1 // Copyright 2016 The SwiftShader Authors. All Rights Reserved.
2 //
3 // Licensed under the Apache License, Version 2.0 (the "License");
4 // you may not use this file except in compliance with the License.
5 // You may obtain a copy of the License at
6 //
7 //    http://www.apache.org/licenses/LICENSE-2.0
8 //
9 // Unless required by applicable law or agreed to in writing, software
10 // distributed under the License is distributed on an "AS IS" BASIS,
11 // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12 // See the License for the specific language governing permissions and
13 // limitations under the License.
14 
15 #include "PixelProcessor.hpp"
16 
17 #include "Surface.hpp"
18 #include "Primitive.hpp"
19 #include "Pipeline/PixelProgram.hpp"
20 #include "Pipeline/PixelShader.hpp"
21 #include "Pipeline/Constants.hpp"
22 #include "Vulkan/VkDebug.hpp"
23 
24 #include <string.h>
25 
26 namespace sw
27 {
28 	extern bool complementaryDepthBuffer;
29 	extern TransparencyAntialiasing transparencyAntialiasing;
30 	extern bool perspectiveCorrection;
31 
32 	bool precachePixel = false;
33 
computeHash()34 	unsigned int PixelProcessor::States::computeHash()
35 	{
36 		unsigned int *state = (unsigned int*)this;
37 		unsigned int hash = 0;
38 
39 		for(unsigned int i = 0; i < sizeof(States) / 4; i++)
40 		{
41 			hash ^= state[i];
42 		}
43 
44 		return hash;
45 	}
46 
State()47 	PixelProcessor::State::State()
48 	{
49 		memset(this, 0, sizeof(State));
50 	}
51 
operator ==(const State & state) const52 	bool PixelProcessor::State::operator==(const State &state) const
53 	{
54 		if(hash != state.hash)
55 		{
56 			return false;
57 		}
58 
59 		return memcmp(static_cast<const States*>(this), static_cast<const States*>(&state), sizeof(States)) == 0;
60 	}
61 
UniformBufferInfo()62 	PixelProcessor::UniformBufferInfo::UniformBufferInfo()
63 	{
64 		buffer = nullptr;
65 		offset = 0;
66 	}
67 
PixelProcessor(Context * context)68 	PixelProcessor::PixelProcessor(Context *context) : context(context)
69 	{
70 		routineCache = nullptr;
71 		setRoutineCacheSize(1024);
72 	}
73 
~PixelProcessor()74 	PixelProcessor::~PixelProcessor()
75 	{
76 		delete routineCache;
77 		routineCache = nullptr;
78 	}
79 
setFloatConstant(unsigned int index,const float value[4])80 	void PixelProcessor::setFloatConstant(unsigned int index, const float value[4])
81 	{
82 		if(index < FRAGMENT_UNIFORM_VECTORS)
83 		{
84 			c[index][0] = value[0];
85 			c[index][1] = value[1];
86 			c[index][2] = value[2];
87 			c[index][3] = value[3];
88 		}
89 		else ASSERT(false);
90 	}
91 
setIntegerConstant(unsigned int index,const int value[4])92 	void PixelProcessor::setIntegerConstant(unsigned int index, const int value[4])
93 	{
94 		if(index < 16)
95 		{
96 			i[index][0] = value[0];
97 			i[index][1] = value[1];
98 			i[index][2] = value[2];
99 			i[index][3] = value[3];
100 		}
101 		else ASSERT(false);
102 	}
103 
setBooleanConstant(unsigned int index,int boolean)104 	void PixelProcessor::setBooleanConstant(unsigned int index, int boolean)
105 	{
106 		if(index < 16)
107 		{
108 			b[index] = boolean != 0;
109 		}
110 		else ASSERT(false);
111 	}
112 
setUniformBuffer(int index,sw::Resource * buffer,int offset)113 	void PixelProcessor::setUniformBuffer(int index, sw::Resource* buffer, int offset)
114 	{
115 		uniformBufferInfo[index].buffer = buffer;
116 		uniformBufferInfo[index].offset = offset;
117 	}
118 
lockUniformBuffers(byte ** u,sw::Resource * uniformBuffers[])119 	void PixelProcessor::lockUniformBuffers(byte** u, sw::Resource* uniformBuffers[])
120 	{
121 		for(int i = 0; i < MAX_UNIFORM_BUFFER_BINDINGS; ++i)
122 		{
123 			u[i] = uniformBufferInfo[i].buffer ? static_cast<byte*>(uniformBufferInfo[i].buffer->lock(PUBLIC, PRIVATE)) + uniformBufferInfo[i].offset : nullptr;
124 			uniformBuffers[i] = uniformBufferInfo[i].buffer;
125 		}
126 	}
127 
setRenderTarget(int index,Surface * renderTarget,unsigned int layer)128 	void PixelProcessor::setRenderTarget(int index, Surface *renderTarget, unsigned int layer)
129 	{
130 		context->renderTarget[index] = renderTarget;
131 		context->renderTargetLayer[index] = layer;
132 	}
133 
setDepthBuffer(Surface * depthBuffer,unsigned int layer)134 	void PixelProcessor::setDepthBuffer(Surface *depthBuffer, unsigned int layer)
135 	{
136 		context->depthBuffer = depthBuffer;
137 		context->depthBufferLayer = layer;
138 	}
139 
setStencilBuffer(Surface * stencilBuffer,unsigned int layer)140 	void PixelProcessor::setStencilBuffer(Surface *stencilBuffer, unsigned int layer)
141 	{
142 		context->stencilBuffer = stencilBuffer;
143 		context->stencilBufferLayer = layer;
144 	}
145 
setTextureFilter(unsigned int sampler,FilterType textureFilter)146 	void PixelProcessor::setTextureFilter(unsigned int sampler, FilterType textureFilter)
147 	{
148 		if(sampler < TEXTURE_IMAGE_UNITS)
149 		{
150 			context->sampler[sampler].setTextureFilter(textureFilter);
151 		}
152 		else ASSERT(false);
153 	}
154 
setMipmapFilter(unsigned int sampler,MipmapType mipmapFilter)155 	void PixelProcessor::setMipmapFilter(unsigned int sampler, MipmapType mipmapFilter)
156 	{
157 		if(sampler < TEXTURE_IMAGE_UNITS)
158 		{
159 			context->sampler[sampler].setMipmapFilter(mipmapFilter);
160 		}
161 		else ASSERT(false);
162 	}
163 
setGatherEnable(unsigned int sampler,bool enable)164 	void PixelProcessor::setGatherEnable(unsigned int sampler, bool enable)
165 	{
166 		if(sampler < TEXTURE_IMAGE_UNITS)
167 		{
168 			context->sampler[sampler].setGatherEnable(enable);
169 		}
170 		else ASSERT(false);
171 	}
172 
setAddressingModeU(unsigned int sampler,AddressingMode addressMode)173 	void PixelProcessor::setAddressingModeU(unsigned int sampler, AddressingMode addressMode)
174 	{
175 		if(sampler < TEXTURE_IMAGE_UNITS)
176 		{
177 			context->sampler[sampler].setAddressingModeU(addressMode);
178 		}
179 		else ASSERT(false);
180 	}
181 
setAddressingModeV(unsigned int sampler,AddressingMode addressMode)182 	void PixelProcessor::setAddressingModeV(unsigned int sampler, AddressingMode addressMode)
183 	{
184 		if(sampler < TEXTURE_IMAGE_UNITS)
185 		{
186 			context->sampler[sampler].setAddressingModeV(addressMode);
187 		}
188 		else ASSERT(false);
189 	}
190 
setAddressingModeW(unsigned int sampler,AddressingMode addressMode)191 	void PixelProcessor::setAddressingModeW(unsigned int sampler, AddressingMode addressMode)
192 	{
193 		if(sampler < TEXTURE_IMAGE_UNITS)
194 		{
195 			context->sampler[sampler].setAddressingModeW(addressMode);
196 		}
197 		else ASSERT(false);
198 	}
199 
setReadSRGB(unsigned int sampler,bool sRGB)200 	void PixelProcessor::setReadSRGB(unsigned int sampler, bool sRGB)
201 	{
202 		if(sampler < TEXTURE_IMAGE_UNITS)
203 		{
204 			context->sampler[sampler].setReadSRGB(sRGB);
205 		}
206 		else ASSERT(false);
207 	}
208 
setMipmapLOD(unsigned int sampler,float bias)209 	void PixelProcessor::setMipmapLOD(unsigned int sampler, float bias)
210 	{
211 		if(sampler < TEXTURE_IMAGE_UNITS)
212 		{
213 			context->sampler[sampler].setMipmapLOD(bias);
214 		}
215 		else ASSERT(false);
216 	}
217 
setBorderColor(unsigned int sampler,const Color<float> & borderColor)218 	void PixelProcessor::setBorderColor(unsigned int sampler, const Color<float> &borderColor)
219 	{
220 		if(sampler < TEXTURE_IMAGE_UNITS)
221 		{
222 			context->sampler[sampler].setBorderColor(borderColor);
223 		}
224 		else ASSERT(false);
225 	}
226 
setMaxAnisotropy(unsigned int sampler,float maxAnisotropy)227 	void PixelProcessor::setMaxAnisotropy(unsigned int sampler, float maxAnisotropy)
228 	{
229 		if(sampler < TEXTURE_IMAGE_UNITS)
230 		{
231 			context->sampler[sampler].setMaxAnisotropy(maxAnisotropy);
232 		}
233 		else ASSERT(false);
234 	}
235 
setHighPrecisionFiltering(unsigned int sampler,bool highPrecisionFiltering)236 	void PixelProcessor::setHighPrecisionFiltering(unsigned int sampler, bool highPrecisionFiltering)
237 	{
238 		if(sampler < TEXTURE_IMAGE_UNITS)
239 		{
240 			context->sampler[sampler].setHighPrecisionFiltering(highPrecisionFiltering);
241 		}
242 		else ASSERT(false);
243 	}
244 
setSwizzleR(unsigned int sampler,SwizzleType swizzleR)245 	void PixelProcessor::setSwizzleR(unsigned int sampler, SwizzleType swizzleR)
246 	{
247 		if(sampler < TEXTURE_IMAGE_UNITS)
248 		{
249 			context->sampler[sampler].setSwizzleR(swizzleR);
250 		}
251 		else ASSERT(false);
252 	}
253 
setSwizzleG(unsigned int sampler,SwizzleType swizzleG)254 	void PixelProcessor::setSwizzleG(unsigned int sampler, SwizzleType swizzleG)
255 	{
256 		if(sampler < TEXTURE_IMAGE_UNITS)
257 		{
258 			context->sampler[sampler].setSwizzleG(swizzleG);
259 		}
260 		else ASSERT(false);
261 	}
262 
setSwizzleB(unsigned int sampler,SwizzleType swizzleB)263 	void PixelProcessor::setSwizzleB(unsigned int sampler, SwizzleType swizzleB)
264 	{
265 		if(sampler < TEXTURE_IMAGE_UNITS)
266 		{
267 			context->sampler[sampler].setSwizzleB(swizzleB);
268 		}
269 		else ASSERT(false);
270 	}
271 
setSwizzleA(unsigned int sampler,SwizzleType swizzleA)272 	void PixelProcessor::setSwizzleA(unsigned int sampler, SwizzleType swizzleA)
273 	{
274 		if(sampler < TEXTURE_IMAGE_UNITS)
275 		{
276 			context->sampler[sampler].setSwizzleA(swizzleA);
277 		}
278 		else ASSERT(false);
279 	}
280 
setCompareFunc(unsigned int sampler,CompareFunc compFunc)281 	void PixelProcessor::setCompareFunc(unsigned int sampler, CompareFunc compFunc)
282 	{
283 		if(sampler < TEXTURE_IMAGE_UNITS)
284 		{
285 			context->sampler[sampler].setCompareFunc(compFunc);
286 		}
287 		else ASSERT(false);
288 	}
289 
setBaseLevel(unsigned int sampler,int baseLevel)290 	void PixelProcessor::setBaseLevel(unsigned int sampler, int baseLevel)
291 	{
292 		if(sampler < TEXTURE_IMAGE_UNITS)
293 		{
294 			context->sampler[sampler].setBaseLevel(baseLevel);
295 		}
296 		else ASSERT(false);
297 	}
298 
setMaxLevel(unsigned int sampler,int maxLevel)299 	void PixelProcessor::setMaxLevel(unsigned int sampler, int maxLevel)
300 	{
301 		if(sampler < TEXTURE_IMAGE_UNITS)
302 		{
303 			context->sampler[sampler].setMaxLevel(maxLevel);
304 		}
305 		else ASSERT(false);
306 	}
307 
setMinLod(unsigned int sampler,float minLod)308 	void PixelProcessor::setMinLod(unsigned int sampler, float minLod)
309 	{
310 		if(sampler < TEXTURE_IMAGE_UNITS)
311 		{
312 			context->sampler[sampler].setMinLod(minLod);
313 		}
314 		else ASSERT(false);
315 	}
316 
setMaxLod(unsigned int sampler,float maxLod)317 	void PixelProcessor::setMaxLod(unsigned int sampler, float maxLod)
318 	{
319 		if(sampler < TEXTURE_IMAGE_UNITS)
320 		{
321 			context->sampler[sampler].setMaxLod(maxLod);
322 		}
323 		else ASSERT(false);
324 	}
325 
setWriteSRGB(bool sRGB)326 	void PixelProcessor::setWriteSRGB(bool sRGB)
327 	{
328 		context->setWriteSRGB(sRGB);
329 	}
330 
setColorLogicOpEnabled(bool colorLogicOpEnabled)331 	void PixelProcessor::setColorLogicOpEnabled(bool colorLogicOpEnabled)
332 	{
333 		context->setColorLogicOpEnabled(colorLogicOpEnabled);
334 	}
335 
setLogicalOperation(VkLogicOp logicalOperation)336 	void PixelProcessor::setLogicalOperation(VkLogicOp logicalOperation)
337 	{
338 		context->setLogicalOperation(logicalOperation);
339 	}
340 
setDepthBufferEnable(bool depthBufferEnable)341 	void PixelProcessor::setDepthBufferEnable(bool depthBufferEnable)
342 	{
343 		context->setDepthBufferEnable(depthBufferEnable);
344 	}
345 
setDepthCompare(VkCompareOp depthCompareMode)346 	void PixelProcessor::setDepthCompare(VkCompareOp depthCompareMode)
347 	{
348 		context->depthCompareMode = depthCompareMode;
349 	}
350 
setAlphaCompare(VkCompareOp alphaCompareMode)351 	void PixelProcessor::setAlphaCompare(VkCompareOp alphaCompareMode)
352 	{
353 		context->alphaCompareMode = alphaCompareMode;
354 	}
355 
setDepthWriteEnable(bool depthWriteEnable)356 	void PixelProcessor::setDepthWriteEnable(bool depthWriteEnable)
357 	{
358 		context->depthWriteEnable = depthWriteEnable;
359 	}
360 
setAlphaTestEnable(bool alphaTestEnable)361 	void PixelProcessor::setAlphaTestEnable(bool alphaTestEnable)
362 	{
363 		context->alphaTestEnable = alphaTestEnable;
364 	}
365 
setCullMode(CullMode cullMode,bool frontFacingCCW)366 	void PixelProcessor::setCullMode(CullMode cullMode, bool frontFacingCCW)
367 	{
368 		context->cullMode = cullMode;
369 		context->frontFacingCCW = frontFacingCCW;
370 	}
371 
setColorWriteMask(int index,int rgbaMask)372 	void PixelProcessor::setColorWriteMask(int index, int rgbaMask)
373 	{
374 		context->setColorWriteMask(index, rgbaMask);
375 	}
376 
setStencilEnable(bool stencilEnable)377 	void PixelProcessor::setStencilEnable(bool stencilEnable)
378 	{
379 		context->stencilEnable = stencilEnable;
380 	}
381 
setStencilCompare(VkCompareOp stencilCompareMode)382 	void PixelProcessor::setStencilCompare(VkCompareOp stencilCompareMode)
383 	{
384 		context->stencilCompareMode = stencilCompareMode;
385 	}
386 
setStencilReference(int stencilReference)387 	void PixelProcessor::setStencilReference(int stencilReference)
388 	{
389 		context->stencilReference = stencilReference;
390 		stencil.set(stencilReference, context->stencilMask, context->stencilWriteMask);
391 	}
392 
setStencilReferenceCCW(int stencilReferenceCCW)393 	void PixelProcessor::setStencilReferenceCCW(int stencilReferenceCCW)
394 	{
395 		context->stencilReferenceCCW = stencilReferenceCCW;
396 		stencilCCW.set(stencilReferenceCCW, context->stencilMaskCCW, context->stencilWriteMaskCCW);
397 	}
398 
setStencilMask(int stencilMask)399 	void PixelProcessor::setStencilMask(int stencilMask)
400 	{
401 		context->stencilMask = stencilMask;
402 		stencil.set(context->stencilReference, stencilMask, context->stencilWriteMask);
403 	}
404 
setStencilMaskCCW(int stencilMaskCCW)405 	void PixelProcessor::setStencilMaskCCW(int stencilMaskCCW)
406 	{
407 		context->stencilMaskCCW = stencilMaskCCW;
408 		stencilCCW.set(context->stencilReferenceCCW, stencilMaskCCW, context->stencilWriteMaskCCW);
409 	}
410 
setStencilFailOperation(VkStencilOp stencilFailOperation)411 	void PixelProcessor::setStencilFailOperation(VkStencilOp stencilFailOperation)
412 	{
413 		context->stencilFailOperation = stencilFailOperation;
414 	}
415 
setStencilPassOperation(VkStencilOp stencilPassOperation)416 	void PixelProcessor::setStencilPassOperation(VkStencilOp stencilPassOperation)
417 	{
418 		context->stencilPassOperation = stencilPassOperation;
419 	}
420 
setStencilZFailOperation(VkStencilOp stencilZFailOperation)421 	void PixelProcessor::setStencilZFailOperation(VkStencilOp stencilZFailOperation)
422 	{
423 		context->stencilZFailOperation = stencilZFailOperation;
424 	}
425 
setStencilWriteMask(int stencilWriteMask)426 	void PixelProcessor::setStencilWriteMask(int stencilWriteMask)
427 	{
428 		context->stencilWriteMask = stencilWriteMask;
429 		stencil.set(context->stencilReference, context->stencilMask, stencilWriteMask);
430 	}
431 
setStencilWriteMaskCCW(int stencilWriteMaskCCW)432 	void PixelProcessor::setStencilWriteMaskCCW(int stencilWriteMaskCCW)
433 	{
434 		context->stencilWriteMaskCCW = stencilWriteMaskCCW;
435 		stencilCCW.set(context->stencilReferenceCCW, context->stencilMaskCCW, stencilWriteMaskCCW);
436 	}
437 
setTwoSidedStencil(bool enable)438 	void PixelProcessor::setTwoSidedStencil(bool enable)
439 	{
440 		context->twoSidedStencil = enable;
441 	}
442 
setStencilCompareCCW(VkCompareOp stencilCompareMode)443 	void PixelProcessor::setStencilCompareCCW(VkCompareOp stencilCompareMode)
444 	{
445 		context->stencilCompareModeCCW = stencilCompareMode;
446 	}
447 
setStencilFailOperationCCW(VkStencilOp stencilFailOperation)448 	void PixelProcessor::setStencilFailOperationCCW(VkStencilOp stencilFailOperation)
449 	{
450 		context->stencilFailOperationCCW = stencilFailOperation;
451 	}
452 
setStencilPassOperationCCW(VkStencilOp stencilPassOperation)453 	void PixelProcessor::setStencilPassOperationCCW(VkStencilOp stencilPassOperation)
454 	{
455 		context->stencilPassOperationCCW = stencilPassOperation;
456 	}
457 
setStencilZFailOperationCCW(VkStencilOp stencilZFailOperation)458 	void PixelProcessor::setStencilZFailOperationCCW(VkStencilOp stencilZFailOperation)
459 	{
460 		context->stencilZFailOperationCCW = stencilZFailOperation;
461 	}
462 
setBlendConstant(const Color<float> & blendConstant)463 	void PixelProcessor::setBlendConstant(const Color<float> &blendConstant)
464 	{
465 		// FIXME: Compact into generic function   // FIXME: Clamp
466 		short blendConstantR = iround(65535 * blendConstant.r);
467 		short blendConstantG = iround(65535 * blendConstant.g);
468 		short blendConstantB = iround(65535 * blendConstant.b);
469 		short blendConstantA = iround(65535 * blendConstant.a);
470 
471 		factor.blendConstant4W[0][0] = blendConstantR;
472 		factor.blendConstant4W[0][1] = blendConstantR;
473 		factor.blendConstant4W[0][2] = blendConstantR;
474 		factor.blendConstant4W[0][3] = blendConstantR;
475 
476 		factor.blendConstant4W[1][0] = blendConstantG;
477 		factor.blendConstant4W[1][1] = blendConstantG;
478 		factor.blendConstant4W[1][2] = blendConstantG;
479 		factor.blendConstant4W[1][3] = blendConstantG;
480 
481 		factor.blendConstant4W[2][0] = blendConstantB;
482 		factor.blendConstant4W[2][1] = blendConstantB;
483 		factor.blendConstant4W[2][2] = blendConstantB;
484 		factor.blendConstant4W[2][3] = blendConstantB;
485 
486 		factor.blendConstant4W[3][0] = blendConstantA;
487 		factor.blendConstant4W[3][1] = blendConstantA;
488 		factor.blendConstant4W[3][2] = blendConstantA;
489 		factor.blendConstant4W[3][3] = blendConstantA;
490 
491 		// FIXME: Compact into generic function   // FIXME: Clamp
492 		short invBlendConstantR = iround(65535 * (1 - blendConstant.r));
493 		short invBlendConstantG = iround(65535 * (1 - blendConstant.g));
494 		short invBlendConstantB = iround(65535 * (1 - blendConstant.b));
495 		short invBlendConstantA = iround(65535 * (1 - blendConstant.a));
496 
497 		factor.invBlendConstant4W[0][0] = invBlendConstantR;
498 		factor.invBlendConstant4W[0][1] = invBlendConstantR;
499 		factor.invBlendConstant4W[0][2] = invBlendConstantR;
500 		factor.invBlendConstant4W[0][3] = invBlendConstantR;
501 
502 		factor.invBlendConstant4W[1][0] = invBlendConstantG;
503 		factor.invBlendConstant4W[1][1] = invBlendConstantG;
504 		factor.invBlendConstant4W[1][2] = invBlendConstantG;
505 		factor.invBlendConstant4W[1][3] = invBlendConstantG;
506 
507 		factor.invBlendConstant4W[2][0] = invBlendConstantB;
508 		factor.invBlendConstant4W[2][1] = invBlendConstantB;
509 		factor.invBlendConstant4W[2][2] = invBlendConstantB;
510 		factor.invBlendConstant4W[2][3] = invBlendConstantB;
511 
512 		factor.invBlendConstant4W[3][0] = invBlendConstantA;
513 		factor.invBlendConstant4W[3][1] = invBlendConstantA;
514 		factor.invBlendConstant4W[3][2] = invBlendConstantA;
515 		factor.invBlendConstant4W[3][3] = invBlendConstantA;
516 
517 		factor.blendConstant4F[0][0] = blendConstant.r;
518 		factor.blendConstant4F[0][1] = blendConstant.r;
519 		factor.blendConstant4F[0][2] = blendConstant.r;
520 		factor.blendConstant4F[0][3] = blendConstant.r;
521 
522 		factor.blendConstant4F[1][0] = blendConstant.g;
523 		factor.blendConstant4F[1][1] = blendConstant.g;
524 		factor.blendConstant4F[1][2] = blendConstant.g;
525 		factor.blendConstant4F[1][3] = blendConstant.g;
526 
527 		factor.blendConstant4F[2][0] = blendConstant.b;
528 		factor.blendConstant4F[2][1] = blendConstant.b;
529 		factor.blendConstant4F[2][2] = blendConstant.b;
530 		factor.blendConstant4F[2][3] = blendConstant.b;
531 
532 		factor.blendConstant4F[3][0] = blendConstant.a;
533 		factor.blendConstant4F[3][1] = blendConstant.a;
534 		factor.blendConstant4F[3][2] = blendConstant.a;
535 		factor.blendConstant4F[3][3] = blendConstant.a;
536 
537 		factor.invBlendConstant4F[0][0] = 1 - blendConstant.r;
538 		factor.invBlendConstant4F[0][1] = 1 - blendConstant.r;
539 		factor.invBlendConstant4F[0][2] = 1 - blendConstant.r;
540 		factor.invBlendConstant4F[0][3] = 1 - blendConstant.r;
541 
542 		factor.invBlendConstant4F[1][0] = 1 - blendConstant.g;
543 		factor.invBlendConstant4F[1][1] = 1 - blendConstant.g;
544 		factor.invBlendConstant4F[1][2] = 1 - blendConstant.g;
545 		factor.invBlendConstant4F[1][3] = 1 - blendConstant.g;
546 
547 		factor.invBlendConstant4F[2][0] = 1 - blendConstant.b;
548 		factor.invBlendConstant4F[2][1] = 1 - blendConstant.b;
549 		factor.invBlendConstant4F[2][2] = 1 - blendConstant.b;
550 		factor.invBlendConstant4F[2][3] = 1 - blendConstant.b;
551 
552 		factor.invBlendConstant4F[3][0] = 1 - blendConstant.a;
553 		factor.invBlendConstant4F[3][1] = 1 - blendConstant.a;
554 		factor.invBlendConstant4F[3][2] = 1 - blendConstant.a;
555 		factor.invBlendConstant4F[3][3] = 1 - blendConstant.a;
556 	}
557 
setAlphaBlendEnable(bool alphaBlendEnable)558 	void PixelProcessor::setAlphaBlendEnable(bool alphaBlendEnable)
559 	{
560 		context->setAlphaBlendEnable(alphaBlendEnable);
561 	}
562 
setSourceBlendFactor(VkBlendFactor sourceBlendFactor)563 	void PixelProcessor::setSourceBlendFactor(VkBlendFactor sourceBlendFactor)
564 	{
565 		context->setSourceBlendFactor(sourceBlendFactor);
566 	}
567 
setDestBlendFactor(VkBlendFactor destBlendFactor)568 	void PixelProcessor::setDestBlendFactor(VkBlendFactor destBlendFactor)
569 	{
570 		context->setDestBlendFactor(destBlendFactor);
571 	}
572 
setBlendOperation(VkBlendOp blendOperation)573 	void PixelProcessor::setBlendOperation(VkBlendOp blendOperation)
574 	{
575 		context->setBlendOperation(blendOperation);
576 	}
577 
setSeparateAlphaBlendEnable(bool separateAlphaBlendEnable)578 	void PixelProcessor::setSeparateAlphaBlendEnable(bool separateAlphaBlendEnable)
579 	{
580 		context->setSeparateAlphaBlendEnable(separateAlphaBlendEnable);
581 	}
582 
setSourceBlendFactorAlpha(VkBlendFactor sourceBlendFactorAlpha)583 	void PixelProcessor::setSourceBlendFactorAlpha(VkBlendFactor sourceBlendFactorAlpha)
584 	{
585 		context->setSourceBlendFactorAlpha(sourceBlendFactorAlpha);
586 	}
587 
setDestBlendFactorAlpha(VkBlendFactor destBlendFactorAlpha)588 	void PixelProcessor::setDestBlendFactorAlpha(VkBlendFactor destBlendFactorAlpha)
589 	{
590 		context->setDestBlendFactorAlpha(destBlendFactorAlpha);
591 	}
592 
setBlendOperationAlpha(VkBlendOp blendOperationAlpha)593 	void PixelProcessor::setBlendOperationAlpha(VkBlendOp blendOperationAlpha)
594 	{
595 		context->setBlendOperationAlpha(blendOperationAlpha);
596 	}
597 
setAlphaReference(float alphaReference)598 	void PixelProcessor::setAlphaReference(float alphaReference)
599 	{
600 		context->alphaReference = alphaReference;
601 
602 		factor.alphaReference4[0] = (word)iround(alphaReference * 0x1000 / 0xFF);
603 		factor.alphaReference4[1] = (word)iround(alphaReference * 0x1000 / 0xFF);
604 		factor.alphaReference4[2] = (word)iround(alphaReference * 0x1000 / 0xFF);
605 		factor.alphaReference4[3] = (word)iround(alphaReference * 0x1000 / 0xFF);
606 	}
607 
setPerspectiveCorrection(bool perspectiveEnable)608 	void PixelProcessor::setPerspectiveCorrection(bool perspectiveEnable)
609 	{
610 		perspectiveCorrection = perspectiveEnable;
611 	}
612 
setOcclusionEnabled(bool enable)613 	void PixelProcessor::setOcclusionEnabled(bool enable)
614 	{
615 		context->occlusionEnabled = enable;
616 	}
617 
setRoutineCacheSize(int cacheSize)618 	void PixelProcessor::setRoutineCacheSize(int cacheSize)
619 	{
620 		delete routineCache;
621 		routineCache = new RoutineCache<State>(clamp(cacheSize, 1, 65536), precachePixel ? "sw-pixel" : 0);
622 	}
623 
update() const624 	const PixelProcessor::State PixelProcessor::update() const
625 	{
626 		State state;
627 
628 		if(context->pixelShader)
629 		{
630 			state.shaderID = context->pixelShader->getSerialID();
631 		}
632 		else
633 		{
634 			state.shaderID = 0;
635 		}
636 
637 		state.depthOverride = context->pixelShader && context->pixelShader->depthOverride();
638 		state.shaderContainsKill = context->pixelShader ? context->pixelShader->containsKill() : false;
639 
640 		if(context->alphaTestActive())
641 		{
642 			state.alphaCompareMode = context->alphaCompareMode;
643 
644 			state.transparencyAntialiasing = context->getMultiSampleCount() > 1 ? transparencyAntialiasing : TRANSPARENCY_NONE;
645 		}
646 
647 		state.depthWriteEnable = context->depthWriteActive();
648 
649 		if(context->stencilActive())
650 		{
651 			state.stencilActive = true;
652 			state.stencilCompareMode = context->stencilCompareMode;
653 			state.stencilFailOperation = context->stencilFailOperation;
654 			state.stencilPassOperation = context->stencilPassOperation;
655 			state.stencilZFailOperation = context->stencilZFailOperation;
656 			state.noStencilMask = (context->stencilMask == 0xFF);
657 			state.noStencilWriteMask = (context->stencilWriteMask == 0xFF);
658 			state.stencilWriteMasked = (context->stencilWriteMask == 0x00);
659 
660 			state.twoSidedStencil = context->twoSidedStencil;
661 			state.stencilCompareModeCCW = context->twoSidedStencil ? context->stencilCompareModeCCW : state.stencilCompareMode;
662 			state.stencilFailOperationCCW = context->twoSidedStencil ? context->stencilFailOperationCCW : state.stencilFailOperation;
663 			state.stencilPassOperationCCW = context->twoSidedStencil ? context->stencilPassOperationCCW : state.stencilPassOperation;
664 			state.stencilZFailOperationCCW = context->twoSidedStencil ? context->stencilZFailOperationCCW : state.stencilZFailOperation;
665 			state.noStencilMaskCCW = context->twoSidedStencil ? (context->stencilMaskCCW == 0xFF) : state.noStencilMask;
666 			state.noStencilWriteMaskCCW = context->twoSidedStencil ? (context->stencilWriteMaskCCW == 0xFF) : state.noStencilWriteMask;
667 			state.stencilWriteMaskedCCW = context->twoSidedStencil ? (context->stencilWriteMaskCCW == 0x00) : state.stencilWriteMasked;
668 		}
669 
670 		if(context->depthBufferActive())
671 		{
672 			state.depthTestActive = true;
673 			state.depthCompareMode = context->depthCompareMode;
674 			state.quadLayoutDepthBuffer = Surface::hasQuadLayout(context->depthBuffer->getInternalFormat());
675 		}
676 
677 		state.occlusionEnabled = context->occlusionEnabled;
678 
679 		state.perspective = context->perspectiveActive();
680 		state.depthClamp = (context->depthBias != 0.0f) || (context->slopeDepthBias != 0.0f);
681 
682 		if(context->alphaBlendActive())
683 		{
684 			state.alphaBlendActive = true;
685 			state.sourceBlendFactor = context->sourceBlendFactor();
686 			state.destBlendFactor = context->destBlendFactor();
687 			state.blendOperation = context->blendOperation();
688 			state.sourceBlendFactorAlpha = context->sourceBlendFactorAlpha();
689 			state.destBlendFactorAlpha = context->destBlendFactorAlpha();
690 			state.blendOperationAlpha = context->blendOperationAlpha();
691 		}
692 
693 		state.logicalOperation = context->colorLogicOp();
694 
695 		for(int i = 0; i < RENDERTARGETS; i++)
696 		{
697 			state.colorWriteMask |= context->colorWriteActive(i) << (4 * i);
698 			state.targetFormat[i] = context->renderTargetInternalFormat(i);
699 		}
700 
701 		state.writeSRGB	= context->writeSRGB && context->renderTarget[0] && Surface::isSRGBwritable(context->renderTarget[0]->getExternalFormat());
702 		state.multiSample = context->getMultiSampleCount();
703 		state.multiSampleMask = context->multiSampleMask;
704 
705 		if(state.multiSample > 1 && context->pixelShader)
706 		{
707 			state.centroid = context->pixelShader->containsCentroid();
708 		}
709 
710 		state.frontFaceCCW = context->frontFacingCCW;
711 
712 
713 		for(unsigned int i = 0; i < 16; i++)
714 		{
715 			if(context->pixelShader)
716 			{
717 				if(context->pixelShader->usesSampler(i))
718 				{
719 					state.sampler[i] = context->sampler[i].samplerState();
720 				}
721 			}
722 		}
723 
724 		const bool point = context->isDrawPoint();
725 
726 		for(int interpolant = 0; interpolant < MAX_FRAGMENT_INPUTS; interpolant++)
727 		{
728 			for(int component = 0; component < 4; component++)
729 			{
730 				const Shader::Semantic &semantic = context->pixelShader->getInput(interpolant, component);
731 
732 				if(semantic.active())
733 				{
734 					bool flat = point;
735 
736 					switch(semantic.usage)
737 					{
738 					case Shader::USAGE_TEXCOORD: flat = false;                  break;
739 					case Shader::USAGE_COLOR:    flat = semantic.flat || point; break;
740 					}
741 
742 					state.interpolant[interpolant].component |= 1 << component;
743 
744 					if(flat)
745 					{
746 						state.interpolant[interpolant].flat |= 1 << component;
747 					}
748 				}
749 			}
750 		}
751 
752 		if(state.centroid)
753 		{
754 			for(int interpolant = 0; interpolant < MAX_FRAGMENT_INPUTS; interpolant++)
755 			{
756 				for(int component = 0; component < 4; component++)
757 				{
758 					state.interpolant[interpolant].centroid = context->pixelShader->getInput(interpolant, 0).centroid;
759 				}
760 			}
761 		}
762 
763 		state.hash = state.computeHash();
764 
765 		return state;
766 	}
767 
routine(const State & state)768 	Routine *PixelProcessor::routine(const State &state)
769 	{
770 		Routine *routine = routineCache->query(state);
771 
772 		if(!routine)
773 		{
774 			const bool integerPipeline = (context->pixelShaderModel() <= 0x0104);
775 			QuadRasterizer *generator = new PixelProgram(state, context->pixelShader);
776 			generator->generate();
777 			routine = (*generator)("PixelRoutine_%0.8X", state.shaderID);
778 			delete generator;
779 
780 			routineCache->add(state, routine);
781 		}
782 
783 		return routine;
784 	}
785 }
786