braid and other screensavers lock up system when compiz activated [-intel, -nvidia]

Bug #101943 reported by Vlad Socaciu
74
This bug affects 2 people
Affects Status Importance Assigned to Milestone
X.Org X server
Fix Released
Unknown
xf86-video-intel
Won't Fix
Medium
compiz (Ubuntu)
Fix Released
High
Unassigned

Bug Description

Selecting Braid screensaver when Desktop effects are activated runs my system unusable. I'm running Fiesty on a dual core amd64, but i guess these kind of details are now included automatically in the report, right?

So what happens is that the screen freezes and i can only move the mouse and hear gaim playing specific sounds. the keyboard is completely unresponsive, not even the caps lock or num lock buttons don't work. I can only press the reset button on my computer to get back. Obviously, going back to screen saver preferences puts me in the same situation. i must first deactivate desktop effects and only then change the screensaver.

I also tried, after a crash, to let it launch by itself the selected screen saver - Braid - but that too freezes the computer after it displays those coloured circles on full screen.
This is the only screen saver with this problem. I checked all the others one by one.

Known graphic cards on which the bug can be reproduced :
01:00.0 VGA compatible controller [0300]: nVidia Corporation G70 [GeForce Go 7600] [10de:0398] (rev a1)
01:00.0 VGA compatible controller [0300]: nVidia Corporation GeForce 8300 GS [10de:0423] (rev a1)
01:00.0 VGA compatible controller [0300]: nVidia Corporation GeForce 8500 GT [10de:0421] (rev a1)
00:12.0 VGA compatible controller [0300]: nVidia Corporation Unknown device [10de:0533] (rev a2)
00:02.0 VGA compatible controller [0300]: Intel Corporation Mobile 945GM/GMS, 943/940GML Express Integrated Graphics Controller [8086:27a2] (rev 03)
04:00.0 VGA compatible controller [0300]: ATI Technologies Inc RV410 [Radeon X700 (PCIE)] [1002:5e4d] (prog-if 00 [VGA])
01:00.0 VGA compatible controller [0300]: nVidia Corporation Quadro FX 570M [10de:040c] (rev a1)

ProblemType: Bug
Architecture: amd64
Date: Mon Apr 2 18:30:51 2007
DistroRelease: Ubuntu 7.04
Uname: Linux vlad-desktop 2.6.20-13-generic #2 SMP Sat Mar 24 23:12:02 UTC 2007 x86_64 GNU/Linux

Related branches

Revision history for this message
António Lima (amrlima) wrote :

I have several issues with screensavers and compiz enabled. Almost al GL screensavers don't work with desktop effects enabled, and they also looks X, and I have to make a ctrl+alt+backspc.

Running on P4 with ait oss drivers

Revision history for this message
Mallard (stuartbrockman) wrote :

Have the same issue running Beryl, although I have been able to "escape" when previewing in Screensaver Preferences, but not when the screensaver is running full-screen.
All other screensavers work fine.
Alt-Ctrl-Backspace doesn't work, have to reset.

As a workaround, how do I disable the specific screensaver from being used (I like to use the "random" screensaver option)?

Running Dual-Core Pentium D (32-bit mode), NVidia binary drivers with TwinView.

Revision history for this message
Phasmus (phasmus) wrote :

The WhirlwindWarp screensaver, viewed in the screensaver selection window, also crashes my system when compiz is enabled. Braid has the same effect.

ctrl-alt-backspc works, but only a few minutes after I hit it. Until then it looks like a hard freeze.

I'm running Feisty on a Pentium-D in 32 bit mode with NVidia-new drivers.

Mallard, this thread has some useful information on replacing gnome-screensaver with xscreensaver, which (among other functional advantages) allows disabling individual screensavers: http://ubuntuforums.org/showthread.php?t=195557

Revision history for this message
Saivann Carignan (oxmosys) wrote :

As this bug can be reproduced with core components of Linux Ubuntu Gutsy, I set the importance of this bug to Medium and I set the package to xscreensaver.

Changed in xscreensaver:
importance: Medium → High
Revision history for this message
Saivann Carignan (oxmosys) wrote :

increase importance to high

Revision history for this message
Cursim (cursim) wrote :

Hi all. This is my first time reporting on a bug. Just wanted to point out that I do not think the package is Xscreensaver, since I don't have that one installed. I have a stock Xubuntu installation, with Compiz enabled and the gtk-window-decorator. All of the screensavers installed by default work EXCEPT Braid. They're all fine, no lockups or crashes. However, if the random screensaver comes up as braid, X locks up and I have to reboot. If I choose it in the screensaver config app, the app locks up, and must be killed.

I have the following packages installed:

gnome-screensaver
screensavers-default-images
xscreensaver-gl (the culprit?)
xdg-utils

I hope this is helpful.

Revision history for this message
Saivann Carignan (oxmosys) wrote :

Thanks for your comment.

What you say is right, the package that contains braid screensaver is xscreensaver-data which is installed by default. However, Launchpad works with source package names, so the source package name for xscreensaver-data is xscreensaver.

Revision history for this message
Cursim (cursim) wrote :

That would be it then. Sorry for the newb comment :). I ended up deleting the braid screensaver, and now all is well.

Revision history for this message
syko21 (syko21) wrote :

Braid can be easily removed with
sudo rm `locate braid`

Revision history for this message
nonZero (udioron) wrote :

Confirmed on gutsy.

Revision history for this message
Tormod Volden (tormodvolden) wrote :

Can all of you who have problems with "braid" (and not general problems) please report what kind of video card you have:
 lspci -nn|grep VGA
and which driver you are using:
 sed -ne '/Section "Device"/,/EndSection/p' /etc/X11/xorg.conf

Please also attach as a separate file Xorg.0.log from the crash

Changed in xscreensaver:
assignee: nobody → tormodvolden
status: Confirmed → Incomplete
Revision history for this message
Saivann Carignan (oxmosys) wrote :

It happens on :

01:00.0 VGA compatible controller [0300]: nVidia Corporation GeForce 8300 GS [10de:0423] (rev a1)

Section "Device"
        Identifier "nVidia Corporation G80 [GeForce 8300 GS]"
        Driver "nvidia"
        Busid "PCI:1:0:0"
        Option "AddARGBVisuals" "True"
        Option "AddARGBGLXVisuals" "True"
        Option "NoLogo" "True"
EndSection

compiz : 1:0.6.2+git20071119-0ubuntu1~gutsy1
xserver-xorg : 1:7.2-5ubuntu13

Compiz doesn't crash so there's no backtrace for compiz.

The X server freeze but doesn't crash. I reproduced the problem with xserver-xorg-dbg installed and attached Xorg.0.log.

When the X server freeze, it's still possible to move the mouse and it's still possible to press CTRL+ALT+PrntScreen and CTRL+ALT+F6 to switch on a tty console.

Revision history for this message
Colin Darie (colinux) wrote :

Hi,

It happens too here :

01:00.0 VGA compatible controller [0300]: nVidia Corporation G70 [GeForce Go 7600] [10de:0398] (rev a1)

Section "Device"
        Identifier "nVidia Corporation G70 [GeForce Go 7600]"
        Driver "nvidia"
        Busid "PCI:1:0:0"
EndSection

Compiz is not activated. As Saîvann said, it's possible to move the mouse. Alt-Ctrl-Backspace has no effect.
X also freezes on preview of the window where we're chosing the screensaver.

There is no problem with the nv driver.

Sorry for my english :)

Revision history for this message
Saivann Carignan (oxmosys) wrote :

Does NOT happen with this configuration :

01:00.0 VGA compatible controller [0300]: ATI Technologies Inc Radeon RV100 QY [Radeon 7000/VE] [1002:5159]

Section "Device"
        Identifier "ATI Technologies Inc Radeon RV100 QY [Radeon 7000/VE]"
        Driver "ati"
        BusID "PCI:1:0:0"
EndSection

Changed in xscreensaver:
assignee: tormodvolden → nobody
status: Incomplete → Confirmed
Revision history for this message
Davide Saurino (dsaurino) wrote : Re: [nvidia] braid screensaver crashes system with compiz activated

Same problem for me with Braid screensaver on gutsy....it freezed the system and I had to reboot the system.
I removed braid as syko21 said.

Revision history for this message
mrAshley (launchpad-mrashley) wrote :

Video Card

01:00.0 VGA compatible controller [0300]: nVidia Corporation GeForce 8500 GT [10de:0421] (rev a1)

xorg.conf Device Section

Section "Device"
    Identifier "nVidia Corporation G80 [GeForce 8500 GT]"
    Driver "nvidia"
    BoardName "nv"
    Screen 0
EndSection

X.org Log added.

***It doesn't seem to have been mentioned before, but this does not cause X to crash, but just really really slow down. You can wake X up from this screensaver, if you wait a good long time, it has been reported to me (by the user of this computer).

Revision history for this message
Lagos (artificial-music) wrote :

Same problem here.
I have gutsy running compiz on an Nvidia 7600GT card, with the latest 169.7 driver.

Mouse works, but the rest of the system appears to be frozen. If i press the power button (I have it preset to shutdown) the system will start a normal shutdown. So that proves the whole system is not fully locked up, just the gui portion.

All of the other screen savers work perfectly. Worked around the problem by deleting the braid screen saver.

Revision history for this message
kornelix (mikecorn-deactivatedaccount) wrote :

I run no screensavers and I have the same problem.

I have a graphics application that produces the same symptoms: desktop frozen, KB frozen, mouse moves but clicks to nothing. Ctrl-Alt-Backspace does nothing. PC must be rebooted to escape.

The graphics application runs normally with Visual Effects set to None. If Visual Effects are set to Normal, the lock-up happens within seconds.

Suggestion: try running the problem screensavers with Visual Effects set to None. Does it now run OK?

My graphics program has the following characteristics, which may be a clue to the underlying bug: hundreds of small screen updates per second (using gdk_draw_drawable()). This is also a typical screensaver characteristic.

I can supply source and build script for the graphics program if wanted.

Revision history for this message
Vincent Lönngren (bice77) wrote :

64 bit Gutsy in Intel Core 2 Duo, same problem. Braid hangs exactly as described above, and only when visual effects are on.

Video card:

00:02.0 VGA compatible controller [0300]: Intel Corporation Mobile 945GM/GMS, 943/940GML Express Integrated Graphics Controller [8086:27a2] (rev 03)

Driver:

Section "Device"
        Identifier "Intel Corporation Mobile 945GM/GMS/940GML Express Integrated Graphics Controller"
        Driver "intel"
        # BusID "PCI:0:2:0"
EndSection

Revision history for this message
Nick B. (futurepilot) wrote :

I have the same problem with a Nvidia GeForce 6600 and 8400 GS. Besides the Braid screensaver there are also others in the xscreensaver-data-extra xscreensaver-gl-extra packages that will do the same. However I think this is a Compiz bug because all screensavers work fine with Compiz turned off.

Revision history for this message
Kavok (cjc-brimd) wrote :

Also affected by this bug:

lspci -nn|grep VGA
00:12.0 VGA compatible controller [0300]: nVidia Corporation Unknown device [10de:0533] (rev a2)
My video card is a Nvidia Geforce 7000m

AMD Athlon X2 dual core processor TK-53

Running Gutsy 7.10

Section "Device"
    Identifier "Generic Video Card"
    Driver "nvidia"
    Option "AddARGBVisuals" "True"
    Option "AddARGBGLXVisuals" "True"
    Option "NoLogo" "True"
    BusID "PCI:0:18:0"
EndSection
Section "Device"
    Identifier "Videocard0"
    Driver "nvidia"
    VendorName "NVIDIA Corporation"
    BoardName "GeForce 7000M / nForce 610M"
EndSection

Using the Nvidia driver

Revision history for this message
Nando (will-crackers) wrote :

I am so disappointed that it was so easy to freeze up this system. I just moved from Windows XP to Ubuntu and I've been meaning to really like it but there are so many problems upon problems. I thought I had just finished up configuring everything and my system was running smoothly and everything and now this happens to me -- and this was posted 8 months ago and the problem hasn't been solved??

What happens with the "Linux is open source and bugs are corrected promptly" kind of deal? I've been telling all my friends how much I like Linux -- and I do -- but things like this just bother me a lot. Now every time I open the screen saver configuration application my system freezes. And the worst is that there is absolutely NOTHING you can do about it, no control alt backspace, no praying, no pressing all the keys randomly, zip, you are done for, your only choice is hitting that power bottom.

So much for Linux being a very robust system, I am happy I didn't bring a friend over to show him or her how amazing this system is when this happened.

By the way, I read many threads online about Ubuntu being notorious for freezing up as opposed to other distributions. I am firm believer in commitment, so I'll stick with this one for a while, but hopefully things get better.

Ubuntu is like Jurassic Park, it is really cool and fun but it has a ton of issues. Oh well, at least I won't get eaten by a Dinosaur here -- unless of course he is inside the screen saver.

Revision history for this message
Saivann Carignan (oxmosys) wrote :

This bug affect nvidia driver, so canonical and the ubuntu community can't fix the problem, that's why screensavers and not activated by default. I think that we should let this bug as it is here and it will help developers.

Revision history for this message
Nando (will-crackers) wrote :

Sorry about that, I was frustrated at the moment. I knew after reading a couple of posts that it only happened to nvidia but I didn't know it was their problem to fix it, I am surprised they haven't done anything about it. I find it quite serious.

I think future releases of Ubuntu should either take out that screen saver (as I just did) or rebuild the GUI or system or whatever it is so that there is a way out other than turning off the computer.

Sometimes is good to think of new users :-) After all they are part of your future, you want to keep them happy!

Let me also add that overall I am very happy with Ubuntu, much more than how I was with Windows, at least here you know where the problems are .-)

Thanks again Saivann and the Ubuntu team for everything.

Revision history for this message
Didier L (l-farquaad) wrote :

I think I have similar problems with Substrat (in the xscreensaver-data-extra package).

For those who cannot go in the sreensaver configuration screen, disable Desktop Effects before launching that tool, and it should just run fine.

Revision history for this message
Timo Aaltonen (tjaalton) wrote :

Confirmed with hardy and nvidia driver 169.07.

Revision history for this message
Timo Aaltonen (tjaalton) wrote :

I've reported this upstream (linux-bugs at nvidia.com, mentioned on the release notes), and they've reproduced it too and opened an internal bug report about it. Stay tuned.

Revision history for this message
Vincent Lönngren (bice77) wrote :

I have an Intel card, and I've posted the lspci output previously. Yet I experience exactly the same problem. It's not all nvidia related.

Revision history for this message
Saivann Carignan (oxmosys) wrote :

Thanks to Vincent Lönngren for finding the issue with intel cards, I confirm that this bug also happens with this intel card :
00:02.0 VGA compatible controller: Intel Corporation Mobile 945GM/GMS, 943/940GML Express Integrated Graphics Controller (rev 03)

However, this bug does not happen with all Intel cards.

Therefore, Even if graphical drivers should probably still be fixed, braid screensaver apparently cause problems with different drivers so I believe that a part of the problem should be fixed here too. I added xscreensaver to affected packages and set Status to High.

Changed in xscreensaver:
importance: Undecided → High
status: New → Confirmed
Revision history for this message
Saivann Carignan (oxmosys) wrote :

Also thanks to Time Aaltonen for sending the issue to nvidia, this will give more change to this bug to get fixed soon and well.

description: updated
Revision history for this message
Tormod Volden (tormodvolden) wrote : Re: braid screensaver crashes system with compiz activated

The screensaver just draws to the screen, so if that makes the machine crash, it is a graphics driver problem. Anyway, to reduce the damage, we can demote the braid hack to the -data-extra package, so that it is not installed by default.

Revision history for this message
Nando (will-crackers) wrote :

That is an excellent idea Tormod, It is one of the less attractive screen savers anyway, no one will miss it.

I use an nVidia geforce go 7600 and my computer turns into a freak show when I use desktop effects for a while. Some problems I could not correct:

- The screen flickers black from time to time, It doesn't do it all the time but some times, I notice that it doesn't happen when desktop effects are off. I also tried changing the refresh rate and do other things but it didn't fix the problem.

- Some times when I maximize Firefox it is black on the way there. That is not a biog problem, but it looks ugly.

- I have an Intel centrino Wireless ipw3945 and every time the screen goes off by itself the wireless stops working, I don't think this has anything to do with the graphic card but I thought I would mention it as well just in case.

- After prolonged use and for no apparent reason when I open videos I see pink and bluish scribbles but no image, this doesn't happen all the time either. (When desktop effects are on after a while)

-- Also, and this couldn't be considered a problem but a nuisance, I have to load the desktop-settings every time after the screensaver goes off because it overrides my color corrections.

I can't give specific steps to reproduce these problems because I am not sure what creates them.

If you have my same card (and a laptop, I have HP pavillion series dv8000) you can turn your effects to the maximum and wait for the show to start and if it doesn't happen, maybe its me -- although this didn't happen with Windows XP.

I checked the Linux driver for my card and the last update is from early 2006, that seems like a long time to me without updating a driver.

Hope since Timo reported this, they will read what I am saying now and reproduce it as well.

Thank you all for your interest in this issue!

Revision history for this message
Nando (will-crackers) wrote :

To clarify what I mean by the "pink and bluish scribbles" here is a screen shot -- yes it just happened again.

The interesting thing is that when I choose to capture the screen shot from the movie player menu the image comes out okay :? I think it happens when I change from desktop without effects to desktop with effects. It was working fine a little while ago.

Here are the comments about people who had the same problem with the flickering black screen.

http://ohioloco.ubuntuforums.org/showthread.php?t=612144

Let me point out I tried some of those solutions and it didn't work for me. (When I have the desktop effects on)

I am trying the desktop effects without the window wobbling to see if that's the problem. I'll post the results later.

I think nVidia is part of the problem but I am also inclined to believe that the Desktop effects need some serious work done to them. It is horrible to get people all excited about this candy and then give them all kinds of issues :-D So far it's like having a relationship with a Playboy model.

Revision history for this message
Timo Aaltonen (tjaalton) wrote :

Nando: those issues have nothing to do with this bug report. Let's keep one issue per bug please.

Revision history for this message
Nando (will-crackers) wrote :

I apologize, I didn't mean to clutter things. However the screen saver relates to the desktop effects issue and that relates to driver and all of those issues just prove that the thing needs to be fixed, so in my opinion it does relate, its not that I was talking about the keyboard or fish and chips.

Sorry once more, I won't post again after this.

Revision history for this message
Vincent Lönngren (bice77) wrote :

If I understand this correctly, the problem is that when Compiz is some applications crash, among them braid and some other screensavers. So far it has only been reported with the proprietary NVIDIA driver ("nvidia") and the Intel driver ("intel"). Does anyone have solid information that this bug does not in fact manifest with other drivers? If so, then concentrating on isolating and fixing some problem in these two drivers seems like the right thing to do. If not, then it seems to me that Compiz is a better suspect than the screensavers, since they've been working just fine forever and ever. I guess it's possible that previously unknown bugs in these hacks are exposed this way, but it seems a bit far-fetched.

Revision history for this message
Saivann Carignan (oxmosys) wrote :

This bug does not happen with all intel cards and it does not happen with ATI cards that I tested so far, it seems to happen with all nvidia cards. That's why drivers should be fixed, this is clear. However, I also think that we should suspect the quality of the screensaver, I don't have sufficient knowledge to comment beyond that point.

Revision history for this message
Tormod Volden (tormodvolden) wrote :

Sorry for the noise, I think this has been pointed out earlier by Vincent and myself: There is no reason to suspect the quality of the screensaver hack. It has worked well with all cards and if your obscure closed-source proprietary graphics driver crashes with the problem-child that is compiz, the screensaver is the last one to blame. The one unconfirmed occurrence on an intel card can just as well be another bug.

Revision history for this message
Saivann Carignan (oxmosys) wrote :

According to Tormod comment, I change xscreensaver package to xserver-xorg-video-intel which has been confirmed to have problems by two persons. Thanks to Tormod for these clear informations.

Revision history for this message
Rob van der Linde (robvdl) wrote :

This may actually be a Compiz problem, because when the Braid screensaver kicks in, the machine doesn't quite crash fully, it just runs out of RAM and becomes terribly slow.

pressing CTRL+ALT+F1 and opening a terminal (or logging in remoely via SSH) shows compiz.real using all of my 1GB of RAM and all my swap, and it continues climbing.

Looks like a memory leak in Compiz to me

Revision history for this message
Dmitriy Geels (dmig) wrote :

Nando, your screenshot shows known nvidia driver bug, fixed in 169.xx version.

Revision history for this message
Dmitriy Geels (dmig) wrote :

I seen that huge memory leak. It persists even after killing compiz. And no processes use more memory than usual.
So leak must be somewhere in video driver.

Can anyone tell, how to determine memory usage by kernel and all it's modules?

Revision history for this message
Takahani (simon-esneault) wrote :

Hi, also happen here, not only with spraid screensaver, but randomly with GL ones:
Ubuntu 7.10 64 bits
Dual Core Xeon 5110
Nvidia Quadro 4500 (dual screen)
07:00.0 VGA compatible controller [0300]: nVidia Corporation G70GL [Quadro FX 4500] [10de:009d] (rev a1)

dSection/p' /etc/X11/xorg.conf
Section "Device"
    Identifier "Videocard0"
    Driver "nvidia"
    VendorName "NVIDIA Corporation"
    BoardName "Quadro FX 4500"
EndSection
Section "Device"
    Identifier "Videocard1"
    Driver "nvidia"
    VendorName "NVIDIA Corporation"
    BoardName "Quadro FX 4500"
    BusID "PCI:7:0:0"
    Screen 1
EndSection

Sometimes, mouse can still move, but both screen freeze and won't come back. Everytime need to ssh from a remote control, and /etc/init.d/gdm restart
Thanks

Revision history for this message
Tom V (tvice) wrote :

I have similar problems with an Nvidia 7300 GT graphics card. Nvidia driver is 100.14.19. Sometimes if I walk away for a few minutes I come back and the screen saver has not activated but the screen is black and the keyboard is unresponsive. My only recourse it to power cycle the PC. This machine started out 18 months ago with Dapper and has been upgraded to successive Ubuntu versions since, to Edgy, then Feisty and now Gutsy. I did not experience this behavior in any previous version.

Revision history for this message
Noel J. Bergman (noeljb) wrote :

Hardy Alpha 4 with all updates applied as of posting date.

 $ lspci -nn | grep VGA
 01:00.0 VGA compatible controller [0300]: nVidia Corporation Quadro FX 570M [10de:040c] (rev a1)

 $ sed -ne '/Section "Device"/,/EndSection/p' /etc/X11/xorg.conf
 Section "Device"
     Identifier "Videocard1"
     Driver "nvidia"
     VendorName "NVIDIA Corporation"
     BoardName "Quadro FX 570M"
     BusID "PCI:1:0:0"
     Screen 1
 EndSection
 Section "Device"
     Identifier "Videocard0"
     Driver "nvidia"
     VendorName "NVIDIA Corporation"
     BoardName "Quadro FX 570M"
 EndSection

I see the memory leak reflected in compiz.real that others have mentioned.

Revision history for this message
Noel J. Bergman (noeljb) wrote :

By the way, I've seen entire multi-page threads on the forums about how to selectively disable screensavers with gnome-screensaver. I don't know why the obvious hasn't been pointed out:

  chmod -x /usr/lib/xscreensaver/${screensaver-to-disable}

Hope that helps some of you while we wait for this thing to be fixed.

Revision history for this message
Dave Vree (hdave) wrote :

I can confirm this bug on Gutsy with an nVidia Geforce Go 7400. I upgraded to 169 of the nVidia driver...no difference. Upgraded to Hardy kernel (2.6.24), no difference.

Revision history for this message
khensucat (khensucat) wrote :

I had this same problem on Gutsy.

Now, on a fresh install of Hardy Alpha 5, both updates in update manager applied, nvidia-glx-new drivers (169) from the restricted drivers manager. No other software installed.

Intel E2140
2GB Kingston Ram
Asrok 775DUAL-VSTA
GeForceFX 5500 AGP

With "desktop effects" on, Braid will completely freeze my system *every* time it is clicked on, wether in the screensaver preview or full screen. With "desktop effects" off, Braid runs fine. I can also run it error free in at least 6 other distros on this same computer, without trouble, using the same nvidia drivers ;-)

Obviously, the problem isn't in an *older* version of nvidia drivers, and obviously the new nvidia drivers *don't* fix the problem. Compiz on - crashes. Compiz off - stable. I think it's pretty safe to go with compiz, and not nvidia drivers, as the culprit, and work for a fix from there. To recap - it happens on at least *one other video platform*, it happens on multiple Ubuntu versions, and is both reproducable and (workaround)fixable solely by turning Compiz on and off. Anyone seeing a pattern here?

Revision history for this message
Saivann Carignan (oxmosys) wrote :

khensucat : This bug does not happen with compiz and intel/ATI cards, except one intel card so compiz can't be the complete cause of that problem. However since compiz is directly associated with that bug, I will add the package at least to make it more visible for compiz developers, they will be able to consider if it must be set invalid or confirmed for compiz. Thanks for your comment!

Changed in compiz:
importance: Undecided → High
Revision history for this message
Travis Watkins (amaranth) wrote :

Compiz cannot cause a screensaver to freeze your system. Only the driver can do that. I've left the compiz task on here for tracking but lowered the priority so it doesn't mess with our workflow.

Changed in compiz:
importance: High → Low
Revision history for this message
Noel J. Bergman (noeljb) wrote :

> Compiz cannot cause a screensaver to freeze your system

Several of us have reported that we see the compiz.real process expand to fill available memory, which would lead to at the least the perception of a freeze. We can test this just using the preview window.

The relationship appears to be enabling compiz, so (not having looked at the code), do the screensavers behave differently, or is the graphics driver called different?

Revision history for this message
Tormod Volden (tormodvolden) wrote :

Is anyone able to do some debugging of Compiz when it explodes in RAM usage? Running gdb on it from a remote computer would maybe be the best way.

Revision history for this message
Dmitriy Geels (dmig) wrote :

Tormod, I can do that, but I need some instructions...

Revision history for this message
Tormod Volden (tormodvolden) wrote :
Revision history for this message
Saivann Carignan (oxmosys) wrote :

Tormod Volden : Doing a backtrace on compiz over SSH does not give result, compiz won't crash. However, after 2 minutes, compiz use 65% of all the computer memory and 31% of the CPU ressources. After 5 minutes, the computer hard drive is working hard and commands over ssh are VERY slow to achieve, ps aux take 10 seconds to show all active process, each process appear one by one, slowly. It looks like compiz take all available memory and expand until all the memory and swap get completely full. This test has been realized on a 2 Gb memory computer, 2,40 Ghz Core 2 duo CPU.

Revision history for this message
Saivann Carignan (oxmosys) wrote :
Revision history for this message
Travis Watkins (amaranth) wrote :

You should run `gdb -p <processid>` to attach to the running compiz.real when it gets in this state and then type 'where' to see what it is doing. You need to have the dbgsym packages installed for compiz and Xorg.

Revision history for this message
Dmitriy Geels (dmig) wrote :

There is no -dbg packages for compiz in repository. Where to get them?

Revision history for this message
Saivann Carignan (oxmosys) wrote :

I did it 5 times during the time compiz was over-loading the computer memory. the 3 first look very similar, but the 4 and 5 are different. compiz and Xorg dbg/dbgsym packages were installed.

Revision history for this message
Saivann Carignan (oxmosys) wrote :
Revision history for this message
Saivann Carignan (oxmosys) wrote :
Revision history for this message
Saivann Carignan (oxmosys) wrote :
Revision history for this message
Saivann Carignan (oxmosys) wrote :
Revision history for this message
Tormod Volden (tormodvolden) wrote :

That's great Saïvann, some compiz developer should be able to get something out of this. You can maybe try to narrow down what's looping by standing in the right "frame" and going "next" repetitively. If you're a frame too high, "next" will not return when it hits the loop. In the right frame you will see a pattern where it's returning to the same spot after a number of "next" commands.

Revision history for this message
Travis Watkins (amaranth) wrote :

That looks about like what I expected, it is damaging the screen every time the screensaver updates (as it should). Do you have Unredirect Fullscreen Windows enabled? If so, does enabling Sync to VBlank help? These are in ccsm in the General Options inder the General and Display Settings tabs, respectively.

Revision history for this message
Travis Watkins (amaranth) wrote :

Oh, and the rapidly rising memory usage is almost certainly the same old problem we've had with nvidia when you run out of Video RAM. There is no way that code path in compiz leaks, we would have noticed since it gets called every single time your screen updates on every single machine using compiz.

Revision history for this message
Tormod Volden (tormodvolden) wrote :

> There is no way that code path in compiz leaks

Is it normal that compiz.real shows the rising memory usage, like Saïvann and Rob have reported?

Revision history for this message
Travis Watkins (amaranth) wrote :

If the nvidia driver is dumping textures into compiz.real's memory space (which it does) and not freeing them (and it doesn't) then yes, unless you look closely it will look like compiz.real is taking more and more memory.

Revision history for this message
Noel J. Bergman (noeljb) wrote :

> Do you have Unredirect Fullscreen Windows enabled? If so, does enabling Sync to VBlank help?

I did have the former enabled, not the latter. I've now tried with both off and both on. Not noticed a beneficial effect vis-a-vis this defect.

> These are in ccsm ...

Please note: Ubuntu does not include ccsm. Users must install either compizconfig-settings-manager or simple-ccsm. I installed the former, which is the one that I expect you meant.

NOW ... I do have some additional information. I walked through EVERY screensaver that comes with Hardy, using the preview window. Braid UNIQUELY and IMMEDIATELY caused both cores to be consumed. Most interestingly, even other high CPU usage screensavers, such as Helios, demonstrated the following characteristic:

  Cpu0 : 8.6%us, 1.0%sy, 76.4%ni, 13.0%id, 0.0%wa, 0.3%hi, 0.7%si, 0.0%st
  Cpu1 : 12.7%us, 2.7%sy, 21.3%ni, 62.7%id, 0.0%wa, 0.0%hi, 0.7%si, 0.0%st

NOTICE that the vast majority of CPU consumption is niced. BRAID uniquely had all of its CPU in the user column, consuming both cores:

  Cpu0 : 99.7%us, 0.0%sy, 0.0%ni, 0.0%id, 0.0%wa, 0.0%hi, 0.3%si, 0.0%st
  Cpu1 : 84.2%us, 2.0%sy, 10.6%ni, 2.9%id, 0.0%wa, 0.0%hi, 0.3%si, 0.0%st

I could consistently recover the system by clicking on another screensaver, and waiting the several minutes it would take for the system to recognize the click. From what I can see, this CPU consumption is the immediate concern, and we can look into any memory issues after it is resolved.

Revision history for this message
Vincent Lönngren (bice77) wrote :

I may have misunderstood something, but it seems weird to me that this rather simple screensaver should consume all the video memory of these cards.

description: updated
Revision history for this message
Saivann Carignan (oxmosys) wrote :

Travis Watkins : It makes a lot of sense. However, it has been reproduced on several nvidia cards, but also one Intel integrated card and I just got to reproduce this also on a ATI card with the opensource ati driver (I added the card to the list in the description) on a computer I had access today for a short time. With the ATI card, the system also become unusable after some minutes.

The issue can be the drivers, but it seems a bit strange that three different drivers on three different brand of cards have the same problem to me. Don't hesitate to ask if I can provide more information (except about the ATI card, I won't have access to that card in the future). Thanks for your work on this.

Noel J. Bergman (noeljb)
description: updated
Revision history for this message
Cursim (cursim) wrote : Re: [Bug 101943] Re: braid screensaver crashes system with compizactivated

I recently witnessed this bug on Fedora 8, running compiz-fusion and an Nvidia geforce LE 7300. So, its not just an ubuntu issue.
Sent via BlackBerry from T-Mobile

-----Original Message-----
From: "Noel J. Bergman" <email address hidden>

Date: Wed, 27 Feb 2008 14:01:17
To:<email address hidden>
Subject: [Bug 101943] Re: braid screensaver crashes system with compiz
 activated

** Description changed:

  Selecting Braid screensaver when Desktop effects are activated runs my
  system unusable. I'm running Fiesty on a dual core amd64, but i guess
  these kind of details are now included automatically in the report,
  right?

  So what happens is that the screen freezes and i can only move the mouse
  and hear gaim playing specific sounds. the keyboard is completely
  unresponsive, not even the caps lock or num lock buttons don't work. I
  can only press the reset button on my computer to get back. Obviously,
  going back to screen saver preferences puts me in the same situation. i
  must first deactivate desktop effects and only then change the
  screensaver.

  I also tried, after a crash, to let it launch by itself the selected screen saver - Braid - but that too freezes the computer after it displays those coloured circles on full screen.
  This is the only screen saver with this problem. I checked all the others one by one.

  Known graphic cards on which the bug can be reproduced :
  01:00.0 VGA compatible controller [0300]: nVidia Corporation G70 [GeForce Go 7600] [10de:0398] (rev a1)
  01:00.0 VGA compatible controller [0300]: nVidia Corporation GeForce 8300 GS [10de:0423] (rev a1)
  01:00.0 VGA compatible controller [0300]: nVidia Corporation GeForce 8500 GT [10de:0421] (rev a1)
  00:12.0 VGA compatible controller [0300]: nVidia Corporation Unknown device [10de:0533] (rev a2)
  00:02.0 VGA compatible controller [0300]: Intel Corporation Mobile 945GM/GMS, 943/940GML Express Integrated Graphics Controller [8086:27a2] (rev 03)
  04:00.0 VGA compatible controller [0300]: ATI Technologies Inc RV410 [Radeon X700 (PCIE)] [1002:5e4d] (prog-if 00 [VGA])
+ 01:00.0 VGA compatible controller [0300]: nVidia Corporation Quadro FX 570M [10de:040c] (rev a1)

  ProblemType: Bug
  Architecture: amd64
  Date: Mon Apr 2 18:30:51 2007
  DistroRelease: Ubuntu 7.04
  Uname: Linux vlad-desktop 2.6.20-13-generic #2 SMP Sat Mar 24 23:12:02 UTC 2007 x86_64 GNU/Linux

--
braid screensaver crashes system with compiz activated
https://bugs.launchpad.net/bugs/101943
You received this bug notification because you are a direct subscriber
of the bug.

Revision history for this message
Travis Watkins (amaranth) wrote : Re: braid screensaver crashes system with compiz activated

This seems to be a generic texture_from_pixmap problem. If I run metacity w/compositor it works fine, same with metacity w/xcompmgr. When I try kwin with 2D (Render) compositing it works too.

Now the fun parts. Using kwin with OpenGL texture_from_pixmap compositing dies, same as compiz. Using kwin with OpenGL shared memory compositing ran like it was right on the edge of just locking up.

So, this is still probably a driver problem but not one that is easily fixed. I'm pretty sure the constant updates are the problem. This screensaver seems to be a bit like glxgears, it just runs as fast as possible all the while melting your system.

Revision history for this message
Noel J. Bergman (noeljb) wrote :

> This screensaver seems to be a bit like glxgears, it just runs as fast as possible all the while melting your system.

:-)

For what it is worth, this is typical from glxgears on my system:

  Cpu0 : 44.8%us, 32.8%sy, 0.0%ni, 12.0%id, 0.0%wa, 0.0%hi, 10.4%si, 0.0%st
  Cpu1 : 41.2%us, 25.2%sy, 0.0%ni, 20.3%id, 0.0%wa, 0.0%hi, 13.3%si, 0.0%st

I don't see a memory explosion, nor do I see the same CPU consumption and denial of use that I get from braid.

Revision history for this message
Tormod Volden (tormodvolden) wrote :

From the workaround dept.: braid will be demoted to -data-extra (and thus not be installed by default) in xscreensaver 5.04-3.

Revision history for this message
Noel J. Bergman (noeljb) wrote :

> This seems to be a generic texture_from_pixmap problem

Please see http://www.nvidia.com/object/linux_display_amd64_169.12.html:

   "- Further improved GLX_EXT_texture_from_pixmap out-of-memory handling."

Let's get that pulled into Hardy. I have posted Bug 197182 for this issue.

Revision history for this message
Noel J. Bergman (noeljb) wrote :

Well, with 169.12, I still saw compiz.real go from ~200M to 1.4GB in a few minutes, during which braid caused compiz.real and XOrg to consume 99%+ CPU, but (still) unlike others such as Helios, in %us not %ni. On the other hand, there were more braid images displayed, and the system seemed more stable.

I still think that there's an unsolved issue here. What results are others seeing, after upgrading Hardy?

Changed in compiz:
status: New → Invalid
Revision history for this message
Saivann Carignan (oxmosys) wrote :

I did not look the memory/CPU use through ssh but the latest 169.12 in Hardy did not change anything to the symptoms, I still get a unusable system after running braid preview during 4 seconds.

Revision history for this message
Vincent Lönngren (bice77) wrote :

This is from /var/log/Xorg.0.log:

(II) intel(0): Memory allocation layout:
(II) intel(0): 0x00000000-0x0001ffff: ring buffer (128 kB)
(II) intel(0): 0x00020000-0x00029fff: HW cursors (40 kB, 0x 7f820000 physical)
(II) intel(0): 0x0002a000-0x00031fff: logical 3D context (32 kB)
(II) intel(0): 0x00032000-0x00032fff: overlay registers (4 kB, 0x 7f832000 physical)
(II) intel(0): 0x00040000-0x033f7fff: front buffer (52960 kB)
(II) intel(0): 0x007bf000: end of stolen memory
(II) intel(0): 0x033f8000-0x03407fff: xaa scratch (64 kB)
(II) intel(0): 0x04000000-0x04ffffff: back buffer (11264 kB)
(II) intel(0): 0x05000000-0x05ffffff: depth buffer (11264 kB)
(II) intel(0): 0x06000000-0x07ffffff: DRI memory manager (32768 kB)
(II) intel(0): 0x08000000-0x09ffffff: textures (32768 kB)
(II) intel(0): 0x10000000: end of aperture

Is there anything that can be done to avoid a simple hack like braid to cause video memory shortage? BIOS sets the aperture size to the maximum possible, 224 MB. What else can be done, as a workaround until the error handling is fixed?

Revision history for this message
Noel J. Bergman (noeljb) wrote :

I've been spending some time with the code today, and the problem seems related and proportional to the "cycles" option.

Running braid from the command line:

 $ ./braid [the default is 100 cycles]
 braid: warning: only 4% idle over the last 10 secs (at 41.5 FPS)
 braid: warning: only 6% idle over the last 10 secs (at 61.3 FPS)
 braid: warning: blocked event processing for 22.1 secs!
 braid: warning: only 2% idle over the last 32 secs (at 18.5 FPS)

 $ ./braid -cycles 1000
 braid: warning: only 4% idle over the last 10 secs (at 35.4 FPS)
 braid: warning: only 3% idle over the last 10 secs (at 27.9 FPS)
 braid: warning: only 2% idle over the last 10 secs (at 23.0 FPS)
 braid: warning: only 3% idle over the last 10 secs (at 30.0 FPS)
 braid: warning: blocked event processing for 97.9 secs!
 braid: warning: only 0% idle over the last 108 secs (at 1.0 FPS)

But ...

 $ ./braid -cycles 10 -- about 50% CPU
 $ ./braid -cycles 1 -- about 25% CPU

The only thing that cycles appears to impact is how often draw_braid calls braid-erase() and init_braid, and I haven't had time to figure out why doing that more often results in lower CPU usage. Nor have I figured out why

 $ nice -n 15 ./helios
 $ nice -n 15 ./braid -delay 10000 -cycles 2

exhibit totally different behavior, with the former (and every other screensaver) showing CPU consumption in nice, and the latter not.

Revision history for this message
denzilla74 (denzilla74-deactivatedaccount) wrote :

I'm still getting hard-locks from this in Hardy as of today. Will there be a fix?

Ti4600
512MB RAM
XP 2000+ CPU

Running nvidia-glx and Compiz. I've had to Kill the power to my PC 3 times since it completely hard locks the system. You have to turn off Compiz in order to be able to go into the screensaver selection and select another screensaver.

Revision history for this message
Erik Hill (thepurr) wrote :

Just to add, I also have this problem, so it's not an nVidia problem.

i945GM, and the description exactly matches what goes on on my Acer laptop (5630) when I run the braid screensaver.

Bug continues as of today (April 1st) on an updated Hardy Beta.

Revision history for this message
Saivann Carignan (oxmosys) wrote :

Tormod : Since Hardy RC is in few weeks, can you remove braid as you suggested before?

Revision history for this message
Tormod Volden (tormodvolden) wrote :

Saïvann, I submitted the merge request and debdiff for the new version a long time ago in bug #201626. That's all I can do. Some core developer has to upload it.

Revision history for this message
pingnak (pingnak) wrote :

I just discovered this for myself, and deleted the braid screensaver myself, then came here to log the bug.

For me, on my Vostro 1700 with 'nVidia Corporation GeForce 8600M GT', and Compiz, the keyboard is completely locked out. Press caps-lock, no light. The mouse cursor moves, and I suspect there may still be things going on in there, but there's no way to prove that because there's just this pretty braid on the screen that won't go away. I had to press and hold the power button to shut it off cold several times before I made the connection and realized it was the screen saver. I went into the 'Screensaver' settings and tracked down which one it was, and lo and behold, the system hard-locked shortly after its preview popped up.

I recommend this be upgraded to 'critical', since it makes your computer DIE HORRIBLY.

A simple enough patch to REMOVE braid until the core problem is discovered seems reasonable to me. You can always put it back later if the bug gets sorted out. Until then, it makes Ubuntu UNSTABLE. If you're not going to carry drivers for things just because they're closed-source, why bundle things that RELIABLY KILL THE COMPUTER?

sudo rm /usr/lib/xscreensaver/braid

Perhaps a feature request:

Screensaver should certainly have a means to disable naughty screen savers, and detect when they're 'bad'. One good detection method might be write a little record when it launches a screen saver, and delete that record when the screen saver shuts down cleanly. In that way, the screen saver could 'know' which plugin isn't exiting cleanly, and black-list it automatically. Then instead of a permanent mystery crash, you'd get a one-time crash with a pop-up note explaining how it was fixed if you go into the Screensaver settings to see.

Revision history for this message
Bryce Harrington (bryce) wrote :

pingnak, nearly all X video driver bugs result in systems that become unusable; so obviously it wouldn't make sense to list them all as critical! :-)

For video driver bugs, 'High' is sufficient. We reserve this priority for bugs that cause serious system crashes, data loss, affect a lot of users, haven't got an available workaround, etc.

I think your suggestion about adding to xscreensaver to auto-disable naughty screen savers is quite excellent; I would strongly encourage you to file a Wishlist feature request against xscreensaver for implementing this capability for Intrepid or Intrepid+1, and speak with Ted Gould about it.

Revision history for this message
Mauro Torrez (maurotorrez) wrote :

I'm using Debian etch/sid, and all this thing with Braid and Compiz happens to me too.
I'm not sure i should report this here or not... anyways, this happens to me not only with Braid but also with "Interaggregate", "Vermiculate" and "Whirlwindwrap" screensavers. I'm sure there were even more, but i can't remember them right now. It used to be just a "20 second" freeze, but last times i tried the computer completely hanged and i had to shutdown.

Output of lspci -nn|grep VGA:
00:02.0 VGA compatible controller [0300]: Intel Corporation 82G965 Integrated Graphics Controller [8086:29a2] (rev 02)

lspci -nn|grep VGA:
00:02.0 VGA compatible controller [0300]: Intel Corporation 82G965 Integrated Graphics Controller [8086:29a2] (rev 02)

Revision history for this message
Vincent Lönngren (bice77) wrote :

I've had Compiz disabled as a workaround, so I didn't notice. Now I tried it with the screensavers Mauro Torrez mentions, a sure enough - it leads to the same bitter end. I can add 'zoom' to the list, which is of special interest since it always seems to end up as the screensaver in the preview window when random mode is selected.

Revision history for this message
Mauro Torrez (maurotorrez) wrote : Re: [Bug 101943] Re: braid screensaver crashes system with compiz activated

Curiously, "Zoom" doesn't hang my machine. This is so vague. I can't
find an error message anywhere... When i run braid or interaggregate on
a terminal window, then press Ctrl-C after a while this is what happens:

~$ /usr/lib/xscreensaver/interaggregate
interaggregate: warning: only 4% idle over the last 10 secs (at 2.3 FPS)
interaggregate: warning: only 2% idle over the last 11 secs (at 1.3 FPS)

Hope it helps.

Revision history for this message
Bryce Harrington (bryce) wrote : Re: braid screensaver crashes system with compiz activated

I wonder if this has anything to do with bug 181121? Symptoms are different and of course it's a different video card, but the fact that we're seeing sooo many xscreensaver issues is notable.

Revision history for this message
Noel J. Bergman (noeljb) wrote :

I don't see this with all GL screensavers, and I am using an Intel Core 2 Duo, not an AMD. I am, however, running 64 bit.

Revision history for this message
Vincent Lönngren (bice77) wrote : Re: [Bug 101943] Re: braid screensaver crashes system with compiz activated

lör 2008-04-05 klockan 01:30 +0000 skrev Noel J. Bergman:
> I don't see this with all GL screensavers, and I am using an Intel Core
> 2 Duo, not an AMD. I am, however, running 64 bit.

I don't see it with any GL screensavers either.
--
Vincent Lönngren <email address hidden>

Revision history for this message
Tormod Volden (tormodvolden) wrote : Re: braid screensaver crashes system with compiz activated

This has nothing to do with the GL crashers in 181121 AFAIK. braid doesn't even use GL. BTW, I submitted a xscreensaver package to the sponsoring queue a month ago that works around this issue.

Revision history for this message
Saivann Carignan (oxmosys) wrote :

Now braid is not anymore included in Hardy, until fixes are released for the problematics drivers.

Revision history for this message
Vincent Lönngren (bice77) wrote :

This problem is a lot less severe for me since I upgraded to Hardy. The hacks don't lock up the system for nearly as long time at a time, so after 5-10 seconds or something like that it's possible to terminate them.

Braid is actually still present on my system after the upgrade. It doesn't matter that much to me since the same problem is in some other screensavers as well and it's less annoying than before the upgrade, but I thought I'd mention it anyway.

Revision history for this message
miguelquiros (mquiros) wrote :

Just confirm that I've been hited by this bug, "nvidia" driver and it does not happen if visual effects are not enabled. I have tested all the screensavers (from a terminal, just keying /usr/lib/xscreensavers/whatever) and the following ones display the bug:
xrayswarm
interaggregate
braid
whirlwindwarp
substrate (this one works for a few seconds before hanging)
zoom (this does not exactly hang but is very very slow)
imsmap
mismunch
I have used gconf-editor to remove all these from the ramdom screensaver (not a very good solution, since the configuration xml file is reset just opening the gnome-screensaver configuration and all changes are lost). All are included in xscreensaver-data-extra package.

Revision history for this message
Saivann Carignan (oxmosys) wrote :

miguelquiros : You seem to hit a specific bug since a lot of screensavers have the problem in your case. Can you open a new bug report for linux-restricted-modules and attach the resulting file from that command in a terminal : sudo lspci -nnvv > lspci-nnvv.txt ?

Revision history for this message
In , Bryce Harrington (bryce) wrote :

I'm forwarding this Ubuntu bug report:
https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/101943

To reproduce on Ubuntu, install the following packages: xscreensaver, xscreensaver-data-extra. Then run one of the problem screensavers (such as braid) via /usr/lib/xscreensaver/braid. With compiz disabled, it will run quickly with modest processor usage, but with compiz enabled it runs the cpu to 95% and bogs the system down significantly (sometimes you can ctrl-C out, other times you must powercycle). For testing purposes, running 'braid -cycles 10' is enough to demonstrate the issue but avoid the lockups.

Commenting out the XDrawLine() calls in the screensaver prevents the lockups from happening. strace also confirms the screensaver is doing a lot of xserver calls (running the screensaver under strace also prevents the lockups, presumably because it throttles the calls down?)

The effect has also been seen with -nvidia, but not with -ati or -fglrx.

Revision history for this message
In , Bryce Harrington (bryce) wrote :

Fwiw, other screensavers reported to show this issue include interaggregate, vermiculate, whirlwind, imsmap, zoom, xrayswarm, substgrate, and mismuch. Others, such as crystal, are making XDrawLine() calls but don't seem to be triggering the lockups or exhibiting slowdowns.

Revision history for this message
In , Michel-tungstengraphics (michel-tungstengraphics) wrote :

> The effect has also been seen with -nvidia, but not with -ati or -fglrx.

Are you sure? I was able to reproduce the problems described in http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=478836 with the radeon driver.

Your analysis sounds like the server could simply be getting swamped with draw line requests and not dealing with it gracefully.

Revision history for this message
Bryce Harrington (bryce) wrote : Re: braid screensaver crashes system with compiz activated

I can reproduce this easily on a i965 laptop with Compiz turned on so am marking it Triaged.

I tried running 'strace braid' but it didn't lock up at all - presumably strace slows it down enough to not get into the locked up state.

In reading the man page for xscreensaver, I noted this passage which might be of relevance:

       Machine Load
               Although this program ``nices'' the subprocesses that it
               starts, graphics-intensive subprograms can still overload the
               machine by causing the X server process itself (which is not
               ``niced'') to consume many cycles. Care has been taken in all
               the modules shipped with xscreensaver to sleep periodically,
               and not run full tilt, so as not to cause appreciable load.

               However, if you are running the OpenGL-based screen savers on a
               machine that does not have a video card with 3D acceleration,
               they will make your machine slow, despite nice(1).

               Your options are: don't use the OpenGL display modes; or, col-
               lect the spare change hidden under the cushions of your couch,
               and use it to buy a video card manufactured after 1998. (It
               doesn't even need to be fast 3D hardware: the problem will be
               fixed if there is any 3D hardware at all.)

Changed in xserver-xorg-video-intel:
status: Confirmed → Triaged
Revision history for this message
Bryce Harrington (bryce) wrote :

Profile data. unsurprisingly it's spending nearly 100% of its time in the draw function.

Flat profile:

Each sample counts as 0.01 seconds.
  % cumulative self self total
 time seconds seconds calls us/call us/call name
100.00 0.01 0.01 479 20.88 20.88 draw_braid
  0.00 0.01 0.00 62964 0.00 0.00 applywordbackto
  0.00 0.01 0.00 8866 0.00 0.00 xlockmore_event
  0.00 0.01 0.00 479 0.00 20.88 xlockmore_draw
  0.00 0.01 0.00 6 0.00 0.00 xlockmore_reshape
  0.00 0.01 0.00 3 0.00 0.00 init_braid
  0.00 0.01 0.00 1 0.00 0.00 init_window
  0.00 0.01 0.00 1 0.00 0.00 make_shell
  0.00 0.01 0.00 1 0.00 0.00 pick_visual
  0.00 0.01 0.00 1 0.00 0.00 release_braid
  0.00 0.01 0.00 1 0.00 0.00 xlockmore_free
  0.00 0.01 0.00 1 0.00 0.00 xlockmore_init
  0.00 0.01 0.00 1 0.00 0.00 xlockmore_setup

Revision history for this message
Bryce Harrington (bryce) wrote :

I commented out the XDrawLine() xlib calls and the lockups stopped. The code for XDrawLine doesn't look suspicious, and other screensavers using that call (like crystal) seem to be working okay, so it must be something with the way that braid is making the call. I verified that interaggregate also shows the same symptoms, which also uses XDrawLine().

It would be nice to have a simpler test case to demonstrate the problem, but the braid code is probably concise enough already. I'll go ahead and forward it upstream.

Changed in xserver-xorg-video-intel:
status: Triaged → New
Changed in xscreensaver:
status: New → Triaged
Changed in xorg-server:
status: Unknown → Confirmed
Changed in xserver-xorg-video-intel:
status: Unknown → Confirmed
Revision history for this message
In , Bryce Harrington (bryce) wrote :

I also wondered if it was just swamping the xserver, however in testing it with compiz turned off, the behavior is notably different. If it was simply swamping the xserver, I'd expect that to occur with about the same performance in both situations.

I've not tried it myself on -ati so couldn't confirm that, but other commenters on the original report claimed they couldn't reproduce the issue on ATI hardware, but didn't provide much detail.

Revision history for this message
In , Michel-tungstengraphics (michel-tungstengraphics) wrote :

(In reply to comment #3)
> If it was simply swamping the xserver, I'd expect that to occur with about the
> same performance in both situations.

Well, compiz does incur some overhead and generally has an impact on how things are drawn, but yeah. Also, the Debian bug report mentions that usually the bad behaviour persists after killing the application, in which case only killing compiz recovers. This made me wonder if there could be some kind of damage event related feedback loop between the X server and compiz.

Revision history for this message
In , Michael Fu (michael-fu-intel) wrote :

sounds like not a driver bug? component field updated.

Revision history for this message
Bryce Harrington (bryce) wrote :

Upstream believes this to be a server issue. Despite what others have reported, upstream says they could reproduce it with the radeon driver.

Revision history for this message
In , Michel-tungstengraphics (michel-tungstengraphics) wrote :

Created an attachment (id=17815)
Contain number of screen damage rects in compiz

It does look like compiz gets overwhelmed by lots of damage events generated by small lines / points. This compiz patch avoids the problem for me with interagreggate.

Can you update http://bugs.opencompositing.org/show_bug.cgi?id=961 with this information?

Revision history for this message
spiro2903 (spiro2903) wrote :

Hi guys, I'm kinda new at this, so here it goes:
I've just installed Ubuntu on my laptop:
Acer aspire 3020
Sempron 3000
ATI X600
512MB RAM
And I'm loving it, I'm hoping never to go back to XP or vista. Linux seem to save your nerves by at least 80%. But I'm having one problem. It seems to be the same as yours here, except it happens with any screensaver and on my ati card. It just locks up. I can move the mouse cusor but that's about it, I can't click on anything as nothing happens and after a while even the mouse locks up. I'll disabled the screensaver on System->Prefrences->Screensaver and I'll leave it over night to see will it happen now, and I'll repost. I did what Tormod Volden said and got:
01:00.0 VGA compatible controller [0300]: ATI Technologies Inc M24 1P [Radeon Mobility X600] [1002:3150]

Section "Device"
        Identifier "Configured Video Device"
        Driver "fglrx"
EndSection

compiz 0.7.4

Linux 2.6.24-19-generic
I don't know where to find Xorg.0.log
Hope this helps, and I hope the Bug gets neutered...I mean fixed :)

BTW Does anyone know where I can find anything about IR port setup, it doesn't seem to be working on my laptop...

Revision history for this message
spiro2903 (spiro2903) wrote :

Me again :)
Guess what, just as I finished posting it happened again! While I was working, so I guess it's probably not a screensaver issue (at least not for me). But everything did lock up, now that I think about it I think it was downloading updates every time it happened, I'm not sure...hmm can anyone point me in the right direction???

Revision history for this message
Saul (creative-balm) wrote : Re: [Bug 101943] Re: braid and other screensavers lock up system when compiz activated [-intel, -nvidia]

I was pretty much new to this as well- mostly Ubuntu works wonders - but
this particular screen-saver killed my computer, needing many reboots. I
think now they have removed it (it wasn't great by the way). If you
have it on the system - just remove it. I have a link somewhere which I
will forward on - abit too busy tonight - but I hope this helps!

Baul

spiro2903 wrote:
> Hi guys, I'm kinda new at this, so here it goes:
> I've just installed Ubuntu on my laptop:
> Acer aspire 3020
> Sempron 3000
> ATI X600
> 512MB RAM
> And I'm loving it, I'm hoping never to go back to XP or vista. Linux seem to save your nerves by at least 80%. But I'm having one problem. It seems to be the same as yours here, except it happens with any screensaver and on my ati card. It just locks up. I can move the mouse cusor but that's about it, I can't click on anything as nothing happens and after a while even the mouse locks up. I'll disabled the screensaver on System->Prefrences->Screensaver and I'll leave it over night to see will it happen now, and I'll repost. I did what Tormod Volden said and got:
> 01:00.0 VGA compatible controller [0300]: ATI Technologies Inc M24 1P [Radeon Mobility X600] [1002:3150]
>
> Section "Device"
> Identifier "Configured Video Device"
> Driver "fglrx"
> EndSection
>
> compiz 0.7.4
>
> Linux 2.6.24-19-generic
> I don't know where to find Xorg.0.log
> Hope this helps, and I hope the Bug gets neutered...I mean fixed :)
>
> BTW Does anyone know where I can find anything about IR port setup, it
> doesn't seem to be working on my laptop...
>
>

Revision history for this message
Saivann Carignan (oxmosys) wrote :

spiro2903 : Thanks for your comment. Your first bug (with screensaver) may be another bug since it happens with all screensavers and since you can still move your mouse (this was not the case yet with this bug). According to this, I suggest that you also open a new bug report and that you post the link to your new bug so we can take a look at it. The second issue that you get (same lock effect, but without screensaver) is probably related, so I suggest to also speak about it in your new bug report.

Before opening a new bug, please make sure that restricted drivers are properly installed in System / Administration / Hardware drivers. Also, it would be interesting if you can specify in that bug report if you can reproduce the bug when desktop effects are disabled (without compiz). Most probably that this bug does not happen when compiz is disabled, ATI drivers are known to have problems with compiz.

May be related to bug 206845 in case if ati driver is used instead of fglrx.

Revision history for this message
Vincent Lönngren (bice77) wrote :

I have found that the effects of this bug are a lot less severe if driconf is used to turn on vertical blanking synchronization. It's kind of a half-workaround. The screensavers are still really slow, but they terminate pretty soon (after a few seconds) on mouse movement.

Revision history for this message
miguelquiros (mquiros) wrote :

spiro2903: I think you have been hitted by bug 352218 (marked at duplicate of 95569) due to a problem in frequency scaling reported and confirmed long time ago but still unsolved. A workaround is to set the frequency governor in the "conservative" state. This may slow your computer a bit but it is better than rebooting every now and them.
For changing this, go to a terminal and type gconf-editor
Open apps -> gnome-power-manager -> cpufreq
and set policy_ac and policy_battery items to "conservative" (I think that "performance" also works and is faster for a desktop, probably too power-consuming for a laptop).
Good luck.

Revision history for this message
miguelquiros (mquiros) wrote :

Sorry, in previous post I have said "bug 352218" where I should have said "item discussed in forum thread 355218".

Revision history for this message
Martin Olsson (mnemo) wrote :

I installed xscreensaver-data-extra activated desktop effects and tried braid. I can confirm that this issue affects both the ATI proprietary and the Intel open source driver.

On my desktop box I see a lockup directly when I select "braid" in the list (X.org doesn't even redraw my "conky" CPU load graphs and whatnot, all screen updates stop completely. I can still login to the machine using ssh though and top shows that both compiz.real and Xorg is hogging maximum CPU (almost 100% each) while braid is taking like 3-5 percent. When I let it stay frozen for a long time eventually I see single frames of the braid screensaver in the preview box (which is otherwise just pitch black for at least 60 seconds in the beginning). While having had the machine in this "frozen with high CPU load" state for a couple of minutes I still don't see any absurd memory usage though. This machine has the following graphics card: 01:00.0 VGA compatible controller [0300]: ATI Technologies Inc RV350 AP [Radeon 9600] [1002:4150] and if I do dpkg -l I have version 1:6.8.0-1 of the xserver-xorg-video-ati package (I think I have the proprietary driver installed right now).

On my laptop (which has an Intel X3100 graphics card with 965GM chipset) I also get the lockup directly when I click on "braid" to see the preview box. Box does visual updates but is still responsive over ssh. top shows Xorg and compiz.real hogging CPU. However, one big difference is that with the intel driver (which was version 2:2.2.1-1ubuntu13.6 by the way) there is significant problems with the memory. Basically both Xorg and compiz.real grows really big (especially compiz.real has a huge resident chunk of memory). After letting the laptop machine stay in "frozen with CPU hogging" mode I saw these values in top:
Xorg VIRT=439m RES=117m MEM%=5.8
compiz.real VIRT=705m RES=698m MEM%=34%

Revision history for this message
In , Danny Baumann (dannybaumann) wrote :

(In reply to comment #6)
> Created an attachment (id=17815) [details]
> Contain number of screen damage rects in compiz
>
> It does look like compiz gets overwhelmed by lots of damage events generated by
> small lines / points. This compiz patch avoids the problem for me with
> interagreggate.
>
> Can you update http://bugs.opencompositing.org/show_bug.cgi?id=961 with this
> information?

Thanks much Michel for that patch, I have applied it (slightly adapted) in
http://gitweb.freedesktop.org/?p=xorg/app/compiz.git;a=commit;h=aed97c441881d9c382c7865d0305fc8f884c10ac

Revision history for this message
In , Michel-tungstengraphics (michel-tungstengraphics) wrote :

Thanks Danny.

Changed in xorg-server:
status: Confirmed → Fix Released
Revision history for this message
Bryce Harrington (bryce) wrote :

Upstream says this is fixed in the following commit:

http://gitweb.freedesktop.org/?p=xorg/app/compiz.git;a=commit;h=aed97c441881d9c382c7865d0305fc8f884c10ac

We should verify we have this version of compiz in intrepid, and if not, pull in this patch.

Changed in compiz:
importance: Low → High
milestone: none → intrepid-alpha-6
status: Invalid → New
Revision history for this message
Bryce Harrington (bryce) wrote :

Patch applied upstream against compiz indicates this is a compiz bug rather than xorg; reassigning.

Changed in xorg-server:
status: Triaged → Invalid
Changed in xserver-xorg-video-intel:
status: Confirmed → Invalid
Revision history for this message
Tormod Volden (tormodvolden) wrote :

There are compiz packages for Intrepid in my PPA which include the upstream fix. Can you please test it?

Changed in compiz:
status: New → Fix Committed
Revision history for this message
Travis Watkins (amaranth) wrote :

The upload before your last one to intrepid disabled this very patch because it causes redraw issues with emacs. You should have spotted it when you were adding this one.

Revision history for this message
Saivann Carignan (oxmosys) wrote :

I confirm that the bug cannot be reproduced with Tormod Volden compiz build 1:0.7.7+git20080807-0ubuntu6~damrect

Tested opensource ATI driver and closed-source nvidia driver. Everything works fine.

Revision history for this message
Tormod Volden (tormodvolden) wrote :

Travis, that was not the same patch.

Revision history for this message
Tormod Volden (tormodvolden) wrote :

Does anyone see redraw issues with emacs using my test package? Are these troubles documented in a bug report?

Revision history for this message
Saivann Carignan (oxmosys) wrote :

Tormod Volden : I re-installed your compiz package and tried to play with emacs, and I don't see specific problems, everything looks well (however I didn't test everything, only a few things). This was done on a opensource ATI driver, compiz 1:0.7.7+git20080807-0ubuntu6~damrect

Revision history for this message
Tormod Volden (tormodvolden) wrote :

Yes, I installed emacs myself and could not see any issues. Travis or mvo, any plans to integrate this patch/branch?

Revision history for this message
Saivann Carignan (oxmosys) wrote :

The compiz upstream fix is already in current jaunty compiz package. Can someone confirm if this bug has been fixed with braid in a up-to-date jaunty installation?

Changed in compiz:
milestone: intrepid-alpha-6 → none
status: Fix Committed → Fix Released
Revision history for this message
Saivann Carignan (oxmosys) wrote :

I tested the bug on jaunty with compiz 1:0.7.8-0ubuntu9 and a 00:02.0 VGA compatible controller [0300]: Intel Corporation 82915G/GV/910GL Integrated Graphics Controller [8086:2582] (rev 04) graphic card and I wasn't able to reproduce this bug.

Revision history for this message
Noel J. Bergman (noeljb) wrote :

Saïvann, I just checked against Jaunty. No problem, and I was one of the ones able to easily reproduce it. I can check against Intrepid the next time I feel like rebooting, but definitely not present in Jaunty.

Revision history for this message
Saivann Carignan (oxmosys) wrote :

Noel J. Bergman : Thank you very much for your testing. I don't think that fixing/testing this bug in intrepid is absolutely needed since braid screensaver isn't installed anymore.

Tormod : Any thoughts about adding braid back to jaunty?

Changed in xserver-xorg-video-intel:
importance: Unknown → Medium
status: Invalid → Won't Fix
Changed in xserver-xorg-video-intel:
importance: Medium → Unknown
Changed in xserver-xorg-video-intel:
importance: Unknown → Medium
Revision history for this message
inte (inte) wrote :

I want to confirm this bug for Ubuntu 12.04 LTS.
System freezes with:
-Braid
-Interaggregate
-WhirlWindWarp

I'm using Nvidia graphics:
01:00.0 VGA compatible controller [0300]: NVIDIA Corporation GF116 [GeForce GTS 450] [10de:1245] (rev a1)
Nvidia driver version is:
 nvidia-current 295.40-0ubuntu1

Thanks for your support!!

no longer affects: ubuntu
no longer affects: xorg-server (Ubuntu)
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Duplicates of this bug

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.