Jason White <jasonjgw@pacific.net.au> writes:
> I am still experiencing memory faults, but I suspect they may be more
> connected with the berkeley database than with svn proper. So far I
> haven't been able to track down the cause. Here are the symptoms:
>
> Running various svn commands (co, ci, log etc.) sometimes, for no
> predictable reason, induces a "memory fault" error message. The
> database is always in an inconsistent state thereafter and must be
> recovered.
> Example:
> [tmp_at_jdc]$ svn co file:///home/jason/svn/test
> Memory fault
>
> The closest I managed to get to this while running gdb was a hang of
> some sort, which had to be interrupted from the keyboard:
>
> Program received signal SIGINT, Interrupt.
> [Switching to Thread 1024 (LWP 1991)]
> 0x400b95c1 in __db_shalloc_4000 () from /usr/lib/libdb-4.0.so
>
> I don't have any reason to suspect any hardware problems; everything
> else on my system is running normally, with no crashes or other
> difficulties.
>
> [jason@jdc]$ uname -a
> Linux jdc 2.4.17-xfs #1 Sat Jan 19 16:47:16 EST 2002 i686
>
> Any suggestions?
Do you get a core file when you get "memory fault"? If not then you
may need to enable core files (see the ulimit command). Once you have
a core file do
$ gdb /path/to/svn core
(gdb) bt
to get a backtrace.
I'm not sure what causes "memory fault", is it a SIGBUS? I see
"2.4.17-xfs", are you using XFS? Anyone know if XFS causes db4 and/or
mmap problems? You could try running the db4 testsuite (I've never
done that).
--
Philip Martin
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Fri Aug 16 02:11:24 2002