[syslog-ng] syslog-ng af_sql driver experiencing segfault
Balazs Scheidler
bazsi at balabit.hu
Sun Feb 6 10:34:41 CET 2011
On Fri, 2011-02-04 at 15:27 -0800, Matthew Hall wrote:
> Program received signal SIGSEGV, Segmentation fault.
> [Switching to Thread 0xf7b12b90 (LWP 23608)]
> 0x00c5c7dc in memcpy () from /lib/libc.so.6
> (gdb) bt
> #0 0x00c5c7dc in memcpy () from /lib/libc.so.6
> #1 0xf7e21fed in g_string_insert_len () from /home/y/lib/libglib-2.0.so.0
> #2 0xf7e223e8 in g_string_append_len () from /home/y/lib/libglib-2.0.so.0
> #3 0xf7edc1d6 in result_append (result=0x8cc1780, sstr=0x910ffaf "", len=3160627, escape=0) at templates.c:213
> #4 0xf7ede535 in log_template_append_format_with_context (self=0x8ba0098, messages=0xf7b12204, num_messages=1, opts=0x8bb8aec, tz=1, seq_num=945, result=0x8cc1780) at templates.c:1058
> #5 0xf7ede83f in log_template_append_format (self=0x8ba0098, lm=0x90d93b0, opts=0x8bb8aec, tz=1, seq_num=945, result=0x8cc1780) at templates.c:1109
> #6 0xf7ede898 in log_template_format (self=0x8ba0098, lm=0x90d93b0, opts=0x8bb8aec, tz=1, seq_num=945, result=0x8cc1780) at templates.c:1116
> #7 0xf7b459e6 in afsql_dd_insert_db (self=0x8bb8a58) at afsql.c:681
> #8 0xf7b45fd9 in afsql_dd_database_thread (arg=0x8bb8a58) at afsql.c:823
> #9 0xf7ed1405 in worker_thread_func (st=0x8ba4238) at misc.c:593
> #10 0xf7e28c51 in ?? () from /home/y/lib/libglib-2.0.so.0
> #11 0x08ba4238 in ?? ()
> #12 0x08ba4248 in ?? ()
> #13 0x00000005 in ?? ()
> #14 0xf7e28b06 in ?? () from /home/y/lib/libglib-2.0.so.0
> #15 0x00d352c6 in ?? () from /lib/libpthread.so.0
> #16 0x00000000 in ?? ()
> (gdb)
>
> Looks like I am running into problems with the database code.
>
> I'll investigate but it's going to take a while so I hope somebody else might have an idea.
Not nice. Can you send me the core file and the binaries in private? The
length of the string to be appended to the result certainly looks
suspicios being 3160627 bytes.
Can you show me the template that is being expanded here?
The message syslog-ng is trying to operate on is probably bogus, but in
order to diagnose I'd probably need the core file.
--
Bazsi
More information about the syslog-ng
mailing list