fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2026-04-21 14:10:37 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions 3

nir/lower_mem_access_bit_sizes: add partial support for offset_shift

Browse source 
Note: this was implemented and tested for ir3. The code paths that are
never used there [1] seem non-trivial to implement. Since they cannot be
easily tested, asserts and TODOs are added to ensure we don't
accidentally hit them for intrinsics with offset_shift.

[1]: these paths are never used on ir3 since lower_mem_access_bit_sizes
is only used for SSBO accesses to lower 64b accesses (which are 64b
aligned) to 32b ones. So we'll never request an increase of alignment.

Signed-off-by: Job Noorman <jnoorman@igalia.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35092>
This commit is contained in:
Job Noorman 2025-08-19 10:41:46 +02:00 • committed by Marge Bot
parent cd72d8e366
commit 7fe484e373
1 changed files with 28 additions and 7 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

35
src/compiler/nir/nir_lower_mem_access_bit_sizes.c
View file

@ -28,7 +28,7 @@
static nir_intrinsic_instr *
dup_mem_intrinsic(nir_builder *b, nir_intrinsic_instr *intrin,
nir_def *offset,
nir_io_offset offset,
unsigned align_mul, unsigned align_offset,
nir_def *data,
unsigned num_components, unsigned bit_size)
@ -45,7 +45,7 @@ dup_mem_intrinsic(nir_builder *b, nir_intrinsic_instr *intrin,
assert(&intrin->src[i] != intrin_offset_src);
dup->src[i] = nir_src_for_ssa(data);
} else if (&intrin->src[i] == intrin_offset_src) {
dup->src[i] = nir_src_for_ssa(offset);
/* Handled by nir_set_io_offset below. */
} else {
dup->src[i] = nir_src_for_ssa(intrin->src[i].ssa);
}
@ -55,6 +55,7 @@ dup_mem_intrinsic(nir_builder *b, nir_intrinsic_instr *intrin,
for (unsigned i = 0; i < info->num_indices; i++)
dup->const_index[i] = intrin->const_index[i];
nir_set_io_offset(dup, offset);
nir_intrinsic_set_align(dup, align_mul, align_offset);
if (info->has_dest) {
@ -208,7 +209,14 @@ lower_mem_load(nir_builder *b, nir_intrinsic_instr *intrin,
uint64_t align_mask = requested.align - 1;
nir_def *chunk_offset = nir_iadd_imm(b, offset, chunk_start);
nir_def *aligned_offset = nir_iand_imm(b, chunk_offset, ~align_mask);
/* TODO add support for offset_shift. */
assert(!nir_intrinsic_has_offset_shift(intrin) ||
nir_intrinsic_offset_shift(intrin) == 0);
nir_io_offset aligned_offset = (nir_io_offset){
.def = nir_iand_imm(b, chunk_offset, ~align_mask),
.shift = 0,
};
nir_intrinsic_instr *load =
dup_mem_intrinsic(b, intrin, aligned_offset,
@ -238,8 +246,15 @@ lower_mem_load(nir_builder *b, nir_intrinsic_instr *intrin,
} else if (chunk_align_offset % requested.align) {
/* In this case, we know how much to adjust the offset */
uint32_t delta = chunk_align_offset % requested.align;
nir_def *load_offset =
nir_iadd_imm(b, offset, (int64_t)chunk_start - (int64_t)delta);
/* TODO add support for offset_shift. */
assert(!nir_intrinsic_has_offset_shift(intrin) ||
nir_intrinsic_offset_shift(intrin) == 0);
nir_io_offset load_offset = (nir_io_offset){
.def =
nir_iadd_imm(b, offset, (int64_t)chunk_start - (int64_t)delta),
.shift = 0,
};
const uint32_t load_align_offset =
(chunk_align_offset - delta) % align_mul;
@ -269,7 +284,8 @@ lower_mem_load(nir_builder *b, nir_intrinsic_instr *intrin,
1, chunk_bit_size);
}
} else {
nir_def *chunk_offset = nir_iadd_imm(b, offset, chunk_start);
nir_io_offset chunk_offset =
nir_io_offset_iadd(b, intrin, chunk_start);
nir_intrinsic_instr *load =
dup_mem_intrinsic(b, intrin, chunk_offset,
align_mul, chunk_align_offset, NULL,
@ -381,6 +397,10 @@ lower_mem_store(nir_builder *b, nir_intrinsic_instr *intrin,
};
uint64_t align_mask = requested.align - 1;
/* TODO add support for offset_shift. */
assert(!nir_intrinsic_has_offset_shift(intrin) ||
nir_intrinsic_offset_shift(intrin) == 0);
nir_def *chunk_offset = nir_iadd_imm(b, offset, chunk_start);
nir_def *pad = chunk_align < 4 ? nir_iand_imm(b, chunk_offset, align_mask) : nir_imm_intN_t(b, 0, chunk_offset->bit_size);
chunk_offset = nir_iand_imm(b, chunk_offset, ~align_mask);
@ -449,7 +469,8 @@ lower_mem_store(nir_builder *b, nir_intrinsic_instr *intrin,
requested.num_components,
requested.bit_size);
nir_def *chunk_offset = nir_iadd_imm(b, offset, chunk_start);
nir_io_offset chunk_offset =
nir_io_offset_iadd(b, intrin, chunk_start);
dup_mem_intrinsic(b, intrin, chunk_offset,
align_mul, chunk_align_offset, packed,
requested.num_components, requested.bit_size);