bug#9006: 24.0.50; Abort in unshow_buffer/kill-buffer

[Eli Zaretskii]

> From: Stephen Berman <stephen.berman@gmx.net>
> Date: Wed, 13 Jul 2011 08:53:19 +0200
> Cc: 9006@debbugs.gnu.org
> 
> I was in the process of replying that I was well past the twelfth hour
> and Emacs was still going strong -- when Emacs crashed :-(.  
> But: I think this crash is not related to your patch to
> vertical_motion -- it happened in compute_line_metrics -- so I still
> think you should install it.

The crash happens during redisplay of mode line.

> PS: On the chance that this crash could be relevant, here are the
> circumstances and the backtrace.  I was in Gnus, had just opened your
> post again to reply, and somehow I mistakenly must have type `C-s SPC',
> because all the spaces in the Summary buffer became covered with
> lazy-highlight face.  I typed C-g two or three times -- and Emacs
> crashed hard (segfault); bt full gave no Lisp backtrace.  I have not
> been able to reproduce this.
> 
> #0  0x080735ac in compute_line_metrics (it=0xbfffb004)
>     at /data/steve/bzr/emacs/quickfixes/src/xdisp.c:16789

Please tell what source line corresponds to the PC address 0x080735ac
in your version of the sources.  It's not clear what code caused the
segfault, and in an optimized build one cannot trust line numbers
reported by the debugger in a backtrace.  You may need to disassemble
the code in GDB to see what source line crashed.

Also, please go to frame #1 (assuming that crashed session still
exists), and type "p *it.glyph_row", then post here what GDB prints.