Introduce system_state variable.

Use it to replace x86-specific early_boot boolean variable.

Also use it to detect suspend/resume case during cpu offline/online
to avoid unnecessarily breaking vcpu and cpupool affinities.

Signed-off-by: Keir Fraser <keir@xen.org>
Acked-by: Juergen Gross <juergen.gross@ts.fujitsu.com>
xen-unstable changeset:   25079:d5ccb2d1dbd1
xen-unstable date:        Thu Mar 22 12:20:13 2012 +0000

Backport requested by Juergen to fix crash on poweroff when using
cpupools.

10 files changed, 33 insertions, 9 deletions

diff --git a/xen/arch/x86/acpi/power.c b/xen/arch/x86/acpi/power.c
index 935c9a1101..53b5e17c44 100644
--- a/xen/arch/x86/acpi/power.c
+++ b/xen/arch/x86/acpi/power.c
@@ -135,6 +135,9 @@ static int enter_state(u32 state)
     if ( !spin_trylock(&pm_lock) )
         return -EBUSY;
 
+    BUG_ON(system_state != SYS_STATE_active);
+    system_state = SYS_STATE_suspend;
+
     printk(XENLOG_INFO "Preparing system for ACPI S%d state.\n", state);
 
     freeze_domains();
@@ -142,7 +145,10 @@ static int enter_state(u32 state)
     acpi_dmar_reinstate();
 
     if ( (error = disable_nonboot_cpus()) )
+    {
+        system_state = SYS_STATE_resume;
         goto enable_cpu;
+    }
 
     cpufreq_del_cpu(0);
 
@@ -159,6 +165,7 @@ static int enter_state(u32 state)
     if ( (error = device_power_down()) )
     {
         printk(XENLOG_ERR "Some devices failed to power down.");
+        system_state = SYS_STATE_resume;
         goto done;
     }
 
@@ -179,6 +186,8 @@ static int enter_state(u32 state)
         break;
     }
 
+    system_state = SYS_STATE_resume;
+
     /* Restore CR4 and EFER from cached values. */
     cr4 = read_cr4();
     write_cr4(cr4 & ~X86_CR4_MCE);
@@ -212,6 +221,7 @@ static int enter_state(u32 state)
     mtrr_aps_sync_end();
     acpi_dmar_zap();
     thaw_domains();
+    system_state = SYS_STATE_active;
     spin_unlock(&pm_lock);
     return error;
 }
diff --git a/xen/arch/x86/mm.c b/xen/arch/x86/mm.c
index fd37a6aa7e..cbdf065909 100644
--- a/xen/arch/x86/mm.c
+++ b/xen/arch/x86/mm.c
@@ -5123,7 +5123,7 @@ int ptwr_do_page_fault(struct vcpu *v, unsigned long addr,
 
 void free_xen_pagetable(void *v)
 {
-    if ( early_boot )
+    if ( system_state == SYS_STATE_early_boot )
         return;
 
     if ( is_xen_heap_page(virt_to_page(v)) )
diff --git a/xen/arch/x86/setup.c b/xen/arch/x86/setup.c
index 1777ffec51..d88884e4cb 100644
--- a/xen/arch/x86/setup.c
+++ b/xen/arch/x86/setup.c
@@ -87,8 +87,6 @@ boolean_param("noapic", skip_ioapic_setup);
 s8 __read_mostly xen_cpuidle = -1;
 boolean_param("cpuidle", xen_cpuidle);
 
-bool_t __read_mostly early_boot = 1;
-
 cpumask_t __read_mostly cpu_present_map;
 
 unsigned long __read_mostly xen_phys_start;
@@ -275,7 +273,7 @@ static void *__init bootstrap_map(const module_t *mod)
     void *ret;
 
 #ifdef __x86_64__
-    if ( !early_boot )
+    if ( system_state != SYS_STATE_early_boot )
         return mod ? mfn_to_virt(mod->mod_start) : NULL;
 #endif
 
@@ -1142,7 +1140,7 @@ void __init __start_xen(unsigned long mbi_p)
 #endif
 
     end_boot_allocator();
-    early_boot = 0;
+    system_state = SYS_STATE_boot;
 
 #if defined(CONFIG_X86_64)
     vesa_init();
@@ -1341,6 +1339,8 @@ void __init __start_xen(unsigned long mbi_p)
     /* Hide UART from DOM0 if we're using it */
     serial_endboot();
 
+    system_state = SYS_STATE_active;
+
     domain_unpause_by_systemcontroller(dom0);
 
     reset_stack_and_jump(init_done);
diff --git a/xen/arch/x86/x86_32/mm.c b/xen/arch/x86/x86_32/mm.c
index 73a2a9bcbc..16ae743462 100644
--- a/xen/arch/x86/x86_32/mm.c
+++ b/xen/arch/x86/x86_32/mm.c
@@ -45,7 +45,7 @@ void *alloc_xen_pagetable(void)
 {
     unsigned long mfn;
 
-    if ( !early_boot )
+    if ( system_state != SYS_STATE_early_boot )
     {
         void *v = alloc_xenheap_page();
 
diff --git a/xen/arch/x86/x86_64/mm.c b/xen/arch/x86/x86_64/mm.c
index 46c55080bc..72cbca4cfd 100644
--- a/xen/arch/x86/x86_64/mm.c
+++ b/xen/arch/x86/x86_64/mm.c
@@ -81,7 +81,7 @@ void *alloc_xen_pagetable(void)
 {
     unsigned long mfn;
 
-    if ( !early_boot )
+    if ( system_state != SYS_STATE_early_boot )
     {
         struct page_info *pg = alloc_domheap_page(NULL, 0);
 
diff --git a/xen/common/cpupool.c b/xen/common/cpupool.c
index c705475538..cafb4be792 100644
--- a/xen/common/cpupool.c
+++ b/xen/common/cpupool.c
@@ -616,6 +616,10 @@ static int cpu_callback(
     unsigned int cpu = (unsigned long)hcpu;
     int rc = 0;
 
+    if ( (system_state == SYS_STATE_suspend) ||
+         (system_state == SYS_STATE_resume) )
+        goto out;
+
     switch ( action )
     {
     case CPU_DOWN_FAILED:
@@ -629,6 +633,7 @@ static int cpu_callback(
         break;
     }
 
+out:
     return !rc ? NOTIFY_DONE : notifier_from_errno(rc);
 }
 
diff --git a/xen/common/kernel.c b/xen/common/kernel.c
index c061a0d838..3ce1dcc7a9 100644
--- a/xen/common/kernel.c
+++ b/xen/common/kernel.c
@@ -22,6 +22,8 @@
 
 #ifndef COMPAT
 
+enum system_state system_state = SYS_STATE_early_boot;
+
 int tainted;
 
 xen_commandline_t saved_cmdline;
diff --git a/xen/common/schedule.c b/xen/common/schedule.c
index 907ea09a76..1ed1bbb4ad 100644
--- a/xen/common/schedule.c
+++ b/xen/common/schedule.c
@@ -546,7 +546,7 @@ int cpu_disable_scheduler(unsigned int cpu)
     bool_t affinity_broken;
 
     c = per_cpu(cpupool, cpu);
-    if ( c == NULL )
+    if ( (c == NULL) || (system_state == SYS_STATE_suspend) )
         return ret;
 
     for_each_domain ( d )
diff --git a/xen/include/asm-x86/setup.h b/xen/include/asm-x86/setup.h
index 4d1ce7202f..58199476f8 100644
--- a/xen/include/asm-x86/setup.h
+++ b/xen/include/asm-x86/setup.h
@@ -3,7 +3,6 @@
 
 #include <xen/multiboot.h>
 
-extern bool_t early_boot;
 extern unsigned long xenheap_initial_phys_start;
 
 void init_done(void);
diff --git a/xen/include/xen/kernel.h b/xen/include/xen/kernel.h
index fd03f74c8b..8e12a52507 100644
--- a/xen/include/xen/kernel.h
+++ b/xen/include/xen/kernel.h
@@ -81,5 +81,13 @@ extern char _sinittext[], _einittext[];
     (__p >= _sinittext) && (__p <= _einittext); \
 })
 
+extern enum system_state {
+    SYS_STATE_early_boot,
+    SYS_STATE_boot,
+    SYS_STATE_active,
+    SYS_STATE_suspend,
+    SYS_STATE_resume
+} system_state;
+
 #endif /* _LINUX_KERNEL_H */