732511 - Firefox 10.0.2.4428 consuming 100% CPU - fails to fully exit.

Status: RESOLVED INCOMPLETE

(Firefox :: Untriaged, defect)

Description

[Eric Harding]

User Agent: Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/535.11 (KHTML, like Gecko) Chrome/17.0.963.56 Safari/535.11

Steps to reproduce:

Firefox has been running for 3 days.  Mostly used for gmail and google calendar.  I noticed that Firefox was taking up 100% CPU and hit close (intending to just restart the browser).


Actual results:

Firefox.exe did not terminate.


Expected results:

Firefox.exe should have terminated.

Comment 1

[Eric Harding]

Additional details:

0:020:x86> !runaway
 User Mode Time
  Thread       Time
  20:1dd0      0 days 3:54:49.542
   0:1068      0 days 0:14:22.092
  19:1e98      0 days 0:02:35.501
   3:1224      0 days 0:00:03.962
   8:10c8      0 days 0:00:03.728
   7:1350      0 days 0:00:01.357
  17:14d4      0 days 0:00:01.279
  10:1438      0 days 0:00:00.031
  21:c90       0 days 0:00:00.015
  24:18bc      0 days 0:00:00.000
  23:be8       0 days 0:00:00.000
  22:11a0      0 days 0:00:00.000
  18:1f60      0 days 0:00:00.000
  16:700       0 days 0:00:00.000
  15:f70       0 days 0:00:00.000
  14:b60       0 days 0:00:00.000
  13:724       0 days 0:00:00.000
  12:c58       0 days 0:00:00.000
  11:1514      0 days 0:00:00.000
   9:13d4      0 days 0:00:00.000
   6:13c4      0 days 0:00:00.000
   5:5e0       0 days 0:00:00.000
   4:1214      0 days 0:00:00.000
   2:11e0      0 days 0:00:00.000
   1:1060      0 days 0:00:00.000
0:020:x86> vertarget
Windows 7 Version 7601 (Service Pack 1) MP (8 procs) Free x64
Product: WinNt, suite: SingleUserTS
kernel32.dll version: 6.1.7601.17651 (win7sp1_gdr.110715-1504)
Machine Name:
Debug session time: Fri Mar  2 14:23:24.614 2012 (UTC - 5:00)
System Uptime: 3 days 5:31:02.984
Process Uptime: 3 days 5:30:08.834
  Kernel time: 0 days 0:01:59.200
  User time: 0 days 4:14:00.283
0:020:x86> ~20k
ChildEBP RetAddr  
WARNING: Stack unwind information not available. Following frames may be wrong.
1e32fb9c 726226fc ntdll_779d0000!RtlEnterCriticalSection+0x34
1e32fbbc 6eb575e3 nspr4!PR_EnterMonitor+0x3c [e:\builds\moz2_slave\rel-m-rel-w32-bld\build\nsprpub\pr\src\threads\prmon.c @ 99]
1e32fbe0 6ecfb0be xul!nsThread::ProcessNextEvent+0xf3 [e:\builds\moz2_slave\rel-m-rel-w32-bld\build\xpcom\threads\nsthread.cpp @ 614]
1e32fc24 6eff383e xul!nsThread::GetObserver+0x25 [e:\builds\moz2_slave\rel-m-rel-w32-bld\build\xpcom\threads\nsthread.cpp @ 705]
1e32fc3c 6f4889c8 xul!nsBuiltinDecoderStateMachine::StopAudioThread+0x3e [e:\builds\moz2_slave\rel-m-rel-w32-bld\build\content\media\nsbuiltindecoderstatemachine.cpp @ 1075]
1e32fc6c 6f4cb52a xul!nsBuiltinDecoderStateMachine::DecodeSeek+0xab [e:\builds\moz2_slave\rel-m-rel-w32-bld\build\content\media\nsbuiltindecoderstatemachine.cpp @ 1324]
1e32fc88 6ef12081 xul!nsBuiltinDecoderStateMachine::DecodeThreadRun+0x51 [e:\builds\moz2_slave\rel-m-rel-w32-bld\build\content\media\nsbuiltindecoderstatemachine.cpp @ 310]
1e32fc8c 6eb5764f xul!nsRunnableMethodImpl<void (__thiscall nsBuiltinDecoder::*)(void),1>::Run+0xe [e:\builds\moz2_slave\rel-m-rel-w32-bld\build\obj-firefox\dist\include\nsthreadutils.h @ 346]
1e32fcb4 6ecfc566 xul!nsThread::ProcessNextEvent+0x15f [e:\builds\moz2_slave\rel-m-rel-w32-bld\build\xpcom\threads\nsthread.cpp @ 637]
1e32fcdc 72622150 xul!nsThreadStartupEvent::Run+0x25 [e:\builds\moz2_slave\rel-m-rel-w32-bld\build\xpcom\threads\nsthread.cpp @ 201]
1e32fcf8 735a2e24 nspr4!_PR_NativeRunThread+0x120 [e:\builds\moz2_slave\rel-m-rel-w32-bld\build\nsprpub\pr\src\threads\combined\pruthr.c @ 448]
1e32fd04 7262173d MSVCR80!initptd+0x131
1e32fd0c 735a29bb nspr4!pr_root+0xd [e:\builds\moz2_slave\rel-m-rel-w32-bld\build\nsprpub\pr\src\md\windows\w95thred.c @ 122]
1e32fd44 735a2a47 MSVCR80!endthreadex+0x3b
1e32fd58 77a09ef2 MSVCR80!endthreadex+0xc7
1e32fd98 77a09ec5 ntdll_779d0000!RtlInitializeExceptionChain+0x63
1e32fdb0 00000000 ntdll_779d0000!RtlInitializeExceptionChain+0x36


I've got a dump file here if you want but since my firefox is only v10 I'm going to guess somebody has already fixed the error and won't waste your disk space.  Reply to the bug thread if you would like me to upload the dump.

Comment 2

[Tyler Downer [He/Him]]

Please create a new profile, and test there.
https://support.mozilla.org/en-US/kb/Managing-profiles.

Also, please go to about:support, and copy and paste the info from there into a text document, and then attach it to this bug. Please also update all your plugins (Flash, Quicktime, adobe reader, Java etc.) to the most recent version available from the developer's website.

Comment 3

[Eric Harding]

Attachment: about:support

Comment 4

[Eric Harding]

Attachment: about:support - better format

Comment 5

[Tyler Downer [He/Him]]

Also, ensure your graphics card drive is up to date as well.

Comment 6

[Eric Harding]

I'm not sure creating a new profile and testing there is useful since this bug only happened once. 

There are no plugins on the offending stack trace.  The only plugins that I have installed are "Google Update" and "Silverlight 4.0.60831.0" all the rest are disabled.

Comment 7

[Eric Harding]

I realize you probably tell that to everyone as a base line diagnostic but I'm not going to change my video driver.  It's the latest update from windows update and has been stable for a good while.  This isn't a crash which is more likely from a bad video driver and the driver isn't on the stack. 

If you would like me to upload the 380 meg crash dump I can do that but I thought it would be fairly obvious to someone with the stack trace above.

Comment 8

[Tyler Downer [He/Him]]

Well the stack trace isn't too terribly helpful.
Please read https://developer.mozilla.org/en/How_to_get_a_stacktrace_with_WinDbg, and get a complete stacktrace for a hang there.

Since this isn't an easily reproducible hang, it will be hard, which is why I suggest that you get new graphics driver's from Nvidia's website (official ones, not beta) to help reduce the chance of a hardware acceleration issue, updating plugins to ensure they aren't interfering in some way, and testing with a fresh profile to ensure any user made settings or extensions aren't interfering.

Because you said this was over 3 days, it is very hard to collect reliable data, and since I probably won't be able to reproduce it, we have to try to get all the data we can from you to determine if it is an issue with Firefox, or third party software.

Comment 9

[Eric Harding]

Attachment: Additional stacks

Comment 10

[Eric Harding]

I've attached the additional stacks specified in your link as well as the list of modules.

Thread 20 is the one taking all the CPU time - you'll notice that it's in a slightly different place under StopAudioThread() than the first stack that I pasted.  After taking the dump I let the process spin for a few more seconds and stopped it again to produce that stack.

Comment 11

[Eric Harding]

Are the symbols from http://symbols.mozilla.org/firefox stripped? I can't seem to get type information for locals on the stack.

04 1e32fc24 6eff383e xul!nsThread::GetObserver+0x25 [e:\builds\moz2_slave\rel-m-rel-w32-bld\build\xpcom\threads\nsthread.cpp @ 705]
Type information missing error for this
Type information missing error for obs
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
05 1e32fc3c 6f4889c8 xul!nsBuiltinDecoderStateMachine::StopAudioThread+0x3e [e:\builds\moz2_slave\rel-m-rel-w32-bld\build\content\media\nsbuiltindecoderstatemachine.cpp @ 1075]
Type information missing error for this
Type information missing error for exitMon

Comment 12

[Eric Harding]

I just remembered - a couple days ago I unplugged a set of headphones from the port on the back of this machine.  I don't really have time right now to crawl under the machine and plug/unplug a few times but if you have the resources it might help to reproduce the problem.  Maybe some device that the audio thread wants no longer exists?

Comment 13

[Wayne Mery (:wsmwk)]

Eric, do you still see this?

(In reply to Eric Harding from comment #12)
> I just remembered - a couple days ago I unplugged a set of headphones from
> the port on the back of this machine.  I don't really have time right now to
> crawl under the machine and plug/unplug a few times but if you have the
> resources it might help to reproduce the problem.  Maybe some device that
> the audio thread wants no longer exists?

Comment 14

[Phoenix]

Resolved per whiteboard