random X freeze with latest ati driver

[biasquez]

system: ubuntu 8.10
ati driver (opensource): compiled from git (latest version)
device: ati radeon mobility 9700

xorg.conf

Driver "ati"
Option "RenderAccel" "true"
Option "AccelMethod" "EXA"
Option "DynamicClocks" "true"
Option "EnablePageFlip" "true"
Option "DMAForXv" "true"
Option "AGPMode" "4"
Option "ColorTiling" "true"
Option "AccelDFS" "true"
Option "EXAVSync" "true"

when i see a video or i'm using firefox or in other situation, with and without compiz, X freeze, cpu run at 100% and i must force shutdown from power button. how can i fix this problem?

[agd5f]

does removing the EXAVSync option help?

[biasquez]

no, same error....i added recently EXAVSync in my xorg.conf

[Stormking]

I've got the same problem. I am using the radeon driver (from the ubuntu repositories) since fglrx broke compatibility with my Mobility Radeon 9700 three months ago. But I had "MigrationHeuristics" set to "Greedy" all the time. After reading a bit in your blog, I updated to the latest snapshot (from launchpad.net/~tormodvolden), commented out the "MigrationHeuristics" setting and added EXAVSync to my xorg.conf.

Soon,I experienced the first lockups. Mostly when scrolling in Firefox, but also in other situations. When it happens, I cannot switch to the console, ssh into my system from another computer or even use the Magic-SysRq-Keys.

First, I disabled the EXAVSync settings, than I reverted back to radeon from ubuntu. Didn't help. Only re-inserting the "MigrationHeuristics" setting did. As I understand it, that effectivly disables any acceleration.

[biasquez]

EDIT:
with option migrationheuristic (greedy), freeze appears with gnome mplayer, with vlc no problem

[aljaz]

does removing the EXAVSync option help?

With me it does help. My problem is, that after a while, on a random moment, everything freezes except for the mouse. Mouse is allowed to move slowly and jerky, while keyboard and the rest of gnome is completely frozen.
No information in logs, however, so not sure how to file a bug report.
I have an RV515 (Gigabyte X1300PRO).

[Stormking]

With me it does help. My problem is, that after a while, on a random moment, everything freezes except for the mouse. Mouse is allowed to move slowly and jerky, while keyboard and the rest of gnome is completely frozen.
I had this problem some time ago, with the OSS driver *and* fglrx. The solution was to provide better cooling to the GPU.

[aljaz]

I had this problem some time ago, with the OSS driver *and* fglrx. The solution was to provide better cooling to the GPU.

Hmmm. This happened on three different machines (one desktop and two HP laptops) so I don't think it's a cooling problem.

[BlackStar]

I can confirm the exact same issue with a 9600Pro (Ubuntu 8.10 x86, vanilla "ati" driver and xorg settings.)

I will test with fglrx and see if the issue persists.

[DuSTman]

I've been having similar problems, myself. This occurs with both 6.9.0 and the new release candidate driver. This is with Xserver 1.5.3 and a radeon x850.

In particular, the problem seems to be provoked by mozilla applications - with firefox 2, it would fairly reliably crash within a few minutes of use, though I haven't noticed the same problem in the last few days (since I upgraded to firefox 3). Thunderbird 2, however, does still cause this issue.

When it happens, the mouse still moves (albeit jerkily - updated ~3Hz), but no other response to any other input device. The CPU usually gets pegged at 100% (from what I can tell - the CPU usage generally ceases updating about five seconds after going into this state).

[f0x_]

I have this problem too.
Random X freeze but only with Desktop Effects enabled in kde4.
Can't switch to console.

system: gentoo
driver: xf86-video-ati and mesa compiled from git
xorg-server: 1.5.3

xorg.conf:
Identifier "Card0"
Driver "radeon"
VendorName "ATI Technologies Inc"
BoardName "M56P [Radeon Mobility X1600]"
BusID "PCI:1:0:0"
Option "AccelMethod" "EXA"

Everything freezes except for the mouse cursor.

[yotambien]

Mobility radeon 9600/9700, no composite thing, no fancy xorg.conf options, laptop fails to resume from suspend to ram like 20% of the times. This started a couple of weeks ago when I removed fglrx and left only radeon on the system (version from Debian sid).

[Stormking]

Does anyone investigate this issue? How can I help? I'd really like to activate acceleration because scrolling in firefox is *much* smoother this way. But when I take "MigrationHeuristics Greedy" out of my xorg.conf, X freezes after about 15 minutes.

[bridgman]

Stormking, you mentioned earlier that improving cooling to the GPU fixed the problem; did it come back ?

[Stormking]

Stormking, you mentioned earlier that improving cooling to the GPU fixed the problem; did it come back ?
There seems to be a misunderstanding. Cooling the GPU helped me with the X-freezes-but-mouse-pointer-can-still-be-moved-problem, two years ago. Now, there's a X-freezes-completely-when-acceleration-is-activated-problem.

[f0x_]

It's not a cooling problem. In my opinion there's a bug related to (compiz, kde desktop effects) acceleration. I don't know how to file a bug report cause the freeze can't make me switch to console.

[bridgman]

There seems to be a misunderstanding. Cooling the GPU helped me with the X-freezes-but-mouse-pointer-can-still-be-moved-problem, two years ago. Now, there's a X-freezes-completely-when-acceleration-is-activated-problem.

Got it. Thanks !

[Stormking]

I'd like to repeat my question: Is anyone investigating this issue? And how can I help?

[Ren Höek]

I also have freezes with my Ati Mobility Radeon 9700 and the open source radeon driver. With Desktop Effects enabled X freezes immediately, without it system sometimes holds out for some minutes, but most of the time just a few seconds after login.

Solution was to set AGPMode to x1 :
Option "AGPMode" "1"

I don't know if this is really a "real" solution or just a dirty fix. Default value for this card is 4 and as far as I could measure I lose some fps, but everything is stable.

I hope it can help.
Or at least someone can tell, why default value doesn't work...

[xeros]

I had this problem some time ago, with the OSS driver *and* fglrx. The solution was to provide better cooling to the GPU.

The same with my Radeon 9600 and my friends Radeon 9250 - after opening the computer I noticed the graphic card is very hot (no fan on card). With computer cover open I've noticed a lot better stability on the open source drivers. Both systems use Kubuntu 8.10 and the default xorg.conf.

[xeros]

It's not a cooling problem. In my opinion there's a bug related to (compiz, kde desktop effects) acceleration. I don't know how to file a bug report cause the freeze can't make me switch to console.

It might be a cooling problem - it's because desktop effects use more of the GPU power all the time. In my case after running for few hours KDE4 with desktop effects and now I start Nexuiz I can get freeze after few seconds (sometimes I can do SysRQ sometimes not) but when I just start my computer and then start Nexuiz I can play the game without problems for few hours. One time I've noticed image dissortion in Firefox and some minutes later a freeze.

[biasquez]

i have strange problem with my 9700 mobility. when i use compiz, i don't have freeze, when i use metacity, after 2-3 minutes, pc freeze and i must force power off. in my xorg.log i have this:

[2273 sec: 110791 usec](II) RADEON(0): [RESUME] Attempting to re-init Radeon hardware.
[2273 sec: 110850 usec](**) RADEON(0): Using AGP 4x
[2273 sec: 110899 usec](II) RADEON(0): [agp] Mode 0x1f000207 [AGP 0x8086/0x3340; Card 0x1002/0x4e50 0x1734/0x106b]

[agd5f]

For those of you having lockup or corruption problems does this patch help?

http://www.botchco.com/alex/xorg/radeon_engine_idle.diff

[suokko]

@agd5f: Maybe this is related to horrible way how mesa dri drivers are writen. I just noticed the there is huge amount of debug output inside drm lock. That means for me that if I have that debug output going to terminal my system had high chance of freeze. Easy way to reproduce that with r200 is doing R200_DEBUG=all glxgears

I did code a work around so that this blocking operation is detected and any program would crash. MY code to do it was that I added to r200_lock.c/h 2 functions. setAlarm() and freeAlarm() that I canlled just after acquiring the lock and just before releasing. setAlarm() did settup SIGALRM handler when it was first time called and everytime called alarm(1). freeAlarm() did just call alarm(0). Signal handler just called abort().

That way I could get coredump from freeze. Of course I know that this isn't going to be general purpose debugging way because of limitations in alarm but maybe something similar could be done with timers so it won't interfere with application code.

Just a warning: DON'T do local debugging using gdb with this trick. It is going to freeze X also. Remote debugging using ssh works fine.

[dungeon]

+1

I quite occasionally notice this maybe 3 years ago. Tried to debug some fallbacks with R200_DEBUG=fall, do a typo and X just freeze.

[suokko]

If you have problems with heat DynaimicClocks might help you

man radeon says:
Option "DynamicClocks" "boolean"
Enable dynamic clock scaling. The on-chip clocks will scale dynamically based on usage. This can help
reduce heat and increase battery life by reducing power usage. Some users report reduced 3D performance
with this enabled. The default is off.

[Stormking]

For those of you having lockup or corruption problems does this patch help?

http://www.botchco.com/alex/xorg/radeon_engine_idle.diff
Should I Still try this one? I've seen that you reverted this patch in your repository.

By the way, there seems to be a new problem. Sometimes, XV output crashes X. I can play five, six movies after another and the seventh makes X restart after about one or two seconds. Happens both with "classic" overlay and Textured Video.

[DuSTman]

I can confirm that my lockup issues still persist with the 6.11.0 driver. I'll try agd5fs patch and report back.

Note that when my system enters this lockup-mode, I can still SSH into it - upon attaching gdb to it, it consistently gives the following backtrace:

#0 0x00007ff2b9c7c777 in ioctl () from /lib/libc.so.6
#1 0x00007ff2b8953b15 in drmDMA () from /usr/lib/libdrm.so.2
#2 0x00007ff2b80621de in RADEONCPGetBuffer (pScrn=0x7fe9b0) at radeon_accel.c:593
#3 0x00007ff2b8062370 in RADEONCPFlushIndirect (pScrn=0x7fe9b0, discard=1) at radeon_accel.c:647
#4 0x00007ff2b80ab4a8 in RADEONPrepareSolidCP (pPix=<value optimized out>, alu=3, pm=4294967295, fg=3432552600) at radeon_exa_funcs.c:103
#5 0x00007ff2b77ea528 in ?? () from /usr/lib64/xorg/modules//libexa.so
#6 0x00007ff2b77eae4b in ?? () from /usr/lib64/xorg/modules//libexa.so
#7 0x00000000005201f4 in ?? ()
#8 0x0000000000508fc1 in miCompositeRects ()
#9 0x0000000000510581 in ?? ()
#10 0x0000000000449654 in Dispatch ()
#11 0x00000000004314ba in main ()

[agd5f]

I can confirm that my lockup issues still persist with the 6.11.0 driver. I'll try agd5fs patch and report back.

Note that when my system enters this lockup-mode, I can still SSH into it - upon attaching gdb to it, it consistently gives the following backtrace:

#0 0x00007ff2b9c7c777 in ioctl () from /lib/libc.so.6
#1 0x00007ff2b8953b15 in drmDMA () from /usr/lib/libdrm.so.2
#2 0x00007ff2b80621de in RADEONCPGetBuffer (pScrn=0x7fe9b0) at radeon_accel.c:593
#3 0x00007ff2b8062370 in RADEONCPFlushIndirect (pScrn=0x7fe9b0, discard=1) at radeon_accel.c:647
#4 0x00007ff2b80ab4a8 in RADEONPrepareSolidCP (pPix=<value optimized out>, alu=3, pm=4294967295, fg=3432552600) at radeon_exa_funcs.c:103
#5 0x00007ff2b77ea528 in ?? () from /usr/lib64/xorg/modules//libexa.so
#6 0x00007ff2b77eae4b in ?? () from /usr/lib64/xorg/modules//libexa.so
#7 0x00000000005201f4 in ?? ()
#8 0x0000000000508fc1 in miCompositeRects ()
#9 0x0000000000510581 in ?? ()
#10 0x0000000000449654 in Dispatch ()
#11 0x00000000004314ba in main ()

This is a GPU lockup. The engine has hung and is no longer processing dma buffers so the fences never come up and the buffers can't be freed and the driver keeps querying for free dma buffers.

[DuSTman]

I did a git bisect on this issue - the results:

0d5e0347af4322713075193154b8a348de4a0b52 is first bad commit
commit 0d5e0347af4322713075193154b8a348de4a0b52
Author: Alex Deucher <alexdeucher@gmail.com>
Date: Wed Aug 13 14:17:34 2008 -0400

Remove reset of 3D scissor registers when using the CP in the ddx

They should only affect 3D and init3d() should take care of that case
noticed by libv on IRC.

:040000 040000 628f00b2656f9697532fbbc31d17f508c483622a c8794ffeeae7823dd45c27e6aada97ff21826ca M src


I don't know how many other people here are affected by this (I get the feeling different people in this thread are encountering slightly different bugs).
For reference, my hardware is a Radeon x850 - kernel is 2.6.27-gentoo-r7

[agd5f]

this patch should re-enable it:
http://www.botchco.com/alex/xorg/write_sc_in_refresh_cp.diff

Does that fix it for you on git master?

[DuSTman]

this patch should re-enable it:
http://www.botchco.com/alex/xorg/write_sc_in_refresh_cp.diff

Does that fix it for you on git master?

No, Error still crops up with that one..

[DuSTman]

Gah. Double post.

[agd5f]

how about this:
http://www.botchco.com/alex/xorg/write_sc_after_every_prim.diff

[DuSTman]

No, that doesn't work either..

[DuSTman]

As requested (on IRC) I did a further git bisect, this time with the first patch applied, which isolated this commit (http://cgit.freedesktop.org/xorg/driver/xf86-video-ati/commit/?id=fc079c5267baf431bbecee7744e484783d393152) as introducing the issue.

[mitrol]

My X also freezes randomly, Ctrl+Alt+Del, AltGr+Sys+Req not doing nothing. Sometimes the mouse still moves (but slowly) and the rest off the desktop is frozen, sometimes all the desktop gets frozen.All of this happens every time and not long after login.

I'm using kde 4.2 with desktop effects enabled.
The problem occurs with radeon and radeonhd when using EXA , I installed both from git today. When using XAA the desktop doesn't freeze.

ATI Mobility Radeon x1600
X.Org X Server 1.5.2
openSUSE 11.1

[bridgman]

How new is your drm code ? Are you running drm from the openSuSE kernel or did you build and install drm from git as well ?

[slacker]

My X also freezes randomly, Ctrl+Alt+Del, AltGr+Sys+Req not doing nothing. Sometimes the mouse still moves (but slowly) and the rest off the desktop is frozen, sometimes all the desktop gets frozen.All of this happens every time and not long after login.

I'm using kde 4.2 with desktop effects enabled.
The problem occurs with radeon and radeonhd when using EXA , I installed both from git today. When using XAA the desktop doesn't freeze.

ATI Mobility Radeon x1600
X.Org X Server 1.5.2
openSUSE 11.1

Seems the same lockup I've been having since R500 3D support was merged, back in May. Here's my bug report: http://bugs.freedesktop.org/show_bug.cgi?id=16198

[mitrol]

Today I build and installed drm from git and I had the same lockup.

[Blacksmith]

I was getting random lockups on Debian with an X800GT using the xserver-xorg-ati-video_6.9.0 driver, back dating the driver to 6.6.3-2 from debian 4.0 completely cured the problem.
Have not tried the latest version "If its not broken, don't fix it"

[DuSTman]

I've tried reversing the change to the RADEON_SWITCH_TO_2D and RADEON_SWITCH_TO_3D calls in 6.11.0 and added the scissor register reset code back in. This time Xserver bricked about 1 second after starting. Way to go, me!

[agd5f]

You might try an updated mesa. Looks like we weren't setting the number of pipes properly in mesa for r3xx-r4xx cards. I've fixed this in mesa master and the 7.4 branch.

[agd5f]

does this patch help?

http://www.botchco.com/alex/xorg/no_wait_if_engine_idle.diff

It's against the r6xx-r7xx-support branch of the drm, but you should be able to apply it to other drms easily (just two lines to add).

[DuSTman]

Sorry, no. Neither an updated mesa nor the libdrm improvement seemed to help.

[agd5f]

Sorry, no. Neither an updated mesa nor the libdrm improvement seemed to help.

Just to clarify, this isn't a change to libdrm, it's a change to the radeon drm kernel module. Make sure you build and install the kernel modules (cd linux-core; make drm.o radeon.ko; cp drm.ko radeon.ko /path/to/your/kernel/modules) rather than libdrm.

[DuSTman]

Sorry about that - unfortunately the updated kernel modules don't seem to help either.

[agd5f]

Several people have reported improved behavior using the drm-next branch of Dave's drm-2.6 tree.

http://git.kernel.org/?p=linux/kernel/git/airlied/drm-2.6.git;a=summary

[DuSTman]

Again, no joy.

[juhuuu]

Hi there
I recently encountered this problem with Ubuntu jaunty beta release.

System repeatingly freezes with Kwin Desktop Effects on after 10-30 Minutes mostly when opening new windows/sub-windows (ie "browse file").
With Desktop effects off everything works absolutely fine, can run glexgears, vlc 720p movie and Goofge earth at the same time for hours...... .

Besides that I have massive tearing issues with Kwin + Desktop Effects and google earth - any ideas?

Bug is files under Ubuntu #351990 and 350766

greetz

[suokko]

Hi there
I recently encountered this problem with Ubuntu jaunty beta release.

System repeatingly freezes with Kwin Desktop Effects on after 10-30 Minutes mostly when opening new windows/sub-windows (ie "browse file").
With Desktop effects off everything works absolutely fine, can run glexgears, vlc 720p movie and Goofge earth at the same time for hours...... .

Besides that I have massive tearing issues with Kwin + Desktop Effects and google earth - any ideas?

Bug is files under Ubuntu #351990 and 350766

greetz

Sounds like same bug that is causing freeze in many opengl games. I noticed it first with xmoto.

for more details about bug report: https://bugs.launchpad.net/bugs/348450

[RealNC]

I don't have random crashes, but I have a way to reproduce a crash with Firefox and Gecko Mediaplayer (an MPlayer Firefox plugin). If I go to a site that has video (not Flash, but normal ones that require a "media player plugin") and the plugin is configured to render video using the gl mplayer renderer, it plays fine, but when closing Firefox, the system hangs.

Since MPlayer and Firefox are userland, the one responsible for the hang has to be the driver.

[agd5f]

Does this patch (against ati git master) help?

http://www.botchco.com/alex/xorg/flush_after_each_sequence.diff

[grantek]

hmm, I'm getting random hangs on a HD4850, ie. compiz etc. not even supported. Has anyone seen hangs without a compositing desktop?

Currently using the jaunty radeon package, with no options in xorg.conf (exa acceleration according to the logs)

[RealNC]

I don't have random crashes, but I have a way to reproduce a crash with Firefox and Gecko Mediaplayer (an MPlayer Firefox plugin). If I go to a site that has video (not Flash, but normal ones that require a "media player plugin") and the plugin is configured to render video using the gl mplayer renderer, it plays fine, but when closing Firefox, the system hangs.

Since MPlayer and Firefox are userland, the one responsible for the hang has to be the driver.

Gah! Due to stupidity, I thought this thread was about Catalyst and not the open source driver...

Sorry for the noise :p

[highlandsun]

Slightly related, suspend/resume doesn't work for me using Ubuntu 9.04's Xorg stuff and my own compile of 2.6.29 kernel. Usually it just hangs on suspend. Sometimes I can run s2ram from a command line and it will suspend, but on resume it only partially redraws the screen and then it hangs again. (HD3450 in HP dv5z laptop.)


Hm, never mind. I just updated my git pulls and rebuilt, and it all worked.

[raven9]

Kubuntu 9.04 beta 64 bit.
Also experiencing freeze with desktop effects turned on.
I did some research and come to this conclusion:
On Kubuntu Alpha 6 everything worked fine, but with beta came kernel update and that is the time when problem has start to show up.
I tried to use Compiz and worked flawlessly.
Also when using Compositing type: XRender don't have any problems with randomly freezing.
Hope this help to someone who is smarter then me, and solve the problem.

[raven9]

Has anyone find solution? :)

[jbernardo]

Here I was thinking it was happening only to me... Radeon 9600 Pro (AGP), clean xorg.conf, kubuntu jaunty beta x86_64

This is how the card is identified in xorg.0.log:
PCI:*(0@1:0:0) ATI Technologies Inc RV350 AP [Radeon 9600] rev 0, Mem @ 0xc0000000/268435456, 0xe0030000/65536, I/O @ 0x0000c000/256, BIOS @ 0x????????/131072

The most frustrating part is that there is nothing in Xorg.0.log which points to any error.

Subscribed bug 351990, lets see if there are any improvements.

[bitnick]

Since upgrading from xserver 1.3.0/xf86-video-ati-6.12 to 1.5.3/xf86-video-ati-6.12.1 I have the same issue:

When using firefox-2 or thunderbird I get very frequent hangs. The screen freezes and I hear a faint whistling noise from the computer (some CPU/GPU at 100%?). The mouse pointer is still alive but moves jerkily, updated a few times per second. I can sync disks and reboot with SysRq.

(Firefox 3 does not seem to trigger the bug.)

This is on an R420 JI, ChipID 0x4a49 (Radeon X800 Pro), AGP.

[raven9]

The only thing that is helped me is to install new kenel 2.6.29 to be precise. I also get some information that you could fix this bug by using options noapic acpi=off when boot, but i didn't try this. Hope this help
@juuhu
Thank you for information.

Also i thank to Fedora community.

[DuSTman]

When using firefox-2 or thunderbird I get very frequent hangs. The screen freezes and I hear a faint whistling noise from the computer (some CPU/GPU at 100%?). The mouse pointer is still alive but moves jerkily, updated a few times per second. I can sync disks and reboot with SysRq.

(Firefox 3 does not seem to trigger the bug.)

This is on an R420 JI, ChipID 0x4a49 (Radeon X800 Pro), AGP.

This, in particular seems identical to what happened on my rig (Firefox 2 would crash frequently, but not v3. Thunderbird, in particular, reliably crashed in this manner when popping up a password dialog box.

Personally, I've switched versions (am now running the GEM/KMS/DRI2 branches of the DRM, ddx and mesa from git), and find that these don't exhibit the same problem.

[agd5f]

Does the patch I attached to comment #21 in bug 16198 help?
https://bugs.freedesktop.org/show_bug.cgi?id=16198

[suokko]

The level "beginers journey three" in xmoto extra level set is sure way to lock graphic system inside drm calls. (Using Ubuntu jaunty and radeon-rewrite+modesetting-gem) I haven't yet find any others as reliable way to reproduce the hang. Intresting point is that there is 3 applications locking in heavy drm lock. Xorg and xmoto is there but I don't know what is the 3rd application (compiz is disabled).

When drm is locked gdb is often unable to attach to a blocked process.

Card is mobility x1400 with turion processor.
EXA, migration smart, dynamic clocks and page flip are set in xorg.conf.

[highlandsun]

I just did a fresh pull from git yesterday and the resulting driver locked up on me within about 10 seconds of the desktop being drawn. i reverted back to the April 21 version and it's running for now. (Commit 228b9ac87...) i haven't dug any further to see exactly what broke.

[Stormking]

Does anybody of the developers still investigate this issue? I still get lockups, every day. Mostly when browsing the web using firefox.

Sometimes, my system works for 10 hours, sometimes it locks up three times in twenty minutes. The situation is completely unacceptable, esspecially now, when my hardware is not supported by fglrx, anymore.

So please, look into this! Let me know all the information you need. I am able to compile a kernel or kernel module, I am able to apply a patch, but you might have to direct me to the right sources.

[slacker]

I suppose you have R500 or R690 chipset?
If so, check https://bugs.freedesktop.org/show_bug.cgi?id=16198
To minimize the occurrence of the lockups you should be using xf86-video-ati from git. For me it still happen, but much less frequently.

[Stormking]

I suppose you have R500 or R690 chipset?
No, R300 (Mobility Radeon 9700/64MB)


To minimize the occurrence of the lockups you should be using xf86-video-ati from git. For me it still happen, but much less frequently.
I use the latest builds from Tormod Volden's PPA (currently git20090531.7599dc40).

[mitrol]

opensuse 11.1 KDE 4.3svn
xorg 1.6.1
ati mobility radeon x1600

I updated mesa+drm+radeon (from git master) on the 2009-05-21 and since than I never experienced a X server crash again. Used to crash every time I login after some minute/hours.

[agd5f]

You might want to try the drm patch in bug 21849:
http://bugs.freedesktop.org/show_bug.cgi?id=21849
Dave just pushed it to Linus in his last round of drm fixes.

[Stormking]

You might want to try the drm patch in bug 21849:
http://bugs.freedesktop.org/show_bug.cgi?id=21849
Dave just pushed it to Linus in his last round of drm fixes.
Okay, I applied the patch to vanilla 2.6.30-rc8, compiled and installed the whole thing. Now I'll wait and see ...

[Kano]

Did you see the drm intel changes in rc8-git1?

[Stormking]

Five minutes ago I wanted to write about how things were looking promising but than ... it happened again!

Until you come up with another idea or patch, I will continue using the patched 2.6.30-rc8 kernel. By tomorrow I should know if the lockups at least happen less frequently.

[Stormking]

After a few days of monitoring, I have to say that this patch does not help. I compared kernel 2.6.30-rc8's (patched) drm module (1) to the one checked out from git://anongit.freedesktop.org/mesa/drm (2). I still have to test the unpatched kernel-2.6.30-module.

(1) usually crashes after 30-45 minutes.

(2) usually crashes after 60-90 minutes.

One time, (1) crashed about one minute after logging in. To be more precise, the first opened window after login triggered the freeze. (2) gives me at least 20 minutes before freezing.

The longest period of uninterrupted use with (1) was about 2 hours. (2) sometimes (very rarely) lets me work for 4 to 6 hours before a freeze occurs.

Freezes are always triggered by some kind of interaction with the GUI (opening, closing or switching tabs in Firefox, clicking on links, opening and closing other windows with or without Firefox running etc.) I can build huge software packages that keep compiling for hours as long as I leave the machine alone. I can also watch several two-hour-movies one after another without having to reboot. But browsing the web for half an hour always feels like gambling.

When the system is frozen (again), there is no moving the mouse pointer, no ssh'ing into the machine, no CTRL-Alt-Backspace, no Magic-SysRq-Keys. I have to press and hold the On/Off-Button to switch the computer off.

[Stormking]

I feel like being ignored. The problem is still there. Any new suggestions?

[hugommg]

I feel like being ignored. The problem is still there. Any new suggestions?

Happens to me too. I gave up and just use XAA instead of EXA.

[bridgman]

Sorry, but agd5f is on vacation and does not have computer access. He's not actually ignoring you :D

[Stormking]

Sorry, but agd5f is on vacation and does not have computer access.
Something like that should not be allowed as long as there are critical bugs! ;-)


He's not actually ignoring you :D
You mean, he's not *actively* ignoring me.

[hugommg]

BTW, does anyone even know what the problem is?
Is it radeonhd/radeon? Mesa? DRM?

:S

[Stormking]

Happens to me too. I gave up and just use XAA instead of EXA.
That's not an option for me as I use rotation and XAA is painfully slow, this way.

[slacker]

I feel like being ignored. The problem is still there. Any new suggestions?

I opened a bug ( http://bugs.freedesktop.org/show_bug.cgi?id=16198 ) about these lockups a year ago, so what should I say?? :D
I've been living with these hard lockups for a long time, and now they seem even more frequent with 2.6.31 :(
XAA isn't an option for me either.

[Stormking]

The strange thing about this freeze is that it's completely unpredictable. I had only one crash the whole weekend, today and yesterday the system freezes at least once per hour.

[Stormking]

Sorry, but agd5f is on vacation and does not have computer access. He's not actually ignoring you :D
He seems to be back.

@agd5f: I hope you had a nice and relaxing vacation. But now, please look into this issue! When ATI dropped the support for what they consider legacy hardware, it was promised that the Open Source Driver would get better, fast.

But the current situation is unacceptable. Imagine doing your work on a machine that is about to freeze, any minute. Every single action can trigger the bug, every window you open can mean the end to your current session. It's really, really, really annoying!

[bridgman]

When ATI dropped the support for what they consider legacy hardware, it was promised that the Open Source Driver would get better, fast.

With respect, that is *not* what was said. Please don't misrepresent my words.

We said that the open source drivers would continue to improve, that out developers would help in addition to the community, and that other than 3D we expected the open source drivers would make most users happy already.

[Stormking]

With respect, that is *not* what was said. Please don't misrepresent my words.

We said that the open source drivers would continue to improve, that out developers would help in addition to the community, and that other than 3D we expected the open source drivers would make most users happy already.
I don't care because I am unhappy. Very unhappy!

All I see is work being done on all that fancy stuff (like KMS or adding support for hardware that is actually usuable with fglrx) while I have to be afraid of opening another browser tab!

[bridgman]

I don't care because I am unhappy. Very unhappy!

Understood, but misrepresenting what we said isn't going to help.

We're dealing with a problem that isn't occurring on developer systems AFAIK, so if this is going to get fixed it will be the combination of good guesses on the developers part and precise information from you.

For example, you said that reverting to the older radeon driver which had been reliable before didn't seem to help but that setting MigrationHeuristic to greedy again eliminated the crashes; is that still true ?

[Stormking]

We're dealing with a problem that isn't occurring on developer systems AFAIK, so if this is going to get fixed it will be the combination of good guesses on the developers part and precise information from you.
I will try to help as much as I can. But in the past, it always happened like this:

Me: X is freezing, randomly.
Developers: Try this patch and tell us if it's gone.
Me: I applied the patch but the bug is still there.

At this point, I usually see or hear nothing from you for a few weeks. Instead, I have to bump this thread a few times to get your attention back. It's like talking to someone who falls asleep after each sentence.


For example, you said that reverting to an older radeon driver didn't seem to help but that setting MigrationHeuristic to greedy eliminated the crashes, is that still true ?
Don't know, will try immediately. Expect a report in a few hours.

By the way, I think I can eliminate a heat problem from the list of subjects. A week ago, I've thrown out my WLAN-Mini-PCI-Card and installed an extra heat sink exactly above the graphics hardware. When taking for granted that this machine was ever supposed to work (and it did), I can't imagine it running hot, now.

[bridgman]

What you're seeing is the developers running out of ideas to try. Unless you maintain an up-to-date "this configuration works, that configuration doesn't" picture for the devs they're not going to have anything to base the next idea on.

[Stormking]

What you're seeing is the developers running out of ideas to try. Unless you maintain an up-to-date "this configuration works, that configuration doesn't" picture for the devs they're not going to have anything to base the next idea on.
I will try that but the problem is, there hasn't been a working configuration for a long time.

At the moment, I am running:

- Kernel 2.6.30
- DRM-Module from kernel 2.6.30, no further patches applied
- xserver-xorg-radeon 6.12.99-git20090629.bb04b450-0ubuntu0tormod
- libgl1-mesa-* 7.6.0~git20090620.3b08a43f-0ubuntu0tormod~radeon
- Compiz 0.8.2-0ubuntu8.1

Relevant options from xorg.conf (AFAIK):

- "AccelMethod" "EXA"
- "AccelDFS" "true"
- "DynamicClocks" "false"
- "EnablePageFlip" "false" (because of rotation)
- "MigrationHeuristics" "greedy" (since your posting)

The last days and weeks, I've been running the same configuration (except xserver-xorg-radeon, which was always the lastest one from Tormod's PPA) with MigrationHeuristics set to default (option commented out in xorg.conf). X froze about every two hours (average), sometimes I could work for 4 to 6 hours without a freeze, sometimes it froze five minutes after logging in. But it's safe to say it usually happens after two hours.

[bridgman]

Just checking, have you already tried the usual suggestions, eg taking AGP mode down to 1 etc ?

If you're running a post-6.12.2 driver it might be useful to try ForceLowPowerMode just in case cooling is a factor.

[Stormking]

Just checking, have you already tried the usual suggestions, eg taking AGP mode down to 1 etc ?
Yes, I did. I can't remember, exactly, but I think I can only take it down to 4, 1 seems not to be allowed with my hardware. Freezes still occured.


If you're running a post-6.12.2 driver it might be useful to try ForceLowPowerMode just in case cooling is a factor.
I'll try that after the next freeze.

But as I said before, I don't believe it's a heat problem. If the GPU is running too hot, now, it would never have worked in the first place and I would have brought the laptop back to my dealer a few years ago, when I still had warranty.

[hugommg]

I don't know if this is helpfull but since this bug also affects me here are my settings:
# lspci | grep VGA
01:00.0 VGA compatible controller: ATI Technologies Inc M56P [Radeon Mobility X1600]

I'm using gentoo on a Toshiba A100 laptop:
# uname -a
Linux jenny 2.6.30-gentoo-r1 #1 SMP PREEMPT Mon Jun 15 10:43:35 WEST 2009 i686 Genuine Intel(R) CPU T2400 @ 1.83GHz GenuineIntel GNU/Linux

and this is my xorg.conf:
Section "ServerLayout"
Identifier "Main Layout"
Screen 0 "aticonfig-Screen[0]-0" 0 0
Option "AIGLX" "on"
# Option "AIGLX" "off"
EndSection

Section "Files"
# FontPath "/usr/share/fonts/local"
FontPath "/usr/share/fonts/misc"
FontPath "/usr/share/fonts/Type1"
FontPath "/usr/share/fonts/TTF"
FontPath "/usr/share/fonts/75dpi"
FontPath "/usr/share/fonts/100dpi"
EndSection

Section "Module"
# Load "dbe"
SubSection "extmod"
Option "omit xfree86-dga"
EndSubSection
Load "freetype"
Load "glx"
# Load "xtrap"
Load "extmod"
Load "GLcore"
Load "dri"
Load "dri2"
Load "record"
EndSection

Section "ServerFlags"
Option "DontZap" "false"
Option "blank time" "10" # 10 minutes
Option "standby time" "20"
Option "suspend time" "30"
Option "off time" "60"
EndSection

Section "InputDevice"
Identifier "Mouse1"
Driver "mouse"
Option "Protocol" "Microsoft"
Option "Device" "/dev/mouse"
EndSection

Section "InputDevice"
Identifier "Mouse2"
Driver "mouse"
Option "Protocol" "MouseMan"
Option "Device" "/dev/mouse2"
EndSection

Section "Monitor"
Identifier "aticonfig-Monitor[0]-0"
Option "VendorName" "ATI Proprietary Driver"
Option "ModelName" "Generic Autodetecting Monitor"
Option "DPMS" "true"

EndSection

Section "Device"
Identifier "aticonfig-Device[0]-0"
Driver "radeonhd"
VendorName "ATI Technologies Inc"
BoardName "Radeon Mobility X1600"
BusID "PCI:1:0:0"

#radeonhd settings
Option "AccelMethod" "EXA"
# Option "ExaNoComposite" "true"
# Option "AccelMethod" "XAA"
Option "DRI" "true"
# Option "DRI2" "true"
Option "NoDDC"


# flgrx settings
# Option "DRI" "true"
# Option "XAANoOffscreenPixmaps" "on"
# Option "TexturedVideo" "on"
# Option "VideoOverlay" "off"
# Option "OpenGLOverlay" "off"
# Option "Textured2D" "on"
# Option "TexturedXrender" "on"
# Option "UseFastTLS" "1"
# Option "BackingStore" "on"
# Option "TexturedVideoSync" "True"
# Option "EnableMonitor" "tv"

EndSection

Section "Screen"
Identifier "aticonfig-Screen[0]-0"
Device "aticonfig-Device[0]-0"
Monitor "aticonfig-Monitor[0]-0"
DefaultDepth 24
SubSection "Display"
Viewport 0 0
Depth 24
EndSubSection
EndSection

Section "DRI"
Group 0
Mode 0666
EndSection

Section "Extensions"
Option "DAMAGE" "true"
Option "RENDER" "true"
Option "Composite" "true"
# Option "XVideo" "Enable"
EndSection


XAA works fine, EXA locks up after a while.

Hope it helps.

[tormod]

Yes, I did. I can't remember, exactly, but I think I can only take it down to 4, 1 seems not to be allowed with my hardware. Freezes still occured.
Did you try all choices 4, 2 and 1 (or 8 and 4 for AGPv3 cards)? AGPMode mismatch is the number one cause for crashes on AGP cards. Lately there have also been some issues in 2.6.29 (as agd5f mentioned) which got fixed in 2.6.30 final. Then there have been reported some random crashes with ext4. The best would be if we could track your issues in a bug report so that we can see all your logs and relevant information without wading through a busy discussion forum.

[Stormking]

Did you try all choices 4, 2 and 1 (or 8 and 4 for AGPv3 cards)? AGPMode mismatch is the number one cause for crashes on AGP cards.
I think I have an AGPv3 card and I tested 4 and tried to test 1 (which didn't work). 8 seems to be the default. Anyway, it didn't stop X from freezing.


Lately there have also been some issues in 2.6.29 (as agd5f mentioned) which got fixed in 2.6.30 final. Then there have been reported some random crashes with ext4.

Sorry, only ext2 and ext3, here.

The best would be if we could track your issues in a bug report so that we can see all your logs and relevant information without wading through a busy discussion forum.
Should I use the entry mentioned above (http://bugs.freedesktop.org/show_bug.cgi?id=16198) or open a new one?

Btw: Nearly 4 hours since setting MigrationHeuristics to "greedy" and no freeze, yet.

[tormod]

Should I use the entry mentioned above (http://bugs.freedesktop.org/show_bug.cgi?id=16198) or open a new one?
Please file your own bug. It is not sure you have the same issue as Mikko, and your card is different. Attach your logs and explain everything you have tried, and last good version if there was any.

[Stormking]

Please file your own bug. It is not sure you have the same issue as Mikko, and your card is different. Attach your logs and explain everything you have tried,
I'll collect all the information, tonight. Which logs apart from X and kernel do you need? And btw., I never found anything in there.


and last good version if there was any.
That was around the time when TV-Out for R300 was added, I think.

[Exershio]

I guess I'll add to this thread in hopes it will be fixed eventually.

I'm using a Radeon 9550 (R300) 256mb AGP 4x card with the open source ati drivers. Whenever I run a 3d intensive application, it'll render properly for a few seconds, then my screen goes black, my monitor loses signal, and my computer completely freezes. It takes no input from keyboard/mouse, audio starts stuttering and sounds like a buzzsaw, and I'm forced to shutdown with the power button.

Whenever I'm playing a game that uses harware acceleration but is not very intensive (such as StepMania), the game runs fine for 3-5 minutes and then the same problem I stated previously occurs. Screen goes black, unresponsive, etc.

Whenever I'm just running a simple KDE desktop without any desktop effects, I experience no problems. It's only when I'm running an OpenGL application does the problem happen.

I'm currently using the latest GIT versions of:
xf86-video-ati-git
mesa-git
dri2proto-git
libdrm-git
glproto-git

The freezing has been occuring for at least a year now, even to this day with the latest updates.

System Specs:
Intel P4 2.0ghz (845g motherboard)
ATI Radeon 9550 (R300) 256mb AGP 4x
Arch Linux with latest updates

By the way, this problem does not occur AT ALL on Windows XP with the same AGP Mode (4x). I can even overclock this card heavily with Windows XP without problems, so I doubt it's a heat issue

Here is my Xorg.conf:
Section "ServerLayout"
Identifier "X.org Configured"
Screen 0 "Screen0" 0 0
InputDevice "Mouse0" "CorePointer"
InputDevice "Keyboard0" "CoreKeyboard"
EndSection

Section "Files"
ModulePath "/usr/lib/xorg/modules"
FontPath "/usr/share/fonts/misc"
FontPath "/usr/share/fonts/100dpi:unscaled"
FontPath "/usr/share/fonts/75dpi:unscaled"
FontPath "/usr/share/fonts/TTF"
FontPath "/usr/share/fonts/Type1"
EndSection

Section "Module"
Load "dbe"
Load "dri2"
Load "record"
Load "glx"
Load "dri"
Load "extmod"
EndSection

Section "InputDevice"
Identifier "Keyboard0"
Driver "kbd"
EndSection

Section "InputDevice"
Identifier "Mouse0"
Driver "mouse"
Option "Protocol" "auto"
Option "Device" "/dev/input/mice"
Option "ZAxisMapping" "4 5 6 7"
EndSection

Section "Monitor"
#DisplaySize 380 300 # mm
Identifier "Monitor0"
VendorName "NUL"
ModelName "19''"
HorizSync 24.0 - 80.0
VertRefresh 50.0 - 75.0
Option "DPMS"
EndSection

Section "Device"
Option "AGPMode" "4"
Identifier "Card0"
Driver "radeon"
VendorName "ATI Technologies Inc"
BoardName "RV350 AS [Radeon 9550]"
BusID "PCI:1:0:0"
EndSection

Section "Screen"
Identifier "Screen0"
Device "Card0"
Monitor "Monitor0"
DefaultDepth 24
SubSection "Display"
Viewport 0 0
Depth 1
EndSubSection
SubSection "Display"
Viewport 0 0
Depth 4
EndSubSection
SubSection "Display"
Viewport 0 0
Depth 8
EndSubSection
SubSection "Display"
Viewport 0 0
Depth 15
EndSubSection
SubSection "Display"
Viewport 0 0
Depth 16
EndSubSection
SubSection "Display"
Viewport 0 0
Depth 24
Modes "1280x1024" "1024x768" "800x600" "640x480"
EndSubSection
EndSection

[tormod]

I'm using a Radeon 9550 (R300) 256mb AGP 4x card with the open source ati drivers. Whenever I run a 3d intensive application, it'll render properly for a few seconds, then my screen goes black, my monitor loses signal, and my computer completely freezes. It takes no input from keyboard/mouse, audio starts stuttering and sounds like a buzzsaw, and I'm forced to shutdown with the power button.

Did you try AGPMode 2 and 1?

[Exershio]

Did you try AGPMode 2 and 1?
I'll be damned. I started up StepMania and played it for about 20 minutes with AGP Mode @ 1x and it didn't crash at all. With AGP Mode @ 4x, StepMania would crash within 2-3 minutes at most usually.

I haven't tested it with any really heavy 3d apps because I don't have any installed atm, but I'll do that asap and report back tomorrow.

Thanks for the tip. I hope it continues to work. This is the only problem holding me off Windows for good. If I can get Neverwinter Nights/StepMania/etc to play in Linux without crashing that'd be great. :D

[hugommg]

I moved to xf86-video-ati from xf86-video-radeonhd (gentoo), then I deleted xorg.conf, made X create a new one and added the EXA configuration to it.
After that I did an emerge -e world and no more lockups.