nir_intrinsic_memory_barrier has the same semantic as memoryBarrier()
in GLSL, which is:
GLSL 4.60, 4.10. "Memory Qualifiers":
"The built-in function memoryBarrier() can be used if needed to
guarantee the completion and relative ordering of memory accesses
performed by a single shader invocation."
GLSL 4.60, 8.17. "Shader Memory Control Functions":
"The built-in functions memoryBarrier() and groupMemoryBarrier() wait
for the completion of accesses to all of the above variable types."
Fixes tests:
dEQP-VK.memory_model.message_passing.core11.u32.coherent.fence_fence.atomicwrite.device.payload_local.image.guard_nonlocal.workgroup.comp
dEQP-VK.memory_model.message_passing.core11.u32.coherent.fence_fence.atomicwrite.device.payload_nonlocal.workgroup.guard_local.image.comp
Fixes:
819a613a ("freedreno/ir3: moar better scheduler")
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Reviewed-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9054>
dEQP-VK.image.subresource_layout.3d.all_levels.r8_snorm,Fail
dEQP-VK.image.subresource_layout.3d.all_levels.r8g8b8a8_snorm,Fail
dEQP-VK.info.device_mandatory_features,Fail
-dEQP-VK.memory_model.message_passing.core11.u32.coherent.fence_fence.atomicwrite.device.payload_local.image.guard_nonlocal.workgroup.comp,Fail
-dEQP-VK.memory_model.message_passing.core11.u32.coherent.fence_fence.atomicwrite.device.payload_nonlocal.workgroup.guard_local.image.comp,Fail
dEQP-VK.pipeline.framebuffer_attachment.diff_attachments_2d_19x27_32x32_ms,Fail
dEQP-VK.pipeline.push_descriptor.compute.binding0_numcalls2_combined_image_sampler,Crash
dEQP-VK.pipeline.push_descriptor.compute.binding0_numcalls2_sampled_image,Crash
* shared image atomic SSBO everything
* barrier()/ - R/W R/W R/W R/W X
* groupMemoryBarrier()
- * memoryBarrier() - R/W R/W
+ * memoryBarrier()
* (but only images declared coherent?)
* memoryBarrierAtomic() - R/W
* memoryBarrierBuffer() - R/W
barrier->flags = IR3_INSTR_SS | IR3_INSTR_SY;
barrier->barrier_class = IR3_BARRIER_EVERYTHING;
break;
- case nir_intrinsic_memory_barrier:
- barrier = ir3_FENCE(b);
- barrier->cat7.g = true;
- barrier->cat7.r = true;
- barrier->cat7.w = true;
- barrier->cat7.l = true;
- barrier->barrier_class = IR3_BARRIER_IMAGE_W |
- IR3_BARRIER_BUFFER_W;
- barrier->barrier_conflict =
- IR3_BARRIER_IMAGE_R | IR3_BARRIER_IMAGE_W |
- IR3_BARRIER_BUFFER_R | IR3_BARRIER_BUFFER_W;
- break;
case nir_intrinsic_memory_barrier_buffer:
barrier = ir3_FENCE(b);
barrier->cat7.g = true;
barrier->barrier_conflict = IR3_BARRIER_SHARED_R |
IR3_BARRIER_SHARED_W;
break;
+ case nir_intrinsic_memory_barrier:
case nir_intrinsic_group_memory_barrier:
barrier = ir3_FENCE(b);
barrier->cat7.g = true;