src/gpu/d3d/GrD3DGpu.cpp

/*
 * Copyright 2020 Google LLC
 *
 * Use of this source code is governed by a BSD-style license that can be
 * found in the LICENSE file.
 */

#include "src/gpu/d3d/GrD3DGpu.h"

#include "include/gpu/GrBackendSurface.h"
#include "include/gpu/d3d/GrD3DBackendContext.h"
#include "src/core/SkConvertPixels.h"
#include "src/core/SkMipmap.h"
#include "src/gpu/GrBackendUtils.h"
#include "src/gpu/GrDataUtils.h"
#include "src/gpu/GrTexture.h"
#include "src/gpu/GrThreadSafePipelineBuilder.h"
#include "src/gpu/d3d/GrD3DAMDMemoryAllocator.h"
#include "src/gpu/d3d/GrD3DAttachment.h"
#include "src/gpu/d3d/GrD3DBuffer.h"
#include "src/gpu/d3d/GrD3DCaps.h"
#include "src/gpu/d3d/GrD3DOpsRenderPass.h"
#include "src/gpu/d3d/GrD3DSemaphore.h"
#include "src/gpu/d3d/GrD3DTexture.h"
#include "src/gpu/d3d/GrD3DTextureRenderTarget.h"
#include "src/gpu/d3d/GrD3DUtil.h"
#include "src/sksl/SkSLCompiler.h"

#if GR_TEST_UTILS
#include <DXProgrammableCapture.h>
#endif

GrThreadSafePipelineBuilder* GrD3DGpu::pipelineBuilder() {
    return nullptr;
}

sk_sp<GrThreadSafePipelineBuilder> GrD3DGpu::refPipelineBuilder() {
    return nullptr;
}


sk_sp<GrGpu> GrD3DGpu::Make(const GrD3DBackendContext& backendContext,
                            const GrContextOptions& contextOptions, GrDirectContext* direct) {
    sk_sp<GrD3DMemoryAllocator> memoryAllocator = backendContext.fMemoryAllocator;
    if (!memoryAllocator) {
        // We were not given a memory allocator at creation
        memoryAllocator = GrD3DAMDMemoryAllocator::Make(
                backendContext.fAdapter.get(), backendContext.fDevice.get());
    }
    if (!memoryAllocator) {
        SkDEBUGFAIL("No supplied Direct3D memory allocator and unable to create one internally.");
        return nullptr;
    }

    return sk_sp<GrGpu>(new GrD3DGpu(direct, contextOptions, backendContext, memoryAllocator));
}

// This constant determines how many OutstandingCommandLists are allocated together as a block in
// the deque. As such it needs to balance allocating too much memory vs. incurring
// allocation/deallocation thrashing. It should roughly correspond to the max number of outstanding
// command lists we expect to see.
static const int kDefaultOutstandingAllocCnt = 8;

// constants have to be aligned to 256
constexpr int kConstantAlignment = 256;

GrD3DGpu::GrD3DGpu(GrDirectContext* direct, const GrContextOptions& contextOptions,
                   const GrD3DBackendContext& backendContext,
                   sk_sp<GrD3DMemoryAllocator> allocator)
        : INHERITED(direct)
        , fDevice(backendContext.fDevice)
        , fQueue(backendContext.fQueue)
        , fMemoryAllocator(std::move(allocator))
        , fResourceProvider(this)
        , fStagingBufferManager(this)
        , fConstantsRingBuffer(this, 128 * 1024, kConstantAlignment, GrGpuBufferType::kVertex)
        , fOutstandingCommandLists(sizeof(OutstandingCommandList), kDefaultOutstandingAllocCnt) {
    this->initCapsAndCompiler(sk_make_sp<GrD3DCaps>(contextOptions,
                                                    backendContext.fAdapter.get(),
                                                    backendContext.fDevice.get()));

    fCurrentDirectCommandList = fResourceProvider.findOrCreateDirectCommandList();
    SkASSERT(fCurrentDirectCommandList);

    SkASSERT(fCurrentFenceValue == 0);
    GR_D3D_CALL_ERRCHECK(fDevice->CreateFence(fCurrentFenceValue, D3D12_FENCE_FLAG_NONE,
                                              IID_PPV_ARGS(&fFence)));

#if GR_TEST_UTILS
    HRESULT getAnalysis = DXGIGetDebugInterface1(0, IID_PPV_ARGS(&fGraphicsAnalysis));
    if (FAILED(getAnalysis)) {
        fGraphicsAnalysis = nullptr;
    }
#endif
}

GrD3DGpu::~GrD3DGpu() {
    this->destroyResources();
}

void GrD3DGpu::destroyResources() {
    if (fCurrentDirectCommandList) {
        fCurrentDirectCommandList->close();
        fCurrentDirectCommandList->reset();
    }

    // We need to make sure everything has finished on the queue.
    this->waitForQueueCompletion();

    SkDEBUGCODE(uint64_t fenceValue = fFence->GetCompletedValue();)

    // We used a placement new for each object in fOutstandingCommandLists, so we're responsible
    // for calling the destructor on each of them as well.
    while (!fOutstandingCommandLists.empty()) {
        OutstandingCommandList* list = (OutstandingCommandList*)fOutstandingCommandLists.front();
        SkASSERT(list->fFenceValue <= fenceValue);
        // No reason to recycle the command lists since we are destroying all resources anyways.
        list->~OutstandingCommandList();
        fOutstandingCommandLists.pop_front();
    }

    fStagingBufferManager.reset();

    fResourceProvider.destroyResources();
}

GrOpsRenderPass* GrD3DGpu::onGetOpsRenderPass(
        GrRenderTarget* rt,
        bool /*useMSAASurface*/,
        GrAttachment*,
        GrSurfaceOrigin origin,
        const SkIRect& bounds,
        const GrOpsRenderPass::LoadAndStoreInfo& colorInfo,
        const GrOpsRenderPass::StencilLoadAndStoreInfo& stencilInfo,
        const SkTArray<GrSurfaceProxy*, true>& sampledProxies,
        GrXferBarrierFlags renderPassXferBarriers) {
    if (!fCachedOpsRenderPass) {
        fCachedOpsRenderPass.reset(new GrD3DOpsRenderPass(this));
    }

    if (!fCachedOpsRenderPass->set(rt, origin, bounds, colorInfo, stencilInfo, sampledProxies)) {
        return nullptr;
    }
    return fCachedOpsRenderPass.get();
}

bool GrD3DGpu::submitDirectCommandList(SyncQueue sync) {
    SkASSERT(fCurrentDirectCommandList);

    fResourceProvider.prepForSubmit();
    for (int i = 0; i < fMipmapCPUDescriptors.count(); ++i) {
        fResourceProvider.recycleShaderView(fMipmapCPUDescriptors[i]);
    }
    fMipmapCPUDescriptors.reset();

    GrD3DDirectCommandList::SubmitResult result = fCurrentDirectCommandList->submit(fQueue.get());
    if (result == GrD3DDirectCommandList::SubmitResult::kFailure) {
        return false;
    } else if (result == GrD3DDirectCommandList::SubmitResult::kNoWork) {
        if (sync == SyncQueue::kForce) {
            this->waitForQueueCompletion();
            this->checkForFinishedCommandLists();
        }
        return true;
    }

    // We just submitted the command list so make sure all GrD3DPipelineState's mark their cached
    // uniform data as dirty.
    fResourceProvider.markPipelineStateUniformsDirty();

    GrFence fence = this->insertFence();
    new (fOutstandingCommandLists.push_back()) OutstandingCommandList(
            std::move(fCurrentDirectCommandList), fence);

    if (sync == SyncQueue::kForce) {
        this->waitForQueueCompletion();
    }

    fCurrentDirectCommandList = fResourceProvider.findOrCreateDirectCommandList();

    // This should be done after we have a new command list in case the freeing of any resources
    // held by a finished command list causes us send a new command to the gpu (like changing the
    // resource state.
    this->checkForFinishedCommandLists();

    SkASSERT(fCurrentDirectCommandList);
    return true;
}

void GrD3DGpu::checkForFinishedCommandLists() {
    uint64_t currentFenceValue = fFence->GetCompletedValue();

    // Iterate over all the outstanding command lists to see if any have finished. The commands
    // lists are in order from oldest to newest, so we start at the front to check if their fence
    // value is less than the last signaled value. If so we pop it off and move onto the next.
    // Repeat till we find a command list that has not finished yet (and all others afterwards are
    // also guaranteed to not have finished).
    OutstandingCommandList* front = (OutstandingCommandList*)fOutstandingCommandLists.front();
    while (front && front->fFenceValue <= currentFenceValue) {
        std::unique_ptr<GrD3DDirectCommandList> currList(std::move(front->fCommandList));
        // Since we used placement new we are responsible for calling the destructor manually.
        front->~OutstandingCommandList();
        fOutstandingCommandLists.pop_front();
        fResourceProvider.recycleDirectCommandList(std::move(currList));
        front = (OutstandingCommandList*)fOutstandingCommandLists.front();
    }
}

void GrD3DGpu::waitForQueueCompletion() {
    if (fFence->GetCompletedValue() < fCurrentFenceValue) {
        HANDLE fenceEvent;
        fenceEvent = CreateEvent(nullptr, FALSE, FALSE, nullptr);
        SkASSERT(fenceEvent);
        GR_D3D_CALL_ERRCHECK(fFence->SetEventOnCompletion(fCurrentFenceValue, fenceEvent));
        WaitForSingleObject(fenceEvent, INFINITE);
        CloseHandle(fenceEvent);
    }
}

void GrD3DGpu::submit(GrOpsRenderPass* renderPass) {
    SkASSERT(fCachedOpsRenderPass.get() == renderPass);

    fCachedOpsRenderPass->submit();
    fCachedOpsRenderPass.reset();
}

void GrD3DGpu::endRenderPass(GrRenderTarget* target, GrSurfaceOrigin origin,
                             const SkIRect& bounds) {
    this->didWriteToSurface(target, origin, &bounds);
}

void GrD3DGpu::addFinishedProc(GrGpuFinishedProc finishedProc,
                               GrGpuFinishedContext finishedContext) {
    SkASSERT(finishedProc);
    this->addFinishedCallback(GrRefCntedCallback::Make(finishedProc, finishedContext));
}

void GrD3DGpu::addFinishedCallback(sk_sp<GrRefCntedCallback> finishedCallback) {
    SkASSERT(finishedCallback);
    // Besides the current command list, we also add the finishedCallback to the newest outstanding
    // command list. Our contract for calling the proc is that all previous submitted command lists
    // have finished when we call it. However, if our current command list has no work when it is
    // flushed it will drop its ref to the callback immediately. But the previous work may not have
    // finished. It is safe to only add the proc to the newest outstanding commandlist cause that
    // must finish after all previously submitted command lists.
    OutstandingCommandList* back = (OutstandingCommandList*)fOutstandingCommandLists.back();
    if (back) {
        back->fCommandList->addFinishedCallback(finishedCallback);
    }
    fCurrentDirectCommandList->addFinishedCallback(std::move(finishedCallback));
}

sk_sp<GrD3DTexture> GrD3DGpu::createD3DTexture(SkISize dimensions,
                                               DXGI_FORMAT dxgiFormat,
                                               GrRenderable renderable,
                                               int renderTargetSampleCnt,
                                               SkBudgeted budgeted,
                                               GrProtected isProtected,
                                               int mipLevelCount,
                                               GrMipmapStatus mipmapStatus) {
    D3D12_RESOURCE_FLAGS usageFlags = D3D12_RESOURCE_FLAG_NONE;
    if (renderable == GrRenderable::kYes) {
        usageFlags |= D3D12_RESOURCE_FLAG_ALLOW_RENDER_TARGET;
    }

    // This desc refers to a texture that will be read by the client. Thus even if msaa is
    // requested, this describes the resolved texture. Therefore we always have samples set
    // to 1.
    SkASSERT(mipLevelCount > 0);
    D3D12_RESOURCE_DESC resourceDesc = {};
    resourceDesc.Dimension = D3D12_RESOURCE_DIMENSION_TEXTURE2D;
    // TODO: will use 4MB alignment for MSAA textures and 64KB for everything else
    //       might want to manually set alignment to 4KB for smaller textures
    resourceDesc.Alignment = 0;
    resourceDesc.Width = dimensions.fWidth;
    resourceDesc.Height = dimensions.fHeight;
    resourceDesc.DepthOrArraySize = 1;
    resourceDesc.MipLevels = mipLevelCount;
    resourceDesc.Format = dxgiFormat;
    resourceDesc.SampleDesc.Count = 1;
    resourceDesc.SampleDesc.Quality = DXGI_STANDARD_MULTISAMPLE_QUALITY_PATTERN;
    resourceDesc.Layout = D3D12_TEXTURE_LAYOUT_UNKNOWN;  // use driver-selected swizzle
    resourceDesc.Flags = usageFlags;

    if (renderable == GrRenderable::kYes) {
        return GrD3DTextureRenderTarget::MakeNewTextureRenderTarget(
                this, budgeted, dimensions, renderTargetSampleCnt, resourceDesc, isProtected,
                mipmapStatus);
    } else {
        return GrD3DTexture::MakeNewTexture(this, budgeted, dimensions, resourceDesc, isProtected,
                                            mipmapStatus);
    }
}

sk_sp<GrTexture> GrD3DGpu::onCreateTexture(SkISize dimensions,
                                           const GrBackendFormat& format,
                                           GrRenderable renderable,
                                           int renderTargetSampleCnt,
                                           SkBudgeted budgeted,
                                           GrProtected isProtected,
                                           int mipLevelCount,
                                           uint32_t levelClearMask) {
    DXGI_FORMAT dxgiFormat;
    SkAssertResult(format.asDxgiFormat(&dxgiFormat));
    SkASSERT(!GrDxgiFormatIsCompressed(dxgiFormat));

    GrMipmapStatus mipmapStatus = mipLevelCount > 1 ? GrMipmapStatus::kDirty
                                                    : GrMipmapStatus::kNotAllocated;

    sk_sp<GrD3DTexture> tex = this->createD3DTexture(dimensions, dxgiFormat, renderable,
                                                     renderTargetSampleCnt, budgeted, isProtected,
                                                     mipLevelCount, mipmapStatus);
    if (!tex) {
        return nullptr;
    }

    if (levelClearMask) {
        // TODO
    }

    return std::move(tex);
}

static void copy_compressed_data(char* mapPtr, DXGI_FORMAT dxgiFormat,
                                 D3D12_PLACED_SUBRESOURCE_FOOTPRINT* placedFootprints,
                                 UINT* numRows, UINT64* rowSizeInBytes,
                                 const void* compressedData, int numMipLevels) {
    SkASSERT(compressedData && numMipLevels);
    SkASSERT(GrDxgiFormatIsCompressed(dxgiFormat));
    SkASSERT(mapPtr);

    const char* src = static_cast<const char*>(compressedData);
    for (int currentMipLevel = 0; currentMipLevel < numMipLevels; currentMipLevel++) {
        // copy data into the buffer, skipping any trailing bytes
        char* dst = mapPtr + placedFootprints[currentMipLevel].Offset;
        SkRectMemcpy(dst, placedFootprints[currentMipLevel].Footprint.RowPitch,
                     src, rowSizeInBytes[currentMipLevel], rowSizeInBytes[currentMipLevel],
                     numRows[currentMipLevel]);
        src += numRows[currentMipLevel] * rowSizeInBytes[currentMipLevel];
    }
}

sk_sp<GrTexture> GrD3DGpu::onCreateCompressedTexture(SkISize dimensions,
                                                     const GrBackendFormat& format,
                                                     SkBudgeted budgeted,
                                                     GrMipmapped mipMapped,
                                                     GrProtected isProtected,
                                                     const void* data, size_t dataSize) {
    DXGI_FORMAT dxgiFormat;
    SkAssertResult(format.asDxgiFormat(&dxgiFormat));
    SkASSERT(GrDxgiFormatIsCompressed(dxgiFormat));

    SkDEBUGCODE(SkImage::CompressionType compression = GrBackendFormatToCompressionType(format));
    SkASSERT(dataSize == SkCompressedFormatDataSize(compression, dimensions,
                                                    mipMapped == GrMipmapped::kYes));

    int mipLevelCount = 1;
    if (mipMapped == GrMipmapped::kYes) {
        mipLevelCount = SkMipmap::ComputeLevelCount(dimensions.width(), dimensions.height()) + 1;
    }
    GrMipmapStatus mipmapStatus = mipLevelCount > 1 ? GrMipmapStatus::kValid
                                                    : GrMipmapStatus::kNotAllocated;

    sk_sp<GrD3DTexture> d3dTex = this->createD3DTexture(dimensions, dxgiFormat, GrRenderable::kNo,
                                                     1, budgeted, isProtected,
                                                     mipLevelCount, mipmapStatus);
    if (!d3dTex) {
        return nullptr;
    }

    ID3D12Resource* d3dResource = d3dTex->d3dResource();
    SkASSERT(d3dResource);
    D3D12_RESOURCE_DESC desc = d3dResource->GetDesc();
    // Either upload only the first miplevel or all miplevels
    SkASSERT(1 == mipLevelCount || mipLevelCount == (int)desc.MipLevels);

    SkAutoTMalloc<D3D12_PLACED_SUBRESOURCE_FOOTPRINT> placedFootprints(mipLevelCount);
    SkAutoTMalloc<UINT> numRows(mipLevelCount);
    SkAutoTMalloc<UINT64> rowSizeInBytes(mipLevelCount);
    UINT64 combinedBufferSize;
    // We reset the width and height in the description to match our subrectangle size
    // so we don't end up allocating more space than we need.
    desc.Width = dimensions.width();
    desc.Height = dimensions.height();
    fDevice->GetCopyableFootprints(&desc, 0, mipLevelCount, 0, placedFootprints.get(),
                                   numRows.get(), rowSizeInBytes.get(), &combinedBufferSize);
    SkASSERT(combinedBufferSize);

    GrStagingBufferManager::Slice slice = fStagingBufferManager.allocateStagingBufferSlice(
            combinedBufferSize, D3D12_TEXTURE_DATA_PLACEMENT_ALIGNMENT);
    if (!slice.fBuffer) {
        return nullptr;
    }

    char* bufferData = (char*)slice.fOffsetMapPtr;

    copy_compressed_data(bufferData, desc.Format, placedFootprints.get(), numRows.get(),
                         rowSizeInBytes.get(), data, mipLevelCount);

    // Update the offsets in the footprints to be relative to the slice's offset
    for (int i = 0; i < mipLevelCount; ++i) {
        placedFootprints[i].Offset += slice.fOffset;
    }

    ID3D12Resource* d3dBuffer = static_cast<GrD3DBuffer*>(slice.fBuffer)->d3dResource();
    fCurrentDirectCommandList->copyBufferToTexture(d3dBuffer, d3dTex.get(), mipLevelCount,
                                                   placedFootprints.get(), 0, 0);

    return std::move(d3dTex);
}

static int get_surface_sample_cnt(GrSurface* surf) {
    if (const GrRenderTarget* rt = surf->asRenderTarget()) {
        return rt->numSamples();
    }
    return 0;
}

bool GrD3DGpu::onCopySurface(GrSurface* dst, GrSurface* src, const SkIRect& srcRect,
                   const SkIPoint& dstPoint) {

    if (src->isProtected() && !dst->isProtected()) {
        SkDebugf("Can't copy from protected memory to non-protected");
        return false;
    }

    int dstSampleCnt = get_surface_sample_cnt(dst);
    int srcSampleCnt = get_surface_sample_cnt(src);

    GrD3DTextureResource* dstTexResource;
    GrD3DTextureResource* srcTexResource;
    GrRenderTarget* dstRT = dst->asRenderTarget();
    if (dstRT) {
        GrD3DRenderTarget* d3dRT = static_cast<GrD3DRenderTarget*>(dstRT);
        dstTexResource = d3dRT->numSamples() > 1 ? d3dRT->msaaTextureResource() : d3dRT;
    } else {
        SkASSERT(dst->asTexture());
        dstTexResource = static_cast<GrD3DTexture*>(dst->asTexture());
    }
    GrRenderTarget* srcRT = src->asRenderTarget();
    if (srcRT) {
        GrD3DRenderTarget* d3dRT = static_cast<GrD3DRenderTarget*>(srcRT);
        srcTexResource = d3dRT->numSamples() > 1 ? d3dRT->msaaTextureResource() : d3dRT;
    } else {
        SkASSERT(src->asTexture());
        srcTexResource = static_cast<GrD3DTexture*>(src->asTexture());
    }

    DXGI_FORMAT dstFormat = dstTexResource->dxgiFormat();
    DXGI_FORMAT srcFormat = srcTexResource->dxgiFormat();

    if (this->d3dCaps().canCopyAsResolve(dstFormat, dstSampleCnt, srcFormat, srcSampleCnt)) {
        this->copySurfaceAsResolve(dst, src, srcRect, dstPoint);
        return true;
    }

    if (this->d3dCaps().canCopyTexture(dstFormat, dstSampleCnt, srcFormat, srcSampleCnt)) {
        this->copySurfaceAsCopyTexture(dst, src, dstTexResource, srcTexResource, srcRect, dstPoint);
        return true;
    }

    return false;
}

void GrD3DGpu::copySurfaceAsCopyTexture(GrSurface* dst, GrSurface* src,
                                        GrD3DTextureResource* dstResource,
                                        GrD3DTextureResource* srcResource,
                                        const SkIRect& srcRect, const SkIPoint& dstPoint) {
#ifdef SK_DEBUG
    int dstSampleCnt = get_surface_sample_cnt(dst);
    int srcSampleCnt = get_surface_sample_cnt(src);
    DXGI_FORMAT dstFormat = dstResource->dxgiFormat();
    DXGI_FORMAT srcFormat;
    SkAssertResult(dst->backendFormat().asDxgiFormat(&srcFormat));
    SkASSERT(this->d3dCaps().canCopyTexture(dstFormat, dstSampleCnt, srcFormat, srcSampleCnt));
#endif
    if (src->isProtected() && !dst->isProtected()) {
        SkDebugf("Can't copy from protected memory to non-protected");
        return;
    }

    dstResource->setResourceState(this, D3D12_RESOURCE_STATE_COPY_DEST);
    srcResource->setResourceState(this, D3D12_RESOURCE_STATE_COPY_SOURCE);

    D3D12_TEXTURE_COPY_LOCATION dstLocation = {};
    dstLocation.pResource = dstResource->d3dResource();
    dstLocation.Type = D3D12_TEXTURE_COPY_TYPE_SUBRESOURCE_INDEX;
    dstLocation.SubresourceIndex = 0;

    D3D12_TEXTURE_COPY_LOCATION srcLocation = {};
    srcLocation.pResource = srcResource->d3dResource();
    srcLocation.Type = D3D12_TEXTURE_COPY_TYPE_SUBRESOURCE_INDEX;
    srcLocation.SubresourceIndex = 0;

    D3D12_BOX srcBox = {};
    srcBox.left = srcRect.fLeft;
    srcBox.top = srcRect.fTop;
    srcBox.right = srcRect.fRight;
    srcBox.bottom = srcRect.fBottom;
    srcBox.front = 0;
    srcBox.back = 1;
    // TODO: use copyResource if copying full resource and sizes match
    fCurrentDirectCommandList->copyTextureRegionToTexture(dstResource->resource(),
                                                          &dstLocation,
                                                          dstPoint.fX, dstPoint.fY,
                                                          srcResource->resource(),
                                                          &srcLocation,
                                                          &srcBox);

    SkIRect dstRect = SkIRect::MakeXYWH(dstPoint.fX, dstPoint.fY,
                                        srcRect.width(), srcRect.height());
    // The rect is already in device space so we pass in kTopLeft so no flip is done.
    this->didWriteToSurface(dst, kTopLeft_GrSurfaceOrigin, &dstRect);
}

void GrD3DGpu::copySurfaceAsResolve(GrSurface* dst, GrSurface* src, const SkIRect& srcRect,
                                    const SkIPoint& dstPoint) {
    GrD3DRenderTarget* srcRT = static_cast<GrD3DRenderTarget*>(src->asRenderTarget());
    SkASSERT(srcRT);

    this->resolveTexture(dst, dstPoint.fX, dstPoint.fY, srcRT, srcRect);
    SkIRect dstRect = SkIRect::MakeXYWH(dstPoint.fX, dstPoint.fY,
                                        srcRect.width(), srcRect.height());
    // The rect is already in device space so we pass in kTopLeft so no flip is done.
    this->didWriteToSurface(dst, kTopLeft_GrSurfaceOrigin, &dstRect);
}

void GrD3DGpu::resolveTexture(GrSurface* dst, int32_t dstX, int32_t dstY,
                              GrD3DRenderTarget* src, const SkIRect& srcIRect) {
    SkASSERT(dst);
    SkASSERT(src && src->numSamples() > 1 && src->msaaTextureResource());

    D3D12_RECT srcRect = { srcIRect.fLeft, srcIRect.fTop, srcIRect.fRight, srcIRect.fBottom };

    GrD3DTextureResource* dstTextureResource;
    GrRenderTarget* dstRT = dst->asRenderTarget();
    if (dstRT) {
        dstTextureResource = static_cast<GrD3DRenderTarget*>(dstRT);
    } else {
        SkASSERT(dst->asTexture());
        dstTextureResource = static_cast<GrD3DTexture*>(dst->asTexture());
    }

    dstTextureResource->setResourceState(this, D3D12_RESOURCE_STATE_RESOLVE_DEST);
    src->msaaTextureResource()->setResourceState(this, D3D12_RESOURCE_STATE_RESOLVE_SOURCE);

    fCurrentDirectCommandList->resolveSubresourceRegion(dstTextureResource, dstX, dstY,
                                                        src->msaaTextureResource(), &srcRect);
}

void GrD3DGpu::onResolveRenderTarget(GrRenderTarget* target, const SkIRect& resolveRect) {
    SkASSERT(target->numSamples() > 1);
    GrD3DRenderTarget* rt = static_cast<GrD3DRenderTarget*>(target);
    SkASSERT(rt->msaaTextureResource() && rt != rt->msaaTextureResource());

    this->resolveTexture(target, resolveRect.fLeft, resolveRect.fTop, rt, resolveRect);
}

bool GrD3DGpu::onReadPixels(GrSurface* surface,
                            SkIRect rect,
                            GrColorType surfaceColorType,
                            GrColorType dstColorType,
                            void* buffer,
                            size_t rowBytes) {
    SkASSERT(surface);

    if (surfaceColorType != dstColorType) {
        return false;
    }

    GrD3DTextureResource* texResource = nullptr;
    GrD3DRenderTarget* rt = static_cast<GrD3DRenderTarget*>(surface->asRenderTarget());
    if (rt) {
        texResource = rt;
    } else {
        texResource = static_cast<GrD3DTexture*>(surface->asTexture());
    }

    if (!texResource) {
        return false;
    }

    D3D12_RESOURCE_DESC desc = texResource->d3dResource()->GetDesc();
    D3D12_PLACED_SUBRESOURCE_FOOTPRINT placedFootprint;
    UINT64 transferTotalBytes;
    fDevice->GetCopyableFootprints(&desc, 0, 1, 0, &placedFootprint,
                                   nullptr, nullptr, &transferTotalBytes);
    SkASSERT(transferTotalBytes);
    // TODO: implement some way of reusing buffers instead of making a new one every time.
    sk_sp<GrGpuBuffer> transferBuffer = this->createBuffer(transferTotalBytes,
                                                           GrGpuBufferType::kXferGpuToCpu,
                                                           kDynamic_GrAccessPattern);

    this->readOrTransferPixels(texResource, rect, transferBuffer, placedFootprint);
    this->submitDirectCommandList(SyncQueue::kForce);

    // Copy back to CPU buffer
    size_t bpp = GrColorTypeBytesPerPixel(dstColorType);
    if (GrDxgiFormatBytesPerBlock(texResource->dxgiFormat()) != bpp) {
        return false;
    }
    size_t tightRowBytes = bpp * rect.width();

    const void* mappedMemory = transferBuffer->map();

    SkRectMemcpy(buffer,
                 rowBytes,
                 mappedMemory,
                 placedFootprint.Footprint.RowPitch,
                 tightRowBytes,
                 rect.height());

    transferBuffer->unmap();

    return true;
}

void GrD3DGpu::readOrTransferPixels(GrD3DTextureResource* texResource,
                                    SkIRect rect,
                                    sk_sp<GrGpuBuffer> transferBuffer,
                                    const D3D12_PLACED_SUBRESOURCE_FOOTPRINT& placedFootprint) {
    // Set up src location and box
    D3D12_TEXTURE_COPY_LOCATION srcLocation = {};
    srcLocation.pResource = texResource->d3dResource();
    SkASSERT(srcLocation.pResource);
    srcLocation.Type = D3D12_TEXTURE_COPY_TYPE_SUBRESOURCE_INDEX;
    srcLocation.SubresourceIndex = 0;

    D3D12_BOX srcBox = {};
    srcBox.left = rect.left();
    srcBox.top = rect.top();
    srcBox.right = rect.right();
    srcBox.bottom = rect.bottom();
    srcBox.front = 0;
    srcBox.back = 1;

    // Set up dst location
    D3D12_TEXTURE_COPY_LOCATION dstLocation = {};
    dstLocation.Type = D3D12_TEXTURE_COPY_TYPE_PLACED_FOOTPRINT;
    dstLocation.PlacedFootprint = placedFootprint;
    GrD3DBuffer* d3dBuf = static_cast<GrD3DBuffer*>(transferBuffer.get());
    dstLocation.pResource = d3dBuf->d3dResource();

    // Need to change the resource state to COPY_SOURCE in order to download from it
    texResource->setResourceState(this, D3D12_RESOURCE_STATE_COPY_SOURCE);

    fCurrentDirectCommandList->copyTextureRegionToBuffer(transferBuffer, &dstLocation, 0, 0,
                                                         texResource->resource(), &srcLocation,
                                                         &srcBox);
}

bool GrD3DGpu::onWritePixels(GrSurface* surface,
                             SkIRect rect,
                             GrColorType surfaceColorType,
                             GrColorType srcColorType,
                             const GrMipLevel texels[],
                             int mipLevelCount,
                             bool prepForTexSampling) {
    GrD3DTexture* d3dTex = static_cast<GrD3DTexture*>(surface->asTexture());
    if (!d3dTex) {
        return false;
    }

    // Make sure we have at least the base level
    if (!mipLevelCount || !texels[0].fPixels) {
        return false;
    }

    SkASSERT(!GrDxgiFormatIsCompressed(d3dTex->dxgiFormat()));
    bool success = false;

    // Need to change the resource state to COPY_DEST in order to upload to it
    d3dTex->setResourceState(this, D3D12_RESOURCE_STATE_COPY_DEST);

    SkASSERT(mipLevelCount <= d3dTex->maxMipmapLevel() + 1);
    success = this->uploadToTexture(d3dTex, rect, srcColorType, texels, mipLevelCount);

    if (prepForTexSampling) {
        d3dTex->setResourceState(this, D3D12_RESOURCE_STATE_PIXEL_SHADER_RESOURCE);
    }

    return success;
}

bool GrD3DGpu::uploadToTexture(GrD3DTexture* tex,
                               SkIRect rect,
                               GrColorType colorType,
                               const GrMipLevel* texels,
                               int mipLevelCount) {
    SkASSERT(this->d3dCaps().isFormatTexturable(tex->dxgiFormat()));
    // The assumption is either that we have no mipmaps, or that our rect is the entire texture
    SkASSERT(mipLevelCount == 1 || rect == SkIRect::MakeSize(tex->dimensions()));

    // We assume that if the texture has mip levels, we either upload to all the levels or just the
    // first.
    SkASSERT(mipLevelCount == 1 || mipLevelCount == (tex->maxMipmapLevel() + 1));

    if (rect.isEmpty()) {
        return false;
    }

    SkASSERT(this->d3dCaps().surfaceSupportsWritePixels(tex));
    SkASSERT(this->d3dCaps().areColorTypeAndFormatCompatible(colorType, tex->backendFormat()));

    ID3D12Resource* d3dResource = tex->d3dResource();
    SkASSERT(d3dResource);
    D3D12_RESOURCE_DESC desc = d3dResource->GetDesc();
    // Either upload only the first miplevel or all miplevels
    SkASSERT(1 == mipLevelCount || mipLevelCount == (int)desc.MipLevels);

    if (1 == mipLevelCount && !texels[0].fPixels) {
        return true;   // no data to upload
    }

    for (int i = 0; i < mipLevelCount; ++i) {
        // We do not allow any gaps in the mip data
        if (!texels[i].fPixels) {
            return false;
        }
    }

    SkAutoTMalloc<D3D12_PLACED_SUBRESOURCE_FOOTPRINT> placedFootprints(mipLevelCount);
    UINT64 combinedBufferSize;
    // We reset the width and height in the description to match our subrectangle size
    // so we don't end up allocating more space than we need.
    desc.Width = rect.width();
    desc.Height = rect.height();
    fDevice->GetCopyableFootprints(&desc, 0, mipLevelCount, 0, placedFootprints.get(),
                                   nullptr, nullptr, &combinedBufferSize);
    size_t bpp = GrColorTypeBytesPerPixel(colorType);
    SkASSERT(combinedBufferSize);

    GrStagingBufferManager::Slice slice = fStagingBufferManager.allocateStagingBufferSlice(
            combinedBufferSize, D3D12_TEXTURE_DATA_PLACEMENT_ALIGNMENT);
    if (!slice.fBuffer) {
        return false;
    }

    char* bufferData = (char*)slice.fOffsetMapPtr;

    int currentWidth = rect.width();
    int currentHeight = rect.height();
    for (int currentMipLevel = 0; currentMipLevel < mipLevelCount; currentMipLevel++) {
        if (texels[currentMipLevel].fPixels) {

            const size_t trimRowBytes = currentWidth * bpp;
            const size_t srcRowBytes = texels[currentMipLevel].fRowBytes;

            char* dst = bufferData + placedFootprints[currentMipLevel].Offset;

            // copy data into the buffer, skipping any trailing bytes
            const char* src = (const char*)texels[currentMipLevel].fPixels;
            SkRectMemcpy(dst, placedFootprints[currentMipLevel].Footprint.RowPitch,
                         src, srcRowBytes, trimRowBytes, currentHeight);
        }
        currentWidth = std::max(1, currentWidth / 2);
        currentHeight = std::max(1, currentHeight / 2);
    }

    // Update the offsets in the footprints to be relative to the slice's offset
    for (int i = 0; i < mipLevelCount; ++i) {
        placedFootprints[i].Offset += slice.fOffset;
    }

    ID3D12Resource* d3dBuffer = static_cast<GrD3DBuffer*>(slice.fBuffer)->d3dResource();
    fCurrentDirectCommandList->copyBufferToTexture(d3dBuffer,
                                                   tex,
                                                   mipLevelCount,
                                                   placedFootprints.get(),
                                                   rect.left(),
                                                   rect.top());

    if (mipLevelCount < (int)desc.MipLevels) {
        tex->markMipmapsDirty();
    }

    return true;
}

bool GrD3DGpu::onTransferPixelsTo(GrTexture* texture,
                                  SkIRect rect,
                                  GrColorType surfaceColorType,
                                  GrColorType bufferColorType,
                                  sk_sp<GrGpuBuffer> transferBuffer,
                                  size_t bufferOffset,
                                  size_t rowBytes) {
    if (!this->currentCommandList()) {
        return false;
    }

    if (!transferBuffer) {
        return false;
    }

    size_t bpp = GrColorTypeBytesPerPixel(bufferColorType);
    if (GrBackendFormatBytesPerPixel(texture->backendFormat()) != bpp) {
        return false;
    }

    // D3D requires offsets for texture transfers to be aligned to this value
    if (SkToBool(bufferOffset & (D3D12_TEXTURE_DATA_PLACEMENT_ALIGNMENT-1))) {
        return false;
    }

    GrD3DTexture* d3dTex = static_cast<GrD3DTexture*>(texture);
    if (!d3dTex) {
        return false;
    }

    SkDEBUGCODE(DXGI_FORMAT format = d3dTex->dxgiFormat());

    // Can't transfer compressed data
    SkASSERT(!GrDxgiFormatIsCompressed(format));

    SkASSERT(GrDxgiFormatBytesPerBlock(format) == GrColorTypeBytesPerPixel(bufferColorType));

    SkASSERT(SkIRect::MakeSize(texture->dimensions()).contains(rect));

    // Set up copy region
    D3D12_PLACED_SUBRESOURCE_FOOTPRINT placedFootprint = {};
    ID3D12Resource* d3dResource = d3dTex->d3dResource();
    SkASSERT(d3dResource);
    D3D12_RESOURCE_DESC desc = d3dResource->GetDesc();
    desc.Width = rect.width();
    desc.Height = rect.height();
    UINT64 totalBytes;
    fDevice->GetCopyableFootprints(&desc, 0, 1, 0, &placedFootprint,
                                   nullptr, nullptr, &totalBytes);
    placedFootprint.Offset = bufferOffset;

    // Change state of our target so it can be copied to
    d3dTex->setResourceState(this, D3D12_RESOURCE_STATE_COPY_DEST);

    // Copy the buffer to the image.
    ID3D12Resource* d3dBuffer = static_cast<GrD3DBuffer*>(transferBuffer.get())->d3dResource();
    fCurrentDirectCommandList->copyBufferToTexture(d3dBuffer,
                                                   d3dTex,
                                                   1,
                                                   &placedFootprint,
                                                   rect.left(),
                                                   rect.top());
    this->currentCommandList()->addGrBuffer(std::move(transferBuffer));

    d3dTex->markMipmapsDirty();
    return true;
}

bool GrD3DGpu::onTransferPixelsFrom(GrSurface* surface,
                                    SkIRect rect,
                                    GrColorType surfaceColorType,
                                    GrColorType bufferColorType,
                                    sk_sp<GrGpuBuffer> transferBuffer,
                                    size_t offset) {
    if (!this->currentCommandList()) {
        return false;
    }
    SkASSERT(surface);
    SkASSERT(transferBuffer);
    // TODO
    //if (fProtectedContext == GrProtected::kYes) {
    //    return false;
    //}

    // D3D requires offsets for texture transfers to be aligned to this value
    if (SkToBool(offset & (D3D12_TEXTURE_DATA_PLACEMENT_ALIGNMENT-1))) {
        return false;
    }

    GrD3DTextureResource* texResource = nullptr;
    GrD3DRenderTarget* rt = static_cast<GrD3DRenderTarget*>(surface->asRenderTarget());
    if (rt) {
        texResource = rt;
    } else {
        texResource = static_cast<GrD3DTexture*>(surface->asTexture());
    }

    if (!texResource) {
        return false;
    }

    SkDEBUGCODE(DXGI_FORMAT format = texResource->dxgiFormat());
    SkASSERT(GrDxgiFormatBytesPerBlock(format) == GrColorTypeBytesPerPixel(bufferColorType));

    D3D12_RESOURCE_DESC desc = texResource->d3dResource()->GetDesc();
    desc.Width = rect.width();
    desc.Height = rect.height();
    D3D12_PLACED_SUBRESOURCE_FOOTPRINT placedFootprint;
    UINT64 transferTotalBytes;
    fDevice->GetCopyableFootprints(&desc, 0, 1, offset, &placedFootprint,
                                   nullptr, nullptr, &transferTotalBytes);
    SkASSERT(transferTotalBytes);

    this->readOrTransferPixels(texResource, rect, transferBuffer, placedFootprint);

    // TODO: It's not clear how to ensure the transfer is done before we read from the buffer,
    // other than maybe doing a resource state transition.

    return true;
}

static bool check_resource_info(const GrD3DTextureResourceInfo& info) {
    if (!info.fResource.get()) {
        return false;
    }
    return true;
}

static bool check_tex_resource_info(const GrD3DCaps& caps, const GrD3DTextureResourceInfo& info) {
    if (!caps.isFormatTexturable(info.fFormat)) {
        return false;
    }
    // We don't support sampling from multisampled textures.
    if (info.fSampleCount != 1) {
        return false;
    }
    return true;
}

static bool check_rt_resource_info(const GrD3DCaps& caps, const GrD3DTextureResourceInfo& info,
                                int sampleCnt) {
    if (!caps.isFormatRenderable(info.fFormat, sampleCnt)) {
        return false;
    }
    return true;
}

sk_sp<GrTexture> GrD3DGpu::onWrapBackendTexture(const GrBackendTexture& tex,
                                                GrWrapOwnership,
                                                GrWrapCacheable wrapType,
                                                GrIOType ioType) {
    GrD3DTextureResourceInfo textureInfo;
    if (!tex.getD3DTextureResourceInfo(&textureInfo)) {
        return nullptr;
    }

    if (!check_resource_info(textureInfo)) {
        return nullptr;
    }

    if (!check_tex_resource_info(this->d3dCaps(), textureInfo)) {
        return nullptr;
    }

    // TODO: support protected context
    if (tex.isProtected()) {
        return nullptr;
    }

    sk_sp<GrD3DResourceState> state = tex.getGrD3DResourceState();
    SkASSERT(state);
    return GrD3DTexture::MakeWrappedTexture(this, tex.dimensions(), wrapType, ioType, textureInfo,
                                            std::move(state));
}

sk_sp<GrTexture> GrD3DGpu::onWrapCompressedBackendTexture(const GrBackendTexture& tex,
                                                          GrWrapOwnership ownership,
                                                          GrWrapCacheable wrapType) {
    return this->onWrapBackendTexture(tex, ownership, wrapType, kRead_GrIOType);
}

sk_sp<GrTexture> GrD3DGpu::onWrapRenderableBackendTexture(const GrBackendTexture& tex,
                                                          int sampleCnt,
                                                          GrWrapOwnership ownership,
                                                          GrWrapCacheable cacheable) {
    GrD3DTextureResourceInfo textureInfo;
    if (!tex.getD3DTextureResourceInfo(&textureInfo)) {
        return nullptr;
    }

    if (!check_resource_info(textureInfo)) {
        return nullptr;
    }

    if (!check_tex_resource_info(this->d3dCaps(), textureInfo)) {
        return nullptr;
    }
    if (!check_rt_resource_info(this->d3dCaps(), textureInfo, sampleCnt)) {
        return nullptr;
    }

    // TODO: support protected context
    if (tex.isProtected()) {
        return nullptr;
    }

    sampleCnt = this->d3dCaps().getRenderTargetSampleCount(sampleCnt, textureInfo.fFormat);

    sk_sp<GrD3DResourceState> state = tex.getGrD3DResourceState();
    SkASSERT(state);

    return GrD3DTextureRenderTarget::MakeWrappedTextureRenderTarget(this, tex.dimensions(),
                                                                    sampleCnt, cacheable,
                                                                    textureInfo, std::move(state));
}

sk_sp<GrRenderTarget> GrD3DGpu::onWrapBackendRenderTarget(const GrBackendRenderTarget& rt) {
    GrD3DTextureResourceInfo info;
    if (!rt.getD3DTextureResourceInfo(&info)) {
        return nullptr;
    }

    if (!check_resource_info(info)) {
        return nullptr;
    }

    if (!check_rt_resource_info(this->d3dCaps(), info, rt.sampleCnt())) {
        return nullptr;
    }

    // TODO: support protected context
    if (rt.isProtected()) {
        return nullptr;
    }

    sk_sp<GrD3DResourceState> state = rt.getGrD3DResourceState();

    sk_sp<GrD3DRenderTarget> tgt = GrD3DRenderTarget::MakeWrappedRenderTarget(
            this, rt.dimensions(), rt.sampleCnt(), info, std::move(state));

    // We don't allow the client to supply a premade stencil buffer. We always create one if needed.
    SkASSERT(!rt.stencilBits());
    if (tgt) {
        SkASSERT(tgt->canAttemptStencilAttachment(tgt->numSamples() > 1));
    }

    return std::move(tgt);
}

static bool is_odd(int x) {
    return x > 1 && SkToBool(x & 0x1);
}

bool GrD3DGpu::onRegenerateMipMapLevels(GrTexture * tex) {
    auto * d3dTex = static_cast<GrD3DTexture*>(tex);
    SkASSERT(tex->textureType() == GrTextureType::k2D);
    int width = tex->width();
    int height = tex->height();

    // determine if we can read from and mipmap this format
    const GrD3DCaps & caps = this->d3dCaps();
    if (!caps.isFormatTexturable(d3dTex->dxgiFormat()) ||
        !caps.mipmapSupport()) {
        return false;
    }

    sk_sp<GrD3DTexture> uavTexture;
    // if the format is unordered accessible and resource flag is set, use resource for uav
    if (caps.isFormatUnorderedAccessible(d3dTex->dxgiFormat()) &&
        (d3dTex->d3dResource()->GetDesc().Flags & D3D12_RESOURCE_FLAG_ALLOW_UNORDERED_ACCESS)) {
        uavTexture = sk_ref_sp(d3dTex);
    } else {
        // need to make a copy and use that for our uav
        D3D12_RESOURCE_DESC uavDesc = d3dTex->d3dResource()->GetDesc();
        uavDesc.Flags |= D3D12_RESOURCE_FLAG_ALLOW_UNORDERED_ACCESS;
        // if the format is unordered accessible, copy to resource with same format and flag set
        if (!caps.isFormatUnorderedAccessible(d3dTex->dxgiFormat())) {
            // TODO: support BGR and sRGB
            return false;
        }
        // TODO: make this a scratch texture
        GrProtected grProtected = tex->isProtected() ? GrProtected::kYes : GrProtected::kNo;
        uavTexture = GrD3DTexture::MakeNewTexture(this, SkBudgeted::kNo, tex->dimensions(),
                                                  uavDesc, grProtected, GrMipmapStatus::kDirty);
        if (!uavTexture) {
            return false;
        }

        d3dTex->setResourceState(this, D3D12_RESOURCE_STATE_COPY_SOURCE);
        // copy top miplevel to uavTexture
        uavTexture->setResourceState(this, D3D12_RESOURCE_STATE_COPY_DEST);
        this->currentCommandList()->copyTextureToTexture(uavTexture.get(), d3dTex, 0);
    }

    uint32_t levelCount = d3dTex->mipLevels();
    // SkMipmap doesn't include the base level in the level count so we have to add 1
    SkASSERT((int)levelCount == SkMipmap::ComputeLevelCount(tex->width(), tex->height()) + 1);

    sk_sp<GrD3DRootSignature> rootSig = fResourceProvider.findOrCreateRootSignature(1, 1);
    this->currentCommandList()->setComputeRootSignature(rootSig);

    // TODO: use linear vs. srgb shader based on texture format
    sk_sp<GrD3DPipeline> pipeline = this->resourceProvider().findOrCreateMipmapPipeline();
    SkASSERT(pipeline);
    this->currentCommandList()->setPipelineState(std::move(pipeline));

    // set sampler
    GrSamplerState samplerState(SkFilterMode::kLinear, SkMipmapMode::kNearest);
    std::vector<D3D12_CPU_DESCRIPTOR_HANDLE> samplers(1);
    samplers[0] = fResourceProvider.findOrCreateCompatibleSampler(samplerState);
    this->currentCommandList()->addSampledTextureRef(uavTexture.get());
    sk_sp<GrD3DDescriptorTable> samplerTable = fResourceProvider.findOrCreateSamplerTable(samplers);
    this->currentCommandList()->setComputeRootDescriptorTable(
            static_cast<unsigned int>(GrD3DRootSignature::ParamIndex::kSamplerDescriptorTable),
            samplerTable->baseGpuDescriptor());

    // Transition the top subresource to be readable in the compute shader
    D3D12_RESOURCE_STATES currentResourceState = uavTexture->currentState();
    D3D12_RESOURCE_TRANSITION_BARRIER barrier;
    barrier.pResource = uavTexture->d3dResource();
    barrier.Subresource = 0;
    barrier.StateBefore = currentResourceState;
    barrier.StateAfter = D3D12_RESOURCE_STATE_NON_PIXEL_SHADER_RESOURCE;
    this->addResourceBarriers(uavTexture->resource(), 1, &barrier);

    // Generate the miplevels
    for (unsigned int dstMip = 1; dstMip < levelCount; ++dstMip) {
        unsigned int srcMip = dstMip - 1;
        width = std::max(1, width / 2);
        height = std::max(1, height / 2);

        unsigned int sampleMode = 0;
        if (is_odd(width) && is_odd(height)) {
            sampleMode = 1;
        } else if (is_odd(width)) {
            sampleMode = 2;
        } else if (is_odd(height)) {
            sampleMode = 3;
        }

        // set constants
        struct {
            SkSize inverseSize;
            uint32_t mipLevel;
            uint32_t sampleMode;
        } constantData = { {1.f / width, 1.f / height}, srcMip, sampleMode };

        D3D12_GPU_VIRTUAL_ADDRESS constantsAddress =
            fResourceProvider.uploadConstantData(&constantData, sizeof(constantData));
        this->currentCommandList()->setComputeRootConstantBufferView(
                (unsigned int)GrD3DRootSignature::ParamIndex::kConstantBufferView,
                constantsAddress);

        std::vector<D3D12_CPU_DESCRIPTOR_HANDLE> shaderViews;
        // create SRV
        GrD3DDescriptorHeap::CPUHandle srvHandle =
                fResourceProvider.createShaderResourceView(uavTexture->d3dResource(), srcMip, 1);
        shaderViews.push_back(srvHandle.fHandle);
        fMipmapCPUDescriptors.push_back(srvHandle);
        // create UAV
        GrD3DDescriptorHeap::CPUHandle uavHandle =
                fResourceProvider.createUnorderedAccessView(uavTexture->d3dResource(), dstMip);
        shaderViews.push_back(uavHandle.fHandle);
        fMipmapCPUDescriptors.push_back(uavHandle);

        // set up and bind shaderView descriptor table
        sk_sp<GrD3DDescriptorTable> srvTable =
                fResourceProvider.findOrCreateShaderViewTable(shaderViews);
        this->currentCommandList()->setComputeRootDescriptorTable(
                (unsigned int)GrD3DRootSignature::ParamIndex::kShaderViewDescriptorTable,
                srvTable->baseGpuDescriptor());

        // Transition resource state of dstMip subresource so we can write to it
        barrier.Subresource = dstMip;
        barrier.StateBefore = currentResourceState;
        barrier.StateAfter = D3D12_RESOURCE_STATE_UNORDERED_ACCESS;
        this->addResourceBarriers(uavTexture->resource(), 1, &barrier);

        // Using the form (x+7)/8 ensures that the remainder is covered as well
        this->currentCommandList()->dispatch((width+7)/8, (height+7)/8);

        // guarantee UAV writes have completed
        this->currentCommandList()->uavBarrier(uavTexture->resource(), uavTexture->d3dResource());

        // Transition resource state of dstMip subresource so we can read it in the next stage
        barrier.StateBefore = D3D12_RESOURCE_STATE_UNORDERED_ACCESS;
        barrier.StateAfter = D3D12_RESOURCE_STATE_NON_PIXEL_SHADER_RESOURCE;
        this->addResourceBarriers(uavTexture->resource(), 1, &barrier);
    }

    // copy back if necessary
    if (uavTexture.get() != d3dTex) {
        d3dTex->setResourceState(this, D3D12_RESOURCE_STATE_COPY_DEST);
        barrier.Subresource = D3D12_RESOURCE_BARRIER_ALL_SUBRESOURCES;
        barrier.StateBefore = D3D12_RESOURCE_STATE_NON_PIXEL_SHADER_RESOURCE;
        barrier.StateAfter = D3D12_RESOURCE_STATE_COPY_SOURCE;
        // TODO: support BGR and sRGB
        this->addResourceBarriers(uavTexture->resource(), 1, &barrier);
        this->currentCommandList()->copyTextureToTexture(d3dTex, uavTexture.get());
    } else {
        // For simplicity our resource state tracking considers all subresources to have the same
        // state. However, we've changed that state one subresource at a time without going through
        // the tracking system, so we need to patch up the resource states back to the original.
        barrier.Subresource = D3D12_RESOURCE_BARRIER_ALL_SUBRESOURCES;
        barrier.StateBefore = D3D12_RESOURCE_STATE_NON_PIXEL_SHADER_RESOURCE;
        barrier.StateAfter = currentResourceState;
        this->addResourceBarriers(d3dTex->resource(), 1, &barrier);
    }

    return true;
}

sk_sp<GrGpuBuffer> GrD3DGpu::onCreateBuffer(size_t sizeInBytes, GrGpuBufferType type,
                                             GrAccessPattern accessPattern, const void* data) {
    sk_sp<GrD3DBuffer> buffer = GrD3DBuffer::Make(this, sizeInBytes, type, accessPattern);
    if (data && buffer) {
        buffer->updateData(data, sizeInBytes);
    }

    return std::move(buffer);
}

sk_sp<GrAttachment> GrD3DGpu::makeStencilAttachment(const GrBackendFormat& /*colorFormat*/,
                                                    SkISize dimensions, int numStencilSamples) {
    DXGI_FORMAT sFmt = this->d3dCaps().preferredStencilFormat();

    fStats.incStencilAttachmentCreates();
    return GrD3DAttachment::MakeStencil(this, dimensions, numStencilSamples, sFmt);
}

bool GrD3DGpu::createTextureResourceForBackendSurface(DXGI_FORMAT dxgiFormat,
                                                      SkISize dimensions,
                                                      GrTexturable texturable,
                                                      GrRenderable renderable,
                                                      GrMipmapped mipMapped,
                                                      int sampleCnt,
                                                      GrD3DTextureResourceInfo* info,
                                                      GrProtected isProtected) {
    SkASSERT(texturable == GrTexturable::kYes || renderable == GrRenderable::kYes);

    if (this->protectedContext() != (isProtected == GrProtected::kYes)) {
        return false;
    }

    if (texturable == GrTexturable::kYes && !this->d3dCaps().isFormatTexturable(dxgiFormat)) {
        return false;
    }

    if (renderable == GrRenderable::kYes && !this->d3dCaps().isFormatRenderable(dxgiFormat, 1)) {
        return false;
    }

    int numMipLevels = 1;
    if (mipMapped == GrMipmapped::kYes) {
        numMipLevels = SkMipmap::ComputeLevelCount(dimensions.width(), dimensions.height()) + 1;
    }

    // create the texture
    D3D12_RESOURCE_FLAGS usageFlags = D3D12_RESOURCE_FLAG_NONE;
    if (renderable == GrRenderable::kYes) {
        usageFlags |= D3D12_RESOURCE_FLAG_ALLOW_RENDER_TARGET;
    }

    D3D12_RESOURCE_DESC resourceDesc = {};
    resourceDesc.Dimension = D3D12_RESOURCE_DIMENSION_TEXTURE2D;
    resourceDesc.Alignment = 0;  // use default alignment
    resourceDesc.Width = dimensions.fWidth;
    resourceDesc.Height = dimensions.fHeight;
    resourceDesc.DepthOrArraySize = 1;
    resourceDesc.MipLevels = numMipLevels;
    resourceDesc.Format = dxgiFormat;
    resourceDesc.SampleDesc.Count = sampleCnt;
    resourceDesc.SampleDesc.Quality = DXGI_STANDARD_MULTISAMPLE_QUALITY_PATTERN;
    resourceDesc.Layout = D3D12_TEXTURE_LAYOUT_UNKNOWN;  // use driver-selected swizzle
    resourceDesc.Flags = usageFlags;

    D3D12_CLEAR_VALUE* clearValuePtr = nullptr;
    D3D12_CLEAR_VALUE clearValue = {};
    if (renderable == GrRenderable::kYes) {
        clearValue.Format = dxgiFormat;
        // Assume transparent black
        clearValue.Color[0] = 0;
        clearValue.Color[1] = 0;
        clearValue.Color[2] = 0;
        clearValue.Color[3] = 0;
        clearValuePtr = &clearValue;
    }

    D3D12_RESOURCE_STATES initialState = (renderable == GrRenderable::kYes)
                                                 ? D3D12_RESOURCE_STATE_RENDER_TARGET
                                                 : D3D12_RESOURCE_STATE_COPY_DEST;
    if (!GrD3DTextureResource::InitTextureResourceInfo(this, resourceDesc, initialState,
                                                       isProtected, clearValuePtr, info)) {
        SkDebugf("Failed to init texture resource info\n");
        return false;
    }

    return true;
}

GrBackendTexture GrD3DGpu::onCreateBackendTexture(SkISize dimensions,
                                                  const GrBackendFormat& format,
                                                  GrRenderable renderable,
                                                  GrMipmapped mipMapped,
                                                  GrProtected isProtected) {
    const GrD3DCaps& caps = this->d3dCaps();

    if (this->protectedContext() != (isProtected == GrProtected::kYes)) {
        return {};
    }

    DXGI_FORMAT dxgiFormat;
    if (!format.asDxgiFormat(&dxgiFormat)) {
        return {};
    }

    // TODO: move the texturability check up to GrGpu::createBackendTexture and just assert here
    if (!caps.isFormatTexturable(dxgiFormat)) {
        return {};
    }

    GrD3DTextureResourceInfo info;
    if (!this->createTextureResourceForBackendSurface(dxgiFormat, dimensions, GrTexturable::kYes,
                                                      renderable, mipMapped, 1, &info,
                                                      isProtected)) {
        return {};
    }

    return GrBackendTexture(dimensions.width(), dimensions.height(), info);
}

static bool copy_color_data(const GrD3DCaps& caps,
                            char* mapPtr,
                            DXGI_FORMAT dxgiFormat,
                            SkISize dimensions,
                            D3D12_PLACED_SUBRESOURCE_FOOTPRINT* placedFootprints,
                            std::array<float, 4> color) {
    auto colorType = caps.getFormatColorType(dxgiFormat);
    if (colorType == GrColorType::kUnknown) {
        return false;
    }
    GrImageInfo ii(colorType, kUnpremul_SkAlphaType, nullptr, dimensions);
    if (!GrClearImage(ii, mapPtr, placedFootprints[0].Footprint.RowPitch, color)) {
        return false;
    }

    return true;
}

bool GrD3DGpu::onClearBackendTexture(const GrBackendTexture& backendTexture,
                                     sk_sp<GrRefCntedCallback> finishedCallback,
                                     std::array<float, 4> color) {
    GrD3DTextureResourceInfo info;
    SkAssertResult(backendTexture.getD3DTextureResourceInfo(&info));
    SkASSERT(!GrDxgiFormatIsCompressed(info.fFormat));

    sk_sp<GrD3DResourceState> state = backendTexture.getGrD3DResourceState();
    SkASSERT(state);
    sk_sp<GrD3DTexture> texture =
            GrD3DTexture::MakeWrappedTexture(this, backendTexture.dimensions(),
                                             GrWrapCacheable::kNo,
                                             kRW_GrIOType, info, std::move(state));
    if (!texture) {
        return false;
    }

    GrD3DDirectCommandList* cmdList = this->currentCommandList();
    if (!cmdList) {
        return false;
    }

    texture->setResourceState(this, D3D12_RESOURCE_STATE_COPY_DEST);

    ID3D12Resource* d3dResource = texture->d3dResource();
    SkASSERT(d3dResource);
    D3D12_RESOURCE_DESC desc = d3dResource->GetDesc();
    unsigned int mipLevelCount = 1;
    if (backendTexture.fMipmapped == GrMipmapped::kYes) {
        mipLevelCount = SkMipmap::ComputeLevelCount(backendTexture.dimensions()) + 1;
    }
    SkASSERT(mipLevelCount == info.fLevelCount);
    SkAutoSTMalloc<15, D3D12_PLACED_SUBRESOURCE_FOOTPRINT> placedFootprints(mipLevelCount);
    UINT numRows;
    UINT64 rowSizeInBytes;
    UINT64 combinedBufferSize;
    // We reuse the same top-level buffer area for all levels, hence passing 1 for level count.
    fDevice->GetCopyableFootprints(&desc,
                                   /* first resource  */ 0,
                                   /* mip level count */ 1,
                                   /* base offset     */ 0,
                                   placedFootprints.get(),
                                   &numRows,
                                   &rowSizeInBytes,
                                   &combinedBufferSize);
    SkASSERT(combinedBufferSize);

    GrStagingBufferManager::Slice slice = fStagingBufferManager.allocateStagingBufferSlice(
            combinedBufferSize, D3D12_TEXTURE_DATA_PLACEMENT_ALIGNMENT);
    if (!slice.fBuffer) {
        return false;
    }

    char* bufferData = (char*)slice.fOffsetMapPtr;
    SkASSERT(bufferData);
    if (!copy_color_data(this->d3dCaps(),
                         bufferData,
                         info.fFormat,
                         backendTexture.dimensions(),
                         placedFootprints,
                         color)) {
        return false;
    }
    // Update the offsets in the footprint to be relative to the slice's offset
    placedFootprints[0].Offset += slice.fOffset;
    // Since we're sharing data for all the levels, set all the upper level footprints to the base.
    UINT w = placedFootprints[0].Footprint.Width;
    UINT h = placedFootprints[0].Footprint.Height;
    for (unsigned int i = 1; i < mipLevelCount; ++i) {
        w = std::max(1U, w/2);
        h = std::max(1U, h/2);
        placedFootprints[i].Offset = placedFootprints[0].Offset;
        placedFootprints[i].Footprint.Format   = placedFootprints[0].Footprint.Format;
        placedFootprints[i].Footprint.Width    = w;
        placedFootprints[i].Footprint.Height   = h;
        placedFootprints[i].Footprint.Depth    = 1;
        placedFootprints[i].Footprint.RowPitch = placedFootprints[0].Footprint.RowPitch;
    }

    ID3D12Resource* d3dBuffer = static_cast<GrD3DBuffer*>(slice.fBuffer)->d3dResource();
    cmdList->copyBufferToTexture(d3dBuffer,
                                 texture.get(),
                                 mipLevelCount,
                                 placedFootprints.get(),
                                 /*left*/ 0,
                                 /*top */ 0);

    if (finishedCallback) {
        this->addFinishedCallback(std::move(finishedCallback));
    }

    return true;
}

GrBackendTexture GrD3DGpu::onCreateCompressedBackendTexture(
    SkISize dimensions, const GrBackendFormat& format, GrMipmapped mipMapped,
    GrProtected isProtected) {
    return this->onCreateBackendTexture(dimensions, format, GrRenderable::kNo, mipMapped,
                                        isProtected);
}

bool GrD3DGpu::onUpdateCompressedBackendTexture(const GrBackendTexture& backendTexture,
                                                sk_sp<GrRefCntedCallback> finishedCallback,
                                                const void* data,
                                                size_t size) {
    GrD3DTextureResourceInfo info;
    SkAssertResult(backendTexture.getD3DTextureResourceInfo(&info));

    sk_sp<GrD3DResourceState> state = backendTexture.getGrD3DResourceState();
    SkASSERT(state);
    sk_sp<GrD3DTexture> texture = GrD3DTexture::MakeWrappedTexture(this,
                                                                   backendTexture.dimensions(),
                                                                   GrWrapCacheable::kNo,
                                                                   kRW_GrIOType,
                                                                   info,
                                                                   std::move(state));
    if (!texture) {
        return false;
    }

    GrD3DDirectCommandList* cmdList = this->currentCommandList();
    if (!cmdList) {
        return false;
    }

    texture->setResourceState(this, D3D12_RESOURCE_STATE_COPY_DEST);

    ID3D12Resource* d3dResource = texture->d3dResource();
    SkASSERT(d3dResource);
    D3D12_RESOURCE_DESC desc = d3dResource->GetDesc();
    unsigned int mipLevelCount = 1;
    if (backendTexture.hasMipmaps()) {
        mipLevelCount = SkMipmap::ComputeLevelCount(backendTexture.dimensions().width(),
                                                    backendTexture.dimensions().height()) + 1;
    }
    SkASSERT(mipLevelCount == info.fLevelCount);
    SkAutoTMalloc<D3D12_PLACED_SUBRESOURCE_FOOTPRINT> placedFootprints(mipLevelCount);
    UINT64 combinedBufferSize;
    SkAutoTMalloc<UINT> numRows(mipLevelCount);
    SkAutoTMalloc<UINT64> rowSizeInBytes(mipLevelCount);
    fDevice->GetCopyableFootprints(&desc,
                                   0,
                                   mipLevelCount,
                                   0,
                                   placedFootprints.get(),
                                   numRows.get(),
                                   rowSizeInBytes.get(),
                                   &combinedBufferSize);
    SkASSERT(combinedBufferSize);
    SkASSERT(GrDxgiFormatIsCompressed(info.fFormat));

    GrStagingBufferManager::Slice slice = fStagingBufferManager.allocateStagingBufferSlice(
            combinedBufferSize, D3D12_TEXTURE_DATA_PLACEMENT_ALIGNMENT);
    if (!slice.fBuffer) {
        return false;
    }

    char* bufferData = (char*)slice.fOffsetMapPtr;
    SkASSERT(bufferData);
    copy_compressed_data(bufferData,
                         info.fFormat,
                         placedFootprints.get(),
                         numRows.get(),
                         rowSizeInBytes.get(),
                         data,
                         info.fLevelCount);

    // Update the offsets in the footprints to be relative to the slice's offset
    for (unsigned int i = 0; i < mipLevelCount; ++i) {
        placedFootprints[i].Offset += slice.fOffset;
    }

    ID3D12Resource* d3dBuffer = static_cast<GrD3DBuffer*>(slice.fBuffer)->d3dResource();
    cmdList->copyBufferToTexture(d3dBuffer,
                                 texture.get(),
                                 mipLevelCount,
                                 placedFootprints.get(),
                                 0,
                                 0);

    if (finishedCallback) {
        this->addFinishedCallback(std::move(finishedCallback));
    }

    return true;
}

void GrD3DGpu::deleteBackendTexture(const GrBackendTexture& tex) {
    SkASSERT(GrBackendApi::kDirect3D == tex.fBackend);
    // Nothing to do here, will get cleaned up when the GrBackendTexture object goes away
}

bool GrD3DGpu::compile(const GrProgramDesc&, const GrProgramInfo&) {
    return false;
}

#if GR_TEST_UTILS
bool GrD3DGpu::isTestingOnlyBackendTexture(const GrBackendTexture& tex) const {
    SkASSERT(GrBackendApi::kDirect3D == tex.backend());

    GrD3DTextureResourceInfo info;
    if (!tex.getD3DTextureResourceInfo(&info)) {
        return false;
    }
    ID3D12Resource* textureResource = info.fResource.get();
    if (!textureResource) {
        return false;
    }
    return !(textureResource->GetDesc().Flags & D3D12_RESOURCE_FLAG_DENY_SHADER_RESOURCE);
}

GrBackendRenderTarget GrD3DGpu::createTestingOnlyBackendRenderTarget(SkISize dimensions,
                                                                     GrColorType colorType,
                                                                     int sampleCnt,
                                                                     GrProtected isProtected) {
    if (dimensions.width()  > this->caps()->maxRenderTargetSize() ||
        dimensions.height() > this->caps()->maxRenderTargetSize()) {
        return {};
    }

    DXGI_FORMAT dxgiFormat = this->d3dCaps().getFormatFromColorType(colorType);

    GrD3DTextureResourceInfo info;
    if (!this->createTextureResourceForBackendSurface(dxgiFormat, dimensions, GrTexturable::kNo,
                                                      GrRenderable::kYes, GrMipmapped::kNo,
                                                      sampleCnt, &info, isProtected)) {
        return {};
    }

    return GrBackendRenderTarget(dimensions.width(), dimensions.height(), info);
}

void GrD3DGpu::deleteTestingOnlyBackendRenderTarget(const GrBackendRenderTarget& rt) {
    SkASSERT(GrBackendApi::kDirect3D == rt.backend());

    GrD3DTextureResourceInfo info;
    if (rt.getD3DTextureResourceInfo(&info)) {
        this->submitToGpu(true);
        // Nothing else to do here, will get cleaned up when the GrBackendRenderTarget
        // is deleted.
    }
}

void GrD3DGpu::testingOnly_startCapture() {
    if (fGraphicsAnalysis) {
        fGraphicsAnalysis->BeginCapture();
    }
}

void GrD3DGpu::testingOnly_endCapture() {
    if (fGraphicsAnalysis) {
        fGraphicsAnalysis->EndCapture();
    }
}
#endif

///////////////////////////////////////////////////////////////////////////////

void GrD3DGpu::addResourceBarriers(sk_sp<GrManagedResource> resource,
                                   int numBarriers,
                                   D3D12_RESOURCE_TRANSITION_BARRIER* barriers) const {
    SkASSERT(fCurrentDirectCommandList);
    SkASSERT(resource);

    fCurrentDirectCommandList->resourceBarrier(std::move(resource), numBarriers, barriers);
}

void GrD3DGpu::addBufferResourceBarriers(GrD3DBuffer* buffer,
                                         int numBarriers,
                                         D3D12_RESOURCE_TRANSITION_BARRIER* barriers) const {
    SkASSERT(fCurrentDirectCommandList);
    SkASSERT(buffer);

    fCurrentDirectCommandList->resourceBarrier(nullptr, numBarriers, barriers);
    fCurrentDirectCommandList->addGrBuffer(sk_ref_sp<const GrBuffer>(buffer));
}


void GrD3DGpu::prepareSurfacesForBackendAccessAndStateUpdates(
        SkSpan<GrSurfaceProxy*> proxies,
        SkSurface::BackendSurfaceAccess access,
        const GrBackendSurfaceMutableState* newState) {
    // prepare proxies by transitioning to PRESENT renderState
    if (!proxies.empty() && access == SkSurface::BackendSurfaceAccess::kPresent) {
        GrD3DTextureResource* resource;
        for (GrSurfaceProxy* proxy : proxies) {
            SkASSERT(proxy->isInstantiated());
            if (GrTexture* tex = proxy->peekTexture()) {
                resource = static_cast<GrD3DTexture*>(tex);
            } else {
                GrRenderTarget* rt = proxy->peekRenderTarget();
                SkASSERT(rt);
                resource = static_cast<GrD3DRenderTarget*>(rt);
            }
            resource->prepareForPresent(this);
        }
    }
}

void GrD3DGpu::takeOwnershipOfBuffer(sk_sp<GrGpuBuffer> buffer) {
    fCurrentDirectCommandList->addGrBuffer(std::move(buffer));
}

bool GrD3DGpu::onSubmitToGpu(bool syncCpu) {
    if (syncCpu) {
        return this->submitDirectCommandList(SyncQueue::kForce);
    } else {
        return this->submitDirectCommandList(SyncQueue::kSkip);
    }
}

std::unique_ptr<GrSemaphore> SK_WARN_UNUSED_RESULT GrD3DGpu::makeSemaphore(bool) {
    return GrD3DSemaphore::Make(this);
}
std::unique_ptr<GrSemaphore> GrD3DGpu::wrapBackendSemaphore(const GrBackendSemaphore& semaphore,
                                                            GrSemaphoreWrapType /* wrapType */,
                                                            GrWrapOwnership /* ownership */) {
    SkASSERT(this->caps()->semaphoreSupport());
    GrD3DFenceInfo fenceInfo;
    if (!semaphore.getD3DFenceInfo(&fenceInfo)) {
        return nullptr;
    }
    return GrD3DSemaphore::MakeWrapped(fenceInfo);
}

void GrD3DGpu::insertSemaphore(GrSemaphore* semaphore) {
    SkASSERT(semaphore);
    GrD3DSemaphore* d3dSem = static_cast<GrD3DSemaphore*>(semaphore);
    // TODO: Do we need to track the lifetime of this? How do we know it's done?
    fQueue->Signal(d3dSem->fence(), d3dSem->value());
}

void GrD3DGpu::waitSemaphore(GrSemaphore* semaphore) {
    SkASSERT(semaphore);
    GrD3DSemaphore* d3dSem = static_cast<GrD3DSemaphore*>(semaphore);
    // TODO: Do we need to track the lifetime of this?
    fQueue->Wait(d3dSem->fence(), d3dSem->value());
}

GrFence SK_WARN_UNUSED_RESULT GrD3DGpu::insertFence() {
    GR_D3D_CALL_ERRCHECK(fQueue->Signal(fFence.get(), ++fCurrentFenceValue));
    return fCurrentFenceValue;
}

bool GrD3DGpu::waitFence(GrFence fence) {
    return (fFence->GetCompletedValue() >= fence);
}

void GrD3DGpu::finishOutstandingGpuWork() {
    this->waitForQueueCompletion();
}