[Mesa-dev] [PATCH 0/7] gallium: catch up with mesa's system values
Roland Scheidegger
sroland at vmware.com
Fri Apr 25 07:31:51 PDT 2014
Am 24.04.2014 05:48, schrieb Ilia Mirkin:
> This adds support for gl_SampleMaskIn and gl_InvocationID system values to
> gallium, mesa/st, and nvc0. The relevant piglit tests pass, except 2
> gl_InvocationID-related ones that fail due to the linking being delayed by
> mesa/st (I think). I wrote a lame one for gl_SampleMaskIn that just makes sure
> that it's == 1 in a non-ms scenario, as nothing better seemed to be available.
>
> None of this functionality is actually available yet since it's all behind
> ARB_gpu_shader5. However you can test it out by adding
> MESA_EXTENSION_OVERRIDE=GL_ARB_gpu_shader5
>
> Note that this patchset applies on top of my ARB_sample_shading patch because
> that one adds the earlier system values.
>
> Ilia Mirkin (7):
> gallium: add INVOCATIONID semantic
> gallium: add GS_INVOCATIONS property
> mesa/st: translate gl_SampleMaskIn to SAMPLEMASK semantic
> mesa/st: translate gl_InvocationID to INVOCATIONID semantic
> nvc0/ir: add support for SAMPLEMASK sysval
> nvc0/ir: add support for INVOCATIONID system value
> nvc0/ir: set instance count based on the GS_INSTANCES property
>
> src/gallium/auxiliary/tgsi/tgsi_strings.c | 3 ++-
> src/gallium/auxiliary/tgsi/tgsi_ureg.c | 9 +++++++++
> src/gallium/docs/source/tgsi.rst | 6 ++++++
> src/gallium/drivers/nouveau/codegen/nv50_ir.h | 1 +
> src/gallium/drivers/nouveau/codegen/nv50_ir_from_tgsi.cpp | 6 +++---
> src/gallium/drivers/nouveau/codegen/nv50_ir_lowering_nvc0.cpp | 4 ++++
> src/gallium/drivers/nouveau/codegen/nv50_ir_print.cpp | 1 +
> src/gallium/drivers/nouveau/codegen/nv50_ir_target_nvc0.cpp | 1 +
> src/gallium/drivers/nouveau/nvc0/nvc0_program.c | 1 -
> src/gallium/include/pipe/p_shader_tokens.h | 6 ++++--
> src/mesa/state_tracker/st_glsl_to_tgsi.cpp | 2 ++
> 11 files changed, 33 insertions(+), 7 deletions(-)
>
Series 1-4 looks good to me too. You are right that properties aren't
all documented, consider that a bug :-).
Roland
More information about the mesa-dev
mailing list