fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2026-05-03 20:48:08 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions 2

freedreno/a6xx: Better program state size calc

Browse source 
Most of the time we were significantly over-allocating the size of
program stateobjs.  Except when the shader had a very large # of
immediates, in which case we were under-allocating (and crashing).

Fixes: 598928d7e7 ("nir/loop_analyze: determine whether all control flow gets eliminated upon loop unrolling")
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/14731
Signed-off-by: Rob Clark <rob.clark@oss.qualcomm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/39545>
(cherry picked from commit 670ded35c1)
This commit is contained in:
Rob Clark 2026-01-26 14:19:18 -08:00 • committed by Eric Engestrom
parent 3ef0b4b27a
commit cda3f42323
2 changed files with 47 additions and 3 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
.pick_status.json
View file

@ -1364,7 +1364,7 @@
"description": "freedreno/a6xx: Better program state size calc",
"nominated": true,
"nomination_type": 2,
"resolution": 0,
"resolution": 1,
"main_sha": null,
"because_sha": "598928d7e75577832056826b205a34aeeffcc363",
"notes": null

48
src/gallium/drivers/freedreno/a6xx/fd6_program.cc
View file

@ -1456,6 +1456,46 @@ emit_interp_state(fd_crb &crb, const struct fd6_program_state *state,
crb.add(VPC_VARYING_REPLACE_MODE_MODE(CHIP, i, vpsrepl[i]));
}
static unsigned
xs_immediate_size_dwords(const struct ir3_shader_variant *v)
{
const struct ir3_const_state *const_state = ir3_const_state(v);
uint32_t base = const_state->allocs.max_const_offset_vec4;
const struct ir3_imm_const_state *imm_state = &v->imm_state;
int32_t size = DIV_ROUND_UP(imm_state->count, 4);
/* truncate size to avoid writing constants that shader
* does not use:
*/
size = MIN2(size + base, v->constlen) - base;
return MAX2(size, 0) * 4;
}
static unsigned
xs_size(const struct ir3_shader_variant *v)
{
if (!v)
return 0;
const struct ir3_const_state *const_state = ir3_const_state(v);
unsigned size = 128;
size += 64 + (v->stream_output.num_outputs != 0 ? 256 : 0);
/* Variable number of UBO upload ranges. */
size += 4 * const_state->ubo_state.num_enabled;
/* Variable number of dwords for the primitive map */
size += v->input_size;
/* Immediates: */
size += xs_immediate_size_dwords(v);
return size * 4;
}
template <chip CHIP>
static struct ir3_program_state *
fd6_program_create(void *data, const struct ir3_shader_variant *bs,
@ -1482,8 +1522,12 @@ fd6_program_create(void *data, const struct ir3_shader_variant *bs,
state->ds = ds;
state->gs = gs;
state->fs = fs;
state->binning_stateobj = fd_ringbuffer_new_object(ctx->pipe, 0x1000);
state->stateobj = fd_ringbuffer_new_object(ctx->pipe, 0x1000);
unsigned stateobj_size = xs_size(vs) + xs_size(hs) +
xs_size(ds) + xs_size(gs) + xs_size(fs);
state->binning_stateobj = fd_ringbuffer_new_object(ctx->pipe, stateobj_size);
state->stateobj = fd_ringbuffer_new_object(ctx->pipe, stateobj_size);
if (hs) {
/* Allocate the fixed-size tess factor BO globally on the screen. This