On May 23, 08 18:44:58 +0200, Matthias Hopf wrote:

This sounds promising that there is a general 64bit issue in the drm/dri framework. I'm about to finish a 64bit installation to verify this.

I just verified that this is a 64bit issue. I'm now able to debug this. The symptoms are exactly what has been discussed here. The most interesting thing is that the very first invocation of a 3D client works as anticipated. So if you *don't* invoke glxinfo first, but start right away with any 3D client, it should work fine this one time. This won't happen before Monday, though, as in Erlangen (that's near Nuremberg) a Comic Saloon is happening during the Weekend... CU Matthias -- Matthias Hopf __ __ __ Maxfeldstr. 5 / 90409 Nuernberg (_ | | (_ |__ mat@mshopf.de Phone +49-911-74053-715 __) |_| __) |__ R & D www.mshopf.de -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org

On May 23, 08 20:45:41 +0200, Sebastian Brocks wrote:

Matthias Hopf schrieb:

...

So if you *don't* invoke glxinfo first, but start right away with any 3D client, it should work fine this one time.

This does not work for me. If I just start X, and run no other programs, and then try to start glxgears, I still get the same error message.

Interesting. I still think this is the same issue, some memory gets overridden, in your case probably earlier than in mine. Matthias -- Matthias Hopf __ __ __ Maxfeldstr. 5 / 90409 Nuernberg (_ | | (_ |__ mat@mshopf.de Phone +49-911-74053-715 __) |_| __) |__ R & D www.mshopf.de -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org

On May 27, 08 09:46:56 +1000, Dave Airlie wrote:

Just FYI, not setting this on Fedora 9 64-bit with radeon and rs690. So its something that is probably fixed in a newer mesa/server.

Thanks for verification, Dave. I used r500-support, latest bits, so it's probably something in the Xserver that changed recently. Most presumably the culling of visual initialization from the Xserver (commit 7d7469053) and following. Matthias -- Matthias Hopf __ __ __ Maxfeldstr. 5 / 90409 Nuernberg (_ | | (_ |__ mat@mshopf.de Phone +49-911-74053-715 __) |_| __) |__ R & D www.mshopf.de -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org

On Tue, May 27, 2008 at 4:09 PM, Russ Dill wrote:

On Tue, May 27, 2008 at 3:12 AM, Matthias Hopf wrote:

...

On May 27, 08 09:46:56 +1000, Dave Airlie wrote:

...

Just FYI, not setting this on Fedora 9 64-bit with radeon and rs690. So its something that is probably fixed in a newer mesa/server.

Thanks for verification, Dave. I used r500-support, latest bits, so it's probably something in the Xserver that changed recently. Most presumably the culling of visual initialization from the Xserver (commit 7d7469053) and following.

xorg-edgers uploaded a new version of the xserver (2:1.4.99.902+git20080524+server-1.5-branch), but there are some conflicts involving the meta package xserver-xorg-input-2 and xserver-xorg-video-2.

With this new xserver (and a manually compiled ati driver) 3d is now working, even with openarena. I have an amd64, so there doesn't seem to be any serious bug with 64 bit systems. I had to rebuild the ati driver manually because the packaged version is using the wrong abi version. -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org