br_startoff &&
+	    left.br_blockcount + new->br_blockcount <= MAXEXTLEN)
+		state |= BMAP_LEFT_CONTIG;
+
+	if ((state & BMAP_RIGHT_VALID) && (state & BMAP_RIGHT_DELAY) &&
+	    new->br_startoff + new->br_blockcount == right.br_startoff &&
+	    new->br_blockcount + right.br_blockcount <= MAXEXTLEN &&
+	    (!(state & BMAP_LEFT_CONTIG) ||
+	     (left.br_blockcount + new->br_blockcount +
+	      right.br_blockcount <= MAXEXTLEN)))
+		state |= BMAP_RIGHT_CONTIG;
+
+	/*
+	 * Switch out based on the contiguity flags.
+	 */
+	switch (state & (BMAP_LEFT_CONTIG | BMAP_RIGHT_CONTIG)) {
+	case BMAP_LEFT_CONTIG | BMAP_RIGHT_CONTIG:
+		/*
+		 * New allocation is contiguous with delayed allocations
+		 * on the left and on the right.
+		 * Merge all three into a single extent record.
+		 */
+		--*idx;
+		temp = left.br_blockcount + new->br_blockcount +
+			right.br_blockcount;
+
+		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
+		xfs_bmbt_set_blockcount(xfs_iext_get_ext(ifp, *idx), temp);
+		oldlen = startblockval(left.br_startblock) +
+			startblockval(new->br_startblock) +
+			startblockval(right.br_startblock);
+		newlen = xfs_bmap_worst_indlen(ip, temp);
+		xfs_bmbt_set_startblock(xfs_iext_get_ext(ifp, *idx),
+			nullstartblock((int)newlen));
+		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+
+		xfs_iext_remove(ip, *idx + 1, 1, state);
+		break;
+
+	case BMAP_LEFT_CONTIG:
+		/*
+		 * New allocation is contiguous with a delayed allocation
+		 * on the left.
+		 * Merge the new allocation with the left neighbor.
+		 */
+		--*idx;
+		temp = left.br_blockcount + new->br_blockcount;
+
+		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
+		xfs_bmbt_set_blockcount(xfs_iext_get_ext(ifp, *idx), temp);
+		oldlen = startblockval(left.br_startblock) +
+			startblockval(new->br_startblock);
+		newlen = xfs_bmap_worst_indlen(ip, temp);
+		xfs_bmbt_set_startblock(xfs_iext_get_ext(ifp, *idx),
+			nullstartblock((int)newlen));
+		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+		break;
+
+	case BMAP_RIGHT_CONTIG:
+		/*
+		 * New allocation is contiguous with a delayed allocation
+		 * on the right.
+		 * Merge the new allocation with the right neighbor.
+		 */
+		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
+		temp = new->br_blockcount + right.br_blockcount;
+		oldlen = startblockval(new->br_startblock) +
+			startblockval(right.br_startblock);
+		newlen = xfs_bmap_worst_indlen(ip, temp);
+		xfs_bmbt_set_allf(xfs_iext_get_ext(ifp, *idx),
+			new->br_startoff,
+			nullstartblock((int)newlen), temp, right.br_state);
+		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+		break;
+
+	case 0:
+		/*
+		 * New allocation is not contiguous with another
+		 * delayed allocation.
+		 * Insert a new entry.
+		 */
+		oldlen = newlen = 0;
+		xfs_iext_insert(ip, *idx, 1, new, state);
+		break;
+	}
+	if (oldlen != newlen) {
+		ASSERT(oldlen > newlen);
+		xfs_icsb_modify_counters(ip->i_mount, XFS_SBS_FDBLOCKS,
+			(int64_t)(oldlen - newlen), 0);
+		/*
+		 * Nothing to do for disk quota accounting here.
+		 */
+	}
+	*logflagsp = 0;
+	return 0;
+}
+
+/*
+ * Called by xfs_bmap_add_extent to handle cases converting a hole
+ * to a real allocation.
+ */
+STATIC int				/* error */
+xfs_bmap_add_extent_hole_real(
+	xfs_inode_t		*ip,	/* incore inode pointer */
+	xfs_extnum_t		*idx,	/* extent number to update/insert */
+	xfs_btree_cur_t		*cur,	/* if null, not a btree */
+	xfs_bmbt_irec_t		*new,	/* new data to add to file extents */
+	int			*logflagsp, /* inode logging flags */
+	int			whichfork) /* data or attr fork */
+{
+	int			error;	/* error return value */
+	int			i;	/* temp state */
+	xfs_ifork_t		*ifp;	/* inode fork pointer */
+	xfs_bmbt_irec_t		left;	/* left neighbor extent entry */
+	xfs_bmbt_irec_t		right;	/* right neighbor extent entry */
+	int			rval=0;	/* return value (logging flags) */
+	int			state;	/* state bits, accessed thru macros */
+
+	ifp = XFS_IFORK_PTR(ip, whichfork);
+	ASSERT(*idx <= ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t));
+	state = 0;
+
+	if (whichfork == XFS_ATTR_FORK)
+		state |= BMAP_ATTRFORK;
+
+	/*
+	 * Check and set flags if this segment has a left neighbor.
+	 */
+	if (*idx > 0) {
+		state |= BMAP_LEFT_VALID;
+		xfs_bmbt_get_all(xfs_iext_get_ext(ifp, *idx - 1), &left);
+		if (isnullstartblock(left.br_startblock))
+			state |= BMAP_LEFT_DELAY;
+	}
+
+	/*
+	 * Check and set flags if this segment has a current value.
+	 * Not true if we're inserting into the "hole" at eof.
+	 */
+	if (*idx < ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t)) {
+		state |= BMAP_RIGHT_VALID;
+		xfs_bmbt_get_all(xfs_iext_get_ext(ifp, *idx), &right);
+		if (isnullstartblock(right.br_startblock))
+			state |= BMAP_RIGHT_DELAY;
+	}
+
+	/*
+	 * We're inserting a real allocation between "left" and "right".
+	 * Set the contiguity flags.  Don't let extents get too large.
+	 */
+	if ((state & BMAP_LEFT_VALID) && !(state & BMAP_LEFT_DELAY) &&
+	    left.br_startoff + left.br_blockcount == new->br_startoff &&
+	    left.br_startblock + left.br_blockcount == new->br_startblock &&
+	    left.br_state == new->br_state &&
+	    left.br_blockcount + new->br_blockcount <= MAXEXTLEN)
+		state |= BMAP_LEFT_CONTIG;
+
+	if ((state & BMAP_RIGHT_VALID) && !(state & BMAP_RIGHT_DELAY) &&
+	    new->br_startoff + new->br_blockcount == right.br_startoff &&
+	    new->br_startblock + new->br_blockcount == right.br_startblock &&
+	    new->br_state == right.br_state &&
+	    new->br_blockcount + right.br_blockcount <= MAXEXTLEN &&
+	    (!(state & BMAP_LEFT_CONTIG) ||
+	     left.br_blockcount + new->br_blockcount +
+	     right.br_blockcount <= MAXEXTLEN))
+		state |= BMAP_RIGHT_CONTIG;
+
+	error = 0;
+	/*
+	 * Select which case we're in here, and implement it.
+	 */
+	switch (state & (BMAP_LEFT_CONTIG | BMAP_RIGHT_CONTIG)) {
+	case BMAP_LEFT_CONTIG | BMAP_RIGHT_CONTIG:
+		/*
+		 * New allocation is contiguous with real allocations on the
+		 * left and on the right.
+		 * Merge all three into a single extent record.
+		 */
+		--*idx;
+		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
+		xfs_bmbt_set_blockcount(xfs_iext_get_ext(ifp, *idx),
+			left.br_blockcount + new->br_blockcount +
+			right.br_blockcount);
+		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+
+		xfs_iext_remove(ip, *idx + 1, 1, state);
+
+		XFS_IFORK_NEXT_SET(ip, whichfork,
+			XFS_IFORK_NEXTENTS(ip, whichfork) - 1);
+		if (cur == NULL) {
+			rval = XFS_ILOG_CORE | xfs_ilog_fext(whichfork);
+		} else {
+			rval = XFS_ILOG_CORE;
+			if ((error = xfs_bmbt_lookup_eq(cur,
+					right.br_startoff,
+					right.br_startblock,
+					right.br_blockcount, &i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+			if ((error = xfs_btree_delete(cur, &i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+			if ((error = xfs_btree_decrement(cur, 0, &i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+			if ((error = xfs_bmbt_update(cur, left.br_startoff,
+					left.br_startblock,
+					left.br_blockcount +
+						new->br_blockcount +
+						right.br_blockcount,
+					left.br_state)))
+				goto done;
+		}
+		break;
+
+	case BMAP_LEFT_CONTIG:
+		/*
+		 * New allocation is contiguous with a real allocation
+		 * on the left.
+		 * Merge the new allocation with the left neighbor.
+		 */
+		--*idx;
+		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
+		xfs_bmbt_set_blockcount(xfs_iext_get_ext(ifp, *idx),
+			left.br_blockcount + new->br_blockcount);
+		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+
+		if (cur == NULL) {
+			rval = xfs_ilog_fext(whichfork);
+		} else {
+			rval = 0;
+			if ((error = xfs_bmbt_lookup_eq(cur,
+					left.br_startoff,
+					left.br_startblock,
+					left.br_blockcount, &i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+			if ((error = xfs_bmbt_update(cur, left.br_startoff,
+					left.br_startblock,
+					left.br_blockcount +
+						new->br_blockcount,
+					left.br_state)))
+				goto done;
+		}
+		break;
+
+	case BMAP_RIGHT_CONTIG:
+		/*
+		 * New allocation is contiguous with a real allocation
+		 * on the right.
+		 * Merge the new allocation with the right neighbor.
+		 */
+		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
+		xfs_bmbt_set_allf(xfs_iext_get_ext(ifp, *idx),
+			new->br_startoff, new->br_startblock,
+			new->br_blockcount + right.br_blockcount,
+			right.br_state);
+		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+
+		if (cur == NULL) {
+			rval = xfs_ilog_fext(whichfork);
+		} else {
+			rval = 0;
+			if ((error = xfs_bmbt_lookup_eq(cur,
+					right.br_startoff,
+					right.br_startblock,
+					right.br_blockcount, &i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+			if ((error = xfs_bmbt_update(cur, new->br_startoff,
+					new->br_startblock,
+					new->br_blockcount +
+						right.br_blockcount,
+					right.br_state)))
+				goto done;
+		}
+		break;
+
+	case 0:
+		/*
+		 * New allocation is not contiguous with another
+		 * real allocation.
+		 * Insert a new entry.
+		 */
+		xfs_iext_insert(ip, *idx, 1, new, state);
+		XFS_IFORK_NEXT_SET(ip, whichfork,
+			XFS_IFORK_NEXTENTS(ip, whichfork) + 1);
+		if (cur == NULL) {
+			rval = XFS_ILOG_CORE | xfs_ilog_fext(whichfork);
+		} else {
+			rval = XFS_ILOG_CORE;
+			if ((error = xfs_bmbt_lookup_eq(cur,
+					new->br_startoff,
+					new->br_startblock,
+					new->br_blockcount, &i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 0, done);
+			cur->bc_rec.b.br_state = new->br_state;
+			if ((error = xfs_btree_insert(cur, &i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+		}
+		break;
+	}
+done:
+	*logflagsp = rval;
+	return error;
+}
+
+/*
+ * Adjust the size of the new extent based on di_extsize and rt extsize.
+ */
+STATIC int
+xfs_bmap_extsize_align(
+	xfs_mount_t	*mp,
+	xfs_bmbt_irec_t	*gotp,		/* next extent pointer */
+	xfs_bmbt_irec_t	*prevp,		/* previous extent pointer */
+	xfs_extlen_t	extsz,		/* align to this extent size */
+	int		rt,		/* is this a realtime inode? */
+	int		eof,		/* is extent at end-of-file? */
+	int		delay,		/* creating delalloc extent? */
+	int		convert,	/* overwriting unwritten extent? */
+	xfs_fileoff_t	*offp,		/* in/out: aligned offset */
+	xfs_extlen_t	*lenp)		/* in/out: aligned length */
+{
+	xfs_fileoff_t	orig_off;	/* original offset */
+	xfs_extlen_t	orig_alen;	/* original length */
+	xfs_fileoff_t	orig_end;	/* original off+len */
+	xfs_fileoff_t	nexto;		/* next file offset */
+	xfs_fileoff_t	prevo;		/* previous file offset */
+	xfs_fileoff_t	align_off;	/* temp for offset */
+	xfs_extlen_t	align_alen;	/* temp for length */
+	xfs_extlen_t	temp;		/* temp for calculations */
+
+	if (convert)
+		return 0;
+
+	orig_off = align_off = *offp;
+	orig_alen = align_alen = *lenp;
+	orig_end = orig_off + orig_alen;
+
+	/*
+	 * If this request overlaps an existing extent, then don't
+	 * attempt to perform any additional alignment.
+	 */
+	if (!delay && !eof &&
+	    (orig_off >= gotp->br_startoff) &&
+	    (orig_end <= gotp->br_startoff + gotp->br_blockcount)) {
+		return 0;
+	}
+
+	/*
+	 * If the file offset is unaligned vs. the extent size
+	 * we need to align it.  This will be possible unless
+	 * the file was previously written with a kernel that didn't
+	 * perform this alignment, or if a truncate shot us in the
+	 * foot.
+	 */
+	temp = do_mod(orig_off, extsz);
+	if (temp) {
+		align_alen += temp;
+		align_off -= temp;
+	}
+	/*
+	 * Same adjustment for the end of the requested area.
+	 */
+	if ((temp = (align_alen % extsz))) {
+		align_alen += extsz - temp;
+	}
+	/*
+	 * If the previous block overlaps with this proposed allocation
+	 * then move the start forward without adjusting the length.
+	 */
+	if (prevp->br_startoff != NULLFILEOFF) {
+		if (prevp->br_startblock == HOLESTARTBLOCK)
+			prevo = prevp->br_startoff;
+		else
+			prevo = prevp->br_startoff + prevp->br_blockcount;
+	} else
+		prevo = 0;
+	if (align_off != orig_off && align_off < prevo)
+		align_off = prevo;
+	/*
+	 * If the next block overlaps with this proposed allocation
+	 * then move the start back without adjusting the length,
+	 * but not before offset 0.
+	 * This may of course make the start overlap previous block,
+	 * and if we hit the offset 0 limit then the next block
+	 * can still overlap too.
+	 */
+	if (!eof && gotp->br_startoff != NULLFILEOFF) {
+		if ((delay && gotp->br_startblock == HOLESTARTBLOCK) ||
+		    (!delay && gotp->br_startblock == DELAYSTARTBLOCK))
+			nexto = gotp->br_startoff + gotp->br_blockcount;
+		else
+			nexto = gotp->br_startoff;
+	} else
+		nexto = NULLFILEOFF;
+	if (!eof &&
+	    align_off + align_alen != orig_end &&
+	    align_off + align_alen > nexto)
+		align_off = nexto > align_alen ? nexto - align_alen : 0;
+	/*
+	 * If we're now overlapping the next or previous extent that
+	 * means we can't fit an extsz piece in this hole.  Just move
+	 * the start forward to the first valid spot and set
+	 * the length so we hit the end.
+	 */
+	if (align_off != orig_off && align_off < prevo)
+		align_off = prevo;
+	if (align_off + align_alen != orig_end &&
+	    align_off + align_alen > nexto &&
+	    nexto != NULLFILEOFF) {
+		ASSERT(nexto > prevo);
+		align_alen = nexto - align_off;
+	}
+
+	/*
+	 * If realtime, and the result isn't a multiple of the realtime
+	 * extent size we need to remove blocks until it is.
+	 */
+	if (rt && (temp = (align_alen % mp->m_sb.sb_rextsize))) {
+		/*
+		 * We're not covering the original request, or
+		 * we won't be able to once we fix the length.
+		 */
+		if (orig_off < align_off ||
+		    orig_end > align_off + align_alen ||
+		    align_alen - temp < orig_alen)
+			return XFS_ERROR(EINVAL);
+		/*
+		 * Try to fix it by moving the start up.
+		 */
+		if (align_off + temp <= orig_off) {
+			align_alen -= temp;
+			align_off += temp;
+		}
+		/*
+		 * Try to fix it by moving the end in.
+		 */
+		else if (align_off + align_alen - temp >= orig_end)
+			align_alen -= temp;
+		/*
+		 * Set the start to the minimum then trim the length.
+		 */
+		else {
+			align_alen -= orig_off - align_off;
+			align_off = orig_off;
+			align_alen -= align_alen % mp->m_sb.sb_rextsize;
+		}
+		/*
+		 * Result doesn't cover the request, fail it.
+		 */
+		if (orig_off < align_off || orig_end > align_off + align_alen)
+			return XFS_ERROR(EINVAL);
+	} else {
+		ASSERT(orig_off >= align_off);
+		ASSERT(orig_end <= align_off + align_alen);
+	}
+
+#ifdef DEBUG
+	if (!eof && gotp->br_startoff != NULLFILEOFF)
+		ASSERT(align_off + align_alen <= gotp->br_startoff);
+	if (prevp->br_startoff != NULLFILEOFF)
+		ASSERT(align_off >= prevp->br_startoff + prevp->br_blockcount);
+#endif
+
+	*lenp = align_alen;
+	*offp = align_off;
+	return 0;
+}
+
+#define XFS_ALLOC_GAP_UNITS	4
+
+STATIC void
+xfs_bmap_adjacent(
+	xfs_bmalloca_t	*ap)		/* bmap alloc argument struct */
+{
+	xfs_fsblock_t	adjust;		/* adjustment to block numbers */
+	xfs_agnumber_t	fb_agno;	/* ag number of ap->firstblock */
+	xfs_mount_t	*mp;		/* mount point structure */
+	int		nullfb;		/* true if ap->firstblock isn't set */
+	int		rt;		/* true if inode is realtime */
+
+#define	ISVALID(x,y)	\
+	(rt ? \
+		(x) < mp->m_sb.sb_rblocks : \
+		XFS_FSB_TO_AGNO(mp, x) == XFS_FSB_TO_AGNO(mp, y) && \
+		XFS_FSB_TO_AGNO(mp, x) < mp->m_sb.sb_agcount && \
+		XFS_FSB_TO_AGBNO(mp, x) < mp->m_sb.sb_agblocks)
+
+	mp = ap->ip->i_mount;
+	nullfb = ap->firstblock == NULLFSBLOCK;
+	rt = XFS_IS_REALTIME_INODE(ap->ip) && ap->userdata;
+	fb_agno = nullfb ? NULLAGNUMBER : XFS_FSB_TO_AGNO(mp, ap->firstblock);
+	/*
+	 * If allocating at eof, and there's a previous real block,
+	 * try to use its last block as our starting point.
+	 */
+	if (ap->eof && ap->prevp->br_startoff != NULLFILEOFF &&
+	    !isnullstartblock(ap->prevp->br_startblock) &&
+	    ISVALID(ap->prevp->br_startblock + ap->prevp->br_blockcount,
+		    ap->prevp->br_startblock)) {
+		ap->rval = ap->prevp->br_startblock + ap->prevp->br_blockcount;
+		/*
+		 * Adjust for the gap between prevp and us.
+		 */
+		adjust = ap->off -
+			(ap->prevp->br_startoff + ap->prevp->br_blockcount);
+		if (adjust &&
+		    ISVALID(ap->rval + adjust, ap->prevp->br_startblock))
+			ap->rval += adjust;
+	}
+	/*
+	 * If not at eof, then compare the two neighbor blocks.
+	 * Figure out whether either one gives us a good starting point,
+	 * and pick the better one.
+	 */
+	else if (!ap->eof) {
+		xfs_fsblock_t	gotbno;		/* right side block number */
+		xfs_fsblock_t	gotdiff=0;	/* right side difference */
+		xfs_fsblock_t	prevbno;	/* left side block number */
+		xfs_fsblock_t	prevdiff=0;	/* left side difference */
+
+		/*
+		 * If there's a previous (left) block, select a requested
+		 * start block based on it.
+		 */
+		if (ap->prevp->br_startoff != NULLFILEOFF &&
+		    !isnullstartblock(ap->prevp->br_startblock) &&
+		    (prevbno = ap->prevp->br_startblock +
+			       ap->prevp->br_blockcount) &&
+		    ISVALID(prevbno, ap->prevp->br_startblock)) {
+			/*
+			 * Calculate gap to end of previous block.
+			 */
+			adjust = prevdiff = ap->off -
+				(ap->prevp->br_startoff +
+				 ap->prevp->br_blockcount);
+			/*
+			 * Figure the startblock based on the previous block's
+			 * end and the gap size.
+			 * Heuristic!
+			 * If the gap is large relative to the piece we're
+			 * allocating, or using it gives us an invalid block
+			 * number, then just use the end of the previous block.
+			 */
+			if (prevdiff <= XFS_ALLOC_GAP_UNITS * ap->alen &&
+			    ISVALID(prevbno + prevdiff,
+				    ap->prevp->br_startblock))
+				prevbno += adjust;
+			else
+				prevdiff += adjust;
+			/*
+			 * If the firstblock forbids it, can't use it,
+			 * must use default.
+			 */
+			if (!rt && !nullfb &&
+			    XFS_FSB_TO_AGNO(mp, prevbno) != fb_agno)
+				prevbno = NULLFSBLOCK;
+		}
+		/*
+		 * No previous block or can't follow it, just default.
+		 */
+		else
+			prevbno = NULLFSBLOCK;
+		/*
+		 * If there's a following (right) block, select a requested
+		 * start block based on it.
+		 */
+		if (!isnullstartblock(ap->gotp->br_startblock)) {
+			/*
+			 * Calculate gap to start of next block.
+			 */
+			adjust = gotdiff = ap->gotp->br_startoff - ap->off;
+			/*
+			 * Figure the startblock based on the next block's
+			 * start and the gap size.
+			 */
+			gotbno = ap->gotp->br_startblock;
+			/*
+			 * Heuristic!
+			 * If the gap is large relative to the piece we're
+			 * allocating, or using it gives us an invalid block
+			 * number, then just use the start of the next block
+			 * offset by our length.
+			 */
+			if (gotdiff <= XFS_ALLOC_GAP_UNITS * ap->alen &&
+			    ISVALID(gotbno - gotdiff, gotbno))
+				gotbno -= adjust;
+			else if (ISVALID(gotbno - ap->alen, gotbno)) {
+				gotbno -= ap->alen;
+				gotdiff += adjust - ap->alen;
+			} else
+				gotdiff += adjust;
+			/*
+			 * If the firstblock forbids it, can't use it,
+			 * must use default.
+			 */
+			if (!rt && !nullfb &&
+			    XFS_FSB_TO_AGNO(mp, gotbno) != fb_agno)
+				gotbno = NULLFSBLOCK;
+		}
+		/*
+		 * No next block, just default.
+		 */
+		else
+			gotbno = NULLFSBLOCK;
+		/*
+		 * If both valid, pick the better one, else the only good
+		 * one, else ap->rval is already set (to 0 or the inode block).
+		 */
+		if (prevbno != NULLFSBLOCK && gotbno != NULLFSBLOCK)
+			ap->rval = prevdiff <= gotdiff ? prevbno : gotbno;
+		else if (prevbno != NULLFSBLOCK)
+			ap->rval = prevbno;
+		else if (gotbno != NULLFSBLOCK)
+			ap->rval = gotbno;
+	}
+#undef ISVALID
+}
+
+STATIC int
+xfs_bmap_rtalloc(
+	xfs_bmalloca_t	*ap)		/* bmap alloc argument struct */
+{
+	xfs_alloctype_t	atype = 0;	/* type for allocation routines */
+	int		error;		/* error return value */
+	xfs_mount_t	*mp;		/* mount point structure */
+	xfs_extlen_t	prod = 0;	/* product factor for allocators */
+	xfs_extlen_t	ralen = 0;	/* realtime allocation length */
+	xfs_extlen_t	align;		/* minimum allocation alignment */
+	xfs_rtblock_t	rtb;
+
+	mp = ap->ip->i_mount;
+	align = xfs_get_extsz_hint(ap->ip);
+	prod = align / mp->m_sb.sb_rextsize;
+	error = xfs_bmap_extsize_align(mp, ap->gotp, ap->prevp,
+					align, 1, ap->eof, 0,
+					ap->conv, &ap->off, &ap->alen);
+	if (error)
+		return error;
+	ASSERT(ap->alen);
+	ASSERT(ap->alen % mp->m_sb.sb_rextsize == 0);
+
+	/*
+	 * If the offset & length are not perfectly aligned
+	 * then kill prod, it will just get us in trouble.
+	 */
+	if (do_mod(ap->off, align) || ap->alen % align)
+		prod = 1;
+	/*
+	 * Set ralen to be the actual requested length in rtextents.
+	 */
+	ralen = ap->alen / mp->m_sb.sb_rextsize;
+	/*
+	 * If the old value was close enough to MAXEXTLEN that
+	 * we rounded up to it, cut it back so it's valid again.
+	 * Note that if it's a really large request (bigger than
+	 * MAXEXTLEN), we don't hear about that number, and can't
+	 * adjust the starting point to match it.
+	 */
+	if (ralen * mp->m_sb.sb_rextsize >= MAXEXTLEN)
+		ralen = MAXEXTLEN / mp->m_sb.sb_rextsize;
+
+	/*
+	 * Lock out other modifications to the RT bitmap inode.
+	 */
+	xfs_ilock(mp->m_rbmip, XFS_ILOCK_EXCL);
+	xfs_trans_ijoin_ref(ap->tp, mp->m_rbmip, XFS_ILOCK_EXCL);
+
+	/*
+	 * If it's an allocation to an empty file at offset 0,
+	 * pick an extent that will space things out in the rt area.
+	 */
+	if (ap->eof && ap->off == 0) {
+		xfs_rtblock_t uninitialized_var(rtx); /* realtime extent no */
+
+		error = xfs_rtpick_extent(mp, ap->tp, ralen, &rtx);
+		if (error)
+			return error;
+		ap->rval = rtx * mp->m_sb.sb_rextsize;
+	} else {
+		ap->rval = 0;
+	}
+
+	xfs_bmap_adjacent(ap);
+
+	/*
+	 * Realtime allocation, done through xfs_rtallocate_extent.
+	 */
+	atype = ap->rval == 0 ?  XFS_ALLOCTYPE_ANY_AG : XFS_ALLOCTYPE_NEAR_BNO;
+	do_div(ap->rval, mp->m_sb.sb_rextsize);
+	rtb = ap->rval;
+	ap->alen = ralen;
+	if ((error = xfs_rtallocate_extent(ap->tp, ap->rval, 1, ap->alen,
+				&ralen, atype, ap->wasdel, prod, &rtb)))
+		return error;
+	if (rtb == NULLFSBLOCK && prod > 1 &&
+	    (error = xfs_rtallocate_extent(ap->tp, ap->rval, 1,
+					   ap->alen, &ralen, atype,
+					   ap->wasdel, 1, &rtb)))
+		return error;
+	ap->rval = rtb;
+	if (ap->rval != NULLFSBLOCK) {
+		ap->rval *= mp->m_sb.sb_rextsize;
+		ralen *= mp->m_sb.sb_rextsize;
+		ap->alen = ralen;
+		ap->ip->i_d.di_nblocks += ralen;
+		xfs_trans_log_inode(ap->tp, ap->ip, XFS_ILOG_CORE);
+		if (ap->wasdel)
+			ap->ip->i_delayed_blks -= ralen;
+		/*
+		 * Adjust the disk quota also. This was reserved
+		 * earlier.
+		 */
+		xfs_trans_mod_dquot_byino(ap->tp, ap->ip,
+			ap->wasdel ? XFS_TRANS_DQ_DELRTBCOUNT :
+					XFS_TRANS_DQ_RTBCOUNT, (long) ralen);
+	} else {
+		ap->alen = 0;
+	}
+	return 0;
+}
+
+STATIC int
+xfs_bmap_btalloc_nullfb(
+	struct xfs_bmalloca	*ap,
+	struct xfs_alloc_arg	*args,
+	xfs_extlen_t		*blen)
+{
+	struct xfs_mount	*mp = ap->ip->i_mount;
+	struct xfs_perag	*pag;
+	xfs_agnumber_t		ag, startag;
+	int			notinit = 0;
+	int			error;
+
+	if (ap->userdata && xfs_inode_is_filestream(ap->ip))
+		args->type = XFS_ALLOCTYPE_NEAR_BNO;
+	else
+		args->type = XFS_ALLOCTYPE_START_BNO;
+	args->total = ap->total;
+
+	/*
+	 * Search for an allocation group with a single extent large enough
+	 * for the request.  If one isn't found, then adjust the minimum
+	 * allocation size to the largest space found.
+	 */
+	startag = ag = XFS_FSB_TO_AGNO(mp, args->fsbno);
+	if (startag == NULLAGNUMBER)
+		startag = ag = 0;
+
+	pag = xfs_perag_get(mp, ag);
+	while (*blen < args->maxlen) {
+		if (!pag->pagf_init) {
+			error = xfs_alloc_pagf_init(mp, args->tp, ag,
+						    XFS_ALLOC_FLAG_TRYLOCK);
+			if (error) {
+				xfs_perag_put(pag);
+				return error;
+			}
+		}
+
+		/*
+		 * See xfs_alloc_fix_freelist...
+		 */
+		if (pag->pagf_init) {
+			xfs_extlen_t	longest;
+			longest = xfs_alloc_longest_free_extent(mp, pag);
+			if (*blen < longest)
+				*blen = longest;
+		} else
+			notinit = 1;
+
+		if (xfs_inode_is_filestream(ap->ip)) {
+			if (*blen >= args->maxlen)
+				break;
+
+			if (ap->userdata) {
+				/*
+				 * If startag is an invalid AG, we've
+				 * come here once before and
+				 * xfs_filestream_new_ag picked the
+				 * best currently available.
+				 *
+				 * Don't continue looping, since we
+				 * could loop forever.
+				 */
+				if (startag == NULLAGNUMBER)
+					break;
+
+				error = xfs_filestream_new_ag(ap, &ag);
+				xfs_perag_put(pag);
+				if (error)
+					return error;
+
+				/* loop again to set 'blen'*/
+				startag = NULLAGNUMBER;
+				pag = xfs_perag_get(mp, ag);
+				continue;
+			}
+		}
+		if (++ag == mp->m_sb.sb_agcount)
+			ag = 0;
+		if (ag == startag)
+			break;
+		xfs_perag_put(pag);
+		pag = xfs_perag_get(mp, ag);
+	}
+	xfs_perag_put(pag);
+
+	/*
+	 * Since the above loop did a BUF_TRYLOCK, it is
+	 * possible that there is space for this request.
+	 */
+	if (notinit || *blen < ap->minlen)
+		args->minlen = ap->minlen;
+	/*
+	 * If the best seen length is less than the request
+	 * length, use the best as the minimum.
+	 */
+	else if (*blen < args->maxlen)
+		args->minlen = *blen;
+	/*
+	 * Otherwise we've seen an extent as big as maxlen,
+	 * use that as the minimum.
+	 */
+	else
+		args->minlen = args->maxlen;
+
+	/*
+	 * set the failure fallback case to look in the selected
+	 * AG as the stream may have moved.
+	 */
+	if (xfs_inode_is_filestream(ap->ip))
+		ap->rval = args->fsbno = XFS_AGB_TO_FSB(mp, ag, 0);
+
+	return 0;
+}
+
+STATIC int
+xfs_bmap_btalloc(
+	xfs_bmalloca_t	*ap)		/* bmap alloc argument struct */
+{
+	xfs_mount_t	*mp;		/* mount point structure */
+	xfs_alloctype_t	atype = 0;	/* type for allocation routines */
+	xfs_extlen_t	align;		/* minimum allocation alignment */
+	xfs_agnumber_t	fb_agno;	/* ag number of ap->firstblock */
+	xfs_agnumber_t	ag;
+	xfs_alloc_arg_t	args;
+	xfs_extlen_t	blen;
+	xfs_extlen_t	nextminlen = 0;
+	int		nullfb;		/* true if ap->firstblock isn't set */
+	int		isaligned;
+	int		tryagain;
+	int		error;
+
+	mp = ap->ip->i_mount;
+	align = ap->userdata ? xfs_get_extsz_hint(ap->ip) : 0;
+	if (unlikely(align)) {
+		error = xfs_bmap_extsize_align(mp, ap->gotp, ap->prevp,
+						align, 0, ap->eof, 0, ap->conv,
+						&ap->off, &ap->alen);
+		ASSERT(!error);
+		ASSERT(ap->alen);
+	}
+	nullfb = ap->firstblock == NULLFSBLOCK;
+	fb_agno = nullfb ? NULLAGNUMBER : XFS_FSB_TO_AGNO(mp, ap->firstblock);
+	if (nullfb) {
+		if (ap->userdata && xfs_inode_is_filestream(ap->ip)) {
+			ag = xfs_filestream_lookup_ag(ap->ip);
+			ag = (ag != NULLAGNUMBER) ? ag : 0;
+			ap->rval = XFS_AGB_TO_FSB(mp, ag, 0);
+		} else {
+			ap->rval = XFS_INO_TO_FSB(mp, ap->ip->i_ino);
+		}
+	} else
+		ap->rval = ap->firstblock;
+
+	xfs_bmap_adjacent(ap);
+
+	/*
+	 * If allowed, use ap->rval; otherwise must use firstblock since
+	 * it's in the right allocation group.
+	 */
+	if (nullfb || XFS_FSB_TO_AGNO(mp, ap->rval) == fb_agno)
+		;
+	else
+		ap->rval = ap->firstblock;
+	/*
+	 * Normal allocation, done through xfs_alloc_vextent.
+	 */
+	tryagain = isaligned = 0;
+	args.tp = ap->tp;
+	args.mp = mp;
+	args.fsbno = ap->rval;
+
+	/* Trim the allocation back to the maximum an AG can fit. */
+	args.maxlen = MIN(ap->alen, XFS_ALLOC_AG_MAX_USABLE(mp));
+	args.firstblock = ap->firstblock;
+	blen = 0;
+	if (nullfb) {
+		error = xfs_bmap_btalloc_nullfb(ap, &args, &blen);
+		if (error)
+			return error;
+	} else if (ap->low) {
+		if (xfs_inode_is_filestream(ap->ip))
+			args.type = XFS_ALLOCTYPE_FIRST_AG;
+		else
+			args.type = XFS_ALLOCTYPE_START_BNO;
+		args.total = args.minlen = ap->minlen;
+	} else {
+		args.type = XFS_ALLOCTYPE_NEAR_BNO;
+		args.total = ap->total;
+		args.minlen = ap->minlen;
+	}
+	/* apply extent size hints if obtained earlier */
+	if (unlikely(align)) {
+		args.prod = align;
+		if ((args.mod = (xfs_extlen_t)do_mod(ap->off, args.prod)))
+			args.mod = (xfs_extlen_t)(args.prod - args.mod);
+	} else if (mp->m_sb.sb_blocksize >= PAGE_CACHE_SIZE) {
+		args.prod = 1;
+		args.mod = 0;
+	} else {
+		args.prod = PAGE_CACHE_SIZE >> mp->m_sb.sb_blocklog;
+		if ((args.mod = (xfs_extlen_t)(do_mod(ap->off, args.prod))))
+			args.mod = (xfs_extlen_t)(args.prod - args.mod);
+	}
+	/*
+	 * If we are not low on available data blocks, and the
+	 * underlying logical volume manager is a stripe, and
+	 * the file offset is zero then try to allocate data
+	 * blocks on stripe unit boundary.
+	 * NOTE: ap->aeof is only set if the allocation length
+	 * is >= the stripe unit and the allocation offset is
+	 * at the end of file.
+	 */
+	if (!ap->low && ap->aeof) {
+		if (!ap->off) {
+			args.alignment = mp->m_dalign;
+			atype = args.type;
+			isaligned = 1;
+			/*
+			 * Adjust for alignment
+			 */
+			if (blen > args.alignment && blen <= args.maxlen)
+				args.minlen = blen - args.alignment;
+			args.minalignslop = 0;
+		} else {
+			/*
+			 * First try an exact bno allocation.
+			 * If it fails then do a near or start bno
+			 * allocation with alignment turned on.
+			 */
+			atype = args.type;
+			tryagain = 1;
+			args.type = XFS_ALLOCTYPE_THIS_BNO;
+			args.alignment = 1;
+			/*
+			 * Compute the minlen+alignment for the
+			 * next case.  Set slop so that the value
+			 * of minlen+alignment+slop doesn't go up
+			 * between the calls.
+			 */
+			if (blen > mp->m_dalign && blen <= args.maxlen)
+				nextminlen = blen - mp->m_dalign;
+			else
+				nextminlen = args.minlen;
+			if (nextminlen + mp->m_dalign > args.minlen + 1)
+				args.minalignslop =
+					nextminlen + mp->m_dalign -
+					args.minlen - 1;
+			else
+				args.minalignslop = 0;
+		}
+	} else {
+		args.alignment = 1;
+		args.minalignslop = 0;
+	}
+	args.minleft = ap->minleft;
+	args.wasdel = ap->wasdel;
+	args.isfl = 0;
+	args.userdata = ap->userdata;
+	if ((error = xfs_alloc_vextent(&args)))
+		return error;
+	if (tryagain && args.fsbno == NULLFSBLOCK) {
+		/*
+		 * Exact allocation failed. Now try with alignment
+		 * turned on.
+		 */
+		args.type = atype;
+		args.fsbno = ap->rval;
+		args.alignment = mp->m_dalign;
+		args.minlen = nextminlen;
+		args.minalignslop = 0;
+		isaligned = 1;
+		if ((error = xfs_alloc_vextent(&args)))
+			return error;
+	}
+	if (isaligned && args.fsbno == NULLFSBLOCK) {
+		/*
+		 * allocation failed, so turn off alignment and
+		 * try again.
+		 */
+		args.type = atype;
+		args.fsbno = ap->rval;
+		args.alignment = 0;
+		if ((error = xfs_alloc_vextent(&args)))
+			return error;
+	}
+	if (args.fsbno == NULLFSBLOCK && nullfb &&
+	    args.minlen > ap->minlen) {
+		args.minlen = ap->minlen;
+		args.type = XFS_ALLOCTYPE_START_BNO;
+		args.fsbno = ap->rval;
+		if ((error = xfs_alloc_vextent(&args)))
+			return error;
+	}
+	if (args.fsbno == NULLFSBLOCK && nullfb) {
+		args.fsbno = 0;
+		args.type = XFS_ALLOCTYPE_FIRST_AG;
+		args.total = ap->minlen;
+		args.minleft = 0;
+		if ((error = xfs_alloc_vextent(&args)))
+			return error;
+		ap->low = 1;
+	}
+	if (args.fsbno != NULLFSBLOCK) {
+		ap->firstblock = ap->rval = args.fsbno;
+		ASSERT(nullfb || fb_agno == args.agno ||
+		       (ap->low && fb_agno < args.agno));
+		ap->alen = args.len;
+		ap->ip->i_d.di_nblocks += args.len;
+		xfs_trans_log_inode(ap->tp, ap->ip, XFS_ILOG_CORE);
+		if (ap->wasdel)
+			ap->ip->i_delayed_blks -= args.len;
+		/*
+		 * Adjust the disk quota also. This was reserved
+		 * earlier.
+		 */
+		xfs_trans_mod_dquot_byino(ap->tp, ap->ip,
+			ap->wasdel ? XFS_TRANS_DQ_DELBCOUNT :
+					XFS_TRANS_DQ_BCOUNT,
+			(long) args.len);
+	} else {
+		ap->rval = NULLFSBLOCK;
+		ap->alen = 0;
+	}
+	return 0;
+}
+
+/*
+ * xfs_bmap_alloc is called by xfs_bmapi to allocate an extent for a file.
+ * It figures out where to ask the underlying allocator to put the new extent.
+ */
+STATIC int
+xfs_bmap_alloc(
+	xfs_bmalloca_t	*ap)		/* bmap alloc argument struct */
+{
+	if (XFS_IS_REALTIME_INODE(ap->ip) && ap->userdata)
+		return xfs_bmap_rtalloc(ap);
+	return xfs_bmap_btalloc(ap);
+}
+
+/*
+ * Transform a btree format file with only one leaf node, where the
+ * extents list will fit in the inode, into an extents format file.
+ * Since the file extents are already in-core, all we have to do is
+ * give up the space for the btree root and pitch the leaf block.
+ */
+STATIC int				/* error */
+xfs_bmap_btree_to_extents(
+	xfs_trans_t		*tp,	/* transaction pointer */
+	xfs_inode_t		*ip,	/* incore inode pointer */
+	xfs_btree_cur_t		*cur,	/* btree cursor */
+	int			*logflagsp, /* inode logging flags */
+	int			whichfork)  /* data or attr fork */
+{
+	/* REFERENCED */
+	struct xfs_btree_block	*cblock;/* child btree block */
+	xfs_fsblock_t		cbno;	/* child block number */
+	xfs_buf_t		*cbp;	/* child block's buffer */
+	int			error;	/* error return value */
+	xfs_ifork_t		*ifp;	/* inode fork data */
+	xfs_mount_t		*mp;	/* mount point structure */
+	__be64			*pp;	/* ptr to block address */
+	struct xfs_btree_block	*rblock;/* root btree block */
+
+	mp = ip->i_mount;
+	ifp = XFS_IFORK_PTR(ip, whichfork);
+	ASSERT(ifp->if_flags & XFS_IFEXTENTS);
+	ASSERT(XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_BTREE);
+	rblock = ifp->if_broot;
+	ASSERT(be16_to_cpu(rblock->bb_level) == 1);
+	ASSERT(be16_to_cpu(rblock->bb_numrecs) == 1);
+	ASSERT(xfs_bmbt_maxrecs(mp, ifp->if_broot_bytes, 0) == 1);
+	pp = XFS_BMAP_BROOT_PTR_ADDR(mp, rblock, 1, ifp->if_broot_bytes);
+	cbno = be64_to_cpu(*pp);
+	*logflagsp = 0;
+#ifdef DEBUG
+	if ((error = xfs_btree_check_lptr(cur, cbno, 1)))
+		return error;
+#endif
+	if ((error = xfs_btree_read_bufl(mp, tp, cbno, 0, &cbp,
+			XFS_BMAP_BTREE_REF)))
+		return error;
+	cblock = XFS_BUF_TO_BLOCK(cbp);
+	if ((error = xfs_btree_check_block(cur, cblock, 0, cbp)))
+		return error;
+	xfs_bmap_add_free(cbno, 1, cur->bc_private.b.flist, mp);
+	ip->i_d.di_nblocks--;
+	xfs_trans_mod_dquot_byino(tp, ip, XFS_TRANS_DQ_BCOUNT, -1L);
+	xfs_trans_binval(tp, cbp);
+	if (cur->bc_bufs[0] == cbp)
+		cur->bc_bufs[0] = NULL;
+	xfs_iroot_realloc(ip, -1, whichfork);
+	ASSERT(ifp->if_broot == NULL);
+	ASSERT((ifp->if_flags & XFS_IFBROOT) == 0);
+	XFS_IFORK_FMT_SET(ip, whichfork, XFS_DINODE_FMT_EXTENTS);
+	*logflagsp = XFS_ILOG_CORE | xfs_ilog_fext(whichfork);
+	return 0;
+}
+
+/*
+ * Called by xfs_bmapi to update file extent records and the btree
+ * after removing space (or undoing a delayed allocation).
+ */
+STATIC int				/* error */
+xfs_bmap_del_extent(
+	xfs_inode_t		*ip,	/* incore inode pointer */
+	xfs_trans_t		*tp,	/* current transaction pointer */
+	xfs_extnum_t		*idx,	/* extent number to update/delete */
+	xfs_bmap_free_t		*flist,	/* list of extents to be freed */
+	xfs_btree_cur_t		*cur,	/* if null, not a btree */
+	xfs_bmbt_irec_t		*del,	/* data to remove from extents */
+	int			*logflagsp, /* inode logging flags */
+	int			whichfork) /* data or attr fork */
+{
+	xfs_filblks_t		da_new;	/* new delay-alloc indirect blocks */
+	xfs_filblks_t		da_old;	/* old delay-alloc indirect blocks */
+	xfs_fsblock_t		del_endblock=0;	/* first block past del */
+	xfs_fileoff_t		del_endoff;	/* first offset past del */
+	int			delay;	/* current block is delayed allocated */
+	int			do_fx;	/* free extent at end of routine */
+	xfs_bmbt_rec_host_t	*ep;	/* current extent entry pointer */
+	int			error;	/* error return value */
+	int			flags;	/* inode logging flags */
+	xfs_bmbt_irec_t		got;	/* current extent entry */
+	xfs_fileoff_t		got_endoff;	/* first offset past got */
+	int			i;	/* temp state */
+	xfs_ifork_t		*ifp;	/* inode fork pointer */
+	xfs_mount_t		*mp;	/* mount structure */
+	xfs_filblks_t		nblks;	/* quota/sb block count */
+	xfs_bmbt_irec_t		new;	/* new record to be inserted */
+	/* REFERENCED */
+	uint			qfield;	/* quota field to update */
+	xfs_filblks_t		temp;	/* for indirect length calculations */
+	xfs_filblks_t		temp2;	/* for indirect length calculations */
+	int			state = 0;
+
+	XFS_STATS_INC(xs_del_exlist);
+
+	if (whichfork == XFS_ATTR_FORK)
+		state |= BMAP_ATTRFORK;
+
+	mp = ip->i_mount;
+	ifp = XFS_IFORK_PTR(ip, whichfork);
+	ASSERT((*idx >= 0) && (*idx < ifp->if_bytes /
+		(uint)sizeof(xfs_bmbt_rec_t)));
+	ASSERT(del->br_blockcount > 0);
+	ep = xfs_iext_get_ext(ifp, *idx);
+	xfs_bmbt_get_all(ep, &got);
+	ASSERT(got.br_startoff <= del->br_startoff);
+	del_endoff = del->br_startoff + del->br_blockcount;
+	got_endoff = got.br_startoff + got.br_blockcount;
+	ASSERT(got_endoff >= del_endoff);
+	delay = isnullstartblock(got.br_startblock);
+	ASSERT(isnullstartblock(del->br_startblock) == delay);
+	flags = 0;
+	qfield = 0;
+	error = 0;
+	/*
+	 * If deleting a real allocation, must free up the disk space.
+	 */
+	if (!delay) {
+		flags = XFS_ILOG_CORE;
+		/*
+		 * Realtime allocation.  Free it and record di_nblocks update.
+		 */
+		if (whichfork == XFS_DATA_FORK && XFS_IS_REALTIME_INODE(ip)) {
+			xfs_fsblock_t	bno;
+			xfs_filblks_t	len;
+
+			ASSERT(do_mod(del->br_blockcount,
+				      mp->m_sb.sb_rextsize) == 0);
+			ASSERT(do_mod(del->br_startblock,
+				      mp->m_sb.sb_rextsize) == 0);
+			bno = del->br_startblock;
+			len = del->br_blockcount;
+			do_div(bno, mp->m_sb.sb_rextsize);
+			do_div(len, mp->m_sb.sb_rextsize);
+			if ((error = xfs_rtfree_extent(ip->i_transp, bno,
+					(xfs_extlen_t)len)))
+				goto done;
+			do_fx = 0;
+			nblks = len * mp->m_sb.sb_rextsize;
+			qfield = XFS_TRANS_DQ_RTBCOUNT;
+		}
+		/*
+		 * Ordinary allocation.
+		 */
+		else {
+			do_fx = 1;
+			nblks = del->br_blockcount;
+			qfield = XFS_TRANS_DQ_BCOUNT;
+		}
+		/*
+		 * Set up del_endblock and cur for later.
+		 */
+		del_endblock = del->br_startblock + del->br_blockcount;
+		if (cur) {
+			if ((error = xfs_bmbt_lookup_eq(cur, got.br_startoff,
+					got.br_startblock, got.br_blockcount,
+					&i)))
+				goto done;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+		}
+		da_old = da_new = 0;
+	} else {
+		da_old = startblockval(got.br_startblock);
+		da_new = 0;
+		nblks = 0;
+		do_fx = 0;
+	}
+	/*
+	 * Set flag value to use in switch statement.
+	 * Left-contig is 2, right-contig is 1.
+	 */
+	switch (((got.br_startoff == del->br_startoff) << 1) |
+		(got_endoff == del_endoff)) {
+	case 3:
+		/*
+		 * Matches the whole extent.  Delete the entry.
+		 */
+		xfs_iext_remove(ip, *idx, 1,
+				whichfork == XFS_ATTR_FORK ? BMAP_ATTRFORK : 0);
+		--*idx;
+		if (delay)
+			break;
+
+		XFS_IFORK_NEXT_SET(ip, whichfork,
+			XFS_IFORK_NEXTENTS(ip, whichfork) - 1);
+		flags |= XFS_ILOG_CORE;
+		if (!cur) {
+			flags |= xfs_ilog_fext(whichfork);
+			break;
+		}
+		if ((error = xfs_btree_delete(cur, &i)))
+			goto done;
+		XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+		break;
+
+	case 2:
+		/*
+		 * Deleting the first part of the extent.
+		 */
+		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
+		xfs_bmbt_set_startoff(ep, del_endoff);
+		temp = got.br_blockcount - del->br_blockcount;
+		xfs_bmbt_set_blockcount(ep, temp);
+		if (delay) {
+			temp = XFS_FILBLKS_MIN(xfs_bmap_worst_indlen(ip, temp),
+				da_old);
+			xfs_bmbt_set_startblock(ep, nullstartblock((int)temp));
+			trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+			da_new = temp;
+			break;
+		}
+		xfs_bmbt_set_startblock(ep, del_endblock);
+		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+		if (!cur) {
+			flags |= xfs_ilog_fext(whichfork);
+			break;
+		}
+		if ((error = xfs_bmbt_update(cur, del_endoff, del_endblock,
+				got.br_blockcount - del->br_blockcount,
+				got.br_state)))
+			goto done;
+		break;
+
+	case 1:
+		/*
+		 * Deleting the last part of the extent.
+		 */
+		temp = got.br_blockcount - del->br_blockcount;
+		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
+		xfs_bmbt_set_blockcount(ep, temp);
+		if (delay) {
+			temp = XFS_FILBLKS_MIN(xfs_bmap_worst_indlen(ip, temp),
+				da_old);
+			xfs_bmbt_set_startblock(ep, nullstartblock((int)temp));
+			trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+			da_new = temp;
+			break;
+		}
+		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+		if (!cur) {
+			flags |= xfs_ilog_fext(whichfork);
+			break;
+		}
+		if ((error = xfs_bmbt_update(cur, got.br_startoff,
+				got.br_startblock,
+				got.br_blockcount - del->br_blockcount,
+				got.br_state)))
+			goto done;
+		break;
+
+	case 0:
+		/*
+		 * Deleting the middle of the extent.
+		 */
+		temp = del->br_startoff - got.br_startoff;
+		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
+		xfs_bmbt_set_blockcount(ep, temp);
+		new.br_startoff = del_endoff;
+		temp2 = got_endoff - del_endoff;
+		new.br_blockcount = temp2;
+		new.br_state = got.br_state;
+		if (!delay) {
+			new.br_startblock = del_endblock;
+			flags |= XFS_ILOG_CORE;
+			if (cur) {
+				if ((error = xfs_bmbt_update(cur,
+						got.br_startoff,
+						got.br_startblock, temp,
+						got.br_state)))
+					goto done;
+				if ((error = xfs_btree_increment(cur, 0, &i)))
+					goto done;
+				cur->bc_rec.b = new;
+				error = xfs_btree_insert(cur, &i);
+				if (error && error != ENOSPC)
+					goto done;
+				/*
+				 * If get no-space back from btree insert,
+				 * it tried a split, and we have a zero
+				 * block reservation.
+				 * Fix up our state and return the error.
+				 */
+				if (error == ENOSPC) {
+					/*
+					 * Reset the cursor, don't trust
+					 * it after any insert operation.
+					 */
+					if ((error = xfs_bmbt_lookup_eq(cur,
+							got.br_startoff,
+							got.br_startblock,
+							temp, &i)))
+						goto done;
+					XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+					/*
+					 * Update the btree record back
+					 * to the original value.
+					 */
+					if ((error = xfs_bmbt_update(cur,
+							got.br_startoff,
+							got.br_startblock,
+							got.br_blockcount,
+							got.br_state)))
+						goto done;
+					/*
+					 * Reset the extent record back
+					 * to the original value.
+					 */
+					xfs_bmbt_set_blockcount(ep,
+						got.br_blockcount);
+					flags = 0;
+					error = XFS_ERROR(ENOSPC);
+					goto done;
+				}
+				XFS_WANT_CORRUPTED_GOTO(i == 1, done);
+			} else
+				flags |= xfs_ilog_fext(whichfork);
+			XFS_IFORK_NEXT_SET(ip, whichfork,
+				XFS_IFORK_NEXTENTS(ip, whichfork) + 1);
+		} else {
+			ASSERT(whichfork == XFS_DATA_FORK);
+			temp = xfs_bmap_worst_indlen(ip, temp);
+			xfs_bmbt_set_startblock(ep, nullstartblock((int)temp));
+			temp2 = xfs_bmap_worst_indlen(ip, temp2);
+			new.br_startblock = nullstartblock((int)temp2);
+			da_new = temp + temp2;
+			while (da_new > da_old) {
+				if (temp) {
+					temp--;
+					da_new--;
+					xfs_bmbt_set_startblock(ep,
+						nullstartblock((int)temp));
+				}
+				if (da_new == da_old)
+					break;
+				if (temp2) {
+					temp2--;
+					da_new--;
+					new.br_startblock =
+						nullstartblock((int)temp2);
+				}
+			}
+		}
+		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+		xfs_iext_insert(ip, *idx + 1, 1, &new, state);
+		++*idx;
+		break;
+	}
+	/*
+	 * If we need to, add to list of extents to delete.
+	 */
+	if (do_fx)
+		xfs_bmap_add_free(del->br_startblock, del->br_blockcount, flist,
+			mp);
+	/*
+	 * Adjust inode # blocks in the file.
+	 */
+	if (nblks)
+		ip->i_d.di_nblocks -= nblks;
+	/*
+	 * Adjust quota data.
+	 */
+	if (qfield)
+		xfs_trans_mod_dquot_byino(tp, ip, qfield, (long)-nblks);
+
+	/*
+	 * Account for change in delayed indirect blocks.
+	 * Nothing to do for disk quota accounting here.
+	 */
+	ASSERT(da_old >= da_new);
+	if (da_old > da_new) {
+		xfs_icsb_modify_counters(mp, XFS_SBS_FDBLOCKS,
+			(int64_t)(da_old - da_new), 0);
+	}
+done:
+	*logflagsp = flags;
+	return error;
+}
+
+/*
+ * Remove the entry "free" from the free item list.  Prev points to the
+ * previous entry, unless "free" is the head of the list.
+ */
+STATIC void
+xfs_bmap_del_free(
+	xfs_bmap_free_t		*flist,	/* free item list header */
+	xfs_bmap_free_item_t	*prev,	/* previous item on list, if any */
+	xfs_bmap_free_item_t	*free)	/* list item to be freed */
+{
+	if (prev)
+		prev->xbfi_next = free->xbfi_next;
+	else
+		flist->xbf_first = free->xbfi_next;
+	flist->xbf_count--;
+	kmem_zone_free(xfs_bmap_free_item_zone, free);
+}
+
+/*
+ * Convert an extents-format file into a btree-format file.
+ * The new file will have a root block (in the inode) and a single child block.
+ */
+STATIC int					/* error */
+xfs_bmap_extents_to_btree(
+	xfs_trans_t		*tp,		/* transaction pointer */
+	xfs_inode_t		*ip,		/* incore inode pointer */
+	xfs_fsblock_t		*firstblock,	/* first-block-allocated */
+	xfs_bmap_free_t		*flist,		/* blocks freed in xaction */
+	xfs_btree_cur_t		**curp,		/* cursor returned to caller */
+	int			wasdel,		/* converting a delayed alloc */
+	int			*logflagsp,	/* inode logging flags */
+	int			whichfork)	/* data or attr fork */
+{
+	struct xfs_btree_block	*ablock;	/* allocated (child) bt block */
+	xfs_buf_t		*abp;		/* buffer for ablock */
+	xfs_alloc_arg_t		args;		/* allocation arguments */
+	xfs_bmbt_rec_t		*arp;		/* child record pointer */
+	struct xfs_btree_block	*block;		/* btree root block */
+	xfs_btree_cur_t		*cur;		/* bmap btree cursor */
+	xfs_bmbt_rec_host_t	*ep;		/* extent record pointer */
+	int			error;		/* error return value */
+	xfs_extnum_t		i, cnt;		/* extent record index */
+	xfs_ifork_t		*ifp;		/* inode fork pointer */
+	xfs_bmbt_key_t		*kp;		/* root block key pointer */
+	xfs_mount_t		*mp;		/* mount structure */
+	xfs_extnum_t		nextents;	/* number of file extents */
+	xfs_bmbt_ptr_t		*pp;		/* root block address pointer */
+
+	ifp = XFS_IFORK_PTR(ip, whichfork);
+	ASSERT(XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_EXTENTS);
+	ASSERT(ifp->if_ext_max ==
+	       XFS_IFORK_SIZE(ip, whichfork) / (uint)sizeof(xfs_bmbt_rec_t));
+	/*
+	 * Make space in the inode incore.
+	 */
+	xfs_iroot_realloc(ip, 1, whichfork);
+	ifp->if_flags |= XFS_IFBROOT;
+
+	/*
+	 * Fill in the root.
+	 */
+	block = ifp->if_broot;
+	block->bb_magic = cpu_to_be32(XFS_BMAP_MAGIC);
+	block->bb_level = cpu_to_be16(1);
+	block->bb_numrecs = cpu_to_be16(1);
+	block->bb_u.l.bb_leftsib = cpu_to_be64(NULLDFSBNO);
+	block->bb_u.l.bb_rightsib = cpu_to_be64(NULLDFSBNO);
+
+	/*
+	 * Need a cursor.  Can't allocate until bb_level is filled in.
+	 */
+	mp = ip->i_mount;
+	cur = xfs_bmbt_init_cursor(mp, tp, ip, whichfork);
+	cur->bc_private.b.firstblock = *firstblock;
+	cur->bc_private.b.flist = flist;
+	cur->bc_private.b.flags = wasdel ? XFS_BTCUR_BPRV_WASDEL : 0;
+	/*
+	 * Convert to a btree with two levels, one record in root.
+	 */
+	XFS_IFORK_FMT_SET(ip, whichfork, XFS_DINODE_FMT_BTREE);
+	args.tp = tp;
+	args.mp = mp;
+	args.firstblock = *firstblock;
+	if (*firstblock == NULLFSBLOCK) {
+		args.type = XFS_ALLOCTYPE_START_BNO;
+		args.fsbno = XFS_INO_TO_FSB(mp, ip->i_ino);
+	} else if (flist->xbf_low) {
+		args.type = XFS_ALLOCTYPE_START_BNO;
+		args.fsbno = *firstblock;
+	} else {
+		args.type = XFS_ALLOCTYPE_NEAR_BNO;
+		args.fsbno = *firstblock;
+	}
+	args.minlen = args.maxlen = args.prod = 1;
+	args.total = args.minleft = args.alignment = args.mod = args.isfl =
+		args.minalignslop = 0;
+	args.wasdel = wasdel;
+	*logflagsp = 0;
+	if ((error = xfs_alloc_vextent(&args))) {
+		xfs_iroot_realloc(ip, -1, whichfork);
+		xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
+		return error;
+	}
+	/*
+	 * Allocation can't fail, the space was reserved.
+	 */
+	ASSERT(args.fsbno != NULLFSBLOCK);
+	ASSERT(*firstblock == NULLFSBLOCK ||
+	       args.agno == XFS_FSB_TO_AGNO(mp, *firstblock) ||
+	       (flist->xbf_low &&
+		args.agno > XFS_FSB_TO_AGNO(mp, *firstblock)));
+	*firstblock = cur->bc_private.b.firstblock = args.fsbno;
+	cur->bc_private.b.allocated++;
+	ip->i_d.di_nblocks++;
+	xfs_trans_mod_dquot_byino(tp, ip, XFS_TRANS_DQ_BCOUNT, 1L);
+	abp = xfs_btree_get_bufl(mp, tp, args.fsbno, 0);
+	/*
+	 * Fill in the child block.
+	 */
+	ablock = XFS_BUF_TO_BLOCK(abp);
+	ablock->bb_magic = cpu_to_be32(XFS_BMAP_MAGIC);
+	ablock->bb_level = 0;
+	ablock->bb_u.l.bb_leftsib = cpu_to_be64(NULLDFSBNO);
+	ablock->bb_u.l.bb_rightsib = cpu_to_be64(NULLDFSBNO);
+	arp = XFS_BMBT_REC_ADDR(mp, ablock, 1);
+	nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
+	for (cnt = i = 0; i < nextents; i++) {
+		ep = xfs_iext_get_ext(ifp, i);
+		if (!isnullstartblock(xfs_bmbt_get_startblock(ep))) {
+			arp->l0 = cpu_to_be64(ep->l0);
+			arp->l1 = cpu_to_be64(ep->l1);
+			arp++; cnt++;
+		}
+	}
+	ASSERT(cnt == XFS_IFORK_NEXTENTS(ip, whichfork));
+	xfs_btree_set_numrecs(ablock, cnt);
+
+	/*
+	 * Fill in the root key and pointer.
+	 */
+	kp = XFS_BMBT_KEY_ADDR(mp, block, 1);
+	arp = XFS_BMBT_REC_ADDR(mp, ablock, 1);
+	kp->br_startoff = cpu_to_be64(xfs_bmbt_disk_get_startoff(arp));
+	pp = XFS_BMBT_PTR_ADDR(mp, block, 1, xfs_bmbt_get_maxrecs(cur,
+						be16_to_cpu(block->bb_level)));
+	*pp = cpu_to_be64(args.fsbno);
+
+	/*
+	 * Do all this logging at the end so that
+	 * the root is at the right level.
+	 */
+	xfs_btree_log_block(cur, abp, XFS_BB_ALL_BITS);
+	xfs_btree_log_recs(cur, abp, 1, be16_to_cpu(ablock->bb_numrecs));
+	ASSERT(*curp == NULL);
+	*curp = cur;
+	*logflagsp = XFS_ILOG_CORE | xfs_ilog_fbroot(whichfork);
+	return 0;
+}
+
+/*
+ * Calculate the default attribute fork offset for newly created inodes.
+ */
+uint
+xfs_default_attroffset(
+	struct xfs_inode	*ip)
+{
+	struct xfs_mount	*mp = ip->i_mount;
+	uint			offset;
+
+	if (mp->m_sb.sb_inodesize == 256) {
+		offset = XFS_LITINO(mp) -
+				XFS_BMDR_SPACE_CALC(MINABTPTRS);
+	} else {
+		offset = XFS_BMDR_SPACE_CALC(6 * MINABTPTRS);
+	}
+
+	ASSERT(offset < XFS_LITINO(mp));
+	return offset;
+}
+
+/*
+ * Helper routine to reset inode di_forkoff field when switching
+ * attribute fork from local to extent format - we reset it where
+ * possible to make space available for inline data fork extents.
+ */
+STATIC void
+xfs_bmap_forkoff_reset(
+	xfs_mount_t	*mp,
+	xfs_inode_t	*ip,
+	int		whichfork)
+{
+	if (whichfork == XFS_ATTR_FORK &&
+	    ip->i_d.di_format != XFS_DINODE_FMT_DEV &&
+	    ip->i_d.di_format != XFS_DINODE_FMT_UUID &&
+	    ip->i_d.di_format != XFS_DINODE_FMT_BTREE) {
+		uint	dfl_forkoff = xfs_default_attroffset(ip) >> 3;
+
+		if (dfl_forkoff > ip->i_d.di_forkoff) {
+			ip->i_d.di_forkoff = dfl_forkoff;
+			ip->i_df.if_ext_max =
+				XFS_IFORK_DSIZE(ip) / sizeof(xfs_bmbt_rec_t);
+			ip->i_afp->if_ext_max =
+				XFS_IFORK_ASIZE(ip) / sizeof(xfs_bmbt_rec_t);
+		}
+	}
+}
+
+/*
+ * Convert a local file to an extents file.
+ * This code is out of bounds for data forks of regular files,
+ * since the file data needs to get logged so things will stay consistent.
+ * (The bmap-level manipulations are ok, though).
+ */
+STATIC int				/* error */
+xfs_bmap_local_to_extents(
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_inode_t	*ip,		/* incore inode pointer */
+	xfs_fsblock_t	*firstblock,	/* first block allocated in xaction */
+	xfs_extlen_t	total,		/* total blocks needed by transaction */
+	int		*logflagsp,	/* inode logging flags */
+	int		whichfork)	/* data or attr fork */
+{
+	int		error;		/* error return value */
+	int		flags;		/* logging flags returned */
+	xfs_ifork_t	*ifp;		/* inode fork pointer */
+
+	/*
+	 * We don't want to deal with the case of keeping inode data inline yet.
+	 * So sending the data fork of a regular inode is invalid.
+	 */
+	ASSERT(!((ip->i_d.di_mode & S_IFMT) == S_IFREG &&
+		 whichfork == XFS_DATA_FORK));
+	ifp = XFS_IFORK_PTR(ip, whichfork);
+	ASSERT(XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_LOCAL);
+	flags = 0;
+	error = 0;
+	if (ifp->if_bytes) {
+		xfs_alloc_arg_t	args;	/* allocation arguments */
+		xfs_buf_t	*bp;	/* buffer for extent block */
+		xfs_bmbt_rec_host_t *ep;/* extent record pointer */
+
+		args.tp = tp;
+		args.mp = ip->i_mount;
+		args.firstblock = *firstblock;
+		ASSERT((ifp->if_flags &
+			(XFS_IFINLINE|XFS_IFEXTENTS|XFS_IFEXTIREC)) == XFS_IFINLINE);
+		/*
+		 * Allocate a block.  We know we need only one, since the
+		 * file currently fits in an inode.
+		 */
+		if (*firstblock == NULLFSBLOCK) {
+			args.fsbno = XFS_INO_TO_FSB(args.mp, ip->i_ino);
+			args.type = XFS_ALLOCTYPE_START_BNO;
+		} else {
+			args.fsbno = *firstblock;
+			args.type = XFS_ALLOCTYPE_NEAR_BNO;
+		}
+		args.total = total;
+		args.mod = args.minleft = args.alignment = args.wasdel =
+			args.isfl = args.minalignslop = 0;
+		args.minlen = args.maxlen = args.prod = 1;
+		if ((error = xfs_alloc_vextent(&args)))
+			goto done;
+		/*
+		 * Can't fail, the space was reserved.
+		 */
+		ASSERT(args.fsbno != NULLFSBLOCK);
+		ASSERT(args.len == 1);
+		*firstblock = args.fsbno;
+		bp = xfs_btree_get_bufl(args.mp, tp, args.fsbno, 0);
+		memcpy((char *)XFS_BUF_PTR(bp), ifp->if_u1.if_data,
+			ifp->if_bytes);
+		xfs_trans_log_buf(tp, bp, 0, ifp->if_bytes - 1);
+		xfs_bmap_forkoff_reset(args.mp, ip, whichfork);
+		xfs_idata_realloc(ip, -ifp->if_bytes, whichfork);
+		xfs_iext_add(ifp, 0, 1);
+		ep = xfs_iext_get_ext(ifp, 0);
+		xfs_bmbt_set_allf(ep, 0, args.fsbno, 1, XFS_EXT_NORM);
+		trace_xfs_bmap_post_update(ip, 0,
+				whichfork == XFS_ATTR_FORK ? BMAP_ATTRFORK : 0,
+				_THIS_IP_);
+		XFS_IFORK_NEXT_SET(ip, whichfork, 1);
+		ip->i_d.di_nblocks = 1;
+		xfs_trans_mod_dquot_byino(tp, ip,
+			XFS_TRANS_DQ_BCOUNT, 1L);
+		flags |= xfs_ilog_fext(whichfork);
+	} else {
+		ASSERT(XFS_IFORK_NEXTENTS(ip, whichfork) == 0);
+		xfs_bmap_forkoff_reset(ip->i_mount, ip, whichfork);
+	}
+	ifp->if_flags &= ~XFS_IFINLINE;
+	ifp->if_flags |= XFS_IFEXTENTS;
+	XFS_IFORK_FMT_SET(ip, whichfork, XFS_DINODE_FMT_EXTENTS);
+	flags |= XFS_ILOG_CORE;
+done:
+	*logflagsp = flags;
+	return error;
+}
+
+/*
+ * Search the extent records for the entry containing block bno.
+ * If bno lies in a hole, point to the next entry.  If bno lies
+ * past eof, *eofp will be set, and *prevp will contain the last
+ * entry (null if none).  Else, *lastxp will be set to the index
+ * of the found entry; *gotp will contain the entry.
+ */
+STATIC xfs_bmbt_rec_host_t *		/* pointer to found extent entry */
+xfs_bmap_search_multi_extents(
+	xfs_ifork_t	*ifp,		/* inode fork pointer */
+	xfs_fileoff_t	bno,		/* block number searched for */
+	int		*eofp,		/* out: end of file found */
+	xfs_extnum_t	*lastxp,	/* out: last extent index */
+	xfs_bmbt_irec_t	*gotp,		/* out: extent entry found */
+	xfs_bmbt_irec_t	*prevp)		/* out: previous extent entry found */
+{
+	xfs_bmbt_rec_host_t *ep;		/* extent record pointer */
+	xfs_extnum_t	lastx;		/* last extent index */
+
+	/*
+	 * Initialize the extent entry structure to catch access to
+	 * uninitialized br_startblock field.
+	 */
+	gotp->br_startoff = 0xffa5a5a5a5a5a5a5LL;
+	gotp->br_blockcount = 0xa55a5a5a5a5a5a5aLL;
+	gotp->br_state = XFS_EXT_INVALID;
+#if XFS_BIG_BLKNOS
+	gotp->br_startblock = 0xffffa5a5a5a5a5a5LL;
+#else
+	gotp->br_startblock = 0xffffa5a5;
+#endif
+	prevp->br_startoff = NULLFILEOFF;
+
+	ep = xfs_iext_bno_to_ext(ifp, bno, &lastx);
+	if (lastx > 0) {
+		xfs_bmbt_get_all(xfs_iext_get_ext(ifp, lastx - 1), prevp);
+	}
+	if (lastx < (ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t))) {
+		xfs_bmbt_get_all(ep, gotp);
+		*eofp = 0;
+	} else {
+		if (lastx > 0) {
+			*gotp = *prevp;
+		}
+		*eofp = 1;
+		ep = NULL;
+	}
+	*lastxp = lastx;
+	return ep;
+}
+
+/*
+ * Search the extents list for the inode, for the extent containing bno.
+ * If bno lies in a hole, point to the next entry.  If bno lies past eof,
+ * *eofp will be set, and *prevp will contain the last entry (null if none).
+ * Else, *lastxp will be set to the index of the found
+ * entry; *gotp will contain the entry.
+ */
+STATIC xfs_bmbt_rec_host_t *                 /* pointer to found extent entry */
+xfs_bmap_search_extents(
+	xfs_inode_t     *ip,            /* incore inode pointer */
+	xfs_fileoff_t   bno,            /* block number searched for */
+	int             fork,      	/* data or attr fork */
+	int             *eofp,          /* out: end of file found */
+	xfs_extnum_t    *lastxp,        /* out: last extent index */
+	xfs_bmbt_irec_t *gotp,          /* out: extent entry found */
+	xfs_bmbt_irec_t *prevp)         /* out: previous extent entry found */
+{
+	xfs_ifork_t	*ifp;		/* inode fork pointer */
+	xfs_bmbt_rec_host_t  *ep;            /* extent record pointer */
+
+	XFS_STATS_INC(xs_look_exlist);
+	ifp = XFS_IFORK_PTR(ip, fork);
+
+	ep = xfs_bmap_search_multi_extents(ifp, bno, eofp, lastxp, gotp, prevp);
+
+	if (unlikely(!(gotp->br_startblock) && (*lastxp != NULLEXTNUM) &&
+		     !(XFS_IS_REALTIME_INODE(ip) && fork == XFS_DATA_FORK))) {
+		xfs_alert_tag(ip->i_mount, XFS_PTAG_FSBLOCK_ZERO,
+				"Access to block zero in inode %llu "
+				"start_block: %llx start_off: %llx "
+				"blkcnt: %llx extent-state: %x lastx: %x\n",
+			(unsigned long long)ip->i_ino,
+			(unsigned long long)gotp->br_startblock,
+			(unsigned long long)gotp->br_startoff,
+			(unsigned long long)gotp->br_blockcount,
+			gotp->br_state, *lastxp);
+		*lastxp = NULLEXTNUM;
+		*eofp = 1;
+		return NULL;
+	}
+	return ep;
+}
+
+/*
+ * Compute the worst-case number of indirect blocks that will be used
+ * for ip's delayed extent of length "len".
+ */
+STATIC xfs_filblks_t
+xfs_bmap_worst_indlen(
+	xfs_inode_t	*ip,		/* incore inode pointer */
+	xfs_filblks_t	len)		/* delayed extent length */
+{
+	int		level;		/* btree level number */
+	int		maxrecs;	/* maximum record count at this level */
+	xfs_mount_t	*mp;		/* mount structure */
+	xfs_filblks_t	rval;		/* return value */
+
+	mp = ip->i_mount;
+	maxrecs = mp->m_bmap_dmxr[0];
+	for (level = 0, rval = 0;
+	     level < XFS_BM_MAXLEVELS(mp, XFS_DATA_FORK);
+	     level++) {
+		len += maxrecs - 1;
+		do_div(len, maxrecs);
+		rval += len;
+		if (len == 1)
+			return rval + XFS_BM_MAXLEVELS(mp, XFS_DATA_FORK) -
+				level - 1;
+		if (level == 0)
+			maxrecs = mp->m_bmap_dmxr[1];
+	}
+	return rval;
+}
+
+/*
+ * Convert inode from non-attributed to attributed.
+ * Must not be in a transaction, ip must not be locked.
+ */
+int						/* error code */
+xfs_bmap_add_attrfork(
+	xfs_inode_t		*ip,		/* incore inode pointer */
+	int			size,		/* space new attribute needs */
+	int			rsvd)		/* xact may use reserved blks */
+{
+	xfs_fsblock_t		firstblock;	/* 1st block/ag allocated */
+	xfs_bmap_free_t		flist;		/* freed extent records */
+	xfs_mount_t		*mp;		/* mount structure */
+	xfs_trans_t		*tp;		/* transaction pointer */
+	int			blks;		/* space reservation */
+	int			version = 1;	/* superblock attr version */
+	int			committed;	/* xaction was committed */
+	int			logflags;	/* logging flags */
+	int			error;		/* error return value */
+
+	ASSERT(XFS_IFORK_Q(ip) == 0);
+	ASSERT(ip->i_df.if_ext_max ==
+	       XFS_IFORK_DSIZE(ip) / (uint)sizeof(xfs_bmbt_rec_t));
+
+	mp = ip->i_mount;
+	ASSERT(!XFS_NOT_DQATTACHED(mp, ip));
+	tp = xfs_trans_alloc(mp, XFS_TRANS_ADDAFORK);
+	blks = XFS_ADDAFORK_SPACE_RES(mp);
+	if (rsvd)
+		tp->t_flags |= XFS_TRANS_RESERVE;
+	if ((error = xfs_trans_reserve(tp, blks, XFS_ADDAFORK_LOG_RES(mp), 0,
+			XFS_TRANS_PERM_LOG_RES, XFS_ADDAFORK_LOG_COUNT)))
+		goto error0;
+	xfs_ilock(ip, XFS_ILOCK_EXCL);
+	error = xfs_trans_reserve_quota_nblks(tp, ip, blks, 0, rsvd ?
+			XFS_QMOPT_RES_REGBLKS | XFS_QMOPT_FORCE_RES :
+			XFS_QMOPT_RES_REGBLKS);
+	if (error) {
+		xfs_iunlock(ip, XFS_ILOCK_EXCL);
+		xfs_trans_cancel(tp, XFS_TRANS_RELEASE_LOG_RES);
+		return error;
+	}
+	if (XFS_IFORK_Q(ip))
+		goto error1;
+	if (ip->i_d.di_aformat != XFS_DINODE_FMT_EXTENTS) {
+		/*
+		 * For inodes coming from pre-6.2 filesystems.
+		 */
+		ASSERT(ip->i_d.di_aformat == 0);
+		ip->i_d.di_aformat = XFS_DINODE_FMT_EXTENTS;
+	}
+	ASSERT(ip->i_d.di_anextents == 0);
+
+	xfs_trans_ijoin_ref(tp, ip, XFS_ILOCK_EXCL);
+	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
+
+	switch (ip->i_d.di_format) {
+	case XFS_DINODE_FMT_DEV:
+		ip->i_d.di_forkoff = roundup(sizeof(xfs_dev_t), 8) >> 3;
+		break;
+	case XFS_DINODE_FMT_UUID:
+		ip->i_d.di_forkoff = roundup(sizeof(uuid_t), 8) >> 3;
+		break;
+	case XFS_DINODE_FMT_LOCAL:
+	case XFS_DINODE_FMT_EXTENTS:
+	case XFS_DINODE_FMT_BTREE:
+		ip->i_d.di_forkoff = xfs_attr_shortform_bytesfit(ip, size);
+		if (!ip->i_d.di_forkoff)
+			ip->i_d.di_forkoff = xfs_default_attroffset(ip) >> 3;
+		else if (mp->m_flags & XFS_MOUNT_ATTR2)
+			version = 2;
+		break;
+	default:
+		ASSERT(0);
+		error = XFS_ERROR(EINVAL);
+		goto error1;
+	}
+	ip->i_df.if_ext_max =
+		XFS_IFORK_DSIZE(ip) / (uint)sizeof(xfs_bmbt_rec_t);
+	ASSERT(ip->i_afp == NULL);
+	ip->i_afp = kmem_zone_zalloc(xfs_ifork_zone, KM_SLEEP);
+	ip->i_afp->if_ext_max =
+		XFS_IFORK_ASIZE(ip) / (uint)sizeof(xfs_bmbt_rec_t);
+	ip->i_afp->if_flags = XFS_IFEXTENTS;
+	logflags = 0;
+	xfs_bmap_init(&flist, &firstblock);
+	switch (ip->i_d.di_format) {
+	case XFS_DINODE_FMT_LOCAL:
+		error = xfs_bmap_add_attrfork_local(tp, ip, &firstblock, &flist,
+			&logflags);
+		break;
+	case XFS_DINODE_FMT_EXTENTS:
+		error = xfs_bmap_add_attrfork_extents(tp, ip, &firstblock,
+			&flist, &logflags);
+		break;
+	case XFS_DINODE_FMT_BTREE:
+		error = xfs_bmap_add_attrfork_btree(tp, ip, &firstblock, &flist,
+			&logflags);
+		break;
+	default:
+		error = 0;
+		break;
+	}
+	if (logflags)
+		xfs_trans_log_inode(tp, ip, logflags);
+	if (error)
+		goto error2;
+	if (!xfs_sb_version_hasattr(&mp->m_sb) ||
+	   (!xfs_sb_version_hasattr2(&mp->m_sb) && version == 2)) {
+		__int64_t sbfields = 0;
+
+		spin_lock(&mp->m_sb_lock);
+		if (!xfs_sb_version_hasattr(&mp->m_sb)) {
+			xfs_sb_version_addattr(&mp->m_sb);
+			sbfields |= XFS_SB_VERSIONNUM;
+		}
+		if (!xfs_sb_version_hasattr2(&mp->m_sb) && version == 2) {
+			xfs_sb_version_addattr2(&mp->m_sb);
+			sbfields |= (XFS_SB_VERSIONNUM | XFS_SB_FEATURES2);
+		}
+		if (sbfields) {
+			spin_unlock(&mp->m_sb_lock);
+			xfs_mod_sb(tp, sbfields);
+		} else
+			spin_unlock(&mp->m_sb_lock);
+	}
+	if ((error = xfs_bmap_finish(&tp, &flist, &committed)))
+		goto error2;
+	error = xfs_trans_commit(tp, XFS_TRANS_RELEASE_LOG_RES);
+	ASSERT(ip->i_df.if_ext_max ==
+	       XFS_IFORK_DSIZE(ip) / (uint)sizeof(xfs_bmbt_rec_t));
+	return error;
+error2:
+	xfs_bmap_cancel(&flist);
+error1:
+	xfs_iunlock(ip, XFS_ILOCK_EXCL);
+error0:
+	xfs_trans_cancel(tp, XFS_TRANS_RELEASE_LOG_RES|XFS_TRANS_ABORT);
+	ASSERT(ip->i_df.if_ext_max ==
+	       XFS_IFORK_DSIZE(ip) / (uint)sizeof(xfs_bmbt_rec_t));
+	return error;
+}
+
+/*
+ * Add the extent to the list of extents to be free at transaction end.
+ * The list is maintained sorted (by block number).
+ */
+/* ARGSUSED */
+void
+xfs_bmap_add_free(
+	xfs_fsblock_t		bno,		/* fs block number of extent */
+	xfs_filblks_t		len,		/* length of extent */
+	xfs_bmap_free_t		*flist,		/* list of extents */
+	xfs_mount_t		*mp)		/* mount point structure */
+{
+	xfs_bmap_free_item_t	*cur;		/* current (next) element */
+	xfs_bmap_free_item_t	*new;		/* new element */
+	xfs_bmap_free_item_t	*prev;		/* previous element */
+#ifdef DEBUG
+	xfs_agnumber_t		agno;
+	xfs_agblock_t		agbno;
+
+	ASSERT(bno != NULLFSBLOCK);
+	ASSERT(len > 0);
+	ASSERT(len <= MAXEXTLEN);
+	ASSERT(!isnullstartblock(bno));
+	agno = XFS_FSB_TO_AGNO(mp, bno);
+	agbno = XFS_FSB_TO_AGBNO(mp, bno);
+	ASSERT(agno < mp->m_sb.sb_agcount);
+	ASSERT(agbno < mp->m_sb.sb_agblocks);
+	ASSERT(len < mp->m_sb.sb_agblocks);
+	ASSERT(agbno + len <= mp->m_sb.sb_agblocks);
+#endif
+	ASSERT(xfs_bmap_free_item_zone != NULL);
+	new = kmem_zone_alloc(xfs_bmap_free_item_zone, KM_SLEEP);
+	new->xbfi_startblock = bno;
+	new->xbfi_blockcount = (xfs_extlen_t)len;
+	for (prev = NULL, cur = flist->xbf_first;
+	     cur != NULL;
+	     prev = cur, cur = cur->xbfi_next) {
+		if (cur->xbfi_startblock >= bno)
+			break;
+	}
+	if (prev)
+		prev->xbfi_next = new;
+	else
+		flist->xbf_first = new;
+	new->xbfi_next = cur;
+	flist->xbf_count++;
+}
+
+/*
+ * Compute and fill in the value of the maximum depth of a bmap btree
+ * in this filesystem.  Done once, during mount.
+ */
+void
+xfs_bmap_compute_maxlevels(
+	xfs_mount_t	*mp,		/* file system mount structure */
+	int		whichfork)	/* data or attr fork */
+{
+	int		level;		/* btree level */
+	uint		maxblocks;	/* max blocks at this level */
+	uint		maxleafents;	/* max leaf entries possible */
+	int		maxrootrecs;	/* max records in root block */
+	int		minleafrecs;	/* min records in leaf block */
+	int		minnoderecs;	/* min records in node block */
+	int		sz;		/* root block size */
+
+	/*
+	 * The maximum number of extents in a file, hence the maximum
+	 * number of leaf entries, is controlled by the type of di_nextents
+	 * (a signed 32-bit number, xfs_extnum_t), or by di_anextents
+	 * (a signed 16-bit number, xfs_aextnum_t).
+	 *
+	 * Note that we can no longer assume that if we are in ATTR1 that
+	 * the fork offset of all the inodes will be
+	 * (xfs_default_attroffset(ip) >> 3) because we could have mounted
+	 * with ATTR2 and then mounted back with ATTR1, keeping the
+	 * di_forkoff's fixed but probably at various positions. Therefore,
+	 * for both ATTR1 and ATTR2 we have to assume the worst case scenario
+	 * of a minimum size available.
+	 */
+	if (whichfork == XFS_DATA_FORK) {
+		maxleafents = MAXEXTNUM;
+		sz = XFS_BMDR_SPACE_CALC(MINDBTPTRS);
+	} else {
+		maxleafents = MAXAEXTNUM;
+		sz = XFS_BMDR_SPACE_CALC(MINABTPTRS);
+	}
+	maxrootrecs = xfs_bmdr_maxrecs(mp, sz, 0);
+	minleafrecs = mp->m_bmap_dmnr[0];
+	minnoderecs = mp->m_bmap_dmnr[1];
+	maxblocks = (maxleafents + minleafrecs - 1) / minleafrecs;
+	for (level = 1; maxblocks > 1; level++) {
+		if (maxblocks <= maxrootrecs)
+			maxblocks = 1;
+		else
+			maxblocks = (maxblocks + minnoderecs - 1) / minnoderecs;
+	}
+	mp->m_bm_maxlevels[whichfork] = level;
+}
+
+/*
+ * Routine to be called at transaction's end by xfs_bmapi, xfs_bunmapi
+ * caller.  Frees all the extents that need freeing, which must be done
+ * last due to locking considerations.  We never free any extents in
+ * the first transaction.
+ *
+ * Return 1 if the given transaction was committed and a new one
+ * started, and 0 otherwise in the committed parameter.
+ */
+int						/* error */
+xfs_bmap_finish(
+	xfs_trans_t		**tp,		/* transaction pointer addr */
+	xfs_bmap_free_t		*flist,		/* i/o: list extents to free */
+	int			*committed)	/* xact committed or not */
+{
+	xfs_efd_log_item_t	*efd;		/* extent free data */
+	xfs_efi_log_item_t	*efi;		/* extent free intention */
+	int			error;		/* error return value */
+	xfs_bmap_free_item_t	*free;		/* free extent item */
+	unsigned int		logres;		/* new log reservation */
+	unsigned int		logcount;	/* new log count */
+	xfs_mount_t		*mp;		/* filesystem mount structure */
+	xfs_bmap_free_item_t	*next;		/* next item on free list */
+	xfs_trans_t		*ntp;		/* new transaction pointer */
+
+	ASSERT((*tp)->t_flags & XFS_TRANS_PERM_LOG_RES);
+	if (flist->xbf_count == 0) {
+		*committed = 0;
+		return 0;
+	}
+	ntp = *tp;
+	efi = xfs_trans_get_efi(ntp, flist->xbf_count);
+	for (free = flist->xbf_first; free; free = free->xbfi_next)
+		xfs_trans_log_efi_extent(ntp, efi, free->xbfi_startblock,
+			free->xbfi_blockcount);
+	logres = ntp->t_log_res;
+	logcount = ntp->t_log_count;
+	ntp = xfs_trans_dup(*tp);
+	error = xfs_trans_commit(*tp, 0);
+	*tp = ntp;
+	*committed = 1;
+	/*
+	 * We have a new transaction, so we should return committed=1,
+	 * even though we're returning an error.
+	 */
+	if (error)
+		return error;
+
+	/*
+	 * transaction commit worked ok so we can drop the extra ticket
+	 * reference that we gained in xfs_trans_dup()
+	 */
+	xfs_log_ticket_put(ntp->t_ticket);
+
+	if ((error = xfs_trans_reserve(ntp, 0, logres, 0, XFS_TRANS_PERM_LOG_RES,
+			logcount)))
+		return error;
+	efd = xfs_trans_get_efd(ntp, efi, flist->xbf_count);
+	for (free = flist->xbf_first; free != NULL; free = next) {
+		next = free->xbfi_next;
+		if ((error = xfs_free_extent(ntp, free->xbfi_startblock,
+				free->xbfi_blockcount))) {
+			/*
+			 * The bmap free list will be cleaned up at a
+			 * higher level.  The EFI will be canceled when
+			 * this transaction is aborted.
+			 * Need to force shutdown here to make sure it
+			 * happens, since this transaction may not be
+			 * dirty yet.
+			 */
+			mp = ntp->t_mountp;
+			if (!XFS_FORCED_SHUTDOWN(mp))
+				xfs_force_shutdown(mp,
+						   (error == EFSCORRUPTED) ?
+						   SHUTDOWN_CORRUPT_INCORE :
+						   SHUTDOWN_META_IO_ERROR);
+			return error;
+		}
+		xfs_trans_log_efd_extent(ntp, efd, free->xbfi_startblock,
+			free->xbfi_blockcount);
+		xfs_bmap_del_free(flist, NULL, free);
+	}
+	return 0;
+}
+
+/*
+ * Free up any items left in the list.
+ */
+void
+xfs_bmap_cancel(
+	xfs_bmap_free_t		*flist)	/* list of bmap_free_items */
+{
+	xfs_bmap_free_item_t	*free;	/* free list item */
+	xfs_bmap_free_item_t	*next;
+
+	if (flist->xbf_count == 0)
+		return;
+	ASSERT(flist->xbf_first != NULL);
+	for (free = flist->xbf_first; free; free = next) {
+		next = free->xbfi_next;
+		xfs_bmap_del_free(flist, NULL, free);
+	}
+	ASSERT(flist->xbf_count == 0);
+}
+
+/*
+ * Returns the file-relative block number of the first unused block(s)
+ * in the file with at least "len" logically contiguous blocks free.
+ * This is the lowest-address hole if the file has holes, else the first block
+ * past the end of file.
+ * Return 0 if the file is currently local (in-inode).
+ */
+int						/* error */
+xfs_bmap_first_unused(
+	xfs_trans_t	*tp,			/* transaction pointer */
+	xfs_inode_t	*ip,			/* incore inode */
+	xfs_extlen_t	len,			/* size of hole to find */
+	xfs_fileoff_t	*first_unused,		/* unused block */
+	int		whichfork)		/* data or attr fork */
+{
+	int		error;			/* error return value */
+	int		idx;			/* extent record index */
+	xfs_ifork_t	*ifp;			/* inode fork pointer */
+	xfs_fileoff_t	lastaddr;		/* last block number seen */
+	xfs_fileoff_t	lowest;			/* lowest useful block */
+	xfs_fileoff_t	max;			/* starting useful block */
+	xfs_fileoff_t	off;			/* offset for this block */
+	xfs_extnum_t	nextents;		/* number of extent entries */
+
+	ASSERT(XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_BTREE ||
+	       XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_EXTENTS ||
+	       XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_LOCAL);
+	if (XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_LOCAL) {
+		*first_unused = 0;
+		return 0;
+	}
+	ifp = XFS_IFORK_PTR(ip, whichfork);
+	if (!(ifp->if_flags & XFS_IFEXTENTS) &&
+	    (error = xfs_iread_extents(tp, ip, whichfork)))
+		return error;
+	lowest = *first_unused;
+	nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
+	for (idx = 0, lastaddr = 0, max = lowest; idx < nextents; idx++) {
+		xfs_bmbt_rec_host_t *ep = xfs_iext_get_ext(ifp, idx);
+		off = xfs_bmbt_get_startoff(ep);
+		/*
+		 * See if the hole before this extent will work.
+		 */
+		if (off >= lowest + len && off - max >= len) {
+			*first_unused = max;
+			return 0;
+		}
+		lastaddr = off + xfs_bmbt_get_blockcount(ep);
+		max = XFS_FILEOFF_MAX(lastaddr, lowest);
+	}
+	*first_unused = max;
+	return 0;
+}
+
+/*
+ * Returns the file-relative block number of the last block + 1 before
+ * last_block (input value) in the file.
+ * This is not based on i_size, it is based on the extent records.
+ * Returns 0 for local files, as they do not have extent records.
+ */
+int						/* error */
+xfs_bmap_last_before(
+	xfs_trans_t	*tp,			/* transaction pointer */
+	xfs_inode_t	*ip,			/* incore inode */
+	xfs_fileoff_t	*last_block,		/* last block */
+	int		whichfork)		/* data or attr fork */
+{
+	xfs_fileoff_t	bno;			/* input file offset */
+	int		eof;			/* hit end of file */
+	xfs_bmbt_rec_host_t *ep;		/* pointer to last extent */
+	int		error;			/* error return value */
+	xfs_bmbt_irec_t	got;			/* current extent value */
+	xfs_ifork_t	*ifp;			/* inode fork pointer */
+	xfs_extnum_t	lastx;			/* last extent used */
+	xfs_bmbt_irec_t	prev;			/* previous extent value */
+
+	if (XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_BTREE &&
+	    XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_EXTENTS &&
+	    XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_LOCAL)
+	       return XFS_ERROR(EIO);
+	if (XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_LOCAL) {
+		*last_block = 0;
+		return 0;
+	}
+	ifp = XFS_IFORK_PTR(ip, whichfork);
+	if (!(ifp->if_flags & XFS_IFEXTENTS) &&
+	    (error = xfs_iread_extents(tp, ip, whichfork)))
+		return error;
+	bno = *last_block - 1;
+	ep = xfs_bmap_search_extents(ip, bno, whichfork, &eof, &lastx, &got,
+		&prev);
+	if (eof || xfs_bmbt_get_startoff(ep) > bno) {
+		if (prev.br_startoff == NULLFILEOFF)
+			*last_block = 0;
+		else
+			*last_block = prev.br_startoff + prev.br_blockcount;
+	}
+	/*
+	 * Otherwise *last_block is already the right answer.
+	 */
+	return 0;
+}
+
+/*
+ * Returns the file-relative block number of the first block past eof in
+ * the file.  This is not based on i_size, it is based on the extent records.
+ * Returns 0 for local files, as they do not have extent records.
+ */
+int						/* error */
+xfs_bmap_last_offset(
+	xfs_trans_t	*tp,			/* transaction pointer */
+	xfs_inode_t	*ip,			/* incore inode */
+	xfs_fileoff_t	*last_block,		/* last block */
+	int		whichfork)		/* data or attr fork */
+{
+	xfs_bmbt_rec_host_t *ep;		/* pointer to last extent */
+	int		error;			/* error return value */
+	xfs_ifork_t	*ifp;			/* inode fork pointer */
+	xfs_extnum_t	nextents;		/* number of extent entries */
+
+	if (XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_BTREE &&
+	    XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_EXTENTS &&
+	    XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_LOCAL)
+	       return XFS_ERROR(EIO);
+	if (XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_LOCAL) {
+		*last_block = 0;
+		return 0;
+	}
+	ifp = XFS_IFORK_PTR(ip, whichfork);
+	if (!(ifp->if_flags & XFS_IFEXTENTS) &&
+	    (error = xfs_iread_extents(tp, ip, whichfork)))
+		return error;
+	nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
+	if (!nextents) {
+		*last_block = 0;
+		return 0;
+	}
+	ep = xfs_iext_get_ext(ifp, nextents - 1);
+	*last_block = xfs_bmbt_get_startoff(ep) + xfs_bmbt_get_blockcount(ep);
+	return 0;
+}
+
+/*
+ * Returns whether the selected fork of the inode has exactly one
+ * block or not.  For the data fork we check this matches di_size,
+ * implying the file's range is 0..bsize-1.
+ */
+int					/* 1=>1 block, 0=>otherwise */
+xfs_bmap_one_block(
+	xfs_inode_t	*ip,		/* incore inode */
+	int		whichfork)	/* data or attr fork */
+{
+	xfs_bmbt_rec_host_t *ep;	/* ptr to fork's extent */
+	xfs_ifork_t	*ifp;		/* inode fork pointer */
+	int		rval;		/* return value */
+	xfs_bmbt_irec_t	s;		/* internal version of extent */
+
+#ifndef DEBUG
+	if (whichfork == XFS_DATA_FORK) {
+		return ((ip->i_d.di_mode & S_IFMT) == S_IFREG) ?
+			(ip->i_size == ip->i_mount->m_sb.sb_blocksize) :
+			(ip->i_d.di_size == ip->i_mount->m_sb.sb_blocksize);
+	}
+#endif	/* !DEBUG */
+	if (XFS_IFORK_NEXTENTS(ip, whichfork) != 1)
+		return 0;
+	if (XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_EXTENTS)
+		return 0;
+	ifp = XFS_IFORK_PTR(ip, whichfork);
+	ASSERT(ifp->if_flags & XFS_IFEXTENTS);
+	ep = xfs_iext_get_ext(ifp, 0);
+	xfs_bmbt_get_all(ep, &s);
+	rval = s.br_startoff == 0 && s.br_blockcount == 1;
+	if (rval && whichfork == XFS_DATA_FORK)
+		ASSERT(ip->i_size == ip->i_mount->m_sb.sb_blocksize);
+	return rval;
+}
+
+STATIC int
+xfs_bmap_sanity_check(
+	struct xfs_mount	*mp,
+	struct xfs_buf		*bp,
+	int			level)
+{
+	struct xfs_btree_block  *block = XFS_BUF_TO_BLOCK(bp);
+
+	if (be32_to_cpu(block->bb_magic) != XFS_BMAP_MAGIC ||
+	    be16_to_cpu(block->bb_level) != level ||
+	    be16_to_cpu(block->bb_numrecs) == 0 ||
+	    be16_to_cpu(block->bb_numrecs) > mp->m_bmap_dmxr[level != 0])
+		return 0;
+	return 1;
+}
+
+/*
+ * Read in the extents to if_extents.
+ * All inode fields are set up by caller, we just traverse the btree
+ * and copy the records in. If the file system cannot contain unwritten
+ * extents, the records are checked for no "state" flags.
+ */
+int					/* error */
+xfs_bmap_read_extents(
+	xfs_trans_t		*tp,	/* transaction pointer */
+	xfs_inode_t		*ip,	/* incore inode */
+	int			whichfork) /* data or attr fork */
+{
+	struct xfs_btree_block	*block;	/* current btree block */
+	xfs_fsblock_t		bno;	/* block # of "block" */
+	xfs_buf_t		*bp;	/* buffer for "block" */
+	int			error;	/* error return value */
+	xfs_exntfmt_t		exntf;	/* XFS_EXTFMT_NOSTATE, if checking */
+	xfs_extnum_t		i, j;	/* index into the extents list */
+	xfs_ifork_t		*ifp;	/* fork structure */
+	int			level;	/* btree level, for checking */
+	xfs_mount_t		*mp;	/* file system mount structure */
+	__be64			*pp;	/* pointer to block address */
+	/* REFERENCED */
+	xfs_extnum_t		room;	/* number of entries there's room for */
+
+	bno = NULLFSBLOCK;
+	mp = ip->i_mount;
+	ifp = XFS_IFORK_PTR(ip, whichfork);
+	exntf = (whichfork != XFS_DATA_FORK) ? XFS_EXTFMT_NOSTATE :
+					XFS_EXTFMT_INODE(ip);
+	block = ifp->if_broot;
+	/*
+	 * Root level must use BMAP_BROOT_PTR_ADDR macro to get ptr out.
+	 */
+	level = be16_to_cpu(block->bb_level);
+	ASSERT(level > 0);
+	pp = XFS_BMAP_BROOT_PTR_ADDR(mp, block, 1, ifp->if_broot_bytes);
+	bno = be64_to_cpu(*pp);
+	ASSERT(bno != NULLDFSBNO);
+	ASSERT(XFS_FSB_TO_AGNO(mp, bno) < mp->m_sb.sb_agcount);
+	ASSERT(XFS_FSB_TO_AGBNO(mp, bno) < mp->m_sb.sb_agblocks);
+	/*
+	 * Go down the tree until leaf level is reached, following the first
+	 * pointer (leftmost) at each level.
+	 */
+	while (level-- > 0) {
+		if ((error = xfs_btree_read_bufl(mp, tp, bno, 0, &bp,
+				XFS_BMAP_BTREE_REF)))
+			return error;
+		block = XFS_BUF_TO_BLOCK(bp);
+		XFS_WANT_CORRUPTED_GOTO(
+			xfs_bmap_sanity_check(mp, bp, level),
+			error0);
+		if (level == 0)
+			break;
+		pp = XFS_BMBT_PTR_ADDR(mp, block, 1, mp->m_bmap_dmxr[1]);
+		bno = be64_to_cpu(*pp);
+		XFS_WANT_CORRUPTED_GOTO(XFS_FSB_SANITY_CHECK(mp, bno), error0);
+		xfs_trans_brelse(tp, bp);
+	}
+	/*
+	 * Here with bp and block set to the leftmost leaf node in the tree.
+	 */
+	room = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
+	i = 0;
+	/*
+	 * Loop over all leaf nodes.  Copy information to the extent records.
+	 */
+	for (;;) {
+		xfs_bmbt_rec_t	*frp;
+		xfs_fsblock_t	nextbno;
+		xfs_extnum_t	num_recs;
+		xfs_extnum_t	start;
+
+
+		num_recs = xfs_btree_get_numrecs(block);
+		if (unlikely(i + num_recs > room)) {
+			ASSERT(i + num_recs <= room);
+			xfs_warn(ip->i_mount,
+				"corrupt dinode %Lu, (btree extents).",
+				(unsigned long long) ip->i_ino);
+			XFS_CORRUPTION_ERROR("xfs_bmap_read_extents(1)",
+				XFS_ERRLEVEL_LOW, ip->i_mount, block);
+			goto error0;
+		}
+		XFS_WANT_CORRUPTED_GOTO(
+			xfs_bmap_sanity_check(mp, bp, 0),
+			error0);
+		/*
+		 * Read-ahead the next leaf block, if any.
+		 */
+		nextbno = be64_to_cpu(block->bb_u.l.bb_rightsib);
+		if (nextbno != NULLFSBLOCK)
+			xfs_btree_reada_bufl(mp, nextbno, 1);
+		/*
+		 * Copy records into the extent records.
+		 */
+		frp = XFS_BMBT_REC_ADDR(mp, block, 1);
+		start = i;
+		for (j = 0; j < num_recs; j++, i++, frp++) {
+			xfs_bmbt_rec_host_t *trp = xfs_iext_get_ext(ifp, i);
+			trp->l0 = be64_to_cpu(frp->l0);
+			trp->l1 = be64_to_cpu(frp->l1);
+		}
+		if (exntf == XFS_EXTFMT_NOSTATE) {
+			/*
+			 * Check all attribute bmap btree records and
+			 * any "older" data bmap btree records for a
+			 * set bit in the "extent flag" position.
+			 */
+			if (unlikely(xfs_check_nostate_extents(ifp,
+					start, num_recs))) {
+				XFS_ERROR_REPORT("xfs_bmap_read_extents(2)",
+						 XFS_ERRLEVEL_LOW,
+						 ip->i_mount);
+				goto error0;
+			}
+		}
+		xfs_trans_brelse(tp, bp);
+		bno = nextbno;
+		/*
+		 * If we've reached the end, stop.
+		 */
+		if (bno == NULLFSBLOCK)
+			break;
+		if ((error = xfs_btree_read_bufl(mp, tp, bno, 0, &bp,
+				XFS_BMAP_BTREE_REF)))
+			return error;
+		block = XFS_BUF_TO_BLOCK(bp);
+	}
+	ASSERT(i == (ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t)));
+	ASSERT(i == XFS_IFORK_NEXTENTS(ip, whichfork));
+	XFS_BMAP_TRACE_EXLIST(ip, i, whichfork);
+	return 0;
+error0:
+	xfs_trans_brelse(tp, bp);
+	return XFS_ERROR(EFSCORRUPTED);
+}
+
+#ifdef DEBUG
+/*
+ * Add bmap trace insert entries for all the contents of the extent records.
+ */
+void
+xfs_bmap_trace_exlist(
+	xfs_inode_t	*ip,		/* incore inode pointer */
+	xfs_extnum_t	cnt,		/* count of entries in the list */
+	int		whichfork,	/* data or attr fork */
+	unsigned long	caller_ip)
+{
+	xfs_extnum_t	idx;		/* extent record index */
+	xfs_ifork_t	*ifp;		/* inode fork pointer */
+	int		state = 0;
+
+	if (whichfork == XFS_ATTR_FORK)
+		state |= BMAP_ATTRFORK;
+
+	ifp = XFS_IFORK_PTR(ip, whichfork);
+	ASSERT(cnt == (ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t)));
+	for (idx = 0; idx < cnt; idx++)
+		trace_xfs_extlist(ip, idx, whichfork, caller_ip);
+}
+
+/*
+ * Validate that the bmbt_irecs being returned from bmapi are valid
+ * given the callers original parameters.  Specifically check the
+ * ranges of the returned irecs to ensure that they only extent beyond
+ * the given parameters if the XFS_BMAPI_ENTIRE flag was set.
+ */
+STATIC void
+xfs_bmap_validate_ret(
+	xfs_fileoff_t		bno,
+	xfs_filblks_t		len,
+	int			flags,
+	xfs_bmbt_irec_t		*mval,
+	int			nmap,
+	int			ret_nmap)
+{
+	int			i;		/* index to map values */
+
+	ASSERT(ret_nmap <= nmap);
+
+	for (i = 0; i < ret_nmap; i++) {
+		ASSERT(mval[i].br_blockcount > 0);
+		if (!(flags & XFS_BMAPI_ENTIRE)) {
+			ASSERT(mval[i].br_startoff >= bno);
+			ASSERT(mval[i].br_blockcount <= len);
+			ASSERT(mval[i].br_startoff + mval[i].br_blockcount <=
+			       bno + len);
+		} else {
+			ASSERT(mval[i].br_startoff < bno + len);
+			ASSERT(mval[i].br_startoff + mval[i].br_blockcount >
+			       bno);
+		}
+		ASSERT(i == 0 ||
+		       mval[i - 1].br_startoff + mval[i - 1].br_blockcount ==
+		       mval[i].br_startoff);
+		if ((flags & XFS_BMAPI_WRITE) && !(flags & XFS_BMAPI_DELAY))
+			ASSERT(mval[i].br_startblock != DELAYSTARTBLOCK &&
+			       mval[i].br_startblock != HOLESTARTBLOCK);
+		ASSERT(mval[i].br_state == XFS_EXT_NORM ||
+		       mval[i].br_state == XFS_EXT_UNWRITTEN);
+	}
+}
+#endif /* DEBUG */
+
+
+/*
+ * Map file blocks to filesystem blocks.
+ * File range is given by the bno/len pair.
+ * Adds blocks to file if a write ("flags & XFS_BMAPI_WRITE" set)
+ * into a hole or past eof.
+ * Only allocates blocks from a single allocation group,
+ * to avoid locking problems.
+ * The returned value in "firstblock" from the first call in a transaction
+ * must be remembered and presented to subsequent calls in "firstblock".
+ * An upper bound for the number of blocks to be allocated is supplied to
+ * the first call in "total"; if no allocation group has that many free
+ * blocks then the call will fail (return NULLFSBLOCK in "firstblock").
+ */
+int					/* error */
+xfs_bmapi(
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_inode_t	*ip,		/* incore inode */
+	xfs_fileoff_t	bno,		/* starting file offs. mapped */
+	xfs_filblks_t	len,		/* length to map in file */
+	int		flags,		/* XFS_BMAPI_... */
+	xfs_fsblock_t	*firstblock,	/* first allocated block
+					   controls a.g. for allocs */
+	xfs_extlen_t	total,		/* total blocks needed */
+	xfs_bmbt_irec_t	*mval,		/* output: map values */
+	int		*nmap,		/* i/o: mval size/count */
+	xfs_bmap_free_t	*flist)		/* i/o: list extents to free */
+{
+	xfs_fsblock_t	abno;		/* allocated block number */
+	xfs_extlen_t	alen;		/* allocated extent length */
+	xfs_fileoff_t	aoff;		/* allocated file offset */
+	xfs_bmalloca_t	bma = { 0 };	/* args for xfs_bmap_alloc */
+	xfs_btree_cur_t	*cur;		/* bmap btree cursor */
+	xfs_fileoff_t	end;		/* end of mapped file region */
+	int		eof;		/* we've hit the end of extents */
+	xfs_bmbt_rec_host_t *ep;	/* extent record pointer */
+	int		error;		/* error return */
+	xfs_bmbt_irec_t	got;		/* current file extent record */
+	xfs_ifork_t	*ifp;		/* inode fork pointer */
+	xfs_extlen_t	indlen;		/* indirect blocks length */
+	xfs_extnum_t	lastx;		/* last useful extent number */
+	int		logflags;	/* flags for transaction logging */
+	xfs_extlen_t	minleft;	/* min blocks left after allocation */
+	xfs_extlen_t	minlen;		/* min allocation size */
+	xfs_mount_t	*mp;		/* xfs mount structure */
+	int		n;		/* current extent index */
+	int		nallocs;	/* number of extents alloc'd */
+	xfs_extnum_t	nextents;	/* number of extents in file */
+	xfs_fileoff_t	obno;		/* old block number (offset) */
+	xfs_bmbt_irec_t	prev;		/* previous file extent record */
+	int		tmp_logflags;	/* temp flags holder */
+	int		whichfork;	/* data or attr fork */
+	char		inhole;		/* current location is hole in file */
+	char		wasdelay;	/* old extent was delayed */
+	char		wr;		/* this is a write request */
+	char		rt;		/* this is a realtime file */
+#ifdef DEBUG
+	xfs_fileoff_t	orig_bno;	/* original block number value */
+	int		orig_flags;	/* original flags arg value */
+	xfs_filblks_t	orig_len;	/* original value of len arg */
+	xfs_bmbt_irec_t	*orig_mval;	/* original value of mval */
+	int		orig_nmap;	/* original value of *nmap */
+
+	orig_bno = bno;
+	orig_len = len;
+	orig_flags = flags;
+	orig_mval = mval;
+	orig_nmap = *nmap;
+#endif
+	ASSERT(*nmap >= 1);
+	ASSERT(*nmap <= XFS_BMAP_MAX_NMAP || !(flags & XFS_BMAPI_WRITE));
+	whichfork = (flags & XFS_BMAPI_ATTRFORK) ?
+		XFS_ATTR_FORK : XFS_DATA_FORK;
+	mp = ip->i_mount;
+	if (unlikely(XFS_TEST_ERROR(
+	    (XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_EXTENTS &&
+	     XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_BTREE &&
+	     XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_LOCAL),
+	     mp, XFS_ERRTAG_BMAPIFORMAT, XFS_RANDOM_BMAPIFORMAT))) {
+		XFS_ERROR_REPORT("xfs_bmapi", XFS_ERRLEVEL_LOW, mp);
+		return XFS_ERROR(EFSCORRUPTED);
+	}
+	if (XFS_FORCED_SHUTDOWN(mp))
+		return XFS_ERROR(EIO);
+	rt = (whichfork == XFS_DATA_FORK) && XFS_IS_REALTIME_INODE(ip);
+	ifp = XFS_IFORK_PTR(ip, whichfork);
+	ASSERT(ifp->if_ext_max ==
+	       XFS_IFORK_SIZE(ip, whichfork) / (uint)sizeof(xfs_bmbt_rec_t));
+	if ((wr = (flags & XFS_BMAPI_WRITE)) != 0)
+		XFS_STATS_INC(xs_blk_mapw);
+	else
+		XFS_STATS_INC(xs_blk_mapr);
+	/*
+	 * IGSTATE flag is used to combine extents which
+	 * differ only due to the state of the extents.
+	 * This technique is used from xfs_getbmap()
+	 * when the caller does not wish to see the
+	 * separation (which is the default).
+	 *
+	 * This technique is also used when writing a
+	 * buffer which has been partially written,
+	 * (usually by being flushed during a chunkread),
+	 * to ensure one write takes place. This also
+	 * prevents a change in the xfs inode extents at
+	 * this time, intentionally. This change occurs
+	 * on completion of the write operation, in
+	 * xfs_strat_comp(), where the xfs_bmapi() call
+	 * is transactioned, and the extents combined.
+	 */
+	if ((flags & XFS_BMAPI_IGSTATE) && wr)	/* if writing unwritten space */
+		wr = 0;				/* no allocations are allowed */
+	ASSERT(wr || !(flags & XFS_BMAPI_DELAY));
+	logflags = 0;
+	nallocs = 0;
+	cur = NULL;
+	if (XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_LOCAL) {
+		ASSERT(wr && tp);
+		if ((error = xfs_bmap_local_to_extents(tp, ip,
+				firstblock, total, &logflags, whichfork)))
+			goto error0;
+	}
+	if (wr && *firstblock == NULLFSBLOCK) {
+		if (XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_BTREE)
+			minleft = be16_to_cpu(ifp->if_broot->bb_level) + 1;
+		else
+			minleft = 1;
+	} else
+		minleft = 0;
+	if (!(ifp->if_flags & XFS_IFEXTENTS) &&
+	    (error = xfs_iread_extents(tp, ip, whichfork)))
+		goto error0;
+	ep = xfs_bmap_search_extents(ip, bno, whichfork, &eof, &lastx, &got,
+		&prev);
+	nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
+	n = 0;
+	end = bno + len;
+	obno = bno;
+	bma.ip = NULL;
+
+	while (bno < end && n < *nmap) {
+		/*
+		 * Reading past eof, act as though there's a hole
+		 * up to end.
+		 */
+		if (eof && !wr)
+			got.br_startoff = end;
+		inhole = eof || got.br_startoff > bno;
+		wasdelay = wr && !inhole && !(flags & XFS_BMAPI_DELAY) &&
+			isnullstartblock(got.br_startblock);
+		/*
+		 * First, deal with the hole before the allocated space
+		 * that we found, if any.
+		 */
+		if (wr && (inhole || wasdelay)) {
+			/*
+			 * For the wasdelay case, we could also just
+			 * allocate the stuff asked for in this bmap call
+			 * but that wouldn't be as good.
+			 */
+			if (wasdelay) {
+				alen = (xfs_extlen_t)got.br_blockcount;
+				aoff = got.br_startoff;
+				if (lastx != NULLEXTNUM && lastx) {
+					ep = xfs_iext_get_ext(ifp, lastx - 1);
+					xfs_bmbt_get_all(ep, &prev);
+				}
+			} else {
+				alen = (xfs_extlen_t)
+					XFS_FILBLKS_MIN(len, MAXEXTLEN);
+				if (!eof)
+					alen = (xfs_extlen_t)
+						XFS_FILBLKS_MIN(alen,
+							got.br_startoff - bno);
+				aoff = bno;
+			}
+			minlen = (flags & XFS_BMAPI_CONTIG) ? alen : 1;
+			if (flags & XFS_BMAPI_DELAY) {
+				xfs_extlen_t	extsz;
+
+				/* Figure out the extent size, adjust alen */
+				extsz = xfs_get_extsz_hint(ip);
+				if (extsz) {
+					/*
+					 * make sure we don't exceed a single
+					 * extent length when we align the
+					 * extent by reducing length we are
+					 * going to allocate by the maximum
+					 * amount extent size aligment may
+					 * require.
+					 */
+					alen = XFS_FILBLKS_MIN(len,
+						   MAXEXTLEN - (2 * extsz - 1));
+					error = xfs_bmap_extsize_align(mp,
+							&got, &prev, extsz,
+							rt, eof,
+							flags&XFS_BMAPI_DELAY,
+							flags&XFS_BMAPI_CONVERT,
+							&aoff, &alen);
+					ASSERT(!error);
+				}
+
+				if (rt)
+					extsz = alen / mp->m_sb.sb_rextsize;
+
+				/*
+				 * Make a transaction-less quota reservation for
+				 * delayed allocation blocks. This number gets
+				 * adjusted later.  We return if we haven't
+				 * allocated blocks already inside this loop.
+				 */
+				error = xfs_trans_reserve_quota_nblks(
+						NULL, ip, (long)alen, 0,
+						rt ? XFS_QMOPT_RES_RTBLKS :
+						     XFS_QMOPT_RES_REGBLKS);
+				if (error) {
+					if (n == 0) {
+						*nmap = 0;
+						ASSERT(cur == NULL);
+						return error;
+					}
+					break;
+				}
+
+				/*
+				 * Split changing sb for alen and indlen since
+				 * they could be coming from different places.
+				 */
+				indlen = (xfs_extlen_t)
+					xfs_bmap_worst_indlen(ip, alen);
+				ASSERT(indlen > 0);
+
+				if (rt) {
+					error = xfs_mod_incore_sb(mp,
+							XFS_SBS_FREXTENTS,
+							-((int64_t)extsz), 0);
+				} else {
+					error = xfs_icsb_modify_counters(mp,
+							XFS_SBS_FDBLOCKS,
+							-((int64_t)alen), 0);
+				}
+				if (!error) {
+					error = xfs_icsb_modify_counters(mp,
+							XFS_SBS_FDBLOCKS,
+							-((int64_t)indlen), 0);
+					if (error && rt)
+						xfs_mod_incore_sb(mp,
+							XFS_SBS_FREXTENTS,
+							(int64_t)extsz, 0);
+					else if (error)
+						xfs_icsb_modify_counters(mp,
+							XFS_SBS_FDBLOCKS,
+							(int64_t)alen, 0);
+				}
+
+				if (error) {
+					if (XFS_IS_QUOTA_ON(mp))
+						/* unreserve the blocks now */
+						(void)
+						xfs_trans_unreserve_quota_nblks(
+							NULL, ip,
+							(long)alen, 0, rt ?
+							XFS_QMOPT_RES_RTBLKS :
+							XFS_QMOPT_RES_REGBLKS);
+					break;
+				}
+
+				ip->i_delayed_blks += alen;
+				abno = nullstartblock(indlen);
+			} else {
+				/*
+				 * If first time, allocate and fill in
+				 * once-only bma fields.
+				 */
+				if (bma.ip == NULL) {
+					bma.tp = tp;
+					bma.ip = ip;
+					bma.prevp = &prev;
+					bma.gotp = &got;
+					bma.total = total;
+					bma.userdata = 0;
+				}
+				/* Indicate if this is the first user data
+				 * in the file, or just any user data.
+				 */
+				if (!(flags & XFS_BMAPI_METADATA)) {
+					bma.userdata = (aoff == 0) ?
+						XFS_ALLOC_INITIAL_USER_DATA :
+						XFS_ALLOC_USERDATA;
+				}
+				/*
+				 * Fill in changeable bma fields.
+				 */
+				bma.eof = eof;
+				bma.firstblock = *firstblock;
+				bma.alen = alen;
+				bma.off = aoff;
+				bma.conv = !!(flags & XFS_BMAPI_CONVERT);
+				bma.wasdel = wasdelay;
+				bma.minlen = minlen;
+				bma.low = flist->xbf_low;
+				bma.minleft = minleft;
+				/*
+				 * Only want to do the alignment at the
+				 * eof if it is userdata and allocation length
+				 * is larger than a stripe unit.
+				 */
+				if (mp->m_dalign && alen >= mp->m_dalign &&
+				    (!(flags & XFS_BMAPI_METADATA)) &&
+				    (whichfork == XFS_DATA_FORK)) {
+					if ((error = xfs_bmap_isaeof(ip, aoff,
+							whichfork, &bma.aeof)))
+						goto error0;
+				} else
+					bma.aeof = 0;
+				/*
+				 * Call allocator.
+				 */
+				if ((error = xfs_bmap_alloc(&bma)))
+					goto error0;
+				/*
+				 * Copy out result fields.
+				 */
+				abno = bma.rval;
+				if ((flist->xbf_low = bma.low))
+					minleft = 0;
+				alen = bma.alen;
+				aoff = bma.off;
+				ASSERT(*firstblock == NULLFSBLOCK ||
+				       XFS_FSB_TO_AGNO(mp, *firstblock) ==
+				       XFS_FSB_TO_AGNO(mp, bma.firstblock) ||
+				       (flist->xbf_low &&
+					XFS_FSB_TO_AGNO(mp, *firstblock) <
+					XFS_FSB_TO_AGNO(mp, bma.firstblock)));
+				*firstblock = bma.firstblock;
+				if (cur)
+					cur->bc_private.b.firstblock =
+						*firstblock;
+				if (abno == NULLFSBLOCK)
+					break;
+				if ((ifp->if_flags & XFS_IFBROOT) && !cur) {
+					cur = xfs_bmbt_init_cursor(mp, tp,
+						ip, whichfork);
+					cur->bc_private.b.firstblock =
+						*firstblock;
+					cur->bc_private.b.flist = flist;
+				}
+				/*
+				 * Bump the number of extents we've allocated
+				 * in this call.
+				 */
+				nallocs++;
+			}
+			if (cur)
+				cur->bc_private.b.flags =
+					wasdelay ? XFS_BTCUR_BPRV_WASDEL : 0;
+			got.br_startoff = aoff;
+			got.br_startblock = abno;
+			got.br_blockcount = alen;
+			got.br_state = XFS_EXT_NORM;	/* assume normal */
+			/*
+			 * Determine state of extent, and the filesystem.
+			 * A wasdelay extent has been initialized, so
+			 * shouldn't be flagged as unwritten.
+			 */
+			if (wr && xfs_sb_version_hasextflgbit(&mp->m_sb)) {
+				if (!wasdelay && (flags & XFS_BMAPI_PREALLOC))
+					got.br_state = XFS_EXT_UNWRITTEN;
+			}
+			error = xfs_bmap_add_extent(ip, &lastx, &cur, &got,
+				firstblock, flist, &tmp_logflags,
+				whichfork);
+			logflags |= tmp_logflags;
+			if (error)
+				goto error0;
+			ep = xfs_iext_get_ext(ifp, lastx);
+			nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
+			xfs_bmbt_get_all(ep, &got);
+			ASSERT(got.br_startoff <= aoff);
+			ASSERT(got.br_startoff + got.br_blockcount >=
+				aoff + alen);
+#ifdef DEBUG
+			if (flags & XFS_BMAPI_DELAY) {
+				ASSERT(isnullstartblock(got.br_startblock));
+				ASSERT(startblockval(got.br_startblock) > 0);
+			}
+			ASSERT(got.br_state == XFS_EXT_NORM ||
+			       got.br_state == XFS_EXT_UNWRITTEN);
+#endif
+			/*
+			 * Fall down into the found allocated space case.
+			 */
+		} else if (inhole) {
+			/*
+			 * Reading in a hole.
+			 */
+			mval->br_startoff = bno;
+			mval->br_startblock = HOLESTARTBLOCK;
+			mval->br_blockcount =
+				XFS_FILBLKS_MIN(len, got.br_startoff - bno);
+			mval->br_state = XFS_EXT_NORM;
+			bno += mval->br_blockcount;
+			len -= mval->br_blockcount;
+			mval++;
+			n++;
+			continue;
+		}
+		/*
+		 * Then deal with the allocated space we found.
+		 */
+		ASSERT(ep != NULL);
+		if (!(flags & XFS_BMAPI_ENTIRE) &&
+		    (got.br_startoff + got.br_blockcount > obno)) {
+			if (obno > bno)
+				bno = obno;
+			ASSERT((bno >= obno) || (n == 0));
+			ASSERT(bno < end);
+			mval->br_startoff = bno;
+			if (isnullstartblock(got.br_startblock)) {
+				ASSERT(!wr || (flags & XFS_BMAPI_DELAY));
+				mval->br_startblock = DELAYSTARTBLOCK;
+			} else
+				mval->br_startblock =
+					got.br_startblock +
+					(bno - got.br_startoff);
+			/*
+			 * Return the minimum of what we got and what we
+			 * asked for for the length.  We can use the len
+			 * variable here because it is modified below
+			 * and we could have been there before coming
+			 * here if the first part of the allocation
+			 * didn't overlap what was asked for.
+			 */
+			mval->br_blockcount =
+				XFS_FILBLKS_MIN(end - bno, got.br_blockcount -
+					(bno - got.br_startoff));
+			mval->br_state = got.br_state;
+			ASSERT(mval->br_blockcount <= len);
+		} else {
+			*mval = got;
+			if (isnullstartblock(mval->br_startblock)) {
+				ASSERT(!wr || (flags & XFS_BMAPI_DELAY));
+				mval->br_startblock = DELAYSTARTBLOCK;
+			}
+		}
+
+		/*
+		 * Check if writing previously allocated but
+		 * unwritten extents.
+		 */
+		if (wr &&
+		    ((mval->br_state == XFS_EXT_UNWRITTEN &&
+		      ((flags & (XFS_BMAPI_PREALLOC|XFS_BMAPI_DELAY)) == 0)) ||
+		     (mval->br_state == XFS_EXT_NORM &&
+		      ((flags & (XFS_BMAPI_PREALLOC|XFS_BMAPI_CONVERT)) ==
+				(XFS_BMAPI_PREALLOC|XFS_BMAPI_CONVERT))))) {
+			/*
+			 * Modify (by adding) the state flag, if writing.
+			 */
+			ASSERT(mval->br_blockcount <= len);
+			if ((ifp->if_flags & XFS_IFBROOT) && !cur) {
+				cur = xfs_bmbt_init_cursor(mp,
+					tp, ip, whichfork);
+				cur->bc_private.b.firstblock =
+					*firstblock;
+				cur->bc_private.b.flist = flist;
+			}
+			mval->br_state = (mval->br_state == XFS_EXT_UNWRITTEN)
+						? XFS_EXT_NORM
+						: XFS_EXT_UNWRITTEN;
+			error = xfs_bmap_add_extent(ip, &lastx, &cur, mval,
+				firstblock, flist, &tmp_logflags,
+				whichfork);
+			logflags |= tmp_logflags;
+			if (error)
+				goto error0;
+			ep = xfs_iext_get_ext(ifp, lastx);
+			nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
+			xfs_bmbt_get_all(ep, &got);
+			/*
+			 * We may have combined previously unwritten
+			 * space with written space, so generate
+			 * another request.
+			 */
+			if (mval->br_blockcount < len)
+				continue;
+		}
+
+		ASSERT((flags & XFS_BMAPI_ENTIRE) ||
+		       ((mval->br_startoff + mval->br_blockcount) <= end));
+		ASSERT((flags & XFS_BMAPI_ENTIRE) ||
+		       (mval->br_blockcount <= len) ||
+		       (mval->br_startoff < obno));
+		bno = mval->br_startoff + mval->br_blockcount;
+		len = end - bno;
+		if (n > 0 && mval->br_startoff == mval[-1].br_startoff) {
+			ASSERT(mval->br_startblock == mval[-1].br_startblock);
+			ASSERT(mval->br_blockcount > mval[-1].br_blockcount);
+			ASSERT(mval->br_state == mval[-1].br_state);
+			mval[-1].br_blockcount = mval->br_blockcount;
+			mval[-1].br_state = mval->br_state;
+		} else if (n > 0 && mval->br_startblock != DELAYSTARTBLOCK &&
+			   mval[-1].br_startblock != DELAYSTARTBLOCK &&
+			   mval[-1].br_startblock != HOLESTARTBLOCK &&
+			   mval->br_startblock ==
+			   mval[-1].br_startblock + mval[-1].br_blockcount &&
+			   ((flags & XFS_BMAPI_IGSTATE) ||
+				mval[-1].br_state == mval->br_state)) {
+			ASSERT(mval->br_startoff ==
+			       mval[-1].br_startoff + mval[-1].br_blockcount);
+			mval[-1].br_blockcount += mval->br_blockcount;
+		} else if (n > 0 &&
+			   mval->br_startblock == DELAYSTARTBLOCK &&
+			   mval[-1].br_startblock == DELAYSTARTBLOCK &&
+			   mval->br_startoff ==
+			   mval[-1].br_startoff + mval[-1].br_blockcount) {
+			mval[-1].br_blockcount += mval->br_blockcount;
+			mval[-1].br_state = mval->br_state;
+		} else if (!((n == 0) &&
+			     ((mval->br_startoff + mval->br_blockcount) <=
+			      obno))) {
+			mval++;
+			n++;
+		}
+		/*
+		 * If we're done, stop now.  Stop when we've allocated
+		 * XFS_BMAP_MAX_NMAP extents no matter what.  Otherwise
+		 * the transaction may get too big.
+		 */
+		if (bno >= end || n >= *nmap || nallocs >= *nmap)
+			break;
+		/*
+		 * Else go on to the next record.
+		 */
+		prev = got;
+		if (++lastx < nextents) {
+			ep = xfs_iext_get_ext(ifp, lastx);
+			xfs_bmbt_get_all(ep, &got);
+		} else {
+			eof = 1;
+		}
+	}
+	*nmap = n;
+	/*
+	 * Transform from btree to extents, give it cur.
+	 */
+	if (tp && XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_BTREE &&
+	    XFS_IFORK_NEXTENTS(ip, whichfork) <= ifp->if_ext_max) {
+		ASSERT(wr && cur);
+		error = xfs_bmap_btree_to_extents(tp, ip, cur,
+			&tmp_logflags, whichfork);
+		logflags |= tmp_logflags;
+		if (error)
+			goto error0;
+	}
+	ASSERT(ifp->if_ext_max ==
+	       XFS_IFORK_SIZE(ip, whichfork) / (uint)sizeof(xfs_bmbt_rec_t));
+	ASSERT(XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_BTREE ||
+	       XFS_IFORK_NEXTENTS(ip, whichfork) > ifp->if_ext_max);
+	error = 0;
+error0:
+	/*
+	 * Log everything.  Do this after conversion, there's no point in
+	 * logging the extent records if we've converted to btree format.
+	 */
+	if ((logflags & xfs_ilog_fext(whichfork)) &&
+	    XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_EXTENTS)
+		logflags &= ~xfs_ilog_fext(whichfork);
+	else if ((logflags & xfs_ilog_fbroot(whichfork)) &&
+		 XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_BTREE)
+		logflags &= ~xfs_ilog_fbroot(whichfork);
+	/*
+	 * Log whatever the flags say, even if error.  Otherwise we might miss
+	 * detecting a case where the data is changed, there's an error,
+	 * and it's not logged so we don't shutdown when we should.
+	 */
+	if (logflags) {
+		ASSERT(tp && wr);
+		xfs_trans_log_inode(tp, ip, logflags);
+	}
+	if (cur) {
+		if (!error) {
+			ASSERT(*firstblock == NULLFSBLOCK ||
+			       XFS_FSB_TO_AGNO(mp, *firstblock) ==
+			       XFS_FSB_TO_AGNO(mp,
+				       cur->bc_private.b.firstblock) ||
+			       (flist->xbf_low &&
+				XFS_FSB_TO_AGNO(mp, *firstblock) <
+				XFS_FSB_TO_AGNO(mp,
+					cur->bc_private.b.firstblock)));
+			*firstblock = cur->bc_private.b.firstblock;
+		}
+		xfs_btree_del_cursor(cur,
+			error ? XFS_BTREE_ERROR : XFS_BTREE_NOERROR);
+	}
+	if (!error)
+		xfs_bmap_validate_ret(orig_bno, orig_len, orig_flags, orig_mval,
+			orig_nmap, *nmap);
+	return error;
+}
+
+/*
+ * Map file blocks to filesystem blocks, simple version.
+ * One block (extent) only, read-only.
+ * For flags, only the XFS_BMAPI_ATTRFORK flag is examined.
+ * For the other flag values, the effect is as if XFS_BMAPI_METADATA
+ * was set and all the others were clear.
+ */
+int						/* error */
+xfs_bmapi_single(
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_inode_t	*ip,		/* incore inode */
+	int		whichfork,	/* data or attr fork */
+	xfs_fsblock_t	*fsb,		/* output: mapped block */
+	xfs_fileoff_t	bno)		/* starting file offs. mapped */
+{
+	int		eof;		/* we've hit the end of extents */
+	int		error;		/* error return */
+	xfs_bmbt_irec_t	got;		/* current file extent record */
+	xfs_ifork_t	*ifp;		/* inode fork pointer */
+	xfs_extnum_t	lastx;		/* last useful extent number */
+	xfs_bmbt_irec_t	prev;		/* previous file extent record */
+
+	ifp = XFS_IFORK_PTR(ip, whichfork);
+	if (unlikely(
+	    XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_BTREE &&
+	    XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_EXTENTS)) {
+	       XFS_ERROR_REPORT("xfs_bmapi_single", XFS_ERRLEVEL_LOW,
+				ip->i_mount);
+	       return XFS_ERROR(EFSCORRUPTED);
+	}
+	if (XFS_FORCED_SHUTDOWN(ip->i_mount))
+		return XFS_ERROR(EIO);
+	XFS_STATS_INC(xs_blk_mapr);
+	if (!(ifp->if_flags & XFS_IFEXTENTS) &&
+	    (error = xfs_iread_extents(tp, ip, whichfork)))
+		return error;
+	(void)xfs_bmap_search_extents(ip, bno, whichfork, &eof, &lastx, &got,
+		&prev);
+	/*
+	 * Reading past eof, act as though there's a hole
+	 * up to end.
+	 */
+	if (eof || got.br_startoff > bno) {
+		*fsb = NULLFSBLOCK;
+		return 0;
+	}
+	ASSERT(!isnullstartblock(got.br_startblock));
+	ASSERT(bno < got.br_startoff + got.br_blockcount);
+	*fsb = got.br_startblock + (bno - got.br_startoff);
+	return 0;
+}
+
+/*
+ * Unmap (remove) blocks from a file.
+ * If nexts is nonzero then the number of extents to remove is limited to
+ * that value.  If not all extents in the block range can be removed then
+ * *done is set.
+ */
+int						/* error */
+xfs_bunmapi(
+	xfs_trans_t		*tp,		/* transaction pointer */
+	struct xfs_inode	*ip,		/* incore inode */
+	xfs_fileoff_t		bno,		/* starting offset to unmap */
+	xfs_filblks_t		len,		/* length to unmap in file */
+	int			flags,		/* misc flags */
+	xfs_extnum_t		nexts,		/* number of extents max */
+	xfs_fsblock_t		*firstblock,	/* first allocated block
+						   controls a.g. for allocs */
+	xfs_bmap_free_t		*flist,		/* i/o: list extents to free */
+	int			*done)		/* set if not done yet */
+{
+	xfs_btree_cur_t		*cur;		/* bmap btree cursor */
+	xfs_bmbt_irec_t		del;		/* extent being deleted */
+	int			eof;		/* is deleting at eof */
+	xfs_bmbt_rec_host_t	*ep;		/* extent record pointer */
+	int			error;		/* error return value */
+	xfs_extnum_t		extno;		/* extent number in list */
+	xfs_bmbt_irec_t		got;		/* current extent record */
+	xfs_ifork_t		*ifp;		/* inode fork pointer */
+	int			isrt;		/* freeing in rt area */
+	xfs_extnum_t		lastx;		/* last extent index used */
+	int			logflags;	/* transaction logging flags */
+	xfs_extlen_t		mod;		/* rt extent offset */
+	xfs_mount_t		*mp;		/* mount structure */
+	xfs_extnum_t		nextents;	/* number of file extents */
+	xfs_bmbt_irec_t		prev;		/* previous extent record */
+	xfs_fileoff_t		start;		/* first file offset deleted */
+	int			tmp_logflags;	/* partial logging flags */
+	int			wasdel;		/* was a delayed alloc extent */
+	int			whichfork;	/* data or attribute fork */
+	xfs_fsblock_t		sum;
+
+	trace_xfs_bunmap(ip, bno, len, flags, _RET_IP_);
+
+	whichfork = (flags & XFS_BMAPI_ATTRFORK) ?
+		XFS_ATTR_FORK : XFS_DATA_FORK;
+	ifp = XFS_IFORK_PTR(ip, whichfork);
+	if (unlikely(
+	    XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_EXTENTS &&
+	    XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_BTREE)) {
+		XFS_ERROR_REPORT("xfs_bunmapi", XFS_ERRLEVEL_LOW,
+				 ip->i_mount);
+		return XFS_ERROR(EFSCORRUPTED);
+	}
+	mp = ip->i_mount;
+	if (XFS_FORCED_SHUTDOWN(mp))
+		return XFS_ERROR(EIO);
+
+	ASSERT(len > 0);
+	ASSERT(nexts >= 0);
+	ASSERT(ifp->if_ext_max ==
+	       XFS_IFORK_SIZE(ip, whichfork) / (uint)sizeof(xfs_bmbt_rec_t));
+	if (!(ifp->if_flags & XFS_IFEXTENTS) &&
+	    (error = xfs_iread_extents(tp, ip, whichfork)))
+		return error;
+	nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
+	if (nextents == 0) {
+		*done = 1;
+		return 0;
+	}
+	XFS_STATS_INC(xs_blk_unmap);
+	isrt = (whichfork == XFS_DATA_FORK) && XFS_IS_REALTIME_INODE(ip);
+	start = bno;
+	bno = start + len - 1;
+	ep = xfs_bmap_search_extents(ip, bno, whichfork, &eof, &lastx, &got,
+		&prev);
+
+	/*
+	 * Check to see if the given block number is past the end of the
+	 * file, back up to the last block if so...
+	 */
+	if (eof) {
+		ep = xfs_iext_get_ext(ifp, --lastx);
+		xfs_bmbt_get_all(ep, &got);
+		bno = got.br_startoff + got.br_blockcount - 1;
+	}
+	logflags = 0;
+	if (ifp->if_flags & XFS_IFBROOT) {
+		ASSERT(XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_BTREE);
+		cur = xfs_bmbt_init_cursor(mp, tp, ip, whichfork);
+		cur->bc_private.b.firstblock = *firstblock;
+		cur->bc_private.b.flist = flist;
+		cur->bc_private.b.flags = 0;
+	} else
+		cur = NULL;
+	extno = 0;
+	while (bno != (xfs_fileoff_t)-1 && bno >= start && lastx >= 0 &&
+	       (nexts == 0 || extno < nexts)) {
+		/*
+		 * Is the found extent after a hole in which bno lives?
+		 * Just back up to the previous extent, if so.
+		 */
+		if (got.br_startoff > bno) {
+			if (--lastx < 0)
+				break;
+			ep = xfs_iext_get_ext(ifp, lastx);
+			xfs_bmbt_get_all(ep, &got);
+		}
+		/*
+		 * Is the last block of this extent before the range
+		 * we're supposed to delete?  If so, we're done.
+		 */
+		bno = XFS_FILEOFF_MIN(bno,
+			got.br_startoff + got.br_blockcount - 1);
+		if (bno < start)
+			break;
+		/*
+		 * Then deal with the (possibly delayed) allocated space
+		 * we found.
+		 */
+		ASSERT(ep != NULL);
+		del = got;
+		wasdel = isnullstartblock(del.br_startblock);
+		if (got.br_startoff < start) {
+			del.br_startoff = start;
+			del.br_blockcount -= start - got.br_startoff;
+			if (!wasdel)
+				del.br_startblock += start - got.br_startoff;
+		}
+		if (del.br_startoff + del.br_blockcount > bno + 1)
+			del.br_blockcount = bno + 1 - del.br_startoff;
+		sum = del.br_startblock + del.br_blockcount;
+		if (isrt &&
+		    (mod = do_mod(sum, mp->m_sb.sb_rextsize))) {
+			/*
+			 * Realtime extent not lined up at the end.
+			 * The extent could have been split into written
+			 * and unwritten pieces, or we could just be
+			 * unmapping part of it.  But we can't really
+			 * get rid of part of a realtime extent.
+			 */
+			if (del.br_state == XFS_EXT_UNWRITTEN ||
+			    !xfs_sb_version_hasextflgbit(&mp->m_sb)) {
+				/*
+				 * This piece is unwritten, or we're not
+				 * using unwritten extents.  Skip over it.
+				 */
+				ASSERT(bno >= mod);
+				bno -= mod > del.br_blockcount ?
+					del.br_blockcount : mod;
+				if (bno < got.br_startoff) {
+					if (--lastx >= 0)
+						xfs_bmbt_get_all(xfs_iext_get_ext(
+							ifp, lastx), &got);
+				}
+				continue;
+			}
+			/*
+			 * It's written, turn it unwritten.
+			 * This is better than zeroing it.
+			 */
+			ASSERT(del.br_state == XFS_EXT_NORM);
+			ASSERT(xfs_trans_get_block_res(tp) > 0);
+			/*
+			 * If this spans a realtime extent boundary,
+			 * chop it back to the start of the one we end at.
+			 */
+			if (del.br_blockcount > mod) {
+				del.br_startoff += del.br_blockcount - mod;
+				del.br_startblock += del.br_blockcount - mod;
+				del.br_blockcount = mod;
+			}
+			del.br_state = XFS_EXT_UNWRITTEN;
+			error = xfs_bmap_add_extent(ip, &lastx, &cur, &del,
+				firstblock, flist, &logflags,
+				XFS_DATA_FORK);
+			if (error)
+				goto error0;
+			goto nodelete;
+		}
+		if (isrt && (mod = do_mod(del.br_startblock, mp->m_sb.sb_rextsize))) {
+			/*
+			 * Realtime extent is lined up at the end but not
+			 * at the front.  We'll get rid of full extents if
+			 * we can.
+			 */
+			mod = mp->m_sb.sb_rextsize - mod;
+			if (del.br_blockcount > mod) {
+				del.br_blockcount -= mod;
+				del.br_startoff += mod;
+				del.br_startblock += mod;
+			} else if ((del.br_startoff == start &&
+				    (del.br_state == XFS_EXT_UNWRITTEN ||
+				     xfs_trans_get_block_res(tp) == 0)) ||
+				   !xfs_sb_version_hasextflgbit(&mp->m_sb)) {
+				/*
+				 * Can't make it unwritten.  There isn't
+				 * a full extent here so just skip it.
+				 */
+				ASSERT(bno >= del.br_blockcount);
+				bno -= del.br_blockcount;
+				if (got.br_startoff > bno) {
+					if (--lastx >= 0) {
+						ep = xfs_iext_get_ext(ifp,
+								      lastx);
+						xfs_bmbt_get_all(ep, &got);
+					}
+				}
+				continue;
+			} else if (del.br_state == XFS_EXT_UNWRITTEN) {
+				/*
+				 * This one is already unwritten.
+				 * It must have a written left neighbor.
+				 * Unwrite the killed part of that one and
+				 * try again.
+				 */
+				ASSERT(lastx > 0);
+				xfs_bmbt_get_all(xfs_iext_get_ext(ifp,
+						lastx - 1), &prev);
+				ASSERT(prev.br_state == XFS_EXT_NORM);
+				ASSERT(!isnullstartblock(prev.br_startblock));
+				ASSERT(del.br_startblock ==
+				       prev.br_startblock + prev.br_blockcount);
+				if (prev.br_startoff < start) {
+					mod = start - prev.br_startoff;
+					prev.br_blockcount -= mod;
+					prev.br_startblock += mod;
+					prev.br_startoff = start;
+				}
+				prev.br_state = XFS_EXT_UNWRITTEN;
+				lastx--;
+				error = xfs_bmap_add_extent(ip, &lastx, &cur,
+					&prev, firstblock, flist, &logflags,
+					XFS_DATA_FORK);
+				if (error)
+					goto error0;
+				goto nodelete;
+			} else {
+				ASSERT(del.br_state == XFS_EXT_NORM);
+				del.br_state = XFS_EXT_UNWRITTEN;
+				error = xfs_bmap_add_extent(ip, &lastx, &cur,
+					&del, firstblock, flist, &logflags,
+					XFS_DATA_FORK);
+				if (error)
+					goto error0;
+				goto nodelete;
+			}
+		}
+		if (wasdel) {
+			ASSERT(startblockval(del.br_startblock) > 0);
+			/* Update realtime/data freespace, unreserve quota */
+			if (isrt) {
+				xfs_filblks_t rtexts;
+
+				rtexts = XFS_FSB_TO_B(mp, del.br_blockcount);
+				do_div(rtexts, mp->m_sb.sb_rextsize);
+				xfs_mod_incore_sb(mp, XFS_SBS_FREXTENTS,
+						(int64_t)rtexts, 0);
+				(void)xfs_trans_reserve_quota_nblks(NULL,
+					ip, -((long)del.br_blockcount), 0,
+					XFS_QMOPT_RES_RTBLKS);
+			} else {
+				xfs_icsb_modify_counters(mp, XFS_SBS_FDBLOCKS,
+						(int64_t)del.br_blockcount, 0);
+				(void)xfs_trans_reserve_quota_nblks(NULL,
+					ip, -((long)del.br_blockcount), 0,
+					XFS_QMOPT_RES_REGBLKS);
+			}
+			ip->i_delayed_blks -= del.br_blockcount;
+			if (cur)
+				cur->bc_private.b.flags |=
+					XFS_BTCUR_BPRV_WASDEL;
+		} else if (cur)
+			cur->bc_private.b.flags &= ~XFS_BTCUR_BPRV_WASDEL;
+		/*
+		 * If it's the case where the directory code is running
+		 * with no block reservation, and the deleted block is in
+		 * the middle of its extent, and the resulting insert
+		 * of an extent would cause transformation to btree format,
+		 * then reject it.  The calling code will then swap
+		 * blocks around instead.
+		 * We have to do this now, rather than waiting for the
+		 * conversion to btree format, since the transaction
+		 * will be dirty.
+		 */
+		if (!wasdel && xfs_trans_get_block_res(tp) == 0 &&
+		    XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_EXTENTS &&
+		    XFS_IFORK_NEXTENTS(ip, whichfork) >= ifp->if_ext_max &&
+		    del.br_startoff > got.br_startoff &&
+		    del.br_startoff + del.br_blockcount <
+		    got.br_startoff + got.br_blockcount) {
+			error = XFS_ERROR(ENOSPC);
+			goto error0;
+		}
+		error = xfs_bmap_del_extent(ip, tp, &lastx, flist, cur, &del,
+				&tmp_logflags, whichfork);
+		logflags |= tmp_logflags;
+		if (error)
+			goto error0;
+		bno = del.br_startoff - 1;
+nodelete:
+		/*
+		 * If not done go on to the next (previous) record.
+		 */
+		if (bno != (xfs_fileoff_t)-1 && bno >= start) {
+			if (lastx >= 0) {
+				ep = xfs_iext_get_ext(ifp, lastx);
+				if (xfs_bmbt_get_startoff(ep) > bno) {
+					if (--lastx >= 0)
+						ep = xfs_iext_get_ext(ifp,
+								      lastx);
+				}
+				xfs_bmbt_get_all(ep, &got);
+			}
+			extno++;
+		}
+	}
+	*done = bno == (xfs_fileoff_t)-1 || bno < start || lastx < 0;
+	ASSERT(ifp->if_ext_max ==
+	       XFS_IFORK_SIZE(ip, whichfork) / (uint)sizeof(xfs_bmbt_rec_t));
+	/*
+	 * Convert to a btree if necessary.
+	 */
+	if (XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_EXTENTS &&
+	    XFS_IFORK_NEXTENTS(ip, whichfork) > ifp->if_ext_max) {
+		ASSERT(cur == NULL);
+		error = xfs_bmap_extents_to_btree(tp, ip, firstblock, flist,
+			&cur, 0, &tmp_logflags, whichfork);
+		logflags |= tmp_logflags;
+		if (error)
+			goto error0;
+	}
+	/*
+	 * transform from btree to extents, give it cur
+	 */
+	else if (XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_BTREE &&
+		 XFS_IFORK_NEXTENTS(ip, whichfork) <= ifp->if_ext_max) {
+		ASSERT(cur != NULL);
+		error = xfs_bmap_btree_to_extents(tp, ip, cur, &tmp_logflags,
+			whichfork);
+		logflags |= tmp_logflags;
+		if (error)
+			goto error0;
+	}
+	/*
+	 * transform from extents to local?
+	 */
+	ASSERT(ifp->if_ext_max ==
+	       XFS_IFORK_SIZE(ip, whichfork) / (uint)sizeof(xfs_bmbt_rec_t));
+	error = 0;
+error0:
+	/*
+	 * Log everything.  Do this after conversion, there's no point in
+	 * logging the extent records if we've converted to btree format.
+	 */
+	if ((logflags & xfs_ilog_fext(whichfork)) &&
+	    XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_EXTENTS)
+		logflags &= ~xfs_ilog_fext(whichfork);
+	else if ((logflags & xfs_ilog_fbroot(whichfork)) &&
+		 XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_BTREE)
+		logflags &= ~xfs_ilog_fbroot(whichfork);
+	/*
+	 * Log inode even in the error case, if the transaction
+	 * is dirty we'll need to shut down the filesystem.
+	 */
+	if (logflags)
+		xfs_trans_log_inode(tp, ip, logflags);
+	if (cur) {
+		if (!error) {
+			*firstblock = cur->bc_private.b.firstblock;
+			cur->bc_private.b.allocated = 0;
+		}
+		xfs_btree_del_cursor(cur,
+			error ? XFS_BTREE_ERROR : XFS_BTREE_NOERROR);
+	}
+	return error;
+}
+
+/*
+ * returns 1 for success, 0 if we failed to map the extent.
+ */
+STATIC int
+xfs_getbmapx_fix_eof_hole(
+	xfs_inode_t		*ip,		/* xfs incore inode pointer */
+	struct getbmapx		*out,		/* output structure */
+	int			prealloced,	/* this is a file with
+						 * preallocated data space */
+	__int64_t		end,		/* last block requested */
+	xfs_fsblock_t		startblock)
+{
+	__int64_t		fixlen;
+	xfs_mount_t		*mp;		/* file system mount point */
+	xfs_ifork_t		*ifp;		/* inode fork pointer */
+	xfs_extnum_t		lastx;		/* last extent pointer */
+	xfs_fileoff_t		fileblock;
+
+	if (startblock == HOLESTARTBLOCK) {
+		mp = ip->i_mount;
+		out->bmv_block = -1;
+		fixlen = XFS_FSB_TO_BB(mp, XFS_B_TO_FSB(mp, ip->i_size));
+		fixlen -= out->bmv_offset;
+		if (prealloced && out->bmv_offset + out->bmv_length == end) {
+			/* Came to hole at EOF. Trim it. */
+			if (fixlen <= 0)
+				return 0;
+			out->bmv_length = fixlen;
+		}
+	} else {
+		if (startblock == DELAYSTARTBLOCK)
+			out->bmv_block = -2;
+		else
+			out->bmv_block = xfs_fsb_to_db(ip, startblock);
+		fileblock = XFS_BB_TO_FSB(ip->i_mount, out->bmv_offset);
+		ifp = XFS_IFORK_PTR(ip, XFS_DATA_FORK);
+		if (xfs_iext_bno_to_ext(ifp, fileblock, &lastx) &&
+		   (lastx == (ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t))-1))
+			out->bmv_oflags |= BMV_OF_LAST;
+	}
+
+	return 1;
+}
+
+/*
+ * Get inode's extents as described in bmv, and format for output.
+ * Calls formatter to fill the user's buffer until all extents
+ * are mapped, until the passed-in bmv->bmv_count slots have
+ * been filled, or until the formatter short-circuits the loop,
+ * if it is tracking filled-in extents on its own.
+ */
+int						/* error code */
+xfs_getbmap(
+	xfs_inode_t		*ip,
+	struct getbmapx		*bmv,		/* user bmap structure */
+	xfs_bmap_format_t	formatter,	/* format to user */
+	void			*arg)		/* formatter arg */
+{
+	__int64_t		bmvend;		/* last block requested */
+	int			error = 0;	/* return value */
+	__int64_t		fixlen;		/* length for -1 case */
+	int			i;		/* extent number */
+	int			lock;		/* lock state */
+	xfs_bmbt_irec_t		*map;		/* buffer for user's data */
+	xfs_mount_t		*mp;		/* file system mount point */
+	int			nex;		/* # of user extents can do */
+	int			nexleft;	/* # of user extents left */
+	int			subnex;		/* # of bmapi's can do */
+	int			nmap;		/* number of map entries */
+	struct getbmapx		*out;		/* output structure */
+	int			whichfork;	/* data or attr fork */
+	int			prealloced;	/* this is a file with
+						 * preallocated data space */
+	int			iflags;		/* interface flags */
+	int			bmapi_flags;	/* flags for xfs_bmapi */
+	int			cur_ext = 0;
+
+	mp = ip->i_mount;
+	iflags = bmv->bmv_iflags;
+	whichfork = iflags & BMV_IF_ATTRFORK ? XFS_ATTR_FORK : XFS_DATA_FORK;
+
+	if (whichfork == XFS_ATTR_FORK) {
+		if (XFS_IFORK_Q(ip)) {
+			if (ip->i_d.di_aformat != XFS_DINODE_FMT_EXTENTS &&
+			    ip->i_d.di_aformat != XFS_DINODE_FMT_BTREE &&
+			    ip->i_d.di_aformat != XFS_DINODE_FMT_LOCAL)
+				return XFS_ERROR(EINVAL);
+		} else if (unlikely(
+			   ip->i_d.di_aformat != 0 &&
+			   ip->i_d.di_aformat != XFS_DINODE_FMT_EXTENTS)) {
+			XFS_ERROR_REPORT("xfs_getbmap", XFS_ERRLEVEL_LOW,
+					 ip->i_mount);
+			return XFS_ERROR(EFSCORRUPTED);
+		}
+
+		prealloced = 0;
+		fixlen = 1LL << 32;
+	} else {
+		if (ip->i_d.di_format != XFS_DINODE_FMT_EXTENTS &&
+		    ip->i_d.di_format != XFS_DINODE_FMT_BTREE &&
+		    ip->i_d.di_format != XFS_DINODE_FMT_LOCAL)
+			return XFS_ERROR(EINVAL);
+
+		if (xfs_get_extsz_hint(ip) ||
+		    ip->i_d.di_flags & (XFS_DIFLAG_PREALLOC|XFS_DIFLAG_APPEND)){
+			prealloced = 1;
+			fixlen = XFS_MAXIOFFSET(mp);
+		} else {
+			prealloced = 0;
+			fixlen = ip->i_size;
+		}
+	}
+
+	if (bmv->bmv_length == -1) {
+		fixlen = XFS_FSB_TO_BB(mp, XFS_B_TO_FSB(mp, fixlen));
+		bmv->bmv_length =
+			max_t(__int64_t, fixlen - bmv->bmv_offset, 0);
+	} else if (bmv->bmv_length == 0) {
+		bmv->bmv_entries = 0;
+		return 0;
+	} else if (bmv->bmv_length < 0) {
+		return XFS_ERROR(EINVAL);
+	}
+
+	nex = bmv->bmv_count - 1;
+	if (nex <= 0)
+		return XFS_ERROR(EINVAL);
+	bmvend = bmv->bmv_offset + bmv->bmv_length;
+
+
+	if (bmv->bmv_count > ULONG_MAX / sizeof(struct getbmapx))
+		return XFS_ERROR(ENOMEM);
+	out = kmem_zalloc(bmv->bmv_count * sizeof(struct getbmapx), KM_MAYFAIL);
+	if (!out)
+		return XFS_ERROR(ENOMEM);
+
+	xfs_ilock(ip, XFS_IOLOCK_SHARED);
+	if (whichfork == XFS_DATA_FORK && !(iflags & BMV_IF_DELALLOC)) {
+		if (ip->i_delayed_blks || ip->i_size > ip->i_d.di_size) {
+			error = xfs_flush_pages(ip, 0, -1, 0, FI_REMAPF);
+			if (error)
+				goto out_unlock_iolock;
+		}
+		/*
+		 * even after flushing the inode, there can still be delalloc
+		 * blocks on the inode beyond EOF due to speculative
+		 * preallocation. These are not removed until the release
+		 * function is called or the inode is inactivated. Hence we
+		 * cannot assert here that ip->i_delayed_blks == 0.
+		 */
+	}
+
+	lock = xfs_ilock_map_shared(ip);
+
+	/*
+	 * Don't let nex be bigger than the number of extents
+	 * we can have assuming alternating holes and real extents.
+	 */
+	if (nex > XFS_IFORK_NEXTENTS(ip, whichfork) * 2 + 1)
+		nex = XFS_IFORK_NEXTENTS(ip, whichfork) * 2 + 1;
+
+	bmapi_flags = xfs_bmapi_aflag(whichfork);
+	if (!(iflags & BMV_IF_PREALLOC))
+		bmapi_flags |= XFS_BMAPI_IGSTATE;
+
+	/*
+	 * Allocate enough space to handle "subnex" maps at a time.
+	 */
+	error = ENOMEM;
+	subnex = 16;
+	map = kmem_alloc(subnex * sizeof(*map), KM_MAYFAIL | KM_NOFS);
+	if (!map)
+		goto out_unlock_ilock;
+
+	bmv->bmv_entries = 0;
+
+	if (XFS_IFORK_NEXTENTS(ip, whichfork) == 0 &&
+	    (whichfork == XFS_ATTR_FORK || !(iflags & BMV_IF_DELALLOC))) {
+		error = 0;
+		goto out_free_map;
+	}
+
+	nexleft = nex;
+
+	do {
+		nmap = (nexleft > subnex) ? subnex : nexleft;
+		error = xfs_bmapi(NULL, ip, XFS_BB_TO_FSBT(mp, bmv->bmv_offset),
+				  XFS_BB_TO_FSB(mp, bmv->bmv_length),
+				  bmapi_flags, NULL, 0, map, &nmap,
+				  NULL);
+		if (error)
+			goto out_free_map;
+		ASSERT(nmap <= subnex);
+
+		for (i = 0; i < nmap && nexleft && bmv->bmv_length; i++) {
+			out[cur_ext].bmv_oflags = 0;
+			if (map[i].br_state == XFS_EXT_UNWRITTEN)
+				out[cur_ext].bmv_oflags |= BMV_OF_PREALLOC;
+			else if (map[i].br_startblock == DELAYSTARTBLOCK)
+				out[cur_ext].bmv_oflags |= BMV_OF_DELALLOC;
+			out[cur_ext].bmv_offset =
+				XFS_FSB_TO_BB(mp, map[i].br_startoff);
+			out[cur_ext].bmv_length =
+				XFS_FSB_TO_BB(mp, map[i].br_blockcount);
+			out[cur_ext].bmv_unused1 = 0;
+			out[cur_ext].bmv_unused2 = 0;
+			ASSERT(((iflags & BMV_IF_DELALLOC) != 0) ||
+			      (map[i].br_startblock != DELAYSTARTBLOCK));
+                        if (map[i].br_startblock == HOLESTARTBLOCK &&
+			    whichfork == XFS_ATTR_FORK) {
+				/* came to the end of attribute fork */
+				out[cur_ext].bmv_oflags |= BMV_OF_LAST;
+				goto out_free_map;
+			}
+
+			if (!xfs_getbmapx_fix_eof_hole(ip, &out[cur_ext],
+					prealloced, bmvend,
+					map[i].br_startblock))
+				goto out_free_map;
+
+			bmv->bmv_offset =
+				out[cur_ext].bmv_offset +
+				out[cur_ext].bmv_length;
+			bmv->bmv_length =
+				max_t(__int64_t, 0, bmvend - bmv->bmv_offset);
+
+			/*
+			 * In case we don't want to return the hole,
+			 * don't increase cur_ext so that we can reuse
+			 * it in the next loop.
+			 */
+			if ((iflags & BMV_IF_NO_HOLES) &&
+			    map[i].br_startblock == HOLESTARTBLOCK) {
+				memset(&out[cur_ext], 0, sizeof(out[cur_ext]));
+				continue;
+			}
+
+			nexleft--;
+			bmv->bmv_entries++;
+			cur_ext++;
+		}
+	} while (nmap && nexleft && bmv->bmv_length);
+
+ out_free_map:
+	kmem_free(map);
+ out_unlock_ilock:
+	xfs_iunlock_map_shared(ip, lock);
+ out_unlock_iolock:
+	xfs_iunlock(ip, XFS_IOLOCK_SHARED);
+
+	for (i = 0; i < cur_ext; i++) {
+		int full = 0;	/* user array is full */
+
+		/* format results & advance arg */
+		error = formatter(&arg, &out[i], &full);
+		if (error || full)
+			break;
+	}
+
+	kmem_free(out);
+	return error;
+}
+
+/*
+ * Check the last inode extent to determine whether this allocation will result
+ * in blocks being allocated at the end of the file. When we allocate new data
+ * blocks at the end of the file which do not start at the previous data block,
+ * we will try to align the new blocks at stripe unit boundaries.
+ */
+STATIC int				/* error */
+xfs_bmap_isaeof(
+	xfs_inode_t	*ip,		/* incore inode pointer */
+	xfs_fileoff_t   off,		/* file offset in fsblocks */
+	int             whichfork,	/* data or attribute fork */
+	char		*aeof)		/* return value */
+{
+	int		error;		/* error return value */
+	xfs_ifork_t	*ifp;		/* inode fork pointer */
+	xfs_bmbt_rec_host_t *lastrec;	/* extent record pointer */
+	xfs_extnum_t	nextents;	/* number of file extents */
+	xfs_bmbt_irec_t	s;		/* expanded extent record */
+
+	ASSERT(whichfork == XFS_DATA_FORK);
+	ifp = XFS_IFORK_PTR(ip, whichfork);
+	if (!(ifp->if_flags & XFS_IFEXTENTS) &&
+	    (error = xfs_iread_extents(NULL, ip, whichfork)))
+		return error;
+	nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
+	if (nextents == 0) {
+		*aeof = 1;
+		return 0;
+	}
+	/*
+	 * Go to the last extent
+	 */
+	lastrec = xfs_iext_get_ext(ifp, nextents - 1);
+	xfs_bmbt_get_all(lastrec, &s);
+	/*
+	 * Check we are allocating in the last extent (for delayed allocations)
+	 * or past the last extent for non-delayed allocations.
+	 */
+	*aeof = (off >= s.br_startoff &&
+		 off < s.br_startoff + s.br_blockcount &&
+		 isnullstartblock(s.br_startblock)) ||
+		off >= s.br_startoff + s.br_blockcount;
+	return 0;
+}
+
+/*
+ * Check if the endoff is outside the last extent. If so the caller will grow
+ * the allocation to a stripe unit boundary.
+ */
+int					/* error */
+xfs_bmap_eof(
+	xfs_inode_t	*ip,		/* incore inode pointer */
+	xfs_fileoff_t	endoff,		/* file offset in fsblocks */
+	int		whichfork,	/* data or attribute fork */
+	int		*eof)		/* result value */
+{
+	xfs_fsblock_t	blockcount;	/* extent block count */
+	int		error;		/* error return value */
+	xfs_ifork_t	*ifp;		/* inode fork pointer */
+	xfs_bmbt_rec_host_t *lastrec;	/* extent record pointer */
+	xfs_extnum_t	nextents;	/* number of file extents */
+	xfs_fileoff_t	startoff;	/* extent starting file offset */
+
+	ASSERT(whichfork == XFS_DATA_FORK);
+	ifp = XFS_IFORK_PTR(ip, whichfork);
+	if (!(ifp->if_flags & XFS_IFEXTENTS) &&
+	    (error = xfs_iread_extents(NULL, ip, whichfork)))
+		return error;
+	nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
+	if (nextents == 0) {
+		*eof = 1;
+		return 0;
+	}
+	/*
+	 * Go to the last extent
+	 */
+	lastrec = xfs_iext_get_ext(ifp, nextents - 1);
+	startoff = xfs_bmbt_get_startoff(lastrec);
+	blockcount = xfs_bmbt_get_blockcount(lastrec);
+	*eof = endoff >= startoff + blockcount;
+	return 0;
+}
+
+#ifdef DEBUG
+STATIC struct xfs_buf *
+xfs_bmap_get_bp(
+	struct xfs_btree_cur	*cur,
+	xfs_fsblock_t		bno)
+{
+	struct xfs_log_item_desc *lidp;
+	int			i;
+
+	if (!cur)
+		return NULL;
+
+	for (i = 0; i < XFS_BTREE_MAXLEVELS; i++) {
+		if (!cur->bc_bufs[i])
+			break;
+		if (XFS_BUF_ADDR(cur->bc_bufs[i]) == bno)
+			return cur->bc_bufs[i];
+	}
+
+	/* Chase down all the log items to see if the bp is there */
+	list_for_each_entry(lidp, &cur->bc_tp->t_items, lid_trans) {
+		struct xfs_buf_log_item	*bip;
+		bip = (struct xfs_buf_log_item *)lidp->lid_item;
+		if (bip->bli_item.li_type == XFS_LI_BUF &&
+		    XFS_BUF_ADDR(bip->bli_buf) == bno)
+			return bip->bli_buf;
+	}
+
+	return NULL;
+}
+
+STATIC void
+xfs_check_block(
+	struct xfs_btree_block	*block,
+	xfs_mount_t		*mp,
+	int			root,
+	short			sz)
+{
+	int			i, j, dmxr;
+	__be64			*pp, *thispa;	/* pointer to block address */
+	xfs_bmbt_key_t		*prevp, *keyp;
+
+	ASSERT(be16_to_cpu(block->bb_level) > 0);
+
+	prevp = NULL;
+	for( i = 1; i <= xfs_btree_get_numrecs(block); i++) {
+		dmxr = mp->m_bmap_dmxr[0];
+		keyp = XFS_BMBT_KEY_ADDR(mp, block, i);
+
+		if (prevp) {
+			ASSERT(be64_to_cpu(prevp->br_startoff) <
+			       be64_to_cpu(keyp->br_startoff));
+		}
+		prevp = keyp;
+
+		/*
+		 * Compare the block numbers to see if there are dups.
+		 */
+		if (root)
+			pp = XFS_BMAP_BROOT_PTR_ADDR(mp, block, i, sz);
+		else
+			pp = XFS_BMBT_PTR_ADDR(mp, block, i, dmxr);
+
+		for (j = i+1; j <= be16_to_cpu(block->bb_numrecs); j++) {
+			if (root)
+				thispa = XFS_BMAP_BROOT_PTR_ADDR(mp, block, j, sz);
+			else
+				thispa = XFS_BMBT_PTR_ADDR(mp, block, j, dmxr);
+			if (*thispa == *pp) {
+				xfs_warn(mp, "%s: thispa(%d) == pp(%d) %Ld",
+					__func__, j, i,
+					(unsigned long long)be64_to_cpu(*thispa));
+				panic("%s: ptrs are equal in node\n",
+					__func__);
+			}
+		}
+	}
+}
+
+/*
+ * Check that the extents for the inode ip are in the right order in all
+ * btree leaves.
+ */
+
+STATIC void
+xfs_bmap_check_leaf_extents(
+	xfs_btree_cur_t		*cur,	/* btree cursor or null */
+	xfs_inode_t		*ip,		/* incore inode pointer */
+	int			whichfork)	/* data or attr fork */
+{
+	struct xfs_btree_block	*block;	/* current btree block */
+	xfs_fsblock_t		bno;	/* block # of "block" */
+	xfs_buf_t		*bp;	/* buffer for "block" */
+	int			error;	/* error return value */
+	xfs_extnum_t		i=0, j;	/* index into the extents list */
+	xfs_ifork_t		*ifp;	/* fork structure */
+	int			level;	/* btree level, for checking */
+	xfs_mount_t		*mp;	/* file system mount structure */
+	__be64			*pp;	/* pointer to block address */
+	xfs_bmbt_rec_t		*ep;	/* pointer to current extent */
+	xfs_bmbt_rec_t		last = {0, 0}; /* last extent in prev block */
+	xfs_bmbt_rec_t		*nextp;	/* pointer to next extent */
+	int			bp_release = 0;
+
+	if (XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_BTREE) {
+		return;
+	}
+
+	bno = NULLFSBLOCK;
+	mp = ip->i_mount;
+	ifp = XFS_IFORK_PTR(ip, whichfork);
+	block = ifp->if_broot;
+	/*
+	 * Root level must use BMAP_BROOT_PTR_ADDR macro to get ptr out.
+	 */
+	level = be16_to_cpu(block->bb_level);
+	ASSERT(level > 0);
+	xfs_check_block(block, mp, 1, ifp->if_broot_bytes);
+	pp = XFS_BMAP_BROOT_PTR_ADDR(mp, block, 1, ifp->if_broot_bytes);
+	bno = be64_to_cpu(*pp);
+
+	ASSERT(bno != NULLDFSBNO);
+	ASSERT(XFS_FSB_TO_AGNO(mp, bno) < mp->m_sb.sb_agcount);
+	ASSERT(XFS_FSB_TO_AGBNO(mp, bno) < mp->m_sb.sb_agblocks);
+
+	/*
+	 * Go down the tree until leaf level is reached, following the first
+	 * pointer (leftmost) at each level.
+	 */
+	while (level-- > 0) {
+		/* See if buf is in cur first */
+		bp = xfs_bmap_get_bp(cur, XFS_FSB_TO_DADDR(mp, bno));
+		if (bp) {
+			bp_release = 0;
+		} else {
+			bp_release = 1;
+		}
+		if (!bp && (error = xfs_btree_read_bufl(mp, NULL, bno, 0, &bp,
+				XFS_BMAP_BTREE_REF)))
+			goto error_norelse;
+		block = XFS_BUF_TO_BLOCK(bp);
+		XFS_WANT_CORRUPTED_GOTO(
+			xfs_bmap_sanity_check(mp, bp, level),
+			error0);
+		if (level == 0)
+			break;
+
+		/*
+		 * Check this block for basic sanity (increasing keys and
+		 * no duplicate blocks).
+		 */
+
+		xfs_check_block(block, mp, 0, 0);
+		pp = XFS_BMBT_PTR_ADDR(mp, block, 1, mp->m_bmap_dmxr[1]);
+		bno = be64_to_cpu(*pp);
+		XFS_WANT_CORRUPTED_GOTO(XFS_FSB_SANITY_CHECK(mp, bno), error0);
+		if (bp_release) {
+			bp_release = 0;
+			xfs_trans_brelse(NULL, bp);
+		}
+	}
+
+	/*
+	 * Here with bp and block set to the leftmost leaf node in the tree.
+	 */
+	i = 0;
+
+	/*
+	 * Loop over all leaf nodes checking that all extents are in the right order.
+	 */
+	for (;;) {
+		xfs_fsblock_t	nextbno;
+		xfs_extnum_t	num_recs;
+
+
+		num_recs = xfs_btree_get_numrecs(block);
+
+		/*
+		 * Read-ahead the next leaf block, if any.
+		 */
+
+		nextbno = be64_to_cpu(block->bb_u.l.bb_rightsib);
+
+		/*
+		 * Check all the extents to make sure they are OK.
+		 * If we had a previous block, the last entry should
+		 * conform with the first entry in this one.
+		 */
+
+		ep = XFS_BMBT_REC_ADDR(mp, block, 1);
+		if (i) {
+			ASSERT(xfs_bmbt_disk_get_startoff(&last) +
+			       xfs_bmbt_disk_get_blockcount(&last) <=
+			       xfs_bmbt_disk_get_startoff(ep));
+		}
+		for (j = 1; j < num_recs; j++) {
+			nextp = XFS_BMBT_REC_ADDR(mp, block, j + 1);
+			ASSERT(xfs_bmbt_disk_get_startoff(ep) +
+			       xfs_bmbt_disk_get_blockcount(ep) <=
+			       xfs_bmbt_disk_get_startoff(nextp));
+			ep = nextp;
+		}
+
+		last = *ep;
+		i += num_recs;
+		if (bp_release) {
+			bp_release = 0;
+			xfs_trans_brelse(NULL, bp);
+		}
+		bno = nextbno;
+		/*
+		 * If we've reached the end, stop.
+		 */
+		if (bno == NULLFSBLOCK)
+			break;
+
+		bp = xfs_bmap_get_bp(cur, XFS_FSB_TO_DADDR(mp, bno));
+		if (bp) {
+			bp_release = 0;
+		} else {
+			bp_release = 1;
+		}
+		if (!bp && (error = xfs_btree_read_bufl(mp, NULL, bno, 0, &bp,
+				XFS_BMAP_BTREE_REF)))
+			goto error_norelse;
+		block = XFS_BUF_TO_BLOCK(bp);
+	}
+	if (bp_release) {
+		bp_release = 0;
+		xfs_trans_brelse(NULL, bp);
+	}
+	return;
+
+error0:
+	xfs_warn(mp, "%s: at error0", __func__);
+	if (bp_release)
+		xfs_trans_brelse(NULL, bp);
+error_norelse:
+	xfs_warn(mp, "%s: BAD after btree leaves for %d extents",
+		__func__, i);
+	panic("%s: CORRUPTED BTREE OR SOMETHING", __func__);
+	return;
+}
+#endif
+
+/*
+ * Count fsblocks of the given fork.
+ */
+int						/* error */
+xfs_bmap_count_blocks(
+	xfs_trans_t		*tp,		/* transaction pointer */
+	xfs_inode_t		*ip,		/* incore inode */
+	int			whichfork,	/* data or attr fork */
+	int			*count)		/* out: count of blocks */
+{
+	struct xfs_btree_block	*block;	/* current btree block */
+	xfs_fsblock_t		bno;	/* block # of "block" */
+	xfs_ifork_t		*ifp;	/* fork structure */
+	int			level;	/* btree level, for checking */
+	xfs_mount_t		*mp;	/* file system mount structure */
+	__be64			*pp;	/* pointer to block address */
+
+	bno = NULLFSBLOCK;
+	mp = ip->i_mount;
+	ifp = XFS_IFORK_PTR(ip, whichfork);
+	if ( XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_EXTENTS ) {
+		xfs_bmap_count_leaves(ifp, 0,
+			ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t),
+			count);
+		return 0;
+	}
+
+	/*
+	 * Root level must use BMAP_BROOT_PTR_ADDR macro to get ptr out.
+	 */
+	block = ifp->if_broot;
+	level = be16_to_cpu(block->bb_level);
+	ASSERT(level > 0);
+	pp = XFS_BMAP_BROOT_PTR_ADDR(mp, block, 1, ifp->if_broot_bytes);
+	bno = be64_to_cpu(*pp);
+	ASSERT(bno != NULLDFSBNO);
+	ASSERT(XFS_FSB_TO_AGNO(mp, bno) < mp->m_sb.sb_agcount);
+	ASSERT(XFS_FSB_TO_AGBNO(mp, bno) < mp->m_sb.sb_agblocks);
+
+	if (unlikely(xfs_bmap_count_tree(mp, tp, ifp, bno, level, count) < 0)) {
+		XFS_ERROR_REPORT("xfs_bmap_count_blocks(2)", XFS_ERRLEVEL_LOW,
+				 mp);
+		return XFS_ERROR(EFSCORRUPTED);
+	}
+
+	return 0;
+}
+
+/*
+ * Recursively walks each level of a btree
+ * to count total fsblocks is use.
+ */
+STATIC int                                     /* error */
+xfs_bmap_count_tree(
+	xfs_mount_t     *mp,            /* file system mount point */
+	xfs_trans_t     *tp,            /* transaction pointer */
+	xfs_ifork_t	*ifp,		/* inode fork pointer */
+	xfs_fsblock_t   blockno,	/* file system block number */
+	int             levelin,	/* level in btree */
+	int		*count)		/* Count of blocks */
+{
+	int			error;
+	xfs_buf_t		*bp, *nbp;
+	int			level = levelin;
+	__be64			*pp;
+	xfs_fsblock_t           bno = blockno;
+	xfs_fsblock_t		nextbno;
+	struct xfs_btree_block	*block, *nextblock;
+	int			numrecs;
+
+	if ((error = xfs_btree_read_bufl(mp, tp, bno, 0, &bp, XFS_BMAP_BTREE_REF)))
+		return error;
+	*count += 1;
+	block = XFS_BUF_TO_BLOCK(bp);
+
+	if (--level) {
+		/* Not at node above leaves, count this level of nodes */
+		nextbno = be64_to_cpu(block->bb_u.l.bb_rightsib);
+		while (nextbno != NULLFSBLOCK) {
+			if ((error = xfs_btree_read_bufl(mp, tp, nextbno,
+				0, &nbp, XFS_BMAP_BTREE_REF)))
+				return error;
+			*count += 1;
+			nextblock = XFS_BUF_TO_BLOCK(nbp);
+			nextbno = be64_to_cpu(nextblock->bb_u.l.bb_rightsib);
+			xfs_trans_brelse(tp, nbp);
+		}
+
+		/* Dive to the next level */
+		pp = XFS_BMBT_PTR_ADDR(mp, block, 1, mp->m_bmap_dmxr[1]);
+		bno = be64_to_cpu(*pp);
+		if (unlikely((error =
+		     xfs_bmap_count_tree(mp, tp, ifp, bno, level, count)) < 0)) {
+			xfs_trans_brelse(tp, bp);
+			XFS_ERROR_REPORT("xfs_bmap_count_tree(1)",
+					 XFS_ERRLEVEL_LOW, mp);
+			return XFS_ERROR(EFSCORRUPTED);
+		}
+		xfs_trans_brelse(tp, bp);
+	} else {
+		/* count all level 1 nodes and their leaves */
+		for (;;) {
+			nextbno = be64_to_cpu(block->bb_u.l.bb_rightsib);
+			numrecs = be16_to_cpu(block->bb_numrecs);
+			xfs_bmap_disk_count_leaves(mp, block, numrecs, count);
+			xfs_trans_brelse(tp, bp);
+			if (nextbno == NULLFSBLOCK)
+				break;
+			bno = nextbno;
+			if ((error = xfs_btree_read_bufl(mp, tp, bno, 0, &bp,
+				XFS_BMAP_BTREE_REF)))
+				return error;
+			*count += 1;
+			block = XFS_BUF_TO_BLOCK(bp);
+		}
+	}
+	return 0;
+}
+
+/*
+ * Count leaf blocks given a range of extent records.
+ */
+STATIC void
+xfs_bmap_count_leaves(
+	xfs_ifork_t		*ifp,
+	xfs_extnum_t		idx,
+	int			numrecs,
+	int			*count)
+{
+	int		b;
+
+	for (b = 0; b < numrecs; b++) {
+		xfs_bmbt_rec_host_t *frp = xfs_iext_get_ext(ifp, idx + b);
+		*count += xfs_bmbt_get_blockcount(frp);
+	}
+}
+
+/*
+ * Count leaf blocks given a range of extent records originally
+ * in btree format.
+ */
+STATIC void
+xfs_bmap_disk_count_leaves(
+	struct xfs_mount	*mp,
+	struct xfs_btree_block	*block,
+	int			numrecs,
+	int			*count)
+{
+	int		b;
+	xfs_bmbt_rec_t	*frp;
+
+	for (b = 1; b <= numrecs; b++) {
+		frp = XFS_BMBT_REC_ADDR(mp, block, b);
+		*count += xfs_bmbt_disk_get_blockcount(frp);
+	}
+}
+
+/*
+ * dead simple method of punching delalyed allocation blocks from a range in
+ * the inode. Walks a block at a time so will be slow, but is only executed in
+ * rare error cases so the overhead is not critical. This will alays punch out
+ * both the start and end blocks, even if the ranges only partially overlap
+ * them, so it is up to the caller to ensure that partial blocks are not
+ * passed in.
+ */
+int
+xfs_bmap_punch_delalloc_range(
+	struct xfs_inode	*ip,
+	xfs_fileoff_t		start_fsb,
+	xfs_fileoff_t		length)
+{
+	xfs_fileoff_t		remaining = length;
+	int			error = 0;
+
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
+
+	do {
+		int		done;
+		xfs_bmbt_irec_t	imap;
+		int		nimaps = 1;
+		xfs_fsblock_t	firstblock;
+		xfs_bmap_free_t flist;
+
+		/*
+		 * Map the range first and check that it is a delalloc extent
+		 * before trying to unmap the range. Otherwise we will be
+		 * trying to remove a real extent (which requires a
+		 * transaction) or a hole, which is probably a bad idea...
+		 */
+		error = xfs_bmapi(NULL, ip, start_fsb, 1,
+				XFS_BMAPI_ENTIRE,  NULL, 0, &imap,
+				&nimaps, NULL);
+
+		if (error) {
+			/* something screwed, just bail */
+			if (!XFS_FORCED_SHUTDOWN(ip->i_mount)) {
+				xfs_alert(ip->i_mount,
+			"Failed delalloc mapping lookup ino %lld fsb %lld.",
+						ip->i_ino, start_fsb);
+			}
+			break;
+		}
+		if (!nimaps) {
+			/* nothing there */
+			goto next_block;
+		}
+		if (imap.br_startblock != DELAYSTARTBLOCK) {
+			/* been converted, ignore */
+			goto next_block;
+		}
+		WARN_ON(imap.br_blockcount == 0);
+
+		/*
+		 * Note: while we initialise the firstblock/flist pair, they
+		 * should never be used because blocks should never be
+		 * allocated or freed for a delalloc extent and hence we need
+		 * don't cancel or finish them after the xfs_bunmapi() call.
+		 */
+		xfs_bmap_init(&flist, &firstblock);
+		error = xfs_bunmapi(NULL, ip, start_fsb, 1, 0, 1, &firstblock,
+					&flist, &done);
+		if (error)
+			break;
+
+		ASSERT(!flist.xbf_count && !flist.xbf_first);
+next_block:
+		start_fsb++;
+		remaining--;
+	} while(remaining > 0);
+
+	return error;
+}
diff --git a/fs/xfs/xfs_bmap.h b/fs/xfs/xfs_bmap.h
new file mode 100644
index 00000000..c62234bd
--- /dev/null
+++ b/fs/xfs/xfs_bmap.h
@@ -0,0 +1,402 @@
+/*
+ * Copyright (c) 2000-2006 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_BMAP_H__
+#define	__XFS_BMAP_H__
+
+struct getbmap;
+struct xfs_bmbt_irec;
+struct xfs_ifork;
+struct xfs_inode;
+struct xfs_mount;
+struct xfs_trans;
+
+extern kmem_zone_t	*xfs_bmap_free_item_zone;
+
+/*
+ * List of extents to be free "later".
+ * The list is kept sorted on xbf_startblock.
+ */
+typedef struct xfs_bmap_free_item
+{
+	xfs_fsblock_t		xbfi_startblock;/* starting fs block number */
+	xfs_extlen_t		xbfi_blockcount;/* number of blocks in extent */
+	struct xfs_bmap_free_item *xbfi_next;	/* link to next entry */
+} xfs_bmap_free_item_t;
+
+/*
+ * Header for free extent list.
+ *
+ * xbf_low is used by the allocator to activate the lowspace algorithm -
+ * when free space is running low the extent allocator may choose to
+ * allocate an extent from an AG without leaving sufficient space for
+ * a btree split when inserting the new extent.  In this case the allocator
+ * will enable the lowspace algorithm which is supposed to allow further
+ * allocations (such as btree splits and newroots) to allocate from
+ * sequential AGs.  In order to avoid locking AGs out of order the lowspace
+ * algorithm will start searching for free space from AG 0.  If the correct
+ * transaction reservations have been made then this algorithm will eventually
+ * find all the space it needs.
+ */
+typedef	struct xfs_bmap_free
+{
+	xfs_bmap_free_item_t	*xbf_first;	/* list of to-be-free extents */
+	int			xbf_count;	/* count of items on list */
+	int			xbf_low;	/* alloc in low mode */
+} xfs_bmap_free_t;
+
+#define	XFS_BMAP_MAX_NMAP	4
+
+/*
+ * Flags for xfs_bmapi
+ */
+#define	XFS_BMAPI_WRITE		0x001	/* write operation: allocate space */
+#define XFS_BMAPI_DELAY		0x002	/* delayed write operation */
+#define XFS_BMAPI_ENTIRE	0x004	/* return entire extent, not trimmed */
+#define XFS_BMAPI_METADATA	0x008	/* mapping metadata not user data */
+#define XFS_BMAPI_ATTRFORK	0x010	/* use attribute fork not data */
+#define	XFS_BMAPI_PREALLOC	0x040	/* preallocation op: unwritten space */
+#define	XFS_BMAPI_IGSTATE	0x080	/* Ignore state - */
+					/* combine contig. space */
+#define	XFS_BMAPI_CONTIG	0x100	/* must allocate only one extent */
+/*
+ * unwritten extent conversion - this needs write cache flushing and no additional
+ * allocation alignments. When specified with XFS_BMAPI_PREALLOC it converts
+ * from written to unwritten, otherwise convert from unwritten to written.
+ */
+#define XFS_BMAPI_CONVERT	0x200
+
+#define XFS_BMAPI_FLAGS \
+	{ XFS_BMAPI_WRITE,	"WRITE" }, \
+	{ XFS_BMAPI_DELAY,	"DELAY" }, \
+	{ XFS_BMAPI_ENTIRE,	"ENTIRE" }, \
+	{ XFS_BMAPI_METADATA,	"METADATA" }, \
+	{ XFS_BMAPI_ATTRFORK,	"ATTRFORK" }, \
+	{ XFS_BMAPI_PREALLOC,	"PREALLOC" }, \
+	{ XFS_BMAPI_IGSTATE,	"IGSTATE" }, \
+	{ XFS_BMAPI_CONTIG,	"CONTIG" }, \
+	{ XFS_BMAPI_CONVERT,	"CONVERT" }
+
+
+static inline int xfs_bmapi_aflag(int w)
+{
+	return (w == XFS_ATTR_FORK ? XFS_BMAPI_ATTRFORK : 0);
+}
+
+/*
+ * Special values for xfs_bmbt_irec_t br_startblock field.
+ */
+#define	DELAYSTARTBLOCK		((xfs_fsblock_t)-1LL)
+#define	HOLESTARTBLOCK		((xfs_fsblock_t)-2LL)
+
+static inline void xfs_bmap_init(xfs_bmap_free_t *flp, xfs_fsblock_t *fbp)
+{
+	((flp)->xbf_first = NULL, (flp)->xbf_count = 0, \
+		(flp)->xbf_low = 0, *(fbp) = NULLFSBLOCK);
+}
+
+/*
+ * Argument structure for xfs_bmap_alloc.
+ */
+typedef struct xfs_bmalloca {
+	xfs_fsblock_t		firstblock; /* i/o first block allocated */
+	xfs_fsblock_t		rval;	/* starting block of new extent */
+	xfs_fileoff_t		off;	/* offset in file filling in */
+	struct xfs_trans	*tp;	/* transaction pointer */
+	struct xfs_inode	*ip;	/* incore inode pointer */
+	struct xfs_bmbt_irec	*prevp;	/* extent before the new one */
+	struct xfs_bmbt_irec	*gotp;	/* extent after, or delayed */
+	xfs_extlen_t		alen;	/* i/o length asked/allocated */
+	xfs_extlen_t		total;	/* total blocks needed for xaction */
+	xfs_extlen_t		minlen;	/* minimum allocation size (blocks) */
+	xfs_extlen_t		minleft; /* amount must be left after alloc */
+	char			eof;	/* set if allocating past last extent */
+	char			wasdel;	/* replacing a delayed allocation */
+	char			userdata;/* set if is user data */
+	char			low;	/* low on space, using seq'l ags */
+	char			aeof;	/* allocated space at eof */
+	char			conv;	/* overwriting unwritten extents */
+} xfs_bmalloca_t;
+
+/*
+ * Flags for xfs_bmap_add_extent*.
+ */
+#define BMAP_LEFT_CONTIG	(1 << 0)
+#define BMAP_RIGHT_CONTIG	(1 << 1)
+#define BMAP_LEFT_FILLING	(1 << 2)
+#define BMAP_RIGHT_FILLING	(1 << 3)
+#define BMAP_LEFT_DELAY		(1 << 4)
+#define BMAP_RIGHT_DELAY	(1 << 5)
+#define BMAP_LEFT_VALID		(1 << 6)
+#define BMAP_RIGHT_VALID	(1 << 7)
+#define BMAP_ATTRFORK		(1 << 8)
+
+#define XFS_BMAP_EXT_FLAGS \
+	{ BMAP_LEFT_CONTIG,	"LC" }, \
+	{ BMAP_RIGHT_CONTIG,	"RC" }, \
+	{ BMAP_LEFT_FILLING,	"LF" }, \
+	{ BMAP_RIGHT_FILLING,	"RF" }, \
+	{ BMAP_ATTRFORK,	"ATTR" }
+
+/*
+ * Add bmap trace insert entries for all the contents of the extent list.
+ *
+ * Quite excessive tracing.  Only do this for debug builds.
+ */
+#if defined(__KERNEL) && defined(DEBUG)
+void
+xfs_bmap_trace_exlist(
+	struct xfs_inode	*ip,		/* incore inode pointer */
+	xfs_extnum_t		cnt,		/* count of entries in list */
+	int			whichfork,
+	unsigned long		caller_ip);	/* data or attr fork */
+#define	XFS_BMAP_TRACE_EXLIST(ip,c,w)	\
+	xfs_bmap_trace_exlist(ip,c,w, _THIS_IP_)
+#else
+#define	XFS_BMAP_TRACE_EXLIST(ip,c,w)
+#endif
+
+/*
+ * Convert inode from non-attributed to attributed.
+ * Must not be in a transaction, ip must not be locked.
+ */
+int					/* error code */
+xfs_bmap_add_attrfork(
+	struct xfs_inode	*ip,	/* incore inode pointer */
+	int			size,	/* space needed for new attribute */
+	int			rsvd);	/* flag for reserved block allocation */
+
+/*
+ * Add the extent to the list of extents to be free at transaction end.
+ * The list is maintained sorted (by block number).
+ */
+void
+xfs_bmap_add_free(
+	xfs_fsblock_t		bno,		/* fs block number of extent */
+	xfs_filblks_t		len,		/* length of extent */
+	xfs_bmap_free_t		*flist,		/* list of extents */
+	struct xfs_mount	*mp);		/* mount point structure */
+
+/*
+ * Routine to clean up the free list data structure when
+ * an error occurs during a transaction.
+ */
+void
+xfs_bmap_cancel(
+	xfs_bmap_free_t		*flist);	/* free list to clean up */
+
+/*
+ * Compute and fill in the value of the maximum depth of a bmap btree
+ * in this filesystem.  Done once, during mount.
+ */
+void
+xfs_bmap_compute_maxlevels(
+	struct xfs_mount	*mp,	/* file system mount structure */
+	int			whichfork);	/* data or attr fork */
+
+/*
+ * Returns the file-relative block number of the first unused block in the file.
+ * This is the lowest-address hole if the file has holes, else the first block
+ * past the end of file.
+ */
+int						/* error */
+xfs_bmap_first_unused(
+	struct xfs_trans	*tp,		/* transaction pointer */
+	struct xfs_inode	*ip,		/* incore inode */
+	xfs_extlen_t		len,		/* size of hole to find */
+	xfs_fileoff_t		*unused,	/* unused block num */
+	int			whichfork);	/* data or attr fork */
+
+/*
+ * Returns the file-relative block number of the last block + 1 before
+ * last_block (input value) in the file.
+ * This is not based on i_size, it is based on the extent list.
+ * Returns 0 for local files, as they do not have an extent list.
+ */
+int						/* error */
+xfs_bmap_last_before(
+	struct xfs_trans	*tp,		/* transaction pointer */
+	struct xfs_inode	*ip,		/* incore inode */
+	xfs_fileoff_t		*last_block,	/* last block */
+	int			whichfork);	/* data or attr fork */
+
+/*
+ * Returns the file-relative block number of the first block past eof in
+ * the file.  This is not based on i_size, it is based on the extent list.
+ * Returns 0 for local files, as they do not have an extent list.
+ */
+int						/* error */
+xfs_bmap_last_offset(
+	struct xfs_trans	*tp,		/* transaction pointer */
+	struct xfs_inode	*ip,		/* incore inode */
+	xfs_fileoff_t		*unused,	/* last block num */
+	int			whichfork);	/* data or attr fork */
+
+/*
+ * Returns whether the selected fork of the inode has exactly one
+ * block or not.  For the data fork we check this matches di_size,
+ * implying the file's range is 0..bsize-1.
+ */
+int
+xfs_bmap_one_block(
+	struct xfs_inode	*ip,		/* incore inode */
+	int			whichfork);	/* data or attr fork */
+
+/*
+ * Read in the extents to iu_extents.
+ * All inode fields are set up by caller, we just traverse the btree
+ * and copy the records in.
+ */
+int						/* error */
+xfs_bmap_read_extents(
+	struct xfs_trans	*tp,		/* transaction pointer */
+	struct xfs_inode	*ip,		/* incore inode */
+	int			whichfork);	/* data or attr fork */
+
+/*
+ * Map file blocks to filesystem blocks.
+ * File range is given by the bno/len pair.
+ * Adds blocks to file if a write ("flags & XFS_BMAPI_WRITE" set)
+ * into a hole or past eof.
+ * Only allocates blocks from a single allocation group,
+ * to avoid locking problems.
+ * The returned value in "firstblock" from the first call in a transaction
+ * must be remembered and presented to subsequent calls in "firstblock".
+ * An upper bound for the number of blocks to be allocated is supplied to
+ * the first call in "total"; if no allocation group has that many free
+ * blocks then the call will fail (return NULLFSBLOCK in "firstblock").
+ */
+int						/* error */
+xfs_bmapi(
+	struct xfs_trans	*tp,		/* transaction pointer */
+	struct xfs_inode	*ip,		/* incore inode */
+	xfs_fileoff_t		bno,		/* starting file offs. mapped */
+	xfs_filblks_t		len,		/* length to map in file */
+	int			flags,		/* XFS_BMAPI_... */
+	xfs_fsblock_t		*firstblock,	/* first allocated block
+						   controls a.g. for allocs */
+	xfs_extlen_t		total,		/* total blocks needed */
+	struct xfs_bmbt_irec	*mval,		/* output: map values */
+	int			*nmap,		/* i/o: mval size/count */
+	xfs_bmap_free_t		*flist);	/* i/o: list extents to free */
+
+/*
+ * Map file blocks to filesystem blocks, simple version.
+ * One block only, read-only.
+ * For flags, only the XFS_BMAPI_ATTRFORK flag is examined.
+ * For the other flag values, the effect is as if XFS_BMAPI_METADATA
+ * was set and all the others were clear.
+ */
+int						/* error */
+xfs_bmapi_single(
+	struct xfs_trans	*tp,		/* transaction pointer */
+	struct xfs_inode	*ip,		/* incore inode */
+	int			whichfork,	/* data or attr fork */
+	xfs_fsblock_t		*fsb,		/* output: mapped block */
+	xfs_fileoff_t		bno);		/* starting file offs. mapped */
+
+/*
+ * Unmap (remove) blocks from a file.
+ * If nexts is nonzero then the number of extents to remove is limited to
+ * that value.  If not all extents in the block range can be removed then
+ * *done is set.
+ */
+int						/* error */
+xfs_bunmapi(
+	struct xfs_trans	*tp,		/* transaction pointer */
+	struct xfs_inode	*ip,		/* incore inode */
+	xfs_fileoff_t		bno,		/* starting offset to unmap */
+	xfs_filblks_t		len,		/* length to unmap in file */
+	int			flags,		/* XFS_BMAPI_... */
+	xfs_extnum_t		nexts,		/* number of extents max */
+	xfs_fsblock_t		*firstblock,	/* first allocated block
+						   controls a.g. for allocs */
+	xfs_bmap_free_t		*flist,		/* i/o: list extents to free */
+	int			*done);		/* set if not done yet */
+
+/*
+ * Check an extent list, which has just been read, for
+ * any bit in the extent flag field.
+ */
+int
+xfs_check_nostate_extents(
+	struct xfs_ifork	*ifp,
+	xfs_extnum_t		idx,
+	xfs_extnum_t		num);
+
+uint
+xfs_default_attroffset(
+	struct xfs_inode	*ip);
+
+#ifdef __KERNEL__
+
+/*
+ * Routine to be called at transaction's end by xfs_bmapi, xfs_bunmapi
+ * caller.  Frees all the extents that need freeing, which must be done
+ * last due to locking considerations.
+ *
+ * Return 1 if the given transaction was committed and a new one allocated,
+ * and 0 otherwise.
+ */
+int						/* error */
+xfs_bmap_finish(
+	struct xfs_trans	**tp,		/* transaction pointer addr */
+	xfs_bmap_free_t		*flist,		/* i/o: list extents to free */
+	int			*committed);	/* xact committed or not */
+
+/* bmap to userspace formatter - copy to user & advance pointer */
+typedef int (*xfs_bmap_format_t)(void **, struct getbmapx *, int *);
+
+/*
+ * Get inode's extents as described in bmv, and format for output.
+ */
+int						/* error code */
+xfs_getbmap(
+	xfs_inode_t		*ip,
+	struct getbmapx		*bmv,		/* user bmap structure */
+	xfs_bmap_format_t	formatter,	/* format to user */
+	void			*arg);		/* formatter arg */
+
+/*
+ * Check if the endoff is outside the last extent. If so the caller will grow
+ * the allocation to a stripe unit boundary
+ */
+int
+xfs_bmap_eof(
+	struct xfs_inode        *ip,
+	xfs_fileoff_t           endoff,
+	int                     whichfork,
+	int                     *eof);
+
+/*
+ * Count fsblocks of the given fork.
+ */
+int
+xfs_bmap_count_blocks(
+	xfs_trans_t		*tp,
+	struct xfs_inode	*ip,
+	int			whichfork,
+	int			*count);
+
+int
+xfs_bmap_punch_delalloc_range(
+	struct xfs_inode	*ip,
+	xfs_fileoff_t		start_fsb,
+	xfs_fileoff_t		length);
+#endif	/* __KERNEL__ */
+
+#endif	/* __XFS_BMAP_H__ */
diff --git a/fs/xfs/xfs_bmap_btree.c b/fs/xfs/xfs_bmap_btree.c
new file mode 100644
index 00000000..87d3c10b
--- /dev/null
+++ b/fs/xfs/xfs_bmap_btree.c
@@ -0,0 +1,919 @@
+/*
+ * Copyright (c) 2000-2003,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_mount.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_alloc_btree.h"
+#include "xfs_ialloc_btree.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_inode_item.h"
+#include "xfs_alloc.h"
+#include "xfs_btree.h"
+#include "xfs_btree_trace.h"
+#include "xfs_itable.h"
+#include "xfs_bmap.h"
+#include "xfs_error.h"
+#include "xfs_quota.h"
+
+/*
+ * Determine the extent state.
+ */
+/* ARGSUSED */
+STATIC xfs_exntst_t
+xfs_extent_state(
+	xfs_filblks_t		blks,
+	int			extent_flag)
+{
+	if (extent_flag) {
+		ASSERT(blks != 0);	/* saved for DMIG */
+		return XFS_EXT_UNWRITTEN;
+	}
+	return XFS_EXT_NORM;
+}
+
+/*
+ * Convert on-disk form of btree root to in-memory form.
+ */
+void
+xfs_bmdr_to_bmbt(
+	struct xfs_mount	*mp,
+	xfs_bmdr_block_t	*dblock,
+	int			dblocklen,
+	struct xfs_btree_block	*rblock,
+	int			rblocklen)
+{
+	int			dmxr;
+	xfs_bmbt_key_t		*fkp;
+	__be64			*fpp;
+	xfs_bmbt_key_t		*tkp;
+	__be64			*tpp;
+
+	rblock->bb_magic = cpu_to_be32(XFS_BMAP_MAGIC);
+	rblock->bb_level = dblock->bb_level;
+	ASSERT(be16_to_cpu(rblock->bb_level) > 0);
+	rblock->bb_numrecs = dblock->bb_numrecs;
+	rblock->bb_u.l.bb_leftsib = cpu_to_be64(NULLDFSBNO);
+	rblock->bb_u.l.bb_rightsib = cpu_to_be64(NULLDFSBNO);
+	dmxr = xfs_bmdr_maxrecs(mp, dblocklen, 0);
+	fkp = XFS_BMDR_KEY_ADDR(dblock, 1);
+	tkp = XFS_BMBT_KEY_ADDR(mp, rblock, 1);
+	fpp = XFS_BMDR_PTR_ADDR(dblock, 1, dmxr);
+	tpp = XFS_BMAP_BROOT_PTR_ADDR(mp, rblock, 1, rblocklen);
+	dmxr = be16_to_cpu(dblock->bb_numrecs);
+	memcpy(tkp, fkp, sizeof(*fkp) * dmxr);
+	memcpy(tpp, fpp, sizeof(*fpp) * dmxr);
+}
+
+/*
+ * Convert a compressed bmap extent record to an uncompressed form.
+ * This code must be in sync with the routines xfs_bmbt_get_startoff,
+ * xfs_bmbt_get_startblock, xfs_bmbt_get_blockcount and xfs_bmbt_get_state.
+ */
+STATIC void
+__xfs_bmbt_get_all(
+		__uint64_t l0,
+		__uint64_t l1,
+		xfs_bmbt_irec_t *s)
+{
+	int	ext_flag;
+	xfs_exntst_t st;
+
+	ext_flag = (int)(l0 >> (64 - BMBT_EXNTFLAG_BITLEN));
+	s->br_startoff = ((xfs_fileoff_t)l0 &
+			   xfs_mask64lo(64 - BMBT_EXNTFLAG_BITLEN)) >> 9;
+#if XFS_BIG_BLKNOS
+	s->br_startblock = (((xfs_fsblock_t)l0 & xfs_mask64lo(9)) << 43) |
+			   (((xfs_fsblock_t)l1) >> 21);
+#else
+#ifdef DEBUG
+	{
+		xfs_dfsbno_t	b;
+
+		b = (((xfs_dfsbno_t)l0 & xfs_mask64lo(9)) << 43) |
+		    (((xfs_dfsbno_t)l1) >> 21);
+		ASSERT((b >> 32) == 0 || isnulldstartblock(b));
+		s->br_startblock = (xfs_fsblock_t)b;
+	}
+#else	/* !DEBUG */
+	s->br_startblock = (xfs_fsblock_t)(((xfs_dfsbno_t)l1) >> 21);
+#endif	/* DEBUG */
+#endif	/* XFS_BIG_BLKNOS */
+	s->br_blockcount = (xfs_filblks_t)(l1 & xfs_mask64lo(21));
+	/* This is xfs_extent_state() in-line */
+	if (ext_flag) {
+		ASSERT(s->br_blockcount != 0);	/* saved for DMIG */
+		st = XFS_EXT_UNWRITTEN;
+	} else
+		st = XFS_EXT_NORM;
+	s->br_state = st;
+}
+
+void
+xfs_bmbt_get_all(
+	xfs_bmbt_rec_host_t *r,
+	xfs_bmbt_irec_t *s)
+{
+	__xfs_bmbt_get_all(r->l0, r->l1, s);
+}
+
+/*
+ * Extract the blockcount field from an in memory bmap extent record.
+ */
+xfs_filblks_t
+xfs_bmbt_get_blockcount(
+	xfs_bmbt_rec_host_t	*r)
+{
+	return (xfs_filblks_t)(r->l1 & xfs_mask64lo(21));
+}
+
+/*
+ * Extract the startblock field from an in memory bmap extent record.
+ */
+xfs_fsblock_t
+xfs_bmbt_get_startblock(
+	xfs_bmbt_rec_host_t	*r)
+{
+#if XFS_BIG_BLKNOS
+	return (((xfs_fsblock_t)r->l0 & xfs_mask64lo(9)) << 43) |
+	       (((xfs_fsblock_t)r->l1) >> 21);
+#else
+#ifdef DEBUG
+	xfs_dfsbno_t	b;
+
+	b = (((xfs_dfsbno_t)r->l0 & xfs_mask64lo(9)) << 43) |
+	    (((xfs_dfsbno_t)r->l1) >> 21);
+	ASSERT((b >> 32) == 0 || isnulldstartblock(b));
+	return (xfs_fsblock_t)b;
+#else	/* !DEBUG */
+	return (xfs_fsblock_t)(((xfs_dfsbno_t)r->l1) >> 21);
+#endif	/* DEBUG */
+#endif	/* XFS_BIG_BLKNOS */
+}
+
+/*
+ * Extract the startoff field from an in memory bmap extent record.
+ */
+xfs_fileoff_t
+xfs_bmbt_get_startoff(
+	xfs_bmbt_rec_host_t	*r)
+{
+	return ((xfs_fileoff_t)r->l0 &
+		 xfs_mask64lo(64 - BMBT_EXNTFLAG_BITLEN)) >> 9;
+}
+
+xfs_exntst_t
+xfs_bmbt_get_state(
+	xfs_bmbt_rec_host_t	*r)
+{
+	int	ext_flag;
+
+	ext_flag = (int)((r->l0) >> (64 - BMBT_EXNTFLAG_BITLEN));
+	return xfs_extent_state(xfs_bmbt_get_blockcount(r),
+				ext_flag);
+}
+
+/*
+ * Extract the blockcount field from an on disk bmap extent record.
+ */
+xfs_filblks_t
+xfs_bmbt_disk_get_blockcount(
+	xfs_bmbt_rec_t	*r)
+{
+	return (xfs_filblks_t)(be64_to_cpu(r->l1) & xfs_mask64lo(21));
+}
+
+/*
+ * Extract the startoff field from a disk format bmap extent record.
+ */
+xfs_fileoff_t
+xfs_bmbt_disk_get_startoff(
+	xfs_bmbt_rec_t	*r)
+{
+	return ((xfs_fileoff_t)be64_to_cpu(r->l0) &
+		 xfs_mask64lo(64 - BMBT_EXNTFLAG_BITLEN)) >> 9;
+}
+
+
+/*
+ * Set all the fields in a bmap extent record from the arguments.
+ */
+void
+xfs_bmbt_set_allf(
+	xfs_bmbt_rec_host_t	*r,
+	xfs_fileoff_t		startoff,
+	xfs_fsblock_t		startblock,
+	xfs_filblks_t		blockcount,
+	xfs_exntst_t		state)
+{
+	int		extent_flag = (state == XFS_EXT_NORM) ? 0 : 1;
+
+	ASSERT(state == XFS_EXT_NORM || state == XFS_EXT_UNWRITTEN);
+	ASSERT((startoff & xfs_mask64hi(64-BMBT_STARTOFF_BITLEN)) == 0);
+	ASSERT((blockcount & xfs_mask64hi(64-BMBT_BLOCKCOUNT_BITLEN)) == 0);
+
+#if XFS_BIG_BLKNOS
+	ASSERT((startblock & xfs_mask64hi(64-BMBT_STARTBLOCK_BITLEN)) == 0);
+
+	r->l0 = ((xfs_bmbt_rec_base_t)extent_flag << 63) |
+		((xfs_bmbt_rec_base_t)startoff << 9) |
+		((xfs_bmbt_rec_base_t)startblock >> 43);
+	r->l1 = ((xfs_bmbt_rec_base_t)startblock << 21) |
+		((xfs_bmbt_rec_base_t)blockcount &
+		(xfs_bmbt_rec_base_t)xfs_mask64lo(21));
+#else	/* !XFS_BIG_BLKNOS */
+	if (isnullstartblock(startblock)) {
+		r->l0 = ((xfs_bmbt_rec_base_t)extent_flag << 63) |
+			((xfs_bmbt_rec_base_t)startoff << 9) |
+			 (xfs_bmbt_rec_base_t)xfs_mask64lo(9);
+		r->l1 = xfs_mask64hi(11) |
+			  ((xfs_bmbt_rec_base_t)startblock << 21) |
+			  ((xfs_bmbt_rec_base_t)blockcount &
+			   (xfs_bmbt_rec_base_t)xfs_mask64lo(21));
+	} else {
+		r->l0 = ((xfs_bmbt_rec_base_t)extent_flag << 63) |
+			((xfs_bmbt_rec_base_t)startoff << 9);
+		r->l1 = ((xfs_bmbt_rec_base_t)startblock << 21) |
+			 ((xfs_bmbt_rec_base_t)blockcount &
+			 (xfs_bmbt_rec_base_t)xfs_mask64lo(21));
+	}
+#endif	/* XFS_BIG_BLKNOS */
+}
+
+/*
+ * Set all the fields in a bmap extent record from the uncompressed form.
+ */
+void
+xfs_bmbt_set_all(
+	xfs_bmbt_rec_host_t *r,
+	xfs_bmbt_irec_t	*s)
+{
+	xfs_bmbt_set_allf(r, s->br_startoff, s->br_startblock,
+			     s->br_blockcount, s->br_state);
+}
+
+
+/*
+ * Set all the fields in a disk format bmap extent record from the arguments.
+ */
+void
+xfs_bmbt_disk_set_allf(
+	xfs_bmbt_rec_t		*r,
+	xfs_fileoff_t		startoff,
+	xfs_fsblock_t		startblock,
+	xfs_filblks_t		blockcount,
+	xfs_exntst_t		state)
+{
+	int			extent_flag = (state == XFS_EXT_NORM) ? 0 : 1;
+
+	ASSERT(state == XFS_EXT_NORM || state == XFS_EXT_UNWRITTEN);
+	ASSERT((startoff & xfs_mask64hi(64-BMBT_STARTOFF_BITLEN)) == 0);
+	ASSERT((blockcount & xfs_mask64hi(64-BMBT_BLOCKCOUNT_BITLEN)) == 0);
+
+#if XFS_BIG_BLKNOS
+	ASSERT((startblock & xfs_mask64hi(64-BMBT_STARTBLOCK_BITLEN)) == 0);
+
+	r->l0 = cpu_to_be64(
+		((xfs_bmbt_rec_base_t)extent_flag << 63) |
+		 ((xfs_bmbt_rec_base_t)startoff << 9) |
+		 ((xfs_bmbt_rec_base_t)startblock >> 43));
+	r->l1 = cpu_to_be64(
+		((xfs_bmbt_rec_base_t)startblock << 21) |
+		 ((xfs_bmbt_rec_base_t)blockcount &
+		  (xfs_bmbt_rec_base_t)xfs_mask64lo(21)));
+#else	/* !XFS_BIG_BLKNOS */
+	if (isnullstartblock(startblock)) {
+		r->l0 = cpu_to_be64(
+			((xfs_bmbt_rec_base_t)extent_flag << 63) |
+			 ((xfs_bmbt_rec_base_t)startoff << 9) |
+			  (xfs_bmbt_rec_base_t)xfs_mask64lo(9));
+		r->l1 = cpu_to_be64(xfs_mask64hi(11) |
+			  ((xfs_bmbt_rec_base_t)startblock << 21) |
+			  ((xfs_bmbt_rec_base_t)blockcount &
+			   (xfs_bmbt_rec_base_t)xfs_mask64lo(21)));
+	} else {
+		r->l0 = cpu_to_be64(
+			((xfs_bmbt_rec_base_t)extent_flag << 63) |
+			 ((xfs_bmbt_rec_base_t)startoff << 9));
+		r->l1 = cpu_to_be64(
+			((xfs_bmbt_rec_base_t)startblock << 21) |
+			 ((xfs_bmbt_rec_base_t)blockcount &
+			  (xfs_bmbt_rec_base_t)xfs_mask64lo(21)));
+	}
+#endif	/* XFS_BIG_BLKNOS */
+}
+
+/*
+ * Set all the fields in a bmap extent record from the uncompressed form.
+ */
+STATIC void
+xfs_bmbt_disk_set_all(
+	xfs_bmbt_rec_t	*r,
+	xfs_bmbt_irec_t *s)
+{
+	xfs_bmbt_disk_set_allf(r, s->br_startoff, s->br_startblock,
+				  s->br_blockcount, s->br_state);
+}
+
+/*
+ * Set the blockcount field in a bmap extent record.
+ */
+void
+xfs_bmbt_set_blockcount(
+	xfs_bmbt_rec_host_t *r,
+	xfs_filblks_t	v)
+{
+	ASSERT((v & xfs_mask64hi(43)) == 0);
+	r->l1 = (r->l1 & (xfs_bmbt_rec_base_t)xfs_mask64hi(43)) |
+		  (xfs_bmbt_rec_base_t)(v & xfs_mask64lo(21));
+}
+
+/*
+ * Set the startblock field in a bmap extent record.
+ */
+void
+xfs_bmbt_set_startblock(
+	xfs_bmbt_rec_host_t *r,
+	xfs_fsblock_t	v)
+{
+#if XFS_BIG_BLKNOS
+	ASSERT((v & xfs_mask64hi(12)) == 0);
+	r->l0 = (r->l0 & (xfs_bmbt_rec_base_t)xfs_mask64hi(55)) |
+		  (xfs_bmbt_rec_base_t)(v >> 43);
+	r->l1 = (r->l1 & (xfs_bmbt_rec_base_t)xfs_mask64lo(21)) |
+		  (xfs_bmbt_rec_base_t)(v << 21);
+#else	/* !XFS_BIG_BLKNOS */
+	if (isnullstartblock(v)) {
+		r->l0 |= (xfs_bmbt_rec_base_t)xfs_mask64lo(9);
+		r->l1 = (xfs_bmbt_rec_base_t)xfs_mask64hi(11) |
+			  ((xfs_bmbt_rec_base_t)v << 21) |
+			  (r->l1 & (xfs_bmbt_rec_base_t)xfs_mask64lo(21));
+	} else {
+		r->l0 &= ~(xfs_bmbt_rec_base_t)xfs_mask64lo(9);
+		r->l1 = ((xfs_bmbt_rec_base_t)v << 21) |
+			  (r->l1 & (xfs_bmbt_rec_base_t)xfs_mask64lo(21));
+	}
+#endif	/* XFS_BIG_BLKNOS */
+}
+
+/*
+ * Set the startoff field in a bmap extent record.
+ */
+void
+xfs_bmbt_set_startoff(
+	xfs_bmbt_rec_host_t *r,
+	xfs_fileoff_t	v)
+{
+	ASSERT((v & xfs_mask64hi(9)) == 0);
+	r->l0 = (r->l0 & (xfs_bmbt_rec_base_t) xfs_mask64hi(1)) |
+		((xfs_bmbt_rec_base_t)v << 9) |
+		  (r->l0 & (xfs_bmbt_rec_base_t)xfs_mask64lo(9));
+}
+
+/*
+ * Set the extent state field in a bmap extent record.
+ */
+void
+xfs_bmbt_set_state(
+	xfs_bmbt_rec_host_t *r,
+	xfs_exntst_t	v)
+{
+	ASSERT(v == XFS_EXT_NORM || v == XFS_EXT_UNWRITTEN);
+	if (v == XFS_EXT_NORM)
+		r->l0 &= xfs_mask64lo(64 - BMBT_EXNTFLAG_BITLEN);
+	else
+		r->l0 |= xfs_mask64hi(BMBT_EXNTFLAG_BITLEN);
+}
+
+/*
+ * Convert in-memory form of btree root to on-disk form.
+ */
+void
+xfs_bmbt_to_bmdr(
+	struct xfs_mount	*mp,
+	struct xfs_btree_block	*rblock,
+	int			rblocklen,
+	xfs_bmdr_block_t	*dblock,
+	int			dblocklen)
+{
+	int			dmxr;
+	xfs_bmbt_key_t		*fkp;
+	__be64			*fpp;
+	xfs_bmbt_key_t		*tkp;
+	__be64			*tpp;
+
+	ASSERT(be32_to_cpu(rblock->bb_magic) == XFS_BMAP_MAGIC);
+	ASSERT(be64_to_cpu(rblock->bb_u.l.bb_leftsib) == NULLDFSBNO);
+	ASSERT(be64_to_cpu(rblock->bb_u.l.bb_rightsib) == NULLDFSBNO);
+	ASSERT(be16_to_cpu(rblock->bb_level) > 0);
+	dblock->bb_level = rblock->bb_level;
+	dblock->bb_numrecs = rblock->bb_numrecs;
+	dmxr = xfs_bmdr_maxrecs(mp, dblocklen, 0);
+	fkp = XFS_BMBT_KEY_ADDR(mp, rblock, 1);
+	tkp = XFS_BMDR_KEY_ADDR(dblock, 1);
+	fpp = XFS_BMAP_BROOT_PTR_ADDR(mp, rblock, 1, rblocklen);
+	tpp = XFS_BMDR_PTR_ADDR(dblock, 1, dmxr);
+	dmxr = be16_to_cpu(dblock->bb_numrecs);
+	memcpy(tkp, fkp, sizeof(*fkp) * dmxr);
+	memcpy(tpp, fpp, sizeof(*fpp) * dmxr);
+}
+
+/*
+ * Check extent records, which have just been read, for
+ * any bit in the extent flag field. ASSERT on debug
+ * kernels, as this condition should not occur.
+ * Return an error condition (1) if any flags found,
+ * otherwise return 0.
+ */
+
+int
+xfs_check_nostate_extents(
+	xfs_ifork_t		*ifp,
+	xfs_extnum_t		idx,
+	xfs_extnum_t		num)
+{
+	for (; num > 0; num--, idx++) {
+		xfs_bmbt_rec_host_t *ep = xfs_iext_get_ext(ifp, idx);
+		if ((ep->l0 >>
+		     (64 - BMBT_EXNTFLAG_BITLEN)) != 0) {
+			ASSERT(0);
+			return 1;
+		}
+	}
+	return 0;
+}
+
+
+STATIC struct xfs_btree_cur *
+xfs_bmbt_dup_cursor(
+	struct xfs_btree_cur	*cur)
+{
+	struct xfs_btree_cur	*new;
+
+	new = xfs_bmbt_init_cursor(cur->bc_mp, cur->bc_tp,
+			cur->bc_private.b.ip, cur->bc_private.b.whichfork);
+
+	/*
+	 * Copy the firstblock, flist, and flags values,
+	 * since init cursor doesn't get them.
+	 */
+	new->bc_private.b.firstblock = cur->bc_private.b.firstblock;
+	new->bc_private.b.flist = cur->bc_private.b.flist;
+	new->bc_private.b.flags = cur->bc_private.b.flags;
+
+	return new;
+}
+
+STATIC void
+xfs_bmbt_update_cursor(
+	struct xfs_btree_cur	*src,
+	struct xfs_btree_cur	*dst)
+{
+	ASSERT((dst->bc_private.b.firstblock != NULLFSBLOCK) ||
+	       (dst->bc_private.b.ip->i_d.di_flags & XFS_DIFLAG_REALTIME));
+	ASSERT(dst->bc_private.b.flist == src->bc_private.b.flist);
+
+	dst->bc_private.b.allocated += src->bc_private.b.allocated;
+	dst->bc_private.b.firstblock = src->bc_private.b.firstblock;
+
+	src->bc_private.b.allocated = 0;
+}
+
+STATIC int
+xfs_bmbt_alloc_block(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*start,
+	union xfs_btree_ptr	*new,
+	int			length,
+	int			*stat)
+{
+	xfs_alloc_arg_t		args;		/* block allocation args */
+	int			error;		/* error return value */
+
+	memset(&args, 0, sizeof(args));
+	args.tp = cur->bc_tp;
+	args.mp = cur->bc_mp;
+	args.fsbno = cur->bc_private.b.firstblock;
+	args.firstblock = args.fsbno;
+
+	if (args.fsbno == NULLFSBLOCK) {
+		args.fsbno = be64_to_cpu(start->l);
+		args.type = XFS_ALLOCTYPE_START_BNO;
+		/*
+		 * Make sure there is sufficient room left in the AG to
+		 * complete a full tree split for an extent insert.  If
+		 * we are converting the middle part of an extent then
+		 * we may need space for two tree splits.
+		 *
+		 * We are relying on the caller to make the correct block
+		 * reservation for this operation to succeed.  If the
+		 * reservation amount is insufficient then we may fail a
+		 * block allocation here and corrupt the filesystem.
+		 */
+		args.minleft = xfs_trans_get_block_res(args.tp);
+	} else if (cur->bc_private.b.flist->xbf_low) {
+		args.type = XFS_ALLOCTYPE_START_BNO;
+	} else {
+		args.type = XFS_ALLOCTYPE_NEAR_BNO;
+	}
+
+	args.minlen = args.maxlen = args.prod = 1;
+	args.wasdel = cur->bc_private.b.flags & XFS_BTCUR_BPRV_WASDEL;
+	if (!args.wasdel && xfs_trans_get_block_res(args.tp) == 0) {
+		error = XFS_ERROR(ENOSPC);
+		goto error0;
+	}
+	error = xfs_alloc_vextent(&args);
+	if (error)
+		goto error0;
+
+	if (args.fsbno == NULLFSBLOCK && args.minleft) {
+		/*
+		 * Could not find an AG with enough free space to satisfy
+		 * a full btree split.  Try again without minleft and if
+		 * successful activate the lowspace algorithm.
+		 */
+		args.fsbno = 0;
+		args.type = XFS_ALLOCTYPE_FIRST_AG;
+		args.minleft = 0;
+		error = xfs_alloc_vextent(&args);
+		if (error)
+			goto error0;
+		cur->bc_private.b.flist->xbf_low = 1;
+	}
+	if (args.fsbno == NULLFSBLOCK) {
+		XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+		*stat = 0;
+		return 0;
+	}
+	ASSERT(args.len == 1);
+	cur->bc_private.b.firstblock = args.fsbno;
+	cur->bc_private.b.allocated++;
+	cur->bc_private.b.ip->i_d.di_nblocks++;
+	xfs_trans_log_inode(args.tp, cur->bc_private.b.ip, XFS_ILOG_CORE);
+	xfs_trans_mod_dquot_byino(args.tp, cur->bc_private.b.ip,
+			XFS_TRANS_DQ_BCOUNT, 1L);
+
+	new->l = cpu_to_be64(args.fsbno);
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	*stat = 1;
+	return 0;
+
+ error0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
+	return error;
+}
+
+STATIC int
+xfs_bmbt_free_block(
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*bp)
+{
+	struct xfs_mount	*mp = cur->bc_mp;
+	struct xfs_inode	*ip = cur->bc_private.b.ip;
+	struct xfs_trans	*tp = cur->bc_tp;
+	xfs_fsblock_t		fsbno = XFS_DADDR_TO_FSB(mp, XFS_BUF_ADDR(bp));
+
+	xfs_bmap_add_free(fsbno, 1, cur->bc_private.b.flist, mp);
+	ip->i_d.di_nblocks--;
+
+	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
+	xfs_trans_mod_dquot_byino(tp, ip, XFS_TRANS_DQ_BCOUNT, -1L);
+	xfs_trans_binval(tp, bp);
+	return 0;
+}
+
+STATIC int
+xfs_bmbt_get_minrecs(
+	struct xfs_btree_cur	*cur,
+	int			level)
+{
+	if (level == cur->bc_nlevels - 1) {
+		struct xfs_ifork	*ifp;
+
+		ifp = XFS_IFORK_PTR(cur->bc_private.b.ip,
+				    cur->bc_private.b.whichfork);
+
+		return xfs_bmbt_maxrecs(cur->bc_mp,
+					ifp->if_broot_bytes, level == 0) / 2;
+	}
+
+	return cur->bc_mp->m_bmap_dmnr[level != 0];
+}
+
+int
+xfs_bmbt_get_maxrecs(
+	struct xfs_btree_cur	*cur,
+	int			level)
+{
+	if (level == cur->bc_nlevels - 1) {
+		struct xfs_ifork	*ifp;
+
+		ifp = XFS_IFORK_PTR(cur->bc_private.b.ip,
+				    cur->bc_private.b.whichfork);
+
+		return xfs_bmbt_maxrecs(cur->bc_mp,
+					ifp->if_broot_bytes, level == 0);
+	}
+
+	return cur->bc_mp->m_bmap_dmxr[level != 0];
+
+}
+
+/*
+ * Get the maximum records we could store in the on-disk format.
+ *
+ * For non-root nodes this is equivalent to xfs_bmbt_get_maxrecs, but
+ * for the root node this checks the available space in the dinode fork
+ * so that we can resize the in-memory buffer to match it.  After a
+ * resize to the maximum size this function returns the same value
+ * as xfs_bmbt_get_maxrecs for the root node, too.
+ */
+STATIC int
+xfs_bmbt_get_dmaxrecs(
+	struct xfs_btree_cur	*cur,
+	int			level)
+{
+	if (level != cur->bc_nlevels - 1)
+		return cur->bc_mp->m_bmap_dmxr[level != 0];
+	return xfs_bmdr_maxrecs(cur->bc_mp, cur->bc_private.b.forksize,
+				level == 0);
+}
+
+STATIC void
+xfs_bmbt_init_key_from_rec(
+	union xfs_btree_key	*key,
+	union xfs_btree_rec	*rec)
+{
+	key->bmbt.br_startoff =
+		cpu_to_be64(xfs_bmbt_disk_get_startoff(&rec->bmbt));
+}
+
+STATIC void
+xfs_bmbt_init_rec_from_key(
+	union xfs_btree_key	*key,
+	union xfs_btree_rec	*rec)
+{
+	ASSERT(key->bmbt.br_startoff != 0);
+
+	xfs_bmbt_disk_set_allf(&rec->bmbt, be64_to_cpu(key->bmbt.br_startoff),
+			       0, 0, XFS_EXT_NORM);
+}
+
+STATIC void
+xfs_bmbt_init_rec_from_cur(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_rec	*rec)
+{
+	xfs_bmbt_disk_set_all(&rec->bmbt, &cur->bc_rec.b);
+}
+
+STATIC void
+xfs_bmbt_init_ptr_from_cur(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*ptr)
+{
+	ptr->l = 0;
+}
+
+STATIC __int64_t
+xfs_bmbt_key_diff(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_key	*key)
+{
+	return (__int64_t)be64_to_cpu(key->bmbt.br_startoff) -
+				      cur->bc_rec.b.br_startoff;
+}
+
+#ifdef DEBUG
+STATIC int
+xfs_bmbt_keys_inorder(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_key	*k1,
+	union xfs_btree_key	*k2)
+{
+	return be64_to_cpu(k1->bmbt.br_startoff) <
+		be64_to_cpu(k2->bmbt.br_startoff);
+}
+
+STATIC int
+xfs_bmbt_recs_inorder(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_rec	*r1,
+	union xfs_btree_rec	*r2)
+{
+	return xfs_bmbt_disk_get_startoff(&r1->bmbt) +
+		xfs_bmbt_disk_get_blockcount(&r1->bmbt) <=
+		xfs_bmbt_disk_get_startoff(&r2->bmbt);
+}
+#endif	/* DEBUG */
+
+#ifdef XFS_BTREE_TRACE
+ktrace_t	*xfs_bmbt_trace_buf;
+
+STATIC void
+xfs_bmbt_trace_enter(
+	struct xfs_btree_cur	*cur,
+	const char		*func,
+	char			*s,
+	int			type,
+	int			line,
+	__psunsigned_t		a0,
+	__psunsigned_t		a1,
+	__psunsigned_t		a2,
+	__psunsigned_t		a3,
+	__psunsigned_t		a4,
+	__psunsigned_t		a5,
+	__psunsigned_t		a6,
+	__psunsigned_t		a7,
+	__psunsigned_t		a8,
+	__psunsigned_t		a9,
+	__psunsigned_t		a10)
+{
+	struct xfs_inode	*ip = cur->bc_private.b.ip;
+	int			whichfork = cur->bc_private.b.whichfork;
+
+	ktrace_enter(xfs_bmbt_trace_buf,
+		(void *)((__psint_t)type | (whichfork << 8) | (line << 16)),
+		(void *)func, (void *)s, (void *)ip, (void *)cur,
+		(void *)a0, (void *)a1, (void *)a2, (void *)a3,
+		(void *)a4, (void *)a5, (void *)a6, (void *)a7,
+		(void *)a8, (void *)a9, (void *)a10);
+}
+
+STATIC void
+xfs_bmbt_trace_cursor(
+	struct xfs_btree_cur	*cur,
+	__uint32_t		*s0,
+	__uint64_t		*l0,
+	__uint64_t		*l1)
+{
+	struct xfs_bmbt_rec_host r;
+
+	xfs_bmbt_set_all(&r, &cur->bc_rec.b);
+
+	*s0 = (cur->bc_nlevels << 24) |
+	      (cur->bc_private.b.flags << 16) |
+	       cur->bc_private.b.allocated;
+	*l0 = r.l0;
+	*l1 = r.l1;
+}
+
+STATIC void
+xfs_bmbt_trace_key(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_key	*key,
+	__uint64_t		*l0,
+	__uint64_t		*l1)
+{
+	*l0 = be64_to_cpu(key->bmbt.br_startoff);
+	*l1 = 0;
+}
+
+/* Endian flipping versions of the bmbt extraction functions */
+STATIC void
+xfs_bmbt_disk_get_all(
+	xfs_bmbt_rec_t	*r,
+	xfs_bmbt_irec_t *s)
+{
+	__xfs_bmbt_get_all(get_unaligned_be64(&r->l0),
+				get_unaligned_be64(&r->l1), s);
+}
+
+STATIC void
+xfs_bmbt_trace_record(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_rec	*rec,
+	__uint64_t		*l0,
+	__uint64_t		*l1,
+	__uint64_t		*l2)
+{
+	struct xfs_bmbt_irec	irec;
+
+	xfs_bmbt_disk_get_all(&rec->bmbt, &irec);
+	*l0 = irec.br_startoff;
+	*l1 = irec.br_startblock;
+	*l2 = irec.br_blockcount;
+}
+#endif /* XFS_BTREE_TRACE */
+
+static const struct xfs_btree_ops xfs_bmbt_ops = {
+	.rec_len		= sizeof(xfs_bmbt_rec_t),
+	.key_len		= sizeof(xfs_bmbt_key_t),
+
+	.dup_cursor		= xfs_bmbt_dup_cursor,
+	.update_cursor		= xfs_bmbt_update_cursor,
+	.alloc_block		= xfs_bmbt_alloc_block,
+	.free_block		= xfs_bmbt_free_block,
+	.get_maxrecs		= xfs_bmbt_get_maxrecs,
+	.get_minrecs		= xfs_bmbt_get_minrecs,
+	.get_dmaxrecs		= xfs_bmbt_get_dmaxrecs,
+	.init_key_from_rec	= xfs_bmbt_init_key_from_rec,
+	.init_rec_from_key	= xfs_bmbt_init_rec_from_key,
+	.init_rec_from_cur	= xfs_bmbt_init_rec_from_cur,
+	.init_ptr_from_cur	= xfs_bmbt_init_ptr_from_cur,
+	.key_diff		= xfs_bmbt_key_diff,
+
+#ifdef DEBUG
+	.keys_inorder		= xfs_bmbt_keys_inorder,
+	.recs_inorder		= xfs_bmbt_recs_inorder,
+#endif
+
+#ifdef XFS_BTREE_TRACE
+	.trace_enter		= xfs_bmbt_trace_enter,
+	.trace_cursor		= xfs_bmbt_trace_cursor,
+	.trace_key		= xfs_bmbt_trace_key,
+	.trace_record		= xfs_bmbt_trace_record,
+#endif
+};
+
+/*
+ * Allocate a new bmap btree cursor.
+ */
+struct xfs_btree_cur *				/* new bmap btree cursor */
+xfs_bmbt_init_cursor(
+	struct xfs_mount	*mp,		/* file system mount point */
+	struct xfs_trans	*tp,		/* transaction pointer */
+	struct xfs_inode	*ip,		/* inode owning the btree */
+	int			whichfork)	/* data or attr fork */
+{
+	struct xfs_ifork	*ifp = XFS_IFORK_PTR(ip, whichfork);
+	struct xfs_btree_cur	*cur;
+
+	cur = kmem_zone_zalloc(xfs_btree_cur_zone, KM_SLEEP);
+
+	cur->bc_tp = tp;
+	cur->bc_mp = mp;
+	cur->bc_nlevels = be16_to_cpu(ifp->if_broot->bb_level) + 1;
+	cur->bc_btnum = XFS_BTNUM_BMAP;
+	cur->bc_blocklog = mp->m_sb.sb_blocklog;
+
+	cur->bc_ops = &xfs_bmbt_ops;
+	cur->bc_flags = XFS_BTREE_LONG_PTRS | XFS_BTREE_ROOT_IN_INODE;
+
+	cur->bc_private.b.forksize = XFS_IFORK_SIZE(ip, whichfork);
+	cur->bc_private.b.ip = ip;
+	cur->bc_private.b.firstblock = NULLFSBLOCK;
+	cur->bc_private.b.flist = NULL;
+	cur->bc_private.b.allocated = 0;
+	cur->bc_private.b.flags = 0;
+	cur->bc_private.b.whichfork = whichfork;
+
+	return cur;
+}
+
+/*
+ * Calculate number of records in a bmap btree block.
+ */
+int
+xfs_bmbt_maxrecs(
+	struct xfs_mount	*mp,
+	int			blocklen,
+	int			leaf)
+{
+	blocklen -= XFS_BMBT_BLOCK_LEN(mp);
+
+	if (leaf)
+		return blocklen / sizeof(xfs_bmbt_rec_t);
+	return blocklen / (sizeof(xfs_bmbt_key_t) + sizeof(xfs_bmbt_ptr_t));
+}
+
+/*
+ * Calculate number of records in a bmap btree inode root.
+ */
+int
+xfs_bmdr_maxrecs(
+	struct xfs_mount	*mp,
+	int			blocklen,
+	int			leaf)
+{
+	blocklen -= sizeof(xfs_bmdr_block_t);
+
+	if (leaf)
+		return blocklen / sizeof(xfs_bmdr_rec_t);
+	return blocklen / (sizeof(xfs_bmdr_key_t) + sizeof(xfs_bmdr_ptr_t));
+}
diff --git a/fs/xfs/xfs_bmap_btree.h b/fs/xfs/xfs_bmap_btree.h
new file mode 100644
index 00000000..0e66c4ea
--- /dev/null
+++ b/fs/xfs/xfs_bmap_btree.h
@@ -0,0 +1,240 @@
+/*
+ * Copyright (c) 2000,2002-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_BMAP_BTREE_H__
+#define __XFS_BMAP_BTREE_H__
+
+#define XFS_BMAP_MAGIC	0x424d4150	/* 'BMAP' */
+
+struct xfs_btree_cur;
+struct xfs_btree_block;
+struct xfs_mount;
+struct xfs_inode;
+struct xfs_trans;
+
+/*
+ * Bmap root header, on-disk form only.
+ */
+typedef struct xfs_bmdr_block {
+	__be16		bb_level;	/* 0 is a leaf */
+	__be16		bb_numrecs;	/* current # of data records */
+} xfs_bmdr_block_t;
+
+/*
+ * Bmap btree record and extent descriptor.
+ *  l0:63 is an extent flag (value 1 indicates non-normal).
+ *  l0:9-62 are startoff.
+ *  l0:0-8 and l1:21-63 are startblock.
+ *  l1:0-20 are blockcount.
+ */
+#define BMBT_EXNTFLAG_BITLEN	1
+#define BMBT_STARTOFF_BITLEN	54
+#define BMBT_STARTBLOCK_BITLEN	52
+#define BMBT_BLOCKCOUNT_BITLEN	21
+
+typedef struct xfs_bmbt_rec {
+	__be64			l0, l1;
+} xfs_bmbt_rec_t;
+
+typedef __uint64_t	xfs_bmbt_rec_base_t;	/* use this for casts */
+typedef xfs_bmbt_rec_t xfs_bmdr_rec_t;
+
+typedef struct xfs_bmbt_rec_host {
+	__uint64_t		l0, l1;
+} xfs_bmbt_rec_host_t;
+
+/*
+ * Values and macros for delayed-allocation startblock fields.
+ */
+#define STARTBLOCKVALBITS	17
+#define STARTBLOCKMASKBITS	(15 + XFS_BIG_BLKNOS * 20)
+#define DSTARTBLOCKMASKBITS	(15 + 20)
+#define STARTBLOCKMASK		\
+	(((((xfs_fsblock_t)1) << STARTBLOCKMASKBITS) - 1) << STARTBLOCKVALBITS)
+#define DSTARTBLOCKMASK		\
+	(((((xfs_dfsbno_t)1) << DSTARTBLOCKMASKBITS) - 1) << STARTBLOCKVALBITS)
+
+static inline int isnullstartblock(xfs_fsblock_t x)
+{
+	return ((x) & STARTBLOCKMASK) == STARTBLOCKMASK;
+}
+
+static inline int isnulldstartblock(xfs_dfsbno_t x)
+{
+	return ((x) & DSTARTBLOCKMASK) == DSTARTBLOCKMASK;
+}
+
+static inline xfs_fsblock_t nullstartblock(int k)
+{
+	ASSERT(k < (1 << STARTBLOCKVALBITS));
+	return STARTBLOCKMASK | (k);
+}
+
+static inline xfs_filblks_t startblockval(xfs_fsblock_t x)
+{
+	return (xfs_filblks_t)((x) & ~STARTBLOCKMASK);
+}
+
+/*
+ * Possible extent formats.
+ */
+typedef enum {
+	XFS_EXTFMT_NOSTATE = 0,
+	XFS_EXTFMT_HASSTATE
+} xfs_exntfmt_t;
+
+/*
+ * Possible extent states.
+ */
+typedef enum {
+	XFS_EXT_NORM, XFS_EXT_UNWRITTEN,
+	XFS_EXT_DMAPI_OFFLINE, XFS_EXT_INVALID
+} xfs_exntst_t;
+
+/*
+ * Extent state and extent format macros.
+ */
+#define XFS_EXTFMT_INODE(x)	\
+	(xfs_sb_version_hasextflgbit(&((x)->i_mount->m_sb)) ? \
+		XFS_EXTFMT_HASSTATE : XFS_EXTFMT_NOSTATE)
+#define ISUNWRITTEN(x)	((x)->br_state == XFS_EXT_UNWRITTEN)
+
+/*
+ * Incore version of above.
+ */
+typedef struct xfs_bmbt_irec
+{
+	xfs_fileoff_t	br_startoff;	/* starting file offset */
+	xfs_fsblock_t	br_startblock;	/* starting block number */
+	xfs_filblks_t	br_blockcount;	/* number of blocks */
+	xfs_exntst_t	br_state;	/* extent state */
+} xfs_bmbt_irec_t;
+
+/*
+ * Key structure for non-leaf levels of the tree.
+ */
+typedef struct xfs_bmbt_key {
+	__be64		br_startoff;	/* starting file offset */
+} xfs_bmbt_key_t, xfs_bmdr_key_t;
+
+/* btree pointer type */
+typedef __be64 xfs_bmbt_ptr_t, xfs_bmdr_ptr_t;
+
+/*
+ * Btree block header size depends on a superblock flag.
+ *
+ * (not quite yet, but soon)
+ */
+#define XFS_BMBT_BLOCK_LEN(mp)	XFS_BTREE_LBLOCK_LEN
+
+#define XFS_BMBT_REC_ADDR(mp, block, index) \
+	((xfs_bmbt_rec_t *) \
+		((char *)(block) + \
+		 XFS_BMBT_BLOCK_LEN(mp) + \
+		 ((index) - 1) * sizeof(xfs_bmbt_rec_t)))
+
+#define XFS_BMBT_KEY_ADDR(mp, block, index) \
+	((xfs_bmbt_key_t *) \
+		((char *)(block) + \
+		 XFS_BMBT_BLOCK_LEN(mp) + \
+		 ((index) - 1) * sizeof(xfs_bmbt_key_t)))
+
+#define XFS_BMBT_PTR_ADDR(mp, block, index, maxrecs) \
+	((xfs_bmbt_ptr_t *) \
+		((char *)(block) + \
+		 XFS_BMBT_BLOCK_LEN(mp) + \
+		 (maxrecs) * sizeof(xfs_bmbt_key_t) + \
+		 ((index) - 1) * sizeof(xfs_bmbt_ptr_t)))
+
+#define XFS_BMDR_REC_ADDR(block, index) \
+	((xfs_bmdr_rec_t *) \
+		((char *)(block) + \
+		 sizeof(struct xfs_bmdr_block) + \
+	         ((index) - 1) * sizeof(xfs_bmdr_rec_t)))
+
+#define XFS_BMDR_KEY_ADDR(block, index) \
+	((xfs_bmdr_key_t *) \
+		((char *)(block) + \
+		 sizeof(struct xfs_bmdr_block) + \
+		 ((index) - 1) * sizeof(xfs_bmdr_key_t)))
+
+#define XFS_BMDR_PTR_ADDR(block, index, maxrecs) \
+	((xfs_bmdr_ptr_t *) \
+		((char *)(block) + \
+		 sizeof(struct xfs_bmdr_block) + \
+		 (maxrecs) * sizeof(xfs_bmdr_key_t) + \
+		 ((index) - 1) * sizeof(xfs_bmdr_ptr_t)))
+
+/*
+ * These are to be used when we know the size of the block and
+ * we don't have a cursor.
+ */
+#define XFS_BMAP_BROOT_PTR_ADDR(mp, bb, i, sz) \
+	XFS_BMBT_PTR_ADDR(mp, bb, i, xfs_bmbt_maxrecs(mp, sz, 0))
+
+#define XFS_BMAP_BROOT_SPACE_CALC(nrecs) \
+	(int)(XFS_BTREE_LBLOCK_LEN + \
+	       ((nrecs) * (sizeof(xfs_bmbt_key_t) + sizeof(xfs_bmbt_ptr_t))))
+
+#define XFS_BMAP_BROOT_SPACE(bb) \
+	(XFS_BMAP_BROOT_SPACE_CALC(be16_to_cpu((bb)->bb_numrecs)))
+#define XFS_BMDR_SPACE_CALC(nrecs) \
+	(int)(sizeof(xfs_bmdr_block_t) + \
+	       ((nrecs) * (sizeof(xfs_bmbt_key_t) + sizeof(xfs_bmbt_ptr_t))))
+
+/*
+ * Maximum number of bmap btree levels.
+ */
+#define XFS_BM_MAXLEVELS(mp,w)		((mp)->m_bm_maxlevels[(w)])
+
+/*
+ * Prototypes for xfs_bmap.c to call.
+ */
+extern void xfs_bmdr_to_bmbt(struct xfs_mount *, xfs_bmdr_block_t *, int,
+			struct xfs_btree_block *, int);
+extern void xfs_bmbt_get_all(xfs_bmbt_rec_host_t *r, xfs_bmbt_irec_t *s);
+extern xfs_filblks_t xfs_bmbt_get_blockcount(xfs_bmbt_rec_host_t *r);
+extern xfs_fsblock_t xfs_bmbt_get_startblock(xfs_bmbt_rec_host_t *r);
+extern xfs_fileoff_t xfs_bmbt_get_startoff(xfs_bmbt_rec_host_t *r);
+extern xfs_exntst_t xfs_bmbt_get_state(xfs_bmbt_rec_host_t *r);
+
+extern xfs_filblks_t xfs_bmbt_disk_get_blockcount(xfs_bmbt_rec_t *r);
+extern xfs_fileoff_t xfs_bmbt_disk_get_startoff(xfs_bmbt_rec_t *r);
+
+extern void xfs_bmbt_set_all(xfs_bmbt_rec_host_t *r, xfs_bmbt_irec_t *s);
+extern void xfs_bmbt_set_allf(xfs_bmbt_rec_host_t *r, xfs_fileoff_t o,
+			xfs_fsblock_t b, xfs_filblks_t c, xfs_exntst_t v);
+extern void xfs_bmbt_set_blockcount(xfs_bmbt_rec_host_t *r, xfs_filblks_t v);
+extern void xfs_bmbt_set_startblock(xfs_bmbt_rec_host_t *r, xfs_fsblock_t v);
+extern void xfs_bmbt_set_startoff(xfs_bmbt_rec_host_t *r, xfs_fileoff_t v);
+extern void xfs_bmbt_set_state(xfs_bmbt_rec_host_t *r, xfs_exntst_t v);
+
+extern void xfs_bmbt_disk_set_allf(xfs_bmbt_rec_t *r, xfs_fileoff_t o,
+			xfs_fsblock_t b, xfs_filblks_t c, xfs_exntst_t v);
+
+extern void xfs_bmbt_to_bmdr(struct xfs_mount *, struct xfs_btree_block *, int,
+			xfs_bmdr_block_t *, int);
+
+extern int xfs_bmbt_get_maxrecs(struct xfs_btree_cur *, int level);
+extern int xfs_bmdr_maxrecs(struct xfs_mount *, int blocklen, int leaf);
+extern int xfs_bmbt_maxrecs(struct xfs_mount *, int blocklen, int leaf);
+
+extern struct xfs_btree_cur *xfs_bmbt_init_cursor(struct xfs_mount *,
+		struct xfs_trans *, struct xfs_inode *, int);
+
+
+#endif	/* __XFS_BMAP_BTREE_H__ */
diff --git a/fs/xfs/xfs_btree.c b/fs/xfs/xfs_btree.c
new file mode 100644
index 00000000..2f9e97c1
--- /dev/null
+++ b/fs/xfs/xfs_btree.c
@@ -0,0 +1,3679 @@
+/*
+ * Copyright (c) 2000-2002,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_mount.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_alloc_btree.h"
+#include "xfs_ialloc_btree.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_inode_item.h"
+#include "xfs_btree.h"
+#include "xfs_btree_trace.h"
+#include "xfs_error.h"
+#include "xfs_trace.h"
+
+/*
+ * Cursor allocation zone.
+ */
+kmem_zone_t	*xfs_btree_cur_zone;
+
+/*
+ * Btree magic numbers.
+ */
+const __uint32_t xfs_magics[XFS_BTNUM_MAX] = {
+	XFS_ABTB_MAGIC, XFS_ABTC_MAGIC, XFS_BMAP_MAGIC, XFS_IBT_MAGIC
+};
+
+
+STATIC int				/* error (0 or EFSCORRUPTED) */
+xfs_btree_check_lblock(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	struct xfs_btree_block	*block,	/* btree long form block pointer */
+	int			level,	/* level of the btree block */
+	struct xfs_buf		*bp)	/* buffer for block, if any */
+{
+	int			lblock_ok; /* block passes checks */
+	struct xfs_mount	*mp;	/* file system mount point */
+
+	mp = cur->bc_mp;
+	lblock_ok =
+		be32_to_cpu(block->bb_magic) == xfs_magics[cur->bc_btnum] &&
+		be16_to_cpu(block->bb_level) == level &&
+		be16_to_cpu(block->bb_numrecs) <=
+			cur->bc_ops->get_maxrecs(cur, level) &&
+		block->bb_u.l.bb_leftsib &&
+		(be64_to_cpu(block->bb_u.l.bb_leftsib) == NULLDFSBNO ||
+		 XFS_FSB_SANITY_CHECK(mp,
+		 	be64_to_cpu(block->bb_u.l.bb_leftsib))) &&
+		block->bb_u.l.bb_rightsib &&
+		(be64_to_cpu(block->bb_u.l.bb_rightsib) == NULLDFSBNO ||
+		 XFS_FSB_SANITY_CHECK(mp,
+		 	be64_to_cpu(block->bb_u.l.bb_rightsib)));
+	if (unlikely(XFS_TEST_ERROR(!lblock_ok, mp,
+			XFS_ERRTAG_BTREE_CHECK_LBLOCK,
+			XFS_RANDOM_BTREE_CHECK_LBLOCK))) {
+		if (bp)
+			trace_xfs_btree_corrupt(bp, _RET_IP_);
+		XFS_ERROR_REPORT("xfs_btree_check_lblock", XFS_ERRLEVEL_LOW,
+				 mp);
+		return XFS_ERROR(EFSCORRUPTED);
+	}
+	return 0;
+}
+
+STATIC int				/* error (0 or EFSCORRUPTED) */
+xfs_btree_check_sblock(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	struct xfs_btree_block	*block,	/* btree short form block pointer */
+	int			level,	/* level of the btree block */
+	struct xfs_buf		*bp)	/* buffer containing block */
+{
+	struct xfs_buf		*agbp;	/* buffer for ag. freespace struct */
+	struct xfs_agf		*agf;	/* ag. freespace structure */
+	xfs_agblock_t		agflen;	/* native ag. freespace length */
+	int			sblock_ok; /* block passes checks */
+
+	agbp = cur->bc_private.a.agbp;
+	agf = XFS_BUF_TO_AGF(agbp);
+	agflen = be32_to_cpu(agf->agf_length);
+	sblock_ok =
+		be32_to_cpu(block->bb_magic) == xfs_magics[cur->bc_btnum] &&
+		be16_to_cpu(block->bb_level) == level &&
+		be16_to_cpu(block->bb_numrecs) <=
+			cur->bc_ops->get_maxrecs(cur, level) &&
+		(be32_to_cpu(block->bb_u.s.bb_leftsib) == NULLAGBLOCK ||
+		 be32_to_cpu(block->bb_u.s.bb_leftsib) < agflen) &&
+		block->bb_u.s.bb_leftsib &&
+		(be32_to_cpu(block->bb_u.s.bb_rightsib) == NULLAGBLOCK ||
+		 be32_to_cpu(block->bb_u.s.bb_rightsib) < agflen) &&
+		block->bb_u.s.bb_rightsib;
+	if (unlikely(XFS_TEST_ERROR(!sblock_ok, cur->bc_mp,
+			XFS_ERRTAG_BTREE_CHECK_SBLOCK,
+			XFS_RANDOM_BTREE_CHECK_SBLOCK))) {
+		if (bp)
+			trace_xfs_btree_corrupt(bp, _RET_IP_);
+		XFS_CORRUPTION_ERROR("xfs_btree_check_sblock",
+			XFS_ERRLEVEL_LOW, cur->bc_mp, block);
+		return XFS_ERROR(EFSCORRUPTED);
+	}
+	return 0;
+}
+
+/*
+ * Debug routine: check that block header is ok.
+ */
+int
+xfs_btree_check_block(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	struct xfs_btree_block	*block,	/* generic btree block pointer */
+	int			level,	/* level of the btree block */
+	struct xfs_buf		*bp)	/* buffer containing block, if any */
+{
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
+		return xfs_btree_check_lblock(cur, block, level, bp);
+	else
+		return xfs_btree_check_sblock(cur, block, level, bp);
+}
+
+/*
+ * Check that (long) pointer is ok.
+ */
+int					/* error (0 or EFSCORRUPTED) */
+xfs_btree_check_lptr(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_dfsbno_t		bno,	/* btree block disk address */
+	int			level)	/* btree block level */
+{
+	XFS_WANT_CORRUPTED_RETURN(
+		level > 0 &&
+		bno != NULLDFSBNO &&
+		XFS_FSB_SANITY_CHECK(cur->bc_mp, bno));
+	return 0;
+}
+
+#ifdef DEBUG
+/*
+ * Check that (short) pointer is ok.
+ */
+STATIC int				/* error (0 or EFSCORRUPTED) */
+xfs_btree_check_sptr(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_agblock_t		bno,	/* btree block disk address */
+	int			level)	/* btree block level */
+{
+	xfs_agblock_t		agblocks = cur->bc_mp->m_sb.sb_agblocks;
+
+	XFS_WANT_CORRUPTED_RETURN(
+		level > 0 &&
+		bno != NULLAGBLOCK &&
+		bno != 0 &&
+		bno < agblocks);
+	return 0;
+}
+
+/*
+ * Check that block ptr is ok.
+ */
+STATIC int				/* error (0 or EFSCORRUPTED) */
+xfs_btree_check_ptr(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	union xfs_btree_ptr	*ptr,	/* btree block disk address */
+	int			index,	/* offset from ptr to check */
+	int			level)	/* btree block level */
+{
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
+		return xfs_btree_check_lptr(cur,
+				be64_to_cpu((&ptr->l)[index]), level);
+	} else {
+		return xfs_btree_check_sptr(cur,
+				be32_to_cpu((&ptr->s)[index]), level);
+	}
+}
+#endif
+
+/*
+ * Delete the btree cursor.
+ */
+void
+xfs_btree_del_cursor(
+	xfs_btree_cur_t	*cur,		/* btree cursor */
+	int		error)		/* del because of error */
+{
+	int		i;		/* btree level */
+
+	/*
+	 * Clear the buffer pointers, and release the buffers.
+	 * If we're doing this in the face of an error, we
+	 * need to make sure to inspect all of the entries
+	 * in the bc_bufs array for buffers to be unlocked.
+	 * This is because some of the btree code works from
+	 * level n down to 0, and if we get an error along
+	 * the way we won't have initialized all the entries
+	 * down to 0.
+	 */
+	for (i = 0; i < cur->bc_nlevels; i++) {
+		if (cur->bc_bufs[i])
+			xfs_trans_brelse(cur->bc_tp, cur->bc_bufs[i]);
+		else if (!error)
+			break;
+	}
+	/*
+	 * Can't free a bmap cursor without having dealt with the
+	 * allocated indirect blocks' accounting.
+	 */
+	ASSERT(cur->bc_btnum != XFS_BTNUM_BMAP ||
+	       cur->bc_private.b.allocated == 0);
+	/*
+	 * Free the cursor.
+	 */
+	kmem_zone_free(xfs_btree_cur_zone, cur);
+}
+
+/*
+ * Duplicate the btree cursor.
+ * Allocate a new one, copy the record, re-get the buffers.
+ */
+int					/* error */
+xfs_btree_dup_cursor(
+	xfs_btree_cur_t	*cur,		/* input cursor */
+	xfs_btree_cur_t	**ncur)		/* output cursor */
+{
+	xfs_buf_t	*bp;		/* btree block's buffer pointer */
+	int		error;		/* error return value */
+	int		i;		/* level number of btree block */
+	xfs_mount_t	*mp;		/* mount structure for filesystem */
+	xfs_btree_cur_t	*new;		/* new cursor value */
+	xfs_trans_t	*tp;		/* transaction pointer, can be NULL */
+
+	tp = cur->bc_tp;
+	mp = cur->bc_mp;
+
+	/*
+	 * Allocate a new cursor like the old one.
+	 */
+	new = cur->bc_ops->dup_cursor(cur);
+
+	/*
+	 * Copy the record currently in the cursor.
+	 */
+	new->bc_rec = cur->bc_rec;
+
+	/*
+	 * For each level current, re-get the buffer and copy the ptr value.
+	 */
+	for (i = 0; i < new->bc_nlevels; i++) {
+		new->bc_ptrs[i] = cur->bc_ptrs[i];
+		new->bc_ra[i] = cur->bc_ra[i];
+		if ((bp = cur->bc_bufs[i])) {
+			if ((error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp,
+				XFS_BUF_ADDR(bp), mp->m_bsize, 0, &bp))) {
+				xfs_btree_del_cursor(new, error);
+				*ncur = NULL;
+				return error;
+			}
+			new->bc_bufs[i] = bp;
+			ASSERT(bp);
+			ASSERT(!XFS_BUF_GETERROR(bp));
+		} else
+			new->bc_bufs[i] = NULL;
+	}
+	*ncur = new;
+	return 0;
+}
+
+/*
+ * XFS btree block layout and addressing:
+ *
+ * There are two types of blocks in the btree: leaf and non-leaf blocks.
+ *
+ * The leaf record start with a header then followed by records containing
+ * the values.  A non-leaf block also starts with the same header, and
+ * then first contains lookup keys followed by an equal number of pointers
+ * to the btree blocks at the previous level.
+ *
+ *		+--------+-------+-------+-------+-------+-------+-------+
+ * Leaf:	| header | rec 1 | rec 2 | rec 3 | rec 4 | rec 5 | rec N |
+ *		+--------+-------+-------+-------+-------+-------+-------+
+ *
+ *		+--------+-------+-------+-------+-------+-------+-------+
+ * Non-Leaf:	| header | key 1 | key 2 | key N | ptr 1 | ptr 2 | ptr N |
+ *		+--------+-------+-------+-------+-------+-------+-------+
+ *
+ * The header is called struct xfs_btree_block for reasons better left unknown
+ * and comes in different versions for short (32bit) and long (64bit) block
+ * pointers.  The record and key structures are defined by the btree instances
+ * and opaque to the btree core.  The block pointers are simple disk endian
+ * integers, available in a short (32bit) and long (64bit) variant.
+ *
+ * The helpers below calculate the offset of a given record, key or pointer
+ * into a btree block (xfs_btree_*_offset) or return a pointer to the given
+ * record, key or pointer (xfs_btree_*_addr).  Note that all addressing
+ * inside the btree block is done using indices starting at one, not zero!
+ */
+
+/*
+ * Return size of the btree block header for this btree instance.
+ */
+static inline size_t xfs_btree_block_len(struct xfs_btree_cur *cur)
+{
+	return (cur->bc_flags & XFS_BTREE_LONG_PTRS) ?
+		XFS_BTREE_LBLOCK_LEN :
+		XFS_BTREE_SBLOCK_LEN;
+}
+
+/*
+ * Return size of btree block pointers for this btree instance.
+ */
+static inline size_t xfs_btree_ptr_len(struct xfs_btree_cur *cur)
+{
+	return (cur->bc_flags & XFS_BTREE_LONG_PTRS) ?
+		sizeof(__be64) : sizeof(__be32);
+}
+
+/*
+ * Calculate offset of the n-th record in a btree block.
+ */
+STATIC size_t
+xfs_btree_rec_offset(
+	struct xfs_btree_cur	*cur,
+	int			n)
+{
+	return xfs_btree_block_len(cur) +
+		(n - 1) * cur->bc_ops->rec_len;
+}
+
+/*
+ * Calculate offset of the n-th key in a btree block.
+ */
+STATIC size_t
+xfs_btree_key_offset(
+	struct xfs_btree_cur	*cur,
+	int			n)
+{
+	return xfs_btree_block_len(cur) +
+		(n - 1) * cur->bc_ops->key_len;
+}
+
+/*
+ * Calculate offset of the n-th block pointer in a btree block.
+ */
+STATIC size_t
+xfs_btree_ptr_offset(
+	struct xfs_btree_cur	*cur,
+	int			n,
+	int			level)
+{
+	return xfs_btree_block_len(cur) +
+		cur->bc_ops->get_maxrecs(cur, level) * cur->bc_ops->key_len +
+		(n - 1) * xfs_btree_ptr_len(cur);
+}
+
+/*
+ * Return a pointer to the n-th record in the btree block.
+ */
+STATIC union xfs_btree_rec *
+xfs_btree_rec_addr(
+	struct xfs_btree_cur	*cur,
+	int			n,
+	struct xfs_btree_block	*block)
+{
+	return (union xfs_btree_rec *)
+		((char *)block + xfs_btree_rec_offset(cur, n));
+}
+
+/*
+ * Return a pointer to the n-th key in the btree block.
+ */
+STATIC union xfs_btree_key *
+xfs_btree_key_addr(
+	struct xfs_btree_cur	*cur,
+	int			n,
+	struct xfs_btree_block	*block)
+{
+	return (union xfs_btree_key *)
+		((char *)block + xfs_btree_key_offset(cur, n));
+}
+
+/*
+ * Return a pointer to the n-th block pointer in the btree block.
+ */
+STATIC union xfs_btree_ptr *
+xfs_btree_ptr_addr(
+	struct xfs_btree_cur	*cur,
+	int			n,
+	struct xfs_btree_block	*block)
+{
+	int			level = xfs_btree_get_level(block);
+
+	ASSERT(block->bb_level != 0);
+
+	return (union xfs_btree_ptr *)
+		((char *)block + xfs_btree_ptr_offset(cur, n, level));
+}
+
+/*
+ * Get a the root block which is stored in the inode.
+ *
+ * For now this btree implementation assumes the btree root is always
+ * stored in the if_broot field of an inode fork.
+ */
+STATIC struct xfs_btree_block *
+xfs_btree_get_iroot(
+       struct xfs_btree_cur    *cur)
+{
+       struct xfs_ifork        *ifp;
+
+       ifp = XFS_IFORK_PTR(cur->bc_private.b.ip, cur->bc_private.b.whichfork);
+       return (struct xfs_btree_block *)ifp->if_broot;
+}
+
+/*
+ * Retrieve the block pointer from the cursor at the given level.
+ * This may be an inode btree root or from a buffer.
+ */
+STATIC struct xfs_btree_block *		/* generic btree block pointer */
+xfs_btree_get_block(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	int			level,	/* level in btree */
+	struct xfs_buf		**bpp)	/* buffer containing the block */
+{
+	if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
+	    (level == cur->bc_nlevels - 1)) {
+		*bpp = NULL;
+		return xfs_btree_get_iroot(cur);
+	}
+
+	*bpp = cur->bc_bufs[level];
+	return XFS_BUF_TO_BLOCK(*bpp);
+}
+
+/*
+ * Get a buffer for the block, return it with no data read.
+ * Long-form addressing.
+ */
+xfs_buf_t *				/* buffer for fsbno */
+xfs_btree_get_bufl(
+	xfs_mount_t	*mp,		/* file system mount point */
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_fsblock_t	fsbno,		/* file system block number */
+	uint		lock)		/* lock flags for get_buf */
+{
+	xfs_buf_t	*bp;		/* buffer pointer (return value) */
+	xfs_daddr_t		d;		/* real disk block address */
+
+	ASSERT(fsbno != NULLFSBLOCK);
+	d = XFS_FSB_TO_DADDR(mp, fsbno);
+	bp = xfs_trans_get_buf(tp, mp->m_ddev_targp, d, mp->m_bsize, lock);
+	ASSERT(bp);
+	ASSERT(!XFS_BUF_GETERROR(bp));
+	return bp;
+}
+
+/*
+ * Get a buffer for the block, return it with no data read.
+ * Short-form addressing.
+ */
+xfs_buf_t *				/* buffer for agno/agbno */
+xfs_btree_get_bufs(
+	xfs_mount_t	*mp,		/* file system mount point */
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_agnumber_t	agno,		/* allocation group number */
+	xfs_agblock_t	agbno,		/* allocation group block number */
+	uint		lock)		/* lock flags for get_buf */
+{
+	xfs_buf_t	*bp;		/* buffer pointer (return value) */
+	xfs_daddr_t		d;		/* real disk block address */
+
+	ASSERT(agno != NULLAGNUMBER);
+	ASSERT(agbno != NULLAGBLOCK);
+	d = XFS_AGB_TO_DADDR(mp, agno, agbno);
+	bp = xfs_trans_get_buf(tp, mp->m_ddev_targp, d, mp->m_bsize, lock);
+	ASSERT(bp);
+	ASSERT(!XFS_BUF_GETERROR(bp));
+	return bp;
+}
+
+/*
+ * Check for the cursor referring to the last block at the given level.
+ */
+int					/* 1=is last block, 0=not last block */
+xfs_btree_islastblock(
+	xfs_btree_cur_t		*cur,	/* btree cursor */
+	int			level)	/* level to check */
+{
+	struct xfs_btree_block	*block;	/* generic btree block pointer */
+	xfs_buf_t		*bp;	/* buffer containing block */
+
+	block = xfs_btree_get_block(cur, level, &bp);
+	xfs_btree_check_block(cur, block, level, bp);
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
+		return be64_to_cpu(block->bb_u.l.bb_rightsib) == NULLDFSBNO;
+	else
+		return be32_to_cpu(block->bb_u.s.bb_rightsib) == NULLAGBLOCK;
+}
+
+/*
+ * Change the cursor to point to the first record at the given level.
+ * Other levels are unaffected.
+ */
+STATIC int				/* success=1, failure=0 */
+xfs_btree_firstrec(
+	xfs_btree_cur_t		*cur,	/* btree cursor */
+	int			level)	/* level to change */
+{
+	struct xfs_btree_block	*block;	/* generic btree block pointer */
+	xfs_buf_t		*bp;	/* buffer containing block */
+
+	/*
+	 * Get the block pointer for this level.
+	 */
+	block = xfs_btree_get_block(cur, level, &bp);
+	xfs_btree_check_block(cur, block, level, bp);
+	/*
+	 * It's empty, there is no such record.
+	 */
+	if (!block->bb_numrecs)
+		return 0;
+	/*
+	 * Set the ptr value to 1, that's the first record/key.
+	 */
+	cur->bc_ptrs[level] = 1;
+	return 1;
+}
+
+/*
+ * Change the cursor to point to the last record in the current block
+ * at the given level.  Other levels are unaffected.
+ */
+STATIC int				/* success=1, failure=0 */
+xfs_btree_lastrec(
+	xfs_btree_cur_t		*cur,	/* btree cursor */
+	int			level)	/* level to change */
+{
+	struct xfs_btree_block	*block;	/* generic btree block pointer */
+	xfs_buf_t		*bp;	/* buffer containing block */
+
+	/*
+	 * Get the block pointer for this level.
+	 */
+	block = xfs_btree_get_block(cur, level, &bp);
+	xfs_btree_check_block(cur, block, level, bp);
+	/*
+	 * It's empty, there is no such record.
+	 */
+	if (!block->bb_numrecs)
+		return 0;
+	/*
+	 * Set the ptr value to numrecs, that's the last record/key.
+	 */
+	cur->bc_ptrs[level] = be16_to_cpu(block->bb_numrecs);
+	return 1;
+}
+
+/*
+ * Compute first and last byte offsets for the fields given.
+ * Interprets the offsets table, which contains struct field offsets.
+ */
+void
+xfs_btree_offsets(
+	__int64_t	fields,		/* bitmask of fields */
+	const short	*offsets,	/* table of field offsets */
+	int		nbits,		/* number of bits to inspect */
+	int		*first,		/* output: first byte offset */
+	int		*last)		/* output: last byte offset */
+{
+	int		i;		/* current bit number */
+	__int64_t	imask;		/* mask for current bit number */
+
+	ASSERT(fields != 0);
+	/*
+	 * Find the lowest bit, so the first byte offset.
+	 */
+	for (i = 0, imask = 1LL; ; i++, imask <<= 1) {
+		if (imask & fields) {
+			*first = offsets[i];
+			break;
+		}
+	}
+	/*
+	 * Find the highest bit, so the last byte offset.
+	 */
+	for (i = nbits - 1, imask = 1LL << i; ; i--, imask >>= 1) {
+		if (imask & fields) {
+			*last = offsets[i + 1] - 1;
+			break;
+		}
+	}
+}
+
+/*
+ * Get a buffer for the block, return it read in.
+ * Long-form addressing.
+ */
+int					/* error */
+xfs_btree_read_bufl(
+	xfs_mount_t	*mp,		/* file system mount point */
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_fsblock_t	fsbno,		/* file system block number */
+	uint		lock,		/* lock flags for read_buf */
+	xfs_buf_t	**bpp,		/* buffer for fsbno */
+	int		refval)		/* ref count value for buffer */
+{
+	xfs_buf_t	*bp;		/* return value */
+	xfs_daddr_t		d;		/* real disk block address */
+	int		error;
+
+	ASSERT(fsbno != NULLFSBLOCK);
+	d = XFS_FSB_TO_DADDR(mp, fsbno);
+	if ((error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp, d,
+			mp->m_bsize, lock, &bp))) {
+		return error;
+	}
+	ASSERT(!bp || !XFS_BUF_GETERROR(bp));
+	if (bp)
+		XFS_BUF_SET_VTYPE_REF(bp, B_FS_MAP, refval);
+	*bpp = bp;
+	return 0;
+}
+
+/*
+ * Read-ahead the block, don't wait for it, don't return a buffer.
+ * Long-form addressing.
+ */
+/* ARGSUSED */
+void
+xfs_btree_reada_bufl(
+	xfs_mount_t	*mp,		/* file system mount point */
+	xfs_fsblock_t	fsbno,		/* file system block number */
+	xfs_extlen_t	count)		/* count of filesystem blocks */
+{
+	xfs_daddr_t		d;
+
+	ASSERT(fsbno != NULLFSBLOCK);
+	d = XFS_FSB_TO_DADDR(mp, fsbno);
+	xfs_buf_readahead(mp->m_ddev_targp, d, mp->m_bsize * count);
+}
+
+/*
+ * Read-ahead the block, don't wait for it, don't return a buffer.
+ * Short-form addressing.
+ */
+/* ARGSUSED */
+void
+xfs_btree_reada_bufs(
+	xfs_mount_t	*mp,		/* file system mount point */
+	xfs_agnumber_t	agno,		/* allocation group number */
+	xfs_agblock_t	agbno,		/* allocation group block number */
+	xfs_extlen_t	count)		/* count of filesystem blocks */
+{
+	xfs_daddr_t		d;
+
+	ASSERT(agno != NULLAGNUMBER);
+	ASSERT(agbno != NULLAGBLOCK);
+	d = XFS_AGB_TO_DADDR(mp, agno, agbno);
+	xfs_buf_readahead(mp->m_ddev_targp, d, mp->m_bsize * count);
+}
+
+STATIC int
+xfs_btree_readahead_lblock(
+	struct xfs_btree_cur	*cur,
+	int			lr,
+	struct xfs_btree_block	*block)
+{
+	int			rval = 0;
+	xfs_dfsbno_t		left = be64_to_cpu(block->bb_u.l.bb_leftsib);
+	xfs_dfsbno_t		right = be64_to_cpu(block->bb_u.l.bb_rightsib);
+
+	if ((lr & XFS_BTCUR_LEFTRA) && left != NULLDFSBNO) {
+		xfs_btree_reada_bufl(cur->bc_mp, left, 1);
+		rval++;
+	}
+
+	if ((lr & XFS_BTCUR_RIGHTRA) && right != NULLDFSBNO) {
+		xfs_btree_reada_bufl(cur->bc_mp, right, 1);
+		rval++;
+	}
+
+	return rval;
+}
+
+STATIC int
+xfs_btree_readahead_sblock(
+	struct xfs_btree_cur	*cur,
+	int			lr,
+	struct xfs_btree_block *block)
+{
+	int			rval = 0;
+	xfs_agblock_t		left = be32_to_cpu(block->bb_u.s.bb_leftsib);
+	xfs_agblock_t		right = be32_to_cpu(block->bb_u.s.bb_rightsib);
+
+
+	if ((lr & XFS_BTCUR_LEFTRA) && left != NULLAGBLOCK) {
+		xfs_btree_reada_bufs(cur->bc_mp, cur->bc_private.a.agno,
+				     left, 1);
+		rval++;
+	}
+
+	if ((lr & XFS_BTCUR_RIGHTRA) && right != NULLAGBLOCK) {
+		xfs_btree_reada_bufs(cur->bc_mp, cur->bc_private.a.agno,
+				     right, 1);
+		rval++;
+	}
+
+	return rval;
+}
+
+/*
+ * Read-ahead btree blocks, at the given level.
+ * Bits in lr are set from XFS_BTCUR_{LEFT,RIGHT}RA.
+ */
+STATIC int
+xfs_btree_readahead(
+	struct xfs_btree_cur	*cur,		/* btree cursor */
+	int			lev,		/* level in btree */
+	int			lr)		/* left/right bits */
+{
+	struct xfs_btree_block	*block;
+
+	/*
+	 * No readahead needed if we are at the root level and the
+	 * btree root is stored in the inode.
+	 */
+	if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
+	    (lev == cur->bc_nlevels - 1))
+		return 0;
+
+	if ((cur->bc_ra[lev] | lr) == cur->bc_ra[lev])
+		return 0;
+
+	cur->bc_ra[lev] |= lr;
+	block = XFS_BUF_TO_BLOCK(cur->bc_bufs[lev]);
+
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
+		return xfs_btree_readahead_lblock(cur, lr, block);
+	return xfs_btree_readahead_sblock(cur, lr, block);
+}
+
+/*
+ * Set the buffer for level "lev" in the cursor to bp, releasing
+ * any previous buffer.
+ */
+STATIC void
+xfs_btree_setbuf(
+	xfs_btree_cur_t		*cur,	/* btree cursor */
+	int			lev,	/* level in btree */
+	xfs_buf_t		*bp)	/* new buffer to set */
+{
+	struct xfs_btree_block	*b;	/* btree block */
+
+	if (cur->bc_bufs[lev])
+		xfs_trans_brelse(cur->bc_tp, cur->bc_bufs[lev]);
+	cur->bc_bufs[lev] = bp;
+	cur->bc_ra[lev] = 0;
+
+	b = XFS_BUF_TO_BLOCK(bp);
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
+		if (be64_to_cpu(b->bb_u.l.bb_leftsib) == NULLDFSBNO)
+			cur->bc_ra[lev] |= XFS_BTCUR_LEFTRA;
+		if (be64_to_cpu(b->bb_u.l.bb_rightsib) == NULLDFSBNO)
+			cur->bc_ra[lev] |= XFS_BTCUR_RIGHTRA;
+	} else {
+		if (be32_to_cpu(b->bb_u.s.bb_leftsib) == NULLAGBLOCK)
+			cur->bc_ra[lev] |= XFS_BTCUR_LEFTRA;
+		if (be32_to_cpu(b->bb_u.s.bb_rightsib) == NULLAGBLOCK)
+			cur->bc_ra[lev] |= XFS_BTCUR_RIGHTRA;
+	}
+}
+
+STATIC int
+xfs_btree_ptr_is_null(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*ptr)
+{
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
+		return be64_to_cpu(ptr->l) == NULLDFSBNO;
+	else
+		return be32_to_cpu(ptr->s) == NULLAGBLOCK;
+}
+
+STATIC void
+xfs_btree_set_ptr_null(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*ptr)
+{
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
+		ptr->l = cpu_to_be64(NULLDFSBNO);
+	else
+		ptr->s = cpu_to_be32(NULLAGBLOCK);
+}
+
+/*
+ * Get/set/init sibling pointers
+ */
+STATIC void
+xfs_btree_get_sibling(
+	struct xfs_btree_cur	*cur,
+	struct xfs_btree_block	*block,
+	union xfs_btree_ptr	*ptr,
+	int			lr)
+{
+	ASSERT(lr == XFS_BB_LEFTSIB || lr == XFS_BB_RIGHTSIB);
+
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
+		if (lr == XFS_BB_RIGHTSIB)
+			ptr->l = block->bb_u.l.bb_rightsib;
+		else
+			ptr->l = block->bb_u.l.bb_leftsib;
+	} else {
+		if (lr == XFS_BB_RIGHTSIB)
+			ptr->s = block->bb_u.s.bb_rightsib;
+		else
+			ptr->s = block->bb_u.s.bb_leftsib;
+	}
+}
+
+STATIC void
+xfs_btree_set_sibling(
+	struct xfs_btree_cur	*cur,
+	struct xfs_btree_block	*block,
+	union xfs_btree_ptr	*ptr,
+	int			lr)
+{
+	ASSERT(lr == XFS_BB_LEFTSIB || lr == XFS_BB_RIGHTSIB);
+
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
+		if (lr == XFS_BB_RIGHTSIB)
+			block->bb_u.l.bb_rightsib = ptr->l;
+		else
+			block->bb_u.l.bb_leftsib = ptr->l;
+	} else {
+		if (lr == XFS_BB_RIGHTSIB)
+			block->bb_u.s.bb_rightsib = ptr->s;
+		else
+			block->bb_u.s.bb_leftsib = ptr->s;
+	}
+}
+
+STATIC void
+xfs_btree_init_block(
+	struct xfs_btree_cur	*cur,
+	int			level,
+	int			numrecs,
+	struct xfs_btree_block	*new)	/* new block */
+{
+	new->bb_magic = cpu_to_be32(xfs_magics[cur->bc_btnum]);
+	new->bb_level = cpu_to_be16(level);
+	new->bb_numrecs = cpu_to_be16(numrecs);
+
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
+		new->bb_u.l.bb_leftsib = cpu_to_be64(NULLDFSBNO);
+		new->bb_u.l.bb_rightsib = cpu_to_be64(NULLDFSBNO);
+	} else {
+		new->bb_u.s.bb_leftsib = cpu_to_be32(NULLAGBLOCK);
+		new->bb_u.s.bb_rightsib = cpu_to_be32(NULLAGBLOCK);
+	}
+}
+
+/*
+ * Return true if ptr is the last record in the btree and
+ * we need to track updateѕ to this record.  The decision
+ * will be further refined in the update_lastrec method.
+ */
+STATIC int
+xfs_btree_is_lastrec(
+	struct xfs_btree_cur	*cur,
+	struct xfs_btree_block	*block,
+	int			level)
+{
+	union xfs_btree_ptr	ptr;
+
+	if (level > 0)
+		return 0;
+	if (!(cur->bc_flags & XFS_BTREE_LASTREC_UPDATE))
+		return 0;
+
+	xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_RIGHTSIB);
+	if (!xfs_btree_ptr_is_null(cur, &ptr))
+		return 0;
+	return 1;
+}
+
+STATIC void
+xfs_btree_buf_to_ptr(
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*bp,
+	union xfs_btree_ptr	*ptr)
+{
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
+		ptr->l = cpu_to_be64(XFS_DADDR_TO_FSB(cur->bc_mp,
+					XFS_BUF_ADDR(bp)));
+	else {
+		ptr->s = cpu_to_be32(xfs_daddr_to_agbno(cur->bc_mp,
+					XFS_BUF_ADDR(bp)));
+	}
+}
+
+STATIC xfs_daddr_t
+xfs_btree_ptr_to_daddr(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*ptr)
+{
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
+		ASSERT(be64_to_cpu(ptr->l) != NULLDFSBNO);
+
+		return XFS_FSB_TO_DADDR(cur->bc_mp, be64_to_cpu(ptr->l));
+	} else {
+		ASSERT(cur->bc_private.a.agno != NULLAGNUMBER);
+		ASSERT(be32_to_cpu(ptr->s) != NULLAGBLOCK);
+
+		return XFS_AGB_TO_DADDR(cur->bc_mp, cur->bc_private.a.agno,
+					be32_to_cpu(ptr->s));
+	}
+}
+
+STATIC void
+xfs_btree_set_refs(
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*bp)
+{
+	switch (cur->bc_btnum) {
+	case XFS_BTNUM_BNO:
+	case XFS_BTNUM_CNT:
+		XFS_BUF_SET_VTYPE_REF(bp, B_FS_MAP, XFS_ALLOC_BTREE_REF);
+		break;
+	case XFS_BTNUM_INO:
+		XFS_BUF_SET_VTYPE_REF(bp, B_FS_INOMAP, XFS_INO_BTREE_REF);
+		break;
+	case XFS_BTNUM_BMAP:
+		XFS_BUF_SET_VTYPE_REF(bp, B_FS_MAP, XFS_BMAP_BTREE_REF);
+		break;
+	default:
+		ASSERT(0);
+	}
+}
+
+STATIC int
+xfs_btree_get_buf_block(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*ptr,
+	int			flags,
+	struct xfs_btree_block	**block,
+	struct xfs_buf		**bpp)
+{
+	struct xfs_mount	*mp = cur->bc_mp;
+	xfs_daddr_t		d;
+
+	/* need to sort out how callers deal with failures first */
+	ASSERT(!(flags & XBF_TRYLOCK));
+
+	d = xfs_btree_ptr_to_daddr(cur, ptr);
+	*bpp = xfs_trans_get_buf(cur->bc_tp, mp->m_ddev_targp, d,
+				 mp->m_bsize, flags);
+
+	ASSERT(*bpp);
+	ASSERT(!XFS_BUF_GETERROR(*bpp));
+
+	*block = XFS_BUF_TO_BLOCK(*bpp);
+	return 0;
+}
+
+/*
+ * Read in the buffer at the given ptr and return the buffer and
+ * the block pointer within the buffer.
+ */
+STATIC int
+xfs_btree_read_buf_block(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*ptr,
+	int			level,
+	int			flags,
+	struct xfs_btree_block	**block,
+	struct xfs_buf		**bpp)
+{
+	struct xfs_mount	*mp = cur->bc_mp;
+	xfs_daddr_t		d;
+	int			error;
+
+	/* need to sort out how callers deal with failures first */
+	ASSERT(!(flags & XBF_TRYLOCK));
+
+	d = xfs_btree_ptr_to_daddr(cur, ptr);
+	error = xfs_trans_read_buf(mp, cur->bc_tp, mp->m_ddev_targp, d,
+				   mp->m_bsize, flags, bpp);
+	if (error)
+		return error;
+
+	ASSERT(*bpp != NULL);
+	ASSERT(!XFS_BUF_GETERROR(*bpp));
+
+	xfs_btree_set_refs(cur, *bpp);
+	*block = XFS_BUF_TO_BLOCK(*bpp);
+
+	error = xfs_btree_check_block(cur, *block, level, *bpp);
+	if (error)
+		xfs_trans_brelse(cur->bc_tp, *bpp);
+	return error;
+}
+
+/*
+ * Copy keys from one btree block to another.
+ */
+STATIC void
+xfs_btree_copy_keys(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_key	*dst_key,
+	union xfs_btree_key	*src_key,
+	int			numkeys)
+{
+	ASSERT(numkeys >= 0);
+	memcpy(dst_key, src_key, numkeys * cur->bc_ops->key_len);
+}
+
+/*
+ * Copy records from one btree block to another.
+ */
+STATIC void
+xfs_btree_copy_recs(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_rec	*dst_rec,
+	union xfs_btree_rec	*src_rec,
+	int			numrecs)
+{
+	ASSERT(numrecs >= 0);
+	memcpy(dst_rec, src_rec, numrecs * cur->bc_ops->rec_len);
+}
+
+/*
+ * Copy block pointers from one btree block to another.
+ */
+STATIC void
+xfs_btree_copy_ptrs(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*dst_ptr,
+	union xfs_btree_ptr	*src_ptr,
+	int			numptrs)
+{
+	ASSERT(numptrs >= 0);
+	memcpy(dst_ptr, src_ptr, numptrs * xfs_btree_ptr_len(cur));
+}
+
+/*
+ * Shift keys one index left/right inside a single btree block.
+ */
+STATIC void
+xfs_btree_shift_keys(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_key	*key,
+	int			dir,
+	int			numkeys)
+{
+	char			*dst_key;
+
+	ASSERT(numkeys >= 0);
+	ASSERT(dir == 1 || dir == -1);
+
+	dst_key = (char *)key + (dir * cur->bc_ops->key_len);
+	memmove(dst_key, key, numkeys * cur->bc_ops->key_len);
+}
+
+/*
+ * Shift records one index left/right inside a single btree block.
+ */
+STATIC void
+xfs_btree_shift_recs(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_rec	*rec,
+	int			dir,
+	int			numrecs)
+{
+	char			*dst_rec;
+
+	ASSERT(numrecs >= 0);
+	ASSERT(dir == 1 || dir == -1);
+
+	dst_rec = (char *)rec + (dir * cur->bc_ops->rec_len);
+	memmove(dst_rec, rec, numrecs * cur->bc_ops->rec_len);
+}
+
+/*
+ * Shift block pointers one index left/right inside a single btree block.
+ */
+STATIC void
+xfs_btree_shift_ptrs(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*ptr,
+	int			dir,
+	int			numptrs)
+{
+	char			*dst_ptr;
+
+	ASSERT(numptrs >= 0);
+	ASSERT(dir == 1 || dir == -1);
+
+	dst_ptr = (char *)ptr + (dir * xfs_btree_ptr_len(cur));
+	memmove(dst_ptr, ptr, numptrs * xfs_btree_ptr_len(cur));
+}
+
+/*
+ * Log key values from the btree block.
+ */
+STATIC void
+xfs_btree_log_keys(
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*bp,
+	int			first,
+	int			last)
+{
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+	XFS_BTREE_TRACE_ARGBII(cur, bp, first, last);
+
+	if (bp) {
+		xfs_trans_log_buf(cur->bc_tp, bp,
+				  xfs_btree_key_offset(cur, first),
+				  xfs_btree_key_offset(cur, last + 1) - 1);
+	} else {
+		xfs_trans_log_inode(cur->bc_tp, cur->bc_private.b.ip,
+				xfs_ilog_fbroot(cur->bc_private.b.whichfork));
+	}
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+}
+
+/*
+ * Log record values from the btree block.
+ */
+void
+xfs_btree_log_recs(
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*bp,
+	int			first,
+	int			last)
+{
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+	XFS_BTREE_TRACE_ARGBII(cur, bp, first, last);
+
+	xfs_trans_log_buf(cur->bc_tp, bp,
+			  xfs_btree_rec_offset(cur, first),
+			  xfs_btree_rec_offset(cur, last + 1) - 1);
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+}
+
+/*
+ * Log block pointer fields from a btree block (nonleaf).
+ */
+STATIC void
+xfs_btree_log_ptrs(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	struct xfs_buf		*bp,	/* buffer containing btree block */
+	int			first,	/* index of first pointer to log */
+	int			last)	/* index of last pointer to log */
+{
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+	XFS_BTREE_TRACE_ARGBII(cur, bp, first, last);
+
+	if (bp) {
+		struct xfs_btree_block	*block = XFS_BUF_TO_BLOCK(bp);
+		int			level = xfs_btree_get_level(block);
+
+		xfs_trans_log_buf(cur->bc_tp, bp,
+				xfs_btree_ptr_offset(cur, first, level),
+				xfs_btree_ptr_offset(cur, last + 1, level) - 1);
+	} else {
+		xfs_trans_log_inode(cur->bc_tp, cur->bc_private.b.ip,
+			xfs_ilog_fbroot(cur->bc_private.b.whichfork));
+	}
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+}
+
+/*
+ * Log fields from a btree block header.
+ */
+void
+xfs_btree_log_block(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	struct xfs_buf		*bp,	/* buffer containing btree block */
+	int			fields)	/* mask of fields: XFS_BB_... */
+{
+	int			first;	/* first byte offset logged */
+	int			last;	/* last byte offset logged */
+	static const short	soffsets[] = {	/* table of offsets (short) */
+		offsetof(struct xfs_btree_block, bb_magic),
+		offsetof(struct xfs_btree_block, bb_level),
+		offsetof(struct xfs_btree_block, bb_numrecs),
+		offsetof(struct xfs_btree_block, bb_u.s.bb_leftsib),
+		offsetof(struct xfs_btree_block, bb_u.s.bb_rightsib),
+		XFS_BTREE_SBLOCK_LEN
+	};
+	static const short	loffsets[] = {	/* table of offsets (long) */
+		offsetof(struct xfs_btree_block, bb_magic),
+		offsetof(struct xfs_btree_block, bb_level),
+		offsetof(struct xfs_btree_block, bb_numrecs),
+		offsetof(struct xfs_btree_block, bb_u.l.bb_leftsib),
+		offsetof(struct xfs_btree_block, bb_u.l.bb_rightsib),
+		XFS_BTREE_LBLOCK_LEN
+	};
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+	XFS_BTREE_TRACE_ARGBI(cur, bp, fields);
+
+	if (bp) {
+		xfs_btree_offsets(fields,
+				  (cur->bc_flags & XFS_BTREE_LONG_PTRS) ?
+					loffsets : soffsets,
+				  XFS_BB_NUM_BITS, &first, &last);
+		xfs_trans_log_buf(cur->bc_tp, bp, first, last);
+	} else {
+		xfs_trans_log_inode(cur->bc_tp, cur->bc_private.b.ip,
+			xfs_ilog_fbroot(cur->bc_private.b.whichfork));
+	}
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+}
+
+/*
+ * Increment cursor by one record at the level.
+ * For nonzero levels the leaf-ward information is untouched.
+ */
+int						/* error */
+xfs_btree_increment(
+	struct xfs_btree_cur	*cur,
+	int			level,
+	int			*stat)		/* success/failure */
+{
+	struct xfs_btree_block	*block;
+	union xfs_btree_ptr	ptr;
+	struct xfs_buf		*bp;
+	int			error;		/* error return value */
+	int			lev;
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+	XFS_BTREE_TRACE_ARGI(cur, level);
+
+	ASSERT(level < cur->bc_nlevels);
+
+	/* Read-ahead to the right at this level. */
+	xfs_btree_readahead(cur, level, XFS_BTCUR_RIGHTRA);
+
+	/* Get a pointer to the btree block. */
+	block = xfs_btree_get_block(cur, level, &bp);
+
+#ifdef DEBUG
+	error = xfs_btree_check_block(cur, block, level, bp);
+	if (error)
+		goto error0;
+#endif
+
+	/* We're done if we remain in the block after the increment. */
+	if (++cur->bc_ptrs[level] <= xfs_btree_get_numrecs(block))
+		goto out1;
+
+	/* Fail if we just went off the right edge of the tree. */
+	xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_RIGHTSIB);
+	if (xfs_btree_ptr_is_null(cur, &ptr))
+		goto out0;
+
+	XFS_BTREE_STATS_INC(cur, increment);
+
+	/*
+	 * March up the tree incrementing pointers.
+	 * Stop when we don't go off the right edge of a block.
+	 */
+	for (lev = level + 1; lev < cur->bc_nlevels; lev++) {
+		block = xfs_btree_get_block(cur, lev, &bp);
+
+#ifdef DEBUG
+		error = xfs_btree_check_block(cur, block, lev, bp);
+		if (error)
+			goto error0;
+#endif
+
+		if (++cur->bc_ptrs[lev] <= xfs_btree_get_numrecs(block))
+			break;
+
+		/* Read-ahead the right block for the next loop. */
+		xfs_btree_readahead(cur, lev, XFS_BTCUR_RIGHTRA);
+	}
+
+	/*
+	 * If we went off the root then we are either seriously
+	 * confused or have the tree root in an inode.
+	 */
+	if (lev == cur->bc_nlevels) {
+		if (cur->bc_flags & XFS_BTREE_ROOT_IN_INODE)
+			goto out0;
+		ASSERT(0);
+		error = EFSCORRUPTED;
+		goto error0;
+	}
+	ASSERT(lev < cur->bc_nlevels);
+
+	/*
+	 * Now walk back down the tree, fixing up the cursor's buffer
+	 * pointers and key numbers.
+	 */
+	for (block = xfs_btree_get_block(cur, lev, &bp); lev > level; ) {
+		union xfs_btree_ptr	*ptrp;
+
+		ptrp = xfs_btree_ptr_addr(cur, cur->bc_ptrs[lev], block);
+		error = xfs_btree_read_buf_block(cur, ptrp, --lev,
+							0, &block, &bp);
+		if (error)
+			goto error0;
+
+		xfs_btree_setbuf(cur, lev, bp);
+		cur->bc_ptrs[lev] = 1;
+	}
+out1:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	*stat = 1;
+	return 0;
+
+out0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	*stat = 0;
+	return 0;
+
+error0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
+	return error;
+}
+
+/*
+ * Decrement cursor by one record at the level.
+ * For nonzero levels the leaf-ward information is untouched.
+ */
+int						/* error */
+xfs_btree_decrement(
+	struct xfs_btree_cur	*cur,
+	int			level,
+	int			*stat)		/* success/failure */
+{
+	struct xfs_btree_block	*block;
+	xfs_buf_t		*bp;
+	int			error;		/* error return value */
+	int			lev;
+	union xfs_btree_ptr	ptr;
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+	XFS_BTREE_TRACE_ARGI(cur, level);
+
+	ASSERT(level < cur->bc_nlevels);
+
+	/* Read-ahead to the left at this level. */
+	xfs_btree_readahead(cur, level, XFS_BTCUR_LEFTRA);
+
+	/* We're done if we remain in the block after the decrement. */
+	if (--cur->bc_ptrs[level] > 0)
+		goto out1;
+
+	/* Get a pointer to the btree block. */
+	block = xfs_btree_get_block(cur, level, &bp);
+
+#ifdef DEBUG
+	error = xfs_btree_check_block(cur, block, level, bp);
+	if (error)
+		goto error0;
+#endif
+
+	/* Fail if we just went off the left edge of the tree. */
+	xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_LEFTSIB);
+	if (xfs_btree_ptr_is_null(cur, &ptr))
+		goto out0;
+
+	XFS_BTREE_STATS_INC(cur, decrement);
+
+	/*
+	 * March up the tree decrementing pointers.
+	 * Stop when we don't go off the left edge of a block.
+	 */
+	for (lev = level + 1; lev < cur->bc_nlevels; lev++) {
+		if (--cur->bc_ptrs[lev] > 0)
+			break;
+		/* Read-ahead the left block for the next loop. */
+		xfs_btree_readahead(cur, lev, XFS_BTCUR_LEFTRA);
+	}
+
+	/*
+	 * If we went off the root then we are seriously confused.
+	 * or the root of the tree is in an inode.
+	 */
+	if (lev == cur->bc_nlevels) {
+		if (cur->bc_flags & XFS_BTREE_ROOT_IN_INODE)
+			goto out0;
+		ASSERT(0);
+		error = EFSCORRUPTED;
+		goto error0;
+	}
+	ASSERT(lev < cur->bc_nlevels);
+
+	/*
+	 * Now walk back down the tree, fixing up the cursor's buffer
+	 * pointers and key numbers.
+	 */
+	for (block = xfs_btree_get_block(cur, lev, &bp); lev > level; ) {
+		union xfs_btree_ptr	*ptrp;
+
+		ptrp = xfs_btree_ptr_addr(cur, cur->bc_ptrs[lev], block);
+		error = xfs_btree_read_buf_block(cur, ptrp, --lev,
+							0, &block, &bp);
+		if (error)
+			goto error0;
+		xfs_btree_setbuf(cur, lev, bp);
+		cur->bc_ptrs[lev] = xfs_btree_get_numrecs(block);
+	}
+out1:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	*stat = 1;
+	return 0;
+
+out0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	*stat = 0;
+	return 0;
+
+error0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
+	return error;
+}
+
+STATIC int
+xfs_btree_lookup_get_block(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	int			level,	/* level in the btree */
+	union xfs_btree_ptr	*pp,	/* ptr to btree block */
+	struct xfs_btree_block	**blkp) /* return btree block */
+{
+	struct xfs_buf		*bp;	/* buffer pointer for btree block */
+	int			error = 0;
+
+	/* special case the root block if in an inode */
+	if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
+	    (level == cur->bc_nlevels - 1)) {
+		*blkp = xfs_btree_get_iroot(cur);
+		return 0;
+	}
+
+	/*
+	 * If the old buffer at this level for the disk address we are
+	 * looking for re-use it.
+	 *
+	 * Otherwise throw it away and get a new one.
+	 */
+	bp = cur->bc_bufs[level];
+	if (bp && XFS_BUF_ADDR(bp) == xfs_btree_ptr_to_daddr(cur, pp)) {
+		*blkp = XFS_BUF_TO_BLOCK(bp);
+		return 0;
+	}
+
+	error = xfs_btree_read_buf_block(cur, pp, level, 0, blkp, &bp);
+	if (error)
+		return error;
+
+	xfs_btree_setbuf(cur, level, bp);
+	return 0;
+}
+
+/*
+ * Get current search key.  For level 0 we don't actually have a key
+ * structure so we make one up from the record.  For all other levels
+ * we just return the right key.
+ */
+STATIC union xfs_btree_key *
+xfs_lookup_get_search_key(
+	struct xfs_btree_cur	*cur,
+	int			level,
+	int			keyno,
+	struct xfs_btree_block	*block,
+	union xfs_btree_key	*kp)
+{
+	if (level == 0) {
+		cur->bc_ops->init_key_from_rec(kp,
+				xfs_btree_rec_addr(cur, keyno, block));
+		return kp;
+	}
+
+	return xfs_btree_key_addr(cur, keyno, block);
+}
+
+/*
+ * Lookup the record.  The cursor is made to point to it, based on dir.
+ * Return 0 if can't find any such record, 1 for success.
+ */
+int					/* error */
+xfs_btree_lookup(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_lookup_t		dir,	/* <=, ==, or >= */
+	int			*stat)	/* success/failure */
+{
+	struct xfs_btree_block	*block;	/* current btree block */
+	__int64_t		diff;	/* difference for the current key */
+	int			error;	/* error return value */
+	int			keyno;	/* current key number */
+	int			level;	/* level in the btree */
+	union xfs_btree_ptr	*pp;	/* ptr to btree block */
+	union xfs_btree_ptr	ptr;	/* ptr to btree block */
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+	XFS_BTREE_TRACE_ARGI(cur, dir);
+
+	XFS_BTREE_STATS_INC(cur, lookup);
+
+	block = NULL;
+	keyno = 0;
+
+	/* initialise start pointer from cursor */
+	cur->bc_ops->init_ptr_from_cur(cur, &ptr);
+	pp = &ptr;
+
+	/*
+	 * Iterate over each level in the btree, starting at the root.
+	 * For each level above the leaves, find the key we need, based
+	 * on the lookup record, then follow the corresponding block
+	 * pointer down to the next level.
+	 */
+	for (level = cur->bc_nlevels - 1, diff = 1; level >= 0; level--) {
+		/* Get the block we need to do the lookup on. */
+		error = xfs_btree_lookup_get_block(cur, level, pp, &block);
+		if (error)
+			goto error0;
+
+		if (diff == 0) {
+			/*
+			 * If we already had a key match at a higher level, we
+			 * know we need to use the first entry in this block.
+			 */
+			keyno = 1;
+		} else {
+			/* Otherwise search this block. Do a binary search. */
+
+			int	high;	/* high entry number */
+			int	low;	/* low entry number */
+
+			/* Set low and high entry numbers, 1-based. */
+			low = 1;
+			high = xfs_btree_get_numrecs(block);
+			if (!high) {
+				/* Block is empty, must be an empty leaf. */
+				ASSERT(level == 0 && cur->bc_nlevels == 1);
+
+				cur->bc_ptrs[0] = dir != XFS_LOOKUP_LE;
+				XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+				*stat = 0;
+				return 0;
+			}
+
+			/* Binary search the block. */
+			while (low <= high) {
+				union xfs_btree_key	key;
+				union xfs_btree_key	*kp;
+
+				XFS_BTREE_STATS_INC(cur, compare);
+
+				/* keyno is average of low and high. */
+				keyno = (low + high) >> 1;
+
+				/* Get current search key */
+				kp = xfs_lookup_get_search_key(cur, level,
+						keyno, block, &key);
+
+				/*
+				 * Compute difference to get next direction:
+				 *  - less than, move right
+				 *  - greater than, move left
+				 *  - equal, we're done
+				 */
+				diff = cur->bc_ops->key_diff(cur, kp);
+				if (diff < 0)
+					low = keyno + 1;
+				else if (diff > 0)
+					high = keyno - 1;
+				else
+					break;
+			}
+		}
+
+		/*
+		 * If there are more levels, set up for the next level
+		 * by getting the block number and filling in the cursor.
+		 */
+		if (level > 0) {
+			/*
+			 * If we moved left, need the previous key number,
+			 * unless there isn't one.
+			 */
+			if (diff > 0 && --keyno < 1)
+				keyno = 1;
+			pp = xfs_btree_ptr_addr(cur, keyno, block);
+
+#ifdef DEBUG
+			error = xfs_btree_check_ptr(cur, pp, 0, level);
+			if (error)
+				goto error0;
+#endif
+			cur->bc_ptrs[level] = keyno;
+		}
+	}
+
+	/* Done with the search. See if we need to adjust the results. */
+	if (dir != XFS_LOOKUP_LE && diff < 0) {
+		keyno++;
+		/*
+		 * If ge search and we went off the end of the block, but it's
+		 * not the last block, we're in the wrong block.
+		 */
+		xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_RIGHTSIB);
+		if (dir == XFS_LOOKUP_GE &&
+		    keyno > xfs_btree_get_numrecs(block) &&
+		    !xfs_btree_ptr_is_null(cur, &ptr)) {
+			int	i;
+
+			cur->bc_ptrs[0] = keyno;
+			error = xfs_btree_increment(cur, 0, &i);
+			if (error)
+				goto error0;
+			XFS_WANT_CORRUPTED_RETURN(i == 1);
+			XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+			*stat = 1;
+			return 0;
+		}
+	} else if (dir == XFS_LOOKUP_LE && diff > 0)
+		keyno--;
+	cur->bc_ptrs[0] = keyno;
+
+	/* Return if we succeeded or not. */
+	if (keyno == 0 || keyno > xfs_btree_get_numrecs(block))
+		*stat = 0;
+	else if (dir != XFS_LOOKUP_EQ || diff == 0)
+		*stat = 1;
+	else
+		*stat = 0;
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	return 0;
+
+error0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
+	return error;
+}
+
+/*
+ * Update keys at all levels from here to the root along the cursor's path.
+ */
+STATIC int
+xfs_btree_updkey(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_key	*keyp,
+	int			level)
+{
+	struct xfs_btree_block	*block;
+	struct xfs_buf		*bp;
+	union xfs_btree_key	*kp;
+	int			ptr;
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+	XFS_BTREE_TRACE_ARGIK(cur, level, keyp);
+
+	ASSERT(!(cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) || level >= 1);
+
+	/*
+	 * Go up the tree from this level toward the root.
+	 * At each level, update the key value to the value input.
+	 * Stop when we reach a level where the cursor isn't pointing
+	 * at the first entry in the block.
+	 */
+	for (ptr = 1; ptr == 1 && level < cur->bc_nlevels; level++) {
+#ifdef DEBUG
+		int		error;
+#endif
+		block = xfs_btree_get_block(cur, level, &bp);
+#ifdef DEBUG
+		error = xfs_btree_check_block(cur, block, level, bp);
+		if (error) {
+			XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
+			return error;
+		}
+#endif
+		ptr = cur->bc_ptrs[level];
+		kp = xfs_btree_key_addr(cur, ptr, block);
+		xfs_btree_copy_keys(cur, kp, keyp, 1);
+		xfs_btree_log_keys(cur, bp, ptr, ptr);
+	}
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	return 0;
+}
+
+/*
+ * Update the record referred to by cur to the value in the
+ * given record. This either works (return 0) or gets an
+ * EFSCORRUPTED error.
+ */
+int
+xfs_btree_update(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_rec	*rec)
+{
+	struct xfs_btree_block	*block;
+	struct xfs_buf		*bp;
+	int			error;
+	int			ptr;
+	union xfs_btree_rec	*rp;
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+	XFS_BTREE_TRACE_ARGR(cur, rec);
+
+	/* Pick up the current block. */
+	block = xfs_btree_get_block(cur, 0, &bp);
+
+#ifdef DEBUG
+	error = xfs_btree_check_block(cur, block, 0, bp);
+	if (error)
+		goto error0;
+#endif
+	/* Get the address of the rec to be updated. */
+	ptr = cur->bc_ptrs[0];
+	rp = xfs_btree_rec_addr(cur, ptr, block);
+
+	/* Fill in the new contents and log them. */
+	xfs_btree_copy_recs(cur, rp, rec, 1);
+	xfs_btree_log_recs(cur, bp, ptr, ptr);
+
+	/*
+	 * If we are tracking the last record in the tree and
+	 * we are at the far right edge of the tree, update it.
+	 */
+	if (xfs_btree_is_lastrec(cur, block, 0)) {
+		cur->bc_ops->update_lastrec(cur, block, rec,
+					    ptr, LASTREC_UPDATE);
+	}
+
+	/* Updating first rec in leaf. Pass new key value up to our parent. */
+	if (ptr == 1) {
+		union xfs_btree_key	key;
+
+		cur->bc_ops->init_key_from_rec(&key, rec);
+		error = xfs_btree_updkey(cur, &key, 1);
+		if (error)
+			goto error0;
+	}
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	return 0;
+
+error0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
+	return error;
+}
+
+/*
+ * Move 1 record left from cur/level if possible.
+ * Update cur to reflect the new path.
+ */
+STATIC int					/* error */
+xfs_btree_lshift(
+	struct xfs_btree_cur	*cur,
+	int			level,
+	int			*stat)		/* success/failure */
+{
+	union xfs_btree_key	key;		/* btree key */
+	struct xfs_buf		*lbp;		/* left buffer pointer */
+	struct xfs_btree_block	*left;		/* left btree block */
+	int			lrecs;		/* left record count */
+	struct xfs_buf		*rbp;		/* right buffer pointer */
+	struct xfs_btree_block	*right;		/* right btree block */
+	int			rrecs;		/* right record count */
+	union xfs_btree_ptr	lptr;		/* left btree pointer */
+	union xfs_btree_key	*rkp = NULL;	/* right btree key */
+	union xfs_btree_ptr	*rpp = NULL;	/* right address pointer */
+	union xfs_btree_rec	*rrp = NULL;	/* right record pointer */
+	int			error;		/* error return value */
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+	XFS_BTREE_TRACE_ARGI(cur, level);
+
+	if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
+	    level == cur->bc_nlevels - 1)
+		goto out0;
+
+	/* Set up variables for this block as "right". */
+	right = xfs_btree_get_block(cur, level, &rbp);
+
+#ifdef DEBUG
+	error = xfs_btree_check_block(cur, right, level, rbp);
+	if (error)
+		goto error0;
+#endif
+
+	/* If we've got no left sibling then we can't shift an entry left. */
+	xfs_btree_get_sibling(cur, right, &lptr, XFS_BB_LEFTSIB);
+	if (xfs_btree_ptr_is_null(cur, &lptr))
+		goto out0;
+
+	/*
+	 * If the cursor entry is the one that would be moved, don't
+	 * do it... it's too complicated.
+	 */
+	if (cur->bc_ptrs[level] <= 1)
+		goto out0;
+
+	/* Set up the left neighbor as "left". */
+	error = xfs_btree_read_buf_block(cur, &lptr, level, 0, &left, &lbp);
+	if (error)
+		goto error0;
+
+	/* If it's full, it can't take another entry. */
+	lrecs = xfs_btree_get_numrecs(left);
+	if (lrecs == cur->bc_ops->get_maxrecs(cur, level))
+		goto out0;
+
+	rrecs = xfs_btree_get_numrecs(right);
+
+	/*
+	 * We add one entry to the left side and remove one for the right side.
+	 * Account for it here, the changes will be updated on disk and logged
+	 * later.
+	 */
+	lrecs++;
+	rrecs--;
+
+	XFS_BTREE_STATS_INC(cur, lshift);
+	XFS_BTREE_STATS_ADD(cur, moves, 1);
+
+	/*
+	 * If non-leaf, copy a key and a ptr to the left block.
+	 * Log the changes to the left block.
+	 */
+	if (level > 0) {
+		/* It's a non-leaf.  Move keys and pointers. */
+		union xfs_btree_key	*lkp;	/* left btree key */
+		union xfs_btree_ptr	*lpp;	/* left address pointer */
+
+		lkp = xfs_btree_key_addr(cur, lrecs, left);
+		rkp = xfs_btree_key_addr(cur, 1, right);
+
+		lpp = xfs_btree_ptr_addr(cur, lrecs, left);
+		rpp = xfs_btree_ptr_addr(cur, 1, right);
+#ifdef DEBUG
+		error = xfs_btree_check_ptr(cur, rpp, 0, level);
+		if (error)
+			goto error0;
+#endif
+		xfs_btree_copy_keys(cur, lkp, rkp, 1);
+		xfs_btree_copy_ptrs(cur, lpp, rpp, 1);
+
+		xfs_btree_log_keys(cur, lbp, lrecs, lrecs);
+		xfs_btree_log_ptrs(cur, lbp, lrecs, lrecs);
+
+		ASSERT(cur->bc_ops->keys_inorder(cur,
+			xfs_btree_key_addr(cur, lrecs - 1, left), lkp));
+	} else {
+		/* It's a leaf.  Move records.  */
+		union xfs_btree_rec	*lrp;	/* left record pointer */
+
+		lrp = xfs_btree_rec_addr(cur, lrecs, left);
+		rrp = xfs_btree_rec_addr(cur, 1, right);
+
+		xfs_btree_copy_recs(cur, lrp, rrp, 1);
+		xfs_btree_log_recs(cur, lbp, lrecs, lrecs);
+
+		ASSERT(cur->bc_ops->recs_inorder(cur,
+			xfs_btree_rec_addr(cur, lrecs - 1, left), lrp));
+	}
+
+	xfs_btree_set_numrecs(left, lrecs);
+	xfs_btree_log_block(cur, lbp, XFS_BB_NUMRECS);
+
+	xfs_btree_set_numrecs(right, rrecs);
+	xfs_btree_log_block(cur, rbp, XFS_BB_NUMRECS);
+
+	/*
+	 * Slide the contents of right down one entry.
+	 */
+	XFS_BTREE_STATS_ADD(cur, moves, rrecs - 1);
+	if (level > 0) {
+		/* It's a nonleaf. operate on keys and ptrs */
+#ifdef DEBUG
+		int			i;		/* loop index */
+
+		for (i = 0; i < rrecs; i++) {
+			error = xfs_btree_check_ptr(cur, rpp, i + 1, level);
+			if (error)
+				goto error0;
+		}
+#endif
+		xfs_btree_shift_keys(cur,
+				xfs_btree_key_addr(cur, 2, right),
+				-1, rrecs);
+		xfs_btree_shift_ptrs(cur,
+				xfs_btree_ptr_addr(cur, 2, right),
+				-1, rrecs);
+
+		xfs_btree_log_keys(cur, rbp, 1, rrecs);
+		xfs_btree_log_ptrs(cur, rbp, 1, rrecs);
+	} else {
+		/* It's a leaf. operate on records */
+		xfs_btree_shift_recs(cur,
+			xfs_btree_rec_addr(cur, 2, right),
+			-1, rrecs);
+		xfs_btree_log_recs(cur, rbp, 1, rrecs);
+
+		/*
+		 * If it's the first record in the block, we'll need a key
+		 * structure to pass up to the next level (updkey).
+		 */
+		cur->bc_ops->init_key_from_rec(&key,
+			xfs_btree_rec_addr(cur, 1, right));
+		rkp = &key;
+	}
+
+	/* Update the parent key values of right. */
+	error = xfs_btree_updkey(cur, rkp, level + 1);
+	if (error)
+		goto error0;
+
+	/* Slide the cursor value left one. */
+	cur->bc_ptrs[level]--;
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	*stat = 1;
+	return 0;
+
+out0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	*stat = 0;
+	return 0;
+
+error0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
+	return error;
+}
+
+/*
+ * Move 1 record right from cur/level if possible.
+ * Update cur to reflect the new path.
+ */
+STATIC int					/* error */
+xfs_btree_rshift(
+	struct xfs_btree_cur	*cur,
+	int			level,
+	int			*stat)		/* success/failure */
+{
+	union xfs_btree_key	key;		/* btree key */
+	struct xfs_buf		*lbp;		/* left buffer pointer */
+	struct xfs_btree_block	*left;		/* left btree block */
+	struct xfs_buf		*rbp;		/* right buffer pointer */
+	struct xfs_btree_block	*right;		/* right btree block */
+	struct xfs_btree_cur	*tcur;		/* temporary btree cursor */
+	union xfs_btree_ptr	rptr;		/* right block pointer */
+	union xfs_btree_key	*rkp;		/* right btree key */
+	int			rrecs;		/* right record count */
+	int			lrecs;		/* left record count */
+	int			error;		/* error return value */
+	int			i;		/* loop counter */
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+	XFS_BTREE_TRACE_ARGI(cur, level);
+
+	if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
+	    (level == cur->bc_nlevels - 1))
+		goto out0;
+
+	/* Set up variables for this block as "left". */
+	left = xfs_btree_get_block(cur, level, &lbp);
+
+#ifdef DEBUG
+	error = xfs_btree_check_block(cur, left, level, lbp);
+	if (error)
+		goto error0;
+#endif
+
+	/* If we've got no right sibling then we can't shift an entry right. */
+	xfs_btree_get_sibling(cur, left, &rptr, XFS_BB_RIGHTSIB);
+	if (xfs_btree_ptr_is_null(cur, &rptr))
+		goto out0;
+
+	/*
+	 * If the cursor entry is the one that would be moved, don't
+	 * do it... it's too complicated.
+	 */
+	lrecs = xfs_btree_get_numrecs(left);
+	if (cur->bc_ptrs[level] >= lrecs)
+		goto out0;
+
+	/* Set up the right neighbor as "right". */
+	error = xfs_btree_read_buf_block(cur, &rptr, level, 0, &right, &rbp);
+	if (error)
+		goto error0;
+
+	/* If it's full, it can't take another entry. */
+	rrecs = xfs_btree_get_numrecs(right);
+	if (rrecs == cur->bc_ops->get_maxrecs(cur, level))
+		goto out0;
+
+	XFS_BTREE_STATS_INC(cur, rshift);
+	XFS_BTREE_STATS_ADD(cur, moves, rrecs);
+
+	/*
+	 * Make a hole at the start of the right neighbor block, then
+	 * copy the last left block entry to the hole.
+	 */
+	if (level > 0) {
+		/* It's a nonleaf. make a hole in the keys and ptrs */
+		union xfs_btree_key	*lkp;
+		union xfs_btree_ptr	*lpp;
+		union xfs_btree_ptr	*rpp;
+
+		lkp = xfs_btree_key_addr(cur, lrecs, left);
+		lpp = xfs_btree_ptr_addr(cur, lrecs, left);
+		rkp = xfs_btree_key_addr(cur, 1, right);
+		rpp = xfs_btree_ptr_addr(cur, 1, right);
+
+#ifdef DEBUG
+		for (i = rrecs - 1; i >= 0; i--) {
+			error = xfs_btree_check_ptr(cur, rpp, i, level);
+			if (error)
+				goto error0;
+		}
+#endif
+
+		xfs_btree_shift_keys(cur, rkp, 1, rrecs);
+		xfs_btree_shift_ptrs(cur, rpp, 1, rrecs);
+
+#ifdef DEBUG
+		error = xfs_btree_check_ptr(cur, lpp, 0, level);
+		if (error)
+			goto error0;
+#endif
+
+		/* Now put the new data in, and log it. */
+		xfs_btree_copy_keys(cur, rkp, lkp, 1);
+		xfs_btree_copy_ptrs(cur, rpp, lpp, 1);
+
+		xfs_btree_log_keys(cur, rbp, 1, rrecs + 1);
+		xfs_btree_log_ptrs(cur, rbp, 1, rrecs + 1);
+
+		ASSERT(cur->bc_ops->keys_inorder(cur, rkp,
+			xfs_btree_key_addr(cur, 2, right)));
+	} else {
+		/* It's a leaf. make a hole in the records */
+		union xfs_btree_rec	*lrp;
+		union xfs_btree_rec	*rrp;
+
+		lrp = xfs_btree_rec_addr(cur, lrecs, left);
+		rrp = xfs_btree_rec_addr(cur, 1, right);
+
+		xfs_btree_shift_recs(cur, rrp, 1, rrecs);
+
+		/* Now put the new data in, and log it. */
+		xfs_btree_copy_recs(cur, rrp, lrp, 1);
+		xfs_btree_log_recs(cur, rbp, 1, rrecs + 1);
+
+		cur->bc_ops->init_key_from_rec(&key, rrp);
+		rkp = &key;
+
+		ASSERT(cur->bc_ops->recs_inorder(cur, rrp,
+			xfs_btree_rec_addr(cur, 2, right)));
+	}
+
+	/*
+	 * Decrement and log left's numrecs, bump and log right's numrecs.
+	 */
+	xfs_btree_set_numrecs(left, --lrecs);
+	xfs_btree_log_block(cur, lbp, XFS_BB_NUMRECS);
+
+	xfs_btree_set_numrecs(right, ++rrecs);
+	xfs_btree_log_block(cur, rbp, XFS_BB_NUMRECS);
+
+	/*
+	 * Using a temporary cursor, update the parent key values of the
+	 * block on the right.
+	 */
+	error = xfs_btree_dup_cursor(cur, &tcur);
+	if (error)
+		goto error0;
+	i = xfs_btree_lastrec(tcur, level);
+	XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+
+	error = xfs_btree_increment(tcur, level, &i);
+	if (error)
+		goto error1;
+
+	error = xfs_btree_updkey(tcur, rkp, level + 1);
+	if (error)
+		goto error1;
+
+	xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	*stat = 1;
+	return 0;
+
+out0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	*stat = 0;
+	return 0;
+
+error0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
+	return error;
+
+error1:
+	XFS_BTREE_TRACE_CURSOR(tcur, XBT_ERROR);
+	xfs_btree_del_cursor(tcur, XFS_BTREE_ERROR);
+	return error;
+}
+
+/*
+ * Split cur/level block in half.
+ * Return new block number and the key to its first
+ * record (to be inserted into parent).
+ */
+STATIC int					/* error */
+xfs_btree_split(
+	struct xfs_btree_cur	*cur,
+	int			level,
+	union xfs_btree_ptr	*ptrp,
+	union xfs_btree_key	*key,
+	struct xfs_btree_cur	**curp,
+	int			*stat)		/* success/failure */
+{
+	union xfs_btree_ptr	lptr;		/* left sibling block ptr */
+	struct xfs_buf		*lbp;		/* left buffer pointer */
+	struct xfs_btree_block	*left;		/* left btree block */
+	union xfs_btree_ptr	rptr;		/* right sibling block ptr */
+	struct xfs_buf		*rbp;		/* right buffer pointer */
+	struct xfs_btree_block	*right;		/* right btree block */
+	union xfs_btree_ptr	rrptr;		/* right-right sibling ptr */
+	struct xfs_buf		*rrbp;		/* right-right buffer pointer */
+	struct xfs_btree_block	*rrblock;	/* right-right btree block */
+	int			lrecs;
+	int			rrecs;
+	int			src_index;
+	int			error;		/* error return value */
+#ifdef DEBUG
+	int			i;
+#endif
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+	XFS_BTREE_TRACE_ARGIPK(cur, level, *ptrp, key);
+
+	XFS_BTREE_STATS_INC(cur, split);
+
+	/* Set up left block (current one). */
+	left = xfs_btree_get_block(cur, level, &lbp);
+
+#ifdef DEBUG
+	error = xfs_btree_check_block(cur, left, level, lbp);
+	if (error)
+		goto error0;
+#endif
+
+	xfs_btree_buf_to_ptr(cur, lbp, &lptr);
+
+	/* Allocate the new block. If we can't do it, we're toast. Give up. */
+	error = cur->bc_ops->alloc_block(cur, &lptr, &rptr, 1, stat);
+	if (error)
+		goto error0;
+	if (*stat == 0)
+		goto out0;
+	XFS_BTREE_STATS_INC(cur, alloc);
+
+	/* Set up the new block as "right". */
+	error = xfs_btree_get_buf_block(cur, &rptr, 0, &right, &rbp);
+	if (error)
+		goto error0;
+
+	/* Fill in the btree header for the new right block. */
+	xfs_btree_init_block(cur, xfs_btree_get_level(left), 0, right);
+
+	/*
+	 * Split the entries between the old and the new block evenly.
+	 * Make sure that if there's an odd number of entries now, that
+	 * each new block will have the same number of entries.
+	 */
+	lrecs = xfs_btree_get_numrecs(left);
+	rrecs = lrecs / 2;
+	if ((lrecs & 1) && cur->bc_ptrs[level] <= rrecs + 1)
+		rrecs++;
+	src_index = (lrecs - rrecs + 1);
+
+	XFS_BTREE_STATS_ADD(cur, moves, rrecs);
+
+	/*
+	 * Copy btree block entries from the left block over to the
+	 * new block, the right. Update the right block and log the
+	 * changes.
+	 */
+	if (level > 0) {
+		/* It's a non-leaf.  Move keys and pointers. */
+		union xfs_btree_key	*lkp;	/* left btree key */
+		union xfs_btree_ptr	*lpp;	/* left address pointer */
+		union xfs_btree_key	*rkp;	/* right btree key */
+		union xfs_btree_ptr	*rpp;	/* right address pointer */
+
+		lkp = xfs_btree_key_addr(cur, src_index, left);
+		lpp = xfs_btree_ptr_addr(cur, src_index, left);
+		rkp = xfs_btree_key_addr(cur, 1, right);
+		rpp = xfs_btree_ptr_addr(cur, 1, right);
+
+#ifdef DEBUG
+		for (i = src_index; i < rrecs; i++) {
+			error = xfs_btree_check_ptr(cur, lpp, i, level);
+			if (error)
+				goto error0;
+		}
+#endif
+
+		xfs_btree_copy_keys(cur, rkp, lkp, rrecs);
+		xfs_btree_copy_ptrs(cur, rpp, lpp, rrecs);
+
+		xfs_btree_log_keys(cur, rbp, 1, rrecs);
+		xfs_btree_log_ptrs(cur, rbp, 1, rrecs);
+
+		/* Grab the keys to the entries moved to the right block */
+		xfs_btree_copy_keys(cur, key, rkp, 1);
+	} else {
+		/* It's a leaf.  Move records.  */
+		union xfs_btree_rec	*lrp;	/* left record pointer */
+		union xfs_btree_rec	*rrp;	/* right record pointer */
+
+		lrp = xfs_btree_rec_addr(cur, src_index, left);
+		rrp = xfs_btree_rec_addr(cur, 1, right);
+
+		xfs_btree_copy_recs(cur, rrp, lrp, rrecs);
+		xfs_btree_log_recs(cur, rbp, 1, rrecs);
+
+		cur->bc_ops->init_key_from_rec(key,
+			xfs_btree_rec_addr(cur, 1, right));
+	}
+
+
+	/*
+	 * Find the left block number by looking in the buffer.
+	 * Adjust numrecs, sibling pointers.
+	 */
+	xfs_btree_get_sibling(cur, left, &rrptr, XFS_BB_RIGHTSIB);
+	xfs_btree_set_sibling(cur, right, &rrptr, XFS_BB_RIGHTSIB);
+	xfs_btree_set_sibling(cur, right, &lptr, XFS_BB_LEFTSIB);
+	xfs_btree_set_sibling(cur, left, &rptr, XFS_BB_RIGHTSIB);
+
+	lrecs -= rrecs;
+	xfs_btree_set_numrecs(left, lrecs);
+	xfs_btree_set_numrecs(right, xfs_btree_get_numrecs(right) + rrecs);
+
+	xfs_btree_log_block(cur, rbp, XFS_BB_ALL_BITS);
+	xfs_btree_log_block(cur, lbp, XFS_BB_NUMRECS | XFS_BB_RIGHTSIB);
+
+	/*
+	 * If there's a block to the new block's right, make that block
+	 * point back to right instead of to left.
+	 */
+	if (!xfs_btree_ptr_is_null(cur, &rrptr)) {
+		error = xfs_btree_read_buf_block(cur, &rrptr, level,
+							0, &rrblock, &rrbp);
+		if (error)
+			goto error0;
+		xfs_btree_set_sibling(cur, rrblock, &rptr, XFS_BB_LEFTSIB);
+		xfs_btree_log_block(cur, rrbp, XFS_BB_LEFTSIB);
+	}
+	/*
+	 * If the cursor is really in the right block, move it there.
+	 * If it's just pointing past the last entry in left, then we'll
+	 * insert there, so don't change anything in that case.
+	 */
+	if (cur->bc_ptrs[level] > lrecs + 1) {
+		xfs_btree_setbuf(cur, level, rbp);
+		cur->bc_ptrs[level] -= lrecs;
+	}
+	/*
+	 * If there are more levels, we'll need another cursor which refers
+	 * the right block, no matter where this cursor was.
+	 */
+	if (level + 1 < cur->bc_nlevels) {
+		error = xfs_btree_dup_cursor(cur, curp);
+		if (error)
+			goto error0;
+		(*curp)->bc_ptrs[level + 1]++;
+	}
+	*ptrp = rptr;
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	*stat = 1;
+	return 0;
+out0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	*stat = 0;
+	return 0;
+
+error0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
+	return error;
+}
+
+/*
+ * Copy the old inode root contents into a real block and make the
+ * broot point to it.
+ */
+int						/* error */
+xfs_btree_new_iroot(
+	struct xfs_btree_cur	*cur,		/* btree cursor */
+	int			*logflags,	/* logging flags for inode */
+	int			*stat)		/* return status - 0 fail */
+{
+	struct xfs_buf		*cbp;		/* buffer for cblock */
+	struct xfs_btree_block	*block;		/* btree block */
+	struct xfs_btree_block	*cblock;	/* child btree block */
+	union xfs_btree_key	*ckp;		/* child key pointer */
+	union xfs_btree_ptr	*cpp;		/* child ptr pointer */
+	union xfs_btree_key	*kp;		/* pointer to btree key */
+	union xfs_btree_ptr	*pp;		/* pointer to block addr */
+	union xfs_btree_ptr	nptr;		/* new block addr */
+	int			level;		/* btree level */
+	int			error;		/* error return code */
+#ifdef DEBUG
+	int			i;		/* loop counter */
+#endif
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+	XFS_BTREE_STATS_INC(cur, newroot);
+
+	ASSERT(cur->bc_flags & XFS_BTREE_ROOT_IN_INODE);
+
+	level = cur->bc_nlevels - 1;
+
+	block = xfs_btree_get_iroot(cur);
+	pp = xfs_btree_ptr_addr(cur, 1, block);
+
+	/* Allocate the new block. If we can't do it, we're toast. Give up. */
+	error = cur->bc_ops->alloc_block(cur, pp, &nptr, 1, stat);
+	if (error)
+		goto error0;
+	if (*stat == 0) {
+		XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+		return 0;
+	}
+	XFS_BTREE_STATS_INC(cur, alloc);
+
+	/* Copy the root into a real block. */
+	error = xfs_btree_get_buf_block(cur, &nptr, 0, &cblock, &cbp);
+	if (error)
+		goto error0;
+
+	memcpy(cblock, block, xfs_btree_block_len(cur));
+
+	be16_add_cpu(&block->bb_level, 1);
+	xfs_btree_set_numrecs(block, 1);
+	cur->bc_nlevels++;
+	cur->bc_ptrs[level + 1] = 1;
+
+	kp = xfs_btree_key_addr(cur, 1, block);
+	ckp = xfs_btree_key_addr(cur, 1, cblock);
+	xfs_btree_copy_keys(cur, ckp, kp, xfs_btree_get_numrecs(cblock));
+
+	cpp = xfs_btree_ptr_addr(cur, 1, cblock);
+#ifdef DEBUG
+	for (i = 0; i < be16_to_cpu(cblock->bb_numrecs); i++) {
+		error = xfs_btree_check_ptr(cur, pp, i, level);
+		if (error)
+			goto error0;
+	}
+#endif
+	xfs_btree_copy_ptrs(cur, cpp, pp, xfs_btree_get_numrecs(cblock));
+
+#ifdef DEBUG
+	error = xfs_btree_check_ptr(cur, &nptr, 0, level);
+	if (error)
+		goto error0;
+#endif
+	xfs_btree_copy_ptrs(cur, pp, &nptr, 1);
+
+	xfs_iroot_realloc(cur->bc_private.b.ip,
+			  1 - xfs_btree_get_numrecs(cblock),
+			  cur->bc_private.b.whichfork);
+
+	xfs_btree_setbuf(cur, level, cbp);
+
+	/*
+	 * Do all this logging at the end so that
+	 * the root is at the right level.
+	 */
+	xfs_btree_log_block(cur, cbp, XFS_BB_ALL_BITS);
+	xfs_btree_log_keys(cur, cbp, 1, be16_to_cpu(cblock->bb_numrecs));
+	xfs_btree_log_ptrs(cur, cbp, 1, be16_to_cpu(cblock->bb_numrecs));
+
+	*logflags |=
+		XFS_ILOG_CORE | xfs_ilog_fbroot(cur->bc_private.b.whichfork);
+	*stat = 1;
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	return 0;
+error0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
+	return error;
+}
+
+/*
+ * Allocate a new root block, fill it in.
+ */
+STATIC int				/* error */
+xfs_btree_new_root(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	int			*stat)	/* success/failure */
+{
+	struct xfs_btree_block	*block;	/* one half of the old root block */
+	struct xfs_buf		*bp;	/* buffer containing block */
+	int			error;	/* error return value */
+	struct xfs_buf		*lbp;	/* left buffer pointer */
+	struct xfs_btree_block	*left;	/* left btree block */
+	struct xfs_buf		*nbp;	/* new (root) buffer */
+	struct xfs_btree_block	*new;	/* new (root) btree block */
+	int			nptr;	/* new value for key index, 1 or 2 */
+	struct xfs_buf		*rbp;	/* right buffer pointer */
+	struct xfs_btree_block	*right;	/* right btree block */
+	union xfs_btree_ptr	rptr;
+	union xfs_btree_ptr	lptr;
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+	XFS_BTREE_STATS_INC(cur, newroot);
+
+	/* initialise our start point from the cursor */
+	cur->bc_ops->init_ptr_from_cur(cur, &rptr);
+
+	/* Allocate the new block. If we can't do it, we're toast. Give up. */
+	error = cur->bc_ops->alloc_block(cur, &rptr, &lptr, 1, stat);
+	if (error)
+		goto error0;
+	if (*stat == 0)
+		goto out0;
+	XFS_BTREE_STATS_INC(cur, alloc);
+
+	/* Set up the new block. */
+	error = xfs_btree_get_buf_block(cur, &lptr, 0, &new, &nbp);
+	if (error)
+		goto error0;
+
+	/* Set the root in the holding structure  increasing the level by 1. */
+	cur->bc_ops->set_root(cur, &lptr, 1);
+
+	/*
+	 * At the previous root level there are now two blocks: the old root,
+	 * and the new block generated when it was split.  We don't know which
+	 * one the cursor is pointing at, so we set up variables "left" and
+	 * "right" for each case.
+	 */
+	block = xfs_btree_get_block(cur, cur->bc_nlevels - 1, &bp);
+
+#ifdef DEBUG
+	error = xfs_btree_check_block(cur, block, cur->bc_nlevels - 1, bp);
+	if (error)
+		goto error0;
+#endif
+
+	xfs_btree_get_sibling(cur, block, &rptr, XFS_BB_RIGHTSIB);
+	if (!xfs_btree_ptr_is_null(cur, &rptr)) {
+		/* Our block is left, pick up the right block. */
+		lbp = bp;
+		xfs_btree_buf_to_ptr(cur, lbp, &lptr);
+		left = block;
+		error = xfs_btree_read_buf_block(cur, &rptr,
+					cur->bc_nlevels - 1, 0, &right, &rbp);
+		if (error)
+			goto error0;
+		bp = rbp;
+		nptr = 1;
+	} else {
+		/* Our block is right, pick up the left block. */
+		rbp = bp;
+		xfs_btree_buf_to_ptr(cur, rbp, &rptr);
+		right = block;
+		xfs_btree_get_sibling(cur, right, &lptr, XFS_BB_LEFTSIB);
+		error = xfs_btree_read_buf_block(cur, &lptr,
+					cur->bc_nlevels - 1, 0, &left, &lbp);
+		if (error)
+			goto error0;
+		bp = lbp;
+		nptr = 2;
+	}
+	/* Fill in the new block's btree header and log it. */
+	xfs_btree_init_block(cur, cur->bc_nlevels, 2, new);
+	xfs_btree_log_block(cur, nbp, XFS_BB_ALL_BITS);
+	ASSERT(!xfs_btree_ptr_is_null(cur, &lptr) &&
+			!xfs_btree_ptr_is_null(cur, &rptr));
+
+	/* Fill in the key data in the new root. */
+	if (xfs_btree_get_level(left) > 0) {
+		xfs_btree_copy_keys(cur,
+				xfs_btree_key_addr(cur, 1, new),
+				xfs_btree_key_addr(cur, 1, left), 1);
+		xfs_btree_copy_keys(cur,
+				xfs_btree_key_addr(cur, 2, new),
+				xfs_btree_key_addr(cur, 1, right), 1);
+	} else {
+		cur->bc_ops->init_key_from_rec(
+				xfs_btree_key_addr(cur, 1, new),
+				xfs_btree_rec_addr(cur, 1, left));
+		cur->bc_ops->init_key_from_rec(
+				xfs_btree_key_addr(cur, 2, new),
+				xfs_btree_rec_addr(cur, 1, right));
+	}
+	xfs_btree_log_keys(cur, nbp, 1, 2);
+
+	/* Fill in the pointer data in the new root. */
+	xfs_btree_copy_ptrs(cur,
+		xfs_btree_ptr_addr(cur, 1, new), &lptr, 1);
+	xfs_btree_copy_ptrs(cur,
+		xfs_btree_ptr_addr(cur, 2, new), &rptr, 1);
+	xfs_btree_log_ptrs(cur, nbp, 1, 2);
+
+	/* Fix up the cursor. */
+	xfs_btree_setbuf(cur, cur->bc_nlevels, nbp);
+	cur->bc_ptrs[cur->bc_nlevels] = nptr;
+	cur->bc_nlevels++;
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	*stat = 1;
+	return 0;
+error0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
+	return error;
+out0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	*stat = 0;
+	return 0;
+}
+
+STATIC int
+xfs_btree_make_block_unfull(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	int			level,	/* btree level */
+	int			numrecs,/* # of recs in block */
+	int			*oindex,/* old tree index */
+	int			*index,	/* new tree index */
+	union xfs_btree_ptr	*nptr,	/* new btree ptr */
+	struct xfs_btree_cur	**ncur,	/* new btree cursor */
+	union xfs_btree_rec	*nrec,	/* new record */
+	int			*stat)
+{
+	union xfs_btree_key	key;	/* new btree key value */
+	int			error = 0;
+
+	if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
+	    level == cur->bc_nlevels - 1) {
+	    	struct xfs_inode *ip = cur->bc_private.b.ip;
+
+		if (numrecs < cur->bc_ops->get_dmaxrecs(cur, level)) {
+			/* A root block that can be made bigger. */
+
+			xfs_iroot_realloc(ip, 1, cur->bc_private.b.whichfork);
+		} else {
+			/* A root block that needs replacing */
+			int	logflags = 0;
+
+			error = xfs_btree_new_iroot(cur, &logflags, stat);
+			if (error || *stat == 0)
+				return error;
+
+			xfs_trans_log_inode(cur->bc_tp, ip, logflags);
+		}
+
+		return 0;
+	}
+
+	/* First, try shifting an entry to the right neighbor. */
+	error = xfs_btree_rshift(cur, level, stat);
+	if (error || *stat)
+		return error;
+
+	/* Next, try shifting an entry to the left neighbor. */
+	error = xfs_btree_lshift(cur, level, stat);
+	if (error)
+		return error;
+
+	if (*stat) {
+		*oindex = *index = cur->bc_ptrs[level];
+		return 0;
+	}
+
+	/*
+	 * Next, try splitting the current block in half.
+	 *
+	 * If this works we have to re-set our variables because we
+	 * could be in a different block now.
+	 */
+	error = xfs_btree_split(cur, level, nptr, &key, ncur, stat);
+	if (error || *stat == 0)
+		return error;
+
+
+	*index = cur->bc_ptrs[level];
+	cur->bc_ops->init_rec_from_key(&key, nrec);
+	return 0;
+}
+
+/*
+ * Insert one record/level.  Return information to the caller
+ * allowing the next level up to proceed if necessary.
+ */
+STATIC int
+xfs_btree_insrec(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	int			level,	/* level to insert record at */
+	union xfs_btree_ptr	*ptrp,	/* i/o: block number inserted */
+	union xfs_btree_rec	*recp,	/* i/o: record data inserted */
+	struct xfs_btree_cur	**curp,	/* output: new cursor replacing cur */
+	int			*stat)	/* success/failure */
+{
+	struct xfs_btree_block	*block;	/* btree block */
+	struct xfs_buf		*bp;	/* buffer for block */
+	union xfs_btree_key	key;	/* btree key */
+	union xfs_btree_ptr	nptr;	/* new block ptr */
+	struct xfs_btree_cur	*ncur;	/* new btree cursor */
+	union xfs_btree_rec	nrec;	/* new record count */
+	int			optr;	/* old key/record index */
+	int			ptr;	/* key/record index */
+	int			numrecs;/* number of records */
+	int			error;	/* error return value */
+#ifdef DEBUG
+	int			i;
+#endif
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+	XFS_BTREE_TRACE_ARGIPR(cur, level, *ptrp, recp);
+
+	ncur = NULL;
+
+	/*
+	 * If we have an external root pointer, and we've made it to the
+	 * root level, allocate a new root block and we're done.
+	 */
+	if (!(cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
+	    (level >= cur->bc_nlevels)) {
+		error = xfs_btree_new_root(cur, stat);
+		xfs_btree_set_ptr_null(cur, ptrp);
+
+		XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+		return error;
+	}
+
+	/* If we're off the left edge, return failure. */
+	ptr = cur->bc_ptrs[level];
+	if (ptr == 0) {
+		XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+		*stat = 0;
+		return 0;
+	}
+
+	/* Make a key out of the record data to be inserted, and save it. */
+	cur->bc_ops->init_key_from_rec(&key, recp);
+
+	optr = ptr;
+
+	XFS_BTREE_STATS_INC(cur, insrec);
+
+	/* Get pointers to the btree buffer and block. */
+	block = xfs_btree_get_block(cur, level, &bp);
+	numrecs = xfs_btree_get_numrecs(block);
+
+#ifdef DEBUG
+	error = xfs_btree_check_block(cur, block, level, bp);
+	if (error)
+		goto error0;
+
+	/* Check that the new entry is being inserted in the right place. */
+	if (ptr <= numrecs) {
+		if (level == 0) {
+			ASSERT(cur->bc_ops->recs_inorder(cur, recp,
+				xfs_btree_rec_addr(cur, ptr, block)));
+		} else {
+			ASSERT(cur->bc_ops->keys_inorder(cur, &key,
+				xfs_btree_key_addr(cur, ptr, block)));
+		}
+	}
+#endif
+
+	/*
+	 * If the block is full, we can't insert the new entry until we
+	 * make the block un-full.
+	 */
+	xfs_btree_set_ptr_null(cur, &nptr);
+	if (numrecs == cur->bc_ops->get_maxrecs(cur, level)) {
+		error = xfs_btree_make_block_unfull(cur, level, numrecs,
+					&optr, &ptr, &nptr, &ncur, &nrec, stat);
+		if (error || *stat == 0)
+			goto error0;
+	}
+
+	/*
+	 * The current block may have changed if the block was
+	 * previously full and we have just made space in it.
+	 */
+	block = xfs_btree_get_block(cur, level, &bp);
+	numrecs = xfs_btree_get_numrecs(block);
+
+#ifdef DEBUG
+	error = xfs_btree_check_block(cur, block, level, bp);
+	if (error)
+		return error;
+#endif
+
+	/*
+	 * At this point we know there's room for our new entry in the block
+	 * we're pointing at.
+	 */
+	XFS_BTREE_STATS_ADD(cur, moves, numrecs - ptr + 1);
+
+	if (level > 0) {
+		/* It's a nonleaf. make a hole in the keys and ptrs */
+		union xfs_btree_key	*kp;
+		union xfs_btree_ptr	*pp;
+
+		kp = xfs_btree_key_addr(cur, ptr, block);
+		pp = xfs_btree_ptr_addr(cur, ptr, block);
+
+#ifdef DEBUG
+		for (i = numrecs - ptr; i >= 0; i--) {
+			error = xfs_btree_check_ptr(cur, pp, i, level);
+			if (error)
+				return error;
+		}
+#endif
+
+		xfs_btree_shift_keys(cur, kp, 1, numrecs - ptr + 1);
+		xfs_btree_shift_ptrs(cur, pp, 1, numrecs - ptr + 1);
+
+#ifdef DEBUG
+		error = xfs_btree_check_ptr(cur, ptrp, 0, level);
+		if (error)
+			goto error0;
+#endif
+
+		/* Now put the new data in, bump numrecs and log it. */
+		xfs_btree_copy_keys(cur, kp, &key, 1);
+		xfs_btree_copy_ptrs(cur, pp, ptrp, 1);
+		numrecs++;
+		xfs_btree_set_numrecs(block, numrecs);
+		xfs_btree_log_ptrs(cur, bp, ptr, numrecs);
+		xfs_btree_log_keys(cur, bp, ptr, numrecs);
+#ifdef DEBUG
+		if (ptr < numrecs) {
+			ASSERT(cur->bc_ops->keys_inorder(cur, kp,
+				xfs_btree_key_addr(cur, ptr + 1, block)));
+		}
+#endif
+	} else {
+		/* It's a leaf. make a hole in the records */
+		union xfs_btree_rec             *rp;
+
+		rp = xfs_btree_rec_addr(cur, ptr, block);
+
+		xfs_btree_shift_recs(cur, rp, 1, numrecs - ptr + 1);
+
+		/* Now put the new data in, bump numrecs and log it. */
+		xfs_btree_copy_recs(cur, rp, recp, 1);
+		xfs_btree_set_numrecs(block, ++numrecs);
+		xfs_btree_log_recs(cur, bp, ptr, numrecs);
+#ifdef DEBUG
+		if (ptr < numrecs) {
+			ASSERT(cur->bc_ops->recs_inorder(cur, rp,
+				xfs_btree_rec_addr(cur, ptr + 1, block)));
+		}
+#endif
+	}
+
+	/* Log the new number of records in the btree header. */
+	xfs_btree_log_block(cur, bp, XFS_BB_NUMRECS);
+
+	/* If we inserted at the start of a block, update the parents' keys. */
+	if (optr == 1) {
+		error = xfs_btree_updkey(cur, &key, level + 1);
+		if (error)
+			goto error0;
+	}
+
+	/*
+	 * If we are tracking the last record in the tree and
+	 * we are at the far right edge of the tree, update it.
+	 */
+	if (xfs_btree_is_lastrec(cur, block, level)) {
+		cur->bc_ops->update_lastrec(cur, block, recp,
+					    ptr, LASTREC_INSREC);
+	}
+
+	/*
+	 * Return the new block number, if any.
+	 * If there is one, give back a record value and a cursor too.
+	 */
+	*ptrp = nptr;
+	if (!xfs_btree_ptr_is_null(cur, &nptr)) {
+		*recp = nrec;
+		*curp = ncur;
+	}
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	*stat = 1;
+	return 0;
+
+error0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
+	return error;
+}
+
+/*
+ * Insert the record at the point referenced by cur.
+ *
+ * A multi-level split of the tree on insert will invalidate the original
+ * cursor.  All callers of this function should assume that the cursor is
+ * no longer valid and revalidate it.
+ */
+int
+xfs_btree_insert(
+	struct xfs_btree_cur	*cur,
+	int			*stat)
+{
+	int			error;	/* error return value */
+	int			i;	/* result value, 0 for failure */
+	int			level;	/* current level number in btree */
+	union xfs_btree_ptr	nptr;	/* new block number (split result) */
+	struct xfs_btree_cur	*ncur;	/* new cursor (split result) */
+	struct xfs_btree_cur	*pcur;	/* previous level's cursor */
+	union xfs_btree_rec	rec;	/* record to insert */
+
+	level = 0;
+	ncur = NULL;
+	pcur = cur;
+
+	xfs_btree_set_ptr_null(cur, &nptr);
+	cur->bc_ops->init_rec_from_cur(cur, &rec);
+
+	/*
+	 * Loop going up the tree, starting at the leaf level.
+	 * Stop when we don't get a split block, that must mean that
+	 * the insert is finished with this level.
+	 */
+	do {
+		/*
+		 * Insert nrec/nptr into this level of the tree.
+		 * Note if we fail, nptr will be null.
+		 */
+		error = xfs_btree_insrec(pcur, level, &nptr, &rec, &ncur, &i);
+		if (error) {
+			if (pcur != cur)
+				xfs_btree_del_cursor(pcur, XFS_BTREE_ERROR);
+			goto error0;
+		}
+
+		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+		level++;
+
+		/*
+		 * See if the cursor we just used is trash.
+		 * Can't trash the caller's cursor, but otherwise we should
+		 * if ncur is a new cursor or we're about to be done.
+		 */
+		if (pcur != cur &&
+		    (ncur || xfs_btree_ptr_is_null(cur, &nptr))) {
+			/* Save the state from the cursor before we trash it */
+			if (cur->bc_ops->update_cursor)
+				cur->bc_ops->update_cursor(pcur, cur);
+			cur->bc_nlevels = pcur->bc_nlevels;
+			xfs_btree_del_cursor(pcur, XFS_BTREE_NOERROR);
+		}
+		/* If we got a new cursor, switch to it. */
+		if (ncur) {
+			pcur = ncur;
+			ncur = NULL;
+		}
+	} while (!xfs_btree_ptr_is_null(cur, &nptr));
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	*stat = i;
+	return 0;
+error0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
+	return error;
+}
+
+/*
+ * Try to merge a non-leaf block back into the inode root.
+ *
+ * Note: the killroot names comes from the fact that we're effectively
+ * killing the old root block.  But because we can't just delete the
+ * inode we have to copy the single block it was pointing to into the
+ * inode.
+ */
+STATIC int
+xfs_btree_kill_iroot(
+	struct xfs_btree_cur	*cur)
+{
+	int			whichfork = cur->bc_private.b.whichfork;
+	struct xfs_inode	*ip = cur->bc_private.b.ip;
+	struct xfs_ifork	*ifp = XFS_IFORK_PTR(ip, whichfork);
+	struct xfs_btree_block	*block;
+	struct xfs_btree_block	*cblock;
+	union xfs_btree_key	*kp;
+	union xfs_btree_key	*ckp;
+	union xfs_btree_ptr	*pp;
+	union xfs_btree_ptr	*cpp;
+	struct xfs_buf		*cbp;
+	int			level;
+	int			index;
+	int			numrecs;
+#ifdef DEBUG
+	union xfs_btree_ptr	ptr;
+	int			i;
+#endif
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+
+	ASSERT(cur->bc_flags & XFS_BTREE_ROOT_IN_INODE);
+	ASSERT(cur->bc_nlevels > 1);
+
+	/*
+	 * Don't deal with the root block needs to be a leaf case.
+	 * We're just going to turn the thing back into extents anyway.
+	 */
+	level = cur->bc_nlevels - 1;
+	if (level == 1)
+		goto out0;
+
+	/*
+	 * Give up if the root has multiple children.
+	 */
+	block = xfs_btree_get_iroot(cur);
+	if (xfs_btree_get_numrecs(block) != 1)
+		goto out0;
+
+	cblock = xfs_btree_get_block(cur, level - 1, &cbp);
+	numrecs = xfs_btree_get_numrecs(cblock);
+
+	/*
+	 * Only do this if the next level will fit.
+	 * Then the data must be copied up to the inode,
+	 * instead of freeing the root you free the next level.
+	 */
+	if (numrecs > cur->bc_ops->get_dmaxrecs(cur, level))
+		goto out0;
+
+	XFS_BTREE_STATS_INC(cur, killroot);
+
+#ifdef DEBUG
+	xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_LEFTSIB);
+	ASSERT(xfs_btree_ptr_is_null(cur, &ptr));
+	xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_RIGHTSIB);
+	ASSERT(xfs_btree_ptr_is_null(cur, &ptr));
+#endif
+
+	index = numrecs - cur->bc_ops->get_maxrecs(cur, level);
+	if (index) {
+		xfs_iroot_realloc(cur->bc_private.b.ip, index,
+				  cur->bc_private.b.whichfork);
+		block = ifp->if_broot;
+	}
+
+	be16_add_cpu(&block->bb_numrecs, index);
+	ASSERT(block->bb_numrecs == cblock->bb_numrecs);
+
+	kp = xfs_btree_key_addr(cur, 1, block);
+	ckp = xfs_btree_key_addr(cur, 1, cblock);
+	xfs_btree_copy_keys(cur, kp, ckp, numrecs);
+
+	pp = xfs_btree_ptr_addr(cur, 1, block);
+	cpp = xfs_btree_ptr_addr(cur, 1, cblock);
+#ifdef DEBUG
+	for (i = 0; i < numrecs; i++) {
+		int		error;
+
+		error = xfs_btree_check_ptr(cur, cpp, i, level - 1);
+		if (error) {
+			XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
+			return error;
+		}
+	}
+#endif
+	xfs_btree_copy_ptrs(cur, pp, cpp, numrecs);
+
+	cur->bc_ops->free_block(cur, cbp);
+	XFS_BTREE_STATS_INC(cur, free);
+
+	cur->bc_bufs[level - 1] = NULL;
+	be16_add_cpu(&block->bb_level, -1);
+	xfs_trans_log_inode(cur->bc_tp, ip,
+		XFS_ILOG_CORE | xfs_ilog_fbroot(cur->bc_private.b.whichfork));
+	cur->bc_nlevels--;
+out0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	return 0;
+}
+
+/*
+ * Kill the current root node, and replace it with it's only child node.
+ */
+STATIC int
+xfs_btree_kill_root(
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*bp,
+	int			level,
+	union xfs_btree_ptr	*newroot)
+{
+	int			error;
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+	XFS_BTREE_STATS_INC(cur, killroot);
+
+	/*
+	 * Update the root pointer, decreasing the level by 1 and then
+	 * free the old root.
+	 */
+	cur->bc_ops->set_root(cur, newroot, -1);
+
+	error = cur->bc_ops->free_block(cur, bp);
+	if (error) {
+		XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
+		return error;
+	}
+
+	XFS_BTREE_STATS_INC(cur, free);
+
+	cur->bc_bufs[level] = NULL;
+	cur->bc_ra[level] = 0;
+	cur->bc_nlevels--;
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	return 0;
+}
+
+STATIC int
+xfs_btree_dec_cursor(
+	struct xfs_btree_cur	*cur,
+	int			level,
+	int			*stat)
+{
+	int			error;
+	int			i;
+
+	if (level > 0) {
+		error = xfs_btree_decrement(cur, level, &i);
+		if (error)
+			return error;
+	}
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	*stat = 1;
+	return 0;
+}
+
+/*
+ * Single level of the btree record deletion routine.
+ * Delete record pointed to by cur/level.
+ * Remove the record from its block then rebalance the tree.
+ * Return 0 for error, 1 for done, 2 to go on to the next level.
+ */
+STATIC int					/* error */
+xfs_btree_delrec(
+	struct xfs_btree_cur	*cur,		/* btree cursor */
+	int			level,		/* level removing record from */
+	int			*stat)		/* fail/done/go-on */
+{
+	struct xfs_btree_block	*block;		/* btree block */
+	union xfs_btree_ptr	cptr;		/* current block ptr */
+	struct xfs_buf		*bp;		/* buffer for block */
+	int			error;		/* error return value */
+	int			i;		/* loop counter */
+	union xfs_btree_key	key;		/* storage for keyp */
+	union xfs_btree_key	*keyp = &key;	/* passed to the next level */
+	union xfs_btree_ptr	lptr;		/* left sibling block ptr */
+	struct xfs_buf		*lbp;		/* left buffer pointer */
+	struct xfs_btree_block	*left;		/* left btree block */
+	int			lrecs = 0;	/* left record count */
+	int			ptr;		/* key/record index */
+	union xfs_btree_ptr	rptr;		/* right sibling block ptr */
+	struct xfs_buf		*rbp;		/* right buffer pointer */
+	struct xfs_btree_block	*right;		/* right btree block */
+	struct xfs_btree_block	*rrblock;	/* right-right btree block */
+	struct xfs_buf		*rrbp;		/* right-right buffer pointer */
+	int			rrecs = 0;	/* right record count */
+	struct xfs_btree_cur	*tcur;		/* temporary btree cursor */
+	int			numrecs;	/* temporary numrec count */
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+	XFS_BTREE_TRACE_ARGI(cur, level);
+
+	tcur = NULL;
+
+	/* Get the index of the entry being deleted, check for nothing there. */
+	ptr = cur->bc_ptrs[level];
+	if (ptr == 0) {
+		XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+		*stat = 0;
+		return 0;
+	}
+
+	/* Get the buffer & block containing the record or key/ptr. */
+	block = xfs_btree_get_block(cur, level, &bp);
+	numrecs = xfs_btree_get_numrecs(block);
+
+#ifdef DEBUG
+	error = xfs_btree_check_block(cur, block, level, bp);
+	if (error)
+		goto error0;
+#endif
+
+	/* Fail if we're off the end of the block. */
+	if (ptr > numrecs) {
+		XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+		*stat = 0;
+		return 0;
+	}
+
+	XFS_BTREE_STATS_INC(cur, delrec);
+	XFS_BTREE_STATS_ADD(cur, moves, numrecs - ptr);
+
+	/* Excise the entries being deleted. */
+	if (level > 0) {
+		/* It's a nonleaf. operate on keys and ptrs */
+		union xfs_btree_key	*lkp;
+		union xfs_btree_ptr	*lpp;
+
+		lkp = xfs_btree_key_addr(cur, ptr + 1, block);
+		lpp = xfs_btree_ptr_addr(cur, ptr + 1, block);
+
+#ifdef DEBUG
+		for (i = 0; i < numrecs - ptr; i++) {
+			error = xfs_btree_check_ptr(cur, lpp, i, level);
+			if (error)
+				goto error0;
+		}
+#endif
+
+		if (ptr < numrecs) {
+			xfs_btree_shift_keys(cur, lkp, -1, numrecs - ptr);
+			xfs_btree_shift_ptrs(cur, lpp, -1, numrecs - ptr);
+			xfs_btree_log_keys(cur, bp, ptr, numrecs - 1);
+			xfs_btree_log_ptrs(cur, bp, ptr, numrecs - 1);
+		}
+
+		/*
+		 * If it's the first record in the block, we'll need to pass a
+		 * key up to the next level (updkey).
+		 */
+		if (ptr == 1)
+			keyp = xfs_btree_key_addr(cur, 1, block);
+	} else {
+		/* It's a leaf. operate on records */
+		if (ptr < numrecs) {
+			xfs_btree_shift_recs(cur,
+				xfs_btree_rec_addr(cur, ptr + 1, block),
+				-1, numrecs - ptr);
+			xfs_btree_log_recs(cur, bp, ptr, numrecs - 1);
+		}
+
+		/*
+		 * If it's the first record in the block, we'll need a key
+		 * structure to pass up to the next level (updkey).
+		 */
+		if (ptr == 1) {
+			cur->bc_ops->init_key_from_rec(&key,
+					xfs_btree_rec_addr(cur, 1, block));
+			keyp = &key;
+		}
+	}
+
+	/*
+	 * Decrement and log the number of entries in the block.
+	 */
+	xfs_btree_set_numrecs(block, --numrecs);
+	xfs_btree_log_block(cur, bp, XFS_BB_NUMRECS);
+
+	/*
+	 * If we are tracking the last record in the tree and
+	 * we are at the far right edge of the tree, update it.
+	 */
+	if (xfs_btree_is_lastrec(cur, block, level)) {
+		cur->bc_ops->update_lastrec(cur, block, NULL,
+					    ptr, LASTREC_DELREC);
+	}
+
+	/*
+	 * We're at the root level.  First, shrink the root block in-memory.
+	 * Try to get rid of the next level down.  If we can't then there's
+	 * nothing left to do.
+	 */
+	if (level == cur->bc_nlevels - 1) {
+		if (cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) {
+			xfs_iroot_realloc(cur->bc_private.b.ip, -1,
+					  cur->bc_private.b.whichfork);
+
+			error = xfs_btree_kill_iroot(cur);
+			if (error)
+				goto error0;
+
+			error = xfs_btree_dec_cursor(cur, level, stat);
+			if (error)
+				goto error0;
+			*stat = 1;
+			return 0;
+		}
+
+		/*
+		 * If this is the root level, and there's only one entry left,
+		 * and it's NOT the leaf level, then we can get rid of this
+		 * level.
+		 */
+		if (numrecs == 1 && level > 0) {
+			union xfs_btree_ptr	*pp;
+			/*
+			 * pp is still set to the first pointer in the block.
+			 * Make it the new root of the btree.
+			 */
+			pp = xfs_btree_ptr_addr(cur, 1, block);
+			error = xfs_btree_kill_root(cur, bp, level, pp);
+			if (error)
+				goto error0;
+		} else if (level > 0) {
+			error = xfs_btree_dec_cursor(cur, level, stat);
+			if (error)
+				goto error0;
+		}
+		*stat = 1;
+		return 0;
+	}
+
+	/*
+	 * If we deleted the leftmost entry in the block, update the
+	 * key values above us in the tree.
+	 */
+	if (ptr == 1) {
+		error = xfs_btree_updkey(cur, keyp, level + 1);
+		if (error)
+			goto error0;
+	}
+
+	/*
+	 * If the number of records remaining in the block is at least
+	 * the minimum, we're done.
+	 */
+	if (numrecs >= cur->bc_ops->get_minrecs(cur, level)) {
+		error = xfs_btree_dec_cursor(cur, level, stat);
+		if (error)
+			goto error0;
+		return 0;
+	}
+
+	/*
+	 * Otherwise, we have to move some records around to keep the
+	 * tree balanced.  Look at the left and right sibling blocks to
+	 * see if we can re-balance by moving only one record.
+	 */
+	xfs_btree_get_sibling(cur, block, &rptr, XFS_BB_RIGHTSIB);
+	xfs_btree_get_sibling(cur, block, &lptr, XFS_BB_LEFTSIB);
+
+	if (cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) {
+		/*
+		 * One child of root, need to get a chance to copy its contents
+		 * into the root and delete it. Can't go up to next level,
+		 * there's nothing to delete there.
+		 */
+		if (xfs_btree_ptr_is_null(cur, &rptr) &&
+		    xfs_btree_ptr_is_null(cur, &lptr) &&
+		    level == cur->bc_nlevels - 2) {
+			error = xfs_btree_kill_iroot(cur);
+			if (!error)
+				error = xfs_btree_dec_cursor(cur, level, stat);
+			if (error)
+				goto error0;
+			return 0;
+		}
+	}
+
+	ASSERT(!xfs_btree_ptr_is_null(cur, &rptr) ||
+	       !xfs_btree_ptr_is_null(cur, &lptr));
+
+	/*
+	 * Duplicate the cursor so our btree manipulations here won't
+	 * disrupt the next level up.
+	 */
+	error = xfs_btree_dup_cursor(cur, &tcur);
+	if (error)
+		goto error0;
+
+	/*
+	 * If there's a right sibling, see if it's ok to shift an entry
+	 * out of it.
+	 */
+	if (!xfs_btree_ptr_is_null(cur, &rptr)) {
+		/*
+		 * Move the temp cursor to the last entry in the next block.
+		 * Actually any entry but the first would suffice.
+		 */
+		i = xfs_btree_lastrec(tcur, level);
+		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+
+		error = xfs_btree_increment(tcur, level, &i);
+		if (error)
+			goto error0;
+		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+
+		i = xfs_btree_lastrec(tcur, level);
+		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+
+		/* Grab a pointer to the block. */
+		right = xfs_btree_get_block(tcur, level, &rbp);
+#ifdef DEBUG
+		error = xfs_btree_check_block(tcur, right, level, rbp);
+		if (error)
+			goto error0;
+#endif
+		/* Grab the current block number, for future use. */
+		xfs_btree_get_sibling(tcur, right, &cptr, XFS_BB_LEFTSIB);
+
+		/*
+		 * If right block is full enough so that removing one entry
+		 * won't make it too empty, and left-shifting an entry out
+		 * of right to us works, we're done.
+		 */
+		if (xfs_btree_get_numrecs(right) - 1 >=
+		    cur->bc_ops->get_minrecs(tcur, level)) {
+			error = xfs_btree_lshift(tcur, level, &i);
+			if (error)
+				goto error0;
+			if (i) {
+				ASSERT(xfs_btree_get_numrecs(block) >=
+				       cur->bc_ops->get_minrecs(tcur, level));
+
+				xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
+				tcur = NULL;
+
+				error = xfs_btree_dec_cursor(cur, level, stat);
+				if (error)
+					goto error0;
+				return 0;
+			}
+		}
+
+		/*
+		 * Otherwise, grab the number of records in right for
+		 * future reference, and fix up the temp cursor to point
+		 * to our block again (last record).
+		 */
+		rrecs = xfs_btree_get_numrecs(right);
+		if (!xfs_btree_ptr_is_null(cur, &lptr)) {
+			i = xfs_btree_firstrec(tcur, level);
+			XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+
+			error = xfs_btree_decrement(tcur, level, &i);
+			if (error)
+				goto error0;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+		}
+	}
+
+	/*
+	 * If there's a left sibling, see if it's ok to shift an entry
+	 * out of it.
+	 */
+	if (!xfs_btree_ptr_is_null(cur, &lptr)) {
+		/*
+		 * Move the temp cursor to the first entry in the
+		 * previous block.
+		 */
+		i = xfs_btree_firstrec(tcur, level);
+		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+
+		error = xfs_btree_decrement(tcur, level, &i);
+		if (error)
+			goto error0;
+		i = xfs_btree_firstrec(tcur, level);
+		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+
+		/* Grab a pointer to the block. */
+		left = xfs_btree_get_block(tcur, level, &lbp);
+#ifdef DEBUG
+		error = xfs_btree_check_block(cur, left, level, lbp);
+		if (error)
+			goto error0;
+#endif
+		/* Grab the current block number, for future use. */
+		xfs_btree_get_sibling(tcur, left, &cptr, XFS_BB_RIGHTSIB);
+
+		/*
+		 * If left block is full enough so that removing one entry
+		 * won't make it too empty, and right-shifting an entry out
+		 * of left to us works, we're done.
+		 */
+		if (xfs_btree_get_numrecs(left) - 1 >=
+		    cur->bc_ops->get_minrecs(tcur, level)) {
+			error = xfs_btree_rshift(tcur, level, &i);
+			if (error)
+				goto error0;
+			if (i) {
+				ASSERT(xfs_btree_get_numrecs(block) >=
+				       cur->bc_ops->get_minrecs(tcur, level));
+				xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
+				tcur = NULL;
+				if (level == 0)
+					cur->bc_ptrs[0]++;
+				XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+				*stat = 1;
+				return 0;
+			}
+		}
+
+		/*
+		 * Otherwise, grab the number of records in right for
+		 * future reference.
+		 */
+		lrecs = xfs_btree_get_numrecs(left);
+	}
+
+	/* Delete the temp cursor, we're done with it. */
+	xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
+	tcur = NULL;
+
+	/* If here, we need to do a join to keep the tree balanced. */
+	ASSERT(!xfs_btree_ptr_is_null(cur, &cptr));
+
+	if (!xfs_btree_ptr_is_null(cur, &lptr) &&
+	    lrecs + xfs_btree_get_numrecs(block) <=
+			cur->bc_ops->get_maxrecs(cur, level)) {
+		/*
+		 * Set "right" to be the starting block,
+		 * "left" to be the left neighbor.
+		 */
+		rptr = cptr;
+		right = block;
+		rbp = bp;
+		error = xfs_btree_read_buf_block(cur, &lptr, level,
+							0, &left, &lbp);
+		if (error)
+			goto error0;
+
+	/*
+	 * If that won't work, see if we can join with the right neighbor block.
+	 */
+	} else if (!xfs_btree_ptr_is_null(cur, &rptr) &&
+		   rrecs + xfs_btree_get_numrecs(block) <=
+			cur->bc_ops->get_maxrecs(cur, level)) {
+		/*
+		 * Set "left" to be the starting block,
+		 * "right" to be the right neighbor.
+		 */
+		lptr = cptr;
+		left = block;
+		lbp = bp;
+		error = xfs_btree_read_buf_block(cur, &rptr, level,
+							0, &right, &rbp);
+		if (error)
+			goto error0;
+
+	/*
+	 * Otherwise, we can't fix the imbalance.
+	 * Just return.  This is probably a logic error, but it's not fatal.
+	 */
+	} else {
+		error = xfs_btree_dec_cursor(cur, level, stat);
+		if (error)
+			goto error0;
+		return 0;
+	}
+
+	rrecs = xfs_btree_get_numrecs(right);
+	lrecs = xfs_btree_get_numrecs(left);
+
+	/*
+	 * We're now going to join "left" and "right" by moving all the stuff
+	 * in "right" to "left" and deleting "right".
+	 */
+	XFS_BTREE_STATS_ADD(cur, moves, rrecs);
+	if (level > 0) {
+		/* It's a non-leaf.  Move keys and pointers. */
+		union xfs_btree_key	*lkp;	/* left btree key */
+		union xfs_btree_ptr	*lpp;	/* left address pointer */
+		union xfs_btree_key	*rkp;	/* right btree key */
+		union xfs_btree_ptr	*rpp;	/* right address pointer */
+
+		lkp = xfs_btree_key_addr(cur, lrecs + 1, left);
+		lpp = xfs_btree_ptr_addr(cur, lrecs + 1, left);
+		rkp = xfs_btree_key_addr(cur, 1, right);
+		rpp = xfs_btree_ptr_addr(cur, 1, right);
+#ifdef DEBUG
+		for (i = 1; i < rrecs; i++) {
+			error = xfs_btree_check_ptr(cur, rpp, i, level);
+			if (error)
+				goto error0;
+		}
+#endif
+		xfs_btree_copy_keys(cur, lkp, rkp, rrecs);
+		xfs_btree_copy_ptrs(cur, lpp, rpp, rrecs);
+
+		xfs_btree_log_keys(cur, lbp, lrecs + 1, lrecs + rrecs);
+		xfs_btree_log_ptrs(cur, lbp, lrecs + 1, lrecs + rrecs);
+	} else {
+		/* It's a leaf.  Move records.  */
+		union xfs_btree_rec	*lrp;	/* left record pointer */
+		union xfs_btree_rec	*rrp;	/* right record pointer */
+
+		lrp = xfs_btree_rec_addr(cur, lrecs + 1, left);
+		rrp = xfs_btree_rec_addr(cur, 1, right);
+
+		xfs_btree_copy_recs(cur, lrp, rrp, rrecs);
+		xfs_btree_log_recs(cur, lbp, lrecs + 1, lrecs + rrecs);
+	}
+
+	XFS_BTREE_STATS_INC(cur, join);
+
+	/*
+	 * Fix up the number of records and right block pointer in the
+	 * surviving block, and log it.
+	 */
+	xfs_btree_set_numrecs(left, lrecs + rrecs);
+	xfs_btree_get_sibling(cur, right, &cptr, XFS_BB_RIGHTSIB),
+	xfs_btree_set_sibling(cur, left, &cptr, XFS_BB_RIGHTSIB);
+	xfs_btree_log_block(cur, lbp, XFS_BB_NUMRECS | XFS_BB_RIGHTSIB);
+
+	/* If there is a right sibling, point it to the remaining block. */
+	xfs_btree_get_sibling(cur, left, &cptr, XFS_BB_RIGHTSIB);
+	if (!xfs_btree_ptr_is_null(cur, &cptr)) {
+		error = xfs_btree_read_buf_block(cur, &cptr, level,
+							0, &rrblock, &rrbp);
+		if (error)
+			goto error0;
+		xfs_btree_set_sibling(cur, rrblock, &lptr, XFS_BB_LEFTSIB);
+		xfs_btree_log_block(cur, rrbp, XFS_BB_LEFTSIB);
+	}
+
+	/* Free the deleted block. */
+	error = cur->bc_ops->free_block(cur, rbp);
+	if (error)
+		goto error0;
+	XFS_BTREE_STATS_INC(cur, free);
+
+	/*
+	 * If we joined with the left neighbor, set the buffer in the
+	 * cursor to the left block, and fix up the index.
+	 */
+	if (bp != lbp) {
+		cur->bc_bufs[level] = lbp;
+		cur->bc_ptrs[level] += lrecs;
+		cur->bc_ra[level] = 0;
+	}
+	/*
+	 * If we joined with the right neighbor and there's a level above
+	 * us, increment the cursor at that level.
+	 */
+	else if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) ||
+		   (level + 1 < cur->bc_nlevels)) {
+		error = xfs_btree_increment(cur, level + 1, &i);
+		if (error)
+			goto error0;
+	}
+
+	/*
+	 * Readjust the ptr at this level if it's not a leaf, since it's
+	 * still pointing at the deletion point, which makes the cursor
+	 * inconsistent.  If this makes the ptr 0, the caller fixes it up.
+	 * We can't use decrement because it would change the next level up.
+	 */
+	if (level > 0)
+		cur->bc_ptrs[level]--;
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	/* Return value means the next level up has something to do. */
+	*stat = 2;
+	return 0;
+
+error0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
+	if (tcur)
+		xfs_btree_del_cursor(tcur, XFS_BTREE_ERROR);
+	return error;
+}
+
+/*
+ * Delete the record pointed to by cur.
+ * The cursor refers to the place where the record was (could be inserted)
+ * when the operation returns.
+ */
+int					/* error */
+xfs_btree_delete(
+	struct xfs_btree_cur	*cur,
+	int			*stat)	/* success/failure */
+{
+	int			error;	/* error return value */
+	int			level;
+	int			i;
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+
+	/*
+	 * Go up the tree, starting at leaf level.
+	 *
+	 * If 2 is returned then a join was done; go to the next level.
+	 * Otherwise we are done.
+	 */
+	for (level = 0, i = 2; i == 2; level++) {
+		error = xfs_btree_delrec(cur, level, &i);
+		if (error)
+			goto error0;
+	}
+
+	if (i == 0) {
+		for (level = 1; level < cur->bc_nlevels; level++) {
+			if (cur->bc_ptrs[level] == 0) {
+				error = xfs_btree_decrement(cur, level, &i);
+				if (error)
+					goto error0;
+				break;
+			}
+		}
+	}
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	*stat = i;
+	return 0;
+error0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
+	return error;
+}
+
+/*
+ * Get the data from the pointed-to record.
+ */
+int					/* error */
+xfs_btree_get_rec(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	union xfs_btree_rec	**recp,	/* output: btree record */
+	int			*stat)	/* output: success/failure */
+{
+	struct xfs_btree_block	*block;	/* btree block */
+	struct xfs_buf		*bp;	/* buffer pointer */
+	int			ptr;	/* record number */
+#ifdef DEBUG
+	int			error;	/* error return value */
+#endif
+
+	ptr = cur->bc_ptrs[0];
+	block = xfs_btree_get_block(cur, 0, &bp);
+
+#ifdef DEBUG
+	error = xfs_btree_check_block(cur, block, 0, bp);
+	if (error)
+		return error;
+#endif
+
+	/*
+	 * Off the right end or left end, return failure.
+	 */
+	if (ptr > xfs_btree_get_numrecs(block) || ptr <= 0) {
+		*stat = 0;
+		return 0;
+	}
+
+	/*
+	 * Point to the record and extract its data.
+	 */
+	*recp = xfs_btree_rec_addr(cur, ptr, block);
+	*stat = 1;
+	return 0;
+}
diff --git a/fs/xfs/xfs_btree.h b/fs/xfs/xfs_btree.h
new file mode 100644
index 00000000..82fafc66
--- /dev/null
+++ b/fs/xfs/xfs_btree.h
@@ -0,0 +1,455 @@
+/*
+ * Copyright (c) 2000-2001,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_BTREE_H__
+#define	__XFS_BTREE_H__
+
+struct xfs_buf;
+struct xfs_bmap_free;
+struct xfs_inode;
+struct xfs_mount;
+struct xfs_trans;
+
+extern kmem_zone_t	*xfs_btree_cur_zone;
+
+/*
+ * This nonsense is to make -wlint happy.
+ */
+#define	XFS_LOOKUP_EQ	((xfs_lookup_t)XFS_LOOKUP_EQi)
+#define	XFS_LOOKUP_LE	((xfs_lookup_t)XFS_LOOKUP_LEi)
+#define	XFS_LOOKUP_GE	((xfs_lookup_t)XFS_LOOKUP_GEi)
+
+#define	XFS_BTNUM_BNO	((xfs_btnum_t)XFS_BTNUM_BNOi)
+#define	XFS_BTNUM_CNT	((xfs_btnum_t)XFS_BTNUM_CNTi)
+#define	XFS_BTNUM_BMAP	((xfs_btnum_t)XFS_BTNUM_BMAPi)
+#define	XFS_BTNUM_INO	((xfs_btnum_t)XFS_BTNUM_INOi)
+
+/*
+ * Generic btree header.
+ *
+ * This is a combination of the actual format used on disk for short and long
+ * format btrees.  The first three fields are shared by both format, but
+ * the pointers are different and should be used with care.
+ *
+ * To get the size of the actual short or long form headers please use
+ * the size macros below.  Never use sizeof(xfs_btree_block).
+ */
+struct xfs_btree_block {
+	__be32		bb_magic;	/* magic number for block type */
+	__be16		bb_level;	/* 0 is a leaf */
+	__be16		bb_numrecs;	/* current # of data records */
+	union {
+		struct {
+			__be32		bb_leftsib;
+			__be32		bb_rightsib;
+		} s;			/* short form pointers */
+		struct	{
+			__be64		bb_leftsib;
+			__be64		bb_rightsib;
+		} l;			/* long form pointers */
+	} bb_u;				/* rest */
+};
+
+#define XFS_BTREE_SBLOCK_LEN	16	/* size of a short form block */
+#define XFS_BTREE_LBLOCK_LEN	24	/* size of a long form block */
+
+
+/*
+ * Generic key, ptr and record wrapper structures.
+ *
+ * These are disk format structures, and are converted where necessary
+ * by the btree specific code that needs to interpret them.
+ */
+union xfs_btree_ptr {
+	__be32			s;	/* short form ptr */
+	__be64			l;	/* long form ptr */
+};
+
+union xfs_btree_key {
+	xfs_bmbt_key_t		bmbt;
+	xfs_bmdr_key_t		bmbr;	/* bmbt root block */
+	xfs_alloc_key_t		alloc;
+	xfs_inobt_key_t		inobt;
+};
+
+union xfs_btree_rec {
+	xfs_bmbt_rec_t		bmbt;
+	xfs_bmdr_rec_t		bmbr;	/* bmbt root block */
+	xfs_alloc_rec_t		alloc;
+	xfs_inobt_rec_t		inobt;
+};
+
+/*
+ * For logging record fields.
+ */
+#define	XFS_BB_MAGIC		0x01
+#define	XFS_BB_LEVEL		0x02
+#define	XFS_BB_NUMRECS		0x04
+#define	XFS_BB_LEFTSIB		0x08
+#define	XFS_BB_RIGHTSIB		0x10
+#define	XFS_BB_NUM_BITS		5
+#define	XFS_BB_ALL_BITS		((1 << XFS_BB_NUM_BITS) - 1)
+
+/*
+ * Magic numbers for btree blocks.
+ */
+extern const __uint32_t	xfs_magics[];
+
+/*
+ * Generic stats interface
+ */
+#define __XFS_BTREE_STATS_INC(type, stat) \
+	XFS_STATS_INC(xs_ ## type ## _2_ ## stat)
+#define XFS_BTREE_STATS_INC(cur, stat)  \
+do {    \
+	switch (cur->bc_btnum) {  \
+	case XFS_BTNUM_BNO: __XFS_BTREE_STATS_INC(abtb, stat); break;	\
+	case XFS_BTNUM_CNT: __XFS_BTREE_STATS_INC(abtc, stat); break;	\
+	case XFS_BTNUM_BMAP: __XFS_BTREE_STATS_INC(bmbt, stat); break;	\
+	case XFS_BTNUM_INO: __XFS_BTREE_STATS_INC(ibt, stat); break;	\
+	case XFS_BTNUM_MAX: ASSERT(0); /* fucking gcc */ ; break;	\
+	}       \
+} while (0)
+
+#define __XFS_BTREE_STATS_ADD(type, stat, val) \
+	XFS_STATS_ADD(xs_ ## type ## _2_ ## stat, val)
+#define XFS_BTREE_STATS_ADD(cur, stat, val)  \
+do {    \
+	switch (cur->bc_btnum) {  \
+	case XFS_BTNUM_BNO: __XFS_BTREE_STATS_ADD(abtb, stat, val); break; \
+	case XFS_BTNUM_CNT: __XFS_BTREE_STATS_ADD(abtc, stat, val); break; \
+	case XFS_BTNUM_BMAP: __XFS_BTREE_STATS_ADD(bmbt, stat, val); break; \
+	case XFS_BTNUM_INO: __XFS_BTREE_STATS_ADD(ibt, stat, val); break; \
+	case XFS_BTNUM_MAX: ASSERT(0); /* fucking gcc */ ; break;	\
+	}       \
+} while (0)
+
+#define	XFS_BTREE_MAXLEVELS	8	/* max of all btrees */
+
+struct xfs_btree_ops {
+	/* size of the key and record structures */
+	size_t	key_len;
+	size_t	rec_len;
+
+	/* cursor operations */
+	struct xfs_btree_cur *(*dup_cursor)(struct xfs_btree_cur *);
+	void	(*update_cursor)(struct xfs_btree_cur *src,
+				 struct xfs_btree_cur *dst);
+
+	/* update btree root pointer */
+	void	(*set_root)(struct xfs_btree_cur *cur,
+			    union xfs_btree_ptr *nptr, int level_change);
+
+	/* block allocation / freeing */
+	int	(*alloc_block)(struct xfs_btree_cur *cur,
+			       union xfs_btree_ptr *start_bno,
+			       union xfs_btree_ptr *new_bno,
+			       int length, int *stat);
+	int	(*free_block)(struct xfs_btree_cur *cur, struct xfs_buf *bp);
+
+	/* update last record information */
+	void	(*update_lastrec)(struct xfs_btree_cur *cur,
+				  struct xfs_btree_block *block,
+				  union xfs_btree_rec *rec,
+				  int ptr, int reason);
+
+	/* records in block/level */
+	int	(*get_minrecs)(struct xfs_btree_cur *cur, int level);
+	int	(*get_maxrecs)(struct xfs_btree_cur *cur, int level);
+
+	/* records on disk.  Matter for the root in inode case. */
+	int	(*get_dmaxrecs)(struct xfs_btree_cur *cur, int level);
+
+	/* init values of btree structures */
+	void	(*init_key_from_rec)(union xfs_btree_key *key,
+				     union xfs_btree_rec *rec);
+	void	(*init_rec_from_key)(union xfs_btree_key *key,
+				     union xfs_btree_rec *rec);
+	void	(*init_rec_from_cur)(struct xfs_btree_cur *cur,
+				     union xfs_btree_rec *rec);
+	void	(*init_ptr_from_cur)(struct xfs_btree_cur *cur,
+				     union xfs_btree_ptr *ptr);
+
+	/* difference between key value and cursor value */
+	__int64_t (*key_diff)(struct xfs_btree_cur *cur,
+			      union xfs_btree_key *key);
+
+#ifdef DEBUG
+	/* check that k1 is lower than k2 */
+	int	(*keys_inorder)(struct xfs_btree_cur *cur,
+				union xfs_btree_key *k1,
+				union xfs_btree_key *k2);
+
+	/* check that r1 is lower than r2 */
+	int	(*recs_inorder)(struct xfs_btree_cur *cur,
+				union xfs_btree_rec *r1,
+				union xfs_btree_rec *r2);
+#endif
+
+	/* btree tracing */
+#ifdef XFS_BTREE_TRACE
+	void		(*trace_enter)(struct xfs_btree_cur *, const char *,
+				       char *, int, int, __psunsigned_t,
+				       __psunsigned_t, __psunsigned_t,
+				       __psunsigned_t, __psunsigned_t,
+				       __psunsigned_t, __psunsigned_t,
+				       __psunsigned_t, __psunsigned_t,
+				       __psunsigned_t, __psunsigned_t);
+	void		(*trace_cursor)(struct xfs_btree_cur *, __uint32_t *,
+					__uint64_t *, __uint64_t *);
+	void		(*trace_key)(struct xfs_btree_cur *,
+				     union xfs_btree_key *, __uint64_t *,
+				     __uint64_t *);
+	void		(*trace_record)(struct xfs_btree_cur *,
+					union xfs_btree_rec *, __uint64_t *,
+					__uint64_t *, __uint64_t *);
+#endif
+};
+
+/*
+ * Reasons for the update_lastrec method to be called.
+ */
+#define LASTREC_UPDATE	0
+#define LASTREC_INSREC	1
+#define LASTREC_DELREC	2
+
+
+/*
+ * Btree cursor structure.
+ * This collects all information needed by the btree code in one place.
+ */
+typedef struct xfs_btree_cur
+{
+	struct xfs_trans	*bc_tp;	/* transaction we're in, if any */
+	struct xfs_mount	*bc_mp;	/* file system mount struct */
+	const struct xfs_btree_ops *bc_ops;
+	uint			bc_flags; /* btree features - below */
+	union {
+		xfs_alloc_rec_incore_t	a;
+		xfs_bmbt_irec_t		b;
+		xfs_inobt_rec_incore_t	i;
+	}		bc_rec;		/* current insert/search record value */
+	struct xfs_buf	*bc_bufs[XFS_BTREE_MAXLEVELS];	/* buf ptr per level */
+	int		bc_ptrs[XFS_BTREE_MAXLEVELS];	/* key/record # */
+	__uint8_t	bc_ra[XFS_BTREE_MAXLEVELS];	/* readahead bits */
+#define	XFS_BTCUR_LEFTRA	1	/* left sibling has been read-ahead */
+#define	XFS_BTCUR_RIGHTRA	2	/* right sibling has been read-ahead */
+	__uint8_t	bc_nlevels;	/* number of levels in the tree */
+	__uint8_t	bc_blocklog;	/* log2(blocksize) of btree blocks */
+	xfs_btnum_t	bc_btnum;	/* identifies which btree type */
+	union {
+		struct {			/* needed for BNO, CNT, INO */
+			struct xfs_buf	*agbp;	/* agf/agi buffer pointer */
+			xfs_agnumber_t	agno;	/* ag number */
+		} a;
+		struct {			/* needed for BMAP */
+			struct xfs_inode *ip;	/* pointer to our inode */
+			struct xfs_bmap_free *flist;	/* list to free after */
+			xfs_fsblock_t	firstblock;	/* 1st blk allocated */
+			int		allocated;	/* count of alloced */
+			short		forksize;	/* fork's inode space */
+			char		whichfork;	/* data or attr fork */
+			char		flags;		/* flags */
+#define	XFS_BTCUR_BPRV_WASDEL	1			/* was delayed */
+		} b;
+	}		bc_private;	/* per-btree type data */
+} xfs_btree_cur_t;
+
+/* cursor flags */
+#define XFS_BTREE_LONG_PTRS		(1<<0)	/* pointers are 64bits long */
+#define XFS_BTREE_ROOT_IN_INODE		(1<<1)	/* root may be variable size */
+#define XFS_BTREE_LASTREC_UPDATE	(1<<2)	/* track last rec externally */
+
+
+#define	XFS_BTREE_NOERROR	0
+#define	XFS_BTREE_ERROR		1
+
+/*
+ * Convert from buffer to btree block header.
+ */
+#define	XFS_BUF_TO_BLOCK(bp)	((struct xfs_btree_block *)XFS_BUF_PTR(bp))
+
+
+/*
+ * Check that block header is ok.
+ */
+int
+xfs_btree_check_block(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	struct xfs_btree_block	*block,	/* generic btree block pointer */
+	int			level,	/* level of the btree block */
+	struct xfs_buf		*bp);	/* buffer containing block, if any */
+
+/*
+ * Check that (long) pointer is ok.
+ */
+int					/* error (0 or EFSCORRUPTED) */
+xfs_btree_check_lptr(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_dfsbno_t		ptr,	/* btree block disk address */
+	int			level);	/* btree block level */
+
+/*
+ * Delete the btree cursor.
+ */
+void
+xfs_btree_del_cursor(
+	xfs_btree_cur_t		*cur,	/* btree cursor */
+	int			error);	/* del because of error */
+
+/*
+ * Duplicate the btree cursor.
+ * Allocate a new one, copy the record, re-get the buffers.
+ */
+int					/* error */
+xfs_btree_dup_cursor(
+	xfs_btree_cur_t		*cur,	/* input cursor */
+	xfs_btree_cur_t		**ncur);/* output cursor */
+
+/*
+ * Get a buffer for the block, return it with no data read.
+ * Long-form addressing.
+ */
+struct xfs_buf *				/* buffer for fsbno */
+xfs_btree_get_bufl(
+	struct xfs_mount	*mp,	/* file system mount point */
+	struct xfs_trans	*tp,	/* transaction pointer */
+	xfs_fsblock_t		fsbno,	/* file system block number */
+	uint			lock);	/* lock flags for get_buf */
+
+/*
+ * Get a buffer for the block, return it with no data read.
+ * Short-form addressing.
+ */
+struct xfs_buf *				/* buffer for agno/agbno */
+xfs_btree_get_bufs(
+	struct xfs_mount	*mp,	/* file system mount point */
+	struct xfs_trans	*tp,	/* transaction pointer */
+	xfs_agnumber_t		agno,	/* allocation group number */
+	xfs_agblock_t		agbno,	/* allocation group block number */
+	uint			lock);	/* lock flags for get_buf */
+
+/*
+ * Check for the cursor referring to the last block at the given level.
+ */
+int					/* 1=is last block, 0=not last block */
+xfs_btree_islastblock(
+	xfs_btree_cur_t		*cur,	/* btree cursor */
+	int			level);	/* level to check */
+
+/*
+ * Compute first and last byte offsets for the fields given.
+ * Interprets the offsets table, which contains struct field offsets.
+ */
+void
+xfs_btree_offsets(
+	__int64_t		fields,	/* bitmask of fields */
+	const short		*offsets,/* table of field offsets */
+	int			nbits,	/* number of bits to inspect */
+	int			*first,	/* output: first byte offset */
+	int			*last);	/* output: last byte offset */
+
+/*
+ * Get a buffer for the block, return it read in.
+ * Long-form addressing.
+ */
+int					/* error */
+xfs_btree_read_bufl(
+	struct xfs_mount	*mp,	/* file system mount point */
+	struct xfs_trans	*tp,	/* transaction pointer */
+	xfs_fsblock_t		fsbno,	/* file system block number */
+	uint			lock,	/* lock flags for read_buf */
+	struct xfs_buf		**bpp,	/* buffer for fsbno */
+	int			refval);/* ref count value for buffer */
+
+/*
+ * Read-ahead the block, don't wait for it, don't return a buffer.
+ * Long-form addressing.
+ */
+void					/* error */
+xfs_btree_reada_bufl(
+	struct xfs_mount	*mp,	/* file system mount point */
+	xfs_fsblock_t		fsbno,	/* file system block number */
+	xfs_extlen_t		count);	/* count of filesystem blocks */
+
+/*
+ * Read-ahead the block, don't wait for it, don't return a buffer.
+ * Short-form addressing.
+ */
+void					/* error */
+xfs_btree_reada_bufs(
+	struct xfs_mount	*mp,	/* file system mount point */
+	xfs_agnumber_t		agno,	/* allocation group number */
+	xfs_agblock_t		agbno,	/* allocation group block number */
+	xfs_extlen_t		count);	/* count of filesystem blocks */
+
+
+/*
+ * Common btree core entry points.
+ */
+int xfs_btree_increment(struct xfs_btree_cur *, int, int *);
+int xfs_btree_decrement(struct xfs_btree_cur *, int, int *);
+int xfs_btree_lookup(struct xfs_btree_cur *, xfs_lookup_t, int *);
+int xfs_btree_update(struct xfs_btree_cur *, union xfs_btree_rec *);
+int xfs_btree_new_iroot(struct xfs_btree_cur *, int *, int *);
+int xfs_btree_insert(struct xfs_btree_cur *, int *);
+int xfs_btree_delete(struct xfs_btree_cur *, int *);
+int xfs_btree_get_rec(struct xfs_btree_cur *, union xfs_btree_rec **, int *);
+
+/*
+ * Internal btree helpers also used by xfs_bmap.c.
+ */
+void xfs_btree_log_block(struct xfs_btree_cur *, struct xfs_buf *, int);
+void xfs_btree_log_recs(struct xfs_btree_cur *, struct xfs_buf *, int, int);
+
+/*
+ * Helpers.
+ */
+static inline int xfs_btree_get_numrecs(struct xfs_btree_block *block)
+{
+	return be16_to_cpu(block->bb_numrecs);
+}
+
+static inline void xfs_btree_set_numrecs(struct xfs_btree_block *block,
+		__uint16_t numrecs)
+{
+	block->bb_numrecs = cpu_to_be16(numrecs);
+}
+
+static inline int xfs_btree_get_level(struct xfs_btree_block *block)
+{
+	return be16_to_cpu(block->bb_level);
+}
+
+
+/*
+ * Min and max functions for extlen, agblock, fileoff, and filblks types.
+ */
+#define	XFS_EXTLEN_MIN(a,b)	min_t(xfs_extlen_t, (a), (b))
+#define	XFS_EXTLEN_MAX(a,b)	max_t(xfs_extlen_t, (a), (b))
+#define	XFS_AGBLOCK_MIN(a,b)	min_t(xfs_agblock_t, (a), (b))
+#define	XFS_AGBLOCK_MAX(a,b)	max_t(xfs_agblock_t, (a), (b))
+#define	XFS_FILEOFF_MIN(a,b)	min_t(xfs_fileoff_t, (a), (b))
+#define	XFS_FILEOFF_MAX(a,b)	max_t(xfs_fileoff_t, (a), (b))
+#define	XFS_FILBLKS_MIN(a,b)	min_t(xfs_filblks_t, (a), (b))
+#define	XFS_FILBLKS_MAX(a,b)	max_t(xfs_filblks_t, (a), (b))
+
+#define	XFS_FSB_SANITY_CHECK(mp,fsb)	\
+	(XFS_FSB_TO_AGNO(mp, fsb) < mp->m_sb.sb_agcount && \
+		XFS_FSB_TO_AGBNO(mp, fsb) < mp->m_sb.sb_agblocks)
+
+#endif	/* __XFS_BTREE_H__ */
diff --git a/fs/xfs/xfs_btree_trace.c b/fs/xfs/xfs_btree_trace.c
new file mode 100644
index 00000000..44ff942a
--- /dev/null
+++ b/fs/xfs/xfs_btree_trace.c
@@ -0,0 +1,249 @@
+/*
+ * Copyright (c) 2008 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_types.h"
+#include "xfs_inum.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_alloc_btree.h"
+#include "xfs_ialloc_btree.h"
+#include "xfs_inode.h"
+#include "xfs_btree.h"
+#include "xfs_btree_trace.h"
+
+STATIC void
+xfs_btree_trace_ptr(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	ptr,
+	__psunsigned_t		*high,
+	__psunsigned_t		*low)
+{
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
+		__u64 val = be64_to_cpu(ptr.l);
+		*high = val >> 32;
+		*low = (int)val;
+	} else {
+		*high = 0;
+		*low = be32_to_cpu(ptr.s);
+	}
+}
+
+/*
+ * Add a trace buffer entry for arguments, for a buffer & 1 integer arg.
+ */
+void
+xfs_btree_trace_argbi(
+	const char		*func,
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*b,
+	int			i,
+	int			line)
+{
+	cur->bc_ops->trace_enter(cur, func, XBT_ARGS, XFS_BTREE_KTRACE_ARGBI,
+				 line, (__psunsigned_t)b, i, 0, 0, 0, 0, 0,
+				 0, 0, 0, 0);
+}
+
+/*
+ * Add a trace buffer entry for arguments, for a buffer & 2 integer args.
+ */
+void
+xfs_btree_trace_argbii(
+	const char		*func,
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*b,
+	int			i0,
+	int			i1,
+	int			line)
+{
+	cur->bc_ops->trace_enter(cur, func, XBT_ARGS, XFS_BTREE_KTRACE_ARGBII,
+				 line, (__psunsigned_t)b, i0, i1, 0, 0, 0, 0,
+				 0, 0, 0, 0);
+}
+
+/*
+ * Add a trace buffer entry for arguments, for 3 block-length args
+ * and an integer arg.
+ */
+void
+xfs_btree_trace_argfffi(
+	const char		*func,
+	struct xfs_btree_cur	*cur,
+	xfs_dfiloff_t		o,
+	xfs_dfsbno_t		b,
+	xfs_dfilblks_t		i,
+	int			j,
+	int			line)
+{
+	cur->bc_ops->trace_enter(cur, func, XBT_ARGS, XFS_BTREE_KTRACE_ARGFFFI,
+				 line,
+				 o >> 32, (int)o,
+				 b >> 32, (int)b,
+				 i >> 32, (int)i,
+				 (int)j, 0, 0, 0, 0);
+}
+
+/*
+ * Add a trace buffer entry for arguments, for one integer arg.
+ */
+void
+xfs_btree_trace_argi(
+	const char		*func,
+	struct xfs_btree_cur	*cur,
+	int			i,
+	int			line)
+{
+	cur->bc_ops->trace_enter(cur, func, XBT_ARGS, XFS_BTREE_KTRACE_ARGI,
+				 line, i, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
+}
+
+/*
+ * Add a trace buffer entry for arguments, for int, fsblock, key.
+ */
+void
+xfs_btree_trace_argipk(
+	const char		*func,
+	struct xfs_btree_cur	*cur,
+	int			i,
+	union xfs_btree_ptr	ptr,
+	union xfs_btree_key	*key,
+	int			line)
+{
+	__psunsigned_t		high, low;
+	__uint64_t		l0, l1;
+
+	xfs_btree_trace_ptr(cur, ptr, &high, &low);
+	cur->bc_ops->trace_key(cur, key, &l0, &l1);
+	cur->bc_ops->trace_enter(cur, func, XBT_ARGS, XFS_BTREE_KTRACE_ARGIPK,
+				 line, i, high, low,
+				 l0 >> 32, (int)l0,
+				 l1 >> 32, (int)l1,
+				 0, 0, 0, 0);
+}
+
+/*
+ * Add a trace buffer entry for arguments, for int, fsblock, rec.
+ */
+void
+xfs_btree_trace_argipr(
+	const char		*func,
+	struct xfs_btree_cur	*cur,
+	int			i,
+	union xfs_btree_ptr	ptr,
+	union xfs_btree_rec	*rec,
+	int			line)
+{
+	__psunsigned_t		high, low;
+	__uint64_t		l0, l1, l2;
+
+	xfs_btree_trace_ptr(cur, ptr, &high, &low);
+	cur->bc_ops->trace_record(cur, rec, &l0, &l1, &l2);
+	cur->bc_ops->trace_enter(cur, func, XBT_ARGS, XFS_BTREE_KTRACE_ARGIPR,
+			      line, i,
+			      high, low,
+			      l0 >> 32, (int)l0,
+			      l1 >> 32, (int)l1,
+			      l2 >> 32, (int)l2,
+			      0, 0);
+}
+
+/*
+ * Add a trace buffer entry for arguments, for int, key.
+ */
+void
+xfs_btree_trace_argik(
+	const char		*func,
+	struct xfs_btree_cur	*cur,
+	int			i,
+	union xfs_btree_key	*key,
+	int			line)
+{
+	__uint64_t		l0, l1;
+
+	cur->bc_ops->trace_key(cur, key, &l0, &l1);
+	cur->bc_ops->trace_enter(cur, func, XBT_ARGS, XFS_BTREE_KTRACE_ARGIK,
+				 line, i,
+				 l0 >> 32, (int)l0,
+				 l1 >> 32, (int)l1,
+				 0, 0, 0, 0, 0, 0);
+}
+
+/*
+ * Add a trace buffer entry for arguments, for record.
+ */
+void
+xfs_btree_trace_argr(
+	const char		*func,
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_rec	*rec,
+	int			line)
+{
+	__uint64_t		l0, l1, l2;
+
+	cur->bc_ops->trace_record(cur, rec, &l0, &l1, &l2);
+	cur->bc_ops->trace_enter(cur, func, XBT_ARGS, XFS_BTREE_KTRACE_ARGR,
+			      line,
+			      l0 >> 32, (int)l0,
+			      l1 >> 32, (int)l1,
+			      l2 >> 32, (int)l2,
+			      0, 0, 0, 0, 0);
+}
+
+/*
+ * Add a trace buffer entry for the cursor/operation.
+ */
+void
+xfs_btree_trace_cursor(
+	const char		*func,
+	struct xfs_btree_cur	*cur,
+	int			type,
+	int			line)
+{
+	__uint32_t		s0;
+	__uint64_t		l0, l1;
+	char			*s;
+
+	switch (type) {
+	case XBT_ARGS:
+		s = "args";
+		break;
+	case XBT_ENTRY:
+		s = "entry";
+		break;
+	case XBT_ERROR:
+		s = "error";
+		break;
+	case XBT_EXIT:
+		s = "exit";
+		break;
+	default:
+		s = "unknown";
+		break;
+	}
+
+	cur->bc_ops->trace_cursor(cur, &s0, &l0, &l1);
+	cur->bc_ops->trace_enter(cur, func, s, XFS_BTREE_KTRACE_CUR, line,
+				 s0,
+				 l0 >> 32, (int)l0,
+				 l1 >> 32, (int)l1,
+				 (__psunsigned_t)cur->bc_bufs[0],
+				 (__psunsigned_t)cur->bc_bufs[1],
+				 (__psunsigned_t)cur->bc_bufs[2],
+				 (__psunsigned_t)cur->bc_bufs[3],
+				 (cur->bc_ptrs[0] << 16) | cur->bc_ptrs[1],
+				 (cur->bc_ptrs[2] << 16) | cur->bc_ptrs[3]);
+}
diff --git a/fs/xfs/xfs_btree_trace.h b/fs/xfs/xfs_btree_trace.h
new file mode 100644
index 00000000..2d8a3098
--- /dev/null
+++ b/fs/xfs/xfs_btree_trace.h
@@ -0,0 +1,99 @@
+/*
+ * Copyright (c) 2008 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_BTREE_TRACE_H__
+#define	__XFS_BTREE_TRACE_H__
+
+struct xfs_btree_cur;
+struct xfs_buf;
+
+
+/*
+ * Trace hooks.
+ * i,j = integer (32 bit)
+ * b = btree block buffer (xfs_buf_t)
+ * p = btree ptr
+ * r = btree record
+ * k = btree key
+ */
+
+#ifdef XFS_BTREE_TRACE
+
+/*
+ * Trace buffer entry types.
+ */
+#define XFS_BTREE_KTRACE_ARGBI   1
+#define XFS_BTREE_KTRACE_ARGBII  2
+#define XFS_BTREE_KTRACE_ARGFFFI 3
+#define XFS_BTREE_KTRACE_ARGI    4
+#define XFS_BTREE_KTRACE_ARGIPK  5
+#define XFS_BTREE_KTRACE_ARGIPR  6
+#define XFS_BTREE_KTRACE_ARGIK   7
+#define XFS_BTREE_KTRACE_ARGR	 8
+#define XFS_BTREE_KTRACE_CUR     9
+
+/*
+ * Sub-types for cursor traces.
+ */
+#define XBT_ARGS	0
+#define XBT_ENTRY	1
+#define XBT_ERROR	2
+#define XBT_EXIT	3
+
+void xfs_btree_trace_argbi(const char *, struct xfs_btree_cur *,
+		struct xfs_buf *, int, int);
+void xfs_btree_trace_argbii(const char *, struct xfs_btree_cur *,
+		struct xfs_buf *, int, int, int);
+void xfs_btree_trace_argi(const char *, struct xfs_btree_cur *, int, int);
+void xfs_btree_trace_argipk(const char *, struct xfs_btree_cur *, int,
+		union xfs_btree_ptr, union xfs_btree_key *, int);
+void xfs_btree_trace_argipr(const char *, struct xfs_btree_cur *, int,
+		union xfs_btree_ptr, union xfs_btree_rec *, int);
+void xfs_btree_trace_argik(const char *, struct xfs_btree_cur *, int,
+		union xfs_btree_key *, int);
+void xfs_btree_trace_argr(const char *, struct xfs_btree_cur *,
+		union xfs_btree_rec *, int);
+void xfs_btree_trace_cursor(const char *, struct xfs_btree_cur *, int, int);
+
+#define	XFS_BTREE_TRACE_ARGBI(c, b, i)	\
+	xfs_btree_trace_argbi(__func__, c, b, i, __LINE__)
+#define	XFS_BTREE_TRACE_ARGBII(c, b, i, j)	\
+	xfs_btree_trace_argbii(__func__, c, b, i, j, __LINE__)
+#define	XFS_BTREE_TRACE_ARGI(c, i)	\
+	xfs_btree_trace_argi(__func__, c, i, __LINE__)
+#define	XFS_BTREE_TRACE_ARGIPK(c, i, p, k)	\
+	xfs_btree_trace_argipk(__func__, c, i, p, k, __LINE__)
+#define	XFS_BTREE_TRACE_ARGIPR(c, i, p, r)	\
+	xfs_btree_trace_argipr(__func__, c, i, p, r, __LINE__)
+#define	XFS_BTREE_TRACE_ARGIK(c, i, k)	\
+	xfs_btree_trace_argik(__func__, c, i, k, __LINE__)
+#define XFS_BTREE_TRACE_ARGR(c, r)	\
+	xfs_btree_trace_argr(__func__, c, r, __LINE__)
+#define	XFS_BTREE_TRACE_CURSOR(c, t)	\
+	xfs_btree_trace_cursor(__func__, c, t, __LINE__)
+#else
+#define	XFS_BTREE_TRACE_ARGBI(c, b, i)
+#define	XFS_BTREE_TRACE_ARGBII(c, b, i, j)
+#define	XFS_BTREE_TRACE_ARGI(c, i)
+#define	XFS_BTREE_TRACE_ARGIPK(c, i, p, s)
+#define	XFS_BTREE_TRACE_ARGIPR(c, i, p, r)
+#define	XFS_BTREE_TRACE_ARGIK(c, i, k)
+#define XFS_BTREE_TRACE_ARGR(c, r)
+#define	XFS_BTREE_TRACE_CURSOR(c, t)
+#endif	/* XFS_BTREE_TRACE */
+
+#endif /* __XFS_BTREE_TRACE_H__ */
diff --git a/fs/xfs/xfs_buf_item.c b/fs/xfs/xfs_buf_item.c
new file mode 100644
index 00000000..7888a756
--- /dev/null
+++ b/fs/xfs/xfs_buf_item.c
@@ -0,0 +1,1064 @@
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_mount.h"
+#include "xfs_buf_item.h"
+#include "xfs_trans_priv.h"
+#include "xfs_error.h"
+#include "xfs_trace.h"
+
+
+kmem_zone_t	*xfs_buf_item_zone;
+
+static inline struct xfs_buf_log_item *BUF_ITEM(struct xfs_log_item *lip)
+{
+	return container_of(lip, struct xfs_buf_log_item, bli_item);
+}
+
+
+#ifdef XFS_TRANS_DEBUG
+/*
+ * This function uses an alternate strategy for tracking the bytes
+ * that the user requests to be logged.  This can then be used
+ * in conjunction with the bli_orig array in the buf log item to
+ * catch bugs in our callers' code.
+ *
+ * We also double check the bits set in xfs_buf_item_log using a
+ * simple algorithm to check that every byte is accounted for.
+ */
+STATIC void
+xfs_buf_item_log_debug(
+	xfs_buf_log_item_t	*bip,
+	uint			first,
+	uint			last)
+{
+	uint	x;
+	uint	byte;
+	uint	nbytes;
+	uint	chunk_num;
+	uint	word_num;
+	uint	bit_num;
+	uint	bit_set;
+	uint	*wordp;
+
+	ASSERT(bip->bli_logged != NULL);
+	byte = first;
+	nbytes = last - first + 1;
+	bfset(bip->bli_logged, first, nbytes);
+	for (x = 0; x < nbytes; x++) {
+		chunk_num = byte >> XFS_BLF_SHIFT;
+		word_num = chunk_num >> BIT_TO_WORD_SHIFT;
+		bit_num = chunk_num & (NBWORD - 1);
+		wordp = &(bip->bli_format.blf_data_map[word_num]);
+		bit_set = *wordp & (1 << bit_num);
+		ASSERT(bit_set);
+		byte++;
+	}
+}
+
+/*
+ * This function is called when we flush something into a buffer without
+ * logging it.  This happens for things like inodes which are logged
+ * separately from the buffer.
+ */
+void
+xfs_buf_item_flush_log_debug(
+	xfs_buf_t	*bp,
+	uint		first,
+	uint		last)
+{
+	xfs_buf_log_item_t	*bip;
+	uint			nbytes;
+
+	bip = XFS_BUF_FSPRIVATE(bp, xfs_buf_log_item_t*);
+	if ((bip == NULL) || (bip->bli_item.li_type != XFS_LI_BUF)) {
+		return;
+	}
+
+	ASSERT(bip->bli_logged != NULL);
+	nbytes = last - first + 1;
+	bfset(bip->bli_logged, first, nbytes);
+}
+
+/*
+ * This function is called to verify that our callers have logged
+ * all the bytes that they changed.
+ *
+ * It does this by comparing the original copy of the buffer stored in
+ * the buf log item's bli_orig array to the current copy of the buffer
+ * and ensuring that all bytes which mismatch are set in the bli_logged
+ * array of the buf log item.
+ */
+STATIC void
+xfs_buf_item_log_check(
+	xfs_buf_log_item_t	*bip)
+{
+	char		*orig;
+	char		*buffer;
+	int		x;
+	xfs_buf_t	*bp;
+
+	ASSERT(bip->bli_orig != NULL);
+	ASSERT(bip->bli_logged != NULL);
+
+	bp = bip->bli_buf;
+	ASSERT(XFS_BUF_COUNT(bp) > 0);
+	ASSERT(XFS_BUF_PTR(bp) != NULL);
+	orig = bip->bli_orig;
+	buffer = XFS_BUF_PTR(bp);
+	for (x = 0; x < XFS_BUF_COUNT(bp); x++) {
+		if (orig[x] != buffer[x] && !btst(bip->bli_logged, x)) {
+			xfs_emerg(bp->b_mount,
+				"%s: bip %x buffer %x orig %x index %d",
+				__func__, bip, bp, orig, x);
+			ASSERT(0);
+		}
+	}
+}
+#else
+#define		xfs_buf_item_log_debug(x,y,z)
+#define		xfs_buf_item_log_check(x)
+#endif
+
+STATIC void	xfs_buf_do_callbacks(struct xfs_buf *bp);
+
+/*
+ * This returns the number of log iovecs needed to log the
+ * given buf log item.
+ *
+ * It calculates this as 1 iovec for the buf log format structure
+ * and 1 for each stretch of non-contiguous chunks to be logged.
+ * Contiguous chunks are logged in a single iovec.
+ *
+ * If the XFS_BLI_STALE flag has been set, then log nothing.
+ */
+STATIC uint
+xfs_buf_item_size(
+	struct xfs_log_item	*lip)
+{
+	struct xfs_buf_log_item	*bip = BUF_ITEM(lip);
+	struct xfs_buf		*bp = bip->bli_buf;
+	uint			nvecs;
+	int			next_bit;
+	int			last_bit;
+
+	ASSERT(atomic_read(&bip->bli_refcount) > 0);
+	if (bip->bli_flags & XFS_BLI_STALE) {
+		/*
+		 * The buffer is stale, so all we need to log
+		 * is the buf log format structure with the
+		 * cancel flag in it.
+		 */
+		trace_xfs_buf_item_size_stale(bip);
+		ASSERT(bip->bli_format.blf_flags & XFS_BLF_CANCEL);
+		return 1;
+	}
+
+	ASSERT(bip->bli_flags & XFS_BLI_LOGGED);
+	nvecs = 1;
+	last_bit = xfs_next_bit(bip->bli_format.blf_data_map,
+					 bip->bli_format.blf_map_size, 0);
+	ASSERT(last_bit != -1);
+	nvecs++;
+	while (last_bit != -1) {
+		/*
+		 * This takes the bit number to start looking from and
+		 * returns the next set bit from there.  It returns -1
+		 * if there are no more bits set or the start bit is
+		 * beyond the end of the bitmap.
+		 */
+		next_bit = xfs_next_bit(bip->bli_format.blf_data_map,
+						 bip->bli_format.blf_map_size,
+						 last_bit + 1);
+		/*
+		 * If we run out of bits, leave the loop,
+		 * else if we find a new set of bits bump the number of vecs,
+		 * else keep scanning the current set of bits.
+		 */
+		if (next_bit == -1) {
+			last_bit = -1;
+		} else if (next_bit != last_bit + 1) {
+			last_bit = next_bit;
+			nvecs++;
+		} else if (xfs_buf_offset(bp, next_bit * XFS_BLF_CHUNK) !=
+			   (xfs_buf_offset(bp, last_bit * XFS_BLF_CHUNK) +
+			    XFS_BLF_CHUNK)) {
+			last_bit = next_bit;
+			nvecs++;
+		} else {
+			last_bit++;
+		}
+	}
+
+	trace_xfs_buf_item_size(bip);
+	return nvecs;
+}
+
+/*
+ * This is called to fill in the vector of log iovecs for the
+ * given log buf item.  It fills the first entry with a buf log
+ * format structure, and the rest point to contiguous chunks
+ * within the buffer.
+ */
+STATIC void
+xfs_buf_item_format(
+	struct xfs_log_item	*lip,
+	struct xfs_log_iovec	*vecp)
+{
+	struct xfs_buf_log_item	*bip = BUF_ITEM(lip);
+	struct xfs_buf	*bp = bip->bli_buf;
+	uint		base_size;
+	uint		nvecs;
+	int		first_bit;
+	int		last_bit;
+	int		next_bit;
+	uint		nbits;
+	uint		buffer_offset;
+
+	ASSERT(atomic_read(&bip->bli_refcount) > 0);
+	ASSERT((bip->bli_flags & XFS_BLI_LOGGED) ||
+	       (bip->bli_flags & XFS_BLI_STALE));
+
+	/*
+	 * The size of the base structure is the size of the
+	 * declared structure plus the space for the extra words
+	 * of the bitmap.  We subtract one from the map size, because
+	 * the first element of the bitmap is accounted for in the
+	 * size of the base structure.
+	 */
+	base_size =
+		(uint)(sizeof(xfs_buf_log_format_t) +
+		       ((bip->bli_format.blf_map_size - 1) * sizeof(uint)));
+	vecp->i_addr = &bip->bli_format;
+	vecp->i_len = base_size;
+	vecp->i_type = XLOG_REG_TYPE_BFORMAT;
+	vecp++;
+	nvecs = 1;
+
+	/*
+	 * If it is an inode buffer, transfer the in-memory state to the
+	 * format flags and clear the in-memory state. We do not transfer
+	 * this state if the inode buffer allocation has not yet been committed
+	 * to the log as setting the XFS_BLI_INODE_BUF flag will prevent
+	 * correct replay of the inode allocation.
+	 */
+	if (bip->bli_flags & XFS_BLI_INODE_BUF) {
+		if (!((bip->bli_flags & XFS_BLI_INODE_ALLOC_BUF) &&
+		      xfs_log_item_in_current_chkpt(lip)))
+			bip->bli_format.blf_flags |= XFS_BLF_INODE_BUF;
+		bip->bli_flags &= ~XFS_BLI_INODE_BUF;
+	}
+
+	if (bip->bli_flags & XFS_BLI_STALE) {
+		/*
+		 * The buffer is stale, so all we need to log
+		 * is the buf log format structure with the
+		 * cancel flag in it.
+		 */
+		trace_xfs_buf_item_format_stale(bip);
+		ASSERT(bip->bli_format.blf_flags & XFS_BLF_CANCEL);
+		bip->bli_format.blf_size = nvecs;
+		return;
+	}
+
+	/*
+	 * Fill in an iovec for each set of contiguous chunks.
+	 */
+	first_bit = xfs_next_bit(bip->bli_format.blf_data_map,
+					 bip->bli_format.blf_map_size, 0);
+	ASSERT(first_bit != -1);
+	last_bit = first_bit;
+	nbits = 1;
+	for (;;) {
+		/*
+		 * This takes the bit number to start looking from and
+		 * returns the next set bit from there.  It returns -1
+		 * if there are no more bits set or the start bit is
+		 * beyond the end of the bitmap.
+		 */
+		next_bit = xfs_next_bit(bip->bli_format.blf_data_map,
+						 bip->bli_format.blf_map_size,
+						 (uint)last_bit + 1);
+		/*
+		 * If we run out of bits fill in the last iovec and get
+		 * out of the loop.
+		 * Else if we start a new set of bits then fill in the
+		 * iovec for the series we were looking at and start
+		 * counting the bits in the new one.
+		 * Else we're still in the same set of bits so just
+		 * keep counting and scanning.
+		 */
+		if (next_bit == -1) {
+			buffer_offset = first_bit * XFS_BLF_CHUNK;
+			vecp->i_addr = xfs_buf_offset(bp, buffer_offset);
+			vecp->i_len = nbits * XFS_BLF_CHUNK;
+			vecp->i_type = XLOG_REG_TYPE_BCHUNK;
+			nvecs++;
+			break;
+		} else if (next_bit != last_bit + 1) {
+			buffer_offset = first_bit * XFS_BLF_CHUNK;
+			vecp->i_addr = xfs_buf_offset(bp, buffer_offset);
+			vecp->i_len = nbits * XFS_BLF_CHUNK;
+			vecp->i_type = XLOG_REG_TYPE_BCHUNK;
+			nvecs++;
+			vecp++;
+			first_bit = next_bit;
+			last_bit = next_bit;
+			nbits = 1;
+		} else if (xfs_buf_offset(bp, next_bit << XFS_BLF_SHIFT) !=
+			   (xfs_buf_offset(bp, last_bit << XFS_BLF_SHIFT) +
+			    XFS_BLF_CHUNK)) {
+			buffer_offset = first_bit * XFS_BLF_CHUNK;
+			vecp->i_addr = xfs_buf_offset(bp, buffer_offset);
+			vecp->i_len = nbits * XFS_BLF_CHUNK;
+			vecp->i_type = XLOG_REG_TYPE_BCHUNK;
+/* You would think we need to bump the nvecs here too, but we do not
+ * this number is used by recovery, and it gets confused by the boundary
+ * split here
+ *			nvecs++;
+ */
+			vecp++;
+			first_bit = next_bit;
+			last_bit = next_bit;
+			nbits = 1;
+		} else {
+			last_bit++;
+			nbits++;
+		}
+	}
+	bip->bli_format.blf_size = nvecs;
+
+	/*
+	 * Check to make sure everything is consistent.
+	 */
+	trace_xfs_buf_item_format(bip);
+	xfs_buf_item_log_check(bip);
+}
+
+/*
+ * This is called to pin the buffer associated with the buf log item in memory
+ * so it cannot be written out.
+ *
+ * We also always take a reference to the buffer log item here so that the bli
+ * is held while the item is pinned in memory. This means that we can
+ * unconditionally drop the reference count a transaction holds when the
+ * transaction is completed.
+ */
+STATIC void
+xfs_buf_item_pin(
+	struct xfs_log_item	*lip)
+{
+	struct xfs_buf_log_item	*bip = BUF_ITEM(lip);
+
+	ASSERT(XFS_BUF_ISBUSY(bip->bli_buf));
+	ASSERT(atomic_read(&bip->bli_refcount) > 0);
+	ASSERT((bip->bli_flags & XFS_BLI_LOGGED) ||
+	       (bip->bli_flags & XFS_BLI_STALE));
+
+	trace_xfs_buf_item_pin(bip);
+
+	atomic_inc(&bip->bli_refcount);
+	atomic_inc(&bip->bli_buf->b_pin_count);
+}
+
+/*
+ * This is called to unpin the buffer associated with the buf log
+ * item which was previously pinned with a call to xfs_buf_item_pin().
+ *
+ * Also drop the reference to the buf item for the current transaction.
+ * If the XFS_BLI_STALE flag is set and we are the last reference,
+ * then free up the buf log item and unlock the buffer.
+ *
+ * If the remove flag is set we are called from uncommit in the
+ * forced-shutdown path.  If that is true and the reference count on
+ * the log item is going to drop to zero we need to free the item's
+ * descriptor in the transaction.
+ */
+STATIC void
+xfs_buf_item_unpin(
+	struct xfs_log_item	*lip,
+	int			remove)
+{
+	struct xfs_buf_log_item	*bip = BUF_ITEM(lip);
+	xfs_buf_t	*bp = bip->bli_buf;
+	struct xfs_ail	*ailp = lip->li_ailp;
+	int		stale = bip->bli_flags & XFS_BLI_STALE;
+	int		freed;
+
+	ASSERT(XFS_BUF_FSPRIVATE(bp, xfs_buf_log_item_t *) == bip);
+	ASSERT(atomic_read(&bip->bli_refcount) > 0);
+
+	trace_xfs_buf_item_unpin(bip);
+
+	freed = atomic_dec_and_test(&bip->bli_refcount);
+
+	if (atomic_dec_and_test(&bp->b_pin_count))
+		wake_up_all(&bp->b_waiters);
+
+	if (freed && stale) {
+		ASSERT(bip->bli_flags & XFS_BLI_STALE);
+		ASSERT(XFS_BUF_VALUSEMA(bp) <= 0);
+		ASSERT(!(XFS_BUF_ISDELAYWRITE(bp)));
+		ASSERT(XFS_BUF_ISSTALE(bp));
+		ASSERT(bip->bli_format.blf_flags & XFS_BLF_CANCEL);
+
+		trace_xfs_buf_item_unpin_stale(bip);
+
+		if (remove) {
+			/*
+			 * If we are in a transaction context, we have to
+			 * remove the log item from the transaction as we are
+			 * about to release our reference to the buffer.  If we
+			 * don't, the unlock that occurs later in
+			 * xfs_trans_uncommit() will try to reference the
+			 * buffer which we no longer have a hold on.
+			 */
+			if (lip->li_desc)
+				xfs_trans_del_item(lip);
+
+			/*
+			 * Since the transaction no longer refers to the buffer,
+			 * the buffer should no longer refer to the transaction.
+			 */
+			XFS_BUF_SET_FSPRIVATE2(bp, NULL);
+		}
+
+		/*
+		 * If we get called here because of an IO error, we may
+		 * or may not have the item on the AIL. xfs_trans_ail_delete()
+		 * will take care of that situation.
+		 * xfs_trans_ail_delete() drops the AIL lock.
+		 */
+		if (bip->bli_flags & XFS_BLI_STALE_INODE) {
+			xfs_buf_do_callbacks(bp);
+			XFS_BUF_SET_FSPRIVATE(bp, NULL);
+			XFS_BUF_CLR_IODONE_FUNC(bp);
+		} else {
+			spin_lock(&ailp->xa_lock);
+			xfs_trans_ail_delete(ailp, (xfs_log_item_t *)bip);
+			xfs_buf_item_relse(bp);
+			ASSERT(XFS_BUF_FSPRIVATE(bp, void *) == NULL);
+		}
+		xfs_buf_relse(bp);
+	}
+}
+
+/*
+ * This is called to attempt to lock the buffer associated with this
+ * buf log item.  Don't sleep on the buffer lock.  If we can't get
+ * the lock right away, return 0.  If we can get the lock, take a
+ * reference to the buffer. If this is a delayed write buffer that
+ * needs AIL help to be written back, invoke the pushbuf routine
+ * rather than the normal success path.
+ */
+STATIC uint
+xfs_buf_item_trylock(
+	struct xfs_log_item	*lip)
+{
+	struct xfs_buf_log_item	*bip = BUF_ITEM(lip);
+	struct xfs_buf		*bp = bip->bli_buf;
+
+	if (XFS_BUF_ISPINNED(bp))
+		return XFS_ITEM_PINNED;
+	if (!XFS_BUF_CPSEMA(bp))
+		return XFS_ITEM_LOCKED;
+
+	/* take a reference to the buffer.  */
+	XFS_BUF_HOLD(bp);
+
+	ASSERT(!(bip->bli_flags & XFS_BLI_STALE));
+	trace_xfs_buf_item_trylock(bip);
+	if (XFS_BUF_ISDELAYWRITE(bp))
+		return XFS_ITEM_PUSHBUF;
+	return XFS_ITEM_SUCCESS;
+}
+
+/*
+ * Release the buffer associated with the buf log item.  If there is no dirty
+ * logged data associated with the buffer recorded in the buf log item, then
+ * free the buf log item and remove the reference to it in the buffer.
+ *
+ * This call ignores the recursion count.  It is only called when the buffer
+ * should REALLY be unlocked, regardless of the recursion count.
+ *
+ * We unconditionally drop the transaction's reference to the log item. If the
+ * item was logged, then another reference was taken when it was pinned, so we
+ * can safely drop the transaction reference now.  This also allows us to avoid
+ * potential races with the unpin code freeing the bli by not referencing the
+ * bli after we've dropped the reference count.
+ *
+ * If the XFS_BLI_HOLD flag is set in the buf log item, then free the log item
+ * if necessary but do not unlock the buffer.  This is for support of
+ * xfs_trans_bhold(). Make sure the XFS_BLI_HOLD field is cleared if we don't
+ * free the item.
+ */
+STATIC void
+xfs_buf_item_unlock(
+	struct xfs_log_item	*lip)
+{
+	struct xfs_buf_log_item	*bip = BUF_ITEM(lip);
+	struct xfs_buf		*bp = bip->bli_buf;
+	int			aborted;
+	uint			hold;
+
+	/* Clear the buffer's association with this transaction. */
+	XFS_BUF_SET_FSPRIVATE2(bp, NULL);
+
+	/*
+	 * If this is a transaction abort, don't return early.  Instead, allow
+	 * the brelse to happen.  Normally it would be done for stale
+	 * (cancelled) buffers at unpin time, but we'll never go through the
+	 * pin/unpin cycle if we abort inside commit.
+	 */
+	aborted = (lip->li_flags & XFS_LI_ABORTED) != 0;
+
+	/*
+	 * Before possibly freeing the buf item, determine if we should
+	 * release the buffer at the end of this routine.
+	 */
+	hold = bip->bli_flags & XFS_BLI_HOLD;
+
+	/* Clear the per transaction state. */
+	bip->bli_flags &= ~(XFS_BLI_LOGGED | XFS_BLI_HOLD);
+
+	/*
+	 * If the buf item is marked stale, then don't do anything.  We'll
+	 * unlock the buffer and free the buf item when the buffer is unpinned
+	 * for the last time.
+	 */
+	if (bip->bli_flags & XFS_BLI_STALE) {
+		trace_xfs_buf_item_unlock_stale(bip);
+		ASSERT(bip->bli_format.blf_flags & XFS_BLF_CANCEL);
+		if (!aborted) {
+			atomic_dec(&bip->bli_refcount);
+			return;
+		}
+	}
+
+	trace_xfs_buf_item_unlock(bip);
+
+	/*
+	 * If the buf item isn't tracking any data, free it, otherwise drop the
+	 * reference we hold to it.
+	 */
+	if (xfs_bitmap_empty(bip->bli_format.blf_data_map,
+			     bip->bli_format.blf_map_size))
+		xfs_buf_item_relse(bp);
+	else
+		atomic_dec(&bip->bli_refcount);
+
+	if (!hold)
+		xfs_buf_relse(bp);
+}
+
+/*
+ * This is called to find out where the oldest active copy of the
+ * buf log item in the on disk log resides now that the last log
+ * write of it completed at the given lsn.
+ * We always re-log all the dirty data in a buffer, so usually the
+ * latest copy in the on disk log is the only one that matters.  For
+ * those cases we simply return the given lsn.
+ *
+ * The one exception to this is for buffers full of newly allocated
+ * inodes.  These buffers are only relogged with the XFS_BLI_INODE_BUF
+ * flag set, indicating that only the di_next_unlinked fields from the
+ * inodes in the buffers will be replayed during recovery.  If the
+ * original newly allocated inode images have not yet been flushed
+ * when the buffer is so relogged, then we need to make sure that we
+ * keep the old images in the 'active' portion of the log.  We do this
+ * by returning the original lsn of that transaction here rather than
+ * the current one.
+ */
+STATIC xfs_lsn_t
+xfs_buf_item_committed(
+	struct xfs_log_item	*lip,
+	xfs_lsn_t		lsn)
+{
+	struct xfs_buf_log_item	*bip = BUF_ITEM(lip);
+
+	trace_xfs_buf_item_committed(bip);
+
+	if ((bip->bli_flags & XFS_BLI_INODE_ALLOC_BUF) && lip->li_lsn != 0)
+		return lip->li_lsn;
+	return lsn;
+}
+
+/*
+ * The buffer is locked, but is not a delayed write buffer. This happens
+ * if we race with IO completion and hence we don't want to try to write it
+ * again. Just release the buffer.
+ */
+STATIC void
+xfs_buf_item_push(
+	struct xfs_log_item	*lip)
+{
+	struct xfs_buf_log_item	*bip = BUF_ITEM(lip);
+	struct xfs_buf		*bp = bip->bli_buf;
+
+	ASSERT(!(bip->bli_flags & XFS_BLI_STALE));
+	ASSERT(!XFS_BUF_ISDELAYWRITE(bp));
+
+	trace_xfs_buf_item_push(bip);
+
+	xfs_buf_relse(bp);
+}
+
+/*
+ * The buffer is locked and is a delayed write buffer. Promote the buffer
+ * in the delayed write queue as the caller knows that they must invoke
+ * the xfsbufd to get this buffer written. We have to unlock the buffer
+ * to allow the xfsbufd to write it, too.
+ */
+STATIC bool
+xfs_buf_item_pushbuf(
+	struct xfs_log_item	*lip)
+{
+	struct xfs_buf_log_item	*bip = BUF_ITEM(lip);
+	struct xfs_buf		*bp = bip->bli_buf;
+
+	ASSERT(!(bip->bli_flags & XFS_BLI_STALE));
+	ASSERT(XFS_BUF_ISDELAYWRITE(bp));
+
+	trace_xfs_buf_item_pushbuf(bip);
+
+	xfs_buf_delwri_promote(bp);
+	xfs_buf_relse(bp);
+	return true;
+}
+
+STATIC void
+xfs_buf_item_committing(
+	struct xfs_log_item	*lip,
+	xfs_lsn_t		commit_lsn)
+{
+}
+
+/*
+ * This is the ops vector shared by all buf log items.
+ */
+static struct xfs_item_ops xfs_buf_item_ops = {
+	.iop_size	= xfs_buf_item_size,
+	.iop_format	= xfs_buf_item_format,
+	.iop_pin	= xfs_buf_item_pin,
+	.iop_unpin	= xfs_buf_item_unpin,
+	.iop_trylock	= xfs_buf_item_trylock,
+	.iop_unlock	= xfs_buf_item_unlock,
+	.iop_committed	= xfs_buf_item_committed,
+	.iop_push	= xfs_buf_item_push,
+	.iop_pushbuf	= xfs_buf_item_pushbuf,
+	.iop_committing = xfs_buf_item_committing
+};
+
+
+/*
+ * Allocate a new buf log item to go with the given buffer.
+ * Set the buffer's b_fsprivate field to point to the new
+ * buf log item.  If there are other item's attached to the
+ * buffer (see xfs_buf_attach_iodone() below), then put the
+ * buf log item at the front.
+ */
+void
+xfs_buf_item_init(
+	xfs_buf_t	*bp,
+	xfs_mount_t	*mp)
+{
+	xfs_log_item_t		*lip;
+	xfs_buf_log_item_t	*bip;
+	int			chunks;
+	int			map_size;
+
+	/*
+	 * Check to see if there is already a buf log item for
+	 * this buffer.  If there is, it is guaranteed to be
+	 * the first.  If we do already have one, there is
+	 * nothing to do here so return.
+	 */
+	ASSERT(bp->b_target->bt_mount == mp);
+	if (XFS_BUF_FSPRIVATE(bp, void *) != NULL) {
+		lip = XFS_BUF_FSPRIVATE(bp, xfs_log_item_t *);
+		if (lip->li_type == XFS_LI_BUF) {
+			return;
+		}
+	}
+
+	/*
+	 * chunks is the number of XFS_BLF_CHUNK size pieces
+	 * the buffer can be divided into. Make sure not to
+	 * truncate any pieces.  map_size is the size of the
+	 * bitmap needed to describe the chunks of the buffer.
+	 */
+	chunks = (int)((XFS_BUF_COUNT(bp) + (XFS_BLF_CHUNK - 1)) >> XFS_BLF_SHIFT);
+	map_size = (int)((chunks + NBWORD) >> BIT_TO_WORD_SHIFT);
+
+	bip = (xfs_buf_log_item_t*)kmem_zone_zalloc(xfs_buf_item_zone,
+						    KM_SLEEP);
+	xfs_log_item_init(mp, &bip->bli_item, XFS_LI_BUF, &xfs_buf_item_ops);
+	bip->bli_buf = bp;
+	xfs_buf_hold(bp);
+	bip->bli_format.blf_type = XFS_LI_BUF;
+	bip->bli_format.blf_blkno = (__int64_t)XFS_BUF_ADDR(bp);
+	bip->bli_format.blf_len = (ushort)BTOBB(XFS_BUF_COUNT(bp));
+	bip->bli_format.blf_map_size = map_size;
+
+#ifdef XFS_TRANS_DEBUG
+	/*
+	 * Allocate the arrays for tracking what needs to be logged
+	 * and what our callers request to be logged.  bli_orig
+	 * holds a copy of the original, clean buffer for comparison
+	 * against, and bli_logged keeps a 1 bit flag per byte in
+	 * the buffer to indicate which bytes the callers have asked
+	 * to have logged.
+	 */
+	bip->bli_orig = (char *)kmem_alloc(XFS_BUF_COUNT(bp), KM_SLEEP);
+	memcpy(bip->bli_orig, XFS_BUF_PTR(bp), XFS_BUF_COUNT(bp));
+	bip->bli_logged = (char *)kmem_zalloc(XFS_BUF_COUNT(bp) / NBBY, KM_SLEEP);
+#endif
+
+	/*
+	 * Put the buf item into the list of items attached to the
+	 * buffer at the front.
+	 */
+	if (XFS_BUF_FSPRIVATE(bp, void *) != NULL) {
+		bip->bli_item.li_bio_list =
+				XFS_BUF_FSPRIVATE(bp, xfs_log_item_t *);
+	}
+	XFS_BUF_SET_FSPRIVATE(bp, bip);
+}
+
+
+/*
+ * Mark bytes first through last inclusive as dirty in the buf
+ * item's bitmap.
+ */
+void
+xfs_buf_item_log(
+	xfs_buf_log_item_t	*bip,
+	uint			first,
+	uint			last)
+{
+	uint		first_bit;
+	uint		last_bit;
+	uint		bits_to_set;
+	uint		bits_set;
+	uint		word_num;
+	uint		*wordp;
+	uint		bit;
+	uint		end_bit;
+	uint		mask;
+
+	/*
+	 * Mark the item as having some dirty data for
+	 * quick reference in xfs_buf_item_dirty.
+	 */
+	bip->bli_flags |= XFS_BLI_DIRTY;
+
+	/*
+	 * Convert byte offsets to bit numbers.
+	 */
+	first_bit = first >> XFS_BLF_SHIFT;
+	last_bit = last >> XFS_BLF_SHIFT;
+
+	/*
+	 * Calculate the total number of bits to be set.
+	 */
+	bits_to_set = last_bit - first_bit + 1;
+
+	/*
+	 * Get a pointer to the first word in the bitmap
+	 * to set a bit in.
+	 */
+	word_num = first_bit >> BIT_TO_WORD_SHIFT;
+	wordp = &(bip->bli_format.blf_data_map[word_num]);
+
+	/*
+	 * Calculate the starting bit in the first word.
+	 */
+	bit = first_bit & (uint)(NBWORD - 1);
+
+	/*
+	 * First set any bits in the first word of our range.
+	 * If it starts at bit 0 of the word, it will be
+	 * set below rather than here.  That is what the variable
+	 * bit tells us. The variable bits_set tracks the number
+	 * of bits that have been set so far.  End_bit is the number
+	 * of the last bit to be set in this word plus one.
+	 */
+	if (bit) {
+		end_bit = MIN(bit + bits_to_set, (uint)NBWORD);
+		mask = ((1 << (end_bit - bit)) - 1) << bit;
+		*wordp |= mask;
+		wordp++;
+		bits_set = end_bit - bit;
+	} else {
+		bits_set = 0;
+	}
+
+	/*
+	 * Now set bits a whole word at a time that are between
+	 * first_bit and last_bit.
+	 */
+	while ((bits_to_set - bits_set) >= NBWORD) {
+		*wordp |= 0xffffffff;
+		bits_set += NBWORD;
+		wordp++;
+	}
+
+	/*
+	 * Finally, set any bits left to be set in one last partial word.
+	 */
+	end_bit = bits_to_set - bits_set;
+	if (end_bit) {
+		mask = (1 << end_bit) - 1;
+		*wordp |= mask;
+	}
+
+	xfs_buf_item_log_debug(bip, first, last);
+}
+
+
+/*
+ * Return 1 if the buffer has some data that has been logged (at any
+ * point, not just the current transaction) and 0 if not.
+ */
+uint
+xfs_buf_item_dirty(
+	xfs_buf_log_item_t	*bip)
+{
+	return (bip->bli_flags & XFS_BLI_DIRTY);
+}
+
+STATIC void
+xfs_buf_item_free(
+	xfs_buf_log_item_t	*bip)
+{
+#ifdef XFS_TRANS_DEBUG
+	kmem_free(bip->bli_orig);
+	kmem_free(bip->bli_logged);
+#endif /* XFS_TRANS_DEBUG */
+
+	kmem_zone_free(xfs_buf_item_zone, bip);
+}
+
+/*
+ * This is called when the buf log item is no longer needed.  It should
+ * free the buf log item associated with the given buffer and clear
+ * the buffer's pointer to the buf log item.  If there are no more
+ * items in the list, clear the b_iodone field of the buffer (see
+ * xfs_buf_attach_iodone() below).
+ */
+void
+xfs_buf_item_relse(
+	xfs_buf_t	*bp)
+{
+	xfs_buf_log_item_t	*bip;
+
+	trace_xfs_buf_item_relse(bp, _RET_IP_);
+
+	bip = XFS_BUF_FSPRIVATE(bp, xfs_buf_log_item_t*);
+	XFS_BUF_SET_FSPRIVATE(bp, bip->bli_item.li_bio_list);
+	if ((XFS_BUF_FSPRIVATE(bp, void *) == NULL) &&
+	    (XFS_BUF_IODONE_FUNC(bp) != NULL)) {
+		XFS_BUF_CLR_IODONE_FUNC(bp);
+	}
+	xfs_buf_rele(bp);
+	xfs_buf_item_free(bip);
+}
+
+
+/*
+ * Add the given log item with its callback to the list of callbacks
+ * to be called when the buffer's I/O completes.  If it is not set
+ * already, set the buffer's b_iodone() routine to be
+ * xfs_buf_iodone_callbacks() and link the log item into the list of
+ * items rooted at b_fsprivate.  Items are always added as the second
+ * entry in the list if there is a first, because the buf item code
+ * assumes that the buf log item is first.
+ */
+void
+xfs_buf_attach_iodone(
+	xfs_buf_t	*bp,
+	void		(*cb)(xfs_buf_t *, xfs_log_item_t *),
+	xfs_log_item_t	*lip)
+{
+	xfs_log_item_t	*head_lip;
+
+	ASSERT(XFS_BUF_ISBUSY(bp));
+	ASSERT(XFS_BUF_VALUSEMA(bp) <= 0);
+
+	lip->li_cb = cb;
+	if (XFS_BUF_FSPRIVATE(bp, void *) != NULL) {
+		head_lip = XFS_BUF_FSPRIVATE(bp, xfs_log_item_t *);
+		lip->li_bio_list = head_lip->li_bio_list;
+		head_lip->li_bio_list = lip;
+	} else {
+		XFS_BUF_SET_FSPRIVATE(bp, lip);
+	}
+
+	ASSERT((XFS_BUF_IODONE_FUNC(bp) == xfs_buf_iodone_callbacks) ||
+	       (XFS_BUF_IODONE_FUNC(bp) == NULL));
+	XFS_BUF_SET_IODONE_FUNC(bp, xfs_buf_iodone_callbacks);
+}
+
+/*
+ * We can have many callbacks on a buffer. Running the callbacks individually
+ * can cause a lot of contention on the AIL lock, so we allow for a single
+ * callback to be able to scan the remaining lip->li_bio_list for other items
+ * of the same type and callback to be processed in the first call.
+ *
+ * As a result, the loop walking the callback list below will also modify the
+ * list. it removes the first item from the list and then runs the callback.
+ * The loop then restarts from the new head of the list. This allows the
+ * callback to scan and modify the list attached to the buffer and we don't
+ * have to care about maintaining a next item pointer.
+ */
+STATIC void
+xfs_buf_do_callbacks(
+	struct xfs_buf		*bp)
+{
+	struct xfs_log_item	*lip;
+
+	while ((lip = XFS_BUF_FSPRIVATE(bp, xfs_log_item_t *)) != NULL) {
+		XFS_BUF_SET_FSPRIVATE(bp, lip->li_bio_list);
+		ASSERT(lip->li_cb != NULL);
+		/*
+		 * Clear the next pointer so we don't have any
+		 * confusion if the item is added to another buf.
+		 * Don't touch the log item after calling its
+		 * callback, because it could have freed itself.
+		 */
+		lip->li_bio_list = NULL;
+		lip->li_cb(bp, lip);
+	}
+}
+
+/*
+ * This is the iodone() function for buffers which have had callbacks
+ * attached to them by xfs_buf_attach_iodone().  It should remove each
+ * log item from the buffer's list and call the callback of each in turn.
+ * When done, the buffer's fsprivate field is set to NULL and the buffer
+ * is unlocked with a call to iodone().
+ */
+void
+xfs_buf_iodone_callbacks(
+	struct xfs_buf		*bp)
+{
+	struct xfs_log_item	*lip = bp->b_fspriv;
+	struct xfs_mount	*mp = lip->li_mountp;
+	static ulong		lasttime;
+	static xfs_buftarg_t	*lasttarg;
+
+	if (likely(!XFS_BUF_GETERROR(bp)))
+		goto do_callbacks;
+
+	/*
+	 * If we've already decided to shutdown the filesystem because of
+	 * I/O errors, there's no point in giving this a retry.
+	 */
+	if (XFS_FORCED_SHUTDOWN(mp)) {
+		XFS_BUF_SUPER_STALE(bp);
+		trace_xfs_buf_item_iodone(bp, _RET_IP_);
+		goto do_callbacks;
+	}
+
+	if (XFS_BUF_TARGET(bp) != lasttarg ||
+	    time_after(jiffies, (lasttime + 5*HZ))) {
+		lasttime = jiffies;
+		xfs_alert(mp, "Device %s: metadata write error block 0x%llx",
+			XFS_BUFTARG_NAME(XFS_BUF_TARGET(bp)),
+		      (__uint64_t)XFS_BUF_ADDR(bp));
+	}
+	lasttarg = XFS_BUF_TARGET(bp);
+
+	/*
+	 * If the write was asynchronous then no one will be looking for the
+	 * error.  Clear the error state and write the buffer out again.
+	 *
+	 * During sync or umount we'll write all pending buffers again
+	 * synchronous, which will catch these errors if they keep hanging
+	 * around.
+	 */
+	if (XFS_BUF_ISASYNC(bp)) {
+		XFS_BUF_ERROR(bp, 0); /* errno of 0 unsets the flag */
+
+		if (!XFS_BUF_ISSTALE(bp)) {
+			XFS_BUF_DELAYWRITE(bp);
+			XFS_BUF_DONE(bp);
+			XFS_BUF_SET_START(bp);
+		}
+		ASSERT(XFS_BUF_IODONE_FUNC(bp));
+		trace_xfs_buf_item_iodone_async(bp, _RET_IP_);
+		xfs_buf_relse(bp);
+		return;
+	}
+
+	/*
+	 * If the write of the buffer was synchronous, we want to make
+	 * sure to return the error to the caller of xfs_bwrite().
+	 */
+	XFS_BUF_STALE(bp);
+	XFS_BUF_DONE(bp);
+	XFS_BUF_UNDELAYWRITE(bp);
+
+	trace_xfs_buf_error_relse(bp, _RET_IP_);
+
+do_callbacks:
+	xfs_buf_do_callbacks(bp);
+	XFS_BUF_SET_FSPRIVATE(bp, NULL);
+	XFS_BUF_CLR_IODONE_FUNC(bp);
+	xfs_buf_ioend(bp, 0);
+}
+
+/*
+ * This is the iodone() function for buffers which have been
+ * logged.  It is called when they are eventually flushed out.
+ * It should remove the buf item from the AIL, and free the buf item.
+ * It is called by xfs_buf_iodone_callbacks() above which will take
+ * care of cleaning up the buffer itself.
+ */
+void
+xfs_buf_iodone(
+	struct xfs_buf		*bp,
+	struct xfs_log_item	*lip)
+{
+	struct xfs_ail		*ailp = lip->li_ailp;
+
+	ASSERT(BUF_ITEM(lip)->bli_buf == bp);
+
+	xfs_buf_rele(bp);
+
+	/*
+	 * If we are forcibly shutting down, this may well be
+	 * off the AIL already. That's because we simulate the
+	 * log-committed callbacks to unpin these buffers. Or we may never
+	 * have put this item on AIL because of the transaction was
+	 * aborted forcibly. xfs_trans_ail_delete() takes care of these.
+	 *
+	 * Either way, AIL is useless if we're forcing a shutdown.
+	 */
+	spin_lock(&ailp->xa_lock);
+	xfs_trans_ail_delete(ailp, lip);
+	xfs_buf_item_free(BUF_ITEM(lip));
+}
diff --git a/fs/xfs/xfs_buf_item.h b/fs/xfs/xfs_buf_item.h
new file mode 100644
index 00000000..b6ecd206
--- /dev/null
+++ b/fs/xfs/xfs_buf_item.h
@@ -0,0 +1,130 @@
+/*
+ * Copyright (c) 2000-2001,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef	__XFS_BUF_ITEM_H__
+#define	__XFS_BUF_ITEM_H__
+
+extern kmem_zone_t	*xfs_buf_item_zone;
+
+/*
+ * This is the structure used to lay out a buf log item in the
+ * log.  The data map describes which 128 byte chunks of the buffer
+ * have been logged.
+ * For 6.2 and beyond, this is XFS_LI_BUF.  We use this to log everything.
+ */
+typedef struct xfs_buf_log_format {
+	unsigned short	blf_type;	/* buf log item type indicator */
+	unsigned short	blf_size;	/* size of this item */
+	ushort		blf_flags;	/* misc state */
+	ushort		blf_len;	/* number of blocks in this buf */
+	__int64_t	blf_blkno;	/* starting blkno of this buf */
+	unsigned int	blf_map_size;	/* size of data bitmap in words */
+	unsigned int	blf_data_map[1];/* variable size bitmap of */
+					/*   regions of buffer in this item */
+} xfs_buf_log_format_t;
+
+/*
+ * This flag indicates that the buffer contains on disk inodes
+ * and requires special recovery handling.
+ */
+#define	XFS_BLF_INODE_BUF	0x1
+/*
+ * This flag indicates that the buffer should not be replayed
+ * during recovery because its blocks are being freed.
+ */
+#define	XFS_BLF_CANCEL		0x2
+/*
+ * This flag indicates that the buffer contains on disk
+ * user or group dquots and may require special recovery handling.
+ */
+#define	XFS_BLF_UDQUOT_BUF	0x4
+#define XFS_BLF_PDQUOT_BUF	0x8
+#define	XFS_BLF_GDQUOT_BUF	0x10
+
+#define	XFS_BLF_CHUNK		128
+#define	XFS_BLF_SHIFT		7
+#define	BIT_TO_WORD_SHIFT	5
+#define	NBWORD			(NBBY * sizeof(unsigned int))
+
+/*
+ * buf log item flags
+ */
+#define	XFS_BLI_HOLD		0x01
+#define	XFS_BLI_DIRTY		0x02
+#define	XFS_BLI_STALE		0x04
+#define	XFS_BLI_LOGGED		0x08
+#define	XFS_BLI_INODE_ALLOC_BUF	0x10
+#define XFS_BLI_STALE_INODE	0x20
+#define	XFS_BLI_INODE_BUF	0x40
+
+#define XFS_BLI_FLAGS \
+	{ XFS_BLI_HOLD,		"HOLD" }, \
+	{ XFS_BLI_DIRTY,	"DIRTY" }, \
+	{ XFS_BLI_STALE,	"STALE" }, \
+	{ XFS_BLI_LOGGED,	"LOGGED" }, \
+	{ XFS_BLI_INODE_ALLOC_BUF, "INODE_ALLOC" }, \
+	{ XFS_BLI_STALE_INODE,	"STALE_INODE" }, \
+	{ XFS_BLI_INODE_BUF,	"INODE_BUF" }
+
+
+#ifdef __KERNEL__
+
+struct xfs_buf;
+struct xfs_mount;
+struct xfs_buf_log_item;
+
+/*
+ * This is the in core log item structure used to track information
+ * needed to log buffers.  It tracks how many times the lock has been
+ * locked, and which 128 byte chunks of the buffer are dirty.
+ */
+typedef struct xfs_buf_log_item {
+	xfs_log_item_t		bli_item;	/* common item structure */
+	struct xfs_buf		*bli_buf;	/* real buffer pointer */
+	unsigned int		bli_flags;	/* misc flags */
+	unsigned int		bli_recur;	/* lock recursion count */
+	atomic_t		bli_refcount;	/* cnt of tp refs */
+#ifdef XFS_TRANS_DEBUG
+	char			*bli_orig;	/* original buffer copy */
+	char			*bli_logged;	/* bytes logged (bitmap) */
+#endif
+	xfs_buf_log_format_t	bli_format;	/* in-log header */
+} xfs_buf_log_item_t;
+
+void	xfs_buf_item_init(struct xfs_buf *, struct xfs_mount *);
+void	xfs_buf_item_relse(struct xfs_buf *);
+void	xfs_buf_item_log(xfs_buf_log_item_t *, uint, uint);
+uint	xfs_buf_item_dirty(xfs_buf_log_item_t *);
+void	xfs_buf_attach_iodone(struct xfs_buf *,
+			      void(*)(struct xfs_buf *, xfs_log_item_t *),
+			      xfs_log_item_t *);
+void	xfs_buf_iodone_callbacks(struct xfs_buf *);
+void	xfs_buf_iodone(struct xfs_buf *, struct xfs_log_item *);
+
+#ifdef XFS_TRANS_DEBUG
+void
+xfs_buf_item_flush_log_debug(
+	struct xfs_buf *bp,
+	uint	first,
+	uint	last);
+#else
+#define	xfs_buf_item_flush_log_debug(bp, first, last)
+#endif
+
+#endif	/* __KERNEL__ */
+
+#endif	/* __XFS_BUF_ITEM_H__ */
diff --git a/fs/xfs/xfs_da_btree.c b/fs/xfs/xfs_da_btree.c
new file mode 100644
index 00000000..6102ac6d
--- /dev/null
+++ b/fs/xfs/xfs_da_btree.c
@@ -0,0 +1,2463 @@
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_dir2.h"
+#include "xfs_mount.h"
+#include "xfs_da_btree.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_dir2_sf.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_inode_item.h"
+#include "xfs_alloc.h"
+#include "xfs_bmap.h"
+#include "xfs_attr.h"
+#include "xfs_attr_leaf.h"
+#include "xfs_dir2_data.h"
+#include "xfs_dir2_leaf.h"
+#include "xfs_dir2_block.h"
+#include "xfs_dir2_node.h"
+#include "xfs_error.h"
+#include "xfs_trace.h"
+
+/*
+ * xfs_da_btree.c
+ *
+ * Routines to implement directories as Btrees of hashed names.
+ */
+
+/*========================================================================
+ * Function prototypes for the kernel.
+ *========================================================================*/
+
+/*
+ * Routines used for growing the Btree.
+ */
+STATIC int xfs_da_root_split(xfs_da_state_t *state,
+					    xfs_da_state_blk_t *existing_root,
+					    xfs_da_state_blk_t *new_child);
+STATIC int xfs_da_node_split(xfs_da_state_t *state,
+					    xfs_da_state_blk_t *existing_blk,
+					    xfs_da_state_blk_t *split_blk,
+					    xfs_da_state_blk_t *blk_to_add,
+					    int treelevel,
+					    int *result);
+STATIC void xfs_da_node_rebalance(xfs_da_state_t *state,
+					 xfs_da_state_blk_t *node_blk_1,
+					 xfs_da_state_blk_t *node_blk_2);
+STATIC void xfs_da_node_add(xfs_da_state_t *state,
+				   xfs_da_state_blk_t *old_node_blk,
+				   xfs_da_state_blk_t *new_node_blk);
+
+/*
+ * Routines used for shrinking the Btree.
+ */
+STATIC int xfs_da_root_join(xfs_da_state_t *state,
+					   xfs_da_state_blk_t *root_blk);
+STATIC int xfs_da_node_toosmall(xfs_da_state_t *state, int *retval);
+STATIC void xfs_da_node_remove(xfs_da_state_t *state,
+					      xfs_da_state_blk_t *drop_blk);
+STATIC void xfs_da_node_unbalance(xfs_da_state_t *state,
+					 xfs_da_state_blk_t *src_node_blk,
+					 xfs_da_state_blk_t *dst_node_blk);
+
+/*
+ * Utility routines.
+ */
+STATIC uint	xfs_da_node_lasthash(xfs_dabuf_t *bp, int *count);
+STATIC int	xfs_da_node_order(xfs_dabuf_t *node1_bp, xfs_dabuf_t *node2_bp);
+STATIC xfs_dabuf_t *xfs_da_buf_make(int nbuf, xfs_buf_t **bps, inst_t *ra);
+STATIC int	xfs_da_blk_unlink(xfs_da_state_t *state,
+				  xfs_da_state_blk_t *drop_blk,
+				  xfs_da_state_blk_t *save_blk);
+STATIC void	xfs_da_state_kill_altpath(xfs_da_state_t *state);
+
+/*========================================================================
+ * Routines used for growing the Btree.
+ *========================================================================*/
+
+/*
+ * Create the initial contents of an intermediate node.
+ */
+int
+xfs_da_node_create(xfs_da_args_t *args, xfs_dablk_t blkno, int level,
+				 xfs_dabuf_t **bpp, int whichfork)
+{
+	xfs_da_intnode_t *node;
+	xfs_dabuf_t *bp;
+	int error;
+	xfs_trans_t *tp;
+
+	tp = args->trans;
+	error = xfs_da_get_buf(tp, args->dp, blkno, -1, &bp, whichfork);
+	if (error)
+		return(error);
+	ASSERT(bp != NULL);
+	node = bp->data;
+	node->hdr.info.forw = 0;
+	node->hdr.info.back = 0;
+	node->hdr.info.magic = cpu_to_be16(XFS_DA_NODE_MAGIC);
+	node->hdr.info.pad = 0;
+	node->hdr.count = 0;
+	node->hdr.level = cpu_to_be16(level);
+
+	xfs_da_log_buf(tp, bp,
+		XFS_DA_LOGRANGE(node, &node->hdr, sizeof(node->hdr)));
+
+	*bpp = bp;
+	return(0);
+}
+
+/*
+ * Split a leaf node, rebalance, then possibly split
+ * intermediate nodes, rebalance, etc.
+ */
+int							/* error */
+xfs_da_split(xfs_da_state_t *state)
+{
+	xfs_da_state_blk_t *oldblk, *newblk, *addblk;
+	xfs_da_intnode_t *node;
+	xfs_dabuf_t *bp;
+	int max, action, error, i;
+
+	/*
+	 * Walk back up the tree splitting/inserting/adjusting as necessary.
+	 * If we need to insert and there isn't room, split the node, then
+	 * decide which fragment to insert the new block from below into.
+	 * Note that we may split the root this way, but we need more fixup.
+	 */
+	max = state->path.active - 1;
+	ASSERT((max >= 0) && (max < XFS_DA_NODE_MAXDEPTH));
+	ASSERT(state->path.blk[max].magic == XFS_ATTR_LEAF_MAGIC ||
+	       state->path.blk[max].magic == XFS_DIR2_LEAFN_MAGIC);
+
+	addblk = &state->path.blk[max];		/* initial dummy value */
+	for (i = max; (i >= 0) && addblk; state->path.active--, i--) {
+		oldblk = &state->path.blk[i];
+		newblk = &state->altpath.blk[i];
+
+		/*
+		 * If a leaf node then
+		 *     Allocate a new leaf node, then rebalance across them.
+		 * else if an intermediate node then
+		 *     We split on the last layer, must we split the node?
+		 */
+		switch (oldblk->magic) {
+		case XFS_ATTR_LEAF_MAGIC:
+			error = xfs_attr_leaf_split(state, oldblk, newblk);
+			if ((error != 0) && (error != ENOSPC)) {
+				return(error);	/* GROT: attr is inconsistent */
+			}
+			if (!error) {
+				addblk = newblk;
+				break;
+			}
+			/*
+			 * Entry wouldn't fit, split the leaf again.
+			 */
+			state->extravalid = 1;
+			if (state->inleaf) {
+				state->extraafter = 0;	/* before newblk */
+				error = xfs_attr_leaf_split(state, oldblk,
+							    &state->extrablk);
+			} else {
+				state->extraafter = 1;	/* after newblk */
+				error = xfs_attr_leaf_split(state, newblk,
+							    &state->extrablk);
+			}
+			if (error)
+				return(error);	/* GROT: attr inconsistent */
+			addblk = newblk;
+			break;
+		case XFS_DIR2_LEAFN_MAGIC:
+			error = xfs_dir2_leafn_split(state, oldblk, newblk);
+			if (error)
+				return error;
+			addblk = newblk;
+			break;
+		case XFS_DA_NODE_MAGIC:
+			error = xfs_da_node_split(state, oldblk, newblk, addblk,
+							 max - i, &action);
+			xfs_da_buf_done(addblk->bp);
+			addblk->bp = NULL;
+			if (error)
+				return(error);	/* GROT: dir is inconsistent */
+			/*
+			 * Record the newly split block for the next time thru?
+			 */
+			if (action)
+				addblk = newblk;
+			else
+				addblk = NULL;
+			break;
+		}
+
+		/*
+		 * Update the btree to show the new hashval for this child.
+		 */
+		xfs_da_fixhashpath(state, &state->path);
+		/*
+		 * If we won't need this block again, it's getting dropped
+		 * from the active path by the loop control, so we need
+		 * to mark it done now.
+		 */
+		if (i > 0 || !addblk)
+			xfs_da_buf_done(oldblk->bp);
+	}
+	if (!addblk)
+		return(0);
+
+	/*
+	 * Split the root node.
+	 */
+	ASSERT(state->path.active == 0);
+	oldblk = &state->path.blk[0];
+	error = xfs_da_root_split(state, oldblk, addblk);
+	if (error) {
+		xfs_da_buf_done(oldblk->bp);
+		xfs_da_buf_done(addblk->bp);
+		addblk->bp = NULL;
+		return(error);	/* GROT: dir is inconsistent */
+	}
+
+	/*
+	 * Update pointers to the node which used to be block 0 and
+	 * just got bumped because of the addition of a new root node.
+	 * There might be three blocks involved if a double split occurred,
+	 * and the original block 0 could be at any position in the list.
+	 */
+
+	node = oldblk->bp->data;
+	if (node->hdr.info.forw) {
+		if (be32_to_cpu(node->hdr.info.forw) == addblk->blkno) {
+			bp = addblk->bp;
+		} else {
+			ASSERT(state->extravalid);
+			bp = state->extrablk.bp;
+		}
+		node = bp->data;
+		node->hdr.info.back = cpu_to_be32(oldblk->blkno);
+		xfs_da_log_buf(state->args->trans, bp,
+		    XFS_DA_LOGRANGE(node, &node->hdr.info,
+		    sizeof(node->hdr.info)));
+	}
+	node = oldblk->bp->data;
+	if (node->hdr.info.back) {
+		if (be32_to_cpu(node->hdr.info.back) == addblk->blkno) {
+			bp = addblk->bp;
+		} else {
+			ASSERT(state->extravalid);
+			bp = state->extrablk.bp;
+		}
+		node = bp->data;
+		node->hdr.info.forw = cpu_to_be32(oldblk->blkno);
+		xfs_da_log_buf(state->args->trans, bp,
+		    XFS_DA_LOGRANGE(node, &node->hdr.info,
+		    sizeof(node->hdr.info)));
+	}
+	xfs_da_buf_done(oldblk->bp);
+	xfs_da_buf_done(addblk->bp);
+	addblk->bp = NULL;
+	return(0);
+}
+
+/*
+ * Split the root.  We have to create a new root and point to the two
+ * parts (the split old root) that we just created.  Copy block zero to
+ * the EOF, extending the inode in process.
+ */
+STATIC int						/* error */
+xfs_da_root_split(xfs_da_state_t *state, xfs_da_state_blk_t *blk1,
+				 xfs_da_state_blk_t *blk2)
+{
+	xfs_da_intnode_t *node, *oldroot;
+	xfs_da_args_t *args;
+	xfs_dablk_t blkno;
+	xfs_dabuf_t *bp;
+	int error, size;
+	xfs_inode_t *dp;
+	xfs_trans_t *tp;
+	xfs_mount_t *mp;
+	xfs_dir2_leaf_t *leaf;
+
+	/*
+	 * Copy the existing (incorrect) block from the root node position
+	 * to a free space somewhere.
+	 */
+	args = state->args;
+	ASSERT(args != NULL);
+	error = xfs_da_grow_inode(args, &blkno);
+	if (error)
+		return(error);
+	dp = args->dp;
+	tp = args->trans;
+	mp = state->mp;
+	error = xfs_da_get_buf(tp, dp, blkno, -1, &bp, args->whichfork);
+	if (error)
+		return(error);
+	ASSERT(bp != NULL);
+	node = bp->data;
+	oldroot = blk1->bp->data;
+	if (be16_to_cpu(oldroot->hdr.info.magic) == XFS_DA_NODE_MAGIC) {
+		size = (int)((char *)&oldroot->btree[be16_to_cpu(oldroot->hdr.count)] -
+			     (char *)oldroot);
+	} else {
+		ASSERT(be16_to_cpu(oldroot->hdr.info.magic) == XFS_DIR2_LEAFN_MAGIC);
+		leaf = (xfs_dir2_leaf_t *)oldroot;
+		size = (int)((char *)&leaf->ents[be16_to_cpu(leaf->hdr.count)] -
+			     (char *)leaf);
+	}
+	memcpy(node, oldroot, size);
+	xfs_da_log_buf(tp, bp, 0, size - 1);
+	xfs_da_buf_done(blk1->bp);
+	blk1->bp = bp;
+	blk1->blkno = blkno;
+
+	/*
+	 * Set up the new root node.
+	 */
+	error = xfs_da_node_create(args,
+		(args->whichfork == XFS_DATA_FORK) ? mp->m_dirleafblk : 0,
+		be16_to_cpu(node->hdr.level) + 1, &bp, args->whichfork);
+	if (error)
+		return(error);
+	node = bp->data;
+	node->btree[0].hashval = cpu_to_be32(blk1->hashval);
+	node->btree[0].before = cpu_to_be32(blk1->blkno);
+	node->btree[1].hashval = cpu_to_be32(blk2->hashval);
+	node->btree[1].before = cpu_to_be32(blk2->blkno);
+	node->hdr.count = cpu_to_be16(2);
+
+#ifdef DEBUG
+	if (be16_to_cpu(oldroot->hdr.info.magic) == XFS_DIR2_LEAFN_MAGIC) {
+		ASSERT(blk1->blkno >= mp->m_dirleafblk &&
+		       blk1->blkno < mp->m_dirfreeblk);
+		ASSERT(blk2->blkno >= mp->m_dirleafblk &&
+		       blk2->blkno < mp->m_dirfreeblk);
+	}
+#endif
+
+	/* Header is already logged by xfs_da_node_create */
+	xfs_da_log_buf(tp, bp,
+		XFS_DA_LOGRANGE(node, node->btree,
+			sizeof(xfs_da_node_entry_t) * 2));
+	xfs_da_buf_done(bp);
+
+	return(0);
+}
+
+/*
+ * Split the node, rebalance, then add the new entry.
+ */
+STATIC int						/* error */
+xfs_da_node_split(xfs_da_state_t *state, xfs_da_state_blk_t *oldblk,
+				 xfs_da_state_blk_t *newblk,
+				 xfs_da_state_blk_t *addblk,
+				 int treelevel, int *result)
+{
+	xfs_da_intnode_t *node;
+	xfs_dablk_t blkno;
+	int newcount, error;
+	int useextra;
+
+	node = oldblk->bp->data;
+	ASSERT(be16_to_cpu(node->hdr.info.magic) == XFS_DA_NODE_MAGIC);
+
+	/*
+	 * With V2 dirs the extra block is data or freespace.
+	 */
+	useextra = state->extravalid && state->args->whichfork == XFS_ATTR_FORK;
+	newcount = 1 + useextra;
+	/*
+	 * Do we have to split the node?
+	 */
+	if ((be16_to_cpu(node->hdr.count) + newcount) > state->node_ents) {
+		/*
+		 * Allocate a new node, add to the doubly linked chain of
+		 * nodes, then move some of our excess entries into it.
+		 */
+		error = xfs_da_grow_inode(state->args, &blkno);
+		if (error)
+			return(error);	/* GROT: dir is inconsistent */
+
+		error = xfs_da_node_create(state->args, blkno, treelevel,
+					   &newblk->bp, state->args->whichfork);
+		if (error)
+			return(error);	/* GROT: dir is inconsistent */
+		newblk->blkno = blkno;
+		newblk->magic = XFS_DA_NODE_MAGIC;
+		xfs_da_node_rebalance(state, oldblk, newblk);
+		error = xfs_da_blk_link(state, oldblk, newblk);
+		if (error)
+			return(error);
+		*result = 1;
+	} else {
+		*result = 0;
+	}
+
+	/*
+	 * Insert the new entry(s) into the correct block
+	 * (updating last hashval in the process).
+	 *
+	 * xfs_da_node_add() inserts BEFORE the given index,
+	 * and as a result of using node_lookup_int() we always
+	 * point to a valid entry (not after one), but a split
+	 * operation always results in a new block whose hashvals
+	 * FOLLOW the current block.
+	 *
+	 * If we had double-split op below us, then add the extra block too.
+	 */
+	node = oldblk->bp->data;
+	if (oldblk->index <= be16_to_cpu(node->hdr.count)) {
+		oldblk->index++;
+		xfs_da_node_add(state, oldblk, addblk);
+		if (useextra) {
+			if (state->extraafter)
+				oldblk->index++;
+			xfs_da_node_add(state, oldblk, &state->extrablk);
+			state->extravalid = 0;
+		}
+	} else {
+		newblk->index++;
+		xfs_da_node_add(state, newblk, addblk);
+		if (useextra) {
+			if (state->extraafter)
+				newblk->index++;
+			xfs_da_node_add(state, newblk, &state->extrablk);
+			state->extravalid = 0;
+		}
+	}
+
+	return(0);
+}
+
+/*
+ * Balance the btree elements between two intermediate nodes,
+ * usually one full and one empty.
+ *
+ * NOTE: if blk2 is empty, then it will get the upper half of blk1.
+ */
+STATIC void
+xfs_da_node_rebalance(xfs_da_state_t *state, xfs_da_state_blk_t *blk1,
+				     xfs_da_state_blk_t *blk2)
+{
+	xfs_da_intnode_t *node1, *node2, *tmpnode;
+	xfs_da_node_entry_t *btree_s, *btree_d;
+	int count, tmp;
+	xfs_trans_t *tp;
+
+	node1 = blk1->bp->data;
+	node2 = blk2->bp->data;
+	/*
+	 * Figure out how many entries need to move, and in which direction.
+	 * Swap the nodes around if that makes it simpler.
+	 */
+	if ((be16_to_cpu(node1->hdr.count) > 0) && (be16_to_cpu(node2->hdr.count) > 0) &&
+	    ((be32_to_cpu(node2->btree[0].hashval) < be32_to_cpu(node1->btree[0].hashval)) ||
+	     (be32_to_cpu(node2->btree[be16_to_cpu(node2->hdr.count)-1].hashval) <
+	      be32_to_cpu(node1->btree[be16_to_cpu(node1->hdr.count)-1].hashval)))) {
+		tmpnode = node1;
+		node1 = node2;
+		node2 = tmpnode;
+	}
+	ASSERT(be16_to_cpu(node1->hdr.info.magic) == XFS_DA_NODE_MAGIC);
+	ASSERT(be16_to_cpu(node2->hdr.info.magic) == XFS_DA_NODE_MAGIC);
+	count = (be16_to_cpu(node1->hdr.count) - be16_to_cpu(node2->hdr.count)) / 2;
+	if (count == 0)
+		return;
+	tp = state->args->trans;
+	/*
+	 * Two cases: high-to-low and low-to-high.
+	 */
+	if (count > 0) {
+		/*
+		 * Move elements in node2 up to make a hole.
+		 */
+		if ((tmp = be16_to_cpu(node2->hdr.count)) > 0) {
+			tmp *= (uint)sizeof(xfs_da_node_entry_t);
+			btree_s = &node2->btree[0];
+			btree_d = &node2->btree[count];
+			memmove(btree_d, btree_s, tmp);
+		}
+
+		/*
+		 * Move the req'd B-tree elements from high in node1 to
+		 * low in node2.
+		 */
+		be16_add_cpu(&node2->hdr.count, count);
+		tmp = count * (uint)sizeof(xfs_da_node_entry_t);
+		btree_s = &node1->btree[be16_to_cpu(node1->hdr.count) - count];
+		btree_d = &node2->btree[0];
+		memcpy(btree_d, btree_s, tmp);
+		be16_add_cpu(&node1->hdr.count, -count);
+	} else {
+		/*
+		 * Move the req'd B-tree elements from low in node2 to
+		 * high in node1.
+		 */
+		count = -count;
+		tmp = count * (uint)sizeof(xfs_da_node_entry_t);
+		btree_s = &node2->btree[0];
+		btree_d = &node1->btree[be16_to_cpu(node1->hdr.count)];
+		memcpy(btree_d, btree_s, tmp);
+		be16_add_cpu(&node1->hdr.count, count);
+		xfs_da_log_buf(tp, blk1->bp,
+			XFS_DA_LOGRANGE(node1, btree_d, tmp));
+
+		/*
+		 * Move elements in node2 down to fill the hole.
+		 */
+		tmp  = be16_to_cpu(node2->hdr.count) - count;
+		tmp *= (uint)sizeof(xfs_da_node_entry_t);
+		btree_s = &node2->btree[count];
+		btree_d = &node2->btree[0];
+		memmove(btree_d, btree_s, tmp);
+		be16_add_cpu(&node2->hdr.count, -count);
+	}
+
+	/*
+	 * Log header of node 1 and all current bits of node 2.
+	 */
+	xfs_da_log_buf(tp, blk1->bp,
+		XFS_DA_LOGRANGE(node1, &node1->hdr, sizeof(node1->hdr)));
+	xfs_da_log_buf(tp, blk2->bp,
+		XFS_DA_LOGRANGE(node2, &node2->hdr,
+			sizeof(node2->hdr) +
+			sizeof(node2->btree[0]) * be16_to_cpu(node2->hdr.count)));
+
+	/*
+	 * Record the last hashval from each block for upward propagation.
+	 * (note: don't use the swapped node pointers)
+	 */
+	node1 = blk1->bp->data;
+	node2 = blk2->bp->data;
+	blk1->hashval = be32_to_cpu(node1->btree[be16_to_cpu(node1->hdr.count)-1].hashval);
+	blk2->hashval = be32_to_cpu(node2->btree[be16_to_cpu(node2->hdr.count)-1].hashval);
+
+	/*
+	 * Adjust the expected index for insertion.
+	 */
+	if (blk1->index >= be16_to_cpu(node1->hdr.count)) {
+		blk2->index = blk1->index - be16_to_cpu(node1->hdr.count);
+		blk1->index = be16_to_cpu(node1->hdr.count) + 1;	/* make it invalid */
+	}
+}
+
+/*
+ * Add a new entry to an intermediate node.
+ */
+STATIC void
+xfs_da_node_add(xfs_da_state_t *state, xfs_da_state_blk_t *oldblk,
+			       xfs_da_state_blk_t *newblk)
+{
+	xfs_da_intnode_t *node;
+	xfs_da_node_entry_t *btree;
+	int tmp;
+
+	node = oldblk->bp->data;
+	ASSERT(be16_to_cpu(node->hdr.info.magic) == XFS_DA_NODE_MAGIC);
+	ASSERT((oldblk->index >= 0) && (oldblk->index <= be16_to_cpu(node->hdr.count)));
+	ASSERT(newblk->blkno != 0);
+	if (state->args->whichfork == XFS_DATA_FORK)
+		ASSERT(newblk->blkno >= state->mp->m_dirleafblk &&
+		       newblk->blkno < state->mp->m_dirfreeblk);
+
+	/*
+	 * We may need to make some room before we insert the new node.
+	 */
+	tmp = 0;
+	btree = &node->btree[ oldblk->index ];
+	if (oldblk->index < be16_to_cpu(node->hdr.count)) {
+		tmp = (be16_to_cpu(node->hdr.count) - oldblk->index) * (uint)sizeof(*btree);
+		memmove(btree + 1, btree, tmp);
+	}
+	btree->hashval = cpu_to_be32(newblk->hashval);
+	btree->before = cpu_to_be32(newblk->blkno);
+	xfs_da_log_buf(state->args->trans, oldblk->bp,
+		XFS_DA_LOGRANGE(node, btree, tmp + sizeof(*btree)));
+	be16_add_cpu(&node->hdr.count, 1);
+	xfs_da_log_buf(state->args->trans, oldblk->bp,
+		XFS_DA_LOGRANGE(node, &node->hdr, sizeof(node->hdr)));
+
+	/*
+	 * Copy the last hash value from the oldblk to propagate upwards.
+	 */
+	oldblk->hashval = be32_to_cpu(node->btree[be16_to_cpu(node->hdr.count)-1 ].hashval);
+}
+
+/*========================================================================
+ * Routines used for shrinking the Btree.
+ *========================================================================*/
+
+/*
+ * Deallocate an empty leaf node, remove it from its parent,
+ * possibly deallocating that block, etc...
+ */
+int
+xfs_da_join(xfs_da_state_t *state)
+{
+	xfs_da_state_blk_t *drop_blk, *save_blk;
+	int action, error;
+
+	action = 0;
+	drop_blk = &state->path.blk[ state->path.active-1 ];
+	save_blk = &state->altpath.blk[ state->path.active-1 ];
+	ASSERT(state->path.blk[0].magic == XFS_DA_NODE_MAGIC);
+	ASSERT(drop_blk->magic == XFS_ATTR_LEAF_MAGIC ||
+	       drop_blk->magic == XFS_DIR2_LEAFN_MAGIC);
+
+	/*
+	 * Walk back up the tree joining/deallocating as necessary.
+	 * When we stop dropping blocks, break out.
+	 */
+	for (  ; state->path.active >= 2; drop_blk--, save_blk--,
+		 state->path.active--) {
+		/*
+		 * See if we can combine the block with a neighbor.
+		 *   (action == 0) => no options, just leave
+		 *   (action == 1) => coalesce, then unlink
+		 *   (action == 2) => block empty, unlink it
+		 */
+		switch (drop_blk->magic) {
+		case XFS_ATTR_LEAF_MAGIC:
+			error = xfs_attr_leaf_toosmall(state, &action);
+			if (error)
+				return(error);
+			if (action == 0)
+				return(0);
+			xfs_attr_leaf_unbalance(state, drop_blk, save_blk);
+			break;
+		case XFS_DIR2_LEAFN_MAGIC:
+			error = xfs_dir2_leafn_toosmall(state, &action);
+			if (error)
+				return error;
+			if (action == 0)
+				return 0;
+			xfs_dir2_leafn_unbalance(state, drop_blk, save_blk);
+			break;
+		case XFS_DA_NODE_MAGIC:
+			/*
+			 * Remove the offending node, fixup hashvals,
+			 * check for a toosmall neighbor.
+			 */
+			xfs_da_node_remove(state, drop_blk);
+			xfs_da_fixhashpath(state, &state->path);
+			error = xfs_da_node_toosmall(state, &action);
+			if (error)
+				return(error);
+			if (action == 0)
+				return 0;
+			xfs_da_node_unbalance(state, drop_blk, save_blk);
+			break;
+		}
+		xfs_da_fixhashpath(state, &state->altpath);
+		error = xfs_da_blk_unlink(state, drop_blk, save_blk);
+		xfs_da_state_kill_altpath(state);
+		if (error)
+			return(error);
+		error = xfs_da_shrink_inode(state->args, drop_blk->blkno,
+							 drop_blk->bp);
+		drop_blk->bp = NULL;
+		if (error)
+			return(error);
+	}
+	/*
+	 * We joined all the way to the top.  If it turns out that
+	 * we only have one entry in the root, make the child block
+	 * the new root.
+	 */
+	xfs_da_node_remove(state, drop_blk);
+	xfs_da_fixhashpath(state, &state->path);
+	error = xfs_da_root_join(state, &state->path.blk[0]);
+	return(error);
+}
+
+/*
+ * We have only one entry in the root.  Copy the only remaining child of
+ * the old root to block 0 as the new root node.
+ */
+STATIC int
+xfs_da_root_join(xfs_da_state_t *state, xfs_da_state_blk_t *root_blk)
+{
+	xfs_da_intnode_t *oldroot;
+	/* REFERENCED */
+	xfs_da_blkinfo_t *blkinfo;
+	xfs_da_args_t *args;
+	xfs_dablk_t child;
+	xfs_dabuf_t *bp;
+	int error;
+
+	args = state->args;
+	ASSERT(args != NULL);
+	ASSERT(root_blk->magic == XFS_DA_NODE_MAGIC);
+	oldroot = root_blk->bp->data;
+	ASSERT(be16_to_cpu(oldroot->hdr.info.magic) == XFS_DA_NODE_MAGIC);
+	ASSERT(!oldroot->hdr.info.forw);
+	ASSERT(!oldroot->hdr.info.back);
+
+	/*
+	 * If the root has more than one child, then don't do anything.
+	 */
+	if (be16_to_cpu(oldroot->hdr.count) > 1)
+		return(0);
+
+	/*
+	 * Read in the (only) child block, then copy those bytes into
+	 * the root block's buffer and free the original child block.
+	 */
+	child = be32_to_cpu(oldroot->btree[0].before);
+	ASSERT(child != 0);
+	error = xfs_da_read_buf(args->trans, args->dp, child, -1, &bp,
+					     args->whichfork);
+	if (error)
+		return(error);
+	ASSERT(bp != NULL);
+	blkinfo = bp->data;
+	if (be16_to_cpu(oldroot->hdr.level) == 1) {
+		ASSERT(be16_to_cpu(blkinfo->magic) == XFS_DIR2_LEAFN_MAGIC ||
+		       be16_to_cpu(blkinfo->magic) == XFS_ATTR_LEAF_MAGIC);
+	} else {
+		ASSERT(be16_to_cpu(blkinfo->magic) == XFS_DA_NODE_MAGIC);
+	}
+	ASSERT(!blkinfo->forw);
+	ASSERT(!blkinfo->back);
+	memcpy(root_blk->bp->data, bp->data, state->blocksize);
+	xfs_da_log_buf(args->trans, root_blk->bp, 0, state->blocksize - 1);
+	error = xfs_da_shrink_inode(args, child, bp);
+	return(error);
+}
+
+/*
+ * Check a node block and its neighbors to see if the block should be
+ * collapsed into one or the other neighbor.  Always keep the block
+ * with the smaller block number.
+ * If the current block is over 50% full, don't try to join it, return 0.
+ * If the block is empty, fill in the state structure and return 2.
+ * If it can be collapsed, fill in the state structure and return 1.
+ * If nothing can be done, return 0.
+ */
+STATIC int
+xfs_da_node_toosmall(xfs_da_state_t *state, int *action)
+{
+	xfs_da_intnode_t *node;
+	xfs_da_state_blk_t *blk;
+	xfs_da_blkinfo_t *info;
+	int count, forward, error, retval, i;
+	xfs_dablk_t blkno;
+	xfs_dabuf_t *bp;
+
+	/*
+	 * Check for the degenerate case of the block being over 50% full.
+	 * If so, it's not worth even looking to see if we might be able
+	 * to coalesce with a sibling.
+	 */
+	blk = &state->path.blk[ state->path.active-1 ];
+	info = blk->bp->data;
+	ASSERT(be16_to_cpu(info->magic) == XFS_DA_NODE_MAGIC);
+	node = (xfs_da_intnode_t *)info;
+	count = be16_to_cpu(node->hdr.count);
+	if (count > (state->node_ents >> 1)) {
+		*action = 0;	/* blk over 50%, don't try to join */
+		return(0);	/* blk over 50%, don't try to join */
+	}
+
+	/*
+	 * Check for the degenerate case of the block being empty.
+	 * If the block is empty, we'll simply delete it, no need to
+	 * coalesce it with a sibling block.  We choose (arbitrarily)
+	 * to merge with the forward block unless it is NULL.
+	 */
+	if (count == 0) {
+		/*
+		 * Make altpath point to the block we want to keep and
+		 * path point to the block we want to drop (this one).
+		 */
+		forward = (info->forw != 0);
+		memcpy(&state->altpath, &state->path, sizeof(state->path));
+		error = xfs_da_path_shift(state, &state->altpath, forward,
+						 0, &retval);
+		if (error)
+			return(error);
+		if (retval) {
+			*action = 0;
+		} else {
+			*action = 2;
+		}
+		return(0);
+	}
+
+	/*
+	 * Examine each sibling block to see if we can coalesce with
+	 * at least 25% free space to spare.  We need to figure out
+	 * whether to merge with the forward or the backward block.
+	 * We prefer coalescing with the lower numbered sibling so as
+	 * to shrink a directory over time.
+	 */
+	/* start with smaller blk num */
+	forward = (be32_to_cpu(info->forw) < be32_to_cpu(info->back));
+	for (i = 0; i < 2; forward = !forward, i++) {
+		if (forward)
+			blkno = be32_to_cpu(info->forw);
+		else
+			blkno = be32_to_cpu(info->back);
+		if (blkno == 0)
+			continue;
+		error = xfs_da_read_buf(state->args->trans, state->args->dp,
+					blkno, -1, &bp, state->args->whichfork);
+		if (error)
+			return(error);
+		ASSERT(bp != NULL);
+
+		node = (xfs_da_intnode_t *)info;
+		count  = state->node_ents;
+		count -= state->node_ents >> 2;
+		count -= be16_to_cpu(node->hdr.count);
+		node = bp->data;
+		ASSERT(be16_to_cpu(node->hdr.info.magic) == XFS_DA_NODE_MAGIC);
+		count -= be16_to_cpu(node->hdr.count);
+		xfs_da_brelse(state->args->trans, bp);
+		if (count >= 0)
+			break;	/* fits with at least 25% to spare */
+	}
+	if (i >= 2) {
+		*action = 0;
+		return(0);
+	}
+
+	/*
+	 * Make altpath point to the block we want to keep (the lower
+	 * numbered block) and path point to the block we want to drop.
+	 */
+	memcpy(&state->altpath, &state->path, sizeof(state->path));
+	if (blkno < blk->blkno) {
+		error = xfs_da_path_shift(state, &state->altpath, forward,
+						 0, &retval);
+		if (error) {
+			return(error);
+		}
+		if (retval) {
+			*action = 0;
+			return(0);
+		}
+	} else {
+		error = xfs_da_path_shift(state, &state->path, forward,
+						 0, &retval);
+		if (error) {
+			return(error);
+		}
+		if (retval) {
+			*action = 0;
+			return(0);
+		}
+	}
+	*action = 1;
+	return(0);
+}
+
+/*
+ * Walk back up the tree adjusting hash values as necessary,
+ * when we stop making changes, return.
+ */
+void
+xfs_da_fixhashpath(xfs_da_state_t *state, xfs_da_state_path_t *path)
+{
+	xfs_da_state_blk_t *blk;
+	xfs_da_intnode_t *node;
+	xfs_da_node_entry_t *btree;
+	xfs_dahash_t lasthash=0;
+	int level, count;
+
+	level = path->active-1;
+	blk = &path->blk[ level ];
+	switch (blk->magic) {
+	case XFS_ATTR_LEAF_MAGIC:
+		lasthash = xfs_attr_leaf_lasthash(blk->bp, &count);
+		if (count == 0)
+			return;
+		break;
+	case XFS_DIR2_LEAFN_MAGIC:
+		lasthash = xfs_dir2_leafn_lasthash(blk->bp, &count);
+		if (count == 0)
+			return;
+		break;
+	case XFS_DA_NODE_MAGIC:
+		lasthash = xfs_da_node_lasthash(blk->bp, &count);
+		if (count == 0)
+			return;
+		break;
+	}
+	for (blk--, level--; level >= 0; blk--, level--) {
+		node = blk->bp->data;
+		ASSERT(be16_to_cpu(node->hdr.info.magic) == XFS_DA_NODE_MAGIC);
+		btree = &node->btree[ blk->index ];
+		if (be32_to_cpu(btree->hashval) == lasthash)
+			break;
+		blk->hashval = lasthash;
+		btree->hashval = cpu_to_be32(lasthash);
+		xfs_da_log_buf(state->args->trans, blk->bp,
+				  XFS_DA_LOGRANGE(node, btree, sizeof(*btree)));
+
+		lasthash = be32_to_cpu(node->btree[be16_to_cpu(node->hdr.count)-1].hashval);
+	}
+}
+
+/*
+ * Remove an entry from an intermediate node.
+ */
+STATIC void
+xfs_da_node_remove(xfs_da_state_t *state, xfs_da_state_blk_t *drop_blk)
+{
+	xfs_da_intnode_t *node;
+	xfs_da_node_entry_t *btree;
+	int tmp;
+
+	node = drop_blk->bp->data;
+	ASSERT(drop_blk->index < be16_to_cpu(node->hdr.count));
+	ASSERT(drop_blk->index >= 0);
+
+	/*
+	 * Copy over the offending entry, or just zero it out.
+	 */
+	btree = &node->btree[drop_blk->index];
+	if (drop_blk->index < (be16_to_cpu(node->hdr.count)-1)) {
+		tmp  = be16_to_cpu(node->hdr.count) - drop_blk->index - 1;
+		tmp *= (uint)sizeof(xfs_da_node_entry_t);
+		memmove(btree, btree + 1, tmp);
+		xfs_da_log_buf(state->args->trans, drop_blk->bp,
+		    XFS_DA_LOGRANGE(node, btree, tmp));
+		btree = &node->btree[be16_to_cpu(node->hdr.count)-1];
+	}
+	memset((char *)btree, 0, sizeof(xfs_da_node_entry_t));
+	xfs_da_log_buf(state->args->trans, drop_blk->bp,
+	    XFS_DA_LOGRANGE(node, btree, sizeof(*btree)));
+	be16_add_cpu(&node->hdr.count, -1);
+	xfs_da_log_buf(state->args->trans, drop_blk->bp,
+	    XFS_DA_LOGRANGE(node, &node->hdr, sizeof(node->hdr)));
+
+	/*
+	 * Copy the last hash value from the block to propagate upwards.
+	 */
+	btree--;
+	drop_blk->hashval = be32_to_cpu(btree->hashval);
+}
+
+/*
+ * Unbalance the btree elements between two intermediate nodes,
+ * move all Btree elements from one node into another.
+ */
+STATIC void
+xfs_da_node_unbalance(xfs_da_state_t *state, xfs_da_state_blk_t *drop_blk,
+				     xfs_da_state_blk_t *save_blk)
+{
+	xfs_da_intnode_t *drop_node, *save_node;
+	xfs_da_node_entry_t *btree;
+	int tmp;
+	xfs_trans_t *tp;
+
+	drop_node = drop_blk->bp->data;
+	save_node = save_blk->bp->data;
+	ASSERT(be16_to_cpu(drop_node->hdr.info.magic) == XFS_DA_NODE_MAGIC);
+	ASSERT(be16_to_cpu(save_node->hdr.info.magic) == XFS_DA_NODE_MAGIC);
+	tp = state->args->trans;
+
+	/*
+	 * If the dying block has lower hashvals, then move all the
+	 * elements in the remaining block up to make a hole.
+	 */
+	if ((be32_to_cpu(drop_node->btree[0].hashval) < be32_to_cpu(save_node->btree[ 0 ].hashval)) ||
+	    (be32_to_cpu(drop_node->btree[be16_to_cpu(drop_node->hdr.count)-1].hashval) <
+	     be32_to_cpu(save_node->btree[be16_to_cpu(save_node->hdr.count)-1].hashval)))
+	{
+		btree = &save_node->btree[be16_to_cpu(drop_node->hdr.count)];
+		tmp = be16_to_cpu(save_node->hdr.count) * (uint)sizeof(xfs_da_node_entry_t);
+		memmove(btree, &save_node->btree[0], tmp);
+		btree = &save_node->btree[0];
+		xfs_da_log_buf(tp, save_blk->bp,
+			XFS_DA_LOGRANGE(save_node, btree,
+				(be16_to_cpu(save_node->hdr.count) + be16_to_cpu(drop_node->hdr.count)) *
+				sizeof(xfs_da_node_entry_t)));
+	} else {
+		btree = &save_node->btree[be16_to_cpu(save_node->hdr.count)];
+		xfs_da_log_buf(tp, save_blk->bp,
+			XFS_DA_LOGRANGE(save_node, btree,
+				be16_to_cpu(drop_node->hdr.count) *
+				sizeof(xfs_da_node_entry_t)));
+	}
+
+	/*
+	 * Move all the B-tree elements from drop_blk to save_blk.
+	 */
+	tmp = be16_to_cpu(drop_node->hdr.count) * (uint)sizeof(xfs_da_node_entry_t);
+	memcpy(btree, &drop_node->btree[0], tmp);
+	be16_add_cpu(&save_node->hdr.count, be16_to_cpu(drop_node->hdr.count));
+
+	xfs_da_log_buf(tp, save_blk->bp,
+		XFS_DA_LOGRANGE(save_node, &save_node->hdr,
+			sizeof(save_node->hdr)));
+
+	/*
+	 * Save the last hashval in the remaining block for upward propagation.
+	 */
+	save_blk->hashval = be32_to_cpu(save_node->btree[be16_to_cpu(save_node->hdr.count)-1].hashval);
+}
+
+/*========================================================================
+ * Routines used for finding things in the Btree.
+ *========================================================================*/
+
+/*
+ * Walk down the Btree looking for a particular filename, filling
+ * in the state structure as we go.
+ *
+ * We will set the state structure to point to each of the elements
+ * in each of the nodes where either the hashval is or should be.
+ *
+ * We support duplicate hashval's so for each entry in the current
+ * node that could contain the desired hashval, descend.  This is a
+ * pruned depth-first tree search.
+ */
+int							/* error */
+xfs_da_node_lookup_int(xfs_da_state_t *state, int *result)
+{
+	xfs_da_state_blk_t *blk;
+	xfs_da_blkinfo_t *curr;
+	xfs_da_intnode_t *node;
+	xfs_da_node_entry_t *btree;
+	xfs_dablk_t blkno;
+	int probe, span, max, error, retval;
+	xfs_dahash_t hashval, btreehashval;
+	xfs_da_args_t *args;
+
+	args = state->args;
+
+	/*
+	 * Descend thru the B-tree searching each level for the right
+	 * node to use, until the right hashval is found.
+	 */
+	blkno = (args->whichfork == XFS_DATA_FORK)? state->mp->m_dirleafblk : 0;
+	for (blk = &state->path.blk[0], state->path.active = 1;
+			 state->path.active <= XFS_DA_NODE_MAXDEPTH;
+			 blk++, state->path.active++) {
+		/*
+		 * Read the next node down in the tree.
+		 */
+		blk->blkno = blkno;
+		error = xfs_da_read_buf(args->trans, args->dp, blkno,
+					-1, &blk->bp, args->whichfork);
+		if (error) {
+			blk->blkno = 0;
+			state->path.active--;
+			return(error);
+		}
+		curr = blk->bp->data;
+		blk->magic = be16_to_cpu(curr->magic);
+		ASSERT(blk->magic == XFS_DA_NODE_MAGIC ||
+		       blk->magic == XFS_DIR2_LEAFN_MAGIC ||
+		       blk->magic == XFS_ATTR_LEAF_MAGIC);
+
+		/*
+		 * Search an intermediate node for a match.
+		 */
+		if (blk->magic == XFS_DA_NODE_MAGIC) {
+			node = blk->bp->data;
+			max = be16_to_cpu(node->hdr.count);
+			blk->hashval = be32_to_cpu(node->btree[max-1].hashval);
+
+			/*
+			 * Binary search.  (note: small blocks will skip loop)
+			 */
+			probe = span = max / 2;
+			hashval = args->hashval;
+			for (btree = &node->btree[probe]; span > 4;
+				   btree = &node->btree[probe]) {
+				span /= 2;
+				btreehashval = be32_to_cpu(btree->hashval);
+				if (btreehashval < hashval)
+					probe += span;
+				else if (btreehashval > hashval)
+					probe -= span;
+				else
+					break;
+			}
+			ASSERT((probe >= 0) && (probe < max));
+			ASSERT((span <= 4) || (be32_to_cpu(btree->hashval) == hashval));
+
+			/*
+			 * Since we may have duplicate hashval's, find the first
+			 * matching hashval in the node.
+			 */
+			while ((probe > 0) && (be32_to_cpu(btree->hashval) >= hashval)) {
+				btree--;
+				probe--;
+			}
+			while ((probe < max) && (be32_to_cpu(btree->hashval) < hashval)) {
+				btree++;
+				probe++;
+			}
+
+			/*
+			 * Pick the right block to descend on.
+			 */
+			if (probe == max) {
+				blk->index = max-1;
+				blkno = be32_to_cpu(node->btree[max-1].before);
+			} else {
+				blk->index = probe;
+				blkno = be32_to_cpu(btree->before);
+			}
+		} else if (blk->magic == XFS_ATTR_LEAF_MAGIC) {
+			blk->hashval = xfs_attr_leaf_lasthash(blk->bp, NULL);
+			break;
+		} else if (blk->magic == XFS_DIR2_LEAFN_MAGIC) {
+			blk->hashval = xfs_dir2_leafn_lasthash(blk->bp, NULL);
+			break;
+		}
+	}
+
+	/*
+	 * A leaf block that ends in the hashval that we are interested in
+	 * (final hashval == search hashval) means that the next block may
+	 * contain more entries with the same hashval, shift upward to the
+	 * next leaf and keep searching.
+	 */
+	for (;;) {
+		if (blk->magic == XFS_DIR2_LEAFN_MAGIC) {
+			retval = xfs_dir2_leafn_lookup_int(blk->bp, args,
+							&blk->index, state);
+		} else if (blk->magic == XFS_ATTR_LEAF_MAGIC) {
+			retval = xfs_attr_leaf_lookup_int(blk->bp, args);
+			blk->index = args->index;
+			args->blkno = blk->blkno;
+		} else {
+			ASSERT(0);
+			return XFS_ERROR(EFSCORRUPTED);
+		}
+		if (((retval == ENOENT) || (retval == ENOATTR)) &&
+		    (blk->hashval == args->hashval)) {
+			error = xfs_da_path_shift(state, &state->path, 1, 1,
+							 &retval);
+			if (error)
+				return(error);
+			if (retval == 0) {
+				continue;
+			} else if (blk->magic == XFS_ATTR_LEAF_MAGIC) {
+				/* path_shift() gives ENOENT */
+				retval = XFS_ERROR(ENOATTR);
+			}
+		}
+		break;
+	}
+	*result = retval;
+	return(0);
+}
+
+/*========================================================================
+ * Utility routines.
+ *========================================================================*/
+
+/*
+ * Link a new block into a doubly linked list of blocks (of whatever type).
+ */
+int							/* error */
+xfs_da_blk_link(xfs_da_state_t *state, xfs_da_state_blk_t *old_blk,
+			       xfs_da_state_blk_t *new_blk)
+{
+	xfs_da_blkinfo_t *old_info, *new_info, *tmp_info;
+	xfs_da_args_t *args;
+	int before=0, error;
+	xfs_dabuf_t *bp;
+
+	/*
+	 * Set up environment.
+	 */
+	args = state->args;
+	ASSERT(args != NULL);
+	old_info = old_blk->bp->data;
+	new_info = new_blk->bp->data;
+	ASSERT(old_blk->magic == XFS_DA_NODE_MAGIC ||
+	       old_blk->magic == XFS_DIR2_LEAFN_MAGIC ||
+	       old_blk->magic == XFS_ATTR_LEAF_MAGIC);
+	ASSERT(old_blk->magic == be16_to_cpu(old_info->magic));
+	ASSERT(new_blk->magic == be16_to_cpu(new_info->magic));
+	ASSERT(old_blk->magic == new_blk->magic);
+
+	switch (old_blk->magic) {
+	case XFS_ATTR_LEAF_MAGIC:
+		before = xfs_attr_leaf_order(old_blk->bp, new_blk->bp);
+		break;
+	case XFS_DIR2_LEAFN_MAGIC:
+		before = xfs_dir2_leafn_order(old_blk->bp, new_blk->bp);
+		break;
+	case XFS_DA_NODE_MAGIC:
+		before = xfs_da_node_order(old_blk->bp, new_blk->bp);
+		break;
+	}
+
+	/*
+	 * Link blocks in appropriate order.
+	 */
+	if (before) {
+		/*
+		 * Link new block in before existing block.
+		 */
+		new_info->forw = cpu_to_be32(old_blk->blkno);
+		new_info->back = old_info->back;
+		if (old_info->back) {
+			error = xfs_da_read_buf(args->trans, args->dp,
+						be32_to_cpu(old_info->back),
+						-1, &bp, args->whichfork);
+			if (error)
+				return(error);
+			ASSERT(bp != NULL);
+			tmp_info = bp->data;
+			ASSERT(be16_to_cpu(tmp_info->magic) == be16_to_cpu(old_info->magic));
+			ASSERT(be32_to_cpu(tmp_info->forw) == old_blk->blkno);
+			tmp_info->forw = cpu_to_be32(new_blk->blkno);
+			xfs_da_log_buf(args->trans, bp, 0, sizeof(*tmp_info)-1);
+			xfs_da_buf_done(bp);
+		}
+		old_info->back = cpu_to_be32(new_blk->blkno);
+	} else {
+		/*
+		 * Link new block in after existing block.
+		 */
+		new_info->forw = old_info->forw;
+		new_info->back = cpu_to_be32(old_blk->blkno);
+		if (old_info->forw) {
+			error = xfs_da_read_buf(args->trans, args->dp,
+						be32_to_cpu(old_info->forw),
+						-1, &bp, args->whichfork);
+			if (error)
+				return(error);
+			ASSERT(bp != NULL);
+			tmp_info = bp->data;
+			ASSERT(tmp_info->magic == old_info->magic);
+			ASSERT(be32_to_cpu(tmp_info->back) == old_blk->blkno);
+			tmp_info->back = cpu_to_be32(new_blk->blkno);
+			xfs_da_log_buf(args->trans, bp, 0, sizeof(*tmp_info)-1);
+			xfs_da_buf_done(bp);
+		}
+		old_info->forw = cpu_to_be32(new_blk->blkno);
+	}
+
+	xfs_da_log_buf(args->trans, old_blk->bp, 0, sizeof(*tmp_info) - 1);
+	xfs_da_log_buf(args->trans, new_blk->bp, 0, sizeof(*tmp_info) - 1);
+	return(0);
+}
+
+/*
+ * Compare two intermediate nodes for "order".
+ */
+STATIC int
+xfs_da_node_order(xfs_dabuf_t *node1_bp, xfs_dabuf_t *node2_bp)
+{
+	xfs_da_intnode_t *node1, *node2;
+
+	node1 = node1_bp->data;
+	node2 = node2_bp->data;
+	ASSERT((be16_to_cpu(node1->hdr.info.magic) == XFS_DA_NODE_MAGIC) &&
+	       (be16_to_cpu(node2->hdr.info.magic) == XFS_DA_NODE_MAGIC));
+	if ((be16_to_cpu(node1->hdr.count) > 0) && (be16_to_cpu(node2->hdr.count) > 0) &&
+	    ((be32_to_cpu(node2->btree[0].hashval) <
+	      be32_to_cpu(node1->btree[0].hashval)) ||
+	     (be32_to_cpu(node2->btree[be16_to_cpu(node2->hdr.count)-1].hashval) <
+	      be32_to_cpu(node1->btree[be16_to_cpu(node1->hdr.count)-1].hashval)))) {
+		return(1);
+	}
+	return(0);
+}
+
+/*
+ * Pick up the last hashvalue from an intermediate node.
+ */
+STATIC uint
+xfs_da_node_lasthash(xfs_dabuf_t *bp, int *count)
+{
+	xfs_da_intnode_t *node;
+
+	node = bp->data;
+	ASSERT(be16_to_cpu(node->hdr.info.magic) == XFS_DA_NODE_MAGIC);
+	if (count)
+		*count = be16_to_cpu(node->hdr.count);
+	if (!node->hdr.count)
+		return(0);
+	return be32_to_cpu(node->btree[be16_to_cpu(node->hdr.count)-1].hashval);
+}
+
+/*
+ * Unlink a block from a doubly linked list of blocks.
+ */
+STATIC int						/* error */
+xfs_da_blk_unlink(xfs_da_state_t *state, xfs_da_state_blk_t *drop_blk,
+				 xfs_da_state_blk_t *save_blk)
+{
+	xfs_da_blkinfo_t *drop_info, *save_info, *tmp_info;
+	xfs_da_args_t *args;
+	xfs_dabuf_t *bp;
+	int error;
+
+	/*
+	 * Set up environment.
+	 */
+	args = state->args;
+	ASSERT(args != NULL);
+	save_info = save_blk->bp->data;
+	drop_info = drop_blk->bp->data;
+	ASSERT(save_blk->magic == XFS_DA_NODE_MAGIC ||
+	       save_blk->magic == XFS_DIR2_LEAFN_MAGIC ||
+	       save_blk->magic == XFS_ATTR_LEAF_MAGIC);
+	ASSERT(save_blk->magic == be16_to_cpu(save_info->magic));
+	ASSERT(drop_blk->magic == be16_to_cpu(drop_info->magic));
+	ASSERT(save_blk->magic == drop_blk->magic);
+	ASSERT((be32_to_cpu(save_info->forw) == drop_blk->blkno) ||
+	       (be32_to_cpu(save_info->back) == drop_blk->blkno));
+	ASSERT((be32_to_cpu(drop_info->forw) == save_blk->blkno) ||
+	       (be32_to_cpu(drop_info->back) == save_blk->blkno));
+
+	/*
+	 * Unlink the leaf block from the doubly linked chain of leaves.
+	 */
+	if (be32_to_cpu(save_info->back) == drop_blk->blkno) {
+		save_info->back = drop_info->back;
+		if (drop_info->back) {
+			error = xfs_da_read_buf(args->trans, args->dp,
+						be32_to_cpu(drop_info->back),
+						-1, &bp, args->whichfork);
+			if (error)
+				return(error);
+			ASSERT(bp != NULL);
+			tmp_info = bp->data;
+			ASSERT(tmp_info->magic == save_info->magic);
+			ASSERT(be32_to_cpu(tmp_info->forw) == drop_blk->blkno);
+			tmp_info->forw = cpu_to_be32(save_blk->blkno);
+			xfs_da_log_buf(args->trans, bp, 0,
+						    sizeof(*tmp_info) - 1);
+			xfs_da_buf_done(bp);
+		}
+	} else {
+		save_info->forw = drop_info->forw;
+		if (drop_info->forw) {
+			error = xfs_da_read_buf(args->trans, args->dp,
+						be32_to_cpu(drop_info->forw),
+						-1, &bp, args->whichfork);
+			if (error)
+				return(error);
+			ASSERT(bp != NULL);
+			tmp_info = bp->data;
+			ASSERT(tmp_info->magic == save_info->magic);
+			ASSERT(be32_to_cpu(tmp_info->back) == drop_blk->blkno);
+			tmp_info->back = cpu_to_be32(save_blk->blkno);
+			xfs_da_log_buf(args->trans, bp, 0,
+						    sizeof(*tmp_info) - 1);
+			xfs_da_buf_done(bp);
+		}
+	}
+
+	xfs_da_log_buf(args->trans, save_blk->bp, 0, sizeof(*save_info) - 1);
+	return(0);
+}
+
+/*
+ * Move a path "forward" or "!forward" one block at the current level.
+ *
+ * This routine will adjust a "path" to point to the next block
+ * "forward" (higher hashvalues) or "!forward" (lower hashvals) in the
+ * Btree, including updating pointers to the intermediate nodes between
+ * the new bottom and the root.
+ */
+int							/* error */
+xfs_da_path_shift(xfs_da_state_t *state, xfs_da_state_path_t *path,
+				 int forward, int release, int *result)
+{
+	xfs_da_state_blk_t *blk;
+	xfs_da_blkinfo_t *info;
+	xfs_da_intnode_t *node;
+	xfs_da_args_t *args;
+	xfs_dablk_t blkno=0;
+	int level, error;
+
+	/*
+	 * Roll up the Btree looking for the first block where our
+	 * current index is not at the edge of the block.  Note that
+	 * we skip the bottom layer because we want the sibling block.
+	 */
+	args = state->args;
+	ASSERT(args != NULL);
+	ASSERT(path != NULL);
+	ASSERT((path->active > 0) && (path->active < XFS_DA_NODE_MAXDEPTH));
+	level = (path->active-1) - 1;	/* skip bottom layer in path */
+	for (blk = &path->blk[level]; level >= 0; blk--, level--) {
+		ASSERT(blk->bp != NULL);
+		node = blk->bp->data;
+		ASSERT(be16_to_cpu(node->hdr.info.magic) == XFS_DA_NODE_MAGIC);
+		if (forward && (blk->index < be16_to_cpu(node->hdr.count)-1)) {
+			blk->index++;
+			blkno = be32_to_cpu(node->btree[blk->index].before);
+			break;
+		} else if (!forward && (blk->index > 0)) {
+			blk->index--;
+			blkno = be32_to_cpu(node->btree[blk->index].before);
+			break;
+		}
+	}
+	if (level < 0) {
+		*result = XFS_ERROR(ENOENT);	/* we're out of our tree */
+		ASSERT(args->op_flags & XFS_DA_OP_OKNOENT);
+		return(0);
+	}
+
+	/*
+	 * Roll down the edge of the subtree until we reach the
+	 * same depth we were at originally.
+	 */
+	for (blk++, level++; level < path->active; blk++, level++) {
+		/*
+		 * Release the old block.
+		 * (if it's dirty, trans won't actually let go)
+		 */
+		if (release)
+			xfs_da_brelse(args->trans, blk->bp);
+
+		/*
+		 * Read the next child block.
+		 */
+		blk->blkno = blkno;
+		error = xfs_da_read_buf(args->trans, args->dp, blkno, -1,
+						     &blk->bp, args->whichfork);
+		if (error)
+			return(error);
+		ASSERT(blk->bp != NULL);
+		info = blk->bp->data;
+		ASSERT(be16_to_cpu(info->magic) == XFS_DA_NODE_MAGIC ||
+		       be16_to_cpu(info->magic) == XFS_DIR2_LEAFN_MAGIC ||
+		       be16_to_cpu(info->magic) == XFS_ATTR_LEAF_MAGIC);
+		blk->magic = be16_to_cpu(info->magic);
+		if (blk->magic == XFS_DA_NODE_MAGIC) {
+			node = (xfs_da_intnode_t *)info;
+			blk->hashval = be32_to_cpu(node->btree[be16_to_cpu(node->hdr.count)-1].hashval);
+			if (forward)
+				blk->index = 0;
+			else
+				blk->index = be16_to_cpu(node->hdr.count)-1;
+			blkno = be32_to_cpu(node->btree[blk->index].before);
+		} else {
+			ASSERT(level == path->active-1);
+			blk->index = 0;
+			switch(blk->magic) {
+			case XFS_ATTR_LEAF_MAGIC:
+				blk->hashval = xfs_attr_leaf_lasthash(blk->bp,
+								      NULL);
+				break;
+			case XFS_DIR2_LEAFN_MAGIC:
+				blk->hashval = xfs_dir2_leafn_lasthash(blk->bp,
+								       NULL);
+				break;
+			default:
+				ASSERT(blk->magic == XFS_ATTR_LEAF_MAGIC ||
+				       blk->magic == XFS_DIR2_LEAFN_MAGIC);
+				break;
+			}
+		}
+	}
+	*result = 0;
+	return(0);
+}
+
+
+/*========================================================================
+ * Utility routines.
+ *========================================================================*/
+
+/*
+ * Implement a simple hash on a character string.
+ * Rotate the hash value by 7 bits, then XOR each character in.
+ * This is implemented with some source-level loop unrolling.
+ */
+xfs_dahash_t
+xfs_da_hashname(const __uint8_t *name, int namelen)
+{
+	xfs_dahash_t hash;
+
+	/*
+	 * Do four characters at a time as long as we can.
+	 */
+	for (hash = 0; namelen >= 4; namelen -= 4, name += 4)
+		hash = (name[0] << 21) ^ (name[1] << 14) ^ (name[2] << 7) ^
+		       (name[3] << 0) ^ rol32(hash, 7 * 4);
+
+	/*
+	 * Now do the rest of the characters.
+	 */
+	switch (namelen) {
+	case 3:
+		return (name[0] << 14) ^ (name[1] << 7) ^ (name[2] << 0) ^
+		       rol32(hash, 7 * 3);
+	case 2:
+		return (name[0] << 7) ^ (name[1] << 0) ^ rol32(hash, 7 * 2);
+	case 1:
+		return (name[0] << 0) ^ rol32(hash, 7 * 1);
+	default: /* case 0: */
+		return hash;
+	}
+}
+
+enum xfs_dacmp
+xfs_da_compname(
+	struct xfs_da_args *args,
+	const unsigned char *name,
+	int		len)
+{
+	return (args->namelen == len && memcmp(args->name, name, len) == 0) ?
+					XFS_CMP_EXACT : XFS_CMP_DIFFERENT;
+}
+
+static xfs_dahash_t
+xfs_default_hashname(
+	struct xfs_name	*name)
+{
+	return xfs_da_hashname(name->name, name->len);
+}
+
+const struct xfs_nameops xfs_default_nameops = {
+	.hashname	= xfs_default_hashname,
+	.compname	= xfs_da_compname
+};
+
+/*
+ * Add a block to the btree ahead of the file.
+ * Return the new block number to the caller.
+ */
+int
+xfs_da_grow_inode(xfs_da_args_t *args, xfs_dablk_t *new_blkno)
+{
+	xfs_fileoff_t bno, b;
+	xfs_bmbt_irec_t map;
+	xfs_bmbt_irec_t	*mapp;
+	xfs_inode_t *dp;
+	int nmap, error, w, count, c, got, i, mapi;
+	xfs_trans_t *tp;
+	xfs_mount_t *mp;
+	xfs_drfsbno_t	nblks;
+
+	dp = args->dp;
+	mp = dp->i_mount;
+	w = args->whichfork;
+	tp = args->trans;
+	nblks = dp->i_d.di_nblocks;
+
+	/*
+	 * For new directories adjust the file offset and block count.
+	 */
+	if (w == XFS_DATA_FORK) {
+		bno = mp->m_dirleafblk;
+		count = mp->m_dirblkfsbs;
+	} else {
+		bno = 0;
+		count = 1;
+	}
+	/*
+	 * Find a spot in the file space to put the new block.
+	 */
+	if ((error = xfs_bmap_first_unused(tp, dp, count, &bno, w)))
+		return error;
+	if (w == XFS_DATA_FORK)
+		ASSERT(bno >= mp->m_dirleafblk && bno < mp->m_dirfreeblk);
+	/*
+	 * Try mapping it in one filesystem block.
+	 */
+	nmap = 1;
+	ASSERT(args->firstblock != NULL);
+	if ((error = xfs_bmapi(tp, dp, bno, count,
+			xfs_bmapi_aflag(w)|XFS_BMAPI_WRITE|XFS_BMAPI_METADATA|
+			XFS_BMAPI_CONTIG,
+			args->firstblock, args->total, &map, &nmap,
+			args->flist))) {
+		return error;
+	}
+	ASSERT(nmap <= 1);
+	if (nmap == 1) {
+		mapp = &map;
+		mapi = 1;
+	}
+	/*
+	 * If we didn't get it and the block might work if fragmented,
+	 * try without the CONTIG flag.  Loop until we get it all.
+	 */
+	else if (nmap == 0 && count > 1) {
+		mapp = kmem_alloc(sizeof(*mapp) * count, KM_SLEEP);
+		for (b = bno, mapi = 0; b < bno + count; ) {
+			nmap = MIN(XFS_BMAP_MAX_NMAP, count);
+			c = (int)(bno + count - b);
+			if ((error = xfs_bmapi(tp, dp, b, c,
+					xfs_bmapi_aflag(w)|XFS_BMAPI_WRITE|
+					XFS_BMAPI_METADATA,
+					args->firstblock, args->total,
+					&mapp[mapi], &nmap, args->flist))) {
+				kmem_free(mapp);
+				return error;
+			}
+			if (nmap < 1)
+				break;
+			mapi += nmap;
+			b = mapp[mapi - 1].br_startoff +
+			    mapp[mapi - 1].br_blockcount;
+		}
+	} else {
+		mapi = 0;
+		mapp = NULL;
+	}
+	/*
+	 * Count the blocks we got, make sure it matches the total.
+	 */
+	for (i = 0, got = 0; i < mapi; i++)
+		got += mapp[i].br_blockcount;
+	if (got != count || mapp[0].br_startoff != bno ||
+	    mapp[mapi - 1].br_startoff + mapp[mapi - 1].br_blockcount !=
+	    bno + count) {
+		if (mapp != &map)
+			kmem_free(mapp);
+		return XFS_ERROR(ENOSPC);
+	}
+	if (mapp != &map)
+		kmem_free(mapp);
+	/* account for newly allocated blocks in reserved blocks total */
+	args->total -= dp->i_d.di_nblocks - nblks;
+	*new_blkno = (xfs_dablk_t)bno;
+	return 0;
+}
+
+/*
+ * Ick.  We need to always be able to remove a btree block, even
+ * if there's no space reservation because the filesystem is full.
+ * This is called if xfs_bunmapi on a btree block fails due to ENOSPC.
+ * It swaps the target block with the last block in the file.  The
+ * last block in the file can always be removed since it can't cause
+ * a bmap btree split to do that.
+ */
+STATIC int
+xfs_da_swap_lastblock(xfs_da_args_t *args, xfs_dablk_t *dead_blknop,
+		      xfs_dabuf_t **dead_bufp)
+{
+	xfs_dablk_t dead_blkno, last_blkno, sib_blkno, par_blkno;
+	xfs_dabuf_t *dead_buf, *last_buf, *sib_buf, *par_buf;
+	xfs_fileoff_t lastoff;
+	xfs_inode_t *ip;
+	xfs_trans_t *tp;
+	xfs_mount_t *mp;
+	int error, w, entno, level, dead_level;
+	xfs_da_blkinfo_t *dead_info, *sib_info;
+	xfs_da_intnode_t *par_node, *dead_node;
+	xfs_dir2_leaf_t *dead_leaf2;
+	xfs_dahash_t dead_hash;
+
+	dead_buf = *dead_bufp;
+	dead_blkno = *dead_blknop;
+	tp = args->trans;
+	ip = args->dp;
+	w = args->whichfork;
+	ASSERT(w == XFS_DATA_FORK);
+	mp = ip->i_mount;
+	lastoff = mp->m_dirfreeblk;
+	error = xfs_bmap_last_before(tp, ip, &lastoff, w);
+	if (error)
+		return error;
+	if (unlikely(lastoff == 0)) {
+		XFS_ERROR_REPORT("xfs_da_swap_lastblock(1)", XFS_ERRLEVEL_LOW,
+				 mp);
+		return XFS_ERROR(EFSCORRUPTED);
+	}
+	/*
+	 * Read the last block in the btree space.
+	 */
+	last_blkno = (xfs_dablk_t)lastoff - mp->m_dirblkfsbs;
+	if ((error = xfs_da_read_buf(tp, ip, last_blkno, -1, &last_buf, w)))
+		return error;
+	/*
+	 * Copy the last block into the dead buffer and log it.
+	 */
+	memcpy(dead_buf->data, last_buf->data, mp->m_dirblksize);
+	xfs_da_log_buf(tp, dead_buf, 0, mp->m_dirblksize - 1);
+	dead_info = dead_buf->data;
+	/*
+	 * Get values from the moved block.
+	 */
+	if (be16_to_cpu(dead_info->magic) == XFS_DIR2_LEAFN_MAGIC) {
+		dead_leaf2 = (xfs_dir2_leaf_t *)dead_info;
+		dead_level = 0;
+		dead_hash = be32_to_cpu(dead_leaf2->ents[be16_to_cpu(dead_leaf2->hdr.count) - 1].hashval);
+	} else {
+		ASSERT(be16_to_cpu(dead_info->magic) == XFS_DA_NODE_MAGIC);
+		dead_node = (xfs_da_intnode_t *)dead_info;
+		dead_level = be16_to_cpu(dead_node->hdr.level);
+		dead_hash = be32_to_cpu(dead_node->btree[be16_to_cpu(dead_node->hdr.count) - 1].hashval);
+	}
+	sib_buf = par_buf = NULL;
+	/*
+	 * If the moved block has a left sibling, fix up the pointers.
+	 */
+	if ((sib_blkno = be32_to_cpu(dead_info->back))) {
+		if ((error = xfs_da_read_buf(tp, ip, sib_blkno, -1, &sib_buf, w)))
+			goto done;
+		sib_info = sib_buf->data;
+		if (unlikely(
+		    be32_to_cpu(sib_info->forw) != last_blkno ||
+		    sib_info->magic != dead_info->magic)) {
+			XFS_ERROR_REPORT("xfs_da_swap_lastblock(2)",
+					 XFS_ERRLEVEL_LOW, mp);
+			error = XFS_ERROR(EFSCORRUPTED);
+			goto done;
+		}
+		sib_info->forw = cpu_to_be32(dead_blkno);
+		xfs_da_log_buf(tp, sib_buf,
+			XFS_DA_LOGRANGE(sib_info, &sib_info->forw,
+					sizeof(sib_info->forw)));
+		xfs_da_buf_done(sib_buf);
+		sib_buf = NULL;
+	}
+	/*
+	 * If the moved block has a right sibling, fix up the pointers.
+	 */
+	if ((sib_blkno = be32_to_cpu(dead_info->forw))) {
+		if ((error = xfs_da_read_buf(tp, ip, sib_blkno, -1, &sib_buf, w)))
+			goto done;
+		sib_info = sib_buf->data;
+		if (unlikely(
+		       be32_to_cpu(sib_info->back) != last_blkno ||
+		       sib_info->magic != dead_info->magic)) {
+			XFS_ERROR_REPORT("xfs_da_swap_lastblock(3)",
+					 XFS_ERRLEVEL_LOW, mp);
+			error = XFS_ERROR(EFSCORRUPTED);
+			goto done;
+		}
+		sib_info->back = cpu_to_be32(dead_blkno);
+		xfs_da_log_buf(tp, sib_buf,
+			XFS_DA_LOGRANGE(sib_info, &sib_info->back,
+					sizeof(sib_info->back)));
+		xfs_da_buf_done(sib_buf);
+		sib_buf = NULL;
+	}
+	par_blkno = mp->m_dirleafblk;
+	level = -1;
+	/*
+	 * Walk down the tree looking for the parent of the moved block.
+	 */
+	for (;;) {
+		if ((error = xfs_da_read_buf(tp, ip, par_blkno, -1, &par_buf, w)))
+			goto done;
+		par_node = par_buf->data;
+		if (unlikely(
+		    be16_to_cpu(par_node->hdr.info.magic) != XFS_DA_NODE_MAGIC ||
+		    (level >= 0 && level != be16_to_cpu(par_node->hdr.level) + 1))) {
+			XFS_ERROR_REPORT("xfs_da_swap_lastblock(4)",
+					 XFS_ERRLEVEL_LOW, mp);
+			error = XFS_ERROR(EFSCORRUPTED);
+			goto done;
+		}
+		level = be16_to_cpu(par_node->hdr.level);
+		for (entno = 0;
+		     entno < be16_to_cpu(par_node->hdr.count) &&
+		     be32_to_cpu(par_node->btree[entno].hashval) < dead_hash;
+		     entno++)
+			continue;
+		if (unlikely(entno == be16_to_cpu(par_node->hdr.count))) {
+			XFS_ERROR_REPORT("xfs_da_swap_lastblock(5)",
+					 XFS_ERRLEVEL_LOW, mp);
+			error = XFS_ERROR(EFSCORRUPTED);
+			goto done;
+		}
+		par_blkno = be32_to_cpu(par_node->btree[entno].before);
+		if (level == dead_level + 1)
+			break;
+		xfs_da_brelse(tp, par_buf);
+		par_buf = NULL;
+	}
+	/*
+	 * We're in the right parent block.
+	 * Look for the right entry.
+	 */
+	for (;;) {
+		for (;
+		     entno < be16_to_cpu(par_node->hdr.count) &&
+		     be32_to_cpu(par_node->btree[entno].before) != last_blkno;
+		     entno++)
+			continue;
+		if (entno < be16_to_cpu(par_node->hdr.count))
+			break;
+		par_blkno = be32_to_cpu(par_node->hdr.info.forw);
+		xfs_da_brelse(tp, par_buf);
+		par_buf = NULL;
+		if (unlikely(par_blkno == 0)) {
+			XFS_ERROR_REPORT("xfs_da_swap_lastblock(6)",
+					 XFS_ERRLEVEL_LOW, mp);
+			error = XFS_ERROR(EFSCORRUPTED);
+			goto done;
+		}
+		if ((error = xfs_da_read_buf(tp, ip, par_blkno, -1, &par_buf, w)))
+			goto done;
+		par_node = par_buf->data;
+		if (unlikely(
+		    be16_to_cpu(par_node->hdr.level) != level ||
+		    be16_to_cpu(par_node->hdr.info.magic) != XFS_DA_NODE_MAGIC)) {
+			XFS_ERROR_REPORT("xfs_da_swap_lastblock(7)",
+					 XFS_ERRLEVEL_LOW, mp);
+			error = XFS_ERROR(EFSCORRUPTED);
+			goto done;
+		}
+		entno = 0;
+	}
+	/*
+	 * Update the parent entry pointing to the moved block.
+	 */
+	par_node->btree[entno].before = cpu_to_be32(dead_blkno);
+	xfs_da_log_buf(tp, par_buf,
+		XFS_DA_LOGRANGE(par_node, &par_node->btree[entno].before,
+				sizeof(par_node->btree[entno].before)));
+	xfs_da_buf_done(par_buf);
+	xfs_da_buf_done(dead_buf);
+	*dead_blknop = last_blkno;
+	*dead_bufp = last_buf;
+	return 0;
+done:
+	if (par_buf)
+		xfs_da_brelse(tp, par_buf);
+	if (sib_buf)
+		xfs_da_brelse(tp, sib_buf);
+	xfs_da_brelse(tp, last_buf);
+	return error;
+}
+
+/*
+ * Remove a btree block from a directory or attribute.
+ */
+int
+xfs_da_shrink_inode(xfs_da_args_t *args, xfs_dablk_t dead_blkno,
+		    xfs_dabuf_t *dead_buf)
+{
+	xfs_inode_t *dp;
+	int done, error, w, count;
+	xfs_trans_t *tp;
+	xfs_mount_t *mp;
+
+	dp = args->dp;
+	w = args->whichfork;
+	tp = args->trans;
+	mp = dp->i_mount;
+	if (w == XFS_DATA_FORK)
+		count = mp->m_dirblkfsbs;
+	else
+		count = 1;
+	for (;;) {
+		/*
+		 * Remove extents.  If we get ENOSPC for a dir we have to move
+		 * the last block to the place we want to kill.
+		 */
+		if ((error = xfs_bunmapi(tp, dp, dead_blkno, count,
+				xfs_bmapi_aflag(w)|XFS_BMAPI_METADATA,
+				0, args->firstblock, args->flist,
+				&done)) == ENOSPC) {
+			if (w != XFS_DATA_FORK)
+				break;
+			if ((error = xfs_da_swap_lastblock(args, &dead_blkno,
+					&dead_buf)))
+				break;
+		} else {
+			break;
+		}
+	}
+	xfs_da_binval(tp, dead_buf);
+	return error;
+}
+
+/*
+ * See if the mapping(s) for this btree block are valid, i.e.
+ * don't contain holes, are logically contiguous, and cover the whole range.
+ */
+STATIC int
+xfs_da_map_covers_blocks(
+	int		nmap,
+	xfs_bmbt_irec_t	*mapp,
+	xfs_dablk_t	bno,
+	int		count)
+{
+	int		i;
+	xfs_fileoff_t	off;
+
+	for (i = 0, off = bno; i < nmap; i++) {
+		if (mapp[i].br_startblock == HOLESTARTBLOCK ||
+		    mapp[i].br_startblock == DELAYSTARTBLOCK) {
+			return 0;
+		}
+		if (off != mapp[i].br_startoff) {
+			return 0;
+		}
+		off += mapp[i].br_blockcount;
+	}
+	return off == bno + count;
+}
+
+/*
+ * Make a dabuf.
+ * Used for get_buf, read_buf, read_bufr, and reada_buf.
+ */
+STATIC int
+xfs_da_do_buf(
+	xfs_trans_t	*trans,
+	xfs_inode_t	*dp,
+	xfs_dablk_t	bno,
+	xfs_daddr_t	*mappedbnop,
+	xfs_dabuf_t	**bpp,
+	int		whichfork,
+	int		caller,
+	inst_t		*ra)
+{
+	xfs_buf_t	*bp = NULL;
+	xfs_buf_t	**bplist;
+	int		error=0;
+	int		i;
+	xfs_bmbt_irec_t	map;
+	xfs_bmbt_irec_t	*mapp;
+	xfs_daddr_t	mappedbno;
+	xfs_mount_t	*mp;
+	int		nbplist=0;
+	int		nfsb;
+	int		nmap;
+	xfs_dabuf_t	*rbp;
+
+	mp = dp->i_mount;
+	nfsb = (whichfork == XFS_DATA_FORK) ? mp->m_dirblkfsbs : 1;
+	mappedbno = *mappedbnop;
+	/*
+	 * Caller doesn't have a mapping.  -2 means don't complain
+	 * if we land in a hole.
+	 */
+	if (mappedbno == -1 || mappedbno == -2) {
+		/*
+		 * Optimize the one-block case.
+		 */
+		if (nfsb == 1) {
+			xfs_fsblock_t	fsb;
+
+			if ((error =
+			    xfs_bmapi_single(trans, dp, whichfork, &fsb,
+				    (xfs_fileoff_t)bno))) {
+				return error;
+			}
+			mapp = &map;
+			if (fsb == NULLFSBLOCK) {
+				nmap = 0;
+			} else {
+				map.br_startblock = fsb;
+				map.br_startoff = (xfs_fileoff_t)bno;
+				map.br_blockcount = 1;
+				nmap = 1;
+			}
+		} else {
+			mapp = kmem_alloc(sizeof(*mapp) * nfsb, KM_SLEEP);
+			nmap = nfsb;
+			if ((error = xfs_bmapi(trans, dp, (xfs_fileoff_t)bno,
+					nfsb,
+					XFS_BMAPI_METADATA |
+						xfs_bmapi_aflag(whichfork),
+					NULL, 0, mapp, &nmap, NULL)))
+				goto exit0;
+		}
+	} else {
+		map.br_startblock = XFS_DADDR_TO_FSB(mp, mappedbno);
+		map.br_startoff = (xfs_fileoff_t)bno;
+		map.br_blockcount = nfsb;
+		mapp = &map;
+		nmap = 1;
+	}
+	if (!xfs_da_map_covers_blocks(nmap, mapp, bno, nfsb)) {
+		error = mappedbno == -2 ? 0 : XFS_ERROR(EFSCORRUPTED);
+		if (unlikely(error == EFSCORRUPTED)) {
+			if (xfs_error_level >= XFS_ERRLEVEL_LOW) {
+				xfs_alert(mp, "%s: bno %lld dir: inode %lld",
+					__func__, (long long)bno,
+					(long long)dp->i_ino);
+				for (i = 0; i < nmap; i++) {
+					xfs_alert(mp,
+"[%02d] br_startoff %lld br_startblock %lld br_blockcount %lld br_state %d",
+						i,
+						(long long)mapp[i].br_startoff,
+						(long long)mapp[i].br_startblock,
+						(long long)mapp[i].br_blockcount,
+						mapp[i].br_state);
+				}
+			}
+			XFS_ERROR_REPORT("xfs_da_do_buf(1)",
+					 XFS_ERRLEVEL_LOW, mp);
+		}
+		goto exit0;
+	}
+	if (caller != 3 && nmap > 1) {
+		bplist = kmem_alloc(sizeof(*bplist) * nmap, KM_SLEEP);
+		nbplist = 0;
+	} else
+		bplist = NULL;
+	/*
+	 * Turn the mapping(s) into buffer(s).
+	 */
+	for (i = 0; i < nmap; i++) {
+		int	nmapped;
+
+		mappedbno = XFS_FSB_TO_DADDR(mp, mapp[i].br_startblock);
+		if (i == 0)
+			*mappedbnop = mappedbno;
+		nmapped = (int)XFS_FSB_TO_BB(mp, mapp[i].br_blockcount);
+		switch (caller) {
+		case 0:
+			bp = xfs_trans_get_buf(trans, mp->m_ddev_targp,
+				mappedbno, nmapped, 0);
+			error = bp ? XFS_BUF_GETERROR(bp) : XFS_ERROR(EIO);
+			break;
+		case 1:
+		case 2:
+			bp = NULL;
+			error = xfs_trans_read_buf(mp, trans, mp->m_ddev_targp,
+				mappedbno, nmapped, 0, &bp);
+			break;
+		case 3:
+			xfs_buf_readahead(mp->m_ddev_targp, mappedbno, nmapped);
+			error = 0;
+			bp = NULL;
+			break;
+		}
+		if (error) {
+			if (bp)
+				xfs_trans_brelse(trans, bp);
+			goto exit1;
+		}
+		if (!bp)
+			continue;
+		if (caller == 1) {
+			if (whichfork == XFS_ATTR_FORK) {
+				XFS_BUF_SET_VTYPE_REF(bp, B_FS_ATTR_BTREE,
+						XFS_ATTR_BTREE_REF);
+			} else {
+				XFS_BUF_SET_VTYPE_REF(bp, B_FS_DIR_BTREE,
+						XFS_DIR_BTREE_REF);
+			}
+		}
+		if (bplist) {
+			bplist[nbplist++] = bp;
+		}
+	}
+	/*
+	 * Build a dabuf structure.
+	 */
+	if (bplist) {
+		rbp = xfs_da_buf_make(nbplist, bplist, ra);
+	} else if (bp)
+		rbp = xfs_da_buf_make(1, &bp, ra);
+	else
+		rbp = NULL;
+	/*
+	 * For read_buf, check the magic number.
+	 */
+	if (caller == 1) {
+		xfs_dir2_data_t		*data;
+		xfs_dir2_free_t		*free;
+		xfs_da_blkinfo_t	*info;
+		uint			magic, magic1;
+
+		info = rbp->data;
+		data = rbp->data;
+		free = rbp->data;
+		magic = be16_to_cpu(info->magic);
+		magic1 = be32_to_cpu(data->hdr.magic);
+		if (unlikely(
+		    XFS_TEST_ERROR((magic != XFS_DA_NODE_MAGIC) &&
+				   (magic != XFS_ATTR_LEAF_MAGIC) &&
+				   (magic != XFS_DIR2_LEAF1_MAGIC) &&
+				   (magic != XFS_DIR2_LEAFN_MAGIC) &&
+				   (magic1 != XFS_DIR2_BLOCK_MAGIC) &&
+				   (magic1 != XFS_DIR2_DATA_MAGIC) &&
+				   (be32_to_cpu(free->hdr.magic) != XFS_DIR2_FREE_MAGIC),
+				mp, XFS_ERRTAG_DA_READ_BUF,
+				XFS_RANDOM_DA_READ_BUF))) {
+			trace_xfs_da_btree_corrupt(rbp->bps[0], _RET_IP_);
+			XFS_CORRUPTION_ERROR("xfs_da_do_buf(2)",
+					     XFS_ERRLEVEL_LOW, mp, info);
+			error = XFS_ERROR(EFSCORRUPTED);
+			xfs_da_brelse(trans, rbp);
+			nbplist = 0;
+			goto exit1;
+		}
+	}
+	if (bplist) {
+		kmem_free(bplist);
+	}
+	if (mapp != &map) {
+		kmem_free(mapp);
+	}
+	if (bpp)
+		*bpp = rbp;
+	return 0;
+exit1:
+	if (bplist) {
+		for (i = 0; i < nbplist; i++)
+			xfs_trans_brelse(trans, bplist[i]);
+		kmem_free(bplist);
+	}
+exit0:
+	if (mapp != &map)
+		kmem_free(mapp);
+	if (bpp)
+		*bpp = NULL;
+	return error;
+}
+
+/*
+ * Get a buffer for the dir/attr block.
+ */
+int
+xfs_da_get_buf(
+	xfs_trans_t	*trans,
+	xfs_inode_t	*dp,
+	xfs_dablk_t	bno,
+	xfs_daddr_t		mappedbno,
+	xfs_dabuf_t	**bpp,
+	int		whichfork)
+{
+	return xfs_da_do_buf(trans, dp, bno, &mappedbno, bpp, whichfork, 0,
+						 (inst_t *)__return_address);
+}
+
+/*
+ * Get a buffer for the dir/attr block, fill in the contents.
+ */
+int
+xfs_da_read_buf(
+	xfs_trans_t	*trans,
+	xfs_inode_t	*dp,
+	xfs_dablk_t	bno,
+	xfs_daddr_t		mappedbno,
+	xfs_dabuf_t	**bpp,
+	int		whichfork)
+{
+	return xfs_da_do_buf(trans, dp, bno, &mappedbno, bpp, whichfork, 1,
+		(inst_t *)__return_address);
+}
+
+/*
+ * Readahead the dir/attr block.
+ */
+xfs_daddr_t
+xfs_da_reada_buf(
+	xfs_trans_t	*trans,
+	xfs_inode_t	*dp,
+	xfs_dablk_t	bno,
+	int		whichfork)
+{
+	xfs_daddr_t		rval;
+
+	rval = -1;
+	if (xfs_da_do_buf(trans, dp, bno, &rval, NULL, whichfork, 3,
+			(inst_t *)__return_address))
+		return -1;
+	else
+		return rval;
+}
+
+kmem_zone_t *xfs_da_state_zone;	/* anchor for state struct zone */
+kmem_zone_t *xfs_dabuf_zone;		/* dabuf zone */
+
+/*
+ * Allocate a dir-state structure.
+ * We don't put them on the stack since they're large.
+ */
+xfs_da_state_t *
+xfs_da_state_alloc(void)
+{
+	return kmem_zone_zalloc(xfs_da_state_zone, KM_NOFS);
+}
+
+/*
+ * Kill the altpath contents of a da-state structure.
+ */
+STATIC void
+xfs_da_state_kill_altpath(xfs_da_state_t *state)
+{
+	int	i;
+
+	for (i = 0; i < state->altpath.active; i++) {
+		if (state->altpath.blk[i].bp) {
+			if (state->altpath.blk[i].bp != state->path.blk[i].bp)
+				xfs_da_buf_done(state->altpath.blk[i].bp);
+			state->altpath.blk[i].bp = NULL;
+		}
+	}
+	state->altpath.active = 0;
+}
+
+/*
+ * Free a da-state structure.
+ */
+void
+xfs_da_state_free(xfs_da_state_t *state)
+{
+	int	i;
+
+	xfs_da_state_kill_altpath(state);
+	for (i = 0; i < state->path.active; i++) {
+		if (state->path.blk[i].bp)
+			xfs_da_buf_done(state->path.blk[i].bp);
+	}
+	if (state->extravalid && state->extrablk.bp)
+		xfs_da_buf_done(state->extrablk.bp);
+#ifdef DEBUG
+	memset((char *)state, 0, sizeof(*state));
+#endif /* DEBUG */
+	kmem_zone_free(xfs_da_state_zone, state);
+}
+
+#ifdef XFS_DABUF_DEBUG
+xfs_dabuf_t	*xfs_dabuf_global_list;
+static DEFINE_SPINLOCK(xfs_dabuf_global_lock);
+#endif
+
+/*
+ * Create a dabuf.
+ */
+/* ARGSUSED */
+STATIC xfs_dabuf_t *
+xfs_da_buf_make(int nbuf, xfs_buf_t **bps, inst_t *ra)
+{
+	xfs_buf_t	*bp;
+	xfs_dabuf_t	*dabuf;
+	int		i;
+	int		off;
+
+	if (nbuf == 1)
+		dabuf = kmem_zone_alloc(xfs_dabuf_zone, KM_NOFS);
+	else
+		dabuf = kmem_alloc(XFS_DA_BUF_SIZE(nbuf), KM_NOFS);
+	dabuf->dirty = 0;
+#ifdef XFS_DABUF_DEBUG
+	dabuf->ra = ra;
+	dabuf->target = XFS_BUF_TARGET(bps[0]);
+	dabuf->blkno = XFS_BUF_ADDR(bps[0]);
+#endif
+	if (nbuf == 1) {
+		dabuf->nbuf = 1;
+		bp = bps[0];
+		dabuf->bbcount = (short)BTOBB(XFS_BUF_COUNT(bp));
+		dabuf->data = XFS_BUF_PTR(bp);
+		dabuf->bps[0] = bp;
+	} else {
+		dabuf->nbuf = nbuf;
+		for (i = 0, dabuf->bbcount = 0; i < nbuf; i++) {
+			dabuf->bps[i] = bp = bps[i];
+			dabuf->bbcount += BTOBB(XFS_BUF_COUNT(bp));
+		}
+		dabuf->data = kmem_alloc(BBTOB(dabuf->bbcount), KM_SLEEP);
+		for (i = off = 0; i < nbuf; i++, off += XFS_BUF_COUNT(bp)) {
+			bp = bps[i];
+			memcpy((char *)dabuf->data + off, XFS_BUF_PTR(bp),
+				XFS_BUF_COUNT(bp));
+		}
+	}
+#ifdef XFS_DABUF_DEBUG
+	{
+		xfs_dabuf_t	*p;
+
+		spin_lock(&xfs_dabuf_global_lock);
+		for (p = xfs_dabuf_global_list; p; p = p->next) {
+			ASSERT(p->blkno != dabuf->blkno ||
+			       p->target != dabuf->target);
+		}
+		dabuf->prev = NULL;
+		if (xfs_dabuf_global_list)
+			xfs_dabuf_global_list->prev = dabuf;
+		dabuf->next = xfs_dabuf_global_list;
+		xfs_dabuf_global_list = dabuf;
+		spin_unlock(&xfs_dabuf_global_lock);
+	}
+#endif
+	return dabuf;
+}
+
+/*
+ * Un-dirty a dabuf.
+ */
+STATIC void
+xfs_da_buf_clean(xfs_dabuf_t *dabuf)
+{
+	xfs_buf_t	*bp;
+	int		i;
+	int		off;
+
+	if (dabuf->dirty) {
+		ASSERT(dabuf->nbuf > 1);
+		dabuf->dirty = 0;
+		for (i = off = 0; i < dabuf->nbuf;
+				i++, off += XFS_BUF_COUNT(bp)) {
+			bp = dabuf->bps[i];
+			memcpy(XFS_BUF_PTR(bp), (char *)dabuf->data + off,
+				XFS_BUF_COUNT(bp));
+		}
+	}
+}
+
+/*
+ * Release a dabuf.
+ */
+void
+xfs_da_buf_done(xfs_dabuf_t *dabuf)
+{
+	ASSERT(dabuf);
+	ASSERT(dabuf->nbuf && dabuf->data && dabuf->bbcount && dabuf->bps[0]);
+	if (dabuf->dirty)
+		xfs_da_buf_clean(dabuf);
+	if (dabuf->nbuf > 1)
+		kmem_free(dabuf->data);
+#ifdef XFS_DABUF_DEBUG
+	{
+		spin_lock(&xfs_dabuf_global_lock);
+		if (dabuf->prev)
+			dabuf->prev->next = dabuf->next;
+		else
+			xfs_dabuf_global_list = dabuf->next;
+		if (dabuf->next)
+			dabuf->next->prev = dabuf->prev;
+		spin_unlock(&xfs_dabuf_global_lock);
+	}
+	memset(dabuf, 0, XFS_DA_BUF_SIZE(dabuf->nbuf));
+#endif
+	if (dabuf->nbuf == 1)
+		kmem_zone_free(xfs_dabuf_zone, dabuf);
+	else
+		kmem_free(dabuf);
+}
+
+/*
+ * Log transaction from a dabuf.
+ */
+void
+xfs_da_log_buf(xfs_trans_t *tp, xfs_dabuf_t *dabuf, uint first, uint last)
+{
+	xfs_buf_t	*bp;
+	uint		f;
+	int		i;
+	uint		l;
+	int		off;
+
+	ASSERT(dabuf->nbuf && dabuf->data && dabuf->bbcount && dabuf->bps[0]);
+	if (dabuf->nbuf == 1) {
+		ASSERT(dabuf->data == (void *)XFS_BUF_PTR(dabuf->bps[0]));
+		xfs_trans_log_buf(tp, dabuf->bps[0], first, last);
+		return;
+	}
+	dabuf->dirty = 1;
+	ASSERT(first <= last);
+	for (i = off = 0; i < dabuf->nbuf; i++, off += XFS_BUF_COUNT(bp)) {
+		bp = dabuf->bps[i];
+		f = off;
+		l = f + XFS_BUF_COUNT(bp) - 1;
+		if (f < first)
+			f = first;
+		if (l > last)
+			l = last;
+		if (f <= l)
+			xfs_trans_log_buf(tp, bp, f - off, l - off);
+		/*
+		 * B_DONE is set by xfs_trans_log buf.
+		 * If we don't set it on a new buffer (get not read)
+		 * then if we don't put anything in the buffer it won't
+		 * be set, and at commit it it released into the cache,
+		 * and then a read will fail.
+		 */
+		else if (!(XFS_BUF_ISDONE(bp)))
+		  XFS_BUF_DONE(bp);
+	}
+	ASSERT(last < off);
+}
+
+/*
+ * Release dabuf from a transaction.
+ * Have to free up the dabuf before the buffers are released,
+ * since the synchronization on the dabuf is really the lock on the buffer.
+ */
+void
+xfs_da_brelse(xfs_trans_t *tp, xfs_dabuf_t *dabuf)
+{
+	xfs_buf_t	*bp;
+	xfs_buf_t	**bplist;
+	int		i;
+	int		nbuf;
+
+	ASSERT(dabuf->nbuf && dabuf->data && dabuf->bbcount && dabuf->bps[0]);
+	if ((nbuf = dabuf->nbuf) == 1) {
+		bplist = &bp;
+		bp = dabuf->bps[0];
+	} else {
+		bplist = kmem_alloc(nbuf * sizeof(*bplist), KM_SLEEP);
+		memcpy(bplist, dabuf->bps, nbuf * sizeof(*bplist));
+	}
+	xfs_da_buf_done(dabuf);
+	for (i = 0; i < nbuf; i++)
+		xfs_trans_brelse(tp, bplist[i]);
+	if (bplist != &bp)
+		kmem_free(bplist);
+}
+
+/*
+ * Invalidate dabuf from a transaction.
+ */
+void
+xfs_da_binval(xfs_trans_t *tp, xfs_dabuf_t *dabuf)
+{
+	xfs_buf_t	*bp;
+	xfs_buf_t	**bplist;
+	int		i;
+	int		nbuf;
+
+	ASSERT(dabuf->nbuf && dabuf->data && dabuf->bbcount && dabuf->bps[0]);
+	if ((nbuf = dabuf->nbuf) == 1) {
+		bplist = &bp;
+		bp = dabuf->bps[0];
+	} else {
+		bplist = kmem_alloc(nbuf * sizeof(*bplist), KM_SLEEP);
+		memcpy(bplist, dabuf->bps, nbuf * sizeof(*bplist));
+	}
+	xfs_da_buf_done(dabuf);
+	for (i = 0; i < nbuf; i++)
+		xfs_trans_binval(tp, bplist[i]);
+	if (bplist != &bp)
+		kmem_free(bplist);
+}
+
+/*
+ * Get the first daddr from a dabuf.
+ */
+xfs_daddr_t
+xfs_da_blkno(xfs_dabuf_t *dabuf)
+{
+	ASSERT(dabuf->nbuf);
+	ASSERT(dabuf->data);
+	return XFS_BUF_ADDR(dabuf->bps[0]);
+}
diff --git a/fs/xfs/xfs_da_btree.h b/fs/xfs/xfs_da_btree.h
new file mode 100644
index 00000000..fe9f5a8c
--- /dev/null
+++ b/fs/xfs/xfs_da_btree.h
@@ -0,0 +1,281 @@
+/*
+ * Copyright (c) 2000,2002,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_DA_BTREE_H__
+#define	__XFS_DA_BTREE_H__
+
+struct xfs_buf;
+struct xfs_bmap_free;
+struct xfs_inode;
+struct xfs_mount;
+struct xfs_trans;
+struct zone;
+
+/*========================================================================
+ * Directory Structure when greater than XFS_LBSIZE(mp) bytes.
+ *========================================================================*/
+
+/*
+ * This structure is common to both leaf nodes and non-leaf nodes in the Btree.
+ *
+ * Is is used to manage a doubly linked list of all blocks at the same
+ * level in the Btree, and to identify which type of block this is.
+ */
+#define XFS_DA_NODE_MAGIC	0xfebe	/* magic number: non-leaf blocks */
+#define XFS_ATTR_LEAF_MAGIC	0xfbee	/* magic number: attribute leaf blks */
+#define	XFS_DIR2_LEAF1_MAGIC	0xd2f1	/* magic number: v2 dirlf single blks */
+#define	XFS_DIR2_LEAFN_MAGIC	0xd2ff	/* magic number: v2 dirlf multi blks */
+
+typedef struct xfs_da_blkinfo {
+	__be32		forw;			/* previous block in list */
+	__be32		back;			/* following block in list */
+	__be16		magic;			/* validity check on block */
+	__be16		pad;			/* unused */
+} xfs_da_blkinfo_t;
+
+/*
+ * This is the structure of the root and intermediate nodes in the Btree.
+ * The leaf nodes are defined above.
+ *
+ * Entries are not packed.
+ *
+ * Since we have duplicate keys, use a binary search but always follow
+ * all match in the block, not just the first match found.
+ */
+#define	XFS_DA_NODE_MAXDEPTH	5	/* max depth of Btree */
+
+typedef struct xfs_da_intnode {
+	struct xfs_da_node_hdr {	/* constant-structure header block */
+		xfs_da_blkinfo_t info;	/* block type, links, etc. */
+		__be16	count;		/* count of active entries */
+		__be16	level;		/* level above leaves (leaf == 0) */
+	} hdr;
+	struct xfs_da_node_entry {
+		__be32	hashval;	/* hash value for this descendant */
+		__be32	before;		/* Btree block before this key */
+	} btree[1];			/* variable sized array of keys */
+} xfs_da_intnode_t;
+typedef struct xfs_da_node_hdr xfs_da_node_hdr_t;
+typedef struct xfs_da_node_entry xfs_da_node_entry_t;
+
+#define	XFS_LBSIZE(mp)	(mp)->m_sb.sb_blocksize
+
+/*========================================================================
+ * Btree searching and modification structure definitions.
+ *========================================================================*/
+
+/*
+ * Search comparison results
+ */
+enum xfs_dacmp {
+	XFS_CMP_DIFFERENT,	/* names are completely different */
+	XFS_CMP_EXACT,		/* names are exactly the same */
+	XFS_CMP_CASE		/* names are same but differ in case */
+};
+
+/*
+ * Structure to ease passing around component names.
+ */
+typedef struct xfs_da_args {
+	const __uint8_t	*name;		/* string (maybe not NULL terminated) */
+	int		namelen;	/* length of string (maybe no NULL) */
+	__uint8_t	*value;		/* set of bytes (maybe contain NULLs) */
+	int		valuelen;	/* length of value */
+	int		flags;		/* argument flags (eg: ATTR_NOCREATE) */
+	xfs_dahash_t	hashval;	/* hash value of name */
+	xfs_ino_t	inumber;	/* input/output inode number */
+	struct xfs_inode *dp;		/* directory inode to manipulate */
+	xfs_fsblock_t	*firstblock;	/* ptr to firstblock for bmap calls */
+	struct xfs_bmap_free *flist;	/* ptr to freelist for bmap_finish */
+	struct xfs_trans *trans;	/* current trans (changes over time) */
+	xfs_extlen_t	total;		/* total blocks needed, for 1st bmap */
+	int		whichfork;	/* data or attribute fork */
+	xfs_dablk_t	blkno;		/* blkno of attr leaf of interest */
+	int		index;		/* index of attr of interest in blk */
+	xfs_dablk_t	rmtblkno;	/* remote attr value starting blkno */
+	int		rmtblkcnt;	/* remote attr value block count */
+	xfs_dablk_t	blkno2;		/* blkno of 2nd attr leaf of interest */
+	int		index2;		/* index of 2nd attr in blk */
+	xfs_dablk_t	rmtblkno2;	/* remote attr value starting blkno */
+	int		rmtblkcnt2;	/* remote attr value block count */
+	int		op_flags;	/* operation flags */
+	enum xfs_dacmp	cmpresult;	/* name compare result for lookups */
+} xfs_da_args_t;
+
+/*
+ * Operation flags:
+ */
+#define XFS_DA_OP_JUSTCHECK	0x0001	/* check for ok with no space */
+#define XFS_DA_OP_RENAME	0x0002	/* this is an atomic rename op */
+#define XFS_DA_OP_ADDNAME	0x0004	/* this is an add operation */
+#define XFS_DA_OP_OKNOENT	0x0008	/* lookup/add op, ENOENT ok, else die */
+#define XFS_DA_OP_CILOOKUP	0x0010	/* lookup to return CI name if found */
+
+#define XFS_DA_OP_FLAGS \
+	{ XFS_DA_OP_JUSTCHECK,	"JUSTCHECK" }, \
+	{ XFS_DA_OP_RENAME,	"RENAME" }, \
+	{ XFS_DA_OP_ADDNAME,	"ADDNAME" }, \
+	{ XFS_DA_OP_OKNOENT,	"OKNOENT" }, \
+	{ XFS_DA_OP_CILOOKUP,	"CILOOKUP" }
+
+/*
+ * Structure to describe buffer(s) for a block.
+ * This is needed in the directory version 2 format case, when
+ * multiple non-contiguous fsblocks might be needed to cover one
+ * logical directory block.
+ * If the buffer count is 1 then the data pointer points to the
+ * same place as the b_addr field for the buffer, else to kmem_alloced memory.
+ */
+typedef struct xfs_dabuf {
+	int		nbuf;		/* number of buffer pointers present */
+	short		dirty;		/* data needs to be copied back */
+	short		bbcount;	/* how large is data in bbs */
+	void		*data;		/* pointer for buffers' data */
+#ifdef XFS_DABUF_DEBUG
+	inst_t		*ra;		/* return address of caller to make */
+	struct xfs_dabuf *next;		/* next in global chain */
+	struct xfs_dabuf *prev;		/* previous in global chain */
+	struct xfs_buftarg *target;	/* device for buffer */
+	xfs_daddr_t	blkno;		/* daddr first in bps[0] */
+#endif
+	struct xfs_buf	*bps[1];	/* actually nbuf of these */
+} xfs_dabuf_t;
+#define	XFS_DA_BUF_SIZE(n)	\
+	(sizeof(xfs_dabuf_t) + sizeof(struct xfs_buf *) * ((n) - 1))
+
+#ifdef XFS_DABUF_DEBUG
+extern xfs_dabuf_t	*xfs_dabuf_global_list;
+#endif
+
+/*
+ * Storage for holding state during Btree searches and split/join ops.
+ *
+ * Only need space for 5 intermediate nodes.  With a minimum of 62-way
+ * fanout to the Btree, we can support over 900 million directory blocks,
+ * which is slightly more than enough.
+ */
+typedef struct xfs_da_state_blk {
+	xfs_dabuf_t	*bp;		/* buffer containing block */
+	xfs_dablk_t	blkno;		/* filesystem blkno of buffer */
+	xfs_daddr_t	disk_blkno;	/* on-disk blkno (in BBs) of buffer */
+	int		index;		/* relevant index into block */
+	xfs_dahash_t	hashval;	/* last hash value in block */
+	int		magic;		/* blk's magic number, ie: blk type */
+} xfs_da_state_blk_t;
+
+typedef struct xfs_da_state_path {
+	int			active;		/* number of active levels */
+	xfs_da_state_blk_t	blk[XFS_DA_NODE_MAXDEPTH];
+} xfs_da_state_path_t;
+
+typedef struct xfs_da_state {
+	xfs_da_args_t		*args;		/* filename arguments */
+	struct xfs_mount	*mp;		/* filesystem mount point */
+	unsigned int		blocksize;	/* logical block size */
+	unsigned int		node_ents;	/* how many entries in danode */
+	xfs_da_state_path_t	path;		/* search/split paths */
+	xfs_da_state_path_t	altpath;	/* alternate path for join */
+	unsigned char		inleaf;		/* insert into 1->lf, 0->splf */
+	unsigned char		extravalid;	/* T/F: extrablk is in use */
+	unsigned char		extraafter;	/* T/F: extrablk is after new */
+	xfs_da_state_blk_t	extrablk;	/* for double-splits on leaves */
+						/* for dirv2 extrablk is data */
+} xfs_da_state_t;
+
+/*
+ * Utility macros to aid in logging changed structure fields.
+ */
+#define XFS_DA_LOGOFF(BASE, ADDR)	((char *)(ADDR) - (char *)(BASE))
+#define XFS_DA_LOGRANGE(BASE, ADDR, SIZE)	\
+		(uint)(XFS_DA_LOGOFF(BASE, ADDR)), \
+		(uint)(XFS_DA_LOGOFF(BASE, ADDR)+(SIZE)-1)
+
+/*
+ * Name ops for directory and/or attr name operations
+ */
+struct xfs_nameops {
+	xfs_dahash_t	(*hashname)(struct xfs_name *);
+	enum xfs_dacmp	(*compname)(struct xfs_da_args *,
+					const unsigned char *, int);
+};
+
+
+/*========================================================================
+ * Function prototypes.
+ *========================================================================*/
+
+/*
+ * Routines used for growing the Btree.
+ */
+int	xfs_da_node_create(xfs_da_args_t *args, xfs_dablk_t blkno, int level,
+					 xfs_dabuf_t **bpp, int whichfork);
+int	xfs_da_split(xfs_da_state_t *state);
+
+/*
+ * Routines used for shrinking the Btree.
+ */
+int	xfs_da_join(xfs_da_state_t *state);
+void	xfs_da_fixhashpath(xfs_da_state_t *state,
+					  xfs_da_state_path_t *path_to_to_fix);
+
+/*
+ * Routines used for finding things in the Btree.
+ */
+int	xfs_da_node_lookup_int(xfs_da_state_t *state, int *result);
+int	xfs_da_path_shift(xfs_da_state_t *state, xfs_da_state_path_t *path,
+					 int forward, int release, int *result);
+/*
+ * Utility routines.
+ */
+int	xfs_da_blk_link(xfs_da_state_t *state, xfs_da_state_blk_t *old_blk,
+				       xfs_da_state_blk_t *new_blk);
+
+/*
+ * Utility routines.
+ */
+int	xfs_da_grow_inode(xfs_da_args_t *args, xfs_dablk_t *new_blkno);
+int	xfs_da_get_buf(struct xfs_trans *trans, struct xfs_inode *dp,
+			      xfs_dablk_t bno, xfs_daddr_t mappedbno,
+			      xfs_dabuf_t **bp, int whichfork);
+int	xfs_da_read_buf(struct xfs_trans *trans, struct xfs_inode *dp,
+			       xfs_dablk_t bno, xfs_daddr_t mappedbno,
+			       xfs_dabuf_t **bpp, int whichfork);
+xfs_daddr_t	xfs_da_reada_buf(struct xfs_trans *trans, struct xfs_inode *dp,
+			xfs_dablk_t bno, int whichfork);
+int	xfs_da_shrink_inode(xfs_da_args_t *args, xfs_dablk_t dead_blkno,
+					  xfs_dabuf_t *dead_buf);
+
+uint xfs_da_hashname(const __uint8_t *name_string, int name_length);
+enum xfs_dacmp xfs_da_compname(struct xfs_da_args *args,
+				const unsigned char *name, int len);
+
+
+xfs_da_state_t *xfs_da_state_alloc(void);
+void xfs_da_state_free(xfs_da_state_t *state);
+
+void xfs_da_buf_done(xfs_dabuf_t *dabuf);
+void xfs_da_log_buf(struct xfs_trans *tp, xfs_dabuf_t *dabuf, uint first,
+			   uint last);
+void xfs_da_brelse(struct xfs_trans *tp, xfs_dabuf_t *dabuf);
+void xfs_da_binval(struct xfs_trans *tp, xfs_dabuf_t *dabuf);
+xfs_daddr_t xfs_da_blkno(xfs_dabuf_t *dabuf);
+
+extern struct kmem_zone *xfs_da_state_zone;
+extern struct kmem_zone *xfs_dabuf_zone;
+extern const struct xfs_nameops xfs_default_nameops;
+
+#endif	/* __XFS_DA_BTREE_H__ */
diff --git a/fs/xfs/xfs_dfrag.c b/fs/xfs/xfs_dfrag.c
new file mode 100644
index 00000000..9a84a85c
--- /dev/null
+++ b/fs/xfs/xfs_dfrag.c
@@ -0,0 +1,457 @@
+/*
+ * Copyright (c) 2000-2006 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_mount.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_inode_item.h"
+#include "xfs_bmap.h"
+#include "xfs_itable.h"
+#include "xfs_dfrag.h"
+#include "xfs_error.h"
+#include "xfs_vnodeops.h"
+#include "xfs_trace.h"
+
+
+static int xfs_swap_extents(
+	xfs_inode_t	*ip,	/* target inode */
+	xfs_inode_t	*tip,	/* tmp inode */
+	xfs_swapext_t	*sxp);
+
+/*
+ * ioctl interface for swapext
+ */
+int
+xfs_swapext(
+	xfs_swapext_t	*sxp)
+{
+	xfs_inode_t     *ip, *tip;
+	struct file	*file, *tmp_file;
+	int		error = 0;
+
+	/* Pull information for the target fd */
+	file = fget((int)sxp->sx_fdtarget);
+	if (!file) {
+		error = XFS_ERROR(EINVAL);
+		goto out;
+	}
+
+	if (!(file->f_mode & FMODE_WRITE) ||
+	    !(file->f_mode & FMODE_READ) ||
+	    (file->f_flags & O_APPEND)) {
+		error = XFS_ERROR(EBADF);
+		goto out_put_file;
+	}
+
+	tmp_file = fget((int)sxp->sx_fdtmp);
+	if (!tmp_file) {
+		error = XFS_ERROR(EINVAL);
+		goto out_put_file;
+	}
+
+	if (!(tmp_file->f_mode & FMODE_WRITE) ||
+	    !(tmp_file->f_mode & FMODE_READ) ||
+	    (tmp_file->f_flags & O_APPEND)) {
+		error = XFS_ERROR(EBADF);
+		goto out_put_tmp_file;
+	}
+
+	if (IS_SWAPFILE(file->f_path.dentry->d_inode) ||
+	    IS_SWAPFILE(tmp_file->f_path.dentry->d_inode)) {
+		error = XFS_ERROR(EINVAL);
+		goto out_put_tmp_file;
+	}
+
+	ip = XFS_I(file->f_path.dentry->d_inode);
+	tip = XFS_I(tmp_file->f_path.dentry->d_inode);
+
+	if (ip->i_mount != tip->i_mount) {
+		error = XFS_ERROR(EINVAL);
+		goto out_put_tmp_file;
+	}
+
+	if (ip->i_ino == tip->i_ino) {
+		error = XFS_ERROR(EINVAL);
+		goto out_put_tmp_file;
+	}
+
+	if (XFS_FORCED_SHUTDOWN(ip->i_mount)) {
+		error = XFS_ERROR(EIO);
+		goto out_put_tmp_file;
+	}
+
+	error = xfs_swap_extents(ip, tip, sxp);
+
+ out_put_tmp_file:
+	fput(tmp_file);
+ out_put_file:
+	fput(file);
+ out:
+	return error;
+}
+
+/*
+ * We need to check that the format of the data fork in the temporary inode is
+ * valid for the target inode before doing the swap. This is not a problem with
+ * attr1 because of the fixed fork offset, but attr2 has a dynamically sized
+ * data fork depending on the space the attribute fork is taking so we can get
+ * invalid formats on the target inode.
+ *
+ * E.g. target has space for 7 extents in extent format, temp inode only has
+ * space for 6.  If we defragment down to 7 extents, then the tmp format is a
+ * btree, but when swapped it needs to be in extent format. Hence we can't just
+ * blindly swap data forks on attr2 filesystems.
+ *
+ * Note that we check the swap in both directions so that we don't end up with
+ * a corrupt temporary inode, either.
+ *
+ * Note that fixing the way xfs_fsr sets up the attribute fork in the source
+ * inode will prevent this situation from occurring, so all we do here is
+ * reject and log the attempt. basically we are putting the responsibility on
+ * userspace to get this right.
+ */
+static int
+xfs_swap_extents_check_format(
+	xfs_inode_t	*ip,	/* target inode */
+	xfs_inode_t	*tip)	/* tmp inode */
+{
+
+	/* Should never get a local format */
+	if (ip->i_d.di_format == XFS_DINODE_FMT_LOCAL ||
+	    tip->i_d.di_format == XFS_DINODE_FMT_LOCAL)
+		return EINVAL;
+
+	/*
+	 * if the target inode has less extents that then temporary inode then
+	 * why did userspace call us?
+	 */
+	if (ip->i_d.di_nextents < tip->i_d.di_nextents)
+		return EINVAL;
+
+	/*
+	 * if the target inode is in extent form and the temp inode is in btree
+	 * form then we will end up with the target inode in the wrong format
+	 * as we already know there are less extents in the temp inode.
+	 */
+	if (ip->i_d.di_format == XFS_DINODE_FMT_EXTENTS &&
+	    tip->i_d.di_format == XFS_DINODE_FMT_BTREE)
+		return EINVAL;
+
+	/* Check temp in extent form to max in target */
+	if (tip->i_d.di_format == XFS_DINODE_FMT_EXTENTS &&
+	    XFS_IFORK_NEXTENTS(tip, XFS_DATA_FORK) > ip->i_df.if_ext_max)
+		return EINVAL;
+
+	/* Check target in extent form to max in temp */
+	if (ip->i_d.di_format == XFS_DINODE_FMT_EXTENTS &&
+	    XFS_IFORK_NEXTENTS(ip, XFS_DATA_FORK) > tip->i_df.if_ext_max)
+		return EINVAL;
+
+	/*
+	 * If we are in a btree format, check that the temp root block will fit
+	 * in the target and that it has enough extents to be in btree format
+	 * in the target.
+	 *
+	 * Note that we have to be careful to allow btree->extent conversions
+	 * (a common defrag case) which will occur when the temp inode is in
+	 * extent format...
+	 */
+	if (tip->i_d.di_format == XFS_DINODE_FMT_BTREE &&
+	    ((XFS_IFORK_BOFF(ip) &&
+	      tip->i_df.if_broot_bytes > XFS_IFORK_BOFF(ip)) ||
+	     XFS_IFORK_NEXTENTS(tip, XFS_DATA_FORK) <= ip->i_df.if_ext_max))
+		return EINVAL;
+
+	/* Reciprocal target->temp btree format checks */
+	if (ip->i_d.di_format == XFS_DINODE_FMT_BTREE &&
+	    ((XFS_IFORK_BOFF(tip) &&
+	      ip->i_df.if_broot_bytes > XFS_IFORK_BOFF(tip)) ||
+	     XFS_IFORK_NEXTENTS(ip, XFS_DATA_FORK) <= tip->i_df.if_ext_max))
+		return EINVAL;
+
+	return 0;
+}
+
+static int
+xfs_swap_extents(
+	xfs_inode_t	*ip,	/* target inode */
+	xfs_inode_t	*tip,	/* tmp inode */
+	xfs_swapext_t	*sxp)
+{
+	xfs_mount_t	*mp = ip->i_mount;
+	xfs_trans_t	*tp;
+	xfs_bstat_t	*sbp = &sxp->sx_stat;
+	xfs_ifork_t	*tempifp, *ifp, *tifp;
+	int		ilf_fields, tilf_fields;
+	int		error = 0;
+	int		aforkblks = 0;
+	int		taforkblks = 0;
+	__uint64_t	tmp;
+
+	tempifp = kmem_alloc(sizeof(xfs_ifork_t), KM_MAYFAIL);
+	if (!tempifp) {
+		error = XFS_ERROR(ENOMEM);
+		goto out;
+	}
+
+	/*
+	 * we have to do two separate lock calls here to keep lockdep
+	 * happy. If we try to get all the locks in one call, lock will
+	 * report false positives when we drop the ILOCK and regain them
+	 * below.
+	 */
+	xfs_lock_two_inodes(ip, tip, XFS_IOLOCK_EXCL);
+	xfs_lock_two_inodes(ip, tip, XFS_ILOCK_EXCL);
+
+	/* Verify that both files have the same format */
+	if ((ip->i_d.di_mode & S_IFMT) != (tip->i_d.di_mode & S_IFMT)) {
+		error = XFS_ERROR(EINVAL);
+		goto out_unlock;
+	}
+
+	/* Verify both files are either real-time or non-realtime */
+	if (XFS_IS_REALTIME_INODE(ip) != XFS_IS_REALTIME_INODE(tip)) {
+		error = XFS_ERROR(EINVAL);
+		goto out_unlock;
+	}
+
+	if (VN_CACHED(VFS_I(tip)) != 0) {
+		error = xfs_flushinval_pages(tip, 0, -1,
+				FI_REMAPF_LOCKED);
+		if (error)
+			goto out_unlock;
+	}
+
+	/* Verify O_DIRECT for ftmp */
+	if (VN_CACHED(VFS_I(tip)) != 0) {
+		error = XFS_ERROR(EINVAL);
+		goto out_unlock;
+	}
+
+	/* Verify all data are being swapped */
+	if (sxp->sx_offset != 0 ||
+	    sxp->sx_length != ip->i_d.di_size ||
+	    sxp->sx_length != tip->i_d.di_size) {
+		error = XFS_ERROR(EFAULT);
+		goto out_unlock;
+	}
+
+	trace_xfs_swap_extent_before(ip, 0);
+	trace_xfs_swap_extent_before(tip, 1);
+
+	/* check inode formats now that data is flushed */
+	error = xfs_swap_extents_check_format(ip, tip);
+	if (error) {
+		xfs_notice(mp,
+		    "%s: inode 0x%llx format is incompatible for exchanging.",
+				__func__, ip->i_ino);
+		goto out_unlock;
+	}
+
+	/*
+	 * Compare the current change & modify times with that
+	 * passed in.  If they differ, we abort this swap.
+	 * This is the mechanism used to ensure the calling
+	 * process that the file was not changed out from
+	 * under it.
+	 */
+	if ((sbp->bs_ctime.tv_sec != VFS_I(ip)->i_ctime.tv_sec) ||
+	    (sbp->bs_ctime.tv_nsec != VFS_I(ip)->i_ctime.tv_nsec) ||
+	    (sbp->bs_mtime.tv_sec != VFS_I(ip)->i_mtime.tv_sec) ||
+	    (sbp->bs_mtime.tv_nsec != VFS_I(ip)->i_mtime.tv_nsec)) {
+		error = XFS_ERROR(EBUSY);
+		goto out_unlock;
+	}
+
+	/* We need to fail if the file is memory mapped.  Once we have tossed
+	 * all existing pages, the page fault will have no option
+	 * but to go to the filesystem for pages. By making the page fault call
+	 * vop_read (or write in the case of autogrow) they block on the iolock
+	 * until we have switched the extents.
+	 */
+	if (VN_MAPPED(VFS_I(ip))) {
+		error = XFS_ERROR(EBUSY);
+		goto out_unlock;
+	}
+
+	xfs_iunlock(ip, XFS_ILOCK_EXCL);
+	xfs_iunlock(tip, XFS_ILOCK_EXCL);
+
+	/*
+	 * There is a race condition here since we gave up the
+	 * ilock.  However, the data fork will not change since
+	 * we have the iolock (locked for truncation too) so we
+	 * are safe.  We don't really care if non-io related
+	 * fields change.
+	 */
+
+	xfs_tosspages(ip, 0, -1, FI_REMAPF);
+
+	tp = xfs_trans_alloc(mp, XFS_TRANS_SWAPEXT);
+	if ((error = xfs_trans_reserve(tp, 0,
+				     XFS_ICHANGE_LOG_RES(mp), 0,
+				     0, 0))) {
+		xfs_iunlock(ip,  XFS_IOLOCK_EXCL);
+		xfs_iunlock(tip, XFS_IOLOCK_EXCL);
+		xfs_trans_cancel(tp, 0);
+		goto out;
+	}
+	xfs_lock_two_inodes(ip, tip, XFS_ILOCK_EXCL);
+
+	/*
+	 * Count the number of extended attribute blocks
+	 */
+	if ( ((XFS_IFORK_Q(ip) != 0) && (ip->i_d.di_anextents > 0)) &&
+	     (ip->i_d.di_aformat != XFS_DINODE_FMT_LOCAL)) {
+		error = xfs_bmap_count_blocks(tp, ip, XFS_ATTR_FORK, &aforkblks);
+		if (error)
+			goto out_trans_cancel;
+	}
+	if ( ((XFS_IFORK_Q(tip) != 0) && (tip->i_d.di_anextents > 0)) &&
+	     (tip->i_d.di_aformat != XFS_DINODE_FMT_LOCAL)) {
+		error = xfs_bmap_count_blocks(tp, tip, XFS_ATTR_FORK,
+			&taforkblks);
+		if (error)
+			goto out_trans_cancel;
+	}
+
+	/*
+	 * Swap the data forks of the inodes
+	 */
+	ifp = &ip->i_df;
+	tifp = &tip->i_df;
+	*tempifp = *ifp;	/* struct copy */
+	*ifp = *tifp;		/* struct copy */
+	*tifp = *tempifp;	/* struct copy */
+
+	/*
+	 * Fix the in-memory data fork values that are dependent on the fork
+	 * offset in the inode. We can't assume they remain the same as attr2
+	 * has dynamic fork offsets.
+	 */
+	ifp->if_ext_max = XFS_IFORK_SIZE(ip, XFS_DATA_FORK) /
+					(uint)sizeof(xfs_bmbt_rec_t);
+	tifp->if_ext_max = XFS_IFORK_SIZE(tip, XFS_DATA_FORK) /
+					(uint)sizeof(xfs_bmbt_rec_t);
+
+	/*
+	 * Fix the on-disk inode values
+	 */
+	tmp = (__uint64_t)ip->i_d.di_nblocks;
+	ip->i_d.di_nblocks = tip->i_d.di_nblocks - taforkblks + aforkblks;
+	tip->i_d.di_nblocks = tmp + taforkblks - aforkblks;
+
+	tmp = (__uint64_t) ip->i_d.di_nextents;
+	ip->i_d.di_nextents = tip->i_d.di_nextents;
+	tip->i_d.di_nextents = tmp;
+
+	tmp = (__uint64_t) ip->i_d.di_format;
+	ip->i_d.di_format = tip->i_d.di_format;
+	tip->i_d.di_format = tmp;
+
+	/*
+	 * The extents in the source inode could still contain speculative
+	 * preallocation beyond EOF (e.g. the file is open but not modified
+	 * while defrag is in progress). In that case, we need to copy over the
+	 * number of delalloc blocks the data fork in the source inode is
+	 * tracking beyond EOF so that when the fork is truncated away when the
+	 * temporary inode is unlinked we don't underrun the i_delayed_blks
+	 * counter on that inode.
+	 */
+	ASSERT(tip->i_delayed_blks == 0);
+	tip->i_delayed_blks = ip->i_delayed_blks;
+	ip->i_delayed_blks = 0;
+
+	ilf_fields = XFS_ILOG_CORE;
+
+	switch(ip->i_d.di_format) {
+	case XFS_DINODE_FMT_EXTENTS:
+		/* If the extents fit in the inode, fix the
+		 * pointer.  Otherwise it's already NULL or
+		 * pointing to the extent.
+		 */
+		if (ip->i_d.di_nextents <= XFS_INLINE_EXTS) {
+			ifp->if_u1.if_extents =
+				ifp->if_u2.if_inline_ext;
+		}
+		ilf_fields |= XFS_ILOG_DEXT;
+		break;
+	case XFS_DINODE_FMT_BTREE:
+		ilf_fields |= XFS_ILOG_DBROOT;
+		break;
+	}
+
+	tilf_fields = XFS_ILOG_CORE;
+
+	switch(tip->i_d.di_format) {
+	case XFS_DINODE_FMT_EXTENTS:
+		/* If the extents fit in the inode, fix the
+		 * pointer.  Otherwise it's already NULL or
+		 * pointing to the extent.
+		 */
+		if (tip->i_d.di_nextents <= XFS_INLINE_EXTS) {
+			tifp->if_u1.if_extents =
+				tifp->if_u2.if_inline_ext;
+		}
+		tilf_fields |= XFS_ILOG_DEXT;
+		break;
+	case XFS_DINODE_FMT_BTREE:
+		tilf_fields |= XFS_ILOG_DBROOT;
+		break;
+	}
+
+
+	xfs_trans_ijoin_ref(tp, ip, XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL);
+	xfs_trans_ijoin_ref(tp, tip, XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL);
+
+	xfs_trans_log_inode(tp, ip,  ilf_fields);
+	xfs_trans_log_inode(tp, tip, tilf_fields);
+
+	/*
+	 * If this is a synchronous mount, make sure that the
+	 * transaction goes to disk before returning to the user.
+	 */
+	if (mp->m_flags & XFS_MOUNT_WSYNC)
+		xfs_trans_set_sync(tp);
+
+	error = xfs_trans_commit(tp, XFS_TRANS_SWAPEXT);
+
+	trace_xfs_swap_extent_after(ip, 0);
+	trace_xfs_swap_extent_after(tip, 1);
+out:
+	kmem_free(tempifp);
+	return error;
+
+out_unlock:
+	xfs_iunlock(ip,  XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL);
+	xfs_iunlock(tip, XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL);
+	goto out;
+
+out_trans_cancel:
+	xfs_trans_cancel(tp, 0);
+	goto out_unlock;
+}
diff --git a/fs/xfs/xfs_dfrag.h b/fs/xfs/xfs_dfrag.h
new file mode 100644
index 00000000..20bdd935
--- /dev/null
+++ b/fs/xfs/xfs_dfrag.h
@@ -0,0 +1,53 @@
+/*
+ * Copyright (c) 2000,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_DFRAG_H__
+#define	__XFS_DFRAG_H__
+
+/*
+ * Structure passed to xfs_swapext
+ */
+
+typedef struct xfs_swapext
+{
+	__int64_t	sx_version;	/* version */
+	__int64_t	sx_fdtarget;	/* fd of target file */
+	__int64_t	sx_fdtmp;	/* fd of tmp file */
+	xfs_off_t	sx_offset;	/* offset into file */
+	xfs_off_t	sx_length;	/* leng from offset */
+	char		sx_pad[16];	/* pad space, unused */
+	xfs_bstat_t	sx_stat;	/* stat of target b4 copy */
+} xfs_swapext_t;
+
+/*
+ * Version flag
+ */
+#define XFS_SX_VERSION		0
+
+#ifdef __KERNEL__
+/*
+ * Prototypes for visible xfs_dfrag.c routines.
+ */
+
+/*
+ * Syscall interface for xfs_swapext
+ */
+int	xfs_swapext(struct xfs_swapext *sx);
+
+#endif	/* __KERNEL__ */
+
+#endif	/* __XFS_DFRAG_H__ */
diff --git a/fs/xfs/xfs_dinode.h b/fs/xfs/xfs_dinode.h
new file mode 100644
index 00000000..dffba9ba
--- /dev/null
+++ b/fs/xfs/xfs_dinode.h
@@ -0,0 +1,216 @@
+/*
+ * Copyright (c) 2000,2002,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_DINODE_H__
+#define	__XFS_DINODE_H__
+
+#define	XFS_DINODE_MAGIC		0x494e	/* 'IN' */
+#define XFS_DINODE_GOOD_VERSION(v)	(((v) == 1 || (v) == 2))
+
+typedef struct xfs_timestamp {
+	__be32		t_sec;		/* timestamp seconds */
+	__be32		t_nsec;		/* timestamp nanoseconds */
+} xfs_timestamp_t;
+
+/*
+ * On-disk inode structure.
+ *
+ * This is just the header or "dinode core", the inode is expanded to fill a
+ * variable size the leftover area split into a data and an attribute fork.
+ * The format of the data and attribute fork depends on the format of the
+ * inode as indicated by di_format and di_aformat.  To access the data and
+ * attribute use the XFS_DFORK_PTR, XFS_DFORK_DPTR, and XFS_DFORK_PTR macros
+ * below.
+ *
+ * There is a very similar struct icdinode in xfs_inode which matches the
+ * layout of the first 96 bytes of this structure, but is kept in native
+ * format instead of big endian.
+ */
+typedef struct xfs_dinode {
+	__be16		di_magic;	/* inode magic # = XFS_DINODE_MAGIC */
+	__be16		di_mode;	/* mode and type of file */
+	__u8		di_version;	/* inode version */
+	__u8		di_format;	/* format of di_c data */
+	__be16		di_onlink;	/* old number of links to file */
+	__be32		di_uid;		/* owner's user id */
+	__be32		di_gid;		/* owner's group id */
+	__be32		di_nlink;	/* number of links to file */
+	__be16		di_projid_lo;	/* lower part of owner's project id */
+	__be16		di_projid_hi;	/* higher part owner's project id */
+	__u8		di_pad[6];	/* unused, zeroed space */
+	__be16		di_flushiter;	/* incremented on flush */
+	xfs_timestamp_t	di_atime;	/* time last accessed */
+	xfs_timestamp_t	di_mtime;	/* time last modified */
+	xfs_timestamp_t	di_ctime;	/* time created/inode modified */
+	__be64		di_size;	/* number of bytes in file */
+	__be64		di_nblocks;	/* # of direct & btree blocks used */
+	__be32		di_extsize;	/* basic/minimum extent size for file */
+	__be32		di_nextents;	/* number of extents in data fork */
+	__be16		di_anextents;	/* number of extents in attribute fork*/
+	__u8		di_forkoff;	/* attr fork offs, <<3 for 64b align */
+	__s8		di_aformat;	/* format of attr fork's data */
+	__be32		di_dmevmask;	/* DMIG event mask */
+	__be16		di_dmstate;	/* DMIG state info */
+	__be16		di_flags;	/* random flags, XFS_DIFLAG_... */
+	__be32		di_gen;		/* generation number */
+
+	/* di_next_unlinked is the only non-core field in the old dinode */
+	__be32		di_next_unlinked;/* agi unlinked list ptr */
+} __attribute__((packed)) xfs_dinode_t;
+
+#define DI_MAX_FLUSH 0xffff
+
+/*
+ * The 32 bit link count in the inode theoretically maxes out at UINT_MAX.
+ * Since the pathconf interface is signed, we use 2^31 - 1 instead.
+ * The old inode format had a 16 bit link count, so its maximum is USHRT_MAX.
+ */
+#define	XFS_MAXLINK		((1U << 31) - 1U)
+#define	XFS_MAXLINK_1		65535U
+
+/*
+ * Values for di_format
+ */
+typedef enum xfs_dinode_fmt {
+	XFS_DINODE_FMT_DEV,		/* xfs_dev_t */
+	XFS_DINODE_FMT_LOCAL,		/* bulk data */
+	XFS_DINODE_FMT_EXTENTS,		/* struct xfs_bmbt_rec */
+	XFS_DINODE_FMT_BTREE,		/* struct xfs_bmdr_block */
+	XFS_DINODE_FMT_UUID		/* uuid_t */
+} xfs_dinode_fmt_t;
+
+/*
+ * Inode minimum and maximum sizes.
+ */
+#define	XFS_DINODE_MIN_LOG	8
+#define	XFS_DINODE_MAX_LOG	11
+#define	XFS_DINODE_MIN_SIZE	(1 << XFS_DINODE_MIN_LOG)
+#define	XFS_DINODE_MAX_SIZE	(1 << XFS_DINODE_MAX_LOG)
+
+/*
+ * Inode size for given fs.
+ */
+#define XFS_LITINO(mp) \
+	((int)(((mp)->m_sb.sb_inodesize) - sizeof(struct xfs_dinode)))
+
+#define	XFS_BROOT_SIZE_ADJ	\
+	(XFS_BTREE_LBLOCK_LEN - sizeof(xfs_bmdr_block_t))
+
+/*
+ * Inode data & attribute fork sizes, per inode.
+ */
+#define XFS_DFORK_Q(dip)		((dip)->di_forkoff != 0)
+#define XFS_DFORK_BOFF(dip)		((int)((dip)->di_forkoff << 3))
+
+#define XFS_DFORK_DSIZE(dip,mp) \
+	(XFS_DFORK_Q(dip) ? \
+		XFS_DFORK_BOFF(dip) : \
+		XFS_LITINO(mp))
+#define XFS_DFORK_ASIZE(dip,mp) \
+	(XFS_DFORK_Q(dip) ? \
+		XFS_LITINO(mp) - XFS_DFORK_BOFF(dip) : \
+		0)
+#define XFS_DFORK_SIZE(dip,mp,w) \
+	((w) == XFS_DATA_FORK ? \
+		XFS_DFORK_DSIZE(dip, mp) : \
+		XFS_DFORK_ASIZE(dip, mp))
+
+/*
+ * Return pointers to the data or attribute forks.
+ */
+#define XFS_DFORK_DPTR(dip) \
+	((char *)(dip) + sizeof(struct xfs_dinode))
+#define XFS_DFORK_APTR(dip)	\
+	(XFS_DFORK_DPTR(dip) + XFS_DFORK_BOFF(dip))
+#define XFS_DFORK_PTR(dip,w)	\
+	((w) == XFS_DATA_FORK ? XFS_DFORK_DPTR(dip) : XFS_DFORK_APTR(dip))
+
+#define XFS_DFORK_FORMAT(dip,w) \
+	((w) == XFS_DATA_FORK ? \
+		(dip)->di_format : \
+		(dip)->di_aformat)
+#define XFS_DFORK_NEXTENTS(dip,w) \
+	((w) == XFS_DATA_FORK ? \
+		be32_to_cpu((dip)->di_nextents) : \
+		be16_to_cpu((dip)->di_anextents))
+
+#define	XFS_BUF_TO_DINODE(bp)	((xfs_dinode_t *)XFS_BUF_PTR(bp))
+
+/*
+ * For block and character special files the 32bit dev_t is stored at the
+ * beginning of the data fork.
+ */
+static inline xfs_dev_t xfs_dinode_get_rdev(struct xfs_dinode *dip)
+{
+	return be32_to_cpu(*(__be32 *)XFS_DFORK_DPTR(dip));
+}
+
+static inline void xfs_dinode_put_rdev(struct xfs_dinode *dip, xfs_dev_t rdev)
+{
+	*(__be32 *)XFS_DFORK_DPTR(dip) = cpu_to_be32(rdev);
+}
+
+/*
+ * Values for di_flags
+ * There should be a one-to-one correspondence between these flags and the
+ * XFS_XFLAG_s.
+ */
+#define XFS_DIFLAG_REALTIME_BIT  0	/* file's blocks come from rt area */
+#define XFS_DIFLAG_PREALLOC_BIT  1	/* file space has been preallocated */
+#define XFS_DIFLAG_NEWRTBM_BIT   2	/* for rtbitmap inode, new format */
+#define XFS_DIFLAG_IMMUTABLE_BIT 3	/* inode is immutable */
+#define XFS_DIFLAG_APPEND_BIT    4	/* inode is append-only */
+#define XFS_DIFLAG_SYNC_BIT      5	/* inode is written synchronously */
+#define XFS_DIFLAG_NOATIME_BIT   6	/* do not update atime */
+#define XFS_DIFLAG_NODUMP_BIT    7	/* do not dump */
+#define XFS_DIFLAG_RTINHERIT_BIT 8	/* create with realtime bit set */
+#define XFS_DIFLAG_PROJINHERIT_BIT   9	/* create with parents projid */
+#define XFS_DIFLAG_NOSYMLINKS_BIT   10	/* disallow symlink creation */
+#define XFS_DIFLAG_EXTSIZE_BIT      11	/* inode extent size allocator hint */
+#define XFS_DIFLAG_EXTSZINHERIT_BIT 12	/* inherit inode extent size */
+#define XFS_DIFLAG_NODEFRAG_BIT     13	/* do not reorganize/defragment */
+#define XFS_DIFLAG_FILESTREAM_BIT   14  /* use filestream allocator */
+#define XFS_DIFLAG_REALTIME      (1 << XFS_DIFLAG_REALTIME_BIT)
+#define XFS_DIFLAG_PREALLOC      (1 << XFS_DIFLAG_PREALLOC_BIT)
+#define XFS_DIFLAG_NEWRTBM       (1 << XFS_DIFLAG_NEWRTBM_BIT)
+#define XFS_DIFLAG_IMMUTABLE     (1 << XFS_DIFLAG_IMMUTABLE_BIT)
+#define XFS_DIFLAG_APPEND        (1 << XFS_DIFLAG_APPEND_BIT)
+#define XFS_DIFLAG_SYNC          (1 << XFS_DIFLAG_SYNC_BIT)
+#define XFS_DIFLAG_NOATIME       (1 << XFS_DIFLAG_NOATIME_BIT)
+#define XFS_DIFLAG_NODUMP        (1 << XFS_DIFLAG_NODUMP_BIT)
+#define XFS_DIFLAG_RTINHERIT     (1 << XFS_DIFLAG_RTINHERIT_BIT)
+#define XFS_DIFLAG_PROJINHERIT   (1 << XFS_DIFLAG_PROJINHERIT_BIT)
+#define XFS_DIFLAG_NOSYMLINKS    (1 << XFS_DIFLAG_NOSYMLINKS_BIT)
+#define XFS_DIFLAG_EXTSIZE       (1 << XFS_DIFLAG_EXTSIZE_BIT)
+#define XFS_DIFLAG_EXTSZINHERIT  (1 << XFS_DIFLAG_EXTSZINHERIT_BIT)
+#define XFS_DIFLAG_NODEFRAG      (1 << XFS_DIFLAG_NODEFRAG_BIT)
+#define XFS_DIFLAG_FILESTREAM    (1 << XFS_DIFLAG_FILESTREAM_BIT)
+
+#ifdef CONFIG_XFS_RT
+#define XFS_IS_REALTIME_INODE(ip) ((ip)->i_d.di_flags & XFS_DIFLAG_REALTIME)
+#else
+#define XFS_IS_REALTIME_INODE(ip) (0)
+#endif
+
+#define XFS_DIFLAG_ANY \
+	(XFS_DIFLAG_REALTIME | XFS_DIFLAG_PREALLOC | XFS_DIFLAG_NEWRTBM | \
+	 XFS_DIFLAG_IMMUTABLE | XFS_DIFLAG_APPEND | XFS_DIFLAG_SYNC | \
+	 XFS_DIFLAG_NOATIME | XFS_DIFLAG_NODUMP | XFS_DIFLAG_RTINHERIT | \
+	 XFS_DIFLAG_PROJINHERIT | XFS_DIFLAG_NOSYMLINKS | XFS_DIFLAG_EXTSIZE | \
+	 XFS_DIFLAG_EXTSZINHERIT | XFS_DIFLAG_NODEFRAG | XFS_DIFLAG_FILESTREAM)
+
+#endif	/* __XFS_DINODE_H__ */
diff --git a/fs/xfs/xfs_dir2.c b/fs/xfs/xfs_dir2.c
new file mode 100644
index 00000000..dba7a71c
--- /dev/null
+++ b/fs/xfs/xfs_dir2.c
@@ -0,0 +1,764 @@
+/*
+ * Copyright (c) 2000-2001,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_dir2.h"
+#include "xfs_mount.h"
+#include "xfs_da_btree.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_alloc_btree.h"
+#include "xfs_dir2_sf.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_inode_item.h"
+#include "xfs_bmap.h"
+#include "xfs_dir2_data.h"
+#include "xfs_dir2_leaf.h"
+#include "xfs_dir2_block.h"
+#include "xfs_dir2_node.h"
+#include "xfs_error.h"
+#include "xfs_vnodeops.h"
+#include "xfs_trace.h"
+
+struct xfs_name xfs_name_dotdot = { (unsigned char *)"..", 2};
+
+/*
+ * ASCII case-insensitive (ie. A-Z) support for directories that was
+ * used in IRIX.
+ */
+STATIC xfs_dahash_t
+xfs_ascii_ci_hashname(
+	struct xfs_name	*name)
+{
+	xfs_dahash_t	hash;
+	int		i;
+
+	for (i = 0, hash = 0; i < name->len; i++)
+		hash = tolower(name->name[i]) ^ rol32(hash, 7);
+
+	return hash;
+}
+
+STATIC enum xfs_dacmp
+xfs_ascii_ci_compname(
+	struct xfs_da_args *args,
+	const unsigned char *name,
+	int		len)
+{
+	enum xfs_dacmp	result;
+	int		i;
+
+	if (args->namelen != len)
+		return XFS_CMP_DIFFERENT;
+
+	result = XFS_CMP_EXACT;
+	for (i = 0; i < len; i++) {
+		if (args->name[i] == name[i])
+			continue;
+		if (tolower(args->name[i]) != tolower(name[i]))
+			return XFS_CMP_DIFFERENT;
+		result = XFS_CMP_CASE;
+	}
+
+	return result;
+}
+
+static struct xfs_nameops xfs_ascii_ci_nameops = {
+	.hashname	= xfs_ascii_ci_hashname,
+	.compname	= xfs_ascii_ci_compname,
+};
+
+void
+xfs_dir_mount(
+	xfs_mount_t	*mp)
+{
+	ASSERT(xfs_sb_version_hasdirv2(&mp->m_sb));
+	ASSERT((1 << (mp->m_sb.sb_blocklog + mp->m_sb.sb_dirblklog)) <=
+	       XFS_MAX_BLOCKSIZE);
+	mp->m_dirblksize = 1 << (mp->m_sb.sb_blocklog + mp->m_sb.sb_dirblklog);
+	mp->m_dirblkfsbs = 1 << mp->m_sb.sb_dirblklog;
+	mp->m_dirdatablk = xfs_dir2_db_to_da(mp, XFS_DIR2_DATA_FIRSTDB(mp));
+	mp->m_dirleafblk = xfs_dir2_db_to_da(mp, XFS_DIR2_LEAF_FIRSTDB(mp));
+	mp->m_dirfreeblk = xfs_dir2_db_to_da(mp, XFS_DIR2_FREE_FIRSTDB(mp));
+	mp->m_attr_node_ents =
+		(mp->m_sb.sb_blocksize - (uint)sizeof(xfs_da_node_hdr_t)) /
+		(uint)sizeof(xfs_da_node_entry_t);
+	mp->m_dir_node_ents =
+		(mp->m_dirblksize - (uint)sizeof(xfs_da_node_hdr_t)) /
+		(uint)sizeof(xfs_da_node_entry_t);
+	mp->m_dir_magicpct = (mp->m_dirblksize * 37) / 100;
+	if (xfs_sb_version_hasasciici(&mp->m_sb))
+		mp->m_dirnameops = &xfs_ascii_ci_nameops;
+	else
+		mp->m_dirnameops = &xfs_default_nameops;
+}
+
+/*
+ * Return 1 if directory contains only "." and "..".
+ */
+int
+xfs_dir_isempty(
+	xfs_inode_t	*dp)
+{
+	xfs_dir2_sf_t	*sfp;
+
+	ASSERT((dp->i_d.di_mode & S_IFMT) == S_IFDIR);
+	if (dp->i_d.di_size == 0)	/* might happen during shutdown. */
+		return 1;
+	if (dp->i_d.di_size > XFS_IFORK_DSIZE(dp))
+		return 0;
+	sfp = (xfs_dir2_sf_t *)dp->i_df.if_u1.if_data;
+	return !sfp->hdr.count;
+}
+
+/*
+ * Validate a given inode number.
+ */
+int
+xfs_dir_ino_validate(
+	xfs_mount_t	*mp,
+	xfs_ino_t	ino)
+{
+	xfs_agblock_t	agblkno;
+	xfs_agino_t	agino;
+	xfs_agnumber_t	agno;
+	int		ino_ok;
+	int		ioff;
+
+	agno = XFS_INO_TO_AGNO(mp, ino);
+	agblkno = XFS_INO_TO_AGBNO(mp, ino);
+	ioff = XFS_INO_TO_OFFSET(mp, ino);
+	agino = XFS_OFFBNO_TO_AGINO(mp, agblkno, ioff);
+	ino_ok =
+		agno < mp->m_sb.sb_agcount &&
+		agblkno < mp->m_sb.sb_agblocks &&
+		agblkno != 0 &&
+		ioff < (1 << mp->m_sb.sb_inopblog) &&
+		XFS_AGINO_TO_INO(mp, agno, agino) == ino;
+	if (unlikely(XFS_TEST_ERROR(!ino_ok, mp, XFS_ERRTAG_DIR_INO_VALIDATE,
+			XFS_RANDOM_DIR_INO_VALIDATE))) {
+		xfs_warn(mp, "Invalid inode number 0x%Lx",
+				(unsigned long long) ino);
+		XFS_ERROR_REPORT("xfs_dir_ino_validate", XFS_ERRLEVEL_LOW, mp);
+		return XFS_ERROR(EFSCORRUPTED);
+	}
+	return 0;
+}
+
+/*
+ * Initialize a directory with its "." and ".." entries.
+ */
+int
+xfs_dir_init(
+	xfs_trans_t	*tp,
+	xfs_inode_t	*dp,
+	xfs_inode_t	*pdp)
+{
+	xfs_da_args_t	args;
+	int		error;
+
+	memset((char *)&args, 0, sizeof(args));
+	args.dp = dp;
+	args.trans = tp;
+	ASSERT((dp->i_d.di_mode & S_IFMT) == S_IFDIR);
+	if ((error = xfs_dir_ino_validate(tp->t_mountp, pdp->i_ino)))
+		return error;
+	return xfs_dir2_sf_create(&args, pdp->i_ino);
+}
+
+/*
+  Enter a name in a directory.
+ */
+int
+xfs_dir_createname(
+	xfs_trans_t		*tp,
+	xfs_inode_t		*dp,
+	struct xfs_name		*name,
+	xfs_ino_t		inum,		/* new entry inode number */
+	xfs_fsblock_t		*first,		/* bmap's firstblock */
+	xfs_bmap_free_t		*flist,		/* bmap's freeblock list */
+	xfs_extlen_t		total)		/* bmap's total block count */
+{
+	xfs_da_args_t		args;
+	int			rval;
+	int			v;		/* type-checking value */
+
+	ASSERT((dp->i_d.di_mode & S_IFMT) == S_IFDIR);
+	if ((rval = xfs_dir_ino_validate(tp->t_mountp, inum)))
+		return rval;
+	XFS_STATS_INC(xs_dir_create);
+
+	memset(&args, 0, sizeof(xfs_da_args_t));
+	args.name = name->name;
+	args.namelen = name->len;
+	args.hashval = dp->i_mount->m_dirnameops->hashname(name);
+	args.inumber = inum;
+	args.dp = dp;
+	args.firstblock = first;
+	args.flist = flist;
+	args.total = total;
+	args.whichfork = XFS_DATA_FORK;
+	args.trans = tp;
+	args.op_flags = XFS_DA_OP_ADDNAME | XFS_DA_OP_OKNOENT;
+
+	if (dp->i_d.di_format == XFS_DINODE_FMT_LOCAL)
+		rval = xfs_dir2_sf_addname(&args);
+	else if ((rval = xfs_dir2_isblock(tp, dp, &v)))
+		return rval;
+	else if (v)
+		rval = xfs_dir2_block_addname(&args);
+	else if ((rval = xfs_dir2_isleaf(tp, dp, &v)))
+		return rval;
+	else if (v)
+		rval = xfs_dir2_leaf_addname(&args);
+	else
+		rval = xfs_dir2_node_addname(&args);
+	return rval;
+}
+
+/*
+ * If doing a CI lookup and case-insensitive match, dup actual name into
+ * args.value. Return EEXIST for success (ie. name found) or an error.
+ */
+int
+xfs_dir_cilookup_result(
+	struct xfs_da_args *args,
+	const unsigned char *name,
+	int		len)
+{
+	if (args->cmpresult == XFS_CMP_DIFFERENT)
+		return ENOENT;
+	if (args->cmpresult != XFS_CMP_CASE ||
+					!(args->op_flags & XFS_DA_OP_CILOOKUP))
+		return EEXIST;
+
+	args->value = kmem_alloc(len, KM_NOFS | KM_MAYFAIL);
+	if (!args->value)
+		return ENOMEM;
+
+	memcpy(args->value, name, len);
+	args->valuelen = len;
+	return EEXIST;
+}
+
+/*
+ * Lookup a name in a directory, give back the inode number.
+ * If ci_name is not NULL, returns the actual name in ci_name if it differs
+ * to name, or ci_name->name is set to NULL for an exact match.
+ */
+
+int
+xfs_dir_lookup(
+	xfs_trans_t	*tp,
+	xfs_inode_t	*dp,
+	struct xfs_name	*name,
+	xfs_ino_t	*inum,		/* out: inode number */
+	struct xfs_name *ci_name)	/* out: actual name if CI match */
+{
+	xfs_da_args_t	args;
+	int		rval;
+	int		v;		/* type-checking value */
+
+	ASSERT((dp->i_d.di_mode & S_IFMT) == S_IFDIR);
+	XFS_STATS_INC(xs_dir_lookup);
+
+	memset(&args, 0, sizeof(xfs_da_args_t));
+	args.name = name->name;
+	args.namelen = name->len;
+	args.hashval = dp->i_mount->m_dirnameops->hashname(name);
+	args.dp = dp;
+	args.whichfork = XFS_DATA_FORK;
+	args.trans = tp;
+	args.op_flags = XFS_DA_OP_OKNOENT;
+	if (ci_name)
+		args.op_flags |= XFS_DA_OP_CILOOKUP;
+
+	if (dp->i_d.di_format == XFS_DINODE_FMT_LOCAL)
+		rval = xfs_dir2_sf_lookup(&args);
+	else if ((rval = xfs_dir2_isblock(tp, dp, &v)))
+		return rval;
+	else if (v)
+		rval = xfs_dir2_block_lookup(&args);
+	else if ((rval = xfs_dir2_isleaf(tp, dp, &v)))
+		return rval;
+	else if (v)
+		rval = xfs_dir2_leaf_lookup(&args);
+	else
+		rval = xfs_dir2_node_lookup(&args);
+	if (rval == EEXIST)
+		rval = 0;
+	if (!rval) {
+		*inum = args.inumber;
+		if (ci_name) {
+			ci_name->name = args.value;
+			ci_name->len = args.valuelen;
+		}
+	}
+	return rval;
+}
+
+/*
+ * Remove an entry from a directory.
+ */
+int
+xfs_dir_removename(
+	xfs_trans_t	*tp,
+	xfs_inode_t	*dp,
+	struct xfs_name	*name,
+	xfs_ino_t	ino,
+	xfs_fsblock_t	*first,		/* bmap's firstblock */
+	xfs_bmap_free_t	*flist,		/* bmap's freeblock list */
+	xfs_extlen_t	total)		/* bmap's total block count */
+{
+	xfs_da_args_t	args;
+	int		rval;
+	int		v;		/* type-checking value */
+
+	ASSERT((dp->i_d.di_mode & S_IFMT) == S_IFDIR);
+	XFS_STATS_INC(xs_dir_remove);
+
+	memset(&args, 0, sizeof(xfs_da_args_t));
+	args.name = name->name;
+	args.namelen = name->len;
+	args.hashval = dp->i_mount->m_dirnameops->hashname(name);
+	args.inumber = ino;
+	args.dp = dp;
+	args.firstblock = first;
+	args.flist = flist;
+	args.total = total;
+	args.whichfork = XFS_DATA_FORK;
+	args.trans = tp;
+
+	if (dp->i_d.di_format == XFS_DINODE_FMT_LOCAL)
+		rval = xfs_dir2_sf_removename(&args);
+	else if ((rval = xfs_dir2_isblock(tp, dp, &v)))
+		return rval;
+	else if (v)
+		rval = xfs_dir2_block_removename(&args);
+	else if ((rval = xfs_dir2_isleaf(tp, dp, &v)))
+		return rval;
+	else if (v)
+		rval = xfs_dir2_leaf_removename(&args);
+	else
+		rval = xfs_dir2_node_removename(&args);
+	return rval;
+}
+
+/*
+ * Read a directory.
+ */
+int
+xfs_readdir(
+	xfs_inode_t	*dp,
+	void		*dirent,
+	size_t		bufsize,
+	xfs_off_t	*offset,
+	filldir_t	filldir)
+{
+	int		rval;		/* return value */
+	int		v;		/* type-checking value */
+
+	trace_xfs_readdir(dp);
+
+	if (XFS_FORCED_SHUTDOWN(dp->i_mount))
+		return XFS_ERROR(EIO);
+
+	ASSERT((dp->i_d.di_mode & S_IFMT) == S_IFDIR);
+	XFS_STATS_INC(xs_dir_getdents);
+
+	if (dp->i_d.di_format == XFS_DINODE_FMT_LOCAL)
+		rval = xfs_dir2_sf_getdents(dp, dirent, offset, filldir);
+	else if ((rval = xfs_dir2_isblock(NULL, dp, &v)))
+		;
+	else if (v)
+		rval = xfs_dir2_block_getdents(dp, dirent, offset, filldir);
+	else
+		rval = xfs_dir2_leaf_getdents(dp, dirent, bufsize, offset,
+					      filldir);
+	return rval;
+}
+
+/*
+ * Replace the inode number of a directory entry.
+ */
+int
+xfs_dir_replace(
+	xfs_trans_t	*tp,
+	xfs_inode_t	*dp,
+	struct xfs_name	*name,		/* name of entry to replace */
+	xfs_ino_t	inum,		/* new inode number */
+	xfs_fsblock_t	*first,		/* bmap's firstblock */
+	xfs_bmap_free_t	*flist,		/* bmap's freeblock list */
+	xfs_extlen_t	total)		/* bmap's total block count */
+{
+	xfs_da_args_t	args;
+	int		rval;
+	int		v;		/* type-checking value */
+
+	ASSERT((dp->i_d.di_mode & S_IFMT) == S_IFDIR);
+
+	if ((rval = xfs_dir_ino_validate(tp->t_mountp, inum)))
+		return rval;
+
+	memset(&args, 0, sizeof(xfs_da_args_t));
+	args.name = name->name;
+	args.namelen = name->len;
+	args.hashval = dp->i_mount->m_dirnameops->hashname(name);
+	args.inumber = inum;
+	args.dp = dp;
+	args.firstblock = first;
+	args.flist = flist;
+	args.total = total;
+	args.whichfork = XFS_DATA_FORK;
+	args.trans = tp;
+
+	if (dp->i_d.di_format == XFS_DINODE_FMT_LOCAL)
+		rval = xfs_dir2_sf_replace(&args);
+	else if ((rval = xfs_dir2_isblock(tp, dp, &v)))
+		return rval;
+	else if (v)
+		rval = xfs_dir2_block_replace(&args);
+	else if ((rval = xfs_dir2_isleaf(tp, dp, &v)))
+		return rval;
+	else if (v)
+		rval = xfs_dir2_leaf_replace(&args);
+	else
+		rval = xfs_dir2_node_replace(&args);
+	return rval;
+}
+
+/*
+ * See if this entry can be added to the directory without allocating space.
+ * First checks that the caller couldn't reserve enough space (resblks = 0).
+ */
+int
+xfs_dir_canenter(
+	xfs_trans_t	*tp,
+	xfs_inode_t	*dp,
+	struct xfs_name	*name,		/* name of entry to add */
+	uint		resblks)
+{
+	xfs_da_args_t	args;
+	int		rval;
+	int		v;		/* type-checking value */
+
+	if (resblks)
+		return 0;
+
+	ASSERT((dp->i_d.di_mode & S_IFMT) == S_IFDIR);
+
+	memset(&args, 0, sizeof(xfs_da_args_t));
+	args.name = name->name;
+	args.namelen = name->len;
+	args.hashval = dp->i_mount->m_dirnameops->hashname(name);
+	args.dp = dp;
+	args.whichfork = XFS_DATA_FORK;
+	args.trans = tp;
+	args.op_flags = XFS_DA_OP_JUSTCHECK | XFS_DA_OP_ADDNAME |
+							XFS_DA_OP_OKNOENT;
+
+	if (dp->i_d.di_format == XFS_DINODE_FMT_LOCAL)
+		rval = xfs_dir2_sf_addname(&args);
+	else if ((rval = xfs_dir2_isblock(tp, dp, &v)))
+		return rval;
+	else if (v)
+		rval = xfs_dir2_block_addname(&args);
+	else if ((rval = xfs_dir2_isleaf(tp, dp, &v)))
+		return rval;
+	else if (v)
+		rval = xfs_dir2_leaf_addname(&args);
+	else
+		rval = xfs_dir2_node_addname(&args);
+	return rval;
+}
+
+/*
+ * Utility routines.
+ */
+
+/*
+ * Add a block to the directory.
+ * This routine is for data and free blocks, not leaf/node blocks
+ * which are handled by xfs_da_grow_inode.
+ */
+int
+xfs_dir2_grow_inode(
+	xfs_da_args_t	*args,
+	int		space,		/* v2 dir's space XFS_DIR2_xxx_SPACE */
+	xfs_dir2_db_t	*dbp)		/* out: block number added */
+{
+	xfs_fileoff_t	bno;		/* directory offset of new block */
+	int		count;		/* count of filesystem blocks */
+	xfs_inode_t	*dp;		/* incore directory inode */
+	int		error;
+	int		got;		/* blocks actually mapped */
+	int		i;
+	xfs_bmbt_irec_t	map;		/* single structure for bmap */
+	int		mapi;		/* mapping index */
+	xfs_bmbt_irec_t	*mapp;		/* bmap mapping structure(s) */
+	xfs_mount_t	*mp;
+	int		nmap;		/* number of bmap entries */
+	xfs_trans_t	*tp;
+	xfs_drfsbno_t	nblks;
+
+	trace_xfs_dir2_grow_inode(args, space);
+
+	dp = args->dp;
+	tp = args->trans;
+	mp = dp->i_mount;
+	nblks = dp->i_d.di_nblocks;
+	/*
+	 * Set lowest possible block in the space requested.
+	 */
+	bno = XFS_B_TO_FSBT(mp, space * XFS_DIR2_SPACE_SIZE);
+	count = mp->m_dirblkfsbs;
+	/*
+	 * Find the first hole for our block.
+	 */
+	if ((error = xfs_bmap_first_unused(tp, dp, count, &bno, XFS_DATA_FORK)))
+		return error;
+	nmap = 1;
+	ASSERT(args->firstblock != NULL);
+	/*
+	 * Try mapping the new block contiguously (one extent).
+	 */
+	if ((error = xfs_bmapi(tp, dp, bno, count,
+			XFS_BMAPI_WRITE|XFS_BMAPI_METADATA|XFS_BMAPI_CONTIG,
+			args->firstblock, args->total, &map, &nmap,
+			args->flist)))
+		return error;
+	ASSERT(nmap <= 1);
+	if (nmap == 1) {
+		mapp = &map;
+		mapi = 1;
+	}
+	/*
+	 * Didn't work and this is a multiple-fsb directory block.
+	 * Try again with contiguous flag turned on.
+	 */
+	else if (nmap == 0 && count > 1) {
+		xfs_fileoff_t	b;	/* current file offset */
+
+		/*
+		 * Space for maximum number of mappings.
+		 */
+		mapp = kmem_alloc(sizeof(*mapp) * count, KM_SLEEP);
+		/*
+		 * Iterate until we get to the end of our block.
+		 */
+		for (b = bno, mapi = 0; b < bno + count; ) {
+			int	c;	/* current fsb count */
+
+			/*
+			 * Can't map more than MAX_NMAP at once.
+			 */
+			nmap = MIN(XFS_BMAP_MAX_NMAP, count);
+			c = (int)(bno + count - b);
+			if ((error = xfs_bmapi(tp, dp, b, c,
+					XFS_BMAPI_WRITE|XFS_BMAPI_METADATA,
+					args->firstblock, args->total,
+					&mapp[mapi], &nmap, args->flist))) {
+				kmem_free(mapp);
+				return error;
+			}
+			if (nmap < 1)
+				break;
+			/*
+			 * Add this bunch into our table, go to the next offset.
+			 */
+			mapi += nmap;
+			b = mapp[mapi - 1].br_startoff +
+			    mapp[mapi - 1].br_blockcount;
+		}
+	}
+	/*
+	 * Didn't work.
+	 */
+	else {
+		mapi = 0;
+		mapp = NULL;
+	}
+	/*
+	 * See how many fsb's we got.
+	 */
+	for (i = 0, got = 0; i < mapi; i++)
+		got += mapp[i].br_blockcount;
+	/*
+	 * Didn't get enough fsb's, or the first/last block's are wrong.
+	 */
+	if (got != count || mapp[0].br_startoff != bno ||
+	    mapp[mapi - 1].br_startoff + mapp[mapi - 1].br_blockcount !=
+	    bno + count) {
+		if (mapp != &map)
+			kmem_free(mapp);
+		return XFS_ERROR(ENOSPC);
+	}
+	/*
+	 * Done with the temporary mapping table.
+	 */
+	if (mapp != &map)
+		kmem_free(mapp);
+
+	/* account for newly allocated blocks in reserved blocks total */
+	args->total -= dp->i_d.di_nblocks - nblks;
+	*dbp = xfs_dir2_da_to_db(mp, (xfs_dablk_t)bno);
+
+	/*
+	 * Update file's size if this is the data space and it grew.
+	 */
+	if (space == XFS_DIR2_DATA_SPACE) {
+		xfs_fsize_t	size;		/* directory file (data) size */
+
+		size = XFS_FSB_TO_B(mp, bno + count);
+		if (size > dp->i_d.di_size) {
+			dp->i_d.di_size = size;
+			xfs_trans_log_inode(tp, dp, XFS_ILOG_CORE);
+		}
+	}
+	return 0;
+}
+
+/*
+ * See if the directory is a single-block form directory.
+ */
+int
+xfs_dir2_isblock(
+	xfs_trans_t	*tp,
+	xfs_inode_t	*dp,
+	int		*vp)		/* out: 1 is block, 0 is not block */
+{
+	xfs_fileoff_t	last;		/* last file offset */
+	xfs_mount_t	*mp;
+	int		rval;
+
+	mp = dp->i_mount;
+	if ((rval = xfs_bmap_last_offset(tp, dp, &last, XFS_DATA_FORK)))
+		return rval;
+	rval = XFS_FSB_TO_B(mp, last) == mp->m_dirblksize;
+	ASSERT(rval == 0 || dp->i_d.di_size == mp->m_dirblksize);
+	*vp = rval;
+	return 0;
+}
+
+/*
+ * See if the directory is a single-leaf form directory.
+ */
+int
+xfs_dir2_isleaf(
+	xfs_trans_t	*tp,
+	xfs_inode_t	*dp,
+	int		*vp)		/* out: 1 is leaf, 0 is not leaf */
+{
+	xfs_fileoff_t	last;		/* last file offset */
+	xfs_mount_t	*mp;
+	int		rval;
+
+	mp = dp->i_mount;
+	if ((rval = xfs_bmap_last_offset(tp, dp, &last, XFS_DATA_FORK)))
+		return rval;
+	*vp = last == mp->m_dirleafblk + (1 << mp->m_sb.sb_dirblklog);
+	return 0;
+}
+
+/*
+ * Remove the given block from the directory.
+ * This routine is used for data and free blocks, leaf/node are done
+ * by xfs_da_shrink_inode.
+ */
+int
+xfs_dir2_shrink_inode(
+	xfs_da_args_t	*args,
+	xfs_dir2_db_t	db,
+	xfs_dabuf_t	*bp)
+{
+	xfs_fileoff_t	bno;		/* directory file offset */
+	xfs_dablk_t	da;		/* directory file offset */
+	int		done;		/* bunmap is finished */
+	xfs_inode_t	*dp;
+	int		error;
+	xfs_mount_t	*mp;
+	xfs_trans_t	*tp;
+
+	trace_xfs_dir2_shrink_inode(args, db);
+
+	dp = args->dp;
+	mp = dp->i_mount;
+	tp = args->trans;
+	da = xfs_dir2_db_to_da(mp, db);
+	/*
+	 * Unmap the fsblock(s).
+	 */
+	if ((error = xfs_bunmapi(tp, dp, da, mp->m_dirblkfsbs,
+			XFS_BMAPI_METADATA, 0, args->firstblock, args->flist,
+			&done))) {
+		/*
+		 * ENOSPC actually can happen if we're in a removename with
+		 * no space reservation, and the resulting block removal
+		 * would cause a bmap btree split or conversion from extents
+		 * to btree.  This can only happen for un-fragmented
+		 * directory blocks, since you need to be punching out
+		 * the middle of an extent.
+		 * In this case we need to leave the block in the file,
+		 * and not binval it.
+		 * So the block has to be in a consistent empty state
+		 * and appropriately logged.
+		 * We don't free up the buffer, the caller can tell it
+		 * hasn't happened since it got an error back.
+		 */
+		return error;
+	}
+	ASSERT(done);
+	/*
+	 * Invalidate the buffer from the transaction.
+	 */
+	xfs_da_binval(tp, bp);
+	/*
+	 * If it's not a data block, we're done.
+	 */
+	if (db >= XFS_DIR2_LEAF_FIRSTDB(mp))
+		return 0;
+	/*
+	 * If the block isn't the last one in the directory, we're done.
+	 */
+	if (dp->i_d.di_size > xfs_dir2_db_off_to_byte(mp, db + 1, 0))
+		return 0;
+	bno = da;
+	if ((error = xfs_bmap_last_before(tp, dp, &bno, XFS_DATA_FORK))) {
+		/*
+		 * This can't really happen unless there's kernel corruption.
+		 */
+		return error;
+	}
+	if (db == mp->m_dirdatablk)
+		ASSERT(bno == 0);
+	else
+		ASSERT(bno > 0);
+	/*
+	 * Set the size to the new last block.
+	 */
+	dp->i_d.di_size = XFS_FSB_TO_B(mp, bno);
+	xfs_trans_log_inode(tp, dp, XFS_ILOG_CORE);
+	return 0;
+}
diff --git a/fs/xfs/xfs_dir2.h b/fs/xfs/xfs_dir2.h
new file mode 100644
index 00000000..74a3b105
--- /dev/null
+++ b/fs/xfs/xfs_dir2.h
@@ -0,0 +1,106 @@
+/*
+ * Copyright (c) 2000-2001,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_DIR2_H__
+#define	__XFS_DIR2_H__
+
+struct uio;
+struct xfs_dabuf;
+struct xfs_da_args;
+struct xfs_dir2_put_args;
+struct xfs_bmap_free;
+struct xfs_inode;
+struct xfs_mount;
+struct xfs_trans;
+
+/*
+ * Directory version 2.
+ * There are 4 possible formats:
+ *	shortform
+ *	single block - data with embedded leaf at the end
+ *	multiple data blocks, single leaf+freeindex block
+ *	data blocks, node&leaf blocks (btree), freeindex blocks
+ *
+ *	The shortform format is in xfs_dir2_sf.h.
+ *	The single block format is in xfs_dir2_block.h.
+ *	The data block format is in xfs_dir2_data.h.
+ *	The leaf and freeindex block formats are in xfs_dir2_leaf.h.
+ *	Node blocks are the same as the other version, in xfs_da_btree.h.
+ */
+
+/*
+ * Byte offset in data block and shortform entry.
+ */
+typedef	__uint16_t	xfs_dir2_data_off_t;
+#define	NULLDATAOFF	0xffffU
+typedef uint		xfs_dir2_data_aoff_t;	/* argument form */
+
+/*
+ * Directory block number (logical dirblk in file)
+ */
+typedef	__uint32_t	xfs_dir2_db_t;
+
+/*
+ * Byte offset in a directory.
+ */
+typedef	xfs_off_t	xfs_dir2_off_t;
+
+extern struct xfs_name	xfs_name_dotdot;
+
+/*
+ * Generic directory interface routines
+ */
+extern void xfs_dir_startup(void);
+extern void xfs_dir_mount(struct xfs_mount *mp);
+extern int xfs_dir_isempty(struct xfs_inode *dp);
+extern int xfs_dir_init(struct xfs_trans *tp, struct xfs_inode *dp,
+				struct xfs_inode *pdp);
+extern int xfs_dir_createname(struct xfs_trans *tp, struct xfs_inode *dp,
+				struct xfs_name *name, xfs_ino_t inum,
+				xfs_fsblock_t *first,
+				struct xfs_bmap_free *flist, xfs_extlen_t tot);
+extern int xfs_dir_lookup(struct xfs_trans *tp, struct xfs_inode *dp,
+				struct xfs_name *name, xfs_ino_t *inum,
+				struct xfs_name *ci_name);
+extern int xfs_dir_removename(struct xfs_trans *tp, struct xfs_inode *dp,
+				struct xfs_name *name, xfs_ino_t ino,
+				xfs_fsblock_t *first,
+				struct xfs_bmap_free *flist, xfs_extlen_t tot);
+extern int xfs_dir_replace(struct xfs_trans *tp, struct xfs_inode *dp,
+				struct xfs_name *name, xfs_ino_t inum,
+				xfs_fsblock_t *first,
+				struct xfs_bmap_free *flist, xfs_extlen_t tot);
+extern int xfs_dir_canenter(struct xfs_trans *tp, struct xfs_inode *dp,
+				struct xfs_name *name, uint resblks);
+extern int xfs_dir_ino_validate(struct xfs_mount *mp, xfs_ino_t ino);
+
+/*
+ * Utility routines for v2 directories.
+ */
+extern int xfs_dir2_grow_inode(struct xfs_da_args *args, int space,
+				xfs_dir2_db_t *dbp);
+extern int xfs_dir2_isblock(struct xfs_trans *tp, struct xfs_inode *dp,
+				int *vp);
+extern int xfs_dir2_isleaf(struct xfs_trans *tp, struct xfs_inode *dp,
+				int *vp);
+extern int xfs_dir2_shrink_inode(struct xfs_da_args *args, xfs_dir2_db_t db,
+				struct xfs_dabuf *bp);
+
+extern int xfs_dir_cilookup_result(struct xfs_da_args *args,
+				const unsigned char *name, int len);
+
+#endif	/* __XFS_DIR2_H__ */
diff --git a/fs/xfs/xfs_dir2_block.c b/fs/xfs/xfs_dir2_block.c
new file mode 100644
index 00000000..580d99ce
--- /dev/null
+++ b/fs/xfs/xfs_dir2_block.c
@@ -0,0 +1,1233 @@
+/*
+ * Copyright (c) 2000-2003,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_dir2.h"
+#include "xfs_mount.h"
+#include "xfs_da_btree.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_dir2_sf.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_inode_item.h"
+#include "xfs_dir2_data.h"
+#include "xfs_dir2_leaf.h"
+#include "xfs_dir2_block.h"
+#include "xfs_error.h"
+#include "xfs_trace.h"
+
+/*
+ * Local function prototypes.
+ */
+static void xfs_dir2_block_log_leaf(xfs_trans_t *tp, xfs_dabuf_t *bp, int first,
+				    int last);
+static void xfs_dir2_block_log_tail(xfs_trans_t *tp, xfs_dabuf_t *bp);
+static int xfs_dir2_block_lookup_int(xfs_da_args_t *args, xfs_dabuf_t **bpp,
+				     int *entno);
+static int xfs_dir2_block_sort(const void *a, const void *b);
+
+static xfs_dahash_t xfs_dir_hash_dot, xfs_dir_hash_dotdot;
+
+/*
+ * One-time startup routine called from xfs_init().
+ */
+void
+xfs_dir_startup(void)
+{
+	xfs_dir_hash_dot = xfs_da_hashname((unsigned char *)".", 1);
+	xfs_dir_hash_dotdot = xfs_da_hashname((unsigned char *)"..", 2);
+}
+
+/*
+ * Add an entry to a block directory.
+ */
+int						/* error */
+xfs_dir2_block_addname(
+	xfs_da_args_t		*args)		/* directory op arguments */
+{
+	xfs_dir2_data_free_t	*bf;		/* bestfree table in block */
+	xfs_dir2_block_t	*block;		/* directory block structure */
+	xfs_dir2_leaf_entry_t	*blp;		/* block leaf entries */
+	xfs_dabuf_t		*bp;		/* buffer for block */
+	xfs_dir2_block_tail_t	*btp;		/* block tail */
+	int			compact;	/* need to compact leaf ents */
+	xfs_dir2_data_entry_t	*dep;		/* block data entry */
+	xfs_inode_t		*dp;		/* directory inode */
+	xfs_dir2_data_unused_t	*dup;		/* block unused entry */
+	int			error;		/* error return value */
+	xfs_dir2_data_unused_t	*enddup=NULL;	/* unused at end of data */
+	xfs_dahash_t		hash;		/* hash value of found entry */
+	int			high;		/* high index for binary srch */
+	int			highstale;	/* high stale index */
+	int			lfloghigh=0;	/* last final leaf to log */
+	int			lfloglow=0;	/* first final leaf to log */
+	int			len;		/* length of the new entry */
+	int			low;		/* low index for binary srch */
+	int			lowstale;	/* low stale index */
+	int			mid=0;		/* midpoint for binary srch */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+	int			needlog;	/* need to log header */
+	int			needscan;	/* need to rescan freespace */
+	__be16			*tagp;		/* pointer to tag value */
+	xfs_trans_t		*tp;		/* transaction structure */
+
+	trace_xfs_dir2_block_addname(args);
+
+	dp = args->dp;
+	tp = args->trans;
+	mp = dp->i_mount;
+	/*
+	 * Read the (one and only) directory block into dabuf bp.
+	 */
+	if ((error =
+	    xfs_da_read_buf(tp, dp, mp->m_dirdatablk, -1, &bp, XFS_DATA_FORK))) {
+		return error;
+	}
+	ASSERT(bp != NULL);
+	block = bp->data;
+	/*
+	 * Check the magic number, corrupted if wrong.
+	 */
+	if (unlikely(be32_to_cpu(block->hdr.magic) != XFS_DIR2_BLOCK_MAGIC)) {
+		XFS_CORRUPTION_ERROR("xfs_dir2_block_addname",
+				     XFS_ERRLEVEL_LOW, mp, block);
+		xfs_da_brelse(tp, bp);
+		return XFS_ERROR(EFSCORRUPTED);
+	}
+	len = xfs_dir2_data_entsize(args->namelen);
+	/*
+	 * Set up pointers to parts of the block.
+	 */
+	bf = block->hdr.bestfree;
+	btp = xfs_dir2_block_tail_p(mp, block);
+	blp = xfs_dir2_block_leaf_p(btp);
+	/*
+	 * No stale entries?  Need space for entry and new leaf.
+	 */
+	if (!btp->stale) {
+		/*
+		 * Tag just before the first leaf entry.
+		 */
+		tagp = (__be16 *)blp - 1;
+		/*
+		 * Data object just before the first leaf entry.
+		 */
+		enddup = (xfs_dir2_data_unused_t *)((char *)block + be16_to_cpu(*tagp));
+		/*
+		 * If it's not free then can't do this add without cleaning up:
+		 * the space before the first leaf entry needs to be free so it
+		 * can be expanded to hold the pointer to the new entry.
+		 */
+		if (be16_to_cpu(enddup->freetag) != XFS_DIR2_DATA_FREE_TAG)
+			dup = enddup = NULL;
+		/*
+		 * Check out the biggest freespace and see if it's the same one.
+		 */
+		else {
+			dup = (xfs_dir2_data_unused_t *)
+			      ((char *)block + be16_to_cpu(bf[0].offset));
+			if (dup == enddup) {
+				/*
+				 * It is the biggest freespace, is it too small
+				 * to hold the new leaf too?
+				 */
+				if (be16_to_cpu(dup->length) < len + (uint)sizeof(*blp)) {
+					/*
+					 * Yes, we use the second-largest
+					 * entry instead if it works.
+					 */
+					if (be16_to_cpu(bf[1].length) >= len)
+						dup = (xfs_dir2_data_unused_t *)
+						      ((char *)block +
+						       be16_to_cpu(bf[1].offset));
+					else
+						dup = NULL;
+				}
+			} else {
+				/*
+				 * Not the same free entry,
+				 * just check its length.
+				 */
+				if (be16_to_cpu(dup->length) < len) {
+					dup = NULL;
+				}
+			}
+		}
+		compact = 0;
+	}
+	/*
+	 * If there are stale entries we'll use one for the leaf.
+	 * Is the biggest entry enough to avoid compaction?
+	 */
+	else if (be16_to_cpu(bf[0].length) >= len) {
+		dup = (xfs_dir2_data_unused_t *)
+		      ((char *)block + be16_to_cpu(bf[0].offset));
+		compact = 0;
+	}
+	/*
+	 * Will need to compact to make this work.
+	 */
+	else {
+		/*
+		 * Tag just before the first leaf entry.
+		 */
+		tagp = (__be16 *)blp - 1;
+		/*
+		 * Data object just before the first leaf entry.
+		 */
+		dup = (xfs_dir2_data_unused_t *)((char *)block + be16_to_cpu(*tagp));
+		/*
+		 * If it's not free then the data will go where the
+		 * leaf data starts now, if it works at all.
+		 */
+		if (be16_to_cpu(dup->freetag) == XFS_DIR2_DATA_FREE_TAG) {
+			if (be16_to_cpu(dup->length) + (be32_to_cpu(btp->stale) - 1) *
+			    (uint)sizeof(*blp) < len)
+				dup = NULL;
+		} else if ((be32_to_cpu(btp->stale) - 1) * (uint)sizeof(*blp) < len)
+			dup = NULL;
+		else
+			dup = (xfs_dir2_data_unused_t *)blp;
+		compact = 1;
+	}
+	/*
+	 * If this isn't a real add, we're done with the buffer.
+	 */
+	if (args->op_flags & XFS_DA_OP_JUSTCHECK)
+		xfs_da_brelse(tp, bp);
+	/*
+	 * If we don't have space for the new entry & leaf ...
+	 */
+	if (!dup) {
+		/*
+		 * Not trying to actually do anything, or don't have
+		 * a space reservation: return no-space.
+		 */
+		if ((args->op_flags & XFS_DA_OP_JUSTCHECK) || args->total == 0)
+			return XFS_ERROR(ENOSPC);
+		/*
+		 * Convert to the next larger format.
+		 * Then add the new entry in that format.
+		 */
+		error = xfs_dir2_block_to_leaf(args, bp);
+		xfs_da_buf_done(bp);
+		if (error)
+			return error;
+		return xfs_dir2_leaf_addname(args);
+	}
+	/*
+	 * Just checking, and it would work, so say so.
+	 */
+	if (args->op_flags & XFS_DA_OP_JUSTCHECK)
+		return 0;
+	needlog = needscan = 0;
+	/*
+	 * If need to compact the leaf entries, do it now.
+	 * Leave the highest-numbered stale entry stale.
+	 * XXX should be the one closest to mid but mid is not yet computed.
+	 */
+	if (compact) {
+		int	fromidx;		/* source leaf index */
+		int	toidx;			/* target leaf index */
+
+		for (fromidx = toidx = be32_to_cpu(btp->count) - 1,
+			highstale = lfloghigh = -1;
+		     fromidx >= 0;
+		     fromidx--) {
+			if (be32_to_cpu(blp[fromidx].address) == XFS_DIR2_NULL_DATAPTR) {
+				if (highstale == -1)
+					highstale = toidx;
+				else {
+					if (lfloghigh == -1)
+						lfloghigh = toidx;
+					continue;
+				}
+			}
+			if (fromidx < toidx)
+				blp[toidx] = blp[fromidx];
+			toidx--;
+		}
+		lfloglow = toidx + 1 - (be32_to_cpu(btp->stale) - 1);
+		lfloghigh -= be32_to_cpu(btp->stale) - 1;
+		be32_add_cpu(&btp->count, -(be32_to_cpu(btp->stale) - 1));
+		xfs_dir2_data_make_free(tp, bp,
+			(xfs_dir2_data_aoff_t)((char *)blp - (char *)block),
+			(xfs_dir2_data_aoff_t)((be32_to_cpu(btp->stale) - 1) * sizeof(*blp)),
+			&needlog, &needscan);
+		blp += be32_to_cpu(btp->stale) - 1;
+		btp->stale = cpu_to_be32(1);
+		/*
+		 * If we now need to rebuild the bestfree map, do so.
+		 * This needs to happen before the next call to use_free.
+		 */
+		if (needscan) {
+			xfs_dir2_data_freescan(mp, (xfs_dir2_data_t *)block, &needlog);
+			needscan = 0;
+		}
+	}
+	/*
+	 * Set leaf logging boundaries to impossible state.
+	 * For the no-stale case they're set explicitly.
+	 */
+	else if (btp->stale) {
+		lfloglow = be32_to_cpu(btp->count);
+		lfloghigh = -1;
+	}
+	/*
+	 * Find the slot that's first lower than our hash value, -1 if none.
+	 */
+	for (low = 0, high = be32_to_cpu(btp->count) - 1; low <= high; ) {
+		mid = (low + high) >> 1;
+		if ((hash = be32_to_cpu(blp[mid].hashval)) == args->hashval)
+			break;
+		if (hash < args->hashval)
+			low = mid + 1;
+		else
+			high = mid - 1;
+	}
+	while (mid >= 0 && be32_to_cpu(blp[mid].hashval) >= args->hashval) {
+		mid--;
+	}
+	/*
+	 * No stale entries, will use enddup space to hold new leaf.
+	 */
+	if (!btp->stale) {
+		/*
+		 * Mark the space needed for the new leaf entry, now in use.
+		 */
+		xfs_dir2_data_use_free(tp, bp, enddup,
+			(xfs_dir2_data_aoff_t)
+			((char *)enddup - (char *)block + be16_to_cpu(enddup->length) -
+			 sizeof(*blp)),
+			(xfs_dir2_data_aoff_t)sizeof(*blp),
+			&needlog, &needscan);
+		/*
+		 * Update the tail (entry count).
+		 */
+		be32_add_cpu(&btp->count, 1);
+		/*
+		 * If we now need to rebuild the bestfree map, do so.
+		 * This needs to happen before the next call to use_free.
+		 */
+		if (needscan) {
+			xfs_dir2_data_freescan(mp, (xfs_dir2_data_t *)block,
+				&needlog);
+			needscan = 0;
+		}
+		/*
+		 * Adjust pointer to the first leaf entry, we're about to move
+		 * the table up one to open up space for the new leaf entry.
+		 * Then adjust our index to match.
+		 */
+		blp--;
+		mid++;
+		if (mid)
+			memmove(blp, &blp[1], mid * sizeof(*blp));
+		lfloglow = 0;
+		lfloghigh = mid;
+	}
+	/*
+	 * Use a stale leaf for our new entry.
+	 */
+	else {
+		for (lowstale = mid;
+		     lowstale >= 0 &&
+			be32_to_cpu(blp[lowstale].address) != XFS_DIR2_NULL_DATAPTR;
+		     lowstale--)
+			continue;
+		for (highstale = mid + 1;
+		     highstale < be32_to_cpu(btp->count) &&
+			be32_to_cpu(blp[highstale].address) != XFS_DIR2_NULL_DATAPTR &&
+			(lowstale < 0 || mid - lowstale > highstale - mid);
+		     highstale++)
+			continue;
+		/*
+		 * Move entries toward the low-numbered stale entry.
+		 */
+		if (lowstale >= 0 &&
+		    (highstale == be32_to_cpu(btp->count) ||
+		     mid - lowstale <= highstale - mid)) {
+			if (mid - lowstale)
+				memmove(&blp[lowstale], &blp[lowstale + 1],
+					(mid - lowstale) * sizeof(*blp));
+			lfloglow = MIN(lowstale, lfloglow);
+			lfloghigh = MAX(mid, lfloghigh);
+		}
+		/*
+		 * Move entries toward the high-numbered stale entry.
+		 */
+		else {
+			ASSERT(highstale < be32_to_cpu(btp->count));
+			mid++;
+			if (highstale - mid)
+				memmove(&blp[mid + 1], &blp[mid],
+					(highstale - mid) * sizeof(*blp));
+			lfloglow = MIN(mid, lfloglow);
+			lfloghigh = MAX(highstale, lfloghigh);
+		}
+		be32_add_cpu(&btp->stale, -1);
+	}
+	/*
+	 * Point to the new data entry.
+	 */
+	dep = (xfs_dir2_data_entry_t *)dup;
+	/*
+	 * Fill in the leaf entry.
+	 */
+	blp[mid].hashval = cpu_to_be32(args->hashval);
+	blp[mid].address = cpu_to_be32(xfs_dir2_byte_to_dataptr(mp,
+				(char *)dep - (char *)block));
+	xfs_dir2_block_log_leaf(tp, bp, lfloglow, lfloghigh);
+	/*
+	 * Mark space for the data entry used.
+	 */
+	xfs_dir2_data_use_free(tp, bp, dup,
+		(xfs_dir2_data_aoff_t)((char *)dup - (char *)block),
+		(xfs_dir2_data_aoff_t)len, &needlog, &needscan);
+	/*
+	 * Create the new data entry.
+	 */
+	dep->inumber = cpu_to_be64(args->inumber);
+	dep->namelen = args->namelen;
+	memcpy(dep->name, args->name, args->namelen);
+	tagp = xfs_dir2_data_entry_tag_p(dep);
+	*tagp = cpu_to_be16((char *)dep - (char *)block);
+	/*
+	 * Clean up the bestfree array and log the header, tail, and entry.
+	 */
+	if (needscan)
+		xfs_dir2_data_freescan(mp, (xfs_dir2_data_t *)block, &needlog);
+	if (needlog)
+		xfs_dir2_data_log_header(tp, bp);
+	xfs_dir2_block_log_tail(tp, bp);
+	xfs_dir2_data_log_entry(tp, bp, dep);
+	xfs_dir2_data_check(dp, bp);
+	xfs_da_buf_done(bp);
+	return 0;
+}
+
+/*
+ * Readdir for block directories.
+ */
+int						/* error */
+xfs_dir2_block_getdents(
+	xfs_inode_t		*dp,		/* incore inode */
+	void			*dirent,
+	xfs_off_t		*offset,
+	filldir_t		filldir)
+{
+	xfs_dir2_block_t	*block;		/* directory block structure */
+	xfs_dabuf_t		*bp;		/* buffer for block */
+	xfs_dir2_block_tail_t	*btp;		/* block tail */
+	xfs_dir2_data_entry_t	*dep;		/* block data entry */
+	xfs_dir2_data_unused_t	*dup;		/* block unused entry */
+	char			*endptr;	/* end of the data entries */
+	int			error;		/* error return value */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+	char			*ptr;		/* current data entry */
+	int			wantoff;	/* starting block offset */
+	xfs_off_t		cook;
+
+	mp = dp->i_mount;
+	/*
+	 * If the block number in the offset is out of range, we're done.
+	 */
+	if (xfs_dir2_dataptr_to_db(mp, *offset) > mp->m_dirdatablk) {
+		return 0;
+	}
+	/*
+	 * Can't read the block, give up, else get dabuf in bp.
+	 */
+	error = xfs_da_read_buf(NULL, dp, mp->m_dirdatablk, -1,
+				&bp, XFS_DATA_FORK);
+	if (error)
+		return error;
+
+	ASSERT(bp != NULL);
+	/*
+	 * Extract the byte offset we start at from the seek pointer.
+	 * We'll skip entries before this.
+	 */
+	wantoff = xfs_dir2_dataptr_to_off(mp, *offset);
+	block = bp->data;
+	xfs_dir2_data_check(dp, bp);
+	/*
+	 * Set up values for the loop.
+	 */
+	btp = xfs_dir2_block_tail_p(mp, block);
+	ptr = (char *)block->u;
+	endptr = (char *)xfs_dir2_block_leaf_p(btp);
+
+	/*
+	 * Loop over the data portion of the block.
+	 * Each object is a real entry (dep) or an unused one (dup).
+	 */
+	while (ptr < endptr) {
+		dup = (xfs_dir2_data_unused_t *)ptr;
+		/*
+		 * Unused, skip it.
+		 */
+		if (be16_to_cpu(dup->freetag) == XFS_DIR2_DATA_FREE_TAG) {
+			ptr += be16_to_cpu(dup->length);
+			continue;
+		}
+
+		dep = (xfs_dir2_data_entry_t *)ptr;
+
+		/*
+		 * Bump pointer for the next iteration.
+		 */
+		ptr += xfs_dir2_data_entsize(dep->namelen);
+		/*
+		 * The entry is before the desired starting point, skip it.
+		 */
+		if ((char *)dep - (char *)block < wantoff)
+			continue;
+
+		cook = xfs_dir2_db_off_to_dataptr(mp, mp->m_dirdatablk,
+					    (char *)dep - (char *)block);
+
+		/*
+		 * If it didn't fit, set the final offset to here & return.
+		 */
+		if (filldir(dirent, (char *)dep->name, dep->namelen,
+			    cook & 0x7fffffff, be64_to_cpu(dep->inumber),
+			    DT_UNKNOWN)) {
+			*offset = cook & 0x7fffffff;
+			xfs_da_brelse(NULL, bp);
+			return 0;
+		}
+	}
+
+	/*
+	 * Reached the end of the block.
+	 * Set the offset to a non-existent block 1 and return.
+	 */
+	*offset = xfs_dir2_db_off_to_dataptr(mp, mp->m_dirdatablk + 1, 0) &
+			0x7fffffff;
+	xfs_da_brelse(NULL, bp);
+	return 0;
+}
+
+/*
+ * Log leaf entries from the block.
+ */
+static void
+xfs_dir2_block_log_leaf(
+	xfs_trans_t		*tp,		/* transaction structure */
+	xfs_dabuf_t		*bp,		/* block buffer */
+	int			first,		/* index of first logged leaf */
+	int			last)		/* index of last logged leaf */
+{
+	xfs_dir2_block_t	*block;		/* directory block structure */
+	xfs_dir2_leaf_entry_t	*blp;		/* block leaf entries */
+	xfs_dir2_block_tail_t	*btp;		/* block tail */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+
+	mp = tp->t_mountp;
+	block = bp->data;
+	btp = xfs_dir2_block_tail_p(mp, block);
+	blp = xfs_dir2_block_leaf_p(btp);
+	xfs_da_log_buf(tp, bp, (uint)((char *)&blp[first] - (char *)block),
+		(uint)((char *)&blp[last + 1] - (char *)block - 1));
+}
+
+/*
+ * Log the block tail.
+ */
+static void
+xfs_dir2_block_log_tail(
+	xfs_trans_t		*tp,		/* transaction structure */
+	xfs_dabuf_t		*bp)		/* block buffer */
+{
+	xfs_dir2_block_t	*block;		/* directory block structure */
+	xfs_dir2_block_tail_t	*btp;		/* block tail */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+
+	mp = tp->t_mountp;
+	block = bp->data;
+	btp = xfs_dir2_block_tail_p(mp, block);
+	xfs_da_log_buf(tp, bp, (uint)((char *)btp - (char *)block),
+		(uint)((char *)(btp + 1) - (char *)block - 1));
+}
+
+/*
+ * Look up an entry in the block.  This is the external routine,
+ * xfs_dir2_block_lookup_int does the real work.
+ */
+int						/* error */
+xfs_dir2_block_lookup(
+	xfs_da_args_t		*args)		/* dir lookup arguments */
+{
+	xfs_dir2_block_t	*block;		/* block structure */
+	xfs_dir2_leaf_entry_t	*blp;		/* block leaf entries */
+	xfs_dabuf_t		*bp;		/* block buffer */
+	xfs_dir2_block_tail_t	*btp;		/* block tail */
+	xfs_dir2_data_entry_t	*dep;		/* block data entry */
+	xfs_inode_t		*dp;		/* incore inode */
+	int			ent;		/* entry index */
+	int			error;		/* error return value */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+
+	trace_xfs_dir2_block_lookup(args);
+
+	/*
+	 * Get the buffer, look up the entry.
+	 * If not found (ENOENT) then return, have no buffer.
+	 */
+	if ((error = xfs_dir2_block_lookup_int(args, &bp, &ent)))
+		return error;
+	dp = args->dp;
+	mp = dp->i_mount;
+	block = bp->data;
+	xfs_dir2_data_check(dp, bp);
+	btp = xfs_dir2_block_tail_p(mp, block);
+	blp = xfs_dir2_block_leaf_p(btp);
+	/*
+	 * Get the offset from the leaf entry, to point to the data.
+	 */
+	dep = (xfs_dir2_data_entry_t *)((char *)block +
+		xfs_dir2_dataptr_to_off(mp, be32_to_cpu(blp[ent].address)));
+	/*
+	 * Fill in inode number, CI name if appropriate, release the block.
+	 */
+	args->inumber = be64_to_cpu(dep->inumber);
+	error = xfs_dir_cilookup_result(args, dep->name, dep->namelen);
+	xfs_da_brelse(args->trans, bp);
+	return XFS_ERROR(error);
+}
+
+/*
+ * Internal block lookup routine.
+ */
+static int					/* error */
+xfs_dir2_block_lookup_int(
+	xfs_da_args_t		*args,		/* dir lookup arguments */
+	xfs_dabuf_t		**bpp,		/* returned block buffer */
+	int			*entno)		/* returned entry number */
+{
+	xfs_dir2_dataptr_t	addr;		/* data entry address */
+	xfs_dir2_block_t	*block;		/* block structure */
+	xfs_dir2_leaf_entry_t	*blp;		/* block leaf entries */
+	xfs_dabuf_t		*bp;		/* block buffer */
+	xfs_dir2_block_tail_t	*btp;		/* block tail */
+	xfs_dir2_data_entry_t	*dep;		/* block data entry */
+	xfs_inode_t		*dp;		/* incore inode */
+	int			error;		/* error return value */
+	xfs_dahash_t		hash;		/* found hash value */
+	int			high;		/* binary search high index */
+	int			low;		/* binary search low index */
+	int			mid;		/* binary search current idx */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+	xfs_trans_t		*tp;		/* transaction pointer */
+	enum xfs_dacmp		cmp;		/* comparison result */
+
+	dp = args->dp;
+	tp = args->trans;
+	mp = dp->i_mount;
+	/*
+	 * Read the buffer, return error if we can't get it.
+	 */
+	if ((error =
+	    xfs_da_read_buf(tp, dp, mp->m_dirdatablk, -1, &bp, XFS_DATA_FORK))) {
+		return error;
+	}
+	ASSERT(bp != NULL);
+	block = bp->data;
+	xfs_dir2_data_check(dp, bp);
+	btp = xfs_dir2_block_tail_p(mp, block);
+	blp = xfs_dir2_block_leaf_p(btp);
+	/*
+	 * Loop doing a binary search for our hash value.
+	 * Find our entry, ENOENT if it's not there.
+	 */
+	for (low = 0, high = be32_to_cpu(btp->count) - 1; ; ) {
+		ASSERT(low <= high);
+		mid = (low + high) >> 1;
+		if ((hash = be32_to_cpu(blp[mid].hashval)) == args->hashval)
+			break;
+		if (hash < args->hashval)
+			low = mid + 1;
+		else
+			high = mid - 1;
+		if (low > high) {
+			ASSERT(args->op_flags & XFS_DA_OP_OKNOENT);
+			xfs_da_brelse(tp, bp);
+			return XFS_ERROR(ENOENT);
+		}
+	}
+	/*
+	 * Back up to the first one with the right hash value.
+	 */
+	while (mid > 0 && be32_to_cpu(blp[mid - 1].hashval) == args->hashval) {
+		mid--;
+	}
+	/*
+	 * Now loop forward through all the entries with the
+	 * right hash value looking for our name.
+	 */
+	do {
+		if ((addr = be32_to_cpu(blp[mid].address)) == XFS_DIR2_NULL_DATAPTR)
+			continue;
+		/*
+		 * Get pointer to the entry from the leaf.
+		 */
+		dep = (xfs_dir2_data_entry_t *)
+			((char *)block + xfs_dir2_dataptr_to_off(mp, addr));
+		/*
+		 * Compare name and if it's an exact match, return the index
+		 * and buffer. If it's the first case-insensitive match, store
+		 * the index and buffer and continue looking for an exact match.
+		 */
+		cmp = mp->m_dirnameops->compname(args, dep->name, dep->namelen);
+		if (cmp != XFS_CMP_DIFFERENT && cmp != args->cmpresult) {
+			args->cmpresult = cmp;
+			*bpp = bp;
+			*entno = mid;
+			if (cmp == XFS_CMP_EXACT)
+				return 0;
+		}
+	} while (++mid < be32_to_cpu(btp->count) &&
+			be32_to_cpu(blp[mid].hashval) == hash);
+
+	ASSERT(args->op_flags & XFS_DA_OP_OKNOENT);
+	/*
+	 * Here, we can only be doing a lookup (not a rename or replace).
+	 * If a case-insensitive match was found earlier, return success.
+	 */
+	if (args->cmpresult == XFS_CMP_CASE)
+		return 0;
+	/*
+	 * No match, release the buffer and return ENOENT.
+	 */
+	xfs_da_brelse(tp, bp);
+	return XFS_ERROR(ENOENT);
+}
+
+/*
+ * Remove an entry from a block format directory.
+ * If that makes the block small enough to fit in shortform, transform it.
+ */
+int						/* error */
+xfs_dir2_block_removename(
+	xfs_da_args_t		*args)		/* directory operation args */
+{
+	xfs_dir2_block_t	*block;		/* block structure */
+	xfs_dir2_leaf_entry_t	*blp;		/* block leaf pointer */
+	xfs_dabuf_t		*bp;		/* block buffer */
+	xfs_dir2_block_tail_t	*btp;		/* block tail */
+	xfs_dir2_data_entry_t	*dep;		/* block data entry */
+	xfs_inode_t		*dp;		/* incore inode */
+	int			ent;		/* block leaf entry index */
+	int			error;		/* error return value */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+	int			needlog;	/* need to log block header */
+	int			needscan;	/* need to fixup bestfree */
+	xfs_dir2_sf_hdr_t	sfh;		/* shortform header */
+	int			size;		/* shortform size */
+	xfs_trans_t		*tp;		/* transaction pointer */
+
+	trace_xfs_dir2_block_removename(args);
+
+	/*
+	 * Look up the entry in the block.  Gets the buffer and entry index.
+	 * It will always be there, the vnodeops level does a lookup first.
+	 */
+	if ((error = xfs_dir2_block_lookup_int(args, &bp, &ent))) {
+		return error;
+	}
+	dp = args->dp;
+	tp = args->trans;
+	mp = dp->i_mount;
+	block = bp->data;
+	btp = xfs_dir2_block_tail_p(mp, block);
+	blp = xfs_dir2_block_leaf_p(btp);
+	/*
+	 * Point to the data entry using the leaf entry.
+	 */
+	dep = (xfs_dir2_data_entry_t *)
+	      ((char *)block + xfs_dir2_dataptr_to_off(mp, be32_to_cpu(blp[ent].address)));
+	/*
+	 * Mark the data entry's space free.
+	 */
+	needlog = needscan = 0;
+	xfs_dir2_data_make_free(tp, bp,
+		(xfs_dir2_data_aoff_t)((char *)dep - (char *)block),
+		xfs_dir2_data_entsize(dep->namelen), &needlog, &needscan);
+	/*
+	 * Fix up the block tail.
+	 */
+	be32_add_cpu(&btp->stale, 1);
+	xfs_dir2_block_log_tail(tp, bp);
+	/*
+	 * Remove the leaf entry by marking it stale.
+	 */
+	blp[ent].address = cpu_to_be32(XFS_DIR2_NULL_DATAPTR);
+	xfs_dir2_block_log_leaf(tp, bp, ent, ent);
+	/*
+	 * Fix up bestfree, log the header if necessary.
+	 */
+	if (needscan)
+		xfs_dir2_data_freescan(mp, (xfs_dir2_data_t *)block, &needlog);
+	if (needlog)
+		xfs_dir2_data_log_header(tp, bp);
+	xfs_dir2_data_check(dp, bp);
+	/*
+	 * See if the size as a shortform is good enough.
+	 */
+	if ((size = xfs_dir2_block_sfsize(dp, block, &sfh)) >
+	    XFS_IFORK_DSIZE(dp)) {
+		xfs_da_buf_done(bp);
+		return 0;
+	}
+	/*
+	 * If it works, do the conversion.
+	 */
+	return xfs_dir2_block_to_sf(args, bp, size, &sfh);
+}
+
+/*
+ * Replace an entry in a V2 block directory.
+ * Change the inode number to the new value.
+ */
+int						/* error */
+xfs_dir2_block_replace(
+	xfs_da_args_t		*args)		/* directory operation args */
+{
+	xfs_dir2_block_t	*block;		/* block structure */
+	xfs_dir2_leaf_entry_t	*blp;		/* block leaf entries */
+	xfs_dabuf_t		*bp;		/* block buffer */
+	xfs_dir2_block_tail_t	*btp;		/* block tail */
+	xfs_dir2_data_entry_t	*dep;		/* block data entry */
+	xfs_inode_t		*dp;		/* incore inode */
+	int			ent;		/* leaf entry index */
+	int			error;		/* error return value */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+
+	trace_xfs_dir2_block_replace(args);
+
+	/*
+	 * Lookup the entry in the directory.  Get buffer and entry index.
+	 * This will always succeed since the caller has already done a lookup.
+	 */
+	if ((error = xfs_dir2_block_lookup_int(args, &bp, &ent))) {
+		return error;
+	}
+	dp = args->dp;
+	mp = dp->i_mount;
+	block = bp->data;
+	btp = xfs_dir2_block_tail_p(mp, block);
+	blp = xfs_dir2_block_leaf_p(btp);
+	/*
+	 * Point to the data entry we need to change.
+	 */
+	dep = (xfs_dir2_data_entry_t *)
+	      ((char *)block + xfs_dir2_dataptr_to_off(mp, be32_to_cpu(blp[ent].address)));
+	ASSERT(be64_to_cpu(dep->inumber) != args->inumber);
+	/*
+	 * Change the inode number to the new value.
+	 */
+	dep->inumber = cpu_to_be64(args->inumber);
+	xfs_dir2_data_log_entry(args->trans, bp, dep);
+	xfs_dir2_data_check(dp, bp);
+	xfs_da_buf_done(bp);
+	return 0;
+}
+
+/*
+ * Qsort comparison routine for the block leaf entries.
+ */
+static int					/* sort order */
+xfs_dir2_block_sort(
+	const void			*a,	/* first leaf entry */
+	const void			*b)	/* second leaf entry */
+{
+	const xfs_dir2_leaf_entry_t	*la;	/* first leaf entry */
+	const xfs_dir2_leaf_entry_t	*lb;	/* second leaf entry */
+
+	la = a;
+	lb = b;
+	return be32_to_cpu(la->hashval) < be32_to_cpu(lb->hashval) ? -1 :
+		(be32_to_cpu(la->hashval) > be32_to_cpu(lb->hashval) ? 1 : 0);
+}
+
+/*
+ * Convert a V2 leaf directory to a V2 block directory if possible.
+ */
+int						/* error */
+xfs_dir2_leaf_to_block(
+	xfs_da_args_t		*args,		/* operation arguments */
+	xfs_dabuf_t		*lbp,		/* leaf buffer */
+	xfs_dabuf_t		*dbp)		/* data buffer */
+{
+	__be16			*bestsp;	/* leaf bests table */
+	xfs_dir2_block_t	*block;		/* block structure */
+	xfs_dir2_block_tail_t	*btp;		/* block tail */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	xfs_dir2_data_unused_t	*dup;		/* unused data entry */
+	int			error;		/* error return value */
+	int			from;		/* leaf from index */
+	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
+	xfs_dir2_leaf_entry_t	*lep;		/* leaf entry */
+	xfs_dir2_leaf_tail_t	*ltp;		/* leaf tail structure */
+	xfs_mount_t		*mp;		/* file system mount point */
+	int			needlog;	/* need to log data header */
+	int			needscan;	/* need to scan for bestfree */
+	xfs_dir2_sf_hdr_t	sfh;		/* shortform header */
+	int			size;		/* bytes used */
+	__be16			*tagp;		/* end of entry (tag) */
+	int			to;		/* block/leaf to index */
+	xfs_trans_t		*tp;		/* transaction pointer */
+
+	trace_xfs_dir2_leaf_to_block(args);
+
+	dp = args->dp;
+	tp = args->trans;
+	mp = dp->i_mount;
+	leaf = lbp->data;
+	ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_DIR2_LEAF1_MAGIC);
+	ltp = xfs_dir2_leaf_tail_p(mp, leaf);
+	/*
+	 * If there are data blocks other than the first one, take this
+	 * opportunity to remove trailing empty data blocks that may have
+	 * been left behind during no-space-reservation operations.
+	 * These will show up in the leaf bests table.
+	 */
+	while (dp->i_d.di_size > mp->m_dirblksize) {
+		bestsp = xfs_dir2_leaf_bests_p(ltp);
+		if (be16_to_cpu(bestsp[be32_to_cpu(ltp->bestcount) - 1]) ==
+		    mp->m_dirblksize - (uint)sizeof(block->hdr)) {
+			if ((error =
+			    xfs_dir2_leaf_trim_data(args, lbp,
+				    (xfs_dir2_db_t)(be32_to_cpu(ltp->bestcount) - 1))))
+				goto out;
+		} else {
+			error = 0;
+			goto out;
+		}
+	}
+	/*
+	 * Read the data block if we don't already have it, give up if it fails.
+	 */
+	if (dbp == NULL &&
+	    (error = xfs_da_read_buf(tp, dp, mp->m_dirdatablk, -1, &dbp,
+		    XFS_DATA_FORK))) {
+		goto out;
+	}
+	block = dbp->data;
+	ASSERT(be32_to_cpu(block->hdr.magic) == XFS_DIR2_DATA_MAGIC);
+	/*
+	 * Size of the "leaf" area in the block.
+	 */
+	size = (uint)sizeof(block->tail) +
+	       (uint)sizeof(*lep) * (be16_to_cpu(leaf->hdr.count) - be16_to_cpu(leaf->hdr.stale));
+	/*
+	 * Look at the last data entry.
+	 */
+	tagp = (__be16 *)((char *)block + mp->m_dirblksize) - 1;
+	dup = (xfs_dir2_data_unused_t *)((char *)block + be16_to_cpu(*tagp));
+	/*
+	 * If it's not free or is too short we can't do it.
+	 */
+	if (be16_to_cpu(dup->freetag) != XFS_DIR2_DATA_FREE_TAG ||
+	    be16_to_cpu(dup->length) < size) {
+		error = 0;
+		goto out;
+	}
+	/*
+	 * Start converting it to block form.
+	 */
+	block->hdr.magic = cpu_to_be32(XFS_DIR2_BLOCK_MAGIC);
+	needlog = 1;
+	needscan = 0;
+	/*
+	 * Use up the space at the end of the block (blp/btp).
+	 */
+	xfs_dir2_data_use_free(tp, dbp, dup, mp->m_dirblksize - size, size,
+		&needlog, &needscan);
+	/*
+	 * Initialize the block tail.
+	 */
+	btp = xfs_dir2_block_tail_p(mp, block);
+	btp->count = cpu_to_be32(be16_to_cpu(leaf->hdr.count) - be16_to_cpu(leaf->hdr.stale));
+	btp->stale = 0;
+	xfs_dir2_block_log_tail(tp, dbp);
+	/*
+	 * Initialize the block leaf area.  We compact out stale entries.
+	 */
+	lep = xfs_dir2_block_leaf_p(btp);
+	for (from = to = 0; from < be16_to_cpu(leaf->hdr.count); from++) {
+		if (be32_to_cpu(leaf->ents[from].address) == XFS_DIR2_NULL_DATAPTR)
+			continue;
+		lep[to++] = leaf->ents[from];
+	}
+	ASSERT(to == be32_to_cpu(btp->count));
+	xfs_dir2_block_log_leaf(tp, dbp, 0, be32_to_cpu(btp->count) - 1);
+	/*
+	 * Scan the bestfree if we need it and log the data block header.
+	 */
+	if (needscan)
+		xfs_dir2_data_freescan(mp, (xfs_dir2_data_t *)block, &needlog);
+	if (needlog)
+		xfs_dir2_data_log_header(tp, dbp);
+	/*
+	 * Pitch the old leaf block.
+	 */
+	error = xfs_da_shrink_inode(args, mp->m_dirleafblk, lbp);
+	lbp = NULL;
+	if (error) {
+		goto out;
+	}
+	/*
+	 * Now see if the resulting block can be shrunken to shortform.
+	 */
+	if ((size = xfs_dir2_block_sfsize(dp, block, &sfh)) >
+	    XFS_IFORK_DSIZE(dp)) {
+		error = 0;
+		goto out;
+	}
+	return xfs_dir2_block_to_sf(args, dbp, size, &sfh);
+out:
+	if (lbp)
+		xfs_da_buf_done(lbp);
+	if (dbp)
+		xfs_da_buf_done(dbp);
+	return error;
+}
+
+/*
+ * Convert the shortform directory to block form.
+ */
+int						/* error */
+xfs_dir2_sf_to_block(
+	xfs_da_args_t		*args)		/* operation arguments */
+{
+	xfs_dir2_db_t		blkno;		/* dir-relative block # (0) */
+	xfs_dir2_block_t	*block;		/* block structure */
+	xfs_dir2_leaf_entry_t	*blp;		/* block leaf entries */
+	xfs_dabuf_t		*bp;		/* block buffer */
+	xfs_dir2_block_tail_t	*btp;		/* block tail pointer */
+	char			*buf;		/* sf buffer */
+	int			buf_len;
+	xfs_dir2_data_entry_t	*dep;		/* data entry pointer */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	int			dummy;		/* trash */
+	xfs_dir2_data_unused_t	*dup;		/* unused entry pointer */
+	int			endoffset;	/* end of data objects */
+	int			error;		/* error return value */
+	int			i;		/* index */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+	int			needlog;	/* need to log block header */
+	int			needscan;	/* need to scan block freespc */
+	int			newoffset;	/* offset from current entry */
+	int			offset;		/* target block offset */
+	xfs_dir2_sf_entry_t	*sfep;		/* sf entry pointer */
+	xfs_dir2_sf_t		*sfp;		/* shortform structure */
+	__be16			*tagp;		/* end of data entry */
+	xfs_trans_t		*tp;		/* transaction pointer */
+	struct xfs_name		name;
+
+	trace_xfs_dir2_sf_to_block(args);
+
+	dp = args->dp;
+	tp = args->trans;
+	mp = dp->i_mount;
+	ASSERT(dp->i_df.if_flags & XFS_IFINLINE);
+	/*
+	 * Bomb out if the shortform directory is way too short.
+	 */
+	if (dp->i_d.di_size < offsetof(xfs_dir2_sf_hdr_t, parent)) {
+		ASSERT(XFS_FORCED_SHUTDOWN(mp));
+		return XFS_ERROR(EIO);
+	}
+	ASSERT(dp->i_df.if_bytes == dp->i_d.di_size);
+	ASSERT(dp->i_df.if_u1.if_data != NULL);
+	sfp = (xfs_dir2_sf_t *)dp->i_df.if_u1.if_data;
+	ASSERT(dp->i_d.di_size >= xfs_dir2_sf_hdr_size(sfp->hdr.i8count));
+	/*
+	 * Copy the directory into the stack buffer.
+	 * Then pitch the incore inode data so we can make extents.
+	 */
+
+	buf_len = dp->i_df.if_bytes;
+	buf = kmem_alloc(buf_len, KM_SLEEP);
+
+	memcpy(buf, sfp, buf_len);
+	xfs_idata_realloc(dp, -buf_len, XFS_DATA_FORK);
+	dp->i_d.di_size = 0;
+	xfs_trans_log_inode(tp, dp, XFS_ILOG_CORE);
+	/*
+	 * Reset pointer - old sfp is gone.
+	 */
+	sfp = (xfs_dir2_sf_t *)buf;
+	/*
+	 * Add block 0 to the inode.
+	 */
+	error = xfs_dir2_grow_inode(args, XFS_DIR2_DATA_SPACE, &blkno);
+	if (error) {
+		kmem_free(buf);
+		return error;
+	}
+	/*
+	 * Initialize the data block.
+	 */
+	error = xfs_dir2_data_init(args, blkno, &bp);
+	if (error) {
+		kmem_free(buf);
+		return error;
+	}
+	block = bp->data;
+	block->hdr.magic = cpu_to_be32(XFS_DIR2_BLOCK_MAGIC);
+	/*
+	 * Compute size of block "tail" area.
+	 */
+	i = (uint)sizeof(*btp) +
+	    (sfp->hdr.count + 2) * (uint)sizeof(xfs_dir2_leaf_entry_t);
+	/*
+	 * The whole thing is initialized to free by the init routine.
+	 * Say we're using the leaf and tail area.
+	 */
+	dup = (xfs_dir2_data_unused_t *)block->u;
+	needlog = needscan = 0;
+	xfs_dir2_data_use_free(tp, bp, dup, mp->m_dirblksize - i, i, &needlog,
+		&needscan);
+	ASSERT(needscan == 0);
+	/*
+	 * Fill in the tail.
+	 */
+	btp = xfs_dir2_block_tail_p(mp, block);
+	btp->count = cpu_to_be32(sfp->hdr.count + 2);	/* ., .. */
+	btp->stale = 0;
+	blp = xfs_dir2_block_leaf_p(btp);
+	endoffset = (uint)((char *)blp - (char *)block);
+	/*
+	 * Remove the freespace, we'll manage it.
+	 */
+	xfs_dir2_data_use_free(tp, bp, dup,
+		(xfs_dir2_data_aoff_t)((char *)dup - (char *)block),
+		be16_to_cpu(dup->length), &needlog, &needscan);
+	/*
+	 * Create entry for .
+	 */
+	dep = (xfs_dir2_data_entry_t *)
+	      ((char *)block + XFS_DIR2_DATA_DOT_OFFSET);
+	dep->inumber = cpu_to_be64(dp->i_ino);
+	dep->namelen = 1;
+	dep->name[0] = '.';
+	tagp = xfs_dir2_data_entry_tag_p(dep);
+	*tagp = cpu_to_be16((char *)dep - (char *)block);
+	xfs_dir2_data_log_entry(tp, bp, dep);
+	blp[0].hashval = cpu_to_be32(xfs_dir_hash_dot);
+	blp[0].address = cpu_to_be32(xfs_dir2_byte_to_dataptr(mp,
+				(char *)dep - (char *)block));
+	/*
+	 * Create entry for ..
+	 */
+	dep = (xfs_dir2_data_entry_t *)
+		((char *)block + XFS_DIR2_DATA_DOTDOT_OFFSET);
+	dep->inumber = cpu_to_be64(xfs_dir2_sf_get_inumber(sfp, &sfp->hdr.parent));
+	dep->namelen = 2;
+	dep->name[0] = dep->name[1] = '.';
+	tagp = xfs_dir2_data_entry_tag_p(dep);
+	*tagp = cpu_to_be16((char *)dep - (char *)block);
+	xfs_dir2_data_log_entry(tp, bp, dep);
+	blp[1].hashval = cpu_to_be32(xfs_dir_hash_dotdot);
+	blp[1].address = cpu_to_be32(xfs_dir2_byte_to_dataptr(mp,
+				(char *)dep - (char *)block));
+	offset = XFS_DIR2_DATA_FIRST_OFFSET;
+	/*
+	 * Loop over existing entries, stuff them in.
+	 */
+	if ((i = 0) == sfp->hdr.count)
+		sfep = NULL;
+	else
+		sfep = xfs_dir2_sf_firstentry(sfp);
+	/*
+	 * Need to preserve the existing offset values in the sf directory.
+	 * Insert holes (unused entries) where necessary.
+	 */
+	while (offset < endoffset) {
+		/*
+		 * sfep is null when we reach the end of the list.
+		 */
+		if (sfep == NULL)
+			newoffset = endoffset;
+		else
+			newoffset = xfs_dir2_sf_get_offset(sfep);
+		/*
+		 * There should be a hole here, make one.
+		 */
+		if (offset < newoffset) {
+			dup = (xfs_dir2_data_unused_t *)
+			      ((char *)block + offset);
+			dup->freetag = cpu_to_be16(XFS_DIR2_DATA_FREE_TAG);
+			dup->length = cpu_to_be16(newoffset - offset);
+			*xfs_dir2_data_unused_tag_p(dup) = cpu_to_be16(
+				((char *)dup - (char *)block));
+			xfs_dir2_data_log_unused(tp, bp, dup);
+			(void)xfs_dir2_data_freeinsert((xfs_dir2_data_t *)block,
+				dup, &dummy);
+			offset += be16_to_cpu(dup->length);
+			continue;
+		}
+		/*
+		 * Copy a real entry.
+		 */
+		dep = (xfs_dir2_data_entry_t *)((char *)block + newoffset);
+		dep->inumber = cpu_to_be64(xfs_dir2_sf_get_inumber(sfp,
+				xfs_dir2_sf_inumberp(sfep)));
+		dep->namelen = sfep->namelen;
+		memcpy(dep->name, sfep->name, dep->namelen);
+		tagp = xfs_dir2_data_entry_tag_p(dep);
+		*tagp = cpu_to_be16((char *)dep - (char *)block);
+		xfs_dir2_data_log_entry(tp, bp, dep);
+		name.name = sfep->name;
+		name.len = sfep->namelen;
+		blp[2 + i].hashval = cpu_to_be32(mp->m_dirnameops->
+							hashname(&name));
+		blp[2 + i].address = cpu_to_be32(xfs_dir2_byte_to_dataptr(mp,
+						 (char *)dep - (char *)block));
+		offset = (int)((char *)(tagp + 1) - (char *)block);
+		if (++i == sfp->hdr.count)
+			sfep = NULL;
+		else
+			sfep = xfs_dir2_sf_nextentry(sfp, sfep);
+	}
+	/* Done with the temporary buffer */
+	kmem_free(buf);
+	/*
+	 * Sort the leaf entries by hash value.
+	 */
+	xfs_sort(blp, be32_to_cpu(btp->count), sizeof(*blp), xfs_dir2_block_sort);
+	/*
+	 * Log the leaf entry area and tail.
+	 * Already logged the header in data_init, ignore needlog.
+	 */
+	ASSERT(needscan == 0);
+	xfs_dir2_block_log_leaf(tp, bp, 0, be32_to_cpu(btp->count) - 1);
+	xfs_dir2_block_log_tail(tp, bp);
+	xfs_dir2_data_check(dp, bp);
+	xfs_da_buf_done(bp);
+	return 0;
+}
diff --git a/fs/xfs/xfs_dir2_block.h b/fs/xfs/xfs_dir2_block.h
new file mode 100644
index 00000000..10e68967
--- /dev/null
+++ b/fs/xfs/xfs_dir2_block.h
@@ -0,0 +1,92 @@
+/*
+ * Copyright (c) 2000-2001,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_DIR2_BLOCK_H__
+#define	__XFS_DIR2_BLOCK_H__
+
+/*
+ * xfs_dir2_block.h
+ * Directory version 2, single block format structures
+ */
+
+struct uio;
+struct xfs_dabuf;
+struct xfs_da_args;
+struct xfs_dir2_data_hdr;
+struct xfs_dir2_leaf_entry;
+struct xfs_inode;
+struct xfs_mount;
+struct xfs_trans;
+
+/*
+ * The single block format is as follows:
+ * xfs_dir2_data_hdr_t structure
+ * xfs_dir2_data_entry_t and xfs_dir2_data_unused_t structures
+ * xfs_dir2_leaf_entry_t structures
+ * xfs_dir2_block_tail_t structure
+ */
+
+#define	XFS_DIR2_BLOCK_MAGIC	0x58443242	/* XD2B: for one block dirs */
+
+typedef struct xfs_dir2_block_tail {
+	__be32		count;			/* count of leaf entries */
+	__be32		stale;			/* count of stale lf entries */
+} xfs_dir2_block_tail_t;
+
+/*
+ * Generic single-block structure, for xfs_db.
+ */
+typedef struct xfs_dir2_block {
+	xfs_dir2_data_hdr_t	hdr;		/* magic XFS_DIR2_BLOCK_MAGIC */
+	xfs_dir2_data_union_t	u[1];
+	xfs_dir2_leaf_entry_t	leaf[1];
+	xfs_dir2_block_tail_t	tail;
+} xfs_dir2_block_t;
+
+/*
+ * Pointer to the leaf header embedded in a data block (1-block format)
+ */
+static inline xfs_dir2_block_tail_t *
+xfs_dir2_block_tail_p(struct xfs_mount *mp, xfs_dir2_block_t *block)
+{
+	return (((xfs_dir2_block_tail_t *)
+		((char *)(block) + (mp)->m_dirblksize)) - 1);
+}
+
+/*
+ * Pointer to the leaf entries embedded in a data block (1-block format)
+ */
+static inline struct xfs_dir2_leaf_entry *
+xfs_dir2_block_leaf_p(xfs_dir2_block_tail_t *btp)
+{
+	return ((struct xfs_dir2_leaf_entry *)btp) - be32_to_cpu(btp->count);
+}
+
+/*
+ * Function declarations.
+ */
+extern int xfs_dir2_block_addname(struct xfs_da_args *args);
+extern int xfs_dir2_block_getdents(struct xfs_inode *dp, void *dirent,
+				   xfs_off_t *offset, filldir_t filldir);
+extern int xfs_dir2_block_lookup(struct xfs_da_args *args);
+extern int xfs_dir2_block_removename(struct xfs_da_args *args);
+extern int xfs_dir2_block_replace(struct xfs_da_args *args);
+extern int xfs_dir2_leaf_to_block(struct xfs_da_args *args,
+				  struct xfs_dabuf *lbp, struct xfs_dabuf *dbp);
+extern int xfs_dir2_sf_to_block(struct xfs_da_args *args);
+
+#endif	/* __XFS_DIR2_BLOCK_H__ */
diff --git a/fs/xfs/xfs_dir2_data.c b/fs/xfs/xfs_dir2_data.c
new file mode 100644
index 00000000..921595b8
--- /dev/null
+++ b/fs/xfs/xfs_dir2_data.c
@@ -0,0 +1,833 @@
+/*
+ * Copyright (c) 2000-2002,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_dir2.h"
+#include "xfs_mount.h"
+#include "xfs_da_btree.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_dir2_sf.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_dir2_data.h"
+#include "xfs_dir2_leaf.h"
+#include "xfs_dir2_block.h"
+#include "xfs_error.h"
+
+#ifdef DEBUG
+/*
+ * Check the consistency of the data block.
+ * The input can also be a block-format directory.
+ * Pop an assert if we find anything bad.
+ */
+void
+xfs_dir2_data_check(
+	xfs_inode_t		*dp,		/* incore inode pointer */
+	xfs_dabuf_t		*bp)		/* data block's buffer */
+{
+	xfs_dir2_dataptr_t	addr;		/* addr for leaf lookup */
+	xfs_dir2_data_free_t	*bf;		/* bestfree table */
+	xfs_dir2_block_tail_t	*btp=NULL;	/* block tail */
+	int			count;		/* count of entries found */
+	xfs_dir2_data_t		*d;		/* data block pointer */
+	xfs_dir2_data_entry_t	*dep;		/* data entry */
+	xfs_dir2_data_free_t	*dfp;		/* bestfree entry */
+	xfs_dir2_data_unused_t	*dup;		/* unused entry */
+	char			*endp;		/* end of useful data */
+	int			freeseen;	/* mask of bestfrees seen */
+	xfs_dahash_t		hash;		/* hash of current name */
+	int			i;		/* leaf index */
+	int			lastfree;	/* last entry was unused */
+	xfs_dir2_leaf_entry_t	*lep=NULL;	/* block leaf entries */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+	char			*p;		/* current data position */
+	int			stale;		/* count of stale leaves */
+	struct xfs_name		name;
+
+	mp = dp->i_mount;
+	d = bp->data;
+	ASSERT(be32_to_cpu(d->hdr.magic) == XFS_DIR2_DATA_MAGIC ||
+	       be32_to_cpu(d->hdr.magic) == XFS_DIR2_BLOCK_MAGIC);
+	bf = d->hdr.bestfree;
+	p = (char *)d->u;
+	if (be32_to_cpu(d->hdr.magic) == XFS_DIR2_BLOCK_MAGIC) {
+		btp = xfs_dir2_block_tail_p(mp, (xfs_dir2_block_t *)d);
+		lep = xfs_dir2_block_leaf_p(btp);
+		endp = (char *)lep;
+	} else
+		endp = (char *)d + mp->m_dirblksize;
+	count = lastfree = freeseen = 0;
+	/*
+	 * Account for zero bestfree entries.
+	 */
+	if (!bf[0].length) {
+		ASSERT(!bf[0].offset);
+		freeseen |= 1 << 0;
+	}
+	if (!bf[1].length) {
+		ASSERT(!bf[1].offset);
+		freeseen |= 1 << 1;
+	}
+	if (!bf[2].length) {
+		ASSERT(!bf[2].offset);
+		freeseen |= 1 << 2;
+	}
+	ASSERT(be16_to_cpu(bf[0].length) >= be16_to_cpu(bf[1].length));
+	ASSERT(be16_to_cpu(bf[1].length) >= be16_to_cpu(bf[2].length));
+	/*
+	 * Loop over the data/unused entries.
+	 */
+	while (p < endp) {
+		dup = (xfs_dir2_data_unused_t *)p;
+		/*
+		 * If it's unused, look for the space in the bestfree table.
+		 * If we find it, account for that, else make sure it
+		 * doesn't need to be there.
+		 */
+		if (be16_to_cpu(dup->freetag) == XFS_DIR2_DATA_FREE_TAG) {
+			ASSERT(lastfree == 0);
+			ASSERT(be16_to_cpu(*xfs_dir2_data_unused_tag_p(dup)) ==
+			       (char *)dup - (char *)d);
+			dfp = xfs_dir2_data_freefind(d, dup);
+			if (dfp) {
+				i = (int)(dfp - bf);
+				ASSERT((freeseen & (1 << i)) == 0);
+				freeseen |= 1 << i;
+			} else {
+				ASSERT(be16_to_cpu(dup->length) <=
+				       be16_to_cpu(bf[2].length));
+			}
+			p += be16_to_cpu(dup->length);
+			lastfree = 1;
+			continue;
+		}
+		/*
+		 * It's a real entry.  Validate the fields.
+		 * If this is a block directory then make sure it's
+		 * in the leaf section of the block.
+		 * The linear search is crude but this is DEBUG code.
+		 */
+		dep = (xfs_dir2_data_entry_t *)p;
+		ASSERT(dep->namelen != 0);
+		ASSERT(xfs_dir_ino_validate(mp, be64_to_cpu(dep->inumber)) == 0);
+		ASSERT(be16_to_cpu(*xfs_dir2_data_entry_tag_p(dep)) ==
+		       (char *)dep - (char *)d);
+		count++;
+		lastfree = 0;
+		if (be32_to_cpu(d->hdr.magic) == XFS_DIR2_BLOCK_MAGIC) {
+			addr = xfs_dir2_db_off_to_dataptr(mp, mp->m_dirdatablk,
+				(xfs_dir2_data_aoff_t)
+				((char *)dep - (char *)d));
+			name.name = dep->name;
+			name.len = dep->namelen;
+			hash = mp->m_dirnameops->hashname(&name);
+			for (i = 0; i < be32_to_cpu(btp->count); i++) {
+				if (be32_to_cpu(lep[i].address) == addr &&
+				    be32_to_cpu(lep[i].hashval) == hash)
+					break;
+			}
+			ASSERT(i < be32_to_cpu(btp->count));
+		}
+		p += xfs_dir2_data_entsize(dep->namelen);
+	}
+	/*
+	 * Need to have seen all the entries and all the bestfree slots.
+	 */
+	ASSERT(freeseen == 7);
+	if (be32_to_cpu(d->hdr.magic) == XFS_DIR2_BLOCK_MAGIC) {
+		for (i = stale = 0; i < be32_to_cpu(btp->count); i++) {
+			if (be32_to_cpu(lep[i].address) == XFS_DIR2_NULL_DATAPTR)
+				stale++;
+			if (i > 0)
+				ASSERT(be32_to_cpu(lep[i].hashval) >= be32_to_cpu(lep[i - 1].hashval));
+		}
+		ASSERT(count == be32_to_cpu(btp->count) - be32_to_cpu(btp->stale));
+		ASSERT(stale == be32_to_cpu(btp->stale));
+	}
+}
+#endif
+
+/*
+ * Given a data block and an unused entry from that block,
+ * return the bestfree entry if any that corresponds to it.
+ */
+xfs_dir2_data_free_t *
+xfs_dir2_data_freefind(
+	xfs_dir2_data_t		*d,		/* data block */
+	xfs_dir2_data_unused_t	*dup)		/* data unused entry */
+{
+	xfs_dir2_data_free_t	*dfp;		/* bestfree entry */
+	xfs_dir2_data_aoff_t	off;		/* offset value needed */
+#if defined(DEBUG) && defined(__KERNEL__)
+	int			matched;	/* matched the value */
+	int			seenzero;	/* saw a 0 bestfree entry */
+#endif
+
+	off = (xfs_dir2_data_aoff_t)((char *)dup - (char *)d);
+#if defined(DEBUG) && defined(__KERNEL__)
+	/*
+	 * Validate some consistency in the bestfree table.
+	 * Check order, non-overlapping entries, and if we find the
+	 * one we're looking for it has to be exact.
+	 */
+	ASSERT(be32_to_cpu(d->hdr.magic) == XFS_DIR2_DATA_MAGIC ||
+	       be32_to_cpu(d->hdr.magic) == XFS_DIR2_BLOCK_MAGIC);
+	for (dfp = &d->hdr.bestfree[0], seenzero = matched = 0;
+	     dfp < &d->hdr.bestfree[XFS_DIR2_DATA_FD_COUNT];
+	     dfp++) {
+		if (!dfp->offset) {
+			ASSERT(!dfp->length);
+			seenzero = 1;
+			continue;
+		}
+		ASSERT(seenzero == 0);
+		if (be16_to_cpu(dfp->offset) == off) {
+			matched = 1;
+			ASSERT(dfp->length == dup->length);
+		} else if (off < be16_to_cpu(dfp->offset))
+			ASSERT(off + be16_to_cpu(dup->length) <= be16_to_cpu(dfp->offset));
+		else
+			ASSERT(be16_to_cpu(dfp->offset) + be16_to_cpu(dfp->length) <= off);
+		ASSERT(matched || be16_to_cpu(dfp->length) >= be16_to_cpu(dup->length));
+		if (dfp > &d->hdr.bestfree[0])
+			ASSERT(be16_to_cpu(dfp[-1].length) >= be16_to_cpu(dfp[0].length));
+	}
+#endif
+	/*
+	 * If this is smaller than the smallest bestfree entry,
+	 * it can't be there since they're sorted.
+	 */
+	if (be16_to_cpu(dup->length) <
+	    be16_to_cpu(d->hdr.bestfree[XFS_DIR2_DATA_FD_COUNT - 1].length))
+		return NULL;
+	/*
+	 * Look at the three bestfree entries for our guy.
+	 */
+	for (dfp = &d->hdr.bestfree[0];
+	     dfp < &d->hdr.bestfree[XFS_DIR2_DATA_FD_COUNT];
+	     dfp++) {
+		if (!dfp->offset)
+			return NULL;
+		if (be16_to_cpu(dfp->offset) == off)
+			return dfp;
+	}
+	/*
+	 * Didn't find it.  This only happens if there are duplicate lengths.
+	 */
+	return NULL;
+}
+
+/*
+ * Insert an unused-space entry into the bestfree table.
+ */
+xfs_dir2_data_free_t *				/* entry inserted */
+xfs_dir2_data_freeinsert(
+	xfs_dir2_data_t		*d,		/* data block pointer */
+	xfs_dir2_data_unused_t	*dup,		/* unused space */
+	int			*loghead)	/* log the data header (out) */
+{
+	xfs_dir2_data_free_t	*dfp;		/* bestfree table pointer */
+	xfs_dir2_data_free_t	new;		/* new bestfree entry */
+
+#ifdef __KERNEL__
+	ASSERT(be32_to_cpu(d->hdr.magic) == XFS_DIR2_DATA_MAGIC ||
+	       be32_to_cpu(d->hdr.magic) == XFS_DIR2_BLOCK_MAGIC);
+#endif
+	dfp = d->hdr.bestfree;
+	new.length = dup->length;
+	new.offset = cpu_to_be16((char *)dup - (char *)d);
+	/*
+	 * Insert at position 0, 1, or 2; or not at all.
+	 */
+	if (be16_to_cpu(new.length) > be16_to_cpu(dfp[0].length)) {
+		dfp[2] = dfp[1];
+		dfp[1] = dfp[0];
+		dfp[0] = new;
+		*loghead = 1;
+		return &dfp[0];
+	}
+	if (be16_to_cpu(new.length) > be16_to_cpu(dfp[1].length)) {
+		dfp[2] = dfp[1];
+		dfp[1] = new;
+		*loghead = 1;
+		return &dfp[1];
+	}
+	if (be16_to_cpu(new.length) > be16_to_cpu(dfp[2].length)) {
+		dfp[2] = new;
+		*loghead = 1;
+		return &dfp[2];
+	}
+	return NULL;
+}
+
+/*
+ * Remove a bestfree entry from the table.
+ */
+STATIC void
+xfs_dir2_data_freeremove(
+	xfs_dir2_data_t		*d,		/* data block pointer */
+	xfs_dir2_data_free_t	*dfp,		/* bestfree entry pointer */
+	int			*loghead)	/* out: log data header */
+{
+#ifdef __KERNEL__
+	ASSERT(be32_to_cpu(d->hdr.magic) == XFS_DIR2_DATA_MAGIC ||
+	       be32_to_cpu(d->hdr.magic) == XFS_DIR2_BLOCK_MAGIC);
+#endif
+	/*
+	 * It's the first entry, slide the next 2 up.
+	 */
+	if (dfp == &d->hdr.bestfree[0]) {
+		d->hdr.bestfree[0] = d->hdr.bestfree[1];
+		d->hdr.bestfree[1] = d->hdr.bestfree[2];
+	}
+	/*
+	 * It's the second entry, slide the 3rd entry up.
+	 */
+	else if (dfp == &d->hdr.bestfree[1])
+		d->hdr.bestfree[1] = d->hdr.bestfree[2];
+	/*
+	 * Must be the last entry.
+	 */
+	else
+		ASSERT(dfp == &d->hdr.bestfree[2]);
+	/*
+	 * Clear the 3rd entry, must be zero now.
+	 */
+	d->hdr.bestfree[2].length = 0;
+	d->hdr.bestfree[2].offset = 0;
+	*loghead = 1;
+}
+
+/*
+ * Given a data block, reconstruct its bestfree map.
+ */
+void
+xfs_dir2_data_freescan(
+	xfs_mount_t		*mp,		/* filesystem mount point */
+	xfs_dir2_data_t		*d,		/* data block pointer */
+	int			*loghead)	/* out: log data header */
+{
+	xfs_dir2_block_tail_t	*btp;		/* block tail */
+	xfs_dir2_data_entry_t	*dep;		/* active data entry */
+	xfs_dir2_data_unused_t	*dup;		/* unused data entry */
+	char			*endp;		/* end of block's data */
+	char			*p;		/* current entry pointer */
+
+#ifdef __KERNEL__
+	ASSERT(be32_to_cpu(d->hdr.magic) == XFS_DIR2_DATA_MAGIC ||
+	       be32_to_cpu(d->hdr.magic) == XFS_DIR2_BLOCK_MAGIC);
+#endif
+	/*
+	 * Start by clearing the table.
+	 */
+	memset(d->hdr.bestfree, 0, sizeof(d->hdr.bestfree));
+	*loghead = 1;
+	/*
+	 * Set up pointers.
+	 */
+	p = (char *)d->u;
+	if (be32_to_cpu(d->hdr.magic) == XFS_DIR2_BLOCK_MAGIC) {
+		btp = xfs_dir2_block_tail_p(mp, (xfs_dir2_block_t *)d);
+		endp = (char *)xfs_dir2_block_leaf_p(btp);
+	} else
+		endp = (char *)d + mp->m_dirblksize;
+	/*
+	 * Loop over the block's entries.
+	 */
+	while (p < endp) {
+		dup = (xfs_dir2_data_unused_t *)p;
+		/*
+		 * If it's a free entry, insert it.
+		 */
+		if (be16_to_cpu(dup->freetag) == XFS_DIR2_DATA_FREE_TAG) {
+			ASSERT((char *)dup - (char *)d ==
+			       be16_to_cpu(*xfs_dir2_data_unused_tag_p(dup)));
+			xfs_dir2_data_freeinsert(d, dup, loghead);
+			p += be16_to_cpu(dup->length);
+		}
+		/*
+		 * For active entries, check their tags and skip them.
+		 */
+		else {
+			dep = (xfs_dir2_data_entry_t *)p;
+			ASSERT((char *)dep - (char *)d ==
+			       be16_to_cpu(*xfs_dir2_data_entry_tag_p(dep)));
+			p += xfs_dir2_data_entsize(dep->namelen);
+		}
+	}
+}
+
+/*
+ * Initialize a data block at the given block number in the directory.
+ * Give back the buffer for the created block.
+ */
+int						/* error */
+xfs_dir2_data_init(
+	xfs_da_args_t		*args,		/* directory operation args */
+	xfs_dir2_db_t		blkno,		/* logical dir block number */
+	xfs_dabuf_t		**bpp)		/* output block buffer */
+{
+	xfs_dabuf_t		*bp;		/* block buffer */
+	xfs_dir2_data_t		*d;		/* pointer to block */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	xfs_dir2_data_unused_t	*dup;		/* unused entry pointer */
+	int			error;		/* error return value */
+	int			i;		/* bestfree index */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+	xfs_trans_t		*tp;		/* transaction pointer */
+	int                     t;              /* temp */
+
+	dp = args->dp;
+	mp = dp->i_mount;
+	tp = args->trans;
+	/*
+	 * Get the buffer set up for the block.
+	 */
+	error = xfs_da_get_buf(tp, dp, xfs_dir2_db_to_da(mp, blkno), -1, &bp,
+		XFS_DATA_FORK);
+	if (error) {
+		return error;
+	}
+	ASSERT(bp != NULL);
+	/*
+	 * Initialize the header.
+	 */
+	d = bp->data;
+	d->hdr.magic = cpu_to_be32(XFS_DIR2_DATA_MAGIC);
+	d->hdr.bestfree[0].offset = cpu_to_be16(sizeof(d->hdr));
+	for (i = 1; i < XFS_DIR2_DATA_FD_COUNT; i++) {
+		d->hdr.bestfree[i].length = 0;
+		d->hdr.bestfree[i].offset = 0;
+	}
+	/*
+	 * Set up an unused entry for the block's body.
+	 */
+	dup = &d->u[0].unused;
+	dup->freetag = cpu_to_be16(XFS_DIR2_DATA_FREE_TAG);
+
+	t=mp->m_dirblksize - (uint)sizeof(d->hdr);
+	d->hdr.bestfree[0].length = cpu_to_be16(t);
+	dup->length = cpu_to_be16(t);
+	*xfs_dir2_data_unused_tag_p(dup) = cpu_to_be16((char *)dup - (char *)d);
+	/*
+	 * Log it and return it.
+	 */
+	xfs_dir2_data_log_header(tp, bp);
+	xfs_dir2_data_log_unused(tp, bp, dup);
+	*bpp = bp;
+	return 0;
+}
+
+/*
+ * Log an active data entry from the block.
+ */
+void
+xfs_dir2_data_log_entry(
+	xfs_trans_t		*tp,		/* transaction pointer */
+	xfs_dabuf_t		*bp,		/* block buffer */
+	xfs_dir2_data_entry_t	*dep)		/* data entry pointer */
+{
+	xfs_dir2_data_t		*d;		/* data block pointer */
+
+	d = bp->data;
+	ASSERT(be32_to_cpu(d->hdr.magic) == XFS_DIR2_DATA_MAGIC ||
+	       be32_to_cpu(d->hdr.magic) == XFS_DIR2_BLOCK_MAGIC);
+	xfs_da_log_buf(tp, bp, (uint)((char *)dep - (char *)d),
+		(uint)((char *)(xfs_dir2_data_entry_tag_p(dep) + 1) -
+		       (char *)d - 1));
+}
+
+/*
+ * Log a data block header.
+ */
+void
+xfs_dir2_data_log_header(
+	xfs_trans_t		*tp,		/* transaction pointer */
+	xfs_dabuf_t		*bp)		/* block buffer */
+{
+	xfs_dir2_data_t		*d;		/* data block pointer */
+
+	d = bp->data;
+	ASSERT(be32_to_cpu(d->hdr.magic) == XFS_DIR2_DATA_MAGIC ||
+	       be32_to_cpu(d->hdr.magic) == XFS_DIR2_BLOCK_MAGIC);
+	xfs_da_log_buf(tp, bp, (uint)((char *)&d->hdr - (char *)d),
+		(uint)(sizeof(d->hdr) - 1));
+}
+
+/*
+ * Log a data unused entry.
+ */
+void
+xfs_dir2_data_log_unused(
+	xfs_trans_t		*tp,		/* transaction pointer */
+	xfs_dabuf_t		*bp,		/* block buffer */
+	xfs_dir2_data_unused_t	*dup)		/* data unused pointer */
+{
+	xfs_dir2_data_t		*d;		/* data block pointer */
+
+	d = bp->data;
+	ASSERT(be32_to_cpu(d->hdr.magic) == XFS_DIR2_DATA_MAGIC ||
+	       be32_to_cpu(d->hdr.magic) == XFS_DIR2_BLOCK_MAGIC);
+	/*
+	 * Log the first part of the unused entry.
+	 */
+	xfs_da_log_buf(tp, bp, (uint)((char *)dup - (char *)d),
+		(uint)((char *)&dup->length + sizeof(dup->length) -
+		       1 - (char *)d));
+	/*
+	 * Log the end (tag) of the unused entry.
+	 */
+	xfs_da_log_buf(tp, bp,
+		(uint)((char *)xfs_dir2_data_unused_tag_p(dup) - (char *)d),
+		(uint)((char *)xfs_dir2_data_unused_tag_p(dup) - (char *)d +
+		       sizeof(xfs_dir2_data_off_t) - 1));
+}
+
+/*
+ * Make a byte range in the data block unused.
+ * Its current contents are unimportant.
+ */
+void
+xfs_dir2_data_make_free(
+	xfs_trans_t		*tp,		/* transaction pointer */
+	xfs_dabuf_t		*bp,		/* block buffer */
+	xfs_dir2_data_aoff_t	offset,		/* starting byte offset */
+	xfs_dir2_data_aoff_t	len,		/* length in bytes */
+	int			*needlogp,	/* out: log header */
+	int			*needscanp)	/* out: regen bestfree */
+{
+	xfs_dir2_data_t		*d;		/* data block pointer */
+	xfs_dir2_data_free_t	*dfp;		/* bestfree pointer */
+	char			*endptr;	/* end of data area */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+	int			needscan;	/* need to regen bestfree */
+	xfs_dir2_data_unused_t	*newdup;	/* new unused entry */
+	xfs_dir2_data_unused_t	*postdup;	/* unused entry after us */
+	xfs_dir2_data_unused_t	*prevdup;	/* unused entry before us */
+
+	mp = tp->t_mountp;
+	d = bp->data;
+	/*
+	 * Figure out where the end of the data area is.
+	 */
+	if (be32_to_cpu(d->hdr.magic) == XFS_DIR2_DATA_MAGIC)
+		endptr = (char *)d + mp->m_dirblksize;
+	else {
+		xfs_dir2_block_tail_t	*btp;	/* block tail */
+
+		ASSERT(be32_to_cpu(d->hdr.magic) == XFS_DIR2_BLOCK_MAGIC);
+		btp = xfs_dir2_block_tail_p(mp, (xfs_dir2_block_t *)d);
+		endptr = (char *)xfs_dir2_block_leaf_p(btp);
+	}
+	/*
+	 * If this isn't the start of the block, then back up to
+	 * the previous entry and see if it's free.
+	 */
+	if (offset > sizeof(d->hdr)) {
+		__be16			*tagp;	/* tag just before us */
+
+		tagp = (__be16 *)((char *)d + offset) - 1;
+		prevdup = (xfs_dir2_data_unused_t *)((char *)d + be16_to_cpu(*tagp));
+		if (be16_to_cpu(prevdup->freetag) != XFS_DIR2_DATA_FREE_TAG)
+			prevdup = NULL;
+	} else
+		prevdup = NULL;
+	/*
+	 * If this isn't the end of the block, see if the entry after
+	 * us is free.
+	 */
+	if ((char *)d + offset + len < endptr) {
+		postdup =
+			(xfs_dir2_data_unused_t *)((char *)d + offset + len);
+		if (be16_to_cpu(postdup->freetag) != XFS_DIR2_DATA_FREE_TAG)
+			postdup = NULL;
+	} else
+		postdup = NULL;
+	ASSERT(*needscanp == 0);
+	needscan = 0;
+	/*
+	 * Previous and following entries are both free,
+	 * merge everything into a single free entry.
+	 */
+	if (prevdup && postdup) {
+		xfs_dir2_data_free_t	*dfp2;	/* another bestfree pointer */
+
+		/*
+		 * See if prevdup and/or postdup are in bestfree table.
+		 */
+		dfp = xfs_dir2_data_freefind(d, prevdup);
+		dfp2 = xfs_dir2_data_freefind(d, postdup);
+		/*
+		 * We need a rescan unless there are exactly 2 free entries
+		 * namely our two.  Then we know what's happening, otherwise
+		 * since the third bestfree is there, there might be more
+		 * entries.
+		 */
+		needscan = (d->hdr.bestfree[2].length != 0);
+		/*
+		 * Fix up the new big freespace.
+		 */
+		be16_add_cpu(&prevdup->length, len + be16_to_cpu(postdup->length));
+		*xfs_dir2_data_unused_tag_p(prevdup) =
+			cpu_to_be16((char *)prevdup - (char *)d);
+		xfs_dir2_data_log_unused(tp, bp, prevdup);
+		if (!needscan) {
+			/*
+			 * Has to be the case that entries 0 and 1 are
+			 * dfp and dfp2 (don't know which is which), and
+			 * entry 2 is empty.
+			 * Remove entry 1 first then entry 0.
+			 */
+			ASSERT(dfp && dfp2);
+			if (dfp == &d->hdr.bestfree[1]) {
+				dfp = &d->hdr.bestfree[0];
+				ASSERT(dfp2 == dfp);
+				dfp2 = &d->hdr.bestfree[1];
+			}
+			xfs_dir2_data_freeremove(d, dfp2, needlogp);
+			xfs_dir2_data_freeremove(d, dfp, needlogp);
+			/*
+			 * Now insert the new entry.
+			 */
+			dfp = xfs_dir2_data_freeinsert(d, prevdup, needlogp);
+			ASSERT(dfp == &d->hdr.bestfree[0]);
+			ASSERT(dfp->length == prevdup->length);
+			ASSERT(!dfp[1].length);
+			ASSERT(!dfp[2].length);
+		}
+	}
+	/*
+	 * The entry before us is free, merge with it.
+	 */
+	else if (prevdup) {
+		dfp = xfs_dir2_data_freefind(d, prevdup);
+		be16_add_cpu(&prevdup->length, len);
+		*xfs_dir2_data_unused_tag_p(prevdup) =
+			cpu_to_be16((char *)prevdup - (char *)d);
+		xfs_dir2_data_log_unused(tp, bp, prevdup);
+		/*
+		 * If the previous entry was in the table, the new entry
+		 * is longer, so it will be in the table too.  Remove
+		 * the old one and add the new one.
+		 */
+		if (dfp) {
+			xfs_dir2_data_freeremove(d, dfp, needlogp);
+			(void)xfs_dir2_data_freeinsert(d, prevdup, needlogp);
+		}
+		/*
+		 * Otherwise we need a scan if the new entry is big enough.
+		 */
+		else {
+			needscan = be16_to_cpu(prevdup->length) >
+				   be16_to_cpu(d->hdr.bestfree[2].length);
+		}
+	}
+	/*
+	 * The following entry is free, merge with it.
+	 */
+	else if (postdup) {
+		dfp = xfs_dir2_data_freefind(d, postdup);
+		newdup = (xfs_dir2_data_unused_t *)((char *)d + offset);
+		newdup->freetag = cpu_to_be16(XFS_DIR2_DATA_FREE_TAG);
+		newdup->length = cpu_to_be16(len + be16_to_cpu(postdup->length));
+		*xfs_dir2_data_unused_tag_p(newdup) =
+			cpu_to_be16((char *)newdup - (char *)d);
+		xfs_dir2_data_log_unused(tp, bp, newdup);
+		/*
+		 * If the following entry was in the table, the new entry
+		 * is longer, so it will be in the table too.  Remove
+		 * the old one and add the new one.
+		 */
+		if (dfp) {
+			xfs_dir2_data_freeremove(d, dfp, needlogp);
+			(void)xfs_dir2_data_freeinsert(d, newdup, needlogp);
+		}
+		/*
+		 * Otherwise we need a scan if the new entry is big enough.
+		 */
+		else {
+			needscan = be16_to_cpu(newdup->length) >
+				   be16_to_cpu(d->hdr.bestfree[2].length);
+		}
+	}
+	/*
+	 * Neither neighbor is free.  Make a new entry.
+	 */
+	else {
+		newdup = (xfs_dir2_data_unused_t *)((char *)d + offset);
+		newdup->freetag = cpu_to_be16(XFS_DIR2_DATA_FREE_TAG);
+		newdup->length = cpu_to_be16(len);
+		*xfs_dir2_data_unused_tag_p(newdup) =
+			cpu_to_be16((char *)newdup - (char *)d);
+		xfs_dir2_data_log_unused(tp, bp, newdup);
+		(void)xfs_dir2_data_freeinsert(d, newdup, needlogp);
+	}
+	*needscanp = needscan;
+}
+
+/*
+ * Take a byte range out of an existing unused space and make it un-free.
+ */
+void
+xfs_dir2_data_use_free(
+	xfs_trans_t		*tp,		/* transaction pointer */
+	xfs_dabuf_t		*bp,		/* data block buffer */
+	xfs_dir2_data_unused_t	*dup,		/* unused entry */
+	xfs_dir2_data_aoff_t	offset,		/* starting offset to use */
+	xfs_dir2_data_aoff_t	len,		/* length to use */
+	int			*needlogp,	/* out: need to log header */
+	int			*needscanp)	/* out: need regen bestfree */
+{
+	xfs_dir2_data_t		*d;		/* data block */
+	xfs_dir2_data_free_t	*dfp;		/* bestfree pointer */
+	int			matchback;	/* matches end of freespace */
+	int			matchfront;	/* matches start of freespace */
+	int			needscan;	/* need to regen bestfree */
+	xfs_dir2_data_unused_t	*newdup;	/* new unused entry */
+	xfs_dir2_data_unused_t	*newdup2;	/* another new unused entry */
+	int			oldlen;		/* old unused entry's length */
+
+	d = bp->data;
+	ASSERT(be32_to_cpu(d->hdr.magic) == XFS_DIR2_DATA_MAGIC ||
+	       be32_to_cpu(d->hdr.magic) == XFS_DIR2_BLOCK_MAGIC);
+	ASSERT(be16_to_cpu(dup->freetag) == XFS_DIR2_DATA_FREE_TAG);
+	ASSERT(offset >= (char *)dup - (char *)d);
+	ASSERT(offset + len <= (char *)dup + be16_to_cpu(dup->length) - (char *)d);
+	ASSERT((char *)dup - (char *)d == be16_to_cpu(*xfs_dir2_data_unused_tag_p(dup)));
+	/*
+	 * Look up the entry in the bestfree table.
+	 */
+	dfp = xfs_dir2_data_freefind(d, dup);
+	oldlen = be16_to_cpu(dup->length);
+	ASSERT(dfp || oldlen <= be16_to_cpu(d->hdr.bestfree[2].length));
+	/*
+	 * Check for alignment with front and back of the entry.
+	 */
+	matchfront = (char *)dup - (char *)d == offset;
+	matchback = (char *)dup + oldlen - (char *)d == offset + len;
+	ASSERT(*needscanp == 0);
+	needscan = 0;
+	/*
+	 * If we matched it exactly we just need to get rid of it from
+	 * the bestfree table.
+	 */
+	if (matchfront && matchback) {
+		if (dfp) {
+			needscan = (d->hdr.bestfree[2].offset != 0);
+			if (!needscan)
+				xfs_dir2_data_freeremove(d, dfp, needlogp);
+		}
+	}
+	/*
+	 * We match the first part of the entry.
+	 * Make a new entry with the remaining freespace.
+	 */
+	else if (matchfront) {
+		newdup = (xfs_dir2_data_unused_t *)((char *)d + offset + len);
+		newdup->freetag = cpu_to_be16(XFS_DIR2_DATA_FREE_TAG);
+		newdup->length = cpu_to_be16(oldlen - len);
+		*xfs_dir2_data_unused_tag_p(newdup) =
+			cpu_to_be16((char *)newdup - (char *)d);
+		xfs_dir2_data_log_unused(tp, bp, newdup);
+		/*
+		 * If it was in the table, remove it and add the new one.
+		 */
+		if (dfp) {
+			xfs_dir2_data_freeremove(d, dfp, needlogp);
+			dfp = xfs_dir2_data_freeinsert(d, newdup, needlogp);
+			ASSERT(dfp != NULL);
+			ASSERT(dfp->length == newdup->length);
+			ASSERT(be16_to_cpu(dfp->offset) == (char *)newdup - (char *)d);
+			/*
+			 * If we got inserted at the last slot,
+			 * that means we don't know if there was a better
+			 * choice for the last slot, or not.  Rescan.
+			 */
+			needscan = dfp == &d->hdr.bestfree[2];
+		}
+	}
+	/*
+	 * We match the last part of the entry.
+	 * Trim the allocated space off the tail of the entry.
+	 */
+	else if (matchback) {
+		newdup = dup;
+		newdup->length = cpu_to_be16(((char *)d + offset) - (char *)newdup);
+		*xfs_dir2_data_unused_tag_p(newdup) =
+			cpu_to_be16((char *)newdup - (char *)d);
+		xfs_dir2_data_log_unused(tp, bp, newdup);
+		/*
+		 * If it was in the table, remove it and add the new one.
+		 */
+		if (dfp) {
+			xfs_dir2_data_freeremove(d, dfp, needlogp);
+			dfp = xfs_dir2_data_freeinsert(d, newdup, needlogp);
+			ASSERT(dfp != NULL);
+			ASSERT(dfp->length == newdup->length);
+			ASSERT(be16_to_cpu(dfp->offset) == (char *)newdup - (char *)d);
+			/*
+			 * If we got inserted at the last slot,
+			 * that means we don't know if there was a better
+			 * choice for the last slot, or not.  Rescan.
+			 */
+			needscan = dfp == &d->hdr.bestfree[2];
+		}
+	}
+	/*
+	 * Poking out the middle of an entry.
+	 * Make two new entries.
+	 */
+	else {
+		newdup = dup;
+		newdup->length = cpu_to_be16(((char *)d + offset) - (char *)newdup);
+		*xfs_dir2_data_unused_tag_p(newdup) =
+			cpu_to_be16((char *)newdup - (char *)d);
+		xfs_dir2_data_log_unused(tp, bp, newdup);
+		newdup2 = (xfs_dir2_data_unused_t *)((char *)d + offset + len);
+		newdup2->freetag = cpu_to_be16(XFS_DIR2_DATA_FREE_TAG);
+		newdup2->length = cpu_to_be16(oldlen - len - be16_to_cpu(newdup->length));
+		*xfs_dir2_data_unused_tag_p(newdup2) =
+			cpu_to_be16((char *)newdup2 - (char *)d);
+		xfs_dir2_data_log_unused(tp, bp, newdup2);
+		/*
+		 * If the old entry was in the table, we need to scan
+		 * if the 3rd entry was valid, since these entries
+		 * are smaller than the old one.
+		 * If we don't need to scan that means there were 1 or 2
+		 * entries in the table, and removing the old and adding
+		 * the 2 new will work.
+		 */
+		if (dfp) {
+			needscan = (d->hdr.bestfree[2].length != 0);
+			if (!needscan) {
+				xfs_dir2_data_freeremove(d, dfp, needlogp);
+				(void)xfs_dir2_data_freeinsert(d, newdup,
+					needlogp);
+				(void)xfs_dir2_data_freeinsert(d, newdup2,
+					needlogp);
+			}
+		}
+	}
+	*needscanp = needscan;
+}
diff --git a/fs/xfs/xfs_dir2_data.h b/fs/xfs/xfs_dir2_data.h
new file mode 100644
index 00000000..efbc290c
--- /dev/null
+++ b/fs/xfs/xfs_dir2_data.h
@@ -0,0 +1,184 @@
+/*
+ * Copyright (c) 2000,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_DIR2_DATA_H__
+#define	__XFS_DIR2_DATA_H__
+
+/*
+ * Directory format 2, data block structures.
+ */
+
+struct xfs_dabuf;
+struct xfs_da_args;
+struct xfs_inode;
+struct xfs_trans;
+
+/*
+ * Constants.
+ */
+#define	XFS_DIR2_DATA_MAGIC	0x58443244	/* XD2D: for multiblock dirs */
+#define	XFS_DIR2_DATA_ALIGN_LOG	3		/* i.e., 8 bytes */
+#define	XFS_DIR2_DATA_ALIGN	(1 << XFS_DIR2_DATA_ALIGN_LOG)
+#define	XFS_DIR2_DATA_FREE_TAG	0xffff
+#define	XFS_DIR2_DATA_FD_COUNT	3
+
+/*
+ * Directory address space divided into sections,
+ * spaces separated by 32GB.
+ */
+#define	XFS_DIR2_SPACE_SIZE	(1ULL << (32 + XFS_DIR2_DATA_ALIGN_LOG))
+#define	XFS_DIR2_DATA_SPACE	0
+#define	XFS_DIR2_DATA_OFFSET	(XFS_DIR2_DATA_SPACE * XFS_DIR2_SPACE_SIZE)
+#define	XFS_DIR2_DATA_FIRSTDB(mp)	\
+	xfs_dir2_byte_to_db(mp, XFS_DIR2_DATA_OFFSET)
+
+/*
+ * Offsets of . and .. in data space (always block 0)
+ */
+#define	XFS_DIR2_DATA_DOT_OFFSET	\
+	((xfs_dir2_data_aoff_t)sizeof(xfs_dir2_data_hdr_t))
+#define	XFS_DIR2_DATA_DOTDOT_OFFSET	\
+	(XFS_DIR2_DATA_DOT_OFFSET + xfs_dir2_data_entsize(1))
+#define	XFS_DIR2_DATA_FIRST_OFFSET		\
+	(XFS_DIR2_DATA_DOTDOT_OFFSET + xfs_dir2_data_entsize(2))
+
+/*
+ * Structures.
+ */
+
+/*
+ * Describe a free area in the data block.
+ * The freespace will be formatted as a xfs_dir2_data_unused_t.
+ */
+typedef struct xfs_dir2_data_free {
+	__be16			offset;		/* start of freespace */
+	__be16			length;		/* length of freespace */
+} xfs_dir2_data_free_t;
+
+/*
+ * Header for the data blocks.
+ * Always at the beginning of a directory-sized block.
+ * The code knows that XFS_DIR2_DATA_FD_COUNT is 3.
+ */
+typedef struct xfs_dir2_data_hdr {
+	__be32			magic;		/* XFS_DIR2_DATA_MAGIC */
+						/* or XFS_DIR2_BLOCK_MAGIC */
+	xfs_dir2_data_free_t	bestfree[XFS_DIR2_DATA_FD_COUNT];
+} xfs_dir2_data_hdr_t;
+
+/*
+ * Active entry in a data block.  Aligned to 8 bytes.
+ * Tag appears as the last 2 bytes.
+ */
+typedef struct xfs_dir2_data_entry {
+	__be64			inumber;	/* inode number */
+	__u8			namelen;	/* name length */
+	__u8			name[1];	/* name bytes, no null */
+						/* variable offset */
+	__be16			tag;		/* starting offset of us */
+} xfs_dir2_data_entry_t;
+
+/*
+ * Unused entry in a data block.  Aligned to 8 bytes.
+ * Tag appears as the last 2 bytes.
+ */
+typedef struct xfs_dir2_data_unused {
+	__be16			freetag;	/* XFS_DIR2_DATA_FREE_TAG */
+	__be16			length;		/* total free length */
+						/* variable offset */
+	__be16			tag;		/* starting offset of us */
+} xfs_dir2_data_unused_t;
+
+typedef union {
+	xfs_dir2_data_entry_t	entry;
+	xfs_dir2_data_unused_t	unused;
+} xfs_dir2_data_union_t;
+
+/*
+ * Generic data block structure, for xfs_db.
+ */
+typedef struct xfs_dir2_data {
+	xfs_dir2_data_hdr_t	hdr;		/* magic XFS_DIR2_DATA_MAGIC */
+	xfs_dir2_data_union_t	u[1];
+} xfs_dir2_data_t;
+
+/*
+ * Macros.
+ */
+
+/*
+ * Size of a data entry.
+ */
+static inline int xfs_dir2_data_entsize(int n)
+{
+	return (int)roundup(offsetof(xfs_dir2_data_entry_t, name[0]) + (n) + \
+		 (uint)sizeof(xfs_dir2_data_off_t), XFS_DIR2_DATA_ALIGN);
+}
+
+/*
+ * Pointer to an entry's tag word.
+ */
+static inline __be16 *
+xfs_dir2_data_entry_tag_p(xfs_dir2_data_entry_t *dep)
+{
+	return (__be16 *)((char *)dep +
+		xfs_dir2_data_entsize(dep->namelen) - sizeof(__be16));
+}
+
+/*
+ * Pointer to a freespace's tag word.
+ */
+static inline __be16 *
+xfs_dir2_data_unused_tag_p(xfs_dir2_data_unused_t *dup)
+{
+	return (__be16 *)((char *)dup +
+			be16_to_cpu(dup->length) - sizeof(__be16));
+}
+
+/*
+ * Function declarations.
+ */
+#ifdef DEBUG
+extern void xfs_dir2_data_check(struct xfs_inode *dp, struct xfs_dabuf *bp);
+#else
+#define	xfs_dir2_data_check(dp,bp)
+#endif
+extern xfs_dir2_data_free_t *xfs_dir2_data_freefind(xfs_dir2_data_t *d,
+				xfs_dir2_data_unused_t *dup);
+extern xfs_dir2_data_free_t *xfs_dir2_data_freeinsert(xfs_dir2_data_t *d,
+				xfs_dir2_data_unused_t *dup, int *loghead);
+extern void xfs_dir2_data_freescan(struct xfs_mount *mp, xfs_dir2_data_t *d,
+				int *loghead);
+extern int xfs_dir2_data_init(struct xfs_da_args *args, xfs_dir2_db_t blkno,
+				struct xfs_dabuf **bpp);
+extern void xfs_dir2_data_log_entry(struct xfs_trans *tp, struct xfs_dabuf *bp,
+				xfs_dir2_data_entry_t *dep);
+extern void xfs_dir2_data_log_header(struct xfs_trans *tp,
+				struct xfs_dabuf *bp);
+extern void xfs_dir2_data_log_unused(struct xfs_trans *tp, struct xfs_dabuf *bp,
+				xfs_dir2_data_unused_t *dup);
+extern void xfs_dir2_data_make_free(struct xfs_trans *tp, struct xfs_dabuf *bp,
+				xfs_dir2_data_aoff_t offset,
+				xfs_dir2_data_aoff_t len, int *needlogp,
+				int *needscanp);
+extern void xfs_dir2_data_use_free(struct xfs_trans *tp, struct xfs_dabuf *bp,
+			       xfs_dir2_data_unused_t *dup,
+			       xfs_dir2_data_aoff_t offset,
+			       xfs_dir2_data_aoff_t len, int *needlogp,
+			       int *needscanp);
+
+#endif	/* __XFS_DIR2_DATA_H__ */
diff --git a/fs/xfs/xfs_dir2_leaf.c b/fs/xfs/xfs_dir2_leaf.c
new file mode 100644
index 00000000..ae891223
--- /dev/null
+++ b/fs/xfs/xfs_dir2_leaf.c
@@ -0,0 +1,1881 @@
+/*
+ * Copyright (c) 2000-2003,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_dir2.h"
+#include "xfs_mount.h"
+#include "xfs_da_btree.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_dir2_sf.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_bmap.h"
+#include "xfs_dir2_data.h"
+#include "xfs_dir2_leaf.h"
+#include "xfs_dir2_block.h"
+#include "xfs_dir2_node.h"
+#include "xfs_error.h"
+#include "xfs_trace.h"
+
+/*
+ * Local function declarations.
+ */
+#ifdef DEBUG
+static void xfs_dir2_leaf_check(xfs_inode_t *dp, xfs_dabuf_t *bp);
+#else
+#define	xfs_dir2_leaf_check(dp, bp)
+#endif
+static int xfs_dir2_leaf_lookup_int(xfs_da_args_t *args, xfs_dabuf_t **lbpp,
+				    int *indexp, xfs_dabuf_t **dbpp);
+static void xfs_dir2_leaf_log_bests(struct xfs_trans *tp, struct xfs_dabuf *bp,
+				    int first, int last);
+static void xfs_dir2_leaf_log_tail(struct xfs_trans *tp, struct xfs_dabuf *bp);
+
+
+/*
+ * Convert a block form directory to a leaf form directory.
+ */
+int						/* error */
+xfs_dir2_block_to_leaf(
+	xfs_da_args_t		*args,		/* operation arguments */
+	xfs_dabuf_t		*dbp)		/* input block's buffer */
+{
+	__be16			*bestsp;	/* leaf's bestsp entries */
+	xfs_dablk_t		blkno;		/* leaf block's bno */
+	xfs_dir2_block_t	*block;		/* block structure */
+	xfs_dir2_leaf_entry_t	*blp;		/* block's leaf entries */
+	xfs_dir2_block_tail_t	*btp;		/* block's tail */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	int			error;		/* error return code */
+	xfs_dabuf_t		*lbp;		/* leaf block's buffer */
+	xfs_dir2_db_t		ldb;		/* leaf block's bno */
+	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
+	xfs_dir2_leaf_tail_t	*ltp;		/* leaf's tail */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+	int			needlog;	/* need to log block header */
+	int			needscan;	/* need to rescan bestfree */
+	xfs_trans_t		*tp;		/* transaction pointer */
+
+	trace_xfs_dir2_block_to_leaf(args);
+
+	dp = args->dp;
+	mp = dp->i_mount;
+	tp = args->trans;
+	/*
+	 * Add the leaf block to the inode.
+	 * This interface will only put blocks in the leaf/node range.
+	 * Since that's empty now, we'll get the root (block 0 in range).
+	 */
+	if ((error = xfs_da_grow_inode(args, &blkno))) {
+		return error;
+	}
+	ldb = xfs_dir2_da_to_db(mp, blkno);
+	ASSERT(ldb == XFS_DIR2_LEAF_FIRSTDB(mp));
+	/*
+	 * Initialize the leaf block, get a buffer for it.
+	 */
+	if ((error = xfs_dir2_leaf_init(args, ldb, &lbp, XFS_DIR2_LEAF1_MAGIC))) {
+		return error;
+	}
+	ASSERT(lbp != NULL);
+	leaf = lbp->data;
+	block = dbp->data;
+	xfs_dir2_data_check(dp, dbp);
+	btp = xfs_dir2_block_tail_p(mp, block);
+	blp = xfs_dir2_block_leaf_p(btp);
+	/*
+	 * Set the counts in the leaf header.
+	 */
+	leaf->hdr.count = cpu_to_be16(be32_to_cpu(btp->count));
+	leaf->hdr.stale = cpu_to_be16(be32_to_cpu(btp->stale));
+	/*
+	 * Could compact these but I think we always do the conversion
+	 * after squeezing out stale entries.
+	 */
+	memcpy(leaf->ents, blp, be32_to_cpu(btp->count) * sizeof(xfs_dir2_leaf_entry_t));
+	xfs_dir2_leaf_log_ents(tp, lbp, 0, be16_to_cpu(leaf->hdr.count) - 1);
+	needscan = 0;
+	needlog = 1;
+	/*
+	 * Make the space formerly occupied by the leaf entries and block
+	 * tail be free.
+	 */
+	xfs_dir2_data_make_free(tp, dbp,
+		(xfs_dir2_data_aoff_t)((char *)blp - (char *)block),
+		(xfs_dir2_data_aoff_t)((char *)block + mp->m_dirblksize -
+				       (char *)blp),
+		&needlog, &needscan);
+	/*
+	 * Fix up the block header, make it a data block.
+	 */
+	block->hdr.magic = cpu_to_be32(XFS_DIR2_DATA_MAGIC);
+	if (needscan)
+		xfs_dir2_data_freescan(mp, (xfs_dir2_data_t *)block, &needlog);
+	/*
+	 * Set up leaf tail and bests table.
+	 */
+	ltp = xfs_dir2_leaf_tail_p(mp, leaf);
+	ltp->bestcount = cpu_to_be32(1);
+	bestsp = xfs_dir2_leaf_bests_p(ltp);
+	bestsp[0] =  block->hdr.bestfree[0].length;
+	/*
+	 * Log the data header and leaf bests table.
+	 */
+	if (needlog)
+		xfs_dir2_data_log_header(tp, dbp);
+	xfs_dir2_leaf_check(dp, lbp);
+	xfs_dir2_data_check(dp, dbp);
+	xfs_dir2_leaf_log_bests(tp, lbp, 0, 0);
+	xfs_da_buf_done(lbp);
+	return 0;
+}
+
+/*
+ * Add an entry to a leaf form directory.
+ */
+int						/* error */
+xfs_dir2_leaf_addname(
+	xfs_da_args_t		*args)		/* operation arguments */
+{
+	__be16			*bestsp;	/* freespace table in leaf */
+	int			compact;	/* need to compact leaves */
+	xfs_dir2_data_t		*data;		/* data block structure */
+	xfs_dabuf_t		*dbp;		/* data block buffer */
+	xfs_dir2_data_entry_t	*dep;		/* data block entry */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	xfs_dir2_data_unused_t	*dup;		/* data unused entry */
+	int			error;		/* error return value */
+	int			grown;		/* allocated new data block */
+	int			highstale;	/* index of next stale leaf */
+	int			i;		/* temporary, index */
+	int			index;		/* leaf table position */
+	xfs_dabuf_t		*lbp;		/* leaf's buffer */
+	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
+	int			length;		/* length of new entry */
+	xfs_dir2_leaf_entry_t	*lep;		/* leaf entry table pointer */
+	int			lfloglow;	/* low leaf logging index */
+	int			lfloghigh;	/* high leaf logging index */
+	int			lowstale;	/* index of prev stale leaf */
+	xfs_dir2_leaf_tail_t	*ltp;		/* leaf tail pointer */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+	int			needbytes;	/* leaf block bytes needed */
+	int			needlog;	/* need to log data header */
+	int			needscan;	/* need to rescan data free */
+	__be16			*tagp;		/* end of data entry */
+	xfs_trans_t		*tp;		/* transaction pointer */
+	xfs_dir2_db_t		use_block;	/* data block number */
+
+	trace_xfs_dir2_leaf_addname(args);
+
+	dp = args->dp;
+	tp = args->trans;
+	mp = dp->i_mount;
+	/*
+	 * Read the leaf block.
+	 */
+	error = xfs_da_read_buf(tp, dp, mp->m_dirleafblk, -1, &lbp,
+		XFS_DATA_FORK);
+	if (error) {
+		return error;
+	}
+	ASSERT(lbp != NULL);
+	/*
+	 * Look up the entry by hash value and name.
+	 * We know it's not there, our caller has already done a lookup.
+	 * So the index is of the entry to insert in front of.
+	 * But if there are dup hash values the index is of the first of those.
+	 */
+	index = xfs_dir2_leaf_search_hash(args, lbp);
+	leaf = lbp->data;
+	ltp = xfs_dir2_leaf_tail_p(mp, leaf);
+	bestsp = xfs_dir2_leaf_bests_p(ltp);
+	length = xfs_dir2_data_entsize(args->namelen);
+	/*
+	 * See if there are any entries with the same hash value
+	 * and space in their block for the new entry.
+	 * This is good because it puts multiple same-hash value entries
+	 * in a data block, improving the lookup of those entries.
+	 */
+	for (use_block = -1, lep = &leaf->ents[index];
+	     index < be16_to_cpu(leaf->hdr.count) && be32_to_cpu(lep->hashval) == args->hashval;
+	     index++, lep++) {
+		if (be32_to_cpu(lep->address) == XFS_DIR2_NULL_DATAPTR)
+			continue;
+		i = xfs_dir2_dataptr_to_db(mp, be32_to_cpu(lep->address));
+		ASSERT(i < be32_to_cpu(ltp->bestcount));
+		ASSERT(be16_to_cpu(bestsp[i]) != NULLDATAOFF);
+		if (be16_to_cpu(bestsp[i]) >= length) {
+			use_block = i;
+			break;
+		}
+	}
+	/*
+	 * Didn't find a block yet, linear search all the data blocks.
+	 */
+	if (use_block == -1) {
+		for (i = 0; i < be32_to_cpu(ltp->bestcount); i++) {
+			/*
+			 * Remember a block we see that's missing.
+			 */
+			if (be16_to_cpu(bestsp[i]) == NULLDATAOFF && use_block == -1)
+				use_block = i;
+			else if (be16_to_cpu(bestsp[i]) >= length) {
+				use_block = i;
+				break;
+			}
+		}
+	}
+	/*
+	 * How many bytes do we need in the leaf block?
+	 */
+	needbytes =
+		(leaf->hdr.stale ? 0 : (uint)sizeof(leaf->ents[0])) +
+		(use_block != -1 ? 0 : (uint)sizeof(leaf->bests[0]));
+	/*
+	 * Now kill use_block if it refers to a missing block, so we
+	 * can use it as an indication of allocation needed.
+	 */
+	if (use_block != -1 && be16_to_cpu(bestsp[use_block]) == NULLDATAOFF)
+		use_block = -1;
+	/*
+	 * If we don't have enough free bytes but we can make enough
+	 * by compacting out stale entries, we'll do that.
+	 */
+	if ((char *)bestsp - (char *)&leaf->ents[be16_to_cpu(leaf->hdr.count)] <
+				needbytes && be16_to_cpu(leaf->hdr.stale) > 1) {
+		compact = 1;
+	}
+	/*
+	 * Otherwise if we don't have enough free bytes we need to
+	 * convert to node form.
+	 */
+	else if ((char *)bestsp - (char *)&leaf->ents[be16_to_cpu(
+						leaf->hdr.count)] < needbytes) {
+		/*
+		 * Just checking or no space reservation, give up.
+		 */
+		if ((args->op_flags & XFS_DA_OP_JUSTCHECK) ||
+							args->total == 0) {
+			xfs_da_brelse(tp, lbp);
+			return XFS_ERROR(ENOSPC);
+		}
+		/*
+		 * Convert to node form.
+		 */
+		error = xfs_dir2_leaf_to_node(args, lbp);
+		xfs_da_buf_done(lbp);
+		if (error)
+			return error;
+		/*
+		 * Then add the new entry.
+		 */
+		return xfs_dir2_node_addname(args);
+	}
+	/*
+	 * Otherwise it will fit without compaction.
+	 */
+	else
+		compact = 0;
+	/*
+	 * If just checking, then it will fit unless we needed to allocate
+	 * a new data block.
+	 */
+	if (args->op_flags & XFS_DA_OP_JUSTCHECK) {
+		xfs_da_brelse(tp, lbp);
+		return use_block == -1 ? XFS_ERROR(ENOSPC) : 0;
+	}
+	/*
+	 * If no allocations are allowed, return now before we've
+	 * changed anything.
+	 */
+	if (args->total == 0 && use_block == -1) {
+		xfs_da_brelse(tp, lbp);
+		return XFS_ERROR(ENOSPC);
+	}
+	/*
+	 * Need to compact the leaf entries, removing stale ones.
+	 * Leave one stale entry behind - the one closest to our
+	 * insertion index - and we'll shift that one to our insertion
+	 * point later.
+	 */
+	if (compact) {
+		xfs_dir2_leaf_compact_x1(lbp, &index, &lowstale, &highstale,
+			&lfloglow, &lfloghigh);
+	}
+	/*
+	 * There are stale entries, so we'll need log-low and log-high
+	 * impossibly bad values later.
+	 */
+	else if (be16_to_cpu(leaf->hdr.stale)) {
+		lfloglow = be16_to_cpu(leaf->hdr.count);
+		lfloghigh = -1;
+	}
+	/*
+	 * If there was no data block space found, we need to allocate
+	 * a new one.
+	 */
+	if (use_block == -1) {
+		/*
+		 * Add the new data block.
+		 */
+		if ((error = xfs_dir2_grow_inode(args, XFS_DIR2_DATA_SPACE,
+				&use_block))) {
+			xfs_da_brelse(tp, lbp);
+			return error;
+		}
+		/*
+		 * Initialize the block.
+		 */
+		if ((error = xfs_dir2_data_init(args, use_block, &dbp))) {
+			xfs_da_brelse(tp, lbp);
+			return error;
+		}
+		/*
+		 * If we're adding a new data block on the end we need to
+		 * extend the bests table.  Copy it up one entry.
+		 */
+		if (use_block >= be32_to_cpu(ltp->bestcount)) {
+			bestsp--;
+			memmove(&bestsp[0], &bestsp[1],
+				be32_to_cpu(ltp->bestcount) * sizeof(bestsp[0]));
+			be32_add_cpu(&ltp->bestcount, 1);
+			xfs_dir2_leaf_log_tail(tp, lbp);
+			xfs_dir2_leaf_log_bests(tp, lbp, 0, be32_to_cpu(ltp->bestcount) - 1);
+		}
+		/*
+		 * If we're filling in a previously empty block just log it.
+		 */
+		else
+			xfs_dir2_leaf_log_bests(tp, lbp, use_block, use_block);
+		data = dbp->data;
+		bestsp[use_block] = data->hdr.bestfree[0].length;
+		grown = 1;
+	}
+	/*
+	 * Already had space in some data block.
+	 * Just read that one in.
+	 */
+	else {
+		if ((error =
+		    xfs_da_read_buf(tp, dp, xfs_dir2_db_to_da(mp, use_block),
+			    -1, &dbp, XFS_DATA_FORK))) {
+			xfs_da_brelse(tp, lbp);
+			return error;
+		}
+		data = dbp->data;
+		grown = 0;
+	}
+	xfs_dir2_data_check(dp, dbp);
+	/*
+	 * Point to the biggest freespace in our data block.
+	 */
+	dup = (xfs_dir2_data_unused_t *)
+	      ((char *)data + be16_to_cpu(data->hdr.bestfree[0].offset));
+	ASSERT(be16_to_cpu(dup->length) >= length);
+	needscan = needlog = 0;
+	/*
+	 * Mark the initial part of our freespace in use for the new entry.
+	 */
+	xfs_dir2_data_use_free(tp, dbp, dup,
+		(xfs_dir2_data_aoff_t)((char *)dup - (char *)data), length,
+		&needlog, &needscan);
+	/*
+	 * Initialize our new entry (at last).
+	 */
+	dep = (xfs_dir2_data_entry_t *)dup;
+	dep->inumber = cpu_to_be64(args->inumber);
+	dep->namelen = args->namelen;
+	memcpy(dep->name, args->name, dep->namelen);
+	tagp = xfs_dir2_data_entry_tag_p(dep);
+	*tagp = cpu_to_be16((char *)dep - (char *)data);
+	/*
+	 * Need to scan fix up the bestfree table.
+	 */
+	if (needscan)
+		xfs_dir2_data_freescan(mp, data, &needlog);
+	/*
+	 * Need to log the data block's header.
+	 */
+	if (needlog)
+		xfs_dir2_data_log_header(tp, dbp);
+	xfs_dir2_data_log_entry(tp, dbp, dep);
+	/*
+	 * If the bests table needs to be changed, do it.
+	 * Log the change unless we've already done that.
+	 */
+	if (be16_to_cpu(bestsp[use_block]) != be16_to_cpu(data->hdr.bestfree[0].length)) {
+		bestsp[use_block] = data->hdr.bestfree[0].length;
+		if (!grown)
+			xfs_dir2_leaf_log_bests(tp, lbp, use_block, use_block);
+	}
+	/*
+	 * Now we need to make room to insert the leaf entry.
+	 * If there are no stale entries, we just insert a hole at index.
+	 */
+	if (!leaf->hdr.stale) {
+		/*
+		 * lep is still good as the index leaf entry.
+		 */
+		if (index < be16_to_cpu(leaf->hdr.count))
+			memmove(lep + 1, lep,
+				(be16_to_cpu(leaf->hdr.count) - index) * sizeof(*lep));
+		/*
+		 * Record low and high logging indices for the leaf.
+		 */
+		lfloglow = index;
+		lfloghigh = be16_to_cpu(leaf->hdr.count);
+		be16_add_cpu(&leaf->hdr.count, 1);
+	}
+	/*
+	 * There are stale entries.
+	 * We will use one of them for the new entry.
+	 * It's probably not at the right location, so we'll have to
+	 * shift some up or down first.
+	 */
+	else {
+		/*
+		 * If we didn't compact before, we need to find the nearest
+		 * stale entries before and after our insertion point.
+		 */
+		if (compact == 0) {
+			/*
+			 * Find the first stale entry before the insertion
+			 * point, if any.
+			 */
+			for (lowstale = index - 1;
+			     lowstale >= 0 &&
+				be32_to_cpu(leaf->ents[lowstale].address) !=
+				XFS_DIR2_NULL_DATAPTR;
+			     lowstale--)
+				continue;
+			/*
+			 * Find the next stale entry at or after the insertion
+			 * point, if any.   Stop if we go so far that the
+			 * lowstale entry would be better.
+			 */
+			for (highstale = index;
+			     highstale < be16_to_cpu(leaf->hdr.count) &&
+				be32_to_cpu(leaf->ents[highstale].address) !=
+				XFS_DIR2_NULL_DATAPTR &&
+				(lowstale < 0 ||
+				 index - lowstale - 1 >= highstale - index);
+			     highstale++)
+				continue;
+		}
+		/*
+		 * If the low one is better, use it.
+		 */
+		if (lowstale >= 0 &&
+		    (highstale == be16_to_cpu(leaf->hdr.count) ||
+		     index - lowstale - 1 < highstale - index)) {
+			ASSERT(index - lowstale - 1 >= 0);
+			ASSERT(be32_to_cpu(leaf->ents[lowstale].address) ==
+			       XFS_DIR2_NULL_DATAPTR);
+			/*
+			 * Copy entries up to cover the stale entry
+			 * and make room for the new entry.
+			 */
+			if (index - lowstale - 1 > 0)
+				memmove(&leaf->ents[lowstale],
+					&leaf->ents[lowstale + 1],
+					(index - lowstale - 1) * sizeof(*lep));
+			lep = &leaf->ents[index - 1];
+			lfloglow = MIN(lowstale, lfloglow);
+			lfloghigh = MAX(index - 1, lfloghigh);
+		}
+		/*
+		 * The high one is better, so use that one.
+		 */
+		else {
+			ASSERT(highstale - index >= 0);
+			ASSERT(be32_to_cpu(leaf->ents[highstale].address) ==
+			       XFS_DIR2_NULL_DATAPTR);
+			/*
+			 * Copy entries down to cover the stale entry
+			 * and make room for the new entry.
+			 */
+			if (highstale - index > 0)
+				memmove(&leaf->ents[index + 1],
+					&leaf->ents[index],
+					(highstale - index) * sizeof(*lep));
+			lep = &leaf->ents[index];
+			lfloglow = MIN(index, lfloglow);
+			lfloghigh = MAX(highstale, lfloghigh);
+		}
+		be16_add_cpu(&leaf->hdr.stale, -1);
+	}
+	/*
+	 * Fill in the new leaf entry.
+	 */
+	lep->hashval = cpu_to_be32(args->hashval);
+	lep->address = cpu_to_be32(xfs_dir2_db_off_to_dataptr(mp, use_block,
+				be16_to_cpu(*tagp)));
+	/*
+	 * Log the leaf fields and give up the buffers.
+	 */
+	xfs_dir2_leaf_log_header(tp, lbp);
+	xfs_dir2_leaf_log_ents(tp, lbp, lfloglow, lfloghigh);
+	xfs_dir2_leaf_check(dp, lbp);
+	xfs_da_buf_done(lbp);
+	xfs_dir2_data_check(dp, dbp);
+	xfs_da_buf_done(dbp);
+	return 0;
+}
+
+#ifdef DEBUG
+/*
+ * Check the internal consistency of a leaf1 block.
+ * Pop an assert if something is wrong.
+ */
+STATIC void
+xfs_dir2_leaf_check(
+	xfs_inode_t		*dp,		/* incore directory inode */
+	xfs_dabuf_t		*bp)		/* leaf's buffer */
+{
+	int			i;		/* leaf index */
+	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
+	xfs_dir2_leaf_tail_t	*ltp;		/* leaf tail pointer */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+	int			stale;		/* count of stale leaves */
+
+	leaf = bp->data;
+	mp = dp->i_mount;
+	ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_DIR2_LEAF1_MAGIC);
+	/*
+	 * This value is not restrictive enough.
+	 * Should factor in the size of the bests table as well.
+	 * We can deduce a value for that from di_size.
+	 */
+	ASSERT(be16_to_cpu(leaf->hdr.count) <= xfs_dir2_max_leaf_ents(mp));
+	ltp = xfs_dir2_leaf_tail_p(mp, leaf);
+	/*
+	 * Leaves and bests don't overlap.
+	 */
+	ASSERT((char *)&leaf->ents[be16_to_cpu(leaf->hdr.count)] <=
+	       (char *)xfs_dir2_leaf_bests_p(ltp));
+	/*
+	 * Check hash value order, count stale entries.
+	 */
+	for (i = stale = 0; i < be16_to_cpu(leaf->hdr.count); i++) {
+		if (i + 1 < be16_to_cpu(leaf->hdr.count))
+			ASSERT(be32_to_cpu(leaf->ents[i].hashval) <=
+			       be32_to_cpu(leaf->ents[i + 1].hashval));
+		if (be32_to_cpu(leaf->ents[i].address) == XFS_DIR2_NULL_DATAPTR)
+			stale++;
+	}
+	ASSERT(be16_to_cpu(leaf->hdr.stale) == stale);
+}
+#endif	/* DEBUG */
+
+/*
+ * Compact out any stale entries in the leaf.
+ * Log the header and changed leaf entries, if any.
+ */
+void
+xfs_dir2_leaf_compact(
+	xfs_da_args_t	*args,		/* operation arguments */
+	xfs_dabuf_t	*bp)		/* leaf buffer */
+{
+	int		from;		/* source leaf index */
+	xfs_dir2_leaf_t	*leaf;		/* leaf structure */
+	int		loglow;		/* first leaf entry to log */
+	int		to;		/* target leaf index */
+
+	leaf = bp->data;
+	if (!leaf->hdr.stale) {
+		return;
+	}
+	/*
+	 * Compress out the stale entries in place.
+	 */
+	for (from = to = 0, loglow = -1; from < be16_to_cpu(leaf->hdr.count); from++) {
+		if (be32_to_cpu(leaf->ents[from].address) == XFS_DIR2_NULL_DATAPTR)
+			continue;
+		/*
+		 * Only actually copy the entries that are different.
+		 */
+		if (from > to) {
+			if (loglow == -1)
+				loglow = to;
+			leaf->ents[to] = leaf->ents[from];
+		}
+		to++;
+	}
+	/*
+	 * Update and log the header, log the leaf entries.
+	 */
+	ASSERT(be16_to_cpu(leaf->hdr.stale) == from - to);
+	be16_add_cpu(&leaf->hdr.count, -(be16_to_cpu(leaf->hdr.stale)));
+	leaf->hdr.stale = 0;
+	xfs_dir2_leaf_log_header(args->trans, bp);
+	if (loglow != -1)
+		xfs_dir2_leaf_log_ents(args->trans, bp, loglow, to - 1);
+}
+
+/*
+ * Compact the leaf entries, removing stale ones.
+ * Leave one stale entry behind - the one closest to our
+ * insertion index - and the caller will shift that one to our insertion
+ * point later.
+ * Return new insertion index, where the remaining stale entry is,
+ * and leaf logging indices.
+ */
+void
+xfs_dir2_leaf_compact_x1(
+	xfs_dabuf_t	*bp,		/* leaf buffer */
+	int		*indexp,	/* insertion index */
+	int		*lowstalep,	/* out: stale entry before us */
+	int		*highstalep,	/* out: stale entry after us */
+	int		*lowlogp,	/* out: low log index */
+	int		*highlogp)	/* out: high log index */
+{
+	int		from;		/* source copy index */
+	int		highstale;	/* stale entry at/after index */
+	int		index;		/* insertion index */
+	int		keepstale;	/* source index of kept stale */
+	xfs_dir2_leaf_t	*leaf;		/* leaf structure */
+	int		lowstale;	/* stale entry before index */
+	int		newindex=0;	/* new insertion index */
+	int		to;		/* destination copy index */
+
+	leaf = bp->data;
+	ASSERT(be16_to_cpu(leaf->hdr.stale) > 1);
+	index = *indexp;
+	/*
+	 * Find the first stale entry before our index, if any.
+	 */
+	for (lowstale = index - 1;
+	     lowstale >= 0 &&
+		be32_to_cpu(leaf->ents[lowstale].address) != XFS_DIR2_NULL_DATAPTR;
+	     lowstale--)
+		continue;
+	/*
+	 * Find the first stale entry at or after our index, if any.
+	 * Stop if the answer would be worse than lowstale.
+	 */
+	for (highstale = index;
+	     highstale < be16_to_cpu(leaf->hdr.count) &&
+		be32_to_cpu(leaf->ents[highstale].address) != XFS_DIR2_NULL_DATAPTR &&
+		(lowstale < 0 || index - lowstale > highstale - index);
+	     highstale++)
+		continue;
+	/*
+	 * Pick the better of lowstale and highstale.
+	 */
+	if (lowstale >= 0 &&
+	    (highstale == be16_to_cpu(leaf->hdr.count) ||
+	     index - lowstale <= highstale - index))
+		keepstale = lowstale;
+	else
+		keepstale = highstale;
+	/*
+	 * Copy the entries in place, removing all the stale entries
+	 * except keepstale.
+	 */
+	for (from = to = 0; from < be16_to_cpu(leaf->hdr.count); from++) {
+		/*
+		 * Notice the new value of index.
+		 */
+		if (index == from)
+			newindex = to;
+		if (from != keepstale &&
+		    be32_to_cpu(leaf->ents[from].address) == XFS_DIR2_NULL_DATAPTR) {
+			if (from == to)
+				*lowlogp = to;
+			continue;
+		}
+		/*
+		 * Record the new keepstale value for the insertion.
+		 */
+		if (from == keepstale)
+			lowstale = highstale = to;
+		/*
+		 * Copy only the entries that have moved.
+		 */
+		if (from > to)
+			leaf->ents[to] = leaf->ents[from];
+		to++;
+	}
+	ASSERT(from > to);
+	/*
+	 * If the insertion point was past the last entry,
+	 * set the new insertion point accordingly.
+	 */
+	if (index == from)
+		newindex = to;
+	*indexp = newindex;
+	/*
+	 * Adjust the leaf header values.
+	 */
+	be16_add_cpu(&leaf->hdr.count, -(from - to));
+	leaf->hdr.stale = cpu_to_be16(1);
+	/*
+	 * Remember the low/high stale value only in the "right"
+	 * direction.
+	 */
+	if (lowstale >= newindex)
+		lowstale = -1;
+	else
+		highstale = be16_to_cpu(leaf->hdr.count);
+	*highlogp = be16_to_cpu(leaf->hdr.count) - 1;
+	*lowstalep = lowstale;
+	*highstalep = highstale;
+}
+
+/*
+ * Getdents (readdir) for leaf and node directories.
+ * This reads the data blocks only, so is the same for both forms.
+ */
+int						/* error */
+xfs_dir2_leaf_getdents(
+	xfs_inode_t		*dp,		/* incore directory inode */
+	void			*dirent,
+	size_t			bufsize,
+	xfs_off_t		*offset,
+	filldir_t		filldir)
+{
+	xfs_dabuf_t		*bp;		/* data block buffer */
+	int			byteoff;	/* offset in current block */
+	xfs_dir2_db_t		curdb;		/* db for current block */
+	xfs_dir2_off_t		curoff;		/* current overall offset */
+	xfs_dir2_data_t		*data;		/* data block structure */
+	xfs_dir2_data_entry_t	*dep;		/* data entry */
+	xfs_dir2_data_unused_t	*dup;		/* unused entry */
+	int			error = 0;	/* error return value */
+	int			i;		/* temporary loop index */
+	int			j;		/* temporary loop index */
+	int			length;		/* temporary length value */
+	xfs_bmbt_irec_t		*map;		/* map vector for blocks */
+	xfs_extlen_t		map_blocks;	/* number of fsbs in map */
+	xfs_dablk_t		map_off;	/* last mapped file offset */
+	int			map_size;	/* total entries in *map */
+	int			map_valid;	/* valid entries in *map */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+	xfs_dir2_off_t		newoff;		/* new curoff after new blk */
+	int			nmap;		/* mappings to ask xfs_bmapi */
+	char			*ptr = NULL;	/* pointer to current data */
+	int			ra_current;	/* number of read-ahead blks */
+	int			ra_index;	/* *map index for read-ahead */
+	int			ra_offset;	/* map entry offset for ra */
+	int			ra_want;	/* readahead count wanted */
+
+	/*
+	 * If the offset is at or past the largest allowed value,
+	 * give up right away.
+	 */
+	if (*offset >= XFS_DIR2_MAX_DATAPTR)
+		return 0;
+
+	mp = dp->i_mount;
+
+	/*
+	 * Set up to bmap a number of blocks based on the caller's
+	 * buffer size, the directory block size, and the filesystem
+	 * block size.
+	 */
+	map_size = howmany(bufsize + mp->m_dirblksize, mp->m_sb.sb_blocksize);
+	map = kmem_alloc(map_size * sizeof(*map), KM_SLEEP);
+	map_valid = ra_index = ra_offset = ra_current = map_blocks = 0;
+	bp = NULL;
+
+	/*
+	 * Inside the loop we keep the main offset value as a byte offset
+	 * in the directory file.
+	 */
+	curoff = xfs_dir2_dataptr_to_byte(mp, *offset);
+
+	/*
+	 * Force this conversion through db so we truncate the offset
+	 * down to get the start of the data block.
+	 */
+	map_off = xfs_dir2_db_to_da(mp, xfs_dir2_byte_to_db(mp, curoff));
+	/*
+	 * Loop over directory entries until we reach the end offset.
+	 * Get more blocks and readahead as necessary.
+	 */
+	while (curoff < XFS_DIR2_LEAF_OFFSET) {
+		/*
+		 * If we have no buffer, or we're off the end of the
+		 * current buffer, need to get another one.
+		 */
+		if (!bp || ptr >= (char *)bp->data + mp->m_dirblksize) {
+			/*
+			 * If we have a buffer, we need to release it and
+			 * take it out of the mapping.
+			 */
+			if (bp) {
+				xfs_da_brelse(NULL, bp);
+				bp = NULL;
+				map_blocks -= mp->m_dirblkfsbs;
+				/*
+				 * Loop to get rid of the extents for the
+				 * directory block.
+				 */
+				for (i = mp->m_dirblkfsbs; i > 0; ) {
+					j = MIN((int)map->br_blockcount, i);
+					map->br_blockcount -= j;
+					map->br_startblock += j;
+					map->br_startoff += j;
+					/*
+					 * If mapping is done, pitch it from
+					 * the table.
+					 */
+					if (!map->br_blockcount && --map_valid)
+						memmove(&map[0], &map[1],
+							sizeof(map[0]) *
+							map_valid);
+					i -= j;
+				}
+			}
+			/*
+			 * Recalculate the readahead blocks wanted.
+			 */
+			ra_want = howmany(bufsize + mp->m_dirblksize,
+					  mp->m_sb.sb_blocksize) - 1;
+			ASSERT(ra_want >= 0);
+
+			/*
+			 * If we don't have as many as we want, and we haven't
+			 * run out of data blocks, get some more mappings.
+			 */
+			if (1 + ra_want > map_blocks &&
+			    map_off <
+			    xfs_dir2_byte_to_da(mp, XFS_DIR2_LEAF_OFFSET)) {
+				/*
+				 * Get more bmaps, fill in after the ones
+				 * we already have in the table.
+				 */
+				nmap = map_size - map_valid;
+				error = xfs_bmapi(NULL, dp,
+					map_off,
+					xfs_dir2_byte_to_da(mp,
+						XFS_DIR2_LEAF_OFFSET) - map_off,
+					XFS_BMAPI_METADATA, NULL, 0,
+					&map[map_valid], &nmap, NULL);
+				/*
+				 * Don't know if we should ignore this or
+				 * try to return an error.
+				 * The trouble with returning errors
+				 * is that readdir will just stop without
+				 * actually passing the error through.
+				 */
+				if (error)
+					break;	/* XXX */
+				/*
+				 * If we got all the mappings we asked for,
+				 * set the final map offset based on the
+				 * last bmap value received.
+				 * Otherwise, we've reached the end.
+				 */
+				if (nmap == map_size - map_valid)
+					map_off =
+					map[map_valid + nmap - 1].br_startoff +
+					map[map_valid + nmap - 1].br_blockcount;
+				else
+					map_off =
+						xfs_dir2_byte_to_da(mp,
+							XFS_DIR2_LEAF_OFFSET);
+				/*
+				 * Look for holes in the mapping, and
+				 * eliminate them.  Count up the valid blocks.
+				 */
+				for (i = map_valid; i < map_valid + nmap; ) {
+					if (map[i].br_startblock ==
+					    HOLESTARTBLOCK) {
+						nmap--;
+						length = map_valid + nmap - i;
+						if (length)
+							memmove(&map[i],
+								&map[i + 1],
+								sizeof(map[i]) *
+								length);
+					} else {
+						map_blocks +=
+							map[i].br_blockcount;
+						i++;
+					}
+				}
+				map_valid += nmap;
+			}
+			/*
+			 * No valid mappings, so no more data blocks.
+			 */
+			if (!map_valid) {
+				curoff = xfs_dir2_da_to_byte(mp, map_off);
+				break;
+			}
+			/*
+			 * Read the directory block starting at the first
+			 * mapping.
+			 */
+			curdb = xfs_dir2_da_to_db(mp, map->br_startoff);
+			error = xfs_da_read_buf(NULL, dp, map->br_startoff,
+				map->br_blockcount >= mp->m_dirblkfsbs ?
+				    XFS_FSB_TO_DADDR(mp, map->br_startblock) :
+				    -1,
+				&bp, XFS_DATA_FORK);
+			/*
+			 * Should just skip over the data block instead
+			 * of giving up.
+			 */
+			if (error)
+				break;	/* XXX */
+			/*
+			 * Adjust the current amount of read-ahead: we just
+			 * read a block that was previously ra.
+			 */
+			if (ra_current)
+				ra_current -= mp->m_dirblkfsbs;
+			/*
+			 * Do we need more readahead?
+			 */
+			for (ra_index = ra_offset = i = 0;
+			     ra_want > ra_current && i < map_blocks;
+			     i += mp->m_dirblkfsbs) {
+				ASSERT(ra_index < map_valid);
+				/*
+				 * Read-ahead a contiguous directory block.
+				 */
+				if (i > ra_current &&
+				    map[ra_index].br_blockcount >=
+				    mp->m_dirblkfsbs) {
+					xfs_buf_readahead(mp->m_ddev_targp,
+						XFS_FSB_TO_DADDR(mp,
+						   map[ra_index].br_startblock +
+						   ra_offset),
+						(int)BTOBB(mp->m_dirblksize));
+					ra_current = i;
+				}
+				/*
+				 * Read-ahead a non-contiguous directory block.
+				 * This doesn't use our mapping, but this
+				 * is a very rare case.
+				 */
+				else if (i > ra_current) {
+					(void)xfs_da_reada_buf(NULL, dp,
+						map[ra_index].br_startoff +
+						ra_offset, XFS_DATA_FORK);
+					ra_current = i;
+				}
+				/*
+				 * Advance offset through the mapping table.
+				 */
+				for (j = 0; j < mp->m_dirblkfsbs; j++) {
+					/*
+					 * The rest of this extent but not
+					 * more than a dir block.
+					 */
+					length = MIN(mp->m_dirblkfsbs,
+						(int)(map[ra_index].br_blockcount -
+						ra_offset));
+					j += length;
+					ra_offset += length;
+					/*
+					 * Advance to the next mapping if
+					 * this one is used up.
+					 */
+					if (ra_offset ==
+					    map[ra_index].br_blockcount) {
+						ra_offset = 0;
+						ra_index++;
+					}
+				}
+			}
+			/*
+			 * Having done a read, we need to set a new offset.
+			 */
+			newoff = xfs_dir2_db_off_to_byte(mp, curdb, 0);
+			/*
+			 * Start of the current block.
+			 */
+			if (curoff < newoff)
+				curoff = newoff;
+			/*
+			 * Make sure we're in the right block.
+			 */
+			else if (curoff > newoff)
+				ASSERT(xfs_dir2_byte_to_db(mp, curoff) ==
+				       curdb);
+			data = bp->data;
+			xfs_dir2_data_check(dp, bp);
+			/*
+			 * Find our position in the block.
+			 */
+			ptr = (char *)&data->u;
+			byteoff = xfs_dir2_byte_to_off(mp, curoff);
+			/*
+			 * Skip past the header.
+			 */
+			if (byteoff == 0)
+				curoff += (uint)sizeof(data->hdr);
+			/*
+			 * Skip past entries until we reach our offset.
+			 */
+			else {
+				while ((char *)ptr - (char *)data < byteoff) {
+					dup = (xfs_dir2_data_unused_t *)ptr;
+
+					if (be16_to_cpu(dup->freetag)
+						  == XFS_DIR2_DATA_FREE_TAG) {
+
+						length = be16_to_cpu(dup->length);
+						ptr += length;
+						continue;
+					}
+					dep = (xfs_dir2_data_entry_t *)ptr;
+					length =
+					   xfs_dir2_data_entsize(dep->namelen);
+					ptr += length;
+				}
+				/*
+				 * Now set our real offset.
+				 */
+				curoff =
+					xfs_dir2_db_off_to_byte(mp,
+					    xfs_dir2_byte_to_db(mp, curoff),
+					    (char *)ptr - (char *)data);
+				if (ptr >= (char *)data + mp->m_dirblksize) {
+					continue;
+				}
+			}
+		}
+		/*
+		 * We have a pointer to an entry.
+		 * Is it a live one?
+		 */
+		dup = (xfs_dir2_data_unused_t *)ptr;
+		/*
+		 * No, it's unused, skip over it.
+		 */
+		if (be16_to_cpu(dup->freetag) == XFS_DIR2_DATA_FREE_TAG) {
+			length = be16_to_cpu(dup->length);
+			ptr += length;
+			curoff += length;
+			continue;
+		}
+
+		dep = (xfs_dir2_data_entry_t *)ptr;
+		length = xfs_dir2_data_entsize(dep->namelen);
+
+		if (filldir(dirent, (char *)dep->name, dep->namelen,
+			    xfs_dir2_byte_to_dataptr(mp, curoff) & 0x7fffffff,
+			    be64_to_cpu(dep->inumber), DT_UNKNOWN))
+			break;
+
+		/*
+		 * Advance to next entry in the block.
+		 */
+		ptr += length;
+		curoff += length;
+		/* bufsize may have just been a guess; don't go negative */
+		bufsize = bufsize > length ? bufsize - length : 0;
+	}
+
+	/*
+	 * All done.  Set output offset value to current offset.
+	 */
+	if (curoff > xfs_dir2_dataptr_to_byte(mp, XFS_DIR2_MAX_DATAPTR))
+		*offset = XFS_DIR2_MAX_DATAPTR & 0x7fffffff;
+	else
+		*offset = xfs_dir2_byte_to_dataptr(mp, curoff) & 0x7fffffff;
+	kmem_free(map);
+	if (bp)
+		xfs_da_brelse(NULL, bp);
+	return error;
+}
+
+/*
+ * Initialize a new leaf block, leaf1 or leafn magic accepted.
+ */
+int
+xfs_dir2_leaf_init(
+	xfs_da_args_t		*args,		/* operation arguments */
+	xfs_dir2_db_t		bno,		/* directory block number */
+	xfs_dabuf_t		**bpp,		/* out: leaf buffer */
+	int			magic)		/* magic number for block */
+{
+	xfs_dabuf_t		*bp;		/* leaf buffer */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	int			error;		/* error return code */
+	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
+	xfs_dir2_leaf_tail_t	*ltp;		/* leaf tail structure */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+	xfs_trans_t		*tp;		/* transaction pointer */
+
+	dp = args->dp;
+	ASSERT(dp != NULL);
+	tp = args->trans;
+	mp = dp->i_mount;
+	ASSERT(bno >= XFS_DIR2_LEAF_FIRSTDB(mp) &&
+	       bno < XFS_DIR2_FREE_FIRSTDB(mp));
+	/*
+	 * Get the buffer for the block.
+	 */
+	error = xfs_da_get_buf(tp, dp, xfs_dir2_db_to_da(mp, bno), -1, &bp,
+		XFS_DATA_FORK);
+	if (error) {
+		return error;
+	}
+	ASSERT(bp != NULL);
+	leaf = bp->data;
+	/*
+	 * Initialize the header.
+	 */
+	leaf->hdr.info.magic = cpu_to_be16(magic);
+	leaf->hdr.info.forw = 0;
+	leaf->hdr.info.back = 0;
+	leaf->hdr.count = 0;
+	leaf->hdr.stale = 0;
+	xfs_dir2_leaf_log_header(tp, bp);
+	/*
+	 * If it's a leaf-format directory initialize the tail.
+	 * In this case our caller has the real bests table to copy into
+	 * the block.
+	 */
+	if (magic == XFS_DIR2_LEAF1_MAGIC) {
+		ltp = xfs_dir2_leaf_tail_p(mp, leaf);
+		ltp->bestcount = 0;
+		xfs_dir2_leaf_log_tail(tp, bp);
+	}
+	*bpp = bp;
+	return 0;
+}
+
+/*
+ * Log the bests entries indicated from a leaf1 block.
+ */
+static void
+xfs_dir2_leaf_log_bests(
+	xfs_trans_t		*tp,		/* transaction pointer */
+	xfs_dabuf_t		*bp,		/* leaf buffer */
+	int			first,		/* first entry to log */
+	int			last)		/* last entry to log */
+{
+	__be16			*firstb;	/* pointer to first entry */
+	__be16			*lastb;		/* pointer to last entry */
+	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
+	xfs_dir2_leaf_tail_t	*ltp;		/* leaf tail structure */
+
+	leaf = bp->data;
+	ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_DIR2_LEAF1_MAGIC);
+	ltp = xfs_dir2_leaf_tail_p(tp->t_mountp, leaf);
+	firstb = xfs_dir2_leaf_bests_p(ltp) + first;
+	lastb = xfs_dir2_leaf_bests_p(ltp) + last;
+	xfs_da_log_buf(tp, bp, (uint)((char *)firstb - (char *)leaf),
+		(uint)((char *)lastb - (char *)leaf + sizeof(*lastb) - 1));
+}
+
+/*
+ * Log the leaf entries indicated from a leaf1 or leafn block.
+ */
+void
+xfs_dir2_leaf_log_ents(
+	xfs_trans_t		*tp,		/* transaction pointer */
+	xfs_dabuf_t		*bp,		/* leaf buffer */
+	int			first,		/* first entry to log */
+	int			last)		/* last entry to log */
+{
+	xfs_dir2_leaf_entry_t	*firstlep;	/* pointer to first entry */
+	xfs_dir2_leaf_entry_t	*lastlep;	/* pointer to last entry */
+	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
+
+	leaf = bp->data;
+	ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_DIR2_LEAF1_MAGIC ||
+	       be16_to_cpu(leaf->hdr.info.magic) == XFS_DIR2_LEAFN_MAGIC);
+	firstlep = &leaf->ents[first];
+	lastlep = &leaf->ents[last];
+	xfs_da_log_buf(tp, bp, (uint)((char *)firstlep - (char *)leaf),
+		(uint)((char *)lastlep - (char *)leaf + sizeof(*lastlep) - 1));
+}
+
+/*
+ * Log the header of the leaf1 or leafn block.
+ */
+void
+xfs_dir2_leaf_log_header(
+	xfs_trans_t		*tp,		/* transaction pointer */
+	xfs_dabuf_t		*bp)		/* leaf buffer */
+{
+	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
+
+	leaf = bp->data;
+	ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_DIR2_LEAF1_MAGIC ||
+	       be16_to_cpu(leaf->hdr.info.magic) == XFS_DIR2_LEAFN_MAGIC);
+	xfs_da_log_buf(tp, bp, (uint)((char *)&leaf->hdr - (char *)leaf),
+		(uint)(sizeof(leaf->hdr) - 1));
+}
+
+/*
+ * Log the tail of the leaf1 block.
+ */
+STATIC void
+xfs_dir2_leaf_log_tail(
+	xfs_trans_t		*tp,		/* transaction pointer */
+	xfs_dabuf_t		*bp)		/* leaf buffer */
+{
+	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
+	xfs_dir2_leaf_tail_t	*ltp;		/* leaf tail structure */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+
+	mp = tp->t_mountp;
+	leaf = bp->data;
+	ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_DIR2_LEAF1_MAGIC);
+	ltp = xfs_dir2_leaf_tail_p(mp, leaf);
+	xfs_da_log_buf(tp, bp, (uint)((char *)ltp - (char *)leaf),
+		(uint)(mp->m_dirblksize - 1));
+}
+
+/*
+ * Look up the entry referred to by args in the leaf format directory.
+ * Most of the work is done by the xfs_dir2_leaf_lookup_int routine which
+ * is also used by the node-format code.
+ */
+int
+xfs_dir2_leaf_lookup(
+	xfs_da_args_t		*args)		/* operation arguments */
+{
+	xfs_dabuf_t		*dbp;		/* data block buffer */
+	xfs_dir2_data_entry_t	*dep;		/* data block entry */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	int			error;		/* error return code */
+	int			index;		/* found entry index */
+	xfs_dabuf_t		*lbp;		/* leaf buffer */
+	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
+	xfs_dir2_leaf_entry_t	*lep;		/* leaf entry */
+	xfs_trans_t		*tp;		/* transaction pointer */
+
+	trace_xfs_dir2_leaf_lookup(args);
+
+	/*
+	 * Look up name in the leaf block, returning both buffers and index.
+	 */
+	if ((error = xfs_dir2_leaf_lookup_int(args, &lbp, &index, &dbp))) {
+		return error;
+	}
+	tp = args->trans;
+	dp = args->dp;
+	xfs_dir2_leaf_check(dp, lbp);
+	leaf = lbp->data;
+	/*
+	 * Get to the leaf entry and contained data entry address.
+	 */
+	lep = &leaf->ents[index];
+	/*
+	 * Point to the data entry.
+	 */
+	dep = (xfs_dir2_data_entry_t *)
+	      ((char *)dbp->data +
+	       xfs_dir2_dataptr_to_off(dp->i_mount, be32_to_cpu(lep->address)));
+	/*
+	 * Return the found inode number & CI name if appropriate
+	 */
+	args->inumber = be64_to_cpu(dep->inumber);
+	error = xfs_dir_cilookup_result(args, dep->name, dep->namelen);
+	xfs_da_brelse(tp, dbp);
+	xfs_da_brelse(tp, lbp);
+	return XFS_ERROR(error);
+}
+
+/*
+ * Look up name/hash in the leaf block.
+ * Fill in indexp with the found index, and dbpp with the data buffer.
+ * If not found dbpp will be NULL, and ENOENT comes back.
+ * lbpp will always be filled in with the leaf buffer unless there's an error.
+ */
+static int					/* error */
+xfs_dir2_leaf_lookup_int(
+	xfs_da_args_t		*args,		/* operation arguments */
+	xfs_dabuf_t		**lbpp,		/* out: leaf buffer */
+	int			*indexp,	/* out: index in leaf block */
+	xfs_dabuf_t		**dbpp)		/* out: data buffer */
+{
+	xfs_dir2_db_t		curdb = -1;	/* current data block number */
+	xfs_dabuf_t		*dbp = NULL;	/* data buffer */
+	xfs_dir2_data_entry_t	*dep;		/* data entry */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	int			error;		/* error return code */
+	int			index;		/* index in leaf block */
+	xfs_dabuf_t		*lbp;		/* leaf buffer */
+	xfs_dir2_leaf_entry_t	*lep;		/* leaf entry */
+	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+	xfs_dir2_db_t		newdb;		/* new data block number */
+	xfs_trans_t		*tp;		/* transaction pointer */
+	xfs_dir2_db_t		cidb = -1;	/* case match data block no. */
+	enum xfs_dacmp		cmp;		/* name compare result */
+
+	dp = args->dp;
+	tp = args->trans;
+	mp = dp->i_mount;
+	/*
+	 * Read the leaf block into the buffer.
+	 */
+	error = xfs_da_read_buf(tp, dp, mp->m_dirleafblk, -1, &lbp,
+							XFS_DATA_FORK);
+	if (error)
+		return error;
+	*lbpp = lbp;
+	leaf = lbp->data;
+	xfs_dir2_leaf_check(dp, lbp);
+	/*
+	 * Look for the first leaf entry with our hash value.
+	 */
+	index = xfs_dir2_leaf_search_hash(args, lbp);
+	/*
+	 * Loop over all the entries with the right hash value
+	 * looking to match the name.
+	 */
+	for (lep = &leaf->ents[index]; index < be16_to_cpu(leaf->hdr.count) &&
+				be32_to_cpu(lep->hashval) == args->hashval;
+				lep++, index++) {
+		/*
+		 * Skip over stale leaf entries.
+		 */
+		if (be32_to_cpu(lep->address) == XFS_DIR2_NULL_DATAPTR)
+			continue;
+		/*
+		 * Get the new data block number.
+		 */
+		newdb = xfs_dir2_dataptr_to_db(mp, be32_to_cpu(lep->address));
+		/*
+		 * If it's not the same as the old data block number,
+		 * need to pitch the old one and read the new one.
+		 */
+		if (newdb != curdb) {
+			if (dbp)
+				xfs_da_brelse(tp, dbp);
+			error = xfs_da_read_buf(tp, dp,
+						xfs_dir2_db_to_da(mp, newdb),
+						-1, &dbp, XFS_DATA_FORK);
+			if (error) {
+				xfs_da_brelse(tp, lbp);
+				return error;
+			}
+			xfs_dir2_data_check(dp, dbp);
+			curdb = newdb;
+		}
+		/*
+		 * Point to the data entry.
+		 */
+		dep = (xfs_dir2_data_entry_t *)((char *)dbp->data +
+			xfs_dir2_dataptr_to_off(mp, be32_to_cpu(lep->address)));
+		/*
+		 * Compare name and if it's an exact match, return the index
+		 * and buffer. If it's the first case-insensitive match, store
+		 * the index and buffer and continue looking for an exact match.
+		 */
+		cmp = mp->m_dirnameops->compname(args, dep->name, dep->namelen);
+		if (cmp != XFS_CMP_DIFFERENT && cmp != args->cmpresult) {
+			args->cmpresult = cmp;
+			*indexp = index;
+			/* case exact match: return the current buffer. */
+			if (cmp == XFS_CMP_EXACT) {
+				*dbpp = dbp;
+				return 0;
+			}
+			cidb = curdb;
+		}
+	}
+	ASSERT(args->op_flags & XFS_DA_OP_OKNOENT);
+	/*
+	 * Here, we can only be doing a lookup (not a rename or remove).
+	 * If a case-insensitive match was found earlier, re-read the
+	 * appropriate data block if required and return it.
+	 */
+	if (args->cmpresult == XFS_CMP_CASE) {
+		ASSERT(cidb != -1);
+		if (cidb != curdb) {
+			xfs_da_brelse(tp, dbp);
+			error = xfs_da_read_buf(tp, dp,
+						xfs_dir2_db_to_da(mp, cidb),
+						-1, &dbp, XFS_DATA_FORK);
+			if (error) {
+				xfs_da_brelse(tp, lbp);
+				return error;
+			}
+		}
+		*dbpp = dbp;
+		return 0;
+	}
+	/*
+	 * No match found, return ENOENT.
+	 */
+	ASSERT(cidb == -1);
+	if (dbp)
+		xfs_da_brelse(tp, dbp);
+	xfs_da_brelse(tp, lbp);
+	return XFS_ERROR(ENOENT);
+}
+
+/*
+ * Remove an entry from a leaf format directory.
+ */
+int						/* error */
+xfs_dir2_leaf_removename(
+	xfs_da_args_t		*args)		/* operation arguments */
+{
+	__be16			*bestsp;	/* leaf block best freespace */
+	xfs_dir2_data_t		*data;		/* data block structure */
+	xfs_dir2_db_t		db;		/* data block number */
+	xfs_dabuf_t		*dbp;		/* data block buffer */
+	xfs_dir2_data_entry_t	*dep;		/* data entry structure */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	int			error;		/* error return code */
+	xfs_dir2_db_t		i;		/* temporary data block # */
+	int			index;		/* index into leaf entries */
+	xfs_dabuf_t		*lbp;		/* leaf buffer */
+	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
+	xfs_dir2_leaf_entry_t	*lep;		/* leaf entry */
+	xfs_dir2_leaf_tail_t	*ltp;		/* leaf tail structure */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+	int			needlog;	/* need to log data header */
+	int			needscan;	/* need to rescan data frees */
+	xfs_dir2_data_off_t	oldbest;	/* old value of best free */
+	xfs_trans_t		*tp;		/* transaction pointer */
+
+	trace_xfs_dir2_leaf_removename(args);
+
+	/*
+	 * Lookup the leaf entry, get the leaf and data blocks read in.
+	 */
+	if ((error = xfs_dir2_leaf_lookup_int(args, &lbp, &index, &dbp))) {
+		return error;
+	}
+	dp = args->dp;
+	tp = args->trans;
+	mp = dp->i_mount;
+	leaf = lbp->data;
+	data = dbp->data;
+	xfs_dir2_data_check(dp, dbp);
+	/*
+	 * Point to the leaf entry, use that to point to the data entry.
+	 */
+	lep = &leaf->ents[index];
+	db = xfs_dir2_dataptr_to_db(mp, be32_to_cpu(lep->address));
+	dep = (xfs_dir2_data_entry_t *)
+	      ((char *)data + xfs_dir2_dataptr_to_off(mp, be32_to_cpu(lep->address)));
+	needscan = needlog = 0;
+	oldbest = be16_to_cpu(data->hdr.bestfree[0].length);
+	ltp = xfs_dir2_leaf_tail_p(mp, leaf);
+	bestsp = xfs_dir2_leaf_bests_p(ltp);
+	ASSERT(be16_to_cpu(bestsp[db]) == oldbest);
+	/*
+	 * Mark the former data entry unused.
+	 */
+	xfs_dir2_data_make_free(tp, dbp,
+		(xfs_dir2_data_aoff_t)((char *)dep - (char *)data),
+		xfs_dir2_data_entsize(dep->namelen), &needlog, &needscan);
+	/*
+	 * We just mark the leaf entry stale by putting a null in it.
+	 */
+	be16_add_cpu(&leaf->hdr.stale, 1);
+	xfs_dir2_leaf_log_header(tp, lbp);
+	lep->address = cpu_to_be32(XFS_DIR2_NULL_DATAPTR);
+	xfs_dir2_leaf_log_ents(tp, lbp, index, index);
+	/*
+	 * Scan the freespace in the data block again if necessary,
+	 * log the data block header if necessary.
+	 */
+	if (needscan)
+		xfs_dir2_data_freescan(mp, data, &needlog);
+	if (needlog)
+		xfs_dir2_data_log_header(tp, dbp);
+	/*
+	 * If the longest freespace in the data block has changed,
+	 * put the new value in the bests table and log that.
+	 */
+	if (be16_to_cpu(data->hdr.bestfree[0].length) != oldbest) {
+		bestsp[db] = data->hdr.bestfree[0].length;
+		xfs_dir2_leaf_log_bests(tp, lbp, db, db);
+	}
+	xfs_dir2_data_check(dp, dbp);
+	/*
+	 * If the data block is now empty then get rid of the data block.
+	 */
+	if (be16_to_cpu(data->hdr.bestfree[0].length) ==
+	    mp->m_dirblksize - (uint)sizeof(data->hdr)) {
+		ASSERT(db != mp->m_dirdatablk);
+		if ((error = xfs_dir2_shrink_inode(args, db, dbp))) {
+			/*
+			 * Nope, can't get rid of it because it caused
+			 * allocation of a bmap btree block to do so.
+			 * Just go on, returning success, leaving the
+			 * empty block in place.
+			 */
+			if (error == ENOSPC && args->total == 0) {
+				xfs_da_buf_done(dbp);
+				error = 0;
+			}
+			xfs_dir2_leaf_check(dp, lbp);
+			xfs_da_buf_done(lbp);
+			return error;
+		}
+		dbp = NULL;
+		/*
+		 * If this is the last data block then compact the
+		 * bests table by getting rid of entries.
+		 */
+		if (db == be32_to_cpu(ltp->bestcount) - 1) {
+			/*
+			 * Look for the last active entry (i).
+			 */
+			for (i = db - 1; i > 0; i--) {
+				if (be16_to_cpu(bestsp[i]) != NULLDATAOFF)
+					break;
+			}
+			/*
+			 * Copy the table down so inactive entries at the
+			 * end are removed.
+			 */
+			memmove(&bestsp[db - i], bestsp,
+				(be32_to_cpu(ltp->bestcount) - (db - i)) * sizeof(*bestsp));
+			be32_add_cpu(&ltp->bestcount, -(db - i));
+			xfs_dir2_leaf_log_tail(tp, lbp);
+			xfs_dir2_leaf_log_bests(tp, lbp, 0, be32_to_cpu(ltp->bestcount) - 1);
+		} else
+			bestsp[db] = cpu_to_be16(NULLDATAOFF);
+	}
+	/*
+	 * If the data block was not the first one, drop it.
+	 */
+	else if (db != mp->m_dirdatablk && dbp != NULL) {
+		xfs_da_buf_done(dbp);
+		dbp = NULL;
+	}
+	xfs_dir2_leaf_check(dp, lbp);
+	/*
+	 * See if we can convert to block form.
+	 */
+	return xfs_dir2_leaf_to_block(args, lbp, dbp);
+}
+
+/*
+ * Replace the inode number in a leaf format directory entry.
+ */
+int						/* error */
+xfs_dir2_leaf_replace(
+	xfs_da_args_t		*args)		/* operation arguments */
+{
+	xfs_dabuf_t		*dbp;		/* data block buffer */
+	xfs_dir2_data_entry_t	*dep;		/* data block entry */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	int			error;		/* error return code */
+	int			index;		/* index of leaf entry */
+	xfs_dabuf_t		*lbp;		/* leaf buffer */
+	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
+	xfs_dir2_leaf_entry_t	*lep;		/* leaf entry */
+	xfs_trans_t		*tp;		/* transaction pointer */
+
+	trace_xfs_dir2_leaf_replace(args);
+
+	/*
+	 * Look up the entry.
+	 */
+	if ((error = xfs_dir2_leaf_lookup_int(args, &lbp, &index, &dbp))) {
+		return error;
+	}
+	dp = args->dp;
+	leaf = lbp->data;
+	/*
+	 * Point to the leaf entry, get data address from it.
+	 */
+	lep = &leaf->ents[index];
+	/*
+	 * Point to the data entry.
+	 */
+	dep = (xfs_dir2_data_entry_t *)
+	      ((char *)dbp->data +
+	       xfs_dir2_dataptr_to_off(dp->i_mount, be32_to_cpu(lep->address)));
+	ASSERT(args->inumber != be64_to_cpu(dep->inumber));
+	/*
+	 * Put the new inode number in, log it.
+	 */
+	dep->inumber = cpu_to_be64(args->inumber);
+	tp = args->trans;
+	xfs_dir2_data_log_entry(tp, dbp, dep);
+	xfs_da_buf_done(dbp);
+	xfs_dir2_leaf_check(dp, lbp);
+	xfs_da_brelse(tp, lbp);
+	return 0;
+}
+
+/*
+ * Return index in the leaf block (lbp) which is either the first
+ * one with this hash value, or if there are none, the insert point
+ * for that hash value.
+ */
+int						/* index value */
+xfs_dir2_leaf_search_hash(
+	xfs_da_args_t		*args,		/* operation arguments */
+	xfs_dabuf_t		*lbp)		/* leaf buffer */
+{
+	xfs_dahash_t		hash=0;		/* hash from this entry */
+	xfs_dahash_t		hashwant;	/* hash value looking for */
+	int			high;		/* high leaf index */
+	int			low;		/* low leaf index */
+	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
+	xfs_dir2_leaf_entry_t	*lep;		/* leaf entry */
+	int			mid=0;		/* current leaf index */
+
+	leaf = lbp->data;
+#ifndef __KERNEL__
+	if (!leaf->hdr.count)
+		return 0;
+#endif
+	/*
+	 * Note, the table cannot be empty, so we have to go through the loop.
+	 * Binary search the leaf entries looking for our hash value.
+	 */
+	for (lep = leaf->ents, low = 0, high = be16_to_cpu(leaf->hdr.count) - 1,
+		hashwant = args->hashval;
+	     low <= high; ) {
+		mid = (low + high) >> 1;
+		if ((hash = be32_to_cpu(lep[mid].hashval)) == hashwant)
+			break;
+		if (hash < hashwant)
+			low = mid + 1;
+		else
+			high = mid - 1;
+	}
+	/*
+	 * Found one, back up through all the equal hash values.
+	 */
+	if (hash == hashwant) {
+		while (mid > 0 && be32_to_cpu(lep[mid - 1].hashval) == hashwant) {
+			mid--;
+		}
+	}
+	/*
+	 * Need to point to an entry higher than ours.
+	 */
+	else if (hash < hashwant)
+		mid++;
+	return mid;
+}
+
+/*
+ * Trim off a trailing data block.  We know it's empty since the leaf
+ * freespace table says so.
+ */
+int						/* error */
+xfs_dir2_leaf_trim_data(
+	xfs_da_args_t		*args,		/* operation arguments */
+	xfs_dabuf_t		*lbp,		/* leaf buffer */
+	xfs_dir2_db_t		db)		/* data block number */
+{
+	__be16			*bestsp;	/* leaf bests table */
+#ifdef DEBUG
+	xfs_dir2_data_t		*data;		/* data block structure */
+#endif
+	xfs_dabuf_t		*dbp;		/* data block buffer */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	int			error;		/* error return value */
+	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
+	xfs_dir2_leaf_tail_t	*ltp;		/* leaf tail structure */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+	xfs_trans_t		*tp;		/* transaction pointer */
+
+	dp = args->dp;
+	mp = dp->i_mount;
+	tp = args->trans;
+	/*
+	 * Read the offending data block.  We need its buffer.
+	 */
+	if ((error = xfs_da_read_buf(tp, dp, xfs_dir2_db_to_da(mp, db), -1, &dbp,
+			XFS_DATA_FORK))) {
+		return error;
+	}
+#ifdef DEBUG
+	data = dbp->data;
+	ASSERT(be32_to_cpu(data->hdr.magic) == XFS_DIR2_DATA_MAGIC);
+#endif
+	/* this seems to be an error
+	 * data is only valid if DEBUG is defined?
+	 * RMC 09/08/1999
+	 */
+
+	leaf = lbp->data;
+	ltp = xfs_dir2_leaf_tail_p(mp, leaf);
+	ASSERT(be16_to_cpu(data->hdr.bestfree[0].length) ==
+	       mp->m_dirblksize - (uint)sizeof(data->hdr));
+	ASSERT(db == be32_to_cpu(ltp->bestcount) - 1);
+	/*
+	 * Get rid of the data block.
+	 */
+	if ((error = xfs_dir2_shrink_inode(args, db, dbp))) {
+		ASSERT(error != ENOSPC);
+		xfs_da_brelse(tp, dbp);
+		return error;
+	}
+	/*
+	 * Eliminate the last bests entry from the table.
+	 */
+	bestsp = xfs_dir2_leaf_bests_p(ltp);
+	be32_add_cpu(&ltp->bestcount, -1);
+	memmove(&bestsp[1], &bestsp[0], be32_to_cpu(ltp->bestcount) * sizeof(*bestsp));
+	xfs_dir2_leaf_log_tail(tp, lbp);
+	xfs_dir2_leaf_log_bests(tp, lbp, 0, be32_to_cpu(ltp->bestcount) - 1);
+	return 0;
+}
+
+/*
+ * Convert node form directory to leaf form directory.
+ * The root of the node form dir needs to already be a LEAFN block.
+ * Just return if we can't do anything.
+ */
+int						/* error */
+xfs_dir2_node_to_leaf(
+	xfs_da_state_t		*state)		/* directory operation state */
+{
+	xfs_da_args_t		*args;		/* operation arguments */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	int			error;		/* error return code */
+	xfs_dabuf_t		*fbp;		/* buffer for freespace block */
+	xfs_fileoff_t		fo;		/* freespace file offset */
+	xfs_dir2_free_t		*free;		/* freespace structure */
+	xfs_dabuf_t		*lbp;		/* buffer for leaf block */
+	xfs_dir2_leaf_tail_t	*ltp;		/* tail of leaf structure */
+	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+	int			rval;		/* successful free trim? */
+	xfs_trans_t		*tp;		/* transaction pointer */
+
+	/*
+	 * There's more than a leaf level in the btree, so there must
+	 * be multiple leafn blocks.  Give up.
+	 */
+	if (state->path.active > 1)
+		return 0;
+	args = state->args;
+
+	trace_xfs_dir2_node_to_leaf(args);
+
+	mp = state->mp;
+	dp = args->dp;
+	tp = args->trans;
+	/*
+	 * Get the last offset in the file.
+	 */
+	if ((error = xfs_bmap_last_offset(tp, dp, &fo, XFS_DATA_FORK))) {
+		return error;
+	}
+	fo -= mp->m_dirblkfsbs;
+	/*
+	 * If there are freespace blocks other than the first one,
+	 * take this opportunity to remove trailing empty freespace blocks
+	 * that may have been left behind during no-space-reservation
+	 * operations.
+	 */
+	while (fo > mp->m_dirfreeblk) {
+		if ((error = xfs_dir2_node_trim_free(args, fo, &rval))) {
+			return error;
+		}
+		if (rval)
+			fo -= mp->m_dirblkfsbs;
+		else
+			return 0;
+	}
+	/*
+	 * Now find the block just before the freespace block.
+	 */
+	if ((error = xfs_bmap_last_before(tp, dp, &fo, XFS_DATA_FORK))) {
+		return error;
+	}
+	/*
+	 * If it's not the single leaf block, give up.
+	 */
+	if (XFS_FSB_TO_B(mp, fo) > XFS_DIR2_LEAF_OFFSET + mp->m_dirblksize)
+		return 0;
+	lbp = state->path.blk[0].bp;
+	leaf = lbp->data;
+	ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_DIR2_LEAFN_MAGIC);
+	/*
+	 * Read the freespace block.
+	 */
+	if ((error = xfs_da_read_buf(tp, dp, mp->m_dirfreeblk, -1, &fbp,
+			XFS_DATA_FORK))) {
+		return error;
+	}
+	free = fbp->data;
+	ASSERT(be32_to_cpu(free->hdr.magic) == XFS_DIR2_FREE_MAGIC);
+	ASSERT(!free->hdr.firstdb);
+	/*
+	 * Now see if the leafn and free data will fit in a leaf1.
+	 * If not, release the buffer and give up.
+	 */
+	if ((uint)sizeof(leaf->hdr) +
+	    (be16_to_cpu(leaf->hdr.count) - be16_to_cpu(leaf->hdr.stale)) * (uint)sizeof(leaf->ents[0]) +
+	    be32_to_cpu(free->hdr.nvalid) * (uint)sizeof(leaf->bests[0]) +
+	    (uint)sizeof(leaf->tail) >
+	    mp->m_dirblksize) {
+		xfs_da_brelse(tp, fbp);
+		return 0;
+	}
+	/*
+	 * If the leaf has any stale entries in it, compress them out.
+	 * The compact routine will log the header.
+	 */
+	if (be16_to_cpu(leaf->hdr.stale))
+		xfs_dir2_leaf_compact(args, lbp);
+	else
+		xfs_dir2_leaf_log_header(tp, lbp);
+	leaf->hdr.info.magic = cpu_to_be16(XFS_DIR2_LEAF1_MAGIC);
+	/*
+	 * Set up the leaf tail from the freespace block.
+	 */
+	ltp = xfs_dir2_leaf_tail_p(mp, leaf);
+	ltp->bestcount = free->hdr.nvalid;
+	/*
+	 * Set up the leaf bests table.
+	 */
+	memcpy(xfs_dir2_leaf_bests_p(ltp), free->bests,
+		be32_to_cpu(ltp->bestcount) * sizeof(leaf->bests[0]));
+	xfs_dir2_leaf_log_bests(tp, lbp, 0, be32_to_cpu(ltp->bestcount) - 1);
+	xfs_dir2_leaf_log_tail(tp, lbp);
+	xfs_dir2_leaf_check(dp, lbp);
+	/*
+	 * Get rid of the freespace block.
+	 */
+	error = xfs_dir2_shrink_inode(args, XFS_DIR2_FREE_FIRSTDB(mp), fbp);
+	if (error) {
+		/*
+		 * This can't fail here because it can only happen when
+		 * punching out the middle of an extent, and this is an
+		 * isolated block.
+		 */
+		ASSERT(error != ENOSPC);
+		return error;
+	}
+	fbp = NULL;
+	/*
+	 * Now see if we can convert the single-leaf directory
+	 * down to a block form directory.
+	 * This routine always kills the dabuf for the leaf, so
+	 * eliminate it from the path.
+	 */
+	error = xfs_dir2_leaf_to_block(args, lbp, NULL);
+	state->path.blk[0].bp = NULL;
+	return error;
+}
diff --git a/fs/xfs/xfs_dir2_leaf.h b/fs/xfs/xfs_dir2_leaf.h
new file mode 100644
index 00000000..6c9539f0
--- /dev/null
+++ b/fs/xfs/xfs_dir2_leaf.h
@@ -0,0 +1,253 @@
+/*
+ * Copyright (c) 2000-2001,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_DIR2_LEAF_H__
+#define	__XFS_DIR2_LEAF_H__
+
+struct uio;
+struct xfs_dabuf;
+struct xfs_da_args;
+struct xfs_inode;
+struct xfs_mount;
+struct xfs_trans;
+
+/*
+ * Offset of the leaf/node space.  First block in this space
+ * is the btree root.
+ */
+#define	XFS_DIR2_LEAF_SPACE	1
+#define	XFS_DIR2_LEAF_OFFSET	(XFS_DIR2_LEAF_SPACE * XFS_DIR2_SPACE_SIZE)
+#define	XFS_DIR2_LEAF_FIRSTDB(mp)	\
+	xfs_dir2_byte_to_db(mp, XFS_DIR2_LEAF_OFFSET)
+
+/*
+ * Offset in data space of a data entry.
+ */
+typedef	__uint32_t	xfs_dir2_dataptr_t;
+#define	XFS_DIR2_MAX_DATAPTR	((xfs_dir2_dataptr_t)0xffffffff)
+#define	XFS_DIR2_NULL_DATAPTR	((xfs_dir2_dataptr_t)0)
+
+/*
+ * Leaf block header.
+ */
+typedef struct xfs_dir2_leaf_hdr {
+	xfs_da_blkinfo_t	info;		/* header for da routines */
+	__be16			count;		/* count of entries */
+	__be16			stale;		/* count of stale entries */
+} xfs_dir2_leaf_hdr_t;
+
+/*
+ * Leaf block entry.
+ */
+typedef struct xfs_dir2_leaf_entry {
+	__be32			hashval;	/* hash value of name */
+	__be32			address;	/* address of data entry */
+} xfs_dir2_leaf_entry_t;
+
+/*
+ * Leaf block tail.
+ */
+typedef struct xfs_dir2_leaf_tail {
+	__be32			bestcount;
+} xfs_dir2_leaf_tail_t;
+
+/*
+ * Leaf block.
+ * bests and tail are at the end of the block for single-leaf only
+ * (magic = XFS_DIR2_LEAF1_MAGIC not XFS_DIR2_LEAFN_MAGIC).
+ */
+typedef struct xfs_dir2_leaf {
+	xfs_dir2_leaf_hdr_t	hdr;		/* leaf header */
+	xfs_dir2_leaf_entry_t	ents[1];	/* entries */
+						/* ... */
+	xfs_dir2_data_off_t	bests[1];	/* best free counts */
+	xfs_dir2_leaf_tail_t	tail;		/* leaf tail */
+} xfs_dir2_leaf_t;
+
+/*
+ * DB blocks here are logical directory block numbers, not filesystem blocks.
+ */
+
+static inline int xfs_dir2_max_leaf_ents(struct xfs_mount *mp)
+{
+	return (int)(((mp)->m_dirblksize - (uint)sizeof(xfs_dir2_leaf_hdr_t)) /
+	       (uint)sizeof(xfs_dir2_leaf_entry_t));
+}
+
+/*
+ * Get address of the bestcount field in the single-leaf block.
+ */
+static inline xfs_dir2_leaf_tail_t *
+xfs_dir2_leaf_tail_p(struct xfs_mount *mp, xfs_dir2_leaf_t *lp)
+{
+	return (xfs_dir2_leaf_tail_t *)
+		((char *)(lp) + (mp)->m_dirblksize - 
+		  (uint)sizeof(xfs_dir2_leaf_tail_t));
+}
+
+/*
+ * Get address of the bests array in the single-leaf block.
+ */
+static inline __be16 *
+xfs_dir2_leaf_bests_p(xfs_dir2_leaf_tail_t *ltp)
+{
+	return (__be16 *)ltp - be32_to_cpu(ltp->bestcount);
+}
+
+/*
+ * Convert dataptr to byte in file space
+ */
+static inline xfs_dir2_off_t
+xfs_dir2_dataptr_to_byte(struct xfs_mount *mp, xfs_dir2_dataptr_t dp)
+{
+	return (xfs_dir2_off_t)(dp) << XFS_DIR2_DATA_ALIGN_LOG;
+}
+
+/*
+ * Convert byte in file space to dataptr.  It had better be aligned.
+ */
+static inline xfs_dir2_dataptr_t
+xfs_dir2_byte_to_dataptr(struct xfs_mount *mp, xfs_dir2_off_t by)
+{
+	return (xfs_dir2_dataptr_t)((by) >> XFS_DIR2_DATA_ALIGN_LOG);
+}
+
+/*
+ * Convert byte in space to (DB) block
+ */
+static inline xfs_dir2_db_t
+xfs_dir2_byte_to_db(struct xfs_mount *mp, xfs_dir2_off_t by)
+{
+	return (xfs_dir2_db_t)((by) >> \
+		 ((mp)->m_sb.sb_blocklog + (mp)->m_sb.sb_dirblklog));
+}
+
+/*
+ * Convert dataptr to a block number
+ */
+static inline xfs_dir2_db_t
+xfs_dir2_dataptr_to_db(struct xfs_mount *mp, xfs_dir2_dataptr_t dp)
+{
+	return xfs_dir2_byte_to_db(mp, xfs_dir2_dataptr_to_byte(mp, dp));
+}
+
+/*
+ * Convert byte in space to offset in a block
+ */
+static inline xfs_dir2_data_aoff_t
+xfs_dir2_byte_to_off(struct xfs_mount *mp, xfs_dir2_off_t by)
+{
+	return (xfs_dir2_data_aoff_t)((by) & \
+		((1 << ((mp)->m_sb.sb_blocklog + (mp)->m_sb.sb_dirblklog)) - 1));
+}
+
+/*
+ * Convert dataptr to a byte offset in a block
+ */
+static inline xfs_dir2_data_aoff_t
+xfs_dir2_dataptr_to_off(struct xfs_mount *mp, xfs_dir2_dataptr_t dp)
+{
+	return xfs_dir2_byte_to_off(mp, xfs_dir2_dataptr_to_byte(mp, dp));
+}
+
+/*
+ * Convert block and offset to byte in space
+ */
+static inline xfs_dir2_off_t
+xfs_dir2_db_off_to_byte(struct xfs_mount *mp, xfs_dir2_db_t db,
+			xfs_dir2_data_aoff_t o)
+{
+	return ((xfs_dir2_off_t)(db) << \
+		((mp)->m_sb.sb_blocklog + (mp)->m_sb.sb_dirblklog)) + (o);
+}
+
+/*
+ * Convert block (DB) to block (dablk)
+ */
+static inline xfs_dablk_t
+xfs_dir2_db_to_da(struct xfs_mount *mp, xfs_dir2_db_t db)
+{
+	return (xfs_dablk_t)((db) << (mp)->m_sb.sb_dirblklog);
+}
+
+/*
+ * Convert byte in space to (DA) block
+ */
+static inline xfs_dablk_t
+xfs_dir2_byte_to_da(struct xfs_mount *mp, xfs_dir2_off_t by)
+{
+	return xfs_dir2_db_to_da(mp, xfs_dir2_byte_to_db(mp, by));
+}
+
+/*
+ * Convert block and offset to dataptr
+ */
+static inline xfs_dir2_dataptr_t
+xfs_dir2_db_off_to_dataptr(struct xfs_mount *mp, xfs_dir2_db_t db,
+			   xfs_dir2_data_aoff_t o)
+{
+	return xfs_dir2_byte_to_dataptr(mp, xfs_dir2_db_off_to_byte(mp, db, o));
+}
+
+/*
+ * Convert block (dablk) to block (DB)
+ */
+static inline xfs_dir2_db_t
+xfs_dir2_da_to_db(struct xfs_mount *mp, xfs_dablk_t da)
+{
+	return (xfs_dir2_db_t)((da) >> (mp)->m_sb.sb_dirblklog);
+}
+
+/*
+ * Convert block (dablk) to byte offset in space
+ */
+static inline xfs_dir2_off_t
+xfs_dir2_da_to_byte(struct xfs_mount *mp, xfs_dablk_t da)
+{
+	return xfs_dir2_db_off_to_byte(mp, xfs_dir2_da_to_db(mp, da), 0);
+}
+
+/*
+ * Function declarations.
+ */
+extern int xfs_dir2_block_to_leaf(struct xfs_da_args *args,
+				  struct xfs_dabuf *dbp);
+extern int xfs_dir2_leaf_addname(struct xfs_da_args *args);
+extern void xfs_dir2_leaf_compact(struct xfs_da_args *args,
+				  struct xfs_dabuf *bp);
+extern void xfs_dir2_leaf_compact_x1(struct xfs_dabuf *bp, int *indexp,
+				     int *lowstalep, int *highstalep,
+				     int *lowlogp, int *highlogp);
+extern int xfs_dir2_leaf_getdents(struct xfs_inode *dp, void *dirent,
+				  size_t bufsize, xfs_off_t *offset,
+				  filldir_t filldir);
+extern int xfs_dir2_leaf_init(struct xfs_da_args *args, xfs_dir2_db_t bno,
+			      struct xfs_dabuf **bpp, int magic);
+extern void xfs_dir2_leaf_log_ents(struct xfs_trans *tp, struct xfs_dabuf *bp,
+				   int first, int last);
+extern void xfs_dir2_leaf_log_header(struct xfs_trans *tp,
+				     struct xfs_dabuf *bp);
+extern int xfs_dir2_leaf_lookup(struct xfs_da_args *args);
+extern int xfs_dir2_leaf_removename(struct xfs_da_args *args);
+extern int xfs_dir2_leaf_replace(struct xfs_da_args *args);
+extern int xfs_dir2_leaf_search_hash(struct xfs_da_args *args,
+				     struct xfs_dabuf *lbp);
+extern int xfs_dir2_leaf_trim_data(struct xfs_da_args *args,
+				   struct xfs_dabuf *lbp, xfs_dir2_db_t db);
+extern int xfs_dir2_node_to_leaf(struct xfs_da_state *state);
+
+#endif	/* __XFS_DIR2_LEAF_H__ */
diff --git a/fs/xfs/xfs_dir2_node.c b/fs/xfs/xfs_dir2_node.c
new file mode 100644
index 00000000..a0aab7d3
--- /dev/null
+++ b/fs/xfs/xfs_dir2_node.c
@@ -0,0 +1,2076 @@
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_dir2.h"
+#include "xfs_mount.h"
+#include "xfs_da_btree.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_dir2_sf.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_bmap.h"
+#include "xfs_dir2_data.h"
+#include "xfs_dir2_leaf.h"
+#include "xfs_dir2_block.h"
+#include "xfs_dir2_node.h"
+#include "xfs_error.h"
+#include "xfs_trace.h"
+
+/*
+ * Function declarations.
+ */
+static void xfs_dir2_free_log_header(xfs_trans_t *tp, xfs_dabuf_t *bp);
+static int xfs_dir2_leafn_add(xfs_dabuf_t *bp, xfs_da_args_t *args, int index);
+#ifdef DEBUG
+static void xfs_dir2_leafn_check(xfs_inode_t *dp, xfs_dabuf_t *bp);
+#else
+#define	xfs_dir2_leafn_check(dp, bp)
+#endif
+static void xfs_dir2_leafn_moveents(xfs_da_args_t *args, xfs_dabuf_t *bp_s,
+				    int start_s, xfs_dabuf_t *bp_d, int start_d,
+				    int count);
+static void xfs_dir2_leafn_rebalance(xfs_da_state_t *state,
+				     xfs_da_state_blk_t *blk1,
+				     xfs_da_state_blk_t *blk2);
+static int xfs_dir2_leafn_remove(xfs_da_args_t *args, xfs_dabuf_t *bp,
+				 int index, xfs_da_state_blk_t *dblk,
+				 int *rval);
+static int xfs_dir2_node_addname_int(xfs_da_args_t *args,
+				     xfs_da_state_blk_t *fblk);
+
+/*
+ * Log entries from a freespace block.
+ */
+STATIC void
+xfs_dir2_free_log_bests(
+	xfs_trans_t		*tp,		/* transaction pointer */
+	xfs_dabuf_t		*bp,		/* freespace buffer */
+	int			first,		/* first entry to log */
+	int			last)		/* last entry to log */
+{
+	xfs_dir2_free_t		*free;		/* freespace structure */
+
+	free = bp->data;
+	ASSERT(be32_to_cpu(free->hdr.magic) == XFS_DIR2_FREE_MAGIC);
+	xfs_da_log_buf(tp, bp,
+		(uint)((char *)&free->bests[first] - (char *)free),
+		(uint)((char *)&free->bests[last] - (char *)free +
+		       sizeof(free->bests[0]) - 1));
+}
+
+/*
+ * Log header from a freespace block.
+ */
+static void
+xfs_dir2_free_log_header(
+	xfs_trans_t		*tp,		/* transaction pointer */
+	xfs_dabuf_t		*bp)		/* freespace buffer */
+{
+	xfs_dir2_free_t		*free;		/* freespace structure */
+
+	free = bp->data;
+	ASSERT(be32_to_cpu(free->hdr.magic) == XFS_DIR2_FREE_MAGIC);
+	xfs_da_log_buf(tp, bp, (uint)((char *)&free->hdr - (char *)free),
+		(uint)(sizeof(xfs_dir2_free_hdr_t) - 1));
+}
+
+/*
+ * Convert a leaf-format directory to a node-format directory.
+ * We need to change the magic number of the leaf block, and copy
+ * the freespace table out of the leaf block into its own block.
+ */
+int						/* error */
+xfs_dir2_leaf_to_node(
+	xfs_da_args_t		*args,		/* operation arguments */
+	xfs_dabuf_t		*lbp)		/* leaf buffer */
+{
+	xfs_inode_t		*dp;		/* incore directory inode */
+	int			error;		/* error return value */
+	xfs_dabuf_t		*fbp;		/* freespace buffer */
+	xfs_dir2_db_t		fdb;		/* freespace block number */
+	xfs_dir2_free_t		*free;		/* freespace structure */
+	__be16			*from;		/* pointer to freespace entry */
+	int			i;		/* leaf freespace index */
+	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
+	xfs_dir2_leaf_tail_t	*ltp;		/* leaf tail structure */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+	int			n;		/* count of live freespc ents */
+	xfs_dir2_data_off_t	off;		/* freespace entry value */
+	__be16			*to;		/* pointer to freespace entry */
+	xfs_trans_t		*tp;		/* transaction pointer */
+
+	trace_xfs_dir2_leaf_to_node(args);
+
+	dp = args->dp;
+	mp = dp->i_mount;
+	tp = args->trans;
+	/*
+	 * Add a freespace block to the directory.
+	 */
+	if ((error = xfs_dir2_grow_inode(args, XFS_DIR2_FREE_SPACE, &fdb))) {
+		return error;
+	}
+	ASSERT(fdb == XFS_DIR2_FREE_FIRSTDB(mp));
+	/*
+	 * Get the buffer for the new freespace block.
+	 */
+	if ((error = xfs_da_get_buf(tp, dp, xfs_dir2_db_to_da(mp, fdb), -1, &fbp,
+			XFS_DATA_FORK))) {
+		return error;
+	}
+	ASSERT(fbp != NULL);
+	free = fbp->data;
+	leaf = lbp->data;
+	ltp = xfs_dir2_leaf_tail_p(mp, leaf);
+	/*
+	 * Initialize the freespace block header.
+	 */
+	free->hdr.magic = cpu_to_be32(XFS_DIR2_FREE_MAGIC);
+	free->hdr.firstdb = 0;
+	ASSERT(be32_to_cpu(ltp->bestcount) <= (uint)dp->i_d.di_size / mp->m_dirblksize);
+	free->hdr.nvalid = ltp->bestcount;
+	/*
+	 * Copy freespace entries from the leaf block to the new block.
+	 * Count active entries.
+	 */
+	for (i = n = 0, from = xfs_dir2_leaf_bests_p(ltp), to = free->bests;
+	     i < be32_to_cpu(ltp->bestcount); i++, from++, to++) {
+		if ((off = be16_to_cpu(*from)) != NULLDATAOFF)
+			n++;
+		*to = cpu_to_be16(off);
+	}
+	free->hdr.nused = cpu_to_be32(n);
+	leaf->hdr.info.magic = cpu_to_be16(XFS_DIR2_LEAFN_MAGIC);
+	/*
+	 * Log everything.
+	 */
+	xfs_dir2_leaf_log_header(tp, lbp);
+	xfs_dir2_free_log_header(tp, fbp);
+	xfs_dir2_free_log_bests(tp, fbp, 0, be32_to_cpu(free->hdr.nvalid) - 1);
+	xfs_da_buf_done(fbp);
+	xfs_dir2_leafn_check(dp, lbp);
+	return 0;
+}
+
+/*
+ * Add a leaf entry to a leaf block in a node-form directory.
+ * The other work necessary is done from the caller.
+ */
+static int					/* error */
+xfs_dir2_leafn_add(
+	xfs_dabuf_t		*bp,		/* leaf buffer */
+	xfs_da_args_t		*args,		/* operation arguments */
+	int			index)		/* insertion pt for new entry */
+{
+	int			compact;	/* compacting stale leaves */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	int			highstale;	/* next stale entry */
+	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
+	xfs_dir2_leaf_entry_t	*lep;		/* leaf entry */
+	int			lfloghigh;	/* high leaf entry logging */
+	int			lfloglow;	/* low leaf entry logging */
+	int			lowstale;	/* previous stale entry */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+	xfs_trans_t		*tp;		/* transaction pointer */
+
+	trace_xfs_dir2_leafn_add(args, index);
+
+	dp = args->dp;
+	mp = dp->i_mount;
+	tp = args->trans;
+	leaf = bp->data;
+
+	/*
+	 * Quick check just to make sure we are not going to index
+	 * into other peoples memory
+	 */
+	if (index < 0)
+		return XFS_ERROR(EFSCORRUPTED);
+
+	/*
+	 * If there are already the maximum number of leaf entries in
+	 * the block, if there are no stale entries it won't fit.
+	 * Caller will do a split.  If there are stale entries we'll do
+	 * a compact.
+	 */
+
+	if (be16_to_cpu(leaf->hdr.count) == xfs_dir2_max_leaf_ents(mp)) {
+		if (!leaf->hdr.stale)
+			return XFS_ERROR(ENOSPC);
+		compact = be16_to_cpu(leaf->hdr.stale) > 1;
+	} else
+		compact = 0;
+	ASSERT(index == 0 || be32_to_cpu(leaf->ents[index - 1].hashval) <= args->hashval);
+	ASSERT(index == be16_to_cpu(leaf->hdr.count) ||
+	       be32_to_cpu(leaf->ents[index].hashval) >= args->hashval);
+
+	if (args->op_flags & XFS_DA_OP_JUSTCHECK)
+		return 0;
+
+	/*
+	 * Compact out all but one stale leaf entry.  Leaves behind
+	 * the entry closest to index.
+	 */
+	if (compact) {
+		xfs_dir2_leaf_compact_x1(bp, &index, &lowstale, &highstale,
+			&lfloglow, &lfloghigh);
+	}
+	/*
+	 * Set impossible logging indices for this case.
+	 */
+	else if (leaf->hdr.stale) {
+		lfloglow = be16_to_cpu(leaf->hdr.count);
+		lfloghigh = -1;
+	}
+	/*
+	 * No stale entries, just insert a space for the new entry.
+	 */
+	if (!leaf->hdr.stale) {
+		lep = &leaf->ents[index];
+		if (index < be16_to_cpu(leaf->hdr.count))
+			memmove(lep + 1, lep,
+				(be16_to_cpu(leaf->hdr.count) - index) * sizeof(*lep));
+		lfloglow = index;
+		lfloghigh = be16_to_cpu(leaf->hdr.count);
+		be16_add_cpu(&leaf->hdr.count, 1);
+	}
+	/*
+	 * There are stale entries.  We'll use one for the new entry.
+	 */
+	else {
+		/*
+		 * If we didn't do a compact then we need to figure out
+		 * which stale entry will be used.
+		 */
+		if (compact == 0) {
+			/*
+			 * Find first stale entry before our insertion point.
+			 */
+			for (lowstale = index - 1;
+			     lowstale >= 0 &&
+				be32_to_cpu(leaf->ents[lowstale].address) !=
+				XFS_DIR2_NULL_DATAPTR;
+			     lowstale--)
+				continue;
+			/*
+			 * Find next stale entry after insertion point.
+			 * Stop looking if the answer would be worse than
+			 * lowstale already found.
+			 */
+			for (highstale = index;
+			     highstale < be16_to_cpu(leaf->hdr.count) &&
+				be32_to_cpu(leaf->ents[highstale].address) !=
+				XFS_DIR2_NULL_DATAPTR &&
+				(lowstale < 0 ||
+				 index - lowstale - 1 >= highstale - index);
+			     highstale++)
+				continue;
+		}
+		/*
+		 * Using the low stale entry.
+		 * Shift entries up toward the stale slot.
+		 */
+		if (lowstale >= 0 &&
+		    (highstale == be16_to_cpu(leaf->hdr.count) ||
+		     index - lowstale - 1 < highstale - index)) {
+			ASSERT(be32_to_cpu(leaf->ents[lowstale].address) ==
+			       XFS_DIR2_NULL_DATAPTR);
+			ASSERT(index - lowstale - 1 >= 0);
+			if (index - lowstale - 1 > 0)
+				memmove(&leaf->ents[lowstale],
+					&leaf->ents[lowstale + 1],
+					(index - lowstale - 1) * sizeof(*lep));
+			lep = &leaf->ents[index - 1];
+			lfloglow = MIN(lowstale, lfloglow);
+			lfloghigh = MAX(index - 1, lfloghigh);
+		}
+		/*
+		 * Using the high stale entry.
+		 * Shift entries down toward the stale slot.
+		 */
+		else {
+			ASSERT(be32_to_cpu(leaf->ents[highstale].address) ==
+			       XFS_DIR2_NULL_DATAPTR);
+			ASSERT(highstale - index >= 0);
+			if (highstale - index > 0)
+				memmove(&leaf->ents[index + 1],
+					&leaf->ents[index],
+					(highstale - index) * sizeof(*lep));
+			lep = &leaf->ents[index];
+			lfloglow = MIN(index, lfloglow);
+			lfloghigh = MAX(highstale, lfloghigh);
+		}
+		be16_add_cpu(&leaf->hdr.stale, -1);
+	}
+	/*
+	 * Insert the new entry, log everything.
+	 */
+	lep->hashval = cpu_to_be32(args->hashval);
+	lep->address = cpu_to_be32(xfs_dir2_db_off_to_dataptr(mp,
+				args->blkno, args->index));
+	xfs_dir2_leaf_log_header(tp, bp);
+	xfs_dir2_leaf_log_ents(tp, bp, lfloglow, lfloghigh);
+	xfs_dir2_leafn_check(dp, bp);
+	return 0;
+}
+
+#ifdef DEBUG
+/*
+ * Check internal consistency of a leafn block.
+ */
+void
+xfs_dir2_leafn_check(
+	xfs_inode_t	*dp,			/* incore directory inode */
+	xfs_dabuf_t	*bp)			/* leaf buffer */
+{
+	int		i;			/* leaf index */
+	xfs_dir2_leaf_t	*leaf;			/* leaf structure */
+	xfs_mount_t	*mp;			/* filesystem mount point */
+	int		stale;			/* count of stale leaves */
+
+	leaf = bp->data;
+	mp = dp->i_mount;
+	ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_DIR2_LEAFN_MAGIC);
+	ASSERT(be16_to_cpu(leaf->hdr.count) <= xfs_dir2_max_leaf_ents(mp));
+	for (i = stale = 0; i < be16_to_cpu(leaf->hdr.count); i++) {
+		if (i + 1 < be16_to_cpu(leaf->hdr.count)) {
+			ASSERT(be32_to_cpu(leaf->ents[i].hashval) <=
+			       be32_to_cpu(leaf->ents[i + 1].hashval));
+		}
+		if (be32_to_cpu(leaf->ents[i].address) == XFS_DIR2_NULL_DATAPTR)
+			stale++;
+	}
+	ASSERT(be16_to_cpu(leaf->hdr.stale) == stale);
+}
+#endif	/* DEBUG */
+
+/*
+ * Return the last hash value in the leaf.
+ * Stale entries are ok.
+ */
+xfs_dahash_t					/* hash value */
+xfs_dir2_leafn_lasthash(
+	xfs_dabuf_t	*bp,			/* leaf buffer */
+	int		*count)			/* count of entries in leaf */
+{
+	xfs_dir2_leaf_t	*leaf;			/* leaf structure */
+
+	leaf = bp->data;
+	ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_DIR2_LEAFN_MAGIC);
+	if (count)
+		*count = be16_to_cpu(leaf->hdr.count);
+	if (!leaf->hdr.count)
+		return 0;
+	return be32_to_cpu(leaf->ents[be16_to_cpu(leaf->hdr.count) - 1].hashval);
+}
+
+/*
+ * Look up a leaf entry for space to add a name in a node-format leaf block.
+ * The extrablk in state is a freespace block.
+ */
+STATIC int
+xfs_dir2_leafn_lookup_for_addname(
+	xfs_dabuf_t		*bp,		/* leaf buffer */
+	xfs_da_args_t		*args,		/* operation arguments */
+	int			*indexp,	/* out: leaf entry index */
+	xfs_da_state_t		*state)		/* state to fill in */
+{
+	xfs_dabuf_t		*curbp = NULL;	/* current data/free buffer */
+	xfs_dir2_db_t		curdb = -1;	/* current data block number */
+	xfs_dir2_db_t		curfdb = -1;	/* current free block number */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	int			error;		/* error return value */
+	int			fi;		/* free entry index */
+	xfs_dir2_free_t		*free = NULL;	/* free block structure */
+	int			index;		/* leaf entry index */
+	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
+	int			length;		/* length of new data entry */
+	xfs_dir2_leaf_entry_t	*lep;		/* leaf entry */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+	xfs_dir2_db_t		newdb;		/* new data block number */
+	xfs_dir2_db_t		newfdb;		/* new free block number */
+	xfs_trans_t		*tp;		/* transaction pointer */
+
+	dp = args->dp;
+	tp = args->trans;
+	mp = dp->i_mount;
+	leaf = bp->data;
+	ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_DIR2_LEAFN_MAGIC);
+#ifdef __KERNEL__
+	ASSERT(be16_to_cpu(leaf->hdr.count) > 0);
+#endif
+	xfs_dir2_leafn_check(dp, bp);
+	/*
+	 * Look up the hash value in the leaf entries.
+	 */
+	index = xfs_dir2_leaf_search_hash(args, bp);
+	/*
+	 * Do we have a buffer coming in?
+	 */
+	if (state->extravalid) {
+		/* If so, it's a free block buffer, get the block number. */
+		curbp = state->extrablk.bp;
+		curfdb = state->extrablk.blkno;
+		free = curbp->data;
+		ASSERT(be32_to_cpu(free->hdr.magic) == XFS_DIR2_FREE_MAGIC);
+	}
+	length = xfs_dir2_data_entsize(args->namelen);
+	/*
+	 * Loop over leaf entries with the right hash value.
+	 */
+	for (lep = &leaf->ents[index]; index < be16_to_cpu(leaf->hdr.count) &&
+				be32_to_cpu(lep->hashval) == args->hashval;
+				lep++, index++) {
+		/*
+		 * Skip stale leaf entries.
+		 */
+		if (be32_to_cpu(lep->address) == XFS_DIR2_NULL_DATAPTR)
+			continue;
+		/*
+		 * Pull the data block number from the entry.
+		 */
+		newdb = xfs_dir2_dataptr_to_db(mp, be32_to_cpu(lep->address));
+		/*
+		 * For addname, we're looking for a place to put the new entry.
+		 * We want to use a data block with an entry of equal
+		 * hash value to ours if there is one with room.
+		 *
+		 * If this block isn't the data block we already have
+		 * in hand, take a look at it.
+		 */
+		if (newdb != curdb) {
+			curdb = newdb;
+			/*
+			 * Convert the data block to the free block
+			 * holding its freespace information.
+			 */
+			newfdb = xfs_dir2_db_to_fdb(mp, newdb);
+			/*
+			 * If it's not the one we have in hand, read it in.
+			 */
+			if (newfdb != curfdb) {
+				/*
+				 * If we had one before, drop it.
+				 */
+				if (curbp)
+					xfs_da_brelse(tp, curbp);
+				/*
+				 * Read the free block.
+				 */
+				error = xfs_da_read_buf(tp, dp,
+						xfs_dir2_db_to_da(mp, newfdb),
+						-1, &curbp, XFS_DATA_FORK);
+				if (error)
+					return error;
+				free = curbp->data;
+				ASSERT(be32_to_cpu(free->hdr.magic) ==
+					XFS_DIR2_FREE_MAGIC);
+				ASSERT((be32_to_cpu(free->hdr.firstdb) %
+					XFS_DIR2_MAX_FREE_BESTS(mp)) == 0);
+				ASSERT(be32_to_cpu(free->hdr.firstdb) <= curdb);
+				ASSERT(curdb < be32_to_cpu(free->hdr.firstdb) +
+					be32_to_cpu(free->hdr.nvalid));
+			}
+			/*
+			 * Get the index for our entry.
+			 */
+			fi = xfs_dir2_db_to_fdindex(mp, curdb);
+			/*
+			 * If it has room, return it.
+			 */
+			if (unlikely(be16_to_cpu(free->bests[fi]) == NULLDATAOFF)) {
+				XFS_ERROR_REPORT("xfs_dir2_leafn_lookup_int",
+							XFS_ERRLEVEL_LOW, mp);
+				if (curfdb != newfdb)
+					xfs_da_brelse(tp, curbp);
+				return XFS_ERROR(EFSCORRUPTED);
+			}
+			curfdb = newfdb;
+			if (be16_to_cpu(free->bests[fi]) >= length)
+				goto out;
+		}
+	}
+	/* Didn't find any space */
+	fi = -1;
+out:
+	ASSERT(args->op_flags & XFS_DA_OP_OKNOENT);
+	if (curbp) {
+		/* Giving back a free block. */
+		state->extravalid = 1;
+		state->extrablk.bp = curbp;
+		state->extrablk.index = fi;
+		state->extrablk.blkno = curfdb;
+		state->extrablk.magic = XFS_DIR2_FREE_MAGIC;
+	} else {
+		state->extravalid = 0;
+	}
+	/*
+	 * Return the index, that will be the insertion point.
+	 */
+	*indexp = index;
+	return XFS_ERROR(ENOENT);
+}
+
+/*
+ * Look up a leaf entry in a node-format leaf block.
+ * The extrablk in state a data block.
+ */
+STATIC int
+xfs_dir2_leafn_lookup_for_entry(
+	xfs_dabuf_t		*bp,		/* leaf buffer */
+	xfs_da_args_t		*args,		/* operation arguments */
+	int			*indexp,	/* out: leaf entry index */
+	xfs_da_state_t		*state)		/* state to fill in */
+{
+	xfs_dabuf_t		*curbp = NULL;	/* current data/free buffer */
+	xfs_dir2_db_t		curdb = -1;	/* current data block number */
+	xfs_dir2_data_entry_t	*dep;		/* data block entry */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	int			error;		/* error return value */
+	int			index;		/* leaf entry index */
+	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
+	xfs_dir2_leaf_entry_t	*lep;		/* leaf entry */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+	xfs_dir2_db_t		newdb;		/* new data block number */
+	xfs_trans_t		*tp;		/* transaction pointer */
+	enum xfs_dacmp		cmp;		/* comparison result */
+
+	dp = args->dp;
+	tp = args->trans;
+	mp = dp->i_mount;
+	leaf = bp->data;
+	ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_DIR2_LEAFN_MAGIC);
+#ifdef __KERNEL__
+	ASSERT(be16_to_cpu(leaf->hdr.count) > 0);
+#endif
+	xfs_dir2_leafn_check(dp, bp);
+	/*
+	 * Look up the hash value in the leaf entries.
+	 */
+	index = xfs_dir2_leaf_search_hash(args, bp);
+	/*
+	 * Do we have a buffer coming in?
+	 */
+	if (state->extravalid) {
+		curbp = state->extrablk.bp;
+		curdb = state->extrablk.blkno;
+	}
+	/*
+	 * Loop over leaf entries with the right hash value.
+	 */
+	for (lep = &leaf->ents[index]; index < be16_to_cpu(leaf->hdr.count) &&
+				be32_to_cpu(lep->hashval) == args->hashval;
+				lep++, index++) {
+		/*
+		 * Skip stale leaf entries.
+		 */
+		if (be32_to_cpu(lep->address) == XFS_DIR2_NULL_DATAPTR)
+			continue;
+		/*
+		 * Pull the data block number from the entry.
+		 */
+		newdb = xfs_dir2_dataptr_to_db(mp, be32_to_cpu(lep->address));
+		/*
+		 * Not adding a new entry, so we really want to find
+		 * the name given to us.
+		 *
+		 * If it's a different data block, go get it.
+		 */
+		if (newdb != curdb) {
+			/*
+			 * If we had a block before that we aren't saving
+			 * for a CI name, drop it
+			 */
+			if (curbp && (args->cmpresult == XFS_CMP_DIFFERENT ||
+						curdb != state->extrablk.blkno))
+				xfs_da_brelse(tp, curbp);
+			/*
+			 * If needing the block that is saved with a CI match,
+			 * use it otherwise read in the new data block.
+			 */
+			if (args->cmpresult != XFS_CMP_DIFFERENT &&
+					newdb == state->extrablk.blkno) {
+				ASSERT(state->extravalid);
+				curbp = state->extrablk.bp;
+			} else {
+				error = xfs_da_read_buf(tp, dp,
+						xfs_dir2_db_to_da(mp, newdb),
+						-1, &curbp, XFS_DATA_FORK);
+				if (error)
+					return error;
+			}
+			xfs_dir2_data_check(dp, curbp);
+			curdb = newdb;
+		}
+		/*
+		 * Point to the data entry.
+		 */
+		dep = (xfs_dir2_data_entry_t *)((char *)curbp->data +
+			xfs_dir2_dataptr_to_off(mp, be32_to_cpu(lep->address)));
+		/*
+		 * Compare the entry and if it's an exact match, return
+		 * EEXIST immediately. If it's the first case-insensitive
+		 * match, store the block & inode number and continue looking.
+		 */
+		cmp = mp->m_dirnameops->compname(args, dep->name, dep->namelen);
+		if (cmp != XFS_CMP_DIFFERENT && cmp != args->cmpresult) {
+			/* If there is a CI match block, drop it */
+			if (args->cmpresult != XFS_CMP_DIFFERENT &&
+						curdb != state->extrablk.blkno)
+				xfs_da_brelse(tp, state->extrablk.bp);
+			args->cmpresult = cmp;
+			args->inumber = be64_to_cpu(dep->inumber);
+			*indexp = index;
+			state->extravalid = 1;
+			state->extrablk.bp = curbp;
+			state->extrablk.blkno = curdb;
+			state->extrablk.index = (int)((char *)dep -
+							(char *)curbp->data);
+			state->extrablk.magic = XFS_DIR2_DATA_MAGIC;
+			if (cmp == XFS_CMP_EXACT)
+				return XFS_ERROR(EEXIST);
+		}
+	}
+	ASSERT(index == be16_to_cpu(leaf->hdr.count) ||
+					(args->op_flags & XFS_DA_OP_OKNOENT));
+	if (curbp) {
+		if (args->cmpresult == XFS_CMP_DIFFERENT) {
+			/* Giving back last used data block. */
+			state->extravalid = 1;
+			state->extrablk.bp = curbp;
+			state->extrablk.index = -1;
+			state->extrablk.blkno = curdb;
+			state->extrablk.magic = XFS_DIR2_DATA_MAGIC;
+		} else {
+			/* If the curbp is not the CI match block, drop it */
+			if (state->extrablk.bp != curbp)
+				xfs_da_brelse(tp, curbp);
+		}
+	} else {
+		state->extravalid = 0;
+	}
+	*indexp = index;
+	return XFS_ERROR(ENOENT);
+}
+
+/*
+ * Look up a leaf entry in a node-format leaf block.
+ * If this is an addname then the extrablk in state is a freespace block,
+ * otherwise it's a data block.
+ */
+int
+xfs_dir2_leafn_lookup_int(
+	xfs_dabuf_t		*bp,		/* leaf buffer */
+	xfs_da_args_t		*args,		/* operation arguments */
+	int			*indexp,	/* out: leaf entry index */
+	xfs_da_state_t		*state)		/* state to fill in */
+{
+	if (args->op_flags & XFS_DA_OP_ADDNAME)
+		return xfs_dir2_leafn_lookup_for_addname(bp, args, indexp,
+							state);
+	return xfs_dir2_leafn_lookup_for_entry(bp, args, indexp, state);
+}
+
+/*
+ * Move count leaf entries from source to destination leaf.
+ * Log entries and headers.  Stale entries are preserved.
+ */
+static void
+xfs_dir2_leafn_moveents(
+	xfs_da_args_t	*args,			/* operation arguments */
+	xfs_dabuf_t	*bp_s,			/* source leaf buffer */
+	int		start_s,		/* source leaf index */
+	xfs_dabuf_t	*bp_d,			/* destination leaf buffer */
+	int		start_d,		/* destination leaf index */
+	int		count)			/* count of leaves to copy */
+{
+	xfs_dir2_leaf_t	*leaf_d;		/* destination leaf structure */
+	xfs_dir2_leaf_t	*leaf_s;		/* source leaf structure */
+	int		stale;			/* count stale leaves copied */
+	xfs_trans_t	*tp;			/* transaction pointer */
+
+	trace_xfs_dir2_leafn_moveents(args, start_s, start_d, count);
+
+	/*
+	 * Silently return if nothing to do.
+	 */
+	if (count == 0) {
+		return;
+	}
+	tp = args->trans;
+	leaf_s = bp_s->data;
+	leaf_d = bp_d->data;
+	/*
+	 * If the destination index is not the end of the current
+	 * destination leaf entries, open up a hole in the destination
+	 * to hold the new entries.
+	 */
+	if (start_d < be16_to_cpu(leaf_d->hdr.count)) {
+		memmove(&leaf_d->ents[start_d + count], &leaf_d->ents[start_d],
+			(be16_to_cpu(leaf_d->hdr.count) - start_d) *
+			sizeof(xfs_dir2_leaf_entry_t));
+		xfs_dir2_leaf_log_ents(tp, bp_d, start_d + count,
+			count + be16_to_cpu(leaf_d->hdr.count) - 1);
+	}
+	/*
+	 * If the source has stale leaves, count the ones in the copy range
+	 * so we can update the header correctly.
+	 */
+	if (leaf_s->hdr.stale) {
+		int	i;			/* temp leaf index */
+
+		for (i = start_s, stale = 0; i < start_s + count; i++) {
+			if (be32_to_cpu(leaf_s->ents[i].address) == XFS_DIR2_NULL_DATAPTR)
+				stale++;
+		}
+	} else
+		stale = 0;
+	/*
+	 * Copy the leaf entries from source to destination.
+	 */
+	memcpy(&leaf_d->ents[start_d], &leaf_s->ents[start_s],
+		count * sizeof(xfs_dir2_leaf_entry_t));
+	xfs_dir2_leaf_log_ents(tp, bp_d, start_d, start_d + count - 1);
+	/*
+	 * If there are source entries after the ones we copied,
+	 * delete the ones we copied by sliding the next ones down.
+	 */
+	if (start_s + count < be16_to_cpu(leaf_s->hdr.count)) {
+		memmove(&leaf_s->ents[start_s], &leaf_s->ents[start_s + count],
+			count * sizeof(xfs_dir2_leaf_entry_t));
+		xfs_dir2_leaf_log_ents(tp, bp_s, start_s, start_s + count - 1);
+	}
+	/*
+	 * Update the headers and log them.
+	 */
+	be16_add_cpu(&leaf_s->hdr.count, -(count));
+	be16_add_cpu(&leaf_s->hdr.stale, -(stale));
+	be16_add_cpu(&leaf_d->hdr.count, count);
+	be16_add_cpu(&leaf_d->hdr.stale, stale);
+	xfs_dir2_leaf_log_header(tp, bp_s);
+	xfs_dir2_leaf_log_header(tp, bp_d);
+	xfs_dir2_leafn_check(args->dp, bp_s);
+	xfs_dir2_leafn_check(args->dp, bp_d);
+}
+
+/*
+ * Determine the sort order of two leaf blocks.
+ * Returns 1 if both are valid and leaf2 should be before leaf1, else 0.
+ */
+int						/* sort order */
+xfs_dir2_leafn_order(
+	xfs_dabuf_t	*leaf1_bp,		/* leaf1 buffer */
+	xfs_dabuf_t	*leaf2_bp)		/* leaf2 buffer */
+{
+	xfs_dir2_leaf_t	*leaf1;			/* leaf1 structure */
+	xfs_dir2_leaf_t	*leaf2;			/* leaf2 structure */
+
+	leaf1 = leaf1_bp->data;
+	leaf2 = leaf2_bp->data;
+	ASSERT(be16_to_cpu(leaf1->hdr.info.magic) == XFS_DIR2_LEAFN_MAGIC);
+	ASSERT(be16_to_cpu(leaf2->hdr.info.magic) == XFS_DIR2_LEAFN_MAGIC);
+	if (be16_to_cpu(leaf1->hdr.count) > 0 &&
+	    be16_to_cpu(leaf2->hdr.count) > 0 &&
+	    (be32_to_cpu(leaf2->ents[0].hashval) < be32_to_cpu(leaf1->ents[0].hashval) ||
+	     be32_to_cpu(leaf2->ents[be16_to_cpu(leaf2->hdr.count) - 1].hashval) <
+	     be32_to_cpu(leaf1->ents[be16_to_cpu(leaf1->hdr.count) - 1].hashval)))
+		return 1;
+	return 0;
+}
+
+/*
+ * Rebalance leaf entries between two leaf blocks.
+ * This is actually only called when the second block is new,
+ * though the code deals with the general case.
+ * A new entry will be inserted in one of the blocks, and that
+ * entry is taken into account when balancing.
+ */
+static void
+xfs_dir2_leafn_rebalance(
+	xfs_da_state_t		*state,		/* btree cursor */
+	xfs_da_state_blk_t	*blk1,		/* first btree block */
+	xfs_da_state_blk_t	*blk2)		/* second btree block */
+{
+	xfs_da_args_t		*args;		/* operation arguments */
+	int			count;		/* count (& direction) leaves */
+	int			isleft;		/* new goes in left leaf */
+	xfs_dir2_leaf_t		*leaf1;		/* first leaf structure */
+	xfs_dir2_leaf_t		*leaf2;		/* second leaf structure */
+	int			mid;		/* midpoint leaf index */
+#ifdef DEBUG
+	int			oldstale;	/* old count of stale leaves */
+#endif
+	int			oldsum;		/* old total leaf count */
+	int			swap;		/* swapped leaf blocks */
+
+	args = state->args;
+	/*
+	 * If the block order is wrong, swap the arguments.
+	 */
+	if ((swap = xfs_dir2_leafn_order(blk1->bp, blk2->bp))) {
+		xfs_da_state_blk_t	*tmp;	/* temp for block swap */
+
+		tmp = blk1;
+		blk1 = blk2;
+		blk2 = tmp;
+	}
+	leaf1 = blk1->bp->data;
+	leaf2 = blk2->bp->data;
+	oldsum = be16_to_cpu(leaf1->hdr.count) + be16_to_cpu(leaf2->hdr.count);
+#ifdef DEBUG
+	oldstale = be16_to_cpu(leaf1->hdr.stale) + be16_to_cpu(leaf2->hdr.stale);
+#endif
+	mid = oldsum >> 1;
+	/*
+	 * If the old leaf count was odd then the new one will be even,
+	 * so we need to divide the new count evenly.
+	 */
+	if (oldsum & 1) {
+		xfs_dahash_t	midhash;	/* middle entry hash value */
+
+		if (mid >= be16_to_cpu(leaf1->hdr.count))
+			midhash = be32_to_cpu(leaf2->ents[mid - be16_to_cpu(leaf1->hdr.count)].hashval);
+		else
+			midhash = be32_to_cpu(leaf1->ents[mid].hashval);
+		isleft = args->hashval <= midhash;
+	}
+	/*
+	 * If the old count is even then the new count is odd, so there's
+	 * no preferred side for the new entry.
+	 * Pick the left one.
+	 */
+	else
+		isleft = 1;
+	/*
+	 * Calculate moved entry count.  Positive means left-to-right,
+	 * negative means right-to-left.  Then move the entries.
+	 */
+	count = be16_to_cpu(leaf1->hdr.count) - mid + (isleft == 0);
+	if (count > 0)
+		xfs_dir2_leafn_moveents(args, blk1->bp,
+			be16_to_cpu(leaf1->hdr.count) - count, blk2->bp, 0, count);
+	else if (count < 0)
+		xfs_dir2_leafn_moveents(args, blk2->bp, 0, blk1->bp,
+			be16_to_cpu(leaf1->hdr.count), count);
+	ASSERT(be16_to_cpu(leaf1->hdr.count) + be16_to_cpu(leaf2->hdr.count) == oldsum);
+	ASSERT(be16_to_cpu(leaf1->hdr.stale) + be16_to_cpu(leaf2->hdr.stale) == oldstale);
+	/*
+	 * Mark whether we're inserting into the old or new leaf.
+	 */
+	if (be16_to_cpu(leaf1->hdr.count) < be16_to_cpu(leaf2->hdr.count))
+		state->inleaf = swap;
+	else if (be16_to_cpu(leaf1->hdr.count) > be16_to_cpu(leaf2->hdr.count))
+		state->inleaf = !swap;
+	else
+		state->inleaf =
+			swap ^ (blk1->index <= be16_to_cpu(leaf1->hdr.count));
+	/*
+	 * Adjust the expected index for insertion.
+	 */
+	if (!state->inleaf)
+		blk2->index = blk1->index - be16_to_cpu(leaf1->hdr.count);
+
+	/*
+	 * Finally sanity check just to make sure we are not returning a
+	 * negative index
+	 */
+	if(blk2->index < 0) {
+		state->inleaf = 1;
+		blk2->index = 0;
+		xfs_alert(args->dp->i_mount,
+	"%s: picked the wrong leaf? reverting original leaf: blk1->index %d\n",
+			__func__, blk1->index);
+	}
+}
+
+/*
+ * Remove an entry from a node directory.
+ * This removes the leaf entry and the data entry,
+ * and updates the free block if necessary.
+ */
+static int					/* error */
+xfs_dir2_leafn_remove(
+	xfs_da_args_t		*args,		/* operation arguments */
+	xfs_dabuf_t		*bp,		/* leaf buffer */
+	int			index,		/* leaf entry index */
+	xfs_da_state_blk_t	*dblk,		/* data block */
+	int			*rval)		/* resulting block needs join */
+{
+	xfs_dir2_data_t		*data;		/* data block structure */
+	xfs_dir2_db_t		db;		/* data block number */
+	xfs_dabuf_t		*dbp;		/* data block buffer */
+	xfs_dir2_data_entry_t	*dep;		/* data block entry */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
+	xfs_dir2_leaf_entry_t	*lep;		/* leaf entry */
+	int			longest;	/* longest data free entry */
+	int			off;		/* data block entry offset */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+	int			needlog;	/* need to log data header */
+	int			needscan;	/* need to rescan data frees */
+	xfs_trans_t		*tp;		/* transaction pointer */
+
+	trace_xfs_dir2_leafn_remove(args, index);
+
+	dp = args->dp;
+	tp = args->trans;
+	mp = dp->i_mount;
+	leaf = bp->data;
+	ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_DIR2_LEAFN_MAGIC);
+	/*
+	 * Point to the entry we're removing.
+	 */
+	lep = &leaf->ents[index];
+	/*
+	 * Extract the data block and offset from the entry.
+	 */
+	db = xfs_dir2_dataptr_to_db(mp, be32_to_cpu(lep->address));
+	ASSERT(dblk->blkno == db);
+	off = xfs_dir2_dataptr_to_off(mp, be32_to_cpu(lep->address));
+	ASSERT(dblk->index == off);
+	/*
+	 * Kill the leaf entry by marking it stale.
+	 * Log the leaf block changes.
+	 */
+	be16_add_cpu(&leaf->hdr.stale, 1);
+	xfs_dir2_leaf_log_header(tp, bp);
+	lep->address = cpu_to_be32(XFS_DIR2_NULL_DATAPTR);
+	xfs_dir2_leaf_log_ents(tp, bp, index, index);
+	/*
+	 * Make the data entry free.  Keep track of the longest freespace
+	 * in the data block in case it changes.
+	 */
+	dbp = dblk->bp;
+	data = dbp->data;
+	dep = (xfs_dir2_data_entry_t *)((char *)data + off);
+	longest = be16_to_cpu(data->hdr.bestfree[0].length);
+	needlog = needscan = 0;
+	xfs_dir2_data_make_free(tp, dbp, off,
+		xfs_dir2_data_entsize(dep->namelen), &needlog, &needscan);
+	/*
+	 * Rescan the data block freespaces for bestfree.
+	 * Log the data block header if needed.
+	 */
+	if (needscan)
+		xfs_dir2_data_freescan(mp, data, &needlog);
+	if (needlog)
+		xfs_dir2_data_log_header(tp, dbp);
+	xfs_dir2_data_check(dp, dbp);
+	/*
+	 * If the longest data block freespace changes, need to update
+	 * the corresponding freeblock entry.
+	 */
+	if (longest < be16_to_cpu(data->hdr.bestfree[0].length)) {
+		int		error;		/* error return value */
+		xfs_dabuf_t	*fbp;		/* freeblock buffer */
+		xfs_dir2_db_t	fdb;		/* freeblock block number */
+		int		findex;		/* index in freeblock entries */
+		xfs_dir2_free_t	*free;		/* freeblock structure */
+		int		logfree;	/* need to log free entry */
+
+		/*
+		 * Convert the data block number to a free block,
+		 * read in the free block.
+		 */
+		fdb = xfs_dir2_db_to_fdb(mp, db);
+		if ((error = xfs_da_read_buf(tp, dp, xfs_dir2_db_to_da(mp, fdb),
+				-1, &fbp, XFS_DATA_FORK))) {
+			return error;
+		}
+		free = fbp->data;
+		ASSERT(be32_to_cpu(free->hdr.magic) == XFS_DIR2_FREE_MAGIC);
+		ASSERT(be32_to_cpu(free->hdr.firstdb) ==
+		       XFS_DIR2_MAX_FREE_BESTS(mp) *
+		       (fdb - XFS_DIR2_FREE_FIRSTDB(mp)));
+		/*
+		 * Calculate which entry we need to fix.
+		 */
+		findex = xfs_dir2_db_to_fdindex(mp, db);
+		longest = be16_to_cpu(data->hdr.bestfree[0].length);
+		/*
+		 * If the data block is now empty we can get rid of it
+		 * (usually).
+		 */
+		if (longest == mp->m_dirblksize - (uint)sizeof(data->hdr)) {
+			/*
+			 * Try to punch out the data block.
+			 */
+			error = xfs_dir2_shrink_inode(args, db, dbp);
+			if (error == 0) {
+				dblk->bp = NULL;
+				data = NULL;
+			}
+			/*
+			 * We can get ENOSPC if there's no space reservation.
+			 * In this case just drop the buffer and some one else
+			 * will eventually get rid of the empty block.
+			 */
+			else if (error == ENOSPC && args->total == 0)
+				xfs_da_buf_done(dbp);
+			else
+				return error;
+		}
+		/*
+		 * If we got rid of the data block, we can eliminate that entry
+		 * in the free block.
+		 */
+		if (data == NULL) {
+			/*
+			 * One less used entry in the free table.
+			 */
+			be32_add_cpu(&free->hdr.nused, -1);
+			xfs_dir2_free_log_header(tp, fbp);
+			/*
+			 * If this was the last entry in the table, we can
+			 * trim the table size back.  There might be other
+			 * entries at the end referring to non-existent
+			 * data blocks, get those too.
+			 */
+			if (findex == be32_to_cpu(free->hdr.nvalid) - 1) {
+				int	i;		/* free entry index */
+
+				for (i = findex - 1;
+				     i >= 0 && be16_to_cpu(free->bests[i]) == NULLDATAOFF;
+				     i--)
+					continue;
+				free->hdr.nvalid = cpu_to_be32(i + 1);
+				logfree = 0;
+			}
+			/*
+			 * Not the last entry, just punch it out.
+			 */
+			else {
+				free->bests[findex] = cpu_to_be16(NULLDATAOFF);
+				logfree = 1;
+			}
+			/*
+			 * If there are no useful entries left in the block,
+			 * get rid of the block if we can.
+			 */
+			if (!free->hdr.nused) {
+				error = xfs_dir2_shrink_inode(args, fdb, fbp);
+				if (error == 0) {
+					fbp = NULL;
+					logfree = 0;
+				} else if (error != ENOSPC || args->total != 0)
+					return error;
+				/*
+				 * It's possible to get ENOSPC if there is no
+				 * space reservation.  In this case some one
+				 * else will eventually get rid of this block.
+				 */
+			}
+		}
+		/*
+		 * Data block is not empty, just set the free entry to
+		 * the new value.
+		 */
+		else {
+			free->bests[findex] = cpu_to_be16(longest);
+			logfree = 1;
+		}
+		/*
+		 * Log the free entry that changed, unless we got rid of it.
+		 */
+		if (logfree)
+			xfs_dir2_free_log_bests(tp, fbp, findex, findex);
+		/*
+		 * Drop the buffer if we still have it.
+		 */
+		if (fbp)
+			xfs_da_buf_done(fbp);
+	}
+	xfs_dir2_leafn_check(dp, bp);
+	/*
+	 * Return indication of whether this leaf block is empty enough
+	 * to justify trying to join it with a neighbor.
+	 */
+	*rval =
+		((uint)sizeof(leaf->hdr) +
+		 (uint)sizeof(leaf->ents[0]) *
+		 (be16_to_cpu(leaf->hdr.count) - be16_to_cpu(leaf->hdr.stale))) <
+		mp->m_dir_magicpct;
+	return 0;
+}
+
+/*
+ * Split the leaf entries in the old block into old and new blocks.
+ */
+int						/* error */
+xfs_dir2_leafn_split(
+	xfs_da_state_t		*state,		/* btree cursor */
+	xfs_da_state_blk_t	*oldblk,	/* original block */
+	xfs_da_state_blk_t	*newblk)	/* newly created block */
+{
+	xfs_da_args_t		*args;		/* operation arguments */
+	xfs_dablk_t		blkno;		/* new leaf block number */
+	int			error;		/* error return value */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+
+	/*
+	 * Allocate space for a new leaf node.
+	 */
+	args = state->args;
+	mp = args->dp->i_mount;
+	ASSERT(args != NULL);
+	ASSERT(oldblk->magic == XFS_DIR2_LEAFN_MAGIC);
+	error = xfs_da_grow_inode(args, &blkno);
+	if (error) {
+		return error;
+	}
+	/*
+	 * Initialize the new leaf block.
+	 */
+	error = xfs_dir2_leaf_init(args, xfs_dir2_da_to_db(mp, blkno),
+		&newblk->bp, XFS_DIR2_LEAFN_MAGIC);
+	if (error) {
+		return error;
+	}
+	newblk->blkno = blkno;
+	newblk->magic = XFS_DIR2_LEAFN_MAGIC;
+	/*
+	 * Rebalance the entries across the two leaves, link the new
+	 * block into the leaves.
+	 */
+	xfs_dir2_leafn_rebalance(state, oldblk, newblk);
+	error = xfs_da_blk_link(state, oldblk, newblk);
+	if (error) {
+		return error;
+	}
+	/*
+	 * Insert the new entry in the correct block.
+	 */
+	if (state->inleaf)
+		error = xfs_dir2_leafn_add(oldblk->bp, args, oldblk->index);
+	else
+		error = xfs_dir2_leafn_add(newblk->bp, args, newblk->index);
+	/*
+	 * Update last hashval in each block since we added the name.
+	 */
+	oldblk->hashval = xfs_dir2_leafn_lasthash(oldblk->bp, NULL);
+	newblk->hashval = xfs_dir2_leafn_lasthash(newblk->bp, NULL);
+	xfs_dir2_leafn_check(args->dp, oldblk->bp);
+	xfs_dir2_leafn_check(args->dp, newblk->bp);
+	return error;
+}
+
+/*
+ * Check a leaf block and its neighbors to see if the block should be
+ * collapsed into one or the other neighbor.  Always keep the block
+ * with the smaller block number.
+ * If the current block is over 50% full, don't try to join it, return 0.
+ * If the block is empty, fill in the state structure and return 2.
+ * If it can be collapsed, fill in the state structure and return 1.
+ * If nothing can be done, return 0.
+ */
+int						/* error */
+xfs_dir2_leafn_toosmall(
+	xfs_da_state_t		*state,		/* btree cursor */
+	int			*action)	/* resulting action to take */
+{
+	xfs_da_state_blk_t	*blk;		/* leaf block */
+	xfs_dablk_t		blkno;		/* leaf block number */
+	xfs_dabuf_t		*bp;		/* leaf buffer */
+	int			bytes;		/* bytes in use */
+	int			count;		/* leaf live entry count */
+	int			error;		/* error return value */
+	int			forward;	/* sibling block direction */
+	int			i;		/* sibling counter */
+	xfs_da_blkinfo_t	*info;		/* leaf block header */
+	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
+	int			rval;		/* result from path_shift */
+
+	/*
+	 * Check for the degenerate case of the block being over 50% full.
+	 * If so, it's not worth even looking to see if we might be able
+	 * to coalesce with a sibling.
+	 */
+	blk = &state->path.blk[state->path.active - 1];
+	info = blk->bp->data;
+	ASSERT(be16_to_cpu(info->magic) == XFS_DIR2_LEAFN_MAGIC);
+	leaf = (xfs_dir2_leaf_t *)info;
+	count = be16_to_cpu(leaf->hdr.count) - be16_to_cpu(leaf->hdr.stale);
+	bytes = (uint)sizeof(leaf->hdr) + count * (uint)sizeof(leaf->ents[0]);
+	if (bytes > (state->blocksize >> 1)) {
+		/*
+		 * Blk over 50%, don't try to join.
+		 */
+		*action = 0;
+		return 0;
+	}
+	/*
+	 * Check for the degenerate case of the block being empty.
+	 * If the block is empty, we'll simply delete it, no need to
+	 * coalesce it with a sibling block.  We choose (arbitrarily)
+	 * to merge with the forward block unless it is NULL.
+	 */
+	if (count == 0) {
+		/*
+		 * Make altpath point to the block we want to keep and
+		 * path point to the block we want to drop (this one).
+		 */
+		forward = (info->forw != 0);
+		memcpy(&state->altpath, &state->path, sizeof(state->path));
+		error = xfs_da_path_shift(state, &state->altpath, forward, 0,
+			&rval);
+		if (error)
+			return error;
+		*action = rval ? 2 : 0;
+		return 0;
+	}
+	/*
+	 * Examine each sibling block to see if we can coalesce with
+	 * at least 25% free space to spare.  We need to figure out
+	 * whether to merge with the forward or the backward block.
+	 * We prefer coalescing with the lower numbered sibling so as
+	 * to shrink a directory over time.
+	 */
+	forward = be32_to_cpu(info->forw) < be32_to_cpu(info->back);
+	for (i = 0, bp = NULL; i < 2; forward = !forward, i++) {
+		blkno = forward ? be32_to_cpu(info->forw) : be32_to_cpu(info->back);
+		if (blkno == 0)
+			continue;
+		/*
+		 * Read the sibling leaf block.
+		 */
+		if ((error =
+		    xfs_da_read_buf(state->args->trans, state->args->dp, blkno,
+			    -1, &bp, XFS_DATA_FORK))) {
+			return error;
+		}
+		ASSERT(bp != NULL);
+		/*
+		 * Count bytes in the two blocks combined.
+		 */
+		leaf = (xfs_dir2_leaf_t *)info;
+		count = be16_to_cpu(leaf->hdr.count) - be16_to_cpu(leaf->hdr.stale);
+		bytes = state->blocksize - (state->blocksize >> 2);
+		leaf = bp->data;
+		ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_DIR2_LEAFN_MAGIC);
+		count += be16_to_cpu(leaf->hdr.count) - be16_to_cpu(leaf->hdr.stale);
+		bytes -= count * (uint)sizeof(leaf->ents[0]);
+		/*
+		 * Fits with at least 25% to spare.
+		 */
+		if (bytes >= 0)
+			break;
+		xfs_da_brelse(state->args->trans, bp);
+	}
+	/*
+	 * Didn't like either block, give up.
+	 */
+	if (i >= 2) {
+		*action = 0;
+		return 0;
+	}
+	/*
+	 * Done with the sibling leaf block here, drop the dabuf
+	 * so path_shift can get it.
+	 */
+	xfs_da_buf_done(bp);
+	/*
+	 * Make altpath point to the block we want to keep (the lower
+	 * numbered block) and path point to the block we want to drop.
+	 */
+	memcpy(&state->altpath, &state->path, sizeof(state->path));
+	if (blkno < blk->blkno)
+		error = xfs_da_path_shift(state, &state->altpath, forward, 0,
+			&rval);
+	else
+		error = xfs_da_path_shift(state, &state->path, forward, 0,
+			&rval);
+	if (error) {
+		return error;
+	}
+	*action = rval ? 0 : 1;
+	return 0;
+}
+
+/*
+ * Move all the leaf entries from drop_blk to save_blk.
+ * This is done as part of a join operation.
+ */
+void
+xfs_dir2_leafn_unbalance(
+	xfs_da_state_t		*state,		/* cursor */
+	xfs_da_state_blk_t	*drop_blk,	/* dead block */
+	xfs_da_state_blk_t	*save_blk)	/* surviving block */
+{
+	xfs_da_args_t		*args;		/* operation arguments */
+	xfs_dir2_leaf_t		*drop_leaf;	/* dead leaf structure */
+	xfs_dir2_leaf_t		*save_leaf;	/* surviving leaf structure */
+
+	args = state->args;
+	ASSERT(drop_blk->magic == XFS_DIR2_LEAFN_MAGIC);
+	ASSERT(save_blk->magic == XFS_DIR2_LEAFN_MAGIC);
+	drop_leaf = drop_blk->bp->data;
+	save_leaf = save_blk->bp->data;
+	ASSERT(be16_to_cpu(drop_leaf->hdr.info.magic) == XFS_DIR2_LEAFN_MAGIC);
+	ASSERT(be16_to_cpu(save_leaf->hdr.info.magic) == XFS_DIR2_LEAFN_MAGIC);
+	/*
+	 * If there are any stale leaf entries, take this opportunity
+	 * to purge them.
+	 */
+	if (drop_leaf->hdr.stale)
+		xfs_dir2_leaf_compact(args, drop_blk->bp);
+	if (save_leaf->hdr.stale)
+		xfs_dir2_leaf_compact(args, save_blk->bp);
+	/*
+	 * Move the entries from drop to the appropriate end of save.
+	 */
+	drop_blk->hashval = be32_to_cpu(drop_leaf->ents[be16_to_cpu(drop_leaf->hdr.count) - 1].hashval);
+	if (xfs_dir2_leafn_order(save_blk->bp, drop_blk->bp))
+		xfs_dir2_leafn_moveents(args, drop_blk->bp, 0, save_blk->bp, 0,
+			be16_to_cpu(drop_leaf->hdr.count));
+	else
+		xfs_dir2_leafn_moveents(args, drop_blk->bp, 0, save_blk->bp,
+			be16_to_cpu(save_leaf->hdr.count), be16_to_cpu(drop_leaf->hdr.count));
+	save_blk->hashval = be32_to_cpu(save_leaf->ents[be16_to_cpu(save_leaf->hdr.count) - 1].hashval);
+	xfs_dir2_leafn_check(args->dp, save_blk->bp);
+}
+
+/*
+ * Top-level node form directory addname routine.
+ */
+int						/* error */
+xfs_dir2_node_addname(
+	xfs_da_args_t		*args)		/* operation arguments */
+{
+	xfs_da_state_blk_t	*blk;		/* leaf block for insert */
+	int			error;		/* error return value */
+	int			rval;		/* sub-return value */
+	xfs_da_state_t		*state;		/* btree cursor */
+
+	trace_xfs_dir2_node_addname(args);
+
+	/*
+	 * Allocate and initialize the state (btree cursor).
+	 */
+	state = xfs_da_state_alloc();
+	state->args = args;
+	state->mp = args->dp->i_mount;
+	state->blocksize = state->mp->m_dirblksize;
+	state->node_ents = state->mp->m_dir_node_ents;
+	/*
+	 * Look up the name.  We're not supposed to find it, but
+	 * this gives us the insertion point.
+	 */
+	error = xfs_da_node_lookup_int(state, &rval);
+	if (error)
+		rval = error;
+	if (rval != ENOENT) {
+		goto done;
+	}
+	/*
+	 * Add the data entry to a data block.
+	 * Extravalid is set to a freeblock found by lookup.
+	 */
+	rval = xfs_dir2_node_addname_int(args,
+		state->extravalid ? &state->extrablk : NULL);
+	if (rval) {
+		goto done;
+	}
+	blk = &state->path.blk[state->path.active - 1];
+	ASSERT(blk->magic == XFS_DIR2_LEAFN_MAGIC);
+	/*
+	 * Add the new leaf entry.
+	 */
+	rval = xfs_dir2_leafn_add(blk->bp, args, blk->index);
+	if (rval == 0) {
+		/*
+		 * It worked, fix the hash values up the btree.
+		 */
+		if (!(args->op_flags & XFS_DA_OP_JUSTCHECK))
+			xfs_da_fixhashpath(state, &state->path);
+	} else {
+		/*
+		 * It didn't work, we need to split the leaf block.
+		 */
+		if (args->total == 0) {
+			ASSERT(rval == ENOSPC);
+			goto done;
+		}
+		/*
+		 * Split the leaf block and insert the new entry.
+		 */
+		rval = xfs_da_split(state);
+	}
+done:
+	xfs_da_state_free(state);
+	return rval;
+}
+
+/*
+ * Add the data entry for a node-format directory name addition.
+ * The leaf entry is added in xfs_dir2_leafn_add.
+ * We may enter with a freespace block that the lookup found.
+ */
+static int					/* error */
+xfs_dir2_node_addname_int(
+	xfs_da_args_t		*args,		/* operation arguments */
+	xfs_da_state_blk_t	*fblk)		/* optional freespace block */
+{
+	xfs_dir2_data_t		*data;		/* data block structure */
+	xfs_dir2_db_t		dbno;		/* data block number */
+	xfs_dabuf_t		*dbp;		/* data block buffer */
+	xfs_dir2_data_entry_t	*dep;		/* data entry pointer */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	xfs_dir2_data_unused_t	*dup;		/* data unused entry pointer */
+	int			error;		/* error return value */
+	xfs_dir2_db_t		fbno;		/* freespace block number */
+	xfs_dabuf_t		*fbp;		/* freespace buffer */
+	int			findex;		/* freespace entry index */
+	xfs_dir2_free_t		*free=NULL;	/* freespace block structure */
+	xfs_dir2_db_t		ifbno;		/* initial freespace block no */
+	xfs_dir2_db_t		lastfbno=0;	/* highest freespace block no */
+	int			length;		/* length of the new entry */
+	int			logfree;	/* need to log free entry */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+	int			needlog;	/* need to log data header */
+	int			needscan;	/* need to rescan data frees */
+	__be16			*tagp;		/* data entry tag pointer */
+	xfs_trans_t		*tp;		/* transaction pointer */
+
+	dp = args->dp;
+	mp = dp->i_mount;
+	tp = args->trans;
+	length = xfs_dir2_data_entsize(args->namelen);
+	/*
+	 * If we came in with a freespace block that means that lookup
+	 * found an entry with our hash value.  This is the freespace
+	 * block for that data entry.
+	 */
+	if (fblk) {
+		fbp = fblk->bp;
+		/*
+		 * Remember initial freespace block number.
+		 */
+		ifbno = fblk->blkno;
+		free = fbp->data;
+		ASSERT(be32_to_cpu(free->hdr.magic) == XFS_DIR2_FREE_MAGIC);
+		findex = fblk->index;
+		/*
+		 * This means the free entry showed that the data block had
+		 * space for our entry, so we remembered it.
+		 * Use that data block.
+		 */
+		if (findex >= 0) {
+			ASSERT(findex < be32_to_cpu(free->hdr.nvalid));
+			ASSERT(be16_to_cpu(free->bests[findex]) != NULLDATAOFF);
+			ASSERT(be16_to_cpu(free->bests[findex]) >= length);
+			dbno = be32_to_cpu(free->hdr.firstdb) + findex;
+		}
+		/*
+		 * The data block looked at didn't have enough room.
+		 * We'll start at the beginning of the freespace entries.
+		 */
+		else {
+			dbno = -1;
+			findex = 0;
+		}
+	}
+	/*
+	 * Didn't come in with a freespace block, so don't have a data block.
+	 */
+	else {
+		ifbno = dbno = -1;
+		fbp = NULL;
+		findex = 0;
+	}
+	/*
+	 * If we don't have a data block yet, we're going to scan the
+	 * freespace blocks looking for one.  Figure out what the
+	 * highest freespace block number is.
+	 */
+	if (dbno == -1) {
+		xfs_fileoff_t	fo;		/* freespace block number */
+
+		if ((error = xfs_bmap_last_offset(tp, dp, &fo, XFS_DATA_FORK)))
+			return error;
+		lastfbno = xfs_dir2_da_to_db(mp, (xfs_dablk_t)fo);
+		fbno = ifbno;
+	}
+	/*
+	 * While we haven't identified a data block, search the freeblock
+	 * data for a good data block.  If we find a null freeblock entry,
+	 * indicating a hole in the data blocks, remember that.
+	 */
+	while (dbno == -1) {
+		/*
+		 * If we don't have a freeblock in hand, get the next one.
+		 */
+		if (fbp == NULL) {
+			/*
+			 * Happens the first time through unless lookup gave
+			 * us a freespace block to start with.
+			 */
+			if (++fbno == 0)
+				fbno = XFS_DIR2_FREE_FIRSTDB(mp);
+			/*
+			 * If it's ifbno we already looked at it.
+			 */
+			if (fbno == ifbno)
+				fbno++;
+			/*
+			 * If it's off the end we're done.
+			 */
+			if (fbno >= lastfbno)
+				break;
+			/*
+			 * Read the block.  There can be holes in the
+			 * freespace blocks, so this might not succeed.
+			 * This should be really rare, so there's no reason
+			 * to avoid it.
+			 */
+			if ((error = xfs_da_read_buf(tp, dp,
+					xfs_dir2_db_to_da(mp, fbno), -2, &fbp,
+					XFS_DATA_FORK))) {
+				return error;
+			}
+			if (unlikely(fbp == NULL)) {
+				continue;
+			}
+			free = fbp->data;
+			ASSERT(be32_to_cpu(free->hdr.magic) == XFS_DIR2_FREE_MAGIC);
+			findex = 0;
+		}
+		/*
+		 * Look at the current free entry.  Is it good enough?
+		 */
+		if (be16_to_cpu(free->bests[findex]) != NULLDATAOFF &&
+		    be16_to_cpu(free->bests[findex]) >= length)
+			dbno = be32_to_cpu(free->hdr.firstdb) + findex;
+		else {
+			/*
+			 * Are we done with the freeblock?
+			 */
+			if (++findex == be32_to_cpu(free->hdr.nvalid)) {
+				/*
+				 * Drop the block.
+				 */
+				xfs_da_brelse(tp, fbp);
+				fbp = NULL;
+				if (fblk && fblk->bp)
+					fblk->bp = NULL;
+			}
+		}
+	}
+	/*
+	 * If we don't have a data block, we need to allocate one and make
+	 * the freespace entries refer to it.
+	 */
+	if (unlikely(dbno == -1)) {
+		/*
+		 * Not allowed to allocate, return failure.
+		 */
+		if ((args->op_flags & XFS_DA_OP_JUSTCHECK) ||
+							args->total == 0) {
+			/*
+			 * Drop the freespace buffer unless it came from our
+			 * caller.
+			 */
+			if ((fblk == NULL || fblk->bp == NULL) && fbp != NULL)
+				xfs_da_buf_done(fbp);
+			return XFS_ERROR(ENOSPC);
+		}
+		/*
+		 * Allocate and initialize the new data block.
+		 */
+		if (unlikely((error = xfs_dir2_grow_inode(args,
+							 XFS_DIR2_DATA_SPACE,
+							 &dbno)) ||
+		    (error = xfs_dir2_data_init(args, dbno, &dbp)))) {
+			/*
+			 * Drop the freespace buffer unless it came from our
+			 * caller.
+			 */
+			if ((fblk == NULL || fblk->bp == NULL) && fbp != NULL)
+				xfs_da_buf_done(fbp);
+			return error;
+		}
+		/*
+		 * If (somehow) we have a freespace block, get rid of it.
+		 */
+		if (fbp)
+			xfs_da_brelse(tp, fbp);
+		if (fblk && fblk->bp)
+			fblk->bp = NULL;
+
+		/*
+		 * Get the freespace block corresponding to the data block
+		 * that was just allocated.
+		 */
+		fbno = xfs_dir2_db_to_fdb(mp, dbno);
+		if (unlikely(error = xfs_da_read_buf(tp, dp,
+				xfs_dir2_db_to_da(mp, fbno), -2, &fbp,
+				XFS_DATA_FORK))) {
+			xfs_da_buf_done(dbp);
+			return error;
+  		}
+		/*
+		 * If there wasn't a freespace block, the read will
+		 * return a NULL fbp.  Allocate and initialize a new one.
+		 */
+		if( fbp == NULL ) {
+			if ((error = xfs_dir2_grow_inode(args, XFS_DIR2_FREE_SPACE,
+							&fbno))) {
+				return error;
+			}
+
+			if (unlikely(xfs_dir2_db_to_fdb(mp, dbno) != fbno)) {
+				xfs_alert(mp,
+			"%s: dir ino " "%llu needed freesp block %lld for\n"
+			"  data block %lld, got %lld ifbno %llu lastfbno %d",
+					__func__, (unsigned long long)dp->i_ino,
+					(long long)xfs_dir2_db_to_fdb(mp, dbno),
+					(long long)dbno, (long long)fbno,
+					(unsigned long long)ifbno, lastfbno);
+				if (fblk) {
+					xfs_alert(mp,
+				" fblk 0x%p blkno %llu index %d magic 0x%x",
+						fblk,
+						(unsigned long long)fblk->blkno,
+						fblk->index,
+						fblk->magic);
+				} else {
+					xfs_alert(mp, " ... fblk is NULL");
+				}
+				XFS_ERROR_REPORT("xfs_dir2_node_addname_int",
+						 XFS_ERRLEVEL_LOW, mp);
+				return XFS_ERROR(EFSCORRUPTED);
+			}
+
+			/*
+			 * Get a buffer for the new block.
+			 */
+			if ((error = xfs_da_get_buf(tp, dp,
+						   xfs_dir2_db_to_da(mp, fbno),
+						   -1, &fbp, XFS_DATA_FORK))) {
+				return error;
+			}
+			ASSERT(fbp != NULL);
+
+			/*
+			 * Initialize the new block to be empty, and remember
+			 * its first slot as our empty slot.
+			 */
+			free = fbp->data;
+			free->hdr.magic = cpu_to_be32(XFS_DIR2_FREE_MAGIC);
+			free->hdr.firstdb = cpu_to_be32(
+				(fbno - XFS_DIR2_FREE_FIRSTDB(mp)) *
+				XFS_DIR2_MAX_FREE_BESTS(mp));
+			free->hdr.nvalid = 0;
+			free->hdr.nused = 0;
+		} else {
+			free = fbp->data;
+			ASSERT(be32_to_cpu(free->hdr.magic) == XFS_DIR2_FREE_MAGIC);
+		}
+
+		/*
+		 * Set the freespace block index from the data block number.
+		 */
+		findex = xfs_dir2_db_to_fdindex(mp, dbno);
+		/*
+		 * If it's after the end of the current entries in the
+		 * freespace block, extend that table.
+		 */
+		if (findex >= be32_to_cpu(free->hdr.nvalid)) {
+			ASSERT(findex < XFS_DIR2_MAX_FREE_BESTS(mp));
+			free->hdr.nvalid = cpu_to_be32(findex + 1);
+			/*
+			 * Tag new entry so nused will go up.
+			 */
+			free->bests[findex] = cpu_to_be16(NULLDATAOFF);
+		}
+		/*
+		 * If this entry was for an empty data block
+		 * (this should always be true) then update the header.
+		 */
+		if (be16_to_cpu(free->bests[findex]) == NULLDATAOFF) {
+			be32_add_cpu(&free->hdr.nused, 1);
+			xfs_dir2_free_log_header(tp, fbp);
+		}
+		/*
+		 * Update the real value in the table.
+		 * We haven't allocated the data entry yet so this will
+		 * change again.
+		 */
+		data = dbp->data;
+		free->bests[findex] = data->hdr.bestfree[0].length;
+		logfree = 1;
+	}
+	/*
+	 * We had a data block so we don't have to make a new one.
+	 */
+	else {
+		/*
+		 * If just checking, we succeeded.
+		 */
+		if (args->op_flags & XFS_DA_OP_JUSTCHECK) {
+			if ((fblk == NULL || fblk->bp == NULL) && fbp != NULL)
+				xfs_da_buf_done(fbp);
+			return 0;
+		}
+		/*
+		 * Read the data block in.
+		 */
+		if (unlikely(
+		    error = xfs_da_read_buf(tp, dp, xfs_dir2_db_to_da(mp, dbno),
+				-1, &dbp, XFS_DATA_FORK))) {
+			if ((fblk == NULL || fblk->bp == NULL) && fbp != NULL)
+				xfs_da_buf_done(fbp);
+			return error;
+		}
+		data = dbp->data;
+		logfree = 0;
+	}
+	ASSERT(be16_to_cpu(data->hdr.bestfree[0].length) >= length);
+	/*
+	 * Point to the existing unused space.
+	 */
+	dup = (xfs_dir2_data_unused_t *)
+	      ((char *)data + be16_to_cpu(data->hdr.bestfree[0].offset));
+	needscan = needlog = 0;
+	/*
+	 * Mark the first part of the unused space, inuse for us.
+	 */
+	xfs_dir2_data_use_free(tp, dbp, dup,
+		(xfs_dir2_data_aoff_t)((char *)dup - (char *)data), length,
+		&needlog, &needscan);
+	/*
+	 * Fill in the new entry and log it.
+	 */
+	dep = (xfs_dir2_data_entry_t *)dup;
+	dep->inumber = cpu_to_be64(args->inumber);
+	dep->namelen = args->namelen;
+	memcpy(dep->name, args->name, dep->namelen);
+	tagp = xfs_dir2_data_entry_tag_p(dep);
+	*tagp = cpu_to_be16((char *)dep - (char *)data);
+	xfs_dir2_data_log_entry(tp, dbp, dep);
+	/*
+	 * Rescan the block for bestfree if needed.
+	 */
+	if (needscan)
+		xfs_dir2_data_freescan(mp, data, &needlog);
+	/*
+	 * Log the data block header if needed.
+	 */
+	if (needlog)
+		xfs_dir2_data_log_header(tp, dbp);
+	/*
+	 * If the freespace entry is now wrong, update it.
+	 */
+	if (be16_to_cpu(free->bests[findex]) != be16_to_cpu(data->hdr.bestfree[0].length)) {
+		free->bests[findex] = data->hdr.bestfree[0].length;
+		logfree = 1;
+	}
+	/*
+	 * Log the freespace entry if needed.
+	 */
+	if (logfree)
+		xfs_dir2_free_log_bests(tp, fbp, findex, findex);
+	/*
+	 * If the caller didn't hand us the freespace block, drop it.
+	 */
+	if ((fblk == NULL || fblk->bp == NULL) && fbp != NULL)
+		xfs_da_buf_done(fbp);
+	/*
+	 * Return the data block and offset in args, then drop the data block.
+	 */
+	args->blkno = (xfs_dablk_t)dbno;
+	args->index = be16_to_cpu(*tagp);
+	xfs_da_buf_done(dbp);
+	return 0;
+}
+
+/*
+ * Lookup an entry in a node-format directory.
+ * All the real work happens in xfs_da_node_lookup_int.
+ * The only real output is the inode number of the entry.
+ */
+int						/* error */
+xfs_dir2_node_lookup(
+	xfs_da_args_t	*args)			/* operation arguments */
+{
+	int		error;			/* error return value */
+	int		i;			/* btree level */
+	int		rval;			/* operation return value */
+	xfs_da_state_t	*state;			/* btree cursor */
+
+	trace_xfs_dir2_node_lookup(args);
+
+	/*
+	 * Allocate and initialize the btree cursor.
+	 */
+	state = xfs_da_state_alloc();
+	state->args = args;
+	state->mp = args->dp->i_mount;
+	state->blocksize = state->mp->m_dirblksize;
+	state->node_ents = state->mp->m_dir_node_ents;
+	/*
+	 * Fill in the path to the entry in the cursor.
+	 */
+	error = xfs_da_node_lookup_int(state, &rval);
+	if (error)
+		rval = error;
+	else if (rval == ENOENT && args->cmpresult == XFS_CMP_CASE) {
+		/* If a CI match, dup the actual name and return EEXIST */
+		xfs_dir2_data_entry_t	*dep;
+
+		dep = (xfs_dir2_data_entry_t *)((char *)state->extrablk.bp->
+						data + state->extrablk.index);
+		rval = xfs_dir_cilookup_result(args, dep->name, dep->namelen);
+	}
+	/*
+	 * Release the btree blocks and leaf block.
+	 */
+	for (i = 0; i < state->path.active; i++) {
+		xfs_da_brelse(args->trans, state->path.blk[i].bp);
+		state->path.blk[i].bp = NULL;
+	}
+	/*
+	 * Release the data block if we have it.
+	 */
+	if (state->extravalid && state->extrablk.bp) {
+		xfs_da_brelse(args->trans, state->extrablk.bp);
+		state->extrablk.bp = NULL;
+	}
+	xfs_da_state_free(state);
+	return rval;
+}
+
+/*
+ * Remove an entry from a node-format directory.
+ */
+int						/* error */
+xfs_dir2_node_removename(
+	xfs_da_args_t		*args)		/* operation arguments */
+{
+	xfs_da_state_blk_t	*blk;		/* leaf block */
+	int			error;		/* error return value */
+	int			rval;		/* operation return value */
+	xfs_da_state_t		*state;		/* btree cursor */
+
+	trace_xfs_dir2_node_removename(args);
+
+	/*
+	 * Allocate and initialize the btree cursor.
+	 */
+	state = xfs_da_state_alloc();
+	state->args = args;
+	state->mp = args->dp->i_mount;
+	state->blocksize = state->mp->m_dirblksize;
+	state->node_ents = state->mp->m_dir_node_ents;
+	/*
+	 * Look up the entry we're deleting, set up the cursor.
+	 */
+	error = xfs_da_node_lookup_int(state, &rval);
+	if (error)
+		rval = error;
+	/*
+	 * Didn't find it, upper layer screwed up.
+	 */
+	if (rval != EEXIST) {
+		xfs_da_state_free(state);
+		return rval;
+	}
+	blk = &state->path.blk[state->path.active - 1];
+	ASSERT(blk->magic == XFS_DIR2_LEAFN_MAGIC);
+	ASSERT(state->extravalid);
+	/*
+	 * Remove the leaf and data entries.
+	 * Extrablk refers to the data block.
+	 */
+	error = xfs_dir2_leafn_remove(args, blk->bp, blk->index,
+		&state->extrablk, &rval);
+	if (error)
+		return error;
+	/*
+	 * Fix the hash values up the btree.
+	 */
+	xfs_da_fixhashpath(state, &state->path);
+	/*
+	 * If we need to join leaf blocks, do it.
+	 */
+	if (rval && state->path.active > 1)
+		error = xfs_da_join(state);
+	/*
+	 * If no errors so far, try conversion to leaf format.
+	 */
+	if (!error)
+		error = xfs_dir2_node_to_leaf(state);
+	xfs_da_state_free(state);
+	return error;
+}
+
+/*
+ * Replace an entry's inode number in a node-format directory.
+ */
+int						/* error */
+xfs_dir2_node_replace(
+	xfs_da_args_t		*args)		/* operation arguments */
+{
+	xfs_da_state_blk_t	*blk;		/* leaf block */
+	xfs_dir2_data_t		*data;		/* data block structure */
+	xfs_dir2_data_entry_t	*dep;		/* data entry changed */
+	int			error;		/* error return value */
+	int			i;		/* btree level */
+	xfs_ino_t		inum;		/* new inode number */
+	xfs_dir2_leaf_t		*leaf;		/* leaf structure */
+	xfs_dir2_leaf_entry_t	*lep;		/* leaf entry being changed */
+	int			rval;		/* internal return value */
+	xfs_da_state_t		*state;		/* btree cursor */
+
+	trace_xfs_dir2_node_replace(args);
+
+	/*
+	 * Allocate and initialize the btree cursor.
+	 */
+	state = xfs_da_state_alloc();
+	state->args = args;
+	state->mp = args->dp->i_mount;
+	state->blocksize = state->mp->m_dirblksize;
+	state->node_ents = state->mp->m_dir_node_ents;
+	inum = args->inumber;
+	/*
+	 * Lookup the entry to change in the btree.
+	 */
+	error = xfs_da_node_lookup_int(state, &rval);
+	if (error) {
+		rval = error;
+	}
+	/*
+	 * It should be found, since the vnodeops layer has looked it up
+	 * and locked it.  But paranoia is good.
+	 */
+	if (rval == EEXIST) {
+		/*
+		 * Find the leaf entry.
+		 */
+		blk = &state->path.blk[state->path.active - 1];
+		ASSERT(blk->magic == XFS_DIR2_LEAFN_MAGIC);
+		leaf = blk->bp->data;
+		lep = &leaf->ents[blk->index];
+		ASSERT(state->extravalid);
+		/*
+		 * Point to the data entry.
+		 */
+		data = state->extrablk.bp->data;
+		ASSERT(be32_to_cpu(data->hdr.magic) == XFS_DIR2_DATA_MAGIC);
+		dep = (xfs_dir2_data_entry_t *)
+		      ((char *)data +
+		       xfs_dir2_dataptr_to_off(state->mp, be32_to_cpu(lep->address)));
+		ASSERT(inum != be64_to_cpu(dep->inumber));
+		/*
+		 * Fill in the new inode number and log the entry.
+		 */
+		dep->inumber = cpu_to_be64(inum);
+		xfs_dir2_data_log_entry(args->trans, state->extrablk.bp, dep);
+		rval = 0;
+	}
+	/*
+	 * Didn't find it, and we're holding a data block.  Drop it.
+	 */
+	else if (state->extravalid) {
+		xfs_da_brelse(args->trans, state->extrablk.bp);
+		state->extrablk.bp = NULL;
+	}
+	/*
+	 * Release all the buffers in the cursor.
+	 */
+	for (i = 0; i < state->path.active; i++) {
+		xfs_da_brelse(args->trans, state->path.blk[i].bp);
+		state->path.blk[i].bp = NULL;
+	}
+	xfs_da_state_free(state);
+	return rval;
+}
+
+/*
+ * Trim off a trailing empty freespace block.
+ * Return (in rvalp) 1 if we did it, 0 if not.
+ */
+int						/* error */
+xfs_dir2_node_trim_free(
+	xfs_da_args_t		*args,		/* operation arguments */
+	xfs_fileoff_t		fo,		/* free block number */
+	int			*rvalp)		/* out: did something */
+{
+	xfs_dabuf_t		*bp;		/* freespace buffer */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	int			error;		/* error return code */
+	xfs_dir2_free_t		*free;		/* freespace structure */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+	xfs_trans_t		*tp;		/* transaction pointer */
+
+	dp = args->dp;
+	mp = dp->i_mount;
+	tp = args->trans;
+	/*
+	 * Read the freespace block.
+	 */
+	if (unlikely(error = xfs_da_read_buf(tp, dp, (xfs_dablk_t)fo, -2, &bp,
+			XFS_DATA_FORK))) {
+		return error;
+	}
+
+	/*
+	 * There can be holes in freespace.  If fo is a hole, there's
+	 * nothing to do.
+	 */
+	if (bp == NULL) {
+		return 0;
+	}
+	free = bp->data;
+	ASSERT(be32_to_cpu(free->hdr.magic) == XFS_DIR2_FREE_MAGIC);
+	/*
+	 * If there are used entries, there's nothing to do.
+	 */
+	if (be32_to_cpu(free->hdr.nused) > 0) {
+		xfs_da_brelse(tp, bp);
+		*rvalp = 0;
+		return 0;
+	}
+	/*
+	 * Blow the block away.
+	 */
+	if ((error =
+	    xfs_dir2_shrink_inode(args, xfs_dir2_da_to_db(mp, (xfs_dablk_t)fo),
+		    bp))) {
+		/*
+		 * Can't fail with ENOSPC since that only happens with no
+		 * space reservation, when breaking up an extent into two
+		 * pieces.  This is the last block of an extent.
+		 */
+		ASSERT(error != ENOSPC);
+		xfs_da_brelse(tp, bp);
+		return error;
+	}
+	/*
+	 * Return that we succeeded.
+	 */
+	*rvalp = 1;
+	return 0;
+}
diff --git a/fs/xfs/xfs_dir2_node.h b/fs/xfs/xfs_dir2_node.h
new file mode 100644
index 00000000..82dfe714
--- /dev/null
+++ b/fs/xfs/xfs_dir2_node.h
@@ -0,0 +1,100 @@
+/*
+ * Copyright (c) 2000,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_DIR2_NODE_H__
+#define	__XFS_DIR2_NODE_H__
+
+/*
+ * Directory version 2, btree node format structures
+ */
+
+struct uio;
+struct xfs_dabuf;
+struct xfs_da_args;
+struct xfs_da_state;
+struct xfs_da_state_blk;
+struct xfs_inode;
+struct xfs_trans;
+
+/*
+ * Offset of the freespace index.
+ */
+#define	XFS_DIR2_FREE_SPACE	2
+#define	XFS_DIR2_FREE_OFFSET	(XFS_DIR2_FREE_SPACE * XFS_DIR2_SPACE_SIZE)
+#define	XFS_DIR2_FREE_FIRSTDB(mp)	\
+	xfs_dir2_byte_to_db(mp, XFS_DIR2_FREE_OFFSET)
+
+#define	XFS_DIR2_FREE_MAGIC	0x58443246	/* XD2F */
+
+typedef	struct xfs_dir2_free_hdr {
+	__be32			magic;		/* XFS_DIR2_FREE_MAGIC */
+	__be32			firstdb;	/* db of first entry */
+	__be32			nvalid;		/* count of valid entries */
+	__be32			nused;		/* count of used entries */
+} xfs_dir2_free_hdr_t;
+
+typedef struct xfs_dir2_free {
+	xfs_dir2_free_hdr_t	hdr;		/* block header */
+	__be16			bests[1];	/* best free counts */
+						/* unused entries are -1 */
+} xfs_dir2_free_t;
+
+#define	XFS_DIR2_MAX_FREE_BESTS(mp)	\
+	(((mp)->m_dirblksize - (uint)sizeof(xfs_dir2_free_hdr_t)) / \
+	 (uint)sizeof(xfs_dir2_data_off_t))
+
+/*
+ * Convert data space db to the corresponding free db.
+ */
+static inline xfs_dir2_db_t
+xfs_dir2_db_to_fdb(struct xfs_mount *mp, xfs_dir2_db_t db)
+{
+	return (XFS_DIR2_FREE_FIRSTDB(mp) + (db) / XFS_DIR2_MAX_FREE_BESTS(mp));
+}
+
+/*
+ * Convert data space db to the corresponding index in a free db.
+ */
+static inline int
+xfs_dir2_db_to_fdindex(struct xfs_mount *mp, xfs_dir2_db_t db)
+{
+	return ((db) % XFS_DIR2_MAX_FREE_BESTS(mp));
+}
+
+extern int xfs_dir2_leaf_to_node(struct xfs_da_args *args,
+				 struct xfs_dabuf *lbp);
+extern xfs_dahash_t xfs_dir2_leafn_lasthash(struct xfs_dabuf *bp, int *count);
+extern int xfs_dir2_leafn_lookup_int(struct xfs_dabuf *bp,
+				     struct xfs_da_args *args, int *indexp,
+				     struct xfs_da_state *state);
+extern int xfs_dir2_leafn_order(struct xfs_dabuf *leaf1_bp,
+				struct xfs_dabuf *leaf2_bp);
+extern int xfs_dir2_leafn_split(struct xfs_da_state *state,
+				struct xfs_da_state_blk *oldblk,
+				struct xfs_da_state_blk *newblk);
+extern int xfs_dir2_leafn_toosmall(struct xfs_da_state *state, int *action);
+extern void xfs_dir2_leafn_unbalance(struct xfs_da_state *state,
+				     struct xfs_da_state_blk *drop_blk,
+				     struct xfs_da_state_blk *save_blk);
+extern int xfs_dir2_node_addname(struct xfs_da_args *args);
+extern int xfs_dir2_node_lookup(struct xfs_da_args *args);
+extern int xfs_dir2_node_removename(struct xfs_da_args *args);
+extern int xfs_dir2_node_replace(struct xfs_da_args *args);
+extern int xfs_dir2_node_trim_free(struct xfs_da_args *args, xfs_fileoff_t fo,
+				   int *rvalp);
+
+#endif	/* __XFS_DIR2_NODE_H__ */
diff --git a/fs/xfs/xfs_dir2_sf.c b/fs/xfs/xfs_dir2_sf.c
new file mode 100644
index 00000000..b1bae6b1
--- /dev/null
+++ b/fs/xfs/xfs_dir2_sf.c
@@ -0,0 +1,1267 @@
+/*
+ * Copyright (c) 2000-2003,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_dir2.h"
+#include "xfs_mount.h"
+#include "xfs_da_btree.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_dir2_sf.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_inode_item.h"
+#include "xfs_error.h"
+#include "xfs_dir2_data.h"
+#include "xfs_dir2_leaf.h"
+#include "xfs_dir2_block.h"
+#include "xfs_trace.h"
+
+/*
+ * Prototypes for internal functions.
+ */
+static void xfs_dir2_sf_addname_easy(xfs_da_args_t *args,
+				     xfs_dir2_sf_entry_t *sfep,
+				     xfs_dir2_data_aoff_t offset,
+				     int new_isize);
+static void xfs_dir2_sf_addname_hard(xfs_da_args_t *args, int objchange,
+				     int new_isize);
+static int xfs_dir2_sf_addname_pick(xfs_da_args_t *args, int objchange,
+				    xfs_dir2_sf_entry_t **sfepp,
+				    xfs_dir2_data_aoff_t *offsetp);
+#ifdef DEBUG
+static void xfs_dir2_sf_check(xfs_da_args_t *args);
+#else
+#define	xfs_dir2_sf_check(args)
+#endif /* DEBUG */
+#if XFS_BIG_INUMS
+static void xfs_dir2_sf_toino4(xfs_da_args_t *args);
+static void xfs_dir2_sf_toino8(xfs_da_args_t *args);
+#endif /* XFS_BIG_INUMS */
+
+/*
+ * Given a block directory (dp/block), calculate its size as a shortform (sf)
+ * directory and a header for the sf directory, if it will fit it the
+ * space currently present in the inode.  If it won't fit, the output
+ * size is too big (but not accurate).
+ */
+int						/* size for sf form */
+xfs_dir2_block_sfsize(
+	xfs_inode_t		*dp,		/* incore inode pointer */
+	xfs_dir2_block_t	*block,		/* block directory data */
+	xfs_dir2_sf_hdr_t	*sfhp)		/* output: header for sf form */
+{
+	xfs_dir2_dataptr_t	addr;		/* data entry address */
+	xfs_dir2_leaf_entry_t	*blp;		/* leaf area of the block */
+	xfs_dir2_block_tail_t	*btp;		/* tail area of the block */
+	int			count;		/* shortform entry count */
+	xfs_dir2_data_entry_t	*dep;		/* data entry in the block */
+	int			i;		/* block entry index */
+	int			i8count;	/* count of big-inode entries */
+	int			isdot;		/* entry is "." */
+	int			isdotdot;	/* entry is ".." */
+	xfs_mount_t		*mp;		/* mount structure pointer */
+	int			namelen;	/* total name bytes */
+	xfs_ino_t		parent = 0;	/* parent inode number */
+	int			size=0;		/* total computed size */
+
+	mp = dp->i_mount;
+
+	count = i8count = namelen = 0;
+	btp = xfs_dir2_block_tail_p(mp, block);
+	blp = xfs_dir2_block_leaf_p(btp);
+
+	/*
+	 * Iterate over the block's data entries by using the leaf pointers.
+	 */
+	for (i = 0; i < be32_to_cpu(btp->count); i++) {
+		if ((addr = be32_to_cpu(blp[i].address)) == XFS_DIR2_NULL_DATAPTR)
+			continue;
+		/*
+		 * Calculate the pointer to the entry at hand.
+		 */
+		dep = (xfs_dir2_data_entry_t *)
+		      ((char *)block + xfs_dir2_dataptr_to_off(mp, addr));
+		/*
+		 * Detect . and .., so we can special-case them.
+		 * . is not included in sf directories.
+		 * .. is included by just the parent inode number.
+		 */
+		isdot = dep->namelen == 1 && dep->name[0] == '.';
+		isdotdot =
+			dep->namelen == 2 &&
+			dep->name[0] == '.' && dep->name[1] == '.';
+#if XFS_BIG_INUMS
+		if (!isdot)
+			i8count += be64_to_cpu(dep->inumber) > XFS_DIR2_MAX_SHORT_INUM;
+#endif
+		if (!isdot && !isdotdot) {
+			count++;
+			namelen += dep->namelen;
+		} else if (isdotdot)
+			parent = be64_to_cpu(dep->inumber);
+		/*
+		 * Calculate the new size, see if we should give up yet.
+		 */
+		size = xfs_dir2_sf_hdr_size(i8count) +		/* header */
+		       count +					/* namelen */
+		       count * (uint)sizeof(xfs_dir2_sf_off_t) + /* offset */
+		       namelen +				/* name */
+		       (i8count ?				/* inumber */
+				(uint)sizeof(xfs_dir2_ino8_t) * count :
+				(uint)sizeof(xfs_dir2_ino4_t) * count);
+		if (size > XFS_IFORK_DSIZE(dp))
+			return size;		/* size value is a failure */
+	}
+	/*
+	 * Create the output header, if it worked.
+	 */
+	sfhp->count = count;
+	sfhp->i8count = i8count;
+	xfs_dir2_sf_put_inumber((xfs_dir2_sf_t *)sfhp, &parent, &sfhp->parent);
+	return size;
+}
+
+/*
+ * Convert a block format directory to shortform.
+ * Caller has already checked that it will fit, and built us a header.
+ */
+int						/* error */
+xfs_dir2_block_to_sf(
+	xfs_da_args_t		*args,		/* operation arguments */
+	xfs_dabuf_t		*bp,		/* block buffer */
+	int			size,		/* shortform directory size */
+	xfs_dir2_sf_hdr_t	*sfhp)		/* shortform directory hdr */
+{
+	xfs_dir2_block_t	*block;		/* block structure */
+	xfs_dir2_block_tail_t	*btp;		/* block tail pointer */
+	xfs_dir2_data_entry_t	*dep;		/* data entry pointer */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	xfs_dir2_data_unused_t	*dup;		/* unused data pointer */
+	char			*endptr;	/* end of data entries */
+	int			error;		/* error return value */
+	int			logflags;	/* inode logging flags */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+	char			*ptr;		/* current data pointer */
+	xfs_dir2_sf_entry_t	*sfep;		/* shortform entry */
+	xfs_dir2_sf_t		*sfp;		/* shortform structure */
+	xfs_ino_t               temp;
+
+	trace_xfs_dir2_block_to_sf(args);
+
+	dp = args->dp;
+	mp = dp->i_mount;
+
+	/*
+	 * Make a copy of the block data, so we can shrink the inode
+	 * and add local data.
+	 */
+	block = kmem_alloc(mp->m_dirblksize, KM_SLEEP);
+	memcpy(block, bp->data, mp->m_dirblksize);
+	logflags = XFS_ILOG_CORE;
+	if ((error = xfs_dir2_shrink_inode(args, mp->m_dirdatablk, bp))) {
+		ASSERT(error != ENOSPC);
+		goto out;
+	}
+	/*
+	 * The buffer is now unconditionally gone, whether
+	 * xfs_dir2_shrink_inode worked or not.
+	 *
+	 * Convert the inode to local format.
+	 */
+	dp->i_df.if_flags &= ~XFS_IFEXTENTS;
+	dp->i_df.if_flags |= XFS_IFINLINE;
+	dp->i_d.di_format = XFS_DINODE_FMT_LOCAL;
+	ASSERT(dp->i_df.if_bytes == 0);
+	xfs_idata_realloc(dp, size, XFS_DATA_FORK);
+	logflags |= XFS_ILOG_DDATA;
+	/*
+	 * Copy the header into the newly allocate local space.
+	 */
+	sfp = (xfs_dir2_sf_t *)dp->i_df.if_u1.if_data;
+	memcpy(sfp, sfhp, xfs_dir2_sf_hdr_size(sfhp->i8count));
+	dp->i_d.di_size = size;
+	/*
+	 * Set up to loop over the block's entries.
+	 */
+	btp = xfs_dir2_block_tail_p(mp, block);
+	ptr = (char *)block->u;
+	endptr = (char *)xfs_dir2_block_leaf_p(btp);
+	sfep = xfs_dir2_sf_firstentry(sfp);
+	/*
+	 * Loop over the active and unused entries.
+	 * Stop when we reach the leaf/tail portion of the block.
+	 */
+	while (ptr < endptr) {
+		/*
+		 * If it's unused, just skip over it.
+		 */
+		dup = (xfs_dir2_data_unused_t *)ptr;
+		if (be16_to_cpu(dup->freetag) == XFS_DIR2_DATA_FREE_TAG) {
+			ptr += be16_to_cpu(dup->length);
+			continue;
+		}
+		dep = (xfs_dir2_data_entry_t *)ptr;
+		/*
+		 * Skip .
+		 */
+		if (dep->namelen == 1 && dep->name[0] == '.')
+			ASSERT(be64_to_cpu(dep->inumber) == dp->i_ino);
+		/*
+		 * Skip .., but make sure the inode number is right.
+		 */
+		else if (dep->namelen == 2 &&
+			 dep->name[0] == '.' && dep->name[1] == '.')
+			ASSERT(be64_to_cpu(dep->inumber) ==
+			       xfs_dir2_sf_get_inumber(sfp, &sfp->hdr.parent));
+		/*
+		 * Normal entry, copy it into shortform.
+		 */
+		else {
+			sfep->namelen = dep->namelen;
+			xfs_dir2_sf_put_offset(sfep,
+				(xfs_dir2_data_aoff_t)
+				((char *)dep - (char *)block));
+			memcpy(sfep->name, dep->name, dep->namelen);
+			temp = be64_to_cpu(dep->inumber);
+			xfs_dir2_sf_put_inumber(sfp, &temp,
+				xfs_dir2_sf_inumberp(sfep));
+			sfep = xfs_dir2_sf_nextentry(sfp, sfep);
+		}
+		ptr += xfs_dir2_data_entsize(dep->namelen);
+	}
+	ASSERT((char *)sfep - (char *)sfp == size);
+	xfs_dir2_sf_check(args);
+out:
+	xfs_trans_log_inode(args->trans, dp, logflags);
+	kmem_free(block);
+	return error;
+}
+
+/*
+ * Add a name to a shortform directory.
+ * There are two algorithms, "easy" and "hard" which we decide on
+ * before changing anything.
+ * Convert to block form if necessary, if the new entry won't fit.
+ */
+int						/* error */
+xfs_dir2_sf_addname(
+	xfs_da_args_t		*args)		/* operation arguments */
+{
+	int			add_entsize;	/* size of the new entry */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	int			error;		/* error return value */
+	int			incr_isize;	/* total change in size */
+	int			new_isize;	/* di_size after adding name */
+	int			objchange;	/* changing to 8-byte inodes */
+	xfs_dir2_data_aoff_t	offset = 0;	/* offset for new entry */
+	int			old_isize;	/* di_size before adding name */
+	int			pick;		/* which algorithm to use */
+	xfs_dir2_sf_t		*sfp;		/* shortform structure */
+	xfs_dir2_sf_entry_t	*sfep = NULL;	/* shortform entry */
+
+	trace_xfs_dir2_sf_addname(args);
+
+	ASSERT(xfs_dir2_sf_lookup(args) == ENOENT);
+	dp = args->dp;
+	ASSERT(dp->i_df.if_flags & XFS_IFINLINE);
+	/*
+	 * Make sure the shortform value has some of its header.
+	 */
+	if (dp->i_d.di_size < offsetof(xfs_dir2_sf_hdr_t, parent)) {
+		ASSERT(XFS_FORCED_SHUTDOWN(dp->i_mount));
+		return XFS_ERROR(EIO);
+	}
+	ASSERT(dp->i_df.if_bytes == dp->i_d.di_size);
+	ASSERT(dp->i_df.if_u1.if_data != NULL);
+	sfp = (xfs_dir2_sf_t *)dp->i_df.if_u1.if_data;
+	ASSERT(dp->i_d.di_size >= xfs_dir2_sf_hdr_size(sfp->hdr.i8count));
+	/*
+	 * Compute entry (and change in) size.
+	 */
+	add_entsize = xfs_dir2_sf_entsize_byname(sfp, args->namelen);
+	incr_isize = add_entsize;
+	objchange = 0;
+#if XFS_BIG_INUMS
+	/*
+	 * Do we have to change to 8 byte inodes?
+	 */
+	if (args->inumber > XFS_DIR2_MAX_SHORT_INUM && sfp->hdr.i8count == 0) {
+		/*
+		 * Yes, adjust the entry size and the total size.
+		 */
+		add_entsize +=
+			(uint)sizeof(xfs_dir2_ino8_t) -
+			(uint)sizeof(xfs_dir2_ino4_t);
+		incr_isize +=
+			(sfp->hdr.count + 2) *
+			((uint)sizeof(xfs_dir2_ino8_t) -
+			 (uint)sizeof(xfs_dir2_ino4_t));
+		objchange = 1;
+	}
+#endif
+	old_isize = (int)dp->i_d.di_size;
+	new_isize = old_isize + incr_isize;
+	/*
+	 * Won't fit as shortform any more (due to size),
+	 * or the pick routine says it won't (due to offset values).
+	 */
+	if (new_isize > XFS_IFORK_DSIZE(dp) ||
+	    (pick =
+	     xfs_dir2_sf_addname_pick(args, objchange, &sfep, &offset)) == 0) {
+		/*
+		 * Just checking or no space reservation, it doesn't fit.
+		 */
+		if ((args->op_flags & XFS_DA_OP_JUSTCHECK) || args->total == 0)
+			return XFS_ERROR(ENOSPC);
+		/*
+		 * Convert to block form then add the name.
+		 */
+		error = xfs_dir2_sf_to_block(args);
+		if (error)
+			return error;
+		return xfs_dir2_block_addname(args);
+	}
+	/*
+	 * Just checking, it fits.
+	 */
+	if (args->op_flags & XFS_DA_OP_JUSTCHECK)
+		return 0;
+	/*
+	 * Do it the easy way - just add it at the end.
+	 */
+	if (pick == 1)
+		xfs_dir2_sf_addname_easy(args, sfep, offset, new_isize);
+	/*
+	 * Do it the hard way - look for a place to insert the new entry.
+	 * Convert to 8 byte inode numbers first if necessary.
+	 */
+	else {
+		ASSERT(pick == 2);
+#if XFS_BIG_INUMS
+		if (objchange)
+			xfs_dir2_sf_toino8(args);
+#endif
+		xfs_dir2_sf_addname_hard(args, objchange, new_isize);
+	}
+	xfs_trans_log_inode(args->trans, dp, XFS_ILOG_CORE | XFS_ILOG_DDATA);
+	return 0;
+}
+
+/*
+ * Add the new entry the "easy" way.
+ * This is copying the old directory and adding the new entry at the end.
+ * Since it's sorted by "offset" we need room after the last offset
+ * that's already there, and then room to convert to a block directory.
+ * This is already checked by the pick routine.
+ */
+static void
+xfs_dir2_sf_addname_easy(
+	xfs_da_args_t		*args,		/* operation arguments */
+	xfs_dir2_sf_entry_t	*sfep,		/* pointer to new entry */
+	xfs_dir2_data_aoff_t	offset,		/* offset to use for new ent */
+	int			new_isize)	/* new directory size */
+{
+	int			byteoff;	/* byte offset in sf dir */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	xfs_dir2_sf_t		*sfp;		/* shortform structure */
+
+	dp = args->dp;
+
+	sfp = (xfs_dir2_sf_t *)dp->i_df.if_u1.if_data;
+	byteoff = (int)((char *)sfep - (char *)sfp);
+	/*
+	 * Grow the in-inode space.
+	 */
+	xfs_idata_realloc(dp, xfs_dir2_sf_entsize_byname(sfp, args->namelen),
+		XFS_DATA_FORK);
+	/*
+	 * Need to set up again due to realloc of the inode data.
+	 */
+	sfp = (xfs_dir2_sf_t *)dp->i_df.if_u1.if_data;
+	sfep = (xfs_dir2_sf_entry_t *)((char *)sfp + byteoff);
+	/*
+	 * Fill in the new entry.
+	 */
+	sfep->namelen = args->namelen;
+	xfs_dir2_sf_put_offset(sfep, offset);
+	memcpy(sfep->name, args->name, sfep->namelen);
+	xfs_dir2_sf_put_inumber(sfp, &args->inumber,
+		xfs_dir2_sf_inumberp(sfep));
+	/*
+	 * Update the header and inode.
+	 */
+	sfp->hdr.count++;
+#if XFS_BIG_INUMS
+	if (args->inumber > XFS_DIR2_MAX_SHORT_INUM)
+		sfp->hdr.i8count++;
+#endif
+	dp->i_d.di_size = new_isize;
+	xfs_dir2_sf_check(args);
+}
+
+/*
+ * Add the new entry the "hard" way.
+ * The caller has already converted to 8 byte inode numbers if necessary,
+ * in which case we need to leave the i8count at 1.
+ * Find a hole that the new entry will fit into, and copy
+ * the first part of the entries, the new entry, and the last part of
+ * the entries.
+ */
+/* ARGSUSED */
+static void
+xfs_dir2_sf_addname_hard(
+	xfs_da_args_t		*args,		/* operation arguments */
+	int			objchange,	/* changing inode number size */
+	int			new_isize)	/* new directory size */
+{
+	int			add_datasize;	/* data size need for new ent */
+	char			*buf;		/* buffer for old */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	int			eof;		/* reached end of old dir */
+	int			nbytes;		/* temp for byte copies */
+	xfs_dir2_data_aoff_t	new_offset;	/* next offset value */
+	xfs_dir2_data_aoff_t	offset;		/* current offset value */
+	int			old_isize;	/* previous di_size */
+	xfs_dir2_sf_entry_t	*oldsfep;	/* entry in original dir */
+	xfs_dir2_sf_t		*oldsfp;	/* original shortform dir */
+	xfs_dir2_sf_entry_t	*sfep;		/* entry in new dir */
+	xfs_dir2_sf_t		*sfp;		/* new shortform dir */
+
+	/*
+	 * Copy the old directory to the stack buffer.
+	 */
+	dp = args->dp;
+
+	sfp = (xfs_dir2_sf_t *)dp->i_df.if_u1.if_data;
+	old_isize = (int)dp->i_d.di_size;
+	buf = kmem_alloc(old_isize, KM_SLEEP);
+	oldsfp = (xfs_dir2_sf_t *)buf;
+	memcpy(oldsfp, sfp, old_isize);
+	/*
+	 * Loop over the old directory finding the place we're going
+	 * to insert the new entry.
+	 * If it's going to end up at the end then oldsfep will point there.
+	 */
+	for (offset = XFS_DIR2_DATA_FIRST_OFFSET,
+	      oldsfep = xfs_dir2_sf_firstentry(oldsfp),
+	      add_datasize = xfs_dir2_data_entsize(args->namelen),
+	      eof = (char *)oldsfep == &buf[old_isize];
+	     !eof;
+	     offset = new_offset + xfs_dir2_data_entsize(oldsfep->namelen),
+	      oldsfep = xfs_dir2_sf_nextentry(oldsfp, oldsfep),
+	      eof = (char *)oldsfep == &buf[old_isize]) {
+		new_offset = xfs_dir2_sf_get_offset(oldsfep);
+		if (offset + add_datasize <= new_offset)
+			break;
+	}
+	/*
+	 * Get rid of the old directory, then allocate space for
+	 * the new one.  We do this so xfs_idata_realloc won't copy
+	 * the data.
+	 */
+	xfs_idata_realloc(dp, -old_isize, XFS_DATA_FORK);
+	xfs_idata_realloc(dp, new_isize, XFS_DATA_FORK);
+	/*
+	 * Reset the pointer since the buffer was reallocated.
+	 */
+	sfp = (xfs_dir2_sf_t *)dp->i_df.if_u1.if_data;
+	/*
+	 * Copy the first part of the directory, including the header.
+	 */
+	nbytes = (int)((char *)oldsfep - (char *)oldsfp);
+	memcpy(sfp, oldsfp, nbytes);
+	sfep = (xfs_dir2_sf_entry_t *)((char *)sfp + nbytes);
+	/*
+	 * Fill in the new entry, and update the header counts.
+	 */
+	sfep->namelen = args->namelen;
+	xfs_dir2_sf_put_offset(sfep, offset);
+	memcpy(sfep->name, args->name, sfep->namelen);
+	xfs_dir2_sf_put_inumber(sfp, &args->inumber,
+		xfs_dir2_sf_inumberp(sfep));
+	sfp->hdr.count++;
+#if XFS_BIG_INUMS
+	if (args->inumber > XFS_DIR2_MAX_SHORT_INUM && !objchange)
+		sfp->hdr.i8count++;
+#endif
+	/*
+	 * If there's more left to copy, do that.
+	 */
+	if (!eof) {
+		sfep = xfs_dir2_sf_nextentry(sfp, sfep);
+		memcpy(sfep, oldsfep, old_isize - nbytes);
+	}
+	kmem_free(buf);
+	dp->i_d.di_size = new_isize;
+	xfs_dir2_sf_check(args);
+}
+
+/*
+ * Decide if the new entry will fit at all.
+ * If it will fit, pick between adding the new entry to the end (easy)
+ * or somewhere else (hard).
+ * Return 0 (won't fit), 1 (easy), 2 (hard).
+ */
+/*ARGSUSED*/
+static int					/* pick result */
+xfs_dir2_sf_addname_pick(
+	xfs_da_args_t		*args,		/* operation arguments */
+	int			objchange,	/* inode # size changes */
+	xfs_dir2_sf_entry_t	**sfepp,	/* out(1): new entry ptr */
+	xfs_dir2_data_aoff_t	*offsetp)	/* out(1): new offset */
+{
+	xfs_inode_t		*dp;		/* incore directory inode */
+	int			holefit;	/* found hole it will fit in */
+	int			i;		/* entry number */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+	xfs_dir2_data_aoff_t	offset;		/* data block offset */
+	xfs_dir2_sf_entry_t	*sfep;		/* shortform entry */
+	xfs_dir2_sf_t		*sfp;		/* shortform structure */
+	int			size;		/* entry's data size */
+	int			used;		/* data bytes used */
+
+	dp = args->dp;
+	mp = dp->i_mount;
+
+	sfp = (xfs_dir2_sf_t *)dp->i_df.if_u1.if_data;
+	size = xfs_dir2_data_entsize(args->namelen);
+	offset = XFS_DIR2_DATA_FIRST_OFFSET;
+	sfep = xfs_dir2_sf_firstentry(sfp);
+	holefit = 0;
+	/*
+	 * Loop over sf entries.
+	 * Keep track of data offset and whether we've seen a place
+	 * to insert the new entry.
+	 */
+	for (i = 0; i < sfp->hdr.count; i++) {
+		if (!holefit)
+			holefit = offset + size <= xfs_dir2_sf_get_offset(sfep);
+		offset = xfs_dir2_sf_get_offset(sfep) +
+			 xfs_dir2_data_entsize(sfep->namelen);
+		sfep = xfs_dir2_sf_nextentry(sfp, sfep);
+	}
+	/*
+	 * Calculate data bytes used excluding the new entry, if this
+	 * was a data block (block form directory).
+	 */
+	used = offset +
+	       (sfp->hdr.count + 3) * (uint)sizeof(xfs_dir2_leaf_entry_t) +
+	       (uint)sizeof(xfs_dir2_block_tail_t);
+	/*
+	 * If it won't fit in a block form then we can't insert it,
+	 * we'll go back, convert to block, then try the insert and convert
+	 * to leaf.
+	 */
+	if (used + (holefit ? 0 : size) > mp->m_dirblksize)
+		return 0;
+	/*
+	 * If changing the inode number size, do it the hard way.
+	 */
+#if XFS_BIG_INUMS
+	if (objchange) {
+		return 2;
+	}
+#else
+	ASSERT(objchange == 0);
+#endif
+	/*
+	 * If it won't fit at the end then do it the hard way (use the hole).
+	 */
+	if (used + size > mp->m_dirblksize)
+		return 2;
+	/*
+	 * Do it the easy way.
+	 */
+	*sfepp = sfep;
+	*offsetp = offset;
+	return 1;
+}
+
+#ifdef DEBUG
+/*
+ * Check consistency of shortform directory, assert if bad.
+ */
+static void
+xfs_dir2_sf_check(
+	xfs_da_args_t		*args)		/* operation arguments */
+{
+	xfs_inode_t		*dp;		/* incore directory inode */
+	int			i;		/* entry number */
+	int			i8count;	/* number of big inode#s */
+	xfs_ino_t		ino;		/* entry inode number */
+	int			offset;		/* data offset */
+	xfs_dir2_sf_entry_t	*sfep;		/* shortform dir entry */
+	xfs_dir2_sf_t		*sfp;		/* shortform structure */
+
+	dp = args->dp;
+
+	sfp = (xfs_dir2_sf_t *)dp->i_df.if_u1.if_data;
+	offset = XFS_DIR2_DATA_FIRST_OFFSET;
+	ino = xfs_dir2_sf_get_inumber(sfp, &sfp->hdr.parent);
+	i8count = ino > XFS_DIR2_MAX_SHORT_INUM;
+
+	for (i = 0, sfep = xfs_dir2_sf_firstentry(sfp);
+	     i < sfp->hdr.count;
+	     i++, sfep = xfs_dir2_sf_nextentry(sfp, sfep)) {
+		ASSERT(xfs_dir2_sf_get_offset(sfep) >= offset);
+		ino = xfs_dir2_sf_get_inumber(sfp, xfs_dir2_sf_inumberp(sfep));
+		i8count += ino > XFS_DIR2_MAX_SHORT_INUM;
+		offset =
+			xfs_dir2_sf_get_offset(sfep) +
+			xfs_dir2_data_entsize(sfep->namelen);
+	}
+	ASSERT(i8count == sfp->hdr.i8count);
+	ASSERT(XFS_BIG_INUMS || i8count == 0);
+	ASSERT((char *)sfep - (char *)sfp == dp->i_d.di_size);
+	ASSERT(offset +
+	       (sfp->hdr.count + 2) * (uint)sizeof(xfs_dir2_leaf_entry_t) +
+	       (uint)sizeof(xfs_dir2_block_tail_t) <=
+	       dp->i_mount->m_dirblksize);
+}
+#endif	/* DEBUG */
+
+/*
+ * Create a new (shortform) directory.
+ */
+int					/* error, always 0 */
+xfs_dir2_sf_create(
+	xfs_da_args_t	*args,		/* operation arguments */
+	xfs_ino_t	pino)		/* parent inode number */
+{
+	xfs_inode_t	*dp;		/* incore directory inode */
+	int		i8count;	/* parent inode is an 8-byte number */
+	xfs_dir2_sf_t	*sfp;		/* shortform structure */
+	int		size;		/* directory size */
+
+	trace_xfs_dir2_sf_create(args);
+
+	dp = args->dp;
+
+	ASSERT(dp != NULL);
+	ASSERT(dp->i_d.di_size == 0);
+	/*
+	 * If it's currently a zero-length extent file,
+	 * convert it to local format.
+	 */
+	if (dp->i_d.di_format == XFS_DINODE_FMT_EXTENTS) {
+		dp->i_df.if_flags &= ~XFS_IFEXTENTS;	/* just in case */
+		dp->i_d.di_format = XFS_DINODE_FMT_LOCAL;
+		xfs_trans_log_inode(args->trans, dp, XFS_ILOG_CORE);
+		dp->i_df.if_flags |= XFS_IFINLINE;
+	}
+	ASSERT(dp->i_df.if_flags & XFS_IFINLINE);
+	ASSERT(dp->i_df.if_bytes == 0);
+	i8count = pino > XFS_DIR2_MAX_SHORT_INUM;
+	size = xfs_dir2_sf_hdr_size(i8count);
+	/*
+	 * Make a buffer for the data.
+	 */
+	xfs_idata_realloc(dp, size, XFS_DATA_FORK);
+	/*
+	 * Fill in the header,
+	 */
+	sfp = (xfs_dir2_sf_t *)dp->i_df.if_u1.if_data;
+	sfp->hdr.i8count = i8count;
+	/*
+	 * Now can put in the inode number, since i8count is set.
+	 */
+	xfs_dir2_sf_put_inumber(sfp, &pino, &sfp->hdr.parent);
+	sfp->hdr.count = 0;
+	dp->i_d.di_size = size;
+	xfs_dir2_sf_check(args);
+	xfs_trans_log_inode(args->trans, dp, XFS_ILOG_CORE | XFS_ILOG_DDATA);
+	return 0;
+}
+
+int						/* error */
+xfs_dir2_sf_getdents(
+	xfs_inode_t		*dp,		/* incore directory inode */
+	void			*dirent,
+	xfs_off_t		*offset,
+	filldir_t		filldir)
+{
+	int			i;		/* shortform entry number */
+	xfs_mount_t		*mp;		/* filesystem mount point */
+	xfs_dir2_dataptr_t	off;		/* current entry's offset */
+	xfs_dir2_sf_entry_t	*sfep;		/* shortform directory entry */
+	xfs_dir2_sf_t		*sfp;		/* shortform structure */
+	xfs_dir2_dataptr_t	dot_offset;
+	xfs_dir2_dataptr_t	dotdot_offset;
+	xfs_ino_t		ino;
+
+	mp = dp->i_mount;
+
+	ASSERT(dp->i_df.if_flags & XFS_IFINLINE);
+	/*
+	 * Give up if the directory is way too short.
+	 */
+	if (dp->i_d.di_size < offsetof(xfs_dir2_sf_hdr_t, parent)) {
+		ASSERT(XFS_FORCED_SHUTDOWN(mp));
+		return XFS_ERROR(EIO);
+	}
+
+	ASSERT(dp->i_df.if_bytes == dp->i_d.di_size);
+	ASSERT(dp->i_df.if_u1.if_data != NULL);
+
+	sfp = (xfs_dir2_sf_t *)dp->i_df.if_u1.if_data;
+
+	ASSERT(dp->i_d.di_size >= xfs_dir2_sf_hdr_size(sfp->hdr.i8count));
+
+	/*
+	 * If the block number in the offset is out of range, we're done.
+	 */
+	if (xfs_dir2_dataptr_to_db(mp, *offset) > mp->m_dirdatablk)
+		return 0;
+
+	/*
+	 * Precalculate offsets for . and .. as we will always need them.
+	 *
+	 * XXX(hch): the second argument is sometimes 0 and sometimes
+	 * mp->m_dirdatablk.
+	 */
+	dot_offset = xfs_dir2_db_off_to_dataptr(mp, mp->m_dirdatablk,
+					     XFS_DIR2_DATA_DOT_OFFSET);
+	dotdot_offset = xfs_dir2_db_off_to_dataptr(mp, mp->m_dirdatablk,
+						XFS_DIR2_DATA_DOTDOT_OFFSET);
+
+	/*
+	 * Put . entry unless we're starting past it.
+	 */
+	if (*offset <= dot_offset) {
+		if (filldir(dirent, ".", 1, dot_offset & 0x7fffffff, dp->i_ino, DT_DIR)) {
+			*offset = dot_offset & 0x7fffffff;
+			return 0;
+		}
+	}
+
+	/*
+	 * Put .. entry unless we're starting past it.
+	 */
+	if (*offset <= dotdot_offset) {
+		ino = xfs_dir2_sf_get_inumber(sfp, &sfp->hdr.parent);
+		if (filldir(dirent, "..", 2, dotdot_offset & 0x7fffffff, ino, DT_DIR)) {
+			*offset = dotdot_offset & 0x7fffffff;
+			return 0;
+		}
+	}
+
+	/*
+	 * Loop while there are more entries and put'ing works.
+	 */
+	sfep = xfs_dir2_sf_firstentry(sfp);
+	for (i = 0; i < sfp->hdr.count; i++) {
+		off = xfs_dir2_db_off_to_dataptr(mp, mp->m_dirdatablk,
+				xfs_dir2_sf_get_offset(sfep));
+
+		if (*offset > off) {
+			sfep = xfs_dir2_sf_nextentry(sfp, sfep);
+			continue;
+		}
+
+		ino = xfs_dir2_sf_get_inumber(sfp, xfs_dir2_sf_inumberp(sfep));
+		if (filldir(dirent, (char *)sfep->name, sfep->namelen,
+			    off & 0x7fffffff, ino, DT_UNKNOWN)) {
+			*offset = off & 0x7fffffff;
+			return 0;
+		}
+		sfep = xfs_dir2_sf_nextentry(sfp, sfep);
+	}
+
+	*offset = xfs_dir2_db_off_to_dataptr(mp, mp->m_dirdatablk + 1, 0) &
+			0x7fffffff;
+	return 0;
+}
+
+/*
+ * Lookup an entry in a shortform directory.
+ * Returns EEXIST if found, ENOENT if not found.
+ */
+int						/* error */
+xfs_dir2_sf_lookup(
+	xfs_da_args_t		*args)		/* operation arguments */
+{
+	xfs_inode_t		*dp;		/* incore directory inode */
+	int			i;		/* entry index */
+	int			error;
+	xfs_dir2_sf_entry_t	*sfep;		/* shortform directory entry */
+	xfs_dir2_sf_t		*sfp;		/* shortform structure */
+	enum xfs_dacmp		cmp;		/* comparison result */
+	xfs_dir2_sf_entry_t	*ci_sfep;	/* case-insens. entry */
+
+	trace_xfs_dir2_sf_lookup(args);
+
+	xfs_dir2_sf_check(args);
+	dp = args->dp;
+
+	ASSERT(dp->i_df.if_flags & XFS_IFINLINE);
+	/*
+	 * Bail out if the directory is way too short.
+	 */
+	if (dp->i_d.di_size < offsetof(xfs_dir2_sf_hdr_t, parent)) {
+		ASSERT(XFS_FORCED_SHUTDOWN(dp->i_mount));
+		return XFS_ERROR(EIO);
+	}
+	ASSERT(dp->i_df.if_bytes == dp->i_d.di_size);
+	ASSERT(dp->i_df.if_u1.if_data != NULL);
+	sfp = (xfs_dir2_sf_t *)dp->i_df.if_u1.if_data;
+	ASSERT(dp->i_d.di_size >= xfs_dir2_sf_hdr_size(sfp->hdr.i8count));
+	/*
+	 * Special case for .
+	 */
+	if (args->namelen == 1 && args->name[0] == '.') {
+		args->inumber = dp->i_ino;
+		args->cmpresult = XFS_CMP_EXACT;
+		return XFS_ERROR(EEXIST);
+	}
+	/*
+	 * Special case for ..
+	 */
+	if (args->namelen == 2 &&
+	    args->name[0] == '.' && args->name[1] == '.') {
+		args->inumber = xfs_dir2_sf_get_inumber(sfp, &sfp->hdr.parent);
+		args->cmpresult = XFS_CMP_EXACT;
+		return XFS_ERROR(EEXIST);
+	}
+	/*
+	 * Loop over all the entries trying to match ours.
+	 */
+	ci_sfep = NULL;
+	for (i = 0, sfep = xfs_dir2_sf_firstentry(sfp); i < sfp->hdr.count;
+				i++, sfep = xfs_dir2_sf_nextentry(sfp, sfep)) {
+		/*
+		 * Compare name and if it's an exact match, return the inode
+		 * number. If it's the first case-insensitive match, store the
+		 * inode number and continue looking for an exact match.
+		 */
+		cmp = dp->i_mount->m_dirnameops->compname(args, sfep->name,
+								sfep->namelen);
+		if (cmp != XFS_CMP_DIFFERENT && cmp != args->cmpresult) {
+			args->cmpresult = cmp;
+			args->inumber = xfs_dir2_sf_get_inumber(sfp,
+						xfs_dir2_sf_inumberp(sfep));
+			if (cmp == XFS_CMP_EXACT)
+				return XFS_ERROR(EEXIST);
+			ci_sfep = sfep;
+		}
+	}
+	ASSERT(args->op_flags & XFS_DA_OP_OKNOENT);
+	/*
+	 * Here, we can only be doing a lookup (not a rename or replace).
+	 * If a case-insensitive match was not found, return ENOENT.
+	 */
+	if (!ci_sfep)
+		return XFS_ERROR(ENOENT);
+	/* otherwise process the CI match as required by the caller */
+	error = xfs_dir_cilookup_result(args, ci_sfep->name, ci_sfep->namelen);
+	return XFS_ERROR(error);
+}
+
+/*
+ * Remove an entry from a shortform directory.
+ */
+int						/* error */
+xfs_dir2_sf_removename(
+	xfs_da_args_t		*args)
+{
+	int			byteoff;	/* offset of removed entry */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	int			entsize;	/* this entry's size */
+	int			i;		/* shortform entry index */
+	int			newsize;	/* new inode size */
+	int			oldsize;	/* old inode size */
+	xfs_dir2_sf_entry_t	*sfep;		/* shortform directory entry */
+	xfs_dir2_sf_t		*sfp;		/* shortform structure */
+
+	trace_xfs_dir2_sf_removename(args);
+
+	dp = args->dp;
+
+	ASSERT(dp->i_df.if_flags & XFS_IFINLINE);
+	oldsize = (int)dp->i_d.di_size;
+	/*
+	 * Bail out if the directory is way too short.
+	 */
+	if (oldsize < offsetof(xfs_dir2_sf_hdr_t, parent)) {
+		ASSERT(XFS_FORCED_SHUTDOWN(dp->i_mount));
+		return XFS_ERROR(EIO);
+	}
+	ASSERT(dp->i_df.if_bytes == oldsize);
+	ASSERT(dp->i_df.if_u1.if_data != NULL);
+	sfp = (xfs_dir2_sf_t *)dp->i_df.if_u1.if_data;
+	ASSERT(oldsize >= xfs_dir2_sf_hdr_size(sfp->hdr.i8count));
+	/*
+	 * Loop over the old directory entries.
+	 * Find the one we're deleting.
+	 */
+	for (i = 0, sfep = xfs_dir2_sf_firstentry(sfp); i < sfp->hdr.count;
+				i++, sfep = xfs_dir2_sf_nextentry(sfp, sfep)) {
+		if (xfs_da_compname(args, sfep->name, sfep->namelen) ==
+								XFS_CMP_EXACT) {
+			ASSERT(xfs_dir2_sf_get_inumber(sfp,
+						xfs_dir2_sf_inumberp(sfep)) ==
+								args->inumber);
+			break;
+		}
+	}
+	/*
+	 * Didn't find it.
+	 */
+	if (i == sfp->hdr.count)
+		return XFS_ERROR(ENOENT);
+	/*
+	 * Calculate sizes.
+	 */
+	byteoff = (int)((char *)sfep - (char *)sfp);
+	entsize = xfs_dir2_sf_entsize_byname(sfp, args->namelen);
+	newsize = oldsize - entsize;
+	/*
+	 * Copy the part if any after the removed entry, sliding it down.
+	 */
+	if (byteoff + entsize < oldsize)
+		memmove((char *)sfp + byteoff, (char *)sfp + byteoff + entsize,
+			oldsize - (byteoff + entsize));
+	/*
+	 * Fix up the header and file size.
+	 */
+	sfp->hdr.count--;
+	dp->i_d.di_size = newsize;
+	/*
+	 * Reallocate, making it smaller.
+	 */
+	xfs_idata_realloc(dp, newsize - oldsize, XFS_DATA_FORK);
+	sfp = (xfs_dir2_sf_t *)dp->i_df.if_u1.if_data;
+#if XFS_BIG_INUMS
+	/*
+	 * Are we changing inode number size?
+	 */
+	if (args->inumber > XFS_DIR2_MAX_SHORT_INUM) {
+		if (sfp->hdr.i8count == 1)
+			xfs_dir2_sf_toino4(args);
+		else
+			sfp->hdr.i8count--;
+	}
+#endif
+	xfs_dir2_sf_check(args);
+	xfs_trans_log_inode(args->trans, dp, XFS_ILOG_CORE | XFS_ILOG_DDATA);
+	return 0;
+}
+
+/*
+ * Replace the inode number of an entry in a shortform directory.
+ */
+int						/* error */
+xfs_dir2_sf_replace(
+	xfs_da_args_t		*args)		/* operation arguments */
+{
+	xfs_inode_t		*dp;		/* incore directory inode */
+	int			i;		/* entry index */
+#if XFS_BIG_INUMS || defined(DEBUG)
+	xfs_ino_t		ino=0;		/* entry old inode number */
+#endif
+#if XFS_BIG_INUMS
+	int			i8elevated;	/* sf_toino8 set i8count=1 */
+#endif
+	xfs_dir2_sf_entry_t	*sfep;		/* shortform directory entry */
+	xfs_dir2_sf_t		*sfp;		/* shortform structure */
+
+	trace_xfs_dir2_sf_replace(args);
+
+	dp = args->dp;
+
+	ASSERT(dp->i_df.if_flags & XFS_IFINLINE);
+	/*
+	 * Bail out if the shortform directory is way too small.
+	 */
+	if (dp->i_d.di_size < offsetof(xfs_dir2_sf_hdr_t, parent)) {
+		ASSERT(XFS_FORCED_SHUTDOWN(dp->i_mount));
+		return XFS_ERROR(EIO);
+	}
+	ASSERT(dp->i_df.if_bytes == dp->i_d.di_size);
+	ASSERT(dp->i_df.if_u1.if_data != NULL);
+	sfp = (xfs_dir2_sf_t *)dp->i_df.if_u1.if_data;
+	ASSERT(dp->i_d.di_size >= xfs_dir2_sf_hdr_size(sfp->hdr.i8count));
+#if XFS_BIG_INUMS
+	/*
+	 * New inode number is large, and need to convert to 8-byte inodes.
+	 */
+	if (args->inumber > XFS_DIR2_MAX_SHORT_INUM && sfp->hdr.i8count == 0) {
+		int	error;			/* error return value */
+		int	newsize;		/* new inode size */
+
+		newsize =
+			dp->i_df.if_bytes +
+			(sfp->hdr.count + 1) *
+			((uint)sizeof(xfs_dir2_ino8_t) -
+			 (uint)sizeof(xfs_dir2_ino4_t));
+		/*
+		 * Won't fit as shortform, convert to block then do replace.
+		 */
+		if (newsize > XFS_IFORK_DSIZE(dp)) {
+			error = xfs_dir2_sf_to_block(args);
+			if (error) {
+				return error;
+			}
+			return xfs_dir2_block_replace(args);
+		}
+		/*
+		 * Still fits, convert to 8-byte now.
+		 */
+		xfs_dir2_sf_toino8(args);
+		i8elevated = 1;
+		sfp = (xfs_dir2_sf_t *)dp->i_df.if_u1.if_data;
+	} else
+		i8elevated = 0;
+#endif
+	ASSERT(args->namelen != 1 || args->name[0] != '.');
+	/*
+	 * Replace ..'s entry.
+	 */
+	if (args->namelen == 2 &&
+	    args->name[0] == '.' && args->name[1] == '.') {
+#if XFS_BIG_INUMS || defined(DEBUG)
+		ino = xfs_dir2_sf_get_inumber(sfp, &sfp->hdr.parent);
+		ASSERT(args->inumber != ino);
+#endif
+		xfs_dir2_sf_put_inumber(sfp, &args->inumber, &sfp->hdr.parent);
+	}
+	/*
+	 * Normal entry, look for the name.
+	 */
+	else {
+		for (i = 0, sfep = xfs_dir2_sf_firstentry(sfp);
+				i < sfp->hdr.count;
+				i++, sfep = xfs_dir2_sf_nextentry(sfp, sfep)) {
+			if (xfs_da_compname(args, sfep->name, sfep->namelen) ==
+								XFS_CMP_EXACT) {
+#if XFS_BIG_INUMS || defined(DEBUG)
+				ino = xfs_dir2_sf_get_inumber(sfp,
+					xfs_dir2_sf_inumberp(sfep));
+				ASSERT(args->inumber != ino);
+#endif
+				xfs_dir2_sf_put_inumber(sfp, &args->inumber,
+					xfs_dir2_sf_inumberp(sfep));
+				break;
+			}
+		}
+		/*
+		 * Didn't find it.
+		 */
+		if (i == sfp->hdr.count) {
+			ASSERT(args->op_flags & XFS_DA_OP_OKNOENT);
+#if XFS_BIG_INUMS
+			if (i8elevated)
+				xfs_dir2_sf_toino4(args);
+#endif
+			return XFS_ERROR(ENOENT);
+		}
+	}
+#if XFS_BIG_INUMS
+	/*
+	 * See if the old number was large, the new number is small.
+	 */
+	if (ino > XFS_DIR2_MAX_SHORT_INUM &&
+	    args->inumber <= XFS_DIR2_MAX_SHORT_INUM) {
+		/*
+		 * And the old count was one, so need to convert to small.
+		 */
+		if (sfp->hdr.i8count == 1)
+			xfs_dir2_sf_toino4(args);
+		else
+			sfp->hdr.i8count--;
+	}
+	/*
+	 * See if the old number was small, the new number is large.
+	 */
+	if (ino <= XFS_DIR2_MAX_SHORT_INUM &&
+	    args->inumber > XFS_DIR2_MAX_SHORT_INUM) {
+		/*
+		 * add to the i8count unless we just converted to 8-byte
+		 * inodes (which does an implied i8count = 1)
+		 */
+		ASSERT(sfp->hdr.i8count != 0);
+		if (!i8elevated)
+			sfp->hdr.i8count++;
+	}
+#endif
+	xfs_dir2_sf_check(args);
+	xfs_trans_log_inode(args->trans, dp, XFS_ILOG_DDATA);
+	return 0;
+}
+
+#if XFS_BIG_INUMS
+/*
+ * Convert from 8-byte inode numbers to 4-byte inode numbers.
+ * The last 8-byte inode number is gone, but the count is still 1.
+ */
+static void
+xfs_dir2_sf_toino4(
+	xfs_da_args_t		*args)		/* operation arguments */
+{
+	char			*buf;		/* old dir's buffer */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	int			i;		/* entry index */
+	xfs_ino_t		ino;		/* entry inode number */
+	int			newsize;	/* new inode size */
+	xfs_dir2_sf_entry_t	*oldsfep;	/* old sf entry */
+	xfs_dir2_sf_t		*oldsfp;	/* old sf directory */
+	int			oldsize;	/* old inode size */
+	xfs_dir2_sf_entry_t	*sfep;		/* new sf entry */
+	xfs_dir2_sf_t		*sfp;		/* new sf directory */
+
+	trace_xfs_dir2_sf_toino4(args);
+
+	dp = args->dp;
+
+	/*
+	 * Copy the old directory to the buffer.
+	 * Then nuke it from the inode, and add the new buffer to the inode.
+	 * Don't want xfs_idata_realloc copying the data here.
+	 */
+	oldsize = dp->i_df.if_bytes;
+	buf = kmem_alloc(oldsize, KM_SLEEP);
+	oldsfp = (xfs_dir2_sf_t *)dp->i_df.if_u1.if_data;
+	ASSERT(oldsfp->hdr.i8count == 1);
+	memcpy(buf, oldsfp, oldsize);
+	/*
+	 * Compute the new inode size.
+	 */
+	newsize =
+		oldsize -
+		(oldsfp->hdr.count + 1) *
+		((uint)sizeof(xfs_dir2_ino8_t) - (uint)sizeof(xfs_dir2_ino4_t));
+	xfs_idata_realloc(dp, -oldsize, XFS_DATA_FORK);
+	xfs_idata_realloc(dp, newsize, XFS_DATA_FORK);
+	/*
+	 * Reset our pointers, the data has moved.
+	 */
+	oldsfp = (xfs_dir2_sf_t *)buf;
+	sfp = (xfs_dir2_sf_t *)dp->i_df.if_u1.if_data;
+	/*
+	 * Fill in the new header.
+	 */
+	sfp->hdr.count = oldsfp->hdr.count;
+	sfp->hdr.i8count = 0;
+	ino = xfs_dir2_sf_get_inumber(oldsfp, &oldsfp->hdr.parent);
+	xfs_dir2_sf_put_inumber(sfp, &ino, &sfp->hdr.parent);
+	/*
+	 * Copy the entries field by field.
+	 */
+	for (i = 0, sfep = xfs_dir2_sf_firstentry(sfp),
+		    oldsfep = xfs_dir2_sf_firstentry(oldsfp);
+	     i < sfp->hdr.count;
+	     i++, sfep = xfs_dir2_sf_nextentry(sfp, sfep),
+		  oldsfep = xfs_dir2_sf_nextentry(oldsfp, oldsfep)) {
+		sfep->namelen = oldsfep->namelen;
+		sfep->offset = oldsfep->offset;
+		memcpy(sfep->name, oldsfep->name, sfep->namelen);
+		ino = xfs_dir2_sf_get_inumber(oldsfp,
+			xfs_dir2_sf_inumberp(oldsfep));
+		xfs_dir2_sf_put_inumber(sfp, &ino, xfs_dir2_sf_inumberp(sfep));
+	}
+	/*
+	 * Clean up the inode.
+	 */
+	kmem_free(buf);
+	dp->i_d.di_size = newsize;
+	xfs_trans_log_inode(args->trans, dp, XFS_ILOG_CORE | XFS_ILOG_DDATA);
+}
+
+/*
+ * Convert from 4-byte inode numbers to 8-byte inode numbers.
+ * The new 8-byte inode number is not there yet, we leave with the
+ * count 1 but no corresponding entry.
+ */
+static void
+xfs_dir2_sf_toino8(
+	xfs_da_args_t		*args)		/* operation arguments */
+{
+	char			*buf;		/* old dir's buffer */
+	xfs_inode_t		*dp;		/* incore directory inode */
+	int			i;		/* entry index */
+	xfs_ino_t		ino;		/* entry inode number */
+	int			newsize;	/* new inode size */
+	xfs_dir2_sf_entry_t	*oldsfep;	/* old sf entry */
+	xfs_dir2_sf_t		*oldsfp;	/* old sf directory */
+	int			oldsize;	/* old inode size */
+	xfs_dir2_sf_entry_t	*sfep;		/* new sf entry */
+	xfs_dir2_sf_t		*sfp;		/* new sf directory */
+
+	trace_xfs_dir2_sf_toino8(args);
+
+	dp = args->dp;
+
+	/*
+	 * Copy the old directory to the buffer.
+	 * Then nuke it from the inode, and add the new buffer to the inode.
+	 * Don't want xfs_idata_realloc copying the data here.
+	 */
+	oldsize = dp->i_df.if_bytes;
+	buf = kmem_alloc(oldsize, KM_SLEEP);
+	oldsfp = (xfs_dir2_sf_t *)dp->i_df.if_u1.if_data;
+	ASSERT(oldsfp->hdr.i8count == 0);
+	memcpy(buf, oldsfp, oldsize);
+	/*
+	 * Compute the new inode size.
+	 */
+	newsize =
+		oldsize +
+		(oldsfp->hdr.count + 1) *
+		((uint)sizeof(xfs_dir2_ino8_t) - (uint)sizeof(xfs_dir2_ino4_t));
+	xfs_idata_realloc(dp, -oldsize, XFS_DATA_FORK);
+	xfs_idata_realloc(dp, newsize, XFS_DATA_FORK);
+	/*
+	 * Reset our pointers, the data has moved.
+	 */
+	oldsfp = (xfs_dir2_sf_t *)buf;
+	sfp = (xfs_dir2_sf_t *)dp->i_df.if_u1.if_data;
+	/*
+	 * Fill in the new header.
+	 */
+	sfp->hdr.count = oldsfp->hdr.count;
+	sfp->hdr.i8count = 1;
+	ino = xfs_dir2_sf_get_inumber(oldsfp, &oldsfp->hdr.parent);
+	xfs_dir2_sf_put_inumber(sfp, &ino, &sfp->hdr.parent);
+	/*
+	 * Copy the entries field by field.
+	 */
+	for (i = 0, sfep = xfs_dir2_sf_firstentry(sfp),
+		    oldsfep = xfs_dir2_sf_firstentry(oldsfp);
+	     i < sfp->hdr.count;
+	     i++, sfep = xfs_dir2_sf_nextentry(sfp, sfep),
+		  oldsfep = xfs_dir2_sf_nextentry(oldsfp, oldsfep)) {
+		sfep->namelen = oldsfep->namelen;
+		sfep->offset = oldsfep->offset;
+		memcpy(sfep->name, oldsfep->name, sfep->namelen);
+		ino = xfs_dir2_sf_get_inumber(oldsfp,
+			xfs_dir2_sf_inumberp(oldsfep));
+		xfs_dir2_sf_put_inumber(sfp, &ino, xfs_dir2_sf_inumberp(sfep));
+	}
+	/*
+	 * Clean up the inode.
+	 */
+	kmem_free(buf);
+	dp->i_d.di_size = newsize;
+	xfs_trans_log_inode(args->trans, dp, XFS_ILOG_CORE | XFS_ILOG_DDATA);
+}
+#endif	/* XFS_BIG_INUMS */
diff --git a/fs/xfs/xfs_dir2_sf.h b/fs/xfs/xfs_dir2_sf.h
new file mode 100644
index 00000000..6ac44b55
--- /dev/null
+++ b/fs/xfs/xfs_dir2_sf.h
@@ -0,0 +1,171 @@
+/*
+ * Copyright (c) 2000-2001,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_DIR2_SF_H__
+#define	__XFS_DIR2_SF_H__
+
+/*
+ * Directory layout when stored internal to an inode.
+ *
+ * Small directories are packed as tightly as possible so as to
+ * fit into the literal area of the inode.
+ */
+
+struct uio;
+struct xfs_dabuf;
+struct xfs_da_args;
+struct xfs_dir2_block;
+struct xfs_inode;
+struct xfs_mount;
+struct xfs_trans;
+
+/*
+ * Inode number stored as 8 8-bit values.
+ */
+typedef	struct { __uint8_t i[8]; } xfs_dir2_ino8_t;
+
+/*
+ * Inode number stored as 4 8-bit values.
+ * Works a lot of the time, when all the inode numbers in a directory
+ * fit in 32 bits.
+ */
+typedef struct { __uint8_t i[4]; } xfs_dir2_ino4_t;
+
+typedef union {
+	xfs_dir2_ino8_t	i8;
+	xfs_dir2_ino4_t	i4;
+} xfs_dir2_inou_t;
+#define	XFS_DIR2_MAX_SHORT_INUM	((xfs_ino_t)0xffffffffULL)
+
+/*
+ * Normalized offset (in a data block) of the entry, really xfs_dir2_data_off_t.
+ * Only need 16 bits, this is the byte offset into the single block form.
+ */
+typedef struct { __uint8_t i[2]; } __arch_pack xfs_dir2_sf_off_t;
+
+/*
+ * The parent directory has a dedicated field, and the self-pointer must
+ * be calculated on the fly.
+ *
+ * Entries are packed toward the top as tightly as possible.  The header
+ * and the elements must be memcpy'd out into a work area to get correct
+ * alignment for the inode number fields.
+ */
+typedef struct xfs_dir2_sf_hdr {
+	__uint8_t		count;		/* count of entries */
+	__uint8_t		i8count;	/* count of 8-byte inode #s */
+	xfs_dir2_inou_t		parent;		/* parent dir inode number */
+} __arch_pack xfs_dir2_sf_hdr_t;
+
+typedef struct xfs_dir2_sf_entry {
+	__uint8_t		namelen;	/* actual name length */
+	xfs_dir2_sf_off_t	offset;		/* saved offset */
+	__uint8_t		name[1];	/* name, variable size */
+	xfs_dir2_inou_t		inumber;	/* inode number, var. offset */
+} __arch_pack xfs_dir2_sf_entry_t; 
+
+typedef struct xfs_dir2_sf {
+	xfs_dir2_sf_hdr_t	hdr;		/* shortform header */
+	xfs_dir2_sf_entry_t	list[1];	/* shortform entries */
+} xfs_dir2_sf_t;
+
+static inline int xfs_dir2_sf_hdr_size(int i8count)
+{
+	return ((uint)sizeof(xfs_dir2_sf_hdr_t) - \
+		((i8count) == 0) * \
+		((uint)sizeof(xfs_dir2_ino8_t) - (uint)sizeof(xfs_dir2_ino4_t)));
+}
+
+static inline xfs_dir2_inou_t *xfs_dir2_sf_inumberp(xfs_dir2_sf_entry_t *sfep)
+{
+	return (xfs_dir2_inou_t *)&(sfep)->name[(sfep)->namelen];
+}
+
+static inline xfs_intino_t
+xfs_dir2_sf_get_inumber(xfs_dir2_sf_t *sfp, xfs_dir2_inou_t *from)
+{
+	return ((sfp)->hdr.i8count == 0 ? \
+		(xfs_intino_t)XFS_GET_DIR_INO4((from)->i4) : \
+		(xfs_intino_t)XFS_GET_DIR_INO8((from)->i8));
+}
+
+static inline void xfs_dir2_sf_put_inumber(xfs_dir2_sf_t *sfp, xfs_ino_t *from,
+						xfs_dir2_inou_t *to)
+{
+	if ((sfp)->hdr.i8count == 0)
+		XFS_PUT_DIR_INO4(*(from), (to)->i4);
+	else
+		XFS_PUT_DIR_INO8(*(from), (to)->i8);
+}
+
+static inline xfs_dir2_data_aoff_t
+xfs_dir2_sf_get_offset(xfs_dir2_sf_entry_t *sfep)
+{
+	return INT_GET_UNALIGNED_16_BE(&(sfep)->offset.i);
+}
+
+static inline void
+xfs_dir2_sf_put_offset(xfs_dir2_sf_entry_t *sfep, xfs_dir2_data_aoff_t off)
+{
+	INT_SET_UNALIGNED_16_BE(&(sfep)->offset.i, off);
+}
+
+static inline int xfs_dir2_sf_entsize_byname(xfs_dir2_sf_t *sfp, int len)
+{
+	return ((uint)sizeof(xfs_dir2_sf_entry_t) - 1 + (len) - \
+		((sfp)->hdr.i8count == 0) * \
+		((uint)sizeof(xfs_dir2_ino8_t) - (uint)sizeof(xfs_dir2_ino4_t)));
+}
+
+static inline int
+xfs_dir2_sf_entsize_byentry(xfs_dir2_sf_t *sfp, xfs_dir2_sf_entry_t *sfep)
+{
+	return ((uint)sizeof(xfs_dir2_sf_entry_t) - 1 + (sfep)->namelen - \
+		((sfp)->hdr.i8count == 0) * \
+		((uint)sizeof(xfs_dir2_ino8_t) - (uint)sizeof(xfs_dir2_ino4_t)));
+}
+
+static inline xfs_dir2_sf_entry_t *xfs_dir2_sf_firstentry(xfs_dir2_sf_t *sfp)
+{
+	return ((xfs_dir2_sf_entry_t *) \
+		((char *)(sfp) + xfs_dir2_sf_hdr_size(sfp->hdr.i8count)));
+}
+
+static inline xfs_dir2_sf_entry_t *
+xfs_dir2_sf_nextentry(xfs_dir2_sf_t *sfp, xfs_dir2_sf_entry_t *sfep)
+{
+	return ((xfs_dir2_sf_entry_t *) \
+		((char *)(sfep) + xfs_dir2_sf_entsize_byentry(sfp,sfep)));
+}
+
+/*
+ * Functions.
+ */
+extern int xfs_dir2_block_sfsize(struct xfs_inode *dp,
+				 struct xfs_dir2_block *block,
+				 xfs_dir2_sf_hdr_t *sfhp);
+extern int xfs_dir2_block_to_sf(struct xfs_da_args *args, struct xfs_dabuf *bp,
+				int size, xfs_dir2_sf_hdr_t *sfhp);
+extern int xfs_dir2_sf_addname(struct xfs_da_args *args);
+extern int xfs_dir2_sf_create(struct xfs_da_args *args, xfs_ino_t pino);
+extern int xfs_dir2_sf_getdents(struct xfs_inode *dp, void *dirent,
+				xfs_off_t *offset, filldir_t filldir);
+extern int xfs_dir2_sf_lookup(struct xfs_da_args *args);
+extern int xfs_dir2_sf_removename(struct xfs_da_args *args);
+extern int xfs_dir2_sf_replace(struct xfs_da_args *args);
+
+#endif	/* __XFS_DIR2_SF_H__ */
diff --git a/fs/xfs/xfs_error.c b/fs/xfs/xfs_error.c
new file mode 100644
index 00000000..39f06336
--- /dev/null
+++ b/fs/xfs/xfs_error.c
@@ -0,0 +1,185 @@
+/*
+ * Copyright (c) 2000-2001,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_mount.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_utils.h"
+#include "xfs_error.h"
+
+#ifdef DEBUG
+
+int	xfs_etrap[XFS_ERROR_NTRAP] = {
+	0,
+};
+
+int
+xfs_error_trap(int e)
+{
+	int i;
+
+	if (!e)
+		return 0;
+	for (i = 0; i < XFS_ERROR_NTRAP; i++) {
+		if (xfs_etrap[i] == 0)
+			break;
+		if (e != xfs_etrap[i])
+			continue;
+		xfs_notice(NULL, "%s: error %d", __func__, e);
+		BUG();
+		break;
+	}
+	return e;
+}
+
+int	xfs_etest[XFS_NUM_INJECT_ERROR];
+int64_t	xfs_etest_fsid[XFS_NUM_INJECT_ERROR];
+char *	xfs_etest_fsname[XFS_NUM_INJECT_ERROR];
+int	xfs_error_test_active;
+
+int
+xfs_error_test(int error_tag, int *fsidp, char *expression,
+	       int line, char *file, unsigned long randfactor)
+{
+	int i;
+	int64_t fsid;
+
+	if (random32() % randfactor)
+		return 0;
+
+	memcpy(&fsid, fsidp, sizeof(xfs_fsid_t));
+
+	for (i = 0; i < XFS_NUM_INJECT_ERROR; i++)  {
+		if (xfs_etest[i] == error_tag && xfs_etest_fsid[i] == fsid) {
+			xfs_warn(NULL,
+	"Injecting error (%s) at file %s, line %d, on filesystem \"%s\"",
+				expression, file, line, xfs_etest_fsname[i]);
+			return 1;
+		}
+	}
+
+	return 0;
+}
+
+int
+xfs_errortag_add(int error_tag, xfs_mount_t *mp)
+{
+	int i;
+	int len;
+	int64_t fsid;
+
+	memcpy(&fsid, mp->m_fixedfsid, sizeof(xfs_fsid_t));
+
+	for (i = 0; i < XFS_NUM_INJECT_ERROR; i++)  {
+		if (xfs_etest_fsid[i] == fsid && xfs_etest[i] == error_tag) {
+			xfs_warn(mp, "error tag #%d on", error_tag);
+			return 0;
+		}
+	}
+
+	for (i = 0; i < XFS_NUM_INJECT_ERROR; i++)  {
+		if (xfs_etest[i] == 0) {
+			xfs_warn(mp, "Turned on XFS error tag #%d",
+				error_tag);
+			xfs_etest[i] = error_tag;
+			xfs_etest_fsid[i] = fsid;
+			len = strlen(mp->m_fsname);
+			xfs_etest_fsname[i] = kmem_alloc(len + 1, KM_SLEEP);
+			strcpy(xfs_etest_fsname[i], mp->m_fsname);
+			xfs_error_test_active++;
+			return 0;
+		}
+	}
+
+	xfs_warn(mp, "error tag overflow, too many turned on");
+
+	return 1;
+}
+
+int
+xfs_errortag_clearall(xfs_mount_t *mp, int loud)
+{
+	int64_t fsid;
+	int cleared = 0;
+	int i;
+
+	memcpy(&fsid, mp->m_fixedfsid, sizeof(xfs_fsid_t));
+
+
+	for (i = 0; i < XFS_NUM_INJECT_ERROR; i++) {
+		if ((fsid == 0LL || xfs_etest_fsid[i] == fsid) &&
+		     xfs_etest[i] != 0) {
+			cleared = 1;
+			xfs_warn(mp, "Clearing XFS error tag #%d",
+				xfs_etest[i]);
+			xfs_etest[i] = 0;
+			xfs_etest_fsid[i] = 0LL;
+			kmem_free(xfs_etest_fsname[i]);
+			xfs_etest_fsname[i] = NULL;
+			xfs_error_test_active--;
+		}
+	}
+
+	if (loud || cleared)
+		xfs_warn(mp, "Cleared all XFS error tags for filesystem");
+
+	return 0;
+}
+#endif /* DEBUG */
+
+void
+xfs_error_report(
+	const char		*tag,
+	int			level,
+	struct xfs_mount	*mp,
+	const char		*filename,
+	int			linenum,
+	inst_t			*ra)
+{
+	if (level <= xfs_error_level) {
+		xfs_alert_tag(mp, XFS_PTAG_ERROR_REPORT,
+		"Internal error %s at line %d of file %s.  Caller 0x%p\n",
+			    tag, linenum, filename, ra);
+
+		xfs_stack_trace();
+	}
+}
+
+void
+xfs_corruption_error(
+	const char		*tag,
+	int			level,
+	struct xfs_mount	*mp,
+	void			*p,
+	const char		*filename,
+	int			linenum,
+	inst_t			*ra)
+{
+	if (level <= xfs_error_level)
+		xfs_hex_dump(p, 16);
+	xfs_error_report(tag, level, mp, filename, linenum, ra);
+	xfs_alert(mp, "Corruption detected. Unmount and run xfs_repair");
+}
diff --git a/fs/xfs/xfs_error.h b/fs/xfs/xfs_error.h
new file mode 100644
index 00000000..079a367f
--- /dev/null
+++ b/fs/xfs/xfs_error.h
@@ -0,0 +1,161 @@
+/*
+ * Copyright (c) 2000-2002,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef	__XFS_ERROR_H__
+#define	__XFS_ERROR_H__
+
+#ifdef DEBUG
+#define	XFS_ERROR_NTRAP	10
+extern int	xfs_etrap[XFS_ERROR_NTRAP];
+extern int	xfs_error_trap(int);
+#define	XFS_ERROR(e)	xfs_error_trap(e)
+#else
+#define	XFS_ERROR(e)	(e)
+#endif
+
+struct xfs_mount;
+
+extern void xfs_error_report(const char *tag, int level, struct xfs_mount *mp,
+			const char *filename, int linenum, inst_t *ra);
+extern void xfs_corruption_error(const char *tag, int level,
+			struct xfs_mount *mp, void *p, const char *filename,
+			int linenum, inst_t *ra);
+
+#define	XFS_ERROR_REPORT(e, lvl, mp)	\
+	xfs_error_report(e, lvl, mp, __FILE__, __LINE__, __return_address)
+#define	XFS_CORRUPTION_ERROR(e, lvl, mp, mem)	\
+	xfs_corruption_error(e, lvl, mp, mem, \
+			     __FILE__, __LINE__, __return_address)
+
+#define XFS_ERRLEVEL_OFF	0
+#define XFS_ERRLEVEL_LOW	1
+#define XFS_ERRLEVEL_HIGH	5
+
+/*
+ * Macros to set EFSCORRUPTED & return/branch.
+ */
+#define	XFS_WANT_CORRUPTED_GOTO(x,l)	\
+	{ \
+		int fs_is_ok = (x); \
+		ASSERT(fs_is_ok); \
+		if (unlikely(!fs_is_ok)) { \
+			XFS_ERROR_REPORT("XFS_WANT_CORRUPTED_GOTO", \
+					 XFS_ERRLEVEL_LOW, NULL); \
+			error = XFS_ERROR(EFSCORRUPTED); \
+			goto l; \
+		} \
+	}
+
+#define	XFS_WANT_CORRUPTED_RETURN(x)	\
+	{ \
+		int fs_is_ok = (x); \
+		ASSERT(fs_is_ok); \
+		if (unlikely(!fs_is_ok)) { \
+			XFS_ERROR_REPORT("XFS_WANT_CORRUPTED_RETURN", \
+					 XFS_ERRLEVEL_LOW, NULL); \
+			return XFS_ERROR(EFSCORRUPTED); \
+		} \
+	}
+
+/*
+ * error injection tags - the labels can be anything you want
+ * but each tag should have its own unique number
+ */
+
+#define XFS_ERRTAG_NOERROR				0
+#define XFS_ERRTAG_IFLUSH_1				1
+#define XFS_ERRTAG_IFLUSH_2				2
+#define XFS_ERRTAG_IFLUSH_3				3
+#define XFS_ERRTAG_IFLUSH_4				4
+#define XFS_ERRTAG_IFLUSH_5				5
+#define XFS_ERRTAG_IFLUSH_6				6
+#define	XFS_ERRTAG_DA_READ_BUF				7
+#define	XFS_ERRTAG_BTREE_CHECK_LBLOCK			8
+#define	XFS_ERRTAG_BTREE_CHECK_SBLOCK			9
+#define	XFS_ERRTAG_ALLOC_READ_AGF			10
+#define	XFS_ERRTAG_IALLOC_READ_AGI			11
+#define	XFS_ERRTAG_ITOBP_INOTOBP			12
+#define	XFS_ERRTAG_IUNLINK				13
+#define	XFS_ERRTAG_IUNLINK_REMOVE			14
+#define	XFS_ERRTAG_DIR_INO_VALIDATE			15
+#define XFS_ERRTAG_BULKSTAT_READ_CHUNK			16
+#define XFS_ERRTAG_IODONE_IOERR				17
+#define XFS_ERRTAG_STRATREAD_IOERR			18
+#define XFS_ERRTAG_STRATCMPL_IOERR			19
+#define XFS_ERRTAG_DIOWRITE_IOERR			20
+#define XFS_ERRTAG_BMAPIFORMAT				21
+#define XFS_ERRTAG_MAX					22
+
+/*
+ * Random factors for above tags, 1 means always, 2 means 1/2 time, etc.
+ */
+#define XFS_RANDOM_DEFAULT				100
+#define XFS_RANDOM_IFLUSH_1				XFS_RANDOM_DEFAULT
+#define XFS_RANDOM_IFLUSH_2				XFS_RANDOM_DEFAULT
+#define XFS_RANDOM_IFLUSH_3				XFS_RANDOM_DEFAULT
+#define XFS_RANDOM_IFLUSH_4				XFS_RANDOM_DEFAULT
+#define XFS_RANDOM_IFLUSH_5				XFS_RANDOM_DEFAULT
+#define XFS_RANDOM_IFLUSH_6				XFS_RANDOM_DEFAULT
+#define XFS_RANDOM_DA_READ_BUF				XFS_RANDOM_DEFAULT
+#define XFS_RANDOM_BTREE_CHECK_LBLOCK			(XFS_RANDOM_DEFAULT/4)
+#define XFS_RANDOM_BTREE_CHECK_SBLOCK			XFS_RANDOM_DEFAULT
+#define XFS_RANDOM_ALLOC_READ_AGF			XFS_RANDOM_DEFAULT
+#define XFS_RANDOM_IALLOC_READ_AGI			XFS_RANDOM_DEFAULT
+#define XFS_RANDOM_ITOBP_INOTOBP			XFS_RANDOM_DEFAULT
+#define XFS_RANDOM_IUNLINK				XFS_RANDOM_DEFAULT
+#define XFS_RANDOM_IUNLINK_REMOVE			XFS_RANDOM_DEFAULT
+#define XFS_RANDOM_DIR_INO_VALIDATE			XFS_RANDOM_DEFAULT
+#define XFS_RANDOM_BULKSTAT_READ_CHUNK			XFS_RANDOM_DEFAULT
+#define XFS_RANDOM_IODONE_IOERR				(XFS_RANDOM_DEFAULT/10)
+#define XFS_RANDOM_STRATREAD_IOERR			(XFS_RANDOM_DEFAULT/10)
+#define XFS_RANDOM_STRATCMPL_IOERR			(XFS_RANDOM_DEFAULT/10)
+#define XFS_RANDOM_DIOWRITE_IOERR			(XFS_RANDOM_DEFAULT/10)
+#define	XFS_RANDOM_BMAPIFORMAT				XFS_RANDOM_DEFAULT
+
+#ifdef DEBUG
+extern int xfs_error_test_active;
+extern int xfs_error_test(int, int *, char *, int, char *, unsigned long);
+
+#define	XFS_NUM_INJECT_ERROR				10
+#define XFS_TEST_ERROR(expr, mp, tag, rf)		\
+	((expr) || (xfs_error_test_active && \
+	 xfs_error_test((tag), (mp)->m_fixedfsid, "expr", __LINE__, __FILE__, \
+			(rf))))
+
+extern int xfs_errortag_add(int error_tag, struct xfs_mount *mp);
+extern int xfs_errortag_clearall(struct xfs_mount *mp, int loud);
+#else
+#define XFS_TEST_ERROR(expr, mp, tag, rf)	(expr)
+#define xfs_errortag_add(tag, mp)		(ENOSYS)
+#define xfs_errortag_clearall(mp, loud)		(ENOSYS)
+#endif /* DEBUG */
+
+/*
+ * XFS panic tags -- allow a call to xfs_alert_tag() be turned into
+ *			a panic by setting xfs_panic_mask in a sysctl.
+ */
+#define		XFS_NO_PTAG			0
+#define		XFS_PTAG_IFLUSH			0x00000001
+#define		XFS_PTAG_LOGRES			0x00000002
+#define		XFS_PTAG_AILDELETE		0x00000004
+#define		XFS_PTAG_ERROR_REPORT		0x00000008
+#define		XFS_PTAG_SHUTDOWN_CORRUPT	0x00000010
+#define		XFS_PTAG_SHUTDOWN_IOERROR	0x00000020
+#define		XFS_PTAG_SHUTDOWN_LOGERROR	0x00000040
+#define		XFS_PTAG_FSBLOCK_ZERO		0x00000080
+
+#endif	/* __XFS_ERROR_H__ */
diff --git a/fs/xfs/xfs_extfree_item.c b/fs/xfs/xfs_extfree_item.c
new file mode 100644
index 00000000..d22e6262
--- /dev/null
+++ b/fs/xfs/xfs_extfree_item.c
@@ -0,0 +1,520 @@
+/*
+ * Copyright (c) 2000-2001,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_buf_item.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_mount.h"
+#include "xfs_trans_priv.h"
+#include "xfs_extfree_item.h"
+
+
+kmem_zone_t	*xfs_efi_zone;
+kmem_zone_t	*xfs_efd_zone;
+
+static inline struct xfs_efi_log_item *EFI_ITEM(struct xfs_log_item *lip)
+{
+	return container_of(lip, struct xfs_efi_log_item, efi_item);
+}
+
+void
+xfs_efi_item_free(
+	struct xfs_efi_log_item	*efip)
+{
+	if (efip->efi_format.efi_nextents > XFS_EFI_MAX_FAST_EXTENTS)
+		kmem_free(efip);
+	else
+		kmem_zone_free(xfs_efi_zone, efip);
+}
+
+/*
+ * Freeing the efi requires that we remove it from the AIL if it has already
+ * been placed there. However, the EFI may not yet have been placed in the AIL
+ * when called by xfs_efi_release() from EFD processing due to the ordering of
+ * committed vs unpin operations in bulk insert operations. Hence the
+ * test_and_clear_bit(XFS_EFI_COMMITTED) to ensure only the last caller frees
+ * the EFI.
+ */
+STATIC void
+__xfs_efi_release(
+	struct xfs_efi_log_item	*efip)
+{
+	struct xfs_ail		*ailp = efip->efi_item.li_ailp;
+
+	if (!test_and_clear_bit(XFS_EFI_COMMITTED, &efip->efi_flags)) {
+		spin_lock(&ailp->xa_lock);
+		/* xfs_trans_ail_delete() drops the AIL lock. */
+		xfs_trans_ail_delete(ailp, &efip->efi_item);
+		xfs_efi_item_free(efip);
+	}
+}
+
+/*
+ * This returns the number of iovecs needed to log the given efi item.
+ * We only need 1 iovec for an efi item.  It just logs the efi_log_format
+ * structure.
+ */
+STATIC uint
+xfs_efi_item_size(
+	struct xfs_log_item	*lip)
+{
+	return 1;
+}
+
+/*
+ * This is called to fill in the vector of log iovecs for the
+ * given efi log item. We use only 1 iovec, and we point that
+ * at the efi_log_format structure embedded in the efi item.
+ * It is at this point that we assert that all of the extent
+ * slots in the efi item have been filled.
+ */
+STATIC void
+xfs_efi_item_format(
+	struct xfs_log_item	*lip,
+	struct xfs_log_iovec	*log_vector)
+{
+	struct xfs_efi_log_item	*efip = EFI_ITEM(lip);
+	uint			size;
+
+	ASSERT(atomic_read(&efip->efi_next_extent) ==
+				efip->efi_format.efi_nextents);
+
+	efip->efi_format.efi_type = XFS_LI_EFI;
+
+	size = sizeof(xfs_efi_log_format_t);
+	size += (efip->efi_format.efi_nextents - 1) * sizeof(xfs_extent_t);
+	efip->efi_format.efi_size = 1;
+
+	log_vector->i_addr = &efip->efi_format;
+	log_vector->i_len = size;
+	log_vector->i_type = XLOG_REG_TYPE_EFI_FORMAT;
+	ASSERT(size >= sizeof(xfs_efi_log_format_t));
+}
+
+
+/*
+ * Pinning has no meaning for an efi item, so just return.
+ */
+STATIC void
+xfs_efi_item_pin(
+	struct xfs_log_item	*lip)
+{
+}
+
+/*
+ * While EFIs cannot really be pinned, the unpin operation is the last place at
+ * which the EFI is manipulated during a transaction.  If we are being asked to
+ * remove the EFI it's because the transaction has been cancelled and by
+ * definition that means the EFI cannot be in the AIL so remove it from the
+ * transaction and free it.  Otherwise coordinate with xfs_efi_release() (via
+ * XFS_EFI_COMMITTED) to determine who gets to free the EFI.
+ */
+STATIC void
+xfs_efi_item_unpin(
+	struct xfs_log_item	*lip,
+	int			remove)
+{
+	struct xfs_efi_log_item	*efip = EFI_ITEM(lip);
+
+	if (remove) {
+		ASSERT(!(lip->li_flags & XFS_LI_IN_AIL));
+		if (lip->li_desc)
+			xfs_trans_del_item(lip);
+		xfs_efi_item_free(efip);
+		return;
+	}
+	__xfs_efi_release(efip);
+}
+
+/*
+ * Efi items have no locking or pushing.  However, since EFIs are
+ * pulled from the AIL when their corresponding EFDs are committed
+ * to disk, their situation is very similar to being pinned.  Return
+ * XFS_ITEM_PINNED so that the caller will eventually flush the log.
+ * This should help in getting the EFI out of the AIL.
+ */
+STATIC uint
+xfs_efi_item_trylock(
+	struct xfs_log_item	*lip)
+{
+	return XFS_ITEM_PINNED;
+}
+
+/*
+ * Efi items have no locking, so just return.
+ */
+STATIC void
+xfs_efi_item_unlock(
+	struct xfs_log_item	*lip)
+{
+	if (lip->li_flags & XFS_LI_ABORTED)
+		xfs_efi_item_free(EFI_ITEM(lip));
+}
+
+/*
+ * The EFI is logged only once and cannot be moved in the log, so simply return
+ * the lsn at which it's been logged.  For bulk transaction committed
+ * processing, the EFI may be processed but not yet unpinned prior to the EFD
+ * being processed. Set the XFS_EFI_COMMITTED flag so this case can be detected
+ * when processing the EFD.
+ */
+STATIC xfs_lsn_t
+xfs_efi_item_committed(
+	struct xfs_log_item	*lip,
+	xfs_lsn_t		lsn)
+{
+	struct xfs_efi_log_item	*efip = EFI_ITEM(lip);
+
+	set_bit(XFS_EFI_COMMITTED, &efip->efi_flags);
+	return lsn;
+}
+
+/*
+ * There isn't much you can do to push on an efi item.  It is simply
+ * stuck waiting for all of its corresponding efd items to be
+ * committed to disk.
+ */
+STATIC void
+xfs_efi_item_push(
+	struct xfs_log_item	*lip)
+{
+}
+
+/*
+ * The EFI dependency tracking op doesn't do squat.  It can't because
+ * it doesn't know where the free extent is coming from.  The dependency
+ * tracking has to be handled by the "enclosing" metadata object.  For
+ * example, for inodes, the inode is locked throughout the extent freeing
+ * so the dependency should be recorded there.
+ */
+STATIC void
+xfs_efi_item_committing(
+	struct xfs_log_item	*lip,
+	xfs_lsn_t		lsn)
+{
+}
+
+/*
+ * This is the ops vector shared by all efi log items.
+ */
+static struct xfs_item_ops xfs_efi_item_ops = {
+	.iop_size	= xfs_efi_item_size,
+	.iop_format	= xfs_efi_item_format,
+	.iop_pin	= xfs_efi_item_pin,
+	.iop_unpin	= xfs_efi_item_unpin,
+	.iop_trylock	= xfs_efi_item_trylock,
+	.iop_unlock	= xfs_efi_item_unlock,
+	.iop_committed	= xfs_efi_item_committed,
+	.iop_push	= xfs_efi_item_push,
+	.iop_committing = xfs_efi_item_committing
+};
+
+
+/*
+ * Allocate and initialize an efi item with the given number of extents.
+ */
+struct xfs_efi_log_item *
+xfs_efi_init(
+	struct xfs_mount	*mp,
+	uint			nextents)
+
+{
+	struct xfs_efi_log_item	*efip;
+	uint			size;
+
+	ASSERT(nextents > 0);
+	if (nextents > XFS_EFI_MAX_FAST_EXTENTS) {
+		size = (uint)(sizeof(xfs_efi_log_item_t) +
+			((nextents - 1) * sizeof(xfs_extent_t)));
+		efip = kmem_zalloc(size, KM_SLEEP);
+	} else {
+		efip = kmem_zone_zalloc(xfs_efi_zone, KM_SLEEP);
+	}
+
+	xfs_log_item_init(mp, &efip->efi_item, XFS_LI_EFI, &xfs_efi_item_ops);
+	efip->efi_format.efi_nextents = nextents;
+	efip->efi_format.efi_id = (__psint_t)(void*)efip;
+	atomic_set(&efip->efi_next_extent, 0);
+
+	return efip;
+}
+
+/*
+ * Copy an EFI format buffer from the given buf, and into the destination
+ * EFI format structure.
+ * The given buffer can be in 32 bit or 64 bit form (which has different padding),
+ * one of which will be the native format for this kernel.
+ * It will handle the conversion of formats if necessary.
+ */
+int
+xfs_efi_copy_format(xfs_log_iovec_t *buf, xfs_efi_log_format_t *dst_efi_fmt)
+{
+	xfs_efi_log_format_t *src_efi_fmt = buf->i_addr;
+	uint i;
+	uint len = sizeof(xfs_efi_log_format_t) + 
+		(src_efi_fmt->efi_nextents - 1) * sizeof(xfs_extent_t);  
+	uint len32 = sizeof(xfs_efi_log_format_32_t) + 
+		(src_efi_fmt->efi_nextents - 1) * sizeof(xfs_extent_32_t);  
+	uint len64 = sizeof(xfs_efi_log_format_64_t) + 
+		(src_efi_fmt->efi_nextents - 1) * sizeof(xfs_extent_64_t);  
+
+	if (buf->i_len == len) {
+		memcpy((char *)dst_efi_fmt, (char*)src_efi_fmt, len);
+		return 0;
+	} else if (buf->i_len == len32) {
+		xfs_efi_log_format_32_t *src_efi_fmt_32 = buf->i_addr;
+
+		dst_efi_fmt->efi_type     = src_efi_fmt_32->efi_type;
+		dst_efi_fmt->efi_size     = src_efi_fmt_32->efi_size;
+		dst_efi_fmt->efi_nextents = src_efi_fmt_32->efi_nextents;
+		dst_efi_fmt->efi_id       = src_efi_fmt_32->efi_id;
+		for (i = 0; i < dst_efi_fmt->efi_nextents; i++) {
+			dst_efi_fmt->efi_extents[i].ext_start =
+				src_efi_fmt_32->efi_extents[i].ext_start;
+			dst_efi_fmt->efi_extents[i].ext_len =
+				src_efi_fmt_32->efi_extents[i].ext_len;
+		}
+		return 0;
+	} else if (buf->i_len == len64) {
+		xfs_efi_log_format_64_t *src_efi_fmt_64 = buf->i_addr;
+
+		dst_efi_fmt->efi_type     = src_efi_fmt_64->efi_type;
+		dst_efi_fmt->efi_size     = src_efi_fmt_64->efi_size;
+		dst_efi_fmt->efi_nextents = src_efi_fmt_64->efi_nextents;
+		dst_efi_fmt->efi_id       = src_efi_fmt_64->efi_id;
+		for (i = 0; i < dst_efi_fmt->efi_nextents; i++) {
+			dst_efi_fmt->efi_extents[i].ext_start =
+				src_efi_fmt_64->efi_extents[i].ext_start;
+			dst_efi_fmt->efi_extents[i].ext_len =
+				src_efi_fmt_64->efi_extents[i].ext_len;
+		}
+		return 0;
+	}
+	return EFSCORRUPTED;
+}
+
+/*
+ * This is called by the efd item code below to release references to the given
+ * efi item.  Each efd calls this with the number of extents that it has
+ * logged, and when the sum of these reaches the total number of extents logged
+ * by this efi item we can free the efi item.
+ */
+void
+xfs_efi_release(xfs_efi_log_item_t	*efip,
+		uint			nextents)
+{
+	ASSERT(atomic_read(&efip->efi_next_extent) >= nextents);
+	if (atomic_sub_and_test(nextents, &efip->efi_next_extent))
+		__xfs_efi_release(efip);
+}
+
+static inline struct xfs_efd_log_item *EFD_ITEM(struct xfs_log_item *lip)
+{
+	return container_of(lip, struct xfs_efd_log_item, efd_item);
+}
+
+STATIC void
+xfs_efd_item_free(struct xfs_efd_log_item *efdp)
+{
+	if (efdp->efd_format.efd_nextents > XFS_EFD_MAX_FAST_EXTENTS)
+		kmem_free(efdp);
+	else
+		kmem_zone_free(xfs_efd_zone, efdp);
+}
+
+/*
+ * This returns the number of iovecs needed to log the given efd item.
+ * We only need 1 iovec for an efd item.  It just logs the efd_log_format
+ * structure.
+ */
+STATIC uint
+xfs_efd_item_size(
+	struct xfs_log_item	*lip)
+{
+	return 1;
+}
+
+/*
+ * This is called to fill in the vector of log iovecs for the
+ * given efd log item. We use only 1 iovec, and we point that
+ * at the efd_log_format structure embedded in the efd item.
+ * It is at this point that we assert that all of the extent
+ * slots in the efd item have been filled.
+ */
+STATIC void
+xfs_efd_item_format(
+	struct xfs_log_item	*lip,
+	struct xfs_log_iovec	*log_vector)
+{
+	struct xfs_efd_log_item	*efdp = EFD_ITEM(lip);
+	uint			size;
+
+	ASSERT(efdp->efd_next_extent == efdp->efd_format.efd_nextents);
+
+	efdp->efd_format.efd_type = XFS_LI_EFD;
+
+	size = sizeof(xfs_efd_log_format_t);
+	size += (efdp->efd_format.efd_nextents - 1) * sizeof(xfs_extent_t);
+	efdp->efd_format.efd_size = 1;
+
+	log_vector->i_addr = &efdp->efd_format;
+	log_vector->i_len = size;
+	log_vector->i_type = XLOG_REG_TYPE_EFD_FORMAT;
+	ASSERT(size >= sizeof(xfs_efd_log_format_t));
+}
+
+/*
+ * Pinning has no meaning for an efd item, so just return.
+ */
+STATIC void
+xfs_efd_item_pin(
+	struct xfs_log_item	*lip)
+{
+}
+
+/*
+ * Since pinning has no meaning for an efd item, unpinning does
+ * not either.
+ */
+STATIC void
+xfs_efd_item_unpin(
+	struct xfs_log_item	*lip,
+	int			remove)
+{
+}
+
+/*
+ * Efd items have no locking, so just return success.
+ */
+STATIC uint
+xfs_efd_item_trylock(
+	struct xfs_log_item	*lip)
+{
+	return XFS_ITEM_LOCKED;
+}
+
+/*
+ * Efd items have no locking or pushing, so return failure
+ * so that the caller doesn't bother with us.
+ */
+STATIC void
+xfs_efd_item_unlock(
+	struct xfs_log_item	*lip)
+{
+	if (lip->li_flags & XFS_LI_ABORTED)
+		xfs_efd_item_free(EFD_ITEM(lip));
+}
+
+/*
+ * When the efd item is committed to disk, all we need to do
+ * is delete our reference to our partner efi item and then
+ * free ourselves.  Since we're freeing ourselves we must
+ * return -1 to keep the transaction code from further referencing
+ * this item.
+ */
+STATIC xfs_lsn_t
+xfs_efd_item_committed(
+	struct xfs_log_item	*lip,
+	xfs_lsn_t		lsn)
+{
+	struct xfs_efd_log_item	*efdp = EFD_ITEM(lip);
+
+	/*
+	 * If we got a log I/O error, it's always the case that the LR with the
+	 * EFI got unpinned and freed before the EFD got aborted.
+	 */
+	if (!(lip->li_flags & XFS_LI_ABORTED))
+		xfs_efi_release(efdp->efd_efip, efdp->efd_format.efd_nextents);
+
+	xfs_efd_item_free(efdp);
+	return (xfs_lsn_t)-1;
+}
+
+/*
+ * There isn't much you can do to push on an efd item.  It is simply
+ * stuck waiting for the log to be flushed to disk.
+ */
+STATIC void
+xfs_efd_item_push(
+	struct xfs_log_item	*lip)
+{
+}
+
+/*
+ * The EFD dependency tracking op doesn't do squat.  It can't because
+ * it doesn't know where the free extent is coming from.  The dependency
+ * tracking has to be handled by the "enclosing" metadata object.  For
+ * example, for inodes, the inode is locked throughout the extent freeing
+ * so the dependency should be recorded there.
+ */
+STATIC void
+xfs_efd_item_committing(
+	struct xfs_log_item	*lip,
+	xfs_lsn_t		lsn)
+{
+}
+
+/*
+ * This is the ops vector shared by all efd log items.
+ */
+static struct xfs_item_ops xfs_efd_item_ops = {
+	.iop_size	= xfs_efd_item_size,
+	.iop_format	= xfs_efd_item_format,
+	.iop_pin	= xfs_efd_item_pin,
+	.iop_unpin	= xfs_efd_item_unpin,
+	.iop_trylock	= xfs_efd_item_trylock,
+	.iop_unlock	= xfs_efd_item_unlock,
+	.iop_committed	= xfs_efd_item_committed,
+	.iop_push	= xfs_efd_item_push,
+	.iop_committing = xfs_efd_item_committing
+};
+
+/*
+ * Allocate and initialize an efd item with the given number of extents.
+ */
+struct xfs_efd_log_item *
+xfs_efd_init(
+	struct xfs_mount	*mp,
+	struct xfs_efi_log_item	*efip,
+	uint			nextents)
+
+{
+	struct xfs_efd_log_item	*efdp;
+	uint			size;
+
+	ASSERT(nextents > 0);
+	if (nextents > XFS_EFD_MAX_FAST_EXTENTS) {
+		size = (uint)(sizeof(xfs_efd_log_item_t) +
+			((nextents - 1) * sizeof(xfs_extent_t)));
+		efdp = kmem_zalloc(size, KM_SLEEP);
+	} else {
+		efdp = kmem_zone_zalloc(xfs_efd_zone, KM_SLEEP);
+	}
+
+	xfs_log_item_init(mp, &efdp->efd_item, XFS_LI_EFD, &xfs_efd_item_ops);
+	efdp->efd_efip = efip;
+	efdp->efd_format.efd_nextents = nextents;
+	efdp->efd_format.efd_efi_id = efip->efi_format.efi_id;
+
+	return efdp;
+}
diff --git a/fs/xfs/xfs_extfree_item.h b/fs/xfs/xfs_extfree_item.h
new file mode 100644
index 00000000..375f68e4
--- /dev/null
+++ b/fs/xfs/xfs_extfree_item.h
@@ -0,0 +1,161 @@
+/*
+ * Copyright (c) 2000,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef	__XFS_EXTFREE_ITEM_H__
+#define	__XFS_EXTFREE_ITEM_H__
+
+struct xfs_mount;
+struct kmem_zone;
+
+typedef struct xfs_extent {
+	xfs_dfsbno_t	ext_start;
+	xfs_extlen_t	ext_len;
+} xfs_extent_t;
+
+/*
+ * Since an xfs_extent_t has types (start:64, len: 32)
+ * there are different alignments on 32 bit and 64 bit kernels.
+ * So we provide the different variants for use by a
+ * conversion routine.
+ */
+
+typedef struct xfs_extent_32 {
+	__uint64_t	ext_start;
+	__uint32_t	ext_len;
+} __attribute__((packed)) xfs_extent_32_t;
+
+typedef struct xfs_extent_64 {
+	__uint64_t	ext_start;
+	__uint32_t	ext_len;
+	__uint32_t	ext_pad;
+} xfs_extent_64_t;
+
+/*
+ * This is the structure used to lay out an efi log item in the
+ * log.  The efi_extents field is a variable size array whose
+ * size is given by efi_nextents.
+ */
+typedef struct xfs_efi_log_format {
+	__uint16_t		efi_type;	/* efi log item type */
+	__uint16_t		efi_size;	/* size of this item */
+	__uint32_t		efi_nextents;	/* # extents to free */
+	__uint64_t		efi_id;		/* efi identifier */
+	xfs_extent_t		efi_extents[1];	/* array of extents to free */
+} xfs_efi_log_format_t;
+
+typedef struct xfs_efi_log_format_32 {
+	__uint16_t		efi_type;	/* efi log item type */
+	__uint16_t		efi_size;	/* size of this item */
+	__uint32_t		efi_nextents;	/* # extents to free */
+	__uint64_t		efi_id;		/* efi identifier */
+	xfs_extent_32_t		efi_extents[1];	/* array of extents to free */
+} __attribute__((packed)) xfs_efi_log_format_32_t;
+
+typedef struct xfs_efi_log_format_64 {
+	__uint16_t		efi_type;	/* efi log item type */
+	__uint16_t		efi_size;	/* size of this item */
+	__uint32_t		efi_nextents;	/* # extents to free */
+	__uint64_t		efi_id;		/* efi identifier */
+	xfs_extent_64_t		efi_extents[1];	/* array of extents to free */
+} xfs_efi_log_format_64_t;
+
+/*
+ * This is the structure used to lay out an efd log item in the
+ * log.  The efd_extents array is a variable size array whose
+ * size is given by efd_nextents;
+ */
+typedef struct xfs_efd_log_format {
+	__uint16_t		efd_type;	/* efd log item type */
+	__uint16_t		efd_size;	/* size of this item */
+	__uint32_t		efd_nextents;	/* # of extents freed */
+	__uint64_t		efd_efi_id;	/* id of corresponding efi */
+	xfs_extent_t		efd_extents[1];	/* array of extents freed */
+} xfs_efd_log_format_t;
+
+typedef struct xfs_efd_log_format_32 {
+	__uint16_t		efd_type;	/* efd log item type */
+	__uint16_t		efd_size;	/* size of this item */
+	__uint32_t		efd_nextents;	/* # of extents freed */
+	__uint64_t		efd_efi_id;	/* id of corresponding efi */
+	xfs_extent_32_t		efd_extents[1];	/* array of extents freed */
+} __attribute__((packed)) xfs_efd_log_format_32_t;
+
+typedef struct xfs_efd_log_format_64 {
+	__uint16_t		efd_type;	/* efd log item type */
+	__uint16_t		efd_size;	/* size of this item */
+	__uint32_t		efd_nextents;	/* # of extents freed */
+	__uint64_t		efd_efi_id;	/* id of corresponding efi */
+	xfs_extent_64_t		efd_extents[1];	/* array of extents freed */
+} xfs_efd_log_format_64_t;
+
+
+#ifdef __KERNEL__
+
+/*
+ * Max number of extents in fast allocation path.
+ */
+#define	XFS_EFI_MAX_FAST_EXTENTS	16
+
+/*
+ * Define EFI flag bits. Manipulated by set/clear/test_bit operators.
+ */
+#define	XFS_EFI_RECOVERED	1
+#define	XFS_EFI_COMMITTED	2
+
+/*
+ * This is the "extent free intention" log item.  It is used
+ * to log the fact that some extents need to be free.  It is
+ * used in conjunction with the "extent free done" log item
+ * described below.
+ */
+typedef struct xfs_efi_log_item {
+	xfs_log_item_t		efi_item;
+	atomic_t		efi_next_extent;
+	unsigned long		efi_flags;	/* misc flags */
+	xfs_efi_log_format_t	efi_format;
+} xfs_efi_log_item_t;
+
+/*
+ * This is the "extent free done" log item.  It is used to log
+ * the fact that some extents earlier mentioned in an efi item
+ * have been freed.
+ */
+typedef struct xfs_efd_log_item {
+	xfs_log_item_t		efd_item;
+	xfs_efi_log_item_t	*efd_efip;
+	uint			efd_next_extent;
+	xfs_efd_log_format_t	efd_format;
+} xfs_efd_log_item_t;
+
+/*
+ * Max number of extents in fast allocation path.
+ */
+#define	XFS_EFD_MAX_FAST_EXTENTS	16
+
+extern struct kmem_zone	*xfs_efi_zone;
+extern struct kmem_zone	*xfs_efd_zone;
+
+xfs_efi_log_item_t	*xfs_efi_init(struct xfs_mount *, uint);
+xfs_efd_log_item_t	*xfs_efd_init(struct xfs_mount *, xfs_efi_log_item_t *,
+				      uint);
+int			xfs_efi_copy_format(xfs_log_iovec_t *buf,
+					    xfs_efi_log_format_t *dst_efi_fmt);
+void			xfs_efi_item_free(xfs_efi_log_item_t *);
+
+#endif	/* __KERNEL__ */
+
+#endif	/* __XFS_EXTFREE_ITEM_H__ */
diff --git a/fs/xfs/xfs_filestream.c b/fs/xfs/xfs_filestream.c
new file mode 100644
index 00000000..9124425b
--- /dev/null
+++ b/fs/xfs/xfs_filestream.c
@@ -0,0 +1,824 @@
+/*
+ * Copyright (c) 2006-2007 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_inum.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_ag.h"
+#include "xfs_log.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_mount.h"
+#include "xfs_bmap.h"
+#include "xfs_alloc.h"
+#include "xfs_utils.h"
+#include "xfs_mru_cache.h"
+#include "xfs_filestream.h"
+#include "xfs_trace.h"
+
+#ifdef XFS_FILESTREAMS_TRACE
+
+ktrace_t *xfs_filestreams_trace_buf;
+
+STATIC void
+xfs_filestreams_trace(
+	xfs_mount_t	*mp,	/* mount point */
+	int		type,	/* type of trace */
+	const char	*func,	/* source function */
+	int		line,	/* source line number */
+	__psunsigned_t	arg0,
+	__psunsigned_t	arg1,
+	__psunsigned_t	arg2,
+	__psunsigned_t	arg3,
+	__psunsigned_t	arg4,
+	__psunsigned_t	arg5)
+{
+	ktrace_enter(xfs_filestreams_trace_buf,
+		(void *)(__psint_t)(type | (line << 16)),
+		(void *)func,
+		(void *)(__psunsigned_t)current_pid(),
+		(void *)mp,
+		(void *)(__psunsigned_t)arg0,
+		(void *)(__psunsigned_t)arg1,
+		(void *)(__psunsigned_t)arg2,
+		(void *)(__psunsigned_t)arg3,
+		(void *)(__psunsigned_t)arg4,
+		(void *)(__psunsigned_t)arg5,
+		NULL, NULL, NULL, NULL, NULL, NULL);
+}
+
+#define TRACE0(mp,t)			TRACE6(mp,t,0,0,0,0,0,0)
+#define TRACE1(mp,t,a0)			TRACE6(mp,t,a0,0,0,0,0,0)
+#define TRACE2(mp,t,a0,a1)		TRACE6(mp,t,a0,a1,0,0,0,0)
+#define TRACE3(mp,t,a0,a1,a2)		TRACE6(mp,t,a0,a1,a2,0,0,0)
+#define TRACE4(mp,t,a0,a1,a2,a3)	TRACE6(mp,t,a0,a1,a2,a3,0,0)
+#define TRACE5(mp,t,a0,a1,a2,a3,a4)	TRACE6(mp,t,a0,a1,a2,a3,a4,0)
+#define TRACE6(mp,t,a0,a1,a2,a3,a4,a5) \
+	xfs_filestreams_trace(mp, t, __func__, __LINE__, \
+				(__psunsigned_t)a0, (__psunsigned_t)a1, \
+				(__psunsigned_t)a2, (__psunsigned_t)a3, \
+				(__psunsigned_t)a4, (__psunsigned_t)a5)
+
+#define TRACE_AG_SCAN(mp, ag, ag2) \
+		TRACE2(mp, XFS_FSTRM_KTRACE_AGSCAN, ag, ag2);
+#define TRACE_AG_PICK1(mp, max_ag, maxfree) \
+		TRACE2(mp, XFS_FSTRM_KTRACE_AGPICK1, max_ag, maxfree);
+#define TRACE_AG_PICK2(mp, ag, ag2, cnt, free, scan, flag) \
+		TRACE6(mp, XFS_FSTRM_KTRACE_AGPICK2, ag, ag2, \
+			 cnt, free, scan, flag)
+#define TRACE_UPDATE(mp, ip, ag, cnt, ag2, cnt2) \
+		TRACE5(mp, XFS_FSTRM_KTRACE_UPDATE, ip, ag, cnt, ag2, cnt2)
+#define TRACE_FREE(mp, ip, pip, ag, cnt) \
+		TRACE4(mp, XFS_FSTRM_KTRACE_FREE, ip, pip, ag, cnt)
+#define TRACE_LOOKUP(mp, ip, pip, ag, cnt) \
+		TRACE4(mp, XFS_FSTRM_KTRACE_ITEM_LOOKUP, ip, pip, ag, cnt)
+#define TRACE_ASSOCIATE(mp, ip, pip, ag, cnt) \
+		TRACE4(mp, XFS_FSTRM_KTRACE_ASSOCIATE, ip, pip, ag, cnt)
+#define TRACE_MOVEAG(mp, ip, pip, oag, ocnt, nag, ncnt) \
+		TRACE6(mp, XFS_FSTRM_KTRACE_MOVEAG, ip, pip, oag, ocnt, nag, ncnt)
+#define TRACE_ORPHAN(mp, ip, ag) \
+		TRACE2(mp, XFS_FSTRM_KTRACE_ORPHAN, ip, ag);
+
+
+#else
+#define TRACE_AG_SCAN(mp, ag, ag2)
+#define TRACE_AG_PICK1(mp, max_ag, maxfree)
+#define TRACE_AG_PICK2(mp, ag, ag2, cnt, free, scan, flag)
+#define TRACE_UPDATE(mp, ip, ag, cnt, ag2, cnt2)
+#define TRACE_FREE(mp, ip, pip, ag, cnt)
+#define TRACE_LOOKUP(mp, ip, pip, ag, cnt)
+#define TRACE_ASSOCIATE(mp, ip, pip, ag, cnt)
+#define TRACE_MOVEAG(mp, ip, pip, oag, ocnt, nag, ncnt)
+#define TRACE_ORPHAN(mp, ip, ag)
+#endif
+
+static kmem_zone_t *item_zone;
+
+/*
+ * Structure for associating a file or a directory with an allocation group.
+ * The parent directory pointer is only needed for files, but since there will
+ * generally be vastly more files than directories in the cache, using the same
+ * data structure simplifies the code with very little memory overhead.
+ */
+typedef struct fstrm_item
+{
+	xfs_agnumber_t	ag;	/* AG currently in use for the file/directory. */
+	xfs_inode_t	*ip;	/* inode self-pointer. */
+	xfs_inode_t	*pip;	/* Parent directory inode pointer. */
+} fstrm_item_t;
+
+/*
+ * Allocation group filestream associations are tracked with per-ag atomic
+ * counters.  These counters allow _xfs_filestream_pick_ag() to tell whether a
+ * particular AG already has active filestreams associated with it. The mount
+ * point's m_peraglock is used to protect these counters from per-ag array
+ * re-allocation during a growfs operation.  When xfs_growfs_data_private() is
+ * about to reallocate the array, it calls xfs_filestream_flush() with the
+ * m_peraglock held in write mode.
+ *
+ * Since xfs_mru_cache_flush() guarantees that all the free functions for all
+ * the cache elements have finished executing before it returns, it's safe for
+ * the free functions to use the atomic counters without m_peraglock protection.
+ * This allows the implementation of xfs_fstrm_free_func() to be agnostic about
+ * whether it was called with the m_peraglock held in read mode, write mode or
+ * not held at all.  The race condition this addresses is the following:
+ *
+ *  - The work queue scheduler fires and pulls a filestream directory cache
+ *    element off the LRU end of the cache for deletion, then gets pre-empted.
+ *  - A growfs operation grabs the m_peraglock in write mode, flushes all the
+ *    remaining items from the cache and reallocates the mount point's per-ag
+ *    array, resetting all the counters to zero.
+ *  - The work queue thread resumes and calls the free function for the element
+ *    it started cleaning up earlier.  In the process it decrements the
+ *    filestreams counter for an AG that now has no references.
+ *
+ * With a shrinkfs feature, the above scenario could panic the system.
+ *
+ * All other uses of the following macros should be protected by either the
+ * m_peraglock held in read mode, or the cache's internal locking exposed by the
+ * interval between a call to xfs_mru_cache_lookup() and a call to
+ * xfs_mru_cache_done().  In addition, the m_peraglock must be held in read mode
+ * when new elements are added to the cache.
+ *
+ * Combined, these locking rules ensure that no associations will ever exist in
+ * the cache that reference per-ag array elements that have since been
+ * reallocated.
+ */
+static int
+xfs_filestream_peek_ag(
+	xfs_mount_t	*mp,
+	xfs_agnumber_t	agno)
+{
+	struct xfs_perag *pag;
+	int		ret;
+
+	pag = xfs_perag_get(mp, agno);
+	ret = atomic_read(&pag->pagf_fstrms);
+	xfs_perag_put(pag);
+	return ret;
+}
+
+static int
+xfs_filestream_get_ag(
+	xfs_mount_t	*mp,
+	xfs_agnumber_t	agno)
+{
+	struct xfs_perag *pag;
+	int		ret;
+
+	pag = xfs_perag_get(mp, agno);
+	ret = atomic_inc_return(&pag->pagf_fstrms);
+	xfs_perag_put(pag);
+	return ret;
+}
+
+static void
+xfs_filestream_put_ag(
+	xfs_mount_t	*mp,
+	xfs_agnumber_t	agno)
+{
+	struct xfs_perag *pag;
+
+	pag = xfs_perag_get(mp, agno);
+	atomic_dec(&pag->pagf_fstrms);
+	xfs_perag_put(pag);
+}
+
+/*
+ * Scan the AGs starting at startag looking for an AG that isn't in use and has
+ * at least minlen blocks free.
+ */
+static int
+_xfs_filestream_pick_ag(
+	xfs_mount_t	*mp,
+	xfs_agnumber_t	startag,
+	xfs_agnumber_t	*agp,
+	int		flags,
+	xfs_extlen_t	minlen)
+{
+	int		streams, max_streams;
+	int		err, trylock, nscan;
+	xfs_extlen_t	longest, free, minfree, maxfree = 0;
+	xfs_agnumber_t	ag, max_ag = NULLAGNUMBER;
+	struct xfs_perag *pag;
+
+	/* 2% of an AG's blocks must be free for it to be chosen. */
+	minfree = mp->m_sb.sb_agblocks / 50;
+
+	ag = startag;
+	*agp = NULLAGNUMBER;
+
+	/* For the first pass, don't sleep trying to init the per-AG. */
+	trylock = XFS_ALLOC_FLAG_TRYLOCK;
+
+	for (nscan = 0; 1; nscan++) {
+		pag = xfs_perag_get(mp, ag);
+		TRACE_AG_SCAN(mp, ag, atomic_read(&pag->pagf_fstrms));
+
+		if (!pag->pagf_init) {
+			err = xfs_alloc_pagf_init(mp, NULL, ag, trylock);
+			if (err && !trylock) {
+				xfs_perag_put(pag);
+				return err;
+			}
+		}
+
+		/* Might fail sometimes during the 1st pass with trylock set. */
+		if (!pag->pagf_init)
+			goto next_ag;
+
+		/* Keep track of the AG with the most free blocks. */
+		if (pag->pagf_freeblks > maxfree) {
+			maxfree = pag->pagf_freeblks;
+			max_streams = atomic_read(&pag->pagf_fstrms);
+			max_ag = ag;
+		}
+
+		/*
+		 * The AG reference count does two things: it enforces mutual
+		 * exclusion when examining the suitability of an AG in this
+		 * loop, and it guards against two filestreams being established
+		 * in the same AG as each other.
+		 */
+		if (xfs_filestream_get_ag(mp, ag) > 1) {
+			xfs_filestream_put_ag(mp, ag);
+			goto next_ag;
+		}
+
+		longest = xfs_alloc_longest_free_extent(mp, pag);
+		if (((minlen && longest >= minlen) ||
+		     (!minlen && pag->pagf_freeblks >= minfree)) &&
+		    (!pag->pagf_metadata || !(flags & XFS_PICK_USERDATA) ||
+		     (flags & XFS_PICK_LOWSPACE))) {
+
+			/* Break out, retaining the reference on the AG. */
+			free = pag->pagf_freeblks;
+			streams = atomic_read(&pag->pagf_fstrms);
+			xfs_perag_put(pag);
+			*agp = ag;
+			break;
+		}
+
+		/* Drop the reference on this AG, it's not usable. */
+		xfs_filestream_put_ag(mp, ag);
+next_ag:
+		xfs_perag_put(pag);
+		/* Move to the next AG, wrapping to AG 0 if necessary. */
+		if (++ag >= mp->m_sb.sb_agcount)
+			ag = 0;
+
+		/* If a full pass of the AGs hasn't been done yet, continue. */
+		if (ag != startag)
+			continue;
+
+		/* Allow sleeping in xfs_alloc_pagf_init() on the 2nd pass. */
+		if (trylock != 0) {
+			trylock = 0;
+			continue;
+		}
+
+		/* Finally, if lowspace wasn't set, set it for the 3rd pass. */
+		if (!(flags & XFS_PICK_LOWSPACE)) {
+			flags |= XFS_PICK_LOWSPACE;
+			continue;
+		}
+
+		/*
+		 * Take the AG with the most free space, regardless of whether
+		 * it's already in use by another filestream.
+		 */
+		if (max_ag != NULLAGNUMBER) {
+			xfs_filestream_get_ag(mp, max_ag);
+			TRACE_AG_PICK1(mp, max_ag, maxfree);
+			streams = max_streams;
+			free = maxfree;
+			*agp = max_ag;
+			break;
+		}
+
+		/* take AG 0 if none matched */
+		TRACE_AG_PICK1(mp, max_ag, maxfree);
+		*agp = 0;
+		return 0;
+	}
+
+	TRACE_AG_PICK2(mp, startag, *agp, streams, free, nscan, flags);
+
+	return 0;
+}
+
+/*
+ * Set the allocation group number for a file or a directory, updating inode
+ * references and per-AG references as appropriate.
+ */
+static int
+_xfs_filestream_update_ag(
+	xfs_inode_t	*ip,
+	xfs_inode_t	*pip,
+	xfs_agnumber_t	ag)
+{
+	int		err = 0;
+	xfs_mount_t	*mp;
+	xfs_mru_cache_t	*cache;
+	fstrm_item_t	*item;
+	xfs_agnumber_t	old_ag;
+	xfs_inode_t	*old_pip;
+
+	/*
+	 * Either ip is a regular file and pip is a directory, or ip is a
+	 * directory and pip is NULL.
+	 */
+	ASSERT(ip && (((ip->i_d.di_mode & S_IFREG) && pip &&
+	               (pip->i_d.di_mode & S_IFDIR)) ||
+	              ((ip->i_d.di_mode & S_IFDIR) && !pip)));
+
+	mp = ip->i_mount;
+	cache = mp->m_filestream;
+
+	item = xfs_mru_cache_lookup(cache, ip->i_ino);
+	if (item) {
+		ASSERT(item->ip == ip);
+		old_ag = item->ag;
+		item->ag = ag;
+		old_pip = item->pip;
+		item->pip = pip;
+		xfs_mru_cache_done(cache);
+
+		/*
+		 * If the AG has changed, drop the old ref and take a new one,
+		 * effectively transferring the reference from old to new AG.
+		 */
+		if (ag != old_ag) {
+			xfs_filestream_put_ag(mp, old_ag);
+			xfs_filestream_get_ag(mp, ag);
+		}
+
+		/*
+		 * If ip is a file and its pip has changed, drop the old ref and
+		 * take a new one.
+		 */
+		if (pip && pip != old_pip) {
+			IRELE(old_pip);
+			IHOLD(pip);
+		}
+
+		TRACE_UPDATE(mp, ip, old_ag, xfs_filestream_peek_ag(mp, old_ag),
+				ag, xfs_filestream_peek_ag(mp, ag));
+		return 0;
+	}
+
+	item = kmem_zone_zalloc(item_zone, KM_MAYFAIL);
+	if (!item)
+		return ENOMEM;
+
+	item->ag = ag;
+	item->ip = ip;
+	item->pip = pip;
+
+	err = xfs_mru_cache_insert(cache, ip->i_ino, item);
+	if (err) {
+		kmem_zone_free(item_zone, item);
+		return err;
+	}
+
+	/* Take a reference on the AG. */
+	xfs_filestream_get_ag(mp, ag);
+
+	/*
+	 * Take a reference on the inode itself regardless of whether it's a
+	 * regular file or a directory.
+	 */
+	IHOLD(ip);
+
+	/*
+	 * In the case of a regular file, take a reference on the parent inode
+	 * as well to ensure it remains in-core.
+	 */
+	if (pip)
+		IHOLD(pip);
+
+	TRACE_UPDATE(mp, ip, ag, xfs_filestream_peek_ag(mp, ag),
+			ag, xfs_filestream_peek_ag(mp, ag));
+
+	return 0;
+}
+
+/* xfs_fstrm_free_func(): callback for freeing cached stream items. */
+STATIC void
+xfs_fstrm_free_func(
+	unsigned long	ino,
+	void		*data)
+{
+	fstrm_item_t	*item  = (fstrm_item_t *)data;
+	xfs_inode_t	*ip = item->ip;
+
+	ASSERT(ip->i_ino == ino);
+
+	xfs_iflags_clear(ip, XFS_IFILESTREAM);
+
+	/* Drop the reference taken on the AG when the item was added. */
+	xfs_filestream_put_ag(ip->i_mount, item->ag);
+
+	TRACE_FREE(ip->i_mount, ip, item->pip, item->ag,
+		xfs_filestream_peek_ag(ip->i_mount, item->ag));
+
+	/*
+	 * _xfs_filestream_update_ag() always takes a reference on the inode
+	 * itself, whether it's a file or a directory.  Release it here.
+	 * This can result in the inode being freed and so we must
+	 * not hold any inode locks when freeing filesstreams objects
+	 * otherwise we can deadlock here.
+	 */
+	IRELE(ip);
+
+	/*
+	 * In the case of a regular file, _xfs_filestream_update_ag() also
+	 * takes a ref on the parent inode to keep it in-core.  Release that
+	 * too.
+	 */
+	if (item->pip)
+		IRELE(item->pip);
+
+	/* Finally, free the memory allocated for the item. */
+	kmem_zone_free(item_zone, item);
+}
+
+/*
+ * xfs_filestream_init() is called at xfs initialisation time to set up the
+ * memory zone that will be used for filestream data structure allocation.
+ */
+int
+xfs_filestream_init(void)
+{
+	item_zone = kmem_zone_init(sizeof(fstrm_item_t), "fstrm_item");
+	if (!item_zone)
+		return -ENOMEM;
+
+	return 0;
+}
+
+/*
+ * xfs_filestream_uninit() is called at xfs termination time to destroy the
+ * memory zone that was used for filestream data structure allocation.
+ */
+void
+xfs_filestream_uninit(void)
+{
+	kmem_zone_destroy(item_zone);
+}
+
+/*
+ * xfs_filestream_mount() is called when a file system is mounted with the
+ * filestream option.  It is responsible for allocating the data structures
+ * needed to track the new file system's file streams.
+ */
+int
+xfs_filestream_mount(
+	xfs_mount_t	*mp)
+{
+	int		err;
+	unsigned int	lifetime, grp_count;
+
+	/*
+	 * The filestream timer tunable is currently fixed within the range of
+	 * one second to four minutes, with five seconds being the default.  The
+	 * group count is somewhat arbitrary, but it'd be nice to adhere to the
+	 * timer tunable to within about 10 percent.  This requires at least 10
+	 * groups.
+	 */
+	lifetime  = xfs_fstrm_centisecs * 10;
+	grp_count = 10;
+
+	err = xfs_mru_cache_create(&mp->m_filestream, lifetime, grp_count,
+	                     xfs_fstrm_free_func);
+
+	return err;
+}
+
+/*
+ * xfs_filestream_unmount() is called when a file system that was mounted with
+ * the filestream option is unmounted.  It drains the data structures created
+ * to track the file system's file streams and frees all the memory that was
+ * allocated.
+ */
+void
+xfs_filestream_unmount(
+	xfs_mount_t	*mp)
+{
+	xfs_mru_cache_destroy(mp->m_filestream);
+}
+
+/*
+ * Return the AG of the filestream the file or directory belongs to, or
+ * NULLAGNUMBER otherwise.
+ */
+xfs_agnumber_t
+xfs_filestream_lookup_ag(
+	xfs_inode_t	*ip)
+{
+	xfs_mru_cache_t	*cache;
+	fstrm_item_t	*item;
+	xfs_agnumber_t	ag;
+	int		ref;
+
+	if (!(ip->i_d.di_mode & (S_IFREG | S_IFDIR))) {
+		ASSERT(0);
+		return NULLAGNUMBER;
+	}
+
+	cache = ip->i_mount->m_filestream;
+	item = xfs_mru_cache_lookup(cache, ip->i_ino);
+	if (!item) {
+		TRACE_LOOKUP(ip->i_mount, ip, NULL, NULLAGNUMBER, 0);
+		return NULLAGNUMBER;
+	}
+
+	ASSERT(ip == item->ip);
+	ag = item->ag;
+	ref = xfs_filestream_peek_ag(ip->i_mount, ag);
+	xfs_mru_cache_done(cache);
+
+	TRACE_LOOKUP(ip->i_mount, ip, item->pip, ag, ref);
+	return ag;
+}
+
+/*
+ * xfs_filestream_associate() should only be called to associate a regular file
+ * with its parent directory.  Calling it with a child directory isn't
+ * appropriate because filestreams don't apply to entire directory hierarchies.
+ * Creating a file in a child directory of an existing filestream directory
+ * starts a new filestream with its own allocation group association.
+ *
+ * Returns < 0 on error, 0 if successful association occurred, > 0 if
+ * we failed to get an association because of locking issues.
+ */
+int
+xfs_filestream_associate(
+	xfs_inode_t	*pip,
+	xfs_inode_t	*ip)
+{
+	xfs_mount_t	*mp;
+	xfs_mru_cache_t	*cache;
+	fstrm_item_t	*item;
+	xfs_agnumber_t	ag, rotorstep, startag;
+	int		err = 0;
+
+	ASSERT(pip->i_d.di_mode & S_IFDIR);
+	ASSERT(ip->i_d.di_mode & S_IFREG);
+	if (!(pip->i_d.di_mode & S_IFDIR) || !(ip->i_d.di_mode & S_IFREG))
+		return -EINVAL;
+
+	mp = pip->i_mount;
+	cache = mp->m_filestream;
+
+	/*
+	 * We have a problem, Houston.
+	 *
+	 * Taking the iolock here violates inode locking order - we already
+	 * hold the ilock. Hence if we block getting this lock we may never
+	 * wake. Unfortunately, that means if we can't get the lock, we're
+	 * screwed in terms of getting a stream association - we can't spin
+	 * waiting for the lock because someone else is waiting on the lock we
+	 * hold and we cannot drop that as we are in a transaction here.
+	 *
+	 * Lucky for us, this inversion is not a problem because it's a
+	 * directory inode that we are trying to lock here.
+	 *
+	 * So, if we can't get the iolock without sleeping then just give up
+	 */
+	if (!xfs_ilock_nowait(pip, XFS_IOLOCK_EXCL))
+		return 1;
+
+	/* If the parent directory is already in the cache, use its AG. */
+	item = xfs_mru_cache_lookup(cache, pip->i_ino);
+	if (item) {
+		ASSERT(item->ip == pip);
+		ag = item->ag;
+		xfs_mru_cache_done(cache);
+
+		TRACE_LOOKUP(mp, pip, pip, ag, xfs_filestream_peek_ag(mp, ag));
+		err = _xfs_filestream_update_ag(ip, pip, ag);
+
+		goto exit;
+	}
+
+	/*
+	 * Set the starting AG using the rotor for inode32, otherwise
+	 * use the directory inode's AG.
+	 */
+	if (mp->m_flags & XFS_MOUNT_32BITINODES) {
+		rotorstep = xfs_rotorstep;
+		startag = (mp->m_agfrotor / rotorstep) % mp->m_sb.sb_agcount;
+		mp->m_agfrotor = (mp->m_agfrotor + 1) %
+		                 (mp->m_sb.sb_agcount * rotorstep);
+	} else
+		startag = XFS_INO_TO_AGNO(mp, pip->i_ino);
+
+	/* Pick a new AG for the parent inode starting at startag. */
+	err = _xfs_filestream_pick_ag(mp, startag, &ag, 0, 0);
+	if (err || ag == NULLAGNUMBER)
+		goto exit_did_pick;
+
+	/* Associate the parent inode with the AG. */
+	err = _xfs_filestream_update_ag(pip, NULL, ag);
+	if (err)
+		goto exit_did_pick;
+
+	/* Associate the file inode with the AG. */
+	err = _xfs_filestream_update_ag(ip, pip, ag);
+	if (err)
+		goto exit_did_pick;
+
+	TRACE_ASSOCIATE(mp, ip, pip, ag, xfs_filestream_peek_ag(mp, ag));
+
+exit_did_pick:
+	/*
+	 * If _xfs_filestream_pick_ag() returned a valid AG, remove the
+	 * reference it took on it, since the file and directory will have taken
+	 * their own now if they were successfully cached.
+	 */
+	if (ag != NULLAGNUMBER)
+		xfs_filestream_put_ag(mp, ag);
+
+exit:
+	xfs_iunlock(pip, XFS_IOLOCK_EXCL);
+	return -err;
+}
+
+/*
+ * Pick a new allocation group for the current file and its file stream.  This
+ * function is called by xfs_bmap_filestreams() with the mount point's per-ag
+ * lock held.
+ */
+int
+xfs_filestream_new_ag(
+	xfs_bmalloca_t	*ap,
+	xfs_agnumber_t	*agp)
+{
+	int		flags, err;
+	xfs_inode_t	*ip, *pip = NULL;
+	xfs_mount_t	*mp;
+	xfs_mru_cache_t	*cache;
+	xfs_extlen_t	minlen;
+	fstrm_item_t	*dir, *file;
+	xfs_agnumber_t	ag = NULLAGNUMBER;
+
+	ip = ap->ip;
+	mp = ip->i_mount;
+	cache = mp->m_filestream;
+	minlen = ap->alen;
+	*agp = NULLAGNUMBER;
+
+	/*
+	 * Look for the file in the cache, removing it if it's found.  Doing
+	 * this allows it to be held across the dir lookup that follows.
+	 */
+	file = xfs_mru_cache_remove(cache, ip->i_ino);
+	if (file) {
+		ASSERT(ip == file->ip);
+
+		/* Save the file's parent inode and old AG number for later. */
+		pip = file->pip;
+		ag = file->ag;
+
+		/* Look for the file's directory in the cache. */
+		dir = xfs_mru_cache_lookup(cache, pip->i_ino);
+		if (dir) {
+			ASSERT(pip == dir->ip);
+
+			/*
+			 * If the directory has already moved on to a new AG,
+			 * use that AG as the new AG for the file. Don't
+			 * forget to twiddle the AG refcounts to match the
+			 * movement.
+			 */
+			if (dir->ag != file->ag) {
+				xfs_filestream_put_ag(mp, file->ag);
+				xfs_filestream_get_ag(mp, dir->ag);
+				*agp = file->ag = dir->ag;
+			}
+
+			xfs_mru_cache_done(cache);
+		}
+
+		/*
+		 * Put the file back in the cache.  If this fails, the free
+		 * function needs to be called to tidy up in the same way as if
+		 * the item had simply expired from the cache.
+		 */
+		err = xfs_mru_cache_insert(cache, ip->i_ino, file);
+		if (err) {
+			xfs_fstrm_free_func(ip->i_ino, file);
+			return err;
+		}
+
+		/*
+		 * If the file's AG was moved to the directory's new AG, there's
+		 * nothing more to be done.
+		 */
+		if (*agp != NULLAGNUMBER) {
+			TRACE_MOVEAG(mp, ip, pip,
+					ag, xfs_filestream_peek_ag(mp, ag),
+					*agp, xfs_filestream_peek_ag(mp, *agp));
+			return 0;
+		}
+	}
+
+	/*
+	 * If the file's parent directory is known, take its iolock in exclusive
+	 * mode to prevent two sibling files from racing each other to migrate
+	 * themselves and their parent to different AGs.
+	 *
+	 * Note that we lock the parent directory iolock inside the child
+	 * iolock here.  That's fine as we never hold both parent and child
+	 * iolock in any other place.  This is different from the ilock,
+	 * which requires locking of the child after the parent for namespace
+	 * operations.
+	 */
+	if (pip)
+		xfs_ilock(pip, XFS_IOLOCK_EXCL | XFS_IOLOCK_PARENT);
+
+	/*
+	 * A new AG needs to be found for the file.  If the file's parent
+	 * directory is also known, it will be moved to the new AG as well to
+	 * ensure that files created inside it in future use the new AG.
+	 */
+	ag = (ag == NULLAGNUMBER) ? 0 : (ag + 1) % mp->m_sb.sb_agcount;
+	flags = (ap->userdata ? XFS_PICK_USERDATA : 0) |
+	        (ap->low ? XFS_PICK_LOWSPACE : 0);
+
+	err = _xfs_filestream_pick_ag(mp, ag, agp, flags, minlen);
+	if (err || *agp == NULLAGNUMBER)
+		goto exit;
+
+	/*
+	 * If the file wasn't found in the file cache, then its parent directory
+	 * inode isn't known.  For this to have happened, the file must either
+	 * be pre-existing, or it was created long enough ago that its cache
+	 * entry has expired.  This isn't the sort of usage that the filestreams
+	 * allocator is trying to optimise, so there's no point trying to track
+	 * its new AG somehow in the filestream data structures.
+	 */
+	if (!pip) {
+		TRACE_ORPHAN(mp, ip, *agp);
+		goto exit;
+	}
+
+	/* Associate the parent inode with the AG. */
+	err = _xfs_filestream_update_ag(pip, NULL, *agp);
+	if (err)
+		goto exit;
+
+	/* Associate the file inode with the AG. */
+	err = _xfs_filestream_update_ag(ip, pip, *agp);
+	if (err)
+		goto exit;
+
+	TRACE_MOVEAG(mp, ip, pip, NULLAGNUMBER, 0,
+			*agp, xfs_filestream_peek_ag(mp, *agp));
+
+exit:
+	/*
+	 * If _xfs_filestream_pick_ag() returned a valid AG, remove the
+	 * reference it took on it, since the file and directory will have taken
+	 * their own now if they were successfully cached.
+	 */
+	if (*agp != NULLAGNUMBER)
+		xfs_filestream_put_ag(mp, *agp);
+	else
+		*agp = 0;
+
+	if (pip)
+		xfs_iunlock(pip, XFS_IOLOCK_EXCL);
+
+	return err;
+}
+
+/*
+ * Remove an association between an inode and a filestream object.
+ * Typically this is done on last close of an unlinked file.
+ */
+void
+xfs_filestream_deassociate(
+	xfs_inode_t	*ip)
+{
+	xfs_mru_cache_t	*cache = ip->i_mount->m_filestream;
+
+	xfs_mru_cache_delete(cache, ip->i_ino);
+}
diff --git a/fs/xfs/xfs_filestream.h b/fs/xfs/xfs_filestream.h
new file mode 100644
index 00000000..09dd9af4
--- /dev/null
+++ b/fs/xfs/xfs_filestream.h
@@ -0,0 +1,74 @@
+/*
+ * Copyright (c) 2006-2007 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_FILESTREAM_H__
+#define __XFS_FILESTREAM_H__
+
+#ifdef __KERNEL__
+
+struct xfs_mount;
+struct xfs_inode;
+struct xfs_perag;
+struct xfs_bmalloca;
+
+#ifdef XFS_FILESTREAMS_TRACE
+#define XFS_FSTRM_KTRACE_INFO		1
+#define XFS_FSTRM_KTRACE_AGSCAN		2
+#define XFS_FSTRM_KTRACE_AGPICK1	3
+#define XFS_FSTRM_KTRACE_AGPICK2	4
+#define XFS_FSTRM_KTRACE_UPDATE		5
+#define XFS_FSTRM_KTRACE_FREE		6
+#define	XFS_FSTRM_KTRACE_ITEM_LOOKUP	7
+#define	XFS_FSTRM_KTRACE_ASSOCIATE	8
+#define	XFS_FSTRM_KTRACE_MOVEAG		9
+#define	XFS_FSTRM_KTRACE_ORPHAN		10
+
+#define XFS_FSTRM_KTRACE_SIZE	16384
+extern ktrace_t *xfs_filestreams_trace_buf;
+
+#endif
+
+/* allocation selection flags */
+typedef enum xfs_fstrm_alloc {
+	XFS_PICK_USERDATA = 1,
+	XFS_PICK_LOWSPACE = 2,
+} xfs_fstrm_alloc_t;
+
+/* prototypes for filestream.c */
+int xfs_filestream_init(void);
+void xfs_filestream_uninit(void);
+int xfs_filestream_mount(struct xfs_mount *mp);
+void xfs_filestream_unmount(struct xfs_mount *mp);
+xfs_agnumber_t xfs_filestream_lookup_ag(struct xfs_inode *ip);
+int xfs_filestream_associate(struct xfs_inode *dip, struct xfs_inode *ip);
+void xfs_filestream_deassociate(struct xfs_inode *ip);
+int xfs_filestream_new_ag(struct xfs_bmalloca *ap, xfs_agnumber_t *agp);
+
+
+/* filestreams for the inode? */
+static inline int
+xfs_inode_is_filestream(
+	struct xfs_inode	*ip)
+{
+	return (ip->i_mount->m_flags & XFS_MOUNT_FILESTREAMS) ||
+		xfs_iflags_test(ip, XFS_IFILESTREAM) ||
+		(ip->i_d.di_flags & XFS_DIFLAG_FILESTREAM);
+}
+
+#endif /* __KERNEL__ */
+
+#endif /* __XFS_FILESTREAM_H__ */
diff --git a/fs/xfs/xfs_fs.h b/fs/xfs/xfs_fs.h
new file mode 100644
index 00000000..8f6fc1a9
--- /dev/null
+++ b/fs/xfs/xfs_fs.h
@@ -0,0 +1,501 @@
+/*
+ * Copyright (c) 1995-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public License
+ * as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_FS_H__
+#define __XFS_FS_H__
+
+/*
+ * SGI's XFS filesystem's major stuff (constants, structures)
+ */
+
+/*
+ * Direct I/O attribute record used with XFS_IOC_DIOINFO
+ * d_miniosz is the min xfer size, xfer size multiple and file seek offset
+ * alignment.
+ */
+#ifndef HAVE_DIOATTR
+struct dioattr {
+	__u32		d_mem;		/* data buffer memory alignment */
+	__u32		d_miniosz;	/* min xfer size		*/
+	__u32		d_maxiosz;	/* max xfer size		*/
+};
+#endif
+
+/*
+ * Structure for XFS_IOC_FSGETXATTR[A] and XFS_IOC_FSSETXATTR.
+ */
+#ifndef HAVE_FSXATTR
+struct fsxattr {
+	__u32		fsx_xflags;	/* xflags field value (get/set) */
+	__u32		fsx_extsize;	/* extsize field value (get/set)*/
+	__u32		fsx_nextents;	/* nextents field value (get)	*/
+	__u32		fsx_projid;	/* project identifier (get/set) */
+	unsigned char	fsx_pad[12];
+};
+#endif
+
+/*
+ * Flags for the bs_xflags/fsx_xflags field
+ * There should be a one-to-one correspondence between these flags and the
+ * XFS_DIFLAG_s.
+ */
+#define XFS_XFLAG_REALTIME	0x00000001	/* data in realtime volume */
+#define XFS_XFLAG_PREALLOC	0x00000002	/* preallocated file extents */
+#define XFS_XFLAG_IMMUTABLE	0x00000008	/* file cannot be modified */
+#define XFS_XFLAG_APPEND	0x00000010	/* all writes append */
+#define XFS_XFLAG_SYNC		0x00000020	/* all writes synchronous */
+#define XFS_XFLAG_NOATIME	0x00000040	/* do not update access time */
+#define XFS_XFLAG_NODUMP	0x00000080	/* do not include in backups */
+#define XFS_XFLAG_RTINHERIT	0x00000100	/* create with rt bit set */
+#define XFS_XFLAG_PROJINHERIT	0x00000200	/* create with parents projid */
+#define XFS_XFLAG_NOSYMLINKS	0x00000400	/* disallow symlink creation */
+#define XFS_XFLAG_EXTSIZE	0x00000800	/* extent size allocator hint */
+#define XFS_XFLAG_EXTSZINHERIT	0x00001000	/* inherit inode extent size */
+#define XFS_XFLAG_NODEFRAG	0x00002000  	/* do not defragment */
+#define XFS_XFLAG_FILESTREAM	0x00004000	/* use filestream allocator */
+#define XFS_XFLAG_HASATTR	0x80000000	/* no DIFLAG for this	*/
+
+/*
+ * Structure for XFS_IOC_GETBMAP.
+ * On input, fill in bmv_offset and bmv_length of the first structure
+ * to indicate the area of interest in the file, and bmv_entries with
+ * the number of array elements given back.  The first structure is
+ * updated on return to give the offset and length for the next call.
+ */
+#ifndef HAVE_GETBMAP
+struct getbmap {
+	__s64		bmv_offset;	/* file offset of segment in blocks */
+	__s64		bmv_block;	/* starting block (64-bit daddr_t)  */
+	__s64		bmv_length;	/* length of segment, blocks	    */
+	__s32		bmv_count;	/* # of entries in array incl. 1st  */
+	__s32		bmv_entries;	/* # of entries filled in (output)  */
+};
+#endif
+
+/*
+ *	Structure for XFS_IOC_GETBMAPX.	 Fields bmv_offset through bmv_entries
+ *	are used exactly as in the getbmap structure.  The getbmapx structure
+ *	has additional bmv_iflags and bmv_oflags fields. The bmv_iflags field
+ *	is only used for the first structure.  It contains input flags
+ *	specifying XFS_IOC_GETBMAPX actions.  The bmv_oflags field is filled
+ *	in by the XFS_IOC_GETBMAPX command for each returned structure after
+ *	the first.
+ */
+#ifndef HAVE_GETBMAPX
+struct getbmapx {
+	__s64		bmv_offset;	/* file offset of segment in blocks */
+	__s64		bmv_block;	/* starting block (64-bit daddr_t)  */
+	__s64		bmv_length;	/* length of segment, blocks	    */
+	__s32		bmv_count;	/* # of entries in array incl. 1st  */
+	__s32		bmv_entries;	/* # of entries filled in (output). */
+	__s32		bmv_iflags;	/* input flags (1st structure)	    */
+	__s32		bmv_oflags;	/* output flags (after 1st structure)*/
+	__s32		bmv_unused1;	/* future use			    */
+	__s32		bmv_unused2;	/* future use			    */
+};
+#endif
+
+/*	bmv_iflags values - set by XFS_IOC_GETBMAPX caller.	*/
+#define BMV_IF_ATTRFORK		0x1	/* return attr fork rather than data */
+#define BMV_IF_NO_DMAPI_READ	0x2	/* Do not generate DMAPI read event  */
+#define BMV_IF_PREALLOC		0x4	/* rtn status BMV_OF_PREALLOC if req */
+#define BMV_IF_DELALLOC		0x8	/* rtn status BMV_OF_DELALLOC if req */
+#define BMV_IF_NO_HOLES		0x10	/* Do not return holes */
+#define BMV_IF_VALID	\
+	(BMV_IF_ATTRFORK|BMV_IF_NO_DMAPI_READ|BMV_IF_PREALLOC|	\
+	 BMV_IF_DELALLOC|BMV_IF_NO_HOLES)
+
+/*	bmv_oflags values - returned for each non-header segment */
+#define BMV_OF_PREALLOC		0x1	/* segment = unwritten pre-allocation */
+#define BMV_OF_DELALLOC		0x2	/* segment = delayed allocation */
+#define BMV_OF_LAST		0x4	/* segment is the last in the file */
+
+/*
+ * Structure for XFS_IOC_FSSETDM.
+ * For use by backup and restore programs to set the XFS on-disk inode
+ * fields di_dmevmask and di_dmstate.  These must be set to exactly and
+ * only values previously obtained via xfs_bulkstat!  (Specifically the
+ * xfs_bstat_t fields bs_dmevmask and bs_dmstate.)
+ */
+#ifndef HAVE_FSDMIDATA
+struct fsdmidata {
+	__u32		fsd_dmevmask;	/* corresponds to di_dmevmask */
+	__u16		fsd_padding;
+	__u16		fsd_dmstate;	/* corresponds to di_dmstate  */
+};
+#endif
+
+/*
+ * File segment locking set data type for 64 bit access.
+ * Also used for all the RESV/FREE interfaces.
+ */
+typedef struct xfs_flock64 {
+	__s16		l_type;
+	__s16		l_whence;
+	__s64		l_start;
+	__s64		l_len;		/* len == 0 means until end of file */
+	__s32		l_sysid;
+	__u32		l_pid;
+	__s32		l_pad[4];	/* reserve area			    */
+} xfs_flock64_t;
+
+/*
+ * Output for XFS_IOC_FSGEOMETRY_V1
+ */
+typedef struct xfs_fsop_geom_v1 {
+	__u32		blocksize;	/* filesystem (data) block size */
+	__u32		rtextsize;	/* realtime extent size		*/
+	__u32		agblocks;	/* fsblocks in an AG		*/
+	__u32		agcount;	/* number of allocation groups	*/
+	__u32		logblocks;	/* fsblocks in the log		*/
+	__u32		sectsize;	/* (data) sector size, bytes	*/
+	__u32		inodesize;	/* inode size in bytes		*/
+	__u32		imaxpct;	/* max allowed inode space(%)	*/
+	__u64		datablocks;	/* fsblocks in data subvolume	*/
+	__u64		rtblocks;	/* fsblocks in realtime subvol	*/
+	__u64		rtextents;	/* rt extents in realtime subvol*/
+	__u64		logstart;	/* starting fsblock of the log	*/
+	unsigned char	uuid[16];	/* unique id of the filesystem	*/
+	__u32		sunit;		/* stripe unit, fsblocks	*/
+	__u32		swidth;		/* stripe width, fsblocks	*/
+	__s32		version;	/* structure version		*/
+	__u32		flags;		/* superblock version flags	*/
+	__u32		logsectsize;	/* log sector size, bytes	*/
+	__u32		rtsectsize;	/* realtime sector size, bytes	*/
+	__u32		dirblocksize;	/* directory block size, bytes	*/
+} xfs_fsop_geom_v1_t;
+
+/*
+ * Output for XFS_IOC_FSGEOMETRY
+ */
+typedef struct xfs_fsop_geom {
+	__u32		blocksize;	/* filesystem (data) block size */
+	__u32		rtextsize;	/* realtime extent size		*/
+	__u32		agblocks;	/* fsblocks in an AG		*/
+	__u32		agcount;	/* number of allocation groups	*/
+	__u32		logblocks;	/* fsblocks in the log		*/
+	__u32		sectsize;	/* (data) sector size, bytes	*/
+	__u32		inodesize;	/* inode size in bytes		*/
+	__u32		imaxpct;	/* max allowed inode space(%)	*/
+	__u64		datablocks;	/* fsblocks in data subvolume	*/
+	__u64		rtblocks;	/* fsblocks in realtime subvol	*/
+	__u64		rtextents;	/* rt extents in realtime subvol*/
+	__u64		logstart;	/* starting fsblock of the log	*/
+	unsigned char	uuid[16];	/* unique id of the filesystem	*/
+	__u32		sunit;		/* stripe unit, fsblocks	*/
+	__u32		swidth;		/* stripe width, fsblocks	*/
+	__s32		version;	/* structure version		*/
+	__u32		flags;		/* superblock version flags	*/
+	__u32		logsectsize;	/* log sector size, bytes	*/
+	__u32		rtsectsize;	/* realtime sector size, bytes	*/
+	__u32		dirblocksize;	/* directory block size, bytes	*/
+	__u32		logsunit;	/* log stripe unit, bytes */
+} xfs_fsop_geom_t;
+
+/* Output for XFS_FS_COUNTS */
+typedef struct xfs_fsop_counts {
+	__u64	freedata;	/* free data section blocks */
+	__u64	freertx;	/* free rt extents */
+	__u64	freeino;	/* free inodes */
+	__u64	allocino;	/* total allocated inodes */
+} xfs_fsop_counts_t;
+
+/* Input/Output for XFS_GET_RESBLKS and XFS_SET_RESBLKS */
+typedef struct xfs_fsop_resblks {
+	__u64  resblks;
+	__u64  resblks_avail;
+} xfs_fsop_resblks_t;
+
+#define XFS_FSOP_GEOM_VERSION	0
+
+#define XFS_FSOP_GEOM_FLAGS_ATTR	0x0001	/* attributes in use	*/
+#define XFS_FSOP_GEOM_FLAGS_NLINK	0x0002	/* 32-bit nlink values	*/
+#define XFS_FSOP_GEOM_FLAGS_QUOTA	0x0004	/* quotas enabled	*/
+#define XFS_FSOP_GEOM_FLAGS_IALIGN	0x0008	/* inode alignment	*/
+#define XFS_FSOP_GEOM_FLAGS_DALIGN	0x0010	/* large data alignment */
+#define XFS_FSOP_GEOM_FLAGS_SHARED	0x0020	/* read-only shared	*/
+#define XFS_FSOP_GEOM_FLAGS_EXTFLG	0x0040	/* special extent flag	*/
+#define XFS_FSOP_GEOM_FLAGS_DIRV2	0x0080	/* directory version 2	*/
+#define XFS_FSOP_GEOM_FLAGS_LOGV2	0x0100	/* log format version 2	*/
+#define XFS_FSOP_GEOM_FLAGS_SECTOR	0x0200	/* sector sizes >1BB	*/
+#define XFS_FSOP_GEOM_FLAGS_ATTR2	0x0400	/* inline attributes rework */
+#define XFS_FSOP_GEOM_FLAGS_DIRV2CI	0x1000	/* ASCII only CI names */
+#define XFS_FSOP_GEOM_FLAGS_LAZYSB	0x4000	/* lazy superblock counters */
+
+
+/*
+ * Minimum and maximum sizes need for growth checks
+ */
+#define XFS_MIN_AG_BLOCKS	64
+#define XFS_MIN_LOG_BLOCKS	512ULL
+#define XFS_MAX_LOG_BLOCKS	(1024 * 1024ULL)
+#define XFS_MIN_LOG_BYTES	(10 * 1024 * 1024ULL)
+
+/* keep the maximum size under 2^31 by a small amount */
+#define XFS_MAX_LOG_BYTES \
+	((2 * 1024 * 1024 * 1024ULL) - XFS_MIN_LOG_BYTES)
+
+/*
+ * Structures for XFS_IOC_FSGROWFSDATA, XFS_IOC_FSGROWFSLOG & XFS_IOC_FSGROWFSRT
+ */
+typedef struct xfs_growfs_data {
+	__u64		newblocks;	/* new data subvol size, fsblocks */
+	__u32		imaxpct;	/* new inode space percentage limit */
+} xfs_growfs_data_t;
+
+typedef struct xfs_growfs_log {
+	__u32		newblocks;	/* new log size, fsblocks */
+	__u32		isint;		/* 1 if new log is internal */
+} xfs_growfs_log_t;
+
+typedef struct xfs_growfs_rt {
+	__u64		newblocks;	/* new realtime size, fsblocks */
+	__u32		extsize;	/* new realtime extent size, fsblocks */
+} xfs_growfs_rt_t;
+
+
+/*
+ * Structures returned from ioctl XFS_IOC_FSBULKSTAT & XFS_IOC_FSBULKSTAT_SINGLE
+ */
+typedef struct xfs_bstime {
+	time_t		tv_sec;		/* seconds		*/
+	__s32		tv_nsec;	/* and nanoseconds	*/
+} xfs_bstime_t;
+
+typedef struct xfs_bstat {
+	__u64		bs_ino;		/* inode number			*/
+	__u16		bs_mode;	/* type and mode		*/
+	__u16		bs_nlink;	/* number of links		*/
+	__u32		bs_uid;		/* user id			*/
+	__u32		bs_gid;		/* group id			*/
+	__u32		bs_rdev;	/* device value			*/
+	__s32		bs_blksize;	/* block size			*/
+	__s64		bs_size;	/* file size			*/
+	xfs_bstime_t	bs_atime;	/* access time			*/
+	xfs_bstime_t	bs_mtime;	/* modify time			*/
+	xfs_bstime_t	bs_ctime;	/* inode change time		*/
+	int64_t		bs_blocks;	/* number of blocks		*/
+	__u32		bs_xflags;	/* extended flags		*/
+	__s32		bs_extsize;	/* extent size			*/
+	__s32		bs_extents;	/* number of extents		*/
+	__u32		bs_gen;		/* generation count		*/
+	__u16		bs_projid_lo;	/* lower part of project id	*/
+#define	bs_projid	bs_projid_lo	/* (previously just bs_projid)	*/
+	__u16		bs_forkoff;	/* inode fork offset in bytes	*/
+	__u16		bs_projid_hi;	/* higher part of project id	*/
+	unsigned char	bs_pad[10];	/* pad space, unused		*/
+	__u32		bs_dmevmask;	/* DMIG event mask		*/
+	__u16		bs_dmstate;	/* DMIG state info		*/
+	__u16		bs_aextents;	/* attribute number of extents	*/
+} xfs_bstat_t;
+
+/*
+ * The user-level BulkStat Request interface structure.
+ */
+typedef struct xfs_fsop_bulkreq {
+	__u64		__user *lastip;	/* last inode # pointer		*/
+	__s32		icount;		/* count of entries in buffer	*/
+	void		__user *ubuffer;/* user buffer for inode desc.	*/
+	__s32		__user *ocount;	/* output count pointer		*/
+} xfs_fsop_bulkreq_t;
+
+
+/*
+ * Structures returned from xfs_inumbers routine (XFS_IOC_FSINUMBERS).
+ */
+typedef struct xfs_inogrp {
+	__u64		xi_startino;	/* starting inode number	*/
+	__s32		xi_alloccount;	/* # bits set in allocmask	*/
+	__u64		xi_allocmask;	/* mask of allocated inodes	*/
+} xfs_inogrp_t;
+
+
+/*
+ * Error injection.
+ */
+typedef struct xfs_error_injection {
+	__s32		fd;
+	__s32		errtag;
+} xfs_error_injection_t;
+
+
+/*
+ * The user-level Handle Request interface structure.
+ */
+typedef struct xfs_fsop_handlereq {
+	__u32		fd;		/* fd for FD_TO_HANDLE		*/
+	void		__user *path;	/* user pathname		*/
+	__u32		oflags;		/* open flags			*/
+	void		__user *ihandle;/* user supplied handle		*/
+	__u32		ihandlen;	/* user supplied length		*/
+	void		__user *ohandle;/* user buffer for handle	*/
+	__u32		__user *ohandlen;/* user buffer length		*/
+} xfs_fsop_handlereq_t;
+
+/*
+ * Compound structures for passing args through Handle Request interfaces
+ * xfs_fssetdm_by_handle, xfs_attrlist_by_handle, xfs_attrmulti_by_handle
+ * - ioctls: XFS_IOC_FSSETDM_BY_HANDLE, XFS_IOC_ATTRLIST_BY_HANDLE, and
+ *	     XFS_IOC_ATTRMULTI_BY_HANDLE
+ */
+
+typedef struct xfs_fsop_setdm_handlereq {
+	struct xfs_fsop_handlereq	hreq;	/* handle information	*/
+	struct fsdmidata		__user *data;	/* DMAPI data	*/
+} xfs_fsop_setdm_handlereq_t;
+
+typedef struct xfs_attrlist_cursor {
+	__u32		opaque[4];
+} xfs_attrlist_cursor_t;
+
+typedef struct xfs_fsop_attrlist_handlereq {
+	struct xfs_fsop_handlereq	hreq; /* handle interface structure */
+	struct xfs_attrlist_cursor	pos; /* opaque cookie, list offset */
+	__u32				flags;	/* which namespace to use */
+	__u32				buflen;	/* length of buffer supplied */
+	void				__user *buffer;	/* returned names */
+} xfs_fsop_attrlist_handlereq_t;
+
+typedef struct xfs_attr_multiop {
+	__u32		am_opcode;
+#define ATTR_OP_GET	1	/* return the indicated attr's value */
+#define ATTR_OP_SET	2	/* set/create the indicated attr/value pair */
+#define ATTR_OP_REMOVE	3	/* remove the indicated attr */
+	__s32		am_error;
+	void		__user *am_attrname;
+	void		__user *am_attrvalue;
+	__u32		am_length;
+	__u32		am_flags;
+} xfs_attr_multiop_t;
+
+typedef struct xfs_fsop_attrmulti_handlereq {
+	struct xfs_fsop_handlereq	hreq; /* handle interface structure */
+	__u32				opcount;/* count of following multiop */
+	struct xfs_attr_multiop		__user *ops; /* attr_multi data */
+} xfs_fsop_attrmulti_handlereq_t;
+
+/*
+ * per machine unique filesystem identifier types.
+ */
+typedef struct { __u32 val[2]; } xfs_fsid_t; /* file system id type */
+
+typedef struct xfs_fid {
+	__u16	fid_len;		/* length of remainder	*/
+	__u16	fid_pad;
+	__u32	fid_gen;		/* generation number	*/
+	__u64	fid_ino;		/* 64 bits inode number */
+} xfs_fid_t;
+
+typedef struct xfs_handle {
+	union {
+		__s64	    align;	/* force alignment of ha_fid	 */
+		xfs_fsid_t  _ha_fsid;	/* unique file system identifier */
+	} ha_u;
+	xfs_fid_t	ha_fid;		/* file system specific file ID	 */
+} xfs_handle_t;
+#define ha_fsid ha_u._ha_fsid
+
+#define XFS_HSIZE(handle)	(((char *) &(handle).ha_fid.fid_pad	 \
+				 - (char *) &(handle))			  \
+				 + (handle).ha_fid.fid_len)
+
+/*
+ * Flags for going down operation
+ */
+#define XFS_FSOP_GOING_FLAGS_DEFAULT		0x0	/* going down */
+#define XFS_FSOP_GOING_FLAGS_LOGFLUSH		0x1	/* flush log but not data */
+#define XFS_FSOP_GOING_FLAGS_NOLOGFLUSH		0x2	/* don't flush log nor data */
+
+/*
+ * ioctl commands that are used by Linux filesystems
+ */
+#define XFS_IOC_GETXFLAGS	FS_IOC_GETFLAGS
+#define XFS_IOC_SETXFLAGS	FS_IOC_SETFLAGS
+#define XFS_IOC_GETVERSION	FS_IOC_GETVERSION
+
+/*
+ * ioctl commands that replace IRIX fcntl()'s
+ * For 'documentation' purposed more than anything else,
+ * the "cmd #" field reflects the IRIX fcntl number.
+ */
+#define XFS_IOC_ALLOCSP		_IOW ('X', 10, struct xfs_flock64)
+#define XFS_IOC_FREESP		_IOW ('X', 11, struct xfs_flock64)
+#define XFS_IOC_DIOINFO		_IOR ('X', 30, struct dioattr)
+#define XFS_IOC_FSGETXATTR	_IOR ('X', 31, struct fsxattr)
+#define XFS_IOC_FSSETXATTR	_IOW ('X', 32, struct fsxattr)
+#define XFS_IOC_ALLOCSP64	_IOW ('X', 36, struct xfs_flock64)
+#define XFS_IOC_FREESP64	_IOW ('X', 37, struct xfs_flock64)
+#define XFS_IOC_GETBMAP		_IOWR('X', 38, struct getbmap)
+#define XFS_IOC_FSSETDM		_IOW ('X', 39, struct fsdmidata)
+#define XFS_IOC_RESVSP		_IOW ('X', 40, struct xfs_flock64)
+#define XFS_IOC_UNRESVSP	_IOW ('X', 41, struct xfs_flock64)
+#define XFS_IOC_RESVSP64	_IOW ('X', 42, struct xfs_flock64)
+#define XFS_IOC_UNRESVSP64	_IOW ('X', 43, struct xfs_flock64)
+#define XFS_IOC_GETBMAPA	_IOWR('X', 44, struct getbmap)
+#define XFS_IOC_FSGETXATTRA	_IOR ('X', 45, struct fsxattr)
+/*	XFS_IOC_SETBIOSIZE ---- deprecated 46	   */
+/*	XFS_IOC_GETBIOSIZE ---- deprecated 47	   */
+#define XFS_IOC_GETBMAPX	_IOWR('X', 56, struct getbmap)
+#define XFS_IOC_ZERO_RANGE	_IOW ('X', 57, struct xfs_flock64)
+
+/*
+ * ioctl commands that replace IRIX syssgi()'s
+ */
+#define XFS_IOC_FSGEOMETRY_V1	     _IOR ('X', 100, struct xfs_fsop_geom_v1)
+#define XFS_IOC_FSBULKSTAT	     _IOWR('X', 101, struct xfs_fsop_bulkreq)
+#define XFS_IOC_FSBULKSTAT_SINGLE    _IOWR('X', 102, struct xfs_fsop_bulkreq)
+#define XFS_IOC_FSINUMBERS	     _IOWR('X', 103, struct xfs_fsop_bulkreq)
+#define XFS_IOC_PATH_TO_FSHANDLE     _IOWR('X', 104, struct xfs_fsop_handlereq)
+#define XFS_IOC_PATH_TO_HANDLE	     _IOWR('X', 105, struct xfs_fsop_handlereq)
+#define XFS_IOC_FD_TO_HANDLE	     _IOWR('X', 106, struct xfs_fsop_handlereq)
+#define XFS_IOC_OPEN_BY_HANDLE	     _IOWR('X', 107, struct xfs_fsop_handlereq)
+#define XFS_IOC_READLINK_BY_HANDLE   _IOWR('X', 108, struct xfs_fsop_handlereq)
+#define XFS_IOC_SWAPEXT		     _IOWR('X', 109, struct xfs_swapext)
+#define XFS_IOC_FSGROWFSDATA	     _IOW ('X', 110, struct xfs_growfs_data)
+#define XFS_IOC_FSGROWFSLOG	     _IOW ('X', 111, struct xfs_growfs_log)
+#define XFS_IOC_FSGROWFSRT	     _IOW ('X', 112, struct xfs_growfs_rt)
+#define XFS_IOC_FSCOUNTS	     _IOR ('X', 113, struct xfs_fsop_counts)
+#define XFS_IOC_SET_RESBLKS	     _IOWR('X', 114, struct xfs_fsop_resblks)
+#define XFS_IOC_GET_RESBLKS	     _IOR ('X', 115, struct xfs_fsop_resblks)
+#define XFS_IOC_ERROR_INJECTION	     _IOW ('X', 116, struct xfs_error_injection)
+#define XFS_IOC_ERROR_CLEARALL	     _IOW ('X', 117, struct xfs_error_injection)
+/*	XFS_IOC_ATTRCTL_BY_HANDLE -- deprecated 118	 */
+/*	XFS_IOC_FREEZE		  -- FIFREEZE   119	 */
+/*	XFS_IOC_THAW		  -- FITHAW     120	 */
+#define XFS_IOC_FSSETDM_BY_HANDLE    _IOW ('X', 121, struct xfs_fsop_setdm_handlereq)
+#define XFS_IOC_ATTRLIST_BY_HANDLE   _IOW ('X', 122, struct xfs_fsop_attrlist_handlereq)
+#define XFS_IOC_ATTRMULTI_BY_HANDLE  _IOW ('X', 123, struct xfs_fsop_attrmulti_handlereq)
+#define XFS_IOC_FSGEOMETRY	     _IOR ('X', 124, struct xfs_fsop_geom)
+#define XFS_IOC_GOINGDOWN	     _IOR ('X', 125, __uint32_t)
+/*	XFS_IOC_GETFSUUID ---------- deprecated 140	 */
+
+
+#ifndef HAVE_BBMACROS
+/*
+ * Block I/O parameterization.	A basic block (BB) is the lowest size of
+ * filesystem allocation, and must equal 512.  Length units given to bio
+ * routines are in BB's.
+ */
+#define BBSHIFT		9
+#define BBSIZE		(1<<BBSHIFT)
+#define BBMASK		(BBSIZE-1)
+#define BTOBB(bytes)	(((__u64)(bytes) + BBSIZE - 1) >> BBSHIFT)
+#define BTOBBT(bytes)	((__u64)(bytes) >> BBSHIFT)
+#define BBTOB(bbs)	((bbs) << BBSHIFT)
+#endif
+
+#endif	/* __XFS_FS_H__ */
diff --git a/fs/xfs/xfs_fsops.c b/fs/xfs/xfs_fsops.c
new file mode 100644
index 00000000..9153d2c7
--- /dev/null
+++ b/fs/xfs/xfs_fsops.c
@@ -0,0 +1,671 @@
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_bit.h"
+#include "xfs_inum.h"
+#include "xfs_log.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_mount.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_alloc_btree.h"
+#include "xfs_ialloc_btree.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_inode_item.h"
+#include "xfs_btree.h"
+#include "xfs_error.h"
+#include "xfs_alloc.h"
+#include "xfs_ialloc.h"
+#include "xfs_fsops.h"
+#include "xfs_itable.h"
+#include "xfs_trans_space.h"
+#include "xfs_rtalloc.h"
+#include "xfs_rw.h"
+#include "xfs_filestream.h"
+#include "xfs_trace.h"
+
+/*
+ * File system operations
+ */
+
+int
+xfs_fs_geometry(
+	xfs_mount_t		*mp,
+	xfs_fsop_geom_t		*geo,
+	int			new_version)
+{
+
+	memset(geo, 0, sizeof(*geo));
+
+	geo->blocksize = mp->m_sb.sb_blocksize;
+	geo->rtextsize = mp->m_sb.sb_rextsize;
+	geo->agblocks = mp->m_sb.sb_agblocks;
+	geo->agcount = mp->m_sb.sb_agcount;
+	geo->logblocks = mp->m_sb.sb_logblocks;
+	geo->sectsize = mp->m_sb.sb_sectsize;
+	geo->inodesize = mp->m_sb.sb_inodesize;
+	geo->imaxpct = mp->m_sb.sb_imax_pct;
+	geo->datablocks = mp->m_sb.sb_dblocks;
+	geo->rtblocks = mp->m_sb.sb_rblocks;
+	geo->rtextents = mp->m_sb.sb_rextents;
+	geo->logstart = mp->m_sb.sb_logstart;
+	ASSERT(sizeof(geo->uuid)==sizeof(mp->m_sb.sb_uuid));
+	memcpy(geo->uuid, &mp->m_sb.sb_uuid, sizeof(mp->m_sb.sb_uuid));
+	if (new_version >= 2) {
+		geo->sunit = mp->m_sb.sb_unit;
+		geo->swidth = mp->m_sb.sb_width;
+	}
+	if (new_version >= 3) {
+		geo->version = XFS_FSOP_GEOM_VERSION;
+		geo->flags =
+			(xfs_sb_version_hasattr(&mp->m_sb) ?
+				XFS_FSOP_GEOM_FLAGS_ATTR : 0) |
+			(xfs_sb_version_hasnlink(&mp->m_sb) ?
+				XFS_FSOP_GEOM_FLAGS_NLINK : 0) |
+			(xfs_sb_version_hasquota(&mp->m_sb) ?
+				XFS_FSOP_GEOM_FLAGS_QUOTA : 0) |
+			(xfs_sb_version_hasalign(&mp->m_sb) ?
+				XFS_FSOP_GEOM_FLAGS_IALIGN : 0) |
+			(xfs_sb_version_hasdalign(&mp->m_sb) ?
+				XFS_FSOP_GEOM_FLAGS_DALIGN : 0) |
+			(xfs_sb_version_hasshared(&mp->m_sb) ?
+				XFS_FSOP_GEOM_FLAGS_SHARED : 0) |
+			(xfs_sb_version_hasextflgbit(&mp->m_sb) ?
+				XFS_FSOP_GEOM_FLAGS_EXTFLG : 0) |
+			(xfs_sb_version_hasdirv2(&mp->m_sb) ?
+				XFS_FSOP_GEOM_FLAGS_DIRV2 : 0) |
+			(xfs_sb_version_hassector(&mp->m_sb) ?
+				XFS_FSOP_GEOM_FLAGS_SECTOR : 0) |
+			(xfs_sb_version_hasasciici(&mp->m_sb) ?
+				XFS_FSOP_GEOM_FLAGS_DIRV2CI : 0) |
+			(xfs_sb_version_haslazysbcount(&mp->m_sb) ?
+				XFS_FSOP_GEOM_FLAGS_LAZYSB : 0) |
+			(xfs_sb_version_hasattr2(&mp->m_sb) ?
+				XFS_FSOP_GEOM_FLAGS_ATTR2 : 0);
+		geo->logsectsize = xfs_sb_version_hassector(&mp->m_sb) ?
+				mp->m_sb.sb_logsectsize : BBSIZE;
+		geo->rtsectsize = mp->m_sb.sb_blocksize;
+		geo->dirblocksize = mp->m_dirblksize;
+	}
+	if (new_version >= 4) {
+		geo->flags |=
+			(xfs_sb_version_haslogv2(&mp->m_sb) ?
+				XFS_FSOP_GEOM_FLAGS_LOGV2 : 0);
+		geo->logsunit = mp->m_sb.sb_logsunit;
+	}
+	return 0;
+}
+
+static int
+xfs_growfs_data_private(
+	xfs_mount_t		*mp,		/* mount point for filesystem */
+	xfs_growfs_data_t	*in)		/* growfs data input struct */
+{
+	xfs_agf_t		*agf;
+	xfs_agi_t		*agi;
+	xfs_agnumber_t		agno;
+	xfs_extlen_t		agsize;
+	xfs_extlen_t		tmpsize;
+	xfs_alloc_rec_t		*arec;
+	struct xfs_btree_block	*block;
+	xfs_buf_t		*bp;
+	int			bucket;
+	int			dpct;
+	int			error;
+	xfs_agnumber_t		nagcount;
+	xfs_agnumber_t		nagimax = 0;
+	xfs_rfsblock_t		nb, nb_mod;
+	xfs_rfsblock_t		new;
+	xfs_rfsblock_t		nfree;
+	xfs_agnumber_t		oagcount;
+	int			pct;
+	xfs_trans_t		*tp;
+
+	nb = in->newblocks;
+	pct = in->imaxpct;
+	if (nb < mp->m_sb.sb_dblocks || pct < 0 || pct > 100)
+		return XFS_ERROR(EINVAL);
+	if ((error = xfs_sb_validate_fsb_count(&mp->m_sb, nb)))
+		return error;
+	dpct = pct - mp->m_sb.sb_imax_pct;
+	bp = xfs_buf_read_uncached(mp, mp->m_ddev_targp,
+				XFS_FSB_TO_BB(mp, nb) - XFS_FSS_TO_BB(mp, 1),
+				BBTOB(XFS_FSS_TO_BB(mp, 1)), 0);
+	if (!bp)
+		return EIO;
+	xfs_buf_relse(bp);
+
+	new = nb;	/* use new as a temporary here */
+	nb_mod = do_div(new, mp->m_sb.sb_agblocks);
+	nagcount = new + (nb_mod != 0);
+	if (nb_mod && nb_mod < XFS_MIN_AG_BLOCKS) {
+		nagcount--;
+		nb = (xfs_rfsblock_t)nagcount * mp->m_sb.sb_agblocks;
+		if (nb < mp->m_sb.sb_dblocks)
+			return XFS_ERROR(EINVAL);
+	}
+	new = nb - mp->m_sb.sb_dblocks;
+	oagcount = mp->m_sb.sb_agcount;
+
+	/* allocate the new per-ag structures */
+	if (nagcount > oagcount) {
+		error = xfs_initialize_perag(mp, nagcount, &nagimax);
+		if (error)
+			return error;
+	}
+
+	tp = xfs_trans_alloc(mp, XFS_TRANS_GROWFS);
+	tp->t_flags |= XFS_TRANS_RESERVE;
+	if ((error = xfs_trans_reserve(tp, XFS_GROWFS_SPACE_RES(mp),
+			XFS_GROWDATA_LOG_RES(mp), 0, 0, 0))) {
+		xfs_trans_cancel(tp, 0);
+		return error;
+	}
+
+	/*
+	 * Write new AG headers to disk. Non-transactional, but written
+	 * synchronously so they are completed prior to the growfs transaction
+	 * being logged.
+	 */
+	nfree = 0;
+	for (agno = nagcount - 1; agno >= oagcount; agno--, new -= agsize) {
+		/*
+		 * AG freelist header block
+		 */
+		bp = xfs_buf_get(mp->m_ddev_targp,
+				 XFS_AG_DADDR(mp, agno, XFS_AGF_DADDR(mp)),
+				 XFS_FSS_TO_BB(mp, 1), XBF_LOCK | XBF_MAPPED);
+		agf = XFS_BUF_TO_AGF(bp);
+		memset(agf, 0, mp->m_sb.sb_sectsize);
+		agf->agf_magicnum = cpu_to_be32(XFS_AGF_MAGIC);
+		agf->agf_versionnum = cpu_to_be32(XFS_AGF_VERSION);
+		agf->agf_seqno = cpu_to_be32(agno);
+		if (agno == nagcount - 1)
+			agsize =
+				nb -
+				(agno * (xfs_rfsblock_t)mp->m_sb.sb_agblocks);
+		else
+			agsize = mp->m_sb.sb_agblocks;
+		agf->agf_length = cpu_to_be32(agsize);
+		agf->agf_roots[XFS_BTNUM_BNOi] = cpu_to_be32(XFS_BNO_BLOCK(mp));
+		agf->agf_roots[XFS_BTNUM_CNTi] = cpu_to_be32(XFS_CNT_BLOCK(mp));
+		agf->agf_levels[XFS_BTNUM_BNOi] = cpu_to_be32(1);
+		agf->agf_levels[XFS_BTNUM_CNTi] = cpu_to_be32(1);
+		agf->agf_flfirst = 0;
+		agf->agf_fllast = cpu_to_be32(XFS_AGFL_SIZE(mp) - 1);
+		agf->agf_flcount = 0;
+		tmpsize = agsize - XFS_PREALLOC_BLOCKS(mp);
+		agf->agf_freeblks = cpu_to_be32(tmpsize);
+		agf->agf_longest = cpu_to_be32(tmpsize);
+		error = xfs_bwrite(mp, bp);
+		if (error) {
+			goto error0;
+		}
+		/*
+		 * AG inode header block
+		 */
+		bp = xfs_buf_get(mp->m_ddev_targp,
+				 XFS_AG_DADDR(mp, agno, XFS_AGI_DADDR(mp)),
+				 XFS_FSS_TO_BB(mp, 1), XBF_LOCK | XBF_MAPPED);
+		agi = XFS_BUF_TO_AGI(bp);
+		memset(agi, 0, mp->m_sb.sb_sectsize);
+		agi->agi_magicnum = cpu_to_be32(XFS_AGI_MAGIC);
+		agi->agi_versionnum = cpu_to_be32(XFS_AGI_VERSION);
+		agi->agi_seqno = cpu_to_be32(agno);
+		agi->agi_length = cpu_to_be32(agsize);
+		agi->agi_count = 0;
+		agi->agi_root = cpu_to_be32(XFS_IBT_BLOCK(mp));
+		agi->agi_level = cpu_to_be32(1);
+		agi->agi_freecount = 0;
+		agi->agi_newino = cpu_to_be32(NULLAGINO);
+		agi->agi_dirino = cpu_to_be32(NULLAGINO);
+		for (bucket = 0; bucket < XFS_AGI_UNLINKED_BUCKETS; bucket++)
+			agi->agi_unlinked[bucket] = cpu_to_be32(NULLAGINO);
+		error = xfs_bwrite(mp, bp);
+		if (error) {
+			goto error0;
+		}
+		/*
+		 * BNO btree root block
+		 */
+		bp = xfs_buf_get(mp->m_ddev_targp,
+				 XFS_AGB_TO_DADDR(mp, agno, XFS_BNO_BLOCK(mp)),
+				 BTOBB(mp->m_sb.sb_blocksize),
+				 XBF_LOCK | XBF_MAPPED);
+		block = XFS_BUF_TO_BLOCK(bp);
+		memset(block, 0, mp->m_sb.sb_blocksize);
+		block->bb_magic = cpu_to_be32(XFS_ABTB_MAGIC);
+		block->bb_level = 0;
+		block->bb_numrecs = cpu_to_be16(1);
+		block->bb_u.s.bb_leftsib = cpu_to_be32(NULLAGBLOCK);
+		block->bb_u.s.bb_rightsib = cpu_to_be32(NULLAGBLOCK);
+		arec = XFS_ALLOC_REC_ADDR(mp, block, 1);
+		arec->ar_startblock = cpu_to_be32(XFS_PREALLOC_BLOCKS(mp));
+		arec->ar_blockcount = cpu_to_be32(
+			agsize - be32_to_cpu(arec->ar_startblock));
+		error = xfs_bwrite(mp, bp);
+		if (error) {
+			goto error0;
+		}
+		/*
+		 * CNT btree root block
+		 */
+		bp = xfs_buf_get(mp->m_ddev_targp,
+				 XFS_AGB_TO_DADDR(mp, agno, XFS_CNT_BLOCK(mp)),
+				 BTOBB(mp->m_sb.sb_blocksize),
+				 XBF_LOCK | XBF_MAPPED);
+		block = XFS_BUF_TO_BLOCK(bp);
+		memset(block, 0, mp->m_sb.sb_blocksize);
+		block->bb_magic = cpu_to_be32(XFS_ABTC_MAGIC);
+		block->bb_level = 0;
+		block->bb_numrecs = cpu_to_be16(1);
+		block->bb_u.s.bb_leftsib = cpu_to_be32(NULLAGBLOCK);
+		block->bb_u.s.bb_rightsib = cpu_to_be32(NULLAGBLOCK);
+		arec = XFS_ALLOC_REC_ADDR(mp, block, 1);
+		arec->ar_startblock = cpu_to_be32(XFS_PREALLOC_BLOCKS(mp));
+		arec->ar_blockcount = cpu_to_be32(
+			agsize - be32_to_cpu(arec->ar_startblock));
+		nfree += be32_to_cpu(arec->ar_blockcount);
+		error = xfs_bwrite(mp, bp);
+		if (error) {
+			goto error0;
+		}
+		/*
+		 * INO btree root block
+		 */
+		bp = xfs_buf_get(mp->m_ddev_targp,
+				 XFS_AGB_TO_DADDR(mp, agno, XFS_IBT_BLOCK(mp)),
+				 BTOBB(mp->m_sb.sb_blocksize),
+				 XBF_LOCK | XBF_MAPPED);
+		block = XFS_BUF_TO_BLOCK(bp);
+		memset(block, 0, mp->m_sb.sb_blocksize);
+		block->bb_magic = cpu_to_be32(XFS_IBT_MAGIC);
+		block->bb_level = 0;
+		block->bb_numrecs = 0;
+		block->bb_u.s.bb_leftsib = cpu_to_be32(NULLAGBLOCK);
+		block->bb_u.s.bb_rightsib = cpu_to_be32(NULLAGBLOCK);
+		error = xfs_bwrite(mp, bp);
+		if (error) {
+			goto error0;
+		}
+	}
+	xfs_trans_agblocks_delta(tp, nfree);
+	/*
+	 * There are new blocks in the old last a.g.
+	 */
+	if (new) {
+		/*
+		 * Change the agi length.
+		 */
+		error = xfs_ialloc_read_agi(mp, tp, agno, &bp);
+		if (error) {
+			goto error0;
+		}
+		ASSERT(bp);
+		agi = XFS_BUF_TO_AGI(bp);
+		be32_add_cpu(&agi->agi_length, new);
+		ASSERT(nagcount == oagcount ||
+		       be32_to_cpu(agi->agi_length) == mp->m_sb.sb_agblocks);
+		xfs_ialloc_log_agi(tp, bp, XFS_AGI_LENGTH);
+		/*
+		 * Change agf length.
+		 */
+		error = xfs_alloc_read_agf(mp, tp, agno, 0, &bp);
+		if (error) {
+			goto error0;
+		}
+		ASSERT(bp);
+		agf = XFS_BUF_TO_AGF(bp);
+		be32_add_cpu(&agf->agf_length, new);
+		ASSERT(be32_to_cpu(agf->agf_length) ==
+		       be32_to_cpu(agi->agi_length));
+
+		xfs_alloc_log_agf(tp, bp, XFS_AGF_LENGTH);
+		/*
+		 * Free the new space.
+		 */
+		error = xfs_free_extent(tp, XFS_AGB_TO_FSB(mp, agno,
+			be32_to_cpu(agf->agf_length) - new), new);
+		if (error) {
+			goto error0;
+		}
+	}
+
+	/*
+	 * Update changed superblock fields transactionally. These are not
+	 * seen by the rest of the world until the transaction commit applies
+	 * them atomically to the superblock.
+	 */
+	if (nagcount > oagcount)
+		xfs_trans_mod_sb(tp, XFS_TRANS_SB_AGCOUNT, nagcount - oagcount);
+	if (nb > mp->m_sb.sb_dblocks)
+		xfs_trans_mod_sb(tp, XFS_TRANS_SB_DBLOCKS,
+				 nb - mp->m_sb.sb_dblocks);
+	if (nfree)
+		xfs_trans_mod_sb(tp, XFS_TRANS_SB_FDBLOCKS, nfree);
+	if (dpct)
+		xfs_trans_mod_sb(tp, XFS_TRANS_SB_IMAXPCT, dpct);
+	error = xfs_trans_commit(tp, 0);
+	if (error)
+		return error;
+
+	/* New allocation groups fully initialized, so update mount struct */
+	if (nagimax)
+		mp->m_maxagi = nagimax;
+	if (mp->m_sb.sb_imax_pct) {
+		__uint64_t icount = mp->m_sb.sb_dblocks * mp->m_sb.sb_imax_pct;
+		do_div(icount, 100);
+		mp->m_maxicount = icount << mp->m_sb.sb_inopblog;
+	} else
+		mp->m_maxicount = 0;
+	xfs_set_low_space_thresholds(mp);
+
+	/* update secondary superblocks. */
+	for (agno = 1; agno < nagcount; agno++) {
+		error = xfs_read_buf(mp, mp->m_ddev_targp,
+				  XFS_AGB_TO_DADDR(mp, agno, XFS_SB_BLOCK(mp)),
+				  XFS_FSS_TO_BB(mp, 1), 0, &bp);
+		if (error) {
+			xfs_warn(mp,
+		"error %d reading secondary superblock for ag %d",
+				error, agno);
+			break;
+		}
+		xfs_sb_to_disk(XFS_BUF_TO_SBP(bp), &mp->m_sb, XFS_SB_ALL_BITS);
+		/*
+		 * If we get an error writing out the alternate superblocks,
+		 * just issue a warning and continue.  The real work is
+		 * already done and committed.
+		 */
+		if (!(error = xfs_bwrite(mp, bp))) {
+			continue;
+		} else {
+			xfs_warn(mp,
+		"write error %d updating secondary superblock for ag %d",
+				error, agno);
+			break; /* no point in continuing */
+		}
+	}
+	return 0;
+
+ error0:
+	xfs_trans_cancel(tp, XFS_TRANS_ABORT);
+	return error;
+}
+
+static int
+xfs_growfs_log_private(
+	xfs_mount_t		*mp,	/* mount point for filesystem */
+	xfs_growfs_log_t	*in)	/* growfs log input struct */
+{
+	xfs_extlen_t		nb;
+
+	nb = in->newblocks;
+	if (nb < XFS_MIN_LOG_BLOCKS || nb < XFS_B_TO_FSB(mp, XFS_MIN_LOG_BYTES))
+		return XFS_ERROR(EINVAL);
+	if (nb == mp->m_sb.sb_logblocks &&
+	    in->isint == (mp->m_sb.sb_logstart != 0))
+		return XFS_ERROR(EINVAL);
+	/*
+	 * Moving the log is hard, need new interfaces to sync
+	 * the log first, hold off all activity while moving it.
+	 * Can have shorter or longer log in the same space,
+	 * or transform internal to external log or vice versa.
+	 */
+	return XFS_ERROR(ENOSYS);
+}
+
+/*
+ * protected versions of growfs function acquire and release locks on the mount
+ * point - exported through ioctls: XFS_IOC_FSGROWFSDATA, XFS_IOC_FSGROWFSLOG,
+ * XFS_IOC_FSGROWFSRT
+ */
+
+
+int
+xfs_growfs_data(
+	xfs_mount_t		*mp,
+	xfs_growfs_data_t	*in)
+{
+	int error;
+
+	if (!capable(CAP_SYS_ADMIN))
+		return XFS_ERROR(EPERM);
+	if (!mutex_trylock(&mp->m_growlock))
+		return XFS_ERROR(EWOULDBLOCK);
+	error = xfs_growfs_data_private(mp, in);
+	mutex_unlock(&mp->m_growlock);
+	return error;
+}
+
+int
+xfs_growfs_log(
+	xfs_mount_t		*mp,
+	xfs_growfs_log_t	*in)
+{
+	int error;
+
+	if (!capable(CAP_SYS_ADMIN))
+		return XFS_ERROR(EPERM);
+	if (!mutex_trylock(&mp->m_growlock))
+		return XFS_ERROR(EWOULDBLOCK);
+	error = xfs_growfs_log_private(mp, in);
+	mutex_unlock(&mp->m_growlock);
+	return error;
+}
+
+/*
+ * exported through ioctl XFS_IOC_FSCOUNTS
+ */
+
+int
+xfs_fs_counts(
+	xfs_mount_t		*mp,
+	xfs_fsop_counts_t	*cnt)
+{
+	xfs_icsb_sync_counters(mp, XFS_ICSB_LAZY_COUNT);
+	spin_lock(&mp->m_sb_lock);
+	cnt->freedata = mp->m_sb.sb_fdblocks - XFS_ALLOC_SET_ASIDE(mp);
+	cnt->freertx = mp->m_sb.sb_frextents;
+	cnt->freeino = mp->m_sb.sb_ifree;
+	cnt->allocino = mp->m_sb.sb_icount;
+	spin_unlock(&mp->m_sb_lock);
+	return 0;
+}
+
+/*
+ * exported through ioctl XFS_IOC_SET_RESBLKS & XFS_IOC_GET_RESBLKS
+ *
+ * xfs_reserve_blocks is called to set m_resblks
+ * in the in-core mount table. The number of unused reserved blocks
+ * is kept in m_resblks_avail.
+ *
+ * Reserve the requested number of blocks if available. Otherwise return
+ * as many as possible to satisfy the request. The actual number
+ * reserved are returned in outval
+ *
+ * A null inval pointer indicates that only the current reserved blocks
+ * available  should  be returned no settings are changed.
+ */
+
+int
+xfs_reserve_blocks(
+	xfs_mount_t             *mp,
+	__uint64_t              *inval,
+	xfs_fsop_resblks_t      *outval)
+{
+	__int64_t		lcounter, delta, fdblks_delta;
+	__uint64_t		request;
+
+	/* If inval is null, report current values and return */
+	if (inval == (__uint64_t *)NULL) {
+		if (!outval)
+			return EINVAL;
+		outval->resblks = mp->m_resblks;
+		outval->resblks_avail = mp->m_resblks_avail;
+		return 0;
+	}
+
+	request = *inval;
+
+	/*
+	 * With per-cpu counters, this becomes an interesting
+	 * problem. we needto work out if we are freeing or allocation
+	 * blocks first, then we can do the modification as necessary.
+	 *
+	 * We do this under the m_sb_lock so that if we are near
+	 * ENOSPC, we will hold out any changes while we work out
+	 * what to do. This means that the amount of free space can
+	 * change while we do this, so we need to retry if we end up
+	 * trying to reserve more space than is available.
+	 *
+	 * We also use the xfs_mod_incore_sb() interface so that we
+	 * don't have to care about whether per cpu counter are
+	 * enabled, disabled or even compiled in....
+	 */
+retry:
+	spin_lock(&mp->m_sb_lock);
+	xfs_icsb_sync_counters_locked(mp, 0);
+
+	/*
+	 * If our previous reservation was larger than the current value,
+	 * then move any unused blocks back to the free pool.
+	 */
+	fdblks_delta = 0;
+	if (mp->m_resblks > request) {
+		lcounter = mp->m_resblks_avail - request;
+		if (lcounter  > 0) {		/* release unused blocks */
+			fdblks_delta = lcounter;
+			mp->m_resblks_avail -= lcounter;
+		}
+		mp->m_resblks = request;
+	} else {
+		__int64_t	free;
+
+		free =  mp->m_sb.sb_fdblocks - XFS_ALLOC_SET_ASIDE(mp);
+		if (!free)
+			goto out; /* ENOSPC and fdblks_delta = 0 */
+
+		delta = request - mp->m_resblks;
+		lcounter = free - delta;
+		if (lcounter < 0) {
+			/* We can't satisfy the request, just get what we can */
+			mp->m_resblks += free;
+			mp->m_resblks_avail += free;
+			fdblks_delta = -free;
+		} else {
+			fdblks_delta = -delta;
+			mp->m_resblks = request;
+			mp->m_resblks_avail += delta;
+		}
+	}
+out:
+	if (outval) {
+		outval->resblks = mp->m_resblks;
+		outval->resblks_avail = mp->m_resblks_avail;
+	}
+	spin_unlock(&mp->m_sb_lock);
+
+	if (fdblks_delta) {
+		/*
+		 * If we are putting blocks back here, m_resblks_avail is
+		 * already at its max so this will put it in the free pool.
+		 *
+		 * If we need space, we'll either succeed in getting it
+		 * from the free block count or we'll get an enospc. If
+		 * we get a ENOSPC, it means things changed while we were
+		 * calculating fdblks_delta and so we should try again to
+		 * see if there is anything left to reserve.
+		 *
+		 * Don't set the reserved flag here - we don't want to reserve
+		 * the extra reserve blocks from the reserve.....
+		 */
+		int error;
+		error = xfs_icsb_modify_counters(mp, XFS_SBS_FDBLOCKS,
+						 fdblks_delta, 0);
+		if (error == ENOSPC)
+			goto retry;
+	}
+	return 0;
+}
+
+/*
+ * Dump a transaction into the log that contains no real change. This is needed
+ * to be able to make the log dirty or stamp the current tail LSN into the log
+ * during the covering operation.
+ *
+ * We cannot use an inode here for this - that will push dirty state back up
+ * into the VFS and then periodic inode flushing will prevent log covering from
+ * making progress. Hence we log a field in the superblock instead and use a
+ * synchronous transaction to ensure the superblock is immediately unpinned
+ * and can be written back.
+ */
+int
+xfs_fs_log_dummy(
+	xfs_mount_t	*mp)
+{
+	xfs_trans_t	*tp;
+	int		error;
+
+	tp = _xfs_trans_alloc(mp, XFS_TRANS_DUMMY1, KM_SLEEP);
+	error = xfs_trans_reserve(tp, 0, mp->m_sb.sb_sectsize + 128, 0, 0,
+					XFS_DEFAULT_LOG_COUNT);
+	if (error) {
+		xfs_trans_cancel(tp, 0);
+		return error;
+	}
+
+	/* log the UUID because it is an unchanging field */
+	xfs_mod_sb(tp, XFS_SB_UUID);
+	xfs_trans_set_sync(tp);
+	return xfs_trans_commit(tp, 0);
+}
+
+int
+xfs_fs_goingdown(
+	xfs_mount_t	*mp,
+	__uint32_t	inflags)
+{
+	switch (inflags) {
+	case XFS_FSOP_GOING_FLAGS_DEFAULT: {
+		struct super_block *sb = freeze_bdev(mp->m_super->s_bdev);
+
+		if (sb && !IS_ERR(sb)) {
+			xfs_force_shutdown(mp, SHUTDOWN_FORCE_UMOUNT);
+			thaw_bdev(sb->s_bdev, sb);
+		}
+
+		break;
+	}
+	case XFS_FSOP_GOING_FLAGS_LOGFLUSH:
+		xfs_force_shutdown(mp, SHUTDOWN_FORCE_UMOUNT);
+		break;
+	case XFS_FSOP_GOING_FLAGS_NOLOGFLUSH:
+		xfs_force_shutdown(mp,
+				SHUTDOWN_FORCE_UMOUNT | SHUTDOWN_LOG_IO_ERROR);
+		break;
+	default:
+		return XFS_ERROR(EINVAL);
+	}
+
+	return 0;
+}
diff --git a/fs/xfs/xfs_fsops.h b/fs/xfs/xfs_fsops.h
new file mode 100644
index 00000000..1b6a98b6
--- /dev/null
+++ b/fs/xfs/xfs_fsops.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright (c) 2000-2001,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_FSOPS_H__
+#define	__XFS_FSOPS_H__
+
+extern int xfs_fs_geometry(xfs_mount_t *mp, xfs_fsop_geom_t *geo, int nversion);
+extern int xfs_growfs_data(xfs_mount_t *mp, xfs_growfs_data_t *in);
+extern int xfs_growfs_log(xfs_mount_t *mp, xfs_growfs_log_t *in);
+extern int xfs_fs_counts(xfs_mount_t *mp, xfs_fsop_counts_t *cnt);
+extern int xfs_reserve_blocks(xfs_mount_t *mp, __uint64_t *inval,
+				xfs_fsop_resblks_t *outval);
+extern int xfs_fs_goingdown(xfs_mount_t *mp, __uint32_t inflags);
+extern int xfs_fs_log_dummy(struct xfs_mount *mp);
+
+#endif	/* __XFS_FSOPS_H__ */
diff --git a/fs/xfs/xfs_ialloc.c b/fs/xfs/xfs_ialloc.c
new file mode 100644
index 00000000..84ebeec1
--- /dev/null
+++ b/fs/xfs/xfs_ialloc.c
@@ -0,0 +1,1563 @@
+/*
+ * Copyright (c) 2000-2002,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_mount.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_alloc_btree.h"
+#include "xfs_ialloc_btree.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_btree.h"
+#include "xfs_ialloc.h"
+#include "xfs_alloc.h"
+#include "xfs_rtalloc.h"
+#include "xfs_error.h"
+#include "xfs_bmap.h"
+
+
+/*
+ * Allocation group level functions.
+ */
+static inline int
+xfs_ialloc_cluster_alignment(
+	xfs_alloc_arg_t	*args)
+{
+	if (xfs_sb_version_hasalign(&args->mp->m_sb) &&
+	    args->mp->m_sb.sb_inoalignmt >=
+	     XFS_B_TO_FSBT(args->mp, XFS_INODE_CLUSTER_SIZE(args->mp)))
+		return args->mp->m_sb.sb_inoalignmt;
+	return 1;
+}
+
+/*
+ * Lookup a record by ino in the btree given by cur.
+ */
+int					/* error */
+xfs_inobt_lookup(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_agino_t		ino,	/* starting inode of chunk */
+	xfs_lookup_t		dir,	/* <=, >=, == */
+	int			*stat)	/* success/failure */
+{
+	cur->bc_rec.i.ir_startino = ino;
+	cur->bc_rec.i.ir_freecount = 0;
+	cur->bc_rec.i.ir_free = 0;
+	return xfs_btree_lookup(cur, dir, stat);
+}
+
+/*
+ * Update the record referred to by cur to the value given.
+ * This either works (return 0) or gets an EFSCORRUPTED error.
+ */
+STATIC int				/* error */
+xfs_inobt_update(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_inobt_rec_incore_t	*irec)	/* btree record */
+{
+	union xfs_btree_rec	rec;
+
+	rec.inobt.ir_startino = cpu_to_be32(irec->ir_startino);
+	rec.inobt.ir_freecount = cpu_to_be32(irec->ir_freecount);
+	rec.inobt.ir_free = cpu_to_be64(irec->ir_free);
+	return xfs_btree_update(cur, &rec);
+}
+
+/*
+ * Get the data from the pointed-to record.
+ */
+int					/* error */
+xfs_inobt_get_rec(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_inobt_rec_incore_t	*irec,	/* btree record */
+	int			*stat)	/* output: success/failure */
+{
+	union xfs_btree_rec	*rec;
+	int			error;
+
+	error = xfs_btree_get_rec(cur, &rec, stat);
+	if (!error && *stat == 1) {
+		irec->ir_startino = be32_to_cpu(rec->inobt.ir_startino);
+		irec->ir_freecount = be32_to_cpu(rec->inobt.ir_freecount);
+		irec->ir_free = be64_to_cpu(rec->inobt.ir_free);
+	}
+	return error;
+}
+
+/*
+ * Verify that the number of free inodes in the AGI is correct.
+ */
+#ifdef DEBUG
+STATIC int
+xfs_check_agi_freecount(
+	struct xfs_btree_cur	*cur,
+	struct xfs_agi		*agi)
+{
+	if (cur->bc_nlevels == 1) {
+		xfs_inobt_rec_incore_t rec;
+		int		freecount = 0;
+		int		error;
+		int		i;
+
+		error = xfs_inobt_lookup(cur, 0, XFS_LOOKUP_GE, &i);
+		if (error)
+			return error;
+
+		do {
+			error = xfs_inobt_get_rec(cur, &rec, &i);
+			if (error)
+				return error;
+
+			if (i) {
+				freecount += rec.ir_freecount;
+				error = xfs_btree_increment(cur, 0, &i);
+				if (error)
+					return error;
+			}
+		} while (i == 1);
+
+		if (!XFS_FORCED_SHUTDOWN(cur->bc_mp))
+			ASSERT(freecount == be32_to_cpu(agi->agi_freecount));
+	}
+	return 0;
+}
+#else
+#define xfs_check_agi_freecount(cur, agi)	0
+#endif
+
+/*
+ * Initialise a new set of inodes.
+ */
+STATIC void
+xfs_ialloc_inode_init(
+	struct xfs_mount	*mp,
+	struct xfs_trans	*tp,
+	xfs_agnumber_t		agno,
+	xfs_agblock_t		agbno,
+	xfs_agblock_t		length,
+	unsigned int		gen)
+{
+	struct xfs_buf		*fbuf;
+	struct xfs_dinode	*free;
+	int			blks_per_cluster, nbufs, ninodes;
+	int			version;
+	int			i, j;
+	xfs_daddr_t		d;
+
+	/*
+	 * Loop over the new block(s), filling in the inodes.
+	 * For small block sizes, manipulate the inodes in buffers
+	 * which are multiples of the blocks size.
+	 */
+	if (mp->m_sb.sb_blocksize >= XFS_INODE_CLUSTER_SIZE(mp)) {
+		blks_per_cluster = 1;
+		nbufs = length;
+		ninodes = mp->m_sb.sb_inopblock;
+	} else {
+		blks_per_cluster = XFS_INODE_CLUSTER_SIZE(mp) /
+				   mp->m_sb.sb_blocksize;
+		nbufs = length / blks_per_cluster;
+		ninodes = blks_per_cluster * mp->m_sb.sb_inopblock;
+	}
+
+	/*
+	 * Figure out what version number to use in the inodes we create.
+	 * If the superblock version has caught up to the one that supports
+	 * the new inode format, then use the new inode version.  Otherwise
+	 * use the old version so that old kernels will continue to be
+	 * able to use the file system.
+	 */
+	if (xfs_sb_version_hasnlink(&mp->m_sb))
+		version = 2;
+	else
+		version = 1;
+
+	for (j = 0; j < nbufs; j++) {
+		/*
+		 * Get the block.
+		 */
+		d = XFS_AGB_TO_DADDR(mp, agno, agbno + (j * blks_per_cluster));
+		fbuf = xfs_trans_get_buf(tp, mp->m_ddev_targp, d,
+					 mp->m_bsize * blks_per_cluster,
+					 XBF_LOCK);
+		ASSERT(fbuf);
+		ASSERT(!XFS_BUF_GETERROR(fbuf));
+
+		/*
+		 * Initialize all inodes in this buffer and then log them.
+		 *
+		 * XXX: It would be much better if we had just one transaction
+		 *	to log a whole cluster of inodes instead of all the
+		 *	individual transactions causing a lot of log traffic.
+		 */
+		xfs_buf_zero(fbuf, 0, ninodes << mp->m_sb.sb_inodelog);
+		for (i = 0; i < ninodes; i++) {
+			int	ioffset = i << mp->m_sb.sb_inodelog;
+			uint	isize = sizeof(struct xfs_dinode);
+
+			free = xfs_make_iptr(mp, fbuf, i);
+			free->di_magic = cpu_to_be16(XFS_DINODE_MAGIC);
+			free->di_version = version;
+			free->di_gen = cpu_to_be32(gen);
+			free->di_next_unlinked = cpu_to_be32(NULLAGINO);
+			xfs_trans_log_buf(tp, fbuf, ioffset, ioffset + isize - 1);
+		}
+		xfs_trans_inode_alloc_buf(tp, fbuf);
+	}
+}
+
+/*
+ * Allocate new inodes in the allocation group specified by agbp.
+ * Return 0 for success, else error code.
+ */
+STATIC int				/* error code or 0 */
+xfs_ialloc_ag_alloc(
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_buf_t	*agbp,		/* alloc group buffer */
+	int		*alloc)
+{
+	xfs_agi_t	*agi;		/* allocation group header */
+	xfs_alloc_arg_t	args;		/* allocation argument structure */
+	xfs_btree_cur_t	*cur;		/* inode btree cursor */
+	xfs_agnumber_t	agno;
+	int		error;
+	int		i;
+	xfs_agino_t	newino;		/* new first inode's number */
+	xfs_agino_t	newlen;		/* new number of inodes */
+	xfs_agino_t	thisino;	/* current inode number, for loop */
+	int		isaligned = 0;	/* inode allocation at stripe unit */
+					/* boundary */
+	struct xfs_perag *pag;
+
+	args.tp = tp;
+	args.mp = tp->t_mountp;
+
+	/*
+	 * Locking will ensure that we don't have two callers in here
+	 * at one time.
+	 */
+	newlen = XFS_IALLOC_INODES(args.mp);
+	if (args.mp->m_maxicount &&
+	    args.mp->m_sb.sb_icount + newlen > args.mp->m_maxicount)
+		return XFS_ERROR(ENOSPC);
+	args.minlen = args.maxlen = XFS_IALLOC_BLOCKS(args.mp);
+	/*
+	 * First try to allocate inodes contiguous with the last-allocated
+	 * chunk of inodes.  If the filesystem is striped, this will fill
+	 * an entire stripe unit with inodes.
+ 	 */
+	agi = XFS_BUF_TO_AGI(agbp);
+	newino = be32_to_cpu(agi->agi_newino);
+	agno = be32_to_cpu(agi->agi_seqno);
+	args.agbno = XFS_AGINO_TO_AGBNO(args.mp, newino) +
+			XFS_IALLOC_BLOCKS(args.mp);
+	if (likely(newino != NULLAGINO &&
+		  (args.agbno < be32_to_cpu(agi->agi_length)))) {
+		args.fsbno = XFS_AGB_TO_FSB(args.mp, agno, args.agbno);
+		args.type = XFS_ALLOCTYPE_THIS_BNO;
+		args.mod = args.total = args.wasdel = args.isfl =
+			args.userdata = args.minalignslop = 0;
+		args.prod = 1;
+
+		/*
+		 * We need to take into account alignment here to ensure that
+		 * we don't modify the free list if we fail to have an exact
+		 * block. If we don't have an exact match, and every oher
+		 * attempt allocation attempt fails, we'll end up cancelling
+		 * a dirty transaction and shutting down.
+		 *
+		 * For an exact allocation, alignment must be 1,
+		 * however we need to take cluster alignment into account when
+		 * fixing up the freelist. Use the minalignslop field to
+		 * indicate that extra blocks might be required for alignment,
+		 * but not to use them in the actual exact allocation.
+		 */
+		args.alignment = 1;
+		args.minalignslop = xfs_ialloc_cluster_alignment(&args) - 1;
+
+		/* Allow space for the inode btree to split. */
+		args.minleft = args.mp->m_in_maxlevels - 1;
+		if ((error = xfs_alloc_vextent(&args)))
+			return error;
+	} else
+		args.fsbno = NULLFSBLOCK;
+
+	if (unlikely(args.fsbno == NULLFSBLOCK)) {
+		/*
+		 * Set the alignment for the allocation.
+		 * If stripe alignment is turned on then align at stripe unit
+		 * boundary.
+		 * If the cluster size is smaller than a filesystem block
+		 * then we're doing I/O for inodes in filesystem block size
+		 * pieces, so don't need alignment anyway.
+		 */
+		isaligned = 0;
+		if (args.mp->m_sinoalign) {
+			ASSERT(!(args.mp->m_flags & XFS_MOUNT_NOALIGN));
+			args.alignment = args.mp->m_dalign;
+			isaligned = 1;
+		} else
+			args.alignment = xfs_ialloc_cluster_alignment(&args);
+		/*
+		 * Need to figure out where to allocate the inode blocks.
+		 * Ideally they should be spaced out through the a.g.
+		 * For now, just allocate blocks up front.
+		 */
+		args.agbno = be32_to_cpu(agi->agi_root);
+		args.fsbno = XFS_AGB_TO_FSB(args.mp, agno, args.agbno);
+		/*
+		 * Allocate a fixed-size extent of inodes.
+		 */
+		args.type = XFS_ALLOCTYPE_NEAR_BNO;
+		args.mod = args.total = args.wasdel = args.isfl =
+			args.userdata = args.minalignslop = 0;
+		args.prod = 1;
+		/*
+		 * Allow space for the inode btree to split.
+		 */
+		args.minleft = args.mp->m_in_maxlevels - 1;
+		if ((error = xfs_alloc_vextent(&args)))
+			return error;
+	}
+
+	/*
+	 * If stripe alignment is turned on, then try again with cluster
+	 * alignment.
+	 */
+	if (isaligned && args.fsbno == NULLFSBLOCK) {
+		args.type = XFS_ALLOCTYPE_NEAR_BNO;
+		args.agbno = be32_to_cpu(agi->agi_root);
+		args.fsbno = XFS_AGB_TO_FSB(args.mp, agno, args.agbno);
+		args.alignment = xfs_ialloc_cluster_alignment(&args);
+		if ((error = xfs_alloc_vextent(&args)))
+			return error;
+	}
+
+	if (args.fsbno == NULLFSBLOCK) {
+		*alloc = 0;
+		return 0;
+	}
+	ASSERT(args.len == args.minlen);
+
+	/*
+	 * Stamp and write the inode buffers.
+	 *
+	 * Seed the new inode cluster with a random generation number. This
+	 * prevents short-term reuse of generation numbers if a chunk is
+	 * freed and then immediately reallocated. We use random numbers
+	 * rather than a linear progression to prevent the next generation
+	 * number from being easily guessable.
+	 */
+	xfs_ialloc_inode_init(args.mp, tp, agno, args.agbno, args.len,
+			      random32());
+
+	/*
+	 * Convert the results.
+	 */
+	newino = XFS_OFFBNO_TO_AGINO(args.mp, args.agbno, 0);
+	be32_add_cpu(&agi->agi_count, newlen);
+	be32_add_cpu(&agi->agi_freecount, newlen);
+	pag = xfs_perag_get(args.mp, agno);
+	pag->pagi_freecount += newlen;
+	xfs_perag_put(pag);
+	agi->agi_newino = cpu_to_be32(newino);
+
+	/*
+	 * Insert records describing the new inode chunk into the btree.
+	 */
+	cur = xfs_inobt_init_cursor(args.mp, tp, agbp, agno);
+	for (thisino = newino;
+	     thisino < newino + newlen;
+	     thisino += XFS_INODES_PER_CHUNK) {
+		cur->bc_rec.i.ir_startino = thisino;
+		cur->bc_rec.i.ir_freecount = XFS_INODES_PER_CHUNK;
+		cur->bc_rec.i.ir_free = XFS_INOBT_ALL_FREE;
+		error = xfs_btree_lookup(cur, XFS_LOOKUP_EQ, &i);
+		if (error) {
+			xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
+			return error;
+		}
+		ASSERT(i == 0);
+		error = xfs_btree_insert(cur, &i);
+		if (error) {
+			xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
+			return error;
+		}
+		ASSERT(i == 1);
+	}
+	xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+	/*
+	 * Log allocation group header fields
+	 */
+	xfs_ialloc_log_agi(tp, agbp,
+		XFS_AGI_COUNT | XFS_AGI_FREECOUNT | XFS_AGI_NEWINO);
+	/*
+	 * Modify/log superblock values for inode count and inode free count.
+	 */
+	xfs_trans_mod_sb(tp, XFS_TRANS_SB_ICOUNT, (long)newlen);
+	xfs_trans_mod_sb(tp, XFS_TRANS_SB_IFREE, (long)newlen);
+	*alloc = 1;
+	return 0;
+}
+
+STATIC xfs_agnumber_t
+xfs_ialloc_next_ag(
+	xfs_mount_t	*mp)
+{
+	xfs_agnumber_t	agno;
+
+	spin_lock(&mp->m_agirotor_lock);
+	agno = mp->m_agirotor;
+	if (++mp->m_agirotor == mp->m_maxagi)
+		mp->m_agirotor = 0;
+	spin_unlock(&mp->m_agirotor_lock);
+
+	return agno;
+}
+
+/*
+ * Select an allocation group to look for a free inode in, based on the parent
+ * inode and then mode.  Return the allocation group buffer.
+ */
+STATIC xfs_buf_t *			/* allocation group buffer */
+xfs_ialloc_ag_select(
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_ino_t	parent,		/* parent directory inode number */
+	mode_t		mode,		/* bits set to indicate file type */
+	int		okalloc)	/* ok to allocate more space */
+{
+	xfs_buf_t	*agbp;		/* allocation group header buffer */
+	xfs_agnumber_t	agcount;	/* number of ag's in the filesystem */
+	xfs_agnumber_t	agno;		/* current ag number */
+	int		flags;		/* alloc buffer locking flags */
+	xfs_extlen_t	ineed;		/* blocks needed for inode allocation */
+	xfs_extlen_t	longest = 0;	/* longest extent available */
+	xfs_mount_t	*mp;		/* mount point structure */
+	int		needspace;	/* file mode implies space allocated */
+	xfs_perag_t	*pag;		/* per allocation group data */
+	xfs_agnumber_t	pagno;		/* parent (starting) ag number */
+
+	/*
+	 * Files of these types need at least one block if length > 0
+	 * (and they won't fit in the inode, but that's hard to figure out).
+	 */
+	needspace = S_ISDIR(mode) || S_ISREG(mode) || S_ISLNK(mode);
+	mp = tp->t_mountp;
+	agcount = mp->m_maxagi;
+	if (S_ISDIR(mode))
+		pagno = xfs_ialloc_next_ag(mp);
+	else {
+		pagno = XFS_INO_TO_AGNO(mp, parent);
+		if (pagno >= agcount)
+			pagno = 0;
+	}
+	ASSERT(pagno < agcount);
+	/*
+	 * Loop through allocation groups, looking for one with a little
+	 * free space in it.  Note we don't look for free inodes, exactly.
+	 * Instead, we include whether there is a need to allocate inodes
+	 * to mean that blocks must be allocated for them,
+	 * if none are currently free.
+	 */
+	agno = pagno;
+	flags = XFS_ALLOC_FLAG_TRYLOCK;
+	for (;;) {
+		pag = xfs_perag_get(mp, agno);
+		if (!pag->pagi_init) {
+			if (xfs_ialloc_read_agi(mp, tp, agno, &agbp)) {
+				agbp = NULL;
+				goto nextag;
+			}
+		} else
+			agbp = NULL;
+
+		if (!pag->pagi_inodeok) {
+			xfs_ialloc_next_ag(mp);
+			goto unlock_nextag;
+		}
+
+		/*
+		 * Is there enough free space for the file plus a block
+		 * of inodes (if we need to allocate some)?
+		 */
+		ineed = pag->pagi_freecount ? 0 : XFS_IALLOC_BLOCKS(mp);
+		if (ineed && !pag->pagf_init) {
+			if (agbp == NULL &&
+			    xfs_ialloc_read_agi(mp, tp, agno, &agbp)) {
+				agbp = NULL;
+				goto nextag;
+			}
+			(void)xfs_alloc_pagf_init(mp, tp, agno, flags);
+		}
+		if (!ineed || pag->pagf_init) {
+			if (ineed && !(longest = pag->pagf_longest))
+				longest = pag->pagf_flcount > 0;
+			if (!ineed ||
+			    (pag->pagf_freeblks >= needspace + ineed &&
+			     longest >= ineed &&
+			     okalloc)) {
+				if (agbp == NULL &&
+				    xfs_ialloc_read_agi(mp, tp, agno, &agbp)) {
+					agbp = NULL;
+					goto nextag;
+				}
+				xfs_perag_put(pag);
+				return agbp;
+			}
+		}
+unlock_nextag:
+		if (agbp)
+			xfs_trans_brelse(tp, agbp);
+nextag:
+		xfs_perag_put(pag);
+		/*
+		 * No point in iterating over the rest, if we're shutting
+		 * down.
+		 */
+		if (XFS_FORCED_SHUTDOWN(mp))
+			return NULL;
+		agno++;
+		if (agno >= agcount)
+			agno = 0;
+		if (agno == pagno) {
+			if (flags == 0)
+				return NULL;
+			flags = 0;
+		}
+	}
+}
+
+/*
+ * Try to retrieve the next record to the left/right from the current one.
+ */
+STATIC int
+xfs_ialloc_next_rec(
+	struct xfs_btree_cur	*cur,
+	xfs_inobt_rec_incore_t	*rec,
+	int			*done,
+	int			left)
+{
+	int                     error;
+	int			i;
+
+	if (left)
+		error = xfs_btree_decrement(cur, 0, &i);
+	else
+		error = xfs_btree_increment(cur, 0, &i);
+
+	if (error)
+		return error;
+	*done = !i;
+	if (i) {
+		error = xfs_inobt_get_rec(cur, rec, &i);
+		if (error)
+			return error;
+		XFS_WANT_CORRUPTED_RETURN(i == 1);
+	}
+
+	return 0;
+}
+
+STATIC int
+xfs_ialloc_get_rec(
+	struct xfs_btree_cur	*cur,
+	xfs_agino_t		agino,
+	xfs_inobt_rec_incore_t	*rec,
+	int			*done,
+	int			left)
+{
+	int                     error;
+	int			i;
+
+	error = xfs_inobt_lookup(cur, agino, XFS_LOOKUP_EQ, &i);
+	if (error)
+		return error;
+	*done = !i;
+	if (i) {
+		error = xfs_inobt_get_rec(cur, rec, &i);
+		if (error)
+			return error;
+		XFS_WANT_CORRUPTED_RETURN(i == 1);
+	}
+
+	return 0;
+}
+
+/*
+ * Visible inode allocation functions.
+ */
+
+/*
+ * Allocate an inode on disk.
+ * Mode is used to tell whether the new inode will need space, and whether
+ * it is a directory.
+ *
+ * The arguments IO_agbp and alloc_done are defined to work within
+ * the constraint of one allocation per transaction.
+ * xfs_dialloc() is designed to be called twice if it has to do an
+ * allocation to make more free inodes.  On the first call,
+ * IO_agbp should be set to NULL. If an inode is available,
+ * i.e., xfs_dialloc() did not need to do an allocation, an inode
+ * number is returned.  In this case, IO_agbp would be set to the
+ * current ag_buf and alloc_done set to false.
+ * If an allocation needed to be done, xfs_dialloc would return
+ * the current ag_buf in IO_agbp and set alloc_done to true.
+ * The caller should then commit the current transaction, allocate a new
+ * transaction, and call xfs_dialloc() again, passing in the previous
+ * value of IO_agbp.  IO_agbp should be held across the transactions.
+ * Since the agbp is locked across the two calls, the second call is
+ * guaranteed to have a free inode available.
+ *
+ * Once we successfully pick an inode its number is returned and the
+ * on-disk data structures are updated.  The inode itself is not read
+ * in, since doing so would break ordering constraints with xfs_reclaim.
+ */
+int
+xfs_dialloc(
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_ino_t	parent,		/* parent inode (directory) */
+	mode_t		mode,		/* mode bits for new inode */
+	int		okalloc,	/* ok to allocate more space */
+	xfs_buf_t	**IO_agbp,	/* in/out ag header's buffer */
+	boolean_t	*alloc_done,	/* true if we needed to replenish
+					   inode freelist */
+	xfs_ino_t	*inop)		/* inode number allocated */
+{
+	xfs_agnumber_t	agcount;	/* number of allocation groups */
+	xfs_buf_t	*agbp;		/* allocation group header's buffer */
+	xfs_agnumber_t	agno;		/* allocation group number */
+	xfs_agi_t	*agi;		/* allocation group header structure */
+	xfs_btree_cur_t	*cur;		/* inode allocation btree cursor */
+	int		error;		/* error return value */
+	int		i;		/* result code */
+	int		ialloced;	/* inode allocation status */
+	int		noroom = 0;	/* no space for inode blk allocation */
+	xfs_ino_t	ino;		/* fs-relative inode to be returned */
+	/* REFERENCED */
+	int		j;		/* result code */
+	xfs_mount_t	*mp;		/* file system mount structure */
+	int		offset;		/* index of inode in chunk */
+	xfs_agino_t	pagino;		/* parent's AG relative inode # */
+	xfs_agnumber_t	pagno;		/* parent's AG number */
+	xfs_inobt_rec_incore_t rec;	/* inode allocation record */
+	xfs_agnumber_t	tagno;		/* testing allocation group number */
+	xfs_btree_cur_t	*tcur;		/* temp cursor */
+	xfs_inobt_rec_incore_t trec;	/* temp inode allocation record */
+	struct xfs_perag *pag;
+
+
+	if (*IO_agbp == NULL) {
+		/*
+		 * We do not have an agbp, so select an initial allocation
+		 * group for inode allocation.
+		 */
+		agbp = xfs_ialloc_ag_select(tp, parent, mode, okalloc);
+		/*
+		 * Couldn't find an allocation group satisfying the
+		 * criteria, give up.
+		 */
+		if (!agbp) {
+			*inop = NULLFSINO;
+			return 0;
+		}
+		agi = XFS_BUF_TO_AGI(agbp);
+		ASSERT(be32_to_cpu(agi->agi_magicnum) == XFS_AGI_MAGIC);
+	} else {
+		/*
+		 * Continue where we left off before.  In this case, we
+		 * know that the allocation group has free inodes.
+		 */
+		agbp = *IO_agbp;
+		agi = XFS_BUF_TO_AGI(agbp);
+		ASSERT(be32_to_cpu(agi->agi_magicnum) == XFS_AGI_MAGIC);
+		ASSERT(be32_to_cpu(agi->agi_freecount) > 0);
+	}
+	mp = tp->t_mountp;
+	agcount = mp->m_sb.sb_agcount;
+	agno = be32_to_cpu(agi->agi_seqno);
+	tagno = agno;
+	pagno = XFS_INO_TO_AGNO(mp, parent);
+	pagino = XFS_INO_TO_AGINO(mp, parent);
+
+	/*
+	 * If we have already hit the ceiling of inode blocks then clear
+	 * okalloc so we scan all available agi structures for a free
+	 * inode.
+	 */
+
+	if (mp->m_maxicount &&
+	    mp->m_sb.sb_icount + XFS_IALLOC_INODES(mp) > mp->m_maxicount) {
+		noroom = 1;
+		okalloc = 0;
+	}
+
+	/*
+	 * Loop until we find an allocation group that either has free inodes
+	 * or in which we can allocate some inodes.  Iterate through the
+	 * allocation groups upward, wrapping at the end.
+	 */
+	*alloc_done = B_FALSE;
+	while (!agi->agi_freecount) {
+		/*
+		 * Don't do anything if we're not supposed to allocate
+		 * any blocks, just go on to the next ag.
+		 */
+		if (okalloc) {
+			/*
+			 * Try to allocate some new inodes in the allocation
+			 * group.
+			 */
+			if ((error = xfs_ialloc_ag_alloc(tp, agbp, &ialloced))) {
+				xfs_trans_brelse(tp, agbp);
+				if (error == ENOSPC) {
+					*inop = NULLFSINO;
+					return 0;
+				} else
+					return error;
+			}
+			if (ialloced) {
+				/*
+				 * We successfully allocated some inodes, return
+				 * the current context to the caller so that it
+				 * can commit the current transaction and call
+				 * us again where we left off.
+				 */
+				ASSERT(be32_to_cpu(agi->agi_freecount) > 0);
+				*alloc_done = B_TRUE;
+				*IO_agbp = agbp;
+				*inop = NULLFSINO;
+				return 0;
+			}
+		}
+		/*
+		 * If it failed, give up on this ag.
+		 */
+		xfs_trans_brelse(tp, agbp);
+		/*
+		 * Go on to the next ag: get its ag header.
+		 */
+nextag:
+		if (++tagno == agcount)
+			tagno = 0;
+		if (tagno == agno) {
+			*inop = NULLFSINO;
+			return noroom ? ENOSPC : 0;
+		}
+		pag = xfs_perag_get(mp, tagno);
+		if (pag->pagi_inodeok == 0) {
+			xfs_perag_put(pag);
+			goto nextag;
+		}
+		error = xfs_ialloc_read_agi(mp, tp, tagno, &agbp);
+		xfs_perag_put(pag);
+		if (error)
+			goto nextag;
+		agi = XFS_BUF_TO_AGI(agbp);
+		ASSERT(be32_to_cpu(agi->agi_magicnum) == XFS_AGI_MAGIC);
+	}
+	/*
+	 * Here with an allocation group that has a free inode.
+	 * Reset agno since we may have chosen a new ag in the
+	 * loop above.
+	 */
+	agno = tagno;
+	*IO_agbp = NULL;
+	pag = xfs_perag_get(mp, agno);
+
+ restart_pagno:
+	cur = xfs_inobt_init_cursor(mp, tp, agbp, be32_to_cpu(agi->agi_seqno));
+	/*
+	 * If pagino is 0 (this is the root inode allocation) use newino.
+	 * This must work because we've just allocated some.
+	 */
+	if (!pagino)
+		pagino = be32_to_cpu(agi->agi_newino);
+
+	error = xfs_check_agi_freecount(cur, agi);
+	if (error)
+		goto error0;
+
+	/*
+	 * If in the same AG as the parent, try to get near the parent.
+	 */
+	if (pagno == agno) {
+		int		doneleft;	/* done, to the left */
+		int		doneright;	/* done, to the right */
+		int		searchdistance = 10;
+
+		error = xfs_inobt_lookup(cur, pagino, XFS_LOOKUP_LE, &i);
+		if (error)
+			goto error0;
+		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+
+		error = xfs_inobt_get_rec(cur, &rec, &j);
+		if (error)
+			goto error0;
+		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+
+		if (rec.ir_freecount > 0) {
+			/*
+			 * Found a free inode in the same chunk
+			 * as the parent, done.
+			 */
+			goto alloc_inode;
+		}
+
+
+		/*
+		 * In the same AG as parent, but parent's chunk is full.
+		 */
+
+		/* duplicate the cursor, search left & right simultaneously */
+		error = xfs_btree_dup_cursor(cur, &tcur);
+		if (error)
+			goto error0;
+
+		/*
+		 * Skip to last blocks looked up if same parent inode.
+		 */
+		if (pagino != NULLAGINO &&
+		    pag->pagl_pagino == pagino &&
+		    pag->pagl_leftrec != NULLAGINO &&
+		    pag->pagl_rightrec != NULLAGINO) {
+			error = xfs_ialloc_get_rec(tcur, pag->pagl_leftrec,
+						   &trec, &doneleft, 1);
+			if (error)
+				goto error1;
+
+			error = xfs_ialloc_get_rec(cur, pag->pagl_rightrec,
+						   &rec, &doneright, 0);
+			if (error)
+				goto error1;
+		} else {
+			/* search left with tcur, back up 1 record */
+			error = xfs_ialloc_next_rec(tcur, &trec, &doneleft, 1);
+			if (error)
+				goto error1;
+
+			/* search right with cur, go forward 1 record. */
+			error = xfs_ialloc_next_rec(cur, &rec, &doneright, 0);
+			if (error)
+				goto error1;
+		}
+
+		/*
+		 * Loop until we find an inode chunk with a free inode.
+		 */
+		while (!doneleft || !doneright) {
+			int	useleft;  /* using left inode chunk this time */
+
+			if (!--searchdistance) {
+				/*
+				 * Not in range - save last search
+				 * location and allocate a new inode
+				 */
+				xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
+				pag->pagl_leftrec = trec.ir_startino;
+				pag->pagl_rightrec = rec.ir_startino;
+				pag->pagl_pagino = pagino;
+				goto newino;
+			}
+
+			/* figure out the closer block if both are valid. */
+			if (!doneleft && !doneright) {
+				useleft = pagino -
+				 (trec.ir_startino + XFS_INODES_PER_CHUNK - 1) <
+				  rec.ir_startino - pagino;
+			} else {
+				useleft = !doneleft;
+			}
+
+			/* free inodes to the left? */
+			if (useleft && trec.ir_freecount) {
+				rec = trec;
+				xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+				cur = tcur;
+
+				pag->pagl_leftrec = trec.ir_startino;
+				pag->pagl_rightrec = rec.ir_startino;
+				pag->pagl_pagino = pagino;
+				goto alloc_inode;
+			}
+
+			/* free inodes to the right? */
+			if (!useleft && rec.ir_freecount) {
+				xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
+
+				pag->pagl_leftrec = trec.ir_startino;
+				pag->pagl_rightrec = rec.ir_startino;
+				pag->pagl_pagino = pagino;
+				goto alloc_inode;
+			}
+
+			/* get next record to check */
+			if (useleft) {
+				error = xfs_ialloc_next_rec(tcur, &trec,
+								 &doneleft, 1);
+			} else {
+				error = xfs_ialloc_next_rec(cur, &rec,
+								 &doneright, 0);
+			}
+			if (error)
+				goto error1;
+		}
+
+		/*
+		 * We've reached the end of the btree. because
+		 * we are only searching a small chunk of the
+		 * btree each search, there is obviously free
+		 * inodes closer to the parent inode than we
+		 * are now. restart the search again.
+		 */
+		pag->pagl_pagino = NULLAGINO;
+		pag->pagl_leftrec = NULLAGINO;
+		pag->pagl_rightrec = NULLAGINO;
+		xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
+		xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+		goto restart_pagno;
+	}
+
+	/*
+	 * In a different AG from the parent.
+	 * See if the most recently allocated block has any free.
+	 */
+newino:
+	if (be32_to_cpu(agi->agi_newino) != NULLAGINO) {
+		error = xfs_inobt_lookup(cur, be32_to_cpu(agi->agi_newino),
+					 XFS_LOOKUP_EQ, &i);
+		if (error)
+			goto error0;
+
+		if (i == 1) {
+			error = xfs_inobt_get_rec(cur, &rec, &j);
+			if (error)
+				goto error0;
+
+			if (j == 1 && rec.ir_freecount > 0) {
+				/*
+				 * The last chunk allocated in the group
+				 * still has a free inode.
+				 */
+				goto alloc_inode;
+			}
+		}
+	}
+
+	/*
+	 * None left in the last group, search the whole AG
+	 */
+	error = xfs_inobt_lookup(cur, 0, XFS_LOOKUP_GE, &i);
+	if (error)
+		goto error0;
+	XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+
+	for (;;) {
+		error = xfs_inobt_get_rec(cur, &rec, &i);
+		if (error)
+			goto error0;
+		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+		if (rec.ir_freecount > 0)
+			break;
+		error = xfs_btree_increment(cur, 0, &i);
+		if (error)
+			goto error0;
+		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+	}
+
+alloc_inode:
+	offset = xfs_ialloc_find_free(&rec.ir_free);
+	ASSERT(offset >= 0);
+	ASSERT(offset < XFS_INODES_PER_CHUNK);
+	ASSERT((XFS_AGINO_TO_OFFSET(mp, rec.ir_startino) %
+				   XFS_INODES_PER_CHUNK) == 0);
+	ino = XFS_AGINO_TO_INO(mp, agno, rec.ir_startino + offset);
+	rec.ir_free &= ~XFS_INOBT_MASK(offset);
+	rec.ir_freecount--;
+	error = xfs_inobt_update(cur, &rec);
+	if (error)
+		goto error0;
+	be32_add_cpu(&agi->agi_freecount, -1);
+	xfs_ialloc_log_agi(tp, agbp, XFS_AGI_FREECOUNT);
+	pag->pagi_freecount--;
+
+	error = xfs_check_agi_freecount(cur, agi);
+	if (error)
+		goto error0;
+
+	xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+	xfs_trans_mod_sb(tp, XFS_TRANS_SB_IFREE, -1);
+	xfs_perag_put(pag);
+	*inop = ino;
+	return 0;
+error1:
+	xfs_btree_del_cursor(tcur, XFS_BTREE_ERROR);
+error0:
+	xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
+	xfs_perag_put(pag);
+	return error;
+}
+
+/*
+ * Free disk inode.  Carefully avoids touching the incore inode, all
+ * manipulations incore are the caller's responsibility.
+ * The on-disk inode is not changed by this operation, only the
+ * btree (free inode mask) is changed.
+ */
+int
+xfs_difree(
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_ino_t	inode,		/* inode to be freed */
+	xfs_bmap_free_t	*flist,		/* extents to free */
+	int		*delete,	/* set if inode cluster was deleted */
+	xfs_ino_t	*first_ino)	/* first inode in deleted cluster */
+{
+	/* REFERENCED */
+	xfs_agblock_t	agbno;	/* block number containing inode */
+	xfs_buf_t	*agbp;	/* buffer containing allocation group header */
+	xfs_agino_t	agino;	/* inode number relative to allocation group */
+	xfs_agnumber_t	agno;	/* allocation group number */
+	xfs_agi_t	*agi;	/* allocation group header */
+	xfs_btree_cur_t	*cur;	/* inode btree cursor */
+	int		error;	/* error return value */
+	int		i;	/* result code */
+	int		ilen;	/* inodes in an inode cluster */
+	xfs_mount_t	*mp;	/* mount structure for filesystem */
+	int		off;	/* offset of inode in inode chunk */
+	xfs_inobt_rec_incore_t rec;	/* btree record */
+	struct xfs_perag *pag;
+
+	mp = tp->t_mountp;
+
+	/*
+	 * Break up inode number into its components.
+	 */
+	agno = XFS_INO_TO_AGNO(mp, inode);
+	if (agno >= mp->m_sb.sb_agcount)  {
+		xfs_warn(mp, "%s: agno >= mp->m_sb.sb_agcount (%d >= %d).",
+			__func__, agno, mp->m_sb.sb_agcount);
+		ASSERT(0);
+		return XFS_ERROR(EINVAL);
+	}
+	agino = XFS_INO_TO_AGINO(mp, inode);
+	if (inode != XFS_AGINO_TO_INO(mp, agno, agino))  {
+		xfs_warn(mp, "%s: inode != XFS_AGINO_TO_INO() (%llu != %llu).",
+			__func__, (unsigned long long)inode,
+			(unsigned long long)XFS_AGINO_TO_INO(mp, agno, agino));
+		ASSERT(0);
+		return XFS_ERROR(EINVAL);
+	}
+	agbno = XFS_AGINO_TO_AGBNO(mp, agino);
+	if (agbno >= mp->m_sb.sb_agblocks)  {
+		xfs_warn(mp, "%s: agbno >= mp->m_sb.sb_agblocks (%d >= %d).",
+			__func__, agbno, mp->m_sb.sb_agblocks);
+		ASSERT(0);
+		return XFS_ERROR(EINVAL);
+	}
+	/*
+	 * Get the allocation group header.
+	 */
+	error = xfs_ialloc_read_agi(mp, tp, agno, &agbp);
+	if (error) {
+		xfs_warn(mp, "%s: xfs_ialloc_read_agi() returned error %d.",
+			__func__, error);
+		return error;
+	}
+	agi = XFS_BUF_TO_AGI(agbp);
+	ASSERT(be32_to_cpu(agi->agi_magicnum) == XFS_AGI_MAGIC);
+	ASSERT(agbno < be32_to_cpu(agi->agi_length));
+	/*
+	 * Initialize the cursor.
+	 */
+	cur = xfs_inobt_init_cursor(mp, tp, agbp, agno);
+
+	error = xfs_check_agi_freecount(cur, agi);
+	if (error)
+		goto error0;
+
+	/*
+	 * Look for the entry describing this inode.
+	 */
+	if ((error = xfs_inobt_lookup(cur, agino, XFS_LOOKUP_LE, &i))) {
+		xfs_warn(mp, "%s: xfs_inobt_lookup() returned error %d.",
+			__func__, error);
+		goto error0;
+	}
+	XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+	error = xfs_inobt_get_rec(cur, &rec, &i);
+	if (error) {
+		xfs_warn(mp, "%s: xfs_inobt_get_rec() returned error %d.",
+			__func__, error);
+		goto error0;
+	}
+	XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+	/*
+	 * Get the offset in the inode chunk.
+	 */
+	off = agino - rec.ir_startino;
+	ASSERT(off >= 0 && off < XFS_INODES_PER_CHUNK);
+	ASSERT(!(rec.ir_free & XFS_INOBT_MASK(off)));
+	/*
+	 * Mark the inode free & increment the count.
+	 */
+	rec.ir_free |= XFS_INOBT_MASK(off);
+	rec.ir_freecount++;
+
+	/*
+	 * When an inode cluster is free, it becomes eligible for removal
+	 */
+	if (!(mp->m_flags & XFS_MOUNT_IKEEP) &&
+	    (rec.ir_freecount == XFS_IALLOC_INODES(mp))) {
+
+		*delete = 1;
+		*first_ino = XFS_AGINO_TO_INO(mp, agno, rec.ir_startino);
+
+		/*
+		 * Remove the inode cluster from the AGI B+Tree, adjust the
+		 * AGI and Superblock inode counts, and mark the disk space
+		 * to be freed when the transaction is committed.
+		 */
+		ilen = XFS_IALLOC_INODES(mp);
+		be32_add_cpu(&agi->agi_count, -ilen);
+		be32_add_cpu(&agi->agi_freecount, -(ilen - 1));
+		xfs_ialloc_log_agi(tp, agbp, XFS_AGI_COUNT | XFS_AGI_FREECOUNT);
+		pag = xfs_perag_get(mp, agno);
+		pag->pagi_freecount -= ilen - 1;
+		xfs_perag_put(pag);
+		xfs_trans_mod_sb(tp, XFS_TRANS_SB_ICOUNT, -ilen);
+		xfs_trans_mod_sb(tp, XFS_TRANS_SB_IFREE, -(ilen - 1));
+
+		if ((error = xfs_btree_delete(cur, &i))) {
+			xfs_warn(mp, "%s: xfs_btree_delete returned error %d.",
+				__func__, error);
+			goto error0;
+		}
+
+		xfs_bmap_add_free(XFS_AGB_TO_FSB(mp,
+				agno, XFS_INO_TO_AGBNO(mp,rec.ir_startino)),
+				XFS_IALLOC_BLOCKS(mp), flist, mp);
+	} else {
+		*delete = 0;
+
+		error = xfs_inobt_update(cur, &rec);
+		if (error) {
+			xfs_warn(mp, "%s: xfs_inobt_update returned error %d.",
+				__func__, error);
+			goto error0;
+		}
+
+		/* 
+		 * Change the inode free counts and log the ag/sb changes.
+		 */
+		be32_add_cpu(&agi->agi_freecount, 1);
+		xfs_ialloc_log_agi(tp, agbp, XFS_AGI_FREECOUNT);
+		pag = xfs_perag_get(mp, agno);
+		pag->pagi_freecount++;
+		xfs_perag_put(pag);
+		xfs_trans_mod_sb(tp, XFS_TRANS_SB_IFREE, 1);
+	}
+
+	error = xfs_check_agi_freecount(cur, agi);
+	if (error)
+		goto error0;
+
+	xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+	return 0;
+
+error0:
+	xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
+	return error;
+}
+
+STATIC int
+xfs_imap_lookup(
+	struct xfs_mount	*mp,
+	struct xfs_trans	*tp,
+	xfs_agnumber_t		agno,
+	xfs_agino_t		agino,
+	xfs_agblock_t		agbno,
+	xfs_agblock_t		*chunk_agbno,
+	xfs_agblock_t		*offset_agbno,
+	int			flags)
+{
+	struct xfs_inobt_rec_incore rec;
+	struct xfs_btree_cur	*cur;
+	struct xfs_buf		*agbp;
+	int			error;
+	int			i;
+
+	error = xfs_ialloc_read_agi(mp, tp, agno, &agbp);
+	if (error) {
+		xfs_alert(mp,
+			"%s: xfs_ialloc_read_agi() returned error %d, agno %d",
+			__func__, error, agno);
+		return error;
+	}
+
+	/*
+	 * Lookup the inode record for the given agino. If the record cannot be
+	 * found, then it's an invalid inode number and we should abort. Once
+	 * we have a record, we need to ensure it contains the inode number
+	 * we are looking up.
+	 */
+	cur = xfs_inobt_init_cursor(mp, tp, agbp, agno);
+	error = xfs_inobt_lookup(cur, agino, XFS_LOOKUP_LE, &i);
+	if (!error) {
+		if (i)
+			error = xfs_inobt_get_rec(cur, &rec, &i);
+		if (!error && i == 0)
+			error = EINVAL;
+	}
+
+	xfs_trans_brelse(tp, agbp);
+	xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+	if (error)
+		return error;
+
+	/* check that the returned record contains the required inode */
+	if (rec.ir_startino > agino ||
+	    rec.ir_startino + XFS_IALLOC_INODES(mp) <= agino)
+		return EINVAL;
+
+	/* for untrusted inodes check it is allocated first */
+	if ((flags & XFS_IGET_UNTRUSTED) &&
+	    (rec.ir_free & XFS_INOBT_MASK(agino - rec.ir_startino)))
+		return EINVAL;
+
+	*chunk_agbno = XFS_AGINO_TO_AGBNO(mp, rec.ir_startino);
+	*offset_agbno = agbno - *chunk_agbno;
+	return 0;
+}
+
+/*
+ * Return the location of the inode in imap, for mapping it into a buffer.
+ */
+int
+xfs_imap(
+	xfs_mount_t	 *mp,	/* file system mount structure */
+	xfs_trans_t	 *tp,	/* transaction pointer */
+	xfs_ino_t	ino,	/* inode to locate */
+	struct xfs_imap	*imap,	/* location map structure */
+	uint		flags)	/* flags for inode btree lookup */
+{
+	xfs_agblock_t	agbno;	/* block number of inode in the alloc group */
+	xfs_agino_t	agino;	/* inode number within alloc group */
+	xfs_agnumber_t	agno;	/* allocation group number */
+	int		blks_per_cluster; /* num blocks per inode cluster */
+	xfs_agblock_t	chunk_agbno;	/* first block in inode chunk */
+	xfs_agblock_t	cluster_agbno;	/* first block in inode cluster */
+	int		error;	/* error code */
+	int		offset;	/* index of inode in its buffer */
+	int		offset_agbno;	/* blks from chunk start to inode */
+
+	ASSERT(ino != NULLFSINO);
+
+	/*
+	 * Split up the inode number into its parts.
+	 */
+	agno = XFS_INO_TO_AGNO(mp, ino);
+	agino = XFS_INO_TO_AGINO(mp, ino);
+	agbno = XFS_AGINO_TO_AGBNO(mp, agino);
+	if (agno >= mp->m_sb.sb_agcount || agbno >= mp->m_sb.sb_agblocks ||
+	    ino != XFS_AGINO_TO_INO(mp, agno, agino)) {
+#ifdef DEBUG
+		/*
+		 * Don't output diagnostic information for untrusted inodes
+		 * as they can be invalid without implying corruption.
+		 */
+		if (flags & XFS_IGET_UNTRUSTED)
+			return XFS_ERROR(EINVAL);
+		if (agno >= mp->m_sb.sb_agcount) {
+			xfs_alert(mp,
+				"%s: agno (%d) >= mp->m_sb.sb_agcount (%d)",
+				__func__, agno, mp->m_sb.sb_agcount);
+		}
+		if (agbno >= mp->m_sb.sb_agblocks) {
+			xfs_alert(mp,
+		"%s: agbno (0x%llx) >= mp->m_sb.sb_agblocks (0x%lx)",
+				__func__, (unsigned long long)agbno,
+				(unsigned long)mp->m_sb.sb_agblocks);
+		}
+		if (ino != XFS_AGINO_TO_INO(mp, agno, agino)) {
+			xfs_alert(mp,
+		"%s: ino (0x%llx) != XFS_AGINO_TO_INO() (0x%llx)",
+				__func__, ino,
+				XFS_AGINO_TO_INO(mp, agno, agino));
+		}
+		xfs_stack_trace();
+#endif /* DEBUG */
+		return XFS_ERROR(EINVAL);
+	}
+
+	blks_per_cluster = XFS_INODE_CLUSTER_SIZE(mp) >> mp->m_sb.sb_blocklog;
+
+	/*
+	 * For bulkstat and handle lookups, we have an untrusted inode number
+	 * that we have to verify is valid. We cannot do this just by reading
+	 * the inode buffer as it may have been unlinked and removed leaving
+	 * inodes in stale state on disk. Hence we have to do a btree lookup
+	 * in all cases where an untrusted inode number is passed.
+	 */
+	if (flags & XFS_IGET_UNTRUSTED) {
+		error = xfs_imap_lookup(mp, tp, agno, agino, agbno,
+					&chunk_agbno, &offset_agbno, flags);
+		if (error)
+			return error;
+		goto out_map;
+	}
+
+	/*
+	 * If the inode cluster size is the same as the blocksize or
+	 * smaller we get to the buffer by simple arithmetics.
+	 */
+	if (XFS_INODE_CLUSTER_SIZE(mp) <= mp->m_sb.sb_blocksize) {
+		offset = XFS_INO_TO_OFFSET(mp, ino);
+		ASSERT(offset < mp->m_sb.sb_inopblock);
+
+		imap->im_blkno = XFS_AGB_TO_DADDR(mp, agno, agbno);
+		imap->im_len = XFS_FSB_TO_BB(mp, 1);
+		imap->im_boffset = (ushort)(offset << mp->m_sb.sb_inodelog);
+		return 0;
+	}
+
+	/*
+	 * If the inode chunks are aligned then use simple maths to
+	 * find the location. Otherwise we have to do a btree
+	 * lookup to find the location.
+	 */
+	if (mp->m_inoalign_mask) {
+		offset_agbno = agbno & mp->m_inoalign_mask;
+		chunk_agbno = agbno - offset_agbno;
+	} else {
+		error = xfs_imap_lookup(mp, tp, agno, agino, agbno,
+					&chunk_agbno, &offset_agbno, flags);
+		if (error)
+			return error;
+	}
+
+out_map:
+	ASSERT(agbno >= chunk_agbno);
+	cluster_agbno = chunk_agbno +
+		((offset_agbno / blks_per_cluster) * blks_per_cluster);
+	offset = ((agbno - cluster_agbno) * mp->m_sb.sb_inopblock) +
+		XFS_INO_TO_OFFSET(mp, ino);
+
+	imap->im_blkno = XFS_AGB_TO_DADDR(mp, agno, cluster_agbno);
+	imap->im_len = XFS_FSB_TO_BB(mp, blks_per_cluster);
+	imap->im_boffset = (ushort)(offset << mp->m_sb.sb_inodelog);
+
+	/*
+	 * If the inode number maps to a block outside the bounds
+	 * of the file system then return NULL rather than calling
+	 * read_buf and panicing when we get an error from the
+	 * driver.
+	 */
+	if ((imap->im_blkno + imap->im_len) >
+	    XFS_FSB_TO_BB(mp, mp->m_sb.sb_dblocks)) {
+		xfs_alert(mp,
+	"%s: (im_blkno (0x%llx) + im_len (0x%llx)) > sb_dblocks (0x%llx)",
+			__func__, (unsigned long long) imap->im_blkno,
+			(unsigned long long) imap->im_len,
+			XFS_FSB_TO_BB(mp, mp->m_sb.sb_dblocks));
+		return XFS_ERROR(EINVAL);
+	}
+	return 0;
+}
+
+/*
+ * Compute and fill in value of m_in_maxlevels.
+ */
+void
+xfs_ialloc_compute_maxlevels(
+	xfs_mount_t	*mp)		/* file system mount structure */
+{
+	int		level;
+	uint		maxblocks;
+	uint		maxleafents;
+	int		minleafrecs;
+	int		minnoderecs;
+
+	maxleafents = (1LL << XFS_INO_AGINO_BITS(mp)) >>
+		XFS_INODES_PER_CHUNK_LOG;
+	minleafrecs = mp->m_alloc_mnr[0];
+	minnoderecs = mp->m_alloc_mnr[1];
+	maxblocks = (maxleafents + minleafrecs - 1) / minleafrecs;
+	for (level = 1; maxblocks > 1; level++)
+		maxblocks = (maxblocks + minnoderecs - 1) / minnoderecs;
+	mp->m_in_maxlevels = level;
+}
+
+/*
+ * Log specified fields for the ag hdr (inode section)
+ */
+void
+xfs_ialloc_log_agi(
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_buf_t	*bp,		/* allocation group header buffer */
+	int		fields)		/* bitmask of fields to log */
+{
+	int			first;		/* first byte number */
+	int			last;		/* last byte number */
+	static const short	offsets[] = {	/* field starting offsets */
+					/* keep in sync with bit definitions */
+		offsetof(xfs_agi_t, agi_magicnum),
+		offsetof(xfs_agi_t, agi_versionnum),
+		offsetof(xfs_agi_t, agi_seqno),
+		offsetof(xfs_agi_t, agi_length),
+		offsetof(xfs_agi_t, agi_count),
+		offsetof(xfs_agi_t, agi_root),
+		offsetof(xfs_agi_t, agi_level),
+		offsetof(xfs_agi_t, agi_freecount),
+		offsetof(xfs_agi_t, agi_newino),
+		offsetof(xfs_agi_t, agi_dirino),
+		offsetof(xfs_agi_t, agi_unlinked),
+		sizeof(xfs_agi_t)
+	};
+#ifdef DEBUG
+	xfs_agi_t		*agi;	/* allocation group header */
+
+	agi = XFS_BUF_TO_AGI(bp);
+	ASSERT(be32_to_cpu(agi->agi_magicnum) == XFS_AGI_MAGIC);
+#endif
+	/*
+	 * Compute byte offsets for the first and last fields.
+	 */
+	xfs_btree_offsets(fields, offsets, XFS_AGI_NUM_BITS, &first, &last);
+	/*
+	 * Log the allocation group inode header buffer.
+	 */
+	xfs_trans_log_buf(tp, bp, first, last);
+}
+
+#ifdef DEBUG
+STATIC void
+xfs_check_agi_unlinked(
+	struct xfs_agi		*agi)
+{
+	int			i;
+
+	for (i = 0; i < XFS_AGI_UNLINKED_BUCKETS; i++)
+		ASSERT(agi->agi_unlinked[i]);
+}
+#else
+#define xfs_check_agi_unlinked(agi)
+#endif
+
+/*
+ * Read in the allocation group header (inode allocation section)
+ */
+int
+xfs_read_agi(
+	struct xfs_mount	*mp,	/* file system mount structure */
+	struct xfs_trans	*tp,	/* transaction pointer */
+	xfs_agnumber_t		agno,	/* allocation group number */
+	struct xfs_buf		**bpp)	/* allocation group hdr buf */
+{
+	struct xfs_agi		*agi;	/* allocation group header */
+	int			agi_ok;	/* agi is consistent */
+	int			error;
+
+	ASSERT(agno != NULLAGNUMBER);
+
+	error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp,
+			XFS_AG_DADDR(mp, agno, XFS_AGI_DADDR(mp)),
+			XFS_FSS_TO_BB(mp, 1), 0, bpp);
+	if (error)
+		return error;
+
+	ASSERT(*bpp && !XFS_BUF_GETERROR(*bpp));
+	agi = XFS_BUF_TO_AGI(*bpp);
+
+	/*
+	 * Validate the magic number of the agi block.
+	 */
+	agi_ok = be32_to_cpu(agi->agi_magicnum) == XFS_AGI_MAGIC &&
+		XFS_AGI_GOOD_VERSION(be32_to_cpu(agi->agi_versionnum)) &&
+		be32_to_cpu(agi->agi_seqno) == agno;
+	if (unlikely(XFS_TEST_ERROR(!agi_ok, mp, XFS_ERRTAG_IALLOC_READ_AGI,
+			XFS_RANDOM_IALLOC_READ_AGI))) {
+		XFS_CORRUPTION_ERROR("xfs_read_agi", XFS_ERRLEVEL_LOW,
+				     mp, agi);
+		xfs_trans_brelse(tp, *bpp);
+		return XFS_ERROR(EFSCORRUPTED);
+	}
+
+	XFS_BUF_SET_VTYPE_REF(*bpp, B_FS_AGI, XFS_AGI_REF);
+
+	xfs_check_agi_unlinked(agi);
+	return 0;
+}
+
+int
+xfs_ialloc_read_agi(
+	struct xfs_mount	*mp,	/* file system mount structure */
+	struct xfs_trans	*tp,	/* transaction pointer */
+	xfs_agnumber_t		agno,	/* allocation group number */
+	struct xfs_buf		**bpp)	/* allocation group hdr buf */
+{
+	struct xfs_agi		*agi;	/* allocation group header */
+	struct xfs_perag	*pag;	/* per allocation group data */
+	int			error;
+
+	error = xfs_read_agi(mp, tp, agno, bpp);
+	if (error)
+		return error;
+
+	agi = XFS_BUF_TO_AGI(*bpp);
+	pag = xfs_perag_get(mp, agno);
+	if (!pag->pagi_init) {
+		pag->pagi_freecount = be32_to_cpu(agi->agi_freecount);
+		pag->pagi_count = be32_to_cpu(agi->agi_count);
+		pag->pagi_init = 1;
+	}
+
+	/*
+	 * It's possible for these to be out of sync if
+	 * we are in the middle of a forced shutdown.
+	 */
+	ASSERT(pag->pagi_freecount == be32_to_cpu(agi->agi_freecount) ||
+		XFS_FORCED_SHUTDOWN(mp));
+	xfs_perag_put(pag);
+	return 0;
+}
+
+/*
+ * Read in the agi to initialise the per-ag data in the mount structure
+ */
+int
+xfs_ialloc_pagi_init(
+	xfs_mount_t	*mp,		/* file system mount structure */
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_agnumber_t	agno)		/* allocation group number */
+{
+	xfs_buf_t	*bp = NULL;
+	int		error;
+
+	error = xfs_ialloc_read_agi(mp, tp, agno, &bp);
+	if (error)
+		return error;
+	if (bp)
+		xfs_trans_brelse(tp, bp);
+	return 0;
+}
diff --git a/fs/xfs/xfs_ialloc.h b/fs/xfs/xfs_ialloc.h
new file mode 100644
index 00000000..bb538547
--- /dev/null
+++ b/fs/xfs/xfs_ialloc.h
@@ -0,0 +1,164 @@
+/*
+ * Copyright (c) 2000,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_IALLOC_H__
+#define	__XFS_IALLOC_H__
+
+struct xfs_buf;
+struct xfs_dinode;
+struct xfs_imap;
+struct xfs_mount;
+struct xfs_trans;
+
+/*
+ * Allocation parameters for inode allocation.
+ */
+#define	XFS_IALLOC_INODES(mp)	(mp)->m_ialloc_inos
+#define	XFS_IALLOC_BLOCKS(mp)	(mp)->m_ialloc_blks
+
+/*
+ * Move inodes in clusters of this size.
+ */
+#define	XFS_INODE_BIG_CLUSTER_SIZE	8192
+#define	XFS_INODE_CLUSTER_SIZE(mp)	(mp)->m_inode_cluster_size
+
+/*
+ * Make an inode pointer out of the buffer/offset.
+ */
+static inline struct xfs_dinode *
+xfs_make_iptr(struct xfs_mount *mp, struct xfs_buf *b, int o)
+{
+	return (xfs_dinode_t *)
+		(xfs_buf_offset(b, o << (mp)->m_sb.sb_inodelog));
+}
+
+/*
+ * Find a free (set) bit in the inode bitmask.
+ */
+static inline int xfs_ialloc_find_free(xfs_inofree_t *fp)
+{
+	return xfs_lowbit64(*fp);
+}
+
+
+/*
+ * Allocate an inode on disk.
+ * Mode is used to tell whether the new inode will need space, and whether
+ * it is a directory.
+ *
+ * To work within the constraint of one allocation per transaction,
+ * xfs_dialloc() is designed to be called twice if it has to do an
+ * allocation to make more free inodes.  If an inode is
+ * available without an allocation, agbp would be set to the current
+ * agbp and alloc_done set to false.
+ * If an allocation needed to be done, agbp would be set to the
+ * inode header of the allocation group and alloc_done set to true.
+ * The caller should then commit the current transaction and allocate a new
+ * transaction.  xfs_dialloc() should then be called again with
+ * the agbp value returned from the previous call.
+ *
+ * Once we successfully pick an inode its number is returned and the
+ * on-disk data structures are updated.  The inode itself is not read
+ * in, since doing so would break ordering constraints with xfs_reclaim.
+ *
+ * *agbp should be set to NULL on the first call, *alloc_done set to FALSE.
+ */
+int					/* error */
+xfs_dialloc(
+	struct xfs_trans *tp,		/* transaction pointer */
+	xfs_ino_t	parent,		/* parent inode (directory) */
+	mode_t		mode,		/* mode bits for new inode */
+	int		okalloc,	/* ok to allocate more space */
+	struct xfs_buf	**agbp,		/* buf for a.g. inode header */
+	boolean_t	*alloc_done,	/* an allocation was done to replenish
+					   the free inodes */
+	xfs_ino_t	*inop);		/* inode number allocated */
+
+/*
+ * Free disk inode.  Carefully avoids touching the incore inode, all
+ * manipulations incore are the caller's responsibility.
+ * The on-disk inode is not changed by this operation, only the
+ * btree (free inode mask) is changed.
+ */
+int					/* error */
+xfs_difree(
+	struct xfs_trans *tp,		/* transaction pointer */
+	xfs_ino_t	inode,		/* inode to be freed */
+	struct xfs_bmap_free *flist,	/* extents to free */
+	int		*delete,	/* set if inode cluster was deleted */
+	xfs_ino_t	*first_ino);	/* first inode in deleted cluster */
+
+/*
+ * Return the location of the inode in imap, for mapping it into a buffer.
+ */
+int
+xfs_imap(
+	struct xfs_mount *mp,		/* file system mount structure */
+	struct xfs_trans *tp,		/* transaction pointer */
+	xfs_ino_t	ino,		/* inode to locate */
+	struct xfs_imap	*imap,		/* location map structure */
+	uint		flags);		/* flags for inode btree lookup */
+
+/*
+ * Compute and fill in value of m_in_maxlevels.
+ */
+void
+xfs_ialloc_compute_maxlevels(
+	struct xfs_mount *mp);		/* file system mount structure */
+
+/*
+ * Log specified fields for the ag hdr (inode section)
+ */
+void
+xfs_ialloc_log_agi(
+	struct xfs_trans *tp,		/* transaction pointer */
+	struct xfs_buf	*bp,		/* allocation group header buffer */
+	int		fields);	/* bitmask of fields to log */
+
+/*
+ * Read in the allocation group header (inode allocation section)
+ */
+int					/* error */
+xfs_ialloc_read_agi(
+	struct xfs_mount *mp,		/* file system mount structure */
+	struct xfs_trans *tp,		/* transaction pointer */
+	xfs_agnumber_t	agno,		/* allocation group number */
+	struct xfs_buf	**bpp);		/* allocation group hdr buf */
+
+/*
+ * Read in the allocation group header to initialise the per-ag data
+ * in the mount structure
+ */
+int
+xfs_ialloc_pagi_init(
+	struct xfs_mount *mp,		/* file system mount structure */
+	struct xfs_trans *tp,		/* transaction pointer */
+        xfs_agnumber_t  agno);		/* allocation group number */
+
+/*
+ * Lookup a record by ino in the btree given by cur.
+ */
+int xfs_inobt_lookup(struct xfs_btree_cur *cur, xfs_agino_t ino,
+		xfs_lookup_t dir, int *stat);
+
+/*
+ * Get the data from the pointed-to record.
+ */
+extern int xfs_inobt_get_rec(struct xfs_btree_cur *cur,
+		xfs_inobt_rec_incore_t *rec, int *stat);
+
+#endif	/* __XFS_IALLOC_H__ */
diff --git a/fs/xfs/xfs_ialloc_btree.c b/fs/xfs/xfs_ialloc_btree.c
new file mode 100644
index 00000000..16921f55
--- /dev/null
+++ b/fs/xfs/xfs_ialloc_btree.c
@@ -0,0 +1,346 @@
+/*
+ * Copyright (c) 2000-2001,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_mount.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_alloc_btree.h"
+#include "xfs_ialloc_btree.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_btree.h"
+#include "xfs_btree_trace.h"
+#include "xfs_ialloc.h"
+#include "xfs_alloc.h"
+#include "xfs_error.h"
+
+
+STATIC int
+xfs_inobt_get_minrecs(
+	struct xfs_btree_cur	*cur,
+	int			level)
+{
+	return cur->bc_mp->m_inobt_mnr[level != 0];
+}
+
+STATIC struct xfs_btree_cur *
+xfs_inobt_dup_cursor(
+	struct xfs_btree_cur	*cur)
+{
+	return xfs_inobt_init_cursor(cur->bc_mp, cur->bc_tp,
+			cur->bc_private.a.agbp, cur->bc_private.a.agno);
+}
+
+STATIC void
+xfs_inobt_set_root(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*nptr,
+	int			inc)	/* level change */
+{
+	struct xfs_buf		*agbp = cur->bc_private.a.agbp;
+	struct xfs_agi		*agi = XFS_BUF_TO_AGI(agbp);
+
+	agi->agi_root = nptr->s;
+	be32_add_cpu(&agi->agi_level, inc);
+	xfs_ialloc_log_agi(cur->bc_tp, agbp, XFS_AGI_ROOT | XFS_AGI_LEVEL);
+}
+
+STATIC int
+xfs_inobt_alloc_block(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*start,
+	union xfs_btree_ptr	*new,
+	int			length,
+	int			*stat)
+{
+	xfs_alloc_arg_t		args;		/* block allocation args */
+	int			error;		/* error return value */
+	xfs_agblock_t		sbno = be32_to_cpu(start->s);
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+
+	memset(&args, 0, sizeof(args));
+	args.tp = cur->bc_tp;
+	args.mp = cur->bc_mp;
+	args.fsbno = XFS_AGB_TO_FSB(args.mp, cur->bc_private.a.agno, sbno);
+	args.minlen = 1;
+	args.maxlen = 1;
+	args.prod = 1;
+	args.type = XFS_ALLOCTYPE_NEAR_BNO;
+
+	error = xfs_alloc_vextent(&args);
+	if (error) {
+		XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
+		return error;
+	}
+	if (args.fsbno == NULLFSBLOCK) {
+		XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+		*stat = 0;
+		return 0;
+	}
+	ASSERT(args.len == 1);
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+
+	new->s = cpu_to_be32(XFS_FSB_TO_AGBNO(args.mp, args.fsbno));
+	*stat = 1;
+	return 0;
+}
+
+STATIC int
+xfs_inobt_free_block(
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*bp)
+{
+	xfs_fsblock_t		fsbno;
+	int			error;
+
+	fsbno = XFS_DADDR_TO_FSB(cur->bc_mp, XFS_BUF_ADDR(bp));
+	error = xfs_free_extent(cur->bc_tp, fsbno, 1);
+	if (error)
+		return error;
+
+	xfs_trans_binval(cur->bc_tp, bp);
+	return error;
+}
+
+STATIC int
+xfs_inobt_get_maxrecs(
+	struct xfs_btree_cur	*cur,
+	int			level)
+{
+	return cur->bc_mp->m_inobt_mxr[level != 0];
+}
+
+STATIC void
+xfs_inobt_init_key_from_rec(
+	union xfs_btree_key	*key,
+	union xfs_btree_rec	*rec)
+{
+	key->inobt.ir_startino = rec->inobt.ir_startino;
+}
+
+STATIC void
+xfs_inobt_init_rec_from_key(
+	union xfs_btree_key	*key,
+	union xfs_btree_rec	*rec)
+{
+	rec->inobt.ir_startino = key->inobt.ir_startino;
+}
+
+STATIC void
+xfs_inobt_init_rec_from_cur(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_rec	*rec)
+{
+	rec->inobt.ir_startino = cpu_to_be32(cur->bc_rec.i.ir_startino);
+	rec->inobt.ir_freecount = cpu_to_be32(cur->bc_rec.i.ir_freecount);
+	rec->inobt.ir_free = cpu_to_be64(cur->bc_rec.i.ir_free);
+}
+
+/*
+ * initial value of ptr for lookup
+ */
+STATIC void
+xfs_inobt_init_ptr_from_cur(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*ptr)
+{
+	struct xfs_agi		*agi = XFS_BUF_TO_AGI(cur->bc_private.a.agbp);
+
+	ASSERT(cur->bc_private.a.agno == be32_to_cpu(agi->agi_seqno));
+
+	ptr->s = agi->agi_root;
+}
+
+STATIC __int64_t
+xfs_inobt_key_diff(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_key	*key)
+{
+	return (__int64_t)be32_to_cpu(key->inobt.ir_startino) -
+			  cur->bc_rec.i.ir_startino;
+}
+
+#ifdef DEBUG
+STATIC int
+xfs_inobt_keys_inorder(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_key	*k1,
+	union xfs_btree_key	*k2)
+{
+	return be32_to_cpu(k1->inobt.ir_startino) <
+		be32_to_cpu(k2->inobt.ir_startino);
+}
+
+STATIC int
+xfs_inobt_recs_inorder(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_rec	*r1,
+	union xfs_btree_rec	*r2)
+{
+	return be32_to_cpu(r1->inobt.ir_startino) + XFS_INODES_PER_CHUNK <=
+		be32_to_cpu(r2->inobt.ir_startino);
+}
+#endif	/* DEBUG */
+
+#ifdef XFS_BTREE_TRACE
+ktrace_t	*xfs_inobt_trace_buf;
+
+STATIC void
+xfs_inobt_trace_enter(
+	struct xfs_btree_cur	*cur,
+	const char		*func,
+	char			*s,
+	int			type,
+	int			line,
+	__psunsigned_t		a0,
+	__psunsigned_t		a1,
+	__psunsigned_t		a2,
+	__psunsigned_t		a3,
+	__psunsigned_t		a4,
+	__psunsigned_t		a5,
+	__psunsigned_t		a6,
+	__psunsigned_t		a7,
+	__psunsigned_t		a8,
+	__psunsigned_t		a9,
+	__psunsigned_t		a10)
+{
+	ktrace_enter(xfs_inobt_trace_buf, (void *)(__psint_t)type,
+		(void *)func, (void *)s, NULL, (void *)cur,
+		(void *)a0, (void *)a1, (void *)a2, (void *)a3,
+		(void *)a4, (void *)a5, (void *)a6, (void *)a7,
+		(void *)a8, (void *)a9, (void *)a10);
+}
+
+STATIC void
+xfs_inobt_trace_cursor(
+	struct xfs_btree_cur	*cur,
+	__uint32_t		*s0,
+	__uint64_t		*l0,
+	__uint64_t		*l1)
+{
+	*s0 = cur->bc_private.a.agno;
+	*l0 = cur->bc_rec.i.ir_startino;
+	*l1 = cur->bc_rec.i.ir_free;
+}
+
+STATIC void
+xfs_inobt_trace_key(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_key	*key,
+	__uint64_t		*l0,
+	__uint64_t		*l1)
+{
+	*l0 = be32_to_cpu(key->inobt.ir_startino);
+	*l1 = 0;
+}
+
+STATIC void
+xfs_inobt_trace_record(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_rec	*rec,
+	__uint64_t		*l0,
+	__uint64_t		*l1,
+	__uint64_t		*l2)
+{
+	*l0 = be32_to_cpu(rec->inobt.ir_startino);
+	*l1 = be32_to_cpu(rec->inobt.ir_freecount);
+	*l2 = be64_to_cpu(rec->inobt.ir_free);
+}
+#endif /* XFS_BTREE_TRACE */
+
+static const struct xfs_btree_ops xfs_inobt_ops = {
+	.rec_len		= sizeof(xfs_inobt_rec_t),
+	.key_len		= sizeof(xfs_inobt_key_t),
+
+	.dup_cursor		= xfs_inobt_dup_cursor,
+	.set_root		= xfs_inobt_set_root,
+	.alloc_block		= xfs_inobt_alloc_block,
+	.free_block		= xfs_inobt_free_block,
+	.get_minrecs		= xfs_inobt_get_minrecs,
+	.get_maxrecs		= xfs_inobt_get_maxrecs,
+	.init_key_from_rec	= xfs_inobt_init_key_from_rec,
+	.init_rec_from_key	= xfs_inobt_init_rec_from_key,
+	.init_rec_from_cur	= xfs_inobt_init_rec_from_cur,
+	.init_ptr_from_cur	= xfs_inobt_init_ptr_from_cur,
+	.key_diff		= xfs_inobt_key_diff,
+
+#ifdef DEBUG
+	.keys_inorder		= xfs_inobt_keys_inorder,
+	.recs_inorder		= xfs_inobt_recs_inorder,
+#endif
+
+#ifdef XFS_BTREE_TRACE
+	.trace_enter		= xfs_inobt_trace_enter,
+	.trace_cursor		= xfs_inobt_trace_cursor,
+	.trace_key		= xfs_inobt_trace_key,
+	.trace_record		= xfs_inobt_trace_record,
+#endif
+};
+
+/*
+ * Allocate a new inode btree cursor.
+ */
+struct xfs_btree_cur *				/* new inode btree cursor */
+xfs_inobt_init_cursor(
+	struct xfs_mount	*mp,		/* file system mount point */
+	struct xfs_trans	*tp,		/* transaction pointer */
+	struct xfs_buf		*agbp,		/* buffer for agi structure */
+	xfs_agnumber_t		agno)		/* allocation group number */
+{
+	struct xfs_agi		*agi = XFS_BUF_TO_AGI(agbp);
+	struct xfs_btree_cur	*cur;
+
+	cur = kmem_zone_zalloc(xfs_btree_cur_zone, KM_SLEEP);
+
+	cur->bc_tp = tp;
+	cur->bc_mp = mp;
+	cur->bc_nlevels = be32_to_cpu(agi->agi_level);
+	cur->bc_btnum = XFS_BTNUM_INO;
+	cur->bc_blocklog = mp->m_sb.sb_blocklog;
+
+	cur->bc_ops = &xfs_inobt_ops;
+
+	cur->bc_private.a.agbp = agbp;
+	cur->bc_private.a.agno = agno;
+
+	return cur;
+}
+
+/*
+ * Calculate number of records in an inobt btree block.
+ */
+int
+xfs_inobt_maxrecs(
+	struct xfs_mount	*mp,
+	int			blocklen,
+	int			leaf)
+{
+	blocklen -= XFS_INOBT_BLOCK_LEN(mp);
+
+	if (leaf)
+		return blocklen / sizeof(xfs_inobt_rec_t);
+	return blocklen / (sizeof(xfs_inobt_key_t) + sizeof(xfs_inobt_ptr_t));
+}
diff --git a/fs/xfs/xfs_ialloc_btree.h b/fs/xfs/xfs_ialloc_btree.h
new file mode 100644
index 00000000..f782ad0c
--- /dev/null
+++ b/fs/xfs/xfs_ialloc_btree.h
@@ -0,0 +1,112 @@
+/*
+ * Copyright (c) 2000,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_IALLOC_BTREE_H__
+#define	__XFS_IALLOC_BTREE_H__
+
+/*
+ * Inode map on-disk structures
+ */
+
+struct xfs_buf;
+struct xfs_btree_cur;
+struct xfs_mount;
+
+/*
+ * There is a btree for the inode map per allocation group.
+ */
+#define	XFS_IBT_MAGIC	0x49414254	/* 'IABT' */
+
+typedef	__uint64_t	xfs_inofree_t;
+#define	XFS_INODES_PER_CHUNK		(NBBY * sizeof(xfs_inofree_t))
+#define	XFS_INODES_PER_CHUNK_LOG	(XFS_NBBYLOG + 3)
+#define	XFS_INOBT_ALL_FREE		((xfs_inofree_t)-1)
+#define	XFS_INOBT_MASK(i)		((xfs_inofree_t)1 << (i))
+
+static inline xfs_inofree_t xfs_inobt_maskn(int i, int n)
+{
+	return ((n >= XFS_INODES_PER_CHUNK ? 0 : XFS_INOBT_MASK(n)) - 1) << i;
+}
+
+/*
+ * Data record structure
+ */
+typedef struct xfs_inobt_rec {
+	__be32		ir_startino;	/* starting inode number */
+	__be32		ir_freecount;	/* count of free inodes (set bits) */
+	__be64		ir_free;	/* free inode mask */
+} xfs_inobt_rec_t;
+
+typedef struct xfs_inobt_rec_incore {
+	xfs_agino_t	ir_startino;	/* starting inode number */
+	__int32_t	ir_freecount;	/* count of free inodes (set bits) */
+	xfs_inofree_t	ir_free;	/* free inode mask */
+} xfs_inobt_rec_incore_t;
+
+
+/*
+ * Key structure
+ */
+typedef struct xfs_inobt_key {
+	__be32		ir_startino;	/* starting inode number */
+} xfs_inobt_key_t;
+
+/* btree pointer type */
+typedef __be32 xfs_inobt_ptr_t;
+
+/*
+ * block numbers in the AG.
+ */
+#define	XFS_IBT_BLOCK(mp)		((xfs_agblock_t)(XFS_CNT_BLOCK(mp) + 1))
+#define	XFS_PREALLOC_BLOCKS(mp)		((xfs_agblock_t)(XFS_IBT_BLOCK(mp) + 1))
+
+/*
+ * Btree block header size depends on a superblock flag.
+ *
+ * (not quite yet, but soon)
+ */
+#define XFS_INOBT_BLOCK_LEN(mp)	XFS_BTREE_SBLOCK_LEN
+
+/*
+ * Record, key, and pointer address macros for btree blocks.
+ *
+ * (note that some of these may appear unused, but they are used in userspace)
+ */
+#define XFS_INOBT_REC_ADDR(mp, block, index) \
+	((xfs_inobt_rec_t *) \
+		((char *)(block) + \
+		 XFS_INOBT_BLOCK_LEN(mp) + \
+		 (((index) - 1) * sizeof(xfs_inobt_rec_t))))
+
+#define XFS_INOBT_KEY_ADDR(mp, block, index) \
+	((xfs_inobt_key_t *) \
+		((char *)(block) + \
+		 XFS_INOBT_BLOCK_LEN(mp) + \
+		 ((index) - 1) * sizeof(xfs_inobt_key_t)))
+
+#define XFS_INOBT_PTR_ADDR(mp, block, index, maxrecs) \
+	((xfs_inobt_ptr_t *) \
+		((char *)(block) + \
+		 XFS_INOBT_BLOCK_LEN(mp) + \
+		 (maxrecs) * sizeof(xfs_inobt_key_t) + \
+		 ((index) - 1) * sizeof(xfs_inobt_ptr_t)))
+
+extern struct xfs_btree_cur *xfs_inobt_init_cursor(struct xfs_mount *,
+		struct xfs_trans *, struct xfs_buf *, xfs_agnumber_t);
+extern int xfs_inobt_maxrecs(struct xfs_mount *, int, int);
+
+#endif	/* __XFS_IALLOC_BTREE_H__ */
diff --git a/fs/xfs/xfs_iget.c b/fs/xfs/xfs_iget.c
new file mode 100644
index 00000000..ca752f05
--- /dev/null
+++ b/fs/xfs/xfs_iget.c
@@ -0,0 +1,727 @@
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_acl.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_mount.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_alloc_btree.h"
+#include "xfs_ialloc_btree.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_btree.h"
+#include "xfs_ialloc.h"
+#include "xfs_quota.h"
+#include "xfs_utils.h"
+#include "xfs_trans_priv.h"
+#include "xfs_inode_item.h"
+#include "xfs_bmap.h"
+#include "xfs_btree_trace.h"
+#include "xfs_trace.h"
+
+
+/*
+ * Define xfs inode iolock lockdep classes. We need to ensure that all active
+ * inodes are considered the same for lockdep purposes, including inodes that
+ * are recycled through the XFS_IRECLAIMABLE state. This is the the only way to
+ * guarantee the locks are considered the same when there are multiple lock
+ * initialisation siteѕ. Also, define a reclaimable inode class so it is
+ * obvious in lockdep reports which class the report is against.
+ */
+static struct lock_class_key xfs_iolock_active;
+struct lock_class_key xfs_iolock_reclaimable;
+
+/*
+ * Allocate and initialise an xfs_inode.
+ */
+STATIC struct xfs_inode *
+xfs_inode_alloc(
+	struct xfs_mount	*mp,
+	xfs_ino_t		ino)
+{
+	struct xfs_inode	*ip;
+
+	/*
+	 * if this didn't occur in transactions, we could use
+	 * KM_MAYFAIL and return NULL here on ENOMEM. Set the
+	 * code up to do this anyway.
+	 */
+	ip = kmem_zone_alloc(xfs_inode_zone, KM_SLEEP);
+	if (!ip)
+		return NULL;
+	if (inode_init_always(mp->m_super, VFS_I(ip))) {
+		kmem_zone_free(xfs_inode_zone, ip);
+		return NULL;
+	}
+
+	ASSERT(atomic_read(&ip->i_iocount) == 0);
+	ASSERT(atomic_read(&ip->i_pincount) == 0);
+	ASSERT(!spin_is_locked(&ip->i_flags_lock));
+	ASSERT(completion_done(&ip->i_flush));
+	ASSERT(ip->i_ino == 0);
+
+	mrlock_init(&ip->i_iolock, MRLOCK_BARRIER, "xfsio", ip->i_ino);
+	lockdep_set_class_and_name(&ip->i_iolock.mr_lock,
+			&xfs_iolock_active, "xfs_iolock_active");
+
+	/* initialise the xfs inode */
+	ip->i_ino = ino;
+	ip->i_mount = mp;
+	memset(&ip->i_imap, 0, sizeof(struct xfs_imap));
+	ip->i_afp = NULL;
+	memset(&ip->i_df, 0, sizeof(xfs_ifork_t));
+	ip->i_flags = 0;
+	ip->i_update_core = 0;
+	ip->i_delayed_blks = 0;
+	memset(&ip->i_d, 0, sizeof(xfs_icdinode_t));
+	ip->i_size = 0;
+	ip->i_new_size = 0;
+
+	return ip;
+}
+
+STATIC void
+xfs_inode_free_callback(
+	struct rcu_head		*head)
+{
+	struct inode		*inode = container_of(head, struct inode, i_rcu);
+	struct xfs_inode	*ip = XFS_I(inode);
+
+	INIT_LIST_HEAD(&inode->i_dentry);
+	kmem_zone_free(xfs_inode_zone, ip);
+}
+
+void
+xfs_inode_free(
+	struct xfs_inode	*ip)
+{
+	switch (ip->i_d.di_mode & S_IFMT) {
+	case S_IFREG:
+	case S_IFDIR:
+	case S_IFLNK:
+		xfs_idestroy_fork(ip, XFS_DATA_FORK);
+		break;
+	}
+
+	if (ip->i_afp)
+		xfs_idestroy_fork(ip, XFS_ATTR_FORK);
+
+	if (ip->i_itemp) {
+		/*
+		 * Only if we are shutting down the fs will we see an
+		 * inode still in the AIL. If it is there, we should remove
+		 * it to prevent a use-after-free from occurring.
+		 */
+		xfs_log_item_t	*lip = &ip->i_itemp->ili_item;
+		struct xfs_ail	*ailp = lip->li_ailp;
+
+		ASSERT(((lip->li_flags & XFS_LI_IN_AIL) == 0) ||
+				       XFS_FORCED_SHUTDOWN(ip->i_mount));
+		if (lip->li_flags & XFS_LI_IN_AIL) {
+			spin_lock(&ailp->xa_lock);
+			if (lip->li_flags & XFS_LI_IN_AIL)
+				xfs_trans_ail_delete(ailp, lip);
+			else
+				spin_unlock(&ailp->xa_lock);
+		}
+		xfs_inode_item_destroy(ip);
+		ip->i_itemp = NULL;
+	}
+
+	/* asserts to verify all state is correct here */
+	ASSERT(atomic_read(&ip->i_iocount) == 0);
+	ASSERT(atomic_read(&ip->i_pincount) == 0);
+	ASSERT(!spin_is_locked(&ip->i_flags_lock));
+	ASSERT(completion_done(&ip->i_flush));
+
+	/*
+	 * Because we use RCU freeing we need to ensure the inode always
+	 * appears to be reclaimed with an invalid inode number when in the
+	 * free state. The ip->i_flags_lock provides the barrier against lookup
+	 * races.
+	 */
+	spin_lock(&ip->i_flags_lock);
+	ip->i_flags = XFS_IRECLAIM;
+	ip->i_ino = 0;
+	spin_unlock(&ip->i_flags_lock);
+
+	call_rcu(&VFS_I(ip)->i_rcu, xfs_inode_free_callback);
+}
+
+/*
+ * Check the validity of the inode we just found it the cache
+ */
+static int
+xfs_iget_cache_hit(
+	struct xfs_perag	*pag,
+	struct xfs_inode	*ip,
+	xfs_ino_t		ino,
+	int			flags,
+	int			lock_flags) __releases(RCU)
+{
+	struct inode		*inode = VFS_I(ip);
+	struct xfs_mount	*mp = ip->i_mount;
+	int			error;
+
+	/*
+	 * check for re-use of an inode within an RCU grace period due to the
+	 * radix tree nodes not being updated yet. We monitor for this by
+	 * setting the inode number to zero before freeing the inode structure.
+	 * If the inode has been reallocated and set up, then the inode number
+	 * will not match, so check for that, too.
+	 */
+	spin_lock(&ip->i_flags_lock);
+	if (ip->i_ino != ino) {
+		trace_xfs_iget_skip(ip);
+		XFS_STATS_INC(xs_ig_frecycle);
+		error = EAGAIN;
+		goto out_error;
+	}
+
+
+	/*
+	 * If we are racing with another cache hit that is currently
+	 * instantiating this inode or currently recycling it out of
+	 * reclaimabe state, wait for the initialisation to complete
+	 * before continuing.
+	 *
+	 * XXX(hch): eventually we should do something equivalent to
+	 *	     wait_on_inode to wait for these flags to be cleared
+	 *	     instead of polling for it.
+	 */
+	if (ip->i_flags & (XFS_INEW|XFS_IRECLAIM)) {
+		trace_xfs_iget_skip(ip);
+		XFS_STATS_INC(xs_ig_frecycle);
+		error = EAGAIN;
+		goto out_error;
+	}
+
+	/*
+	 * If lookup is racing with unlink return an error immediately.
+	 */
+	if (ip->i_d.di_mode == 0 && !(flags & XFS_IGET_CREATE)) {
+		error = ENOENT;
+		goto out_error;
+	}
+
+	/*
+	 * If IRECLAIMABLE is set, we've torn down the VFS inode already.
+	 * Need to carefully get it back into useable state.
+	 */
+	if (ip->i_flags & XFS_IRECLAIMABLE) {
+		trace_xfs_iget_reclaim(ip);
+
+		/*
+		 * We need to set XFS_IRECLAIM to prevent xfs_reclaim_inode
+		 * from stomping over us while we recycle the inode.  We can't
+		 * clear the radix tree reclaimable tag yet as it requires
+		 * pag_ici_lock to be held exclusive.
+		 */
+		ip->i_flags |= XFS_IRECLAIM;
+
+		spin_unlock(&ip->i_flags_lock);
+		rcu_read_unlock();
+
+		error = -inode_init_always(mp->m_super, inode);
+		if (error) {
+			/*
+			 * Re-initializing the inode failed, and we are in deep
+			 * trouble.  Try to re-add it to the reclaim list.
+			 */
+			rcu_read_lock();
+			spin_lock(&ip->i_flags_lock);
+
+			ip->i_flags &= ~(XFS_INEW | XFS_IRECLAIM);
+			ASSERT(ip->i_flags & XFS_IRECLAIMABLE);
+			trace_xfs_iget_reclaim_fail(ip);
+			goto out_error;
+		}
+
+		spin_lock(&pag->pag_ici_lock);
+		spin_lock(&ip->i_flags_lock);
+
+		/*
+		 * Clear the per-lifetime state in the inode as we are now
+		 * effectively a new inode and need to return to the initial
+		 * state before reuse occurs.
+		 */
+		ip->i_flags &= ~XFS_IRECLAIM_RESET_FLAGS;
+		ip->i_flags |= XFS_INEW;
+		__xfs_inode_clear_reclaim_tag(mp, pag, ip);
+		inode->i_state = I_NEW;
+
+		ASSERT(!rwsem_is_locked(&ip->i_iolock.mr_lock));
+		mrlock_init(&ip->i_iolock, MRLOCK_BARRIER, "xfsio", ip->i_ino);
+		lockdep_set_class_and_name(&ip->i_iolock.mr_lock,
+				&xfs_iolock_active, "xfs_iolock_active");
+
+		spin_unlock(&ip->i_flags_lock);
+		spin_unlock(&pag->pag_ici_lock);
+	} else {
+		/* If the VFS inode is being torn down, pause and try again. */
+		if (!igrab(inode)) {
+			trace_xfs_iget_skip(ip);
+			error = EAGAIN;
+			goto out_error;
+		}
+
+		/* We've got a live one. */
+		spin_unlock(&ip->i_flags_lock);
+		rcu_read_unlock();
+		trace_xfs_iget_hit(ip);
+	}
+
+	if (lock_flags != 0)
+		xfs_ilock(ip, lock_flags);
+
+	xfs_iflags_clear(ip, XFS_ISTALE);
+	XFS_STATS_INC(xs_ig_found);
+
+	return 0;
+
+out_error:
+	spin_unlock(&ip->i_flags_lock);
+	rcu_read_unlock();
+	return error;
+}
+
+
+static int
+xfs_iget_cache_miss(
+	struct xfs_mount	*mp,
+	struct xfs_perag	*pag,
+	xfs_trans_t		*tp,
+	xfs_ino_t		ino,
+	struct xfs_inode	**ipp,
+	int			flags,
+	int			lock_flags)
+{
+	struct xfs_inode	*ip;
+	int			error;
+	xfs_agino_t		agino = XFS_INO_TO_AGINO(mp, ino);
+
+	ip = xfs_inode_alloc(mp, ino);
+	if (!ip)
+		return ENOMEM;
+
+	error = xfs_iread(mp, tp, ip, flags);
+	if (error)
+		goto out_destroy;
+
+	trace_xfs_iget_miss(ip);
+
+	if ((ip->i_d.di_mode == 0) && !(flags & XFS_IGET_CREATE)) {
+		error = ENOENT;
+		goto out_destroy;
+	}
+
+	/*
+	 * Preload the radix tree so we can insert safely under the
+	 * write spinlock. Note that we cannot sleep inside the preload
+	 * region.
+	 */
+	if (radix_tree_preload(GFP_KERNEL)) {
+		error = EAGAIN;
+		goto out_destroy;
+	}
+
+	/*
+	 * Because the inode hasn't been added to the radix-tree yet it can't
+	 * be found by another thread, so we can do the non-sleeping lock here.
+	 */
+	if (lock_flags) {
+		if (!xfs_ilock_nowait(ip, lock_flags))
+			BUG();
+	}
+
+	/*
+	 * These values must be set before inserting the inode into the radix
+	 * tree as the moment it is inserted a concurrent lookup (allowed by the
+	 * RCU locking mechanism) can find it and that lookup must see that this
+	 * is an inode currently under construction (i.e. that XFS_INEW is set).
+	 * The ip->i_flags_lock that protects the XFS_INEW flag forms the
+	 * memory barrier that ensures this detection works correctly at lookup
+	 * time.
+	 */
+	ip->i_udquot = ip->i_gdquot = NULL;
+	xfs_iflags_set(ip, XFS_INEW);
+
+	/* insert the new inode */
+	spin_lock(&pag->pag_ici_lock);
+	error = radix_tree_insert(&pag->pag_ici_root, agino, ip);
+	if (unlikely(error)) {
+		WARN_ON(error != -EEXIST);
+		XFS_STATS_INC(xs_ig_dup);
+		error = EAGAIN;
+		goto out_preload_end;
+	}
+	spin_unlock(&pag->pag_ici_lock);
+	radix_tree_preload_end();
+
+	*ipp = ip;
+	return 0;
+
+out_preload_end:
+	spin_unlock(&pag->pag_ici_lock);
+	radix_tree_preload_end();
+	if (lock_flags)
+		xfs_iunlock(ip, lock_flags);
+out_destroy:
+	__destroy_inode(VFS_I(ip));
+	xfs_inode_free(ip);
+	return error;
+}
+
+/*
+ * Look up an inode by number in the given file system.
+ * The inode is looked up in the cache held in each AG.
+ * If the inode is found in the cache, initialise the vfs inode
+ * if necessary.
+ *
+ * If it is not in core, read it in from the file system's device,
+ * add it to the cache and initialise the vfs inode.
+ *
+ * The inode is locked according to the value of the lock_flags parameter.
+ * This flag parameter indicates how and if the inode's IO lock and inode lock
+ * should be taken.
+ *
+ * mp -- the mount point structure for the current file system.  It points
+ *       to the inode hash table.
+ * tp -- a pointer to the current transaction if there is one.  This is
+ *       simply passed through to the xfs_iread() call.
+ * ino -- the number of the inode desired.  This is the unique identifier
+ *        within the file system for the inode being requested.
+ * lock_flags -- flags indicating how to lock the inode.  See the comment
+ *		 for xfs_ilock() for a list of valid values.
+ */
+int
+xfs_iget(
+	xfs_mount_t	*mp,
+	xfs_trans_t	*tp,
+	xfs_ino_t	ino,
+	uint		flags,
+	uint		lock_flags,
+	xfs_inode_t	**ipp)
+{
+	xfs_inode_t	*ip;
+	int		error;
+	xfs_perag_t	*pag;
+	xfs_agino_t	agino;
+
+	/* reject inode numbers outside existing AGs */
+	if (!ino || XFS_INO_TO_AGNO(mp, ino) >= mp->m_sb.sb_agcount)
+		return EINVAL;
+
+	/* get the perag structure and ensure that it's inode capable */
+	pag = xfs_perag_get(mp, XFS_INO_TO_AGNO(mp, ino));
+	agino = XFS_INO_TO_AGINO(mp, ino);
+
+again:
+	error = 0;
+	rcu_read_lock();
+	ip = radix_tree_lookup(&pag->pag_ici_root, agino);
+
+	if (ip) {
+		error = xfs_iget_cache_hit(pag, ip, ino, flags, lock_flags);
+		if (error)
+			goto out_error_or_again;
+	} else {
+		rcu_read_unlock();
+		XFS_STATS_INC(xs_ig_missed);
+
+		error = xfs_iget_cache_miss(mp, pag, tp, ino, &ip,
+							flags, lock_flags);
+		if (error)
+			goto out_error_or_again;
+	}
+	xfs_perag_put(pag);
+
+	*ipp = ip;
+
+	ASSERT(ip->i_df.if_ext_max ==
+	       XFS_IFORK_DSIZE(ip) / sizeof(xfs_bmbt_rec_t));
+	/*
+	 * If we have a real type for an on-disk inode, we can set ops(&unlock)
+	 * now.	 If it's a new inode being created, xfs_ialloc will handle it.
+	 */
+	if (xfs_iflags_test(ip, XFS_INEW) && ip->i_d.di_mode != 0)
+		xfs_setup_inode(ip);
+	return 0;
+
+out_error_or_again:
+	if (error == EAGAIN) {
+		delay(1);
+		goto again;
+	}
+	xfs_perag_put(pag);
+	return error;
+}
+
+/*
+ * This is a wrapper routine around the xfs_ilock() routine
+ * used to centralize some grungy code.  It is used in places
+ * that wish to lock the inode solely for reading the extents.
+ * The reason these places can't just call xfs_ilock(SHARED)
+ * is that the inode lock also guards to bringing in of the
+ * extents from disk for a file in b-tree format.  If the inode
+ * is in b-tree format, then we need to lock the inode exclusively
+ * until the extents are read in.  Locking it exclusively all
+ * the time would limit our parallelism unnecessarily, though.
+ * What we do instead is check to see if the extents have been
+ * read in yet, and only lock the inode exclusively if they
+ * have not.
+ *
+ * The function returns a value which should be given to the
+ * corresponding xfs_iunlock_map_shared().  This value is
+ * the mode in which the lock was actually taken.
+ */
+uint
+xfs_ilock_map_shared(
+	xfs_inode_t	*ip)
+{
+	uint	lock_mode;
+
+	if ((ip->i_d.di_format == XFS_DINODE_FMT_BTREE) &&
+	    ((ip->i_df.if_flags & XFS_IFEXTENTS) == 0)) {
+		lock_mode = XFS_ILOCK_EXCL;
+	} else {
+		lock_mode = XFS_ILOCK_SHARED;
+	}
+
+	xfs_ilock(ip, lock_mode);
+
+	return lock_mode;
+}
+
+/*
+ * This is simply the unlock routine to go with xfs_ilock_map_shared().
+ * All it does is call xfs_iunlock() with the given lock_mode.
+ */
+void
+xfs_iunlock_map_shared(
+	xfs_inode_t	*ip,
+	unsigned int	lock_mode)
+{
+	xfs_iunlock(ip, lock_mode);
+}
+
+/*
+ * The xfs inode contains 2 locks: a multi-reader lock called the
+ * i_iolock and a multi-reader lock called the i_lock.  This routine
+ * allows either or both of the locks to be obtained.
+ *
+ * The 2 locks should always be ordered so that the IO lock is
+ * obtained first in order to prevent deadlock.
+ *
+ * ip -- the inode being locked
+ * lock_flags -- this parameter indicates the inode's locks
+ *       to be locked.  It can be:
+ *		XFS_IOLOCK_SHARED,
+ *		XFS_IOLOCK_EXCL,
+ *		XFS_ILOCK_SHARED,
+ *		XFS_ILOCK_EXCL,
+ *		XFS_IOLOCK_SHARED | XFS_ILOCK_SHARED,
+ *		XFS_IOLOCK_SHARED | XFS_ILOCK_EXCL,
+ *		XFS_IOLOCK_EXCL | XFS_ILOCK_SHARED,
+ *		XFS_IOLOCK_EXCL | XFS_ILOCK_EXCL
+ */
+void
+xfs_ilock(
+	xfs_inode_t		*ip,
+	uint			lock_flags)
+{
+	/*
+	 * You can't set both SHARED and EXCL for the same lock,
+	 * and only XFS_IOLOCK_SHARED, XFS_IOLOCK_EXCL, XFS_ILOCK_SHARED,
+	 * and XFS_ILOCK_EXCL are valid values to set in lock_flags.
+	 */
+	ASSERT((lock_flags & (XFS_IOLOCK_SHARED | XFS_IOLOCK_EXCL)) !=
+	       (XFS_IOLOCK_SHARED | XFS_IOLOCK_EXCL));
+	ASSERT((lock_flags & (XFS_ILOCK_SHARED | XFS_ILOCK_EXCL)) !=
+	       (XFS_ILOCK_SHARED | XFS_ILOCK_EXCL));
+	ASSERT((lock_flags & ~(XFS_LOCK_MASK | XFS_LOCK_DEP_MASK)) == 0);
+
+	if (lock_flags & XFS_IOLOCK_EXCL)
+		mrupdate_nested(&ip->i_iolock, XFS_IOLOCK_DEP(lock_flags));
+	else if (lock_flags & XFS_IOLOCK_SHARED)
+		mraccess_nested(&ip->i_iolock, XFS_IOLOCK_DEP(lock_flags));
+
+	if (lock_flags & XFS_ILOCK_EXCL)
+		mrupdate_nested(&ip->i_lock, XFS_ILOCK_DEP(lock_flags));
+	else if (lock_flags & XFS_ILOCK_SHARED)
+		mraccess_nested(&ip->i_lock, XFS_ILOCK_DEP(lock_flags));
+
+	trace_xfs_ilock(ip, lock_flags, _RET_IP_);
+}
+
+/*
+ * This is just like xfs_ilock(), except that the caller
+ * is guaranteed not to sleep.  It returns 1 if it gets
+ * the requested locks and 0 otherwise.  If the IO lock is
+ * obtained but the inode lock cannot be, then the IO lock
+ * is dropped before returning.
+ *
+ * ip -- the inode being locked
+ * lock_flags -- this parameter indicates the inode's locks to be
+ *       to be locked.  See the comment for xfs_ilock() for a list
+ *	 of valid values.
+ */
+int
+xfs_ilock_nowait(
+	xfs_inode_t		*ip,
+	uint			lock_flags)
+{
+	/*
+	 * You can't set both SHARED and EXCL for the same lock,
+	 * and only XFS_IOLOCK_SHARED, XFS_IOLOCK_EXCL, XFS_ILOCK_SHARED,
+	 * and XFS_ILOCK_EXCL are valid values to set in lock_flags.
+	 */
+	ASSERT((lock_flags & (XFS_IOLOCK_SHARED | XFS_IOLOCK_EXCL)) !=
+	       (XFS_IOLOCK_SHARED | XFS_IOLOCK_EXCL));
+	ASSERT((lock_flags & (XFS_ILOCK_SHARED | XFS_ILOCK_EXCL)) !=
+	       (XFS_ILOCK_SHARED | XFS_ILOCK_EXCL));
+	ASSERT((lock_flags & ~(XFS_LOCK_MASK | XFS_LOCK_DEP_MASK)) == 0);
+
+	if (lock_flags & XFS_IOLOCK_EXCL) {
+		if (!mrtryupdate(&ip->i_iolock))
+			goto out;
+	} else if (lock_flags & XFS_IOLOCK_SHARED) {
+		if (!mrtryaccess(&ip->i_iolock))
+			goto out;
+	}
+	if (lock_flags & XFS_ILOCK_EXCL) {
+		if (!mrtryupdate(&ip->i_lock))
+			goto out_undo_iolock;
+	} else if (lock_flags & XFS_ILOCK_SHARED) {
+		if (!mrtryaccess(&ip->i_lock))
+			goto out_undo_iolock;
+	}
+	trace_xfs_ilock_nowait(ip, lock_flags, _RET_IP_);
+	return 1;
+
+ out_undo_iolock:
+	if (lock_flags & XFS_IOLOCK_EXCL)
+		mrunlock_excl(&ip->i_iolock);
+	else if (lock_flags & XFS_IOLOCK_SHARED)
+		mrunlock_shared(&ip->i_iolock);
+ out:
+	return 0;
+}
+
+/*
+ * xfs_iunlock() is used to drop the inode locks acquired with
+ * xfs_ilock() and xfs_ilock_nowait().  The caller must pass
+ * in the flags given to xfs_ilock() or xfs_ilock_nowait() so
+ * that we know which locks to drop.
+ *
+ * ip -- the inode being unlocked
+ * lock_flags -- this parameter indicates the inode's locks to be
+ *       to be unlocked.  See the comment for xfs_ilock() for a list
+ *	 of valid values for this parameter.
+ *
+ */
+void
+xfs_iunlock(
+	xfs_inode_t		*ip,
+	uint			lock_flags)
+{
+	/*
+	 * You can't set both SHARED and EXCL for the same lock,
+	 * and only XFS_IOLOCK_SHARED, XFS_IOLOCK_EXCL, XFS_ILOCK_SHARED,
+	 * and XFS_ILOCK_EXCL are valid values to set in lock_flags.
+	 */
+	ASSERT((lock_flags & (XFS_IOLOCK_SHARED | XFS_IOLOCK_EXCL)) !=
+	       (XFS_IOLOCK_SHARED | XFS_IOLOCK_EXCL));
+	ASSERT((lock_flags & (XFS_ILOCK_SHARED | XFS_ILOCK_EXCL)) !=
+	       (XFS_ILOCK_SHARED | XFS_ILOCK_EXCL));
+	ASSERT((lock_flags & ~(XFS_LOCK_MASK | XFS_IUNLOCK_NONOTIFY |
+			XFS_LOCK_DEP_MASK)) == 0);
+	ASSERT(lock_flags != 0);
+
+	if (lock_flags & XFS_IOLOCK_EXCL)
+		mrunlock_excl(&ip->i_iolock);
+	else if (lock_flags & XFS_IOLOCK_SHARED)
+		mrunlock_shared(&ip->i_iolock);
+
+	if (lock_flags & XFS_ILOCK_EXCL)
+		mrunlock_excl(&ip->i_lock);
+	else if (lock_flags & XFS_ILOCK_SHARED)
+		mrunlock_shared(&ip->i_lock);
+
+	if ((lock_flags & (XFS_ILOCK_SHARED | XFS_ILOCK_EXCL)) &&
+	    !(lock_flags & XFS_IUNLOCK_NONOTIFY) && ip->i_itemp) {
+		/*
+		 * Let the AIL know that this item has been unlocked in case
+		 * it is in the AIL and anyone is waiting on it.  Don't do
+		 * this if the caller has asked us not to.
+		 */
+		xfs_trans_unlocked_item(ip->i_itemp->ili_item.li_ailp,
+					(xfs_log_item_t*)(ip->i_itemp));
+	}
+	trace_xfs_iunlock(ip, lock_flags, _RET_IP_);
+}
+
+/*
+ * give up write locks.  the i/o lock cannot be held nested
+ * if it is being demoted.
+ */
+void
+xfs_ilock_demote(
+	xfs_inode_t		*ip,
+	uint			lock_flags)
+{
+	ASSERT(lock_flags & (XFS_IOLOCK_EXCL|XFS_ILOCK_EXCL));
+	ASSERT((lock_flags & ~(XFS_IOLOCK_EXCL|XFS_ILOCK_EXCL)) == 0);
+
+	if (lock_flags & XFS_ILOCK_EXCL)
+		mrdemote(&ip->i_lock);
+	if (lock_flags & XFS_IOLOCK_EXCL)
+		mrdemote(&ip->i_iolock);
+
+	trace_xfs_ilock_demote(ip, lock_flags, _RET_IP_);
+}
+
+#ifdef DEBUG
+int
+xfs_isilocked(
+	xfs_inode_t		*ip,
+	uint			lock_flags)
+{
+	if (lock_flags & (XFS_ILOCK_EXCL|XFS_ILOCK_SHARED)) {
+		if (!(lock_flags & XFS_ILOCK_SHARED))
+			return !!ip->i_lock.mr_writer;
+		return rwsem_is_locked(&ip->i_lock.mr_lock);
+	}
+
+	if (lock_flags & (XFS_IOLOCK_EXCL|XFS_IOLOCK_SHARED)) {
+		if (!(lock_flags & XFS_IOLOCK_SHARED))
+			return !!ip->i_iolock.mr_writer;
+		return rwsem_is_locked(&ip->i_iolock.mr_lock);
+	}
+
+	ASSERT(0);
+	return 0;
+}
+#endif
diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c
new file mode 100644
index 00000000..57152799
--- /dev/null
+++ b/fs/xfs/xfs_inode.c
@@ -0,0 +1,4145 @@
+/*
+ * Copyright (c) 2000-2006 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include <linux/log2.h>
+
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_trans_priv.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_mount.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_alloc_btree.h"
+#include "xfs_ialloc_btree.h"
+#include "xfs_attr_sf.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_buf_item.h"
+#include "xfs_inode_item.h"
+#include "xfs_btree.h"
+#include "xfs_btree_trace.h"
+#include "xfs_alloc.h"
+#include "xfs_ialloc.h"
+#include "xfs_bmap.h"
+#include "xfs_error.h"
+#include "xfs_utils.h"
+#include "xfs_quota.h"
+#include "xfs_filestream.h"
+#include "xfs_vnodeops.h"
+#include "xfs_trace.h"
+
+kmem_zone_t *xfs_ifork_zone;
+kmem_zone_t *xfs_inode_zone;
+
+/*
+ * Used in xfs_itruncate().  This is the maximum number of extents
+ * freed from a file in a single transaction.
+ */
+#define	XFS_ITRUNC_MAX_EXTENTS	2
+
+STATIC int xfs_iflush_int(xfs_inode_t *, xfs_buf_t *);
+STATIC int xfs_iformat_local(xfs_inode_t *, xfs_dinode_t *, int, int);
+STATIC int xfs_iformat_extents(xfs_inode_t *, xfs_dinode_t *, int);
+STATIC int xfs_iformat_btree(xfs_inode_t *, xfs_dinode_t *, int);
+
+#ifdef DEBUG
+/*
+ * Make sure that the extents in the given memory buffer
+ * are valid.
+ */
+STATIC void
+xfs_validate_extents(
+	xfs_ifork_t		*ifp,
+	int			nrecs,
+	xfs_exntfmt_t		fmt)
+{
+	xfs_bmbt_irec_t		irec;
+	xfs_bmbt_rec_host_t	rec;
+	int			i;
+
+	for (i = 0; i < nrecs; i++) {
+		xfs_bmbt_rec_host_t *ep = xfs_iext_get_ext(ifp, i);
+		rec.l0 = get_unaligned(&ep->l0);
+		rec.l1 = get_unaligned(&ep->l1);
+		xfs_bmbt_get_all(&rec, &irec);
+		if (fmt == XFS_EXTFMT_NOSTATE)
+			ASSERT(irec.br_state == XFS_EXT_NORM);
+	}
+}
+#else /* DEBUG */
+#define xfs_validate_extents(ifp, nrecs, fmt)
+#endif /* DEBUG */
+
+/*
+ * Check that none of the inode's in the buffer have a next
+ * unlinked field of 0.
+ */
+#if defined(DEBUG)
+void
+xfs_inobp_check(
+	xfs_mount_t	*mp,
+	xfs_buf_t	*bp)
+{
+	int		i;
+	int		j;
+	xfs_dinode_t	*dip;
+
+	j = mp->m_inode_cluster_size >> mp->m_sb.sb_inodelog;
+
+	for (i = 0; i < j; i++) {
+		dip = (xfs_dinode_t *)xfs_buf_offset(bp,
+					i * mp->m_sb.sb_inodesize);
+		if (!dip->di_next_unlinked)  {
+			xfs_alert(mp,
+	"Detected bogus zero next_unlinked field in incore inode buffer 0x%p.",
+				bp);
+			ASSERT(dip->di_next_unlinked);
+		}
+	}
+}
+#endif
+
+/*
+ * Find the buffer associated with the given inode map
+ * We do basic validation checks on the buffer once it has been
+ * retrieved from disk.
+ */
+STATIC int
+xfs_imap_to_bp(
+	xfs_mount_t	*mp,
+	xfs_trans_t	*tp,
+	struct xfs_imap	*imap,
+	xfs_buf_t	**bpp,
+	uint		buf_flags,
+	uint		iget_flags)
+{
+	int		error;
+	int		i;
+	int		ni;
+	xfs_buf_t	*bp;
+
+	error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp, imap->im_blkno,
+				   (int)imap->im_len, buf_flags, &bp);
+	if (error) {
+		if (error != EAGAIN) {
+			xfs_warn(mp,
+				"%s: xfs_trans_read_buf() returned error %d.",
+				__func__, error);
+		} else {
+			ASSERT(buf_flags & XBF_TRYLOCK);
+		}
+		return error;
+	}
+
+	/*
+	 * Validate the magic number and version of every inode in the buffer
+	 * (if DEBUG kernel) or the first inode in the buffer, otherwise.
+	 */
+#ifdef DEBUG
+	ni = BBTOB(imap->im_len) >> mp->m_sb.sb_inodelog;
+#else	/* usual case */
+	ni = 1;
+#endif
+
+	for (i = 0; i < ni; i++) {
+		int		di_ok;
+		xfs_dinode_t	*dip;
+
+		dip = (xfs_dinode_t *)xfs_buf_offset(bp,
+					(i << mp->m_sb.sb_inodelog));
+		di_ok = be16_to_cpu(dip->di_magic) == XFS_DINODE_MAGIC &&
+			    XFS_DINODE_GOOD_VERSION(dip->di_version);
+		if (unlikely(XFS_TEST_ERROR(!di_ok, mp,
+						XFS_ERRTAG_ITOBP_INOTOBP,
+						XFS_RANDOM_ITOBP_INOTOBP))) {
+			if (iget_flags & XFS_IGET_UNTRUSTED) {
+				xfs_trans_brelse(tp, bp);
+				return XFS_ERROR(EINVAL);
+			}
+			XFS_CORRUPTION_ERROR("xfs_imap_to_bp",
+						XFS_ERRLEVEL_HIGH, mp, dip);
+#ifdef DEBUG
+			xfs_emerg(mp,
+				"bad inode magic/vsn daddr %lld #%d (magic=%x)",
+				(unsigned long long)imap->im_blkno, i,
+				be16_to_cpu(dip->di_magic));
+			ASSERT(0);
+#endif
+			xfs_trans_brelse(tp, bp);
+			return XFS_ERROR(EFSCORRUPTED);
+		}
+	}
+
+	xfs_inobp_check(mp, bp);
+
+	/*
+	 * Mark the buffer as an inode buffer now that it looks good
+	 */
+	XFS_BUF_SET_VTYPE(bp, B_FS_INO);
+
+	*bpp = bp;
+	return 0;
+}
+
+/*
+ * This routine is called to map an inode number within a file
+ * system to the buffer containing the on-disk version of the
+ * inode.  It returns a pointer to the buffer containing the
+ * on-disk inode in the bpp parameter, and in the dip parameter
+ * it returns a pointer to the on-disk inode within that buffer.
+ *
+ * If a non-zero error is returned, then the contents of bpp and
+ * dipp are undefined.
+ *
+ * Use xfs_imap() to determine the size and location of the
+ * buffer to read from disk.
+ */
+int
+xfs_inotobp(
+	xfs_mount_t	*mp,
+	xfs_trans_t	*tp,
+	xfs_ino_t	ino,
+	xfs_dinode_t	**dipp,
+	xfs_buf_t	**bpp,
+	int		*offset,
+	uint		imap_flags)
+{
+	struct xfs_imap	imap;
+	xfs_buf_t	*bp;
+	int		error;
+
+	imap.im_blkno = 0;
+	error = xfs_imap(mp, tp, ino, &imap, imap_flags);
+	if (error)
+		return error;
+
+	error = xfs_imap_to_bp(mp, tp, &imap, &bp, XBF_LOCK, imap_flags);
+	if (error)
+		return error;
+
+	*dipp = (xfs_dinode_t *)xfs_buf_offset(bp, imap.im_boffset);
+	*bpp = bp;
+	*offset = imap.im_boffset;
+	return 0;
+}
+
+
+/*
+ * This routine is called to map an inode to the buffer containing
+ * the on-disk version of the inode.  It returns a pointer to the
+ * buffer containing the on-disk inode in the bpp parameter, and in
+ * the dip parameter it returns a pointer to the on-disk inode within
+ * that buffer.
+ *
+ * If a non-zero error is returned, then the contents of bpp and
+ * dipp are undefined.
+ *
+ * The inode is expected to already been mapped to its buffer and read
+ * in once, thus we can use the mapping information stored in the inode
+ * rather than calling xfs_imap().  This allows us to avoid the overhead
+ * of looking at the inode btree for small block file systems
+ * (see xfs_imap()).
+ */
+int
+xfs_itobp(
+	xfs_mount_t	*mp,
+	xfs_trans_t	*tp,
+	xfs_inode_t	*ip,
+	xfs_dinode_t	**dipp,
+	xfs_buf_t	**bpp,
+	uint		buf_flags)
+{
+	xfs_buf_t	*bp;
+	int		error;
+
+	ASSERT(ip->i_imap.im_blkno != 0);
+
+	error = xfs_imap_to_bp(mp, tp, &ip->i_imap, &bp, buf_flags, 0);
+	if (error)
+		return error;
+
+	if (!bp) {
+		ASSERT(buf_flags & XBF_TRYLOCK);
+		ASSERT(tp == NULL);
+		*bpp = NULL;
+		return EAGAIN;
+	}
+
+	*dipp = (xfs_dinode_t *)xfs_buf_offset(bp, ip->i_imap.im_boffset);
+	*bpp = bp;
+	return 0;
+}
+
+/*
+ * Move inode type and inode format specific information from the
+ * on-disk inode to the in-core inode.  For fifos, devs, and sockets
+ * this means set if_rdev to the proper value.  For files, directories,
+ * and symlinks this means to bring in the in-line data or extent
+ * pointers.  For a file in B-tree format, only the root is immediately
+ * brought in-core.  The rest will be in-lined in if_extents when it
+ * is first referenced (see xfs_iread_extents()).
+ */
+STATIC int
+xfs_iformat(
+	xfs_inode_t		*ip,
+	xfs_dinode_t		*dip)
+{
+	xfs_attr_shortform_t	*atp;
+	int			size;
+	int			error;
+	xfs_fsize_t             di_size;
+	ip->i_df.if_ext_max =
+		XFS_IFORK_DSIZE(ip) / (uint)sizeof(xfs_bmbt_rec_t);
+	error = 0;
+
+	if (unlikely(be32_to_cpu(dip->di_nextents) +
+		     be16_to_cpu(dip->di_anextents) >
+		     be64_to_cpu(dip->di_nblocks))) {
+		xfs_warn(ip->i_mount,
+			"corrupt dinode %Lu, extent total = %d, nblocks = %Lu.",
+			(unsigned long long)ip->i_ino,
+			(int)(be32_to_cpu(dip->di_nextents) +
+			      be16_to_cpu(dip->di_anextents)),
+			(unsigned long long)
+				be64_to_cpu(dip->di_nblocks));
+		XFS_CORRUPTION_ERROR("xfs_iformat(1)", XFS_ERRLEVEL_LOW,
+				     ip->i_mount, dip);
+		return XFS_ERROR(EFSCORRUPTED);
+	}
+
+	if (unlikely(dip->di_forkoff > ip->i_mount->m_sb.sb_inodesize)) {
+		xfs_warn(ip->i_mount, "corrupt dinode %Lu, forkoff = 0x%x.",
+			(unsigned long long)ip->i_ino,
+			dip->di_forkoff);
+		XFS_CORRUPTION_ERROR("xfs_iformat(2)", XFS_ERRLEVEL_LOW,
+				     ip->i_mount, dip);
+		return XFS_ERROR(EFSCORRUPTED);
+	}
+
+	if (unlikely((ip->i_d.di_flags & XFS_DIFLAG_REALTIME) &&
+		     !ip->i_mount->m_rtdev_targp)) {
+		xfs_warn(ip->i_mount,
+			"corrupt dinode %Lu, has realtime flag set.",
+			ip->i_ino);
+		XFS_CORRUPTION_ERROR("xfs_iformat(realtime)",
+				     XFS_ERRLEVEL_LOW, ip->i_mount, dip);
+		return XFS_ERROR(EFSCORRUPTED);
+	}
+
+	switch (ip->i_d.di_mode & S_IFMT) {
+	case S_IFIFO:
+	case S_IFCHR:
+	case S_IFBLK:
+	case S_IFSOCK:
+		if (unlikely(dip->di_format != XFS_DINODE_FMT_DEV)) {
+			XFS_CORRUPTION_ERROR("xfs_iformat(3)", XFS_ERRLEVEL_LOW,
+					      ip->i_mount, dip);
+			return XFS_ERROR(EFSCORRUPTED);
+		}
+		ip->i_d.di_size = 0;
+		ip->i_size = 0;
+		ip->i_df.if_u2.if_rdev = xfs_dinode_get_rdev(dip);
+		break;
+
+	case S_IFREG:
+	case S_IFLNK:
+	case S_IFDIR:
+		switch (dip->di_format) {
+		case XFS_DINODE_FMT_LOCAL:
+			/*
+			 * no local regular files yet
+			 */
+			if (unlikely((be16_to_cpu(dip->di_mode) & S_IFMT) == S_IFREG)) {
+				xfs_warn(ip->i_mount,
+			"corrupt inode %Lu (local format for regular file).",
+					(unsigned long long) ip->i_ino);
+				XFS_CORRUPTION_ERROR("xfs_iformat(4)",
+						     XFS_ERRLEVEL_LOW,
+						     ip->i_mount, dip);
+				return XFS_ERROR(EFSCORRUPTED);
+			}
+
+			di_size = be64_to_cpu(dip->di_size);
+			if (unlikely(di_size > XFS_DFORK_DSIZE(dip, ip->i_mount))) {
+				xfs_warn(ip->i_mount,
+			"corrupt inode %Lu (bad size %Ld for local inode).",
+					(unsigned long long) ip->i_ino,
+					(long long) di_size);
+				XFS_CORRUPTION_ERROR("xfs_iformat(5)",
+						     XFS_ERRLEVEL_LOW,
+						     ip->i_mount, dip);
+				return XFS_ERROR(EFSCORRUPTED);
+			}
+
+			size = (int)di_size;
+			error = xfs_iformat_local(ip, dip, XFS_DATA_FORK, size);
+			break;
+		case XFS_DINODE_FMT_EXTENTS:
+			error = xfs_iformat_extents(ip, dip, XFS_DATA_FORK);
+			break;
+		case XFS_DINODE_FMT_BTREE:
+			error = xfs_iformat_btree(ip, dip, XFS_DATA_FORK);
+			break;
+		default:
+			XFS_ERROR_REPORT("xfs_iformat(6)", XFS_ERRLEVEL_LOW,
+					 ip->i_mount);
+			return XFS_ERROR(EFSCORRUPTED);
+		}
+		break;
+
+	default:
+		XFS_ERROR_REPORT("xfs_iformat(7)", XFS_ERRLEVEL_LOW, ip->i_mount);
+		return XFS_ERROR(EFSCORRUPTED);
+	}
+	if (error) {
+		return error;
+	}
+	if (!XFS_DFORK_Q(dip))
+		return 0;
+	ASSERT(ip->i_afp == NULL);
+	ip->i_afp = kmem_zone_zalloc(xfs_ifork_zone, KM_SLEEP | KM_NOFS);
+	ip->i_afp->if_ext_max =
+		XFS_IFORK_ASIZE(ip) / (uint)sizeof(xfs_bmbt_rec_t);
+	switch (dip->di_aformat) {
+	case XFS_DINODE_FMT_LOCAL:
+		atp = (xfs_attr_shortform_t *)XFS_DFORK_APTR(dip);
+		size = be16_to_cpu(atp->hdr.totsize);
+
+		if (unlikely(size < sizeof(struct xfs_attr_sf_hdr))) {
+			xfs_warn(ip->i_mount,
+				"corrupt inode %Lu (bad attr fork size %Ld).",
+				(unsigned long long) ip->i_ino,
+				(long long) size);
+			XFS_CORRUPTION_ERROR("xfs_iformat(8)",
+					     XFS_ERRLEVEL_LOW,
+					     ip->i_mount, dip);
+			return XFS_ERROR(EFSCORRUPTED);
+		}
+
+		error = xfs_iformat_local(ip, dip, XFS_ATTR_FORK, size);
+		break;
+	case XFS_DINODE_FMT_EXTENTS:
+		error = xfs_iformat_extents(ip, dip, XFS_ATTR_FORK);
+		break;
+	case XFS_DINODE_FMT_BTREE:
+		error = xfs_iformat_btree(ip, dip, XFS_ATTR_FORK);
+		break;
+	default:
+		error = XFS_ERROR(EFSCORRUPTED);
+		break;
+	}
+	if (error) {
+		kmem_zone_free(xfs_ifork_zone, ip->i_afp);
+		ip->i_afp = NULL;
+		xfs_idestroy_fork(ip, XFS_DATA_FORK);
+	}
+	return error;
+}
+
+/*
+ * The file is in-lined in the on-disk inode.
+ * If it fits into if_inline_data, then copy
+ * it there, otherwise allocate a buffer for it
+ * and copy the data there.  Either way, set
+ * if_data to point at the data.
+ * If we allocate a buffer for the data, make
+ * sure that its size is a multiple of 4 and
+ * record the real size in i_real_bytes.
+ */
+STATIC int
+xfs_iformat_local(
+	xfs_inode_t	*ip,
+	xfs_dinode_t	*dip,
+	int		whichfork,
+	int		size)
+{
+	xfs_ifork_t	*ifp;
+	int		real_size;
+
+	/*
+	 * If the size is unreasonable, then something
+	 * is wrong and we just bail out rather than crash in
+	 * kmem_alloc() or memcpy() below.
+	 */
+	if (unlikely(size > XFS_DFORK_SIZE(dip, ip->i_mount, whichfork))) {
+		xfs_warn(ip->i_mount,
+	"corrupt inode %Lu (bad size %d for local fork, size = %d).",
+			(unsigned long long) ip->i_ino, size,
+			XFS_DFORK_SIZE(dip, ip->i_mount, whichfork));
+		XFS_CORRUPTION_ERROR("xfs_iformat_local", XFS_ERRLEVEL_LOW,
+				     ip->i_mount, dip);
+		return XFS_ERROR(EFSCORRUPTED);
+	}
+	ifp = XFS_IFORK_PTR(ip, whichfork);
+	real_size = 0;
+	if (size == 0)
+		ifp->if_u1.if_data = NULL;
+	else if (size <= sizeof(ifp->if_u2.if_inline_data))
+		ifp->if_u1.if_data = ifp->if_u2.if_inline_data;
+	else {
+		real_size = roundup(size, 4);
+		ifp->if_u1.if_data = kmem_alloc(real_size, KM_SLEEP | KM_NOFS);
+	}
+	ifp->if_bytes = size;
+	ifp->if_real_bytes = real_size;
+	if (size)
+		memcpy(ifp->if_u1.if_data, XFS_DFORK_PTR(dip, whichfork), size);
+	ifp->if_flags &= ~XFS_IFEXTENTS;
+	ifp->if_flags |= XFS_IFINLINE;
+	return 0;
+}
+
+/*
+ * The file consists of a set of extents all
+ * of which fit into the on-disk inode.
+ * If there are few enough extents to fit into
+ * the if_inline_ext, then copy them there.
+ * Otherwise allocate a buffer for them and copy
+ * them into it.  Either way, set if_extents
+ * to point at the extents.
+ */
+STATIC int
+xfs_iformat_extents(
+	xfs_inode_t	*ip,
+	xfs_dinode_t	*dip,
+	int		whichfork)
+{
+	xfs_bmbt_rec_t	*dp;
+	xfs_ifork_t	*ifp;
+	int		nex;
+	int		size;
+	int		i;
+
+	ifp = XFS_IFORK_PTR(ip, whichfork);
+	nex = XFS_DFORK_NEXTENTS(dip, whichfork);
+	size = nex * (uint)sizeof(xfs_bmbt_rec_t);
+
+	/*
+	 * If the number of extents is unreasonable, then something
+	 * is wrong and we just bail out rather than crash in
+	 * kmem_alloc() or memcpy() below.
+	 */
+	if (unlikely(size < 0 || size > XFS_DFORK_SIZE(dip, ip->i_mount, whichfork))) {
+		xfs_warn(ip->i_mount, "corrupt inode %Lu ((a)extents = %d).",
+			(unsigned long long) ip->i_ino, nex);
+		XFS_CORRUPTION_ERROR("xfs_iformat_extents(1)", XFS_ERRLEVEL_LOW,
+				     ip->i_mount, dip);
+		return XFS_ERROR(EFSCORRUPTED);
+	}
+
+	ifp->if_real_bytes = 0;
+	if (nex == 0)
+		ifp->if_u1.if_extents = NULL;
+	else if (nex <= XFS_INLINE_EXTS)
+		ifp->if_u1.if_extents = ifp->if_u2.if_inline_ext;
+	else
+		xfs_iext_add(ifp, 0, nex);
+
+	ifp->if_bytes = size;
+	if (size) {
+		dp = (xfs_bmbt_rec_t *) XFS_DFORK_PTR(dip, whichfork);
+		xfs_validate_extents(ifp, nex, XFS_EXTFMT_INODE(ip));
+		for (i = 0; i < nex; i++, dp++) {
+			xfs_bmbt_rec_host_t *ep = xfs_iext_get_ext(ifp, i);
+			ep->l0 = get_unaligned_be64(&dp->l0);
+			ep->l1 = get_unaligned_be64(&dp->l1);
+		}
+		XFS_BMAP_TRACE_EXLIST(ip, nex, whichfork);
+		if (whichfork != XFS_DATA_FORK ||
+			XFS_EXTFMT_INODE(ip) == XFS_EXTFMT_NOSTATE)
+				if (unlikely(xfs_check_nostate_extents(
+				    ifp, 0, nex))) {
+					XFS_ERROR_REPORT("xfs_iformat_extents(2)",
+							 XFS_ERRLEVEL_LOW,
+							 ip->i_mount);
+					return XFS_ERROR(EFSCORRUPTED);
+				}
+	}
+	ifp->if_flags |= XFS_IFEXTENTS;
+	return 0;
+}
+
+/*
+ * The file has too many extents to fit into
+ * the inode, so they are in B-tree format.
+ * Allocate a buffer for the root of the B-tree
+ * and copy the root into it.  The i_extents
+ * field will remain NULL until all of the
+ * extents are read in (when they are needed).
+ */
+STATIC int
+xfs_iformat_btree(
+	xfs_inode_t		*ip,
+	xfs_dinode_t		*dip,
+	int			whichfork)
+{
+	xfs_bmdr_block_t	*dfp;
+	xfs_ifork_t		*ifp;
+	/* REFERENCED */
+	int			nrecs;
+	int			size;
+
+	ifp = XFS_IFORK_PTR(ip, whichfork);
+	dfp = (xfs_bmdr_block_t *)XFS_DFORK_PTR(dip, whichfork);
+	size = XFS_BMAP_BROOT_SPACE(dfp);
+	nrecs = be16_to_cpu(dfp->bb_numrecs);
+
+	/*
+	 * blow out if -- fork has less extents than can fit in
+	 * fork (fork shouldn't be a btree format), root btree
+	 * block has more records than can fit into the fork,
+	 * or the number of extents is greater than the number of
+	 * blocks.
+	 */
+	if (unlikely(XFS_IFORK_NEXTENTS(ip, whichfork) <= ifp->if_ext_max
+	    || XFS_BMDR_SPACE_CALC(nrecs) >
+			XFS_DFORK_SIZE(dip, ip->i_mount, whichfork)
+	    || XFS_IFORK_NEXTENTS(ip, whichfork) > ip->i_d.di_nblocks)) {
+		xfs_warn(ip->i_mount, "corrupt inode %Lu (btree).",
+			(unsigned long long) ip->i_ino);
+		XFS_CORRUPTION_ERROR("xfs_iformat_btree", XFS_ERRLEVEL_LOW,
+				 ip->i_mount, dip);
+		return XFS_ERROR(EFSCORRUPTED);
+	}
+
+	ifp->if_broot_bytes = size;
+	ifp->if_broot = kmem_alloc(size, KM_SLEEP | KM_NOFS);
+	ASSERT(ifp->if_broot != NULL);
+	/*
+	 * Copy and convert from the on-disk structure
+	 * to the in-memory structure.
+	 */
+	xfs_bmdr_to_bmbt(ip->i_mount, dfp,
+			 XFS_DFORK_SIZE(dip, ip->i_mount, whichfork),
+			 ifp->if_broot, size);
+	ifp->if_flags &= ~XFS_IFEXTENTS;
+	ifp->if_flags |= XFS_IFBROOT;
+
+	return 0;
+}
+
+STATIC void
+xfs_dinode_from_disk(
+	xfs_icdinode_t		*to,
+	xfs_dinode_t		*from)
+{
+	to->di_magic = be16_to_cpu(from->di_magic);
+	to->di_mode = be16_to_cpu(from->di_mode);
+	to->di_version = from ->di_version;
+	to->di_format = from->di_format;
+	to->di_onlink = be16_to_cpu(from->di_onlink);
+	to->di_uid = be32_to_cpu(from->di_uid);
+	to->di_gid = be32_to_cpu(from->di_gid);
+	to->di_nlink = be32_to_cpu(from->di_nlink);
+	to->di_projid_lo = be16_to_cpu(from->di_projid_lo);
+	to->di_projid_hi = be16_to_cpu(from->di_projid_hi);
+	memcpy(to->di_pad, from->di_pad, sizeof(to->di_pad));
+	to->di_flushiter = be16_to_cpu(from->di_flushiter);
+	to->di_atime.t_sec = be32_to_cpu(from->di_atime.t_sec);
+	to->di_atime.t_nsec = be32_to_cpu(from->di_atime.t_nsec);
+	to->di_mtime.t_sec = be32_to_cpu(from->di_mtime.t_sec);
+	to->di_mtime.t_nsec = be32_to_cpu(from->di_mtime.t_nsec);
+	to->di_ctime.t_sec = be32_to_cpu(from->di_ctime.t_sec);
+	to->di_ctime.t_nsec = be32_to_cpu(from->di_ctime.t_nsec);
+	to->di_size = be64_to_cpu(from->di_size);
+	to->di_nblocks = be64_to_cpu(from->di_nblocks);
+	to->di_extsize = be32_to_cpu(from->di_extsize);
+	to->di_nextents = be32_to_cpu(from->di_nextents);
+	to->di_anextents = be16_to_cpu(from->di_anextents);
+	to->di_forkoff = from->di_forkoff;
+	to->di_aformat	= from->di_aformat;
+	to->di_dmevmask	= be32_to_cpu(from->di_dmevmask);
+	to->di_dmstate	= be16_to_cpu(from->di_dmstate);
+	to->di_flags	= be16_to_cpu(from->di_flags);
+	to->di_gen	= be32_to_cpu(from->di_gen);
+}
+
+void
+xfs_dinode_to_disk(
+	xfs_dinode_t		*to,
+	xfs_icdinode_t		*from)
+{
+	to->di_magic = cpu_to_be16(from->di_magic);
+	to->di_mode = cpu_to_be16(from->di_mode);
+	to->di_version = from ->di_version;
+	to->di_format = from->di_format;
+	to->di_onlink = cpu_to_be16(from->di_onlink);
+	to->di_uid = cpu_to_be32(from->di_uid);
+	to->di_gid = cpu_to_be32(from->di_gid);
+	to->di_nlink = cpu_to_be32(from->di_nlink);
+	to->di_projid_lo = cpu_to_be16(from->di_projid_lo);
+	to->di_projid_hi = cpu_to_be16(from->di_projid_hi);
+	memcpy(to->di_pad, from->di_pad, sizeof(to->di_pad));
+	to->di_flushiter = cpu_to_be16(from->di_flushiter);
+	to->di_atime.t_sec = cpu_to_be32(from->di_atime.t_sec);
+	to->di_atime.t_nsec = cpu_to_be32(from->di_atime.t_nsec);
+	to->di_mtime.t_sec = cpu_to_be32(from->di_mtime.t_sec);
+	to->di_mtime.t_nsec = cpu_to_be32(from->di_mtime.t_nsec);
+	to->di_ctime.t_sec = cpu_to_be32(from->di_ctime.t_sec);
+	to->di_ctime.t_nsec = cpu_to_be32(from->di_ctime.t_nsec);
+	to->di_size = cpu_to_be64(from->di_size);
+	to->di_nblocks = cpu_to_be64(from->di_nblocks);
+	to->di_extsize = cpu_to_be32(from->di_extsize);
+	to->di_nextents = cpu_to_be32(from->di_nextents);
+	to->di_anextents = cpu_to_be16(from->di_anextents);
+	to->di_forkoff = from->di_forkoff;
+	to->di_aformat = from->di_aformat;
+	to->di_dmevmask = cpu_to_be32(from->di_dmevmask);
+	to->di_dmstate = cpu_to_be16(from->di_dmstate);
+	to->di_flags = cpu_to_be16(from->di_flags);
+	to->di_gen = cpu_to_be32(from->di_gen);
+}
+
+STATIC uint
+_xfs_dic2xflags(
+	__uint16_t		di_flags)
+{
+	uint			flags = 0;
+
+	if (di_flags & XFS_DIFLAG_ANY) {
+		if (di_flags & XFS_DIFLAG_REALTIME)
+			flags |= XFS_XFLAG_REALTIME;
+		if (di_flags & XFS_DIFLAG_PREALLOC)
+			flags |= XFS_XFLAG_PREALLOC;
+		if (di_flags & XFS_DIFLAG_IMMUTABLE)
+			flags |= XFS_XFLAG_IMMUTABLE;
+		if (di_flags & XFS_DIFLAG_APPEND)
+			flags |= XFS_XFLAG_APPEND;
+		if (di_flags & XFS_DIFLAG_SYNC)
+			flags |= XFS_XFLAG_SYNC;
+		if (di_flags & XFS_DIFLAG_NOATIME)
+			flags |= XFS_XFLAG_NOATIME;
+		if (di_flags & XFS_DIFLAG_NODUMP)
+			flags |= XFS_XFLAG_NODUMP;
+		if (di_flags & XFS_DIFLAG_RTINHERIT)
+			flags |= XFS_XFLAG_RTINHERIT;
+		if (di_flags & XFS_DIFLAG_PROJINHERIT)
+			flags |= XFS_XFLAG_PROJINHERIT;
+		if (di_flags & XFS_DIFLAG_NOSYMLINKS)
+			flags |= XFS_XFLAG_NOSYMLINKS;
+		if (di_flags & XFS_DIFLAG_EXTSIZE)
+			flags |= XFS_XFLAG_EXTSIZE;
+		if (di_flags & XFS_DIFLAG_EXTSZINHERIT)
+			flags |= XFS_XFLAG_EXTSZINHERIT;
+		if (di_flags & XFS_DIFLAG_NODEFRAG)
+			flags |= XFS_XFLAG_NODEFRAG;
+		if (di_flags & XFS_DIFLAG_FILESTREAM)
+			flags |= XFS_XFLAG_FILESTREAM;
+	}
+
+	return flags;
+}
+
+uint
+xfs_ip2xflags(
+	xfs_inode_t		*ip)
+{
+	xfs_icdinode_t		*dic = &ip->i_d;
+
+	return _xfs_dic2xflags(dic->di_flags) |
+				(XFS_IFORK_Q(ip) ? XFS_XFLAG_HASATTR : 0);
+}
+
+uint
+xfs_dic2xflags(
+	xfs_dinode_t		*dip)
+{
+	return _xfs_dic2xflags(be16_to_cpu(dip->di_flags)) |
+				(XFS_DFORK_Q(dip) ? XFS_XFLAG_HASATTR : 0);
+}
+
+/*
+ * Read the disk inode attributes into the in-core inode structure.
+ */
+int
+xfs_iread(
+	xfs_mount_t	*mp,
+	xfs_trans_t	*tp,
+	xfs_inode_t	*ip,
+	uint		iget_flags)
+{
+	xfs_buf_t	*bp;
+	xfs_dinode_t	*dip;
+	int		error;
+
+	/*
+	 * Fill in the location information in the in-core inode.
+	 */
+	error = xfs_imap(mp, tp, ip->i_ino, &ip->i_imap, iget_flags);
+	if (error)
+		return error;
+
+	/*
+	 * Get pointers to the on-disk inode and the buffer containing it.
+	 */
+	error = xfs_imap_to_bp(mp, tp, &ip->i_imap, &bp,
+			       XBF_LOCK, iget_flags);
+	if (error)
+		return error;
+	dip = (xfs_dinode_t *)xfs_buf_offset(bp, ip->i_imap.im_boffset);
+
+	/*
+	 * If we got something that isn't an inode it means someone
+	 * (nfs or dmi) has a stale handle.
+	 */
+	if (be16_to_cpu(dip->di_magic) != XFS_DINODE_MAGIC) {
+#ifdef DEBUG
+		xfs_alert(mp,
+			"%s: dip->di_magic (0x%x) != XFS_DINODE_MAGIC (0x%x)",
+			__func__, be16_to_cpu(dip->di_magic), XFS_DINODE_MAGIC);
+#endif /* DEBUG */
+		error = XFS_ERROR(EINVAL);
+		goto out_brelse;
+	}
+
+	/*
+	 * If the on-disk inode is already linked to a directory
+	 * entry, copy all of the inode into the in-core inode.
+	 * xfs_iformat() handles copying in the inode format
+	 * specific information.
+	 * Otherwise, just get the truly permanent information.
+	 */
+	if (dip->di_mode) {
+		xfs_dinode_from_disk(&ip->i_d, dip);
+		error = xfs_iformat(ip, dip);
+		if (error)  {
+#ifdef DEBUG
+			xfs_alert(mp, "%s: xfs_iformat() returned error %d",
+				__func__, error);
+#endif /* DEBUG */
+			goto out_brelse;
+		}
+	} else {
+		ip->i_d.di_magic = be16_to_cpu(dip->di_magic);
+		ip->i_d.di_version = dip->di_version;
+		ip->i_d.di_gen = be32_to_cpu(dip->di_gen);
+		ip->i_d.di_flushiter = be16_to_cpu(dip->di_flushiter);
+		/*
+		 * Make sure to pull in the mode here as well in
+		 * case the inode is released without being used.
+		 * This ensures that xfs_inactive() will see that
+		 * the inode is already free and not try to mess
+		 * with the uninitialized part of it.
+		 */
+		ip->i_d.di_mode = 0;
+		/*
+		 * Initialize the per-fork minima and maxima for a new
+		 * inode here.  xfs_iformat will do it for old inodes.
+		 */
+		ip->i_df.if_ext_max =
+			XFS_IFORK_DSIZE(ip) / (uint)sizeof(xfs_bmbt_rec_t);
+	}
+
+	/*
+	 * The inode format changed when we moved the link count and
+	 * made it 32 bits long.  If this is an old format inode,
+	 * convert it in memory to look like a new one.  If it gets
+	 * flushed to disk we will convert back before flushing or
+	 * logging it.  We zero out the new projid field and the old link
+	 * count field.  We'll handle clearing the pad field (the remains
+	 * of the old uuid field) when we actually convert the inode to
+	 * the new format. We don't change the version number so that we
+	 * can distinguish this from a real new format inode.
+	 */
+	if (ip->i_d.di_version == 1) {
+		ip->i_d.di_nlink = ip->i_d.di_onlink;
+		ip->i_d.di_onlink = 0;
+		xfs_set_projid(ip, 0);
+	}
+
+	ip->i_delayed_blks = 0;
+	ip->i_size = ip->i_d.di_size;
+
+	/*
+	 * Mark the buffer containing the inode as something to keep
+	 * around for a while.  This helps to keep recently accessed
+	 * meta-data in-core longer.
+	 */
+	xfs_buf_set_ref(bp, XFS_INO_REF);
+
+	/*
+	 * Use xfs_trans_brelse() to release the buffer containing the
+	 * on-disk inode, because it was acquired with xfs_trans_read_buf()
+	 * in xfs_itobp() above.  If tp is NULL, this is just a normal
+	 * brelse().  If we're within a transaction, then xfs_trans_brelse()
+	 * will only release the buffer if it is not dirty within the
+	 * transaction.  It will be OK to release the buffer in this case,
+	 * because inodes on disk are never destroyed and we will be
+	 * locking the new in-core inode before putting it in the hash
+	 * table where other processes can find it.  Thus we don't have
+	 * to worry about the inode being changed just because we released
+	 * the buffer.
+	 */
+ out_brelse:
+	xfs_trans_brelse(tp, bp);
+	return error;
+}
+
+/*
+ * Read in extents from a btree-format inode.
+ * Allocate and fill in if_extents.  Real work is done in xfs_bmap.c.
+ */
+int
+xfs_iread_extents(
+	xfs_trans_t	*tp,
+	xfs_inode_t	*ip,
+	int		whichfork)
+{
+	int		error;
+	xfs_ifork_t	*ifp;
+	xfs_extnum_t	nextents;
+
+	if (unlikely(XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_BTREE)) {
+		XFS_ERROR_REPORT("xfs_iread_extents", XFS_ERRLEVEL_LOW,
+				 ip->i_mount);
+		return XFS_ERROR(EFSCORRUPTED);
+	}
+	nextents = XFS_IFORK_NEXTENTS(ip, whichfork);
+	ifp = XFS_IFORK_PTR(ip, whichfork);
+
+	/*
+	 * We know that the size is valid (it's checked in iformat_btree)
+	 */
+	ifp->if_bytes = ifp->if_real_bytes = 0;
+	ifp->if_flags |= XFS_IFEXTENTS;
+	xfs_iext_add(ifp, 0, nextents);
+	error = xfs_bmap_read_extents(tp, ip, whichfork);
+	if (error) {
+		xfs_iext_destroy(ifp);
+		ifp->if_flags &= ~XFS_IFEXTENTS;
+		return error;
+	}
+	xfs_validate_extents(ifp, nextents, XFS_EXTFMT_INODE(ip));
+	return 0;
+}
+
+/*
+ * Allocate an inode on disk and return a copy of its in-core version.
+ * The in-core inode is locked exclusively.  Set mode, nlink, and rdev
+ * appropriately within the inode.  The uid and gid for the inode are
+ * set according to the contents of the given cred structure.
+ *
+ * Use xfs_dialloc() to allocate the on-disk inode. If xfs_dialloc()
+ * has a free inode available, call xfs_iget()
+ * to obtain the in-core version of the allocated inode.  Finally,
+ * fill in the inode and log its initial contents.  In this case,
+ * ialloc_context would be set to NULL and call_again set to false.
+ *
+ * If xfs_dialloc() does not have an available inode,
+ * it will replenish its supply by doing an allocation. Since we can
+ * only do one allocation within a transaction without deadlocks, we
+ * must commit the current transaction before returning the inode itself.
+ * In this case, therefore, we will set call_again to true and return.
+ * The caller should then commit the current transaction, start a new
+ * transaction, and call xfs_ialloc() again to actually get the inode.
+ *
+ * To ensure that some other process does not grab the inode that
+ * was allocated during the first call to xfs_ialloc(), this routine
+ * also returns the [locked] bp pointing to the head of the freelist
+ * as ialloc_context.  The caller should hold this buffer across
+ * the commit and pass it back into this routine on the second call.
+ *
+ * If we are allocating quota inodes, we do not have a parent inode
+ * to attach to or associate with (i.e. pip == NULL) because they
+ * are not linked into the directory structure - they are attached
+ * directly to the superblock - and so have no parent.
+ */
+int
+xfs_ialloc(
+	xfs_trans_t	*tp,
+	xfs_inode_t	*pip,
+	mode_t		mode,
+	xfs_nlink_t	nlink,
+	xfs_dev_t	rdev,
+	prid_t		prid,
+	int		okalloc,
+	xfs_buf_t	**ialloc_context,
+	boolean_t	*call_again,
+	xfs_inode_t	**ipp)
+{
+	xfs_ino_t	ino;
+	xfs_inode_t	*ip;
+	uint		flags;
+	int		error;
+	timespec_t	tv;
+	int		filestreams = 0;
+
+	/*
+	 * Call the space management code to pick
+	 * the on-disk inode to be allocated.
+	 */
+	error = xfs_dialloc(tp, pip ? pip->i_ino : 0, mode, okalloc,
+			    ialloc_context, call_again, &ino);
+	if (error)
+		return error;
+	if (*call_again || ino == NULLFSINO) {
+		*ipp = NULL;
+		return 0;
+	}
+	ASSERT(*ialloc_context == NULL);
+
+	/*
+	 * Get the in-core inode with the lock held exclusively.
+	 * This is because we're setting fields here we need
+	 * to prevent others from looking at until we're done.
+	 */
+	error = xfs_iget(tp->t_mountp, tp, ino, XFS_IGET_CREATE,
+			 XFS_ILOCK_EXCL, &ip);
+	if (error)
+		return error;
+	ASSERT(ip != NULL);
+
+	ip->i_d.di_mode = (__uint16_t)mode;
+	ip->i_d.di_onlink = 0;
+	ip->i_d.di_nlink = nlink;
+	ASSERT(ip->i_d.di_nlink == nlink);
+	ip->i_d.di_uid = current_fsuid();
+	ip->i_d.di_gid = current_fsgid();
+	xfs_set_projid(ip, prid);
+	memset(&(ip->i_d.di_pad[0]), 0, sizeof(ip->i_d.di_pad));
+
+	/*
+	 * If the superblock version is up to where we support new format
+	 * inodes and this is currently an old format inode, then change
+	 * the inode version number now.  This way we only do the conversion
+	 * here rather than here and in the flush/logging code.
+	 */
+	if (xfs_sb_version_hasnlink(&tp->t_mountp->m_sb) &&
+	    ip->i_d.di_version == 1) {
+		ip->i_d.di_version = 2;
+		/*
+		 * We've already zeroed the old link count, the projid field,
+		 * and the pad field.
+		 */
+	}
+
+	/*
+	 * Project ids won't be stored on disk if we are using a version 1 inode.
+	 */
+	if ((prid != 0) && (ip->i_d.di_version == 1))
+		xfs_bump_ino_vers2(tp, ip);
+
+	if (pip && XFS_INHERIT_GID(pip)) {
+		ip->i_d.di_gid = pip->i_d.di_gid;
+		if ((pip->i_d.di_mode & S_ISGID) && (mode & S_IFMT) == S_IFDIR) {
+			ip->i_d.di_mode |= S_ISGID;
+		}
+	}
+
+	/*
+	 * If the group ID of the new file does not match the effective group
+	 * ID or one of the supplementary group IDs, the S_ISGID bit is cleared
+	 * (and only if the irix_sgid_inherit compatibility variable is set).
+	 */
+	if ((irix_sgid_inherit) &&
+	    (ip->i_d.di_mode & S_ISGID) &&
+	    (!in_group_p((gid_t)ip->i_d.di_gid))) {
+		ip->i_d.di_mode &= ~S_ISGID;
+	}
+
+	ip->i_d.di_size = 0;
+	ip->i_size = 0;
+	ip->i_d.di_nextents = 0;
+	ASSERT(ip->i_d.di_nblocks == 0);
+
+	nanotime(&tv);
+	ip->i_d.di_mtime.t_sec = (__int32_t)tv.tv_sec;
+	ip->i_d.di_mtime.t_nsec = (__int32_t)tv.tv_nsec;
+	ip->i_d.di_atime = ip->i_d.di_mtime;
+	ip->i_d.di_ctime = ip->i_d.di_mtime;
+
+	/*
+	 * di_gen will have been taken care of in xfs_iread.
+	 */
+	ip->i_d.di_extsize = 0;
+	ip->i_d.di_dmevmask = 0;
+	ip->i_d.di_dmstate = 0;
+	ip->i_d.di_flags = 0;
+	flags = XFS_ILOG_CORE;
+	switch (mode & S_IFMT) {
+	case S_IFIFO:
+	case S_IFCHR:
+	case S_IFBLK:
+	case S_IFSOCK:
+		ip->i_d.di_format = XFS_DINODE_FMT_DEV;
+		ip->i_df.if_u2.if_rdev = rdev;
+		ip->i_df.if_flags = 0;
+		flags |= XFS_ILOG_DEV;
+		break;
+	case S_IFREG:
+		/*
+		 * we can't set up filestreams until after the VFS inode
+		 * is set up properly.
+		 */
+		if (pip && xfs_inode_is_filestream(pip))
+			filestreams = 1;
+		/* fall through */
+	case S_IFDIR:
+		if (pip && (pip->i_d.di_flags & XFS_DIFLAG_ANY)) {
+			uint	di_flags = 0;
+
+			if ((mode & S_IFMT) == S_IFDIR) {
+				if (pip->i_d.di_flags & XFS_DIFLAG_RTINHERIT)
+					di_flags |= XFS_DIFLAG_RTINHERIT;
+				if (pip->i_d.di_flags & XFS_DIFLAG_EXTSZINHERIT) {
+					di_flags |= XFS_DIFLAG_EXTSZINHERIT;
+					ip->i_d.di_extsize = pip->i_d.di_extsize;
+				}
+			} else if ((mode & S_IFMT) == S_IFREG) {
+				if (pip->i_d.di_flags & XFS_DIFLAG_RTINHERIT)
+					di_flags |= XFS_DIFLAG_REALTIME;
+				if (pip->i_d.di_flags & XFS_DIFLAG_EXTSZINHERIT) {
+					di_flags |= XFS_DIFLAG_EXTSIZE;
+					ip->i_d.di_extsize = pip->i_d.di_extsize;
+				}
+			}
+			if ((pip->i_d.di_flags & XFS_DIFLAG_NOATIME) &&
+			    xfs_inherit_noatime)
+				di_flags |= XFS_DIFLAG_NOATIME;
+			if ((pip->i_d.di_flags & XFS_DIFLAG_NODUMP) &&
+			    xfs_inherit_nodump)
+				di_flags |= XFS_DIFLAG_NODUMP;
+			if ((pip->i_d.di_flags & XFS_DIFLAG_SYNC) &&
+			    xfs_inherit_sync)
+				di_flags |= XFS_DIFLAG_SYNC;
+			if ((pip->i_d.di_flags & XFS_DIFLAG_NOSYMLINKS) &&
+			    xfs_inherit_nosymlinks)
+				di_flags |= XFS_DIFLAG_NOSYMLINKS;
+			if (pip->i_d.di_flags & XFS_DIFLAG_PROJINHERIT)
+				di_flags |= XFS_DIFLAG_PROJINHERIT;
+			if ((pip->i_d.di_flags & XFS_DIFLAG_NODEFRAG) &&
+			    xfs_inherit_nodefrag)
+				di_flags |= XFS_DIFLAG_NODEFRAG;
+			if (pip->i_d.di_flags & XFS_DIFLAG_FILESTREAM)
+				di_flags |= XFS_DIFLAG_FILESTREAM;
+			ip->i_d.di_flags |= di_flags;
+		}
+		/* FALLTHROUGH */
+	case S_IFLNK:
+		ip->i_d.di_format = XFS_DINODE_FMT_EXTENTS;
+		ip->i_df.if_flags = XFS_IFEXTENTS;
+		ip->i_df.if_bytes = ip->i_df.if_real_bytes = 0;
+		ip->i_df.if_u1.if_extents = NULL;
+		break;
+	default:
+		ASSERT(0);
+	}
+	/*
+	 * Attribute fork settings for new inode.
+	 */
+	ip->i_d.di_aformat = XFS_DINODE_FMT_EXTENTS;
+	ip->i_d.di_anextents = 0;
+
+	/*
+	 * Log the new values stuffed into the inode.
+	 */
+	xfs_trans_ijoin_ref(tp, ip, XFS_ILOCK_EXCL);
+	xfs_trans_log_inode(tp, ip, flags);
+
+	/* now that we have an i_mode we can setup inode ops and unlock */
+	xfs_setup_inode(ip);
+
+	/* now we have set up the vfs inode we can associate the filestream */
+	if (filestreams) {
+		error = xfs_filestream_associate(pip, ip);
+		if (error < 0)
+			return -error;
+		if (!error)
+			xfs_iflags_set(ip, XFS_IFILESTREAM);
+	}
+
+	*ipp = ip;
+	return 0;
+}
+
+/*
+ * Check to make sure that there are no blocks allocated to the
+ * file beyond the size of the file.  We don't check this for
+ * files with fixed size extents or real time extents, but we
+ * at least do it for regular files.
+ */
+#ifdef DEBUG
+void
+xfs_isize_check(
+	xfs_mount_t	*mp,
+	xfs_inode_t	*ip,
+	xfs_fsize_t	isize)
+{
+	xfs_fileoff_t	map_first;
+	int		nimaps;
+	xfs_bmbt_irec_t	imaps[2];
+
+	if ((ip->i_d.di_mode & S_IFMT) != S_IFREG)
+		return;
+
+	if (XFS_IS_REALTIME_INODE(ip))
+		return;
+
+	if (ip->i_d.di_flags & XFS_DIFLAG_EXTSIZE)
+		return;
+
+	nimaps = 2;
+	map_first = XFS_B_TO_FSB(mp, (xfs_ufsize_t)isize);
+	/*
+	 * The filesystem could be shutting down, so bmapi may return
+	 * an error.
+	 */
+	if (xfs_bmapi(NULL, ip, map_first,
+			 (XFS_B_TO_FSB(mp,
+				       (xfs_ufsize_t)XFS_MAXIOFFSET(mp)) -
+			  map_first),
+			 XFS_BMAPI_ENTIRE, NULL, 0, imaps, &nimaps,
+			 NULL))
+	    return;
+	ASSERT(nimaps == 1);
+	ASSERT(imaps[0].br_startblock == HOLESTARTBLOCK);
+}
+#endif	/* DEBUG */
+
+/*
+ * Calculate the last possible buffered byte in a file.  This must
+ * include data that was buffered beyond the EOF by the write code.
+ * This also needs to deal with overflowing the xfs_fsize_t type
+ * which can happen for sizes near the limit.
+ *
+ * We also need to take into account any blocks beyond the EOF.  It
+ * may be the case that they were buffered by a write which failed.
+ * In that case the pages will still be in memory, but the inode size
+ * will never have been updated.
+ */
+STATIC xfs_fsize_t
+xfs_file_last_byte(
+	xfs_inode_t	*ip)
+{
+	xfs_mount_t	*mp;
+	xfs_fsize_t	last_byte;
+	xfs_fileoff_t	last_block;
+	xfs_fileoff_t	size_last_block;
+	int		error;
+
+	ASSERT(xfs_isilocked(ip, XFS_IOLOCK_EXCL|XFS_IOLOCK_SHARED));
+
+	mp = ip->i_mount;
+	/*
+	 * Only check for blocks beyond the EOF if the extents have
+	 * been read in.  This eliminates the need for the inode lock,
+	 * and it also saves us from looking when it really isn't
+	 * necessary.
+	 */
+	if (ip->i_df.if_flags & XFS_IFEXTENTS) {
+		xfs_ilock(ip, XFS_ILOCK_SHARED);
+		error = xfs_bmap_last_offset(NULL, ip, &last_block,
+			XFS_DATA_FORK);
+		xfs_iunlock(ip, XFS_ILOCK_SHARED);
+		if (error) {
+			last_block = 0;
+		}
+	} else {
+		last_block = 0;
+	}
+	size_last_block = XFS_B_TO_FSB(mp, (xfs_ufsize_t)ip->i_size);
+	last_block = XFS_FILEOFF_MAX(last_block, size_last_block);
+
+	last_byte = XFS_FSB_TO_B(mp, last_block);
+	if (last_byte < 0) {
+		return XFS_MAXIOFFSET(mp);
+	}
+	last_byte += (1 << mp->m_writeio_log);
+	if (last_byte < 0) {
+		return XFS_MAXIOFFSET(mp);
+	}
+	return last_byte;
+}
+
+/*
+ * Start the truncation of the file to new_size.  The new size
+ * must be smaller than the current size.  This routine will
+ * clear the buffer and page caches of file data in the removed
+ * range, and xfs_itruncate_finish() will remove the underlying
+ * disk blocks.
+ *
+ * The inode must have its I/O lock locked EXCLUSIVELY, and it
+ * must NOT have the inode lock held at all.  This is because we're
+ * calling into the buffer/page cache code and we can't hold the
+ * inode lock when we do so.
+ *
+ * We need to wait for any direct I/Os in flight to complete before we
+ * proceed with the truncate. This is needed to prevent the extents
+ * being read or written by the direct I/Os from being removed while the
+ * I/O is in flight as there is no other method of synchronising
+ * direct I/O with the truncate operation.  Also, because we hold
+ * the IOLOCK in exclusive mode, we prevent new direct I/Os from being
+ * started until the truncate completes and drops the lock. Essentially,
+ * the xfs_ioend_wait() call forms an I/O barrier that provides strict
+ * ordering between direct I/Os and the truncate operation.
+ *
+ * The flags parameter can have either the value XFS_ITRUNC_DEFINITE
+ * or XFS_ITRUNC_MAYBE.  The XFS_ITRUNC_MAYBE value should be used
+ * in the case that the caller is locking things out of order and
+ * may not be able to call xfs_itruncate_finish() with the inode lock
+ * held without dropping the I/O lock.  If the caller must drop the
+ * I/O lock before calling xfs_itruncate_finish(), then xfs_itruncate_start()
+ * must be called again with all the same restrictions as the initial
+ * call.
+ */
+int
+xfs_itruncate_start(
+	xfs_inode_t	*ip,
+	uint		flags,
+	xfs_fsize_t	new_size)
+{
+	xfs_fsize_t	last_byte;
+	xfs_off_t	toss_start;
+	xfs_mount_t	*mp;
+	int		error = 0;
+
+	ASSERT(xfs_isilocked(ip, XFS_IOLOCK_EXCL));
+	ASSERT((new_size == 0) || (new_size <= ip->i_size));
+	ASSERT((flags == XFS_ITRUNC_DEFINITE) ||
+	       (flags == XFS_ITRUNC_MAYBE));
+
+	mp = ip->i_mount;
+
+	/* wait for the completion of any pending DIOs */
+	if (new_size == 0 || new_size < ip->i_size)
+		xfs_ioend_wait(ip);
+
+	/*
+	 * Call toss_pages or flushinval_pages to get rid of pages
+	 * overlapping the region being removed.  We have to use
+	 * the less efficient flushinval_pages in the case that the
+	 * caller may not be able to finish the truncate without
+	 * dropping the inode's I/O lock.  Make sure
+	 * to catch any pages brought in by buffers overlapping
+	 * the EOF by searching out beyond the isize by our
+	 * block size. We round new_size up to a block boundary
+	 * so that we don't toss things on the same block as
+	 * new_size but before it.
+	 *
+	 * Before calling toss_page or flushinval_pages, make sure to
+	 * call remapf() over the same region if the file is mapped.
+	 * This frees up mapped file references to the pages in the
+	 * given range and for the flushinval_pages case it ensures
+	 * that we get the latest mapped changes flushed out.
+	 */
+	toss_start = XFS_B_TO_FSB(mp, (xfs_ufsize_t)new_size);
+	toss_start = XFS_FSB_TO_B(mp, toss_start);
+	if (toss_start < 0) {
+		/*
+		 * The place to start tossing is beyond our maximum
+		 * file size, so there is no way that the data extended
+		 * out there.
+		 */
+		return 0;
+	}
+	last_byte = xfs_file_last_byte(ip);
+	trace_xfs_itruncate_start(ip, new_size, flags, toss_start, last_byte);
+	if (last_byte > toss_start) {
+		if (flags & XFS_ITRUNC_DEFINITE) {
+			xfs_tosspages(ip, toss_start,
+					-1, FI_REMAPF_LOCKED);
+		} else {
+			error = xfs_flushinval_pages(ip, toss_start,
+					-1, FI_REMAPF_LOCKED);
+		}
+	}
+
+#ifdef DEBUG
+	if (new_size == 0) {
+		ASSERT(VN_CACHED(VFS_I(ip)) == 0);
+	}
+#endif
+	return error;
+}
+
+/*
+ * Shrink the file to the given new_size.  The new size must be smaller than
+ * the current size.  This will free up the underlying blocks in the removed
+ * range after a call to xfs_itruncate_start() or xfs_atruncate_start().
+ *
+ * The transaction passed to this routine must have made a permanent log
+ * reservation of at least XFS_ITRUNCATE_LOG_RES.  This routine may commit the
+ * given transaction and start new ones, so make sure everything involved in
+ * the transaction is tidy before calling here.  Some transaction will be
+ * returned to the caller to be committed.  The incoming transaction must
+ * already include the inode, and both inode locks must be held exclusively.
+ * The inode must also be "held" within the transaction.  On return the inode
+ * will be "held" within the returned transaction.  This routine does NOT
+ * require any disk space to be reserved for it within the transaction.
+ *
+ * The fork parameter must be either xfs_attr_fork or xfs_data_fork, and it
+ * indicates the fork which is to be truncated.  For the attribute fork we only
+ * support truncation to size 0.
+ *
+ * We use the sync parameter to indicate whether or not the first transaction
+ * we perform might have to be synchronous.  For the attr fork, it needs to be
+ * so if the unlink of the inode is not yet known to be permanent in the log.
+ * This keeps us from freeing and reusing the blocks of the attribute fork
+ * before the unlink of the inode becomes permanent.
+ *
+ * For the data fork, we normally have to run synchronously if we're being
+ * called out of the inactive path or we're being called out of the create path
+ * where we're truncating an existing file.  Either way, the truncate needs to
+ * be sync so blocks don't reappear in the file with altered data in case of a
+ * crash.  wsync filesystems can run the first case async because anything that
+ * shrinks the inode has to run sync so by the time we're called here from
+ * inactive, the inode size is permanently set to 0.
+ *
+ * Calls from the truncate path always need to be sync unless we're in a wsync
+ * filesystem and the file has already been unlinked.
+ *
+ * The caller is responsible for correctly setting the sync parameter.  It gets
+ * too hard for us to guess here which path we're being called out of just
+ * based on inode state.
+ *
+ * If we get an error, we must return with the inode locked and linked into the
+ * current transaction. This keeps things simple for the higher level code,
+ * because it always knows that the inode is locked and held in the transaction
+ * that returns to it whether errors occur or not.  We don't mark the inode
+ * dirty on error so that transactions can be easily aborted if possible.
+ */
+int
+xfs_itruncate_finish(
+	xfs_trans_t	**tp,
+	xfs_inode_t	*ip,
+	xfs_fsize_t	new_size,
+	int		fork,
+	int		sync)
+{
+	xfs_fsblock_t	first_block;
+	xfs_fileoff_t	first_unmap_block;
+	xfs_fileoff_t	last_block;
+	xfs_filblks_t	unmap_len=0;
+	xfs_mount_t	*mp;
+	xfs_trans_t	*ntp;
+	int		done;
+	int		committed;
+	xfs_bmap_free_t	free_list;
+	int		error;
+
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL|XFS_IOLOCK_EXCL));
+	ASSERT((new_size == 0) || (new_size <= ip->i_size));
+	ASSERT(*tp != NULL);
+	ASSERT((*tp)->t_flags & XFS_TRANS_PERM_LOG_RES);
+	ASSERT(ip->i_transp == *tp);
+	ASSERT(ip->i_itemp != NULL);
+	ASSERT(ip->i_itemp->ili_lock_flags == 0);
+
+
+	ntp = *tp;
+	mp = (ntp)->t_mountp;
+	ASSERT(! XFS_NOT_DQATTACHED(mp, ip));
+
+	/*
+	 * We only support truncating the entire attribute fork.
+	 */
+	if (fork == XFS_ATTR_FORK) {
+		new_size = 0LL;
+	}
+	first_unmap_block = XFS_B_TO_FSB(mp, (xfs_ufsize_t)new_size);
+	trace_xfs_itruncate_finish_start(ip, new_size);
+
+	/*
+	 * The first thing we do is set the size to new_size permanently
+	 * on disk.  This way we don't have to worry about anyone ever
+	 * being able to look at the data being freed even in the face
+	 * of a crash.  What we're getting around here is the case where
+	 * we free a block, it is allocated to another file, it is written
+	 * to, and then we crash.  If the new data gets written to the
+	 * file but the log buffers containing the free and reallocation
+	 * don't, then we'd end up with garbage in the blocks being freed.
+	 * As long as we make the new_size permanent before actually
+	 * freeing any blocks it doesn't matter if they get written to.
+	 *
+	 * The callers must signal into us whether or not the size
+	 * setting here must be synchronous.  There are a few cases
+	 * where it doesn't have to be synchronous.  Those cases
+	 * occur if the file is unlinked and we know the unlink is
+	 * permanent or if the blocks being truncated are guaranteed
+	 * to be beyond the inode eof (regardless of the link count)
+	 * and the eof value is permanent.  Both of these cases occur
+	 * only on wsync-mounted filesystems.  In those cases, we're
+	 * guaranteed that no user will ever see the data in the blocks
+	 * that are being truncated so the truncate can run async.
+	 * In the free beyond eof case, the file may wind up with
+	 * more blocks allocated to it than it needs if we crash
+	 * and that won't get fixed until the next time the file
+	 * is re-opened and closed but that's ok as that shouldn't
+	 * be too many blocks.
+	 *
+	 * However, we can't just make all wsync xactions run async
+	 * because there's one call out of the create path that needs
+	 * to run sync where it's truncating an existing file to size
+	 * 0 whose size is > 0.
+	 *
+	 * It's probably possible to come up with a test in this
+	 * routine that would correctly distinguish all the above
+	 * cases from the values of the function parameters and the
+	 * inode state but for sanity's sake, I've decided to let the
+	 * layers above just tell us.  It's simpler to correctly figure
+	 * out in the layer above exactly under what conditions we
+	 * can run async and I think it's easier for others read and
+	 * follow the logic in case something has to be changed.
+	 * cscope is your friend -- rcc.
+	 *
+	 * The attribute fork is much simpler.
+	 *
+	 * For the attribute fork we allow the caller to tell us whether
+	 * the unlink of the inode that led to this call is yet permanent
+	 * in the on disk log.  If it is not and we will be freeing extents
+	 * in this inode then we make the first transaction synchronous
+	 * to make sure that the unlink is permanent by the time we free
+	 * the blocks.
+	 */
+	if (fork == XFS_DATA_FORK) {
+		if (ip->i_d.di_nextents > 0) {
+			/*
+			 * If we are not changing the file size then do
+			 * not update the on-disk file size - we may be
+			 * called from xfs_inactive_free_eofblocks().  If we
+			 * update the on-disk file size and then the system
+			 * crashes before the contents of the file are
+			 * flushed to disk then the files may be full of
+			 * holes (ie NULL files bug).
+			 */
+			if (ip->i_size != new_size) {
+				ip->i_d.di_size = new_size;
+				ip->i_size = new_size;
+				xfs_trans_log_inode(ntp, ip, XFS_ILOG_CORE);
+			}
+		}
+	}
+
+	/*
+	 * Since it is possible for space to become allocated beyond
+	 * the end of the file (in a crash where the space is allocated
+	 * but the inode size is not yet updated), simply remove any
+	 * blocks which show up between the new EOF and the maximum
+	 * possible file size.  If the first block to be removed is
+	 * beyond the maximum file size (ie it is the same as last_block),
+	 * then there is nothing to do.
+	 */
+	last_block = XFS_B_TO_FSB(mp, (xfs_ufsize_t)XFS_MAXIOFFSET(mp));
+	ASSERT(first_unmap_block <= last_block);
+	done = 0;
+	if (last_block == first_unmap_block) {
+		done = 1;
+	} else {
+		unmap_len = last_block - first_unmap_block + 1;
+	}
+	while (!done) {
+		/*
+		 * Free up up to XFS_ITRUNC_MAX_EXTENTS.  xfs_bunmapi()
+		 * will tell us whether it freed the entire range or
+		 * not.  If this is a synchronous mount (wsync),
+		 * then we can tell bunmapi to keep all the
+		 * transactions asynchronous since the unlink
+		 * transaction that made this inode inactive has
+		 * already hit the disk.  There's no danger of
+		 * the freed blocks being reused, there being a
+		 * crash, and the reused blocks suddenly reappearing
+		 * in this file with garbage in them once recovery
+		 * runs.
+		 */
+		xfs_bmap_init(&free_list, &first_block);
+		error = xfs_bunmapi(ntp, ip,
+				    first_unmap_block, unmap_len,
+				    xfs_bmapi_aflag(fork),
+				    XFS_ITRUNC_MAX_EXTENTS,
+				    &first_block, &free_list,
+				    &done);
+		if (error) {
+			/*
+			 * If the bunmapi call encounters an error,
+			 * return to the caller where the transaction
+			 * can be properly aborted.  We just need to
+			 * make sure we're not holding any resources
+			 * that we were not when we came in.
+			 */
+			xfs_bmap_cancel(&free_list);
+			return error;
+		}
+
+		/*
+		 * Duplicate the transaction that has the permanent
+		 * reservation and commit the old transaction.
+		 */
+		error = xfs_bmap_finish(tp, &free_list, &committed);
+		ntp = *tp;
+		if (committed)
+			xfs_trans_ijoin(ntp, ip);
+
+		if (error) {
+			/*
+			 * If the bmap finish call encounters an error, return
+			 * to the caller where the transaction can be properly
+			 * aborted.  We just need to make sure we're not
+			 * holding any resources that we were not when we came
+			 * in.
+			 *
+			 * Aborting from this point might lose some blocks in
+			 * the file system, but oh well.
+			 */
+			xfs_bmap_cancel(&free_list);
+			return error;
+		}
+
+		if (committed) {
+			/*
+			 * Mark the inode dirty so it will be logged and
+			 * moved forward in the log as part of every commit.
+			 */
+			xfs_trans_log_inode(ntp, ip, XFS_ILOG_CORE);
+		}
+
+		ntp = xfs_trans_dup(ntp);
+		error = xfs_trans_commit(*tp, 0);
+		*tp = ntp;
+
+		xfs_trans_ijoin(ntp, ip);
+
+		if (error)
+			return error;
+		/*
+		 * transaction commit worked ok so we can drop the extra ticket
+		 * reference that we gained in xfs_trans_dup()
+		 */
+		xfs_log_ticket_put(ntp->t_ticket);
+		error = xfs_trans_reserve(ntp, 0,
+					XFS_ITRUNCATE_LOG_RES(mp), 0,
+					XFS_TRANS_PERM_LOG_RES,
+					XFS_ITRUNCATE_LOG_COUNT);
+		if (error)
+			return error;
+	}
+	/*
+	 * Only update the size in the case of the data fork, but
+	 * always re-log the inode so that our permanent transaction
+	 * can keep on rolling it forward in the log.
+	 */
+	if (fork == XFS_DATA_FORK) {
+		xfs_isize_check(mp, ip, new_size);
+		/*
+		 * If we are not changing the file size then do
+		 * not update the on-disk file size - we may be
+		 * called from xfs_inactive_free_eofblocks().  If we
+		 * update the on-disk file size and then the system
+		 * crashes before the contents of the file are
+		 * flushed to disk then the files may be full of
+		 * holes (ie NULL files bug).
+		 */
+		if (ip->i_size != new_size) {
+			ip->i_d.di_size = new_size;
+			ip->i_size = new_size;
+		}
+	}
+	xfs_trans_log_inode(ntp, ip, XFS_ILOG_CORE);
+	ASSERT((new_size != 0) ||
+	       (fork == XFS_ATTR_FORK) ||
+	       (ip->i_delayed_blks == 0));
+	ASSERT((new_size != 0) ||
+	       (fork == XFS_ATTR_FORK) ||
+	       (ip->i_d.di_nextents == 0));
+	trace_xfs_itruncate_finish_end(ip, new_size);
+	return 0;
+}
+
+/*
+ * This is called when the inode's link count goes to 0.
+ * We place the on-disk inode on a list in the AGI.  It
+ * will be pulled from this list when the inode is freed.
+ */
+int
+xfs_iunlink(
+	xfs_trans_t	*tp,
+	xfs_inode_t	*ip)
+{
+	xfs_mount_t	*mp;
+	xfs_agi_t	*agi;
+	xfs_dinode_t	*dip;
+	xfs_buf_t	*agibp;
+	xfs_buf_t	*ibp;
+	xfs_agino_t	agino;
+	short		bucket_index;
+	int		offset;
+	int		error;
+
+	ASSERT(ip->i_d.di_nlink == 0);
+	ASSERT(ip->i_d.di_mode != 0);
+	ASSERT(ip->i_transp == tp);
+
+	mp = tp->t_mountp;
+
+	/*
+	 * Get the agi buffer first.  It ensures lock ordering
+	 * on the list.
+	 */
+	error = xfs_read_agi(mp, tp, XFS_INO_TO_AGNO(mp, ip->i_ino), &agibp);
+	if (error)
+		return error;
+	agi = XFS_BUF_TO_AGI(agibp);
+
+	/*
+	 * Get the index into the agi hash table for the
+	 * list this inode will go on.
+	 */
+	agino = XFS_INO_TO_AGINO(mp, ip->i_ino);
+	ASSERT(agino != 0);
+	bucket_index = agino % XFS_AGI_UNLINKED_BUCKETS;
+	ASSERT(agi->agi_unlinked[bucket_index]);
+	ASSERT(be32_to_cpu(agi->agi_unlinked[bucket_index]) != agino);
+
+	if (be32_to_cpu(agi->agi_unlinked[bucket_index]) != NULLAGINO) {
+		/*
+		 * There is already another inode in the bucket we need
+		 * to add ourselves to.  Add us at the front of the list.
+		 * Here we put the head pointer into our next pointer,
+		 * and then we fall through to point the head at us.
+		 */
+		error = xfs_itobp(mp, tp, ip, &dip, &ibp, XBF_LOCK);
+		if (error)
+			return error;
+
+		ASSERT(be32_to_cpu(dip->di_next_unlinked) == NULLAGINO);
+		/* both on-disk, don't endian flip twice */
+		dip->di_next_unlinked = agi->agi_unlinked[bucket_index];
+		offset = ip->i_imap.im_boffset +
+			offsetof(xfs_dinode_t, di_next_unlinked);
+		xfs_trans_inode_buf(tp, ibp);
+		xfs_trans_log_buf(tp, ibp, offset,
+				  (offset + sizeof(xfs_agino_t) - 1));
+		xfs_inobp_check(mp, ibp);
+	}
+
+	/*
+	 * Point the bucket head pointer at the inode being inserted.
+	 */
+	ASSERT(agino != 0);
+	agi->agi_unlinked[bucket_index] = cpu_to_be32(agino);
+	offset = offsetof(xfs_agi_t, agi_unlinked) +
+		(sizeof(xfs_agino_t) * bucket_index);
+	xfs_trans_log_buf(tp, agibp, offset,
+			  (offset + sizeof(xfs_agino_t) - 1));
+	return 0;
+}
+
+/*
+ * Pull the on-disk inode from the AGI unlinked list.
+ */
+STATIC int
+xfs_iunlink_remove(
+	xfs_trans_t	*tp,
+	xfs_inode_t	*ip)
+{
+	xfs_ino_t	next_ino;
+	xfs_mount_t	*mp;
+	xfs_agi_t	*agi;
+	xfs_dinode_t	*dip;
+	xfs_buf_t	*agibp;
+	xfs_buf_t	*ibp;
+	xfs_agnumber_t	agno;
+	xfs_agino_t	agino;
+	xfs_agino_t	next_agino;
+	xfs_buf_t	*last_ibp;
+	xfs_dinode_t	*last_dip = NULL;
+	short		bucket_index;
+	int		offset, last_offset = 0;
+	int		error;
+
+	mp = tp->t_mountp;
+	agno = XFS_INO_TO_AGNO(mp, ip->i_ino);
+
+	/*
+	 * Get the agi buffer first.  It ensures lock ordering
+	 * on the list.
+	 */
+	error = xfs_read_agi(mp, tp, agno, &agibp);
+	if (error)
+		return error;
+
+	agi = XFS_BUF_TO_AGI(agibp);
+
+	/*
+	 * Get the index into the agi hash table for the
+	 * list this inode will go on.
+	 */
+	agino = XFS_INO_TO_AGINO(mp, ip->i_ino);
+	ASSERT(agino != 0);
+	bucket_index = agino % XFS_AGI_UNLINKED_BUCKETS;
+	ASSERT(be32_to_cpu(agi->agi_unlinked[bucket_index]) != NULLAGINO);
+	ASSERT(agi->agi_unlinked[bucket_index]);
+
+	if (be32_to_cpu(agi->agi_unlinked[bucket_index]) == agino) {
+		/*
+		 * We're at the head of the list.  Get the inode's
+		 * on-disk buffer to see if there is anyone after us
+		 * on the list.  Only modify our next pointer if it
+		 * is not already NULLAGINO.  This saves us the overhead
+		 * of dealing with the buffer when there is no need to
+		 * change it.
+		 */
+		error = xfs_itobp(mp, tp, ip, &dip, &ibp, XBF_LOCK);
+		if (error) {
+			xfs_warn(mp, "%s: xfs_itobp() returned error %d.",
+				__func__, error);
+			return error;
+		}
+		next_agino = be32_to_cpu(dip->di_next_unlinked);
+		ASSERT(next_agino != 0);
+		if (next_agino != NULLAGINO) {
+			dip->di_next_unlinked = cpu_to_be32(NULLAGINO);
+			offset = ip->i_imap.im_boffset +
+				offsetof(xfs_dinode_t, di_next_unlinked);
+			xfs_trans_inode_buf(tp, ibp);
+			xfs_trans_log_buf(tp, ibp, offset,
+					  (offset + sizeof(xfs_agino_t) - 1));
+			xfs_inobp_check(mp, ibp);
+		} else {
+			xfs_trans_brelse(tp, ibp);
+		}
+		/*
+		 * Point the bucket head pointer at the next inode.
+		 */
+		ASSERT(next_agino != 0);
+		ASSERT(next_agino != agino);
+		agi->agi_unlinked[bucket_index] = cpu_to_be32(next_agino);
+		offset = offsetof(xfs_agi_t, agi_unlinked) +
+			(sizeof(xfs_agino_t) * bucket_index);
+		xfs_trans_log_buf(tp, agibp, offset,
+				  (offset + sizeof(xfs_agino_t) - 1));
+	} else {
+		/*
+		 * We need to search the list for the inode being freed.
+		 */
+		next_agino = be32_to_cpu(agi->agi_unlinked[bucket_index]);
+		last_ibp = NULL;
+		while (next_agino != agino) {
+			/*
+			 * If the last inode wasn't the one pointing to
+			 * us, then release its buffer since we're not
+			 * going to do anything with it.
+			 */
+			if (last_ibp != NULL) {
+				xfs_trans_brelse(tp, last_ibp);
+			}
+			next_ino = XFS_AGINO_TO_INO(mp, agno, next_agino);
+			error = xfs_inotobp(mp, tp, next_ino, &last_dip,
+					    &last_ibp, &last_offset, 0);
+			if (error) {
+				xfs_warn(mp,
+					"%s: xfs_inotobp() returned error %d.",
+					__func__, error);
+				return error;
+			}
+			next_agino = be32_to_cpu(last_dip->di_next_unlinked);
+			ASSERT(next_agino != NULLAGINO);
+			ASSERT(next_agino != 0);
+		}
+		/*
+		 * Now last_ibp points to the buffer previous to us on
+		 * the unlinked list.  Pull us from the list.
+		 */
+		error = xfs_itobp(mp, tp, ip, &dip, &ibp, XBF_LOCK);
+		if (error) {
+			xfs_warn(mp, "%s: xfs_itobp(2) returned error %d.",
+				__func__, error);
+			return error;
+		}
+		next_agino = be32_to_cpu(dip->di_next_unlinked);
+		ASSERT(next_agino != 0);
+		ASSERT(next_agino != agino);
+		if (next_agino != NULLAGINO) {
+			dip->di_next_unlinked = cpu_to_be32(NULLAGINO);
+			offset = ip->i_imap.im_boffset +
+				offsetof(xfs_dinode_t, di_next_unlinked);
+			xfs_trans_inode_buf(tp, ibp);
+			xfs_trans_log_buf(tp, ibp, offset,
+					  (offset + sizeof(xfs_agino_t) - 1));
+			xfs_inobp_check(mp, ibp);
+		} else {
+			xfs_trans_brelse(tp, ibp);
+		}
+		/*
+		 * Point the previous inode on the list to the next inode.
+		 */
+		last_dip->di_next_unlinked = cpu_to_be32(next_agino);
+		ASSERT(next_agino != 0);
+		offset = last_offset + offsetof(xfs_dinode_t, di_next_unlinked);
+		xfs_trans_inode_buf(tp, last_ibp);
+		xfs_trans_log_buf(tp, last_ibp, offset,
+				  (offset + sizeof(xfs_agino_t) - 1));
+		xfs_inobp_check(mp, last_ibp);
+	}
+	return 0;
+}
+
+/*
+ * A big issue when freeing the inode cluster is is that we _cannot_ skip any
+ * inodes that are in memory - they all must be marked stale and attached to
+ * the cluster buffer.
+ */
+STATIC void
+xfs_ifree_cluster(
+	xfs_inode_t	*free_ip,
+	xfs_trans_t	*tp,
+	xfs_ino_t	inum)
+{
+	xfs_mount_t		*mp = free_ip->i_mount;
+	int			blks_per_cluster;
+	int			nbufs;
+	int			ninodes;
+	int			i, j;
+	xfs_daddr_t		blkno;
+	xfs_buf_t		*bp;
+	xfs_inode_t		*ip;
+	xfs_inode_log_item_t	*iip;
+	xfs_log_item_t		*lip;
+	struct xfs_perag	*pag;
+
+	pag = xfs_perag_get(mp, XFS_INO_TO_AGNO(mp, inum));
+	if (mp->m_sb.sb_blocksize >= XFS_INODE_CLUSTER_SIZE(mp)) {
+		blks_per_cluster = 1;
+		ninodes = mp->m_sb.sb_inopblock;
+		nbufs = XFS_IALLOC_BLOCKS(mp);
+	} else {
+		blks_per_cluster = XFS_INODE_CLUSTER_SIZE(mp) /
+					mp->m_sb.sb_blocksize;
+		ninodes = blks_per_cluster * mp->m_sb.sb_inopblock;
+		nbufs = XFS_IALLOC_BLOCKS(mp) / blks_per_cluster;
+	}
+
+	for (j = 0; j < nbufs; j++, inum += ninodes) {
+		blkno = XFS_AGB_TO_DADDR(mp, XFS_INO_TO_AGNO(mp, inum),
+					 XFS_INO_TO_AGBNO(mp, inum));
+
+		/*
+		 * We obtain and lock the backing buffer first in the process
+		 * here, as we have to ensure that any dirty inode that we
+		 * can't get the flush lock on is attached to the buffer.
+		 * If we scan the in-memory inodes first, then buffer IO can
+		 * complete before we get a lock on it, and hence we may fail
+		 * to mark all the active inodes on the buffer stale.
+		 */
+		bp = xfs_trans_get_buf(tp, mp->m_ddev_targp, blkno,
+					mp->m_bsize * blks_per_cluster,
+					XBF_LOCK);
+
+		/*
+		 * Walk the inodes already attached to the buffer and mark them
+		 * stale. These will all have the flush locks held, so an
+		 * in-memory inode walk can't lock them. By marking them all
+		 * stale first, we will not attempt to lock them in the loop
+		 * below as the XFS_ISTALE flag will be set.
+		 */
+		lip = XFS_BUF_FSPRIVATE(bp, xfs_log_item_t *);
+		while (lip) {
+			if (lip->li_type == XFS_LI_INODE) {
+				iip = (xfs_inode_log_item_t *)lip;
+				ASSERT(iip->ili_logged == 1);
+				lip->li_cb = xfs_istale_done;
+				xfs_trans_ail_copy_lsn(mp->m_ail,
+							&iip->ili_flush_lsn,
+							&iip->ili_item.li_lsn);
+				xfs_iflags_set(iip->ili_inode, XFS_ISTALE);
+			}
+			lip = lip->li_bio_list;
+		}
+
+
+		/*
+		 * For each inode in memory attempt to add it to the inode
+		 * buffer and set it up for being staled on buffer IO
+		 * completion.  This is safe as we've locked out tail pushing
+		 * and flushing by locking the buffer.
+		 *
+		 * We have already marked every inode that was part of a
+		 * transaction stale above, which means there is no point in
+		 * even trying to lock them.
+		 */
+		for (i = 0; i < ninodes; i++) {
+retry:
+			rcu_read_lock();
+			ip = radix_tree_lookup(&pag->pag_ici_root,
+					XFS_INO_TO_AGINO(mp, (inum + i)));
+
+			/* Inode not in memory, nothing to do */
+			if (!ip) {
+				rcu_read_unlock();
+				continue;
+			}
+
+			/*
+			 * because this is an RCU protected lookup, we could
+			 * find a recently freed or even reallocated inode
+			 * during the lookup. We need to check under the
+			 * i_flags_lock for a valid inode here. Skip it if it
+			 * is not valid, the wrong inode or stale.
+			 */
+			spin_lock(&ip->i_flags_lock);
+			if (ip->i_ino != inum + i ||
+			    __xfs_iflags_test(ip, XFS_ISTALE)) {
+				spin_unlock(&ip->i_flags_lock);
+				rcu_read_unlock();
+				continue;
+			}
+			spin_unlock(&ip->i_flags_lock);
+
+			/*
+			 * Don't try to lock/unlock the current inode, but we
+			 * _cannot_ skip the other inodes that we did not find
+			 * in the list attached to the buffer and are not
+			 * already marked stale. If we can't lock it, back off
+			 * and retry.
+			 */
+			if (ip != free_ip &&
+			    !xfs_ilock_nowait(ip, XFS_ILOCK_EXCL)) {
+				rcu_read_unlock();
+				delay(1);
+				goto retry;
+			}
+			rcu_read_unlock();
+
+			xfs_iflock(ip);
+			xfs_iflags_set(ip, XFS_ISTALE);
+
+			/*
+			 * we don't need to attach clean inodes or those only
+			 * with unlogged changes (which we throw away, anyway).
+			 */
+			iip = ip->i_itemp;
+			if (!iip || xfs_inode_clean(ip)) {
+				ASSERT(ip != free_ip);
+				ip->i_update_core = 0;
+				xfs_ifunlock(ip);
+				xfs_iunlock(ip, XFS_ILOCK_EXCL);
+				continue;
+			}
+
+			iip->ili_last_fields = iip->ili_format.ilf_fields;
+			iip->ili_format.ilf_fields = 0;
+			iip->ili_logged = 1;
+			xfs_trans_ail_copy_lsn(mp->m_ail, &iip->ili_flush_lsn,
+						&iip->ili_item.li_lsn);
+
+			xfs_buf_attach_iodone(bp, xfs_istale_done,
+						  &iip->ili_item);
+
+			if (ip != free_ip)
+				xfs_iunlock(ip, XFS_ILOCK_EXCL);
+		}
+
+		xfs_trans_stale_inode_buf(tp, bp);
+		xfs_trans_binval(tp, bp);
+	}
+
+	xfs_perag_put(pag);
+}
+
+/*
+ * This is called to return an inode to the inode free list.
+ * The inode should already be truncated to 0 length and have
+ * no pages associated with it.  This routine also assumes that
+ * the inode is already a part of the transaction.
+ *
+ * The on-disk copy of the inode will have been added to the list
+ * of unlinked inodes in the AGI. We need to remove the inode from
+ * that list atomically with respect to freeing it here.
+ */
+int
+xfs_ifree(
+	xfs_trans_t	*tp,
+	xfs_inode_t	*ip,
+	xfs_bmap_free_t	*flist)
+{
+	int			error;
+	int			delete;
+	xfs_ino_t		first_ino;
+	xfs_dinode_t    	*dip;
+	xfs_buf_t       	*ibp;
+
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
+	ASSERT(ip->i_transp == tp);
+	ASSERT(ip->i_d.di_nlink == 0);
+	ASSERT(ip->i_d.di_nextents == 0);
+	ASSERT(ip->i_d.di_anextents == 0);
+	ASSERT((ip->i_d.di_size == 0 && ip->i_size == 0) ||
+	       ((ip->i_d.di_mode & S_IFMT) != S_IFREG));
+	ASSERT(ip->i_d.di_nblocks == 0);
+
+	/*
+	 * Pull the on-disk inode from the AGI unlinked list.
+	 */
+	error = xfs_iunlink_remove(tp, ip);
+	if (error != 0) {
+		return error;
+	}
+
+	error = xfs_difree(tp, ip->i_ino, flist, &delete, &first_ino);
+	if (error != 0) {
+		return error;
+	}
+	ip->i_d.di_mode = 0;		/* mark incore inode as free */
+	ip->i_d.di_flags = 0;
+	ip->i_d.di_dmevmask = 0;
+	ip->i_d.di_forkoff = 0;		/* mark the attr fork not in use */
+	ip->i_df.if_ext_max =
+		XFS_IFORK_DSIZE(ip) / (uint)sizeof(xfs_bmbt_rec_t);
+	ip->i_d.di_format = XFS_DINODE_FMT_EXTENTS;
+	ip->i_d.di_aformat = XFS_DINODE_FMT_EXTENTS;
+	/*
+	 * Bump the generation count so no one will be confused
+	 * by reincarnations of this inode.
+	 */
+	ip->i_d.di_gen++;
+
+	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
+
+	error = xfs_itobp(ip->i_mount, tp, ip, &dip, &ibp, XBF_LOCK);
+	if (error)
+		return error;
+
+        /*
+	* Clear the on-disk di_mode. This is to prevent xfs_bulkstat
+	* from picking up this inode when it is reclaimed (its incore state
+	* initialzed but not flushed to disk yet). The in-core di_mode is
+	* already cleared  and a corresponding transaction logged.
+	* The hack here just synchronizes the in-core to on-disk
+	* di_mode value in advance before the actual inode sync to disk.
+	* This is OK because the inode is already unlinked and would never
+	* change its di_mode again for this inode generation.
+	* This is a temporary hack that would require a proper fix
+	* in the future.
+	*/
+	dip->di_mode = 0;
+
+	if (delete) {
+		xfs_ifree_cluster(ip, tp, first_ino);
+	}
+
+	return 0;
+}
+
+/*
+ * Reallocate the space for if_broot based on the number of records
+ * being added or deleted as indicated in rec_diff.  Move the records
+ * and pointers in if_broot to fit the new size.  When shrinking this
+ * will eliminate holes between the records and pointers created by
+ * the caller.  When growing this will create holes to be filled in
+ * by the caller.
+ *
+ * The caller must not request to add more records than would fit in
+ * the on-disk inode root.  If the if_broot is currently NULL, then
+ * if we adding records one will be allocated.  The caller must also
+ * not request that the number of records go below zero, although
+ * it can go to zero.
+ *
+ * ip -- the inode whose if_broot area is changing
+ * ext_diff -- the change in the number of records, positive or negative,
+ *	 requested for the if_broot array.
+ */
+void
+xfs_iroot_realloc(
+	xfs_inode_t		*ip,
+	int			rec_diff,
+	int			whichfork)
+{
+	struct xfs_mount	*mp = ip->i_mount;
+	int			cur_max;
+	xfs_ifork_t		*ifp;
+	struct xfs_btree_block	*new_broot;
+	int			new_max;
+	size_t			new_size;
+	char			*np;
+	char			*op;
+
+	/*
+	 * Handle the degenerate case quietly.
+	 */
+	if (rec_diff == 0) {
+		return;
+	}
+
+	ifp = XFS_IFORK_PTR(ip, whichfork);
+	if (rec_diff > 0) {
+		/*
+		 * If there wasn't any memory allocated before, just
+		 * allocate it now and get out.
+		 */
+		if (ifp->if_broot_bytes == 0) {
+			new_size = (size_t)XFS_BMAP_BROOT_SPACE_CALC(rec_diff);
+			ifp->if_broot = kmem_alloc(new_size, KM_SLEEP | KM_NOFS);
+			ifp->if_broot_bytes = (int)new_size;
+			return;
+		}
+
+		/*
+		 * If there is already an existing if_broot, then we need
+		 * to realloc() it and shift the pointers to their new
+		 * location.  The records don't change location because
+		 * they are kept butted up against the btree block header.
+		 */
+		cur_max = xfs_bmbt_maxrecs(mp, ifp->if_broot_bytes, 0);
+		new_max = cur_max + rec_diff;
+		new_size = (size_t)XFS_BMAP_BROOT_SPACE_CALC(new_max);
+		ifp->if_broot = kmem_realloc(ifp->if_broot, new_size,
+				(size_t)XFS_BMAP_BROOT_SPACE_CALC(cur_max), /* old size */
+				KM_SLEEP | KM_NOFS);
+		op = (char *)XFS_BMAP_BROOT_PTR_ADDR(mp, ifp->if_broot, 1,
+						     ifp->if_broot_bytes);
+		np = (char *)XFS_BMAP_BROOT_PTR_ADDR(mp, ifp->if_broot, 1,
+						     (int)new_size);
+		ifp->if_broot_bytes = (int)new_size;
+		ASSERT(ifp->if_broot_bytes <=
+			XFS_IFORK_SIZE(ip, whichfork) + XFS_BROOT_SIZE_ADJ);
+		memmove(np, op, cur_max * (uint)sizeof(xfs_dfsbno_t));
+		return;
+	}
+
+	/*
+	 * rec_diff is less than 0.  In this case, we are shrinking the
+	 * if_broot buffer.  It must already exist.  If we go to zero
+	 * records, just get rid of the root and clear the status bit.
+	 */
+	ASSERT((ifp->if_broot != NULL) && (ifp->if_broot_bytes > 0));
+	cur_max = xfs_bmbt_maxrecs(mp, ifp->if_broot_bytes, 0);
+	new_max = cur_max + rec_diff;
+	ASSERT(new_max >= 0);
+	if (new_max > 0)
+		new_size = (size_t)XFS_BMAP_BROOT_SPACE_CALC(new_max);
+	else
+		new_size = 0;
+	if (new_size > 0) {
+		new_broot = kmem_alloc(new_size, KM_SLEEP | KM_NOFS);
+		/*
+		 * First copy over the btree block header.
+		 */
+		memcpy(new_broot, ifp->if_broot, XFS_BTREE_LBLOCK_LEN);
+	} else {
+		new_broot = NULL;
+		ifp->if_flags &= ~XFS_IFBROOT;
+	}
+
+	/*
+	 * Only copy the records and pointers if there are any.
+	 */
+	if (new_max > 0) {
+		/*
+		 * First copy the records.
+		 */
+		op = (char *)XFS_BMBT_REC_ADDR(mp, ifp->if_broot, 1);
+		np = (char *)XFS_BMBT_REC_ADDR(mp, new_broot, 1);
+		memcpy(np, op, new_max * (uint)sizeof(xfs_bmbt_rec_t));
+
+		/*
+		 * Then copy the pointers.
+		 */
+		op = (char *)XFS_BMAP_BROOT_PTR_ADDR(mp, ifp->if_broot, 1,
+						     ifp->if_broot_bytes);
+		np = (char *)XFS_BMAP_BROOT_PTR_ADDR(mp, new_broot, 1,
+						     (int)new_size);
+		memcpy(np, op, new_max * (uint)sizeof(xfs_dfsbno_t));
+	}
+	kmem_free(ifp->if_broot);
+	ifp->if_broot = new_broot;
+	ifp->if_broot_bytes = (int)new_size;
+	ASSERT(ifp->if_broot_bytes <=
+		XFS_IFORK_SIZE(ip, whichfork) + XFS_BROOT_SIZE_ADJ);
+	return;
+}
+
+
+/*
+ * This is called when the amount of space needed for if_data
+ * is increased or decreased.  The change in size is indicated by
+ * the number of bytes that need to be added or deleted in the
+ * byte_diff parameter.
+ *
+ * If the amount of space needed has decreased below the size of the
+ * inline buffer, then switch to using the inline buffer.  Otherwise,
+ * use kmem_realloc() or kmem_alloc() to adjust the size of the buffer
+ * to what is needed.
+ *
+ * ip -- the inode whose if_data area is changing
+ * byte_diff -- the change in the number of bytes, positive or negative,
+ *	 requested for the if_data array.
+ */
+void
+xfs_idata_realloc(
+	xfs_inode_t	*ip,
+	int		byte_diff,
+	int		whichfork)
+{
+	xfs_ifork_t	*ifp;
+	int		new_size;
+	int		real_size;
+
+	if (byte_diff == 0) {
+		return;
+	}
+
+	ifp = XFS_IFORK_PTR(ip, whichfork);
+	new_size = (int)ifp->if_bytes + byte_diff;
+	ASSERT(new_size >= 0);
+
+	if (new_size == 0) {
+		if (ifp->if_u1.if_data != ifp->if_u2.if_inline_data) {
+			kmem_free(ifp->if_u1.if_data);
+		}
+		ifp->if_u1.if_data = NULL;
+		real_size = 0;
+	} else if (new_size <= sizeof(ifp->if_u2.if_inline_data)) {
+		/*
+		 * If the valid extents/data can fit in if_inline_ext/data,
+		 * copy them from the malloc'd vector and free it.
+		 */
+		if (ifp->if_u1.if_data == NULL) {
+			ifp->if_u1.if_data = ifp->if_u2.if_inline_data;
+		} else if (ifp->if_u1.if_data != ifp->if_u2.if_inline_data) {
+			ASSERT(ifp->if_real_bytes != 0);
+			memcpy(ifp->if_u2.if_inline_data, ifp->if_u1.if_data,
+			      new_size);
+			kmem_free(ifp->if_u1.if_data);
+			ifp->if_u1.if_data = ifp->if_u2.if_inline_data;
+		}
+		real_size = 0;
+	} else {
+		/*
+		 * Stuck with malloc/realloc.
+		 * For inline data, the underlying buffer must be
+		 * a multiple of 4 bytes in size so that it can be
+		 * logged and stay on word boundaries.  We enforce
+		 * that here.
+		 */
+		real_size = roundup(new_size, 4);
+		if (ifp->if_u1.if_data == NULL) {
+			ASSERT(ifp->if_real_bytes == 0);
+			ifp->if_u1.if_data = kmem_alloc(real_size,
+							KM_SLEEP | KM_NOFS);
+		} else if (ifp->if_u1.if_data != ifp->if_u2.if_inline_data) {
+			/*
+			 * Only do the realloc if the underlying size
+			 * is really changing.
+			 */
+			if (ifp->if_real_bytes != real_size) {
+				ifp->if_u1.if_data =
+					kmem_realloc(ifp->if_u1.if_data,
+							real_size,
+							ifp->if_real_bytes,
+							KM_SLEEP | KM_NOFS);
+			}
+		} else {
+			ASSERT(ifp->if_real_bytes == 0);
+			ifp->if_u1.if_data = kmem_alloc(real_size,
+							KM_SLEEP | KM_NOFS);
+			memcpy(ifp->if_u1.if_data, ifp->if_u2.if_inline_data,
+				ifp->if_bytes);
+		}
+	}
+	ifp->if_real_bytes = real_size;
+	ifp->if_bytes = new_size;
+	ASSERT(ifp->if_bytes <= XFS_IFORK_SIZE(ip, whichfork));
+}
+
+void
+xfs_idestroy_fork(
+	xfs_inode_t	*ip,
+	int		whichfork)
+{
+	xfs_ifork_t	*ifp;
+
+	ifp = XFS_IFORK_PTR(ip, whichfork);
+	if (ifp->if_broot != NULL) {
+		kmem_free(ifp->if_broot);
+		ifp->if_broot = NULL;
+	}
+
+	/*
+	 * If the format is local, then we can't have an extents
+	 * array so just look for an inline data array.  If we're
+	 * not local then we may or may not have an extents list,
+	 * so check and free it up if we do.
+	 */
+	if (XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_LOCAL) {
+		if ((ifp->if_u1.if_data != ifp->if_u2.if_inline_data) &&
+		    (ifp->if_u1.if_data != NULL)) {
+			ASSERT(ifp->if_real_bytes != 0);
+			kmem_free(ifp->if_u1.if_data);
+			ifp->if_u1.if_data = NULL;
+			ifp->if_real_bytes = 0;
+		}
+	} else if ((ifp->if_flags & XFS_IFEXTENTS) &&
+		   ((ifp->if_flags & XFS_IFEXTIREC) ||
+		    ((ifp->if_u1.if_extents != NULL) &&
+		     (ifp->if_u1.if_extents != ifp->if_u2.if_inline_ext)))) {
+		ASSERT(ifp->if_real_bytes != 0);
+		xfs_iext_destroy(ifp);
+	}
+	ASSERT(ifp->if_u1.if_extents == NULL ||
+	       ifp->if_u1.if_extents == ifp->if_u2.if_inline_ext);
+	ASSERT(ifp->if_real_bytes == 0);
+	if (whichfork == XFS_ATTR_FORK) {
+		kmem_zone_free(xfs_ifork_zone, ip->i_afp);
+		ip->i_afp = NULL;
+	}
+}
+
+/*
+ * This is called to unpin an inode.  The caller must have the inode locked
+ * in at least shared mode so that the buffer cannot be subsequently pinned
+ * once someone is waiting for it to be unpinned.
+ */
+static void
+xfs_iunpin_nowait(
+	struct xfs_inode	*ip)
+{
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL|XFS_ILOCK_SHARED));
+
+	trace_xfs_inode_unpin_nowait(ip, _RET_IP_);
+
+	/* Give the log a push to start the unpinning I/O */
+	xfs_log_force_lsn(ip->i_mount, ip->i_itemp->ili_last_lsn, 0);
+
+}
+
+void
+xfs_iunpin_wait(
+	struct xfs_inode	*ip)
+{
+	if (xfs_ipincount(ip)) {
+		xfs_iunpin_nowait(ip);
+		wait_event(ip->i_ipin_wait, (xfs_ipincount(ip) == 0));
+	}
+}
+
+/*
+ * xfs_iextents_copy()
+ *
+ * This is called to copy the REAL extents (as opposed to the delayed
+ * allocation extents) from the inode into the given buffer.  It
+ * returns the number of bytes copied into the buffer.
+ *
+ * If there are no delayed allocation extents, then we can just
+ * memcpy() the extents into the buffer.  Otherwise, we need to
+ * examine each extent in turn and skip those which are delayed.
+ */
+int
+xfs_iextents_copy(
+	xfs_inode_t		*ip,
+	xfs_bmbt_rec_t		*dp,
+	int			whichfork)
+{
+	int			copied;
+	int			i;
+	xfs_ifork_t		*ifp;
+	int			nrecs;
+	xfs_fsblock_t		start_block;
+
+	ifp = XFS_IFORK_PTR(ip, whichfork);
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL|XFS_ILOCK_SHARED));
+	ASSERT(ifp->if_bytes > 0);
+
+	nrecs = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
+	XFS_BMAP_TRACE_EXLIST(ip, nrecs, whichfork);
+	ASSERT(nrecs > 0);
+
+	/*
+	 * There are some delayed allocation extents in the
+	 * inode, so copy the extents one at a time and skip
+	 * the delayed ones.  There must be at least one
+	 * non-delayed extent.
+	 */
+	copied = 0;
+	for (i = 0; i < nrecs; i++) {
+		xfs_bmbt_rec_host_t *ep = xfs_iext_get_ext(ifp, i);
+		start_block = xfs_bmbt_get_startblock(ep);
+		if (isnullstartblock(start_block)) {
+			/*
+			 * It's a delayed allocation extent, so skip it.
+			 */
+			continue;
+		}
+
+		/* Translate to on disk format */
+		put_unaligned(cpu_to_be64(ep->l0), &dp->l0);
+		put_unaligned(cpu_to_be64(ep->l1), &dp->l1);
+		dp++;
+		copied++;
+	}
+	ASSERT(copied != 0);
+	xfs_validate_extents(ifp, copied, XFS_EXTFMT_INODE(ip));
+
+	return (copied * (uint)sizeof(xfs_bmbt_rec_t));
+}
+
+/*
+ * Each of the following cases stores data into the same region
+ * of the on-disk inode, so only one of them can be valid at
+ * any given time. While it is possible to have conflicting formats
+ * and log flags, e.g. having XFS_ILOG_?DATA set when the fork is
+ * in EXTENTS format, this can only happen when the fork has
+ * changed formats after being modified but before being flushed.
+ * In these cases, the format always takes precedence, because the
+ * format indicates the current state of the fork.
+ */
+/*ARGSUSED*/
+STATIC void
+xfs_iflush_fork(
+	xfs_inode_t		*ip,
+	xfs_dinode_t		*dip,
+	xfs_inode_log_item_t	*iip,
+	int			whichfork,
+	xfs_buf_t		*bp)
+{
+	char			*cp;
+	xfs_ifork_t		*ifp;
+	xfs_mount_t		*mp;
+#ifdef XFS_TRANS_DEBUG
+	int			first;
+#endif
+	static const short	brootflag[2] =
+		{ XFS_ILOG_DBROOT, XFS_ILOG_ABROOT };
+	static const short	dataflag[2] =
+		{ XFS_ILOG_DDATA, XFS_ILOG_ADATA };
+	static const short	extflag[2] =
+		{ XFS_ILOG_DEXT, XFS_ILOG_AEXT };
+
+	if (!iip)
+		return;
+	ifp = XFS_IFORK_PTR(ip, whichfork);
+	/*
+	 * This can happen if we gave up in iformat in an error path,
+	 * for the attribute fork.
+	 */
+	if (!ifp) {
+		ASSERT(whichfork == XFS_ATTR_FORK);
+		return;
+	}
+	cp = XFS_DFORK_PTR(dip, whichfork);
+	mp = ip->i_mount;
+	switch (XFS_IFORK_FORMAT(ip, whichfork)) {
+	case XFS_DINODE_FMT_LOCAL:
+		if ((iip->ili_format.ilf_fields & dataflag[whichfork]) &&
+		    (ifp->if_bytes > 0)) {
+			ASSERT(ifp->if_u1.if_data != NULL);
+			ASSERT(ifp->if_bytes <= XFS_IFORK_SIZE(ip, whichfork));
+			memcpy(cp, ifp->if_u1.if_data, ifp->if_bytes);
+		}
+		break;
+
+	case XFS_DINODE_FMT_EXTENTS:
+		ASSERT((ifp->if_flags & XFS_IFEXTENTS) ||
+		       !(iip->ili_format.ilf_fields & extflag[whichfork]));
+		if ((iip->ili_format.ilf_fields & extflag[whichfork]) &&
+		    (ifp->if_bytes > 0)) {
+			ASSERT(xfs_iext_get_ext(ifp, 0));
+			ASSERT(XFS_IFORK_NEXTENTS(ip, whichfork) > 0);
+			(void)xfs_iextents_copy(ip, (xfs_bmbt_rec_t *)cp,
+				whichfork);
+		}
+		break;
+
+	case XFS_DINODE_FMT_BTREE:
+		if ((iip->ili_format.ilf_fields & brootflag[whichfork]) &&
+		    (ifp->if_broot_bytes > 0)) {
+			ASSERT(ifp->if_broot != NULL);
+			ASSERT(ifp->if_broot_bytes <=
+			       (XFS_IFORK_SIZE(ip, whichfork) +
+				XFS_BROOT_SIZE_ADJ));
+			xfs_bmbt_to_bmdr(mp, ifp->if_broot, ifp->if_broot_bytes,
+				(xfs_bmdr_block_t *)cp,
+				XFS_DFORK_SIZE(dip, mp, whichfork));
+		}
+		break;
+
+	case XFS_DINODE_FMT_DEV:
+		if (iip->ili_format.ilf_fields & XFS_ILOG_DEV) {
+			ASSERT(whichfork == XFS_DATA_FORK);
+			xfs_dinode_put_rdev(dip, ip->i_df.if_u2.if_rdev);
+		}
+		break;
+
+	case XFS_DINODE_FMT_UUID:
+		if (iip->ili_format.ilf_fields & XFS_ILOG_UUID) {
+			ASSERT(whichfork == XFS_DATA_FORK);
+			memcpy(XFS_DFORK_DPTR(dip),
+			       &ip->i_df.if_u2.if_uuid,
+			       sizeof(uuid_t));
+		}
+		break;
+
+	default:
+		ASSERT(0);
+		break;
+	}
+}
+
+STATIC int
+xfs_iflush_cluster(
+	xfs_inode_t	*ip,
+	xfs_buf_t	*bp)
+{
+	xfs_mount_t		*mp = ip->i_mount;
+	struct xfs_perag	*pag;
+	unsigned long		first_index, mask;
+	unsigned long		inodes_per_cluster;
+	int			ilist_size;
+	xfs_inode_t		**ilist;
+	xfs_inode_t		*iq;
+	int			nr_found;
+	int			clcount = 0;
+	int			bufwasdelwri;
+	int			i;
+
+	pag = xfs_perag_get(mp, XFS_INO_TO_AGNO(mp, ip->i_ino));
+
+	inodes_per_cluster = XFS_INODE_CLUSTER_SIZE(mp) >> mp->m_sb.sb_inodelog;
+	ilist_size = inodes_per_cluster * sizeof(xfs_inode_t *);
+	ilist = kmem_alloc(ilist_size, KM_MAYFAIL|KM_NOFS);
+	if (!ilist)
+		goto out_put;
+
+	mask = ~(((XFS_INODE_CLUSTER_SIZE(mp) >> mp->m_sb.sb_inodelog)) - 1);
+	first_index = XFS_INO_TO_AGINO(mp, ip->i_ino) & mask;
+	rcu_read_lock();
+	/* really need a gang lookup range call here */
+	nr_found = radix_tree_gang_lookup(&pag->pag_ici_root, (void**)ilist,
+					first_index, inodes_per_cluster);
+	if (nr_found == 0)
+		goto out_free;
+
+	for (i = 0; i < nr_found; i++) {
+		iq = ilist[i];
+		if (iq == ip)
+			continue;
+
+		/*
+		 * because this is an RCU protected lookup, we could find a
+		 * recently freed or even reallocated inode during the lookup.
+		 * We need to check under the i_flags_lock for a valid inode
+		 * here. Skip it if it is not valid or the wrong inode.
+		 */
+		spin_lock(&ip->i_flags_lock);
+		if (!ip->i_ino ||
+		    (XFS_INO_TO_AGINO(mp, iq->i_ino) & mask) != first_index) {
+			spin_unlock(&ip->i_flags_lock);
+			continue;
+		}
+		spin_unlock(&ip->i_flags_lock);
+
+		/*
+		 * Do an un-protected check to see if the inode is dirty and
+		 * is a candidate for flushing.  These checks will be repeated
+		 * later after the appropriate locks are acquired.
+		 */
+		if (xfs_inode_clean(iq) && xfs_ipincount(iq) == 0)
+			continue;
+
+		/*
+		 * Try to get locks.  If any are unavailable or it is pinned,
+		 * then this inode cannot be flushed and is skipped.
+		 */
+
+		if (!xfs_ilock_nowait(iq, XFS_ILOCK_SHARED))
+			continue;
+		if (!xfs_iflock_nowait(iq)) {
+			xfs_iunlock(iq, XFS_ILOCK_SHARED);
+			continue;
+		}
+		if (xfs_ipincount(iq)) {
+			xfs_ifunlock(iq);
+			xfs_iunlock(iq, XFS_ILOCK_SHARED);
+			continue;
+		}
+
+		/*
+		 * arriving here means that this inode can be flushed.  First
+		 * re-check that it's dirty before flushing.
+		 */
+		if (!xfs_inode_clean(iq)) {
+			int	error;
+			error = xfs_iflush_int(iq, bp);
+			if (error) {
+				xfs_iunlock(iq, XFS_ILOCK_SHARED);
+				goto cluster_corrupt_out;
+			}
+			clcount++;
+		} else {
+			xfs_ifunlock(iq);
+		}
+		xfs_iunlock(iq, XFS_ILOCK_SHARED);
+	}
+
+	if (clcount) {
+		XFS_STATS_INC(xs_icluster_flushcnt);
+		XFS_STATS_ADD(xs_icluster_flushinode, clcount);
+	}
+
+out_free:
+	rcu_read_unlock();
+	kmem_free(ilist);
+out_put:
+	xfs_perag_put(pag);
+	return 0;
+
+
+cluster_corrupt_out:
+	/*
+	 * Corruption detected in the clustering loop.  Invalidate the
+	 * inode buffer and shut down the filesystem.
+	 */
+	rcu_read_unlock();
+	/*
+	 * Clean up the buffer.  If it was B_DELWRI, just release it --
+	 * brelse can handle it with no problems.  If not, shut down the
+	 * filesystem before releasing the buffer.
+	 */
+	bufwasdelwri = XFS_BUF_ISDELAYWRITE(bp);
+	if (bufwasdelwri)
+		xfs_buf_relse(bp);
+
+	xfs_force_shutdown(mp, SHUTDOWN_CORRUPT_INCORE);
+
+	if (!bufwasdelwri) {
+		/*
+		 * Just like incore_relse: if we have b_iodone functions,
+		 * mark the buffer as an error and call them.  Otherwise
+		 * mark it as stale and brelse.
+		 */
+		if (XFS_BUF_IODONE_FUNC(bp)) {
+			XFS_BUF_UNDONE(bp);
+			XFS_BUF_STALE(bp);
+			XFS_BUF_ERROR(bp,EIO);
+			xfs_buf_ioend(bp, 0);
+		} else {
+			XFS_BUF_STALE(bp);
+			xfs_buf_relse(bp);
+		}
+	}
+
+	/*
+	 * Unlocks the flush lock
+	 */
+	xfs_iflush_abort(iq);
+	kmem_free(ilist);
+	xfs_perag_put(pag);
+	return XFS_ERROR(EFSCORRUPTED);
+}
+
+/*
+ * xfs_iflush() will write a modified inode's changes out to the
+ * inode's on disk home.  The caller must have the inode lock held
+ * in at least shared mode and the inode flush completion must be
+ * active as well.  The inode lock will still be held upon return from
+ * the call and the caller is free to unlock it.
+ * The inode flush will be completed when the inode reaches the disk.
+ * The flags indicate how the inode's buffer should be written out.
+ */
+int
+xfs_iflush(
+	xfs_inode_t		*ip,
+	uint			flags)
+{
+	xfs_inode_log_item_t	*iip;
+	xfs_buf_t		*bp;
+	xfs_dinode_t		*dip;
+	xfs_mount_t		*mp;
+	int			error;
+
+	XFS_STATS_INC(xs_iflush_count);
+
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL|XFS_ILOCK_SHARED));
+	ASSERT(!completion_done(&ip->i_flush));
+	ASSERT(ip->i_d.di_format != XFS_DINODE_FMT_BTREE ||
+	       ip->i_d.di_nextents > ip->i_df.if_ext_max);
+
+	iip = ip->i_itemp;
+	mp = ip->i_mount;
+
+	/*
+	 * We can't flush the inode until it is unpinned, so wait for it if we
+	 * are allowed to block.  We know no one new can pin it, because we are
+	 * holding the inode lock shared and you need to hold it exclusively to
+	 * pin the inode.
+	 *
+	 * If we are not allowed to block, force the log out asynchronously so
+	 * that when we come back the inode will be unpinned. If other inodes
+	 * in the same cluster are dirty, they will probably write the inode
+	 * out for us if they occur after the log force completes.
+	 */
+	if (!(flags & SYNC_WAIT) && xfs_ipincount(ip)) {
+		xfs_iunpin_nowait(ip);
+		xfs_ifunlock(ip);
+		return EAGAIN;
+	}
+	xfs_iunpin_wait(ip);
+
+	/*
+	 * For stale inodes we cannot rely on the backing buffer remaining
+	 * stale in cache for the remaining life of the stale inode and so
+	 * xfs_itobp() below may give us a buffer that no longer contains
+	 * inodes below. We have to check this after ensuring the inode is
+	 * unpinned so that it is safe to reclaim the stale inode after the
+	 * flush call.
+	 */
+	if (xfs_iflags_test(ip, XFS_ISTALE)) {
+		xfs_ifunlock(ip);
+		return 0;
+	}
+
+	/*
+	 * This may have been unpinned because the filesystem is shutting
+	 * down forcibly. If that's the case we must not write this inode
+	 * to disk, because the log record didn't make it to disk!
+	 */
+	if (XFS_FORCED_SHUTDOWN(mp)) {
+		ip->i_update_core = 0;
+		if (iip)
+			iip->ili_format.ilf_fields = 0;
+		xfs_ifunlock(ip);
+		return XFS_ERROR(EIO);
+	}
+
+	/*
+	 * Get the buffer containing the on-disk inode.
+	 */
+	error = xfs_itobp(mp, NULL, ip, &dip, &bp,
+				(flags & SYNC_TRYLOCK) ? XBF_TRYLOCK : XBF_LOCK);
+	if (error || !bp) {
+		xfs_ifunlock(ip);
+		return error;
+	}
+
+	/*
+	 * First flush out the inode that xfs_iflush was called with.
+	 */
+	error = xfs_iflush_int(ip, bp);
+	if (error)
+		goto corrupt_out;
+
+	/*
+	 * If the buffer is pinned then push on the log now so we won't
+	 * get stuck waiting in the write for too long.
+	 */
+	if (XFS_BUF_ISPINNED(bp))
+		xfs_log_force(mp, 0);
+
+	/*
+	 * inode clustering:
+	 * see if other inodes can be gathered into this write
+	 */
+	error = xfs_iflush_cluster(ip, bp);
+	if (error)
+		goto cluster_corrupt_out;
+
+	if (flags & SYNC_WAIT)
+		error = xfs_bwrite(mp, bp);
+	else
+		xfs_bdwrite(mp, bp);
+	return error;
+
+corrupt_out:
+	xfs_buf_relse(bp);
+	xfs_force_shutdown(mp, SHUTDOWN_CORRUPT_INCORE);
+cluster_corrupt_out:
+	/*
+	 * Unlocks the flush lock
+	 */
+	xfs_iflush_abort(ip);
+	return XFS_ERROR(EFSCORRUPTED);
+}
+
+
+STATIC int
+xfs_iflush_int(
+	xfs_inode_t		*ip,
+	xfs_buf_t		*bp)
+{
+	xfs_inode_log_item_t	*iip;
+	xfs_dinode_t		*dip;
+	xfs_mount_t		*mp;
+#ifdef XFS_TRANS_DEBUG
+	int			first;
+#endif
+
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL|XFS_ILOCK_SHARED));
+	ASSERT(!completion_done(&ip->i_flush));
+	ASSERT(ip->i_d.di_format != XFS_DINODE_FMT_BTREE ||
+	       ip->i_d.di_nextents > ip->i_df.if_ext_max);
+
+	iip = ip->i_itemp;
+	mp = ip->i_mount;
+
+	/* set *dip = inode's place in the buffer */
+	dip = (xfs_dinode_t *)xfs_buf_offset(bp, ip->i_imap.im_boffset);
+
+	/*
+	 * Clear i_update_core before copying out the data.
+	 * This is for coordination with our timestamp updates
+	 * that don't hold the inode lock. They will always
+	 * update the timestamps BEFORE setting i_update_core,
+	 * so if we clear i_update_core after they set it we
+	 * are guaranteed to see their updates to the timestamps.
+	 * I believe that this depends on strongly ordered memory
+	 * semantics, but we have that.  We use the SYNCHRONIZE
+	 * macro to make sure that the compiler does not reorder
+	 * the i_update_core access below the data copy below.
+	 */
+	ip->i_update_core = 0;
+	SYNCHRONIZE();
+
+	/*
+	 * Make sure to get the latest timestamps from the Linux inode.
+	 */
+	xfs_synchronize_times(ip);
+
+	if (XFS_TEST_ERROR(be16_to_cpu(dip->di_magic) != XFS_DINODE_MAGIC,
+			       mp, XFS_ERRTAG_IFLUSH_1, XFS_RANDOM_IFLUSH_1)) {
+		xfs_alert_tag(mp, XFS_PTAG_IFLUSH,
+			"%s: Bad inode %Lu magic number 0x%x, ptr 0x%p",
+			__func__, ip->i_ino, be16_to_cpu(dip->di_magic), dip);
+		goto corrupt_out;
+	}
+	if (XFS_TEST_ERROR(ip->i_d.di_magic != XFS_DINODE_MAGIC,
+				mp, XFS_ERRTAG_IFLUSH_2, XFS_RANDOM_IFLUSH_2)) {
+		xfs_alert_tag(mp, XFS_PTAG_IFLUSH,
+			"%s: Bad inode %Lu, ptr 0x%p, magic number 0x%x",
+			__func__, ip->i_ino, ip, ip->i_d.di_magic);
+		goto corrupt_out;
+	}
+	if ((ip->i_d.di_mode & S_IFMT) == S_IFREG) {
+		if (XFS_TEST_ERROR(
+		    (ip->i_d.di_format != XFS_DINODE_FMT_EXTENTS) &&
+		    (ip->i_d.di_format != XFS_DINODE_FMT_BTREE),
+		    mp, XFS_ERRTAG_IFLUSH_3, XFS_RANDOM_IFLUSH_3)) {
+			xfs_alert_tag(mp, XFS_PTAG_IFLUSH,
+				"%s: Bad regular inode %Lu, ptr 0x%p",
+				__func__, ip->i_ino, ip);
+			goto corrupt_out;
+		}
+	} else if ((ip->i_d.di_mode & S_IFMT) == S_IFDIR) {
+		if (XFS_TEST_ERROR(
+		    (ip->i_d.di_format != XFS_DINODE_FMT_EXTENTS) &&
+		    (ip->i_d.di_format != XFS_DINODE_FMT_BTREE) &&
+		    (ip->i_d.di_format != XFS_DINODE_FMT_LOCAL),
+		    mp, XFS_ERRTAG_IFLUSH_4, XFS_RANDOM_IFLUSH_4)) {
+			xfs_alert_tag(mp, XFS_PTAG_IFLUSH,
+				"%s: Bad directory inode %Lu, ptr 0x%p",
+				__func__, ip->i_ino, ip);
+			goto corrupt_out;
+		}
+	}
+	if (XFS_TEST_ERROR(ip->i_d.di_nextents + ip->i_d.di_anextents >
+				ip->i_d.di_nblocks, mp, XFS_ERRTAG_IFLUSH_5,
+				XFS_RANDOM_IFLUSH_5)) {
+		xfs_alert_tag(mp, XFS_PTAG_IFLUSH,
+			"%s: detected corrupt incore inode %Lu, "
+			"total extents = %d, nblocks = %Ld, ptr 0x%p",
+			__func__, ip->i_ino,
+			ip->i_d.di_nextents + ip->i_d.di_anextents,
+			ip->i_d.di_nblocks, ip);
+		goto corrupt_out;
+	}
+	if (XFS_TEST_ERROR(ip->i_d.di_forkoff > mp->m_sb.sb_inodesize,
+				mp, XFS_ERRTAG_IFLUSH_6, XFS_RANDOM_IFLUSH_6)) {
+		xfs_alert_tag(mp, XFS_PTAG_IFLUSH,
+			"%s: bad inode %Lu, forkoff 0x%x, ptr 0x%p",
+			__func__, ip->i_ino, ip->i_d.di_forkoff, ip);
+		goto corrupt_out;
+	}
+	/*
+	 * bump the flush iteration count, used to detect flushes which
+	 * postdate a log record during recovery.
+	 */
+
+	ip->i_d.di_flushiter++;
+
+	/*
+	 * Copy the dirty parts of the inode into the on-disk
+	 * inode.  We always copy out the core of the inode,
+	 * because if the inode is dirty at all the core must
+	 * be.
+	 */
+	xfs_dinode_to_disk(dip, &ip->i_d);
+
+	/* Wrap, we never let the log put out DI_MAX_FLUSH */
+	if (ip->i_d.di_flushiter == DI_MAX_FLUSH)
+		ip->i_d.di_flushiter = 0;
+
+	/*
+	 * If this is really an old format inode and the superblock version
+	 * has not been updated to support only new format inodes, then
+	 * convert back to the old inode format.  If the superblock version
+	 * has been updated, then make the conversion permanent.
+	 */
+	ASSERT(ip->i_d.di_version == 1 || xfs_sb_version_hasnlink(&mp->m_sb));
+	if (ip->i_d.di_version == 1) {
+		if (!xfs_sb_version_hasnlink(&mp->m_sb)) {
+			/*
+			 * Convert it back.
+			 */
+			ASSERT(ip->i_d.di_nlink <= XFS_MAXLINK_1);
+			dip->di_onlink = cpu_to_be16(ip->i_d.di_nlink);
+		} else {
+			/*
+			 * The superblock version has already been bumped,
+			 * so just make the conversion to the new inode
+			 * format permanent.
+			 */
+			ip->i_d.di_version = 2;
+			dip->di_version = 2;
+			ip->i_d.di_onlink = 0;
+			dip->di_onlink = 0;
+			memset(&(ip->i_d.di_pad[0]), 0, sizeof(ip->i_d.di_pad));
+			memset(&(dip->di_pad[0]), 0,
+			      sizeof(dip->di_pad));
+			ASSERT(xfs_get_projid(ip) == 0);
+		}
+	}
+
+	xfs_iflush_fork(ip, dip, iip, XFS_DATA_FORK, bp);
+	if (XFS_IFORK_Q(ip))
+		xfs_iflush_fork(ip, dip, iip, XFS_ATTR_FORK, bp);
+	xfs_inobp_check(mp, bp);
+
+	/*
+	 * We've recorded everything logged in the inode, so we'd
+	 * like to clear the ilf_fields bits so we don't log and
+	 * flush things unnecessarily.  However, we can't stop
+	 * logging all this information until the data we've copied
+	 * into the disk buffer is written to disk.  If we did we might
+	 * overwrite the copy of the inode in the log with all the
+	 * data after re-logging only part of it, and in the face of
+	 * a crash we wouldn't have all the data we need to recover.
+	 *
+	 * What we do is move the bits to the ili_last_fields field.
+	 * When logging the inode, these bits are moved back to the
+	 * ilf_fields field.  In the xfs_iflush_done() routine we
+	 * clear ili_last_fields, since we know that the information
+	 * those bits represent is permanently on disk.  As long as
+	 * the flush completes before the inode is logged again, then
+	 * both ilf_fields and ili_last_fields will be cleared.
+	 *
+	 * We can play with the ilf_fields bits here, because the inode
+	 * lock must be held exclusively in order to set bits there
+	 * and the flush lock protects the ili_last_fields bits.
+	 * Set ili_logged so the flush done
+	 * routine can tell whether or not to look in the AIL.
+	 * Also, store the current LSN of the inode so that we can tell
+	 * whether the item has moved in the AIL from xfs_iflush_done().
+	 * In order to read the lsn we need the AIL lock, because
+	 * it is a 64 bit value that cannot be read atomically.
+	 */
+	if (iip != NULL && iip->ili_format.ilf_fields != 0) {
+		iip->ili_last_fields = iip->ili_format.ilf_fields;
+		iip->ili_format.ilf_fields = 0;
+		iip->ili_logged = 1;
+
+		xfs_trans_ail_copy_lsn(mp->m_ail, &iip->ili_flush_lsn,
+					&iip->ili_item.li_lsn);
+
+		/*
+		 * Attach the function xfs_iflush_done to the inode's
+		 * buffer.  This will remove the inode from the AIL
+		 * and unlock the inode's flush lock when the inode is
+		 * completely written to disk.
+		 */
+		xfs_buf_attach_iodone(bp, xfs_iflush_done, &iip->ili_item);
+
+		ASSERT(XFS_BUF_FSPRIVATE(bp, void *) != NULL);
+		ASSERT(XFS_BUF_IODONE_FUNC(bp) != NULL);
+	} else {
+		/*
+		 * We're flushing an inode which is not in the AIL and has
+		 * not been logged but has i_update_core set.  For this
+		 * case we can use a B_DELWRI flush and immediately drop
+		 * the inode flush lock because we can avoid the whole
+		 * AIL state thing.  It's OK to drop the flush lock now,
+		 * because we've already locked the buffer and to do anything
+		 * you really need both.
+		 */
+		if (iip != NULL) {
+			ASSERT(iip->ili_logged == 0);
+			ASSERT(iip->ili_last_fields == 0);
+			ASSERT((iip->ili_item.li_flags & XFS_LI_IN_AIL) == 0);
+		}
+		xfs_ifunlock(ip);
+	}
+
+	return 0;
+
+corrupt_out:
+	return XFS_ERROR(EFSCORRUPTED);
+}
+
+void
+xfs_promote_inode(
+	struct xfs_inode	*ip)
+{
+	struct xfs_buf		*bp;
+
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL|XFS_ILOCK_SHARED));
+
+	bp = xfs_incore(ip->i_mount->m_ddev_targp, ip->i_imap.im_blkno,
+			ip->i_imap.im_len, XBF_TRYLOCK);
+	if (!bp)
+		return;
+
+	if (XFS_BUF_ISDELAYWRITE(bp)) {
+		xfs_buf_delwri_promote(bp);
+		wake_up_process(ip->i_mount->m_ddev_targp->bt_task);
+	}
+
+	xfs_buf_relse(bp);
+}
+
+/*
+ * Return a pointer to the extent record at file index idx.
+ */
+xfs_bmbt_rec_host_t *
+xfs_iext_get_ext(
+	xfs_ifork_t	*ifp,		/* inode fork pointer */
+	xfs_extnum_t	idx)		/* index of target extent */
+{
+	ASSERT(idx >= 0);
+	ASSERT(idx < ifp->if_bytes / sizeof(xfs_bmbt_rec_t));
+
+	if ((ifp->if_flags & XFS_IFEXTIREC) && (idx == 0)) {
+		return ifp->if_u1.if_ext_irec->er_extbuf;
+	} else if (ifp->if_flags & XFS_IFEXTIREC) {
+		xfs_ext_irec_t	*erp;		/* irec pointer */
+		int		erp_idx = 0;	/* irec index */
+		xfs_extnum_t	page_idx = idx;	/* ext index in target list */
+
+		erp = xfs_iext_idx_to_irec(ifp, &page_idx, &erp_idx, 0);
+		return &erp->er_extbuf[page_idx];
+	} else if (ifp->if_bytes) {
+		return &ifp->if_u1.if_extents[idx];
+	} else {
+		return NULL;
+	}
+}
+
+/*
+ * Insert new item(s) into the extent records for incore inode
+ * fork 'ifp'.  'count' new items are inserted at index 'idx'.
+ */
+void
+xfs_iext_insert(
+	xfs_inode_t	*ip,		/* incore inode pointer */
+	xfs_extnum_t	idx,		/* starting index of new items */
+	xfs_extnum_t	count,		/* number of inserted items */
+	xfs_bmbt_irec_t	*new,		/* items to insert */
+	int		state)		/* type of extent conversion */
+{
+	xfs_ifork_t	*ifp = (state & BMAP_ATTRFORK) ? ip->i_afp : &ip->i_df;
+	xfs_extnum_t	i;		/* extent record index */
+
+	trace_xfs_iext_insert(ip, idx, new, state, _RET_IP_);
+
+	ASSERT(ifp->if_flags & XFS_IFEXTENTS);
+	xfs_iext_add(ifp, idx, count);
+	for (i = idx; i < idx + count; i++, new++)
+		xfs_bmbt_set_all(xfs_iext_get_ext(ifp, i), new);
+}
+
+/*
+ * This is called when the amount of space required for incore file
+ * extents needs to be increased. The ext_diff parameter stores the
+ * number of new extents being added and the idx parameter contains
+ * the extent index where the new extents will be added. If the new
+ * extents are being appended, then we just need to (re)allocate and
+ * initialize the space. Otherwise, if the new extents are being
+ * inserted into the middle of the existing entries, a bit more work
+ * is required to make room for the new extents to be inserted. The
+ * caller is responsible for filling in the new extent entries upon
+ * return.
+ */
+void
+xfs_iext_add(
+	xfs_ifork_t	*ifp,		/* inode fork pointer */
+	xfs_extnum_t	idx,		/* index to begin adding exts */
+	int		ext_diff)	/* number of extents to add */
+{
+	int		byte_diff;	/* new bytes being added */
+	int		new_size;	/* size of extents after adding */
+	xfs_extnum_t	nextents;	/* number of extents in file */
+
+	nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
+	ASSERT((idx >= 0) && (idx <= nextents));
+	byte_diff = ext_diff * sizeof(xfs_bmbt_rec_t);
+	new_size = ifp->if_bytes + byte_diff;
+	/*
+	 * If the new number of extents (nextents + ext_diff)
+	 * fits inside the inode, then continue to use the inline
+	 * extent buffer.
+	 */
+	if (nextents + ext_diff <= XFS_INLINE_EXTS) {
+		if (idx < nextents) {
+			memmove(&ifp->if_u2.if_inline_ext[idx + ext_diff],
+				&ifp->if_u2.if_inline_ext[idx],
+				(nextents - idx) * sizeof(xfs_bmbt_rec_t));
+			memset(&ifp->if_u2.if_inline_ext[idx], 0, byte_diff);
+		}
+		ifp->if_u1.if_extents = ifp->if_u2.if_inline_ext;
+		ifp->if_real_bytes = 0;
+	}
+	/*
+	 * Otherwise use a linear (direct) extent list.
+	 * If the extents are currently inside the inode,
+	 * xfs_iext_realloc_direct will switch us from
+	 * inline to direct extent allocation mode.
+	 */
+	else if (nextents + ext_diff <= XFS_LINEAR_EXTS) {
+		xfs_iext_realloc_direct(ifp, new_size);
+		if (idx < nextents) {
+			memmove(&ifp->if_u1.if_extents[idx + ext_diff],
+				&ifp->if_u1.if_extents[idx],
+				(nextents - idx) * sizeof(xfs_bmbt_rec_t));
+			memset(&ifp->if_u1.if_extents[idx], 0, byte_diff);
+		}
+	}
+	/* Indirection array */
+	else {
+		xfs_ext_irec_t	*erp;
+		int		erp_idx = 0;
+		int		page_idx = idx;
+
+		ASSERT(nextents + ext_diff > XFS_LINEAR_EXTS);
+		if (ifp->if_flags & XFS_IFEXTIREC) {
+			erp = xfs_iext_idx_to_irec(ifp, &page_idx, &erp_idx, 1);
+		} else {
+			xfs_iext_irec_init(ifp);
+			ASSERT(ifp->if_flags & XFS_IFEXTIREC);
+			erp = ifp->if_u1.if_ext_irec;
+		}
+		/* Extents fit in target extent page */
+		if (erp && erp->er_extcount + ext_diff <= XFS_LINEAR_EXTS) {
+			if (page_idx < erp->er_extcount) {
+				memmove(&erp->er_extbuf[page_idx + ext_diff],
+					&erp->er_extbuf[page_idx],
+					(erp->er_extcount - page_idx) *
+					sizeof(xfs_bmbt_rec_t));
+				memset(&erp->er_extbuf[page_idx], 0, byte_diff);
+			}
+			erp->er_extcount += ext_diff;
+			xfs_iext_irec_update_extoffs(ifp, erp_idx + 1, ext_diff);
+		}
+		/* Insert a new extent page */
+		else if (erp) {
+			xfs_iext_add_indirect_multi(ifp,
+				erp_idx, page_idx, ext_diff);
+		}
+		/*
+		 * If extent(s) are being appended to the last page in
+		 * the indirection array and the new extent(s) don't fit
+		 * in the page, then erp is NULL and erp_idx is set to
+		 * the next index needed in the indirection array.
+		 */
+		else {
+			int	count = ext_diff;
+
+			while (count) {
+				erp = xfs_iext_irec_new(ifp, erp_idx);
+				erp->er_extcount = count;
+				count -= MIN(count, (int)XFS_LINEAR_EXTS);
+				if (count) {
+					erp_idx++;
+				}
+			}
+		}
+	}
+	ifp->if_bytes = new_size;
+}
+
+/*
+ * This is called when incore extents are being added to the indirection
+ * array and the new extents do not fit in the target extent list. The
+ * erp_idx parameter contains the irec index for the target extent list
+ * in the indirection array, and the idx parameter contains the extent
+ * index within the list. The number of extents being added is stored
+ * in the count parameter.
+ *
+ *    |-------|   |-------|
+ *    |       |   |       |    idx - number of extents before idx
+ *    |  idx  |   | count |
+ *    |       |   |       |    count - number of extents being inserted at idx
+ *    |-------|   |-------|
+ *    | count |   | nex2  |    nex2 - number of extents after idx + count
+ *    |-------|   |-------|
+ */
+void
+xfs_iext_add_indirect_multi(
+	xfs_ifork_t	*ifp,			/* inode fork pointer */
+	int		erp_idx,		/* target extent irec index */
+	xfs_extnum_t	idx,			/* index within target list */
+	int		count)			/* new extents being added */
+{
+	int		byte_diff;		/* new bytes being added */
+	xfs_ext_irec_t	*erp;			/* pointer to irec entry */
+	xfs_extnum_t	ext_diff;		/* number of extents to add */
+	xfs_extnum_t	ext_cnt;		/* new extents still needed */
+	xfs_extnum_t	nex2;			/* extents after idx + count */
+	xfs_bmbt_rec_t	*nex2_ep = NULL;	/* temp list for nex2 extents */
+	int		nlists;			/* number of irec's (lists) */
+
+	ASSERT(ifp->if_flags & XFS_IFEXTIREC);
+	erp = &ifp->if_u1.if_ext_irec[erp_idx];
+	nex2 = erp->er_extcount - idx;
+	nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ;
+
+	/*
+	 * Save second part of target extent list
+	 * (all extents past */
+	if (nex2) {
+		byte_diff = nex2 * sizeof(xfs_bmbt_rec_t);
+		nex2_ep = (xfs_bmbt_rec_t *) kmem_alloc(byte_diff, KM_NOFS);
+		memmove(nex2_ep, &erp->er_extbuf[idx], byte_diff);
+		erp->er_extcount -= nex2;
+		xfs_iext_irec_update_extoffs(ifp, erp_idx + 1, -nex2);
+		memset(&erp->er_extbuf[idx], 0, byte_diff);
+	}
+
+	/*
+	 * Add the new extents to the end of the target
+	 * list, then allocate new irec record(s) and
+	 * extent buffer(s) as needed to store the rest
+	 * of the new extents.
+	 */
+	ext_cnt = count;
+	ext_diff = MIN(ext_cnt, (int)XFS_LINEAR_EXTS - erp->er_extcount);
+	if (ext_diff) {
+		erp->er_extcount += ext_diff;
+		xfs_iext_irec_update_extoffs(ifp, erp_idx + 1, ext_diff);
+		ext_cnt -= ext_diff;
+	}
+	while (ext_cnt) {
+		erp_idx++;
+		erp = xfs_iext_irec_new(ifp, erp_idx);
+		ext_diff = MIN(ext_cnt, (int)XFS_LINEAR_EXTS);
+		erp->er_extcount = ext_diff;
+		xfs_iext_irec_update_extoffs(ifp, erp_idx + 1, ext_diff);
+		ext_cnt -= ext_diff;
+	}
+
+	/* Add nex2 extents back to indirection array */
+	if (nex2) {
+		xfs_extnum_t	ext_avail;
+		int		i;
+
+		byte_diff = nex2 * sizeof(xfs_bmbt_rec_t);
+		ext_avail = XFS_LINEAR_EXTS - erp->er_extcount;
+		i = 0;
+		/*
+		 * If nex2 extents fit in the current page, append
+		 * nex2_ep after the new extents.
+		 */
+		if (nex2 <= ext_avail) {
+			i = erp->er_extcount;
+		}
+		/*
+		 * Otherwise, check if space is available in the
+		 * next page.
+		 */
+		else if ((erp_idx < nlists - 1) &&
+			 (nex2 <= (ext_avail = XFS_LINEAR_EXTS -
+			  ifp->if_u1.if_ext_irec[erp_idx+1].er_extcount))) {
+			erp_idx++;
+			erp++;
+			/* Create a hole for nex2 extents */
+			memmove(&erp->er_extbuf[nex2], erp->er_extbuf,
+				erp->er_extcount * sizeof(xfs_bmbt_rec_t));
+		}
+		/*
+		 * Final choice, create a new extent page for
+		 * nex2 extents.
+		 */
+		else {
+			erp_idx++;
+			erp = xfs_iext_irec_new(ifp, erp_idx);
+		}
+		memmove(&erp->er_extbuf[i], nex2_ep, byte_diff);
+		kmem_free(nex2_ep);
+		erp->er_extcount += nex2;
+		xfs_iext_irec_update_extoffs(ifp, erp_idx + 1, nex2);
+	}
+}
+
+/*
+ * This is called when the amount of space required for incore file
+ * extents needs to be decreased. The ext_diff parameter stores the
+ * number of extents to be removed and the idx parameter contains
+ * the extent index where the extents will be removed from.
+ *
+ * If the amount of space needed has decreased below the linear
+ * limit, XFS_IEXT_BUFSZ, then switch to using the contiguous
+ * extent array.  Otherwise, use kmem_realloc() to adjust the
+ * size to what is needed.
+ */
+void
+xfs_iext_remove(
+	xfs_inode_t	*ip,		/* incore inode pointer */
+	xfs_extnum_t	idx,		/* index to begin removing exts */
+	int		ext_diff,	/* number of extents to remove */
+	int		state)		/* type of extent conversion */
+{
+	xfs_ifork_t	*ifp = (state & BMAP_ATTRFORK) ? ip->i_afp : &ip->i_df;
+	xfs_extnum_t	nextents;	/* number of extents in file */
+	int		new_size;	/* size of extents after removal */
+
+	trace_xfs_iext_remove(ip, idx, state, _RET_IP_);
+
+	ASSERT(ext_diff > 0);
+	nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
+	new_size = (nextents - ext_diff) * sizeof(xfs_bmbt_rec_t);
+
+	if (new_size == 0) {
+		xfs_iext_destroy(ifp);
+	} else if (ifp->if_flags & XFS_IFEXTIREC) {
+		xfs_iext_remove_indirect(ifp, idx, ext_diff);
+	} else if (ifp->if_real_bytes) {
+		xfs_iext_remove_direct(ifp, idx, ext_diff);
+	} else {
+		xfs_iext_remove_inline(ifp, idx, ext_diff);
+	}
+	ifp->if_bytes = new_size;
+}
+
+/*
+ * This removes ext_diff extents from the inline buffer, beginning
+ * at extent index idx.
+ */
+void
+xfs_iext_remove_inline(
+	xfs_ifork_t	*ifp,		/* inode fork pointer */
+	xfs_extnum_t	idx,		/* index to begin removing exts */
+	int		ext_diff)	/* number of extents to remove */
+{
+	int		nextents;	/* number of extents in file */
+
+	ASSERT(!(ifp->if_flags & XFS_IFEXTIREC));
+	ASSERT(idx < XFS_INLINE_EXTS);
+	nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
+	ASSERT(((nextents - ext_diff) > 0) &&
+		(nextents - ext_diff) < XFS_INLINE_EXTS);
+
+	if (idx + ext_diff < nextents) {
+		memmove(&ifp->if_u2.if_inline_ext[idx],
+			&ifp->if_u2.if_inline_ext[idx + ext_diff],
+			(nextents - (idx + ext_diff)) *
+			 sizeof(xfs_bmbt_rec_t));
+		memset(&ifp->if_u2.if_inline_ext[nextents - ext_diff],
+			0, ext_diff * sizeof(xfs_bmbt_rec_t));
+	} else {
+		memset(&ifp->if_u2.if_inline_ext[idx], 0,
+			ext_diff * sizeof(xfs_bmbt_rec_t));
+	}
+}
+
+/*
+ * This removes ext_diff extents from a linear (direct) extent list,
+ * beginning at extent index idx. If the extents are being removed
+ * from the end of the list (ie. truncate) then we just need to re-
+ * allocate the list to remove the extra space. Otherwise, if the
+ * extents are being removed from the middle of the existing extent
+ * entries, then we first need to move the extent records beginning
+ * at idx + ext_diff up in the list to overwrite the records being
+ * removed, then remove the extra space via kmem_realloc.
+ */
+void
+xfs_iext_remove_direct(
+	xfs_ifork_t	*ifp,		/* inode fork pointer */
+	xfs_extnum_t	idx,		/* index to begin removing exts */
+	int		ext_diff)	/* number of extents to remove */
+{
+	xfs_extnum_t	nextents;	/* number of extents in file */
+	int		new_size;	/* size of extents after removal */
+
+	ASSERT(!(ifp->if_flags & XFS_IFEXTIREC));
+	new_size = ifp->if_bytes -
+		(ext_diff * sizeof(xfs_bmbt_rec_t));
+	nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
+
+	if (new_size == 0) {
+		xfs_iext_destroy(ifp);
+		return;
+	}
+	/* Move extents up in the list (if needed) */
+	if (idx + ext_diff < nextents) {
+		memmove(&ifp->if_u1.if_extents[idx],
+			&ifp->if_u1.if_extents[idx + ext_diff],
+			(nextents - (idx + ext_diff)) *
+			 sizeof(xfs_bmbt_rec_t));
+	}
+	memset(&ifp->if_u1.if_extents[nextents - ext_diff],
+		0, ext_diff * sizeof(xfs_bmbt_rec_t));
+	/*
+	 * Reallocate the direct extent list. If the extents
+	 * will fit inside the inode then xfs_iext_realloc_direct
+	 * will switch from direct to inline extent allocation
+	 * mode for us.
+	 */
+	xfs_iext_realloc_direct(ifp, new_size);
+	ifp->if_bytes = new_size;
+}
+
+/*
+ * This is called when incore extents are being removed from the
+ * indirection array and the extents being removed span multiple extent
+ * buffers. The idx parameter contains the file extent index where we
+ * want to begin removing extents, and the count parameter contains
+ * how many extents need to be removed.
+ *
+ *    |-------|   |-------|
+ *    | nex1  |   |       |    nex1 - number of extents before idx
+ *    |-------|   | count |
+ *    |       |   |       |    count - number of extents being removed at idx
+ *    | count |   |-------|
+ *    |       |   | nex2  |    nex2 - number of extents after idx + count
+ *    |-------|   |-------|
+ */
+void
+xfs_iext_remove_indirect(
+	xfs_ifork_t	*ifp,		/* inode fork pointer */
+	xfs_extnum_t	idx,		/* index to begin removing extents */
+	int		count)		/* number of extents to remove */
+{
+	xfs_ext_irec_t	*erp;		/* indirection array pointer */
+	int		erp_idx = 0;	/* indirection array index */
+	xfs_extnum_t	ext_cnt;	/* extents left to remove */
+	xfs_extnum_t	ext_diff;	/* extents to remove in current list */
+	xfs_extnum_t	nex1;		/* number of extents before idx */
+	xfs_extnum_t	nex2;		/* extents after idx + count */
+	int		page_idx = idx;	/* index in target extent list */
+
+	ASSERT(ifp->if_flags & XFS_IFEXTIREC);
+	erp = xfs_iext_idx_to_irec(ifp,  &page_idx, &erp_idx, 0);
+	ASSERT(erp != NULL);
+	nex1 = page_idx;
+	ext_cnt = count;
+	while (ext_cnt) {
+		nex2 = MAX((erp->er_extcount - (nex1 + ext_cnt)), 0);
+		ext_diff = MIN(ext_cnt, (erp->er_extcount - nex1));
+		/*
+		 * Check for deletion of entire list;
+		 * xfs_iext_irec_remove() updates extent offsets.
+		 */
+		if (ext_diff == erp->er_extcount) {
+			xfs_iext_irec_remove(ifp, erp_idx);
+			ext_cnt -= ext_diff;
+			nex1 = 0;
+			if (ext_cnt) {
+				ASSERT(erp_idx < ifp->if_real_bytes /
+					XFS_IEXT_BUFSZ);
+				erp = &ifp->if_u1.if_ext_irec[erp_idx];
+				nex1 = 0;
+				continue;
+			} else {
+				break;
+			}
+		}
+		/* Move extents up (if needed) */
+		if (nex2) {
+			memmove(&erp->er_extbuf[nex1],
+				&erp->er_extbuf[nex1 + ext_diff],
+				nex2 * sizeof(xfs_bmbt_rec_t));
+		}
+		/* Zero out rest of page */
+		memset(&erp->er_extbuf[nex1 + nex2], 0, (XFS_IEXT_BUFSZ -
+			((nex1 + nex2) * sizeof(xfs_bmbt_rec_t))));
+		/* Update remaining counters */
+		erp->er_extcount -= ext_diff;
+		xfs_iext_irec_update_extoffs(ifp, erp_idx + 1, -ext_diff);
+		ext_cnt -= ext_diff;
+		nex1 = 0;
+		erp_idx++;
+		erp++;
+	}
+	ifp->if_bytes -= count * sizeof(xfs_bmbt_rec_t);
+	xfs_iext_irec_compact(ifp);
+}
+
+/*
+ * Create, destroy, or resize a linear (direct) block of extents.
+ */
+void
+xfs_iext_realloc_direct(
+	xfs_ifork_t	*ifp,		/* inode fork pointer */
+	int		new_size)	/* new size of extents */
+{
+	int		rnew_size;	/* real new size of extents */
+
+	rnew_size = new_size;
+
+	ASSERT(!(ifp->if_flags & XFS_IFEXTIREC) ||
+		((new_size >= 0) && (new_size <= XFS_IEXT_BUFSZ) &&
+		 (new_size != ifp->if_real_bytes)));
+
+	/* Free extent records */
+	if (new_size == 0) {
+		xfs_iext_destroy(ifp);
+	}
+	/* Resize direct extent list and zero any new bytes */
+	else if (ifp->if_real_bytes) {
+		/* Check if extents will fit inside the inode */
+		if (new_size <= XFS_INLINE_EXTS * sizeof(xfs_bmbt_rec_t)) {
+			xfs_iext_direct_to_inline(ifp, new_size /
+				(uint)sizeof(xfs_bmbt_rec_t));
+			ifp->if_bytes = new_size;
+			return;
+		}
+		if (!is_power_of_2(new_size)){
+			rnew_size = roundup_pow_of_two(new_size);
+		}
+		if (rnew_size != ifp->if_real_bytes) {
+			ifp->if_u1.if_extents =
+				kmem_realloc(ifp->if_u1.if_extents,
+						rnew_size,
+						ifp->if_real_bytes, KM_NOFS);
+		}
+		if (rnew_size > ifp->if_real_bytes) {
+			memset(&ifp->if_u1.if_extents[ifp->if_bytes /
+				(uint)sizeof(xfs_bmbt_rec_t)], 0,
+				rnew_size - ifp->if_real_bytes);
+		}
+	}
+	/*
+	 * Switch from the inline extent buffer to a direct
+	 * extent list. Be sure to include the inline extent
+	 * bytes in new_size.
+	 */
+	else {
+		new_size += ifp->if_bytes;
+		if (!is_power_of_2(new_size)) {
+			rnew_size = roundup_pow_of_two(new_size);
+		}
+		xfs_iext_inline_to_direct(ifp, rnew_size);
+	}
+	ifp->if_real_bytes = rnew_size;
+	ifp->if_bytes = new_size;
+}
+
+/*
+ * Switch from linear (direct) extent records to inline buffer.
+ */
+void
+xfs_iext_direct_to_inline(
+	xfs_ifork_t	*ifp,		/* inode fork pointer */
+	xfs_extnum_t	nextents)	/* number of extents in file */
+{
+	ASSERT(ifp->if_flags & XFS_IFEXTENTS);
+	ASSERT(nextents <= XFS_INLINE_EXTS);
+	/*
+	 * The inline buffer was zeroed when we switched
+	 * from inline to direct extent allocation mode,
+	 * so we don't need to clear it here.
+	 */
+	memcpy(ifp->if_u2.if_inline_ext, ifp->if_u1.if_extents,
+		nextents * sizeof(xfs_bmbt_rec_t));
+	kmem_free(ifp->if_u1.if_extents);
+	ifp->if_u1.if_extents = ifp->if_u2.if_inline_ext;
+	ifp->if_real_bytes = 0;
+}
+
+/*
+ * Switch from inline buffer to linear (direct) extent records.
+ * new_size should already be rounded up to the next power of 2
+ * by the caller (when appropriate), so use new_size as it is.
+ * However, since new_size may be rounded up, we can't update
+ * if_bytes here. It is the caller's responsibility to update
+ * if_bytes upon return.
+ */
+void
+xfs_iext_inline_to_direct(
+	xfs_ifork_t	*ifp,		/* inode fork pointer */
+	int		new_size)	/* number of extents in file */
+{
+	ifp->if_u1.if_extents = kmem_alloc(new_size, KM_NOFS);
+	memset(ifp->if_u1.if_extents, 0, new_size);
+	if (ifp->if_bytes) {
+		memcpy(ifp->if_u1.if_extents, ifp->if_u2.if_inline_ext,
+			ifp->if_bytes);
+		memset(ifp->if_u2.if_inline_ext, 0, XFS_INLINE_EXTS *
+			sizeof(xfs_bmbt_rec_t));
+	}
+	ifp->if_real_bytes = new_size;
+}
+
+/*
+ * Resize an extent indirection array to new_size bytes.
+ */
+STATIC void
+xfs_iext_realloc_indirect(
+	xfs_ifork_t	*ifp,		/* inode fork pointer */
+	int		new_size)	/* new indirection array size */
+{
+	int		nlists;		/* number of irec's (ex lists) */
+	int		size;		/* current indirection array size */
+
+	ASSERT(ifp->if_flags & XFS_IFEXTIREC);
+	nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ;
+	size = nlists * sizeof(xfs_ext_irec_t);
+	ASSERT(ifp->if_real_bytes);
+	ASSERT((new_size >= 0) && (new_size != size));
+	if (new_size == 0) {
+		xfs_iext_destroy(ifp);
+	} else {
+		ifp->if_u1.if_ext_irec = (xfs_ext_irec_t *)
+			kmem_realloc(ifp->if_u1.if_ext_irec,
+				new_size, size, KM_NOFS);
+	}
+}
+
+/*
+ * Switch from indirection array to linear (direct) extent allocations.
+ */
+STATIC void
+xfs_iext_indirect_to_direct(
+	 xfs_ifork_t	*ifp)		/* inode fork pointer */
+{
+	xfs_bmbt_rec_host_t *ep;	/* extent record pointer */
+	xfs_extnum_t	nextents;	/* number of extents in file */
+	int		size;		/* size of file extents */
+
+	ASSERT(ifp->if_flags & XFS_IFEXTIREC);
+	nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
+	ASSERT(nextents <= XFS_LINEAR_EXTS);
+	size = nextents * sizeof(xfs_bmbt_rec_t);
+
+	xfs_iext_irec_compact_pages(ifp);
+	ASSERT(ifp->if_real_bytes == XFS_IEXT_BUFSZ);
+
+	ep = ifp->if_u1.if_ext_irec->er_extbuf;
+	kmem_free(ifp->if_u1.if_ext_irec);
+	ifp->if_flags &= ~XFS_IFEXTIREC;
+	ifp->if_u1.if_extents = ep;
+	ifp->if_bytes = size;
+	if (nextents < XFS_LINEAR_EXTS) {
+		xfs_iext_realloc_direct(ifp, size);
+	}
+}
+
+/*
+ * Free incore file extents.
+ */
+void
+xfs_iext_destroy(
+	xfs_ifork_t	*ifp)		/* inode fork pointer */
+{
+	if (ifp->if_flags & XFS_IFEXTIREC) {
+		int	erp_idx;
+		int	nlists;
+
+		nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ;
+		for (erp_idx = nlists - 1; erp_idx >= 0 ; erp_idx--) {
+			xfs_iext_irec_remove(ifp, erp_idx);
+		}
+		ifp->if_flags &= ~XFS_IFEXTIREC;
+	} else if (ifp->if_real_bytes) {
+		kmem_free(ifp->if_u1.if_extents);
+	} else if (ifp->if_bytes) {
+		memset(ifp->if_u2.if_inline_ext, 0, XFS_INLINE_EXTS *
+			sizeof(xfs_bmbt_rec_t));
+	}
+	ifp->if_u1.if_extents = NULL;
+	ifp->if_real_bytes = 0;
+	ifp->if_bytes = 0;
+}
+
+/*
+ * Return a pointer to the extent record for file system block bno.
+ */
+xfs_bmbt_rec_host_t *			/* pointer to found extent record */
+xfs_iext_bno_to_ext(
+	xfs_ifork_t	*ifp,		/* inode fork pointer */
+	xfs_fileoff_t	bno,		/* block number to search for */
+	xfs_extnum_t	*idxp)		/* index of target extent */
+{
+	xfs_bmbt_rec_host_t *base;	/* pointer to first extent */
+	xfs_filblks_t	blockcount = 0;	/* number of blocks in extent */
+	xfs_bmbt_rec_host_t *ep = NULL;	/* pointer to target extent */
+	xfs_ext_irec_t	*erp = NULL;	/* indirection array pointer */
+	int		high;		/* upper boundary in search */
+	xfs_extnum_t	idx = 0;	/* index of target extent */
+	int		low;		/* lower boundary in search */
+	xfs_extnum_t	nextents;	/* number of file extents */
+	xfs_fileoff_t	startoff = 0;	/* start offset of extent */
+
+	nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
+	if (nextents == 0) {
+		*idxp = 0;
+		return NULL;
+	}
+	low = 0;
+	if (ifp->if_flags & XFS_IFEXTIREC) {
+		/* Find target extent list */
+		int	erp_idx = 0;
+		erp = xfs_iext_bno_to_irec(ifp, bno, &erp_idx);
+		base = erp->er_extbuf;
+		high = erp->er_extcount - 1;
+	} else {
+		base = ifp->if_u1.if_extents;
+		high = nextents - 1;
+	}
+	/* Binary search extent records */
+	while (low <= high) {
+		idx = (low + high) >> 1;
+		ep = base + idx;
+		startoff = xfs_bmbt_get_startoff(ep);
+		blockcount = xfs_bmbt_get_blockcount(ep);
+		if (bno < startoff) {
+			high = idx - 1;
+		} else if (bno >= startoff + blockcount) {
+			low = idx + 1;
+		} else {
+			/* Convert back to file-based extent index */
+			if (ifp->if_flags & XFS_IFEXTIREC) {
+				idx += erp->er_extoff;
+			}
+			*idxp = idx;
+			return ep;
+		}
+	}
+	/* Convert back to file-based extent index */
+	if (ifp->if_flags & XFS_IFEXTIREC) {
+		idx += erp->er_extoff;
+	}
+	if (bno >= startoff + blockcount) {
+		if (++idx == nextents) {
+			ep = NULL;
+		} else {
+			ep = xfs_iext_get_ext(ifp, idx);
+		}
+	}
+	*idxp = idx;
+	return ep;
+}
+
+/*
+ * Return a pointer to the indirection array entry containing the
+ * extent record for filesystem block bno. Store the index of the
+ * target irec in *erp_idxp.
+ */
+xfs_ext_irec_t *			/* pointer to found extent record */
+xfs_iext_bno_to_irec(
+	xfs_ifork_t	*ifp,		/* inode fork pointer */
+	xfs_fileoff_t	bno,		/* block number to search for */
+	int		*erp_idxp)	/* irec index of target ext list */
+{
+	xfs_ext_irec_t	*erp = NULL;	/* indirection array pointer */
+	xfs_ext_irec_t	*erp_next;	/* next indirection array entry */
+	int		erp_idx;	/* indirection array index */
+	int		nlists;		/* number of extent irec's (lists) */
+	int		high;		/* binary search upper limit */
+	int		low;		/* binary search lower limit */
+
+	ASSERT(ifp->if_flags & XFS_IFEXTIREC);
+	nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ;
+	erp_idx = 0;
+	low = 0;
+	high = nlists - 1;
+	while (low <= high) {
+		erp_idx = (low + high) >> 1;
+		erp = &ifp->if_u1.if_ext_irec[erp_idx];
+		erp_next = erp_idx < nlists - 1 ? erp + 1 : NULL;
+		if (bno < xfs_bmbt_get_startoff(erp->er_extbuf)) {
+			high = erp_idx - 1;
+		} else if (erp_next && bno >=
+			   xfs_bmbt_get_startoff(erp_next->er_extbuf)) {
+			low = erp_idx + 1;
+		} else {
+			break;
+		}
+	}
+	*erp_idxp = erp_idx;
+	return erp;
+}
+
+/*
+ * Return a pointer to the indirection array entry containing the
+ * extent record at file extent index *idxp. Store the index of the
+ * target irec in *erp_idxp and store the page index of the target
+ * extent record in *idxp.
+ */
+xfs_ext_irec_t *
+xfs_iext_idx_to_irec(
+	xfs_ifork_t	*ifp,		/* inode fork pointer */
+	xfs_extnum_t	*idxp,		/* extent index (file -> page) */
+	int		*erp_idxp,	/* pointer to target irec */
+	int		realloc)	/* new bytes were just added */
+{
+	xfs_ext_irec_t	*prev;		/* pointer to previous irec */
+	xfs_ext_irec_t	*erp = NULL;	/* pointer to current irec */
+	int		erp_idx;	/* indirection array index */
+	int		nlists;		/* number of irec's (ex lists) */
+	int		high;		/* binary search upper limit */
+	int		low;		/* binary search lower limit */
+	xfs_extnum_t	page_idx = *idxp; /* extent index in target list */
+
+	ASSERT(ifp->if_flags & XFS_IFEXTIREC);
+	ASSERT(page_idx >= 0);
+	ASSERT(page_idx <= ifp->if_bytes / sizeof(xfs_bmbt_rec_t));
+	ASSERT(page_idx < ifp->if_bytes / sizeof(xfs_bmbt_rec_t) || realloc);
+
+	nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ;
+	erp_idx = 0;
+	low = 0;
+	high = nlists - 1;
+
+	/* Binary search extent irec's */
+	while (low <= high) {
+		erp_idx = (low + high) >> 1;
+		erp = &ifp->if_u1.if_ext_irec[erp_idx];
+		prev = erp_idx > 0 ? erp - 1 : NULL;
+		if (page_idx < erp->er_extoff || (page_idx == erp->er_extoff &&
+		     realloc && prev && prev->er_extcount < XFS_LINEAR_EXTS)) {
+			high = erp_idx - 1;
+		} else if (page_idx > erp->er_extoff + erp->er_extcount ||
+			   (page_idx == erp->er_extoff + erp->er_extcount &&
+			    !realloc)) {
+			low = erp_idx + 1;
+		} else if (page_idx == erp->er_extoff + erp->er_extcount &&
+			   erp->er_extcount == XFS_LINEAR_EXTS) {
+			ASSERT(realloc);
+			page_idx = 0;
+			erp_idx++;
+			erp = erp_idx < nlists ? erp + 1 : NULL;
+			break;
+		} else {
+			page_idx -= erp->er_extoff;
+			break;
+		}
+	}
+	*idxp = page_idx;
+	*erp_idxp = erp_idx;
+	return(erp);
+}
+
+/*
+ * Allocate and initialize an indirection array once the space needed
+ * for incore extents increases above XFS_IEXT_BUFSZ.
+ */
+void
+xfs_iext_irec_init(
+	xfs_ifork_t	*ifp)		/* inode fork pointer */
+{
+	xfs_ext_irec_t	*erp;		/* indirection array pointer */
+	xfs_extnum_t	nextents;	/* number of extents in file */
+
+	ASSERT(!(ifp->if_flags & XFS_IFEXTIREC));
+	nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
+	ASSERT(nextents <= XFS_LINEAR_EXTS);
+
+	erp = kmem_alloc(sizeof(xfs_ext_irec_t), KM_NOFS);
+
+	if (nextents == 0) {
+		ifp->if_u1.if_extents = kmem_alloc(XFS_IEXT_BUFSZ, KM_NOFS);
+	} else if (!ifp->if_real_bytes) {
+		xfs_iext_inline_to_direct(ifp, XFS_IEXT_BUFSZ);
+	} else if (ifp->if_real_bytes < XFS_IEXT_BUFSZ) {
+		xfs_iext_realloc_direct(ifp, XFS_IEXT_BUFSZ);
+	}
+	erp->er_extbuf = ifp->if_u1.if_extents;
+	erp->er_extcount = nextents;
+	erp->er_extoff = 0;
+
+	ifp->if_flags |= XFS_IFEXTIREC;
+	ifp->if_real_bytes = XFS_IEXT_BUFSZ;
+	ifp->if_bytes = nextents * sizeof(xfs_bmbt_rec_t);
+	ifp->if_u1.if_ext_irec = erp;
+
+	return;
+}
+
+/*
+ * Allocate and initialize a new entry in the indirection array.
+ */
+xfs_ext_irec_t *
+xfs_iext_irec_new(
+	xfs_ifork_t	*ifp,		/* inode fork pointer */
+	int		erp_idx)	/* index for new irec */
+{
+	xfs_ext_irec_t	*erp;		/* indirection array pointer */
+	int		i;		/* loop counter */
+	int		nlists;		/* number of irec's (ex lists) */
+
+	ASSERT(ifp->if_flags & XFS_IFEXTIREC);
+	nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ;
+
+	/* Resize indirection array */
+	xfs_iext_realloc_indirect(ifp, ++nlists *
+				  sizeof(xfs_ext_irec_t));
+	/*
+	 * Move records down in the array so the
+	 * new page can use erp_idx.
+	 */
+	erp = ifp->if_u1.if_ext_irec;
+	for (i = nlists - 1; i > erp_idx; i--) {
+		memmove(&erp[i], &erp[i-1], sizeof(xfs_ext_irec_t));
+	}
+	ASSERT(i == erp_idx);
+
+	/* Initialize new extent record */
+	erp = ifp->if_u1.if_ext_irec;
+	erp[erp_idx].er_extbuf = kmem_alloc(XFS_IEXT_BUFSZ, KM_NOFS);
+	ifp->if_real_bytes = nlists * XFS_IEXT_BUFSZ;
+	memset(erp[erp_idx].er_extbuf, 0, XFS_IEXT_BUFSZ);
+	erp[erp_idx].er_extcount = 0;
+	erp[erp_idx].er_extoff = erp_idx > 0 ?
+		erp[erp_idx-1].er_extoff + erp[erp_idx-1].er_extcount : 0;
+	return (&erp[erp_idx]);
+}
+
+/*
+ * Remove a record from the indirection array.
+ */
+void
+xfs_iext_irec_remove(
+	xfs_ifork_t	*ifp,		/* inode fork pointer */
+	int		erp_idx)	/* irec index to remove */
+{
+	xfs_ext_irec_t	*erp;		/* indirection array pointer */
+	int		i;		/* loop counter */
+	int		nlists;		/* number of irec's (ex lists) */
+
+	ASSERT(ifp->if_flags & XFS_IFEXTIREC);
+	nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ;
+	erp = &ifp->if_u1.if_ext_irec[erp_idx];
+	if (erp->er_extbuf) {
+		xfs_iext_irec_update_extoffs(ifp, erp_idx + 1,
+			-erp->er_extcount);
+		kmem_free(erp->er_extbuf);
+	}
+	/* Compact extent records */
+	erp = ifp->if_u1.if_ext_irec;
+	for (i = erp_idx; i < nlists - 1; i++) {
+		memmove(&erp[i], &erp[i+1], sizeof(xfs_ext_irec_t));
+	}
+	/*
+	 * Manually free the last extent record from the indirection
+	 * array.  A call to xfs_iext_realloc_indirect() with a size
+	 * of zero would result in a call to xfs_iext_destroy() which
+	 * would in turn call this function again, creating a nasty
+	 * infinite loop.
+	 */
+	if (--nlists) {
+		xfs_iext_realloc_indirect(ifp,
+			nlists * sizeof(xfs_ext_irec_t));
+	} else {
+		kmem_free(ifp->if_u1.if_ext_irec);
+	}
+	ifp->if_real_bytes = nlists * XFS_IEXT_BUFSZ;
+}
+
+/*
+ * This is called to clean up large amounts of unused memory allocated
+ * by the indirection array.  Before compacting anything though, verify
+ * that the indirection array is still needed and switch back to the
+ * linear extent list (or even the inline buffer) if possible.  The
+ * compaction policy is as follows:
+ *
+ *    Full Compaction: Extents fit into a single page (or inline buffer)
+ * Partial Compaction: Extents occupy less than 50% of allocated space
+ *      No Compaction: Extents occupy at least 50% of allocated space
+ */
+void
+xfs_iext_irec_compact(
+	xfs_ifork_t	*ifp)		/* inode fork pointer */
+{
+	xfs_extnum_t	nextents;	/* number of extents in file */
+	int		nlists;		/* number of irec's (ex lists) */
+
+	ASSERT(ifp->if_flags & XFS_IFEXTIREC);
+	nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ;
+	nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
+
+	if (nextents == 0) {
+		xfs_iext_destroy(ifp);
+	} else if (nextents <= XFS_INLINE_EXTS) {
+		xfs_iext_indirect_to_direct(ifp);
+		xfs_iext_direct_to_inline(ifp, nextents);
+	} else if (nextents <= XFS_LINEAR_EXTS) {
+		xfs_iext_indirect_to_direct(ifp);
+	} else if (nextents < (nlists * XFS_LINEAR_EXTS) >> 1) {
+		xfs_iext_irec_compact_pages(ifp);
+	}
+}
+
+/*
+ * Combine extents from neighboring extent pages.
+ */
+void
+xfs_iext_irec_compact_pages(
+	xfs_ifork_t	*ifp)		/* inode fork pointer */
+{
+	xfs_ext_irec_t	*erp, *erp_next;/* pointers to irec entries */
+	int		erp_idx = 0;	/* indirection array index */
+	int		nlists;		/* number of irec's (ex lists) */
+
+	ASSERT(ifp->if_flags & XFS_IFEXTIREC);
+	nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ;
+	while (erp_idx < nlists - 1) {
+		erp = &ifp->if_u1.if_ext_irec[erp_idx];
+		erp_next = erp + 1;
+		if (erp_next->er_extcount <=
+		    (XFS_LINEAR_EXTS - erp->er_extcount)) {
+			memcpy(&erp->er_extbuf[erp->er_extcount],
+				erp_next->er_extbuf, erp_next->er_extcount *
+				sizeof(xfs_bmbt_rec_t));
+			erp->er_extcount += erp_next->er_extcount;
+			/*
+			 * Free page before removing extent record
+			 * so er_extoffs don't get modified in
+			 * xfs_iext_irec_remove.
+			 */
+			kmem_free(erp_next->er_extbuf);
+			erp_next->er_extbuf = NULL;
+			xfs_iext_irec_remove(ifp, erp_idx + 1);
+			nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ;
+		} else {
+			erp_idx++;
+		}
+	}
+}
+
+/*
+ * This is called to update the er_extoff field in the indirection
+ * array when extents have been added or removed from one of the
+ * extent lists. erp_idx contains the irec index to begin updating
+ * at and ext_diff contains the number of extents that were added
+ * or removed.
+ */
+void
+xfs_iext_irec_update_extoffs(
+	xfs_ifork_t	*ifp,		/* inode fork pointer */
+	int		erp_idx,	/* irec index to update */
+	int		ext_diff)	/* number of new extents */
+{
+	int		i;		/* loop counter */
+	int		nlists;		/* number of irec's (ex lists */
+
+	ASSERT(ifp->if_flags & XFS_IFEXTIREC);
+	nlists = ifp->if_real_bytes / XFS_IEXT_BUFSZ;
+	for (i = erp_idx; i < nlists; i++) {
+		ifp->if_u1.if_ext_irec[i].er_extoff += ext_diff;
+	}
+}
diff --git a/fs/xfs/xfs_inode.h b/fs/xfs/xfs_inode.h
new file mode 100644
index 00000000..28b35964
--- /dev/null
+++ b/fs/xfs/xfs_inode.h
@@ -0,0 +1,600 @@
+/*
+ * Copyright (c) 2000-2003,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef	__XFS_INODE_H__
+#define	__XFS_INODE_H__
+
+struct posix_acl;
+struct xfs_dinode;
+struct xfs_inode;
+
+/*
+ * Fork identifiers.
+ */
+#define	XFS_DATA_FORK	0
+#define	XFS_ATTR_FORK	1
+
+/*
+ * The following xfs_ext_irec_t struct introduces a second (top) level
+ * to the in-core extent allocation scheme. These structs are allocated
+ * in a contiguous block, creating an indirection array where each entry
+ * (irec) contains a pointer to a buffer of in-core extent records which
+ * it manages. Each extent buffer is 4k in size, since 4k is the system
+ * page size on Linux i386 and systems with larger page sizes don't seem
+ * to gain much, if anything, by using their native page size as the
+ * extent buffer size. Also, using 4k extent buffers everywhere provides
+ * a consistent interface for CXFS across different platforms.
+ *
+ * There is currently no limit on the number of irec's (extent lists)
+ * allowed, so heavily fragmented files may require an indirection array
+ * which spans multiple system pages of memory. The number of extents
+ * which would require this amount of contiguous memory is very large
+ * and should not cause problems in the foreseeable future. However,
+ * if the memory needed for the contiguous array ever becomes a problem,
+ * it is possible that a third level of indirection may be required.
+ */
+typedef struct xfs_ext_irec {
+	xfs_bmbt_rec_host_t *er_extbuf;	/* block of extent records */
+	xfs_extnum_t	er_extoff;	/* extent offset in file */
+	xfs_extnum_t	er_extcount;	/* number of extents in page/block */
+} xfs_ext_irec_t;
+
+/*
+ * File incore extent information, present for each of data & attr forks.
+ */
+#define	XFS_IEXT_BUFSZ		4096
+#define	XFS_LINEAR_EXTS		(XFS_IEXT_BUFSZ / (uint)sizeof(xfs_bmbt_rec_t))
+#define	XFS_INLINE_EXTS		2
+#define	XFS_INLINE_DATA		32
+typedef struct xfs_ifork {
+	int			if_bytes;	/* bytes in if_u1 */
+	int			if_real_bytes;	/* bytes allocated in if_u1 */
+	struct xfs_btree_block	*if_broot;	/* file's incore btree root */
+	short			if_broot_bytes;	/* bytes allocated for root */
+	unsigned char		if_flags;	/* per-fork flags */
+	unsigned char		if_ext_max;	/* max # of extent records */
+	union {
+		xfs_bmbt_rec_host_t *if_extents;/* linear map file exts */
+		xfs_ext_irec_t	*if_ext_irec;	/* irec map file exts */
+		char		*if_data;	/* inline file data */
+	} if_u1;
+	union {
+		xfs_bmbt_rec_host_t if_inline_ext[XFS_INLINE_EXTS];
+						/* very small file extents */
+		char		if_inline_data[XFS_INLINE_DATA];
+						/* very small file data */
+		xfs_dev_t	if_rdev;	/* dev number if special */
+		uuid_t		if_uuid;	/* mount point value */
+	} if_u2;
+} xfs_ifork_t;
+
+/*
+ * Inode location information.  Stored in the inode and passed to
+ * xfs_imap_to_bp() to get a buffer and dinode for a given inode.
+ */
+struct xfs_imap {
+	xfs_daddr_t	im_blkno;	/* starting BB of inode chunk */
+	ushort		im_len;		/* length in BBs of inode chunk */
+	ushort		im_boffset;	/* inode offset in block in bytes */
+};
+
+/*
+ * This is the xfs in-core inode structure.
+ * Most of the on-disk inode is embedded in the i_d field.
+ *
+ * The extent pointers/inline file space, however, are managed
+ * separately.  The memory for this information is pointed to by
+ * the if_u1 unions depending on the type of the data.
+ * This is used to linearize the array of extents for fast in-core
+ * access.  This is used until the file's number of extents
+ * surpasses XFS_MAX_INCORE_EXTENTS, at which point all extent pointers
+ * are accessed through the buffer cache.
+ *
+ * Other state kept in the in-core inode is used for identification,
+ * locking, transactional updating, etc of the inode.
+ *
+ * Generally, we do not want to hold the i_rlock while holding the
+ * i_ilock. Hierarchy is i_iolock followed by i_rlock.
+ *
+ * xfs_iptr_t contains all the inode fields up to and including the
+ * i_mnext and i_mprev fields, it is used as a marker in the inode
+ * chain off the mount structure by xfs_sync calls.
+ */
+
+typedef struct xfs_ictimestamp {
+	__int32_t	t_sec;		/* timestamp seconds */
+	__int32_t	t_nsec;		/* timestamp nanoseconds */
+} xfs_ictimestamp_t;
+
+/*
+ * NOTE:  This structure must be kept identical to struct xfs_dinode
+ * 	  in xfs_dinode.h except for the endianness annotations.
+ */
+typedef struct xfs_icdinode {
+	__uint16_t	di_magic;	/* inode magic # = XFS_DINODE_MAGIC */
+	__uint16_t	di_mode;	/* mode and type of file */
+	__int8_t	di_version;	/* inode version */
+	__int8_t	di_format;	/* format of di_c data */
+	__uint16_t	di_onlink;	/* old number of links to file */
+	__uint32_t	di_uid;		/* owner's user id */
+	__uint32_t	di_gid;		/* owner's group id */
+	__uint32_t	di_nlink;	/* number of links to file */
+	__uint16_t	di_projid_lo;	/* lower part of owner's project id */
+	__uint16_t	di_projid_hi;	/* higher part of owner's project id */
+	__uint8_t	di_pad[6];	/* unused, zeroed space */
+	__uint16_t	di_flushiter;	/* incremented on flush */
+	xfs_ictimestamp_t di_atime;	/* time last accessed */
+	xfs_ictimestamp_t di_mtime;	/* time last modified */
+	xfs_ictimestamp_t di_ctime;	/* time created/inode modified */
+	xfs_fsize_t	di_size;	/* number of bytes in file */
+	xfs_drfsbno_t	di_nblocks;	/* # of direct & btree blocks used */
+	xfs_extlen_t	di_extsize;	/* basic/minimum extent size for file */
+	xfs_extnum_t	di_nextents;	/* number of extents in data fork */
+	xfs_aextnum_t	di_anextents;	/* number of extents in attribute fork*/
+	__uint8_t	di_forkoff;	/* attr fork offs, <<3 for 64b align */
+	__int8_t	di_aformat;	/* format of attr fork's data */
+	__uint32_t	di_dmevmask;	/* DMIG event mask */
+	__uint16_t	di_dmstate;	/* DMIG state info */
+	__uint16_t	di_flags;	/* random flags, XFS_DIFLAG_... */
+	__uint32_t	di_gen;		/* generation number */
+} xfs_icdinode_t;
+
+/*
+ * Flags for xfs_ichgtime().
+ */
+#define	XFS_ICHGTIME_MOD	0x1	/* data fork modification timestamp */
+#define	XFS_ICHGTIME_CHG	0x2	/* inode field change timestamp */
+
+/*
+ * Per-fork incore inode flags.
+ */
+#define	XFS_IFINLINE	0x01	/* Inline data is read in */
+#define	XFS_IFEXTENTS	0x02	/* All extent pointers are read in */
+#define	XFS_IFBROOT	0x04	/* i_broot points to the bmap b-tree root */
+#define	XFS_IFEXTIREC	0x08	/* Indirection array of extent blocks */
+
+/*
+ * Fork handling.
+ */
+
+#define XFS_IFORK_Q(ip)			((ip)->i_d.di_forkoff != 0)
+#define XFS_IFORK_BOFF(ip)		((int)((ip)->i_d.di_forkoff << 3))
+
+#define XFS_IFORK_PTR(ip,w)		\
+	((w) == XFS_DATA_FORK ? \
+		&(ip)->i_df : \
+		(ip)->i_afp)
+#define XFS_IFORK_DSIZE(ip) \
+	(XFS_IFORK_Q(ip) ? \
+		XFS_IFORK_BOFF(ip) : \
+		XFS_LITINO((ip)->i_mount))
+#define XFS_IFORK_ASIZE(ip) \
+	(XFS_IFORK_Q(ip) ? \
+		XFS_LITINO((ip)->i_mount) - XFS_IFORK_BOFF(ip) : \
+		0)
+#define XFS_IFORK_SIZE(ip,w) \
+	((w) == XFS_DATA_FORK ? \
+		XFS_IFORK_DSIZE(ip) : \
+		XFS_IFORK_ASIZE(ip))
+#define XFS_IFORK_FORMAT(ip,w) \
+	((w) == XFS_DATA_FORK ? \
+		(ip)->i_d.di_format : \
+		(ip)->i_d.di_aformat)
+#define XFS_IFORK_FMT_SET(ip,w,n) \
+	((w) == XFS_DATA_FORK ? \
+		((ip)->i_d.di_format = (n)) : \
+		((ip)->i_d.di_aformat = (n)))
+#define XFS_IFORK_NEXTENTS(ip,w) \
+	((w) == XFS_DATA_FORK ? \
+		(ip)->i_d.di_nextents : \
+		(ip)->i_d.di_anextents)
+#define XFS_IFORK_NEXT_SET(ip,w,n) \
+	((w) == XFS_DATA_FORK ? \
+		((ip)->i_d.di_nextents = (n)) : \
+		((ip)->i_d.di_anextents = (n)))
+
+
+
+#ifdef __KERNEL__
+
+struct bhv_desc;
+struct xfs_buf;
+struct xfs_bmap_free;
+struct xfs_bmbt_irec;
+struct xfs_inode_log_item;
+struct xfs_mount;
+struct xfs_trans;
+struct xfs_dquot;
+
+typedef struct dm_attrs_s {
+	__uint32_t	da_dmevmask;	/* DMIG event mask */
+	__uint16_t	da_dmstate;	/* DMIG state info */
+	__uint16_t	da_pad;		/* DMIG extra padding */
+} dm_attrs_t;
+
+typedef struct xfs_inode {
+	/* Inode linking and identification information. */
+	struct xfs_mount	*i_mount;	/* fs mount struct ptr */
+	struct xfs_dquot	*i_udquot;	/* user dquot */
+	struct xfs_dquot	*i_gdquot;	/* group dquot */
+
+	/* Inode location stuff */
+	xfs_ino_t		i_ino;		/* inode number (agno/agino)*/
+	struct xfs_imap		i_imap;		/* location for xfs_imap() */
+
+	/* Extent information. */
+	xfs_ifork_t		*i_afp;		/* attribute fork pointer */
+	xfs_ifork_t		i_df;		/* data fork */
+
+	/* Transaction and locking information. */
+	struct xfs_trans	*i_transp;	/* ptr to owning transaction*/
+	struct xfs_inode_log_item *i_itemp;	/* logging information */
+	mrlock_t		i_lock;		/* inode lock */
+	mrlock_t		i_iolock;	/* inode IO lock */
+	struct completion	i_flush;	/* inode flush completion q */
+	atomic_t		i_pincount;	/* inode pin count */
+	wait_queue_head_t	i_ipin_wait;	/* inode pinning wait queue */
+	spinlock_t		i_flags_lock;	/* inode i_flags lock */
+	/* Miscellaneous state. */
+	unsigned short		i_flags;	/* see defined flags below */
+	unsigned char		i_update_core;	/* timestamps/size is dirty */
+	unsigned int		i_delayed_blks;	/* count of delay alloc blks */
+
+	xfs_icdinode_t		i_d;		/* most of ondisk inode */
+
+	xfs_fsize_t		i_size;		/* in-memory size */
+	xfs_fsize_t		i_new_size;	/* size when write completes */
+	atomic_t		i_iocount;	/* outstanding I/O count */
+
+	/* VFS inode */
+	struct inode		i_vnode;	/* embedded VFS inode */
+} xfs_inode_t;
+
+#define XFS_ISIZE(ip)	(((ip)->i_d.di_mode & S_IFMT) == S_IFREG) ? \
+				(ip)->i_size : (ip)->i_d.di_size;
+
+/* Convert from vfs inode to xfs inode */
+static inline struct xfs_inode *XFS_I(struct inode *inode)
+{
+	return container_of(inode, struct xfs_inode, i_vnode);
+}
+
+/* convert from xfs inode to vfs inode */
+static inline struct inode *VFS_I(struct xfs_inode *ip)
+{
+	return &ip->i_vnode;
+}
+
+/*
+ * i_flags helper functions
+ */
+static inline void
+__xfs_iflags_set(xfs_inode_t *ip, unsigned short flags)
+{
+	ip->i_flags |= flags;
+}
+
+static inline void
+xfs_iflags_set(xfs_inode_t *ip, unsigned short flags)
+{
+	spin_lock(&ip->i_flags_lock);
+	__xfs_iflags_set(ip, flags);
+	spin_unlock(&ip->i_flags_lock);
+}
+
+static inline void
+xfs_iflags_clear(xfs_inode_t *ip, unsigned short flags)
+{
+	spin_lock(&ip->i_flags_lock);
+	ip->i_flags &= ~flags;
+	spin_unlock(&ip->i_flags_lock);
+}
+
+static inline int
+__xfs_iflags_test(xfs_inode_t *ip, unsigned short flags)
+{
+	return (ip->i_flags & flags);
+}
+
+static inline int
+xfs_iflags_test(xfs_inode_t *ip, unsigned short flags)
+{
+	int ret;
+	spin_lock(&ip->i_flags_lock);
+	ret = __xfs_iflags_test(ip, flags);
+	spin_unlock(&ip->i_flags_lock);
+	return ret;
+}
+
+static inline int
+xfs_iflags_test_and_clear(xfs_inode_t *ip, unsigned short flags)
+{
+	int ret;
+
+	spin_lock(&ip->i_flags_lock);
+	ret = ip->i_flags & flags;
+	if (ret)
+		ip->i_flags &= ~flags;
+	spin_unlock(&ip->i_flags_lock);
+	return ret;
+}
+
+/*
+ * Project quota id helpers (previously projid was 16bit only
+ * and using two 16bit values to hold new 32bit projid was chosen
+ * to retain compatibility with "old" filesystems).
+ */
+static inline prid_t
+xfs_get_projid(struct xfs_inode *ip)
+{
+	return (prid_t)ip->i_d.di_projid_hi << 16 | ip->i_d.di_projid_lo;
+}
+
+static inline void
+xfs_set_projid(struct xfs_inode *ip,
+		prid_t projid)
+{
+	ip->i_d.di_projid_hi = (__uint16_t) (projid >> 16);
+	ip->i_d.di_projid_lo = (__uint16_t) (projid & 0xffff);
+}
+
+/*
+ * Manage the i_flush queue embedded in the inode.  This completion
+ * queue synchronizes processes attempting to flush the in-core
+ * inode back to disk.
+ */
+static inline void xfs_iflock(xfs_inode_t *ip)
+{
+	wait_for_completion(&ip->i_flush);
+}
+
+static inline int xfs_iflock_nowait(xfs_inode_t *ip)
+{
+	return try_wait_for_completion(&ip->i_flush);
+}
+
+static inline void xfs_ifunlock(xfs_inode_t *ip)
+{
+	complete(&ip->i_flush);
+}
+
+/*
+ * In-core inode flags.
+ */
+#define XFS_IRECLAIM		0x0001  /* started reclaiming this inode */
+#define XFS_ISTALE		0x0002	/* inode has been staled */
+#define XFS_IRECLAIMABLE	0x0004	/* inode can be reclaimed */
+#define XFS_INEW		0x0008	/* inode has just been allocated */
+#define XFS_IFILESTREAM		0x0010	/* inode is in a filestream directory */
+#define XFS_ITRUNCATED		0x0020	/* truncated down so flush-on-close */
+#define XFS_IDIRTY_RELEASE	0x0040	/* dirty release already seen */
+
+/*
+ * Per-lifetime flags need to be reset when re-using a reclaimable inode during
+ * inode lookup. Thi prevents unintended behaviour on the new inode from
+ * ocurring.
+ */
+#define XFS_IRECLAIM_RESET_FLAGS	\
+	(XFS_IRECLAIMABLE | XFS_IRECLAIM | \
+	 XFS_IDIRTY_RELEASE | XFS_ITRUNCATED | \
+	 XFS_IFILESTREAM);
+
+/*
+ * Flags for inode locking.
+ * Bit ranges:	1<<1  - 1<<16-1 -- iolock/ilock modes (bitfield)
+ *		1<<16 - 1<<32-1 -- lockdep annotation (integers)
+ */
+#define	XFS_IOLOCK_EXCL		(1<<0)
+#define	XFS_IOLOCK_SHARED	(1<<1)
+#define	XFS_ILOCK_EXCL		(1<<2)
+#define	XFS_ILOCK_SHARED	(1<<3)
+#define	XFS_IUNLOCK_NONOTIFY	(1<<4)
+
+#define XFS_LOCK_MASK		(XFS_IOLOCK_EXCL | XFS_IOLOCK_SHARED \
+				| XFS_ILOCK_EXCL | XFS_ILOCK_SHARED)
+
+#define XFS_LOCK_FLAGS \
+	{ XFS_IOLOCK_EXCL,	"IOLOCK_EXCL" }, \
+	{ XFS_IOLOCK_SHARED,	"IOLOCK_SHARED" }, \
+	{ XFS_ILOCK_EXCL,	"ILOCK_EXCL" }, \
+	{ XFS_ILOCK_SHARED,	"ILOCK_SHARED" }, \
+	{ XFS_IUNLOCK_NONOTIFY,	"IUNLOCK_NONOTIFY" }
+
+
+/*
+ * Flags for lockdep annotations.
+ *
+ * XFS_LOCK_PARENT - for directory operations that require locking a
+ * parent directory inode and a child entry inode.  The parent gets locked
+ * with this flag so it gets a lockdep subclass of 1 and the child entry
+ * lock will have a lockdep subclass of 0.
+ *
+ * XFS_LOCK_RTBITMAP/XFS_LOCK_RTSUM - the realtime device bitmap and summary
+ * inodes do not participate in the normal lock order, and thus have their
+ * own subclasses.
+ *
+ * XFS_LOCK_INUMORDER - for locking several inodes at the some time
+ * with xfs_lock_inodes().  This flag is used as the starting subclass
+ * and each subsequent lock acquired will increment the subclass by one.
+ * So the first lock acquired will have a lockdep subclass of 4, the
+ * second lock will have a lockdep subclass of 5, and so on. It is
+ * the responsibility of the class builder to shift this to the correct
+ * portion of the lock_mode lockdep mask.
+ */
+#define XFS_LOCK_PARENT		1
+#define XFS_LOCK_RTBITMAP	2
+#define XFS_LOCK_RTSUM		3
+#define XFS_LOCK_INUMORDER	4
+
+#define XFS_IOLOCK_SHIFT	16
+#define	XFS_IOLOCK_PARENT	(XFS_LOCK_PARENT << XFS_IOLOCK_SHIFT)
+
+#define XFS_ILOCK_SHIFT		24
+#define	XFS_ILOCK_PARENT	(XFS_LOCK_PARENT << XFS_ILOCK_SHIFT)
+#define	XFS_ILOCK_RTBITMAP	(XFS_LOCK_RTBITMAP << XFS_ILOCK_SHIFT)
+#define	XFS_ILOCK_RTSUM		(XFS_LOCK_RTSUM << XFS_ILOCK_SHIFT)
+
+#define XFS_IOLOCK_DEP_MASK	0x00ff0000
+#define XFS_ILOCK_DEP_MASK	0xff000000
+#define XFS_LOCK_DEP_MASK	(XFS_IOLOCK_DEP_MASK | XFS_ILOCK_DEP_MASK)
+
+#define XFS_IOLOCK_DEP(flags)	(((flags) & XFS_IOLOCK_DEP_MASK) >> XFS_IOLOCK_SHIFT)
+#define XFS_ILOCK_DEP(flags)	(((flags) & XFS_ILOCK_DEP_MASK) >> XFS_ILOCK_SHIFT)
+
+extern struct lock_class_key xfs_iolock_reclaimable;
+
+/*
+ * Flags for xfs_itruncate_start().
+ */
+#define	XFS_ITRUNC_DEFINITE	0x1
+#define	XFS_ITRUNC_MAYBE	0x2
+
+#define XFS_ITRUNC_FLAGS \
+	{ XFS_ITRUNC_DEFINITE,	"DEFINITE" }, \
+	{ XFS_ITRUNC_MAYBE,	"MAYBE" }
+
+/*
+ * For multiple groups support: if S_ISGID bit is set in the parent
+ * directory, group of new file is set to that of the parent, and
+ * new subdirectory gets S_ISGID bit from parent.
+ */
+#define XFS_INHERIT_GID(pip)	\
+	(((pip)->i_mount->m_flags & XFS_MOUNT_GRPID) || \
+	 ((pip)->i_d.di_mode & S_ISGID))
+
+/*
+ * xfs_iget.c prototypes.
+ */
+int		xfs_iget(struct xfs_mount *, struct xfs_trans *, xfs_ino_t,
+			 uint, uint, xfs_inode_t **);
+void		xfs_ilock(xfs_inode_t *, uint);
+int		xfs_ilock_nowait(xfs_inode_t *, uint);
+void		xfs_iunlock(xfs_inode_t *, uint);
+void		xfs_ilock_demote(xfs_inode_t *, uint);
+int		xfs_isilocked(xfs_inode_t *, uint);
+uint		xfs_ilock_map_shared(xfs_inode_t *);
+void		xfs_iunlock_map_shared(xfs_inode_t *, uint);
+void		xfs_inode_free(struct xfs_inode *ip);
+
+/*
+ * xfs_inode.c prototypes.
+ */
+int		xfs_ialloc(struct xfs_trans *, xfs_inode_t *, mode_t,
+			   xfs_nlink_t, xfs_dev_t, prid_t, int,
+			   struct xfs_buf **, boolean_t *, xfs_inode_t **);
+
+uint		xfs_ip2xflags(struct xfs_inode *);
+uint		xfs_dic2xflags(struct xfs_dinode *);
+int		xfs_ifree(struct xfs_trans *, xfs_inode_t *,
+			   struct xfs_bmap_free *);
+int		xfs_itruncate_start(xfs_inode_t *, uint, xfs_fsize_t);
+int		xfs_itruncate_finish(struct xfs_trans **, xfs_inode_t *,
+				     xfs_fsize_t, int, int);
+int		xfs_iunlink(struct xfs_trans *, xfs_inode_t *);
+
+void		xfs_iext_realloc(xfs_inode_t *, int, int);
+void		xfs_iunpin_wait(xfs_inode_t *);
+int		xfs_iflush(xfs_inode_t *, uint);
+void		xfs_promote_inode(struct xfs_inode *);
+void		xfs_lock_inodes(xfs_inode_t **, int, uint);
+void		xfs_lock_two_inodes(xfs_inode_t *, xfs_inode_t *, uint);
+
+void		xfs_synchronize_times(xfs_inode_t *);
+void		xfs_mark_inode_dirty(xfs_inode_t *);
+void		xfs_mark_inode_dirty_sync(xfs_inode_t *);
+
+#define IHOLD(ip) \
+do { \
+	ASSERT(atomic_read(&VFS_I(ip)->i_count) > 0) ; \
+	ihold(VFS_I(ip)); \
+	trace_xfs_ihold(ip, _THIS_IP_); \
+} while (0)
+
+#define IRELE(ip) \
+do { \
+	trace_xfs_irele(ip, _THIS_IP_); \
+	iput(VFS_I(ip)); \
+} while (0)
+
+#endif /* __KERNEL__ */
+
+/*
+ * Flags for xfs_iget()
+ */
+#define XFS_IGET_CREATE		0x1
+#define XFS_IGET_UNTRUSTED	0x2
+
+int		xfs_inotobp(struct xfs_mount *, struct xfs_trans *,
+			    xfs_ino_t, struct xfs_dinode **,
+			    struct xfs_buf **, int *, uint);
+int		xfs_itobp(struct xfs_mount *, struct xfs_trans *,
+			  struct xfs_inode *, struct xfs_dinode **,
+			  struct xfs_buf **, uint);
+int		xfs_iread(struct xfs_mount *, struct xfs_trans *,
+			  struct xfs_inode *, uint);
+void		xfs_dinode_to_disk(struct xfs_dinode *,
+				   struct xfs_icdinode *);
+void		xfs_idestroy_fork(struct xfs_inode *, int);
+void		xfs_idata_realloc(struct xfs_inode *, int, int);
+void		xfs_iroot_realloc(struct xfs_inode *, int, int);
+int		xfs_iread_extents(struct xfs_trans *, struct xfs_inode *, int);
+int		xfs_iextents_copy(struct xfs_inode *, xfs_bmbt_rec_t *, int);
+
+xfs_bmbt_rec_host_t *xfs_iext_get_ext(xfs_ifork_t *, xfs_extnum_t);
+void		xfs_iext_insert(xfs_inode_t *, xfs_extnum_t, xfs_extnum_t,
+				xfs_bmbt_irec_t *, int);
+void		xfs_iext_add(xfs_ifork_t *, xfs_extnum_t, int);
+void		xfs_iext_add_indirect_multi(xfs_ifork_t *, int, xfs_extnum_t, int);
+void		xfs_iext_remove(xfs_inode_t *, xfs_extnum_t, int, int);
+void		xfs_iext_remove_inline(xfs_ifork_t *, xfs_extnum_t, int);
+void		xfs_iext_remove_direct(xfs_ifork_t *, xfs_extnum_t, int);
+void		xfs_iext_remove_indirect(xfs_ifork_t *, xfs_extnum_t, int);
+void		xfs_iext_realloc_direct(xfs_ifork_t *, int);
+void		xfs_iext_direct_to_inline(xfs_ifork_t *, xfs_extnum_t);
+void		xfs_iext_inline_to_direct(xfs_ifork_t *, int);
+void		xfs_iext_destroy(xfs_ifork_t *);
+xfs_bmbt_rec_host_t *xfs_iext_bno_to_ext(xfs_ifork_t *, xfs_fileoff_t, int *);
+xfs_ext_irec_t	*xfs_iext_bno_to_irec(xfs_ifork_t *, xfs_fileoff_t, int *);
+xfs_ext_irec_t	*xfs_iext_idx_to_irec(xfs_ifork_t *, xfs_extnum_t *, int *, int);
+void		xfs_iext_irec_init(xfs_ifork_t *);
+xfs_ext_irec_t *xfs_iext_irec_new(xfs_ifork_t *, int);
+void		xfs_iext_irec_remove(xfs_ifork_t *, int);
+void		xfs_iext_irec_compact(xfs_ifork_t *);
+void		xfs_iext_irec_compact_pages(xfs_ifork_t *);
+void		xfs_iext_irec_compact_full(xfs_ifork_t *);
+void		xfs_iext_irec_update_extoffs(xfs_ifork_t *, int, int);
+
+#define xfs_ipincount(ip)	((unsigned int) atomic_read(&ip->i_pincount))
+
+#ifdef DEBUG
+void		xfs_isize_check(struct xfs_mount *, struct xfs_inode *,
+				xfs_fsize_t);
+#else	/* DEBUG */
+#define xfs_isize_check(mp, ip, isize)
+#endif	/* DEBUG */
+
+#if defined(DEBUG)
+void		xfs_inobp_check(struct xfs_mount *, struct xfs_buf *);
+#else
+#define	xfs_inobp_check(mp, bp)
+#endif /* DEBUG */
+
+extern struct kmem_zone	*xfs_ifork_zone;
+extern struct kmem_zone	*xfs_inode_zone;
+extern struct kmem_zone	*xfs_ili_zone;
+
+#endif	/* __XFS_INODE_H__ */
diff --git a/fs/xfs/xfs_inode_item.c b/fs/xfs/xfs_inode_item.c
new file mode 100644
index 00000000..391044c6
--- /dev/null
+++ b/fs/xfs/xfs_inode_item.c
@@ -0,0 +1,1060 @@
+/*
+ * Copyright (c) 2000-2002,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_mount.h"
+#include "xfs_trans_priv.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_inode_item.h"
+#include "xfs_error.h"
+#include "xfs_trace.h"
+
+
+kmem_zone_t	*xfs_ili_zone;		/* inode log item zone */
+
+static inline struct xfs_inode_log_item *INODE_ITEM(struct xfs_log_item *lip)
+{
+	return container_of(lip, struct xfs_inode_log_item, ili_item);
+}
+
+
+/*
+ * This returns the number of iovecs needed to log the given inode item.
+ *
+ * We need one iovec for the inode log format structure, one for the
+ * inode core, and possibly one for the inode data/extents/b-tree root
+ * and one for the inode attribute data/extents/b-tree root.
+ */
+STATIC uint
+xfs_inode_item_size(
+	struct xfs_log_item	*lip)
+{
+	struct xfs_inode_log_item *iip = INODE_ITEM(lip);
+	struct xfs_inode	*ip = iip->ili_inode;
+	uint			nvecs = 2;
+
+	/*
+	 * Only log the data/extents/b-tree root if there is something
+	 * left to log.
+	 */
+	iip->ili_format.ilf_fields |= XFS_ILOG_CORE;
+
+	switch (ip->i_d.di_format) {
+	case XFS_DINODE_FMT_EXTENTS:
+		iip->ili_format.ilf_fields &=
+			~(XFS_ILOG_DDATA | XFS_ILOG_DBROOT |
+			  XFS_ILOG_DEV | XFS_ILOG_UUID);
+		if ((iip->ili_format.ilf_fields & XFS_ILOG_DEXT) &&
+		    (ip->i_d.di_nextents > 0) &&
+		    (ip->i_df.if_bytes > 0)) {
+			ASSERT(ip->i_df.if_u1.if_extents != NULL);
+			nvecs++;
+		} else {
+			iip->ili_format.ilf_fields &= ~XFS_ILOG_DEXT;
+		}
+		break;
+
+	case XFS_DINODE_FMT_BTREE:
+		ASSERT(ip->i_df.if_ext_max ==
+		       XFS_IFORK_DSIZE(ip) / (uint)sizeof(xfs_bmbt_rec_t));
+		iip->ili_format.ilf_fields &=
+			~(XFS_ILOG_DDATA | XFS_ILOG_DEXT |
+			  XFS_ILOG_DEV | XFS_ILOG_UUID);
+		if ((iip->ili_format.ilf_fields & XFS_ILOG_DBROOT) &&
+		    (ip->i_df.if_broot_bytes > 0)) {
+			ASSERT(ip->i_df.if_broot != NULL);
+			nvecs++;
+		} else {
+			ASSERT(!(iip->ili_format.ilf_fields &
+				 XFS_ILOG_DBROOT));
+#ifdef XFS_TRANS_DEBUG
+			if (iip->ili_root_size > 0) {
+				ASSERT(iip->ili_root_size ==
+				       ip->i_df.if_broot_bytes);
+				ASSERT(memcmp(iip->ili_orig_root,
+					    ip->i_df.if_broot,
+					    iip->ili_root_size) == 0);
+			} else {
+				ASSERT(ip->i_df.if_broot_bytes == 0);
+			}
+#endif
+			iip->ili_format.ilf_fields &= ~XFS_ILOG_DBROOT;
+		}
+		break;
+
+	case XFS_DINODE_FMT_LOCAL:
+		iip->ili_format.ilf_fields &=
+			~(XFS_ILOG_DEXT | XFS_ILOG_DBROOT |
+			  XFS_ILOG_DEV | XFS_ILOG_UUID);
+		if ((iip->ili_format.ilf_fields & XFS_ILOG_DDATA) &&
+		    (ip->i_df.if_bytes > 0)) {
+			ASSERT(ip->i_df.if_u1.if_data != NULL);
+			ASSERT(ip->i_d.di_size > 0);
+			nvecs++;
+		} else {
+			iip->ili_format.ilf_fields &= ~XFS_ILOG_DDATA;
+		}
+		break;
+
+	case XFS_DINODE_FMT_DEV:
+		iip->ili_format.ilf_fields &=
+			~(XFS_ILOG_DDATA | XFS_ILOG_DBROOT |
+			  XFS_ILOG_DEXT | XFS_ILOG_UUID);
+		break;
+
+	case XFS_DINODE_FMT_UUID:
+		iip->ili_format.ilf_fields &=
+			~(XFS_ILOG_DDATA | XFS_ILOG_DBROOT |
+			  XFS_ILOG_DEXT | XFS_ILOG_DEV);
+		break;
+
+	default:
+		ASSERT(0);
+		break;
+	}
+
+	/*
+	 * If there are no attributes associated with this file,
+	 * then there cannot be anything more to log.
+	 * Clear all attribute-related log flags.
+	 */
+	if (!XFS_IFORK_Q(ip)) {
+		iip->ili_format.ilf_fields &=
+			~(XFS_ILOG_ADATA | XFS_ILOG_ABROOT | XFS_ILOG_AEXT);
+		return nvecs;
+	}
+
+	/*
+	 * Log any necessary attribute data.
+	 */
+	switch (ip->i_d.di_aformat) {
+	case XFS_DINODE_FMT_EXTENTS:
+		iip->ili_format.ilf_fields &=
+			~(XFS_ILOG_ADATA | XFS_ILOG_ABROOT);
+		if ((iip->ili_format.ilf_fields & XFS_ILOG_AEXT) &&
+		    (ip->i_d.di_anextents > 0) &&
+		    (ip->i_afp->if_bytes > 0)) {
+			ASSERT(ip->i_afp->if_u1.if_extents != NULL);
+			nvecs++;
+		} else {
+			iip->ili_format.ilf_fields &= ~XFS_ILOG_AEXT;
+		}
+		break;
+
+	case XFS_DINODE_FMT_BTREE:
+		iip->ili_format.ilf_fields &=
+			~(XFS_ILOG_ADATA | XFS_ILOG_AEXT);
+		if ((iip->ili_format.ilf_fields & XFS_ILOG_ABROOT) &&
+		    (ip->i_afp->if_broot_bytes > 0)) {
+			ASSERT(ip->i_afp->if_broot != NULL);
+			nvecs++;
+		} else {
+			iip->ili_format.ilf_fields &= ~XFS_ILOG_ABROOT;
+		}
+		break;
+
+	case XFS_DINODE_FMT_LOCAL:
+		iip->ili_format.ilf_fields &=
+			~(XFS_ILOG_AEXT | XFS_ILOG_ABROOT);
+		if ((iip->ili_format.ilf_fields & XFS_ILOG_ADATA) &&
+		    (ip->i_afp->if_bytes > 0)) {
+			ASSERT(ip->i_afp->if_u1.if_data != NULL);
+			nvecs++;
+		} else {
+			iip->ili_format.ilf_fields &= ~XFS_ILOG_ADATA;
+		}
+		break;
+
+	default:
+		ASSERT(0);
+		break;
+	}
+
+	return nvecs;
+}
+
+/*
+ * xfs_inode_item_format_extents - convert in-core extents to on-disk form
+ *
+ * For either the data or attr fork in extent format, we need to endian convert
+ * the in-core extent as we place them into the on-disk inode. In this case, we
+ * need to do this conversion before we write the extents into the log. Because
+ * we don't have the disk inode to write into here, we allocate a buffer and
+ * format the extents into it via xfs_iextents_copy(). We free the buffer in
+ * the unlock routine after the copy for the log has been made.
+ *
+ * In the case of the data fork, the in-core and on-disk fork sizes can be
+ * different due to delayed allocation extents. We only log on-disk extents
+ * here, so always use the physical fork size to determine the size of the
+ * buffer we need to allocate.
+ */
+STATIC void
+xfs_inode_item_format_extents(
+	struct xfs_inode	*ip,
+	struct xfs_log_iovec	*vecp,
+	int			whichfork,
+	int			type)
+{
+	xfs_bmbt_rec_t		*ext_buffer;
+
+	ext_buffer = kmem_alloc(XFS_IFORK_SIZE(ip, whichfork), KM_SLEEP);
+	if (whichfork == XFS_DATA_FORK)
+		ip->i_itemp->ili_extents_buf = ext_buffer;
+	else
+		ip->i_itemp->ili_aextents_buf = ext_buffer;
+
+	vecp->i_addr = ext_buffer;
+	vecp->i_len = xfs_iextents_copy(ip, ext_buffer, whichfork);
+	vecp->i_type = type;
+}
+
+/*
+ * This is called to fill in the vector of log iovecs for the
+ * given inode log item.  It fills the first item with an inode
+ * log format structure, the second with the on-disk inode structure,
+ * and a possible third and/or fourth with the inode data/extents/b-tree
+ * root and inode attributes data/extents/b-tree root.
+ */
+STATIC void
+xfs_inode_item_format(
+	struct xfs_log_item	*lip,
+	struct xfs_log_iovec	*vecp)
+{
+	struct xfs_inode_log_item *iip = INODE_ITEM(lip);
+	struct xfs_inode	*ip = iip->ili_inode;
+	uint			nvecs;
+	size_t			data_bytes;
+	xfs_mount_t		*mp;
+
+	vecp->i_addr = &iip->ili_format;
+	vecp->i_len  = sizeof(xfs_inode_log_format_t);
+	vecp->i_type = XLOG_REG_TYPE_IFORMAT;
+	vecp++;
+	nvecs	     = 1;
+
+	/*
+	 * Clear i_update_core if the timestamps (or any other
+	 * non-transactional modification) need flushing/logging
+	 * and we're about to log them with the rest of the core.
+	 *
+	 * This is the same logic as xfs_iflush() but this code can't
+	 * run at the same time as xfs_iflush because we're in commit
+	 * processing here and so we have the inode lock held in
+	 * exclusive mode.  Although it doesn't really matter
+	 * for the timestamps if both routines were to grab the
+	 * timestamps or not.  That would be ok.
+	 *
+	 * We clear i_update_core before copying out the data.
+	 * This is for coordination with our timestamp updates
+	 * that don't hold the inode lock. They will always
+	 * update the timestamps BEFORE setting i_update_core,
+	 * so if we clear i_update_core after they set it we
+	 * are guaranteed to see their updates to the timestamps
+	 * either here.  Likewise, if they set it after we clear it
+	 * here, we'll see it either on the next commit of this
+	 * inode or the next time the inode gets flushed via
+	 * xfs_iflush().  This depends on strongly ordered memory
+	 * semantics, but we have that.  We use the SYNCHRONIZE
+	 * macro to make sure that the compiler does not reorder
+	 * the i_update_core access below the data copy below.
+	 */
+	if (ip->i_update_core)  {
+		ip->i_update_core = 0;
+		SYNCHRONIZE();
+	}
+
+	/*
+	 * Make sure to get the latest timestamps from the Linux inode.
+	 */
+	xfs_synchronize_times(ip);
+
+	vecp->i_addr = &ip->i_d;
+	vecp->i_len  = sizeof(struct xfs_icdinode);
+	vecp->i_type = XLOG_REG_TYPE_ICORE;
+	vecp++;
+	nvecs++;
+	iip->ili_format.ilf_fields |= XFS_ILOG_CORE;
+
+	/*
+	 * If this is really an old format inode, then we need to
+	 * log it as such.  This means that we have to copy the link
+	 * count from the new field to the old.  We don't have to worry
+	 * about the new fields, because nothing trusts them as long as
+	 * the old inode version number is there.  If the superblock already
+	 * has a new version number, then we don't bother converting back.
+	 */
+	mp = ip->i_mount;
+	ASSERT(ip->i_d.di_version == 1 || xfs_sb_version_hasnlink(&mp->m_sb));
+	if (ip->i_d.di_version == 1) {
+		if (!xfs_sb_version_hasnlink(&mp->m_sb)) {
+			/*
+			 * Convert it back.
+			 */
+			ASSERT(ip->i_d.di_nlink <= XFS_MAXLINK_1);
+			ip->i_d.di_onlink = ip->i_d.di_nlink;
+		} else {
+			/*
+			 * The superblock version has already been bumped,
+			 * so just make the conversion to the new inode
+			 * format permanent.
+			 */
+			ip->i_d.di_version = 2;
+			ip->i_d.di_onlink = 0;
+			memset(&(ip->i_d.di_pad[0]), 0, sizeof(ip->i_d.di_pad));
+		}
+	}
+
+	switch (ip->i_d.di_format) {
+	case XFS_DINODE_FMT_EXTENTS:
+		ASSERT(!(iip->ili_format.ilf_fields &
+			 (XFS_ILOG_DDATA | XFS_ILOG_DBROOT |
+			  XFS_ILOG_DEV | XFS_ILOG_UUID)));
+		if (iip->ili_format.ilf_fields & XFS_ILOG_DEXT) {
+			ASSERT(ip->i_df.if_bytes > 0);
+			ASSERT(ip->i_df.if_u1.if_extents != NULL);
+			ASSERT(ip->i_d.di_nextents > 0);
+			ASSERT(iip->ili_extents_buf == NULL);
+			ASSERT((ip->i_df.if_bytes /
+				(uint)sizeof(xfs_bmbt_rec_t)) > 0);
+#ifdef XFS_NATIVE_HOST
+                       if (ip->i_d.di_nextents == ip->i_df.if_bytes /
+                                               (uint)sizeof(xfs_bmbt_rec_t)) {
+				/*
+				 * There are no delayed allocation
+				 * extents, so just point to the
+				 * real extents array.
+				 */
+				vecp->i_addr = ip->i_df.if_u1.if_extents;
+				vecp->i_len = ip->i_df.if_bytes;
+				vecp->i_type = XLOG_REG_TYPE_IEXT;
+			} else
+#endif
+			{
+				xfs_inode_item_format_extents(ip, vecp,
+					XFS_DATA_FORK, XLOG_REG_TYPE_IEXT);
+			}
+			ASSERT(vecp->i_len <= ip->i_df.if_bytes);
+			iip->ili_format.ilf_dsize = vecp->i_len;
+			vecp++;
+			nvecs++;
+		}
+		break;
+
+	case XFS_DINODE_FMT_BTREE:
+		ASSERT(!(iip->ili_format.ilf_fields &
+			 (XFS_ILOG_DDATA | XFS_ILOG_DEXT |
+			  XFS_ILOG_DEV | XFS_ILOG_UUID)));
+		if (iip->ili_format.ilf_fields & XFS_ILOG_DBROOT) {
+			ASSERT(ip->i_df.if_broot_bytes > 0);
+			ASSERT(ip->i_df.if_broot != NULL);
+			vecp->i_addr = ip->i_df.if_broot;
+			vecp->i_len = ip->i_df.if_broot_bytes;
+			vecp->i_type = XLOG_REG_TYPE_IBROOT;
+			vecp++;
+			nvecs++;
+			iip->ili_format.ilf_dsize = ip->i_df.if_broot_bytes;
+		}
+		break;
+
+	case XFS_DINODE_FMT_LOCAL:
+		ASSERT(!(iip->ili_format.ilf_fields &
+			 (XFS_ILOG_DBROOT | XFS_ILOG_DEXT |
+			  XFS_ILOG_DEV | XFS_ILOG_UUID)));
+		if (iip->ili_format.ilf_fields & XFS_ILOG_DDATA) {
+			ASSERT(ip->i_df.if_bytes > 0);
+			ASSERT(ip->i_df.if_u1.if_data != NULL);
+			ASSERT(ip->i_d.di_size > 0);
+
+			vecp->i_addr = ip->i_df.if_u1.if_data;
+			/*
+			 * Round i_bytes up to a word boundary.
+			 * The underlying memory is guaranteed to
+			 * to be there by xfs_idata_realloc().
+			 */
+			data_bytes = roundup(ip->i_df.if_bytes, 4);
+			ASSERT((ip->i_df.if_real_bytes == 0) ||
+			       (ip->i_df.if_real_bytes == data_bytes));
+			vecp->i_len = (int)data_bytes;
+			vecp->i_type = XLOG_REG_TYPE_ILOCAL;
+			vecp++;
+			nvecs++;
+			iip->ili_format.ilf_dsize = (unsigned)data_bytes;
+		}
+		break;
+
+	case XFS_DINODE_FMT_DEV:
+		ASSERT(!(iip->ili_format.ilf_fields &
+			 (XFS_ILOG_DBROOT | XFS_ILOG_DEXT |
+			  XFS_ILOG_DDATA | XFS_ILOG_UUID)));
+		if (iip->ili_format.ilf_fields & XFS_ILOG_DEV) {
+			iip->ili_format.ilf_u.ilfu_rdev =
+				ip->i_df.if_u2.if_rdev;
+		}
+		break;
+
+	case XFS_DINODE_FMT_UUID:
+		ASSERT(!(iip->ili_format.ilf_fields &
+			 (XFS_ILOG_DBROOT | XFS_ILOG_DEXT |
+			  XFS_ILOG_DDATA | XFS_ILOG_DEV)));
+		if (iip->ili_format.ilf_fields & XFS_ILOG_UUID) {
+			iip->ili_format.ilf_u.ilfu_uuid =
+				ip->i_df.if_u2.if_uuid;
+		}
+		break;
+
+	default:
+		ASSERT(0);
+		break;
+	}
+
+	/*
+	 * If there are no attributes associated with the file,
+	 * then we're done.
+	 * Assert that no attribute-related log flags are set.
+	 */
+	if (!XFS_IFORK_Q(ip)) {
+		ASSERT(nvecs == lip->li_desc->lid_size);
+		iip->ili_format.ilf_size = nvecs;
+		ASSERT(!(iip->ili_format.ilf_fields &
+			 (XFS_ILOG_ADATA | XFS_ILOG_ABROOT | XFS_ILOG_AEXT)));
+		return;
+	}
+
+	switch (ip->i_d.di_aformat) {
+	case XFS_DINODE_FMT_EXTENTS:
+		ASSERT(!(iip->ili_format.ilf_fields &
+			 (XFS_ILOG_ADATA | XFS_ILOG_ABROOT)));
+		if (iip->ili_format.ilf_fields & XFS_ILOG_AEXT) {
+#ifdef DEBUG
+			int nrecs = ip->i_afp->if_bytes /
+				(uint)sizeof(xfs_bmbt_rec_t);
+			ASSERT(nrecs > 0);
+			ASSERT(nrecs == ip->i_d.di_anextents);
+			ASSERT(ip->i_afp->if_bytes > 0);
+			ASSERT(ip->i_afp->if_u1.if_extents != NULL);
+			ASSERT(ip->i_d.di_anextents > 0);
+#endif
+#ifdef XFS_NATIVE_HOST
+			/*
+			 * There are not delayed allocation extents
+			 * for attributes, so just point at the array.
+			 */
+			vecp->i_addr = ip->i_afp->if_u1.if_extents;
+			vecp->i_len = ip->i_afp->if_bytes;
+			vecp->i_type = XLOG_REG_TYPE_IATTR_EXT;
+#else
+			ASSERT(iip->ili_aextents_buf == NULL);
+			xfs_inode_item_format_extents(ip, vecp,
+					XFS_ATTR_FORK, XLOG_REG_TYPE_IATTR_EXT);
+#endif
+			iip->ili_format.ilf_asize = vecp->i_len;
+			vecp++;
+			nvecs++;
+		}
+		break;
+
+	case XFS_DINODE_FMT_BTREE:
+		ASSERT(!(iip->ili_format.ilf_fields &
+			 (XFS_ILOG_ADATA | XFS_ILOG_AEXT)));
+		if (iip->ili_format.ilf_fields & XFS_ILOG_ABROOT) {
+			ASSERT(ip->i_afp->if_broot_bytes > 0);
+			ASSERT(ip->i_afp->if_broot != NULL);
+			vecp->i_addr = ip->i_afp->if_broot;
+			vecp->i_len = ip->i_afp->if_broot_bytes;
+			vecp->i_type = XLOG_REG_TYPE_IATTR_BROOT;
+			vecp++;
+			nvecs++;
+			iip->ili_format.ilf_asize = ip->i_afp->if_broot_bytes;
+		}
+		break;
+
+	case XFS_DINODE_FMT_LOCAL:
+		ASSERT(!(iip->ili_format.ilf_fields &
+			 (XFS_ILOG_ABROOT | XFS_ILOG_AEXT)));
+		if (iip->ili_format.ilf_fields & XFS_ILOG_ADATA) {
+			ASSERT(ip->i_afp->if_bytes > 0);
+			ASSERT(ip->i_afp->if_u1.if_data != NULL);
+
+			vecp->i_addr = ip->i_afp->if_u1.if_data;
+			/*
+			 * Round i_bytes up to a word boundary.
+			 * The underlying memory is guaranteed to
+			 * to be there by xfs_idata_realloc().
+			 */
+			data_bytes = roundup(ip->i_afp->if_bytes, 4);
+			ASSERT((ip->i_afp->if_real_bytes == 0) ||
+			       (ip->i_afp->if_real_bytes == data_bytes));
+			vecp->i_len = (int)data_bytes;
+			vecp->i_type = XLOG_REG_TYPE_IATTR_LOCAL;
+			vecp++;
+			nvecs++;
+			iip->ili_format.ilf_asize = (unsigned)data_bytes;
+		}
+		break;
+
+	default:
+		ASSERT(0);
+		break;
+	}
+
+	ASSERT(nvecs == lip->li_desc->lid_size);
+	iip->ili_format.ilf_size = nvecs;
+}
+
+
+/*
+ * This is called to pin the inode associated with the inode log
+ * item in memory so it cannot be written out.
+ */
+STATIC void
+xfs_inode_item_pin(
+	struct xfs_log_item	*lip)
+{
+	struct xfs_inode	*ip = INODE_ITEM(lip)->ili_inode;
+
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
+
+	trace_xfs_inode_pin(ip, _RET_IP_);
+	atomic_inc(&ip->i_pincount);
+}
+
+
+/*
+ * This is called to unpin the inode associated with the inode log
+ * item which was previously pinned with a call to xfs_inode_item_pin().
+ *
+ * Also wake up anyone in xfs_iunpin_wait() if the count goes to 0.
+ */
+STATIC void
+xfs_inode_item_unpin(
+	struct xfs_log_item	*lip,
+	int			remove)
+{
+	struct xfs_inode	*ip = INODE_ITEM(lip)->ili_inode;
+
+	trace_xfs_inode_unpin(ip, _RET_IP_);
+	ASSERT(atomic_read(&ip->i_pincount) > 0);
+	if (atomic_dec_and_test(&ip->i_pincount))
+		wake_up(&ip->i_ipin_wait);
+}
+
+/*
+ * This is called to attempt to lock the inode associated with this
+ * inode log item, in preparation for the push routine which does the actual
+ * iflush.  Don't sleep on the inode lock or the flush lock.
+ *
+ * If the flush lock is already held, indicating that the inode has
+ * been or is in the process of being flushed, then (ideally) we'd like to
+ * see if the inode's buffer is still incore, and if so give it a nudge.
+ * We delay doing so until the pushbuf routine, though, to avoid holding
+ * the AIL lock across a call to the blackhole which is the buffer cache.
+ * Also we don't want to sleep in any device strategy routines, which can happen
+ * if we do the subsequent bawrite in here.
+ */
+STATIC uint
+xfs_inode_item_trylock(
+	struct xfs_log_item	*lip)
+{
+	struct xfs_inode_log_item *iip = INODE_ITEM(lip);
+	struct xfs_inode	*ip = iip->ili_inode;
+
+	if (xfs_ipincount(ip) > 0)
+		return XFS_ITEM_PINNED;
+
+	if (!xfs_ilock_nowait(ip, XFS_ILOCK_SHARED))
+		return XFS_ITEM_LOCKED;
+
+	if (!xfs_iflock_nowait(ip)) {
+		/*
+		 * inode has already been flushed to the backing buffer,
+		 * leave it locked in shared mode, pushbuf routine will
+		 * unlock it.
+		 */
+		return XFS_ITEM_PUSHBUF;
+	}
+
+	/* Stale items should force out the iclog */
+	if (ip->i_flags & XFS_ISTALE) {
+		xfs_ifunlock(ip);
+		/*
+		 * we hold the AIL lock - notify the unlock routine of this
+		 * so it doesn't try to get the lock again.
+		 */
+		xfs_iunlock(ip, XFS_ILOCK_SHARED|XFS_IUNLOCK_NONOTIFY);
+		return XFS_ITEM_PINNED;
+	}
+
+#ifdef DEBUG
+	if (!XFS_FORCED_SHUTDOWN(ip->i_mount)) {
+		ASSERT(iip->ili_format.ilf_fields != 0);
+		ASSERT(iip->ili_logged == 0);
+		ASSERT(lip->li_flags & XFS_LI_IN_AIL);
+	}
+#endif
+	return XFS_ITEM_SUCCESS;
+}
+
+/*
+ * Unlock the inode associated with the inode log item.
+ * Clear the fields of the inode and inode log item that
+ * are specific to the current transaction.  If the
+ * hold flags is set, do not unlock the inode.
+ */
+STATIC void
+xfs_inode_item_unlock(
+	struct xfs_log_item	*lip)
+{
+	struct xfs_inode_log_item *iip = INODE_ITEM(lip);
+	struct xfs_inode	*ip = iip->ili_inode;
+	unsigned short		lock_flags;
+
+	ASSERT(iip->ili_inode->i_itemp != NULL);
+	ASSERT(xfs_isilocked(iip->ili_inode, XFS_ILOCK_EXCL));
+
+	/*
+	 * Clear the transaction pointer in the inode.
+	 */
+	ip->i_transp = NULL;
+
+	/*
+	 * If the inode needed a separate buffer with which to log
+	 * its extents, then free it now.
+	 */
+	if (iip->ili_extents_buf != NULL) {
+		ASSERT(ip->i_d.di_format == XFS_DINODE_FMT_EXTENTS);
+		ASSERT(ip->i_d.di_nextents > 0);
+		ASSERT(iip->ili_format.ilf_fields & XFS_ILOG_DEXT);
+		ASSERT(ip->i_df.if_bytes > 0);
+		kmem_free(iip->ili_extents_buf);
+		iip->ili_extents_buf = NULL;
+	}
+	if (iip->ili_aextents_buf != NULL) {
+		ASSERT(ip->i_d.di_aformat == XFS_DINODE_FMT_EXTENTS);
+		ASSERT(ip->i_d.di_anextents > 0);
+		ASSERT(iip->ili_format.ilf_fields & XFS_ILOG_AEXT);
+		ASSERT(ip->i_afp->if_bytes > 0);
+		kmem_free(iip->ili_aextents_buf);
+		iip->ili_aextents_buf = NULL;
+	}
+
+	lock_flags = iip->ili_lock_flags;
+	iip->ili_lock_flags = 0;
+	if (lock_flags) {
+		xfs_iunlock(iip->ili_inode, lock_flags);
+		IRELE(iip->ili_inode);
+	}
+}
+
+/*
+ * This is called to find out where the oldest active copy of the inode log
+ * item in the on disk log resides now that the last log write of it completed
+ * at the given lsn.  Since we always re-log all dirty data in an inode, the
+ * latest copy in the on disk log is the only one that matters.  Therefore,
+ * simply return the given lsn.
+ *
+ * If the inode has been marked stale because the cluster is being freed, we
+ * don't want to (re-)insert this inode into the AIL. There is a race condition
+ * where the cluster buffer may be unpinned before the inode is inserted into
+ * the AIL during transaction committed processing. If the buffer is unpinned
+ * before the inode item has been committed and inserted, then it is possible
+ * for the buffer to be written and IO completes before the inode is inserted
+ * into the AIL. In that case, we'd be inserting a clean, stale inode into the
+ * AIL which will never get removed. It will, however, get reclaimed which
+ * triggers an assert in xfs_inode_free() complaining about freein an inode
+ * still in the AIL.
+ *
+ * To avoid this, just unpin the inode directly and return a LSN of -1 so the
+ * transaction committed code knows that it does not need to do any further
+ * processing on the item.
+ */
+STATIC xfs_lsn_t
+xfs_inode_item_committed(
+	struct xfs_log_item	*lip,
+	xfs_lsn_t		lsn)
+{
+	struct xfs_inode_log_item *iip = INODE_ITEM(lip);
+	struct xfs_inode	*ip = iip->ili_inode;
+
+	if (xfs_iflags_test(ip, XFS_ISTALE)) {
+		xfs_inode_item_unpin(lip, 0);
+		return -1;
+	}
+	return lsn;
+}
+
+/*
+ * This gets called by xfs_trans_push_ail(), when IOP_TRYLOCK
+ * failed to get the inode flush lock but did get the inode locked SHARED.
+ * Here we're trying to see if the inode buffer is incore, and if so whether it's
+ * marked delayed write. If that's the case, we'll promote it and that will
+ * allow the caller to write the buffer by triggering the xfsbufd to run.
+ */
+STATIC bool
+xfs_inode_item_pushbuf(
+	struct xfs_log_item	*lip)
+{
+	struct xfs_inode_log_item *iip = INODE_ITEM(lip);
+	struct xfs_inode	*ip = iip->ili_inode;
+	struct xfs_buf		*bp;
+	bool			ret = true;
+
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_SHARED));
+
+	/*
+	 * If a flush is not in progress anymore, chances are that the
+	 * inode was taken off the AIL. So, just get out.
+	 */
+	if (completion_done(&ip->i_flush) ||
+	    !(lip->li_flags & XFS_LI_IN_AIL)) {
+		xfs_iunlock(ip, XFS_ILOCK_SHARED);
+		return true;
+	}
+
+	bp = xfs_incore(ip->i_mount->m_ddev_targp, iip->ili_format.ilf_blkno,
+			iip->ili_format.ilf_len, XBF_TRYLOCK);
+
+	xfs_iunlock(ip, XFS_ILOCK_SHARED);
+	if (!bp)
+		return true;
+	if (XFS_BUF_ISDELAYWRITE(bp))
+		xfs_buf_delwri_promote(bp);
+	if (XFS_BUF_ISPINNED(bp))
+		ret = false;
+	xfs_buf_relse(bp);
+	return ret;
+}
+
+/*
+ * This is called to asynchronously write the inode associated with this
+ * inode log item out to disk. The inode will already have been locked by
+ * a successful call to xfs_inode_item_trylock().
+ */
+STATIC void
+xfs_inode_item_push(
+	struct xfs_log_item	*lip)
+{
+	struct xfs_inode_log_item *iip = INODE_ITEM(lip);
+	struct xfs_inode	*ip = iip->ili_inode;
+
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_SHARED));
+	ASSERT(!completion_done(&ip->i_flush));
+
+	/*
+	 * Since we were able to lock the inode's flush lock and
+	 * we found it on the AIL, the inode must be dirty.  This
+	 * is because the inode is removed from the AIL while still
+	 * holding the flush lock in xfs_iflush_done().  Thus, if
+	 * we found it in the AIL and were able to obtain the flush
+	 * lock without sleeping, then there must not have been
+	 * anyone in the process of flushing the inode.
+	 */
+	ASSERT(XFS_FORCED_SHUTDOWN(ip->i_mount) ||
+	       iip->ili_format.ilf_fields != 0);
+
+	/*
+	 * Push the inode to it's backing buffer. This will not remove the
+	 * inode from the AIL - a further push will be required to trigger a
+	 * buffer push. However, this allows all the dirty inodes to be pushed
+	 * to the buffer before it is pushed to disk. The buffer IO completion
+	 * will pull the inode from the AIL, mark it clean and unlock the flush
+	 * lock.
+	 */
+	(void) xfs_iflush(ip, SYNC_TRYLOCK);
+	xfs_iunlock(ip, XFS_ILOCK_SHARED);
+}
+
+/*
+ * XXX rcc - this one really has to do something.  Probably needs
+ * to stamp in a new field in the incore inode.
+ */
+STATIC void
+xfs_inode_item_committing(
+	struct xfs_log_item	*lip,
+	xfs_lsn_t		lsn)
+{
+	INODE_ITEM(lip)->ili_last_lsn = lsn;
+}
+
+/*
+ * This is the ops vector shared by all buf log items.
+ */
+static struct xfs_item_ops xfs_inode_item_ops = {
+	.iop_size	= xfs_inode_item_size,
+	.iop_format	= xfs_inode_item_format,
+	.iop_pin	= xfs_inode_item_pin,
+	.iop_unpin	= xfs_inode_item_unpin,
+	.iop_trylock	= xfs_inode_item_trylock,
+	.iop_unlock	= xfs_inode_item_unlock,
+	.iop_committed	= xfs_inode_item_committed,
+	.iop_push	= xfs_inode_item_push,
+	.iop_pushbuf	= xfs_inode_item_pushbuf,
+	.iop_committing = xfs_inode_item_committing
+};
+
+
+/*
+ * Initialize the inode log item for a newly allocated (in-core) inode.
+ */
+void
+xfs_inode_item_init(
+	struct xfs_inode	*ip,
+	struct xfs_mount	*mp)
+{
+	struct xfs_inode_log_item *iip;
+
+	ASSERT(ip->i_itemp == NULL);
+	iip = ip->i_itemp = kmem_zone_zalloc(xfs_ili_zone, KM_SLEEP);
+
+	iip->ili_inode = ip;
+	xfs_log_item_init(mp, &iip->ili_item, XFS_LI_INODE,
+						&xfs_inode_item_ops);
+	iip->ili_format.ilf_type = XFS_LI_INODE;
+	iip->ili_format.ilf_ino = ip->i_ino;
+	iip->ili_format.ilf_blkno = ip->i_imap.im_blkno;
+	iip->ili_format.ilf_len = ip->i_imap.im_len;
+	iip->ili_format.ilf_boffset = ip->i_imap.im_boffset;
+}
+
+/*
+ * Free the inode log item and any memory hanging off of it.
+ */
+void
+xfs_inode_item_destroy(
+	xfs_inode_t	*ip)
+{
+#ifdef XFS_TRANS_DEBUG
+	if (ip->i_itemp->ili_root_size != 0) {
+		kmem_free(ip->i_itemp->ili_orig_root);
+	}
+#endif
+	kmem_zone_free(xfs_ili_zone, ip->i_itemp);
+}
+
+
+/*
+ * This is the inode flushing I/O completion routine.  It is called
+ * from interrupt level when the buffer containing the inode is
+ * flushed to disk.  It is responsible for removing the inode item
+ * from the AIL if it has not been re-logged, and unlocking the inode's
+ * flush lock.
+ *
+ * To reduce AIL lock traffic as much as possible, we scan the buffer log item
+ * list for other inodes that will run this function. We remove them from the
+ * buffer list so we can process all the inode IO completions in one AIL lock
+ * traversal.
+ */
+void
+xfs_iflush_done(
+	struct xfs_buf		*bp,
+	struct xfs_log_item	*lip)
+{
+	struct xfs_inode_log_item *iip;
+	struct xfs_log_item	*blip;
+	struct xfs_log_item	*next;
+	struct xfs_log_item	*prev;
+	struct xfs_ail		*ailp = lip->li_ailp;
+	int			need_ail = 0;
+
+	/*
+	 * Scan the buffer IO completions for other inodes being completed and
+	 * attach them to the current inode log item.
+	 */
+	blip = XFS_BUF_FSPRIVATE(bp, xfs_log_item_t *);
+	prev = NULL;
+	while (blip != NULL) {
+		if (lip->li_cb != xfs_iflush_done) {
+			prev = blip;
+			blip = blip->li_bio_list;
+			continue;
+		}
+
+		/* remove from list */
+		next = blip->li_bio_list;
+		if (!prev) {
+			XFS_BUF_SET_FSPRIVATE(bp, next);
+		} else {
+			prev->li_bio_list = next;
+		}
+
+		/* add to current list */
+		blip->li_bio_list = lip->li_bio_list;
+		lip->li_bio_list = blip;
+
+		/*
+		 * while we have the item, do the unlocked check for needing
+		 * the AIL lock.
+		 */
+		iip = INODE_ITEM(blip);
+		if (iip->ili_logged && blip->li_lsn == iip->ili_flush_lsn)
+			need_ail++;
+
+		blip = next;
+	}
+
+	/* make sure we capture the state of the initial inode. */
+	iip = INODE_ITEM(lip);
+	if (iip->ili_logged && lip->li_lsn == iip->ili_flush_lsn)
+		need_ail++;
+
+	/*
+	 * We only want to pull the item from the AIL if it is
+	 * actually there and its location in the log has not
+	 * changed since we started the flush.  Thus, we only bother
+	 * if the ili_logged flag is set and the inode's lsn has not
+	 * changed.  First we check the lsn outside
+	 * the lock since it's cheaper, and then we recheck while
+	 * holding the lock before removing the inode from the AIL.
+	 */
+	if (need_ail) {
+		struct xfs_log_item *log_items[need_ail];
+		int i = 0;
+		spin_lock(&ailp->xa_lock);
+		for (blip = lip; blip; blip = blip->li_bio_list) {
+			iip = INODE_ITEM(blip);
+			if (iip->ili_logged &&
+			    blip->li_lsn == iip->ili_flush_lsn) {
+				log_items[i++] = blip;
+			}
+			ASSERT(i <= need_ail);
+		}
+		/* xfs_trans_ail_delete_bulk() drops the AIL lock. */
+		xfs_trans_ail_delete_bulk(ailp, log_items, i);
+	}
+
+
+	/*
+	 * clean up and unlock the flush lock now we are done. We can clear the
+	 * ili_last_fields bits now that we know that the data corresponding to
+	 * them is safely on disk.
+	 */
+	for (blip = lip; blip; blip = next) {
+		next = blip->li_bio_list;
+		blip->li_bio_list = NULL;
+
+		iip = INODE_ITEM(blip);
+		iip->ili_logged = 0;
+		iip->ili_last_fields = 0;
+		xfs_ifunlock(iip->ili_inode);
+	}
+}
+
+/*
+ * This is the inode flushing abort routine.  It is called
+ * from xfs_iflush when the filesystem is shutting down to clean
+ * up the inode state.
+ * It is responsible for removing the inode item
+ * from the AIL if it has not been re-logged, and unlocking the inode's
+ * flush lock.
+ */
+void
+xfs_iflush_abort(
+	xfs_inode_t		*ip)
+{
+	xfs_inode_log_item_t	*iip = ip->i_itemp;
+
+	if (iip) {
+		struct xfs_ail	*ailp = iip->ili_item.li_ailp;
+		if (iip->ili_item.li_flags & XFS_LI_IN_AIL) {
+			spin_lock(&ailp->xa_lock);
+			if (iip->ili_item.li_flags & XFS_LI_IN_AIL) {
+				/* xfs_trans_ail_delete() drops the AIL lock. */
+				xfs_trans_ail_delete(ailp, (xfs_log_item_t *)iip);
+			} else
+				spin_unlock(&ailp->xa_lock);
+		}
+		iip->ili_logged = 0;
+		/*
+		 * Clear the ili_last_fields bits now that we know that the
+		 * data corresponding to them is safely on disk.
+		 */
+		iip->ili_last_fields = 0;
+		/*
+		 * Clear the inode logging fields so no more flushes are
+		 * attempted.
+		 */
+		iip->ili_format.ilf_fields = 0;
+	}
+	/*
+	 * Release the inode's flush lock since we're done with it.
+	 */
+	xfs_ifunlock(ip);
+}
+
+void
+xfs_istale_done(
+	struct xfs_buf		*bp,
+	struct xfs_log_item	*lip)
+{
+	xfs_iflush_abort(INODE_ITEM(lip)->ili_inode);
+}
+
+/*
+ * convert an xfs_inode_log_format struct from either 32 or 64 bit versions
+ * (which can have different field alignments) to the native version
+ */
+int
+xfs_inode_item_format_convert(
+	xfs_log_iovec_t		*buf,
+	xfs_inode_log_format_t	*in_f)
+{
+	if (buf->i_len == sizeof(xfs_inode_log_format_32_t)) {
+		xfs_inode_log_format_32_t *in_f32 = buf->i_addr;
+
+		in_f->ilf_type = in_f32->ilf_type;
+		in_f->ilf_size = in_f32->ilf_size;
+		in_f->ilf_fields = in_f32->ilf_fields;
+		in_f->ilf_asize = in_f32->ilf_asize;
+		in_f->ilf_dsize = in_f32->ilf_dsize;
+		in_f->ilf_ino = in_f32->ilf_ino;
+		/* copy biggest field of ilf_u */
+		memcpy(in_f->ilf_u.ilfu_uuid.__u_bits,
+		       in_f32->ilf_u.ilfu_uuid.__u_bits,
+		       sizeof(uuid_t));
+		in_f->ilf_blkno = in_f32->ilf_blkno;
+		in_f->ilf_len = in_f32->ilf_len;
+		in_f->ilf_boffset = in_f32->ilf_boffset;
+		return 0;
+	} else if (buf->i_len == sizeof(xfs_inode_log_format_64_t)){
+		xfs_inode_log_format_64_t *in_f64 = buf->i_addr;
+
+		in_f->ilf_type = in_f64->ilf_type;
+		in_f->ilf_size = in_f64->ilf_size;
+		in_f->ilf_fields = in_f64->ilf_fields;
+		in_f->ilf_asize = in_f64->ilf_asize;
+		in_f->ilf_dsize = in_f64->ilf_dsize;
+		in_f->ilf_ino = in_f64->ilf_ino;
+		/* copy biggest field of ilf_u */
+		memcpy(in_f->ilf_u.ilfu_uuid.__u_bits,
+		       in_f64->ilf_u.ilfu_uuid.__u_bits,
+		       sizeof(uuid_t));
+		in_f->ilf_blkno = in_f64->ilf_blkno;
+		in_f->ilf_len = in_f64->ilf_len;
+		in_f->ilf_boffset = in_f64->ilf_boffset;
+		return 0;
+	}
+	return EFSCORRUPTED;
+}
diff --git a/fs/xfs/xfs_inode_item.h b/fs/xfs/xfs_inode_item.h
new file mode 100644
index 00000000..d3dee61e
--- /dev/null
+++ b/fs/xfs/xfs_inode_item.h
@@ -0,0 +1,166 @@
+/*
+ * Copyright (c) 2000,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef	__XFS_INODE_ITEM_H__
+#define	__XFS_INODE_ITEM_H__
+
+/*
+ * This is the structure used to lay out an inode log item in the
+ * log.  The size of the inline data/extents/b-tree root to be logged
+ * (if any) is indicated in the ilf_dsize field.  Changes to this structure
+ * must be added on to the end.
+ */
+typedef struct xfs_inode_log_format {
+	__uint16_t		ilf_type;	/* inode log item type */
+	__uint16_t		ilf_size;	/* size of this item */
+	__uint32_t		ilf_fields;	/* flags for fields logged */
+	__uint16_t		ilf_asize;	/* size of attr d/ext/root */
+	__uint16_t		ilf_dsize;	/* size of data/ext/root */
+	__uint64_t		ilf_ino;	/* inode number */
+	union {
+		__uint32_t	ilfu_rdev;	/* rdev value for dev inode*/
+		uuid_t		ilfu_uuid;	/* mount point value */
+	} ilf_u;
+	__int64_t		ilf_blkno;	/* blkno of inode buffer */
+	__int32_t		ilf_len;	/* len of inode buffer */
+	__int32_t		ilf_boffset;	/* off of inode in buffer */
+} xfs_inode_log_format_t;
+
+typedef struct xfs_inode_log_format_32 {
+	__uint16_t		ilf_type;	/* inode log item type */
+	__uint16_t		ilf_size;	/* size of this item */
+	__uint32_t		ilf_fields;	/* flags for fields logged */
+	__uint16_t		ilf_asize;	/* size of attr d/ext/root */
+	__uint16_t		ilf_dsize;	/* size of data/ext/root */
+	__uint64_t		ilf_ino;	/* inode number */
+	union {
+		__uint32_t	ilfu_rdev;	/* rdev value for dev inode*/
+		uuid_t		ilfu_uuid;	/* mount point value */
+	} ilf_u;
+	__int64_t		ilf_blkno;	/* blkno of inode buffer */
+	__int32_t		ilf_len;	/* len of inode buffer */
+	__int32_t		ilf_boffset;	/* off of inode in buffer */
+} __attribute__((packed)) xfs_inode_log_format_32_t;
+
+typedef struct xfs_inode_log_format_64 {
+	__uint16_t		ilf_type;	/* inode log item type */
+	__uint16_t		ilf_size;	/* size of this item */
+	__uint32_t		ilf_fields;	/* flags for fields logged */
+	__uint16_t		ilf_asize;	/* size of attr d/ext/root */
+	__uint16_t		ilf_dsize;	/* size of data/ext/root */
+	__uint32_t		ilf_pad;	/* pad for 64 bit boundary */
+	__uint64_t		ilf_ino;	/* inode number */
+	union {
+		__uint32_t	ilfu_rdev;	/* rdev value for dev inode*/
+		uuid_t		ilfu_uuid;	/* mount point value */
+	} ilf_u;
+	__int64_t		ilf_blkno;	/* blkno of inode buffer */
+	__int32_t		ilf_len;	/* len of inode buffer */
+	__int32_t		ilf_boffset;	/* off of inode in buffer */
+} xfs_inode_log_format_64_t;
+
+/*
+ * Flags for xfs_trans_log_inode flags field.
+ */
+#define	XFS_ILOG_CORE	0x001	/* log standard inode fields */
+#define	XFS_ILOG_DDATA	0x002	/* log i_df.if_data */
+#define	XFS_ILOG_DEXT	0x004	/* log i_df.if_extents */
+#define	XFS_ILOG_DBROOT	0x008	/* log i_df.i_broot */
+#define	XFS_ILOG_DEV	0x010	/* log the dev field */
+#define	XFS_ILOG_UUID	0x020	/* log the uuid field */
+#define	XFS_ILOG_ADATA	0x040	/* log i_af.if_data */
+#define	XFS_ILOG_AEXT	0x080	/* log i_af.if_extents */
+#define	XFS_ILOG_ABROOT	0x100	/* log i_af.i_broot */
+
+#define	XFS_ILOG_NONCORE	(XFS_ILOG_DDATA | XFS_ILOG_DEXT | \
+				 XFS_ILOG_DBROOT | XFS_ILOG_DEV | \
+				 XFS_ILOG_UUID | XFS_ILOG_ADATA | \
+				 XFS_ILOG_AEXT | XFS_ILOG_ABROOT)
+
+#define	XFS_ILOG_DFORK		(XFS_ILOG_DDATA | XFS_ILOG_DEXT | \
+				 XFS_ILOG_DBROOT)
+
+#define	XFS_ILOG_AFORK		(XFS_ILOG_ADATA | XFS_ILOG_AEXT | \
+				 XFS_ILOG_ABROOT)
+
+#define	XFS_ILOG_ALL		(XFS_ILOG_CORE | XFS_ILOG_DDATA | \
+				 XFS_ILOG_DEXT | XFS_ILOG_DBROOT | \
+				 XFS_ILOG_DEV | XFS_ILOG_UUID | \
+				 XFS_ILOG_ADATA | XFS_ILOG_AEXT | \
+				 XFS_ILOG_ABROOT)
+
+static inline int xfs_ilog_fbroot(int w)
+{
+	return (w == XFS_DATA_FORK ? XFS_ILOG_DBROOT : XFS_ILOG_ABROOT);
+}
+
+static inline int xfs_ilog_fext(int w)
+{
+	return (w == XFS_DATA_FORK ? XFS_ILOG_DEXT : XFS_ILOG_AEXT);
+}
+
+static inline int xfs_ilog_fdata(int w)
+{
+	return (w == XFS_DATA_FORK ? XFS_ILOG_DDATA : XFS_ILOG_ADATA);
+}
+
+#ifdef __KERNEL__
+
+struct xfs_buf;
+struct xfs_bmbt_rec;
+struct xfs_inode;
+struct xfs_mount;
+
+
+typedef struct xfs_inode_log_item {
+	xfs_log_item_t		ili_item;	   /* common portion */
+	struct xfs_inode	*ili_inode;	   /* inode ptr */
+	xfs_lsn_t		ili_flush_lsn;	   /* lsn at last flush */
+	xfs_lsn_t		ili_last_lsn;	   /* lsn at last transaction */
+	unsigned short		ili_lock_flags;	   /* lock flags */
+	unsigned short		ili_logged;	   /* flushed logged data */
+	unsigned int		ili_last_fields;   /* fields when flushed */
+	struct xfs_bmbt_rec	*ili_extents_buf;  /* array of logged
+						      data exts */
+	struct xfs_bmbt_rec	*ili_aextents_buf; /* array of logged
+						      attr exts */
+#ifdef XFS_TRANS_DEBUG
+	int			ili_root_size;
+	char			*ili_orig_root;
+#endif
+	xfs_inode_log_format_t	ili_format;	   /* logged structure */
+} xfs_inode_log_item_t;
+
+
+static inline int xfs_inode_clean(xfs_inode_t *ip)
+{
+	return (!ip->i_itemp ||
+		!(ip->i_itemp->ili_format.ilf_fields & XFS_ILOG_ALL)) &&
+	       !ip->i_update_core;
+}
+
+extern void xfs_inode_item_init(struct xfs_inode *, struct xfs_mount *);
+extern void xfs_inode_item_destroy(struct xfs_inode *);
+extern void xfs_iflush_done(struct xfs_buf *, struct xfs_log_item *);
+extern void xfs_istale_done(struct xfs_buf *, struct xfs_log_item *);
+extern void xfs_iflush_abort(struct xfs_inode *);
+extern int xfs_inode_item_format_convert(xfs_log_iovec_t *,
+					 xfs_inode_log_format_t *);
+
+#endif	/* __KERNEL__ */
+
+#endif	/* __XFS_INODE_ITEM_H__ */
diff --git a/fs/xfs/xfs_inum.h b/fs/xfs/xfs_inum.h
new file mode 100644
index 00000000..b8e4ee4e
--- /dev/null
+++ b/fs/xfs/xfs_inum.h
@@ -0,0 +1,80 @@
+/*
+ * Copyright (c) 2000-2003,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_INUM_H__
+#define	__XFS_INUM_H__
+
+/*
+ * Inode number format:
+ * low inopblog bits - offset in block
+ * next agblklog bits - block number in ag
+ * next agno_log bits - ag number
+ * high agno_log-agblklog-inopblog bits - 0
+ */
+
+typedef	__uint32_t	xfs_agino_t;	/* within allocation grp inode number */
+
+/*
+ * Useful inode bits for this kernel.
+ * Used in some places where having 64-bits in the 32-bit kernels
+ * costs too much.
+ */
+#if XFS_BIG_INUMS
+typedef	xfs_ino_t	xfs_intino_t;
+#else
+typedef	__uint32_t	xfs_intino_t;
+#endif
+
+#define	NULLFSINO	((xfs_ino_t)-1)
+#define	NULLAGINO	((xfs_agino_t)-1)
+
+struct xfs_mount;
+
+#define	XFS_INO_MASK(k)			(__uint32_t)((1ULL << (k)) - 1)
+#define	XFS_INO_OFFSET_BITS(mp)		(mp)->m_sb.sb_inopblog
+#define	XFS_INO_AGBNO_BITS(mp)		(mp)->m_sb.sb_agblklog
+#define	XFS_INO_AGINO_BITS(mp)		(mp)->m_agino_log
+#define	XFS_INO_AGNO_BITS(mp)		(mp)->m_agno_log
+#define	XFS_INO_BITS(mp)		\
+	XFS_INO_AGNO_BITS(mp) + XFS_INO_AGINO_BITS(mp)
+#define	XFS_INO_TO_AGNO(mp,i)		\
+	((xfs_agnumber_t)((i) >> XFS_INO_AGINO_BITS(mp)))
+#define	XFS_INO_TO_AGINO(mp,i)		\
+	((xfs_agino_t)(i) & XFS_INO_MASK(XFS_INO_AGINO_BITS(mp)))
+#define	XFS_INO_TO_AGBNO(mp,i)		\
+	(((xfs_agblock_t)(i) >> XFS_INO_OFFSET_BITS(mp)) & \
+		XFS_INO_MASK(XFS_INO_AGBNO_BITS(mp)))
+#define	XFS_INO_TO_OFFSET(mp,i)		\
+	((int)(i) & XFS_INO_MASK(XFS_INO_OFFSET_BITS(mp)))
+#define	XFS_INO_TO_FSB(mp,i)		\
+	XFS_AGB_TO_FSB(mp, XFS_INO_TO_AGNO(mp,i), XFS_INO_TO_AGBNO(mp,i))
+#define	XFS_AGINO_TO_INO(mp,a,i)	\
+	(((xfs_ino_t)(a) << XFS_INO_AGINO_BITS(mp)) | (i))
+#define	XFS_AGINO_TO_AGBNO(mp,i)	((i) >> XFS_INO_OFFSET_BITS(mp))
+#define	XFS_AGINO_TO_OFFSET(mp,i)	\
+	((i) & XFS_INO_MASK(XFS_INO_OFFSET_BITS(mp)))
+#define	XFS_OFFBNO_TO_AGINO(mp,b,o)	\
+	((xfs_agino_t)(((b) << XFS_INO_OFFSET_BITS(mp)) | (o)))
+
+#if XFS_BIG_INUMS
+#define	XFS_MAXINUMBER		((xfs_ino_t)((1ULL << 56) - 1ULL))
+#else
+#define	XFS_MAXINUMBER		((xfs_ino_t)((1ULL << 32) - 1ULL))
+#endif
+#define	XFS_MAXINUMBER_32	((xfs_ino_t)((1ULL << 32) - 1ULL))
+
+#endif	/* __XFS_INUM_H__ */
diff --git a/fs/xfs/xfs_iomap.c b/fs/xfs/xfs_iomap.c
new file mode 100644
index 00000000..091d82b9
--- /dev/null
+++ b/fs/xfs/xfs_iomap.c
@@ -0,0 +1,748 @@
+/*
+ * Copyright (c) 2000-2006 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_alloc.h"
+#include "xfs_quota.h"
+#include "xfs_mount.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_alloc_btree.h"
+#include "xfs_ialloc_btree.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_btree.h"
+#include "xfs_bmap.h"
+#include "xfs_rtalloc.h"
+#include "xfs_error.h"
+#include "xfs_itable.h"
+#include "xfs_rw.h"
+#include "xfs_attr.h"
+#include "xfs_buf_item.h"
+#include "xfs_trans_space.h"
+#include "xfs_utils.h"
+#include "xfs_iomap.h"
+#include "xfs_trace.h"
+
+
+#define XFS_WRITEIO_ALIGN(mp,off)	(((off) >> mp->m_writeio_log) \
+						<< mp->m_writeio_log)
+#define XFS_WRITE_IMAPS		XFS_BMAP_MAX_NMAP
+
+STATIC int
+xfs_iomap_eof_align_last_fsb(
+	xfs_mount_t	*mp,
+	xfs_inode_t	*ip,
+	xfs_extlen_t	extsize,
+	xfs_fileoff_t	*last_fsb)
+{
+	xfs_fileoff_t	new_last_fsb = 0;
+	xfs_extlen_t	align;
+	int		eof, error;
+
+	if (XFS_IS_REALTIME_INODE(ip))
+		;
+	/*
+	 * If mounted with the "-o swalloc" option, roundup the allocation
+	 * request to a stripe width boundary if the file size is >=
+	 * stripe width and we are allocating past the allocation eof.
+	 */
+	else if (mp->m_swidth && (mp->m_flags & XFS_MOUNT_SWALLOC) &&
+	        (ip->i_size >= XFS_FSB_TO_B(mp, mp->m_swidth)))
+		new_last_fsb = roundup_64(*last_fsb, mp->m_swidth);
+	/*
+	 * Roundup the allocation request to a stripe unit (m_dalign) boundary
+	 * if the file size is >= stripe unit size, and we are allocating past
+	 * the allocation eof.
+	 */
+	else if (mp->m_dalign && (ip->i_size >= XFS_FSB_TO_B(mp, mp->m_dalign)))
+		new_last_fsb = roundup_64(*last_fsb, mp->m_dalign);
+
+	/*
+	 * Always round up the allocation request to an extent boundary
+	 * (when file on a real-time subvolume or has di_extsize hint).
+	 */
+	if (extsize) {
+		if (new_last_fsb)
+			align = roundup_64(new_last_fsb, extsize);
+		else
+			align = extsize;
+		new_last_fsb = roundup_64(*last_fsb, align);
+	}
+
+	if (new_last_fsb) {
+		error = xfs_bmap_eof(ip, new_last_fsb, XFS_DATA_FORK, &eof);
+		if (error)
+			return error;
+		if (eof)
+			*last_fsb = new_last_fsb;
+	}
+	return 0;
+}
+
+STATIC int
+xfs_alert_fsblock_zero(
+	xfs_inode_t	*ip,
+	xfs_bmbt_irec_t	*imap)
+{
+	xfs_alert_tag(ip->i_mount, XFS_PTAG_FSBLOCK_ZERO,
+			"Access to block zero in inode %llu "
+			"start_block: %llx start_off: %llx "
+			"blkcnt: %llx extent-state: %x\n",
+		(unsigned long long)ip->i_ino,
+		(unsigned long long)imap->br_startblock,
+		(unsigned long long)imap->br_startoff,
+		(unsigned long long)imap->br_blockcount,
+		imap->br_state);
+	return EFSCORRUPTED;
+}
+
+int
+xfs_iomap_write_direct(
+	xfs_inode_t	*ip,
+	xfs_off_t	offset,
+	size_t		count,
+	xfs_bmbt_irec_t *imap,
+	int		nmaps)
+{
+	xfs_mount_t	*mp = ip->i_mount;
+	xfs_fileoff_t	offset_fsb;
+	xfs_fileoff_t	last_fsb;
+	xfs_filblks_t	count_fsb, resaligned;
+	xfs_fsblock_t	firstfsb;
+	xfs_extlen_t	extsz, temp;
+	int		nimaps;
+	int		bmapi_flag;
+	int		quota_flag;
+	int		rt;
+	xfs_trans_t	*tp;
+	xfs_bmap_free_t free_list;
+	uint		qblocks, resblks, resrtextents;
+	int		committed;
+	int		error;
+
+	/*
+	 * Make sure that the dquots are there. This doesn't hold
+	 * the ilock across a disk read.
+	 */
+	error = xfs_qm_dqattach_locked(ip, 0);
+	if (error)
+		return XFS_ERROR(error);
+
+	rt = XFS_IS_REALTIME_INODE(ip);
+	extsz = xfs_get_extsz_hint(ip);
+
+	offset_fsb = XFS_B_TO_FSBT(mp, offset);
+	last_fsb = XFS_B_TO_FSB(mp, ((xfs_ufsize_t)(offset + count)));
+	if ((offset + count) > ip->i_size) {
+		error = xfs_iomap_eof_align_last_fsb(mp, ip, extsz, &last_fsb);
+		if (error)
+			goto error_out;
+	} else {
+		if (nmaps && (imap->br_startblock == HOLESTARTBLOCK))
+			last_fsb = MIN(last_fsb, (xfs_fileoff_t)
+					imap->br_blockcount +
+					imap->br_startoff);
+	}
+	count_fsb = last_fsb - offset_fsb;
+	ASSERT(count_fsb > 0);
+
+	resaligned = count_fsb;
+	if (unlikely(extsz)) {
+		if ((temp = do_mod(offset_fsb, extsz)))
+			resaligned += temp;
+		if ((temp = do_mod(resaligned, extsz)))
+			resaligned += extsz - temp;
+	}
+
+	if (unlikely(rt)) {
+		resrtextents = qblocks = resaligned;
+		resrtextents /= mp->m_sb.sb_rextsize;
+		resblks = XFS_DIOSTRAT_SPACE_RES(mp, 0);
+		quota_flag = XFS_QMOPT_RES_RTBLKS;
+	} else {
+		resrtextents = 0;
+		resblks = qblocks = XFS_DIOSTRAT_SPACE_RES(mp, resaligned);
+		quota_flag = XFS_QMOPT_RES_REGBLKS;
+	}
+
+	/*
+	 * Allocate and setup the transaction
+	 */
+	xfs_iunlock(ip, XFS_ILOCK_EXCL);
+	tp = xfs_trans_alloc(mp, XFS_TRANS_DIOSTRAT);
+	error = xfs_trans_reserve(tp, resblks,
+			XFS_WRITE_LOG_RES(mp), resrtextents,
+			XFS_TRANS_PERM_LOG_RES,
+			XFS_WRITE_LOG_COUNT);
+	/*
+	 * Check for running out of space, note: need lock to return
+	 */
+	if (error)
+		xfs_trans_cancel(tp, 0);
+	xfs_ilock(ip, XFS_ILOCK_EXCL);
+	if (error)
+		goto error_out;
+
+	error = xfs_trans_reserve_quota_nblks(tp, ip, qblocks, 0, quota_flag);
+	if (error)
+		goto error1;
+
+	xfs_trans_ijoin(tp, ip);
+
+	bmapi_flag = XFS_BMAPI_WRITE;
+	if (offset < ip->i_size || extsz)
+		bmapi_flag |= XFS_BMAPI_PREALLOC;
+
+	/*
+	 * Issue the xfs_bmapi() call to allocate the blocks.
+	 *
+	 * From this point onwards we overwrite the imap pointer that the
+	 * caller gave to us.
+	 */
+	xfs_bmap_init(&free_list, &firstfsb);
+	nimaps = 1;
+	error = xfs_bmapi(tp, ip, offset_fsb, count_fsb, bmapi_flag,
+		&firstfsb, 0, imap, &nimaps, &free_list);
+	if (error)
+		goto error0;
+
+	/*
+	 * Complete the transaction
+	 */
+	error = xfs_bmap_finish(&tp, &free_list, &committed);
+	if (error)
+		goto error0;
+	error = xfs_trans_commit(tp, XFS_TRANS_RELEASE_LOG_RES);
+	if (error)
+		goto error_out;
+
+	/*
+	 * Copy any maps to caller's array and return any error.
+	 */
+	if (nimaps == 0) {
+		error = ENOSPC;
+		goto error_out;
+	}
+
+	if (!(imap->br_startblock || XFS_IS_REALTIME_INODE(ip))) {
+		error = xfs_alert_fsblock_zero(ip, imap);
+		goto error_out;
+	}
+
+	return 0;
+
+error0:	/* Cancel bmap, unlock inode, unreserve quota blocks, cancel trans */
+	xfs_bmap_cancel(&free_list);
+	xfs_trans_unreserve_quota_nblks(tp, ip, qblocks, 0, quota_flag);
+
+error1:	/* Just cancel transaction */
+	xfs_trans_cancel(tp, XFS_TRANS_RELEASE_LOG_RES | XFS_TRANS_ABORT);
+
+error_out:
+	return XFS_ERROR(error);
+}
+
+/*
+ * If the caller is doing a write at the end of the file, then extend the
+ * allocation out to the file system's write iosize.  We clean up any extra
+ * space left over when the file is closed in xfs_inactive().
+ *
+ * If we find we already have delalloc preallocation beyond EOF, don't do more
+ * preallocation as it it not needed.
+ */
+STATIC int
+xfs_iomap_eof_want_preallocate(
+	xfs_mount_t	*mp,
+	xfs_inode_t	*ip,
+	xfs_off_t	offset,
+	size_t		count,
+	xfs_bmbt_irec_t *imap,
+	int		nimaps,
+	int		*prealloc)
+{
+	xfs_fileoff_t   start_fsb;
+	xfs_filblks_t   count_fsb;
+	xfs_fsblock_t	firstblock;
+	int		n, error, imaps;
+	int		found_delalloc = 0;
+
+	*prealloc = 0;
+	if ((offset + count) <= ip->i_size)
+		return 0;
+
+	/*
+	 * If there are any real blocks past eof, then don't
+	 * do any speculative allocation.
+	 */
+	start_fsb = XFS_B_TO_FSBT(mp, ((xfs_ufsize_t)(offset + count - 1)));
+	count_fsb = XFS_B_TO_FSB(mp, (xfs_ufsize_t)XFS_MAXIOFFSET(mp));
+	while (count_fsb > 0) {
+		imaps = nimaps;
+		firstblock = NULLFSBLOCK;
+		error = xfs_bmapi(NULL, ip, start_fsb, count_fsb, 0,
+				  &firstblock, 0, imap, &imaps, NULL);
+		if (error)
+			return error;
+		for (n = 0; n < imaps; n++) {
+			if ((imap[n].br_startblock != HOLESTARTBLOCK) &&
+			    (imap[n].br_startblock != DELAYSTARTBLOCK))
+				return 0;
+			start_fsb += imap[n].br_blockcount;
+			count_fsb -= imap[n].br_blockcount;
+
+			if (imap[n].br_startblock == DELAYSTARTBLOCK)
+				found_delalloc = 1;
+		}
+	}
+	if (!found_delalloc)
+		*prealloc = 1;
+	return 0;
+}
+
+/*
+ * If we don't have a user specified preallocation size, dynamically increase
+ * the preallocation size as the size of the file grows. Cap the maximum size
+ * at a single extent or less if the filesystem is near full. The closer the
+ * filesystem is to full, the smaller the maximum prealocation.
+ */
+STATIC xfs_fsblock_t
+xfs_iomap_prealloc_size(
+	struct xfs_mount	*mp,
+	struct xfs_inode	*ip)
+{
+	xfs_fsblock_t		alloc_blocks = 0;
+
+	if (!(mp->m_flags & XFS_MOUNT_DFLT_IOSIZE)) {
+		int shift = 0;
+		int64_t freesp;
+
+		/*
+		 * rounddown_pow_of_two() returns an undefined result
+		 * if we pass in alloc_blocks = 0. Hence the "+ 1" to
+		 * ensure we always pass in a non-zero value.
+		 */
+		alloc_blocks = XFS_B_TO_FSB(mp, ip->i_size) + 1;
+		alloc_blocks = XFS_FILEOFF_MIN(MAXEXTLEN,
+					rounddown_pow_of_two(alloc_blocks));
+
+		xfs_icsb_sync_counters(mp, XFS_ICSB_LAZY_COUNT);
+		freesp = mp->m_sb.sb_fdblocks;
+		if (freesp < mp->m_low_space[XFS_LOWSP_5_PCNT]) {
+			shift = 2;
+			if (freesp < mp->m_low_space[XFS_LOWSP_4_PCNT])
+				shift++;
+			if (freesp < mp->m_low_space[XFS_LOWSP_3_PCNT])
+				shift++;
+			if (freesp < mp->m_low_space[XFS_LOWSP_2_PCNT])
+				shift++;
+			if (freesp < mp->m_low_space[XFS_LOWSP_1_PCNT])
+				shift++;
+		}
+		if (shift)
+			alloc_blocks >>= shift;
+	}
+
+	if (alloc_blocks < mp->m_writeio_blocks)
+		alloc_blocks = mp->m_writeio_blocks;
+
+	return alloc_blocks;
+}
+
+int
+xfs_iomap_write_delay(
+	xfs_inode_t	*ip,
+	xfs_off_t	offset,
+	size_t		count,
+	xfs_bmbt_irec_t *ret_imap)
+{
+	xfs_mount_t	*mp = ip->i_mount;
+	xfs_fileoff_t	offset_fsb;
+	xfs_fileoff_t	last_fsb;
+	xfs_off_t	aligned_offset;
+	xfs_fileoff_t	ioalign;
+	xfs_fsblock_t	firstblock;
+	xfs_extlen_t	extsz;
+	int		nimaps;
+	xfs_bmbt_irec_t imap[XFS_WRITE_IMAPS];
+	int		prealloc, flushed = 0;
+	int		error;
+
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
+
+	/*
+	 * Make sure that the dquots are there. This doesn't hold
+	 * the ilock across a disk read.
+	 */
+	error = xfs_qm_dqattach_locked(ip, 0);
+	if (error)
+		return XFS_ERROR(error);
+
+	extsz = xfs_get_extsz_hint(ip);
+	offset_fsb = XFS_B_TO_FSBT(mp, offset);
+
+
+	error = xfs_iomap_eof_want_preallocate(mp, ip, offset, count,
+				imap, XFS_WRITE_IMAPS, &prealloc);
+	if (error)
+		return error;
+
+retry:
+	if (prealloc) {
+		xfs_fsblock_t	alloc_blocks = xfs_iomap_prealloc_size(mp, ip);
+
+		aligned_offset = XFS_WRITEIO_ALIGN(mp, (offset + count - 1));
+		ioalign = XFS_B_TO_FSBT(mp, aligned_offset);
+		last_fsb = ioalign + alloc_blocks;
+	} else {
+		last_fsb = XFS_B_TO_FSB(mp, ((xfs_ufsize_t)(offset + count)));
+	}
+
+	if (prealloc || extsz) {
+		error = xfs_iomap_eof_align_last_fsb(mp, ip, extsz, &last_fsb);
+		if (error)
+			return error;
+	}
+
+	nimaps = XFS_WRITE_IMAPS;
+	firstblock = NULLFSBLOCK;
+	error = xfs_bmapi(NULL, ip, offset_fsb,
+			  (xfs_filblks_t)(last_fsb - offset_fsb),
+			  XFS_BMAPI_DELAY | XFS_BMAPI_WRITE |
+			  XFS_BMAPI_ENTIRE, &firstblock, 1, imap,
+			  &nimaps, NULL);
+	switch (error) {
+	case 0:
+	case ENOSPC:
+	case EDQUOT:
+		break;
+	default:
+		return XFS_ERROR(error);
+	}
+
+	/*
+	 * If bmapi returned us nothing, we got either ENOSPC or EDQUOT.  For
+	 * ENOSPC, * flush all other inodes with delalloc blocks to free up
+	 * some of the excess reserved metadata space. For both cases, retry
+	 * without EOF preallocation.
+	 */
+	if (nimaps == 0) {
+		trace_xfs_delalloc_enospc(ip, offset, count);
+		if (flushed)
+			return XFS_ERROR(error ? error : ENOSPC);
+
+		if (error == ENOSPC) {
+			xfs_iunlock(ip, XFS_ILOCK_EXCL);
+			xfs_flush_inodes(ip);
+			xfs_ilock(ip, XFS_ILOCK_EXCL);
+		}
+
+		flushed = 1;
+		error = 0;
+		prealloc = 0;
+		goto retry;
+	}
+
+	if (!(imap[0].br_startblock || XFS_IS_REALTIME_INODE(ip)))
+		return xfs_alert_fsblock_zero(ip, &imap[0]);
+
+	*ret_imap = imap[0];
+	return 0;
+}
+
+/*
+ * Pass in a delayed allocate extent, convert it to real extents;
+ * return to the caller the extent we create which maps on top of
+ * the originating callers request.
+ *
+ * Called without a lock on the inode.
+ *
+ * We no longer bother to look at the incoming map - all we have to
+ * guarantee is that whatever we allocate fills the required range.
+ */
+int
+xfs_iomap_write_allocate(
+	xfs_inode_t	*ip,
+	xfs_off_t	offset,
+	size_t		count,
+	xfs_bmbt_irec_t *imap)
+{
+	xfs_mount_t	*mp = ip->i_mount;
+	xfs_fileoff_t	offset_fsb, last_block;
+	xfs_fileoff_t	end_fsb, map_start_fsb;
+	xfs_fsblock_t	first_block;
+	xfs_bmap_free_t	free_list;
+	xfs_filblks_t	count_fsb;
+	xfs_trans_t	*tp;
+	int		nimaps, committed;
+	int		error = 0;
+	int		nres;
+
+	/*
+	 * Make sure that the dquots are there.
+	 */
+	error = xfs_qm_dqattach(ip, 0);
+	if (error)
+		return XFS_ERROR(error);
+
+	offset_fsb = XFS_B_TO_FSBT(mp, offset);
+	count_fsb = imap->br_blockcount;
+	map_start_fsb = imap->br_startoff;
+
+	XFS_STATS_ADD(xs_xstrat_bytes, XFS_FSB_TO_B(mp, count_fsb));
+
+	while (count_fsb != 0) {
+		/*
+		 * Set up a transaction with which to allocate the
+		 * backing store for the file.  Do allocations in a
+		 * loop until we get some space in the range we are
+		 * interested in.  The other space that might be allocated
+		 * is in the delayed allocation extent on which we sit
+		 * but before our buffer starts.
+		 */
+
+		nimaps = 0;
+		while (nimaps == 0) {
+			tp = xfs_trans_alloc(mp, XFS_TRANS_STRAT_WRITE);
+			tp->t_flags |= XFS_TRANS_RESERVE;
+			nres = XFS_EXTENTADD_SPACE_RES(mp, XFS_DATA_FORK);
+			error = xfs_trans_reserve(tp, nres,
+					XFS_WRITE_LOG_RES(mp),
+					0, XFS_TRANS_PERM_LOG_RES,
+					XFS_WRITE_LOG_COUNT);
+			if (error) {
+				xfs_trans_cancel(tp, 0);
+				return XFS_ERROR(error);
+			}
+			xfs_ilock(ip, XFS_ILOCK_EXCL);
+			xfs_trans_ijoin(tp, ip);
+
+			xfs_bmap_init(&free_list, &first_block);
+
+			/*
+			 * it is possible that the extents have changed since
+			 * we did the read call as we dropped the ilock for a
+			 * while. We have to be careful about truncates or hole
+			 * punchs here - we are not allowed to allocate
+			 * non-delalloc blocks here.
+			 *
+			 * The only protection against truncation is the pages
+			 * for the range we are being asked to convert are
+			 * locked and hence a truncate will block on them
+			 * first.
+			 *
+			 * As a result, if we go beyond the range we really
+			 * need and hit an delalloc extent boundary followed by
+			 * a hole while we have excess blocks in the map, we
+			 * will fill the hole incorrectly and overrun the
+			 * transaction reservation.
+			 *
+			 * Using a single map prevents this as we are forced to
+			 * check each map we look for overlap with the desired
+			 * range and abort as soon as we find it. Also, given
+			 * that we only return a single map, having one beyond
+			 * what we can return is probably a bit silly.
+			 *
+			 * We also need to check that we don't go beyond EOF;
+			 * this is a truncate optimisation as a truncate sets
+			 * the new file size before block on the pages we
+			 * currently have locked under writeback. Because they
+			 * are about to be tossed, we don't need to write them
+			 * back....
+			 */
+			nimaps = 1;
+			end_fsb = XFS_B_TO_FSB(mp, ip->i_size);
+			error = xfs_bmap_last_offset(NULL, ip, &last_block,
+							XFS_DATA_FORK);
+			if (error)
+				goto trans_cancel;
+
+			last_block = XFS_FILEOFF_MAX(last_block, end_fsb);
+			if ((map_start_fsb + count_fsb) > last_block) {
+				count_fsb = last_block - map_start_fsb;
+				if (count_fsb == 0) {
+					error = EAGAIN;
+					goto trans_cancel;
+				}
+			}
+
+			/*
+			 * Go get the actual blocks.
+	 	 	 *
+			 * From this point onwards we overwrite the imap
+			 * pointer that the caller gave to us.
+			 */
+			error = xfs_bmapi(tp, ip, map_start_fsb, count_fsb,
+					XFS_BMAPI_WRITE, &first_block, 1,
+					imap, &nimaps, &free_list);
+			if (error)
+				goto trans_cancel;
+
+			error = xfs_bmap_finish(&tp, &free_list, &committed);
+			if (error)
+				goto trans_cancel;
+
+			error = xfs_trans_commit(tp, XFS_TRANS_RELEASE_LOG_RES);
+			if (error)
+				goto error0;
+
+			xfs_iunlock(ip, XFS_ILOCK_EXCL);
+		}
+
+		/*
+		 * See if we were able to allocate an extent that
+		 * covers at least part of the callers request
+		 */
+		if (!(imap->br_startblock || XFS_IS_REALTIME_INODE(ip)))
+			return xfs_alert_fsblock_zero(ip, imap);
+
+		if ((offset_fsb >= imap->br_startoff) &&
+		    (offset_fsb < (imap->br_startoff +
+				   imap->br_blockcount))) {
+			XFS_STATS_INC(xs_xstrat_quick);
+			return 0;
+		}
+
+		/*
+		 * So far we have not mapped the requested part of the
+		 * file, just surrounding data, try again.
+		 */
+		count_fsb -= imap->br_blockcount;
+		map_start_fsb = imap->br_startoff + imap->br_blockcount;
+	}
+
+trans_cancel:
+	xfs_bmap_cancel(&free_list);
+	xfs_trans_cancel(tp, XFS_TRANS_RELEASE_LOG_RES | XFS_TRANS_ABORT);
+error0:
+	xfs_iunlock(ip, XFS_ILOCK_EXCL);
+	return XFS_ERROR(error);
+}
+
+int
+xfs_iomap_write_unwritten(
+	xfs_inode_t	*ip,
+	xfs_off_t	offset,
+	size_t		count)
+{
+	xfs_mount_t	*mp = ip->i_mount;
+	xfs_fileoff_t	offset_fsb;
+	xfs_filblks_t	count_fsb;
+	xfs_filblks_t	numblks_fsb;
+	xfs_fsblock_t	firstfsb;
+	int		nimaps;
+	xfs_trans_t	*tp;
+	xfs_bmbt_irec_t imap;
+	xfs_bmap_free_t free_list;
+	uint		resblks;
+	int		committed;
+	int		error;
+
+	trace_xfs_unwritten_convert(ip, offset, count);
+
+	offset_fsb = XFS_B_TO_FSBT(mp, offset);
+	count_fsb = XFS_B_TO_FSB(mp, (xfs_ufsize_t)offset + count);
+	count_fsb = (xfs_filblks_t)(count_fsb - offset_fsb);
+
+	/*
+	 * Reserve enough blocks in this transaction for two complete extent
+	 * btree splits.  We may be converting the middle part of an unwritten
+	 * extent and in this case we will insert two new extents in the btree
+	 * each of which could cause a full split.
+	 *
+	 * This reservation amount will be used in the first call to
+	 * xfs_bmbt_split() to select an AG with enough space to satisfy the
+	 * rest of the operation.
+	 */
+	resblks = XFS_DIOSTRAT_SPACE_RES(mp, 0) << 1;
+
+	do {
+		/*
+		 * set up a transaction to convert the range of extents
+		 * from unwritten to real. Do allocations in a loop until
+		 * we have covered the range passed in.
+		 *
+		 * Note that we open code the transaction allocation here
+		 * to pass KM_NOFS--we can't risk to recursing back into
+		 * the filesystem here as we might be asked to write out
+		 * the same inode that we complete here and might deadlock
+		 * on the iolock.
+		 */
+		xfs_wait_for_freeze(mp, SB_FREEZE_TRANS);
+		tp = _xfs_trans_alloc(mp, XFS_TRANS_STRAT_WRITE, KM_NOFS);
+		tp->t_flags |= XFS_TRANS_RESERVE;
+		error = xfs_trans_reserve(tp, resblks,
+				XFS_WRITE_LOG_RES(mp), 0,
+				XFS_TRANS_PERM_LOG_RES,
+				XFS_WRITE_LOG_COUNT);
+		if (error) {
+			xfs_trans_cancel(tp, 0);
+			return XFS_ERROR(error);
+		}
+
+		xfs_ilock(ip, XFS_ILOCK_EXCL);
+		xfs_trans_ijoin(tp, ip);
+
+		/*
+		 * Modify the unwritten extent state of the buffer.
+		 */
+		xfs_bmap_init(&free_list, &firstfsb);
+		nimaps = 1;
+		error = xfs_bmapi(tp, ip, offset_fsb, count_fsb,
+				  XFS_BMAPI_WRITE|XFS_BMAPI_CONVERT, &firstfsb,
+				  1, &imap, &nimaps, &free_list);
+		if (error)
+			goto error_on_bmapi_transaction;
+
+		error = xfs_bmap_finish(&(tp), &(free_list), &committed);
+		if (error)
+			goto error_on_bmapi_transaction;
+
+		error = xfs_trans_commit(tp, XFS_TRANS_RELEASE_LOG_RES);
+		xfs_iunlock(ip, XFS_ILOCK_EXCL);
+		if (error)
+			return XFS_ERROR(error);
+
+		if (!(imap.br_startblock || XFS_IS_REALTIME_INODE(ip)))
+			return xfs_alert_fsblock_zero(ip, &imap);
+
+		if ((numblks_fsb = imap.br_blockcount) == 0) {
+			/*
+			 * The numblks_fsb value should always get
+			 * smaller, otherwise the loop is stuck.
+			 */
+			ASSERT(imap.br_blockcount);
+			break;
+		}
+		offset_fsb += numblks_fsb;
+		count_fsb -= numblks_fsb;
+	} while (count_fsb > 0);
+
+	return 0;
+
+error_on_bmapi_transaction:
+	xfs_bmap_cancel(&free_list);
+	xfs_trans_cancel(tp, (XFS_TRANS_RELEASE_LOG_RES | XFS_TRANS_ABORT));
+	xfs_iunlock(ip, XFS_ILOCK_EXCL);
+	return XFS_ERROR(error);
+}
diff --git a/fs/xfs/xfs_iomap.h b/fs/xfs/xfs_iomap.h
new file mode 100644
index 00000000..80615760
--- /dev/null
+++ b/fs/xfs/xfs_iomap.h
@@ -0,0 +1,32 @@
+/*
+ * Copyright (c) 2003-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_IOMAP_H__
+#define __XFS_IOMAP_H__
+
+struct xfs_inode;
+struct xfs_bmbt_irec;
+
+extern int xfs_iomap_write_direct(struct xfs_inode *, xfs_off_t, size_t,
+			struct xfs_bmbt_irec *, int);
+extern int xfs_iomap_write_delay(struct xfs_inode *, xfs_off_t, size_t,
+			struct xfs_bmbt_irec *);
+extern int xfs_iomap_write_allocate(struct xfs_inode *, xfs_off_t, size_t,
+			struct xfs_bmbt_irec *);
+extern int xfs_iomap_write_unwritten(struct xfs_inode *, xfs_off_t, size_t);
+
+#endif /* __XFS_IOMAP_H__*/
diff --git a/fs/xfs/xfs_itable.c b/fs/xfs/xfs_itable.c
new file mode 100644
index 00000000..751e94fe
--- /dev/null
+++ b/fs/xfs/xfs_itable.c
@@ -0,0 +1,736 @@
+/*
+ * Copyright (c) 2000-2002,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_mount.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_alloc_btree.h"
+#include "xfs_ialloc_btree.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_ialloc.h"
+#include "xfs_itable.h"
+#include "xfs_error.h"
+#include "xfs_btree.h"
+#include "xfs_trace.h"
+
+STATIC int
+xfs_internal_inum(
+	xfs_mount_t	*mp,
+	xfs_ino_t	ino)
+{
+	return (ino == mp->m_sb.sb_rbmino || ino == mp->m_sb.sb_rsumino ||
+		(xfs_sb_version_hasquota(&mp->m_sb) &&
+		 (ino == mp->m_sb.sb_uquotino || ino == mp->m_sb.sb_gquotino)));
+}
+
+/*
+ * Return stat information for one inode.
+ * Return 0 if ok, else errno.
+ */
+int
+xfs_bulkstat_one_int(
+	struct xfs_mount	*mp,		/* mount point for filesystem */
+	xfs_ino_t		ino,		/* inode to get data for */
+	void __user		*buffer,	/* buffer to place output in */
+	int			ubsize,		/* size of buffer */
+	bulkstat_one_fmt_pf	formatter,	/* formatter, copy to user */
+	int			*ubused,	/* bytes used by me */
+	int			*stat)		/* BULKSTAT_RV_... */
+{
+	struct xfs_icdinode	*dic;		/* dinode core info pointer */
+	struct xfs_inode	*ip;		/* incore inode pointer */
+	struct inode		*inode;
+	struct xfs_bstat	*buf;		/* return buffer */
+	int			error = 0;	/* error value */
+
+	*stat = BULKSTAT_RV_NOTHING;
+
+	if (!buffer || xfs_internal_inum(mp, ino))
+		return XFS_ERROR(EINVAL);
+
+	buf = kmem_alloc(sizeof(*buf), KM_SLEEP | KM_MAYFAIL);
+	if (!buf)
+		return XFS_ERROR(ENOMEM);
+
+	error = xfs_iget(mp, NULL, ino,
+			 XFS_IGET_UNTRUSTED, XFS_ILOCK_SHARED, &ip);
+	if (error) {
+		*stat = BULKSTAT_RV_NOTHING;
+		goto out_free;
+	}
+
+	ASSERT(ip != NULL);
+	ASSERT(ip->i_imap.im_blkno != 0);
+
+	dic = &ip->i_d;
+	inode = VFS_I(ip);
+
+	/* xfs_iget returns the following without needing
+	 * further change.
+	 */
+	buf->bs_nlink = dic->di_nlink;
+	buf->bs_projid_lo = dic->di_projid_lo;
+	buf->bs_projid_hi = dic->di_projid_hi;
+	buf->bs_ino = ino;
+	buf->bs_mode = dic->di_mode;
+	buf->bs_uid = dic->di_uid;
+	buf->bs_gid = dic->di_gid;
+	buf->bs_size = dic->di_size;
+
+	/*
+	 * We need to read the timestamps from the Linux inode because
+	 * the VFS keeps writing directly into the inode structure instead
+	 * of telling us about the updates.
+	 */
+	buf->bs_atime.tv_sec = inode->i_atime.tv_sec;
+	buf->bs_atime.tv_nsec = inode->i_atime.tv_nsec;
+	buf->bs_mtime.tv_sec = inode->i_mtime.tv_sec;
+	buf->bs_mtime.tv_nsec = inode->i_mtime.tv_nsec;
+	buf->bs_ctime.tv_sec = inode->i_ctime.tv_sec;
+	buf->bs_ctime.tv_nsec = inode->i_ctime.tv_nsec;
+
+	buf->bs_xflags = xfs_ip2xflags(ip);
+	buf->bs_extsize = dic->di_extsize << mp->m_sb.sb_blocklog;
+	buf->bs_extents = dic->di_nextents;
+	buf->bs_gen = dic->di_gen;
+	memset(buf->bs_pad, 0, sizeof(buf->bs_pad));
+	buf->bs_dmevmask = dic->di_dmevmask;
+	buf->bs_dmstate = dic->di_dmstate;
+	buf->bs_aextents = dic->di_anextents;
+	buf->bs_forkoff = XFS_IFORK_BOFF(ip);
+
+	switch (dic->di_format) {
+	case XFS_DINODE_FMT_DEV:
+		buf->bs_rdev = ip->i_df.if_u2.if_rdev;
+		buf->bs_blksize = BLKDEV_IOSIZE;
+		buf->bs_blocks = 0;
+		break;
+	case XFS_DINODE_FMT_LOCAL:
+	case XFS_DINODE_FMT_UUID:
+		buf->bs_rdev = 0;
+		buf->bs_blksize = mp->m_sb.sb_blocksize;
+		buf->bs_blocks = 0;
+		break;
+	case XFS_DINODE_FMT_EXTENTS:
+	case XFS_DINODE_FMT_BTREE:
+		buf->bs_rdev = 0;
+		buf->bs_blksize = mp->m_sb.sb_blocksize;
+		buf->bs_blocks = dic->di_nblocks + ip->i_delayed_blks;
+		break;
+	}
+	xfs_iunlock(ip, XFS_ILOCK_SHARED);
+	IRELE(ip);
+
+	error = formatter(buffer, ubsize, ubused, buf);
+
+	if (!error)
+		*stat = BULKSTAT_RV_DIDONE;
+
+ out_free:
+	kmem_free(buf);
+	return error;
+}
+
+/* Return 0 on success or positive error */
+STATIC int
+xfs_bulkstat_one_fmt(
+	void			__user *ubuffer,
+	int			ubsize,
+	int			*ubused,
+	const xfs_bstat_t	*buffer)
+{
+	if (ubsize < sizeof(*buffer))
+		return XFS_ERROR(ENOMEM);
+	if (copy_to_user(ubuffer, buffer, sizeof(*buffer)))
+		return XFS_ERROR(EFAULT);
+	if (ubused)
+		*ubused = sizeof(*buffer);
+	return 0;
+}
+
+int
+xfs_bulkstat_one(
+	xfs_mount_t	*mp,		/* mount point for filesystem */
+	xfs_ino_t	ino,		/* inode number to get data for */
+	void		__user *buffer,	/* buffer to place output in */
+	int		ubsize,		/* size of buffer */
+	int		*ubused,	/* bytes used by me */
+	int		*stat)		/* BULKSTAT_RV_... */
+{
+	return xfs_bulkstat_one_int(mp, ino, buffer, ubsize,
+				    xfs_bulkstat_one_fmt, ubused, stat);
+}
+
+#define XFS_BULKSTAT_UBLEFT(ubleft)	((ubleft) >= statstruct_size)
+
+/*
+ * Return stat information in bulk (by-inode) for the filesystem.
+ */
+int					/* error status */
+xfs_bulkstat(
+	xfs_mount_t		*mp,	/* mount point for filesystem */
+	xfs_ino_t		*lastinop, /* last inode returned */
+	int			*ubcountp, /* size of buffer/count returned */
+	bulkstat_one_pf		formatter, /* func that'd fill a single buf */
+	size_t			statstruct_size, /* sizeof struct filling */
+	char			__user *ubuffer, /* buffer with inode stats */
+	int			*done)	/* 1 if there are more stats to get */
+{
+	xfs_agblock_t		agbno=0;/* allocation group block number */
+	xfs_buf_t		*agbp;	/* agi header buffer */
+	xfs_agi_t		*agi;	/* agi header data */
+	xfs_agino_t		agino;	/* inode # in allocation group */
+	xfs_agnumber_t		agno;	/* allocation group number */
+	int			chunkidx; /* current index into inode chunk */
+	int			clustidx; /* current index into inode cluster */
+	xfs_btree_cur_t		*cur;	/* btree cursor for ialloc btree */
+	int			end_of_ag; /* set if we've seen the ag end */
+	int			error;	/* error code */
+	int                     fmterror;/* bulkstat formatter result */
+	int			i;	/* loop index */
+	int			icount;	/* count of inodes good in irbuf */
+	size_t			irbsize; /* size of irec buffer in bytes */
+	xfs_ino_t		ino;	/* inode number (filesystem) */
+	xfs_inobt_rec_incore_t	*irbp;	/* current irec buffer pointer */
+	xfs_inobt_rec_incore_t	*irbuf;	/* start of irec buffer */
+	xfs_inobt_rec_incore_t	*irbufend; /* end of good irec buffer entries */
+	xfs_ino_t		lastino; /* last inode number returned */
+	int			nbcluster; /* # of blocks in a cluster */
+	int			nicluster; /* # of inodes in a cluster */
+	int			nimask;	/* mask for inode clusters */
+	int			nirbuf;	/* size of irbuf */
+	int			rval;	/* return value error code */
+	int			tmp;	/* result value from btree calls */
+	int			ubcount; /* size of user's buffer */
+	int			ubleft;	/* bytes left in user's buffer */
+	char			__user *ubufp;	/* pointer into user's buffer */
+	int			ubelem;	/* spaces used in user's buffer */
+	int			ubused;	/* bytes used by formatter */
+	xfs_buf_t		*bp;	/* ptr to on-disk inode cluster buf */
+
+	/*
+	 * Get the last inode value, see if there's nothing to do.
+	 */
+	ino = (xfs_ino_t)*lastinop;
+	lastino = ino;
+	agno = XFS_INO_TO_AGNO(mp, ino);
+	agino = XFS_INO_TO_AGINO(mp, ino);
+	if (agno >= mp->m_sb.sb_agcount ||
+	    ino != XFS_AGINO_TO_INO(mp, agno, agino)) {
+		*done = 1;
+		*ubcountp = 0;
+		return 0;
+	}
+	if (!ubcountp || *ubcountp <= 0) {
+		return EINVAL;
+	}
+	ubcount = *ubcountp; /* statstruct's */
+	ubleft = ubcount * statstruct_size; /* bytes */
+	*ubcountp = ubelem = 0;
+	*done = 0;
+	fmterror = 0;
+	ubufp = ubuffer;
+	nicluster = mp->m_sb.sb_blocksize >= XFS_INODE_CLUSTER_SIZE(mp) ?
+		mp->m_sb.sb_inopblock :
+		(XFS_INODE_CLUSTER_SIZE(mp) >> mp->m_sb.sb_inodelog);
+	nimask = ~(nicluster - 1);
+	nbcluster = nicluster >> mp->m_sb.sb_inopblog;
+	irbuf = kmem_zalloc_greedy(&irbsize, PAGE_SIZE, PAGE_SIZE * 4);
+	if (!irbuf)
+		return ENOMEM;
+
+	nirbuf = irbsize / sizeof(*irbuf);
+
+	/*
+	 * Loop over the allocation groups, starting from the last
+	 * inode returned; 0 means start of the allocation group.
+	 */
+	rval = 0;
+	while (XFS_BULKSTAT_UBLEFT(ubleft) && agno < mp->m_sb.sb_agcount) {
+		cond_resched();
+		bp = NULL;
+		error = xfs_ialloc_read_agi(mp, NULL, agno, &agbp);
+		if (error) {
+			/*
+			 * Skip this allocation group and go to the next one.
+			 */
+			agno++;
+			agino = 0;
+			continue;
+		}
+		agi = XFS_BUF_TO_AGI(agbp);
+		/*
+		 * Allocate and initialize a btree cursor for ialloc btree.
+		 */
+		cur = xfs_inobt_init_cursor(mp, NULL, agbp, agno);
+		irbp = irbuf;
+		irbufend = irbuf + nirbuf;
+		end_of_ag = 0;
+		/*
+		 * If we're returning in the middle of an allocation group,
+		 * we need to get the remainder of the chunk we're in.
+		 */
+		if (agino > 0) {
+			xfs_inobt_rec_incore_t r;
+
+			/*
+			 * Lookup the inode chunk that this inode lives in.
+			 */
+			error = xfs_inobt_lookup(cur, agino, XFS_LOOKUP_LE,
+						 &tmp);
+			if (!error &&	/* no I/O error */
+			    tmp &&	/* lookup succeeded */
+					/* got the record, should always work */
+			    !(error = xfs_inobt_get_rec(cur, &r, &i)) &&
+			    i == 1 &&
+					/* this is the right chunk */
+			    agino < r.ir_startino + XFS_INODES_PER_CHUNK &&
+					/* lastino was not last in chunk */
+			    (chunkidx = agino - r.ir_startino + 1) <
+				    XFS_INODES_PER_CHUNK &&
+					/* there are some left allocated */
+			    xfs_inobt_maskn(chunkidx,
+				    XFS_INODES_PER_CHUNK - chunkidx) &
+				    ~r.ir_free) {
+				/*
+				 * Grab the chunk record.  Mark all the
+				 * uninteresting inodes (because they're
+				 * before our start point) free.
+				 */
+				for (i = 0; i < chunkidx; i++) {
+					if (XFS_INOBT_MASK(i) & ~r.ir_free)
+						r.ir_freecount++;
+				}
+				r.ir_free |= xfs_inobt_maskn(0, chunkidx);
+				irbp->ir_startino = r.ir_startino;
+				irbp->ir_freecount = r.ir_freecount;
+				irbp->ir_free = r.ir_free;
+				irbp++;
+				agino = r.ir_startino + XFS_INODES_PER_CHUNK;
+				icount = XFS_INODES_PER_CHUNK - r.ir_freecount;
+			} else {
+				/*
+				 * If any of those tests failed, bump the
+				 * inode number (just in case).
+				 */
+				agino++;
+				icount = 0;
+			}
+			/*
+			 * In any case, increment to the next record.
+			 */
+			if (!error)
+				error = xfs_btree_increment(cur, 0, &tmp);
+		} else {
+			/*
+			 * Start of ag.  Lookup the first inode chunk.
+			 */
+			error = xfs_inobt_lookup(cur, 0, XFS_LOOKUP_GE, &tmp);
+			icount = 0;
+		}
+		/*
+		 * Loop through inode btree records in this ag,
+		 * until we run out of inodes or space in the buffer.
+		 */
+		while (irbp < irbufend && icount < ubcount) {
+			xfs_inobt_rec_incore_t r;
+
+			/*
+			 * Loop as long as we're unable to read the
+			 * inode btree.
+			 */
+			while (error) {
+				agino += XFS_INODES_PER_CHUNK;
+				if (XFS_AGINO_TO_AGBNO(mp, agino) >=
+						be32_to_cpu(agi->agi_length))
+					break;
+				error = xfs_inobt_lookup(cur, agino,
+							 XFS_LOOKUP_GE, &tmp);
+				cond_resched();
+			}
+			/*
+			 * If ran off the end of the ag either with an error,
+			 * or the normal way, set end and stop collecting.
+			 */
+			if (error) {
+				end_of_ag = 1;
+				break;
+			}
+
+			error = xfs_inobt_get_rec(cur, &r, &i);
+			if (error || i == 0) {
+				end_of_ag = 1;
+				break;
+			}
+
+			/*
+			 * If this chunk has any allocated inodes, save it.
+			 * Also start read-ahead now for this chunk.
+			 */
+			if (r.ir_freecount < XFS_INODES_PER_CHUNK) {
+				/*
+				 * Loop over all clusters in the next chunk.
+				 * Do a readahead if there are any allocated
+				 * inodes in that cluster.
+				 */
+				agbno = XFS_AGINO_TO_AGBNO(mp, r.ir_startino);
+				for (chunkidx = 0;
+				     chunkidx < XFS_INODES_PER_CHUNK;
+				     chunkidx += nicluster,
+				     agbno += nbcluster) {
+					if (xfs_inobt_maskn(chunkidx, nicluster)
+							& ~r.ir_free)
+						xfs_btree_reada_bufs(mp, agno,
+							agbno, nbcluster);
+				}
+				irbp->ir_startino = r.ir_startino;
+				irbp->ir_freecount = r.ir_freecount;
+				irbp->ir_free = r.ir_free;
+				irbp++;
+				icount += XFS_INODES_PER_CHUNK - r.ir_freecount;
+			}
+			/*
+			 * Set agino to after this chunk and bump the cursor.
+			 */
+			agino = r.ir_startino + XFS_INODES_PER_CHUNK;
+			error = xfs_btree_increment(cur, 0, &tmp);
+			cond_resched();
+		}
+		/*
+		 * Drop the btree buffers and the agi buffer.
+		 * We can't hold any of the locks these represent
+		 * when calling iget.
+		 */
+		xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+		xfs_buf_relse(agbp);
+		/*
+		 * Now format all the good inodes into the user's buffer.
+		 */
+		irbufend = irbp;
+		for (irbp = irbuf;
+		     irbp < irbufend && XFS_BULKSTAT_UBLEFT(ubleft); irbp++) {
+			/*
+			 * Now process this chunk of inodes.
+			 */
+			for (agino = irbp->ir_startino, chunkidx = clustidx = 0;
+			     XFS_BULKSTAT_UBLEFT(ubleft) &&
+				irbp->ir_freecount < XFS_INODES_PER_CHUNK;
+			     chunkidx++, clustidx++, agino++) {
+				ASSERT(chunkidx < XFS_INODES_PER_CHUNK);
+				/*
+				 * Recompute agbno if this is the
+				 * first inode of the cluster.
+				 *
+				 * Careful with clustidx.   There can be
+				 * multiple clusters per chunk, a single
+				 * cluster per chunk or a cluster that has
+				 * inodes represented from several different
+				 * chunks (if blocksize is large).
+				 *
+				 * Because of this, the starting clustidx is
+				 * initialized to zero in this loop but must
+				 * later be reset after reading in the cluster
+				 * buffer.
+				 */
+				if ((chunkidx & (nicluster - 1)) == 0) {
+					agbno = XFS_AGINO_TO_AGBNO(mp,
+							irbp->ir_startino) +
+						((chunkidx & nimask) >>
+						 mp->m_sb.sb_inopblog);
+				}
+				ino = XFS_AGINO_TO_INO(mp, agno, agino);
+				/*
+				 * Skip if this inode is free.
+				 */
+				if (XFS_INOBT_MASK(chunkidx) & irbp->ir_free) {
+					lastino = ino;
+					continue;
+				}
+				/*
+				 * Count used inodes as free so we can tell
+				 * when the chunk is used up.
+				 */
+				irbp->ir_freecount++;
+
+				/*
+				 * Get the inode and fill in a single buffer.
+				 */
+				ubused = statstruct_size;
+				error = formatter(mp, ino, ubufp, ubleft,
+						  &ubused, &fmterror);
+				if (fmterror == BULKSTAT_RV_NOTHING) {
+					if (error && error != ENOENT &&
+						error != EINVAL) {
+						ubleft = 0;
+						rval = error;
+						break;
+					}
+					lastino = ino;
+					continue;
+				}
+				if (fmterror == BULKSTAT_RV_GIVEUP) {
+					ubleft = 0;
+					ASSERT(error);
+					rval = error;
+					break;
+				}
+				if (ubufp)
+					ubufp += ubused;
+				ubleft -= ubused;
+				ubelem++;
+				lastino = ino;
+			}
+
+			cond_resched();
+		}
+
+		if (bp)
+			xfs_buf_relse(bp);
+
+		/*
+		 * Set up for the next loop iteration.
+		 */
+		if (XFS_BULKSTAT_UBLEFT(ubleft)) {
+			if (end_of_ag) {
+				agno++;
+				agino = 0;
+			} else
+				agino = XFS_INO_TO_AGINO(mp, lastino);
+		} else
+			break;
+	}
+	/*
+	 * Done, we're either out of filesystem or space to put the data.
+	 */
+	kmem_free_large(irbuf);
+	*ubcountp = ubelem;
+	/*
+	 * Found some inodes, return them now and return the error next time.
+	 */
+	if (ubelem)
+		rval = 0;
+	if (agno >= mp->m_sb.sb_agcount) {
+		/*
+		 * If we ran out of filesystem, mark lastino as off
+		 * the end of the filesystem, so the next call
+		 * will return immediately.
+		 */
+		*lastinop = (xfs_ino_t)XFS_AGINO_TO_INO(mp, agno, 0);
+		*done = 1;
+	} else
+		*lastinop = (xfs_ino_t)lastino;
+
+	return rval;
+}
+
+/*
+ * Return stat information in bulk (by-inode) for the filesystem.
+ * Special case for non-sequential one inode bulkstat.
+ */
+int					/* error status */
+xfs_bulkstat_single(
+	xfs_mount_t		*mp,	/* mount point for filesystem */
+	xfs_ino_t		*lastinop, /* inode to return */
+	char			__user *buffer, /* buffer with inode stats */
+	int			*done)	/* 1 if there are more stats to get */
+{
+	int			count;	/* count value for bulkstat call */
+	int			error;	/* return value */
+	xfs_ino_t		ino;	/* filesystem inode number */
+	int			res;	/* result from bs1 */
+
+	/*
+	 * note that requesting valid inode numbers which are not allocated
+	 * to inodes will most likely cause xfs_itobp to generate warning
+	 * messages about bad magic numbers. This is ok. The fact that
+	 * the inode isn't actually an inode is handled by the
+	 * error check below. Done this way to make the usual case faster
+	 * at the expense of the error case.
+	 */
+
+	ino = (xfs_ino_t)*lastinop;
+	error = xfs_bulkstat_one(mp, ino, buffer, sizeof(xfs_bstat_t), 0, &res);
+	if (error) {
+		/*
+		 * Special case way failed, do it the "long" way
+		 * to see if that works.
+		 */
+		(*lastinop)--;
+		count = 1;
+		if (xfs_bulkstat(mp, lastinop, &count, xfs_bulkstat_one,
+				sizeof(xfs_bstat_t), buffer, done))
+			return error;
+		if (count == 0 || (xfs_ino_t)*lastinop != ino)
+			return error == EFSCORRUPTED ?
+				XFS_ERROR(EINVAL) : error;
+		else
+			return 0;
+	}
+	*done = 0;
+	return 0;
+}
+
+int
+xfs_inumbers_fmt(
+	void			__user *ubuffer, /* buffer to write to */
+	const xfs_inogrp_t	*buffer,	/* buffer to read from */
+	long			count,		/* # of elements to read */
+	long			*written)	/* # of bytes written */
+{
+	if (copy_to_user(ubuffer, buffer, count * sizeof(*buffer)))
+		return -EFAULT;
+	*written = count * sizeof(*buffer);
+	return 0;
+}
+
+/*
+ * Return inode number table for the filesystem.
+ */
+int					/* error status */
+xfs_inumbers(
+	xfs_mount_t	*mp,		/* mount point for filesystem */
+	xfs_ino_t	*lastino,	/* last inode returned */
+	int		*count,		/* size of buffer/count returned */
+	void		__user *ubuffer,/* buffer with inode descriptions */
+	inumbers_fmt_pf	formatter)
+{
+	xfs_buf_t	*agbp;
+	xfs_agino_t	agino;
+	xfs_agnumber_t	agno;
+	int		bcount;
+	xfs_inogrp_t	*buffer;
+	int		bufidx;
+	xfs_btree_cur_t	*cur;
+	int		error;
+	xfs_inobt_rec_incore_t r;
+	int		i;
+	xfs_ino_t	ino;
+	int		left;
+	int		tmp;
+
+	ino = (xfs_ino_t)*lastino;
+	agno = XFS_INO_TO_AGNO(mp, ino);
+	agino = XFS_INO_TO_AGINO(mp, ino);
+	left = *count;
+	*count = 0;
+	bcount = MIN(left, (int)(PAGE_SIZE / sizeof(*buffer)));
+	buffer = kmem_alloc(bcount * sizeof(*buffer), KM_SLEEP);
+	error = bufidx = 0;
+	cur = NULL;
+	agbp = NULL;
+	while (left > 0 && agno < mp->m_sb.sb_agcount) {
+		if (agbp == NULL) {
+			error = xfs_ialloc_read_agi(mp, NULL, agno, &agbp);
+			if (error) {
+				/*
+				 * If we can't read the AGI of this ag,
+				 * then just skip to the next one.
+				 */
+				ASSERT(cur == NULL);
+				agbp = NULL;
+				agno++;
+				agino = 0;
+				continue;
+			}
+			cur = xfs_inobt_init_cursor(mp, NULL, agbp, agno);
+			error = xfs_inobt_lookup(cur, agino, XFS_LOOKUP_GE,
+						 &tmp);
+			if (error) {
+				xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
+				cur = NULL;
+				xfs_buf_relse(agbp);
+				agbp = NULL;
+				/*
+				 * Move up the last inode in the current
+				 * chunk.  The lookup_ge will always get
+				 * us the first inode in the next chunk.
+				 */
+				agino += XFS_INODES_PER_CHUNK - 1;
+				continue;
+			}
+		}
+		error = xfs_inobt_get_rec(cur, &r, &i);
+		if (error || i == 0) {
+			xfs_buf_relse(agbp);
+			agbp = NULL;
+			xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+			cur = NULL;
+			agno++;
+			agino = 0;
+			continue;
+		}
+		agino = r.ir_startino + XFS_INODES_PER_CHUNK - 1;
+		buffer[bufidx].xi_startino =
+			XFS_AGINO_TO_INO(mp, agno, r.ir_startino);
+		buffer[bufidx].xi_alloccount =
+			XFS_INODES_PER_CHUNK - r.ir_freecount;
+		buffer[bufidx].xi_allocmask = ~r.ir_free;
+		bufidx++;
+		left--;
+		if (bufidx == bcount) {
+			long written;
+			if (formatter(ubuffer, buffer, bufidx, &written)) {
+				error = XFS_ERROR(EFAULT);
+				break;
+			}
+			ubuffer += written;
+			*count += bufidx;
+			bufidx = 0;
+		}
+		if (left) {
+			error = xfs_btree_increment(cur, 0, &tmp);
+			if (error) {
+				xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
+				cur = NULL;
+				xfs_buf_relse(agbp);
+				agbp = NULL;
+				/*
+				 * The agino value has already been bumped.
+				 * Just try to skip up to it.
+				 */
+				agino += XFS_INODES_PER_CHUNK;
+				continue;
+			}
+		}
+	}
+	if (!error) {
+		if (bufidx) {
+			long written;
+			if (formatter(ubuffer, buffer, bufidx, &written))
+				error = XFS_ERROR(EFAULT);
+			else
+				*count += bufidx;
+		}
+		*lastino = XFS_AGINO_TO_INO(mp, agno, agino);
+	}
+	kmem_free(buffer);
+	if (cur)
+		xfs_btree_del_cursor(cur, (error ? XFS_BTREE_ERROR :
+					   XFS_BTREE_NOERROR));
+	if (agbp)
+		xfs_buf_relse(agbp);
+	return error;
+}
diff --git a/fs/xfs/xfs_itable.h b/fs/xfs/xfs_itable.h
new file mode 100644
index 00000000..97295d91
--- /dev/null
+++ b/fs/xfs/xfs_itable.h
@@ -0,0 +1,106 @@
+/*
+ * Copyright (c) 2000-2001 Silicon Graphics, Inc.  All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_ITABLE_H__
+#define	__XFS_ITABLE_H__
+
+/*
+ * xfs_bulkstat() is used to fill in xfs_bstat structures as well as dm_stat
+ * structures (by the dmi library). This is a pointer to a formatter function
+ * that will iget the inode and fill in the appropriate structure.
+ * see xfs_bulkstat_one() and xfs_dm_bulkstat_one() in dmapi_xfs.c
+ */
+typedef int (*bulkstat_one_pf)(struct xfs_mount	*mp,
+			       xfs_ino_t	ino,
+			       void		__user *buffer,
+			       int		ubsize,
+			       int		*ubused,
+			       int		*stat);
+
+/*
+ * Values for stat return value.
+ */
+#define BULKSTAT_RV_NOTHING	0
+#define BULKSTAT_RV_DIDONE	1
+#define BULKSTAT_RV_GIVEUP	2
+
+/*
+ * Return stat information in bulk (by-inode) for the filesystem.
+ */
+int					/* error status */
+xfs_bulkstat(
+	xfs_mount_t	*mp,		/* mount point for filesystem */
+	xfs_ino_t	*lastino,	/* last inode returned */
+	int		*count,		/* size of buffer/count returned */
+	bulkstat_one_pf formatter,	/* func that'd fill a single buf */
+	size_t		statstruct_size,/* sizeof struct that we're filling */
+	char		__user *ubuffer,/* buffer with inode stats */
+	int		*done);		/* 1 if there are more stats to get */
+
+int
+xfs_bulkstat_single(
+	xfs_mount_t		*mp,
+	xfs_ino_t		*lastinop,
+	char			__user *buffer,
+	int			*done);
+
+typedef int (*bulkstat_one_fmt_pf)(  /* used size in bytes or negative error */
+	void			__user *ubuffer, /* buffer to write to */
+	int			ubsize,		 /* remaining user buffer sz */
+	int			*ubused,	 /* bytes used by formatter */
+	const xfs_bstat_t	*buffer);        /* buffer to read from */
+
+int
+xfs_bulkstat_one_int(
+	xfs_mount_t		*mp,
+	xfs_ino_t		ino,
+	void			__user *buffer,
+	int			ubsize,
+	bulkstat_one_fmt_pf	formatter,
+	int			*ubused,
+	int			*stat);
+
+int
+xfs_bulkstat_one(
+	xfs_mount_t		*mp,
+	xfs_ino_t		ino,
+	void			__user *buffer,
+	int			ubsize,
+	int			*ubused,
+	int			*stat);
+
+typedef int (*inumbers_fmt_pf)(
+	void			__user *ubuffer, /* buffer to write to */
+	const xfs_inogrp_t	*buffer,	/* buffer to read from */
+	long			count,		/* # of elements to read */
+	long			*written);	/* # of bytes written */
+
+int
+xfs_inumbers_fmt(
+	void			__user *ubuffer, /* buffer to write to */
+	const xfs_inogrp_t	*buffer,	/* buffer to read from */
+	long			count,		/* # of elements to read */
+	long			*written);	/* # of bytes written */
+
+int					/* error status */
+xfs_inumbers(
+	xfs_mount_t		*mp,	/* mount point for filesystem */
+	xfs_ino_t		*last,	/* last inode returned */
+	int			*count,	/* size of buffer/count returned */
+	void			__user *buffer, /* buffer with inode info */
+	inumbers_fmt_pf		formatter);
+
+#endif	/* __XFS_ITABLE_H__ */
diff --git a/fs/xfs/xfs_log.c b/fs/xfs/xfs_log.c
new file mode 100644
index 00000000..41d5b8f2
--- /dev/null
+++ b/fs/xfs/xfs_log.c
@@ -0,0 +1,3767 @@
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_mount.h"
+#include "xfs_error.h"
+#include "xfs_log_priv.h"
+#include "xfs_buf_item.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_alloc_btree.h"
+#include "xfs_ialloc_btree.h"
+#include "xfs_log_recover.h"
+#include "xfs_trans_priv.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_rw.h"
+#include "xfs_trace.h"
+
+kmem_zone_t	*xfs_log_ticket_zone;
+
+/* Local miscellaneous function prototypes */
+STATIC int	 xlog_commit_record(struct log *log, struct xlog_ticket *ticket,
+				    xlog_in_core_t **, xfs_lsn_t *);
+STATIC xlog_t *  xlog_alloc_log(xfs_mount_t	*mp,
+				xfs_buftarg_t	*log_target,
+				xfs_daddr_t	blk_offset,
+				int		num_bblks);
+STATIC int	 xlog_space_left(struct log *log, atomic64_t *head);
+STATIC int	 xlog_sync(xlog_t *log, xlog_in_core_t *iclog);
+STATIC void	 xlog_dealloc_log(xlog_t *log);
+
+/* local state machine functions */
+STATIC void xlog_state_done_syncing(xlog_in_core_t *iclog, int);
+STATIC void xlog_state_do_callback(xlog_t *log,int aborted, xlog_in_core_t *iclog);
+STATIC int  xlog_state_get_iclog_space(xlog_t		*log,
+				       int		len,
+				       xlog_in_core_t	**iclog,
+				       xlog_ticket_t	*ticket,
+				       int		*continued_write,
+				       int		*logoffsetp);
+STATIC int  xlog_state_release_iclog(xlog_t		*log,
+				     xlog_in_core_t	*iclog);
+STATIC void xlog_state_switch_iclogs(xlog_t		*log,
+				     xlog_in_core_t *iclog,
+				     int		eventual_size);
+STATIC void xlog_state_want_sync(xlog_t	*log, xlog_in_core_t *iclog);
+
+/* local functions to manipulate grant head */
+STATIC int  xlog_grant_log_space(xlog_t		*log,
+				 xlog_ticket_t	*xtic);
+STATIC void xlog_grant_push_ail(struct log	*log,
+				int		need_bytes);
+STATIC void xlog_regrant_reserve_log_space(xlog_t	 *log,
+					   xlog_ticket_t *ticket);
+STATIC int xlog_regrant_write_log_space(xlog_t		*log,
+					 xlog_ticket_t  *ticket);
+STATIC void xlog_ungrant_log_space(xlog_t	 *log,
+				   xlog_ticket_t *ticket);
+
+#if defined(DEBUG)
+STATIC void	xlog_verify_dest_ptr(xlog_t *log, char *ptr);
+STATIC void	xlog_verify_grant_tail(struct log *log);
+STATIC void	xlog_verify_iclog(xlog_t *log, xlog_in_core_t *iclog,
+				  int count, boolean_t syncing);
+STATIC void	xlog_verify_tail_lsn(xlog_t *log, xlog_in_core_t *iclog,
+				     xfs_lsn_t tail_lsn);
+#else
+#define xlog_verify_dest_ptr(a,b)
+#define xlog_verify_grant_tail(a)
+#define xlog_verify_iclog(a,b,c,d)
+#define xlog_verify_tail_lsn(a,b,c)
+#endif
+
+STATIC int	xlog_iclogs_empty(xlog_t *log);
+
+static void
+xlog_grant_sub_space(
+	struct log	*log,
+	atomic64_t	*head,
+	int		bytes)
+{
+	int64_t	head_val = atomic64_read(head);
+	int64_t new, old;
+
+	do {
+		int	cycle, space;
+
+		xlog_crack_grant_head_val(head_val, &cycle, &space);
+
+		space -= bytes;
+		if (space < 0) {
+			space += log->l_logsize;
+			cycle--;
+		}
+
+		old = head_val;
+		new = xlog_assign_grant_head_val(cycle, space);
+		head_val = atomic64_cmpxchg(head, old, new);
+	} while (head_val != old);
+}
+
+static void
+xlog_grant_add_space(
+	struct log	*log,
+	atomic64_t	*head,
+	int		bytes)
+{
+	int64_t	head_val = atomic64_read(head);
+	int64_t new, old;
+
+	do {
+		int		tmp;
+		int		cycle, space;
+
+		xlog_crack_grant_head_val(head_val, &cycle, &space);
+
+		tmp = log->l_logsize - space;
+		if (tmp > bytes)
+			space += bytes;
+		else {
+			space = bytes - tmp;
+			cycle++;
+		}
+
+		old = head_val;
+		new = xlog_assign_grant_head_val(cycle, space);
+		head_val = atomic64_cmpxchg(head, old, new);
+	} while (head_val != old);
+}
+
+static void
+xlog_tic_reset_res(xlog_ticket_t *tic)
+{
+	tic->t_res_num = 0;
+	tic->t_res_arr_sum = 0;
+	tic->t_res_num_ophdrs = 0;
+}
+
+static void
+xlog_tic_add_region(xlog_ticket_t *tic, uint len, uint type)
+{
+	if (tic->t_res_num == XLOG_TIC_LEN_MAX) {
+		/* add to overflow and start again */
+		tic->t_res_o_flow += tic->t_res_arr_sum;
+		tic->t_res_num = 0;
+		tic->t_res_arr_sum = 0;
+	}
+
+	tic->t_res_arr[tic->t_res_num].r_len = len;
+	tic->t_res_arr[tic->t_res_num].r_type = type;
+	tic->t_res_arr_sum += len;
+	tic->t_res_num++;
+}
+
+/*
+ * NOTES:
+ *
+ *	1. currblock field gets updated at startup and after in-core logs
+ *		marked as with WANT_SYNC.
+ */
+
+/*
+ * This routine is called when a user of a log manager ticket is done with
+ * the reservation.  If the ticket was ever used, then a commit record for
+ * the associated transaction is written out as a log operation header with
+ * no data.  The flag XLOG_TIC_INITED is set when the first write occurs with
+ * a given ticket.  If the ticket was one with a permanent reservation, then
+ * a few operations are done differently.  Permanent reservation tickets by
+ * default don't release the reservation.  They just commit the current
+ * transaction with the belief that the reservation is still needed.  A flag
+ * must be passed in before permanent reservations are actually released.
+ * When these type of tickets are not released, they need to be set into
+ * the inited state again.  By doing this, a start record will be written
+ * out when the next write occurs.
+ */
+xfs_lsn_t
+xfs_log_done(
+	struct xfs_mount	*mp,
+	struct xlog_ticket	*ticket,
+	struct xlog_in_core	**iclog,
+	uint			flags)
+{
+	struct log		*log = mp->m_log;
+	xfs_lsn_t		lsn = 0;
+
+	if (XLOG_FORCED_SHUTDOWN(log) ||
+	    /*
+	     * If nothing was ever written, don't write out commit record.
+	     * If we get an error, just continue and give back the log ticket.
+	     */
+	    (((ticket->t_flags & XLOG_TIC_INITED) == 0) &&
+	     (xlog_commit_record(log, ticket, iclog, &lsn)))) {
+		lsn = (xfs_lsn_t) -1;
+		if (ticket->t_flags & XLOG_TIC_PERM_RESERV) {
+			flags |= XFS_LOG_REL_PERM_RESERV;
+		}
+	}
+
+
+	if ((ticket->t_flags & XLOG_TIC_PERM_RESERV) == 0 ||
+	    (flags & XFS_LOG_REL_PERM_RESERV)) {
+		trace_xfs_log_done_nonperm(log, ticket);
+
+		/*
+		 * Release ticket if not permanent reservation or a specific
+		 * request has been made to release a permanent reservation.
+		 */
+		xlog_ungrant_log_space(log, ticket);
+		xfs_log_ticket_put(ticket);
+	} else {
+		trace_xfs_log_done_perm(log, ticket);
+
+		xlog_regrant_reserve_log_space(log, ticket);
+		/* If this ticket was a permanent reservation and we aren't
+		 * trying to release it, reset the inited flags; so next time
+		 * we write, a start record will be written out.
+		 */
+		ticket->t_flags |= XLOG_TIC_INITED;
+	}
+
+	return lsn;
+}
+
+/*
+ * Attaches a new iclog I/O completion callback routine during
+ * transaction commit.  If the log is in error state, a non-zero
+ * return code is handed back and the caller is responsible for
+ * executing the callback at an appropriate time.
+ */
+int
+xfs_log_notify(
+	struct xfs_mount	*mp,
+	struct xlog_in_core	*iclog,
+	xfs_log_callback_t	*cb)
+{
+	int	abortflg;
+
+	spin_lock(&iclog->ic_callback_lock);
+	abortflg = (iclog->ic_state & XLOG_STATE_IOERROR);
+	if (!abortflg) {
+		ASSERT_ALWAYS((iclog->ic_state == XLOG_STATE_ACTIVE) ||
+			      (iclog->ic_state == XLOG_STATE_WANT_SYNC));
+		cb->cb_next = NULL;
+		*(iclog->ic_callback_tail) = cb;
+		iclog->ic_callback_tail = &(cb->cb_next);
+	}
+	spin_unlock(&iclog->ic_callback_lock);
+	return abortflg;
+}
+
+int
+xfs_log_release_iclog(
+	struct xfs_mount	*mp,
+	struct xlog_in_core	*iclog)
+{
+	if (xlog_state_release_iclog(mp->m_log, iclog)) {
+		xfs_force_shutdown(mp, SHUTDOWN_LOG_IO_ERROR);
+		return EIO;
+	}
+
+	return 0;
+}
+
+/*
+ *  1. Reserve an amount of on-disk log space and return a ticket corresponding
+ *	to the reservation.
+ *  2. Potentially, push buffers at tail of log to disk.
+ *
+ * Each reservation is going to reserve extra space for a log record header.
+ * When writes happen to the on-disk log, we don't subtract the length of the
+ * log record header from any reservation.  By wasting space in each
+ * reservation, we prevent over allocation problems.
+ */
+int
+xfs_log_reserve(
+	struct xfs_mount	*mp,
+	int		 	unit_bytes,
+	int		 	cnt,
+	struct xlog_ticket	**ticket,
+	__uint8_t	 	client,
+	uint		 	flags,
+	uint		 	t_type)
+{
+	struct log		*log = mp->m_log;
+	struct xlog_ticket	*internal_ticket;
+	int			retval = 0;
+
+	ASSERT(client == XFS_TRANSACTION || client == XFS_LOG);
+
+	if (XLOG_FORCED_SHUTDOWN(log))
+		return XFS_ERROR(EIO);
+
+	XFS_STATS_INC(xs_try_logspace);
+
+
+	if (*ticket != NULL) {
+		ASSERT(flags & XFS_LOG_PERM_RESERV);
+		internal_ticket = *ticket;
+
+		/*
+		 * this is a new transaction on the ticket, so we need to
+		 * change the transaction ID so that the next transaction has a
+		 * different TID in the log. Just add one to the existing tid
+		 * so that we can see chains of rolling transactions in the log
+		 * easily.
+		 */
+		internal_ticket->t_tid++;
+
+		trace_xfs_log_reserve(log, internal_ticket);
+
+		xlog_grant_push_ail(log, internal_ticket->t_unit_res);
+		retval = xlog_regrant_write_log_space(log, internal_ticket);
+	} else {
+		/* may sleep if need to allocate more tickets */
+		internal_ticket = xlog_ticket_alloc(log, unit_bytes, cnt,
+						  client, flags,
+						  KM_SLEEP|KM_MAYFAIL);
+		if (!internal_ticket)
+			return XFS_ERROR(ENOMEM);
+		internal_ticket->t_trans_type = t_type;
+		*ticket = internal_ticket;
+
+		trace_xfs_log_reserve(log, internal_ticket);
+
+		xlog_grant_push_ail(log,
+				    (internal_ticket->t_unit_res *
+				     internal_ticket->t_cnt));
+		retval = xlog_grant_log_space(log, internal_ticket);
+	}
+
+	return retval;
+}	/* xfs_log_reserve */
+
+
+/*
+ * Mount a log filesystem
+ *
+ * mp		- ubiquitous xfs mount point structure
+ * log_target	- buftarg of on-disk log device
+ * blk_offset	- Start block # where block size is 512 bytes (BBSIZE)
+ * num_bblocks	- Number of BBSIZE blocks in on-disk log
+ *
+ * Return error or zero.
+ */
+int
+xfs_log_mount(
+	xfs_mount_t	*mp,
+	xfs_buftarg_t	*log_target,
+	xfs_daddr_t	blk_offset,
+	int		num_bblks)
+{
+	int		error;
+
+	if (!(mp->m_flags & XFS_MOUNT_NORECOVERY))
+		xfs_notice(mp, "Mounting Filesystem");
+	else {
+		xfs_notice(mp,
+"Mounting filesystem in no-recovery mode.  Filesystem will be inconsistent.");
+		ASSERT(mp->m_flags & XFS_MOUNT_RDONLY);
+	}
+
+	mp->m_log = xlog_alloc_log(mp, log_target, blk_offset, num_bblks);
+	if (IS_ERR(mp->m_log)) {
+		error = -PTR_ERR(mp->m_log);
+		goto out;
+	}
+
+	/*
+	 * Initialize the AIL now we have a log.
+	 */
+	error = xfs_trans_ail_init(mp);
+	if (error) {
+		xfs_warn(mp, "AIL initialisation failed: error %d", error);
+		goto out_free_log;
+	}
+	mp->m_log->l_ailp = mp->m_ail;
+
+	/*
+	 * skip log recovery on a norecovery mount.  pretend it all
+	 * just worked.
+	 */
+	if (!(mp->m_flags & XFS_MOUNT_NORECOVERY)) {
+		int	readonly = (mp->m_flags & XFS_MOUNT_RDONLY);
+
+		if (readonly)
+			mp->m_flags &= ~XFS_MOUNT_RDONLY;
+
+		error = xlog_recover(mp->m_log);
+
+		if (readonly)
+			mp->m_flags |= XFS_MOUNT_RDONLY;
+		if (error) {
+			xfs_warn(mp, "log mount/recovery failed: error %d",
+				error);
+			goto out_destroy_ail;
+		}
+	}
+
+	/* Normal transactions can now occur */
+	mp->m_log->l_flags &= ~XLOG_ACTIVE_RECOVERY;
+
+	/*
+	 * Now the log has been fully initialised and we know were our
+	 * space grant counters are, we can initialise the permanent ticket
+	 * needed for delayed logging to work.
+	 */
+	xlog_cil_init_post_recovery(mp->m_log);
+
+	return 0;
+
+out_destroy_ail:
+	xfs_trans_ail_destroy(mp);
+out_free_log:
+	xlog_dealloc_log(mp->m_log);
+out:
+	return error;
+}
+
+/*
+ * Finish the recovery of the file system.  This is separate from
+ * the xfs_log_mount() call, because it depends on the code in
+ * xfs_mountfs() to read in the root and real-time bitmap inodes
+ * between calling xfs_log_mount() and here.
+ *
+ * mp		- ubiquitous xfs mount point structure
+ */
+int
+xfs_log_mount_finish(xfs_mount_t *mp)
+{
+	int	error;
+
+	if (!(mp->m_flags & XFS_MOUNT_NORECOVERY))
+		error = xlog_recover_finish(mp->m_log);
+	else {
+		error = 0;
+		ASSERT(mp->m_flags & XFS_MOUNT_RDONLY);
+	}
+
+	return error;
+}
+
+/*
+ * Final log writes as part of unmount.
+ *
+ * Mark the filesystem clean as unmount happens.  Note that during relocation
+ * this routine needs to be executed as part of source-bag while the
+ * deallocation must not be done until source-end.
+ */
+
+/*
+ * Unmount record used to have a string "Unmount filesystem--" in the
+ * data section where the "Un" was really a magic number (XLOG_UNMOUNT_TYPE).
+ * We just write the magic number now since that particular field isn't
+ * currently architecture converted and "nUmount" is a bit foo.
+ * As far as I know, there weren't any dependencies on the old behaviour.
+ */
+
+int
+xfs_log_unmount_write(xfs_mount_t *mp)
+{
+	xlog_t		 *log = mp->m_log;
+	xlog_in_core_t	 *iclog;
+#ifdef DEBUG
+	xlog_in_core_t	 *first_iclog;
+#endif
+	xlog_ticket_t	*tic = NULL;
+	xfs_lsn_t	 lsn;
+	int		 error;
+
+	/*
+	 * Don't write out unmount record on read-only mounts.
+	 * Or, if we are doing a forced umount (typically because of IO errors).
+	 */
+	if (mp->m_flags & XFS_MOUNT_RDONLY)
+		return 0;
+
+	error = _xfs_log_force(mp, XFS_LOG_SYNC, NULL);
+	ASSERT(error || !(XLOG_FORCED_SHUTDOWN(log)));
+
+#ifdef DEBUG
+	first_iclog = iclog = log->l_iclog;
+	do {
+		if (!(iclog->ic_state & XLOG_STATE_IOERROR)) {
+			ASSERT(iclog->ic_state & XLOG_STATE_ACTIVE);
+			ASSERT(iclog->ic_offset == 0);
+		}
+		iclog = iclog->ic_next;
+	} while (iclog != first_iclog);
+#endif
+	if (! (XLOG_FORCED_SHUTDOWN(log))) {
+		error = xfs_log_reserve(mp, 600, 1, &tic,
+					XFS_LOG, 0, XLOG_UNMOUNT_REC_TYPE);
+		if (!error) {
+			/* the data section must be 32 bit size aligned */
+			struct {
+			    __uint16_t magic;
+			    __uint16_t pad1;
+			    __uint32_t pad2; /* may as well make it 64 bits */
+			} magic = {
+				.magic = XLOG_UNMOUNT_TYPE,
+			};
+			struct xfs_log_iovec reg = {
+				.i_addr = &magic,
+				.i_len = sizeof(magic),
+				.i_type = XLOG_REG_TYPE_UNMOUNT,
+			};
+			struct xfs_log_vec vec = {
+				.lv_niovecs = 1,
+				.lv_iovecp = &reg,
+			};
+
+			/* remove inited flag */
+			tic->t_flags = 0;
+			error = xlog_write(log, &vec, tic, &lsn,
+					   NULL, XLOG_UNMOUNT_TRANS);
+			/*
+			 * At this point, we're umounting anyway,
+			 * so there's no point in transitioning log state
+			 * to IOERROR. Just continue...
+			 */
+		}
+
+		if (error)
+			xfs_alert(mp, "%s: unmount record failed", __func__);
+
+
+		spin_lock(&log->l_icloglock);
+		iclog = log->l_iclog;
+		atomic_inc(&iclog->ic_refcnt);
+		xlog_state_want_sync(log, iclog);
+		spin_unlock(&log->l_icloglock);
+		error = xlog_state_release_iclog(log, iclog);
+
+		spin_lock(&log->l_icloglock);
+		if (!(iclog->ic_state == XLOG_STATE_ACTIVE ||
+		      iclog->ic_state == XLOG_STATE_DIRTY)) {
+			if (!XLOG_FORCED_SHUTDOWN(log)) {
+				xlog_wait(&iclog->ic_force_wait,
+							&log->l_icloglock);
+			} else {
+				spin_unlock(&log->l_icloglock);
+			}
+		} else {
+			spin_unlock(&log->l_icloglock);
+		}
+		if (tic) {
+			trace_xfs_log_umount_write(log, tic);
+			xlog_ungrant_log_space(log, tic);
+			xfs_log_ticket_put(tic);
+		}
+	} else {
+		/*
+		 * We're already in forced_shutdown mode, couldn't
+		 * even attempt to write out the unmount transaction.
+		 *
+		 * Go through the motions of sync'ing and releasing
+		 * the iclog, even though no I/O will actually happen,
+		 * we need to wait for other log I/Os that may already
+		 * be in progress.  Do this as a separate section of
+		 * code so we'll know if we ever get stuck here that
+		 * we're in this odd situation of trying to unmount
+		 * a file system that went into forced_shutdown as
+		 * the result of an unmount..
+		 */
+		spin_lock(&log->l_icloglock);
+		iclog = log->l_iclog;
+		atomic_inc(&iclog->ic_refcnt);
+
+		xlog_state_want_sync(log, iclog);
+		spin_unlock(&log->l_icloglock);
+		error =  xlog_state_release_iclog(log, iclog);
+
+		spin_lock(&log->l_icloglock);
+
+		if ( ! (   iclog->ic_state == XLOG_STATE_ACTIVE
+			|| iclog->ic_state == XLOG_STATE_DIRTY
+			|| iclog->ic_state == XLOG_STATE_IOERROR) ) {
+
+				xlog_wait(&iclog->ic_force_wait,
+							&log->l_icloglock);
+		} else {
+			spin_unlock(&log->l_icloglock);
+		}
+	}
+
+	return error;
+}	/* xfs_log_unmount_write */
+
+/*
+ * Deallocate log structures for unmount/relocation.
+ *
+ * We need to stop the aild from running before we destroy
+ * and deallocate the log as the aild references the log.
+ */
+void
+xfs_log_unmount(xfs_mount_t *mp)
+{
+	xfs_trans_ail_destroy(mp);
+	xlog_dealloc_log(mp->m_log);
+}
+
+void
+xfs_log_item_init(
+	struct xfs_mount	*mp,
+	struct xfs_log_item	*item,
+	int			type,
+	struct xfs_item_ops	*ops)
+{
+	item->li_mountp = mp;
+	item->li_ailp = mp->m_ail;
+	item->li_type = type;
+	item->li_ops = ops;
+	item->li_lv = NULL;
+
+	INIT_LIST_HEAD(&item->li_ail);
+	INIT_LIST_HEAD(&item->li_cil);
+}
+
+/*
+ * Write region vectors to log.  The write happens using the space reservation
+ * of the ticket (tic).  It is not a requirement that all writes for a given
+ * transaction occur with one call to xfs_log_write(). However, it is important
+ * to note that the transaction reservation code makes an assumption about the
+ * number of log headers a transaction requires that may be violated if you
+ * don't pass all the transaction vectors in one call....
+ */
+int
+xfs_log_write(
+	struct xfs_mount	*mp,
+	struct xfs_log_iovec	reg[],
+	int			nentries,
+	struct xlog_ticket	*tic,
+	xfs_lsn_t		*start_lsn)
+{
+	struct log		*log = mp->m_log;
+	int			error;
+	struct xfs_log_vec	vec = {
+		.lv_niovecs = nentries,
+		.lv_iovecp = reg,
+	};
+
+	if (XLOG_FORCED_SHUTDOWN(log))
+		return XFS_ERROR(EIO);
+
+	error = xlog_write(log, &vec, tic, start_lsn, NULL, 0);
+	if (error)
+		xfs_force_shutdown(mp, SHUTDOWN_LOG_IO_ERROR);
+	return error;
+}
+
+void
+xfs_log_move_tail(xfs_mount_t	*mp,
+		  xfs_lsn_t	tail_lsn)
+{
+	xlog_ticket_t	*tic;
+	xlog_t		*log = mp->m_log;
+	int		need_bytes, free_bytes;
+
+	if (XLOG_FORCED_SHUTDOWN(log))
+		return;
+
+	if (tail_lsn == 0)
+		tail_lsn = atomic64_read(&log->l_last_sync_lsn);
+
+	/* tail_lsn == 1 implies that we weren't passed a valid value.  */
+	if (tail_lsn != 1)
+		atomic64_set(&log->l_tail_lsn, tail_lsn);
+
+	if (!list_empty_careful(&log->l_writeq)) {
+#ifdef DEBUG
+		if (log->l_flags & XLOG_ACTIVE_RECOVERY)
+			panic("Recovery problem");
+#endif
+		spin_lock(&log->l_grant_write_lock);
+		free_bytes = xlog_space_left(log, &log->l_grant_write_head);
+		list_for_each_entry(tic, &log->l_writeq, t_queue) {
+			ASSERT(tic->t_flags & XLOG_TIC_PERM_RESERV);
+
+			if (free_bytes < tic->t_unit_res && tail_lsn != 1)
+				break;
+			tail_lsn = 0;
+			free_bytes -= tic->t_unit_res;
+			trace_xfs_log_regrant_write_wake_up(log, tic);
+			wake_up(&tic->t_wait);
+		}
+		spin_unlock(&log->l_grant_write_lock);
+	}
+
+	if (!list_empty_careful(&log->l_reserveq)) {
+#ifdef DEBUG
+		if (log->l_flags & XLOG_ACTIVE_RECOVERY)
+			panic("Recovery problem");
+#endif
+		spin_lock(&log->l_grant_reserve_lock);
+		free_bytes = xlog_space_left(log, &log->l_grant_reserve_head);
+		list_for_each_entry(tic, &log->l_reserveq, t_queue) {
+			if (tic->t_flags & XLOG_TIC_PERM_RESERV)
+				need_bytes = tic->t_unit_res*tic->t_cnt;
+			else
+				need_bytes = tic->t_unit_res;
+			if (free_bytes < need_bytes && tail_lsn != 1)
+				break;
+			tail_lsn = 0;
+			free_bytes -= need_bytes;
+			trace_xfs_log_grant_wake_up(log, tic);
+			wake_up(&tic->t_wait);
+		}
+		spin_unlock(&log->l_grant_reserve_lock);
+	}
+}
+
+/*
+ * Determine if we have a transaction that has gone to disk
+ * that needs to be covered. To begin the transition to the idle state
+ * firstly the log needs to be idle (no AIL and nothing in the iclogs).
+ * If we are then in a state where covering is needed, the caller is informed
+ * that dummy transactions are required to move the log into the idle state.
+ *
+ * Because this is called as part of the sync process, we should also indicate
+ * that dummy transactions should be issued in anything but the covered or
+ * idle states. This ensures that the log tail is accurately reflected in
+ * the log at the end of the sync, hence if a crash occurrs avoids replay
+ * of transactions where the metadata is already on disk.
+ */
+int
+xfs_log_need_covered(xfs_mount_t *mp)
+{
+	int		needed = 0;
+	xlog_t		*log = mp->m_log;
+
+	if (!xfs_fs_writable(mp))
+		return 0;
+
+	spin_lock(&log->l_icloglock);
+	switch (log->l_covered_state) {
+	case XLOG_STATE_COVER_DONE:
+	case XLOG_STATE_COVER_DONE2:
+	case XLOG_STATE_COVER_IDLE:
+		break;
+	case XLOG_STATE_COVER_NEED:
+	case XLOG_STATE_COVER_NEED2:
+		if (!xfs_ail_min_lsn(log->l_ailp) &&
+		    xlog_iclogs_empty(log)) {
+			if (log->l_covered_state == XLOG_STATE_COVER_NEED)
+				log->l_covered_state = XLOG_STATE_COVER_DONE;
+			else
+				log->l_covered_state = XLOG_STATE_COVER_DONE2;
+		}
+		/* FALLTHRU */
+	default:
+		needed = 1;
+		break;
+	}
+	spin_unlock(&log->l_icloglock);
+	return needed;
+}
+
+/******************************************************************************
+ *
+ *	local routines
+ *
+ ******************************************************************************
+ */
+
+/* xfs_trans_tail_ail returns 0 when there is nothing in the list.
+ * The log manager must keep track of the last LR which was committed
+ * to disk.  The lsn of this LR will become the new tail_lsn whenever
+ * xfs_trans_tail_ail returns 0.  If we don't do this, we run into
+ * the situation where stuff could be written into the log but nothing
+ * was ever in the AIL when asked.  Eventually, we panic since the
+ * tail hits the head.
+ *
+ * We may be holding the log iclog lock upon entering this routine.
+ */
+xfs_lsn_t
+xlog_assign_tail_lsn(
+	struct xfs_mount	*mp)
+{
+	xfs_lsn_t		tail_lsn;
+	struct log		*log = mp->m_log;
+
+	tail_lsn = xfs_ail_min_lsn(mp->m_ail);
+	if (!tail_lsn)
+		tail_lsn = atomic64_read(&log->l_last_sync_lsn);
+
+	atomic64_set(&log->l_tail_lsn, tail_lsn);
+	return tail_lsn;
+}
+
+/*
+ * Return the space in the log between the tail and the head.  The head
+ * is passed in the cycle/bytes formal parms.  In the special case where
+ * the reserve head has wrapped passed the tail, this calculation is no
+ * longer valid.  In this case, just return 0 which means there is no space
+ * in the log.  This works for all places where this function is called
+ * with the reserve head.  Of course, if the write head were to ever
+ * wrap the tail, we should blow up.  Rather than catch this case here,
+ * we depend on other ASSERTions in other parts of the code.   XXXmiken
+ *
+ * This code also handles the case where the reservation head is behind
+ * the tail.  The details of this case are described below, but the end
+ * result is that we return the size of the log as the amount of space left.
+ */
+STATIC int
+xlog_space_left(
+	struct log	*log,
+	atomic64_t	*head)
+{
+	int		free_bytes;
+	int		tail_bytes;
+	int		tail_cycle;
+	int		head_cycle;
+	int		head_bytes;
+
+	xlog_crack_grant_head(head, &head_cycle, &head_bytes);
+	xlog_crack_atomic_lsn(&log->l_tail_lsn, &tail_cycle, &tail_bytes);
+	tail_bytes = BBTOB(tail_bytes);
+	if (tail_cycle == head_cycle && head_bytes >= tail_bytes)
+		free_bytes = log->l_logsize - (head_bytes - tail_bytes);
+	else if (tail_cycle + 1 < head_cycle)
+		return 0;
+	else if (tail_cycle < head_cycle) {
+		ASSERT(tail_cycle == (head_cycle - 1));
+		free_bytes = tail_bytes - head_bytes;
+	} else {
+		/*
+		 * The reservation head is behind the tail.
+		 * In this case we just want to return the size of the
+		 * log as the amount of space left.
+		 */
+		xfs_alert(log->l_mp,
+			"xlog_space_left: head behind tail\n"
+			"  tail_cycle = %d, tail_bytes = %d\n"
+			"  GH   cycle = %d, GH   bytes = %d",
+			tail_cycle, tail_bytes, head_cycle, head_bytes);
+		ASSERT(0);
+		free_bytes = log->l_logsize;
+	}
+	return free_bytes;
+}
+
+
+/*
+ * Log function which is called when an io completes.
+ *
+ * The log manager needs its own routine, in order to control what
+ * happens with the buffer after the write completes.
+ */
+void
+xlog_iodone(xfs_buf_t *bp)
+{
+	xlog_in_core_t	*iclog;
+	xlog_t		*l;
+	int		aborted;
+
+	iclog = XFS_BUF_FSPRIVATE(bp, xlog_in_core_t *);
+	ASSERT(XFS_BUF_FSPRIVATE2(bp, unsigned long) == (unsigned long) 2);
+	XFS_BUF_SET_FSPRIVATE2(bp, (unsigned long)1);
+	aborted = 0;
+	l = iclog->ic_log;
+
+	/*
+	 * Race to shutdown the filesystem if we see an error.
+	 */
+	if (XFS_TEST_ERROR((XFS_BUF_GETERROR(bp)), l->l_mp,
+			XFS_ERRTAG_IODONE_IOERR, XFS_RANDOM_IODONE_IOERR)) {
+		xfs_ioerror_alert("xlog_iodone", l->l_mp, bp, XFS_BUF_ADDR(bp));
+		XFS_BUF_STALE(bp);
+		xfs_force_shutdown(l->l_mp, SHUTDOWN_LOG_IO_ERROR);
+		/*
+		 * This flag will be propagated to the trans-committed
+		 * callback routines to let them know that the log-commit
+		 * didn't succeed.
+		 */
+		aborted = XFS_LI_ABORTED;
+	} else if (iclog->ic_state & XLOG_STATE_IOERROR) {
+		aborted = XFS_LI_ABORTED;
+	}
+
+	/* log I/O is always issued ASYNC */
+	ASSERT(XFS_BUF_ISASYNC(bp));
+	xlog_state_done_syncing(iclog, aborted);
+	/*
+	 * do not reference the buffer (bp) here as we could race
+	 * with it being freed after writing the unmount record to the
+	 * log.
+	 */
+
+}	/* xlog_iodone */
+
+/*
+ * Return size of each in-core log record buffer.
+ *
+ * All machines get 8 x 32kB buffers by default, unless tuned otherwise.
+ *
+ * If the filesystem blocksize is too large, we may need to choose a
+ * larger size since the directory code currently logs entire blocks.
+ */
+
+STATIC void
+xlog_get_iclog_buffer_size(xfs_mount_t	*mp,
+			   xlog_t	*log)
+{
+	int size;
+	int xhdrs;
+
+	if (mp->m_logbufs <= 0)
+		log->l_iclog_bufs = XLOG_MAX_ICLOGS;
+	else
+		log->l_iclog_bufs = mp->m_logbufs;
+
+	/*
+	 * Buffer size passed in from mount system call.
+	 */
+	if (mp->m_logbsize > 0) {
+		size = log->l_iclog_size = mp->m_logbsize;
+		log->l_iclog_size_log = 0;
+		while (size != 1) {
+			log->l_iclog_size_log++;
+			size >>= 1;
+		}
+
+		if (xfs_sb_version_haslogv2(&mp->m_sb)) {
+			/* # headers = size / 32k
+			 * one header holds cycles from 32k of data
+			 */
+
+			xhdrs = mp->m_logbsize / XLOG_HEADER_CYCLE_SIZE;
+			if (mp->m_logbsize % XLOG_HEADER_CYCLE_SIZE)
+				xhdrs++;
+			log->l_iclog_hsize = xhdrs << BBSHIFT;
+			log->l_iclog_heads = xhdrs;
+		} else {
+			ASSERT(mp->m_logbsize <= XLOG_BIG_RECORD_BSIZE);
+			log->l_iclog_hsize = BBSIZE;
+			log->l_iclog_heads = 1;
+		}
+		goto done;
+	}
+
+	/* All machines use 32kB buffers by default. */
+	log->l_iclog_size = XLOG_BIG_RECORD_BSIZE;
+	log->l_iclog_size_log = XLOG_BIG_RECORD_BSHIFT;
+
+	/* the default log size is 16k or 32k which is one header sector */
+	log->l_iclog_hsize = BBSIZE;
+	log->l_iclog_heads = 1;
+
+done:
+	/* are we being asked to make the sizes selected above visible? */
+	if (mp->m_logbufs == 0)
+		mp->m_logbufs = log->l_iclog_bufs;
+	if (mp->m_logbsize == 0)
+		mp->m_logbsize = log->l_iclog_size;
+}	/* xlog_get_iclog_buffer_size */
+
+
+/*
+ * This routine initializes some of the log structure for a given mount point.
+ * Its primary purpose is to fill in enough, so recovery can occur.  However,
+ * some other stuff may be filled in too.
+ */
+STATIC xlog_t *
+xlog_alloc_log(xfs_mount_t	*mp,
+	       xfs_buftarg_t	*log_target,
+	       xfs_daddr_t	blk_offset,
+	       int		num_bblks)
+{
+	xlog_t			*log;
+	xlog_rec_header_t	*head;
+	xlog_in_core_t		**iclogp;
+	xlog_in_core_t		*iclog, *prev_iclog=NULL;
+	xfs_buf_t		*bp;
+	int			i;
+	int			error = ENOMEM;
+	uint			log2_size = 0;
+
+	log = kmem_zalloc(sizeof(xlog_t), KM_MAYFAIL);
+	if (!log) {
+		xfs_warn(mp, "Log allocation failed: No memory!");
+		goto out;
+	}
+
+	log->l_mp	   = mp;
+	log->l_targ	   = log_target;
+	log->l_logsize     = BBTOB(num_bblks);
+	log->l_logBBstart  = blk_offset;
+	log->l_logBBsize   = num_bblks;
+	log->l_covered_state = XLOG_STATE_COVER_IDLE;
+	log->l_flags	   |= XLOG_ACTIVE_RECOVERY;
+
+	log->l_prev_block  = -1;
+	/* log->l_tail_lsn = 0x100000000LL; cycle = 1; current block = 0 */
+	xlog_assign_atomic_lsn(&log->l_tail_lsn, 1, 0);
+	xlog_assign_atomic_lsn(&log->l_last_sync_lsn, 1, 0);
+	log->l_curr_cycle  = 1;	    /* 0 is bad since this is initial value */
+	xlog_assign_grant_head(&log->l_grant_reserve_head, 1, 0);
+	xlog_assign_grant_head(&log->l_grant_write_head, 1, 0);
+	INIT_LIST_HEAD(&log->l_reserveq);
+	INIT_LIST_HEAD(&log->l_writeq);
+	spin_lock_init(&log->l_grant_reserve_lock);
+	spin_lock_init(&log->l_grant_write_lock);
+
+	error = EFSCORRUPTED;
+	if (xfs_sb_version_hassector(&mp->m_sb)) {
+	        log2_size = mp->m_sb.sb_logsectlog;
+		if (log2_size < BBSHIFT) {
+			xfs_warn(mp, "Log sector size too small (0x%x < 0x%x)",
+				log2_size, BBSHIFT);
+			goto out_free_log;
+		}
+
+	        log2_size -= BBSHIFT;
+		if (log2_size > mp->m_sectbb_log) {
+			xfs_warn(mp, "Log sector size too large (0x%x > 0x%x)",
+				log2_size, mp->m_sectbb_log);
+			goto out_free_log;
+		}
+
+		/* for larger sector sizes, must have v2 or external log */
+		if (log2_size && log->l_logBBstart > 0 &&
+			    !xfs_sb_version_haslogv2(&mp->m_sb)) {
+			xfs_warn(mp,
+		"log sector size (0x%x) invalid for configuration.",
+				log2_size);
+			goto out_free_log;
+		}
+	}
+	log->l_sectBBsize = 1 << log2_size;
+
+	xlog_get_iclog_buffer_size(mp, log);
+
+	error = ENOMEM;
+	bp = xfs_buf_get_empty(log->l_iclog_size, mp->m_logdev_targp);
+	if (!bp)
+		goto out_free_log;
+	XFS_BUF_SET_IODONE_FUNC(bp, xlog_iodone);
+	XFS_BUF_SET_FSPRIVATE2(bp, (unsigned long)1);
+	ASSERT(XFS_BUF_ISBUSY(bp));
+	ASSERT(XFS_BUF_VALUSEMA(bp) <= 0);
+	log->l_xbuf = bp;
+
+	spin_lock_init(&log->l_icloglock);
+	init_waitqueue_head(&log->l_flush_wait);
+
+	/* log record size must be multiple of BBSIZE; see xlog_rec_header_t */
+	ASSERT((XFS_BUF_SIZE(bp) & BBMASK) == 0);
+
+	iclogp = &log->l_iclog;
+	/*
+	 * The amount of memory to allocate for the iclog structure is
+	 * rather funky due to the way the structure is defined.  It is
+	 * done this way so that we can use different sizes for machines
+	 * with different amounts of memory.  See the definition of
+	 * xlog_in_core_t in xfs_log_priv.h for details.
+	 */
+	ASSERT(log->l_iclog_size >= 4096);
+	for (i=0; i < log->l_iclog_bufs; i++) {
+		*iclogp = kmem_zalloc(sizeof(xlog_in_core_t), KM_MAYFAIL);
+		if (!*iclogp)
+			goto out_free_iclog;
+
+		iclog = *iclogp;
+		iclog->ic_prev = prev_iclog;
+		prev_iclog = iclog;
+
+		bp = xfs_buf_get_uncached(mp->m_logdev_targp,
+						log->l_iclog_size, 0);
+		if (!bp)
+			goto out_free_iclog;
+		if (!XFS_BUF_CPSEMA(bp))
+			ASSERT(0);
+		XFS_BUF_SET_IODONE_FUNC(bp, xlog_iodone);
+		XFS_BUF_SET_FSPRIVATE2(bp, (unsigned long)1);
+		iclog->ic_bp = bp;
+		iclog->ic_data = bp->b_addr;
+#ifdef DEBUG
+		log->l_iclog_bak[i] = (xfs_caddr_t)&(iclog->ic_header);
+#endif
+		head = &iclog->ic_header;
+		memset(head, 0, sizeof(xlog_rec_header_t));
+		head->h_magicno = cpu_to_be32(XLOG_HEADER_MAGIC_NUM);
+		head->h_version = cpu_to_be32(
+			xfs_sb_version_haslogv2(&log->l_mp->m_sb) ? 2 : 1);
+		head->h_size = cpu_to_be32(log->l_iclog_size);
+		/* new fields */
+		head->h_fmt = cpu_to_be32(XLOG_FMT);
+		memcpy(&head->h_fs_uuid, &mp->m_sb.sb_uuid, sizeof(uuid_t));
+
+		iclog->ic_size = XFS_BUF_SIZE(bp) - log->l_iclog_hsize;
+		iclog->ic_state = XLOG_STATE_ACTIVE;
+		iclog->ic_log = log;
+		atomic_set(&iclog->ic_refcnt, 0);
+		spin_lock_init(&iclog->ic_callback_lock);
+		iclog->ic_callback_tail = &(iclog->ic_callback);
+		iclog->ic_datap = (char *)iclog->ic_data + log->l_iclog_hsize;
+
+		ASSERT(XFS_BUF_ISBUSY(iclog->ic_bp));
+		ASSERT(XFS_BUF_VALUSEMA(iclog->ic_bp) <= 0);
+		init_waitqueue_head(&iclog->ic_force_wait);
+		init_waitqueue_head(&iclog->ic_write_wait);
+
+		iclogp = &iclog->ic_next;
+	}
+	*iclogp = log->l_iclog;			/* complete ring */
+	log->l_iclog->ic_prev = prev_iclog;	/* re-write 1st prev ptr */
+
+	error = xlog_cil_init(log);
+	if (error)
+		goto out_free_iclog;
+	return log;
+
+out_free_iclog:
+	for (iclog = log->l_iclog; iclog; iclog = prev_iclog) {
+		prev_iclog = iclog->ic_next;
+		if (iclog->ic_bp)
+			xfs_buf_free(iclog->ic_bp);
+		kmem_free(iclog);
+	}
+	spinlock_destroy(&log->l_icloglock);
+	xfs_buf_free(log->l_xbuf);
+out_free_log:
+	kmem_free(log);
+out:
+	return ERR_PTR(-error);
+}	/* xlog_alloc_log */
+
+
+/*
+ * Write out the commit record of a transaction associated with the given
+ * ticket.  Return the lsn of the commit record.
+ */
+STATIC int
+xlog_commit_record(
+	struct log		*log,
+	struct xlog_ticket	*ticket,
+	struct xlog_in_core	**iclog,
+	xfs_lsn_t		*commitlsnp)
+{
+	struct xfs_mount *mp = log->l_mp;
+	int	error;
+	struct xfs_log_iovec reg = {
+		.i_addr = NULL,
+		.i_len = 0,
+		.i_type = XLOG_REG_TYPE_COMMIT,
+	};
+	struct xfs_log_vec vec = {
+		.lv_niovecs = 1,
+		.lv_iovecp = &reg,
+	};
+
+	ASSERT_ALWAYS(iclog);
+	error = xlog_write(log, &vec, ticket, commitlsnp, iclog,
+					XLOG_COMMIT_TRANS);
+	if (error)
+		xfs_force_shutdown(mp, SHUTDOWN_LOG_IO_ERROR);
+	return error;
+}
+
+/*
+ * Push on the buffer cache code if we ever use more than 75% of the on-disk
+ * log space.  This code pushes on the lsn which would supposedly free up
+ * the 25% which we want to leave free.  We may need to adopt a policy which
+ * pushes on an lsn which is further along in the log once we reach the high
+ * water mark.  In this manner, we would be creating a low water mark.
+ */
+STATIC void
+xlog_grant_push_ail(
+	struct log	*log,
+	int		need_bytes)
+{
+	xfs_lsn_t	threshold_lsn = 0;
+	xfs_lsn_t	last_sync_lsn;
+	int		free_blocks;
+	int		free_bytes;
+	int		threshold_block;
+	int		threshold_cycle;
+	int		free_threshold;
+
+	ASSERT(BTOBB(need_bytes) < log->l_logBBsize);
+
+	free_bytes = xlog_space_left(log, &log->l_grant_reserve_head);
+	free_blocks = BTOBBT(free_bytes);
+
+	/*
+	 * Set the threshold for the minimum number of free blocks in the
+	 * log to the maximum of what the caller needs, one quarter of the
+	 * log, and 256 blocks.
+	 */
+	free_threshold = BTOBB(need_bytes);
+	free_threshold = MAX(free_threshold, (log->l_logBBsize >> 2));
+	free_threshold = MAX(free_threshold, 256);
+	if (free_blocks >= free_threshold)
+		return;
+
+	xlog_crack_atomic_lsn(&log->l_tail_lsn, &threshold_cycle,
+						&threshold_block);
+	threshold_block += free_threshold;
+	if (threshold_block >= log->l_logBBsize) {
+		threshold_block -= log->l_logBBsize;
+		threshold_cycle += 1;
+	}
+	threshold_lsn = xlog_assign_lsn(threshold_cycle,
+					threshold_block);
+	/*
+	 * Don't pass in an lsn greater than the lsn of the last
+	 * log record known to be on disk. Use a snapshot of the last sync lsn
+	 * so that it doesn't change between the compare and the set.
+	 */
+	last_sync_lsn = atomic64_read(&log->l_last_sync_lsn);
+	if (XFS_LSN_CMP(threshold_lsn, last_sync_lsn) > 0)
+		threshold_lsn = last_sync_lsn;
+
+	/*
+	 * Get the transaction layer to kick the dirty buffers out to
+	 * disk asynchronously. No point in trying to do this if
+	 * the filesystem is shutting down.
+	 */
+	if (!XLOG_FORCED_SHUTDOWN(log))
+		xfs_ail_push(log->l_ailp, threshold_lsn);
+}
+
+/*
+ * The bdstrat callback function for log bufs. This gives us a central
+ * place to trap bufs in case we get hit by a log I/O error and need to
+ * shutdown. Actually, in practice, even when we didn't get a log error,
+ * we transition the iclogs to IOERROR state *after* flushing all existing
+ * iclogs to disk. This is because we don't want anymore new transactions to be
+ * started or completed afterwards.
+ */
+STATIC int
+xlog_bdstrat(
+	struct xfs_buf		*bp)
+{
+	struct xlog_in_core	*iclog;
+
+	iclog = XFS_BUF_FSPRIVATE(bp, xlog_in_core_t *);
+	if (iclog->ic_state & XLOG_STATE_IOERROR) {
+		XFS_BUF_ERROR(bp, EIO);
+		XFS_BUF_STALE(bp);
+		xfs_buf_ioend(bp, 0);
+		/*
+		 * It would seem logical to return EIO here, but we rely on
+		 * the log state machine to propagate I/O errors instead of
+		 * doing it here.
+		 */
+		return 0;
+	}
+
+	bp->b_flags |= _XBF_RUN_QUEUES;
+	xfs_buf_iorequest(bp);
+	return 0;
+}
+
+/*
+ * Flush out the in-core log (iclog) to the on-disk log in an asynchronous 
+ * fashion.  Previously, we should have moved the current iclog
+ * ptr in the log to point to the next available iclog.  This allows further
+ * write to continue while this code syncs out an iclog ready to go.
+ * Before an in-core log can be written out, the data section must be scanned
+ * to save away the 1st word of each BBSIZE block into the header.  We replace
+ * it with the current cycle count.  Each BBSIZE block is tagged with the
+ * cycle count because there in an implicit assumption that drives will
+ * guarantee that entire 512 byte blocks get written at once.  In other words,
+ * we can't have part of a 512 byte block written and part not written.  By
+ * tagging each block, we will know which blocks are valid when recovering
+ * after an unclean shutdown.
+ *
+ * This routine is single threaded on the iclog.  No other thread can be in
+ * this routine with the same iclog.  Changing contents of iclog can there-
+ * fore be done without grabbing the state machine lock.  Updating the global
+ * log will require grabbing the lock though.
+ *
+ * The entire log manager uses a logical block numbering scheme.  Only
+ * log_sync (and then only bwrite()) know about the fact that the log may
+ * not start with block zero on a given device.  The log block start offset
+ * is added immediately before calling bwrite().
+ */
+
+STATIC int
+xlog_sync(xlog_t		*log,
+	  xlog_in_core_t	*iclog)
+{
+	xfs_caddr_t	dptr;		/* pointer to byte sized element */
+	xfs_buf_t	*bp;
+	int		i;
+	uint		count;		/* byte count of bwrite */
+	uint		count_init;	/* initial count before roundup */
+	int		roundoff;       /* roundoff to BB or stripe */
+	int		split = 0;	/* split write into two regions */
+	int		error;
+	int		v2 = xfs_sb_version_haslogv2(&log->l_mp->m_sb);
+
+	XFS_STATS_INC(xs_log_writes);
+	ASSERT(atomic_read(&iclog->ic_refcnt) == 0);
+
+	/* Add for LR header */
+	count_init = log->l_iclog_hsize + iclog->ic_offset;
+
+	/* Round out the log write size */
+	if (v2 && log->l_mp->m_sb.sb_logsunit > 1) {
+		/* we have a v2 stripe unit to use */
+		count = XLOG_LSUNITTOB(log, XLOG_BTOLSUNIT(log, count_init));
+	} else {
+		count = BBTOB(BTOBB(count_init));
+	}
+	roundoff = count - count_init;
+	ASSERT(roundoff >= 0);
+	ASSERT((v2 && log->l_mp->m_sb.sb_logsunit > 1 && 
+                roundoff < log->l_mp->m_sb.sb_logsunit)
+		|| 
+		(log->l_mp->m_sb.sb_logsunit <= 1 && 
+		 roundoff < BBTOB(1)));
+
+	/* move grant heads by roundoff in sync */
+	xlog_grant_add_space(log, &log->l_grant_reserve_head, roundoff);
+	xlog_grant_add_space(log, &log->l_grant_write_head, roundoff);
+
+	/* put cycle number in every block */
+	xlog_pack_data(log, iclog, roundoff); 
+
+	/* real byte length */
+	if (v2) {
+		iclog->ic_header.h_len =
+			cpu_to_be32(iclog->ic_offset + roundoff);
+	} else {
+		iclog->ic_header.h_len =
+			cpu_to_be32(iclog->ic_offset);
+	}
+
+	bp = iclog->ic_bp;
+	ASSERT(XFS_BUF_FSPRIVATE2(bp, unsigned long) == (unsigned long)1);
+	XFS_BUF_SET_FSPRIVATE2(bp, (unsigned long)2);
+	XFS_BUF_SET_ADDR(bp, BLOCK_LSN(be64_to_cpu(iclog->ic_header.h_lsn)));
+
+	XFS_STATS_ADD(xs_log_blocks, BTOBB(count));
+
+	/* Do we need to split this write into 2 parts? */
+	if (XFS_BUF_ADDR(bp) + BTOBB(count) > log->l_logBBsize) {
+		split = count - (BBTOB(log->l_logBBsize - XFS_BUF_ADDR(bp)));
+		count = BBTOB(log->l_logBBsize - XFS_BUF_ADDR(bp));
+		iclog->ic_bwritecnt = 2;	/* split into 2 writes */
+	} else {
+		iclog->ic_bwritecnt = 1;
+	}
+	XFS_BUF_SET_COUNT(bp, count);
+	XFS_BUF_SET_FSPRIVATE(bp, iclog);	/* save for later */
+	XFS_BUF_ZEROFLAGS(bp);
+	XFS_BUF_BUSY(bp);
+	XFS_BUF_ASYNC(bp);
+	bp->b_flags |= XBF_LOG_BUFFER;
+
+	if (log->l_mp->m_flags & XFS_MOUNT_BARRIER) {
+		/*
+		 * If we have an external log device, flush the data device
+		 * before flushing the log to make sure all meta data
+		 * written back from the AIL actually made it to disk
+		 * before writing out the new log tail LSN in the log buffer.
+		 */
+		if (log->l_mp->m_logdev_targp != log->l_mp->m_ddev_targp)
+			xfs_blkdev_issue_flush(log->l_mp->m_ddev_targp);
+		XFS_BUF_ORDERED(bp);
+	}
+
+	ASSERT(XFS_BUF_ADDR(bp) <= log->l_logBBsize-1);
+	ASSERT(XFS_BUF_ADDR(bp) + BTOBB(count) <= log->l_logBBsize);
+
+	xlog_verify_iclog(log, iclog, count, B_TRUE);
+
+	/* account for log which doesn't start at block #0 */
+	XFS_BUF_SET_ADDR(bp, XFS_BUF_ADDR(bp) + log->l_logBBstart);
+	/*
+	 * Don't call xfs_bwrite here. We do log-syncs even when the filesystem
+	 * is shutting down.
+	 */
+	XFS_BUF_WRITE(bp);
+
+	if ((error = xlog_bdstrat(bp))) {
+		xfs_ioerror_alert("xlog_sync", log->l_mp, bp,
+				  XFS_BUF_ADDR(bp));
+		return error;
+	}
+	if (split) {
+		bp = iclog->ic_log->l_xbuf;
+		ASSERT(XFS_BUF_FSPRIVATE2(bp, unsigned long) ==
+							(unsigned long)1);
+		XFS_BUF_SET_FSPRIVATE2(bp, (unsigned long)2);
+		XFS_BUF_SET_ADDR(bp, 0);	     /* logical 0 */
+		XFS_BUF_SET_PTR(bp, (xfs_caddr_t)((__psint_t)&(iclog->ic_header)+
+					    (__psint_t)count), split);
+		XFS_BUF_SET_FSPRIVATE(bp, iclog);
+		XFS_BUF_ZEROFLAGS(bp);
+		XFS_BUF_BUSY(bp);
+		XFS_BUF_ASYNC(bp);
+		bp->b_flags |= XBF_LOG_BUFFER;
+		if (log->l_mp->m_flags & XFS_MOUNT_BARRIER)
+			XFS_BUF_ORDERED(bp);
+		dptr = XFS_BUF_PTR(bp);
+		/*
+		 * Bump the cycle numbers at the start of each block
+		 * since this part of the buffer is at the start of
+		 * a new cycle.  Watch out for the header magic number
+		 * case, though.
+		 */
+		for (i = 0; i < split; i += BBSIZE) {
+			be32_add_cpu((__be32 *)dptr, 1);
+			if (be32_to_cpu(*(__be32 *)dptr) == XLOG_HEADER_MAGIC_NUM)
+				be32_add_cpu((__be32 *)dptr, 1);
+			dptr += BBSIZE;
+		}
+
+		ASSERT(XFS_BUF_ADDR(bp) <= log->l_logBBsize-1);
+		ASSERT(XFS_BUF_ADDR(bp) + BTOBB(count) <= log->l_logBBsize);
+
+		/* account for internal log which doesn't start at block #0 */
+		XFS_BUF_SET_ADDR(bp, XFS_BUF_ADDR(bp) + log->l_logBBstart);
+		XFS_BUF_WRITE(bp);
+		if ((error = xlog_bdstrat(bp))) {
+			xfs_ioerror_alert("xlog_sync (split)", log->l_mp,
+					  bp, XFS_BUF_ADDR(bp));
+			return error;
+		}
+	}
+	return 0;
+}	/* xlog_sync */
+
+
+/*
+ * Deallocate a log structure
+ */
+STATIC void
+xlog_dealloc_log(xlog_t *log)
+{
+	xlog_in_core_t	*iclog, *next_iclog;
+	int		i;
+
+	xlog_cil_destroy(log);
+
+	/*
+	 * always need to ensure that the extra buffer does not point to memory
+	 * owned by another log buffer before we free it.
+	 */
+	xfs_buf_set_empty(log->l_xbuf, log->l_iclog_size);
+	xfs_buf_free(log->l_xbuf);
+
+	iclog = log->l_iclog;
+	for (i=0; i<log->l_iclog_bufs; i++) {
+		xfs_buf_free(iclog->ic_bp);
+		next_iclog = iclog->ic_next;
+		kmem_free(iclog);
+		iclog = next_iclog;
+	}
+	spinlock_destroy(&log->l_icloglock);
+
+	log->l_mp->m_log = NULL;
+	kmem_free(log);
+}	/* xlog_dealloc_log */
+
+/*
+ * Update counters atomically now that memcpy is done.
+ */
+/* ARGSUSED */
+static inline void
+xlog_state_finish_copy(xlog_t		*log,
+		       xlog_in_core_t	*iclog,
+		       int		record_cnt,
+		       int		copy_bytes)
+{
+	spin_lock(&log->l_icloglock);
+
+	be32_add_cpu(&iclog->ic_header.h_num_logops, record_cnt);
+	iclog->ic_offset += copy_bytes;
+
+	spin_unlock(&log->l_icloglock);
+}	/* xlog_state_finish_copy */
+
+
+
+
+/*
+ * print out info relating to regions written which consume
+ * the reservation
+ */
+void
+xlog_print_tic_res(
+	struct xfs_mount	*mp,
+	struct xlog_ticket	*ticket)
+{
+	uint i;
+	uint ophdr_spc = ticket->t_res_num_ophdrs * (uint)sizeof(xlog_op_header_t);
+
+	/* match with XLOG_REG_TYPE_* in xfs_log.h */
+	static char *res_type_str[XLOG_REG_TYPE_MAX] = {
+	    "bformat",
+	    "bchunk",
+	    "efi_format",
+	    "efd_format",
+	    "iformat",
+	    "icore",
+	    "iext",
+	    "ibroot",
+	    "ilocal",
+	    "iattr_ext",
+	    "iattr_broot",
+	    "iattr_local",
+	    "qformat",
+	    "dquot",
+	    "quotaoff",
+	    "LR header",
+	    "unmount",
+	    "commit",
+	    "trans header"
+	};
+	static char *trans_type_str[XFS_TRANS_TYPE_MAX] = {
+	    "SETATTR_NOT_SIZE",
+	    "SETATTR_SIZE",
+	    "INACTIVE",
+	    "CREATE",
+	    "CREATE_TRUNC",
+	    "TRUNCATE_FILE",
+	    "REMOVE",
+	    "LINK",
+	    "RENAME",
+	    "MKDIR",
+	    "RMDIR",
+	    "SYMLINK",
+	    "SET_DMATTRS",
+	    "GROWFS",
+	    "STRAT_WRITE",
+	    "DIOSTRAT",
+	    "WRITE_SYNC",
+	    "WRITEID",
+	    "ADDAFORK",
+	    "ATTRINVAL",
+	    "ATRUNCATE",
+	    "ATTR_SET",
+	    "ATTR_RM",
+	    "ATTR_FLAG",
+	    "CLEAR_AGI_BUCKET",
+	    "QM_SBCHANGE",
+	    "DUMMY1",
+	    "DUMMY2",
+	    "QM_QUOTAOFF",
+	    "QM_DQALLOC",
+	    "QM_SETQLIM",
+	    "QM_DQCLUSTER",
+	    "QM_QINOCREATE",
+	    "QM_QUOTAOFF_END",
+	    "SB_UNIT",
+	    "FSYNC_TS",
+	    "GROWFSRT_ALLOC",
+	    "GROWFSRT_ZERO",
+	    "GROWFSRT_FREE",
+	    "SWAPEXT"
+	};
+
+	xfs_warn(mp,
+		"xfs_log_write: reservation summary:\n"
+		"  trans type  = %s (%u)\n"
+		"  unit res    = %d bytes\n"
+		"  current res = %d bytes\n"
+		"  total reg   = %u bytes (o/flow = %u bytes)\n"
+		"  ophdrs      = %u (ophdr space = %u bytes)\n"
+		"  ophdr + reg = %u bytes\n"
+		"  num regions = %u\n",
+		((ticket->t_trans_type <= 0 ||
+		  ticket->t_trans_type > XFS_TRANS_TYPE_MAX) ?
+		  "bad-trans-type" : trans_type_str[ticket->t_trans_type-1]),
+		ticket->t_trans_type,
+		ticket->t_unit_res,
+		ticket->t_curr_res,
+		ticket->t_res_arr_sum, ticket->t_res_o_flow,
+		ticket->t_res_num_ophdrs, ophdr_spc,
+		ticket->t_res_arr_sum +
+		ticket->t_res_o_flow + ophdr_spc,
+		ticket->t_res_num);
+
+	for (i = 0; i < ticket->t_res_num; i++) {
+		uint r_type = ticket->t_res_arr[i].r_type;
+		xfs_warn(mp, "region[%u]: %s - %u bytes\n", i,
+			    ((r_type <= 0 || r_type > XLOG_REG_TYPE_MAX) ?
+			    "bad-rtype" : res_type_str[r_type-1]),
+			    ticket->t_res_arr[i].r_len);
+	}
+
+	xfs_alert_tag(mp, XFS_PTAG_LOGRES,
+		"xfs_log_write: reservation ran out. Need to up reservation");
+	xfs_force_shutdown(mp, SHUTDOWN_CORRUPT_INCORE);
+}
+
+/*
+ * Calculate the potential space needed by the log vector.  Each region gets
+ * its own xlog_op_header_t and may need to be double word aligned.
+ */
+static int
+xlog_write_calc_vec_length(
+	struct xlog_ticket	*ticket,
+	struct xfs_log_vec	*log_vector)
+{
+	struct xfs_log_vec	*lv;
+	int			headers = 0;
+	int			len = 0;
+	int			i;
+
+	/* acct for start rec of xact */
+	if (ticket->t_flags & XLOG_TIC_INITED)
+		headers++;
+
+	for (lv = log_vector; lv; lv = lv->lv_next) {
+		headers += lv->lv_niovecs;
+
+		for (i = 0; i < lv->lv_niovecs; i++) {
+			struct xfs_log_iovec	*vecp = &lv->lv_iovecp[i];
+
+			len += vecp->i_len;
+			xlog_tic_add_region(ticket, vecp->i_len, vecp->i_type);
+		}
+	}
+
+	ticket->t_res_num_ophdrs += headers;
+	len += headers * sizeof(struct xlog_op_header);
+
+	return len;
+}
+
+/*
+ * If first write for transaction, insert start record  We can't be trying to
+ * commit if we are inited.  We can't have any "partial_copy" if we are inited.
+ */
+static int
+xlog_write_start_rec(
+	struct xlog_op_header	*ophdr,
+	struct xlog_ticket	*ticket)
+{
+	if (!(ticket->t_flags & XLOG_TIC_INITED))
+		return 0;
+
+	ophdr->oh_tid	= cpu_to_be32(ticket->t_tid);
+	ophdr->oh_clientid = ticket->t_clientid;
+	ophdr->oh_len = 0;
+	ophdr->oh_flags = XLOG_START_TRANS;
+	ophdr->oh_res2 = 0;
+
+	ticket->t_flags &= ~XLOG_TIC_INITED;
+
+	return sizeof(struct xlog_op_header);
+}
+
+static xlog_op_header_t *
+xlog_write_setup_ophdr(
+	struct log		*log,
+	struct xlog_op_header	*ophdr,
+	struct xlog_ticket	*ticket,
+	uint			flags)
+{
+	ophdr->oh_tid = cpu_to_be32(ticket->t_tid);
+	ophdr->oh_clientid = ticket->t_clientid;
+	ophdr->oh_res2 = 0;
+
+	/* are we copying a commit or unmount record? */
+	ophdr->oh_flags = flags;
+
+	/*
+	 * We've seen logs corrupted with bad transaction client ids.  This
+	 * makes sure that XFS doesn't generate them on.  Turn this into an EIO
+	 * and shut down the filesystem.
+	 */
+	switch (ophdr->oh_clientid)  {
+	case XFS_TRANSACTION:
+	case XFS_VOLUME:
+	case XFS_LOG:
+		break;
+	default:
+		xfs_warn(log->l_mp,
+			"Bad XFS transaction clientid 0x%x in ticket 0x%p",
+			ophdr->oh_clientid, ticket);
+		return NULL;
+	}
+
+	return ophdr;
+}
+
+/*
+ * Set up the parameters of the region copy into the log. This has
+ * to handle region write split across multiple log buffers - this
+ * state is kept external to this function so that this code can
+ * can be written in an obvious, self documenting manner.
+ */
+static int
+xlog_write_setup_copy(
+	struct xlog_ticket	*ticket,
+	struct xlog_op_header	*ophdr,
+	int			space_available,
+	int			space_required,
+	int			*copy_off,
+	int			*copy_len,
+	int			*last_was_partial_copy,
+	int			*bytes_consumed)
+{
+	int			still_to_copy;
+
+	still_to_copy = space_required - *bytes_consumed;
+	*copy_off = *bytes_consumed;
+
+	if (still_to_copy <= space_available) {
+		/* write of region completes here */
+		*copy_len = still_to_copy;
+		ophdr->oh_len = cpu_to_be32(*copy_len);
+		if (*last_was_partial_copy)
+			ophdr->oh_flags |= (XLOG_END_TRANS|XLOG_WAS_CONT_TRANS);
+		*last_was_partial_copy = 0;
+		*bytes_consumed = 0;
+		return 0;
+	}
+
+	/* partial write of region, needs extra log op header reservation */
+	*copy_len = space_available;
+	ophdr->oh_len = cpu_to_be32(*copy_len);
+	ophdr->oh_flags |= XLOG_CONTINUE_TRANS;
+	if (*last_was_partial_copy)
+		ophdr->oh_flags |= XLOG_WAS_CONT_TRANS;
+	*bytes_consumed += *copy_len;
+	(*last_was_partial_copy)++;
+
+	/* account for new log op header */
+	ticket->t_curr_res -= sizeof(struct xlog_op_header);
+	ticket->t_res_num_ophdrs++;
+
+	return sizeof(struct xlog_op_header);
+}
+
+static int
+xlog_write_copy_finish(
+	struct log		*log,
+	struct xlog_in_core	*iclog,
+	uint			flags,
+	int			*record_cnt,
+	int			*data_cnt,
+	int			*partial_copy,
+	int			*partial_copy_len,
+	int			log_offset,
+	struct xlog_in_core	**commit_iclog)
+{
+	if (*partial_copy) {
+		/*
+		 * This iclog has already been marked WANT_SYNC by
+		 * xlog_state_get_iclog_space.
+		 */
+		xlog_state_finish_copy(log, iclog, *record_cnt, *data_cnt);
+		*record_cnt = 0;
+		*data_cnt = 0;
+		return xlog_state_release_iclog(log, iclog);
+	}
+
+	*partial_copy = 0;
+	*partial_copy_len = 0;
+
+	if (iclog->ic_size - log_offset <= sizeof(xlog_op_header_t)) {
+		/* no more space in this iclog - push it. */
+		xlog_state_finish_copy(log, iclog, *record_cnt, *data_cnt);
+		*record_cnt = 0;
+		*data_cnt = 0;
+
+		spin_lock(&log->l_icloglock);
+		xlog_state_want_sync(log, iclog);
+		spin_unlock(&log->l_icloglock);
+
+		if (!commit_iclog)
+			return xlog_state_release_iclog(log, iclog);
+		ASSERT(flags & XLOG_COMMIT_TRANS);
+		*commit_iclog = iclog;
+	}
+
+	return 0;
+}
+
+/*
+ * Write some region out to in-core log
+ *
+ * This will be called when writing externally provided regions or when
+ * writing out a commit record for a given transaction.
+ *
+ * General algorithm:
+ *	1. Find total length of this write.  This may include adding to the
+ *		lengths passed in.
+ *	2. Check whether we violate the tickets reservation.
+ *	3. While writing to this iclog
+ *	    A. Reserve as much space in this iclog as can get
+ *	    B. If this is first write, save away start lsn
+ *	    C. While writing this region:
+ *		1. If first write of transaction, write start record
+ *		2. Write log operation header (header per region)
+ *		3. Find out if we can fit entire region into this iclog
+ *		4. Potentially, verify destination memcpy ptr
+ *		5. Memcpy (partial) region
+ *		6. If partial copy, release iclog; otherwise, continue
+ *			copying more regions into current iclog
+ *	4. Mark want sync bit (in simulation mode)
+ *	5. Release iclog for potential flush to on-disk log.
+ *
+ * ERRORS:
+ * 1.	Panic if reservation is overrun.  This should never happen since
+ *	reservation amounts are generated internal to the filesystem.
+ * NOTES:
+ * 1. Tickets are single threaded data structures.
+ * 2. The XLOG_END_TRANS & XLOG_CONTINUE_TRANS flags are passed down to the
+ *	syncing routine.  When a single log_write region needs to span
+ *	multiple in-core logs, the XLOG_CONTINUE_TRANS bit should be set
+ *	on all log operation writes which don't contain the end of the
+ *	region.  The XLOG_END_TRANS bit is used for the in-core log
+ *	operation which contains the end of the continued log_write region.
+ * 3. When xlog_state_get_iclog_space() grabs the rest of the current iclog,
+ *	we don't really know exactly how much space will be used.  As a result,
+ *	we don't update ic_offset until the end when we know exactly how many
+ *	bytes have been written out.
+ */
+int
+xlog_write(
+	struct log		*log,
+	struct xfs_log_vec	*log_vector,
+	struct xlog_ticket	*ticket,
+	xfs_lsn_t		*start_lsn,
+	struct xlog_in_core	**commit_iclog,
+	uint			flags)
+{
+	struct xlog_in_core	*iclog = NULL;
+	struct xfs_log_iovec	*vecp;
+	struct xfs_log_vec	*lv;
+	int			len;
+	int			index;
+	int			partial_copy = 0;
+	int			partial_copy_len = 0;
+	int			contwr = 0;
+	int			record_cnt = 0;
+	int			data_cnt = 0;
+	int			error;
+
+	*start_lsn = 0;
+
+	len = xlog_write_calc_vec_length(ticket, log_vector);
+	if (log->l_cilp) {
+		/*
+		 * Region headers and bytes are already accounted for.
+		 * We only need to take into account start records and
+		 * split regions in this function.
+		 */
+		if (ticket->t_flags & XLOG_TIC_INITED)
+			ticket->t_curr_res -= sizeof(xlog_op_header_t);
+
+		/*
+		 * Commit record headers need to be accounted for. These
+		 * come in as separate writes so are easy to detect.
+		 */
+		if (flags & (XLOG_COMMIT_TRANS | XLOG_UNMOUNT_TRANS))
+			ticket->t_curr_res -= sizeof(xlog_op_header_t);
+	} else
+		ticket->t_curr_res -= len;
+
+	if (ticket->t_curr_res < 0)
+		xlog_print_tic_res(log->l_mp, ticket);
+
+	index = 0;
+	lv = log_vector;
+	vecp = lv->lv_iovecp;
+	while (lv && index < lv->lv_niovecs) {
+		void		*ptr;
+		int		log_offset;
+
+		error = xlog_state_get_iclog_space(log, len, &iclog, ticket,
+						   &contwr, &log_offset);
+		if (error)
+			return error;
+
+		ASSERT(log_offset <= iclog->ic_size - 1);
+		ptr = iclog->ic_datap + log_offset;
+
+		/* start_lsn is the first lsn written to. That's all we need. */
+		if (!*start_lsn)
+			*start_lsn = be64_to_cpu(iclog->ic_header.h_lsn);
+
+		/*
+		 * This loop writes out as many regions as can fit in the amount
+		 * of space which was allocated by xlog_state_get_iclog_space().
+		 */
+		while (lv && index < lv->lv_niovecs) {
+			struct xfs_log_iovec	*reg = &vecp[index];
+			struct xlog_op_header	*ophdr;
+			int			start_rec_copy;
+			int			copy_len;
+			int			copy_off;
+
+			ASSERT(reg->i_len % sizeof(__int32_t) == 0);
+			ASSERT((unsigned long)ptr % sizeof(__int32_t) == 0);
+
+			start_rec_copy = xlog_write_start_rec(ptr, ticket);
+			if (start_rec_copy) {
+				record_cnt++;
+				xlog_write_adv_cnt(&ptr, &len, &log_offset,
+						   start_rec_copy);
+			}
+
+			ophdr = xlog_write_setup_ophdr(log, ptr, ticket, flags);
+			if (!ophdr)
+				return XFS_ERROR(EIO);
+
+			xlog_write_adv_cnt(&ptr, &len, &log_offset,
+					   sizeof(struct xlog_op_header));
+
+			len += xlog_write_setup_copy(ticket, ophdr,
+						     iclog->ic_size-log_offset,
+						     reg->i_len,
+						     &copy_off, &copy_len,
+						     &partial_copy,
+						     &partial_copy_len);
+			xlog_verify_dest_ptr(log, ptr);
+
+			/* copy region */
+			ASSERT(copy_len >= 0);
+			memcpy(ptr, reg->i_addr + copy_off, copy_len);
+			xlog_write_adv_cnt(&ptr, &len, &log_offset, copy_len);
+
+			copy_len += start_rec_copy + sizeof(xlog_op_header_t);
+			record_cnt++;
+			data_cnt += contwr ? copy_len : 0;
+
+			error = xlog_write_copy_finish(log, iclog, flags,
+						       &record_cnt, &data_cnt,
+						       &partial_copy,
+						       &partial_copy_len,
+						       log_offset,
+						       commit_iclog);
+			if (error)
+				return error;
+
+			/*
+			 * if we had a partial copy, we need to get more iclog
+			 * space but we don't want to increment the region
+			 * index because there is still more is this region to
+			 * write.
+			 *
+			 * If we completed writing this region, and we flushed
+			 * the iclog (indicated by resetting of the record
+			 * count), then we also need to get more log space. If
+			 * this was the last record, though, we are done and
+			 * can just return.
+			 */
+			if (partial_copy)
+				break;
+
+			if (++index == lv->lv_niovecs) {
+				lv = lv->lv_next;
+				index = 0;
+				if (lv)
+					vecp = lv->lv_iovecp;
+			}
+			if (record_cnt == 0) {
+				if (!lv)
+					return 0;
+				break;
+			}
+		}
+	}
+
+	ASSERT(len == 0);
+
+	xlog_state_finish_copy(log, iclog, record_cnt, data_cnt);
+	if (!commit_iclog)
+		return xlog_state_release_iclog(log, iclog);
+
+	ASSERT(flags & XLOG_COMMIT_TRANS);
+	*commit_iclog = iclog;
+	return 0;
+}
+
+
+/*****************************************************************************
+ *
+ *		State Machine functions
+ *
+ *****************************************************************************
+ */
+
+/* Clean iclogs starting from the head.  This ordering must be
+ * maintained, so an iclog doesn't become ACTIVE beyond one that
+ * is SYNCING.  This is also required to maintain the notion that we use
+ * a ordered wait queue to hold off would be writers to the log when every
+ * iclog is trying to sync to disk.
+ *
+ * State Change: DIRTY -> ACTIVE
+ */
+STATIC void
+xlog_state_clean_log(xlog_t *log)
+{
+	xlog_in_core_t	*iclog;
+	int changed = 0;
+
+	iclog = log->l_iclog;
+	do {
+		if (iclog->ic_state == XLOG_STATE_DIRTY) {
+			iclog->ic_state	= XLOG_STATE_ACTIVE;
+			iclog->ic_offset       = 0;
+			ASSERT(iclog->ic_callback == NULL);
+			/*
+			 * If the number of ops in this iclog indicate it just
+			 * contains the dummy transaction, we can
+			 * change state into IDLE (the second time around).
+			 * Otherwise we should change the state into
+			 * NEED a dummy.
+			 * We don't need to cover the dummy.
+			 */
+			if (!changed &&
+			   (be32_to_cpu(iclog->ic_header.h_num_logops) ==
+			   		XLOG_COVER_OPS)) {
+				changed = 1;
+			} else {
+				/*
+				 * We have two dirty iclogs so start over
+				 * This could also be num of ops indicates
+				 * this is not the dummy going out.
+				 */
+				changed = 2;
+			}
+			iclog->ic_header.h_num_logops = 0;
+			memset(iclog->ic_header.h_cycle_data, 0,
+			      sizeof(iclog->ic_header.h_cycle_data));
+			iclog->ic_header.h_lsn = 0;
+		} else if (iclog->ic_state == XLOG_STATE_ACTIVE)
+			/* do nothing */;
+		else
+			break;	/* stop cleaning */
+		iclog = iclog->ic_next;
+	} while (iclog != log->l_iclog);
+
+	/* log is locked when we are called */
+	/*
+	 * Change state for the dummy log recording.
+	 * We usually go to NEED. But we go to NEED2 if the changed indicates
+	 * we are done writing the dummy record.
+	 * If we are done with the second dummy recored (DONE2), then
+	 * we go to IDLE.
+	 */
+	if (changed) {
+		switch (log->l_covered_state) {
+		case XLOG_STATE_COVER_IDLE:
+		case XLOG_STATE_COVER_NEED:
+		case XLOG_STATE_COVER_NEED2:
+			log->l_covered_state = XLOG_STATE_COVER_NEED;
+			break;
+
+		case XLOG_STATE_COVER_DONE:
+			if (changed == 1)
+				log->l_covered_state = XLOG_STATE_COVER_NEED2;
+			else
+				log->l_covered_state = XLOG_STATE_COVER_NEED;
+			break;
+
+		case XLOG_STATE_COVER_DONE2:
+			if (changed == 1)
+				log->l_covered_state = XLOG_STATE_COVER_IDLE;
+			else
+				log->l_covered_state = XLOG_STATE_COVER_NEED;
+			break;
+
+		default:
+			ASSERT(0);
+		}
+	}
+}	/* xlog_state_clean_log */
+
+STATIC xfs_lsn_t
+xlog_get_lowest_lsn(
+	xlog_t		*log)
+{
+	xlog_in_core_t  *lsn_log;
+	xfs_lsn_t	lowest_lsn, lsn;
+
+	lsn_log = log->l_iclog;
+	lowest_lsn = 0;
+	do {
+	    if (!(lsn_log->ic_state & (XLOG_STATE_ACTIVE|XLOG_STATE_DIRTY))) {
+		lsn = be64_to_cpu(lsn_log->ic_header.h_lsn);
+		if ((lsn && !lowest_lsn) ||
+		    (XFS_LSN_CMP(lsn, lowest_lsn) < 0)) {
+			lowest_lsn = lsn;
+		}
+	    }
+	    lsn_log = lsn_log->ic_next;
+	} while (lsn_log != log->l_iclog);
+	return lowest_lsn;
+}
+
+
+STATIC void
+xlog_state_do_callback(
+	xlog_t		*log,
+	int		aborted,
+	xlog_in_core_t	*ciclog)
+{
+	xlog_in_core_t	   *iclog;
+	xlog_in_core_t	   *first_iclog;	/* used to know when we've
+						 * processed all iclogs once */
+	xfs_log_callback_t *cb, *cb_next;
+	int		   flushcnt = 0;
+	xfs_lsn_t	   lowest_lsn;
+	int		   ioerrors;	/* counter: iclogs with errors */
+	int		   loopdidcallbacks; /* flag: inner loop did callbacks*/
+	int		   funcdidcallbacks; /* flag: function did callbacks */
+	int		   repeats;	/* for issuing console warnings if
+					 * looping too many times */
+	int		   wake = 0;
+
+	spin_lock(&log->l_icloglock);
+	first_iclog = iclog = log->l_iclog;
+	ioerrors = 0;
+	funcdidcallbacks = 0;
+	repeats = 0;
+
+	do {
+		/*
+		 * Scan all iclogs starting with the one pointed to by the
+		 * log.  Reset this starting point each time the log is
+		 * unlocked (during callbacks).
+		 *
+		 * Keep looping through iclogs until one full pass is made
+		 * without running any callbacks.
+		 */
+		first_iclog = log->l_iclog;
+		iclog = log->l_iclog;
+		loopdidcallbacks = 0;
+		repeats++;
+
+		do {
+
+			/* skip all iclogs in the ACTIVE & DIRTY states */
+			if (iclog->ic_state &
+			    (XLOG_STATE_ACTIVE|XLOG_STATE_DIRTY)) {
+				iclog = iclog->ic_next;
+				continue;
+			}
+
+			/*
+			 * Between marking a filesystem SHUTDOWN and stopping
+			 * the log, we do flush all iclogs to disk (if there
+			 * wasn't a log I/O error). So, we do want things to
+			 * go smoothly in case of just a SHUTDOWN  w/o a
+			 * LOG_IO_ERROR.
+			 */
+			if (!(iclog->ic_state & XLOG_STATE_IOERROR)) {
+				/*
+				 * Can only perform callbacks in order.  Since
+				 * this iclog is not in the DONE_SYNC/
+				 * DO_CALLBACK state, we skip the rest and
+				 * just try to clean up.  If we set our iclog
+				 * to DO_CALLBACK, we will not process it when
+				 * we retry since a previous iclog is in the
+				 * CALLBACK and the state cannot change since
+				 * we are holding the l_icloglock.
+				 */
+				if (!(iclog->ic_state &
+					(XLOG_STATE_DONE_SYNC |
+						 XLOG_STATE_DO_CALLBACK))) {
+					if (ciclog && (ciclog->ic_state ==
+							XLOG_STATE_DONE_SYNC)) {
+						ciclog->ic_state = XLOG_STATE_DO_CALLBACK;
+					}
+					break;
+				}
+				/*
+				 * We now have an iclog that is in either the
+				 * DO_CALLBACK or DONE_SYNC states. The other
+				 * states (WANT_SYNC, SYNCING, or CALLBACK were
+				 * caught by the above if and are going to
+				 * clean (i.e. we aren't doing their callbacks)
+				 * see the above if.
+				 */
+
+				/*
+				 * We will do one more check here to see if we
+				 * have chased our tail around.
+				 */
+
+				lowest_lsn = xlog_get_lowest_lsn(log);
+				if (lowest_lsn &&
+				    XFS_LSN_CMP(lowest_lsn,
+						be64_to_cpu(iclog->ic_header.h_lsn)) < 0) {
+					iclog = iclog->ic_next;
+					continue; /* Leave this iclog for
+						   * another thread */
+				}
+
+				iclog->ic_state = XLOG_STATE_CALLBACK;
+
+
+				/*
+				 * update the last_sync_lsn before we drop the
+				 * icloglock to ensure we are the only one that
+				 * can update it.
+				 */
+				ASSERT(XFS_LSN_CMP(atomic64_read(&log->l_last_sync_lsn),
+					be64_to_cpu(iclog->ic_header.h_lsn)) <= 0);
+				atomic64_set(&log->l_last_sync_lsn,
+					be64_to_cpu(iclog->ic_header.h_lsn));
+
+			} else
+				ioerrors++;
+
+			spin_unlock(&log->l_icloglock);
+
+			/*
+			 * Keep processing entries in the callback list until
+			 * we come around and it is empty.  We need to
+			 * atomically see that the list is empty and change the
+			 * state to DIRTY so that we don't miss any more
+			 * callbacks being added.
+			 */
+			spin_lock(&iclog->ic_callback_lock);
+			cb = iclog->ic_callback;
+			while (cb) {
+				iclog->ic_callback_tail = &(iclog->ic_callback);
+				iclog->ic_callback = NULL;
+				spin_unlock(&iclog->ic_callback_lock);
+
+				/* perform callbacks in the order given */
+				for (; cb; cb = cb_next) {
+					cb_next = cb->cb_next;
+					cb->cb_func(cb->cb_arg, aborted);
+				}
+				spin_lock(&iclog->ic_callback_lock);
+				cb = iclog->ic_callback;
+			}
+
+			loopdidcallbacks++;
+			funcdidcallbacks++;
+
+			spin_lock(&log->l_icloglock);
+			ASSERT(iclog->ic_callback == NULL);
+			spin_unlock(&iclog->ic_callback_lock);
+			if (!(iclog->ic_state & XLOG_STATE_IOERROR))
+				iclog->ic_state = XLOG_STATE_DIRTY;
+
+			/*
+			 * Transition from DIRTY to ACTIVE if applicable.
+			 * NOP if STATE_IOERROR.
+			 */
+			xlog_state_clean_log(log);
+
+			/* wake up threads waiting in xfs_log_force() */
+			wake_up_all(&iclog->ic_force_wait);
+
+			iclog = iclog->ic_next;
+		} while (first_iclog != iclog);
+
+		if (repeats > 5000) {
+			flushcnt += repeats;
+			repeats = 0;
+			xfs_warn(log->l_mp,
+				"%s: possible infinite loop (%d iterations)",
+				__func__, flushcnt);
+		}
+	} while (!ioerrors && loopdidcallbacks);
+
+	/*
+	 * make one last gasp attempt to see if iclogs are being left in
+	 * limbo..
+	 */
+#ifdef DEBUG
+	if (funcdidcallbacks) {
+		first_iclog = iclog = log->l_iclog;
+		do {
+			ASSERT(iclog->ic_state != XLOG_STATE_DO_CALLBACK);
+			/*
+			 * Terminate the loop if iclogs are found in states
+			 * which will cause other threads to clean up iclogs.
+			 *
+			 * SYNCING - i/o completion will go through logs
+			 * DONE_SYNC - interrupt thread should be waiting for
+			 *              l_icloglock
+			 * IOERROR - give up hope all ye who enter here
+			 */
+			if (iclog->ic_state == XLOG_STATE_WANT_SYNC ||
+			    iclog->ic_state == XLOG_STATE_SYNCING ||
+			    iclog->ic_state == XLOG_STATE_DONE_SYNC ||
+			    iclog->ic_state == XLOG_STATE_IOERROR )
+				break;
+			iclog = iclog->ic_next;
+		} while (first_iclog != iclog);
+	}
+#endif
+
+	if (log->l_iclog->ic_state & (XLOG_STATE_ACTIVE|XLOG_STATE_IOERROR))
+		wake = 1;
+	spin_unlock(&log->l_icloglock);
+
+	if (wake)
+		wake_up_all(&log->l_flush_wait);
+}
+
+
+/*
+ * Finish transitioning this iclog to the dirty state.
+ *
+ * Make sure that we completely execute this routine only when this is
+ * the last call to the iclog.  There is a good chance that iclog flushes,
+ * when we reach the end of the physical log, get turned into 2 separate
+ * calls to bwrite.  Hence, one iclog flush could generate two calls to this
+ * routine.  By using the reference count bwritecnt, we guarantee that only
+ * the second completion goes through.
+ *
+ * Callbacks could take time, so they are done outside the scope of the
+ * global state machine log lock.
+ */
+STATIC void
+xlog_state_done_syncing(
+	xlog_in_core_t	*iclog,
+	int		aborted)
+{
+	xlog_t		   *log = iclog->ic_log;
+
+	spin_lock(&log->l_icloglock);
+
+	ASSERT(iclog->ic_state == XLOG_STATE_SYNCING ||
+	       iclog->ic_state == XLOG_STATE_IOERROR);
+	ASSERT(atomic_read(&iclog->ic_refcnt) == 0);
+	ASSERT(iclog->ic_bwritecnt == 1 || iclog->ic_bwritecnt == 2);
+
+
+	/*
+	 * If we got an error, either on the first buffer, or in the case of
+	 * split log writes, on the second, we mark ALL iclogs STATE_IOERROR,
+	 * and none should ever be attempted to be written to disk
+	 * again.
+	 */
+	if (iclog->ic_state != XLOG_STATE_IOERROR) {
+		if (--iclog->ic_bwritecnt == 1) {
+			spin_unlock(&log->l_icloglock);
+			return;
+		}
+		iclog->ic_state = XLOG_STATE_DONE_SYNC;
+	}
+
+	/*
+	 * Someone could be sleeping prior to writing out the next
+	 * iclog buffer, we wake them all, one will get to do the
+	 * I/O, the others get to wait for the result.
+	 */
+	wake_up_all(&iclog->ic_write_wait);
+	spin_unlock(&log->l_icloglock);
+	xlog_state_do_callback(log, aborted, iclog);	/* also cleans log */
+}	/* xlog_state_done_syncing */
+
+
+/*
+ * If the head of the in-core log ring is not (ACTIVE or DIRTY), then we must
+ * sleep.  We wait on the flush queue on the head iclog as that should be
+ * the first iclog to complete flushing. Hence if all iclogs are syncing,
+ * we will wait here and all new writes will sleep until a sync completes.
+ *
+ * The in-core logs are used in a circular fashion. They are not used
+ * out-of-order even when an iclog past the head is free.
+ *
+ * return:
+ *	* log_offset where xlog_write() can start writing into the in-core
+ *		log's data space.
+ *	* in-core log pointer to which xlog_write() should write.
+ *	* boolean indicating this is a continued write to an in-core log.
+ *		If this is the last write, then the in-core log's offset field
+ *		needs to be incremented, depending on the amount of data which
+ *		is copied.
+ */
+STATIC int
+xlog_state_get_iclog_space(xlog_t	  *log,
+			   int		  len,
+			   xlog_in_core_t **iclogp,
+			   xlog_ticket_t  *ticket,
+			   int		  *continued_write,
+			   int		  *logoffsetp)
+{
+	int		  log_offset;
+	xlog_rec_header_t *head;
+	xlog_in_core_t	  *iclog;
+	int		  error;
+
+restart:
+	spin_lock(&log->l_icloglock);
+	if (XLOG_FORCED_SHUTDOWN(log)) {
+		spin_unlock(&log->l_icloglock);
+		return XFS_ERROR(EIO);
+	}
+
+	iclog = log->l_iclog;
+	if (iclog->ic_state != XLOG_STATE_ACTIVE) {
+		XFS_STATS_INC(xs_log_noiclogs);
+
+		/* Wait for log writes to have flushed */
+		xlog_wait(&log->l_flush_wait, &log->l_icloglock);
+		goto restart;
+	}
+
+	head = &iclog->ic_header;
+
+	atomic_inc(&iclog->ic_refcnt);	/* prevents sync */
+	log_offset = iclog->ic_offset;
+
+	/* On the 1st write to an iclog, figure out lsn.  This works
+	 * if iclogs marked XLOG_STATE_WANT_SYNC always write out what they are
+	 * committing to.  If the offset is set, that's how many blocks
+	 * must be written.
+	 */
+	if (log_offset == 0) {
+		ticket->t_curr_res -= log->l_iclog_hsize;
+		xlog_tic_add_region(ticket,
+				    log->l_iclog_hsize,
+				    XLOG_REG_TYPE_LRHEADER);
+		head->h_cycle = cpu_to_be32(log->l_curr_cycle);
+		head->h_lsn = cpu_to_be64(
+			xlog_assign_lsn(log->l_curr_cycle, log->l_curr_block));
+		ASSERT(log->l_curr_block >= 0);
+	}
+
+	/* If there is enough room to write everything, then do it.  Otherwise,
+	 * claim the rest of the region and make sure the XLOG_STATE_WANT_SYNC
+	 * bit is on, so this will get flushed out.  Don't update ic_offset
+	 * until you know exactly how many bytes get copied.  Therefore, wait
+	 * until later to update ic_offset.
+	 *
+	 * xlog_write() algorithm assumes that at least 2 xlog_op_header_t's
+	 * can fit into remaining data section.
+	 */
+	if (iclog->ic_size - iclog->ic_offset < 2*sizeof(xlog_op_header_t)) {
+		xlog_state_switch_iclogs(log, iclog, iclog->ic_size);
+
+		/*
+		 * If I'm the only one writing to this iclog, sync it to disk.
+		 * We need to do an atomic compare and decrement here to avoid
+		 * racing with concurrent atomic_dec_and_lock() calls in
+		 * xlog_state_release_iclog() when there is more than one
+		 * reference to the iclog.
+		 */
+		if (!atomic_add_unless(&iclog->ic_refcnt, -1, 1)) {
+			/* we are the only one */
+			spin_unlock(&log->l_icloglock);
+			error = xlog_state_release_iclog(log, iclog);
+			if (error)
+				return error;
+		} else {
+			spin_unlock(&log->l_icloglock);
+		}
+		goto restart;
+	}
+
+	/* Do we have enough room to write the full amount in the remainder
+	 * of this iclog?  Or must we continue a write on the next iclog and
+	 * mark this iclog as completely taken?  In the case where we switch
+	 * iclogs (to mark it taken), this particular iclog will release/sync
+	 * to disk in xlog_write().
+	 */
+	if (len <= iclog->ic_size - iclog->ic_offset) {
+		*continued_write = 0;
+		iclog->ic_offset += len;
+	} else {
+		*continued_write = 1;
+		xlog_state_switch_iclogs(log, iclog, iclog->ic_size);
+	}
+	*iclogp = iclog;
+
+	ASSERT(iclog->ic_offset <= iclog->ic_size);
+	spin_unlock(&log->l_icloglock);
+
+	*logoffsetp = log_offset;
+	return 0;
+}	/* xlog_state_get_iclog_space */
+
+/*
+ * Atomically get the log space required for a log ticket.
+ *
+ * Once a ticket gets put onto the reserveq, it will only return after
+ * the needed reservation is satisfied.
+ *
+ * This function is structured so that it has a lock free fast path. This is
+ * necessary because every new transaction reservation will come through this
+ * path. Hence any lock will be globally hot if we take it unconditionally on
+ * every pass.
+ *
+ * As tickets are only ever moved on and off the reserveq under the
+ * l_grant_reserve_lock, we only need to take that lock if we are going
+ * to add the ticket to the queue and sleep. We can avoid taking the lock if the
+ * ticket was never added to the reserveq because the t_queue list head will be
+ * empty and we hold the only reference to it so it can safely be checked
+ * unlocked.
+ */
+STATIC int
+xlog_grant_log_space(xlog_t	   *log,
+		     xlog_ticket_t *tic)
+{
+	int		 free_bytes;
+	int		 need_bytes;
+
+#ifdef DEBUG
+	if (log->l_flags & XLOG_ACTIVE_RECOVERY)
+		panic("grant Recovery problem");
+#endif
+
+	trace_xfs_log_grant_enter(log, tic);
+
+	need_bytes = tic->t_unit_res;
+	if (tic->t_flags & XFS_LOG_PERM_RESERV)
+		need_bytes *= tic->t_ocnt;
+
+	/* something is already sleeping; insert new transaction at end */
+	if (!list_empty_careful(&log->l_reserveq)) {
+		spin_lock(&log->l_grant_reserve_lock);
+		/* recheck the queue now we are locked */
+		if (list_empty(&log->l_reserveq)) {
+			spin_unlock(&log->l_grant_reserve_lock);
+			goto redo;
+		}
+		list_add_tail(&tic->t_queue, &log->l_reserveq);
+
+		trace_xfs_log_grant_sleep1(log, tic);
+
+		/*
+		 * Gotta check this before going to sleep, while we're
+		 * holding the grant lock.
+		 */
+		if (XLOG_FORCED_SHUTDOWN(log))
+			goto error_return;
+
+		XFS_STATS_INC(xs_sleep_logspace);
+		xlog_wait(&tic->t_wait, &log->l_grant_reserve_lock);
+
+		/*
+		 * If we got an error, and the filesystem is shutting down,
+		 * we'll catch it down below. So just continue...
+		 */
+		trace_xfs_log_grant_wake1(log, tic);
+	}
+
+redo:
+	if (XLOG_FORCED_SHUTDOWN(log))
+		goto error_return_unlocked;
+
+	free_bytes = xlog_space_left(log, &log->l_grant_reserve_head);
+	if (free_bytes < need_bytes) {
+		spin_lock(&log->l_grant_reserve_lock);
+		if (list_empty(&tic->t_queue))
+			list_add_tail(&tic->t_queue, &log->l_reserveq);
+
+		trace_xfs_log_grant_sleep2(log, tic);
+
+		if (XLOG_FORCED_SHUTDOWN(log))
+			goto error_return;
+
+		xlog_grant_push_ail(log, need_bytes);
+
+		XFS_STATS_INC(xs_sleep_logspace);
+		xlog_wait(&tic->t_wait, &log->l_grant_reserve_lock);
+
+		trace_xfs_log_grant_wake2(log, tic);
+		goto redo;
+	}
+
+	if (!list_empty(&tic->t_queue)) {
+		spin_lock(&log->l_grant_reserve_lock);
+		list_del_init(&tic->t_queue);
+		spin_unlock(&log->l_grant_reserve_lock);
+	}
+
+	/* we've got enough space */
+	xlog_grant_add_space(log, &log->l_grant_reserve_head, need_bytes);
+	xlog_grant_add_space(log, &log->l_grant_write_head, need_bytes);
+	trace_xfs_log_grant_exit(log, tic);
+	xlog_verify_grant_tail(log);
+	return 0;
+
+error_return_unlocked:
+	spin_lock(&log->l_grant_reserve_lock);
+error_return:
+	list_del_init(&tic->t_queue);
+	spin_unlock(&log->l_grant_reserve_lock);
+	trace_xfs_log_grant_error(log, tic);
+
+	/*
+	 * If we are failing, make sure the ticket doesn't have any
+	 * current reservations. We don't want to add this back when
+	 * the ticket/transaction gets cancelled.
+	 */
+	tic->t_curr_res = 0;
+	tic->t_cnt = 0; /* ungrant will give back unit_res * t_cnt. */
+	return XFS_ERROR(EIO);
+}	/* xlog_grant_log_space */
+
+
+/*
+ * Replenish the byte reservation required by moving the grant write head.
+ *
+ * Similar to xlog_grant_log_space, the function is structured to have a lock
+ * free fast path.
+ */
+STATIC int
+xlog_regrant_write_log_space(xlog_t	   *log,
+			     xlog_ticket_t *tic)
+{
+	int		free_bytes, need_bytes;
+
+	tic->t_curr_res = tic->t_unit_res;
+	xlog_tic_reset_res(tic);
+
+	if (tic->t_cnt > 0)
+		return 0;
+
+#ifdef DEBUG
+	if (log->l_flags & XLOG_ACTIVE_RECOVERY)
+		panic("regrant Recovery problem");
+#endif
+
+	trace_xfs_log_regrant_write_enter(log, tic);
+	if (XLOG_FORCED_SHUTDOWN(log))
+		goto error_return_unlocked;
+
+	/* If there are other waiters on the queue then give them a
+	 * chance at logspace before us. Wake up the first waiters,
+	 * if we do not wake up all the waiters then go to sleep waiting
+	 * for more free space, otherwise try to get some space for
+	 * this transaction.
+	 */
+	need_bytes = tic->t_unit_res;
+	if (!list_empty_careful(&log->l_writeq)) {
+		struct xlog_ticket *ntic;
+
+		spin_lock(&log->l_grant_write_lock);
+		free_bytes = xlog_space_left(log, &log->l_grant_write_head);
+		list_for_each_entry(ntic, &log->l_writeq, t_queue) {
+			ASSERT(ntic->t_flags & XLOG_TIC_PERM_RESERV);
+
+			if (free_bytes < ntic->t_unit_res)
+				break;
+			free_bytes -= ntic->t_unit_res;
+			wake_up(&ntic->t_wait);
+		}
+
+		if (ntic != list_first_entry(&log->l_writeq,
+						struct xlog_ticket, t_queue)) {
+			if (list_empty(&tic->t_queue))
+				list_add_tail(&tic->t_queue, &log->l_writeq);
+			trace_xfs_log_regrant_write_sleep1(log, tic);
+
+			xlog_grant_push_ail(log, need_bytes);
+
+			XFS_STATS_INC(xs_sleep_logspace);
+			xlog_wait(&tic->t_wait, &log->l_grant_write_lock);
+			trace_xfs_log_regrant_write_wake1(log, tic);
+		} else
+			spin_unlock(&log->l_grant_write_lock);
+	}
+
+redo:
+	if (XLOG_FORCED_SHUTDOWN(log))
+		goto error_return_unlocked;
+
+	free_bytes = xlog_space_left(log, &log->l_grant_write_head);
+	if (free_bytes < need_bytes) {
+		spin_lock(&log->l_grant_write_lock);
+		if (list_empty(&tic->t_queue))
+			list_add_tail(&tic->t_queue, &log->l_writeq);
+
+		if (XLOG_FORCED_SHUTDOWN(log))
+			goto error_return;
+
+		xlog_grant_push_ail(log, need_bytes);
+
+		XFS_STATS_INC(xs_sleep_logspace);
+		trace_xfs_log_regrant_write_sleep2(log, tic);
+		xlog_wait(&tic->t_wait, &log->l_grant_write_lock);
+
+		trace_xfs_log_regrant_write_wake2(log, tic);
+		goto redo;
+	}
+
+	if (!list_empty(&tic->t_queue)) {
+		spin_lock(&log->l_grant_write_lock);
+		list_del_init(&tic->t_queue);
+		spin_unlock(&log->l_grant_write_lock);
+	}
+
+	/* we've got enough space */
+	xlog_grant_add_space(log, &log->l_grant_write_head, need_bytes);
+	trace_xfs_log_regrant_write_exit(log, tic);
+	xlog_verify_grant_tail(log);
+	return 0;
+
+
+ error_return_unlocked:
+	spin_lock(&log->l_grant_write_lock);
+ error_return:
+	list_del_init(&tic->t_queue);
+	spin_unlock(&log->l_grant_write_lock);
+	trace_xfs_log_regrant_write_error(log, tic);
+
+	/*
+	 * If we are failing, make sure the ticket doesn't have any
+	 * current reservations. We don't want to add this back when
+	 * the ticket/transaction gets cancelled.
+	 */
+	tic->t_curr_res = 0;
+	tic->t_cnt = 0; /* ungrant will give back unit_res * t_cnt. */
+	return XFS_ERROR(EIO);
+}	/* xlog_regrant_write_log_space */
+
+
+/* The first cnt-1 times through here we don't need to
+ * move the grant write head because the permanent
+ * reservation has reserved cnt times the unit amount.
+ * Release part of current permanent unit reservation and
+ * reset current reservation to be one units worth.  Also
+ * move grant reservation head forward.
+ */
+STATIC void
+xlog_regrant_reserve_log_space(xlog_t	     *log,
+			       xlog_ticket_t *ticket)
+{
+	trace_xfs_log_regrant_reserve_enter(log, ticket);
+
+	if (ticket->t_cnt > 0)
+		ticket->t_cnt--;
+
+	xlog_grant_sub_space(log, &log->l_grant_reserve_head,
+					ticket->t_curr_res);
+	xlog_grant_sub_space(log, &log->l_grant_write_head,
+					ticket->t_curr_res);
+	ticket->t_curr_res = ticket->t_unit_res;
+	xlog_tic_reset_res(ticket);
+
+	trace_xfs_log_regrant_reserve_sub(log, ticket);
+
+	/* just return if we still have some of the pre-reserved space */
+	if (ticket->t_cnt > 0)
+		return;
+
+	xlog_grant_add_space(log, &log->l_grant_reserve_head,
+					ticket->t_unit_res);
+
+	trace_xfs_log_regrant_reserve_exit(log, ticket);
+
+	ticket->t_curr_res = ticket->t_unit_res;
+	xlog_tic_reset_res(ticket);
+}	/* xlog_regrant_reserve_log_space */
+
+
+/*
+ * Give back the space left from a reservation.
+ *
+ * All the information we need to make a correct determination of space left
+ * is present.  For non-permanent reservations, things are quite easy.  The
+ * count should have been decremented to zero.  We only need to deal with the
+ * space remaining in the current reservation part of the ticket.  If the
+ * ticket contains a permanent reservation, there may be left over space which
+ * needs to be released.  A count of N means that N-1 refills of the current
+ * reservation can be done before we need to ask for more space.  The first
+ * one goes to fill up the first current reservation.  Once we run out of
+ * space, the count will stay at zero and the only space remaining will be
+ * in the current reservation field.
+ */
+STATIC void
+xlog_ungrant_log_space(xlog_t	     *log,
+		       xlog_ticket_t *ticket)
+{
+	int	bytes;
+
+	if (ticket->t_cnt > 0)
+		ticket->t_cnt--;
+
+	trace_xfs_log_ungrant_enter(log, ticket);
+	trace_xfs_log_ungrant_sub(log, ticket);
+
+	/*
+	 * If this is a permanent reservation ticket, we may be able to free
+	 * up more space based on the remaining count.
+	 */
+	bytes = ticket->t_curr_res;
+	if (ticket->t_cnt > 0) {
+		ASSERT(ticket->t_flags & XLOG_TIC_PERM_RESERV);
+		bytes += ticket->t_unit_res*ticket->t_cnt;
+	}
+
+	xlog_grant_sub_space(log, &log->l_grant_reserve_head, bytes);
+	xlog_grant_sub_space(log, &log->l_grant_write_head, bytes);
+
+	trace_xfs_log_ungrant_exit(log, ticket);
+
+	xfs_log_move_tail(log->l_mp, 1);
+}	/* xlog_ungrant_log_space */
+
+
+/*
+ * Flush iclog to disk if this is the last reference to the given iclog and
+ * the WANT_SYNC bit is set.
+ *
+ * When this function is entered, the iclog is not necessarily in the
+ * WANT_SYNC state.  It may be sitting around waiting to get filled.
+ *
+ *
+ */
+STATIC int
+xlog_state_release_iclog(
+	xlog_t		*log,
+	xlog_in_core_t	*iclog)
+{
+	int		sync = 0;	/* do we sync? */
+
+	if (iclog->ic_state & XLOG_STATE_IOERROR)
+		return XFS_ERROR(EIO);
+
+	ASSERT(atomic_read(&iclog->ic_refcnt) > 0);
+	if (!atomic_dec_and_lock(&iclog->ic_refcnt, &log->l_icloglock))
+		return 0;
+
+	if (iclog->ic_state & XLOG_STATE_IOERROR) {
+		spin_unlock(&log->l_icloglock);
+		return XFS_ERROR(EIO);
+	}
+	ASSERT(iclog->ic_state == XLOG_STATE_ACTIVE ||
+	       iclog->ic_state == XLOG_STATE_WANT_SYNC);
+
+	if (iclog->ic_state == XLOG_STATE_WANT_SYNC) {
+		/* update tail before writing to iclog */
+		xfs_lsn_t tail_lsn = xlog_assign_tail_lsn(log->l_mp);
+		sync++;
+		iclog->ic_state = XLOG_STATE_SYNCING;
+		iclog->ic_header.h_tail_lsn = cpu_to_be64(tail_lsn);
+		xlog_verify_tail_lsn(log, iclog, tail_lsn);
+		/* cycle incremented when incrementing curr_block */
+	}
+	spin_unlock(&log->l_icloglock);
+
+	/*
+	 * We let the log lock go, so it's possible that we hit a log I/O
+	 * error or some other SHUTDOWN condition that marks the iclog
+	 * as XLOG_STATE_IOERROR before the bwrite. However, we know that
+	 * this iclog has consistent data, so we ignore IOERROR
+	 * flags after this point.
+	 */
+	if (sync)
+		return xlog_sync(log, iclog);
+	return 0;
+}	/* xlog_state_release_iclog */
+
+
+/*
+ * This routine will mark the current iclog in the ring as WANT_SYNC
+ * and move the current iclog pointer to the next iclog in the ring.
+ * When this routine is called from xlog_state_get_iclog_space(), the
+ * exact size of the iclog has not yet been determined.  All we know is
+ * that every data block.  We have run out of space in this log record.
+ */
+STATIC void
+xlog_state_switch_iclogs(xlog_t		*log,
+			 xlog_in_core_t *iclog,
+			 int		eventual_size)
+{
+	ASSERT(iclog->ic_state == XLOG_STATE_ACTIVE);
+	if (!eventual_size)
+		eventual_size = iclog->ic_offset;
+	iclog->ic_state = XLOG_STATE_WANT_SYNC;
+	iclog->ic_header.h_prev_block = cpu_to_be32(log->l_prev_block);
+	log->l_prev_block = log->l_curr_block;
+	log->l_prev_cycle = log->l_curr_cycle;
+
+	/* roll log?: ic_offset changed later */
+	log->l_curr_block += BTOBB(eventual_size)+BTOBB(log->l_iclog_hsize);
+
+	/* Round up to next log-sunit */
+	if (xfs_sb_version_haslogv2(&log->l_mp->m_sb) &&
+	    log->l_mp->m_sb.sb_logsunit > 1) {
+		__uint32_t sunit_bb = BTOBB(log->l_mp->m_sb.sb_logsunit);
+		log->l_curr_block = roundup(log->l_curr_block, sunit_bb);
+	}
+
+	if (log->l_curr_block >= log->l_logBBsize) {
+		log->l_curr_cycle++;
+		if (log->l_curr_cycle == XLOG_HEADER_MAGIC_NUM)
+			log->l_curr_cycle++;
+		log->l_curr_block -= log->l_logBBsize;
+		ASSERT(log->l_curr_block >= 0);
+	}
+	ASSERT(iclog == log->l_iclog);
+	log->l_iclog = iclog->ic_next;
+}	/* xlog_state_switch_iclogs */
+
+/*
+ * Write out all data in the in-core log as of this exact moment in time.
+ *
+ * Data may be written to the in-core log during this call.  However,
+ * we don't guarantee this data will be written out.  A change from past
+ * implementation means this routine will *not* write out zero length LRs.
+ *
+ * Basically, we try and perform an intelligent scan of the in-core logs.
+ * If we determine there is no flushable data, we just return.  There is no
+ * flushable data if:
+ *
+ *	1. the current iclog is active and has no data; the previous iclog
+ *		is in the active or dirty state.
+ *	2. the current iclog is drity, and the previous iclog is in the
+ *		active or dirty state.
+ *
+ * We may sleep if:
+ *
+ *	1. the current iclog is not in the active nor dirty state.
+ *	2. the current iclog dirty, and the previous iclog is not in the
+ *		active nor dirty state.
+ *	3. the current iclog is active, and there is another thread writing
+ *		to this particular iclog.
+ *	4. a) the current iclog is active and has no other writers
+ *	   b) when we return from flushing out this iclog, it is still
+ *		not in the active nor dirty state.
+ */
+int
+_xfs_log_force(
+	struct xfs_mount	*mp,
+	uint			flags,
+	int			*log_flushed)
+{
+	struct log		*log = mp->m_log;
+	struct xlog_in_core	*iclog;
+	xfs_lsn_t		lsn;
+
+	XFS_STATS_INC(xs_log_force);
+
+	if (log->l_cilp)
+		xlog_cil_force(log);
+
+	spin_lock(&log->l_icloglock);
+
+	iclog = log->l_iclog;
+	if (iclog->ic_state & XLOG_STATE_IOERROR) {
+		spin_unlock(&log->l_icloglock);
+		return XFS_ERROR(EIO);
+	}
+
+	/* If the head iclog is not active nor dirty, we just attach
+	 * ourselves to the head and go to sleep.
+	 */
+	if (iclog->ic_state == XLOG_STATE_ACTIVE ||
+	    iclog->ic_state == XLOG_STATE_DIRTY) {
+		/*
+		 * If the head is dirty or (active and empty), then
+		 * we need to look at the previous iclog.  If the previous
+		 * iclog is active or dirty we are done.  There is nothing
+		 * to sync out.  Otherwise, we attach ourselves to the
+		 * previous iclog and go to sleep.
+		 */
+		if (iclog->ic_state == XLOG_STATE_DIRTY ||
+		    (atomic_read(&iclog->ic_refcnt) == 0
+		     && iclog->ic_offset == 0)) {
+			iclog = iclog->ic_prev;
+			if (iclog->ic_state == XLOG_STATE_ACTIVE ||
+			    iclog->ic_state == XLOG_STATE_DIRTY)
+				goto no_sleep;
+			else
+				goto maybe_sleep;
+		} else {
+			if (atomic_read(&iclog->ic_refcnt) == 0) {
+				/* We are the only one with access to this
+				 * iclog.  Flush it out now.  There should
+				 * be a roundoff of zero to show that someone
+				 * has already taken care of the roundoff from
+				 * the previous sync.
+				 */
+				atomic_inc(&iclog->ic_refcnt);
+				lsn = be64_to_cpu(iclog->ic_header.h_lsn);
+				xlog_state_switch_iclogs(log, iclog, 0);
+				spin_unlock(&log->l_icloglock);
+
+				if (xlog_state_release_iclog(log, iclog))
+					return XFS_ERROR(EIO);
+
+				if (log_flushed)
+					*log_flushed = 1;
+				spin_lock(&log->l_icloglock);
+				if (be64_to_cpu(iclog->ic_header.h_lsn) == lsn &&
+				    iclog->ic_state != XLOG_STATE_DIRTY)
+					goto maybe_sleep;
+				else
+					goto no_sleep;
+			} else {
+				/* Someone else is writing to this iclog.
+				 * Use its call to flush out the data.  However,
+				 * the other thread may not force out this LR,
+				 * so we mark it WANT_SYNC.
+				 */
+				xlog_state_switch_iclogs(log, iclog, 0);
+				goto maybe_sleep;
+			}
+		}
+	}
+
+	/* By the time we come around again, the iclog could've been filled
+	 * which would give it another lsn.  If we have a new lsn, just
+	 * return because the relevant data has been flushed.
+	 */
+maybe_sleep:
+	if (flags & XFS_LOG_SYNC) {
+		/*
+		 * We must check if we're shutting down here, before
+		 * we wait, while we're holding the l_icloglock.
+		 * Then we check again after waking up, in case our
+		 * sleep was disturbed by a bad news.
+		 */
+		if (iclog->ic_state & XLOG_STATE_IOERROR) {
+			spin_unlock(&log->l_icloglock);
+			return XFS_ERROR(EIO);
+		}
+		XFS_STATS_INC(xs_log_force_sleep);
+		xlog_wait(&iclog->ic_force_wait, &log->l_icloglock);
+		/*
+		 * No need to grab the log lock here since we're
+		 * only deciding whether or not to return EIO
+		 * and the memory read should be atomic.
+		 */
+		if (iclog->ic_state & XLOG_STATE_IOERROR)
+			return XFS_ERROR(EIO);
+		if (log_flushed)
+			*log_flushed = 1;
+	} else {
+
+no_sleep:
+		spin_unlock(&log->l_icloglock);
+	}
+	return 0;
+}
+
+/*
+ * Wrapper for _xfs_log_force(), to be used when caller doesn't care
+ * about errors or whether the log was flushed or not. This is the normal
+ * interface to use when trying to unpin items or move the log forward.
+ */
+void
+xfs_log_force(
+	xfs_mount_t	*mp,
+	uint		flags)
+{
+	int	error;
+
+	error = _xfs_log_force(mp, flags, NULL);
+	if (error)
+		xfs_warn(mp, "%s: error %d returned.", __func__, error);
+}
+
+/*
+ * Force the in-core log to disk for a specific LSN.
+ *
+ * Find in-core log with lsn.
+ *	If it is in the DIRTY state, just return.
+ *	If it is in the ACTIVE state, move the in-core log into the WANT_SYNC
+ *		state and go to sleep or return.
+ *	If it is in any other state, go to sleep or return.
+ *
+ * Synchronous forces are implemented with a signal variable. All callers
+ * to force a given lsn to disk will wait on a the sv attached to the
+ * specific in-core log.  When given in-core log finally completes its
+ * write to disk, that thread will wake up all threads waiting on the
+ * sv.
+ */
+int
+_xfs_log_force_lsn(
+	struct xfs_mount	*mp,
+	xfs_lsn_t		lsn,
+	uint			flags,
+	int			*log_flushed)
+{
+	struct log		*log = mp->m_log;
+	struct xlog_in_core	*iclog;
+	int			already_slept = 0;
+
+	ASSERT(lsn != 0);
+
+	XFS_STATS_INC(xs_log_force);
+
+	if (log->l_cilp) {
+		lsn = xlog_cil_force_lsn(log, lsn);
+		if (lsn == NULLCOMMITLSN)
+			return 0;
+	}
+
+try_again:
+	spin_lock(&log->l_icloglock);
+	iclog = log->l_iclog;
+	if (iclog->ic_state & XLOG_STATE_IOERROR) {
+		spin_unlock(&log->l_icloglock);
+		return XFS_ERROR(EIO);
+	}
+
+	do {
+		if (be64_to_cpu(iclog->ic_header.h_lsn) != lsn) {
+			iclog = iclog->ic_next;
+			continue;
+		}
+
+		if (iclog->ic_state == XLOG_STATE_DIRTY) {
+			spin_unlock(&log->l_icloglock);
+			return 0;
+		}
+
+		if (iclog->ic_state == XLOG_STATE_ACTIVE) {
+			/*
+			 * We sleep here if we haven't already slept (e.g.
+			 * this is the first time we've looked at the correct
+			 * iclog buf) and the buffer before us is going to
+			 * be sync'ed. The reason for this is that if we
+			 * are doing sync transactions here, by waiting for
+			 * the previous I/O to complete, we can allow a few
+			 * more transactions into this iclog before we close
+			 * it down.
+			 *
+			 * Otherwise, we mark the buffer WANT_SYNC, and bump
+			 * up the refcnt so we can release the log (which
+			 * drops the ref count).  The state switch keeps new
+			 * transaction commits from using this buffer.  When
+			 * the current commits finish writing into the buffer,
+			 * the refcount will drop to zero and the buffer will
+			 * go out then.
+			 */
+			if (!already_slept &&
+			    (iclog->ic_prev->ic_state &
+			     (XLOG_STATE_WANT_SYNC | XLOG_STATE_SYNCING))) {
+				ASSERT(!(iclog->ic_state & XLOG_STATE_IOERROR));
+
+				XFS_STATS_INC(xs_log_force_sleep);
+
+				xlog_wait(&iclog->ic_prev->ic_write_wait,
+							&log->l_icloglock);
+				if (log_flushed)
+					*log_flushed = 1;
+				already_slept = 1;
+				goto try_again;
+			}
+			atomic_inc(&iclog->ic_refcnt);
+			xlog_state_switch_iclogs(log, iclog, 0);
+			spin_unlock(&log->l_icloglock);
+			if (xlog_state_release_iclog(log, iclog))
+				return XFS_ERROR(EIO);
+			if (log_flushed)
+				*log_flushed = 1;
+			spin_lock(&log->l_icloglock);
+		}
+
+		if ((flags & XFS_LOG_SYNC) && /* sleep */
+		    !(iclog->ic_state &
+		      (XLOG_STATE_ACTIVE | XLOG_STATE_DIRTY))) {
+			/*
+			 * Don't wait on completion if we know that we've
+			 * gotten a log write error.
+			 */
+			if (iclog->ic_state & XLOG_STATE_IOERROR) {
+				spin_unlock(&log->l_icloglock);
+				return XFS_ERROR(EIO);
+			}
+			XFS_STATS_INC(xs_log_force_sleep);
+			xlog_wait(&iclog->ic_force_wait, &log->l_icloglock);
+			/*
+			 * No need to grab the log lock here since we're
+			 * only deciding whether or not to return EIO
+			 * and the memory read should be atomic.
+			 */
+			if (iclog->ic_state & XLOG_STATE_IOERROR)
+				return XFS_ERROR(EIO);
+
+			if (log_flushed)
+				*log_flushed = 1;
+		} else {		/* just return */
+			spin_unlock(&log->l_icloglock);
+		}
+
+		return 0;
+	} while (iclog != log->l_iclog);
+
+	spin_unlock(&log->l_icloglock);
+	return 0;
+}
+
+/*
+ * Wrapper for _xfs_log_force_lsn(), to be used when caller doesn't care
+ * about errors or whether the log was flushed or not. This is the normal
+ * interface to use when trying to unpin items or move the log forward.
+ */
+void
+xfs_log_force_lsn(
+	xfs_mount_t	*mp,
+	xfs_lsn_t	lsn,
+	uint		flags)
+{
+	int	error;
+
+	error = _xfs_log_force_lsn(mp, lsn, flags, NULL);
+	if (error)
+		xfs_warn(mp, "%s: error %d returned.", __func__, error);
+}
+
+/*
+ * Called when we want to mark the current iclog as being ready to sync to
+ * disk.
+ */
+STATIC void
+xlog_state_want_sync(xlog_t *log, xlog_in_core_t *iclog)
+{
+	assert_spin_locked(&log->l_icloglock);
+
+	if (iclog->ic_state == XLOG_STATE_ACTIVE) {
+		xlog_state_switch_iclogs(log, iclog, 0);
+	} else {
+		ASSERT(iclog->ic_state &
+			(XLOG_STATE_WANT_SYNC|XLOG_STATE_IOERROR));
+	}
+}
+
+
+/*****************************************************************************
+ *
+ *		TICKET functions
+ *
+ *****************************************************************************
+ */
+
+/*
+ * Free a used ticket when its refcount falls to zero.
+ */
+void
+xfs_log_ticket_put(
+	xlog_ticket_t	*ticket)
+{
+	ASSERT(atomic_read(&ticket->t_ref) > 0);
+	if (atomic_dec_and_test(&ticket->t_ref))
+		kmem_zone_free(xfs_log_ticket_zone, ticket);
+}
+
+xlog_ticket_t *
+xfs_log_ticket_get(
+	xlog_ticket_t	*ticket)
+{
+	ASSERT(atomic_read(&ticket->t_ref) > 0);
+	atomic_inc(&ticket->t_ref);
+	return ticket;
+}
+
+/*
+ * Allocate and initialise a new log ticket.
+ */
+xlog_ticket_t *
+xlog_ticket_alloc(
+	struct log	*log,
+	int		unit_bytes,
+	int		cnt,
+	char		client,
+	uint		xflags,
+	int		alloc_flags)
+{
+	struct xlog_ticket *tic;
+	uint		num_headers;
+	int		iclog_space;
+
+	tic = kmem_zone_zalloc(xfs_log_ticket_zone, alloc_flags);
+	if (!tic)
+		return NULL;
+
+	/*
+	 * Permanent reservations have up to 'cnt'-1 active log operations
+	 * in the log.  A unit in this case is the amount of space for one
+	 * of these log operations.  Normal reservations have a cnt of 1
+	 * and their unit amount is the total amount of space required.
+	 *
+	 * The following lines of code account for non-transaction data
+	 * which occupy space in the on-disk log.
+	 *
+	 * Normal form of a transaction is:
+	 * <oph><trans-hdr><start-oph><reg1-oph><reg1><reg2-oph>...<commit-oph>
+	 * and then there are LR hdrs, split-recs and roundoff at end of syncs.
+	 *
+	 * We need to account for all the leadup data and trailer data
+	 * around the transaction data.
+	 * And then we need to account for the worst case in terms of using
+	 * more space.
+	 * The worst case will happen if:
+	 * - the placement of the transaction happens to be such that the
+	 *   roundoff is at its maximum
+	 * - the transaction data is synced before the commit record is synced
+	 *   i.e. <transaction-data><roundoff> | <commit-rec><roundoff>
+	 *   Therefore the commit record is in its own Log Record.
+	 *   This can happen as the commit record is called with its
+	 *   own region to xlog_write().
+	 *   This then means that in the worst case, roundoff can happen for
+	 *   the commit-rec as well.
+	 *   The commit-rec is smaller than padding in this scenario and so it is
+	 *   not added separately.
+	 */
+
+	/* for trans header */
+	unit_bytes += sizeof(xlog_op_header_t);
+	unit_bytes += sizeof(xfs_trans_header_t);
+
+	/* for start-rec */
+	unit_bytes += sizeof(xlog_op_header_t);
+
+	/*
+	 * for LR headers - the space for data in an iclog is the size minus
+	 * the space used for the headers. If we use the iclog size, then we
+	 * undercalculate the number of headers required.
+	 *
+	 * Furthermore - the addition of op headers for split-recs might
+	 * increase the space required enough to require more log and op
+	 * headers, so take that into account too.
+	 *
+	 * IMPORTANT: This reservation makes the assumption that if this
+	 * transaction is the first in an iclog and hence has the LR headers
+	 * accounted to it, then the remaining space in the iclog is
+	 * exclusively for this transaction.  i.e. if the transaction is larger
+	 * than the iclog, it will be the only thing in that iclog.
+	 * Fundamentally, this means we must pass the entire log vector to
+	 * xlog_write to guarantee this.
+	 */
+	iclog_space = log->l_iclog_size - log->l_iclog_hsize;
+	num_headers = howmany(unit_bytes, iclog_space);
+
+	/* for split-recs - ophdrs added when data split over LRs */
+	unit_bytes += sizeof(xlog_op_header_t) * num_headers;
+
+	/* add extra header reservations if we overrun */
+	while (!num_headers ||
+	       howmany(unit_bytes, iclog_space) > num_headers) {
+		unit_bytes += sizeof(xlog_op_header_t);
+		num_headers++;
+	}
+	unit_bytes += log->l_iclog_hsize * num_headers;
+
+	/* for commit-rec LR header - note: padding will subsume the ophdr */
+	unit_bytes += log->l_iclog_hsize;
+
+	/* for roundoff padding for transaction data and one for commit record */
+	if (xfs_sb_version_haslogv2(&log->l_mp->m_sb) &&
+	    log->l_mp->m_sb.sb_logsunit > 1) {
+		/* log su roundoff */
+		unit_bytes += 2*log->l_mp->m_sb.sb_logsunit;
+	} else {
+		/* BB roundoff */
+		unit_bytes += 2*BBSIZE;
+        }
+
+	atomic_set(&tic->t_ref, 1);
+	INIT_LIST_HEAD(&tic->t_queue);
+	tic->t_unit_res		= unit_bytes;
+	tic->t_curr_res		= unit_bytes;
+	tic->t_cnt		= cnt;
+	tic->t_ocnt		= cnt;
+	tic->t_tid		= random32();
+	tic->t_clientid		= client;
+	tic->t_flags		= XLOG_TIC_INITED;
+	tic->t_trans_type	= 0;
+	if (xflags & XFS_LOG_PERM_RESERV)
+		tic->t_flags |= XLOG_TIC_PERM_RESERV;
+	init_waitqueue_head(&tic->t_wait);
+
+	xlog_tic_reset_res(tic);
+
+	return tic;
+}
+
+
+/******************************************************************************
+ *
+ *		Log debug routines
+ *
+ ******************************************************************************
+ */
+#if defined(DEBUG)
+/*
+ * Make sure that the destination ptr is within the valid data region of
+ * one of the iclogs.  This uses backup pointers stored in a different
+ * part of the log in case we trash the log structure.
+ */
+void
+xlog_verify_dest_ptr(
+	struct log	*log,
+	char		*ptr)
+{
+	int i;
+	int good_ptr = 0;
+
+	for (i = 0; i < log->l_iclog_bufs; i++) {
+		if (ptr >= log->l_iclog_bak[i] &&
+		    ptr <= log->l_iclog_bak[i] + log->l_iclog_size)
+			good_ptr++;
+	}
+
+	if (!good_ptr)
+		xfs_emerg(log->l_mp, "%s: invalid ptr", __func__);
+}
+
+/*
+ * Check to make sure the grant write head didn't just over lap the tail.  If
+ * the cycles are the same, we can't be overlapping.  Otherwise, make sure that
+ * the cycles differ by exactly one and check the byte count.
+ *
+ * This check is run unlocked, so can give false positives. Rather than assert
+ * on failures, use a warn-once flag and a panic tag to allow the admin to
+ * determine if they want to panic the machine when such an error occurs. For
+ * debug kernels this will have the same effect as using an assert but, unlinke
+ * an assert, it can be turned off at runtime.
+ */
+STATIC void
+xlog_verify_grant_tail(
+	struct log	*log)
+{
+	int		tail_cycle, tail_blocks;
+	int		cycle, space;
+
+	xlog_crack_grant_head(&log->l_grant_write_head, &cycle, &space);
+	xlog_crack_atomic_lsn(&log->l_tail_lsn, &tail_cycle, &tail_blocks);
+	if (tail_cycle != cycle) {
+		if (cycle - 1 != tail_cycle &&
+		    !(log->l_flags & XLOG_TAIL_WARN)) {
+			xfs_alert_tag(log->l_mp, XFS_PTAG_LOGRES,
+				"%s: cycle - 1 != tail_cycle", __func__);
+			log->l_flags |= XLOG_TAIL_WARN;
+		}
+
+		if (space > BBTOB(tail_blocks) &&
+		    !(log->l_flags & XLOG_TAIL_WARN)) {
+			xfs_alert_tag(log->l_mp, XFS_PTAG_LOGRES,
+				"%s: space > BBTOB(tail_blocks)", __func__);
+			log->l_flags |= XLOG_TAIL_WARN;
+		}
+	}
+}
+
+/* check if it will fit */
+STATIC void
+xlog_verify_tail_lsn(xlog_t	    *log,
+		     xlog_in_core_t *iclog,
+		     xfs_lsn_t	    tail_lsn)
+{
+    int blocks;
+
+    if (CYCLE_LSN(tail_lsn) == log->l_prev_cycle) {
+	blocks =
+	    log->l_logBBsize - (log->l_prev_block - BLOCK_LSN(tail_lsn));
+	if (blocks < BTOBB(iclog->ic_offset)+BTOBB(log->l_iclog_hsize))
+		xfs_emerg(log->l_mp, "%s: ran out of log space", __func__);
+    } else {
+	ASSERT(CYCLE_LSN(tail_lsn)+1 == log->l_prev_cycle);
+
+	if (BLOCK_LSN(tail_lsn) == log->l_prev_block)
+		xfs_emerg(log->l_mp, "%s: tail wrapped", __func__);
+
+	blocks = BLOCK_LSN(tail_lsn) - log->l_prev_block;
+	if (blocks < BTOBB(iclog->ic_offset) + 1)
+		xfs_emerg(log->l_mp, "%s: ran out of log space", __func__);
+    }
+}	/* xlog_verify_tail_lsn */
+
+/*
+ * Perform a number of checks on the iclog before writing to disk.
+ *
+ * 1. Make sure the iclogs are still circular
+ * 2. Make sure we have a good magic number
+ * 3. Make sure we don't have magic numbers in the data
+ * 4. Check fields of each log operation header for:
+ *	A. Valid client identifier
+ *	B. tid ptr value falls in valid ptr space (user space code)
+ *	C. Length in log record header is correct according to the
+ *		individual operation headers within record.
+ * 5. When a bwrite will occur within 5 blocks of the front of the physical
+ *	log, check the preceding blocks of the physical log to make sure all
+ *	the cycle numbers agree with the current cycle number.
+ */
+STATIC void
+xlog_verify_iclog(xlog_t	 *log,
+		  xlog_in_core_t *iclog,
+		  int		 count,
+		  boolean_t	 syncing)
+{
+	xlog_op_header_t	*ophead;
+	xlog_in_core_t		*icptr;
+	xlog_in_core_2_t	*xhdr;
+	xfs_caddr_t		ptr;
+	xfs_caddr_t		base_ptr;
+	__psint_t		field_offset;
+	__uint8_t		clientid;
+	int			len, i, j, k, op_len;
+	int			idx;
+
+	/* check validity of iclog pointers */
+	spin_lock(&log->l_icloglock);
+	icptr = log->l_iclog;
+	for (i=0; i < log->l_iclog_bufs; i++) {
+		if (icptr == NULL)
+			xfs_emerg(log->l_mp, "%s: invalid ptr", __func__);
+		icptr = icptr->ic_next;
+	}
+	if (icptr != log->l_iclog)
+		xfs_emerg(log->l_mp, "%s: corrupt iclog ring", __func__);
+	spin_unlock(&log->l_icloglock);
+
+	/* check log magic numbers */
+	if (be32_to_cpu(iclog->ic_header.h_magicno) != XLOG_HEADER_MAGIC_NUM)
+		xfs_emerg(log->l_mp, "%s: invalid magic num", __func__);
+
+	ptr = (xfs_caddr_t) &iclog->ic_header;
+	for (ptr += BBSIZE; ptr < ((xfs_caddr_t)&iclog->ic_header) + count;
+	     ptr += BBSIZE) {
+		if (be32_to_cpu(*(__be32 *)ptr) == XLOG_HEADER_MAGIC_NUM)
+			xfs_emerg(log->l_mp, "%s: unexpected magic num",
+				__func__);
+	}
+
+	/* check fields */
+	len = be32_to_cpu(iclog->ic_header.h_num_logops);
+	ptr = iclog->ic_datap;
+	base_ptr = ptr;
+	ophead = (xlog_op_header_t *)ptr;
+	xhdr = iclog->ic_data;
+	for (i = 0; i < len; i++) {
+		ophead = (xlog_op_header_t *)ptr;
+
+		/* clientid is only 1 byte */
+		field_offset = (__psint_t)
+			       ((xfs_caddr_t)&(ophead->oh_clientid) - base_ptr);
+		if (syncing == B_FALSE || (field_offset & 0x1ff)) {
+			clientid = ophead->oh_clientid;
+		} else {
+			idx = BTOBBT((xfs_caddr_t)&(ophead->oh_clientid) - iclog->ic_datap);
+			if (idx >= (XLOG_HEADER_CYCLE_SIZE / BBSIZE)) {
+				j = idx / (XLOG_HEADER_CYCLE_SIZE / BBSIZE);
+				k = idx % (XLOG_HEADER_CYCLE_SIZE / BBSIZE);
+				clientid = xlog_get_client_id(
+					xhdr[j].hic_xheader.xh_cycle_data[k]);
+			} else {
+				clientid = xlog_get_client_id(
+					iclog->ic_header.h_cycle_data[idx]);
+			}
+		}
+		if (clientid != XFS_TRANSACTION && clientid != XFS_LOG)
+			xfs_warn(log->l_mp,
+				"%s: invalid clientid %d op 0x%p offset 0x%lx",
+				__func__, clientid, ophead,
+				(unsigned long)field_offset);
+
+		/* check length */
+		field_offset = (__psint_t)
+			       ((xfs_caddr_t)&(ophead->oh_len) - base_ptr);
+		if (syncing == B_FALSE || (field_offset & 0x1ff)) {
+			op_len = be32_to_cpu(ophead->oh_len);
+		} else {
+			idx = BTOBBT((__psint_t)&ophead->oh_len -
+				    (__psint_t)iclog->ic_datap);
+			if (idx >= (XLOG_HEADER_CYCLE_SIZE / BBSIZE)) {
+				j = idx / (XLOG_HEADER_CYCLE_SIZE / BBSIZE);
+				k = idx % (XLOG_HEADER_CYCLE_SIZE / BBSIZE);
+				op_len = be32_to_cpu(xhdr[j].hic_xheader.xh_cycle_data[k]);
+			} else {
+				op_len = be32_to_cpu(iclog->ic_header.h_cycle_data[idx]);
+			}
+		}
+		ptr += sizeof(xlog_op_header_t) + op_len;
+	}
+}	/* xlog_verify_iclog */
+#endif
+
+/*
+ * Mark all iclogs IOERROR. l_icloglock is held by the caller.
+ */
+STATIC int
+xlog_state_ioerror(
+	xlog_t	*log)
+{
+	xlog_in_core_t	*iclog, *ic;
+
+	iclog = log->l_iclog;
+	if (! (iclog->ic_state & XLOG_STATE_IOERROR)) {
+		/*
+		 * Mark all the incore logs IOERROR.
+		 * From now on, no log flushes will result.
+		 */
+		ic = iclog;
+		do {
+			ic->ic_state = XLOG_STATE_IOERROR;
+			ic = ic->ic_next;
+		} while (ic != iclog);
+		return 0;
+	}
+	/*
+	 * Return non-zero, if state transition has already happened.
+	 */
+	return 1;
+}
+
+/*
+ * This is called from xfs_force_shutdown, when we're forcibly
+ * shutting down the filesystem, typically because of an IO error.
+ * Our main objectives here are to make sure that:
+ *	a. the filesystem gets marked 'SHUTDOWN' for all interested
+ *	   parties to find out, 'atomically'.
+ *	b. those who're sleeping on log reservations, pinned objects and
+ *	    other resources get woken up, and be told the bad news.
+ *	c. nothing new gets queued up after (a) and (b) are done.
+ *	d. if !logerror, flush the iclogs to disk, then seal them off
+ *	   for business.
+ *
+ * Note: for delayed logging the !logerror case needs to flush the regions
+ * held in memory out to the iclogs before flushing them to disk. This needs
+ * to be done before the log is marked as shutdown, otherwise the flush to the
+ * iclogs will fail.
+ */
+int
+xfs_log_force_umount(
+	struct xfs_mount	*mp,
+	int			logerror)
+{
+	xlog_ticket_t	*tic;
+	xlog_t		*log;
+	int		retval;
+
+	log = mp->m_log;
+
+	/*
+	 * If this happens during log recovery, don't worry about
+	 * locking; the log isn't open for business yet.
+	 */
+	if (!log ||
+	    log->l_flags & XLOG_ACTIVE_RECOVERY) {
+		mp->m_flags |= XFS_MOUNT_FS_SHUTDOWN;
+		if (mp->m_sb_bp)
+			XFS_BUF_DONE(mp->m_sb_bp);
+		return 0;
+	}
+
+	/*
+	 * Somebody could've already done the hard work for us.
+	 * No need to get locks for this.
+	 */
+	if (logerror && log->l_iclog->ic_state & XLOG_STATE_IOERROR) {
+		ASSERT(XLOG_FORCED_SHUTDOWN(log));
+		return 1;
+	}
+	retval = 0;
+
+	/*
+	 * Flush the in memory commit item list before marking the log as
+	 * being shut down. We need to do it in this order to ensure all the
+	 * completed transactions are flushed to disk with the xfs_log_force()
+	 * call below.
+	 */
+	if (!logerror && (mp->m_flags & XFS_MOUNT_DELAYLOG))
+		xlog_cil_force(log);
+
+	/*
+	 * mark the filesystem and the as in a shutdown state and wake
+	 * everybody up to tell them the bad news.
+	 */
+	spin_lock(&log->l_icloglock);
+	mp->m_flags |= XFS_MOUNT_FS_SHUTDOWN;
+	if (mp->m_sb_bp)
+		XFS_BUF_DONE(mp->m_sb_bp);
+
+	/*
+	 * This flag is sort of redundant because of the mount flag, but
+	 * it's good to maintain the separation between the log and the rest
+	 * of XFS.
+	 */
+	log->l_flags |= XLOG_IO_ERROR;
+
+	/*
+	 * If we hit a log error, we want to mark all the iclogs IOERROR
+	 * while we're still holding the loglock.
+	 */
+	if (logerror)
+		retval = xlog_state_ioerror(log);
+	spin_unlock(&log->l_icloglock);
+
+	/*
+	 * We don't want anybody waiting for log reservations after this. That
+	 * means we have to wake up everybody queued up on reserveq as well as
+	 * writeq.  In addition, we make sure in xlog_{re}grant_log_space that
+	 * we don't enqueue anything once the SHUTDOWN flag is set, and this
+	 * action is protected by the grant locks.
+	 */
+	spin_lock(&log->l_grant_reserve_lock);
+	list_for_each_entry(tic, &log->l_reserveq, t_queue)
+		wake_up(&tic->t_wait);
+	spin_unlock(&log->l_grant_reserve_lock);
+
+	spin_lock(&log->l_grant_write_lock);
+	list_for_each_entry(tic, &log->l_writeq, t_queue)
+		wake_up(&tic->t_wait);
+	spin_unlock(&log->l_grant_write_lock);
+
+	if (!(log->l_iclog->ic_state & XLOG_STATE_IOERROR)) {
+		ASSERT(!logerror);
+		/*
+		 * Force the incore logs to disk before shutting the
+		 * log down completely.
+		 */
+		_xfs_log_force(mp, XFS_LOG_SYNC, NULL);
+
+		spin_lock(&log->l_icloglock);
+		retval = xlog_state_ioerror(log);
+		spin_unlock(&log->l_icloglock);
+	}
+	/*
+	 * Wake up everybody waiting on xfs_log_force.
+	 * Callback all log item committed functions as if the
+	 * log writes were completed.
+	 */
+	xlog_state_do_callback(log, XFS_LI_ABORTED, NULL);
+
+#ifdef XFSERRORDEBUG
+	{
+		xlog_in_core_t	*iclog;
+
+		spin_lock(&log->l_icloglock);
+		iclog = log->l_iclog;
+		do {
+			ASSERT(iclog->ic_callback == 0);
+			iclog = iclog->ic_next;
+		} while (iclog != log->l_iclog);
+		spin_unlock(&log->l_icloglock);
+	}
+#endif
+	/* return non-zero if log IOERROR transition had already happened */
+	return retval;
+}
+
+STATIC int
+xlog_iclogs_empty(xlog_t *log)
+{
+	xlog_in_core_t	*iclog;
+
+	iclog = log->l_iclog;
+	do {
+		/* endianness does not matter here, zero is zero in
+		 * any language.
+		 */
+		if (iclog->ic_header.h_num_logops)
+			return 0;
+		iclog = iclog->ic_next;
+	} while (iclog != log->l_iclog);
+	return 1;
+}
diff --git a/fs/xfs/xfs_log.h b/fs/xfs/xfs_log.h
new file mode 100644
index 00000000..78c90399
--- /dev/null
+++ b/fs/xfs/xfs_log.h
@@ -0,0 +1,198 @@
+/*
+ * Copyright (c) 2000-2003,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef	__XFS_LOG_H__
+#define __XFS_LOG_H__
+
+/* get lsn fields */
+#define CYCLE_LSN(lsn) ((uint)((lsn)>>32))
+#define BLOCK_LSN(lsn) ((uint)(lsn))
+
+/* this is used in a spot where we might otherwise double-endian-flip */
+#define CYCLE_LSN_DISK(lsn) (((__be32 *)&(lsn))[0])
+
+#ifdef __KERNEL__
+/*
+ * By comparing each component, we don't have to worry about extra
+ * endian issues in treating two 32 bit numbers as one 64 bit number
+ */
+static inline xfs_lsn_t	_lsn_cmp(xfs_lsn_t lsn1, xfs_lsn_t lsn2)
+{
+	if (CYCLE_LSN(lsn1) != CYCLE_LSN(lsn2))
+		return (CYCLE_LSN(lsn1)<CYCLE_LSN(lsn2))? -999 : 999;
+
+	if (BLOCK_LSN(lsn1) != BLOCK_LSN(lsn2))
+		return (BLOCK_LSN(lsn1)<BLOCK_LSN(lsn2))? -999 : 999;
+
+	return 0;
+}
+
+#define	XFS_LSN_CMP(x,y) _lsn_cmp(x,y)
+
+/*
+ * Macros, structures, prototypes for interface to the log manager.
+ */
+
+/*
+ * Flags to xfs_log_done()
+ */
+#define XFS_LOG_REL_PERM_RESERV	0x1
+
+/*
+ * Flags to xfs_log_reserve()
+ *
+ *	XFS_LOG_PERM_RESERV: Permanent reservation.  When writes are
+ *		performed against this type of reservation, the reservation
+ *		is not decreased.  Long running transactions should use this.
+ */
+#define XFS_LOG_PERM_RESERV	0x2
+
+/*
+ * Flags to xfs_log_force()
+ *
+ *	XFS_LOG_SYNC:	Synchronous force in-core log to disk
+ */
+#define XFS_LOG_SYNC		0x1
+
+#endif	/* __KERNEL__ */
+
+
+/* Log Clients */
+#define XFS_TRANSACTION		0x69
+#define XFS_VOLUME		0x2
+#define XFS_LOG			0xaa
+
+
+/* Region types for iovec's i_type */
+#define XLOG_REG_TYPE_BFORMAT		1
+#define XLOG_REG_TYPE_BCHUNK		2
+#define XLOG_REG_TYPE_EFI_FORMAT	3
+#define XLOG_REG_TYPE_EFD_FORMAT	4
+#define XLOG_REG_TYPE_IFORMAT		5
+#define XLOG_REG_TYPE_ICORE		6
+#define XLOG_REG_TYPE_IEXT		7
+#define XLOG_REG_TYPE_IBROOT		8
+#define XLOG_REG_TYPE_ILOCAL		9
+#define XLOG_REG_TYPE_IATTR_EXT		10
+#define XLOG_REG_TYPE_IATTR_BROOT	11
+#define XLOG_REG_TYPE_IATTR_LOCAL	12
+#define XLOG_REG_TYPE_QFORMAT		13
+#define XLOG_REG_TYPE_DQUOT		14
+#define XLOG_REG_TYPE_QUOTAOFF		15
+#define XLOG_REG_TYPE_LRHEADER		16
+#define XLOG_REG_TYPE_UNMOUNT		17
+#define XLOG_REG_TYPE_COMMIT		18
+#define XLOG_REG_TYPE_TRANSHDR		19
+#define XLOG_REG_TYPE_MAX		19
+
+typedef struct xfs_log_iovec {
+	void		*i_addr;	/* beginning address of region */
+	int		i_len;		/* length in bytes of region */
+	uint		i_type;		/* type of region */
+} xfs_log_iovec_t;
+
+struct xfs_log_vec {
+	struct xfs_log_vec	*lv_next;	/* next lv in build list */
+	int			lv_niovecs;	/* number of iovecs in lv */
+	struct xfs_log_iovec	*lv_iovecp;	/* iovec array */
+	struct xfs_log_item	*lv_item;	/* owner */
+	char			*lv_buf;	/* formatted buffer */
+	int			lv_buf_len;	/* size of formatted buffer */
+};
+
+/*
+ * Structure used to pass callback function and the function's argument
+ * to the log manager.
+ */
+typedef struct xfs_log_callback {
+	struct xfs_log_callback	*cb_next;
+	void			(*cb_func)(void *, int);
+	void			*cb_arg;
+} xfs_log_callback_t;
+
+
+#ifdef __KERNEL__
+/* Log manager interfaces */
+struct xfs_mount;
+struct xlog_in_core;
+struct xlog_ticket;
+struct xfs_log_item;
+struct xfs_item_ops;
+struct xfs_trans;
+
+void	xfs_log_item_init(struct xfs_mount	*mp,
+			struct xfs_log_item	*item,
+			int			type,
+			struct xfs_item_ops	*ops);
+
+xfs_lsn_t xfs_log_done(struct xfs_mount *mp,
+		       struct xlog_ticket *ticket,
+		       struct xlog_in_core **iclog,
+		       uint		flags);
+int	  _xfs_log_force(struct xfs_mount *mp,
+			 uint		flags,
+			 int		*log_forced);
+void	  xfs_log_force(struct xfs_mount	*mp,
+			uint			flags);
+int	  _xfs_log_force_lsn(struct xfs_mount *mp,
+			     xfs_lsn_t		lsn,
+			     uint		flags,
+			     int		*log_forced);
+void	  xfs_log_force_lsn(struct xfs_mount	*mp,
+			    xfs_lsn_t		lsn,
+			    uint		flags);
+int	  xfs_log_mount(struct xfs_mount	*mp,
+			struct xfs_buftarg	*log_target,
+			xfs_daddr_t		start_block,
+			int		 	num_bblocks);
+int	  xfs_log_mount_finish(struct xfs_mount *mp);
+void	  xfs_log_move_tail(struct xfs_mount	*mp,
+			    xfs_lsn_t		tail_lsn);
+int	  xfs_log_notify(struct xfs_mount	*mp,
+			 struct xlog_in_core	*iclog,
+			 xfs_log_callback_t	*callback_entry);
+int	  xfs_log_release_iclog(struct xfs_mount *mp,
+			 struct xlog_in_core	 *iclog);
+int	  xfs_log_reserve(struct xfs_mount *mp,
+			  int		   length,
+			  int		   count,
+			  struct xlog_ticket **ticket,
+			  __uint8_t	   clientid,
+			  uint		   flags,
+			  uint		   t_type);
+int	  xfs_log_write(struct xfs_mount *mp,
+			xfs_log_iovec_t  region[],
+			int		 nentries,
+			struct xlog_ticket *ticket,
+			xfs_lsn_t	 *start_lsn);
+int	  xfs_log_unmount_write(struct xfs_mount *mp);
+void      xfs_log_unmount(struct xfs_mount *mp);
+int	  xfs_log_force_umount(struct xfs_mount *mp, int logerror);
+int	  xfs_log_need_covered(struct xfs_mount *mp);
+
+void	  xlog_iodone(struct xfs_buf *);
+
+struct xlog_ticket *xfs_log_ticket_get(struct xlog_ticket *ticket);
+void	  xfs_log_ticket_put(struct xlog_ticket *ticket);
+
+void	xfs_log_commit_cil(struct xfs_mount *mp, struct xfs_trans *tp,
+				struct xfs_log_vec *log_vector,
+				xfs_lsn_t *commit_lsn, int flags);
+bool	xfs_log_item_in_current_chkpt(struct xfs_log_item *lip);
+
+#endif
+#endif	/* __XFS_LOG_H__ */
diff --git a/fs/xfs/xfs_log_cil.c b/fs/xfs/xfs_log_cil.c
new file mode 100644
index 00000000..c7755d5a
--- /dev/null
+++ b/fs/xfs/xfs_log_cil.c
@@ -0,0 +1,804 @@
+/*
+ * Copyright (c) 2010 Red Hat, Inc. All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_trans_priv.h"
+#include "xfs_log_priv.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_mount.h"
+#include "xfs_error.h"
+#include "xfs_alloc.h"
+#include "xfs_discard.h"
+
+/*
+ * Perform initial CIL structure initialisation. If the CIL is not
+ * enabled in this filesystem, ensure the log->l_cilp is null so
+ * we can check this conditional to determine if we are doing delayed
+ * logging or not.
+ */
+int
+xlog_cil_init(
+	struct log	*log)
+{
+	struct xfs_cil	*cil;
+	struct xfs_cil_ctx *ctx;
+
+	log->l_cilp = NULL;
+	if (!(log->l_mp->m_flags & XFS_MOUNT_DELAYLOG))
+		return 0;
+
+	cil = kmem_zalloc(sizeof(*cil), KM_SLEEP|KM_MAYFAIL);
+	if (!cil)
+		return ENOMEM;
+
+	ctx = kmem_zalloc(sizeof(*ctx), KM_SLEEP|KM_MAYFAIL);
+	if (!ctx) {
+		kmem_free(cil);
+		return ENOMEM;
+	}
+
+	INIT_LIST_HEAD(&cil->xc_cil);
+	INIT_LIST_HEAD(&cil->xc_committing);
+	spin_lock_init(&cil->xc_cil_lock);
+	init_rwsem(&cil->xc_ctx_lock);
+	init_waitqueue_head(&cil->xc_commit_wait);
+
+	INIT_LIST_HEAD(&ctx->committing);
+	INIT_LIST_HEAD(&ctx->busy_extents);
+	ctx->sequence = 1;
+	ctx->cil = cil;
+	cil->xc_ctx = ctx;
+	cil->xc_current_sequence = ctx->sequence;
+
+	cil->xc_log = log;
+	log->l_cilp = cil;
+	return 0;
+}
+
+void
+xlog_cil_destroy(
+	struct log	*log)
+{
+	if (!log->l_cilp)
+		return;
+
+	if (log->l_cilp->xc_ctx) {
+		if (log->l_cilp->xc_ctx->ticket)
+			xfs_log_ticket_put(log->l_cilp->xc_ctx->ticket);
+		kmem_free(log->l_cilp->xc_ctx);
+	}
+
+	ASSERT(list_empty(&log->l_cilp->xc_cil));
+	kmem_free(log->l_cilp);
+}
+
+/*
+ * Allocate a new ticket. Failing to get a new ticket makes it really hard to
+ * recover, so we don't allow failure here. Also, we allocate in a context that
+ * we don't want to be issuing transactions from, so we need to tell the
+ * allocation code this as well.
+ *
+ * We don't reserve any space for the ticket - we are going to steal whatever
+ * space we require from transactions as they commit. To ensure we reserve all
+ * the space required, we need to set the current reservation of the ticket to
+ * zero so that we know to steal the initial transaction overhead from the
+ * first transaction commit.
+ */
+static struct xlog_ticket *
+xlog_cil_ticket_alloc(
+	struct log	*log)
+{
+	struct xlog_ticket *tic;
+
+	tic = xlog_ticket_alloc(log, 0, 1, XFS_TRANSACTION, 0,
+				KM_SLEEP|KM_NOFS);
+	tic->t_trans_type = XFS_TRANS_CHECKPOINT;
+
+	/*
+	 * set the current reservation to zero so we know to steal the basic
+	 * transaction overhead reservation from the first transaction commit.
+	 */
+	tic->t_curr_res = 0;
+	return tic;
+}
+
+/*
+ * After the first stage of log recovery is done, we know where the head and
+ * tail of the log are. We need this log initialisation done before we can
+ * initialise the first CIL checkpoint context.
+ *
+ * Here we allocate a log ticket to track space usage during a CIL push.  This
+ * ticket is passed to xlog_write() directly so that we don't slowly leak log
+ * space by failing to account for space used by log headers and additional
+ * region headers for split regions.
+ */
+void
+xlog_cil_init_post_recovery(
+	struct log	*log)
+{
+	if (!log->l_cilp)
+		return;
+
+	log->l_cilp->xc_ctx->ticket = xlog_cil_ticket_alloc(log);
+	log->l_cilp->xc_ctx->sequence = 1;
+	log->l_cilp->xc_ctx->commit_lsn = xlog_assign_lsn(log->l_curr_cycle,
+								log->l_curr_block);
+}
+
+/*
+ * Format log item into a flat buffers
+ *
+ * For delayed logging, we need to hold a formatted buffer containing all the
+ * changes on the log item. This enables us to relog the item in memory and
+ * write it out asynchronously without needing to relock the object that was
+ * modified at the time it gets written into the iclog.
+ *
+ * This function builds a vector for the changes in each log item in the
+ * transaction. It then works out the length of the buffer needed for each log
+ * item, allocates them and formats the vector for the item into the buffer.
+ * The buffer is then attached to the log item are then inserted into the
+ * Committed Item List for tracking until the next checkpoint is written out.
+ *
+ * We don't set up region headers during this process; we simply copy the
+ * regions into the flat buffer. We can do this because we still have to do a
+ * formatting step to write the regions into the iclog buffer.  Writing the
+ * ophdrs during the iclog write means that we can support splitting large
+ * regions across iclog boundares without needing a change in the format of the
+ * item/region encapsulation.
+ *
+ * Hence what we need to do now is change the rewrite the vector array to point
+ * to the copied region inside the buffer we just allocated. This allows us to
+ * format the regions into the iclog as though they are being formatted
+ * directly out of the objects themselves.
+ */
+static void
+xlog_cil_format_items(
+	struct log		*log,
+	struct xfs_log_vec	*log_vector)
+{
+	struct xfs_log_vec *lv;
+
+	ASSERT(log_vector);
+	for (lv = log_vector; lv; lv = lv->lv_next) {
+		void	*ptr;
+		int	index;
+		int	len = 0;
+
+		/* build the vector array and calculate it's length */
+		IOP_FORMAT(lv->lv_item, lv->lv_iovecp);
+		for (index = 0; index < lv->lv_niovecs; index++)
+			len += lv->lv_iovecp[index].i_len;
+
+		lv->lv_buf_len = len;
+		lv->lv_buf = kmem_alloc(lv->lv_buf_len, KM_SLEEP|KM_NOFS);
+		ptr = lv->lv_buf;
+
+		for (index = 0; index < lv->lv_niovecs; index++) {
+			struct xfs_log_iovec *vec = &lv->lv_iovecp[index];
+
+			memcpy(ptr, vec->i_addr, vec->i_len);
+			vec->i_addr = ptr;
+			ptr += vec->i_len;
+		}
+		ASSERT(ptr == lv->lv_buf + lv->lv_buf_len);
+	}
+}
+
+/*
+ * Prepare the log item for insertion into the CIL. Calculate the difference in
+ * log space and vectors it will consume, and if it is a new item pin it as
+ * well.
+ */
+STATIC void
+xfs_cil_prepare_item(
+	struct log		*log,
+	struct xfs_log_vec	*lv,
+	int			*len,
+	int			*diff_iovecs)
+{
+	struct xfs_log_vec	*old = lv->lv_item->li_lv;
+
+	if (old) {
+		/* existing lv on log item, space used is a delta */
+		ASSERT(!list_empty(&lv->lv_item->li_cil));
+		ASSERT(old->lv_buf && old->lv_buf_len && old->lv_niovecs);
+
+		*len += lv->lv_buf_len - old->lv_buf_len;
+		*diff_iovecs += lv->lv_niovecs - old->lv_niovecs;
+		kmem_free(old->lv_buf);
+		kmem_free(old);
+	} else {
+		/* new lv, must pin the log item */
+		ASSERT(!lv->lv_item->li_lv);
+		ASSERT(list_empty(&lv->lv_item->li_cil));
+
+		*len += lv->lv_buf_len;
+		*diff_iovecs += lv->lv_niovecs;
+		IOP_PIN(lv->lv_item);
+
+	}
+
+	/* attach new log vector to log item */
+	lv->lv_item->li_lv = lv;
+
+	/*
+	 * If this is the first time the item is being committed to the
+	 * CIL, store the sequence number on the log item so we can
+	 * tell in future commits whether this is the first checkpoint
+	 * the item is being committed into.
+	 */
+	if (!lv->lv_item->li_seq)
+		lv->lv_item->li_seq = log->l_cilp->xc_ctx->sequence;
+}
+
+/*
+ * Insert the log items into the CIL and calculate the difference in space
+ * consumed by the item. Add the space to the checkpoint ticket and calculate
+ * if the change requires additional log metadata. If it does, take that space
+ * as well. Remove the amount of space we addded to the checkpoint ticket from
+ * the current transaction ticket so that the accounting works out correctly.
+ */
+static void
+xlog_cil_insert_items(
+	struct log		*log,
+	struct xfs_log_vec	*log_vector,
+	struct xlog_ticket	*ticket)
+{
+	struct xfs_cil		*cil = log->l_cilp;
+	struct xfs_cil_ctx	*ctx = cil->xc_ctx;
+	struct xfs_log_vec	*lv;
+	int			len = 0;
+	int			diff_iovecs = 0;
+	int			iclog_space;
+
+	ASSERT(log_vector);
+
+	/*
+	 * Do all the accounting aggregation and switching of log vectors
+	 * around in a separate loop to the insertion of items into the CIL.
+	 * Then we can do a separate loop to update the CIL within a single
+	 * lock/unlock pair. This reduces the number of round trips on the CIL
+	 * lock from O(nr_logvectors) to O(1) and greatly reduces the overall
+	 * hold time for the transaction commit.
+	 *
+	 * If this is the first time the item is being placed into the CIL in
+	 * this context, pin it so it can't be written to disk until the CIL is
+	 * flushed to the iclog and the iclog written to disk.
+	 *
+	 * We can do this safely because the context can't checkpoint until we
+	 * are done so it doesn't matter exactly how we update the CIL.
+	 */
+	for (lv = log_vector; lv; lv = lv->lv_next)
+		xfs_cil_prepare_item(log, lv, &len, &diff_iovecs);
+
+	/* account for space used by new iovec headers  */
+	len += diff_iovecs * sizeof(xlog_op_header_t);
+
+	spin_lock(&cil->xc_cil_lock);
+
+	/* move the items to the tail of the CIL */
+	for (lv = log_vector; lv; lv = lv->lv_next)
+		list_move_tail(&lv->lv_item->li_cil, &cil->xc_cil);
+
+	ctx->nvecs += diff_iovecs;
+
+	/*
+	 * Now transfer enough transaction reservation to the context ticket
+	 * for the checkpoint. The context ticket is special - the unit
+	 * reservation has to grow as well as the current reservation as we
+	 * steal from tickets so we can correctly determine the space used
+	 * during the transaction commit.
+	 */
+	if (ctx->ticket->t_curr_res == 0) {
+		/* first commit in checkpoint, steal the header reservation */
+		ASSERT(ticket->t_curr_res >= ctx->ticket->t_unit_res + len);
+		ctx->ticket->t_curr_res = ctx->ticket->t_unit_res;
+		ticket->t_curr_res -= ctx->ticket->t_unit_res;
+	}
+
+	/* do we need space for more log record headers? */
+	iclog_space = log->l_iclog_size - log->l_iclog_hsize;
+	if (len > 0 && (ctx->space_used / iclog_space !=
+				(ctx->space_used + len) / iclog_space)) {
+		int hdrs;
+
+		hdrs = (len + iclog_space - 1) / iclog_space;
+		/* need to take into account split region headers, too */
+		hdrs *= log->l_iclog_hsize + sizeof(struct xlog_op_header);
+		ctx->ticket->t_unit_res += hdrs;
+		ctx->ticket->t_curr_res += hdrs;
+		ticket->t_curr_res -= hdrs;
+		ASSERT(ticket->t_curr_res >= len);
+	}
+	ticket->t_curr_res -= len;
+	ctx->space_used += len;
+
+	spin_unlock(&cil->xc_cil_lock);
+}
+
+static void
+xlog_cil_free_logvec(
+	struct xfs_log_vec	*log_vector)
+{
+	struct xfs_log_vec	*lv;
+
+	for (lv = log_vector; lv; ) {
+		struct xfs_log_vec *next = lv->lv_next;
+		kmem_free(lv->lv_buf);
+		kmem_free(lv);
+		lv = next;
+	}
+}
+
+/*
+ * Mark all items committed and clear busy extents. We free the log vector
+ * chains in a separate pass so that we unpin the log items as quickly as
+ * possible.
+ */
+static void
+xlog_cil_committed(
+	void	*args,
+	int	abort)
+{
+	struct xfs_cil_ctx	*ctx = args;
+	struct xfs_mount	*mp = ctx->cil->xc_log->l_mp;
+
+	xfs_trans_committed_bulk(ctx->cil->xc_log->l_ailp, ctx->lv_chain,
+					ctx->start_lsn, abort);
+
+	xfs_alloc_busy_sort(&ctx->busy_extents);
+	xfs_alloc_busy_clear(mp, &ctx->busy_extents,
+			     (mp->m_flags & XFS_MOUNT_DISCARD) && !abort);
+
+	spin_lock(&ctx->cil->xc_cil_lock);
+	list_del(&ctx->committing);
+	spin_unlock(&ctx->cil->xc_cil_lock);
+
+	xlog_cil_free_logvec(ctx->lv_chain);
+
+	if (!list_empty(&ctx->busy_extents)) {
+		ASSERT(mp->m_flags & XFS_MOUNT_DISCARD);
+
+		xfs_discard_extents(mp, &ctx->busy_extents);
+		xfs_alloc_busy_clear(mp, &ctx->busy_extents, false);
+	}
+
+	kmem_free(ctx);
+}
+
+/*
+ * Push the Committed Item List to the log. If @push_seq flag is zero, then it
+ * is a background flush and so we can chose to ignore it. Otherwise, if the
+ * current sequence is the same as @push_seq we need to do a flush. If
+ * @push_seq is less than the current sequence, then it has already been
+ * flushed and we don't need to do anything - the caller will wait for it to
+ * complete if necessary.
+ *
+ * @push_seq is a value rather than a flag because that allows us to do an
+ * unlocked check of the sequence number for a match. Hence we can allows log
+ * forces to run racily and not issue pushes for the same sequence twice. If we
+ * get a race between multiple pushes for the same sequence they will block on
+ * the first one and then abort, hence avoiding needless pushes.
+ */
+STATIC int
+xlog_cil_push(
+	struct log		*log,
+	xfs_lsn_t		push_seq)
+{
+	struct xfs_cil		*cil = log->l_cilp;
+	struct xfs_log_vec	*lv;
+	struct xfs_cil_ctx	*ctx;
+	struct xfs_cil_ctx	*new_ctx;
+	struct xlog_in_core	*commit_iclog;
+	struct xlog_ticket	*tic;
+	int			num_lv;
+	int			num_iovecs;
+	int			len;
+	int			error = 0;
+	struct xfs_trans_header thdr;
+	struct xfs_log_iovec	lhdr;
+	struct xfs_log_vec	lvhdr = { NULL };
+	xfs_lsn_t		commit_lsn;
+
+	if (!cil)
+		return 0;
+
+	ASSERT(!push_seq || push_seq <= cil->xc_ctx->sequence);
+
+	new_ctx = kmem_zalloc(sizeof(*new_ctx), KM_SLEEP|KM_NOFS);
+	new_ctx->ticket = xlog_cil_ticket_alloc(log);
+
+	/*
+	 * Lock out transaction commit, but don't block for background pushes
+	 * unless we are well over the CIL space limit. See the definition of
+	 * XLOG_CIL_HARD_SPACE_LIMIT() for the full explanation of the logic
+	 * used here.
+	 */
+	if (!down_write_trylock(&cil->xc_ctx_lock)) {
+		if (!push_seq &&
+		    cil->xc_ctx->space_used < XLOG_CIL_HARD_SPACE_LIMIT(log))
+			goto out_free_ticket;
+		down_write(&cil->xc_ctx_lock);
+	}
+	ctx = cil->xc_ctx;
+
+	/* check if we've anything to push */
+	if (list_empty(&cil->xc_cil))
+		goto out_skip;
+
+	/* check for spurious background flush */
+	if (!push_seq && cil->xc_ctx->space_used < XLOG_CIL_SPACE_LIMIT(log))
+		goto out_skip;
+
+	/* check for a previously pushed seqeunce */
+	if (push_seq && push_seq < cil->xc_ctx->sequence)
+		goto out_skip;
+
+	/*
+	 * pull all the log vectors off the items in the CIL, and
+	 * remove the items from the CIL. We don't need the CIL lock
+	 * here because it's only needed on the transaction commit
+	 * side which is currently locked out by the flush lock.
+	 */
+	lv = NULL;
+	num_lv = 0;
+	num_iovecs = 0;
+	len = 0;
+	while (!list_empty(&cil->xc_cil)) {
+		struct xfs_log_item	*item;
+		int			i;
+
+		item = list_first_entry(&cil->xc_cil,
+					struct xfs_log_item, li_cil);
+		list_del_init(&item->li_cil);
+		if (!ctx->lv_chain)
+			ctx->lv_chain = item->li_lv;
+		else
+			lv->lv_next = item->li_lv;
+		lv = item->li_lv;
+		item->li_lv = NULL;
+
+		num_lv++;
+		num_iovecs += lv->lv_niovecs;
+		for (i = 0; i < lv->lv_niovecs; i++)
+			len += lv->lv_iovecp[i].i_len;
+	}
+
+	/*
+	 * initialise the new context and attach it to the CIL. Then attach
+	 * the current context to the CIL committing lsit so it can be found
+	 * during log forces to extract the commit lsn of the sequence that
+	 * needs to be forced.
+	 */
+	INIT_LIST_HEAD(&new_ctx->committing);
+	INIT_LIST_HEAD(&new_ctx->busy_extents);
+	new_ctx->sequence = ctx->sequence + 1;
+	new_ctx->cil = cil;
+	cil->xc_ctx = new_ctx;
+
+	/*
+	 * mirror the new sequence into the cil structure so that we can do
+	 * unlocked checks against the current sequence in log forces without
+	 * risking deferencing a freed context pointer.
+	 */
+	cil->xc_current_sequence = new_ctx->sequence;
+
+	/*
+	 * The switch is now done, so we can drop the context lock and move out
+	 * of a shared context. We can't just go straight to the commit record,
+	 * though - we need to synchronise with previous and future commits so
+	 * that the commit records are correctly ordered in the log to ensure
+	 * that we process items during log IO completion in the correct order.
+	 *
+	 * For example, if we get an EFI in one checkpoint and the EFD in the
+	 * next (e.g. due to log forces), we do not want the checkpoint with
+	 * the EFD to be committed before the checkpoint with the EFI.  Hence
+	 * we must strictly order the commit records of the checkpoints so
+	 * that: a) the checkpoint callbacks are attached to the iclogs in the
+	 * correct order; and b) the checkpoints are replayed in correct order
+	 * in log recovery.
+	 *
+	 * Hence we need to add this context to the committing context list so
+	 * that higher sequences will wait for us to write out a commit record
+	 * before they do.
+	 */
+	spin_lock(&cil->xc_cil_lock);
+	list_add(&ctx->committing, &cil->xc_committing);
+	spin_unlock(&cil->xc_cil_lock);
+	up_write(&cil->xc_ctx_lock);
+
+	/*
+	 * Build a checkpoint transaction header and write it to the log to
+	 * begin the transaction. We need to account for the space used by the
+	 * transaction header here as it is not accounted for in xlog_write().
+	 *
+	 * The LSN we need to pass to the log items on transaction commit is
+	 * the LSN reported by the first log vector write. If we use the commit
+	 * record lsn then we can move the tail beyond the grant write head.
+	 */
+	tic = ctx->ticket;
+	thdr.th_magic = XFS_TRANS_HEADER_MAGIC;
+	thdr.th_type = XFS_TRANS_CHECKPOINT;
+	thdr.th_tid = tic->t_tid;
+	thdr.th_num_items = num_iovecs;
+	lhdr.i_addr = &thdr;
+	lhdr.i_len = sizeof(xfs_trans_header_t);
+	lhdr.i_type = XLOG_REG_TYPE_TRANSHDR;
+	tic->t_curr_res -= lhdr.i_len + sizeof(xlog_op_header_t);
+
+	lvhdr.lv_niovecs = 1;
+	lvhdr.lv_iovecp = &lhdr;
+	lvhdr.lv_next = ctx->lv_chain;
+
+	error = xlog_write(log, &lvhdr, tic, &ctx->start_lsn, NULL, 0);
+	if (error)
+		goto out_abort_free_ticket;
+
+	/*
+	 * now that we've written the checkpoint into the log, strictly
+	 * order the commit records so replay will get them in the right order.
+	 */
+restart:
+	spin_lock(&cil->xc_cil_lock);
+	list_for_each_entry(new_ctx, &cil->xc_committing, committing) {
+		/*
+		 * Higher sequences will wait for this one so skip them.
+		 * Don't wait for own own sequence, either.
+		 */
+		if (new_ctx->sequence >= ctx->sequence)
+			continue;
+		if (!new_ctx->commit_lsn) {
+			/*
+			 * It is still being pushed! Wait for the push to
+			 * complete, then start again from the beginning.
+			 */
+			xlog_wait(&cil->xc_commit_wait, &cil->xc_cil_lock);
+			goto restart;
+		}
+	}
+	spin_unlock(&cil->xc_cil_lock);
+
+	/* xfs_log_done always frees the ticket on error. */
+	commit_lsn = xfs_log_done(log->l_mp, tic, &commit_iclog, 0);
+	if (commit_lsn == -1)
+		goto out_abort;
+
+	/* attach all the transactions w/ busy extents to iclog */
+	ctx->log_cb.cb_func = xlog_cil_committed;
+	ctx->log_cb.cb_arg = ctx;
+	error = xfs_log_notify(log->l_mp, commit_iclog, &ctx->log_cb);
+	if (error)
+		goto out_abort;
+
+	/*
+	 * now the checkpoint commit is complete and we've attached the
+	 * callbacks to the iclog we can assign the commit LSN to the context
+	 * and wake up anyone who is waiting for the commit to complete.
+	 */
+	spin_lock(&cil->xc_cil_lock);
+	ctx->commit_lsn = commit_lsn;
+	wake_up_all(&cil->xc_commit_wait);
+	spin_unlock(&cil->xc_cil_lock);
+
+	/* release the hounds! */
+	return xfs_log_release_iclog(log->l_mp, commit_iclog);
+
+out_skip:
+	up_write(&cil->xc_ctx_lock);
+out_free_ticket:
+	xfs_log_ticket_put(new_ctx->ticket);
+	kmem_free(new_ctx);
+	return 0;
+
+out_abort_free_ticket:
+	xfs_log_ticket_put(tic);
+out_abort:
+	xlog_cil_committed(ctx, XFS_LI_ABORTED);
+	return XFS_ERROR(EIO);
+}
+
+/*
+ * Commit a transaction with the given vector to the Committed Item List.
+ *
+ * To do this, we need to format the item, pin it in memory if required and
+ * account for the space used by the transaction. Once we have done that we
+ * need to release the unused reservation for the transaction, attach the
+ * transaction to the checkpoint context so we carry the busy extents through
+ * to checkpoint completion, and then unlock all the items in the transaction.
+ *
+ * For more specific information about the order of operations in
+ * xfs_log_commit_cil() please refer to the comments in
+ * xfs_trans_commit_iclog().
+ *
+ * Called with the context lock already held in read mode to lock out
+ * background commit, returns without it held once background commits are
+ * allowed again.
+ */
+void
+xfs_log_commit_cil(
+	struct xfs_mount	*mp,
+	struct xfs_trans	*tp,
+	struct xfs_log_vec	*log_vector,
+	xfs_lsn_t		*commit_lsn,
+	int			flags)
+{
+	struct log		*log = mp->m_log;
+	int			log_flags = 0;
+	int			push = 0;
+
+	if (flags & XFS_TRANS_RELEASE_LOG_RES)
+		log_flags = XFS_LOG_REL_PERM_RESERV;
+
+	/*
+	 * do all the hard work of formatting items (including memory
+	 * allocation) outside the CIL context lock. This prevents stalling CIL
+	 * pushes when we are low on memory and a transaction commit spends a
+	 * lot of time in memory reclaim.
+	 */
+	xlog_cil_format_items(log, log_vector);
+
+	/* lock out background commit */
+	down_read(&log->l_cilp->xc_ctx_lock);
+	if (commit_lsn)
+		*commit_lsn = log->l_cilp->xc_ctx->sequence;
+
+	xlog_cil_insert_items(log, log_vector, tp->t_ticket);
+
+	/* check we didn't blow the reservation */
+	if (tp->t_ticket->t_curr_res < 0)
+		xlog_print_tic_res(log->l_mp, tp->t_ticket);
+
+	/* attach the transaction to the CIL if it has any busy extents */
+	if (!list_empty(&tp->t_busy)) {
+		spin_lock(&log->l_cilp->xc_cil_lock);
+		list_splice_init(&tp->t_busy,
+					&log->l_cilp->xc_ctx->busy_extents);
+		spin_unlock(&log->l_cilp->xc_cil_lock);
+	}
+
+	tp->t_commit_lsn = *commit_lsn;
+	xfs_log_done(mp, tp->t_ticket, NULL, log_flags);
+	xfs_trans_unreserve_and_mod_sb(tp);
+
+	/*
+	 * Once all the items of the transaction have been copied to the CIL,
+	 * the items can be unlocked and freed.
+	 *
+	 * This needs to be done before we drop the CIL context lock because we
+	 * have to update state in the log items and unlock them before they go
+	 * to disk. If we don't, then the CIL checkpoint can race with us and
+	 * we can run checkpoint completion before we've updated and unlocked
+	 * the log items. This affects (at least) processing of stale buffers,
+	 * inodes and EFIs.
+	 */
+	xfs_trans_free_items(tp, *commit_lsn, 0);
+
+	/* check for background commit before unlock */
+	if (log->l_cilp->xc_ctx->space_used > XLOG_CIL_SPACE_LIMIT(log))
+		push = 1;
+
+	up_read(&log->l_cilp->xc_ctx_lock);
+
+	/*
+	 * We need to push CIL every so often so we don't cache more than we
+	 * can fit in the log. The limit really is that a checkpoint can't be
+	 * more than half the log (the current checkpoint is not allowed to
+	 * overwrite the previous checkpoint), but commit latency and memory
+	 * usage limit this to a smaller size in most cases.
+	 */
+	if (push)
+		xlog_cil_push(log, 0);
+}
+
+/*
+ * Conditionally push the CIL based on the sequence passed in.
+ *
+ * We only need to push if we haven't already pushed the sequence
+ * number given. Hence the only time we will trigger a push here is
+ * if the push sequence is the same as the current context.
+ *
+ * We return the current commit lsn to allow the callers to determine if a
+ * iclog flush is necessary following this call.
+ *
+ * XXX: Initially, just push the CIL unconditionally and return whatever
+ * commit lsn is there. It'll be empty, so this is broken for now.
+ */
+xfs_lsn_t
+xlog_cil_force_lsn(
+	struct log	*log,
+	xfs_lsn_t	sequence)
+{
+	struct xfs_cil		*cil = log->l_cilp;
+	struct xfs_cil_ctx	*ctx;
+	xfs_lsn_t		commit_lsn = NULLCOMMITLSN;
+
+	ASSERT(sequence <= cil->xc_current_sequence);
+
+	/*
+	 * check to see if we need to force out the current context.
+	 * xlog_cil_push() handles racing pushes for the same sequence,
+	 * so no need to deal with it here.
+	 */
+	if (sequence == cil->xc_current_sequence)
+		xlog_cil_push(log, sequence);
+
+	/*
+	 * See if we can find a previous sequence still committing.
+	 * We need to wait for all previous sequence commits to complete
+	 * before allowing the force of push_seq to go ahead. Hence block
+	 * on commits for those as well.
+	 */
+restart:
+	spin_lock(&cil->xc_cil_lock);
+	list_for_each_entry(ctx, &cil->xc_committing, committing) {
+		if (ctx->sequence > sequence)
+			continue;
+		if (!ctx->commit_lsn) {
+			/*
+			 * It is still being pushed! Wait for the push to
+			 * complete, then start again from the beginning.
+			 */
+			xlog_wait(&cil->xc_commit_wait, &cil->xc_cil_lock);
+			goto restart;
+		}
+		if (ctx->sequence != sequence)
+			continue;
+		/* found it! */
+		commit_lsn = ctx->commit_lsn;
+	}
+	spin_unlock(&cil->xc_cil_lock);
+	return commit_lsn;
+}
+
+/*
+ * Check if the current log item was first committed in this sequence.
+ * We can't rely on just the log item being in the CIL, we have to check
+ * the recorded commit sequence number.
+ *
+ * Note: for this to be used in a non-racy manner, it has to be called with
+ * CIL flushing locked out. As a result, it should only be used during the
+ * transaction commit process when deciding what to format into the item.
+ */
+bool
+xfs_log_item_in_current_chkpt(
+	struct xfs_log_item *lip)
+{
+	struct xfs_cil_ctx *ctx;
+
+	if (!(lip->li_mountp->m_flags & XFS_MOUNT_DELAYLOG))
+		return false;
+	if (list_empty(&lip->li_cil))
+		return false;
+
+	ctx = lip->li_mountp->m_log->l_cilp->xc_ctx;
+
+	/*
+	 * li_seq is written on the first commit of a log item to record the
+	 * first checkpoint it is written to. Hence if it is different to the
+	 * current sequence, we're in a new checkpoint.
+	 */
+	if (XFS_LSN_CMP(lip->li_seq, ctx->sequence) != 0)
+		return false;
+	return true;
+}
diff --git a/fs/xfs/xfs_log_priv.h b/fs/xfs/xfs_log_priv.h
new file mode 100644
index 00000000..2d3b6a49
--- /dev/null
+++ b/fs/xfs/xfs_log_priv.h
@@ -0,0 +1,668 @@
+/*
+ * Copyright (c) 2000-2003,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef	__XFS_LOG_PRIV_H__
+#define __XFS_LOG_PRIV_H__
+
+struct xfs_buf;
+struct log;
+struct xlog_ticket;
+struct xfs_mount;
+
+/*
+ * Macros, structures, prototypes for internal log manager use.
+ */
+
+#define XLOG_MIN_ICLOGS		2
+#define XLOG_MAX_ICLOGS		8
+#define XLOG_HEADER_MAGIC_NUM	0xFEEDbabe	/* Invalid cycle number */
+#define XLOG_VERSION_1		1
+#define XLOG_VERSION_2		2		/* Large IClogs, Log sunit */
+#define XLOG_VERSION_OKBITS	(XLOG_VERSION_1 | XLOG_VERSION_2)
+#define XLOG_MIN_RECORD_BSIZE	(16*1024)	/* eventually 32k */
+#define XLOG_BIG_RECORD_BSIZE	(32*1024)	/* 32k buffers */
+#define XLOG_MAX_RECORD_BSIZE	(256*1024)
+#define XLOG_HEADER_CYCLE_SIZE	(32*1024)	/* cycle data in header */
+#define XLOG_MIN_RECORD_BSHIFT	14		/* 16384 == 1 << 14 */
+#define XLOG_BIG_RECORD_BSHIFT	15		/* 32k == 1 << 15 */
+#define XLOG_MAX_RECORD_BSHIFT	18		/* 256k == 1 << 18 */
+#define XLOG_BTOLSUNIT(log, b)  (((b)+(log)->l_mp->m_sb.sb_logsunit-1) / \
+                                 (log)->l_mp->m_sb.sb_logsunit)
+#define XLOG_LSUNITTOB(log, su) ((su) * (log)->l_mp->m_sb.sb_logsunit)
+
+#define XLOG_HEADER_SIZE	512
+
+#define XLOG_REC_SHIFT(log) \
+	BTOBB(1 << (xfs_sb_version_haslogv2(&log->l_mp->m_sb) ? \
+	 XLOG_MAX_RECORD_BSHIFT : XLOG_BIG_RECORD_BSHIFT))
+#define XLOG_TOTAL_REC_SHIFT(log) \
+	BTOBB(XLOG_MAX_ICLOGS << (xfs_sb_version_haslogv2(&log->l_mp->m_sb) ? \
+	 XLOG_MAX_RECORD_BSHIFT : XLOG_BIG_RECORD_BSHIFT))
+
+static inline xfs_lsn_t xlog_assign_lsn(uint cycle, uint block)
+{
+	return ((xfs_lsn_t)cycle << 32) | block;
+}
+
+static inline uint xlog_get_cycle(char *ptr)
+{
+	if (be32_to_cpu(*(__be32 *)ptr) == XLOG_HEADER_MAGIC_NUM)
+		return be32_to_cpu(*((__be32 *)ptr + 1));
+	else
+		return be32_to_cpu(*(__be32 *)ptr);
+}
+
+#define BLK_AVG(blk1, blk2)	((blk1+blk2) >> 1)
+
+#ifdef __KERNEL__
+
+/*
+ * get client id from packed copy.
+ *
+ * this hack is here because the xlog_pack code copies four bytes
+ * of xlog_op_header containing the fields oh_clientid, oh_flags
+ * and oh_res2 into the packed copy.
+ *
+ * later on this four byte chunk is treated as an int and the
+ * client id is pulled out.
+ *
+ * this has endian issues, of course.
+ */
+static inline uint xlog_get_client_id(__be32 i)
+{
+	return be32_to_cpu(i) >> 24;
+}
+
+/*
+ * In core log state
+ */
+#define XLOG_STATE_ACTIVE    0x0001 /* Current IC log being written to */
+#define XLOG_STATE_WANT_SYNC 0x0002 /* Want to sync this iclog; no more writes */
+#define XLOG_STATE_SYNCING   0x0004 /* This IC log is syncing */
+#define XLOG_STATE_DONE_SYNC 0x0008 /* Done syncing to disk */
+#define XLOG_STATE_DO_CALLBACK \
+			     0x0010 /* Process callback functions */
+#define XLOG_STATE_CALLBACK  0x0020 /* Callback functions now */
+#define XLOG_STATE_DIRTY     0x0040 /* Dirty IC log, not ready for ACTIVE status*/
+#define XLOG_STATE_IOERROR   0x0080 /* IO error happened in sync'ing log */
+#define XLOG_STATE_ALL	     0x7FFF /* All possible valid flags */
+#define XLOG_STATE_NOTUSED   0x8000 /* This IC log not being used */
+#endif	/* __KERNEL__ */
+
+/*
+ * Flags to log operation header
+ *
+ * The first write of a new transaction will be preceded with a start
+ * record, XLOG_START_TRANS.  Once a transaction is committed, a commit
+ * record is written, XLOG_COMMIT_TRANS.  If a single region can not fit into
+ * the remainder of the current active in-core log, it is split up into
+ * multiple regions.  Each partial region will be marked with a
+ * XLOG_CONTINUE_TRANS until the last one, which gets marked with XLOG_END_TRANS.
+ *
+ */
+#define XLOG_START_TRANS	0x01	/* Start a new transaction */
+#define XLOG_COMMIT_TRANS	0x02	/* Commit this transaction */
+#define XLOG_CONTINUE_TRANS	0x04	/* Cont this trans into new region */
+#define XLOG_WAS_CONT_TRANS	0x08	/* Cont this trans into new region */
+#define XLOG_END_TRANS		0x10	/* End a continued transaction */
+#define XLOG_UNMOUNT_TRANS	0x20	/* Unmount a filesystem transaction */
+
+#ifdef __KERNEL__
+/*
+ * Flags to log ticket
+ */
+#define XLOG_TIC_INITED		0x1	/* has been initialized */
+#define XLOG_TIC_PERM_RESERV	0x2	/* permanent reservation */
+
+#define XLOG_TIC_FLAGS \
+	{ XLOG_TIC_INITED,	"XLOG_TIC_INITED" }, \
+	{ XLOG_TIC_PERM_RESERV,	"XLOG_TIC_PERM_RESERV" }
+
+#endif	/* __KERNEL__ */
+
+#define XLOG_UNMOUNT_TYPE	0x556e	/* Un for Unmount */
+
+/*
+ * Flags for log structure
+ */
+#define XLOG_CHKSUM_MISMATCH	0x1	/* used only during recovery */
+#define XLOG_ACTIVE_RECOVERY	0x2	/* in the middle of recovery */
+#define	XLOG_RECOVERY_NEEDED	0x4	/* log was recovered */
+#define XLOG_IO_ERROR		0x8	/* log hit an I/O error, and being
+					   shutdown */
+#define XLOG_TAIL_WARN		0x10	/* log tail verify warning issued */
+
+typedef __uint32_t xlog_tid_t;
+
+#ifdef __KERNEL__
+/*
+ * Below are states for covering allocation transactions.
+ * By covering, we mean changing the h_tail_lsn in the last on-disk
+ * log write such that no allocation transactions will be re-done during
+ * recovery after a system crash. Recovery starts at the last on-disk
+ * log write.
+ *
+ * These states are used to insert dummy log entries to cover
+ * space allocation transactions which can undo non-transactional changes
+ * after a crash. Writes to a file with space
+ * already allocated do not result in any transactions. Allocations
+ * might include space beyond the EOF. So if we just push the EOF a
+ * little, the last transaction for the file could contain the wrong
+ * size. If there is no file system activity, after an allocation
+ * transaction, and the system crashes, the allocation transaction
+ * will get replayed and the file will be truncated. This could
+ * be hours/days/... after the allocation occurred.
+ *
+ * The fix for this is to do two dummy transactions when the
+ * system is idle. We need two dummy transaction because the h_tail_lsn
+ * in the log record header needs to point beyond the last possible
+ * non-dummy transaction. The first dummy changes the h_tail_lsn to
+ * the first transaction before the dummy. The second dummy causes
+ * h_tail_lsn to point to the first dummy. Recovery starts at h_tail_lsn.
+ *
+ * These dummy transactions get committed when everything
+ * is idle (after there has been some activity).
+ *
+ * There are 5 states used to control this.
+ *
+ *  IDLE -- no logging has been done on the file system or
+ *		we are done covering previous transactions.
+ *  NEED -- logging has occurred and we need a dummy transaction
+ *		when the log becomes idle.
+ *  DONE -- we were in the NEED state and have committed a dummy
+ *		transaction.
+ *  NEED2 -- we detected that a dummy transaction has gone to the
+ *		on disk log with no other transactions.
+ *  DONE2 -- we committed a dummy transaction when in the NEED2 state.
+ *
+ * There are two places where we switch states:
+ *
+ * 1.) In xfs_sync, when we detect an idle log and are in NEED or NEED2.
+ *	We commit the dummy transaction and switch to DONE or DONE2,
+ *	respectively. In all other states, we don't do anything.
+ *
+ * 2.) When we finish writing the on-disk log (xlog_state_clean_log).
+ *
+ *	No matter what state we are in, if this isn't the dummy
+ *	transaction going out, the next state is NEED.
+ *	So, if we aren't in the DONE or DONE2 states, the next state
+ *	is NEED. We can't be finishing a write of the dummy record
+ *	unless it was committed and the state switched to DONE or DONE2.
+ *
+ *	If we are in the DONE state and this was a write of the
+ *		dummy transaction, we move to NEED2.
+ *
+ *	If we are in the DONE2 state and this was a write of the
+ *		dummy transaction, we move to IDLE.
+ *
+ *
+ * Writing only one dummy transaction can get appended to
+ * one file space allocation. When this happens, the log recovery
+ * code replays the space allocation and a file could be truncated.
+ * This is why we have the NEED2 and DONE2 states before going idle.
+ */
+
+#define XLOG_STATE_COVER_IDLE	0
+#define XLOG_STATE_COVER_NEED	1
+#define XLOG_STATE_COVER_DONE	2
+#define XLOG_STATE_COVER_NEED2	3
+#define XLOG_STATE_COVER_DONE2	4
+
+#define XLOG_COVER_OPS		5
+
+
+/* Ticket reservation region accounting */ 
+#define XLOG_TIC_LEN_MAX	15
+
+/*
+ * Reservation region
+ * As would be stored in xfs_log_iovec but without the i_addr which
+ * we don't care about.
+ */
+typedef struct xlog_res {
+	uint	r_len;	/* region length		:4 */
+	uint	r_type;	/* region's transaction type	:4 */
+} xlog_res_t;
+
+typedef struct xlog_ticket {
+	wait_queue_head_t  t_wait;	 /* ticket wait queue */
+	struct list_head   t_queue;	 /* reserve/write queue */
+	xlog_tid_t	   t_tid;	 /* transaction identifier	 : 4  */
+	atomic_t	   t_ref;	 /* ticket reference count       : 4  */
+	int		   t_curr_res;	 /* current reservation in bytes : 4  */
+	int		   t_unit_res;	 /* unit reservation in bytes    : 4  */
+	char		   t_ocnt;	 /* original count		 : 1  */
+	char		   t_cnt;	 /* current count		 : 1  */
+	char		   t_clientid;	 /* who does this belong to;	 : 1  */
+	char		   t_flags;	 /* properties of reservation	 : 1  */
+	uint		   t_trans_type; /* transaction type             : 4  */
+
+        /* reservation array fields */
+	uint		   t_res_num;                    /* num in array : 4 */
+	uint		   t_res_num_ophdrs;		 /* num op hdrs  : 4 */
+	uint		   t_res_arr_sum;		 /* array sum    : 4 */
+	uint		   t_res_o_flow;		 /* sum overflow : 4 */
+	xlog_res_t	   t_res_arr[XLOG_TIC_LEN_MAX];  /* array of res : 8 * 15 */ 
+} xlog_ticket_t;
+
+#endif
+
+
+typedef struct xlog_op_header {
+	__be32	   oh_tid;	/* transaction id of operation	:  4 b */
+	__be32	   oh_len;	/* bytes in data region		:  4 b */
+	__u8	   oh_clientid;	/* who sent me this		:  1 b */
+	__u8	   oh_flags;	/*				:  1 b */
+	__u16	   oh_res2;	/* 32 bit align			:  2 b */
+} xlog_op_header_t;
+
+
+/* valid values for h_fmt */
+#define XLOG_FMT_UNKNOWN  0
+#define XLOG_FMT_LINUX_LE 1
+#define XLOG_FMT_LINUX_BE 2
+#define XLOG_FMT_IRIX_BE  3
+
+/* our fmt */
+#ifdef XFS_NATIVE_HOST
+#define XLOG_FMT XLOG_FMT_LINUX_BE
+#else
+#define XLOG_FMT XLOG_FMT_LINUX_LE
+#endif
+
+typedef struct xlog_rec_header {
+	__be32	  h_magicno;	/* log record (LR) identifier		:  4 */
+	__be32	  h_cycle;	/* write cycle of log			:  4 */
+	__be32	  h_version;	/* LR version				:  4 */
+	__be32	  h_len;	/* len in bytes; should be 64-bit aligned: 4 */
+	__be64	  h_lsn;	/* lsn of this LR			:  8 */
+	__be64	  h_tail_lsn;	/* lsn of 1st LR w/ buffers not committed: 8 */
+	__be32	  h_chksum;	/* may not be used; non-zero if used	:  4 */
+	__be32	  h_prev_block; /* block number to previous LR		:  4 */
+	__be32	  h_num_logops;	/* number of log operations in this LR	:  4 */
+	__be32	  h_cycle_data[XLOG_HEADER_CYCLE_SIZE / BBSIZE];
+	/* new fields */
+	__be32    h_fmt;        /* format of log record                 :  4 */
+	uuid_t	  h_fs_uuid;    /* uuid of FS                           : 16 */
+	__be32	  h_size;	/* iclog size				:  4 */
+} xlog_rec_header_t;
+
+typedef struct xlog_rec_ext_header {
+	__be32	  xh_cycle;	/* write cycle of log			: 4 */
+	__be32	  xh_cycle_data[XLOG_HEADER_CYCLE_SIZE / BBSIZE]; /*	: 256 */
+} xlog_rec_ext_header_t;
+
+#ifdef __KERNEL__
+
+/*
+ * Quite misnamed, because this union lays out the actual on-disk log buffer.
+ */
+typedef union xlog_in_core2 {
+	xlog_rec_header_t	hic_header;
+	xlog_rec_ext_header_t	hic_xheader;
+	char			hic_sector[XLOG_HEADER_SIZE];
+} xlog_in_core_2_t;
+
+/*
+ * - A log record header is 512 bytes.  There is plenty of room to grow the
+ *	xlog_rec_header_t into the reserved space.
+ * - ic_data follows, so a write to disk can start at the beginning of
+ *	the iclog.
+ * - ic_forcewait is used to implement synchronous forcing of the iclog to disk.
+ * - ic_next is the pointer to the next iclog in the ring.
+ * - ic_bp is a pointer to the buffer used to write this incore log to disk.
+ * - ic_log is a pointer back to the global log structure.
+ * - ic_callback is a linked list of callback function/argument pairs to be
+ *	called after an iclog finishes writing.
+ * - ic_size is the full size of the header plus data.
+ * - ic_offset is the current number of bytes written to in this iclog.
+ * - ic_refcnt is bumped when someone is writing to the log.
+ * - ic_state is the state of the iclog.
+ *
+ * Because of cacheline contention on large machines, we need to separate
+ * various resources onto different cachelines. To start with, make the
+ * structure cacheline aligned. The following fields can be contended on
+ * by independent processes:
+ *
+ *	- ic_callback_*
+ *	- ic_refcnt
+ *	- fields protected by the global l_icloglock
+ *
+ * so we need to ensure that these fields are located in separate cachelines.
+ * We'll put all the read-only and l_icloglock fields in the first cacheline,
+ * and move everything else out to subsequent cachelines.
+ */
+typedef struct xlog_in_core {
+	wait_queue_head_t	ic_force_wait;
+	wait_queue_head_t	ic_write_wait;
+	struct xlog_in_core	*ic_next;
+	struct xlog_in_core	*ic_prev;
+	struct xfs_buf		*ic_bp;
+	struct log		*ic_log;
+	int			ic_size;
+	int			ic_offset;
+	int			ic_bwritecnt;
+	unsigned short		ic_state;
+	char			*ic_datap;	/* pointer to iclog data */
+
+	/* Callback structures need their own cacheline */
+	spinlock_t		ic_callback_lock ____cacheline_aligned_in_smp;
+	xfs_log_callback_t	*ic_callback;
+	xfs_log_callback_t	**ic_callback_tail;
+
+	/* reference counts need their own cacheline */
+	atomic_t		ic_refcnt ____cacheline_aligned_in_smp;
+	xlog_in_core_2_t	*ic_data;
+#define ic_header	ic_data->hic_header
+} xlog_in_core_t;
+
+/*
+ * The CIL context is used to aggregate per-transaction details as well be
+ * passed to the iclog for checkpoint post-commit processing.  After being
+ * passed to the iclog, another context needs to be allocated for tracking the
+ * next set of transactions to be aggregated into a checkpoint.
+ */
+struct xfs_cil;
+
+struct xfs_cil_ctx {
+	struct xfs_cil		*cil;
+	xfs_lsn_t		sequence;	/* chkpt sequence # */
+	xfs_lsn_t		start_lsn;	/* first LSN of chkpt commit */
+	xfs_lsn_t		commit_lsn;	/* chkpt commit record lsn */
+	struct xlog_ticket	*ticket;	/* chkpt ticket */
+	int			nvecs;		/* number of regions */
+	int			space_used;	/* aggregate size of regions */
+	struct list_head	busy_extents;	/* busy extents in chkpt */
+	struct xfs_log_vec	*lv_chain;	/* logvecs being pushed */
+	xfs_log_callback_t	log_cb;		/* completion callback hook. */
+	struct list_head	committing;	/* ctx committing list */
+};
+
+/*
+ * Committed Item List structure
+ *
+ * This structure is used to track log items that have been committed but not
+ * yet written into the log. It is used only when the delayed logging mount
+ * option is enabled.
+ *
+ * This structure tracks the list of committing checkpoint contexts so
+ * we can avoid the problem of having to hold out new transactions during a
+ * flush until we have a the commit record LSN of the checkpoint. We can
+ * traverse the list of committing contexts in xlog_cil_push_lsn() to find a
+ * sequence match and extract the commit LSN directly from there. If the
+ * checkpoint is still in the process of committing, we can block waiting for
+ * the commit LSN to be determined as well. This should make synchronous
+ * operations almost as efficient as the old logging methods.
+ */
+struct xfs_cil {
+	struct log		*xc_log;
+	struct list_head	xc_cil;
+	spinlock_t		xc_cil_lock;
+	struct xfs_cil_ctx	*xc_ctx;
+	struct rw_semaphore	xc_ctx_lock;
+	struct list_head	xc_committing;
+	wait_queue_head_t	xc_commit_wait;
+	xfs_lsn_t		xc_current_sequence;
+};
+
+/*
+ * The amount of log space we allow the CIL to aggregate is difficult to size.
+ * Whatever we choose, we have to make sure we can get a reservation for the
+ * log space effectively, that it is large enough to capture sufficient
+ * relogging to reduce log buffer IO significantly, but it is not too large for
+ * the log or induces too much latency when writing out through the iclogs. We
+ * track both space consumed and the number of vectors in the checkpoint
+ * context, so we need to decide which to use for limiting.
+ *
+ * Every log buffer we write out during a push needs a header reserved, which
+ * is at least one sector and more for v2 logs. Hence we need a reservation of
+ * at least 512 bytes per 32k of log space just for the LR headers. That means
+ * 16KB of reservation per megabyte of delayed logging space we will consume,
+ * plus various headers.  The number of headers will vary based on the num of
+ * io vectors, so limiting on a specific number of vectors is going to result
+ * in transactions of varying size. IOWs, it is more consistent to track and
+ * limit space consumed in the log rather than by the number of objects being
+ * logged in order to prevent checkpoint ticket overruns.
+ *
+ * Further, use of static reservations through the log grant mechanism is
+ * problematic. It introduces a lot of complexity (e.g. reserve grant vs write
+ * grant) and a significant deadlock potential because regranting write space
+ * can block on log pushes. Hence if we have to regrant log space during a log
+ * push, we can deadlock.
+ *
+ * However, we can avoid this by use of a dynamic "reservation stealing"
+ * technique during transaction commit whereby unused reservation space in the
+ * transaction ticket is transferred to the CIL ctx commit ticket to cover the
+ * space needed by the checkpoint transaction. This means that we never need to
+ * specifically reserve space for the CIL checkpoint transaction, nor do we
+ * need to regrant space once the checkpoint completes. This also means the
+ * checkpoint transaction ticket is specific to the checkpoint context, rather
+ * than the CIL itself.
+ *
+ * With dynamic reservations, we can effectively make up arbitrary limits for
+ * the checkpoint size so long as they don't violate any other size rules.
+ * Recovery imposes a rule that no transaction exceed half the log, so we are
+ * limited by that.  Furthermore, the log transaction reservation subsystem
+ * tries to keep 25% of the log free, so we need to keep below that limit or we
+ * risk running out of free log space to start any new transactions.
+ *
+ * In order to keep background CIL push efficient, we will set a lower
+ * threshold at which background pushing is attempted without blocking current
+ * transaction commits.  A separate, higher bound defines when CIL pushes are
+ * enforced to ensure we stay within our maximum checkpoint size bounds.
+ * threshold, yet give us plenty of space for aggregation on large logs.
+ */
+#define XLOG_CIL_SPACE_LIMIT(log)	(log->l_logsize >> 3)
+#define XLOG_CIL_HARD_SPACE_LIMIT(log)	(3 * (log->l_logsize >> 4))
+
+/*
+ * The reservation head lsn is not made up of a cycle number and block number.
+ * Instead, it uses a cycle number and byte number.  Logs don't expect to
+ * overflow 31 bits worth of byte offset, so using a byte number will mean
+ * that round off problems won't occur when releasing partial reservations.
+ */
+typedef struct log {
+	/* The following fields don't need locking */
+	struct xfs_mount	*l_mp;	        /* mount point */
+	struct xfs_ail		*l_ailp;	/* AIL log is working with */
+	struct xfs_cil		*l_cilp;	/* CIL log is working with */
+	struct xfs_buf		*l_xbuf;        /* extra buffer for log
+						 * wrapping */
+	struct xfs_buftarg	*l_targ;        /* buftarg of log */
+	uint			l_flags;
+	uint			l_quotaoffs_flag; /* XFS_DQ_*, for QUOTAOFFs */
+	struct list_head	*l_buf_cancel_table;
+	int			l_iclog_hsize;  /* size of iclog header */
+	int			l_iclog_heads;  /* # of iclog header sectors */
+	uint			l_sectBBsize;   /* sector size in BBs (2^n) */
+	int			l_iclog_size;	/* size of log in bytes */
+	int			l_iclog_size_log; /* log power size of log */
+	int			l_iclog_bufs;	/* number of iclog buffers */
+	xfs_daddr_t		l_logBBstart;   /* start block of log */
+	int			l_logsize;      /* size of log in bytes */
+	int			l_logBBsize;    /* size of log in BB chunks */
+
+	/* The following block of fields are changed while holding icloglock */
+	wait_queue_head_t	l_flush_wait ____cacheline_aligned_in_smp;
+						/* waiting for iclog flush */
+	int			l_covered_state;/* state of "covering disk
+						 * log entries" */
+	xlog_in_core_t		*l_iclog;       /* head log queue	*/
+	spinlock_t		l_icloglock;    /* grab to change iclog state */
+	int			l_curr_cycle;   /* Cycle number of log writes */
+	int			l_prev_cycle;   /* Cycle number before last
+						 * block increment */
+	int			l_curr_block;   /* current logical log block */
+	int			l_prev_block;   /* previous logical log block */
+
+	/*
+	 * l_last_sync_lsn and l_tail_lsn are atomics so they can be set and
+	 * read without needing to hold specific locks. To avoid operations
+	 * contending with other hot objects, place each of them on a separate
+	 * cacheline.
+	 */
+	/* lsn of last LR on disk */
+	atomic64_t		l_last_sync_lsn ____cacheline_aligned_in_smp;
+	/* lsn of 1st LR with unflushed * buffers */
+	atomic64_t		l_tail_lsn ____cacheline_aligned_in_smp;
+
+	/*
+	 * ticket grant locks, queues and accounting have their own cachlines
+	 * as these are quite hot and can be operated on concurrently.
+	 */
+	spinlock_t		l_grant_reserve_lock ____cacheline_aligned_in_smp;
+	struct list_head	l_reserveq;
+	atomic64_t		l_grant_reserve_head;
+
+	spinlock_t		l_grant_write_lock ____cacheline_aligned_in_smp;
+	struct list_head	l_writeq;
+	atomic64_t		l_grant_write_head;
+
+	/* The following field are used for debugging; need to hold icloglock */
+#ifdef DEBUG
+	char			*l_iclog_bak[XLOG_MAX_ICLOGS];
+#endif
+
+} xlog_t;
+
+#define XLOG_BUF_CANCEL_BUCKET(log, blkno) \
+	((log)->l_buf_cancel_table + ((__uint64_t)blkno % XLOG_BC_TABLE_SIZE))
+
+#define XLOG_FORCED_SHUTDOWN(log)	((log)->l_flags & XLOG_IO_ERROR)
+
+/* common routines */
+extern xfs_lsn_t xlog_assign_tail_lsn(struct xfs_mount *mp);
+extern int	 xlog_recover(xlog_t *log);
+extern int	 xlog_recover_finish(xlog_t *log);
+extern void	 xlog_pack_data(xlog_t *log, xlog_in_core_t *iclog, int);
+
+extern kmem_zone_t *xfs_log_ticket_zone;
+struct xlog_ticket *xlog_ticket_alloc(struct log *log, int unit_bytes,
+				int count, char client, uint xflags,
+				int alloc_flags);
+
+
+static inline void
+xlog_write_adv_cnt(void **ptr, int *len, int *off, size_t bytes)
+{
+	*ptr += bytes;
+	*len -= bytes;
+	*off += bytes;
+}
+
+void	xlog_print_tic_res(struct xfs_mount *mp, struct xlog_ticket *ticket);
+int	xlog_write(struct log *log, struct xfs_log_vec *log_vector,
+				struct xlog_ticket *tic, xfs_lsn_t *start_lsn,
+				xlog_in_core_t **commit_iclog, uint flags);
+
+/*
+ * When we crack an atomic LSN, we sample it first so that the value will not
+ * change while we are cracking it into the component values. This means we
+ * will always get consistent component values to work from. This should always
+ * be used to sample and crack LSNs that are stored and updated in atomic
+ * variables.
+ */
+static inline void
+xlog_crack_atomic_lsn(atomic64_t *lsn, uint *cycle, uint *block)
+{
+	xfs_lsn_t val = atomic64_read(lsn);
+
+	*cycle = CYCLE_LSN(val);
+	*block = BLOCK_LSN(val);
+}
+
+/*
+ * Calculate and assign a value to an atomic LSN variable from component pieces.
+ */
+static inline void
+xlog_assign_atomic_lsn(atomic64_t *lsn, uint cycle, uint block)
+{
+	atomic64_set(lsn, xlog_assign_lsn(cycle, block));
+}
+
+/*
+ * When we crack the grant head, we sample it first so that the value will not
+ * change while we are cracking it into the component values. This means we
+ * will always get consistent component values to work from.
+ */
+static inline void
+xlog_crack_grant_head_val(int64_t val, int *cycle, int *space)
+{
+	*cycle = val >> 32;
+	*space = val & 0xffffffff;
+}
+
+static inline void
+xlog_crack_grant_head(atomic64_t *head, int *cycle, int *space)
+{
+	xlog_crack_grant_head_val(atomic64_read(head), cycle, space);
+}
+
+static inline int64_t
+xlog_assign_grant_head_val(int cycle, int space)
+{
+	return ((int64_t)cycle << 32) | space;
+}
+
+static inline void
+xlog_assign_grant_head(atomic64_t *head, int cycle, int space)
+{
+	atomic64_set(head, xlog_assign_grant_head_val(cycle, space));
+}
+
+/*
+ * Committed Item List interfaces
+ */
+int	xlog_cil_init(struct log *log);
+void	xlog_cil_init_post_recovery(struct log *log);
+void	xlog_cil_destroy(struct log *log);
+
+/*
+ * CIL force routines
+ */
+xfs_lsn_t xlog_cil_force_lsn(struct log *log, xfs_lsn_t sequence);
+
+static inline void
+xlog_cil_force(struct log *log)
+{
+	xlog_cil_force_lsn(log, log->l_cilp->xc_current_sequence);
+}
+
+/*
+ * Unmount record type is used as a pseudo transaction type for the ticket.
+ * It's value must be outside the range of XFS_TRANS_* values.
+ */
+#define XLOG_UNMOUNT_REC_TYPE	(-1U)
+
+/*
+ * Wrapper function for waiting on a wait queue serialised against wakeups
+ * by a spinlock. This matches the semantics of all the wait queues used in the
+ * log code.
+ */
+static inline void xlog_wait(wait_queue_head_t *wq, spinlock_t *lock)
+{
+	DECLARE_WAITQUEUE(wait, current);
+
+	add_wait_queue_exclusive(wq, &wait);
+	__set_current_state(TASK_UNINTERRUPTIBLE);
+	spin_unlock(lock);
+	schedule();
+	remove_wait_queue(wq, &wait);
+}
+#endif	/* __KERNEL__ */
+
+#endif	/* __XFS_LOG_PRIV_H__ */
diff --git a/fs/xfs/xfs_log_recover.c b/fs/xfs/xfs_log_recover.c
new file mode 100644
index 00000000..b75fd67c
--- /dev/null
+++ b/fs/xfs/xfs_log_recover.c
@@ -0,0 +1,3843 @@
+/*
+ * Copyright (c) 2000-2006 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_mount.h"
+#include "xfs_error.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_alloc_btree.h"
+#include "xfs_ialloc_btree.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_inode_item.h"
+#include "xfs_alloc.h"
+#include "xfs_ialloc.h"
+#include "xfs_log_priv.h"
+#include "xfs_buf_item.h"
+#include "xfs_log_recover.h"
+#include "xfs_extfree_item.h"
+#include "xfs_trans_priv.h"
+#include "xfs_quota.h"
+#include "xfs_rw.h"
+#include "xfs_utils.h"
+#include "xfs_trace.h"
+
+STATIC int	xlog_find_zeroed(xlog_t *, xfs_daddr_t *);
+STATIC int	xlog_clear_stale_blocks(xlog_t *, xfs_lsn_t);
+#if defined(DEBUG)
+STATIC void	xlog_recover_check_summary(xlog_t *);
+#else
+#define	xlog_recover_check_summary(log)
+#endif
+
+/*
+ * This structure is used during recovery to record the buf log items which
+ * have been canceled and should not be replayed.
+ */
+struct xfs_buf_cancel {
+	xfs_daddr_t		bc_blkno;
+	uint			bc_len;
+	int			bc_refcount;
+	struct list_head	bc_list;
+};
+
+/*
+ * Sector aligned buffer routines for buffer create/read/write/access
+ */
+
+/*
+ * Verify the given count of basic blocks is valid number of blocks
+ * to specify for an operation involving the given XFS log buffer.
+ * Returns nonzero if the count is valid, 0 otherwise.
+ */
+
+static inline int
+xlog_buf_bbcount_valid(
+	xlog_t		*log,
+	int		bbcount)
+{
+	return bbcount > 0 && bbcount <= log->l_logBBsize;
+}
+
+/*
+ * Allocate a buffer to hold log data.  The buffer needs to be able
+ * to map to a range of nbblks basic blocks at any valid (basic
+ * block) offset within the log.
+ */
+STATIC xfs_buf_t *
+xlog_get_bp(
+	xlog_t		*log,
+	int		nbblks)
+{
+	if (!xlog_buf_bbcount_valid(log, nbblks)) {
+		xfs_warn(log->l_mp, "Invalid block length (0x%x) for buffer",
+			nbblks);
+		XFS_ERROR_REPORT(__func__, XFS_ERRLEVEL_HIGH, log->l_mp);
+		return NULL;
+	}
+
+	/*
+	 * We do log I/O in units of log sectors (a power-of-2
+	 * multiple of the basic block size), so we round up the
+	 * requested size to accommodate the basic blocks required
+	 * for complete log sectors.
+	 *
+	 * In addition, the buffer may be used for a non-sector-
+	 * aligned block offset, in which case an I/O of the
+	 * requested size could extend beyond the end of the
+	 * buffer.  If the requested size is only 1 basic block it
+	 * will never straddle a sector boundary, so this won't be
+	 * an issue.  Nor will this be a problem if the log I/O is
+	 * done in basic blocks (sector size 1).  But otherwise we
+	 * extend the buffer by one extra log sector to ensure
+	 * there's space to accommodate this possibility.
+	 */
+	if (nbblks > 1 && log->l_sectBBsize > 1)
+		nbblks += log->l_sectBBsize;
+	nbblks = round_up(nbblks, log->l_sectBBsize);
+
+	return xfs_buf_get_uncached(log->l_mp->m_logdev_targp,
+					BBTOB(nbblks), 0);
+}
+
+STATIC void
+xlog_put_bp(
+	xfs_buf_t	*bp)
+{
+	xfs_buf_free(bp);
+}
+
+/*
+ * Return the address of the start of the given block number's data
+ * in a log buffer.  The buffer covers a log sector-aligned region.
+ */
+STATIC xfs_caddr_t
+xlog_align(
+	xlog_t		*log,
+	xfs_daddr_t	blk_no,
+	int		nbblks,
+	xfs_buf_t	*bp)
+{
+	xfs_daddr_t	offset = blk_no & ((xfs_daddr_t)log->l_sectBBsize - 1);
+
+	ASSERT(BBTOB(offset + nbblks) <= XFS_BUF_SIZE(bp));
+	return XFS_BUF_PTR(bp) + BBTOB(offset);
+}
+
+
+/*
+ * nbblks should be uint, but oh well.  Just want to catch that 32-bit length.
+ */
+STATIC int
+xlog_bread_noalign(
+	xlog_t		*log,
+	xfs_daddr_t	blk_no,
+	int		nbblks,
+	xfs_buf_t	*bp)
+{
+	int		error;
+
+	if (!xlog_buf_bbcount_valid(log, nbblks)) {
+		xfs_warn(log->l_mp, "Invalid block length (0x%x) for buffer",
+			nbblks);
+		XFS_ERROR_REPORT(__func__, XFS_ERRLEVEL_HIGH, log->l_mp);
+		return EFSCORRUPTED;
+	}
+
+	blk_no = round_down(blk_no, log->l_sectBBsize);
+	nbblks = round_up(nbblks, log->l_sectBBsize);
+
+	ASSERT(nbblks > 0);
+	ASSERT(BBTOB(nbblks) <= XFS_BUF_SIZE(bp));
+
+	XFS_BUF_SET_ADDR(bp, log->l_logBBstart + blk_no);
+	XFS_BUF_READ(bp);
+	XFS_BUF_BUSY(bp);
+	XFS_BUF_SET_COUNT(bp, BBTOB(nbblks));
+	XFS_BUF_SET_TARGET(bp, log->l_mp->m_logdev_targp);
+
+	xfsbdstrat(log->l_mp, bp);
+	error = xfs_buf_iowait(bp);
+	if (error)
+		xfs_ioerror_alert("xlog_bread", log->l_mp,
+				  bp, XFS_BUF_ADDR(bp));
+	return error;
+}
+
+STATIC int
+xlog_bread(
+	xlog_t		*log,
+	xfs_daddr_t	blk_no,
+	int		nbblks,
+	xfs_buf_t	*bp,
+	xfs_caddr_t	*offset)
+{
+	int		error;
+
+	error = xlog_bread_noalign(log, blk_no, nbblks, bp);
+	if (error)
+		return error;
+
+	*offset = xlog_align(log, blk_no, nbblks, bp);
+	return 0;
+}
+
+/*
+ * Read at an offset into the buffer. Returns with the buffer in it's original
+ * state regardless of the result of the read.
+ */
+STATIC int
+xlog_bread_offset(
+	xlog_t		*log,
+	xfs_daddr_t	blk_no,		/* block to read from */
+	int		nbblks,		/* blocks to read */
+	xfs_buf_t	*bp,
+	xfs_caddr_t	offset)
+{
+	xfs_caddr_t	orig_offset = XFS_BUF_PTR(bp);
+	int		orig_len = bp->b_buffer_length;
+	int		error, error2;
+
+	error = XFS_BUF_SET_PTR(bp, offset, BBTOB(nbblks));
+	if (error)
+		return error;
+
+	error = xlog_bread_noalign(log, blk_no, nbblks, bp);
+
+	/* must reset buffer pointer even on error */
+	error2 = XFS_BUF_SET_PTR(bp, orig_offset, orig_len);
+	if (error)
+		return error;
+	return error2;
+}
+
+/*
+ * Write out the buffer at the given block for the given number of blocks.
+ * The buffer is kept locked across the write and is returned locked.
+ * This can only be used for synchronous log writes.
+ */
+STATIC int
+xlog_bwrite(
+	xlog_t		*log,
+	xfs_daddr_t	blk_no,
+	int		nbblks,
+	xfs_buf_t	*bp)
+{
+	int		error;
+
+	if (!xlog_buf_bbcount_valid(log, nbblks)) {
+		xfs_warn(log->l_mp, "Invalid block length (0x%x) for buffer",
+			nbblks);
+		XFS_ERROR_REPORT(__func__, XFS_ERRLEVEL_HIGH, log->l_mp);
+		return EFSCORRUPTED;
+	}
+
+	blk_no = round_down(blk_no, log->l_sectBBsize);
+	nbblks = round_up(nbblks, log->l_sectBBsize);
+
+	ASSERT(nbblks > 0);
+	ASSERT(BBTOB(nbblks) <= XFS_BUF_SIZE(bp));
+
+	XFS_BUF_SET_ADDR(bp, log->l_logBBstart + blk_no);
+	XFS_BUF_ZEROFLAGS(bp);
+	XFS_BUF_BUSY(bp);
+	XFS_BUF_HOLD(bp);
+	XFS_BUF_PSEMA(bp, PRIBIO);
+	XFS_BUF_SET_COUNT(bp, BBTOB(nbblks));
+	XFS_BUF_SET_TARGET(bp, log->l_mp->m_logdev_targp);
+
+	if ((error = xfs_bwrite(log->l_mp, bp)))
+		xfs_ioerror_alert("xlog_bwrite", log->l_mp,
+				  bp, XFS_BUF_ADDR(bp));
+	return error;
+}
+
+#ifdef DEBUG
+/*
+ * dump debug superblock and log record information
+ */
+STATIC void
+xlog_header_check_dump(
+	xfs_mount_t		*mp,
+	xlog_rec_header_t	*head)
+{
+	xfs_debug(mp, "%s:  SB : uuid = %pU, fmt = %d\n",
+		__func__, &mp->m_sb.sb_uuid, XLOG_FMT);
+	xfs_debug(mp, "    log : uuid = %pU, fmt = %d\n",
+		&head->h_fs_uuid, be32_to_cpu(head->h_fmt));
+}
+#else
+#define xlog_header_check_dump(mp, head)
+#endif
+
+/*
+ * check log record header for recovery
+ */
+STATIC int
+xlog_header_check_recover(
+	xfs_mount_t		*mp,
+	xlog_rec_header_t	*head)
+{
+	ASSERT(be32_to_cpu(head->h_magicno) == XLOG_HEADER_MAGIC_NUM);
+
+	/*
+	 * IRIX doesn't write the h_fmt field and leaves it zeroed
+	 * (XLOG_FMT_UNKNOWN). This stops us from trying to recover
+	 * a dirty log created in IRIX.
+	 */
+	if (unlikely(be32_to_cpu(head->h_fmt) != XLOG_FMT)) {
+		xfs_warn(mp,
+	"dirty log written in incompatible format - can't recover");
+		xlog_header_check_dump(mp, head);
+		XFS_ERROR_REPORT("xlog_header_check_recover(1)",
+				 XFS_ERRLEVEL_HIGH, mp);
+		return XFS_ERROR(EFSCORRUPTED);
+	} else if (unlikely(!uuid_equal(&mp->m_sb.sb_uuid, &head->h_fs_uuid))) {
+		xfs_warn(mp,
+	"dirty log entry has mismatched uuid - can't recover");
+		xlog_header_check_dump(mp, head);
+		XFS_ERROR_REPORT("xlog_header_check_recover(2)",
+				 XFS_ERRLEVEL_HIGH, mp);
+		return XFS_ERROR(EFSCORRUPTED);
+	}
+	return 0;
+}
+
+/*
+ * read the head block of the log and check the header
+ */
+STATIC int
+xlog_header_check_mount(
+	xfs_mount_t		*mp,
+	xlog_rec_header_t	*head)
+{
+	ASSERT(be32_to_cpu(head->h_magicno) == XLOG_HEADER_MAGIC_NUM);
+
+	if (uuid_is_nil(&head->h_fs_uuid)) {
+		/*
+		 * IRIX doesn't write the h_fs_uuid or h_fmt fields. If
+		 * h_fs_uuid is nil, we assume this log was last mounted
+		 * by IRIX and continue.
+		 */
+		xfs_warn(mp, "nil uuid in log - IRIX style log");
+	} else if (unlikely(!uuid_equal(&mp->m_sb.sb_uuid, &head->h_fs_uuid))) {
+		xfs_warn(mp, "log has mismatched uuid - can't recover");
+		xlog_header_check_dump(mp, head);
+		XFS_ERROR_REPORT("xlog_header_check_mount",
+				 XFS_ERRLEVEL_HIGH, mp);
+		return XFS_ERROR(EFSCORRUPTED);
+	}
+	return 0;
+}
+
+STATIC void
+xlog_recover_iodone(
+	struct xfs_buf	*bp)
+{
+	if (XFS_BUF_GETERROR(bp)) {
+		/*
+		 * We're not going to bother about retrying
+		 * this during recovery. One strike!
+		 */
+		xfs_ioerror_alert("xlog_recover_iodone",
+					bp->b_target->bt_mount, bp,
+					XFS_BUF_ADDR(bp));
+		xfs_force_shutdown(bp->b_target->bt_mount,
+					SHUTDOWN_META_IO_ERROR);
+	}
+	XFS_BUF_CLR_IODONE_FUNC(bp);
+	xfs_buf_ioend(bp, 0);
+}
+
+/*
+ * This routine finds (to an approximation) the first block in the physical
+ * log which contains the given cycle.  It uses a binary search algorithm.
+ * Note that the algorithm can not be perfect because the disk will not
+ * necessarily be perfect.
+ */
+STATIC int
+xlog_find_cycle_start(
+	xlog_t		*log,
+	xfs_buf_t	*bp,
+	xfs_daddr_t	first_blk,
+	xfs_daddr_t	*last_blk,
+	uint		cycle)
+{
+	xfs_caddr_t	offset;
+	xfs_daddr_t	mid_blk;
+	xfs_daddr_t	end_blk;
+	uint		mid_cycle;
+	int		error;
+
+	end_blk = *last_blk;
+	mid_blk = BLK_AVG(first_blk, end_blk);
+	while (mid_blk != first_blk && mid_blk != end_blk) {
+		error = xlog_bread(log, mid_blk, 1, bp, &offset);
+		if (error)
+			return error;
+		mid_cycle = xlog_get_cycle(offset);
+		if (mid_cycle == cycle)
+			end_blk = mid_blk;   /* last_half_cycle == mid_cycle */
+		else
+			first_blk = mid_blk; /* first_half_cycle == mid_cycle */
+		mid_blk = BLK_AVG(first_blk, end_blk);
+	}
+	ASSERT((mid_blk == first_blk && mid_blk+1 == end_blk) ||
+	       (mid_blk == end_blk && mid_blk-1 == first_blk));
+
+	*last_blk = end_blk;
+
+	return 0;
+}
+
+/*
+ * Check that a range of blocks does not contain stop_on_cycle_no.
+ * Fill in *new_blk with the block offset where such a block is
+ * found, or with -1 (an invalid block number) if there is no such
+ * block in the range.  The scan needs to occur from front to back
+ * and the pointer into the region must be updated since a later
+ * routine will need to perform another test.
+ */
+STATIC int
+xlog_find_verify_cycle(
+	xlog_t		*log,
+	xfs_daddr_t	start_blk,
+	int		nbblks,
+	uint		stop_on_cycle_no,
+	xfs_daddr_t	*new_blk)
+{
+	xfs_daddr_t	i, j;
+	uint		cycle;
+	xfs_buf_t	*bp;
+	xfs_daddr_t	bufblks;
+	xfs_caddr_t	buf = NULL;
+	int		error = 0;
+
+	/*
+	 * Greedily allocate a buffer big enough to handle the full
+	 * range of basic blocks we'll be examining.  If that fails,
+	 * try a smaller size.  We need to be able to read at least
+	 * a log sector, or we're out of luck.
+	 */
+	bufblks = 1 << ffs(nbblks);
+	while (!(bp = xlog_get_bp(log, bufblks))) {
+		bufblks >>= 1;
+		if (bufblks < log->l_sectBBsize)
+			return ENOMEM;
+	}
+
+	for (i = start_blk; i < start_blk + nbblks; i += bufblks) {
+		int	bcount;
+
+		bcount = min(bufblks, (start_blk + nbblks - i));
+
+		error = xlog_bread(log, i, bcount, bp, &buf);
+		if (error)
+			goto out;
+
+		for (j = 0; j < bcount; j++) {
+			cycle = xlog_get_cycle(buf);
+			if (cycle == stop_on_cycle_no) {
+				*new_blk = i+j;
+				goto out;
+			}
+
+			buf += BBSIZE;
+		}
+	}
+
+	*new_blk = -1;
+
+out:
+	xlog_put_bp(bp);
+	return error;
+}
+
+/*
+ * Potentially backup over partial log record write.
+ *
+ * In the typical case, last_blk is the number of the block directly after
+ * a good log record.  Therefore, we subtract one to get the block number
+ * of the last block in the given buffer.  extra_bblks contains the number
+ * of blocks we would have read on a previous read.  This happens when the
+ * last log record is split over the end of the physical log.
+ *
+ * extra_bblks is the number of blocks potentially verified on a previous
+ * call to this routine.
+ */
+STATIC int
+xlog_find_verify_log_record(
+	xlog_t			*log,
+	xfs_daddr_t		start_blk,
+	xfs_daddr_t		*last_blk,
+	int			extra_bblks)
+{
+	xfs_daddr_t		i;
+	xfs_buf_t		*bp;
+	xfs_caddr_t		offset = NULL;
+	xlog_rec_header_t	*head = NULL;
+	int			error = 0;
+	int			smallmem = 0;
+	int			num_blks = *last_blk - start_blk;
+	int			xhdrs;
+
+	ASSERT(start_blk != 0 || *last_blk != start_blk);
+
+	if (!(bp = xlog_get_bp(log, num_blks))) {
+		if (!(bp = xlog_get_bp(log, 1)))
+			return ENOMEM;
+		smallmem = 1;
+	} else {
+		error = xlog_bread(log, start_blk, num_blks, bp, &offset);
+		if (error)
+			goto out;
+		offset += ((num_blks - 1) << BBSHIFT);
+	}
+
+	for (i = (*last_blk) - 1; i >= 0; i--) {
+		if (i < start_blk) {
+			/* valid log record not found */
+			xfs_warn(log->l_mp,
+		"Log inconsistent (didn't find previous header)");
+			ASSERT(0);
+			error = XFS_ERROR(EIO);
+			goto out;
+		}
+
+		if (smallmem) {
+			error = xlog_bread(log, i, 1, bp, &offset);
+			if (error)
+				goto out;
+		}
+
+		head = (xlog_rec_header_t *)offset;
+
+		if (XLOG_HEADER_MAGIC_NUM == be32_to_cpu(head->h_magicno))
+			break;
+
+		if (!smallmem)
+			offset -= BBSIZE;
+	}
+
+	/*
+	 * We hit the beginning of the physical log & still no header.  Return
+	 * to caller.  If caller can handle a return of -1, then this routine
+	 * will be called again for the end of the physical log.
+	 */
+	if (i == -1) {
+		error = -1;
+		goto out;
+	}
+
+	/*
+	 * We have the final block of the good log (the first block
+	 * of the log record _before_ the head. So we check the uuid.
+	 */
+	if ((error = xlog_header_check_mount(log->l_mp, head)))
+		goto out;
+
+	/*
+	 * We may have found a log record header before we expected one.
+	 * last_blk will be the 1st block # with a given cycle #.  We may end
+	 * up reading an entire log record.  In this case, we don't want to
+	 * reset last_blk.  Only when last_blk points in the middle of a log
+	 * record do we update last_blk.
+	 */
+	if (xfs_sb_version_haslogv2(&log->l_mp->m_sb)) {
+		uint	h_size = be32_to_cpu(head->h_size);
+
+		xhdrs = h_size / XLOG_HEADER_CYCLE_SIZE;
+		if (h_size % XLOG_HEADER_CYCLE_SIZE)
+			xhdrs++;
+	} else {
+		xhdrs = 1;
+	}
+
+	if (*last_blk - i + extra_bblks !=
+	    BTOBB(be32_to_cpu(head->h_len)) + xhdrs)
+		*last_blk = i;
+
+out:
+	xlog_put_bp(bp);
+	return error;
+}
+
+/*
+ * Head is defined to be the point of the log where the next log write
+ * write could go.  This means that incomplete LR writes at the end are
+ * eliminated when calculating the head.  We aren't guaranteed that previous
+ * LR have complete transactions.  We only know that a cycle number of
+ * current cycle number -1 won't be present in the log if we start writing
+ * from our current block number.
+ *
+ * last_blk contains the block number of the first block with a given
+ * cycle number.
+ *
+ * Return: zero if normal, non-zero if error.
+ */
+STATIC int
+xlog_find_head(
+	xlog_t 		*log,
+	xfs_daddr_t	*return_head_blk)
+{
+	xfs_buf_t	*bp;
+	xfs_caddr_t	offset;
+	xfs_daddr_t	new_blk, first_blk, start_blk, last_blk, head_blk;
+	int		num_scan_bblks;
+	uint		first_half_cycle, last_half_cycle;
+	uint		stop_on_cycle;
+	int		error, log_bbnum = log->l_logBBsize;
+
+	/* Is the end of the log device zeroed? */
+	if ((error = xlog_find_zeroed(log, &first_blk)) == -1) {
+		*return_head_blk = first_blk;
+
+		/* Is the whole lot zeroed? */
+		if (!first_blk) {
+			/* Linux XFS shouldn't generate totally zeroed logs -
+			 * mkfs etc write a dummy unmount record to a fresh
+			 * log so we can store the uuid in there
+			 */
+			xfs_warn(log->l_mp, "totally zeroed log");
+		}
+
+		return 0;
+	} else if (error) {
+		xfs_warn(log->l_mp, "empty log check failed");
+		return error;
+	}
+
+	first_blk = 0;			/* get cycle # of 1st block */
+	bp = xlog_get_bp(log, 1);
+	if (!bp)
+		return ENOMEM;
+
+	error = xlog_bread(log, 0, 1, bp, &offset);
+	if (error)
+		goto bp_err;
+
+	first_half_cycle = xlog_get_cycle(offset);
+
+	last_blk = head_blk = log_bbnum - 1;	/* get cycle # of last block */
+	error = xlog_bread(log, last_blk, 1, bp, &offset);
+	if (error)
+		goto bp_err;
+
+	last_half_cycle = xlog_get_cycle(offset);
+	ASSERT(last_half_cycle != 0);
+
+	/*
+	 * If the 1st half cycle number is equal to the last half cycle number,
+	 * then the entire log is stamped with the same cycle number.  In this
+	 * case, head_blk can't be set to zero (which makes sense).  The below
+	 * math doesn't work out properly with head_blk equal to zero.  Instead,
+	 * we set it to log_bbnum which is an invalid block number, but this
+	 * value makes the math correct.  If head_blk doesn't changed through
+	 * all the tests below, *head_blk is set to zero at the very end rather
+	 * than log_bbnum.  In a sense, log_bbnum and zero are the same block
+	 * in a circular file.
+	 */
+	if (first_half_cycle == last_half_cycle) {
+		/*
+		 * In this case we believe that the entire log should have
+		 * cycle number last_half_cycle.  We need to scan backwards
+		 * from the end verifying that there are no holes still
+		 * containing last_half_cycle - 1.  If we find such a hole,
+		 * then the start of that hole will be the new head.  The
+		 * simple case looks like
+		 *        x | x ... | x - 1 | x
+		 * Another case that fits this picture would be
+		 *        x | x + 1 | x ... | x
+		 * In this case the head really is somewhere at the end of the
+		 * log, as one of the latest writes at the beginning was
+		 * incomplete.
+		 * One more case is
+		 *        x | x + 1 | x ... | x - 1 | x
+		 * This is really the combination of the above two cases, and
+		 * the head has to end up at the start of the x-1 hole at the
+		 * end of the log.
+		 *
+		 * In the 256k log case, we will read from the beginning to the
+		 * end of the log and search for cycle numbers equal to x-1.
+		 * We don't worry about the x+1 blocks that we encounter,
+		 * because we know that they cannot be the head since the log
+		 * started with x.
+		 */
+		head_blk = log_bbnum;
+		stop_on_cycle = last_half_cycle - 1;
+	} else {
+		/*
+		 * In this case we want to find the first block with cycle
+		 * number matching last_half_cycle.  We expect the log to be
+		 * some variation on
+		 *        x + 1 ... | x ... | x
+		 * The first block with cycle number x (last_half_cycle) will
+		 * be where the new head belongs.  First we do a binary search
+		 * for the first occurrence of last_half_cycle.  The binary
+		 * search may not be totally accurate, so then we scan back
+		 * from there looking for occurrences of last_half_cycle before
+		 * us.  If that backwards scan wraps around the beginning of
+		 * the log, then we look for occurrences of last_half_cycle - 1
+		 * at the end of the log.  The cases we're looking for look
+		 * like
+		 *                               v binary search stopped here
+		 *        x + 1 ... | x | x + 1 | x ... | x
+		 *                   ^ but we want to locate this spot
+		 * or
+		 *        <---------> less than scan distance
+		 *        x + 1 ... | x ... | x - 1 | x
+		 *                           ^ we want to locate this spot
+		 */
+		stop_on_cycle = last_half_cycle;
+		if ((error = xlog_find_cycle_start(log, bp, first_blk,
+						&head_blk, last_half_cycle)))
+			goto bp_err;
+	}
+
+	/*
+	 * Now validate the answer.  Scan back some number of maximum possible
+	 * blocks and make sure each one has the expected cycle number.  The
+	 * maximum is determined by the total possible amount of buffering
+	 * in the in-core log.  The following number can be made tighter if
+	 * we actually look at the block size of the filesystem.
+	 */
+	num_scan_bblks = XLOG_TOTAL_REC_SHIFT(log);
+	if (head_blk >= num_scan_bblks) {
+		/*
+		 * We are guaranteed that the entire check can be performed
+		 * in one buffer.
+		 */
+		start_blk = head_blk - num_scan_bblks;
+		if ((error = xlog_find_verify_cycle(log,
+						start_blk, num_scan_bblks,
+						stop_on_cycle, &new_blk)))
+			goto bp_err;
+		if (new_blk != -1)
+			head_blk = new_blk;
+	} else {		/* need to read 2 parts of log */
+		/*
+		 * We are going to scan backwards in the log in two parts.
+		 * First we scan the physical end of the log.  In this part
+		 * of the log, we are looking for blocks with cycle number
+		 * last_half_cycle - 1.
+		 * If we find one, then we know that the log starts there, as
+		 * we've found a hole that didn't get written in going around
+		 * the end of the physical log.  The simple case for this is
+		 *        x + 1 ... | x ... | x - 1 | x
+		 *        <---------> less than scan distance
+		 * If all of the blocks at the end of the log have cycle number
+		 * last_half_cycle, then we check the blocks at the start of
+		 * the log looking for occurrences of last_half_cycle.  If we
+		 * find one, then our current estimate for the location of the
+		 * first occurrence of last_half_cycle is wrong and we move
+		 * back to the hole we've found.  This case looks like
+		 *        x + 1 ... | x | x + 1 | x ...
+		 *                               ^ binary search stopped here
+		 * Another case we need to handle that only occurs in 256k
+		 * logs is
+		 *        x + 1 ... | x ... | x+1 | x ...
+		 *                   ^ binary search stops here
+		 * In a 256k log, the scan at the end of the log will see the
+		 * x + 1 blocks.  We need to skip past those since that is
+		 * certainly not the head of the log.  By searching for
+		 * last_half_cycle-1 we accomplish that.
+		 */
+		ASSERT(head_blk <= INT_MAX &&
+			(xfs_daddr_t) num_scan_bblks >= head_blk);
+		start_blk = log_bbnum - (num_scan_bblks - head_blk);
+		if ((error = xlog_find_verify_cycle(log, start_blk,
+					num_scan_bblks - (int)head_blk,
+					(stop_on_cycle - 1), &new_blk)))
+			goto bp_err;
+		if (new_blk != -1) {
+			head_blk = new_blk;
+			goto validate_head;
+		}
+
+		/*
+		 * Scan beginning of log now.  The last part of the physical
+		 * log is good.  This scan needs to verify that it doesn't find
+		 * the last_half_cycle.
+		 */
+		start_blk = 0;
+		ASSERT(head_blk <= INT_MAX);
+		if ((error = xlog_find_verify_cycle(log,
+					start_blk, (int)head_blk,
+					stop_on_cycle, &new_blk)))
+			goto bp_err;
+		if (new_blk != -1)
+			head_blk = new_blk;
+	}
+
+validate_head:
+	/*
+	 * Now we need to make sure head_blk is not pointing to a block in
+	 * the middle of a log record.
+	 */
+	num_scan_bblks = XLOG_REC_SHIFT(log);
+	if (head_blk >= num_scan_bblks) {
+		start_blk = head_blk - num_scan_bblks; /* don't read head_blk */
+
+		/* start ptr at last block ptr before head_blk */
+		if ((error = xlog_find_verify_log_record(log, start_blk,
+							&head_blk, 0)) == -1) {
+			error = XFS_ERROR(EIO);
+			goto bp_err;
+		} else if (error)
+			goto bp_err;
+	} else {
+		start_blk = 0;
+		ASSERT(head_blk <= INT_MAX);
+		if ((error = xlog_find_verify_log_record(log, start_blk,
+							&head_blk, 0)) == -1) {
+			/* We hit the beginning of the log during our search */
+			start_blk = log_bbnum - (num_scan_bblks - head_blk);
+			new_blk = log_bbnum;
+			ASSERT(start_blk <= INT_MAX &&
+				(xfs_daddr_t) log_bbnum-start_blk >= 0);
+			ASSERT(head_blk <= INT_MAX);
+			if ((error = xlog_find_verify_log_record(log,
+							start_blk, &new_blk,
+							(int)head_blk)) == -1) {
+				error = XFS_ERROR(EIO);
+				goto bp_err;
+			} else if (error)
+				goto bp_err;
+			if (new_blk != log_bbnum)
+				head_blk = new_blk;
+		} else if (error)
+			goto bp_err;
+	}
+
+	xlog_put_bp(bp);
+	if (head_blk == log_bbnum)
+		*return_head_blk = 0;
+	else
+		*return_head_blk = head_blk;
+	/*
+	 * When returning here, we have a good block number.  Bad block
+	 * means that during a previous crash, we didn't have a clean break
+	 * from cycle number N to cycle number N-1.  In this case, we need
+	 * to find the first block with cycle number N-1.
+	 */
+	return 0;
+
+ bp_err:
+	xlog_put_bp(bp);
+
+	if (error)
+		xfs_warn(log->l_mp, "failed to find log head");
+	return error;
+}
+
+/*
+ * Find the sync block number or the tail of the log.
+ *
+ * This will be the block number of the last record to have its
+ * associated buffers synced to disk.  Every log record header has
+ * a sync lsn embedded in it.  LSNs hold block numbers, so it is easy
+ * to get a sync block number.  The only concern is to figure out which
+ * log record header to believe.
+ *
+ * The following algorithm uses the log record header with the largest
+ * lsn.  The entire log record does not need to be valid.  We only care
+ * that the header is valid.
+ *
+ * We could speed up search by using current head_blk buffer, but it is not
+ * available.
+ */
+STATIC int
+xlog_find_tail(
+	xlog_t			*log,
+	xfs_daddr_t		*head_blk,
+	xfs_daddr_t		*tail_blk)
+{
+	xlog_rec_header_t	*rhead;
+	xlog_op_header_t	*op_head;
+	xfs_caddr_t		offset = NULL;
+	xfs_buf_t		*bp;
+	int			error, i, found;
+	xfs_daddr_t		umount_data_blk;
+	xfs_daddr_t		after_umount_blk;
+	xfs_lsn_t		tail_lsn;
+	int			hblks;
+
+	found = 0;
+
+	/*
+	 * Find previous log record
+	 */
+	if ((error = xlog_find_head(log, head_blk)))
+		return error;
+
+	bp = xlog_get_bp(log, 1);
+	if (!bp)
+		return ENOMEM;
+	if (*head_blk == 0) {				/* special case */
+		error = xlog_bread(log, 0, 1, bp, &offset);
+		if (error)
+			goto done;
+
+		if (xlog_get_cycle(offset) == 0) {
+			*tail_blk = 0;
+			/* leave all other log inited values alone */
+			goto done;
+		}
+	}
+
+	/*
+	 * Search backwards looking for log record header block
+	 */
+	ASSERT(*head_blk < INT_MAX);
+	for (i = (int)(*head_blk) - 1; i >= 0; i--) {
+		error = xlog_bread(log, i, 1, bp, &offset);
+		if (error)
+			goto done;
+
+		if (XLOG_HEADER_MAGIC_NUM == be32_to_cpu(*(__be32 *)offset)) {
+			found = 1;
+			break;
+		}
+	}
+	/*
+	 * If we haven't found the log record header block, start looking
+	 * again from the end of the physical log.  XXXmiken: There should be
+	 * a check here to make sure we didn't search more than N blocks in
+	 * the previous code.
+	 */
+	if (!found) {
+		for (i = log->l_logBBsize - 1; i >= (int)(*head_blk); i--) {
+			error = xlog_bread(log, i, 1, bp, &offset);
+			if (error)
+				goto done;
+
+			if (XLOG_HEADER_MAGIC_NUM ==
+			    be32_to_cpu(*(__be32 *)offset)) {
+				found = 2;
+				break;
+			}
+		}
+	}
+	if (!found) {
+		xfs_warn(log->l_mp, "%s: couldn't find sync record", __func__);
+		ASSERT(0);
+		return XFS_ERROR(EIO);
+	}
+
+	/* find blk_no of tail of log */
+	rhead = (xlog_rec_header_t *)offset;
+	*tail_blk = BLOCK_LSN(be64_to_cpu(rhead->h_tail_lsn));
+
+	/*
+	 * Reset log values according to the state of the log when we
+	 * crashed.  In the case where head_blk == 0, we bump curr_cycle
+	 * one because the next write starts a new cycle rather than
+	 * continuing the cycle of the last good log record.  At this
+	 * point we have guaranteed that all partial log records have been
+	 * accounted for.  Therefore, we know that the last good log record
+	 * written was complete and ended exactly on the end boundary
+	 * of the physical log.
+	 */
+	log->l_prev_block = i;
+	log->l_curr_block = (int)*head_blk;
+	log->l_curr_cycle = be32_to_cpu(rhead->h_cycle);
+	if (found == 2)
+		log->l_curr_cycle++;
+	atomic64_set(&log->l_tail_lsn, be64_to_cpu(rhead->h_tail_lsn));
+	atomic64_set(&log->l_last_sync_lsn, be64_to_cpu(rhead->h_lsn));
+	xlog_assign_grant_head(&log->l_grant_reserve_head, log->l_curr_cycle,
+					BBTOB(log->l_curr_block));
+	xlog_assign_grant_head(&log->l_grant_write_head, log->l_curr_cycle,
+					BBTOB(log->l_curr_block));
+
+	/*
+	 * Look for unmount record.  If we find it, then we know there
+	 * was a clean unmount.  Since 'i' could be the last block in
+	 * the physical log, we convert to a log block before comparing
+	 * to the head_blk.
+	 *
+	 * Save the current tail lsn to use to pass to
+	 * xlog_clear_stale_blocks() below.  We won't want to clear the
+	 * unmount record if there is one, so we pass the lsn of the
+	 * unmount record rather than the block after it.
+	 */
+	if (xfs_sb_version_haslogv2(&log->l_mp->m_sb)) {
+		int	h_size = be32_to_cpu(rhead->h_size);
+		int	h_version = be32_to_cpu(rhead->h_version);
+
+		if ((h_version & XLOG_VERSION_2) &&
+		    (h_size > XLOG_HEADER_CYCLE_SIZE)) {
+			hblks = h_size / XLOG_HEADER_CYCLE_SIZE;
+			if (h_size % XLOG_HEADER_CYCLE_SIZE)
+				hblks++;
+		} else {
+			hblks = 1;
+		}
+	} else {
+		hblks = 1;
+	}
+	after_umount_blk = (i + hblks + (int)
+		BTOBB(be32_to_cpu(rhead->h_len))) % log->l_logBBsize;
+	tail_lsn = atomic64_read(&log->l_tail_lsn);
+	if (*head_blk == after_umount_blk &&
+	    be32_to_cpu(rhead->h_num_logops) == 1) {
+		umount_data_blk = (i + hblks) % log->l_logBBsize;
+		error = xlog_bread(log, umount_data_blk, 1, bp, &offset);
+		if (error)
+			goto done;
+
+		op_head = (xlog_op_header_t *)offset;
+		if (op_head->oh_flags & XLOG_UNMOUNT_TRANS) {
+			/*
+			 * Set tail and last sync so that newly written
+			 * log records will point recovery to after the
+			 * current unmount record.
+			 */
+			xlog_assign_atomic_lsn(&log->l_tail_lsn,
+					log->l_curr_cycle, after_umount_blk);
+			xlog_assign_atomic_lsn(&log->l_last_sync_lsn,
+					log->l_curr_cycle, after_umount_blk);
+			*tail_blk = after_umount_blk;
+
+			/*
+			 * Note that the unmount was clean. If the unmount
+			 * was not clean, we need to know this to rebuild the
+			 * superblock counters from the perag headers if we
+			 * have a filesystem using non-persistent counters.
+			 */
+			log->l_mp->m_flags |= XFS_MOUNT_WAS_CLEAN;
+		}
+	}
+
+	/*
+	 * Make sure that there are no blocks in front of the head
+	 * with the same cycle number as the head.  This can happen
+	 * because we allow multiple outstanding log writes concurrently,
+	 * and the later writes might make it out before earlier ones.
+	 *
+	 * We use the lsn from before modifying it so that we'll never
+	 * overwrite the unmount record after a clean unmount.
+	 *
+	 * Do this only if we are going to recover the filesystem
+	 *
+	 * NOTE: This used to say "if (!readonly)"
+	 * However on Linux, we can & do recover a read-only filesystem.
+	 * We only skip recovery if NORECOVERY is specified on mount,
+	 * in which case we would not be here.
+	 *
+	 * But... if the -device- itself is readonly, just skip this.
+	 * We can't recover this device anyway, so it won't matter.
+	 */
+	if (!xfs_readonly_buftarg(log->l_mp->m_logdev_targp))
+		error = xlog_clear_stale_blocks(log, tail_lsn);
+
+done:
+	xlog_put_bp(bp);
+
+	if (error)
+		xfs_warn(log->l_mp, "failed to locate log tail");
+	return error;
+}
+
+/*
+ * Is the log zeroed at all?
+ *
+ * The last binary search should be changed to perform an X block read
+ * once X becomes small enough.  You can then search linearly through
+ * the X blocks.  This will cut down on the number of reads we need to do.
+ *
+ * If the log is partially zeroed, this routine will pass back the blkno
+ * of the first block with cycle number 0.  It won't have a complete LR
+ * preceding it.
+ *
+ * Return:
+ *	0  => the log is completely written to
+ *	-1 => use *blk_no as the first block of the log
+ *	>0 => error has occurred
+ */
+STATIC int
+xlog_find_zeroed(
+	xlog_t		*log,
+	xfs_daddr_t	*blk_no)
+{
+	xfs_buf_t	*bp;
+	xfs_caddr_t	offset;
+	uint	        first_cycle, last_cycle;
+	xfs_daddr_t	new_blk, last_blk, start_blk;
+	xfs_daddr_t     num_scan_bblks;
+	int	        error, log_bbnum = log->l_logBBsize;
+
+	*blk_no = 0;
+
+	/* check totally zeroed log */
+	bp = xlog_get_bp(log, 1);
+	if (!bp)
+		return ENOMEM;
+	error = xlog_bread(log, 0, 1, bp, &offset);
+	if (error)
+		goto bp_err;
+
+	first_cycle = xlog_get_cycle(offset);
+	if (first_cycle == 0) {		/* completely zeroed log */
+		*blk_no = 0;
+		xlog_put_bp(bp);
+		return -1;
+	}
+
+	/* check partially zeroed log */
+	error = xlog_bread(log, log_bbnum-1, 1, bp, &offset);
+	if (error)
+		goto bp_err;
+
+	last_cycle = xlog_get_cycle(offset);
+	if (last_cycle != 0) {		/* log completely written to */
+		xlog_put_bp(bp);
+		return 0;
+	} else if (first_cycle != 1) {
+		/*
+		 * If the cycle of the last block is zero, the cycle of
+		 * the first block must be 1. If it's not, maybe we're
+		 * not looking at a log... Bail out.
+		 */
+		xfs_warn(log->l_mp,
+			"Log inconsistent or not a log (last==0, first!=1)");
+		return XFS_ERROR(EINVAL);
+	}
+
+	/* we have a partially zeroed log */
+	last_blk = log_bbnum-1;
+	if ((error = xlog_find_cycle_start(log, bp, 0, &last_blk, 0)))
+		goto bp_err;
+
+	/*
+	 * Validate the answer.  Because there is no way to guarantee that
+	 * the entire log is made up of log records which are the same size,
+	 * we scan over the defined maximum blocks.  At this point, the maximum
+	 * is not chosen to mean anything special.   XXXmiken
+	 */
+	num_scan_bblks = XLOG_TOTAL_REC_SHIFT(log);
+	ASSERT(num_scan_bblks <= INT_MAX);
+
+	if (last_blk < num_scan_bblks)
+		num_scan_bblks = last_blk;
+	start_blk = last_blk - num_scan_bblks;
+
+	/*
+	 * We search for any instances of cycle number 0 that occur before
+	 * our current estimate of the head.  What we're trying to detect is
+	 *        1 ... | 0 | 1 | 0...
+	 *                       ^ binary search ends here
+	 */
+	if ((error = xlog_find_verify_cycle(log, start_blk,
+					 (int)num_scan_bblks, 0, &new_blk)))
+		goto bp_err;
+	if (new_blk != -1)
+		last_blk = new_blk;
+
+	/*
+	 * Potentially backup over partial log record write.  We don't need
+	 * to search the end of the log because we know it is zero.
+	 */
+	if ((error = xlog_find_verify_log_record(log, start_blk,
+				&last_blk, 0)) == -1) {
+	    error = XFS_ERROR(EIO);
+	    goto bp_err;
+	} else if (error)
+	    goto bp_err;
+
+	*blk_no = last_blk;
+bp_err:
+	xlog_put_bp(bp);
+	if (error)
+		return error;
+	return -1;
+}
+
+/*
+ * These are simple subroutines used by xlog_clear_stale_blocks() below
+ * to initialize a buffer full of empty log record headers and write
+ * them into the log.
+ */
+STATIC void
+xlog_add_record(
+	xlog_t			*log,
+	xfs_caddr_t		buf,
+	int			cycle,
+	int			block,
+	int			tail_cycle,
+	int			tail_block)
+{
+	xlog_rec_header_t	*recp = (xlog_rec_header_t *)buf;
+
+	memset(buf, 0, BBSIZE);
+	recp->h_magicno = cpu_to_be32(XLOG_HEADER_MAGIC_NUM);
+	recp->h_cycle = cpu_to_be32(cycle);
+	recp->h_version = cpu_to_be32(
+			xfs_sb_version_haslogv2(&log->l_mp->m_sb) ? 2 : 1);
+	recp->h_lsn = cpu_to_be64(xlog_assign_lsn(cycle, block));
+	recp->h_tail_lsn = cpu_to_be64(xlog_assign_lsn(tail_cycle, tail_block));
+	recp->h_fmt = cpu_to_be32(XLOG_FMT);
+	memcpy(&recp->h_fs_uuid, &log->l_mp->m_sb.sb_uuid, sizeof(uuid_t));
+}
+
+STATIC int
+xlog_write_log_records(
+	xlog_t		*log,
+	int		cycle,
+	int		start_block,
+	int		blocks,
+	int		tail_cycle,
+	int		tail_block)
+{
+	xfs_caddr_t	offset;
+	xfs_buf_t	*bp;
+	int		balign, ealign;
+	int		sectbb = log->l_sectBBsize;
+	int		end_block = start_block + blocks;
+	int		bufblks;
+	int		error = 0;
+	int		i, j = 0;
+
+	/*
+	 * Greedily allocate a buffer big enough to handle the full
+	 * range of basic blocks to be written.  If that fails, try
+	 * a smaller size.  We need to be able to write at least a
+	 * log sector, or we're out of luck.
+	 */
+	bufblks = 1 << ffs(blocks);
+	while (!(bp = xlog_get_bp(log, bufblks))) {
+		bufblks >>= 1;
+		if (bufblks < sectbb)
+			return ENOMEM;
+	}
+
+	/* We may need to do a read at the start to fill in part of
+	 * the buffer in the starting sector not covered by the first
+	 * write below.
+	 */
+	balign = round_down(start_block, sectbb);
+	if (balign != start_block) {
+		error = xlog_bread_noalign(log, start_block, 1, bp);
+		if (error)
+			goto out_put_bp;
+
+		j = start_block - balign;
+	}
+
+	for (i = start_block; i < end_block; i += bufblks) {
+		int		bcount, endcount;
+
+		bcount = min(bufblks, end_block - start_block);
+		endcount = bcount - j;
+
+		/* We may need to do a read at the end to fill in part of
+		 * the buffer in the final sector not covered by the write.
+		 * If this is the same sector as the above read, skip it.
+		 */
+		ealign = round_down(end_block, sectbb);
+		if (j == 0 && (start_block + endcount > ealign)) {
+			offset = XFS_BUF_PTR(bp) + BBTOB(ealign - start_block);
+			error = xlog_bread_offset(log, ealign, sectbb,
+							bp, offset);
+			if (error)
+				break;
+
+		}
+
+		offset = xlog_align(log, start_block, endcount, bp);
+		for (; j < endcount; j++) {
+			xlog_add_record(log, offset, cycle, i+j,
+					tail_cycle, tail_block);
+			offset += BBSIZE;
+		}
+		error = xlog_bwrite(log, start_block, endcount, bp);
+		if (error)
+			break;
+		start_block += endcount;
+		j = 0;
+	}
+
+ out_put_bp:
+	xlog_put_bp(bp);
+	return error;
+}
+
+/*
+ * This routine is called to blow away any incomplete log writes out
+ * in front of the log head.  We do this so that we won't become confused
+ * if we come up, write only a little bit more, and then crash again.
+ * If we leave the partial log records out there, this situation could
+ * cause us to think those partial writes are valid blocks since they
+ * have the current cycle number.  We get rid of them by overwriting them
+ * with empty log records with the old cycle number rather than the
+ * current one.
+ *
+ * The tail lsn is passed in rather than taken from
+ * the log so that we will not write over the unmount record after a
+ * clean unmount in a 512 block log.  Doing so would leave the log without
+ * any valid log records in it until a new one was written.  If we crashed
+ * during that time we would not be able to recover.
+ */
+STATIC int
+xlog_clear_stale_blocks(
+	xlog_t		*log,
+	xfs_lsn_t	tail_lsn)
+{
+	int		tail_cycle, head_cycle;
+	int		tail_block, head_block;
+	int		tail_distance, max_distance;
+	int		distance;
+	int		error;
+
+	tail_cycle = CYCLE_LSN(tail_lsn);
+	tail_block = BLOCK_LSN(tail_lsn);
+	head_cycle = log->l_curr_cycle;
+	head_block = log->l_curr_block;
+
+	/*
+	 * Figure out the distance between the new head of the log
+	 * and the tail.  We want to write over any blocks beyond the
+	 * head that we may have written just before the crash, but
+	 * we don't want to overwrite the tail of the log.
+	 */
+	if (head_cycle == tail_cycle) {
+		/*
+		 * The tail is behind the head in the physical log,
+		 * so the distance from the head to the tail is the
+		 * distance from the head to the end of the log plus
+		 * the distance from the beginning of the log to the
+		 * tail.
+		 */
+		if (unlikely(head_block < tail_block || head_block >= log->l_logBBsize)) {
+			XFS_ERROR_REPORT("xlog_clear_stale_blocks(1)",
+					 XFS_ERRLEVEL_LOW, log->l_mp);
+			return XFS_ERROR(EFSCORRUPTED);
+		}
+		tail_distance = tail_block + (log->l_logBBsize - head_block);
+	} else {
+		/*
+		 * The head is behind the tail in the physical log,
+		 * so the distance from the head to the tail is just
+		 * the tail block minus the head block.
+		 */
+		if (unlikely(head_block >= tail_block || head_cycle != (tail_cycle + 1))){
+			XFS_ERROR_REPORT("xlog_clear_stale_blocks(2)",
+					 XFS_ERRLEVEL_LOW, log->l_mp);
+			return XFS_ERROR(EFSCORRUPTED);
+		}
+		tail_distance = tail_block - head_block;
+	}
+
+	/*
+	 * If the head is right up against the tail, we can't clear
+	 * anything.
+	 */
+	if (tail_distance <= 0) {
+		ASSERT(tail_distance == 0);
+		return 0;
+	}
+
+	max_distance = XLOG_TOTAL_REC_SHIFT(log);
+	/*
+	 * Take the smaller of the maximum amount of outstanding I/O
+	 * we could have and the distance to the tail to clear out.
+	 * We take the smaller so that we don't overwrite the tail and
+	 * we don't waste all day writing from the head to the tail
+	 * for no reason.
+	 */
+	max_distance = MIN(max_distance, tail_distance);
+
+	if ((head_block + max_distance) <= log->l_logBBsize) {
+		/*
+		 * We can stomp all the blocks we need to without
+		 * wrapping around the end of the log.  Just do it
+		 * in a single write.  Use the cycle number of the
+		 * current cycle minus one so that the log will look like:
+		 *     n ... | n - 1 ...
+		 */
+		error = xlog_write_log_records(log, (head_cycle - 1),
+				head_block, max_distance, tail_cycle,
+				tail_block);
+		if (error)
+			return error;
+	} else {
+		/*
+		 * We need to wrap around the end of the physical log in
+		 * order to clear all the blocks.  Do it in two separate
+		 * I/Os.  The first write should be from the head to the
+		 * end of the physical log, and it should use the current
+		 * cycle number minus one just like above.
+		 */
+		distance = log->l_logBBsize - head_block;
+		error = xlog_write_log_records(log, (head_cycle - 1),
+				head_block, distance, tail_cycle,
+				tail_block);
+
+		if (error)
+			return error;
+
+		/*
+		 * Now write the blocks at the start of the physical log.
+		 * This writes the remainder of the blocks we want to clear.
+		 * It uses the current cycle number since we're now on the
+		 * same cycle as the head so that we get:
+		 *    n ... n ... | n - 1 ...
+		 *    ^^^^^ blocks we're writing
+		 */
+		distance = max_distance - (log->l_logBBsize - head_block);
+		error = xlog_write_log_records(log, head_cycle, 0, distance,
+				tail_cycle, tail_block);
+		if (error)
+			return error;
+	}
+
+	return 0;
+}
+
+/******************************************************************************
+ *
+ *		Log recover routines
+ *
+ ******************************************************************************
+ */
+
+STATIC xlog_recover_t *
+xlog_recover_find_tid(
+	struct hlist_head	*head,
+	xlog_tid_t		tid)
+{
+	xlog_recover_t		*trans;
+	struct hlist_node	*n;
+
+	hlist_for_each_entry(trans, n, head, r_list) {
+		if (trans->r_log_tid == tid)
+			return trans;
+	}
+	return NULL;
+}
+
+STATIC void
+xlog_recover_new_tid(
+	struct hlist_head	*head,
+	xlog_tid_t		tid,
+	xfs_lsn_t		lsn)
+{
+	xlog_recover_t		*trans;
+
+	trans = kmem_zalloc(sizeof(xlog_recover_t), KM_SLEEP);
+	trans->r_log_tid   = tid;
+	trans->r_lsn	   = lsn;
+	INIT_LIST_HEAD(&trans->r_itemq);
+
+	INIT_HLIST_NODE(&trans->r_list);
+	hlist_add_head(&trans->r_list, head);
+}
+
+STATIC void
+xlog_recover_add_item(
+	struct list_head	*head)
+{
+	xlog_recover_item_t	*item;
+
+	item = kmem_zalloc(sizeof(xlog_recover_item_t), KM_SLEEP);
+	INIT_LIST_HEAD(&item->ri_list);
+	list_add_tail(&item->ri_list, head);
+}
+
+STATIC int
+xlog_recover_add_to_cont_trans(
+	struct log		*log,
+	xlog_recover_t		*trans,
+	xfs_caddr_t		dp,
+	int			len)
+{
+	xlog_recover_item_t	*item;
+	xfs_caddr_t		ptr, old_ptr;
+	int			old_len;
+
+	if (list_empty(&trans->r_itemq)) {
+		/* finish copying rest of trans header */
+		xlog_recover_add_item(&trans->r_itemq);
+		ptr = (xfs_caddr_t) &trans->r_theader +
+				sizeof(xfs_trans_header_t) - len;
+		memcpy(ptr, dp, len); /* d, s, l */
+		return 0;
+	}
+	/* take the tail entry */
+	item = list_entry(trans->r_itemq.prev, xlog_recover_item_t, ri_list);
+
+	old_ptr = item->ri_buf[item->ri_cnt-1].i_addr;
+	old_len = item->ri_buf[item->ri_cnt-1].i_len;
+
+	ptr = kmem_realloc(old_ptr, len+old_len, old_len, 0u);
+	memcpy(&ptr[old_len], dp, len); /* d, s, l */
+	item->ri_buf[item->ri_cnt-1].i_len += len;
+	item->ri_buf[item->ri_cnt-1].i_addr = ptr;
+	trace_xfs_log_recover_item_add_cont(log, trans, item, 0);
+	return 0;
+}
+
+/*
+ * The next region to add is the start of a new region.  It could be
+ * a whole region or it could be the first part of a new region.  Because
+ * of this, the assumption here is that the type and size fields of all
+ * format structures fit into the first 32 bits of the structure.
+ *
+ * This works because all regions must be 32 bit aligned.  Therefore, we
+ * either have both fields or we have neither field.  In the case we have
+ * neither field, the data part of the region is zero length.  We only have
+ * a log_op_header and can throw away the header since a new one will appear
+ * later.  If we have at least 4 bytes, then we can determine how many regions
+ * will appear in the current log item.
+ */
+STATIC int
+xlog_recover_add_to_trans(
+	struct log		*log,
+	xlog_recover_t		*trans,
+	xfs_caddr_t		dp,
+	int			len)
+{
+	xfs_inode_log_format_t	*in_f;			/* any will do */
+	xlog_recover_item_t	*item;
+	xfs_caddr_t		ptr;
+
+	if (!len)
+		return 0;
+	if (list_empty(&trans->r_itemq)) {
+		/* we need to catch log corruptions here */
+		if (*(uint *)dp != XFS_TRANS_HEADER_MAGIC) {
+			xfs_warn(log->l_mp, "%s: bad header magic number",
+				__func__);
+			ASSERT(0);
+			return XFS_ERROR(EIO);
+		}
+		if (len == sizeof(xfs_trans_header_t))
+			xlog_recover_add_item(&trans->r_itemq);
+		memcpy(&trans->r_theader, dp, len); /* d, s, l */
+		return 0;
+	}
+
+	ptr = kmem_alloc(len, KM_SLEEP);
+	memcpy(ptr, dp, len);
+	in_f = (xfs_inode_log_format_t *)ptr;
+
+	/* take the tail entry */
+	item = list_entry(trans->r_itemq.prev, xlog_recover_item_t, ri_list);
+	if (item->ri_total != 0 &&
+	     item->ri_total == item->ri_cnt) {
+		/* tail item is in use, get a new one */
+		xlog_recover_add_item(&trans->r_itemq);
+		item = list_entry(trans->r_itemq.prev,
+					xlog_recover_item_t, ri_list);
+	}
+
+	if (item->ri_total == 0) {		/* first region to be added */
+		if (in_f->ilf_size == 0 ||
+		    in_f->ilf_size > XLOG_MAX_REGIONS_IN_ITEM) {
+			xfs_warn(log->l_mp,
+		"bad number of regions (%d) in inode log format",
+				  in_f->ilf_size);
+			ASSERT(0);
+			return XFS_ERROR(EIO);
+		}
+
+		item->ri_total = in_f->ilf_size;
+		item->ri_buf =
+			kmem_zalloc(item->ri_total * sizeof(xfs_log_iovec_t),
+				    KM_SLEEP);
+	}
+	ASSERT(item->ri_total > item->ri_cnt);
+	/* Description region is ri_buf[0] */
+	item->ri_buf[item->ri_cnt].i_addr = ptr;
+	item->ri_buf[item->ri_cnt].i_len  = len;
+	item->ri_cnt++;
+	trace_xfs_log_recover_item_add(log, trans, item, 0);
+	return 0;
+}
+
+/*
+ * Sort the log items in the transaction. Cancelled buffers need
+ * to be put first so they are processed before any items that might
+ * modify the buffers. If they are cancelled, then the modifications
+ * don't need to be replayed.
+ */
+STATIC int
+xlog_recover_reorder_trans(
+	struct log		*log,
+	xlog_recover_t		*trans,
+	int			pass)
+{
+	xlog_recover_item_t	*item, *n;
+	LIST_HEAD(sort_list);
+
+	list_splice_init(&trans->r_itemq, &sort_list);
+	list_for_each_entry_safe(item, n, &sort_list, ri_list) {
+		xfs_buf_log_format_t	*buf_f = item->ri_buf[0].i_addr;
+
+		switch (ITEM_TYPE(item)) {
+		case XFS_LI_BUF:
+			if (!(buf_f->blf_flags & XFS_BLF_CANCEL)) {
+				trace_xfs_log_recover_item_reorder_head(log,
+							trans, item, pass);
+				list_move(&item->ri_list, &trans->r_itemq);
+				break;
+			}
+		case XFS_LI_INODE:
+		case XFS_LI_DQUOT:
+		case XFS_LI_QUOTAOFF:
+		case XFS_LI_EFD:
+		case XFS_LI_EFI:
+			trace_xfs_log_recover_item_reorder_tail(log,
+							trans, item, pass);
+			list_move_tail(&item->ri_list, &trans->r_itemq);
+			break;
+		default:
+			xfs_warn(log->l_mp,
+				"%s: unrecognized type of log operation",
+				__func__);
+			ASSERT(0);
+			return XFS_ERROR(EIO);
+		}
+	}
+	ASSERT(list_empty(&sort_list));
+	return 0;
+}
+
+/*
+ * Build up the table of buf cancel records so that we don't replay
+ * cancelled data in the second pass.  For buffer records that are
+ * not cancel records, there is nothing to do here so we just return.
+ *
+ * If we get a cancel record which is already in the table, this indicates
+ * that the buffer was cancelled multiple times.  In order to ensure
+ * that during pass 2 we keep the record in the table until we reach its
+ * last occurrence in the log, we keep a reference count in the cancel
+ * record in the table to tell us how many times we expect to see this
+ * record during the second pass.
+ */
+STATIC int
+xlog_recover_buffer_pass1(
+	struct log		*log,
+	xlog_recover_item_t	*item)
+{
+	xfs_buf_log_format_t	*buf_f = item->ri_buf[0].i_addr;
+	struct list_head	*bucket;
+	struct xfs_buf_cancel	*bcp;
+
+	/*
+	 * If this isn't a cancel buffer item, then just return.
+	 */
+	if (!(buf_f->blf_flags & XFS_BLF_CANCEL)) {
+		trace_xfs_log_recover_buf_not_cancel(log, buf_f);
+		return 0;
+	}
+
+	/*
+	 * Insert an xfs_buf_cancel record into the hash table of them.
+	 * If there is already an identical record, bump its reference count.
+	 */
+	bucket = XLOG_BUF_CANCEL_BUCKET(log, buf_f->blf_blkno);
+	list_for_each_entry(bcp, bucket, bc_list) {
+		if (bcp->bc_blkno == buf_f->blf_blkno &&
+		    bcp->bc_len == buf_f->blf_len) {
+			bcp->bc_refcount++;
+			trace_xfs_log_recover_buf_cancel_ref_inc(log, buf_f);
+			return 0;
+		}
+	}
+
+	bcp = kmem_alloc(sizeof(struct xfs_buf_cancel), KM_SLEEP);
+	bcp->bc_blkno = buf_f->blf_blkno;
+	bcp->bc_len = buf_f->blf_len;
+	bcp->bc_refcount = 1;
+	list_add_tail(&bcp->bc_list, bucket);
+
+	trace_xfs_log_recover_buf_cancel_add(log, buf_f);
+	return 0;
+}
+
+/*
+ * Check to see whether the buffer being recovered has a corresponding
+ * entry in the buffer cancel record table.  If it does then return 1
+ * so that it will be cancelled, otherwise return 0.  If the buffer is
+ * actually a buffer cancel item (XFS_BLF_CANCEL is set), then decrement
+ * the refcount on the entry in the table and remove it from the table
+ * if this is the last reference.
+ *
+ * We remove the cancel record from the table when we encounter its
+ * last occurrence in the log so that if the same buffer is re-used
+ * again after its last cancellation we actually replay the changes
+ * made at that point.
+ */
+STATIC int
+xlog_check_buffer_cancelled(
+	struct log		*log,
+	xfs_daddr_t		blkno,
+	uint			len,
+	ushort			flags)
+{
+	struct list_head	*bucket;
+	struct xfs_buf_cancel	*bcp;
+
+	if (log->l_buf_cancel_table == NULL) {
+		/*
+		 * There is nothing in the table built in pass one,
+		 * so this buffer must not be cancelled.
+		 */
+		ASSERT(!(flags & XFS_BLF_CANCEL));
+		return 0;
+	}
+
+	/*
+	 * Search for an entry in the  cancel table that matches our buffer.
+	 */
+	bucket = XLOG_BUF_CANCEL_BUCKET(log, blkno);
+	list_for_each_entry(bcp, bucket, bc_list) {
+		if (bcp->bc_blkno == blkno && bcp->bc_len == len)
+			goto found;
+	}
+
+	/*
+	 * We didn't find a corresponding entry in the table, so return 0 so
+	 * that the buffer is NOT cancelled.
+	 */
+	ASSERT(!(flags & XFS_BLF_CANCEL));
+	return 0;
+
+found:
+	/*
+	 * We've go a match, so return 1 so that the recovery of this buffer
+	 * is cancelled.  If this buffer is actually a buffer cancel log
+	 * item, then decrement the refcount on the one in the table and
+	 * remove it if this is the last reference.
+	 */
+	if (flags & XFS_BLF_CANCEL) {
+		if (--bcp->bc_refcount == 0) {
+			list_del(&bcp->bc_list);
+			kmem_free(bcp);
+		}
+	}
+	return 1;
+}
+
+/*
+ * Perform recovery for a buffer full of inodes.  In these buffers, the only
+ * data which should be recovered is that which corresponds to the
+ * di_next_unlinked pointers in the on disk inode structures.  The rest of the
+ * data for the inodes is always logged through the inodes themselves rather
+ * than the inode buffer and is recovered in xlog_recover_inode_pass2().
+ *
+ * The only time when buffers full of inodes are fully recovered is when the
+ * buffer is full of newly allocated inodes.  In this case the buffer will
+ * not be marked as an inode buffer and so will be sent to
+ * xlog_recover_do_reg_buffer() below during recovery.
+ */
+STATIC int
+xlog_recover_do_inode_buffer(
+	struct xfs_mount	*mp,
+	xlog_recover_item_t	*item,
+	struct xfs_buf		*bp,
+	xfs_buf_log_format_t	*buf_f)
+{
+	int			i;
+	int			item_index = 0;
+	int			bit = 0;
+	int			nbits = 0;
+	int			reg_buf_offset = 0;
+	int			reg_buf_bytes = 0;
+	int			next_unlinked_offset;
+	int			inodes_per_buf;
+	xfs_agino_t		*logged_nextp;
+	xfs_agino_t		*buffer_nextp;
+
+	trace_xfs_log_recover_buf_inode_buf(mp->m_log, buf_f);
+
+	inodes_per_buf = XFS_BUF_COUNT(bp) >> mp->m_sb.sb_inodelog;
+	for (i = 0; i < inodes_per_buf; i++) {
+		next_unlinked_offset = (i * mp->m_sb.sb_inodesize) +
+			offsetof(xfs_dinode_t, di_next_unlinked);
+
+		while (next_unlinked_offset >=
+		       (reg_buf_offset + reg_buf_bytes)) {
+			/*
+			 * The next di_next_unlinked field is beyond
+			 * the current logged region.  Find the next
+			 * logged region that contains or is beyond
+			 * the current di_next_unlinked field.
+			 */
+			bit += nbits;
+			bit = xfs_next_bit(buf_f->blf_data_map,
+					   buf_f->blf_map_size, bit);
+
+			/*
+			 * If there are no more logged regions in the
+			 * buffer, then we're done.
+			 */
+			if (bit == -1)
+				return 0;
+
+			nbits = xfs_contig_bits(buf_f->blf_data_map,
+						buf_f->blf_map_size, bit);
+			ASSERT(nbits > 0);
+			reg_buf_offset = bit << XFS_BLF_SHIFT;
+			reg_buf_bytes = nbits << XFS_BLF_SHIFT;
+			item_index++;
+		}
+
+		/*
+		 * If the current logged region starts after the current
+		 * di_next_unlinked field, then move on to the next
+		 * di_next_unlinked field.
+		 */
+		if (next_unlinked_offset < reg_buf_offset)
+			continue;
+
+		ASSERT(item->ri_buf[item_index].i_addr != NULL);
+		ASSERT((item->ri_buf[item_index].i_len % XFS_BLF_CHUNK) == 0);
+		ASSERT((reg_buf_offset + reg_buf_bytes) <= XFS_BUF_COUNT(bp));
+
+		/*
+		 * The current logged region contains a copy of the
+		 * current di_next_unlinked field.  Extract its value
+		 * and copy it to the buffer copy.
+		 */
+		logged_nextp = item->ri_buf[item_index].i_addr +
+				next_unlinked_offset - reg_buf_offset;
+		if (unlikely(*logged_nextp == 0)) {
+			xfs_alert(mp,
+		"Bad inode buffer log record (ptr = 0x%p, bp = 0x%p). "
+		"Trying to replay bad (0) inode di_next_unlinked field.",
+				item, bp);
+			XFS_ERROR_REPORT("xlog_recover_do_inode_buf",
+					 XFS_ERRLEVEL_LOW, mp);
+			return XFS_ERROR(EFSCORRUPTED);
+		}
+
+		buffer_nextp = (xfs_agino_t *)xfs_buf_offset(bp,
+					      next_unlinked_offset);
+		*buffer_nextp = *logged_nextp;
+	}
+
+	return 0;
+}
+
+/*
+ * Perform a 'normal' buffer recovery.  Each logged region of the
+ * buffer should be copied over the corresponding region in the
+ * given buffer.  The bitmap in the buf log format structure indicates
+ * where to place the logged data.
+ */
+STATIC void
+xlog_recover_do_reg_buffer(
+	struct xfs_mount	*mp,
+	xlog_recover_item_t	*item,
+	struct xfs_buf		*bp,
+	xfs_buf_log_format_t	*buf_f)
+{
+	int			i;
+	int			bit;
+	int			nbits;
+	int                     error;
+
+	trace_xfs_log_recover_buf_reg_buf(mp->m_log, buf_f);
+
+	bit = 0;
+	i = 1;  /* 0 is the buf format structure */
+	while (1) {
+		bit = xfs_next_bit(buf_f->blf_data_map,
+				   buf_f->blf_map_size, bit);
+		if (bit == -1)
+			break;
+		nbits = xfs_contig_bits(buf_f->blf_data_map,
+					buf_f->blf_map_size, bit);
+		ASSERT(nbits > 0);
+		ASSERT(item->ri_buf[i].i_addr != NULL);
+		ASSERT(item->ri_buf[i].i_len % XFS_BLF_CHUNK == 0);
+		ASSERT(XFS_BUF_COUNT(bp) >=
+		       ((uint)bit << XFS_BLF_SHIFT)+(nbits<<XFS_BLF_SHIFT));
+
+		/*
+		 * Do a sanity check if this is a dquot buffer. Just checking
+		 * the first dquot in the buffer should do. XXXThis is
+		 * probably a good thing to do for other buf types also.
+		 */
+		error = 0;
+		if (buf_f->blf_flags &
+		   (XFS_BLF_UDQUOT_BUF|XFS_BLF_PDQUOT_BUF|XFS_BLF_GDQUOT_BUF)) {
+			if (item->ri_buf[i].i_addr == NULL) {
+				xfs_alert(mp,
+					"XFS: NULL dquot in %s.", __func__);
+				goto next;
+			}
+			if (item->ri_buf[i].i_len < sizeof(xfs_disk_dquot_t)) {
+				xfs_alert(mp,
+					"XFS: dquot too small (%d) in %s.",
+					item->ri_buf[i].i_len, __func__);
+				goto next;
+			}
+			error = xfs_qm_dqcheck(mp, item->ri_buf[i].i_addr,
+					       -1, 0, XFS_QMOPT_DOWARN,
+					       "dquot_buf_recover");
+			if (error)
+				goto next;
+		}
+
+		memcpy(xfs_buf_offset(bp,
+			(uint)bit << XFS_BLF_SHIFT),	/* dest */
+			item->ri_buf[i].i_addr,		/* source */
+			nbits<<XFS_BLF_SHIFT);		/* length */
+ next:
+		i++;
+		bit += nbits;
+	}
+
+	/* Shouldn't be any more regions */
+	ASSERT(i == item->ri_total);
+}
+
+/*
+ * Do some primitive error checking on ondisk dquot data structures.
+ */
+int
+xfs_qm_dqcheck(
+	struct xfs_mount *mp,
+	xfs_disk_dquot_t *ddq,
+	xfs_dqid_t	 id,
+	uint		 type,	  /* used only when IO_dorepair is true */
+	uint		 flags,
+	char		 *str)
+{
+	xfs_dqblk_t	 *d = (xfs_dqblk_t *)ddq;
+	int		errs = 0;
+
+	/*
+	 * We can encounter an uninitialized dquot buffer for 2 reasons:
+	 * 1. If we crash while deleting the quotainode(s), and those blks got
+	 *    used for user data. This is because we take the path of regular
+	 *    file deletion; however, the size field of quotainodes is never
+	 *    updated, so all the tricks that we play in itruncate_finish
+	 *    don't quite matter.
+	 *
+	 * 2. We don't play the quota buffers when there's a quotaoff logitem.
+	 *    But the allocation will be replayed so we'll end up with an
+	 *    uninitialized quota block.
+	 *
+	 * This is all fine; things are still consistent, and we haven't lost
+	 * any quota information. Just don't complain about bad dquot blks.
+	 */
+	if (be16_to_cpu(ddq->d_magic) != XFS_DQUOT_MAGIC) {
+		if (flags & XFS_QMOPT_DOWARN)
+			xfs_alert(mp,
+			"%s : XFS dquot ID 0x%x, magic 0x%x != 0x%x",
+			str, id, be16_to_cpu(ddq->d_magic), XFS_DQUOT_MAGIC);
+		errs++;
+	}
+	if (ddq->d_version != XFS_DQUOT_VERSION) {
+		if (flags & XFS_QMOPT_DOWARN)
+			xfs_alert(mp,
+			"%s : XFS dquot ID 0x%x, version 0x%x != 0x%x",
+			str, id, ddq->d_version, XFS_DQUOT_VERSION);
+		errs++;
+	}
+
+	if (ddq->d_flags != XFS_DQ_USER &&
+	    ddq->d_flags != XFS_DQ_PROJ &&
+	    ddq->d_flags != XFS_DQ_GROUP) {
+		if (flags & XFS_QMOPT_DOWARN)
+			xfs_alert(mp,
+			"%s : XFS dquot ID 0x%x, unknown flags 0x%x",
+			str, id, ddq->d_flags);
+		errs++;
+	}
+
+	if (id != -1 && id != be32_to_cpu(ddq->d_id)) {
+		if (flags & XFS_QMOPT_DOWARN)
+			xfs_alert(mp,
+			"%s : ondisk-dquot 0x%p, ID mismatch: "
+			"0x%x expected, found id 0x%x",
+			str, ddq, id, be32_to_cpu(ddq->d_id));
+		errs++;
+	}
+
+	if (!errs && ddq->d_id) {
+		if (ddq->d_blk_softlimit &&
+		    be64_to_cpu(ddq->d_bcount) >=
+				be64_to_cpu(ddq->d_blk_softlimit)) {
+			if (!ddq->d_btimer) {
+				if (flags & XFS_QMOPT_DOWARN)
+					xfs_alert(mp,
+			"%s : Dquot ID 0x%x (0x%p) BLK TIMER NOT STARTED",
+					str, (int)be32_to_cpu(ddq->d_id), ddq);
+				errs++;
+			}
+		}
+		if (ddq->d_ino_softlimit &&
+		    be64_to_cpu(ddq->d_icount) >=
+				be64_to_cpu(ddq->d_ino_softlimit)) {
+			if (!ddq->d_itimer) {
+				if (flags & XFS_QMOPT_DOWARN)
+					xfs_alert(mp,
+			"%s : Dquot ID 0x%x (0x%p) INODE TIMER NOT STARTED",
+					str, (int)be32_to_cpu(ddq->d_id), ddq);
+				errs++;
+			}
+		}
+		if (ddq->d_rtb_softlimit &&
+		    be64_to_cpu(ddq->d_rtbcount) >=
+				be64_to_cpu(ddq->d_rtb_softlimit)) {
+			if (!ddq->d_rtbtimer) {
+				if (flags & XFS_QMOPT_DOWARN)
+					xfs_alert(mp,
+			"%s : Dquot ID 0x%x (0x%p) RTBLK TIMER NOT STARTED",
+					str, (int)be32_to_cpu(ddq->d_id), ddq);
+				errs++;
+			}
+		}
+	}
+
+	if (!errs || !(flags & XFS_QMOPT_DQREPAIR))
+		return errs;
+
+	if (flags & XFS_QMOPT_DOWARN)
+		xfs_notice(mp, "Re-initializing dquot ID 0x%x", id);
+
+	/*
+	 * Typically, a repair is only requested by quotacheck.
+	 */
+	ASSERT(id != -1);
+	ASSERT(flags & XFS_QMOPT_DQREPAIR);
+	memset(d, 0, sizeof(xfs_dqblk_t));
+
+	d->dd_diskdq.d_magic = cpu_to_be16(XFS_DQUOT_MAGIC);
+	d->dd_diskdq.d_version = XFS_DQUOT_VERSION;
+	d->dd_diskdq.d_flags = type;
+	d->dd_diskdq.d_id = cpu_to_be32(id);
+
+	return errs;
+}
+
+/*
+ * Perform a dquot buffer recovery.
+ * Simple algorithm: if we have found a QUOTAOFF logitem of the same type
+ * (ie. USR or GRP), then just toss this buffer away; don't recover it.
+ * Else, treat it as a regular buffer and do recovery.
+ */
+STATIC void
+xlog_recover_do_dquot_buffer(
+	xfs_mount_t		*mp,
+	xlog_t			*log,
+	xlog_recover_item_t	*item,
+	xfs_buf_t		*bp,
+	xfs_buf_log_format_t	*buf_f)
+{
+	uint			type;
+
+	trace_xfs_log_recover_buf_dquot_buf(log, buf_f);
+
+	/*
+	 * Filesystems are required to send in quota flags at mount time.
+	 */
+	if (mp->m_qflags == 0) {
+		return;
+	}
+
+	type = 0;
+	if (buf_f->blf_flags & XFS_BLF_UDQUOT_BUF)
+		type |= XFS_DQ_USER;
+	if (buf_f->blf_flags & XFS_BLF_PDQUOT_BUF)
+		type |= XFS_DQ_PROJ;
+	if (buf_f->blf_flags & XFS_BLF_GDQUOT_BUF)
+		type |= XFS_DQ_GROUP;
+	/*
+	 * This type of quotas was turned off, so ignore this buffer
+	 */
+	if (log->l_quotaoffs_flag & type)
+		return;
+
+	xlog_recover_do_reg_buffer(mp, item, bp, buf_f);
+}
+
+/*
+ * This routine replays a modification made to a buffer at runtime.
+ * There are actually two types of buffer, regular and inode, which
+ * are handled differently.  Inode buffers are handled differently
+ * in that we only recover a specific set of data from them, namely
+ * the inode di_next_unlinked fields.  This is because all other inode
+ * data is actually logged via inode records and any data we replay
+ * here which overlaps that may be stale.
+ *
+ * When meta-data buffers are freed at run time we log a buffer item
+ * with the XFS_BLF_CANCEL bit set to indicate that previous copies
+ * of the buffer in the log should not be replayed at recovery time.
+ * This is so that if the blocks covered by the buffer are reused for
+ * file data before we crash we don't end up replaying old, freed
+ * meta-data into a user's file.
+ *
+ * To handle the cancellation of buffer log items, we make two passes
+ * over the log during recovery.  During the first we build a table of
+ * those buffers which have been cancelled, and during the second we
+ * only replay those buffers which do not have corresponding cancel
+ * records in the table.  See xlog_recover_do_buffer_pass[1,2] above
+ * for more details on the implementation of the table of cancel records.
+ */
+STATIC int
+xlog_recover_buffer_pass2(
+	xlog_t			*log,
+	xlog_recover_item_t	*item)
+{
+	xfs_buf_log_format_t	*buf_f = item->ri_buf[0].i_addr;
+	xfs_mount_t		*mp = log->l_mp;
+	xfs_buf_t		*bp;
+	int			error;
+	uint			buf_flags;
+
+	/*
+	 * In this pass we only want to recover all the buffers which have
+	 * not been cancelled and are not cancellation buffers themselves.
+	 */
+	if (xlog_check_buffer_cancelled(log, buf_f->blf_blkno,
+			buf_f->blf_len, buf_f->blf_flags)) {
+		trace_xfs_log_recover_buf_cancel(log, buf_f);
+		return 0;
+	}
+
+	trace_xfs_log_recover_buf_recover(log, buf_f);
+
+	buf_flags = XBF_LOCK;
+	if (!(buf_f->blf_flags & XFS_BLF_INODE_BUF))
+		buf_flags |= XBF_MAPPED;
+
+	bp = xfs_buf_read(mp->m_ddev_targp, buf_f->blf_blkno, buf_f->blf_len,
+			  buf_flags);
+	if (XFS_BUF_ISERROR(bp)) {
+		xfs_ioerror_alert("xlog_recover_do..(read#1)", mp,
+				  bp, buf_f->blf_blkno);
+		error = XFS_BUF_GETERROR(bp);
+		xfs_buf_relse(bp);
+		return error;
+	}
+
+	error = 0;
+	if (buf_f->blf_flags & XFS_BLF_INODE_BUF) {
+		error = xlog_recover_do_inode_buffer(mp, item, bp, buf_f);
+	} else if (buf_f->blf_flags &
+		  (XFS_BLF_UDQUOT_BUF|XFS_BLF_PDQUOT_BUF|XFS_BLF_GDQUOT_BUF)) {
+		xlog_recover_do_dquot_buffer(mp, log, item, bp, buf_f);
+	} else {
+		xlog_recover_do_reg_buffer(mp, item, bp, buf_f);
+	}
+	if (error)
+		return XFS_ERROR(error);
+
+	/*
+	 * Perform delayed write on the buffer.  Asynchronous writes will be
+	 * slower when taking into account all the buffers to be flushed.
+	 *
+	 * Also make sure that only inode buffers with good sizes stay in
+	 * the buffer cache.  The kernel moves inodes in buffers of 1 block
+	 * or XFS_INODE_CLUSTER_SIZE bytes, whichever is bigger.  The inode
+	 * buffers in the log can be a different size if the log was generated
+	 * by an older kernel using unclustered inode buffers or a newer kernel
+	 * running with a different inode cluster size.  Regardless, if the
+	 * the inode buffer size isn't MAX(blocksize, XFS_INODE_CLUSTER_SIZE)
+	 * for *our* value of XFS_INODE_CLUSTER_SIZE, then we need to keep
+	 * the buffer out of the buffer cache so that the buffer won't
+	 * overlap with future reads of those inodes.
+	 */
+	if (XFS_DINODE_MAGIC ==
+	    be16_to_cpu(*((__be16 *)xfs_buf_offset(bp, 0))) &&
+	    (XFS_BUF_COUNT(bp) != MAX(log->l_mp->m_sb.sb_blocksize,
+			(__uint32_t)XFS_INODE_CLUSTER_SIZE(log->l_mp)))) {
+		XFS_BUF_STALE(bp);
+		error = xfs_bwrite(mp, bp);
+	} else {
+		ASSERT(bp->b_target->bt_mount == mp);
+		XFS_BUF_SET_IODONE_FUNC(bp, xlog_recover_iodone);
+		xfs_bdwrite(mp, bp);
+	}
+
+	return (error);
+}
+
+STATIC int
+xlog_recover_inode_pass2(
+	xlog_t			*log,
+	xlog_recover_item_t	*item)
+{
+	xfs_inode_log_format_t	*in_f;
+	xfs_mount_t		*mp = log->l_mp;
+	xfs_buf_t		*bp;
+	xfs_dinode_t		*dip;
+	int			len;
+	xfs_caddr_t		src;
+	xfs_caddr_t		dest;
+	int			error;
+	int			attr_index;
+	uint			fields;
+	xfs_icdinode_t		*dicp;
+	int			need_free = 0;
+
+	if (item->ri_buf[0].i_len == sizeof(xfs_inode_log_format_t)) {
+		in_f = item->ri_buf[0].i_addr;
+	} else {
+		in_f = kmem_alloc(sizeof(xfs_inode_log_format_t), KM_SLEEP);
+		need_free = 1;
+		error = xfs_inode_item_format_convert(&item->ri_buf[0], in_f);
+		if (error)
+			goto error;
+	}
+
+	/*
+	 * Inode buffers can be freed, look out for it,
+	 * and do not replay the inode.
+	 */
+	if (xlog_check_buffer_cancelled(log, in_f->ilf_blkno,
+					in_f->ilf_len, 0)) {
+		error = 0;
+		trace_xfs_log_recover_inode_cancel(log, in_f);
+		goto error;
+	}
+	trace_xfs_log_recover_inode_recover(log, in_f);
+
+	bp = xfs_buf_read(mp->m_ddev_targp, in_f->ilf_blkno, in_f->ilf_len,
+			  XBF_LOCK);
+	if (XFS_BUF_ISERROR(bp)) {
+		xfs_ioerror_alert("xlog_recover_do..(read#2)", mp,
+				  bp, in_f->ilf_blkno);
+		error = XFS_BUF_GETERROR(bp);
+		xfs_buf_relse(bp);
+		goto error;
+	}
+	error = 0;
+	ASSERT(in_f->ilf_fields & XFS_ILOG_CORE);
+	dip = (xfs_dinode_t *)xfs_buf_offset(bp, in_f->ilf_boffset);
+
+	/*
+	 * Make sure the place we're flushing out to really looks
+	 * like an inode!
+	 */
+	if (unlikely(be16_to_cpu(dip->di_magic) != XFS_DINODE_MAGIC)) {
+		xfs_buf_relse(bp);
+		xfs_alert(mp,
+	"%s: Bad inode magic number, dip = 0x%p, dino bp = 0x%p, ino = %Ld",
+			__func__, dip, bp, in_f->ilf_ino);
+		XFS_ERROR_REPORT("xlog_recover_inode_pass2(1)",
+				 XFS_ERRLEVEL_LOW, mp);
+		error = EFSCORRUPTED;
+		goto error;
+	}
+	dicp = item->ri_buf[1].i_addr;
+	if (unlikely(dicp->di_magic != XFS_DINODE_MAGIC)) {
+		xfs_buf_relse(bp);
+		xfs_alert(mp,
+			"%s: Bad inode log record, rec ptr 0x%p, ino %Ld",
+			__func__, item, in_f->ilf_ino);
+		XFS_ERROR_REPORT("xlog_recover_inode_pass2(2)",
+				 XFS_ERRLEVEL_LOW, mp);
+		error = EFSCORRUPTED;
+		goto error;
+	}
+
+	/* Skip replay when the on disk inode is newer than the log one */
+	if (dicp->di_flushiter < be16_to_cpu(dip->di_flushiter)) {
+		/*
+		 * Deal with the wrap case, DI_MAX_FLUSH is less
+		 * than smaller numbers
+		 */
+		if (be16_to_cpu(dip->di_flushiter) == DI_MAX_FLUSH &&
+		    dicp->di_flushiter < (DI_MAX_FLUSH >> 1)) {
+			/* do nothing */
+		} else {
+			xfs_buf_relse(bp);
+			trace_xfs_log_recover_inode_skip(log, in_f);
+			error = 0;
+			goto error;
+		}
+	}
+	/* Take the opportunity to reset the flush iteration count */
+	dicp->di_flushiter = 0;
+
+	if (unlikely((dicp->di_mode & S_IFMT) == S_IFREG)) {
+		if ((dicp->di_format != XFS_DINODE_FMT_EXTENTS) &&
+		    (dicp->di_format != XFS_DINODE_FMT_BTREE)) {
+			XFS_CORRUPTION_ERROR("xlog_recover_inode_pass2(3)",
+					 XFS_ERRLEVEL_LOW, mp, dicp);
+			xfs_buf_relse(bp);
+			xfs_alert(mp,
+		"%s: Bad regular inode log record, rec ptr 0x%p, "
+		"ino ptr = 0x%p, ino bp = 0x%p, ino %Ld",
+				__func__, item, dip, bp, in_f->ilf_ino);
+			error = EFSCORRUPTED;
+			goto error;
+		}
+	} else if (unlikely((dicp->di_mode & S_IFMT) == S_IFDIR)) {
+		if ((dicp->di_format != XFS_DINODE_FMT_EXTENTS) &&
+		    (dicp->di_format != XFS_DINODE_FMT_BTREE) &&
+		    (dicp->di_format != XFS_DINODE_FMT_LOCAL)) {
+			XFS_CORRUPTION_ERROR("xlog_recover_inode_pass2(4)",
+					     XFS_ERRLEVEL_LOW, mp, dicp);
+			xfs_buf_relse(bp);
+			xfs_alert(mp,
+		"%s: Bad dir inode log record, rec ptr 0x%p, "
+		"ino ptr = 0x%p, ino bp = 0x%p, ino %Ld",
+				__func__, item, dip, bp, in_f->ilf_ino);
+			error = EFSCORRUPTED;
+			goto error;
+		}
+	}
+	if (unlikely(dicp->di_nextents + dicp->di_anextents > dicp->di_nblocks)){
+		XFS_CORRUPTION_ERROR("xlog_recover_inode_pass2(5)",
+				     XFS_ERRLEVEL_LOW, mp, dicp);
+		xfs_buf_relse(bp);
+		xfs_alert(mp,
+	"%s: Bad inode log record, rec ptr 0x%p, dino ptr 0x%p, "
+	"dino bp 0x%p, ino %Ld, total extents = %d, nblocks = %Ld",
+			__func__, item, dip, bp, in_f->ilf_ino,
+			dicp->di_nextents + dicp->di_anextents,
+			dicp->di_nblocks);
+		error = EFSCORRUPTED;
+		goto error;
+	}
+	if (unlikely(dicp->di_forkoff > mp->m_sb.sb_inodesize)) {
+		XFS_CORRUPTION_ERROR("xlog_recover_inode_pass2(6)",
+				     XFS_ERRLEVEL_LOW, mp, dicp);
+		xfs_buf_relse(bp);
+		xfs_alert(mp,
+	"%s: Bad inode log record, rec ptr 0x%p, dino ptr 0x%p, "
+	"dino bp 0x%p, ino %Ld, forkoff 0x%x", __func__,
+			item, dip, bp, in_f->ilf_ino, dicp->di_forkoff);
+		error = EFSCORRUPTED;
+		goto error;
+	}
+	if (unlikely(item->ri_buf[1].i_len > sizeof(struct xfs_icdinode))) {
+		XFS_CORRUPTION_ERROR("xlog_recover_inode_pass2(7)",
+				     XFS_ERRLEVEL_LOW, mp, dicp);
+		xfs_buf_relse(bp);
+		xfs_alert(mp,
+			"%s: Bad inode log record length %d, rec ptr 0x%p",
+			__func__, item->ri_buf[1].i_len, item);
+		error = EFSCORRUPTED;
+		goto error;
+	}
+
+	/* The core is in in-core format */
+	xfs_dinode_to_disk(dip, item->ri_buf[1].i_addr);
+
+	/* the rest is in on-disk format */
+	if (item->ri_buf[1].i_len > sizeof(struct xfs_icdinode)) {
+		memcpy((xfs_caddr_t) dip + sizeof(struct xfs_icdinode),
+			item->ri_buf[1].i_addr + sizeof(struct xfs_icdinode),
+			item->ri_buf[1].i_len  - sizeof(struct xfs_icdinode));
+	}
+
+	fields = in_f->ilf_fields;
+	switch (fields & (XFS_ILOG_DEV | XFS_ILOG_UUID)) {
+	case XFS_ILOG_DEV:
+		xfs_dinode_put_rdev(dip, in_f->ilf_u.ilfu_rdev);
+		break;
+	case XFS_ILOG_UUID:
+		memcpy(XFS_DFORK_DPTR(dip),
+		       &in_f->ilf_u.ilfu_uuid,
+		       sizeof(uuid_t));
+		break;
+	}
+
+	if (in_f->ilf_size == 2)
+		goto write_inode_buffer;
+	len = item->ri_buf[2].i_len;
+	src = item->ri_buf[2].i_addr;
+	ASSERT(in_f->ilf_size <= 4);
+	ASSERT((in_f->ilf_size == 3) || (fields & XFS_ILOG_AFORK));
+	ASSERT(!(fields & XFS_ILOG_DFORK) ||
+	       (len == in_f->ilf_dsize));
+
+	switch (fields & XFS_ILOG_DFORK) {
+	case XFS_ILOG_DDATA:
+	case XFS_ILOG_DEXT:
+		memcpy(XFS_DFORK_DPTR(dip), src, len);
+		break;
+
+	case XFS_ILOG_DBROOT:
+		xfs_bmbt_to_bmdr(mp, (struct xfs_btree_block *)src, len,
+				 (xfs_bmdr_block_t *)XFS_DFORK_DPTR(dip),
+				 XFS_DFORK_DSIZE(dip, mp));
+		break;
+
+	default:
+		/*
+		 * There are no data fork flags set.
+		 */
+		ASSERT((fields & XFS_ILOG_DFORK) == 0);
+		break;
+	}
+
+	/*
+	 * If we logged any attribute data, recover it.  There may or
+	 * may not have been any other non-core data logged in this
+	 * transaction.
+	 */
+	if (in_f->ilf_fields & XFS_ILOG_AFORK) {
+		if (in_f->ilf_fields & XFS_ILOG_DFORK) {
+			attr_index = 3;
+		} else {
+			attr_index = 2;
+		}
+		len = item->ri_buf[attr_index].i_len;
+		src = item->ri_buf[attr_index].i_addr;
+		ASSERT(len == in_f->ilf_asize);
+
+		switch (in_f->ilf_fields & XFS_ILOG_AFORK) {
+		case XFS_ILOG_ADATA:
+		case XFS_ILOG_AEXT:
+			dest = XFS_DFORK_APTR(dip);
+			ASSERT(len <= XFS_DFORK_ASIZE(dip, mp));
+			memcpy(dest, src, len);
+			break;
+
+		case XFS_ILOG_ABROOT:
+			dest = XFS_DFORK_APTR(dip);
+			xfs_bmbt_to_bmdr(mp, (struct xfs_btree_block *)src,
+					 len, (xfs_bmdr_block_t*)dest,
+					 XFS_DFORK_ASIZE(dip, mp));
+			break;
+
+		default:
+			xfs_warn(log->l_mp, "%s: Invalid flag", __func__);
+			ASSERT(0);
+			xfs_buf_relse(bp);
+			error = EIO;
+			goto error;
+		}
+	}
+
+write_inode_buffer:
+	ASSERT(bp->b_target->bt_mount == mp);
+	XFS_BUF_SET_IODONE_FUNC(bp, xlog_recover_iodone);
+	xfs_bdwrite(mp, bp);
+error:
+	if (need_free)
+		kmem_free(in_f);
+	return XFS_ERROR(error);
+}
+
+/*
+ * Recover QUOTAOFF records. We simply make a note of it in the xlog_t
+ * structure, so that we know not to do any dquot item or dquot buffer recovery,
+ * of that type.
+ */
+STATIC int
+xlog_recover_quotaoff_pass1(
+	xlog_t			*log,
+	xlog_recover_item_t	*item)
+{
+	xfs_qoff_logformat_t	*qoff_f = item->ri_buf[0].i_addr;
+	ASSERT(qoff_f);
+
+	/*
+	 * The logitem format's flag tells us if this was user quotaoff,
+	 * group/project quotaoff or both.
+	 */
+	if (qoff_f->qf_flags & XFS_UQUOTA_ACCT)
+		log->l_quotaoffs_flag |= XFS_DQ_USER;
+	if (qoff_f->qf_flags & XFS_PQUOTA_ACCT)
+		log->l_quotaoffs_flag |= XFS_DQ_PROJ;
+	if (qoff_f->qf_flags & XFS_GQUOTA_ACCT)
+		log->l_quotaoffs_flag |= XFS_DQ_GROUP;
+
+	return (0);
+}
+
+/*
+ * Recover a dquot record
+ */
+STATIC int
+xlog_recover_dquot_pass2(
+	xlog_t			*log,
+	xlog_recover_item_t	*item)
+{
+	xfs_mount_t		*mp = log->l_mp;
+	xfs_buf_t		*bp;
+	struct xfs_disk_dquot	*ddq, *recddq;
+	int			error;
+	xfs_dq_logformat_t	*dq_f;
+	uint			type;
+
+
+	/*
+	 * Filesystems are required to send in quota flags at mount time.
+	 */
+	if (mp->m_qflags == 0)
+		return (0);
+
+	recddq = item->ri_buf[1].i_addr;
+	if (recddq == NULL) {
+		xfs_alert(log->l_mp, "NULL dquot in %s.", __func__);
+		return XFS_ERROR(EIO);
+	}
+	if (item->ri_buf[1].i_len < sizeof(xfs_disk_dquot_t)) {
+		xfs_alert(log->l_mp, "dquot too small (%d) in %s.",
+			item->ri_buf[1].i_len, __func__);
+		return XFS_ERROR(EIO);
+	}
+
+	/*
+	 * This type of quotas was turned off, so ignore this record.
+	 */
+	type = recddq->d_flags & (XFS_DQ_USER | XFS_DQ_PROJ | XFS_DQ_GROUP);
+	ASSERT(type);
+	if (log->l_quotaoffs_flag & type)
+		return (0);
+
+	/*
+	 * At this point we know that quota was _not_ turned off.
+	 * Since the mount flags are not indicating to us otherwise, this
+	 * must mean that quota is on, and the dquot needs to be replayed.
+	 * Remember that we may not have fully recovered the superblock yet,
+	 * so we can't do the usual trick of looking at the SB quota bits.
+	 *
+	 * The other possibility, of course, is that the quota subsystem was
+	 * removed since the last mount - ENOSYS.
+	 */
+	dq_f = item->ri_buf[0].i_addr;
+	ASSERT(dq_f);
+	error = xfs_qm_dqcheck(mp, recddq, dq_f->qlf_id, 0, XFS_QMOPT_DOWARN,
+			   "xlog_recover_dquot_pass2 (log copy)");
+	if (error)
+		return XFS_ERROR(EIO);
+	ASSERT(dq_f->qlf_len == 1);
+
+	error = xfs_read_buf(mp, mp->m_ddev_targp,
+			     dq_f->qlf_blkno,
+			     XFS_FSB_TO_BB(mp, dq_f->qlf_len),
+			     0, &bp);
+	if (error) {
+		xfs_ioerror_alert("xlog_recover_do..(read#3)", mp,
+				  bp, dq_f->qlf_blkno);
+		return error;
+	}
+	ASSERT(bp);
+	ddq = (xfs_disk_dquot_t *)xfs_buf_offset(bp, dq_f->qlf_boffset);
+
+	/*
+	 * At least the magic num portion should be on disk because this
+	 * was among a chunk of dquots created earlier, and we did some
+	 * minimal initialization then.
+	 */
+	error = xfs_qm_dqcheck(mp, ddq, dq_f->qlf_id, 0, XFS_QMOPT_DOWARN,
+			   "xlog_recover_dquot_pass2");
+	if (error) {
+		xfs_buf_relse(bp);
+		return XFS_ERROR(EIO);
+	}
+
+	memcpy(ddq, recddq, item->ri_buf[1].i_len);
+
+	ASSERT(dq_f->qlf_size == 2);
+	ASSERT(bp->b_target->bt_mount == mp);
+	XFS_BUF_SET_IODONE_FUNC(bp, xlog_recover_iodone);
+	xfs_bdwrite(mp, bp);
+
+	return (0);
+}
+
+/*
+ * This routine is called to create an in-core extent free intent
+ * item from the efi format structure which was logged on disk.
+ * It allocates an in-core efi, copies the extents from the format
+ * structure into it, and adds the efi to the AIL with the given
+ * LSN.
+ */
+STATIC int
+xlog_recover_efi_pass2(
+	xlog_t			*log,
+	xlog_recover_item_t	*item,
+	xfs_lsn_t		lsn)
+{
+	int			error;
+	xfs_mount_t		*mp = log->l_mp;
+	xfs_efi_log_item_t	*efip;
+	xfs_efi_log_format_t	*efi_formatp;
+
+	efi_formatp = item->ri_buf[0].i_addr;
+
+	efip = xfs_efi_init(mp, efi_formatp->efi_nextents);
+	if ((error = xfs_efi_copy_format(&(item->ri_buf[0]),
+					 &(efip->efi_format)))) {
+		xfs_efi_item_free(efip);
+		return error;
+	}
+	atomic_set(&efip->efi_next_extent, efi_formatp->efi_nextents);
+
+	spin_lock(&log->l_ailp->xa_lock);
+	/*
+	 * xfs_trans_ail_update() drops the AIL lock.
+	 */
+	xfs_trans_ail_update(log->l_ailp, &efip->efi_item, lsn);
+	return 0;
+}
+
+
+/*
+ * This routine is called when an efd format structure is found in
+ * a committed transaction in the log.  It's purpose is to cancel
+ * the corresponding efi if it was still in the log.  To do this
+ * it searches the AIL for the efi with an id equal to that in the
+ * efd format structure.  If we find it, we remove the efi from the
+ * AIL and free it.
+ */
+STATIC int
+xlog_recover_efd_pass2(
+	xlog_t			*log,
+	xlog_recover_item_t	*item)
+{
+	xfs_efd_log_format_t	*efd_formatp;
+	xfs_efi_log_item_t	*efip = NULL;
+	xfs_log_item_t		*lip;
+	__uint64_t		efi_id;
+	struct xfs_ail_cursor	cur;
+	struct xfs_ail		*ailp = log->l_ailp;
+
+	efd_formatp = item->ri_buf[0].i_addr;
+	ASSERT((item->ri_buf[0].i_len == (sizeof(xfs_efd_log_format_32_t) +
+		((efd_formatp->efd_nextents - 1) * sizeof(xfs_extent_32_t)))) ||
+	       (item->ri_buf[0].i_len == (sizeof(xfs_efd_log_format_64_t) +
+		((efd_formatp->efd_nextents - 1) * sizeof(xfs_extent_64_t)))));
+	efi_id = efd_formatp->efd_efi_id;
+
+	/*
+	 * Search for the efi with the id in the efd format structure
+	 * in the AIL.
+	 */
+	spin_lock(&ailp->xa_lock);
+	lip = xfs_trans_ail_cursor_first(ailp, &cur, 0);
+	while (lip != NULL) {
+		if (lip->li_type == XFS_LI_EFI) {
+			efip = (xfs_efi_log_item_t *)lip;
+			if (efip->efi_format.efi_id == efi_id) {
+				/*
+				 * xfs_trans_ail_delete() drops the
+				 * AIL lock.
+				 */
+				xfs_trans_ail_delete(ailp, lip);
+				xfs_efi_item_free(efip);
+				spin_lock(&ailp->xa_lock);
+				break;
+			}
+		}
+		lip = xfs_trans_ail_cursor_next(ailp, &cur);
+	}
+	xfs_trans_ail_cursor_done(ailp, &cur);
+	spin_unlock(&ailp->xa_lock);
+
+	return 0;
+}
+
+/*
+ * Free up any resources allocated by the transaction
+ *
+ * Remember that EFIs, EFDs, and IUNLINKs are handled later.
+ */
+STATIC void
+xlog_recover_free_trans(
+	struct xlog_recover	*trans)
+{
+	xlog_recover_item_t	*item, *n;
+	int			i;
+
+	list_for_each_entry_safe(item, n, &trans->r_itemq, ri_list) {
+		/* Free the regions in the item. */
+		list_del(&item->ri_list);
+		for (i = 0; i < item->ri_cnt; i++)
+			kmem_free(item->ri_buf[i].i_addr);
+		/* Free the item itself */
+		kmem_free(item->ri_buf);
+		kmem_free(item);
+	}
+	/* Free the transaction recover structure */
+	kmem_free(trans);
+}
+
+STATIC int
+xlog_recover_commit_pass1(
+	struct log		*log,
+	struct xlog_recover	*trans,
+	xlog_recover_item_t	*item)
+{
+	trace_xfs_log_recover_item_recover(log, trans, item, XLOG_RECOVER_PASS1);
+
+	switch (ITEM_TYPE(item)) {
+	case XFS_LI_BUF:
+		return xlog_recover_buffer_pass1(log, item);
+	case XFS_LI_QUOTAOFF:
+		return xlog_recover_quotaoff_pass1(log, item);
+	case XFS_LI_INODE:
+	case XFS_LI_EFI:
+	case XFS_LI_EFD:
+	case XFS_LI_DQUOT:
+		/* nothing to do in pass 1 */
+		return 0;
+	default:
+		xfs_warn(log->l_mp, "%s: invalid item type (%d)",
+			__func__, ITEM_TYPE(item));
+		ASSERT(0);
+		return XFS_ERROR(EIO);
+	}
+}
+
+STATIC int
+xlog_recover_commit_pass2(
+	struct log		*log,
+	struct xlog_recover	*trans,
+	xlog_recover_item_t	*item)
+{
+	trace_xfs_log_recover_item_recover(log, trans, item, XLOG_RECOVER_PASS2);
+
+	switch (ITEM_TYPE(item)) {
+	case XFS_LI_BUF:
+		return xlog_recover_buffer_pass2(log, item);
+	case XFS_LI_INODE:
+		return xlog_recover_inode_pass2(log, item);
+	case XFS_LI_EFI:
+		return xlog_recover_efi_pass2(log, item, trans->r_lsn);
+	case XFS_LI_EFD:
+		return xlog_recover_efd_pass2(log, item);
+	case XFS_LI_DQUOT:
+		return xlog_recover_dquot_pass2(log, item);
+	case XFS_LI_QUOTAOFF:
+		/* nothing to do in pass2 */
+		return 0;
+	default:
+		xfs_warn(log->l_mp, "%s: invalid item type (%d)",
+			__func__, ITEM_TYPE(item));
+		ASSERT(0);
+		return XFS_ERROR(EIO);
+	}
+}
+
+/*
+ * Perform the transaction.
+ *
+ * If the transaction modifies a buffer or inode, do it now.  Otherwise,
+ * EFIs and EFDs get queued up by adding entries into the AIL for them.
+ */
+STATIC int
+xlog_recover_commit_trans(
+	struct log		*log,
+	struct xlog_recover	*trans,
+	int			pass)
+{
+	int			error = 0;
+	xlog_recover_item_t	*item;
+
+	hlist_del(&trans->r_list);
+
+	error = xlog_recover_reorder_trans(log, trans, pass);
+	if (error)
+		return error;
+
+	list_for_each_entry(item, &trans->r_itemq, ri_list) {
+		if (pass == XLOG_RECOVER_PASS1)
+			error = xlog_recover_commit_pass1(log, trans, item);
+		else
+			error = xlog_recover_commit_pass2(log, trans, item);
+		if (error)
+			return error;
+	}
+
+	xlog_recover_free_trans(trans);
+	return 0;
+}
+
+STATIC int
+xlog_recover_unmount_trans(
+	struct log		*log,
+	xlog_recover_t		*trans)
+{
+	/* Do nothing now */
+	xfs_warn(log->l_mp, "%s: Unmount LR", __func__);
+	return 0;
+}
+
+/*
+ * There are two valid states of the r_state field.  0 indicates that the
+ * transaction structure is in a normal state.  We have either seen the
+ * start of the transaction or the last operation we added was not a partial
+ * operation.  If the last operation we added to the transaction was a
+ * partial operation, we need to mark r_state with XLOG_WAS_CONT_TRANS.
+ *
+ * NOTE: skip LRs with 0 data length.
+ */
+STATIC int
+xlog_recover_process_data(
+	xlog_t			*log,
+	struct hlist_head	rhash[],
+	xlog_rec_header_t	*rhead,
+	xfs_caddr_t		dp,
+	int			pass)
+{
+	xfs_caddr_t		lp;
+	int			num_logops;
+	xlog_op_header_t	*ohead;
+	xlog_recover_t		*trans;
+	xlog_tid_t		tid;
+	int			error;
+	unsigned long		hash;
+	uint			flags;
+
+	lp = dp + be32_to_cpu(rhead->h_len);
+	num_logops = be32_to_cpu(rhead->h_num_logops);
+
+	/* check the log format matches our own - else we can't recover */
+	if (xlog_header_check_recover(log->l_mp, rhead))
+		return (XFS_ERROR(EIO));
+
+	while ((dp < lp) && num_logops) {
+		ASSERT(dp + sizeof(xlog_op_header_t) <= lp);
+		ohead = (xlog_op_header_t *)dp;
+		dp += sizeof(xlog_op_header_t);
+		if (ohead->oh_clientid != XFS_TRANSACTION &&
+		    ohead->oh_clientid != XFS_LOG) {
+			xfs_warn(log->l_mp, "%s: bad clientid 0x%x",
+					__func__, ohead->oh_clientid);
+			ASSERT(0);
+			return (XFS_ERROR(EIO));
+		}
+		tid = be32_to_cpu(ohead->oh_tid);
+		hash = XLOG_RHASH(tid);
+		trans = xlog_recover_find_tid(&rhash[hash], tid);
+		if (trans == NULL) {		   /* not found; add new tid */
+			if (ohead->oh_flags & XLOG_START_TRANS)
+				xlog_recover_new_tid(&rhash[hash], tid,
+					be64_to_cpu(rhead->h_lsn));
+		} else {
+			if (dp + be32_to_cpu(ohead->oh_len) > lp) {
+				xfs_warn(log->l_mp, "%s: bad length 0x%x",
+					__func__, be32_to_cpu(ohead->oh_len));
+				WARN_ON(1);
+				return (XFS_ERROR(EIO));
+			}
+			flags = ohead->oh_flags & ~XLOG_END_TRANS;
+			if (flags & XLOG_WAS_CONT_TRANS)
+				flags &= ~XLOG_CONTINUE_TRANS;
+			switch (flags) {
+			case XLOG_COMMIT_TRANS:
+				error = xlog_recover_commit_trans(log,
+								trans, pass);
+				break;
+			case XLOG_UNMOUNT_TRANS:
+				error = xlog_recover_unmount_trans(log, trans);
+				break;
+			case XLOG_WAS_CONT_TRANS:
+				error = xlog_recover_add_to_cont_trans(log,
+						trans, dp,
+						be32_to_cpu(ohead->oh_len));
+				break;
+			case XLOG_START_TRANS:
+				xfs_warn(log->l_mp, "%s: bad transaction",
+					__func__);
+				ASSERT(0);
+				error = XFS_ERROR(EIO);
+				break;
+			case 0:
+			case XLOG_CONTINUE_TRANS:
+				error = xlog_recover_add_to_trans(log, trans,
+						dp, be32_to_cpu(ohead->oh_len));
+				break;
+			default:
+				xfs_warn(log->l_mp, "%s: bad flag 0x%x",
+					__func__, flags);
+				ASSERT(0);
+				error = XFS_ERROR(EIO);
+				break;
+			}
+			if (error)
+				return error;
+		}
+		dp += be32_to_cpu(ohead->oh_len);
+		num_logops--;
+	}
+	return 0;
+}
+
+/*
+ * Process an extent free intent item that was recovered from
+ * the log.  We need to free the extents that it describes.
+ */
+STATIC int
+xlog_recover_process_efi(
+	xfs_mount_t		*mp,
+	xfs_efi_log_item_t	*efip)
+{
+	xfs_efd_log_item_t	*efdp;
+	xfs_trans_t		*tp;
+	int			i;
+	int			error = 0;
+	xfs_extent_t		*extp;
+	xfs_fsblock_t		startblock_fsb;
+
+	ASSERT(!test_bit(XFS_EFI_RECOVERED, &efip->efi_flags));
+
+	/*
+	 * First check the validity of the extents described by the
+	 * EFI.  If any are bad, then assume that all are bad and
+	 * just toss the EFI.
+	 */
+	for (i = 0; i < efip->efi_format.efi_nextents; i++) {
+		extp = &(efip->efi_format.efi_extents[i]);
+		startblock_fsb = XFS_BB_TO_FSB(mp,
+				   XFS_FSB_TO_DADDR(mp, extp->ext_start));
+		if ((startblock_fsb == 0) ||
+		    (extp->ext_len == 0) ||
+		    (startblock_fsb >= mp->m_sb.sb_dblocks) ||
+		    (extp->ext_len >= mp->m_sb.sb_agblocks)) {
+			/*
+			 * This will pull the EFI from the AIL and
+			 * free the memory associated with it.
+			 */
+			xfs_efi_release(efip, efip->efi_format.efi_nextents);
+			return XFS_ERROR(EIO);
+		}
+	}
+
+	tp = xfs_trans_alloc(mp, 0);
+	error = xfs_trans_reserve(tp, 0, XFS_ITRUNCATE_LOG_RES(mp), 0, 0, 0);
+	if (error)
+		goto abort_error;
+	efdp = xfs_trans_get_efd(tp, efip, efip->efi_format.efi_nextents);
+
+	for (i = 0; i < efip->efi_format.efi_nextents; i++) {
+		extp = &(efip->efi_format.efi_extents[i]);
+		error = xfs_free_extent(tp, extp->ext_start, extp->ext_len);
+		if (error)
+			goto abort_error;
+		xfs_trans_log_efd_extent(tp, efdp, extp->ext_start,
+					 extp->ext_len);
+	}
+
+	set_bit(XFS_EFI_RECOVERED, &efip->efi_flags);
+	error = xfs_trans_commit(tp, 0);
+	return error;
+
+abort_error:
+	xfs_trans_cancel(tp, XFS_TRANS_ABORT);
+	return error;
+}
+
+/*
+ * When this is called, all of the EFIs which did not have
+ * corresponding EFDs should be in the AIL.  What we do now
+ * is free the extents associated with each one.
+ *
+ * Since we process the EFIs in normal transactions, they
+ * will be removed at some point after the commit.  This prevents
+ * us from just walking down the list processing each one.
+ * We'll use a flag in the EFI to skip those that we've already
+ * processed and use the AIL iteration mechanism's generation
+ * count to try to speed this up at least a bit.
+ *
+ * When we start, we know that the EFIs are the only things in
+ * the AIL.  As we process them, however, other items are added
+ * to the AIL.  Since everything added to the AIL must come after
+ * everything already in the AIL, we stop processing as soon as
+ * we see something other than an EFI in the AIL.
+ */
+STATIC int
+xlog_recover_process_efis(
+	xlog_t			*log)
+{
+	xfs_log_item_t		*lip;
+	xfs_efi_log_item_t	*efip;
+	int			error = 0;
+	struct xfs_ail_cursor	cur;
+	struct xfs_ail		*ailp;
+
+	ailp = log->l_ailp;
+	spin_lock(&ailp->xa_lock);
+	lip = xfs_trans_ail_cursor_first(ailp, &cur, 0);
+	while (lip != NULL) {
+		/*
+		 * We're done when we see something other than an EFI.
+		 * There should be no EFIs left in the AIL now.
+		 */
+		if (lip->li_type != XFS_LI_EFI) {
+#ifdef DEBUG
+			for (; lip; lip = xfs_trans_ail_cursor_next(ailp, &cur))
+				ASSERT(lip->li_type != XFS_LI_EFI);
+#endif
+			break;
+		}
+
+		/*
+		 * Skip EFIs that we've already processed.
+		 */
+		efip = (xfs_efi_log_item_t *)lip;
+		if (test_bit(XFS_EFI_RECOVERED, &efip->efi_flags)) {
+			lip = xfs_trans_ail_cursor_next(ailp, &cur);
+			continue;
+		}
+
+		spin_unlock(&ailp->xa_lock);
+		error = xlog_recover_process_efi(log->l_mp, efip);
+		spin_lock(&ailp->xa_lock);
+		if (error)
+			goto out;
+		lip = xfs_trans_ail_cursor_next(ailp, &cur);
+	}
+out:
+	xfs_trans_ail_cursor_done(ailp, &cur);
+	spin_unlock(&ailp->xa_lock);
+	return error;
+}
+
+/*
+ * This routine performs a transaction to null out a bad inode pointer
+ * in an agi unlinked inode hash bucket.
+ */
+STATIC void
+xlog_recover_clear_agi_bucket(
+	xfs_mount_t	*mp,
+	xfs_agnumber_t	agno,
+	int		bucket)
+{
+	xfs_trans_t	*tp;
+	xfs_agi_t	*agi;
+	xfs_buf_t	*agibp;
+	int		offset;
+	int		error;
+
+	tp = xfs_trans_alloc(mp, XFS_TRANS_CLEAR_AGI_BUCKET);
+	error = xfs_trans_reserve(tp, 0, XFS_CLEAR_AGI_BUCKET_LOG_RES(mp),
+				  0, 0, 0);
+	if (error)
+		goto out_abort;
+
+	error = xfs_read_agi(mp, tp, agno, &agibp);
+	if (error)
+		goto out_abort;
+
+	agi = XFS_BUF_TO_AGI(agibp);
+	agi->agi_unlinked[bucket] = cpu_to_be32(NULLAGINO);
+	offset = offsetof(xfs_agi_t, agi_unlinked) +
+		 (sizeof(xfs_agino_t) * bucket);
+	xfs_trans_log_buf(tp, agibp, offset,
+			  (offset + sizeof(xfs_agino_t) - 1));
+
+	error = xfs_trans_commit(tp, 0);
+	if (error)
+		goto out_error;
+	return;
+
+out_abort:
+	xfs_trans_cancel(tp, XFS_TRANS_ABORT);
+out_error:
+	xfs_warn(mp, "%s: failed to clear agi %d. Continuing.", __func__, agno);
+	return;
+}
+
+STATIC xfs_agino_t
+xlog_recover_process_one_iunlink(
+	struct xfs_mount		*mp,
+	xfs_agnumber_t			agno,
+	xfs_agino_t			agino,
+	int				bucket)
+{
+	struct xfs_buf			*ibp;
+	struct xfs_dinode		*dip;
+	struct xfs_inode		*ip;
+	xfs_ino_t			ino;
+	int				error;
+
+	ino = XFS_AGINO_TO_INO(mp, agno, agino);
+	error = xfs_iget(mp, NULL, ino, 0, 0, &ip);
+	if (error)
+		goto fail;
+
+	/*
+	 * Get the on disk inode to find the next inode in the bucket.
+	 */
+	error = xfs_itobp(mp, NULL, ip, &dip, &ibp, XBF_LOCK);
+	if (error)
+		goto fail_iput;
+
+	ASSERT(ip->i_d.di_nlink == 0);
+	ASSERT(ip->i_d.di_mode != 0);
+
+	/* setup for the next pass */
+	agino = be32_to_cpu(dip->di_next_unlinked);
+	xfs_buf_relse(ibp);
+
+	/*
+	 * Prevent any DMAPI event from being sent when the reference on
+	 * the inode is dropped.
+	 */
+	ip->i_d.di_dmevmask = 0;
+
+	IRELE(ip);
+	return agino;
+
+ fail_iput:
+	IRELE(ip);
+ fail:
+	/*
+	 * We can't read in the inode this bucket points to, or this inode
+	 * is messed up.  Just ditch this bucket of inodes.  We will lose
+	 * some inodes and space, but at least we won't hang.
+	 *
+	 * Call xlog_recover_clear_agi_bucket() to perform a transaction to
+	 * clear the inode pointer in the bucket.
+	 */
+	xlog_recover_clear_agi_bucket(mp, agno, bucket);
+	return NULLAGINO;
+}
+
+/*
+ * xlog_iunlink_recover
+ *
+ * This is called during recovery to process any inodes which
+ * we unlinked but not freed when the system crashed.  These
+ * inodes will be on the lists in the AGI blocks.  What we do
+ * here is scan all the AGIs and fully truncate and free any
+ * inodes found on the lists.  Each inode is removed from the
+ * lists when it has been fully truncated and is freed.  The
+ * freeing of the inode and its removal from the list must be
+ * atomic.
+ */
+STATIC void
+xlog_recover_process_iunlinks(
+	xlog_t		*log)
+{
+	xfs_mount_t	*mp;
+	xfs_agnumber_t	agno;
+	xfs_agi_t	*agi;
+	xfs_buf_t	*agibp;
+	xfs_agino_t	agino;
+	int		bucket;
+	int		error;
+	uint		mp_dmevmask;
+
+	mp = log->l_mp;
+
+	/*
+	 * Prevent any DMAPI event from being sent while in this function.
+	 */
+	mp_dmevmask = mp->m_dmevmask;
+	mp->m_dmevmask = 0;
+
+	for (agno = 0; agno < mp->m_sb.sb_agcount; agno++) {
+		/*
+		 * Find the agi for this ag.
+		 */
+		error = xfs_read_agi(mp, NULL, agno, &agibp);
+		if (error) {
+			/*
+			 * AGI is b0rked. Don't process it.
+			 *
+			 * We should probably mark the filesystem as corrupt
+			 * after we've recovered all the ag's we can....
+			 */
+			continue;
+		}
+		/*
+		 * Unlock the buffer so that it can be acquired in the normal
+		 * course of the transaction to truncate and free each inode.
+		 * Because we are not racing with anyone else here for the AGI
+		 * buffer, we don't even need to hold it locked to read the
+		 * initial unlinked bucket entries out of the buffer. We keep
+		 * buffer reference though, so that it stays pinned in memory
+		 * while we need the buffer.
+		 */
+		agi = XFS_BUF_TO_AGI(agibp);
+		xfs_buf_unlock(agibp);
+
+		for (bucket = 0; bucket < XFS_AGI_UNLINKED_BUCKETS; bucket++) {
+			agino = be32_to_cpu(agi->agi_unlinked[bucket]);
+			while (agino != NULLAGINO) {
+				agino = xlog_recover_process_one_iunlink(mp,
+							agno, agino, bucket);
+			}
+		}
+		xfs_buf_rele(agibp);
+	}
+
+	mp->m_dmevmask = mp_dmevmask;
+}
+
+
+#ifdef DEBUG
+STATIC void
+xlog_pack_data_checksum(
+	xlog_t		*log,
+	xlog_in_core_t	*iclog,
+	int		size)
+{
+	int		i;
+	__be32		*up;
+	uint		chksum = 0;
+
+	up = (__be32 *)iclog->ic_datap;
+	/* divide length by 4 to get # words */
+	for (i = 0; i < (size >> 2); i++) {
+		chksum ^= be32_to_cpu(*up);
+		up++;
+	}
+	iclog->ic_header.h_chksum = cpu_to_be32(chksum);
+}
+#else
+#define xlog_pack_data_checksum(log, iclog, size)
+#endif
+
+/*
+ * Stamp cycle number in every block
+ */
+void
+xlog_pack_data(
+	xlog_t			*log,
+	xlog_in_core_t		*iclog,
+	int			roundoff)
+{
+	int			i, j, k;
+	int			size = iclog->ic_offset + roundoff;
+	__be32			cycle_lsn;
+	xfs_caddr_t		dp;
+
+	xlog_pack_data_checksum(log, iclog, size);
+
+	cycle_lsn = CYCLE_LSN_DISK(iclog->ic_header.h_lsn);
+
+	dp = iclog->ic_datap;
+	for (i = 0; i < BTOBB(size) &&
+		i < (XLOG_HEADER_CYCLE_SIZE / BBSIZE); i++) {
+		iclog->ic_header.h_cycle_data[i] = *(__be32 *)dp;
+		*(__be32 *)dp = cycle_lsn;
+		dp += BBSIZE;
+	}
+
+	if (xfs_sb_version_haslogv2(&log->l_mp->m_sb)) {
+		xlog_in_core_2_t *xhdr = iclog->ic_data;
+
+		for ( ; i < BTOBB(size); i++) {
+			j = i / (XLOG_HEADER_CYCLE_SIZE / BBSIZE);
+			k = i % (XLOG_HEADER_CYCLE_SIZE / BBSIZE);
+			xhdr[j].hic_xheader.xh_cycle_data[k] = *(__be32 *)dp;
+			*(__be32 *)dp = cycle_lsn;
+			dp += BBSIZE;
+		}
+
+		for (i = 1; i < log->l_iclog_heads; i++) {
+			xhdr[i].hic_xheader.xh_cycle = cycle_lsn;
+		}
+	}
+}
+
+STATIC void
+xlog_unpack_data(
+	xlog_rec_header_t	*rhead,
+	xfs_caddr_t		dp,
+	xlog_t			*log)
+{
+	int			i, j, k;
+
+	for (i = 0; i < BTOBB(be32_to_cpu(rhead->h_len)) &&
+		  i < (XLOG_HEADER_CYCLE_SIZE / BBSIZE); i++) {
+		*(__be32 *)dp = *(__be32 *)&rhead->h_cycle_data[i];
+		dp += BBSIZE;
+	}
+
+	if (xfs_sb_version_haslogv2(&log->l_mp->m_sb)) {
+		xlog_in_core_2_t *xhdr = (xlog_in_core_2_t *)rhead;
+		for ( ; i < BTOBB(be32_to_cpu(rhead->h_len)); i++) {
+			j = i / (XLOG_HEADER_CYCLE_SIZE / BBSIZE);
+			k = i % (XLOG_HEADER_CYCLE_SIZE / BBSIZE);
+			*(__be32 *)dp = xhdr[j].hic_xheader.xh_cycle_data[k];
+			dp += BBSIZE;
+		}
+	}
+}
+
+STATIC int
+xlog_valid_rec_header(
+	xlog_t			*log,
+	xlog_rec_header_t	*rhead,
+	xfs_daddr_t		blkno)
+{
+	int			hlen;
+
+	if (unlikely(be32_to_cpu(rhead->h_magicno) != XLOG_HEADER_MAGIC_NUM)) {
+		XFS_ERROR_REPORT("xlog_valid_rec_header(1)",
+				XFS_ERRLEVEL_LOW, log->l_mp);
+		return XFS_ERROR(EFSCORRUPTED);
+	}
+	if (unlikely(
+	    (!rhead->h_version ||
+	    (be32_to_cpu(rhead->h_version) & (~XLOG_VERSION_OKBITS))))) {
+		xfs_warn(log->l_mp, "%s: unrecognised log version (%d).",
+			__func__, be32_to_cpu(rhead->h_version));
+		return XFS_ERROR(EIO);
+	}
+
+	/* LR body must have data or it wouldn't have been written */
+	hlen = be32_to_cpu(rhead->h_len);
+	if (unlikely( hlen <= 0 || hlen > INT_MAX )) {
+		XFS_ERROR_REPORT("xlog_valid_rec_header(2)",
+				XFS_ERRLEVEL_LOW, log->l_mp);
+		return XFS_ERROR(EFSCORRUPTED);
+	}
+	if (unlikely( blkno > log->l_logBBsize || blkno > INT_MAX )) {
+		XFS_ERROR_REPORT("xlog_valid_rec_header(3)",
+				XFS_ERRLEVEL_LOW, log->l_mp);
+		return XFS_ERROR(EFSCORRUPTED);
+	}
+	return 0;
+}
+
+/*
+ * Read the log from tail to head and process the log records found.
+ * Handle the two cases where the tail and head are in the same cycle
+ * and where the active portion of the log wraps around the end of
+ * the physical log separately.  The pass parameter is passed through
+ * to the routines called to process the data and is not looked at
+ * here.
+ */
+STATIC int
+xlog_do_recovery_pass(
+	xlog_t			*log,
+	xfs_daddr_t		head_blk,
+	xfs_daddr_t		tail_blk,
+	int			pass)
+{
+	xlog_rec_header_t	*rhead;
+	xfs_daddr_t		blk_no;
+	xfs_caddr_t		offset;
+	xfs_buf_t		*hbp, *dbp;
+	int			error = 0, h_size;
+	int			bblks, split_bblks;
+	int			hblks, split_hblks, wrapped_hblks;
+	struct hlist_head	rhash[XLOG_RHASH_SIZE];
+
+	ASSERT(head_blk != tail_blk);
+
+	/*
+	 * Read the header of the tail block and get the iclog buffer size from
+	 * h_size.  Use this to tell how many sectors make up the log header.
+	 */
+	if (xfs_sb_version_haslogv2(&log->l_mp->m_sb)) {
+		/*
+		 * When using variable length iclogs, read first sector of
+		 * iclog header and extract the header size from it.  Get a
+		 * new hbp that is the correct size.
+		 */
+		hbp = xlog_get_bp(log, 1);
+		if (!hbp)
+			return ENOMEM;
+
+		error = xlog_bread(log, tail_blk, 1, hbp, &offset);
+		if (error)
+			goto bread_err1;
+
+		rhead = (xlog_rec_header_t *)offset;
+		error = xlog_valid_rec_header(log, rhead, tail_blk);
+		if (error)
+			goto bread_err1;
+		h_size = be32_to_cpu(rhead->h_size);
+		if ((be32_to_cpu(rhead->h_version) & XLOG_VERSION_2) &&
+		    (h_size > XLOG_HEADER_CYCLE_SIZE)) {
+			hblks = h_size / XLOG_HEADER_CYCLE_SIZE;
+			if (h_size % XLOG_HEADER_CYCLE_SIZE)
+				hblks++;
+			xlog_put_bp(hbp);
+			hbp = xlog_get_bp(log, hblks);
+		} else {
+			hblks = 1;
+		}
+	} else {
+		ASSERT(log->l_sectBBsize == 1);
+		hblks = 1;
+		hbp = xlog_get_bp(log, 1);
+		h_size = XLOG_BIG_RECORD_BSIZE;
+	}
+
+	if (!hbp)
+		return ENOMEM;
+	dbp = xlog_get_bp(log, BTOBB(h_size));
+	if (!dbp) {
+		xlog_put_bp(hbp);
+		return ENOMEM;
+	}
+
+	memset(rhash, 0, sizeof(rhash));
+	if (tail_blk <= head_blk) {
+		for (blk_no = tail_blk; blk_no < head_blk; ) {
+			error = xlog_bread(log, blk_no, hblks, hbp, &offset);
+			if (error)
+				goto bread_err2;
+
+			rhead = (xlog_rec_header_t *)offset;
+			error = xlog_valid_rec_header(log, rhead, blk_no);
+			if (error)
+				goto bread_err2;
+
+			/* blocks in data section */
+			bblks = (int)BTOBB(be32_to_cpu(rhead->h_len));
+			error = xlog_bread(log, blk_no + hblks, bblks, dbp,
+					   &offset);
+			if (error)
+				goto bread_err2;
+
+			xlog_unpack_data(rhead, offset, log);
+			if ((error = xlog_recover_process_data(log,
+						rhash, rhead, offset, pass)))
+				goto bread_err2;
+			blk_no += bblks + hblks;
+		}
+	} else {
+		/*
+		 * Perform recovery around the end of the physical log.
+		 * When the head is not on the same cycle number as the tail,
+		 * we can't do a sequential recovery as above.
+		 */
+		blk_no = tail_blk;
+		while (blk_no < log->l_logBBsize) {
+			/*
+			 * Check for header wrapping around physical end-of-log
+			 */
+			offset = XFS_BUF_PTR(hbp);
+			split_hblks = 0;
+			wrapped_hblks = 0;
+			if (blk_no + hblks <= log->l_logBBsize) {
+				/* Read header in one read */
+				error = xlog_bread(log, blk_no, hblks, hbp,
+						   &offset);
+				if (error)
+					goto bread_err2;
+			} else {
+				/* This LR is split across physical log end */
+				if (blk_no != log->l_logBBsize) {
+					/* some data before physical log end */
+					ASSERT(blk_no <= INT_MAX);
+					split_hblks = log->l_logBBsize - (int)blk_no;
+					ASSERT(split_hblks > 0);
+					error = xlog_bread(log, blk_no,
+							   split_hblks, hbp,
+							   &offset);
+					if (error)
+						goto bread_err2;
+				}
+
+				/*
+				 * Note: this black magic still works with
+				 * large sector sizes (non-512) only because:
+				 * - we increased the buffer size originally
+				 *   by 1 sector giving us enough extra space
+				 *   for the second read;
+				 * - the log start is guaranteed to be sector
+				 *   aligned;
+				 * - we read the log end (LR header start)
+				 *   _first_, then the log start (LR header end)
+				 *   - order is important.
+				 */
+				wrapped_hblks = hblks - split_hblks;
+				error = xlog_bread_offset(log, 0,
+						wrapped_hblks, hbp,
+						offset + BBTOB(split_hblks));
+				if (error)
+					goto bread_err2;
+			}
+			rhead = (xlog_rec_header_t *)offset;
+			error = xlog_valid_rec_header(log, rhead,
+						split_hblks ? blk_no : 0);
+			if (error)
+				goto bread_err2;
+
+			bblks = (int)BTOBB(be32_to_cpu(rhead->h_len));
+			blk_no += hblks;
+
+			/* Read in data for log record */
+			if (blk_no + bblks <= log->l_logBBsize) {
+				error = xlog_bread(log, blk_no, bblks, dbp,
+						   &offset);
+				if (error)
+					goto bread_err2;
+			} else {
+				/* This log record is split across the
+				 * physical end of log */
+				offset = XFS_BUF_PTR(dbp);
+				split_bblks = 0;
+				if (blk_no != log->l_logBBsize) {
+					/* some data is before the physical
+					 * end of log */
+					ASSERT(!wrapped_hblks);
+					ASSERT(blk_no <= INT_MAX);
+					split_bblks =
+						log->l_logBBsize - (int)blk_no;
+					ASSERT(split_bblks > 0);
+					error = xlog_bread(log, blk_no,
+							split_bblks, dbp,
+							&offset);
+					if (error)
+						goto bread_err2;
+				}
+
+				/*
+				 * Note: this black magic still works with
+				 * large sector sizes (non-512) only because:
+				 * - we increased the buffer size originally
+				 *   by 1 sector giving us enough extra space
+				 *   for the second read;
+				 * - the log start is guaranteed to be sector
+				 *   aligned;
+				 * - we read the log end (LR header start)
+				 *   _first_, then the log start (LR header end)
+				 *   - order is important.
+				 */
+				error = xlog_bread_offset(log, 0,
+						bblks - split_bblks, hbp,
+						offset + BBTOB(split_bblks));
+				if (error)
+					goto bread_err2;
+			}
+			xlog_unpack_data(rhead, offset, log);
+			if ((error = xlog_recover_process_data(log, rhash,
+							rhead, offset, pass)))
+				goto bread_err2;
+			blk_no += bblks;
+		}
+
+		ASSERT(blk_no >= log->l_logBBsize);
+		blk_no -= log->l_logBBsize;
+
+		/* read first part of physical log */
+		while (blk_no < head_blk) {
+			error = xlog_bread(log, blk_no, hblks, hbp, &offset);
+			if (error)
+				goto bread_err2;
+
+			rhead = (xlog_rec_header_t *)offset;
+			error = xlog_valid_rec_header(log, rhead, blk_no);
+			if (error)
+				goto bread_err2;
+
+			bblks = (int)BTOBB(be32_to_cpu(rhead->h_len));
+			error = xlog_bread(log, blk_no+hblks, bblks, dbp,
+					   &offset);
+			if (error)
+				goto bread_err2;
+
+			xlog_unpack_data(rhead, offset, log);
+			if ((error = xlog_recover_process_data(log, rhash,
+							rhead, offset, pass)))
+				goto bread_err2;
+			blk_no += bblks + hblks;
+		}
+	}
+
+ bread_err2:
+	xlog_put_bp(dbp);
+ bread_err1:
+	xlog_put_bp(hbp);
+	return error;
+}
+
+/*
+ * Do the recovery of the log.  We actually do this in two phases.
+ * The two passes are necessary in order to implement the function
+ * of cancelling a record written into the log.  The first pass
+ * determines those things which have been cancelled, and the
+ * second pass replays log items normally except for those which
+ * have been cancelled.  The handling of the replay and cancellations
+ * takes place in the log item type specific routines.
+ *
+ * The table of items which have cancel records in the log is allocated
+ * and freed at this level, since only here do we know when all of
+ * the log recovery has been completed.
+ */
+STATIC int
+xlog_do_log_recovery(
+	xlog_t		*log,
+	xfs_daddr_t	head_blk,
+	xfs_daddr_t	tail_blk)
+{
+	int		error, i;
+
+	ASSERT(head_blk != tail_blk);
+
+	/*
+	 * First do a pass to find all of the cancelled buf log items.
+	 * Store them in the buf_cancel_table for use in the second pass.
+	 */
+	log->l_buf_cancel_table = kmem_zalloc(XLOG_BC_TABLE_SIZE *
+						 sizeof(struct list_head),
+						 KM_SLEEP);
+	for (i = 0; i < XLOG_BC_TABLE_SIZE; i++)
+		INIT_LIST_HEAD(&log->l_buf_cancel_table[i]);
+
+	error = xlog_do_recovery_pass(log, head_blk, tail_blk,
+				      XLOG_RECOVER_PASS1);
+	if (error != 0) {
+		kmem_free(log->l_buf_cancel_table);
+		log->l_buf_cancel_table = NULL;
+		return error;
+	}
+	/*
+	 * Then do a second pass to actually recover the items in the log.
+	 * When it is complete free the table of buf cancel items.
+	 */
+	error = xlog_do_recovery_pass(log, head_blk, tail_blk,
+				      XLOG_RECOVER_PASS2);
+#ifdef DEBUG
+	if (!error) {
+		int	i;
+
+		for (i = 0; i < XLOG_BC_TABLE_SIZE; i++)
+			ASSERT(list_empty(&log->l_buf_cancel_table[i]));
+	}
+#endif	/* DEBUG */
+
+	kmem_free(log->l_buf_cancel_table);
+	log->l_buf_cancel_table = NULL;
+
+	return error;
+}
+
+/*
+ * Do the actual recovery
+ */
+STATIC int
+xlog_do_recover(
+	xlog_t		*log,
+	xfs_daddr_t	head_blk,
+	xfs_daddr_t	tail_blk)
+{
+	int		error;
+	xfs_buf_t	*bp;
+	xfs_sb_t	*sbp;
+
+	/*
+	 * First replay the images in the log.
+	 */
+	error = xlog_do_log_recovery(log, head_blk, tail_blk);
+	if (error) {
+		return error;
+	}
+
+	XFS_bflush(log->l_mp->m_ddev_targp);
+
+	/*
+	 * If IO errors happened during recovery, bail out.
+	 */
+	if (XFS_FORCED_SHUTDOWN(log->l_mp)) {
+		return (EIO);
+	}
+
+	/*
+	 * We now update the tail_lsn since much of the recovery has completed
+	 * and there may be space available to use.  If there were no extent
+	 * or iunlinks, we can free up the entire log and set the tail_lsn to
+	 * be the last_sync_lsn.  This was set in xlog_find_tail to be the
+	 * lsn of the last known good LR on disk.  If there are extent frees
+	 * or iunlinks they will have some entries in the AIL; so we look at
+	 * the AIL to determine how to set the tail_lsn.
+	 */
+	xlog_assign_tail_lsn(log->l_mp);
+
+	/*
+	 * Now that we've finished replaying all buffer and inode
+	 * updates, re-read in the superblock.
+	 */
+	bp = xfs_getsb(log->l_mp, 0);
+	XFS_BUF_UNDONE(bp);
+	ASSERT(!(XFS_BUF_ISWRITE(bp)));
+	ASSERT(!(XFS_BUF_ISDELAYWRITE(bp)));
+	XFS_BUF_READ(bp);
+	XFS_BUF_UNASYNC(bp);
+	xfsbdstrat(log->l_mp, bp);
+	error = xfs_buf_iowait(bp);
+	if (error) {
+		xfs_ioerror_alert("xlog_do_recover",
+				  log->l_mp, bp, XFS_BUF_ADDR(bp));
+		ASSERT(0);
+		xfs_buf_relse(bp);
+		return error;
+	}
+
+	/* Convert superblock from on-disk format */
+	sbp = &log->l_mp->m_sb;
+	xfs_sb_from_disk(sbp, XFS_BUF_TO_SBP(bp));
+	ASSERT(sbp->sb_magicnum == XFS_SB_MAGIC);
+	ASSERT(xfs_sb_good_version(sbp));
+	xfs_buf_relse(bp);
+
+	/* We've re-read the superblock so re-initialize per-cpu counters */
+	xfs_icsb_reinit_counters(log->l_mp);
+
+	xlog_recover_check_summary(log);
+
+	/* Normal transactions can now occur */
+	log->l_flags &= ~XLOG_ACTIVE_RECOVERY;
+	return 0;
+}
+
+/*
+ * Perform recovery and re-initialize some log variables in xlog_find_tail.
+ *
+ * Return error or zero.
+ */
+int
+xlog_recover(
+	xlog_t		*log)
+{
+	xfs_daddr_t	head_blk, tail_blk;
+	int		error;
+
+	/* find the tail of the log */
+	if ((error = xlog_find_tail(log, &head_blk, &tail_blk)))
+		return error;
+
+	if (tail_blk != head_blk) {
+		/* There used to be a comment here:
+		 *
+		 * disallow recovery on read-only mounts.  note -- mount
+		 * checks for ENOSPC and turns it into an intelligent
+		 * error message.
+		 * ...but this is no longer true.  Now, unless you specify
+		 * NORECOVERY (in which case this function would never be
+		 * called), we just go ahead and recover.  We do this all
+		 * under the vfs layer, so we can get away with it unless
+		 * the device itself is read-only, in which case we fail.
+		 */
+		if ((error = xfs_dev_is_read_only(log->l_mp, "recovery"))) {
+			return error;
+		}
+
+		xfs_notice(log->l_mp, "Starting recovery (logdev: %s)",
+				log->l_mp->m_logname ? log->l_mp->m_logname
+						     : "internal");
+
+		error = xlog_do_recover(log, head_blk, tail_blk);
+		log->l_flags |= XLOG_RECOVERY_NEEDED;
+	}
+	return error;
+}
+
+/*
+ * In the first part of recovery we replay inodes and buffers and build
+ * up the list of extent free items which need to be processed.  Here
+ * we process the extent free items and clean up the on disk unlinked
+ * inode lists.  This is separated from the first part of recovery so
+ * that the root and real-time bitmap inodes can be read in from disk in
+ * between the two stages.  This is necessary so that we can free space
+ * in the real-time portion of the file system.
+ */
+int
+xlog_recover_finish(
+	xlog_t		*log)
+{
+	/*
+	 * Now we're ready to do the transactions needed for the
+	 * rest of recovery.  Start with completing all the extent
+	 * free intent records and then process the unlinked inode
+	 * lists.  At this point, we essentially run in normal mode
+	 * except that we're still performing recovery actions
+	 * rather than accepting new requests.
+	 */
+	if (log->l_flags & XLOG_RECOVERY_NEEDED) {
+		int	error;
+		error = xlog_recover_process_efis(log);
+		if (error) {
+			xfs_alert(log->l_mp, "Failed to recover EFIs");
+			return error;
+		}
+		/*
+		 * Sync the log to get all the EFIs out of the AIL.
+		 * This isn't absolutely necessary, but it helps in
+		 * case the unlink transactions would have problems
+		 * pushing the EFIs out of the way.
+		 */
+		xfs_log_force(log->l_mp, XFS_LOG_SYNC);
+
+		xlog_recover_process_iunlinks(log);
+
+		xlog_recover_check_summary(log);
+
+		xfs_notice(log->l_mp, "Ending recovery (logdev: %s)",
+				log->l_mp->m_logname ? log->l_mp->m_logname
+						     : "internal");
+		log->l_flags &= ~XLOG_RECOVERY_NEEDED;
+	} else {
+		xfs_info(log->l_mp, "Ending clean mount");
+	}
+	return 0;
+}
+
+
+#if defined(DEBUG)
+/*
+ * Read all of the agf and agi counters and check that they
+ * are consistent with the superblock counters.
+ */
+void
+xlog_recover_check_summary(
+	xlog_t		*log)
+{
+	xfs_mount_t	*mp;
+	xfs_agf_t	*agfp;
+	xfs_buf_t	*agfbp;
+	xfs_buf_t	*agibp;
+	xfs_agnumber_t	agno;
+	__uint64_t	freeblks;
+	__uint64_t	itotal;
+	__uint64_t	ifree;
+	int		error;
+
+	mp = log->l_mp;
+
+	freeblks = 0LL;
+	itotal = 0LL;
+	ifree = 0LL;
+	for (agno = 0; agno < mp->m_sb.sb_agcount; agno++) {
+		error = xfs_read_agf(mp, NULL, agno, 0, &agfbp);
+		if (error) {
+			xfs_alert(mp, "%s agf read failed agno %d error %d",
+						__func__, agno, error);
+		} else {
+			agfp = XFS_BUF_TO_AGF(agfbp);
+			freeblks += be32_to_cpu(agfp->agf_freeblks) +
+				    be32_to_cpu(agfp->agf_flcount);
+			xfs_buf_relse(agfbp);
+		}
+
+		error = xfs_read_agi(mp, NULL, agno, &agibp);
+		if (error) {
+			xfs_alert(mp, "%s agi read failed agno %d error %d",
+						__func__, agno, error);
+		} else {
+			struct xfs_agi	*agi = XFS_BUF_TO_AGI(agibp);
+
+			itotal += be32_to_cpu(agi->agi_count);
+			ifree += be32_to_cpu(agi->agi_freecount);
+			xfs_buf_relse(agibp);
+		}
+	}
+}
+#endif /* DEBUG */
diff --git a/fs/xfs/xfs_log_recover.h b/fs/xfs/xfs_log_recover.h
new file mode 100644
index 00000000..1c55ccbb
--- /dev/null
+++ b/fs/xfs/xfs_log_recover.h
@@ -0,0 +1,66 @@
+/*
+ * Copyright (c) 2000,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef	__XFS_LOG_RECOVER_H__
+#define __XFS_LOG_RECOVER_H__
+
+/*
+ * Macros, structures, prototypes for internal log manager use.
+ */
+
+#define XLOG_RHASH_BITS  4
+#define XLOG_RHASH_SIZE	16
+#define XLOG_RHASH_SHIFT 2
+#define XLOG_RHASH(tid)	\
+	((((__uint32_t)tid)>>XLOG_RHASH_SHIFT) & (XLOG_RHASH_SIZE-1))
+
+#define XLOG_MAX_REGIONS_IN_ITEM   (XFS_MAX_BLOCKSIZE / XFS_BLF_CHUNK / 2 + 1)
+
+
+/*
+ * item headers are in ri_buf[0].  Additional buffers follow.
+ */
+typedef struct xlog_recover_item {
+	struct list_head	ri_list;
+	int			ri_type;
+	int			ri_cnt;	/* count of regions found */
+	int			ri_total;	/* total regions */
+	xfs_log_iovec_t		*ri_buf;	/* ptr to regions buffer */
+} xlog_recover_item_t;
+
+struct xlog_tid;
+typedef struct xlog_recover {
+	struct hlist_node	r_list;
+	xlog_tid_t		r_log_tid;	/* log's transaction id */
+	xfs_trans_header_t	r_theader;	/* trans header for partial */
+	int			r_state;	/* not needed */
+	xfs_lsn_t		r_lsn;		/* xact lsn */
+	struct list_head	r_itemq;	/* q for items */
+} xlog_recover_t;
+
+#define ITEM_TYPE(i)	(*(ushort *)(i)->ri_buf[0].i_addr)
+
+/*
+ * This is the number of entries in the l_buf_cancel_table used during
+ * recovery.
+ */
+#define	XLOG_BC_TABLE_SIZE	64
+
+#define	XLOG_RECOVER_PASS1	1
+#define	XLOG_RECOVER_PASS2	2
+
+#endif	/* __XFS_LOG_RECOVER_H__ */
diff --git a/fs/xfs/xfs_mount.c b/fs/xfs/xfs_mount.c
new file mode 100644
index 00000000..9afdd497
--- /dev/null
+++ b/fs/xfs/xfs_mount.c
@@ -0,0 +1,2585 @@
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_dir2.h"
+#include "xfs_mount.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_alloc_btree.h"
+#include "xfs_ialloc_btree.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_btree.h"
+#include "xfs_ialloc.h"
+#include "xfs_alloc.h"
+#include "xfs_rtalloc.h"
+#include "xfs_bmap.h"
+#include "xfs_error.h"
+#include "xfs_rw.h"
+#include "xfs_quota.h"
+#include "xfs_fsops.h"
+#include "xfs_utils.h"
+#include "xfs_trace.h"
+
+
+#ifdef HAVE_PERCPU_SB
+STATIC void	xfs_icsb_balance_counter(xfs_mount_t *, xfs_sb_field_t,
+						int);
+STATIC void	xfs_icsb_balance_counter_locked(xfs_mount_t *, xfs_sb_field_t,
+						int);
+STATIC void	xfs_icsb_disable_counter(xfs_mount_t *, xfs_sb_field_t);
+#else
+
+#define xfs_icsb_balance_counter(mp, a, b)		do { } while (0)
+#define xfs_icsb_balance_counter_locked(mp, a, b)	do { } while (0)
+#endif
+
+static const struct {
+	short offset;
+	short type;	/* 0 = integer
+			 * 1 = binary / string (no translation)
+			 */
+} xfs_sb_info[] = {
+    { offsetof(xfs_sb_t, sb_magicnum),   0 },
+    { offsetof(xfs_sb_t, sb_blocksize),  0 },
+    { offsetof(xfs_sb_t, sb_dblocks),    0 },
+    { offsetof(xfs_sb_t, sb_rblocks),    0 },
+    { offsetof(xfs_sb_t, sb_rextents),   0 },
+    { offsetof(xfs_sb_t, sb_uuid),       1 },
+    { offsetof(xfs_sb_t, sb_logstart),   0 },
+    { offsetof(xfs_sb_t, sb_rootino),    0 },
+    { offsetof(xfs_sb_t, sb_rbmino),     0 },
+    { offsetof(xfs_sb_t, sb_rsumino),    0 },
+    { offsetof(xfs_sb_t, sb_rextsize),   0 },
+    { offsetof(xfs_sb_t, sb_agblocks),   0 },
+    { offsetof(xfs_sb_t, sb_agcount),    0 },
+    { offsetof(xfs_sb_t, sb_rbmblocks),  0 },
+    { offsetof(xfs_sb_t, sb_logblocks),  0 },
+    { offsetof(xfs_sb_t, sb_versionnum), 0 },
+    { offsetof(xfs_sb_t, sb_sectsize),   0 },
+    { offsetof(xfs_sb_t, sb_inodesize),  0 },
+    { offsetof(xfs_sb_t, sb_inopblock),  0 },
+    { offsetof(xfs_sb_t, sb_fname[0]),   1 },
+    { offsetof(xfs_sb_t, sb_blocklog),   0 },
+    { offsetof(xfs_sb_t, sb_sectlog),    0 },
+    { offsetof(xfs_sb_t, sb_inodelog),   0 },
+    { offsetof(xfs_sb_t, sb_inopblog),   0 },
+    { offsetof(xfs_sb_t, sb_agblklog),   0 },
+    { offsetof(xfs_sb_t, sb_rextslog),   0 },
+    { offsetof(xfs_sb_t, sb_inprogress), 0 },
+    { offsetof(xfs_sb_t, sb_imax_pct),   0 },
+    { offsetof(xfs_sb_t, sb_icount),     0 },
+    { offsetof(xfs_sb_t, sb_ifree),      0 },
+    { offsetof(xfs_sb_t, sb_fdblocks),   0 },
+    { offsetof(xfs_sb_t, sb_frextents),  0 },
+    { offsetof(xfs_sb_t, sb_uquotino),   0 },
+    { offsetof(xfs_sb_t, sb_gquotino),   0 },
+    { offsetof(xfs_sb_t, sb_qflags),     0 },
+    { offsetof(xfs_sb_t, sb_flags),      0 },
+    { offsetof(xfs_sb_t, sb_shared_vn),  0 },
+    { offsetof(xfs_sb_t, sb_inoalignmt), 0 },
+    { offsetof(xfs_sb_t, sb_unit),	 0 },
+    { offsetof(xfs_sb_t, sb_width),	 0 },
+    { offsetof(xfs_sb_t, sb_dirblklog),	 0 },
+    { offsetof(xfs_sb_t, sb_logsectlog), 0 },
+    { offsetof(xfs_sb_t, sb_logsectsize),0 },
+    { offsetof(xfs_sb_t, sb_logsunit),	 0 },
+    { offsetof(xfs_sb_t, sb_features2),	 0 },
+    { offsetof(xfs_sb_t, sb_bad_features2), 0 },
+    { sizeof(xfs_sb_t),			 0 }
+};
+
+static DEFINE_MUTEX(xfs_uuid_table_mutex);
+static int xfs_uuid_table_size;
+static uuid_t *xfs_uuid_table;
+
+/*
+ * See if the UUID is unique among mounted XFS filesystems.
+ * Mount fails if UUID is nil or a FS with the same UUID is already mounted.
+ */
+STATIC int
+xfs_uuid_mount(
+	struct xfs_mount	*mp)
+{
+	uuid_t			*uuid = &mp->m_sb.sb_uuid;
+	int			hole, i;
+
+	if (mp->m_flags & XFS_MOUNT_NOUUID)
+		return 0;
+
+	if (uuid_is_nil(uuid)) {
+		xfs_warn(mp, "Filesystem has nil UUID - can't mount");
+		return XFS_ERROR(EINVAL);
+	}
+
+	mutex_lock(&xfs_uuid_table_mutex);
+	for (i = 0, hole = -1; i < xfs_uuid_table_size; i++) {
+		if (uuid_is_nil(&xfs_uuid_table[i])) {
+			hole = i;
+			continue;
+		}
+		if (uuid_equal(uuid, &xfs_uuid_table[i]))
+			goto out_duplicate;
+	}
+
+	if (hole < 0) {
+		xfs_uuid_table = kmem_realloc(xfs_uuid_table,
+			(xfs_uuid_table_size + 1) * sizeof(*xfs_uuid_table),
+			xfs_uuid_table_size  * sizeof(*xfs_uuid_table),
+			KM_SLEEP);
+		hole = xfs_uuid_table_size++;
+	}
+	xfs_uuid_table[hole] = *uuid;
+	mutex_unlock(&xfs_uuid_table_mutex);
+
+	return 0;
+
+ out_duplicate:
+	mutex_unlock(&xfs_uuid_table_mutex);
+	xfs_warn(mp, "Filesystem has duplicate UUID - can't mount");
+	return XFS_ERROR(EINVAL);
+}
+
+STATIC void
+xfs_uuid_unmount(
+	struct xfs_mount	*mp)
+{
+	uuid_t			*uuid = &mp->m_sb.sb_uuid;
+	int			i;
+
+	if (mp->m_flags & XFS_MOUNT_NOUUID)
+		return;
+
+	mutex_lock(&xfs_uuid_table_mutex);
+	for (i = 0; i < xfs_uuid_table_size; i++) {
+		if (uuid_is_nil(&xfs_uuid_table[i]))
+			continue;
+		if (!uuid_equal(uuid, &xfs_uuid_table[i]))
+			continue;
+		memset(&xfs_uuid_table[i], 0, sizeof(uuid_t));
+		break;
+	}
+	ASSERT(i < xfs_uuid_table_size);
+	mutex_unlock(&xfs_uuid_table_mutex);
+}
+
+
+/*
+ * Reference counting access wrappers to the perag structures.
+ * Because we never free per-ag structures, the only thing we
+ * have to protect against changes is the tree structure itself.
+ */
+struct xfs_perag *
+xfs_perag_get(struct xfs_mount *mp, xfs_agnumber_t agno)
+{
+	struct xfs_perag	*pag;
+	int			ref = 0;
+
+	rcu_read_lock();
+	pag = radix_tree_lookup(&mp->m_perag_tree, agno);
+	if (pag) {
+		ASSERT(atomic_read(&pag->pag_ref) >= 0);
+		ref = atomic_inc_return(&pag->pag_ref);
+	}
+	rcu_read_unlock();
+	trace_xfs_perag_get(mp, agno, ref, _RET_IP_);
+	return pag;
+}
+
+/*
+ * search from @first to find the next perag with the given tag set.
+ */
+struct xfs_perag *
+xfs_perag_get_tag(
+	struct xfs_mount	*mp,
+	xfs_agnumber_t		first,
+	int			tag)
+{
+	struct xfs_perag	*pag;
+	int			found;
+	int			ref;
+
+	rcu_read_lock();
+	found = radix_tree_gang_lookup_tag(&mp->m_perag_tree,
+					(void **)&pag, first, 1, tag);
+	if (found <= 0) {
+		rcu_read_unlock();
+		return NULL;
+	}
+	ref = atomic_inc_return(&pag->pag_ref);
+	rcu_read_unlock();
+	trace_xfs_perag_get_tag(mp, pag->pag_agno, ref, _RET_IP_);
+	return pag;
+}
+
+void
+xfs_perag_put(struct xfs_perag *pag)
+{
+	int	ref;
+
+	ASSERT(atomic_read(&pag->pag_ref) > 0);
+	ref = atomic_dec_return(&pag->pag_ref);
+	trace_xfs_perag_put(pag->pag_mount, pag->pag_agno, ref, _RET_IP_);
+}
+
+STATIC void
+__xfs_free_perag(
+	struct rcu_head	*head)
+{
+	struct xfs_perag *pag = container_of(head, struct xfs_perag, rcu_head);
+
+	ASSERT(atomic_read(&pag->pag_ref) == 0);
+	kmem_free(pag);
+}
+
+/*
+ * Free up the per-ag resources associated with the mount structure.
+ */
+STATIC void
+xfs_free_perag(
+	xfs_mount_t	*mp)
+{
+	xfs_agnumber_t	agno;
+	struct xfs_perag *pag;
+
+	for (agno = 0; agno < mp->m_sb.sb_agcount; agno++) {
+		spin_lock(&mp->m_perag_lock);
+		pag = radix_tree_delete(&mp->m_perag_tree, agno);
+		spin_unlock(&mp->m_perag_lock);
+		ASSERT(pag);
+		ASSERT(atomic_read(&pag->pag_ref) == 0);
+		call_rcu(&pag->rcu_head, __xfs_free_perag);
+	}
+}
+
+/*
+ * Check size of device based on the (data/realtime) block count.
+ * Note: this check is used by the growfs code as well as mount.
+ */
+int
+xfs_sb_validate_fsb_count(
+	xfs_sb_t	*sbp,
+	__uint64_t	nblocks)
+{
+	ASSERT(PAGE_SHIFT >= sbp->sb_blocklog);
+	ASSERT(sbp->sb_blocklog >= BBSHIFT);
+
+#if XFS_BIG_BLKNOS     /* Limited by ULONG_MAX of page cache index */
+	if (nblocks >> (PAGE_CACHE_SHIFT - sbp->sb_blocklog) > ULONG_MAX)
+		return EFBIG;
+#else                  /* Limited by UINT_MAX of sectors */
+	if (nblocks << (sbp->sb_blocklog - BBSHIFT) > UINT_MAX)
+		return EFBIG;
+#endif
+	return 0;
+}
+
+/*
+ * Check the validity of the SB found.
+ */
+STATIC int
+xfs_mount_validate_sb(
+	xfs_mount_t	*mp,
+	xfs_sb_t	*sbp,
+	int		flags)
+{
+	int		loud = !(flags & XFS_MFSI_QUIET);
+
+	/*
+	 * If the log device and data device have the
+	 * same device number, the log is internal.
+	 * Consequently, the sb_logstart should be non-zero.  If
+	 * we have a zero sb_logstart in this case, we may be trying to mount
+	 * a volume filesystem in a non-volume manner.
+	 */
+	if (sbp->sb_magicnum != XFS_SB_MAGIC) {
+		if (loud)
+			xfs_warn(mp, "bad magic number");
+		return XFS_ERROR(EWRONGFS);
+	}
+
+	if (!xfs_sb_good_version(sbp)) {
+		if (loud)
+			xfs_warn(mp, "bad version");
+		return XFS_ERROR(EWRONGFS);
+	}
+
+	if (unlikely(
+	    sbp->sb_logstart == 0 && mp->m_logdev_targp == mp->m_ddev_targp)) {
+		if (loud)
+			xfs_warn(mp,
+		"filesystem is marked as having an external log; "
+		"specify logdev on the mount command line.");
+		return XFS_ERROR(EINVAL);
+	}
+
+	if (unlikely(
+	    sbp->sb_logstart != 0 && mp->m_logdev_targp != mp->m_ddev_targp)) {
+		if (loud)
+			xfs_warn(mp,
+		"filesystem is marked as having an internal log; "
+		"do not specify logdev on the mount command line.");
+		return XFS_ERROR(EINVAL);
+	}
+
+	/*
+	 * More sanity checking. These were stolen directly from
+	 * xfs_repair.
+	 */
+	if (unlikely(
+	    sbp->sb_agcount <= 0					||
+	    sbp->sb_sectsize < XFS_MIN_SECTORSIZE			||
+	    sbp->sb_sectsize > XFS_MAX_SECTORSIZE			||
+	    sbp->sb_sectlog < XFS_MIN_SECTORSIZE_LOG			||
+	    sbp->sb_sectlog > XFS_MAX_SECTORSIZE_LOG			||
+	    sbp->sb_sectsize != (1 << sbp->sb_sectlog)			||
+	    sbp->sb_blocksize < XFS_MIN_BLOCKSIZE			||
+	    sbp->sb_blocksize > XFS_MAX_BLOCKSIZE			||
+	    sbp->sb_blocklog < XFS_MIN_BLOCKSIZE_LOG			||
+	    sbp->sb_blocklog > XFS_MAX_BLOCKSIZE_LOG			||
+	    sbp->sb_blocksize != (1 << sbp->sb_blocklog)		||
+	    sbp->sb_inodesize < XFS_DINODE_MIN_SIZE			||
+	    sbp->sb_inodesize > XFS_DINODE_MAX_SIZE			||
+	    sbp->sb_inodelog < XFS_DINODE_MIN_LOG			||
+	    sbp->sb_inodelog > XFS_DINODE_MAX_LOG			||
+	    sbp->sb_inodesize != (1 << sbp->sb_inodelog)		||
+	    (sbp->sb_blocklog - sbp->sb_inodelog != sbp->sb_inopblog)	||
+	    (sbp->sb_rextsize * sbp->sb_blocksize > XFS_MAX_RTEXTSIZE)	||
+	    (sbp->sb_rextsize * sbp->sb_blocksize < XFS_MIN_RTEXTSIZE)	||
+	    (sbp->sb_imax_pct > 100 /* zero sb_imax_pct is valid */))) {
+		if (loud)
+			xfs_warn(mp, "SB sanity check 1 failed");
+		return XFS_ERROR(EFSCORRUPTED);
+	}
+
+	/*
+	 * Sanity check AG count, size fields against data size field
+	 */
+	if (unlikely(
+	    sbp->sb_dblocks == 0 ||
+	    sbp->sb_dblocks >
+	     (xfs_drfsbno_t)sbp->sb_agcount * sbp->sb_agblocks ||
+	    sbp->sb_dblocks < (xfs_drfsbno_t)(sbp->sb_agcount - 1) *
+			      sbp->sb_agblocks + XFS_MIN_AG_BLOCKS)) {
+		if (loud)
+			xfs_warn(mp, "SB sanity check 2 failed");
+		return XFS_ERROR(EFSCORRUPTED);
+	}
+
+	/*
+	 * Until this is fixed only page-sized or smaller data blocks work.
+	 */
+	if (unlikely(sbp->sb_blocksize > PAGE_SIZE)) {
+		if (loud) {
+			xfs_warn(mp,
+		"File system with blocksize %d bytes. "
+		"Only pagesize (%ld) or less will currently work.",
+				sbp->sb_blocksize, PAGE_SIZE);
+		}
+		return XFS_ERROR(ENOSYS);
+	}
+
+	/*
+	 * Currently only very few inode sizes are supported.
+	 */
+	switch (sbp->sb_inodesize) {
+	case 256:
+	case 512:
+	case 1024:
+	case 2048:
+		break;
+	default:
+		if (loud)
+			xfs_warn(mp, "inode size of %d bytes not supported",
+				sbp->sb_inodesize);
+		return XFS_ERROR(ENOSYS);
+	}
+
+	if (xfs_sb_validate_fsb_count(sbp, sbp->sb_dblocks) ||
+	    xfs_sb_validate_fsb_count(sbp, sbp->sb_rblocks)) {
+		if (loud)
+			xfs_warn(mp,
+		"file system too large to be mounted on this system.");
+		return XFS_ERROR(EFBIG);
+	}
+
+	if (unlikely(sbp->sb_inprogress)) {
+		if (loud)
+			xfs_warn(mp, "file system busy");
+		return XFS_ERROR(EFSCORRUPTED);
+	}
+
+	/*
+	 * Version 1 directory format has never worked on Linux.
+	 */
+	if (unlikely(!xfs_sb_version_hasdirv2(sbp))) {
+		if (loud)
+			xfs_warn(mp,
+				"file system using version 1 directory format");
+		return XFS_ERROR(ENOSYS);
+	}
+
+	return 0;
+}
+
+int
+xfs_initialize_perag(
+	xfs_mount_t	*mp,
+	xfs_agnumber_t	agcount,
+	xfs_agnumber_t	*maxagi)
+{
+	xfs_agnumber_t	index, max_metadata;
+	xfs_agnumber_t	first_initialised = 0;
+	xfs_perag_t	*pag;
+	xfs_agino_t	agino;
+	xfs_ino_t	ino;
+	xfs_sb_t	*sbp = &mp->m_sb;
+	int		error = -ENOMEM;
+
+	/*
+	 * Walk the current per-ag tree so we don't try to initialise AGs
+	 * that already exist (growfs case). Allocate and insert all the
+	 * AGs we don't find ready for initialisation.
+	 */
+	for (index = 0; index < agcount; index++) {
+		pag = xfs_perag_get(mp, index);
+		if (pag) {
+			xfs_perag_put(pag);
+			continue;
+		}
+		if (!first_initialised)
+			first_initialised = index;
+
+		pag = kmem_zalloc(sizeof(*pag), KM_MAYFAIL);
+		if (!pag)
+			goto out_unwind;
+		pag->pag_agno = index;
+		pag->pag_mount = mp;
+		spin_lock_init(&pag->pag_ici_lock);
+		mutex_init(&pag->pag_ici_reclaim_lock);
+		INIT_RADIX_TREE(&pag->pag_ici_root, GFP_ATOMIC);
+		spin_lock_init(&pag->pag_buf_lock);
+		pag->pag_buf_tree = RB_ROOT;
+
+		if (radix_tree_preload(GFP_NOFS))
+			goto out_unwind;
+
+		spin_lock(&mp->m_perag_lock);
+		if (radix_tree_insert(&mp->m_perag_tree, index, pag)) {
+			BUG();
+			spin_unlock(&mp->m_perag_lock);
+			radix_tree_preload_end();
+			error = -EEXIST;
+			goto out_unwind;
+		}
+		spin_unlock(&mp->m_perag_lock);
+		radix_tree_preload_end();
+	}
+
+	/*
+	 * If we mount with the inode64 option, or no inode overflows
+	 * the legacy 32-bit address space clear the inode32 option.
+	 */
+	agino = XFS_OFFBNO_TO_AGINO(mp, sbp->sb_agblocks - 1, 0);
+	ino = XFS_AGINO_TO_INO(mp, agcount - 1, agino);
+
+	if ((mp->m_flags & XFS_MOUNT_SMALL_INUMS) && ino > XFS_MAXINUMBER_32)
+		mp->m_flags |= XFS_MOUNT_32BITINODES;
+	else
+		mp->m_flags &= ~XFS_MOUNT_32BITINODES;
+
+	if (mp->m_flags & XFS_MOUNT_32BITINODES) {
+		/*
+		 * Calculate how much should be reserved for inodes to meet
+		 * the max inode percentage.
+		 */
+		if (mp->m_maxicount) {
+			__uint64_t	icount;
+
+			icount = sbp->sb_dblocks * sbp->sb_imax_pct;
+			do_div(icount, 100);
+			icount += sbp->sb_agblocks - 1;
+			do_div(icount, sbp->sb_agblocks);
+			max_metadata = icount;
+		} else {
+			max_metadata = agcount;
+		}
+
+		for (index = 0; index < agcount; index++) {
+			ino = XFS_AGINO_TO_INO(mp, index, agino);
+			if (ino > XFS_MAXINUMBER_32) {
+				index++;
+				break;
+			}
+
+			pag = xfs_perag_get(mp, index);
+			pag->pagi_inodeok = 1;
+			if (index < max_metadata)
+				pag->pagf_metadata = 1;
+			xfs_perag_put(pag);
+		}
+	} else {
+		for (index = 0; index < agcount; index++) {
+			pag = xfs_perag_get(mp, index);
+			pag->pagi_inodeok = 1;
+			xfs_perag_put(pag);
+		}
+	}
+
+	if (maxagi)
+		*maxagi = index;
+	return 0;
+
+out_unwind:
+	kmem_free(pag);
+	for (; index > first_initialised; index--) {
+		pag = radix_tree_delete(&mp->m_perag_tree, index);
+		kmem_free(pag);
+	}
+	return error;
+}
+
+void
+xfs_sb_from_disk(
+	xfs_sb_t	*to,
+	xfs_dsb_t	*from)
+{
+	to->sb_magicnum = be32_to_cpu(from->sb_magicnum);
+	to->sb_blocksize = be32_to_cpu(from->sb_blocksize);
+	to->sb_dblocks = be64_to_cpu(from->sb_dblocks);
+	to->sb_rblocks = be64_to_cpu(from->sb_rblocks);
+	to->sb_rextents = be64_to_cpu(from->sb_rextents);
+	memcpy(&to->sb_uuid, &from->sb_uuid, sizeof(to->sb_uuid));
+	to->sb_logstart = be64_to_cpu(from->sb_logstart);
+	to->sb_rootino = be64_to_cpu(from->sb_rootino);
+	to->sb_rbmino = be64_to_cpu(from->sb_rbmino);
+	to->sb_rsumino = be64_to_cpu(from->sb_rsumino);
+	to->sb_rextsize = be32_to_cpu(from->sb_rextsize);
+	to->sb_agblocks = be32_to_cpu(from->sb_agblocks);
+	to->sb_agcount = be32_to_cpu(from->sb_agcount);
+	to->sb_rbmblocks = be32_to_cpu(from->sb_rbmblocks);
+	to->sb_logblocks = be32_to_cpu(from->sb_logblocks);
+	to->sb_versionnum = be16_to_cpu(from->sb_versionnum);
+	to->sb_sectsize = be16_to_cpu(from->sb_sectsize);
+	to->sb_inodesize = be16_to_cpu(from->sb_inodesize);
+	to->sb_inopblock = be16_to_cpu(from->sb_inopblock);
+	memcpy(&to->sb_fname, &from->sb_fname, sizeof(to->sb_fname));
+	to->sb_blocklog = from->sb_blocklog;
+	to->sb_sectlog = from->sb_sectlog;
+	to->sb_inodelog = from->sb_inodelog;
+	to->sb_inopblog = from->sb_inopblog;
+	to->sb_agblklog = from->sb_agblklog;
+	to->sb_rextslog = from->sb_rextslog;
+	to->sb_inprogress = from->sb_inprogress;
+	to->sb_imax_pct = from->sb_imax_pct;
+	to->sb_icount = be64_to_cpu(from->sb_icount);
+	to->sb_ifree = be64_to_cpu(from->sb_ifree);
+	to->sb_fdblocks = be64_to_cpu(from->sb_fdblocks);
+	to->sb_frextents = be64_to_cpu(from->sb_frextents);
+	to->sb_uquotino = be64_to_cpu(from->sb_uquotino);
+	to->sb_gquotino = be64_to_cpu(from->sb_gquotino);
+	to->sb_qflags = be16_to_cpu(from->sb_qflags);
+	to->sb_flags = from->sb_flags;
+	to->sb_shared_vn = from->sb_shared_vn;
+	to->sb_inoalignmt = be32_to_cpu(from->sb_inoalignmt);
+	to->sb_unit = be32_to_cpu(from->sb_unit);
+	to->sb_width = be32_to_cpu(from->sb_width);
+	to->sb_dirblklog = from->sb_dirblklog;
+	to->sb_logsectlog = from->sb_logsectlog;
+	to->sb_logsectsize = be16_to_cpu(from->sb_logsectsize);
+	to->sb_logsunit = be32_to_cpu(from->sb_logsunit);
+	to->sb_features2 = be32_to_cpu(from->sb_features2);
+	to->sb_bad_features2 = be32_to_cpu(from->sb_bad_features2);
+}
+
+/*
+ * Copy in core superblock to ondisk one.
+ *
+ * The fields argument is mask of superblock fields to copy.
+ */
+void
+xfs_sb_to_disk(
+	xfs_dsb_t	*to,
+	xfs_sb_t	*from,
+	__int64_t	fields)
+{
+	xfs_caddr_t	to_ptr = (xfs_caddr_t)to;
+	xfs_caddr_t	from_ptr = (xfs_caddr_t)from;
+	xfs_sb_field_t	f;
+	int		first;
+	int		size;
+
+	ASSERT(fields);
+	if (!fields)
+		return;
+
+	while (fields) {
+		f = (xfs_sb_field_t)xfs_lowbit64((__uint64_t)fields);
+		first = xfs_sb_info[f].offset;
+		size = xfs_sb_info[f + 1].offset - first;
+
+		ASSERT(xfs_sb_info[f].type == 0 || xfs_sb_info[f].type == 1);
+
+		if (size == 1 || xfs_sb_info[f].type == 1) {
+			memcpy(to_ptr + first, from_ptr + first, size);
+		} else {
+			switch (size) {
+			case 2:
+				*(__be16 *)(to_ptr + first) =
+					cpu_to_be16(*(__u16 *)(from_ptr + first));
+				break;
+			case 4:
+				*(__be32 *)(to_ptr + first) =
+					cpu_to_be32(*(__u32 *)(from_ptr + first));
+				break;
+			case 8:
+				*(__be64 *)(to_ptr + first) =
+					cpu_to_be64(*(__u64 *)(from_ptr + first));
+				break;
+			default:
+				ASSERT(0);
+			}
+		}
+
+		fields &= ~(1LL << f);
+	}
+}
+
+/*
+ * xfs_readsb
+ *
+ * Does the initial read of the superblock.
+ */
+int
+xfs_readsb(xfs_mount_t *mp, int flags)
+{
+	unsigned int	sector_size;
+	xfs_buf_t	*bp;
+	int		error;
+	int		loud = !(flags & XFS_MFSI_QUIET);
+
+	ASSERT(mp->m_sb_bp == NULL);
+	ASSERT(mp->m_ddev_targp != NULL);
+
+	/*
+	 * Allocate a (locked) buffer to hold the superblock.
+	 * This will be kept around at all times to optimize
+	 * access to the superblock.
+	 */
+	sector_size = xfs_getsize_buftarg(mp->m_ddev_targp);
+
+reread:
+	bp = xfs_buf_read_uncached(mp, mp->m_ddev_targp,
+					XFS_SB_DADDR, sector_size, 0);
+	if (!bp) {
+		if (loud)
+			xfs_warn(mp, "SB buffer read failed");
+		return EIO;
+	}
+
+	/*
+	 * Initialize the mount structure from the superblock.
+	 * But first do some basic consistency checking.
+	 */
+	xfs_sb_from_disk(&mp->m_sb, XFS_BUF_TO_SBP(bp));
+	error = xfs_mount_validate_sb(mp, &(mp->m_sb), flags);
+	if (error) {
+		if (loud)
+			xfs_warn(mp, "SB validate failed");
+		goto release_buf;
+	}
+
+	/*
+	 * We must be able to do sector-sized and sector-aligned IO.
+	 */
+	if (sector_size > mp->m_sb.sb_sectsize) {
+		if (loud)
+			xfs_warn(mp, "device supports %u byte sectors (not %u)",
+				sector_size, mp->m_sb.sb_sectsize);
+		error = ENOSYS;
+		goto release_buf;
+	}
+
+	/*
+	 * If device sector size is smaller than the superblock size,
+	 * re-read the superblock so the buffer is correctly sized.
+	 */
+	if (sector_size < mp->m_sb.sb_sectsize) {
+		xfs_buf_relse(bp);
+		sector_size = mp->m_sb.sb_sectsize;
+		goto reread;
+	}
+
+	/* Initialize per-cpu counters */
+	xfs_icsb_reinit_counters(mp);
+
+	mp->m_sb_bp = bp;
+	xfs_buf_unlock(bp);
+	return 0;
+
+release_buf:
+	xfs_buf_relse(bp);
+	return error;
+}
+
+
+/*
+ * xfs_mount_common
+ *
+ * Mount initialization code establishing various mount
+ * fields from the superblock associated with the given
+ * mount structure
+ */
+STATIC void
+xfs_mount_common(xfs_mount_t *mp, xfs_sb_t *sbp)
+{
+	mp->m_agfrotor = mp->m_agirotor = 0;
+	spin_lock_init(&mp->m_agirotor_lock);
+	mp->m_maxagi = mp->m_sb.sb_agcount;
+	mp->m_blkbit_log = sbp->sb_blocklog + XFS_NBBYLOG;
+	mp->m_blkbb_log = sbp->sb_blocklog - BBSHIFT;
+	mp->m_sectbb_log = sbp->sb_sectlog - BBSHIFT;
+	mp->m_agno_log = xfs_highbit32(sbp->sb_agcount - 1) + 1;
+	mp->m_agino_log = sbp->sb_inopblog + sbp->sb_agblklog;
+	mp->m_blockmask = sbp->sb_blocksize - 1;
+	mp->m_blockwsize = sbp->sb_blocksize >> XFS_WORDLOG;
+	mp->m_blockwmask = mp->m_blockwsize - 1;
+
+	mp->m_alloc_mxr[0] = xfs_allocbt_maxrecs(mp, sbp->sb_blocksize, 1);
+	mp->m_alloc_mxr[1] = xfs_allocbt_maxrecs(mp, sbp->sb_blocksize, 0);
+	mp->m_alloc_mnr[0] = mp->m_alloc_mxr[0] / 2;
+	mp->m_alloc_mnr[1] = mp->m_alloc_mxr[1] / 2;
+
+	mp->m_inobt_mxr[0] = xfs_inobt_maxrecs(mp, sbp->sb_blocksize, 1);
+	mp->m_inobt_mxr[1] = xfs_inobt_maxrecs(mp, sbp->sb_blocksize, 0);
+	mp->m_inobt_mnr[0] = mp->m_inobt_mxr[0] / 2;
+	mp->m_inobt_mnr[1] = mp->m_inobt_mxr[1] / 2;
+
+	mp->m_bmap_dmxr[0] = xfs_bmbt_maxrecs(mp, sbp->sb_blocksize, 1);
+	mp->m_bmap_dmxr[1] = xfs_bmbt_maxrecs(mp, sbp->sb_blocksize, 0);
+	mp->m_bmap_dmnr[0] = mp->m_bmap_dmxr[0] / 2;
+	mp->m_bmap_dmnr[1] = mp->m_bmap_dmxr[1] / 2;
+
+	mp->m_bsize = XFS_FSB_TO_BB(mp, 1);
+	mp->m_ialloc_inos = (int)MAX((__uint16_t)XFS_INODES_PER_CHUNK,
+					sbp->sb_inopblock);
+	mp->m_ialloc_blks = mp->m_ialloc_inos >> sbp->sb_inopblog;
+}
+
+/*
+ * xfs_initialize_perag_data
+ *
+ * Read in each per-ag structure so we can count up the number of
+ * allocated inodes, free inodes and used filesystem blocks as this
+ * information is no longer persistent in the superblock. Once we have
+ * this information, write it into the in-core superblock structure.
+ */
+STATIC int
+xfs_initialize_perag_data(xfs_mount_t *mp, xfs_agnumber_t agcount)
+{
+	xfs_agnumber_t	index;
+	xfs_perag_t	*pag;
+	xfs_sb_t	*sbp = &mp->m_sb;
+	uint64_t	ifree = 0;
+	uint64_t	ialloc = 0;
+	uint64_t	bfree = 0;
+	uint64_t	bfreelst = 0;
+	uint64_t	btree = 0;
+	int		error;
+
+	for (index = 0; index < agcount; index++) {
+		/*
+		 * read the agf, then the agi. This gets us
+		 * all the information we need and populates the
+		 * per-ag structures for us.
+		 */
+		error = xfs_alloc_pagf_init(mp, NULL, index, 0);
+		if (error)
+			return error;
+
+		error = xfs_ialloc_pagi_init(mp, NULL, index);
+		if (error)
+			return error;
+		pag = xfs_perag_get(mp, index);
+		ifree += pag->pagi_freecount;
+		ialloc += pag->pagi_count;
+		bfree += pag->pagf_freeblks;
+		bfreelst += pag->pagf_flcount;
+		btree += pag->pagf_btreeblks;
+		xfs_perag_put(pag);
+	}
+	/*
+	 * Overwrite incore superblock counters with just-read data
+	 */
+	spin_lock(&mp->m_sb_lock);
+	sbp->sb_ifree = ifree;
+	sbp->sb_icount = ialloc;
+	sbp->sb_fdblocks = bfree + bfreelst + btree;
+	spin_unlock(&mp->m_sb_lock);
+
+	/* Fixup the per-cpu counters as well. */
+	xfs_icsb_reinit_counters(mp);
+
+	return 0;
+}
+
+/*
+ * Update alignment values based on mount options and sb values
+ */
+STATIC int
+xfs_update_alignment(xfs_mount_t *mp)
+{
+	xfs_sb_t	*sbp = &(mp->m_sb);
+
+	if (mp->m_dalign) {
+		/*
+		 * If stripe unit and stripe width are not multiples
+		 * of the fs blocksize turn off alignment.
+		 */
+		if ((BBTOB(mp->m_dalign) & mp->m_blockmask) ||
+		    (BBTOB(mp->m_swidth) & mp->m_blockmask)) {
+			if (mp->m_flags & XFS_MOUNT_RETERR) {
+				xfs_warn(mp, "alignment check 1 failed");
+				return XFS_ERROR(EINVAL);
+			}
+			mp->m_dalign = mp->m_swidth = 0;
+		} else {
+			/*
+			 * Convert the stripe unit and width to FSBs.
+			 */
+			mp->m_dalign = XFS_BB_TO_FSBT(mp, mp->m_dalign);
+			if (mp->m_dalign && (sbp->sb_agblocks % mp->m_dalign)) {
+				if (mp->m_flags & XFS_MOUNT_RETERR) {
+					return XFS_ERROR(EINVAL);
+				}
+				xfs_warn(mp,
+		"stripe alignment turned off: sunit(%d)/swidth(%d) "
+		"incompatible with agsize(%d)",
+					mp->m_dalign, mp->m_swidth,
+					sbp->sb_agblocks);
+
+				mp->m_dalign = 0;
+				mp->m_swidth = 0;
+			} else if (mp->m_dalign) {
+				mp->m_swidth = XFS_BB_TO_FSBT(mp, mp->m_swidth);
+			} else {
+				if (mp->m_flags & XFS_MOUNT_RETERR) {
+					xfs_warn(mp,
+		"stripe alignment turned off: sunit(%d) less than bsize(%d)",
+						mp->m_dalign,
+						mp->m_blockmask +1);
+					return XFS_ERROR(EINVAL);
+				}
+				mp->m_swidth = 0;
+			}
+		}
+
+		/*
+		 * Update superblock with new values
+		 * and log changes
+		 */
+		if (xfs_sb_version_hasdalign(sbp)) {
+			if (sbp->sb_unit != mp->m_dalign) {
+				sbp->sb_unit = mp->m_dalign;
+				mp->m_update_flags |= XFS_SB_UNIT;
+			}
+			if (sbp->sb_width != mp->m_swidth) {
+				sbp->sb_width = mp->m_swidth;
+				mp->m_update_flags |= XFS_SB_WIDTH;
+			}
+		}
+	} else if ((mp->m_flags & XFS_MOUNT_NOALIGN) != XFS_MOUNT_NOALIGN &&
+		    xfs_sb_version_hasdalign(&mp->m_sb)) {
+			mp->m_dalign = sbp->sb_unit;
+			mp->m_swidth = sbp->sb_width;
+	}
+
+	return 0;
+}
+
+/*
+ * Set the maximum inode count for this filesystem
+ */
+STATIC void
+xfs_set_maxicount(xfs_mount_t *mp)
+{
+	xfs_sb_t	*sbp = &(mp->m_sb);
+	__uint64_t	icount;
+
+	if (sbp->sb_imax_pct) {
+		/*
+		 * Make sure the maximum inode count is a multiple
+		 * of the units we allocate inodes in.
+		 */
+		icount = sbp->sb_dblocks * sbp->sb_imax_pct;
+		do_div(icount, 100);
+		do_div(icount, mp->m_ialloc_blks);
+		mp->m_maxicount = (icount * mp->m_ialloc_blks)  <<
+				   sbp->sb_inopblog;
+	} else {
+		mp->m_maxicount = 0;
+	}
+}
+
+/*
+ * Set the default minimum read and write sizes unless
+ * already specified in a mount option.
+ * We use smaller I/O sizes when the file system
+ * is being used for NFS service (wsync mount option).
+ */
+STATIC void
+xfs_set_rw_sizes(xfs_mount_t *mp)
+{
+	xfs_sb_t	*sbp = &(mp->m_sb);
+	int		readio_log, writeio_log;
+
+	if (!(mp->m_flags & XFS_MOUNT_DFLT_IOSIZE)) {
+		if (mp->m_flags & XFS_MOUNT_WSYNC) {
+			readio_log = XFS_WSYNC_READIO_LOG;
+			writeio_log = XFS_WSYNC_WRITEIO_LOG;
+		} else {
+			readio_log = XFS_READIO_LOG_LARGE;
+			writeio_log = XFS_WRITEIO_LOG_LARGE;
+		}
+	} else {
+		readio_log = mp->m_readio_log;
+		writeio_log = mp->m_writeio_log;
+	}
+
+	if (sbp->sb_blocklog > readio_log) {
+		mp->m_readio_log = sbp->sb_blocklog;
+	} else {
+		mp->m_readio_log = readio_log;
+	}
+	mp->m_readio_blocks = 1 << (mp->m_readio_log - sbp->sb_blocklog);
+	if (sbp->sb_blocklog > writeio_log) {
+		mp->m_writeio_log = sbp->sb_blocklog;
+	} else {
+		mp->m_writeio_log = writeio_log;
+	}
+	mp->m_writeio_blocks = 1 << (mp->m_writeio_log - sbp->sb_blocklog);
+}
+
+/*
+ * precalculate the low space thresholds for dynamic speculative preallocation.
+ */
+void
+xfs_set_low_space_thresholds(
+	struct xfs_mount	*mp)
+{
+	int i;
+
+	for (i = 0; i < XFS_LOWSP_MAX; i++) {
+		__uint64_t space = mp->m_sb.sb_dblocks;
+
+		do_div(space, 100);
+		mp->m_low_space[i] = space * (i + 1);
+	}
+}
+
+
+/*
+ * Set whether we're using inode alignment.
+ */
+STATIC void
+xfs_set_inoalignment(xfs_mount_t *mp)
+{
+	if (xfs_sb_version_hasalign(&mp->m_sb) &&
+	    mp->m_sb.sb_inoalignmt >=
+	    XFS_B_TO_FSBT(mp, mp->m_inode_cluster_size))
+		mp->m_inoalign_mask = mp->m_sb.sb_inoalignmt - 1;
+	else
+		mp->m_inoalign_mask = 0;
+	/*
+	 * If we are using stripe alignment, check whether
+	 * the stripe unit is a multiple of the inode alignment
+	 */
+	if (mp->m_dalign && mp->m_inoalign_mask &&
+	    !(mp->m_dalign & mp->m_inoalign_mask))
+		mp->m_sinoalign = mp->m_dalign;
+	else
+		mp->m_sinoalign = 0;
+}
+
+/*
+ * Check that the data (and log if separate) are an ok size.
+ */
+STATIC int
+xfs_check_sizes(xfs_mount_t *mp)
+{
+	xfs_buf_t	*bp;
+	xfs_daddr_t	d;
+
+	d = (xfs_daddr_t)XFS_FSB_TO_BB(mp, mp->m_sb.sb_dblocks);
+	if (XFS_BB_TO_FSB(mp, d) != mp->m_sb.sb_dblocks) {
+		xfs_warn(mp, "filesystem size mismatch detected");
+		return XFS_ERROR(EFBIG);
+	}
+	bp = xfs_buf_read_uncached(mp, mp->m_ddev_targp,
+					d - XFS_FSS_TO_BB(mp, 1),
+					BBTOB(XFS_FSS_TO_BB(mp, 1)), 0);
+	if (!bp) {
+		xfs_warn(mp, "last sector read failed");
+		return EIO;
+	}
+	xfs_buf_relse(bp);
+
+	if (mp->m_logdev_targp != mp->m_ddev_targp) {
+		d = (xfs_daddr_t)XFS_FSB_TO_BB(mp, mp->m_sb.sb_logblocks);
+		if (XFS_BB_TO_FSB(mp, d) != mp->m_sb.sb_logblocks) {
+			xfs_warn(mp, "log size mismatch detected");
+			return XFS_ERROR(EFBIG);
+		}
+		bp = xfs_buf_read_uncached(mp, mp->m_logdev_targp,
+					d - XFS_FSB_TO_BB(mp, 1),
+					XFS_FSB_TO_B(mp, 1), 0);
+		if (!bp) {
+			xfs_warn(mp, "log device read failed");
+			return EIO;
+		}
+		xfs_buf_relse(bp);
+	}
+	return 0;
+}
+
+/*
+ * Clear the quotaflags in memory and in the superblock.
+ */
+int
+xfs_mount_reset_sbqflags(
+	struct xfs_mount	*mp)
+{
+	int			error;
+	struct xfs_trans	*tp;
+
+	mp->m_qflags = 0;
+
+	/*
+	 * It is OK to look at sb_qflags here in mount path,
+	 * without m_sb_lock.
+	 */
+	if (mp->m_sb.sb_qflags == 0)
+		return 0;
+	spin_lock(&mp->m_sb_lock);
+	mp->m_sb.sb_qflags = 0;
+	spin_unlock(&mp->m_sb_lock);
+
+	/*
+	 * If the fs is readonly, let the incore superblock run
+	 * with quotas off but don't flush the update out to disk
+	 */
+	if (mp->m_flags & XFS_MOUNT_RDONLY)
+		return 0;
+
+#ifdef QUOTADEBUG
+	xfs_notice(mp, "Writing superblock quota changes");
+#endif
+
+	tp = xfs_trans_alloc(mp, XFS_TRANS_QM_SBCHANGE);
+	error = xfs_trans_reserve(tp, 0, mp->m_sb.sb_sectsize + 128, 0, 0,
+				      XFS_DEFAULT_LOG_COUNT);
+	if (error) {
+		xfs_trans_cancel(tp, 0);
+		xfs_alert(mp, "%s: Superblock update failed!", __func__);
+		return error;
+	}
+
+	xfs_mod_sb(tp, XFS_SB_QFLAGS);
+	return xfs_trans_commit(tp, 0);
+}
+
+__uint64_t
+xfs_default_resblks(xfs_mount_t *mp)
+{
+	__uint64_t resblks;
+
+	/*
+	 * We default to 5% or 8192 fsbs of space reserved, whichever is
+	 * smaller.  This is intended to cover concurrent allocation
+	 * transactions when we initially hit enospc. These each require a 4
+	 * block reservation. Hence by default we cover roughly 2000 concurrent
+	 * allocation reservations.
+	 */
+	resblks = mp->m_sb.sb_dblocks;
+	do_div(resblks, 20);
+	resblks = min_t(__uint64_t, resblks, 8192);
+	return resblks;
+}
+
+/*
+ * This function does the following on an initial mount of a file system:
+ *	- reads the superblock from disk and init the mount struct
+ *	- if we're a 32-bit kernel, do a size check on the superblock
+ *		so we don't mount terabyte filesystems
+ *	- init mount struct realtime fields
+ *	- allocate inode hash table for fs
+ *	- init directory manager
+ *	- perform recovery and init the log manager
+ */
+int
+xfs_mountfs(
+	xfs_mount_t	*mp)
+{
+	xfs_sb_t	*sbp = &(mp->m_sb);
+	xfs_inode_t	*rip;
+	__uint64_t	resblks;
+	uint		quotamount = 0;
+	uint		quotaflags = 0;
+	int		error = 0;
+
+	xfs_mount_common(mp, sbp);
+
+	/*
+	 * Check for a mismatched features2 values.  Older kernels
+	 * read & wrote into the wrong sb offset for sb_features2
+	 * on some platforms due to xfs_sb_t not being 64bit size aligned
+	 * when sb_features2 was added, which made older superblock
+	 * reading/writing routines swap it as a 64-bit value.
+	 *
+	 * For backwards compatibility, we make both slots equal.
+	 *
+	 * If we detect a mismatched field, we OR the set bits into the
+	 * existing features2 field in case it has already been modified; we
+	 * don't want to lose any features.  We then update the bad location
+	 * with the ORed value so that older kernels will see any features2
+	 * flags, and mark the two fields as needing updates once the
+	 * transaction subsystem is online.
+	 */
+	if (xfs_sb_has_mismatched_features2(sbp)) {
+		xfs_warn(mp, "correcting sb_features alignment problem");
+		sbp->sb_features2 |= sbp->sb_bad_features2;
+		sbp->sb_bad_features2 = sbp->sb_features2;
+		mp->m_update_flags |= XFS_SB_FEATURES2 | XFS_SB_BAD_FEATURES2;
+
+		/*
+		 * Re-check for ATTR2 in case it was found in bad_features2
+		 * slot.
+		 */
+		if (xfs_sb_version_hasattr2(&mp->m_sb) &&
+		   !(mp->m_flags & XFS_MOUNT_NOATTR2))
+			mp->m_flags |= XFS_MOUNT_ATTR2;
+	}
+
+	if (xfs_sb_version_hasattr2(&mp->m_sb) &&
+	   (mp->m_flags & XFS_MOUNT_NOATTR2)) {
+		xfs_sb_version_removeattr2(&mp->m_sb);
+		mp->m_update_flags |= XFS_SB_FEATURES2;
+
+		/* update sb_versionnum for the clearing of the morebits */
+		if (!sbp->sb_features2)
+			mp->m_update_flags |= XFS_SB_VERSIONNUM;
+	}
+
+	/*
+	 * Check if sb_agblocks is aligned at stripe boundary
+	 * If sb_agblocks is NOT aligned turn off m_dalign since
+	 * allocator alignment is within an ag, therefore ag has
+	 * to be aligned at stripe boundary.
+	 */
+	error = xfs_update_alignment(mp);
+	if (error)
+		goto out;
+
+	xfs_alloc_compute_maxlevels(mp);
+	xfs_bmap_compute_maxlevels(mp, XFS_DATA_FORK);
+	xfs_bmap_compute_maxlevels(mp, XFS_ATTR_FORK);
+	xfs_ialloc_compute_maxlevels(mp);
+
+	xfs_set_maxicount(mp);
+
+	mp->m_maxioffset = xfs_max_file_offset(sbp->sb_blocklog);
+
+	error = xfs_uuid_mount(mp);
+	if (error)
+		goto out;
+
+	/*
+	 * Set the minimum read and write sizes
+	 */
+	xfs_set_rw_sizes(mp);
+
+	/* set the low space thresholds for dynamic preallocation */
+	xfs_set_low_space_thresholds(mp);
+
+	/*
+	 * Set the inode cluster size.
+	 * This may still be overridden by the file system
+	 * block size if it is larger than the chosen cluster size.
+	 */
+	mp->m_inode_cluster_size = XFS_INODE_BIG_CLUSTER_SIZE;
+
+	/*
+	 * Set inode alignment fields
+	 */
+	xfs_set_inoalignment(mp);
+
+	/*
+	 * Check that the data (and log if separate) are an ok size.
+	 */
+	error = xfs_check_sizes(mp);
+	if (error)
+		goto out_remove_uuid;
+
+	/*
+	 * Initialize realtime fields in the mount structure
+	 */
+	error = xfs_rtmount_init(mp);
+	if (error) {
+		xfs_warn(mp, "RT mount failed");
+		goto out_remove_uuid;
+	}
+
+	/*
+	 *  Copies the low order bits of the timestamp and the randomly
+	 *  set "sequence" number out of a UUID.
+	 */
+	uuid_getnodeuniq(&sbp->sb_uuid, mp->m_fixedfsid);
+
+	mp->m_dmevmask = 0;	/* not persistent; set after each mount */
+
+	xfs_dir_mount(mp);
+
+	/*
+	 * Initialize the attribute manager's entries.
+	 */
+	mp->m_attr_magicpct = (mp->m_sb.sb_blocksize * 37) / 100;
+
+	/*
+	 * Initialize the precomputed transaction reservations values.
+	 */
+	xfs_trans_init(mp);
+
+	/*
+	 * Allocate and initialize the per-ag data.
+	 */
+	spin_lock_init(&mp->m_perag_lock);
+	INIT_RADIX_TREE(&mp->m_perag_tree, GFP_ATOMIC);
+	error = xfs_initialize_perag(mp, sbp->sb_agcount, &mp->m_maxagi);
+	if (error) {
+		xfs_warn(mp, "Failed per-ag init: %d", error);
+		goto out_remove_uuid;
+	}
+
+	if (!sbp->sb_logblocks) {
+		xfs_warn(mp, "no log defined");
+		XFS_ERROR_REPORT("xfs_mountfs", XFS_ERRLEVEL_LOW, mp);
+		error = XFS_ERROR(EFSCORRUPTED);
+		goto out_free_perag;
+	}
+
+	/*
+	 * log's mount-time initialization. Perform 1st part recovery if needed
+	 */
+	error = xfs_log_mount(mp, mp->m_logdev_targp,
+			      XFS_FSB_TO_DADDR(mp, sbp->sb_logstart),
+			      XFS_FSB_TO_BB(mp, sbp->sb_logblocks));
+	if (error) {
+		xfs_warn(mp, "log mount failed");
+		goto out_free_perag;
+	}
+
+	/*
+	 * Now the log is mounted, we know if it was an unclean shutdown or
+	 * not. If it was, with the first phase of recovery has completed, we
+	 * have consistent AG blocks on disk. We have not recovered EFIs yet,
+	 * but they are recovered transactionally in the second recovery phase
+	 * later.
+	 *
+	 * Hence we can safely re-initialise incore superblock counters from
+	 * the per-ag data. These may not be correct if the filesystem was not
+	 * cleanly unmounted, so we need to wait for recovery to finish before
+	 * doing this.
+	 *
+	 * If the filesystem was cleanly unmounted, then we can trust the
+	 * values in the superblock to be correct and we don't need to do
+	 * anything here.
+	 *
+	 * If we are currently making the filesystem, the initialisation will
+	 * fail as the perag data is in an undefined state.
+	 */
+	if (xfs_sb_version_haslazysbcount(&mp->m_sb) &&
+	    !XFS_LAST_UNMOUNT_WAS_CLEAN(mp) &&
+	     !mp->m_sb.sb_inprogress) {
+		error = xfs_initialize_perag_data(mp, sbp->sb_agcount);
+		if (error)
+			goto out_free_perag;
+	}
+
+	/*
+	 * Get and sanity-check the root inode.
+	 * Save the pointer to it in the mount structure.
+	 */
+	error = xfs_iget(mp, NULL, sbp->sb_rootino, 0, XFS_ILOCK_EXCL, &rip);
+	if (error) {
+		xfs_warn(mp, "failed to read root inode");
+		goto out_log_dealloc;
+	}
+
+	ASSERT(rip != NULL);
+
+	if (unlikely((rip->i_d.di_mode & S_IFMT) != S_IFDIR)) {
+		xfs_warn(mp, "corrupted root inode %llu: not a directory",
+			(unsigned long long)rip->i_ino);
+		xfs_iunlock(rip, XFS_ILOCK_EXCL);
+		XFS_ERROR_REPORT("xfs_mountfs_int(2)", XFS_ERRLEVEL_LOW,
+				 mp);
+		error = XFS_ERROR(EFSCORRUPTED);
+		goto out_rele_rip;
+	}
+	mp->m_rootip = rip;	/* save it */
+
+	xfs_iunlock(rip, XFS_ILOCK_EXCL);
+
+	/*
+	 * Initialize realtime inode pointers in the mount structure
+	 */
+	error = xfs_rtmount_inodes(mp);
+	if (error) {
+		/*
+		 * Free up the root inode.
+		 */
+		xfs_warn(mp, "failed to read RT inodes");
+		goto out_rele_rip;
+	}
+
+	/*
+	 * If this is a read-only mount defer the superblock updates until
+	 * the next remount into writeable mode.  Otherwise we would never
+	 * perform the update e.g. for the root filesystem.
+	 */
+	if (mp->m_update_flags && !(mp->m_flags & XFS_MOUNT_RDONLY)) {
+		error = xfs_mount_log_sb(mp, mp->m_update_flags);
+		if (error) {
+			xfs_warn(mp, "failed to write sb changes");
+			goto out_rtunmount;
+		}
+	}
+
+	/*
+	 * Initialise the XFS quota management subsystem for this mount
+	 */
+	if (XFS_IS_QUOTA_RUNNING(mp)) {
+		error = xfs_qm_newmount(mp, &quotamount, &quotaflags);
+		if (error)
+			goto out_rtunmount;
+	} else {
+		ASSERT(!XFS_IS_QUOTA_ON(mp));
+
+		/*
+		 * If a file system had quotas running earlier, but decided to
+		 * mount without -o uquota/pquota/gquota options, revoke the
+		 * quotachecked license.
+		 */
+		if (mp->m_sb.sb_qflags & XFS_ALL_QUOTA_ACCT) {
+			xfs_notice(mp, "resetting quota flags");
+			error = xfs_mount_reset_sbqflags(mp);
+			if (error)
+				return error;
+		}
+	}
+
+	/*
+	 * Finish recovering the file system.  This part needed to be
+	 * delayed until after the root and real-time bitmap inodes
+	 * were consistently read in.
+	 */
+	error = xfs_log_mount_finish(mp);
+	if (error) {
+		xfs_warn(mp, "log mount finish failed");
+		goto out_rtunmount;
+	}
+
+	/*
+	 * Complete the quota initialisation, post-log-replay component.
+	 */
+	if (quotamount) {
+		ASSERT(mp->m_qflags == 0);
+		mp->m_qflags = quotaflags;
+
+		xfs_qm_mount_quotas(mp);
+	}
+
+	/*
+	 * Now we are mounted, reserve a small amount of unused space for
+	 * privileged transactions. This is needed so that transaction
+	 * space required for critical operations can dip into this pool
+	 * when at ENOSPC. This is needed for operations like create with
+	 * attr, unwritten extent conversion at ENOSPC, etc. Data allocations
+	 * are not allowed to use this reserved space.
+	 *
+	 * This may drive us straight to ENOSPC on mount, but that implies
+	 * we were already there on the last unmount. Warn if this occurs.
+	 */
+	if (!(mp->m_flags & XFS_MOUNT_RDONLY)) {
+		resblks = xfs_default_resblks(mp);
+		error = xfs_reserve_blocks(mp, &resblks, NULL);
+		if (error)
+			xfs_warn(mp,
+	"Unable to allocate reserve blocks. Continuing without reserve pool.");
+	}
+
+	return 0;
+
+ out_rtunmount:
+	xfs_rtunmount_inodes(mp);
+ out_rele_rip:
+	IRELE(rip);
+ out_log_dealloc:
+	xfs_log_unmount(mp);
+ out_free_perag:
+	xfs_free_perag(mp);
+ out_remove_uuid:
+	xfs_uuid_unmount(mp);
+ out:
+	return error;
+}
+
+/*
+ * This flushes out the inodes,dquots and the superblock, unmounts the
+ * log and makes sure that incore structures are freed.
+ */
+void
+xfs_unmountfs(
+	struct xfs_mount	*mp)
+{
+	__uint64_t		resblks;
+	int			error;
+
+	xfs_qm_unmount_quotas(mp);
+	xfs_rtunmount_inodes(mp);
+	IRELE(mp->m_rootip);
+
+	/*
+	 * We can potentially deadlock here if we have an inode cluster
+	 * that has been freed has its buffer still pinned in memory because
+	 * the transaction is still sitting in a iclog. The stale inodes
+	 * on that buffer will have their flush locks held until the
+	 * transaction hits the disk and the callbacks run. the inode
+	 * flush takes the flush lock unconditionally and with nothing to
+	 * push out the iclog we will never get that unlocked. hence we
+	 * need to force the log first.
+	 */
+	xfs_log_force(mp, XFS_LOG_SYNC);
+
+	/*
+	 * Do a delwri reclaim pass first so that as many dirty inodes are
+	 * queued up for IO as possible. Then flush the buffers before making
+	 * a synchronous path to catch all the remaining inodes are reclaimed.
+	 * This makes the reclaim process as quick as possible by avoiding
+	 * synchronous writeout and blocking on inodes already in the delwri
+	 * state as much as possible.
+	 */
+	xfs_reclaim_inodes(mp, 0);
+	XFS_bflush(mp->m_ddev_targp);
+	xfs_reclaim_inodes(mp, SYNC_WAIT);
+
+	xfs_qm_unmount(mp);
+
+	/*
+	 * Flush out the log synchronously so that we know for sure
+	 * that nothing is pinned.  This is important because bflush()
+	 * will skip pinned buffers.
+	 */
+	xfs_log_force(mp, XFS_LOG_SYNC);
+
+	/*
+	 * Unreserve any blocks we have so that when we unmount we don't account
+	 * the reserved free space as used. This is really only necessary for
+	 * lazy superblock counting because it trusts the incore superblock
+	 * counters to be absolutely correct on clean unmount.
+	 *
+	 * We don't bother correcting this elsewhere for lazy superblock
+	 * counting because on mount of an unclean filesystem we reconstruct the
+	 * correct counter value and this is irrelevant.
+	 *
+	 * For non-lazy counter filesystems, this doesn't matter at all because
+	 * we only every apply deltas to the superblock and hence the incore
+	 * value does not matter....
+	 */
+	resblks = 0;
+	error = xfs_reserve_blocks(mp, &resblks, NULL);
+	if (error)
+		xfs_warn(mp, "Unable to free reserved block pool. "
+				"Freespace may not be correct on next mount.");
+
+	error = xfs_log_sbcount(mp, 1);
+	if (error)
+		xfs_warn(mp, "Unable to update superblock counters. "
+				"Freespace may not be correct on next mount.");
+	xfs_unmountfs_writesb(mp);
+
+	/*
+	 * Make sure all buffers have been flushed and completed before
+	 * unmounting the log.
+	 */
+	error = xfs_flush_buftarg(mp->m_ddev_targp, 1);
+	if (error)
+		xfs_warn(mp, "%d busy buffers during unmount.", error);
+	xfs_wait_buftarg(mp->m_ddev_targp);
+
+	xfs_log_unmount_write(mp);
+	xfs_log_unmount(mp);
+	xfs_uuid_unmount(mp);
+
+#if defined(DEBUG)
+	xfs_errortag_clearall(mp, 0);
+#endif
+	xfs_free_perag(mp);
+}
+
+int
+xfs_fs_writable(xfs_mount_t *mp)
+{
+	return !(xfs_test_for_freeze(mp) || XFS_FORCED_SHUTDOWN(mp) ||
+		(mp->m_flags & XFS_MOUNT_RDONLY));
+}
+
+/*
+ * xfs_log_sbcount
+ *
+ * Called either periodically to keep the on disk superblock values
+ * roughly up to date or from unmount to make sure the values are
+ * correct on a clean unmount.
+ *
+ * Note this code can be called during the process of freezing, so
+ * we may need to use the transaction allocator which does not not
+ * block when the transaction subsystem is in its frozen state.
+ */
+int
+xfs_log_sbcount(
+	xfs_mount_t	*mp,
+	uint		sync)
+{
+	xfs_trans_t	*tp;
+	int		error;
+
+	if (!xfs_fs_writable(mp))
+		return 0;
+
+	xfs_icsb_sync_counters(mp, 0);
+
+	/*
+	 * we don't need to do this if we are updating the superblock
+	 * counters on every modification.
+	 */
+	if (!xfs_sb_version_haslazysbcount(&mp->m_sb))
+		return 0;
+
+	tp = _xfs_trans_alloc(mp, XFS_TRANS_SB_COUNT, KM_SLEEP);
+	error = xfs_trans_reserve(tp, 0, mp->m_sb.sb_sectsize + 128, 0, 0,
+					XFS_DEFAULT_LOG_COUNT);
+	if (error) {
+		xfs_trans_cancel(tp, 0);
+		return error;
+	}
+
+	xfs_mod_sb(tp, XFS_SB_IFREE | XFS_SB_ICOUNT | XFS_SB_FDBLOCKS);
+	if (sync)
+		xfs_trans_set_sync(tp);
+	error = xfs_trans_commit(tp, 0);
+	return error;
+}
+
+int
+xfs_unmountfs_writesb(xfs_mount_t *mp)
+{
+	xfs_buf_t	*sbp;
+	int		error = 0;
+
+	/*
+	 * skip superblock write if fs is read-only, or
+	 * if we are doing a forced umount.
+	 */
+	if (!((mp->m_flags & XFS_MOUNT_RDONLY) ||
+		XFS_FORCED_SHUTDOWN(mp))) {
+
+		sbp = xfs_getsb(mp, 0);
+
+		XFS_BUF_UNDONE(sbp);
+		XFS_BUF_UNREAD(sbp);
+		XFS_BUF_UNDELAYWRITE(sbp);
+		XFS_BUF_WRITE(sbp);
+		XFS_BUF_UNASYNC(sbp);
+		ASSERT(XFS_BUF_TARGET(sbp) == mp->m_ddev_targp);
+		xfsbdstrat(mp, sbp);
+		error = xfs_buf_iowait(sbp);
+		if (error)
+			xfs_ioerror_alert("xfs_unmountfs_writesb",
+					  mp, sbp, XFS_BUF_ADDR(sbp));
+		xfs_buf_relse(sbp);
+	}
+	return error;
+}
+
+/*
+ * xfs_mod_sb() can be used to copy arbitrary changes to the
+ * in-core superblock into the superblock buffer to be logged.
+ * It does not provide the higher level of locking that is
+ * needed to protect the in-core superblock from concurrent
+ * access.
+ */
+void
+xfs_mod_sb(xfs_trans_t *tp, __int64_t fields)
+{
+	xfs_buf_t	*bp;
+	int		first;
+	int		last;
+	xfs_mount_t	*mp;
+	xfs_sb_field_t	f;
+
+	ASSERT(fields);
+	if (!fields)
+		return;
+	mp = tp->t_mountp;
+	bp = xfs_trans_getsb(tp, mp, 0);
+	first = sizeof(xfs_sb_t);
+	last = 0;
+
+	/* translate/copy */
+
+	xfs_sb_to_disk(XFS_BUF_TO_SBP(bp), &mp->m_sb, fields);
+
+	/* find modified range */
+	f = (xfs_sb_field_t)xfs_highbit64((__uint64_t)fields);
+	ASSERT((1LL << f) & XFS_SB_MOD_BITS);
+	last = xfs_sb_info[f + 1].offset - 1;
+
+	f = (xfs_sb_field_t)xfs_lowbit64((__uint64_t)fields);
+	ASSERT((1LL << f) & XFS_SB_MOD_BITS);
+	first = xfs_sb_info[f].offset;
+
+	xfs_trans_log_buf(tp, bp, first, last);
+}
+
+
+/*
+ * xfs_mod_incore_sb_unlocked() is a utility routine common used to apply
+ * a delta to a specified field in the in-core superblock.  Simply
+ * switch on the field indicated and apply the delta to that field.
+ * Fields are not allowed to dip below zero, so if the delta would
+ * do this do not apply it and return EINVAL.
+ *
+ * The m_sb_lock must be held when this routine is called.
+ */
+STATIC int
+xfs_mod_incore_sb_unlocked(
+	xfs_mount_t	*mp,
+	xfs_sb_field_t	field,
+	int64_t		delta,
+	int		rsvd)
+{
+	int		scounter;	/* short counter for 32 bit fields */
+	long long	lcounter;	/* long counter for 64 bit fields */
+	long long	res_used, rem;
+
+	/*
+	 * With the in-core superblock spin lock held, switch
+	 * on the indicated field.  Apply the delta to the
+	 * proper field.  If the fields value would dip below
+	 * 0, then do not apply the delta and return EINVAL.
+	 */
+	switch (field) {
+	case XFS_SBS_ICOUNT:
+		lcounter = (long long)mp->m_sb.sb_icount;
+		lcounter += delta;
+		if (lcounter < 0) {
+			ASSERT(0);
+			return XFS_ERROR(EINVAL);
+		}
+		mp->m_sb.sb_icount = lcounter;
+		return 0;
+	case XFS_SBS_IFREE:
+		lcounter = (long long)mp->m_sb.sb_ifree;
+		lcounter += delta;
+		if (lcounter < 0) {
+			ASSERT(0);
+			return XFS_ERROR(EINVAL);
+		}
+		mp->m_sb.sb_ifree = lcounter;
+		return 0;
+	case XFS_SBS_FDBLOCKS:
+		lcounter = (long long)
+			mp->m_sb.sb_fdblocks - XFS_ALLOC_SET_ASIDE(mp);
+		res_used = (long long)(mp->m_resblks - mp->m_resblks_avail);
+
+		if (delta > 0) {		/* Putting blocks back */
+			if (res_used > delta) {
+				mp->m_resblks_avail += delta;
+			} else {
+				rem = delta - res_used;
+				mp->m_resblks_avail = mp->m_resblks;
+				lcounter += rem;
+			}
+		} else {				/* Taking blocks away */
+			lcounter += delta;
+			if (lcounter >= 0) {
+				mp->m_sb.sb_fdblocks = lcounter +
+							XFS_ALLOC_SET_ASIDE(mp);
+				return 0;
+			}
+
+			/*
+			 * We are out of blocks, use any available reserved
+			 * blocks if were allowed to.
+			 */
+			if (!rsvd)
+				return XFS_ERROR(ENOSPC);
+
+			lcounter = (long long)mp->m_resblks_avail + delta;
+			if (lcounter >= 0) {
+				mp->m_resblks_avail = lcounter;
+				return 0;
+			}
+			printk_once(KERN_WARNING
+				"Filesystem \"%s\": reserve blocks depleted! "
+				"Consider increasing reserve pool size.",
+				mp->m_fsname);
+			return XFS_ERROR(ENOSPC);
+		}
+
+		mp->m_sb.sb_fdblocks = lcounter + XFS_ALLOC_SET_ASIDE(mp);
+		return 0;
+	case XFS_SBS_FREXTENTS:
+		lcounter = (long long)mp->m_sb.sb_frextents;
+		lcounter += delta;
+		if (lcounter < 0) {
+			return XFS_ERROR(ENOSPC);
+		}
+		mp->m_sb.sb_frextents = lcounter;
+		return 0;
+	case XFS_SBS_DBLOCKS:
+		lcounter = (long long)mp->m_sb.sb_dblocks;
+		lcounter += delta;
+		if (lcounter < 0) {
+			ASSERT(0);
+			return XFS_ERROR(EINVAL);
+		}
+		mp->m_sb.sb_dblocks = lcounter;
+		return 0;
+	case XFS_SBS_AGCOUNT:
+		scounter = mp->m_sb.sb_agcount;
+		scounter += delta;
+		if (scounter < 0) {
+			ASSERT(0);
+			return XFS_ERROR(EINVAL);
+		}
+		mp->m_sb.sb_agcount = scounter;
+		return 0;
+	case XFS_SBS_IMAX_PCT:
+		scounter = mp->m_sb.sb_imax_pct;
+		scounter += delta;
+		if (scounter < 0) {
+			ASSERT(0);
+			return XFS_ERROR(EINVAL);
+		}
+		mp->m_sb.sb_imax_pct = scounter;
+		return 0;
+	case XFS_SBS_REXTSIZE:
+		scounter = mp->m_sb.sb_rextsize;
+		scounter += delta;
+		if (scounter < 0) {
+			ASSERT(0);
+			return XFS_ERROR(EINVAL);
+		}
+		mp->m_sb.sb_rextsize = scounter;
+		return 0;
+	case XFS_SBS_RBMBLOCKS:
+		scounter = mp->m_sb.sb_rbmblocks;
+		scounter += delta;
+		if (scounter < 0) {
+			ASSERT(0);
+			return XFS_ERROR(EINVAL);
+		}
+		mp->m_sb.sb_rbmblocks = scounter;
+		return 0;
+	case XFS_SBS_RBLOCKS:
+		lcounter = (long long)mp->m_sb.sb_rblocks;
+		lcounter += delta;
+		if (lcounter < 0) {
+			ASSERT(0);
+			return XFS_ERROR(EINVAL);
+		}
+		mp->m_sb.sb_rblocks = lcounter;
+		return 0;
+	case XFS_SBS_REXTENTS:
+		lcounter = (long long)mp->m_sb.sb_rextents;
+		lcounter += delta;
+		if (lcounter < 0) {
+			ASSERT(0);
+			return XFS_ERROR(EINVAL);
+		}
+		mp->m_sb.sb_rextents = lcounter;
+		return 0;
+	case XFS_SBS_REXTSLOG:
+		scounter = mp->m_sb.sb_rextslog;
+		scounter += delta;
+		if (scounter < 0) {
+			ASSERT(0);
+			return XFS_ERROR(EINVAL);
+		}
+		mp->m_sb.sb_rextslog = scounter;
+		return 0;
+	default:
+		ASSERT(0);
+		return XFS_ERROR(EINVAL);
+	}
+}
+
+/*
+ * xfs_mod_incore_sb() is used to change a field in the in-core
+ * superblock structure by the specified delta.  This modification
+ * is protected by the m_sb_lock.  Just use the xfs_mod_incore_sb_unlocked()
+ * routine to do the work.
+ */
+int
+xfs_mod_incore_sb(
+	struct xfs_mount	*mp,
+	xfs_sb_field_t		field,
+	int64_t			delta,
+	int			rsvd)
+{
+	int			status;
+
+#ifdef HAVE_PERCPU_SB
+	ASSERT(field < XFS_SBS_ICOUNT || field > XFS_SBS_FDBLOCKS);
+#endif
+	spin_lock(&mp->m_sb_lock);
+	status = xfs_mod_incore_sb_unlocked(mp, field, delta, rsvd);
+	spin_unlock(&mp->m_sb_lock);
+
+	return status;
+}
+
+/*
+ * Change more than one field in the in-core superblock structure at a time.
+ *
+ * The fields and changes to those fields are specified in the array of
+ * xfs_mod_sb structures passed in.  Either all of the specified deltas
+ * will be applied or none of them will.  If any modified field dips below 0,
+ * then all modifications will be backed out and EINVAL will be returned.
+ *
+ * Note that this function may not be used for the superblock values that
+ * are tracked with the in-memory per-cpu counters - a direct call to
+ * xfs_icsb_modify_counters is required for these.
+ */
+int
+xfs_mod_incore_sb_batch(
+	struct xfs_mount	*mp,
+	xfs_mod_sb_t		*msb,
+	uint			nmsb,
+	int			rsvd)
+{
+	xfs_mod_sb_t		*msbp;
+	int			error = 0;
+
+	/*
+	 * Loop through the array of mod structures and apply each individually.
+	 * If any fail, then back out all those which have already been applied.
+	 * Do all of this within the scope of the m_sb_lock so that all of the
+	 * changes will be atomic.
+	 */
+	spin_lock(&mp->m_sb_lock);
+	for (msbp = msb; msbp < (msb + nmsb); msbp++) {
+		ASSERT(msbp->msb_field < XFS_SBS_ICOUNT ||
+		       msbp->msb_field > XFS_SBS_FDBLOCKS);
+
+		error = xfs_mod_incore_sb_unlocked(mp, msbp->msb_field,
+						   msbp->msb_delta, rsvd);
+		if (error)
+			goto unwind;
+	}
+	spin_unlock(&mp->m_sb_lock);
+	return 0;
+
+unwind:
+	while (--msbp >= msb) {
+		error = xfs_mod_incore_sb_unlocked(mp, msbp->msb_field,
+						   -msbp->msb_delta, rsvd);
+		ASSERT(error == 0);
+	}
+	spin_unlock(&mp->m_sb_lock);
+	return error;
+}
+
+/*
+ * xfs_getsb() is called to obtain the buffer for the superblock.
+ * The buffer is returned locked and read in from disk.
+ * The buffer should be released with a call to xfs_brelse().
+ *
+ * If the flags parameter is BUF_TRYLOCK, then we'll only return
+ * the superblock buffer if it can be locked without sleeping.
+ * If it can't then we'll return NULL.
+ */
+xfs_buf_t *
+xfs_getsb(
+	xfs_mount_t	*mp,
+	int		flags)
+{
+	xfs_buf_t	*bp;
+
+	ASSERT(mp->m_sb_bp != NULL);
+	bp = mp->m_sb_bp;
+	if (flags & XBF_TRYLOCK) {
+		if (!XFS_BUF_CPSEMA(bp)) {
+			return NULL;
+		}
+	} else {
+		XFS_BUF_PSEMA(bp, PRIBIO);
+	}
+	XFS_BUF_HOLD(bp);
+	ASSERT(XFS_BUF_ISDONE(bp));
+	return bp;
+}
+
+/*
+ * Used to free the superblock along various error paths.
+ */
+void
+xfs_freesb(
+	struct xfs_mount	*mp)
+{
+	struct xfs_buf		*bp = mp->m_sb_bp;
+
+	xfs_buf_lock(bp);
+	mp->m_sb_bp = NULL;
+	xfs_buf_relse(bp);
+}
+
+/*
+ * Used to log changes to the superblock unit and width fields which could
+ * be altered by the mount options, as well as any potential sb_features2
+ * fixup. Only the first superblock is updated.
+ */
+int
+xfs_mount_log_sb(
+	xfs_mount_t	*mp,
+	__int64_t	fields)
+{
+	xfs_trans_t	*tp;
+	int		error;
+
+	ASSERT(fields & (XFS_SB_UNIT | XFS_SB_WIDTH | XFS_SB_UUID |
+			 XFS_SB_FEATURES2 | XFS_SB_BAD_FEATURES2 |
+			 XFS_SB_VERSIONNUM));
+
+	tp = xfs_trans_alloc(mp, XFS_TRANS_SB_UNIT);
+	error = xfs_trans_reserve(tp, 0, mp->m_sb.sb_sectsize + 128, 0, 0,
+				XFS_DEFAULT_LOG_COUNT);
+	if (error) {
+		xfs_trans_cancel(tp, 0);
+		return error;
+	}
+	xfs_mod_sb(tp, fields);
+	error = xfs_trans_commit(tp, 0);
+	return error;
+}
+
+/*
+ * If the underlying (data/log/rt) device is readonly, there are some
+ * operations that cannot proceed.
+ */
+int
+xfs_dev_is_read_only(
+	struct xfs_mount	*mp,
+	char			*message)
+{
+	if (xfs_readonly_buftarg(mp->m_ddev_targp) ||
+	    xfs_readonly_buftarg(mp->m_logdev_targp) ||
+	    (mp->m_rtdev_targp && xfs_readonly_buftarg(mp->m_rtdev_targp))) {
+		xfs_notice(mp, "%s required on read-only device.", message);
+		xfs_notice(mp, "write access unavailable, cannot proceed.");
+		return EROFS;
+	}
+	return 0;
+}
+
+#ifdef HAVE_PERCPU_SB
+/*
+ * Per-cpu incore superblock counters
+ *
+ * Simple concept, difficult implementation
+ *
+ * Basically, replace the incore superblock counters with a distributed per cpu
+ * counter for contended fields (e.g.  free block count).
+ *
+ * Difficulties arise in that the incore sb is used for ENOSPC checking, and
+ * hence needs to be accurately read when we are running low on space. Hence
+ * there is a method to enable and disable the per-cpu counters based on how
+ * much "stuff" is available in them.
+ *
+ * Basically, a counter is enabled if there is enough free resource to justify
+ * running a per-cpu fast-path. If the per-cpu counter runs out (i.e. a local
+ * ENOSPC), then we disable the counters to synchronise all callers and
+ * re-distribute the available resources.
+ *
+ * If, once we redistributed the available resources, we still get a failure,
+ * we disable the per-cpu counter and go through the slow path.
+ *
+ * The slow path is the current xfs_mod_incore_sb() function.  This means that
+ * when we disable a per-cpu counter, we need to drain its resources back to
+ * the global superblock. We do this after disabling the counter to prevent
+ * more threads from queueing up on the counter.
+ *
+ * Essentially, this means that we still need a lock in the fast path to enable
+ * synchronisation between the global counters and the per-cpu counters. This
+ * is not a problem because the lock will be local to a CPU almost all the time
+ * and have little contention except when we get to ENOSPC conditions.
+ *
+ * Basically, this lock becomes a barrier that enables us to lock out the fast
+ * path while we do things like enabling and disabling counters and
+ * synchronising the counters.
+ *
+ * Locking rules:
+ *
+ * 	1. m_sb_lock before picking up per-cpu locks
+ * 	2. per-cpu locks always picked up via for_each_online_cpu() order
+ * 	3. accurate counter sync requires m_sb_lock + per cpu locks
+ * 	4. modifying per-cpu counters requires holding per-cpu lock
+ * 	5. modifying global counters requires holding m_sb_lock
+ *	6. enabling or disabling a counter requires holding the m_sb_lock 
+ *	   and _none_ of the per-cpu locks.
+ *
+ * Disabled counters are only ever re-enabled by a balance operation
+ * that results in more free resources per CPU than a given threshold.
+ * To ensure counters don't remain disabled, they are rebalanced when
+ * the global resource goes above a higher threshold (i.e. some hysteresis
+ * is present to prevent thrashing).
+ */
+
+#ifdef CONFIG_HOTPLUG_CPU
+/*
+ * hot-plug CPU notifier support.
+ *
+ * We need a notifier per filesystem as we need to be able to identify
+ * the filesystem to balance the counters out. This is achieved by
+ * having a notifier block embedded in the xfs_mount_t and doing pointer
+ * magic to get the mount pointer from the notifier block address.
+ */
+STATIC int
+xfs_icsb_cpu_notify(
+	struct notifier_block *nfb,
+	unsigned long action,
+	void *hcpu)
+{
+	xfs_icsb_cnts_t *cntp;
+	xfs_mount_t	*mp;
+
+	mp = (xfs_mount_t *)container_of(nfb, xfs_mount_t, m_icsb_notifier);
+	cntp = (xfs_icsb_cnts_t *)
+			per_cpu_ptr(mp->m_sb_cnts, (unsigned long)hcpu);
+	switch (action) {
+	case CPU_UP_PREPARE:
+	case CPU_UP_PREPARE_FROZEN:
+		/* Easy Case - initialize the area and locks, and
+		 * then rebalance when online does everything else for us. */
+		memset(cntp, 0, sizeof(xfs_icsb_cnts_t));
+		break;
+	case CPU_ONLINE:
+	case CPU_ONLINE_FROZEN:
+		xfs_icsb_lock(mp);
+		xfs_icsb_balance_counter(mp, XFS_SBS_ICOUNT, 0);
+		xfs_icsb_balance_counter(mp, XFS_SBS_IFREE, 0);
+		xfs_icsb_balance_counter(mp, XFS_SBS_FDBLOCKS, 0);
+		xfs_icsb_unlock(mp);
+		break;
+	case CPU_DEAD:
+	case CPU_DEAD_FROZEN:
+		/* Disable all the counters, then fold the dead cpu's
+		 * count into the total on the global superblock and
+		 * re-enable the counters. */
+		xfs_icsb_lock(mp);
+		spin_lock(&mp->m_sb_lock);
+		xfs_icsb_disable_counter(mp, XFS_SBS_ICOUNT);
+		xfs_icsb_disable_counter(mp, XFS_SBS_IFREE);
+		xfs_icsb_disable_counter(mp, XFS_SBS_FDBLOCKS);
+
+		mp->m_sb.sb_icount += cntp->icsb_icount;
+		mp->m_sb.sb_ifree += cntp->icsb_ifree;
+		mp->m_sb.sb_fdblocks += cntp->icsb_fdblocks;
+
+		memset(cntp, 0, sizeof(xfs_icsb_cnts_t));
+
+		xfs_icsb_balance_counter_locked(mp, XFS_SBS_ICOUNT, 0);
+		xfs_icsb_balance_counter_locked(mp, XFS_SBS_IFREE, 0);
+		xfs_icsb_balance_counter_locked(mp, XFS_SBS_FDBLOCKS, 0);
+		spin_unlock(&mp->m_sb_lock);
+		xfs_icsb_unlock(mp);
+		break;
+	}
+
+	return NOTIFY_OK;
+}
+#endif /* CONFIG_HOTPLUG_CPU */
+
+int
+xfs_icsb_init_counters(
+	xfs_mount_t	*mp)
+{
+	xfs_icsb_cnts_t *cntp;
+	int		i;
+
+	mp->m_sb_cnts = alloc_percpu(xfs_icsb_cnts_t);
+	if (mp->m_sb_cnts == NULL)
+		return -ENOMEM;
+
+#ifdef CONFIG_HOTPLUG_CPU
+	mp->m_icsb_notifier.notifier_call = xfs_icsb_cpu_notify;
+	mp->m_icsb_notifier.priority = 0;
+	register_hotcpu_notifier(&mp->m_icsb_notifier);
+#endif /* CONFIG_HOTPLUG_CPU */
+
+	for_each_online_cpu(i) {
+		cntp = (xfs_icsb_cnts_t *)per_cpu_ptr(mp->m_sb_cnts, i);
+		memset(cntp, 0, sizeof(xfs_icsb_cnts_t));
+	}
+
+	mutex_init(&mp->m_icsb_mutex);
+
+	/*
+	 * start with all counters disabled so that the
+	 * initial balance kicks us off correctly
+	 */
+	mp->m_icsb_counters = -1;
+	return 0;
+}
+
+void
+xfs_icsb_reinit_counters(
+	xfs_mount_t	*mp)
+{
+	xfs_icsb_lock(mp);
+	/*
+	 * start with all counters disabled so that the
+	 * initial balance kicks us off correctly
+	 */
+	mp->m_icsb_counters = -1;
+	xfs_icsb_balance_counter(mp, XFS_SBS_ICOUNT, 0);
+	xfs_icsb_balance_counter(mp, XFS_SBS_IFREE, 0);
+	xfs_icsb_balance_counter(mp, XFS_SBS_FDBLOCKS, 0);
+	xfs_icsb_unlock(mp);
+}
+
+void
+xfs_icsb_destroy_counters(
+	xfs_mount_t	*mp)
+{
+	if (mp->m_sb_cnts) {
+		unregister_hotcpu_notifier(&mp->m_icsb_notifier);
+		free_percpu(mp->m_sb_cnts);
+	}
+	mutex_destroy(&mp->m_icsb_mutex);
+}
+
+STATIC void
+xfs_icsb_lock_cntr(
+	xfs_icsb_cnts_t	*icsbp)
+{
+	while (test_and_set_bit(XFS_ICSB_FLAG_LOCK, &icsbp->icsb_flags)) {
+		ndelay(1000);
+	}
+}
+
+STATIC void
+xfs_icsb_unlock_cntr(
+	xfs_icsb_cnts_t	*icsbp)
+{
+	clear_bit(XFS_ICSB_FLAG_LOCK, &icsbp->icsb_flags);
+}
+
+
+STATIC void
+xfs_icsb_lock_all_counters(
+	xfs_mount_t	*mp)
+{
+	xfs_icsb_cnts_t *cntp;
+	int		i;
+
+	for_each_online_cpu(i) {
+		cntp = (xfs_icsb_cnts_t *)per_cpu_ptr(mp->m_sb_cnts, i);
+		xfs_icsb_lock_cntr(cntp);
+	}
+}
+
+STATIC void
+xfs_icsb_unlock_all_counters(
+	xfs_mount_t	*mp)
+{
+	xfs_icsb_cnts_t *cntp;
+	int		i;
+
+	for_each_online_cpu(i) {
+		cntp = (xfs_icsb_cnts_t *)per_cpu_ptr(mp->m_sb_cnts, i);
+		xfs_icsb_unlock_cntr(cntp);
+	}
+}
+
+STATIC void
+xfs_icsb_count(
+	xfs_mount_t	*mp,
+	xfs_icsb_cnts_t	*cnt,
+	int		flags)
+{
+	xfs_icsb_cnts_t *cntp;
+	int		i;
+
+	memset(cnt, 0, sizeof(xfs_icsb_cnts_t));
+
+	if (!(flags & XFS_ICSB_LAZY_COUNT))
+		xfs_icsb_lock_all_counters(mp);
+
+	for_each_online_cpu(i) {
+		cntp = (xfs_icsb_cnts_t *)per_cpu_ptr(mp->m_sb_cnts, i);
+		cnt->icsb_icount += cntp->icsb_icount;
+		cnt->icsb_ifree += cntp->icsb_ifree;
+		cnt->icsb_fdblocks += cntp->icsb_fdblocks;
+	}
+
+	if (!(flags & XFS_ICSB_LAZY_COUNT))
+		xfs_icsb_unlock_all_counters(mp);
+}
+
+STATIC int
+xfs_icsb_counter_disabled(
+	xfs_mount_t	*mp,
+	xfs_sb_field_t	field)
+{
+	ASSERT((field >= XFS_SBS_ICOUNT) && (field <= XFS_SBS_FDBLOCKS));
+	return test_bit(field, &mp->m_icsb_counters);
+}
+
+STATIC void
+xfs_icsb_disable_counter(
+	xfs_mount_t	*mp,
+	xfs_sb_field_t	field)
+{
+	xfs_icsb_cnts_t	cnt;
+
+	ASSERT((field >= XFS_SBS_ICOUNT) && (field <= XFS_SBS_FDBLOCKS));
+
+	/*
+	 * If we are already disabled, then there is nothing to do
+	 * here. We check before locking all the counters to avoid
+	 * the expensive lock operation when being called in the
+	 * slow path and the counter is already disabled. This is
+	 * safe because the only time we set or clear this state is under
+	 * the m_icsb_mutex.
+	 */
+	if (xfs_icsb_counter_disabled(mp, field))
+		return;
+
+	xfs_icsb_lock_all_counters(mp);
+	if (!test_and_set_bit(field, &mp->m_icsb_counters)) {
+		/* drain back to superblock */
+
+		xfs_icsb_count(mp, &cnt, XFS_ICSB_LAZY_COUNT);
+		switch(field) {
+		case XFS_SBS_ICOUNT:
+			mp->m_sb.sb_icount = cnt.icsb_icount;
+			break;
+		case XFS_SBS_IFREE:
+			mp->m_sb.sb_ifree = cnt.icsb_ifree;
+			break;
+		case XFS_SBS_FDBLOCKS:
+			mp->m_sb.sb_fdblocks = cnt.icsb_fdblocks;
+			break;
+		default:
+			BUG();
+		}
+	}
+
+	xfs_icsb_unlock_all_counters(mp);
+}
+
+STATIC void
+xfs_icsb_enable_counter(
+	xfs_mount_t	*mp,
+	xfs_sb_field_t	field,
+	uint64_t	count,
+	uint64_t	resid)
+{
+	xfs_icsb_cnts_t	*cntp;
+	int		i;
+
+	ASSERT((field >= XFS_SBS_ICOUNT) && (field <= XFS_SBS_FDBLOCKS));
+
+	xfs_icsb_lock_all_counters(mp);
+	for_each_online_cpu(i) {
+		cntp = per_cpu_ptr(mp->m_sb_cnts, i);
+		switch (field) {
+		case XFS_SBS_ICOUNT:
+			cntp->icsb_icount = count + resid;
+			break;
+		case XFS_SBS_IFREE:
+			cntp->icsb_ifree = count + resid;
+			break;
+		case XFS_SBS_FDBLOCKS:
+			cntp->icsb_fdblocks = count + resid;
+			break;
+		default:
+			BUG();
+			break;
+		}
+		resid = 0;
+	}
+	clear_bit(field, &mp->m_icsb_counters);
+	xfs_icsb_unlock_all_counters(mp);
+}
+
+void
+xfs_icsb_sync_counters_locked(
+	xfs_mount_t	*mp,
+	int		flags)
+{
+	xfs_icsb_cnts_t	cnt;
+
+	xfs_icsb_count(mp, &cnt, flags);
+
+	if (!xfs_icsb_counter_disabled(mp, XFS_SBS_ICOUNT))
+		mp->m_sb.sb_icount = cnt.icsb_icount;
+	if (!xfs_icsb_counter_disabled(mp, XFS_SBS_IFREE))
+		mp->m_sb.sb_ifree = cnt.icsb_ifree;
+	if (!xfs_icsb_counter_disabled(mp, XFS_SBS_FDBLOCKS))
+		mp->m_sb.sb_fdblocks = cnt.icsb_fdblocks;
+}
+
+/*
+ * Accurate update of per-cpu counters to incore superblock
+ */
+void
+xfs_icsb_sync_counters(
+	xfs_mount_t	*mp,
+	int		flags)
+{
+	spin_lock(&mp->m_sb_lock);
+	xfs_icsb_sync_counters_locked(mp, flags);
+	spin_unlock(&mp->m_sb_lock);
+}
+
+/*
+ * Balance and enable/disable counters as necessary.
+ *
+ * Thresholds for re-enabling counters are somewhat magic.  inode counts are
+ * chosen to be the same number as single on disk allocation chunk per CPU, and
+ * free blocks is something far enough zero that we aren't going thrash when we
+ * get near ENOSPC. We also need to supply a minimum we require per cpu to
+ * prevent looping endlessly when xfs_alloc_space asks for more than will
+ * be distributed to a single CPU but each CPU has enough blocks to be
+ * reenabled.
+ *
+ * Note that we can be called when counters are already disabled.
+ * xfs_icsb_disable_counter() optimises the counter locking in this case to
+ * prevent locking every per-cpu counter needlessly.
+ */
+
+#define XFS_ICSB_INO_CNTR_REENABLE	(uint64_t)64
+#define XFS_ICSB_FDBLK_CNTR_REENABLE(mp) \
+		(uint64_t)(512 + XFS_ALLOC_SET_ASIDE(mp))
+STATIC void
+xfs_icsb_balance_counter_locked(
+	xfs_mount_t	*mp,
+	xfs_sb_field_t  field,
+	int		min_per_cpu)
+{
+	uint64_t	count, resid;
+	int		weight = num_online_cpus();
+	uint64_t	min = (uint64_t)min_per_cpu;
+
+	/* disable counter and sync counter */
+	xfs_icsb_disable_counter(mp, field);
+
+	/* update counters  - first CPU gets residual*/
+	switch (field) {
+	case XFS_SBS_ICOUNT:
+		count = mp->m_sb.sb_icount;
+		resid = do_div(count, weight);
+		if (count < max(min, XFS_ICSB_INO_CNTR_REENABLE))
+			return;
+		break;
+	case XFS_SBS_IFREE:
+		count = mp->m_sb.sb_ifree;
+		resid = do_div(count, weight);
+		if (count < max(min, XFS_ICSB_INO_CNTR_REENABLE))
+			return;
+		break;
+	case XFS_SBS_FDBLOCKS:
+		count = mp->m_sb.sb_fdblocks;
+		resid = do_div(count, weight);
+		if (count < max(min, XFS_ICSB_FDBLK_CNTR_REENABLE(mp)))
+			return;
+		break;
+	default:
+		BUG();
+		count = resid = 0;	/* quiet, gcc */
+		break;
+	}
+
+	xfs_icsb_enable_counter(mp, field, count, resid);
+}
+
+STATIC void
+xfs_icsb_balance_counter(
+	xfs_mount_t	*mp,
+	xfs_sb_field_t  fields,
+	int		min_per_cpu)
+{
+	spin_lock(&mp->m_sb_lock);
+	xfs_icsb_balance_counter_locked(mp, fields, min_per_cpu);
+	spin_unlock(&mp->m_sb_lock);
+}
+
+int
+xfs_icsb_modify_counters(
+	xfs_mount_t	*mp,
+	xfs_sb_field_t	field,
+	int64_t		delta,
+	int		rsvd)
+{
+	xfs_icsb_cnts_t	*icsbp;
+	long long	lcounter;	/* long counter for 64 bit fields */
+	int		ret = 0;
+
+	might_sleep();
+again:
+	preempt_disable();
+	icsbp = this_cpu_ptr(mp->m_sb_cnts);
+
+	/*
+	 * if the counter is disabled, go to slow path
+	 */
+	if (unlikely(xfs_icsb_counter_disabled(mp, field)))
+		goto slow_path;
+	xfs_icsb_lock_cntr(icsbp);
+	if (unlikely(xfs_icsb_counter_disabled(mp, field))) {
+		xfs_icsb_unlock_cntr(icsbp);
+		goto slow_path;
+	}
+
+	switch (field) {
+	case XFS_SBS_ICOUNT:
+		lcounter = icsbp->icsb_icount;
+		lcounter += delta;
+		if (unlikely(lcounter < 0))
+			goto balance_counter;
+		icsbp->icsb_icount = lcounter;
+		break;
+
+	case XFS_SBS_IFREE:
+		lcounter = icsbp->icsb_ifree;
+		lcounter += delta;
+		if (unlikely(lcounter < 0))
+			goto balance_counter;
+		icsbp->icsb_ifree = lcounter;
+		break;
+
+	case XFS_SBS_FDBLOCKS:
+		BUG_ON((mp->m_resblks - mp->m_resblks_avail) != 0);
+
+		lcounter = icsbp->icsb_fdblocks - XFS_ALLOC_SET_ASIDE(mp);
+		lcounter += delta;
+		if (unlikely(lcounter < 0))
+			goto balance_counter;
+		icsbp->icsb_fdblocks = lcounter + XFS_ALLOC_SET_ASIDE(mp);
+		break;
+	default:
+		BUG();
+		break;
+	}
+	xfs_icsb_unlock_cntr(icsbp);
+	preempt_enable();
+	return 0;
+
+slow_path:
+	preempt_enable();
+
+	/*
+	 * serialise with a mutex so we don't burn lots of cpu on
+	 * the superblock lock. We still need to hold the superblock
+	 * lock, however, when we modify the global structures.
+	 */
+	xfs_icsb_lock(mp);
+
+	/*
+	 * Now running atomically.
+	 *
+	 * If the counter is enabled, someone has beaten us to rebalancing.
+	 * Drop the lock and try again in the fast path....
+	 */
+	if (!(xfs_icsb_counter_disabled(mp, field))) {
+		xfs_icsb_unlock(mp);
+		goto again;
+	}
+
+	/*
+	 * The counter is currently disabled. Because we are
+	 * running atomically here, we know a rebalance cannot
+	 * be in progress. Hence we can go straight to operating
+	 * on the global superblock. We do not call xfs_mod_incore_sb()
+	 * here even though we need to get the m_sb_lock. Doing so
+	 * will cause us to re-enter this function and deadlock.
+	 * Hence we get the m_sb_lock ourselves and then call
+	 * xfs_mod_incore_sb_unlocked() as the unlocked path operates
+	 * directly on the global counters.
+	 */
+	spin_lock(&mp->m_sb_lock);
+	ret = xfs_mod_incore_sb_unlocked(mp, field, delta, rsvd);
+	spin_unlock(&mp->m_sb_lock);
+
+	/*
+	 * Now that we've modified the global superblock, we
+	 * may be able to re-enable the distributed counters
+	 * (e.g. lots of space just got freed). After that
+	 * we are done.
+	 */
+	if (ret != ENOSPC)
+		xfs_icsb_balance_counter(mp, field, 0);
+	xfs_icsb_unlock(mp);
+	return ret;
+
+balance_counter:
+	xfs_icsb_unlock_cntr(icsbp);
+	preempt_enable();
+
+	/*
+	 * We may have multiple threads here if multiple per-cpu
+	 * counters run dry at the same time. This will mean we can
+	 * do more balances than strictly necessary but it is not
+	 * the common slowpath case.
+	 */
+	xfs_icsb_lock(mp);
+
+	/*
+	 * running atomically.
+	 *
+	 * This will leave the counter in the correct state for future
+	 * accesses. After the rebalance, we simply try again and our retry
+	 * will either succeed through the fast path or slow path without
+	 * another balance operation being required.
+	 */
+	xfs_icsb_balance_counter(mp, field, delta);
+	xfs_icsb_unlock(mp);
+	goto again;
+}
+
+#endif
diff --git a/fs/xfs/xfs_mount.h b/fs/xfs/xfs_mount.h
new file mode 100644
index 00000000..3d68bb26
--- /dev/null
+++ b/fs/xfs/xfs_mount.h
@@ -0,0 +1,402 @@
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_MOUNT_H__
+#define	__XFS_MOUNT_H__
+
+typedef struct xfs_trans_reservations {
+	uint	tr_write;	/* extent alloc trans */
+	uint	tr_itruncate;	/* truncate trans */
+	uint	tr_rename;	/* rename trans */
+	uint	tr_link;	/* link trans */
+	uint	tr_remove;	/* unlink trans */
+	uint	tr_symlink;	/* symlink trans */
+	uint	tr_create;	/* create trans */
+	uint	tr_mkdir;	/* mkdir trans */
+	uint	tr_ifree;	/* inode free trans */
+	uint	tr_ichange;	/* inode update trans */
+	uint	tr_growdata;	/* fs data section grow trans */
+	uint	tr_swrite;	/* sync write inode trans */
+	uint	tr_addafork;	/* cvt inode to attributed trans */
+	uint	tr_writeid;	/* write setuid/setgid file */
+	uint	tr_attrinval;	/* attr fork buffer invalidation */
+	uint	tr_attrset;	/* set/create an attribute */
+	uint	tr_attrrm;	/* remove an attribute */
+	uint	tr_clearagi;	/* clear bad agi unlinked ino bucket */
+	uint	tr_growrtalloc;	/* grow realtime allocations */
+	uint	tr_growrtzero;	/* grow realtime zeroing */
+	uint	tr_growrtfree;	/* grow realtime freeing */
+} xfs_trans_reservations_t;
+
+#ifndef __KERNEL__
+
+#define xfs_daddr_to_agno(mp,d) \
+	((xfs_agnumber_t)(XFS_BB_TO_FSBT(mp, d) / (mp)->m_sb.sb_agblocks))
+#define xfs_daddr_to_agbno(mp,d) \
+	((xfs_agblock_t)(XFS_BB_TO_FSBT(mp, d) % (mp)->m_sb.sb_agblocks))
+
+#else /* __KERNEL__ */
+
+#include "xfs_sync.h"
+
+struct log;
+struct xfs_mount_args;
+struct xfs_inode;
+struct xfs_bmbt_irec;
+struct xfs_bmap_free;
+struct xfs_extdelta;
+struct xfs_swapext;
+struct xfs_mru_cache;
+struct xfs_nameops;
+struct xfs_ail;
+struct xfs_quotainfo;
+
+#ifdef HAVE_PERCPU_SB
+
+/*
+ * Valid per-cpu incore superblock counters. Note that if you add new counters,
+ * you may need to define new counter disabled bit field descriptors as there
+ * are more possible fields in the superblock that can fit in a bitfield on a
+ * 32 bit platform. The XFS_SBS_* values for the current current counters just
+ * fit.
+ */
+typedef struct xfs_icsb_cnts {
+	uint64_t	icsb_fdblocks;
+	uint64_t	icsb_ifree;
+	uint64_t	icsb_icount;
+	unsigned long	icsb_flags;
+} xfs_icsb_cnts_t;
+
+#define XFS_ICSB_FLAG_LOCK	(1 << 0)	/* counter lock bit */
+
+#define XFS_ICSB_LAZY_COUNT	(1 << 1)	/* accuracy not needed */
+
+extern int	xfs_icsb_init_counters(struct xfs_mount *);
+extern void	xfs_icsb_reinit_counters(struct xfs_mount *);
+extern void	xfs_icsb_destroy_counters(struct xfs_mount *);
+extern void	xfs_icsb_sync_counters(struct xfs_mount *, int);
+extern void	xfs_icsb_sync_counters_locked(struct xfs_mount *, int);
+extern int	xfs_icsb_modify_counters(struct xfs_mount *, xfs_sb_field_t,
+						int64_t, int);
+
+#else
+#define xfs_icsb_init_counters(mp)		(0)
+#define xfs_icsb_destroy_counters(mp)		do { } while (0)
+#define xfs_icsb_reinit_counters(mp)		do { } while (0)
+#define xfs_icsb_sync_counters(mp, flags)	do { } while (0)
+#define xfs_icsb_sync_counters_locked(mp, flags) do { } while (0)
+#define xfs_icsb_modify_counters(mp, field, delta, rsvd) \
+	xfs_mod_incore_sb(mp, field, delta, rsvd)
+#endif
+
+/* dynamic preallocation free space thresholds, 5% down to 1% */
+enum {
+	XFS_LOWSP_1_PCNT = 0,
+	XFS_LOWSP_2_PCNT,
+	XFS_LOWSP_3_PCNT,
+	XFS_LOWSP_4_PCNT,
+	XFS_LOWSP_5_PCNT,
+	XFS_LOWSP_MAX,
+};
+
+typedef struct xfs_mount {
+	struct super_block	*m_super;
+	xfs_tid_t		m_tid;		/* next unused tid for fs */
+	struct xfs_ail		*m_ail;		/* fs active log item list */
+	xfs_sb_t		m_sb;		/* copy of fs superblock */
+	spinlock_t		m_sb_lock;	/* sb counter lock */
+	struct xfs_buf		*m_sb_bp;	/* buffer for superblock */
+	char			*m_fsname;	/* filesystem name */
+	int			m_fsname_len;	/* strlen of fs name */
+	char			*m_rtname;	/* realtime device name */
+	char			*m_logname;	/* external log device name */
+	int			m_bsize;	/* fs logical block size */
+	xfs_agnumber_t		m_agfrotor;	/* last ag where space found */
+	xfs_agnumber_t		m_agirotor;	/* last ag dir inode alloced */
+	spinlock_t		m_agirotor_lock;/* .. and lock protecting it */
+	xfs_agnumber_t		m_maxagi;	/* highest inode alloc group */
+	uint			m_readio_log;	/* min read size log bytes */
+	uint			m_readio_blocks; /* min read size blocks */
+	uint			m_writeio_log;	/* min write size log bytes */
+	uint			m_writeio_blocks; /* min write size blocks */
+	struct log		*m_log;		/* log specific stuff */
+	int			m_logbufs;	/* number of log buffers */
+	int			m_logbsize;	/* size of each log buffer */
+	uint			m_rsumlevels;	/* rt summary levels */
+	uint			m_rsumsize;	/* size of rt summary, bytes */
+	struct xfs_inode	*m_rbmip;	/* pointer to bitmap inode */
+	struct xfs_inode	*m_rsumip;	/* pointer to summary inode */
+	struct xfs_inode	*m_rootip;	/* pointer to root directory */
+	struct xfs_quotainfo	*m_quotainfo;	/* disk quota information */
+	xfs_buftarg_t		*m_ddev_targp;	/* saves taking the address */
+	xfs_buftarg_t		*m_logdev_targp;/* ptr to log device */
+	xfs_buftarg_t		*m_rtdev_targp;	/* ptr to rt device */
+	__uint8_t		m_blkbit_log;	/* blocklog + NBBY */
+	__uint8_t		m_blkbb_log;	/* blocklog - BBSHIFT */
+	__uint8_t		m_agno_log;	/* log #ag's */
+	__uint8_t		m_agino_log;	/* #bits for agino in inum */
+	__uint16_t		m_inode_cluster_size;/* min inode buf size */
+	uint			m_blockmask;	/* sb_blocksize-1 */
+	uint			m_blockwsize;	/* sb_blocksize in words */
+	uint			m_blockwmask;	/* blockwsize-1 */
+	uint			m_alloc_mxr[2];	/* max alloc btree records */
+	uint			m_alloc_mnr[2];	/* min alloc btree records */
+	uint			m_bmap_dmxr[2];	/* max bmap btree records */
+	uint			m_bmap_dmnr[2];	/* min bmap btree records */
+	uint			m_inobt_mxr[2];	/* max inobt btree records */
+	uint			m_inobt_mnr[2];	/* min inobt btree records */
+	uint			m_ag_maxlevels;	/* XFS_AG_MAXLEVELS */
+	uint			m_bm_maxlevels[2]; /* XFS_BM_MAXLEVELS */
+	uint			m_in_maxlevels;	/* max inobt btree levels. */
+	struct radix_tree_root	m_perag_tree;	/* per-ag accounting info */
+	spinlock_t		m_perag_lock;	/* lock for m_perag_tree */
+	struct mutex		m_growlock;	/* growfs mutex */
+	int			m_fixedfsid[2];	/* unchanged for life of FS */
+	uint			m_dmevmask;	/* DMI events for this FS */
+	__uint64_t		m_flags;	/* global mount flags */
+	uint			m_dir_node_ents; /* #entries in a dir danode */
+	uint			m_attr_node_ents; /* #entries in attr danode */
+	int			m_ialloc_inos;	/* inodes in inode allocation */
+	int			m_ialloc_blks;	/* blocks in inode allocation */
+	int			m_inoalign_mask;/* mask sb_inoalignmt if used */
+	uint			m_qflags;	/* quota status flags */
+	xfs_trans_reservations_t m_reservations;/* precomputed res values */
+	__uint64_t		m_maxicount;	/* maximum inode count */
+	__uint64_t		m_maxioffset;	/* maximum inode offset */
+	__uint64_t		m_resblks;	/* total reserved blocks */
+	__uint64_t		m_resblks_avail;/* available reserved blocks */
+	__uint64_t		m_resblks_save;	/* reserved blks @ remount,ro */
+	int			m_dalign;	/* stripe unit */
+	int			m_swidth;	/* stripe width */
+	int			m_sinoalign;	/* stripe unit inode alignment */
+	int			m_attr_magicpct;/* 37% of the blocksize */
+	int			m_dir_magicpct;	/* 37% of the dir blocksize */
+	__uint8_t		m_sectbb_log;	/* sectlog - BBSHIFT */
+	const struct xfs_nameops *m_dirnameops;	/* vector of dir name ops */
+	int			m_dirblksize;	/* directory block sz--bytes */
+	int			m_dirblkfsbs;	/* directory block sz--fsbs */
+	xfs_dablk_t		m_dirdatablk;	/* blockno of dir data v2 */
+	xfs_dablk_t		m_dirleafblk;	/* blockno of dir non-data v2 */
+	xfs_dablk_t		m_dirfreeblk;	/* blockno of dirfreeindex v2 */
+	uint			m_chsize;	/* size of next field */
+	struct xfs_chash	*m_chash;	/* fs private inode per-cluster
+						 * hash table */
+	atomic_t		m_active_trans;	/* number trans frozen */
+#ifdef HAVE_PERCPU_SB
+	xfs_icsb_cnts_t __percpu *m_sb_cnts;	/* per-cpu superblock counters */
+	unsigned long		m_icsb_counters; /* disabled per-cpu counters */
+	struct notifier_block	m_icsb_notifier; /* hotplug cpu notifier */
+	struct mutex		m_icsb_mutex;	/* balancer sync lock */
+#endif
+	struct xfs_mru_cache	*m_filestream;  /* per-mount filestream data */
+	struct delayed_work	m_sync_work;	/* background sync work */
+	struct delayed_work	m_reclaim_work;	/* background inode reclaim */
+	struct work_struct	m_flush_work;	/* background inode flush */
+	__int64_t		m_update_flags;	/* sb flags we need to update
+						   on the next remount,rw */
+	struct shrinker		m_inode_shrink;	/* inode reclaim shrinker */
+	int64_t			m_low_space[XFS_LOWSP_MAX];
+						/* low free space thresholds */
+} xfs_mount_t;
+
+/*
+ * Flags for m_flags.
+ */
+#define XFS_MOUNT_WSYNC		(1ULL << 0)	/* for nfs - all metadata ops
+						   must be synchronous except
+						   for space allocations */
+#define XFS_MOUNT_DELAYLOG	(1ULL << 1)	/* delayed logging is enabled */
+#define XFS_MOUNT_WAS_CLEAN	(1ULL << 3)
+#define XFS_MOUNT_FS_SHUTDOWN	(1ULL << 4)	/* atomic stop of all filesystem
+						   operations, typically for
+						   disk errors in metadata */
+#define XFS_MOUNT_DISCARD	(1ULL << 5)	/* discard unused blocks */
+#define XFS_MOUNT_RETERR	(1ULL << 6)     /* return alignment errors to
+						   user */
+#define XFS_MOUNT_NOALIGN	(1ULL << 7)	/* turn off stripe alignment
+						   allocations */
+#define XFS_MOUNT_ATTR2		(1ULL << 8)	/* allow use of attr2 format */
+#define XFS_MOUNT_GRPID		(1ULL << 9)	/* group-ID assigned from directory */
+#define XFS_MOUNT_NORECOVERY	(1ULL << 10)	/* no recovery - dirty fs */
+#define XFS_MOUNT_DFLT_IOSIZE	(1ULL << 12)	/* set default i/o size */
+#define XFS_MOUNT_32BITINODES	(1ULL << 14)	/* do not create inodes above
+						 * 32 bits in size */
+#define XFS_MOUNT_SMALL_INUMS	(1ULL << 15)	/* users wants 32bit inodes */
+#define XFS_MOUNT_NOUUID	(1ULL << 16)	/* ignore uuid during mount */
+#define XFS_MOUNT_BARRIER	(1ULL << 17)
+#define XFS_MOUNT_IKEEP		(1ULL << 18)	/* keep empty inode clusters*/
+#define XFS_MOUNT_SWALLOC	(1ULL << 19)	/* turn on stripe width
+						 * allocation */
+#define XFS_MOUNT_RDONLY	(1ULL << 20)	/* read-only fs */
+#define XFS_MOUNT_DIRSYNC	(1ULL << 21)	/* synchronous directory ops */
+#define XFS_MOUNT_COMPAT_IOSIZE	(1ULL << 22)	/* don't report large preferred
+						 * I/O size in stat() */
+#define XFS_MOUNT_FILESTREAMS	(1ULL << 24)	/* enable the filestreams
+						   allocator */
+#define XFS_MOUNT_NOATTR2	(1ULL << 25)	/* disable use of attr2 format */
+
+
+/*
+ * Default minimum read and write sizes.
+ */
+#define XFS_READIO_LOG_LARGE	16
+#define XFS_WRITEIO_LOG_LARGE	16
+
+/*
+ * Max and min values for mount-option defined I/O
+ * preallocation sizes.
+ */
+#define XFS_MAX_IO_LOG		30	/* 1G */
+#define XFS_MIN_IO_LOG		PAGE_SHIFT
+
+/*
+ * Synchronous read and write sizes.  This should be
+ * better for NFSv2 wsync filesystems.
+ */
+#define	XFS_WSYNC_READIO_LOG	15	/* 32k */
+#define	XFS_WSYNC_WRITEIO_LOG	14	/* 16k */
+
+/*
+ * Allow large block sizes to be reported to userspace programs if the
+ * "largeio" mount option is used.
+ *
+ * If compatibility mode is specified, simply return the basic unit of caching
+ * so that we don't get inefficient read/modify/write I/O from user apps.
+ * Otherwise....
+ *
+ * If the underlying volume is a stripe, then return the stripe width in bytes
+ * as the recommended I/O size. It is not a stripe and we've set a default
+ * buffered I/O size, return that, otherwise return the compat default.
+ */
+static inline unsigned long
+xfs_preferred_iosize(xfs_mount_t *mp)
+{
+	if (mp->m_flags & XFS_MOUNT_COMPAT_IOSIZE)
+		return PAGE_CACHE_SIZE;
+	return (mp->m_swidth ?
+		(mp->m_swidth << mp->m_sb.sb_blocklog) :
+		((mp->m_flags & XFS_MOUNT_DFLT_IOSIZE) ?
+			(1 << (int)MAX(mp->m_readio_log, mp->m_writeio_log)) :
+			PAGE_CACHE_SIZE));
+}
+
+#define XFS_MAXIOFFSET(mp)	((mp)->m_maxioffset)
+
+#define XFS_LAST_UNMOUNT_WAS_CLEAN(mp)	\
+				((mp)->m_flags & XFS_MOUNT_WAS_CLEAN)
+#define XFS_FORCED_SHUTDOWN(mp)	((mp)->m_flags & XFS_MOUNT_FS_SHUTDOWN)
+void xfs_do_force_shutdown(struct xfs_mount *mp, int flags, char *fname,
+		int lnnum);
+#define xfs_force_shutdown(m,f)	\
+	xfs_do_force_shutdown(m, f, __FILE__, __LINE__)
+
+#define SHUTDOWN_META_IO_ERROR	0x0001	/* write attempt to metadata failed */
+#define SHUTDOWN_LOG_IO_ERROR	0x0002	/* write attempt to the log failed */
+#define SHUTDOWN_FORCE_UMOUNT	0x0004	/* shutdown from a forced unmount */
+#define SHUTDOWN_CORRUPT_INCORE	0x0008	/* corrupt in-memory data structures */
+#define SHUTDOWN_REMOTE_REQ	0x0010	/* shutdown came from remote cell */
+#define SHUTDOWN_DEVICE_REQ	0x0020	/* failed all paths to the device */
+
+#define xfs_test_for_freeze(mp)		((mp)->m_super->s_frozen)
+#define xfs_wait_for_freeze(mp,l)	vfs_check_frozen((mp)->m_super, (l))
+
+/*
+ * Flags for xfs_mountfs
+ */
+#define XFS_MFSI_QUIET		0x40	/* Be silent if mount errors found */
+
+static inline xfs_agnumber_t
+xfs_daddr_to_agno(struct xfs_mount *mp, xfs_daddr_t d)
+{
+	xfs_daddr_t ld = XFS_BB_TO_FSBT(mp, d);
+	do_div(ld, mp->m_sb.sb_agblocks);
+	return (xfs_agnumber_t) ld;
+}
+
+static inline xfs_agblock_t
+xfs_daddr_to_agbno(struct xfs_mount *mp, xfs_daddr_t d)
+{
+	xfs_daddr_t ld = XFS_BB_TO_FSBT(mp, d);
+	return (xfs_agblock_t) do_div(ld, mp->m_sb.sb_agblocks);
+}
+
+/*
+ * perag get/put wrappers for ref counting
+ */
+struct xfs_perag *xfs_perag_get(struct xfs_mount *mp, xfs_agnumber_t agno);
+struct xfs_perag *xfs_perag_get_tag(struct xfs_mount *mp, xfs_agnumber_t agno,
+					int tag);
+void	xfs_perag_put(struct xfs_perag *pag);
+
+/*
+ * Per-cpu superblock locking functions
+ */
+#ifdef HAVE_PERCPU_SB
+static inline void
+xfs_icsb_lock(xfs_mount_t *mp)
+{
+	mutex_lock(&mp->m_icsb_mutex);
+}
+
+static inline void
+xfs_icsb_unlock(xfs_mount_t *mp)
+{
+	mutex_unlock(&mp->m_icsb_mutex);
+}
+#else
+#define xfs_icsb_lock(mp)
+#define xfs_icsb_unlock(mp)
+#endif
+
+/*
+ * This structure is for use by the xfs_mod_incore_sb_batch() routine.
+ * xfs_growfs can specify a few fields which are more than int limit
+ */
+typedef struct xfs_mod_sb {
+	xfs_sb_field_t	msb_field;	/* Field to modify, see below */
+	int64_t		msb_delta;	/* Change to make to specified field */
+} xfs_mod_sb_t;
+
+extern int	xfs_log_sbcount(xfs_mount_t *, uint);
+extern __uint64_t xfs_default_resblks(xfs_mount_t *mp);
+extern int	xfs_mountfs(xfs_mount_t *mp);
+
+extern void	xfs_unmountfs(xfs_mount_t *);
+extern int	xfs_unmountfs_writesb(xfs_mount_t *);
+extern int	xfs_mod_incore_sb(xfs_mount_t *, xfs_sb_field_t, int64_t, int);
+extern int	xfs_mod_incore_sb_batch(xfs_mount_t *, xfs_mod_sb_t *,
+			uint, int);
+extern int	xfs_mount_log_sb(xfs_mount_t *, __int64_t);
+extern struct xfs_buf *xfs_getsb(xfs_mount_t *, int);
+extern int	xfs_readsb(xfs_mount_t *, int);
+extern void	xfs_freesb(xfs_mount_t *);
+extern int	xfs_fs_writable(xfs_mount_t *);
+extern int	xfs_sb_validate_fsb_count(struct xfs_sb *, __uint64_t);
+
+extern int	xfs_dev_is_read_only(struct xfs_mount *, char *);
+
+extern void	xfs_set_low_space_thresholds(struct xfs_mount *);
+
+#endif	/* __KERNEL__ */
+
+extern void	xfs_mod_sb(struct xfs_trans *, __int64_t);
+extern int	xfs_initialize_perag(struct xfs_mount *, xfs_agnumber_t,
+					xfs_agnumber_t *);
+extern void	xfs_sb_from_disk(struct xfs_sb *, struct xfs_dsb *);
+extern void	xfs_sb_to_disk(struct xfs_dsb *, struct xfs_sb *, __int64_t);
+
+#endif	/* __XFS_MOUNT_H__ */
diff --git a/fs/xfs/xfs_mru_cache.c b/fs/xfs/xfs_mru_cache.c
new file mode 100644
index 00000000..4aff5639
--- /dev/null
+++ b/fs/xfs/xfs_mru_cache.c
@@ -0,0 +1,576 @@
+/*
+ * Copyright (c) 2006-2007 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_mru_cache.h"
+
+/*
+ * The MRU Cache data structure consists of a data store, an array of lists and
+ * a lock to protect its internal state.  At initialisation time, the client
+ * supplies an element lifetime in milliseconds and a group count, as well as a
+ * function pointer to call when deleting elements.  A data structure for
+ * queueing up work in the form of timed callbacks is also included.
+ *
+ * The group count controls how many lists are created, and thereby how finely
+ * the elements are grouped in time.  When reaping occurs, all the elements in
+ * all the lists whose time has expired are deleted.
+ *
+ * To give an example of how this works in practice, consider a client that
+ * initialises an MRU Cache with a lifetime of ten seconds and a group count of
+ * five.  Five internal lists will be created, each representing a two second
+ * period in time.  When the first element is added, time zero for the data
+ * structure is initialised to the current time.
+ *
+ * All the elements added in the first two seconds are appended to the first
+ * list.  Elements added in the third second go into the second list, and so on.
+ * If an element is accessed at any point, it is removed from its list and
+ * inserted at the head of the current most-recently-used list.
+ *
+ * The reaper function will have nothing to do until at least twelve seconds
+ * have elapsed since the first element was added.  The reason for this is that
+ * if it were called at t=11s, there could be elements in the first list that
+ * have only been inactive for nine seconds, so it still does nothing.  If it is
+ * called anywhere between t=12 and t=14 seconds, it will delete all the
+ * elements that remain in the first list.  It's therefore possible for elements
+ * to remain in the data store even after they've been inactive for up to
+ * (t + t/g) seconds, where t is the inactive element lifetime and g is the
+ * number of groups.
+ *
+ * The above example assumes that the reaper function gets called at least once
+ * every (t/g) seconds.  If it is called less frequently, unused elements will
+ * accumulate in the reap list until the reaper function is eventually called.
+ * The current implementation uses work queue callbacks to carefully time the
+ * reaper function calls, so this should happen rarely, if at all.
+ *
+ * From a design perspective, the primary reason for the choice of a list array
+ * representing discrete time intervals is that it's only practical to reap
+ * expired elements in groups of some appreciable size.  This automatically
+ * introduces a granularity to element lifetimes, so there's no point storing an
+ * individual timeout with each element that specifies a more precise reap time.
+ * The bonus is a saving of sizeof(long) bytes of memory per element stored.
+ *
+ * The elements could have been stored in just one list, but an array of
+ * counters or pointers would need to be maintained to allow them to be divided
+ * up into discrete time groups.  More critically, the process of touching or
+ * removing an element would involve walking large portions of the entire list,
+ * which would have a detrimental effect on performance.  The additional memory
+ * requirement for the array of list heads is minimal.
+ *
+ * When an element is touched or deleted, it needs to be removed from its
+ * current list.  Doubly linked lists are used to make the list maintenance
+ * portion of these operations O(1).  Since reaper timing can be imprecise,
+ * inserts and lookups can occur when there are no free lists available.  When
+ * this happens, all the elements on the LRU list need to be migrated to the end
+ * of the reap list.  To keep the list maintenance portion of these operations
+ * O(1) also, list tails need to be accessible without walking the entire list.
+ * This is the reason why doubly linked list heads are used.
+ */
+
+/*
+ * An MRU Cache is a dynamic data structure that stores its elements in a way
+ * that allows efficient lookups, but also groups them into discrete time
+ * intervals based on insertion time.  This allows elements to be efficiently
+ * and automatically reaped after a fixed period of inactivity.
+ *
+ * When a client data pointer is stored in the MRU Cache it needs to be added to
+ * both the data store and to one of the lists.  It must also be possible to
+ * access each of these entries via the other, i.e. to:
+ *
+ *    a) Walk a list, removing the corresponding data store entry for each item.
+ *    b) Look up a data store entry, then access its list entry directly.
+ *
+ * To achieve both of these goals, each entry must contain both a list entry and
+ * a key, in addition to the user's data pointer.  Note that it's not a good
+ * idea to have the client embed one of these structures at the top of their own
+ * data structure, because inserting the same item more than once would most
+ * likely result in a loop in one of the lists.  That's a sure-fire recipe for
+ * an infinite loop in the code.
+ */
+typedef struct xfs_mru_cache_elem
+{
+	struct list_head list_node;
+	unsigned long	key;
+	void		*value;
+} xfs_mru_cache_elem_t;
+
+static kmem_zone_t		*xfs_mru_elem_zone;
+static struct workqueue_struct	*xfs_mru_reap_wq;
+
+/*
+ * When inserting, destroying or reaping, it's first necessary to update the
+ * lists relative to a particular time.  In the case of destroying, that time
+ * will be well in the future to ensure that all items are moved to the reap
+ * list.  In all other cases though, the time will be the current time.
+ *
+ * This function enters a loop, moving the contents of the LRU list to the reap
+ * list again and again until either a) the lists are all empty, or b) time zero
+ * has been advanced sufficiently to be within the immediate element lifetime.
+ *
+ * Case a) above is detected by counting how many groups are migrated and
+ * stopping when they've all been moved.  Case b) is detected by monitoring the
+ * time_zero field, which is updated as each group is migrated.
+ *
+ * The return value is the earliest time that more migration could be needed, or
+ * zero if there's no need to schedule more work because the lists are empty.
+ */
+STATIC unsigned long
+_xfs_mru_cache_migrate(
+	xfs_mru_cache_t	*mru,
+	unsigned long	now)
+{
+	unsigned int	grp;
+	unsigned int	migrated = 0;
+	struct list_head *lru_list;
+
+	/* Nothing to do if the data store is empty. */
+	if (!mru->time_zero)
+		return 0;
+
+	/* While time zero is older than the time spanned by all the lists. */
+	while (mru->time_zero <= now - mru->grp_count * mru->grp_time) {
+
+		/*
+		 * If the LRU list isn't empty, migrate its elements to the tail
+		 * of the reap list.
+		 */
+		lru_list = mru->lists + mru->lru_grp;
+		if (!list_empty(lru_list))
+			list_splice_init(lru_list, mru->reap_list.prev);
+
+		/*
+		 * Advance the LRU group number, freeing the old LRU list to
+		 * become the new MRU list; advance time zero accordingly.
+		 */
+		mru->lru_grp = (mru->lru_grp + 1) % mru->grp_count;
+		mru->time_zero += mru->grp_time;
+
+		/*
+		 * If reaping is so far behind that all the elements on all the
+		 * lists have been migrated to the reap list, it's now empty.
+		 */
+		if (++migrated == mru->grp_count) {
+			mru->lru_grp = 0;
+			mru->time_zero = 0;
+			return 0;
+		}
+	}
+
+	/* Find the first non-empty list from the LRU end. */
+	for (grp = 0; grp < mru->grp_count; grp++) {
+
+		/* Check the grp'th list from the LRU end. */
+		lru_list = mru->lists + ((mru->lru_grp + grp) % mru->grp_count);
+		if (!list_empty(lru_list))
+			return mru->time_zero +
+			       (mru->grp_count + grp) * mru->grp_time;
+	}
+
+	/* All the lists must be empty. */
+	mru->lru_grp = 0;
+	mru->time_zero = 0;
+	return 0;
+}
+
+/*
+ * When inserting or doing a lookup, an element needs to be inserted into the
+ * MRU list.  The lists must be migrated first to ensure that they're
+ * up-to-date, otherwise the new element could be given a shorter lifetime in
+ * the cache than it should.
+ */
+STATIC void
+_xfs_mru_cache_list_insert(
+	xfs_mru_cache_t		*mru,
+	xfs_mru_cache_elem_t	*elem)
+{
+	unsigned int	grp = 0;
+	unsigned long	now = jiffies;
+
+	/*
+	 * If the data store is empty, initialise time zero, leave grp set to
+	 * zero and start the work queue timer if necessary.  Otherwise, set grp
+	 * to the number of group times that have elapsed since time zero.
+	 */
+	if (!_xfs_mru_cache_migrate(mru, now)) {
+		mru->time_zero = now;
+		if (!mru->queued) {
+			mru->queued = 1;
+			queue_delayed_work(xfs_mru_reap_wq, &mru->work,
+			                   mru->grp_count * mru->grp_time);
+		}
+	} else {
+		grp = (now - mru->time_zero) / mru->grp_time;
+		grp = (mru->lru_grp + grp) % mru->grp_count;
+	}
+
+	/* Insert the element at the tail of the corresponding list. */
+	list_add_tail(&elem->list_node, mru->lists + grp);
+}
+
+/*
+ * When destroying or reaping, all the elements that were migrated to the reap
+ * list need to be deleted.  For each element this involves removing it from the
+ * data store, removing it from the reap list, calling the client's free
+ * function and deleting the element from the element zone.
+ *
+ * We get called holding the mru->lock, which we drop and then reacquire.
+ * Sparse need special help with this to tell it we know what we are doing.
+ */
+STATIC void
+_xfs_mru_cache_clear_reap_list(
+	xfs_mru_cache_t		*mru) __releases(mru->lock) __acquires(mru->lock)
+
+{
+	xfs_mru_cache_elem_t	*elem, *next;
+	struct list_head	tmp;
+
+	INIT_LIST_HEAD(&tmp);
+	list_for_each_entry_safe(elem, next, &mru->reap_list, list_node) {
+
+		/* Remove the element from the data store. */
+		radix_tree_delete(&mru->store, elem->key);
+
+		/*
+		 * remove to temp list so it can be freed without
+		 * needing to hold the lock
+		 */
+		list_move(&elem->list_node, &tmp);
+	}
+	spin_unlock(&mru->lock);
+
+	list_for_each_entry_safe(elem, next, &tmp, list_node) {
+
+		/* Remove the element from the reap list. */
+		list_del_init(&elem->list_node);
+
+		/* Call the client's free function with the key and value pointer. */
+		mru->free_func(elem->key, elem->value);
+
+		/* Free the element structure. */
+		kmem_zone_free(xfs_mru_elem_zone, elem);
+	}
+
+	spin_lock(&mru->lock);
+}
+
+/*
+ * We fire the reap timer every group expiry interval so
+ * we always have a reaper ready to run. This makes shutdown
+ * and flushing of the reaper easy to do. Hence we need to
+ * keep when the next reap must occur so we can determine
+ * at each interval whether there is anything we need to do.
+ */
+STATIC void
+_xfs_mru_cache_reap(
+	struct work_struct	*work)
+{
+	xfs_mru_cache_t		*mru = container_of(work, xfs_mru_cache_t, work.work);
+	unsigned long		now, next;
+
+	ASSERT(mru && mru->lists);
+	if (!mru || !mru->lists)
+		return;
+
+	spin_lock(&mru->lock);
+	next = _xfs_mru_cache_migrate(mru, jiffies);
+	_xfs_mru_cache_clear_reap_list(mru);
+
+	mru->queued = next;
+	if ((mru->queued > 0)) {
+		now = jiffies;
+		if (next <= now)
+			next = 0;
+		else
+			next -= now;
+		queue_delayed_work(xfs_mru_reap_wq, &mru->work, next);
+	}
+
+	spin_unlock(&mru->lock);
+}
+
+int
+xfs_mru_cache_init(void)
+{
+	xfs_mru_elem_zone = kmem_zone_init(sizeof(xfs_mru_cache_elem_t),
+	                                 "xfs_mru_cache_elem");
+	if (!xfs_mru_elem_zone)
+		goto out;
+
+	xfs_mru_reap_wq = alloc_workqueue("xfs_mru_cache", WQ_MEM_RECLAIM, 1);
+	if (!xfs_mru_reap_wq)
+		goto out_destroy_mru_elem_zone;
+
+	return 0;
+
+ out_destroy_mru_elem_zone:
+	kmem_zone_destroy(xfs_mru_elem_zone);
+ out:
+	return -ENOMEM;
+}
+
+void
+xfs_mru_cache_uninit(void)
+{
+	destroy_workqueue(xfs_mru_reap_wq);
+	kmem_zone_destroy(xfs_mru_elem_zone);
+}
+
+/*
+ * To initialise a struct xfs_mru_cache pointer, call xfs_mru_cache_create()
+ * with the address of the pointer, a lifetime value in milliseconds, a group
+ * count and a free function to use when deleting elements.  This function
+ * returns 0 if the initialisation was successful.
+ */
+int
+xfs_mru_cache_create(
+	xfs_mru_cache_t		**mrup,
+	unsigned int		lifetime_ms,
+	unsigned int		grp_count,
+	xfs_mru_cache_free_func_t free_func)
+{
+	xfs_mru_cache_t	*mru = NULL;
+	int		err = 0, grp;
+	unsigned int	grp_time;
+
+	if (mrup)
+		*mrup = NULL;
+
+	if (!mrup || !grp_count || !lifetime_ms || !free_func)
+		return EINVAL;
+
+	if (!(grp_time = msecs_to_jiffies(lifetime_ms) / grp_count))
+		return EINVAL;
+
+	if (!(mru = kmem_zalloc(sizeof(*mru), KM_SLEEP)))
+		return ENOMEM;
+
+	/* An extra list is needed to avoid reaping up to a grp_time early. */
+	mru->grp_count = grp_count + 1;
+	mru->lists = kmem_zalloc(mru->grp_count * sizeof(*mru->lists), KM_SLEEP);
+
+	if (!mru->lists) {
+		err = ENOMEM;
+		goto exit;
+	}
+
+	for (grp = 0; grp < mru->grp_count; grp++)
+		INIT_LIST_HEAD(mru->lists + grp);
+
+	/*
+	 * We use GFP_KERNEL radix tree preload and do inserts under a
+	 * spinlock so GFP_ATOMIC is appropriate for the radix tree itself.
+	 */
+	INIT_RADIX_TREE(&mru->store, GFP_ATOMIC);
+	INIT_LIST_HEAD(&mru->reap_list);
+	spin_lock_init(&mru->lock);
+	INIT_DELAYED_WORK(&mru->work, _xfs_mru_cache_reap);
+
+	mru->grp_time  = grp_time;
+	mru->free_func = free_func;
+
+	*mrup = mru;
+
+exit:
+	if (err && mru && mru->lists)
+		kmem_free(mru->lists);
+	if (err && mru)
+		kmem_free(mru);
+
+	return err;
+}
+
+/*
+ * Call xfs_mru_cache_flush() to flush out all cached entries, calling their
+ * free functions as they're deleted.  When this function returns, the caller is
+ * guaranteed that all the free functions for all the elements have finished
+ * executing and the reaper is not running.
+ */
+static void
+xfs_mru_cache_flush(
+	xfs_mru_cache_t		*mru)
+{
+	if (!mru || !mru->lists)
+		return;
+
+	spin_lock(&mru->lock);
+	if (mru->queued) {
+		spin_unlock(&mru->lock);
+		cancel_delayed_work_sync(&mru->work);
+		spin_lock(&mru->lock);
+	}
+
+	_xfs_mru_cache_migrate(mru, jiffies + mru->grp_count * mru->grp_time);
+	_xfs_mru_cache_clear_reap_list(mru);
+
+	spin_unlock(&mru->lock);
+}
+
+void
+xfs_mru_cache_destroy(
+	xfs_mru_cache_t		*mru)
+{
+	if (!mru || !mru->lists)
+		return;
+
+	xfs_mru_cache_flush(mru);
+
+	kmem_free(mru->lists);
+	kmem_free(mru);
+}
+
+/*
+ * To insert an element, call xfs_mru_cache_insert() with the data store, the
+ * element's key and the client data pointer.  This function returns 0 on
+ * success or ENOMEM if memory for the data element couldn't be allocated.
+ */
+int
+xfs_mru_cache_insert(
+	xfs_mru_cache_t	*mru,
+	unsigned long	key,
+	void		*value)
+{
+	xfs_mru_cache_elem_t *elem;
+
+	ASSERT(mru && mru->lists);
+	if (!mru || !mru->lists)
+		return EINVAL;
+
+	elem = kmem_zone_zalloc(xfs_mru_elem_zone, KM_SLEEP);
+	if (!elem)
+		return ENOMEM;
+
+	if (radix_tree_preload(GFP_KERNEL)) {
+		kmem_zone_free(xfs_mru_elem_zone, elem);
+		return ENOMEM;
+	}
+
+	INIT_LIST_HEAD(&elem->list_node);
+	elem->key = key;
+	elem->value = value;
+
+	spin_lock(&mru->lock);
+
+	radix_tree_insert(&mru->store, key, elem);
+	radix_tree_preload_end();
+	_xfs_mru_cache_list_insert(mru, elem);
+
+	spin_unlock(&mru->lock);
+
+	return 0;
+}
+
+/*
+ * To remove an element without calling the free function, call
+ * xfs_mru_cache_remove() with the data store and the element's key.  On success
+ * the client data pointer for the removed element is returned, otherwise this
+ * function will return a NULL pointer.
+ */
+void *
+xfs_mru_cache_remove(
+	xfs_mru_cache_t	*mru,
+	unsigned long	key)
+{
+	xfs_mru_cache_elem_t *elem;
+	void		*value = NULL;
+
+	ASSERT(mru && mru->lists);
+	if (!mru || !mru->lists)
+		return NULL;
+
+	spin_lock(&mru->lock);
+	elem = radix_tree_delete(&mru->store, key);
+	if (elem) {
+		value = elem->value;
+		list_del(&elem->list_node);
+	}
+
+	spin_unlock(&mru->lock);
+
+	if (elem)
+		kmem_zone_free(xfs_mru_elem_zone, elem);
+
+	return value;
+}
+
+/*
+ * To remove and element and call the free function, call xfs_mru_cache_delete()
+ * with the data store and the element's key.
+ */
+void
+xfs_mru_cache_delete(
+	xfs_mru_cache_t	*mru,
+	unsigned long	key)
+{
+	void		*value = xfs_mru_cache_remove(mru, key);
+
+	if (value)
+		mru->free_func(key, value);
+}
+
+/*
+ * To look up an element using its key, call xfs_mru_cache_lookup() with the
+ * data store and the element's key.  If found, the element will be moved to the
+ * head of the MRU list to indicate that it's been touched.
+ *
+ * The internal data structures are protected by a spinlock that is STILL HELD
+ * when this function returns.  Call xfs_mru_cache_done() to release it.  Note
+ * that it is not safe to call any function that might sleep in the interim.
+ *
+ * The implementation could have used reference counting to avoid this
+ * restriction, but since most clients simply want to get, set or test a member
+ * of the returned data structure, the extra per-element memory isn't warranted.
+ *
+ * If the element isn't found, this function returns NULL and the spinlock is
+ * released.  xfs_mru_cache_done() should NOT be called when this occurs.
+ *
+ * Because sparse isn't smart enough to know about conditional lock return
+ * status, we need to help it get it right by annotating the path that does
+ * not release the lock.
+ */
+void *
+xfs_mru_cache_lookup(
+	xfs_mru_cache_t	*mru,
+	unsigned long	key)
+{
+	xfs_mru_cache_elem_t *elem;
+
+	ASSERT(mru && mru->lists);
+	if (!mru || !mru->lists)
+		return NULL;
+
+	spin_lock(&mru->lock);
+	elem = radix_tree_lookup(&mru->store, key);
+	if (elem) {
+		list_del(&elem->list_node);
+		_xfs_mru_cache_list_insert(mru, elem);
+		__release(mru_lock); /* help sparse not be stupid */
+	} else
+		spin_unlock(&mru->lock);
+
+	return elem ? elem->value : NULL;
+}
+
+/*
+ * To release the internal data structure spinlock after having performed an
+ * xfs_mru_cache_lookup() or an xfs_mru_cache_peek(), call xfs_mru_cache_done()
+ * with the data store pointer.
+ */
+void
+xfs_mru_cache_done(
+	xfs_mru_cache_t	*mru) __releases(mru->lock)
+{
+	spin_unlock(&mru->lock);
+}
diff --git a/fs/xfs/xfs_mru_cache.h b/fs/xfs/xfs_mru_cache.h
new file mode 100644
index 00000000..36dd3ec8
--- /dev/null
+++ b/fs/xfs/xfs_mru_cache.h
@@ -0,0 +1,53 @@
+/*
+ * Copyright (c) 2006-2007 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_MRU_CACHE_H__
+#define __XFS_MRU_CACHE_H__
+
+
+/* Function pointer type for callback to free a client's data pointer. */
+typedef void (*xfs_mru_cache_free_func_t)(unsigned long, void*);
+
+typedef struct xfs_mru_cache
+{
+	struct radix_tree_root	store;     /* Core storage data structure.  */
+	struct list_head	*lists;    /* Array of lists, one per grp.  */
+	struct list_head	reap_list; /* Elements overdue for reaping. */
+	spinlock_t		lock;      /* Lock to protect this struct.  */
+	unsigned int		grp_count; /* Number of discrete groups.    */
+	unsigned int		grp_time;  /* Time period spanned by grps.  */
+	unsigned int		lru_grp;   /* Group containing time zero.   */
+	unsigned long		time_zero; /* Time first element was added. */
+	xfs_mru_cache_free_func_t free_func; /* Function pointer for freeing. */
+	struct delayed_work	work;      /* Workqueue data for reaping.   */
+	unsigned int		queued;	   /* work has been queued */
+} xfs_mru_cache_t;
+
+int xfs_mru_cache_init(void);
+void xfs_mru_cache_uninit(void);
+int xfs_mru_cache_create(struct xfs_mru_cache **mrup, unsigned int lifetime_ms,
+			     unsigned int grp_count,
+			     xfs_mru_cache_free_func_t free_func);
+void xfs_mru_cache_destroy(struct xfs_mru_cache *mru);
+int xfs_mru_cache_insert(struct xfs_mru_cache *mru, unsigned long key,
+				void *value);
+void * xfs_mru_cache_remove(struct xfs_mru_cache *mru, unsigned long key);
+void xfs_mru_cache_delete(struct xfs_mru_cache *mru, unsigned long key);
+void *xfs_mru_cache_lookup(struct xfs_mru_cache *mru, unsigned long key);
+void xfs_mru_cache_done(struct xfs_mru_cache *mru);
+
+#endif /* __XFS_MRU_CACHE_H__ */
diff --git a/fs/xfs/xfs_quota.h b/fs/xfs/xfs_quota.h
new file mode 100644
index 00000000..a595f295
--- /dev/null
+++ b/fs/xfs/xfs_quota.h
@@ -0,0 +1,390 @@
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_QUOTA_H__
+#define __XFS_QUOTA_H__
+
+struct xfs_trans;
+
+/*
+ * The ondisk form of a dquot structure.
+ */
+#define XFS_DQUOT_MAGIC		0x4451		/* 'DQ' */
+#define XFS_DQUOT_VERSION	(u_int8_t)0x01	/* latest version number */
+
+/*
+ * uid_t and gid_t are hard-coded to 32 bits in the inode.
+ * Hence, an 'id' in a dquot is 32 bits..
+ */
+typedef __uint32_t	xfs_dqid_t;
+
+/*
+ * Even though users may not have quota limits occupying all 64-bits,
+ * they may need 64-bit accounting. Hence, 64-bit quota-counters,
+ * and quota-limits. This is a waste in the common case, but hey ...
+ */
+typedef __uint64_t	xfs_qcnt_t;
+typedef __uint16_t	xfs_qwarncnt_t;
+
+/*
+ * This is the main portion of the on-disk representation of quota
+ * information for a user. This is the q_core of the xfs_dquot_t that
+ * is kept in kernel memory. We pad this with some more expansion room
+ * to construct the on disk structure.
+ */
+typedef struct	xfs_disk_dquot {
+	__be16		d_magic;	/* dquot magic = XFS_DQUOT_MAGIC */
+	__u8		d_version;	/* dquot version */
+	__u8		d_flags;	/* XFS_DQ_USER/PROJ/GROUP */
+	__be32		d_id;		/* user,project,group id */
+	__be64		d_blk_hardlimit;/* absolute limit on disk blks */
+	__be64		d_blk_softlimit;/* preferred limit on disk blks */
+	__be64		d_ino_hardlimit;/* maximum # allocated inodes */
+	__be64		d_ino_softlimit;/* preferred inode limit */
+	__be64		d_bcount;	/* disk blocks owned by the user */
+	__be64		d_icount;	/* inodes owned by the user */
+	__be32		d_itimer;	/* zero if within inode limits if not,
+					   this is when we refuse service */
+	__be32		d_btimer;	/* similar to above; for disk blocks */
+	__be16		d_iwarns;	/* warnings issued wrt num inodes */
+	__be16		d_bwarns;	/* warnings issued wrt disk blocks */
+	__be32		d_pad0;		/* 64 bit align */
+	__be64		d_rtb_hardlimit;/* absolute limit on realtime blks */
+	__be64		d_rtb_softlimit;/* preferred limit on RT disk blks */
+	__be64		d_rtbcount;	/* realtime blocks owned */
+	__be32		d_rtbtimer;	/* similar to above; for RT disk blocks */
+	__be16		d_rtbwarns;	/* warnings issued wrt RT disk blocks */
+	__be16		d_pad;
+} xfs_disk_dquot_t;
+
+/*
+ * This is what goes on disk. This is separated from the xfs_disk_dquot because
+ * carrying the unnecessary padding would be a waste of memory.
+ */
+typedef struct xfs_dqblk {
+	xfs_disk_dquot_t  dd_diskdq;	/* portion that lives incore as well */
+	char		  dd_fill[32];	/* filling for posterity */
+} xfs_dqblk_t;
+
+/*
+ * flags for q_flags field in the dquot.
+ */
+#define XFS_DQ_USER		0x0001		/* a user quota */
+#define XFS_DQ_PROJ		0x0002		/* project quota */
+#define XFS_DQ_GROUP		0x0004		/* a group quota */
+#define XFS_DQ_DIRTY		0x0008		/* dquot is dirty */
+#define XFS_DQ_WANT		0x0010		/* for lookup/reclaim race */
+#define XFS_DQ_INACTIVE		0x0020		/* dq off mplist & hashlist */
+
+#define XFS_DQ_ALLTYPES		(XFS_DQ_USER|XFS_DQ_PROJ|XFS_DQ_GROUP)
+
+#define XFS_DQ_FLAGS \
+	{ XFS_DQ_USER,		"USER" }, \
+	{ XFS_DQ_PROJ,		"PROJ" }, \
+	{ XFS_DQ_GROUP,		"GROUP" }, \
+	{ XFS_DQ_DIRTY,		"DIRTY" }, \
+	{ XFS_DQ_WANT,		"WANT" }, \
+	{ XFS_DQ_INACTIVE,	"INACTIVE" }
+
+/*
+ * In the worst case, when both user and group quotas are on,
+ * we can have a max of three dquots changing in a single transaction.
+ */
+#define XFS_DQUOT_LOGRES(mp)	(sizeof(xfs_disk_dquot_t) * 3)
+
+
+/*
+ * These are the structures used to lay out dquots and quotaoff
+ * records on the log. Quite similar to those of inodes.
+ */
+
+/*
+ * log format struct for dquots.
+ * The first two fields must be the type and size fitting into
+ * 32 bits : log_recovery code assumes that.
+ */
+typedef struct xfs_dq_logformat {
+	__uint16_t		qlf_type;      /* dquot log item type */
+	__uint16_t		qlf_size;      /* size of this item */
+	xfs_dqid_t		qlf_id;	       /* usr/grp/proj id : 32 bits */
+	__int64_t		qlf_blkno;     /* blkno of dquot buffer */
+	__int32_t		qlf_len;       /* len of dquot buffer */
+	__uint32_t		qlf_boffset;   /* off of dquot in buffer */
+} xfs_dq_logformat_t;
+
+/*
+ * log format struct for QUOTAOFF records.
+ * The first two fields must be the type and size fitting into
+ * 32 bits : log_recovery code assumes that.
+ * We write two LI_QUOTAOFF logitems per quotaoff, the last one keeps a pointer
+ * to the first and ensures that the first logitem is taken out of the AIL
+ * only when the last one is securely committed.
+ */
+typedef struct xfs_qoff_logformat {
+	unsigned short		qf_type;	/* quotaoff log item type */
+	unsigned short		qf_size;	/* size of this item */
+	unsigned int		qf_flags;	/* USR and/or GRP */
+	char			qf_pad[12];	/* padding for future */
+} xfs_qoff_logformat_t;
+
+
+/*
+ * Disk quotas status in m_qflags, and also sb_qflags. 16 bits.
+ */
+#define XFS_UQUOTA_ACCT	0x0001  /* user quota accounting ON */
+#define XFS_UQUOTA_ENFD	0x0002  /* user quota limits enforced */
+#define XFS_UQUOTA_CHKD	0x0004  /* quotacheck run on usr quotas */
+#define XFS_PQUOTA_ACCT	0x0008  /* project quota accounting ON */
+#define XFS_OQUOTA_ENFD	0x0010  /* other (grp/prj) quota limits enforced */
+#define XFS_OQUOTA_CHKD	0x0020  /* quotacheck run on other (grp/prj) quotas */
+#define XFS_GQUOTA_ACCT	0x0040  /* group quota accounting ON */
+
+/*
+ * Quota Accounting/Enforcement flags
+ */
+#define XFS_ALL_QUOTA_ACCT	\
+		(XFS_UQUOTA_ACCT | XFS_GQUOTA_ACCT | XFS_PQUOTA_ACCT)
+#define XFS_ALL_QUOTA_ENFD	(XFS_UQUOTA_ENFD | XFS_OQUOTA_ENFD)
+#define XFS_ALL_QUOTA_CHKD	(XFS_UQUOTA_CHKD | XFS_OQUOTA_CHKD)
+
+#define XFS_IS_QUOTA_RUNNING(mp)	((mp)->m_qflags & XFS_ALL_QUOTA_ACCT)
+#define XFS_IS_UQUOTA_RUNNING(mp)	((mp)->m_qflags & XFS_UQUOTA_ACCT)
+#define XFS_IS_PQUOTA_RUNNING(mp)	((mp)->m_qflags & XFS_PQUOTA_ACCT)
+#define XFS_IS_GQUOTA_RUNNING(mp)	((mp)->m_qflags & XFS_GQUOTA_ACCT)
+#define XFS_IS_UQUOTA_ENFORCED(mp)	((mp)->m_qflags & XFS_UQUOTA_ENFD)
+#define XFS_IS_OQUOTA_ENFORCED(mp)	((mp)->m_qflags & XFS_OQUOTA_ENFD)
+
+/*
+ * Incore only flags for quotaoff - these bits get cleared when quota(s)
+ * are in the process of getting turned off. These flags are in m_qflags but
+ * never in sb_qflags.
+ */
+#define XFS_UQUOTA_ACTIVE	0x0100  /* uquotas are being turned off */
+#define XFS_PQUOTA_ACTIVE	0x0200  /* pquotas are being turned off */
+#define XFS_GQUOTA_ACTIVE	0x0400  /* gquotas are being turned off */
+
+/*
+ * Checking XFS_IS_*QUOTA_ON() while holding any inode lock guarantees
+ * quota will be not be switched off as long as that inode lock is held.
+ */
+#define XFS_IS_QUOTA_ON(mp)	((mp)->m_qflags & (XFS_UQUOTA_ACTIVE | \
+						   XFS_GQUOTA_ACTIVE | \
+						   XFS_PQUOTA_ACTIVE))
+#define XFS_IS_OQUOTA_ON(mp)	((mp)->m_qflags & (XFS_GQUOTA_ACTIVE | \
+						   XFS_PQUOTA_ACTIVE))
+#define XFS_IS_UQUOTA_ON(mp)	((mp)->m_qflags & XFS_UQUOTA_ACTIVE)
+#define XFS_IS_GQUOTA_ON(mp)	((mp)->m_qflags & XFS_GQUOTA_ACTIVE)
+#define XFS_IS_PQUOTA_ON(mp)	((mp)->m_qflags & XFS_PQUOTA_ACTIVE)
+
+/*
+ * Flags to tell various functions what to do. Not all of these are meaningful
+ * to a single function. None of these XFS_QMOPT_* flags are meant to have
+ * persistent values (ie. their values can and will change between versions)
+ */
+#define XFS_QMOPT_DQALLOC	0x0000002 /* alloc dquot ondisk if needed */
+#define XFS_QMOPT_UQUOTA	0x0000004 /* user dquot requested */
+#define XFS_QMOPT_PQUOTA	0x0000008 /* project dquot requested */
+#define XFS_QMOPT_FORCE_RES	0x0000010 /* ignore quota limits */
+#define XFS_QMOPT_DQSUSER	0x0000020 /* don't cache super users dquot */
+#define XFS_QMOPT_SBVERSION	0x0000040 /* change superblock version num */
+#define XFS_QMOPT_DOWARN        0x0000400 /* increase warning cnt if needed */
+#define XFS_QMOPT_DQREPAIR	0x0001000 /* repair dquot if damaged */
+#define XFS_QMOPT_GQUOTA	0x0002000 /* group dquot requested */
+#define XFS_QMOPT_ENOSPC	0x0004000 /* enospc instead of edquot (prj) */
+
+/*
+ * flags to xfs_trans_mod_dquot to indicate which field needs to be
+ * modified.
+ */
+#define XFS_QMOPT_RES_REGBLKS	0x0010000
+#define XFS_QMOPT_RES_RTBLKS	0x0020000
+#define XFS_QMOPT_BCOUNT	0x0040000
+#define XFS_QMOPT_ICOUNT	0x0080000
+#define XFS_QMOPT_RTBCOUNT	0x0100000
+#define XFS_QMOPT_DELBCOUNT	0x0200000
+#define XFS_QMOPT_DELRTBCOUNT	0x0400000
+#define XFS_QMOPT_RES_INOS	0x0800000
+
+/*
+ * flags for dqalloc.
+ */
+#define XFS_QMOPT_INHERIT	0x1000000
+
+/*
+ * flags to xfs_trans_mod_dquot.
+ */
+#define XFS_TRANS_DQ_RES_BLKS	XFS_QMOPT_RES_REGBLKS
+#define XFS_TRANS_DQ_RES_RTBLKS	XFS_QMOPT_RES_RTBLKS
+#define XFS_TRANS_DQ_RES_INOS	XFS_QMOPT_RES_INOS
+#define XFS_TRANS_DQ_BCOUNT	XFS_QMOPT_BCOUNT
+#define XFS_TRANS_DQ_DELBCOUNT	XFS_QMOPT_DELBCOUNT
+#define XFS_TRANS_DQ_ICOUNT	XFS_QMOPT_ICOUNT
+#define XFS_TRANS_DQ_RTBCOUNT	XFS_QMOPT_RTBCOUNT
+#define XFS_TRANS_DQ_DELRTBCOUNT XFS_QMOPT_DELRTBCOUNT
+
+
+#define XFS_QMOPT_QUOTALL	\
+		(XFS_QMOPT_UQUOTA | XFS_QMOPT_PQUOTA | XFS_QMOPT_GQUOTA)
+#define XFS_QMOPT_RESBLK_MASK	(XFS_QMOPT_RES_REGBLKS | XFS_QMOPT_RES_RTBLKS)
+
+#ifdef __KERNEL__
+/*
+ * This check is done typically without holding the inode lock;
+ * that may seem racy, but it is harmless in the context that it is used.
+ * The inode cannot go inactive as long a reference is kept, and
+ * therefore if dquot(s) were attached, they'll stay consistent.
+ * If, for example, the ownership of the inode changes while
+ * we didn't have the inode locked, the appropriate dquot(s) will be
+ * attached atomically.
+ */
+#define XFS_NOT_DQATTACHED(mp, ip) ((XFS_IS_UQUOTA_ON(mp) &&\
+				     (ip)->i_udquot == NULL) || \
+				    (XFS_IS_OQUOTA_ON(mp) && \
+				     (ip)->i_gdquot == NULL))
+
+#define XFS_QM_NEED_QUOTACHECK(mp) \
+	((XFS_IS_UQUOTA_ON(mp) && \
+		(mp->m_sb.sb_qflags & XFS_UQUOTA_CHKD) == 0) || \
+	 (XFS_IS_GQUOTA_ON(mp) && \
+		((mp->m_sb.sb_qflags & XFS_OQUOTA_CHKD) == 0 || \
+		 (mp->m_sb.sb_qflags & XFS_PQUOTA_ACCT))) || \
+	 (XFS_IS_PQUOTA_ON(mp) && \
+		((mp->m_sb.sb_qflags & XFS_OQUOTA_CHKD) == 0 || \
+		 (mp->m_sb.sb_qflags & XFS_GQUOTA_ACCT))))
+
+#define XFS_MOUNT_QUOTA_SET1	(XFS_UQUOTA_ACCT|XFS_UQUOTA_ENFD|\
+				 XFS_UQUOTA_CHKD|XFS_PQUOTA_ACCT|\
+				 XFS_OQUOTA_ENFD|XFS_OQUOTA_CHKD)
+
+#define XFS_MOUNT_QUOTA_SET2	(XFS_UQUOTA_ACCT|XFS_UQUOTA_ENFD|\
+				 XFS_UQUOTA_CHKD|XFS_GQUOTA_ACCT|\
+				 XFS_OQUOTA_ENFD|XFS_OQUOTA_CHKD)
+
+#define XFS_MOUNT_QUOTA_ALL	(XFS_UQUOTA_ACCT|XFS_UQUOTA_ENFD|\
+				 XFS_UQUOTA_CHKD|XFS_PQUOTA_ACCT|\
+				 XFS_OQUOTA_ENFD|XFS_OQUOTA_CHKD|\
+				 XFS_GQUOTA_ACCT)
+
+
+/*
+ * The structure kept inside the xfs_trans_t keep track of dquot changes
+ * within a transaction and apply them later.
+ */
+typedef struct xfs_dqtrx {
+	struct xfs_dquot *qt_dquot;	  /* the dquot this refers to */
+	ulong		qt_blk_res;	  /* blks reserved on a dquot */
+	ulong		qt_blk_res_used;  /* blks used from the reservation */
+	ulong		qt_ino_res;	  /* inode reserved on a dquot */
+	ulong		qt_ino_res_used;  /* inodes used from the reservation */
+	long		qt_bcount_delta;  /* dquot blk count changes */
+	long		qt_delbcnt_delta; /* delayed dquot blk count changes */
+	long		qt_icount_delta;  /* dquot inode count changes */
+	ulong		qt_rtblk_res;	  /* # blks reserved on a dquot */
+	ulong		qt_rtblk_res_used;/* # blks used from reservation */
+	long		qt_rtbcount_delta;/* dquot realtime blk changes */
+	long		qt_delrtb_delta;  /* delayed RT blk count changes */
+} xfs_dqtrx_t;
+
+#ifdef CONFIG_XFS_QUOTA
+extern void xfs_trans_dup_dqinfo(struct xfs_trans *, struct xfs_trans *);
+extern void xfs_trans_free_dqinfo(struct xfs_trans *);
+extern void xfs_trans_mod_dquot_byino(struct xfs_trans *, struct xfs_inode *,
+		uint, long);
+extern void xfs_trans_apply_dquot_deltas(struct xfs_trans *);
+extern void xfs_trans_unreserve_and_mod_dquots(struct xfs_trans *);
+extern int xfs_trans_reserve_quota_nblks(struct xfs_trans *,
+		struct xfs_inode *, long, long, uint);
+extern int xfs_trans_reserve_quota_bydquots(struct xfs_trans *,
+		struct xfs_mount *, struct xfs_dquot *,
+		struct xfs_dquot *, long, long, uint);
+
+extern int xfs_qm_vop_dqalloc(struct xfs_inode *, uid_t, gid_t, prid_t, uint,
+		struct xfs_dquot **, struct xfs_dquot **);
+extern void xfs_qm_vop_create_dqattach(struct xfs_trans *, struct xfs_inode *,
+		struct xfs_dquot *, struct xfs_dquot *);
+extern int xfs_qm_vop_rename_dqattach(struct xfs_inode **);
+extern struct xfs_dquot *xfs_qm_vop_chown(struct xfs_trans *,
+		struct xfs_inode *, struct xfs_dquot **, struct xfs_dquot *);
+extern int xfs_qm_vop_chown_reserve(struct xfs_trans *, struct xfs_inode *,
+		struct xfs_dquot *, struct xfs_dquot *, uint);
+extern int xfs_qm_dqattach(struct xfs_inode *, uint);
+extern int xfs_qm_dqattach_locked(struct xfs_inode *, uint);
+extern void xfs_qm_dqdetach(struct xfs_inode *);
+extern void xfs_qm_dqrele(struct xfs_dquot *);
+extern void xfs_qm_statvfs(struct xfs_inode *, struct kstatfs *);
+extern int xfs_qm_sync(struct xfs_mount *, int);
+extern int xfs_qm_newmount(struct xfs_mount *, uint *, uint *);
+extern void xfs_qm_mount_quotas(struct xfs_mount *);
+extern void xfs_qm_unmount(struct xfs_mount *);
+extern void xfs_qm_unmount_quotas(struct xfs_mount *);
+
+#else
+static inline int
+xfs_qm_vop_dqalloc(struct xfs_inode *ip, uid_t uid, gid_t gid, prid_t prid,
+		uint flags, struct xfs_dquot **udqp, struct xfs_dquot **gdqp)
+{
+	*udqp = NULL;
+	*gdqp = NULL;
+	return 0;
+}
+#define xfs_trans_dup_dqinfo(tp, tp2)
+#define xfs_trans_free_dqinfo(tp)
+#define xfs_trans_mod_dquot_byino(tp, ip, fields, delta)
+#define xfs_trans_apply_dquot_deltas(tp)
+#define xfs_trans_unreserve_and_mod_dquots(tp)
+static inline int xfs_trans_reserve_quota_nblks(struct xfs_trans *tp,
+		struct xfs_inode *ip, long nblks, long ninos, uint flags)
+{
+	return 0;
+}
+static inline int xfs_trans_reserve_quota_bydquots(struct xfs_trans *tp,
+		struct xfs_mount *mp, struct xfs_dquot *udqp,
+		struct xfs_dquot *gdqp, long nblks, long nions, uint flags)
+{
+	return 0;
+}
+#define xfs_qm_vop_create_dqattach(tp, ip, u, g)
+#define xfs_qm_vop_rename_dqattach(it)					(0)
+#define xfs_qm_vop_chown(tp, ip, old, new)				(NULL)
+#define xfs_qm_vop_chown_reserve(tp, ip, u, g, fl)			(0)
+#define xfs_qm_dqattach(ip, fl)						(0)
+#define xfs_qm_dqattach_locked(ip, fl)					(0)
+#define xfs_qm_dqdetach(ip)
+#define xfs_qm_dqrele(d)
+#define xfs_qm_statvfs(ip, s)
+static inline int xfs_qm_sync(struct xfs_mount *mp, int flags)
+{
+	return 0;
+}
+#define xfs_qm_newmount(mp, a, b)					(0)
+#define xfs_qm_mount_quotas(mp)
+#define xfs_qm_unmount(mp)
+#define xfs_qm_unmount_quotas(mp)
+#endif /* CONFIG_XFS_QUOTA */
+
+#define xfs_trans_unreserve_quota_nblks(tp, ip, nblks, ninos, flags) \
+	xfs_trans_reserve_quota_nblks(tp, ip, -(nblks), -(ninos), flags)
+#define xfs_trans_reserve_quota(tp, mp, ud, gd, nb, ni, f) \
+	xfs_trans_reserve_quota_bydquots(tp, mp, ud, gd, nb, ni, \
+				f | XFS_QMOPT_RES_REGBLKS)
+
+extern int xfs_qm_dqcheck(struct xfs_mount *, xfs_disk_dquot_t *,
+				xfs_dqid_t, uint, uint, char *);
+extern int xfs_mount_reset_sbqflags(struct xfs_mount *);
+
+#endif	/* __KERNEL__ */
+#endif	/* __XFS_QUOTA_H__ */
diff --git a/fs/xfs/xfs_rename.c b/fs/xfs/xfs_rename.c
new file mode 100644
index 00000000..77a59891
--- /dev/null
+++ b/fs/xfs/xfs_rename.c
@@ -0,0 +1,358 @@
+/*
+ * Copyright (c) 2000-2003,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_dir2.h"
+#include "xfs_mount.h"
+#include "xfs_da_btree.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_inode_item.h"
+#include "xfs_bmap.h"
+#include "xfs_error.h"
+#include "xfs_quota.h"
+#include "xfs_utils.h"
+#include "xfs_trans_space.h"
+#include "xfs_vnodeops.h"
+#include "xfs_trace.h"
+
+
+/*
+ * Enter all inodes for a rename transaction into a sorted array.
+ */
+STATIC void
+xfs_sort_for_rename(
+	xfs_inode_t	*dp1,	/* in: old (source) directory inode */
+	xfs_inode_t	*dp2,	/* in: new (target) directory inode */
+	xfs_inode_t	*ip1,	/* in: inode of old entry */
+	xfs_inode_t	*ip2,	/* in: inode of new entry, if it
+				   already exists, NULL otherwise. */
+	xfs_inode_t	**i_tab,/* out: array of inode returned, sorted */
+	int		*num_inodes)  /* out: number of inodes in array */
+{
+	xfs_inode_t		*temp;
+	int			i, j;
+
+	/*
+	 * i_tab contains a list of pointers to inodes.  We initialize
+	 * the table here & we'll sort it.  We will then use it to
+	 * order the acquisition of the inode locks.
+	 *
+	 * Note that the table may contain duplicates.  e.g., dp1 == dp2.
+	 */
+	i_tab[0] = dp1;
+	i_tab[1] = dp2;
+	i_tab[2] = ip1;
+	if (ip2) {
+		*num_inodes = 4;
+		i_tab[3] = ip2;
+	} else {
+		*num_inodes = 3;
+		i_tab[3] = NULL;
+	}
+
+	/*
+	 * Sort the elements via bubble sort.  (Remember, there are at
+	 * most 4 elements to sort, so this is adequate.)
+	 */
+	for (i = 0; i < *num_inodes; i++) {
+		for (j = 1; j < *num_inodes; j++) {
+			if (i_tab[j]->i_ino < i_tab[j-1]->i_ino) {
+				temp = i_tab[j];
+				i_tab[j] = i_tab[j-1];
+				i_tab[j-1] = temp;
+			}
+		}
+	}
+}
+
+/*
+ * xfs_rename
+ */
+int
+xfs_rename(
+	xfs_inode_t	*src_dp,
+	struct xfs_name	*src_name,
+	xfs_inode_t	*src_ip,
+	xfs_inode_t	*target_dp,
+	struct xfs_name	*target_name,
+	xfs_inode_t	*target_ip)
+{
+	xfs_trans_t	*tp = NULL;
+	xfs_mount_t	*mp = src_dp->i_mount;
+	int		new_parent;		/* moving to a new dir */
+	int		src_is_directory;	/* src_name is a directory */
+	int		error;
+	xfs_bmap_free_t free_list;
+	xfs_fsblock_t   first_block;
+	int		cancel_flags;
+	int		committed;
+	xfs_inode_t	*inodes[4];
+	int		spaceres;
+	int		num_inodes;
+
+	trace_xfs_rename(src_dp, target_dp, src_name, target_name);
+
+	new_parent = (src_dp != target_dp);
+	src_is_directory = ((src_ip->i_d.di_mode & S_IFMT) == S_IFDIR);
+
+	if (src_is_directory) {
+		/*
+		 * Check for link count overflow on target_dp
+		 */
+		if (target_ip == NULL && new_parent &&
+		    target_dp->i_d.di_nlink >= XFS_MAXLINK) {
+			error = XFS_ERROR(EMLINK);
+			goto std_return;
+		}
+	}
+
+	xfs_sort_for_rename(src_dp, target_dp, src_ip, target_ip,
+				inodes, &num_inodes);
+
+	xfs_bmap_init(&free_list, &first_block);
+	tp = xfs_trans_alloc(mp, XFS_TRANS_RENAME);
+	cancel_flags = XFS_TRANS_RELEASE_LOG_RES;
+	spaceres = XFS_RENAME_SPACE_RES(mp, target_name->len);
+	error = xfs_trans_reserve(tp, spaceres, XFS_RENAME_LOG_RES(mp), 0,
+			XFS_TRANS_PERM_LOG_RES, XFS_RENAME_LOG_COUNT);
+	if (error == ENOSPC) {
+		spaceres = 0;
+		error = xfs_trans_reserve(tp, 0, XFS_RENAME_LOG_RES(mp), 0,
+				XFS_TRANS_PERM_LOG_RES, XFS_RENAME_LOG_COUNT);
+	}
+	if (error) {
+		xfs_trans_cancel(tp, 0);
+		goto std_return;
+	}
+
+	/*
+	 * Attach the dquots to the inodes
+	 */
+	error = xfs_qm_vop_rename_dqattach(inodes);
+	if (error) {
+		xfs_trans_cancel(tp, cancel_flags);
+		goto std_return;
+	}
+
+	/*
+	 * Lock all the participating inodes. Depending upon whether
+	 * the target_name exists in the target directory, and
+	 * whether the target directory is the same as the source
+	 * directory, we can lock from 2 to 4 inodes.
+	 */
+	xfs_lock_inodes(inodes, num_inodes, XFS_ILOCK_EXCL);
+
+	/*
+	 * Join all the inodes to the transaction. From this point on,
+	 * we can rely on either trans_commit or trans_cancel to unlock
+	 * them.
+	 */
+	xfs_trans_ijoin_ref(tp, src_dp, XFS_ILOCK_EXCL);
+	if (new_parent)
+		xfs_trans_ijoin_ref(tp, target_dp, XFS_ILOCK_EXCL);
+	xfs_trans_ijoin_ref(tp, src_ip, XFS_ILOCK_EXCL);
+	if (target_ip)
+		xfs_trans_ijoin_ref(tp, target_ip, XFS_ILOCK_EXCL);
+
+	/*
+	 * If we are using project inheritance, we only allow renames
+	 * into our tree when the project IDs are the same; else the
+	 * tree quota mechanism would be circumvented.
+	 */
+	if (unlikely((target_dp->i_d.di_flags & XFS_DIFLAG_PROJINHERIT) &&
+		     (xfs_get_projid(target_dp) != xfs_get_projid(src_ip)))) {
+		error = XFS_ERROR(EXDEV);
+		goto error_return;
+	}
+
+	/*
+	 * Set up the target.
+	 */
+	if (target_ip == NULL) {
+		/*
+		 * If there's no space reservation, check the entry will
+		 * fit before actually inserting it.
+		 */
+		error = xfs_dir_canenter(tp, target_dp, target_name, spaceres);
+		if (error)
+			goto error_return;
+		/*
+		 * If target does not exist and the rename crosses
+		 * directories, adjust the target directory link count
+		 * to account for the ".." reference from the new entry.
+		 */
+		error = xfs_dir_createname(tp, target_dp, target_name,
+						src_ip->i_ino, &first_block,
+						&free_list, spaceres);
+		if (error == ENOSPC)
+			goto error_return;
+		if (error)
+			goto abort_return;
+
+		xfs_trans_ichgtime(tp, target_dp,
+					XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG);
+
+		if (new_parent && src_is_directory) {
+			error = xfs_bumplink(tp, target_dp);
+			if (error)
+				goto abort_return;
+		}
+	} else { /* target_ip != NULL */
+		/*
+		 * If target exists and it's a directory, check that both
+		 * target and source are directories and that target can be
+		 * destroyed, or that neither is a directory.
+		 */
+		if ((target_ip->i_d.di_mode & S_IFMT) == S_IFDIR) {
+			/*
+			 * Make sure target dir is empty.
+			 */
+			if (!(xfs_dir_isempty(target_ip)) ||
+			    (target_ip->i_d.di_nlink > 2)) {
+				error = XFS_ERROR(EEXIST);
+				goto error_return;
+			}
+		}
+
+		/*
+		 * Link the source inode under the target name.
+		 * If the source inode is a directory and we are moving
+		 * it across directories, its ".." entry will be
+		 * inconsistent until we replace that down below.
+		 *
+		 * In case there is already an entry with the same
+		 * name at the destination directory, remove it first.
+		 */
+		error = xfs_dir_replace(tp, target_dp, target_name,
+					src_ip->i_ino,
+					&first_block, &free_list, spaceres);
+		if (error)
+			goto abort_return;
+
+		xfs_trans_ichgtime(tp, target_dp,
+					XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG);
+
+		/*
+		 * Decrement the link count on the target since the target
+		 * dir no longer points to it.
+		 */
+		error = xfs_droplink(tp, target_ip);
+		if (error)
+			goto abort_return;
+
+		if (src_is_directory) {
+			/*
+			 * Drop the link from the old "." entry.
+			 */
+			error = xfs_droplink(tp, target_ip);
+			if (error)
+				goto abort_return;
+		}
+	} /* target_ip != NULL */
+
+	/*
+	 * Remove the source.
+	 */
+	if (new_parent && src_is_directory) {
+		/*
+		 * Rewrite the ".." entry to point to the new
+		 * directory.
+		 */
+		error = xfs_dir_replace(tp, src_ip, &xfs_name_dotdot,
+					target_dp->i_ino,
+					&first_block, &free_list, spaceres);
+		ASSERT(error != EEXIST);
+		if (error)
+			goto abort_return;
+	}
+
+	/*
+	 * We always want to hit the ctime on the source inode.
+	 *
+	 * This isn't strictly required by the standards since the source
+	 * inode isn't really being changed, but old unix file systems did
+	 * it and some incremental backup programs won't work without it.
+	 */
+	xfs_trans_ichgtime(tp, src_ip, XFS_ICHGTIME_CHG);
+	xfs_trans_log_inode(tp, src_ip, XFS_ILOG_CORE);
+
+	/*
+	 * Adjust the link count on src_dp.  This is necessary when
+	 * renaming a directory, either within one parent when
+	 * the target existed, or across two parent directories.
+	 */
+	if (src_is_directory && (new_parent || target_ip != NULL)) {
+
+		/*
+		 * Decrement link count on src_directory since the
+		 * entry that's moved no longer points to it.
+		 */
+		error = xfs_droplink(tp, src_dp);
+		if (error)
+			goto abort_return;
+	}
+
+	error = xfs_dir_removename(tp, src_dp, src_name, src_ip->i_ino,
+					&first_block, &free_list, spaceres);
+	if (error)
+		goto abort_return;
+
+	xfs_trans_ichgtime(tp, src_dp, XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG);
+	xfs_trans_log_inode(tp, src_dp, XFS_ILOG_CORE);
+	if (new_parent)
+		xfs_trans_log_inode(tp, target_dp, XFS_ILOG_CORE);
+
+	/*
+	 * If this is a synchronous mount, make sure that the
+	 * rename transaction goes to disk before returning to
+	 * the user.
+	 */
+	if (mp->m_flags & (XFS_MOUNT_WSYNC|XFS_MOUNT_DIRSYNC)) {
+		xfs_trans_set_sync(tp);
+	}
+
+	error = xfs_bmap_finish(&tp, &free_list, &committed);
+	if (error) {
+		xfs_bmap_cancel(&free_list);
+		xfs_trans_cancel(tp, (XFS_TRANS_RELEASE_LOG_RES |
+				 XFS_TRANS_ABORT));
+		goto std_return;
+	}
+
+	/*
+	 * trans_commit will unlock src_ip, target_ip & decrement
+	 * the vnode references.
+	 */
+	return xfs_trans_commit(tp, XFS_TRANS_RELEASE_LOG_RES);
+
+ abort_return:
+	cancel_flags |= XFS_TRANS_ABORT;
+ error_return:
+	xfs_bmap_cancel(&free_list);
+	xfs_trans_cancel(tp, cancel_flags);
+ std_return:
+	return error;
+}
diff --git a/fs/xfs/xfs_rtalloc.c b/fs/xfs/xfs_rtalloc.c
new file mode 100644
index 00000000..8f76fdff
--- /dev/null
+++ b/fs/xfs/xfs_rtalloc.c
@@ -0,0 +1,2325 @@
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_dir2.h"
+#include "xfs_mount.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_alloc.h"
+#include "xfs_bmap.h"
+#include "xfs_rtalloc.h"
+#include "xfs_fsops.h"
+#include "xfs_error.h"
+#include "xfs_rw.h"
+#include "xfs_inode_item.h"
+#include "xfs_trans_space.h"
+#include "xfs_utils.h"
+#include "xfs_trace.h"
+#include "xfs_buf.h"
+
+
+/*
+ * Prototypes for internal functions.
+ */
+
+
+STATIC int xfs_rtallocate_range(xfs_mount_t *, xfs_trans_t *, xfs_rtblock_t,
+		xfs_extlen_t, xfs_buf_t **, xfs_fsblock_t *);
+STATIC int xfs_rtany_summary(xfs_mount_t *, xfs_trans_t *, int, int,
+		xfs_rtblock_t, xfs_buf_t **, xfs_fsblock_t *, int *);
+STATIC int xfs_rtcheck_range(xfs_mount_t *, xfs_trans_t *, xfs_rtblock_t,
+		xfs_extlen_t, int, xfs_rtblock_t *, int *);
+STATIC int xfs_rtfind_back(xfs_mount_t *, xfs_trans_t *, xfs_rtblock_t,
+		xfs_rtblock_t, xfs_rtblock_t *);
+STATIC int xfs_rtfind_forw(xfs_mount_t *, xfs_trans_t *, xfs_rtblock_t,
+		xfs_rtblock_t, xfs_rtblock_t *);
+STATIC int xfs_rtget_summary( xfs_mount_t *, xfs_trans_t *, int,
+		xfs_rtblock_t, xfs_buf_t **, xfs_fsblock_t *, xfs_suminfo_t *);
+STATIC int xfs_rtmodify_range(xfs_mount_t *, xfs_trans_t *, xfs_rtblock_t,
+		xfs_extlen_t, int);
+STATIC int xfs_rtmodify_summary(xfs_mount_t *, xfs_trans_t *, int,
+		xfs_rtblock_t, int, xfs_buf_t **, xfs_fsblock_t *);
+
+/*
+ * Internal functions.
+ */
+
+/*
+ * Allocate space to the bitmap or summary file, and zero it, for growfs.
+ */
+STATIC int				/* error */
+xfs_growfs_rt_alloc(
+	xfs_mount_t	*mp,		/* file system mount point */
+	xfs_extlen_t	oblocks,	/* old count of blocks */
+	xfs_extlen_t	nblocks,	/* new count of blocks */
+	xfs_inode_t	*ip)		/* inode (bitmap/summary) */
+{
+	xfs_fileoff_t	bno;		/* block number in file */
+	xfs_buf_t	*bp;		/* temporary buffer for zeroing */
+	int		committed;	/* transaction committed flag */
+	xfs_daddr_t	d;		/* disk block address */
+	int		error;		/* error return value */
+	xfs_fsblock_t	firstblock;	/* first block allocated in xaction */
+	xfs_bmap_free_t	flist;		/* list of freed blocks */
+	xfs_fsblock_t	fsbno;		/* filesystem block for bno */
+	xfs_bmbt_irec_t	map;		/* block map output */
+	int		nmap;		/* number of block maps */
+	int		resblks;	/* space reservation */
+
+	/*
+	 * Allocate space to the file, as necessary.
+	 */
+	while (oblocks < nblocks) {
+		int		cancelflags = 0;
+		xfs_trans_t	*tp;
+
+		tp = xfs_trans_alloc(mp, XFS_TRANS_GROWFSRT_ALLOC);
+		resblks = XFS_GROWFSRT_SPACE_RES(mp, nblocks - oblocks);
+		/*
+		 * Reserve space & log for one extent added to the file.
+		 */
+		if ((error = xfs_trans_reserve(tp, resblks,
+				XFS_GROWRTALLOC_LOG_RES(mp), 0,
+				XFS_TRANS_PERM_LOG_RES,
+				XFS_DEFAULT_PERM_LOG_COUNT)))
+			goto error_cancel;
+		cancelflags = XFS_TRANS_RELEASE_LOG_RES;
+		/*
+		 * Lock the inode.
+		 */
+		xfs_ilock(ip, XFS_ILOCK_EXCL);
+		xfs_trans_ijoin_ref(tp, ip, XFS_ILOCK_EXCL);
+
+		xfs_bmap_init(&flist, &firstblock);
+		/*
+		 * Allocate blocks to the bitmap file.
+		 */
+		nmap = 1;
+		cancelflags |= XFS_TRANS_ABORT;
+		error = xfs_bmapi(tp, ip, oblocks, nblocks - oblocks,
+			XFS_BMAPI_WRITE | XFS_BMAPI_METADATA, &firstblock,
+			resblks, &map, &nmap, &flist);
+		if (!error && nmap < 1)
+			error = XFS_ERROR(ENOSPC);
+		if (error)
+			goto error_cancel;
+		/*
+		 * Free any blocks freed up in the transaction, then commit.
+		 */
+		error = xfs_bmap_finish(&tp, &flist, &committed);
+		if (error)
+			goto error_cancel;
+		error = xfs_trans_commit(tp, XFS_TRANS_RELEASE_LOG_RES);
+		if (error)
+			goto error;
+		/*
+		 * Now we need to clear the allocated blocks.
+		 * Do this one block per transaction, to keep it simple.
+		 */
+		cancelflags = 0;
+		for (bno = map.br_startoff, fsbno = map.br_startblock;
+		     bno < map.br_startoff + map.br_blockcount;
+		     bno++, fsbno++) {
+			tp = xfs_trans_alloc(mp, XFS_TRANS_GROWFSRT_ZERO);
+			/*
+			 * Reserve log for one block zeroing.
+			 */
+			if ((error = xfs_trans_reserve(tp, 0,
+					XFS_GROWRTZERO_LOG_RES(mp), 0, 0, 0)))
+				goto error_cancel;
+			/*
+			 * Lock the bitmap inode.
+			 */
+			xfs_ilock(ip, XFS_ILOCK_EXCL);
+			xfs_trans_ijoin_ref(tp, ip, XFS_ILOCK_EXCL);
+			/*
+			 * Get a buffer for the block.
+			 */
+			d = XFS_FSB_TO_DADDR(mp, fsbno);
+			bp = xfs_trans_get_buf(tp, mp->m_ddev_targp, d,
+				mp->m_bsize, 0);
+			if (bp == NULL) {
+				error = XFS_ERROR(EIO);
+error_cancel:
+				xfs_trans_cancel(tp, cancelflags);
+				goto error;
+			}
+			memset(XFS_BUF_PTR(bp), 0, mp->m_sb.sb_blocksize);
+			xfs_trans_log_buf(tp, bp, 0, mp->m_sb.sb_blocksize - 1);
+			/*
+			 * Commit the transaction.
+			 */
+			error = xfs_trans_commit(tp, 0);
+			if (error)
+				goto error;
+		}
+		/*
+		 * Go on to the next extent, if any.
+		 */
+		oblocks = map.br_startoff + map.br_blockcount;
+	}
+	return 0;
+error:
+	return error;
+}
+
+/*
+ * Attempt to allocate an extent minlen<=len<=maxlen starting from
+ * bitmap block bbno.  If we don't get maxlen then use prod to trim
+ * the length, if given.  Returns error; returns starting block in *rtblock.
+ * The lengths are all in rtextents.
+ */
+STATIC int				/* error */
+xfs_rtallocate_extent_block(
+	xfs_mount_t	*mp,		/* file system mount point */
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_rtblock_t	bbno,		/* bitmap block number */
+	xfs_extlen_t	minlen,		/* minimum length to allocate */
+	xfs_extlen_t	maxlen,		/* maximum length to allocate */
+	xfs_extlen_t	*len,		/* out: actual length allocated */
+	xfs_rtblock_t	*nextp,		/* out: next block to try */
+	xfs_buf_t	**rbpp,		/* in/out: summary block buffer */
+	xfs_fsblock_t	*rsb,		/* in/out: summary block number */
+	xfs_extlen_t	prod,		/* extent product factor */
+	xfs_rtblock_t	*rtblock)	/* out: start block allocated */
+{
+	xfs_rtblock_t	besti;		/* best rtblock found so far */
+	xfs_rtblock_t	bestlen;	/* best length found so far */
+	xfs_rtblock_t	end;		/* last rtblock in chunk */
+	int		error;		/* error value */
+	xfs_rtblock_t	i;		/* current rtblock trying */
+	xfs_rtblock_t	next;		/* next rtblock to try */
+	int		stat;		/* status from internal calls */
+
+	/*
+	 * Loop over all the extents starting in this bitmap block,
+	 * looking for one that's long enough.
+	 */
+	for (i = XFS_BLOCKTOBIT(mp, bbno), besti = -1, bestlen = 0,
+		end = XFS_BLOCKTOBIT(mp, bbno + 1) - 1;
+	     i <= end;
+	     i++) {
+		/*
+		 * See if there's a free extent of maxlen starting at i.
+		 * If it's not so then next will contain the first non-free.
+		 */
+		error = xfs_rtcheck_range(mp, tp, i, maxlen, 1, &next, &stat);
+		if (error) {
+			return error;
+		}
+		if (stat) {
+			/*
+			 * i for maxlen is all free, allocate and return that.
+			 */
+			error = xfs_rtallocate_range(mp, tp, i, maxlen, rbpp,
+				rsb);
+			if (error) {
+				return error;
+			}
+			*len = maxlen;
+			*rtblock = i;
+			return 0;
+		}
+		/*
+		 * In the case where we have a variable-sized allocation
+		 * request, figure out how big this free piece is,
+		 * and if it's big enough for the minimum, and the best
+		 * so far, remember it.
+		 */
+		if (minlen < maxlen) {
+			xfs_rtblock_t	thislen;	/* this extent size */
+
+			thislen = next - i;
+			if (thislen >= minlen && thislen > bestlen) {
+				besti = i;
+				bestlen = thislen;
+			}
+		}
+		/*
+		 * If not done yet, find the start of the next free space.
+		 */
+		if (next < end) {
+			error = xfs_rtfind_forw(mp, tp, next, end, &i);
+			if (error) {
+				return error;
+			}
+		} else
+			break;
+	}
+	/*
+	 * Searched the whole thing & didn't find a maxlen free extent.
+	 */
+	if (minlen < maxlen && besti != -1) {
+		xfs_extlen_t	p;	/* amount to trim length by */
+
+		/*
+		 * If size should be a multiple of prod, make that so.
+		 */
+		if (prod > 1 && (p = do_mod(bestlen, prod)))
+			bestlen -= p;
+		/*
+		 * Allocate besti for bestlen & return that.
+		 */
+		error = xfs_rtallocate_range(mp, tp, besti, bestlen, rbpp, rsb);
+		if (error) {
+			return error;
+		}
+		*len = bestlen;
+		*rtblock = besti;
+		return 0;
+	}
+	/*
+	 * Allocation failed.  Set *nextp to the next block to try.
+	 */
+	*nextp = next;
+	*rtblock = NULLRTBLOCK;
+	return 0;
+}
+
+/*
+ * Allocate an extent of length minlen<=len<=maxlen, starting at block
+ * bno.  If we don't get maxlen then use prod to trim the length, if given.
+ * Returns error; returns starting block in *rtblock.
+ * The lengths are all in rtextents.
+ */
+STATIC int				/* error */
+xfs_rtallocate_extent_exact(
+	xfs_mount_t	*mp,		/* file system mount point */
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_rtblock_t	bno,		/* starting block number to allocate */
+	xfs_extlen_t	minlen,		/* minimum length to allocate */
+	xfs_extlen_t	maxlen,		/* maximum length to allocate */
+	xfs_extlen_t	*len,		/* out: actual length allocated */
+	xfs_buf_t	**rbpp,		/* in/out: summary block buffer */
+	xfs_fsblock_t	*rsb,		/* in/out: summary block number */
+	xfs_extlen_t	prod,		/* extent product factor */
+	xfs_rtblock_t	*rtblock)	/* out: start block allocated */
+{
+	int		error;		/* error value */
+	xfs_extlen_t	i;		/* extent length trimmed due to prod */
+	int		isfree;		/* extent is free */
+	xfs_rtblock_t	next;		/* next block to try (dummy) */
+
+	ASSERT(minlen % prod == 0 && maxlen % prod == 0);
+	/*
+	 * Check if the range in question (for maxlen) is free.
+	 */
+	error = xfs_rtcheck_range(mp, tp, bno, maxlen, 1, &next, &isfree);
+	if (error) {
+		return error;
+	}
+	if (isfree) {
+		/*
+		 * If it is, allocate it and return success.
+		 */
+		error = xfs_rtallocate_range(mp, tp, bno, maxlen, rbpp, rsb);
+		if (error) {
+			return error;
+		}
+		*len = maxlen;
+		*rtblock = bno;
+		return 0;
+	}
+	/*
+	 * If not, allocate what there is, if it's at least minlen.
+	 */
+	maxlen = next - bno;
+	if (maxlen < minlen) {
+		/*
+		 * Failed, return failure status.
+		 */
+		*rtblock = NULLRTBLOCK;
+		return 0;
+	}
+	/*
+	 * Trim off tail of extent, if prod is specified.
+	 */
+	if (prod > 1 && (i = maxlen % prod)) {
+		maxlen -= i;
+		if (maxlen < minlen) {
+			/*
+			 * Now we can't do it, return failure status.
+			 */
+			*rtblock = NULLRTBLOCK;
+			return 0;
+		}
+	}
+	/*
+	 * Allocate what we can and return it.
+	 */
+	error = xfs_rtallocate_range(mp, tp, bno, maxlen, rbpp, rsb);
+	if (error) {
+		return error;
+	}
+	*len = maxlen;
+	*rtblock = bno;
+	return 0;
+}
+
+/*
+ * Allocate an extent of length minlen<=len<=maxlen, starting as near
+ * to bno as possible.  If we don't get maxlen then use prod to trim
+ * the length, if given.  The lengths are all in rtextents.
+ */
+STATIC int				/* error */
+xfs_rtallocate_extent_near(
+	xfs_mount_t	*mp,		/* file system mount point */
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_rtblock_t	bno,		/* starting block number to allocate */
+	xfs_extlen_t	minlen,		/* minimum length to allocate */
+	xfs_extlen_t	maxlen,		/* maximum length to allocate */
+	xfs_extlen_t	*len,		/* out: actual length allocated */
+	xfs_buf_t	**rbpp,		/* in/out: summary block buffer */
+	xfs_fsblock_t	*rsb,		/* in/out: summary block number */
+	xfs_extlen_t	prod,		/* extent product factor */
+	xfs_rtblock_t	*rtblock)	/* out: start block allocated */
+{
+	int		any;		/* any useful extents from summary */
+	xfs_rtblock_t	bbno;		/* bitmap block number */
+	int		error;		/* error value */
+	int		i;		/* bitmap block offset (loop control) */
+	int		j;		/* secondary loop control */
+	int		log2len;	/* log2 of minlen */
+	xfs_rtblock_t	n;		/* next block to try */
+	xfs_rtblock_t	r;		/* result block */
+
+	ASSERT(minlen % prod == 0 && maxlen % prod == 0);
+	/*
+	 * If the block number given is off the end, silently set it to
+	 * the last block.
+	 */
+	if (bno >= mp->m_sb.sb_rextents)
+		bno = mp->m_sb.sb_rextents - 1;
+	/*
+	 * Try the exact allocation first.
+	 */
+	error = xfs_rtallocate_extent_exact(mp, tp, bno, minlen, maxlen, len,
+		rbpp, rsb, prod, &r);
+	if (error) {
+		return error;
+	}
+	/*
+	 * If the exact allocation worked, return that.
+	 */
+	if (r != NULLRTBLOCK) {
+		*rtblock = r;
+		return 0;
+	}
+	bbno = XFS_BITTOBLOCK(mp, bno);
+	i = 0;
+	ASSERT(minlen != 0);
+	log2len = xfs_highbit32(minlen);
+	/*
+	 * Loop over all bitmap blocks (bbno + i is current block).
+	 */
+	for (;;) {
+		/*
+		 * Get summary information of extents of all useful levels
+		 * starting in this bitmap block.
+		 */
+		error = xfs_rtany_summary(mp, tp, log2len, mp->m_rsumlevels - 1,
+			bbno + i, rbpp, rsb, &any);
+		if (error) {
+			return error;
+		}
+		/*
+		 * If there are any useful extents starting here, try
+		 * allocating one.
+		 */
+		if (any) {
+			/*
+			 * On the positive side of the starting location.
+			 */
+			if (i >= 0) {
+				/*
+				 * Try to allocate an extent starting in
+				 * this block.
+				 */
+				error = xfs_rtallocate_extent_block(mp, tp,
+					bbno + i, minlen, maxlen, len, &n, rbpp,
+					rsb, prod, &r);
+				if (error) {
+					return error;
+				}
+				/*
+				 * If it worked, return it.
+				 */
+				if (r != NULLRTBLOCK) {
+					*rtblock = r;
+					return 0;
+				}
+			}
+			/*
+			 * On the negative side of the starting location.
+			 */
+			else {		/* i < 0 */
+				/*
+				 * Loop backwards through the bitmap blocks from
+				 * the starting point-1 up to where we are now.
+				 * There should be an extent which ends in this
+				 * bitmap block and is long enough.
+				 */
+				for (j = -1; j > i; j--) {
+					/*
+					 * Grab the summary information for
+					 * this bitmap block.
+					 */
+					error = xfs_rtany_summary(mp, tp,
+						log2len, mp->m_rsumlevels - 1,
+						bbno + j, rbpp, rsb, &any);
+					if (error) {
+						return error;
+					}
+					/*
+					 * If there's no extent given in the
+					 * summary that means the extent we
+					 * found must carry over from an
+					 * earlier block.  If there is an
+					 * extent given, we've already tried
+					 * that allocation, don't do it again.
+					 */
+					if (any)
+						continue;
+					error = xfs_rtallocate_extent_block(mp,
+						tp, bbno + j, minlen, maxlen,
+						len, &n, rbpp, rsb, prod, &r);
+					if (error) {
+						return error;
+					}
+					/*
+					 * If it works, return the extent.
+					 */
+					if (r != NULLRTBLOCK) {
+						*rtblock = r;
+						return 0;
+					}
+				}
+				/*
+				 * There weren't intervening bitmap blocks
+				 * with a long enough extent, or the
+				 * allocation didn't work for some reason
+				 * (i.e. it's a little * too short).
+				 * Try to allocate from the summary block
+				 * that we found.
+				 */
+				error = xfs_rtallocate_extent_block(mp, tp,
+					bbno + i, minlen, maxlen, len, &n, rbpp,
+					rsb, prod, &r);
+				if (error) {
+					return error;
+				}
+				/*
+				 * If it works, return the extent.
+				 */
+				if (r != NULLRTBLOCK) {
+					*rtblock = r;
+					return 0;
+				}
+			}
+		}
+		/*
+		 * Loop control.  If we were on the positive side, and there's
+		 * still more blocks on the negative side, go there.
+		 */
+		if (i > 0 && (int)bbno - i >= 0)
+			i = -i;
+		/*
+		 * If positive, and no more negative, but there are more
+		 * positive, go there.
+		 */
+		else if (i > 0 && (int)bbno + i < mp->m_sb.sb_rbmblocks - 1)
+			i++;
+		/*
+		 * If negative or 0 (just started), and there are positive
+		 * blocks to go, go there.  The 0 case moves to block 1.
+		 */
+		else if (i <= 0 && (int)bbno - i < mp->m_sb.sb_rbmblocks - 1)
+			i = 1 - i;
+		/*
+		 * If negative or 0 and there are more negative blocks,
+		 * go there.
+		 */
+		else if (i <= 0 && (int)bbno + i > 0)
+			i--;
+		/*
+		 * Must be done.  Return failure.
+		 */
+		else
+			break;
+	}
+	*rtblock = NULLRTBLOCK;
+	return 0;
+}
+
+/*
+ * Allocate an extent of length minlen<=len<=maxlen, with no position
+ * specified.  If we don't get maxlen then use prod to trim
+ * the length, if given.  The lengths are all in rtextents.
+ */
+STATIC int				/* error */
+xfs_rtallocate_extent_size(
+	xfs_mount_t	*mp,		/* file system mount point */
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_extlen_t	minlen,		/* minimum length to allocate */
+	xfs_extlen_t	maxlen,		/* maximum length to allocate */
+	xfs_extlen_t	*len,		/* out: actual length allocated */
+	xfs_buf_t	**rbpp,		/* in/out: summary block buffer */
+	xfs_fsblock_t	*rsb,		/* in/out: summary block number */
+	xfs_extlen_t	prod,		/* extent product factor */
+	xfs_rtblock_t	*rtblock)	/* out: start block allocated */
+{
+	int		error;		/* error value */
+	int		i;		/* bitmap block number */
+	int		l;		/* level number (loop control) */
+	xfs_rtblock_t	n;		/* next block to be tried */
+	xfs_rtblock_t	r;		/* result block number */
+	xfs_suminfo_t	sum;		/* summary information for extents */
+
+	ASSERT(minlen % prod == 0 && maxlen % prod == 0);
+	ASSERT(maxlen != 0);
+
+	/*
+	 * Loop over all the levels starting with maxlen.
+	 * At each level, look at all the bitmap blocks, to see if there
+	 * are extents starting there that are long enough (>= maxlen).
+	 * Note, only on the initial level can the allocation fail if
+	 * the summary says there's an extent.
+	 */
+	for (l = xfs_highbit32(maxlen); l < mp->m_rsumlevels; l++) {
+		/*
+		 * Loop over all the bitmap blocks.
+		 */
+		for (i = 0; i < mp->m_sb.sb_rbmblocks; i++) {
+			/*
+			 * Get the summary for this level/block.
+			 */
+			error = xfs_rtget_summary(mp, tp, l, i, rbpp, rsb,
+				&sum);
+			if (error) {
+				return error;
+			}
+			/*
+			 * Nothing there, on to the next block.
+			 */
+			if (!sum)
+				continue;
+			/*
+			 * Try allocating the extent.
+			 */
+			error = xfs_rtallocate_extent_block(mp, tp, i, maxlen,
+				maxlen, len, &n, rbpp, rsb, prod, &r);
+			if (error) {
+				return error;
+			}
+			/*
+			 * If it worked, return that.
+			 */
+			if (r != NULLRTBLOCK) {
+				*rtblock = r;
+				return 0;
+			}
+			/*
+			 * If the "next block to try" returned from the
+			 * allocator is beyond the next bitmap block,
+			 * skip to that bitmap block.
+			 */
+			if (XFS_BITTOBLOCK(mp, n) > i + 1)
+				i = XFS_BITTOBLOCK(mp, n) - 1;
+		}
+	}
+	/*
+	 * Didn't find any maxlen blocks.  Try smaller ones, unless
+	 * we're asking for a fixed size extent.
+	 */
+	if (minlen > --maxlen) {
+		*rtblock = NULLRTBLOCK;
+		return 0;
+	}
+	ASSERT(minlen != 0);
+	ASSERT(maxlen != 0);
+
+	/*
+	 * Loop over sizes, from maxlen down to minlen.
+	 * This time, when we do the allocations, allow smaller ones
+	 * to succeed.
+	 */
+	for (l = xfs_highbit32(maxlen); l >= xfs_highbit32(minlen); l--) {
+		/*
+		 * Loop over all the bitmap blocks, try an allocation
+		 * starting in that block.
+		 */
+		for (i = 0; i < mp->m_sb.sb_rbmblocks; i++) {
+			/*
+			 * Get the summary information for this level/block.
+			 */
+			error =	xfs_rtget_summary(mp, tp, l, i, rbpp, rsb,
+						  &sum);
+			if (error) {
+				return error;
+			}
+			/*
+			 * If nothing there, go on to next.
+			 */
+			if (!sum)
+				continue;
+			/*
+			 * Try the allocation.  Make sure the specified
+			 * minlen/maxlen are in the possible range for
+			 * this summary level.
+			 */
+			error = xfs_rtallocate_extent_block(mp, tp, i,
+					XFS_RTMAX(minlen, 1 << l),
+					XFS_RTMIN(maxlen, (1 << (l + 1)) - 1),
+					len, &n, rbpp, rsb, prod, &r);
+			if (error) {
+				return error;
+			}
+			/*
+			 * If it worked, return that extent.
+			 */
+			if (r != NULLRTBLOCK) {
+				*rtblock = r;
+				return 0;
+			}
+			/*
+			 * If the "next block to try" returned from the
+			 * allocator is beyond the next bitmap block,
+			 * skip to that bitmap block.
+			 */
+			if (XFS_BITTOBLOCK(mp, n) > i + 1)
+				i = XFS_BITTOBLOCK(mp, n) - 1;
+		}
+	}
+	/*
+	 * Got nothing, return failure.
+	 */
+	*rtblock = NULLRTBLOCK;
+	return 0;
+}
+
+/*
+ * Mark an extent specified by start and len allocated.
+ * Updates all the summary information as well as the bitmap.
+ */
+STATIC int				/* error */
+xfs_rtallocate_range(
+	xfs_mount_t	*mp,		/* file system mount point */
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_rtblock_t	start,		/* start block to allocate */
+	xfs_extlen_t	len,		/* length to allocate */
+	xfs_buf_t	**rbpp,		/* in/out: summary block buffer */
+	xfs_fsblock_t	*rsb)		/* in/out: summary block number */
+{
+	xfs_rtblock_t	end;		/* end of the allocated extent */
+	int		error;		/* error value */
+	xfs_rtblock_t	postblock;	/* first block allocated > end */
+	xfs_rtblock_t	preblock;	/* first block allocated < start */
+
+	end = start + len - 1;
+	/*
+	 * Assume we're allocating out of the middle of a free extent.
+	 * We need to find the beginning and end of the extent so we can
+	 * properly update the summary.
+	 */
+	error = xfs_rtfind_back(mp, tp, start, 0, &preblock);
+	if (error) {
+		return error;
+	}
+	/*
+	 * Find the next allocated block (end of free extent).
+	 */
+	error = xfs_rtfind_forw(mp, tp, end, mp->m_sb.sb_rextents - 1,
+		&postblock);
+	if (error) {
+		return error;
+	}
+	/*
+	 * Decrement the summary information corresponding to the entire
+	 * (old) free extent.
+	 */
+	error = xfs_rtmodify_summary(mp, tp,
+		XFS_RTBLOCKLOG(postblock + 1 - preblock),
+		XFS_BITTOBLOCK(mp, preblock), -1, rbpp, rsb);
+	if (error) {
+		return error;
+	}
+	/*
+	 * If there are blocks not being allocated at the front of the
+	 * old extent, add summary data for them to be free.
+	 */
+	if (preblock < start) {
+		error = xfs_rtmodify_summary(mp, tp,
+			XFS_RTBLOCKLOG(start - preblock),
+			XFS_BITTOBLOCK(mp, preblock), 1, rbpp, rsb);
+		if (error) {
+			return error;
+		}
+	}
+	/*
+	 * If there are blocks not being allocated at the end of the
+	 * old extent, add summary data for them to be free.
+	 */
+	if (postblock > end) {
+		error = xfs_rtmodify_summary(mp, tp,
+			XFS_RTBLOCKLOG(postblock - end),
+			XFS_BITTOBLOCK(mp, end + 1), 1, rbpp, rsb);
+		if (error) {
+			return error;
+		}
+	}
+	/*
+	 * Modify the bitmap to mark this extent allocated.
+	 */
+	error = xfs_rtmodify_range(mp, tp, start, len, 0);
+	return error;
+}
+
+/*
+ * Return whether there are any free extents in the size range given
+ * by low and high, for the bitmap block bbno.
+ */
+STATIC int				/* error */
+xfs_rtany_summary(
+	xfs_mount_t	*mp,		/* file system mount structure */
+	xfs_trans_t	*tp,		/* transaction pointer */
+	int		low,		/* low log2 extent size */
+	int		high,		/* high log2 extent size */
+	xfs_rtblock_t	bbno,		/* bitmap block number */
+	xfs_buf_t	**rbpp,		/* in/out: summary block buffer */
+	xfs_fsblock_t	*rsb,		/* in/out: summary block number */
+	int		*stat)		/* out: any good extents here? */
+{
+	int		error;		/* error value */
+	int		log;		/* loop counter, log2 of ext. size */
+	xfs_suminfo_t	sum;		/* summary data */
+
+	/*
+	 * Loop over logs of extent sizes.  Order is irrelevant.
+	 */
+	for (log = low; log <= high; log++) {
+		/*
+		 * Get one summary datum.
+		 */
+		error = xfs_rtget_summary(mp, tp, log, bbno, rbpp, rsb, &sum);
+		if (error) {
+			return error;
+		}
+		/*
+		 * If there are any, return success.
+		 */
+		if (sum) {
+			*stat = 1;
+			return 0;
+		}
+	}
+	/*
+	 * Found nothing, return failure.
+	 */
+	*stat = 0;
+	return 0;
+}
+
+/*
+ * Get a buffer for the bitmap or summary file block specified.
+ * The buffer is returned read and locked.
+ */
+STATIC int				/* error */
+xfs_rtbuf_get(
+	xfs_mount_t	*mp,		/* file system mount structure */
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_rtblock_t	block,		/* block number in bitmap or summary */
+	int		issum,		/* is summary not bitmap */
+	xfs_buf_t	**bpp)		/* output: buffer for the block */
+{
+	xfs_buf_t	*bp;		/* block buffer, result */
+	xfs_daddr_t	d;		/* disk addr of block */
+	int		error;		/* error value */
+	xfs_fsblock_t	fsb;		/* fs block number for block */
+	xfs_inode_t	*ip;		/* bitmap or summary inode */
+
+	ip = issum ? mp->m_rsumip : mp->m_rbmip;
+	/*
+	 * Map from the file offset (block) and inode number to the
+	 * file system block.
+	 */
+	error = xfs_bmapi_single(tp, ip, XFS_DATA_FORK, &fsb, block);
+	if (error) {
+		return error;
+	}
+	ASSERT(fsb != NULLFSBLOCK);
+	/*
+	 * Convert to disk address for buffer cache.
+	 */
+	d = XFS_FSB_TO_DADDR(mp, fsb);
+	/*
+	 * Read the buffer.
+	 */
+	error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp, d,
+				   mp->m_bsize, 0, &bp);
+	if (error) {
+		return error;
+	}
+	ASSERT(bp && !XFS_BUF_GETERROR(bp));
+	*bpp = bp;
+	return 0;
+}
+
+#ifdef DEBUG
+/*
+ * Check that the given extent (block range) is allocated already.
+ */
+STATIC int				/* error */
+xfs_rtcheck_alloc_range(
+	xfs_mount_t	*mp,		/* file system mount point */
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_rtblock_t	bno,		/* starting block number of extent */
+	xfs_extlen_t	len,		/* length of extent */
+	int		*stat)		/* out: 1 for allocated, 0 for not */
+{
+	xfs_rtblock_t	new;		/* dummy for xfs_rtcheck_range */
+
+	return xfs_rtcheck_range(mp, tp, bno, len, 0, &new, stat);
+}
+#endif
+
+/*
+ * Check that the given range is either all allocated (val = 0) or
+ * all free (val = 1).
+ */
+STATIC int				/* error */
+xfs_rtcheck_range(
+	xfs_mount_t	*mp,		/* file system mount point */
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_rtblock_t	start,		/* starting block number of extent */
+	xfs_extlen_t	len,		/* length of extent */
+	int		val,		/* 1 for free, 0 for allocated */
+	xfs_rtblock_t	*new,		/* out: first block not matching */
+	int		*stat)		/* out: 1 for matches, 0 for not */
+{
+	xfs_rtword_t	*b;		/* current word in buffer */
+	int		bit;		/* bit number in the word */
+	xfs_rtblock_t	block;		/* bitmap block number */
+	xfs_buf_t	*bp;		/* buf for the block */
+	xfs_rtword_t	*bufp;		/* starting word in buffer */
+	int		error;		/* error value */
+	xfs_rtblock_t	i;		/* current bit number rel. to start */
+	xfs_rtblock_t	lastbit;	/* last useful bit in word */
+	xfs_rtword_t	mask;		/* mask of relevant bits for value */
+	xfs_rtword_t	wdiff;		/* difference from wanted value */
+	int		word;		/* word number in the buffer */
+
+	/*
+	 * Compute starting bitmap block number
+	 */
+	block = XFS_BITTOBLOCK(mp, start);
+	/*
+	 * Read the bitmap block.
+	 */
+	error = xfs_rtbuf_get(mp, tp, block, 0, &bp);
+	if (error) {
+		return error;
+	}
+	bufp = (xfs_rtword_t *)XFS_BUF_PTR(bp);
+	/*
+	 * Compute the starting word's address, and starting bit.
+	 */
+	word = XFS_BITTOWORD(mp, start);
+	b = &bufp[word];
+	bit = (int)(start & (XFS_NBWORD - 1));
+	/*
+	 * 0 (allocated) => all zero's; 1 (free) => all one's.
+	 */
+	val = -val;
+	/*
+	 * If not starting on a word boundary, deal with the first
+	 * (partial) word.
+	 */
+	if (bit) {
+		/*
+		 * Compute first bit not examined.
+		 */
+		lastbit = XFS_RTMIN(bit + len, XFS_NBWORD);
+		/*
+		 * Mask of relevant bits.
+		 */
+		mask = (((xfs_rtword_t)1 << (lastbit - bit)) - 1) << bit;
+		/*
+		 * Compute difference between actual and desired value.
+		 */
+		if ((wdiff = (*b ^ val) & mask)) {
+			/*
+			 * Different, compute first wrong bit and return.
+			 */
+			xfs_trans_brelse(tp, bp);
+			i = XFS_RTLOBIT(wdiff) - bit;
+			*new = start + i;
+			*stat = 0;
+			return 0;
+		}
+		i = lastbit - bit;
+		/*
+		 * Go on to next block if that's where the next word is
+		 * and we need the next word.
+		 */
+		if (++word == XFS_BLOCKWSIZE(mp) && i < len) {
+			/*
+			 * If done with this block, get the next one.
+			 */
+			xfs_trans_brelse(tp, bp);
+			error = xfs_rtbuf_get(mp, tp, ++block, 0, &bp);
+			if (error) {
+				return error;
+			}
+			b = bufp = (xfs_rtword_t *)XFS_BUF_PTR(bp);
+			word = 0;
+		} else {
+			/*
+			 * Go on to the next word in the buffer.
+			 */
+			b++;
+		}
+	} else {
+		/*
+		 * Starting on a word boundary, no partial word.
+		 */
+		i = 0;
+	}
+	/*
+	 * Loop over whole words in buffers.  When we use up one buffer
+	 * we move on to the next one.
+	 */
+	while (len - i >= XFS_NBWORD) {
+		/*
+		 * Compute difference between actual and desired value.
+		 */
+		if ((wdiff = *b ^ val)) {
+			/*
+			 * Different, compute first wrong bit and return.
+			 */
+			xfs_trans_brelse(tp, bp);
+			i += XFS_RTLOBIT(wdiff);
+			*new = start + i;
+			*stat = 0;
+			return 0;
+		}
+		i += XFS_NBWORD;
+		/*
+		 * Go on to next block if that's where the next word is
+		 * and we need the next word.
+		 */
+		if (++word == XFS_BLOCKWSIZE(mp) && i < len) {
+			/*
+			 * If done with this block, get the next one.
+			 */
+			xfs_trans_brelse(tp, bp);
+			error = xfs_rtbuf_get(mp, tp, ++block, 0, &bp);
+			if (error) {
+				return error;
+			}
+			b = bufp = (xfs_rtword_t *)XFS_BUF_PTR(bp);
+			word = 0;
+		} else {
+			/*
+			 * Go on to the next word in the buffer.
+			 */
+			b++;
+		}
+	}
+	/*
+	 * If not ending on a word boundary, deal with the last
+	 * (partial) word.
+	 */
+	if ((lastbit = len - i)) {
+		/*
+		 * Mask of relevant bits.
+		 */
+		mask = ((xfs_rtword_t)1 << lastbit) - 1;
+		/*
+		 * Compute difference between actual and desired value.
+		 */
+		if ((wdiff = (*b ^ val) & mask)) {
+			/*
+			 * Different, compute first wrong bit and return.
+			 */
+			xfs_trans_brelse(tp, bp);
+			i += XFS_RTLOBIT(wdiff);
+			*new = start + i;
+			*stat = 0;
+			return 0;
+		} else
+			i = len;
+	}
+	/*
+	 * Successful, return.
+	 */
+	xfs_trans_brelse(tp, bp);
+	*new = start + i;
+	*stat = 1;
+	return 0;
+}
+
+/*
+ * Copy and transform the summary file, given the old and new
+ * parameters in the mount structures.
+ */
+STATIC int				/* error */
+xfs_rtcopy_summary(
+	xfs_mount_t	*omp,		/* old file system mount point */
+	xfs_mount_t	*nmp,		/* new file system mount point */
+	xfs_trans_t	*tp)		/* transaction pointer */
+{
+	xfs_rtblock_t	bbno;		/* bitmap block number */
+	xfs_buf_t	*bp;		/* summary buffer */
+	int		error;		/* error return value */
+	int		log;		/* summary level number (log length) */
+	xfs_suminfo_t	sum;		/* summary data */
+	xfs_fsblock_t	sumbno;		/* summary block number */
+
+	bp = NULL;
+	for (log = omp->m_rsumlevels - 1; log >= 0; log--) {
+		for (bbno = omp->m_sb.sb_rbmblocks - 1;
+		     (xfs_srtblock_t)bbno >= 0;
+		     bbno--) {
+			error = xfs_rtget_summary(omp, tp, log, bbno, &bp,
+				&sumbno, &sum);
+			if (error)
+				return error;
+			if (sum == 0)
+				continue;
+			error = xfs_rtmodify_summary(omp, tp, log, bbno, -sum,
+				&bp, &sumbno);
+			if (error)
+				return error;
+			error = xfs_rtmodify_summary(nmp, tp, log, bbno, sum,
+				&bp, &sumbno);
+			if (error)
+				return error;
+			ASSERT(sum > 0);
+		}
+	}
+	return 0;
+}
+
+/*
+ * Searching backward from start to limit, find the first block whose
+ * allocated/free state is different from start's.
+ */
+STATIC int				/* error */
+xfs_rtfind_back(
+	xfs_mount_t	*mp,		/* file system mount point */
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_rtblock_t	start,		/* starting block to look at */
+	xfs_rtblock_t	limit,		/* last block to look at */
+	xfs_rtblock_t	*rtblock)	/* out: start block found */
+{
+	xfs_rtword_t	*b;		/* current word in buffer */
+	int		bit;		/* bit number in the word */
+	xfs_rtblock_t	block;		/* bitmap block number */
+	xfs_buf_t	*bp;		/* buf for the block */
+	xfs_rtword_t	*bufp;		/* starting word in buffer */
+	int		error;		/* error value */
+	xfs_rtblock_t	firstbit;	/* first useful bit in the word */
+	xfs_rtblock_t	i;		/* current bit number rel. to start */
+	xfs_rtblock_t	len;		/* length of inspected area */
+	xfs_rtword_t	mask;		/* mask of relevant bits for value */
+	xfs_rtword_t	want;		/* mask for "good" values */
+	xfs_rtword_t	wdiff;		/* difference from wanted value */
+	int		word;		/* word number in the buffer */
+
+	/*
+	 * Compute and read in starting bitmap block for starting block.
+	 */
+	block = XFS_BITTOBLOCK(mp, start);
+	error = xfs_rtbuf_get(mp, tp, block, 0, &bp);
+	if (error) {
+		return error;
+	}
+	bufp = (xfs_rtword_t *)XFS_BUF_PTR(bp);
+	/*
+	 * Get the first word's index & point to it.
+	 */
+	word = XFS_BITTOWORD(mp, start);
+	b = &bufp[word];
+	bit = (int)(start & (XFS_NBWORD - 1));
+	len = start - limit + 1;
+	/*
+	 * Compute match value, based on the bit at start: if 1 (free)
+	 * then all-ones, else all-zeroes.
+	 */
+	want = (*b & ((xfs_rtword_t)1 << bit)) ? -1 : 0;
+	/*
+	 * If the starting position is not word-aligned, deal with the
+	 * partial word.
+	 */
+	if (bit < XFS_NBWORD - 1) {
+		/*
+		 * Calculate first (leftmost) bit number to look at,
+		 * and mask for all the relevant bits in this word.
+		 */
+		firstbit = XFS_RTMAX((xfs_srtblock_t)(bit - len + 1), 0);
+		mask = (((xfs_rtword_t)1 << (bit - firstbit + 1)) - 1) <<
+			firstbit;
+		/*
+		 * Calculate the difference between the value there
+		 * and what we're looking for.
+		 */
+		if ((wdiff = (*b ^ want) & mask)) {
+			/*
+			 * Different.  Mark where we are and return.
+			 */
+			xfs_trans_brelse(tp, bp);
+			i = bit - XFS_RTHIBIT(wdiff);
+			*rtblock = start - i + 1;
+			return 0;
+		}
+		i = bit - firstbit + 1;
+		/*
+		 * Go on to previous block if that's where the previous word is
+		 * and we need the previous word.
+		 */
+		if (--word == -1 && i < len) {
+			/*
+			 * If done with this block, get the previous one.
+			 */
+			xfs_trans_brelse(tp, bp);
+			error = xfs_rtbuf_get(mp, tp, --block, 0, &bp);
+			if (error) {
+				return error;
+			}
+			bufp = (xfs_rtword_t *)XFS_BUF_PTR(bp);
+			word = XFS_BLOCKWMASK(mp);
+			b = &bufp[word];
+		} else {
+			/*
+			 * Go on to the previous word in the buffer.
+			 */
+			b--;
+		}
+	} else {
+		/*
+		 * Starting on a word boundary, no partial word.
+		 */
+		i = 0;
+	}
+	/*
+	 * Loop over whole words in buffers.  When we use up one buffer
+	 * we move on to the previous one.
+	 */
+	while (len - i >= XFS_NBWORD) {
+		/*
+		 * Compute difference between actual and desired value.
+		 */
+		if ((wdiff = *b ^ want)) {
+			/*
+			 * Different, mark where we are and return.
+			 */
+			xfs_trans_brelse(tp, bp);
+			i += XFS_NBWORD - 1 - XFS_RTHIBIT(wdiff);
+			*rtblock = start - i + 1;
+			return 0;
+		}
+		i += XFS_NBWORD;
+		/*
+		 * Go on to previous block if that's where the previous word is
+		 * and we need the previous word.
+		 */
+		if (--word == -1 && i < len) {
+			/*
+			 * If done with this block, get the previous one.
+			 */
+			xfs_trans_brelse(tp, bp);
+			error = xfs_rtbuf_get(mp, tp, --block, 0, &bp);
+			if (error) {
+				return error;
+			}
+			bufp = (xfs_rtword_t *)XFS_BUF_PTR(bp);
+			word = XFS_BLOCKWMASK(mp);
+			b = &bufp[word];
+		} else {
+			/*
+			 * Go on to the previous word in the buffer.
+			 */
+			b--;
+		}
+	}
+	/*
+	 * If not ending on a word boundary, deal with the last
+	 * (partial) word.
+	 */
+	if (len - i) {
+		/*
+		 * Calculate first (leftmost) bit number to look at,
+		 * and mask for all the relevant bits in this word.
+		 */
+		firstbit = XFS_NBWORD - (len - i);
+		mask = (((xfs_rtword_t)1 << (len - i)) - 1) << firstbit;
+		/*
+		 * Compute difference between actual and desired value.
+		 */
+		if ((wdiff = (*b ^ want) & mask)) {
+			/*
+			 * Different, mark where we are and return.
+			 */
+			xfs_trans_brelse(tp, bp);
+			i += XFS_NBWORD - 1 - XFS_RTHIBIT(wdiff);
+			*rtblock = start - i + 1;
+			return 0;
+		} else
+			i = len;
+	}
+	/*
+	 * No match, return that we scanned the whole area.
+	 */
+	xfs_trans_brelse(tp, bp);
+	*rtblock = start - i + 1;
+	return 0;
+}
+
+/*
+ * Searching forward from start to limit, find the first block whose
+ * allocated/free state is different from start's.
+ */
+STATIC int				/* error */
+xfs_rtfind_forw(
+	xfs_mount_t	*mp,		/* file system mount point */
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_rtblock_t	start,		/* starting block to look at */
+	xfs_rtblock_t	limit,		/* last block to look at */
+	xfs_rtblock_t	*rtblock)	/* out: start block found */
+{
+	xfs_rtword_t	*b;		/* current word in buffer */
+	int		bit;		/* bit number in the word */
+	xfs_rtblock_t	block;		/* bitmap block number */
+	xfs_buf_t	*bp;		/* buf for the block */
+	xfs_rtword_t	*bufp;		/* starting word in buffer */
+	int		error;		/* error value */
+	xfs_rtblock_t	i;		/* current bit number rel. to start */
+	xfs_rtblock_t	lastbit;	/* last useful bit in the word */
+	xfs_rtblock_t	len;		/* length of inspected area */
+	xfs_rtword_t	mask;		/* mask of relevant bits for value */
+	xfs_rtword_t	want;		/* mask for "good" values */
+	xfs_rtword_t	wdiff;		/* difference from wanted value */
+	int		word;		/* word number in the buffer */
+
+	/*
+	 * Compute and read in starting bitmap block for starting block.
+	 */
+	block = XFS_BITTOBLOCK(mp, start);
+	error = xfs_rtbuf_get(mp, tp, block, 0, &bp);
+	if (error) {
+		return error;
+	}
+	bufp = (xfs_rtword_t *)XFS_BUF_PTR(bp);
+	/*
+	 * Get the first word's index & point to it.
+	 */
+	word = XFS_BITTOWORD(mp, start);
+	b = &bufp[word];
+	bit = (int)(start & (XFS_NBWORD - 1));
+	len = limit - start + 1;
+	/*
+	 * Compute match value, based on the bit at start: if 1 (free)
+	 * then all-ones, else all-zeroes.
+	 */
+	want = (*b & ((xfs_rtword_t)1 << bit)) ? -1 : 0;
+	/*
+	 * If the starting position is not word-aligned, deal with the
+	 * partial word.
+	 */
+	if (bit) {
+		/*
+		 * Calculate last (rightmost) bit number to look at,
+		 * and mask for all the relevant bits in this word.
+		 */
+		lastbit = XFS_RTMIN(bit + len, XFS_NBWORD);
+		mask = (((xfs_rtword_t)1 << (lastbit - bit)) - 1) << bit;
+		/*
+		 * Calculate the difference between the value there
+		 * and what we're looking for.
+		 */
+		if ((wdiff = (*b ^ want) & mask)) {
+			/*
+			 * Different.  Mark where we are and return.
+			 */
+			xfs_trans_brelse(tp, bp);
+			i = XFS_RTLOBIT(wdiff) - bit;
+			*rtblock = start + i - 1;
+			return 0;
+		}
+		i = lastbit - bit;
+		/*
+		 * Go on to next block if that's where the next word is
+		 * and we need the next word.
+		 */
+		if (++word == XFS_BLOCKWSIZE(mp) && i < len) {
+			/*
+			 * If done with this block, get the previous one.
+			 */
+			xfs_trans_brelse(tp, bp);
+			error = xfs_rtbuf_get(mp, tp, ++block, 0, &bp);
+			if (error) {
+				return error;
+			}
+			b = bufp = (xfs_rtword_t *)XFS_BUF_PTR(bp);
+			word = 0;
+		} else {
+			/*
+			 * Go on to the previous word in the buffer.
+			 */
+			b++;
+		}
+	} else {
+		/*
+		 * Starting on a word boundary, no partial word.
+		 */
+		i = 0;
+	}
+	/*
+	 * Loop over whole words in buffers.  When we use up one buffer
+	 * we move on to the next one.
+	 */
+	while (len - i >= XFS_NBWORD) {
+		/*
+		 * Compute difference between actual and desired value.
+		 */
+		if ((wdiff = *b ^ want)) {
+			/*
+			 * Different, mark where we are and return.
+			 */
+			xfs_trans_brelse(tp, bp);
+			i += XFS_RTLOBIT(wdiff);
+			*rtblock = start + i - 1;
+			return 0;
+		}
+		i += XFS_NBWORD;
+		/*
+		 * Go on to next block if that's where the next word is
+		 * and we need the next word.
+		 */
+		if (++word == XFS_BLOCKWSIZE(mp) && i < len) {
+			/*
+			 * If done with this block, get the next one.
+			 */
+			xfs_trans_brelse(tp, bp);
+			error = xfs_rtbuf_get(mp, tp, ++block, 0, &bp);
+			if (error) {
+				return error;
+			}
+			b = bufp = (xfs_rtword_t *)XFS_BUF_PTR(bp);
+			word = 0;
+		} else {
+			/*
+			 * Go on to the next word in the buffer.
+			 */
+			b++;
+		}
+	}
+	/*
+	 * If not ending on a word boundary, deal with the last
+	 * (partial) word.
+	 */
+	if ((lastbit = len - i)) {
+		/*
+		 * Calculate mask for all the relevant bits in this word.
+		 */
+		mask = ((xfs_rtword_t)1 << lastbit) - 1;
+		/*
+		 * Compute difference between actual and desired value.
+		 */
+		if ((wdiff = (*b ^ want) & mask)) {
+			/*
+			 * Different, mark where we are and return.
+			 */
+			xfs_trans_brelse(tp, bp);
+			i += XFS_RTLOBIT(wdiff);
+			*rtblock = start + i - 1;
+			return 0;
+		} else
+			i = len;
+	}
+	/*
+	 * No match, return that we scanned the whole area.
+	 */
+	xfs_trans_brelse(tp, bp);
+	*rtblock = start + i - 1;
+	return 0;
+}
+
+/*
+ * Mark an extent specified by start and len freed.
+ * Updates all the summary information as well as the bitmap.
+ */
+STATIC int				/* error */
+xfs_rtfree_range(
+	xfs_mount_t	*mp,		/* file system mount point */
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_rtblock_t	start,		/* starting block to free */
+	xfs_extlen_t	len,		/* length to free */
+	xfs_buf_t	**rbpp,		/* in/out: summary block buffer */
+	xfs_fsblock_t	*rsb)		/* in/out: summary block number */
+{
+	xfs_rtblock_t	end;		/* end of the freed extent */
+	int		error;		/* error value */
+	xfs_rtblock_t	postblock;	/* first block freed > end */
+	xfs_rtblock_t	preblock;	/* first block freed < start */
+
+	end = start + len - 1;
+	/*
+	 * Modify the bitmap to mark this extent freed.
+	 */
+	error = xfs_rtmodify_range(mp, tp, start, len, 1);
+	if (error) {
+		return error;
+	}
+	/*
+	 * Assume we're freeing out of the middle of an allocated extent.
+	 * We need to find the beginning and end of the extent so we can
+	 * properly update the summary.
+	 */
+	error = xfs_rtfind_back(mp, tp, start, 0, &preblock);
+	if (error) {
+		return error;
+	}
+	/*
+	 * Find the next allocated block (end of allocated extent).
+	 */
+	error = xfs_rtfind_forw(mp, tp, end, mp->m_sb.sb_rextents - 1,
+		&postblock);
+	if (error)
+		return error;
+	/*
+	 * If there are blocks not being freed at the front of the
+	 * old extent, add summary data for them to be allocated.
+	 */
+	if (preblock < start) {
+		error = xfs_rtmodify_summary(mp, tp,
+			XFS_RTBLOCKLOG(start - preblock),
+			XFS_BITTOBLOCK(mp, preblock), -1, rbpp, rsb);
+		if (error) {
+			return error;
+		}
+	}
+	/*
+	 * If there are blocks not being freed at the end of the
+	 * old extent, add summary data for them to be allocated.
+	 */
+	if (postblock > end) {
+		error = xfs_rtmodify_summary(mp, tp,
+			XFS_RTBLOCKLOG(postblock - end),
+			XFS_BITTOBLOCK(mp, end + 1), -1, rbpp, rsb);
+		if (error) {
+			return error;
+		}
+	}
+	/*
+	 * Increment the summary information corresponding to the entire
+	 * (new) free extent.
+	 */
+	error = xfs_rtmodify_summary(mp, tp,
+		XFS_RTBLOCKLOG(postblock + 1 - preblock),
+		XFS_BITTOBLOCK(mp, preblock), 1, rbpp, rsb);
+	return error;
+}
+
+/*
+ * Read and return the summary information for a given extent size,
+ * bitmap block combination.
+ * Keeps track of a current summary block, so we don't keep reading
+ * it from the buffer cache.
+ */
+STATIC int				/* error */
+xfs_rtget_summary(
+	xfs_mount_t	*mp,		/* file system mount structure */
+	xfs_trans_t	*tp,		/* transaction pointer */
+	int		log,		/* log2 of extent size */
+	xfs_rtblock_t	bbno,		/* bitmap block number */
+	xfs_buf_t	**rbpp,		/* in/out: summary block buffer */
+	xfs_fsblock_t	*rsb,		/* in/out: summary block number */
+	xfs_suminfo_t	*sum)		/* out: summary info for this block */
+{
+	xfs_buf_t	*bp;		/* buffer for summary block */
+	int		error;		/* error value */
+	xfs_fsblock_t	sb;		/* summary fsblock */
+	int		so;		/* index into the summary file */
+	xfs_suminfo_t	*sp;		/* pointer to returned data */
+
+	/*
+	 * Compute entry number in the summary file.
+	 */
+	so = XFS_SUMOFFS(mp, log, bbno);
+	/*
+	 * Compute the block number in the summary file.
+	 */
+	sb = XFS_SUMOFFSTOBLOCK(mp, so);
+	/*
+	 * If we have an old buffer, and the block number matches, use that.
+	 */
+	if (rbpp && *rbpp && *rsb == sb)
+		bp = *rbpp;
+	/*
+	 * Otherwise we have to get the buffer.
+	 */
+	else {
+		/*
+		 * If there was an old one, get rid of it first.
+		 */
+		if (rbpp && *rbpp)
+			xfs_trans_brelse(tp, *rbpp);
+		error = xfs_rtbuf_get(mp, tp, sb, 1, &bp);
+		if (error) {
+			return error;
+		}
+		/*
+		 * Remember this buffer and block for the next call.
+		 */
+		if (rbpp) {
+			*rbpp = bp;
+			*rsb = sb;
+		}
+	}
+	/*
+	 * Point to the summary information & copy it out.
+	 */
+	sp = XFS_SUMPTR(mp, bp, so);
+	*sum = *sp;
+	/*
+	 * Drop the buffer if we're not asked to remember it.
+	 */
+	if (!rbpp)
+		xfs_trans_brelse(tp, bp);
+	return 0;
+}
+
+/*
+ * Set the given range of bitmap bits to the given value.
+ * Do whatever I/O and logging is required.
+ */
+STATIC int				/* error */
+xfs_rtmodify_range(
+	xfs_mount_t	*mp,		/* file system mount point */
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_rtblock_t	start,		/* starting block to modify */
+	xfs_extlen_t	len,		/* length of extent to modify */
+	int		val)		/* 1 for free, 0 for allocated */
+{
+	xfs_rtword_t	*b;		/* current word in buffer */
+	int		bit;		/* bit number in the word */
+	xfs_rtblock_t	block;		/* bitmap block number */
+	xfs_buf_t	*bp;		/* buf for the block */
+	xfs_rtword_t	*bufp;		/* starting word in buffer */
+	int		error;		/* error value */
+	xfs_rtword_t	*first;		/* first used word in the buffer */
+	int		i;		/* current bit number rel. to start */
+	int		lastbit;	/* last useful bit in word */
+	xfs_rtword_t	mask;		/* mask o frelevant bits for value */
+	int		word;		/* word number in the buffer */
+
+	/*
+	 * Compute starting bitmap block number.
+	 */
+	block = XFS_BITTOBLOCK(mp, start);
+	/*
+	 * Read the bitmap block, and point to its data.
+	 */
+	error = xfs_rtbuf_get(mp, tp, block, 0, &bp);
+	if (error) {
+		return error;
+	}
+	bufp = (xfs_rtword_t *)XFS_BUF_PTR(bp);
+	/*
+	 * Compute the starting word's address, and starting bit.
+	 */
+	word = XFS_BITTOWORD(mp, start);
+	first = b = &bufp[word];
+	bit = (int)(start & (XFS_NBWORD - 1));
+	/*
+	 * 0 (allocated) => all zeroes; 1 (free) => all ones.
+	 */
+	val = -val;
+	/*
+	 * If not starting on a word boundary, deal with the first
+	 * (partial) word.
+	 */
+	if (bit) {
+		/*
+		 * Compute first bit not changed and mask of relevant bits.
+		 */
+		lastbit = XFS_RTMIN(bit + len, XFS_NBWORD);
+		mask = (((xfs_rtword_t)1 << (lastbit - bit)) - 1) << bit;
+		/*
+		 * Set/clear the active bits.
+		 */
+		if (val)
+			*b |= mask;
+		else
+			*b &= ~mask;
+		i = lastbit - bit;
+		/*
+		 * Go on to the next block if that's where the next word is
+		 * and we need the next word.
+		 */
+		if (++word == XFS_BLOCKWSIZE(mp) && i < len) {
+			/*
+			 * Log the changed part of this block.
+			 * Get the next one.
+			 */
+			xfs_trans_log_buf(tp, bp,
+				(uint)((char *)first - (char *)bufp),
+				(uint)((char *)b - (char *)bufp));
+			error = xfs_rtbuf_get(mp, tp, ++block, 0, &bp);
+			if (error) {
+				return error;
+			}
+			first = b = bufp = (xfs_rtword_t *)XFS_BUF_PTR(bp);
+			word = 0;
+		} else {
+			/*
+			 * Go on to the next word in the buffer
+			 */
+			b++;
+		}
+	} else {
+		/*
+		 * Starting on a word boundary, no partial word.
+		 */
+		i = 0;
+	}
+	/*
+	 * Loop over whole words in buffers.  When we use up one buffer
+	 * we move on to the next one.
+	 */
+	while (len - i >= XFS_NBWORD) {
+		/*
+		 * Set the word value correctly.
+		 */
+		*b = val;
+		i += XFS_NBWORD;
+		/*
+		 * Go on to the next block if that's where the next word is
+		 * and we need the next word.
+		 */
+		if (++word == XFS_BLOCKWSIZE(mp) && i < len) {
+			/*
+			 * Log the changed part of this block.
+			 * Get the next one.
+			 */
+			xfs_trans_log_buf(tp, bp,
+				(uint)((char *)first - (char *)bufp),
+				(uint)((char *)b - (char *)bufp));
+			error = xfs_rtbuf_get(mp, tp, ++block, 0, &bp);
+			if (error) {
+				return error;
+			}
+			first = b = bufp = (xfs_rtword_t *)XFS_BUF_PTR(bp);
+			word = 0;
+		} else {
+			/*
+			 * Go on to the next word in the buffer
+			 */
+			b++;
+		}
+	}
+	/*
+	 * If not ending on a word boundary, deal with the last
+	 * (partial) word.
+	 */
+	if ((lastbit = len - i)) {
+		/*
+		 * Compute a mask of relevant bits.
+		 */
+		bit = 0;
+		mask = ((xfs_rtword_t)1 << lastbit) - 1;
+		/*
+		 * Set/clear the active bits.
+		 */
+		if (val)
+			*b |= mask;
+		else
+			*b &= ~mask;
+		b++;
+	}
+	/*
+	 * Log any remaining changed bytes.
+	 */
+	if (b > first)
+		xfs_trans_log_buf(tp, bp, (uint)((char *)first - (char *)bufp),
+			(uint)((char *)b - (char *)bufp - 1));
+	return 0;
+}
+
+/*
+ * Read and modify the summary information for a given extent size,
+ * bitmap block combination.
+ * Keeps track of a current summary block, so we don't keep reading
+ * it from the buffer cache.
+ */
+STATIC int				/* error */
+xfs_rtmodify_summary(
+	xfs_mount_t	*mp,		/* file system mount point */
+	xfs_trans_t	*tp,		/* transaction pointer */
+	int		log,		/* log2 of extent size */
+	xfs_rtblock_t	bbno,		/* bitmap block number */
+	int		delta,		/* change to make to summary info */
+	xfs_buf_t	**rbpp,		/* in/out: summary block buffer */
+	xfs_fsblock_t	*rsb)		/* in/out: summary block number */
+{
+	xfs_buf_t	*bp;		/* buffer for the summary block */
+	int		error;		/* error value */
+	xfs_fsblock_t	sb;		/* summary fsblock */
+	int		so;		/* index into the summary file */
+	xfs_suminfo_t	*sp;		/* pointer to returned data */
+
+	/*
+	 * Compute entry number in the summary file.
+	 */
+	so = XFS_SUMOFFS(mp, log, bbno);
+	/*
+	 * Compute the block number in the summary file.
+	 */
+	sb = XFS_SUMOFFSTOBLOCK(mp, so);
+	/*
+	 * If we have an old buffer, and the block number matches, use that.
+	 */
+	if (rbpp && *rbpp && *rsb == sb)
+		bp = *rbpp;
+	/*
+	 * Otherwise we have to get the buffer.
+	 */
+	else {
+		/*
+		 * If there was an old one, get rid of it first.
+		 */
+		if (rbpp && *rbpp)
+			xfs_trans_brelse(tp, *rbpp);
+		error = xfs_rtbuf_get(mp, tp, sb, 1, &bp);
+		if (error) {
+			return error;
+		}
+		/*
+		 * Remember this buffer and block for the next call.
+		 */
+		if (rbpp) {
+			*rbpp = bp;
+			*rsb = sb;
+		}
+	}
+	/*
+	 * Point to the summary information, modify and log it.
+	 */
+	sp = XFS_SUMPTR(mp, bp, so);
+	*sp += delta;
+	xfs_trans_log_buf(tp, bp, (uint)((char *)sp - (char *)XFS_BUF_PTR(bp)),
+		(uint)((char *)sp - (char *)XFS_BUF_PTR(bp) + sizeof(*sp) - 1));
+	return 0;
+}
+
+/*
+ * Visible (exported) functions.
+ */
+
+/*
+ * Grow the realtime area of the filesystem.
+ */
+int
+xfs_growfs_rt(
+	xfs_mount_t	*mp,		/* mount point for filesystem */
+	xfs_growfs_rt_t	*in)		/* growfs rt input struct */
+{
+	xfs_rtblock_t	bmbno;		/* bitmap block number */
+	xfs_buf_t	*bp;		/* temporary buffer */
+	int		error;		/* error return value */
+	xfs_mount_t	*nmp;		/* new (fake) mount structure */
+	xfs_drfsbno_t	nrblocks;	/* new number of realtime blocks */
+	xfs_extlen_t	nrbmblocks;	/* new number of rt bitmap blocks */
+	xfs_drtbno_t	nrextents;	/* new number of realtime extents */
+	uint8_t		nrextslog;	/* new log2 of sb_rextents */
+	xfs_extlen_t	nrsumblocks;	/* new number of summary blocks */
+	uint		nrsumlevels;	/* new rt summary levels */
+	uint		nrsumsize;	/* new size of rt summary, bytes */
+	xfs_sb_t	*nsbp;		/* new superblock */
+	xfs_extlen_t	rbmblocks;	/* current number of rt bitmap blocks */
+	xfs_extlen_t	rsumblocks;	/* current number of rt summary blks */
+	xfs_sb_t	*sbp;		/* old superblock */
+	xfs_fsblock_t	sumbno;		/* summary block number */
+
+	sbp = &mp->m_sb;
+	/*
+	 * Initial error checking.
+	 */
+	if (!capable(CAP_SYS_ADMIN))
+		return XFS_ERROR(EPERM);
+	if (mp->m_rtdev_targp == NULL || mp->m_rbmip == NULL ||
+	    (nrblocks = in->newblocks) <= sbp->sb_rblocks ||
+	    (sbp->sb_rblocks && (in->extsize != sbp->sb_rextsize)))
+		return XFS_ERROR(EINVAL);
+	if ((error = xfs_sb_validate_fsb_count(sbp, nrblocks)))
+		return error;
+	/*
+	 * Read in the last block of the device, make sure it exists.
+	 */
+	bp = xfs_buf_read_uncached(mp, mp->m_rtdev_targp,
+				XFS_FSB_TO_BB(mp, nrblocks - 1),
+				XFS_FSB_TO_B(mp, 1), 0);
+	if (!bp)
+		return EIO;
+	xfs_buf_relse(bp);
+
+	/*
+	 * Calculate new parameters.  These are the final values to be reached.
+	 */
+	nrextents = nrblocks;
+	do_div(nrextents, in->extsize);
+	nrbmblocks = howmany_64(nrextents, NBBY * sbp->sb_blocksize);
+	nrextslog = xfs_highbit32(nrextents);
+	nrsumlevels = nrextslog + 1;
+	nrsumsize = (uint)sizeof(xfs_suminfo_t) * nrsumlevels * nrbmblocks;
+	nrsumblocks = XFS_B_TO_FSB(mp, nrsumsize);
+	nrsumsize = XFS_FSB_TO_B(mp, nrsumblocks);
+	/*
+	 * New summary size can't be more than half the size of
+	 * the log.  This prevents us from getting a log overflow,
+	 * since we'll log basically the whole summary file at once.
+	 */
+	if (nrsumblocks > (mp->m_sb.sb_logblocks >> 1))
+		return XFS_ERROR(EINVAL);
+	/*
+	 * Get the old block counts for bitmap and summary inodes.
+	 * These can't change since other growfs callers are locked out.
+	 */
+	rbmblocks = XFS_B_TO_FSB(mp, mp->m_rbmip->i_d.di_size);
+	rsumblocks = XFS_B_TO_FSB(mp, mp->m_rsumip->i_d.di_size);
+	/*
+	 * Allocate space to the bitmap and summary files, as necessary.
+	 */
+	error = xfs_growfs_rt_alloc(mp, rbmblocks, nrbmblocks, mp->m_rbmip);
+	if (error)
+		return error;
+	error = xfs_growfs_rt_alloc(mp, rsumblocks, nrsumblocks, mp->m_rsumip);
+	if (error)
+		return error;
+	/*
+	 * Allocate a new (fake) mount/sb.
+	 */
+	nmp = kmem_alloc(sizeof(*nmp), KM_SLEEP);
+	/*
+	 * Loop over the bitmap blocks.
+	 * We will do everything one bitmap block at a time.
+	 * Skip the current block if it is exactly full.
+	 * This also deals with the case where there were no rtextents before.
+	 */
+	for (bmbno = sbp->sb_rbmblocks -
+		     ((sbp->sb_rextents & ((1 << mp->m_blkbit_log) - 1)) != 0);
+	     bmbno < nrbmblocks;
+	     bmbno++) {
+		xfs_trans_t	*tp;
+		int		cancelflags = 0;
+
+		*nmp = *mp;
+		nsbp = &nmp->m_sb;
+		/*
+		 * Calculate new sb and mount fields for this round.
+		 */
+		nsbp->sb_rextsize = in->extsize;
+		nsbp->sb_rbmblocks = bmbno + 1;
+		nsbp->sb_rblocks =
+			XFS_RTMIN(nrblocks,
+				  nsbp->sb_rbmblocks * NBBY *
+				  nsbp->sb_blocksize * nsbp->sb_rextsize);
+		nsbp->sb_rextents = nsbp->sb_rblocks;
+		do_div(nsbp->sb_rextents, nsbp->sb_rextsize);
+		ASSERT(nsbp->sb_rextents != 0);
+		nsbp->sb_rextslog = xfs_highbit32(nsbp->sb_rextents);
+		nrsumlevels = nmp->m_rsumlevels = nsbp->sb_rextslog + 1;
+		nrsumsize =
+			(uint)sizeof(xfs_suminfo_t) * nrsumlevels *
+			nsbp->sb_rbmblocks;
+		nrsumblocks = XFS_B_TO_FSB(mp, nrsumsize);
+		nmp->m_rsumsize = nrsumsize = XFS_FSB_TO_B(mp, nrsumblocks);
+		/*
+		 * Start a transaction, get the log reservation.
+		 */
+		tp = xfs_trans_alloc(mp, XFS_TRANS_GROWFSRT_FREE);
+		if ((error = xfs_trans_reserve(tp, 0,
+				XFS_GROWRTFREE_LOG_RES(nmp), 0, 0, 0)))
+			goto error_cancel;
+		/*
+		 * Lock out other callers by grabbing the bitmap inode lock.
+		 */
+		xfs_ilock(mp->m_rbmip, XFS_ILOCK_EXCL);
+		xfs_trans_ijoin_ref(tp, mp->m_rbmip, XFS_ILOCK_EXCL);
+		/*
+		 * Update the bitmap inode's size.
+		 */
+		mp->m_rbmip->i_d.di_size =
+			nsbp->sb_rbmblocks * nsbp->sb_blocksize;
+		xfs_trans_log_inode(tp, mp->m_rbmip, XFS_ILOG_CORE);
+		cancelflags |= XFS_TRANS_ABORT;
+		/*
+		 * Get the summary inode into the transaction.
+		 */
+		xfs_ilock(mp->m_rsumip, XFS_ILOCK_EXCL);
+		xfs_trans_ijoin_ref(tp, mp->m_rsumip, XFS_ILOCK_EXCL);
+		/*
+		 * Update the summary inode's size.
+		 */
+		mp->m_rsumip->i_d.di_size = nmp->m_rsumsize;
+		xfs_trans_log_inode(tp, mp->m_rsumip, XFS_ILOG_CORE);
+		/*
+		 * Copy summary data from old to new sizes.
+		 * Do this when the real size (not block-aligned) changes.
+		 */
+		if (sbp->sb_rbmblocks != nsbp->sb_rbmblocks ||
+		    mp->m_rsumlevels != nmp->m_rsumlevels) {
+			error = xfs_rtcopy_summary(mp, nmp, tp);
+			if (error)
+				goto error_cancel;
+		}
+		/*
+		 * Update superblock fields.
+		 */
+		if (nsbp->sb_rextsize != sbp->sb_rextsize)
+			xfs_trans_mod_sb(tp, XFS_TRANS_SB_REXTSIZE,
+				nsbp->sb_rextsize - sbp->sb_rextsize);
+		if (nsbp->sb_rbmblocks != sbp->sb_rbmblocks)
+			xfs_trans_mod_sb(tp, XFS_TRANS_SB_RBMBLOCKS,
+				nsbp->sb_rbmblocks - sbp->sb_rbmblocks);
+		if (nsbp->sb_rblocks != sbp->sb_rblocks)
+			xfs_trans_mod_sb(tp, XFS_TRANS_SB_RBLOCKS,
+				nsbp->sb_rblocks - sbp->sb_rblocks);
+		if (nsbp->sb_rextents != sbp->sb_rextents)
+			xfs_trans_mod_sb(tp, XFS_TRANS_SB_REXTENTS,
+				nsbp->sb_rextents - sbp->sb_rextents);
+		if (nsbp->sb_rextslog != sbp->sb_rextslog)
+			xfs_trans_mod_sb(tp, XFS_TRANS_SB_REXTSLOG,
+				nsbp->sb_rextslog - sbp->sb_rextslog);
+		/*
+		 * Free new extent.
+		 */
+		bp = NULL;
+		error = xfs_rtfree_range(nmp, tp, sbp->sb_rextents,
+			nsbp->sb_rextents - sbp->sb_rextents, &bp, &sumbno);
+		if (error) {
+error_cancel:
+			xfs_trans_cancel(tp, cancelflags);
+			break;
+		}
+		/*
+		 * Mark more blocks free in the superblock.
+		 */
+		xfs_trans_mod_sb(tp, XFS_TRANS_SB_FREXTENTS,
+			nsbp->sb_rextents - sbp->sb_rextents);
+		/*
+		 * Update mp values into the real mp structure.
+		 */
+		mp->m_rsumlevels = nrsumlevels;
+		mp->m_rsumsize = nrsumsize;
+
+		error = xfs_trans_commit(tp, 0);
+		if (error)
+			break;
+	}
+
+	/*
+	 * Free the fake mp structure.
+	 */
+	kmem_free(nmp);
+
+	return error;
+}
+
+/*
+ * Allocate an extent in the realtime subvolume, with the usual allocation
+ * parameters.  The length units are all in realtime extents, as is the
+ * result block number.
+ */
+int					/* error */
+xfs_rtallocate_extent(
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_rtblock_t	bno,		/* starting block number to allocate */
+	xfs_extlen_t	minlen,		/* minimum length to allocate */
+	xfs_extlen_t	maxlen,		/* maximum length to allocate */
+	xfs_extlen_t	*len,		/* out: actual length allocated */
+	xfs_alloctype_t	type,		/* allocation type XFS_ALLOCTYPE... */
+	int		wasdel,		/* was a delayed allocation extent */
+	xfs_extlen_t	prod,		/* extent product factor */
+	xfs_rtblock_t	*rtblock)	/* out: start block allocated */
+{
+	xfs_mount_t	*mp = tp->t_mountp;
+	int		error;		/* error value */
+	xfs_rtblock_t	r;		/* result allocated block */
+	xfs_fsblock_t	sb;		/* summary file block number */
+	xfs_buf_t	*sumbp;		/* summary file block buffer */
+
+	ASSERT(xfs_isilocked(mp->m_rbmip, XFS_ILOCK_EXCL));
+	ASSERT(minlen > 0 && minlen <= maxlen);
+
+	/*
+	 * If prod is set then figure out what to do to minlen and maxlen.
+	 */
+	if (prod > 1) {
+		xfs_extlen_t	i;
+
+		if ((i = maxlen % prod))
+			maxlen -= i;
+		if ((i = minlen % prod))
+			minlen += prod - i;
+		if (maxlen < minlen) {
+			*rtblock = NULLRTBLOCK;
+			return 0;
+		}
+	}
+
+	sumbp = NULL;
+	/*
+	 * Allocate by size, or near another block, or exactly at some block.
+	 */
+	switch (type) {
+	case XFS_ALLOCTYPE_ANY_AG:
+		error = xfs_rtallocate_extent_size(mp, tp, minlen, maxlen, len,
+				&sumbp,	&sb, prod, &r);
+		break;
+	case XFS_ALLOCTYPE_NEAR_BNO:
+		error = xfs_rtallocate_extent_near(mp, tp, bno, minlen, maxlen,
+				len, &sumbp, &sb, prod, &r);
+		break;
+	case XFS_ALLOCTYPE_THIS_BNO:
+		error = xfs_rtallocate_extent_exact(mp, tp, bno, minlen, maxlen,
+				len, &sumbp, &sb, prod, &r);
+		break;
+	default:
+		error = EIO;
+		ASSERT(0);
+	}
+	if (error)
+		return error;
+
+	/*
+	 * If it worked, update the superblock.
+	 */
+	if (r != NULLRTBLOCK) {
+		long	slen = (long)*len;
+
+		ASSERT(*len >= minlen && *len <= maxlen);
+		if (wasdel)
+			xfs_trans_mod_sb(tp, XFS_TRANS_SB_RES_FREXTENTS, -slen);
+		else
+			xfs_trans_mod_sb(tp, XFS_TRANS_SB_FREXTENTS, -slen);
+	}
+	*rtblock = r;
+	return 0;
+}
+
+/*
+ * Free an extent in the realtime subvolume.  Length is expressed in
+ * realtime extents, as is the block number.
+ */
+int					/* error */
+xfs_rtfree_extent(
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_rtblock_t	bno,		/* starting block number to free */
+	xfs_extlen_t	len)		/* length of extent freed */
+{
+	int		error;		/* error value */
+	xfs_mount_t	*mp;		/* file system mount structure */
+	xfs_fsblock_t	sb;		/* summary file block number */
+	xfs_buf_t	*sumbp;		/* summary file block buffer */
+
+	mp = tp->t_mountp;
+	/*
+	 * Synchronize by locking the bitmap inode.
+	 */
+	xfs_ilock(mp->m_rbmip, XFS_ILOCK_EXCL);
+	xfs_trans_ijoin_ref(tp, mp->m_rbmip, XFS_ILOCK_EXCL);
+
+#if defined(__KERNEL__) && defined(DEBUG)
+	/*
+	 * Check to see that this whole range is currently allocated.
+	 */
+	{
+		int	stat;		/* result from checking range */
+
+		error = xfs_rtcheck_alloc_range(mp, tp, bno, len, &stat);
+		if (error) {
+			return error;
+		}
+		ASSERT(stat);
+	}
+#endif
+	sumbp = NULL;
+	/*
+	 * Free the range of realtime blocks.
+	 */
+	error = xfs_rtfree_range(mp, tp, bno, len, &sumbp, &sb);
+	if (error) {
+		return error;
+	}
+	/*
+	 * Mark more blocks free in the superblock.
+	 */
+	xfs_trans_mod_sb(tp, XFS_TRANS_SB_FREXTENTS, (long)len);
+	/*
+	 * If we've now freed all the blocks, reset the file sequence
+	 * number to 0.
+	 */
+	if (tp->t_frextents_delta + mp->m_sb.sb_frextents ==
+	    mp->m_sb.sb_rextents) {
+		if (!(mp->m_rbmip->i_d.di_flags & XFS_DIFLAG_NEWRTBM))
+			mp->m_rbmip->i_d.di_flags |= XFS_DIFLAG_NEWRTBM;
+		*(__uint64_t *)&mp->m_rbmip->i_d.di_atime = 0;
+		xfs_trans_log_inode(tp, mp->m_rbmip, XFS_ILOG_CORE);
+	}
+	return 0;
+}
+
+/*
+ * Initialize realtime fields in the mount structure.
+ */
+int				/* error */
+xfs_rtmount_init(
+	xfs_mount_t	*mp)	/* file system mount structure */
+{
+	xfs_buf_t	*bp;	/* buffer for last block of subvolume */
+	xfs_daddr_t	d;	/* address of last block of subvolume */
+	xfs_sb_t	*sbp;	/* filesystem superblock copy in mount */
+
+	sbp = &mp->m_sb;
+	if (sbp->sb_rblocks == 0)
+		return 0;
+	if (mp->m_rtdev_targp == NULL) {
+		xfs_warn(mp,
+	"Filesystem has a realtime volume, use rtdev=device option");
+		return XFS_ERROR(ENODEV);
+	}
+	mp->m_rsumlevels = sbp->sb_rextslog + 1;
+	mp->m_rsumsize =
+		(uint)sizeof(xfs_suminfo_t) * mp->m_rsumlevels *
+		sbp->sb_rbmblocks;
+	mp->m_rsumsize = roundup(mp->m_rsumsize, sbp->sb_blocksize);
+	mp->m_rbmip = mp->m_rsumip = NULL;
+	/*
+	 * Check that the realtime section is an ok size.
+	 */
+	d = (xfs_daddr_t)XFS_FSB_TO_BB(mp, mp->m_sb.sb_rblocks);
+	if (XFS_BB_TO_FSB(mp, d) != mp->m_sb.sb_rblocks) {
+		xfs_warn(mp, "realtime mount -- %llu != %llu",
+			(unsigned long long) XFS_BB_TO_FSB(mp, d),
+			(unsigned long long) mp->m_sb.sb_rblocks);
+		return XFS_ERROR(EFBIG);
+	}
+	bp = xfs_buf_read_uncached(mp, mp->m_rtdev_targp,
+					d - XFS_FSB_TO_BB(mp, 1),
+					XFS_FSB_TO_B(mp, 1), 0);
+	if (!bp) {
+		xfs_warn(mp, "realtime device size check failed");
+		return EIO;
+	}
+	xfs_buf_relse(bp);
+	return 0;
+}
+
+/*
+ * Get the bitmap and summary inodes into the mount structure
+ * at mount time.
+ */
+int					/* error */
+xfs_rtmount_inodes(
+	xfs_mount_t	*mp)		/* file system mount structure */
+{
+	int		error;		/* error return value */
+	xfs_sb_t	*sbp;
+
+	sbp = &mp->m_sb;
+	if (sbp->sb_rbmino == NULLFSINO)
+		return 0;
+	error = xfs_iget(mp, NULL, sbp->sb_rbmino, 0, 0, &mp->m_rbmip);
+	if (error)
+		return error;
+	ASSERT(mp->m_rbmip != NULL);
+	ASSERT(sbp->sb_rsumino != NULLFSINO);
+	error = xfs_iget(mp, NULL, sbp->sb_rsumino, 0, 0, &mp->m_rsumip);
+	if (error) {
+		IRELE(mp->m_rbmip);
+		return error;
+	}
+	ASSERT(mp->m_rsumip != NULL);
+	return 0;
+}
+
+void
+xfs_rtunmount_inodes(
+	struct xfs_mount	*mp)
+{
+	if (mp->m_rbmip)
+		IRELE(mp->m_rbmip);
+	if (mp->m_rsumip)
+		IRELE(mp->m_rsumip);
+}
+
+/*
+ * Pick an extent for allocation at the start of a new realtime file.
+ * Use the sequence number stored in the atime field of the bitmap inode.
+ * Translate this to a fraction of the rtextents, and return the product
+ * of rtextents and the fraction.
+ * The fraction sequence is 0, 1/2, 1/4, 3/4, 1/8, ..., 7/8, 1/16, ...
+ */
+int					/* error */
+xfs_rtpick_extent(
+	xfs_mount_t	*mp,		/* file system mount point */
+	xfs_trans_t	*tp,		/* transaction pointer */
+	xfs_extlen_t	len,		/* allocation length (rtextents) */
+	xfs_rtblock_t	*pick)		/* result rt extent */
+{
+	xfs_rtblock_t	b;		/* result block */
+	int		log2;		/* log of sequence number */
+	__uint64_t	resid;		/* residual after log removed */
+	__uint64_t	seq;		/* sequence number of file creation */
+	__uint64_t	*seqp;		/* pointer to seqno in inode */
+
+	ASSERT(xfs_isilocked(mp->m_rbmip, XFS_ILOCK_EXCL));
+
+	seqp = (__uint64_t *)&mp->m_rbmip->i_d.di_atime;
+	if (!(mp->m_rbmip->i_d.di_flags & XFS_DIFLAG_NEWRTBM)) {
+		mp->m_rbmip->i_d.di_flags |= XFS_DIFLAG_NEWRTBM;
+		*seqp = 0;
+	}
+	seq = *seqp;
+	if ((log2 = xfs_highbit64(seq)) == -1)
+		b = 0;
+	else {
+		resid = seq - (1ULL << log2);
+		b = (mp->m_sb.sb_rextents * ((resid << 1) + 1ULL)) >>
+		    (log2 + 1);
+		if (b >= mp->m_sb.sb_rextents)
+			b = do_mod(b, mp->m_sb.sb_rextents);
+		if (b + len > mp->m_sb.sb_rextents)
+			b = mp->m_sb.sb_rextents - len;
+	}
+	*seqp = seq + 1;
+	xfs_trans_log_inode(tp, mp->m_rbmip, XFS_ILOG_CORE);
+	*pick = b;
+	return 0;
+}
diff --git a/fs/xfs/xfs_rtalloc.h b/fs/xfs/xfs_rtalloc.h
new file mode 100644
index 00000000..09e1f4f3
--- /dev/null
+++ b/fs/xfs/xfs_rtalloc.h
@@ -0,0 +1,166 @@
+/*
+ * Copyright (c) 2000-2003,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_RTALLOC_H__
+#define	__XFS_RTALLOC_H__
+
+struct xfs_mount;
+struct xfs_trans;
+
+/* Min and max rt extent sizes, specified in bytes */
+#define	XFS_MAX_RTEXTSIZE	(1024 * 1024 * 1024)	/* 1GB */
+#define	XFS_DFL_RTEXTSIZE	(64 * 1024)	        /* 64kB */
+#define	XFS_MIN_RTEXTSIZE	(4 * 1024)		/* 4kB */
+
+/*
+ * Constants for bit manipulations.
+ */
+#define	XFS_NBBYLOG	3		/* log2(NBBY) */
+#define	XFS_WORDLOG	2		/* log2(sizeof(xfs_rtword_t)) */
+#define	XFS_NBWORDLOG	(XFS_NBBYLOG + XFS_WORDLOG)
+#define	XFS_NBWORD	(1 << XFS_NBWORDLOG)
+#define	XFS_WORDMASK	((1 << XFS_WORDLOG) - 1)
+
+#define	XFS_BLOCKSIZE(mp)	((mp)->m_sb.sb_blocksize)
+#define	XFS_BLOCKMASK(mp)	((mp)->m_blockmask)
+#define	XFS_BLOCKWSIZE(mp)	((mp)->m_blockwsize)
+#define	XFS_BLOCKWMASK(mp)	((mp)->m_blockwmask)
+
+/*
+ * Summary and bit manipulation macros.
+ */
+#define	XFS_SUMOFFS(mp,ls,bb)	((int)((ls) * (mp)->m_sb.sb_rbmblocks + (bb)))
+#define	XFS_SUMOFFSTOBLOCK(mp,s)	\
+	(((s) * (uint)sizeof(xfs_suminfo_t)) >> (mp)->m_sb.sb_blocklog)
+#define	XFS_SUMPTR(mp,bp,so)	\
+	((xfs_suminfo_t *)((char *)XFS_BUF_PTR(bp) + \
+		(((so) * (uint)sizeof(xfs_suminfo_t)) & XFS_BLOCKMASK(mp))))
+
+#define	XFS_BITTOBLOCK(mp,bi)	((bi) >> (mp)->m_blkbit_log)
+#define	XFS_BLOCKTOBIT(mp,bb)	((bb) << (mp)->m_blkbit_log)
+#define	XFS_BITTOWORD(mp,bi)	\
+	((int)(((bi) >> XFS_NBWORDLOG) & XFS_BLOCKWMASK(mp)))
+
+#define	XFS_RTMIN(a,b)	((a) < (b) ? (a) : (b))
+#define	XFS_RTMAX(a,b)	((a) > (b) ? (a) : (b))
+
+#define	XFS_RTLOBIT(w)	xfs_lowbit32(w)
+#define	XFS_RTHIBIT(w)	xfs_highbit32(w)
+
+#if XFS_BIG_BLKNOS
+#define	XFS_RTBLOCKLOG(b)	xfs_highbit64(b)
+#else
+#define	XFS_RTBLOCKLOG(b)	xfs_highbit32(b)
+#endif
+
+
+#ifdef __KERNEL__
+
+#ifdef CONFIG_XFS_RT
+/*
+ * Function prototypes for exported functions.
+ */
+
+/*
+ * Allocate an extent in the realtime subvolume, with the usual allocation
+ * parameters.  The length units are all in realtime extents, as is the
+ * result block number.
+ */
+int					/* error */
+xfs_rtallocate_extent(
+	struct xfs_trans	*tp,	/* transaction pointer */
+	xfs_rtblock_t		bno,	/* starting block number to allocate */
+	xfs_extlen_t		minlen,	/* minimum length to allocate */
+	xfs_extlen_t		maxlen,	/* maximum length to allocate */
+	xfs_extlen_t		*len,	/* out: actual length allocated */
+	xfs_alloctype_t		type,	/* allocation type XFS_ALLOCTYPE... */
+	int			wasdel,	/* was a delayed allocation extent */
+	xfs_extlen_t		prod,	/* extent product factor */
+	xfs_rtblock_t		*rtblock); /* out: start block allocated */
+
+/*
+ * Free an extent in the realtime subvolume.  Length is expressed in
+ * realtime extents, as is the block number.
+ */
+int					/* error */
+xfs_rtfree_extent(
+	struct xfs_trans	*tp,	/* transaction pointer */
+	xfs_rtblock_t		bno,	/* starting block number to free */
+	xfs_extlen_t		len);	/* length of extent freed */
+
+/*
+ * Initialize realtime fields in the mount structure.
+ */
+int					/* error */
+xfs_rtmount_init(
+	struct xfs_mount	*mp);	/* file system mount structure */
+void
+xfs_rtunmount_inodes(
+	struct xfs_mount	*mp);
+
+/*
+ * Get the bitmap and summary inodes into the mount structure
+ * at mount time.
+ */
+int					/* error */
+xfs_rtmount_inodes(
+	struct xfs_mount	*mp);	/* file system mount structure */
+
+/*
+ * Pick an extent for allocation at the start of a new realtime file.
+ * Use the sequence number stored in the atime field of the bitmap inode.
+ * Translate this to a fraction of the rtextents, and return the product
+ * of rtextents and the fraction.
+ * The fraction sequence is 0, 1/2, 1/4, 3/4, 1/8, ..., 7/8, 1/16, ...
+ */
+int					/* error */
+xfs_rtpick_extent(
+	struct xfs_mount	*mp,	/* file system mount point */
+	struct xfs_trans	*tp,	/* transaction pointer */
+	xfs_extlen_t		len,	/* allocation length (rtextents) */
+	xfs_rtblock_t		*pick);	/* result rt extent */
+
+/*
+ * Grow the realtime area of the filesystem.
+ */
+int
+xfs_growfs_rt(
+	struct xfs_mount	*mp,	/* file system mount structure */
+	xfs_growfs_rt_t		*in);	/* user supplied growfs struct */
+
+#else
+# define xfs_rtallocate_extent(t,b,min,max,l,a,f,p,rb)  (ENOSYS)
+# define xfs_rtfree_extent(t,b,l)                       (ENOSYS)
+# define xfs_rtpick_extent(m,t,l,rb)                    (ENOSYS)
+# define xfs_growfs_rt(mp,in)                           (ENOSYS)
+static inline int		/* error */
+xfs_rtmount_init(
+	xfs_mount_t	*mp)	/* file system mount structure */
+{
+	if (mp->m_sb.sb_rblocks == 0)
+		return 0;
+
+	xfs_warn(mp, "Not built with CONFIG_XFS_RT");
+	return ENOSYS;
+}
+# define xfs_rtmount_inodes(m)  (((mp)->m_sb.sb_rblocks == 0)? 0 : (ENOSYS))
+# define xfs_rtunmount_inodes(m)
+#endif	/* CONFIG_XFS_RT */
+
+#endif	/* __KERNEL__ */
+
+#endif	/* __XFS_RTALLOC_H__ */
diff --git a/fs/xfs/xfs_rw.c b/fs/xfs/xfs_rw.c
new file mode 100644
index 00000000..d6d6fdfe
--- /dev/null
+++ b/fs/xfs/xfs_rw.c
@@ -0,0 +1,175 @@
+/*
+ * Copyright (c) 2000-2006 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_mount.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_error.h"
+#include "xfs_rw.h"
+
+/*
+ * Force a shutdown of the filesystem instantly while keeping
+ * the filesystem consistent. We don't do an unmount here; just shutdown
+ * the shop, make sure that absolutely nothing persistent happens to
+ * this filesystem after this point.
+ */
+void
+xfs_do_force_shutdown(
+	xfs_mount_t	*mp,
+	int		flags,
+	char		*fname,
+	int		lnnum)
+{
+	int		logerror;
+
+	logerror = flags & SHUTDOWN_LOG_IO_ERROR;
+
+	if (!(flags & SHUTDOWN_FORCE_UMOUNT)) {
+		xfs_notice(mp,
+	"%s(0x%x) called from line %d of file %s.  Return address = 0x%p",
+			__func__, flags, lnnum, fname, __return_address);
+	}
+	/*
+	 * No need to duplicate efforts.
+	 */
+	if (XFS_FORCED_SHUTDOWN(mp) && !logerror)
+		return;
+
+	/*
+	 * This flags XFS_MOUNT_FS_SHUTDOWN, makes sure that we don't
+	 * queue up anybody new on the log reservations, and wakes up
+	 * everybody who's sleeping on log reservations to tell them
+	 * the bad news.
+	 */
+	if (xfs_log_force_umount(mp, logerror))
+		return;
+
+	if (flags & SHUTDOWN_CORRUPT_INCORE) {
+		xfs_alert_tag(mp, XFS_PTAG_SHUTDOWN_CORRUPT,
+    "Corruption of in-memory data detected.  Shutting down filesystem");
+		if (XFS_ERRLEVEL_HIGH <= xfs_error_level)
+			xfs_stack_trace();
+	} else if (!(flags & SHUTDOWN_FORCE_UMOUNT)) {
+		if (logerror) {
+			xfs_alert_tag(mp, XFS_PTAG_SHUTDOWN_LOGERROR,
+		"Log I/O Error Detected.  Shutting down filesystem");
+		} else if (flags & SHUTDOWN_DEVICE_REQ) {
+			xfs_alert_tag(mp, XFS_PTAG_SHUTDOWN_IOERROR,
+		"All device paths lost.  Shutting down filesystem");
+		} else if (!(flags & SHUTDOWN_REMOTE_REQ)) {
+			xfs_alert_tag(mp, XFS_PTAG_SHUTDOWN_IOERROR,
+		"I/O Error Detected. Shutting down filesystem");
+		}
+	}
+	if (!(flags & SHUTDOWN_FORCE_UMOUNT)) {
+		xfs_alert(mp,
+	"Please umount the filesystem and rectify the problem(s)");
+	}
+}
+
+/*
+ * Prints out an ALERT message about I/O error.
+ */
+void
+xfs_ioerror_alert(
+	char			*func,
+	struct xfs_mount	*mp,
+	xfs_buf_t		*bp,
+	xfs_daddr_t		blkno)
+{
+	xfs_alert(mp,
+		 "I/O error occurred: meta-data dev %s block 0x%llx"
+		 "       (\"%s\") error %d buf count %zd",
+		XFS_BUFTARG_NAME(XFS_BUF_TARGET(bp)),
+		(__uint64_t)blkno, func,
+		XFS_BUF_GETERROR(bp), XFS_BUF_COUNT(bp));
+}
+
+/*
+ * This isn't an absolute requirement, but it is
+ * just a good idea to call xfs_read_buf instead of
+ * directly doing a read_buf call. For one, we shouldn't
+ * be doing this disk read if we are in SHUTDOWN state anyway,
+ * so this stops that from happening. Secondly, this does all
+ * the error checking stuff and the brelse if appropriate for
+ * the caller, so the code can be a little leaner.
+ */
+
+int
+xfs_read_buf(
+	struct xfs_mount *mp,
+	xfs_buftarg_t	 *target,
+	xfs_daddr_t	 blkno,
+	int              len,
+	uint             flags,
+	xfs_buf_t	 **bpp)
+{
+	xfs_buf_t	 *bp;
+	int		 error;
+
+	if (!flags)
+		flags = XBF_LOCK | XBF_MAPPED;
+
+	bp = xfs_buf_read(target, blkno, len, flags);
+	if (!bp)
+		return XFS_ERROR(EIO);
+	error = XFS_BUF_GETERROR(bp);
+	if (bp && !error && !XFS_FORCED_SHUTDOWN(mp)) {
+		*bpp = bp;
+	} else {
+		*bpp = NULL;
+		if (error) {
+			xfs_ioerror_alert("xfs_read_buf", mp, bp, XFS_BUF_ADDR(bp));
+		} else {
+			error = XFS_ERROR(EIO);
+		}
+		if (bp) {
+			XFS_BUF_UNDONE(bp);
+			XFS_BUF_UNDELAYWRITE(bp);
+			XFS_BUF_STALE(bp);
+			/*
+			 * brelse clears B_ERROR and b_error
+			 */
+			xfs_buf_relse(bp);
+		}
+	}
+	return (error);
+}
+
+/*
+ * helper function to extract extent size hint from inode
+ */
+xfs_extlen_t
+xfs_get_extsz_hint(
+	struct xfs_inode	*ip)
+{
+	if ((ip->i_d.di_flags & XFS_DIFLAG_EXTSIZE) && ip->i_d.di_extsize)
+		return ip->i_d.di_extsize;
+	if (XFS_IS_REALTIME_INODE(ip))
+		return ip->i_mount->m_sb.sb_rextsize;
+	return 0;
+}
diff --git a/fs/xfs/xfs_rw.h b/fs/xfs/xfs_rw.h
new file mode 100644
index 00000000..11c41ec6
--- /dev/null
+++ b/fs/xfs/xfs_rw.h
@@ -0,0 +1,49 @@
+/*
+ * Copyright (c) 2000-2006 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef	__XFS_RW_H__
+#define	__XFS_RW_H__
+
+struct xfs_buf;
+struct xfs_inode;
+struct xfs_mount;
+
+/*
+ * Convert the given file system block to a disk block.
+ * We have to treat it differently based on whether the
+ * file is a real time file or not, because the bmap code
+ * does.
+ */
+static inline xfs_daddr_t
+xfs_fsb_to_db(struct xfs_inode *ip, xfs_fsblock_t fsb)
+{
+	return (XFS_IS_REALTIME_INODE(ip) ? \
+		 (xfs_daddr_t)XFS_FSB_TO_BB((ip)->i_mount, (fsb)) : \
+		 XFS_FSB_TO_DADDR((ip)->i_mount, (fsb)));
+}
+
+/*
+ * Prototypes for functions in xfs_rw.c.
+ */
+extern int xfs_read_buf(struct xfs_mount *mp, xfs_buftarg_t *btp,
+			xfs_daddr_t blkno, int len, uint flags,
+			struct xfs_buf **bpp);
+extern void xfs_ioerror_alert(char *func, struct xfs_mount *mp,
+				xfs_buf_t *bp, xfs_daddr_t blkno);
+extern xfs_extlen_t xfs_get_extsz_hint(struct xfs_inode *ip);
+
+#endif /* __XFS_RW_H__ */
diff --git a/fs/xfs/xfs_sb.h b/fs/xfs/xfs_sb.h
new file mode 100644
index 00000000..1eb2ba58
--- /dev/null
+++ b/fs/xfs/xfs_sb.h
@@ -0,0 +1,543 @@
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_SB_H__
+#define	__XFS_SB_H__
+
+/*
+ * Super block
+ * Fits into a sector-sized buffer at address 0 of each allocation group.
+ * Only the first of these is ever updated except during growfs.
+ */
+
+struct xfs_buf;
+struct xfs_mount;
+
+#define	XFS_SB_MAGIC		0x58465342	/* 'XFSB' */
+#define	XFS_SB_VERSION_1	1		/* 5.3, 6.0.1, 6.1 */
+#define	XFS_SB_VERSION_2	2		/* 6.2 - attributes */
+#define	XFS_SB_VERSION_3	3		/* 6.2 - new inode version */
+#define	XFS_SB_VERSION_4	4		/* 6.2+ - bitmask version */
+#define	XFS_SB_VERSION_NUMBITS		0x000f
+#define	XFS_SB_VERSION_ALLFBITS		0xfff0
+#define	XFS_SB_VERSION_SASHFBITS	0xf000
+#define	XFS_SB_VERSION_REALFBITS	0x0ff0
+#define	XFS_SB_VERSION_ATTRBIT		0x0010
+#define	XFS_SB_VERSION_NLINKBIT		0x0020
+#define	XFS_SB_VERSION_QUOTABIT		0x0040
+#define	XFS_SB_VERSION_ALIGNBIT		0x0080
+#define	XFS_SB_VERSION_DALIGNBIT	0x0100
+#define	XFS_SB_VERSION_SHAREDBIT	0x0200
+#define XFS_SB_VERSION_LOGV2BIT		0x0400
+#define XFS_SB_VERSION_SECTORBIT	0x0800
+#define	XFS_SB_VERSION_EXTFLGBIT	0x1000
+#define	XFS_SB_VERSION_DIRV2BIT		0x2000
+#define	XFS_SB_VERSION_BORGBIT		0x4000	/* ASCII only case-insens. */
+#define	XFS_SB_VERSION_MOREBITSBIT	0x8000
+#define	XFS_SB_VERSION_OKSASHFBITS	\
+	(XFS_SB_VERSION_EXTFLGBIT | \
+	 XFS_SB_VERSION_DIRV2BIT | \
+	 XFS_SB_VERSION_BORGBIT)
+#define	XFS_SB_VERSION_OKREALFBITS	\
+	(XFS_SB_VERSION_ATTRBIT | \
+	 XFS_SB_VERSION_NLINKBIT | \
+	 XFS_SB_VERSION_QUOTABIT | \
+	 XFS_SB_VERSION_ALIGNBIT | \
+	 XFS_SB_VERSION_DALIGNBIT | \
+	 XFS_SB_VERSION_SHAREDBIT | \
+	 XFS_SB_VERSION_LOGV2BIT | \
+	 XFS_SB_VERSION_SECTORBIT | \
+	 XFS_SB_VERSION_MOREBITSBIT)
+#define	XFS_SB_VERSION_OKREALBITS	\
+	(XFS_SB_VERSION_NUMBITS | \
+	 XFS_SB_VERSION_OKREALFBITS | \
+	 XFS_SB_VERSION_OKSASHFBITS)
+
+/*
+ * There are two words to hold XFS "feature" bits: the original
+ * word, sb_versionnum, and sb_features2.  Whenever a bit is set in
+ * sb_features2, the feature bit XFS_SB_VERSION_MOREBITSBIT must be set.
+ *
+ * These defines represent bits in sb_features2.
+ */
+#define XFS_SB_VERSION2_REALFBITS	0x00ffffff	/* Mask: features */
+#define XFS_SB_VERSION2_RESERVED1BIT	0x00000001
+#define XFS_SB_VERSION2_LAZYSBCOUNTBIT	0x00000002	/* Superblk counters */
+#define XFS_SB_VERSION2_RESERVED4BIT	0x00000004
+#define XFS_SB_VERSION2_ATTR2BIT	0x00000008	/* Inline attr rework */
+#define XFS_SB_VERSION2_PARENTBIT	0x00000010	/* parent pointers */
+#define XFS_SB_VERSION2_PROJID32BIT	0x00000080	/* 32 bit project id */
+
+#define	XFS_SB_VERSION2_OKREALFBITS	\
+	(XFS_SB_VERSION2_LAZYSBCOUNTBIT	| \
+	 XFS_SB_VERSION2_ATTR2BIT	| \
+	 XFS_SB_VERSION2_PROJID32BIT)
+#define	XFS_SB_VERSION2_OKSASHFBITS	\
+	(0)
+#define XFS_SB_VERSION2_OKREALBITS	\
+	(XFS_SB_VERSION2_OKREALFBITS |	\
+	 XFS_SB_VERSION2_OKSASHFBITS )
+
+/*
+ * Superblock - in core version.  Must match the ondisk version below.
+ * Must be padded to 64 bit alignment.
+ */
+typedef struct xfs_sb {
+	__uint32_t	sb_magicnum;	/* magic number == XFS_SB_MAGIC */
+	__uint32_t	sb_blocksize;	/* logical block size, bytes */
+	xfs_drfsbno_t	sb_dblocks;	/* number of data blocks */
+	xfs_drfsbno_t	sb_rblocks;	/* number of realtime blocks */
+	xfs_drtbno_t	sb_rextents;	/* number of realtime extents */
+	uuid_t		sb_uuid;	/* file system unique id */
+	xfs_dfsbno_t	sb_logstart;	/* starting block of log if internal */
+	xfs_ino_t	sb_rootino;	/* root inode number */
+	xfs_ino_t	sb_rbmino;	/* bitmap inode for realtime extents */
+	xfs_ino_t	sb_rsumino;	/* summary inode for rt bitmap */
+	xfs_agblock_t	sb_rextsize;	/* realtime extent size, blocks */
+	xfs_agblock_t	sb_agblocks;	/* size of an allocation group */
+	xfs_agnumber_t	sb_agcount;	/* number of allocation groups */
+	xfs_extlen_t	sb_rbmblocks;	/* number of rt bitmap blocks */
+	xfs_extlen_t	sb_logblocks;	/* number of log blocks */
+	__uint16_t	sb_versionnum;	/* header version == XFS_SB_VERSION */
+	__uint16_t	sb_sectsize;	/* volume sector size, bytes */
+	__uint16_t	sb_inodesize;	/* inode size, bytes */
+	__uint16_t	sb_inopblock;	/* inodes per block */
+	char		sb_fname[12];	/* file system name */
+	__uint8_t	sb_blocklog;	/* log2 of sb_blocksize */
+	__uint8_t	sb_sectlog;	/* log2 of sb_sectsize */
+	__uint8_t	sb_inodelog;	/* log2 of sb_inodesize */
+	__uint8_t	sb_inopblog;	/* log2 of sb_inopblock */
+	__uint8_t	sb_agblklog;	/* log2 of sb_agblocks (rounded up) */
+	__uint8_t	sb_rextslog;	/* log2 of sb_rextents */
+	__uint8_t	sb_inprogress;	/* mkfs is in progress, don't mount */
+	__uint8_t	sb_imax_pct;	/* max % of fs for inode space */
+					/* statistics */
+	/*
+	 * These fields must remain contiguous.  If you really
+	 * want to change their layout, make sure you fix the
+	 * code in xfs_trans_apply_sb_deltas().
+	 */
+	__uint64_t	sb_icount;	/* allocated inodes */
+	__uint64_t	sb_ifree;	/* free inodes */
+	__uint64_t	sb_fdblocks;	/* free data blocks */
+	__uint64_t	sb_frextents;	/* free realtime extents */
+	/*
+	 * End contiguous fields.
+	 */
+	xfs_ino_t	sb_uquotino;	/* user quota inode */
+	xfs_ino_t	sb_gquotino;	/* group quota inode */
+	__uint16_t	sb_qflags;	/* quota flags */
+	__uint8_t	sb_flags;	/* misc. flags */
+	__uint8_t	sb_shared_vn;	/* shared version number */
+	xfs_extlen_t	sb_inoalignmt;	/* inode chunk alignment, fsblocks */
+	__uint32_t	sb_unit;	/* stripe or raid unit */
+	__uint32_t	sb_width;	/* stripe or raid width */
+	__uint8_t	sb_dirblklog;	/* log2 of dir block size (fsbs) */
+	__uint8_t	sb_logsectlog;	/* log2 of the log sector size */
+	__uint16_t	sb_logsectsize;	/* sector size for the log, bytes */
+	__uint32_t	sb_logsunit;	/* stripe unit size for the log */
+	__uint32_t	sb_features2;	/* additional feature bits */
+
+	/*
+	 * bad features2 field as a result of failing to pad the sb
+	 * structure to 64 bits. Some machines will be using this field
+	 * for features2 bits. Easiest just to mark it bad and not use
+	 * it for anything else.
+	 */
+	__uint32_t	sb_bad_features2;
+
+	/* must be padded to 64 bit alignment */
+} xfs_sb_t;
+
+/*
+ * Superblock - on disk version.  Must match the in core version above.
+ * Must be padded to 64 bit alignment.
+ */
+typedef struct xfs_dsb {
+	__be32		sb_magicnum;	/* magic number == XFS_SB_MAGIC */
+	__be32		sb_blocksize;	/* logical block size, bytes */
+	__be64		sb_dblocks;	/* number of data blocks */
+	__be64		sb_rblocks;	/* number of realtime blocks */
+	__be64		sb_rextents;	/* number of realtime extents */
+	uuid_t		sb_uuid;	/* file system unique id */
+	__be64		sb_logstart;	/* starting block of log if internal */
+	__be64		sb_rootino;	/* root inode number */
+	__be64		sb_rbmino;	/* bitmap inode for realtime extents */
+	__be64		sb_rsumino;	/* summary inode for rt bitmap */
+	__be32		sb_rextsize;	/* realtime extent size, blocks */
+	__be32		sb_agblocks;	/* size of an allocation group */
+	__be32		sb_agcount;	/* number of allocation groups */
+	__be32		sb_rbmblocks;	/* number of rt bitmap blocks */
+	__be32		sb_logblocks;	/* number of log blocks */
+	__be16		sb_versionnum;	/* header version == XFS_SB_VERSION */
+	__be16		sb_sectsize;	/* volume sector size, bytes */
+	__be16		sb_inodesize;	/* inode size, bytes */
+	__be16		sb_inopblock;	/* inodes per block */
+	char		sb_fname[12];	/* file system name */
+	__u8		sb_blocklog;	/* log2 of sb_blocksize */
+	__u8		sb_sectlog;	/* log2 of sb_sectsize */
+	__u8		sb_inodelog;	/* log2 of sb_inodesize */
+	__u8		sb_inopblog;	/* log2 of sb_inopblock */
+	__u8		sb_agblklog;	/* log2 of sb_agblocks (rounded up) */
+	__u8		sb_rextslog;	/* log2 of sb_rextents */
+	__u8		sb_inprogress;	/* mkfs is in progress, don't mount */
+	__u8		sb_imax_pct;	/* max % of fs for inode space */
+					/* statistics */
+	/*
+	 * These fields must remain contiguous.  If you really
+	 * want to change their layout, make sure you fix the
+	 * code in xfs_trans_apply_sb_deltas().
+	 */
+	__be64		sb_icount;	/* allocated inodes */
+	__be64		sb_ifree;	/* free inodes */
+	__be64		sb_fdblocks;	/* free data blocks */
+	__be64		sb_frextents;	/* free realtime extents */
+	/*
+	 * End contiguous fields.
+	 */
+	__be64		sb_uquotino;	/* user quota inode */
+	__be64		sb_gquotino;	/* group quota inode */
+	__be16		sb_qflags;	/* quota flags */
+	__u8		sb_flags;	/* misc. flags */
+	__u8		sb_shared_vn;	/* shared version number */
+	__be32		sb_inoalignmt;	/* inode chunk alignment, fsblocks */
+	__be32		sb_unit;	/* stripe or raid unit */
+	__be32		sb_width;	/* stripe or raid width */
+	__u8		sb_dirblklog;	/* log2 of dir block size (fsbs) */
+	__u8		sb_logsectlog;	/* log2 of the log sector size */
+	__be16		sb_logsectsize;	/* sector size for the log, bytes */
+	__be32		sb_logsunit;	/* stripe unit size for the log */
+	__be32		sb_features2;	/* additional feature bits */
+	/*
+	 * bad features2 field as a result of failing to pad the sb
+	 * structure to 64 bits. Some machines will be using this field
+	 * for features2 bits. Easiest just to mark it bad and not use
+	 * it for anything else.
+	 */
+	__be32	sb_bad_features2;
+
+	/* must be padded to 64 bit alignment */
+} xfs_dsb_t;
+
+/*
+ * Sequence number values for the fields.
+ */
+typedef enum {
+	XFS_SBS_MAGICNUM, XFS_SBS_BLOCKSIZE, XFS_SBS_DBLOCKS, XFS_SBS_RBLOCKS,
+	XFS_SBS_REXTENTS, XFS_SBS_UUID, XFS_SBS_LOGSTART, XFS_SBS_ROOTINO,
+	XFS_SBS_RBMINO, XFS_SBS_RSUMINO, XFS_SBS_REXTSIZE, XFS_SBS_AGBLOCKS,
+	XFS_SBS_AGCOUNT, XFS_SBS_RBMBLOCKS, XFS_SBS_LOGBLOCKS,
+	XFS_SBS_VERSIONNUM, XFS_SBS_SECTSIZE, XFS_SBS_INODESIZE,
+	XFS_SBS_INOPBLOCK, XFS_SBS_FNAME, XFS_SBS_BLOCKLOG,
+	XFS_SBS_SECTLOG, XFS_SBS_INODELOG, XFS_SBS_INOPBLOG, XFS_SBS_AGBLKLOG,
+	XFS_SBS_REXTSLOG, XFS_SBS_INPROGRESS, XFS_SBS_IMAX_PCT, XFS_SBS_ICOUNT,
+	XFS_SBS_IFREE, XFS_SBS_FDBLOCKS, XFS_SBS_FREXTENTS, XFS_SBS_UQUOTINO,
+	XFS_SBS_GQUOTINO, XFS_SBS_QFLAGS, XFS_SBS_FLAGS, XFS_SBS_SHARED_VN,
+	XFS_SBS_INOALIGNMT, XFS_SBS_UNIT, XFS_SBS_WIDTH, XFS_SBS_DIRBLKLOG,
+	XFS_SBS_LOGSECTLOG, XFS_SBS_LOGSECTSIZE, XFS_SBS_LOGSUNIT,
+	XFS_SBS_FEATURES2, XFS_SBS_BAD_FEATURES2,
+	XFS_SBS_FIELDCOUNT
+} xfs_sb_field_t;
+
+/*
+ * Mask values, defined based on the xfs_sb_field_t values.
+ * Only define the ones we're using.
+ */
+#define	XFS_SB_MVAL(x)		(1LL << XFS_SBS_ ## x)
+#define	XFS_SB_UUID		XFS_SB_MVAL(UUID)
+#define	XFS_SB_FNAME		XFS_SB_MVAL(FNAME)
+#define	XFS_SB_ROOTINO		XFS_SB_MVAL(ROOTINO)
+#define	XFS_SB_RBMINO		XFS_SB_MVAL(RBMINO)
+#define	XFS_SB_RSUMINO		XFS_SB_MVAL(RSUMINO)
+#define	XFS_SB_VERSIONNUM	XFS_SB_MVAL(VERSIONNUM)
+#define XFS_SB_UQUOTINO		XFS_SB_MVAL(UQUOTINO)
+#define XFS_SB_GQUOTINO		XFS_SB_MVAL(GQUOTINO)
+#define XFS_SB_QFLAGS		XFS_SB_MVAL(QFLAGS)
+#define XFS_SB_SHARED_VN	XFS_SB_MVAL(SHARED_VN)
+#define XFS_SB_UNIT		XFS_SB_MVAL(UNIT)
+#define XFS_SB_WIDTH		XFS_SB_MVAL(WIDTH)
+#define XFS_SB_ICOUNT		XFS_SB_MVAL(ICOUNT)
+#define XFS_SB_IFREE		XFS_SB_MVAL(IFREE)
+#define XFS_SB_FDBLOCKS		XFS_SB_MVAL(FDBLOCKS)
+#define XFS_SB_FEATURES2	XFS_SB_MVAL(FEATURES2)
+#define XFS_SB_BAD_FEATURES2	XFS_SB_MVAL(BAD_FEATURES2)
+#define	XFS_SB_NUM_BITS		((int)XFS_SBS_FIELDCOUNT)
+#define	XFS_SB_ALL_BITS		((1LL << XFS_SB_NUM_BITS) - 1)
+#define	XFS_SB_MOD_BITS		\
+	(XFS_SB_UUID | XFS_SB_ROOTINO | XFS_SB_RBMINO | XFS_SB_RSUMINO | \
+	 XFS_SB_VERSIONNUM | XFS_SB_UQUOTINO | XFS_SB_GQUOTINO | \
+	 XFS_SB_QFLAGS | XFS_SB_SHARED_VN | XFS_SB_UNIT | XFS_SB_WIDTH | \
+	 XFS_SB_ICOUNT | XFS_SB_IFREE | XFS_SB_FDBLOCKS | XFS_SB_FEATURES2 | \
+	 XFS_SB_BAD_FEATURES2)
+
+
+/*
+ * Misc. Flags - warning - these will be cleared by xfs_repair unless
+ * a feature bit is set when the flag is used.
+ */
+#define XFS_SBF_NOFLAGS		0x00	/* no flags set */
+#define XFS_SBF_READONLY	0x01	/* only read-only mounts allowed */
+
+/*
+ * define max. shared version we can interoperate with
+ */
+#define XFS_SB_MAX_SHARED_VN	0
+
+#define	XFS_SB_VERSION_NUM(sbp)	((sbp)->sb_versionnum & XFS_SB_VERSION_NUMBITS)
+
+static inline int xfs_sb_good_version(xfs_sb_t *sbp)
+{
+	/* We always support version 1-3 */
+	if (sbp->sb_versionnum >= XFS_SB_VERSION_1 &&
+	    sbp->sb_versionnum <= XFS_SB_VERSION_3)
+		return 1;
+
+	/* We support version 4 if all feature bits are supported */
+	if (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4) {
+		if ((sbp->sb_versionnum & ~XFS_SB_VERSION_OKREALBITS) ||
+		    ((sbp->sb_versionnum & XFS_SB_VERSION_MOREBITSBIT) &&
+		     (sbp->sb_features2 & ~XFS_SB_VERSION2_OKREALBITS)))
+			return 0;
+
+#ifdef __KERNEL__
+		if (sbp->sb_shared_vn > XFS_SB_MAX_SHARED_VN)
+			return 0;
+#else
+		if ((sbp->sb_versionnum & XFS_SB_VERSION_SHAREDBIT) &&
+		    sbp->sb_shared_vn > XFS_SB_MAX_SHARED_VN)
+			return 0;
+#endif
+
+		return 1;
+	}
+
+	return 0;
+}
+
+/*
+ * Detect a mismatched features2 field.  Older kernels read/wrote
+ * this into the wrong slot, so to be safe we keep them in sync.
+ */
+static inline int xfs_sb_has_mismatched_features2(xfs_sb_t *sbp)
+{
+	return (sbp->sb_bad_features2 != sbp->sb_features2);
+}
+
+static inline unsigned xfs_sb_version_tonew(unsigned v)
+{
+	if (v == XFS_SB_VERSION_1)
+		return XFS_SB_VERSION_4;
+
+	if (v == XFS_SB_VERSION_2)
+		return XFS_SB_VERSION_4 | XFS_SB_VERSION_ATTRBIT;
+
+	return XFS_SB_VERSION_4 | XFS_SB_VERSION_ATTRBIT |
+		XFS_SB_VERSION_NLINKBIT;
+}
+
+static inline unsigned xfs_sb_version_toold(unsigned v)
+{
+	if (v & (XFS_SB_VERSION_QUOTABIT | XFS_SB_VERSION_ALIGNBIT))
+		return 0;
+	if (v & XFS_SB_VERSION_NLINKBIT)
+		return XFS_SB_VERSION_3;
+	if (v & XFS_SB_VERSION_ATTRBIT)
+		return XFS_SB_VERSION_2;
+	return XFS_SB_VERSION_1;
+}
+
+static inline int xfs_sb_version_hasattr(xfs_sb_t *sbp)
+{
+	return sbp->sb_versionnum == XFS_SB_VERSION_2 ||
+		sbp->sb_versionnum == XFS_SB_VERSION_3 ||
+		(XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
+		 (sbp->sb_versionnum & XFS_SB_VERSION_ATTRBIT));
+}
+
+static inline void xfs_sb_version_addattr(xfs_sb_t *sbp)
+{
+	if (sbp->sb_versionnum == XFS_SB_VERSION_1)
+		sbp->sb_versionnum = XFS_SB_VERSION_2;
+	else if (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4)
+		sbp->sb_versionnum |= XFS_SB_VERSION_ATTRBIT;
+	else
+		sbp->sb_versionnum = XFS_SB_VERSION_4 | XFS_SB_VERSION_ATTRBIT;
+}
+
+static inline int xfs_sb_version_hasnlink(xfs_sb_t *sbp)
+{
+	return sbp->sb_versionnum == XFS_SB_VERSION_3 ||
+		 (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
+		  (sbp->sb_versionnum & XFS_SB_VERSION_NLINKBIT));
+}
+
+static inline void xfs_sb_version_addnlink(xfs_sb_t *sbp)
+{
+	if (sbp->sb_versionnum <= XFS_SB_VERSION_2)
+		sbp->sb_versionnum = XFS_SB_VERSION_3;
+	else
+		sbp->sb_versionnum |= XFS_SB_VERSION_NLINKBIT;
+}
+
+static inline int xfs_sb_version_hasquota(xfs_sb_t *sbp)
+{
+	return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
+		(sbp->sb_versionnum & XFS_SB_VERSION_QUOTABIT);
+}
+
+static inline void xfs_sb_version_addquota(xfs_sb_t *sbp)
+{
+	if (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4)
+		sbp->sb_versionnum |= XFS_SB_VERSION_QUOTABIT;
+	else
+		sbp->sb_versionnum = xfs_sb_version_tonew(sbp->sb_versionnum) |
+					XFS_SB_VERSION_QUOTABIT;
+}
+
+static inline int xfs_sb_version_hasalign(xfs_sb_t *sbp)
+{
+	return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
+		(sbp->sb_versionnum & XFS_SB_VERSION_ALIGNBIT);
+}
+
+static inline int xfs_sb_version_hasdalign(xfs_sb_t *sbp)
+{
+	return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
+		(sbp->sb_versionnum & XFS_SB_VERSION_DALIGNBIT);
+}
+
+static inline int xfs_sb_version_hasshared(xfs_sb_t *sbp)
+{
+	return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
+		(sbp->sb_versionnum & XFS_SB_VERSION_SHAREDBIT);
+}
+
+static inline int xfs_sb_version_hasdirv2(xfs_sb_t *sbp)
+{
+	return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
+		(sbp->sb_versionnum & XFS_SB_VERSION_DIRV2BIT);
+}
+
+static inline int xfs_sb_version_haslogv2(xfs_sb_t *sbp)
+{
+	return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
+		(sbp->sb_versionnum & XFS_SB_VERSION_LOGV2BIT);
+}
+
+static inline int xfs_sb_version_hasextflgbit(xfs_sb_t *sbp)
+{
+	return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
+		(sbp->sb_versionnum & XFS_SB_VERSION_EXTFLGBIT);
+}
+
+static inline int xfs_sb_version_hassector(xfs_sb_t *sbp)
+{
+	return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
+		(sbp->sb_versionnum & XFS_SB_VERSION_SECTORBIT);
+}
+
+static inline int xfs_sb_version_hasasciici(xfs_sb_t *sbp)
+{
+	return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
+		(sbp->sb_versionnum & XFS_SB_VERSION_BORGBIT);
+}
+
+static inline int xfs_sb_version_hasmorebits(xfs_sb_t *sbp)
+{
+	return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
+		(sbp->sb_versionnum & XFS_SB_VERSION_MOREBITSBIT);
+}
+
+/*
+ * sb_features2 bit version macros.
+ *
+ * For example, for a bit defined as XFS_SB_VERSION2_FUNBIT, has a macro:
+ *
+ * SB_VERSION_HASFUNBIT(xfs_sb_t *sbp)
+ *	((xfs_sb_version_hasmorebits(sbp) &&
+ *	 ((sbp)->sb_features2 & XFS_SB_VERSION2_FUNBIT)
+ */
+
+static inline int xfs_sb_version_haslazysbcount(xfs_sb_t *sbp)
+{
+	return xfs_sb_version_hasmorebits(sbp) &&
+		(sbp->sb_features2 & XFS_SB_VERSION2_LAZYSBCOUNTBIT);
+}
+
+static inline int xfs_sb_version_hasattr2(xfs_sb_t *sbp)
+{
+	return xfs_sb_version_hasmorebits(sbp) &&
+		(sbp->sb_features2 & XFS_SB_VERSION2_ATTR2BIT);
+}
+
+static inline void xfs_sb_version_addattr2(xfs_sb_t *sbp)
+{
+	sbp->sb_versionnum |= XFS_SB_VERSION_MOREBITSBIT;
+	sbp->sb_features2 |= XFS_SB_VERSION2_ATTR2BIT;
+}
+
+static inline void xfs_sb_version_removeattr2(xfs_sb_t *sbp)
+{
+	sbp->sb_features2 &= ~XFS_SB_VERSION2_ATTR2BIT;
+	if (!sbp->sb_features2)
+		sbp->sb_versionnum &= ~XFS_SB_VERSION_MOREBITSBIT;
+}
+
+static inline int xfs_sb_version_hasprojid32bit(xfs_sb_t *sbp)
+{
+	return xfs_sb_version_hasmorebits(sbp) &&
+		(sbp->sb_features2 & XFS_SB_VERSION2_PROJID32BIT);
+}
+
+/*
+ * end of superblock version macros
+ */
+
+#define XFS_SB_DADDR		((xfs_daddr_t)0) /* daddr in filesystem/ag */
+#define	XFS_SB_BLOCK(mp)	XFS_HDR_BLOCK(mp, XFS_SB_DADDR)
+#define XFS_BUF_TO_SBP(bp)	((xfs_dsb_t *)XFS_BUF_PTR(bp))
+
+#define	XFS_HDR_BLOCK(mp,d)	((xfs_agblock_t)XFS_BB_TO_FSBT(mp,d))
+#define	XFS_DADDR_TO_FSB(mp,d)	XFS_AGB_TO_FSB(mp, \
+			xfs_daddr_to_agno(mp,d), xfs_daddr_to_agbno(mp,d))
+#define	XFS_FSB_TO_DADDR(mp,fsbno)	XFS_AGB_TO_DADDR(mp, \
+			XFS_FSB_TO_AGNO(mp,fsbno), XFS_FSB_TO_AGBNO(mp,fsbno))
+
+/*
+ * File system sector to basic block conversions.
+ */
+#define XFS_FSS_TO_BB(mp,sec)	((sec) << (mp)->m_sectbb_log)
+
+/*
+ * File system block to basic block conversions.
+ */
+#define	XFS_FSB_TO_BB(mp,fsbno)	((fsbno) << (mp)->m_blkbb_log)
+#define	XFS_BB_TO_FSB(mp,bb)	\
+	(((bb) + (XFS_FSB_TO_BB(mp,1) - 1)) >> (mp)->m_blkbb_log)
+#define	XFS_BB_TO_FSBT(mp,bb)	((bb) >> (mp)->m_blkbb_log)
+#define	XFS_BB_FSB_OFFSET(mp,bb) ((bb) & ((mp)->m_bsize - 1))
+
+/*
+ * File system block to byte conversions.
+ */
+#define XFS_FSB_TO_B(mp,fsbno)	((xfs_fsize_t)(fsbno) << (mp)->m_sb.sb_blocklog)
+#define XFS_B_TO_FSB(mp,b)	\
+	((((__uint64_t)(b)) + (mp)->m_blockmask) >> (mp)->m_sb.sb_blocklog)
+#define XFS_B_TO_FSBT(mp,b)	(((__uint64_t)(b)) >> (mp)->m_sb.sb_blocklog)
+#define XFS_B_FSB_OFFSET(mp,b)	((b) & (mp)->m_blockmask)
+
+#endif	/* __XFS_SB_H__ */
diff --git a/fs/xfs/xfs_trans.c b/fs/xfs/xfs_trans.c
new file mode 100644
index 00000000..efc147f0
--- /dev/null
+++ b/fs/xfs/xfs_trans.c
@@ -0,0 +1,2026 @@
+/*
+ * Copyright (c) 2000-2003,2005 Silicon Graphics, Inc.
+ * Copyright (C) 2010 Red Hat, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_mount.h"
+#include "xfs_error.h"
+#include "xfs_da_btree.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_alloc_btree.h"
+#include "xfs_ialloc_btree.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_btree.h"
+#include "xfs_ialloc.h"
+#include "xfs_alloc.h"
+#include "xfs_bmap.h"
+#include "xfs_quota.h"
+#include "xfs_trans_priv.h"
+#include "xfs_trans_space.h"
+#include "xfs_inode_item.h"
+#include "xfs_trace.h"
+
+kmem_zone_t	*xfs_trans_zone;
+kmem_zone_t	*xfs_log_item_desc_zone;
+
+
+/*
+ * Various log reservation values.
+ *
+ * These are based on the size of the file system block because that is what
+ * most transactions manipulate.  Each adds in an additional 128 bytes per
+ * item logged to try to account for the overhead of the transaction mechanism.
+ *
+ * Note:  Most of the reservations underestimate the number of allocation
+ * groups into which they could free extents in the xfs_bmap_finish() call.
+ * This is because the number in the worst case is quite high and quite
+ * unusual.  In order to fix this we need to change xfs_bmap_finish() to free
+ * extents in only a single AG at a time.  This will require changes to the
+ * EFI code as well, however, so that the EFI for the extents not freed is
+ * logged again in each transaction.  See SGI PV #261917.
+ *
+ * Reservation functions here avoid a huge stack in xfs_trans_init due to
+ * register overflow from temporaries in the calculations.
+ */
+
+
+/*
+ * In a write transaction we can allocate a maximum of 2
+ * extents.  This gives:
+ *    the inode getting the new extents: inode size
+ *    the inode's bmap btree: max depth * block size
+ *    the agfs of the ags from which the extents are allocated: 2 * sector
+ *    the superblock free block counter: sector size
+ *    the allocation btrees: 2 exts * 2 trees * (2 * max depth - 1) * block size
+ * And the bmap_finish transaction can free bmap blocks in a join:
+ *    the agfs of the ags containing the blocks: 2 * sector size
+ *    the agfls of the ags containing the blocks: 2 * sector size
+ *    the super block free block counter: sector size
+ *    the allocation btrees: 2 exts * 2 trees * (2 * max depth - 1) * block size
+ */
+STATIC uint
+xfs_calc_write_reservation(
+	struct xfs_mount	*mp)
+{
+	return XFS_DQUOT_LOGRES(mp) +
+		MAX((mp->m_sb.sb_inodesize +
+		     XFS_FSB_TO_B(mp, XFS_BM_MAXLEVELS(mp, XFS_DATA_FORK)) +
+		     2 * mp->m_sb.sb_sectsize +
+		     mp->m_sb.sb_sectsize +
+		     XFS_ALLOCFREE_LOG_RES(mp, 2) +
+		     128 * (4 + XFS_BM_MAXLEVELS(mp, XFS_DATA_FORK) +
+			    XFS_ALLOCFREE_LOG_COUNT(mp, 2))),
+		    (2 * mp->m_sb.sb_sectsize +
+		     2 * mp->m_sb.sb_sectsize +
+		     mp->m_sb.sb_sectsize +
+		     XFS_ALLOCFREE_LOG_RES(mp, 2) +
+		     128 * (5 + XFS_ALLOCFREE_LOG_COUNT(mp, 2))));
+}
+
+/*
+ * In truncating a file we free up to two extents at once.  We can modify:
+ *    the inode being truncated: inode size
+ *    the inode's bmap btree: (max depth + 1) * block size
+ * And the bmap_finish transaction can free the blocks and bmap blocks:
+ *    the agf for each of the ags: 4 * sector size
+ *    the agfl for each of the ags: 4 * sector size
+ *    the super block to reflect the freed blocks: sector size
+ *    worst case split in allocation btrees per extent assuming 4 extents:
+ *		4 exts * 2 trees * (2 * max depth - 1) * block size
+ *    the inode btree: max depth * blocksize
+ *    the allocation btrees: 2 trees * (max depth - 1) * block size
+ */
+STATIC uint
+xfs_calc_itruncate_reservation(
+	struct xfs_mount	*mp)
+{
+	return XFS_DQUOT_LOGRES(mp) +
+		MAX((mp->m_sb.sb_inodesize +
+		     XFS_FSB_TO_B(mp, XFS_BM_MAXLEVELS(mp, XFS_DATA_FORK) + 1) +
+		     128 * (2 + XFS_BM_MAXLEVELS(mp, XFS_DATA_FORK))),
+		    (4 * mp->m_sb.sb_sectsize +
+		     4 * mp->m_sb.sb_sectsize +
+		     mp->m_sb.sb_sectsize +
+		     XFS_ALLOCFREE_LOG_RES(mp, 4) +
+		     128 * (9 + XFS_ALLOCFREE_LOG_COUNT(mp, 4)) +
+		     128 * 5 +
+		     XFS_ALLOCFREE_LOG_RES(mp, 1) +
+		     128 * (2 + XFS_IALLOC_BLOCKS(mp) + mp->m_in_maxlevels +
+			    XFS_ALLOCFREE_LOG_COUNT(mp, 1))));
+}
+
+/*
+ * In renaming a files we can modify:
+ *    the four inodes involved: 4 * inode size
+ *    the two directory btrees: 2 * (max depth + v2) * dir block size
+ *    the two directory bmap btrees: 2 * max depth * block size
+ * And the bmap_finish transaction can free dir and bmap blocks (two sets
+ *	of bmap blocks) giving:
+ *    the agf for the ags in which the blocks live: 3 * sector size
+ *    the agfl for the ags in which the blocks live: 3 * sector size
+ *    the superblock for the free block count: sector size
+ *    the allocation btrees: 3 exts * 2 trees * (2 * max depth - 1) * block size
+ */
+STATIC uint
+xfs_calc_rename_reservation(
+	struct xfs_mount	*mp)
+{
+	return XFS_DQUOT_LOGRES(mp) +
+		MAX((4 * mp->m_sb.sb_inodesize +
+		     2 * XFS_DIROP_LOG_RES(mp) +
+		     128 * (4 + 2 * XFS_DIROP_LOG_COUNT(mp))),
+		    (3 * mp->m_sb.sb_sectsize +
+		     3 * mp->m_sb.sb_sectsize +
+		     mp->m_sb.sb_sectsize +
+		     XFS_ALLOCFREE_LOG_RES(mp, 3) +
+		     128 * (7 + XFS_ALLOCFREE_LOG_COUNT(mp, 3))));
+}
+
+/*
+ * For creating a link to an inode:
+ *    the parent directory inode: inode size
+ *    the linked inode: inode size
+ *    the directory btree could split: (max depth + v2) * dir block size
+ *    the directory bmap btree could join or split: (max depth + v2) * blocksize
+ * And the bmap_finish transaction can free some bmap blocks giving:
+ *    the agf for the ag in which the blocks live: sector size
+ *    the agfl for the ag in which the blocks live: sector size
+ *    the superblock for the free block count: sector size
+ *    the allocation btrees: 2 trees * (2 * max depth - 1) * block size
+ */
+STATIC uint
+xfs_calc_link_reservation(
+	struct xfs_mount	*mp)
+{
+	return XFS_DQUOT_LOGRES(mp) +
+		MAX((mp->m_sb.sb_inodesize +
+		     mp->m_sb.sb_inodesize +
+		     XFS_DIROP_LOG_RES(mp) +
+		     128 * (2 + XFS_DIROP_LOG_COUNT(mp))),
+		    (mp->m_sb.sb_sectsize +
+		     mp->m_sb.sb_sectsize +
+		     mp->m_sb.sb_sectsize +
+		     XFS_ALLOCFREE_LOG_RES(mp, 1) +
+		     128 * (3 + XFS_ALLOCFREE_LOG_COUNT(mp, 1))));
+}
+
+/*
+ * For removing a directory entry we can modify:
+ *    the parent directory inode: inode size
+ *    the removed inode: inode size
+ *    the directory btree could join: (max depth + v2) * dir block size
+ *    the directory bmap btree could join or split: (max depth + v2) * blocksize
+ * And the bmap_finish transaction can free the dir and bmap blocks giving:
+ *    the agf for the ag in which the blocks live: 2 * sector size
+ *    the agfl for the ag in which the blocks live: 2 * sector size
+ *    the superblock for the free block count: sector size
+ *    the allocation btrees: 2 exts * 2 trees * (2 * max depth - 1) * block size
+ */
+STATIC uint
+xfs_calc_remove_reservation(
+	struct xfs_mount	*mp)
+{
+	return XFS_DQUOT_LOGRES(mp) +
+		MAX((mp->m_sb.sb_inodesize +
+		     mp->m_sb.sb_inodesize +
+		     XFS_DIROP_LOG_RES(mp) +
+		     128 * (2 + XFS_DIROP_LOG_COUNT(mp))),
+		    (2 * mp->m_sb.sb_sectsize +
+		     2 * mp->m_sb.sb_sectsize +
+		     mp->m_sb.sb_sectsize +
+		     XFS_ALLOCFREE_LOG_RES(mp, 2) +
+		     128 * (5 + XFS_ALLOCFREE_LOG_COUNT(mp, 2))));
+}
+
+/*
+ * For symlink we can modify:
+ *    the parent directory inode: inode size
+ *    the new inode: inode size
+ *    the inode btree entry: 1 block
+ *    the directory btree: (max depth + v2) * dir block size
+ *    the directory inode's bmap btree: (max depth + v2) * block size
+ *    the blocks for the symlink: 1 kB
+ * Or in the first xact we allocate some inodes giving:
+ *    the agi and agf of the ag getting the new inodes: 2 * sectorsize
+ *    the inode blocks allocated: XFS_IALLOC_BLOCKS * blocksize
+ *    the inode btree: max depth * blocksize
+ *    the allocation btrees: 2 trees * (2 * max depth - 1) * block size
+ */
+STATIC uint
+xfs_calc_symlink_reservation(
+	struct xfs_mount	*mp)
+{
+	return XFS_DQUOT_LOGRES(mp) +
+		MAX((mp->m_sb.sb_inodesize +
+		     mp->m_sb.sb_inodesize +
+		     XFS_FSB_TO_B(mp, 1) +
+		     XFS_DIROP_LOG_RES(mp) +
+		     1024 +
+		     128 * (4 + XFS_DIROP_LOG_COUNT(mp))),
+		    (2 * mp->m_sb.sb_sectsize +
+		     XFS_FSB_TO_B(mp, XFS_IALLOC_BLOCKS(mp)) +
+		     XFS_FSB_TO_B(mp, mp->m_in_maxlevels) +
+		     XFS_ALLOCFREE_LOG_RES(mp, 1) +
+		     128 * (2 + XFS_IALLOC_BLOCKS(mp) + mp->m_in_maxlevels +
+			    XFS_ALLOCFREE_LOG_COUNT(mp, 1))));
+}
+
+/*
+ * For create we can modify:
+ *    the parent directory inode: inode size
+ *    the new inode: inode size
+ *    the inode btree entry: block size
+ *    the superblock for the nlink flag: sector size
+ *    the directory btree: (max depth + v2) * dir block size
+ *    the directory inode's bmap btree: (max depth + v2) * block size
+ * Or in the first xact we allocate some inodes giving:
+ *    the agi and agf of the ag getting the new inodes: 2 * sectorsize
+ *    the superblock for the nlink flag: sector size
+ *    the inode blocks allocated: XFS_IALLOC_BLOCKS * blocksize
+ *    the inode btree: max depth * blocksize
+ *    the allocation btrees: 2 trees * (max depth - 1) * block size
+ */
+STATIC uint
+xfs_calc_create_reservation(
+	struct xfs_mount	*mp)
+{
+	return XFS_DQUOT_LOGRES(mp) +
+		MAX((mp->m_sb.sb_inodesize +
+		     mp->m_sb.sb_inodesize +
+		     mp->m_sb.sb_sectsize +
+		     XFS_FSB_TO_B(mp, 1) +
+		     XFS_DIROP_LOG_RES(mp) +
+		     128 * (3 + XFS_DIROP_LOG_COUNT(mp))),
+		    (3 * mp->m_sb.sb_sectsize +
+		     XFS_FSB_TO_B(mp, XFS_IALLOC_BLOCKS(mp)) +
+		     XFS_FSB_TO_B(mp, mp->m_in_maxlevels) +
+		     XFS_ALLOCFREE_LOG_RES(mp, 1) +
+		     128 * (2 + XFS_IALLOC_BLOCKS(mp) + mp->m_in_maxlevels +
+			    XFS_ALLOCFREE_LOG_COUNT(mp, 1))));
+}
+
+/*
+ * Making a new directory is the same as creating a new file.
+ */
+STATIC uint
+xfs_calc_mkdir_reservation(
+	struct xfs_mount	*mp)
+{
+	return xfs_calc_create_reservation(mp);
+}
+
+/*
+ * In freeing an inode we can modify:
+ *    the inode being freed: inode size
+ *    the super block free inode counter: sector size
+ *    the agi hash list and counters: sector size
+ *    the inode btree entry: block size
+ *    the on disk inode before ours in the agi hash list: inode cluster size
+ *    the inode btree: max depth * blocksize
+ *    the allocation btrees: 2 trees * (max depth - 1) * block size
+ */
+STATIC uint
+xfs_calc_ifree_reservation(
+	struct xfs_mount	*mp)
+{
+	return XFS_DQUOT_LOGRES(mp) +
+		mp->m_sb.sb_inodesize +
+		mp->m_sb.sb_sectsize +
+		mp->m_sb.sb_sectsize +
+		XFS_FSB_TO_B(mp, 1) +
+		MAX((__uint16_t)XFS_FSB_TO_B(mp, 1),
+		    XFS_INODE_CLUSTER_SIZE(mp)) +
+		128 * 5 +
+		XFS_ALLOCFREE_LOG_RES(mp, 1) +
+		128 * (2 + XFS_IALLOC_BLOCKS(mp) + mp->m_in_maxlevels +
+		       XFS_ALLOCFREE_LOG_COUNT(mp, 1));
+}
+
+/*
+ * When only changing the inode we log the inode and possibly the superblock
+ * We also add a bit of slop for the transaction stuff.
+ */
+STATIC uint
+xfs_calc_ichange_reservation(
+	struct xfs_mount	*mp)
+{
+	return XFS_DQUOT_LOGRES(mp) +
+		mp->m_sb.sb_inodesize +
+		mp->m_sb.sb_sectsize +
+		512;
+
+}
+
+/*
+ * Growing the data section of the filesystem.
+ *	superblock
+ *	agi and agf
+ *	allocation btrees
+ */
+STATIC uint
+xfs_calc_growdata_reservation(
+	struct xfs_mount	*mp)
+{
+	return mp->m_sb.sb_sectsize * 3 +
+		XFS_ALLOCFREE_LOG_RES(mp, 1) +
+		128 * (3 + XFS_ALLOCFREE_LOG_COUNT(mp, 1));
+}
+
+/*
+ * Growing the rt section of the filesystem.
+ * In the first set of transactions (ALLOC) we allocate space to the
+ * bitmap or summary files.
+ *	superblock: sector size
+ *	agf of the ag from which the extent is allocated: sector size
+ *	bmap btree for bitmap/summary inode: max depth * blocksize
+ *	bitmap/summary inode: inode size
+ *	allocation btrees for 1 block alloc: 2 * (2 * maxdepth - 1) * blocksize
+ */
+STATIC uint
+xfs_calc_growrtalloc_reservation(
+	struct xfs_mount	*mp)
+{
+	return 2 * mp->m_sb.sb_sectsize +
+		XFS_FSB_TO_B(mp, XFS_BM_MAXLEVELS(mp, XFS_DATA_FORK)) +
+		mp->m_sb.sb_inodesize +
+		XFS_ALLOCFREE_LOG_RES(mp, 1) +
+		128 * (3 + XFS_BM_MAXLEVELS(mp, XFS_DATA_FORK) +
+		       XFS_ALLOCFREE_LOG_COUNT(mp, 1));
+}
+
+/*
+ * Growing the rt section of the filesystem.
+ * In the second set of transactions (ZERO) we zero the new metadata blocks.
+ *	one bitmap/summary block: blocksize
+ */
+STATIC uint
+xfs_calc_growrtzero_reservation(
+	struct xfs_mount	*mp)
+{
+	return mp->m_sb.sb_blocksize + 128;
+}
+
+/*
+ * Growing the rt section of the filesystem.
+ * In the third set of transactions (FREE) we update metadata without
+ * allocating any new blocks.
+ *	superblock: sector size
+ *	bitmap inode: inode size
+ *	summary inode: inode size
+ *	one bitmap block: blocksize
+ *	summary blocks: new summary size
+ */
+STATIC uint
+xfs_calc_growrtfree_reservation(
+	struct xfs_mount	*mp)
+{
+	return mp->m_sb.sb_sectsize +
+		2 * mp->m_sb.sb_inodesize +
+		mp->m_sb.sb_blocksize +
+		mp->m_rsumsize +
+		128 * 5;
+}
+
+/*
+ * Logging the inode modification timestamp on a synchronous write.
+ *	inode
+ */
+STATIC uint
+xfs_calc_swrite_reservation(
+	struct xfs_mount	*mp)
+{
+	return mp->m_sb.sb_inodesize + 128;
+}
+
+/*
+ * Logging the inode mode bits when writing a setuid/setgid file
+ *	inode
+ */
+STATIC uint
+xfs_calc_writeid_reservation(xfs_mount_t *mp)
+{
+	return mp->m_sb.sb_inodesize + 128;
+}
+
+/*
+ * Converting the inode from non-attributed to attributed.
+ *	the inode being converted: inode size
+ *	agf block and superblock (for block allocation)
+ *	the new block (directory sized)
+ *	bmap blocks for the new directory block
+ *	allocation btrees
+ */
+STATIC uint
+xfs_calc_addafork_reservation(
+	struct xfs_mount	*mp)
+{
+	return XFS_DQUOT_LOGRES(mp) +
+		mp->m_sb.sb_inodesize +
+		mp->m_sb.sb_sectsize * 2 +
+		mp->m_dirblksize +
+		XFS_FSB_TO_B(mp, XFS_DAENTER_BMAP1B(mp, XFS_DATA_FORK) + 1) +
+		XFS_ALLOCFREE_LOG_RES(mp, 1) +
+		128 * (4 + XFS_DAENTER_BMAP1B(mp, XFS_DATA_FORK) + 1 +
+		       XFS_ALLOCFREE_LOG_COUNT(mp, 1));
+}
+
+/*
+ * Removing the attribute fork of a file
+ *    the inode being truncated: inode size
+ *    the inode's bmap btree: max depth * block size
+ * And the bmap_finish transaction can free the blocks and bmap blocks:
+ *    the agf for each of the ags: 4 * sector size
+ *    the agfl for each of the ags: 4 * sector size
+ *    the super block to reflect the freed blocks: sector size
+ *    worst case split in allocation btrees per extent assuming 4 extents:
+ *		4 exts * 2 trees * (2 * max depth - 1) * block size
+ */
+STATIC uint
+xfs_calc_attrinval_reservation(
+	struct xfs_mount	*mp)
+{
+	return MAX((mp->m_sb.sb_inodesize +
+		    XFS_FSB_TO_B(mp, XFS_BM_MAXLEVELS(mp, XFS_ATTR_FORK)) +
+		    128 * (1 + XFS_BM_MAXLEVELS(mp, XFS_ATTR_FORK))),
+		   (4 * mp->m_sb.sb_sectsize +
+		    4 * mp->m_sb.sb_sectsize +
+		    mp->m_sb.sb_sectsize +
+		    XFS_ALLOCFREE_LOG_RES(mp, 4) +
+		    128 * (9 + XFS_ALLOCFREE_LOG_COUNT(mp, 4))));
+}
+
+/*
+ * Setting an attribute.
+ *	the inode getting the attribute
+ *	the superblock for allocations
+ *	the agfs extents are allocated from
+ *	the attribute btree * max depth
+ *	the inode allocation btree
+ * Since attribute transaction space is dependent on the size of the attribute,
+ * the calculation is done partially at mount time and partially at runtime.
+ */
+STATIC uint
+xfs_calc_attrset_reservation(
+	struct xfs_mount	*mp)
+{
+	return XFS_DQUOT_LOGRES(mp) +
+		mp->m_sb.sb_inodesize +
+		mp->m_sb.sb_sectsize +
+		XFS_FSB_TO_B(mp, XFS_DA_NODE_MAXDEPTH) +
+		128 * (2 + XFS_DA_NODE_MAXDEPTH);
+}
+
+/*
+ * Removing an attribute.
+ *    the inode: inode size
+ *    the attribute btree could join: max depth * block size
+ *    the inode bmap btree could join or split: max depth * block size
+ * And the bmap_finish transaction can free the attr blocks freed giving:
+ *    the agf for the ag in which the blocks live: 2 * sector size
+ *    the agfl for the ag in which the blocks live: 2 * sector size
+ *    the superblock for the free block count: sector size
+ *    the allocation btrees: 2 exts * 2 trees * (2 * max depth - 1) * block size
+ */
+STATIC uint
+xfs_calc_attrrm_reservation(
+	struct xfs_mount	*mp)
+{
+	return XFS_DQUOT_LOGRES(mp) +
+		MAX((mp->m_sb.sb_inodesize +
+		     XFS_FSB_TO_B(mp, XFS_DA_NODE_MAXDEPTH) +
+		     XFS_FSB_TO_B(mp, XFS_BM_MAXLEVELS(mp, XFS_ATTR_FORK)) +
+		     128 * (1 + XFS_DA_NODE_MAXDEPTH +
+			    XFS_BM_MAXLEVELS(mp, XFS_DATA_FORK))),
+		    (2 * mp->m_sb.sb_sectsize +
+		     2 * mp->m_sb.sb_sectsize +
+		     mp->m_sb.sb_sectsize +
+		     XFS_ALLOCFREE_LOG_RES(mp, 2) +
+		     128 * (5 + XFS_ALLOCFREE_LOG_COUNT(mp, 2))));
+}
+
+/*
+ * Clearing a bad agino number in an agi hash bucket.
+ */
+STATIC uint
+xfs_calc_clear_agi_bucket_reservation(
+	struct xfs_mount	*mp)
+{
+	return mp->m_sb.sb_sectsize + 128;
+}
+
+/*
+ * Initialize the precomputed transaction reservation values
+ * in the mount structure.
+ */
+void
+xfs_trans_init(
+	struct xfs_mount	*mp)
+{
+	struct xfs_trans_reservations *resp = &mp->m_reservations;
+
+	resp->tr_write = xfs_calc_write_reservation(mp);
+	resp->tr_itruncate = xfs_calc_itruncate_reservation(mp);
+	resp->tr_rename = xfs_calc_rename_reservation(mp);
+	resp->tr_link = xfs_calc_link_reservation(mp);
+	resp->tr_remove = xfs_calc_remove_reservation(mp);
+	resp->tr_symlink = xfs_calc_symlink_reservation(mp);
+	resp->tr_create = xfs_calc_create_reservation(mp);
+	resp->tr_mkdir = xfs_calc_mkdir_reservation(mp);
+	resp->tr_ifree = xfs_calc_ifree_reservation(mp);
+	resp->tr_ichange = xfs_calc_ichange_reservation(mp);
+	resp->tr_growdata = xfs_calc_growdata_reservation(mp);
+	resp->tr_swrite = xfs_calc_swrite_reservation(mp);
+	resp->tr_writeid = xfs_calc_writeid_reservation(mp);
+	resp->tr_addafork = xfs_calc_addafork_reservation(mp);
+	resp->tr_attrinval = xfs_calc_attrinval_reservation(mp);
+	resp->tr_attrset = xfs_calc_attrset_reservation(mp);
+	resp->tr_attrrm = xfs_calc_attrrm_reservation(mp);
+	resp->tr_clearagi = xfs_calc_clear_agi_bucket_reservation(mp);
+	resp->tr_growrtalloc = xfs_calc_growrtalloc_reservation(mp);
+	resp->tr_growrtzero = xfs_calc_growrtzero_reservation(mp);
+	resp->tr_growrtfree = xfs_calc_growrtfree_reservation(mp);
+}
+
+/*
+ * This routine is called to allocate a transaction structure.
+ * The type parameter indicates the type of the transaction.  These
+ * are enumerated in xfs_trans.h.
+ *
+ * Dynamically allocate the transaction structure from the transaction
+ * zone, initialize it, and return it to the caller.
+ */
+xfs_trans_t *
+xfs_trans_alloc(
+	xfs_mount_t	*mp,
+	uint		type)
+{
+	xfs_wait_for_freeze(mp, SB_FREEZE_TRANS);
+	return _xfs_trans_alloc(mp, type, KM_SLEEP);
+}
+
+xfs_trans_t *
+_xfs_trans_alloc(
+	xfs_mount_t	*mp,
+	uint		type,
+	uint		memflags)
+{
+	xfs_trans_t	*tp;
+
+	atomic_inc(&mp->m_active_trans);
+
+	tp = kmem_zone_zalloc(xfs_trans_zone, memflags);
+	tp->t_magic = XFS_TRANS_MAGIC;
+	tp->t_type = type;
+	tp->t_mountp = mp;
+	INIT_LIST_HEAD(&tp->t_items);
+	INIT_LIST_HEAD(&tp->t_busy);
+	return tp;
+}
+
+/*
+ * Free the transaction structure.  If there is more clean up
+ * to do when the structure is freed, add it here.
+ */
+STATIC void
+xfs_trans_free(
+	struct xfs_trans	*tp)
+{
+	xfs_alloc_busy_sort(&tp->t_busy);
+	xfs_alloc_busy_clear(tp->t_mountp, &tp->t_busy, false);
+
+	atomic_dec(&tp->t_mountp->m_active_trans);
+	xfs_trans_free_dqinfo(tp);
+	kmem_zone_free(xfs_trans_zone, tp);
+}
+
+/*
+ * This is called to create a new transaction which will share the
+ * permanent log reservation of the given transaction.  The remaining
+ * unused block and rt extent reservations are also inherited.  This
+ * implies that the original transaction is no longer allowed to allocate
+ * blocks.  Locks and log items, however, are no inherited.  They must
+ * be added to the new transaction explicitly.
+ */
+xfs_trans_t *
+xfs_trans_dup(
+	xfs_trans_t	*tp)
+{
+	xfs_trans_t	*ntp;
+
+	ntp = kmem_zone_zalloc(xfs_trans_zone, KM_SLEEP);
+
+	/*
+	 * Initialize the new transaction structure.
+	 */
+	ntp->t_magic = XFS_TRANS_MAGIC;
+	ntp->t_type = tp->t_type;
+	ntp->t_mountp = tp->t_mountp;
+	INIT_LIST_HEAD(&ntp->t_items);
+	INIT_LIST_HEAD(&ntp->t_busy);
+
+	ASSERT(tp->t_flags & XFS_TRANS_PERM_LOG_RES);
+	ASSERT(tp->t_ticket != NULL);
+
+	ntp->t_flags = XFS_TRANS_PERM_LOG_RES | (tp->t_flags & XFS_TRANS_RESERVE);
+	ntp->t_ticket = xfs_log_ticket_get(tp->t_ticket);
+	ntp->t_blk_res = tp->t_blk_res - tp->t_blk_res_used;
+	tp->t_blk_res = tp->t_blk_res_used;
+	ntp->t_rtx_res = tp->t_rtx_res - tp->t_rtx_res_used;
+	tp->t_rtx_res = tp->t_rtx_res_used;
+	ntp->t_pflags = tp->t_pflags;
+
+	xfs_trans_dup_dqinfo(tp, ntp);
+
+	atomic_inc(&tp->t_mountp->m_active_trans);
+	return ntp;
+}
+
+/*
+ * This is called to reserve free disk blocks and log space for the
+ * given transaction.  This must be done before allocating any resources
+ * within the transaction.
+ *
+ * This will return ENOSPC if there are not enough blocks available.
+ * It will sleep waiting for available log space.
+ * The only valid value for the flags parameter is XFS_RES_LOG_PERM, which
+ * is used by long running transactions.  If any one of the reservations
+ * fails then they will all be backed out.
+ *
+ * This does not do quota reservations. That typically is done by the
+ * caller afterwards.
+ */
+int
+xfs_trans_reserve(
+	xfs_trans_t	*tp,
+	uint		blocks,
+	uint		logspace,
+	uint		rtextents,
+	uint		flags,
+	uint		logcount)
+{
+	int		log_flags;
+	int		error = 0;
+	int		rsvd = (tp->t_flags & XFS_TRANS_RESERVE) != 0;
+
+	/* Mark this thread as being in a transaction */
+	current_set_flags_nested(&tp->t_pflags, PF_FSTRANS);
+
+	/*
+	 * Attempt to reserve the needed disk blocks by decrementing
+	 * the number needed from the number available.  This will
+	 * fail if the count would go below zero.
+	 */
+	if (blocks > 0) {
+		error = xfs_icsb_modify_counters(tp->t_mountp, XFS_SBS_FDBLOCKS,
+					  -((int64_t)blocks), rsvd);
+		if (error != 0) {
+			current_restore_flags_nested(&tp->t_pflags, PF_FSTRANS);
+			return (XFS_ERROR(ENOSPC));
+		}
+		tp->t_blk_res += blocks;
+	}
+
+	/*
+	 * Reserve the log space needed for this transaction.
+	 */
+	if (logspace > 0) {
+		ASSERT((tp->t_log_res == 0) || (tp->t_log_res == logspace));
+		ASSERT((tp->t_log_count == 0) ||
+			(tp->t_log_count == logcount));
+		if (flags & XFS_TRANS_PERM_LOG_RES) {
+			log_flags = XFS_LOG_PERM_RESERV;
+			tp->t_flags |= XFS_TRANS_PERM_LOG_RES;
+		} else {
+			ASSERT(tp->t_ticket == NULL);
+			ASSERT(!(tp->t_flags & XFS_TRANS_PERM_LOG_RES));
+			log_flags = 0;
+		}
+
+		error = xfs_log_reserve(tp->t_mountp, logspace, logcount,
+					&tp->t_ticket,
+					XFS_TRANSACTION, log_flags, tp->t_type);
+		if (error) {
+			goto undo_blocks;
+		}
+		tp->t_log_res = logspace;
+		tp->t_log_count = logcount;
+	}
+
+	/*
+	 * Attempt to reserve the needed realtime extents by decrementing
+	 * the number needed from the number available.  This will
+	 * fail if the count would go below zero.
+	 */
+	if (rtextents > 0) {
+		error = xfs_mod_incore_sb(tp->t_mountp, XFS_SBS_FREXTENTS,
+					  -((int64_t)rtextents), rsvd);
+		if (error) {
+			error = XFS_ERROR(ENOSPC);
+			goto undo_log;
+		}
+		tp->t_rtx_res += rtextents;
+	}
+
+	return 0;
+
+	/*
+	 * Error cases jump to one of these labels to undo any
+	 * reservations which have already been performed.
+	 */
+undo_log:
+	if (logspace > 0) {
+		if (flags & XFS_TRANS_PERM_LOG_RES) {
+			log_flags = XFS_LOG_REL_PERM_RESERV;
+		} else {
+			log_flags = 0;
+		}
+		xfs_log_done(tp->t_mountp, tp->t_ticket, NULL, log_flags);
+		tp->t_ticket = NULL;
+		tp->t_log_res = 0;
+		tp->t_flags &= ~XFS_TRANS_PERM_LOG_RES;
+	}
+
+undo_blocks:
+	if (blocks > 0) {
+		xfs_icsb_modify_counters(tp->t_mountp, XFS_SBS_FDBLOCKS,
+					 (int64_t)blocks, rsvd);
+		tp->t_blk_res = 0;
+	}
+
+	current_restore_flags_nested(&tp->t_pflags, PF_FSTRANS);
+
+	return error;
+}
+
+/*
+ * Record the indicated change to the given field for application
+ * to the file system's superblock when the transaction commits.
+ * For now, just store the change in the transaction structure.
+ *
+ * Mark the transaction structure to indicate that the superblock
+ * needs to be updated before committing.
+ *
+ * Because we may not be keeping track of allocated/free inodes and
+ * used filesystem blocks in the superblock, we do not mark the
+ * superblock dirty in this transaction if we modify these fields.
+ * We still need to update the transaction deltas so that they get
+ * applied to the incore superblock, but we don't want them to
+ * cause the superblock to get locked and logged if these are the
+ * only fields in the superblock that the transaction modifies.
+ */
+void
+xfs_trans_mod_sb(
+	xfs_trans_t	*tp,
+	uint		field,
+	int64_t		delta)
+{
+	uint32_t	flags = (XFS_TRANS_DIRTY|XFS_TRANS_SB_DIRTY);
+	xfs_mount_t	*mp = tp->t_mountp;
+
+	switch (field) {
+	case XFS_TRANS_SB_ICOUNT:
+		tp->t_icount_delta += delta;
+		if (xfs_sb_version_haslazysbcount(&mp->m_sb))
+			flags &= ~XFS_TRANS_SB_DIRTY;
+		break;
+	case XFS_TRANS_SB_IFREE:
+		tp->t_ifree_delta += delta;
+		if (xfs_sb_version_haslazysbcount(&mp->m_sb))
+			flags &= ~XFS_TRANS_SB_DIRTY;
+		break;
+	case XFS_TRANS_SB_FDBLOCKS:
+		/*
+		 * Track the number of blocks allocated in the
+		 * transaction.  Make sure it does not exceed the
+		 * number reserved.
+		 */
+		if (delta < 0) {
+			tp->t_blk_res_used += (uint)-delta;
+			ASSERT(tp->t_blk_res_used <= tp->t_blk_res);
+		}
+		tp->t_fdblocks_delta += delta;
+		if (xfs_sb_version_haslazysbcount(&mp->m_sb))
+			flags &= ~XFS_TRANS_SB_DIRTY;
+		break;
+	case XFS_TRANS_SB_RES_FDBLOCKS:
+		/*
+		 * The allocation has already been applied to the
+		 * in-core superblock's counter.  This should only
+		 * be applied to the on-disk superblock.
+		 */
+		ASSERT(delta < 0);
+		tp->t_res_fdblocks_delta += delta;
+		if (xfs_sb_version_haslazysbcount(&mp->m_sb))
+			flags &= ~XFS_TRANS_SB_DIRTY;
+		break;
+	case XFS_TRANS_SB_FREXTENTS:
+		/*
+		 * Track the number of blocks allocated in the
+		 * transaction.  Make sure it does not exceed the
+		 * number reserved.
+		 */
+		if (delta < 0) {
+			tp->t_rtx_res_used += (uint)-delta;
+			ASSERT(tp->t_rtx_res_used <= tp->t_rtx_res);
+		}
+		tp->t_frextents_delta += delta;
+		break;
+	case XFS_TRANS_SB_RES_FREXTENTS:
+		/*
+		 * The allocation has already been applied to the
+		 * in-core superblock's counter.  This should only
+		 * be applied to the on-disk superblock.
+		 */
+		ASSERT(delta < 0);
+		tp->t_res_frextents_delta += delta;
+		break;
+	case XFS_TRANS_SB_DBLOCKS:
+		ASSERT(delta > 0);
+		tp->t_dblocks_delta += delta;
+		break;
+	case XFS_TRANS_SB_AGCOUNT:
+		ASSERT(delta > 0);
+		tp->t_agcount_delta += delta;
+		break;
+	case XFS_TRANS_SB_IMAXPCT:
+		tp->t_imaxpct_delta += delta;
+		break;
+	case XFS_TRANS_SB_REXTSIZE:
+		tp->t_rextsize_delta += delta;
+		break;
+	case XFS_TRANS_SB_RBMBLOCKS:
+		tp->t_rbmblocks_delta += delta;
+		break;
+	case XFS_TRANS_SB_RBLOCKS:
+		tp->t_rblocks_delta += delta;
+		break;
+	case XFS_TRANS_SB_REXTENTS:
+		tp->t_rextents_delta += delta;
+		break;
+	case XFS_TRANS_SB_REXTSLOG:
+		tp->t_rextslog_delta += delta;
+		break;
+	default:
+		ASSERT(0);
+		return;
+	}
+
+	tp->t_flags |= flags;
+}
+
+/*
+ * xfs_trans_apply_sb_deltas() is called from the commit code
+ * to bring the superblock buffer into the current transaction
+ * and modify it as requested by earlier calls to xfs_trans_mod_sb().
+ *
+ * For now we just look at each field allowed to change and change
+ * it if necessary.
+ */
+STATIC void
+xfs_trans_apply_sb_deltas(
+	xfs_trans_t	*tp)
+{
+	xfs_dsb_t	*sbp;
+	xfs_buf_t	*bp;
+	int		whole = 0;
+
+	bp = xfs_trans_getsb(tp, tp->t_mountp, 0);
+	sbp = XFS_BUF_TO_SBP(bp);
+
+	/*
+	 * Check that superblock mods match the mods made to AGF counters.
+	 */
+	ASSERT((tp->t_fdblocks_delta + tp->t_res_fdblocks_delta) ==
+	       (tp->t_ag_freeblks_delta + tp->t_ag_flist_delta +
+		tp->t_ag_btree_delta));
+
+	/*
+	 * Only update the superblock counters if we are logging them
+	 */
+	if (!xfs_sb_version_haslazysbcount(&(tp->t_mountp->m_sb))) {
+		if (tp->t_icount_delta)
+			be64_add_cpu(&sbp->sb_icount, tp->t_icount_delta);
+		if (tp->t_ifree_delta)
+			be64_add_cpu(&sbp->sb_ifree, tp->t_ifree_delta);
+		if (tp->t_fdblocks_delta)
+			be64_add_cpu(&sbp->sb_fdblocks, tp->t_fdblocks_delta);
+		if (tp->t_res_fdblocks_delta)
+			be64_add_cpu(&sbp->sb_fdblocks, tp->t_res_fdblocks_delta);
+	}
+
+	if (tp->t_frextents_delta)
+		be64_add_cpu(&sbp->sb_frextents, tp->t_frextents_delta);
+	if (tp->t_res_frextents_delta)
+		be64_add_cpu(&sbp->sb_frextents, tp->t_res_frextents_delta);
+
+	if (tp->t_dblocks_delta) {
+		be64_add_cpu(&sbp->sb_dblocks, tp->t_dblocks_delta);
+		whole = 1;
+	}
+	if (tp->t_agcount_delta) {
+		be32_add_cpu(&sbp->sb_agcount, tp->t_agcount_delta);
+		whole = 1;
+	}
+	if (tp->t_imaxpct_delta) {
+		sbp->sb_imax_pct += tp->t_imaxpct_delta;
+		whole = 1;
+	}
+	if (tp->t_rextsize_delta) {
+		be32_add_cpu(&sbp->sb_rextsize, tp->t_rextsize_delta);
+		whole = 1;
+	}
+	if (tp->t_rbmblocks_delta) {
+		be32_add_cpu(&sbp->sb_rbmblocks, tp->t_rbmblocks_delta);
+		whole = 1;
+	}
+	if (tp->t_rblocks_delta) {
+		be64_add_cpu(&sbp->sb_rblocks, tp->t_rblocks_delta);
+		whole = 1;
+	}
+	if (tp->t_rextents_delta) {
+		be64_add_cpu(&sbp->sb_rextents, tp->t_rextents_delta);
+		whole = 1;
+	}
+	if (tp->t_rextslog_delta) {
+		sbp->sb_rextslog += tp->t_rextslog_delta;
+		whole = 1;
+	}
+
+	if (whole)
+		/*
+		 * Log the whole thing, the fields are noncontiguous.
+		 */
+		xfs_trans_log_buf(tp, bp, 0, sizeof(xfs_dsb_t) - 1);
+	else
+		/*
+		 * Since all the modifiable fields are contiguous, we
+		 * can get away with this.
+		 */
+		xfs_trans_log_buf(tp, bp, offsetof(xfs_dsb_t, sb_icount),
+				  offsetof(xfs_dsb_t, sb_frextents) +
+				  sizeof(sbp->sb_frextents) - 1);
+}
+
+/*
+ * xfs_trans_unreserve_and_mod_sb() is called to release unused reservations
+ * and apply superblock counter changes to the in-core superblock.  The
+ * t_res_fdblocks_delta and t_res_frextents_delta fields are explicitly NOT
+ * applied to the in-core superblock.  The idea is that that has already been
+ * done.
+ *
+ * This is done efficiently with a single call to xfs_mod_incore_sb_batch().
+ * However, we have to ensure that we only modify each superblock field only
+ * once because the application of the delta values may not be atomic. That can
+ * lead to ENOSPC races occurring if we have two separate modifcations of the
+ * free space counter to put back the entire reservation and then take away
+ * what we used.
+ *
+ * If we are not logging superblock counters, then the inode allocated/free and
+ * used block counts are not updated in the on disk superblock. In this case,
+ * XFS_TRANS_SB_DIRTY will not be set when the transaction is updated but we
+ * still need to update the incore superblock with the changes.
+ */
+void
+xfs_trans_unreserve_and_mod_sb(
+	xfs_trans_t	*tp)
+{
+	xfs_mod_sb_t	msb[9];	/* If you add cases, add entries */
+	xfs_mod_sb_t	*msbp;
+	xfs_mount_t	*mp = tp->t_mountp;
+	/* REFERENCED */
+	int		error;
+	int		rsvd;
+	int64_t		blkdelta = 0;
+	int64_t		rtxdelta = 0;
+	int64_t		idelta = 0;
+	int64_t		ifreedelta = 0;
+
+	msbp = msb;
+	rsvd = (tp->t_flags & XFS_TRANS_RESERVE) != 0;
+
+	/* calculate deltas */
+	if (tp->t_blk_res > 0)
+		blkdelta = tp->t_blk_res;
+	if ((tp->t_fdblocks_delta != 0) &&
+	    (xfs_sb_version_haslazysbcount(&mp->m_sb) ||
+	     (tp->t_flags & XFS_TRANS_SB_DIRTY)))
+	        blkdelta += tp->t_fdblocks_delta;
+
+	if (tp->t_rtx_res > 0)
+		rtxdelta = tp->t_rtx_res;
+	if ((tp->t_frextents_delta != 0) &&
+	    (tp->t_flags & XFS_TRANS_SB_DIRTY))
+		rtxdelta += tp->t_frextents_delta;
+
+	if (xfs_sb_version_haslazysbcount(&mp->m_sb) ||
+	     (tp->t_flags & XFS_TRANS_SB_DIRTY)) {
+		idelta = tp->t_icount_delta;
+		ifreedelta = tp->t_ifree_delta;
+	}
+
+	/* apply the per-cpu counters */
+	if (blkdelta) {
+		error = xfs_icsb_modify_counters(mp, XFS_SBS_FDBLOCKS,
+						 blkdelta, rsvd);
+		if (error)
+			goto out;
+	}
+
+	if (idelta) {
+		error = xfs_icsb_modify_counters(mp, XFS_SBS_ICOUNT,
+						 idelta, rsvd);
+		if (error)
+			goto out_undo_fdblocks;
+	}
+
+	if (ifreedelta) {
+		error = xfs_icsb_modify_counters(mp, XFS_SBS_IFREE,
+						 ifreedelta, rsvd);
+		if (error)
+			goto out_undo_icount;
+	}
+
+	/* apply remaining deltas */
+	if (rtxdelta != 0) {
+		msbp->msb_field = XFS_SBS_FREXTENTS;
+		msbp->msb_delta = rtxdelta;
+		msbp++;
+	}
+
+	if (tp->t_flags & XFS_TRANS_SB_DIRTY) {
+		if (tp->t_dblocks_delta != 0) {
+			msbp->msb_field = XFS_SBS_DBLOCKS;
+			msbp->msb_delta = tp->t_dblocks_delta;
+			msbp++;
+		}
+		if (tp->t_agcount_delta != 0) {
+			msbp->msb_field = XFS_SBS_AGCOUNT;
+			msbp->msb_delta = tp->t_agcount_delta;
+			msbp++;
+		}
+		if (tp->t_imaxpct_delta != 0) {
+			msbp->msb_field = XFS_SBS_IMAX_PCT;
+			msbp->msb_delta = tp->t_imaxpct_delta;
+			msbp++;
+		}
+		if (tp->t_rextsize_delta != 0) {
+			msbp->msb_field = XFS_SBS_REXTSIZE;
+			msbp->msb_delta = tp->t_rextsize_delta;
+			msbp++;
+		}
+		if (tp->t_rbmblocks_delta != 0) {
+			msbp->msb_field = XFS_SBS_RBMBLOCKS;
+			msbp->msb_delta = tp->t_rbmblocks_delta;
+			msbp++;
+		}
+		if (tp->t_rblocks_delta != 0) {
+			msbp->msb_field = XFS_SBS_RBLOCKS;
+			msbp->msb_delta = tp->t_rblocks_delta;
+			msbp++;
+		}
+		if (tp->t_rextents_delta != 0) {
+			msbp->msb_field = XFS_SBS_REXTENTS;
+			msbp->msb_delta = tp->t_rextents_delta;
+			msbp++;
+		}
+		if (tp->t_rextslog_delta != 0) {
+			msbp->msb_field = XFS_SBS_REXTSLOG;
+			msbp->msb_delta = tp->t_rextslog_delta;
+			msbp++;
+		}
+	}
+
+	/*
+	 * If we need to change anything, do it.
+	 */
+	if (msbp > msb) {
+		error = xfs_mod_incore_sb_batch(tp->t_mountp, msb,
+			(uint)(msbp - msb), rsvd);
+		if (error)
+			goto out_undo_ifreecount;
+	}
+
+	return;
+
+out_undo_ifreecount:
+	if (ifreedelta)
+		xfs_icsb_modify_counters(mp, XFS_SBS_IFREE, -ifreedelta, rsvd);
+out_undo_icount:
+	if (idelta)
+		xfs_icsb_modify_counters(mp, XFS_SBS_ICOUNT, -idelta, rsvd);
+out_undo_fdblocks:
+	if (blkdelta)
+		xfs_icsb_modify_counters(mp, XFS_SBS_FDBLOCKS, -blkdelta, rsvd);
+out:
+	ASSERT(error == 0);
+	return;
+}
+
+/*
+ * Add the given log item to the transaction's list of log items.
+ *
+ * The log item will now point to its new descriptor with its li_desc field.
+ */
+void
+xfs_trans_add_item(
+	struct xfs_trans	*tp,
+	struct xfs_log_item	*lip)
+{
+	struct xfs_log_item_desc *lidp;
+
+	ASSERT(lip->li_mountp = tp->t_mountp);
+	ASSERT(lip->li_ailp = tp->t_mountp->m_ail);
+
+	lidp = kmem_zone_zalloc(xfs_log_item_desc_zone, KM_SLEEP | KM_NOFS);
+
+	lidp->lid_item = lip;
+	lidp->lid_flags = 0;
+	lidp->lid_size = 0;
+	list_add_tail(&lidp->lid_trans, &tp->t_items);
+
+	lip->li_desc = lidp;
+}
+
+STATIC void
+xfs_trans_free_item_desc(
+	struct xfs_log_item_desc *lidp)
+{
+	list_del_init(&lidp->lid_trans);
+	kmem_zone_free(xfs_log_item_desc_zone, lidp);
+}
+
+/*
+ * Unlink and free the given descriptor.
+ */
+void
+xfs_trans_del_item(
+	struct xfs_log_item	*lip)
+{
+	xfs_trans_free_item_desc(lip->li_desc);
+	lip->li_desc = NULL;
+}
+
+/*
+ * Unlock all of the items of a transaction and free all the descriptors
+ * of that transaction.
+ */
+void
+xfs_trans_free_items(
+	struct xfs_trans	*tp,
+	xfs_lsn_t		commit_lsn,
+	int			flags)
+{
+	struct xfs_log_item_desc *lidp, *next;
+
+	list_for_each_entry_safe(lidp, next, &tp->t_items, lid_trans) {
+		struct xfs_log_item	*lip = lidp->lid_item;
+
+		lip->li_desc = NULL;
+
+		if (commit_lsn != NULLCOMMITLSN)
+			IOP_COMMITTING(lip, commit_lsn);
+		if (flags & XFS_TRANS_ABORT)
+			lip->li_flags |= XFS_LI_ABORTED;
+		IOP_UNLOCK(lip);
+
+		xfs_trans_free_item_desc(lidp);
+	}
+}
+
+/*
+ * Unlock the items associated with a transaction.
+ *
+ * Items which were not logged should be freed.  Those which were logged must
+ * still be tracked so they can be unpinned when the transaction commits.
+ */
+STATIC void
+xfs_trans_unlock_items(
+	struct xfs_trans	*tp,
+	xfs_lsn_t		commit_lsn)
+{
+	struct xfs_log_item_desc *lidp, *next;
+
+	list_for_each_entry_safe(lidp, next, &tp->t_items, lid_trans) {
+		struct xfs_log_item	*lip = lidp->lid_item;
+
+		lip->li_desc = NULL;
+
+		if (commit_lsn != NULLCOMMITLSN)
+			IOP_COMMITTING(lip, commit_lsn);
+		IOP_UNLOCK(lip);
+
+		/*
+		 * Free the descriptor if the item is not dirty
+		 * within this transaction.
+		 */
+		if (!(lidp->lid_flags & XFS_LID_DIRTY))
+			xfs_trans_free_item_desc(lidp);
+	}
+}
+
+/*
+ * Total up the number of log iovecs needed to commit this
+ * transaction.  The transaction itself needs one for the
+ * transaction header.  Ask each dirty item in turn how many
+ * it needs to get the total.
+ */
+static uint
+xfs_trans_count_vecs(
+	struct xfs_trans	*tp)
+{
+	int			nvecs;
+	struct xfs_log_item_desc *lidp;
+
+	nvecs = 1;
+
+	/* In the non-debug case we need to start bailing out if we
+	 * didn't find a log_item here, return zero and let trans_commit
+	 * deal with it.
+	 */
+	if (list_empty(&tp->t_items)) {
+		ASSERT(0);
+		return 0;
+	}
+
+	list_for_each_entry(lidp, &tp->t_items, lid_trans) {
+		/*
+		 * Skip items which aren't dirty in this transaction.
+		 */
+		if (!(lidp->lid_flags & XFS_LID_DIRTY))
+			continue;
+		lidp->lid_size = IOP_SIZE(lidp->lid_item);
+		nvecs += lidp->lid_size;
+	}
+
+	return nvecs;
+}
+
+/*
+ * Fill in the vector with pointers to data to be logged
+ * by this transaction.  The transaction header takes
+ * the first vector, and then each dirty item takes the
+ * number of vectors it indicated it needed in xfs_trans_count_vecs().
+ *
+ * As each item fills in the entries it needs, also pin the item
+ * so that it cannot be flushed out until the log write completes.
+ */
+static void
+xfs_trans_fill_vecs(
+	struct xfs_trans	*tp,
+	struct xfs_log_iovec	*log_vector)
+{
+	struct xfs_log_item_desc *lidp;
+	struct xfs_log_iovec	*vecp;
+	uint			nitems;
+
+	/*
+	 * Skip over the entry for the transaction header, we'll
+	 * fill that in at the end.
+	 */
+	vecp = log_vector + 1;
+
+	nitems = 0;
+	ASSERT(!list_empty(&tp->t_items));
+	list_for_each_entry(lidp, &tp->t_items, lid_trans) {
+		/* Skip items which aren't dirty in this transaction. */
+		if (!(lidp->lid_flags & XFS_LID_DIRTY))
+			continue;
+
+		/*
+		 * The item may be marked dirty but not log anything.  This can
+		 * be used to get called when a transaction is committed.
+		 */
+		if (lidp->lid_size)
+			nitems++;
+		IOP_FORMAT(lidp->lid_item, vecp);
+		vecp += lidp->lid_size;
+		IOP_PIN(lidp->lid_item);
+	}
+
+	/*
+	 * Now that we've counted the number of items in this transaction, fill
+	 * in the transaction header. Note that the transaction header does not
+	 * have a log item.
+	 */
+	tp->t_header.th_magic = XFS_TRANS_HEADER_MAGIC;
+	tp->t_header.th_type = tp->t_type;
+	tp->t_header.th_num_items = nitems;
+	log_vector->i_addr = (xfs_caddr_t)&tp->t_header;
+	log_vector->i_len = sizeof(xfs_trans_header_t);
+	log_vector->i_type = XLOG_REG_TYPE_TRANSHDR;
+}
+
+/*
+ * The committed item processing consists of calling the committed routine of
+ * each logged item, updating the item's position in the AIL if necessary, and
+ * unpinning each item.  If the committed routine returns -1, then do nothing
+ * further with the item because it may have been freed.
+ *
+ * Since items are unlocked when they are copied to the incore log, it is
+ * possible for two transactions to be completing and manipulating the same
+ * item simultaneously.  The AIL lock will protect the lsn field of each item.
+ * The value of this field can never go backwards.
+ *
+ * We unpin the items after repositioning them in the AIL, because otherwise
+ * they could be immediately flushed and we'd have to race with the flusher
+ * trying to pull the item from the AIL as we add it.
+ */
+static void
+xfs_trans_item_committed(
+	struct xfs_log_item	*lip,
+	xfs_lsn_t		commit_lsn,
+	int			aborted)
+{
+	xfs_lsn_t		item_lsn;
+	struct xfs_ail		*ailp;
+
+	if (aborted)
+		lip->li_flags |= XFS_LI_ABORTED;
+	item_lsn = IOP_COMMITTED(lip, commit_lsn);
+
+	/* item_lsn of -1 means the item needs no further processing */
+	if (XFS_LSN_CMP(item_lsn, (xfs_lsn_t)-1) == 0)
+		return;
+
+	/*
+	 * If the returned lsn is greater than what it contained before, update
+	 * the location of the item in the AIL.  If it is not, then do nothing.
+	 * Items can never move backwards in the AIL.
+	 *
+	 * While the new lsn should usually be greater, it is possible that a
+	 * later transaction completing simultaneously with an earlier one
+	 * using the same item could complete first with a higher lsn.  This
+	 * would cause the earlier transaction to fail the test below.
+	 */
+	ailp = lip->li_ailp;
+	spin_lock(&ailp->xa_lock);
+	if (XFS_LSN_CMP(item_lsn, lip->li_lsn) > 0) {
+		/*
+		 * This will set the item's lsn to item_lsn and update the
+		 * position of the item in the AIL.
+		 *
+		 * xfs_trans_ail_update() drops the AIL lock.
+		 */
+		xfs_trans_ail_update(ailp, lip, item_lsn);
+	} else {
+		spin_unlock(&ailp->xa_lock);
+	}
+
+	/*
+	 * Now that we've repositioned the item in the AIL, unpin it so it can
+	 * be flushed. Pass information about buffer stale state down from the
+	 * log item flags, if anyone else stales the buffer we do not want to
+	 * pay any attention to it.
+	 */
+	IOP_UNPIN(lip, 0);
+}
+
+/*
+ * This is typically called by the LM when a transaction has been fully
+ * committed to disk.  It needs to unpin the items which have
+ * been logged by the transaction and update their positions
+ * in the AIL if necessary.
+ *
+ * This also gets called when the transactions didn't get written out
+ * because of an I/O error. Abortflag & XFS_LI_ABORTED is set then.
+ */
+STATIC void
+xfs_trans_committed(
+	void			*arg,
+	int			abortflag)
+{
+	struct xfs_trans	*tp = arg;
+	struct xfs_log_item_desc *lidp, *next;
+
+	list_for_each_entry_safe(lidp, next, &tp->t_items, lid_trans) {
+		xfs_trans_item_committed(lidp->lid_item, tp->t_lsn, abortflag);
+		xfs_trans_free_item_desc(lidp);
+	}
+
+	xfs_trans_free(tp);
+}
+
+static inline void
+xfs_log_item_batch_insert(
+	struct xfs_ail		*ailp,
+	struct xfs_ail_cursor	*cur,
+	struct xfs_log_item	**log_items,
+	int			nr_items,
+	xfs_lsn_t		commit_lsn)
+{
+	int	i;
+
+	spin_lock(&ailp->xa_lock);
+	/* xfs_trans_ail_update_bulk drops ailp->xa_lock */
+	xfs_trans_ail_update_bulk(ailp, cur, log_items, nr_items, commit_lsn);
+
+	for (i = 0; i < nr_items; i++)
+		IOP_UNPIN(log_items[i], 0);
+}
+
+/*
+ * Bulk operation version of xfs_trans_committed that takes a log vector of
+ * items to insert into the AIL. This uses bulk AIL insertion techniques to
+ * minimise lock traffic.
+ *
+ * If we are called with the aborted flag set, it is because a log write during
+ * a CIL checkpoint commit has failed. In this case, all the items in the
+ * checkpoint have already gone through IOP_COMMITED and IOP_UNLOCK, which
+ * means that checkpoint commit abort handling is treated exactly the same
+ * as an iclog write error even though we haven't started any IO yet. Hence in
+ * this case all we need to do is IOP_COMMITTED processing, followed by an
+ * IOP_UNPIN(aborted) call.
+ *
+ * The AIL cursor is used to optimise the insert process. If commit_lsn is not
+ * at the end of the AIL, the insert cursor avoids the need to walk
+ * the AIL to find the insertion point on every xfs_log_item_batch_insert()
+ * call. This saves a lot of needless list walking and is a net win, even
+ * though it slightly increases that amount of AIL lock traffic to set it up
+ * and tear it down.
+ */
+void
+xfs_trans_committed_bulk(
+	struct xfs_ail		*ailp,
+	struct xfs_log_vec	*log_vector,
+	xfs_lsn_t		commit_lsn,
+	int			aborted)
+{
+#define LOG_ITEM_BATCH_SIZE	32
+	struct xfs_log_item	*log_items[LOG_ITEM_BATCH_SIZE];
+	struct xfs_log_vec	*lv;
+	struct xfs_ail_cursor	cur;
+	int			i = 0;
+
+	spin_lock(&ailp->xa_lock);
+	xfs_trans_ail_cursor_last(ailp, &cur, commit_lsn);
+	spin_unlock(&ailp->xa_lock);
+
+	/* unpin all the log items */
+	for (lv = log_vector; lv; lv = lv->lv_next ) {
+		struct xfs_log_item	*lip = lv->lv_item;
+		xfs_lsn_t		item_lsn;
+
+		if (aborted)
+			lip->li_flags |= XFS_LI_ABORTED;
+		item_lsn = IOP_COMMITTED(lip, commit_lsn);
+
+		/* item_lsn of -1 means the item needs no further processing */
+		if (XFS_LSN_CMP(item_lsn, (xfs_lsn_t)-1) == 0)
+			continue;
+
+		/*
+		 * if we are aborting the operation, no point in inserting the
+		 * object into the AIL as we are in a shutdown situation.
+		 */
+		if (aborted) {
+			ASSERT(XFS_FORCED_SHUTDOWN(ailp->xa_mount));
+			IOP_UNPIN(lip, 1);
+			continue;
+		}
+
+		if (item_lsn != commit_lsn) {
+
+			/*
+			 * Not a bulk update option due to unusual item_lsn.
+			 * Push into AIL immediately, rechecking the lsn once
+			 * we have the ail lock. Then unpin the item. This does
+			 * not affect the AIL cursor the bulk insert path is
+			 * using.
+			 */
+			spin_lock(&ailp->xa_lock);
+			if (XFS_LSN_CMP(item_lsn, lip->li_lsn) > 0)
+				xfs_trans_ail_update(ailp, lip, item_lsn);
+			else
+				spin_unlock(&ailp->xa_lock);
+			IOP_UNPIN(lip, 0);
+			continue;
+		}
+
+		/* Item is a candidate for bulk AIL insert.  */
+		log_items[i++] = lv->lv_item;
+		if (i >= LOG_ITEM_BATCH_SIZE) {
+			xfs_log_item_batch_insert(ailp, &cur, log_items,
+					LOG_ITEM_BATCH_SIZE, commit_lsn);
+			i = 0;
+		}
+	}
+
+	/* make sure we insert the remainder! */
+	if (i)
+		xfs_log_item_batch_insert(ailp, &cur, log_items, i, commit_lsn);
+
+	spin_lock(&ailp->xa_lock);
+	xfs_trans_ail_cursor_done(ailp, &cur);
+	spin_unlock(&ailp->xa_lock);
+}
+
+/*
+ * Called from the trans_commit code when we notice that the filesystem is in
+ * the middle of a forced shutdown.
+ *
+ * When we are called here, we have already pinned all the items in the
+ * transaction. However, neither IOP_COMMITTING or IOP_UNLOCK has been called
+ * so we can simply walk the items in the transaction, unpin them with an abort
+ * flag and then free the items. Note that unpinning the items can result in
+ * them being freed immediately, so we need to use a safe list traversal method
+ * here.
+ */
+STATIC void
+xfs_trans_uncommit(
+	struct xfs_trans	*tp,
+	uint			flags)
+{
+	struct xfs_log_item_desc *lidp, *n;
+
+	list_for_each_entry_safe(lidp, n, &tp->t_items, lid_trans) {
+		if (lidp->lid_flags & XFS_LID_DIRTY)
+			IOP_UNPIN(lidp->lid_item, 1);
+	}
+
+	xfs_trans_unreserve_and_mod_sb(tp);
+	xfs_trans_unreserve_and_mod_dquots(tp);
+
+	xfs_trans_free_items(tp, NULLCOMMITLSN, flags);
+	xfs_trans_free(tp);
+}
+
+/*
+ * Format the transaction direct to the iclog. This isolates the physical
+ * transaction commit operation from the logical operation and hence allows
+ * other methods to be introduced without affecting the existing commit path.
+ */
+static int
+xfs_trans_commit_iclog(
+	struct xfs_mount	*mp,
+	struct xfs_trans	*tp,
+	xfs_lsn_t		*commit_lsn,
+	int			flags)
+{
+	int			shutdown;
+	int			error;
+	int			log_flags = 0;
+	struct xlog_in_core	*commit_iclog;
+#define XFS_TRANS_LOGVEC_COUNT  16
+	struct xfs_log_iovec	log_vector_fast[XFS_TRANS_LOGVEC_COUNT];
+	struct xfs_log_iovec	*log_vector;
+	uint			nvec;
+
+
+	/*
+	 * Ask each log item how many log_vector entries it will
+	 * need so we can figure out how many to allocate.
+	 * Try to avoid the kmem_alloc() call in the common case
+	 * by using a vector from the stack when it fits.
+	 */
+	nvec = xfs_trans_count_vecs(tp);
+	if (nvec == 0) {
+		return ENOMEM;	/* triggers a shutdown! */
+	} else if (nvec <= XFS_TRANS_LOGVEC_COUNT) {
+		log_vector = log_vector_fast;
+	} else {
+		log_vector = (xfs_log_iovec_t *)kmem_alloc(nvec *
+						   sizeof(xfs_log_iovec_t),
+						   KM_SLEEP);
+	}
+
+	/*
+	 * Fill in the log_vector and pin the logged items, and
+	 * then write the transaction to the log.
+	 */
+	xfs_trans_fill_vecs(tp, log_vector);
+
+	if (flags & XFS_TRANS_RELEASE_LOG_RES)
+		log_flags = XFS_LOG_REL_PERM_RESERV;
+
+	error = xfs_log_write(mp, log_vector, nvec, tp->t_ticket, &(tp->t_lsn));
+
+	/*
+	 * The transaction is committed incore here, and can go out to disk
+	 * at any time after this call.  However, all the items associated
+	 * with the transaction are still locked and pinned in memory.
+	 */
+	*commit_lsn = xfs_log_done(mp, tp->t_ticket, &commit_iclog, log_flags);
+
+	tp->t_commit_lsn = *commit_lsn;
+	trace_xfs_trans_commit_lsn(tp);
+
+	if (nvec > XFS_TRANS_LOGVEC_COUNT)
+		kmem_free(log_vector);
+
+	/*
+	 * If we got a log write error. Unpin the logitems that we
+	 * had pinned, clean up, free trans structure, and return error.
+	 */
+	if (error || *commit_lsn == -1) {
+		current_restore_flags_nested(&tp->t_pflags, PF_FSTRANS);
+		xfs_trans_uncommit(tp, flags|XFS_TRANS_ABORT);
+		return XFS_ERROR(EIO);
+	}
+
+	/*
+	 * Once the transaction has committed, unused
+	 * reservations need to be released and changes to
+	 * the superblock need to be reflected in the in-core
+	 * version.  Do that now.
+	 */
+	xfs_trans_unreserve_and_mod_sb(tp);
+
+	/*
+	 * Tell the LM to call the transaction completion routine
+	 * when the log write with LSN commit_lsn completes (e.g.
+	 * when the transaction commit really hits the on-disk log).
+	 * After this call we cannot reference tp, because the call
+	 * can happen at any time and the call will free the transaction
+	 * structure pointed to by tp.  The only case where we call
+	 * the completion routine (xfs_trans_committed) directly is
+	 * if the log is turned off on a debug kernel or we're
+	 * running in simulation mode (the log is explicitly turned
+	 * off).
+	 */
+	tp->t_logcb.cb_func = xfs_trans_committed;
+	tp->t_logcb.cb_arg = tp;
+
+	/*
+	 * We need to pass the iclog buffer which was used for the
+	 * transaction commit record into this function, and attach
+	 * the callback to it. The callback must be attached before
+	 * the items are unlocked to avoid racing with other threads
+	 * waiting for an item to unlock.
+	 */
+	shutdown = xfs_log_notify(mp, commit_iclog, &(tp->t_logcb));
+
+	/*
+	 * Mark this thread as no longer being in a transaction
+	 */
+	current_restore_flags_nested(&tp->t_pflags, PF_FSTRANS);
+
+	/*
+	 * Once all the items of the transaction have been copied
+	 * to the in core log and the callback is attached, the
+	 * items can be unlocked.
+	 *
+	 * This will free descriptors pointing to items which were
+	 * not logged since there is nothing more to do with them.
+	 * For items which were logged, we will keep pointers to them
+	 * so they can be unpinned after the transaction commits to disk.
+	 * This will also stamp each modified meta-data item with
+	 * the commit lsn of this transaction for dependency tracking
+	 * purposes.
+	 */
+	xfs_trans_unlock_items(tp, *commit_lsn);
+
+	/*
+	 * If we detected a log error earlier, finish committing
+	 * the transaction now (unpin log items, etc).
+	 *
+	 * Order is critical here, to avoid using the transaction
+	 * pointer after its been freed (by xfs_trans_committed
+	 * either here now, or as a callback).  We cannot do this
+	 * step inside xfs_log_notify as was done earlier because
+	 * of this issue.
+	 */
+	if (shutdown)
+		xfs_trans_committed(tp, XFS_LI_ABORTED);
+
+	/*
+	 * Now that the xfs_trans_committed callback has been attached,
+	 * and the items are released we can finally allow the iclog to
+	 * go to disk.
+	 */
+	return xfs_log_release_iclog(mp, commit_iclog);
+}
+
+/*
+ * Walk the log items and allocate log vector structures for
+ * each item large enough to fit all the vectors they require.
+ * Note that this format differs from the old log vector format in
+ * that there is no transaction header in these log vectors.
+ */
+STATIC struct xfs_log_vec *
+xfs_trans_alloc_log_vecs(
+	xfs_trans_t	*tp)
+{
+	struct xfs_log_item_desc *lidp;
+	struct xfs_log_vec	*lv = NULL;
+	struct xfs_log_vec	*ret_lv = NULL;
+
+
+	/* Bail out if we didn't find a log item.  */
+	if (list_empty(&tp->t_items)) {
+		ASSERT(0);
+		return NULL;
+	}
+
+	list_for_each_entry(lidp, &tp->t_items, lid_trans) {
+		struct xfs_log_vec *new_lv;
+
+		/* Skip items which aren't dirty in this transaction. */
+		if (!(lidp->lid_flags & XFS_LID_DIRTY))
+			continue;
+
+		/* Skip items that do not have any vectors for writing */
+		lidp->lid_size = IOP_SIZE(lidp->lid_item);
+		if (!lidp->lid_size)
+			continue;
+
+		new_lv = kmem_zalloc(sizeof(*new_lv) +
+				lidp->lid_size * sizeof(struct xfs_log_iovec),
+				KM_SLEEP);
+
+		/* The allocated iovec region lies beyond the log vector. */
+		new_lv->lv_iovecp = (struct xfs_log_iovec *)&new_lv[1];
+		new_lv->lv_niovecs = lidp->lid_size;
+		new_lv->lv_item = lidp->lid_item;
+		if (!ret_lv)
+			ret_lv = new_lv;
+		else
+			lv->lv_next = new_lv;
+		lv = new_lv;
+	}
+
+	return ret_lv;
+}
+
+static int
+xfs_trans_commit_cil(
+	struct xfs_mount	*mp,
+	struct xfs_trans	*tp,
+	xfs_lsn_t		*commit_lsn,
+	int			flags)
+{
+	struct xfs_log_vec	*log_vector;
+
+	/*
+	 * Get each log item to allocate a vector structure for
+	 * the log item to to pass to the log write code. The
+	 * CIL commit code will format the vector and save it away.
+	 */
+	log_vector = xfs_trans_alloc_log_vecs(tp);
+	if (!log_vector)
+		return ENOMEM;
+
+	xfs_log_commit_cil(mp, tp, log_vector, commit_lsn, flags);
+
+	current_restore_flags_nested(&tp->t_pflags, PF_FSTRANS);
+	xfs_trans_free(tp);
+	return 0;
+}
+
+/*
+ * xfs_trans_commit
+ *
+ * Commit the given transaction to the log a/synchronously.
+ *
+ * XFS disk error handling mechanism is not based on a typical
+ * transaction abort mechanism. Logically after the filesystem
+ * gets marked 'SHUTDOWN', we can't let any new transactions
+ * be durable - ie. committed to disk - because some metadata might
+ * be inconsistent. In such cases, this returns an error, and the
+ * caller may assume that all locked objects joined to the transaction
+ * have already been unlocked as if the commit had succeeded.
+ * Do not reference the transaction structure after this call.
+ */
+int
+_xfs_trans_commit(
+	struct xfs_trans	*tp,
+	uint			flags,
+	int			*log_flushed)
+{
+	struct xfs_mount	*mp = tp->t_mountp;
+	xfs_lsn_t		commit_lsn = -1;
+	int			error = 0;
+	int			log_flags = 0;
+	int			sync = tp->t_flags & XFS_TRANS_SYNC;
+
+	/*
+	 * Determine whether this commit is releasing a permanent
+	 * log reservation or not.
+	 */
+	if (flags & XFS_TRANS_RELEASE_LOG_RES) {
+		ASSERT(tp->t_flags & XFS_TRANS_PERM_LOG_RES);
+		log_flags = XFS_LOG_REL_PERM_RESERV;
+	}
+
+	/*
+	 * If there is nothing to be logged by the transaction,
+	 * then unlock all of the items associated with the
+	 * transaction and free the transaction structure.
+	 * Also make sure to return any reserved blocks to
+	 * the free pool.
+	 */
+	if (!(tp->t_flags & XFS_TRANS_DIRTY))
+		goto out_unreserve;
+
+	if (XFS_FORCED_SHUTDOWN(mp)) {
+		error = XFS_ERROR(EIO);
+		goto out_unreserve;
+	}
+
+	ASSERT(tp->t_ticket != NULL);
+
+	/*
+	 * If we need to update the superblock, then do it now.
+	 */
+	if (tp->t_flags & XFS_TRANS_SB_DIRTY)
+		xfs_trans_apply_sb_deltas(tp);
+	xfs_trans_apply_dquot_deltas(tp);
+
+	if (mp->m_flags & XFS_MOUNT_DELAYLOG)
+		error = xfs_trans_commit_cil(mp, tp, &commit_lsn, flags);
+	else
+		error = xfs_trans_commit_iclog(mp, tp, &commit_lsn, flags);
+
+	if (error == ENOMEM) {
+		xfs_force_shutdown(mp, SHUTDOWN_LOG_IO_ERROR);
+		error = XFS_ERROR(EIO);
+		goto out_unreserve;
+	}
+
+	/*
+	 * If the transaction needs to be synchronous, then force the
+	 * log out now and wait for it.
+	 */
+	if (sync) {
+		if (!error) {
+			error = _xfs_log_force_lsn(mp, commit_lsn,
+				      XFS_LOG_SYNC, log_flushed);
+		}
+		XFS_STATS_INC(xs_trans_sync);
+	} else {
+		XFS_STATS_INC(xs_trans_async);
+	}
+
+	return error;
+
+out_unreserve:
+	xfs_trans_unreserve_and_mod_sb(tp);
+
+	/*
+	 * It is indeed possible for the transaction to be not dirty but
+	 * the dqinfo portion to be.  All that means is that we have some
+	 * (non-persistent) quota reservations that need to be unreserved.
+	 */
+	xfs_trans_unreserve_and_mod_dquots(tp);
+	if (tp->t_ticket) {
+		commit_lsn = xfs_log_done(mp, tp->t_ticket, NULL, log_flags);
+		if (commit_lsn == -1 && !error)
+			error = XFS_ERROR(EIO);
+	}
+	current_restore_flags_nested(&tp->t_pflags, PF_FSTRANS);
+	xfs_trans_free_items(tp, NULLCOMMITLSN, error ? XFS_TRANS_ABORT : 0);
+	xfs_trans_free(tp);
+
+	XFS_STATS_INC(xs_trans_empty);
+	return error;
+}
+
+/*
+ * Unlock all of the transaction's items and free the transaction.
+ * The transaction must not have modified any of its items, because
+ * there is no way to restore them to their previous state.
+ *
+ * If the transaction has made a log reservation, make sure to release
+ * it as well.
+ */
+void
+xfs_trans_cancel(
+	xfs_trans_t		*tp,
+	int			flags)
+{
+	int			log_flags;
+	xfs_mount_t		*mp = tp->t_mountp;
+
+	/*
+	 * See if the caller is being too lazy to figure out if
+	 * the transaction really needs an abort.
+	 */
+	if ((flags & XFS_TRANS_ABORT) && !(tp->t_flags & XFS_TRANS_DIRTY))
+		flags &= ~XFS_TRANS_ABORT;
+	/*
+	 * See if the caller is relying on us to shut down the
+	 * filesystem.  This happens in paths where we detect
+	 * corruption and decide to give up.
+	 */
+	if ((tp->t_flags & XFS_TRANS_DIRTY) && !XFS_FORCED_SHUTDOWN(mp)) {
+		XFS_ERROR_REPORT("xfs_trans_cancel", XFS_ERRLEVEL_LOW, mp);
+		xfs_force_shutdown(mp, SHUTDOWN_CORRUPT_INCORE);
+	}
+#ifdef DEBUG
+	if (!(flags & XFS_TRANS_ABORT) && !XFS_FORCED_SHUTDOWN(mp)) {
+		struct xfs_log_item_desc *lidp;
+
+		list_for_each_entry(lidp, &tp->t_items, lid_trans)
+			ASSERT(!(lidp->lid_item->li_type == XFS_LI_EFD));
+	}
+#endif
+	xfs_trans_unreserve_and_mod_sb(tp);
+	xfs_trans_unreserve_and_mod_dquots(tp);
+
+	if (tp->t_ticket) {
+		if (flags & XFS_TRANS_RELEASE_LOG_RES) {
+			ASSERT(tp->t_flags & XFS_TRANS_PERM_LOG_RES);
+			log_flags = XFS_LOG_REL_PERM_RESERV;
+		} else {
+			log_flags = 0;
+		}
+		xfs_log_done(mp, tp->t_ticket, NULL, log_flags);
+	}
+
+	/* mark this thread as no longer being in a transaction */
+	current_restore_flags_nested(&tp->t_pflags, PF_FSTRANS);
+
+	xfs_trans_free_items(tp, NULLCOMMITLSN, flags);
+	xfs_trans_free(tp);
+}
+
+/*
+ * Roll from one trans in the sequence of PERMANENT transactions to
+ * the next: permanent transactions are only flushed out when
+ * committed with XFS_TRANS_RELEASE_LOG_RES, but we still want as soon
+ * as possible to let chunks of it go to the log. So we commit the
+ * chunk we've been working on and get a new transaction to continue.
+ */
+int
+xfs_trans_roll(
+	struct xfs_trans	**tpp,
+	struct xfs_inode	*dp)
+{
+	struct xfs_trans	*trans;
+	unsigned int		logres, count;
+	int			error;
+
+	/*
+	 * Ensure that the inode is always logged.
+	 */
+	trans = *tpp;
+	xfs_trans_log_inode(trans, dp, XFS_ILOG_CORE);
+
+	/*
+	 * Copy the critical parameters from one trans to the next.
+	 */
+	logres = trans->t_log_res;
+	count = trans->t_log_count;
+	*tpp = xfs_trans_dup(trans);
+
+	/*
+	 * Commit the current transaction.
+	 * If this commit failed, then it'd just unlock those items that
+	 * are not marked ihold. That also means that a filesystem shutdown
+	 * is in progress. The caller takes the responsibility to cancel
+	 * the duplicate transaction that gets returned.
+	 */
+	error = xfs_trans_commit(trans, 0);
+	if (error)
+		return (error);
+
+	trans = *tpp;
+
+	/*
+	 * transaction commit worked ok so we can drop the extra ticket
+	 * reference that we gained in xfs_trans_dup()
+	 */
+	xfs_log_ticket_put(trans->t_ticket);
+
+
+	/*
+	 * Reserve space in the log for th next transaction.
+	 * This also pushes items in the "AIL", the list of logged items,
+	 * out to disk if they are taking up space at the tail of the log
+	 * that we want to use.  This requires that either nothing be locked
+	 * across this call, or that anything that is locked be logged in
+	 * the prior and the next transactions.
+	 */
+	error = xfs_trans_reserve(trans, 0, logres, 0,
+				  XFS_TRANS_PERM_LOG_RES, count);
+	/*
+	 *  Ensure that the inode is in the new transaction and locked.
+	 */
+	if (error)
+		return error;
+
+	xfs_trans_ijoin(trans, dp);
+	return 0;
+}
diff --git a/fs/xfs/xfs_trans.h b/fs/xfs/xfs_trans.h
new file mode 100644
index 00000000..53597f4d
--- /dev/null
+++ b/fs/xfs/xfs_trans.h
@@ -0,0 +1,506 @@
+/*
+ * Copyright (c) 2000-2002,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef	__XFS_TRANS_H__
+#define	__XFS_TRANS_H__
+
+struct xfs_log_item;
+
+/*
+ * This is the structure written in the log at the head of
+ * every transaction. It identifies the type and id of the
+ * transaction, and contains the number of items logged by
+ * the transaction so we know how many to expect during recovery.
+ *
+ * Do not change the below structure without redoing the code in
+ * xlog_recover_add_to_trans() and xlog_recover_add_to_cont_trans().
+ */
+typedef struct xfs_trans_header {
+	uint		th_magic;		/* magic number */
+	uint		th_type;		/* transaction type */
+	__int32_t	th_tid;			/* transaction id (unused) */
+	uint		th_num_items;		/* num items logged by trans */
+} xfs_trans_header_t;
+
+#define	XFS_TRANS_HEADER_MAGIC	0x5452414e	/* TRAN */
+
+/*
+ * Log item types.
+ */
+#define	XFS_LI_EFI		0x1236
+#define	XFS_LI_EFD		0x1237
+#define	XFS_LI_IUNLINK		0x1238
+#define	XFS_LI_INODE		0x123b	/* aligned ino chunks, var-size ibufs */
+#define	XFS_LI_BUF		0x123c	/* v2 bufs, variable sized inode bufs */
+#define	XFS_LI_DQUOT		0x123d
+#define	XFS_LI_QUOTAOFF		0x123e
+
+#define XFS_LI_TYPE_DESC \
+	{ XFS_LI_EFI,		"XFS_LI_EFI" }, \
+	{ XFS_LI_EFD,		"XFS_LI_EFD" }, \
+	{ XFS_LI_IUNLINK,	"XFS_LI_IUNLINK" }, \
+	{ XFS_LI_INODE,		"XFS_LI_INODE" }, \
+	{ XFS_LI_BUF,		"XFS_LI_BUF" }, \
+	{ XFS_LI_DQUOT,		"XFS_LI_DQUOT" }, \
+	{ XFS_LI_QUOTAOFF,	"XFS_LI_QUOTAOFF" }
+
+/*
+ * Transaction types.  Used to distinguish types of buffers.
+ */
+#define XFS_TRANS_SETATTR_NOT_SIZE	1
+#define XFS_TRANS_SETATTR_SIZE		2
+#define XFS_TRANS_INACTIVE		3
+#define XFS_TRANS_CREATE		4
+#define XFS_TRANS_CREATE_TRUNC		5
+#define XFS_TRANS_TRUNCATE_FILE		6
+#define XFS_TRANS_REMOVE		7
+#define XFS_TRANS_LINK			8
+#define XFS_TRANS_RENAME		9
+#define XFS_TRANS_MKDIR			10
+#define XFS_TRANS_RMDIR			11
+#define XFS_TRANS_SYMLINK		12
+#define XFS_TRANS_SET_DMATTRS		13
+#define XFS_TRANS_GROWFS		14
+#define XFS_TRANS_STRAT_WRITE		15
+#define XFS_TRANS_DIOSTRAT		16
+/* 17 was XFS_TRANS_WRITE_SYNC */
+#define	XFS_TRANS_WRITEID		18
+#define	XFS_TRANS_ADDAFORK		19
+#define	XFS_TRANS_ATTRINVAL		20
+#define	XFS_TRANS_ATRUNCATE		21
+#define	XFS_TRANS_ATTR_SET		22
+#define	XFS_TRANS_ATTR_RM		23
+#define	XFS_TRANS_ATTR_FLAG		24
+#define	XFS_TRANS_CLEAR_AGI_BUCKET	25
+#define XFS_TRANS_QM_SBCHANGE		26
+/*
+ * Dummy entries since we use the transaction type to index into the
+ * trans_type[] in xlog_recover_print_trans_head()
+ */
+#define XFS_TRANS_DUMMY1		27
+#define XFS_TRANS_DUMMY2		28
+#define XFS_TRANS_QM_QUOTAOFF		29
+#define XFS_TRANS_QM_DQALLOC		30
+#define XFS_TRANS_QM_SETQLIM		31
+#define XFS_TRANS_QM_DQCLUSTER		32
+#define XFS_TRANS_QM_QINOCREATE		33
+#define XFS_TRANS_QM_QUOTAOFF_END	34
+#define XFS_TRANS_SB_UNIT		35
+#define XFS_TRANS_FSYNC_TS		36
+#define	XFS_TRANS_GROWFSRT_ALLOC	37
+#define	XFS_TRANS_GROWFSRT_ZERO		38
+#define	XFS_TRANS_GROWFSRT_FREE		39
+#define	XFS_TRANS_SWAPEXT		40
+#define	XFS_TRANS_SB_COUNT		41
+#define	XFS_TRANS_CHECKPOINT		42
+#define	XFS_TRANS_TYPE_MAX		42
+/* new transaction types need to be reflected in xfs_logprint(8) */
+
+#define XFS_TRANS_TYPES \
+	{ XFS_TRANS_SETATTR_NOT_SIZE,	"SETATTR_NOT_SIZE" }, \
+	{ XFS_TRANS_SETATTR_SIZE,	"SETATTR_SIZE" }, \
+	{ XFS_TRANS_INACTIVE,		"INACTIVE" }, \
+	{ XFS_TRANS_CREATE,		"CREATE" }, \
+	{ XFS_TRANS_CREATE_TRUNC,	"CREATE_TRUNC" }, \
+	{ XFS_TRANS_TRUNCATE_FILE,	"TRUNCATE_FILE" }, \
+	{ XFS_TRANS_REMOVE,		"REMOVE" }, \
+	{ XFS_TRANS_LINK,		"LINK" }, \
+	{ XFS_TRANS_RENAME,		"RENAME" }, \
+	{ XFS_TRANS_MKDIR,		"MKDIR" }, \
+	{ XFS_TRANS_RMDIR,		"RMDIR" }, \
+	{ XFS_TRANS_SYMLINK,		"SYMLINK" }, \
+	{ XFS_TRANS_SET_DMATTRS,	"SET_DMATTRS" }, \
+	{ XFS_TRANS_GROWFS,		"GROWFS" }, \
+	{ XFS_TRANS_STRAT_WRITE,	"STRAT_WRITE" }, \
+	{ XFS_TRANS_DIOSTRAT,		"DIOSTRAT" }, \
+	{ XFS_TRANS_WRITEID,		"WRITEID" }, \
+	{ XFS_TRANS_ADDAFORK,		"ADDAFORK" }, \
+	{ XFS_TRANS_ATTRINVAL,		"ATTRINVAL" }, \
+	{ XFS_TRANS_ATRUNCATE,		"ATRUNCATE" }, \
+	{ XFS_TRANS_ATTR_SET,		"ATTR_SET" }, \
+	{ XFS_TRANS_ATTR_RM,		"ATTR_RM" }, \
+	{ XFS_TRANS_ATTR_FLAG,		"ATTR_FLAG" }, \
+	{ XFS_TRANS_CLEAR_AGI_BUCKET,	"CLEAR_AGI_BUCKET" }, \
+	{ XFS_TRANS_QM_SBCHANGE,	"QM_SBCHANGE" }, \
+	{ XFS_TRANS_QM_QUOTAOFF,	"QM_QUOTAOFF" }, \
+	{ XFS_TRANS_QM_DQALLOC,		"QM_DQALLOC" }, \
+	{ XFS_TRANS_QM_SETQLIM,		"QM_SETQLIM" }, \
+	{ XFS_TRANS_QM_DQCLUSTER,	"QM_DQCLUSTER" }, \
+	{ XFS_TRANS_QM_QINOCREATE,	"QM_QINOCREATE" }, \
+	{ XFS_TRANS_QM_QUOTAOFF_END,	"QM_QOFF_END" }, \
+	{ XFS_TRANS_SB_UNIT,		"SB_UNIT" }, \
+	{ XFS_TRANS_FSYNC_TS,		"FSYNC_TS" }, \
+	{ XFS_TRANS_GROWFSRT_ALLOC,	"GROWFSRT_ALLOC" }, \
+	{ XFS_TRANS_GROWFSRT_ZERO,	"GROWFSRT_ZERO" }, \
+	{ XFS_TRANS_GROWFSRT_FREE,	"GROWFSRT_FREE" }, \
+	{ XFS_TRANS_SWAPEXT,		"SWAPEXT" }, \
+	{ XFS_TRANS_SB_COUNT,		"SB_COUNT" }, \
+	{ XFS_TRANS_CHECKPOINT,		"CHECKPOINT" }, \
+	{ XFS_TRANS_DUMMY1,		"DUMMY1" }, \
+	{ XFS_TRANS_DUMMY2,		"DUMMY2" }, \
+	{ XLOG_UNMOUNT_REC_TYPE,	"UNMOUNT" }
+
+/*
+ * This structure is used to track log items associated with
+ * a transaction.  It points to the log item and keeps some
+ * flags to track the state of the log item.  It also tracks
+ * the amount of space needed to log the item it describes
+ * once we get to commit processing (see xfs_trans_commit()).
+ */
+struct xfs_log_item_desc {
+	struct xfs_log_item	*lid_item;
+	ushort			lid_size;
+	unsigned char		lid_flags;
+	struct list_head	lid_trans;
+};
+
+#define XFS_LID_DIRTY		0x1
+
+#define	XFS_TRANS_MAGIC		0x5452414E	/* 'TRAN' */
+/*
+ * Values for t_flags.
+ */
+#define	XFS_TRANS_DIRTY		0x01	/* something needs to be logged */
+#define	XFS_TRANS_SB_DIRTY	0x02	/* superblock is modified */
+#define	XFS_TRANS_PERM_LOG_RES	0x04	/* xact took a permanent log res */
+#define	XFS_TRANS_SYNC		0x08	/* make commit synchronous */
+#define XFS_TRANS_DQ_DIRTY	0x10	/* at least one dquot in trx dirty */
+#define XFS_TRANS_RESERVE	0x20    /* OK to use reserved data blocks */
+
+/*
+ * Values for call flags parameter.
+ */
+#define	XFS_TRANS_RELEASE_LOG_RES	0x4
+#define	XFS_TRANS_ABORT			0x8
+
+/*
+ * Field values for xfs_trans_mod_sb.
+ */
+#define	XFS_TRANS_SB_ICOUNT		0x00000001
+#define	XFS_TRANS_SB_IFREE		0x00000002
+#define	XFS_TRANS_SB_FDBLOCKS		0x00000004
+#define	XFS_TRANS_SB_RES_FDBLOCKS	0x00000008
+#define	XFS_TRANS_SB_FREXTENTS		0x00000010
+#define	XFS_TRANS_SB_RES_FREXTENTS	0x00000020
+#define	XFS_TRANS_SB_DBLOCKS		0x00000040
+#define	XFS_TRANS_SB_AGCOUNT		0x00000080
+#define	XFS_TRANS_SB_IMAXPCT		0x00000100
+#define	XFS_TRANS_SB_REXTSIZE		0x00000200
+#define	XFS_TRANS_SB_RBMBLOCKS		0x00000400
+#define	XFS_TRANS_SB_RBLOCKS		0x00000800
+#define	XFS_TRANS_SB_REXTENTS		0x00001000
+#define	XFS_TRANS_SB_REXTSLOG		0x00002000
+
+
+/*
+ * Per-extent log reservation for the allocation btree changes
+ * involved in freeing or allocating an extent.
+ * 2 trees * (2 blocks/level * max depth - 1) * block size
+ */
+#define	XFS_ALLOCFREE_LOG_RES(mp,nx) \
+	((nx) * (2 * XFS_FSB_TO_B((mp), 2 * XFS_AG_MAXLEVELS(mp) - 1)))
+#define	XFS_ALLOCFREE_LOG_COUNT(mp,nx) \
+	((nx) * (2 * (2 * XFS_AG_MAXLEVELS(mp) - 1)))
+
+/*
+ * Per-directory log reservation for any directory change.
+ * dir blocks: (1 btree block per level + data block + free block) * dblock size
+ * bmap btree: (levels + 2) * max depth * block size
+ * v2 directory blocks can be fragmented below the dirblksize down to the fsb
+ * size, so account for that in the DAENTER macros.
+ */
+#define	XFS_DIROP_LOG_RES(mp)	\
+	(XFS_FSB_TO_B(mp, XFS_DAENTER_BLOCKS(mp, XFS_DATA_FORK)) + \
+	 (XFS_FSB_TO_B(mp, XFS_DAENTER_BMAPS(mp, XFS_DATA_FORK) + 1)))
+#define	XFS_DIROP_LOG_COUNT(mp)	\
+	(XFS_DAENTER_BLOCKS(mp, XFS_DATA_FORK) + \
+	 XFS_DAENTER_BMAPS(mp, XFS_DATA_FORK) + 1)
+
+
+#define	XFS_WRITE_LOG_RES(mp)	((mp)->m_reservations.tr_write)
+#define	XFS_ITRUNCATE_LOG_RES(mp)   ((mp)->m_reservations.tr_itruncate)
+#define	XFS_RENAME_LOG_RES(mp)	((mp)->m_reservations.tr_rename)
+#define	XFS_LINK_LOG_RES(mp)	((mp)->m_reservations.tr_link)
+#define	XFS_REMOVE_LOG_RES(mp)	((mp)->m_reservations.tr_remove)
+#define	XFS_SYMLINK_LOG_RES(mp)	((mp)->m_reservations.tr_symlink)
+#define	XFS_CREATE_LOG_RES(mp)	((mp)->m_reservations.tr_create)
+#define	XFS_MKDIR_LOG_RES(mp)	((mp)->m_reservations.tr_mkdir)
+#define	XFS_IFREE_LOG_RES(mp)	((mp)->m_reservations.tr_ifree)
+#define	XFS_ICHANGE_LOG_RES(mp)	((mp)->m_reservations.tr_ichange)
+#define	XFS_GROWDATA_LOG_RES(mp)    ((mp)->m_reservations.tr_growdata)
+#define	XFS_GROWRTALLOC_LOG_RES(mp)	((mp)->m_reservations.tr_growrtalloc)
+#define	XFS_GROWRTZERO_LOG_RES(mp)	((mp)->m_reservations.tr_growrtzero)
+#define	XFS_GROWRTFREE_LOG_RES(mp)	((mp)->m_reservations.tr_growrtfree)
+#define	XFS_SWRITE_LOG_RES(mp)	((mp)->m_reservations.tr_swrite)
+/*
+ * Logging the inode timestamps on an fsync -- same as SWRITE
+ * as long as SWRITE logs the entire inode core
+ */
+#define XFS_FSYNC_TS_LOG_RES(mp)        ((mp)->m_reservations.tr_swrite)
+#define	XFS_WRITEID_LOG_RES(mp)	((mp)->m_reservations.tr_swrite)
+#define	XFS_ADDAFORK_LOG_RES(mp)	((mp)->m_reservations.tr_addafork)
+#define	XFS_ATTRINVAL_LOG_RES(mp)	((mp)->m_reservations.tr_attrinval)
+#define	XFS_ATTRSET_LOG_RES(mp, ext)	\
+	((mp)->m_reservations.tr_attrset + \
+	 (ext * (mp)->m_sb.sb_sectsize) + \
+	 (ext * XFS_FSB_TO_B((mp), XFS_BM_MAXLEVELS(mp, XFS_ATTR_FORK))) + \
+	 (128 * (ext + (ext * XFS_BM_MAXLEVELS(mp, XFS_ATTR_FORK)))))
+#define	XFS_ATTRRM_LOG_RES(mp)	((mp)->m_reservations.tr_attrrm)
+#define	XFS_CLEAR_AGI_BUCKET_LOG_RES(mp)  ((mp)->m_reservations.tr_clearagi)
+
+
+/*
+ * Various log count values.
+ */
+#define	XFS_DEFAULT_LOG_COUNT		1
+#define	XFS_DEFAULT_PERM_LOG_COUNT	2
+#define	XFS_ITRUNCATE_LOG_COUNT		2
+#define XFS_INACTIVE_LOG_COUNT		2
+#define	XFS_CREATE_LOG_COUNT		2
+#define	XFS_MKDIR_LOG_COUNT		3
+#define	XFS_SYMLINK_LOG_COUNT		3
+#define	XFS_REMOVE_LOG_COUNT		2
+#define	XFS_LINK_LOG_COUNT		2
+#define	XFS_RENAME_LOG_COUNT		2
+#define	XFS_WRITE_LOG_COUNT		2
+#define	XFS_ADDAFORK_LOG_COUNT		2
+#define	XFS_ATTRINVAL_LOG_COUNT		1
+#define	XFS_ATTRSET_LOG_COUNT		3
+#define	XFS_ATTRRM_LOG_COUNT		3
+
+/*
+ * Here we centralize the specification of XFS meta-data buffer
+ * reference count values.  This determine how hard the buffer
+ * cache tries to hold onto the buffer.
+ */
+#define	XFS_AGF_REF		4
+#define	XFS_AGI_REF		4
+#define	XFS_AGFL_REF		3
+#define	XFS_INO_BTREE_REF	3
+#define	XFS_ALLOC_BTREE_REF	2
+#define	XFS_BMAP_BTREE_REF	2
+#define	XFS_DIR_BTREE_REF	2
+#define	XFS_INO_REF		2
+#define	XFS_ATTR_BTREE_REF	1
+#define	XFS_DQUOT_REF		1
+
+#ifdef __KERNEL__
+
+struct xfs_buf;
+struct xfs_buftarg;
+struct xfs_efd_log_item;
+struct xfs_efi_log_item;
+struct xfs_inode;
+struct xfs_item_ops;
+struct xfs_log_iovec;
+struct xfs_log_item_desc;
+struct xfs_mount;
+struct xfs_trans;
+struct xfs_dquot_acct;
+struct xfs_busy_extent;
+
+typedef struct xfs_log_item {
+	struct list_head		li_ail;		/* AIL pointers */
+	xfs_lsn_t			li_lsn;		/* last on-disk lsn */
+	struct xfs_log_item_desc	*li_desc;	/* ptr to current desc*/
+	struct xfs_mount		*li_mountp;	/* ptr to fs mount */
+	struct xfs_ail			*li_ailp;	/* ptr to AIL */
+	uint				li_type;	/* item type */
+	uint				li_flags;	/* misc flags */
+	struct xfs_log_item		*li_bio_list;	/* buffer item list */
+	void				(*li_cb)(struct xfs_buf *,
+						 struct xfs_log_item *);
+							/* buffer item iodone */
+							/* callback func */
+	struct xfs_item_ops		*li_ops;	/* function list */
+
+	/* delayed logging */
+	struct list_head		li_cil;		/* CIL pointers */
+	struct xfs_log_vec		*li_lv;		/* active log vector */
+	xfs_lsn_t			li_seq;		/* CIL commit seq */
+} xfs_log_item_t;
+
+#define	XFS_LI_IN_AIL	0x1
+#define XFS_LI_ABORTED	0x2
+
+#define XFS_LI_FLAGS \
+	{ XFS_LI_IN_AIL,	"IN_AIL" }, \
+	{ XFS_LI_ABORTED,	"ABORTED" }
+
+typedef struct xfs_item_ops {
+	uint (*iop_size)(xfs_log_item_t *);
+	void (*iop_format)(xfs_log_item_t *, struct xfs_log_iovec *);
+	void (*iop_pin)(xfs_log_item_t *);
+	void (*iop_unpin)(xfs_log_item_t *, int remove);
+	uint (*iop_trylock)(xfs_log_item_t *);
+	void (*iop_unlock)(xfs_log_item_t *);
+	xfs_lsn_t (*iop_committed)(xfs_log_item_t *, xfs_lsn_t);
+	void (*iop_push)(xfs_log_item_t *);
+	bool (*iop_pushbuf)(xfs_log_item_t *);
+	void (*iop_committing)(xfs_log_item_t *, xfs_lsn_t);
+} xfs_item_ops_t;
+
+#define IOP_SIZE(ip)		(*(ip)->li_ops->iop_size)(ip)
+#define IOP_FORMAT(ip,vp)	(*(ip)->li_ops->iop_format)(ip, vp)
+#define IOP_PIN(ip)		(*(ip)->li_ops->iop_pin)(ip)
+#define IOP_UNPIN(ip, remove)	(*(ip)->li_ops->iop_unpin)(ip, remove)
+#define IOP_TRYLOCK(ip)		(*(ip)->li_ops->iop_trylock)(ip)
+#define IOP_UNLOCK(ip)		(*(ip)->li_ops->iop_unlock)(ip)
+#define IOP_COMMITTED(ip, lsn)	(*(ip)->li_ops->iop_committed)(ip, lsn)
+#define IOP_PUSH(ip)		(*(ip)->li_ops->iop_push)(ip)
+#define IOP_PUSHBUF(ip)		(*(ip)->li_ops->iop_pushbuf)(ip)
+#define IOP_COMMITTING(ip, lsn) (*(ip)->li_ops->iop_committing)(ip, lsn)
+
+/*
+ * Return values for the IOP_TRYLOCK() routines.
+ */
+#define	XFS_ITEM_SUCCESS	0
+#define	XFS_ITEM_PINNED		1
+#define	XFS_ITEM_LOCKED		2
+#define XFS_ITEM_PUSHBUF	3
+
+/*
+ * This is the type of function which can be given to xfs_trans_callback()
+ * to be called upon the transaction's commit to disk.
+ */
+typedef void (*xfs_trans_callback_t)(struct xfs_trans *, void *);
+
+/*
+ * This is the structure maintained for every active transaction.
+ */
+typedef struct xfs_trans {
+	unsigned int		t_magic;	/* magic number */
+	xfs_log_callback_t	t_logcb;	/* log callback struct */
+	unsigned int		t_type;		/* transaction type */
+	unsigned int		t_log_res;	/* amt of log space resvd */
+	unsigned int		t_log_count;	/* count for perm log res */
+	unsigned int		t_blk_res;	/* # of blocks resvd */
+	unsigned int		t_blk_res_used;	/* # of resvd blocks used */
+	unsigned int		t_rtx_res;	/* # of rt extents resvd */
+	unsigned int		t_rtx_res_used;	/* # of resvd rt extents used */
+	struct xlog_ticket	*t_ticket;	/* log mgr ticket */
+	xfs_lsn_t		t_lsn;		/* log seq num of start of
+						 * transaction. */
+	xfs_lsn_t		t_commit_lsn;	/* log seq num of end of
+						 * transaction. */
+	struct xfs_mount	*t_mountp;	/* ptr to fs mount struct */
+	struct xfs_dquot_acct   *t_dqinfo;	/* acctg info for dquots */
+	unsigned int		t_flags;	/* misc flags */
+	int64_t			t_icount_delta;	/* superblock icount change */
+	int64_t			t_ifree_delta;	/* superblock ifree change */
+	int64_t			t_fdblocks_delta; /* superblock fdblocks chg */
+	int64_t			t_res_fdblocks_delta; /* on-disk only chg */
+	int64_t			t_frextents_delta;/* superblock freextents chg*/
+	int64_t			t_res_frextents_delta; /* on-disk only chg */
+#ifdef DEBUG
+	int64_t			t_ag_freeblks_delta; /* debugging counter */
+	int64_t			t_ag_flist_delta; /* debugging counter */
+	int64_t			t_ag_btree_delta; /* debugging counter */
+#endif
+	int64_t			t_dblocks_delta;/* superblock dblocks change */
+	int64_t			t_agcount_delta;/* superblock agcount change */
+	int64_t			t_imaxpct_delta;/* superblock imaxpct change */
+	int64_t			t_rextsize_delta;/* superblock rextsize chg */
+	int64_t			t_rbmblocks_delta;/* superblock rbmblocks chg */
+	int64_t			t_rblocks_delta;/* superblock rblocks change */
+	int64_t			t_rextents_delta;/* superblocks rextents chg */
+	int64_t			t_rextslog_delta;/* superblocks rextslog chg */
+	struct list_head	t_items;	/* log item descriptors */
+	xfs_trans_header_t	t_header;	/* header for in-log trans */
+	struct list_head	t_busy;		/* list of busy extents */
+	unsigned long		t_pflags;	/* saved process flags state */
+} xfs_trans_t;
+
+/*
+ * XFS transaction mechanism exported interfaces that are
+ * actually macros.
+ */
+#define	xfs_trans_get_log_res(tp)	((tp)->t_log_res)
+#define	xfs_trans_get_log_count(tp)	((tp)->t_log_count)
+#define	xfs_trans_get_block_res(tp)	((tp)->t_blk_res)
+#define	xfs_trans_set_sync(tp)		((tp)->t_flags |= XFS_TRANS_SYNC)
+
+#ifdef DEBUG
+#define	xfs_trans_agblocks_delta(tp, d)	((tp)->t_ag_freeblks_delta += (int64_t)d)
+#define	xfs_trans_agflist_delta(tp, d)	((tp)->t_ag_flist_delta += (int64_t)d)
+#define	xfs_trans_agbtree_delta(tp, d)	((tp)->t_ag_btree_delta += (int64_t)d)
+#else
+#define	xfs_trans_agblocks_delta(tp, d)
+#define	xfs_trans_agflist_delta(tp, d)
+#define	xfs_trans_agbtree_delta(tp, d)
+#endif
+
+/*
+ * XFS transaction mechanism exported interfaces.
+ */
+xfs_trans_t	*xfs_trans_alloc(struct xfs_mount *, uint);
+xfs_trans_t	*_xfs_trans_alloc(struct xfs_mount *, uint, uint);
+xfs_trans_t	*xfs_trans_dup(xfs_trans_t *);
+int		xfs_trans_reserve(xfs_trans_t *, uint, uint, uint,
+				  uint, uint);
+void		xfs_trans_mod_sb(xfs_trans_t *, uint, int64_t);
+struct xfs_buf	*xfs_trans_get_buf(xfs_trans_t *, struct xfs_buftarg *, xfs_daddr_t,
+				   int, uint);
+int		xfs_trans_read_buf(struct xfs_mount *, xfs_trans_t *,
+				   struct xfs_buftarg *, xfs_daddr_t, int, uint,
+				   struct xfs_buf **);
+struct xfs_buf	*xfs_trans_getsb(xfs_trans_t *, struct xfs_mount *, int);
+
+void		xfs_trans_brelse(xfs_trans_t *, struct xfs_buf *);
+void		xfs_trans_bjoin(xfs_trans_t *, struct xfs_buf *);
+void		xfs_trans_bhold(xfs_trans_t *, struct xfs_buf *);
+void		xfs_trans_bhold_release(xfs_trans_t *, struct xfs_buf *);
+void		xfs_trans_binval(xfs_trans_t *, struct xfs_buf *);
+void		xfs_trans_inode_buf(xfs_trans_t *, struct xfs_buf *);
+void		xfs_trans_stale_inode_buf(xfs_trans_t *, struct xfs_buf *);
+void		xfs_trans_dquot_buf(xfs_trans_t *, struct xfs_buf *, uint);
+void		xfs_trans_inode_alloc_buf(xfs_trans_t *, struct xfs_buf *);
+void		xfs_trans_ichgtime(struct xfs_trans *, struct xfs_inode *, int);
+void		xfs_trans_ijoin_ref(struct xfs_trans *, struct xfs_inode *, uint);
+void		xfs_trans_ijoin(struct xfs_trans *, struct xfs_inode *);
+void		xfs_trans_log_buf(xfs_trans_t *, struct xfs_buf *, uint, uint);
+void		xfs_trans_log_inode(xfs_trans_t *, struct xfs_inode *, uint);
+struct xfs_efi_log_item	*xfs_trans_get_efi(xfs_trans_t *, uint);
+void		xfs_efi_release(struct xfs_efi_log_item *, uint);
+void		xfs_trans_log_efi_extent(xfs_trans_t *,
+					 struct xfs_efi_log_item *,
+					 xfs_fsblock_t,
+					 xfs_extlen_t);
+struct xfs_efd_log_item	*xfs_trans_get_efd(xfs_trans_t *,
+				  struct xfs_efi_log_item *,
+				  uint);
+void		xfs_trans_log_efd_extent(xfs_trans_t *,
+					 struct xfs_efd_log_item *,
+					 xfs_fsblock_t,
+					 xfs_extlen_t);
+int		_xfs_trans_commit(xfs_trans_t *,
+				  uint flags,
+				  int *);
+#define xfs_trans_commit(tp, flags)	_xfs_trans_commit(tp, flags, NULL)
+void		xfs_trans_cancel(xfs_trans_t *, int);
+int		xfs_trans_ail_init(struct xfs_mount *);
+void		xfs_trans_ail_destroy(struct xfs_mount *);
+
+extern kmem_zone_t	*xfs_trans_zone;
+extern kmem_zone_t	*xfs_log_item_desc_zone;
+
+#endif	/* __KERNEL__ */
+
+void		xfs_trans_init(struct xfs_mount *);
+int		xfs_trans_roll(struct xfs_trans **, struct xfs_inode *);
+
+#endif	/* __XFS_TRANS_H__ */
diff --git a/fs/xfs/xfs_trans_ail.c b/fs/xfs/xfs_trans_ail.c
new file mode 100644
index 00000000..a4c281bf
--- /dev/null
+++ b/fs/xfs/xfs_trans_ail.c
@@ -0,0 +1,890 @@
+/*
+ * Copyright (c) 2000-2002,2005 Silicon Graphics, Inc.
+ * Copyright (c) 2008 Dave Chinner
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_mount.h"
+#include "xfs_trans_priv.h"
+#include "xfs_error.h"
+
+#ifdef DEBUG
+/*
+ * Check that the list is sorted as it should be.
+ */
+STATIC void
+xfs_ail_check(
+	struct xfs_ail	*ailp,
+	xfs_log_item_t	*lip)
+{
+	xfs_log_item_t	*prev_lip;
+
+	if (list_empty(&ailp->xa_ail))
+		return;
+
+	/*
+	 * Check the next and previous entries are valid.
+	 */
+	ASSERT((lip->li_flags & XFS_LI_IN_AIL) != 0);
+	prev_lip = list_entry(lip->li_ail.prev, xfs_log_item_t, li_ail);
+	if (&prev_lip->li_ail != &ailp->xa_ail)
+		ASSERT(XFS_LSN_CMP(prev_lip->li_lsn, lip->li_lsn) <= 0);
+
+	prev_lip = list_entry(lip->li_ail.next, xfs_log_item_t, li_ail);
+	if (&prev_lip->li_ail != &ailp->xa_ail)
+		ASSERT(XFS_LSN_CMP(prev_lip->li_lsn, lip->li_lsn) >= 0);
+
+
+#ifdef XFS_TRANS_DEBUG
+	/*
+	 * Walk the list checking lsn ordering, and that every entry has the
+	 * XFS_LI_IN_AIL flag set. This is really expensive, so only do it
+	 * when specifically debugging the transaction subsystem.
+	 */
+	prev_lip = list_entry(&ailp->xa_ail, xfs_log_item_t, li_ail);
+	list_for_each_entry(lip, &ailp->xa_ail, li_ail) {
+		if (&prev_lip->li_ail != &ailp->xa_ail)
+			ASSERT(XFS_LSN_CMP(prev_lip->li_lsn, lip->li_lsn) <= 0);
+		ASSERT((lip->li_flags & XFS_LI_IN_AIL) != 0);
+		prev_lip = lip;
+	}
+#endif /* XFS_TRANS_DEBUG */
+}
+#else /* !DEBUG */
+#define	xfs_ail_check(a,l)
+#endif /* DEBUG */
+
+/*
+ * Return a pointer to the first item in the AIL.  If the AIL is empty, then
+ * return NULL.
+ */
+static xfs_log_item_t *
+xfs_ail_min(
+	struct xfs_ail  *ailp)
+{
+	if (list_empty(&ailp->xa_ail))
+		return NULL;
+
+	return list_first_entry(&ailp->xa_ail, xfs_log_item_t, li_ail);
+}
+
+ /*
+ * Return a pointer to the last item in the AIL.  If the AIL is empty, then
+ * return NULL.
+ */
+static xfs_log_item_t *
+xfs_ail_max(
+	struct xfs_ail  *ailp)
+{
+	if (list_empty(&ailp->xa_ail))
+		return NULL;
+
+	return list_entry(ailp->xa_ail.prev, xfs_log_item_t, li_ail);
+}
+
+/*
+ * Return a pointer to the item which follows the given item in the AIL.  If
+ * the given item is the last item in the list, then return NULL.
+ */
+static xfs_log_item_t *
+xfs_ail_next(
+	struct xfs_ail  *ailp,
+	xfs_log_item_t  *lip)
+{
+	if (lip->li_ail.next == &ailp->xa_ail)
+		return NULL;
+
+	return list_first_entry(&lip->li_ail, xfs_log_item_t, li_ail);
+}
+
+/*
+ * This is called by the log manager code to determine the LSN of the tail of
+ * the log.  This is exactly the LSN of the first item in the AIL.  If the AIL
+ * is empty, then this function returns 0.
+ *
+ * We need the AIL lock in order to get a coherent read of the lsn of the last
+ * item in the AIL.
+ */
+xfs_lsn_t
+xfs_ail_min_lsn(
+	struct xfs_ail	*ailp)
+{
+	xfs_lsn_t	lsn = 0;
+	xfs_log_item_t	*lip;
+
+	spin_lock(&ailp->xa_lock);
+	lip = xfs_ail_min(ailp);
+	if (lip)
+		lsn = lip->li_lsn;
+	spin_unlock(&ailp->xa_lock);
+
+	return lsn;
+}
+
+/*
+ * Return the maximum lsn held in the AIL, or zero if the AIL is empty.
+ */
+static xfs_lsn_t
+xfs_ail_max_lsn(
+	struct xfs_ail  *ailp)
+{
+	xfs_lsn_t       lsn = 0;
+	xfs_log_item_t  *lip;
+
+	spin_lock(&ailp->xa_lock);
+	lip = xfs_ail_max(ailp);
+	if (lip)
+		lsn = lip->li_lsn;
+	spin_unlock(&ailp->xa_lock);
+
+	return lsn;
+}
+
+/*
+ * AIL traversal cursor initialisation.
+ *
+ * The cursor keeps track of where our current traversal is up
+ * to by tracking the next ƣtem in the list for us. However, for
+ * this to be safe, removing an object from the AIL needs to invalidate
+ * any cursor that points to it. hence the traversal cursor needs to
+ * be linked to the struct xfs_ail so that deletion can search all the
+ * active cursors for invalidation.
+ *
+ * We don't link the push cursor because it is embedded in the struct
+ * xfs_ail and hence easily findable.
+ */
+STATIC void
+xfs_trans_ail_cursor_init(
+	struct xfs_ail		*ailp,
+	struct xfs_ail_cursor	*cur)
+{
+	cur->item = NULL;
+	if (cur == &ailp->xa_cursors)
+		return;
+
+	cur->next = ailp->xa_cursors.next;
+	ailp->xa_cursors.next = cur;
+}
+
+/*
+ * Set the cursor to the next item, because when we look
+ * up the cursor the current item may have been freed.
+ */
+STATIC void
+xfs_trans_ail_cursor_set(
+	struct xfs_ail		*ailp,
+	struct xfs_ail_cursor	*cur,
+	struct xfs_log_item	*lip)
+{
+	if (lip)
+		cur->item = xfs_ail_next(ailp, lip);
+}
+
+/*
+ * Get the next item in the traversal and advance the cursor.
+ * If the cursor was invalidated (inidicated by a lip of 1),
+ * restart the traversal.
+ */
+struct xfs_log_item *
+xfs_trans_ail_cursor_next(
+	struct xfs_ail		*ailp,
+	struct xfs_ail_cursor	*cur)
+{
+	struct xfs_log_item	*lip = cur->item;
+
+	if ((__psint_t)lip & 1)
+		lip = xfs_ail_min(ailp);
+	xfs_trans_ail_cursor_set(ailp, cur, lip);
+	return lip;
+}
+
+/*
+ * Now that the traversal is complete, we need to remove the cursor
+ * from the list of traversing cursors. Avoid removing the embedded
+ * push cursor, but use the fact it is always present to make the
+ * list deletion simple.
+ */
+void
+xfs_trans_ail_cursor_done(
+	struct xfs_ail		*ailp,
+	struct xfs_ail_cursor	*done)
+{
+	struct xfs_ail_cursor	*prev = NULL;
+	struct xfs_ail_cursor	*cur;
+
+	done->item = NULL;
+	if (done == &ailp->xa_cursors)
+		return;
+	prev = &ailp->xa_cursors;
+	for (cur = prev->next; cur; prev = cur, cur = prev->next) {
+		if (cur == done) {
+			prev->next = cur->next;
+			break;
+		}
+	}
+	ASSERT(cur);
+}
+
+/*
+ * Invalidate any cursor that is pointing to this item. This is
+ * called when an item is removed from the AIL. Any cursor pointing
+ * to this object is now invalid and the traversal needs to be
+ * terminated so it doesn't reference a freed object. We set the
+ * cursor item to a value of 1 so we can distinguish between an
+ * invalidation and the end of the list when getting the next item
+ * from the cursor.
+ */
+STATIC void
+xfs_trans_ail_cursor_clear(
+	struct xfs_ail		*ailp,
+	struct xfs_log_item	*lip)
+{
+	struct xfs_ail_cursor	*cur;
+
+	/* need to search all cursors */
+	for (cur = &ailp->xa_cursors; cur; cur = cur->next) {
+		if (cur->item == lip)
+			cur->item = (struct xfs_log_item *)
+					((__psint_t)cur->item | 1);
+	}
+}
+
+/*
+ * Initialise the cursor to the first item in the AIL with the given @lsn.
+ * This searches the list from lowest LSN to highest. Pass a @lsn of zero
+ * to initialise the cursor to the first item in the AIL.
+ */
+xfs_log_item_t *
+xfs_trans_ail_cursor_first(
+	struct xfs_ail		*ailp,
+	struct xfs_ail_cursor	*cur,
+	xfs_lsn_t		lsn)
+{
+	xfs_log_item_t		*lip;
+
+	xfs_trans_ail_cursor_init(ailp, cur);
+	lip = xfs_ail_min(ailp);
+	if (lsn == 0)
+		goto out;
+
+	list_for_each_entry(lip, &ailp->xa_ail, li_ail) {
+		if (XFS_LSN_CMP(lip->li_lsn, lsn) >= 0)
+			goto out;
+	}
+	lip = NULL;
+out:
+	xfs_trans_ail_cursor_set(ailp, cur, lip);
+	return lip;
+}
+
+/*
+ * Initialise the cursor to the last item in the AIL with the given @lsn.
+ * This searches the list from highest LSN to lowest. If there is no item with
+ * the value of @lsn, then it sets the cursor to the last item with an LSN lower
+ * than @lsn.
+ */
+static struct xfs_log_item *
+__xfs_trans_ail_cursor_last(
+	struct xfs_ail		*ailp,
+	xfs_lsn_t		lsn)
+{
+	xfs_log_item_t		*lip;
+
+	list_for_each_entry_reverse(lip, &ailp->xa_ail, li_ail) {
+		if (XFS_LSN_CMP(lip->li_lsn, lsn) <= 0)
+			return lip;
+	}
+	return NULL;
+}
+
+/*
+ * Initialise the cursor to the last item in the AIL with the given @lsn.
+ * This searches the list from highest LSN to lowest.
+ */
+struct xfs_log_item *
+xfs_trans_ail_cursor_last(
+	struct xfs_ail		*ailp,
+	struct xfs_ail_cursor	*cur,
+	xfs_lsn_t		lsn)
+{
+	xfs_trans_ail_cursor_init(ailp, cur);
+	cur->item = __xfs_trans_ail_cursor_last(ailp, lsn);
+	return cur->item;
+}
+
+/*
+ * splice the log item list into the AIL at the given LSN. We splice to the
+ * tail of the given LSN to maintain insert order for push traversals. The
+ * cursor is optional, allowing repeated updates to the same LSN to avoid
+ * repeated traversals.
+ */
+static void
+xfs_ail_splice(
+	struct xfs_ail		*ailp,
+	struct xfs_ail_cursor	*cur,
+	struct list_head	*list,
+	xfs_lsn_t		lsn)
+{
+	struct xfs_log_item	*lip = cur ? cur->item : NULL;
+	struct xfs_log_item	*next_lip;
+
+	/*
+	 * Get a new cursor if we don't have a placeholder or the existing one
+	 * has been invalidated.
+	 */
+	if (!lip || (__psint_t)lip & 1) {
+		lip = __xfs_trans_ail_cursor_last(ailp, lsn);
+
+		if (!lip) {
+			/* The list is empty, so just splice and return.  */
+			if (cur)
+				cur->item = NULL;
+			list_splice(list, &ailp->xa_ail);
+			return;
+		}
+	}
+
+	/*
+	 * Our cursor points to the item we want to insert _after_, so we have
+	 * to update the cursor to point to the end of the list we are splicing
+	 * in so that it points to the correct location for the next splice.
+	 * i.e. before the splice
+	 *
+	 *  lsn -> lsn -> lsn + x -> lsn + x ...
+	 *          ^
+	 *          | cursor points here
+	 *
+	 * After the splice we have:
+	 *
+	 *  lsn -> lsn -> lsn -> lsn -> .... -> lsn -> lsn + x -> lsn + x ...
+	 *          ^                            ^
+	 *          | cursor points here         | needs to move here
+	 *
+	 * So we set the cursor to the last item in the list to be spliced
+	 * before we execute the splice, resulting in the cursor pointing to
+	 * the correct item after the splice occurs.
+	 */
+	if (cur) {
+		next_lip = list_entry(list->prev, struct xfs_log_item, li_ail);
+		cur->item = next_lip;
+	}
+	list_splice(list, &lip->li_ail);
+}
+
+/*
+ * Delete the given item from the AIL.  Return a pointer to the item.
+ */
+static void
+xfs_ail_delete(
+	struct xfs_ail  *ailp,
+	xfs_log_item_t  *lip)
+{
+	xfs_ail_check(ailp, lip);
+	list_del(&lip->li_ail);
+	xfs_trans_ail_cursor_clear(ailp, lip);
+}
+
+static long
+xfsaild_push(
+	struct xfs_ail		*ailp)
+{
+	xfs_mount_t		*mp = ailp->xa_mount;
+	struct xfs_ail_cursor	*cur = &ailp->xa_cursors;
+	xfs_log_item_t		*lip;
+	xfs_lsn_t		lsn;
+	xfs_lsn_t		target;
+	long			tout = 10;
+	int			flush_log = 0;
+	int			stuck = 0;
+	int			count = 0;
+	int			push_xfsbufd = 0;
+
+	spin_lock(&ailp->xa_lock);
+	target = ailp->xa_target;
+	xfs_trans_ail_cursor_init(ailp, cur);
+	lip = xfs_trans_ail_cursor_first(ailp, cur, ailp->xa_last_pushed_lsn);
+	if (!lip || XFS_FORCED_SHUTDOWN(mp)) {
+		/*
+		 * AIL is empty or our push has reached the end.
+		 */
+		xfs_trans_ail_cursor_done(ailp, cur);
+		spin_unlock(&ailp->xa_lock);
+		goto out_done;
+	}
+
+	XFS_STATS_INC(xs_push_ail);
+
+	/*
+	 * While the item we are looking at is below the given threshold
+	 * try to flush it out. We'd like not to stop until we've at least
+	 * tried to push on everything in the AIL with an LSN less than
+	 * the given threshold.
+	 *
+	 * However, we will stop after a certain number of pushes and wait
+	 * for a reduced timeout to fire before pushing further. This
+	 * prevents use from spinning when we can't do anything or there is
+	 * lots of contention on the AIL lists.
+	 */
+	lsn = lip->li_lsn;
+	while ((XFS_LSN_CMP(lip->li_lsn, target) <= 0)) {
+		int	lock_result;
+		/*
+		 * If we can lock the item without sleeping, unlock the AIL
+		 * lock and flush the item.  Then re-grab the AIL lock so we
+		 * can look for the next item on the AIL. List changes are
+		 * handled by the AIL lookup functions internally
+		 *
+		 * If we can't lock the item, either its holder will flush it
+		 * or it is already being flushed or it is being relogged.  In
+		 * any of these case it is being taken care of and we can just
+		 * skip to the next item in the list.
+		 */
+		lock_result = IOP_TRYLOCK(lip);
+		spin_unlock(&ailp->xa_lock);
+		switch (lock_result) {
+		case XFS_ITEM_SUCCESS:
+			XFS_STATS_INC(xs_push_ail_success);
+			IOP_PUSH(lip);
+			ailp->xa_last_pushed_lsn = lsn;
+			break;
+
+		case XFS_ITEM_PUSHBUF:
+			XFS_STATS_INC(xs_push_ail_pushbuf);
+
+			if (!IOP_PUSHBUF(lip)) {
+				stuck++;
+				flush_log = 1;
+			} else {
+				ailp->xa_last_pushed_lsn = lsn;
+			}
+			push_xfsbufd = 1;
+			break;
+
+		case XFS_ITEM_PINNED:
+			XFS_STATS_INC(xs_push_ail_pinned);
+			stuck++;
+			flush_log = 1;
+			break;
+
+		case XFS_ITEM_LOCKED:
+			XFS_STATS_INC(xs_push_ail_locked);
+			stuck++;
+			break;
+
+		default:
+			ASSERT(0);
+			break;
+		}
+
+		spin_lock(&ailp->xa_lock);
+		/* should we bother continuing? */
+		if (XFS_FORCED_SHUTDOWN(mp))
+			break;
+		ASSERT(mp->m_log);
+
+		count++;
+
+		/*
+		 * Are there too many items we can't do anything with?
+		 * If we we are skipping too many items because we can't flush
+		 * them or they are already being flushed, we back off and
+		 * given them time to complete whatever operation is being
+		 * done. i.e. remove pressure from the AIL while we can't make
+		 * progress so traversals don't slow down further inserts and
+		 * removals to/from the AIL.
+		 *
+		 * The value of 100 is an arbitrary magic number based on
+		 * observation.
+		 */
+		if (stuck > 100)
+			break;
+
+		lip = xfs_trans_ail_cursor_next(ailp, cur);
+		if (lip == NULL)
+			break;
+		lsn = lip->li_lsn;
+	}
+	xfs_trans_ail_cursor_done(ailp, cur);
+	spin_unlock(&ailp->xa_lock);
+
+	if (flush_log) {
+		/*
+		 * If something we need to push out was pinned, then
+		 * push out the log so it will become unpinned and
+		 * move forward in the AIL.
+		 */
+		XFS_STATS_INC(xs_push_ail_flush);
+		xfs_log_force(mp, 0);
+	}
+
+	if (push_xfsbufd) {
+		/* we've got delayed write buffers to flush */
+		wake_up_process(mp->m_ddev_targp->bt_task);
+	}
+
+	/* assume we have more work to do in a short while */
+out_done:
+	if (!count) {
+		/* We're past our target or empty, so idle */
+		ailp->xa_last_pushed_lsn = 0;
+
+		tout = 50;
+	} else if (XFS_LSN_CMP(lsn, target) >= 0) {
+		/*
+		 * We reached the target so wait a bit longer for I/O to
+		 * complete and remove pushed items from the AIL before we
+		 * start the next scan from the start of the AIL.
+		 */
+		tout = 50;
+		ailp->xa_last_pushed_lsn = 0;
+	} else if ((stuck * 100) / count > 90) {
+		/*
+		 * Either there is a lot of contention on the AIL or we
+		 * are stuck due to operations in progress. "Stuck" in this
+		 * case is defined as >90% of the items we tried to push
+		 * were stuck.
+		 *
+		 * Backoff a bit more to allow some I/O to complete before
+		 * continuing from where we were.
+		 */
+		tout = 20;
+	}
+
+	return tout;
+}
+
+static int
+xfsaild(
+	void		*data)
+{
+	struct xfs_ail	*ailp = data;
+	long		tout = 0;	/* milliseconds */
+
+	while (!kthread_should_stop()) {
+		if (tout && tout <= 20)
+			__set_current_state(TASK_KILLABLE);
+		else
+			__set_current_state(TASK_INTERRUPTIBLE);
+		schedule_timeout(tout ?
+				 msecs_to_jiffies(tout) : MAX_SCHEDULE_TIMEOUT);
+
+		try_to_freeze();
+
+		tout = xfsaild_push(ailp);
+	}
+
+	return 0;
+}
+
+/*
+ * This routine is called to move the tail of the AIL forward.  It does this by
+ * trying to flush items in the AIL whose lsns are below the given
+ * threshold_lsn.
+ *
+ * The push is run asynchronously in a workqueue, which means the caller needs
+ * to handle waiting on the async flush for space to become available.
+ * We don't want to interrupt any push that is in progress, hence we only queue
+ * work if we set the pushing bit approriately.
+ *
+ * We do this unlocked - we only need to know whether there is anything in the
+ * AIL at the time we are called. We don't need to access the contents of
+ * any of the objects, so the lock is not needed.
+ */
+void
+xfs_ail_push(
+	struct xfs_ail	*ailp,
+	xfs_lsn_t	threshold_lsn)
+{
+	xfs_log_item_t	*lip;
+
+	lip = xfs_ail_min(ailp);
+	if (!lip || XFS_FORCED_SHUTDOWN(ailp->xa_mount) ||
+	    XFS_LSN_CMP(threshold_lsn, ailp->xa_target) <= 0)
+		return;
+
+	/*
+	 * Ensure that the new target is noticed in push code before it clears
+	 * the XFS_AIL_PUSHING_BIT.
+	 */
+	smp_wmb();
+	xfs_trans_ail_copy_lsn(ailp, &ailp->xa_target, &threshold_lsn);
+	smp_wmb();
+
+	wake_up_process(ailp->xa_task);
+}
+
+/*
+ * Push out all items in the AIL immediately
+ */
+void
+xfs_ail_push_all(
+	struct xfs_ail  *ailp)
+{
+	xfs_lsn_t       threshold_lsn = xfs_ail_max_lsn(ailp);
+
+	if (threshold_lsn)
+		xfs_ail_push(ailp, threshold_lsn);
+}
+
+/*
+ * This is to be called when an item is unlocked that may have
+ * been in the AIL.  It will wake up the first member of the AIL
+ * wait list if this item's unlocking might allow it to progress.
+ * If the item is in the AIL, then we need to get the AIL lock
+ * while doing our checking so we don't race with someone going
+ * to sleep waiting for this event in xfs_trans_push_ail().
+ */
+void
+xfs_trans_unlocked_item(
+	struct xfs_ail	*ailp,
+	xfs_log_item_t	*lip)
+{
+	xfs_log_item_t	*min_lip;
+
+	/*
+	 * If we're forcibly shutting down, we may have
+	 * unlocked log items arbitrarily. The last thing
+	 * we want to do is to move the tail of the log
+	 * over some potentially valid data.
+	 */
+	if (!(lip->li_flags & XFS_LI_IN_AIL) ||
+	    XFS_FORCED_SHUTDOWN(ailp->xa_mount)) {
+		return;
+	}
+
+	/*
+	 * This is the one case where we can call into xfs_ail_min()
+	 * without holding the AIL lock because we only care about the
+	 * case where we are at the tail of the AIL.  If the object isn't
+	 * at the tail, it doesn't matter what result we get back.  This
+	 * is slightly racy because since we were just unlocked, we could
+	 * go to sleep between the call to xfs_ail_min and the call to
+	 * xfs_log_move_tail, have someone else lock us, commit to us disk,
+	 * move us out of the tail of the AIL, and then we wake up.  However,
+	 * the call to xfs_log_move_tail() doesn't do anything if there's
+	 * not enough free space to wake people up so we're safe calling it.
+	 */
+	min_lip = xfs_ail_min(ailp);
+
+	if (min_lip == lip)
+		xfs_log_move_tail(ailp->xa_mount, 1);
+}	/* xfs_trans_unlocked_item */
+
+/*
+ * xfs_trans_ail_update - bulk AIL insertion operation.
+ *
+ * @xfs_trans_ail_update takes an array of log items that all need to be
+ * positioned at the same LSN in the AIL. If an item is not in the AIL, it will
+ * be added.  Otherwise, it will be repositioned  by removing it and re-adding
+ * it to the AIL. If we move the first item in the AIL, update the log tail to
+ * match the new minimum LSN in the AIL.
+ *
+ * This function takes the AIL lock once to execute the update operations on
+ * all the items in the array, and as such should not be called with the AIL
+ * lock held. As a result, once we have the AIL lock, we need to check each log
+ * item LSN to confirm it needs to be moved forward in the AIL.
+ *
+ * To optimise the insert operation, we delete all the items from the AIL in
+ * the first pass, moving them into a temporary list, then splice the temporary
+ * list into the correct position in the AIL. This avoids needing to do an
+ * insert operation on every item.
+ *
+ * This function must be called with the AIL lock held.  The lock is dropped
+ * before returning.
+ */
+void
+xfs_trans_ail_update_bulk(
+	struct xfs_ail		*ailp,
+	struct xfs_ail_cursor	*cur,
+	struct xfs_log_item	**log_items,
+	int			nr_items,
+	xfs_lsn_t		lsn) __releases(ailp->xa_lock)
+{
+	xfs_log_item_t		*mlip;
+	xfs_lsn_t		tail_lsn;
+	int			mlip_changed = 0;
+	int			i;
+	LIST_HEAD(tmp);
+
+	mlip = xfs_ail_min(ailp);
+
+	for (i = 0; i < nr_items; i++) {
+		struct xfs_log_item *lip = log_items[i];
+		if (lip->li_flags & XFS_LI_IN_AIL) {
+			/* check if we really need to move the item */
+			if (XFS_LSN_CMP(lsn, lip->li_lsn) <= 0)
+				continue;
+
+			xfs_ail_delete(ailp, lip);
+			if (mlip == lip)
+				mlip_changed = 1;
+		} else {
+			lip->li_flags |= XFS_LI_IN_AIL;
+		}
+		lip->li_lsn = lsn;
+		list_add(&lip->li_ail, &tmp);
+	}
+
+	xfs_ail_splice(ailp, cur, &tmp, lsn);
+
+	if (!mlip_changed) {
+		spin_unlock(&ailp->xa_lock);
+		return;
+	}
+
+	/*
+	 * It is not safe to access mlip after the AIL lock is dropped, so we
+	 * must get a copy of li_lsn before we do so.  This is especially
+	 * important on 32-bit platforms where accessing and updating 64-bit
+	 * values like li_lsn is not atomic.
+	 */
+	mlip = xfs_ail_min(ailp);
+	tail_lsn = mlip->li_lsn;
+	spin_unlock(&ailp->xa_lock);
+	xfs_log_move_tail(ailp->xa_mount, tail_lsn);
+}
+
+/*
+ * xfs_trans_ail_delete_bulk - remove multiple log items from the AIL
+ *
+ * @xfs_trans_ail_delete_bulk takes an array of log items that all need to
+ * removed from the AIL. The caller is already holding the AIL lock, and done
+ * all the checks necessary to ensure the items passed in via @log_items are
+ * ready for deletion. This includes checking that the items are in the AIL.
+ *
+ * For each log item to be removed, unlink it  from the AIL, clear the IN_AIL
+ * flag from the item and reset the item's lsn to 0. If we remove the first
+ * item in the AIL, update the log tail to match the new minimum LSN in the
+ * AIL.
+ *
+ * This function will not drop the AIL lock until all items are removed from
+ * the AIL to minimise the amount of lock traffic on the AIL. This does not
+ * greatly increase the AIL hold time, but does significantly reduce the amount
+ * of traffic on the lock, especially during IO completion.
+ *
+ * This function must be called with the AIL lock held.  The lock is dropped
+ * before returning.
+ */
+void
+xfs_trans_ail_delete_bulk(
+	struct xfs_ail		*ailp,
+	struct xfs_log_item	**log_items,
+	int			nr_items) __releases(ailp->xa_lock)
+{
+	xfs_log_item_t		*mlip;
+	xfs_lsn_t		tail_lsn;
+	int			mlip_changed = 0;
+	int			i;
+
+	mlip = xfs_ail_min(ailp);
+
+	for (i = 0; i < nr_items; i++) {
+		struct xfs_log_item *lip = log_items[i];
+		if (!(lip->li_flags & XFS_LI_IN_AIL)) {
+			struct xfs_mount	*mp = ailp->xa_mount;
+
+			spin_unlock(&ailp->xa_lock);
+			if (!XFS_FORCED_SHUTDOWN(mp)) {
+				xfs_alert_tag(mp, XFS_PTAG_AILDELETE,
+		"%s: attempting to delete a log item that is not in the AIL",
+						__func__);
+				xfs_force_shutdown(mp, SHUTDOWN_CORRUPT_INCORE);
+			}
+			return;
+		}
+
+		xfs_ail_delete(ailp, lip);
+		lip->li_flags &= ~XFS_LI_IN_AIL;
+		lip->li_lsn = 0;
+		if (mlip == lip)
+			mlip_changed = 1;
+	}
+
+	if (!mlip_changed) {
+		spin_unlock(&ailp->xa_lock);
+		return;
+	}
+
+	/*
+	 * It is not safe to access mlip after the AIL lock is dropped, so we
+	 * must get a copy of li_lsn before we do so.  This is especially
+	 * important on 32-bit platforms where accessing and updating 64-bit
+	 * values like li_lsn is not atomic. It is possible we've emptied the
+	 * AIL here, so if that is the case, pass an LSN of 0 to the tail move.
+	 */
+	mlip = xfs_ail_min(ailp);
+	tail_lsn = mlip ? mlip->li_lsn : 0;
+	spin_unlock(&ailp->xa_lock);
+	xfs_log_move_tail(ailp->xa_mount, tail_lsn);
+}
+
+/*
+ * The active item list (AIL) is a doubly linked list of log
+ * items sorted by ascending lsn.  The base of the list is
+ * a forw/back pointer pair embedded in the xfs mount structure.
+ * The base is initialized with both pointers pointing to the
+ * base.  This case always needs to be distinguished, because
+ * the base has no lsn to look at.  We almost always insert
+ * at the end of the list, so on inserts we search from the
+ * end of the list to find where the new item belongs.
+ */
+
+/*
+ * Initialize the doubly linked list to point only to itself.
+ */
+int
+xfs_trans_ail_init(
+	xfs_mount_t	*mp)
+{
+	struct xfs_ail	*ailp;
+
+	ailp = kmem_zalloc(sizeof(struct xfs_ail), KM_MAYFAIL);
+	if (!ailp)
+		return ENOMEM;
+
+	ailp->xa_mount = mp;
+	INIT_LIST_HEAD(&ailp->xa_ail);
+	spin_lock_init(&ailp->xa_lock);
+
+	ailp->xa_task = kthread_run(xfsaild, ailp, "xfsaild/%s",
+			ailp->xa_mount->m_fsname);
+	if (IS_ERR(ailp->xa_task))
+		goto out_free_ailp;
+
+	mp->m_ail = ailp;
+	return 0;
+
+out_free_ailp:
+	kmem_free(ailp);
+	return ENOMEM;
+}
+
+void
+xfs_trans_ail_destroy(
+	xfs_mount_t	*mp)
+{
+	struct xfs_ail	*ailp = mp->m_ail;
+
+	kthread_stop(ailp->xa_task);
+	kmem_free(ailp);
+}
diff --git a/fs/xfs/xfs_trans_buf.c b/fs/xfs/xfs_trans_buf.c
new file mode 100644
index 00000000..03b3b7f8
--- /dev/null
+++ b/fs/xfs/xfs_trans_buf.c
@@ -0,0 +1,879 @@
+/*
+ * Copyright (c) 2000-2002,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_mount.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_alloc_btree.h"
+#include "xfs_ialloc_btree.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_buf_item.h"
+#include "xfs_trans_priv.h"
+#include "xfs_error.h"
+#include "xfs_rw.h"
+#include "xfs_trace.h"
+
+/*
+ * Check to see if a buffer matching the given parameters is already
+ * a part of the given transaction.
+ */
+STATIC struct xfs_buf *
+xfs_trans_buf_item_match(
+	struct xfs_trans	*tp,
+	struct xfs_buftarg	*target,
+	xfs_daddr_t		blkno,
+	int			len)
+{
+	struct xfs_log_item_desc *lidp;
+	struct xfs_buf_log_item	*blip;
+
+	len = BBTOB(len);
+	list_for_each_entry(lidp, &tp->t_items, lid_trans) {
+		blip = (struct xfs_buf_log_item *)lidp->lid_item;
+		if (blip->bli_item.li_type == XFS_LI_BUF &&
+		    XFS_BUF_TARGET(blip->bli_buf) == target &&
+		    XFS_BUF_ADDR(blip->bli_buf) == blkno &&
+		    XFS_BUF_COUNT(blip->bli_buf) == len)
+			return blip->bli_buf;
+	}
+
+	return NULL;
+}
+
+/*
+ * Add the locked buffer to the transaction.
+ *
+ * The buffer must be locked, and it cannot be associated with any
+ * transaction.
+ *
+ * If the buffer does not yet have a buf log item associated with it,
+ * then allocate one for it.  Then add the buf item to the transaction.
+ */
+STATIC void
+_xfs_trans_bjoin(
+	struct xfs_trans	*tp,
+	struct xfs_buf		*bp,
+	int			reset_recur)
+{
+	struct xfs_buf_log_item	*bip;
+
+	ASSERT(XFS_BUF_ISBUSY(bp));
+	ASSERT(XFS_BUF_FSPRIVATE2(bp, void *) == NULL);
+
+	/*
+	 * The xfs_buf_log_item pointer is stored in b_fsprivate.  If
+	 * it doesn't have one yet, then allocate one and initialize it.
+	 * The checks to see if one is there are in xfs_buf_item_init().
+	 */
+	xfs_buf_item_init(bp, tp->t_mountp);
+	bip = XFS_BUF_FSPRIVATE(bp, xfs_buf_log_item_t *);
+	ASSERT(!(bip->bli_flags & XFS_BLI_STALE));
+	ASSERT(!(bip->bli_format.blf_flags & XFS_BLF_CANCEL));
+	ASSERT(!(bip->bli_flags & XFS_BLI_LOGGED));
+	if (reset_recur)
+		bip->bli_recur = 0;
+
+	/*
+	 * Take a reference for this transaction on the buf item.
+	 */
+	atomic_inc(&bip->bli_refcount);
+
+	/*
+	 * Get a log_item_desc to point at the new item.
+	 */
+	xfs_trans_add_item(tp, &bip->bli_item);
+
+	/*
+	 * Initialize b_fsprivate2 so we can find it with incore_match()
+	 * in xfs_trans_get_buf() and friends above.
+	 */
+	XFS_BUF_SET_FSPRIVATE2(bp, tp);
+
+}
+
+void
+xfs_trans_bjoin(
+	struct xfs_trans	*tp,
+	struct xfs_buf		*bp)
+{
+	_xfs_trans_bjoin(tp, bp, 0);
+	trace_xfs_trans_bjoin(bp->b_fspriv);
+}
+
+/*
+ * Get and lock the buffer for the caller if it is not already
+ * locked within the given transaction.  If it is already locked
+ * within the transaction, just increment its lock recursion count
+ * and return a pointer to it.
+ *
+ * If the transaction pointer is NULL, make this just a normal
+ * get_buf() call.
+ */
+xfs_buf_t *
+xfs_trans_get_buf(xfs_trans_t	*tp,
+		  xfs_buftarg_t	*target_dev,
+		  xfs_daddr_t	blkno,
+		  int		len,
+		  uint		flags)
+{
+	xfs_buf_t		*bp;
+	xfs_buf_log_item_t	*bip;
+
+	if (flags == 0)
+		flags = XBF_LOCK | XBF_MAPPED;
+
+	/*
+	 * Default to a normal get_buf() call if the tp is NULL.
+	 */
+	if (tp == NULL)
+		return xfs_buf_get(target_dev, blkno, len,
+				   flags | XBF_DONT_BLOCK);
+
+	/*
+	 * If we find the buffer in the cache with this transaction
+	 * pointer in its b_fsprivate2 field, then we know we already
+	 * have it locked.  In this case we just increment the lock
+	 * recursion count and return the buffer to the caller.
+	 */
+	bp = xfs_trans_buf_item_match(tp, target_dev, blkno, len);
+	if (bp != NULL) {
+		ASSERT(XFS_BUF_VALUSEMA(bp) <= 0);
+		if (XFS_FORCED_SHUTDOWN(tp->t_mountp))
+			XFS_BUF_SUPER_STALE(bp);
+
+		/*
+		 * If the buffer is stale then it was binval'ed
+		 * since last read.  This doesn't matter since the
+		 * caller isn't allowed to use the data anyway.
+		 */
+		else if (XFS_BUF_ISSTALE(bp))
+			ASSERT(!XFS_BUF_ISDELAYWRITE(bp));
+
+		ASSERT(XFS_BUF_FSPRIVATE2(bp, xfs_trans_t *) == tp);
+		bip = XFS_BUF_FSPRIVATE(bp, xfs_buf_log_item_t *);
+		ASSERT(bip != NULL);
+		ASSERT(atomic_read(&bip->bli_refcount) > 0);
+		bip->bli_recur++;
+		trace_xfs_trans_get_buf_recur(bip);
+		return (bp);
+	}
+
+	/*
+	 * We always specify the XBF_DONT_BLOCK flag within a transaction
+	 * so that get_buf does not try to push out a delayed write buffer
+	 * which might cause another transaction to take place (if the
+	 * buffer was delayed alloc).  Such recursive transactions can
+	 * easily deadlock with our current transaction as well as cause
+	 * us to run out of stack space.
+	 */
+	bp = xfs_buf_get(target_dev, blkno, len, flags | XBF_DONT_BLOCK);
+	if (bp == NULL) {
+		return NULL;
+	}
+
+	ASSERT(!XFS_BUF_GETERROR(bp));
+
+	_xfs_trans_bjoin(tp, bp, 1);
+	trace_xfs_trans_get_buf(bp->b_fspriv);
+	return (bp);
+}
+
+/*
+ * Get and lock the superblock buffer of this file system for the
+ * given transaction.
+ *
+ * We don't need to use incore_match() here, because the superblock
+ * buffer is a private buffer which we keep a pointer to in the
+ * mount structure.
+ */
+xfs_buf_t *
+xfs_trans_getsb(xfs_trans_t	*tp,
+		struct xfs_mount *mp,
+		int		flags)
+{
+	xfs_buf_t		*bp;
+	xfs_buf_log_item_t	*bip;
+
+	/*
+	 * Default to just trying to lock the superblock buffer
+	 * if tp is NULL.
+	 */
+	if (tp == NULL) {
+		return (xfs_getsb(mp, flags));
+	}
+
+	/*
+	 * If the superblock buffer already has this transaction
+	 * pointer in its b_fsprivate2 field, then we know we already
+	 * have it locked.  In this case we just increment the lock
+	 * recursion count and return the buffer to the caller.
+	 */
+	bp = mp->m_sb_bp;
+	if (XFS_BUF_FSPRIVATE2(bp, xfs_trans_t *) == tp) {
+		bip = XFS_BUF_FSPRIVATE(bp, xfs_buf_log_item_t*);
+		ASSERT(bip != NULL);
+		ASSERT(atomic_read(&bip->bli_refcount) > 0);
+		bip->bli_recur++;
+		trace_xfs_trans_getsb_recur(bip);
+		return (bp);
+	}
+
+	bp = xfs_getsb(mp, flags);
+	if (bp == NULL)
+		return NULL;
+
+	_xfs_trans_bjoin(tp, bp, 1);
+	trace_xfs_trans_getsb(bp->b_fspriv);
+	return (bp);
+}
+
+#ifdef DEBUG
+xfs_buftarg_t *xfs_error_target;
+int	xfs_do_error;
+int	xfs_req_num;
+int	xfs_error_mod = 33;
+#endif
+
+/*
+ * Get and lock the buffer for the caller if it is not already
+ * locked within the given transaction.  If it has not yet been
+ * read in, read it from disk. If it is already locked
+ * within the transaction and already read in, just increment its
+ * lock recursion count and return a pointer to it.
+ *
+ * If the transaction pointer is NULL, make this just a normal
+ * read_buf() call.
+ */
+int
+xfs_trans_read_buf(
+	xfs_mount_t	*mp,
+	xfs_trans_t	*tp,
+	xfs_buftarg_t	*target,
+	xfs_daddr_t	blkno,
+	int		len,
+	uint		flags,
+	xfs_buf_t	**bpp)
+{
+	xfs_buf_t		*bp;
+	xfs_buf_log_item_t	*bip;
+	int			error;
+
+	if (flags == 0)
+		flags = XBF_LOCK | XBF_MAPPED;
+
+	/*
+	 * Default to a normal get_buf() call if the tp is NULL.
+	 */
+	if (tp == NULL) {
+		bp = xfs_buf_read(target, blkno, len, flags | XBF_DONT_BLOCK);
+		if (!bp)
+			return (flags & XBF_TRYLOCK) ?
+					EAGAIN : XFS_ERROR(ENOMEM);
+
+		if (XFS_BUF_GETERROR(bp) != 0) {
+			xfs_ioerror_alert("xfs_trans_read_buf", mp,
+					  bp, blkno);
+			error = XFS_BUF_GETERROR(bp);
+			xfs_buf_relse(bp);
+			return error;
+		}
+#ifdef DEBUG
+		if (xfs_do_error) {
+			if (xfs_error_target == target) {
+				if (((xfs_req_num++) % xfs_error_mod) == 0) {
+					xfs_buf_relse(bp);
+					xfs_debug(mp, "Returning error!");
+					return XFS_ERROR(EIO);
+				}
+			}
+		}
+#endif
+		if (XFS_FORCED_SHUTDOWN(mp))
+			goto shutdown_abort;
+		*bpp = bp;
+		return 0;
+	}
+
+	/*
+	 * If we find the buffer in the cache with this transaction
+	 * pointer in its b_fsprivate2 field, then we know we already
+	 * have it locked.  If it is already read in we just increment
+	 * the lock recursion count and return the buffer to the caller.
+	 * If the buffer is not yet read in, then we read it in, increment
+	 * the lock recursion count, and return it to the caller.
+	 */
+	bp = xfs_trans_buf_item_match(tp, target, blkno, len);
+	if (bp != NULL) {
+		ASSERT(XFS_BUF_VALUSEMA(bp) <= 0);
+		ASSERT(XFS_BUF_FSPRIVATE2(bp, xfs_trans_t *) == tp);
+		ASSERT(XFS_BUF_FSPRIVATE(bp, void *) != NULL);
+		ASSERT((XFS_BUF_ISERROR(bp)) == 0);
+		if (!(XFS_BUF_ISDONE(bp))) {
+			trace_xfs_trans_read_buf_io(bp, _RET_IP_);
+			ASSERT(!XFS_BUF_ISASYNC(bp));
+			XFS_BUF_READ(bp);
+			xfsbdstrat(tp->t_mountp, bp);
+			error = xfs_buf_iowait(bp);
+			if (error) {
+				xfs_ioerror_alert("xfs_trans_read_buf", mp,
+						  bp, blkno);
+				xfs_buf_relse(bp);
+				/*
+				 * We can gracefully recover from most read
+				 * errors. Ones we can't are those that happen
+				 * after the transaction's already dirty.
+				 */
+				if (tp->t_flags & XFS_TRANS_DIRTY)
+					xfs_force_shutdown(tp->t_mountp,
+							SHUTDOWN_META_IO_ERROR);
+				return error;
+			}
+		}
+		/*
+		 * We never locked this buf ourselves, so we shouldn't
+		 * brelse it either. Just get out.
+		 */
+		if (XFS_FORCED_SHUTDOWN(mp)) {
+			trace_xfs_trans_read_buf_shut(bp, _RET_IP_);
+			*bpp = NULL;
+			return XFS_ERROR(EIO);
+		}
+
+
+		bip = XFS_BUF_FSPRIVATE(bp, xfs_buf_log_item_t*);
+		bip->bli_recur++;
+
+		ASSERT(atomic_read(&bip->bli_refcount) > 0);
+		trace_xfs_trans_read_buf_recur(bip);
+		*bpp = bp;
+		return 0;
+	}
+
+	/*
+	 * We always specify the XBF_DONT_BLOCK flag within a transaction
+	 * so that get_buf does not try to push out a delayed write buffer
+	 * which might cause another transaction to take place (if the
+	 * buffer was delayed alloc).  Such recursive transactions can
+	 * easily deadlock with our current transaction as well as cause
+	 * us to run out of stack space.
+	 */
+	bp = xfs_buf_read(target, blkno, len, flags | XBF_DONT_BLOCK);
+	if (bp == NULL) {
+		*bpp = NULL;
+		return (flags & XBF_TRYLOCK) ?
+					0 : XFS_ERROR(ENOMEM);
+	}
+	if (XFS_BUF_GETERROR(bp) != 0) {
+	    XFS_BUF_SUPER_STALE(bp);
+		error = XFS_BUF_GETERROR(bp);
+
+		xfs_ioerror_alert("xfs_trans_read_buf", mp,
+				  bp, blkno);
+		if (tp->t_flags & XFS_TRANS_DIRTY)
+			xfs_force_shutdown(tp->t_mountp, SHUTDOWN_META_IO_ERROR);
+		xfs_buf_relse(bp);
+		return error;
+	}
+#ifdef DEBUG
+	if (xfs_do_error && !(tp->t_flags & XFS_TRANS_DIRTY)) {
+		if (xfs_error_target == target) {
+			if (((xfs_req_num++) % xfs_error_mod) == 0) {
+				xfs_force_shutdown(tp->t_mountp,
+						   SHUTDOWN_META_IO_ERROR);
+				xfs_buf_relse(bp);
+				xfs_debug(mp, "Returning trans error!");
+				return XFS_ERROR(EIO);
+			}
+		}
+	}
+#endif
+	if (XFS_FORCED_SHUTDOWN(mp))
+		goto shutdown_abort;
+
+	_xfs_trans_bjoin(tp, bp, 1);
+	trace_xfs_trans_read_buf(bp->b_fspriv);
+
+	*bpp = bp;
+	return 0;
+
+shutdown_abort:
+	/*
+	 * the theory here is that buffer is good but we're
+	 * bailing out because the filesystem is being forcibly
+	 * shut down.  So we should leave the b_flags alone since
+	 * the buffer's not staled and just get out.
+	 */
+#if defined(DEBUG)
+	if (XFS_BUF_ISSTALE(bp) && XFS_BUF_ISDELAYWRITE(bp))
+		xfs_notice(mp, "about to pop assert, bp == 0x%p", bp);
+#endif
+	ASSERT((XFS_BUF_BFLAGS(bp) & (XBF_STALE|XBF_DELWRI)) !=
+				     (XBF_STALE|XBF_DELWRI));
+
+	trace_xfs_trans_read_buf_shut(bp, _RET_IP_);
+	xfs_buf_relse(bp);
+	*bpp = NULL;
+	return XFS_ERROR(EIO);
+}
+
+
+/*
+ * Release the buffer bp which was previously acquired with one of the
+ * xfs_trans_... buffer allocation routines if the buffer has not
+ * been modified within this transaction.  If the buffer is modified
+ * within this transaction, do decrement the recursion count but do
+ * not release the buffer even if the count goes to 0.  If the buffer is not
+ * modified within the transaction, decrement the recursion count and
+ * release the buffer if the recursion count goes to 0.
+ *
+ * If the buffer is to be released and it was not modified before
+ * this transaction began, then free the buf_log_item associated with it.
+ *
+ * If the transaction pointer is NULL, make this just a normal
+ * brelse() call.
+ */
+void
+xfs_trans_brelse(xfs_trans_t	*tp,
+		 xfs_buf_t	*bp)
+{
+	xfs_buf_log_item_t	*bip;
+	xfs_log_item_t		*lip;
+
+	/*
+	 * Default to a normal brelse() call if the tp is NULL.
+	 */
+	if (tp == NULL) {
+		ASSERT(XFS_BUF_FSPRIVATE2(bp, void *) == NULL);
+		/*
+		 * If there's a buf log item attached to the buffer,
+		 * then let the AIL know that the buffer is being
+		 * unlocked.
+		 */
+		if (XFS_BUF_FSPRIVATE(bp, void *) != NULL) {
+			lip = XFS_BUF_FSPRIVATE(bp, xfs_log_item_t *);
+			if (lip->li_type == XFS_LI_BUF) {
+				bip = XFS_BUF_FSPRIVATE(bp,xfs_buf_log_item_t*);
+				xfs_trans_unlocked_item(bip->bli_item.li_ailp,
+							lip);
+			}
+		}
+		xfs_buf_relse(bp);
+		return;
+	}
+
+	ASSERT(XFS_BUF_FSPRIVATE2(bp, xfs_trans_t *) == tp);
+	bip = XFS_BUF_FSPRIVATE(bp, xfs_buf_log_item_t *);
+	ASSERT(bip->bli_item.li_type == XFS_LI_BUF);
+	ASSERT(!(bip->bli_flags & XFS_BLI_STALE));
+	ASSERT(!(bip->bli_format.blf_flags & XFS_BLF_CANCEL));
+	ASSERT(atomic_read(&bip->bli_refcount) > 0);
+
+	trace_xfs_trans_brelse(bip);
+
+	/*
+	 * If the release is just for a recursive lock,
+	 * then decrement the count and return.
+	 */
+	if (bip->bli_recur > 0) {
+		bip->bli_recur--;
+		return;
+	}
+
+	/*
+	 * If the buffer is dirty within this transaction, we can't
+	 * release it until we commit.
+	 */
+	if (bip->bli_item.li_desc->lid_flags & XFS_LID_DIRTY)
+		return;
+
+	/*
+	 * If the buffer has been invalidated, then we can't release
+	 * it until the transaction commits to disk unless it is re-dirtied
+	 * as part of this transaction.  This prevents us from pulling
+	 * the item from the AIL before we should.
+	 */
+	if (bip->bli_flags & XFS_BLI_STALE)
+		return;
+
+	ASSERT(!(bip->bli_flags & XFS_BLI_LOGGED));
+
+	/*
+	 * Free up the log item descriptor tracking the released item.
+	 */
+	xfs_trans_del_item(&bip->bli_item);
+
+	/*
+	 * Clear the hold flag in the buf log item if it is set.
+	 * We wouldn't want the next user of the buffer to
+	 * get confused.
+	 */
+	if (bip->bli_flags & XFS_BLI_HOLD) {
+		bip->bli_flags &= ~XFS_BLI_HOLD;
+	}
+
+	/*
+	 * Drop our reference to the buf log item.
+	 */
+	atomic_dec(&bip->bli_refcount);
+
+	/*
+	 * If the buf item is not tracking data in the log, then
+	 * we must free it before releasing the buffer back to the
+	 * free pool.  Before releasing the buffer to the free pool,
+	 * clear the transaction pointer in b_fsprivate2 to dissolve
+	 * its relation to this transaction.
+	 */
+	if (!xfs_buf_item_dirty(bip)) {
+/***
+		ASSERT(bp->b_pincount == 0);
+***/
+		ASSERT(atomic_read(&bip->bli_refcount) == 0);
+		ASSERT(!(bip->bli_item.li_flags & XFS_LI_IN_AIL));
+		ASSERT(!(bip->bli_flags & XFS_BLI_INODE_ALLOC_BUF));
+		xfs_buf_item_relse(bp);
+		bip = NULL;
+	}
+	XFS_BUF_SET_FSPRIVATE2(bp, NULL);
+
+	/*
+	 * If we've still got a buf log item on the buffer, then
+	 * tell the AIL that the buffer is being unlocked.
+	 */
+	if (bip != NULL) {
+		xfs_trans_unlocked_item(bip->bli_item.li_ailp,
+					(xfs_log_item_t*)bip);
+	}
+
+	xfs_buf_relse(bp);
+	return;
+}
+
+/*
+ * Mark the buffer as not needing to be unlocked when the buf item's
+ * IOP_UNLOCK() routine is called.  The buffer must already be locked
+ * and associated with the given transaction.
+ */
+/* ARGSUSED */
+void
+xfs_trans_bhold(xfs_trans_t	*tp,
+		xfs_buf_t	*bp)
+{
+	xfs_buf_log_item_t	*bip;
+
+	ASSERT(XFS_BUF_ISBUSY(bp));
+	ASSERT(XFS_BUF_FSPRIVATE2(bp, xfs_trans_t *) == tp);
+	ASSERT(XFS_BUF_FSPRIVATE(bp, void *) != NULL);
+
+	bip = XFS_BUF_FSPRIVATE(bp, xfs_buf_log_item_t *);
+	ASSERT(!(bip->bli_flags & XFS_BLI_STALE));
+	ASSERT(!(bip->bli_format.blf_flags & XFS_BLF_CANCEL));
+	ASSERT(atomic_read(&bip->bli_refcount) > 0);
+	bip->bli_flags |= XFS_BLI_HOLD;
+	trace_xfs_trans_bhold(bip);
+}
+
+/*
+ * Cancel the previous buffer hold request made on this buffer
+ * for this transaction.
+ */
+void
+xfs_trans_bhold_release(xfs_trans_t	*tp,
+			xfs_buf_t	*bp)
+{
+	xfs_buf_log_item_t	*bip;
+
+	ASSERT(XFS_BUF_ISBUSY(bp));
+	ASSERT(XFS_BUF_FSPRIVATE2(bp, xfs_trans_t *) == tp);
+	ASSERT(XFS_BUF_FSPRIVATE(bp, void *) != NULL);
+
+	bip = XFS_BUF_FSPRIVATE(bp, xfs_buf_log_item_t *);
+	ASSERT(!(bip->bli_flags & XFS_BLI_STALE));
+	ASSERT(!(bip->bli_format.blf_flags & XFS_BLF_CANCEL));
+	ASSERT(atomic_read(&bip->bli_refcount) > 0);
+	ASSERT(bip->bli_flags & XFS_BLI_HOLD);
+	bip->bli_flags &= ~XFS_BLI_HOLD;
+
+	trace_xfs_trans_bhold_release(bip);
+}
+
+/*
+ * This is called to mark bytes first through last inclusive of the given
+ * buffer as needing to be logged when the transaction is committed.
+ * The buffer must already be associated with the given transaction.
+ *
+ * First and last are numbers relative to the beginning of this buffer,
+ * so the first byte in the buffer is numbered 0 regardless of the
+ * value of b_blkno.
+ */
+void
+xfs_trans_log_buf(xfs_trans_t	*tp,
+		  xfs_buf_t	*bp,
+		  uint		first,
+		  uint		last)
+{
+	xfs_buf_log_item_t	*bip;
+
+	ASSERT(XFS_BUF_ISBUSY(bp));
+	ASSERT(XFS_BUF_FSPRIVATE2(bp, xfs_trans_t *) == tp);
+	ASSERT(XFS_BUF_FSPRIVATE(bp, void *) != NULL);
+	ASSERT((first <= last) && (last < XFS_BUF_COUNT(bp)));
+	ASSERT((XFS_BUF_IODONE_FUNC(bp) == NULL) ||
+	       (XFS_BUF_IODONE_FUNC(bp) == xfs_buf_iodone_callbacks));
+
+	/*
+	 * Mark the buffer as needing to be written out eventually,
+	 * and set its iodone function to remove the buffer's buf log
+	 * item from the AIL and free it when the buffer is flushed
+	 * to disk.  See xfs_buf_attach_iodone() for more details
+	 * on li_cb and xfs_buf_iodone_callbacks().
+	 * If we end up aborting this transaction, we trap this buffer
+	 * inside the b_bdstrat callback so that this won't get written to
+	 * disk.
+	 */
+	XFS_BUF_DELAYWRITE(bp);
+	XFS_BUF_DONE(bp);
+
+	bip = XFS_BUF_FSPRIVATE(bp, xfs_buf_log_item_t *);
+	ASSERT(atomic_read(&bip->bli_refcount) > 0);
+	XFS_BUF_SET_IODONE_FUNC(bp, xfs_buf_iodone_callbacks);
+	bip->bli_item.li_cb = xfs_buf_iodone;
+
+	trace_xfs_trans_log_buf(bip);
+
+	/*
+	 * If we invalidated the buffer within this transaction, then
+	 * cancel the invalidation now that we're dirtying the buffer
+	 * again.  There are no races with the code in xfs_buf_item_unpin(),
+	 * because we have a reference to the buffer this entire time.
+	 */
+	if (bip->bli_flags & XFS_BLI_STALE) {
+		bip->bli_flags &= ~XFS_BLI_STALE;
+		ASSERT(XFS_BUF_ISSTALE(bp));
+		XFS_BUF_UNSTALE(bp);
+		bip->bli_format.blf_flags &= ~XFS_BLF_CANCEL;
+	}
+
+	tp->t_flags |= XFS_TRANS_DIRTY;
+	bip->bli_item.li_desc->lid_flags |= XFS_LID_DIRTY;
+	bip->bli_flags |= XFS_BLI_LOGGED;
+	xfs_buf_item_log(bip, first, last);
+}
+
+
+/*
+ * This called to invalidate a buffer that is being used within
+ * a transaction.  Typically this is because the blocks in the
+ * buffer are being freed, so we need to prevent it from being
+ * written out when we're done.  Allowing it to be written again
+ * might overwrite data in the free blocks if they are reallocated
+ * to a file.
+ *
+ * We prevent the buffer from being written out by clearing the
+ * B_DELWRI flag.  We can't always
+ * get rid of the buf log item at this point, though, because
+ * the buffer may still be pinned by another transaction.  If that
+ * is the case, then we'll wait until the buffer is committed to
+ * disk for the last time (we can tell by the ref count) and
+ * free it in xfs_buf_item_unpin().  Until it is cleaned up we
+ * will keep the buffer locked so that the buffer and buf log item
+ * are not reused.
+ */
+void
+xfs_trans_binval(
+	xfs_trans_t	*tp,
+	xfs_buf_t	*bp)
+{
+	xfs_buf_log_item_t	*bip;
+
+	ASSERT(XFS_BUF_ISBUSY(bp));
+	ASSERT(XFS_BUF_FSPRIVATE2(bp, xfs_trans_t *) == tp);
+	ASSERT(XFS_BUF_FSPRIVATE(bp, void *) != NULL);
+
+	bip = XFS_BUF_FSPRIVATE(bp, xfs_buf_log_item_t *);
+	ASSERT(atomic_read(&bip->bli_refcount) > 0);
+
+	trace_xfs_trans_binval(bip);
+
+	if (bip->bli_flags & XFS_BLI_STALE) {
+		/*
+		 * If the buffer is already invalidated, then
+		 * just return.
+		 */
+		ASSERT(!(XFS_BUF_ISDELAYWRITE(bp)));
+		ASSERT(XFS_BUF_ISSTALE(bp));
+		ASSERT(!(bip->bli_flags & (XFS_BLI_LOGGED | XFS_BLI_DIRTY)));
+		ASSERT(!(bip->bli_format.blf_flags & XFS_BLF_INODE_BUF));
+		ASSERT(bip->bli_format.blf_flags & XFS_BLF_CANCEL);
+		ASSERT(bip->bli_item.li_desc->lid_flags & XFS_LID_DIRTY);
+		ASSERT(tp->t_flags & XFS_TRANS_DIRTY);
+		return;
+	}
+
+	/*
+	 * Clear the dirty bit in the buffer and set the STALE flag
+	 * in the buf log item.  The STALE flag will be used in
+	 * xfs_buf_item_unpin() to determine if it should clean up
+	 * when the last reference to the buf item is given up.
+	 * We set the XFS_BLF_CANCEL flag in the buf log format structure
+	 * and log the buf item.  This will be used at recovery time
+	 * to determine that copies of the buffer in the log before
+	 * this should not be replayed.
+	 * We mark the item descriptor and the transaction dirty so
+	 * that we'll hold the buffer until after the commit.
+	 *
+	 * Since we're invalidating the buffer, we also clear the state
+	 * about which parts of the buffer have been logged.  We also
+	 * clear the flag indicating that this is an inode buffer since
+	 * the data in the buffer will no longer be valid.
+	 *
+	 * We set the stale bit in the buffer as well since we're getting
+	 * rid of it.
+	 */
+	XFS_BUF_UNDELAYWRITE(bp);
+	XFS_BUF_STALE(bp);
+	bip->bli_flags |= XFS_BLI_STALE;
+	bip->bli_flags &= ~(XFS_BLI_INODE_BUF | XFS_BLI_LOGGED | XFS_BLI_DIRTY);
+	bip->bli_format.blf_flags &= ~XFS_BLF_INODE_BUF;
+	bip->bli_format.blf_flags |= XFS_BLF_CANCEL;
+	memset((char *)(bip->bli_format.blf_data_map), 0,
+	      (bip->bli_format.blf_map_size * sizeof(uint)));
+	bip->bli_item.li_desc->lid_flags |= XFS_LID_DIRTY;
+	tp->t_flags |= XFS_TRANS_DIRTY;
+}
+
+/*
+ * This call is used to indicate that the buffer contains on-disk inodes which
+ * must be handled specially during recovery.  They require special handling
+ * because only the di_next_unlinked from the inodes in the buffer should be
+ * recovered.  The rest of the data in the buffer is logged via the inodes
+ * themselves.
+ *
+ * All we do is set the XFS_BLI_INODE_BUF flag in the items flags so it can be
+ * transferred to the buffer's log format structure so that we'll know what to
+ * do at recovery time.
+ */
+void
+xfs_trans_inode_buf(
+	xfs_trans_t	*tp,
+	xfs_buf_t	*bp)
+{
+	xfs_buf_log_item_t	*bip;
+
+	ASSERT(XFS_BUF_ISBUSY(bp));
+	ASSERT(XFS_BUF_FSPRIVATE2(bp, xfs_trans_t *) == tp);
+	ASSERT(XFS_BUF_FSPRIVATE(bp, void *) != NULL);
+
+	bip = XFS_BUF_FSPRIVATE(bp, xfs_buf_log_item_t *);
+	ASSERT(atomic_read(&bip->bli_refcount) > 0);
+
+	bip->bli_flags |= XFS_BLI_INODE_BUF;
+}
+
+/*
+ * This call is used to indicate that the buffer is going to
+ * be staled and was an inode buffer. This means it gets
+ * special processing during unpin - where any inodes 
+ * associated with the buffer should be removed from ail.
+ * There is also special processing during recovery,
+ * any replay of the inodes in the buffer needs to be
+ * prevented as the buffer may have been reused.
+ */
+void
+xfs_trans_stale_inode_buf(
+	xfs_trans_t	*tp,
+	xfs_buf_t	*bp)
+{
+	xfs_buf_log_item_t	*bip;
+
+	ASSERT(XFS_BUF_ISBUSY(bp));
+	ASSERT(XFS_BUF_FSPRIVATE2(bp, xfs_trans_t *) == tp);
+	ASSERT(XFS_BUF_FSPRIVATE(bp, void *) != NULL);
+
+	bip = XFS_BUF_FSPRIVATE(bp, xfs_buf_log_item_t *);
+	ASSERT(atomic_read(&bip->bli_refcount) > 0);
+
+	bip->bli_flags |= XFS_BLI_STALE_INODE;
+	bip->bli_item.li_cb = xfs_buf_iodone;
+}
+
+/*
+ * Mark the buffer as being one which contains newly allocated
+ * inodes.  We need to make sure that even if this buffer is
+ * relogged as an 'inode buf' we still recover all of the inode
+ * images in the face of a crash.  This works in coordination with
+ * xfs_buf_item_committed() to ensure that the buffer remains in the
+ * AIL at its original location even after it has been relogged.
+ */
+/* ARGSUSED */
+void
+xfs_trans_inode_alloc_buf(
+	xfs_trans_t	*tp,
+	xfs_buf_t	*bp)
+{
+	xfs_buf_log_item_t	*bip;
+
+	ASSERT(XFS_BUF_ISBUSY(bp));
+	ASSERT(XFS_BUF_FSPRIVATE2(bp, xfs_trans_t *) == tp);
+	ASSERT(XFS_BUF_FSPRIVATE(bp, void *) != NULL);
+
+	bip = XFS_BUF_FSPRIVATE(bp, xfs_buf_log_item_t *);
+	ASSERT(atomic_read(&bip->bli_refcount) > 0);
+
+	bip->bli_flags |= XFS_BLI_INODE_ALLOC_BUF;
+}
+
+
+/*
+ * Similar to xfs_trans_inode_buf(), this marks the buffer as a cluster of
+ * dquots. However, unlike in inode buffer recovery, dquot buffers get
+ * recovered in their entirety. (Hence, no XFS_BLI_DQUOT_ALLOC_BUF flag).
+ * The only thing that makes dquot buffers different from regular
+ * buffers is that we must not replay dquot bufs when recovering
+ * if a _corresponding_ quotaoff has happened. We also have to distinguish
+ * between usr dquot bufs and grp dquot bufs, because usr and grp quotas
+ * can be turned off independently.
+ */
+/* ARGSUSED */
+void
+xfs_trans_dquot_buf(
+	xfs_trans_t	*tp,
+	xfs_buf_t	*bp,
+	uint		type)
+{
+	xfs_buf_log_item_t	*bip;
+
+	ASSERT(XFS_BUF_ISBUSY(bp));
+	ASSERT(XFS_BUF_FSPRIVATE2(bp, xfs_trans_t *) == tp);
+	ASSERT(XFS_BUF_FSPRIVATE(bp, void *) != NULL);
+	ASSERT(type == XFS_BLF_UDQUOT_BUF ||
+	       type == XFS_BLF_PDQUOT_BUF ||
+	       type == XFS_BLF_GDQUOT_BUF);
+
+	bip = XFS_BUF_FSPRIVATE(bp, xfs_buf_log_item_t *);
+	ASSERT(atomic_read(&bip->bli_refcount) > 0);
+
+	bip->bli_format.blf_flags |= type;
+}
diff --git a/fs/xfs/xfs_trans_extfree.c b/fs/xfs/xfs_trans_extfree.c
new file mode 100644
index 00000000..f7590f5b
--- /dev/null
+++ b/fs/xfs/xfs_trans_extfree.c
@@ -0,0 +1,134 @@
+/*
+ * Copyright (c) 2000,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_mount.h"
+#include "xfs_trans_priv.h"
+#include "xfs_extfree_item.h"
+
+/*
+ * This routine is called to allocate an "extent free intention"
+ * log item that will hold nextents worth of extents.  The
+ * caller must use all nextents extents, because we are not
+ * flexible about this at all.
+ */
+xfs_efi_log_item_t *
+xfs_trans_get_efi(xfs_trans_t	*tp,
+		  uint		nextents)
+{
+	xfs_efi_log_item_t	*efip;
+
+	ASSERT(tp != NULL);
+	ASSERT(nextents > 0);
+
+	efip = xfs_efi_init(tp->t_mountp, nextents);
+	ASSERT(efip != NULL);
+
+	/*
+	 * Get a log_item_desc to point at the new item.
+	 */
+	xfs_trans_add_item(tp, &efip->efi_item);
+	return efip;
+}
+
+/*
+ * This routine is called to indicate that the described
+ * extent is to be logged as needing to be freed.  It should
+ * be called once for each extent to be freed.
+ */
+void
+xfs_trans_log_efi_extent(xfs_trans_t		*tp,
+			 xfs_efi_log_item_t	*efip,
+			 xfs_fsblock_t		start_block,
+			 xfs_extlen_t		ext_len)
+{
+	uint			next_extent;
+	xfs_extent_t		*extp;
+
+	tp->t_flags |= XFS_TRANS_DIRTY;
+	efip->efi_item.li_desc->lid_flags |= XFS_LID_DIRTY;
+
+	/*
+	 * atomic_inc_return gives us the value after the increment;
+	 * we want to use it as an array index so we need to subtract 1 from
+	 * it.
+	 */
+	next_extent = atomic_inc_return(&efip->efi_next_extent) - 1;
+	ASSERT(next_extent < efip->efi_format.efi_nextents);
+	extp = &(efip->efi_format.efi_extents[next_extent]);
+	extp->ext_start = start_block;
+	extp->ext_len = ext_len;
+}
+
+
+/*
+ * This routine is called to allocate an "extent free done"
+ * log item that will hold nextents worth of extents.  The
+ * caller must use all nextents extents, because we are not
+ * flexible about this at all.
+ */
+xfs_efd_log_item_t *
+xfs_trans_get_efd(xfs_trans_t		*tp,
+		  xfs_efi_log_item_t	*efip,
+		  uint			nextents)
+{
+	xfs_efd_log_item_t	*efdp;
+
+	ASSERT(tp != NULL);
+	ASSERT(nextents > 0);
+
+	efdp = xfs_efd_init(tp->t_mountp, efip, nextents);
+	ASSERT(efdp != NULL);
+
+	/*
+	 * Get a log_item_desc to point at the new item.
+	 */
+	xfs_trans_add_item(tp, &efdp->efd_item);
+	return efdp;
+}
+
+/*
+ * This routine is called to indicate that the described
+ * extent is to be logged as having been freed.  It should
+ * be called once for each extent freed.
+ */
+void
+xfs_trans_log_efd_extent(xfs_trans_t		*tp,
+			 xfs_efd_log_item_t	*efdp,
+			 xfs_fsblock_t		start_block,
+			 xfs_extlen_t		ext_len)
+{
+	uint			next_extent;
+	xfs_extent_t		*extp;
+
+	tp->t_flags |= XFS_TRANS_DIRTY;
+	efdp->efd_item.li_desc->lid_flags |= XFS_LID_DIRTY;
+
+	next_extent = efdp->efd_next_extent;
+	ASSERT(next_extent < efdp->efd_format.efd_nextents);
+	extp = &(efdp->efd_format.efd_extents[next_extent]);
+	extp->ext_start = start_block;
+	extp->ext_len = ext_len;
+	efdp->efd_next_extent++;
+}
diff --git a/fs/xfs/xfs_trans_inode.c b/fs/xfs/xfs_trans_inode.c
new file mode 100644
index 00000000..048b0c68
--- /dev/null
+++ b/fs/xfs/xfs_trans_inode.c
@@ -0,0 +1,190 @@
+/*
+ * Copyright (c) 2000,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_mount.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_alloc_btree.h"
+#include "xfs_ialloc_btree.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_btree.h"
+#include "xfs_trans_priv.h"
+#include "xfs_inode_item.h"
+#include "xfs_trace.h"
+
+#ifdef XFS_TRANS_DEBUG
+STATIC void
+xfs_trans_inode_broot_debug(
+	xfs_inode_t	*ip);
+#else
+#define	xfs_trans_inode_broot_debug(ip)
+#endif
+
+/*
+ * Add a locked inode to the transaction.
+ *
+ * The inode must be locked, and it cannot be associated with any transaction.
+ */
+void
+xfs_trans_ijoin(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*ip)
+{
+	xfs_inode_log_item_t	*iip;
+
+	ASSERT(ip->i_transp == NULL);
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
+	if (ip->i_itemp == NULL)
+		xfs_inode_item_init(ip, ip->i_mount);
+	iip = ip->i_itemp;
+	ASSERT(iip->ili_lock_flags == 0);
+
+	/*
+	 * Get a log_item_desc to point at the new item.
+	 */
+	xfs_trans_add_item(tp, &iip->ili_item);
+
+	xfs_trans_inode_broot_debug(ip);
+
+	/*
+	 * Initialize i_transp so we can find it with xfs_inode_incore()
+	 * in xfs_trans_iget() above.
+	 */
+	ip->i_transp = tp;
+}
+
+/*
+ * Add a locked inode to the transaction.
+ *
+ *
+ * Grabs a reference to the inode which will be dropped when the transaction
+ * is committed.  The inode will also be unlocked at that point.  The inode
+ * must be locked, and it cannot be associated with any transaction.
+ */
+void
+xfs_trans_ijoin_ref(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*ip,
+	uint			lock_flags)
+{
+	xfs_trans_ijoin(tp, ip);
+	IHOLD(ip);
+	ip->i_itemp->ili_lock_flags = lock_flags;
+}
+
+/*
+ * Transactional inode timestamp update. Requires the inode to be locked and
+ * joined to the transaction supplied. Relies on the transaction subsystem to
+ * track dirty state and update/writeback the inode accordingly.
+ */
+void
+xfs_trans_ichgtime(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*ip,
+	int			flags)
+{
+	struct inode		*inode = VFS_I(ip);
+	timespec_t		tv;
+
+	ASSERT(tp);
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
+	ASSERT(ip->i_transp == tp);
+
+	tv = current_fs_time(inode->i_sb);
+
+	if ((flags & XFS_ICHGTIME_MOD) &&
+	    !timespec_equal(&inode->i_mtime, &tv)) {
+		inode->i_mtime = tv;
+	}
+	if ((flags & XFS_ICHGTIME_CHG) &&
+	    !timespec_equal(&inode->i_ctime, &tv)) {
+		inode->i_ctime = tv;
+	}
+}
+
+/*
+ * This is called to mark the fields indicated in fieldmask as needing
+ * to be logged when the transaction is committed.  The inode must
+ * already be associated with the given transaction.
+ *
+ * The values for fieldmask are defined in xfs_inode_item.h.  We always
+ * log all of the core inode if any of it has changed, and we always log
+ * all of the inline data/extents/b-tree root if any of them has changed.
+ */
+void
+xfs_trans_log_inode(
+	xfs_trans_t	*tp,
+	xfs_inode_t	*ip,
+	uint		flags)
+{
+	ASSERT(ip->i_transp == tp);
+	ASSERT(ip->i_itemp != NULL);
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
+
+	tp->t_flags |= XFS_TRANS_DIRTY;
+	ip->i_itemp->ili_item.li_desc->lid_flags |= XFS_LID_DIRTY;
+
+	/*
+	 * Always OR in the bits from the ili_last_fields field.
+	 * This is to coordinate with the xfs_iflush() and xfs_iflush_done()
+	 * routines in the eventual clearing of the ilf_fields bits.
+	 * See the big comment in xfs_iflush() for an explanation of
+	 * this coordination mechanism.
+	 */
+	flags |= ip->i_itemp->ili_last_fields;
+	ip->i_itemp->ili_format.ilf_fields |= flags;
+}
+
+#ifdef XFS_TRANS_DEBUG
+/*
+ * Keep track of the state of the inode btree root to make sure we
+ * log it properly.
+ */
+STATIC void
+xfs_trans_inode_broot_debug(
+	xfs_inode_t	*ip)
+{
+	xfs_inode_log_item_t	*iip;
+
+	ASSERT(ip->i_itemp != NULL);
+	iip = ip->i_itemp;
+	if (iip->ili_root_size != 0) {
+		ASSERT(iip->ili_orig_root != NULL);
+		kmem_free(iip->ili_orig_root);
+		iip->ili_root_size = 0;
+		iip->ili_orig_root = NULL;
+	}
+	if (ip->i_d.di_format == XFS_DINODE_FMT_BTREE) {
+		ASSERT((ip->i_df.if_broot != NULL) &&
+		       (ip->i_df.if_broot_bytes > 0));
+		iip->ili_root_size = ip->i_df.if_broot_bytes;
+		iip->ili_orig_root =
+			(char*)kmem_alloc(iip->ili_root_size, KM_SLEEP);
+		memcpy(iip->ili_orig_root, (char*)(ip->i_df.if_broot),
+		      iip->ili_root_size);
+	}
+}
+#endif
diff --git a/fs/xfs/xfs_trans_priv.h b/fs/xfs/xfs_trans_priv.h
new file mode 100644
index 00000000..fe2e3cbc
--- /dev/null
+++ b/fs/xfs/xfs_trans_priv.h
@@ -0,0 +1,143 @@
+/*
+ * Copyright (c) 2000,2002,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_TRANS_PRIV_H__
+#define	__XFS_TRANS_PRIV_H__
+
+struct xfs_log_item;
+struct xfs_log_item_desc;
+struct xfs_mount;
+struct xfs_trans;
+struct xfs_ail;
+struct xfs_log_vec;
+
+void	xfs_trans_add_item(struct xfs_trans *, struct xfs_log_item *);
+void	xfs_trans_del_item(struct xfs_log_item *);
+void	xfs_trans_free_items(struct xfs_trans *tp, xfs_lsn_t commit_lsn,
+				int flags);
+void	xfs_trans_unreserve_and_mod_sb(struct xfs_trans *tp);
+
+void	xfs_trans_committed_bulk(struct xfs_ail *ailp, struct xfs_log_vec *lv,
+				xfs_lsn_t commit_lsn, int aborted);
+/*
+ * AIL traversal cursor.
+ *
+ * Rather than using a generation number for detecting changes in the ail, use
+ * a cursor that is protected by the ail lock. The aild cursor exists in the
+ * struct xfs_ail, but other traversals can declare it on the stack and link it
+ * to the ail list.
+ *
+ * When an object is deleted from or moved int the AIL, the cursor list is
+ * searched to see if the object is a designated cursor item. If it is, it is
+ * deleted from the cursor so that the next time the cursor is used traversal
+ * will return to the start.
+ *
+ * This means a traversal colliding with a removal will cause a restart of the
+ * list scan, rather than any insertion or deletion anywhere in the list. The
+ * low bit of the item pointer is set if the cursor has been invalidated so
+ * that we can tell the difference between invalidation and reaching the end
+ * of the list to trigger traversal restarts.
+ */
+struct xfs_ail_cursor {
+	struct xfs_ail_cursor	*next;
+	struct xfs_log_item	*item;
+};
+
+/*
+ * Private AIL structures.
+ *
+ * Eventually we need to drive the locking in here as well.
+ */
+struct xfs_ail {
+	struct xfs_mount	*xa_mount;
+	struct task_struct	*xa_task;
+	struct list_head	xa_ail;
+	xfs_lsn_t		xa_target;
+	struct xfs_ail_cursor	xa_cursors;
+	spinlock_t		xa_lock;
+	xfs_lsn_t		xa_last_pushed_lsn;
+};
+
+/*
+ * From xfs_trans_ail.c
+ */
+void	xfs_trans_ail_update_bulk(struct xfs_ail *ailp,
+				struct xfs_ail_cursor *cur,
+				struct xfs_log_item **log_items, int nr_items,
+				xfs_lsn_t lsn) __releases(ailp->xa_lock);
+static inline void
+xfs_trans_ail_update(
+	struct xfs_ail		*ailp,
+	struct xfs_log_item	*lip,
+	xfs_lsn_t		lsn) __releases(ailp->xa_lock)
+{
+	xfs_trans_ail_update_bulk(ailp, NULL, &lip, 1, lsn);
+}
+
+void	xfs_trans_ail_delete_bulk(struct xfs_ail *ailp,
+				struct xfs_log_item **log_items, int nr_items)
+				__releases(ailp->xa_lock);
+static inline void
+xfs_trans_ail_delete(
+	struct xfs_ail	*ailp,
+	xfs_log_item_t	*lip) __releases(ailp->xa_lock)
+{
+	xfs_trans_ail_delete_bulk(ailp, &lip, 1);
+}
+
+void			xfs_ail_push(struct xfs_ail *, xfs_lsn_t);
+void			xfs_ail_push_all(struct xfs_ail *);
+xfs_lsn_t		xfs_ail_min_lsn(struct xfs_ail *ailp);
+
+void			xfs_trans_unlocked_item(struct xfs_ail *,
+					xfs_log_item_t *);
+
+struct xfs_log_item *	xfs_trans_ail_cursor_first(struct xfs_ail *ailp,
+					struct xfs_ail_cursor *cur,
+					xfs_lsn_t lsn);
+struct xfs_log_item *	xfs_trans_ail_cursor_last(struct xfs_ail *ailp,
+					struct xfs_ail_cursor *cur,
+					xfs_lsn_t lsn);
+struct xfs_log_item *	xfs_trans_ail_cursor_next(struct xfs_ail *ailp,
+					struct xfs_ail_cursor *cur);
+void			xfs_trans_ail_cursor_done(struct xfs_ail *ailp,
+					struct xfs_ail_cursor *cur);
+
+#if BITS_PER_LONG != 64
+static inline void
+xfs_trans_ail_copy_lsn(
+	struct xfs_ail	*ailp,
+	xfs_lsn_t	*dst,
+	xfs_lsn_t	*src)
+{
+	ASSERT(sizeof(xfs_lsn_t) == 8);	/* don't lock if it shrinks */
+	spin_lock(&ailp->xa_lock);
+	*dst = *src;
+	spin_unlock(&ailp->xa_lock);
+}
+#else
+static inline void
+xfs_trans_ail_copy_lsn(
+	struct xfs_ail	*ailp,
+	xfs_lsn_t	*dst,
+	xfs_lsn_t	*src)
+{
+	ASSERT(sizeof(xfs_lsn_t) == 8);
+	*dst = *src;
+}
+#endif
+#endif	/* __XFS_TRANS_PRIV_H__ */
diff --git a/fs/xfs/xfs_trans_space.h b/fs/xfs/xfs_trans_space.h
new file mode 100644
index 00000000..7d2c920d
--- /dev/null
+++ b/fs/xfs/xfs_trans_space.h
@@ -0,0 +1,86 @@
+/*
+ * Copyright (c) 2000,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_TRANS_SPACE_H__
+#define __XFS_TRANS_SPACE_H__
+
+/*
+ * Components of space reservations.
+ */
+#define XFS_MAX_CONTIG_EXTENTS_PER_BLOCK(mp)    \
+		(((mp)->m_alloc_mxr[0]) - ((mp)->m_alloc_mnr[0]))
+#define	XFS_EXTENTADD_SPACE_RES(mp,w)	(XFS_BM_MAXLEVELS(mp,w) - 1)
+#define XFS_NEXTENTADD_SPACE_RES(mp,b,w)\
+	(((b + XFS_MAX_CONTIG_EXTENTS_PER_BLOCK(mp) - 1) / \
+	  XFS_MAX_CONTIG_EXTENTS_PER_BLOCK(mp)) * \
+	  XFS_EXTENTADD_SPACE_RES(mp,w))
+#define	XFS_DAENTER_1B(mp,w)	((w) == XFS_DATA_FORK ? (mp)->m_dirblkfsbs : 1)
+#define	XFS_DAENTER_DBS(mp,w)	\
+	(XFS_DA_NODE_MAXDEPTH + (((w) == XFS_DATA_FORK) ? 2 : 0))
+#define	XFS_DAENTER_BLOCKS(mp,w)	\
+	(XFS_DAENTER_1B(mp,w) * XFS_DAENTER_DBS(mp,w))
+#define	XFS_DAENTER_BMAP1B(mp,w)	\
+	XFS_NEXTENTADD_SPACE_RES(mp, XFS_DAENTER_1B(mp, w), w)
+#define	XFS_DAENTER_BMAPS(mp,w)		\
+	(XFS_DAENTER_DBS(mp,w) * XFS_DAENTER_BMAP1B(mp,w))
+#define	XFS_DAENTER_SPACE_RES(mp,w)	\
+	(XFS_DAENTER_BLOCKS(mp,w) + XFS_DAENTER_BMAPS(mp,w))
+#define	XFS_DAREMOVE_SPACE_RES(mp,w)	XFS_DAENTER_BMAPS(mp,w)
+#define	XFS_DIRENTER_MAX_SPLIT(mp,nl)	1
+#define	XFS_DIRENTER_SPACE_RES(mp,nl)	\
+	(XFS_DAENTER_SPACE_RES(mp, XFS_DATA_FORK) * \
+	 XFS_DIRENTER_MAX_SPLIT(mp,nl))
+#define	XFS_DIRREMOVE_SPACE_RES(mp)	\
+	XFS_DAREMOVE_SPACE_RES(mp, XFS_DATA_FORK)
+#define	XFS_IALLOC_SPACE_RES(mp)	\
+	(XFS_IALLOC_BLOCKS(mp) + (mp)->m_in_maxlevels - 1)
+
+/*
+ * Space reservation values for various transactions.
+ */
+#define	XFS_ADDAFORK_SPACE_RES(mp)	\
+	((mp)->m_dirblkfsbs + XFS_DAENTER_BMAP1B(mp, XFS_DATA_FORK))
+#define	XFS_ATTRRM_SPACE_RES(mp)	\
+	XFS_DAREMOVE_SPACE_RES(mp, XFS_ATTR_FORK)
+/* This macro is not used - see inline code in xfs_attr_set */
+#define	XFS_ATTRSET_SPACE_RES(mp, v)	\
+	(XFS_DAENTER_SPACE_RES(mp, XFS_ATTR_FORK) + XFS_B_TO_FSB(mp, v))
+#define	XFS_CREATE_SPACE_RES(mp,nl)	\
+	(XFS_IALLOC_SPACE_RES(mp) + XFS_DIRENTER_SPACE_RES(mp,nl))
+#define	XFS_DIOSTRAT_SPACE_RES(mp, v)	\
+	(XFS_EXTENTADD_SPACE_RES(mp, XFS_DATA_FORK) + (v))
+#define	XFS_GROWFS_SPACE_RES(mp)	\
+	(2 * XFS_AG_MAXLEVELS(mp))
+#define	XFS_GROWFSRT_SPACE_RES(mp,b)	\
+	((b) + XFS_EXTENTADD_SPACE_RES(mp, XFS_DATA_FORK))
+#define	XFS_LINK_SPACE_RES(mp,nl)	\
+	XFS_DIRENTER_SPACE_RES(mp,nl)
+#define	XFS_MKDIR_SPACE_RES(mp,nl)	\
+	(XFS_IALLOC_SPACE_RES(mp) + XFS_DIRENTER_SPACE_RES(mp,nl))
+#define	XFS_QM_DQALLOC_SPACE_RES(mp)	\
+	(XFS_EXTENTADD_SPACE_RES(mp, XFS_DATA_FORK) + \
+	 XFS_DQUOT_CLUSTER_SIZE_FSB)
+#define	XFS_QM_QINOCREATE_SPACE_RES(mp)	\
+	XFS_IALLOC_SPACE_RES(mp)
+#define	XFS_REMOVE_SPACE_RES(mp)	\
+	XFS_DIRREMOVE_SPACE_RES(mp)
+#define	XFS_RENAME_SPACE_RES(mp,nl)	\
+	(XFS_DIRREMOVE_SPACE_RES(mp) + XFS_DIRENTER_SPACE_RES(mp,nl))
+#define	XFS_SYMLINK_SPACE_RES(mp,nl,b)	\
+	(XFS_IALLOC_SPACE_RES(mp) + XFS_DIRENTER_SPACE_RES(mp,nl) + (b))
+
+#endif	/* __XFS_TRANS_SPACE_H__ */
diff --git a/fs/xfs/xfs_types.h b/fs/xfs/xfs_types.h
new file mode 100644
index 00000000..65584b55
--- /dev/null
+++ b/fs/xfs/xfs_types.h
@@ -0,0 +1,156 @@
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_TYPES_H__
+#define	__XFS_TYPES_H__
+
+#ifdef __KERNEL__
+
+/*
+ * Additional type declarations for XFS
+ */
+typedef signed char		__int8_t;
+typedef unsigned char		__uint8_t;
+typedef signed short int	__int16_t;
+typedef unsigned short int	__uint16_t;
+typedef signed int		__int32_t;
+typedef unsigned int		__uint32_t;
+typedef signed long long int	__int64_t;
+typedef unsigned long long int	__uint64_t;
+
+typedef enum { B_FALSE,B_TRUE }	boolean_t;
+typedef __uint32_t		prid_t;		/* project ID */
+typedef __uint32_t		inst_t;		/* an instruction */
+
+typedef __s64			xfs_off_t;	/* <file offset> type */
+typedef unsigned long long	xfs_ino_t;	/* <inode> type */
+typedef __s64			xfs_daddr_t;	/* <disk address> type */
+typedef char *			xfs_caddr_t;	/* <core address> type */
+typedef __u32			xfs_dev_t;
+typedef __u32			xfs_nlink_t;
+
+/* __psint_t is the same size as a pointer */
+#if (BITS_PER_LONG == 32)
+typedef __int32_t __psint_t;
+typedef __uint32_t __psunsigned_t;
+#elif (BITS_PER_LONG == 64)
+typedef __int64_t __psint_t;
+typedef __uint64_t __psunsigned_t;
+#else
+#error BITS_PER_LONG must be 32 or 64
+#endif
+
+#endif	/* __KERNEL__ */
+
+typedef __uint32_t	xfs_agblock_t;	/* blockno in alloc. group */
+typedef	__uint32_t	xfs_extlen_t;	/* extent length in blocks */
+typedef	__uint32_t	xfs_agnumber_t;	/* allocation group number */
+typedef __int32_t	xfs_extnum_t;	/* # of extents in a file */
+typedef __int16_t	xfs_aextnum_t;	/* # extents in an attribute fork */
+typedef	__int64_t	xfs_fsize_t;	/* bytes in a file */
+typedef __uint64_t	xfs_ufsize_t;	/* unsigned bytes in a file */
+
+typedef	__int32_t	xfs_suminfo_t;	/* type of bitmap summary info */
+typedef	__int32_t	xfs_rtword_t;	/* word type for bitmap manipulations */
+
+typedef	__int64_t	xfs_lsn_t;	/* log sequence number */
+typedef	__int32_t	xfs_tid_t;	/* transaction identifier */
+
+typedef	__uint32_t	xfs_dablk_t;	/* dir/attr block number (in file) */
+typedef	__uint32_t	xfs_dahash_t;	/* dir/attr hash value */
+
+/*
+ * These types are 64 bits on disk but are either 32 or 64 bits in memory.
+ * Disk based types:
+ */
+typedef __uint64_t	xfs_dfsbno_t;	/* blockno in filesystem (agno|agbno) */
+typedef __uint64_t	xfs_drfsbno_t;	/* blockno in filesystem (raw) */
+typedef	__uint64_t	xfs_drtbno_t;	/* extent (block) in realtime area */
+typedef	__uint64_t	xfs_dfiloff_t;	/* block number in a file */
+typedef	__uint64_t	xfs_dfilblks_t;	/* number of blocks in a file */
+
+/*
+ * Memory based types are conditional.
+ */
+#if XFS_BIG_BLKNOS
+typedef	__uint64_t	xfs_fsblock_t;	/* blockno in filesystem (agno|agbno) */
+typedef __uint64_t	xfs_rfsblock_t;	/* blockno in filesystem (raw) */
+typedef __uint64_t	xfs_rtblock_t;	/* extent (block) in realtime area */
+typedef	__int64_t	xfs_srtblock_t;	/* signed version of xfs_rtblock_t */
+#else
+typedef	__uint32_t	xfs_fsblock_t;	/* blockno in filesystem (agno|agbno) */
+typedef __uint32_t	xfs_rfsblock_t;	/* blockno in filesystem (raw) */
+typedef __uint32_t	xfs_rtblock_t;	/* extent (block) in realtime area */
+typedef	__int32_t	xfs_srtblock_t;	/* signed version of xfs_rtblock_t */
+#endif
+typedef __uint64_t	xfs_fileoff_t;	/* block number in a file */
+typedef __int64_t	xfs_sfiloff_t;	/* signed block number in a file */
+typedef __uint64_t	xfs_filblks_t;	/* number of blocks in a file */
+
+/*
+ * Null values for the types.
+ */
+#define	NULLDFSBNO	((xfs_dfsbno_t)-1)
+#define	NULLDRFSBNO	((xfs_drfsbno_t)-1)
+#define	NULLDRTBNO	((xfs_drtbno_t)-1)
+#define	NULLDFILOFF	((xfs_dfiloff_t)-1)
+
+#define	NULLFSBLOCK	((xfs_fsblock_t)-1)
+#define	NULLRFSBLOCK	((xfs_rfsblock_t)-1)
+#define	NULLRTBLOCK	((xfs_rtblock_t)-1)
+#define	NULLFILEOFF	((xfs_fileoff_t)-1)
+
+#define	NULLAGBLOCK	((xfs_agblock_t)-1)
+#define	NULLAGNUMBER	((xfs_agnumber_t)-1)
+#define	NULLEXTNUM	((xfs_extnum_t)-1)
+
+#define NULLCOMMITLSN	((xfs_lsn_t)-1)
+
+/*
+ * Max values for extlen, extnum, aextnum.
+ */
+#define	MAXEXTLEN	((xfs_extlen_t)0x001fffff)	/* 21 bits */
+#define	MAXEXTNUM	((xfs_extnum_t)0x7fffffff)	/* signed int */
+#define	MAXAEXTNUM	((xfs_aextnum_t)0x7fff)		/* signed short */
+
+/*
+ * Min numbers of data/attr fork btree root pointers.
+ */
+#define MINDBTPTRS	3
+#define MINABTPTRS	2
+
+/*
+ * MAXNAMELEN is the length (including the terminating null) of
+ * the longest permissible file (component) name.
+ */
+#define MAXNAMELEN	256
+
+typedef enum {
+	XFS_LOOKUP_EQi, XFS_LOOKUP_LEi, XFS_LOOKUP_GEi
+} xfs_lookup_t;
+
+typedef enum {
+	XFS_BTNUM_BNOi, XFS_BTNUM_CNTi, XFS_BTNUM_BMAPi, XFS_BTNUM_INOi,
+	XFS_BTNUM_MAX
+} xfs_btnum_t;
+
+struct xfs_name {
+	const unsigned char	*name;
+	int			len;
+};
+
+#endif	/* __XFS_TYPES_H__ */
diff --git a/fs/xfs/xfs_utils.c b/fs/xfs/xfs_utils.c
new file mode 100644
index 00000000..8b32d1a4
--- /dev/null
+++ b/fs/xfs/xfs_utils.c
@@ -0,0 +1,321 @@
+/*
+ * Copyright (c) 2000-2002,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_dir2.h"
+#include "xfs_mount.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_inode_item.h"
+#include "xfs_bmap.h"
+#include "xfs_error.h"
+#include "xfs_quota.h"
+#include "xfs_itable.h"
+#include "xfs_utils.h"
+
+
+/*
+ * Allocates a new inode from disk and return a pointer to the
+ * incore copy. This routine will internally commit the current
+ * transaction and allocate a new one if the Space Manager needed
+ * to do an allocation to replenish the inode free-list.
+ *
+ * This routine is designed to be called from xfs_create and
+ * xfs_create_dir.
+ *
+ */
+int
+xfs_dir_ialloc(
+	xfs_trans_t	**tpp,		/* input: current transaction;
+					   output: may be a new transaction. */
+	xfs_inode_t	*dp,		/* directory within whose allocate
+					   the inode. */
+	mode_t		mode,
+	xfs_nlink_t	nlink,
+	xfs_dev_t	rdev,
+	prid_t		prid,		/* project id */
+	int		okalloc,	/* ok to allocate new space */
+	xfs_inode_t	**ipp,		/* pointer to inode; it will be
+					   locked. */
+	int		*committed)
+
+{
+	xfs_trans_t	*tp;
+	xfs_trans_t	*ntp;
+	xfs_inode_t	*ip;
+	xfs_buf_t	*ialloc_context = NULL;
+	boolean_t	call_again = B_FALSE;
+	int		code;
+	uint		log_res;
+	uint		log_count;
+	void		*dqinfo;
+	uint		tflags;
+
+	tp = *tpp;
+	ASSERT(tp->t_flags & XFS_TRANS_PERM_LOG_RES);
+
+	/*
+	 * xfs_ialloc will return a pointer to an incore inode if
+	 * the Space Manager has an available inode on the free
+	 * list. Otherwise, it will do an allocation and replenish
+	 * the freelist.  Since we can only do one allocation per
+	 * transaction without deadlocks, we will need to commit the
+	 * current transaction and start a new one.  We will then
+	 * need to call xfs_ialloc again to get the inode.
+	 *
+	 * If xfs_ialloc did an allocation to replenish the freelist,
+	 * it returns the bp containing the head of the freelist as
+	 * ialloc_context. We will hold a lock on it across the
+	 * transaction commit so that no other process can steal
+	 * the inode(s) that we've just allocated.
+	 */
+	code = xfs_ialloc(tp, dp, mode, nlink, rdev, prid, okalloc,
+			  &ialloc_context, &call_again, &ip);
+
+	/*
+	 * Return an error if we were unable to allocate a new inode.
+	 * This should only happen if we run out of space on disk or
+	 * encounter a disk error.
+	 */
+	if (code) {
+		*ipp = NULL;
+		return code;
+	}
+	if (!call_again && (ip == NULL)) {
+		*ipp = NULL;
+		return XFS_ERROR(ENOSPC);
+	}
+
+	/*
+	 * If call_again is set, then we were unable to get an
+	 * inode in one operation.  We need to commit the current
+	 * transaction and call xfs_ialloc() again.  It is guaranteed
+	 * to succeed the second time.
+	 */
+	if (call_again) {
+
+		/*
+		 * Normally, xfs_trans_commit releases all the locks.
+		 * We call bhold to hang on to the ialloc_context across
+		 * the commit.  Holding this buffer prevents any other
+		 * processes from doing any allocations in this
+		 * allocation group.
+		 */
+		xfs_trans_bhold(tp, ialloc_context);
+		/*
+		 * Save the log reservation so we can use
+		 * them in the next transaction.
+		 */
+		log_res = xfs_trans_get_log_res(tp);
+		log_count = xfs_trans_get_log_count(tp);
+
+		/*
+		 * We want the quota changes to be associated with the next
+		 * transaction, NOT this one. So, detach the dqinfo from this
+		 * and attach it to the next transaction.
+		 */
+		dqinfo = NULL;
+		tflags = 0;
+		if (tp->t_dqinfo) {
+			dqinfo = (void *)tp->t_dqinfo;
+			tp->t_dqinfo = NULL;
+			tflags = tp->t_flags & XFS_TRANS_DQ_DIRTY;
+			tp->t_flags &= ~(XFS_TRANS_DQ_DIRTY);
+		}
+
+		ntp = xfs_trans_dup(tp);
+		code = xfs_trans_commit(tp, 0);
+		tp = ntp;
+		if (committed != NULL) {
+			*committed = 1;
+		}
+		/*
+		 * If we get an error during the commit processing,
+		 * release the buffer that is still held and return
+		 * to the caller.
+		 */
+		if (code) {
+			xfs_buf_relse(ialloc_context);
+			if (dqinfo) {
+				tp->t_dqinfo = dqinfo;
+				xfs_trans_free_dqinfo(tp);
+			}
+			*tpp = ntp;
+			*ipp = NULL;
+			return code;
+		}
+
+		/*
+		 * transaction commit worked ok so we can drop the extra ticket
+		 * reference that we gained in xfs_trans_dup()
+		 */
+		xfs_log_ticket_put(tp->t_ticket);
+		code = xfs_trans_reserve(tp, 0, log_res, 0,
+					 XFS_TRANS_PERM_LOG_RES, log_count);
+		/*
+		 * Re-attach the quota info that we detached from prev trx.
+		 */
+		if (dqinfo) {
+			tp->t_dqinfo = dqinfo;
+			tp->t_flags |= tflags;
+		}
+
+		if (code) {
+			xfs_buf_relse(ialloc_context);
+			*tpp = ntp;
+			*ipp = NULL;
+			return code;
+		}
+		xfs_trans_bjoin(tp, ialloc_context);
+
+		/*
+		 * Call ialloc again. Since we've locked out all
+		 * other allocations in this allocation group,
+		 * this call should always succeed.
+		 */
+		code = xfs_ialloc(tp, dp, mode, nlink, rdev, prid,
+				  okalloc, &ialloc_context, &call_again, &ip);
+
+		/*
+		 * If we get an error at this point, return to the caller
+		 * so that the current transaction can be aborted.
+		 */
+		if (code) {
+			*tpp = tp;
+			*ipp = NULL;
+			return code;
+		}
+		ASSERT ((!call_again) && (ip != NULL));
+
+	} else {
+		if (committed != NULL) {
+			*committed = 0;
+		}
+	}
+
+	*ipp = ip;
+	*tpp = tp;
+
+	return 0;
+}
+
+/*
+ * Decrement the link count on an inode & log the change.
+ * If this causes the link count to go to zero, initiate the
+ * logging activity required to truncate a file.
+ */
+int				/* error */
+xfs_droplink(
+	xfs_trans_t *tp,
+	xfs_inode_t *ip)
+{
+	int	error;
+
+	xfs_trans_ichgtime(tp, ip, XFS_ICHGTIME_CHG);
+
+	ASSERT (ip->i_d.di_nlink > 0);
+	ip->i_d.di_nlink--;
+	drop_nlink(VFS_I(ip));
+	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
+
+	error = 0;
+	if (ip->i_d.di_nlink == 0) {
+		/*
+		 * We're dropping the last link to this file.
+		 * Move the on-disk inode to the AGI unlinked list.
+		 * From xfs_inactive() we will pull the inode from
+		 * the list and free it.
+		 */
+		error = xfs_iunlink(tp, ip);
+	}
+	return error;
+}
+
+/*
+ * This gets called when the inode's version needs to be changed from 1 to 2.
+ * Currently this happens when the nlink field overflows the old 16-bit value
+ * or when chproj is called to change the project for the first time.
+ * As a side effect the superblock version will also get rev'd
+ * to contain the NLINK bit.
+ */
+void
+xfs_bump_ino_vers2(
+	xfs_trans_t	*tp,
+	xfs_inode_t	*ip)
+{
+	xfs_mount_t	*mp;
+
+	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
+	ASSERT(ip->i_d.di_version == 1);
+
+	ip->i_d.di_version = 2;
+	ip->i_d.di_onlink = 0;
+	memset(&(ip->i_d.di_pad[0]), 0, sizeof(ip->i_d.di_pad));
+	mp = tp->t_mountp;
+	if (!xfs_sb_version_hasnlink(&mp->m_sb)) {
+		spin_lock(&mp->m_sb_lock);
+		if (!xfs_sb_version_hasnlink(&mp->m_sb)) {
+			xfs_sb_version_addnlink(&mp->m_sb);
+			spin_unlock(&mp->m_sb_lock);
+			xfs_mod_sb(tp, XFS_SB_VERSIONNUM);
+		} else {
+			spin_unlock(&mp->m_sb_lock);
+		}
+	}
+	/* Caller must log the inode */
+}
+
+/*
+ * Increment the link count on an inode & log the change.
+ */
+int
+xfs_bumplink(
+	xfs_trans_t *tp,
+	xfs_inode_t *ip)
+{
+	if (ip->i_d.di_nlink >= XFS_MAXLINK)
+		return XFS_ERROR(EMLINK);
+	xfs_trans_ichgtime(tp, ip, XFS_ICHGTIME_CHG);
+
+	ASSERT(ip->i_d.di_nlink > 0);
+	ip->i_d.di_nlink++;
+	inc_nlink(VFS_I(ip));
+	if ((ip->i_d.di_version == 1) &&
+	    (ip->i_d.di_nlink > XFS_MAXLINK_1)) {
+		/*
+		 * The inode has increased its number of links beyond
+		 * what can fit in an old format inode.  It now needs
+		 * to be converted to a version 2 inode with a 32 bit
+		 * link count.  If this is the first inode in the file
+		 * system to do this, then we need to bump the superblock
+		 * version number as well.
+		 */
+		xfs_bump_ino_vers2(tp, ip);
+	}
+
+	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
+	return 0;
+}
diff --git a/fs/xfs/xfs_utils.h b/fs/xfs/xfs_utils.h
new file mode 100644
index 00000000..456fca31
--- /dev/null
+++ b/fs/xfs/xfs_utils.h
@@ -0,0 +1,27 @@
+/*
+ * Copyright (c) 2000-2002,2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_UTILS_H__
+#define __XFS_UTILS_H__
+
+extern int xfs_dir_ialloc(xfs_trans_t **, xfs_inode_t *, mode_t, xfs_nlink_t,
+				xfs_dev_t, prid_t, int, xfs_inode_t **, int *);
+extern int xfs_droplink(xfs_trans_t *, xfs_inode_t *);
+extern int xfs_bumplink(xfs_trans_t *, xfs_inode_t *);
+extern void xfs_bump_ino_vers2(xfs_trans_t *, xfs_inode_t *);
+
+#endif	/* __XFS_UTILS_H__ */
diff --git a/fs/xfs/xfs_vnodeops.c b/fs/xfs/xfs_vnodeops.c
new file mode 100644
index 00000000..59509ae0
--- /dev/null
+++ b/fs/xfs/xfs_vnodeops.c
@@ -0,0 +1,2852 @@
+/*
+ * Copyright (c) 2000-2006 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_dir2.h"
+#include "xfs_mount.h"
+#include "xfs_da_btree.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_ialloc_btree.h"
+#include "xfs_dinode.h"
+#include "xfs_inode.h"
+#include "xfs_inode_item.h"
+#include "xfs_itable.h"
+#include "xfs_ialloc.h"
+#include "xfs_alloc.h"
+#include "xfs_bmap.h"
+#include "xfs_acl.h"
+#include "xfs_attr.h"
+#include "xfs_rw.h"
+#include "xfs_error.h"
+#include "xfs_quota.h"
+#include "xfs_utils.h"
+#include "xfs_rtalloc.h"
+#include "xfs_trans_space.h"
+#include "xfs_log_priv.h"
+#include "xfs_filestream.h"
+#include "xfs_vnodeops.h"
+#include "xfs_trace.h"
+
+int
+xfs_setattr(
+	struct xfs_inode	*ip,
+	struct iattr		*iattr,
+	int			flags)
+{
+	xfs_mount_t		*mp = ip->i_mount;
+	struct inode		*inode = VFS_I(ip);
+	int			mask = iattr->ia_valid;
+	xfs_trans_t		*tp;
+	int			code;
+	uint			lock_flags;
+	uint			commit_flags=0;
+	uid_t			uid=0, iuid=0;
+	gid_t			gid=0, igid=0;
+	struct xfs_dquot	*udqp, *gdqp, *olddquot1, *olddquot2;
+	int			need_iolock = 1;
+
+	trace_xfs_setattr(ip);
+
+	if (mp->m_flags & XFS_MOUNT_RDONLY)
+		return XFS_ERROR(EROFS);
+
+	if (XFS_FORCED_SHUTDOWN(mp))
+		return XFS_ERROR(EIO);
+
+	code = -inode_change_ok(inode, iattr);
+	if (code)
+		return code;
+
+	olddquot1 = olddquot2 = NULL;
+	udqp = gdqp = NULL;
+
+	/*
+	 * If disk quotas is on, we make sure that the dquots do exist on disk,
+	 * before we start any other transactions. Trying to do this later
+	 * is messy. We don't care to take a readlock to look at the ids
+	 * in inode here, because we can't hold it across the trans_reserve.
+	 * If the IDs do change before we take the ilock, we're covered
+	 * because the i_*dquot fields will get updated anyway.
+	 */
+	if (XFS_IS_QUOTA_ON(mp) && (mask & (ATTR_UID|ATTR_GID))) {
+		uint	qflags = 0;
+
+		if ((mask & ATTR_UID) && XFS_IS_UQUOTA_ON(mp)) {
+			uid = iattr->ia_uid;
+			qflags |= XFS_QMOPT_UQUOTA;
+		} else {
+			uid = ip->i_d.di_uid;
+		}
+		if ((mask & ATTR_GID) && XFS_IS_GQUOTA_ON(mp)) {
+			gid = iattr->ia_gid;
+			qflags |= XFS_QMOPT_GQUOTA;
+		}  else {
+			gid = ip->i_d.di_gid;
+		}
+
+		/*
+		 * We take a reference when we initialize udqp and gdqp,
+		 * so it is important that we never blindly double trip on
+		 * the same variable. See xfs_create() for an example.
+		 */
+		ASSERT(udqp == NULL);
+		ASSERT(gdqp == NULL);
+		code = xfs_qm_vop_dqalloc(ip, uid, gid, xfs_get_projid(ip),
+					 qflags, &udqp, &gdqp);
+		if (code)
+			return code;
+	}
+
+	/*
+	 * For the other attributes, we acquire the inode lock and
+	 * first do an error checking pass.
+	 */
+	tp = NULL;
+	lock_flags = XFS_ILOCK_EXCL;
+	if (flags & XFS_ATTR_NOLOCK)
+		need_iolock = 0;
+	if (!(mask & ATTR_SIZE)) {
+		tp = xfs_trans_alloc(mp, XFS_TRANS_SETATTR_NOT_SIZE);
+		commit_flags = 0;
+		code = xfs_trans_reserve(tp, 0, XFS_ICHANGE_LOG_RES(mp),
+					 0, 0, 0);
+		if (code) {
+			lock_flags = 0;
+			goto error_return;
+		}
+	} else {
+		if (need_iolock)
+			lock_flags |= XFS_IOLOCK_EXCL;
+	}
+
+	xfs_ilock(ip, lock_flags);
+
+	/*
+	 * Change file ownership.  Must be the owner or privileged.
+	 */
+	if (mask & (ATTR_UID|ATTR_GID)) {
+		/*
+		 * These IDs could have changed since we last looked at them.
+		 * But, we're assured that if the ownership did change
+		 * while we didn't have the inode locked, inode's dquot(s)
+		 * would have changed also.
+		 */
+		iuid = ip->i_d.di_uid;
+		igid = ip->i_d.di_gid;
+		gid = (mask & ATTR_GID) ? iattr->ia_gid : igid;
+		uid = (mask & ATTR_UID) ? iattr->ia_uid : iuid;
+
+		/*
+		 * Do a quota reservation only if uid/gid is actually
+		 * going to change.
+		 */
+		if (XFS_IS_QUOTA_RUNNING(mp) &&
+		    ((XFS_IS_UQUOTA_ON(mp) && iuid != uid) ||
+		     (XFS_IS_GQUOTA_ON(mp) && igid != gid))) {
+			ASSERT(tp);
+			code = xfs_qm_vop_chown_reserve(tp, ip, udqp, gdqp,
+						capable(CAP_FOWNER) ?
+						XFS_QMOPT_FORCE_RES : 0);
+			if (code)	/* out of quota */
+				goto error_return;
+		}
+	}
+
+	/*
+	 * Truncate file.  Must have write permission and not be a directory.
+	 */
+	if (mask & ATTR_SIZE) {
+		/* Short circuit the truncate case for zero length files */
+		if (iattr->ia_size == 0 &&
+		    ip->i_size == 0 && ip->i_d.di_nextents == 0) {
+			xfs_iunlock(ip, XFS_ILOCK_EXCL);
+			lock_flags &= ~XFS_ILOCK_EXCL;
+			if (mask & ATTR_CTIME) {
+				inode->i_mtime = inode->i_ctime =
+						current_fs_time(inode->i_sb);
+				xfs_mark_inode_dirty_sync(ip);
+			}
+			code = 0;
+			goto error_return;
+		}
+
+		if (S_ISDIR(ip->i_d.di_mode)) {
+			code = XFS_ERROR(EISDIR);
+			goto error_return;
+		} else if (!S_ISREG(ip->i_d.di_mode)) {
+			code = XFS_ERROR(EINVAL);
+			goto error_return;
+		}
+
+		/*
+		 * Make sure that the dquots are attached to the inode.
+		 */
+		code = xfs_qm_dqattach_locked(ip, 0);
+		if (code)
+			goto error_return;
+
+		/*
+		 * Now we can make the changes.  Before we join the inode
+		 * to the transaction, if ATTR_SIZE is set then take care of
+		 * the part of the truncation that must be done without the
+		 * inode lock.  This needs to be done before joining the inode
+		 * to the transaction, because the inode cannot be unlocked
+		 * once it is a part of the transaction.
+		 */
+		if (iattr->ia_size > ip->i_size) {
+			/*
+			 * Do the first part of growing a file: zero any data
+			 * in the last block that is beyond the old EOF.  We
+			 * need to do this before the inode is joined to the
+			 * transaction to modify the i_size.
+			 */
+			code = xfs_zero_eof(ip, iattr->ia_size, ip->i_size);
+			if (code)
+				goto error_return;
+		}
+		xfs_iunlock(ip, XFS_ILOCK_EXCL);
+		lock_flags &= ~XFS_ILOCK_EXCL;
+
+		/*
+		 * We are going to log the inode size change in this
+		 * transaction so any previous writes that are beyond the on
+		 * disk EOF and the new EOF that have not been written out need
+		 * to be written here. If we do not write the data out, we
+		 * expose ourselves to the null files problem.
+		 *
+		 * Only flush from the on disk size to the smaller of the in
+		 * memory file size or the new size as that's the range we
+		 * really care about here and prevents waiting for other data
+		 * not within the range we care about here.
+		 */
+		if (ip->i_size != ip->i_d.di_size &&
+		    iattr->ia_size > ip->i_d.di_size) {
+			code = xfs_flush_pages(ip,
+					ip->i_d.di_size, iattr->ia_size,
+					XBF_ASYNC, FI_NONE);
+			if (code)
+				goto error_return;
+		}
+
+		/* wait for all I/O to complete */
+		xfs_ioend_wait(ip);
+
+		code = -block_truncate_page(inode->i_mapping, iattr->ia_size,
+					    xfs_get_blocks);
+		if (code)
+			goto error_return;
+
+		tp = xfs_trans_alloc(mp, XFS_TRANS_SETATTR_SIZE);
+		code = xfs_trans_reserve(tp, 0, XFS_ITRUNCATE_LOG_RES(mp), 0,
+					 XFS_TRANS_PERM_LOG_RES,
+					 XFS_ITRUNCATE_LOG_COUNT);
+		if (code)
+			goto error_return;
+
+		truncate_setsize(inode, iattr->ia_size);
+
+		commit_flags = XFS_TRANS_RELEASE_LOG_RES;
+		lock_flags |= XFS_ILOCK_EXCL;
+
+		xfs_ilock(ip, XFS_ILOCK_EXCL);
+
+		xfs_trans_ijoin(tp, ip);
+
+		/*
+		 * Only change the c/mtime if we are changing the size
+		 * or we are explicitly asked to change it. This handles
+		 * the semantic difference between truncate() and ftruncate()
+		 * as implemented in the VFS.
+		 *
+		 * The regular truncate() case without ATTR_CTIME and ATTR_MTIME
+		 * is a special case where we need to update the times despite
+		 * not having these flags set.  For all other operations the
+		 * VFS set these flags explicitly if it wants a timestamp
+		 * update.
+		 */
+		if (iattr->ia_size != ip->i_size &&
+		    (!(mask & (ATTR_CTIME | ATTR_MTIME)))) {
+			iattr->ia_ctime = iattr->ia_mtime =
+				current_fs_time(inode->i_sb);
+			mask |= ATTR_CTIME | ATTR_MTIME;
+		}
+
+		if (iattr->ia_size > ip->i_size) {
+			ip->i_d.di_size = iattr->ia_size;
+			ip->i_size = iattr->ia_size;
+			xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
+		} else if (iattr->ia_size <= ip->i_size ||
+			   (iattr->ia_size == 0 && ip->i_d.di_nextents)) {
+			/*
+			 * signal a sync transaction unless
+			 * we're truncating an already unlinked
+			 * file on a wsync filesystem
+			 */
+			code = xfs_itruncate_finish(&tp, ip, iattr->ia_size,
+					    XFS_DATA_FORK,
+					    ((ip->i_d.di_nlink != 0 ||
+					      !(mp->m_flags & XFS_MOUNT_WSYNC))
+					     ? 1 : 0));
+			if (code)
+				goto abort_return;
+			/*
+			 * Truncated "down", so we're removing references
+			 * to old data here - if we now delay flushing for
+			 * a long time, we expose ourselves unduly to the
+			 * notorious NULL files problem.  So, we mark this
+			 * vnode and flush it when the file is closed, and
+			 * do not wait the usual (long) time for writeout.
+			 */
+			xfs_iflags_set(ip, XFS_ITRUNCATED);
+		}
+	} else if (tp) {
+		xfs_trans_ijoin(tp, ip);
+	}
+
+	/*
+	 * Change file ownership.  Must be the owner or privileged.
+	 */
+	if (mask & (ATTR_UID|ATTR_GID)) {
+		/*
+		 * CAP_FSETID overrides the following restrictions:
+		 *
+		 * The set-user-ID and set-group-ID bits of a file will be
+		 * cleared upon successful return from chown()
+		 */
+		if ((ip->i_d.di_mode & (S_ISUID|S_ISGID)) &&
+		    !capable(CAP_FSETID)) {
+			ip->i_d.di_mode &= ~(S_ISUID|S_ISGID);
+		}
+
+		/*
+		 * Change the ownerships and register quota modifications
+		 * in the transaction.
+		 */
+		if (iuid != uid) {
+			if (XFS_IS_QUOTA_RUNNING(mp) && XFS_IS_UQUOTA_ON(mp)) {
+				ASSERT(mask & ATTR_UID);
+				ASSERT(udqp);
+				olddquot1 = xfs_qm_vop_chown(tp, ip,
+							&ip->i_udquot, udqp);
+			}
+			ip->i_d.di_uid = uid;
+			inode->i_uid = uid;
+		}
+		if (igid != gid) {
+			if (XFS_IS_QUOTA_RUNNING(mp) && XFS_IS_GQUOTA_ON(mp)) {
+				ASSERT(!XFS_IS_PQUOTA_ON(mp));
+				ASSERT(mask & ATTR_GID);
+				ASSERT(gdqp);
+				olddquot2 = xfs_qm_vop_chown(tp, ip,
+							&ip->i_gdquot, gdqp);
+			}
+			ip->i_d.di_gid = gid;
+			inode->i_gid = gid;
+		}
+	}
+
+	/*
+	 * Change file access modes.
+	 */
+	if (mask & ATTR_MODE) {
+		umode_t mode = iattr->ia_mode;
+
+		if (!in_group_p(inode->i_gid) && !capable(CAP_FSETID))
+			mode &= ~S_ISGID;
+
+		ip->i_d.di_mode &= S_IFMT;
+		ip->i_d.di_mode |= mode & ~S_IFMT;
+
+		inode->i_mode &= S_IFMT;
+		inode->i_mode |= mode & ~S_IFMT;
+	}
+
+	/*
+	 * Change file access or modified times.
+	 */
+	if (mask & ATTR_ATIME) {
+		inode->i_atime = iattr->ia_atime;
+		ip->i_d.di_atime.t_sec = iattr->ia_atime.tv_sec;
+		ip->i_d.di_atime.t_nsec = iattr->ia_atime.tv_nsec;
+		ip->i_update_core = 1;
+	}
+	if (mask & ATTR_CTIME) {
+		inode->i_ctime = iattr->ia_ctime;
+		ip->i_d.di_ctime.t_sec = iattr->ia_ctime.tv_sec;
+		ip->i_d.di_ctime.t_nsec = iattr->ia_ctime.tv_nsec;
+		ip->i_update_core = 1;
+	}
+	if (mask & ATTR_MTIME) {
+		inode->i_mtime = iattr->ia_mtime;
+		ip->i_d.di_mtime.t_sec = iattr->ia_mtime.tv_sec;
+		ip->i_d.di_mtime.t_nsec = iattr->ia_mtime.tv_nsec;
+		ip->i_update_core = 1;
+	}
+
+	/*
+	 * And finally, log the inode core if any attribute in it
+	 * has been changed.
+	 */
+	if (mask & (ATTR_UID|ATTR_GID|ATTR_MODE|
+		    ATTR_ATIME|ATTR_CTIME|ATTR_MTIME))
+		xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
+
+	XFS_STATS_INC(xs_ig_attrchg);
+
+	/*
+	 * If this is a synchronous mount, make sure that the
+	 * transaction goes to disk before returning to the user.
+	 * This is slightly sub-optimal in that truncates require
+	 * two sync transactions instead of one for wsync filesystems.
+	 * One for the truncate and one for the timestamps since we
+	 * don't want to change the timestamps unless we're sure the
+	 * truncate worked.  Truncates are less than 1% of the laddis
+	 * mix so this probably isn't worth the trouble to optimize.
+	 */
+	code = 0;
+	if (mp->m_flags & XFS_MOUNT_WSYNC)
+		xfs_trans_set_sync(tp);
+
+	code = xfs_trans_commit(tp, commit_flags);
+
+	xfs_iunlock(ip, lock_flags);
+
+	/*
+	 * Release any dquot(s) the inode had kept before chown.
+	 */
+	xfs_qm_dqrele(olddquot1);
+	xfs_qm_dqrele(olddquot2);
+	xfs_qm_dqrele(udqp);
+	xfs_qm_dqrele(gdqp);
+
+	if (code)
+		return code;
+
+	/*
+	 * XXX(hch): Updating the ACL entries is not atomic vs the i_mode
+	 * 	     update.  We could avoid this with linked transactions
+	 * 	     and passing down the transaction pointer all the way
+	 *	     to attr_set.  No previous user of the generic
+	 * 	     Posix ACL code seems to care about this issue either.
+	 */
+	if ((mask & ATTR_MODE) && !(flags & XFS_ATTR_NOACL)) {
+		code = -xfs_acl_chmod(inode);
+		if (code)
+			return XFS_ERROR(code);
+	}
+
+	return 0;
+
+ abort_return:
+	commit_flags |= XFS_TRANS_ABORT;
+ error_return:
+	xfs_qm_dqrele(udqp);
+	xfs_qm_dqrele(gdqp);
+	if (tp) {
+		xfs_trans_cancel(tp, commit_flags);
+	}
+	if (lock_flags != 0) {
+		xfs_iunlock(ip, lock_flags);
+	}
+	return code;
+}
+
+/*
+ * The maximum pathlen is 1024 bytes. Since the minimum file system
+ * blocksize is 512 bytes, we can get a max of 2 extents back from
+ * bmapi.
+ */
+#define SYMLINK_MAPS 2
+
+STATIC int
+xfs_readlink_bmap(
+	xfs_inode_t	*ip,
+	char		*link)
+{
+	xfs_mount_t	*mp = ip->i_mount;
+	int		pathlen = ip->i_d.di_size;
+	int             nmaps = SYMLINK_MAPS;
+	xfs_bmbt_irec_t mval[SYMLINK_MAPS];
+	xfs_daddr_t	d;
+	int		byte_cnt;
+	int		n;
+	xfs_buf_t	*bp;
+	int		error = 0;
+
+	error = xfs_bmapi(NULL, ip, 0, XFS_B_TO_FSB(mp, pathlen), 0, NULL, 0,
+			mval, &nmaps, NULL);
+	if (error)
+		goto out;
+
+	for (n = 0; n < nmaps; n++) {
+		d = XFS_FSB_TO_DADDR(mp, mval[n].br_startblock);
+		byte_cnt = XFS_FSB_TO_B(mp, mval[n].br_blockcount);
+
+		bp = xfs_buf_read(mp->m_ddev_targp, d, BTOBB(byte_cnt),
+				  XBF_LOCK | XBF_MAPPED | XBF_DONT_BLOCK);
+		error = XFS_BUF_GETERROR(bp);
+		if (error) {
+			xfs_ioerror_alert("xfs_readlink",
+				  ip->i_mount, bp, XFS_BUF_ADDR(bp));
+			xfs_buf_relse(bp);
+			goto out;
+		}
+		if (pathlen < byte_cnt)
+			byte_cnt = pathlen;
+		pathlen -= byte_cnt;
+
+		memcpy(link, XFS_BUF_PTR(bp), byte_cnt);
+		xfs_buf_relse(bp);
+	}
+
+	link[ip->i_d.di_size] = '\0';
+	error = 0;
+
+ out:
+	return error;
+}
+
+int
+xfs_readlink(
+	xfs_inode_t     *ip,
+	char		*link)
+{
+	xfs_mount_t	*mp = ip->i_mount;
+	xfs_fsize_t	pathlen;
+	int		error = 0;
+
+	trace_xfs_readlink(ip);
+
+	if (XFS_FORCED_SHUTDOWN(mp))
+		return XFS_ERROR(EIO);
+
+	xfs_ilock(ip, XFS_ILOCK_SHARED);
+
+	pathlen = ip->i_d.di_size;
+	if (!pathlen)
+		goto out;
+
+	if (pathlen < 0 || pathlen > MAXPATHLEN) {
+		xfs_alert(mp, "%s: inode (%llu) bad symlink length (%lld)",
+			 __func__, (unsigned long long) ip->i_ino,
+			 (long long) pathlen);
+		ASSERT(0);
+		error = XFS_ERROR(EFSCORRUPTED);
+		goto out;
+	}
+
+
+	if (ip->i_df.if_flags & XFS_IFINLINE) {
+		memcpy(link, ip->i_df.if_u1.if_data, pathlen);
+		link[pathlen] = '\0';
+	} else {
+		error = xfs_readlink_bmap(ip, link);
+	}
+
+ out:
+	xfs_iunlock(ip, XFS_ILOCK_SHARED);
+	return error;
+}
+
+/*
+ * Flags for xfs_free_eofblocks
+ */
+#define XFS_FREE_EOF_TRYLOCK	(1<<0)
+
+/*
+ * This is called by xfs_inactive to free any blocks beyond eof
+ * when the link count isn't zero and by xfs_dm_punch_hole() when
+ * punching a hole to EOF.
+ */
+STATIC int
+xfs_free_eofblocks(
+	xfs_mount_t	*mp,
+	xfs_inode_t	*ip,
+	int		flags)
+{
+	xfs_trans_t	*tp;
+	int		error;
+	xfs_fileoff_t	end_fsb;
+	xfs_fileoff_t	last_fsb;
+	xfs_filblks_t	map_len;
+	int		nimaps;
+	xfs_bmbt_irec_t	imap;
+
+	/*
+	 * Figure out if there are any blocks beyond the end
+	 * of the file.  If not, then there is nothing to do.
+	 */
+	end_fsb = XFS_B_TO_FSB(mp, ((xfs_ufsize_t)ip->i_size));
+	last_fsb = XFS_B_TO_FSB(mp, (xfs_ufsize_t)XFS_MAXIOFFSET(mp));
+	if (last_fsb <= end_fsb)
+		return 0;
+	map_len = last_fsb - end_fsb;
+
+	nimaps = 1;
+	xfs_ilock(ip, XFS_ILOCK_SHARED);
+	error = xfs_bmapi(NULL, ip, end_fsb, map_len, 0,
+			  NULL, 0, &imap, &nimaps, NULL);
+	xfs_iunlock(ip, XFS_ILOCK_SHARED);
+
+	if (!error && (nimaps != 0) &&
+	    (imap.br_startblock != HOLESTARTBLOCK ||
+	     ip->i_delayed_blks)) {
+		/*
+		 * Attach the dquots to the inode up front.
+		 */
+		error = xfs_qm_dqattach(ip, 0);
+		if (error)
+			return error;
+
+		/*
+		 * There are blocks after the end of file.
+		 * Free them up now by truncating the file to
+		 * its current size.
+		 */
+		tp = xfs_trans_alloc(mp, XFS_TRANS_INACTIVE);
+
+		/*
+		 * Do the xfs_itruncate_start() call before
+		 * reserving any log space because
+		 * itruncate_start will call into the buffer
+		 * cache and we can't
+		 * do that within a transaction.
+		 */
+		if (flags & XFS_FREE_EOF_TRYLOCK) {
+			if (!xfs_ilock_nowait(ip, XFS_IOLOCK_EXCL)) {
+				xfs_trans_cancel(tp, 0);
+				return 0;
+			}
+		} else {
+			xfs_ilock(ip, XFS_IOLOCK_EXCL);
+		}
+		error = xfs_itruncate_start(ip, XFS_ITRUNC_DEFINITE,
+				    ip->i_size);
+		if (error) {
+			xfs_trans_cancel(tp, 0);
+			xfs_iunlock(ip, XFS_IOLOCK_EXCL);
+			return error;
+		}
+
+		error = xfs_trans_reserve(tp, 0,
+					  XFS_ITRUNCATE_LOG_RES(mp),
+					  0, XFS_TRANS_PERM_LOG_RES,
+					  XFS_ITRUNCATE_LOG_COUNT);
+		if (error) {
+			ASSERT(XFS_FORCED_SHUTDOWN(mp));
+			xfs_trans_cancel(tp, 0);
+			xfs_iunlock(ip, XFS_IOLOCK_EXCL);
+			return error;
+		}
+
+		xfs_ilock(ip, XFS_ILOCK_EXCL);
+		xfs_trans_ijoin(tp, ip);
+
+		error = xfs_itruncate_finish(&tp, ip,
+					     ip->i_size,
+					     XFS_DATA_FORK,
+					     0);
+		/*
+		 * If we get an error at this point we
+		 * simply don't bother truncating the file.
+		 */
+		if (error) {
+			xfs_trans_cancel(tp,
+					 (XFS_TRANS_RELEASE_LOG_RES |
+					  XFS_TRANS_ABORT));
+		} else {
+			error = xfs_trans_commit(tp,
+						XFS_TRANS_RELEASE_LOG_RES);
+		}
+		xfs_iunlock(ip, XFS_IOLOCK_EXCL|XFS_ILOCK_EXCL);
+	}
+	return error;
+}
+
+/*
+ * Free a symlink that has blocks associated with it.
+ */
+STATIC int
+xfs_inactive_symlink_rmt(
+	xfs_inode_t	*ip,
+	xfs_trans_t	**tpp)
+{
+	xfs_buf_t	*bp;
+	int		committed;
+	int		done;
+	int		error;
+	xfs_fsblock_t	first_block;
+	xfs_bmap_free_t	free_list;
+	int		i;
+	xfs_mount_t	*mp;
+	xfs_bmbt_irec_t	mval[SYMLINK_MAPS];
+	int		nmaps;
+	xfs_trans_t	*ntp;
+	int		size;
+	xfs_trans_t	*tp;
+
+	tp = *tpp;
+	mp = ip->i_mount;
+	ASSERT(ip->i_d.di_size > XFS_IFORK_DSIZE(ip));
+	/*
+	 * We're freeing a symlink that has some
+	 * blocks allocated to it.  Free the
+	 * blocks here.  We know that we've got
+	 * either 1 or 2 extents and that we can
+	 * free them all in one bunmapi call.
+	 */
+	ASSERT(ip->i_d.di_nextents > 0 && ip->i_d.di_nextents <= 2);
+	if ((error = xfs_trans_reserve(tp, 0, XFS_ITRUNCATE_LOG_RES(mp), 0,
+			XFS_TRANS_PERM_LOG_RES, XFS_ITRUNCATE_LOG_COUNT))) {
+		ASSERT(XFS_FORCED_SHUTDOWN(mp));
+		xfs_trans_cancel(tp, 0);
+		*tpp = NULL;
+		return error;
+	}
+	/*
+	 * Lock the inode, fix the size, and join it to the transaction.
+	 * Hold it so in the normal path, we still have it locked for
+	 * the second transaction.  In the error paths we need it
+	 * held so the cancel won't rele it, see below.
+	 */
+	xfs_ilock(ip, XFS_IOLOCK_EXCL | XFS_ILOCK_EXCL);
+	size = (int)ip->i_d.di_size;
+	ip->i_d.di_size = 0;
+	xfs_trans_ijoin(tp, ip);
+	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
+	/*
+	 * Find the block(s) so we can inval and unmap them.
+	 */
+	done = 0;
+	xfs_bmap_init(&free_list, &first_block);
+	nmaps = ARRAY_SIZE(mval);
+	if ((error = xfs_bmapi(tp, ip, 0, XFS_B_TO_FSB(mp, size),
+			XFS_BMAPI_METADATA, &first_block, 0, mval, &nmaps,
+			&free_list)))
+		goto error0;
+	/*
+	 * Invalidate the block(s).
+	 */
+	for (i = 0; i < nmaps; i++) {
+		bp = xfs_trans_get_buf(tp, mp->m_ddev_targp,
+			XFS_FSB_TO_DADDR(mp, mval[i].br_startblock),
+			XFS_FSB_TO_BB(mp, mval[i].br_blockcount), 0);
+		xfs_trans_binval(tp, bp);
+	}
+	/*
+	 * Unmap the dead block(s) to the free_list.
+	 */
+	if ((error = xfs_bunmapi(tp, ip, 0, size, XFS_BMAPI_METADATA, nmaps,
+			&first_block, &free_list, &done)))
+		goto error1;
+	ASSERT(done);
+	/*
+	 * Commit the first transaction.  This logs the EFI and the inode.
+	 */
+	if ((error = xfs_bmap_finish(&tp, &free_list, &committed)))
+		goto error1;
+	/*
+	 * The transaction must have been committed, since there were
+	 * actually extents freed by xfs_bunmapi.  See xfs_bmap_finish.
+	 * The new tp has the extent freeing and EFDs.
+	 */
+	ASSERT(committed);
+	/*
+	 * The first xact was committed, so add the inode to the new one.
+	 * Mark it dirty so it will be logged and moved forward in the log as
+	 * part of every commit.
+	 */
+	xfs_trans_ijoin(tp, ip);
+	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
+	/*
+	 * Get a new, empty transaction to return to our caller.
+	 */
+	ntp = xfs_trans_dup(tp);
+	/*
+	 * Commit the transaction containing extent freeing and EFDs.
+	 * If we get an error on the commit here or on the reserve below,
+	 * we need to unlock the inode since the new transaction doesn't
+	 * have the inode attached.
+	 */
+	error = xfs_trans_commit(tp, 0);
+	tp = ntp;
+	if (error) {
+		ASSERT(XFS_FORCED_SHUTDOWN(mp));
+		goto error0;
+	}
+	/*
+	 * transaction commit worked ok so we can drop the extra ticket
+	 * reference that we gained in xfs_trans_dup()
+	 */
+	xfs_log_ticket_put(tp->t_ticket);
+
+	/*
+	 * Remove the memory for extent descriptions (just bookkeeping).
+	 */
+	if (ip->i_df.if_bytes)
+		xfs_idata_realloc(ip, -ip->i_df.if_bytes, XFS_DATA_FORK);
+	ASSERT(ip->i_df.if_bytes == 0);
+	/*
+	 * Put an itruncate log reservation in the new transaction
+	 * for our caller.
+	 */
+	if ((error = xfs_trans_reserve(tp, 0, XFS_ITRUNCATE_LOG_RES(mp), 0,
+			XFS_TRANS_PERM_LOG_RES, XFS_ITRUNCATE_LOG_COUNT))) {
+		ASSERT(XFS_FORCED_SHUTDOWN(mp));
+		goto error0;
+	}
+	/*
+	 * Return with the inode locked but not joined to the transaction.
+	 */
+	*tpp = tp;
+	return 0;
+
+ error1:
+	xfs_bmap_cancel(&free_list);
+ error0:
+	/*
+	 * Have to come here with the inode locked and either
+	 * (held and in the transaction) or (not in the transaction).
+	 * If the inode isn't held then cancel would iput it, but
+	 * that's wrong since this is inactive and the vnode ref
+	 * count is 0 already.
+	 * Cancel won't do anything to the inode if held, but it still
+	 * needs to be locked until the cancel is done, if it was
+	 * joined to the transaction.
+	 */
+	xfs_trans_cancel(tp, XFS_TRANS_RELEASE_LOG_RES | XFS_TRANS_ABORT);
+	xfs_iunlock(ip, XFS_IOLOCK_EXCL | XFS_ILOCK_EXCL);
+	*tpp = NULL;
+	return error;
+
+}
+
+STATIC int
+xfs_inactive_symlink_local(
+	xfs_inode_t	*ip,
+	xfs_trans_t	**tpp)
+{
+	int		error;
+
+	ASSERT(ip->i_d.di_size <= XFS_IFORK_DSIZE(ip));
+	/*
+	 * We're freeing a symlink which fit into
+	 * the inode.  Just free the memory used
+	 * to hold the old symlink.
+	 */
+	error = xfs_trans_reserve(*tpp, 0,
+				  XFS_ITRUNCATE_LOG_RES(ip->i_mount),
+				  0, XFS_TRANS_PERM_LOG_RES,
+				  XFS_ITRUNCATE_LOG_COUNT);
+
+	if (error) {
+		xfs_trans_cancel(*tpp, 0);
+		*tpp = NULL;
+		return error;
+	}
+	xfs_ilock(ip, XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL);
+
+	/*
+	 * Zero length symlinks _can_ exist.
+	 */
+	if (ip->i_df.if_bytes > 0) {
+		xfs_idata_realloc(ip,
+				  -(ip->i_df.if_bytes),
+				  XFS_DATA_FORK);
+		ASSERT(ip->i_df.if_bytes == 0);
+	}
+	return 0;
+}
+
+STATIC int
+xfs_inactive_attrs(
+	xfs_inode_t	*ip,
+	xfs_trans_t	**tpp)
+{
+	xfs_trans_t	*tp;
+	int		error;
+	xfs_mount_t	*mp;
+
+	ASSERT(xfs_isilocked(ip, XFS_IOLOCK_EXCL));
+	tp = *tpp;
+	mp = ip->i_mount;
+	ASSERT(ip->i_d.di_forkoff != 0);
+	error = xfs_trans_commit(tp, XFS_TRANS_RELEASE_LOG_RES);
+	xfs_iunlock(ip, XFS_ILOCK_EXCL);
+	if (error)
+		goto error_unlock;
+
+	error = xfs_attr_inactive(ip);
+	if (error)
+		goto error_unlock;
+
+	tp = xfs_trans_alloc(mp, XFS_TRANS_INACTIVE);
+	error = xfs_trans_reserve(tp, 0,
+				  XFS_IFREE_LOG_RES(mp),
+				  0, XFS_TRANS_PERM_LOG_RES,
+				  XFS_INACTIVE_LOG_COUNT);
+	if (error)
+		goto error_cancel;
+
+	xfs_ilock(ip, XFS_ILOCK_EXCL);
+	xfs_trans_ijoin(tp, ip);
+	xfs_idestroy_fork(ip, XFS_ATTR_FORK);
+
+	ASSERT(ip->i_d.di_anextents == 0);
+
+	*tpp = tp;
+	return 0;
+
+error_cancel:
+	ASSERT(XFS_FORCED_SHUTDOWN(mp));
+	xfs_trans_cancel(tp, 0);
+error_unlock:
+	*tpp = NULL;
+	xfs_iunlock(ip, XFS_IOLOCK_EXCL);
+	return error;
+}
+
+int
+xfs_release(
+	xfs_inode_t	*ip)
+{
+	xfs_mount_t	*mp = ip->i_mount;
+	int		error;
+
+	if (!S_ISREG(ip->i_d.di_mode) || (ip->i_d.di_mode == 0))
+		return 0;
+
+	/* If this is a read-only mount, don't do this (would generate I/O) */
+	if (mp->m_flags & XFS_MOUNT_RDONLY)
+		return 0;
+
+	if (!XFS_FORCED_SHUTDOWN(mp)) {
+		int truncated;
+
+		/*
+		 * If we are using filestreams, and we have an unlinked
+		 * file that we are processing the last close on, then nothing
+		 * will be able to reopen and write to this file. Purge this
+		 * inode from the filestreams cache so that it doesn't delay
+		 * teardown of the inode.
+		 */
+		if ((ip->i_d.di_nlink == 0) && xfs_inode_is_filestream(ip))
+			xfs_filestream_deassociate(ip);
+
+		/*
+		 * If we previously truncated this file and removed old data
+		 * in the process, we want to initiate "early" writeout on
+		 * the last close.  This is an attempt to combat the notorious
+		 * NULL files problem which is particularly noticeable from a
+		 * truncate down, buffered (re-)write (delalloc), followed by
+		 * a crash.  What we are effectively doing here is
+		 * significantly reducing the time window where we'd otherwise
+		 * be exposed to that problem.
+		 */
+		truncated = xfs_iflags_test_and_clear(ip, XFS_ITRUNCATED);
+		if (truncated) {
+			xfs_iflags_clear(ip, XFS_IDIRTY_RELEASE);
+			if (VN_DIRTY(VFS_I(ip)) && ip->i_delayed_blks > 0)
+				xfs_flush_pages(ip, 0, -1, XBF_ASYNC, FI_NONE);
+		}
+	}
+
+	if (ip->i_d.di_nlink == 0)
+		return 0;
+
+	if ((((ip->i_d.di_mode & S_IFMT) == S_IFREG) &&
+	     ((ip->i_size > 0) || (VN_CACHED(VFS_I(ip)) > 0 ||
+	       ip->i_delayed_blks > 0)) &&
+	     (ip->i_df.if_flags & XFS_IFEXTENTS))  &&
+	    (!(ip->i_d.di_flags & (XFS_DIFLAG_PREALLOC | XFS_DIFLAG_APPEND)))) {
+
+		/*
+		 * If we can't get the iolock just skip truncating the blocks
+		 * past EOF because we could deadlock with the mmap_sem
+		 * otherwise.  We'll get another chance to drop them once the
+		 * last reference to the inode is dropped, so we'll never leak
+		 * blocks permanently.
+		 *
+		 * Further, check if the inode is being opened, written and
+		 * closed frequently and we have delayed allocation blocks
+		 * outstanding (e.g. streaming writes from the NFS server),
+		 * truncating the blocks past EOF will cause fragmentation to
+		 * occur.
+		 *
+		 * In this case don't do the truncation, either, but we have to
+		 * be careful how we detect this case. Blocks beyond EOF show
+		 * up as i_delayed_blks even when the inode is clean, so we
+		 * need to truncate them away first before checking for a dirty
+		 * release. Hence on the first dirty close we will still remove
+		 * the speculative allocation, but after that we will leave it
+		 * in place.
+		 */
+		if (xfs_iflags_test(ip, XFS_IDIRTY_RELEASE))
+			return 0;
+
+		error = xfs_free_eofblocks(mp, ip,
+					   XFS_FREE_EOF_TRYLOCK);
+		if (error)
+			return error;
+
+		/* delalloc blocks after truncation means it really is dirty */
+		if (ip->i_delayed_blks)
+			xfs_iflags_set(ip, XFS_IDIRTY_RELEASE);
+	}
+	return 0;
+}
+
+/*
+ * xfs_inactive
+ *
+ * This is called when the vnode reference count for the vnode
+ * goes to zero.  If the file has been unlinked, then it must
+ * now be truncated.  Also, we clear all of the read-ahead state
+ * kept for the inode here since the file is now closed.
+ */
+int
+xfs_inactive(
+	xfs_inode_t	*ip)
+{
+	xfs_bmap_free_t	free_list;
+	xfs_fsblock_t	first_block;
+	int		committed;
+	xfs_trans_t	*tp;
+	xfs_mount_t	*mp;
+	int		error;
+	int		truncate;
+
+	/*
+	 * If the inode is already free, then there can be nothing
+	 * to clean up here.
+	 */
+	if (ip->i_d.di_mode == 0 || is_bad_inode(VFS_I(ip))) {
+		ASSERT(ip->i_df.if_real_bytes == 0);
+		ASSERT(ip->i_df.if_broot_bytes == 0);
+		return VN_INACTIVE_CACHE;
+	}
+
+	/*
+	 * Only do a truncate if it's a regular file with
+	 * some actual space in it.  It's OK to look at the
+	 * inode's fields without the lock because we're the
+	 * only one with a reference to the inode.
+	 */
+	truncate = ((ip->i_d.di_nlink == 0) &&
+	    ((ip->i_d.di_size != 0) || (ip->i_size != 0) ||
+	     (ip->i_d.di_nextents > 0) || (ip->i_delayed_blks > 0)) &&
+	    ((ip->i_d.di_mode & S_IFMT) == S_IFREG));
+
+	mp = ip->i_mount;
+
+	error = 0;
+
+	/* If this is a read-only mount, don't do this (would generate I/O) */
+	if (mp->m_flags & XFS_MOUNT_RDONLY)
+		goto out;
+
+	if (ip->i_d.di_nlink != 0) {
+		if ((((ip->i_d.di_mode & S_IFMT) == S_IFREG) &&
+                     ((ip->i_size > 0) || (VN_CACHED(VFS_I(ip)) > 0 ||
+                       ip->i_delayed_blks > 0)) &&
+		      (ip->i_df.if_flags & XFS_IFEXTENTS) &&
+		     (!(ip->i_d.di_flags &
+				(XFS_DIFLAG_PREALLOC | XFS_DIFLAG_APPEND)) ||
+		      (ip->i_delayed_blks != 0)))) {
+			error = xfs_free_eofblocks(mp, ip, 0);
+			if (error)
+				return VN_INACTIVE_CACHE;
+		}
+		goto out;
+	}
+
+	ASSERT(ip->i_d.di_nlink == 0);
+
+	error = xfs_qm_dqattach(ip, 0);
+	if (error)
+		return VN_INACTIVE_CACHE;
+
+	tp = xfs_trans_alloc(mp, XFS_TRANS_INACTIVE);
+	if (truncate) {
+		/*
+		 * Do the xfs_itruncate_start() call before
+		 * reserving any log space because itruncate_start
+		 * will call into the buffer cache and we can't
+		 * do that within a transaction.
+		 */
+		xfs_ilock(ip, XFS_IOLOCK_EXCL);
+
+		error = xfs_itruncate_start(ip, XFS_ITRUNC_DEFINITE, 0);
+		if (error) {
+			xfs_trans_cancel(tp, 0);
+			xfs_iunlock(ip, XFS_IOLOCK_EXCL);
+			return VN_INACTIVE_CACHE;
+		}
+
+		error = xfs_trans_reserve(tp, 0,
+					  XFS_ITRUNCATE_LOG_RES(mp),
+					  0, XFS_TRANS_PERM_LOG_RES,
+					  XFS_ITRUNCATE_LOG_COUNT);
+		if (error) {
+			/* Don't call itruncate_cleanup */
+			ASSERT(XFS_FORCED_SHUTDOWN(mp));
+			xfs_trans_cancel(tp, 0);
+			xfs_iunlock(ip, XFS_IOLOCK_EXCL);
+			return VN_INACTIVE_CACHE;
+		}
+
+		xfs_ilock(ip, XFS_ILOCK_EXCL);
+		xfs_trans_ijoin(tp, ip);
+
+		/*
+		 * normally, we have to run xfs_itruncate_finish sync.
+		 * But if filesystem is wsync and we're in the inactive
+		 * path, then we know that nlink == 0, and that the
+		 * xaction that made nlink == 0 is permanently committed
+		 * since xfs_remove runs as a synchronous transaction.
+		 */
+		error = xfs_itruncate_finish(&tp, ip, 0, XFS_DATA_FORK,
+				(!(mp->m_flags & XFS_MOUNT_WSYNC) ? 1 : 0));
+
+		if (error) {
+			xfs_trans_cancel(tp,
+				XFS_TRANS_RELEASE_LOG_RES | XFS_TRANS_ABORT);
+			xfs_iunlock(ip, XFS_IOLOCK_EXCL | XFS_ILOCK_EXCL);
+			return VN_INACTIVE_CACHE;
+		}
+	} else if ((ip->i_d.di_mode & S_IFMT) == S_IFLNK) {
+
+		/*
+		 * If we get an error while cleaning up a
+		 * symlink we bail out.
+		 */
+		error = (ip->i_d.di_size > XFS_IFORK_DSIZE(ip)) ?
+			xfs_inactive_symlink_rmt(ip, &tp) :
+			xfs_inactive_symlink_local(ip, &tp);
+
+		if (error) {
+			ASSERT(tp == NULL);
+			return VN_INACTIVE_CACHE;
+		}
+
+		xfs_trans_ijoin(tp, ip);
+	} else {
+		error = xfs_trans_reserve(tp, 0,
+					  XFS_IFREE_LOG_RES(mp),
+					  0, XFS_TRANS_PERM_LOG_RES,
+					  XFS_INACTIVE_LOG_COUNT);
+		if (error) {
+			ASSERT(XFS_FORCED_SHUTDOWN(mp));
+			xfs_trans_cancel(tp, 0);
+			return VN_INACTIVE_CACHE;
+		}
+
+		xfs_ilock(ip, XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL);
+		xfs_trans_ijoin(tp, ip);
+	}
+
+	/*
+	 * If there are attributes associated with the file
+	 * then blow them away now.  The code calls a routine
+	 * that recursively deconstructs the attribute fork.
+	 * We need to just commit the current transaction
+	 * because we can't use it for xfs_attr_inactive().
+	 */
+	if (ip->i_d.di_anextents > 0) {
+		error = xfs_inactive_attrs(ip, &tp);
+		/*
+		 * If we got an error, the transaction is already
+		 * cancelled, and the inode is unlocked. Just get out.
+		 */
+		 if (error)
+			 return VN_INACTIVE_CACHE;
+	} else if (ip->i_afp) {
+		xfs_idestroy_fork(ip, XFS_ATTR_FORK);
+	}
+
+	/*
+	 * Free the inode.
+	 */
+	xfs_bmap_init(&free_list, &first_block);
+	error = xfs_ifree(tp, ip, &free_list);
+	if (error) {
+		/*
+		 * If we fail to free the inode, shut down.  The cancel
+		 * might do that, we need to make sure.  Otherwise the
+		 * inode might be lost for a long time or forever.
+		 */
+		if (!XFS_FORCED_SHUTDOWN(mp)) {
+			xfs_notice(mp, "%s: xfs_ifree returned error %d",
+				__func__, error);
+			xfs_force_shutdown(mp, SHUTDOWN_META_IO_ERROR);
+		}
+		xfs_trans_cancel(tp, XFS_TRANS_RELEASE_LOG_RES|XFS_TRANS_ABORT);
+	} else {
+		/*
+		 * Credit the quota account(s). The inode is gone.
+		 */
+		xfs_trans_mod_dquot_byino(tp, ip, XFS_TRANS_DQ_ICOUNT, -1);
+
+		/*
+		 * Just ignore errors at this point.  There is nothing we can
+		 * do except to try to keep going. Make sure it's not a silent
+		 * error.
+		 */
+		error = xfs_bmap_finish(&tp,  &free_list, &committed);
+		if (error)
+			xfs_notice(mp, "%s: xfs_bmap_finish returned error %d",
+				__func__, error);
+		error = xfs_trans_commit(tp, XFS_TRANS_RELEASE_LOG_RES);
+		if (error)
+			xfs_notice(mp, "%s: xfs_trans_commit returned error %d",
+				__func__, error);
+	}
+
+	/*
+	 * Release the dquots held by inode, if any.
+	 */
+	xfs_qm_dqdetach(ip);
+	xfs_iunlock(ip, XFS_IOLOCK_EXCL | XFS_ILOCK_EXCL);
+
+ out:
+	return VN_INACTIVE_CACHE;
+}
+
+/*
+ * Lookups up an inode from "name". If ci_name is not NULL, then a CI match
+ * is allowed, otherwise it has to be an exact match. If a CI match is found,
+ * ci_name->name will point to a the actual name (caller must free) or
+ * will be set to NULL if an exact match is found.
+ */
+int
+xfs_lookup(
+	xfs_inode_t		*dp,
+	struct xfs_name		*name,
+	xfs_inode_t		**ipp,
+	struct xfs_name		*ci_name)
+{
+	xfs_ino_t		inum;
+	int			error;
+	uint			lock_mode;
+
+	trace_xfs_lookup(dp, name);
+
+	if (XFS_FORCED_SHUTDOWN(dp->i_mount))
+		return XFS_ERROR(EIO);
+
+	lock_mode = xfs_ilock_map_shared(dp);
+	error = xfs_dir_lookup(NULL, dp, name, &inum, ci_name);
+	xfs_iunlock_map_shared(dp, lock_mode);
+
+	if (error)
+		goto out;
+
+	error = xfs_iget(dp->i_mount, NULL, inum, 0, 0, ipp);
+	if (error)
+		goto out_free_name;
+
+	return 0;
+
+out_free_name:
+	if (ci_name)
+		kmem_free(ci_name->name);
+out:
+	*ipp = NULL;
+	return error;
+}
+
+int
+xfs_create(
+	xfs_inode_t		*dp,
+	struct xfs_name		*name,
+	mode_t			mode,
+	xfs_dev_t		rdev,
+	xfs_inode_t		**ipp)
+{
+	int			is_dir = S_ISDIR(mode);
+	struct xfs_mount	*mp = dp->i_mount;
+	struct xfs_inode	*ip = NULL;
+	struct xfs_trans	*tp = NULL;
+	int			error;
+	xfs_bmap_free_t		free_list;
+	xfs_fsblock_t		first_block;
+	boolean_t		unlock_dp_on_error = B_FALSE;
+	uint			cancel_flags;
+	int			committed;
+	prid_t			prid;
+	struct xfs_dquot	*udqp = NULL;
+	struct xfs_dquot	*gdqp = NULL;
+	uint			resblks;
+	uint			log_res;
+	uint			log_count;
+
+	trace_xfs_create(dp, name);
+
+	if (XFS_FORCED_SHUTDOWN(mp))
+		return XFS_ERROR(EIO);
+
+	if (dp->i_d.di_flags & XFS_DIFLAG_PROJINHERIT)
+		prid = xfs_get_projid(dp);
+	else
+		prid = XFS_PROJID_DEFAULT;
+
+	/*
+	 * Make sure that we have allocated dquot(s) on disk.
+	 */
+	error = xfs_qm_vop_dqalloc(dp, current_fsuid(), current_fsgid(), prid,
+			XFS_QMOPT_QUOTALL | XFS_QMOPT_INHERIT, &udqp, &gdqp);
+	if (error)
+		return error;
+
+	if (is_dir) {
+		rdev = 0;
+		resblks = XFS_MKDIR_SPACE_RES(mp, name->len);
+		log_res = XFS_MKDIR_LOG_RES(mp);
+		log_count = XFS_MKDIR_LOG_COUNT;
+		tp = xfs_trans_alloc(mp, XFS_TRANS_MKDIR);
+	} else {
+		resblks = XFS_CREATE_SPACE_RES(mp, name->len);
+		log_res = XFS_CREATE_LOG_RES(mp);
+		log_count = XFS_CREATE_LOG_COUNT;
+		tp = xfs_trans_alloc(mp, XFS_TRANS_CREATE);
+	}
+
+	cancel_flags = XFS_TRANS_RELEASE_LOG_RES;
+
+	/*
+	 * Initially assume that the file does not exist and
+	 * reserve the resources for that case.  If that is not
+	 * the case we'll drop the one we have and get a more
+	 * appropriate transaction later.
+	 */
+	error = xfs_trans_reserve(tp, resblks, log_res, 0,
+			XFS_TRANS_PERM_LOG_RES, log_count);
+	if (error == ENOSPC) {
+		/* flush outstanding delalloc blocks and retry */
+		xfs_flush_inodes(dp);
+		error = xfs_trans_reserve(tp, resblks, log_res, 0,
+				XFS_TRANS_PERM_LOG_RES, log_count);
+	}
+	if (error == ENOSPC) {
+		/* No space at all so try a "no-allocation" reservation */
+		resblks = 0;
+		error = xfs_trans_reserve(tp, 0, log_res, 0,
+				XFS_TRANS_PERM_LOG_RES, log_count);
+	}
+	if (error) {
+		cancel_flags = 0;
+		goto out_trans_cancel;
+	}
+
+	xfs_ilock(dp, XFS_ILOCK_EXCL | XFS_ILOCK_PARENT);
+	unlock_dp_on_error = B_TRUE;
+
+	/*
+	 * Check for directory link count overflow.
+	 */
+	if (is_dir && dp->i_d.di_nlink >= XFS_MAXLINK) {
+		error = XFS_ERROR(EMLINK);
+		goto out_trans_cancel;
+	}
+
+	xfs_bmap_init(&free_list, &first_block);
+
+	/*
+	 * Reserve disk quota and the inode.
+	 */
+	error = xfs_trans_reserve_quota(tp, mp, udqp, gdqp, resblks, 1, 0);
+	if (error)
+		goto out_trans_cancel;
+
+	error = xfs_dir_canenter(tp, dp, name, resblks);
+	if (error)
+		goto out_trans_cancel;
+
+	/*
+	 * A newly created regular or special file just has one directory
+	 * entry pointing to them, but a directory also the "." entry
+	 * pointing to itself.
+	 */
+	error = xfs_dir_ialloc(&tp, dp, mode, is_dir ? 2 : 1, rdev,
+			       prid, resblks > 0, &ip, &committed);
+	if (error) {
+		if (error == ENOSPC)
+			goto out_trans_cancel;
+		goto out_trans_abort;
+	}
+
+	/*
+	 * Now we join the directory inode to the transaction.  We do not do it
+	 * earlier because xfs_dir_ialloc might commit the previous transaction
+	 * (and release all the locks).  An error from here on will result in
+	 * the transaction cancel unlocking dp so don't do it explicitly in the
+	 * error path.
+	 */
+	xfs_trans_ijoin_ref(tp, dp, XFS_ILOCK_EXCL);
+	unlock_dp_on_error = B_FALSE;
+
+	error = xfs_dir_createname(tp, dp, name, ip->i_ino,
+					&first_block, &free_list, resblks ?
+					resblks - XFS_IALLOC_SPACE_RES(mp) : 0);
+	if (error) {
+		ASSERT(error != ENOSPC);
+		goto out_trans_abort;
+	}
+	xfs_trans_ichgtime(tp, dp, XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG);
+	xfs_trans_log_inode(tp, dp, XFS_ILOG_CORE);
+
+	if (is_dir) {
+		error = xfs_dir_init(tp, ip, dp);
+		if (error)
+			goto out_bmap_cancel;
+
+		error = xfs_bumplink(tp, dp);
+		if (error)
+			goto out_bmap_cancel;
+	}
+
+	/*
+	 * If this is a synchronous mount, make sure that the
+	 * create transaction goes to disk before returning to
+	 * the user.
+	 */
+	if (mp->m_flags & (XFS_MOUNT_WSYNC|XFS_MOUNT_DIRSYNC))
+		xfs_trans_set_sync(tp);
+
+	/*
+	 * Attach the dquot(s) to the inodes and modify them incore.
+	 * These ids of the inode couldn't have changed since the new
+	 * inode has been locked ever since it was created.
+	 */
+	xfs_qm_vop_create_dqattach(tp, ip, udqp, gdqp);
+
+	error = xfs_bmap_finish(&tp, &free_list, &committed);
+	if (error)
+		goto out_bmap_cancel;
+
+	error = xfs_trans_commit(tp, XFS_TRANS_RELEASE_LOG_RES);
+	if (error)
+		goto out_release_inode;
+
+	xfs_qm_dqrele(udqp);
+	xfs_qm_dqrele(gdqp);
+
+	*ipp = ip;
+	return 0;
+
+ out_bmap_cancel:
+	xfs_bmap_cancel(&free_list);
+ out_trans_abort:
+	cancel_flags |= XFS_TRANS_ABORT;
+ out_trans_cancel:
+	xfs_trans_cancel(tp, cancel_flags);
+ out_release_inode:
+	/*
+	 * Wait until after the current transaction is aborted to
+	 * release the inode.  This prevents recursive transactions
+	 * and deadlocks from xfs_inactive.
+	 */
+	if (ip)
+		IRELE(ip);
+
+	xfs_qm_dqrele(udqp);
+	xfs_qm_dqrele(gdqp);
+
+	if (unlock_dp_on_error)
+		xfs_iunlock(dp, XFS_ILOCK_EXCL);
+	return error;
+}
+
+#ifdef DEBUG
+int xfs_locked_n;
+int xfs_small_retries;
+int xfs_middle_retries;
+int xfs_lots_retries;
+int xfs_lock_delays;
+#endif
+
+/*
+ * Bump the subclass so xfs_lock_inodes() acquires each lock with
+ * a different value
+ */
+static inline int
+xfs_lock_inumorder(int lock_mode, int subclass)
+{
+	if (lock_mode & (XFS_IOLOCK_SHARED|XFS_IOLOCK_EXCL))
+		lock_mode |= (subclass + XFS_LOCK_INUMORDER) << XFS_IOLOCK_SHIFT;
+	if (lock_mode & (XFS_ILOCK_SHARED|XFS_ILOCK_EXCL))
+		lock_mode |= (subclass + XFS_LOCK_INUMORDER) << XFS_ILOCK_SHIFT;
+
+	return lock_mode;
+}
+
+/*
+ * The following routine will lock n inodes in exclusive mode.
+ * We assume the caller calls us with the inodes in i_ino order.
+ *
+ * We need to detect deadlock where an inode that we lock
+ * is in the AIL and we start waiting for another inode that is locked
+ * by a thread in a long running transaction (such as truncate). This can
+ * result in deadlock since the long running trans might need to wait
+ * for the inode we just locked in order to push the tail and free space
+ * in the log.
+ */
+void
+xfs_lock_inodes(
+	xfs_inode_t	**ips,
+	int		inodes,
+	uint		lock_mode)
+{
+	int		attempts = 0, i, j, try_lock;
+	xfs_log_item_t	*lp;
+
+	ASSERT(ips && (inodes >= 2)); /* we need at least two */
+
+	try_lock = 0;
+	i = 0;
+
+again:
+	for (; i < inodes; i++) {
+		ASSERT(ips[i]);
+
+		if (i && (ips[i] == ips[i-1]))	/* Already locked */
+			continue;
+
+		/*
+		 * If try_lock is not set yet, make sure all locked inodes
+		 * are not in the AIL.
+		 * If any are, set try_lock to be used later.
+		 */
+
+		if (!try_lock) {
+			for (j = (i - 1); j >= 0 && !try_lock; j--) {
+				lp = (xfs_log_item_t *)ips[j]->i_itemp;
+				if (lp && (lp->li_flags & XFS_LI_IN_AIL)) {
+					try_lock++;
+				}
+			}
+		}
+
+		/*
+		 * If any of the previous locks we have locked is in the AIL,
+		 * we must TRY to get the second and subsequent locks. If
+		 * we can't get any, we must release all we have
+		 * and try again.
+		 */
+
+		if (try_lock) {
+			/* try_lock must be 0 if i is 0. */
+			/*
+			 * try_lock means we have an inode locked
+			 * that is in the AIL.
+			 */
+			ASSERT(i != 0);
+			if (!xfs_ilock_nowait(ips[i], xfs_lock_inumorder(lock_mode, i))) {
+				attempts++;
+
+				/*
+				 * Unlock all previous guys and try again.
+				 * xfs_iunlock will try to push the tail
+				 * if the inode is in the AIL.
+				 */
+
+				for(j = i - 1; j >= 0; j--) {
+
+					/*
+					 * Check to see if we've already
+					 * unlocked this one.
+					 * Not the first one going back,
+					 * and the inode ptr is the same.
+					 */
+					if ((j != (i - 1)) && ips[j] ==
+								ips[j+1])
+						continue;
+
+					xfs_iunlock(ips[j], lock_mode);
+				}
+
+				if ((attempts % 5) == 0) {
+					delay(1); /* Don't just spin the CPU */
+#ifdef DEBUG
+					xfs_lock_delays++;
+#endif
+				}
+				i = 0;
+				try_lock = 0;
+				goto again;
+			}
+		} else {
+			xfs_ilock(ips[i], xfs_lock_inumorder(lock_mode, i));
+		}
+	}
+
+#ifdef DEBUG
+	if (attempts) {
+		if (attempts < 5) xfs_small_retries++;
+		else if (attempts < 100) xfs_middle_retries++;
+		else xfs_lots_retries++;
+	} else {
+		xfs_locked_n++;
+	}
+#endif
+}
+
+/*
+ * xfs_lock_two_inodes() can only be used to lock one type of lock
+ * at a time - the iolock or the ilock, but not both at once. If
+ * we lock both at once, lockdep will report false positives saying
+ * we have violated locking orders.
+ */
+void
+xfs_lock_two_inodes(
+	xfs_inode_t		*ip0,
+	xfs_inode_t		*ip1,
+	uint			lock_mode)
+{
+	xfs_inode_t		*temp;
+	int			attempts = 0;
+	xfs_log_item_t		*lp;
+
+	if (lock_mode & (XFS_IOLOCK_SHARED|XFS_IOLOCK_EXCL))
+		ASSERT((lock_mode & (XFS_ILOCK_SHARED|XFS_ILOCK_EXCL)) == 0);
+	ASSERT(ip0->i_ino != ip1->i_ino);
+
+	if (ip0->i_ino > ip1->i_ino) {
+		temp = ip0;
+		ip0 = ip1;
+		ip1 = temp;
+	}
+
+ again:
+	xfs_ilock(ip0, xfs_lock_inumorder(lock_mode, 0));
+
+	/*
+	 * If the first lock we have locked is in the AIL, we must TRY to get
+	 * the second lock. If we can't get it, we must release the first one
+	 * and try again.
+	 */
+	lp = (xfs_log_item_t *)ip0->i_itemp;
+	if (lp && (lp->li_flags & XFS_LI_IN_AIL)) {
+		if (!xfs_ilock_nowait(ip1, xfs_lock_inumorder(lock_mode, 1))) {
+			xfs_iunlock(ip0, lock_mode);
+			if ((++attempts % 5) == 0)
+				delay(1); /* Don't just spin the CPU */
+			goto again;
+		}
+	} else {
+		xfs_ilock(ip1, xfs_lock_inumorder(lock_mode, 1));
+	}
+}
+
+int
+xfs_remove(
+	xfs_inode_t             *dp,
+	struct xfs_name		*name,
+	xfs_inode_t		*ip)
+{
+	xfs_mount_t		*mp = dp->i_mount;
+	xfs_trans_t             *tp = NULL;
+	int			is_dir = S_ISDIR(ip->i_d.di_mode);
+	int                     error = 0;
+	xfs_bmap_free_t         free_list;
+	xfs_fsblock_t           first_block;
+	int			cancel_flags;
+	int			committed;
+	int			link_zero;
+	uint			resblks;
+	uint			log_count;
+
+	trace_xfs_remove(dp, name);
+
+	if (XFS_FORCED_SHUTDOWN(mp))
+		return XFS_ERROR(EIO);
+
+	error = xfs_qm_dqattach(dp, 0);
+	if (error)
+		goto std_return;
+
+	error = xfs_qm_dqattach(ip, 0);
+	if (error)
+		goto std_return;
+
+	if (is_dir) {
+		tp = xfs_trans_alloc(mp, XFS_TRANS_RMDIR);
+		log_count = XFS_DEFAULT_LOG_COUNT;
+	} else {
+		tp = xfs_trans_alloc(mp, XFS_TRANS_REMOVE);
+		log_count = XFS_REMOVE_LOG_COUNT;
+	}
+	cancel_flags = XFS_TRANS_RELEASE_LOG_RES;
+
+	/*
+	 * We try to get the real space reservation first,
+	 * allowing for directory btree deletion(s) implying
+	 * possible bmap insert(s).  If we can't get the space
+	 * reservation then we use 0 instead, and avoid the bmap
+	 * btree insert(s) in the directory code by, if the bmap
+	 * insert tries to happen, instead trimming the LAST
+	 * block from the directory.
+	 */
+	resblks = XFS_REMOVE_SPACE_RES(mp);
+	error = xfs_trans_reserve(tp, resblks, XFS_REMOVE_LOG_RES(mp), 0,
+				  XFS_TRANS_PERM_LOG_RES, log_count);
+	if (error == ENOSPC) {
+		resblks = 0;
+		error = xfs_trans_reserve(tp, 0, XFS_REMOVE_LOG_RES(mp), 0,
+					  XFS_TRANS_PERM_LOG_RES, log_count);
+	}
+	if (error) {
+		ASSERT(error != ENOSPC);
+		cancel_flags = 0;
+		goto out_trans_cancel;
+	}
+
+	xfs_lock_two_inodes(dp, ip, XFS_ILOCK_EXCL);
+
+	xfs_trans_ijoin_ref(tp, dp, XFS_ILOCK_EXCL);
+	xfs_trans_ijoin_ref(tp, ip, XFS_ILOCK_EXCL);
+
+	/*
+	 * If we're removing a directory perform some additional validation.
+	 */
+	if (is_dir) {
+		ASSERT(ip->i_d.di_nlink >= 2);
+		if (ip->i_d.di_nlink != 2) {
+			error = XFS_ERROR(ENOTEMPTY);
+			goto out_trans_cancel;
+		}
+		if (!xfs_dir_isempty(ip)) {
+			error = XFS_ERROR(ENOTEMPTY);
+			goto out_trans_cancel;
+		}
+	}
+
+	xfs_bmap_init(&free_list, &first_block);
+	error = xfs_dir_removename(tp, dp, name, ip->i_ino,
+					&first_block, &free_list, resblks);
+	if (error) {
+		ASSERT(error != ENOENT);
+		goto out_bmap_cancel;
+	}
+	xfs_trans_ichgtime(tp, dp, XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG);
+
+	if (is_dir) {
+		/*
+		 * Drop the link from ip's "..".
+		 */
+		error = xfs_droplink(tp, dp);
+		if (error)
+			goto out_bmap_cancel;
+
+		/*
+		 * Drop the "." link from ip to self.
+		 */
+		error = xfs_droplink(tp, ip);
+		if (error)
+			goto out_bmap_cancel;
+	} else {
+		/*
+		 * When removing a non-directory we need to log the parent
+		 * inode here.  For a directory this is done implicitly
+		 * by the xfs_droplink call for the ".." entry.
+		 */
+		xfs_trans_log_inode(tp, dp, XFS_ILOG_CORE);
+	}
+
+	/*
+	 * Drop the link from dp to ip.
+	 */
+	error = xfs_droplink(tp, ip);
+	if (error)
+		goto out_bmap_cancel;
+
+	/*
+	 * Determine if this is the last link while
+	 * we are in the transaction.
+	 */
+	link_zero = (ip->i_d.di_nlink == 0);
+
+	/*
+	 * If this is a synchronous mount, make sure that the
+	 * remove transaction goes to disk before returning to
+	 * the user.
+	 */
+	if (mp->m_flags & (XFS_MOUNT_WSYNC|XFS_MOUNT_DIRSYNC))
+		xfs_trans_set_sync(tp);
+
+	error = xfs_bmap_finish(&tp, &free_list, &committed);
+	if (error)
+		goto out_bmap_cancel;
+
+	error = xfs_trans_commit(tp, XFS_TRANS_RELEASE_LOG_RES);
+	if (error)
+		goto std_return;
+
+	/*
+	 * If we are using filestreams, kill the stream association.
+	 * If the file is still open it may get a new one but that
+	 * will get killed on last close in xfs_close() so we don't
+	 * have to worry about that.
+	 */
+	if (!is_dir && link_zero && xfs_inode_is_filestream(ip))
+		xfs_filestream_deassociate(ip);
+
+	return 0;
+
+ out_bmap_cancel:
+	xfs_bmap_cancel(&free_list);
+	cancel_flags |= XFS_TRANS_ABORT;
+ out_trans_cancel:
+	xfs_trans_cancel(tp, cancel_flags);
+ std_return:
+	return error;
+}
+
+int
+xfs_link(
+	xfs_inode_t		*tdp,
+	xfs_inode_t		*sip,
+	struct xfs_name		*target_name)
+{
+	xfs_mount_t		*mp = tdp->i_mount;
+	xfs_trans_t		*tp;
+	int			error;
+	xfs_bmap_free_t         free_list;
+	xfs_fsblock_t           first_block;
+	int			cancel_flags;
+	int			committed;
+	int			resblks;
+
+	trace_xfs_link(tdp, target_name);
+
+	ASSERT(!S_ISDIR(sip->i_d.di_mode));
+
+	if (XFS_FORCED_SHUTDOWN(mp))
+		return XFS_ERROR(EIO);
+
+	error = xfs_qm_dqattach(sip, 0);
+	if (error)
+		goto std_return;
+
+	error = xfs_qm_dqattach(tdp, 0);
+	if (error)
+		goto std_return;
+
+	tp = xfs_trans_alloc(mp, XFS_TRANS_LINK);
+	cancel_flags = XFS_TRANS_RELEASE_LOG_RES;
+	resblks = XFS_LINK_SPACE_RES(mp, target_name->len);
+	error = xfs_trans_reserve(tp, resblks, XFS_LINK_LOG_RES(mp), 0,
+			XFS_TRANS_PERM_LOG_RES, XFS_LINK_LOG_COUNT);
+	if (error == ENOSPC) {
+		resblks = 0;
+		error = xfs_trans_reserve(tp, 0, XFS_LINK_LOG_RES(mp), 0,
+				XFS_TRANS_PERM_LOG_RES, XFS_LINK_LOG_COUNT);
+	}
+	if (error) {
+		cancel_flags = 0;
+		goto error_return;
+	}
+
+	xfs_lock_two_inodes(sip, tdp, XFS_ILOCK_EXCL);
+
+	xfs_trans_ijoin_ref(tp, sip, XFS_ILOCK_EXCL);
+	xfs_trans_ijoin_ref(tp, tdp, XFS_ILOCK_EXCL);
+
+	/*
+	 * If the source has too many links, we can't make any more to it.
+	 */
+	if (sip->i_d.di_nlink >= XFS_MAXLINK) {
+		error = XFS_ERROR(EMLINK);
+		goto error_return;
+	}
+
+	/*
+	 * If we are using project inheritance, we only allow hard link
+	 * creation in our tree when the project IDs are the same; else
+	 * the tree quota mechanism could be circumvented.
+	 */
+	if (unlikely((tdp->i_d.di_flags & XFS_DIFLAG_PROJINHERIT) &&
+		     (xfs_get_projid(tdp) != xfs_get_projid(sip)))) {
+		error = XFS_ERROR(EXDEV);
+		goto error_return;
+	}
+
+	error = xfs_dir_canenter(tp, tdp, target_name, resblks);
+	if (error)
+		goto error_return;
+
+	xfs_bmap_init(&free_list, &first_block);
+
+	error = xfs_dir_createname(tp, tdp, target_name, sip->i_ino,
+					&first_block, &free_list, resblks);
+	if (error)
+		goto abort_return;
+	xfs_trans_ichgtime(tp, tdp, XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG);
+	xfs_trans_log_inode(tp, tdp, XFS_ILOG_CORE);
+
+	error = xfs_bumplink(tp, sip);
+	if (error)
+		goto abort_return;
+
+	/*
+	 * If this is a synchronous mount, make sure that the
+	 * link transaction goes to disk before returning to
+	 * the user.
+	 */
+	if (mp->m_flags & (XFS_MOUNT_WSYNC|XFS_MOUNT_DIRSYNC)) {
+		xfs_trans_set_sync(tp);
+	}
+
+	error = xfs_bmap_finish (&tp, &free_list, &committed);
+	if (error) {
+		xfs_bmap_cancel(&free_list);
+		goto abort_return;
+	}
+
+	return xfs_trans_commit(tp, XFS_TRANS_RELEASE_LOG_RES);
+
+ abort_return:
+	cancel_flags |= XFS_TRANS_ABORT;
+ error_return:
+	xfs_trans_cancel(tp, cancel_flags);
+ std_return:
+	return error;
+}
+
+int
+xfs_symlink(
+	xfs_inode_t		*dp,
+	struct xfs_name		*link_name,
+	const char		*target_path,
+	mode_t			mode,
+	xfs_inode_t		**ipp)
+{
+	xfs_mount_t		*mp = dp->i_mount;
+	xfs_trans_t		*tp;
+	xfs_inode_t		*ip;
+	int			error;
+	int			pathlen;
+	xfs_bmap_free_t		free_list;
+	xfs_fsblock_t		first_block;
+	boolean_t		unlock_dp_on_error = B_FALSE;
+	uint			cancel_flags;
+	int			committed;
+	xfs_fileoff_t		first_fsb;
+	xfs_filblks_t		fs_blocks;
+	int			nmaps;
+	xfs_bmbt_irec_t		mval[SYMLINK_MAPS];
+	xfs_daddr_t		d;
+	const char		*cur_chunk;
+	int			byte_cnt;
+	int			n;
+	xfs_buf_t		*bp;
+	prid_t			prid;
+	struct xfs_dquot	*udqp, *gdqp;
+	uint			resblks;
+
+	*ipp = NULL;
+	error = 0;
+	ip = NULL;
+	tp = NULL;
+
+	trace_xfs_symlink(dp, link_name);
+
+	if (XFS_FORCED_SHUTDOWN(mp))
+		return XFS_ERROR(EIO);
+
+	/*
+	 * Check component lengths of the target path name.
+	 */
+	pathlen = strlen(target_path);
+	if (pathlen >= MAXPATHLEN)      /* total string too long */
+		return XFS_ERROR(ENAMETOOLONG);
+
+	udqp = gdqp = NULL;
+	if (dp->i_d.di_flags & XFS_DIFLAG_PROJINHERIT)
+		prid = xfs_get_projid(dp);
+	else
+		prid = XFS_PROJID_DEFAULT;
+
+	/*
+	 * Make sure that we have allocated dquot(s) on disk.
+	 */
+	error = xfs_qm_vop_dqalloc(dp, current_fsuid(), current_fsgid(), prid,
+			XFS_QMOPT_QUOTALL | XFS_QMOPT_INHERIT, &udqp, &gdqp);
+	if (error)
+		goto std_return;
+
+	tp = xfs_trans_alloc(mp, XFS_TRANS_SYMLINK);
+	cancel_flags = XFS_TRANS_RELEASE_LOG_RES;
+	/*
+	 * The symlink will fit into the inode data fork?
+	 * There can't be any attributes so we get the whole variable part.
+	 */
+	if (pathlen <= XFS_LITINO(mp))
+		fs_blocks = 0;
+	else
+		fs_blocks = XFS_B_TO_FSB(mp, pathlen);
+	resblks = XFS_SYMLINK_SPACE_RES(mp, link_name->len, fs_blocks);
+	error = xfs_trans_reserve(tp, resblks, XFS_SYMLINK_LOG_RES(mp), 0,
+			XFS_TRANS_PERM_LOG_RES, XFS_SYMLINK_LOG_COUNT);
+	if (error == ENOSPC && fs_blocks == 0) {
+		resblks = 0;
+		error = xfs_trans_reserve(tp, 0, XFS_SYMLINK_LOG_RES(mp), 0,
+				XFS_TRANS_PERM_LOG_RES, XFS_SYMLINK_LOG_COUNT);
+	}
+	if (error) {
+		cancel_flags = 0;
+		goto error_return;
+	}
+
+	xfs_ilock(dp, XFS_ILOCK_EXCL | XFS_ILOCK_PARENT);
+	unlock_dp_on_error = B_TRUE;
+
+	/*
+	 * Check whether the directory allows new symlinks or not.
+	 */
+	if (dp->i_d.di_flags & XFS_DIFLAG_NOSYMLINKS) {
+		error = XFS_ERROR(EPERM);
+		goto error_return;
+	}
+
+	/*
+	 * Reserve disk quota : blocks and inode.
+	 */
+	error = xfs_trans_reserve_quota(tp, mp, udqp, gdqp, resblks, 1, 0);
+	if (error)
+		goto error_return;
+
+	/*
+	 * Check for ability to enter directory entry, if no space reserved.
+	 */
+	error = xfs_dir_canenter(tp, dp, link_name, resblks);
+	if (error)
+		goto error_return;
+	/*
+	 * Initialize the bmap freelist prior to calling either
+	 * bmapi or the directory create code.
+	 */
+	xfs_bmap_init(&free_list, &first_block);
+
+	/*
+	 * Allocate an inode for the symlink.
+	 */
+	error = xfs_dir_ialloc(&tp, dp, S_IFLNK | (mode & ~S_IFMT), 1, 0,
+			       prid, resblks > 0, &ip, NULL);
+	if (error) {
+		if (error == ENOSPC)
+			goto error_return;
+		goto error1;
+	}
+
+	/*
+	 * An error after we've joined dp to the transaction will result in the
+	 * transaction cancel unlocking dp so don't do it explicitly in the
+	 * error path.
+	 */
+	xfs_trans_ijoin_ref(tp, dp, XFS_ILOCK_EXCL);
+	unlock_dp_on_error = B_FALSE;
+
+	/*
+	 * Also attach the dquot(s) to it, if applicable.
+	 */
+	xfs_qm_vop_create_dqattach(tp, ip, udqp, gdqp);
+
+	if (resblks)
+		resblks -= XFS_IALLOC_SPACE_RES(mp);
+	/*
+	 * If the symlink will fit into the inode, write it inline.
+	 */
+	if (pathlen <= XFS_IFORK_DSIZE(ip)) {
+		xfs_idata_realloc(ip, pathlen, XFS_DATA_FORK);
+		memcpy(ip->i_df.if_u1.if_data, target_path, pathlen);
+		ip->i_d.di_size = pathlen;
+
+		/*
+		 * The inode was initially created in extent format.
+		 */
+		ip->i_df.if_flags &= ~(XFS_IFEXTENTS | XFS_IFBROOT);
+		ip->i_df.if_flags |= XFS_IFINLINE;
+
+		ip->i_d.di_format = XFS_DINODE_FMT_LOCAL;
+		xfs_trans_log_inode(tp, ip, XFS_ILOG_DDATA | XFS_ILOG_CORE);
+
+	} else {
+		first_fsb = 0;
+		nmaps = SYMLINK_MAPS;
+
+		error = xfs_bmapi(tp, ip, first_fsb, fs_blocks,
+				  XFS_BMAPI_WRITE | XFS_BMAPI_METADATA,
+				  &first_block, resblks, mval, &nmaps,
+				  &free_list);
+		if (error)
+			goto error2;
+
+		if (resblks)
+			resblks -= fs_blocks;
+		ip->i_d.di_size = pathlen;
+		xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
+
+		cur_chunk = target_path;
+		for (n = 0; n < nmaps; n++) {
+			d = XFS_FSB_TO_DADDR(mp, mval[n].br_startblock);
+			byte_cnt = XFS_FSB_TO_B(mp, mval[n].br_blockcount);
+			bp = xfs_trans_get_buf(tp, mp->m_ddev_targp, d,
+					       BTOBB(byte_cnt), 0);
+			ASSERT(bp && !XFS_BUF_GETERROR(bp));
+			if (pathlen < byte_cnt) {
+				byte_cnt = pathlen;
+			}
+			pathlen -= byte_cnt;
+
+			memcpy(XFS_BUF_PTR(bp), cur_chunk, byte_cnt);
+			cur_chunk += byte_cnt;
+
+			xfs_trans_log_buf(tp, bp, 0, byte_cnt - 1);
+		}
+	}
+
+	/*
+	 * Create the directory entry for the symlink.
+	 */
+	error = xfs_dir_createname(tp, dp, link_name, ip->i_ino,
+					&first_block, &free_list, resblks);
+	if (error)
+		goto error2;
+	xfs_trans_ichgtime(tp, dp, XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG);
+	xfs_trans_log_inode(tp, dp, XFS_ILOG_CORE);
+
+	/*
+	 * If this is a synchronous mount, make sure that the
+	 * symlink transaction goes to disk before returning to
+	 * the user.
+	 */
+	if (mp->m_flags & (XFS_MOUNT_WSYNC|XFS_MOUNT_DIRSYNC)) {
+		xfs_trans_set_sync(tp);
+	}
+
+	error = xfs_bmap_finish(&tp, &free_list, &committed);
+	if (error) {
+		goto error2;
+	}
+	error = xfs_trans_commit(tp, XFS_TRANS_RELEASE_LOG_RES);
+	xfs_qm_dqrele(udqp);
+	xfs_qm_dqrele(gdqp);
+
+	*ipp = ip;
+	return 0;
+
+ error2:
+	IRELE(ip);
+ error1:
+	xfs_bmap_cancel(&free_list);
+	cancel_flags |= XFS_TRANS_ABORT;
+ error_return:
+	xfs_trans_cancel(tp, cancel_flags);
+	xfs_qm_dqrele(udqp);
+	xfs_qm_dqrele(gdqp);
+
+	if (unlock_dp_on_error)
+		xfs_iunlock(dp, XFS_ILOCK_EXCL);
+ std_return:
+	return error;
+}
+
+int
+xfs_set_dmattrs(
+	xfs_inode_t     *ip,
+	u_int		evmask,
+	u_int16_t	state)
+{
+	xfs_mount_t	*mp = ip->i_mount;
+	xfs_trans_t	*tp;
+	int		error;
+
+	if (!capable(CAP_SYS_ADMIN))
+		return XFS_ERROR(EPERM);
+
+	if (XFS_FORCED_SHUTDOWN(mp))
+		return XFS_ERROR(EIO);
+
+	tp = xfs_trans_alloc(mp, XFS_TRANS_SET_DMATTRS);
+	error = xfs_trans_reserve(tp, 0, XFS_ICHANGE_LOG_RES (mp), 0, 0, 0);
+	if (error) {
+		xfs_trans_cancel(tp, 0);
+		return error;
+	}
+	xfs_ilock(ip, XFS_ILOCK_EXCL);
+	xfs_trans_ijoin_ref(tp, ip, XFS_ILOCK_EXCL);
+
+	ip->i_d.di_dmevmask = evmask;
+	ip->i_d.di_dmstate  = state;
+
+	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
+	error = xfs_trans_commit(tp, 0);
+
+	return error;
+}
+
+/*
+ * xfs_alloc_file_space()
+ *      This routine allocates disk space for the given file.
+ *
+ *	If alloc_type == 0, this request is for an ALLOCSP type
+ *	request which will change the file size.  In this case, no
+ *	DMAPI event will be generated by the call.  A TRUNCATE event
+ *	will be generated later by xfs_setattr.
+ *
+ *	If alloc_type != 0, this request is for a RESVSP type
+ *	request, and a DMAPI DM_EVENT_WRITE will be generated if the
+ *	lower block boundary byte address is less than the file's
+ *	length.
+ *
+ * RETURNS:
+ *       0 on success
+ *      errno on error
+ *
+ */
+STATIC int
+xfs_alloc_file_space(
+	xfs_inode_t		*ip,
+	xfs_off_t		offset,
+	xfs_off_t		len,
+	int			alloc_type,
+	int			attr_flags)
+{
+	xfs_mount_t		*mp = ip->i_mount;
+	xfs_off_t		count;
+	xfs_filblks_t		allocated_fsb;
+	xfs_filblks_t		allocatesize_fsb;
+	xfs_extlen_t		extsz, temp;
+	xfs_fileoff_t		startoffset_fsb;
+	xfs_fsblock_t		firstfsb;
+	int			nimaps;
+	int			bmapi_flag;
+	int			quota_flag;
+	int			rt;
+	xfs_trans_t		*tp;
+	xfs_bmbt_irec_t		imaps[1], *imapp;
+	xfs_bmap_free_t		free_list;
+	uint			qblocks, resblks, resrtextents;
+	int			committed;
+	int			error;
+
+	trace_xfs_alloc_file_space(ip);
+
+	if (XFS_FORCED_SHUTDOWN(mp))
+		return XFS_ERROR(EIO);
+
+	error = xfs_qm_dqattach(ip, 0);
+	if (error)
+		return error;
+
+	if (len <= 0)
+		return XFS_ERROR(EINVAL);
+
+	rt = XFS_IS_REALTIME_INODE(ip);
+	extsz = xfs_get_extsz_hint(ip);
+
+	count = len;
+	imapp = &imaps[0];
+	nimaps = 1;
+	bmapi_flag = XFS_BMAPI_WRITE | alloc_type;
+	startoffset_fsb	= XFS_B_TO_FSBT(mp, offset);
+	allocatesize_fsb = XFS_B_TO_FSB(mp, count);
+
+	/*
+	 * Allocate file space until done or until there is an error
+	 */
+	while (allocatesize_fsb && !error) {
+		xfs_fileoff_t	s, e;
+
+		/*
+		 * Determine space reservations for data/realtime.
+		 */
+		if (unlikely(extsz)) {
+			s = startoffset_fsb;
+			do_div(s, extsz);
+			s *= extsz;
+			e = startoffset_fsb + allocatesize_fsb;
+			if ((temp = do_mod(startoffset_fsb, extsz)))
+				e += temp;
+			if ((temp = do_mod(e, extsz)))
+				e += extsz - temp;
+		} else {
+			s = 0;
+			e = allocatesize_fsb;
+		}
+
+		/*
+		 * The transaction reservation is limited to a 32-bit block
+		 * count, hence we need to limit the number of blocks we are
+		 * trying to reserve to avoid an overflow. We can't allocate
+		 * more than @nimaps extents, and an extent is limited on disk
+		 * to MAXEXTLEN (21 bits), so use that to enforce the limit.
+		 */
+		resblks = min_t(xfs_fileoff_t, (e - s), (MAXEXTLEN * nimaps));
+		if (unlikely(rt)) {
+			resrtextents = qblocks = resblks;
+			resrtextents /= mp->m_sb.sb_rextsize;
+			resblks = XFS_DIOSTRAT_SPACE_RES(mp, 0);
+			quota_flag = XFS_QMOPT_RES_RTBLKS;
+		} else {
+			resrtextents = 0;
+			resblks = qblocks = XFS_DIOSTRAT_SPACE_RES(mp, resblks);
+			quota_flag = XFS_QMOPT_RES_REGBLKS;
+		}
+
+		/*
+		 * Allocate and setup the transaction.
+		 */
+		tp = xfs_trans_alloc(mp, XFS_TRANS_DIOSTRAT);
+		error = xfs_trans_reserve(tp, resblks,
+					  XFS_WRITE_LOG_RES(mp), resrtextents,
+					  XFS_TRANS_PERM_LOG_RES,
+					  XFS_WRITE_LOG_COUNT);
+		/*
+		 * Check for running out of space
+		 */
+		if (error) {
+			/*
+			 * Free the transaction structure.
+			 */
+			ASSERT(error == ENOSPC || XFS_FORCED_SHUTDOWN(mp));
+			xfs_trans_cancel(tp, 0);
+			break;
+		}
+		xfs_ilock(ip, XFS_ILOCK_EXCL);
+		error = xfs_trans_reserve_quota_nblks(tp, ip, qblocks,
+						      0, quota_flag);
+		if (error)
+			goto error1;
+
+		xfs_trans_ijoin(tp, ip);
+
+		/*
+		 * Issue the xfs_bmapi() call to allocate the blocks
+		 */
+		xfs_bmap_init(&free_list, &firstfsb);
+		error = xfs_bmapi(tp, ip, startoffset_fsb,
+				  allocatesize_fsb, bmapi_flag,
+				  &firstfsb, 0, imapp, &nimaps,
+				  &free_list);
+		if (error) {
+			goto error0;
+		}
+
+		/*
+		 * Complete the transaction
+		 */
+		error = xfs_bmap_finish(&tp, &free_list, &committed);
+		if (error) {
+			goto error0;
+		}
+
+		error = xfs_trans_commit(tp, XFS_TRANS_RELEASE_LOG_RES);
+		xfs_iunlock(ip, XFS_ILOCK_EXCL);
+		if (error) {
+			break;
+		}
+
+		allocated_fsb = imapp->br_blockcount;
+
+		if (nimaps == 0) {
+			error = XFS_ERROR(ENOSPC);
+			break;
+		}
+
+		startoffset_fsb += allocated_fsb;
+		allocatesize_fsb -= allocated_fsb;
+	}
+
+	return error;
+
+error0:	/* Cancel bmap, unlock inode, unreserve quota blocks, cancel trans */
+	xfs_bmap_cancel(&free_list);
+	xfs_trans_unreserve_quota_nblks(tp, ip, qblocks, 0, quota_flag);
+
+error1:	/* Just cancel transaction */
+	xfs_trans_cancel(tp, XFS_TRANS_RELEASE_LOG_RES | XFS_TRANS_ABORT);
+	xfs_iunlock(ip, XFS_ILOCK_EXCL);
+	return error;
+}
+
+/*
+ * Zero file bytes between startoff and endoff inclusive.
+ * The iolock is held exclusive and no blocks are buffered.
+ *
+ * This function is used by xfs_free_file_space() to zero
+ * partial blocks when the range to free is not block aligned.
+ * When unreserving space with boundaries that are not block
+ * aligned we round up the start and round down the end
+ * boundaries and then use this function to zero the parts of
+ * the blocks that got dropped during the rounding.
+ */
+STATIC int
+xfs_zero_remaining_bytes(
+	xfs_inode_t		*ip,
+	xfs_off_t		startoff,
+	xfs_off_t		endoff)
+{
+	xfs_bmbt_irec_t		imap;
+	xfs_fileoff_t		offset_fsb;
+	xfs_off_t		lastoffset;
+	xfs_off_t		offset;
+	xfs_buf_t		*bp;
+	xfs_mount_t		*mp = ip->i_mount;
+	int			nimap;
+	int			error = 0;
+
+	/*
+	 * Avoid doing I/O beyond eof - it's not necessary
+	 * since nothing can read beyond eof.  The space will
+	 * be zeroed when the file is extended anyway.
+	 */
+	if (startoff >= ip->i_size)
+		return 0;
+
+	if (endoff > ip->i_size)
+		endoff = ip->i_size;
+
+	bp = xfs_buf_get_uncached(XFS_IS_REALTIME_INODE(ip) ?
+					mp->m_rtdev_targp : mp->m_ddev_targp,
+				mp->m_sb.sb_blocksize, XBF_DONT_BLOCK);
+	if (!bp)
+		return XFS_ERROR(ENOMEM);
+
+	for (offset = startoff; offset <= endoff; offset = lastoffset + 1) {
+		offset_fsb = XFS_B_TO_FSBT(mp, offset);
+		nimap = 1;
+		error = xfs_bmapi(NULL, ip, offset_fsb, 1, 0,
+			NULL, 0, &imap, &nimap, NULL);
+		if (error || nimap < 1)
+			break;
+		ASSERT(imap.br_blockcount >= 1);
+		ASSERT(imap.br_startoff == offset_fsb);
+		lastoffset = XFS_FSB_TO_B(mp, imap.br_startoff + 1) - 1;
+		if (lastoffset > endoff)
+			lastoffset = endoff;
+		if (imap.br_startblock == HOLESTARTBLOCK)
+			continue;
+		ASSERT(imap.br_startblock != DELAYSTARTBLOCK);
+		if (imap.br_state == XFS_EXT_UNWRITTEN)
+			continue;
+		XFS_BUF_UNDONE(bp);
+		XFS_BUF_UNWRITE(bp);
+		XFS_BUF_READ(bp);
+		XFS_BUF_SET_ADDR(bp, xfs_fsb_to_db(ip, imap.br_startblock));
+		xfsbdstrat(mp, bp);
+		error = xfs_buf_iowait(bp);
+		if (error) {
+			xfs_ioerror_alert("xfs_zero_remaining_bytes(read)",
+					  mp, bp, XFS_BUF_ADDR(bp));
+			break;
+		}
+		memset(XFS_BUF_PTR(bp) +
+			(offset - XFS_FSB_TO_B(mp, imap.br_startoff)),
+		      0, lastoffset - offset + 1);
+		XFS_BUF_UNDONE(bp);
+		XFS_BUF_UNREAD(bp);
+		XFS_BUF_WRITE(bp);
+		xfsbdstrat(mp, bp);
+		error = xfs_buf_iowait(bp);
+		if (error) {
+			xfs_ioerror_alert("xfs_zero_remaining_bytes(write)",
+					  mp, bp, XFS_BUF_ADDR(bp));
+			break;
+		}
+	}
+	xfs_buf_free(bp);
+	return error;
+}
+
+/*
+ * xfs_free_file_space()
+ *      This routine frees disk space for the given file.
+ *
+ *	This routine is only called by xfs_change_file_space
+ *	for an UNRESVSP type call.
+ *
+ * RETURNS:
+ *       0 on success
+ *      errno on error
+ *
+ */
+STATIC int
+xfs_free_file_space(
+	xfs_inode_t		*ip,
+	xfs_off_t		offset,
+	xfs_off_t		len,
+	int			attr_flags)
+{
+	int			committed;
+	int			done;
+	xfs_fileoff_t		endoffset_fsb;
+	int			error;
+	xfs_fsblock_t		firstfsb;
+	xfs_bmap_free_t		free_list;
+	xfs_bmbt_irec_t		imap;
+	xfs_off_t		ioffset;
+	xfs_extlen_t		mod=0;
+	xfs_mount_t		*mp;
+	int			nimap;
+	uint			resblks;
+	uint			rounding;
+	int			rt;
+	xfs_fileoff_t		startoffset_fsb;
+	xfs_trans_t		*tp;
+	int			need_iolock = 1;
+
+	mp = ip->i_mount;
+
+	trace_xfs_free_file_space(ip);
+
+	error = xfs_qm_dqattach(ip, 0);
+	if (error)
+		return error;
+
+	error = 0;
+	if (len <= 0)	/* if nothing being freed */
+		return error;
+	rt = XFS_IS_REALTIME_INODE(ip);
+	startoffset_fsb	= XFS_B_TO_FSB(mp, offset);
+	endoffset_fsb = XFS_B_TO_FSBT(mp, offset + len);
+
+	if (attr_flags & XFS_ATTR_NOLOCK)
+		need_iolock = 0;
+	if (need_iolock) {
+		xfs_ilock(ip, XFS_IOLOCK_EXCL);
+		/* wait for the completion of any pending DIOs */
+		xfs_ioend_wait(ip);
+	}
+
+	rounding = max_t(uint, 1 << mp->m_sb.sb_blocklog, PAGE_CACHE_SIZE);
+	ioffset = offset & ~(rounding - 1);
+
+	if (VN_CACHED(VFS_I(ip)) != 0) {
+		error = xfs_flushinval_pages(ip, ioffset, -1, FI_REMAPF_LOCKED);
+		if (error)
+			goto out_unlock_iolock;
+	}
+
+	/*
+	 * Need to zero the stuff we're not freeing, on disk.
+	 * If it's a realtime file & can't use unwritten extents then we
+	 * actually need to zero the extent edges.  Otherwise xfs_bunmapi
+	 * will take care of it for us.
+	 */
+	if (rt && !xfs_sb_version_hasextflgbit(&mp->m_sb)) {
+		nimap = 1;
+		error = xfs_bmapi(NULL, ip, startoffset_fsb,
+			1, 0, NULL, 0, &imap, &nimap, NULL);
+		if (error)
+			goto out_unlock_iolock;
+		ASSERT(nimap == 0 || nimap == 1);
+		if (nimap && imap.br_startblock != HOLESTARTBLOCK) {
+			xfs_daddr_t	block;
+
+			ASSERT(imap.br_startblock != DELAYSTARTBLOCK);
+			block = imap.br_startblock;
+			mod = do_div(block, mp->m_sb.sb_rextsize);
+			if (mod)
+				startoffset_fsb += mp->m_sb.sb_rextsize - mod;
+		}
+		nimap = 1;
+		error = xfs_bmapi(NULL, ip, endoffset_fsb - 1,
+			1, 0, NULL, 0, &imap, &nimap, NULL);
+		if (error)
+			goto out_unlock_iolock;
+		ASSERT(nimap == 0 || nimap == 1);
+		if (nimap && imap.br_startblock != HOLESTARTBLOCK) {
+			ASSERT(imap.br_startblock != DELAYSTARTBLOCK);
+			mod++;
+			if (mod && (mod != mp->m_sb.sb_rextsize))
+				endoffset_fsb -= mod;
+		}
+	}
+	if ((done = (endoffset_fsb <= startoffset_fsb)))
+		/*
+		 * One contiguous piece to clear
+		 */
+		error = xfs_zero_remaining_bytes(ip, offset, offset + len - 1);
+	else {
+		/*
+		 * Some full blocks, possibly two pieces to clear
+		 */
+		if (offset < XFS_FSB_TO_B(mp, startoffset_fsb))
+			error = xfs_zero_remaining_bytes(ip, offset,
+				XFS_FSB_TO_B(mp, startoffset_fsb) - 1);
+		if (!error &&
+		    XFS_FSB_TO_B(mp, endoffset_fsb) < offset + len)
+			error = xfs_zero_remaining_bytes(ip,
+				XFS_FSB_TO_B(mp, endoffset_fsb),
+				offset + len - 1);
+	}
+
+	/*
+	 * free file space until done or until there is an error
+	 */
+	resblks = XFS_DIOSTRAT_SPACE_RES(mp, 0);
+	while (!error && !done) {
+
+		/*
+		 * allocate and setup the transaction. Allow this
+		 * transaction to dip into the reserve blocks to ensure
+		 * the freeing of the space succeeds at ENOSPC.
+		 */
+		tp = xfs_trans_alloc(mp, XFS_TRANS_DIOSTRAT);
+		tp->t_flags |= XFS_TRANS_RESERVE;
+		error = xfs_trans_reserve(tp,
+					  resblks,
+					  XFS_WRITE_LOG_RES(mp),
+					  0,
+					  XFS_TRANS_PERM_LOG_RES,
+					  XFS_WRITE_LOG_COUNT);
+
+		/*
+		 * check for running out of space
+		 */
+		if (error) {
+			/*
+			 * Free the transaction structure.
+			 */
+			ASSERT(error == ENOSPC || XFS_FORCED_SHUTDOWN(mp));
+			xfs_trans_cancel(tp, 0);
+			break;
+		}
+		xfs_ilock(ip, XFS_ILOCK_EXCL);
+		error = xfs_trans_reserve_quota(tp, mp,
+				ip->i_udquot, ip->i_gdquot,
+				resblks, 0, XFS_QMOPT_RES_REGBLKS);
+		if (error)
+			goto error1;
+
+		xfs_trans_ijoin(tp, ip);
+
+		/*
+		 * issue the bunmapi() call to free the blocks
+		 */
+		xfs_bmap_init(&free_list, &firstfsb);
+		error = xfs_bunmapi(tp, ip, startoffset_fsb,
+				  endoffset_fsb - startoffset_fsb,
+				  0, 2, &firstfsb, &free_list, &done);
+		if (error) {
+			goto error0;
+		}
+
+		/*
+		 * complete the transaction
+		 */
+		error = xfs_bmap_finish(&tp, &free_list, &committed);
+		if (error) {
+			goto error0;
+		}
+
+		error = xfs_trans_commit(tp, XFS_TRANS_RELEASE_LOG_RES);
+		xfs_iunlock(ip, XFS_ILOCK_EXCL);
+	}
+
+ out_unlock_iolock:
+	if (need_iolock)
+		xfs_iunlock(ip, XFS_IOLOCK_EXCL);
+	return error;
+
+ error0:
+	xfs_bmap_cancel(&free_list);
+ error1:
+	xfs_trans_cancel(tp, XFS_TRANS_RELEASE_LOG_RES | XFS_TRANS_ABORT);
+	xfs_iunlock(ip, need_iolock ? (XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL) :
+		    XFS_ILOCK_EXCL);
+	return error;
+}
+
+/*
+ * xfs_change_file_space()
+ *      This routine allocates or frees disk space for the given file.
+ *      The user specified parameters are checked for alignment and size
+ *      limitations.
+ *
+ * RETURNS:
+ *       0 on success
+ *      errno on error
+ *
+ */
+int
+xfs_change_file_space(
+	xfs_inode_t	*ip,
+	int		cmd,
+	xfs_flock64_t	*bf,
+	xfs_off_t	offset,
+	int		attr_flags)
+{
+	xfs_mount_t	*mp = ip->i_mount;
+	int		clrprealloc;
+	int		error;
+	xfs_fsize_t	fsize;
+	int		setprealloc;
+	xfs_off_t	startoffset;
+	xfs_off_t	llen;
+	xfs_trans_t	*tp;
+	struct iattr	iattr;
+	int		prealloc_type;
+
+	if (!S_ISREG(ip->i_d.di_mode))
+		return XFS_ERROR(EINVAL);
+
+	switch (bf->l_whence) {
+	case 0: /*SEEK_SET*/
+		break;
+	case 1: /*SEEK_CUR*/
+		bf->l_start += offset;
+		break;
+	case 2: /*SEEK_END*/
+		bf->l_start += ip->i_size;
+		break;
+	default:
+		return XFS_ERROR(EINVAL);
+	}
+
+	llen = bf->l_len > 0 ? bf->l_len - 1 : bf->l_len;
+
+	if (   (bf->l_start < 0)
+	    || (bf->l_start > XFS_MAXIOFFSET(mp))
+	    || (bf->l_start + llen < 0)
+	    || (bf->l_start + llen > XFS_MAXIOFFSET(mp)))
+		return XFS_ERROR(EINVAL);
+
+	bf->l_whence = 0;
+
+	startoffset = bf->l_start;
+	fsize = ip->i_size;
+
+	/*
+	 * XFS_IOC_RESVSP and XFS_IOC_UNRESVSP will reserve or unreserve
+	 * file space.
+	 * These calls do NOT zero the data space allocated to the file,
+	 * nor do they change the file size.
+	 *
+	 * XFS_IOC_ALLOCSP and XFS_IOC_FREESP will allocate and free file
+	 * space.
+	 * These calls cause the new file data to be zeroed and the file
+	 * size to be changed.
+	 */
+	setprealloc = clrprealloc = 0;
+	prealloc_type = XFS_BMAPI_PREALLOC;
+
+	switch (cmd) {
+	case XFS_IOC_ZERO_RANGE:
+		prealloc_type |= XFS_BMAPI_CONVERT;
+		xfs_tosspages(ip, startoffset, startoffset + bf->l_len, 0);
+		/* FALLTHRU */
+	case XFS_IOC_RESVSP:
+	case XFS_IOC_RESVSP64:
+		error = xfs_alloc_file_space(ip, startoffset, bf->l_len,
+						prealloc_type, attr_flags);
+		if (error)
+			return error;
+		setprealloc = 1;
+		break;
+
+	case XFS_IOC_UNRESVSP:
+	case XFS_IOC_UNRESVSP64:
+		if ((error = xfs_free_file_space(ip, startoffset, bf->l_len,
+								attr_flags)))
+			return error;
+		break;
+
+	case XFS_IOC_ALLOCSP:
+	case XFS_IOC_ALLOCSP64:
+	case XFS_IOC_FREESP:
+	case XFS_IOC_FREESP64:
+		if (startoffset > fsize) {
+			error = xfs_alloc_file_space(ip, fsize,
+					startoffset - fsize, 0, attr_flags);
+			if (error)
+				break;
+		}
+
+		iattr.ia_valid = ATTR_SIZE;
+		iattr.ia_size = startoffset;
+
+		error = xfs_setattr(ip, &iattr, attr_flags);
+
+		if (error)
+			return error;
+
+		clrprealloc = 1;
+		break;
+
+	default:
+		ASSERT(0);
+		return XFS_ERROR(EINVAL);
+	}
+
+	/*
+	 * update the inode timestamp, mode, and prealloc flag bits
+	 */
+	tp = xfs_trans_alloc(mp, XFS_TRANS_WRITEID);
+
+	if ((error = xfs_trans_reserve(tp, 0, XFS_WRITEID_LOG_RES(mp),
+				      0, 0, 0))) {
+		/* ASSERT(0); */
+		xfs_trans_cancel(tp, 0);
+		return error;
+	}
+
+	xfs_ilock(ip, XFS_ILOCK_EXCL);
+
+	xfs_trans_ijoin(tp, ip);
+
+	if ((attr_flags & XFS_ATTR_DMI) == 0) {
+		ip->i_d.di_mode &= ~S_ISUID;
+
+		/*
+		 * Note that we don't have to worry about mandatory
+		 * file locking being disabled here because we only
+		 * clear the S_ISGID bit if the Group execute bit is
+		 * on, but if it was on then mandatory locking wouldn't
+		 * have been enabled.
+		 */
+		if (ip->i_d.di_mode & S_IXGRP)
+			ip->i_d.di_mode &= ~S_ISGID;
+
+		xfs_trans_ichgtime(tp, ip, XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG);
+	}
+	if (setprealloc)
+		ip->i_d.di_flags |= XFS_DIFLAG_PREALLOC;
+	else if (clrprealloc)
+		ip->i_d.di_flags &= ~XFS_DIFLAG_PREALLOC;
+
+	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
+	if (attr_flags & XFS_ATTR_SYNC)
+		xfs_trans_set_sync(tp);
+
+	error = xfs_trans_commit(tp, 0);
+
+	xfs_iunlock(ip, XFS_ILOCK_EXCL);
+
+	return error;
+}
diff --git a/fs/xfs/xfs_vnodeops.h b/fs/xfs/xfs_vnodeops.h
new file mode 100644
index 00000000..3bcd2335
--- /dev/null
+++ b/fs/xfs/xfs_vnodeops.h
@@ -0,0 +1,63 @@
+#ifndef _XFS_VNODEOPS_H
+#define _XFS_VNODEOPS_H 1
+
+struct attrlist_cursor_kern;
+struct file;
+struct iattr;
+struct inode;
+struct iovec;
+struct kiocb;
+struct pipe_inode_info;
+struct uio;
+struct xfs_inode;
+struct xfs_iomap;
+
+
+int xfs_setattr(struct xfs_inode *ip, struct iattr *vap, int flags);
+#define	XFS_ATTR_DMI		0x01	/* invocation from a DMI function */
+#define	XFS_ATTR_NONBLOCK	0x02	/* return EAGAIN if operation would block */
+#define XFS_ATTR_NOLOCK		0x04	/* Don't grab any conflicting locks */
+#define XFS_ATTR_NOACL		0x08	/* Don't call xfs_acl_chmod */
+#define XFS_ATTR_SYNC		0x10	/* synchronous operation required */
+
+int xfs_readlink(struct xfs_inode *ip, char *link);
+int xfs_release(struct xfs_inode *ip);
+int xfs_inactive(struct xfs_inode *ip);
+int xfs_lookup(struct xfs_inode *dp, struct xfs_name *name,
+		struct xfs_inode **ipp, struct xfs_name *ci_name);
+int xfs_create(struct xfs_inode *dp, struct xfs_name *name, mode_t mode,
+		xfs_dev_t rdev, struct xfs_inode **ipp);
+int xfs_remove(struct xfs_inode *dp, struct xfs_name *name,
+		struct xfs_inode *ip);
+int xfs_link(struct xfs_inode *tdp, struct xfs_inode *sip,
+		struct xfs_name *target_name);
+int xfs_readdir(struct xfs_inode	*dp, void *dirent, size_t bufsize,
+		       xfs_off_t *offset, filldir_t filldir);
+int xfs_symlink(struct xfs_inode *dp, struct xfs_name *link_name,
+		const char *target_path, mode_t mode, struct xfs_inode **ipp);
+int xfs_set_dmattrs(struct xfs_inode *ip, u_int evmask, u_int16_t state);
+int xfs_change_file_space(struct xfs_inode *ip, int cmd,
+		xfs_flock64_t *bf, xfs_off_t offset, int attr_flags);
+int xfs_rename(struct xfs_inode *src_dp, struct xfs_name *src_name,
+		struct xfs_inode *src_ip, struct xfs_inode *target_dp,
+		struct xfs_name *target_name, struct xfs_inode *target_ip);
+int xfs_attr_get(struct xfs_inode *ip, const unsigned char *name,
+		unsigned char *value, int *valuelenp, int flags);
+int xfs_attr_set(struct xfs_inode *dp, const unsigned char *name,
+		unsigned char *value, int valuelen, int flags);
+int xfs_attr_remove(struct xfs_inode *dp, const unsigned char *name, int flags);
+int xfs_attr_list(struct xfs_inode *dp, char *buffer, int bufsize,
+		int flags, struct attrlist_cursor_kern *cursor);
+int xfs_bmap(struct xfs_inode *ip, xfs_off_t offset, ssize_t count,
+		int flags, struct xfs_iomap *iomapp, int *niomaps);
+void xfs_tosspages(struct xfs_inode *inode, xfs_off_t first,
+		xfs_off_t last, int fiopt);
+int xfs_flushinval_pages(struct xfs_inode *ip, xfs_off_t first,
+		xfs_off_t last, int fiopt);
+int xfs_flush_pages(struct xfs_inode *ip, xfs_off_t first,
+		xfs_off_t last, uint64_t flags, int fiopt);
+int xfs_wait_on_pages(struct xfs_inode *ip, xfs_off_t first, xfs_off_t last);
+
+int xfs_zero_eof(struct xfs_inode *, xfs_off_t, xfs_fsize_t);
+
+#endif /* _XFS_VNODEOPS_H */