Distribution: Ubuntu 4.2 Warty Hardware Environment: Averatec 3225 Software Environment: Linux 2.6.9 2.6.8.1 had no support for via_agp suspend/resume, but 2.6.9 added such support. It works (my machine no longer freezes on resume), but now there is still an issue. Basically, SOME (but not all) 3D applications become distorted after the resume. The distortion is described by the attached screenshots of foobillard and tuxracer. I am actually able to fix this issue by taking the following steps: (1) shut down X (/etc/init.d/gdm stop) (2) rmmod -f via_agp (3) suspend my machine (4) resume my machine (5) modprobe via_agp (6) start x (/etc/init.d/gdm start) If I do those steps, the distortion goes away. Obviously an unweildy six steps, but doesn't it indicate that via_agp still isn't suspending or resuming properly? I'm not sure because I'm unfamiliar with Linux agp code. If you want to reproduce these problems, simply suspend and resume a VIA-based machine and try the above apps. Strangely enough, about HALF the time the suspend/resume cycle does NOT cause the above problems, but half the time it does. Thanks.
Created attachment 4223 [details] tuxracer after resume
Created attachment 4224 [details] foobillard after resume
the attachments are html files that point to a http://www.pixelmonkey.org webserver which doesn't have those files. Please attach the actual files, and not forwarders like this. Is this still a problem with 2.6.15 ?
Created attachment 6941 [details] tux "inversion" after resume in tuxracer
Created attachment 6942 [details] foobillard table disappearance after resume
Created attachment 6943 [details] glxgears "inversion"
Sorry about the old links, I posted this bug awhile ago and changed web server since then. There are the new ones. I'm currently running 2.6.11, and still have the same issues. I guess you want me to try the DRI support in 2.6.15. I downloaded the new kernel yesterday, so hopefully I'll give it a shot soon and tell you if the problems are still there.
Andrew, what is the status of this issue in kernel 2.6.17?
Please reopen this bug if it's still present in kernel 2.6.17.
Sorry guys, I actually switched away from my old laptop, so it's difficult for me to say whether the bug still exists in 2.6.17.