[Date Prev][Date Next] [Chronological] [Thread] [Top]

Re: slapadd segfaults with sub,approx indexing on UTF-8 data (ITS#1215)

pmasarati@bci.it wrote:

> Stig@OpenLDAP.org wrote:
> >
> > Not sure what I was doing yesterday. I managed to reproduce the problem
> > today, and I think I've fixed it. Please try latest CVS version. Hope-
> > fully the changes haven't introduced new bugs.
> I son't think it's your fault. I was afraid it was mine,
> after I committed the back-monitor stuff, though. I checked
> your changes and I don't think you caused any change in
> such code (of course a subtle side-effect can always happen).
> The problem occurs when closing the backend: in
> libraries/libldbm/ldbm.c
> void ldbm_shutdown_env(DB_ENV *env)
> {
>         env->close( env, 0 );
> }
> is called with a NULL env.  My question is: is this correct
> (at least in TOOL mode) or is something causing the env to
> be corrupted right before shutting down the backend? Notice
> that by skipping this call when env is NULL makes all the
> tests succeeed regularly.
> Ando
> --
> Dr. Pierangelo Masarati    mailto:ando@sys-net.it
> Developer, SysNet s.n.c.   http://www.sys-net.it

This looks like it could be the cache changes I submitted recently.  Kurt
mentioned he had seen a problem when multiple LDBM databases were defined.
I will review what I did.