[Mesa-dev] [PATCH 1/2] glsl: Add uniform_locations_assigned parameter to do_dead_code opt pass

[Brian Paul]

On 10/21/2011 01:23 PM, Brian Paul wrote:
> On 10/21/2011 12:49 PM, Ian Romanick wrote:
>> From: Ian Romanick<ian.d.romanick at intel.com>
>>
>> Setting this flag prevents declarations of uniforms from being removed
>> from the IR. Since the IR is directly used by several API functions
>> that query uniforms in shaders, uniform declarations cannot be removed
>> after the locations have been set. However, it should still be safe
>> to reorder the declarations (this is not tested).
>>
>> Signed-off-by: Ian Romanick<ian.d.romanick at intel.com>
>> Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=41980
>> Cc: Brian Paul<brianp at vmware.com>
>> Cc: Bryan Cain<bryancain3 at gmail.com>
>> Cc: Vinson Lee<vlee at vmware.com>
>> Cc: José Fonseca<jfonseca at vmware.com>
>> Cc: Kenneth Graunke<kenneth at whitecape.org>
>
> I applied these two patches to my tree but I'm still seeing the
> GL_INVALID_OPERATION error in glUniform() calls that I mentioned in
> the bug report (param->DataType = 0).
>
> The viewperf test I'm using has a fairly complex vertex shader with an
> array of structures of lighting info containing a "vec3 position" field.
>
> The test calls glGetUniform(shader, "LW_LightSource1[0].position").
> When glUniform3f() is called to set that uniform, the
> GL_INVALID_OPERATION errro is raised.
>
> I'll try a bisection to see where this started failing.

The problem started at commit 58a7461e1672935e7d30780a4dd40c00abbc28a5

I'm attaching a piglit shader test that exercises the bug.

-Brian

-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: glsl-vs-uniform-struct-array.shader_test
URL: <http://lists.freedesktop.org/archives/mesa-dev/attachments/20111021/faaf9eb6/attachment.asc>