Re: segv in 21.1 on Irix

[Eli Zaretskii]

On 11 Dec 2001, Rohan Hart wrote:

> > Could you please provide a C-level backtrace?
> 
>  warning: Warning: GDB can't find the start of the function at
> 0xfb65c1c.
> 
> Yes, that is the C level backtrace!  It might be crashing in a system
> call but neither gdb nor dbx can tell me.

Is it possible to use the Lisp backtrace to guess where in the Emacs C
dode does it crash, approximately, then put a breakpoint on the
suspect line(s), and produce a backtrace _before_ the offending
function is called?

> > Also, does it help to run Emacs in the (default) multibyte mode?  This
> > information:
> > 
> > >   default-enable-multibyte-characters: nil
> > 
> > seems to indicate that you are running "emacs --unibyte".
> 
> I'm not using --unibyte and haven't deliberately turned it off

Well, is default-enable-multibyte-characters nil in "emacs -q
--no-start-file"?  If not, something in your .emacs turns of multibyte
characters.

> though I haven't got around to installing leim.

Leim has nothing to do with the multibyteness of your session.

> > In any case, your Emacs customizations are too elaborate for anyone
> > but yourself to be able to debug that.  Is it possible to find a much
> > simpler environment where the problem happens, something much closer
> > to "emacs -q --no-site-file"?
> 
> Knew there was something I'd forgotten - the crash still occurs with
> -q --no-site-file plus a little lisp to get vm and bbdb configured to
> work together.

Please post the customizations required to get vm and bbdb to work,
the minimal set of customizations that allows to reproduce the bug.

Btw, is it certain that the problem doesn't happen without bbdb?