[opensuse] A New ATI Driver Problem
Hi, I just noticed that my kernel log (/var/log/messages and virtual console 10) are being flooded with messages like this: Feb 9 12:24:18 twain atieventsd[4149]: Too many Event Notification connection errors. Giving up! Searching the Web is not proving fruitful. Has anyone seen this? Does anyone know what it means or might mean? Any idea how to fix it? Randall Schulz -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Randall R Schulz wrote:
Hi,
I just noticed that my kernel log (/var/log/messages and virtual console 10) are being flooded with messages like this:
Feb 9 12:24:18 twain atieventsd[4149]: Too many Event Notification connection errors. Giving up!
Searching the Web is not proving fruitful. Has anyone seen this? Does anyone know what it means or might mean? Any idea how to fix it?
Randall Schulz
Never seen that before. Did find a link or two: http://www.phoronix.com/forums/showthread.php?t=10090 http://groups.google.com/group/linux.debian.changes.devel/browse_thread/thre... Looks like some folks were deleting atieventsd. Which driver are you running? -- David C. Rankin, J.D.,P.E. Rankin Law Firm, PLLC 510 Ochiltree Street Nacogdoches, Texas 75961 Telephone: (936) 715-9333 Facsimile: (936) 715-9339 www.rankinlawfirm.com -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Monday February 9 2009, David C. Rankin wrote:
Randall R Schulz wrote:
Hi,
I just noticed that my kernel log (/var/log/messages and virtual console 10) are being flooded with messages like this:
Feb 9 12:24:18 twain atieventsd[4149]: Too many Event Notification connection errors. Giving up!
...
Randall Schulz
Never seen that before. Did find a link or two:
http://www.phoronix.com/forums/showthread.php?t=10090 http://groups.google.com/group/linux.debian.changes.devel/browse_thre ad/thread/002c535876669683
Thanks.
Looks like some folks were deleting atieventsd. Which driver are you running?
% rpm -q --whatprovides $(type -p atieventsd) x11-video-fglrxG01-8.561-1.1 ... and / or: % rpm -qa |egrep -i fglrx ati-fglrxG01-kmp-pae-8.561_2.6.27.7_9.1-1.1 x11-video-fglrxG01-8.561-1.1 (From the AMD / ATI repository for openSUSE 11.1.)
-- David C. Rankin
Randall Schulz -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Randall R Schulz wrote:
On Monday February 9 2009, David C. Rankin wrote:
Randall R Schulz wrote:
Hi,
I just noticed that my kernel log (/var/log/messages and virtual console 10) are being flooded with messages like this:
Feb 9 12:24:18 twain atieventsd[4149]: Too many Event Notification connection errors. Giving up!
...
Randall Schulz Never seen that before. Did find a link or two:
http://www.phoronix.com/forums/showthread.php?t=10090 http://groups.google.com/group/linux.debian.changes.devel/browse_thre ad/thread/002c535876669683
Thanks.
Looks like some folks were deleting atieventsd. Which driver are you running?
% rpm -q --whatprovides $(type -p atieventsd) x11-video-fglrxG01-8.561-1.1
... and / or:
% rpm -qa |egrep -i fglrx ati-fglrxG01-kmp-pae-8.561_2.6.27.7_9.1-1.1 x11-video-fglrxG01-8.561-1.1
(From the AMD / ATI repository for openSUSE 11.1.)
OK, still on the 8-12 driver from the repository and not from the ati-driver-package. Huh? This is a crappy answer, but I have got to hand over the reins to you on this one, you be in uncharted waters, well beyond the charted seas of my knowledge. I could only suggest, you try the driver package from the ati site and see if the problem persists. My understanding from the links above is that the atieventd daemon/module whatever it is, is the part of the package that responds to machine/user input events like 'lid closed' 'sleep' 'suspend' and for whatever reason the driver you have on your box is interpreting something as an event that it shouldn't even be listening to and whatever it is, the conversation 'it' and atieventd seems quite vivacious. You also might want to research the 'aticonfig --set-policy=STRING' string.
From the aticonfig man page it discusses:
External Events Daemon Options: Following options will not change the config file. They are used to send commands to the atieventsd external events daemon. --set-policy=STRING Sets the event policy for the daemon to be STRING. See the atieventsd(8) manpage for further details. Maybe you can turn the listening off? -- David C. Rankin, J.D.,P.E. Rankin Law Firm, PLLC 510 Ochiltree Street Nacogdoches, Texas 75961 Telephone: (936) 715-9333 Facsimile: (936) 715-9339 www.rankinlawfirm.com -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Monday February 9 2009, David C. Rankin wrote:
Randall R Schulz wrote:
...
% rpm -qa |egrep -i fglrx ati-fglrxG01-kmp-pae-8.561_2.6.27.7_9.1-1.1 x11-video-fglrxG01-8.561-1.1
(From the AMD / ATI repository for openSUSE 11.1.)
OK, still on the 8-12 driver from the repository and not from the ati-driver-package. Huh? This is a crappy answer, but I have got to hand over the reins to you on this one, you be in uncharted waters, well beyond the charted seas of my knowledge.
You shouldn't feel obligated nor should you apologize.
I could only suggest, you try the driver package from the ati site and see if the problem persists.
I tried (very clumsily) to switch to the radeon driver, and was sorry. I was picking up the pieces for quite a while. I have, frankly, never had a good experience trying to modify or reconfigure anything at the X11 server or video-card driver level, and if "once bitten, twice shy," then I'm very, very shy about this.
My understanding from the links above is that the atieventd daemon/module whatever it is, is the part of the package that responds to machine/user input events like 'lid closed' 'sleep' 'suspend' and for whatever reason the driver you have on your box is interpreting something as an event that it shouldn't even be listening to and whatever it is, the conversation 'it' and atieventd seems quite vivacious.
That might make it a culprit the other symptom I reported but about which I received no replies: At certain intervals after the machine becomes idle, the X11 server begins consuming 100%, mostly kernel mode. That condition comes and goes a few times (I don't know the pattern or timing very precisely 'cause it happens on intervals of 10s of minutes and I haven't had the patience to monitor it carefully or long enough) but eventually becomes "permanent" (instead of intermittent). As soon as I touch the mouse or keyboard, everything goes back to normal (I can watch the patterns using htop from another system logged in via ssh). Another thing I noticed is that the /etc/init.d script for atieventsd seems to have some bugs (or not be properly written), 'cause while I can see the process running, YaST's System Services (Runlevel) module reports it as stopped.
You also might want to research the 'aticonfig --set-policy=STRING' string. From the aticonfig man page it discusses:
External Events Daemon Options: Following options will not change the config file. They are used to send commands to the atieventsd external events daemon. --set-policy=STRING Sets the event policy for the daemon to be STRING. See the atieventsd(8) manpage for further details.
Maybe you can turn the listening off?
At the moment, I just did a "/etc/init.d/atieventsd stop". I'll see what consequences that has...
-- David C. Rankin
Randall Schulz -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Randall R Schulz wrote:
That might make it a culprit the other symptom I reported but about which I received no replies: At certain intervals after the machine becomes idle, the X11 server begins consuming 100%, mostly kernel mode. That condition comes and goes a few times (I don't know the pattern or timing very precisely 'cause it happens on intervals of 10s of minutes and I haven't had the patience to monitor it carefully or long enough) but eventually becomes "permanent" (instead of intermittent). As soon as I touch the mouse or keyboard, everything goes back to normal (I can watch the patterns using htop from another system logged in via ssh).
What is your screensaver interval? kpowersaved suspend interval? It might be that your box is trying to go into sleep mode and fails and then has to come back to life. Maybe X consuming 100% is the machine state being processed for suspend or sleep, then the suspend or sleep fails, the machine is back up, then the next sleep interval later, the process repeats itself? Just a SWAG at it. (swinging wild ass guess) -- David C. Rankin, J.D.,P.E. Rankin Law Firm, PLLC 510 Ochiltree Street Nacogdoches, Texas 75961 Telephone: (936) 715-9333 Facsimile: (936) 715-9339 www.rankinlawfirm.com -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Tuesday February 10 2009, David C. Rankin wrote:
Randall R Schulz wrote:
That might make it a culprit the other symptom I reported but about which I received no replies: At certain intervals after the machine becomes idle, the X11 server begins consuming 100%, mostly kernel mode. That condition comes and goes a few times (I don't know the pattern or timing very precisely 'cause it happens on intervals of 10s of minutes and I haven't had the patience to monitor it carefully or long enough) but eventually becomes "permanent" (instead of intermittent). As soon as I touch the mouse or keyboard, everything goes back to normal (I can watch the patterns using htop from another system logged in via ssh).
What is your screensaver interval? kpowersaved suspend interval?
The symptom (kernel-mode CPU saturation of one core) does not correspond to the screen saver engaging. Nor does it correspond in any straightforward to any of the intervals of the successive DPMS power-save modes. In fact, it happens even when I use the "Presentation" mode. Also, the high CPU use comes and goes a few times before becoming permanent.
It might be that your box is trying to go into sleep mode and fails and then has to come back to life. Maybe X consuming 100% is the machine state being processed for suspend or sleep, then the suspend or sleep fails, the machine is back up, then the next sleep interval later, the process repeats itself?
Just a SWAG at it. (swinging wild ass guess)
I suspected (and still suspect) that its related to DPMS, but I can't see a clear relationship.
-- David C. Rankin
Randall Schulz -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Monday February 9 2009, Randall R Schulz wrote:
Hi,
I just noticed that my kernel log (/var/log/messages and virtual console 10) are being flooded with messages like this:
Feb 9 12:24:18 twain atieventsd[4149]: Too many Event Notification connection errors. Giving up!
Also, oddly enough, this began only three days ago, though I've been running the ATI / AMD 8.561 driver for a few weeks, now. Randall Schulz -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
participants (2)
-
David C. Rankin
-
Randall R Schulz