Created attachment 102104 [details]
R9 270X card, just updated my software. When restart happened, with a few kernels tried, the screen is corrupt the same every time, and is buggy rendering absolutely everything. Pretty easy to spot. :)
Created attachment 102105 [details]
Screen, looks the same each time.
Created attachment 102106 [details]
My hardware and system info.
Created attachment 102107 [details]
Screenshot of my screen on boot.
What components did you update? kernel? glamor? llvm? mesa? Please attach your xorg log as well.
Created attachment 102138 [details]
Packages installed before rendering problems.
Created attachment 102139 [details]
(In reply to comment #4)
> What components did you update? kernel? glamor? llvm? mesa? Please
> attach your xorg log as well.
Can you narrow down what component caused the regression?
(In reply to comment #7)
> Can you narrow down what component caused the regression?
I'm honestly not too sure, I believe it was the radeon xorg-xserver updates, but can't say for 100% sure at all. I'm checking Mesa's git and updates like always, and it seems the build system was fixed for all modules into one module, this could have updated in the middle of all that for my card. But yet, it doesn't have any errors in xorg and everything seems to have loaded and executed fine, so I sadly can't really pin point one exact component to why this has broken my rendering. My best guess is xorg, but there's not many updates and such there for radeon. So I'm basically stumped as to what is really broken, honestly.
Does reverting 5d5c20920e0e570742a497aa047e99a2fa3c04f2 from mesa help?
(In reply to comment #9)
> Does reverting 5d5c20920e0e570742a497aa047e99a2fa3c04f2 from mesa help?
Well, I tried to test it, but with no luck. Mesa failed to install correctly. Directories were completely wrong (Putting all of them in /usr/local/? HUH?), so they weren't loaded on the DE start. Now my desktop is in fall back because I tried xorg edgers, but their radeon is basically not supported anymore so I'm in fall back. Oh well. If someone else could test, that'd be great. Maybe later I can try to test it, I'll probably reinstall tonight in about...5 hours or so. Sorry about that.
It was most likely commit 5d5c20920e0e570742a497aa047e99a2fa3c04f2, which I've now reverted. Please reopen if the problem persists with the revert.
Author: Michel Dänzer <firstname.lastname@example.org>
Date: Thu Jul 3 10:58:57 2014 +0900
Revert "radeonsi: Use dma_copy when possible for si_blit."