1 files changed, 78 insertions, 5 deletions

diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp
index 7bced675c..edf527f2d 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer.cpp
+++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp
@@ -24,6 +24,7 @@
 #include "video_core/renderer_opengl/gl_query_cache.h"
 #include "video_core/renderer_opengl/gl_rasterizer.h"
 #include "video_core/renderer_opengl/gl_shader_cache.h"
+#include "video_core/renderer_opengl/gl_staging_buffer_pool.h"
 #include "video_core/renderer_opengl/gl_texture_cache.h"
 #include "video_core/renderer_opengl/maxwell_to_gl.h"
 #include "video_core/renderer_opengl/renderer_opengl.h"
@@ -58,12 +59,13 @@ RasterizerOpenGL::RasterizerOpenGL(Core::Frontend::EmuWindow& emu_window_, Tegra
                                    StateTracker& state_tracker_)
     : RasterizerAccelerated(cpu_memory_), gpu(gpu_), device(device_), screen_info(screen_info_),
       program_manager(program_manager_), state_tracker(state_tracker_),
-      texture_cache_runtime(device, program_manager, state_tracker),
-      texture_cache(texture_cache_runtime, *this), buffer_cache_runtime(device),
+      texture_cache_runtime(device, program_manager, state_tracker, staging_buffer_pool),
+      texture_cache(texture_cache_runtime, *this),
+      buffer_cache_runtime(device, staging_buffer_pool),
       buffer_cache(*this, cpu_memory_, buffer_cache_runtime),
       shader_cache(*this, emu_window_, device, texture_cache, buffer_cache, program_manager,
                    state_tracker, gpu.ShaderNotify()),
-      query_cache(*this), accelerate_dma(buffer_cache),
+      query_cache(*this, cpu_memory_), accelerate_dma(buffer_cache, texture_cache),
       fence_manager(*this, gpu, texture_cache, buffer_cache, query_cache),
       blit_image(program_manager_) {}
 
@@ -220,6 +222,9 @@ void RasterizerOpenGL::PrepareDraw(bool is_indexed, Func&& draw_func) {
     gpu.TickWork();
 
     std::scoped_lock lock{buffer_cache.mutex, texture_cache.mutex};
+    if (pipeline->UsesLocalMemory()) {
+        program_manager.LocalMemoryWarmup();
+    }
     pipeline->SetEngine(maxwell3d, gpu_memory);
     pipeline->Configure(is_indexed);
 
@@ -357,6 +362,7 @@ void RasterizerOpenGL::DrawTexture() {
                                         .y = static_cast<s32>(draw_texture_state.src_y1)}};
         blit_image.BlitColor(texture_cache.GetFramebuffer()->Handle(), texture.DefaultHandle(),
                              sampler->Handle(), dst_region, src_region, texture.size);
+        state_tracker.InvalidateState();
     }
 
     ++num_queued_commands;
@@ -368,6 +374,9 @@ void RasterizerOpenGL::DispatchCompute() {
     if (!pipeline) {
         return;
     }
+    if (pipeline->UsesLocalMemory()) {
+        program_manager.LocalMemoryWarmup();
+    }
     pipeline->SetEngine(kepler_compute, gpu_memory);
     pipeline->Configure();
     const auto& qmd{kepler_compute->launch_description};
@@ -432,6 +441,29 @@ bool RasterizerOpenGL::MustFlushRegion(VAddr addr, u64 size, VideoCommon::CacheT
     return false;
 }
 
+VideoCore::RasterizerDownloadArea RasterizerOpenGL::GetFlushArea(VAddr addr, u64 size) {
+    {
+        std::scoped_lock lock{texture_cache.mutex};
+        auto area = texture_cache.GetFlushArea(addr, size);
+        if (area) {
+            return *area;
+        }
+    }
+    {
+        std::scoped_lock lock{buffer_cache.mutex};
+        auto area = buffer_cache.GetFlushArea(addr, size);
+        if (area) {
+            return *area;
+        }
+    }
+    VideoCore::RasterizerDownloadArea new_area{
+        .start_address = Common::AlignDown(addr, Core::Memory::YUZU_PAGESIZE),
+        .end_address = Common::AlignUp(addr + size, Core::Memory::YUZU_PAGESIZE),
+        .preemtive = true,
+    };
+    return new_area;
+}
+
 void RasterizerOpenGL::InvalidateRegion(VAddr addr, u64 size, VideoCommon::CacheType which) {
     MICROPROFILE_SCOPE(OpenGL_CacheManagement);
     if (addr == 0 || size == 0) {
@@ -576,7 +608,7 @@ bool RasterizerOpenGL::AccelerateConditionalRendering() {
         // Reimplement Host conditional rendering.
         return false;
     }
-    // Medium / Low Hack: stub any checks on queries writen into the buffer cache.
+    // Medium / Low Hack: stub any checks on queries written into the buffer cache.
     const GPUVAddr condition_address{maxwell3d->regs.render_enable.Address()};
     Maxwell::ReportSemaphore::Compare cmp;
     if (gpu_memory->IsMemoryDirty(condition_address, sizeof(cmp),
@@ -1262,7 +1294,8 @@ void RasterizerOpenGL::ReleaseChannel(s32 channel_id) {
     query_cache.EraseChannel(channel_id);
 }
 
-AccelerateDMA::AccelerateDMA(BufferCache& buffer_cache_) : buffer_cache{buffer_cache_} {}
+AccelerateDMA::AccelerateDMA(BufferCache& buffer_cache_, TextureCache& texture_cache_)
+    : buffer_cache{buffer_cache_}, texture_cache{texture_cache_} {}
 
 bool AccelerateDMA::BufferCopy(GPUVAddr src_address, GPUVAddr dest_address, u64 amount) {
     std::scoped_lock lock{buffer_cache.mutex};
@@ -1274,4 +1307,44 @@ bool AccelerateDMA::BufferClear(GPUVAddr src_address, u64 amount, u32 value) {
     return buffer_cache.DMAClear(src_address, amount, value);
 }
 
+template <bool IS_IMAGE_UPLOAD>
+bool AccelerateDMA::DmaBufferImageCopy(const Tegra::DMA::ImageCopy& copy_info,
+                                       const Tegra::DMA::BufferOperand& buffer_operand,
+                                       const Tegra::DMA::ImageOperand& image_operand) {
+    std::scoped_lock lock{buffer_cache.mutex, texture_cache.mutex};
+    const auto image_id = texture_cache.DmaImageId(image_operand, IS_IMAGE_UPLOAD);
+    if (image_id == VideoCommon::NULL_IMAGE_ID) {
+        return false;
+    }
+    const u32 buffer_size = static_cast<u32>(buffer_operand.pitch * buffer_operand.height);
+    static constexpr auto sync_info = VideoCommon::ObtainBufferSynchronize::FullSynchronize;
+    const auto post_op = VideoCommon::ObtainBufferOperation::DoNothing;
+    const auto [buffer, offset] =
+        buffer_cache.ObtainBuffer(buffer_operand.address, buffer_size, sync_info, post_op);
+
+    const auto [image, copy] = texture_cache.DmaBufferImageCopy(
+        copy_info, buffer_operand, image_operand, image_id, IS_IMAGE_UPLOAD);
+    const std::span copy_span{&copy, 1};
+
+    if constexpr (IS_IMAGE_UPLOAD) {
+        image->UploadMemory(buffer->Handle(), offset, copy_span);
+    } else {
+        texture_cache.DownloadImageIntoBuffer(image, buffer->Handle(), offset, copy_span,
+                                              buffer_operand.address, buffer_size);
+    }
+    return true;
+}
+
+bool AccelerateDMA::ImageToBuffer(const Tegra::DMA::ImageCopy& copy_info,
+                                  const Tegra::DMA::ImageOperand& image_operand,
+                                  const Tegra::DMA::BufferOperand& buffer_operand) {
+    return DmaBufferImageCopy<false>(copy_info, buffer_operand, image_operand);
+}
+
+bool AccelerateDMA::BufferToImage(const Tegra::DMA::ImageCopy& copy_info,
+                                  const Tegra::DMA::BufferOperand& buffer_operand,
+                                  const Tegra::DMA::ImageOperand& image_operand) {
+    return DmaBufferImageCopy<true>(copy_info, buffer_operand, image_operand);
+}
+
 } // namespace OpenGL