[Mesa-dev] [PATCH v6 2/2] anv: enable VK_EXT_shader_stencil_export
Jason Ekstrand
jason at jlekstrand.net
Fri May 11 20:24:47 UTC 2018
On Fri, May 11, 2018 at 12:15 PM, Caio Marcelo de Oliveira Filho <
caio.oliveira at intel.com> wrote:
> From: Gustavo Lima Chaves <gustavo.lima.chaves at intel.com>
>
> v2:
> An attempt to support SpvExecutionModeStencilRefReplacingEXT's behavior
> also follows, with the interpretation to said mode being we prevent
> writes to the built-in FragStencilRefEXT variable when the execution
> mode isn't set.
>
> v3:
> A more cautious reading of 1db44252d01bf7539452ccc2b5210c74b8dcd573 led
> me to a missing change that would stop (what I later discovered were)
> GPU hangs on the CTS test written to exercize this.
>
> v4:
> Turn FragStencilRefEXT decoration usage without StencilRefReplacingEXT
> mode into a warning, instead of trying to make the variable read-only.
> If we are to follow the originating extension on GL, the built-in
> variable in question should never be readable anyway.
> ---
>
> This is the original patch from Gustavo Lima Chaves (v5 was another
> rebase), with some trivial changes to rebase.
>
> I'm updating the test on Vulkan CTS to make sure we have some coverage
> of this feature; so this should only land once that gets accepted, but
> is ready for review.
>
> src/compiler/shader_info.h | 2 ++
> src/compiler/spirv/spirv_to_nir.c | 4 ++++
> src/compiler/spirv/vtn_variables.c | 5 +++++
> src/intel/vulkan/anv_extensions.py | 1 +
> src/intel/vulkan/anv_pipeline.c | 1 +
> src/intel/vulkan/genX_pipeline.c | 1 +
> 6 files changed, 14 insertions(+)
>
> diff --git a/src/compiler/shader_info.h b/src/compiler/shader_info.h
> index 81f844d36ae..e9141261e8b 100644
> --- a/src/compiler/shader_info.h
> +++ b/src/compiler/shader_info.h
> @@ -180,6 +180,8 @@ typedef struct shader_info {
>
> bool pixel_center_integer;
>
> + bool outputs_stencil;
>
This isn't used by anything outside of spirv_to_nir so I don't think it
makes sense in shader_info. If user wants to know if the output is
written, they can look at outputs_written. If you want to have the warning
below (which seems good to me), maybe a boolean in vtn_builder instead?
> +
> /** gl_FragDepth layout for ARB_conservative_depth. */
> enum gl_frag_depth_layout depth_layout;
> } fs;
> diff --git a/src/compiler/spirv/spirv_to_nir.c
> b/src/compiler/spirv/spirv_to_nir.c
> index 6c0551603ea..81c5e890071 100644
> --- a/src/compiler/spirv/spirv_to_nir.c
> +++ b/src/compiler/spirv/spirv_to_nir.c
> @@ -3577,6 +3577,10 @@ vtn_handle_execution_mode(struct vtn_builder *b,
> struct vtn_value *entry_point,
> case SpvExecutionModeContractionOff:
> break; /* OpenCL */
>
> + case SpvExecutionModeStencilRefReplacingEXT:
> + b->shader->info.fs.outputs_stencil = true;
> + break;
> +
> default:
> vtn_fail("Unhandled execution mode");
> }
> diff --git a/src/compiler/spirv/vtn_variables.c b/src/compiler/spirv/vtn_
> variables.c
> index 53bee1b9288..0a3108bdc43 100644
> --- a/src/compiler/spirv/vtn_variables.c
> +++ b/src/compiler/spirv/vtn_variables.c
> @@ -1429,6 +1429,11 @@ apply_var_decoration(struct vtn_builder *b,
> nir_variable *nir_var,
> case SpvBuiltInSamplePosition:
> nir_var->data.origin_upper_left = b->origin_upper_left;
> break;
> + case SpvBuiltInFragStencilRefEXT:
> + if (!b->shader->info.fs.outputs_stencil)
> + vtn_warn("The StencilRefReplacingEXT mode should be declared
> when"
> + " the decoration FragStencilRefEXT is used on a
> variable");
> + break;
>
This stuff should go in the previous patch.
> default:
> break;
> }
> diff --git a/src/intel/vulkan/anv_extensions.py b/src/intel/vulkan/anv_
> extensions.py
> index b5bee0881ce..8160864685f 100644
> --- a/src/intel/vulkan/anv_extensions.py
> +++ b/src/intel/vulkan/anv_extensions.py
> @@ -112,6 +112,7 @@ EXTENSIONS = [
> Extension('VK_EXT_global_priority', 1,
> 'device->has_context_priority'),
> Extension('VK_EXT_shader_viewport_index_layer', 1, True),
> + Extension('VK_EXT_shader_stencil_export', 1,
> 'device->info.gen >= 9'),
> ]
>
> class VkVersion:
> diff --git a/src/intel/vulkan/anv_pipeline.c b/src/intel/vulkan/anv_
> pipeline.c
> index 8f30136b100..c37b9b96e11 100644
> --- a/src/intel/vulkan/anv_pipeline.c
> +++ b/src/intel/vulkan/anv_pipeline.c
> @@ -152,6 +152,7 @@ anv_shader_compile_to_nir(struct anv_pipeline
> *pipeline,
> .subgroup_quad = true,
> .subgroup_shuffle = true,
> .subgroup_vote = true,
> + .stencil_export = true,
> },
> };
>
> diff --git a/src/intel/vulkan/genX_pipeline.c b/src/intel/vulkan/genX_
> pipeline.c
> index 6016d257584..462c59451cc 100644
> --- a/src/intel/vulkan/genX_pipeline.c
> +++ b/src/intel/vulkan/genX_pipeline.c
> @@ -1600,6 +1600,7 @@ emit_3dstate_ps_extra(struct anv_pipeline *pipeline,
> ps.PixelShaderHasUAV = true;
>
> #if GEN_GEN >= 9
> + ps.PixelShaderComputesStencil = wm_prog_data->computed_stencil;
> ps.PixelShaderPullsBary = wm_prog_data->pulls_bary;
> ps.InputCoverageMaskState = wm_prog_data->uses_sample_mask ?
> ICMS_INNER_CONSERVATIVE : ICMS_NONE;
> --
> 2.17.0
>
> _______________________________________________
> mesa-dev mailing list
> mesa-dev at lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/mesa-dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/mesa-dev/attachments/20180511/e20954ca/attachment.html>
More information about the mesa-dev
mailing list