Created attachment 372820 [details] Xorg.log with KMS enabled Description of problem: I have an LCD monitor working at a 1920x1080 resolution. I am using an ATI HD3450 (RV620) card. When disabling Kernel Mode Setting (nomodeset) Xorg automatically detects and use the native resolution 1920x1080. When using KMS plymouth correctly detects and use the native 1920x1080 resolution but Xorg does not : it uses a low resolution and s-c-d does not offer me the 1920x1080 resolution. Version-Release number of selected component: I am using the following package version from updates-testing : kernel-2.6.31.6-134.fc12.x86_64 xorg-x11-server-Xorg-1.7.1-7.fc12.x86_64 xorg-x11-drv-ati-6.13.0-0.11.20091119git437113124.fc12.x86_64 Additional info: Similar issue reported in Bug 496335 with intel driver. But it is not clear whether the bug is solved or not, even if closed ... I am not using any Xorg.conf, I use the Xorg's auto detection. See attached Xorg.log (with KMS enable/disable) I have attached the 2
Created attachment 372821 [details] Xorg.log with KMS disabled
Hi all ! This is just to notify that the problem still exists with latest kernel 2.6.31.6-145.fc12
Hi Charles-Henri, Could you please attach /var/log/dmesg after booting with drm.debug=1 ? Thanks in advance, -- Fedora Bugzappers volunteer triage team https://fedoraproject.org/wiki/BugZappers
Created attachment 381103 [details] dmesg with drm.debug=1
Hi François, Please find attached the requested dmesg with drm.debug=1. Please note that KMS was disabled at that time. I do not know if it matters or not ... If so do not hesitate to request another log file with KMS enabled and drm.debug=1 Thanks a lot for your help here ! Cheers, CH
Hi Charles-Henri, Sorry, I should have been more specific: yes, we need it without "nomodeset", i.e. with KMS. Otherwise there won't be anything useful in it. Could you attach another log file without nomodeset and with drm.debug=1 ? Thank you! -- Fedora Bugzappers volunteer triage team https://fedoraproject.org/wiki/BugZappers
Created attachment 381662 [details] dmesg with drm.debug=1 and KMS enabled No problem ! I should have given both :-) Here is the requested log file with KMS enabled. Thanks again for your help !
Charles-Henri, Thank you for the logs. Note: this could be related to 538486. -- Fedora Bugzappers volunteer triage team https://fedoraproject.org/wiki/BugZappers
Hello, See also bug #553037 : Using KMS with mesa-dri-experimental makes Xorg crash.
Please attach Xorg log and dmesg with KMS enabled and drm.debug=15 thanks.
Created attachment 383352 [details] dmesg with drm.debug=15 and KMS enabled and dri-experimental Please note that I have mesa-dri-drivers-experimental installed (see bug #553037).
Created attachment 383353 [details] Xorg.log with KMS and drm.debug=15 and mesa-dri-drivers-experimental Please note that Xorg segfault. I can install -debug package if you want. Just tell me those you are interested in ...
If you need more info or testing packages from updates-testing do not hesitate to ask ! Again : thanks a lot for your time and help on this issue !!!
If you use package from update testing you will need to upgrade all libdrm and all mesa and all ddx package. Otherwise it will segault.
Created attachment 383551 [details] Xorg.log with KMS and drm.debug=15 and mesa-dri-drivers-experimental and updates from updates-testing
Created attachment 383552 [details] dmesg with drm.debug=15 and KMS enabled and dri-experimental and upated packages
Hello ! I have some more news. I have updated key components from updates-testing : kernel.x86_64 2.6.31.6-162.fc12 @updates libdrm.x86_64 2.4.17-1.fc12 @updates-testing mesa-dri-drivers.x86_64 7.7-2.fc12 @updates-testing mesa-dri-drivers-experimental.x86_64 7.7-2.fc12 @updates-testing mesa-libGL.x86_64 7.7-2.fc12 @updates-testing mesa-libGLU.x86_64 7.7-2.fc12 @updates-testing plymouth.x86_64 0.8.0-0.2009.29.09.19.1.fc12 @updates-testing xorg-x11-drivers.x86_64 7.3-13.fc12 @anaconda- xorg-x11-server-Xorg.x86_64 1.7.4-1.fc12 @updates xorg-x11-server-common.x86_64 1.7.4-1.fc12 @updates xorg-x11-drv-ati.x86_64 6.13.0-0.20.20091221git4b05c47ac.fc12 @updates-testing As expected by Jerome Xorg no more segfault. I will close bug #553037 since the 3D works great with mesa-dri-drivers-experimental and KMS enabled : I have only tried Gnome Shell (awesome btw ...) and playing high quality video in full screen but all went fine so far ! However : I still cannot get my screen to its full resolution with KMS enabled. More ever I do not think that KMS is working well since I have the same result, on the speed point of view, when switching to a VT or to another user whatever KMS is enabled or not. For a daily usage I have to deactivate KMS (nomodset) unfortunatly, since the resolution is far too low. I have attached new Xorg.log and dmesg using the updated packages above, KMS enabled and drm.debug=15. Sorry for having messed-up things with my mix of updates and updates-testing ... I will be more carefull next time ! I hope it helped ! Cheers, Charles-Henri
Created attachment 384416 [details] patch against latest kernel Hello, Can you please try the attached patch? I think it should fix your problem. Joshua Roys
Hello ! Congratulation Joshua, you did the right patch ! A big thank you ! I have patched kernel-2.6.31.9-174.fc12 with the attached patch and it worked great : The screen resolution is correctly detected with KMS enabled, KMS work great when switching user or to a VT, 3D work with mesa-dri-drivers-experimental , my RV620 ATI GPU and free Xorg ATI driver. It is Christmas every day with the Fedora Project !!! I have a few questions regarding this patch : 1- Should this bug be assigned to kernel component instead of xorg-x11-drv-ati ? 2- My hardware does not have a given feature (integrated sync apparently and stereo, no idea of what it is), or is the kernel/xorg not able to use one of my hardware feature yet ? 3- Has this patch been submitted to upstream kernel (I did not found it in Linus's tree) yet ? Where can I follow its "life" ? 4- Is there any chance to have this patch applied to F12 kernel or F13 kernel until it is applied upstream ? Again, thanks a lot to all of you for your time and your help ! If you need more testing do not hesitate to ask.
It's in upstream queue : http://git.kernel.org/?p=linux/kernel/git/airlied/drm-2.6.git;a=commit;h=79b7dcb2af7da728819cf060b94ce8c9ead36f7e I will try to get it in f12.
Thanks a lot Jerome ! However no pressure here : this is not a critical bug for me on F12 as long as KMS is not required for critical desktop features. If you cannot get this patch into F12 (time or technical issues) I will recompile the kernel updates until the patch is applied upstream, this is not big deal.
Hello, I have some more news about this subject. When using KMS the patch makes my system unstable with either Xorg freeze or strange memory leak issue. When using firefox and flash enabled website : the memory usage is growing very fast, then the computer is swapping and then it freezes. I do not have such issues with standard kernel. I have tried with kernel-2.6.31.12-174.2.3 and kernel-2.6.31.9-174. I am going to attach dmesg and Xorg.0.log (before the freeze, with modesetting enabled and drm.debug=1) Feel free to ask for more testing or logs. Cheers, CH
Created attachment 386248 [details] dmesg with KMS and drm.debug=15 and attached drm patch applied
Created attachment 386249 [details] Xorg.log with KMS and drm.debug=15 and attached drm patch applied
I also tested that installing or removing mesa-dri-drivers-experimental has no impact here : the issue still occurs. Maybe more quickly without mesa-dri-drivers-experimental, but this is just a not scientific feeling !
THis patch can't be the reason of instability, it just allow some video mode to be accepted (or it means we have serious memory corruption bug in radeon). Which kernel are you using exactly ?
I am using Kernel 2.6.31.12-174.2.3 on F12 x86_64 with the patch applied as the only customization : I just modified the kernel.spec file to add a Patch and Apply patch line, and then built using mock -with-baseonly and -with-headers and -with-firmware (I do not remember the exact mock command but the idea is here). I can easily reproduce the issue : just use my computer for a few minutes with Firefox from F12 stable repositories and x86_64 flash plugin from Adobe. Sometimes I do not even need to use Firefox to get the memory leak issue. I do not have this issue while disabling KMS ... Tell me if you want more testing or log files. Also if you wish I can apply the patch on the coming kernel-2.6.32.4-30.fc12 (http://koji.fedoraproject.org/koji/buildinfo?buildID=152350) and try to reproduce the issue with such Kernel, if 2.6.32.4 can be rebuilt on its own without more package updates ... If the patch is not the cause, then since it makes KMS work with my card, it may be a KMS or Xorg or ATI radeon bug ? Do not hesitate to ask for more information ! Thanks ! CH.
Hello ! As far as I understand the patch has landed in Linus's tree : http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=79b7dcb2af7da728819cf060b94ce8c9ead36f7e And guess what ? Today's update on F12 with kernel 2.6.32.9-67.fc12 seems to fix this issue ! I have not tried again mesa-dri-drivers-experimental, I will as soon as possible. I think this bug can be closed as solved ... A big thank you to all of you for your help and work. Cheers, CH.
Thank you for letting us know.