[PATCH] Optimize PDA accesses slightly

[safe/jmp/linux-2.6] / arch / x86_64 / kernel / setup64.c

diff --git a/arch/x86_64/kernel/setup64.c b/arch/x86_64/kernel/setup64.c
index 66325a7..4913617 100644 (file)
--- a/arch/x86_64/kernel/setup64.c
+++ b/arch/x86_64/kernel/setup64.c
@@ -3,9 +3,7 @@
  * Copyright (C) 1995  Linus Torvalds
  * Copyright 2001, 2002, 2003 SuSE Labs / Andi Kleen.
  * See setup.c for older changelog.
- * $Id: setup64.c,v 1.12 2002/03/21 10:09:17 ak Exp $
  */ 
-#include <linux/config.h>
 #include <linux/init.h>
 #include <linux/kernel.h>
 #include <linux/sched.h>
@@ -26,18 +24,21 @@
 #include <asm/proto.h>
 #include <asm/sections.h>
 
-char x86_boot_params[BOOT_PARAM_SIZE] __initdata = {0,};
+char x86_boot_params[BOOT_PARAM_SIZE] __initdata;
 
 cpumask_t cpu_initialized __cpuinitdata = CPU_MASK_NONE;
 
-struct x8664_pda _cpu_pda[NR_CPUS] __cacheline_aligned;
+struct x8664_pda *_cpu_pda[NR_CPUS] __read_mostly;
+EXPORT_SYMBOL(_cpu_pda);
+struct x8664_pda boot_cpu_pda[NR_CPUS] __cacheline_aligned;
 
-struct desc_ptr idt_descr = { 256 * 16, (unsigned long) idt_table }; 
+struct desc_ptr idt_descr = { 256 * 16 - 1, (unsigned long) idt_table };
 
 char boot_cpu_stack[IRQSTACKSIZE] __attribute__((section(".bss.page_aligned")));
 
 unsigned long __supported_pte_mask __read_mostly = ~0UL;
-static int do_not_nx __initdata = 0;
+EXPORT_SYMBOL(__supported_pte_mask);
+static int do_not_nx __cpuinitdata = 0;
 
 /* noexec=on|off
 Control non executable mappings for 64bit processes.
@@ -45,8 +46,10 @@ Control non executable mappings for 64bit processes.
 on     Enable(default)
 off    Disable
 */ 
-int __init nonx_setup(char *str)
+static int __init nonx_setup(char *str)
 {
+       if (!str)
+               return -EINVAL;
        if (!strncmp(str, "on", 2)) {
                 __supported_pte_mask |= _PAGE_NX; 
                do_not_nx = 0; 
@@ -56,9 +59,9 @@ int __init nonx_setup(char *str)
         }
        return 0;
 } 
-__setup("noexec=", nonx_setup);        /* parsed early actually */
+early_param("noexec", nonx_setup);
 
-int force_personality32 = READ_IMPLIES_EXEC;
+int force_personality32 = 0; 
 
 /* noexec32=on|off
 Control non executable heap for 32bit processes.
@@ -73,7 +76,7 @@ static int __init nonx32_setup(char *str)
                force_personality32 &= ~READ_IMPLIES_EXEC;
        else if (!strcmp(str, "off"))
                force_personality32 |= READ_IMPLIES_EXEC;
-       return 0;
+       return 1;
 }
 __setup("noexec32=", nonx32_setup);
 
@@ -92,12 +95,9 @@ void __init setup_per_cpu_areas(void)
 #endif
 
        /* Copy section for each CPU (we discard the original) */
-       size = ALIGN(__per_cpu_end - __per_cpu_start, SMP_CACHE_BYTES);
-#ifdef CONFIG_MODULES
-       if (size < PERCPU_ENOUGH_ROOM)
-               size = PERCPU_ENOUGH_ROOM;
-#endif
+       size = PERCPU_ENOUGH_ROOM;
 
+       printk(KERN_INFO "PERCPU: Allocating %lu bytes of per cpu data\n", size);
        for_each_cpu_mask (i, cpu_possible_map) {
                char *ptr;
 
@@ -121,7 +121,10 @@ void pda_init(int cpu)
 
        /* Setup up data that may be needed in __get_free_pages early */
        asm volatile("movl %0,%%fs ; movl %0,%%gs" :: "r" (0)); 
+       /* Memory clobbers used to order PDA accessed */
+       mb();
        wrmsrl(MSR_GS_BASE, pda);
+       mb();
 
        pda->cpunumber = cpu; 
        pda->irqcount = -1;
@@ -145,7 +148,7 @@ void pda_init(int cpu)
        pda->irqstackptr += IRQSTACKSIZE-64;
 } 
 
-char boot_exception_stacks[(N_EXCEPTION_STACKS - 2) * EXCEPTION_STKSZ + DEBUG_STKSZ]
+char boot_exception_stacks[(N_EXCEPTION_STACKS - 1) * EXCEPTION_STKSZ + DEBUG_STKSZ]
 __attribute__((section(".bss.page_aligned")));
 
 /* May not be marked __init: used by software suspend */
@@ -188,6 +191,7 @@ void __cpuinit cpu_init (void)
 {
        int cpu = stack_smp_processor_id();
        struct tss_struct *t = &per_cpu(init_tss, cpu);
+       struct orig_ist *orig_ist = &per_cpu(orig_ist, cpu);
        unsigned long v; 
        char *estacks = NULL; 
        struct task_struct *me;
@@ -247,7 +251,7 @@ void __cpuinit cpu_init (void)
                switch (v + 1) {
 #if DEBUG_STKSZ > EXCEPTION_STKSZ
                case DEBUG_STACK:
-                       cpu_pda[cpu].debugstack = (unsigned long)estacks;
+                       cpu_pda(cpu)->debugstack = (unsigned long)estacks;
                        estacks += DEBUG_STKSZ;
                        break;
 #endif
@@ -255,7 +259,7 @@ void __cpuinit cpu_init (void)
                        estacks += EXCEPTION_STKSZ;
                        break;
                }
-               t->ist[v] = (unsigned long)estacks;
+               orig_ist->ist[v] = t->ist[v] = (unsigned long)estacks;
        }
 
        t->io_bitmap_base = offsetof(struct tss_struct, io_bitmap);
@@ -280,12 +284,12 @@ void __cpuinit cpu_init (void)
         * Clear all 6 debug registers:
         */
 
-       set_debug(0UL, 0);
-       set_debug(0UL, 1);
-       set_debug(0UL, 2);
-       set_debug(0UL, 3);
-       set_debug(0UL, 6);
-       set_debug(0UL, 7);
+       set_debugreg(0UL, 0);
+       set_debugreg(0UL, 1);
+       set_debugreg(0UL, 2);
+       set_debugreg(0UL, 3);
+       set_debugreg(0UL, 6);
+       set_debugreg(0UL, 7);
 
        fpu_init(); 
 }