src/gallium/auxiliary/draw/draw_vs_exec.c

   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 #include "tgsi/tgsi_exec.h"
  45
  46
  47 struct exec_vertex_shader {
  48    struct draw_vertex_shader base;
  49    struct tgsi_exec_machine *machine;
  50 };
  51
  52 static struct exec_vertex_shader *exec_vertex_shader( struct draw_vertex_shader *vs )
  53 {
  54    return (struct exec_vertex_shader *)vs;
  55 }
  56
  57
  58 /* Not required for run_linear.
  59  */
  60 static void
  61 vs_exec_prepare( struct draw_vertex_shader *shader,
  62                  struct draw_context *draw )
  63 {
  64    struct exec_vertex_shader *evs = exec_vertex_shader(shader);
  65
  66    /* Specify the vertex program to interpret/execute.
  67     * Avoid rebinding when possible.
  68     */
  69    if (evs->machine->Tokens != shader->state.tokens) {
  70       tgsi_exec_machine_bind_shader(evs->machine,
  71                                     shader->state.tokens,
  72                                     draw->vs.num_samplers,
  73                                     draw->vs.samplers);
  74    }
  75 }
  76
  77
  78
  79
  80 /* Simplified vertex shader interface for the pt paths.  Given the
  81  * complexity of code-generating all the above operations together,
  82  * it's time to try doing all the other stuff separately.
  83  */
  84 static void
  85 vs_exec_run_linear( struct draw_vertex_shader *shader,
  86                     const float (*input)[4],
  87                     float (*output)[4],
  88                     const void *constants[PIPE_MAX_CONSTANT_BUFFERS],
  89                     const unsigned const_size[PIPE_MAX_CONSTANT_BUFFERS],
  90                     unsigned count,
  91                     unsigned input_stride,
  92                     unsigned output_stride )
  93 {
  94    struct exec_vertex_shader *evs = exec_vertex_shader(shader);
  95    struct tgsi_exec_machine *machine = evs->machine;
  96    unsigned int i, j;
  97    unsigned slot;
  98
  99    tgsi_exec_set_constant_buffers(machine, PIPE_MAX_CONSTANT_BUFFERS,
 100                                   constants, const_size);
 101
 102    for (i = 0; i < count; i += MAX_TGSI_VERTICES) {
 103       unsigned int max_vertices = MIN2(MAX_TGSI_VERTICES, count - i);
 104
 105       /* Swizzle inputs.
 106        */
 107       for (j = 0; j < max_vertices; j++) {
 108 #if 0
 109          debug_printf("%d) Input vert:\n", i + j);
 110          for (slot = 0; slot < shader->info.num_inputs; slot++) {
 111             debug_printf("\t%d: %f %f %f %f\n", slot,
 112                          input[slot][0],
 113                          input[slot][1],
 114                          input[slot][2],
 115                          input[slot][3]);
 116          }
 117 #endif
 118
 119          for (slot = 0; slot < shader->info.num_inputs; slot++) {
 120 #if 0
 121             assert(!util_is_inf_or_nan(input[slot][0]));
 122             assert(!util_is_inf_or_nan(input[slot][1]));
 123             assert(!util_is_inf_or_nan(input[slot][2]));
 124             assert(!util_is_inf_or_nan(input[slot][3]));
 125 #endif
 126             machine->Inputs[slot].xyzw[0].f[j] = input[slot][0];
 127             machine->Inputs[slot].xyzw[1].f[j] = input[slot][1];
 128             machine->Inputs[slot].xyzw[2].f[j] = input[slot][2];
 129             machine->Inputs[slot].xyzw[3].f[j] = input[slot][3];
 130          }
 131
 132          input = (const float (*)[4])((const char *)input + input_stride);
 133       }
 134
 135       tgsi_set_exec_mask(machine,
 136                          1,
 137                          max_vertices > 1,
 138                          max_vertices > 2,
 139                          max_vertices > 3);
 140
 141       /* run interpreter */
 142       tgsi_exec_machine_run( machine );
 143
 144       /* Unswizzle all output results.
 145        */
 146       for (j = 0; j < max_vertices; j++) {
 147          for (slot = 0; slot < shader->info.num_outputs; slot++) {
 148             output[slot][0] = machine->Outputs[slot].xyzw[0].f[j];
 149             output[slot][1] = machine->Outputs[slot].xyzw[1].f[j];
 150             output[slot][2] = machine->Outputs[slot].xyzw[2].f[j];
 151             output[slot][3] = machine->Outputs[slot].xyzw[3].f[j];
 152
 153          }
 154
 155 #if 0
 156          debug_printf("%d) Post xform vert:\n", i + j);
 157          for (slot = 0; slot < shader->info.num_outputs; slot++) {
 158             debug_printf("\t%d: %f %f %f %f\n", slot,
 159                          output[slot][0],
 160                          output[slot][1],
 161                          output[slot][2],
 162                          output[slot][3]);
 163             assert(!util_is_inf_or_nan(output[slot][0]));
 164          }
 165 #endif
 166
 167          output = (float (*)[4])((char *)output + output_stride);
 168       }
 169
 170    }
 171 }
 172
 173
 174
 175
 176 static void
 177 vs_exec_delete( struct draw_vertex_shader *dvs )
 178 {
 179    FREE((void*) dvs->state.tokens);
 180    FREE( dvs );
 181 }
 182
 183
 184 struct draw_vertex_shader *
 185 draw_create_vs_exec(struct draw_context *draw,
 186                     const struct pipe_shader_state *state)
 187 {
 188    struct exec_vertex_shader *vs = CALLOC_STRUCT( exec_vertex_shader );
 189
 190    if (vs == NULL)
 191       return NULL;
 192
 193    /* we make a private copy of the tokens */
 194    vs->base.state.tokens = tgsi_dup_tokens(state->tokens);
 195    if (!vs->base.state.tokens) {
 196       FREE(vs);
 197       return NULL;
 198    }
 199
 200    tgsi_scan_shader(state->tokens, &vs->base.info);
 201
 202    vs->base.draw = draw;
 203    vs->base.prepare = vs_exec_prepare;
 204    vs->base.run_linear = vs_exec_run_linear;
 205    vs->base.delete = vs_exec_delete;
 206    vs->base.create_variant = draw_vs_create_variant_generic;
 207    vs->machine = draw->vs.machine;
 208
 209    return &vs->base;
 210 }