Re: emacs from head segfaults when run with -nw

[Eli Zaretskii <eliz@gnu.org>]

> From: Stefan Monnier <monnier@iro.umontreal.ca>
> Cc: ams@gnu.org,  emacs-devel@gnu.org
> Date: Mon, 05 Apr 2010 09:52:43 -0400
> 
> > I think I see a similar problem on MS-Windows, and I'd like to compare
> > the backtraces we see.  The segfault only happens for me in an
> > optimized build, btw.  It seems to happen when GC tries to mark the
> > value of Buffer-menu-mode-map.
> > I cannot reproduce the segfault on GNU/Linux.
> 
> I think I've seen the same crash on my GNU/Linux system.  But I've had
> difficulty reproducing it (it was in an unusal situation and I can't
> quite remember what I did to get there).

It's very elusive; it even managed to defeat "bzr bisect".  I
eventually bisected manually, and it looks like my bidi merge is the
culprit, although it beats me how that could happen, given the
evidence I posted, and given that the bidi code is not even called
once between startup and the crash.  Perhaps my merge rearranged
memory and somehow awoke this sleeping dog.  Or maybe I'm missing
something.

I posted some details of what I succeeded to find out; any ideas about
how to proceed are welcome.