summaryrefslogtreecommitdiff
path: root/src/mesa/pipe/draw/draw_vertex_shader.c
blob: c2e038453ec6dfecf35a7d77ba7225fa67aea4f7 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
/**************************************************************************
 * 
 * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
 * All Rights Reserved.
 * 
 * Permission is hereby granted, free of charge, to any person obtaining a
 * copy of this software and associated documentation files (the
 * "Software"), to deal in the Software without restriction, including
 * without limitation the rights to use, copy, modify, merge, publish,
 * distribute, sub license, and/or sell copies of the Software, and to
 * permit persons to whom the Software is furnished to do so, subject to
 * the following conditions:
 * 
 * The above copyright notice and this permission notice (including the
 * next paragraph) shall be included in all copies or substantial portions
 * of the Software.
 * 
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
 * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
 * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
 * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
 * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
 * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 * 
 **************************************************************************/

 /*
  * Authors:
  *   Keith Whitwell <keith@tungstengraphics.com>
  *   Brian Paul
  */

#include "pipe/p_util.h"
#include "pipe/p_shader_tokens.h"
#if defined(__i386__) || defined(__386__)
#include "pipe/tgsi/exec/tgsi_sse2.h"
#endif
#include "draw_private.h"
#include "draw_context.h"
#include "draw_vertex.h"

#include "x86/rtasm/x86sse.h"
#include "pipe/llvm/gallivm.h"


#define DBG_VS 0


static INLINE unsigned
compute_clipmask(const float *clip, /*const*/ float plane[][4], unsigned nr)
{
   unsigned mask = 0;
   unsigned i;

   /* Do the hardwired planes first:
    */
   if (-clip[0] + clip[3] < 0) mask |= CLIP_RIGHT_BIT;
   if ( clip[0] + clip[3] < 0) mask |= CLIP_LEFT_BIT;
   if (-clip[1] + clip[3] < 0) mask |= CLIP_TOP_BIT;
   if ( clip[1] + clip[3] < 0) mask |= CLIP_BOTTOM_BIT;
   if (-clip[2] + clip[3] < 0) mask |= CLIP_FAR_BIT;
   if ( clip[2] + clip[3] < 0) mask |= CLIP_NEAR_BIT;

   /* Followed by any remaining ones:
    */
   for (i = 6; i < nr; i++) {
      if (dot4(clip, plane[i]) < 0) 
         mask |= (1<<i);
   }

   return mask;
}


typedef void (XSTDCALL *codegen_function) (
   const struct tgsi_exec_vector *input,
   struct tgsi_exec_vector *output,
   float (*constant)[4],
   struct tgsi_exec_vector *temporary );


/**
 * Transform vertices with the current vertex program/shader
 * Up to four vertices can be shaded at a time.
 * \param vbuffer  the input vertex data
 * \param elts  indexes of four input vertices
 * \param count  number of vertices to shade [1..4]
 * \param vOut  array of pointers to four output vertices
 */
static void
run_vertex_program(struct draw_context *draw,
                   unsigned elts[4], unsigned count,
                   struct vertex_header *vOut[])
{
   struct tgsi_exec_machine *machine = &draw->machine;
   unsigned int j;

   ALIGN16_DECL(struct tgsi_exec_vector, inputs, PIPE_ATTRIB_MAX);
   ALIGN16_DECL(struct tgsi_exec_vector, outputs, PIPE_ATTRIB_MAX);
   const float *scale = draw->viewport.scale;
   const float *trans = draw->viewport.translate;

   assert(count <= 4);
   assert(draw->vertex_shader->state->output_semantic_name[0]
          == TGSI_SEMANTIC_POSITION);

   /* Consts does not require 16 byte alignment. */
   machine->Consts = (float (*)[4]) draw->user.constants;

   machine->Inputs = ALIGN16_ASSIGN(inputs);
   machine->Outputs = ALIGN16_ASSIGN(outputs);

   draw_vertex_fetch( draw, machine, elts, count );

   /* run shader */
#if defined(__i386__) || defined(__386__)
   if (draw->use_sse) {
      /* SSE */
      /* cast away const */
      struct draw_vertex_shader *shader
         = (struct draw_vertex_shader *)draw->vertex_shader;
      codegen_function func
         = (codegen_function) x86_get_func( &shader->sse2_program );
      func(
         machine->Inputs,
         machine->Outputs,
         machine->Consts,
         machine->Temps );
   }
   else
#endif
   {
      /* interpreter */
      tgsi_exec_machine_run( machine );
   }

   /* store machine results */
   for (j = 0; j < count; j++) {
      unsigned slot;
      float x, y, z, w;

      /* Handle attr[0] (position) specially:
       *
       * XXX: Computing the clipmask should be done in the vertex
       * program as a set of DP4 instructions appended to the
       * user-provided code.
       */
      x = vOut[j]->clip[0] = machine->Outputs[0].xyzw[0].f[j];
      y = vOut[j]->clip[1] = machine->Outputs[0].xyzw[1].f[j];
      z = vOut[j]->clip[2] = machine->Outputs[0].xyzw[2].f[j];
      w = vOut[j]->clip[3] = machine->Outputs[0].xyzw[3].f[j];

      vOut[j]->clipmask = compute_clipmask(vOut[j]->clip, draw->plane, draw->nr_planes);
      vOut[j]->edgeflag = 1;

      /* divide by w */
      w = 1.0f / w;
      x *= w;
      y *= w;
      z *= w;

      /* Viewport mapping */
      vOut[j]->data[0][0] = x * scale[0] + trans[0];
      vOut[j]->data[0][1] = y * scale[1] + trans[1];
      vOut[j]->data[0][2] = z * scale[2] + trans[2];
      vOut[j]->data[0][3] = w;

#if DBG_VS
      printf("output[%d]win: %f %f %f %f\n", j,
             vOut[j]->data[0][0],
             vOut[j]->data[0][1],
             vOut[j]->data[0][2],
             vOut[j]->data[0][3]);
#endif
      /* Remaining attributes are packed into sequential post-transform
       * vertex attrib slots.
       */
      for (slot = 1; slot < draw->vertex_info.num_attribs; slot++) {
         vOut[j]->data[slot][0] = machine->Outputs[slot].xyzw[0].f[j];
         vOut[j]->data[slot][1] = machine->Outputs[slot].xyzw[1].f[j];
         vOut[j]->data[slot][2] = machine->Outputs[slot].xyzw[2].f[j];
         vOut[j]->data[slot][3] = machine->Outputs[slot].xyzw[3].f[j];
#if DBG_VS
         printf("output[%d][%d]: %f %f %f %f\n", j, slot,
                vOut[j]->data[slot][0],
                vOut[j]->data[slot][1],
                vOut[j]->data[slot][2],
                vOut[j]->data[slot][3]);
#endif
      }
   } /* loop over vertices */
}


/**
 * Run the vertex shader on all vertices in the vertex queue.
 * Called by the draw module when the vertx cache needs to be flushed.
 */
void
draw_vertex_shader_queue_flush(struct draw_context *draw)
{
   unsigned i, j;

//   fprintf(stderr, " q(%d) ", draw->vs.queue_nr );
#ifdef MESA_LLVM
   if (draw->vertex_shader->llvm_prog) {
      draw_vertex_shader_queue_flush_llvm(draw);
      return;
   }
#endif

   /* run vertex shader on vertex cache entries, four per invokation */
   for (i = 0; i < draw->vs.queue_nr; i += 4) {
      struct vertex_header *dests[4];
      unsigned elts[4];
      int n;

      for (j = 0; j < 4; j++) {
         elts[j] = draw->vs.queue[i + j].elt;
         dests[j] = draw->vs.queue[i + j].dest;
      }

      n = MIN2(4, draw->vs.queue_nr - i);
      assert(n > 0);
      assert(n <= 4);

      run_vertex_program(draw, elts, n, dests);
   }

   draw->vs.queue_nr = 0;
}


struct draw_vertex_shader *
draw_create_vertex_shader(struct draw_context *draw,
                          const struct pipe_shader_state *shader)
{
   struct draw_vertex_shader *vs;

   vs = CALLOC_STRUCT( draw_vertex_shader );
   if (vs == NULL) {
      return NULL;
   }

   vs->state = shader;

#ifdef MESA_LLVM
   vs->llvm_prog = gallivm_from_tgsi(shader->tokens, GALLIVM_VS);
   draw->engine = gallivm_global_cpu_engine();
   if (!draw->engine) {
      draw->engine = gallivm_cpu_engine_create(vs->llvm_prog);
   }
   else {
      gallivm_cpu_jit_compile(draw->engine, vs->llvm_prog);
   }
#elif defined(__i386__) || defined(__386__)
   if (draw->use_sse) {
      /* cast-away const */
      struct pipe_shader_state *sh = (struct pipe_shader_state *) shader;

      x86_init_func( &vs->sse2_program );
      tgsi_emit_sse2( (struct tgsi_token *) sh->tokens, &vs->sse2_program );
   }
#endif

   return vs;
}


void
draw_bind_vertex_shader(struct draw_context *draw,
                        struct draw_vertex_shader *dvs)
{
   draw_flush(draw);
   draw->vertex_shader = dvs;

   /* specify the fragment program to interpret/execute */
   tgsi_exec_machine_init(&draw->machine,
                          draw->vertex_shader->state->tokens,
                          PIPE_MAX_SAMPLERS,
                          NULL /*samplers*/ );
}


void
draw_delete_vertex_shader(struct draw_context *draw,
                          struct draw_vertex_shader *dvs)
{
#if defined(__i386__) || defined(__386__)
   x86_release_func( (struct x86_function *) &dvs->sse2_program );
#endif

   FREE( dvs );
}