fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2026-05-05 07:28:11 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions 2

llvmpipe: fix quad group helper invocation masking

Browse source 
https://docs.vulkan.org/spec/latest/chapters/shaders.html#shaders-helper-invocations

Concretely, this fixes implicit derivatives of SSBO
variables on triangle edges, which are used in USD.

cc: mesa-stable

Reviewed-by: Konstantin Seurer <konstantin.seurer@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31047>
(cherry picked from commit 942a584db2)
This commit is contained in:
Aleksi Sapon 2024-09-20 11:36:50 -04:00 • committed by Eric Engestrom
parent 8512574ccc
commit 04f962304d
2 changed files with 36 additions and 17 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
.pick_status.json
View file

@ -344,7 +344,7 @@
"description": "llvmpipe: fix quad group helper invocation masking",
"nominated": true,
"nomination_type": 0,
"resolution": 0,
"resolution": 1,
"main_sha": null,
"because_sha": null,
"notes": null

51
src/gallium/auxiliary/gallivm/lp_bld_nir_soa.c
View file

@ -55,24 +55,45 @@ static int bit_size_to_shift_size(int bit_size)
}
/*
* combine the execution mask if there is one with the current mask.
* Combine the global mask if there is one with the current execution mask.
*/
static LLVMValueRef
mask_vec(struct lp_build_nir_context *bld_base)
{
struct lp_build_nir_soa_context * bld = (struct lp_build_nir_soa_context *)bld_base;
LLVMBuilderRef builder = bld->bld_base.base.gallivm->builder;
struct lp_build_nir_soa_context *bld = (struct lp_build_nir_soa_context *)bld_base;
struct lp_exec_mask *exec_mask = &bld->exec_mask;
LLVMValueRef bld_mask = bld->mask ? lp_build_mask_value(bld->mask) : NULL;
if (!exec_mask->has_mask) {
if (!exec_mask->has_mask)
return bld_mask;
}
if (!bld_mask)
return exec_mask->exec_mask;
LLVMBuilderRef builder = bld->bld_base.base.gallivm->builder;
return LLVMBuildAnd(builder, lp_build_mask_value(bld->mask),
exec_mask->exec_mask, "");
}
/*
* Use the execution mask if there is one, otherwise don't mask (ignore global mask).
* This allows helper invocations to run, which are necessary for correct derivatives.
*/
static LLVMValueRef
mask_vec_with_helpers(struct lp_build_nir_context *bld_base)
{
if (bld_base->shader->info.stage != MESA_SHADER_FRAGMENT)
return mask_vec(bld_base); /* No helper invocations needed. */
struct lp_build_nir_soa_context *bld = (struct lp_build_nir_soa_context *)bld_base;
struct lp_exec_mask *exec_mask = &bld->exec_mask;
if (exec_mask->has_mask)
return exec_mask->exec_mask;
return lp_build_const_int_vec(bld_base->base.gallivm,
bld_base->uint_bld.type, -1);
}
static bool
invocation_0_must_be_active(struct lp_build_nir_context *bld_base)
{
@ -1032,7 +1053,6 @@ static void emit_load_global(struct lp_build_nir_context *bld_base,
LLVMBuilderRef builder = gallivm->builder;
struct lp_build_context *uint_bld = &bld_base->uint_bld;
struct lp_build_context *res_bld;
LLVMValueRef exec_mask = mask_vec(bld_base);
res_bld = get_int_bld(bld_base, true, bit_size);
@ -1055,14 +1075,13 @@ static void emit_load_global(struct lp_build_nir_context *bld_base,
return;
}
LLVMValueRef mask = mask_vec_with_helpers(bld_base);
for (unsigned c = 0; c < nc; c++) {
LLVMValueRef chan_offset = lp_build_const_int_vec(gallivm, uint_bld->type, c * (bit_size / 8));
outval[c] = lp_build_masked_gather(gallivm, res_bld->type.length,
bit_size,
res_bld->vec_type,
outval[c] = lp_build_masked_gather(gallivm, res_bld->type.length, bit_size, res_bld->vec_type,
lp_vec_add_offset_ptr(bld_base, bit_size, addr, chan_offset),
exec_mask);
mask);
outval[c] = LLVMBuildBitCast(builder, outval[c], res_bld->vec_type, "");
}
}
@ -1431,14 +1450,14 @@ static void emit_load_mem(struct lp_build_nir_context *bld_base,
for (unsigned c = 0; c < nc; c++)
result[c] = lp_build_alloca(gallivm, load_bld->vec_type, "");
LLVMValueRef exec_mask = mask_vec(bld_base);
LLVMValueRef cond = LLVMBuildICmp(gallivm->builder, LLVMIntNE, exec_mask, uint_bld->zero, "");
LLVMValueRef gather_mask = mask_vec_with_helpers(bld_base);
LLVMValueRef gather_cond = LLVMBuildICmp(gallivm->builder, LLVMIntNE, gather_mask, uint_bld->zero, "");
for (unsigned i = 0; i < uint_bld->type.length; i++) {
LLVMValueRef counter = lp_build_const_int32(gallivm, i);
LLVMValueRef loop_cond = LLVMBuildExtractElement(gallivm->builder, cond, counter, "");
LLVMValueRef element_gather_cond = LLVMBuildExtractElement(gallivm->builder, gather_cond, counter, "");
struct lp_build_if_state exec_ifthen;
lp_build_if(&exec_ifthen, gallivm, loop_cond);
struct lp_build_if_state if_gather_element;
lp_build_if(&if_gather_element, gallivm, element_gather_cond);
LLVMValueRef ssbo_limit;
LLVMValueRef mem_ptr = mem_access_base_pointer(bld_base, load_bld, bit_size, payload, index,
@ -1473,7 +1492,7 @@ static void emit_load_mem(struct lp_build_nir_context *bld_base,
lp_build_endif(&ifthen);
}
lp_build_endif(&exec_ifthen);
lp_build_endif(&if_gather_element);
}
for (unsigned c = 0; c < nc; c++)
outval[c] = LLVMBuildLoad2(gallivm->builder, load_bld->vec_type, result[c], "");