bug#45117: 28.0.50; process-send-string mysteriously exiting non-locally when called from timer

["João Távora" <joaotavora@gmail.com>]

[ We've been CC-ing bug-gnu-emacs@gnu.org for a while.  My fault, the
typical CC blunder.  Wonder how debbugs was dealing with that so
gracefully tho. ]

Eli Zaretskii <eliz@gnu.org> writes:

>> From: João Távora <joaotavora@gmail.com>
>> Cc: bug-gnu-emacs@gnu.org
>> Date: Tue, 08 Dec 2020 15:56:49 +0000
>> 
>> > Can you elaborate on the evidence you found of this non-local exit?
>> 
>> It was evidenced by M-x trace-function process-send-string RET and also
>> by substituting the snippet I posted earlier with:
> few places we call sys_longjmp, and when it breaks, show the C and
> Lisp backtrace from there.

Right, so I got this setup, compiled Emacs 27.1 with all debugging
flags.  I can reproduce it, and even with GDB attached, great.  The
problem is the breakpoints.

If I set breakpoints at _all_ places where we call sys_longjmp(), I risk
tearing down my X, which I did a couple of times.

So I skip those "dangerous" breakpoints.  I'm guessing one of the
interesting loci to break is unwind_to_catch in eval.c.  Of course that
gets called every dang time a signal is thrown, so it's hard for me to
catch the precise situation, even if I set up nicely and then call M-x
redraw-display, and only then enable the breakpoint.

It breaks near immediately, and the `bt` output I get is always from
some other function that expectedly signalled an error as part of its
normal control flow.  (Yeah, maybe I shouldn't be using signals for
normal control flow, but that's another matter.)

So:

1. I have to find a way to set the unwind_to_catch() breakpoint
   conditional on some Elisp/near-elisp context, in this case something
   inside the Elisp function sly-net-send() or Fprocess_send_string.

   Do you think setting a silly global in Fprocess_send_string() and
   then checking that as the breakpoint condition would be a good idea?
   Where would I reset the flag?  Is there some C-version of
   "unwind-protect"?
   
2. The mysterious long jump may be coming from some other place.  I
   enabled a breakpoint in the sys_longjmp call in
   quit_throw_to_read_char(), but that's not it, I can reproduce the
   error and it doesn't break there.  Then there are some in image.c
   loci that I don't think matter much and the one in alloc.c which
   freezes my X.

3. I set up one of those "tracer" breakpoints for the that you once
   showed me (how did that go?)  They're going to make a LOT of noise,
   but at least we should be able to register the correct
   unwind_to_catch() context.

João