Created attachment 34029 [details]
misrendered GLSL testcase
The attached testcase doesn't render correctly on a GM45.
If you are here, please have a look at https://bugs.freedesktop.org/show_bug.cgi?id=26691 too.
Created attachment 34030 [details]
updated test case
The previous test-case forced SW rendering which hides the bug by rendering correctly.
Author: Eric Anholt <email@example.com>
Date: Wed Mar 10 11:00:40 2010 -0800
i965: Add support for the CMP opcode in the GLSL path.
This would be triggered by use of sqrt() along with control flow.
Fixes piglit-fs-sqrt-branch and a bug in Yo Frankie!.
(In reply to comment #3)
> commit 48dca99feb394febc3af44e14f23fb12a9cc9204
> Author: Eric Anholt <firstname.lastname@example.org>
> Date: Wed Mar 10 11:00:40 2010 -0800
> i965: Add support for the CMP opcode in the GLSL path.
> This would be triggered by use of sqrt() along with control flow.
> Fixes piglit-fs-sqrt-branch and a bug in Yo Frankie!.
I have re-run the test case with today's Mesa master 8df65e98998b4c104db30cbba8a38be7eb2a9acd (including the above referred patch) and drm master c1c8bbf80b1f734e23996bf805dc78f32ebaf56f and the misrendering still exists. The warning has gone away for the sqrt instruction though. Please run the test case with LIBGL_ALWAYS_SOFTWARE=1 ./fsraytrace to what the normal rendering should be.
This is fixed with the merge of GLSL2.
I can confirm, it is working on GM45 with upstream mesa. Nice work! Thanks!