Branko Čibej wrote on Mon, Apr 02, 2012 at 21:04:36 +0200:
> On 02.04.2012 18:46, Philip Martin wrote:
> > I am unable to run the Subversion regression tests for BDB and DAV with
> > Apache 2.4 using the event MPM (the new default MPM in 2.4). Sometimes
> > mod_dav_svn SEGVs causing tests to FAIL and sometimes the tests hang
> > with mod_dav_svn spinning in a loop. The problems go away when I switch
> > to the worker MPM (the old default MPM in 2.2).
> >
> > I'm not sure what the event MPM does that causes the problem. Here is a
> > typical SEGV:
> >
> > #0 0x00007fb2017f8e96 in svn_error_clear (err=0x2031203920312038)
> > at ../src/subversion/libsvn_subr/error.c:341
> > 341 while (err->child)
> > (gdb) p err
> > $1 = (svn_error_t *) 0x2031203920312038
> > (gdb) p err[0]
> > Cannot access memory at address 0x2031203920312038
>
> Looks like the DB_REGISTER dance biting us yet again. Pool lifetimes
> most likely, it's a tricky bit of code. Hard to say for certain since I
> don't know how the event MPM works.
http://httpd.apache.org/docs/current/mod/event.html
"""To solve [the 'keep alive problem'], this MPM uses a dedicated thread
to handle both the Listening sockets, all sockets that are in a Keep
Alive state, and sockets where the handler and protocol filters have
done their work and the only remaining thing to do is send the data to
the client"""
>
> -- Brane
>
Received on 2012-04-02 21:30:41 CEST