[Bug 91926] [SKL bisected] texsubimage pbo intermittent failures

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=91926

--- Comment #40 from Topi Pohjolainen <topi.pohjolainen at intel.com> ---
When I started looking at the bug I was uneasy with a another error message
coming from mesa although this didn't cause the test to fail. Namely:

Mesa: User error: GL_INVALID_OPERATION in glTexSubImage2D(out of bounds PBO
access)

This comes from the second update round when the pixel buffer object is used as
the source for the rectangle to be update in the texture image.

The test is written to try 10 different regions with random width, height and
depth. The error message comes from mesa when one tries to upload with texture
depth zero. This results into no upload as mesa bails out. The reason there is
no piglit failure is because the piglit test _also_ skips to check any pixels
as
the test iterates over the depth. I left this to be fixed later on as I thought
it is separate from the case where we really check pixels and find faulty
square.

However, just out of curiosity I modified the piglit test to skip the failing
attempt to upload when one of the random dimensions (tw, th, td) is zero. With
this change in the test I'm not able to reproduce the failure anymore.

I'll check next if this just a coincidence or if it is somehow connected to the
failure in pixel values.

-- 
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-3d-bugs/attachments/20151023/2b9817d9/attachment.html>