1 files changed, 39 insertions, 24 deletions

diff --git a/arch/x86_64/kernel/entry.S b/arch/x86_64/kernel/entry.S
index aa8d8939abc1..2802524104f3 100644
--- a/arch/x86_64/kernel/entry.S
+++ b/arch/x86_64/kernel/entry.S
@@ -4,8 +4,6 @@
  *  Copyright (C) 1991, 1992  Linus Torvalds
  *  Copyright (C) 2000, 2001, 2002  Andi Kleen SuSE Labs
  *  Copyright (C) 2000  Pavel Machek <pavel@suse.cz>
- * 
- *  $Id$
  */
 
 /*
@@ -22,15 +20,25 @@
  * at the top of the kernel process stack.      
  * - partial stack frame: partially saved registers upto R11.
  * - full stack frame: Like partial stack frame, but all register saved. 
- *      
- * TODO:         
- * - schedule it carefully for the final hardware.
+ *
+ * Some macro usage:
+ * - CFI macros are used to generate dwarf2 unwind information for better
+ * backtraces. They don't change any code.
+ * - SAVE_ALL/RESTORE_ALL - Save/restore all registers
+ * - SAVE_ARGS/RESTORE_ARGS - Save/restore registers that C functions modify.
+ * There are unfortunately lots of special cases where some registers
+ * not touched. The macro is a big mess that should be cleaned up.
+ * - SAVE_REST/RESTORE_REST - Handle the registers not saved by SAVE_ARGS.
+ * Gives a full stack frame.
+ * - ENTRY/END Define functions in the symbol table.
+ * - FIXUP_TOP_OF_STACK/RESTORE_TOP_OF_STACK - Fix up the hardware stack
+ * frame that is otherwise undefined after a SYSCALL
+ * - TRACE_IRQ_* - Trace hard interrupt state for lock debugging.
+ * - errorentry/paranoidentry/zeroentry - Define exception entry points.
  */
 
-#define ASSEMBLY 1
 #include <linux/linkage.h>
 #include <asm/segment.h>
-#include <asm/smp.h>
 #include <asm/cache.h>
 #include <asm/errno.h>
 #include <asm/dwarf2.h>
@@ -115,6 +123,7 @@
         .macro  CFI_DEFAULT_STACK start=1
         .if \start
         CFI_STARTPROC   simple
+        CFI_SIGNAL_FRAME
         CFI_DEF_CFA     rsp,SS+8
         .else
         CFI_DEF_CFA_OFFSET SS+8
@@ -146,6 +155,10 @@
 /* rdi: prev */ 
 ENTRY(ret_from_fork)
         CFI_DEFAULT_STACK
+        push kernel_eflags(%rip)
+        CFI_ADJUST_CFA_OFFSET 4
+        popf                            # reset kernel eflags
+        CFI_ADJUST_CFA_OFFSET -4
         call schedule_tail
         GET_THREAD_INFO(%rcx)
         testl $(_TIF_SYSCALL_TRACE|_TIF_SYSCALL_AUDIT),threadinfo_flags(%rcx)
@@ -199,6 +212,7 @@ END(ret_from_fork)
 
 ENTRY(system_call)
         CFI_STARTPROC   simple
+        CFI_SIGNAL_FRAME
         CFI_DEF_CFA     rsp,PDA_STACKOFFSET
         CFI_REGISTER    rip,rcx
         /*CFI_REGISTER  rflags,r11*/
@@ -316,6 +330,7 @@ END(system_call)
  */     
 ENTRY(int_ret_from_sys_call)
         CFI_STARTPROC   simple
+        CFI_SIGNAL_FRAME
         CFI_DEF_CFA     rsp,SS+8-ARGOFFSET
         /*CFI_REL_OFFSET        ss,SS-ARGOFFSET*/
         CFI_REL_OFFSET  rsp,RSP-ARGOFFSET
@@ -476,6 +491,7 @@ END(stub_rt_sigreturn)
  */
         .macro _frame ref
         CFI_STARTPROC simple
+        CFI_SIGNAL_FRAME
         CFI_DEF_CFA rsp,SS+8-\ref
         /*CFI_REL_OFFSET ss,SS-\ref*/
         CFI_REL_OFFSET rsp,RSP-\ref
@@ -511,7 +527,12 @@ END(stub_rt_sigreturn)
         testl $3,CS(%rdi)
         je 1f
         swapgs  
-1:      incl    %gs:pda_irqcount        # RED-PEN should check preempt count
+        /* irqcount is used to check if a CPU is already on an interrupt
+           stack or not. While this is essentially redundant with preempt_count
+           it is a little cheaper to use a separate counter in the PDA
+           (short of moving irq_enter into assembly, which would be too
+            much work) */
+1:      incl    %gs:pda_irqcount
         cmoveq %gs:pda_irqstackptr,%rsp
         push    %rbp                    # backlink for old unwinder
         /*
@@ -619,8 +640,7 @@ retint_signal:
 #ifdef CONFIG_PREEMPT
         /* Returning to kernel space. Check if we need preemption */
         /* rcx:  threadinfo. interrupts off. */
-        .p2align
-retint_kernel:  
+ENTRY(retint_kernel)
         cmpl $0,threadinfo_preempt_count(%rcx)
         jnz  retint_restore_args
         bt  $TIF_NEED_RESCHED,threadinfo_flags(%rcx)
@@ -679,7 +699,6 @@ ENTRY(call_function_interrupt)
 END(call_function_interrupt)
 #endif
 
-#ifdef CONFIG_X86_LOCAL_APIC    
 ENTRY(apic_timer_interrupt)
         apicinterrupt LOCAL_TIMER_VECTOR,smp_apic_timer_interrupt
 END(apic_timer_interrupt)
@@ -691,7 +710,6 @@ END(error_interrupt)
 ENTRY(spurious_interrupt)
         apicinterrupt SPURIOUS_APIC_VECTOR,smp_spurious_interrupt
 END(spurious_interrupt)
-#endif
                                 
 /*
  * Exception entry points.
@@ -768,7 +786,9 @@ paranoid_exit\trace:
         testl $3,CS(%rsp)
         jnz   paranoid_userspace\trace
 paranoid_swapgs\trace:
+        .if \trace
         TRACE_IRQS_IRETQ 0
+        .endif
         swapgs
 paranoid_restore\trace:
         RESTORE_ALL 8
@@ -814,7 +834,7 @@ paranoid_schedule\trace:
  * Exception entry point. This expects an error code/orig_rax on the stack
  * and the exception handler in %rax.   
  */                                             
-ENTRY(error_entry)
+KPROBE_ENTRY(error_entry)
         _frame RDI
         /* rdi slot contains rax, oldrax contains error code */
         cld     
@@ -898,7 +918,7 @@ error_kernelspace:
         cmpq $gs_change,RIP(%rsp)
         je   error_swapgs
         jmp  error_sti
-END(error_entry)
+KPROBE_END(error_entry)
         
        /* Reload gs selector with exception handling */
        /* edi:  new selector */ 
@@ -1020,8 +1040,7 @@ ENDPROC(execve)
 
 KPROBE_ENTRY(page_fault)
         errorentry do_page_fault
-END(page_fault)
-        .previous .text
+KPROBE_END(page_fault)
 
 ENTRY(coprocessor_error)
         zeroentry do_coprocessor_error
@@ -1042,8 +1061,7 @@ KPROBE_ENTRY(debug)
         CFI_ADJUST_CFA_OFFSET 8         
         paranoidentry do_debug, DEBUG_STACK
         paranoidexit
-END(debug)
-        .previous .text
+KPROBE_END(debug)
 
         /* runs on exception stack */   
 KPROBE_ENTRY(nmi)
@@ -1057,8 +1075,7 @@ KPROBE_ENTRY(nmi)
         jmp paranoid_exit1
         CFI_ENDPROC
 #endif
-END(nmi)
-        .previous .text
+KPROBE_END(nmi)
 
 KPROBE_ENTRY(int3)
         INTR_FRAME
@@ -1067,8 +1084,7 @@ KPROBE_ENTRY(int3)
         paranoidentry do_int3, DEBUG_STACK
         jmp paranoid_exit1
         CFI_ENDPROC
-END(int3)
-        .previous .text
+KPROBE_END(int3)
 
 ENTRY(overflow)
         zeroentry do_overflow
@@ -1116,8 +1132,7 @@ END(stack_segment)
 
 KPROBE_ENTRY(general_protection)
         errorentry do_general_protection
-END(general_protection)
-        .previous .text
+KPROBE_END(general_protection)
 
 ENTRY(alignment_check)
         errorentry do_alignment_check