[PATCH] i386/x86-64: Work around gcc bug with noreturn functions in unwinder

Current gcc generates calls not jumps to noreturn functions. When that happens the
return address can point to the next function, which confuses the unwinder.

This patch works around it by marking asynchronous exception
frames in contrast normal call frames in the unwind information.  Then teach
the unwinder to decode this.

For normal call frames the unwinder now subtracts one from the address which avoids
this problem.  The standard libgcc unwinder uses the same trick.

It doesn't include adjustment of the printed address (i.e. for the original
example, it'd still be kernel_math_error+0 that gets displayed, but the
unwinder wouldn't get confused anymore.

This only works with binutils 2.6.17+ and some versions of H.J.Lu's 2.6.16
unfortunately because earlier binutils don't support .cfi_signal_frame

[AK: added automatic detection of the new binutils and wrote description]

Signed-off-by: Jan Beulich <jbeulich@novell.com>
Signed-off-by: Andi Kleen <ak@suse.de>

11 files changed, 73 insertions, 9 deletions

diff --git a/arch/i386/Makefile b/arch/i386/Makefile
index 508cdbeb3a09..7cc0b189b82b 100644
--- a/arch/i386/Makefile
+++ b/arch/i386/Makefile
@@ -50,6 +50,10 @@ CFLAGS				+= $(shell if [ $(call cc-version) -lt 0400 ] ; then echo $(call cc-op
 cflags-y += $(call as-instr,.cfi_startproc\n.cfi_endproc,-DCONFIG_AS_CFI=1,)
 AFLAGS += $(call as-instr,.cfi_startproc\n.cfi_endproc,-DCONFIG_AS_CFI=1,)
 
+# is .cfi_signal_frame supported too?
+cflags-y += $(call as-instr,.cfi_startproc\n.cfi_endproc,-DCONFIG_AS_CFI=1,)
+AFLAGS += $(call as-instr,.cfi_startproc\n.cfi_endproc,-DCONFIG_AS_CFI=1,)
+
 CFLAGS += $(cflags-y)
 
 # Default subarch .c files
diff --git a/arch/i386/kernel/entry.S b/arch/i386/kernel/entry.S
index 284f2e908ad0..5a63d6fdb70e 100644
--- a/arch/i386/kernel/entry.S
+++ b/arch/i386/kernel/entry.S
@@ -183,18 +183,21 @@ VM_MASK		= 0x00020000
 
 #define RING0_INT_FRAME \
         CFI_STARTPROC simple;\
+        CFI_SIGNAL_FRAME;\
         CFI_DEF_CFA esp, 3*4;\
         /*CFI_OFFSET cs, -2*4;*/\
         CFI_OFFSET eip, -3*4
 
 #define RING0_EC_FRAME \
         CFI_STARTPROC simple;\
+        CFI_SIGNAL_FRAME;\
         CFI_DEF_CFA esp, 4*4;\
         /*CFI_OFFSET cs, -2*4;*/\
         CFI_OFFSET eip, -3*4
 
 #define RING0_PTREGS_FRAME \
         CFI_STARTPROC simple;\
+        CFI_SIGNAL_FRAME;\
         CFI_DEF_CFA esp, OLDESP-EBX;\
         /*CFI_OFFSET cs, CS-OLDESP;*/\
         CFI_OFFSET eip, EIP-OLDESP;\
@@ -275,6 +278,7 @@ need_resched:
         # sysenter call handler stub
 ENTRY(sysenter_entry)
         CFI_STARTPROC simple
+        CFI_SIGNAL_FRAME
         CFI_DEF_CFA esp, 0
         CFI_REGISTER esp, ebp
         movl TSS_sysenter_esp0(%esp),%esp
diff --git a/arch/x86_64/Makefile b/arch/x86_64/Makefile
index 2b8d07c70106..1c0f18d4f887 100644
--- a/arch/x86_64/Makefile
+++ b/arch/x86_64/Makefile
@@ -58,6 +58,10 @@ cflags-y += $(call cc-option,-mno-sse -mno-mmx -mno-sse2 -mno-3dnow,)
 cflags-y += $(call as-instr,.cfi_startproc\n.cfi_endproc,-DCONFIG_AS_CFI=1,)
 AFLAGS += $(call as-instr,.cfi_startproc\n.cfi_endproc,-DCONFIG_AS_CFI=1,)
 
+# is .cfi_signal_frame supported too?
+cflags-y += $(call as-instr,.cfi_startproc\n.cfi_signal_frame\n.cfi_endproc,-DCONFIG_AS_CFI_SIGNAL_FRAME=1,)
+AFLAGS += $(call as-instr,.cfi_startproc\n.cfi_signal_frame\n.cfi_endproc,-DCONFIG_AS_CFI_SIGNAL_FRAME=1,)
+
 cflags-$(CONFIG_CC_STACKPROTECTOR) += $(shell $(CONFIG_SHELL) $(srctree)/scripts/gcc-x86_64-has-stack-protector.sh $(CC) -fstack-protector )
 cflags-$(CONFIG_CC_STACKPROTECTOR_ALL) += $(shell $(CONFIG_SHELL) $(srctree)/scripts/gcc-x86_64-has-stack-protector.sh $(CC) -fstack-protector-all )
 
diff --git a/arch/x86_64/ia32/ia32entry.S b/arch/x86_64/ia32/ia32entry.S
index 32fd32bea07c..b4aa875e175b 100644
--- a/arch/x86_64/ia32/ia32entry.S
+++ b/arch/x86_64/ia32/ia32entry.S
@@ -71,6 +71,7 @@
  */     
 ENTRY(ia32_sysenter_target)
         CFI_STARTPROC32 simple
+        CFI_SIGNAL_FRAME
         CFI_DEF_CFA     rsp,0
         CFI_REGISTER    rsp,rbp
         swapgs
@@ -186,6 +187,7 @@ ENDPROC(ia32_sysenter_target)
  */     
 ENTRY(ia32_cstar_target)
         CFI_STARTPROC32 simple
+        CFI_SIGNAL_FRAME
         CFI_DEF_CFA     rsp,PDA_STACKOFFSET
         CFI_REGISTER    rip,rcx
         /*CFI_REGISTER  rflags,r11*/
@@ -293,6 +295,7 @@ ia32_badarg:
 
 ENTRY(ia32_syscall)
         CFI_STARTPROC   simple
+        CFI_SIGNAL_FRAME
         CFI_DEF_CFA     rsp,SS+8-RIP
         /*CFI_REL_OFFSET        ss,SS-RIP*/
         CFI_REL_OFFSET  rsp,RSP-RIP
@@ -370,6 +373,7 @@ ENTRY(ia32_ptregs_common)
         popq %r11
         CFI_ENDPROC
         CFI_STARTPROC32 simple
+        CFI_SIGNAL_FRAME
         CFI_DEF_CFA     rsp,SS+8-ARGOFFSET
         CFI_REL_OFFSET  rax,RAX-ARGOFFSET
         CFI_REL_OFFSET  rcx,RCX-ARGOFFSET
diff --git a/arch/x86_64/kernel/entry.S b/arch/x86_64/kernel/entry.S
index ea32688386fd..4cbc65290ae7 100644
--- a/arch/x86_64/kernel/entry.S
+++ b/arch/x86_64/kernel/entry.S
@@ -123,6 +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
@@ -207,6 +208,7 @@ 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*/
@@ -324,6 +326,7 @@ 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
@@ -484,6 +487,7 @@ 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
diff --git a/include/asm-i386/dwarf2.h b/include/asm-i386/dwarf2.h
index 5d1a8db5a9b0..6d66398a307d 100644
--- a/include/asm-i386/dwarf2.h
+++ b/include/asm-i386/dwarf2.h
@@ -28,6 +28,12 @@
 #define CFI_RESTORE_STATE .cfi_restore_state
 #define CFI_UNDEFINED .cfi_undefined
 
+#ifdef CONFIG_AS_CFI_SIGNAL_FRAME
+#define CFI_SIGNAL_FRAME .cfi_signal_frame
+#else
+#define CFI_SIGNAL_FRAME
+#endif
+
 #else
 
 /* Due to the structure of pre-exisiting code, don't use assembler line
@@ -48,6 +54,7 @@
 #define CFI_REMEMBER_STATE ignore
 #define CFI_RESTORE_STATE ignore
 #define CFI_UNDEFINED ignore
+#define CFI_SIGNAL_FRAME ignore
 
 #endif
 
diff --git a/include/asm-i386/unwind.h b/include/asm-i386/unwind.h
index f0ac399bae3c..5031d693b89d 100644
--- a/include/asm-i386/unwind.h
+++ b/include/asm-i386/unwind.h
@@ -18,6 +18,7 @@ struct unwind_frame_info
 {
         struct pt_regs regs;
         struct task_struct *task;
+        unsigned call_frame:1;
 };
 
 #define UNW_PC(frame)        (frame)->regs.eip
@@ -44,6 +45,10 @@ struct unwind_frame_info
         PTREGS_INFO(edi), \
         PTREGS_INFO(eip)
 
+#define UNW_DEFAULT_RA(raItem, dataAlign) \
+        ((raItem).where == Memory && \
+         !((raItem).value * (dataAlign) + 4))
+
 static inline void arch_unw_init_frame_info(struct unwind_frame_info *info,
                                             /*const*/ struct pt_regs *regs)
 {
diff --git a/include/asm-x86_64/dwarf2.h b/include/asm-x86_64/dwarf2.h
index 2b9368365fad..eedc08526b0b 100644
--- a/include/asm-x86_64/dwarf2.h
+++ b/include/asm-x86_64/dwarf2.h
@@ -28,6 +28,11 @@
 #define CFI_REMEMBER_STATE .cfi_remember_state
 #define CFI_RESTORE_STATE .cfi_restore_state
 #define CFI_UNDEFINED .cfi_undefined
+#ifdef CONFIG_AS_CFI_SIGNAL_FRAME
+#define CFI_SIGNAL_FRAME .cfi_signal_frame
+#else
+#define CFI_SIGNAL_FRAME
+#endif
 
 #else
 
@@ -45,6 +50,7 @@
 #define CFI_REMEMBER_STATE      #
 #define CFI_RESTORE_STATE       #
 #define CFI_UNDEFINED   #
+#define CFI_SIGNAL_FRAME        #
 
 #endif
 
diff --git a/include/asm-x86_64/unwind.h b/include/asm-x86_64/unwind.h
index 1f6e9bfb569e..b8fa5cb7ff88 100644
--- a/include/asm-x86_64/unwind.h
+++ b/include/asm-x86_64/unwind.h
@@ -18,6 +18,7 @@ struct unwind_frame_info
 {
         struct pt_regs regs;
         struct task_struct *task;
+        unsigned call_frame:1;
 };
 
 #define UNW_PC(frame)        (frame)->regs.rip
@@ -57,6 +58,10 @@ struct unwind_frame_info
         PTREGS_INFO(r15), \
         PTREGS_INFO(rip)
 
+#define UNW_DEFAULT_RA(raItem, dataAlign) \
+        ((raItem).where == Memory && \
+         !((raItem).value * (dataAlign) + 8))
+
 static inline void arch_unw_init_frame_info(struct unwind_frame_info *info,
                                             /*const*/ struct pt_regs *regs)
 {
diff --git a/kernel/unwind.c b/kernel/unwind.c
index f69c804c8e62..3430475fcd88 100644
--- a/kernel/unwind.c
+++ b/kernel/unwind.c
@@ -603,6 +603,7 @@ int unwind(struct unwind_frame_info *frame)
 #define FRAME_REG(r, t) (((t *)frame)[reg_info[r].offs])
         const u32 *fde = NULL, *cie = NULL;
         const u8 *ptr = NULL, *end = NULL;
+        unsigned long pc = UNW_PC(frame) - frame->call_frame;
         unsigned long startLoc = 0, endLoc = 0, cfa;
         unsigned i;
         signed ptrType = -1;
@@ -612,7 +613,7 @@ int unwind(struct unwind_frame_info *frame)
 
         if (UNW_PC(frame) == 0)
                 return -EINVAL;
-        if ((table = find_table(UNW_PC(frame))) != NULL
+        if ((table = find_table(pc)) != NULL
             && !(table->size & (sizeof(*fde) - 1))) {
                 unsigned long tableSize = table->size;
 
@@ -647,7 +648,7 @@ int unwind(struct unwind_frame_info *frame)
                                                 ptrType & DW_EH_PE_indirect
                                                 ? ptrType
                                                 : ptrType & (DW_EH_PE_FORM|DW_EH_PE_signed));
-                        if (UNW_PC(frame) >= startLoc && UNW_PC(frame) < endLoc)
+                        if (pc >= startLoc && pc < endLoc)
                                 break;
                         cie = NULL;
                 }
@@ -657,16 +658,28 @@ int unwind(struct unwind_frame_info *frame)
                 state.cieEnd = ptr; /* keep here temporarily */
                 ptr = (const u8 *)(cie + 2);
                 end = (const u8 *)(cie + 1) + *cie;
+                frame->call_frame = 1;
                 if ((state.version = *ptr) != 1)
                         cie = NULL; /* unsupported version */
                 else if (*++ptr) {
                         /* check if augmentation size is first (and thus present) */
                         if (*ptr == 'z') {
-                                /* check for ignorable (or already handled)
-                                 * nul-terminated augmentation string */
-                                while (++ptr < end && *ptr)
-                                        if (strchr("LPR", *ptr) == NULL)
+                                while (++ptr < end && *ptr) {
+                                        switch(*ptr) {
+                                        /* check for ignorable (or already handled)
+                                         * nul-terminated augmentation string */
+                                        case 'L':
+                                        case 'P':
+                                        case 'R':
+                                                continue;
+                                        case 'S':
+                                                frame->call_frame = 0;
+                                                continue;
+                                        default:
                                                 break;
+                                        }
+                                        break;
+                                }
                         }
                         if (ptr >= end || *ptr)
                                 cie = NULL;
@@ -755,7 +768,7 @@ int unwind(struct unwind_frame_info *frame)
         state.org = startLoc;
         memcpy(&state.cfa, &badCFA, sizeof(state.cfa));
         /* process instructions */
-        if (!processCFI(ptr, end, UNW_PC(frame), ptrType, &state)
+        if (!processCFI(ptr, end, pc, ptrType, &state)
            || state.loc > endLoc
            || state.regs[retAddrReg].where == Nowhere
            || state.cfa.reg >= ARRAY_SIZE(reg_info)
@@ -763,6 +776,11 @@ int unwind(struct unwind_frame_info *frame)
            || state.cfa.offs % sizeof(unsigned long))
                 return -EIO;
         /* update frame */
+#ifndef CONFIG_AS_CFI_SIGNAL_FRAME
+        if(frame->call_frame
+           && !UNW_DEFAULT_RA(state.regs[retAddrReg], state.dataAlign))
+                frame->call_frame = 0;
+#endif
         cfa = FRAME_REG(state.cfa.reg, unsigned long) + state.cfa.offs;
         startLoc = min((unsigned long)UNW_SP(frame), cfa);
         endLoc = max((unsigned long)UNW_SP(frame), cfa);
@@ -866,6 +884,7 @@ int unwind_init_frame_info(struct unwind_frame_info *info,
                            /*const*/ struct pt_regs *regs)
 {
         info->task = tsk;
+        info->call_frame = 0;
         arch_unw_init_frame_info(info, regs);
 
         return 0;
@@ -879,6 +898,7 @@ int unwind_init_blocked(struct unwind_frame_info *info,
                         struct task_struct *tsk)
 {
         info->task = tsk;
+        info->call_frame = 0;
         arch_unw_init_blocked(info);
 
         return 0;
@@ -894,6 +914,7 @@ int unwind_init_running(struct unwind_frame_info *info,
                         void *arg)
 {
         info->task = current;
+        info->call_frame = 0;
 
         return arch_unwind_init_running(info, callback, arg);
 }
diff --git a/scripts/Kbuild.include b/scripts/Kbuild.include
index 7adef12a0c26..4f5ff19b992b 100644
--- a/scripts/Kbuild.include
+++ b/scripts/Kbuild.include
@@ -66,8 +66,8 @@ as-option = $(shell if $(CC) $(CFLAGS) $(1) -Wa,-Z -c -o /dev/null \
 # as-instr
 # Usage: cflags-y += $(call as-instr, instr, option1, option2)
 
-as-instr = $(shell if echo -e "$(1)" | $(AS) -Z -o astest$$$$.out \
-                   2>&1 >/dev/null ; then echo "$(2)"; else echo "$(3)"; fi; \
+as-instr = $(shell if echo -e "$(1)" | $(AS) >/dev/null 2>&1 -W -Z -o astest$$$$.out ; \
+                   then echo "$(2)"; else echo "$(3)"; fi; \
                    rm -f astest$$$$.out)
 
 # cc-option