bug#11653: 23.2; Segfault with csv-mode

["Jordi Gutiérrez Hermoso" <jordigh@octave.org>]

On 8 June 2012 13:49, Eli Zaretskii <eliz@gnu.org> wrote:
>> Date: Fri, 8 Jun 2012 12:29:46 -0400
>> From: Jordi Gutiérrez Hermoso <jordigh@octave.org>
>>
>> (1) Load the following csv-mode.el:
>>
>>     http://jordi.platinum.linux.pl/emacs-bug/csv-mode.el
>>
>> (2) Open the following csv file:
>>
>>     http://jordi.platinum.linux.pl/emacs-bug/foo.csv
>>
>> (3) Align all fields: C-x h C-c C-a
>>
>> (4) Attempt to sort them: M-x csv-sort-fields <RET>
>>
>> In Emacs 23, this incurred an immediate crash. At bzr revno: 108512,
>> after attempting to move the point a little (C-e, C-a, C-p, C-n, maybe
>> C-g), I obtained the following stack trace:
>>
>>
>> Program received signal SIGSEGV, Segmentation fault.
>> 0x0000000000438b03 in handle_face_prop (it=0x7fffffffb0c0) at xdisp.c:3693
>> 3693          = (it->current.overlay_string_index >= 0
>
> I cannot reproduce this with today's trunk (revision 108257).

I can reproduce it quite reliably. I'm not sure I'm at the correct
revision, but I just tried this again at revno 108581 with the same
results.

I tried 108257 (which seems to be from May 17? Not "today"?), and also
got the same result. It seems to take a little while for the segfault
to occur. Trying a few motion keys (C-n, C-e, C-a) seems to trigger it
for me.

>> (gdb) bt
>> #0  0x0000000000438b03 in handle_face_prop (it=0x7fffffffb0c0) at xdisp.c:3693
>
> In frame 0, can you show what does the following GDB command print?

Sure thing:

    Program received signal SIGSEGV, Segmentation fault.
    0x0000000000438649 in handle_face_prop (it=0x7fffffffb180) at xdisp.c:3689
    3689		= (it->current.overlay_string_index >= 0
    (gdb) p it->current
    $1 = {
      pos = {
        charpos = 1,
        bytepos = 1
      },
      overlay_string_index = 1935,
      string_pos = {
        charpos = 0,
        bytepos = 0
      },
      dpvec_index = -1
    }

Thank you for your interest,
- Jordi G. H.