[Bug 503151] New: firefox crashes with scim/skim enabled
http://bugzilla.novell.com/show_bug.cgi?id=503151 Summary: firefox crashes with scim/skim enabled Classification: openSUSE Product: openSUSE 11.2 Version: Factory Platform: Other OS/Version: Other Status: NEW Severity: Major Priority: P5 - None Component: Firefox AssignedTo: bnc-team-mozilla@forge.provo.novell.com ReportedBy: burnus@gmx.de QAContact: qa@suse.de Found By: --- This is with Firefox 3.5 beta4 under KDE 3 or KDE 4 with scim/skim enabled. Result: firefox crashes. Using QT_IM_MODULE= XMODIFIERS= GTK_IM_MODULE= firefox it works. Result of running it with firefox -d gdb -no-remote -safe-mode 0x00007ffff6c4d645 in raise () from /lib64/libc.so.6 (gdb) bt #0 0x00007ffff6c4d645 in raise () from /lib64/libc.so.6 #1 0x00007ffff6c4ec33 in abort () from /lib64/libc.so.6 #2 0x00007ffff6c898e8 in ?? () from /lib64/libc.so.6 #3 0x00007ffff6c8f118 in ?? () from /lib64/libc.so.6 #4 0x00007ffff6c90c76 in free () from /lib64/libc.so.6 #5 0x00007fffddbb2c89 in ?? () from /lib64/libnsl.so.1 #6 0x00007ffff7bcfed3 in pthread_once () from /lib64/libpthread.so.0 #7 0x00007fffddbb2df1 in _nsl_default_nss () from /lib64/libnsl.so.1 #8 0x00007fffdd99db51 in _nss_nis_setpwent () from /lib64/libnss_nis.so.2 #9 0x00007fffdddbd4b9 in ?? () from /lib64/libnss_compat.so.2 #10 0x00007fffdddbd626 in _nss_compat_setpwent () from /lib64/libnss_compat.so.2 #11 0x00007ffff6cfe3aa in ?? () from /lib64/libc.so.6 #12 0x00007ffff6cbb88c in setpwent () from /lib64/libc.so.6 #13 0x00007fffde26c62e in scim::scim_get_home_dir() () from /usr/lib64/libscim-1.0.so.8 #14 0x00007fffde22ed4c in ?? () from /usr/lib64/libscim-1.0.so.8 #15 0x00007fffde230cd5 in scim::scim_global_config_read(std::string const&, int) () from /usr/lib64/libscim-1.0.so.8 #16 0x00007fffde25f7b7 in scim::scim_get_default_socket_timeout() () from /usr/lib64/libscim-1.0.so.8 #17 0x00007fffde25c712 in scim::PanelClient::PanelClientImpl::PanelClientImpl() -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
http://bugzilla.novell.com/show_bug.cgi?id=503151
User wolfgang@rosenauer.org added comment
http://bugzilla.novell.com/show_bug.cgi?id=503151#c1
Wolfgang Rosenauer
http://bugzilla.novell.com/show_bug.cgi?id=503151
User kairo@kairo.at added comment
http://bugzilla.novell.com/show_bug.cgi?id=503151#c2
Robert Kaiser
From what I could see in gdb, it looks like some problem with memory access, and it started happening on the Firefox/Mozilla side (all Mozilla products, actually) with the switch of the main binary to jemalloc - that's why Firefox 3.0.x didn't show it while Firefox 3.5 does.
Even new Firefox/Mozilla source on openSUSE 11.1 works fine, so I suspect that somehow the new libgcc (or something we picked up at a similar time) and jemalloc don't like each other when it comes to memory management. Firefox somehow accesses GTK stuff for print and print preview, AFAIK, maybe that helps to find a clue as to what is exactly happening. In https://bugzilla.mozilla.org/show_bug.cgi?id=473428 where I mentioned this on the Mozilla side, a run under valgrind was requested, but I don't really have a clue how to find anything out there. -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
http://bugzilla.novell.com/show_bug.cgi?id=503151
User wolfgang@rosenauer.org added comment
http://bugzilla.novell.com/show_bug.cgi?id=503151#c3
--- Comment #3 from Wolfgang Rosenauer
http://bugzilla.novell.com/show_bug.cgi?id=503151
User wolfgang@rosenauer.org added comment
http://bugzilla.novell.com/show_bug.cgi?id=503151#c4
Wolfgang Rosenauer
http://bugzilla.novell.com/show_bug.cgi?id=503151
User pbaudis@novell.com added comment
http://bugzilla.novell.com/show_bug.cgi?id=503151#c5
Petr Baudis
http://bugzilla.novell.com/show_bug.cgi?id=503151
User kairo@kairo.at added comment
http://bugzilla.novell.com/show_bug.cgi?id=503151#c6
--- Comment #6 from Robert Kaiser
In contrary to Robert's comment Firefox 3.0.x already uses jemalloc but apparently didn't expose the issue (at least up to 11.1 (not sure for Factory)).
We (mozilla) did use jemalloc there but not for the main binary yet, that was only introduced in Firefox 3.5 et al. and apparently was the point where this was triggered. (In reply to comment #5)
We currently do not support custom malloc() implementation in NSS due to our patch to open NSS modules deep-bound
Is this a patch to the Mozilla NSS or elsewhere (libc or so)? I'm seeing the same problem with my own builds of Mozilla applications (all of Firefox, SeaMonkey, Thunderbird) which don't use system-installed Mozilla NSS but the one that comes with the Mozilla code, i.e. I recompile that one as well. -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
http://bugzilla.novell.com/show_bug.cgi?id=503151
User pbaudis@novell.com added comment
http://bugzilla.novell.com/show_bug.cgi?id=503151#c7
--- Comment #7 from Petr Baudis
http://bugzilla.novell.com/show_bug.cgi?id=503151
User kairo@kairo.at added comment
http://bugzilla.novell.com/show_bug.cgi?id=503151#c8
--- Comment #8 from Robert Kaiser
http://bugzilla.novell.com/show_bug.cgi?id=503151
User dholbert@mozilla.com added comment
http://bugzilla.novell.com/show_bug.cgi?id=503151#c9
Daniel Holbert
In the meantime, can the custom malloc be turned off temporarily?
I'm not sure if this is what you were asking, but you can disable the custom malloc in your own builds by adding: ac_add_options --disable-jemalloc to your .mozconfig file at build time. -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
http://bugzilla.novell.com/show_bug.cgi?id=503151
User wolfgang@rosenauer.org added comment
http://bugzilla.novell.com/show_bug.cgi?id=503151#c10
--- Comment #10 from Wolfgang Rosenauer
(In reply to comment #5)
In the meantime, can the custom malloc be turned off temporarily?
I'm not sure if this is what you were asking, but you can disable the custom malloc in your own builds by adding: ac_add_options --disable-jemalloc to your .mozconfig file at build time.
Petr, as Daniel said but what means "temporarily" in your question to give me an idea if it makes sense to go through that since it really has to be temporary? It's not really an option to ship mozilla like that. -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
http://bugzilla.novell.com/show_bug.cgi?id=503151
User pbaudis@novell.com added comment
http://bugzilla.novell.com/show_bug.cgi?id=503151#c11
Petr Baudis
http://bugzilla.novell.com/show_bug.cgi?id=503151
User dholbert@mozilla.com added comment
http://bugzilla.novell.com/show_bug.cgi?id=503151#c12
--- Comment #12 from Daniel Holbert
http://bugzilla.novell.com/show_bug.cgi?id=503151
User wolfgang@rosenauer.org added comment
http://bugzilla.novell.com/show_bug.cgi?id=503151#c13
--- Comment #13 from Wolfgang Rosenauer
I don't want to argue with you, I'm merely wondering, why isn't it an option? What does the custom malloc() actually do?
Here are some references why mozilla started to use jemalloc as malloc replacement: http://blog.pavlov.net/2008/03/11/firefox-3-memory-usage/ http://blog.pavlov.net/2007/12/04/vlad-and-analysis-of-dtrace-was-used/ http://blog.pavlov.net/2008/02/27/jemalloc-on-trunk-linux-edition/ So it's (probably) not an option because: 1. it helps to keep memory usage under control (at least given the above analysis are correct 2. you never know what can be changed from upstream builds (legally because of their trademark policy) -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
http://bugzilla.novell.com/show_bug.cgi?id=503151
Wolfgang Rosenauer
http://bugzilla.novell.com/show_bug.cgi?id=503151
User wolfgang@rosenauer.org added comment
http://bugzilla.novell.com/show_bug.cgi?id=503151#c14
--- Comment #14 from Wolfgang Rosenauer
http://bugzilla.novell.com/show_bug.cgi?id=503151
User pbaudis@novell.com added comment
http://bugzilla.novell.com/show_bug.cgi?id=503151#c15
--- Comment #15 from Petr Baudis
http://bugzilla.novell.com/show_bug.cgi?id=503151
User dave.plater@yahoo.co.uk added comment
http://bugzilla.novell.com/show_bug.cgi?id=503151#c16
--- Comment #16 from Dave Plater
http://bugzilla.novell.com/show_bug.cgi?id=503151
User kairo@kairo.at added comment
http://bugzilla.novell.com/show_bug.cgi?id=503151#c17
--- Comment #17 from Robert Kaiser
I've run firefox-3.5rc2 from mozilla for three days with no crash yet maybe this is a clue to something?
No, it probably isn't. 1) As long as you don't try things like print or print preview, we know that there's no problem, as you don't interact with glibc memory management. 2) We don't need clues, we know where the problem is. We're just waiting for the relevant people to remove or fix the patch in glibc NSS. Petr, any progress on that? -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
http://bugzilla.novell.com/show_bug.cgi?id=503151
User dave.plater@yahoo.co.uk added comment
http://bugzilla.novell.com/show_bug.cgi?id=503151#c18
--- Comment #18 from Dave Plater
http://bugzilla.novell.com/show_bug.cgi?id=503151
User aj@novell.com added comment
http://bugzilla.novell.com/show_bug.cgi?id=503151#c19
Andreas Jaeger
http://bugzilla.novell.com/show_bug.cgi?id=503151
User pbaudis@novell.com added comment
http://bugzilla.novell.com/show_bug.cgi?id=503151#c20
Petr Baudis
http://bugzilla.novell.com/show_bug.cgi?id=503151
User aj@novell.com added comment
http://bugzilla.novell.com/show_bug.cgi?id=503151#c21
Andreas Jaeger
http://bugzilla.novell.com/show_bug.cgi?id=503151
User pbaudis@novell.com added comment
http://bugzilla.novell.com/show_bug.cgi?id=503151#c22
--- Comment #22 from Petr Baudis
http://bugzilla.novell.com/show_bug.cgi?id=503151
User wolfgang@rosenauer.org added comment
http://bugzilla.novell.com/show_bug.cgi?id=503151#c23
--- Comment #23 from Wolfgang Rosenauer
http://bugzilla.novell.com/show_bug.cgi?id=503151
User wolfgang@rosenauer.org added comment
http://bugzilla.novell.com/show_bug.cgi?id=503151#c24
--- Comment #24 from Wolfgang Rosenauer
http://bugzilla.novell.com/show_bug.cgi?id=503151
User dave.plater@yahoo.co.uk added comment
http://bugzilla.novell.com/show_bug.cgi?id=503151#c25
--- Comment #25 from Dave Plater
http://bugzilla.novell.com/show_bug.cgi?id=503151
User wolfgang@rosenauer.org added comment
http://bugzilla.novell.com/show_bug.cgi?id=503151#c26
--- Comment #26 from Wolfgang Rosenauer
http://bugzilla.novell.com/show_bug.cgi?id=503151
User dave.plater@yahoo.co.uk added comment
http://bugzilla.novell.com/show_bug.cgi?id=503151#c27
--- Comment #27 from Dave Plater
http://bugzilla.novell.com/show_bug.cgi?id=503151
User aj@novell.com added comment
http://bugzilla.novell.com/show_bug.cgi?id=503151#c28
--- Comment #28 from Andreas Jaeger
http://bugzilla.novell.com/show_bug.cgi?id=503151
User wolfgang@rosenauer.org added comment
http://bugzilla.novell.com/show_bug.cgi?id=503151#c29
Wolfgang Rosenauer
http://bugzilla.novell.com/show_bug.cgi?id=503151
User dave.plater@yahoo.co.uk added comment
http://bugzilla.novell.com/show_bug.cgi?id=503151#c30
--- Comment #30 from Dave Plater
http://bugzilla.novell.com/show_bug.cgi?id=503151
User dave.plater@yahoo.co.uk added comment
http://bugzilla.novell.com/show_bug.cgi?id=503151#c31
--- Comment #31 from Dave Plater
http://bugzilla.novell.com/show_bug.cgi?id=503151
User wolfgang@rosenauer.org added comment
http://bugzilla.novell.com/show_bug.cgi?id=503151#c32
Wolfgang Rosenauer
https://bugzilla.novell.com/show_bug.cgi?id=503151
https://bugzilla.novell.com/show_bug.cgi?id=503151#c33
--- Comment #33 from Swamp Workflow Management
http://bugzilla.novell.com/show_bug.cgi?id=503151
http://bugzilla.novell.com/show_bug.cgi?id=503151#c34
--- Comment #34 from Bernhard Wiedemann
participants (1)
-
bugzilla_noreply@novell.com