While we're here, we also fix up a couple of potential ralloc-parenting bugs. In particular, we were calling nir_src_copy with the shader as the mem context instead of the instruction. --- src/glsl/nir/nir_lower_vec_to_movs.c | 22 ++++++++++++++++------ 1 file changed, 16 insertions(+), 6 deletions(-)
diff --git a/src/glsl/nir/nir_lower_vec_to_movs.c b/src/glsl/nir/nir_lower_vec_to_movs.c index b7f096d..e297cb8 100644 --- a/src/glsl/nir/nir_lower_vec_to_movs.c +++ b/src/glsl/nir/nir_lower_vec_to_movs.c @@ -32,6 +32,10 @@ * moves with partial writes. */ +struct vec_to_movs_state { + nir_shader *shader; +}; + static bool src_matches_dest_reg(nir_dest *dest, nir_src *src) { @@ -54,12 +58,12 @@ src_matches_dest_reg(nir_dest *dest, nir_src *src) */ static unsigned insert_mov(nir_alu_instr *vec, unsigned start_channel, - unsigned start_src_idx, void *mem_ctx) + unsigned start_src_idx, nir_shader *shader) { unsigned src_idx = start_src_idx; assert(src_idx < nir_op_infos[vec->op].num_inputs); - nir_alu_instr *mov = nir_alu_instr_create(mem_ctx, nir_op_imov); + nir_alu_instr *mov = nir_alu_instr_create(shader, nir_op_imov); nir_alu_src_copy(&mov->src[0], &vec->src[src_idx], mov); nir_alu_dest_copy(&mov->dest, &vec->dest, mov); @@ -84,8 +88,10 @@ insert_mov(nir_alu_instr *vec, unsigned start_channel, } static bool -lower_vec_to_movs_block(nir_block *block, void *mem_ctx) +lower_vec_to_movs_block(nir_block *block, void *void_state) { + struct vec_to_movs_state *state = void_state; + nir_foreach_instr_safe(block, instr) { if (instr->type != nir_instr_type_alu) continue; @@ -115,7 +121,7 @@ lower_vec_to_movs_block(nir_block *block, void *mem_ctx) continue; if (src_matches_dest_reg(&vec->dest.dest, &vec->src[src_idx].src)) { - finished_write_mask |= insert_mov(vec, i, src_idx, mem_ctx); + finished_write_mask |= insert_mov(vec, i, src_idx, state->shader); break; } src_idx++; @@ -127,7 +133,7 @@ lower_vec_to_movs_block(nir_block *block, void *mem_ctx) continue; if (!(finished_write_mask & (1 << i))) - finished_write_mask |= insert_mov(vec, i, src_idx, mem_ctx); + finished_write_mask |= insert_mov(vec, i, src_idx, state->shader); src_idx++; } @@ -142,7 +148,11 @@ lower_vec_to_movs_block(nir_block *block, void *mem_ctx) static void nir_lower_vec_to_movs_impl(nir_function_impl *impl) { - nir_foreach_block(impl, lower_vec_to_movs_block, ralloc_parent(impl)); + struct vec_to_movs_state state; + + state.shader = impl->overload->function->shader; + + nir_foreach_block(impl, lower_vec_to_movs_block, &state); } void -- 2.5.0.400.gff86faf _______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev