Re: Weirdness while testing 1.4.0-rc2 (svnserve processes that don't die)

Philip Martin <philip@codematters.co.uk> writes:

> $ svnadmin create repo --fs-type bdb
> $ valgrind -q svnserve -Xr.
> $ svn ls svn://localhost/repo
>
> valgrind shows svnserve failing:
>
> ==2314== Invalid read of size 4
> ==2314== at 0x4247B22: pthread_mutex_lock (mutex.c:102)
> ==2314== by 0x41EF6FF: apr_thread_mutex_lock (thread_mutex.c:92)
> ==2314== by 0x43F0D11: acquire_cache_mutex (env.c:443)
> ==2314== by 0x43F0C8D: svn_fs_bdb__close (env.c:579)
> ==2314== by 0x43F97B2: cleanup_fs (fs.c:185)
> ==2314== by 0x43F9846: cleanup_fs_apr (fs.c:291)
> ==2314== by 0x41F211E: run_cleanups (apr_pools.c:2034)
> ==2314== by 0x41F143B: pool_clear_debug (apr_pools.c:1372)
> ==2314== by 0x41F1624: pool_destroy_debug (apr_pools.c:1457)
> ==2314== by 0x41F1423: pool_clear_debug (apr_pools.c:1369)
> ==2314== by 0x41F1624: pool_destroy_debug (apr_pools.c:1457)
> ==2314== by 0x41F16EC: apr_pool_destroy_debug (apr_pools.c:1499)

I don't get that error if I revert r20218:

  r20218 | brane | 2006-06-22 15:38:48 +0100 (Thu, 22 Jun 2006) | 16 lines

  Make sure BDB environments always get closed even when the environment
  cache is destroyed before the last of the FS objects.

I think that's the DB_TXN_NOSYNC lost commits fix.

-- 
Philip Martin
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org