x86-64: Move kernelstack from PDA to per-cpu.

Also clean up PER_CPU_VAR usage in xen-asm_64.S

tj: * remove now unused stack_thread_info()
    * s/kernelstack/kernel_stack/
    * added FIXME comment in xen-asm_64.S

Signed-off-by: Brian Gerst <brgerst@gmail.com>
Signed-off-by: Tejun Heo <tj@kernel.org>

9 files changed, 35 insertions, 38 deletions

diff --git a/arch/x86/ia32/ia32entry.S b/arch/x86/ia32/ia32entry.S
index 256b00b61892..9c79b2477008 100644
--- a/arch/x86/ia32/ia32entry.S
+++ b/arch/x86/ia32/ia32entry.S
@@ -112,8 +112,8 @@ ENTRY(ia32_sysenter_target)
         CFI_DEF_CFA     rsp,0
         CFI_REGISTER    rsp,rbp
         SWAPGS_UNSAFE_STACK
-        movq    %gs:pda_kernelstack, %rsp
-        addq    $(PDA_STACKOFFSET),%rsp 
+        movq    PER_CPU_VAR(kernel_stack), %rsp
+        addq    $(KERNEL_STACK_OFFSET),%rsp
         /*
          * No need to follow this irqs on/off section: the syscall
          * disabled irqs, here we enable it straight after entry:
@@ -273,13 +273,13 @@ ENDPROC(ia32_sysenter_target)
 ENTRY(ia32_cstar_target)
         CFI_STARTPROC32 simple
         CFI_SIGNAL_FRAME
-        CFI_DEF_CFA     rsp,PDA_STACKOFFSET
+        CFI_DEF_CFA     rsp,KERNEL_STACK_OFFSET
         CFI_REGISTER    rip,rcx
         /*CFI_REGISTER  rflags,r11*/
         SWAPGS_UNSAFE_STACK
         movl    %esp,%r8d
         CFI_REGISTER    rsp,r8
-        movq    %gs:pda_kernelstack,%rsp
+        movq    PER_CPU_VAR(kernel_stack),%rsp
         /*
          * No need to follow this irqs on/off section: the syscall
          * disabled irqs and here we enable it straight after entry:
diff --git a/arch/x86/include/asm/pda.h b/arch/x86/include/asm/pda.h
index 7209302d9227..4d28ffba6e1b 100644
--- a/arch/x86/include/asm/pda.h
+++ b/arch/x86/include/asm/pda.h
@@ -13,7 +13,7 @@
 struct x8664_pda {
         unsigned long unused1;
         unsigned long unused2;
-        unsigned long kernelstack;      /* 16 top of kernel stack for current */
+        unsigned long unused3;
         unsigned long oldrsp;           /* 24 user rsp for system call */
         int irqcount;                   /* 32 Irq nesting counter. Starts -1 */
         unsigned int unused6;           /* 36 was cpunumber */
@@ -44,6 +44,4 @@ extern void pda_init(int);
 
 #endif
 
-#define PDA_STACKOFFSET (5*8)
-
 #endif /* _ASM_X86_PDA_H */
diff --git a/arch/x86/include/asm/thread_info.h b/arch/x86/include/asm/thread_info.h
index 98789647baa9..b46f8ca007b5 100644
--- a/arch/x86/include/asm/thread_info.h
+++ b/arch/x86/include/asm/thread_info.h
@@ -194,25 +194,21 @@ static inline struct thread_info *current_thread_info(void)
 
 #else /* X86_32 */
 
-#include <asm/pda.h>
+#include <asm/percpu.h>
+#define KERNEL_STACK_OFFSET (5*8)
 
 /*
  * macros/functions for gaining access to the thread information structure
  * preempt_count needs to be 1 initially, until the scheduler is functional.
  */
 #ifndef __ASSEMBLY__
-static inline struct thread_info *current_thread_info(void)
-{
-        struct thread_info *ti;
-        ti = (void *)(read_pda(kernelstack) + PDA_STACKOFFSET - THREAD_SIZE);
-        return ti;
-}
+DECLARE_PER_CPU(unsigned long, kernel_stack);
 
-/* do not use in interrupt context */
-static inline struct thread_info *stack_thread_info(void)
+static inline struct thread_info *current_thread_info(void)
 {
         struct thread_info *ti;
-        asm("andq %%rsp,%0; " : "=r" (ti) : "0" (~(THREAD_SIZE - 1)));
+        ti = (void *)(percpu_read(kernel_stack) +
+                      KERNEL_STACK_OFFSET - THREAD_SIZE);
         return ti;
 }
 
@@ -220,8 +216,8 @@ static inline struct thread_info *stack_thread_info(void)
 
 /* how to get the thread information struct from ASM */
 #define GET_THREAD_INFO(reg) \
-        movq %gs:pda_kernelstack,reg ; \
-        subq $(THREAD_SIZE-PDA_STACKOFFSET),reg
+        movq PER_CPU_VAR(kernel_stack),reg ; \
+        subq $(THREAD_SIZE-KERNEL_STACK_OFFSET),reg
 
 #endif
 
diff --git a/arch/x86/kernel/asm-offsets_64.c b/arch/x86/kernel/asm-offsets_64.c
index 4f7a210e1e58..cafff5f4a031 100644
--- a/arch/x86/kernel/asm-offsets_64.c
+++ b/arch/x86/kernel/asm-offsets_64.c
@@ -49,7 +49,6 @@ int main(void)
         BLANK();
 #undef ENTRY
 #define ENTRY(entry) DEFINE(pda_ ## entry, offsetof(struct x8664_pda, entry))
-        ENTRY(kernelstack); 
         ENTRY(oldrsp); 
         ENTRY(irqcount);
         DEFINE(pda_size, sizeof(struct x8664_pda));
diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c
index b50e38d16888..06b6290088f4 100644
--- a/arch/x86/kernel/cpu/common.c
+++ b/arch/x86/kernel/cpu/common.c
@@ -889,6 +889,10 @@ DEFINE_PER_CPU(char *, irq_stack_ptr) =
         per_cpu_var(irq_stack) + IRQ_STACK_SIZE - 64;
 #endif
 
+DEFINE_PER_CPU(unsigned long, kernel_stack) =
+        (unsigned long)&init_thread_union - KERNEL_STACK_OFFSET + THREAD_SIZE;
+EXPORT_PER_CPU_SYMBOL(kernel_stack);
+
 void __cpuinit pda_init(int cpu)
 {
         struct x8664_pda *pda = cpu_pda(cpu);
@@ -900,8 +904,6 @@ void __cpuinit pda_init(int cpu)
         load_pda_offset(cpu);
 
         pda->irqcount = -1;
-        pda->kernelstack = (unsigned long)stack_thread_info() -
-                                 PDA_STACKOFFSET + THREAD_SIZE;
 
         if (cpu != 0) {
                 if (pda->nodenumber == 0 && cpu_to_node(cpu) != NUMA_NO_NODE)
diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S
index d22677a66438..0dd45859a7a8 100644
--- a/arch/x86/kernel/entry_64.S
+++ b/arch/x86/kernel/entry_64.S
@@ -468,7 +468,7 @@ END(ret_from_fork)
 ENTRY(system_call)
         CFI_STARTPROC   simple
         CFI_SIGNAL_FRAME
-        CFI_DEF_CFA     rsp,PDA_STACKOFFSET
+        CFI_DEF_CFA     rsp,KERNEL_STACK_OFFSET
         CFI_REGISTER    rip,rcx
         /*CFI_REGISTER  rflags,r11*/
         SWAPGS_UNSAFE_STACK
@@ -480,7 +480,7 @@ ENTRY(system_call)
 ENTRY(system_call_after_swapgs)
 
         movq    %rsp,%gs:pda_oldrsp
-        movq    %gs:pda_kernelstack,%rsp
+        movq    PER_CPU_VAR(kernel_stack),%rsp
         /*
          * No need to follow this irqs off/on section - it's straight
          * and short:
diff --git a/arch/x86/kernel/process_64.c b/arch/x86/kernel/process_64.c
index e00c31a4b3c0..6c5f57602108 100644
--- a/arch/x86/kernel/process_64.c
+++ b/arch/x86/kernel/process_64.c
@@ -620,9 +620,9 @@ __switch_to(struct task_struct *prev_p, struct task_struct *next_p)
         write_pda(oldrsp, next->usersp);
         percpu_write(current_task, next_p);
 
-        write_pda(kernelstack,
+        percpu_write(kernel_stack,
                   (unsigned long)task_stack_page(next_p) +
-                  THREAD_SIZE - PDA_STACKOFFSET);
+                  THREAD_SIZE - KERNEL_STACK_OFFSET);
 #ifdef CONFIG_CC_STACKPROTECTOR
         write_pda(stack_canary, next_p->stack_canary);
         /*
diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c
index 5854be0fb804..869b98840fd0 100644
--- a/arch/x86/kernel/smpboot.c
+++ b/arch/x86/kernel/smpboot.c
@@ -798,6 +798,9 @@ do_rest:
 #else
         clear_tsk_thread_flag(c_idle.idle, TIF_FORK);
         initial_gs = per_cpu_offset(cpu);
+        per_cpu(kernel_stack, cpu) =
+                (unsigned long)task_stack_page(c_idle.idle) -
+                KERNEL_STACK_OFFSET + THREAD_SIZE;
 #endif
         early_gdt_descr.address = (unsigned long)get_cpu_gdt_table(cpu);
         initial_code = (unsigned long)start_secondary;
diff --git a/arch/x86/xen/xen-asm_64.S b/arch/x86/xen/xen-asm_64.S
index 05794c566e87..5a23e8993678 100644
--- a/arch/x86/xen/xen-asm_64.S
+++ b/arch/x86/xen/xen-asm_64.S
@@ -17,6 +17,7 @@
 #include <asm/processor-flags.h>
 #include <asm/errno.h>
 #include <asm/segment.h>
+#include <asm/percpu.h>
 
 #include <xen/interface/xen.h>
 
@@ -28,12 +29,10 @@
 
 #if 1
 /*
-        x86-64 does not yet support direct access to percpu variables
-        via a segment override, so we just need to make sure this code
-        never gets used
+        FIXME: x86_64 now can support direct access to percpu variables
+        via a segment override.  Update xen accordingly.
  */
 #define BUG                     ud2a
-#define PER_CPU_VAR(var, off)   0xdeadbeef
 #endif
 
 /*
@@ -45,14 +44,14 @@ ENTRY(xen_irq_enable_direct)
         BUG
 
         /* Unmask events */
-        movb $0, PER_CPU_VAR(xen_vcpu_info, XEN_vcpu_info_mask)
+        movb $0, PER_CPU_VAR(xen_vcpu_info) + XEN_vcpu_info_mask
 
         /* Preempt here doesn't matter because that will deal with
            any pending interrupts.  The pending check may end up being
            run on the wrong CPU, but that doesn't hurt. */
 
         /* Test for pending */
-        testb $0xff, PER_CPU_VAR(xen_vcpu_info, XEN_vcpu_info_pending)
+        testb $0xff, PER_CPU_VAR(xen_vcpu_info) + XEN_vcpu_info_pending
         jz 1f
 
 2:      call check_events
@@ -69,7 +68,7 @@ ENDPATCH(xen_irq_enable_direct)
 ENTRY(xen_irq_disable_direct)
         BUG
 
-        movb $1, PER_CPU_VAR(xen_vcpu_info, XEN_vcpu_info_mask)
+        movb $1, PER_CPU_VAR(xen_vcpu_info) + XEN_vcpu_info_mask
 ENDPATCH(xen_irq_disable_direct)
         ret
         ENDPROC(xen_irq_disable_direct)
@@ -87,7 +86,7 @@ ENDPATCH(xen_irq_disable_direct)
 ENTRY(xen_save_fl_direct)
         BUG
 
-        testb $0xff, PER_CPU_VAR(xen_vcpu_info, XEN_vcpu_info_mask)
+        testb $0xff, PER_CPU_VAR(xen_vcpu_info) + XEN_vcpu_info_mask
         setz %ah
         addb %ah,%ah
 ENDPATCH(xen_save_fl_direct)
@@ -107,13 +106,13 @@ ENTRY(xen_restore_fl_direct)
         BUG
 
         testb $X86_EFLAGS_IF>>8, %ah
-        setz PER_CPU_VAR(xen_vcpu_info, XEN_vcpu_info_mask)
+        setz PER_CPU_VAR(xen_vcpu_info) + XEN_vcpu_info_mask
         /* Preempt here doesn't matter because that will deal with
            any pending interrupts.  The pending check may end up being
            run on the wrong CPU, but that doesn't hurt. */
 
         /* check for unmasked and pending */
-        cmpw $0x0001, PER_CPU_VAR(xen_vcpu_info, XEN_vcpu_info_pending)
+        cmpw $0x0001, PER_CPU_VAR(xen_vcpu_info) + XEN_vcpu_info_pending
         jz 1f
 2:      call check_events
 1:
@@ -196,7 +195,7 @@ ENTRY(xen_sysret64)
         /* We're already on the usermode stack at this point, but still
            with the kernel gs, so we can easily switch back */
         movq %rsp, %gs:pda_oldrsp
-        movq %gs:pda_kernelstack,%rsp
+        movq PER_CPU_VAR(kernel_stack),%rsp
 
         pushq $__USER_DS
         pushq %gs:pda_oldrsp
@@ -213,7 +212,7 @@ ENTRY(xen_sysret32)
         /* We're already on the usermode stack at this point, but still
            with the kernel gs, so we can easily switch back */
         movq %rsp, %gs:pda_oldrsp
-        movq %gs:pda_kernelstack, %rsp
+        movq PER_CPU_VAR(kernel_stack), %rsp
 
         pushq $__USER32_DS
         pushq %gs:pda_oldrsp