From 26f80bd6a9ab17bc8a60b6092e7c0d05c5927ce5 Mon Sep 17 00:00:00 2001 From: Brian Gerst Date: Mon, 19 Jan 2009 00:38:58 +0900 Subject: x86-64: Convert irqstacks to per-cpu Move the irqstackptr variable from the PDA to per-cpu. Make the stacks themselves per-cpu, removing some specific allocation code. Add a seperate flag (is_boot_cpu) to simplify the per-cpu boot adjustments. tj: * sprinkle some underbars around. * irq_stack_ptr is not used till traps_init(), no reason to initialize it early. On SMP, just leaving it NULL till proper initialization in setup_per_cpu_areas() works. Dropped is_boot_cpu and early irq_stack_ptr initialization. * do DECLARE/DEFINE_PER_CPU(char[IRQ_STACK_SIZE], irq_stack) instead of (char, irq_stack[IRQ_STACK_SIZE]). Signed-off-by: Brian Gerst Signed-off-by: Tejun Heo --- arch/x86/include/asm/processor.h | 3 +++ 1 file changed, 3 insertions(+) (limited to 'arch/x86/include/asm/processor.h') diff --git a/arch/x86/include/asm/processor.h b/arch/x86/include/asm/processor.h index 091cd8855f2e..f511246fa6cd 100644 --- a/arch/x86/include/asm/processor.h +++ b/arch/x86/include/asm/processor.h @@ -378,6 +378,9 @@ union thread_xstate { #ifdef CONFIG_X86_64 DECLARE_PER_CPU(struct orig_ist, orig_ist); + +DECLARE_PER_CPU(char[IRQ_STACK_SIZE], irq_stack); +DECLARE_PER_CPU(char *, irq_stack_ptr); #endif extern void print_cpu_info(struct cpuinfo_x86 *); -- cgit From 947e76cdc34c782fc947313d4331380686eebbad Mon Sep 17 00:00:00 2001 From: Brian Gerst Date: Mon, 19 Jan 2009 12:21:28 +0900 Subject: x86: move stack_canary into irq_stack Impact: x86_64 percpu area layout change, irq_stack now at the beginning Now that the PDA is empty except for the stack canary, it can be removed. The irqstack is moved to the start of the per-cpu section. If the stack protector is enabled, the canary overlaps the bottom 48 bytes of the irqstack. tj: * updated subject * dropped asm relocation of irq_stack_ptr * updated comments a bit * rebased on top of stack canary changes Signed-off-by: Brian Gerst Signed-off-by: Tejun Heo --- arch/x86/include/asm/pda.h | 3 --- arch/x86/include/asm/percpu.h | 6 ------ arch/x86/include/asm/processor.h | 23 ++++++++++++++++++++++- arch/x86/include/asm/stackprotector.h | 6 +++--- arch/x86/include/asm/system.h | 4 ++-- arch/x86/kernel/asm-offsets_64.c | 4 ---- arch/x86/kernel/cpu/common.c | 7 ++++--- arch/x86/kernel/head_64.S | 13 +++++-------- arch/x86/kernel/setup_percpu.c | 34 ++++------------------------------ arch/x86/kernel/vmlinux_64.lds.S | 8 ++++++-- 10 files changed, 46 insertions(+), 62 deletions(-) (limited to 'arch/x86/include/asm/processor.h') diff --git a/arch/x86/include/asm/pda.h b/arch/x86/include/asm/pda.h index b473e952439a..ba46416634f0 100644 --- a/arch/x86/include/asm/pda.h +++ b/arch/x86/include/asm/pda.h @@ -17,9 +17,6 @@ struct x8664_pda { unsigned long unused4; int unused5; unsigned int unused6; /* 36 was cpunumber */ - unsigned long stack_canary; /* 40 stack canary value */ - /* gcc-ABI: this canary MUST be at - offset 40!!! */ short in_bootmem; /* pda lives in bootmem */ } ____cacheline_aligned_in_smp; diff --git a/arch/x86/include/asm/percpu.h b/arch/x86/include/asm/percpu.h index 165d5272ece1..ce980db5e59d 100644 --- a/arch/x86/include/asm/percpu.h +++ b/arch/x86/include/asm/percpu.h @@ -133,12 +133,6 @@ do { \ /* We can use this directly for local CPU (faster). */ DECLARE_PER_CPU(unsigned long, this_cpu_off); -#ifdef CONFIG_X86_64 -extern void load_pda_offset(int cpu); -#else -static inline void load_pda_offset(int cpu) { } -#endif - #endif /* !__ASSEMBLY__ */ #ifdef CONFIG_SMP diff --git a/arch/x86/include/asm/processor.h b/arch/x86/include/asm/processor.h index f511246fa6cd..48676b943b92 100644 --- a/arch/x86/include/asm/processor.h +++ b/arch/x86/include/asm/processor.h @@ -379,8 +379,29 @@ union thread_xstate { #ifdef CONFIG_X86_64 DECLARE_PER_CPU(struct orig_ist, orig_ist); -DECLARE_PER_CPU(char[IRQ_STACK_SIZE], irq_stack); +union irq_stack_union { + char irq_stack[IRQ_STACK_SIZE]; + /* + * GCC hardcodes the stack canary as %gs:40. Since the + * irq_stack is the object at %gs:0, we reserve the bottom + * 48 bytes of the irq stack for the canary. + */ + struct { + char gs_base[40]; + unsigned long stack_canary; + }; +}; + +DECLARE_PER_CPU(union irq_stack_union, irq_stack_union); DECLARE_PER_CPU(char *, irq_stack_ptr); + +static inline void load_gs_base(int cpu) +{ + /* Memory clobbers used to order pda/percpu accesses */ + mb(); + wrmsrl(MSR_GS_BASE, (unsigned long)per_cpu(irq_stack_union.gs_base, cpu)); + mb(); +} #endif extern void print_cpu_info(struct cpuinfo_x86 *); diff --git a/arch/x86/include/asm/stackprotector.h b/arch/x86/include/asm/stackprotector.h index 2383e5bb475c..36a700acaf2b 100644 --- a/arch/x86/include/asm/stackprotector.h +++ b/arch/x86/include/asm/stackprotector.h @@ -2,7 +2,7 @@ #define _ASM_STACKPROTECTOR_H 1 #include -#include +#include /* * Initialize the stackprotector canary value. @@ -19,7 +19,7 @@ static __always_inline void boot_init_stack_canary(void) * Build time only check to make sure the stack_canary is at * offset 40 in the pda; this is a gcc ABI requirement */ - BUILD_BUG_ON(offsetof(struct x8664_pda, stack_canary) != 40); + BUILD_BUG_ON(offsetof(union irq_stack_union, stack_canary) != 40); /* * We both use the random pool and the current TSC as a source @@ -32,7 +32,7 @@ static __always_inline void boot_init_stack_canary(void) canary += tsc + (tsc << 32UL); current->stack_canary = canary; - write_pda(stack_canary, canary); + percpu_write(irq_stack_union.stack_canary, canary); } #endif diff --git a/arch/x86/include/asm/system.h b/arch/x86/include/asm/system.h index b77bd8bd3cc2..52eb748a68af 100644 --- a/arch/x86/include/asm/system.h +++ b/arch/x86/include/asm/system.h @@ -89,10 +89,10 @@ do { \ #ifdef CONFIG_CC_STACKPROTECTOR #define __switch_canary \ "movq %P[task_canary](%%rsi),%%r8\n\t" \ - "movq %%r8,%%gs:%P[pda_canary]\n\t" + "movq %%r8,%%gs:%P[gs_canary]\n\t" #define __switch_canary_param \ , [task_canary] "i" (offsetof(struct task_struct, stack_canary)) \ - , [pda_canary] "i" (offsetof(struct x8664_pda, stack_canary)) + , [gs_canary] "i" (offsetof(union irq_stack_union, stack_canary)) #else /* CC_STACKPROTECTOR */ #define __switch_canary #define __switch_canary_param diff --git a/arch/x86/kernel/asm-offsets_64.c b/arch/x86/kernel/asm-offsets_64.c index 64c834a39aa8..94f9c8b39d20 100644 --- a/arch/x86/kernel/asm-offsets_64.c +++ b/arch/x86/kernel/asm-offsets_64.c @@ -48,10 +48,6 @@ int main(void) #endif BLANK(); #undef ENTRY -#define ENTRY(entry) DEFINE(pda_ ## entry, offsetof(struct x8664_pda, entry)) - DEFINE(pda_size, sizeof(struct x8664_pda)); - BLANK(); -#undef ENTRY #ifdef CONFIG_PARAVIRT BLANK(); OFFSET(PARAVIRT_enabled, pv_info, paravirt_enabled); diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c index f83a4d6160f0..098934e72a16 100644 --- a/arch/x86/kernel/cpu/common.c +++ b/arch/x86/kernel/cpu/common.c @@ -881,12 +881,13 @@ __setup("clearcpuid=", setup_disablecpuid); #ifdef CONFIG_X86_64 struct desc_ptr idt_descr = { 256 * 16 - 1, (unsigned long) idt_table }; -DEFINE_PER_CPU_PAGE_ALIGNED(char[IRQ_STACK_SIZE], irq_stack); +DEFINE_PER_CPU_FIRST(union irq_stack_union, + irq_stack_union) __aligned(PAGE_SIZE); #ifdef CONFIG_SMP DEFINE_PER_CPU(char *, irq_stack_ptr); /* will be set during per cpu init */ #else DEFINE_PER_CPU(char *, irq_stack_ptr) = - per_cpu_var(irq_stack) + IRQ_STACK_SIZE - 64; + per_cpu_var(irq_stack_union.irq_stack) + IRQ_STACK_SIZE - 64; #endif DEFINE_PER_CPU(unsigned long, kernel_stack) = @@ -960,7 +961,7 @@ void __cpuinit cpu_init(void) loadsegment(fs, 0); loadsegment(gs, 0); - load_pda_offset(cpu); + load_gs_base(cpu); #ifdef CONFIG_NUMA if (cpu != 0 && percpu_read(node_number) == 0 && diff --git a/arch/x86/kernel/head_64.S b/arch/x86/kernel/head_64.S index 98ea26a2fca1..a0a2b5ca9b7d 100644 --- a/arch/x86/kernel/head_64.S +++ b/arch/x86/kernel/head_64.S @@ -242,13 +242,10 @@ ENTRY(secondary_startup_64) /* Set up %gs. * - * On SMP, %gs should point to the per-cpu area. For initial - * boot, make %gs point to the init data section. For a - * secondary CPU,initial_gs should be set to its pda address - * before the CPU runs this code. - * - * On UP, initial_gs points to PER_CPU_VAR(__pda) and doesn't - * change. + * The base of %gs always points to the bottom of the irqstack + * union. If the stack protector canary is enabled, it is + * located at %gs:40. Note that, on SMP, the boot cpu uses + * init data section till per cpu areas are set up. */ movl $MSR_GS_BASE,%ecx movq initial_gs(%rip),%rax @@ -281,7 +278,7 @@ ENTRY(secondary_startup_64) #ifdef CONFIG_SMP .quad __per_cpu_load #else - .quad PER_CPU_VAR(__pda) + .quad PER_CPU_VAR(irq_stack_union) #endif __FINITDATA diff --git a/arch/x86/kernel/setup_percpu.c b/arch/x86/kernel/setup_percpu.c index efbafbbff584..90b8e154bb53 100644 --- a/arch/x86/kernel/setup_percpu.c +++ b/arch/x86/kernel/setup_percpu.c @@ -77,30 +77,6 @@ static void __init setup_node_to_cpumask_map(void); static inline void setup_node_to_cpumask_map(void) { } #endif -/* - * Define load_pda_offset() and per-cpu __pda for x86_64. - * load_pda_offset() is responsible for loading the offset of pda into - * %gs. - * - * On SMP, pda offset also duals as percpu base address and thus it - * should be at the start of per-cpu area. To achieve this, it's - * preallocated in vmlinux_64.lds.S directly instead of using - * DEFINE_PER_CPU(). - */ -#ifdef CONFIG_X86_64 -void __cpuinit load_pda_offset(int cpu) -{ - /* Memory clobbers used to order pda/percpu accesses */ - mb(); - wrmsrl(MSR_GS_BASE, cpu_pda(cpu)); - mb(); -} -#ifndef CONFIG_SMP -DEFINE_PER_CPU(struct x8664_pda, __pda); -#endif -EXPORT_PER_CPU_SYMBOL(__pda); -#endif /* CONFIG_SMP && CONFIG_X86_64 */ - #ifdef CONFIG_X86_64 /* correctly size the local cpu masks */ @@ -207,15 +183,13 @@ void __init setup_per_cpu_areas(void) per_cpu(cpu_number, cpu) = cpu; #ifdef CONFIG_X86_64 per_cpu(irq_stack_ptr, cpu) = - (char *)per_cpu(irq_stack, cpu) + IRQ_STACK_SIZE - 64; + per_cpu(irq_stack_union.irq_stack, cpu) + IRQ_STACK_SIZE - 64; /* - * CPU0 modified pda in the init data area, reload pda - * offset for CPU0 and clear the area for others. + * Up to this point, CPU0 has been using .data.init + * area. Reload %gs offset for CPU0. */ if (cpu == 0) - load_pda_offset(0); - else - memset(cpu_pda(cpu), 0, sizeof(*cpu_pda(cpu))); + load_gs_base(cpu); #endif DBG("PERCPU: cpu %4d %p\n", cpu, ptr); diff --git a/arch/x86/kernel/vmlinux_64.lds.S b/arch/x86/kernel/vmlinux_64.lds.S index a09abb8fb97f..c9740996430a 100644 --- a/arch/x86/kernel/vmlinux_64.lds.S +++ b/arch/x86/kernel/vmlinux_64.lds.S @@ -220,8 +220,7 @@ SECTIONS * so that it can be accessed as a percpu variable. */ . = ALIGN(PAGE_SIZE); - PERCPU_VADDR_PREALLOC(0, :percpu, pda_size) - per_cpu____pda = __per_cpu_start; + PERCPU_VADDR(0, :percpu) #else PERCPU(PAGE_SIZE) #endif @@ -262,3 +261,8 @@ SECTIONS */ ASSERT((_end - _text <= KERNEL_IMAGE_SIZE), "kernel image bigger than KERNEL_IMAGE_SIZE") + +#ifdef CONFIG_SMP +ASSERT((per_cpu__irq_stack_union == 0), + "irq_stack_union is not at start of per-cpu area"); +#endif -- cgit From b2d2f4312b117a6cc647c8521e2643a88771f757 Mon Sep 17 00:00:00 2001 From: Brian Gerst Date: Tue, 27 Jan 2009 12:56:48 +0900 Subject: x86: initialize per-cpu GDT segment in per-cpu setup Impact: cleanup Rename init_gdt() to setup_percpu_segment(), and move it to setup_percpu.c. Signed-off-by: Brian Gerst Signed-off-by: Tejun Heo --- arch/x86/include/asm/processor.h | 1 - arch/x86/kernel/Makefile | 3 +-- arch/x86/kernel/setup_percpu.c | 14 ++++++++++++++ arch/x86/kernel/smpboot.c | 4 ---- arch/x86/kernel/smpcommon.c | 25 ------------------------- arch/x86/mach-voyager/voyager_smp.c | 2 -- arch/x86/xen/smp.c | 1 - 7 files changed, 15 insertions(+), 35 deletions(-) delete mode 100644 arch/x86/kernel/smpcommon.c (limited to 'arch/x86/include/asm/processor.h') diff --git a/arch/x86/include/asm/processor.h b/arch/x86/include/asm/processor.h index 48676b943b92..32c30b02b51f 100644 --- a/arch/x86/include/asm/processor.h +++ b/arch/x86/include/asm/processor.h @@ -778,7 +778,6 @@ extern struct desc_ptr early_gdt_descr; extern void cpu_set_gdt(int); extern void switch_to_new_gdt(void); extern void cpu_init(void); -extern void init_gdt(int cpu); static inline unsigned long get_debugctlmsr(void) { diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile index 73de055c29c2..37fa30bada17 100644 --- a/arch/x86/kernel/Makefile +++ b/arch/x86/kernel/Makefile @@ -60,8 +60,7 @@ obj-$(CONFIG_APM) += apm.o obj-$(CONFIG_X86_SMP) += smp.o obj-$(CONFIG_X86_SMP) += smpboot.o tsc_sync.o ipi.o obj-$(CONFIG_SMP) += setup_percpu.o -obj-$(CONFIG_X86_32_SMP) += smpcommon.o -obj-$(CONFIG_X86_64_SMP) += tsc_sync.o smpcommon.o +obj-$(CONFIG_X86_64_SMP) += tsc_sync.o obj-$(CONFIG_X86_TRAMPOLINE) += trampoline_$(BITS).o obj-$(CONFIG_X86_MPPARSE) += mpparse.o obj-$(CONFIG_X86_LOCAL_APIC) += apic.o nmi.o diff --git a/arch/x86/kernel/setup_percpu.c b/arch/x86/kernel/setup_percpu.c index 599dc1cc1da8..bcca3a7b3748 100644 --- a/arch/x86/kernel/setup_percpu.c +++ b/arch/x86/kernel/setup_percpu.c @@ -40,6 +40,19 @@ unsigned long __per_cpu_offset[NR_CPUS] __read_mostly = { }; EXPORT_SYMBOL(__per_cpu_offset); +static inline void setup_percpu_segment(int cpu) +{ +#ifdef CONFIG_X86_32 + struct desc_struct gdt; + + pack_descriptor(&gdt, per_cpu_offset(cpu), 0xFFFFF, + 0x2 | DESCTYPE_S, 0x8); + gdt.s = 1; + write_gdt_entry(get_cpu_gdt_table(cpu), + GDT_ENTRY_PERCPU, &gdt, DESCTYPE_S); +#endif +} + /* * Great future plan: * Declare PDA itself and support (irqstack,tss,pgd) as per cpu data. @@ -81,6 +94,7 @@ void __init setup_per_cpu_areas(void) per_cpu_offset(cpu) = ptr - __per_cpu_start; per_cpu(this_cpu_off, cpu) = per_cpu_offset(cpu); per_cpu(cpu_number, cpu) = cpu; + setup_percpu_segment(cpu); /* * Copy data used in early init routines from the initial arrays to the * per cpu data areas. These arrays then become expendable and the diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c index def770b57b5a..f9dbcff43546 100644 --- a/arch/x86/kernel/smpboot.c +++ b/arch/x86/kernel/smpboot.c @@ -793,7 +793,6 @@ static int __cpuinit do_boot_cpu(int apicid, int cpu) do_rest: per_cpu(current_task, cpu) = c_idle.idle; #ifdef CONFIG_X86_32 - init_gdt(cpu); /* Stack for startup_32 can be just as for start_secondary onwards */ irq_ctx_init(cpu); #else @@ -1186,9 +1185,6 @@ out: void __init native_smp_prepare_boot_cpu(void) { int me = smp_processor_id(); -#ifdef CONFIG_X86_32 - init_gdt(me); -#endif switch_to_new_gdt(); /* already set me in cpu_online_mask in boot_cpu_init() */ cpumask_set_cpu(me, cpu_callout_mask); diff --git a/arch/x86/kernel/smpcommon.c b/arch/x86/kernel/smpcommon.c deleted file mode 100644 index 5ec29a1a8465..000000000000 --- a/arch/x86/kernel/smpcommon.c +++ /dev/null @@ -1,25 +0,0 @@ -/* - * SMP stuff which is common to all sub-architectures. - */ -#include -#include -#include - -#ifdef CONFIG_X86_32 -/* - * Initialize the CPU's GDT. This is either the boot CPU doing itself - * (still using the master per-cpu area), or a CPU doing it for a - * secondary which will soon come up. - */ -__cpuinit void init_gdt(int cpu) -{ - struct desc_struct gdt; - - pack_descriptor(&gdt, __per_cpu_offset[cpu], 0xFFFFF, - 0x2 | DESCTYPE_S, 0x8); - gdt.s = 1; - - write_gdt_entry(get_cpu_gdt_table(cpu), - GDT_ENTRY_PERCPU, &gdt, DESCTYPE_S); -} -#endif diff --git a/arch/x86/mach-voyager/voyager_smp.c b/arch/x86/mach-voyager/voyager_smp.c index dd82f2052f34..331cd6d56483 100644 --- a/arch/x86/mach-voyager/voyager_smp.c +++ b/arch/x86/mach-voyager/voyager_smp.c @@ -530,7 +530,6 @@ static void __init do_boot_cpu(__u8 cpu) /* init_tasks (in sched.c) is indexed logically */ stack_start.sp = (void *)idle->thread.sp; - init_gdt(cpu); per_cpu(current_task, cpu) = idle; early_gdt_descr.address = (unsigned long)get_cpu_gdt_table(cpu); irq_ctx_init(cpu); @@ -1747,7 +1746,6 @@ static void __init voyager_smp_prepare_cpus(unsigned int max_cpus) static void __cpuinit voyager_smp_prepare_boot_cpu(void) { - init_gdt(smp_processor_id()); switch_to_new_gdt(); cpu_set(smp_processor_id(), cpu_online_map); diff --git a/arch/x86/xen/smp.c b/arch/x86/xen/smp.c index 72c2eb9b64cd..7735e3dd359c 100644 --- a/arch/x86/xen/smp.c +++ b/arch/x86/xen/smp.c @@ -281,7 +281,6 @@ static int __cpuinit xen_cpu_up(unsigned int cpu) per_cpu(current_task, cpu) = idle; #ifdef CONFIG_X86_32 - init_gdt(cpu); irq_ctx_init(cpu); #else clear_tsk_thread_flag(idle, TIF_FORK); -- cgit From 1825b8edc2034c012ae48f797d74efd1bd9d4f72 Mon Sep 17 00:00:00 2001 From: Brian Gerst Date: Tue, 27 Jan 2009 12:56:48 +0900 Subject: x86: remove extra barriers from load_gs_base() Impact: optimization mb() generates an mfence instruction, which is not needed here. Only a compiler barrier is needed, and that is handled by the memory clobber in the wrmsrl function. Signed-off-by: Brian Gerst Signed-off-by: Tejun Heo --- arch/x86/include/asm/processor.h | 3 --- 1 file changed, 3 deletions(-) (limited to 'arch/x86/include/asm/processor.h') diff --git a/arch/x86/include/asm/processor.h b/arch/x86/include/asm/processor.h index 32c30b02b51f..794234eba317 100644 --- a/arch/x86/include/asm/processor.h +++ b/arch/x86/include/asm/processor.h @@ -397,10 +397,7 @@ DECLARE_PER_CPU(char *, irq_stack_ptr); static inline void load_gs_base(int cpu) { - /* Memory clobbers used to order pda/percpu accesses */ - mb(); wrmsrl(MSR_GS_BASE, (unsigned long)per_cpu(irq_stack_union.gs_base, cpu)); - mb(); } #endif -- cgit From 2697fbd5faf19c84c17441b1752bdcbdcfd1248c Mon Sep 17 00:00:00 2001 From: Brian Gerst Date: Tue, 27 Jan 2009 12:56:48 +0900 Subject: x86: load new GDT after setting up boot cpu per-cpu area Impact: sync 32 and 64-bit code Merge load_gs_base() into switch_to_new_gdt(). Load the GDT and per-cpu state for the boot cpu when its new area is set up. Signed-off-by: Brian Gerst Signed-off-by: Tejun Heo --- arch/x86/include/asm/processor.h | 5 ----- arch/x86/kernel/cpu/common.c | 15 +++++++++------ arch/x86/kernel/setup_percpu.c | 6 +++--- 3 files changed, 12 insertions(+), 14 deletions(-) (limited to 'arch/x86/include/asm/processor.h') diff --git a/arch/x86/include/asm/processor.h b/arch/x86/include/asm/processor.h index 794234eba317..befa20b4a68c 100644 --- a/arch/x86/include/asm/processor.h +++ b/arch/x86/include/asm/processor.h @@ -394,11 +394,6 @@ union irq_stack_union { DECLARE_PER_CPU(union irq_stack_union, irq_stack_union); DECLARE_PER_CPU(char *, irq_stack_ptr); - -static inline void load_gs_base(int cpu) -{ - wrmsrl(MSR_GS_BASE, (unsigned long)per_cpu(irq_stack_union.gs_base, cpu)); -} #endif extern void print_cpu_info(struct cpuinfo_x86 *); diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c index 67e30c8a282c..0c766b80d915 100644 --- a/arch/x86/kernel/cpu/common.c +++ b/arch/x86/kernel/cpu/common.c @@ -258,12 +258,17 @@ __u32 cleared_cpu_caps[NCAPINTS] __cpuinitdata; void switch_to_new_gdt(void) { struct desc_ptr gdt_descr; + int cpu = smp_processor_id(); - gdt_descr.address = (long)get_cpu_gdt_table(smp_processor_id()); + gdt_descr.address = (long)get_cpu_gdt_table(cpu); gdt_descr.size = GDT_SIZE - 1; load_gdt(&gdt_descr); + /* Reload the per-cpu base */ #ifdef CONFIG_X86_32 - asm("mov %0, %%fs" : : "r" (__KERNEL_PERCPU) : "memory"); + loadsegment(fs, __KERNEL_PERCPU); +#else + loadsegment(gs, 0); + wrmsrl(MSR_GS_BASE, (unsigned long)per_cpu(irq_stack_union.gs_base, cpu)); #endif } @@ -968,10 +973,6 @@ void __cpuinit cpu_init(void) struct task_struct *me; int i; - loadsegment(fs, 0); - loadsegment(gs, 0); - load_gs_base(cpu); - #ifdef CONFIG_NUMA if (cpu != 0 && percpu_read(node_number) == 0 && cpu_to_node(cpu) != NUMA_NO_NODE) @@ -993,6 +994,8 @@ void __cpuinit cpu_init(void) */ switch_to_new_gdt(); + loadsegment(fs, 0); + load_idt((const struct desc_ptr *)&idt_descr); memset(me->thread.tls_array, 0, GDT_ENTRY_TLS_ENTRIES * 8); diff --git a/arch/x86/kernel/setup_percpu.c b/arch/x86/kernel/setup_percpu.c index bcca3a7b3748..4caa78d7cb15 100644 --- a/arch/x86/kernel/setup_percpu.c +++ b/arch/x86/kernel/setup_percpu.c @@ -112,14 +112,14 @@ void __init setup_per_cpu_areas(void) #ifdef CONFIG_NUMA per_cpu(x86_cpu_to_node_map, cpu) = early_per_cpu_map(x86_cpu_to_node_map, cpu); +#endif #endif /* * Up to this point, the boot CPU has been using .data.init - * area. Reload %gs offset for the boot CPU. + * area. Reload any changed state for the boot CPU. */ if (cpu == boot_cpu_id) - load_gs_base(cpu); -#endif + switch_to_new_gdt(); DBG("PERCPU: cpu %4d %p\n", cpu, ptr); } -- cgit