1 /* 2 * Copyright 2016 Broadcom 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 #include "nir.h" 25 #include "nir_builder.h" 26 27 /** @file nir_lower_io_to_scalar.c 28 * 29 * Replaces nir_load_input/nir_store_output operations with num_components != 30 * 1 with individual per-channel operations. 31 */ 32 33 static void 34 lower_load_input_to_scalar(nir_builder *b, nir_intrinsic_instr *intr) 35 { 36 b->cursor = nir_before_instr(&intr->instr); 37 38 assert(intr->dest.is_ssa); 39 40 nir_ssa_def *loads[4]; 41 42 for (unsigned i = 0; i < intr->num_components; i++) { 43 nir_intrinsic_instr *chan_intr = 44 nir_intrinsic_instr_create(b->shader, intr->intrinsic); 45 nir_ssa_dest_init(&chan_intr->instr, &chan_intr->dest, 46 1, intr->dest.ssa.bit_size, NULL); 47 chan_intr->num_components = 1; 48 49 nir_intrinsic_set_base(chan_intr, nir_intrinsic_base(intr)); 50 nir_intrinsic_set_component(chan_intr, nir_intrinsic_component(intr) + i); 51 /* offset */ 52 chan_intr->src[0] = intr->src[0]; 53 54 nir_builder_instr_insert(b, &chan_intr->instr); 55 56 loads[i] = &chan_intr->dest.ssa; 57 } 58 59 nir_ssa_def_rewrite_uses(&intr->dest.ssa, 60 nir_src_for_ssa(nir_vec(b, loads, 61 intr->num_components))); 62 nir_instr_remove(&intr->instr); 63 } 64 65 static void 66 lower_store_output_to_scalar(nir_builder *b, nir_intrinsic_instr *intr) 67 { 68 b->cursor = nir_before_instr(&intr->instr); 69 70 nir_ssa_def *value = nir_ssa_for_src(b, intr->src[0], intr->num_components); 71 72 for (unsigned i = 0; i < intr->num_components; i++) { 73 if (!(nir_intrinsic_write_mask(intr) & (1 << i))) 74 continue; 75 76 nir_intrinsic_instr *chan_intr = 77 nir_intrinsic_instr_create(b->shader, intr->intrinsic); 78 chan_intr->num_components = 1; 79 80 nir_intrinsic_set_base(chan_intr, nir_intrinsic_base(intr)); 81 nir_intrinsic_set_write_mask(chan_intr, 0x1); 82 nir_intrinsic_set_component(chan_intr, nir_intrinsic_component(intr) + i); 83 84 /* value */ 85 chan_intr->src[0] = nir_src_for_ssa(nir_channel(b, value, i)); 86 /* offset */ 87 chan_intr->src[1] = intr->src[1]; 88 89 nir_builder_instr_insert(b, &chan_intr->instr); 90 } 91 92 nir_instr_remove(&intr->instr); 93 } 94 95 void 96 nir_lower_io_to_scalar(nir_shader *shader, nir_variable_mode mask) 97 { 98 nir_foreach_function(function, shader) { 99 if (function->impl) { 100 nir_builder b; 101 nir_builder_init(&b, function->impl); 102 103 nir_foreach_block(block, function->impl) { 104 nir_foreach_instr_safe(instr, block) { 105 if (instr->type != nir_instr_type_intrinsic) 106 continue; 107 108 nir_intrinsic_instr *intr = nir_instr_as_intrinsic(instr); 109 110 if (intr->num_components == 1) 111 continue; 112 113 switch (intr->intrinsic) { 114 case nir_intrinsic_load_input: 115 if (mask & nir_var_shader_in) 116 lower_load_input_to_scalar(&b, intr); 117 break; 118 case nir_intrinsic_store_output: 119 if (mask & nir_var_shader_out) 120 lower_store_output_to_scalar(&b, intr); 121 break; 122 default: 123 break; 124 } 125 } 126 } 127 } 128 } 129 } 130