fbPolyArc calls mi functions which can end up calling the accelerated driver. This patch demonstrates what needs to happen to fbPolyArc to it's wrapped/unwrapped correctly. I've just thrown this together in the hope someone can review for the 7.3 release and get this fixed as it can cause rendering glitches.
Created attachment 11425 [details] [review] fbPolyArc fix here's the patch
Created attachment 11426 [details] [review] slight update to the last one update to the last.
Created attachment 11427 [details] [review] stop rushing....
Created attachment 11429 [details] [review] Fixes for EXA to avoid the problem also This patch is for EXA which bumps into similar issues caused by fbPolyArc. We can't rely on this path as fbPolyArc can call out to mi anyway. So let's just call miPolyArc for all cases.
Note, these haven't been committed to master either.
The 'if else' syntax is wrong due to missing brackets, for me.
Created attachment 11437 [details] [review] an update to fix braces This includes both the fb & EXA fixes. Yeah, in the rush to get something coherent I didn't build test, and it ended up in-coherent. Sorry. Attached is an updated patch.
EXA part pushed to master and no-pci-rework branches.
Commit the fb part and add FB_ACCESS_WRAPPER around them as well.
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.