Re: (ITS#3698) [devel] test035-meta fails

[quanah@stanford.edu]

--On Tuesday, May 03, 2005 11:26 PM +0200 Pierangelo Masarati 
<ando@sys-net.it> wrote:

>>> if it isn't, a SIGSEGV or anythng bad might have occurred.  In either
>>> case, I think I need further info to be able to track what might have
>>> happened.  I've repeatedly run that test under heavy load on a fresh
>>> build
>>> from HEAD for hours this afternoon and I didn't see any problem.  It's
>>> of
>>> paramount importance for me to have this and similar tests running
>>> reliably, so I'd appreciate if you can send as much information as
>>> possible.
>>
>> Program received signal SIGSEGV, Segmentation fault.
>> [Switching to LWP 3]
>> 0xfeb33218 in strlen () from /usr/lib/libc.so.1
>> (gdb) bt
>> # 0  0xfeb33218 in strlen () from /usr/lib/libc.so.1
>> # 1  0xfeb86530 in _doprnt () from /usr/lib/libc.so.1
>> # 2  0xfeb88708 in vsnprintf () from /usr/lib/libc.so.1
>> # 3  0xff3171cc in lutil_debug (debug=-33561296, level=2147483647,
>> fmt=0x121258 "[rw] %s: \"%s\" -> \"%s\"\n") at debug.c:82
>>
>> # 3  0xff3171cc in lutil_debug (debug=-33561296, level=2147483647,
>> fmt=0x121258 "[rw] %s: \"%s\" -> \"%s\"\n") at debug.c:82
>> (gdb) l
>> 82              va_start( vl, fmt );
>> 83              vsnprintf( buffer, sizeof(buffer), fmt, vl );
>> 84              buffer[sizeof(buffer)-1] = '\0';
>> 85              if( log_file != NULL ) {
>> 86                      fputs( buffer, log_file );
>> 87                      fflush( log_file );
>> 88              }
>> 89              fputs( buffer, stderr );
>> 90              va_end( vl );
>> 91      }
>
> That's way more interesting.  Looks like when debugging librewrite stuff
> I'm printing a NULL, which on Linux doesn't harm but might on Solaris.
> I'll recheck all that portion of code; yoru stack trace doesn't indicate
> where the accident occurred in back-meta.  Can you step a bit more
> backwards?  Moreover, you should be able to complete the test if you
> reduce the logs (e.g. with SLAPD_DEBUG=0).

Yeah, it passes fine with DEBUG level 1.

gdb didn't give me any more than the 3 bits of backtrace (it usually gives 
me 6).  I'll try a newer version of GDB.

--Quanah


--
Quanah Gibson-Mount
Principal Software Developer
ITSS/Shared Services
Stanford University
GnuPG Public Key: http://www.stanford.edu/~quanah/pgp.html

"These censorship operations against schools and libraries are stronger
than ever in the present religio-political climate. They often focus on
fantasy and sf books, which foster that deadly enemy to bigotry and blind
faith, the imagination." -- Ursula K. Le Guin