src/gpu/GrResourceProvider.h - skia - Git at Google

 /*
  * Copyright 2015 Google Inc.
  *
  * Use of this source code is governed by a BSD-style license that can be
  * found in the LICENSE file.
  */

 #ifndef GrResourceProvider_DEFINED
 #define GrResourceProvider_DEFINED

 #include "include/gpu/GrContextOptions.h"
 #include "include/private/SkImageInfoPriv.h"
 #include "src/gpu/GrGpuBuffer.h"
 #include "src/gpu/GrResourceCache.h"

 class GrAttachment;
 class GrBackendRenderTarget;
 class GrBackendSemaphore;
 class GrBackendTexture;
 class GrGpu;
 class GrMSAAAttachment;
 class GrPath;
 class GrRenderTarget;
 class GrResourceProviderPriv;
 class GrSemaphore;
 class GrSingleOwner;
 class GrTexture;
 struct GrVkDrawableInfo;

 class GrStyle;
 class SkDescriptor;
 class SkPath;
 class SkTypeface;

 namespace skgpu { struct VertexWriter; }

 /**
  * A factory for arbitrary resource types.
  */
 class GrResourceProvider {
 public:
     GrResourceProvider(GrGpu*, GrResourceCache*, GrSingleOwner*);

     /**
      * Finds a resource in the cache, based on the specified key. Prior to calling this, the caller
      * must be sure that if a resource of exists in the cache with the given unique key then it is
      * of type T.
      */
     template <typename T = GrGpuResource>
     typename std::enable_if<std::is_base_of<GrGpuResource, T>::value, sk_sp<T>>::type
     findByUniqueKey(const skgpu::UniqueKey& key) {
         return sk_sp<T>(static_cast<T*>(this->findResourceByUniqueKey(key).release()));
     }

     ///////////////////////////////////////////////////////////////////////////
     // Textures

     /**
      * Finds a texture that approximately matches the descriptor. Will be at least as large in width
      * and height as desc specifies. If renderable is kYes then the GrTexture will also be a
      * GrRenderTarget. The texture's format and sample count will always match the request.
      * The contents of the texture are undefined.
      */
     sk_sp<GrTexture> createApproxTexture(SkISize dimensions,
                                          const GrBackendFormat& format,
                                          GrTextureType textureType,
                                          GrRenderable renderable,
                                          int renderTargetSampleCnt,
                                          GrProtected isProtected);

     /** Create an exact fit texture with no initial data to upload. */
     sk_sp<GrTexture> createTexture(SkISize dimensions,
                                    const GrBackendFormat& format,
                                    GrTextureType textureType,
                                    GrRenderable renderable,
                                    int renderTargetSampleCnt,
                                    GrMipmapped mipMapped,
                                    SkBudgeted budgeted,
                                    GrProtected isProtected);

     /**
      * Create an exact fit texture with initial data to upload. The color type must be valid
      * for the format and also describe the texel data. This will ensure any conversions that
      * need to get applied to the data before upload are applied.
      */
     sk_sp<GrTexture> createTexture(SkISize dimensions,
                                    const GrBackendFormat& format,
                                    GrTextureType textureType,
                                    GrColorType colorType,
                                    GrRenderable renderable,
                                    int renderTargetSampleCnt,
                                    SkBudgeted budgeted,
                                    GrMipMapped mipMapped,
                                    GrProtected isProtected,
                                    const GrMipLevel texels[]);

     /**
      * Create a potentially loose fit texture with the provided data. The color type must be valid
      * for the format and also describe the texel data. This will ensure any conversions that
      * need to get applied to the data before upload are applied.
      */
     sk_sp<GrTexture> createTexture(SkISize dimensions,
                                    const GrBackendFormat&,
                                    GrTextureType textureType,
                                    GrColorType srcColorType,
                                    GrRenderable,
                                    int renderTargetSampleCnt,
                                    SkBudgeted,
                                    SkBackingFit,
                                    GrProtected,
                                    const GrMipLevel& mipLevel);

     /**
      * Search the cache for a scratch texture matching the provided arguments. Failing that
      * it returns null. If non-null, the resulting texture is always budgeted.
      */
     sk_sp<GrTexture> findAndRefScratchTexture(const skgpu::ScratchKey&);
     sk_sp<GrTexture> findAndRefScratchTexture(SkISize dimensions,
                                               const GrBackendFormat&,
                                               GrTextureType textureType,
                                               GrRenderable,
                                               int renderTargetSampleCnt,
                                               GrMipmapped,
                                               GrProtected);

     /**
      * Creates a compressed texture. The GrGpu must support the SkImageImage::Compression type.
      * It will not be renderable.
      */
     sk_sp<GrTexture> createCompressedTexture(SkISize dimensions,
                                              const GrBackendFormat&,
                                              SkBudgeted,
                                              GrMipmapped,
                                              GrProtected,
                                              SkData* data);

     ///////////////////////////////////////////////////////////////////////////
     // Wrapped Backend Surfaces

     /**
      * Wraps an existing texture with a GrTexture object.
      *
      * GrIOType must either be kRead or kRW. kRead blocks any operations that would modify the
      * pixels (e.g. dst for a copy, regenerating MIP levels, write pixels).
      *
      * OpenGL: if the object is a texture Gr may change its GL texture params
      *         when it is drawn.
      *
      * @return GrTexture object or NULL on failure.
      */
     sk_sp<GrTexture> wrapBackendTexture(const GrBackendTexture& tex,
                                         GrWrapOwnership,
                                         GrWrapCacheable,
                                         GrIOType);

     sk_sp<GrTexture> wrapCompressedBackendTexture(const GrBackendTexture& tex,
                                                   GrWrapOwnership,
                                                   GrWrapCacheable);

     /**
      * This makes the backend texture be renderable. If sampleCnt is > 1 and the underlying API
      * uses separate MSAA render buffers then a MSAA render buffer is created that resolves
      * to the texture.
      */
     sk_sp<GrTexture> wrapRenderableBackendTexture(const GrBackendTexture& tex,
                                                   int sampleCnt,
                                                   GrWrapOwnership,
                                                   GrWrapCacheable);

     /**
      * Wraps an existing render target with a GrRenderTarget object. It is
      * similar to wrapBackendTexture but can be used to draw into surfaces
      * that are not also textures (e.g. FBO 0 in OpenGL, or an MSAA buffer that
      * the client will resolve to a texture). Currently wrapped render targets
      * always use the kBorrow_GrWrapOwnership and GrWrapCacheable::kNo semantics.
      *
      * @return GrRenderTarget object or NULL on failure.
      */
     sk_sp<GrRenderTarget> wrapBackendRenderTarget(const GrBackendRenderTarget&);

     sk_sp<GrRenderTarget> wrapVulkanSecondaryCBAsRenderTarget(const SkImageInfo&,
                                                               const GrVkDrawableInfo&);

     static const int kMinScratchTextureSize;

     /**
      * Either finds and refs a buffer with the given unique key, or creates a new new, fills its
      * contents with the InitializeBufferDataFn() callback, and assigns it the unique key.
      *
      * @param intendedType        hint to the graphics subsystem about how the buffer will be used.
      * @param size                minimum size of buffer to return.
      * @param key                 Key to be assigned to the buffer.
      * @param InitializeBufferFn  callback with which to initialize the buffer.
      *
      * @return The buffer if successful, otherwise nullptr.
      */
     using InitializeBufferFn = void(*)(skgpu::VertexWriter, size_t bufferSize);
     sk_sp<const GrGpuBuffer> findOrMakeStaticBuffer(GrGpuBufferType intendedType,
                                                     size_t size,
                                                     const skgpu::UniqueKey& key,
                                                     InitializeBufferFn);

     /**
      * Either finds and refs, or creates a static buffer with the given parameters and contents.
      *
      * @param intendedType    hint to the graphics subsystem about what the buffer will be used for.
      * @param size            minimum size of buffer to return.
      * @param data            optional data with which to initialize the buffer.
      * @param key             Key to be assigned to the buffer.
      *
      * @return The buffer if successful, otherwise nullptr.
      */
     sk_sp<const GrGpuBuffer> findOrMakeStaticBuffer(GrGpuBufferType intendedType,
                                                     size_t size,
                                                     const void* staticData,
                                                     const skgpu::UniqueKey& key);

     /**
      * Either finds and refs, or creates an index buffer with a repeating pattern for drawing
      * contiguous vertices of a repeated mesh. If the return is non-null, the caller owns a ref on
      * the returned GrBuffer.
      *
      * @param pattern     the pattern of indices to repeat
      * @param patternSize size in bytes of the pattern
      * @param reps        number of times to repeat the pattern
      * @param vertCount   number of vertices the pattern references
      * @param key         Key to be assigned to the index buffer.
      *
      * @return The index buffer if successful, otherwise nullptr.
      */
     sk_sp<const GrGpuBuffer> findOrCreatePatternedIndexBuffer(const uint16_t* pattern,
                                                               int patternSize,
                                                               int reps,
                                                               int vertCount,
                                                               const skgpu::UniqueKey& key) {
         if (auto buffer = this->findByUniqueKey<const GrGpuBuffer>(key)) {
             return buffer;
         }
         return this->createPatternedIndexBuffer(pattern, patternSize, reps, vertCount, &key);
     }

     /**
      * Returns an index buffer that can be used to render non-antialiased quads.
      * Each quad consumes 6 indices (0, 1, 2, 2, 1, 3) and 4 vertices.
      * Call MaxNumNonAAQuads to get the max allowed number of non-AA quads.
      * Draw with GrPrimitiveType::kTriangles
      * @ return the non-AA quad index buffer
      */
     sk_sp<const GrGpuBuffer> refNonAAQuadIndexBuffer() {
         if (!fNonAAQuadIndexBuffer) {
             fNonAAQuadIndexBuffer = this->createNonAAQuadIndexBuffer();
         }
         return fNonAAQuadIndexBuffer;
     }

     static int MaxNumNonAAQuads();
     static int NumVertsPerNonAAQuad();
     static int NumIndicesPerNonAAQuad();

     /**
      * Returns an index buffer that can be used to render antialiased quads.
      * Each quad consumes 30 indices and 8 vertices.
      * Call MaxNumAAQuads to get the max allowed number of AA quads.
      * Draw with GrPrimitiveType::kTriangles
      * @ return the AA quad index buffer
      */
     sk_sp<const GrGpuBuffer> refAAQuadIndexBuffer() {
         if (!fAAQuadIndexBuffer) {
             fAAQuadIndexBuffer = this->createAAQuadIndexBuffer();
         }
         return fAAQuadIndexBuffer;
     }

     static int MaxNumAAQuads();
     static int NumVertsPerAAQuad();
     static int NumIndicesPerAAQuad();

     /**
      * Returns a buffer.
      *
      * @param size            minimum size of buffer to return.
      * @param intendedType    hint to the graphics subsystem about what the buffer will be used for.
      * @param GrAccessPattern hint to the graphics subsystem about how the data will be accessed.
      * @param flags           see Flags enum.
      * @param data            optional data with which to initialize the buffer.
      *
      * @return the buffer if successful, otherwise nullptr.
      */
     sk_sp<GrGpuBuffer> createBuffer(size_t size,
                                     GrGpuBufferType intendedType,
                                     GrAccessPattern,
                                     const void* data = nullptr);

     /**
      * If passed in render target already has a stencil buffer on the specified surface, return
      * true. Otherwise attempt to attach one and return true on success.
      */
     bool attachStencilAttachment(GrRenderTarget* rt, bool useMSAASurface);

     sk_sp<GrAttachment> makeMSAAAttachment(SkISize dimensions,
                                            const GrBackendFormat& format,
                                            int sampleCnt,
                                            GrProtected isProtected,
                                            GrMemoryless isMemoryless);

     /**
      * Gets a GrAttachment that can be used for MSAA rendering. This attachment may be shared by
      * other users. Thus any renderpass that uses the attachment should not assume any specific
      * data at the start and should not try to save written data at the end. Ideally the render pass
      * should discard the data at the end.
      */
     sk_sp<GrAttachment> getDiscardableMSAAAttachment(SkISize dimensions,
                                                      const GrBackendFormat& format,
                                                      int sampleCnt,
                                                      GrProtected isProtected,
                                                      GrMemoryless memoryless);

     /**
      * Assigns a unique key to a resource. If the key is associated with another resource that
      * association is removed and replaced by this resource.
      */
     void assignUniqueKeyToResource(const skgpu::UniqueKey&, GrGpuResource*);

     std::unique_ptr<GrSemaphore> SK_WARN_UNUSED_RESULT makeSemaphore(bool isOwned = true);

     std::unique_ptr<GrSemaphore> wrapBackendSemaphore(const GrBackendSemaphore&,
                                                       GrSemaphoreWrapType,
                                                       GrWrapOwnership = kBorrow_GrWrapOwnership);

     void abandon() {
         fCache = nullptr;
         fGpu = nullptr;
     }

     uint32_t contextUniqueID() const { return fCache->contextUniqueID(); }
     const GrCaps* caps() const { return fCaps.get(); }
     bool overBudget() const { return fCache->overBudget(); }

     static SkISize MakeApprox(SkISize);

     inline GrResourceProviderPriv priv();
     inline const GrResourceProviderPriv priv() const;  // NOLINT(readability-const-return-type)

 private:
     sk_sp<GrGpuResource> findResourceByUniqueKey(const skgpu::UniqueKey&);

     /*
      * Try to find an existing scratch texture that exactly matches 'desc'. If successful
      * update the budgeting accordingly.
      */
     sk_sp<GrTexture> getExactScratch(SkISize dimensions,
                                      const GrBackendFormat&,
                                      GrTextureType,
                                      GrRenderable,
                                      int renderTargetSampleCnt,
                                      SkBudgeted,
                                      GrMipmapped,
                                      GrProtected);

     // Attempts to find a resource in the cache that exactly matches the SkISize. Failing that
     // it returns null. If non-null, the resulting msaa attachment is always budgeted.
     sk_sp<GrAttachment> refScratchMSAAAttachment(SkISize dimensions,
                                                  const GrBackendFormat&,
                                                  int sampleCnt,
                                                  GrProtected,
                                                  GrMemoryless memoryless);

     // Used to perform any conversions necessary to texel data before creating a texture with
     // existing data or uploading to a scratch texture.
     using TempLevels = SkAutoSTArray<14, GrMipLevel>;
     using TempLevelDatas = SkAutoSTArray<14, std::unique_ptr<char[]>>;
     GrColorType prepareLevels(const GrBackendFormat& format,
                               GrColorType,
                               SkISize baseSize,
                               const GrMipLevel texels[],
                               int mipLevelCount,
                               TempLevels*,
                               TempLevelDatas*) const;

     // GrResourceProvider may be asked to "create" a new texture with initial pixel data to populate
     // it. In implementation it may pull an existing texture from GrResourceCache and then write the
     // pixel data to the texture. It takes a width/height for the base level because we may be
     // using an approximate-sized scratch texture. On success the texture is returned and nullptr
     // on failure.
     sk_sp<GrTexture> writePixels(sk_sp<GrTexture> texture,
                                  GrColorType colorType,
                                  SkISize baseSize,
                                  const GrMipLevel texels[],
                                  int mipLevelCount) const;

     GrResourceCache* cache() { return fCache; }
     const GrResourceCache* cache() const { return fCache; }

     friend class GrResourceProviderPriv;

     // Method made available via GrResourceProviderPriv
     GrGpu* gpu() { return fGpu; }
     const GrGpu* gpu() const { return fGpu; }

     bool isAbandoned() const {
         SkASSERT(SkToBool(fGpu) == SkToBool(fCache));
         return !SkToBool(fCache);
     }

     sk_sp<const GrGpuBuffer> createPatternedIndexBuffer(const uint16_t* pattern,
                                                         int patternSize,
                                                         int reps,
                                                         int vertCount,
                                                         const skgpu::UniqueKey* key);

     sk_sp<const GrGpuBuffer> createNonAAQuadIndexBuffer();
     sk_sp<const GrGpuBuffer> createAAQuadIndexBuffer();

     GrResourceCache* fCache;
     GrGpu* fGpu;
     sk_sp<const GrCaps> fCaps;
     sk_sp<const GrGpuBuffer> fNonAAQuadIndexBuffer;
     sk_sp<const GrGpuBuffer> fAAQuadIndexBuffer;

     // In debug builds we guard against improper thread handling
     SkDEBUGCODE(mutable GrSingleOwner* fSingleOwner;)
 };

 #endif
	/*
	* Copyright 2015 Google Inc.
	*
	* Use of this source code is governed by a BSD-style license that can be
	* found in the LICENSE file.
	*/

	#ifndef GrResourceProvider_DEFINED
	#define GrResourceProvider_DEFINED

	#include "include/gpu/GrContextOptions.h"
	#include "include/private/SkImageInfoPriv.h"
	#include "src/gpu/GrGpuBuffer.h"
	#include "src/gpu/GrResourceCache.h"

	class GrAttachment;
	class GrBackendRenderTarget;
	class GrBackendSemaphore;
	class GrBackendTexture;
	class GrGpu;
	class GrMSAAAttachment;
	class GrPath;
	class GrRenderTarget;
	class GrResourceProviderPriv;
	class GrSemaphore;
	class GrSingleOwner;
	class GrTexture;
	struct GrVkDrawableInfo;

	class GrStyle;
	class SkDescriptor;
	class SkPath;
	class SkTypeface;

	namespace skgpu { struct VertexWriter; }

	/**
	* A factory for arbitrary resource types.
	*/
	class GrResourceProvider {
	public:
	GrResourceProvider(GrGpu, GrResourceCache, GrSingleOwner*);

	/**
	* Finds a resource in the cache, based on the specified key. Prior to calling this, the caller
	* must be sure that if a resource of exists in the cache with the given unique key then it is
	* of type T.
	*/
	template <typename T = GrGpuResource>
	typename std::enable_if<std::is_base_of<GrGpuResource, T>::value, sk_sp<T>>::type
	findByUniqueKey(const skgpu::UniqueKey& key) {
	return sk_sp<T>(static_cast<T*>(this->findResourceByUniqueKey(key).release()));
	}

	///////////////////////////////////////////////////////////////////////////
	// Textures

	/**
	* Finds a texture that approximately matches the descriptor. Will be at least as large in width
	* and height as desc specifies. If renderable is kYes then the GrTexture will also be a
	* GrRenderTarget. The texture's format and sample count will always match the request.
	* The contents of the texture are undefined.
	*/
	sk_sp<GrTexture> createApproxTexture(SkISize dimensions,
	const GrBackendFormat& format,
	GrTextureType textureType,
	GrRenderable renderable,
	int renderTargetSampleCnt,
	GrProtected isProtected);

	/** Create an exact fit texture with no initial data to upload. */
	sk_sp<GrTexture> createTexture(SkISize dimensions,
	const GrBackendFormat& format,
	GrTextureType textureType,
	GrRenderable renderable,
	int renderTargetSampleCnt,
	GrMipmapped mipMapped,
	SkBudgeted budgeted,
	GrProtected isProtected);

	/**
	* Create an exact fit texture with initial data to upload. The color type must be valid
	* for the format and also describe the texel data. This will ensure any conversions that
	* need to get applied to the data before upload are applied.
	*/
	sk_sp<GrTexture> createTexture(SkISize dimensions,
	const GrBackendFormat& format,
	GrTextureType textureType,
	GrColorType colorType,
	GrRenderable renderable,
	int renderTargetSampleCnt,
	SkBudgeted budgeted,
	GrMipMapped mipMapped,
	GrProtected isProtected,
	const GrMipLevel texels[]);

	/**
	* Create a potentially loose fit texture with the provided data. The color type must be valid
	* for the format and also describe the texel data. This will ensure any conversions that
	* need to get applied to the data before upload are applied.
	*/
	sk_sp<GrTexture> createTexture(SkISize dimensions,
	const GrBackendFormat&,
	GrTextureType textureType,
	GrColorType srcColorType,
	GrRenderable,
	int renderTargetSampleCnt,
	SkBudgeted,
	SkBackingFit,
	GrProtected,
	const GrMipLevel& mipLevel);

	/**
	* Search the cache for a scratch texture matching the provided arguments. Failing that
	* it returns null. If non-null, the resulting texture is always budgeted.
	*/
	sk_sp<GrTexture> findAndRefScratchTexture(const skgpu::ScratchKey&);
	sk_sp<GrTexture> findAndRefScratchTexture(SkISize dimensions,
	const GrBackendFormat&,
	GrTextureType textureType,
	GrRenderable,
	int renderTargetSampleCnt,
	GrMipmapped,
	GrProtected);

	/**
	* Creates a compressed texture. The GrGpu must support the SkImageImage::Compression type.
	* It will not be renderable.
	*/
	sk_sp<GrTexture> createCompressedTexture(SkISize dimensions,
	const GrBackendFormat&,
	SkBudgeted,
	GrMipmapped,
	GrProtected,
	SkData* data);

	///////////////////////////////////////////////////////////////////////////
	// Wrapped Backend Surfaces

	/**
	* Wraps an existing texture with a GrTexture object.
	*
	* GrIOType must either be kRead or kRW. kRead blocks any operations that would modify the
	* pixels (e.g. dst for a copy, regenerating MIP levels, write pixels).
	*
	* OpenGL: if the object is a texture Gr may change its GL texture params
	* when it is drawn.
	*
	* @return GrTexture object or NULL on failure.
	*/
	sk_sp<GrTexture> wrapBackendTexture(const GrBackendTexture& tex,
	GrWrapOwnership,
	GrWrapCacheable,
	GrIOType);

	sk_sp<GrTexture> wrapCompressedBackendTexture(const GrBackendTexture& tex,
	GrWrapOwnership,
	GrWrapCacheable);

	/**
	* This makes the backend texture be renderable. If sampleCnt is > 1 and the underlying API
	* uses separate MSAA render buffers then a MSAA render buffer is created that resolves
	* to the texture.
	*/
	sk_sp<GrTexture> wrapRenderableBackendTexture(const GrBackendTexture& tex,
	int sampleCnt,
	GrWrapOwnership,
	GrWrapCacheable);

	/**
	* Wraps an existing render target with a GrRenderTarget object. It is
	* similar to wrapBackendTexture but can be used to draw into surfaces
	* that are not also textures (e.g. FBO 0 in OpenGL, or an MSAA buffer that
	* the client will resolve to a texture). Currently wrapped render targets
	* always use the kBorrow_GrWrapOwnership and GrWrapCacheable::kNo semantics.
	*
	* @return GrRenderTarget object or NULL on failure.
	*/
	sk_sp<GrRenderTarget> wrapBackendRenderTarget(const GrBackendRenderTarget&);

	sk_sp<GrRenderTarget> wrapVulkanSecondaryCBAsRenderTarget(const SkImageInfo&,
	const GrVkDrawableInfo&);

	static const int kMinScratchTextureSize;

	/**
	* Either finds and refs a buffer with the given unique key, or creates a new new, fills its
	* contents with the InitializeBufferDataFn() callback, and assigns it the unique key.
	*
	* @param intendedType hint to the graphics subsystem about how the buffer will be used.
	* @param size minimum size of buffer to return.
	* @param key Key to be assigned to the buffer.
	* @param InitializeBufferFn callback with which to initialize the buffer.
	*
	* @return The buffer if successful, otherwise nullptr.
	*/
	using InitializeBufferFn = void(*)(skgpu::VertexWriter, size_t bufferSize);
	sk_sp<const GrGpuBuffer> findOrMakeStaticBuffer(GrGpuBufferType intendedType,
	size_t size,
	const skgpu::UniqueKey& key,
	InitializeBufferFn);

	/**
	* Either finds and refs, or creates a static buffer with the given parameters and contents.
	*
	* @param intendedType hint to the graphics subsystem about what the buffer will be used for.
	* @param size minimum size of buffer to return.
	* @param data optional data with which to initialize the buffer.
	* @param key Key to be assigned to the buffer.
	*
	* @return The buffer if successful, otherwise nullptr.
	*/
	sk_sp<const GrGpuBuffer> findOrMakeStaticBuffer(GrGpuBufferType intendedType,
	size_t size,
	const void* staticData,
	const skgpu::UniqueKey& key);

	/**
	* Either finds and refs, or creates an index buffer with a repeating pattern for drawing
	* contiguous vertices of a repeated mesh. If the return is non-null, the caller owns a ref on
	* the returned GrBuffer.
	*
	* @param pattern the pattern of indices to repeat
	* @param patternSize size in bytes of the pattern
	* @param reps number of times to repeat the pattern
	* @param vertCount number of vertices the pattern references
	* @param key Key to be assigned to the index buffer.
	*
	* @return The index buffer if successful, otherwise nullptr.
	*/
	sk_sp<const GrGpuBuffer> findOrCreatePatternedIndexBuffer(const uint16_t* pattern,
	int patternSize,
	int reps,
	int vertCount,
	const skgpu::UniqueKey& key) {
	if (auto buffer = this->findByUniqueKey<const GrGpuBuffer>(key)) {
	return buffer;
	}
	return this->createPatternedIndexBuffer(pattern, patternSize, reps, vertCount, &key);
	}

	/**
	* Returns an index buffer that can be used to render non-antialiased quads.
	* Each quad consumes 6 indices (0, 1, 2, 2, 1, 3) and 4 vertices.
	* Call MaxNumNonAAQuads to get the max allowed number of non-AA quads.
	* Draw with GrPrimitiveType::kTriangles
	* @ return the non-AA quad index buffer
	*/
	sk_sp<const GrGpuBuffer> refNonAAQuadIndexBuffer() {
	if (!fNonAAQuadIndexBuffer) {
	fNonAAQuadIndexBuffer = this->createNonAAQuadIndexBuffer();
	}
	return fNonAAQuadIndexBuffer;
	}

	static int MaxNumNonAAQuads();
	static int NumVertsPerNonAAQuad();
	static int NumIndicesPerNonAAQuad();

	/**
	* Returns an index buffer that can be used to render antialiased quads.
	* Each quad consumes 30 indices and 8 vertices.
	* Call MaxNumAAQuads to get the max allowed number of AA quads.
	* Draw with GrPrimitiveType::kTriangles
	* @ return the AA quad index buffer
	*/
	sk_sp<const GrGpuBuffer> refAAQuadIndexBuffer() {
	if (!fAAQuadIndexBuffer) {
	fAAQuadIndexBuffer = this->createAAQuadIndexBuffer();
	}
	return fAAQuadIndexBuffer;
	}

	static int MaxNumAAQuads();
	static int NumVertsPerAAQuad();
	static int NumIndicesPerAAQuad();

	/**
	* Returns a buffer.
	*
	* @param size minimum size of buffer to return.
	* @param intendedType hint to the graphics subsystem about what the buffer will be used for.
	* @param GrAccessPattern hint to the graphics subsystem about how the data will be accessed.
	* @param flags see Flags enum.
	* @param data optional data with which to initialize the buffer.
	*
	* @return the buffer if successful, otherwise nullptr.
	*/
	sk_sp<GrGpuBuffer> createBuffer(size_t size,
	GrGpuBufferType intendedType,
	GrAccessPattern,
	const void* data = nullptr);

	/**
	* If passed in render target already has a stencil buffer on the specified surface, return
	* true. Otherwise attempt to attach one and return true on success.
	*/
	bool attachStencilAttachment(GrRenderTarget* rt, bool useMSAASurface);

	sk_sp<GrAttachment> makeMSAAAttachment(SkISize dimensions,
	const GrBackendFormat& format,
	int sampleCnt,
	GrProtected isProtected,
	GrMemoryless isMemoryless);

	/**
	* Gets a GrAttachment that can be used for MSAA rendering. This attachment may be shared by
	* other users. Thus any renderpass that uses the attachment should not assume any specific
	* data at the start and should not try to save written data at the end. Ideally the render pass
	* should discard the data at the end.
	*/
	sk_sp<GrAttachment> getDiscardableMSAAAttachment(SkISize dimensions,
	const GrBackendFormat& format,
	int sampleCnt,
	GrProtected isProtected,
	GrMemoryless memoryless);

	/**
	* Assigns a unique key to a resource. If the key is associated with another resource that
	* association is removed and replaced by this resource.
	*/
	void assignUniqueKeyToResource(const skgpu::UniqueKey&, GrGpuResource*);

	std::unique_ptr<GrSemaphore> SK_WARN_UNUSED_RESULT makeSemaphore(bool isOwned = true);

	std::unique_ptr<GrSemaphore> wrapBackendSemaphore(const GrBackendSemaphore&,
	GrSemaphoreWrapType,
	GrWrapOwnership = kBorrow_GrWrapOwnership);

	void abandon() {
	fCache = nullptr;
	fGpu = nullptr;
	}

	uint32_t contextUniqueID() const { return fCache->contextUniqueID(); }
	const GrCaps* caps() const { return fCaps.get(); }
	bool overBudget() const { return fCache->overBudget(); }

	static SkISize MakeApprox(SkISize);

	inline GrResourceProviderPriv priv();
	inline const GrResourceProviderPriv priv() const; // NOLINT(readability-const-return-type)

	private:
	sk_sp<GrGpuResource> findResourceByUniqueKey(const skgpu::UniqueKey&);

	/*
	* Try to find an existing scratch texture that exactly matches 'desc'. If successful
	* update the budgeting accordingly.
	*/
	sk_sp<GrTexture> getExactScratch(SkISize dimensions,
	const GrBackendFormat&,
	GrTextureType,
	GrRenderable,
	int renderTargetSampleCnt,
	SkBudgeted,
	GrMipmapped,
	GrProtected);

	// Attempts to find a resource in the cache that exactly matches the SkISize. Failing that
	// it returns null. If non-null, the resulting msaa attachment is always budgeted.
	sk_sp<GrAttachment> refScratchMSAAAttachment(SkISize dimensions,
	const GrBackendFormat&,
	int sampleCnt,
	GrProtected,
	GrMemoryless memoryless);

	// Used to perform any conversions necessary to texel data before creating a texture with
	// existing data or uploading to a scratch texture.
	using TempLevels = SkAutoSTArray<14, GrMipLevel>;
	using TempLevelDatas = SkAutoSTArray<14, std::unique_ptr<char[]>>;
	GrColorType prepareLevels(const GrBackendFormat& format,
	GrColorType,
	SkISize baseSize,
	const GrMipLevel texels[],
	int mipLevelCount,
	TempLevels*,
	TempLevelDatas*) const;

	// GrResourceProvider may be asked to "create" a new texture with initial pixel data to populate
	// it. In implementation it may pull an existing texture from GrResourceCache and then write the
	// pixel data to the texture. It takes a width/height for the base level because we may be
	// using an approximate-sized scratch texture. On success the texture is returned and nullptr
	// on failure.
	sk_sp<GrTexture> writePixels(sk_sp<GrTexture> texture,
	GrColorType colorType,
	SkISize baseSize,
	const GrMipLevel texels[],
	int mipLevelCount) const;

	GrResourceCache* cache() { return fCache; }
	const GrResourceCache* cache() const { return fCache; }

	friend class GrResourceProviderPriv;

	// Method made available via GrResourceProviderPriv
	GrGpu* gpu() { return fGpu; }
	const GrGpu* gpu() const { return fGpu; }

	bool isAbandoned() const {
	SkASSERT(SkToBool(fGpu) == SkToBool(fCache));
	return !SkToBool(fCache);
	}

	sk_sp<const GrGpuBuffer> createPatternedIndexBuffer(const uint16_t* pattern,
	int patternSize,
	int reps,
	int vertCount,
	const skgpu::UniqueKey* key);

	sk_sp<const GrGpuBuffer> createNonAAQuadIndexBuffer();
	sk_sp<const GrGpuBuffer> createAAQuadIndexBuffer();

	GrResourceCache* fCache;
	GrGpu* fGpu;
	sk_sp<const GrCaps> fCaps;
	sk_sp<const GrGpuBuffer> fNonAAQuadIndexBuffer;
	sk_sp<const GrGpuBuffer> fAAQuadIndexBuffer;

	// In debug builds we guard against improper thread handling
	SkDEBUGCODE(mutable GrSingleOwner* fSingleOwner;)
	};

	#endif