Hash :
e79ed0dc
Author :
Date :
2022-03-25T17:15:30
Metal: Fix for not submitting command buffers in order First I added a check that CommandBuffers are committed in order as the design requires that they are. This showed several tests asserting including the angle end2end test, OcclusionQueriesNoSurfaceTestES3.SwitchingContextsWithQuery/ES3_Metal and also several others. The check is cheap and helps catch bugs so it seems prudent to have it. Unfortunately, AFAICT, there is no trival fix. The issue is ContextMtl::flushCommandBuffer commits the outstanding commandbuffers but then, if there is/was a query in progress, more work needs to be done. That work calls ContextMtl::getBlitCommandEncoder which calls ContextMtl::ensureCommandBufferReady which calls ProvokingVertexHelper::ensureCommandBufferReady which ends up making a new command buffer. That command buffer should be committed before switching to a new context but the code that would commit it has already executed. It's not at all clear to me how to refactor the code to do this correctly. The simplest solution is to call ContextMlt::flushCommandBuffer twice which I know is gross but at least it fixes the bug and optimizing and/or refactoring can be done separately. Bug: angleproject:7131 Change-Id: Idb11efb35f6ad2fd890a5db15d3791c07586bf34 Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3553939 Reviewed-by: Kenneth Russell <kbr@chromium.org> Commit-Queue: Gregg Tavares <gman@chromium.org>
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652
//
// Copyright 2019 The ANGLE Project Authors. All rights reserved.
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.
//
// mtl_command_buffer.h:
// Defines the wrapper classes for Metal's MTLCommandEncoder, MTLCommandQueue and
// MTLCommandBuffer.
//
#ifndef LIBANGLE_RENDERER_METAL_COMMANDENBUFFERMTL_H_
#define LIBANGLE_RENDERER_METAL_COMMANDENBUFFERMTL_H_
#import <Metal/Metal.h>
#import <QuartzCore/CAMetalLayer.h>
#include <cstdint>
#include <deque>
#include <memory>
#include <mutex>
#include <thread>
#include <unordered_set>
#include <vector>
#include "common/FixedVector.h"
#include "common/angleutils.h"
#include "libANGLE/renderer/metal/mtl_common.h"
#include "libANGLE/renderer/metal/mtl_resources.h"
#include "libANGLE/renderer/metal/mtl_state_cache.h"
namespace rx
{
namespace mtl
{
enum CommandBufferFinishOperation
{
NoWait,
WaitUntilScheduled,
WaitUntilFinished
};
class CommandBuffer;
class CommandEncoder;
class RenderCommandEncoder;
class OcclusionQueryPool;
class CommandQueue final : public WrappedObject<id<MTLCommandQueue>>, angle::NonCopyable
{
public:
void reset();
void set(id<MTLCommandQueue> metalQueue);
void finishAllCommands();
// This method will ensure that every GPU command buffer using this resource will finish before
// returning. Note: this doesn't include the "in-progress" command buffer, i.e. the one hasn't
// been commmitted yet. It's the responsibility of caller to make sure that command buffer is
// commited/flushed first before calling this method.
void ensureResourceReadyForCPU(const ResourceRef &resource);
void ensureResourceReadyForCPU(Resource *resource);
// Check whether the resource is being used by any command buffer still running on GPU.
// This must be called before attempting to read the content of resource on CPU side.
bool isResourceBeingUsedByGPU(const ResourceRef &resource) const
{
return isResourceBeingUsedByGPU(resource.get());
}
bool isResourceBeingUsedByGPU(const Resource *resource) const;
// Checks whether the last command buffer that uses the given resource has been committed or not
bool resourceHasPendingWorks(const Resource *resource) const;
CommandQueue &operator=(id<MTLCommandQueue> metalQueue)
{
set(metalQueue);
return *this;
}
AutoObjCPtr<id<MTLCommandBuffer>> makeMetalCommandBuffer(uint64_t *queueSerialOut);
void onCommandBufferCommitted(id<MTLCommandBuffer> buf, uint64_t serial);
private:
void onCommandBufferCompleted(id<MTLCommandBuffer> buf, uint64_t serial);
using ParentClass = WrappedObject<id<MTLCommandQueue>>;
struct CmdBufferQueueEntry
{
AutoObjCPtr<id<MTLCommandBuffer>> buffer;
uint64_t serial;
};
std::deque<CmdBufferQueueEntry> mMetalCmdBuffers;
uint64_t mQueueSerialCounter = 1;
uint64_t mLastCommittedSerial = 0;
std::atomic<uint64_t> mCommittedBufferSerial{0};
std::atomic<uint64_t> mCompletedBufferSerial{0};
mutable std::mutex mLock;
};
class CommandBuffer final : public WrappedObject<id<MTLCommandBuffer>>, angle::NonCopyable
{
public:
CommandBuffer(CommandQueue *cmdQueue);
~CommandBuffer();
// This method must be called so that command encoder can be used.
void restart();
// Return true if command buffer can be encoded into. Return false if it has been committed
// and hasn't been restarted.
bool ready() const;
void commit(CommandBufferFinishOperation operation);
void present(id<CAMetalDrawable> presentationDrawable);
void setWriteDependency(const ResourceRef &resource);
void setReadDependency(const ResourceRef &resource);
void setReadDependency(Resource *resourcePtr);
void queueEventSignal(const mtl::SharedEventRef &event, uint64_t value);
void serverWaitEvent(const mtl::SharedEventRef &event, uint64_t value);
void insertDebugSign(const std::string &marker);
void pushDebugGroup(const std::string &marker);
void popDebugGroup();
CommandQueue &cmdQueue() { return mCmdQueue; }
// Private use only
void setActiveCommandEncoder(CommandEncoder *encoder);
void invalidateActiveCommandEncoder(CommandEncoder *encoder);
bool needsFlushForDrawCallLimits() const;
private:
void set(id<MTLCommandBuffer> metalBuffer);
void cleanup();
bool readyImpl() const;
bool commitImpl();
void forceEndingCurrentEncoder();
void setPendingEvents();
void setEventImpl(const mtl::SharedEventRef &event, uint64_t value);
void waitEventImpl(const mtl::SharedEventRef &event, uint64_t value);
void pushDebugGroupImpl(const std::string &marker);
void popDebugGroupImpl();
void setResourceUsedByCommandBuffer(const ResourceRef &resource);
void clearResourceListAndSize();
using ParentClass = WrappedObject<id<MTLCommandBuffer>>;
CommandQueue &mCmdQueue;
CommandEncoder *mActiveCommandEncoder = nullptr;
uint64_t mQueueSerial = 0;
mutable std::mutex mLock;
std::vector<std::string> mPendingDebugSigns;
std::vector<std::pair<mtl::SharedEventRef, uint64_t>> mPendingSignalEvents;
std::vector<std::string> mDebugGroups;
std::unordered_set<id> mResourceList;
size_t mWorkingResourceSize = 0;
bool mCommitted = false;
};
class CommandEncoder : public WrappedObject<id<MTLCommandEncoder>>, angle::NonCopyable
{
public:
enum Type
{
RENDER,
BLIT,
COMPUTE,
};
virtual ~CommandEncoder();
virtual void endEncoding();
virtual void reset();
Type getType() const { return mType; }
CommandEncoder &markResourceBeingWrittenByGPU(const BufferRef &buffer);
CommandEncoder &markResourceBeingWrittenByGPU(const TextureRef &texture);
void insertDebugSign(NSString *label);
virtual void pushDebugGroup(NSString *label);
virtual void popDebugGroup();
protected:
using ParentClass = WrappedObject<id<MTLCommandEncoder>>;
CommandEncoder(CommandBuffer *cmdBuffer, Type type);
CommandBuffer &cmdBuffer() { return mCmdBuffer; }
CommandQueue &cmdQueue() { return mCmdBuffer.cmdQueue(); }
void set(id<MTLCommandEncoder> metalCmdEncoder);
virtual void insertDebugSignImpl(NSString *marker);
private:
const Type mType;
CommandBuffer &mCmdBuffer;
};
// Stream to store commands before encoding them into the real MTLCommandEncoder
class IntermediateCommandStream
{
public:
template <typename T>
inline IntermediateCommandStream &push(const T &val)
{
auto ptr = reinterpret_cast<const uint8_t *>(&val);
mBuffer.insert(mBuffer.end(), ptr, ptr + sizeof(T));
return *this;
}
inline IntermediateCommandStream &push(const uint8_t *bytes, size_t len)
{
mBuffer.insert(mBuffer.end(), bytes, bytes + len);
return *this;
}
template <typename T>
inline T peek()
{
ASSERT(mReadPtr <= mBuffer.size() - sizeof(T));
T re;
auto ptr = reinterpret_cast<uint8_t *>(&re);
std::copy(mBuffer.data() + mReadPtr, mBuffer.data() + mReadPtr + sizeof(T), ptr);
return re;
}
template <typename T>
inline T fetch()
{
auto re = peek<T>();
mReadPtr += sizeof(T);
return re;
}
inline const uint8_t *fetch(size_t bytes)
{
ASSERT(mReadPtr <= mBuffer.size() - bytes);
auto cur = mReadPtr;
mReadPtr += bytes;
return mBuffer.data() + cur;
}
inline void clear()
{
mBuffer.clear();
mReadPtr = 0;
}
inline void resetReadPtr(size_t readPtr)
{
ASSERT(readPtr <= mBuffer.size());
mReadPtr = readPtr;
}
inline bool good() const { return mReadPtr < mBuffer.size(); }
private:
std::vector<uint8_t> mBuffer;
size_t mReadPtr = 0;
};
// Per shader stage's states
struct RenderCommandEncoderShaderStates
{
RenderCommandEncoderShaderStates();
void reset();
std::array<id<MTLBuffer>, kMaxShaderBuffers> buffers;
std::array<uint32_t, kMaxShaderBuffers> bufferOffsets;
std::array<id<MTLSamplerState>, kMaxShaderSamplers> samplers;
std::array<Optional<std::pair<float, float>>, kMaxShaderSamplers> samplerLodClamps;
std::array<id<MTLTexture>, kMaxShaderSamplers> textures;
};
// Per render pass's states
struct RenderCommandEncoderStates
{
RenderCommandEncoderStates();
void reset();
id<MTLRenderPipelineState> renderPipeline;
MTLTriangleFillMode triangleFillMode;
MTLWinding winding;
MTLCullMode cullMode;
id<MTLDepthStencilState> depthStencilState;
float depthBias, depthSlopeScale, depthClamp;
uint32_t stencilFrontRef, stencilBackRef;
Optional<MTLViewport> viewport;
Optional<MTLScissorRect> scissorRect;
std::array<float, 4> blendColor;
gl::ShaderMap<RenderCommandEncoderShaderStates> perShaderStates;
MTLVisibilityResultMode visibilityResultMode;
size_t visibilityResultBufferOffset;
};
// Encoder for encoding render commands
class RenderCommandEncoder final : public CommandEncoder
{
public:
RenderCommandEncoder(CommandBuffer *cmdBuffer, const OcclusionQueryPool &queryPool);
~RenderCommandEncoder() override;
// override CommandEncoder
bool valid() const { return mRecording; }
void reset() override;
void endEncoding() override;
// Restart the encoder so that new commands can be encoded.
// NOTE: parent CommandBuffer's restart() must be called before this.
RenderCommandEncoder &restart(const RenderPassDesc &desc);
RenderCommandEncoder &setRenderPipelineState(id<MTLRenderPipelineState> state);
RenderCommandEncoder &setTriangleFillMode(MTLTriangleFillMode mode);
RenderCommandEncoder &setFrontFacingWinding(MTLWinding winding);
RenderCommandEncoder &setCullMode(MTLCullMode mode);
RenderCommandEncoder &setDepthStencilState(id<MTLDepthStencilState> state);
RenderCommandEncoder &setDepthBias(float depthBias, float slopeScale, float clamp);
RenderCommandEncoder &setStencilRefVals(uint32_t frontRef, uint32_t backRef);
RenderCommandEncoder &setStencilRefVal(uint32_t ref);
RenderCommandEncoder &setViewport(const MTLViewport &viewport);
RenderCommandEncoder &setScissorRect(const MTLScissorRect &rect);
RenderCommandEncoder &setBlendColor(float r, float g, float b, float a);
RenderCommandEncoder &setVertexBuffer(const BufferRef &buffer, uint32_t offset, uint32_t index)
{
return setBuffer(gl::ShaderType::Vertex, buffer, offset, index);
}
RenderCommandEncoder &setVertexBytes(const uint8_t *bytes, size_t size, uint32_t index)
{
return setBytes(gl::ShaderType::Vertex, bytes, size, index);
}
template <typename T>
RenderCommandEncoder &setVertexData(const T &data, uint32_t index)
{
return setVertexBytes(reinterpret_cast<const uint8_t *>(&data), sizeof(T), index);
}
RenderCommandEncoder &setVertexSamplerState(id<MTLSamplerState> state,
float lodMinClamp,
float lodMaxClamp,
uint32_t index)
{
return setSamplerState(gl::ShaderType::Vertex, state, lodMinClamp, lodMaxClamp, index);
}
RenderCommandEncoder &setVertexTexture(const TextureRef &texture, uint32_t index)
{
return setTexture(gl::ShaderType::Vertex, texture, index);
}
RenderCommandEncoder &setFragmentBuffer(const BufferRef &buffer,
uint32_t offset,
uint32_t index)
{
return setBuffer(gl::ShaderType::Fragment, buffer, offset, index);
}
RenderCommandEncoder &setFragmentBytes(const uint8_t *bytes, size_t size, uint32_t index)
{
return setBytes(gl::ShaderType::Fragment, bytes, size, index);
}
template <typename T>
RenderCommandEncoder &setFragmentData(const T &data, uint32_t index)
{
return setFragmentBytes(reinterpret_cast<const uint8_t *>(&data), sizeof(T), index);
}
RenderCommandEncoder &setFragmentSamplerState(id<MTLSamplerState> state,
float lodMinClamp,
float lodMaxClamp,
uint32_t index)
{
return setSamplerState(gl::ShaderType::Fragment, state, lodMinClamp, lodMaxClamp, index);
}
RenderCommandEncoder &setFragmentTexture(const TextureRef &texture, uint32_t index)
{
return setTexture(gl::ShaderType::Fragment, texture, index);
}
RenderCommandEncoder &setBuffer(gl::ShaderType shaderType,
const BufferRef &buffer,
uint32_t offset,
uint32_t index);
RenderCommandEncoder &setBufferForWrite(gl::ShaderType shaderType,
const BufferRef &buffer,
uint32_t offset,
uint32_t index);
RenderCommandEncoder &setBytes(gl::ShaderType shaderType,
const uint8_t *bytes,
size_t size,
uint32_t index);
template <typename T>
RenderCommandEncoder &setData(gl::ShaderType shaderType, const T &data, uint32_t index)
{
return setBytes(shaderType, reinterpret_cast<const uint8_t *>(&data), sizeof(T), index);
}
RenderCommandEncoder &setSamplerState(gl::ShaderType shaderType,
id<MTLSamplerState> state,
float lodMinClamp,
float lodMaxClamp,
uint32_t index);
RenderCommandEncoder &setTexture(gl::ShaderType shaderType,
const TextureRef &texture,
uint32_t index);
RenderCommandEncoder &draw(MTLPrimitiveType primitiveType,
uint32_t vertexStart,
uint32_t vertexCount);
RenderCommandEncoder &drawInstanced(MTLPrimitiveType primitiveType,
uint32_t vertexStart,
uint32_t vertexCount,
uint32_t instances);
RenderCommandEncoder &drawInstancedBaseInstance(MTLPrimitiveType primitiveType,
uint32_t vertexStart,
uint32_t vertexCount,
uint32_t instances,
uint32_t baseInstance);
RenderCommandEncoder &drawIndexed(MTLPrimitiveType primitiveType,
uint32_t indexCount,
MTLIndexType indexType,
const BufferRef &indexBuffer,
size_t bufferOffset);
RenderCommandEncoder &drawIndexedInstanced(MTLPrimitiveType primitiveType,
uint32_t indexCount,
MTLIndexType indexType,
const BufferRef &indexBuffer,
size_t bufferOffset,
uint32_t instances);
RenderCommandEncoder &drawIndexedInstancedBaseVertexBaseInstance(MTLPrimitiveType primitiveType,
uint32_t indexCount,
MTLIndexType indexType,
const BufferRef &indexBuffer,
size_t bufferOffset,
uint32_t instances,
uint32_t baseVertex,
uint32_t baseInstance);
RenderCommandEncoder &setVisibilityResultMode(MTLVisibilityResultMode mode, size_t offset);
RenderCommandEncoder &useResource(const BufferRef &resource,
MTLResourceUsage usage,
mtl::RenderStages states);
RenderCommandEncoder &memoryBarrierWithResource(const BufferRef &resource,
mtl::RenderStages after,
mtl::RenderStages before);
RenderCommandEncoder &setColorStoreAction(MTLStoreAction action, uint32_t colorAttachmentIndex);
// Set store action for every color attachment.
RenderCommandEncoder &setColorStoreAction(MTLStoreAction action);
RenderCommandEncoder &setDepthStencilStoreAction(MTLStoreAction depthStoreAction,
MTLStoreAction stencilStoreAction);
RenderCommandEncoder &setDepthStoreAction(MTLStoreAction action);
RenderCommandEncoder &setStencilStoreAction(MTLStoreAction action);
// Set storeaction for every color & depth & stencil attachment.
RenderCommandEncoder &setStoreAction(MTLStoreAction action);
// Change the render pass's loadAction. Note that this operation is only allowed when there
// is no draw call recorded yet.
RenderCommandEncoder &setColorLoadAction(MTLLoadAction action,
const MTLClearColor &clearValue,
uint32_t colorAttachmentIndex);
RenderCommandEncoder &setDepthLoadAction(MTLLoadAction action, double clearValue);
RenderCommandEncoder &setStencilLoadAction(MTLLoadAction action, uint32_t clearValue);
void setLabel(NSString *label);
void pushDebugGroup(NSString *label) override;
void popDebugGroup() override;
const RenderPassDesc &renderPassDesc() const { return mRenderPassDesc; }
bool hasDrawCalls() const { return mHasDrawCalls; }
private:
// Override CommandEncoder
id<MTLRenderCommandEncoder> get()
{
return static_cast<id<MTLRenderCommandEncoder>>(CommandEncoder::get());
}
void insertDebugSignImpl(NSString *label) override;
void initAttachmentWriteDependencyAndScissorRect(const RenderPassAttachmentDesc &attachment);
void initWriteDependency(const TextureRef &texture);
void finalizeLoadStoreAction(MTLRenderPassAttachmentDescriptor *objCRenderPassAttachment);
void encodeMetalEncoder();
void simulateDiscardFramebuffer();
void endEncodingImpl(bool considerDiscardSimulation);
RenderCommandEncoder &commonSetBuffer(gl::ShaderType shaderType,
id<MTLBuffer> mtlBuffer,
uint32_t offset,
uint32_t index);
RenderPassDesc mRenderPassDesc;
// Cached Objective-C render pass desc to avoid re-allocate every frame.
mtl::AutoObjCObj<MTLRenderPassDescriptor> mCachedRenderPassDescObjC;
mtl::AutoObjCObj<NSString> mLabel;
MTLScissorRect mRenderPassMaxScissorRect;
const OcclusionQueryPool &mOcclusionQueryPool;
bool mRecording = false;
bool mHasDrawCalls = false;
IntermediateCommandStream mCommands;
gl::ShaderMap<uint8_t> mSetBufferCmds;
gl::ShaderMap<uint8_t> mSetBufferOffsetCmds;
gl::ShaderMap<uint8_t> mSetBytesCmds;
gl::ShaderMap<uint8_t> mSetTextureCmds;
gl::ShaderMap<uint8_t> mSetSamplerCmds;
RenderCommandEncoderStates mStateCache = {};
bool mPipelineStateSet = false;
};
class BlitCommandEncoder final : public CommandEncoder
{
public:
BlitCommandEncoder(CommandBuffer *cmdBuffer);
~BlitCommandEncoder() override;
// Restart the encoder so that new commands can be encoded.
// NOTE: parent CommandBuffer's restart() must be called before this.
BlitCommandEncoder &restart();
BlitCommandEncoder ©Buffer(const BufferRef &src,
size_t srcOffset,
const BufferRef &dst,
size_t dstOffset,
size_t size);
BlitCommandEncoder ©BufferToTexture(const BufferRef &src,
size_t srcOffset,
size_t srcBytesPerRow,
size_t srcBytesPerImage,
MTLSize srcSize,
const TextureRef &dst,
uint32_t dstSlice,
MipmapNativeLevel dstLevel,
MTLOrigin dstOrigin,
MTLBlitOption blitOption);
BlitCommandEncoder ©TextureToBuffer(const TextureRef &src,
uint32_t srcSlice,
MipmapNativeLevel srcLevel,
MTLOrigin srcOrigin,
MTLSize srcSize,
const BufferRef &dst,
size_t dstOffset,
size_t dstBytesPerRow,
size_t dstBytesPerImage,
MTLBlitOption blitOption);
BlitCommandEncoder ©Texture(const TextureRef &src,
uint32_t srcSlice,
MipmapNativeLevel srcLevel,
const TextureRef &dst,
uint32_t dstSlice,
MipmapNativeLevel dstLevel,
uint32_t sliceCount,
uint32_t levelCount);
BlitCommandEncoder &fillBuffer(const BufferRef &buffer, NSRange range, uint8_t value);
BlitCommandEncoder &generateMipmapsForTexture(const TextureRef &texture);
BlitCommandEncoder &synchronizeResource(Buffer *bufferPtr);
BlitCommandEncoder &synchronizeResource(Texture *texturePtr);
private:
id<MTLBlitCommandEncoder> get()
{
return static_cast<id<MTLBlitCommandEncoder>>(CommandEncoder::get());
}
};
class ComputeCommandEncoder final : public CommandEncoder
{
public:
ComputeCommandEncoder(CommandBuffer *cmdBuffer);
~ComputeCommandEncoder() override;
// Restart the encoder so that new commands can be encoded.
// NOTE: parent CommandBuffer's restart() must be called before this.
ComputeCommandEncoder &restart();
ComputeCommandEncoder &setComputePipelineState(id<MTLComputePipelineState> state);
ComputeCommandEncoder &setBuffer(const BufferRef &buffer, uint32_t offset, uint32_t index);
ComputeCommandEncoder &setBufferForWrite(const BufferRef &buffer,
uint32_t offset,
uint32_t index);
ComputeCommandEncoder &setBytes(const uint8_t *bytes, size_t size, uint32_t index);
template <typename T>
ComputeCommandEncoder &setData(const T &data, uint32_t index)
{
return setBytes(reinterpret_cast<const uint8_t *>(&data), sizeof(T), index);
}
ComputeCommandEncoder &setSamplerState(id<MTLSamplerState> state,
float lodMinClamp,
float lodMaxClamp,
uint32_t index);
ComputeCommandEncoder &setTexture(const TextureRef &texture, uint32_t index);
ComputeCommandEncoder &setTextureForWrite(const TextureRef &texture, uint32_t index);
ComputeCommandEncoder &dispatch(const MTLSize &threadGroupsPerGrid,
const MTLSize &threadsPerGroup);
ComputeCommandEncoder &dispatchNonUniform(const MTLSize &threadsPerGrid,
const MTLSize &threadsPerGroup);
private:
id<MTLComputeCommandEncoder> get()
{
return static_cast<id<MTLComputeCommandEncoder>>(CommandEncoder::get());
}
};
} // namespace mtl
} // namespace rx
#endif /* LIBANGLE_RENDERER_METAL_COMMANDENBUFFERMTL_H_ */