13 files changed, 40 insertions, 219 deletions

diff --git a/src/video_core/renderer_opengl/gl_buffer_cache.cpp b/src/video_core/renderer_opengl/gl_buffer_cache.cpp
index 25652e794..48b86f3bd 100644
--- a/src/video_core/renderer_opengl/gl_buffer_cache.cpp
+++ b/src/video_core/renderer_opengl/gl_buffer_cache.cpp
@@ -71,16 +71,6 @@ GLintptr OGLBufferCache::UploadHostMemory(const void* raw_pointer, std::size_t s
     return uploaded_offset;
 }
 
-std::tuple<u8*, GLintptr> OGLBufferCache::ReserveMemory(std::size_t size, std::size_t alignment) {
-    AlignBuffer(alignment);
-    u8* const uploaded_ptr = buffer_ptr;
-    const GLintptr uploaded_offset = buffer_offset;
-
-    buffer_ptr += size;
-    buffer_offset += size;
-    return std::make_tuple(uploaded_ptr, uploaded_offset);
-}
-
 bool OGLBufferCache::Map(std::size_t max_size) {
     bool invalidate;
     std::tie(buffer_ptr, buffer_offset_base, invalidate) =
diff --git a/src/video_core/renderer_opengl/gl_buffer_cache.h b/src/video_core/renderer_opengl/gl_buffer_cache.h
index f9247a40e..f2347581b 100644
--- a/src/video_core/renderer_opengl/gl_buffer_cache.h
+++ b/src/video_core/renderer_opengl/gl_buffer_cache.h
@@ -61,9 +61,6 @@ public:
     /// Uploads from a host memory. Returns host's buffer offset where it's been allocated.
     GLintptr UploadHostMemory(const void* raw_pointer, std::size_t size, std::size_t alignment = 4);
 
-    /// Reserves memory to be used by host's CPU. Returns mapped address and offset.
-    std::tuple<u8*, GLintptr> ReserveMemory(std::size_t size, std::size_t alignment = 4);
-
     bool Map(std::size_t max_size);
     void Unmap();
 
diff --git a/src/video_core/renderer_opengl/gl_primitive_assembler.cpp b/src/video_core/renderer_opengl/gl_primitive_assembler.cpp
deleted file mode 100644
index c3e94d917..000000000
--- a/src/video_core/renderer_opengl/gl_primitive_assembler.cpp
+++ /dev/null
@@ -1,63 +0,0 @@
-// Copyright 2018 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include <algorithm>
-#include <array>
-#include "common/assert.h"
-#include "common/common_types.h"
-#include "core/core.h"
-#include "video_core/memory_manager.h"
-#include "video_core/renderer_opengl/gl_buffer_cache.h"
-#include "video_core/renderer_opengl/gl_primitive_assembler.h"
-
-namespace OpenGL {
-
-constexpr u32 TRIANGLES_PER_QUAD = 6;
-constexpr std::array<u32, TRIANGLES_PER_QUAD> QUAD_MAP = {0, 1, 2, 0, 2, 3};
-
-PrimitiveAssembler::PrimitiveAssembler(OGLBufferCache& buffer_cache) : buffer_cache(buffer_cache) {}
-
-PrimitiveAssembler::~PrimitiveAssembler() = default;
-
-std::size_t PrimitiveAssembler::CalculateQuadSize(u32 count) const {
-    ASSERT_MSG(count % 4 == 0, "Quad count is expected to be a multiple of 4");
-    return (count / 4) * TRIANGLES_PER_QUAD * sizeof(GLuint);
-}
-
-GLintptr PrimitiveAssembler::MakeQuadArray(u32 first, u32 count) {
-    const std::size_t size{CalculateQuadSize(count)};
-    auto [dst_pointer, index_offset] = buffer_cache.ReserveMemory(size);
-
-    for (u32 primitive = 0; primitive < count / 4; ++primitive) {
-        for (u32 i = 0; i < TRIANGLES_PER_QUAD; ++i) {
-            const u32 index = first + primitive * 4 + QUAD_MAP[i];
-            std::memcpy(dst_pointer, &index, sizeof(index));
-            dst_pointer += sizeof(index);
-        }
-    }
-
-    return index_offset;
-}
-
-GLintptr PrimitiveAssembler::MakeQuadIndexed(GPUVAddr gpu_addr, std::size_t index_size, u32 count) {
-    const std::size_t map_size{CalculateQuadSize(count)};
-    auto [dst_pointer, index_offset] = buffer_cache.ReserveMemory(map_size);
-
-    auto& memory_manager = Core::System::GetInstance().GPU().MemoryManager();
-    const u8* source{memory_manager.GetPointer(gpu_addr)};
-
-    for (u32 primitive = 0; primitive < count / 4; ++primitive) {
-        for (std::size_t i = 0; i < TRIANGLES_PER_QUAD; ++i) {
-            const u32 index = primitive * 4 + QUAD_MAP[i];
-            const u8* src_offset = source + (index * index_size);
-
-            std::memcpy(dst_pointer, src_offset, index_size);
-            dst_pointer += index_size;
-        }
-    }
-
-    return index_offset;
-}
-
-} // namespace OpenGL
diff --git a/src/video_core/renderer_opengl/gl_primitive_assembler.h b/src/video_core/renderer_opengl/gl_primitive_assembler.h
deleted file mode 100644
index 4e87ce4d6..000000000
--- a/src/video_core/renderer_opengl/gl_primitive_assembler.h
+++ /dev/null
@@ -1,31 +0,0 @@
-// Copyright 2018 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include <glad/glad.h>
-
-#include "common/common_types.h"
-
-namespace OpenGL {
-
-class OGLBufferCache;
-
-class PrimitiveAssembler {
-public:
-    explicit PrimitiveAssembler(OGLBufferCache& buffer_cache);
-    ~PrimitiveAssembler();
-
-    /// Calculates the size required by MakeQuadArray and MakeQuadIndexed.
-    std::size_t CalculateQuadSize(u32 count) const;
-
-    GLintptr MakeQuadArray(u32 first, u32 count);
-
-    GLintptr MakeQuadIndexed(GPUVAddr gpu_addr, std::size_t index_size, u32 count);
-
-private:
-    OGLBufferCache& buffer_cache;
-};
-
-} // namespace OpenGL
\ No newline at end of file
diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp
index f9b6dfeea..ca410287a 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer.cpp
+++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp
@@ -246,29 +246,6 @@ DrawParameters RasterizerOpenGL::SetupDraw() {
     DrawParameters params{};
     params.current_instance = gpu.state.current_instance;
 
-    if (regs.draw.topology == Maxwell::PrimitiveTopology::Quads) {
-        MICROPROFILE_SCOPE(OpenGL_PrimitiveAssembly);
-
-        params.use_indexed = true;
-        params.primitive_mode = GL_TRIANGLES;
-
-        if (is_indexed) {
-            params.index_format = MaxwellToGL::IndexFormat(regs.index_array.format);
-            params.count = (regs.index_array.count / 4) * 6;
-            params.index_buffer_offset = primitive_assembler.MakeQuadIndexed(
-                regs.index_array.IndexStart(), regs.index_array.FormatSizeInBytes(),
-                regs.index_array.count);
-            params.base_vertex = static_cast<GLint>(regs.vb_element_base);
-        } else {
-            // MakeQuadArray always generates u32 indexes
-            params.index_format = GL_UNSIGNED_INT;
-            params.count = (regs.vertex_buffer.count / 4) * 6;
-            params.index_buffer_offset = primitive_assembler.MakeQuadArray(
-                regs.vertex_buffer.first, regs.vertex_buffer.count);
-        }
-        return params;
-    }
-
     params.use_indexed = is_indexed;
     params.primitive_mode = MaxwellToGL::PrimitiveTopology(regs.draw.topology);
 
@@ -686,30 +663,19 @@ void RasterizerOpenGL::DrawArrays() {
     SyncCullMode();
     SyncPrimitiveRestart();
     SyncScissorTest(state);
-    // Alpha Testing is synced on shaders.
     SyncTransformFeedback();
     SyncPointState();
-    CheckAlphaTests();
     SyncPolygonOffset();
-    // TODO(bunnei): Sync framebuffer_scale uniform here
-    // TODO(bunnei): Sync scissorbox uniform(s) here
+    SyncAlphaTest();
 
     // Draw the vertex batch
     const bool is_indexed = accelerate_draw == AccelDraw::Indexed;
 
     std::size_t buffer_size = CalculateVertexArraysSize();
 
-    // Add space for index buffer (keeping in mind non-core primitives)
-    switch (regs.draw.topology) {
-    case Maxwell::PrimitiveTopology::Quads:
-        buffer_size = Common::AlignUp(buffer_size, 4) +
-                      primitive_assembler.CalculateQuadSize(regs.vertex_buffer.count);
-        break;
-    default:
-        if (is_indexed) {
-            buffer_size = Common::AlignUp(buffer_size, 4) + CalculateIndexBufferSize();
-        }
-        break;
+    // Add space for index buffer
+    if (is_indexed) {
+        buffer_size = Common::AlignUp(buffer_size, 4) + CalculateIndexBufferSize();
     }
 
     // Uniform space for the 5 shader stages
@@ -1152,10 +1118,17 @@ void RasterizerOpenGL::SyncPolygonOffset() {
     state.polygon_offset.clamp = regs.polygon_offset_clamp;
 }
 
-void RasterizerOpenGL::CheckAlphaTests() {
+void RasterizerOpenGL::SyncAlphaTest() {
     const auto& regs = system.GPU().Maxwell3D().regs;
     UNIMPLEMENTED_IF_MSG(regs.alpha_test_enabled != 0 && regs.rt_control.count > 1,
                          "Alpha Testing is enabled with more than one rendertarget");
+
+    state.alpha_test.enabled = regs.alpha_test_enabled;
+    if (!state.alpha_test.enabled) {
+        return;
+    }
+    state.alpha_test.func = MaxwellToGL::ComparisonOp(regs.alpha_test_func);
+    state.alpha_test.ref = regs.alpha_test_ref;
 }
 
 } // namespace OpenGL
diff --git a/src/video_core/renderer_opengl/gl_rasterizer.h b/src/video_core/renderer_opengl/gl_rasterizer.h
index d78094138..2817f65c9 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer.h
+++ b/src/video_core/renderer_opengl/gl_rasterizer.h
@@ -23,7 +23,6 @@
 #include "video_core/renderer_opengl/gl_buffer_cache.h"
 #include "video_core/renderer_opengl/gl_device.h"
 #include "video_core/renderer_opengl/gl_global_cache.h"
-#include "video_core/renderer_opengl/gl_primitive_assembler.h"
 #include "video_core/renderer_opengl/gl_rasterizer_cache.h"
 #include "video_core/renderer_opengl/gl_resource_manager.h"
 #include "video_core/renderer_opengl/gl_sampler_cache.h"
@@ -167,8 +166,8 @@ private:
     /// Syncs the polygon offsets
     void SyncPolygonOffset();
 
-    /// Check asserts for alpha testing.
-    void CheckAlphaTests();
+    /// Syncs the alpha test state to match the guest state
+    void SyncAlphaTest();
 
     /// Check for extension that are not strictly required
     /// but are needed for correct emulation
@@ -197,7 +196,6 @@ private:
 
     static constexpr std::size_t STREAM_BUFFER_SIZE = 128 * 1024 * 1024;
     OGLBufferCache buffer_cache;
-    PrimitiveAssembler primitive_assembler{buffer_cache};
 
     BindBuffersRangePushBuffer bind_ubo_pushbuffer{GL_UNIFORM_BUFFER};
     BindBuffersRangePushBuffer bind_ssbo_pushbuffer{GL_SHADER_STORAGE_BUFFER};
diff --git a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp
index f2d0722af..739477cc9 100644
--- a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp
+++ b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp
@@ -1467,27 +1467,9 @@ private:
 
         UNIMPLEMENTED_IF_MSG(header.ps.omap.sample_mask != 0, "Sample mask write is unimplemented");
 
-        code.AddLine("if (alpha_test[0] != 0) {{");
-        ++code.scope;
-        // We start on the register containing the alpha value in the first RT.
-        u32 current_reg = 3;
-        for (u32 render_target = 0; render_target < Maxwell::NumRenderTargets; ++render_target) {
-            // TODO(Blinkhawk): verify the behavior of alpha testing on hardware when
-            // multiple render targets are used.
-            if (header.ps.IsColorComponentOutputEnabled(render_target, 0) ||
-                header.ps.IsColorComponentOutputEnabled(render_target, 1) ||
-                header.ps.IsColorComponentOutputEnabled(render_target, 2) ||
-                header.ps.IsColorComponentOutputEnabled(render_target, 3)) {
-                code.AddLine("if (!AlphaFunc({})) discard;", SafeGetRegister(current_reg));
-                current_reg += 4;
-            }
-        }
-        --code.scope;
-        code.AddLine("}}");
-
         // Write the color outputs using the data in the shader registers, disabled
         // rendertargets/components are skipped in the register assignment.
-        current_reg = 0;
+        u32 current_reg = 0;
         for (u32 render_target = 0; render_target < Maxwell::NumRenderTargets; ++render_target) {
             // TODO(Subv): Figure out how dual-source blending is configured in the Switch.
             for (u32 component = 0; component < 4; ++component) {
diff --git a/src/video_core/renderer_opengl/gl_shader_gen.cpp b/src/video_core/renderer_opengl/gl_shader_gen.cpp
index c845b29aa..9148629ec 100644
--- a/src/video_core/renderer_opengl/gl_shader_gen.cpp
+++ b/src/video_core/renderer_opengl/gl_shader_gen.cpp
@@ -26,7 +26,6 @@ ProgramResult GenerateVertexShader(const Device& device, const ShaderSetup& setu
 layout (std140, binding = EMULATION_UBO_BINDING) uniform vs_config {
     vec4 viewport_flip;
     uvec4 config_pack; // instance_id, flip_stage, y_direction, padding
-    uvec4 alpha_test;
 };
 
 )";
@@ -78,7 +77,6 @@ ProgramResult GenerateGeometryShader(const Device& device, const ShaderSetup& se
 layout (std140, binding = EMULATION_UBO_BINDING) uniform gs_config {
     vec4 viewport_flip;
     uvec4 config_pack; // instance_id, flip_stage, y_direction, padding
-    uvec4 alpha_test;
 };
 
 )";
@@ -114,33 +112,8 @@ layout (location = 7) out vec4 FragColor7;
 layout (std140, binding = EMULATION_UBO_BINDING) uniform fs_config {
     vec4 viewport_flip;
     uvec4 config_pack; // instance_id, flip_stage, y_direction, padding
-    uvec4 alpha_test;
 };
 
-bool AlphaFunc(in float value) {
-    float ref = uintBitsToFloat(alpha_test[2]);
-    switch (alpha_test[1]) {
-        case 1:
-            return false;
-        case 2:
-            return value < ref;
-        case 3:
-            return value == ref;
-        case 4:
-            return value <= ref;
-        case 5:
-            return value > ref;
-        case 6:
-            return value != ref;
-        case 7:
-            return value >= ref;
-        case 8:
-            return true;
-        default:
-            return false;
-    }
-}
-
 )";
     const ShaderIR program_ir(setup.program.code, PROGRAM_OFFSET);
     ProgramResult program =
diff --git a/src/video_core/renderer_opengl/gl_shader_manager.cpp b/src/video_core/renderer_opengl/gl_shader_manager.cpp
index 05ab01dcb..b05f90f20 100644
--- a/src/video_core/renderer_opengl/gl_shader_manager.cpp
+++ b/src/video_core/renderer_opengl/gl_shader_manager.cpp
@@ -48,17 +48,6 @@ void MaxwellUniformData::SetFromRegs(const Maxwell3D& maxwell, std::size_t shade
     viewport_flip[0] = regs.viewport_transform[0].scale_x < 0.0 ? -1.0f : 1.0f;
     viewport_flip[1] = regs.viewport_transform[0].scale_y < 0.0 ? -1.0f : 1.0f;
 
-    auto func{static_cast<u32>(regs.alpha_test_func)};
-    // Normalize the gl variants of opCompare to be the same as the normal variants
-    const u32 op_gl_variant_base = static_cast<u32>(Maxwell3D::Regs::ComparisonOp::Never);
-    if (func >= op_gl_variant_base) {
-        func = func - op_gl_variant_base + 1U;
-    }
-
-    alpha_test.enabled = regs.alpha_test_enabled;
-    alpha_test.func = func;
-    alpha_test.ref = regs.alpha_test_ref;
-
     instance_id = state.current_instance;
 
     // Assign in which stage the position has to be flipped
diff --git a/src/video_core/renderer_opengl/gl_shader_manager.h b/src/video_core/renderer_opengl/gl_shader_manager.h
index cec18a832..6961e702a 100644
--- a/src/video_core/renderer_opengl/gl_shader_manager.h
+++ b/src/video_core/renderer_opengl/gl_shader_manager.h
@@ -27,14 +27,8 @@ struct MaxwellUniformData {
         GLuint flip_stage;
         GLfloat y_direction;
     };
-    struct alignas(16) {
-        GLuint enabled;
-        GLuint func;
-        GLfloat ref;
-        GLuint padding;
-    } alpha_test;
 };
-static_assert(sizeof(MaxwellUniformData) == 48, "MaxwellUniformData structure size is incorrect");
+static_assert(sizeof(MaxwellUniformData) == 32, "MaxwellUniformData structure size is incorrect");
 static_assert(sizeof(MaxwellUniformData) < 16384,
               "MaxwellUniformData structure must be less than 16kb as per the OpenGL spec");
 
diff --git a/src/video_core/renderer_opengl/gl_state.cpp b/src/video_core/renderer_opengl/gl_state.cpp
index 7425fbe5d..d86e137ac 100644
--- a/src/video_core/renderer_opengl/gl_state.cpp
+++ b/src/video_core/renderer_opengl/gl_state.cpp
@@ -156,6 +156,10 @@ OpenGLState::OpenGLState() {
     polygon_offset.factor = 0.0f;
     polygon_offset.units = 0.0f;
     polygon_offset.clamp = 0.0f;
+
+    alpha_test.enabled = false;
+    alpha_test.func = GL_ALWAYS;
+    alpha_test.ref = 0.0f;
 }
 
 void OpenGLState::ApplyDefaultState() {
@@ -461,6 +465,14 @@ void OpenGLState::ApplyPolygonOffset() const {
     }
 }
 
+void OpenGLState::ApplyAlphaTest() const {
+    Enable(GL_ALPHA_TEST, cur_state.alpha_test.enabled, alpha_test.enabled);
+    if (UpdateTie(std::tie(cur_state.alpha_test.func, cur_state.alpha_test.ref),
+                  std::tie(alpha_test.func, alpha_test.ref))) {
+        glAlphaFunc(alpha_test.func, alpha_test.ref);
+    }
+}
+
 void OpenGLState::ApplyTextures() const {
     bool has_delta{};
     std::size_t first{};
@@ -533,6 +545,7 @@ void OpenGLState::Apply() const {
     ApplyTextures();
     ApplySamplers();
     ApplyPolygonOffset();
+    ApplyAlphaTest();
 }
 
 void OpenGLState::EmulateViewportWithScissor() {
diff --git a/src/video_core/renderer_opengl/gl_state.h b/src/video_core/renderer_opengl/gl_state.h
index 41418a7b8..b0140495d 100644
--- a/src/video_core/renderer_opengl/gl_state.h
+++ b/src/video_core/renderer_opengl/gl_state.h
@@ -172,6 +172,12 @@ public:
         GLfloat clamp;
     } polygon_offset;
 
+    struct {
+        bool enabled; // GL_ALPHA_TEST
+        GLenum func;  // GL_ALPHA_TEST_FUNC
+        GLfloat ref;  // GL_ALPHA_TEST_REF
+    } alpha_test;
+
     std::array<bool, 8> clip_distance; // GL_CLIP_DISTANCE
 
     OpenGLState();
@@ -215,6 +221,7 @@ public:
     void ApplySamplers() const;
     void ApplyDepthClamp() const;
     void ApplyPolygonOffset() const;
+    void ApplyAlphaTest() const;
 
     /// Set the initial OpenGL state
     static void ApplyDefaultState();
diff --git a/src/video_core/renderer_opengl/maxwell_to_gl.h b/src/video_core/renderer_opengl/maxwell_to_gl.h
index ed7b5cff0..ea77dd211 100644
--- a/src/video_core/renderer_opengl/maxwell_to_gl.h
+++ b/src/video_core/renderer_opengl/maxwell_to_gl.h
@@ -128,6 +128,8 @@ inline GLenum PrimitiveTopology(Maxwell::PrimitiveTopology topology) {
         return GL_TRIANGLE_STRIP;
     case Maxwell::PrimitiveTopology::TriangleFan:
         return GL_TRIANGLE_FAN;
+    case Maxwell::PrimitiveTopology::Quads:
+        return GL_QUADS;
     default:
         LOG_CRITICAL(Render_OpenGL, "Unimplemented topology={}", static_cast<u32>(topology));
         UNREACHABLE();
@@ -173,11 +175,8 @@ inline GLenum WrapMode(Tegra::Texture::WrapMode wrap_mode) {
         return GL_CLAMP_TO_EDGE;
     case Tegra::Texture::WrapMode::Border:
         return GL_CLAMP_TO_BORDER;
-    case Tegra::Texture::WrapMode::ClampOGL:
-        // TODO(Subv): GL_CLAMP was removed as of OpenGL 3.1, to implement GL_CLAMP, we can use
-        // GL_CLAMP_TO_BORDER to get the border color of the texture, and then sample the edge to
-        // manually mix them. However the shader part of this is not yet implemented.
-        return GL_CLAMP_TO_BORDER;
+    case Tegra::Texture::WrapMode::Clamp:
+        return GL_CLAMP;
     case Tegra::Texture::WrapMode::MirrorOnceClampToEdge:
         return GL_MIRROR_CLAMP_TO_EDGE;
     case Tegra::Texture::WrapMode::MirrorOnceBorder: