From: Michel <mi...@da...> - 2003-04-23 23:00:47
|
On Mit, 2003-04-23 at 23:49, Benjamin Herrenschmidt wrote: > On Wed, 2003-04-23 at 22:15, Michel Dänzer wrote: > > On Mit, 2003-04-23 at 21:26, Andreas Stenglein wrote: > > > just built texmem-0-0-1 branch, radeon kernelmodule > > > from texmem-0-0-1 branch, installed both > > > and have similar behavior with my Radeon 7500. > > > > > > dmesg shows lots of > > > [drm:radeon_context_switch_complete] *ERROR* Lock isn't held after > > > context switch > > > and tail -f /var/log/messages shows that there are thousands. > > > > I fixed this on the trunk today. The question is whether the problem Ben > > reported is also related to this or maybe even a side effect of my fix. > > Is your fix generic or does it only apply on r200 ? It's in the X server drm module, so very much driver independent. (And even if it was driver dependent, r100 and r200 use the same 2D driver and DRM :). > Your fix got rid of the above message on my M9 (r200 driver) but the > message is still there on the M6 (r100 driver). You must be running an older X server on the M6. Anyway, it looks like we're really gonna have to fix this in the DRM, apparently the DRM system call implementations Keith removed aren't as unused as he thought. :) Which means we need to deal with the locking issues Linus brought up... -- Earthling Michel Dänzer \ Debian (powerpc), XFree86 and DRI developer Software libre enthusiast \ http://svcs.affero.net/rm.php?r=daenzer |