kmx git

//
// Copyright 2018 The ANGLE Project Authors. All rights reserved.
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.
//
// vk_cache_utils.h:
//    Contains the classes for the Pipeline State Object cache as well as the RenderPass cache.
//    Also contains the structures for the packed descriptions for the RenderPass and Pipeline.
//

#ifndef LIBANGLE_RENDERER_VULKAN_VK_CACHE_UTILS_H_
#define LIBANGLE_RENDERER_VULKAN_VK_CACHE_UTILS_H_

#include "common/Color.h"
#include "libANGLE/renderer/vulkan/vk_utils.h"

namespace rx
{

namespace vk
{
// Packed Vk resource descriptions.
// Most Vk types use many more bits than required to represent the underlying data.
// Since ANGLE wants cache things like RenderPasses and Pipeline State Objects using
// hashing (and also needs to check equality) we can optimize these operations by
// using fewer bits. Hence the packed types.
//
// One implementation note: these types could potentially be improved by using even
// fewer bits. For example, boolean values could be represented by a single bit instead
// of a uint8_t. However at the current time there are concerns about the portability
// of bitfield operators, and complexity issues with using bit mask operations. This is
// something likely we will want to investigate as the Vulkan implementation progresses.
//
// Second implementation note: the struct packing is also a bit fragile, and some of the
// packing requirements depend on using alignas and field ordering to get the result of
// packing nicely into the desired space. This is something we could also potentially fix
// with a redesign to use bitfields or bit mask operations.

struct alignas(4) PackedAttachmentDesc
{
    uint8_t flags;
    uint8_t samples;
    uint16_t format;
};

static_assert(sizeof(PackedAttachmentDesc) == 4, "Size check failed");

class RenderPassDesc final
{
  public:
    RenderPassDesc();
    ~RenderPassDesc();
    RenderPassDesc(const RenderPassDesc &other);
    RenderPassDesc &operator=(const RenderPassDesc &other);

    // Depth stencil attachments must be packed after color attachments.
    void packColorAttachment(const Format &format, GLsizei samples);
    void packDepthStencilAttachment(const Format &format, GLsizei samples);

    size_t hash() const;

    uint32_t attachmentCount() const;
    uint32_t colorAttachmentCount() const;
    uint32_t depthStencilAttachmentCount() const;
    const PackedAttachmentDesc &operator[](size_t index) const;

  private:
    void packAttachment(uint32_t index, const vk::Format &format, GLsizei samples);

    uint32_t mColorAttachmentCount;
    uint32_t mDepthStencilAttachmentCount;
    gl::AttachmentArray<PackedAttachmentDesc> mAttachmentDescs;
    uint32_t mPadding[4];
};

bool operator==(const RenderPassDesc &lhs, const RenderPassDesc &rhs);

static_assert(sizeof(RenderPassDesc) == 64, "Size check failed");

struct alignas(8) PackedAttachmentOpsDesc final
{
    uint8_t loadOp;
    uint8_t storeOp;
    uint8_t stencilLoadOp;
    uint8_t stencilStoreOp;

    // 16-bits to force pad the structure to exactly 8 bytes.
    uint16_t initialLayout;
    uint16_t finalLayout;
};

static_assert(sizeof(PackedAttachmentOpsDesc) == 8, "Size check failed");

class AttachmentOpsArray final
{
  public:
    AttachmentOpsArray();
    ~AttachmentOpsArray();
    AttachmentOpsArray(const AttachmentOpsArray &other);
    AttachmentOpsArray &operator=(const AttachmentOpsArray &other);

    const PackedAttachmentOpsDesc &operator[](size_t index) const;
    PackedAttachmentOpsDesc &operator[](size_t index);

    // Initializes an attachment op with whatever values. Used for compatible RenderPass checks.
    void initDummyOp(size_t index, VkImageLayout initialLayout, VkImageLayout finalLayout);

    size_t hash() const;

  private:
    gl::AttachmentArray<PackedAttachmentOpsDesc> mOps;
};

bool operator==(const AttachmentOpsArray &lhs, const AttachmentOpsArray &rhs);

static_assert(sizeof(AttachmentOpsArray) == 80, "Size check failed");

struct alignas(8) PackedShaderStageInfo final
{
    uint32_t stage;
    uint32_t moduleSerial;
    // TODO(jmadill): Do we want specialization constants?
};

static_assert(sizeof(PackedShaderStageInfo) == 8, "Size check failed");

struct alignas(4) PackedVertexInputBindingDesc final
{
    // Although techncially stride can be any value in ES 2.0, in practice supporting stride
    // greater than MAX_USHORT should not be that helpful. Note that stride limits are
    // introduced in ES 3.1.
    uint16_t stride;
    uint16_t inputRate;
};

static_assert(sizeof(PackedVertexInputBindingDesc) == 4, "Size check failed");

struct alignas(8) PackedVertexInputAttributeDesc final
{
    uint16_t location;
    uint16_t format;
    uint32_t offset;
};

static_assert(sizeof(PackedVertexInputAttributeDesc) == 8, "Size check failed");

struct alignas(8) PackedInputAssemblyInfo
{
    uint32_t topology;
    uint32_t primitiveRestartEnable;
};

static_assert(sizeof(PackedInputAssemblyInfo) == 8, "Size check failed");

struct alignas(32) PackedRasterizationStateInfo
{
    // Padded to ensure there's no gaps in this structure or those that use it.
    uint32_t depthClampEnable;
    uint32_t rasterizationDiscardEnable;
    uint16_t polygonMode;
    uint16_t cullMode;
    uint16_t frontFace;
    uint16_t depthBiasEnable;
    float depthBiasConstantFactor;
    // Note: depth bias clamp is only exposed in a 3.1 extension, but left here for completeness.
    float depthBiasClamp;
    float depthBiasSlopeFactor;
    float lineWidth;
};

static_assert(sizeof(PackedRasterizationStateInfo) == 32, "Size check failed");

struct alignas(16) PackedMultisampleStateInfo final
{
    uint8_t rasterizationSamples;
    uint8_t sampleShadingEnable;
    uint8_t alphaToCoverageEnable;
    uint8_t alphaToOneEnable;
    float minSampleShading;
    uint32_t sampleMask[gl::MAX_SAMPLE_MASK_WORDS];
};

static_assert(sizeof(PackedMultisampleStateInfo) == 16, "Size check failed");

struct alignas(16) PackedStencilOpState final
{
    uint8_t failOp;
    uint8_t passOp;
    uint8_t depthFailOp;
    uint8_t compareOp;
    uint32_t compareMask;
    uint32_t writeMask;
    uint32_t reference;
};

static_assert(sizeof(PackedStencilOpState) == 16, "Size check failed");

struct PackedDepthStencilStateInfo final
{
    uint8_t depthTestEnable;
    uint8_t depthWriteEnable;
    uint8_t depthCompareOp;
    uint8_t depthBoundsTestEnable;
    // 32-bits to pad the alignments.
    uint32_t stencilTestEnable;
    float minDepthBounds;
    float maxDepthBounds;
    PackedStencilOpState front;
    PackedStencilOpState back;
};

static_assert(sizeof(PackedDepthStencilStateInfo) == 48, "Size check failed");

struct alignas(8) PackedColorBlendAttachmentState final
{
    uint8_t blendEnable;
    uint8_t srcColorBlendFactor;
    uint8_t dstColorBlendFactor;
    uint8_t colorBlendOp;
    uint8_t srcAlphaBlendFactor;
    uint8_t dstAlphaBlendFactor;
    uint8_t alphaBlendOp;
    uint8_t colorWriteMask;
};

static_assert(sizeof(PackedColorBlendAttachmentState) == 8, "Size check failed");

struct PackedColorBlendStateInfo final
{
    // Padded to round the strut size.
    uint32_t logicOpEnable;
    uint32_t logicOp;
    uint32_t attachmentCount;
    float blendConstants[4];
    PackedColorBlendAttachmentState attachments[gl::IMPLEMENTATION_MAX_DRAW_BUFFERS];
};

static_assert(sizeof(PackedColorBlendStateInfo) == 96, "Size check failed");

using ShaderStageInfo       = std::array<PackedShaderStageInfo, 2>;
using VertexInputBindings   = gl::AttribArray<PackedVertexInputBindingDesc>;
using VertexInputAttributes = gl::AttribArray<PackedVertexInputAttributeDesc>;

class PipelineDesc final
{
  public:
    // Use aligned allocation and free so we can use the alignas keyword.
    void *operator new(std::size_t size);
    void operator delete(void *ptr);

    PipelineDesc();
    ~PipelineDesc();
    PipelineDesc(const PipelineDesc &other);
    PipelineDesc &operator=(const PipelineDesc &other);

    size_t hash() const;
    bool operator==(const PipelineDesc &other) const;

    void initDefaults();

    Error initializePipeline(VkDevice device,
                             const RenderPass &compatibleRenderPass,
                             const PipelineLayout &pipelineLayout,
                             const gl::AttributesMask &activeAttribLocationsMask,
                             const ShaderModule &vertexModule,
                             const ShaderModule &fragmentModule,
                             Pipeline *pipelineOut) const;

    void updateViewport(const gl::Rectangle &viewport, float nearPlane, float farPlane);

    // Shader stage info
    const ShaderStageInfo &getShaderStageInfo() const;
    void updateShaders(ProgramVk *programVk);

    // Vertex input state
    void updateVertexInputInfo(const VertexInputBindings &bindings,
                               const VertexInputAttributes &attribs);

    // Input assembly info
    void updateTopology(GLenum drawMode);

    // Raster states
    void updateCullMode(const gl::RasterizerState &rasterState);
    void updateFrontFace(const gl::RasterizerState &rasterState);
    void updateLineWidth(float lineWidth);

    // RenderPass description.
    const RenderPassDesc &getRenderPassDesc() const;
    void updateRenderPassDesc(const RenderPassDesc &renderPassDesc);

    // Scissor support
    const VkRect2D &getScissor() const { return mScissor; }
    void updateScissor(const gl::Rectangle &rect);

    // Blend states
    void updateBlendEnabled(bool isBlendEnabled);
    void updateBlendColor(const gl::ColorF &color);
    void updateBlendFuncs(const gl::BlendState &blend_state);
    void updateBlendEquations(const gl::BlendState &blend_state);

    // Depth/stencil states.
    void updateDepthTestEnabled(const gl::DepthStencilState &depthStencilState);
    void updateDepthFunc(const gl::DepthStencilState &depthStencilState);

  private:
    // TODO(jmadill): Handle Geometry/Compute shaders when necessary.
    ShaderStageInfo mShaderStageInfo;
    VertexInputBindings mVertexInputBindings;
    VertexInputAttributes mVertexInputAttribs;
    PackedInputAssemblyInfo mInputAssemblyInfo;
    // TODO(jmadill): Consider using dynamic state for viewport/scissor.
    VkViewport mViewport;
    VkRect2D mScissor;
    PackedRasterizationStateInfo mRasterizationStateInfo;
    PackedMultisampleStateInfo mMultisampleStateInfo;
    PackedDepthStencilStateInfo mDepthStencilStateInfo;
    PackedColorBlendStateInfo mColorBlendStateInfo;
    // TODO(jmadill): Dynamic state.
    // TODO(jmadill): Pipeline layout
    RenderPassDesc mRenderPassDesc;
};

// Verify the packed pipeline description has no gaps in the packing.
// This is not guaranteed by the spec, but is validated by a compile-time check.
// No gaps or padding at the end ensures that hashing and memcmp checks will not run
// into uninitialized memory regions.
constexpr size_t PipelineDescSumOfSizes =
    sizeof(ShaderStageInfo) + sizeof(VertexInputBindings) + sizeof(VertexInputAttributes) +
    sizeof(PackedInputAssemblyInfo) + sizeof(VkViewport) + sizeof(VkRect2D) +
    sizeof(PackedRasterizationStateInfo) + sizeof(PackedMultisampleStateInfo) +
    sizeof(PackedDepthStencilStateInfo) + sizeof(PackedColorBlendStateInfo) +
    sizeof(RenderPassDesc);

static_assert(sizeof(PipelineDesc) == PipelineDescSumOfSizes, "Size mismatch");

using RenderPassAndSerial = ObjectAndSerial<RenderPass>;
using PipelineAndSerial   = ObjectAndSerial<Pipeline>;
}  // namespace vk
}  // namespace rx

// Introduce a std::hash for a RenderPassDesc
namespace std
{
template <>
struct hash<rx::vk::RenderPassDesc>
{
    size_t operator()(const rx::vk::RenderPassDesc &key) const { return key.hash(); }
};

template <>
struct hash<rx::vk::AttachmentOpsArray>
{
    size_t operator()(const rx::vk::AttachmentOpsArray &key) const { return key.hash(); }
};

template <>
struct hash<rx::vk::PipelineDesc>
{
    size_t operator()(const rx::vk::PipelineDesc &key) const { return key.hash(); }
};

}  // namespace std

namespace rx
{
// TODO(jmadill): Add cache trimming/eviction.
class RenderPassCache final : angle::NonCopyable
{
  public:
    RenderPassCache();
    ~RenderPassCache();

    void destroy(VkDevice device);

    vk::Error getCompatibleRenderPass(VkDevice device,
                                      Serial serial,
                                      const vk::RenderPassDesc &desc,
                                      vk::RenderPass **renderPassOut);
    vk::Error getRenderPassWithOps(VkDevice device,
                                   Serial serial,
                                   const vk::RenderPassDesc &desc,
                                   const vk::AttachmentOpsArray &attachmentOps,
                                   vk::RenderPass **renderPassOut);

  private:
    // Use a two-layer caching scheme. The top level matches the "compatible" RenderPass elements.
    // The second layer caches the attachment load/store ops and initial/final layout.
    using InnerCache = std::unordered_map<vk::AttachmentOpsArray, vk::RenderPassAndSerial>;
    using OuterCache = std::unordered_map<vk::RenderPassDesc, InnerCache>;

    OuterCache mPayload;
};

// TODO(jmadill): Add cache trimming/eviction.
class PipelineCache final : angle::NonCopyable
{
  public:
    PipelineCache();
    ~PipelineCache();

    void destroy(VkDevice device);

    void populate(const vk::PipelineDesc &desc, vk::Pipeline &&pipeline);
    vk::Error getPipeline(VkDevice device,
                          const vk::RenderPass &compatibleRenderPass,
                          const vk::PipelineLayout &pipelineLayout,
                          const gl::AttributesMask &activeAttribLocationsMask,
                          const vk::ShaderModule &vertexModule,
                          const vk::ShaderModule &fragmentModule,
                          const vk::PipelineDesc &desc,
                          vk::PipelineAndSerial **pipelineOut);

  private:
    std::unordered_map<vk::PipelineDesc, vk::PipelineAndSerial> mPayload;
};

}  // namespace rx

#endif  // LIBANGLE_RENDERER_VULKAN_VK_CACHE_UTILS_H_
kc3-lang/angle/src/libANGLE/renderer/vulkan/vk_cache_utils.h

Commit

src/libANGLE/renderer/vulkan/vk_cache_utils.h
