summaryrefslogtreecommitdiffstats
path: root/src/vulkan/pipeline.c
blob: 60389e4bbba6aa106a1f3a94cdc5d6ce38eaf8cb (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
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
/*
 * Copyright © 2015 Intel Corporation
 *
 * 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, sublicense,
 * 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 NONINFRINGEMENT.  IN NO EVENT SHALL
 * THE AUTHORS OR COPYRIGHT HOLDERS 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 <assert.h>
#include <stdbool.h>
#include <string.h>
#include <unistd.h>
#include <fcntl.h>

#include "private.h"

// Shader functions

VkResult VKAPI vkCreateShader(
    VkDevice                                    _device,
    const VkShaderCreateInfo*                   pCreateInfo,
    VkShader*                                   pShader)
{
   struct anv_device *device = (struct anv_device *) _device;
   struct anv_shader *shader;

   assert(pCreateInfo->sType == VK_STRUCTURE_TYPE_SHADER_CREATE_INFO);

   shader = anv_device_alloc(device, sizeof(*shader) + pCreateInfo->codeSize, 8,
                               VK_SYSTEM_ALLOC_TYPE_API_OBJECT);
   if (shader == NULL)
      return vk_error(VK_ERROR_OUT_OF_HOST_MEMORY);

   shader->size = pCreateInfo->codeSize;
   memcpy(shader->data, pCreateInfo->pCode, shader->size);

   *pShader = (VkShader) shader;

   return VK_SUCCESS;
}

// Pipeline functions

static void
emit_vertex_input(struct anv_pipeline *pipeline, VkPipelineVertexInputCreateInfo *info)
{
   const uint32_t num_dwords = 1 + info->attributeCount * 2;
   uint32_t *p;
   bool instancing_enable[32];
   
   for (uint32_t i = 0; i < info->bindingCount; i++) {
      const VkVertexInputBindingDescription *desc =
         &info->pVertexBindingDescriptions[i];
      
      pipeline->binding_stride[desc->binding] = desc->strideInBytes;

      /* Step rate is programmed per vertex element (attribute), not
       * binding. Set up a map of which bindings step per instance, for
       * reference by vertex element setup. */
      switch (desc->stepRate) {
      default:
      case VK_VERTEX_INPUT_STEP_RATE_VERTEX:
         instancing_enable[desc->binding] = false;
         break;
      case VK_VERTEX_INPUT_STEP_RATE_INSTANCE:
         instancing_enable[desc->binding] = true;
         break;
      }
   }

   p = anv_batch_emitn(&pipeline->batch, num_dwords,
                       GEN8_3DSTATE_VERTEX_ELEMENTS);

   for (uint32_t i = 0; i < info->attributeCount; i++) {
      const VkVertexInputAttributeDescription *desc =
         &info->pVertexAttributeDescriptions[i];
      const struct anv_format *format = anv_format_for_vk_format(desc->format);

      struct GEN8_VERTEX_ELEMENT_STATE element = {
         .VertexBufferIndex = desc->binding,
         .Valid = true,
         .SourceElementFormat = format->format,
         .EdgeFlagEnable = false,
         .SourceElementOffset = desc->offsetInBytes,
         .Component0Control = VFCOMP_STORE_SRC,
         .Component1Control = format->channels >= 2 ? VFCOMP_STORE_SRC : VFCOMP_STORE_0,
         .Component2Control = format->channels >= 3 ? VFCOMP_STORE_SRC : VFCOMP_STORE_0,
         .Component3Control = format->channels >= 4 ? VFCOMP_STORE_SRC : VFCOMP_STORE_1_FP
      };
      GEN8_VERTEX_ELEMENT_STATE_pack(NULL, &p[1 + i * 2], &element);

      anv_batch_emit(&pipeline->batch, GEN8_3DSTATE_VF_INSTANCING,
                     .InstancingEnable = instancing_enable[desc->binding],
                     .VertexElementIndex = i,
                     /* Vulkan so far doesn't have an instance divisor, so
                      * this is always 1 (ignored if not instancing). */
                     .InstanceDataStepRate = 1);
   }

   anv_batch_emit(&pipeline->batch, GEN8_3DSTATE_VF_SGVS,
                  .VertexIDEnable = pipeline->vs_prog_data.uses_vertexid,
                  .VertexIDComponentNumber = 2,
                  .VertexIDElementOffset = info->bindingCount,
                  .InstanceIDEnable = pipeline->vs_prog_data.uses_instanceid,
                  .InstanceIDComponentNumber = 3,
                  .InstanceIDElementOffset = info->bindingCount);
}

static void
emit_ia_state(struct anv_pipeline *pipeline,
              VkPipelineIaStateCreateInfo *info,
              const struct anv_pipeline_create_info *extra)
{
   static const uint32_t vk_to_gen_primitive_type[] = {
      [VK_PRIMITIVE_TOPOLOGY_POINT_LIST] = _3DPRIM_POINTLIST,
      [VK_PRIMITIVE_TOPOLOGY_LINE_LIST] = _3DPRIM_LINELIST,
      [VK_PRIMITIVE_TOPOLOGY_LINE_STRIP] = _3DPRIM_LINESTRIP,
      [VK_PRIMITIVE_TOPOLOGY_TRIANGLE_LIST] = _3DPRIM_TRILIST,
      [VK_PRIMITIVE_TOPOLOGY_TRIANGLE_STRIP] = _3DPRIM_TRISTRIP,
      [VK_PRIMITIVE_TOPOLOGY_TRIANGLE_FAN] = _3DPRIM_TRIFAN,
      [VK_PRIMITIVE_TOPOLOGY_LINE_LIST_ADJ] = _3DPRIM_LINELIST_ADJ,
      [VK_PRIMITIVE_TOPOLOGY_LINE_STRIP_ADJ] = _3DPRIM_LISTSTRIP_ADJ,
      [VK_PRIMITIVE_TOPOLOGY_TRIANGLE_LIST_ADJ] = _3DPRIM_TRILIST_ADJ,
      [VK_PRIMITIVE_TOPOLOGY_TRIANGLE_STRIP_ADJ] = _3DPRIM_TRISTRIP_ADJ,
      [VK_PRIMITIVE_TOPOLOGY_PATCH] = _3DPRIM_PATCHLIST_1
   };
   uint32_t topology = vk_to_gen_primitive_type[info->topology];

   if (extra && extra->use_rectlist)
      topology = _3DPRIM_RECTLIST;

   anv_batch_emit(&pipeline->batch, GEN8_3DSTATE_VF,
                  .IndexedDrawCutIndexEnable = info->primitiveRestartEnable,
                  .CutIndex = info->primitiveRestartIndex);
   anv_batch_emit(&pipeline->batch, GEN8_3DSTATE_VF_TOPOLOGY,
                  .PrimitiveTopologyType = topology);
}

static void
emit_rs_state(struct anv_pipeline *pipeline, VkPipelineRsStateCreateInfo *info,
              const struct anv_pipeline_create_info *extra)

{
   static const uint32_t vk_to_gen_cullmode[] = {
      [VK_CULL_MODE_NONE] = CULLMODE_NONE,
      [VK_CULL_MODE_FRONT] = CULLMODE_FRONT,
      [VK_CULL_MODE_BACK] = CULLMODE_BACK,
      [VK_CULL_MODE_FRONT_AND_BACK] = CULLMODE_BOTH
   };

   static const uint32_t vk_to_gen_fillmode[] = {
      [VK_FILL_MODE_POINTS] = RASTER_POINT,
      [VK_FILL_MODE_WIREFRAME] = RASTER_WIREFRAME,
      [VK_FILL_MODE_SOLID] = RASTER_SOLID
   };

   static const uint32_t vk_to_gen_front_face[] = {
      [VK_FRONT_FACE_CCW] = CounterClockwise,
      [VK_FRONT_FACE_CW] = Clockwise
   };
   
   static const uint32_t vk_to_gen_coordinate_origin[] = {
      [VK_COORDINATE_ORIGIN_UPPER_LEFT] = UPPERLEFT,
      [VK_COORDINATE_ORIGIN_LOWER_LEFT] = LOWERLEFT
   };

   struct GEN8_3DSTATE_SF sf = {
      GEN8_3DSTATE_SF_header,
      .ViewportTransformEnable = !(extra && extra->disable_viewport),
      .TriangleStripListProvokingVertexSelect =
         info->provokingVertex == VK_PROVOKING_VERTEX_FIRST ? 0 : 2,
      .LineStripListProvokingVertexSelect =
         info->provokingVertex == VK_PROVOKING_VERTEX_FIRST ? 0 : 1,
      .TriangleFanProvokingVertexSelect =
         info->provokingVertex == VK_PROVOKING_VERTEX_FIRST ? 0 : 2,
      .PointWidthSource = info->programPointSize ? Vertex : State,
   };

   /* bool32_t                                    rasterizerDiscardEnable; */


   GEN8_3DSTATE_SF_pack(NULL, pipeline->state_sf, &sf);

   anv_batch_emit(&pipeline->batch, GEN8_3DSTATE_RASTER,
                  .FrontWinding = vk_to_gen_front_face[info->frontFace],
                  .CullMode = vk_to_gen_cullmode[info->cullMode],
                  .FrontFaceFillMode = vk_to_gen_fillmode[info->fillMode],
                  .BackFaceFillMode = vk_to_gen_fillmode[info->fillMode],
                  .ScissorRectangleEnable = !(extra && extra->disable_scissor),
                  .ViewportZClipTestEnable = info->depthClipEnable);

   anv_batch_emit(&pipeline->batch, GEN8_3DSTATE_SBE,
                  .ForceVertexURBEntryReadLength = false,
                  .ForceVertexURBEntryReadOffset = false,
                  .PointSpriteTextureCoordinateOrigin =
                     vk_to_gen_coordinate_origin[info->pointOrigin],
                  .NumberofSFOutputAttributes =
                     pipeline->wm_prog_data.num_varying_inputs);

}

VkResult VKAPI vkCreateGraphicsPipeline(
    VkDevice                                    device,
    const VkGraphicsPipelineCreateInfo*         pCreateInfo,
    VkPipeline*                                 pPipeline)
{
   return anv_pipeline_create(device, pCreateInfo, NULL, pPipeline);
}


VkResult
anv_pipeline_create(
    VkDevice                                    _device,
    const VkGraphicsPipelineCreateInfo*         pCreateInfo,
    const struct anv_pipeline_create_info *     extra,
    VkPipeline*                                 pPipeline)
{
   struct anv_device *device = (struct anv_device *) _device;
   struct anv_pipeline *pipeline;
   const struct anv_common *common;
   VkPipelineShaderStageCreateInfo *shader_create_info;
   VkPipelineIaStateCreateInfo *ia_info;
   VkPipelineRsStateCreateInfo *rs_info;
   VkPipelineVertexInputCreateInfo *vi_info;
   VkResult result;
   uint32_t offset, length;

   assert(pCreateInfo->sType == VK_STRUCTURE_TYPE_GRAPHICS_PIPELINE_CREATE_INFO);
   
   pipeline = anv_device_alloc(device, sizeof(*pipeline), 8,
                               VK_SYSTEM_ALLOC_TYPE_API_OBJECT);
   if (pipeline == NULL)
      return vk_error(VK_ERROR_OUT_OF_HOST_MEMORY);

   pipeline->device = device;
   pipeline->layout = (struct anv_pipeline_layout *) pCreateInfo->layout;
   memset(pipeline->shaders, 0, sizeof(pipeline->shaders));
   result = anv_batch_init(&pipeline->batch, device);
   if (result != VK_SUCCESS)
      goto fail;

   for (common = pCreateInfo->pNext; common; common = common->pNext) {
      switch (common->sType) {
      case VK_STRUCTURE_TYPE_PIPELINE_VERTEX_INPUT_CREATE_INFO:
         vi_info = (VkPipelineVertexInputCreateInfo *) common;
         break;
      case VK_STRUCTURE_TYPE_PIPELINE_IA_STATE_CREATE_INFO:
         ia_info = (VkPipelineIaStateCreateInfo *) common;
         break;
      case VK_STRUCTURE_TYPE_PIPELINE_TESS_STATE_CREATE_INFO:
      case VK_STRUCTURE_TYPE_PIPELINE_VP_STATE_CREATE_INFO:
         break;
      case VK_STRUCTURE_TYPE_PIPELINE_RS_STATE_CREATE_INFO:
         rs_info = (VkPipelineRsStateCreateInfo *) common;
         break;
      case VK_STRUCTURE_TYPE_PIPELINE_MS_STATE_CREATE_INFO:
      case VK_STRUCTURE_TYPE_PIPELINE_CB_STATE_CREATE_INFO:
      case VK_STRUCTURE_TYPE_PIPELINE_DS_STATE_CREATE_INFO:
      case VK_STRUCTURE_TYPE_PIPELINE_SHADER_STAGE_CREATE_INFO:
         shader_create_info = (VkPipelineShaderStageCreateInfo *) common;
         pipeline->shaders[shader_create_info->shader.stage] =
            (struct anv_shader *) shader_create_info->shader.shader;
         break;
      default:
         break;
      }
   }

   pipeline->use_repclear = extra && extra->use_repclear;

   anv_compiler_run(device->compiler, pipeline);

   /* FIXME: The compiler dead-codes FS inputs when we don't have a VS, so we
    * hard code this to num_attributes - 2. This is because the attributes
    * include VUE header and position, which aren't counted as varying
    * inputs. */
   if (pipeline->vs_simd8 == NO_KERNEL)
      pipeline->wm_prog_data.num_varying_inputs = vi_info->attributeCount - 2;

   emit_vertex_input(pipeline, vi_info);
   emit_ia_state(pipeline, ia_info, extra);
   emit_rs_state(pipeline, rs_info, extra);

   anv_batch_emit(&pipeline->batch, GEN8_3DSTATE_CLIP,
                  .ClipEnable = true,
                  .ViewportXYClipTestEnable = !(extra && extra->disable_viewport));

   anv_batch_emit(&pipeline->batch, GEN8_3DSTATE_WM,
                  .StatisticsEnable = true,
                  .LineEndCapAntialiasingRegionWidth = _05pixels,
                  .LineAntialiasingRegionWidth = _10pixels,
                  .EarlyDepthStencilControl = NORMAL,
                  .ForceThreadDispatchEnable = NORMAL,
                  .PointRasterizationRule = RASTRULE_UPPER_RIGHT,
                  .BarycentricInterpolationMode =
                     pipeline->wm_prog_data.barycentric_interp_modes);

   uint32_t samples = 1;
   uint32_t log2_samples = __builtin_ffs(samples) - 1;
   bool enable_sampling = samples > 1 ? true : false;

   anv_batch_emit(&pipeline->batch, GEN8_3DSTATE_MULTISAMPLE,
                  .PixelPositionOffsetEnable = enable_sampling,
                  .PixelLocation = CENTER,
                  .NumberofMultisamples = log2_samples);

   anv_batch_emit(&pipeline->batch, GEN8_3DSTATE_URB_VS,
                  .VSURBStartingAddress = pipeline->urb.vs_start,
                  .VSURBEntryAllocationSize = pipeline->urb.vs_size - 1,
                  .VSNumberofURBEntries = pipeline->urb.nr_vs_entries);

   anv_batch_emit(&pipeline->batch, GEN8_3DSTATE_URB_GS,
                  .GSURBStartingAddress = pipeline->urb.gs_start,
                  .GSURBEntryAllocationSize = pipeline->urb.gs_size - 1,
                  .GSNumberofURBEntries = pipeline->urb.nr_gs_entries);

   anv_batch_emit(&pipeline->batch, GEN8_3DSTATE_URB_HS,
                  .HSURBStartingAddress = pipeline->urb.vs_start,
                  .HSURBEntryAllocationSize = 0,
                  .HSNumberofURBEntries = 0);

   anv_batch_emit(&pipeline->batch, GEN8_3DSTATE_URB_DS,
                  .DSURBStartingAddress = pipeline->urb.vs_start,
                  .DSURBEntryAllocationSize = 0,
                  .DSNumberofURBEntries = 0);

   const struct brw_gs_prog_data *gs_prog_data = &pipeline->gs_prog_data;
   offset = 1;
   length = (gs_prog_data->base.vue_map.num_slots + 1) / 2 - offset;

   if (pipeline->gs_vec4 == NO_KERNEL)
      anv_batch_emit(&pipeline->batch, GEN8_3DSTATE_GS, .Enable = false);
   else
      anv_batch_emit(&pipeline->batch, GEN8_3DSTATE_GS,
                     .SingleProgramFlow = false,
                     .KernelStartPointer = pipeline->gs_vec4,
                     .VectorMaskEnable = Vmask,
                     .SamplerCount = 0,
                     .BindingTableEntryCount = 0,
                     .ExpectedVertexCount = pipeline->gs_vertex_count,
                        
                     .PerThreadScratchSpace = 0,
                     .ScratchSpaceBasePointer = 0,

                     .OutputVertexSize = gs_prog_data->output_vertex_size_hwords * 2 - 1,
                     .OutputTopology = gs_prog_data->output_topology,
                     .VertexURBEntryReadLength = gs_prog_data->base.urb_read_length,
                     .DispatchGRFStartRegisterForURBData =
                        gs_prog_data->base.base.dispatch_grf_start_reg,

                     .MaximumNumberofThreads = device->info.max_gs_threads,
                     .ControlDataHeaderSize = gs_prog_data->control_data_header_size_hwords,
                     //pipeline->gs_prog_data.dispatch_mode |
                     .StatisticsEnable = true,
                     .IncludePrimitiveID = gs_prog_data->include_primitive_id,
                     .ReorderMode = TRAILING,
                     .Enable = true,

                     .ControlDataFormat = gs_prog_data->control_data_format,

                     /* FIXME: mesa sets this based on ctx->Transform.ClipPlanesEnabled:
                      * UserClipDistanceClipTestEnableBitmask_3DSTATE_GS(v)
                      * UserClipDistanceCullTestEnableBitmask(v)
                      */

                     .VertexURBEntryOutputReadOffset = offset,
                     .VertexURBEntryOutputLength = length);

   //trp_generate_blend_hw_cmds(batch, pipeline);

   const struct brw_vue_prog_data *vue_prog_data = &pipeline->vs_prog_data.base;
   /* Skip the VUE header and position slots */
   offset = 1;
   length = (vue_prog_data->vue_map.num_slots + 1) / 2 - offset;

   if (pipeline->vs_simd8 == NO_KERNEL || (extra && extra->disable_vs))
      anv_batch_emit(&pipeline->batch, GEN8_3DSTATE_VS,
                     .FunctionEnable = false,
                     .VertexURBEntryOutputReadOffset = 1,
                     /* Even if VS is disabled, SBE still gets the amount of
                      * vertex data to read from this field. We use attribute
                      * count - 1, as we don't count the VUE header here. */
                     .VertexURBEntryOutputLength =
                        DIV_ROUND_UP(vi_info->attributeCount - 1, 2));
   else
      anv_batch_emit(&pipeline->batch, GEN8_3DSTATE_VS,
                     .KernelStartPointer = pipeline->vs_simd8,
                     .SingleVertexDispatch = Multiple,
                     .VectorMaskEnable = Dmask,
                     .SamplerCount = 0,
                     .BindingTableEntryCount =
                     vue_prog_data->base.binding_table.size_bytes / 4,
                     .ThreadDispatchPriority = Normal,
                     .FloatingPointMode = IEEE754,
                     .IllegalOpcodeExceptionEnable = false,
                     .AccessesUAV = false,
                     .SoftwareExceptionEnable = false,

                     /* FIXME: pointer needs to be assigned outside as it aliases
                      * PerThreadScratchSpace.
                      */
                     .ScratchSpaceBasePointer = 0,
                     .PerThreadScratchSpace = 0,

                     .DispatchGRFStartRegisterForURBData =
                     vue_prog_data->base.dispatch_grf_start_reg,
                     .VertexURBEntryReadLength = vue_prog_data->urb_read_length,
                     .VertexURBEntryReadOffset = 0,

                     .MaximumNumberofThreads = device->info.max_vs_threads - 1,
                     .StatisticsEnable = false,
                     .SIMD8DispatchEnable = true,
                     .VertexCacheDisable = ia_info->disableVertexReuse,
                     .FunctionEnable = true,

                     .VertexURBEntryOutputReadOffset = offset,
                     .VertexURBEntryOutputLength = length,
                     .UserClipDistanceClipTestEnableBitmask = 0,
                     .UserClipDistanceCullTestEnableBitmask = 0);

   const struct brw_wm_prog_data *wm_prog_data = &pipeline->wm_prog_data;
   uint32_t ksp0, ksp2, grf_start0, grf_start2;

   ksp2 = 0;
   grf_start2 = 0;
   if (pipeline->ps_simd8 != NO_KERNEL) {
      ksp0 = pipeline->ps_simd8;
      grf_start0 = wm_prog_data->base.dispatch_grf_start_reg;
      if (pipeline->ps_simd16 != NO_KERNEL) {
         ksp2 = pipeline->ps_simd16;
         grf_start2 = wm_prog_data->dispatch_grf_start_reg_16;
      }
   } else if (pipeline->ps_simd16 != NO_KERNEL) {
      ksp0 = pipeline->ps_simd16;
      grf_start0 = wm_prog_data->dispatch_grf_start_reg_16;
   } else {
      unreachable("no ps shader");
   }

   anv_batch_emit(&pipeline->batch, GEN8_3DSTATE_PS,
                  .KernelStartPointer0 = ksp0,
   
                  .SingleProgramFlow = false,
                  .VectorMaskEnable = true,
                  .SamplerCount = 1,

                  .ScratchSpaceBasePointer = 0,
                  .PerThreadScratchSpace = 0,
                  
                  .MaximumNumberofThreadsPerPSD = 64 - 2,
                  .PositionXYOffsetSelect = wm_prog_data->uses_pos_offset ?
                     POSOFFSET_SAMPLE: POSOFFSET_NONE,
                  .PushConstantEnable = wm_prog_data->base.nr_params > 0,
                  ._8PixelDispatchEnable = pipeline->ps_simd8 != NO_KERNEL,
                  ._16PixelDispatchEnable = pipeline->ps_simd16 != NO_KERNEL,
                  ._32PixelDispatchEnable = false,

                  .DispatchGRFStartRegisterForConstantSetupData0 = grf_start0,
                  .DispatchGRFStartRegisterForConstantSetupData1 = 0,
                  .DispatchGRFStartRegisterForConstantSetupData2 = grf_start2,

                  .KernelStartPointer1 = 0,
                  .KernelStartPointer2 = ksp2);

   bool per_sample_ps = false;
   anv_batch_emit(&pipeline->batch, GEN8_3DSTATE_PS_EXTRA,
                  .PixelShaderValid = true,
                  .PixelShaderKillsPixel = wm_prog_data->uses_kill,
                  .PixelShaderComputedDepthMode = wm_prog_data->computed_depth_mode,
                  .AttributeEnable = wm_prog_data->num_varying_inputs > 0,
                  .oMaskPresenttoRenderTarget = wm_prog_data->uses_omask,
                  .PixelShaderIsPerSample = per_sample_ps);

   *pPipeline = (VkPipeline) pipeline;

   return VK_SUCCESS;

 fail:
   anv_device_free(device, pipeline);
   
   return result;
}

VkResult
anv_pipeline_destroy(struct anv_pipeline *pipeline)
{
   anv_compiler_free(pipeline);
   anv_batch_finish(&pipeline->batch, pipeline->device);
   anv_device_free(pipeline->device, pipeline);

   return VK_SUCCESS;
}

VkResult VKAPI vkCreateGraphicsPipelineDerivative(
    VkDevice                                    device,
    const VkGraphicsPipelineCreateInfo*         pCreateInfo,
    VkPipeline                                  basePipeline,
    VkPipeline*                                 pPipeline)
{
   stub_return(VK_UNSUPPORTED);
}

VkResult VKAPI vkCreateComputePipeline(
    VkDevice                                    device,
    const VkComputePipelineCreateInfo*          pCreateInfo,
    VkPipeline*                                 pPipeline)
{
   stub_return(VK_UNSUPPORTED);
}

VkResult VKAPI vkStorePipeline(
    VkDevice                                    device,
    VkPipeline                                  pipeline,
    size_t*                                     pDataSize,
    void*                                       pData)
{
   stub_return(VK_UNSUPPORTED);
}

VkResult VKAPI vkLoadPipeline(
    VkDevice                                    device,
    size_t                                      dataSize,
    const void*                                 pData,
    VkPipeline*                                 pPipeline)
{
   stub_return(VK_UNSUPPORTED);
}

VkResult VKAPI vkLoadPipelineDerivative(
    VkDevice                                    device,
    size_t                                      dataSize,
    const void*                                 pData,
    VkPipeline                                  basePipeline,
    VkPipeline*                                 pPipeline)
{
   stub_return(VK_UNSUPPORTED);
}

// Pipeline layout functions

VkResult VKAPI vkCreatePipelineLayout(
    VkDevice                                    _device,
    const VkPipelineLayoutCreateInfo*           pCreateInfo,
    VkPipelineLayout*                           pPipelineLayout)
{
   struct anv_device *device = (struct anv_device *) _device;
   struct anv_pipeline_layout *layout;

   assert(pCreateInfo->sType == VK_STRUCTURE_TYPE_PIPELINE_LAYOUT_CREATE_INFO);

   layout = anv_device_alloc(device, sizeof(*layout), 8,
                             VK_SYSTEM_ALLOC_TYPE_API_OBJECT);
   if (layout == NULL)
      return vk_error(VK_ERROR_OUT_OF_HOST_MEMORY);

   layout->num_sets = pCreateInfo->descriptorSetCount;

   uint32_t surface_start[VK_NUM_SHADER_STAGE] = { 0, };
   uint32_t sampler_start[VK_NUM_SHADER_STAGE] = { 0, };

   for (uint32_t s = 0; s < VK_NUM_SHADER_STAGE; s++) {
      layout->stage[s].surface_count = 0;
      layout->stage[s].sampler_count = 0;
   }

   for (uint32_t i = 0; i < pCreateInfo->descriptorSetCount; i++) {
      struct anv_descriptor_set_layout *set_layout =
         (struct anv_descriptor_set_layout *) pCreateInfo->pSetLayouts[i];

      layout->set[i].layout = set_layout;
      for (uint32_t s = 0; s < VK_NUM_SHADER_STAGE; s++) {
         layout->set[i].surface_start[s] = surface_start[s];
         surface_start[s] += set_layout->stage[s].surface_count;
         layout->set[i].sampler_start[s] = sampler_start[s];
         sampler_start[s] += set_layout->stage[s].sampler_count;

         layout->stage[s].surface_count += set_layout->stage[s].surface_count;
         layout->stage[s].sampler_count += set_layout->stage[s].sampler_count;
      }
   }

   *pPipelineLayout = (VkPipelineLayout) layout;

   return VK_SUCCESS;
}