Newer versions of the radeon driver have problems setting the video mode if kernel modesetting is not enabled. The computer hangs for just a while and stays irresponsive; sometimes changes can not be applied (xrandr) which does especially affect game playing. Even worse there is no response in me asking for a fix of kernel modesetting so that I will likely have to boot with it for a long time in the future: "externel monitor does not work with kernel modesetting" - https://bugzilla.novell.com/show_bug.cgi?id=620439. Using: Xorg 1.8.0 2010-04-02 xorg-x11-driver-video-7.5-15.2.x86_64
Please attach your xorg log and dmesg output. Please make sure you aren't trying to mix UMS and KMS. Can you try a newer kernel with KMS? 2.6.36 or 2.6.37?
Unfortunately I did not have time up to now to test with the newest kernel (though I do not expect anything having changed since the described kernel modesetting problem has already remained for month up to when I had filed this bug). Unfortunately I will not be able to do this for the next 5 days where I will remain without internet access. Mixing UMS and KMS? - How could I? If I specify the nomodeset kernel parameter the driver should recognize this. Besides this I wonder how your driver should work with OSes that do not have KMS like FreeBSD this way.
Created attachment 40779 [details] [review] Xorg.0.log of crash during modesetting This time it was no more a hang but something that started as a hang and ended in a crash (for crashes with xorg-x11-driver-video-7.5-15.2 see also Bug 32061).
Well, with the newer 2.6.36 kernels of openSUSE/tumbleweed the situation is like the following: It boots up normally even with kernel modesetting keeping both monitors in function, but then when I try to start an X-server kernel modesetting fails completely: My external screen is suddenly blackened and the color palette of my integrated screen is in the a.. showing pixelry but no cognizable output. So please keep UserLevelModeSetting working! Kernel modesetting will probably not start to work for longingly long epochs. It is always a bad idea to purge support in favour of a prospective feature that yet does not work. People want to work with Xorg now and yet! Older versions are so often not an option due to other flagrant bugs. To me support of an external monitor is coactively necessary as I do not have an own desktop computer but a desktop monitor and a desktop mouse and keyboard as used with a smaller-displayered notebook... and believe me all those guys who have a notebook want to plug it one or another time into their TV to show photos or watch a film - so this is nothing marginal nothing less than a feature of all-out usage, a feature that should not be payed less attention to than implementing new 3d-features used only by a limited number applications. At least people hate to loose something that was already known to work!
So what is the issue now ? UMS is phased out we don't have resource to support it, if it works lucky you, if it doesn't you are on your own, we only care about KMS.
User modesetting has simply been dropped (braking Xorg manys) ?!! How does it then work on other systems like FreeBSD, if I may ask? It does not have kernel modesetting or does it?
As i said if UMS works then it's good, otherwise we don't care and we just encourage other os to move to kms in the meantime they can stick to whatever works for them. If you give me 1millions dollars i am ready to start considering supporting UMS.
Oh, what a delight: Radeon does from now on again work with UMS! (see Bug 32393, Comment #25) (Dacht` ichs mir doch, daß es auch ohne 1 Mio $ geht.)
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.