bug#20748: 25.0.50; Seemingly random segfaults.

[Eli Zaretskii <eliz@gnu.org>]

> From: zack@apertron.net
> Date: Fri, 05 Jun 2015 19:51:43 +0000
> 
> Starting program: /home/zack/ins/bin/emacs 
> Got object file from memory but can't read symbols: File truncated.

A strange message, don't you think?

> Program received signal SIGSEGV, Segmentation fault.
> turn_on_face (f=f@entry=0xcde8f8, face_id=face_id@entry=74) at term.c:1902
> 1902	  unsigned long fg = face->foreground;
> #0  0x00000000004dc28a in turn_on_face (f=f@entry=0xcde8f8, face_id=face_id@entry=74) at term.c:1902
>         face = 0x0

Please show the output of these GDB commands:

 (gdb) frame 0
 (gdb) p f->face_cache->used

The first command assumes that "frame 0" is the one shown above, a
call to turn_on_face.  The frame number is that #0 part at the left
edge of the line.  If the next time it happens you will see the call
to turn_on_face in a different stack frame, use that frame's number
instead of zero in the first command above.

Did you build this Emacs yourself?  If so, can you also build a
non-optimized version, and see if the problem happens there?  To build
a non-optimized Emacs, configure it like this:

  CFLAGS='-O0 -g3' ./configure --enable-checking='yes,glyphs' ...

Replace "..." with the rest of your configure-time options.

Finally, do you have a lot of customizations in your ~/.emacs?  I'm
puzzled by the large number of the face ID (74) in the backtrace -- do
you customize your faces a lot, or use packages that create a lot of
faces?

Thanks.