summaryrefslogtreecommitdiffstats
path: root/src/mesa/drivers/dri/i965/brw_draw.c
blob: 7b33b76e75a5db77c488b8bdd30bb1759afad335 (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
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
/**************************************************************************
 * 
 * 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 <sys/errno.h>

#include "main/glheader.h"
#include "main/context.h"
#include "main/condrender.h"
#include "main/samplerobj.h"
#include "main/state.h"
#include "main/enums.h"
#include "main/macros.h"
#include "main/transformfeedback.h"
#include "tnl/tnl.h"
#include "vbo/vbo_context.h"
#include "swrast/swrast.h"
#include "swrast_setup/swrast_setup.h"
#include "drivers/common/meta.h"

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

#include "intel_batchbuffer.h"
#include "intel_fbo.h"
#include "intel_mipmap_tree.h"
#include "intel_regions.h"

#define FILE_DEBUG_FLAG DEBUG_PRIMS

const GLuint prim_to_hw_prim[GL_TRIANGLE_STRIP_ADJACENCY+1] = {
   _3DPRIM_POINTLIST,
   _3DPRIM_LINELIST,
   _3DPRIM_LINELOOP,
   _3DPRIM_LINESTRIP,
   _3DPRIM_TRILIST,
   _3DPRIM_TRISTRIP,
   _3DPRIM_TRIFAN,
   _3DPRIM_QUADLIST,
   _3DPRIM_QUADSTRIP,
   _3DPRIM_POLYGON,
   _3DPRIM_LINELIST_ADJ,
   _3DPRIM_LINESTRIP_ADJ,
   _3DPRIM_TRILIST_ADJ,
   _3DPRIM_TRISTRIP_ADJ,
};


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 void brw_set_prim(struct brw_context *brw,
                         const struct _mesa_prim *prim)
{
   struct gl_context *ctx = &brw->ctx;
   uint32_t hw_prim = prim_to_hw_prim[prim->mode];

   DBG("PRIM: %s\n", _mesa_lookup_enum_by_nr(prim->mode));

   /* Slight optimization to avoid the GS program when not needed:
    */
   if (prim->mode == GL_QUAD_STRIP &&
       ctx->Light.ShadeModel != GL_FLAT &&
       ctx->Polygon.FrontMode == GL_FILL &&
       ctx->Polygon.BackMode == GL_FILL)
      hw_prim = _3DPRIM_TRISTRIP;

   if (prim->mode == GL_QUADS && prim->count == 4 &&
       ctx->Light.ShadeModel != GL_FLAT &&
       ctx->Polygon.FrontMode == GL_FILL &&
       ctx->Polygon.BackMode == GL_FILL) {
      hw_prim = _3DPRIM_TRIFAN;
   }

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

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

static void gen6_set_prim(struct brw_context *brw,
                          const struct _mesa_prim *prim)
{
   uint32_t hw_prim;

   DBG("PRIM: %s\n", _mesa_lookup_enum_by_nr(prim->mode));

   hw_prim = prim_to_hw_prim[prim->mode];

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


/**
 * The hardware is capable of removing dangling vertices on its own; however,
 * prior to Gen6, we sometimes convert quads into trifans (and quad strips
 * into tristrips), since pre-Gen6 hardware requires a GS to render quads.
 * This function manually trims dangling vertices from a draw call involving
 * quads so that those dangling vertices won't get drawn when we convert to
 * trifans/tristrips.
 */
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,
			  uint32_t hw_prim)
{
   int verts_per_instance;
   int vertex_access_type;
   int start_vertex_location;
   int base_vertex_location;

   DBG("PRIM: %s %d %d\n", _mesa_lookup_enum_by_nr(prim->mode),
       prim->start, prim->count);

   start_vertex_location = prim->start;
   base_vertex_location = prim->basevertex;
   if (prim->indexed) {
      vertex_access_type = brw->gen >= 7 ?
         GEN7_3DPRIM_VERTEXBUFFER_ACCESS_RANDOM :
         GEN4_3DPRIM_VERTEXBUFFER_ACCESS_RANDOM;
      start_vertex_location += brw->ib.start_vertex_offset;
      base_vertex_location += brw->vb.start_vertex_bias;
   } else {
      vertex_access_type = brw->gen >= 7 ?
         GEN7_3DPRIM_VERTEXBUFFER_ACCESS_SEQUENTIAL :
         GEN4_3DPRIM_VERTEXBUFFER_ACCESS_SEQUENTIAL;
      start_vertex_location += brw->vb.start_vertex_bias;
   }

   /* We only need to trim the primitive count on pre-Gen6. */
   if (brw->gen < 6)
      verts_per_instance = trim(prim->mode, prim->count);
   else
      verts_per_instance = prim->count;

   /* If nothing to emit, just return. */
   if (verts_per_instance == 0)
      return;

   /* If we're set to always flush, do it before and after the primitive emit.
    * We want to catch both missed flushes that hurt instruction/state cache
    * and missed flushes of the render cache as it heads to other parts of
    * the besides the draw code.
    */
   if (brw->always_flush_cache) {
      intel_batchbuffer_emit_mi_flush(brw);
   }

   if (brw->gen >= 7) {
      BEGIN_BATCH(7);
      OUT_BATCH(CMD_3D_PRIM << 16 | (7 - 2));
      OUT_BATCH(hw_prim | vertex_access_type);
   } else {
      BEGIN_BATCH(6);
      OUT_BATCH(CMD_3D_PRIM << 16 | (6 - 2) |
                hw_prim << GEN4_3DPRIM_TOPOLOGY_TYPE_SHIFT |
                vertex_access_type);
   }
   OUT_BATCH(verts_per_instance);
   OUT_BATCH(start_vertex_location);
   OUT_BATCH(prim->num_instances);
   OUT_BATCH(prim->base_instance);
   OUT_BATCH(base_vertex_location);
   ADVANCE_BATCH();

   /* Only used on Sandybridge; harmless to set elsewhere. */
   brw->batch.need_workaround_flush = true;

   if (brw->always_flush_cache) {
      intel_batchbuffer_emit_mi_flush(brw);
   }
}


static void brw_merge_inputs( struct brw_context *brw,
		       const struct gl_client_array *arrays[])
{
   GLuint i;

   for (i = 0; i < brw->vb.nr_buffers; i++) {
      drm_intel_bo_unreference(brw->vb.buffers[i].bo);
      brw->vb.buffers[i].bo = NULL;
   }
   brw->vb.nr_buffers = 0;

   for (i = 0; i < VERT_ATTRIB_MAX; i++) {
      brw->vb.inputs[i].buffer = -1;
      brw->vb.inputs[i].glarray = arrays[i];
      brw->vb.inputs[i].attrib = (gl_vert_attrib) i;
   }
}

/*
 * \brief Resolve buffers before drawing.
 *
 * Resolve the depth buffer's HiZ buffer and resolve the depth buffer of each
 * enabled depth texture.
 *
 * (In the future, this will also perform MSAA resolves).
 */
static void
brw_predraw_resolve_buffers(struct brw_context *brw)
{
   struct gl_context *ctx = &brw->ctx;
   struct intel_renderbuffer *depth_irb;
   struct intel_texture_object *tex_obj;

   /* Resolve the depth buffer's HiZ buffer. */
   depth_irb = intel_get_renderbuffer(ctx->DrawBuffer, BUFFER_DEPTH);
   if (depth_irb)
      intel_renderbuffer_resolve_hiz(brw, depth_irb);

   /* Resolve depth buffer of each enabled depth texture, and color buffer of
    * each fast-clear-enabled color texture.
    */
   for (int i = 0; i < BRW_MAX_TEX_UNIT; i++) {
      if (!ctx->Texture.Unit[i]._ReallyEnabled)
	 continue;
      tex_obj = intel_texture_object(ctx->Texture.Unit[i]._Current);
      if (!tex_obj || !tex_obj->mt)
	 continue;
      intel_miptree_all_slices_resolve_depth(brw, tex_obj->mt);
      intel_miptree_resolve_color(brw, tex_obj->mt);
   }
}

/**
 * \brief Call this after drawing to mark which buffers need resolving
 *
 * If the depth buffer was written to and if it has an accompanying HiZ
 * buffer, then mark that it needs a depth resolve.
 *
 * If the color buffer is a multisample window system buffer, then
 * mark that it needs a downsample.
 */
static void brw_postdraw_set_buffers_need_resolve(struct brw_context *brw)
{
   struct gl_context *ctx = &brw->ctx;
   struct gl_framebuffer *fb = ctx->DrawBuffer;

   struct intel_renderbuffer *front_irb = NULL;
   struct intel_renderbuffer *back_irb = intel_get_renderbuffer(fb, BUFFER_BACK_LEFT);
   struct intel_renderbuffer *depth_irb = intel_get_renderbuffer(fb, BUFFER_DEPTH);
   struct gl_renderbuffer_attachment *depth_att = &fb->Attachment[BUFFER_DEPTH];

   if (brw->is_front_buffer_rendering)
      front_irb = intel_get_renderbuffer(fb, BUFFER_FRONT_LEFT);

   if (front_irb)
      intel_renderbuffer_set_needs_downsample(front_irb);
   if (back_irb)
      intel_renderbuffer_set_needs_downsample(back_irb);
   if (depth_irb && ctx->Depth.Mask)
      intel_renderbuffer_att_set_needs_depth_resolve(depth_att);
}

/* May fail if out of video memory for texture or vbo upload, or on
 * fallback conditions.
 */
static bool brw_try_draw_prims( struct gl_context *ctx,
				     const struct gl_client_array *arrays[],
				     const struct _mesa_prim *prims,
				     GLuint nr_prims,
				     const struct _mesa_index_buffer *ib,
				     GLuint min_index,
				     GLuint max_index )
{
   struct brw_context *brw = brw_context(ctx);
   bool retval = true;
   GLuint i;
   bool fail_next = false;

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

   /* Find the highest sampler unit used by each shader program.  A bit-count
    * won't work since ARB programs use the texture unit number as the sampler
    * index.
    */
   brw->wm.base.sampler_count =
      _mesa_fls(ctx->FragmentProgram._Current->Base.SamplersUsed);
   brw->gs.base.sampler_count = ctx->GeometryProgram._Current ?
      _mesa_fls(ctx->GeometryProgram._Current->Base.SamplersUsed) : 0;
   brw->vs.base.sampler_count =
      _mesa_fls(ctx->VertexProgram._Current->Base.SamplersUsed);

   /* We have to validate the textures *before* checking for fallbacks;
    * otherwise, the software fallback won't be able to rely on the
    * texture state, the firstLevel and lastLevel fields won't be
    * set in the intel texture object (they'll both be 0), and the 
    * software fallback will segfault if it attempts to access any
    * texture level other than level 0.
    */
   brw_validate_textures( brw );

   intel_prepare_render(brw);

   /* This workaround has to happen outside of brw_upload_state() because it
    * may flush the batchbuffer for a blit, affecting the state flags.
    */
   brw_workaround_depthstencil_alignment(brw, 0);

   /* Resolves must occur after updating renderbuffers, updating context state,
    * and finalizing textures but before setting up any hardware state for
    * this draw call.
    */
   brw_predraw_resolve_buffers(brw);

   /* Bind all inputs, derive varying and size information:
    */
   brw_merge_inputs( brw, arrays );

   brw->ib.ib = ib;
   brw->state.dirty.brw |= BRW_NEW_INDICES;

   brw->vb.min_index = min_index;
   brw->vb.max_index = max_index;
   brw->state.dirty.brw |= BRW_NEW_VERTICES;

   for (i = 0; i < nr_prims; i++) {
      int estimated_max_prim_size;

      estimated_max_prim_size = 512; /* batchbuffer commands */
      estimated_max_prim_size += (BRW_MAX_TEX_UNIT *
				  (sizeof(struct brw_sampler_state) +
				   sizeof(struct gen5_sampler_default_color)));
      estimated_max_prim_size += 1024; /* gen6 VS push constants */
      estimated_max_prim_size += 1024; /* gen6 WM push constants */
      estimated_max_prim_size += 512; /* misc. pad */

      /* 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.
       */
      intel_batchbuffer_require_space(brw, estimated_max_prim_size, false);
      intel_batchbuffer_save_state(brw);

      if (brw->num_instances != prims[i].num_instances) {
         brw->num_instances = prims[i].num_instances;
         brw->state.dirty.brw |= BRW_NEW_VERTICES;
         brw_merge_inputs(brw, arrays);
      }
      if (brw->basevertex != prims[i].basevertex) {
         brw->basevertex = prims[i].basevertex;
         brw->state.dirty.brw |= BRW_NEW_VERTICES;
         brw_merge_inputs(brw, arrays);
      }
      if (brw->gen < 6)
	 brw_set_prim(brw, &prims[i]);
      else
	 gen6_set_prim(brw, &prims[i]);

retry:
      /* Note that before the loop, brw->state.dirty.brw was set to != 0, and
       * that the state updated in the loop outside of this block is that in
       * *_set_prim or intel_batchbuffer_flush(), which only impacts
       * brw->state.dirty.brw.
       */
      if (brw->state.dirty.brw) {
	 brw->no_batch_wrap = true;
	 brw_upload_state(brw);
      }

      brw_emit_prim(brw, &prims[i], brw->primitive);

      brw->no_batch_wrap = false;

      if (dri_bufmgr_check_aperture_space(&brw->batch.bo, 1)) {
	 if (!fail_next) {
	    intel_batchbuffer_reset_to_saved(brw);
	    intel_batchbuffer_flush(brw);
	    fail_next = true;
	    goto retry;
	 } else {
	    if (intel_batchbuffer_flush(brw) == -ENOSPC) {
	       static bool warned = false;

	       if (!warned) {
		  fprintf(stderr, "i965: Single primitive emit exceeded"
			  "available aperture space\n");
		  warned = true;
	       }

	       retval = false;
	    }
	 }
      }
   }

   if (brw->always_flush_batch)
      intel_batchbuffer_flush(brw);

   brw_state_cache_check_size(brw);
   brw_postdraw_set_buffers_need_resolve(brw);

   return retval;
}

void brw_draw_prims( struct gl_context *ctx,
		     const struct _mesa_prim *prims,
		     GLuint nr_prims,
		     const struct _mesa_index_buffer *ib,
		     GLboolean index_bounds_valid,
		     GLuint min_index,
		     GLuint max_index,
		     struct gl_transform_feedback_object *unused_tfb_object)
{
   struct brw_context *brw = brw_context(ctx);
   const struct gl_client_array **arrays = ctx->Array._DrawArrays;

   assert(unused_tfb_object == NULL);

   if (!_mesa_check_conditional_render(ctx))
      return;

   /* Handle primitive restart if needed */
   if (brw_handle_primitive_restart(ctx, prims, nr_prims, ib)) {
      /* The draw was handled, so we can exit now */
      return;
   }

   /* If we're going to have to upload any of the user's vertex arrays, then
    * get the minimum and maximum of their index buffer so we know what range
    * to upload.
    */
   if (!vbo_all_varyings_in_vbos(arrays) && !index_bounds_valid) {
      perf_debug("Scanning index buffer to compute index buffer bounds.  "
                 "Use glDrawRangeElements() to avoid this.\n");
      vbo_get_minmax_indices(ctx, prims, ib, &min_index, &max_index, nr_prims);
   }

   /* Do GL_SELECT and GL_FEEDBACK rendering using swrast, even though it
    * won't support all the extensions we support.
    */
   if (ctx->RenderMode != GL_RENDER) {
      perf_debug("%s render mode not supported in hardware\n",
                 _mesa_lookup_enum_by_nr(ctx->RenderMode));
      _swsetup_Wakeup(ctx);
      _tnl_wakeup(ctx);
      _tnl_draw_prims(ctx, arrays, prims, nr_prims, ib, min_index, max_index);
      return;
   }

   /* Try drawing with the hardware, but don't do anything else if we can't
    * manage it.  swrast doesn't support our featureset, so we can't fall back
    * to it.
    */
   brw_try_draw_prims(ctx, arrays, prims, nr_prims, ib, min_index, max_index);
}

void brw_draw_init( struct brw_context *brw )
{
   struct gl_context *ctx = &brw->ctx;
   struct vbo_context *vbo = vbo_context(ctx);
   int i;

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

   for (i = 0; i < VERT_ATTRIB_MAX; i++)
      brw->vb.inputs[i].buffer = -1;
   brw->vb.nr_buffers = 0;
   brw->vb.nr_enabled = 0;
}

void brw_draw_destroy( struct brw_context *brw )
{
   int i;

   for (i = 0; i < brw->vb.nr_buffers; i++) {
      drm_intel_bo_unreference(brw->vb.buffers[i].bo);
      brw->vb.buffers[i].bo = NULL;
   }
   brw->vb.nr_buffers = 0;

   for (i = 0; i < brw->vb.nr_enabled; i++) {
      brw->vb.enabled[i]->buffer = -1;
   }
   brw->vb.nr_enabled = 0;

   drm_intel_bo_unreference(brw->ib.bo);
   brw->ib.bo = NULL;
}