S390: Add cfi information for start routines in order to stop unwinding.
Commit Message
Hi,
GDB failed to detect the outermost frame while showing the backtrace
within a thread:
Backtrace stopped: previous frame identical to this frame (corrupt stack?)
Before this patch, the start routines like thread_start had
no cfi information. GDB is then using the prologue unwinder if
no cfi information is available. This unwinder tries to unwind r15
and stops e.g. if r15 was updated or on some jump-instructions.
On older glibc-versions (before commit "Remove cached PID/TID in clone"
c579f48edba88380635ab98cb612030e3ed8691e), the thread_start function
used such a jump-instruction and GDB did not fail with an error.
This patch adds cfi information for _start, thread_start and
__makecontext_ret and marks r14 as undefined which marks the frame
as outermost frame and GDB stops the backtrace. Also tested
different gcc versions in order to test _Unwind_Backtrace() in libgcc
as this is used by backtrace() in glibc.
Bye
Stefan
ChangeLog:
* sysdeps/s390/s390-64/start.S (_start):
Add cfi information for r14.
* sysdeps/s390/s390-32/start.S: (_start): Likewise
* sysdeps/unix/sysv/linux/s390/s390-64/clone.S
(thread_start): Likewise.
* sysdeps/unix/sysv/linux/s390/s390-32/clone.S
(thread_start): Likewise.
* sysdeps/unix/sysv/linux/s390/s390-64/__makecontext_ret.S
(__makecontext_ret): Likewise.
* sysdeps/unix/sysv/linux/s390/s390-32/__makecontext_ret.S
(__makecontext_ret): Likewise.
Comments
On 11/14/2017 10:52 AM, Stefan Liebler wrote:
> Hi,
>
> GDB failed to detect the outermost frame while showing the backtrace
> within a thread:
> Backtrace stopped: previous frame identical to this frame (corrupt stack?)
>
> Before this patch, the start routines like thread_start had
> no cfi information. GDB is then using the prologue unwinder if
> no cfi information is available. This unwinder tries to unwind r15
> and stops e.g. if r15 was updated or on some jump-instructions.
>
> On older glibc-versions (before commit "Remove cached PID/TID in clone"
> c579f48edba88380635ab98cb612030e3ed8691e), the thread_start function
> used such a jump-instruction and GDB did not fail with an error.
>
> This patch adds cfi information for _start, thread_start and
> __makecontext_ret and marks r14 as undefined which marks the frame
> as outermost frame and GDB stops the backtrace. Also tested
> different gcc versions in order to test _Unwind_Backtrace() in libgcc
> as this is used by backtrace() in glibc.
>
> Bye
> Stefan
>
> ChangeLog:
>
> * sysdeps/s390/s390-64/start.S (_start):
> Add cfi information for r14.
> * sysdeps/s390/s390-32/start.S: (_start): Likewise
> * sysdeps/unix/sysv/linux/s390/s390-64/clone.S
> (thread_start): Likewise.
> * sysdeps/unix/sysv/linux/s390/s390-32/clone.S
> (thread_start): Likewise.
> * sysdeps/unix/sysv/linux/s390/s390-64/__makecontext_ret.S
> (__makecontext_ret): Likewise.
> * sysdeps/unix/sysv/linux/s390/s390-32/__makecontext_ret.S
> (__makecontext_ret): Likewise.
Committed
commit f078ec7c04c5b82b98002946d6d7f388187a2e49
Author: Stefan Liebler <stli@linux.vnet.ibm.com>
Date: Fri Nov 10 09:40:03 2017 +0100
S390: Add cfi information for start routines in order to stop unwinding.
GDB failed to detect the outermost frame while showing the backtrace
within a thread:
Backtrace stopped: previous frame identical to this frame (corrupt stack?)
Before this patch, the start routines like thread_start had no cfi information.
GDB is then using the prologue unwinder if no cfi information is available.
This unwinder tries to unwind r15 and stops e.g. if r15 was updated or
on some jump-instructions.
On older glibc-versions (before commit "Remove cached PID/TID in clone"
c579f48edba88380635ab98cb612030e3ed8691e), the thread_start function used
such a jump-instruction and GDB did not fail with an error.
This patch adds cfi information for _start, thread_start and __makecontext_ret
and marks r14 as undefined which marks the frame as outermost frame and GDB
stops the backtrace. Also tested different gcc versions in order to test
_Unwind_Backtrace() in libgcc as this is used by backtrace() in glibc.
ChangeLog:
* sysdeps/s390/s390-64/start.S (_start): Add cfi information for r14.
* sysdeps/s390/s390-32/start.S: (_start): Likewise
* sysdeps/unix/sysv/linux/s390/s390-64/clone.S
(thread_start): Likewise.
* sysdeps/unix/sysv/linux/s390/s390-32/clone.S
(thread_start): Likewise.
* sysdeps/unix/sysv/linux/s390/s390-64/__makecontext_ret.S
(__makecontext_ret): Likewise.
* sysdeps/unix/sysv/linux/s390/s390-32/__makecontext_ret.S
(__makecontext_ret): Likewise.
@@ -34,6 +34,8 @@
License along with the GNU C Library; if not, see
<http://www.gnu.org/licenses/>. */
+#include <sysdep.h>
+
/*
This is the canonical entry point, usually the first thing in the text
segment. Most registers' values are unspecified, except for:
@@ -57,6 +59,10 @@
.globl _start
.type _start,@function
_start:
+ cfi_startproc
+ /* Mark r14 as undefined in order to stop unwinding here! */
+ cfi_undefined (r14)
+
/* Check if the kernel provides highgprs facility if needed by
the binary. */
@@ -188,6 +194,7 @@ _start:
/* crash if __libc_start_main returns */
.word 0
+ cfi_endproc
.Llit:
#ifndef PIC
.L1: .long __libc_csu_init
@@ -34,6 +34,8 @@
License along with the GNU C Library; if not, see
<http://www.gnu.org/licenses/>. */
+#include <sysdep.h>
+
/*
This is the canonical entry point, usually the first thing in the text
segment. Most registers' values are unspecified, except for:
@@ -57,6 +59,9 @@
.globl _start
.type _start,@function
_start:
+ cfi_startproc
+ /* Mark r14 as undefined in order to stop unwinding here! */
+ cfi_undefined (r14)
/* Load argc and argv from stack. */
la %r4,8(%r15) # get argv
lg %r3,0(%r15) # get argc
@@ -91,6 +96,8 @@ _start:
/* Crash if __libc_start_main returns. */
.word 0
+ cfi_endproc
+
/* Define a symbol for the first piece of initialized data. */
.data
.globl __data_start
@@ -17,15 +17,9 @@
#include <sysdep.h>
-/* We do not want .eh_frame info so that __makecontext_ret stops unwinding
- if backtrace was called within a context created by makecontext. (There
- is also no .eh_frame info for _start or thread_start.) */
-#undef cfi_startproc
-#define cfi_startproc
-#undef cfi_endproc
-#define cfi_endproc
-
ENTRY(__makecontext_ret)
+ /* Mark r14 as undefined in order to stop unwinding here! */
+ cfi_undefined (r14)
basr %r14,%r7
ltr %r8,%r8 /* Check whether uc_link is 0. */
jz 1f
@@ -54,12 +54,16 @@ error:
PSEUDO_END (__clone)
thread_start:
+ cfi_startproc
+ /* Mark r14 as undefined in order to stop unwinding here! */
+ cfi_undefined (r14)
/* fn is in gpr 1, arg in gpr 0 */
lr %r2,%r0 /* set first parameter to void *arg */
ahi %r15,-96 /* make room on the stack for the save area */
xc 0(4,%r15),0(%r15)
basr %r14,%r1 /* jump to fn */
DO_CALL (exit, 1)
+ cfi_endproc
libc_hidden_def (__clone)
weak_alias (__clone, clone)
@@ -17,15 +17,9 @@
#include <sysdep.h>
-/* We do not want .eh_frame info so that __makecontext_ret stops unwinding
- if backtrace was called within a context created by makecontext. (There
- is also no .eh_frame info for _start or thread_start.) */
-#undef cfi_startproc
-#define cfi_startproc
-#undef cfi_endproc
-#define cfi_endproc
-
ENTRY(__makecontext_ret)
+ /* Mark r14 as undefined in order to stop unwinding here! */
+ cfi_undefined (r14)
basr %r14,%r7
ltgr %r8,%r8 /* Check whether uc_link is 0. */
jz 1f
@@ -55,12 +55,16 @@ error:
PSEUDO_END (__clone)
thread_start:
+ cfi_startproc
+ /* Mark r14 as undefined in order to stop unwinding here! */
+ cfi_undefined (r14)
/* fn is in gpr 1, arg in gpr 0 */
lgr %r2,%r0 /* set first parameter to void *arg */
aghi %r15,-160 /* make room on the stack for the save area */
xc 0(8,%r15),0(%r15)
basr %r14,%r1 /* jump to fn */
DO_CALL (exit, 1)
+ cfi_endproc
libc_hidden_def (__clone)
weak_alias (__clone, clone)