Andy Piper <andyp(a)bea.com&gt; writes: > Try and get a C stacktrace for the hang (gdb) where #0 0xfec18224 in _poll () from /usr/lib/libc.so.1 #1 0xfebcb808 in select () from /usr/lib/libc.so.1 #2 0xfef1bdc0 in _XWaitForReadable () from /opt/X11R5/lib/libX11.so.4 #3 0xfef1bbd8 in _XRead () from /opt/X11R5/lib/libX11.so.4 #4 0xfef265fc in _XEventsQueued () from /opt/X11R5/lib/libX11.so.4 #5 0xfef35ae4 in XEventsQueued () from /opt/X11R5/lib/libX11.so.4 #6 0x1daed0 in emacs_Xt_quit_p () at event-Xt.c:2829 #7 0xe1130 in event_stream_quit_p () at event-stream.c:596 #8 0x1b13f4 in check_what_happened () at signal.c:497 #9 0x97680 in Ffuncall (nargs=1, args=0xffbee748) at eval.c:3474 #10 0x9bc58 in call1 (fn=5050196, arg0=11554280) at eval.c:4183 #11 0xdc954 in execute_internal_event (event=11554280) at event-stream.c:2939 #12 0xdf544 in Fdispatch_event (event=11554280) at event-stream.c:4170 #13 0x7009c in Fcommand_loop_1 () at cmdloop.c:583 #14 0x70334 in command_loop_1 (dummy=5393664) at cmdloop.c:494 #15 0x934f8 in condition_case_1 (handlers=3018748, bfun=0x702e8 <command_loop_1>, barg=3254276, hfun=0x70394 <cmd_error>, harg=3254276) at eval.c:1651 #16 0x704a4 in command_loop_2 (dummy=3254276) at cmdloop.c:256 #17 0x9c398 in internal_catch (tag=3334340, func=0x70458 <command_loop_2>, arg=3254276, threw=0x0) at eval.c:1317 #18 0x6f7b4 in initial_command_loop (load_me=3254276) at cmdloop.c:305 #19 0x8edc0 in sort_args (argc=3064832, argv=0xffbeedac) at emacs.c:2253 If you look at truss while moving the mouse or typing you see this sort of thing. There is no display activity. poll(0xFFBEE1A8, 1, -1) (sleeping...) Received signal #22, SIGPOLL, in poll() [caught] siginfo: SIGPOLL POLL_IN fd=0 band=0 poll(0xFFBEE1A8, 1, -1) Err#4 EINTR setcontext(0xFFBEDE90) Received signal #22, SIGPOLL [caught] siginfo: SIGPOLL POLL_IN fd=0 band=0 setcontext(0xFFBEDE90) Received signal #22, SIGPOLL, in poll() [caught] siginfo: SIGPOLL POLL_IN fd=0 band=0 poll(0xFFBEE1A8, 1, -1) Err#4 EINTR setcontext(0xFFBEDE90) Received signal #22, SIGPOLL, in poll() [caught] siginfo: SIGPOLL POLL_IN fd=0 band=0 Anything else? > > andy > > At 02:42 PM 4/26/01 -0400, John M. Adams wrote: > >Help, I'm dyin' out here. > > > >I used to use a ga release 21 version of xemacs until I upgraded to > >solaris8. At that time, I started getting random hangs. Since then > >I've used many different betas and have the same experience. I > >reported a bug a while back. > > > >It's easy to reproduce this over the course of a day or two. If > >someone could give me a few hints, I'd be happy to poke around in the > >broken running image. > > > >I'm using xemacs-21.2-b45 which I know is behind. If we think this is > >fixed, fine. Otherwise, I'll be happy to test the latest beta. > > > >Thanks alot. > > > >This problem REALLY hurts. I'd like to help as much as possible. > > > >-- > >John M. Adams -- John M. Adams

Andy Piper <andyp(a)bea.com&gt; writes: > Ben, Gunnar, > > Is this related to the X quit stuff you were looking at? Yes, in the sense that it shows a bug in the X quit handling. Not sure what exactly is happening here, though. emacs_Xt_quit_p() should certainly never block. I had a quick look at Ben's suggestions of fixing drain_X_queue() and I agree with it but haven't gotten around to implementing it yet. I probably won't have much time in the next few weeks either as I have to go to a couple of conferences in the States. Gunnar