Rewrite host GPU abstraction

- Don't have to repeat the same thing for 4 renderers.
 - Add native Metal renderer.

src/util/opengl_pipeline.h

@@ -0,0 +1,107 @@
+// SPDX-FileCopyrightText: 2019-2023 Connor McLaughlin <stenzek@gmail.com>
+// SPDX-License-Identifier: (GPL-3.0 OR CC-BY-NC-ND-4.0)
+
+#pragma once
+
+#include "gpu_device.h"
+#include "gpu_shader_cache.h"
+#include "opengl_loader.h"
+
+class OpenGLDevice;
+
+class OpenGLShader final : public GPUShader
+{
+  friend OpenGLDevice;
+
+public:
+  ~OpenGLShader() override;
+
+  void SetDebugName(const std::string_view& name) override;
+
+  ALWAYS_INLINE GLuint GetGLId() const { return m_id; }
+  ALWAYS_INLINE const GPUShaderCache::CacheIndexKey& GetKey() const { return m_key; }
+
+private:
+  OpenGLShader(GPUShaderStage stage, GLuint id, const GPUShaderCache::CacheIndexKey& key);
+
+  GLuint m_id;
+  GPUShaderCache::CacheIndexKey m_key;
+};
+
+class OpenGLPipeline final : public GPUPipeline
+{
+  friend OpenGLDevice;
+
+public:
+  static constexpr u32 MAX_VERTEX_ATTRIBUTES = 6;
+
+  struct VertexArrayCacheKey
+  {
+    VertexAttribute vertex_attributes[MAX_VERTEX_ATTRIBUTES];
+    u32 vertex_attribute_stride;
+    u32 num_vertex_attributes;
+
+    bool operator==(const VertexArrayCacheKey& rhs) const;
+    bool operator!=(const VertexArrayCacheKey& rhs) const;
+  };
+  struct VertexArrayCacheItem
+  {
+    GLuint vao_id;
+    u32 reference_count;
+  };
+  struct VertexArrayCacheKeyHash
+  {
+    size_t operator()(const VertexArrayCacheKey& k) const;
+  };
+  using VertexArrayCache = std::unordered_map<VertexArrayCacheKey, VertexArrayCacheItem, VertexArrayCacheKeyHash>;
+
+  struct ProgramCacheKey
+  {
+    GPUShaderCache::CacheIndexKey vs_key;
+    GPUShaderCache::CacheIndexKey fs_key;
+    VertexArrayCacheKey va_key;
+
+    bool operator==(const ProgramCacheKey& rhs) const;
+    bool operator!=(const ProgramCacheKey& rhs) const;
+  };
+  static_assert(sizeof(ProgramCacheKey) == 112); // Has no padding
+  struct ProgramCacheKeyHash
+  {
+    size_t operator()(const ProgramCacheKey& k) const;
+  };
+  struct ProgramCacheItem
+  {
+    GLuint program_id;
+    u32 reference_count;
+    GLenum file_format;
+    u32 file_offset;
+    u32 file_compressed_size;
+    u32 file_uncompressed_size;
+  };
+  using ProgramCache = std::unordered_map<ProgramCacheKey, ProgramCacheItem, ProgramCacheKeyHash>;
+
+  static ProgramCacheKey GetProgramCacheKey(const GraphicsConfig& plconfig);
+
+  ~OpenGLPipeline() override;
+
+  ALWAYS_INLINE GLuint GetProgram() const { return m_program; }
+  ALWAYS_INLINE GLuint GetVAO() const { return m_vao; }
+  ALWAYS_INLINE const RasterizationState& GetRasterizationState() const { return m_rasterization_state; }
+  ALWAYS_INLINE const DepthState& GetDepthState() const { return m_depth_state; }
+  ALWAYS_INLINE const BlendState& GetBlendState() const { return m_blend_state; }
+  ALWAYS_INLINE GLenum GetTopology() const { return m_topology; }
+
+  void SetDebugName(const std::string_view& name) override;
+
+private:
+  OpenGLPipeline(const ProgramCacheKey& key, GLuint program, GLuint vao, const RasterizationState& rs,
+                 const DepthState& ds, const BlendState& bs, GLenum topology);
+
+  ProgramCacheKey m_key;
+  GLuint m_program;
+  GLuint m_vao;
+  BlendState m_blend_state;
+  RasterizationState m_rasterization_state;
+  DepthState m_depth_state;
+  GLenum m_topology;
+};