On 30 Jan 2003, Nolan Leake wrote: >While poking around the tree, I happened across the reason why >RENDER acceleration was screwed up on dualhead G[45][05]0s. > >xaaStateChange.c was not calling RestoreAccelState before doing >CPUToScreenTexture and CPUToScreenAlphaTexture. > >This patch fixes that, however it does _NOT_ re-enable MGA's >RENDER accel with dualhead.
The RENDER problems were not just in dualhead, although most people experienced them in dualhead. I've had bug reports of singlehead corruption with RENDER as well, however it is possible the two issues are separate. I've disabled RENDER accel on mga in our builds completely for the last month or so roughly, but I can re-enable it and test your patch out. I'll have to find the CVS commit that disabled render accel on dualhead first and revert it. I'll report back here when I've had a few people bang on it. -- Mike A. Harris ftp://people.redhat.com/mharris OS Systems Engineer - XFree86 maintainer - Red Hat _______________________________________________ Devel mailing list [EMAIL PROTECTED] http://XFree86.Org/mailman/listinfo/devel