please runtime test this, because i'm not sure if it runs OK due to local issues i'm having anyway, it's needed eventually for the spidermonkey-1.8.5 support thanks
Created attachment 292883 [details] gxine output with bt bad for me.
amd64: no pass here either @ago +1
Let's say this report is too vague. I'm on x86 mixed and gxine 0.5.906 seems to work for me, though I don't use it much.
gxine-0.5.906 runs fine here (tried several local mkv, avi, and flv files). This is on an ~amd64 system with gnome-3.2; gxine built with USE="gnome nls udev xcb xinerama -lirc -nsplugin". Samuli, could you post the steps you need to reproduce the bug?
Created attachment 293009 [details] backtrace backtrace after CXXFLAGS="-march=core2 -O1 -pipe -g -ggdb" CFLAGS="-march=core2 -O1 -pipe -g -ggdb" FEATURES="nostrip" USE="debug" emerge -av spidermonkey gxine
Created attachment 293011 [details] emerge info from above system with backtrace also, if I recompile gxine and spidermonkey with -O2 instead of -O1 -ggdb -g I see the mummap chunk like the first paste in this bug
My C{XX}FLAGS are simply '-O2 -march=athlon -mtune=athlon'.
(In reply to comment #6) > #0 0x00007ffff7bbe184 in lexer_finalize_r () from /usr/lib64/libxine.so.1 > #1 0x00007ffff7bbf95c in xml_parser_finalize_r () from /usr/lib64/libxine.so.1 > #2 0x000000000042abc9 in playlist_load (fname=0x1829e00 "/home/ssuominen/.config/gxine/playlist") at playlist.c:529 Could you please append your ~/.config/gxine/playlist? The backtrace seems to indicate that the crash is in the playlist parser, and without your playlist file, debugging it would be difficult.
(In reply to comment #8) > (In reply to comment #6) > > > #0 0x00007ffff7bbe184 in lexer_finalize_r () from /usr/lib64/libxine.so.1 > > #1 0x00007ffff7bbf95c in xml_parser_finalize_r () from /usr/lib64/libxine.so.1 > > #2 0x000000000042abc9 in playlist_load (fname=0x1829e00 "/home/ssuominen/.config/gxine/playlist") at playlist.c:529 > > Could you please append your ~/.config/gxine/playlist? The backtrace seems to > indicate that the crash is in the playlist parser, and without your playlist > file, debugging it would be difficult. My ~/.config/gxine only has one "file" called "socket": ssuominen@null ~/.config $ ls -l *xine* total 0 srwxr-xr-x 1 ssuominen users 0 Nov 24 19:17 socket and the backtrace is always same, before and after rm -rf ~/.config/*xine*
@comment 9: Well.... It's a bit ridiculous, but ~/.config/gxine/playlist missing *is* the reason of this crash. I can reproduce it by simply renaming my ~/.config/gxine/playlist. But at the same time, it means it should be easy to fix, as even 'touch playlist' is enough to bypass it (sure, it complains about such "file" having invalid format, but it doesn't crash). So, it just a matter of g_file_test(<filename>, G_FILE_TEST_EXISTS) (or perhaps G_FILE_TEST_IS_REGULAR) in the correct place.
Created attachment 293735 [details, diff] patch that fixes the crash So, it was both more and less complicated - it sort of lies in http://anonscm.debian.org/hg/xine-lib/gxine/rev/d70688fa8c4a, but actually it seems the problem existed earlier, but was simply hidden. This patch fixes the problem for me. On unrelated note: gxine is awfully noisy during build with spidermonkey 1.8.5 (probably it's gnu99) - would be nice, if its upstream looked into that.
Works for me. Thanks a lot Rafal! Pushed upstream here: http://bugs.xine-project.org/show_bug.cgi?id=438 And in portage: *gxine-0.5.906-r1 (25 Nov 2011) 25 Nov 2011; Samuli Suominen <ssuominen@gentoo.org> +gxine-0.5.906-r1.ebuild, +files/gxine-0.5.906-playlist_segfault.patch: Fix segmentation fault in playlist wrt #390783 by Rafał Mużyło
