merge with xen-unstable.hg

9 files changed, 236 insertions, 65 deletions

diff --git a/tools/libxc/Makefile b/tools/libxc/Makefile
index d685419da6..c68a0f0d9e 100644
--- a/tools/libxc/Makefile
+++ b/tools/libxc/Makefile
@@ -32,7 +32,7 @@ GUEST_SRCS-y += xg_private.c
 GUEST_SRCS-$(CONFIG_MIGRATE) += xc_domain_restore.c xc_domain_save.c
 GUEST_SRCS-$(CONFIG_HVM) += xc_hvm_build.c
 
-VPATH = ../../xen/common/libelf
+vpath %.c ../../xen/common/libelf
 CFLAGS += -I../../xen/common/libelf
 
 GUEST_SRCS-y += libelf-tools.c libelf-loader.c
diff --git a/tools/libxc/ia64/dom_fw_acpi.c b/tools/libxc/ia64/dom_fw_acpi.c
index cb13a94ea5..493a5adbaf 100644
--- a/tools/libxc/ia64/dom_fw_acpi.c
+++ b/tools/libxc/ia64/dom_fw_acpi.c
@@ -1,14 +1,31 @@
 #include <inttypes.h>
 #include "xc_dom_ia64_util.h"
 #include <xen/acpi.h>
+#include <acpi/actables.h>
 
-uint8_t
-generate_acpi_checksum(void *tbl, unsigned long len)
+/* stolen from xen/drivers/acpi/tables/tbutils.c */
+
+/*******************************************************************************
+ *
+ * FUNCTION:    acpi_tb_checksum
+ *
+ * PARAMETERS:  Buffer          - Pointer to memory region to be checked
+ *              Length          - Length of this memory region
+ *
+ * RETURN:      Checksum (u8)
+ *
+ * DESCRIPTION: Calculates circular checksum of memory region.
+ *
+ ******************************************************************************/
+
+u8 acpi_tb_checksum(u8 * buffer, acpi_native_uint length)
 {
-    uint8_t *ptr, sum = 0;
+	u8 sum = 0;
+	u8 *end = buffer + length;
 
-    for ( ptr = tbl; len > 0 ; len--, ptr++ )
-        sum += *ptr;
+	while (buffer < end) {
+		sum = (u8) (sum + *(buffer++));
+	}
 
-    return 0 - sum;
+	return sum;
 }
diff --git a/tools/libxc/xc_cpuid_x86.c b/tools/libxc/xc_cpuid_x86.c
index febf97c607..2f08182bfd 100644
--- a/tools/libxc/xc_cpuid_x86.c
+++ b/tools/libxc/xc_cpuid_x86.c
@@ -25,19 +25,50 @@
 #include <xen/hvm/params.h>
 
 #define bitmaskof(idx)      (1u << ((idx) & 31))
-#define clear_bit(idx, dst) ((dst) &= ~(1u << (idx)))
-#define set_bit(idx, dst)   ((dst) |= (1u << (idx)))
+#define clear_bit(idx, dst) ((dst) &= ~(1u << ((idx) & 31)))
+#define set_bit(idx, dst)   ((dst) |= (1u << ((idx) & 31)))
 
 #define DEF_MAX_BASE 0x00000004u
 #define DEF_MAX_EXT  0x80000008u
 
-static void amd_xc_cpuid_policy(
-    int xc, domid_t domid, const unsigned int *input, unsigned int *regs)
+static int hypervisor_is_64bit(int xc)
 {
-    unsigned long pae = 0;
+    xen_capabilities_info_t xen_caps = "";
+    return ((xc_version(xc, XENVER_capabilities, &xen_caps) == 0) &&
+            (strstr(xen_caps, "x86_64") != NULL));
+}
 
-    xc_get_hvm_param(xc, domid, HVM_PARAM_PAE_ENABLED, &pae);
+static void cpuid(const unsigned int *input, unsigned int *regs)
+{
+    unsigned int count = (input[1] == XEN_CPUID_INPUT_UNUSED) ? 0 : input[1];
+    asm (
+#ifdef __i386__
+        "push %%ebx; cpuid; mov %%ebx,%1; pop %%ebx"
+#else
+        "push %%rbx; cpuid; mov %%ebx,%1; pop %%rbx"
+#endif
+        : "=a" (regs[0]), "=r" (regs[1]), "=c" (regs[2]), "=d" (regs[3])
+        : "0" (input[0]), "2" (count) );
+}
+
+/* Get the manufacturer brand name of the host processor. */
+static void xc_cpuid_brand_get(char *str)
+{
+    unsigned int input[2] = { 0, 0 };
+    unsigned int regs[4];
+
+    cpuid(input, regs);
+
+    *(uint32_t *)(str + 0) = regs[1];
+    *(uint32_t *)(str + 4) = regs[3];
+    *(uint32_t *)(str + 8) = regs[2];
+    str[12] = '\0';
+}
 
+static void amd_xc_cpuid_policy(
+    int xc, domid_t domid, const unsigned int *input, unsigned int *regs,
+    int is_pae)
+{
     switch ( input[0] )
     {
     case 0x00000001:
@@ -52,21 +83,23 @@ static void amd_xc_cpuid_policy(
         regs[0] = regs[1] = regs[2] = 0;
         break;
 
-    case 0x80000001:
-        if ( !pae )
-            clear_bit(X86_FEATURE_PAE & 31, regs[3]);
-        clear_bit(X86_FEATURE_PSE36 & 31, regs[3]);
+    case 0x80000001: {
+        int is_64bit = hypervisor_is_64bit(xc) && is_pae;
+
+        if ( !is_pae )
+            clear_bit(X86_FEATURE_PAE, regs[3]);
+        clear_bit(X86_FEATURE_PSE36, regs[3]);
 
         /* Filter all other features according to a whitelist. */
-        regs[2] &= (bitmaskof(X86_FEATURE_LAHF_LM) |
+        regs[2] &= ((is_64bit ? bitmaskof(X86_FEATURE_LAHF_LM) : 0) |
                     bitmaskof(X86_FEATURE_ALTMOVCR) |
                     bitmaskof(X86_FEATURE_ABM) |
                     bitmaskof(X86_FEATURE_SSE4A) |
                     bitmaskof(X86_FEATURE_MISALIGNSSE) |
                     bitmaskof(X86_FEATURE_3DNOWPF));
         regs[3] &= (0x0183f3ff | /* features shared with 0x00000001:EDX */
-                    bitmaskof(X86_FEATURE_NX) |
-                    bitmaskof(X86_FEATURE_LM) |
+                    (is_pae ? bitmaskof(X86_FEATURE_NX) : 0) |
+                    (is_64bit ? bitmaskof(X86_FEATURE_LM) : 0) |
                     bitmaskof(X86_FEATURE_SYSCALL) |
                     bitmaskof(X86_FEATURE_MP) |
                     bitmaskof(X86_FEATURE_MMXEXT) |
@@ -75,10 +108,12 @@ static void amd_xc_cpuid_policy(
                     bitmaskof(X86_FEATURE_3DNOWEXT));
         break;
     }
+    }
 }
 
 static void intel_xc_cpuid_policy(
-    int xc, domid_t domid, const unsigned int *input, unsigned int *regs)
+    int xc, domid_t domid, const unsigned int *input, unsigned int *regs,
+    int is_pae)
 {
     switch ( input[0] )
     {
@@ -92,52 +127,30 @@ static void intel_xc_cpuid_policy(
         regs[3] &= 0x3FF;
         break;
 
-    case 0x80000001:
+    case 0x80000001: {
+        int is_64bit = hypervisor_is_64bit(xc) && is_pae;
+
         /* Only a few features are advertised in Intel's 0x80000001. */
-        regs[2] &= (bitmaskof(X86_FEATURE_LAHF_LM));
-        regs[3] &= (bitmaskof(X86_FEATURE_NX) |
-                    bitmaskof(X86_FEATURE_LM) |
-                    bitmaskof(X86_FEATURE_SYSCALL));
+        regs[2] &= (is_64bit ? bitmaskof(X86_FEATURE_LAHF_LM) : 0);
+        regs[3] &= ((is_pae ? bitmaskof(X86_FEATURE_NX) : 0) |
+                    (is_64bit ? bitmaskof(X86_FEATURE_LM) : 0) |
+                    (is_64bit ? bitmaskof(X86_FEATURE_SYSCALL) : 0));
         break;
     }
+    }
 }
 
-static void cpuid(const unsigned int *input, unsigned int *regs)
-{
-    unsigned int count = (input[1] == XEN_CPUID_INPUT_UNUSED) ? 0 : input[1];
-    asm (
-#ifdef __i386__
-        "push %%ebx; cpuid; mov %%ebx,%1; pop %%ebx"
-#else
-        "push %%rbx; cpuid; mov %%ebx,%1; pop %%rbx"
-#endif
-        : "=a" (regs[0]), "=r" (regs[1]), "=c" (regs[2]), "=d" (regs[3])
-        : "0" (input[0]), "2" (count) );
-}
-
-/* Get the manufacturer brand name of the host processor. */
-static void xc_cpuid_brand_get(char *str)
-{
-    unsigned int input[2] = { 0, 0 };
-    unsigned int regs[4];
-
-    cpuid(input, regs);
-
-    *(uint32_t *)(str + 0) = regs[1];
-    *(uint32_t *)(str + 4) = regs[3];
-    *(uint32_t *)(str + 8) = regs[2];
-    str[12] = '\0';
-}
-
-static void xc_cpuid_policy(
+static void xc_cpuid_hvm_policy(
     int xc, domid_t domid, const unsigned int *input, unsigned int *regs)
 {
     char brand[13];
     unsigned long pae;
+    int is_pae;
 
     xc_get_hvm_param(xc, domid, HVM_PARAM_PAE_ENABLED, &pae);
+    is_pae = !!pae;
 
-    switch( input[0] )
+    switch ( input[0] )
     {
     case 0x00000000:
         if ( regs[0] > DEF_MAX_BASE )
@@ -177,8 +190,8 @@ static void xc_cpuid_policy(
         /* We always support MTRR MSRs. */
         regs[3] |= bitmaskof(X86_FEATURE_MTRR);
 
-        if ( !pae )
-            clear_bit(X86_FEATURE_PAE & 31, regs[3]);
+        if ( !is_pae )
+            clear_bit(X86_FEATURE_PAE, regs[3]);
         break;
 
     case 0x80000000:
@@ -187,8 +200,8 @@ static void xc_cpuid_policy(
         break;
 
     case 0x80000001:
-        if ( !pae )
-            clear_bit(X86_FEATURE_NX & 31, regs[3]);
+        if ( !is_pae )
+            clear_bit(X86_FEATURE_NX, regs[3]);
         break;
 
 
@@ -212,9 +225,104 @@ static void xc_cpuid_policy(
 
     xc_cpuid_brand_get(brand);
     if ( strstr(brand, "AMD") )
-        amd_xc_cpuid_policy(xc, domid, input, regs);
+        amd_xc_cpuid_policy(xc, domid, input, regs, is_pae);
     else
-        intel_xc_cpuid_policy(xc, domid, input, regs);
+        intel_xc_cpuid_policy(xc, domid, input, regs, is_pae);
+
+}
+
+static void xc_cpuid_pv_policy(
+    int xc, domid_t domid, const unsigned int *input, unsigned int *regs)
+{
+    DECLARE_DOMCTL;
+    int guest_64bit, xen_64bit = hypervisor_is_64bit(xc);
+    char brand[13];
+
+    xc_cpuid_brand_get(brand);
+
+    memset(&domctl, 0, sizeof(domctl));
+    domctl.domain = domid;
+    domctl.cmd = XEN_DOMCTL_get_address_size;
+    do_domctl(xc, &domctl);
+    guest_64bit = (domctl.u.address_size.size == 64);
+
+    if ( (input[0] & 0x7fffffff) == 1 )
+    {
+        clear_bit(X86_FEATURE_VME, regs[3]);
+        clear_bit(X86_FEATURE_PSE, regs[3]);
+        clear_bit(X86_FEATURE_PGE, regs[3]);
+        clear_bit(X86_FEATURE_MCE, regs[3]);
+        clear_bit(X86_FEATURE_MCA, regs[3]);
+        clear_bit(X86_FEATURE_MTRR, regs[3]);
+        clear_bit(X86_FEATURE_PSE36, regs[3]);
+    }
+
+    switch ( input[0] )
+    {
+    case 1:
+        if ( !xen_64bit || strstr(brand, "AMD") )
+            clear_bit(X86_FEATURE_SEP, regs[3]);
+        clear_bit(X86_FEATURE_DS, regs[3]);
+        clear_bit(X86_FEATURE_ACC, regs[3]);
+        clear_bit(X86_FEATURE_PBE, regs[3]);
+
+        clear_bit(X86_FEATURE_DTES64, regs[2]);
+        clear_bit(X86_FEATURE_MWAIT, regs[2]);
+        clear_bit(X86_FEATURE_DSCPL, regs[2]);
+        clear_bit(X86_FEATURE_VMXE, regs[2]);
+        clear_bit(X86_FEATURE_SMXE, regs[2]);
+        clear_bit(X86_FEATURE_EST, regs[2]);
+        clear_bit(X86_FEATURE_TM2, regs[2]);
+        if ( !guest_64bit )
+            clear_bit(X86_FEATURE_CX16, regs[2]);
+        clear_bit(X86_FEATURE_XTPR, regs[2]);
+        clear_bit(X86_FEATURE_PDCM, regs[2]);
+        clear_bit(X86_FEATURE_DCA, regs[2]);
+        break;
+    case 0x80000001:
+        if ( !guest_64bit )
+        {
+            clear_bit(X86_FEATURE_LM, regs[3]);
+            clear_bit(X86_FEATURE_LAHF_LM, regs[2]);
+            if ( !strstr(brand, "AMD") )
+                clear_bit(X86_FEATURE_SYSCALL, regs[3]);
+        }
+        else
+        {
+            set_bit(X86_FEATURE_SYSCALL, regs[3]);
+        }
+        clear_bit(X86_FEATURE_PAGE1GB, regs[3]);
+        clear_bit(X86_FEATURE_RDTSCP, regs[3]);
+
+        clear_bit(X86_FEATURE_SVME, regs[2]);
+        clear_bit(X86_FEATURE_OSVW, regs[2]);
+        clear_bit(X86_FEATURE_IBS, regs[2]);
+        clear_bit(X86_FEATURE_SKINIT, regs[2]);
+        clear_bit(X86_FEATURE_WDT, regs[2]);
+        break;
+    case 5: /* MONITOR/MWAIT */
+    case 0xa: /* Architectural Performance Monitor Features */
+    case 0x8000000a: /* SVM revision and features */
+    case 0x8000001b: /* Instruction Based Sampling */
+        regs[0] = regs[1] = regs[2] = regs[3] = 0;
+        break;
+    }
+}
+
+static int xc_cpuid_policy(
+    int xc, domid_t domid, const unsigned int *input, unsigned int *regs)
+{
+    xc_dominfo_t        info;
+
+    if ( xc_domain_getinfo(xc, domid, 1, &info) == 0 )
+        return -EINVAL;
+
+    if ( info.hvm )
+        xc_cpuid_hvm_policy(xc, domid, input, regs);
+    else
+        xc_cpuid_pv_policy(xc, domid, input, regs);
+
+    return 0;
 }
 
 static int xc_cpuid_do_domctl(
diff --git a/tools/libxc/xc_dom.h b/tools/libxc/xc_dom.h
index 30111c98e2..770e00bba2 100644
--- a/tools/libxc/xc_dom.h
+++ b/tools/libxc/xc_dom.h
@@ -107,6 +107,8 @@ struct xc_dom_image {
 
     /* kernel loader */
     struct xc_dom_arch *arch_hooks;
+    /* allocate up to virt_alloc_end */
+    int (*allocate) (struct xc_dom_image * dom, xen_vaddr_t up_to);
 };
 
 /* --- pluggable kernel loader ------------------------------------- */
@@ -167,6 +169,7 @@ int xc_dom_ramdisk_mem(struct xc_dom_image *dom, const void *mem,
                        size_t memsize);
 
 int xc_dom_parse_image(struct xc_dom_image *dom);
+struct xc_dom_arch *xc_dom_find_arch_hooks(char *guest_type);
 int xc_dom_build_image(struct xc_dom_image *dom);
 int xc_dom_update_guest_p2m(struct xc_dom_image *dom);
 
diff --git a/tools/libxc/xc_dom_core.c b/tools/libxc/xc_dom_core.c
index 9a158ef5e2..2b6f6b6868 100644
--- a/tools/libxc/xc_dom_core.c
+++ b/tools/libxc/xc_dom_core.c
@@ -409,6 +409,8 @@ int xc_dom_alloc_segment(struct xc_dom_image *dom,
     seg->vend = start + pages * page_size;
     seg->pfn = (seg->vstart - dom->parms.virt_base) / page_size;
     dom->virt_alloc_end = seg->vend;
+    if (dom->allocate)
+        dom->allocate(dom, dom->virt_alloc_end);
 
     xc_dom_printf("%-20s:   %-12s : 0x%" PRIx64 " -> 0x%" PRIx64
                   "  (pfn 0x%" PRIpfn " + 0x%" PRIpfn " pages)\n",
@@ -431,6 +433,8 @@ int xc_dom_alloc_page(struct xc_dom_image *dom, char *name)
 
     start = dom->virt_alloc_end;
     dom->virt_alloc_end += page_size;
+    if (dom->allocate)
+        dom->allocate(dom, dom->virt_alloc_end);
     pfn = (start - dom->parms.virt_base) / page_size;
 
     xc_dom_printf("%-20s:   %-12s : 0x%" PRIx64 " (pfn 0x%" PRIpfn ")\n",
@@ -506,7 +510,7 @@ void xc_dom_register_arch_hooks(struct xc_dom_arch *hooks)
     first_hook = hooks;
 }
 
-static struct xc_dom_arch *xc_dom_find_arch_hooks(char *guest_type)
+struct xc_dom_arch *xc_dom_find_arch_hooks(char *guest_type)
 {
     struct xc_dom_arch *hooks = first_hook;
 
diff --git a/tools/libxc/xc_dom_x86.c b/tools/libxc/xc_dom_x86.c
index 2ed298907c..f96ec3f7ab 100644
--- a/tools/libxc/xc_dom_x86.c
+++ b/tools/libxc/xc_dom_x86.c
@@ -418,7 +418,7 @@ static int start_info_x86_32(struct xc_dom_image *dom)
     xc_dom_printf("%s: called\n", __FUNCTION__);
 
     memset(start_info, 0, sizeof(*start_info));
-    sprintf(start_info->magic, dom->guest_type);
+    snprintf(start_info->magic, sizeof(start_info->magic), dom->guest_type);
     start_info->nr_pages = dom->total_pages;
     start_info->shared_info = shinfo << PAGE_SHIFT_X86;
     start_info->pt_base = dom->pgtables_seg.vstart;
@@ -457,7 +457,7 @@ static int start_info_x86_64(struct xc_dom_image *dom)
     xc_dom_printf("%s: called\n", __FUNCTION__);
 
     memset(start_info, 0, sizeof(*start_info));
-    sprintf(start_info->magic, dom->guest_type);
+    snprintf(start_info->magic, sizeof(start_info->magic), dom->guest_type);
     start_info->nr_pages = dom->total_pages;
     start_info->shared_info = shinfo << PAGE_SHIFT_X86;
     start_info->pt_base = dom->pgtables_seg.vstart;
diff --git a/tools/libxc/xc_domain_restore.c b/tools/libxc/xc_domain_restore.c
index b7b74352e1..46859f63cf 100644
--- a/tools/libxc/xc_domain_restore.c
+++ b/tools/libxc/xc_domain_restore.c
@@ -474,6 +474,22 @@ int xc_domain_restore(int xc_handle, int io_fd, uint32_t dom,
             continue;
         }
 
+        if ( j == -3 )
+        {
+            uint64_t ident_pt;
+
+            /* Skip padding 4 bytes then read the EPT identity PT location. */
+            if ( read_exact(io_fd, &ident_pt, sizeof(uint32_t)) ||
+                 read_exact(io_fd, &ident_pt, sizeof(uint64_t)) )
+            {
+                ERROR("error read the address of the EPT identity map");
+                goto out;
+            }
+
+            xc_set_hvm_param(xc_handle, dom, HVM_PARAM_IDENT_PT, ident_pt);
+            continue;
+        }
+
         if ( j == 0 )
             break;  /* our work here is done */
 
diff --git a/tools/libxc/xc_domain_save.c b/tools/libxc/xc_domain_save.c
index 636bb78e7f..7b8a1317a4 100644
--- a/tools/libxc/xc_domain_save.c
+++ b/tools/libxc/xc_domain_save.c
@@ -1429,6 +1429,25 @@ int xc_domain_save(int xc_handle, int io_fd, uint32_t dom, uint32_t max_iters,
         }
     }
 
+    if ( hvm )
+    {
+        struct {
+            int minusthree;
+            uint32_t pad;
+            uint64_t ident_pt;
+        } chunk = { -3, 0 };
+
+        xc_get_hvm_param(xc_handle, dom, HVM_PARAM_IDENT_PT,
+                         (unsigned long *)&chunk.ident_pt);
+
+        if ( (chunk.ident_pt != 0) &&
+             write_exact(io_fd, &chunk, sizeof(chunk)) )
+        {
+            PERROR("Error when writing the ident_pt for EPT guest");
+            goto out;
+        }
+    }
+
     /* Zero terminate */
     i = 0;
     if ( write_exact(io_fd, &i, sizeof(int)) )
diff --git a/tools/libxc/xc_netbsd.c b/tools/libxc/xc_netbsd.c
index 81c86d1064..036f64879c 100644
--- a/tools/libxc/xc_netbsd.c
+++ b/tools/libxc/xc_netbsd.c
@@ -142,9 +142,13 @@ static int do_privcmd(int xc_handle, unsigned int cmd, unsigned long data)
 
 int do_xen_hypercall(int xc_handle, privcmd_hypercall_t *hypercall)
 {
-    return do_privcmd(xc_handle,
+    int error = do_privcmd(xc_handle,
                       IOCTL_PRIVCMD_HYPERCALL,
                       (unsigned long)hypercall);
+    if (error)
+       return error;
+    else
+       return (hypercall->retval);
 }
 
 #define EVTCHN_DEV_NAME  "/dev/xenevt"