Unprovoked shutdowns of Firefox

[shanen (Shannon Jacobs)]

Binary package hint: firefox-3.0

This is in Firefox 3.6.8, but may have begun with 3.6.7 under Ubuntu 9.04. (Later versions of Ubuntu have been tested on the machine, but this seems to be the most stable. The main stability problem is probably related to some kind of display bug (causing a white screen of death) discovered several years ago and still probably unresolved, though very rare under 9.04.)

Current problem is random and sometimes frequent unclean shutdowns of Firefox. No evident provocation. Sometimes nothing seems to be happening, sometimes I'm moving the mouse, or sometimes I've just clicked something, but no clear pattern at all.

I have attempted to upload the apport diagnostic information several times. Several times it clearly seemed to fail, though other times it seemed to succeed. Looking at that information, I was not able to see anything suggestive or helpful. I also studied the various logs without recognizing anything that seemed suspicious or possibly related to the Firefox problems. No new plugins have been added for a long time, though NoScript was recently updated (but without any sign of problems on several other machines that I use). This is the only machine I am running under Ubuntu 9.04, and I currently believe it is probably specific to this version and probably related to poor regression testing, which seems to be a growing problem in Ubuntu.

Having said all of that (and without a crash so far <knock on wood>), I don't much care anymore. I had very high hopes for Ubuntu to become a viable alternative to such ugly options as Windows. I really hoped that Ubuntu could help Linux become more than a tiny niche OS for geeks and nerds. My hopes were not dashed. More like chewed to pieces by catty bugs.

The situation has not improved on Ubuntu 9.10.

[Luis Aguiar (techm3)]

I have the same problem in 10.04.
Sometimes I'm doing important things and it just crashed magically.

[Micah Gersten (micahg)]

Thank you for reporting this to Ubuntu. Can you please generate a backtrace for this issue? https://wiki.ubuntu.com/Backtrace
Please substitute 'firefox -g' for 'gdb <program>' in the instructions.

[shanen (Shannon Jacobs) (shanen)]

No, I cannot do that because your (Micah Gersten's) instructions (in the previous message and in the linked webpage) are incomplete or incorrect. I have just wasted another half hour or so trying to figure out various combinations of the most likely corrections.

I made that much effort in an attempt to be polite, not because I have any residual enthusiasm for helping Ubuntu. I suppose you are sincere and you believed that the instructions were sufficient. However, the net result is to reduce my appreciation for Ubuntu even lower.

[shanen (Shannon Jacobs) (shanen)]

By the way, even if I had succeeded, I still do not have any reliable mechanism with which to trigger the bug. It happens when it feels like happening. The last time just happened to be when I was looking at the email that told me you had posted your comment to Launchpad. Sometimes it happens several times in an hour, and other times it doesn't happen for several hours.

[madbiologist (me-again)]

Please run apport-collect 612131

Is this a desktop or laptop machine?

Please also attach the output of lspci -vvnn

[shanen (Shannon Jacobs) (shanen)]

Thank you (Mr Gersten) for linking from the other bug. If I can at least temporarily overcome my increasingly hopeless feelings towards Ubuntu, I may also attempt the steps recommended by "madbiologist" (since the backtrace approach failed (and going even farther by the way, the backtrace had probably failed even earlier, since the software had already been installed at some point before I attempted to follow the most recent instructions)).

I do note that this bug report seems to be linked to firefox 3.0, and I'm sure that the current version of the bug on the afflicted machine is NOT 3.0, and did not even appear until 3.5 or later. It doesn't appear that any of the latest "Report Problem..." information has been copied over to this bug. However, I've made a note of this bug number, and I should have plenty of opportunities to get that information again. Firefox shut down at least three times this morning during a short session on that machine.

Just in case it's related, I'll report this symptom though I doubt it is directly related to this bug. Rather I think this belongs to an earlier Java-related bug which I never even bothered to report. That Java bug had gotten worse about the time this bug appeared, but after the upgrade from 9.10 to 10.4, it has returned to its previous state. At least I don't remember ever reporting this one, but it goes far enough back that I may have reported it during my 'youthful enthusiasm for Ubuntu' phase. The origins may have involved extensive difficulties installing Java to execute within Firefox for the use of such <sarcasm>trivial and unknown</sarcasm> websites as Yahoo. (Did the sarcasm pseudo-tag survive? If not, take it as a micro-review of the Yahoo portal. Yahoo certainly deserves to be regarded as trivial and unknown for their ongoing support of spammers.) Anyway, Java does work, but there is an initialization delay of several minutes each time before it is ready to function normally. This has been it's stable-but-buggy state for a long time, going back well before this bug. However, at some point, probably close to the appearance of this bug, Java became completely disabled within Firefox. At that point, I simply regarded it as yet another lack-of-usability problem of Ubuntu, and I didn't even report it (unless I was venting and have already forgotten the vent). I simply stopped attempting to use any webpage that used Java on the afflicted machine. (There are no Java-in-Firefox problems on other machines, but I don't recall if I ever tried it under Ubuntu, though I'm sure it was okay for Windows.) As noted early in this paragraph, after the upgrade from 9.10 to 10.4, Java was again running within Firefox, though once again with the long delay for some kind of initialization. I haven't timed it exactly, but it's about three or four minutes. Some kind of network timeout before a fallback? I haven't cared enough to investigate. I just work in another program or on another machine for a while. During this weird initialization period, Firefox is completely unusable on the afflicted machine. (Users are so adaptable, eh? I now apparently regard it as "normal" that hitting a webpage that invokes Java will...

[shanen (Shannon Jacobs) (shanen)]

WA70:~$ lspci -vvnn
00:00.0 Host bridge [0600]: ATI Technologies Inc RS480 Host Bridge [1002:5950] (rev 01)
 Subsystem: Sharp corporation Device [13bd:104c]
 Control: I/O- Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B- DisINTx-
 Status: Cap- 66MHz+ UDF- FastB2B- ParErr- DEVSEL=medium >TAbort- <TAbort- <MAbort+ >SERR- <PERR- INTx-
 Latency: 64
 Kernel modules: ati-agp

00:01.0 PCI bridge [0604]: ATI Technologies Inc RS480 PCI Bridge [1002:5a3f]
 Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR+ FastB2B- DisINTx-
 Status: Cap+ 66MHz+ UDF- FastB2B- ParErr- DEVSEL=medium >TAbort- <TAbort- <MAbort- >SERR- <PERR- INTx-
 Latency: 128
 Bus: primary=00, secondary=01, subordinate=01, sec-latency=32
 I/O behind bridge: 0000c000-0000dfff
 Memory behind bridge: c0000000-cfffffff
 Prefetchable memory behind bridge: 0000000090000000-000000009fffffff
 Secondary status: 66MHz+ FastB2B- ParErr- DEVSEL=medium >TAbort- <TAbort- <MAbort+ <SERR- <PERR-
 BridgeCtl: Parity- SERR+ NoISA- VGA+ MAbort- >Reset- FastB2B-
  PriDiscTmr- SecDiscTmr- DiscTmrStat- DiscTmrSERREn-
 Capabilities: <access denied>
 Kernel modules: shpchp

00:14.0 CardBus bridge [0607]: Texas Instruments PCI4510 PC card Cardbus Controller [104c:ac44] (rev 03)
 Subsystem: Sharp corporation Device [13bd:104c]
 Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B- DisINTx-
 Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=medium >TAbort- <TAbort- <MAbort- >SERR- <PERR- INTx-
 Latency: 168, Cache Line Size: 32 bytes
 Interrupt: pin A routed to IRQ 17
 Region 0: Memory at 28000000 (32-bit, non-prefetchable) [size=4K]
 Bus: primary=00, secondary=03, subordinate=06, sec-latency=176
 Memory window 0: 20000000-23fff000 (prefetchable)
 Memory window 1: 24000000-27fff000
 I/O window 0: 00001800-000018ff
 I/O window 1: 00001c00-00001cff
 BridgeCtl: Parity- SERR- ISA- VGA- MAbort- >Reset+ 16bInt+ PostWrite+
 16-bit legacy interface ports at 0001
 Kernel driver in use: yenta_cardbus
 Kernel modules: yenta_socket

00:14.1 FireWire (IEEE 1394) [0c00]: Texas Instruments PCI4510 IEEE-1394 Controller [104c:8029] (rev 01) (prog-if 10)
 Subsystem: Sharp corporation Device [13bd:104c]
 Control: I/O- Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B- DisINTx-
 Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=medium >TAbort- <TAbort- <MAbort- >SERR- <PERR- INTx-
 Latency: 128 (500ns min, 1000ns max), Cache Line Size: 16 bytes
 Interrupt: pin B routed to IRQ 19
 Region 0: Memory at d0000000 (32-bit, non-prefetchable) [size=2K]
 Region 1: Memory at d0004000 (32-bit, non-prefetchable) [size=16K]
 Capabilities: <access denied>
 Kernel driver in use: ohci1394
 Kernel modules: firewire-ohci, ohci1394

00:15.0 Multimedia video controller [0400]: Conexant Systems, Inc. CX23880/1/2/3 PCI Video and Audio Decoder [14f1:8800] (rev 05)
 Subsystem: Sharp corporation Device [13bd:104c]
 Control: I/O- Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B- DisINTx-
 Status: Cap+ 66MHz- UDF- FastB2B+ ParErr- DEVSEL=medium >TAbort- <TAbort- <MAbort- >SERR- <PERR- INTx-
 Latency: 128 (50...

[shanen (Shannon Jacobs) (shanen)]

The attempt to add the basic "Report a Problem..." information failed, as did the attempt to use the apport-collect 612131. This was after the first unprovoked shutdown of the evening, which was actually quite slow--already over an hour.

[madbiologist (me-again)]

Thanks. It seems you are using a Sharp laptop with an Athlon64 CPU and ATI Radeon XPRESS 200M graphics. Not the most powerful graphics around, but it should be plenty to run firefox and has the benefit of good open-source driver support.

Sorry if I wasn't clear - you can run apport-collect 612131 at any time - you don't need to wait for an unprovoked shutdown. Or perhaps the reason it didn't work was because of the mismatch between your Firefox version and the Firefox version this bug was marked as affecting. I've changed that now, so could you please try again? To make sure I've got the syntax correct I'll run it myself in minute.

[madbiologist (me-again)]

OK, it looks I have the apport-collect syntax correct, but I/we are unable to file a bug because we are using Firefox 3.6.x. I got this error when running apport-collect 612131 in the terminal:

Package firefox-3.5 not installed and no hook available, ignoring

Launchpad does not have a Firefox 3.6 package option.

I'll file a new bug about this.

[madbiologist (me-again)]

According to https://wiki.ubuntu.com/Bugs/FindRightPackage this bug should be filed in the package "firefox". Updating (and cancelling my new bug).

[madbiologist (me-again)]

Architecture: i386
DistroRelease: Ubuntu 10.10
FirefoxPackages:
 firefox 3.6.8+build1+nobinonly-0ubuntu1
 firefox-gnome-support 3.6.8+build1+nobinonly-0ubuntu1
 firefox-branding 3.6.8+build1+nobinonly-0ubuntu1
 abroswer N/A
 abrowser-branding N/A
LiveMediaBuild: Ubuntu 10.10 "Maverick Meerkat" - Alpha i386 (20100630.2)
Package: firefox 3.6.8+build1+nobinonly-0ubuntu1
PackageArchitecture: i386
ProcEnviron:
 LANG=en_US.UTF-8
 SHELL=/bin/bash
ProcVersionSignature: Ubuntu 2.6.35-6.8-generic 2.6.35-rc3
Tags: maverick
Uname: Linux 2.6.35-6-generic i686
UserGroups: adm admin cdrom dialout lpadmin plugdev sambashare

[madbiologist (me-again)]

apport information

[madbiologist (me-again)]

apport information

[madbiologist (me-again)]

apport information

[madbiologist (me-again)]

apport information

[madbiologist (me-again)]

OK, now it works. Can you please run apport-collect 612131 again as per comment 9?

[madbiologist (me-again)]

After a private communication from Micah Gersten I now know that this bug should be filed in the package firefox-3.0, which is the firefox version shipped with Ubuntu 9.04, despite you using Firefox 3.6.x.

The correct command for you to use is apport-collect -p firefox 612131

[shanen (Shannon Jacobs) (shanen)]

Sorry I've been kind of busy and haven't gotten back to you on this. I really do appreciate your attempt to follow up on it, and I value your technical acumen, but like many people, I have competing priorities. I've already noted why Ubuntu has become a relatively low one for me. More importantly, it seems that almost no one else is seeing anything similar enough to this problem to add their comments in this thread, so I'm willing to assume it is just something I'm doing wrong or something unusual about my configuration on that particular machine. If it will help diagnose a more widespread problem, then I'm willing to put more time into it (but I'm still going to be hard pressed to give it much priority).

I don't understand your explanation of the Firefox 3.0 versus 3.5 thing. This machine was almost surely originally a Firefox 2.0 machine, and now that it's been upgraded to 9.10, I don't see how Firefox 3.0 is involved. (However, I do see how that might prevent other people with similar problems from finding this thread, since it is absolutely clear the problem did not appear until the Firefox upgrade, probably to version 3.6.)

What I can say is that this particular shutdown seems to be less frequent now. Unfortunately, I can't correlate it to anything that has changed, and I never had the bug in a can. Without a trigger mechanism to reproduce the bug on demand, who can tell anything? Unfortunately, as I had suspected, the WSOD problem has become more frequent with the updated version of Ubuntu on that machine. Right now both of these crashes are rare enough to tolerate, but I'm knocking on wood. Especially for the WSOD, it wouldn't take too much of an increase to make the machine unusable.

Big picture solution? I'm probably going to try forcing the main experimental partition on the troubled machine to a different version of Ubuntu at some point after the next release (after waiting long enough for the early mortality bugs to be eliminated). However, when I replace the machine, which will probably be early next year, I have no plans to do any serious Ubuntu work on the new one. I probably won't even bother with the virtual machine approach, given the problems encountered on three machines where I've been using Ubuntu in VMware Player virtual machines... (I still have no Windows 7 machines of my own, but at least it isn't an obvious stinker like Vista--and Linux has (in my judgment) returned to the not-ready-for-real-world-use status. I'm saddened and disappointed, but...)

(By the way, what's gone wrong with the BitTorrent download of the basic 10.04? The 9.08 version is apparently still valid, but 10.04 is being disavowed by the Ubuntu.com website? I was still sharing both seeds, but now the 10.04 is just an error message, and I'm sort of guessing that the release version of 10.04 was so bad that it's been replaced, and the Ubuntu people just don't want to talk about it. I haven't been motivated enough to pursue another Ubuntu problem now...)

[shanen (Shannon Jacobs) (shanen)]

Just stopped by to see if anyone else has reported it. Amusingly enough, I asked the other local user of Ubuntu about it, and he said his Firefox does it, too--but he's never bothered to report it. He just figured it was because he has too many Firefox windows open, which got me to try the experiment of running with one window and doing everything in tabs--but it still crashed.

Actually, in my case the crashes have become more frequent again. I still use the Ubuntu machine every day--but not for anything serious or important. I have to assume that Firefox may shut down at any time and for no apparent reason.

Meanwhile, I have started looking at Windows 7 machines, but not too seriously yet... I'm going to wait for one more upgrade cycle (as in next month) to see which way things are going. Unless there is clear evidence that Ubuntu has turned the corner, I'll just give up on the machine and let it take Ubuntu with it.

On the other hand, if I only have direct access to one Firefox on Ubuntu user and if he has a similar problem to this one, maybe I should wonder about the frequency of the problem? Maybe the surviving Ubuntu users are just becoming accustomed to this sort of flakiness?

[madbiologist (me-again)]

Sorry for taking so long to get back to you on this. I think I may have missed the notification email for comment #19 due to my inbox being full a few times recently.

I'll deal with the 10.04 BitTorrent question first. Although I don't use BitTorrent to get Ubuntu releases myself, I strongly suspect that the reason that 10.04 has errored out is because Ubuntu 10.04.1 was released on the 17th August as per the schedule at https://wiki.ubuntu.com/MaverickReleaseSchedule
Note that this is actually the release schedule for the upcoming Ubuntu 10.10 "Maverick Meerkat", but towards the bottom of the Notes column it also mentions 10.04.1. 10.04.1 was planned from the beginning, prior to the release of 10.04, due to 10.04 being an LTS (Long Term Support) release. Note that this does not necessarily mean that the original 10.04 is not available for download, the BitTorrent details may simply have been changed.

If I've followed you correctly, the machine with the crashing Firefox is currently running Ubuntu 9.10. According to https://launchpad.net/ubuntu/+source/firefox-3.5 the default version of Firefox in 9.10 was Firefox 3.5.3, so I'm re-targeting this bug to that package (as shown in the Affects column at the top of the page). Unless I've completely botched things up the correct command for you to use should still be apport-collect -p firefox 612131 (if you wish to).

[madbiologist (me-again)]

I'm tempted to ask what the WSOD (Workspace on demand?) problem is, as I can't see where you mentioned it previously (although it's getting late and I'm tired so I may have overlooked something), however if it's a different bug it doesn't really belong here and should be filed as a separate bug (if you can't find an existing bug report for that issue).

I'm sorry to hear that the Firefox crashes have become more frequent again. Something to note is that although the older ATI Radeon hardware has good open-source driver support (as I mentioned in comment #9), the support for the actual Radeon XPRESS 200M does seem to have a few quirky bugs which have proven difficult to solve. Since writing comment #9 I've come across one or two bugs where quite a few people are having problems with the display of Google Earth on the Radeon XPRESS 200M. I can't remember if the Radeon XPRESS 200M uses sideport memory or not, but there is a DRM fix for sideport memory in kernel 2.6.35, which is the default kernel in the upcoming Ubuntu 10.10 "Maverick Meerkat". If you are feeling adventurous you can try this kernel on your 9.10 installation - the 2.6.35.4 kernel PPA is at http://kernel.ubuntu.com/~kernel-ppa/mainline/

Are you able to try adding a discrete graphics card to that machine to see if things improve? Depending on the motherboard, doing so might automatically disable the integrated Radeon XPRESS 200M, otherwise you will need to disable it in the BIOS and enable the discrete graphics card.

Also, I've used several versions of Firefox on the last 3 versions of Ubuntu (I'm counting the 10.10 alphas and the 10.10 beta as the third version), and prior to that on Windows and never had an issue with regular/frequent Firefox crashes. Recently, for the first time I had quite a few crashes with Firefox 3.6.9, however this has mostly gone away with the latest Firefox 3.6.10 update, so I encourage you to try that.

Hopefully the number of crashes is reduced once Firefox 3.6.10 is installed. Either way, let us know, and then if you are able to, try disabling extensions one by one and see if things improve. Note that the apport-collect -p firefox 612131 command I mentioned in comment #20 will give us a list of all of your extensions, among other things.

I presume your Flash player version is up to date? Also, I just had a thought - are you using Ubuntu 32bit or 64bit? If it's Ubuntu 64bit, are you using the 32bit Flash player or the 64bit Flash player? Adobe neglected 64bit Flash player development for Linux for a while, and people using the old development versions of the Linux 64bit Flash player were reporting crashes. Adobe have recently addressed this situation with a new Linux 64bit Flash player - see http://www.phoronix.com/scan.php?page=news_item&px=ODYwMg

Finally, on the Windows XP (SP2 I think, or maybe SP3?) machines where I work, we use Internet Explorer 8 quite heavily and it is not the most stable application I've seen. Each user including myself (and I always try to be careful) gets about 5 or 6 IE8 crashes per day.