[Piglit] [PATCH] glsl-deriv-varyings: Fixes compilation warnings and adds extra regular subtest

Andres Gomez agomez at igalia.com
Mon Feb 23 06:02:13 PST 2015


Hi,

sorry for the delay in answering I've not been able to devote much time
for this lately.

On Thu, 2015-02-19 at 13:25 -0800, Matt Turner wrote:
> On Thu, Feb 19, 2015 at 10:39 AM, Ian Romanick <idr at freedesktop.org> wrote:
> > So... where did we end up with this?  I just pushed a patch from the bug
> > that makes shaders 4 and 5 actually run.  Both fail, but Matt thought
> > that the fix to Mesa was committed.
> 
> Indeed, it's upstream as commit 8517e665.

When introducing the patch in commit 8517e665 we were fixing fdo#82991.

As there was not a piglit test for that case, in order to track
regressions I extended the piglit "glsl-deriv-varyings" test with 3 more
cases.

The first one of those 3 was failing before mesa commit 8517e665 and is
passing afterwards.

The other 2 cases, however, fail before and after that commit. That
means that, while extending "glsl-deriv-varyings" I found, AFAIU, 2 new
bugs in mesa. I filed fdo#87407 to track those.

However, I've tested those 2 new cases with swrast, nouveau, i965 and
the proprietary NVIDIA driver in GNU/Linux and all of them fail
consistently with the same results. Therefore, I'm not totally sure the
expected results I added for those 2 extra cases are correct and I will
welcome another pair of eyes.

If the expected results are not correct, they should be corrected in the
piglit test and, probably, fdo#87407 should be invalidated.

Other than that, I agree with Ian that this test should be converted to
shader_runner. I will try to find some time to do this.

Hope this clarifies the status.

-- 
Br,

Andres



More information about the Piglit mailing list