v3d: Fix predication with atomic image operations

Fixes dEQP test:
dEQP-GLES31.functional.synchronization.inter_call.with_memory_barrier.image_atomic_multiple_interleaved_write_read

Fixes piglit test:
spec/glsl-es-3.10/execution/cs-image-atomic-if-else.shader_test

Fixes: 6281f26f064ada ("v3d: Add support for shader_image_load_store.")
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Reviewed-by: Eric Anholt <eric@anholt.net>

diff --git a/src/broadcom/compiler/v3d40_tex.c b/src/broadcom/compiler/v3d40_tex.c
index 287116381fb8f8eb6456eab57bf9a6aa7b35b33c..379f90c146a27f465189c55fb9561b71bb6cd9e8 100644 (file)
--- a/src/broadcom/compiler/v3d40_tex.c
+++ b/src/broadcom/compiler/v3d40_tex.c
@@ -392,9 +392,21 @@ v3d40_vir_emit_image_load_store(struct v3d_compile *c,
                 }
         }
 
+        if (vir_in_nonuniform_control_flow(c) &&
+            instr->intrinsic != nir_intrinsic_image_deref_load) {
+           vir_set_pf(vir_MOV_dest(c, vir_nop_reg(), c->execute),
+                      V3D_QPU_PF_PUSHZ);
+        }
+
         vir_TMU_WRITE(c, V3D_QPU_WADDR_TMUSF, ntq_get_src(c, instr->src[1], 0),
                       &tmu_writes);
 
+        if (vir_in_nonuniform_control_flow(c) &&
+            instr->intrinsic != nir_intrinsic_image_deref_load) {
+           struct qinst *last_inst= (struct  qinst *)c->cur_block->instructions.prev;
+           vir_set_cond(last_inst, V3D_QPU_COND_IFA);
+        }
+
         vir_emit_thrsw(c);
 
         /* The input FIFO has 16 slots across all threads, so make sure we