You: > [ Owen, 'the patch' below refers to >http://cgit.freedesktop.org/xorg/xserver/commit/?id=7c8327f0a75087a85864256a9cea80dd4b86def5 > > ] > > > Any corrupted glyph in the cache should get replaced sooner > > > or later with or without the patch. > > > > Is there a way to explicitely invalidate the cache? > > No. The only thing that might indirectly have a similar effect is VT > switching to console and back to X.
I think I already tried without success, but am not really sure. Will take a look at the next time... > > In my observation, once a glyph is corrupted it stays corrupted. > > That could mean either: > > * The actual glyph pixmap gets corrupted, not the glyph cache > pixmap. This seems rather unlikely, as this pixmap is never > rendered to after the glyph is stored in it. > * The glyph cache algorithm never evicts the corrupted glyph from > the cache for some reason. I don't fully understand this > algorithm, you may want to discuss this possibility with Owen > Taylor (CC'd). Third variant, I did not wait long enough ;-). Regardless, I can still see corruption with ExaNoUploadToScreen but it occurs not so often (my stomach says, about 1/5 of the original rate) and indeed normally "heals" quite quickly. > > > Does > > > > > > Option "ExaNoUploadToScreen" > > > > > > work around the problem, with or without the patch? > > > > Currently testing without the patch, corrupted glyphs within 30min > > after server startup. > > Okay, this might be the problem fixed by the patch. Then I'll again try with the patch and report the result. _______________________________________________ xorg-driver-ati mailing list [email protected] http://lists.x.org/mailman/listinfo/xorg-driver-ati
