Created attachment 33950 [details]
xorg.conf

Created attachment 33951 [details]
Xorg.0.log

Created attachment 33952 [details]
testcase

What's the expected output of the test case?

Does the problem still occur with xserver Git?

(In reply to comment #4)
> What's the expected output of the test case?
> 

When any key is pressed on a window, the contents of another window is copied to the window.

(In reply to comment #4)
> Does the problem still occur with xserver Git?
> 

Yes, the problem still exists.

This bug doesn't affect a redirected window.

Created attachment 43831 [details]
testcase

Created attachment 43833 [details]
expected result

Created attachment 43836 [details]
actual result

It seems that composite assumes a source drawable is always at (0, 0).
This is true for pixmap and redirected window, but not for unredirected window.

Created attachment 43840 [details] [review]
possible fix

Does this patch help?  Do solid and copy need something similar?

(In reply to comment #9)
> It seems that composite assumes a source drawable is always at (0, 0).
> This is true for pixmap and redirected window, but not for unredirected window.

Sorry, please ignore this. I am confused.
The patch doesn't fix the problem. It also breaks redirected window.

This is not something the driver should need to be concerned with. If the driver acceleration hooks aren't called properly, that needs to be fixed in EXA or maybe even higher up.

-- GitLab Migration Automatic Message --

This bug has been migrated to freedesktop.org's GitLab instance and has been closed from further activity.

You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.freedesktop.org/xorg/xserver/issues/81.