I was messing around with x11-terms/evilvte-0.5.1 and resizing its window several times over a few seconds when openbox died on SIGSEGV. I reproduced it a few times with the stock ebuild and no savedconfig; it seems to happen with about 25% probability on any mouse-resize, and perhaps more often when using the gtk resize-handle-thingy inside the window, as opposed to the handle on openbox's frame. Resizing slowly seems not to trigger it as much. I tracked the problem back to moveresize_end (openbox/moveresize.c:309) being called with moveresize_client == NULL, which causes client_configure (openbox/client.c:3104) to be called with self == NULL, which causes a segfault when initializing the fdecor variable there. I'll attach a bandaid patch I made which might be helpful. It's entirely possible this is evilvte's fault, but I think openbox shouldn't die because of a buggy app. What now? Is there any way I can help debug this? Reproducible: Sometimes Steps to Reproduce: 1. Run emerge -av1 =x11-terms/evilvte-0.5.1 2. Run evilvte under openbox 3. Drag the bottom right window corner of evilvte around and release 4. Repeat 3 quickly Actual Results: openbox receives SIGSEGV, which brings down X and all other programs using it Expected Results: openbox doesn't crash
Created attachment 330696 [details, diff] resize-crash.patch
(In reply to comment #1) > Created attachment 330696 [details, diff] [details, diff] > bandaid patch Try the stable version first. _p20111019 is a snapshot so it could be broken
It's also very unlikely for me to patch openbox with code that hasn't been accepted upstream first. It's better if you submit your patch in the upstream bugzilla http://openbox.org/wiki/Openbox:Contribute
(In reply to comment #2) > Try the stable version first. _p20111019 is a snapshot so it could be broken Well, it's keyworded stable on x86; I thought it was a gentoo-local patchset, since the ebuild points at http://dev.gentoo.org/~hwoarang/distfiles/ I've reproduced it with =x11-wm/openbox-3.5.0-r1 so I'll take this upstream: (gdb) bt #0 0xb777f424 in __kernel_vsyscall () #1 0xb71b62cf in __GI_raise (sig=6) at ../nptl/sysdeps/unix/sysv/linux/raise.c:64 #2 0xb71b7b65 in __GI_abort () at abort.c:91 #3 0xb77288c2 in sighandler (sig=11) at obt/signal.c:281 #4 <signal handler called> #5 client_configure (self=0x0, x=168, y=148, w=770, h=477, user=1, final=1, force_reply=0) at openbox/client.c:3057 #6 0x0807fe14 in moveresize_end (cancel=1) at openbox/moveresize.c:333 #7 0x08070980 in event_handle_client (e=0xbfa60a3c, client=0x813c378) at openbox/event.c:1491 #8 event_process (ec=0xbfa60ba0, data=0x0) at openbox/event.c:625 #9 0xb77297cc in event_read (source=0x813b710, callback=0, data=0x0) at obt/xqueue.c:338 #10 0xb74c2566 in g_main_context_dispatch () from /usr/lib/libglib-2.0.so.0 #11 0xb74c2905 in ?? () from /usr/lib/libglib-2.0.so.0 #12 0xb74c2d4b in g_main_loop_run () from /usr/lib/libglib-2.0.so.0 #13 0x080565dc in main (argc=1, argv=0xbfa60e24) at openbox/openbox.c:378 (gdb) up 5 #5 client_configure (self=0x0, x=168, y=148, w=770, h=477, user=1, final=1, force_reply=0) at openbox/client.c:3057 3057 guint fdecor = self->frame->decorations;
Did you try 3.5.0_p20121006 as well?
(In reply to comment #5) > Did you try 3.5.0_p20121006 as well? I tried both =x11-wm/openbox-3.5.0_p20121006 and =x11-wm/openbox-9999 (the git master) and was unable to reproduce it with either. I guess this means we need to move to a newer snapshot? Upstream bug is https://bugzilla.icculus.org/show_bug.cgi?id=5809 though there's nothing new there right now.
(In reply to comment #6) > (In reply to comment #5) > > Did you try 3.5.0_p20121006 as well? > I tried both =x11-wm/openbox-3.5.0_p20121006 and =x11-wm/openbox-9999 (the > git master) and was unable to reproduce it with either. I guess this means > we need to move to a newer snapshot? > > Upstream bug is https://bugzilla.icculus.org/show_bug.cgi?id=5809 though > there's nothing new there right now. So there is nothing for upstream to fix it if it works in the latest ebuilds. We just need to stabilize a newer version. Please close the upstream bug so they wont bother investigating (point them to this bug if you want).
Arches, please stabilize =x11-wm/openbox-3.5.0_p20121006
Tested amd64: works fine here. Please, fix bug 445108 at the same time. It will be fast to do :-)
(In reply to comment #7) > Please close the upstream bug Done. Thank you for your time. (FWIW I've been using the git master (which seems to be only 1 day newer than that snapshot) with no problems on x86.)
Stable for HPPA.
arm stable
(In reply to comment #9) > Tested amd64: works fine here. > > Please, fix bug 445108 at the same time. It will be fast to do :-) amd64 done. Thanks Vicente
stable ppc ppc64
x86 stable
alpha/sparc stable