[Mesa-dev] [PATCH] ir_to_mesa: fix copy propagation

[Christoph Bumiller]

On 27.01.2011 23:20, Christoph Bumiller wrote:
> On 27.01.2011 07:34, Eric Anholt wrote:
>> On Wed, 26 Jan 2011 13:30:04 +0100, Christoph Bumiller <e0425955 at student.tuwien.ac.at> wrote:
>>> The current copy propagation code would propagate TEMP[0].x from (6)
>>> into TEMP[1].x from (8) in the following, which is clearly wrong:
>>>
>>>   6:   MOV TEMP[1].x, TEMP[0].xxxx
>>>   7:   MOV TEMP[0].x, TEMP[0].yyyy
>>>   8:   MOV TEMP[0].y, TEMP[1].xxxx
>>>   9:   ADD TEMP[1].x, TEMP[0].zzzz, IMM[0].yyyy
>>>  10:   MOV TEMP[0].z, TEMP[1].xxxx
>>>
>>> Possible patch and a small program to check this attached.
>>> Tested with llvmpipe.
>> Thanks for catching this!  Could you cut it down to a nice small
>> shader_test for piglit so we can regression test?
> Here you are, a piglit test.
>
Sorry, didn't (consciously) know about the shader_runner, here's a
shader_test instead.
>
>
> _______________________________________________
> mesa-dev mailing list
> mesa-dev at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/mesa-dev

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/mesa-dev/attachments/20110127/9124f51b/attachment.htm>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: glsl-copy-propagation-bug.shader_test
URL: <http://lists.freedesktop.org/archives/mesa-dev/attachments/20110127/9124f51b/attachment.txt>