summaryrefslogtreecommitdiff
path: root/src/mesa/pipe/i965simple/brw_wm_decl.c
blob: 5b1abb9d4892ec546318b976c9a2677dff7e5b2e (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
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377

#include "brw_context.h"
#include "brw_eu.h"
#include "brw_wm.h"
#include "pipe/p_util.h"
#include "pipe/p_shader_tokens.h"
#include "pipe/tgsi/util/tgsi_parse.h"

static struct brw_reg alloc_tmp(struct brw_wm_compile *c)
{
   c->tmp_index++;
   c->reg_index = MAX2(c->reg_index, c->tmp_start + c->tmp_index);
   return brw_vec8_grf(c->tmp_start + c->tmp_index, 0);
}

static void release_tmps(struct brw_wm_compile *c)
{
   c->tmp_index = 0;
}



static int is_null( struct brw_reg reg )
{
   return (reg.file == BRW_ARCHITECTURE_REGISTER_FILE &&
	   reg.nr == BRW_ARF_NULL);
}

static void emit_pixel_xy( struct brw_wm_compile *c )
{
   if (is_null(c->pixel_xy[0])) {

      struct brw_compile *p = &c->func;
      struct brw_reg r1_uw = retype(brw_vec1_grf(1, 0), BRW_REGISTER_TYPE_UW);

      c->pixel_xy[0] = vec8(retype(alloc_tmp(c), BRW_REGISTER_TYPE_UW));
      c->pixel_xy[1] = vec8(retype(alloc_tmp(c), BRW_REGISTER_TYPE_UW));

      /* Calculate pixel centers by adding 1 or 0 to each of the
       * micro-tile coordinates passed in r1.
       */
      brw_ADD(p,
	      c->pixel_xy[0],
	      stride(suboffset(r1_uw, 4), 2, 4, 0),
	      brw_imm_v(0x10101010));

      brw_ADD(p,
	      c->pixel_xy[1],
	      stride(suboffset(r1_uw, 5), 2, 4, 0),
	      brw_imm_v(0x11001100));
   }
}






static void emit_delta_xy( struct brw_wm_compile *c )
{
   if (is_null(c->delta_xy[0])) {
      struct brw_compile *p = &c->func;
      struct brw_reg r1 = brw_vec1_grf(1, 0);

      emit_pixel_xy(c);

      c->delta_xy[0] = alloc_tmp(c);
      c->delta_xy[1] = alloc_tmp(c);

      /* Calc delta X,Y by subtracting origin in r1 from the pixel
       * centers.
       */
      brw_ADD(p,
	      c->delta_xy[0],
	      retype(c->pixel_xy[0], BRW_REGISTER_TYPE_UW),
	      negate(r1));

      brw_ADD(p,
	      c->delta_xy[1],
	      retype(c->pixel_xy[1], BRW_REGISTER_TYPE_UW),
	      negate(suboffset(r1,1)));
   }
}



#if 0
static void emit_pixel_w( struct brw_wm_compile *c )
{
   if (is_null(c->pixel_w)) {
      struct brw_compile *p = &c->func;

      struct brw_reg interp_wpos = c->coef_wpos;
      
      c->pixel_w = alloc_tmp(c);

      emit_delta_xy(c);

      /* Calc 1/w - just linterp wpos[3] optimized by putting the
       * result straight into a message reg.
       */
      struct brw_reg interp3 = brw_vec1_grf(interp_wpos.nr+1, 4);
      brw_LINE(p, brw_null_reg(), interp3, c->delta_xy[0]);
      brw_MAC(p, brw_message_reg(2), suboffset(interp3, 1), c->delta_xy[1]);

      /* Calc w */
      brw_math_16( p, 
		   c->pixel_w,
		   BRW_MATH_FUNCTION_INV,
		   BRW_MATH_SATURATE_NONE,
		   2, 
		   brw_null_reg(),
		   BRW_MATH_PRECISION_FULL);
   }
}
#endif


static void emit_cinterp(struct brw_wm_compile *c,
			 int idx,
			 int mask )
{
   struct brw_compile *p = &c->func;
   struct brw_reg interp[4];
   struct brw_reg coef = c->payload_coef[idx];
   int i;

   interp[0] = brw_vec1_grf(coef.nr, 0);
   interp[1] = brw_vec1_grf(coef.nr, 4);
   interp[2] = brw_vec1_grf(coef.nr+1, 0);
   interp[3] = brw_vec1_grf(coef.nr+1, 4);

   for(i = 0; i < 4; i++ ) {
      if (mask & (1<<i)) {
	 struct brw_reg dst = c->wm_regs[TGSI_FILE_INPUT][idx][i];
	 brw_MOV(p, dst, suboffset(interp[i],3));
      }
   }
}

static void emit_linterp(struct brw_wm_compile *c,
			 int idx,
			 int mask )
{
   struct brw_compile *p = &c->func;
   struct brw_reg interp[4];
   struct brw_reg coef = c->payload_coef[idx];
   int i;

   emit_delta_xy(c);

   interp[0] = brw_vec1_grf(coef.nr, 0);
   interp[1] = brw_vec1_grf(coef.nr, 4);
   interp[2] = brw_vec1_grf(coef.nr+1, 0);
   interp[3] = brw_vec1_grf(coef.nr+1, 4);

   for(i = 0; i < 4; i++ ) {
      if (mask & (1<<i)) {
	 struct brw_reg dst = c->wm_regs[TGSI_FILE_INPUT][idx][i];
	 brw_LINE(p, brw_null_reg(), interp[i], c->delta_xy[0]);
	 brw_MAC(p, dst, suboffset(interp[i],1), c->delta_xy[1]);
      }
   }
}

#if 0
static void emit_pinterp(struct brw_wm_compile *c,
			 int idx,
			 int mask )
{
   struct brw_compile *p = &c->func;
   struct brw_reg interp[4];
   struct brw_reg coef = c->payload_coef[idx];
   int i;

   get_delta_xy(c);
   get_pixel_w(c);

   interp[0] = brw_vec1_grf(coef.nr, 0);
   interp[1] = brw_vec1_grf(coef.nr, 4);
   interp[2] = brw_vec1_grf(coef.nr+1, 0);
   interp[3] = brw_vec1_grf(coef.nr+1, 4);

   for(i = 0; i < 4; i++ ) {
      if (mask & (1<<i)) {
	 struct brw_reg dst = allocate_reg(c, TGSI_FILE_INPUT, idx, i);
	 brw_LINE(p, brw_null_reg(), interp[i], c->delta_xy[0]);
	 brw_MAC(p, dst, suboffset(interp[i],1), c->delta_xy[1]);
	 brw_MUL(p, dst, dst, c->pixel_w);
      }
   }
}
#endif



#if 0
static void emit_wpos( )
{ 
   struct prog_dst_register dst = dst_reg(PROGRAM_INPUT, idx);
   struct tgsi_full_src_register interp = src_reg(PROGRAM_PAYLOAD, idx);
   struct tgsi_full_src_register deltas = get_delta_xy(c);
   struct tgsi_full_src_register arg2;
   unsigned opcode;

   opcode = WM_LINTERP;
   arg2 = src_undef();

   /* Have to treat wpos.xy specially:
    */
   emit_op(c,
	   WM_WPOSXY,
	   dst_mask(dst, WRITEMASK_XY),
	   0, 0, 0,
	   get_pixel_xy(c),
	   src_undef(),
	   src_undef());
      
   dst = dst_mask(dst, WRITEMASK_ZW);

   /* PROGRAM_INPUT.attr.xyzw = INTERP payload.interp[attr].x, deltas.xyw
    */
   emit_op(c,
	   WM_LINTERP,
	   dst,
	   0, 0, 0,
	   interp,
	   deltas,
	   arg2);
}
#endif




/* Perform register allocation:
 * 
 *  -- r0???
 *  -- passthrough depth regs (and stencil/aa??)
 *  -- curbe ??
 *  -- inputs (coefficients)
 *
 * Use a totally static register allocation.  This will perform poorly
 * but is an easy way to get started (again).
 */
static void prealloc_reg(struct brw_wm_compile *c)
{
   int i, j;
   int nr_curbe_regs = 0;

   /* R0, then some depth related regs:
    */
   for (i = 0; i < c->key.nr_depth_regs; i++) {
      c->payload_depth[i] =  brw_vec8_grf(i*2, 0);
      c->reg_index += 2;
   }


   /* Then a copy of our part of the CURBE entry:
    */
   {
      int nr_constants = c->fp->info.nr_regs[TGSI_FILE_CONSTANT];
      int index = 0;

      c->prog_data.max_const = 4*nr_constants;
      for (i = 0; i < nr_constants; i++) {
	 for (j = 0; j < 4; j++, index++) 
	    c->wm_regs[TGSI_FILE_CONSTANT][i][j] = brw_vec1_grf(c->reg_index + index/8,
								index%8);
      }

      nr_curbe_regs = 2*((4*nr_constants+15)/16);
      c->reg_index += nr_curbe_regs;
   }

   /* Next we receive the plane coefficients for parameter
    * interpolation:
    */
   for (i = 0; i < c->fp->info.nr_regs[TGSI_FILE_INPUT]; i++) {
      c->payload_coef[i] = brw_vec8_grf(c->reg_index, 0);
      c->reg_index += 2;
   }

   c->prog_data.first_curbe_grf = c->key.nr_depth_regs * 2;
   c->prog_data.urb_read_length = c->fp->program.num_inputs * 2;
   c->prog_data.curb_read_length = nr_curbe_regs;

   /* That's the end of the payload, now we can start allocating registers.
    */
   c->emit_mask_reg = brw_uw1_reg(BRW_GENERAL_REGISTER_FILE, c->reg_index, 0);
   c->reg_index++;

   c->stack = brw_uw16_reg(BRW_GENERAL_REGISTER_FILE, c->reg_index, 0);
   c->reg_index += 2;

   /* Now allocate room for the interpolated inputs and staging
    * registers for the outputs:
    */
   for (i = 0; i < c->fp->info.nr_regs[TGSI_FILE_INPUT]; i++) 
      for (j = 0; j < 4; j++)
	 c->wm_regs[TGSI_FILE_INPUT][i][j] = brw_vec8_grf( c->reg_index++, 0 );

   for (i = 0; i < c->fp->info.nr_regs[TGSI_FILE_OUTPUT]; i++) 
      for (j = 0; j < 4; j++)
	 c->wm_regs[TGSI_FILE_OUTPUT][i][j] = brw_vec8_grf( c->reg_index++, 0 );

   /* Beyond this we should only need registers for internal temporaries:
    */
   c->tmp_start = c->reg_index;
}





/* Need to interpolate fragment program inputs in as a preamble to the
 * shader.  A more sophisticated compiler would do this on demand, but
 * we'll do it up front:
 */
void brw_wm_emit_decls(struct brw_wm_compile *c)
{
   struct tgsi_parse_context parse;
   int done = 0;

   prealloc_reg(c);

   tgsi_parse_init( &parse, c->fp->program.tokens );

   while( !done &&
	  !tgsi_parse_end_of_tokens( &parse ) ) 
   {
      tgsi_parse_token( &parse );

      switch( parse.FullToken.Token.Type ) {
      case TGSI_TOKEN_TYPE_DECLARATION:
      {
	 const struct tgsi_full_declaration *decl = &parse.FullToken.FullDeclaration;
	 unsigned first = decl->u.DeclarationRange.First;
	 unsigned last = decl->u.DeclarationRange.Last;
	 unsigned mask = decl->Declaration.UsageMask; /* ? */
	 unsigned i;

	 if (decl->Declaration.File != TGSI_FILE_INPUT)
	    break;

	 assert(decl->Declaration.Interpolate);

	 for( i = first; i <= last; i++ ) {
	    switch (decl->Interpolation.Interpolate) {
	    case TGSI_INTERPOLATE_CONSTANT:
	       emit_cinterp(c, i, mask);
	       break;

	    case TGSI_INTERPOLATE_LINEAR:
	       emit_linterp(c, i, mask);
	       break;

	    case TGSI_INTERPOLATE_PERSPECTIVE:
	       //emit_pinterp(c, i, mask);
	       emit_linterp(c, i, mask);
	       break;
	    }
	 }
	 break;
      }
      case TGSI_TOKEN_TYPE_IMMEDIATE:
      case TGSI_TOKEN_TYPE_INSTRUCTION:
      default:
         done = 1;
	 break;
      }
   }

   tgsi_parse_free (&parse);
   
   release_tmps(c);
}