Closed Bug 756337 Opened 12 years ago Closed 11 years ago

Periodic crashing with intel_do_flush_locked failed: Input/output error

Categories

(Core :: Graphics, defect)

Product:
Core
Component:
Graphics
Version:
12 Branch
Platform:
x86_64
Linux
Type:
defect
Priority:
Not set
Severity:
critical

Tracking

()

Status:
RESOLVED WORKSFORME

People

(Reporter: web, Unassigned)

Details

(Keywords: crash) 

User Agent: Mozilla/5.0 (X11; Linux x86_64; rv:12.0) Gecko/20100101 Firefox/12.0
Build ID: 20120426131355

Steps to reproduce:

This bug crops up periodically. Once it starts happening, nearly any interaction with Firefox provokes this crash. Sometimes it takes  few minutes, but I've found that Help -> Troubleshooting Information provides a crash almost immediately. Running in -safe-mode does not solve the problem.



User Agent: Mozilla/5.0 (X11; Linux x86_64; rv:12.0) Gecko/20100101 Firefox/12.0
uname -a: Linux 3.3.6-1-ARCH #1 SMP PREEMPT Sun May 13 10:52:32 CEST 2012 x86_64 GNU/Linux

Some initial searching has told me this may be graphics related, so here is my graphics information from about:support

Adapter Description:Tungsten Graphics, Inc -- Mesa DRI Mobile Intel® GM45 Express Chipset
Vendor ID:Tungsten Graphics, Inc
Device ID:Mesa DRI Mobile Intel® GM45 Express Chipset
Driver Version:2.1 Mesa 8.0.2
WebGL Renderer:Tungsten Graphics, Inc -- Mesa DRI Mobile Intel® GM45 Express Chipset  -- 2.1 Mesa 8.0.2
GPU Accelerated Windows:0
AzureBackend: skia






Actual results:

The browser crashes after some relatively short period of time (0-2 minutes), and if launched from a console displays:
intel_do_flush_locked failed: Input/output error

My shell reports a segfault.


Expected results:

I would of course prefer that the crash didn't occur.
Does it happen in Safe Mode (see http://support.mozilla.org/kb/Safe%20Mode)?
Does it happen with a new profile (see http://support.mozilla.org/kb/Managing-profiles)?
Can you provide a stacktrace (see https://developer.mozilla.org/en/How_to_get_a_stacktrace_for_a_bug_report)?
Component: Untriaged → Graphics
Product: Firefox → Core
QA Contact: untriaged → thebes
Severity: normal → critical
Keywords: crash
It does happen in Safe Mode. As soon as it appears next (I haven't figured out how to reproduce it, unfortunately), I will test a fresh profile and provide a stack trace.
(In reply to web from comment #2)
> It does happen in Safe Mode. As soon as it appears next (I haven't figured
> out how to reproduce it, unfortunately), I will test a fresh profile and
> provide a stack trace.

Can you provide link to a crash report? You can get a crash report for this as follows:

1. Go to about:crashes
2. Find the date of when the crash occurred in the list of links
3. Click that link
4. Paste that link here (that will give us a crash report to work off of)
about:crashes gives me a "The URL is not valid and cannot be loaded."

I just checked about:buildconfig, and it appears that for some reason my distro (Arch) built Firefox with --disable-crashreporter.

I can ask on the Arch forums the reasoning behind this, or maybe someone here knows why someone would elect to disable it?
(In reply to web from comment #4)
> about:crashes gives me a "The URL is not valid and cannot be loaded."
> 
> I just checked about:buildconfig, and it appears that for some reason my
> distro (Arch) built Firefox with --disable-crashreporter.
> 
> I can ask on the Arch forums the reasoning behind this, or maybe someone
> here knows why someone would elect to disable it?

Good question. I'd file a bug on this, as this sounds incorrect. File a bug on this in Toolkit --> Breakpad indicating that the crash reporter is disabled on Arch Linux.
I'm getting this as well. Started happening with 20.0 for Ubuntu.

Build identifier: Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:20.0) Gecko/20100101 Firefox/20.0

It's started reliably crashing on various pages. Including internal ones. If I start Firefox from a command line this is what I get:

$ firefox
ATTENTION: default value of option force_s3tc_enable overridden by environment.
intel_do_flush_locked failed: Input/output error
$

One of the pages it reliably crashes on is about:support, which is more than a little annoying.

The crash also happens in safe mode:

$ firefox -safe-mode
ATTENTION: default value of option force_s3tc_enable overridden by environment.
intel_do_flush_locked failed: Input/output error

###!!! [Child][SyncChannel] Error: Channel error: cannot send/recv

about:crashes doesn't list any of these crashes.

I was wondering if it was the hardware acceleration, but I disabled that in preferences and it still crashes.

I tried using a Mozilla binary from tarball and it doesn't work!

$ ./run-mozilla.sh 

run-mozilla.sh: Cannot execute .

$ ./firefox-bin 
XPCOMGlueLoad error for file /home/fun/bin/firefox/libxpcom.so:
libxul.so: cannot open shared object file: No such file or directory
Couldn't load XPCOM.
$ ./firefox
XPCOMGlueLoad error for file /home/fun/bin/firefox/libxpcom.so:
libxul.so: cannot open shared object file: No such file or directory
Couldn't load XPCOM.


Ideas? Move to Chromium? Firefox is randomly unusable, including for internal site I have to use. I'm at a loss at this point.
Please provide a stack trace (see https://developer.mozilla.org/docs/How_to_get_a_stacktrace_for_a_bug_report#Linux).
Flags: needinfo?(web)
So, per my comment, how do I run the mozilla.org binary? Note that I would have provided a stacktrace in the above *if I could have*.
I have the mozilla.org binary running. (The secret: sudo apt-get install ia32-libs.)

I'm not getting the crash reporter. The entire text at the link you give is:

"Note that for most distros the package you need to get symbols for will be something like "xulrunner", not "firefox"."

That's not in any way directions on how to give you what you're asking for.
When I run the Mozilla binary in safe mode, I get a better error now:

$ ./firefox -safe
LoadPlugin: failed to initialize shared library libXt.so [libXt.so: cannot open shared object file: No such file or directory]
LoadPlugin: failed to initialize shared library libXext.so [libXext.so: cannot open shared object file: No such file or directory]
LoadPlugin: failed to initialize shared library /usr/lib/flashplugin-installer/libflashplayer.so [/usr/lib/flashplugin-installer/libflashplayer.so: wrong ELF class: ELFCLASS64]
ATTENTION: default value of option force_s3tc_enable overridden by environment.
intel_do_flush_locked failed: Input/output error
Component: Graphics → Plug-ins
(In reply to David Gerard from comment #10)
> [/usr/lib/flashplugin-installer/libflashplayer.so: wrong ELF class:
> ELFCLASS64]

Looks like you have 64bit Flash installed and are trying to use it from a 32bit Firefox.
Please use matching architectures for both.
Apparently the official download site doesn't offer 64bit builds; these ones should be proper though:
http://releases.mozilla.org/pub/mozilla.org/firefox/releases/latest/linux-x86_64/
I'm confused. If you are using a distro build, you should be able to get the debuginfo package from your distro and catch the crash in a debugger.

Are you experiencing the same crash using the 32-bit mozilla.org build? If so, are you saying that the crash reporter is not showing up?
I am seeing this both in the 64-bit distro build and in the 32-bit mozilla.org build. It crashes reliably on about:support and http://bugs.kde.org, to give two examples.

Georg #11 - please read my comments in full, this has been answered. I even gave the build ID in #6.

The crash reporter is not showing up for either Ubuntu or mozilla.org builds.

This is pretty stock Xubuntu 12.04. I've installed firefox-dbg. This is the result:

$ gdb /usr/lib/firefox/firefox
GNU gdb (Ubuntu/Linaro 7.4-2012.04-0ubuntu2.1) 7.4-2012.04
Copyright (C) 2012 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-linux-gnu".
For bug reporting instructions, please see:
<http://bugs.launchpad.net/gdb-linaro/>...
Reading symbols from /usr/lib/firefox/firefox...Reading symbols from /usr/lib/debug/usr/lib/firefox/firefox...done.
done.
(gdb) run
Starting program: /usr/lib/firefox/firefox 
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
[New Thread 0x7fffea6ec700 (LWP 3113)]
[New Thread 0x7fffe9579700 (LWP 3114)]
[New Thread 0x7fffe8bff700 (LWP 3115)]
[New Thread 0x7fffe7de2700 (LWP 3116)]
[New Thread 0x7fffe5dff700 (LWP 3117)]
[New Thread 0x7fffe51fe700 (LWP 3118)]
[New Thread 0x7fffe49fd700 (LWP 3119)]
[New Thread 0x7fffe3cff700 (LWP 3120)]
[New Thread 0x7fffe2cff700 (LWP 3121)]
[Thread 0x7fffe2cff700 (LWP 3121) exited]
[New Thread 0x7fffe2cff700 (LWP 3122)]
[New Thread 0x7fffdbdff700 (LWP 3123)]
[New Thread 0x7fffdb0ff700 (LWP 3124)]
[New Thread 0x7fffda0fe700 (LWP 3125)]
[New Thread 0x7fffd92ff700 (LWP 3126)]
[Thread 0x7fffe2cff700 (LWP 3122) exited]
[New Thread 0x7fffe2cff700 (LWP 3127)]
[New Thread 0x7fffd88ff700 (LWP 3128)]
[New Thread 0x7fffd80fe700 (LWP 3129)]
[Thread 0x7fffd88ff700 (LWP 3128) exited]
[Thread 0x7fffd80fe700 (LWP 3129) exited]
[Thread 0x7fffd92ff700 (LWP 3126) exited]
[New Thread 0x7fffd92ff700 (LWP 3130)]
[New Thread 0x7fffd88ff700 (LWP 3131)]
[New Thread 0x7fffd80fe700 (LWP 3132)]
[Thread 0x7fffd80fe700 (LWP 3132) exited]
[New Thread 0x7fffd80fe700 (LWP 3133)]
[New Thread 0x7fffd42ff700 (LWP 3134)]
[New Thread 0x7fffd1f9a700 (LWP 3135)]
[New Thread 0x7fffd1799700 (LWP 3136)]
[New Thread 0x7fffd0aff700 (LWP 3137)]
[New Thread 0x7fffce9ff700 (LWP 3138)]
ATTENTION: default value of option force_s3tc_enable overridden by environment.
intel_do_flush_locked failed: Input/output error
[Thread 0x7fffe7de2700 (LWP 3116) exited]
[Thread 0x7fffd42ff700 (LWP 3134) exited]
[Thread 0x7fffd88ff700 (LWP 3131) exited]
[Thread 0x7fffd1799700 (LWP 3136) exited]
[Thread 0x7fffdbdff700 (LWP 3123) exited]
[Thread 0x7fffe49fd700 (LWP 3119) exited]
[Thread 0x7fffda0fe700 (LWP 3125) exited]
[Thread 0x7fffd80fe700 (LWP 3133) exited]
[Thread 0x7fffe51fe700 (LWP 3118) exited]
[Thread 0x7fffd0aff700 (LWP 3137) exited]
[Thread 0x7fffe5dff700 (LWP 3117) exited]
[Thread 0x7fffd92ff700 (LWP 3130) exited]
[Thread 0x7fffe8bff700 (LWP 3115) exited]
[Thread 0x7fffd1f9a700 (LWP 3135) exited]
[Thread 0x7fffdb0ff700 (LWP 3124) exited]
[Thread 0x7fffe9579700 (LWP 3114) exited]
[Thread 0x7fffea6ec700 (LWP 3113) exited]
[Thread 0x7fffe2cff700 (LWP 3127) exited]
[Thread 0x7fffce9ff700 (LWP 3138) exited]
[Thread 0x7fffe3cff700 (LWP 3120) exited]
[Inferior 1 (process 3049) exited with code 01]
(gdb) 

The crash happened when I went to about:support .

Is this what you expect to see? Should I try something else?
From that gdb trace, I don't see any evidence that Firefox crashed. It says that it just "exited with code 01".
Well, it certainly did something indistinguishable from crashing! All Firefox windows disappeared and there's no Firefox process hanging about.
Given that the Flash issue is a separate one and "intel_do_flush_locked failed: input/output error" seems to be connected to process exits over some Linux distributions, Core::Graphics seems to be the better fit.
Component: Plug-ins → Graphics
I'm now on Firefox 21.0 (from the distro) and the problem seems to have stopped. (I can't speak for the original bug poster.) There was also a kernel update, which might contain video driver updates.
Agreed, I no longer experience this bug, I will assume it was some kernel update, and just hope it doesn't return.
Flags: needinfo?(web)
Status: UNCONFIRMED → RESOLVED
Closed: 11 years ago
Resolution: --- → WORKSFORME
FWIW, I think I can narrow down what fixed this.  I was seeing the same error on Fedora 18, with the distro-packaged kernel and Firefox:

Linux 3.6.10-4.fc18.i686
firefox-24.0-1.fc18.i686

I updated the following packages, restarted X, and the crash no longer happened.

xorg-x11-drv-intel.i686:
   old, 0:2.20.14-1.fc18
   new, 0:2.21.12-1.fc18

xorg-x11-server-Xorg.i686:
   old, 0:1.13.0-11.fc18
   new, 0:1.13.3-3.fc18

libunwind.i686:
   old, (not installed)
   new, 0:1.0.1-5.fc18

xorg-x11-server-common.i686:
   old, 0:1.13.0-11.fc18
   new, 0:1.13.3-3.fc18

libXv.i686:
   old, 1.0.7-2.fc18
   new, 1.0.7-5.20130524git50fc4cb18.fc18

libXvMC.i686:
   old, 1.0.7-2.fc18
   new, 1.0.7-6.20130524gite9415ddef.fc18

I suspect it was a bug fix in the Xorg Intel drivers.
Hey guys, can we resolve this bug with a different tag than WORKSFORME, because it seems very clear that this problem occurs with buggy Xorg Intel drivers (I ran into the very same problem), and additionally the crash reporting dialog fails to come up, which seems like a proper bug to me...
Same problem here:

firefox

(process:6030): GLib-CRITICAL **: g_slice_set_config: assertion 'sys_page_size == 0' failed
WARNING: content window passed to PrivateBrowsingUtils.isWindowPrivate. Use isContentWindowPrivate instead (but only for frame scripts).
pbu_isWindowPrivate@resource://gre/modules/PrivateBrowsingUtils.jsm:25:14
nsBrowserAccess.prototype.openURI@chrome://browser/content/browser.js:13858:25
ATTENTION: default value of option force_s3tc_enable overridden by environment.
intel_do_flush_locked failed: Eingabe-/Ausgabefehler
[NPAPI 6076] ###!!! ABORT: Aborting on channel error.: file /build/firefox/src/mozilla-release/ipc/glue/MessageChannel.cpp, line 1512
[NPAPI 6076] ###!!! ABORT: Aborting on channel error.: file /build/firefox/src/mozilla-release/ipc/glue/MessageChannel.cpp, line 1512

Tagging this with worksforme isnt a solutiot...
You need to log in before you can comment on or make changes to this bug.