dsg/mesa
2
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

radv/rt: only run move_rt_instructions() for CPS shaders

Browse source 
move_rt_instructions() only makes sense for CPS recursive shaders, where
later rt_trace_ray calls can overwrite the current shader's RT system
values.

Running it on the function-call path can hoist load_hit_attrib_amd
above merged intersection writes, which corrupts any-hit
hitAttributeEXT. Move the pass into the existing CPS-only
non-intersection branch before nir_lower_shader_calls().

Fixes: c5d796c902 ("radv/rt: Use function call structure in NIR lowering")
Closes: #15074

Reviewed-by: Konstantin Seurer <konstantin.seurer@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40531>
This commit is contained in:
kingstom.chen 2026-03-20 16:38:51 +08:00 • committed by Marge Bot
parent c60e5df798
commit 5a7f4c62d8
1 changed files with 5 additions and 5 deletions
Download patch file Download diff file Expand all files Collapse all files

10
src/amd/vulkan/radv_pipeline_rt.c
View file

@ -411,16 +411,16 @@ radv_rt_nir_to_asm(struct radv_device *device, struct vk_pipeline_cache *cache,
stage->info.inline_push_constant_mask = stage->args.ac.inline_push_const_mask;
stage->info.type = radv_is_traversal_shader(stage->nir) ? RADV_SHADER_TYPE_RT_TRAVERSAL : RADV_SHADER_TYPE_DEFAULT;
/* Move ray tracing system values to the top that are set by rt_trace_ray
* to prevent them from being overwritten by other rt_trace_ray calls.
*/
NIR_PASS(_, stage->nir, move_rt_instructions);
uint32_t num_resume_shaders = 0;
nir_shader **resume_shaders = NULL;
void *mem_ctx = ralloc_context(NULL);
if (stage->stage != MESA_SHADER_INTERSECTION && mode == RADV_RT_LOWERING_MODE_CPS) {
/* Move ray tracing system values to the top that are set by rt_trace_ray
* to prevent them from being overwritten by other rt_trace_ray calls.
*/
NIR_PASS(_, stage->nir, move_rt_instructions);
nir_builder b = nir_builder_at(nir_after_impl(nir_shader_get_entrypoint(stage->nir)));
nir_rt_return_amd(&b);