[Bug 108805] i965 regressions from EXT_texture_sRGB_R8

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=108805

--- Comment #2 from Gert Wollny <gw.fossdev at gmail.com> ---
I tracked down the KHR failures to the patch that changes the error state from
FRAMEBUFFER_UNSUPPORTED to more specific values. For me these tests switch from
"Unsupported (unsupported framebuffer configuration)" to "Internal error
(Error)". 

This is a bit surprising to me because a change in the error number should not
result in a crash, so I digged into the CTS to see what it does: 

external/openglcts/modules/gl/gl4cDirectStateAccessTexturesTests.cpp: 

if (gl.checkFramebufferStatus(GL_FRAMEBUFFER) != GL_FRAMEBUFFER_COMPLETE)
{
if (gl.checkFramebufferStatus(GL_FRAMEBUFFER) == GL_FRAMEBUFFER_UNSUPPORTED)
    throw tcu::NotSupportedError("unsupported framebuffer configuration");
else
    throw 0;
}

Given that RGB formats are not required to be renderable, the error state 
GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT should also be legal, and indeed adding a
check for this latter return value turns the output from "Fail" into
"Unsupported", so this actually seems to be a bug in the CTS.

Regarding the piglit, it says "unknown argument -alpha-one", and then it
passes. 

The other failures on the other hand are probably due to a lax use of the
extension flags (which is indeed my fault), and there is a patch series that
addresses these [1], especially 24/30 should take care of this specific error
[2]. 

[1] https://patchwork.freedesktop.org/series/52689/
[2] https://patchwork.freedesktop.org/patch/262637/

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/intel-3d-bugs/attachments/20181120/18a2829f/attachment.html>