Re: svn commit: r932367 - /subversion/trunk/subversion/libsvn_ra_neon/lock.c

[Philip Martin <ph...@wandisco.com>]

philip@apache.org writes:

> Author: philip
> Date: Fri Apr  9 11:49:40 2010
> New Revision: 932367
>
> URL: http://svn.apache.org/viewvc?rev=932367&view=rev
> Log:
> Fix a SEGV in lock_tests.py 37 when using neon.

I started seeing this when I using apr[util] with pool debugging
enabled.  The test also failes using serf but I don't know whether our
use of serf is wrong or whether the problem is in serf itself:

==3579== Invalid read of size 4
==3579==    at 0x9C86838: serf_bucket_mem_free (allocator.c:223)
==3579==    by 0x9C859F7: serf_response_destroy_and_data (response_buckets.c:86)
==3579==    by 0x9C83599: cancel_request (context.c:437)
==3579==    by 0x9C83A24: serf_connection_close (context.c:1249)
==3579==    by 0x9C83B38: clean_conn (context.c:184)
==3579==    by 0x5E1B96D: pool_clear_debug (apr_pools.c:2314)
==3579==    by 0x5E1BBDD: pool_destroy_debug (apr_pools.c:1600)
==3579==    by 0x5E1B947: pool_clear_debug (apr_pools.c:1512)
==3579==    by 0x5E1BBDD: pool_destroy_debug (apr_pools.c:1600)
==3579==    by 0x413E41: main (main.c:2335)
==3579==  Address 0xabd6810 is 32 bytes inside a block of size 64 free'd

This reminds me of 

http://mail-archives.apache.org/mod_mbox/subversion-dev/201002.mbox/%3C878wb8oria.fsf_-_@stat.home.lan%3E

If someone could raise this on the serf list that would save me from
having to create an account to raise a bug report :)

-- 
Philip