[Bug 93816] 915GM/945GM XRender does not display stretched image when multiple windows overlap on second display.

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=93816

marc-andre.parent at grassvalley.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |marc-andre.parent at grassvall
                   |                            |ey.com

--- Comment #2 from marc-andre.parent at grassvalley.com ---
Created attachment 121466
  --> https://bugs.freedesktop.org/attachment.cgi?id=121466&action=edit
C code to replicate the issue along with driver debug logs

In c_test_app.zip, there is a c application which replicates the issue using
only Xlib and Xrender calls.
To compile: gcc xrenderscaledimage.c -o xrenderscaledimage -lX11 -lm -lXrender

Here's what the application does:
1. Creates two 1920x1080 side-by-side colored windows(1 & 2) and maps them.
2. Creates two smaller white windows(3 & 4) of 150x150 and maps them over the
first two.
3. Creates Pictures for windows 1 and 2.
4. Creates two Pictures for an image and scales them from 300x300 to 600x600.
5. On the 'Expose' event, the image Pictures are XRenderComposite'd over the
window Pictures.
***The problem is that on the second head, the scaled image is not displayed.

If SCALE_FACTOR in xrenderscaledimage.c is set to 1.0 the problem doesn't
occur.

In c_test_app.zip, there is also intel driver debug logs for executing the code
separately on each head.  Strangely, at some point, there seems to be negative
coordinates used for the second head:
sna_blt_copy_boxes src=(0, 0) -> (-1920, 0) x 1, tiling=(0, 1), pitch=(15360,
4096)

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/intel-gfx-bugs/attachments/20160202/75d44f4d/attachment.html>