FS#38085 - Firefox segfaults when opening certain website
Attached to Project:
Arch Linux
Opened by indianahorst (indianahorst) - Monday, 09 December 2013, 22:08 GMT
Last edited by Ionut Biru (wonder) - Monday, 20 January 2014, 09:21 GMT
Opened by indianahorst (indianahorst) - Monday, 09 December 2013, 22:08 GMT
Last edited by Ionut Biru (wonder) - Monday, 20 January 2014, 09:21 GMT
|
Details
Description:
Firefox (without any addons and with all plugins deactivated) throws an segmentation fault when visiting http://www.mtv.de/news/41702-justin-bieber-hasch-drama-in-brisbane Additional info: * package version(s) extra/firefox 25.0.1-1 System info (inxi): CPU~Dual core AMD Athlon 5050e (-MCP-) clocked at 2600.000 Mhz Kernel~3.12.3-1-ARCH x86_64 Up~2 days Mem~1245.6/3831.6MB HDD~1000.2GB(94.9% used) Procs~135 Client~Shell inxi~1.9.17 * config and/or log files etc. $ firefox (process:28445): GLib-CRITICAL **: g_slice_set_config: assertion 'sys_page_size == 0' failed Segmentation fault (core dumped) Steps to reproduce: 1. Start Firefox 2. Open the URL http://www.mtv.de/news/41702-justin-bieber-hasch-drama-in-brisbane 3. Wait until the page is loaded 4. Wait some seconds or scroll down 5. Firefox crashes |
This task depends upon
Closed by Ionut Biru (wonder)
Monday, 20 January 2014, 09:21 GMT
Reason for closing: Fixed
Additional comments about closing: Issue seems to be fixed since the update to nvidia 331.38-1.
Monday, 20 January 2014, 09:21 GMT
Reason for closing: Fixed
Additional comments about closing: Issue seems to be fixed since the update to nvidia 331.38-1.
Firefox (without any addons and with all plugins deactivated) (...)
I have tried it with addons and without addons (and plugins), it doesn't make a difference, Firefox continues crashing after a few seconds on this page.
BTW: my crash is fixed with FF 26 in testing.
$ firefox -P testing
(process:25599): GLib-CRITICAL **: g_slice_set_config: assertion 'sys_page_size == 0' failed
OpenGL version detected: 210
OpenGL version detected: 210
OpenGL version detected: 210
Segmentation fault (core dumped)
Here is the output of gdb. Hope it helps!
$ gdb /usr/bin/firefox
GNU gdb (GDB) 7.6.1
Copyright (C) 2013 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law. Type "show copying"
and "show warranty" for details.
This GDB was configured as "x86_64-unknown-linux-gnu".
For bug reporting instructions, please see:
<http://www.gnu.org/software/gdb/bugs/>...
Reading symbols from /usr/lib/firefox/firefox...(no debugging symbols found)...done.
(gdb) run
Starting program: /usr/bin/firefox
warning: Could not load shared library symbols for linux-vdso.so.1.
Do you need "set solib-search-path" or "set sysroot"?
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/usr/lib/libthread_db.so.1".
[New Thread 0x7fffe7220700 (LWP 17590)]
[Thread 0x7fffe7220700 (LWP 17590) exited]
(process:17580): GLib-CRITICAL **: g_slice_set_config: assertion 'sys_page_size == 0' failed
[New Thread 0x7fffe7220700 (LWP 17608)]
[New Thread 0x7fffe53ff700 (LWP 17620)]
[New Thread 0x7fffe49ff700 (LWP 17629)]
[New Thread 0x7fffe41fe700 (LWP 17630)]
[New Thread 0x7fffe359f700 (LWP 17633)]
[New Thread 0x7fffe1bff700 (LWP 17640)]
[New Thread 0x7fffe13fe700 (LWP 17641)]
[New Thread 0x7fffe09ff700 (LWP 17642)]
[New Thread 0x7fffe01fe700 (LWP 17643)]
[New Thread 0x7fffdf7ff700 (LWP 17644)]
[New Thread 0x7fffd4227700 (LWP 17645)]
[New Thread 0x7fffd36ff700 (LWP 17646)]
[New Thread 0x7fffd2cff700 (LWP 17647)]
[New Thread 0x7fffd1dff700 (LWP 17648)]
[New Thread 0x7fffc52ff700 (LWP 17681)]
[New Thread 0x7fffc45f0700 (LWP 17682)]
[New Thread 0x7fffc377c700 (LWP 17683)]
[New Thread 0x7fffc2f7b700 (LWP 17684)]
[New Thread 0x7fffe0bfd700 (LWP 17685)]
[New Thread 0x7fffc277a700 (LWP 17686)]
[New Thread 0x7fffc1f79700 (LWP 17697)]
[New Thread 0x7fffc1778700 (LWP 17698)]
[New Thread 0x7fffc0d6b700 (LWP 17699)]
[New Thread 0x7fffc056a700 (LWP 17701)]
[New Thread 0x7fffbfd69700 (LWP 17703)]
[New Thread 0x7fffbf362700 (LWP 17709)]
[New Thread 0x7fffbbc40700 (LWP 17727)]
[New Thread 0x7fffbb43f700 (LWP 17728)]
[New Thread 0x7fffbac3e700 (LWP 17729)]
[New Thread 0x7fffba43d700 (LWP 17730)]
[New Thread 0x7fffb9c3c700 (LWP 17733)]
[New Thread 0x7fffb943b700 (LWP 17742)]
[New Thread 0x7fffb8c3a700 (LWP 17743)]
[New Thread 0x7fffb72ff700 (LWP 17768)]
[New Thread 0x7fffb5fff700 (LWP 17769)]
[New Thread 0x7fffb57fe700 (LWP 17770)]
[Thread 0x7fffba43d700 (LWP 17730) exited]
[New Thread 0x7fffba43d700 (LWP 17851)]
[New Thread 0x7fffa98ff700 (LWP 17887)]
[New Thread 0x7fffa90ff700 (LWP 17907)]
[New Thread 0x7fffa88fe700 (LWP 17915)]
[New Thread 0x7fffa80fd700 (LWP 18110)]
[New Thread 0x7fffa78fc700 (LWP 18111)]
[Thread 0x7fffb943b700 (LWP 17742) exited]
[Thread 0x7fffb8c3a700 (LWP 17743) exited]
[Thread 0x7fffbac3e700 (LWP 17729) exited]
OpenGL version detected: 210
OpenGL version detected: 210
OpenGL version detected: 210
Program received signal SIGSEGV, Segmentation fault.
0x00007fffe8a8db60 in ?? () from /usr/lib/libnvidia-glcore.so.304.116
(gdb)
Which program package from the Archlinux Repositories do you mean?
https://www.mozilla.org/en-US/firefox/all/
maybe it's a graphic driver problem, for those who have this bug : try to disable Hardware Acceleration feature in firefox options
Probably there's nothing we or mozilla devs can fix here, as a binary module is involved.
#0 0x00007fffe8a8db60 in ?? () from /usr/lib/libnvidia-glcore.so.304.116
#1 0x00007fffe89e0846 in ?? () from /usr/lib/libnvidia-glcore.so.304.116
#2 0x00007fffe8941c7e in ?? () from /usr/lib/libnvidia-glcore.so.304.116
#3 0x00007fffe87b4db3 in ?? () from /usr/lib/libnvidia-glcore.so.304.116
#4 0x00007ffff3fa5aec in ?? () from /usr/lib/firefox/libxul.so
#5 0x00007ffff3fa5b1f in ?? () from /usr/lib/firefox/libxul.so
#6 0x00007ffff3fa7129 in ?? () from /usr/lib/firefox/libxul.so
#7 0x00007ffff3fa74ff in ?? () from /usr/lib/firefox/libxul.so
#8 0x00007ffff3fa75fb in ?? () from /usr/lib/firefox/libxul.so
#9 0x00007ffff3fa0f21 in ?? () from /usr/lib/firefox/libxul.so
#10 0x00007ffff3fa0f5f in ?? () from /usr/lib/firefox/libxul.so
#11 0x00007ffff3f9949c in ?? () from /usr/lib/firefox/libxul.so
#12 0x00007ffff3f97b15 in ?? () from /usr/lib/firefox/libxul.so
#13 0x00007ffff3f97b85 in ?? () from /usr/lib/firefox/libxul.so
#14 0x00007ffff3293139 in ?? () from /usr/lib/firefox/libxul.so
#15 0x00007ffff3290882 in ?? () from /usr/lib/firefox/libxul.so
#16 0x00007ffff3e7e303 in ?? () from /usr/lib/firefox/libxul.so
#17 0x00007ffff375063d in ?? () from /usr/lib/firefox/libxul.so
#18 0x00007ffff3e729ad in ?? () from /usr/lib/firefox/libxul.so
#19 0x00007ffff3e31649 in ?? () from /usr/lib/firefox/libxul.so
#20 0x00007ffff3a15a1b in ?? () from /usr/lib/firefox/libxul.so
#21 0x00007ffff3ecb6ff in ?? () from /usr/lib/firefox/libxul.so
#22 0x00007ffff3990435 in ?? () from /usr/lib/firefox/libxul.so
#23 0x00007ffff3820d2b in ?? () from /usr/lib/firefox/libxul.so
#24 0x00007ffff2c99c4f in ?? () from /usr/lib/firefox/libxul.so
#25 0x00007ffff2c99eb9 in ?? () from /usr/lib/firefox/libxul.so
#26 0x00007ffff2c9a0e4 in XRE_main () from /usr/lib/firefox/libxul.so
#27 0x0000000000403980 in _start ()
Program received signal SIGSEGV, Segmentation fault.
0x00007ffff58658da in ?? () from /usr/lib/firefox/libxul.so
Full log attached
I tried with official build from Mozilla, works without problem without crashes
I did need to add the following line to the build() function in the PKGBUILD:
export SHELL="/usr/bin/bash"
At the end of the build process, there was a warning:
==> WARNING: Package contains reference to $srcdir
Edit: added Firefox gdb output with backtrace. I don't see anything NVIDIA related in there. There's btw. another user reporting the same issue with an Intel graphics card:
https://plus.google.com/u/0/109543393310967160565/posts/PSec517bYio
More details:
- Firefox crashes randomly, but more often when opening PDFs
- The problem seems to go away when running from inside a terminal
- Downgrading to 26.0-1 appears to solve the problem
#0 0x00007ffff58658da in ?? () from /usr/lib/firefox/libxul.so
#1 0x00007ffff5908115 in ?? () from /usr/lib/firefox/libxul.so
#2 0x00007ffff7e4625c in ?? ()
#3 0x0000000000000000 in ?? ()
Program received signal SIGSEGV, Segmentation fault.
0x00007ffff590f20d in ?? () from /usr/lib/firefox/libxul.so
Crashes with a profile with extensions and inactive plugins
Crashes with a profile with no extension and inactive plugins (and even disabled HW acceleration)
Seems ok in "pure" safe-mode
I agree with rlipman : dialog openings (especially auth) seems to lead to more segfaults than regular browsing.
Running from a terminal seemed to correct the problem yesterday, but today it's a mess too...
$ LANGUAGE=C pacman -Qi firefox nvidia |grep -i version
Version : 26.0-2
Version : 331.20-2
$
gdb stuff (I don't know if it's useful): http://pastebin.com/Xvm9RRHA
PDF viewer seems to always crash even in safe-mode. For example this pdf causes a crash for me: http://www.nxp.com/documents/data_sheet/LPC11U3X.pdf
I reset my Firefox and it didn't help a thing, so now I'm hunting all my addons back.
OpenGL version detected: 330
OpenGL version detected: 330
...
OpenGL version detected: 330
so long as the page in the original report (MTV.de) is open, but it will not crash.
This does seem to be an issue between firefox and nvidia. I'll keep playing around and see if I can get it to segfault again.
As mentioned above, it still crashes with hw acceleration disabled :/
I guess it's related !
https://support.mozilla.org/fr/questions/976711
Waiting for a fix.
For the Mono issues, there's a separate Task at https://bugs.archlinux.org/task/38010 , but in all likelihood this is the same bug.
The driver issues are "tracked" (discussed, rather) upstream at https://devtalk.nvidia.com/default/topic/633706/linux/recent-drivers-cause-applications-to-hang-not-start-at-all-or-compilation-failures/
Another user was so kind to provide PKGBUILDs for the old version of the driver in https://bbs.archlinux.org/viewtopic.php?id=173024
https://projects.archlinux.org/svntogit/packages.git/commit/trunk?h=packages/firefox&id=913909a50e329a03ffec0f6ac4b434a065fb8814
The most important change is that 26.0-2 is linked against a lot more system libraries (as opposed to built-in code provided by Mozilla). In particular, the following lines got added to mozconfig:
+ac_add_options --with-system-nspr
+ac_add_options --with-system-nss
+ac_add_options --with-system-jpeg
+ac_add_options --with-system-zlib
+ac_add_options --with-system-bz2
+ac_add_options --with-system-png
+ac_add_options --with-system-libevent
+ac_add_options --with-system-libvpx
+ac_add_options --enable-system-hunspell
+ac_add_options --enable-system-sqlite
+ac_add_options --enable-system-ffi
+#ac_add_options --enable-system-cairo
Perhaps someone could remove these, re-build the package, and see if Firefox still crashes?
(I plan to do this myself, of course, but I'm in a hurry right now.)
UPDATE: I just verified it's an nvidia issue myself. So, I also think tinkering with Firefox is not really a solution. I-ll be using nvidia-304xx until this is resolved upstream.
This should be changed to a problem with Nvidia packages, not firefox, and merged with the mono bug, seeing as that is probably from the same source (as far as I have deduced from reading forum posts about this topic).
$ LANGUAGE=C pacman -Qi firefox nvidia |grep -i version
Version : 26.0-2
Version : 331.20-3
As this is a Nvidia driver problem, are Nvidia's developers aware of this?
https://devtalk.nvidia.com/default/topic/633706/linux/recent-drivers-cause-applications-to-hang-not-start-at-all-or-compilation-failures/
In the last comment (12/26/2013) they said: "Issue no longer reproduce with our latest driver that will be available soon."
I hope this work.
Edit: added a link to this bug report over at Mozilla's: https://bugzilla.mozilla.org/show_bug.cgi?id=950397
Are you guys all running custom compiles without default CFLAGS/CPPFLAGS?
- noticed -D_FORTIFY_SOURCE=2 was sometimes missing during build process despite being in CPPFLAGS
- googled around and learned that indeed CPPFLAGS can sometimes be ignored altogether
- concluded that some build tools on linux are probably flawed in this regard [it's quite a mess isn't it?]
- decided that it's better to alter CFLAGS and CXXFLAGS and sleep well
Regarding -D_FORTIFY_SOURCE=2, it was set in CPPFLAGS alone on 1 machine, while it was defined in CFLAGS and CXXFLAGS on the other. Building nvidia stuff with only the CPPFLAGS was not enough for the former, so I modified makepkg.conf to match that of the latter, ie. follow revel's advice. Voilà :-)
[Edit] Just had 2 crashes. They didn't occur at startup but a little later. I'll have to test some more.