Michael Richardson writes:
Dear Bug Team!
I have placed a uuencoded file at http://www.sandelman.ca/tmp/26.uu
It is an email file from mh-e/nmh/gnus. When I display it, it causes
Xemacs to crash repeatedly.
Maybe so, but I don't see how it could possibly cause the backtraces
you included, since there are no attachments at all in the message in
that file. Did you send the right message?
If I should report this via debian, let me know.
It would be useful to report to Debian to help other Debian users (I
don't think I've heard of this crash elsewhere). That won't help much
to fix it, though. AFAIK Debian has no patches that would affect
this. If you do report to Debian you can cite this report (it will
probably show up on the tracker as
within a few hours.
You should check that your libpng is up to date; several somewhat
recent versions are known to be buggy enough to cause memory
corruption and/or crash themselves, and there's nothing we can do
about a crash due to a library we don't maintain (except tell you not
to include that feature!) However AFAIK all known bugs in libpng have
been fixed, so if your libpng is up to date, it is most likely an
See also http://tracker.xemacs.org/XEmacs/its/issue805
21.4 can't be build with libpng 1.5.
However, since this is a crash and Lisp can't crash ;-), the problem
must be at the C level, however this:
Cannot access memory at address 0xb777e8e4
#0 0xb7761424 in ?? ()
are useless in debugging. I believe Debian provides debug packages
for all of its binary executables, so that you can load the debug
symbol table into gdb and get a usable backtrace. You'll have to ask
Debian or a Debian user about the procedure for doing that.
I also don't understand why your memory map doesn't show any evidence
of libpng.so despite this report in your Installation file:
Compiling in support for PNG images.
If you can help resolve these issues about your installation and the
email message, we can be of more help in dealing with the crash
XEmacs-Beta mailing list