GNU bug report logs - #75632
31.0.50; igc: Crash report

Previous Next

Full log

View this message in rfc822 format

From: Eli Zaretskii <eliz <at> gnu.org>
To: Pip Cet <pipcet <at> protonmail.com>
Cc: yantar92 <at> posteo.net, 75632 <at> debbugs.gnu.org
Subject: bug#75632: 31.0.50; igc: Crash report
Date: Sun, 19 Jan 2025 07:39:34 +0200

> Date: Sat, 18 Jan 2025 20:40:47 +0000
> From: Pip Cet <pipcet <at> protonmail.com>
> Cc: 75632 <at> debbugs.gnu.org, yantar92 <at> posteo.net
> 
> "Eli Zaretskii" <eliz <at> gnu.org> writes:
> 
> > But if we use !initialized, the code which installs the variable
> > watcher will not run in the dumped Emacs, right?
> 
> Correct.  Variable watchers are symbol properties, and symbol properties
> are restored from the dump, so there is no need to add them again.

Then why do we need to do this in C and not in Lisp?  Can't we set up
the watcher as part of some preloaded Lisp file?

> >> >> IIRC, there are some loops that test Vinhibit_quit or Vquit_flag
> >> >> directly (or use QUITP); we wouldn't be able to break out of one of
> >> >> those if we only set Vinhibit_quit from what's currently
> >> >> store_user_signal_events, would we?
> >> >
> >> > Sorry, I don't follow.  Why does it matter whether we set those
> >> > variables from a signal handler or in gobble_input?
> >>
> >> We might never reach gobble_input if the variable isn't set in the
> >> signal handler: some loops check QUITP (etc.), and if there's a bug in
> >> them and Vinhibit_quit is set, the loop might never call maybe_quit.
> >
> > Then how about setting Vquit_flag and resetting Vinhibit_quit in the
> > handler, but leaving the check whether we should call the debugger to
> > gobble_input?
> 
> 1. We can't do that unconditionally.  An ordinary SIGUSR2 should not
> cause a quit, only a special event should.  So we need to check whether
> the event we received is special or not.
> 
> 2. probably_quit handles the quit flag first, then handles pending
> signals only when it is called again, by the next maybe_quit.  That
> means by the time we reach gobble_input, it's too late to call the
> debugger.  So we need to set debug_* at the same time as setting
> Vquit_flag (or rewrite probably_quit, which probably has very good
> reasons for its peculiar behavior).

OK, but could we at least make this somewhat cleaner?  Instead of
copying the name of the event aside, can we produce a C integer that
identifies the event, and store that number instead?  Accessing
integers from a signal handler should not be a problem, right?  (In
retrospect, we should have probably made debug-on-event a function,
then we could produce that number inside the function, thus avoiding
the need for watching the variable.  Maybe we should introduce the
function now and deprecate the variable?)

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.