11 files changed, 90 insertions, 50 deletions

diff --git a/arch/i386/kernel/cpu/common.c b/arch/i386/kernel/cpu/common.c
index 2203a9d20212..4553ffd94b1f 100644
--- a/arch/i386/kernel/cpu/common.c
+++ b/arch/i386/kernel/cpu/common.c
@@ -435,6 +435,11 @@ void __devinit identify_cpu(struct cpuinfo_x86 *c)
         if (c == &boot_cpu_data)
                 sysenter_setup();
         enable_sep_cpu();
+
+        if (c == &boot_cpu_data)
+                mtrr_bp_init();
+        else
+                mtrr_ap_init();
 }
 
 #ifdef CONFIG_X86_HT
diff --git a/arch/i386/kernel/cpu/intel.c b/arch/i386/kernel/cpu/intel.c
index 96a75d045835..a2c33c1a46c5 100644
--- a/arch/i386/kernel/cpu/intel.c
+++ b/arch/i386/kernel/cpu/intel.c
@@ -25,7 +25,7 @@ extern int trap_init_f00f_bug(void);
 /*
  * Alignment at which movsl is preferred for bulk memory copies.
  */
-struct movsl_mask movsl_mask;
+struct movsl_mask movsl_mask __read_mostly;
 #endif
 
 void __devinit early_intel_workaround(struct cpuinfo_x86 *c)
diff --git a/arch/i386/kernel/cpu/mtrr/generic.c b/arch/i386/kernel/cpu/mtrr/generic.c
index 64d91f73a0a4..169ac8e0db68 100644
--- a/arch/i386/kernel/cpu/mtrr/generic.c
+++ b/arch/i386/kernel/cpu/mtrr/generic.c
@@ -67,13 +67,6 @@ void __init get_mtrr_state(void)
         mtrr_state.enabled = (lo & 0xc00) >> 10;
 }
 
-/*  Free resources associated with a struct mtrr_state  */
-void __init finalize_mtrr_state(void)
-{
-        kfree(mtrr_state.var_ranges);
-        mtrr_state.var_ranges = NULL;
-}
-
 /*  Some BIOS's are fucked and don't set all MTRRs the same!  */
 void __init mtrr_state_warn(void)
 {
@@ -334,6 +327,9 @@ static void generic_set_mtrr(unsigned int reg, unsigned long base,
 */
 {
         unsigned long flags;
+        struct mtrr_var_range *vr;
+
+        vr = &mtrr_state.var_ranges[reg];
 
         local_irq_save(flags);
         prepare_set();
@@ -342,11 +338,15 @@ static void generic_set_mtrr(unsigned int reg, unsigned long base,
                 /* The invalid bit is kept in the mask, so we simply clear the
                    relevant mask register to disable a range. */
                 mtrr_wrmsr(MTRRphysMask_MSR(reg), 0, 0);
+                memset(vr, 0, sizeof(struct mtrr_var_range));
         } else {
-                mtrr_wrmsr(MTRRphysBase_MSR(reg), base << PAGE_SHIFT | type,
-                      (base & size_and_mask) >> (32 - PAGE_SHIFT));
-                mtrr_wrmsr(MTRRphysMask_MSR(reg), -size << PAGE_SHIFT | 0x800,
-                      (-size & size_and_mask) >> (32 - PAGE_SHIFT));
+                vr->base_lo = base << PAGE_SHIFT | type;
+                vr->base_hi = (base & size_and_mask) >> (32 - PAGE_SHIFT);
+                vr->mask_lo = -size << PAGE_SHIFT | 0x800;
+                vr->mask_hi = (-size & size_and_mask) >> (32 - PAGE_SHIFT);
+
+                mtrr_wrmsr(MTRRphysBase_MSR(reg), vr->base_lo, vr->base_hi);
+                mtrr_wrmsr(MTRRphysMask_MSR(reg), vr->mask_lo, vr->mask_hi);
         }
 
         post_set();
diff --git a/arch/i386/kernel/cpu/mtrr/main.c b/arch/i386/kernel/cpu/mtrr/main.c
index d66b09e0c820..764cac64e211 100644
--- a/arch/i386/kernel/cpu/mtrr/main.c
+++ b/arch/i386/kernel/cpu/mtrr/main.c
@@ -332,6 +332,8 @@ int mtrr_add_page(unsigned long base, unsigned long size,
 
         error = -EINVAL;
 
+        /* No CPU hotplug when we change MTRR entries */
+        lock_cpu_hotplug();
         /*  Search for existing MTRR  */
         down(&main_lock);
         for (i = 0; i < num_var_ranges; ++i) {
@@ -372,6 +374,7 @@ int mtrr_add_page(unsigned long base, unsigned long size,
         error = i;
  out:
         up(&main_lock);
+        unlock_cpu_hotplug();
         return error;
 }
 
@@ -461,6 +464,8 @@ int mtrr_del_page(int reg, unsigned long base, unsigned long size)
                 return -ENXIO;
 
         max = num_var_ranges;
+        /* No CPU hotplug when we change MTRR entries */
+        lock_cpu_hotplug();
         down(&main_lock);
         if (reg < 0) {
                 /*  Search for existing MTRR  */
@@ -501,6 +506,7 @@ int mtrr_del_page(int reg, unsigned long base, unsigned long size)
         error = reg;
  out:
         up(&main_lock);
+        unlock_cpu_hotplug();
         return error;
 }
 /**
@@ -544,21 +550,9 @@ static void __init init_ifs(void)
         centaur_init_mtrr();
 }
 
-static void __init init_other_cpus(void)
-{
-        if (use_intel())
-                get_mtrr_state();
-
-        /* bring up the other processors */
-        set_mtrr(~0U,0,0,0);
-
-        if (use_intel()) {
-                finalize_mtrr_state();
-                mtrr_state_warn();
-        }
-}
-
-
+/* The suspend/resume methods are only for CPU without MTRR. CPU using generic
+ * MTRR driver doesn't require this
+ */
 struct mtrr_value {
         mtrr_type       ltype;
         unsigned long   lbase;
@@ -611,13 +605,13 @@ static struct sysdev_driver mtrr_sysdev_driver = {
 
 
 /**
- * mtrr_init - initialize mtrrs on the boot CPU
+ * mtrr_bp_init - initialize mtrrs on the boot CPU
  *
  * This needs to be called early; before any of the other CPUs are 
  * initialized (i.e. before smp_init()).
  * 
  */
-static int __init mtrr_init(void)
+void __init mtrr_bp_init(void)
 {
         init_ifs();
 
@@ -674,12 +668,48 @@ static int __init mtrr_init(void)
         if (mtrr_if) {
                 set_num_var_ranges();
                 init_table();
-                init_other_cpus();
-
-                return sysdev_driver_register(&cpu_sysdev_class,
-                                              &mtrr_sysdev_driver);
+                if (use_intel())
+                        get_mtrr_state();
         }
-        return -ENXIO;
 }
 
-subsys_initcall(mtrr_init);
+void mtrr_ap_init(void)
+{
+        unsigned long flags;
+
+        if (!mtrr_if || !use_intel())
+                return;
+        /*
+         * Ideally we should hold main_lock here to avoid mtrr entries changed,
+         * but this routine will be called in cpu boot time, holding the lock
+         * breaks it. This routine is called in two cases: 1.very earily time
+         * of software resume, when there absolutely isn't mtrr entry changes;
+         * 2.cpu hotadd time. We let mtrr_add/del_page hold cpuhotplug lock to
+         * prevent mtrr entry changes
+         */
+        local_irq_save(flags);
+
+        mtrr_if->set_all();
+
+        local_irq_restore(flags);
+}
+
+static int __init mtrr_init_finialize(void)
+{
+        if (!mtrr_if)
+                return 0;
+        if (use_intel())
+                mtrr_state_warn();
+        else {
+                /* The CPUs haven't MTRR and seemes not support SMP. They have
+                 * specific drivers, we use a tricky method to support
+                 * suspend/resume for them.
+                 * TBD: is there any system with such CPU which supports
+                 * suspend/resume?  if no, we should remove the code.
+                 */
+                sysdev_driver_register(&cpu_sysdev_class,
+                        &mtrr_sysdev_driver);
+        }
+        return 0;
+}
+subsys_initcall(mtrr_init_finialize);
diff --git a/arch/i386/kernel/cpu/mtrr/mtrr.h b/arch/i386/kernel/cpu/mtrr/mtrr.h
index de1351245599..99c9f2682041 100644
--- a/arch/i386/kernel/cpu/mtrr/mtrr.h
+++ b/arch/i386/kernel/cpu/mtrr/mtrr.h
@@ -91,7 +91,6 @@ extern struct mtrr_ops * mtrr_if;
 
 extern unsigned int num_var_ranges;
 
-void finalize_mtrr_state(void);
 void mtrr_state_warn(void);
 char *mtrr_attrib_to_str(int x);
 void mtrr_wrmsr(unsigned, unsigned, unsigned);
diff --git a/arch/i386/kernel/smpboot.c b/arch/i386/kernel/smpboot.c
index d66bf489a2e9..8ac8e9fd5614 100644
--- a/arch/i386/kernel/smpboot.c
+++ b/arch/i386/kernel/smpboot.c
@@ -68,21 +68,21 @@ EXPORT_SYMBOL(smp_num_siblings);
 #endif
 
 /* Package ID of each logical CPU */
-int phys_proc_id[NR_CPUS] = {[0 ... NR_CPUS-1] = BAD_APICID};
+int phys_proc_id[NR_CPUS] __read_mostly = {[0 ... NR_CPUS-1] = BAD_APICID};
 EXPORT_SYMBOL(phys_proc_id);
 
 /* Core ID of each logical CPU */
-int cpu_core_id[NR_CPUS] = {[0 ... NR_CPUS-1] = BAD_APICID};
+int cpu_core_id[NR_CPUS] __read_mostly = {[0 ... NR_CPUS-1] = BAD_APICID};
 EXPORT_SYMBOL(cpu_core_id);
 
-cpumask_t cpu_sibling_map[NR_CPUS];
+cpumask_t cpu_sibling_map[NR_CPUS] __read_mostly;
 EXPORT_SYMBOL(cpu_sibling_map);
 
-cpumask_t cpu_core_map[NR_CPUS];
+cpumask_t cpu_core_map[NR_CPUS] __read_mostly;
 EXPORT_SYMBOL(cpu_core_map);
 
 /* bitmap of online cpus */
-cpumask_t cpu_online_map;
+cpumask_t cpu_online_map __read_mostly;
 EXPORT_SYMBOL(cpu_online_map);
 
 cpumask_t cpu_callin_map;
@@ -100,7 +100,7 @@ static int __devinitdata tsc_sync_disabled;
 struct cpuinfo_x86 cpu_data[NR_CPUS] __cacheline_aligned;
 EXPORT_SYMBOL(cpu_data);
 
-u8 x86_cpu_to_apicid[NR_CPUS] =
+u8 x86_cpu_to_apicid[NR_CPUS] __read_mostly =
                         { [0 ... NR_CPUS-1] = 0xff };
 EXPORT_SYMBOL(x86_cpu_to_apicid);
 
@@ -550,10 +550,10 @@ extern struct {
 #ifdef CONFIG_NUMA
 
 /* which logical CPUs are on which nodes */
-cpumask_t node_2_cpu_mask[MAX_NUMNODES] =
+cpumask_t node_2_cpu_mask[MAX_NUMNODES] __read_mostly =
                                 { [0 ... MAX_NUMNODES-1] = CPU_MASK_NONE };
 /* which node each logical CPU is on */
-int cpu_2_node[NR_CPUS] = { [0 ... NR_CPUS-1] = 0 };
+int cpu_2_node[NR_CPUS] __read_mostly = { [0 ... NR_CPUS-1] = 0 };
 EXPORT_SYMBOL(cpu_2_node);
 
 /* set up a mapping between cpu and node. */
@@ -581,7 +581,7 @@ static inline void unmap_cpu_to_node(int cpu)
 
 #endif /* CONFIG_NUMA */
 
-u8 cpu_2_logical_apicid[NR_CPUS] = { [0 ... NR_CPUS-1] = BAD_APICID };
+u8 cpu_2_logical_apicid[NR_CPUS] __read_mostly = { [0 ... NR_CPUS-1] = BAD_APICID };
 
 static void map_cpu_to_logical_apicid(void)
 {
diff --git a/arch/i386/kernel/time.c b/arch/i386/kernel/time.c
index 2854c357377f..0ee9dee8af06 100644
--- a/arch/i386/kernel/time.c
+++ b/arch/i386/kernel/time.c
@@ -91,7 +91,7 @@ EXPORT_SYMBOL(rtc_lock);
 DEFINE_SPINLOCK(i8253_lock);
 EXPORT_SYMBOL(i8253_lock);
 
-struct timer_opts *cur_timer = &timer_none;
+struct timer_opts *cur_timer __read_mostly = &timer_none;
 
 /*
  * This is a special lock that is owned by the CPU and holds the index
diff --git a/arch/i386/kernel/timers/timer_hpet.c b/arch/i386/kernel/timers/timer_hpet.c
index d766e0963ac1..ef8dac5dd33b 100644
--- a/arch/i386/kernel/timers/timer_hpet.c
+++ b/arch/i386/kernel/timers/timer_hpet.c
@@ -18,7 +18,7 @@
 #include "mach_timer.h"
 #include <asm/hpet.h>
 
-static unsigned long hpet_usec_quotient;        /* convert hpet clks to usec */
+static unsigned long __read_mostly hpet_usec_quotient;  /* convert hpet clks to usec */
 static unsigned long tsc_hpet_quotient;         /* convert tsc to hpet clks */
 static unsigned long hpet_last;         /* hpet counter value at last tick*/
 static unsigned long last_tsc_low;      /* lsb 32 bits of Time Stamp Counter */
@@ -180,7 +180,7 @@ static int __init init_hpet(char* override)
 /************************************************************/
 
 /* tsc timer_opts struct */
-static struct timer_opts timer_hpet = {
+static struct timer_opts timer_hpet __read_mostly = {
         .name =                 "hpet",
         .mark_offset =          mark_offset_hpet,
         .get_offset =           get_offset_hpet,
diff --git a/arch/i386/kernel/vmlinux.lds.S b/arch/i386/kernel/vmlinux.lds.S
index 7e01a528a83a..761972f8cb6c 100644
--- a/arch/i386/kernel/vmlinux.lds.S
+++ b/arch/i386/kernel/vmlinux.lds.S
@@ -57,6 +57,9 @@ SECTIONS
         *(.data.cacheline_aligned)
   }
 
+  /* rarely changed data like cpu maps */
+  . = ALIGN(32);
+  .data.read_mostly : AT(ADDR(.data.read_mostly) - LOAD_OFFSET) { *(.data.read_mostly) }
   _edata = .;                   /* End of data section */
 
   . = ALIGN(THREAD_SIZE);       /* init_task */
diff --git a/arch/i386/mm/ioremap.c b/arch/i386/mm/ioremap.c
index 6b25afc933b6..f379b8d67558 100644
--- a/arch/i386/mm/ioremap.c
+++ b/arch/i386/mm/ioremap.c
@@ -228,7 +228,8 @@ EXPORT_SYMBOL(ioremap_nocache);
 void iounmap(volatile void __iomem *addr)
 {
         struct vm_struct *p;
-        if ((void __force *) addr <= high_memory) 
+
+        if ((void __force *)addr <= high_memory)
                 return;
 
         /*
@@ -241,9 +242,10 @@ void iounmap(volatile void __iomem *addr)
                 return;
 
         write_lock(&vmlist_lock);
-        p = __remove_vm_area((void *) (PAGE_MASK & (unsigned long __force) addr));
+        p = __remove_vm_area((void *)(PAGE_MASK & (unsigned long __force)addr));
         if (!p) { 
                 printk(KERN_WARNING "iounmap: bad address %p\n", addr);
+                dump_stack();
                 goto out_unlock;
         }
 
diff --git a/arch/i386/power/cpu.c b/arch/i386/power/cpu.c
index 0e6b45b61251..c547c1af6fa1 100644
--- a/arch/i386/power/cpu.c
+++ b/arch/i386/power/cpu.c
@@ -137,6 +137,7 @@ void __restore_processor_state(struct saved_context *ctxt)
 
         fix_processor_context();
         do_fpu_end();
+        mtrr_ap_init();
 }
 
 void restore_processor_state(void)