draw_vs_exec.c revision 986d4a9ec60ee52067b2d6f81c05f8ec3b8dc8b1
1/************************************************************************** 2 * 3 * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. 4 * All Rights Reserved. 5 * 6 * Permission is hereby granted, free of charge, to any person obtaining a 7 * copy of this software and associated documentation files (the 8 * "Software"), to deal in the Software without restriction, including 9 * without limitation the rights to use, copy, modify, merge, publish, 10 * distribute, sub license, and/or sell copies of the Software, and to 11 * permit persons to whom the Software is furnished to do so, subject to 12 * the following conditions: 13 * 14 * The above copyright notice and this permission notice (including the 15 * next paragraph) shall be included in all copies or substantial portions 16 * of the Software. 17 * 18 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS 19 * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF 20 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. 21 * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR 22 * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, 23 * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE 24 * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. 25 * 26 **************************************************************************/ 27 28 /* 29 * Authors: 30 * Keith Whitwell <keith@tungstengraphics.com> 31 * Brian Paul 32 */ 33 34#include "util/u_math.h" 35#include "util/u_memory.h" 36#include "pipe/p_shader_tokens.h" 37 38#include "draw_private.h" 39#include "draw_context.h" 40#include "draw_vs.h" 41 42#include "tgsi/tgsi_parse.h" 43#include "tgsi/tgsi_scan.h" 44 45 46struct exec_vertex_shader { 47 struct draw_vertex_shader base; 48 struct tgsi_exec_machine *machine; 49}; 50 51static struct exec_vertex_shader *exec_vertex_shader( struct draw_vertex_shader *vs ) 52{ 53 return (struct exec_vertex_shader *)vs; 54} 55 56 57/* Not required for run_linear. 58 */ 59static void 60vs_exec_prepare( struct draw_vertex_shader *shader, 61 struct draw_context *draw ) 62{ 63 struct exec_vertex_shader *evs = exec_vertex_shader(shader); 64 65 /* Specify the vertex program to interpret/execute. 66 * Avoid rebinding when possible. 67 */ 68 if (evs->machine->Tokens != shader->state.tokens) { 69 tgsi_exec_machine_bind_shader(evs->machine, 70 shader->state.tokens, 71 draw->vs.num_samplers, 72 draw->vs.samplers); 73 } 74} 75 76 77 78 79/* Simplified vertex shader interface for the pt paths. Given the 80 * complexity of code-generating all the above operations together, 81 * it's time to try doing all the other stuff separately. 82 */ 83static void 84vs_exec_run_linear( struct draw_vertex_shader *shader, 85 const float (*input)[4], 86 float (*output)[4], 87 const float (*constants)[4], 88 unsigned count, 89 unsigned input_stride, 90 unsigned output_stride ) 91{ 92 struct exec_vertex_shader *evs = exec_vertex_shader(shader); 93 struct tgsi_exec_machine *machine = evs->machine; 94 unsigned int i, j; 95 unsigned slot; 96 97 machine->Consts = constants; 98 99 for (i = 0; i < count; i += MAX_TGSI_VERTICES) { 100 unsigned int max_vertices = MIN2(MAX_TGSI_VERTICES, count - i); 101 102 /* Swizzle inputs. 103 */ 104 for (j = 0; j < max_vertices; j++) { 105#if 0 106 debug_printf("%d) Input vert:\n", i + j); 107 for (slot = 0; slot < shader->info.num_inputs; slot++) { 108 debug_printf("\t%d: %f %f %f %f\n", slot, 109 input[slot][0], 110 input[slot][1], 111 input[slot][2], 112 input[slot][3]); 113 } 114#endif 115 116 for (slot = 0; slot < shader->info.num_inputs; slot++) { 117#if 0 118 assert(!util_is_inf_or_nan(input[slot][0])); 119 assert(!util_is_inf_or_nan(input[slot][1])); 120 assert(!util_is_inf_or_nan(input[slot][2])); 121 assert(!util_is_inf_or_nan(input[slot][3])); 122#endif 123 machine->Inputs[slot].xyzw[0].f[j] = input[slot][0]; 124 machine->Inputs[slot].xyzw[1].f[j] = input[slot][1]; 125 machine->Inputs[slot].xyzw[2].f[j] = input[slot][2]; 126 machine->Inputs[slot].xyzw[3].f[j] = input[slot][3]; 127 } 128 129 input = (const float (*)[4])((const char *)input + input_stride); 130 } 131 132 tgsi_set_exec_mask(machine, 133 1, 134 max_vertices > 1, 135 max_vertices > 2, 136 max_vertices > 3); 137 138 /* run interpreter */ 139 tgsi_exec_machine_run( machine ); 140 141 /* Unswizzle all output results. 142 */ 143 for (j = 0; j < max_vertices; j++) { 144 for (slot = 0; slot < shader->info.num_outputs; slot++) { 145 output[slot][0] = machine->Outputs[slot].xyzw[0].f[j]; 146 output[slot][1] = machine->Outputs[slot].xyzw[1].f[j]; 147 output[slot][2] = machine->Outputs[slot].xyzw[2].f[j]; 148 output[slot][3] = machine->Outputs[slot].xyzw[3].f[j]; 149 150 } 151 152#if 0 153 debug_printf("%d) Post xform vert:\n", i + j); 154 for (slot = 0; slot < shader->info.num_outputs; slot++) { 155 debug_printf("\t%d: %f %f %f %f\n", slot, 156 output[slot][0], 157 output[slot][1], 158 output[slot][2], 159 output[slot][3]); 160 assert(!util_is_inf_or_nan(output[slot][0])); 161 } 162#endif 163 164 output = (float (*)[4])((char *)output + output_stride); 165 } 166 167 } 168} 169 170 171 172 173static void 174vs_exec_delete( struct draw_vertex_shader *dvs ) 175{ 176 FREE((void*) dvs->state.tokens); 177 FREE( dvs ); 178} 179 180 181struct draw_vertex_shader * 182draw_create_vs_exec(struct draw_context *draw, 183 const struct pipe_shader_state *state) 184{ 185 struct exec_vertex_shader *vs = CALLOC_STRUCT( exec_vertex_shader ); 186 187 if (vs == NULL) 188 return NULL; 189 190 /* we make a private copy of the tokens */ 191 vs->base.state.tokens = tgsi_dup_tokens(state->tokens); 192 if (!vs->base.state.tokens) { 193 FREE(vs); 194 return NULL; 195 } 196 197 tgsi_scan_shader(state->tokens, &vs->base.info); 198 199 vs->base.draw = draw; 200 vs->base.prepare = vs_exec_prepare; 201 vs->base.run_linear = vs_exec_run_linear; 202 vs->base.delete = vs_exec_delete; 203 vs->base.create_varient = draw_vs_varient_generic; 204 vs->machine = &draw->vs.machine; 205 206 return &vs->base; 207} 208