/* * Copyright © 2013 Intel Corporation * Copyright © 2023 Valve Corporation * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and associated documentation files (the "Software"), * to deal in the Software without restriction, including without limitation * the rights to use, copy, modify, merge, publish, distribute, sublicense, * and/or sell copies of the Software, and to permit persons to whom the * Software is furnished to do so, subject to the following conditions: * * The above copyright notice and this permission notice (including the next * paragraph) shall be included in all copies or substantial portions of the * Software. * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER * DEALINGS IN THE SOFTWARE. */ /** * * This lowering pass converts all interface blocks with instance names * into interface blocks without an instance name. * * For example, the following shader: * * out block { * float block_var; * } inst_name; * * main() * { * inst_name.block_var = 0.0; * } * * Is rewritten to: * * out block { * float block_var; * }; * * main() * { * block_var = 0.0; * } * * This takes place after the shader code has already been verified with * the interface name in place. * * The linking phase will use the interface block name rather than the * interface's instance name when linking interfaces. * * This modification to the ir allows our currently existing dead code * elimination to work with interface blocks without changes. */ #include "nir.h" #include "nir_builder.h" #include "nir_deref.h" #include "gl_nir.h" #include "glsl_types.h" #include "main/shader_types.h" struct lower_named_block_state { void *mem_ctx; struct hash_table *interface_namespace; }; static const struct glsl_type * process_array_type(const struct glsl_type *type, unsigned idx) { const struct glsl_type *element_type = glsl_get_array_element(type); unsigned length = glsl_get_length(type); if (glsl_type_is_array(element_type)) { const struct glsl_type *new_array_type = process_array_type(element_type, idx); return glsl_array_type(new_array_type, length, 0); } else { return glsl_array_type(glsl_get_struct_field(element_type, idx), length, 0); } } static nir_deref_instr * process_derefs(nir_builder *b, nir_deref_instr **p, nir_deref_instr *parent) { bool found_ifc = false; for (; *p; p++) { if ((*p)->deref_type == nir_deref_type_array) { parent = nir_build_deref_array(b, parent, (*p)->arr.index.ssa); } else if ((*p)->deref_type == nir_deref_type_struct) { if (!found_ifc) { /* We found the interface block so just skip it */ found_ifc = true; continue; } else { parent = nir_build_deref_struct(b, parent, (*p)->strct.index); } } else if ((*p)->deref_type == nir_deref_type_array_wildcard) { parent = nir_build_deref_array_wildcard(b, parent); } } return parent; } /* Disable packing on varyings used by interpolate functions. This must be * called after lowering blocks to avoid disabling packing on the entire * block. */ static void disable_varying_packing_when_used_by_interpolate_functions(nir_intrinsic_instr *intr, nir_variable *var) { if (intr->intrinsic == nir_intrinsic_interp_deref_at_centroid || intr->intrinsic == nir_intrinsic_interp_deref_at_sample || intr->intrinsic == nir_intrinsic_interp_deref_at_offset) { /* This disables varying packing for this input. */ var->data.must_be_shader_input = 1; } } static char * create_ifc_field_name_str(void *mem_ctx, nir_variable *var, const struct glsl_type *iface_t, unsigned f_idx) { char *iface_field_name = ralloc_asprintf(mem_ctx, "%s %s.%s.%s", var->data.mode == nir_var_shader_in ? "in" : "out", glsl_get_type_name(iface_t), var->name, glsl_get_struct_elem_name(iface_t, f_idx)); return iface_field_name; } static bool flatten_named_interface_deref(void *mem_ctx, nir_builder *b, nir_deref_instr *deref, nir_intrinsic_instr *intr, struct hash_table *interface_namespace, bool is_src0) { nir_variable_mode mask = nir_var_shader_in | nir_var_shader_out; if (!nir_deref_mode_is_one_of(deref, mask)) return false; nir_variable *var = nir_deref_instr_get_variable(deref); const struct glsl_type * iface_t = glsl_without_array(var->type); if (iface_t != var->interface_type) return false; nir_deref_path path; nir_deref_path_init(&path, deref, NULL); assert(path.path[0]->deref_type == nir_deref_type_var); nir_deref_instr **p = &path.path[1]; char *iface_field_name = NULL; for (; *p; p++) { if ((*p)->deref_type == nir_deref_type_struct) { iface_field_name = create_ifc_field_name_str(mem_ctx, var, iface_t, (*p)->strct.index); break; } } assert(iface_field_name); /* Find the variable in the set of flattened interface blocks */ struct hash_entry *entry = _mesa_hash_table_search(interface_namespace, iface_field_name); assert(entry); nir_variable *found_var = (nir_variable *) entry->data; if (intr->intrinsic == nir_intrinsic_store_deref || (intr->intrinsic == nir_intrinsic_copy_deref && is_src0)) found_var->data.assigned = 1; b->cursor = nir_before_instr(&intr->instr); nir_deref_instr *deref_var = nir_build_deref_var(b, found_var); if (glsl_type_is_array(found_var->type) || glsl_type_is_struct(found_var->type) || glsl_type_is_matrix(found_var->type)) { p = &path.path[1]; deref_var = process_derefs(b, p, deref_var); } disable_varying_packing_when_used_by_interpolate_functions(intr, found_var); nir_deref_path_finish(&path); nir_def_rewrite_uses(&deref->def, &deref_var->def); return true; } static bool flatten_named_interface_derefs(nir_builder *b, nir_intrinsic_instr *intr, void *cb_data) { struct lower_named_block_state *state = (struct lower_named_block_state *) cb_data; if (intr->intrinsic != nir_intrinsic_copy_deref && intr->intrinsic != nir_intrinsic_load_deref && intr->intrinsic != nir_intrinsic_store_deref && intr->intrinsic != nir_intrinsic_interp_deref_at_centroid && intr->intrinsic != nir_intrinsic_interp_deref_at_sample && intr->intrinsic != nir_intrinsic_interp_deref_at_offset && intr->intrinsic != nir_intrinsic_interp_deref_at_vertex) return false; nir_deref_instr *deref = nir_src_as_deref(intr->src[0]); bool progress = flatten_named_interface_deref(state->mem_ctx, b, deref, intr, state->interface_namespace, true); if (intr->intrinsic == nir_intrinsic_copy_deref) { deref = nir_src_as_deref(intr->src[1]); progress |= flatten_named_interface_deref(state->mem_ctx, b, deref, intr, state->interface_namespace, false); } return progress; } static void lower_named_interface_blocks(struct gl_linked_shader *sh) { void *mem_ctx = ralloc_context(NULL); struct hash_table *interface_namespace = _mesa_hash_table_create(mem_ctx, _mesa_hash_string, _mesa_key_string_equal); /* First pass: adjust instance block variables with an instance name * to not have an instance name. * * The interface block variables are stored in the interface_namespace * hash table so they can be used in the second pass. */ nir_foreach_variable_with_modes_safe(var, sh->Program->nir, nir_var_shader_in | nir_var_shader_out) { const struct glsl_type * iface_t = glsl_without_array(var->type); if (iface_t != var->interface_type) continue; for (unsigned i = 0; i < iface_t->length; i++) { const char *field_name = glsl_get_struct_elem_name(iface_t, i); char *iface_field_name = create_ifc_field_name_str(mem_ctx, var, iface_t, i); struct hash_entry *entry = _mesa_hash_table_search(interface_namespace, iface_field_name); nir_variable *found_var = entry ? (nir_variable *) entry->data : NULL; if (!found_var) { const struct glsl_struct_field *field_data = glsl_get_struct_field_data(iface_t, i); nir_variable *new_var = rzalloc(sh->Program->nir, nir_variable); new_var->name = ralloc_strdup(new_var, field_name); if (!glsl_type_is_array(var->type)) { new_var->type = glsl_get_struct_field(iface_t, i); } else { new_var->type = process_array_type(var->type, i); } new_var->data.mode = var->data.mode; new_var->data.location = field_data->location; new_var->data.location_frac = field_data->component >= 0 ? field_data->component : 0; new_var->data.explicit_location = (new_var->data.location >= 0); new_var->data.offset = field_data->offset; new_var->data.explicit_offset = (field_data->offset >= 0); new_var->data.xfb.buffer = field_data->xfb_buffer; new_var->data.explicit_xfb_buffer = field_data->explicit_xfb_buffer; new_var->data.interpolation = field_data->interpolation; new_var->data.centroid = field_data->centroid; new_var->data.sample = field_data->sample; new_var->data.patch = field_data->patch; new_var->data.stream = var->data.stream; new_var->data.how_declared = var->data.how_declared; new_var->data.from_named_ifc_block = 1; new_var->interface_type = var->type; _mesa_hash_table_insert(interface_namespace, iface_field_name, new_var); nir_shader_add_variable(sh->Program->nir, new_var); } } } /* Second pass: redirect dereferences to the new vars. */ struct lower_named_block_state state; state.mem_ctx = mem_ctx; state.interface_namespace = interface_namespace; nir_shader_intrinsics_pass(sh->Program->nir, flatten_named_interface_derefs, nir_metadata_control_flow, &state); /* Third pass: Mark now lowered blks as ordinary globals to be dead code * eliminated. Also use this oppotunity to set the compact flag where * needed now that the default interface block has been lowered away. */ nir_foreach_variable_with_modes(var, sh->Program->nir, nir_var_shader_in | nir_var_shader_out) { if (var->data.mode == nir_var_shader_in) { if (sh->Program->nir->info.stage == MESA_SHADER_TESS_EVAL && (var->data.location == VARYING_SLOT_TESS_LEVEL_INNER || var->data.location == VARYING_SLOT_TESS_LEVEL_OUTER)) { var->data.compact = glsl_type_is_scalar(glsl_without_array(var->type)); } if (sh->Program->nir->info.stage > MESA_SHADER_VERTEX && var->data.location >= VARYING_SLOT_CLIP_DIST0 && var->data.location <= VARYING_SLOT_CULL_DIST1) { var->data.compact = glsl_type_is_scalar(glsl_without_array(var->type)); } } else { assert(var->data.mode == nir_var_shader_out); if (sh->Program->nir->info.stage == MESA_SHADER_TESS_CTRL && (var->data.location == VARYING_SLOT_TESS_LEVEL_INNER || var->data.location == VARYING_SLOT_TESS_LEVEL_OUTER)) { var->data.compact = glsl_type_is_scalar(glsl_without_array(var->type)); } if (sh->Program->nir->info.stage <= MESA_SHADER_GEOMETRY && var->data.location >= VARYING_SLOT_CLIP_DIST0 && var->data.location <= VARYING_SLOT_CULL_DIST1) { var->data.compact = glsl_type_is_scalar(glsl_without_array(var->type)); } } const struct glsl_type * iface_t = glsl_without_array(var->type); if (!(iface_t == var->interface_type)) continue; var->data.mode = nir_var_shader_temp; } nir_fixup_deref_modes(sh->Program->nir); ralloc_free(mem_ctx); } void gl_nir_lower_named_interface_blocks(struct gl_shader_program *prog) { for (unsigned int i = 0; i < MESA_SHADER_STAGES; i++) { if (prog->_LinkedShaders[i] != NULL) { NIR_PASS(_, prog->_LinkedShaders[i]->Program->nir, nir_split_var_copies); lower_named_interface_blocks(prog->_LinkedShaders[i]); } } }