Crash to white screen of death (possibly Firefox?)

I forgot a couple of things that might be diagnostic hints. Since upgrading to Gutsy I have noticed a number of problems that seem to involve the underlying windowing system. The most common symptom is an inability to select a window and bring it to the front with the focus. Usually this involves Firefox, which is the only application that I'm using most of the time. There will be two partially overlapping windows, but when I click on the lower window, it does not come up, though it is scrollable at that point. (I can use the mousewheel to make the contents of the partially covered window move up and down.) When that happens, the workaround I've been using is to switch to another application (usually terminal) and then return to the lower window, and it will come to the front properly.

Probably not related, but from tracking the swap file usage it seems to me that Gutsy is also having worse problems with memory leaks. I don't feel like that's related to the problem I'm reporting here. On the other hand, this machine has quite a bit of RAM so maybe it's having memory problems before it ever gets to swapping. I've mostly been watching the memory leak problems on a different machine--but that one has not yet experienced the white screen of death.

Just had another one, and still no idea what's provoking it and no ideas about how to diagnose the cause. Only thing I noticed this time was that Firefox had clearly lost some of the status information that should have been recorded in that session.

No one else is seeing anything like this? No one has any diagnostic hints?

Again not sure if this is related, but if so, it suggests the problem may be deeper than Firefox... Somehow I suspect it may be related to the windowing system itself. Concretely, it also lost the language information. Not just the Japanese stuff, but all of the language support wanted to be reinstalled.

Hi,

  could you have a look at the following Gutsy release notes : https://wiki.ubuntu.com/GutsyGibbon/ReleaseNotes

particularly :Blank screen with some ATI hardware
      People with ATI display adapters may get a blank screen when loading X due to the driver being unable to initialize certain hardware. Upstream is working on it, and hopefully we'll be able to release an update for 7.10 soon after the release. In the meantime, add 'Option "LVDSBiosNativeMode" "false"' to the driver section of xorg.conf. Bug #132716

I read all of those comments, but none of them seemed to apply to this problem on a Sharp notebook (PC-WA70L). However, it hasn't happened recently, so maybe it was repaired in some of the recent patches. It was never very common, though the last few times I saw it, it seemed slightly different. The machine was still completely dead and had to be powered down, but the early times the final image was a completely white screen, while the last few times it was a white screen with black pinstripes.

After the first pinstripe display, I switched to the non-restricted video driver, but that didn't help, and it crashed a couple more times after that, still with the pinstripe display. I was hoping to get some guidance about where to search in the logs for revealing information, but I couldn't find any on my own at the time, and it's almost surely too late now, unless the logs go back a long ways and you can tell me about something very specific to look for.

Hopefully that problem has gone away. It was the most serious new glitch of Gutsy. There are still some other glitches here, but stuff I can live with, though overall the Gutsy experience has been another small step down...

Problem has gotten worse, with two of the crashes in the last few minutes. That was after a long period with no such crashes, so obviously I suspect one of the recent patches.

If this keeps up, I'm going to have to go back to Windows for a while and let y'all try to sort it out. I'm willing to try to do a few diagnostic tests or search through some logs if someone will provide clear directions about how to test or what to look for. I am not willing to make a major effort and spend lots of time searching for your bugs and I am especially not willing to live in fear of losing my work at any moment.

Do you have some kind of coverage matrix about what machines have actually been tested and are known to be safe? My basic conclusion is that this Sharp PC-WA70L is *NOT* a safe machine for use with Ubuntu, though most of the other machines I've worked with seem to be okay. (Can't be sure, since the crash is so sporadic and since this is the machine I've been using most.)

Just had another one. Pretty certain that the only applications that had run since booting were Firefox and terminal. Searched the logs carefully. There was a "MARK" at 5:36, and the reboot began at 5:39. None of the logs seemed to reveal anything about what had gone wrong.

Lost some more work, and this is becoming pretty unacceptable for a "production" system.

I think the one a few minutes ago may have been the first time when the crash happened when I was using some other program, though Firefox was open in the other workspace at the time. Frequency is not too high right now, but still annoying, and stil not able to think of any useful diagnostic steps.

Reading some more stuff, and I'm increasingly suspicious that the actual bug may be in compiz. Is it possible to remove compiz from Gutsy if I'm not using any of the fancy features? Or is it possible to diagnose the crash more precisely if I suspect it's in compiz?

Part of this hypothesis is that the hard crash may be related to focal problems among the windows, which I'm pretty sure are related to compiz. Again, removing compiz might be a solution.

compiz does seem to have some serious stability issues for many people
you can remove it by disabling the visual effects in System->Preferences->Appearance->None

At this point I think compiz is not related to the problem. If I understand the situation correctly, my display settings are not using compiz at all--though some of the component files are presumably present. I'm still seeing the crash from time to time, but I also doubt it is related to Firefox, since there have been a lot of Firefox upgrades in the months since I first saw it. Sometimes I can hope it has gone away--but so far it has always returned. I had another in the last day or two. White screen, fan running, totally frozen. (I hope the running fan means it won't burn up in this condition, but maybe it's symptomatic.)

So right now I'm just hoping it goes away when the Heron is released in April...

Recent patches have evidently made the problem worse. Seeing one of these almost every day. Only new wrinkle was that the last one wasn't completely white. There was a bit of stuff like a bar code in the lower right corner, so apparently part of the video memory had been walked on?

Second time in the hour. The machine is approaching unusable.

I have the same card in some of my computers...the past couple of drivers have been very buggy for this card. The bar code thing was fixed in last months ATI driver, but you won't be able to get it in the gusty repositories...you have to get the driver directly from ATI.
I think the best solution would be to give it a try and see if it fixes the issue. A good how-to is on this website: <a href=http://wiki.cchtml.com/index.php/Ubuntu_Gutsy_Installation_Guide>http://wiki.cchtml.com/index.php/Ubuntu_Gutsy_Installation_Guide</a>

Sorry, wasn't sure how to a link:
http://wiki.cchtml.com/index.php/Ubuntu_Gutsy_Installation_Guide

also by any chance are you using the ubuntu ati restricted driver and if so what version?

Yes, it says that I am using a restricted driver from the Restricted Drivers Manager, but I haven't been able to figure out what version it is. Hardware Manager is apparently not the proper place to look.

Anyway, at this point I'm just waiting for the next release and hoping that will make it better rather than worse. I'm pretty sure I can live with it for another month...

You should probably change the status of this bug report.
The ATI restricted driver is not maintained by AMD not Ubuntu developers.

you can find out what version you are using by typing in a terminal:
dmesg | grep 'fglrx'

you should see a line that looks like this:
[ 52.012785] [fglrx] module loaded - fglrx 8.45.4 [Jan 16 2008] on minor 0
where the 8.45.4 would be where your version number is.

Another thing you could try is to disable the restricted driver and try the vesa driver or the non-restricted driver

I don' t know how to change the status of the bug, but on your advice, I changed to the non-restricted driver--and saw the same crash. Well, pretty sure it was the same one, though this time the white screen had widely spaced little stripes from top to bottom. Anyway, it was a good idea, but doesn't seem to have provided much in the way of valuable data.

Hadn't seen this for a while, but it's back again. At least twice in the last few days. I sure wish there was some way to diagnose it, but it seems that all I can do is keep my eyes open and look for some pattern--and I'm not spotting anything obvious. Something in the officially released version of Firefox 3?

Current frequency seems to be about once per 6 hours. However, at least they've fixed the recovery part via fsck. It doesn't need the manual intervention now, but will recover by itself after two reboots.

copied over from a duplicate:

This is the same bug I reported earlier as the white screen of death:

https://bugs.launchpad.net/ubuntu/+bug/159594

In that report, I speculated that it might be Firefox-related, but now I'm pretty certain that it is not related to Firefox, but is most visible there because that's the main application I use on the afflicted machine.

The crash usually results in a totally white screen, though I have seen at least one case of a bar code like pattern in the lower right, and one case where the entire display was striped with various colors. In every case the machine is completely hosed. There is nothing to be done but cut the power and reboot. Near as I can recall, the fan is always running when it is in that state. I hope that is not symptomatic of any thermal problem or related to the crash, but at least the machine hasn't started smoking (yet).

I have not been able to find any particular stimulus that makes the crash reproducible. However, I am fairly sure it is related to mouse movements and that the crash can only happen when the mouse is moving. The visible display (of the white screen) may be related to a driver conflict, but that's also kind of speculative. However, I strongly suspect the problem is related to window-focus problems the machine exhibits pretty often. These are normally recoverable events, but I suspect the crash is a case when the same root problem is not caught at the recoverable condition, but just goes all the way to death.

The frequency of the crash is around once per day, though sometimes I won't see it for several days, and other days there may be two of them. I don't recall seeing more than two in a short period, but it's already to the point where I feel I can't rely on the machine for any serious work in Linux--it can crash at any time.

As noted in the other bug report, I'm looking for any kind of diagnostic data or some way to make the bug reproducible.

The afflicted machine is my only Sharp running Linux, and is also the only one with an AMD CPU. Various other distinctions, including a TV tuner (which I'm not trying to use from the Linux side).

Hadn't seen it for a while, but just had one a few minutes ago. Still don't know what diagnostic information to collect, and still can't stop any particular trigger event.

Just had another one, and there was a day last week when I had about three of them in just a few hours. Seems like it's getting worse again. However, it also seems like no one else is seeing it... Not sure what that signifies, but I don't feel like this is an especially odd machine. Perhaps the most unusual feature is that the CPU is an AMD Sempron? The ATI graphics are pretty standard...

Getting really bad these days, with increasingly frequent crashes. Minor variant is that not all of them are white screens now, but sometimes pastel colors of various types.

Shannon could you try something:
Add the following line to ~home/.gnomerc

export WINDOW_MANAGER=metacity

If this solves your problem its a compiz issue (I've had the white screen more times than I care to talk about in Gutsy)

At this point, are you still using Gutsy, or have you upgraded to Hardy? Could you also test it with Intrepid alpha 6 (perhaps dual boot?)?

There is no .gnomerc file here, and I have upgraded to Hardy. The problem is rather sporadic, and so far I haven't been able to spot any sure-fire trigger. Sometimes it doesn't appear for weeks, and other times it will happen several times within a few hours.

Create the file, and add the contents to it, then restart GDM 'sudo
/etc/init.d/gdm restart'

----- Original Message -----
From: "shanen (Shannon Jacobs)" <email address hidden>
To: <email address hidden>
Sent: Monday, September 22, 2008 4:43 PM
Subject: [Bug 159594] Re: Crash to white screen of death (possibly Firefox?)

There is no .gnomerc file here, and I have upgraded to Hardy. The
problem is rather sporadic, and so far I haven't been able to spot any
sure-fire trigger. Sometimes it doesn't appear for weeks, and other
times it will happen several times within a few hours.

--
Crash to white screen of death (possibly Firefox?)
https://bugs.launchpad.net/bugs/159594
You received this bug notification because you are a direct subscriber
of the bug.

Just had another one. At least three or four crashes in the last few hours...

I'm trying your .gnomerc solution, though I'm not happy about the idea of winding up in a unique configuration that I don't really know how to maintain....

if i interpret correctly this results in disabling compiz, which can be achieved by disabling visual effects in the appearance properties as well.

additionally it has been suggested:

"Run a full complex pass of memtest and also a stress testing application for the CPU / CACHE and GPU.

If the hardware proves itself to be reliable under stress tests, I would suggest he considers a clean, fresh install of the latest Hardy point release and see if the problem still replicates itself. If it does, steps to replicate the problem need to be narrowed down."

-----Original Message-----
From: <email address hidden> [mailto:<email address hidden>] On Behalf Of Michael Nagel
Sent: Wednesday, September 24, 2008 9:40 AM
To: <email address hidden>
Subject: [Bug 159594] Re: Crash to white screen of death (possibly Firefox?)

>if i interpret correctly this results in disabling compiz, which can be
>achieved by disabling visual effects in the appearance properties as
>well.

-Yep, but what if one can't login due to GUI issues.

>additionally it has been suggested:

>"Run a full complex pass of memtest and also a stress testing
>application for the CPU / CACHE and GPU.

I would definitely do this as well.

>If the hardware proves itself to be reliable under stress tests, I would
>suggest he considers a clean, fresh install of the latest Hardy point
>release and see if the problem still replicates itself. If it does,
>steps to replicate the problem need to be narrowed down."

Before that, Id obtain stack/back traces from X.
http://wiki.ubuntu.com/DebuggingProgramCrash

--
Crash to white screen of death (possibly Firefox?)
https://bugs.launchpad.net/bugs/159594
You received this bug notification because you are a direct subscriber
of the bug.

Well, I can now confirm that the .gnomerc approach did not solve the problem. There was one crash since I added that file. Just to confirm I did it right, here is the cat result:

shanen@shanen-HH-WA70:~$ cat .gnomerc
export WINDOW_MANAGER=metacity

I need to run another memory check from GRUB, but I basically don't believe it is hardware. Not sure because I think the best check is running in Windows, and I haven't been running the machine in Windows very much. What I can say is that I have never seen the crash when I was running in Windows, and there was about a year when I only used the machine in Windows. The crashes started only after Ubuntu was installed.

With regards to the recommended "stack/back traces from X", this part seems too complicated for me, though I may tackle it again. However, what I'd much prefer is that I *NOT* spend lots of time trying to fix Ubuntu's problems. It's bad enough when I lose work in a crash--and I'm getting antsy now and feel the pressure to save this before it crashes.

Look. Ubuntu is realizing that a crash has occurred. After the crash it does the disk scan and an extra reboot before it's ready to go again. That's the place where it should trigger it's debugging procedures and install whatever needs to be installed to do fancy traces or backtraces or whatever. If they can't fully automate it, at least the system should report the problem to the Ubuntu people and they should be able to get back to me (assuming there are enough similar crash events to be worth worrying about).

Anyway, time to save it before it crashes again...

I agree fully in that back traces should be automated, and allot of people work very hard on making ubuntu better, for free. I hope you take the time to get help obtaining the information needed to better diagnose your issue, this is how the open source model works, people write the software for free, people fix the bugs for free, people report and gather information needed for these developers for free.
Good luck with it, and I hope this doesn't cause you to leave Linux behind....

-----Original Message-----
From: <email address hidden> [mailto:<email address hidden>] On Behalf Of shanen (Shannon Jacobs)
Sent: Friday, September 26, 2008 5:49 PM
To: <email address hidden>
Subject: [Bug 159594] Re: Crash to white screen of death (possibly Firefox?)

Well, I can now confirm that the .gnomerc approach did not solve the
problem. There was one crash since I added that file. Just to confirm I
did it right, here is the cat result:

shanen@shanen-HH-WA70:~$ cat .gnomerc
export WINDOW_MANAGER=metacity

I need to run another memory check from GRUB, but I basically don't
believe it is hardware. Not sure because I think the best check is
running in Windows, and I haven't been running the machine in Windows
very much. What I can say is that I have never seen the crash when I was
running in Windows, and there was about a year when I only used the
machine in Windows. The crashes started only after Ubuntu was installed.

With regards to the recommended "stack/back traces from X", this part
seems too complicated for me, though I may tackle it again. However,
what I'd much prefer is that I *NOT* spend lots of time trying to fix
Ubuntu's problems. It's bad enough when I lose work in a crash--and I'm
getting antsy now and feel the pressure to save this before it crashes.

Look. Ubuntu is realizing that a crash has occurred. After the crash it
does the disk scan and an extra reboot before it's ready to go again.
That's the place where it should trigger it's debugging procedures and
install whatever needs to be installed to do fancy traces or backtraces
or whatever. If they can't fully automate it, at least the system should
report the problem to the Ubuntu people and they should be able to get
back to me (assuming there are enough similar crash events to be worth
worrying about).

Anyway, time to save it before it crashes again...

--
Crash to white screen of death (possibly Firefox?)
https://bugs.launchpad.net/bugs/159594
You received this bug notification because you are a direct subscriber
of the bug.

The frequency of the crashes is increasing again after a low-crash period. Lovely pastel death this morning.

Right now I have to regard the machine when running under Ubuntu as unsuitable for any serious work (unless the working program includes a Gmail-like remote backup every few minutes). This problem now goes back at least two major releases, and we're about to enter a third. I sure hope it goes away--but I can't be optimistic. I have been unable to diagnose or fix it myself and unable to help anyone else do anything about it.

I think I understand how open source works, and I even work in the food chain of one of the companies that is a big contributor to open source, especially Apache and Eclipse. We even have an in-house Linux distro that I use some of the time (and don't like much). However, I personally am *NOT* in a position to do much to help you here. Back in my programming days I was just a database programmer. I did a bit of assembly-level hacking, but I was never much good at it. I do think a serious hacker could probably pin this bug down. I don't have a reliably trigger, but the frequency is pretty high...

Unfortunately, it seems that there isn't a serious hacker using Ubuntu on these models of Sharp. As the situation is, I'm actually considering buying a new machine to get away from the problems with Ubuntu. I've also been experimenting with other distros, but haven't found any that I like as much as Ubuntu, so I'm feeling like just throwing money at the problem... (I already have several personal machines, and many others at work, but this Sharp with the AMD Sempron CPU is the only egregious troublemaker.)

Bug has survived the transition to Ibex, apparently. I initially did the upgrade to a scratch partition and ran it for a few days without seeing the crash, but as soon as I upgraded my main working partition, it returned. At least twice in the last few days since I did the upgrade.

I noticed that Ibex had installed a new proprietary driver for the ATI chipset, so I've disabled that now, and I'm waiting to see if the crash recurs. Unfortunately the frequency of the crash is so low that I will never be sure it's gone... At least that's how I feel about it.

Seems to be worse than usual now. Already two crashes this morning.

At this point I'm teetering on giving up Ubuntu. It's not very useful when the machine can crash at any moment, is it?

I suppose the alternative would be to try a different computer, but that sort of cancels the argument that Linux is cost competitive, eh? Actually, I already tried to find some information about computers and brands that are recommended with Linux, but I haven't been about to find much clear information. Ditto on information about what to avoid. Maybe this machine is a bummer because of the AMD CPU? I still suspect it's mostly Sharp's fondness for non-standard tweaks.

Knocking on wood, but it seems that Jackalope has fixed this bug. I haven't seen any of these crashes since I upgraded to Jackalope.

I should have known better to say anything...

It's back! Unprovoked and unexplained white screen of death.

Oh well. Overall I still think Jaunty Jackalope is generally better than the last few releases of Ubuntu for this troublesome machine. Time to try a complete install from scratch?

Thanks for your bug report. The symptoms provided indicate this is a problem with your video drivers and is being reassigned to the xorg package. For more information on problems with visual effects and video drivers see http://wiki.ubuntu.com/VisualEffects.

Just had another one, but that's only the second time since switching to Jackelope, so I feel like it's mostly cured. However, I'll say this was an especially nasty one to fix. The automatic recovery did not work and after several failed iterations, I remembered to try a manual fsck, which seems to have taken care of it (for now).

Please run 'apport-collect 159594' to attach the relevant logs etc to this bug.

I'm afraid to say anything, since I haven't seen the bug in a long time. Let me knock on some wood before continuing...

However, from looking at this thread now, I'm pretty sure you were talking to someone else, probably Bryce Harrington. Do you want me to try to run the same information? I don't know if it would help, because in my case it would now appear to be trying to prove a negative.

<knocks on wood again...> It really was a pretty annoying bug when it was happening.

I finally got this bug into a can yesterday. I thought it had gone away, but it started showing up again, though very rarely. Then yesterday, I found a way to reproduce it. At least it worked twice in a row. If someone will tell me what to do, maybe it will be possible to diagnose it now.

The situation that triggers the bug is on a completely fresh installation of the latest Ubuntu 9.10, with the updates to yesterday. As soon as the machine is booted, start Firefox, and then open a terminal window. The command "sudo update-grub" will ask for the password and then immediately go to the white screen of death. I'm not sure what to do from this position, or how to set up the machine to capture the required information, but this is the first time I've apparently found a way to reproduce the crash.

I want to help, and I'm willing to try to fix this OLD bug--but I admit that at this point I'm increasingly doubtful that Ubuntu is worth the effort. I am pretty strongly motivated by my love of freedom, and my desire to be free of Microsoft and Apple--but Ubuntu seems to be going in the wrong direction in terms of offering a viable option that will bring us more freedom. I absolutely cannot recommend recent versions of Ubuntu to anyone. This 9.10 release is totally crippled in the sound on many machines that I've tested, which already tells me that they testing was totally inadequate. This old bug of the WSOD is still around after a couple of years, so it is obvious that the real cause has never been found and fixed. I desperately want Ubuntu to succeed, but it is increasingly obvious that this economic model is NOT working to provide sufficient testing.

@shanen: Can you reproduce this after uninstalling ureadahead.
I have noticed that when ureadahead does a trace (does it every 30 days or after an update) it can cause some xserver issues including segfaults. See Bug 501715 for more details.

There is a quirk that is applicable to this hardware. I don't know if it would help in your case, but it might be worth testing. Use the 'pci=nomsi' kernel option and see if that makes the issue resolve. Let us know either way.

Also, please run 'apport-collect 159594' and re-test this against lucid. It appears this bug report hasn't gotten as much attention as it could because it lacks necessary files.

Should I do this from the J or K version? Or maybe wait for L? I have both J and K installed here, but Koala had so many problems with the sound card that I still regard it as experimental and I normally run from J.

The problem has become quite infrequent under J, however, so I wasn't too worried about it. However, it has not gone away, and I saw it at least once last month--quite possibly while I was running under Koala. Still, with the crashes so infrequent, it seems really hard to test or know if it is has been eliminated or just waiting. I don't suppose you have any way to trigger the crash before and after changing the pci setting?

I really do wish Ubuntu success, but I am just about on the edge of giving up on it... The sound hassles were quite annoying, and most recently I spent a number of hours trying to read a Star Trek DVD that a friend loaned me... (In Windows, I could easily watch the classic Amok Time.)

Hi Shanen, it should be for Lucid.

We're closing this bug since it is has been some time with no response from the original reporter. However, if the issue still exists please feel free to reopen with the requested information. Also, if you could, please test against the latest development version of Ubuntu, since this confirms the bug is one we may be able to pass upstream for help.

I don't object to your closing it, but I feel obliged to let you know you that I think this bug is still present in Koala. However, the release before Koala seems to be good enough, and that's where I've been staying with the machine for most of the time. I'm going to go ahead and install the new Lynx in the experimental partition and run it for a while, but I'm not optimistic about it. I haven't run Koala very much on this sick Sharp, and I've seen the crash a couple of times...

So far I've only upgraded to Lynx on one machine, where it seems to be working pretty much okay, and done fresh installs on two other machines, where it is having moderate problems already. I hope it isn't as bad as Koala, which gave me big troubles with the sound cards on several machines for several months...

This issue still seems to be present in Lucid - I experienced a number of white (or light grey) screen crashes lately.

As it is rather hot here these days, at first I thought this could be related to an overheating graphics adapter (xserver-xorg-video-ati and the 2.6.32 kernel do not support power management), but I saw the system crash just yesterday after having placed an extra fan above the GPU's heatspreader, so I guess this might rather not be a hardware related issue.

Currently, I am suspecting compiz to be involved in this somehow; since having deactivated it, I haven't experienced the crash again. Crashing generally rather infrequently, of course it is too early to definitely blame the issue on compiz though..

I'll just note that I'm still seeing it, but quite rarely now. In contrast, the new Firefox shutdown problem has become quite frequent...

Having watched this a couple of weeks now, I don't think it is related to Compiz. Actually, I do think it might be related to Firefox and possibly some other interfering component.

I switched to Chromium quite soon after my comment above (#51) and didn't experience any more WSOD crashes since then, even after re-activating Compiz. This is somehow notable as the machine on which I was experiencing the crashes is a 'production' system being used for software development more than ten hours a day, and actually I only remember the system crashing when I was using Firefox; in some, if not all (but I am not sure about this) cases right after Firefox gained focus and I started mouse movement. This would also partially match assumptions in comment #21 and might point to some interference between Firefox, focus handling and mouse movement.