52446 2004-05-30 05:56 0000 Kino segfaults on startup with both gcc 3.3.3-r5 and 3.4.0-r4 2005-01-16 04:17:29 0000 1 1 1 Unclassified Gentoo Linux Applications unspecified AMD64 Linux RESOLVED FIXED P2 normal --- 1 sr@sander.yi.org amd64@gentoo.org m.debruijne@matrict.nl sgtphou@fire-eyes.org sr@sander.yi.org 2004-05-30 05:56:50 0000 kino segfaults on startup. It doesn't seem to matter whether it is compiled with 3.3.3-r5 or 3.4.0-r4. According to the strace output the same problem occurs Reproducible: Always Steps to Reproduce: 1. 2. 3. emerge -pv kino: [ebuild R ] media-video/kino-0.7.1 -quicktime 0 kB last few lines of strace kino: lseek(15, 0, SEEK_SET) = 0 rt_sigprocmask(SIG_BLOCK, NULL, [RTMIN], 8) = 0 read(15, "\377\330\377\340\0\20JFIF\0\1\1\1\0H\0H\0\0\377\376\0\27"..., 65536) = 5132 read(15, "", 57344) = 0 brk(0) = 0x552b399000 brk(0x552b3c3000) = 0x552b3c3000 close(15) = 0 munmap(0x2a9a5b6000, 4096) = 0 write(2, "> Kino Common being built", 25> Kino Common being built) = 25 write(2, "\n", 1 ) = 1 write(2, "> Creating page editor", 22> Creating page editor) = 22 write(2, "\n", 1 ) = 1 --- SIGSEGV (Segmentation fault) @ 0 (0) --- write(3, "\233\21\3\0\3\0\240\1H\0\0\0\233\24\25\0\3\0\240\1\1\0"..., 4756) = 47 56 write(3, " \21\2\0\0\0\0\0", 8) = 8 write(3, "+\21\1\0", 4) = 4 read(3, "\1\1\210\0\0\0\0\0\3\0\0\1\0\0\0\0\0\0\0\0\0\0\0\0X\v\326"..., 32) = 32 fork() = 20208 wait4(20208, [WIFEXITED(s) && WEXITSTATUS(s) == 0], 0, NULL) = 20208 --- SIGCHLD (Child exited) @ 0 (0) --- rt_sigreturn(0x552b03e950) = 20208 _exit(1) = ? kugelfang@gentoo.org 2004-05-30 13:10:45 0000 Interesting... I managed to create a patch that comes around the SIGSEGV. They are caused my bad coding: GDKImage{Filters,Transitions}Repository doesnt set its member "selected_{filter,transition}" to NULL. I will attach a patch for this. On the other side, after eliminating this SIGSEGVs, a SIG32 emerge and its backtrace looks very much like the backtrace of the nautilus/gftp SIG32. Program received signal SIG32, Real-time event 32. 0x0000002a95eb6dfe in pthread_getconcurrency () from /lib/libpthread.so.0 (gdb) bt #0 0x0000002a95eb6dfe in pthread_getconcurrency () from /lib/libpthread.so.0 #1 0x0000002a95eb6c80 in pthread_getconcurrency () from /lib/libpthread.so.0 #2 0x0000002a95eb65b7 in pthread_create () from /lib/libpthread.so.0 #3 0x00000000004b579f in Storyboard::redraw() (this=0xfdcd80) at storyboard.cc:742 #4 0x0000000000480cbd in PageEditor::ResetBar() (this=0xce27f0) at page_editor.cc:1334 #5 0x0000000000480be8 in PageEditor::DrawBar(int) (this=0xce27f0, currentFrame=-1) at page_editor.cc:1308 #6 0x000000000047e62a in PageEditor::showFrame(int, int) (this=0xce27f0, i=-1, no_audio=1) at page_editor.cc:288 #7 0x000000000047adcf in KinoCommon::moveToFrame(int) (this=0xce0340, frame=-1) at kino_common.cc:1183 #8 0x00000000004785e7 in KinoCommon::setCurrentPage(int) (this=0xce0340, page=0) at kino_common.cc:405 #9 0x0000002a967f1381 in _gtk_marshal_VOID__POINTER_UINT (closure=0xcdad00, return_value=0x8, n_param_values=2147483647, param_values=0x7fbfffee20, invocation_hint=0xfea020, marshal_data=0x0) at gtkmarshalers.c:2075 #10 0x0000002a9786975f in g_closure_invoke (closure=0xcdad00, return_value=0x0, n_param_values=3, param_values=0x7fbfffee20, invocation_hint=0x7fbfffece0) at gclosure.c:437 #11 0x0000002a9787b43e in signal_emit_unlocked_R (node=0xa17770, detail=0, instance=0xc3acb0, emission_return=0x0, instance_and_params=0x7fbfffee20) at gsignal.c:2436 #12 0x0000002a9787a34e in g_signal_emit_valist (instance=0xc3acb0, signal_id=3221220896, detail=0, var_args=0x7fbfffefe0) at gsignal.c:2195 #13 0x0000002a9787a6a3 in g_signal_emit (instance=0x7fbfffe970, signal_id=8, detail=2147483647) at gsignal.c:2239 #14 0x0000002a968041ae in gtk_notebook_switch_page (notebook=0xc3acb0, page=0xc476a0, page_num=0) at gtknotebook.c:3866 #15 0x0000000000430e32 in main (argc=1, argv=0x7fbffff188) at main.c:159 kugelfang@gentoo.org 2004-05-30 14:03:39 0000 Created an attachment (id=32337) Preliminary patch to allow to get the nautilus like SIG32 instead SIGSEGVs for bad kino source code herbs@gentoo.org 2004-06-10 03:48:11 0000 seems that people are getting this same crash on x86 - see bug #50572 here's another backtrace if it's any help: #0 0x00000000004a58e9 in GDKImageTransitionRepository::SelectionChange() ( this=0xd00238) at image_transitions.cc:505 #1 0x00000000004a59ec in GDKImageTransitionRepository::Initialise(_GtkOptionMenu*, _GtkBin*) (this=0xd00238, menu=0xb10510, container=0x0) at image_transitions.cc:488 #2 0x0000000000494ad1 in PageMagick (this=0xcfd9b0, common=0x98a970) at page_magick.cc:1365 #3 0x000000000046d9f3 in KinoCommon (this=0xcf4220, widget=0xc1ae00) at kino_common.cc:103 #4 0x0000000000496643 in kinoInitialise (widget=0xc1ae00) at commands.cc:148 #5 0x000000000042f6af in main (argc=1, argv=0x7fbffff128) at main.c:157 kugelfang@gentoo.org 2004-06-10 11:36:23 0000 Herbie, did you use my patch ? herbs@gentoo.org 2004-06-11 07:01:16 0000 No, that was without your patch. Just re-emerged kino and it now seems to be working. I got no sigsegv or sig32's and it ran perfectly. Not too sure what I've changed on my system that could have affected this. I'm using gcc-3.4 btw. kugelfang@gentoo.org 2004-07-03 05:27:08 0000 I still get those SIG32s. I will mail the kino developers about it. dan@dennedy.org 2004-07-12 17:24:46 0000 As the kino developer, I just want to confirm that there is a known bug in Kino 0.7.1, which Danny's patch addresses. We have had the fix in our CVS for some time, but it is a little different: I use a member initializer in the constructor: GDKImageTransitionRepository::GDKImageTransitionRepository() : selected_transition( 0 ) and GDKImageFilterRepository::GDKImageFilterRepository() : selected_filter( 0 ) I am working on a 0.7.2 release, so I am trying to reproduce the bug on his machine. So far, with the slowness of ssh X11 forwarding over the net, it is not reproducing. kugelfang@gentoo.org 2004-07-14 06:32:47 0000 Ok, those SIG32s only get thrown when you run kino via gdb. Seems to be a bug in gdb that is already being workd on. Patch is in CVS by now. jhuebel@gentoo.org 2004-08-11 14:15:48 0000 *** Bug 50572 has been marked as a duplicate of this bug. *** fjmccloud@yahoo.com 2004-12-22 00:07:56 0000 Please reopen this bug. This was never fixed in kino-0.7.1. People are still complaining about segfaults at start on amd64. See: http://forums.gentoo.org/viewtopic.php?p=1903538#1903538 and: http://forums.gentoo.org/viewtopic.php?t=252484&highlight=kino+amd64 Requesting masking kino-0.7.1 as -amd64 and promoting kino-0.7.5 as + in its place. blubb@gentoo.org 2004-12-22 04:06:36 0000 reopened per user request malc@gentoo.org 2005-01-16 04:17:29 0000 0.7.5 marked stable in cvs as most of the forum ppl report success - I use it just fine (now libdv is fixed). 32337 2004-05-30 14:03 0000 Preliminary patch to allow to get the nautilus like SIG32 instead SIGSEGVs for bad kino source code kino-0.7.1-64bit-clean.diff text/plain LS0tIHNyYy9pbWFnZV9maWx0ZXJzLmNjLm9yaWcJMjAwNC0wNS0zMCAyMTo0MzozMC45MzM2NTQ2 MTYgKzAyMDAKKysrIHNyYy9pbWFnZV9maWx0ZXJzLmNjCTIwMDQtMDUtMzAgMjE6NDI6NTcuMDAw ODEzMTkyICswMjAwCkBAIC01NTgsNiArNTU4LDcgQEAKIAkvLyBTdG9yZSB0aGVzZSBmb3IgZnV0 dXJlIHJlZmVyZW5jZQogCXRoaXMtPm1lbnUgPSBtZW51OwogCXRoaXMtPmNvbnRhaW5lciA9IGNv bnRhaW5lcjsKKwl0aGlzLT5zZWxlY3RlZF9maWx0ZXIgPSAwOwogCiAJLy8gQWRkIHRoZSBmaWx0 ZXJzIHRvIHRoZSBtZW51CiAJR3RrTWVudSAqbWVudV9uZXcgPSBHVEtfTUVOVSggZ3RrX21lbnVf bmV3KCApICk7Ci0tLSBzcmMvaW1hZ2VfdHJhbnNpdGlvbnMuY2Mub3JpZwkyMDA0LTA1LTMwIDIx OjUyOjQ1LjcwMDMxNzI2NCArMDIwMAorKysgc3JjL2ltYWdlX3RyYW5zaXRpb25zLmNjCTIwMDQt MDUtMzAgMjE6NTM6MTQuOTc4ODY2MjQ4ICswMjAwCkBAIC00NzEsNiArNDcxLDcgQEAKIAkvLyBT dG9yZSB0aGVzZSBmb3IgZnV0dXJlIHJlZmVyZW5jZQogCXRoaXMtPm1lbnUgPSBtZW51OwogCXRo aXMtPmNvbnRhaW5lciA9IGNvbnRhaW5lcjsKKwl0aGlzLT5zZWxlY3RlZF90cmFuc2l0aW9uID0g MDsKIAogCS8vIEFkZCB0aGUgdHJhbnNpdGlvbnMgdG8gdGhlIG1lbnUKIAlHdGtNZW51ICptZW51 X25ldyA9IEdUS19NRU5VKCBndGtfbWVudV9uZXcoICkgKTsK