Re: svn commit: rev 2521 - trunk/subversion/libsvn_client

Ben Collins-Sussman <sussman@collab.net> writes:

> What I'm getting, however, is a segfault in unlock_dirs() at the very
> end of the commit processing. First, it tries to svn_wc_adm_close()
> the parent access baton... this access baton has a children hash with
> one child in it, so we recurse on the child access baton. But the
> child access baton is total garbage.
>
> I'm guessing that some kind of pool is being freed too early or
> something? Can you reproduce this?

I see a subpool being cleared at libsvn_client/commit.c:942, that's
probably it. Putting my release-manager hat on (hmm, it doesn't seem
to fit...) this code doesn't yet fix any locking bugs, so alpha is
probably not the place to test it :-/

[I think I'll add a pool cleanup handler to fix the problem.]

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