From 0a2536a0df1f4aea406f2132d3edda0430acc9d1 Mon Sep 17 00:00:00 2001 From: Fernando Sahmkow Date: Mon, 25 Dec 2023 07:32:16 +0100 Subject: SMMU: Initial adaptation to video_core. --- src/video_core/renderer_vulkan/renderer_vulkan.cpp | 10 ++-- src/video_core/renderer_vulkan/renderer_vulkan.h | 5 +- src/video_core/renderer_vulkan/vk_blit_screen.cpp | 17 +++---- src/video_core/renderer_vulkan/vk_blit_screen.h | 9 ++-- src/video_core/renderer_vulkan/vk_buffer_cache.cpp | 10 ++-- src/video_core/renderer_vulkan/vk_buffer_cache.h | 7 ++- .../renderer_vulkan/vk_pipeline_cache.cpp | 5 +- src/video_core/renderer_vulkan/vk_pipeline_cache.h | 4 +- src/video_core/renderer_vulkan/vk_query_cache.cpp | 56 ++++++++++++---------- src/video_core/renderer_vulkan/vk_query_cache.h | 4 +- src/video_core/renderer_vulkan/vk_rasterizer.cpp | 43 +++++++++-------- src/video_core/renderer_vulkan/vk_rasterizer.h | 38 ++++++++------- 12 files changed, 109 insertions(+), 99 deletions(-) (limited to 'src/video_core/renderer_vulkan') diff --git a/src/video_core/renderer_vulkan/renderer_vulkan.cpp b/src/video_core/renderer_vulkan/renderer_vulkan.cpp index 100b70918..0e1815076 100644 --- a/src/video_core/renderer_vulkan/renderer_vulkan.cpp +++ b/src/video_core/renderer_vulkan/renderer_vulkan.cpp @@ -82,10 +82,10 @@ Device CreateDevice(const vk::Instance& instance, const vk::InstanceDispatch& dl RendererVulkan::RendererVulkan(Core::TelemetrySession& telemetry_session_, Core::Frontend::EmuWindow& emu_window, - Core::Memory::Memory& cpu_memory_, Tegra::GPU& gpu_, + Tegra::MaxwellDeviceMemoryManager& device_memory_, Tegra::GPU& gpu_, std::unique_ptr context_) try : RendererBase(emu_window, std::move(context_)), telemetry_session(telemetry_session_), - cpu_memory(cpu_memory_), gpu(gpu_), library(OpenLibrary(context.get())), + device_memory(device_memory_), gpu(gpu_), library(OpenLibrary(context.get())), instance(CreateInstance(*library, dld, VK_API_VERSION_1_1, render_window.GetWindowInfo().type, Settings::values.renderer_debug.GetValue())), debug_messenger(Settings::values.renderer_debug ? CreateDebugUtilsCallback(instance) @@ -97,9 +97,9 @@ RendererVulkan::RendererVulkan(Core::TelemetrySession& telemetry_session_, render_window.GetFramebufferLayout().height), present_manager(instance, render_window, device, memory_allocator, scheduler, swapchain, surface), - blit_screen(cpu_memory, render_window, device, memory_allocator, swapchain, present_manager, + blit_screen(device_memory, render_window, device, memory_allocator, swapchain, present_manager, scheduler, screen_info), - rasterizer(render_window, gpu, cpu_memory, screen_info, device, memory_allocator, + rasterizer(render_window, gpu, device_memory, screen_info, device, memory_allocator, state_tracker, scheduler) { if (Settings::values.renderer_force_max_clock.GetValue() && device.ShouldBoostClocks()) { turbo_mode.emplace(instance, dld); @@ -128,7 +128,7 @@ void RendererVulkan::SwapBuffers(const Tegra::FramebufferConfig* framebuffer) { screen_info.width = framebuffer->width; screen_info.height = framebuffer->height; - const VAddr framebuffer_addr = framebuffer->address + framebuffer->offset; + const DAddr framebuffer_addr = framebuffer->address + framebuffer->offset; const bool use_accelerated = rasterizer.AccelerateDisplay(*framebuffer, framebuffer_addr, framebuffer->stride); RenderScreenshot(*framebuffer, use_accelerated); diff --git a/src/video_core/renderer_vulkan/renderer_vulkan.h b/src/video_core/renderer_vulkan/renderer_vulkan.h index 14e257cf7..e5ce4692d 100644 --- a/src/video_core/renderer_vulkan/renderer_vulkan.h +++ b/src/video_core/renderer_vulkan/renderer_vulkan.h @@ -20,6 +20,7 @@ #include "video_core/vulkan_common/vulkan_device.h" #include "video_core/vulkan_common/vulkan_memory_allocator.h" #include "video_core/vulkan_common/vulkan_wrapper.h" +#include "video_core/host1x/gpu_device_memory_manager.h" namespace Core { class TelemetrySession; @@ -42,7 +43,7 @@ class RendererVulkan final : public VideoCore::RendererBase { public: explicit RendererVulkan(Core::TelemetrySession& telemtry_session, Core::Frontend::EmuWindow& emu_window, - Core::Memory::Memory& cpu_memory_, Tegra::GPU& gpu_, + Tegra::MaxwellDeviceMemoryManager& device_memory_, Tegra::GPU& gpu_, std::unique_ptr context_); ~RendererVulkan() override; @@ -62,7 +63,7 @@ private: void RenderScreenshot(const Tegra::FramebufferConfig& framebuffer, bool use_accelerated); Core::TelemetrySession& telemetry_session; - Core::Memory::Memory& cpu_memory; + Tegra::MaxwellDeviceMemoryManager& device_memory; Tegra::GPU& gpu; std::shared_ptr library; diff --git a/src/video_core/renderer_vulkan/vk_blit_screen.cpp b/src/video_core/renderer_vulkan/vk_blit_screen.cpp index 60432f5ad..610f27c84 100644 --- a/src/video_core/renderer_vulkan/vk_blit_screen.cpp +++ b/src/video_core/renderer_vulkan/vk_blit_screen.cpp @@ -14,8 +14,8 @@ #include "common/settings.h" #include "core/core.h" #include "core/frontend/emu_window.h" -#include "core/memory.h" #include "video_core/gpu.h" +#include "video_core/host1x/gpu_device_memory_manager.h" #include "video_core/host_shaders/fxaa_frag_spv.h" #include "video_core/host_shaders/fxaa_vert_spv.h" #include "video_core/host_shaders/present_bicubic_frag_spv.h" @@ -121,11 +121,12 @@ struct BlitScreen::BufferData { // Unaligned image data goes here }; -BlitScreen::BlitScreen(Core::Memory::Memory& cpu_memory_, Core::Frontend::EmuWindow& render_window_, - const Device& device_, MemoryAllocator& memory_allocator_, - Swapchain& swapchain_, PresentManager& present_manager_, - Scheduler& scheduler_, const ScreenInfo& screen_info_) - : cpu_memory{cpu_memory_}, render_window{render_window_}, device{device_}, +BlitScreen::BlitScreen(Tegra::MaxwellDeviceMemoryManager& device_memory_, + Core::Frontend::EmuWindow& render_window_, const Device& device_, + MemoryAllocator& memory_allocator_, Swapchain& swapchain_, + PresentManager& present_manager_, Scheduler& scheduler_, + const ScreenInfo& screen_info_) + : device_memory{device_memory_}, render_window{render_window_}, device{device_}, memory_allocator{memory_allocator_}, swapchain{swapchain_}, present_manager{present_manager_}, scheduler{scheduler_}, image_count{swapchain.GetImageCount()}, screen_info{screen_info_} { resource_ticks.resize(image_count); @@ -219,8 +220,8 @@ void BlitScreen::Draw(const Tegra::FramebufferConfig& framebuffer, if (!use_accelerated) { const u64 image_offset = GetRawImageOffset(framebuffer); - const VAddr framebuffer_addr = framebuffer.address + framebuffer.offset; - const u8* const host_ptr = cpu_memory.GetPointer(framebuffer_addr); + const DAddr framebuffer_addr = framebuffer.address + framebuffer.offset; + const u8* const host_ptr = device_memory.GetPointer(framebuffer_addr); // TODO(Rodrigo): Read this from HLE constexpr u32 block_height_log2 = 4; diff --git a/src/video_core/renderer_vulkan/vk_blit_screen.h b/src/video_core/renderer_vulkan/vk_blit_screen.h index 78b32416d..cb941a956 100644 --- a/src/video_core/renderer_vulkan/vk_blit_screen.h +++ b/src/video_core/renderer_vulkan/vk_blit_screen.h @@ -8,15 +8,12 @@ #include "core/frontend/framebuffer_layout.h" #include "video_core/vulkan_common/vulkan_memory_allocator.h" #include "video_core/vulkan_common/vulkan_wrapper.h" +#include "video_core/host1x/gpu_device_memory_manager.h" namespace Core { class System; } -namespace Core::Memory { -class Memory; -} - namespace Core::Frontend { class EmuWindow; } @@ -56,7 +53,7 @@ struct ScreenInfo { class BlitScreen { public: - explicit BlitScreen(Core::Memory::Memory& cpu_memory, Core::Frontend::EmuWindow& render_window, + explicit BlitScreen(Tegra::MaxwellDeviceMemoryManager& device_memory, Core::Frontend::EmuWindow& render_window, const Device& device, MemoryAllocator& memory_manager, Swapchain& swapchain, PresentManager& present_manager, Scheduler& scheduler, const ScreenInfo& screen_info); @@ -109,7 +106,7 @@ private: u64 CalculateBufferSize(const Tegra::FramebufferConfig& framebuffer) const; u64 GetRawImageOffset(const Tegra::FramebufferConfig& framebuffer) const; - Core::Memory::Memory& cpu_memory; + Tegra::MaxwellDeviceMemoryManager& device_memory; Core::Frontend::EmuWindow& render_window; const Device& device; MemoryAllocator& memory_allocator; diff --git a/src/video_core/renderer_vulkan/vk_buffer_cache.cpp b/src/video_core/renderer_vulkan/vk_buffer_cache.cpp index 3c61799fa..31001d142 100644 --- a/src/video_core/renderer_vulkan/vk_buffer_cache.cpp +++ b/src/video_core/renderer_vulkan/vk_buffer_cache.cpp @@ -79,7 +79,7 @@ vk::Buffer CreateBuffer(const Device& device, const MemoryAllocator& memory_allo } // Anonymous namespace Buffer::Buffer(BufferCacheRuntime& runtime, VideoCommon::NullBufferParams null_params) - : VideoCommon::BufferBase(null_params), tracker{4096} { + : VideoCommon::BufferBase(null_params), tracker{4096} { if (runtime.device.HasNullDescriptor()) { return; } @@ -88,11 +88,9 @@ Buffer::Buffer(BufferCacheRuntime& runtime, VideoCommon::NullBufferParams null_p is_null = true; } -Buffer::Buffer(BufferCacheRuntime& runtime, VideoCore::RasterizerInterface& rasterizer_, - VAddr cpu_addr_, u64 size_bytes_) - : VideoCommon::BufferBase(rasterizer_, cpu_addr_, size_bytes_), - device{&runtime.device}, buffer{CreateBuffer(*device, runtime.memory_allocator, SizeBytes())}, - tracker{SizeBytes()} { +Buffer::Buffer(BufferCacheRuntime& runtime, DAddr cpu_addr_, u64 size_bytes_) + : VideoCommon::BufferBase(cpu_addr_, size_bytes_), device{&runtime.device}, + buffer{CreateBuffer(*device, runtime.memory_allocator, SizeBytes())}, tracker{SizeBytes()} { if (runtime.device.HasDebuggingToolAttached()) { buffer.SetObjectNameEXT(fmt::format("Buffer 0x{:x}", CpuAddr()).c_str()); } diff --git a/src/video_core/renderer_vulkan/vk_buffer_cache.h b/src/video_core/renderer_vulkan/vk_buffer_cache.h index dc300d7cb..e273f4988 100644 --- a/src/video_core/renderer_vulkan/vk_buffer_cache.h +++ b/src/video_core/renderer_vulkan/vk_buffer_cache.h @@ -23,11 +23,10 @@ struct HostVertexBinding; class BufferCacheRuntime; -class Buffer : public VideoCommon::BufferBase { +class Buffer : public VideoCommon::BufferBase { public: explicit Buffer(BufferCacheRuntime&, VideoCommon::NullBufferParams null_params); - explicit Buffer(BufferCacheRuntime& runtime, VideoCore::RasterizerInterface& rasterizer_, - VAddr cpu_addr_, u64 size_bytes_); + explicit Buffer(BufferCacheRuntime& runtime, VAddr cpu_addr_, u64 size_bytes_); [[nodiscard]] VkBufferView View(u32 offset, u32 size, VideoCore::Surface::PixelFormat format); @@ -173,7 +172,7 @@ struct BufferCacheParams { using Runtime = Vulkan::BufferCacheRuntime; using Buffer = Vulkan::Buffer; using Async_Buffer = Vulkan::StagingBufferRef; - using MemoryTracker = VideoCommon::MemoryTrackerBase; + using MemoryTracker = VideoCommon::MemoryTrackerBase; static constexpr bool IS_OPENGL = false; static constexpr bool HAS_PERSISTENT_UNIFORM_BUFFER_BINDINGS = false; diff --git a/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp b/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp index d1841198d..bec20c21a 100644 --- a/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp +++ b/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp @@ -30,7 +30,6 @@ #include "video_core/renderer_vulkan/vk_compute_pipeline.h" #include "video_core/renderer_vulkan/vk_descriptor_pool.h" #include "video_core/renderer_vulkan/vk_pipeline_cache.h" -#include "video_core/renderer_vulkan/vk_rasterizer.h" #include "video_core/renderer_vulkan/vk_scheduler.h" #include "video_core/renderer_vulkan/vk_shader_util.h" #include "video_core/renderer_vulkan/vk_update_descriptor.h" @@ -299,12 +298,12 @@ bool GraphicsPipelineCacheKey::operator==(const GraphicsPipelineCacheKey& rhs) c return std::memcmp(&rhs, this, Size()) == 0; } -PipelineCache::PipelineCache(RasterizerVulkan& rasterizer_, const Device& device_, +PipelineCache::PipelineCache(Tegra::MaxwellDeviceMemoryManager& device_memory_, const Device& device_, Scheduler& scheduler_, DescriptorPool& descriptor_pool_, GuestDescriptorQueue& guest_descriptor_queue_, RenderPassCache& render_pass_cache_, BufferCache& buffer_cache_, TextureCache& texture_cache_, VideoCore::ShaderNotify& shader_notify_) - : VideoCommon::ShaderCache{rasterizer_}, device{device_}, scheduler{scheduler_}, + : VideoCommon::ShaderCache{device_memory_}, device{device_}, scheduler{scheduler_}, descriptor_pool{descriptor_pool_}, guest_descriptor_queue{guest_descriptor_queue_}, render_pass_cache{render_pass_cache_}, buffer_cache{buffer_cache_}, texture_cache{texture_cache_}, shader_notify{shader_notify_}, diff --git a/src/video_core/renderer_vulkan/vk_pipeline_cache.h b/src/video_core/renderer_vulkan/vk_pipeline_cache.h index e323ea0fd..354fdc8ed 100644 --- a/src/video_core/renderer_vulkan/vk_pipeline_cache.h +++ b/src/video_core/renderer_vulkan/vk_pipeline_cache.h @@ -26,6 +26,7 @@ #include "video_core/renderer_vulkan/vk_graphics_pipeline.h" #include "video_core/renderer_vulkan/vk_texture_cache.h" #include "video_core/shader_cache.h" +#include "video_core/host1x/gpu_device_memory_manager.h" namespace Core { class System; @@ -79,7 +80,6 @@ class ComputePipeline; class DescriptorPool; class Device; class PipelineStatistics; -class RasterizerVulkan; class RenderPassCache; class Scheduler; @@ -99,7 +99,7 @@ struct ShaderPools { class PipelineCache : public VideoCommon::ShaderCache { public: - explicit PipelineCache(RasterizerVulkan& rasterizer, const Device& device, Scheduler& scheduler, + explicit PipelineCache(Tegra::MaxwellDeviceMemoryManager& device_memory_, const Device& device, Scheduler& scheduler, DescriptorPool& descriptor_pool, GuestDescriptorQueue& guest_descriptor_queue, RenderPassCache& render_pass_cache, BufferCache& buffer_cache, diff --git a/src/video_core/renderer_vulkan/vk_query_cache.cpp b/src/video_core/renderer_vulkan/vk_query_cache.cpp index ad4caf688..d59fe698c 100644 --- a/src/video_core/renderer_vulkan/vk_query_cache.cpp +++ b/src/video_core/renderer_vulkan/vk_query_cache.cpp @@ -14,7 +14,9 @@ #include "common/bit_util.h" #include "common/common_types.h" #include "core/memory.h" +#include "video_core/rasterizer_interface.h" #include "video_core/engines/draw_manager.h" +#include "video_core/host1x/gpu_device_memory_manager.h" #include "video_core/query_cache/query_cache.h" #include "video_core/renderer_vulkan/vk_buffer_cache.h" #include "video_core/renderer_vulkan/vk_compute_pass.h" @@ -102,7 +104,7 @@ private: using BaseStreamer = VideoCommon::SimpleStreamer; struct HostSyncValues { - VAddr address; + DAddr address; size_t size; size_t offset; @@ -317,7 +319,7 @@ public: pending_sync.clear(); } - size_t WriteCounter(VAddr address, bool has_timestamp, u32 value, + size_t WriteCounter(DAddr address, bool has_timestamp, u32 value, [[maybe_unused]] std::optional subreport) override { PauseCounter(); auto index = BuildQuery(); @@ -738,7 +740,7 @@ public: pending_sync.clear(); } - size_t WriteCounter(VAddr address, bool has_timestamp, u32 value, + size_t WriteCounter(DAddr address, bool has_timestamp, u32 value, std::optional subreport_) override { auto index = BuildQuery(); auto* new_query = GetQuery(index); @@ -769,9 +771,9 @@ public: return index; } - std::optional> GetLastQueryStream(size_t stream) { + std::optional> GetLastQueryStream(size_t stream) { if (last_queries[stream] != 0) { - std::pair result(last_queries[stream], last_queries_stride[stream]); + std::pair result(last_queries[stream], last_queries_stride[stream]); return result; } return std::nullopt; @@ -974,7 +976,7 @@ private: size_t buffers_count{}; std::array counter_buffers{}; std::array offsets{}; - std::array last_queries; + std::array last_queries; std::array last_queries_stride; Maxwell3D::Regs::PrimitiveTopology out_topology; u64 streams_mask; @@ -987,7 +989,7 @@ public: : VideoCommon::QueryBase(0, VideoCommon::QueryFlagBits::IsHostManaged, 0) {} // Parameterized constructor - PrimitivesQueryBase(bool has_timestamp, VAddr address) + PrimitivesQueryBase(bool has_timestamp, DAddr address) : VideoCommon::QueryBase(address, VideoCommon::QueryFlagBits::IsHostManaged, 0) { if (has_timestamp) { flags |= VideoCommon::QueryFlagBits::HasTimestamp; @@ -995,7 +997,7 @@ public: } u64 stride{}; - VAddr dependant_address{}; + DAddr dependant_address{}; Maxwell3D::Regs::PrimitiveTopology topology{Maxwell3D::Regs::PrimitiveTopology::Points}; size_t dependant_index{}; bool dependant_manage{}; @@ -1005,15 +1007,15 @@ class PrimitivesSucceededStreamer : public VideoCommon::SimpleStreamer(id_), runtime{runtime_}, - tfb_streamer{tfb_streamer_}, cpu_memory{cpu_memory_} { + tfb_streamer{tfb_streamer_}, device_memory{device_memory_} { MakeDependent(&tfb_streamer); } ~PrimitivesSucceededStreamer() = default; - size_t WriteCounter(VAddr address, bool has_timestamp, u32 value, + size_t WriteCounter(DAddr address, bool has_timestamp, u32 value, std::optional subreport_) override { auto index = BuildQuery(); auto* new_query = GetQuery(index); @@ -1063,6 +1065,8 @@ public: } }); } + auto* ptr = device_memory.GetPointer(new_query->dependant_address); + ASSERT(ptr != nullptr); new_query->dependant_manage = must_manage_dependance; pending_flush_queries.push_back(index); @@ -1100,7 +1104,7 @@ public: num_vertices = dependant_query->value / query->stride; tfb_streamer.Free(query->dependant_index); } else { - u8* pointer = cpu_memory.GetPointer(query->dependant_address); + u8* pointer = device_memory.GetPointer(query->dependant_address); u32 result; std::memcpy(&result, pointer, sizeof(u32)); num_vertices = static_cast(result) / query->stride; @@ -1137,7 +1141,7 @@ public: private: QueryCacheRuntime& runtime; TFBCounterStreamer& tfb_streamer; - Core::Memory::Memory& cpu_memory; + Tegra::MaxwellDeviceMemoryManager& device_memory; // syncing queue std::vector pending_sync; @@ -1152,12 +1156,12 @@ private: struct QueryCacheRuntimeImpl { QueryCacheRuntimeImpl(QueryCacheRuntime& runtime, VideoCore::RasterizerInterface* rasterizer_, - Core::Memory::Memory& cpu_memory_, Vulkan::BufferCache& buffer_cache_, + Tegra::MaxwellDeviceMemoryManager& device_memory_, Vulkan::BufferCache& buffer_cache_, const Device& device_, const MemoryAllocator& memory_allocator_, Scheduler& scheduler_, StagingBufferPool& staging_pool_, ComputePassDescriptorQueue& compute_pass_descriptor_queue, DescriptorPool& descriptor_pool) - : rasterizer{rasterizer_}, cpu_memory{cpu_memory_}, + : rasterizer{rasterizer_}, device_memory{device_memory_}, buffer_cache{buffer_cache_}, device{device_}, memory_allocator{memory_allocator_}, scheduler{scheduler_}, staging_pool{staging_pool_}, guest_streamer(0, runtime), @@ -1168,7 +1172,7 @@ struct QueryCacheRuntimeImpl { scheduler, memory_allocator, staging_pool), primitives_succeeded_streamer( static_cast(QueryType::StreamingPrimitivesSucceeded), runtime, tfb_streamer, - cpu_memory_), + device_memory_), primitives_needed_minus_succeeded_streamer( static_cast(QueryType::StreamingPrimitivesNeededMinusSucceeded), runtime, 0u), hcr_setup{}, hcr_is_set{}, is_hcr_running{}, maxwell3d{} { @@ -1195,7 +1199,7 @@ struct QueryCacheRuntimeImpl { } VideoCore::RasterizerInterface* rasterizer; - Core::Memory::Memory& cpu_memory; + Tegra::MaxwellDeviceMemoryManager& device_memory; Vulkan::BufferCache& buffer_cache; const Device& device; @@ -1210,7 +1214,7 @@ struct QueryCacheRuntimeImpl { PrimitivesSucceededStreamer primitives_succeeded_streamer; VideoCommon::StubStreamer primitives_needed_minus_succeeded_streamer; - std::vector> little_cache; + std::vector> little_cache; std::vector> buffers_to_upload_to; std::vector redirect_cache; std::vector> copies_setup; @@ -1229,14 +1233,14 @@ struct QueryCacheRuntimeImpl { }; QueryCacheRuntime::QueryCacheRuntime(VideoCore::RasterizerInterface* rasterizer, - Core::Memory::Memory& cpu_memory_, + Tegra::MaxwellDeviceMemoryManager& device_memory_, Vulkan::BufferCache& buffer_cache_, const Device& device_, const MemoryAllocator& memory_allocator_, Scheduler& scheduler_, StagingBufferPool& staging_pool_, ComputePassDescriptorQueue& compute_pass_descriptor_queue, DescriptorPool& descriptor_pool) { impl = std::make_unique( - *this, rasterizer, cpu_memory_, buffer_cache_, device_, memory_allocator_, scheduler_, + *this, rasterizer, device_memory_, buffer_cache_, device_, memory_allocator_, scheduler_, staging_pool_, compute_pass_descriptor_queue, descriptor_pool); } @@ -1309,7 +1313,7 @@ void QueryCacheRuntime::HostConditionalRenderingCompareValueImpl(VideoCommon::Lo ResumeHostConditionalRendering(); } -void QueryCacheRuntime::HostConditionalRenderingCompareBCImpl(VAddr address, bool is_equal) { +void QueryCacheRuntime::HostConditionalRenderingCompareBCImpl(DAddr address, bool is_equal) { VkBuffer to_resolve; u32 to_resolve_offset; { @@ -1350,11 +1354,11 @@ bool QueryCacheRuntime::HostConditionalRenderingCompareValues(VideoCommon::Looku return false; } - const auto check_in_bc = [&](VAddr address) { + const auto check_in_bc = [&](DAddr address) { return impl->buffer_cache.IsRegionGpuModified(address, 8); }; - const auto check_value = [&](VAddr address) { - u8* ptr = impl->cpu_memory.GetPointer(address); + const auto check_value = [&](DAddr address) { + u8* ptr = impl->device_memory.GetPointer(address); u64 value{}; std::memcpy(&value, ptr, sizeof(value)); return value == 0; @@ -1477,8 +1481,8 @@ void QueryCacheRuntime::SyncValues(std::span values, VkBuffer ba for (auto& sync_val : values) { total_size += sync_val.size; bool found = false; - VAddr base = Common::AlignDown(sync_val.address, Core::Memory::YUZU_PAGESIZE); - VAddr base_end = base + Core::Memory::YUZU_PAGESIZE; + DAddr base = Common::AlignDown(sync_val.address, Core::Memory::YUZU_PAGESIZE); + DAddr base_end = base + Core::Memory::YUZU_PAGESIZE; for (size_t i = 0; i < impl->little_cache.size(); i++) { const auto set_found = [&] { impl->redirect_cache.push_back(i); diff --git a/src/video_core/renderer_vulkan/vk_query_cache.h b/src/video_core/renderer_vulkan/vk_query_cache.h index e9a1ea169..f6151123e 100644 --- a/src/video_core/renderer_vulkan/vk_query_cache.h +++ b/src/video_core/renderer_vulkan/vk_query_cache.h @@ -27,7 +27,7 @@ struct QueryCacheRuntimeImpl; class QueryCacheRuntime { public: explicit QueryCacheRuntime(VideoCore::RasterizerInterface* rasterizer, - Core::Memory::Memory& cpu_memory_, + Tegra::MaxwellDeviceMemoryManager& device_memory_, Vulkan::BufferCache& buffer_cache_, const Device& device_, const MemoryAllocator& memory_allocator_, Scheduler& scheduler_, StagingBufferPool& staging_pool_, @@ -61,7 +61,7 @@ public: private: void HostConditionalRenderingCompareValueImpl(VideoCommon::LookupData object, bool is_equal); - void HostConditionalRenderingCompareBCImpl(VAddr address, bool is_equal); + void HostConditionalRenderingCompareBCImpl(DAddr address, bool is_equal); friend struct QueryCacheRuntimeImpl; std::unique_ptr impl; }; diff --git a/src/video_core/renderer_vulkan/vk_rasterizer.cpp b/src/video_core/renderer_vulkan/vk_rasterizer.cpp index 241fc34be..efcc349a0 100644 --- a/src/video_core/renderer_vulkan/vk_rasterizer.cpp +++ b/src/video_core/renderer_vulkan/vk_rasterizer.cpp @@ -18,6 +18,7 @@ #include "video_core/engines/draw_manager.h" #include "video_core/engines/kepler_compute.h" #include "video_core/engines/maxwell_3d.h" +#include "video_core/host1x/gpu_device_memory_manager.h" #include "video_core/renderer_vulkan/blit_image.h" #include "video_core/renderer_vulkan/fixed_pipeline_state.h" #include "video_core/renderer_vulkan/maxwell_to_vk.h" @@ -37,6 +38,7 @@ #include "video_core/vulkan_common/vulkan_device.h" #include "video_core/vulkan_common/vulkan_wrapper.h" + namespace Vulkan { using Maxwell = Tegra::Engines::Maxwell3D::Regs; @@ -163,10 +165,11 @@ DrawParams MakeDrawParams(const MaxwellDrawState& draw_state, u32 num_instances, } // Anonymous namespace RasterizerVulkan::RasterizerVulkan(Core::Frontend::EmuWindow& emu_window_, Tegra::GPU& gpu_, - Core::Memory::Memory& cpu_memory_, ScreenInfo& screen_info_, - const Device& device_, MemoryAllocator& memory_allocator_, - StateTracker& state_tracker_, Scheduler& scheduler_) - : RasterizerAccelerated{cpu_memory_}, gpu{gpu_}, screen_info{screen_info_}, device{device_}, + Tegra::MaxwellDeviceMemoryManager& device_memory_, + ScreenInfo& screen_info_, const Device& device_, + MemoryAllocator& memory_allocator_, StateTracker& state_tracker_, + Scheduler& scheduler_) + : gpu{gpu_}, device_memory{device_memory_}, screen_info{screen_info_}, device{device_}, memory_allocator{memory_allocator_}, state_tracker{state_tracker_}, scheduler{scheduler_}, staging_pool(device, memory_allocator, scheduler), descriptor_pool(device, scheduler), guest_descriptor_queue(device, scheduler), compute_pass_descriptor_queue(device, scheduler), @@ -174,14 +177,14 @@ RasterizerVulkan::RasterizerVulkan(Core::Frontend::EmuWindow& emu_window_, Tegra texture_cache_runtime{ device, scheduler, memory_allocator, staging_pool, blit_image, render_pass_cache, descriptor_pool, compute_pass_descriptor_queue}, - texture_cache(texture_cache_runtime, *this), + texture_cache(texture_cache_runtime, device_memory), buffer_cache_runtime(device, memory_allocator, scheduler, staging_pool, guest_descriptor_queue, compute_pass_descriptor_queue, descriptor_pool), - buffer_cache(*this, cpu_memory_, buffer_cache_runtime), - query_cache_runtime(this, cpu_memory_, buffer_cache, device, memory_allocator, scheduler, + buffer_cache(device_memory, buffer_cache_runtime), + query_cache_runtime(this, device_memory, buffer_cache, device, memory_allocator, scheduler, staging_pool, compute_pass_descriptor_queue, descriptor_pool), - query_cache(gpu, *this, cpu_memory_, query_cache_runtime), - pipeline_cache(*this, device, scheduler, descriptor_pool, guest_descriptor_queue, + query_cache(gpu, *this, device_memory, query_cache_runtime), + pipeline_cache(device_memory, device, scheduler, descriptor_pool, guest_descriptor_queue, render_pass_cache, buffer_cache, texture_cache, gpu.ShaderNotify()), accelerate_dma(buffer_cache, texture_cache, scheduler), fence_manager(*this, gpu, texture_cache, buffer_cache, query_cache, device, scheduler), @@ -508,7 +511,7 @@ void Vulkan::RasterizerVulkan::DisableGraphicsUniformBuffer(size_t stage, u32 in void RasterizerVulkan::FlushAll() {} -void RasterizerVulkan::FlushRegion(VAddr addr, u64 size, VideoCommon::CacheType which) { +void RasterizerVulkan::FlushRegion(DAddr addr, u64 size, VideoCommon::CacheType which) { if (addr == 0 || size == 0) { return; } @@ -525,7 +528,7 @@ void RasterizerVulkan::FlushRegion(VAddr addr, u64 size, VideoCommon::CacheType } } -bool RasterizerVulkan::MustFlushRegion(VAddr addr, u64 size, VideoCommon::CacheType which) { +bool RasterizerVulkan::MustFlushRegion(DAddr addr, u64 size, VideoCommon::CacheType which) { if ((True(which & VideoCommon::CacheType::BufferCache))) { std::scoped_lock lock{buffer_cache.mutex}; if (buffer_cache.IsRegionGpuModified(addr, size)) { @@ -542,7 +545,7 @@ bool RasterizerVulkan::MustFlushRegion(VAddr addr, u64 size, VideoCommon::CacheT return false; } -VideoCore::RasterizerDownloadArea RasterizerVulkan::GetFlushArea(VAddr addr, u64 size) { +VideoCore::RasterizerDownloadArea RasterizerVulkan::GetFlushArea(DAddr addr, u64 size) { { std::scoped_lock lock{texture_cache.mutex}; auto area = texture_cache.GetFlushArea(addr, size); @@ -558,7 +561,7 @@ VideoCore::RasterizerDownloadArea RasterizerVulkan::GetFlushArea(VAddr addr, u64 return new_area; } -void RasterizerVulkan::InvalidateRegion(VAddr addr, u64 size, VideoCommon::CacheType which) { +void RasterizerVulkan::InvalidateRegion(DAddr addr, u64 size, VideoCommon::CacheType which) { if (addr == 0 || size == 0) { return; } @@ -578,7 +581,7 @@ void RasterizerVulkan::InvalidateRegion(VAddr addr, u64 size, VideoCommon::Cache } } -void RasterizerVulkan::InnerInvalidation(std::span> sequences) { +void RasterizerVulkan::InnerInvalidation(std::span> sequences) { { std::scoped_lock lock{texture_cache.mutex}; for (const auto& [addr, size] : sequences) { @@ -599,7 +602,8 @@ void RasterizerVulkan::InnerInvalidation(std::span { public: explicit RasterizerVulkan(Core::Frontend::EmuWindow& emu_window_, Tegra::GPU& gpu_, - Core::Memory::Memory& cpu_memory_, ScreenInfo& screen_info_, - const Device& device_, MemoryAllocator& memory_allocator_, - StateTracker& state_tracker_, Scheduler& scheduler_); + Tegra::MaxwellDeviceMemoryManager& device_memory_, + ScreenInfo& screen_info_, const Device& device_, + MemoryAllocator& memory_allocator_, StateTracker& state_tracker_, + Scheduler& scheduler_); ~RasterizerVulkan() override; void Draw(bool is_indexed, u32 instance_count) override; @@ -90,18 +95,18 @@ public: void BindGraphicsUniformBuffer(size_t stage, u32 index, GPUVAddr gpu_addr, u32 size) override; void DisableGraphicsUniformBuffer(size_t stage, u32 index) override; void FlushAll() override; - void FlushRegion(VAddr addr, u64 size, + void FlushRegion(DAddr addr, u64 size, VideoCommon::CacheType which = VideoCommon::CacheType::All) override; - bool MustFlushRegion(VAddr addr, u64 size, + bool MustFlushRegion(DAddr addr, u64 size, VideoCommon::CacheType which = VideoCommon::CacheType::All) override; - VideoCore::RasterizerDownloadArea GetFlushArea(VAddr addr, u64 size) override; - void InvalidateRegion(VAddr addr, u64 size, + VideoCore::RasterizerDownloadArea GetFlushArea(DAddr addr, u64 size) override; + void InvalidateRegion(DAddr addr, u64 size, VideoCommon::CacheType which = VideoCommon::CacheType::All) override; - void InnerInvalidation(std::span> sequences) override; - void OnCacheInvalidation(VAddr addr, u64 size) override; - bool OnCPUWrite(VAddr addr, u64 size) override; + void InnerInvalidation(std::span> sequences) override; + void OnCacheInvalidation(DAddr addr, u64 size) override; + bool OnCPUWrite(DAddr addr, u64 size) override; void InvalidateGPUCache() override; - void UnmapMemory(VAddr addr, u64 size) override; + void UnmapMemory(DAddr addr, u64 size) override; void ModifyGPUMemory(size_t as_id, GPUVAddr addr, u64 size) override; void SignalFence(std::function&& func) override; void SyncOperation(std::function&& func) override; @@ -109,7 +114,7 @@ public: void SignalReference() override; void ReleaseFences(bool force = true) override; void FlushAndInvalidateRegion( - VAddr addr, u64 size, VideoCommon::CacheType which = VideoCommon::CacheType::All) override; + DAddr addr, u64 size, VideoCommon::CacheType which = VideoCommon::CacheType::All) override; void WaitForIdle() override; void FragmentBarrier() override; void TiledCacheBarrier() override; @@ -122,7 +127,7 @@ public: Tegra::Engines::AccelerateDMAInterface& AccessAccelerateDMA() override; void AccelerateInlineToMemory(GPUVAddr address, size_t copy_size, std::span memory) override; - bool AccelerateDisplay(const Tegra::FramebufferConfig& config, VAddr framebuffer_addr, + bool AccelerateDisplay(const Tegra::FramebufferConfig& config, DAddr framebuffer_addr, u32 pixel_stride) override; void LoadDiskResources(u64 title_id, std::stop_token stop_loading, const VideoCore::DiskResourceLoadCallback& callback) override; @@ -176,6 +181,7 @@ private: void UpdateVertexInput(Tegra::Engines::Maxwell3D::Regs& regs); Tegra::GPU& gpu; + Tegra::MaxwellDeviceMemoryManager& device_memory; ScreenInfo& screen_info; const Device& device; -- cgit v1.2.3 From 34a8d0cc8e04b4b9d8e5a75e552f0adb31b5d718 Mon Sep 17 00:00:00 2001 From: Fernando Sahmkow Date: Fri, 29 Dec 2023 07:53:52 +0100 Subject: SMMU: Implement physical memory mirroring --- src/core/device_memory_manager.h | 27 +++- src/core/device_memory_manager.inc | 154 ++++++++++++++++++++++- src/core/hle/service/nvdrv/core/container.cpp | 6 +- src/core/hle/service/nvdrv/core/nvmap.cpp | 7 +- src/core/hle/service/nvdrv/core/nvmap.h | 6 +- src/core/memory.cpp | 53 ++++---- src/video_core/renderer_opengl/gl_rasterizer.cpp | 7 +- src/video_core/renderer_vulkan/vk_rasterizer.cpp | 6 +- 8 files changed, 226 insertions(+), 40 deletions(-) (limited to 'src/video_core/renderer_vulkan') diff --git a/src/core/device_memory_manager.h b/src/core/device_memory_manager.h index 1a63cbd09..7c7726348 100644 --- a/src/core/device_memory_manager.h +++ b/src/core/device_memory_manager.h @@ -10,8 +10,10 @@ #include #include "common/common_types.h" +#include "common/scratch_buffer.h" #include "common/virtual_buffer.h" + namespace Core { class DeviceMemory; @@ -49,9 +51,25 @@ public: template const T* GetPointer(DAddr address) const; - DAddr GetAddressFromPAddr(PAddr address) const { + template + void ApplyOpOnPAddr(PAddr address, Common::ScratchBuffer& buffer, Func&& operation) { DAddr subbits = static_cast(address & page_mask); - return (static_cast(compressed_device_addr[(address >> page_bits)]) << page_bits) + subbits; + const u32 base = compressed_device_addr[(address >> page_bits)]; + if ((base >> MULTI_FLAG_BITS) == 0) [[likely]] { + const DAddr d_address = static_cast(base << page_bits) + subbits; + operation(d_address); + return; + } + InnerGatherDeviceAddresses(buffer, address); + for (u32 value : buffer) { + operation(static_cast(value << page_bits) + subbits); + } + } + + template + void ApplyOpOnPointer(const u8* p, Common::ScratchBuffer& buffer, Func&& operation) { + PAddr address = GetRawPhysicalAddr(p); + ApplyOpOnPAddr(address, buffer, operation); } PAddr GetPhysicalRawAddressFromDAddr(DAddr address) const { @@ -98,6 +116,9 @@ private: static constexpr size_t page_size = 1ULL << page_bits; static constexpr size_t page_mask = page_size - 1ULL; static constexpr u32 physical_address_base = 1U << page_bits; + static constexpr u32 MULTI_FLAG_BITS = 31; + static constexpr u32 MULTI_FLAG = 1U << MULTI_FLAG_BITS; + static constexpr u32 MULTI_MASK = ~MULTI_FLAG; template T* GetPointerFromRaw(PAddr addr) { @@ -117,6 +138,8 @@ private: void WalkBlock(const DAddr addr, const std::size_t size, auto on_unmapped, auto on_memory, auto increment); + void InnerGatherDeviceAddresses(Common::ScratchBuffer& buffer, PAddr address); + std::unique_ptr> impl; const uintptr_t physical_base; diff --git a/src/core/device_memory_manager.inc b/src/core/device_memory_manager.inc index 8c5f82d31..4fb3ad3ab 100644 --- a/src/core/device_memory_manager.inc +++ b/src/core/device_memory_manager.inc @@ -18,10 +18,117 @@ namespace Core { +namespace { + +class PhysicalAddressContainer { +public: + PhysicalAddressContainer() = default; + ~PhysicalAddressContainer() = default; + + void GatherValues(u32 start_entry, Common::ScratchBuffer& buffer) { + buffer.resize(8); + buffer.resize(0); + size_t index = 0; + const auto add_value = [&](u32 value) { + buffer[index] = value; + index++; + buffer.resize(index); + }; + + u32 iter_entry = start_entry; + Entry* current = &storage[iter_entry - 1]; + add_value(current->value); + while (current->next_entry != 0) { + iter_entry = current->next_entry; + current = &storage[iter_entry - 1]; + add_value(current->value); + } + } + + u32 Register(u32 value) { + return RegisterImplementation(value); + } + + void Register(u32 value, u32 start_entry) { + auto entry_id = RegisterImplementation(value); + u32 iter_entry = start_entry; + Entry* current = &storage[iter_entry - 1]; + while (current->next_entry != 0) { + iter_entry = current->next_entry; + current = &storage[iter_entry - 1]; + } + current->next_entry = entry_id; + } + + std::pair Unregister(u32 value, u32 start_entry) { + u32 iter_entry = start_entry; + Entry* previous{}; + Entry* current = &storage[iter_entry - 1]; + Entry* next{}; + bool more_than_one_remaining = false; + u32 result_start{start_entry}; + size_t count = 0; + while (current->value != value) { + count++; + previous = current; + iter_entry = current->next_entry; + current = &storage[iter_entry - 1]; + } + // Find next + u32 next_entry = current->next_entry; + if (next_entry != 0) { + next = &storage[next_entry - 1]; + more_than_one_remaining = next->next_entry != 0; + } + if (previous) { + previous->next_entry = next_entry; + } else { + result_start = next_entry; + } + free_entries.emplace_back(iter_entry); + return std::make_pair(more_than_one_remaining || count > 1, result_start); + } + + u32 ReleaseEntry(u32 start_entry) { + Entry* current = &storage[start_entry - 1]; + free_entries.emplace_back(start_entry); + return current->value; + } + +private: + u32 RegisterImplementation(u32 value) { + auto entry_id = GetNewEntry(); + auto& entry = storage[entry_id - 1]; + entry.next_entry = 0; + entry.value = value; + return entry_id; + } + u32 GetNewEntry() { + if (!free_entries.empty()) { + u32 result = free_entries.front(); + free_entries.pop_front(); + return result; + } + storage.emplace_back(); + u32 new_entry = static_cast(storage.size()); + return new_entry; + } + + struct Entry { + u32 next_entry{}; + u32 value{}; + }; + + std::deque storage; + std::deque free_entries; +}; + struct EmptyAllocator { EmptyAllocator([[maybe_unused]] DAddr address) {} }; +} // namespace + template struct DeviceMemoryManagerAllocator { static constexpr bool supports_pinning = DTraits::supports_pinning; @@ -38,6 +145,7 @@ struct DeviceMemoryManagerAllocator { std::conditional_t, EmptyAllocator> pin_allocator; Common::FlatAllocator main_allocator; + PhysicalAddressContainer multi_dev_address; /// Returns true when vaddr -> vaddr+size is fully contained in the buffer template @@ -109,6 +217,9 @@ DeviceMemoryManager::DeviceMemoryManager(const DeviceMemory& device_memo cpu_backing_address(device_as_size >> Memory::YUZU_PAGEBITS) { impl = std::make_unique>(); cached_pages = std::make_unique(); + for (size_t i = 0; i < 1ULL << (33 - 12); i++) { + compressed_device_addr[i] = 0; + } } template @@ -155,8 +266,19 @@ void DeviceMemoryManager::Map(DAddr address, VAddr virtual_address, size } auto phys_addr = static_cast(GetRawPhysicalAddr(ptr) >> Memory::YUZU_PAGEBITS) + 1U; compressed_physical_ptr[start_page_d + i] = phys_addr; - compressed_device_addr[phys_addr - 1U] = static_cast(start_page_d + i); InsertCPUBacking(start_page_d + i, new_vaddress, process_id); + const u32 base_dev = compressed_device_addr[phys_addr - 1U]; + const u32 new_dev = static_cast(start_page_d + i); + if (base_dev == 0) [[likely]] { + compressed_device_addr[phys_addr - 1U] = new_dev; + continue; + } + u32 start_id = base_dev & MULTI_MASK; + if ((base_dev >> MULTI_FLAG_BITS) == 0) { + start_id = impl->multi_dev_address.Register(base_dev); + compressed_device_addr[phys_addr - 1U] = MULTI_FLAG | start_id; + } + impl->multi_dev_address.Register(new_dev, start_id); } } @@ -170,12 +292,38 @@ void DeviceMemoryManager::Unmap(DAddr address, size_t size) { auto phys_addr = compressed_physical_ptr[start_page_d + i]; compressed_physical_ptr[start_page_d + i] = 0; cpu_backing_address[start_page_d + i] = 0; - if (phys_addr != 0) { - compressed_device_addr[phys_addr - 1] = 0; + if (phys_addr != 0) [[likely]] { + const u32 base_dev = compressed_device_addr[phys_addr - 1U]; + if ((base_dev >> MULTI_FLAG_BITS) == 0) [[likely]] { + compressed_device_addr[phys_addr - 1] = 0; + continue; + } + const auto [more_entries, new_start] = impl->multi_dev_address.Unregister( + static_cast(start_page_d + i), base_dev & MULTI_MASK); + if (!more_entries) { + compressed_device_addr[phys_addr - 1] = + impl->multi_dev_address.ReleaseEntry(new_start); + continue; + } + compressed_device_addr[phys_addr - 1] = new_start | MULTI_FLAG; } } } +template +void DeviceMemoryManager::InnerGatherDeviceAddresses(Common::ScratchBuffer& buffer, + PAddr address) { + size_t phys_addr = address >> page_bits; + std::scoped_lock lk(mapping_guard); + u32 backing = compressed_device_addr[phys_addr]; + if ((backing >> MULTI_FLAG_BITS) != 0) { + impl->multi_dev_address.GatherValues(backing & MULTI_MASK, buffer); + return; + } + buffer.resize(1); + buffer[0] = backing; +} + template template T* DeviceMemoryManager::GetPointer(DAddr address) { diff --git a/src/core/hle/service/nvdrv/core/container.cpp b/src/core/hle/service/nvdrv/core/container.cpp index 7c2231fe6..e12ce05c1 100644 --- a/src/core/hle/service/nvdrv/core/container.cpp +++ b/src/core/hle/service/nvdrv/core/container.cpp @@ -16,8 +16,8 @@ namespace Service::Nvidia::NvCore { struct ContainerImpl { - explicit ContainerImpl(Tegra::Host1x::Host1x& host1x_) - : host1x{host1x_}, file{host1x_}, manager{host1x_}, device_file_data{} {} + explicit ContainerImpl(Container& core, Tegra::Host1x::Host1x& host1x_) + : host1x{host1x_}, file{core, host1x_}, manager{host1x_}, device_file_data{} {} Tegra::Host1x::Host1x& host1x; NvMap file; SyncpointManager manager; @@ -29,7 +29,7 @@ struct ContainerImpl { }; Container::Container(Tegra::Host1x::Host1x& host1x_) { - impl = std::make_unique(host1x_); + impl = std::make_unique(*this, host1x_); } Container::~Container() = default; diff --git a/src/core/hle/service/nvdrv/core/nvmap.cpp b/src/core/hle/service/nvdrv/core/nvmap.cpp index 7879c6f04..e4168a37c 100644 --- a/src/core/hle/service/nvdrv/core/nvmap.cpp +++ b/src/core/hle/service/nvdrv/core/nvmap.cpp @@ -7,6 +7,7 @@ #include "common/alignment.h" #include "common/assert.h" #include "common/logging/log.h" +#include "core/hle/service/nvdrv/core/container.h" #include "core/hle/service/nvdrv/core/nvmap.h" #include "core/memory.h" #include "video_core/host1x/host1x.h" @@ -64,7 +65,7 @@ NvResult NvMap::Handle::Duplicate(bool internal_session) { return NvResult::Success; } -NvMap::NvMap(Tegra::Host1x::Host1x& host1x_) : host1x{host1x_} {} +NvMap::NvMap(Container& core_, Tegra::Host1x::Host1x& host1x_) : host1x{host1x_}, core{core_} {} void NvMap::AddHandle(std::shared_ptr handle_description) { std::scoped_lock lock(handles_lock); @@ -160,6 +161,8 @@ DAddr NvMap::PinHandle(NvMap::Handle::Id handle, size_t session_id, bool low_are // If not then allocate some space and map it DAddr address{}; auto& smmu = host1x.MemoryManager(); + auto* session = core.GetSession(session_id); + auto allocate = std::bind(&Tegra::MaxwellDeviceMemoryManager::Allocate, &smmu, _1); //: std::bind(&Tegra::MaxwellDeviceMemoryManager::Allocate, &smmu, _1); while ((address = allocate(static_cast(handle_description->aligned_size))) == 0) { @@ -179,7 +182,7 @@ DAddr NvMap::PinHandle(NvMap::Handle::Id handle, size_t session_id, bool low_are handle_description->d_address = address; smmu.Map(address, handle_description->address, handle_description->aligned_size, - session_id); + session->smmu_id); } handle_description->pins++; diff --git a/src/core/hle/service/nvdrv/core/nvmap.h b/src/core/hle/service/nvdrv/core/nvmap.h index e9e9e8b5b..7dd6d26c3 100644 --- a/src/core/hle/service/nvdrv/core/nvmap.h +++ b/src/core/hle/service/nvdrv/core/nvmap.h @@ -25,6 +25,8 @@ class Host1x; } // namespace Tegra namespace Service::Nvidia::NvCore { + +class Container; /** * @brief The nvmap core class holds the global state for nvmap and provides methods to manage * handles @@ -109,7 +111,7 @@ public: bool can_unlock; //!< If the address region is ready to be unlocked }; - explicit NvMap(Tegra::Host1x::Host1x& host1x); + explicit NvMap(Container& core, Tegra::Host1x::Host1x& host1x); /** * @brief Creates an unallocated handle of the given size @@ -173,5 +175,7 @@ private: * @return If the handle was removed from the map */ bool TryRemoveHandle(const Handle& handle_description); + + Container& core; }; } // namespace Service::Nvidia::NvCore diff --git a/src/core/memory.cpp b/src/core/memory.cpp index 609e775ae..f126840cb 100644 --- a/src/core/memory.cpp +++ b/src/core/memory.cpp @@ -44,7 +44,8 @@ bool AddressSpaceContains(const Common::PageTable& table, const Common::ProcessA // from outside classes. This also allows modification to the internals of the memory // subsystem without needing to rebuild all files that make use of the memory interface. struct Memory::Impl { - explicit Impl(Core::System& system_) : system{system_} {} + explicit Impl(Core::System& system_) + : system{system_} {} void SetCurrentPageTable(Kernel::KProcess& process) { current_page_table = &process.GetPageTable().GetImpl(); @@ -817,26 +818,31 @@ struct Memory::Impl { void HandleRasterizerDownload(VAddr v_address, size_t size) { const auto* p = GetPointerImpl( v_address, []() {}, []() {}); - auto& gpu_device_memory = system.Host1x().MemoryManager(); - DAddr address = - gpu_device_memory.GetAddressFromPAddr(system.DeviceMemory().GetRawPhysicalAddr(p)); + if (!gpu_device_memory) [[unlikely]] { + gpu_device_memory = &system.Host1x().MemoryManager(); + } const size_t core = system.GetCurrentHostThreadID(); auto& current_area = rasterizer_read_areas[core]; - const DAddr end_address = address + size; - if (current_area.start_address <= address && end_address <= current_area.end_address) - [[likely]] { - return; - } - current_area = system.GPU().OnCPURead(address, size); + gpu_device_memory->ApplyOpOnPointer( + p, scratch_buffers[core], [&](DAddr address) { + const DAddr end_address = address + size; + if (current_area.start_address <= address && end_address <= current_area.end_address) + [[likely]] { + return; + } + current_area = system.GPU().OnCPURead(address, size); + }); } void HandleRasterizerWrite(VAddr v_address, size_t size) { const auto* p = GetPointerImpl( v_address, []() {}, []() {}); - PAddr address = system.DeviceMemory().GetRawPhysicalAddr(p); constexpr size_t sys_core = Core::Hardware::NUM_CPU_CORES - 1; const size_t core = std::min(system.GetCurrentHostThreadID(), sys_core); // any other calls threads go to syscore. + if (!gpu_device_memory) [[unlikely]] { + gpu_device_memory = &system.Host1x().MemoryManager(); + } // Guard on sys_core; if (core == sys_core) [[unlikely]] { sys_core_guard.lock(); @@ -846,17 +852,20 @@ struct Memory::Impl { sys_core_guard.unlock(); } }); - auto& current_area = rasterizer_write_areas[core]; - PAddr subaddress = address >> YUZU_PAGEBITS; - bool do_collection = current_area.last_address == subaddress; - if (!do_collection) [[unlikely]] { - do_collection = system.GPU().OnCPUWrite(address, size); - if (!do_collection) { - return; + gpu_device_memory->ApplyOpOnPointer( + p, scratch_buffers[core], [&](DAddr address) { + auto& current_area = rasterizer_write_areas[core]; + PAddr subaddress = address >> YUZU_PAGEBITS; + bool do_collection = current_area.last_address == subaddress; + if (!do_collection) [[unlikely]] { + do_collection = system.GPU().OnCPUWrite(address, size); + if (!do_collection) { + return; + } + current_area.last_address = subaddress; } - current_area.last_address = subaddress; - } - gpu_dirty_managers[core].Collect(address, size); + gpu_dirty_managers[core].Collect(address, size); + }); } struct GPUDirtyState { @@ -872,10 +881,12 @@ struct Memory::Impl { } Core::System& system; + Tegra::MaxwellDeviceMemoryManager* gpu_device_memory{}; Common::PageTable* current_page_table = nullptr; std::array rasterizer_read_areas{}; std::array rasterizer_write_areas{}; + std::array, Core::Hardware::NUM_CPU_CORES> scratch_buffers{}; std::span gpu_dirty_managers; std::mutex sys_core_guard; diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp index ca31e2fbd..71b748c74 100644 --- a/src/video_core/renderer_opengl/gl_rasterizer.cpp +++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp @@ -554,9 +554,8 @@ void RasterizerOpenGL::InvalidateRegion(DAddr addr, u64 size, VideoCommon::Cache } } -bool RasterizerOpenGL::OnCPUWrite(PAddr p_addr, u64 size) { +bool RasterizerOpenGL::OnCPUWrite(DAddr addr, u64 size) { MICROPROFILE_SCOPE(OpenGL_CacheManagement); - const DAddr addr = device_memory.GetAddressFromPAddr(p_addr); if (addr == 0 || size == 0) { return false; } @@ -577,9 +576,9 @@ bool RasterizerOpenGL::OnCPUWrite(PAddr p_addr, u64 size) { return false; } -void RasterizerOpenGL::OnCacheInvalidation(PAddr p_addr, u64 size) { +void RasterizerOpenGL::OnCacheInvalidation(DAddr addr, u64 size) { MICROPROFILE_SCOPE(OpenGL_CacheManagement); - const DAddr addr = device_memory.GetAddressFromPAddr(p_addr); + if (addr == 0 || size == 0) { return; } diff --git a/src/video_core/renderer_vulkan/vk_rasterizer.cpp b/src/video_core/renderer_vulkan/vk_rasterizer.cpp index efcc349a0..7db131985 100644 --- a/src/video_core/renderer_vulkan/vk_rasterizer.cpp +++ b/src/video_core/renderer_vulkan/vk_rasterizer.cpp @@ -602,8 +602,7 @@ void RasterizerVulkan::InnerInvalidation(std::span Date: Sun, 31 Dec 2023 20:55:15 +0100 Subject: Core: Clang format and other small issues. --- src/audio_core/device/device_session.cpp | 2 +- .../renderer/command/data_source/decode.cpp | 2 +- src/core/device_memory_manager.h | 7 +-- src/core/device_memory_manager.inc | 50 +++++++++++++++------- src/core/guest_memory.h | 5 ++- src/core/hle/service/hle_ipc.cpp | 2 +- src/core/hle/service/nvdrv/core/container.cpp | 8 +++- src/core/hle/service/nvdrv/core/container.h | 8 ++++ src/core/hle/service/nvdrv/core/heap_mapper.cpp | 25 ++++++----- src/core/hle/service/nvdrv/core/nvmap.cpp | 8 ++-- src/core/hle/service/nvdrv/core/nvmap.h | 11 ++--- .../service/nvdrv/devices/nvhost_nvdec_common.h | 2 +- src/core/hle/service/nvdrv/devices/nvhost_vic.cpp | 2 +- src/core/hle/service/nvdrv/devices/nvmap.cpp | 4 +- .../service/nvnflinger/fb_share_buffer_manager.cpp | 10 +++-- .../service/nvnflinger/fb_share_buffer_manager.h | 3 +- src/core/memory.h | 2 +- src/video_core/buffer_cache/buffer_cache.h | 8 ++-- src/video_core/buffer_cache/word_manager.h | 4 +- src/video_core/dma_pusher.cpp | 4 +- src/video_core/engines/sw_blitter/blitter.cpp | 2 +- src/video_core/guest_memory.h | 3 +- src/video_core/host1x/codecs/vp9.cpp | 5 +-- src/video_core/host1x/vic.cpp | 7 ++- src/video_core/memory_manager.h | 5 ++- src/video_core/query_cache/query_cache_base.h | 5 ++- src/video_core/renderer_opengl/gl_buffer_cache.h | 3 +- src/video_core/renderer_opengl/gl_query_cache.cpp | 3 +- src/video_core/renderer_opengl/gl_query_cache.h | 5 ++- src/video_core/renderer_opengl/gl_rasterizer.h | 6 +-- src/video_core/renderer_vulkan/renderer_vulkan.cpp | 4 +- src/video_core/renderer_vulkan/renderer_vulkan.h | 5 +-- src/video_core/renderer_vulkan/vk_blit_screen.h | 7 +-- .../renderer_vulkan/vk_pipeline_cache.cpp | 5 ++- src/video_core/renderer_vulkan/vk_pipeline_cache.h | 6 +-- src/video_core/renderer_vulkan/vk_query_cache.cpp | 9 ++-- src/video_core/renderer_vulkan/vk_rasterizer.cpp | 1 - src/video_core/renderer_vulkan/vk_rasterizer.h | 5 +-- src/video_core/shader_cache.cpp | 3 +- src/video_core/shader_cache.h | 2 +- 40 files changed, 152 insertions(+), 106 deletions(-) (limited to 'src/video_core/renderer_vulkan') diff --git a/src/audio_core/device/device_session.cpp b/src/audio_core/device/device_session.cpp index d9fc8c3e0..2a1ae1bb3 100644 --- a/src/audio_core/device/device_session.cpp +++ b/src/audio_core/device/device_session.cpp @@ -8,8 +8,8 @@ #include "audio_core/sink/sink_stream.h" #include "core/core.h" #include "core/core_timing.h" -#include "core/memory.h" #include "core/guest_memory.h" +#include "core/memory.h" #include "core/hle/kernel/k_process.h" diff --git a/src/audio_core/renderer/command/data_source/decode.cpp b/src/audio_core/renderer/command/data_source/decode.cpp index 77a33a87a..905613a5a 100644 --- a/src/audio_core/renderer/command/data_source/decode.cpp +++ b/src/audio_core/renderer/command/data_source/decode.cpp @@ -9,8 +9,8 @@ #include "common/fixed_point.h" #include "common/logging/log.h" #include "common/scratch_buffer.h" -#include "core/memory.h" #include "core/guest_memory.h" +#include "core/memory.h" namespace AudioCore::Renderer { diff --git a/src/core/device_memory_manager.h b/src/core/device_memory_manager.h index 0f6599cfe..f9cb13a7a 100644 --- a/src/core/device_memory_manager.h +++ b/src/core/device_memory_manager.h @@ -27,13 +27,13 @@ struct DeviceMemoryManagerAllocator; template class DeviceMemoryManager { using DeviceInterface = typename Traits::DeviceInterface; - using DeviceMethods = Traits::DeviceMethods; + using DeviceMethods = typename Traits::DeviceMethods; public: DeviceMemoryManager(const DeviceMemory& device_memory); ~DeviceMemoryManager(); - void BindInterface(DeviceInterface* interface); + void BindInterface(DeviceInterface* device_inter); DAddr Allocate(size_t size); void AllocateFixed(DAddr start, size_t size); @@ -111,6 +111,7 @@ public: private: static constexpr size_t device_virtual_bits = Traits::device_virtual_bits; static constexpr size_t device_as_size = 1ULL << device_virtual_bits; + static constexpr size_t physical_min_bits = 32; static constexpr size_t physical_max_bits = 33; static constexpr size_t page_bits = 12; static constexpr size_t page_size = 1ULL << page_bits; @@ -143,7 +144,7 @@ private: std::unique_ptr> impl; const uintptr_t physical_base; - DeviceInterface* interface; + DeviceInterface* device_inter; Common::VirtualBuffer compressed_physical_ptr; Common::VirtualBuffer compressed_device_addr; Common::VirtualBuffer continuity_tracker; diff --git a/src/core/device_memory_manager.inc b/src/core/device_memory_manager.inc index 175f0cd5f..a0eb4214e 100644 --- a/src/core/device_memory_manager.inc +++ b/src/core/device_memory_manager.inc @@ -12,6 +12,7 @@ #include "common/assert.h" #include "common/div_ceil.h" #include "common/scope_exit.h" +#include "common/settings.h" #include "core/device_memory.h" #include "core/device_memory_manager.h" #include "core/memory.h" @@ -162,20 +163,39 @@ struct DeviceMemoryManagerAllocator { template DeviceMemoryManager::DeviceMemoryManager(const DeviceMemory& device_memory_) : physical_base{reinterpret_cast(device_memory_.buffer.BackingBasePointer())}, - interface{nullptr}, compressed_physical_ptr(device_as_size >> Memory::YUZU_PAGEBITS), - compressed_device_addr(1ULL << (physical_max_bits - Memory::YUZU_PAGEBITS)), + device_inter{nullptr}, compressed_physical_ptr(device_as_size >> Memory::YUZU_PAGEBITS), + compressed_device_addr(1ULL << ((Settings::values.memory_layout_mode.GetValue() == + Settings::MemoryLayout::Memory_4Gb + ? physical_min_bits + : physical_max_bits) - + Memory::YUZU_PAGEBITS)), continuity_tracker(device_as_size >> Memory::YUZU_PAGEBITS), cpu_backing_address(device_as_size >> Memory::YUZU_PAGEBITS) { impl = std::make_unique>(); cached_pages = std::make_unique(); + + const size_t total_virtual = device_as_size >> Memory::YUZU_PAGEBITS; + for (size_t i = 0; i < total_virtual; i++) { + compressed_physical_ptr[i] = 0; + continuity_tracker[i] = 1; + cpu_backing_address[i] = 0; + } + const size_t total_phys = 1ULL << ((Settings::values.memory_layout_mode.GetValue() == + Settings::MemoryLayout::Memory_4Gb + ? physical_min_bits + : physical_max_bits) - + Memory::YUZU_PAGEBITS); + for (size_t i = 0; i < total_phys; i++) { + compressed_device_addr[i] = 0; + } } template DeviceMemoryManager::~DeviceMemoryManager() = default; template -void DeviceMemoryManager::BindInterface(DeviceInterface* interface_) { - interface = interface_; +void DeviceMemoryManager::BindInterface(DeviceInterface* device_inter_) { + device_inter = device_inter_; } template @@ -232,7 +252,7 @@ template void DeviceMemoryManager::Unmap(DAddr address, size_t size) { size_t start_page_d = address >> Memory::YUZU_PAGEBITS; size_t num_pages = Common::AlignUp(size, Memory::YUZU_PAGESIZE) >> Memory::YUZU_PAGEBITS; - interface->InvalidateRegion(address, size); + device_inter->InvalidateRegion(address, size); std::scoped_lock lk(mapping_guard); for (size_t i = 0; i < num_pages; i++) { auto phys_addr = compressed_physical_ptr[start_page_d + i]; @@ -392,7 +412,7 @@ void DeviceMemoryManager::WalkBlock(DAddr addr, std::size_t size, auto o template void DeviceMemoryManager::ReadBlock(DAddr address, void* dest_pointer, size_t size) { - interface->FlushRegion(address, size); + device_inter->FlushRegion(address, size); WalkBlock( address, size, [&](size_t copy_amount, DAddr current_vaddr) { @@ -426,7 +446,7 @@ void DeviceMemoryManager::WriteBlock(DAddr address, const void* src_poin [&](const std::size_t copy_amount) { src_pointer = static_cast(src_pointer) + copy_amount; }); - interface->InvalidateRegion(address, size); + device_inter->InvalidateRegion(address, size); } template @@ -468,14 +488,14 @@ void DeviceMemoryManager::WriteBlockUnsafe(DAddr address, const void* sr } template -size_t DeviceMemoryManager::RegisterProcess(Memory::Memory* memory_interface) { +size_t DeviceMemoryManager::RegisterProcess(Memory::Memory* memory_device_inter) { size_t new_id; if (!id_pool.empty()) { new_id = id_pool.front(); id_pool.pop_front(); - registered_processes[new_id] = memory_interface; + registered_processes[new_id] = memory_device_inter; } else { - registered_processes.emplace_back(memory_interface); + registered_processes.emplace_back(memory_device_inter); new_id = registered_processes.size() - 1U; } return new_id; @@ -512,7 +532,7 @@ void DeviceMemoryManager::UpdatePagesCachedCount(DAddr addr, size_t size size_t page = addr >> Memory::YUZU_PAGEBITS; auto [process_id, base_vaddress] = ExtractCPUBacking(page); size_t vpage = base_vaddress >> Memory::YUZU_PAGEBITS; - auto* memory_interface = registered_processes[process_id]; + auto* memory_device_inter = registered_processes[process_id]; for (; page != page_end; ++page) { std::atomic_uint8_t& count = cached_pages->at(page >> 3).Count(page); @@ -536,7 +556,7 @@ void DeviceMemoryManager::UpdatePagesCachedCount(DAddr addr, size_t size uncache_bytes += Memory::YUZU_PAGESIZE; } else if (uncache_bytes > 0) { lock(); - MarkRegionCaching(memory_interface, uncache_begin << Memory::YUZU_PAGEBITS, + MarkRegionCaching(memory_device_inter, uncache_begin << Memory::YUZU_PAGEBITS, uncache_bytes, false); uncache_bytes = 0; } @@ -547,7 +567,7 @@ void DeviceMemoryManager::UpdatePagesCachedCount(DAddr addr, size_t size cache_bytes += Memory::YUZU_PAGESIZE; } else if (cache_bytes > 0) { lock(); - MarkRegionCaching(memory_interface, cache_begin << Memory::YUZU_PAGEBITS, cache_bytes, + MarkRegionCaching(memory_device_inter, cache_begin << Memory::YUZU_PAGEBITS, cache_bytes, true); cache_bytes = 0; } @@ -555,12 +575,12 @@ void DeviceMemoryManager::UpdatePagesCachedCount(DAddr addr, size_t size } if (uncache_bytes > 0) { lock(); - MarkRegionCaching(memory_interface, uncache_begin << Memory::YUZU_PAGEBITS, uncache_bytes, + MarkRegionCaching(memory_device_inter, uncache_begin << Memory::YUZU_PAGEBITS, uncache_bytes, false); } if (cache_bytes > 0) { lock(); - MarkRegionCaching(memory_interface, cache_begin << Memory::YUZU_PAGEBITS, cache_bytes, + MarkRegionCaching(memory_device_inter, cache_begin << Memory::YUZU_PAGEBITS, cache_bytes, true); } } diff --git a/src/core/guest_memory.h b/src/core/guest_memory.h index 0b349cc17..8030df73a 100644 --- a/src/core/guest_memory.h +++ b/src/core/guest_memory.h @@ -202,7 +202,8 @@ public: } else { this->m_memory.WriteBlockUnsafe(this->m_addr, this->data(), this->size_bytes()); } - } else if constexpr ((FLAGS & GuestMemoryFlags::Safe) || (FLAGS & GuestMemoryFlags::Cached)) { + } else if constexpr ((FLAGS & GuestMemoryFlags::Safe) || + (FLAGS & GuestMemoryFlags::Cached)) { this->m_memory.InvalidateRegion(this->m_addr, this->size_bytes()); } } @@ -215,4 +216,4 @@ using CpuGuestMemory = GuestMemory; template using CpuGuestMemoryScoped = GuestMemoryScoped; -} // namespace Tegra::Memory +} // namespace Core::Memory diff --git a/src/core/hle/service/hle_ipc.cpp b/src/core/hle/service/hle_ipc.cpp index 9f6274c7d..e491dd260 100644 --- a/src/core/hle/service/hle_ipc.cpp +++ b/src/core/hle/service/hle_ipc.cpp @@ -12,6 +12,7 @@ #include "common/common_types.h" #include "common/logging/log.h" #include "common/scratch_buffer.h" +#include "core/guest_memory.h" #include "core/hle/kernel/k_auto_object.h" #include "core/hle/kernel/k_handle_table.h" #include "core/hle/kernel/k_process.h" @@ -22,7 +23,6 @@ #include "core/hle/service/hle_ipc.h" #include "core/hle/service/ipc_helpers.h" #include "core/memory.h" -#include "core/guest_memory.h" namespace Service { diff --git a/src/core/hle/service/nvdrv/core/container.cpp b/src/core/hle/service/nvdrv/core/container.cpp index 4d3a9d696..d04b7f5ff 100644 --- a/src/core/hle/service/nvdrv/core/container.cpp +++ b/src/core/hle/service/nvdrv/core/container.cpp @@ -16,6 +16,12 @@ namespace Service::Nvidia::NvCore { +Session::Session(size_t id_, Kernel::KProcess* process_, size_t smmu_id_) + : id{id_}, process{process_}, smmu_id{smmu_id_}, + has_preallocated_area{}, mapper{}, is_active{} {} + +Session::~Session() = default; + struct ContainerImpl { explicit ContainerImpl(Container& core, Tegra::Host1x::Host1x& host1x_) : host1x{host1x_}, file{core, host1x_}, manager{host1x_}, device_file_data{} {} @@ -54,8 +60,8 @@ size_t Container::OpenSession(Kernel::KProcess* process) { impl->id_pool.pop_front(); impl->sessions[new_id] = Session{new_id, process, smmu_id}; } else { - impl->sessions.emplace_back(new_id, process, smmu_id); new_id = impl->new_ids++; + impl->sessions.emplace_back(new_id, process, smmu_id); } auto& session = impl->sessions[new_id]; session.is_active = true; diff --git a/src/core/hle/service/nvdrv/core/container.h b/src/core/hle/service/nvdrv/core/container.h index 86705cbc8..4b8452844 100644 --- a/src/core/hle/service/nvdrv/core/container.h +++ b/src/core/hle/service/nvdrv/core/container.h @@ -27,6 +27,14 @@ class SyncpointManager; struct ContainerImpl; struct Session { + Session(size_t id_, Kernel::KProcess* process_, size_t smmu_id_); + ~Session(); + + Session(const Session&) = delete; + Session& operator=(const Session&) = delete; + Session(Session&&) = default; + Session& operator=(Session&&) = default; + size_t id; Kernel::KProcess* process; size_t smmu_id; diff --git a/src/core/hle/service/nvdrv/core/heap_mapper.cpp b/src/core/hle/service/nvdrv/core/heap_mapper.cpp index 59d993bc6..c29191b92 100644 --- a/src/core/hle/service/nvdrv/core/heap_mapper.cpp +++ b/src/core/hle/service/nvdrv/core/heap_mapper.cpp @@ -124,10 +124,11 @@ DAddr HeapMapper::Map(VAddr start, size_t size) { m_internal->base_set.clear(); const IntervalType interval{start, start + size}; m_internal->base_set.insert(interval); - m_internal->ForEachInOverlapCounter(m_internal->mapping_overlaps, start, size, [this](VAddr start_addr, VAddr end_addr, int){ - const IntervalType other{start_addr, end_addr}; - m_internal->base_set.subtract(other); - }); + m_internal->ForEachInOverlapCounter(m_internal->mapping_overlaps, start, size, + [this](VAddr start_addr, VAddr end_addr, int) { + const IntervalType other{start_addr, end_addr}; + m_internal->base_set.subtract(other); + }); if (!m_internal->base_set.empty()) { auto it = m_internal->base_set.begin(); auto end_it = m_internal->base_set.end(); @@ -136,7 +137,8 @@ DAddr HeapMapper::Map(VAddr start, size_t size) { const VAddr inter_addr = it->lower(); const size_t offset = inter_addr - m_vaddress; const size_t sub_size = inter_addr_end - inter_addr; - m_internal->device_memory.Map(m_daddress + offset, m_vaddress + offset, sub_size, m_smmu_id); + m_internal->device_memory.Map(m_daddress + offset, m_vaddress + offset, sub_size, + m_smmu_id); } } m_internal->mapping_overlaps += std::make_pair(interval, 1); @@ -147,12 +149,13 @@ DAddr HeapMapper::Map(VAddr start, size_t size) { void HeapMapper::Unmap(VAddr start, size_t size) { std::scoped_lock lk(m_internal->guard); m_internal->base_set.clear(); - m_internal->ForEachInOverlapCounter(m_internal->mapping_overlaps, start, size, [this](VAddr start_addr, VAddr end_addr, int value) { - if (value <= 1) { - const IntervalType other{start_addr, end_addr}; - m_internal->base_set.insert(other); - } - }); + m_internal->ForEachInOverlapCounter(m_internal->mapping_overlaps, start, size, + [this](VAddr start_addr, VAddr end_addr, int value) { + if (value <= 1) { + const IntervalType other{start_addr, end_addr}; + m_internal->base_set.insert(other); + } + }); if (!m_internal->base_set.empty()) { auto it = m_internal->base_set.begin(); auto end_it = m_internal->base_set.end(); diff --git a/src/core/hle/service/nvdrv/core/nvmap.cpp b/src/core/hle/service/nvdrv/core/nvmap.cpp index 296b4d8d2..6e59d4fe1 100644 --- a/src/core/hle/service/nvdrv/core/nvmap.cpp +++ b/src/core/hle/service/nvdrv/core/nvmap.cpp @@ -13,8 +13,8 @@ #include "core/memory.h" #include "video_core/host1x/host1x.h" - using Core::Memory::YUZU_PAGESIZE; +constexpr size_t BIG_PAGE_SIZE = YUZU_PAGESIZE * 16; namespace Service::Nvidia::NvCore { NvMap::Handle::Handle(u64 size_, Id id_) @@ -96,8 +96,9 @@ void NvMap::UnmapHandle(Handle& handle_description) { const size_t map_size = handle_description.aligned_size; if (!handle_description.in_heap) { auto& smmu = host1x.MemoryManager(); + size_t aligned_up = Common::AlignUp(map_size, BIG_PAGE_SIZE); smmu.Unmap(handle_description.d_address, map_size); - smmu.Free(handle_description.d_address, static_cast(map_size)); + smmu.Free(handle_description.d_address, static_cast(aligned_up)); handle_description.d_address = 0; return; } @@ -206,7 +207,8 @@ DAddr NvMap::PinHandle(NvMap::Handle::Id handle, bool low_area_pin) { handle_description->d_address = session->mapper->Map(vaddress, map_size); handle_description->in_heap = true; } else { - while ((address = smmu.Allocate(map_size)) == 0) { + size_t aligned_up = Common::AlignUp(map_size, BIG_PAGE_SIZE); + while ((address = smmu.Allocate(aligned_up)) == 0) { // Free handles until the allocation succeeds std::scoped_lock queueLock(unmap_queue_lock); if (auto freeHandleDesc{unmap_queue.front()}) { diff --git a/src/core/hle/service/nvdrv/core/nvmap.h b/src/core/hle/service/nvdrv/core/nvmap.h index 119efc38d..aa5cd21ec 100644 --- a/src/core/hle/service/nvdrv/core/nvmap.h +++ b/src/core/hle/service/nvdrv/core/nvmap.h @@ -63,8 +63,8 @@ public: } flags{}; static_assert(sizeof(Flags) == sizeof(u32)); - VAddr address{}; //!< The memory location in the guest's AS that this handle corresponds to, - //!< this can also be in the nvdrv tmem + VAddr address{}; //!< The memory location in the guest's AS that this handle corresponds to, + //!< this can also be in the nvdrv tmem bool is_shared_mem_mapped{}; //!< If this nvmap has been mapped with the MapSharedMem IPC //!< call @@ -73,8 +73,8 @@ public: bool in_heap{}; size_t session_id{}; - DAddr d_address{}; //!< The memory location in the device's AS that this handle corresponds to, - //!< this can also be in the nvdrv tmem + DAddr d_address{}; //!< The memory location in the device's AS that this handle corresponds + //!< to, this can also be in the nvdrv tmem Handle(u64 size, Id id); @@ -82,7 +82,8 @@ public: * @brief Sets up the handle with the given memory config, can allocate memory from the tmem * if a 0 address is passed */ - [[nodiscard]] NvResult Alloc(Flags pFlags, u32 pAlign, u8 pKind, u64 pAddress, size_t pSessionId); + [[nodiscard]] NvResult Alloc(Flags pFlags, u32 pAlign, u8 pKind, u64 pAddress, + size_t pSessionId); /** * @brief Increases the dupe counter of the handle for the given session diff --git a/src/core/hle/service/nvdrv/devices/nvhost_nvdec_common.h b/src/core/hle/service/nvdrv/devices/nvhost_nvdec_common.h index b44b17a82..718e0fecd 100644 --- a/src/core/hle/service/nvdrv/devices/nvhost_nvdec_common.h +++ b/src/core/hle/service/nvdrv/devices/nvhost_nvdec_common.h @@ -4,8 +4,8 @@ #pragma once #include -#include #include +#include #include "common/common_types.h" #include "common/swap.h" diff --git a/src/core/hle/service/nvdrv/devices/nvhost_vic.cpp b/src/core/hle/service/nvdrv/devices/nvhost_vic.cpp index d4c93ea5d..a27bed29b 100644 --- a/src/core/hle/service/nvdrv/devices/nvhost_vic.cpp +++ b/src/core/hle/service/nvdrv/devices/nvhost_vic.cpp @@ -69,7 +69,7 @@ NvResult nvhost_vic::Ioctl3(DeviceFD fd, Ioctl command, std::span inpu } void nvhost_vic::OnOpen(size_t session_id, DeviceFD fd) { - sessions[fd] = session_id; + sessions[fd] = session_id; } void nvhost_vic::OnClose(DeviceFD fd) { diff --git a/src/core/hle/service/nvdrv/devices/nvmap.cpp b/src/core/hle/service/nvdrv/devices/nvmap.cpp index 24f49ddcd..08ee8ec24 100644 --- a/src/core/hle/service/nvdrv/devices/nvmap.cpp +++ b/src/core/hle/service/nvdrv/devices/nvmap.cpp @@ -123,8 +123,8 @@ NvResult nvmap::IocAlloc(IocAllocParams& params, DeviceFD fd) { return NvResult::InsufficientMemory; } - const auto result = - handle_description->Alloc(params.flags, params.align, params.kind, params.address, sessions[fd]); + const auto result = handle_description->Alloc(params.flags, params.align, params.kind, + params.address, sessions[fd]); if (result != NvResult::Success) { LOG_CRITICAL(Service_NVDRV, "Object failed to allocate, handle={:08X}", params.handle); return result; diff --git a/src/core/hle/service/nvnflinger/fb_share_buffer_manager.cpp b/src/core/hle/service/nvnflinger/fb_share_buffer_manager.cpp index d36eff4ec..86e272b41 100644 --- a/src/core/hle/service/nvnflinger/fb_share_buffer_manager.cpp +++ b/src/core/hle/service/nvnflinger/fb_share_buffer_manager.cpp @@ -92,7 +92,8 @@ Result FreeNvMapHandle(Nvidia::Devices::nvmap& nvmap, u32 handle, Nvidia::Device Nvidia::Devices::nvmap::IocFreeParams free_params{ .handle = handle, }; - R_UNLESS(nvmap.IocFree(free_params, nvmap_fd) == Nvidia::NvResult::Success, VI::ResultOperationFailed); + R_UNLESS(nvmap.IocFree(free_params, nvmap_fd) == Nvidia::NvResult::Success, + VI::ResultOperationFailed); // We succeeded. R_SUCCEED(); @@ -109,7 +110,8 @@ Result AllocNvMapHandle(Nvidia::Devices::nvmap& nvmap, u32 handle, Common::Proce .kind = 0, .address = GetInteger(buffer), }; - R_UNLESS(nvmap.IocAlloc(alloc_params, nvmap_fd) == Nvidia::NvResult::Success, VI::ResultOperationFailed); + R_UNLESS(nvmap.IocAlloc(alloc_params, nvmap_fd) == Nvidia::NvResult::Success, + VI::ResultOperationFailed); // We succeeded. R_SUCCEED(); @@ -201,8 +203,8 @@ Result FbShareBufferManager::Initialize(u64* out_buffer_id, u64* out_layer_id, u m_nvmap_fd = m_nvdrv->Open("/dev/nvmap", m_session_id); // Create an nvmap handle for the buffer and assign the memory to it. - R_TRY(AllocateHandleForBuffer(std::addressof(m_buffer_nvmap_handle), *m_nvdrv, m_nvmap_fd, map_address, - SharedBufferSize)); + R_TRY(AllocateHandleForBuffer(std::addressof(m_buffer_nvmap_handle), *m_nvdrv, m_nvmap_fd, + map_address, SharedBufferSize)); // Record the display id. m_display_id = display_id; diff --git a/src/core/hle/service/nvnflinger/fb_share_buffer_manager.h b/src/core/hle/service/nvnflinger/fb_share_buffer_manager.h index 4b1a3d430..d2ec7a9b9 100644 --- a/src/core/hle/service/nvnflinger/fb_share_buffer_manager.h +++ b/src/core/hle/service/nvnflinger/fb_share_buffer_manager.h @@ -4,9 +4,9 @@ #pragma once #include "common/math_util.h" +#include "core/hle/service/nvdrv/nvdata.h" #include "core/hle/service/nvnflinger/nvnflinger.h" #include "core/hle/service/nvnflinger/ui/fence.h" -#include "core/hle/service/nvdrv/nvdata.h" namespace Kernel { class KPageGroup; @@ -62,7 +62,6 @@ private: Core::System& m_system; Nvnflinger& m_flinger; std::shared_ptr m_nvdrv; - }; } // namespace Service::Nvnflinger diff --git a/src/core/memory.h b/src/core/memory.h index 9d29cfd3f..552fd585f 100644 --- a/src/core/memory.h +++ b/src/core/memory.h @@ -488,7 +488,7 @@ public: void SetGPUDirtyManagers(std::span managers); bool InvalidateNCE(Common::ProcessAddress vaddr, size_t size); - + bool InvalidateSeparateHeap(void* fault_address); private: diff --git a/src/video_core/buffer_cache/buffer_cache.h b/src/video_core/buffer_cache/buffer_cache.h index 6fe2e8b93..5325a715a 100644 --- a/src/video_core/buffer_cache/buffer_cache.h +++ b/src/video_core/buffer_cache/buffer_cache.h @@ -240,8 +240,8 @@ bool BufferCache

::DMACopy(GPUVAddr src_address, GPUVAddr dest_address, u64 am memory_tracker.MarkRegionAsGpuModified(*cpu_dest_address, amount); } - Tegra::Memory::DeviceGuestMemoryScoped tmp( - device_memory, *cpu_src_address, amount, &tmp_buffer); + Tegra::Memory::DeviceGuestMemoryScoped + tmp(device_memory, *cpu_src_address, amount, &tmp_buffer); tmp.SetAddressAndSize(*cpu_dest_address, amount); return true; } @@ -1355,7 +1355,7 @@ typename BufferCache

::OverlapResult BufferCache

::ResolveOverlaps(DAddr dev bool has_stream_leap = false; auto expand_begin = [&](DAddr add_value) { static constexpr DAddr min_page = CACHING_PAGESIZE + Core::Memory::YUZU_PAGESIZE; - if (add_value > begin - min_page ) { + if (add_value > begin - min_page) { begin = min_page; device_addr = Core::Memory::YUZU_PAGESIZE; return; @@ -1365,7 +1365,7 @@ typename BufferCache

::OverlapResult BufferCache

::ResolveOverlaps(DAddr dev }; auto expand_end = [&](DAddr add_value) { static constexpr DAddr max_page = 1ULL << Tegra::MaxwellDeviceMemoryManager::AS_BITS; - if (add_value > max_page - end ) { + if (add_value > max_page - end) { end = max_page; return; } diff --git a/src/video_core/buffer_cache/word_manager.h b/src/video_core/buffer_cache/word_manager.h index 56ab4f5f1..1ca333b32 100644 --- a/src/video_core/buffer_cache/word_manager.h +++ b/src/video_core/buffer_cache/word_manager.h @@ -472,8 +472,8 @@ private: u64 changed_bits = (add_to_tracker ? current_bits : ~current_bits) & new_bits; VAddr addr = cpu_addr + word_index * BYTES_PER_WORD; IteratePages(changed_bits, [&](size_t offset, size_t size) { - tracker->UpdatePagesCachedCount(addr + offset * BYTES_PER_PAGE, - size * BYTES_PER_PAGE, add_to_tracker ? 1 : -1); + tracker->UpdatePagesCachedCount(addr + offset * BYTES_PER_PAGE, size * BYTES_PER_PAGE, + add_to_tracker ? 1 : -1); }); } diff --git a/src/video_core/dma_pusher.cpp b/src/video_core/dma_pusher.cpp index d461c5be8..fb2060ca4 100644 --- a/src/video_core/dma_pusher.cpp +++ b/src/video_core/dma_pusher.cpp @@ -86,14 +86,14 @@ bool DmaPusher::Step() { } const auto safe_process = [&] { Tegra::Memory::GpuGuestMemory + Tegra::Memory::GuestMemoryFlags::SafeRead> headers(memory_manager, dma_state.dma_get, command_list_header.size, &command_headers); ProcessCommands(headers); }; const auto unsafe_process = [&] { Tegra::Memory::GpuGuestMemory + Tegra::Memory::GuestMemoryFlags::UnsafeRead> headers(memory_manager, dma_state.dma_get, command_list_header.size, &command_headers); ProcessCommands(headers); diff --git a/src/video_core/engines/sw_blitter/blitter.cpp b/src/video_core/engines/sw_blitter/blitter.cpp index b67589daf..4bc079024 100644 --- a/src/video_core/engines/sw_blitter/blitter.cpp +++ b/src/video_core/engines/sw_blitter/blitter.cpp @@ -8,10 +8,10 @@ #include "common/scratch_buffer.h" #include "video_core/engines/sw_blitter/blitter.h" #include "video_core/engines/sw_blitter/converter.h" +#include "video_core/guest_memory.h" #include "video_core/memory_manager.h" #include "video_core/surface.h" #include "video_core/textures/decoders.h" -#include "video_core/guest_memory.h" namespace Tegra { class MemoryManager; diff --git a/src/video_core/guest_memory.h b/src/video_core/guest_memory.h index a2132f7ea..8b6213172 100644 --- a/src/video_core/guest_memory.h +++ b/src/video_core/guest_memory.h @@ -20,7 +20,8 @@ using GuestMemoryFlags = Core::Memory::GuestMemoryFlags; template using DeviceGuestMemory = Core::Memory::GuestMemory; template -using DeviceGuestMemoryScoped = Core::Memory::GuestMemoryScoped; +using DeviceGuestMemoryScoped = + Core::Memory::GuestMemoryScoped; template using GpuGuestMemory = Core::Memory::GuestMemory; template diff --git a/src/video_core/host1x/codecs/vp9.cpp b/src/video_core/host1x/codecs/vp9.cpp index e2ae1f76d..65d6fb2d5 100644 --- a/src/video_core/host1x/codecs/vp9.cpp +++ b/src/video_core/host1x/codecs/vp9.cpp @@ -383,9 +383,8 @@ Vp9FrameContainer VP9::GetCurrentFrame(const Host1x::NvdecCommon::NvdecRegisters // gpu.SyncGuestHost(); epic, why? current_frame.info = GetVp9PictureInfo(state); current_frame.bit_stream.resize(current_frame.info.bitstream_size); - host1x.GMMU().ReadBlock(state.frame_bitstream_offset, - current_frame.bit_stream.data(), - current_frame.info.bitstream_size); + host1x.GMMU().ReadBlock(state.frame_bitstream_offset, current_frame.bit_stream.data(), + current_frame.info.bitstream_size); } if (!next_frame.bit_stream.empty()) { Vp9FrameContainer temp{ diff --git a/src/video_core/host1x/vic.cpp b/src/video_core/host1x/vic.cpp index 1826211a1..d154746af 100644 --- a/src/video_core/host1x/vic.cpp +++ b/src/video_core/host1x/vic.cpp @@ -167,7 +167,7 @@ void Vic::WriteRGBFrame(std::unique_ptr frame, const VicConfig& c // send pitch linear frame const size_t linear_size = width * height * 4; host1x.GMMU().WriteBlock(output_surface_luma_address, converted_frame_buf_addr, - linear_size); + linear_size); } } @@ -193,8 +193,7 @@ void Vic::WriteYUVFrame(std::unique_ptr frame, const VicConfig& c const std::size_t dst = y * aligned_width; std::memcpy(luma_buffer.data() + dst, luma_src + src, frame_width); } - host1x.GMMU().WriteBlock(output_surface_luma_address, luma_buffer.data(), - luma_buffer.size()); + host1x.GMMU().WriteBlock(output_surface_luma_address, luma_buffer.data(), luma_buffer.size()); // Chroma const std::size_t half_height = frame_height / 2; @@ -234,7 +233,7 @@ void Vic::WriteYUVFrame(std::unique_ptr frame, const VicConfig& c break; } host1x.GMMU().WriteBlock(output_surface_chroma_address, chroma_buffer.data(), - chroma_buffer.size()); + chroma_buffer.size()); } } // namespace Host1x diff --git a/src/video_core/memory_manager.h b/src/video_core/memory_manager.h index 00d64dcce..eb00918fc 100644 --- a/src/video_core/memory_manager.h +++ b/src/video_core/memory_manager.h @@ -38,8 +38,9 @@ class MemoryManager final { public: explicit MemoryManager(Core::System& system_, u64 address_space_bits_ = 40, u64 big_page_bits_ = 16, u64 page_bits_ = 12); - explicit MemoryManager(Core::System& system_, MaxwellDeviceMemoryManager& memory_, u64 address_space_bits_ = 40, - u64 big_page_bits_ = 16, u64 page_bits_ = 12); + explicit MemoryManager(Core::System& system_, MaxwellDeviceMemoryManager& memory_, + u64 address_space_bits_ = 40, u64 big_page_bits_ = 16, + u64 page_bits_ = 12); ~MemoryManager(); size_t GetID() const { diff --git a/src/video_core/query_cache/query_cache_base.h b/src/video_core/query_cache/query_cache_base.h index 7720456ff..3c820b5f2 100644 --- a/src/video_core/query_cache/query_cache_base.h +++ b/src/video_core/query_cache/query_cache_base.h @@ -15,9 +15,9 @@ #include "common/common_types.h" #include "core/memory.h" #include "video_core/control/channel_state_cache.h" +#include "video_core/host1x/gpu_device_memory_manager.h" #include "video_core/query_cache/query_base.h" #include "video_core/query_cache/types.h" -#include "video_core/host1x/gpu_device_memory_manager.h" namespace VideoCore { class RasterizerInterface; @@ -50,7 +50,8 @@ public: }; explicit QueryCacheBase(Tegra::GPU& gpu, VideoCore::RasterizerInterface& rasterizer_, - Tegra::MaxwellDeviceMemoryManager& device_memory_, RuntimeType& runtime_); + Tegra::MaxwellDeviceMemoryManager& device_memory_, + RuntimeType& runtime_); ~QueryCacheBase(); diff --git a/src/video_core/renderer_opengl/gl_buffer_cache.h b/src/video_core/renderer_opengl/gl_buffer_cache.h index e6ad030cb..af34c272b 100644 --- a/src/video_core/renderer_opengl/gl_buffer_cache.h +++ b/src/video_core/renderer_opengl/gl_buffer_cache.h @@ -20,8 +20,7 @@ class BufferCacheRuntime; class Buffer : public VideoCommon::BufferBase { public: - explicit Buffer(BufferCacheRuntime&, DAddr cpu_addr, - u64 size_bytes); + explicit Buffer(BufferCacheRuntime&, DAddr cpu_addr, u64 size_bytes); explicit Buffer(BufferCacheRuntime&, VideoCommon::NullBufferParams); void ImmediateUpload(size_t offset, std::span data) noexcept; diff --git a/src/video_core/renderer_opengl/gl_query_cache.cpp b/src/video_core/renderer_opengl/gl_query_cache.cpp index 567292e1c..2147d587f 100644 --- a/src/video_core/renderer_opengl/gl_query_cache.cpp +++ b/src/video_core/renderer_opengl/gl_query_cache.cpp @@ -35,7 +35,8 @@ constexpr GLenum GetTarget(VideoCore::QueryType type) { } // Anonymous namespace -QueryCache::QueryCache(RasterizerOpenGL& rasterizer_, Tegra::MaxwellDeviceMemoryManager& device_memory_) +QueryCache::QueryCache(RasterizerOpenGL& rasterizer_, + Tegra::MaxwellDeviceMemoryManager& device_memory_) : QueryCacheLegacy(rasterizer_, device_memory_), gl_rasterizer{rasterizer_} { EnableCounters(); } diff --git a/src/video_core/renderer_opengl/gl_query_cache.h b/src/video_core/renderer_opengl/gl_query_cache.h index 04a1b39c9..38118f355 100644 --- a/src/video_core/renderer_opengl/gl_query_cache.h +++ b/src/video_core/renderer_opengl/gl_query_cache.h @@ -8,10 +8,10 @@ #include #include "common/common_types.h" +#include "video_core/host1x/gpu_device_memory_manager.h" #include "video_core/query_cache.h" #include "video_core/rasterizer_interface.h" #include "video_core/renderer_opengl/gl_resource_manager.h" -#include "video_core/host1x/gpu_device_memory_manager.h" namespace Core { class System; @@ -29,7 +29,8 @@ using CounterStream = VideoCommon::CounterStreamBase; class QueryCache final : public VideoCommon::QueryCacheLegacy { public: - explicit QueryCache(RasterizerOpenGL& rasterizer_, Tegra::MaxwellDeviceMemoryManager& device_memory_); + explicit QueryCache(RasterizerOpenGL& rasterizer_, + Tegra::MaxwellDeviceMemoryManager& device_memory_); ~QueryCache(); OGLQuery AllocateQuery(VideoCore::QueryType type); diff --git a/src/video_core/renderer_opengl/gl_rasterizer.h b/src/video_core/renderer_opengl/gl_rasterizer.h index f197774ed..34aa73526 100644 --- a/src/video_core/renderer_opengl/gl_rasterizer.h +++ b/src/video_core/renderer_opengl/gl_rasterizer.h @@ -75,9 +75,9 @@ class RasterizerOpenGL : public VideoCore::RasterizerInterface, protected VideoCommon::ChannelSetupCaches { public: explicit RasterizerOpenGL(Core::Frontend::EmuWindow& emu_window_, Tegra::GPU& gpu_, - Tegra::MaxwellDeviceMemoryManager& device_memory_, const Device& device_, - ScreenInfo& screen_info_, ProgramManager& program_manager_, - StateTracker& state_tracker_); + Tegra::MaxwellDeviceMemoryManager& device_memory_, + const Device& device_, ScreenInfo& screen_info_, + ProgramManager& program_manager_, StateTracker& state_tracker_); ~RasterizerOpenGL() override; void Draw(bool is_indexed, u32 instance_count) override; diff --git a/src/video_core/renderer_vulkan/renderer_vulkan.cpp b/src/video_core/renderer_vulkan/renderer_vulkan.cpp index 0e1815076..1631276c6 100644 --- a/src/video_core/renderer_vulkan/renderer_vulkan.cpp +++ b/src/video_core/renderer_vulkan/renderer_vulkan.cpp @@ -97,8 +97,8 @@ RendererVulkan::RendererVulkan(Core::TelemetrySession& telemetry_session_, render_window.GetFramebufferLayout().height), present_manager(instance, render_window, device, memory_allocator, scheduler, swapchain, surface), - blit_screen(device_memory, render_window, device, memory_allocator, swapchain, present_manager, - scheduler, screen_info), + blit_screen(device_memory, render_window, device, memory_allocator, swapchain, + present_manager, scheduler, screen_info), rasterizer(render_window, gpu, device_memory, screen_info, device, memory_allocator, state_tracker, scheduler) { if (Settings::values.renderer_force_max_clock.GetValue() && device.ShouldBoostClocks()) { diff --git a/src/video_core/renderer_vulkan/renderer_vulkan.h b/src/video_core/renderer_vulkan/renderer_vulkan.h index e5ce4692d..11c52287a 100644 --- a/src/video_core/renderer_vulkan/renderer_vulkan.h +++ b/src/video_core/renderer_vulkan/renderer_vulkan.h @@ -7,12 +7,12 @@ #include #include -#include "video_core/renderer_vulkan/vk_rasterizer.h" - #include "common/dynamic_library.h" +#include "video_core/host1x/gpu_device_memory_manager.h" #include "video_core/renderer_base.h" #include "video_core/renderer_vulkan/vk_blit_screen.h" #include "video_core/renderer_vulkan/vk_present_manager.h" +#include "video_core/renderer_vulkan/vk_rasterizer.h" #include "video_core/renderer_vulkan/vk_scheduler.h" #include "video_core/renderer_vulkan/vk_state_tracker.h" #include "video_core/renderer_vulkan/vk_swapchain.h" @@ -20,7 +20,6 @@ #include "video_core/vulkan_common/vulkan_device.h" #include "video_core/vulkan_common/vulkan_memory_allocator.h" #include "video_core/vulkan_common/vulkan_wrapper.h" -#include "video_core/host1x/gpu_device_memory_manager.h" namespace Core { class TelemetrySession; diff --git a/src/video_core/renderer_vulkan/vk_blit_screen.h b/src/video_core/renderer_vulkan/vk_blit_screen.h index cb941a956..3eff76009 100644 --- a/src/video_core/renderer_vulkan/vk_blit_screen.h +++ b/src/video_core/renderer_vulkan/vk_blit_screen.h @@ -6,9 +6,9 @@ #include #include "core/frontend/framebuffer_layout.h" +#include "video_core/host1x/gpu_device_memory_manager.h" #include "video_core/vulkan_common/vulkan_memory_allocator.h" #include "video_core/vulkan_common/vulkan_wrapper.h" -#include "video_core/host1x/gpu_device_memory_manager.h" namespace Core { class System; @@ -53,8 +53,9 @@ struct ScreenInfo { class BlitScreen { public: - explicit BlitScreen(Tegra::MaxwellDeviceMemoryManager& device_memory, Core::Frontend::EmuWindow& render_window, - const Device& device, MemoryAllocator& memory_manager, Swapchain& swapchain, + explicit BlitScreen(Tegra::MaxwellDeviceMemoryManager& device_memory, + Core::Frontend::EmuWindow& render_window, const Device& device, + MemoryAllocator& memory_manager, Swapchain& swapchain, PresentManager& present_manager, Scheduler& scheduler, const ScreenInfo& screen_info); ~BlitScreen(); diff --git a/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp b/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp index bec20c21a..1e1821b10 100644 --- a/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp +++ b/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp @@ -298,8 +298,9 @@ bool GraphicsPipelineCacheKey::operator==(const GraphicsPipelineCacheKey& rhs) c return std::memcmp(&rhs, this, Size()) == 0; } -PipelineCache::PipelineCache(Tegra::MaxwellDeviceMemoryManager& device_memory_, const Device& device_, - Scheduler& scheduler_, DescriptorPool& descriptor_pool_, +PipelineCache::PipelineCache(Tegra::MaxwellDeviceMemoryManager& device_memory_, + const Device& device_, Scheduler& scheduler_, + DescriptorPool& descriptor_pool_, GuestDescriptorQueue& guest_descriptor_queue_, RenderPassCache& render_pass_cache_, BufferCache& buffer_cache_, TextureCache& texture_cache_, VideoCore::ShaderNotify& shader_notify_) diff --git a/src/video_core/renderer_vulkan/vk_pipeline_cache.h b/src/video_core/renderer_vulkan/vk_pipeline_cache.h index 354fdc8ed..797700128 100644 --- a/src/video_core/renderer_vulkan/vk_pipeline_cache.h +++ b/src/video_core/renderer_vulkan/vk_pipeline_cache.h @@ -20,13 +20,13 @@ #include "shader_recompiler/object_pool.h" #include "shader_recompiler/profile.h" #include "video_core/engines/maxwell_3d.h" +#include "video_core/host1x/gpu_device_memory_manager.h" #include "video_core/renderer_vulkan/fixed_pipeline_state.h" #include "video_core/renderer_vulkan/vk_buffer_cache.h" #include "video_core/renderer_vulkan/vk_compute_pipeline.h" #include "video_core/renderer_vulkan/vk_graphics_pipeline.h" #include "video_core/renderer_vulkan/vk_texture_cache.h" #include "video_core/shader_cache.h" -#include "video_core/host1x/gpu_device_memory_manager.h" namespace Core { class System; @@ -99,8 +99,8 @@ struct ShaderPools { class PipelineCache : public VideoCommon::ShaderCache { public: - explicit PipelineCache(Tegra::MaxwellDeviceMemoryManager& device_memory_, const Device& device, Scheduler& scheduler, - DescriptorPool& descriptor_pool, + explicit PipelineCache(Tegra::MaxwellDeviceMemoryManager& device_memory_, const Device& device, + Scheduler& scheduler, DescriptorPool& descriptor_pool, GuestDescriptorQueue& guest_descriptor_queue, RenderPassCache& render_pass_cache, BufferCache& buffer_cache, TextureCache& texture_cache, VideoCore::ShaderNotify& shader_notify_); diff --git a/src/video_core/renderer_vulkan/vk_query_cache.cpp b/src/video_core/renderer_vulkan/vk_query_cache.cpp index d59fe698c..522f92dae 100644 --- a/src/video_core/renderer_vulkan/vk_query_cache.cpp +++ b/src/video_core/renderer_vulkan/vk_query_cache.cpp @@ -14,10 +14,10 @@ #include "common/bit_util.h" #include "common/common_types.h" #include "core/memory.h" -#include "video_core/rasterizer_interface.h" #include "video_core/engines/draw_manager.h" #include "video_core/host1x/gpu_device_memory_manager.h" #include "video_core/query_cache/query_cache.h" +#include "video_core/rasterizer_interface.h" #include "video_core/renderer_vulkan/vk_buffer_cache.h" #include "video_core/renderer_vulkan/vk_compute_pass.h" #include "video_core/renderer_vulkan/vk_query_cache.h" @@ -1156,9 +1156,10 @@ private: struct QueryCacheRuntimeImpl { QueryCacheRuntimeImpl(QueryCacheRuntime& runtime, VideoCore::RasterizerInterface* rasterizer_, - Tegra::MaxwellDeviceMemoryManager& device_memory_, Vulkan::BufferCache& buffer_cache_, - const Device& device_, const MemoryAllocator& memory_allocator_, - Scheduler& scheduler_, StagingBufferPool& staging_pool_, + Tegra::MaxwellDeviceMemoryManager& device_memory_, + Vulkan::BufferCache& buffer_cache_, const Device& device_, + const MemoryAllocator& memory_allocator_, Scheduler& scheduler_, + StagingBufferPool& staging_pool_, ComputePassDescriptorQueue& compute_pass_descriptor_queue, DescriptorPool& descriptor_pool) : rasterizer{rasterizer_}, device_memory{device_memory_}, diff --git a/src/video_core/renderer_vulkan/vk_rasterizer.cpp b/src/video_core/renderer_vulkan/vk_rasterizer.cpp index 7db131985..874927311 100644 --- a/src/video_core/renderer_vulkan/vk_rasterizer.cpp +++ b/src/video_core/renderer_vulkan/vk_rasterizer.cpp @@ -38,7 +38,6 @@ #include "video_core/vulkan_common/vulkan_device.h" #include "video_core/vulkan_common/vulkan_wrapper.h" - namespace Vulkan { using Maxwell = Tegra::Engines::Maxwell3D::Regs; diff --git a/src/video_core/renderer_vulkan/vk_rasterizer.h b/src/video_core/renderer_vulkan/vk_rasterizer.h index d593f35df..881ee0993 100644 --- a/src/video_core/renderer_vulkan/vk_rasterizer.h +++ b/src/video_core/renderer_vulkan/vk_rasterizer.h @@ -7,13 +7,13 @@ #include -#include "video_core/renderer_vulkan/vk_buffer_cache.h" - #include "common/common_types.h" #include "video_core/control/channel_state_cache.h" #include "video_core/engines/maxwell_dma.h" +#include "video_core/host1x/gpu_device_memory_manager.h" #include "video_core/rasterizer_interface.h" #include "video_core/renderer_vulkan/blit_image.h" +#include "video_core/renderer_vulkan/vk_buffer_cache.h" #include "video_core/renderer_vulkan/vk_descriptor_pool.h" #include "video_core/renderer_vulkan/vk_fence_manager.h" #include "video_core/renderer_vulkan/vk_pipeline_cache.h" @@ -24,7 +24,6 @@ #include "video_core/renderer_vulkan/vk_update_descriptor.h" #include "video_core/vulkan_common/vulkan_memory_allocator.h" #include "video_core/vulkan_common/vulkan_wrapper.h" -#include "video_core/host1x/gpu_device_memory_manager.h" namespace Core { class System; diff --git a/src/video_core/shader_cache.cpp b/src/video_core/shader_cache.cpp index 86fd62428..2af32c8f2 100644 --- a/src/video_core/shader_cache.cpp +++ b/src/video_core/shader_cache.cpp @@ -35,7 +35,8 @@ void ShaderCache::SyncGuestHost() { RemovePendingShaders(); } -ShaderCache::ShaderCache(Tegra::MaxwellDeviceMemoryManager& device_memory_) : device_memory{device_memory_} {} +ShaderCache::ShaderCache(Tegra::MaxwellDeviceMemoryManager& device_memory_) + : device_memory{device_memory_} {} bool ShaderCache::RefreshStages(std::array& unique_hashes) { auto& dirty{maxwell3d->dirty.flags}; diff --git a/src/video_core/shader_cache.h b/src/video_core/shader_cache.h index 02ef39483..fd9bf2562 100644 --- a/src/video_core/shader_cache.h +++ b/src/video_core/shader_cache.h @@ -14,9 +14,9 @@ #include "common/common_types.h" #include "common/polyfill_ranges.h" #include "video_core/control/channel_state_cache.h" +#include "video_core/host1x/gpu_device_memory_manager.h" #include "video_core/rasterizer_interface.h" #include "video_core/shader_environment.h" -#include "video_core/host1x/gpu_device_memory_manager.h" namespace Tegra { class MemoryManager; -- cgit v1.2.3 From 23430e67724d803184b6a861e4bcb3cac0e38cb0 Mon Sep 17 00:00:00 2001 From: Fernando Sahmkow Date: Sun, 7 Jan 2024 05:33:43 +0100 Subject: Core: Eliminate core/memory dependancies. --- src/core/device_memory_manager.h | 4 ++++ src/core/gpu_dirty_memory_manager.h | 4 ++-- src/tests/video_core/memory_tracker.cpp | 7 +++---- src/video_core/buffer_cache/buffer_cache.h | 21 ++++++++++----------- src/video_core/buffer_cache/buffer_cache_base.h | 4 ++-- src/video_core/buffer_cache/word_manager.h | 4 ++-- src/video_core/engines/maxwell_3d.cpp | 1 - src/video_core/engines/maxwell_dma.cpp | 1 - src/video_core/memory_manager.cpp | 8 ++++---- src/video_core/memory_manager.h | 1 - src/video_core/query_cache.h | 1 - src/video_core/query_cache/query_cache.h | 15 +++++++-------- src/video_core/query_cache/query_cache_base.h | 7 +++---- src/video_core/renderer_null/null_rasterizer.cpp | 5 ++--- src/video_core/renderer_opengl/gl_rasterizer.cpp | 4 ++-- src/video_core/renderer_opengl/renderer_opengl.cpp | 1 - src/video_core/renderer_vulkan/pipeline_helper.h | 1 - .../renderer_vulkan/vk_graphics_pipeline.cpp | 1 + src/video_core/renderer_vulkan/vk_query_cache.cpp | 5 ++--- src/video_core/renderer_vulkan/vk_rasterizer.cpp | 4 ++-- src/video_core/texture_cache/util.cpp | 1 - 21 files changed, 46 insertions(+), 54 deletions(-) (limited to 'src/video_core/renderer_vulkan') diff --git a/src/core/device_memory_manager.h b/src/core/device_memory_manager.h index cc9fd023f..6311e9ece 100644 --- a/src/core/device_memory_manager.h +++ b/src/core/device_memory_manager.h @@ -15,6 +15,10 @@ namespace Core { +constexpr size_t DEVICE_PAGEBITS = 12ULL; +constexpr size_t DEVICE_PAGESIZE = 1ULL << DEVICE_PAGEBITS; +constexpr size_t DEVICE_PAGEMASK = DEVICE_PAGESIZE - 1ULL; + class DeviceMemory; namespace Memory { diff --git a/src/core/gpu_dirty_memory_manager.h b/src/core/gpu_dirty_memory_manager.h index f1abf4f83..cc8fc176f 100644 --- a/src/core/gpu_dirty_memory_manager.h +++ b/src/core/gpu_dirty_memory_manager.h @@ -10,7 +10,7 @@ #include #include -#include "core/memory.h" +#include "core/device_memory_manager.h" namespace Core { @@ -80,7 +80,7 @@ private: u32 mask; }; - constexpr static size_t page_bits = Memory::YUZU_PAGEBITS - 1; + constexpr static size_t page_bits = DEVICE_PAGEBITS - 1; constexpr static size_t page_size = 1ULL << page_bits; constexpr static size_t page_mask = page_size - 1; diff --git a/src/tests/video_core/memory_tracker.cpp b/src/tests/video_core/memory_tracker.cpp index 618793668..0e559a590 100644 --- a/src/tests/video_core/memory_tracker.cpp +++ b/src/tests/video_core/memory_tracker.cpp @@ -24,9 +24,8 @@ constexpr VAddr c = 16 * HIGH_PAGE_SIZE; class RasterizerInterface { public: void UpdatePagesCachedCount(VAddr addr, u64 size, int delta) { - const u64 page_start{addr >> Core::Memory::YUZU_PAGEBITS}; - const u64 page_end{(addr + size + Core::Memory::YUZU_PAGESIZE - 1) >> - Core::Memory::YUZU_PAGEBITS}; + const u64 page_start{addr >> Core::DEVICE_PAGEBITS}; + const u64 page_end{(addr + size + Core::DEVICE_PAGESIZE - 1) >> Core::DEVICE_PAGEBITS}; for (u64 page = page_start; page < page_end; ++page) { int& value = page_table[page]; value += delta; @@ -40,7 +39,7 @@ public: } [[nodiscard]] int Count(VAddr addr) const noexcept { - const auto it = page_table.find(addr >> Core::Memory::YUZU_PAGEBITS); + const auto it = page_table.find(addr >> Core::DEVICE_PAGEBITS); return it == page_table.end() ? 0 : it->second; } diff --git a/src/video_core/buffer_cache/buffer_cache.h b/src/video_core/buffer_cache/buffer_cache.h index 5325a715a..b4bf369d1 100644 --- a/src/video_core/buffer_cache/buffer_cache.h +++ b/src/video_core/buffer_cache/buffer_cache.h @@ -13,7 +13,7 @@ namespace VideoCommon { -using Core::Memory::YUZU_PAGESIZE; +using Core::DEVICE_PAGESIZE; template BufferCache

::BufferCache(Tegra::MaxwellDeviceMemoryManager& device_memory_, Runtime& runtime_) @@ -120,8 +120,8 @@ void BufferCache

::CachedWriteMemory(DAddr device_addr, u64 size) { if (!is_dirty) { return; } - DAddr aligned_start = Common::AlignDown(device_addr, YUZU_PAGESIZE); - DAddr aligned_end = Common::AlignUp(device_addr + size, YUZU_PAGESIZE); + DAddr aligned_start = Common::AlignDown(device_addr, DEVICE_PAGESIZE); + DAddr aligned_end = Common::AlignUp(device_addr + size, DEVICE_PAGESIZE); if (!IsRegionGpuModified(aligned_start, aligned_end - aligned_start)) { WriteMemory(device_addr, size); return; @@ -151,9 +151,8 @@ std::optional BufferCache

::GetFlushArea(DA u64 size) { std::optional area{}; area.emplace(); - DAddr device_addr_start_aligned = Common::AlignDown(device_addr, Core::Memory::YUZU_PAGESIZE); - DAddr device_addr_end_aligned = - Common::AlignUp(device_addr + size, Core::Memory::YUZU_PAGESIZE); + DAddr device_addr_start_aligned = Common::AlignDown(device_addr, Core::DEVICE_PAGESIZE); + DAddr device_addr_end_aligned = Common::AlignUp(device_addr + size, Core::DEVICE_PAGESIZE); area->start_address = device_addr_start_aligned; area->end_address = device_addr_end_aligned; if (memory_tracker.IsRegionPreflushable(device_addr, size)) { @@ -1354,10 +1353,10 @@ typename BufferCache

::OverlapResult BufferCache

::ResolveOverlaps(DAddr dev int stream_score = 0; bool has_stream_leap = false; auto expand_begin = [&](DAddr add_value) { - static constexpr DAddr min_page = CACHING_PAGESIZE + Core::Memory::YUZU_PAGESIZE; + static constexpr DAddr min_page = CACHING_PAGESIZE + Core::DEVICE_PAGESIZE; if (add_value > begin - min_page) { begin = min_page; - device_addr = Core::Memory::YUZU_PAGESIZE; + device_addr = Core::DEVICE_PAGESIZE; return; } begin -= add_value; @@ -1587,8 +1586,8 @@ bool BufferCache

::InlineMemory(DAddr dest_address, size_t copy_size, if (!is_dirty) { return false; } - DAddr aligned_start = Common::AlignDown(dest_address, YUZU_PAGESIZE); - DAddr aligned_end = Common::AlignUp(dest_address + copy_size, YUZU_PAGESIZE); + DAddr aligned_start = Common::AlignDown(dest_address, DEVICE_PAGESIZE); + DAddr aligned_end = Common::AlignUp(dest_address + copy_size, DEVICE_PAGESIZE); if (!IsRegionGpuModified(aligned_start, aligned_end - aligned_start)) { return false; } @@ -1786,7 +1785,7 @@ Binding BufferCache

::StorageBufferBinding(GPUVAddr ssbo_addr, u32 cbuf_index, ASSERT_MSG(device_addr, "Unaligned storage buffer address not found for cbuf index {}", cbuf_index); // The end address used for size calculation does not need to be aligned - const DAddr cpu_end = Common::AlignUp(*device_addr + size, Core::Memory::YUZU_PAGESIZE); + const DAddr cpu_end = Common::AlignUp(*device_addr + size, Core::DEVICE_PAGESIZE); const Binding binding{ .device_addr = *aligned_device_addr, diff --git a/src/video_core/buffer_cache/buffer_cache_base.h b/src/video_core/buffer_cache/buffer_cache_base.h index 4074003e4..80dbb81e7 100644 --- a/src/video_core/buffer_cache/buffer_cache_base.h +++ b/src/video_core/buffer_cache/buffer_cache_base.h @@ -449,8 +449,8 @@ private: } static bool IsRangeGranular(DAddr device_addr, size_t size) { - return (device_addr & ~Core::Memory::YUZU_PAGEMASK) == - ((device_addr + size) & ~Core::Memory::YUZU_PAGEMASK); + return (device_addr & ~Core::DEVICE_PAGEMASK) == + ((device_addr + size) & ~Core::DEVICE_PAGEMASK); } void RunGarbageCollector(); diff --git a/src/video_core/buffer_cache/word_manager.h b/src/video_core/buffer_cache/word_manager.h index 1ca333b32..3db9d8b42 100644 --- a/src/video_core/buffer_cache/word_manager.h +++ b/src/video_core/buffer_cache/word_manager.h @@ -13,12 +13,12 @@ #include "common/common_funcs.h" #include "common/common_types.h" #include "common/div_ceil.h" -#include "core/memory.h" +#include "video_core/host1x/gpu_device_memory_manager.h" namespace VideoCommon { constexpr u64 PAGES_PER_WORD = 64; -constexpr u64 BYTES_PER_PAGE = Core::Memory::YUZU_PAGESIZE; +constexpr u64 BYTES_PER_PAGE = Core::DEVICE_PAGESIZE; constexpr u64 BYTES_PER_WORD = PAGES_PER_WORD * BYTES_PER_PAGE; enum class Type { diff --git a/src/video_core/engines/maxwell_3d.cpp b/src/video_core/engines/maxwell_3d.cpp index 95ba4f76c..a94e1f043 100644 --- a/src/video_core/engines/maxwell_3d.cpp +++ b/src/video_core/engines/maxwell_3d.cpp @@ -9,7 +9,6 @@ #include "common/settings.h" #include "core/core.h" #include "core/core_timing.h" -#include "core/memory.h" #include "video_core/dirty_flags.h" #include "video_core/engines/draw_manager.h" #include "video_core/engines/maxwell_3d.h" diff --git a/src/video_core/engines/maxwell_dma.cpp b/src/video_core/engines/maxwell_dma.cpp index 4bf461fb0..2ebd21fc5 100644 --- a/src/video_core/engines/maxwell_dma.cpp +++ b/src/video_core/engines/maxwell_dma.cpp @@ -8,7 +8,6 @@ #include "common/polyfill_ranges.h" #include "common/settings.h" #include "core/core.h" -#include "core/memory.h" #include "video_core/engines/maxwell_3d.h" #include "video_core/engines/maxwell_dma.h" #include "video_core/guest_memory.h" diff --git a/src/video_core/memory_manager.cpp b/src/video_core/memory_manager.cpp index b18b44e42..a52f8e486 100644 --- a/src/video_core/memory_manager.cpp +++ b/src/video_core/memory_manager.cpp @@ -606,14 +606,14 @@ bool MemoryManager::IsGranularRange(GPUVAddr gpu_addr, std::size_t size) const { const std::size_t page{(page_index & big_page_mask) + size}; return page <= big_page_size; } - const std::size_t page{(gpu_addr & Core::Memory::YUZU_PAGEMASK) + size}; - return page <= Core::Memory::YUZU_PAGESIZE; + const std::size_t page{(gpu_addr & Core::DEVICE_PAGEMASK) + size}; + return page <= Core::DEVICE_PAGESIZE; } if (GetEntry(gpu_addr) != EntryType::Mapped) { return false; } - const std::size_t page{(gpu_addr & Core::Memory::YUZU_PAGEMASK) + size}; - return page <= Core::Memory::YUZU_PAGESIZE; + const std::size_t page{(gpu_addr & Core::DEVICE_PAGEMASK) + size}; + return page <= Core::DEVICE_PAGESIZE; } bool MemoryManager::IsContinuousRange(GPUVAddr gpu_addr, std::size_t size) const { diff --git a/src/video_core/memory_manager.h b/src/video_core/memory_manager.h index eb00918fc..c5255f36c 100644 --- a/src/video_core/memory_manager.h +++ b/src/video_core/memory_manager.h @@ -15,7 +15,6 @@ #include "common/range_map.h" #include "common/scratch_buffer.h" #include "common/virtual_buffer.h" -#include "core/memory.h" #include "video_core/cache_types.h" #include "video_core/host1x/gpu_device_memory_manager.h" #include "video_core/pte_kind.h" diff --git a/src/video_core/query_cache.h b/src/video_core/query_cache.h index b01d843e4..4861b123a 100644 --- a/src/video_core/query_cache.h +++ b/src/video_core/query_cache.h @@ -18,7 +18,6 @@ #include "common/assert.h" #include "common/settings.h" -#include "core/memory.h" #include "video_core/control/channel_state_cache.h" #include "video_core/engines/maxwell_3d.h" #include "video_core/host1x/gpu_device_memory_manager.h" diff --git a/src/video_core/query_cache/query_cache.h b/src/video_core/query_cache/query_cache.h index b5e90cf8c..08b779055 100644 --- a/src/video_core/query_cache/query_cache.h +++ b/src/video_core/query_cache/query_cache.h @@ -15,7 +15,6 @@ #include "common/logging/log.h" #include "common/scope_exit.h" #include "common/settings.h" -#include "core/memory.h" #include "video_core/engines/maxwell_3d.h" #include "video_core/gpu.h" #include "video_core/host1x/gpu_device_memory_manager.h" @@ -253,8 +252,8 @@ void QueryCacheBase::CounterReport(GPUVAddr addr, QueryType counter_type query_location.stream_id.Assign(static_cast(streamer_id)); query_location.query_id.Assign(static_cast(new_query_id)); const auto gen_caching_indexing = [](VAddr cur_addr) { - return std::make_pair(cur_addr >> Core::Memory::YUZU_PAGEBITS, - static_cast(cur_addr & Core::Memory::YUZU_PAGEMASK)); + return std::make_pair(cur_addr >> Core::DEVICE_PAGEBITS, + static_cast(cur_addr & Core::DEVICE_PAGEMASK)); }; u8* pointer = impl->device_memory.template GetPointer(cpu_addr); u8* pointer_timestamp = impl->device_memory.template GetPointer(cpu_addr + 8); @@ -325,8 +324,8 @@ void QueryCacheBase::CounterReport(GPUVAddr addr, QueryType counter_type template void QueryCacheBase::UnregisterPending() { const auto gen_caching_indexing = [](VAddr cur_addr) { - return std::make_pair(cur_addr >> Core::Memory::YUZU_PAGEBITS, - static_cast(cur_addr & Core::Memory::YUZU_PAGEMASK)); + return std::make_pair(cur_addr >> Core::DEVICE_PAGEBITS, + static_cast(cur_addr & Core::DEVICE_PAGEMASK)); }; std::scoped_lock lock(cache_mutex); for (QueryLocation loc : impl->pending_unregister) { @@ -390,7 +389,7 @@ bool QueryCacheBase::AccelerateHostConditionalRendering() { } VAddr cpu_addr = *cpu_addr_opt; std::scoped_lock lock(cache_mutex); - auto it1 = cached_queries.find(cpu_addr >> Core::Memory::YUZU_PAGEBITS); + auto it1 = cached_queries.find(cpu_addr >> Core::DEVICE_PAGEBITS); if (it1 == cached_queries.end()) { return VideoCommon::LookupData{ .address = cpu_addr, @@ -398,10 +397,10 @@ bool QueryCacheBase::AccelerateHostConditionalRendering() { }; } auto& sub_container = it1->second; - auto it_current = sub_container.find(cpu_addr & Core::Memory::YUZU_PAGEMASK); + auto it_current = sub_container.find(cpu_addr & Core::DEVICE_PAGEMASK); if (it_current == sub_container.end()) { - auto it_current_2 = sub_container.find((cpu_addr & Core::Memory::YUZU_PAGEMASK) + 4); + auto it_current_2 = sub_container.find((cpu_addr & Core::DEVICE_PAGEMASK) + 4); if (it_current_2 == sub_container.end()) { return VideoCommon::LookupData{ .address = cpu_addr, diff --git a/src/video_core/query_cache/query_cache_base.h b/src/video_core/query_cache/query_cache_base.h index 3c820b5f2..c12fb75ef 100644 --- a/src/video_core/query_cache/query_cache_base.h +++ b/src/video_core/query_cache/query_cache_base.h @@ -13,7 +13,6 @@ #include "common/assert.h" #include "common/bit_field.h" #include "common/common_types.h" -#include "core/memory.h" #include "video_core/control/channel_state_cache.h" #include "video_core/host1x/gpu_device_memory_manager.h" #include "video_core/query_cache/query_base.h" @@ -123,10 +122,10 @@ protected: const u64 addr_begin = addr; const u64 addr_end = addr_begin + size; - const u64 page_end = addr_end >> Core::Memory::YUZU_PAGEBITS; + const u64 page_end = addr_end >> Core::DEVICE_PAGEBITS; std::scoped_lock lock(cache_mutex); - for (u64 page = addr_begin >> Core::Memory::YUZU_PAGEBITS; page <= page_end; ++page) { - const u64 page_start = page << Core::Memory::YUZU_PAGEBITS; + for (u64 page = addr_begin >> Core::DEVICE_PAGEBITS; page <= page_end; ++page) { + const u64 page_start = page << Core::DEVICE_PAGEBITS; const auto in_range = [page_start, addr_begin, addr_end](const u32 query_location) { const u64 cache_begin = page_start + query_location; const u64 cache_end = cache_begin + sizeof(u32); diff --git a/src/video_core/renderer_null/null_rasterizer.cpp b/src/video_core/renderer_null/null_rasterizer.cpp index 11b93fdc9..abfabb65b 100644 --- a/src/video_core/renderer_null/null_rasterizer.cpp +++ b/src/video_core/renderer_null/null_rasterizer.cpp @@ -2,7 +2,6 @@ // SPDX-License-Identifier: GPL-2.0-or-later #include "common/alignment.h" -#include "core/memory.h" #include "video_core/control/channel_state.h" #include "video_core/host1x/host1x.h" #include "video_core/memory_manager.h" @@ -55,8 +54,8 @@ bool RasterizerNull::OnCPUWrite(PAddr addr, u64 size) { void RasterizerNull::OnCacheInvalidation(PAddr addr, u64 size) {} VideoCore::RasterizerDownloadArea RasterizerNull::GetFlushArea(PAddr addr, u64 size) { VideoCore::RasterizerDownloadArea new_area{ - .start_address = Common::AlignDown(addr, Core::Memory::YUZU_PAGESIZE), - .end_address = Common::AlignUp(addr + size, Core::Memory::YUZU_PAGESIZE), + .start_address = Common::AlignDown(addr, Core::DEVICE_PAGESIZE), + .end_address = Common::AlignUp(addr + size, Core::DEVICE_PAGESIZE), .preemtive = true, }; return new_area; diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp index 71b748c74..d5354ef2d 100644 --- a/src/video_core/renderer_opengl/gl_rasterizer.cpp +++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp @@ -526,8 +526,8 @@ VideoCore::RasterizerDownloadArea RasterizerOpenGL::GetFlushArea(DAddr addr, u64 } } VideoCore::RasterizerDownloadArea new_area{ - .start_address = Common::AlignDown(addr, Core::Memory::YUZU_PAGESIZE), - .end_address = Common::AlignUp(addr + size, Core::Memory::YUZU_PAGESIZE), + .start_address = Common::AlignDown(addr, Core::DEVICE_PAGESIZE), + .end_address = Common::AlignUp(addr + size, Core::DEVICE_PAGESIZE), .preemtive = true, }; return new_area; diff --git a/src/video_core/renderer_opengl/renderer_opengl.cpp b/src/video_core/renderer_opengl/renderer_opengl.cpp index 821a045ad..b75376fdb 100644 --- a/src/video_core/renderer_opengl/renderer_opengl.cpp +++ b/src/video_core/renderer_opengl/renderer_opengl.cpp @@ -15,7 +15,6 @@ #include "common/telemetry.h" #include "core/core_timing.h" #include "core/frontend/emu_window.h" -#include "core/memory.h" #include "core/telemetry_session.h" #include "video_core/host_shaders/ffx_a_h.h" #include "video_core/host_shaders/ffx_fsr1_h.h" diff --git a/src/video_core/renderer_vulkan/pipeline_helper.h b/src/video_core/renderer_vulkan/pipeline_helper.h index 71c783709..850c34a3a 100644 --- a/src/video_core/renderer_vulkan/pipeline_helper.h +++ b/src/video_core/renderer_vulkan/pipeline_helper.h @@ -12,7 +12,6 @@ #include "shader_recompiler/shader_info.h" #include "video_core/renderer_vulkan/vk_texture_cache.h" #include "video_core/renderer_vulkan/vk_update_descriptor.h" -#include "video_core/texture_cache/texture_cache.h" #include "video_core/texture_cache/types.h" #include "video_core/vulkan_common/vulkan_device.h" diff --git a/src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp b/src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp index f2fd2670f..ec6b3a4b0 100644 --- a/src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp +++ b/src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp @@ -19,6 +19,7 @@ #include "video_core/renderer_vulkan/vk_texture_cache.h" #include "video_core/renderer_vulkan/vk_update_descriptor.h" #include "video_core/shader_notify.h" +#include "video_core/texture_cache/texture_cache.h" #include "video_core/vulkan_common/vulkan_device.h" #if defined(_MSC_VER) && defined(NDEBUG) diff --git a/src/video_core/renderer_vulkan/vk_query_cache.cpp b/src/video_core/renderer_vulkan/vk_query_cache.cpp index 522f92dae..7cbc9c73c 100644 --- a/src/video_core/renderer_vulkan/vk_query_cache.cpp +++ b/src/video_core/renderer_vulkan/vk_query_cache.cpp @@ -13,7 +13,6 @@ #include "common/bit_util.h" #include "common/common_types.h" -#include "core/memory.h" #include "video_core/engines/draw_manager.h" #include "video_core/host1x/gpu_device_memory_manager.h" #include "video_core/query_cache/query_cache.h" @@ -1482,8 +1481,8 @@ void QueryCacheRuntime::SyncValues(std::span values, VkBuffer ba for (auto& sync_val : values) { total_size += sync_val.size; bool found = false; - DAddr base = Common::AlignDown(sync_val.address, Core::Memory::YUZU_PAGESIZE); - DAddr base_end = base + Core::Memory::YUZU_PAGESIZE; + DAddr base = Common::AlignDown(sync_val.address, Core::DEVICE_PAGESIZE); + DAddr base_end = base + Core::DEVICE_PAGESIZE; for (size_t i = 0; i < impl->little_cache.size(); i++) { const auto set_found = [&] { impl->redirect_cache.push_back(i); diff --git a/src/video_core/renderer_vulkan/vk_rasterizer.cpp b/src/video_core/renderer_vulkan/vk_rasterizer.cpp index 874927311..5bf41b81f 100644 --- a/src/video_core/renderer_vulkan/vk_rasterizer.cpp +++ b/src/video_core/renderer_vulkan/vk_rasterizer.cpp @@ -553,8 +553,8 @@ VideoCore::RasterizerDownloadArea RasterizerVulkan::GetFlushArea(DAddr addr, u64 } } VideoCore::RasterizerDownloadArea new_area{ - .start_address = Common::AlignDown(addr, Core::Memory::YUZU_PAGESIZE), - .end_address = Common::AlignUp(addr + size, Core::Memory::YUZU_PAGESIZE), + .start_address = Common::AlignDown(addr, Core::DEVICE_PAGESIZE), + .end_address = Common::AlignUp(addr + size, Core::DEVICE_PAGESIZE), .preemtive = true, }; return new_area; diff --git a/src/video_core/texture_cache/util.cpp b/src/video_core/texture_cache/util.cpp index 96f04b6c8..1a6f0d1ad 100644 --- a/src/video_core/texture_cache/util.cpp +++ b/src/video_core/texture_cache/util.cpp @@ -20,7 +20,6 @@ #include "common/div_ceil.h" #include "common/scratch_buffer.h" #include "common/settings.h" -#include "core/memory.h" #include "video_core/compatible_formats.h" #include "video_core/engines/maxwell_3d.h" #include "video_core/guest_memory.h" -- cgit v1.2.3