I came across this regression recently and it appears to not been picked up by developers. I guess this is due to it only affecting the RS690 chipset.
If you use XV to display content which has to be scaled it does not work and instead displays a color which changes rapidly depending on what is supposed to be on screen.
Broken since commit:
e932836691aeaec37794fdaed2dabb22710fd171 -- radeon: initial preparation for kms patch.
Last Working commit:
bb04b450ed00ca4b1aa44c33085567d47b33b547 -- Check if the composite op is supported in R200CheckComposite.
duplicate of http://bugs.freedesktop.org/show_bug.cgi?id=22730
(In reply to comment #1)
> duplicate of http://bugs.freedesktop.org/show_bug.cgi?id=22730
Unfortunately although that does appear to be similar to the problem I am experiencing it is no the same.
In my case the corruption only appears if the image if scaled to more than 2x of the original size. This is not the case on the other bug report.
The effect is also different as in my case the result is a single colour and not graphical artifacts as described in the other bug report.
Here are 4 screenshots which demonstrate what I have described above.
As I said these colors will changes rapidly and are consistent if you go over a segment from the video again. As such they are not completely irregular as they are based on the video content.
to me, it looks the exact same bug, just different symptoms, probably due to different chipsets: me and the other guy have R500, you R690.
*** This bug has been marked as a duplicate of bug 22730 ***