Re: SIGSEGV while reading vm mail

[Stefan Monnier]

RMS> It's clear from where this crashed that what happened was
RMS> something had clobbered the free list of conses.  It could have
RMS> been clobbered way down the list.  If that happened, then later
RMS> on after the previons conses had been allocated, Emacs would
RMS> crash at the next call to Fcons.

KFS> I have seen a number of bug reports from you recently with
KFS> crashes in different parts of emacs.  It seems like you are
KFS> pretty much alone in getting these crashes, so I really wonder if
KFS> there is anything special with your setup, or are you just
KFS> particularly good at finding those bugs :-)

> I do spend virtually all my time in emacs (of course, so do lots of
> others).

It could all be a single bug.  If you can easily trigger one of these bugs
(not reliably, but easily), can you try recompiling with something like

  make CFLAGS="-g -Wall -DSYNC_INPUT"

The important part is the -DSYNC_INPUT which moves the async-signal
processing to synchronous code and thus removes one source of
potential bug.  Then tell us if you can still trigger one of those SIGSEGV
crashes or not (most likely it won't make a difference).


        Stefan