bug#13086: 24.2.50; Emacs seems to hang at w32proc.c:1126

[Stephen Powell]

Eli Zaretskii <eliz@gnu.org> writes:

> I also found some potential problem in w32proc.c, and fixed that in
> revision 111132 on the trunk.  Stephen, please try the latest code and
> see if the problem persists.

OK, I tried that revision with the same problem.

I don't really know what I'm doing, but one odd thing I noticed while
playing around with Paul's suggestion to set a breakpoint on
delete_process:

1. Set a breakpoint on process.c:808. Print p.pid and do a xbacktrace.
2. Run gnus.
3. Gnus uses imap.el to get mail from my host.
4. Imap.el calls delete-process from imap-close

$1 = 1356
"delete-process" (0x88bf48)
"imap-close" (0x88c248)
"mail-source-fetch-imap" (0x88c584)

5. Imap.el calls delete-process to delete the same pid from
imap-sentinel

$2 = 1356
"delete-process" (0x88b928)
"imap-sentinel" (0x88bc24)
"delete-process" (0x88bf48)
"imap-close" (0x88c248)
"mail-source-fetch-imap" (0x88c584)

6. Let gnus sit for about a minute.  The error is signalled with the
same pid

Breakpoint 1, terminate_due_to_signal (sig=22,
backtrace_limit=2147483647) at emacs.c:314
314       signal (sig, SIG_DFL);
(gdb) up 2
#2  0x0114a8db in get_child_status (child=1356, status=0x0, options=1,
interruptible=false) at sysdep.c:294
294           eassert (errno == EINTR);