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    if (shader->info.uses_instanceid) {
 103       unsigned i = machine->SysSemanticToIndex[TGSI_SEMANTIC_INSTANCEID];
 104       assert(i < Elements(machine->SystemValue));
 105       machine->SystemValue[i][0] = shader->draw->instance_id;
 106    }
 107
 108    for (i = 0; i < count; i += MAX_TGSI_VERTICES) {
 109       unsigned int max_vertices = MIN2(MAX_TGSI_VERTICES, count - i);
 110
 111       /* Swizzle inputs.
 112        */
 113       for (j = 0; j < max_vertices; j++) {
 114 #if 0
 115          debug_printf("%d) Input vert:\n", i + j);
 116          for (slot = 0; slot < shader->info.num_inputs; slot++) {
 117             debug_printf("\t%d: %f %f %f %f\n", slot,
 118                          input[slot][0],
 119                          input[slot][1],
 120                          input[slot][2],
 121                          input[slot][3]);
 122          }
 123 #endif
 124
 125          for (slot = 0; slot < shader->info.num_inputs; slot++) {
 126 #if 0
 127             assert(!util_is_inf_or_nan(input[slot][0]));
 128             assert(!util_is_inf_or_nan(input[slot][1]));
 129             assert(!util_is_inf_or_nan(input[slot][2]));
 130             assert(!util_is_inf_or_nan(input[slot][3]));
 131 #endif
 132             machine->Inputs[slot].xyzw[0].f[j] = input[slot][0];
 133             machine->Inputs[slot].xyzw[1].f[j] = input[slot][1];
 134             machine->Inputs[slot].xyzw[2].f[j] = input[slot][2];
 135             machine->Inputs[slot].xyzw[3].f[j] = input[slot][3];
 136          }
 137
 138          input = (const float (*)[4])((const char *)input + input_stride);
 139       }
 140
 141       tgsi_set_exec_mask(machine,
 142                          1,
 143                          max_vertices > 1,
 144                          max_vertices > 2,
 145                          max_vertices > 3);
 146
 147       /* run interpreter */
 148       tgsi_exec_machine_run( machine );
 149
 150       /* Unswizzle all output results.
 151        */
 152       for (j = 0; j < max_vertices; j++) {
 153          for (slot = 0; slot < shader->info.num_outputs; slot++) {
 154             output[slot][0] = machine->Outputs[slot].xyzw[0].f[j];
 155             output[slot][1] = machine->Outputs[slot].xyzw[1].f[j];
 156             output[slot][2] = machine->Outputs[slot].xyzw[2].f[j];
 157             output[slot][3] = machine->Outputs[slot].xyzw[3].f[j];
 158
 159          }
 160
 161 #if 0
 162          debug_printf("%d) Post xform vert:\n", i + j);
 163          for (slot = 0; slot < shader->info.num_outputs; slot++) {
 164             debug_printf("\t%d: %f %f %f %f\n", slot,
 165                          output[slot][0],
 166                          output[slot][1],
 167                          output[slot][2],
 168                          output[slot][3]);
 169             assert(!util_is_inf_or_nan(output[slot][0]));
 170          }
 171 #endif
 172
 173          output = (float (*)[4])((char *)output + output_stride);
 174       }
 175
 176    }
 177 }
 178
 179
 180
 181
 182 static void
 183 vs_exec_delete( struct draw_vertex_shader *dvs )
 184 {
 185    FREE((void*) dvs->state.tokens);
 186    FREE( dvs );
 187 }
 188
 189
 190 struct draw_vertex_shader *
 191 draw_create_vs_exec(struct draw_context *draw,
 192                     const struct pipe_shader_state *state)
 193 {
 194    struct exec_vertex_shader *vs = CALLOC_STRUCT( exec_vertex_shader );
 195
 196    if (vs == NULL)
 197       return NULL;
 198
 199    /* we make a private copy of the tokens */
 200    vs->base.state.tokens = tgsi_dup_tokens(state->tokens);
 201    if (!vs->base.state.tokens) {
 202       FREE(vs);
 203       return NULL;
 204    }
 205
 206    tgsi_scan_shader(state->tokens, &vs->base.info);
 207
 208    vs->base.draw = draw;
 209    vs->base.prepare = vs_exec_prepare;
 210    vs->base.run_linear = vs_exec_run_linear;
 211    vs->base.delete = vs_exec_delete;
 212    vs->base.create_variant = draw_vs_create_variant_generic;
 213    vs->machine = draw->vs.machine;
 214
 215    return &vs->base;
 216 }