[PATCH] x86_64: reliable stack trace support (x86-64)

These are the x86_64-specific pieces to enable reliable stack traces. The
only restriction with this is that it currently cannot unwind across the
interrupt->normal stack boundary, as that transition is lacking proper
annotation.

Signed-off-by: Jan Beulich <jbeulich@novell.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

diff --git a/arch/x86_64/kernel/entry.S b/arch/x86_64/kernel/entry.S
index 9999d703b6c42fcd6c36e838b7f482bc4528d99c..6c68beec51b395cb8fe7f5b01a5f01762fb3e9c7 100644 (file)
--- a/arch/x86_64/kernel/entry.S
+++ b/arch/x86_64/kernel/entry.S
@@ -1089,3 +1089,36 @@ ENTRY(call_softirq)
        ret
        CFI_ENDPROC
 ENDPROC(call_softirq)
+
+#ifdef CONFIG_STACK_UNWIND
+ENTRY(arch_unwind_init_running)
+       CFI_STARTPROC
+       movq    %r15, R15(%rdi)
+       movq    %r14, R14(%rdi)
+       xchgq   %rsi, %rdx
+       movq    %r13, R13(%rdi)
+       movq    %r12, R12(%rdi)
+       xorl    %eax, %eax
+       movq    %rbp, RBP(%rdi)
+       movq    %rbx, RBX(%rdi)
+       movq    (%rsp), %rcx
+       movq    %rax, R11(%rdi)
+       movq    %rax, R10(%rdi)
+       movq    %rax, R9(%rdi)
+       movq    %rax, R8(%rdi)
+       movq    %rax, RAX(%rdi)
+       movq    %rax, RCX(%rdi)
+       movq    %rax, RDX(%rdi)
+       movq    %rax, RSI(%rdi)
+       movq    %rax, RDI(%rdi)
+       movq    %rax, ORIG_RAX(%rdi)
+       movq    %rcx, RIP(%rdi)
+       leaq    8(%rsp), %rcx
+       movq    $__KERNEL_CS, CS(%rdi)
+       movq    %rax, EFLAGS(%rdi)
+       movq    %rcx, RSP(%rdi)
+       movq    $__KERNEL_DS, SS(%rdi)
+       jmpq    *%rdx
+       CFI_ENDPROC
+ENDPROC(arch_unwind_init_running)
+#endif

diff --git a/arch/x86_64/kernel/process.c b/arch/x86_64/kernel/process.c
index dcb77b462348f824a7536b6ec0f681b35265c5b5..d6fa41459c8095b7126d5f170352d68f657a422c 100644 (file)
--- a/arch/x86_64/kernel/process.c
+++ b/arch/x86_64/kernel/process.c
@@ -334,7 +334,7 @@ void show_regs(struct pt_regs *regs)
 {
        printk("CPU %d:", smp_processor_id());
        __show_regs(regs);
-       show_trace(&regs->rsp);
+       show_trace(NULL, regs, (void *)(regs + 1));
 }
 
 /*

diff --git a/arch/x86_64/kernel/traps.c b/arch/x86_64/kernel/traps.c
index 9f8f1eff4a6c2806995c7465f69ceaefbeed9cf6..eb1534ff1f5f0cbe527f1689c4a76ed4a657a0b9 100644 (file)
--- a/arch/x86_64/kernel/traps.c
+++ b/arch/x86_64/kernel/traps.c
@@ -29,6 +29,7 @@
 #include <linux/nmi.h>
 #include <linux/kprobes.h>
 #include <linux/kexec.h>
+#include <linux/unwind.h>
 
 #include <asm/system.h>
 #include <asm/uaccess.h>
@@ -39,7 +40,7 @@
 #include <asm/i387.h>
 #include <asm/kdebug.h>
 #include <asm/processor.h>
-
+#include <asm/unwind.h>
 #include <asm/smp.h>
 #include <asm/pgalloc.h>
 #include <asm/pda.h>
@@ -189,6 +190,23 @@ static unsigned long *in_exception_stack(unsigned cpu, unsigned long stack,
        return NULL;
 }
 
+static void show_trace_unwind(struct unwind_frame_info *info, void *context)
+{
+       int i = 11;
+
+       while (unwind(info) == 0 && UNW_PC(info)) {
+               if (i > 50) {
+                       printk("\n       ");
+                       i = 7;
+               } else
+                       i += printk(" ");
+               i += printk_address(UNW_PC(info));
+               if (arch_unw_user_mode(info))
+                       break;
+       }
+       printk("\n");
+}
+
 /*
  * x86-64 can have upto three kernel stacks: 
  * process stack
@@ -196,15 +214,34 @@ static unsigned long *in_exception_stack(unsigned cpu, unsigned long stack,
  * severe exception (double fault, nmi, stack fault, debug, mce) hardware stack
  */
 
-void show_trace(unsigned long *stack)
+void show_trace(struct task_struct *tsk, struct pt_regs *regs, unsigned long * stack)
 {
        const unsigned cpu = safe_smp_processor_id();
        unsigned long *irqstack_end = (unsigned long *)cpu_pda(cpu)->irqstackptr;
        int i;
        unsigned used = 0;
+       struct unwind_frame_info info;
 
        printk("\nCall Trace:");
 
+       if (!tsk)
+               tsk = current;
+
+       if (regs) {
+               if (unwind_init_frame_info(&info, tsk, regs) == 0) {
+                       show_trace_unwind(&info, NULL);
+                       return;
+               }
+       } else if (tsk == current) {
+               if (unwind_init_running(&info, show_trace_unwind, NULL) == 0)
+                       return;
+       } else {
+               if (unwind_init_blocked(&info, tsk) == 0) {
+                       show_trace_unwind(&info, NULL);
+                       return;
+               }
+       }
+
 #define HANDLE_STACK(cond) \
        do while (cond) { \
                unsigned long addr = *stack++; \
@@ -262,7 +299,7 @@ void show_trace(unsigned long *stack)
        printk("\n");
 }
 
-void show_stack(struct task_struct *tsk, unsigned long * rsp)
+static void _show_stack(struct task_struct *tsk, struct pt_regs *regs, unsigned long * rsp)
 {
        unsigned long *stack;
        int i;
@@ -296,7 +333,12 @@ void show_stack(struct task_struct *tsk, unsigned long * rsp)
                printk("%016lx ", *stack++);
                touch_nmi_watchdog();
        }
-       show_trace((unsigned long *)rsp);
+       show_trace(tsk, regs, rsp);
+}
+
+void show_stack(struct task_struct *tsk, unsigned long * rsp)
+{
+       _show_stack(tsk, NULL, rsp);
 }
 
 /*
@@ -305,7 +347,7 @@ void show_stack(struct task_struct *tsk, unsigned long * rsp)
 void dump_stack(void)
 {
        unsigned long dummy;
-       show_trace(&dummy);
+       show_trace(NULL, NULL, &dummy);
 }
 
 EXPORT_SYMBOL(dump_stack);
@@ -332,7 +374,7 @@ void show_registers(struct pt_regs *regs)
        if (in_kernel) {
 
                printk("Stack: ");
-               show_stack(NULL, (unsigned long*)rsp);
+               _show_stack(NULL, regs, (unsigned long*)rsp);
 
                printk("\nCode: ");
                if (regs->rip < PAGE_OFFSET)

diff --git a/arch/x86_64/kernel/vmlinux.lds.S b/arch/x86_64/kernel/vmlinux.lds.S
index 5968c2415da91d83f20ae0e84d5e9bd135e31323..1c6a5f32291920d33954a81cc0c3089a28a56bdd 100644 (file)
--- a/arch/x86_64/kernel/vmlinux.lds.S
+++ b/arch/x86_64/kernel/vmlinux.lds.S
@@ -45,6 +45,15 @@ SECTIONS
 
   RODATA
 
+#ifdef CONFIG_STACK_UNWIND
+  . = ALIGN(8);
+  .eh_frame : AT(ADDR(.eh_frame) - LOAD_OFFSET) {
+       __start_unwind = .;
+       *(.eh_frame)
+       __end_unwind = .;
+  }
+#endif
+
                                /* Data */
   .data : AT(ADDR(.data) - LOAD_OFFSET) {
        *(.data)

diff --git a/include/asm-x86_64/proto.h b/include/asm-x86_64/proto.h
index 1064533e0959852fd19d3637a551fcb0aa64dfb8..105476ff803911a10c9b8f9cd45a105593aa3ba8 100644 (file)
--- a/include/asm-x86_64/proto.h
+++ b/include/asm-x86_64/proto.h
@@ -75,7 +75,7 @@ extern void main_timer_handler(struct pt_regs *regs);
 
 extern unsigned long end_pfn_map; 
 
-extern void show_trace(unsigned long * rsp);
+extern void show_trace(struct task_struct *, struct pt_regs *, unsigned long * rsp);
 extern void show_registers(struct pt_regs *regs);
 
 extern void exception_table_check(void);

diff --git a/include/asm-x86_64/unwind.h b/include/asm-x86_64/unwind.h
new file mode 100644 (file)
index 0000000..4f61de2
--- /dev/null
+++ b/include/asm-x86_64/unwind.h
@@ -0,0 +1,106 @@
+#ifndef _ASM_X86_64_UNWIND_H
+#define _ASM_X86_64_UNWIND_H
+
+/*
+ * Copyright (C) 2002-2006 Novell, Inc.
+ *     Jan Beulich <jbeulich@novell.com>
+ * This code is released under version 2 of the GNU GPL.
+ */
+
+#ifdef CONFIG_STACK_UNWIND
+
+#include <linux/sched.h>
+#include <asm/ptrace.h>
+#include <asm/uaccess.h>
+#include <asm/vsyscall.h>
+
+struct unwind_frame_info
+{
+       struct pt_regs regs;
+       struct task_struct *task;
+};
+
+#define UNW_PC(frame)        (frame)->regs.rip
+#define UNW_SP(frame)        (frame)->regs.rsp
+#ifdef CONFIG_FRAME_POINTER
+#define UNW_FP(frame)        (frame)->regs.rbp
+#define FRAME_RETADDR_OFFSET 8
+#define FRAME_LINK_OFFSET    0
+#define STACK_BOTTOM(tsk)    (((tsk)->thread.rsp0 - 1) & ~(THREAD_SIZE - 1))
+#define STACK_TOP(tsk)       ((tsk)->thread.rsp0)
+#endif
+/* Might need to account for the special exception and interrupt handling
+   stacks here, since normally
+       EXCEPTION_STACK_ORDER < THREAD_ORDER < IRQSTACK_ORDER,
+   but the construct is needed only for getting across the stack switch to
+   the interrupt stack - thus considering the IRQ stack itself is unnecessary,
+   and the overhead of comparing against all exception handling stacks seems
+   not desirable. */
+#define STACK_LIMIT(ptr)     (((ptr) - 1) & ~(THREAD_SIZE - 1))
+
+#define UNW_REGISTER_INFO \
+       PTREGS_INFO(rax), \
+       PTREGS_INFO(rdx), \
+       PTREGS_INFO(rcx), \
+       PTREGS_INFO(rbx), \
+       PTREGS_INFO(rsi), \
+       PTREGS_INFO(rdi), \
+       PTREGS_INFO(rbp), \
+       PTREGS_INFO(rsp), \
+       PTREGS_INFO(r8), \
+       PTREGS_INFO(r9), \
+       PTREGS_INFO(r10), \
+       PTREGS_INFO(r11), \
+       PTREGS_INFO(r12), \
+       PTREGS_INFO(r13), \
+       PTREGS_INFO(r14), \
+       PTREGS_INFO(r15), \
+       PTREGS_INFO(rip)
+
+static inline void arch_unw_init_frame_info(struct unwind_frame_info *info,
+                                            /*const*/ struct pt_regs *regs)
+{
+       info->regs = *regs;
+}
+
+static inline void arch_unw_init_blocked(struct unwind_frame_info *info)
+{
+       extern const char thread_return[];
+
+       memset(&info->regs, 0, sizeof(info->regs));
+       info->regs.rip = (unsigned long)thread_return;
+       info->regs.cs = __KERNEL_CS;
+       __get_user(info->regs.rbp, (unsigned long *)info->task->thread.rsp);
+       info->regs.rsp = info->task->thread.rsp;
+       info->regs.ss = __KERNEL_DS;
+}
+
+extern void arch_unwind_init_running(struct unwind_frame_info *,
+                                     void (*callback)(struct unwind_frame_info *,
+                                                      void *arg),
+                                     void *arg);
+
+static inline int arch_unw_user_mode(const struct unwind_frame_info *info)
+{
+#if 0 /* This can only work when selector register saves/restores
+         are properly annotated (and tracked in UNW_REGISTER_INFO). */
+       return user_mode(&info->regs);
+#else
+       return (long)info->regs.rip >= 0
+              || (info->regs.rip >= VSYSCALL_START && info->regs.rip < VSYSCALL_END)
+              || (long)info->regs.rsp >= 0;
+#endif
+}
+
+#else
+
+#define UNW_PC(frame) ((void)(frame), 0)
+
+static inline int arch_unw_user_mode(const void *info)
+{
+       return 0;
+}
+
+#endif
+
+#endif /* _ASM_X86_64_UNWIND_H */

diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug
index 256b3b805c5c985f154651687111616de420bc8b..bbaed84a686b79c7201b8fb2662d570550e5f5bf 100644 (file)
--- a/lib/Kconfig.debug
+++ b/lib/Kconfig.debug
@@ -199,7 +199,7 @@ config UNWIND_INFO
 config STACK_UNWIND
        bool "Stack unwind support"
        depends on UNWIND_INFO
-       depends on n
+       depends on X86_64
        help
          This enables more precise stack traces, omitting all unrelated
          occurrences of pointers into kernel code from the dump.