[Mesa-dev] [PATCH 1/4] i965: Also examine _BaseFormat when deciding to perform xRGB_alpha fixups

[Ian Romanick]

On 01/22/2013 05:54 PM, Carl Worth wrote:
> Carl Worth <cworth at cworth.org> writes:
>> Note: The test results mentioned here, (and in the following patches), expect
>> the fixes I recently submitted to the fbo-* tests in piglit. That series
>> consists of 5 patches starting with:
>
> While developing this patch series, I did most of my testing with the
> fbo-* tests mentioned above. I had meant to do a more-inclusive run of
> piglit tests before sending this series out, but I forgot to do that.
>
> I've just completed that testing and found the following two changes
> which are not mentioned in my emails for this patch series:
>
> 1. shaders/glsl-fs-user-varying-if
>
> 	This test changes from FAIL to PASS with my patch series. The
> 	output says:
>
> 		"Probe at (.., ..)" returning mismatched results is expected and correct.
> 		Probe at (15,15)
> 		  Expected: 0.000000 1.000000 0.000000
> 		  Observed: 0.000000 0.000000 1.000000
>
> 	I'm not familiar with this test, so without looking closer at
> 	it, I don't know what the above means nor whether the PASS
> 	result is legitimate based on my changes.
>
> 2. EXT_texture_integer/texture_integer_glsl130
>
> 	This test changes from PASS to FAIL. The output is:
>
> 		Errors:
> 		texture-integer: failure with format GL_RG8I:
> 		  texture color = 71, 55, 112, 0
> 		  expected color = 0.25, 0.5, 0, 0
> 		  result color = 0.25098, 0.501961, 0, 1
>
> 	Again, I haven't looked at this test closely yet, but here, at
> 	least, the new results do seem consistent with my patch series,
> 	(alpha changing from 0 to 1 for an RG format). Perhaps the test
> 	should be updated?
>
> Any feedback from anyone with more knowledge of these two tests would be
> appreciated.

These tests intermittently pass / fail for me for quite some time (since 
always?).