src/compiler/nir/nir_lower_outputs_to_temporaries.c

   1 /*
   2  * Copyright © 2015 Intel Corporation
   3  *
   4  * Permission is hereby granted, free of charge, to any person obtaining a
   5  * copy of this software and associated documentation files (the "Software"),
   6  * to deal in the Software without restriction, including without limitation
   7  * the rights to use, copy, modify, merge, publish, distribute, sublicense,
   8  * and/or sell copies of the Software, and to permit persons to whom the
   9  * Software is furnished to do so, subject to the following conditions:
  10  *
  11  * The above copyright notice and this permission notice (including the next
  12  * paragraph) shall be included in all copies or substantial portions of the
  13  * Software.
  14  *
  15  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
  16  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
  17  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
  18  * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
  19  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
  20  * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
  21  * IN THE SOFTWARE.
  22  */
  23
  24 /*
  25  * Implements a pass that lowers output variables to a temporary plus an
  26  * output variable with a single copy at each exit point of the shader.
  27  * This way the output variable is only ever written.
  28  *
  29  * Because valid NIR requires that output variables are never read, this
  30  * pass is more of a helper for NIR producers and must be run before the
  31  * shader is ever validated.
  32  */
  33
  34 #include "nir.h"
  35
  36 struct lower_outputs_state {
  37    nir_shader *shader;
  38    struct exec_list old_outputs;
  39 };
  40
  41 static void
  42 emit_output_copies(nir_cursor cursor, struct lower_outputs_state *state)
  43 {
  44    assert(exec_list_length(&state->shader->outputs) ==
  45           exec_list_length(&state->old_outputs));
  46
  47    foreach_two_lists(out_node, &state->shader->outputs,
  48                      temp_node, &state->old_outputs) {
  49       nir_variable *output = exec_node_data(nir_variable, out_node, node);
  50       nir_variable *temp = exec_node_data(nir_variable, temp_node, node);
  51
  52       nir_intrinsic_instr *copy =
  53          nir_intrinsic_instr_create(state->shader, nir_intrinsic_copy_var);
  54       copy->variables[0] = nir_deref_var_create(copy, output);
  55       copy->variables[1] = nir_deref_var_create(copy, temp);
  56
  57       nir_instr_insert(cursor, &copy->instr);
  58    }
  59 }
  60
  61 static bool
  62 emit_output_copies_block(nir_block *block, void *state)
  63 {
  64    nir_foreach_instr(block, instr) {
  65       if (instr->type != nir_instr_type_intrinsic)
  66          continue;
  67
  68       nir_intrinsic_instr *intrin = nir_instr_as_intrinsic(instr);
  69       if (intrin->intrinsic == nir_intrinsic_emit_vertex)
  70          emit_output_copies(nir_before_instr(&intrin->instr), state);
  71    }
  72
  73    return true;
  74 }
  75
  76 void
  77 nir_lower_outputs_to_temporaries(nir_shader *shader, nir_function *entrypoint)
  78 {
  79    struct lower_outputs_state state;
  80
  81    if (shader->stage == MESA_SHADER_TESS_CTRL)
  82       return;
  83
  84    state.shader = shader;
  85    exec_list_move_nodes_to(&shader->outputs, &state.old_outputs);
  86
  87    /* Walk over all of the outputs turn each output into a temporary and
  88     * make a new variable for the actual output.
  89     */
  90    nir_foreach_variable(var, &state.old_outputs) {
  91       nir_variable *output = ralloc(shader, nir_variable);
  92       memcpy(output, var, sizeof *output);
  93
  94       /* The orignal is now the temporary */
  95       nir_variable *temp = var;
  96
  97       /* Reparent the name to the new variable */
  98       ralloc_steal(output, output->name);
  99
 100       /* Reparent the constant initializer (if any) */
 101       ralloc_steal(output, output->constant_initializer);
 102
 103       /* Give the output a new name with @out-temp appended */
 104       temp->name = ralloc_asprintf(var, "%s@out-temp", output->name);
 105       temp->data.mode = nir_var_global;
 106       temp->constant_initializer = NULL;
 107
 108       exec_list_push_tail(&shader->outputs, &output->node);
 109    }
 110
 111    nir_foreach_function(shader, function) {
 112       if (function->impl == NULL)
 113          continue;
 114
 115       if (shader->stage == MESA_SHADER_GEOMETRY) {
 116          /* For geometry shaders, we have to emit the output copies right
 117           * before each EmitVertex call.
 118           */
 119          nir_foreach_block(function->impl, emit_output_copies_block, &state);
 120       } else if (function == entrypoint) {
 121          /* For all other shader types, we need to do the copies right before
 122           * the jumps to the end block.
 123           */
 124          struct set_entry *block_entry;
 125          set_foreach(function->impl->end_block->predecessors, block_entry) {
 126             struct nir_block *block = (void *)block_entry->key;
 127             emit_output_copies(nir_after_block_before_jump(block), &state);
 128          }
 129       }
 130
 131       nir_metadata_preserve(function->impl, nir_metadata_block_index |
 132                                             nir_metadata_dominance);
 133    }
 134
 135    exec_list_append(&shader->globals, &state.old_outputs);
 136 }