Stubborn Intel Graphics Bug Haunts Ubuntu 12.04

jones_supa writes "The current long-term support version of Ubuntu (12.04) has been experiencing a remarkably tough-to-crack and widely affecting bug related to laptops using an Intel graphics solution. When the lid is closed, every now and then the desktop freezes and only the mouse cursor can be moved. Compiz is usually found hung in the process, switching to a VT afterwards works. The Freedesktop guys are also informed. Have Slashdotters been bitten by this bug and possibly could offer some detective work to help the OSS community find and apply the correct fix?"

Comments

What they are actually reporting an Issue.

[jellomizer]

I have been reporting that problem for a while, but they just assume that I am an idiot who just doesn't know how to use a computer.

Re:What they are actually reporting an Issue.

[Funk_dat69]

Yeah, this issue has been around long before 12.04. Glad its getting some attention. My workaround was to switch to a different distro. :)

Re:What they are actually reporting an Issue.

[muszek]

I have been reporting that problem for a while, but they just assume that I am an idiot who just doesn't know how to use a computer.

Have you tried turning it off and on again?

Re:What they are actually reporting an Issue.

[Anonymous Coward]

but they just assume that I am an idiot who just doesn't know how to use a computer.

And this is why Linux will never be more than just an "alternative" desktop OS. Because its user base is always assumed to just KNOW how everything works, and if you pose a question that some neckbeard thinks is "stupid", he'll let you know it, and you'll be the focal point of mockery and derision. Linux needs to stop being that exclusive club for the cool kids and know-it-alls and start being a more user-oriented (rather than developer-oriented) community.

The Linux community needs a drastic culture-change before it's ready to compete on any meaningful level with Windows and OS X.

Oh, and I WILL be modded down for this, and it WILL validate my point even more.

Re:What they are actually reporting an Issue.

[whistlingtony]

ditto... bug report submitted, nothing done on it. grumble!

Re:What they are actually reporting an Issue.

[Skapare]

Problems like this are the result of closed drivers. Hardware manufacturers need to make their hardware interfaces open (keeping internal firmware closed is OK in this regard) and need to make all code that runs in the host CPU anywhere all open source.

Re:What they are actually reporting an Issue.

[Anonymous Coward]

Except this Intel driver IS open source.

Re:What they are actually reporting an Issue.

[serviscope_minor]

but they just assume that I am an idiot who just doesn't know how to use a computer.

  And this is why Linux will never be more than just an "alternative" desktop OS.

Every time I've tried to get help for something commercial, they assume I'm an idiot who doesn't know how to use a computer. It's a fair assumption, since it is statistically correct.

Linux needs to stop being that exclusive club for the cool kids and know-it-alls and start being a more user-oriented (rather than developer-oriented) community.

Why, what would that get us except more of Gnome3, Unity and Wayland? I rather like Linux: what benefit would it bring to have more users who aren't capable of contributing back on board?

Hardware drivers possibly, but it's not exactly hard to get good hardware which runs Linux excellently already.

Re:What they are actually reporting an Issue.

[Robert Zenz]

As far as I know, you don't get hand-holding with a purchase of Windows. Not sure about OSX.

Re:What they are actually reporting an Issue.

[riondluz]

Tho I prefer E17 to compiz I have CCM running on a toshiba that had this problem. The only fix (for me) that i found is to enable screen-locking.

It brings up the login dialog and restores the desktop

Re:What they are actually reporting an Issue.

[Joce640k]

I have been reporting that problem for a while, but they just assume that I am an idiot who just doesn't know how to use a computer.

Aren't they XKCD/806 compliant?

Re:What they are actually reporting an Issue.

[INowRegretThesePosts]

Yeah, this issue has been around long before 12.04. Glad its getting some attention.

My Core i3 laptop didn't have this problem before 12.04. It appeared with 12.04.

Re:What they are actually reporting an Issue.

[Lumpy]

So do the same with linux, you can call red hat and pay them. or any other linux support company... there are tons to choose from. Linuxcare, redhat, Emperor, etc...

Just like Microsoft and OSX. because they dont give you tech support for free. MSFT requires a credit card, Apple does too if you dont pay for applecare...

Oh wait, we cant compare that way, it would be fair and balanced... we cant have that....

Re:What they are actually reporting an Issue.

[Kelerei]

I have been reporting that problem for a while, but they just assume that I am an idiot who just doesn't know how to use a computer.

I'm guessing that, in their eyes, you didn't ask your question in the proper form.

(I don't necessarily agree with all of ESR's points myself, but his essay is kind of like a creed that the OSS Folks That Matter religiously follow -- so, like it or not, you have to follow it too.)

Re:What they are actually reporting an Issue.

[macbeth66]

Actually, you might be right about him. But that is actually the case in my family. Kids went to Mac, which they griped about for other reasons, my wife, who prefers the old gnome, went back to XP, I'm using KDE and my mother ( 77 ) is just learning KDE. She came from an XP machine. I did put the icons ( about a dozen or so of them ) for her favorite programs right on the desktop. So, Ubuntu lost 60% of the users in my family because of Unity.

Just say no to Unity

Re:What they are actually reporting an Issue.

[INowRegretThesePosts]

but they just assume that I am an idiot who just doesn't know how to use a computer.

And this is why Linux will never be more than just an "alternative" desktop OS. Because its user base is always assumed to just KNOW how everything works, and if you pose a question that some neckbeard thinks is "stupid", he'll let you know it, and you'll be the focal point of mockery and derision.

Some Linux users are like that. But it seems to me that askubuntu.com is quite newbie-friendly.

Oh, and I WILL be modded down for this, and it WILL validate my point even more.

Yeah, right.

Re:What they are actually reporting an Issue.

[h4rr4r]

I honestly though they included that in the sale of the OS.

Then yes, the user should call Canonical and hand over the CC.

Either way asking in forums is not proper support and GP should not complain about free support. Beggars not being chosers and all.

Re:What they are actually reporting an Issue.

[Anonymous Coward]

Wow I didn't know GPUs were programmed in a manor ! How big are the grounds ? Do they have servants too ?

Re:What they are actually reporting an Issue.

[macbeth66]

I have never found that to be particularly true. Yeah, there have been the occasional a**holes, but surprisingly far fewer than with MAC. What a smug self-righteous bunch of turds. As for Windows, there are just so many people using it, that fixes and/or work arounds are found rather quickly. With Windows, skill level varies so much, that the level of expected understanding is quite low.

As for modding you down, when you speak out your *ss and hide behind A/C, what do you expect? However, I see that you have been modded up to 5, Insightful. I bet now you wish you had the balls to stand up and publicly state your case.

Re:What they are actually reporting an Issue.

[mcgrew]

And this is why Linux will never be more than just an "alternative" desktop OS.

No, the reason it will never be more than an alternative OS is because nobody but us nerds have ever heard of it. Windows and iOS have million dollar ad budgets, Linux doesn't.

Because its user base is always assumed to just KNOW how everything works, and if you pose a question that some neckbeard thinks is "stupid", he'll let you know it, and you'll be the focal point of mockery and derision.

That hasn't been my experience. I found that if I asked a legitimate question politely, I'd almost always get a good answer, whether asking in a slashdot comment, help board, or even from my own site I'd get helpful emails from readers.

OTOH I have seen idiots come up with "WTF?? Where's the C: drive in this stupid OS?" I say idiot because you'd be incredibly stupid to expect anything but "RTFM n00b" from a question posed like that. Say "I'm confused, I have two drives in this machine, how do I find them? Thanks." and you'll almost always get a good answer.

When you call a proprietary help desk that have to politely answer questions asked impolitely or even insultingly -- it's sales and money to them. If you're used to getting answers to rude questions, don't bother asking Linux questions until you learn to behave like an adult.

Oh, and I WILL be modded down for this, and it WILL validate my point even more.

Wrong, n00b. You're sitting at +4 right now, as an AC and you started at 0. It's perfectly understandable why you get dissed.

Re:What they are actually reporting an Issue.

[Psyborgue]

If you make a Genius appointment at your local apple store they will help you -- even without apple care.

Re:What they are actually reporting an Issue.

[sl4shd0rk]

And this is why Linux will never be more than just an "alternative" desktop OS.

Things have changed since 1998. There are a ton of moderated Linux forums available these days to help the less experienced. The http://www.stackoverflow.com/ and it's associated channels are a great place to search for, and ask, questions. Chances are, whatever you're struggling with has already been answered there. The moderation/ranking system used keeps the useless garbage cut way down. There's a big enough userbase that questions get answered quickly and accurately. There's no reason to subject yourself to IRC anymore.

Re:What they are actually reporting an Issue.

[Kjella]

No, but all your fellow users will gladly blame Microsoft while Microsoft itself will politely ignore you - at least not rudely trash talk you in an "attack is the best defense" way. You may not get any help in either case, but it'll be a much more pleasant experience. There's a huge difference between not getting your way and not getting your way as well as being insulted. Don't get me wrong I'm sure the Microsoft developers don't hold their users in much higher regard than in open source projects, but if you want to stay employed in customer support you don't tell it to their face.

Re:What they are actually reporting an Issue.

[seandiggity]

I have been reporting that problem for a while, but they just assume that I am an idiot who just doesn't know how to use a computer.

My fix is switching to the 3.4 kernel on the Eee PCs at work, and the Intel graphics problems go away. Needless to say, switching to a bleeding-edge kernel can break things, so be cautious.

I'd try this PPA first with a LiveCD: https://launchpad.net/~xorg-edgers/+archive/ppa

...for your /etc/X11/xorg.conf:

Section "Device"
Identifier "intel"
Driver "intel"
Option "AccelMethod" "uxa"
EndSection

Re:What they are actually reporting an Issue.

[VGPowerlord]

Hell, why not let me edit a text file to change adaptor settings if I want? Text files are easy to backup, easy to script against, easy to work with period. You could even use the silly wizards and dialog boxes while still just writing text files.

If you really want to do it from the command-line, Windows has had netsh for a decade now, which allows you to script and backup network settings.

Re:What they are actually reporting an Issue.

[Qzukk]

+5 informative.

Everything here goes into git (etckeeper), since then nothing has mysteriously stopped working on a restart because we know exactly who and when "nobody" touched something and screwed it up and how to undo it.

Re:What they are actually reporting an Issue.

[Microlith]

Why are you not on their forums having a fucking shitfit?

Cause we aren't all vulgar assholes?

Why are you not telling the devs to stop acting like douchebags and quit fucking with the internals so shit isn't constantly fucking breaking like this?

What internals, precisely, were broken? Oh, closed source drivers from an uncooperative company? You should blame the company rather than demanding others comply to the whims of lazy corporations that refuse to support their products.

Re:What they are actually reporting an Issue.

[Microlith]

The Linux community needs a drastic culture-change before it's ready to compete on any meaningful level with Windows and OS X.

No, companies that ship Linux on their systems should step up and actually support them, rather than push the support off on the communities that surround them. I have met my fair share of dickish Windows and Apple fanboys who attack people with questions rather than help, and seen more than a few people with questions regarding Linux get help.

The users at large are under no obligation to help anyone.

The worst are users that assume that they're entitled to an answer right this moment from someone who's helping others on their own time.

I WILL be modded down for this, and it WILL validate my point even more.

You weren't, but you should be just for adding this stupid passive-aggressive wannabe victim-complex "oh-i'm-so-oppressed" tag.

Re:What they are actually reporting an Issue.

[sa666_666]

I know it's almost impossible for Windows' fans to understand, but sometimes dealing with things not working is still better than dealing with problems in Windows. At least with the former there's a chance that (a) someone will fix it, or (b) you can learn to fix it yourself. That just isn't an option in Windows. If there's a bug and it's closed-source, there's nothing you can do about it.

Now, I understand how someone using Windows without any problems (if that is even possible) would think that all Linux users are masochists, but personally I cannot stand running Windows, and I feel my blood pressure rise any time I'm forced to. I'll deal with Linux issues any day.

Re:What they are actually reporting an Issue.

[Decameron81]

Failure to recognize one's weaknesses is a sure way to fail. When a good number of people keeps telling you they reported errors and got treated as if they were stupid, ignoring them is just another confirmation for them to look elsewhere.

Companies have a human resources department for a reason.

Re:What they are actually reporting an Issue.

[radtea]

Why are you not nailing their balls to the wall? if this were any other product on the planet you would be having massive boycotts and giving them hell. You know that free broken shit is STILL broken shit, correct?

And broken shit is worth what we pay for it.

I've been running Slackware on my netbook lately (13.37) after experiencing Very Bad Things from Ubuntu, Debian, Mint and another distro I can't recall. It's been OK, but just today I had a crash in the Xserver that took me down to the command line when I opened the lid. Besides that, printing still doesn't work (I can see the shared printer on my WinXP box but can't print to it, although I've been chipping away at that issue and it's been getting better). And wireless networking is hit and miss: some networks that I can connect fine with XP don't work, some do. The fancier the encryption the less likely things are to work, it seems.

So the story for "Linux on the desktop" in 2012 is pretty sad: still crashes, only Slackware actually worked out the box (and I've been using it since '94 or so, pre-1.0, so I'm happy tweaking it and building stuff I need from source, etc.) And basic functions like wireless networking and network printing are poor, and power management sucks.

But: it doesn't cost me anything but time (which is a non-trivial concern) and I get a few extra years out of an older machine even as Microsoft is twilighting XP (it's too lame to run Win7 on, but performance with Slackware is pretty good, in terms of raw responsiveness.) I put up with the clunky, broken, barely adequate feature set because of those things.

Re:What they are actually reporting an Issue.

[drinkypoo]

Something everyone with Windows and OSX takes for granted, simply closing the lid and the system goes to sleep, is BROKEN, has been broken for several revs now and its quite clear the devs do not care

If caring were all it took then we'd all have everything we wanted already. Your comment is based on this fallacy, and it is therefore fallacious. Hardware makers do the work of making their hardware work on Windows, not Microsoft. If they spent half as much effort supporting Linux then stuff would work fine on Linux, and when manufacturers have then this has been true. This is, of course, vanishingly rare, due to the massive lead that Microsoft built up by illegally abusing their monopoly position, and through other anticompetitive trade practice. In short, this is not a technical problem, except insofar as that it is not technically trivial to get this stuff right on as much hardware as works with Windows.

It is worth adding that Linux supports vast quantities of hardware that Windows won't run on. In truth, Linux's hardware support is far superior to that of Windows. I have an HP ScanJet 4570c. Once it worked flawlessly, but now SANE is taking a long time to get data back from the scanner. However, it actually works. This scanner is supported only under Windows XP, the version of Windows that you don't want people to be running any more. Maybe Vista, which nobody wants to be running any more. The Windows 7 driver doesn't support it, in spite of the fact that it is based on the same engine as many other scanners, and speaks the same protocol as other scanners which are still supported. HP has chosen not to support this device in order to spur additional sales. A former Windows XP user who upgrades to Windows 7 will find their scanner not working, and it does not matter whether that's HP's fault or Microsoft's. By your own logic, that is Microsoft's responsibility.

Right now Linux is making greater strides than it ever has before, in the form of Android. You might call that not a real Linux, but it is important to the users for the reason that they are getting a device for which they can get kernel source. That means that someone or someones can produce alternate software for the machine because if in no other way it is documented with kernel source. If there is sufficient demand, then it can and likely will happen. The lack of this demand in any particular case is not an indictment against the availability of source code being useful, as you believe it to be. Having the code is not POINTLESS (and capitals don't make your argument more valid, either) but it doesn't solve all your problems; however, it's still better than not having the code. It means that you might get relief for your particular problems, whereas not having the code means you almost certainly won't get what you want if you haven't been able to get it just by spending money.

Your arguments are all essentially fallacious and are based on emotion masquerading as reason.

Dell Netbooks do this

[Urban Garlic]

I have this issue on my Dell mini netbook, it's one of the older ones that actually shipped with Ubuntu back in early 2010.

The problem seemed to gradually get worse for a while, and at one point the graphical start-up screen stopped working, and the thing just booted in text mode. The most egregious symptoms went away with the most recent kernel update, but it's hard to tell if the hang-on-wake problem is actually fixed, because it was intermittent anyways.

Wow! Common bug reports get front-page stories!

[CajunArson]

How about this one:
    KDE 4 has issues with displaying changes made to files in Dolphin. Sometimes the changes show up fine, other times they don't and you have to refresh manually.

Oh and... "Ubuntu" because that magic word has to be inserted for Slashdots "editors" (and I use the term loosely) to care.

There! Now give me a front-page story!

Re:Wow! Common bug reports get front-page stories!

[serviscope_minor]

How about this one:

For better or worse, ubuntu is the most popular distribution and Intel is the most popular vendor of graphics chips. I would say that this is a bug affecting a very substantial portion of the user base.

Re:Yep

[TechMouse]

PATIENT: Doctor! Doctor! It hurts when I do this...

DOCTOR: Well stop doing that then.

Ah, the old ones...

And this is why...

[ilsaloving]

And this is why I use a Mac. I don't have to worry about any of that unreliable unix-y stuff and shoddy quality from Intel!

Re:And this is why...

[WilyCoder]

Oh yeah well at least they don't use Intel chips in Mac!

Re:And this is why...

[ilsaloving]

I think next time I post a comment like this, I'll get a small wind farm set up first. All that woosh must be worth some serious energy.

Slow news day?

[wbr1]

I was under the impression that the ubuntu community had their own channels (forums, lists, etc) for this sort of thing. Are we now the help desk for a linux distribution?
Now if ubuntu had been found to have hidden bestiality videos embedded in it somewhere, that would be news.
Come to think of it, maybe that's what the version names are about. I need to find the hidden porn involving a Hoary Hedgehog or a Precise Pangolin!

12.10 better not be the fix

[Concern]

It's an awful problem.

LTS release that can't reliably suspend (which means, it can't suspend) on Lenovo Thinkpads...

Ubuntu fixes this rapidly, in-stream or they cease to be credible.

Thank you Slashdot, for bringing attention to this.

Re:12.10 better not be the fix

[Anonymous Coward]

My experience has been it's something aggravated by dbus, as turning off zeitgeist almost totally eliminates the problem.
Note: I am not using unity, although again, unity aggravates the issue(s).
Using something like collectd causes an increase in the issue appearing, but at a way lower rate than zeitgeist.
e.g. collectd causes the problem to appear about once a week, vs 2-3 times a day with zeitgeist.
fox/plugin-container /usr/lib/flashplugin-installer/libflashplayer.so -greomni /usr/lib/firefox/omni.ja also seems to
frequently aggravate the issue in question.
I sorta suspect it's something related to kernel polling/caching and resources limits, as changing user limits and/or changing
the shell associated with a uid in /etc/password (e.g. /bin/sh to /bin/nologin or /bin/false) alters the frequency at which
the problem appears. Altering cache sizes sometimes delays the issue. Flushing caches almost always fixes the issue (assuming
one can switch windows before the issue "freezes" things).
Giiven the size of the file cache (4GB) , it might just be due to the time required to flush the cache.
It is as if there's an exclusive lock/resource roadblock for which either dbus can use, collectd can use, or omni.ja can use,
but having more than one going at the same time will result in the issues discussed.
Note, the kernel is a standard, stock ubuntu kernel build.

Re:Irony

[Robert Zenz]

I have to defend the "Works For Me"-closing. It is *very* hard to track bugs which are not showing up on your machine (or any test machine). You never know exactly what that user has already done to the machine prior to the bug occurring and it's hard to get additional information. While it for sure sucks if you get that response, the immediate reaction should be "well, how can I help you find this" and not "I still have that problem" as it happens many times. Also many bug reports which are closed with "works for me" never received any attention from the reporter after reported.

A wider scale problem

[Theril]

My view is that this is only an individual symptom of a larger scale problem. It seems that there are a lot of old, verified, almost showstopper bugs that just get ignored. I'm too busy/lazy to hunt the links at this point, but for example GNOME3 has probably from the beginning had a bug that it gets very sluggish after a few days, at least on some GPU's/drivers, the kernel's trashing behavior in out-of-memory situations is horrible, the audio stack is a horrible mess etc.

It's probably a wider problem of QA, that may be very difficult to solve. At least as a programmer I'm first to admit that I don't want to use my spare time fixing bugs. Debian's almost draconian policies seem to do quite well in terms of stability, but the desktop often lags behind (this may be unavoidable) and the desktop doesn't seem to be a very high priority for them.

PS. This post in no way suggests that things are better, or aren't a lot worse, in Windows-world and OS X with the controlled hardware platform is a very different situation. Maybe I should check if the grass is really greener in the BSD-side.

Re:Irony

[h4rr4r]

Then it is probably another similar bug, but not the same one. When folks like you add "I have this bug too, but everything else is different!" Then you are just adding more noise to the signal that is the bug report.

You have a different problem, it might look similar, report a new bug. Supply the requested information.

I have this problem.

[Psyborgue]

On an Acer Aspire 5313. I've been dealing with it simply by doing as the summary suggests: switching to VT and restarting lighttdm. I'd be happy to help any way I can. The problem doesn't seem to happen every time it suspends, but it happens enough to be annoying. Mostly I just use that laptop for browsing the web so it's no big deal, but I can see it being a major issue if somebody was, for example, working on an important document and hadn't saved.

sudo apt-get install xubuntu-desktop

[tepples]

You didn't really need to switch to Fedora over that. What I did when Unity got on my nerves was sudo apt-get install xubuntu-desktop

Re:sudo apt-get install xubuntu-desktop

[Immerman]

Sure, you could - but why switch to a somewhat under-powered and less-well-supported community branch distro when you could switch to a completely different distro whose direction is more in keeping with your own desires? It's called voting with your feet.

Ubuntu really only has two things going for it - a nice user friendly interface, and a large friendly community that's nice for beginners. If you no longer feel you can have (1) without switching to a branch that will cost you a fair slice of (2), then why stick around at all? Especially if you're no longer a beginner.

Re:sudo apt-get install xubuntu-desktop

[Patch86]

sudo apt-get install kubuntu-desktop
sudo apt-get install gnome-shell

Or if you fancy something different:
sudo add-apt-repository "deb http://packages.mate-desktop.org/repo/ubuntu precise main"
sudo apt-get install mate-desktop-environment

Or...
sudo add-apt-repository ppa:gwendal-lebihan-dev/cinnamon-stable
sudo apt-get install cinnamon

Honestly, it's not like you didn't have plenty of options besides "nuke it from orbit". The Kubuntu and Xubuntu communities in particular are pretty large and friendly.

Not that I'm saying you shouldn't vote with your feet if you feel strongly, and there's certainly nothing wrong with Fedora. But seriously, with the work involved in switching "your whole family" to a new OS, I might have considered something less extreme for the sake of a desktop environment.

Re:sudo apt-get install xubuntu-desktop

[Immerman]

It was a while ago, but offhand there were a couple of panel additions that I consider mandatory that I couldn't find - most importantly hierarchical menu-based file browsing. It actually sounded like one of the base XFCE modules probably provided what I wanted, but didn't appear to be included in Xubuntu or any of the available add-on packages, and I couldn't be bothered with finding a compatible source version and getting it to compile correctly just to see if it actually was suitable. Perhaps the problem was simply that I'm still running 10.04 because I'm unwilling to deal with Unity's idiot-friendly interface. I had actually been testing out the xubuntu desktop for a few days to see if it was minimally acceptable before upgrading to 12.x, sadly it didn't quite pass muster.

I think another problem was a limitation to a single-level heirarchy on the program menu - I've got a couple categories that contain several hundred infrequently entries, without subcategories I'm stuck with either a ridiculously long scrolling menu, or cluttering up the top-level menu with several extra categories that I have to manage by hand.

I tried KDE as well - I believe my problem there was that the screen corners are "numb", all panel buttons start one pixel away where they require attention to click, and the nuisance factor is too great to put up with on things I click dozens of times a day. I'll put up with that sort of idiocy on Windows because there's compatibility gains to be had, but not on a Linux desktop manager.

Re:Irony

[Larry_Dillon]

True, but I see it happen all of the time that a program works on a developer's machine (with all of the development libraries and development tools installed and god-only-knows what custom tweaking).

not just for linux

[glebovitz]

I am not convinced that this is just a linux problem. I have a laptop with a Intel HD Graphics adapter (8086:01116) running Window 7 pro, that experiences similar behavior when coming out of suspend. Some times the screen freezes and the mouse moves, other times the mouse freezes but the screen continues to update (for instance alt-tab navigates windows that still respond to the keyboard).

.

Re:Bug fixing is not fun.

[Panaflex]

Moreover, a kgdb session could likely track this bug down. I'm going to guess that it's a simple locking bug, likely in the intel drivers. Compiz or whatever is performing some operation out of synch with what is "normal" activity and trying to perform a double-lock.

Since cursor operations are tied to a hardware interrupt they still continue to operate.

Another possibility is that the kernel is running at a higher interrupt level in the driver after wakeup, and not locking the iommu/register area away from userspace operations - thus the graphics chip state is getting corrupted and goes into a unknown hardware state. Switching video modes is causing an interrupt which awakens the chip again, and restarting X causes the graphics unit to reset properly.

Yes, I used to be a graphics driver developer for X long, long ago.

grub

[rokj]

I have Intel integrated graphics and I had similar problems, but following line into grub.conf at the end of kernel line fixed it: pcie_aspm=force i915.i915_enable_rc6=1 i915.lvds_downclock=1 i915.i915_enable_fbc=1

Re:Irony

[Robert Zenz]

Yes, that would be helpful. Even better is if you have the machine and the user right next to you and shows you how to reproduce the bug, and then leaves the machine with you (and the user leaves the room...or better, the city).

Don't get me wrong, but it can be freaking hard to track stuff which is not happening on your machine...sometimes it's hard enough even if it is reproducible on a machine you have. Race conditions are such a case. I had some in my application I wrote at work (yeah, totally shame on me for not realizing that obvious thing) and some clients did *never* have a problem and others fell over every few hours. Figuring it out was...tricky...reproducing it even more. Another such case are overflows of all kinds which are not immediately wracking the system, but instead corrupting some parts and then slumber for quite some time.

Ubuntu does what Ubuntu wants

[gottabeme]

Dude, just stop. I have reported many critical Ubuntu bugs to Launchpad--I'm talking about stupid bugs that should never have happened, should never have been released, could quickly be fixed or reverted--but no one at Ubuntu is responsible for fixing them, or for taking the lead on getting the right people to fix it, so nothing happens. That, or 6 months after you report it, a bot says, "Thank you for helping Ubuntu by reporting this bug. Please test the latest version of the software to see if the bug still exists." You confirm it, and 6 months later, same thing. Meanwhile, no actual effort has been expended to investigate or fix the bug. It's like a slap in the face to the bug reporter: it's saying that his time is worthless, when he's already spent time dealing with and reporting the bug.

So get off your high horse. There are plenty of people like me who do exactly that: we file bugs, we complain when they are ignored, we complain when they are not fixed, we complain when stupid regressions appear, we complain when boneheaded decisions are made to release buggy garbage that shouldn't have seen the light of packages.ubuntu.com--but Ubuntu does what it wants to do. New and shiny is more important than stable and reliable and consistent--even for a Long Term Support release, big fat ugly smelly bugs go ignored, and pleas to release the blatantly obvious fix fall on deaf ears.

If it wasn't for dpkg and apt, I'd gladly try another distro. One of these days I'll probably go back to Debian, where at least packages have maintainers who are supposed to be responsible for them, and bug reporters don't get told to test unchanged software over and over again.