summaryrefslogtreecommitdiff
path: root/src/mesa/drivers/dri/i965/brw_draw.c
blob: 0990dcfac423a2df1f958b324717c6a780828164 (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
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
/**************************************************************************
 * 
 * Copyright 2003 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.
 * 
 **************************************************************************/

#include <stdlib.h>

#include "glheader.h"
#include "context.h"
#include "state.h"
#include "api_validate.h"
#include "enums.h"

#include "brw_draw.h"
#include "brw_defines.h"
#include "brw_context.h"
#include "brw_state.h"
#include "brw_fallback.h"

#include "intel_ioctl.h"
#include "intel_batchbuffer.h"
#include "intel_buffer_objects.h"

#include "tnl/tnl.h"
#include "vbo/vbo_context.h"
#include "swrast/swrast.h"
#include "swrast_setup/swrast_setup.h"

#define FILE_DEBUG_FLAG DEBUG_BATCH

static GLuint hw_prim[GL_POLYGON+1] = {
   _3DPRIM_POINTLIST,
   _3DPRIM_LINELIST,
   _3DPRIM_LINELOOP,
   _3DPRIM_LINESTRIP,
   _3DPRIM_TRILIST,
   _3DPRIM_TRISTRIP,
   _3DPRIM_TRIFAN,
   _3DPRIM_QUADLIST,
   _3DPRIM_QUADSTRIP,
   _3DPRIM_POLYGON
};


static const GLenum reduced_prim[GL_POLYGON+1] = {  
   GL_POINTS,
   GL_LINES,
   GL_LINES,
   GL_LINES,
   GL_TRIANGLES,
   GL_TRIANGLES,
   GL_TRIANGLES,
   GL_TRIANGLES,
   GL_TRIANGLES,
   GL_TRIANGLES
};


/* When the primitive changes, set a state bit and re-validate.  Not
 * the nicest and would rather deal with this by having all the
 * programs be immune to the active primitive (ie. cope with all
 * possibilities).  That may not be realistic however.
 */
static GLuint brw_set_prim(struct brw_context *brw, GLenum prim)
{
   if (INTEL_DEBUG & DEBUG_PRIMS)
      _mesa_printf("PRIM: %s\n", _mesa_lookup_enum_by_nr(prim));
   
   /* Slight optimization to avoid the GS program when not needed:
    */
   if (prim == GL_QUAD_STRIP &&
       brw->attribs.Light->ShadeModel != GL_FLAT &&
       brw->attribs.Polygon->FrontMode == GL_FILL &&
       brw->attribs.Polygon->BackMode == GL_FILL)
      prim = GL_TRIANGLE_STRIP;

   if (prim != brw->primitive) {
      brw->primitive = prim;
      brw->state.dirty.brw |= BRW_NEW_PRIMITIVE;

      if (reduced_prim[prim] != brw->intel.reduced_primitive) {
	 brw->intel.reduced_primitive = reduced_prim[prim];
	 brw->state.dirty.brw |= BRW_NEW_REDUCED_PRIMITIVE;
      }

      brw_validate_state(brw);
   }

   return hw_prim[prim];
}


static GLuint trim(GLenum prim, GLuint length)
{
   if (prim == GL_QUAD_STRIP)
      return length > 3 ? (length - length % 2) : 0;
   else if (prim == GL_QUADS)
      return length - length % 4;
   else 
      return length;
}


static void brw_emit_prim( struct brw_context *brw, 
			   const struct _mesa_prim *prim )

{
   struct brw_3d_primitive prim_packet;

   if (INTEL_DEBUG & DEBUG_PRIMS)
      _mesa_printf("PRIM: %s %d %d\n", _mesa_lookup_enum_by_nr(prim->mode), 
		   prim->start, prim->count);

   prim_packet.header.opcode = CMD_3D_PRIM;
   prim_packet.header.length = sizeof(prim_packet)/4 - 2;
   prim_packet.header.pad = 0;
   prim_packet.header.topology = brw_set_prim(brw, prim->mode);
   prim_packet.header.indexed = prim->indexed;

   prim_packet.verts_per_instance = trim(prim->mode, prim->count);
   prim_packet.start_vert_location = prim->start;
   prim_packet.instance_count = 1;
   prim_packet.start_instance_location = 0;
   prim_packet.base_vert_location = 0;

   if (prim_packet.verts_per_instance) {
      intel_batchbuffer_data( brw->intel.batch, &prim_packet,
			      sizeof(prim_packet), LOOP_CLIPRECTS);
   }
}

static void brw_merge_inputs( struct brw_context *brw,
		       const struct gl_client_array *arrays[])
{
   struct brw_vertex_element *inputs = brw->vb.inputs;
   struct brw_vertex_info old = brw->vb.info;
   GLuint i;

   memset(inputs, 0, sizeof(*inputs));
   memset(&brw->vb.info, 0, sizeof(brw->vb.info));

   for (i = 0; i < VERT_ATTRIB_MAX; i++) {
      brw->vb.inputs[i].glarray = arrays[i];

      /* XXX: metaops passes null arrays */
      if (arrays[i]) {
	 if (arrays[i]->StrideB != 0)
	    brw->vb.info.varying |= 1 << i;

	 brw->vb.info.sizes[i/16] |= (inputs[i].glarray->Size - 1) << ((i%16) * 2);
      }
   }

   /* Raise statechanges if input sizes and varying have changed: 
    */
   if (memcmp(brw->vb.info.sizes, old.sizes, sizeof(old.sizes)) != 0)
      brw->state.dirty.brw |= BRW_NEW_INPUT_DIMENSIONS;

   if (brw->vb.info.varying != old.varying)
      brw->state.dirty.brw |= BRW_NEW_INPUT_VARYING;
}

/* XXX: could split the primitive list to fallback only on the
 * non-conformant primitives.
 */
static GLboolean check_fallbacks( struct brw_context *brw,
				  const struct _mesa_prim *prim,
				  GLuint nr_prims )
{
   GLuint i;

   if (!brw->intel.strict_conformance)
      return GL_FALSE;

   if (brw->attribs.Polygon->SmoothFlag) {
      for (i = 0; i < nr_prims; i++)
	 if (reduced_prim[prim[i].mode] == GL_TRIANGLES) 
	    return GL_TRUE;
   }

   /* BRW hardware will do AA lines, but they are non-conformant it
    * seems.  TBD whether we keep this fallback:
    */
   if (brw->attribs.Line->SmoothFlag) {
      for (i = 0; i < nr_prims; i++)
	 if (reduced_prim[prim[i].mode] == GL_LINES) 
	    return GL_TRUE;
   }

   /* Stipple -- these fallbacks could be resolved with a little
    * bit of work?
    */
   if (brw->attribs.Line->StippleFlag) {
      for (i = 0; i < nr_prims; i++) {
	 /* GS doesn't get enough information to know when to reset
	  * the stipple counter?!?
	  */
	 if (prim[i].mode == GL_LINE_LOOP) 
	    return GL_TRUE;
	    
	 if (prim[i].mode == GL_POLYGON &&
	     (brw->attribs.Polygon->FrontMode == GL_LINE ||
	      brw->attribs.Polygon->BackMode == GL_LINE))
	    return GL_TRUE;
      }
   }


   if (brw->attribs.Point->SmoothFlag) {
      for (i = 0; i < nr_prims; i++)
	 if (prim[i].mode == GL_POINTS) 
	    return GL_TRUE;
   }
      
   return GL_FALSE;
}

/* May fail if out of video memory for texture or vbo upload, or on
 * fallback conditions.
 */
static GLboolean brw_try_draw_prims( GLcontext *ctx,
				     const struct gl_client_array *arrays[],
				     const struct _mesa_prim *prim,
				     GLuint nr_prims,
				     const struct _mesa_index_buffer *ib,
				     GLuint min_index,
				     GLuint max_index )
{
   struct intel_context *intel = intel_context(ctx);
   struct brw_context *brw = brw_context(ctx);
   GLboolean retval = GL_FALSE;
   GLuint i;

   if (ctx->NewState)
      _mesa_update_state( ctx );

   brw_validate_textures( brw );

   /* Bind all inputs, derive varying and size information:
    */
   brw_merge_inputs( brw, arrays );
      
   /* Have to validate state quite late.  Will rebuild tnl_program,
    * which depends on varying information.  
    * 
    * Note this is where brw->vs->prog_data.inputs_read is calculated,
    * so can't access it earlier.
    */

   LOCK_HARDWARE(intel);

   if (brw->intel.numClipRects == 0) {
      UNLOCK_HARDWARE(intel);
      return GL_TRUE;
   }

   {
      /* Flush the batch if it's approaching full, so that we don't wrap while
       * we've got validated state that needs to be in the same batch as the
       * primitives.  This fraction is just a guess (minimal full state plus
       * a primitive is around 512 bytes), and would be better if we had
       * an upper bound of how much we might emit in a single
       * brw_try_draw_prims().
       */
      if (intel->batch->ptr - intel->batch->map > intel->batch->size * 3 / 4
	/* brw_emit_prim may change the cliprect_mode to LOOP_CLIPRECTS */
	|| intel->batch->cliprect_mode != LOOP_CLIPRECTS) 
	      intel_batchbuffer_flush(intel->batch);

      brw->no_batch_wrap = GL_TRUE;

      /* Set the first primitive early, ahead of validate_state:
       */
      brw_set_prim(brw, prim[0].mode);

      /* XXX:  Need to separate validate and upload of state.  
       */
      brw_validate_state( brw );

      /* Various fallback checks:
       */
      if (brw->intel.Fallback) 
	 goto out;

      if (check_fallbacks( brw, prim, nr_prims ))
	 goto out;
	  
      /* Upload index, vertex data: 
       */
      if (ib)
	 brw_upload_indices( brw, ib );

      if (!brw_upload_vertices( brw, min_index, max_index)) {
	 goto out;
      }

      for (i = 0; i < nr_prims; i++) {
	 brw_emit_prim(brw, &prim[i]);
      }

      retval = GL_TRUE;
   }

 out:

   brw->no_batch_wrap = GL_FALSE;

   UNLOCK_HARDWARE(intel);

   if (!retval)
      DBG("%s failed\n", __FUNCTION__);

   return retval;
}

static GLboolean brw_need_rebase( GLcontext *ctx,
				  const struct gl_client_array *arrays[],
				  const struct _mesa_index_buffer *ib,
				  GLuint min_index )
{
   if (min_index == 0) 
      return GL_FALSE;

   if (ib) {
      if (!vbo_all_varyings_in_vbos(arrays))
	 return GL_TRUE;
      else
	 return GL_FALSE;
   }
   else {
      /* Hmm.  This isn't quite what I wanted.  BRW can actually
       * handle the mixed case well enough that we shouldn't need to
       * rebase.  However, it's probably not very common, nor hugely
       * expensive to do it this way:
       */
      if (!vbo_all_varyings_in_vbos(arrays))
	 return GL_TRUE;
      else
	 return GL_FALSE;
   }
}
				  

void brw_draw_prims( GLcontext *ctx,
		     const struct gl_client_array *arrays[],
		     const struct _mesa_prim *prim,
		     GLuint nr_prims,
		     const struct _mesa_index_buffer *ib,
		     GLuint min_index,
		     GLuint max_index )
{
   GLboolean retval;

   /* Decide if we want to rebase.  If so we end up recursing once
    * only into this function.
    */
   if (brw_need_rebase( ctx, arrays, ib, min_index )) {
      vbo_rebase_prims( ctx, arrays, 
			prim, nr_prims, 
			ib, min_index, max_index, 
			brw_draw_prims );
      
      return;
   }


   /* Make a first attempt at drawing:
    */
   retval = brw_try_draw_prims(ctx, arrays, prim, nr_prims, ib, min_index, max_index);

   /* Otherwise, we really are out of memory.  Pass the drawing
    * command to the software tnl module and which will in turn call
    * swrast to do the drawing.
    */
   if (!retval) {
       _swsetup_Wakeup(ctx);
      _tnl_draw_prims(ctx, arrays, prim, nr_prims, ib, min_index, max_index);
   }
}

void brw_draw_init( struct brw_context *brw )
{
   GLcontext *ctx = &brw->intel.ctx;
   struct vbo_context *vbo = vbo_context(ctx);

   /* Register our drawing function: 
    */
   vbo->draw_prims = brw_draw_prims;
}

void brw_draw_destroy( struct brw_context *brw )
{
   if (brw->vb.upload.bo != NULL) {
      dri_bo_unreference(brw->vb.upload.bo);
      brw->vb.upload.bo = NULL;
   }
}