Debugging with ltrace and gdb

debug.txt (text/plain), 5.64 KB, created by Lars T. Mikkelsen on 2005-03-21 12:23:08 UTC

If LANG or LC_* are set, programs seg fault inside setlocale().

$ set | grep 'LANG\|LC_'
LANG=en_DK.UTF-8
LC_CTYPE=da_DK.UTF-8
$ ltrace /bin/date 
__libc_start_main(0x8049480, 1, 0xbffff744, 0x804e8b0, 0x804e900 <unfinished ...>
getenv("_POSIX2_VERSION") = NULL
getenv("POSIXLY_CORRECT") = NULL
setlocale(6, "" <unfinished ...>
--- SIGSEGV (Segmentation fault) ---
+++ killed by SIGSEGV +++


According to GDB the program seg faults in _nl_normalize_codeset (), however,
I'm not able to get a stack trace.

$ gdb /bin/date
GNU gdb 6.3
[snip]
(gdb) run
Starting program: /bin/date 
[Thread debugging using libthread_db enabled]
[New Thread 16384 (LWP 26977)]

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 16384 (LWP 26977)]
0xb7ec71e5 in _nl_normalize_codeset () from /lib/libc.so.6
(gdb) bt
#0  0xb7ec71e5 in _nl_normalize_codeset () from /lib/libc.so.6
(gdb) 


If I break at _nl_normalize_codeset() I'm able to get the stack trace, it does,
however, look quite strange and GDB talks about a corrupt stack (why is that?).
If the program is stepped further, it will eventually seg fault.

$ gdb /bin/date
GNU gdb 6.3
[snip]
(gdb) break _nl_normalize_codeset
Function "_nl_normalize_codeset" not defined.
Make breakpoint pending on future shared library load? (y or [n]) y
Breakpoint 1 (_nl_normalize_codeset) pending.
(gdb) run
Starting program: /bin/date 
[Thread debugging using libthread_db enabled]
[New Thread 16384 (LWP 26996)]
Breakpoint 2 at 0xb7ec71a0
Pending breakpoint "_nl_normalize_codeset" resolved
[Switching to Thread 16384 (LWP 26996)]

Breakpoint 2, 0xb7ec71a0 in _nl_normalize_codeset () from /lib/libc.so.6
(gdb) bt
#0  0xb7ec71a0 in _nl_normalize_codeset () from /lib/libc.so.6
#1  0xb7ec1630 in _nl_load_locale_from_archive () from /lib/libc.so.6
#2  0xb7ec03ab in _nl_find_locale () from /lib/libc.so.6
#3  0xb7ebff44 in setlocale () from /lib/libc.so.6
#4  0x08049508 in ?? ()
#5  0x00000006 in ?? ()
#6  0x0804ea4d in _IO_stdin_used ()
#7  0xbffff7c0 in ?? ()
#8  0xbffff7c0 in ?? ()
#9  0x00000002 in ?? ()
#10 0xb7fd05c8 in ?? ()
#11 0x00000001 in ?? ()
#12 0x00000000 in ?? ()
#13 0x00000001 in ?? ()
#14 0xb8001628 in ?? ()
#15 0x00000000 in ?? ()
#16 0x00000000 in ?? ()
#17 0xbffff678 in ?? ()
#18 0x00000000 in ?? ()
#19 0x00000000 in ?? ()
#20 0x00000000 in ?? ()
#21 0x00000000 in ?? ()
#22 0x00000000 in ?? ()
#23 0x00000000 in ?? ()
#24 0x00000000 in ?? ()
#25 0xb7e5787c in __pthread_alt_unlock () from /lib/libpthread.so.0
#26 0xb7e5fc60 in __linuxthreads_initial_report_events () from /lib/libpthread.so.0
#27 0xb7e5fc68 in __pthread_multiple_threads () from /lib/libpthread.so.0
#28 0x00000000 in ?? ()
#29 0x00000000 in ?? ()
#30 0x00000000 in ?? ()
#31 0xb7eaae00 in ?? () from /lib/libc.so.6
#32 0xb7fd0000 in ?? ()
#33 0x00000000 in ?? ()
#34 0x00000000 in ?? ()
#35 0x00000000 in ?? ()
#36 0x00000000 in ?? ()
#37 0xb7e5cff4 in ?? () from /lib/libpthread.so.0
#38 0xb7fcd540 in last_environ () from /lib/libc.so.6
#39 0x00000002 in ?? ()
#40 0xbffff734 in ?? ()
#41 0xb7e54404 in pthread_mutex_unlock () from /lib/libpthread.so.0
Previous frame inner to this frame (corrupt stack?)
(gdb) step
Single stepping until exit from function _nl_normalize_codeset, 
which has no line number information.
0xb7e56b10 in __pthread_internal_tsd_address () from /lib/libpthread.so.0
(gdb) step
Single stepping until exit from function __pthread_internal_tsd_address, 
which has no line number information.
0xb7ec736c in _nl_normalize_codeset () from /lib/libc.so.6
(gdb) step
Single stepping until exit from function _nl_normalize_codeset, 
which has no line number information.

Program received signal SIGSEGV, Segmentation fault.
0xb7ec71e5 in _nl_normalize_codeset () from /lib/libc.so.6
(gdb) 


--------------------------------------------------------------------------------


If LANG or LC_* are not set, programs seg fault inside bindtextdomain().

$ unset LANG LC_CTYPE
$ set | grep 'LANG\|LC_'
$ ltrace /bin/date 
__libc_start_main(0x8049480, 1, 0xbffff774, 0x804e8b0, 0x804e900 <unfinished ...>
getenv("_POSIX2_VERSION") = NULL
getenv("POSIXLY_CORRECT") = NULL
setlocale(6, "")                                   = "C"
bindtextdomain("coreutils", "/usr/share/locale" <unfinished ...>
--- SIGSEGV (Segmentation fault) ---
+++ killed by SIGSEGV +++


Like above, GDB only shows the top stack frame on seg fault.

$ gdb /bin/date
GNU gdb 6.3
[snip]
(gdb) run
Starting program: /bin/date 
[Thread debugging using libthread_db enabled]
[New Thread 16384 (LWP 27035)]

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 16384 (LWP 27035)]
0xb7e56a64 in __pthread_internal_tsd_set () from /lib/libpthread.so.0
(gdb) bt
#0  0xb7e56a64 in __pthread_internal_tsd_set () from /lib/libpthread.so.0
(gdb) 


However, unlike above, I'm not able to get a stack trace by breaking at
__pthread_internal_tsd_set(). GDB seg faults! :-)

$ gdb /bin/date
GNU gdb 6.3
[snip]
(gdb) break __pthread_internal_tsd_set
Function "__pthread_internal_tsd_set" not defined.
Make breakpoint pending on future shared library load? (y or [n]) y
Breakpoint 1 (__pthread_internal_tsd_set) pending.
(gdb) run
Starting program: /bin/date 
[Thread debugging using libthread_db enabled]
[New Thread 16384 (LWP 27041)]
Breakpoint 2 at 0xb7e569c0
Pending breakpoint "__pthread_internal_tsd_set" resolved
[Switching to Thread 16384 (LWP 27041)]

Breakpoint 2, 0xb7e569c0 in __pthread_internal_tsd_set () from /lib/libpthread.so.0
(gdb) step
Single stepping until exit from function __pthread_internal_tsd_set, 
which has no line number information.
0xb7ec29f2 in uselocale () from /lib/libc.so.6
[snip]
(gdb) step
Single stepping until exit from function _init, 
which has no line number information.
Segmentation fault
$