Bug 204810
Summary: | latest compiz locks entire machine | ||||||||
---|---|---|---|---|---|---|---|---|---|
Product: | [Fedora] Fedora | Reporter: | Thomas J. Baker <tjb> | ||||||
Component: | compiz | Assignee: | Kristian Høgsberg <krh> | ||||||
Status: | CLOSED RAWHIDE | QA Contact: | |||||||
Severity: | medium | Docs Contact: | |||||||
Priority: | medium | ||||||||
Version: | rawhide | CC: | clydekunkel7734, matthias, mcepl, ralph+rh-bugzilla, xgl-maint | ||||||
Target Milestone: | --- | Keywords: | Reopened | ||||||
Target Release: | --- | ||||||||
Hardware: | All | ||||||||
OS: | Linux | ||||||||
Whiteboard: | |||||||||
Fixed In Version: | Doc Type: | Bug Fix | |||||||
Doc Text: | Story Points: | --- | |||||||
Clone Of: | Environment: | ||||||||
Last Closed: | 2007-10-02 17:55:30 UTC | Type: | --- | ||||||
Regression: | --- | Mount Type: | --- | ||||||
Documentation: | --- | CRM: | |||||||
Verified Versions: | Category: | --- | |||||||
oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |||||||
Cloudforms Team: | --- | Target Upstream Version: | |||||||
Embargoed: | |||||||||
Attachments: |
|
Description
Thomas J. Baker
2006-08-31 17:52:10 UTC
September 1st rawhide fixed things. compiz works again. After todays large update, this problem is back. Right after the update, I rebooted, logged in, and before anything appeared on the desktop, the whole system hung. I power cycled, reboooted, logged in failsafe, turned off desktop-effects, logged out, logged back in, turned back on desktop-effects and everything was fine. If it's like last time, everything will be fine until some future update. (In reply to comment #2) > After todays large update, this problem is back. Right after the update, I > rebooted, logged in, and before anything appeared on the desktop, the whole > system hung. I power cycled, reboooted, logged in failsafe, turned off > desktop-effects, logged out, logged back in, turned back on desktop-effects and > everything was fine. If it's like last time, everything will be fine until some > future update. How do you disable desktop-effects in failsafe login? I'm having this exact same problem on my laptop and I haven't had time to search BZ until now. My backup plan for when the default Fedora desktop no workie is to fire up XFce and grumble until GNOME works again. I update my system daily, so I think I have the latest version of compiz, I just need to do the disable/enable trick to get it working again (if that worked for you). Log in failsafe and run 'desktop-effects' from the command line. Log out and log back in with the normal gnome session and then, at least for me, you can re-enable desktop-effects and everything works. I just experienced the same problem again after todays updates. Everything went just like comment #2. I even tried twice (login, lock, power cycle, login lock) before using the failsafe recovery technique. The only updates today were these: continuity> rpm -qa --last | more krb5-workstation-1.5-7 Thu 07 Sep 2006 12:54:16 PM EDT desktop-backgrounds-basic-2.0-37 Thu 07 Sep 2006 12:54:15 PM EDT kernel-headers-2.6.17-1.2630.fc6 Thu 07 Sep 2006 12:54:13 PM EDT kernel-devel-2.6.17-1.2630.fc6 Thu 07 Sep 2006 12:54:03 PM EDT policycoreutils-1.30.28-4 Thu 07 Sep 2006 12:53:51 PM EDT kernel-2.6.17-1.2630.fc6 Thu 07 Sep 2006 12:53:32 PM EDT fedora-release-notes-5.92-2 Thu 07 Sep 2006 12:53:26 PM EDT redhat-rpm-config-8.0.45-3 Thu 07 Sep 2006 12:53:24 PM EDT yelp-2.16.0-2.fc6 Thu 07 Sep 2006 12:53:02 PM EDT fedora-logos-1.1.50-1.fc6 Thu 07 Sep 2006 12:52:56 PM EDT krb5-devel-1.5-7 Thu 07 Sep 2006 12:52:52 PM EDT tomboy-0.4.1-1.fc6 Thu 07 Sep 2006 12:52:39 PM EDT rhpl-0.188-3 Thu 07 Sep 2006 12:52:28 PM EDT mkinitrd-5.1.11-1 Thu 07 Sep 2006 12:52:26 PM EDT krb5-libs-1.5-7 Thu 07 Sep 2006 12:52:19 PM EDT Maybe it's related to a new kernel? It's gotton worse for me. Now everytime I boot, I have to do the disable desktop-effects from failsafe trick or my laptop hangs. Verfied this morning before any updates and again after today's updates. I am seeing similar behavior on a Thinkpad T43 w/Chipset ATI Radeon Mobility X300 (M22) 5460 (PCIE) using XAA acceleration. The machine is fully up to date with rawhide as of 9/10/06. The machine hangs after rebooting and attempting to log into Gnome via gdm -- unless you first log into a failsafe terminal and disable/re-enable compiz via desktop effects. Mostly a "me too" here. It happens on my laptop which has a Mobility 9600, but not on my desktop which has a 9200 nor on my other laptop which has an Intel i9xx. After a clean install of fc6t3 this morning and then installing compiz-0.0.13-0.23.20060817git.fc6 from development, I was able to log in with compiz enabled after a reboot. After a couple of reboots this weekend, I still have the problem about half the time. After the clean install on Friday, I didn't do any updates until Sunday night. I have the lockup before and after Sunday night updates about half the time I reboot. So far, there doesn't seem to be any correlation to anything else (like on power/on battery or no network/wired network/wireless network or cold boot/reboot.) It's now pretty consistent that if I boot with the power adapter plugged in and log in with compiz enabled, the system will hang. Boot while on battery and everything works as expected. Seems like a kernel problem to me. Matthais, can you duplicate it? Sorry, I've just switched my laptop for a workstation (with an NVidia card, so no compiz eye candy for me anymore) :-( Well I've just had it lock three times on battery so it's not as simple as I had hoped. Once while logging out and twice while logging in. I'm trying to debug a cpu scaling bug and with cpufreq.debug=7 added to the kernel command line, I'm not having any luck with compiz. I was able to get these errors from dmesg: [drm:radeon_cp_start] *ERROR* radeon_cp_start called without lock held [drm:radeon_cp_idle] *ERROR* radeon_cp_idle called without lock held [drm:radeon_cp_reset] *ERROR* radeon_cp_reset called without lock held [drm:radeon_cp_start] *ERROR* radeon_cp_start called without lock held [drm:radeon_cp_idle] *ERROR* radeon_cp_idle called without lock held [drm:radeon_cp_reset] *ERROR* radeon_cp_reset called without lock held [drm:radeon_cp_start] *ERROR* radeon_cp_start called without lock held [drm:radeon_cp_idle] *ERROR* radeon_cp_idle called without lock held [drm:radeon_cp_reset] *ERROR* radeon_cp_reset called without lock held [drm:radeon_cp_start] *ERROR* radeon_cp_start called without lock held [drm:radeon_cp_idle] *ERROR* radeon_cp_idle called without lock held [drm:radeon_cp_reset] *ERROR* radeon_cp_reset called without lock held [drm:radeon_cp_start] *ERROR* radeon_cp_start called without lock held Then the whole machine locks up. I'm not sure why it's suddenly gotton so much worse but it may be related to yesterdays updates. Created attachment 137337 [details]
Screenshot of compiz weirdness
Screenshot of compiz flaking out on a Dell D600, ATI Radeon Mobility M9.
I'll add a 'me too' to this, with some additional information. Hardware is a Dell D600, Chipset: "ATI Radeon Mobility 9000 (M9) Lf (AGP)" (ChipID = 0x4c66). 'yum update' as of September 28th. If I run desktop-effects from inside a GNOME session, metacity exits and the machine locks hard (alt-sysreq doesn't work at all). If I log in with a 'failsafe' session, start gconfd manually, then start gnome-window-decorator and 'compiz --replace gconf' from the failsafe xterm, compiz starts. However, the right half of the screen has issues that I can only describe via a screenshot. All effects (cube, wobble, etc) appear to work from failsafe. I can't consistently reproduce this crash, though I've gotten it to lock up a few times. Could you guys try adding Option "XaaNoOffscreenPixmaps" to the "Device" section of your xorg.conf files? Thanks, Kristian Unfortunately for debugging, I no longer have that laptop at my disposal. Sorry. Looks like no one else who still has an affected machine is on the CC list either. This seems to work (XaaNoOffscreenPixmaps), but leads to some graphical errors on the Desktop (see attached bling.png file). The machine does not lockup anymore when desktop-effects are enabled. Machine is an IBM T40 with an ATI M9 card - the same as the Dells in this bug. compiz-0.0.13-0.31.20060817git.fc6 xorg-x11-drv-ati-6.6.2-3.fc6 xorg-x11-server-Xorg-1.1.1-43.fc6 mesa-libGLU-6.5.1-7.fc6 mesa-libGL-6.5.1-7.fc6 Bug 207905 is a duplicate of this bug. You can close that, then Created attachment 137581 [details]
compiz working with XaaNoOffscreenPixmaps - but weird effects
This is a screenshot with the proposed Setting XaaNoOffscreenPixmaps in the
Devices Section.
Just forgot: compiz works great, if I run X with 800x600 (had some problems with X today coming up in that resolution instead of 1400x1050 and just had to try it). The ATI has 32MB *** Bug 207905 has been marked as a duplicate of this bug. *** (In reply to comment #19) > This seems to work (XaaNoOffscreenPixmaps), but leads to some graphical errors > on the Desktop (see attached bling.png file). The machine does not lockup > anymore when desktop-effects are enabled. OK, thanks, that's interesting. The artifacts you're seeing now is bug #203282, there's a workaround in that bug. Nope, that workaround doesn't work here. I already have the /etc/drirc in place: [ralph@logout ~]$ ls -l /etc/drirc -rw-r--r-- 1 root root 174 Sep 24 16:04 /etc/drirc [ralph@logout ~]$ cat /etc/drirc <driconf> <device screen="0" driver="radeon"> <application name="all"> <option name="allow_large_textures" value="2" /> </application> </device> </driconf> This bug should be fixed and the new build will be in rawhide soon. In the meantime, you can get it here: http://people.redhat.com/krh/xorg-x11-server-Xorg-1.1.1-44.fc6.i386.rpm Please give it a try and let me know how it goes. BTW, when testing the build in comment #25, make sure to remove the workaround from comment #17. The package from comment #25 solves the lockup problem but I still see the graphics problem from comment #20 - the /etc/drirc doesn't solve that. But it's a great step forward on this machine :) Okay, if I substitute driver="radeon" with driver="r200" in the drirc from comment #24, everything works as expected. From my view the bug can be closed. Following comment 28, I would close this bug. Reporter, do you agree? If you won't reply in one month, I will have to close this bug as INSUFFICIENT_DATA. Thank you. I am the original reporter but as I said in comment #18, I no longer have the laptop to test. Some one with relevant hardware would be better suited to comment. Yes, you are right. Closing because of overwhelming agreement of non-reproducability ;-). |