From eb67a45ca82bc01ac843c853fd3c17f2a90e0250 Mon Sep 17 00:00:00 2001 From: ameerj Date: Mon, 26 Oct 2020 23:07:36 -0400 Subject: video_core: NVDEC Implementation This commit aims to implement the NVDEC (Nvidia Decoder) functionality, with video frame decoding being handled by the FFmpeg library. The process begins with Ioctl commands being sent to the NVDEC and VIC (Video Image Composer) emulated devices. These allocate the necessary GPU buffers for the frame data, along with providing information on the incoming video data. A Submit command then signals the GPU to process and decode the frame data. To decode the frame, the respective codec's header must be manually composed from the information provided by NVDEC, then sent with the raw frame data to the ffmpeg library. Currently, H264 and VP9 are supported, with VP9 having some minor artifacting issues related mainly to the reference frame composition in its uncompressed header. Async GPU is not properly implemented at the moment. Co-Authored-By: David <25727384+ogniK5377@users.noreply.github.com> --- src/video_core/cdma_pusher.h | 138 +++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 138 insertions(+) create mode 100644 src/video_core/cdma_pusher.h (limited to 'src/video_core/cdma_pusher.h') diff --git a/src/video_core/cdma_pusher.h b/src/video_core/cdma_pusher.h new file mode 100644 index 000000000..982f309c5 --- /dev/null +++ b/src/video_core/cdma_pusher.h @@ -0,0 +1,138 @@ +// Copyright 2020 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include +#include +#include +#include + +#include "common/bit_field.h" +#include "common/common_types.h" +#include "video_core/command_classes/sync_manager.h" + +namespace Tegra { + +class GPU; +class Nvdec; +class Vic; +class Host1x; + +enum class ChSubmissionMode : u32 { + SetClass = 0, + Incrementing = 1, + NonIncrementing = 2, + Mask = 3, + Immediate = 4, + Restart = 5, + Gather = 6, +}; + +enum class ChClassId : u32 { + NoClass = 0x0, + Host1x = 0x1, + VideoEncodeMpeg = 0x20, + VideoEncodeNvEnc = 0x21, + VideoStreamingVi = 0x30, + VideoStreamingIsp = 0x32, + VideoStreamingIspB = 0x34, + VideoStreamingViI2c = 0x36, + GraphicsVic = 0x5d, + Graphics3D = 0x60, + GraphicsGpu = 0x61, + Tsec = 0xe0, + TsecB = 0xe1, + NvJpg = 0xc0, + NvDec = 0xf0 +}; + +enum class ChMethod : u32 { + Empty = 0, + SetMethod = 0x10, + SetData = 0x11, +}; + +union ChCommandHeader { + u32 raw; + BitField<0, 16, u32> value; + BitField<16, 12, ChMethod> method_offset; + BitField<28, 4, ChSubmissionMode> submission_mode; +}; +static_assert(sizeof(ChCommandHeader) == sizeof(u32), "ChCommand header is an invalid size"); + +struct ChCommand { + ChClassId class_id{}; + int method_offset{}; + std::vector arguments; +}; + +using ChCommandHeaderList = std::vector; +using ChCommandList = std::vector; + +struct ThiRegisters { + u32_le increment_syncpt{}; + INSERT_PADDING_WORDS(1); + u32_le increment_syncpt_error{}; + u32_le ctx_switch_incremement_syncpt{}; + INSERT_PADDING_WORDS(4); + u32_le ctx_switch{}; + INSERT_PADDING_WORDS(1); + u32_le ctx_syncpt_eof{}; + INSERT_PADDING_WORDS(5); + u32_le method_0{}; + u32_le method_1{}; + INSERT_PADDING_WORDS(12); + u32_le int_status{}; + u32_le int_mask{}; +}; + +enum class ThiMethod : u32 { + IncSyncpt = offsetof(ThiRegisters, increment_syncpt) / sizeof(u32), + SetMethod0 = offsetof(ThiRegisters, method_0) / sizeof(u32), + SetMethod1 = offsetof(ThiRegisters, method_1) / sizeof(u32), +}; + +class CDmaPusher { +public: + explicit CDmaPusher(GPU& gpu); + ~CDmaPusher(); + + /// Push NVDEC command buffer entries into queue + void Push(ChCommandHeaderList&& entries); + + /// Process queued command buffer entries + void DispatchCalls(); + + /// Process one queue element + void Step(); + + /// Invoke command class devices to execute the command based on the current state + void ExecuteCommand(u32 offset, u32 data); + +private: + /// Write arguments value to the ThiRegisters member at the specified offset + void ThiStateWrite(ThiRegisters& state, u32 offset, const std::vector& arguments); + + GPU& gpu; + + std::shared_ptr nvdec_processor; + std::unique_ptr vic_processor; + std::unique_ptr host1x_processor; + std::unique_ptr nvdec_sync; + std::unique_ptr vic_sync; + ChClassId current_class{}; + ThiRegisters vic_thi_state{}; + ThiRegisters nvdec_thi_state{}; + + s32 count{}; + s32 offset{}; + s32 mask{}; + bool incrementing{}; + + // Queue of command lists to be processed + std::queue cdma_queue; +}; + +} // namespace Tegra -- cgit v1.2.3 From 677a8b208d47d0d2397197ce74c7039a8ea79d20 Mon Sep 17 00:00:00 2001 From: Lioncash Date: Fri, 4 Dec 2020 14:39:12 -0500 Subject: video_core: Resolve more variable shadowing scenarios Resolves variable shadowing scenarios up to the end of the OpenGL code to make it nicer to review. The rest will be resolved in a following commit. --- src/video_core/cdma_pusher.cpp | 30 +++++++------- src/video_core/cdma_pusher.h | 16 ++++---- src/video_core/command_classes/codecs/vp9.cpp | 2 +- src/video_core/command_classes/codecs/vp9.h | 2 +- src/video_core/dma_pusher.cpp | 9 +++- src/video_core/dma_pusher.h | 14 +++---- src/video_core/engines/engine_upload.cpp | 8 ++-- src/video_core/engines/engine_upload.h | 4 +- src/video_core/engines/kepler_memory.cpp | 4 +- src/video_core/engines/kepler_memory.h | 2 +- src/video_core/engines/maxwell_dma.cpp | 6 ++- src/video_core/engines/maxwell_dma.h | 4 +- src/video_core/fence_manager.h | 8 ++-- src/video_core/gpu.cpp | 8 +++- src/video_core/gpu.h | 12 +++--- src/video_core/gpu_asynch.cpp | 4 +- src/video_core/gpu_asynch.h | 2 +- src/video_core/gpu_synch.cpp | 2 +- src/video_core/gpu_synch.h | 2 +- src/video_core/gpu_thread.cpp | 16 ++++---- src/video_core/gpu_thread.h | 25 ++++++----- src/video_core/guest_driver.h | 4 +- src/video_core/macro/macro_hle.cpp | 6 +-- src/video_core/macro/macro_hle.h | 2 +- src/video_core/macro/macro_interpreter.cpp | 20 ++++----- src/video_core/macro/macro_interpreter.h | 10 ++--- src/video_core/macro/macro_jit_x64.cpp | 16 ++++---- src/video_core/macro/macro_jit_x64.h | 4 +- src/video_core/memory_manager.h | 4 +- .../renderer_opengl/gl_arb_decompiler.cpp | 48 +++++++++++----------- .../renderer_opengl/gl_fence_manager.cpp | 14 +++---- src/video_core/renderer_opengl/gl_fence_manager.h | 10 ++--- src/video_core/renderer_opengl/gl_query_cache.cpp | 15 +++---- src/video_core/renderer_opengl/gl_query_cache.h | 6 +-- src/video_core/renderer_opengl/gl_shader_cache.cpp | 6 +-- src/video_core/renderer_opengl/gl_shader_cache.h | 2 +- .../renderer_opengl/gl_shader_decompiler.cpp | 35 ++++++++-------- .../renderer_opengl/gl_shader_decompiler.h | 12 +++--- .../renderer_opengl/gl_texture_cache.cpp | 20 ++++----- src/video_core/renderer_opengl/gl_texture_cache.h | 5 ++- src/video_core/renderer_vulkan/vk_command_pool.cpp | 4 +- src/video_core/renderer_vulkan/vk_command_pool.h | 2 +- 42 files changed, 219 insertions(+), 206 deletions(-) (limited to 'src/video_core/cdma_pusher.h') diff --git a/src/video_core/cdma_pusher.cpp b/src/video_core/cdma_pusher.cpp index b60f86260..e3e7432f7 100644 --- a/src/video_core/cdma_pusher.cpp +++ b/src/video_core/cdma_pusher.cpp @@ -29,8 +29,8 @@ #include "video_core/memory_manager.h" namespace Tegra { -CDmaPusher::CDmaPusher(GPU& gpu) - : gpu(gpu), nvdec_processor(std::make_shared(gpu)), +CDmaPusher::CDmaPusher(GPU& gpu_) + : gpu{gpu_}, nvdec_processor(std::make_shared(gpu)), vic_processor(std::make_unique(gpu, nvdec_processor)), host1x_processor(std::make_unique(gpu)), nvdec_sync(std::make_unique(gpu)), @@ -100,11 +100,11 @@ void CDmaPusher::Step() { } } -void CDmaPusher::ExecuteCommand(u32 offset, u32 data) { +void CDmaPusher::ExecuteCommand(u32 state_offset, u32 data) { switch (current_class) { case ChClassId::NvDec: - ThiStateWrite(nvdec_thi_state, offset, {data}); - switch (static_cast(offset)) { + ThiStateWrite(nvdec_thi_state, state_offset, {data}); + switch (static_cast(state_offset)) { case ThiMethod::IncSyncpt: { LOG_DEBUG(Service_NVDRV, "NVDEC Class IncSyncpt Method"); const auto syncpoint_id = static_cast(data & 0xFF); @@ -120,16 +120,16 @@ void CDmaPusher::ExecuteCommand(u32 offset, u32 data) { case ThiMethod::SetMethod1: LOG_DEBUG(Service_NVDRV, "NVDEC method 0x{:X}", static_cast(nvdec_thi_state.method_0)); - nvdec_processor->ProcessMethod( - static_cast(nvdec_thi_state.method_0), {data}); + nvdec_processor->ProcessMethod(static_cast(nvdec_thi_state.method_0), + {data}); break; default: break; } break; case ChClassId::GraphicsVic: - ThiStateWrite(vic_thi_state, static_cast(offset), {data}); - switch (static_cast(offset)) { + ThiStateWrite(vic_thi_state, static_cast(state_offset), {data}); + switch (static_cast(state_offset)) { case ThiMethod::IncSyncpt: { LOG_DEBUG(Service_NVDRV, "VIC Class IncSyncpt Method"); const auto syncpoint_id = static_cast(data & 0xFF); @@ -145,8 +145,7 @@ void CDmaPusher::ExecuteCommand(u32 offset, u32 data) { case ThiMethod::SetMethod1: LOG_DEBUG(Service_NVDRV, "VIC method 0x{:X}, Args=({})", static_cast(vic_thi_state.method_0), data); - vic_processor->ProcessMethod(static_cast(vic_thi_state.method_0), - {data}); + vic_processor->ProcessMethod(static_cast(vic_thi_state.method_0), {data}); break; default: break; @@ -155,7 +154,7 @@ void CDmaPusher::ExecuteCommand(u32 offset, u32 data) { case ChClassId::Host1x: // This device is mainly for syncpoint synchronization LOG_DEBUG(Service_NVDRV, "Host1X Class Method"); - host1x_processor->ProcessMethod(static_cast(offset), {data}); + host1x_processor->ProcessMethod(static_cast(state_offset), {data}); break; default: UNIMPLEMENTED_MSG("Current class not implemented {:X}", static_cast(current_class)); @@ -163,9 +162,10 @@ void CDmaPusher::ExecuteCommand(u32 offset, u32 data) { } } -void CDmaPusher::ThiStateWrite(ThiRegisters& state, u32 offset, const std::vector& arguments) { - u8* const state_offset = reinterpret_cast(&state) + sizeof(u32) * offset; - std::memcpy(state_offset, arguments.data(), sizeof(u32) * arguments.size()); +void CDmaPusher::ThiStateWrite(ThiRegisters& state, u32 state_offset, + const std::vector& arguments) { + u8* const state_offset_ptr = reinterpret_cast(&state) + sizeof(u32) * state_offset; + std::memcpy(state_offset_ptr, arguments.data(), sizeof(u32) * arguments.size()); } } // namespace Tegra diff --git a/src/video_core/cdma_pusher.h b/src/video_core/cdma_pusher.h index 982f309c5..0db1cd646 100644 --- a/src/video_core/cdma_pusher.h +++ b/src/video_core/cdma_pusher.h @@ -68,8 +68,8 @@ struct ChCommand { std::vector arguments; }; -using ChCommandHeaderList = std::vector; -using ChCommandList = std::vector; +using ChCommandHeaderList = std::vector; +using ChCommandList = std::vector; struct ThiRegisters { u32_le increment_syncpt{}; @@ -96,7 +96,7 @@ enum class ThiMethod : u32 { class CDmaPusher { public: - explicit CDmaPusher(GPU& gpu); + explicit CDmaPusher(GPU& gpu_); ~CDmaPusher(); /// Push NVDEC command buffer entries into queue @@ -109,17 +109,17 @@ public: void Step(); /// Invoke command class devices to execute the command based on the current state - void ExecuteCommand(u32 offset, u32 data); + void ExecuteCommand(u32 state_offset, u32 data); private: /// Write arguments value to the ThiRegisters member at the specified offset - void ThiStateWrite(ThiRegisters& state, u32 offset, const std::vector& arguments); + void ThiStateWrite(ThiRegisters& state, u32 state_offset, const std::vector& arguments); GPU& gpu; - std::shared_ptr nvdec_processor; - std::unique_ptr vic_processor; - std::unique_ptr host1x_processor; + std::shared_ptr nvdec_processor; + std::unique_ptr vic_processor; + std::unique_ptr host1x_processor; std::unique_ptr nvdec_sync; std::unique_ptr vic_sync; ChClassId current_class{}; diff --git a/src/video_core/command_classes/codecs/vp9.cpp b/src/video_core/command_classes/codecs/vp9.cpp index 7d8d6ee3c..1771bc939 100644 --- a/src/video_core/command_classes/codecs/vp9.cpp +++ b/src/video_core/command_classes/codecs/vp9.cpp @@ -233,7 +233,7 @@ constexpr std::array map_lut{ } } // Anonymous namespace -VP9::VP9(GPU& gpu) : gpu(gpu) {} +VP9::VP9(GPU& gpu_) : gpu{gpu_} {} VP9::~VP9() = default; diff --git a/src/video_core/command_classes/codecs/vp9.h b/src/video_core/command_classes/codecs/vp9.h index 9ebbbf59e..8396c8105 100644 --- a/src/video_core/command_classes/codecs/vp9.h +++ b/src/video_core/command_classes/codecs/vp9.h @@ -108,7 +108,7 @@ private: class VP9 { public: - explicit VP9(GPU& gpu); + explicit VP9(GPU& gpu_); ~VP9(); VP9(const VP9&) = delete; diff --git a/src/video_core/dma_pusher.cpp b/src/video_core/dma_pusher.cpp index d8801b1f5..2c8b20024 100644 --- a/src/video_core/dma_pusher.cpp +++ b/src/video_core/dma_pusher.cpp @@ -13,7 +13,7 @@ namespace Tegra { -DmaPusher::DmaPusher(Core::System& system, GPU& gpu) : gpu{gpu}, system{system} {} +DmaPusher::DmaPusher(Core::System& system_, GPU& gpu_) : gpu{gpu_}, system{system_} {} DmaPusher::~DmaPusher() = default; @@ -152,7 +152,12 @@ void DmaPusher::SetState(const CommandHeader& command_header) { void DmaPusher::CallMethod(u32 argument) const { if (dma_state.method < non_puller_methods) { - gpu.CallMethod({dma_state.method, argument, dma_state.subchannel, dma_state.method_count}); + gpu.CallMethod(GPU::MethodCall{ + dma_state.method, + argument, + dma_state.subchannel, + dma_state.method_count, + }); } else { subchannels[dma_state.subchannel]->CallMethod(dma_state.method, argument, dma_state.is_last_call); diff --git a/src/video_core/dma_pusher.h b/src/video_core/dma_pusher.h index 96ac267f7..19f286fa7 100644 --- a/src/video_core/dma_pusher.h +++ b/src/video_core/dma_pusher.h @@ -87,11 +87,11 @@ inline CommandHeader BuildCommandHeader(BufferMethods method, u32 arg_count, Sub struct CommandList final { CommandList() = default; explicit CommandList(std::size_t size) : command_lists(size) {} - explicit CommandList(std::vector&& prefetch_command_list) - : prefetch_command_list{std::move(prefetch_command_list)} {} + explicit CommandList(std::vector&& prefetch_command_list_) + : prefetch_command_list{std::move(prefetch_command_list_)} {} - std::vector command_lists; - std::vector prefetch_command_list; + std::vector command_lists; + std::vector prefetch_command_list; }; /** @@ -103,7 +103,7 @@ struct CommandList final { */ class DmaPusher final { public: - explicit DmaPusher(Core::System& system, GPU& gpu); + explicit DmaPusher(Core::System& system_, GPU& gpu_); ~DmaPusher(); void Push(CommandList&& entries) { @@ -112,7 +112,7 @@ public: void DispatchCalls(); - void BindSubchannel(Tegra::Engines::EngineInterface* engine, u32 subchannel_id) { + void BindSubchannel(Engines::EngineInterface* engine, u32 subchannel_id) { subchannels[subchannel_id] = engine; } @@ -145,7 +145,7 @@ private: bool ib_enable{true}; ///< IB mode enabled - std::array subchannels{}; + std::array subchannels{}; GPU& gpu; Core::System& system; diff --git a/src/video_core/engines/engine_upload.cpp b/src/video_core/engines/engine_upload.cpp index d44ad0cd8..71d7e1473 100644 --- a/src/video_core/engines/engine_upload.cpp +++ b/src/video_core/engines/engine_upload.cpp @@ -11,16 +11,16 @@ namespace Tegra::Engines::Upload { -State::State(MemoryManager& memory_manager, Registers& regs) - : regs{regs}, memory_manager{memory_manager} {} +State::State(MemoryManager& memory_manager_, Registers& regs_) + : regs{regs_}, memory_manager{memory_manager_} {} State::~State() = default; -void State::ProcessExec(const bool is_linear) { +void State::ProcessExec(const bool is_linear_) { write_offset = 0; copy_size = regs.line_length_in * regs.line_count; inner_buffer.resize(copy_size); - this->is_linear = is_linear; + is_linear = is_linear_; } void State::ProcessData(const u32 data, const bool is_last_call) { diff --git a/src/video_core/engines/engine_upload.h b/src/video_core/engines/engine_upload.h index 462da419e..1c7f1effa 100644 --- a/src/video_core/engines/engine_upload.h +++ b/src/video_core/engines/engine_upload.h @@ -54,10 +54,10 @@ struct Registers { class State { public: - State(MemoryManager& memory_manager, Registers& regs); + explicit State(MemoryManager& memory_manager_, Registers& regs_); ~State(); - void ProcessExec(bool is_linear); + void ProcessExec(bool is_linear_); void ProcessData(u32 data, bool is_last_call); private: diff --git a/src/video_core/engines/kepler_memory.cpp b/src/video_core/engines/kepler_memory.cpp index dc71b2eec..9911140e9 100644 --- a/src/video_core/engines/kepler_memory.cpp +++ b/src/video_core/engines/kepler_memory.cpp @@ -14,8 +14,8 @@ namespace Tegra::Engines { -KeplerMemory::KeplerMemory(Core::System& system, MemoryManager& memory_manager) - : system{system}, upload_state{memory_manager, regs.upload} {} +KeplerMemory::KeplerMemory(Core::System& system_, MemoryManager& memory_manager) + : system{system_}, upload_state{memory_manager, regs.upload} {} KeplerMemory::~KeplerMemory() = default; diff --git a/src/video_core/engines/kepler_memory.h b/src/video_core/engines/kepler_memory.h index 5b7f71a00..62483589e 100644 --- a/src/video_core/engines/kepler_memory.h +++ b/src/video_core/engines/kepler_memory.h @@ -35,7 +35,7 @@ namespace Tegra::Engines { class KeplerMemory final : public EngineInterface { public: - KeplerMemory(Core::System& system, MemoryManager& memory_manager); + explicit KeplerMemory(Core::System& system_, MemoryManager& memory_manager); ~KeplerMemory(); /// Write the value to the register identified by method. diff --git a/src/video_core/engines/maxwell_dma.cpp b/src/video_core/engines/maxwell_dma.cpp index 8fa359d0a..1c29e895e 100644 --- a/src/video_core/engines/maxwell_dma.cpp +++ b/src/video_core/engines/maxwell_dma.cpp @@ -16,8 +16,10 @@ namespace Tegra::Engines { using namespace Texture; -MaxwellDMA::MaxwellDMA(Core::System& system, MemoryManager& memory_manager) - : system{system}, memory_manager{memory_manager} {} +MaxwellDMA::MaxwellDMA(Core::System& system_, MemoryManager& memory_manager_) + : system{system_}, memory_manager{memory_manager_} {} + +MaxwellDMA::~MaxwellDMA() = default; void MaxwellDMA::CallMethod(u32 method, u32 method_argument, bool is_last_call) { ASSERT_MSG(method < NUM_REGS, "Invalid MaxwellDMA register"); diff --git a/src/video_core/engines/maxwell_dma.h b/src/video_core/engines/maxwell_dma.h index 50f445efc..17bd280c4 100644 --- a/src/video_core/engines/maxwell_dma.h +++ b/src/video_core/engines/maxwell_dma.h @@ -185,8 +185,8 @@ public: }; static_assert(sizeof(RemapConst) == 12); - explicit MaxwellDMA(Core::System& system, MemoryManager& memory_manager); - ~MaxwellDMA() = default; + explicit MaxwellDMA(Core::System& system_, MemoryManager& memory_manager_); + ~MaxwellDMA(); /// Write the value to the register identified by method. void CallMethod(u32 method, u32 method_argument, bool is_last_call) override; diff --git a/src/video_core/fence_manager.h b/src/video_core/fence_manager.h index de6991ef6..c5f26896e 100644 --- a/src/video_core/fence_manager.h +++ b/src/video_core/fence_manager.h @@ -17,11 +17,11 @@ namespace VideoCommon { class FenceBase { public: - FenceBase(u32 payload, bool is_stubbed) - : address{}, payload{payload}, is_semaphore{false}, is_stubbed{is_stubbed} {} + explicit FenceBase(u32 payload_, bool is_stubbed_) + : address{}, payload{payload_}, is_semaphore{false}, is_stubbed{is_stubbed_} {} - FenceBase(GPUVAddr address, u32 payload, bool is_stubbed) - : address{address}, payload{payload}, is_semaphore{true}, is_stubbed{is_stubbed} {} + explicit FenceBase(GPUVAddr address_, u32 payload_, bool is_stubbed_) + : address{address_}, payload{payload_}, is_semaphore{true}, is_stubbed{is_stubbed_} {} GPUVAddr GetAddress() const { return address; diff --git a/src/video_core/gpu.cpp b/src/video_core/gpu.cpp index e91f52938..964b3f3dc 100644 --- a/src/video_core/gpu.cpp +++ b/src/video_core/gpu.cpp @@ -232,8 +232,12 @@ void GPU::CallMultiMethod(u32 method, u32 subchannel, const u32* base_start, u32 CallEngineMultiMethod(method, subchannel, base_start, amount, methods_pending); } else { for (std::size_t i = 0; i < amount; i++) { - CallPullerMethod( - {method, base_start[i], subchannel, methods_pending - static_cast(i)}); + CallPullerMethod(MethodCall{ + method, + base_start[i], + subchannel, + methods_pending - static_cast(i), + }); } } } diff --git a/src/video_core/gpu.h b/src/video_core/gpu.h index 21410e125..660641d04 100644 --- a/src/video_core/gpu.h +++ b/src/video_core/gpu.h @@ -149,16 +149,16 @@ public: u32 subchannel{}; u32 method_count{}; - MethodCall(u32 method, u32 argument, u32 subchannel = 0, u32 method_count = 0) - : method(method), argument(argument), subchannel(subchannel), - method_count(method_count) {} + explicit MethodCall(u32 method_, u32 argument_, u32 subchannel_ = 0, u32 method_count_ = 0) + : method(method_), argument(argument_), subchannel(subchannel_), + method_count(method_count_) {} [[nodiscard]] bool IsLastCall() const { return method_count <= 1; } }; - explicit GPU(Core::System& system, bool is_async, bool use_nvdec); + explicit GPU(Core::System& system_, bool is_async_, bool use_nvdec_); virtual ~GPU(); /// Binds a renderer to the GPU. @@ -414,8 +414,8 @@ private: std::condition_variable sync_cv; struct FlushRequest { - FlushRequest(u64 fence, VAddr addr, std::size_t size) - : fence{fence}, addr{addr}, size{size} {} + explicit FlushRequest(u64 fence_, VAddr addr_, std::size_t size_) + : fence{fence_}, addr{addr_}, size{size_} {} u64 fence; VAddr addr; std::size_t size; diff --git a/src/video_core/gpu_asynch.cpp b/src/video_core/gpu_asynch.cpp index a9baaf7ef..6cc091ecd 100644 --- a/src/video_core/gpu_asynch.cpp +++ b/src/video_core/gpu_asynch.cpp @@ -10,8 +10,8 @@ namespace VideoCommon { -GPUAsynch::GPUAsynch(Core::System& system, bool use_nvdec) - : GPU{system, true, use_nvdec}, gpu_thread{system} {} +GPUAsynch::GPUAsynch(Core::System& system_, bool use_nvdec_) + : GPU{system_, true, use_nvdec_}, gpu_thread{system_} {} GPUAsynch::~GPUAsynch() = default; diff --git a/src/video_core/gpu_asynch.h b/src/video_core/gpu_asynch.h index 0c0872e73..a384113f4 100644 --- a/src/video_core/gpu_asynch.h +++ b/src/video_core/gpu_asynch.h @@ -20,7 +20,7 @@ namespace VideoCommon { /// Implementation of GPU interface that runs the GPU asynchronously class GPUAsynch final : public Tegra::GPU { public: - explicit GPUAsynch(Core::System& system, bool use_nvdec); + explicit GPUAsynch(Core::System& system_, bool use_nvdec_); ~GPUAsynch() override; void Start() override; diff --git a/src/video_core/gpu_synch.cpp b/src/video_core/gpu_synch.cpp index ecf7bbdf3..1e9d4b9b2 100644 --- a/src/video_core/gpu_synch.cpp +++ b/src/video_core/gpu_synch.cpp @@ -7,7 +7,7 @@ namespace VideoCommon { -GPUSynch::GPUSynch(Core::System& system, bool use_nvdec) : GPU{system, false, use_nvdec} {} +GPUSynch::GPUSynch(Core::System& system_, bool use_nvdec_) : GPU{system_, false, use_nvdec_} {} GPUSynch::~GPUSynch() = default; diff --git a/src/video_core/gpu_synch.h b/src/video_core/gpu_synch.h index 9d778c71a..c5904b8db 100644 --- a/src/video_core/gpu_synch.h +++ b/src/video_core/gpu_synch.h @@ -19,7 +19,7 @@ namespace VideoCommon { /// Implementation of GPU interface that runs the GPU synchronously class GPUSynch final : public Tegra::GPU { public: - explicit GPUSynch(Core::System& system, bool use_nvdec); + explicit GPUSynch(Core::System& system_, bool use_nvdec_); ~GPUSynch() override; void Start() override; diff --git a/src/video_core/gpu_thread.cpp b/src/video_core/gpu_thread.cpp index 4b8f58283..e27218b96 100644 --- a/src/video_core/gpu_thread.cpp +++ b/src/video_core/gpu_thread.cpp @@ -39,23 +39,23 @@ static void RunThread(Core::System& system, VideoCore::RendererBase& renderer, CommandDataContainer next; while (state.is_running) { next = state.queue.PopWait(); - if (const auto submit_list = std::get_if(&next.data)) { + if (auto* submit_list = std::get_if(&next.data)) { dma_pusher.Push(std::move(submit_list->entries)); dma_pusher.DispatchCalls(); - } else if (const auto command_list = std::get_if(&next.data)) { + } else if (auto* command_list = std::get_if(&next.data)) { // NVDEC cdma_pusher.Push(std::move(command_list->entries)); cdma_pusher.DispatchCalls(); - } else if (const auto data = std::get_if(&next.data)) { + } else if (const auto* data = std::get_if(&next.data)) { renderer.SwapBuffers(data->framebuffer ? &*data->framebuffer : nullptr); } else if (std::holds_alternative(next.data)) { renderer.Rasterizer().ReleaseFences(); } else if (std::holds_alternative(next.data)) { system.GPU().TickWork(); - } else if (const auto data = std::get_if(&next.data)) { - renderer.Rasterizer().FlushRegion(data->addr, data->size); - } else if (const auto data = std::get_if(&next.data)) { - renderer.Rasterizer().OnCPUWrite(data->addr, data->size); + } else if (const auto* flush = std::get_if(&next.data)) { + renderer.Rasterizer().FlushRegion(flush->addr, flush->size); + } else if (const auto* invalidate = std::get_if(&next.data)) { + renderer.Rasterizer().OnCPUWrite(invalidate->addr, invalidate->size); } else if (std::holds_alternative(next.data)) { return; } else { @@ -65,7 +65,7 @@ static void RunThread(Core::System& system, VideoCore::RendererBase& renderer, } } -ThreadManager::ThreadManager(Core::System& system) : system{system} {} +ThreadManager::ThreadManager(Core::System& system_) : system{system_} {} ThreadManager::~ThreadManager() { if (!thread.joinable()) { diff --git a/src/video_core/gpu_thread.h b/src/video_core/gpu_thread.h index 32a34e3a7..f1c52cd9e 100644 --- a/src/video_core/gpu_thread.h +++ b/src/video_core/gpu_thread.h @@ -32,30 +32,30 @@ struct EndProcessingCommand final {}; /// Command to signal to the GPU thread that a command list is ready for processing struct SubmitListCommand final { - explicit SubmitListCommand(Tegra::CommandList&& entries) : entries{std::move(entries)} {} + explicit SubmitListCommand(Tegra::CommandList&& entries_) : entries{std::move(entries_)} {} Tegra::CommandList entries; }; /// Command to signal to the GPU thread that a cdma command list is ready for processing struct SubmitChCommandEntries final { - explicit SubmitChCommandEntries(Tegra::ChCommandHeaderList&& entries) - : entries{std::move(entries)} {} + explicit SubmitChCommandEntries(Tegra::ChCommandHeaderList&& entries_) + : entries{std::move(entries_)} {} Tegra::ChCommandHeaderList entries; }; /// Command to signal to the GPU thread that a swap buffers is pending struct SwapBuffersCommand final { - explicit SwapBuffersCommand(std::optional framebuffer) - : framebuffer{std::move(framebuffer)} {} + explicit SwapBuffersCommand(std::optional framebuffer_) + : framebuffer{std::move(framebuffer_)} {} std::optional framebuffer; }; /// Command to signal to the GPU thread to flush a region struct FlushRegionCommand final { - explicit constexpr FlushRegionCommand(VAddr addr, u64 size) : addr{addr}, size{size} {} + explicit constexpr FlushRegionCommand(VAddr addr_, u64 size_) : addr{addr_}, size{size_} {} VAddr addr; u64 size; @@ -63,7 +63,7 @@ struct FlushRegionCommand final { /// Command to signal to the GPU thread to invalidate a region struct InvalidateRegionCommand final { - explicit constexpr InvalidateRegionCommand(VAddr addr, u64 size) : addr{addr}, size{size} {} + explicit constexpr InvalidateRegionCommand(VAddr addr_, u64 size_) : addr{addr_}, size{size_} {} VAddr addr; u64 size; @@ -71,8 +71,8 @@ struct InvalidateRegionCommand final { /// Command to signal to the GPU thread to flush and invalidate a region struct FlushAndInvalidateRegionCommand final { - explicit constexpr FlushAndInvalidateRegionCommand(VAddr addr, u64 size) - : addr{addr}, size{size} {} + explicit constexpr FlushAndInvalidateRegionCommand(VAddr addr_, u64 size_) + : addr{addr_}, size{size_} {} VAddr addr; u64 size; @@ -92,8 +92,8 @@ using CommandData = struct CommandDataContainer { CommandDataContainer() = default; - CommandDataContainer(CommandData&& data, u64 next_fence) - : data{std::move(data)}, fence{next_fence} {} + explicit CommandDataContainer(CommandData&& data_, u64 next_fence_) + : data{std::move(data_)}, fence{next_fence_} {} CommandData data; u64 fence{}; @@ -112,7 +112,7 @@ struct SynchState final { /// Class used to manage the GPU thread class ThreadManager final { public: - explicit ThreadManager(Core::System& system); + explicit ThreadManager(Core::System& system_); ~ThreadManager(); /// Creates and starts the GPU thread. @@ -146,7 +146,6 @@ private: /// Pushes a command to be executed by the GPU thread u64 PushCommand(CommandData&& command_data); -private: SynchState state; Core::System& system; std::thread thread; diff --git a/src/video_core/guest_driver.h b/src/video_core/guest_driver.h index 99450777e..21e569ba1 100644 --- a/src/video_core/guest_driver.h +++ b/src/video_core/guest_driver.h @@ -19,8 +19,8 @@ namespace VideoCore { class GuestDriverProfile { public: explicit GuestDriverProfile() = default; - explicit GuestDriverProfile(std::optional texture_handler_size) - : texture_handler_size{texture_handler_size} {} + explicit GuestDriverProfile(std::optional texture_handler_size_) + : texture_handler_size{texture_handler_size_} {} void DeduceTextureHandlerSize(std::vector bound_offsets); diff --git a/src/video_core/macro/macro_hle.cpp b/src/video_core/macro/macro_hle.cpp index df00b57df..70ac7c620 100644 --- a/src/video_core/macro/macro_hle.cpp +++ b/src/video_core/macro/macro_hle.cpp @@ -85,7 +85,7 @@ constexpr std::array, 3> hle_funcs{{ {0x0217920100488FF7, &HLE_0217920100488FF7}, }}; -HLEMacro::HLEMacro(Engines::Maxwell3D& maxwell3d) : maxwell3d(maxwell3d) {} +HLEMacro::HLEMacro(Engines::Maxwell3D& maxwell3d_) : maxwell3d{maxwell3d_} {} HLEMacro::~HLEMacro() = default; std::optional> HLEMacro::GetHLEProgram(u64 hash) const { @@ -99,8 +99,8 @@ std::optional> HLEMacro::GetHLEProgram(u64 hash) co HLEMacroImpl::~HLEMacroImpl() = default; -HLEMacroImpl::HLEMacroImpl(Engines::Maxwell3D& maxwell3d, HLEFunction func) - : maxwell3d(maxwell3d), func(func) {} +HLEMacroImpl::HLEMacroImpl(Engines::Maxwell3D& maxwell3d_, HLEFunction func_) + : maxwell3d{maxwell3d_}, func{func_} {} void HLEMacroImpl::Execute(const std::vector& parameters, u32 method) { func(maxwell3d, parameters); diff --git a/src/video_core/macro/macro_hle.h b/src/video_core/macro/macro_hle.h index 37af875a0..cb3bd1600 100644 --- a/src/video_core/macro/macro_hle.h +++ b/src/video_core/macro/macro_hle.h @@ -20,7 +20,7 @@ using HLEFunction = void (*)(Engines::Maxwell3D& maxwell3d, const std::vector> GetHLEProgram(u64 hash) const; diff --git a/src/video_core/macro/macro_interpreter.cpp b/src/video_core/macro/macro_interpreter.cpp index bd01fd1f2..44a71aa6c 100644 --- a/src/video_core/macro/macro_interpreter.cpp +++ b/src/video_core/macro/macro_interpreter.cpp @@ -11,29 +11,29 @@ MICROPROFILE_DEFINE(MacroInterp, "GPU", "Execute macro interpreter", MP_RGB(128, 128, 192)); namespace Tegra { -MacroInterpreter::MacroInterpreter(Engines::Maxwell3D& maxwell3d) - : MacroEngine::MacroEngine(maxwell3d), maxwell3d(maxwell3d) {} +MacroInterpreter::MacroInterpreter(Engines::Maxwell3D& maxwell3d_) + : MacroEngine{maxwell3d_}, maxwell3d{maxwell3d_} {} std::unique_ptr MacroInterpreter::Compile(const std::vector& code) { return std::make_unique(maxwell3d, code); } -MacroInterpreterImpl::MacroInterpreterImpl(Engines::Maxwell3D& maxwell3d, - const std::vector& code) - : maxwell3d(maxwell3d), code(code) {} +MacroInterpreterImpl::MacroInterpreterImpl(Engines::Maxwell3D& maxwell3d_, + const std::vector& code_) + : maxwell3d{maxwell3d_}, code{code_} {} -void MacroInterpreterImpl::Execute(const std::vector& parameters, u32 method) { +void MacroInterpreterImpl::Execute(const std::vector& params, u32 method) { MICROPROFILE_SCOPE(MacroInterp); Reset(); - registers[1] = parameters[0]; - num_parameters = parameters.size(); + registers[1] = params[0]; + num_parameters = params.size(); if (num_parameters > parameters_capacity) { parameters_capacity = num_parameters; - this->parameters = std::make_unique(num_parameters); + parameters = std::make_unique(num_parameters); } - std::memcpy(this->parameters.get(), parameters.data(), num_parameters * sizeof(u32)); + std::memcpy(parameters.get(), params.data(), num_parameters * sizeof(u32)); // Execute the code until we hit an exit condition. bool keep_executing = true; diff --git a/src/video_core/macro/macro_interpreter.h b/src/video_core/macro/macro_interpreter.h index 90217fc89..d50c619ce 100644 --- a/src/video_core/macro/macro_interpreter.h +++ b/src/video_core/macro/macro_interpreter.h @@ -17,7 +17,7 @@ class Maxwell3D; class MacroInterpreter final : public MacroEngine { public: - explicit MacroInterpreter(Engines::Maxwell3D& maxwell3d); + explicit MacroInterpreter(Engines::Maxwell3D& maxwell3d_); protected: std::unique_ptr Compile(const std::vector& code) override; @@ -28,8 +28,8 @@ private: class MacroInterpreterImpl : public CachedMacro { public: - MacroInterpreterImpl(Engines::Maxwell3D& maxwell3d, const std::vector& code); - void Execute(const std::vector& parameters, u32 method) override; + explicit MacroInterpreterImpl(Engines::Maxwell3D& maxwell3d_, const std::vector& code_); + void Execute(const std::vector& params, u32 method) override; private: /// Resets the execution engine state, zeroing registers, etc. @@ -38,9 +38,9 @@ private: /** * Executes a single macro instruction located at the current program counter. Returns whether * the interpreter should keep running. - * @param offset Offset to start execution at. + * * @param is_delay_slot Whether the current step is being executed due to a delay slot in a - * previous instruction. + * previous instruction. */ bool Step(bool is_delay_slot); diff --git a/src/video_core/macro/macro_jit_x64.cpp b/src/video_core/macro/macro_jit_x64.cpp index 954b87515..c82bb987f 100644 --- a/src/video_core/macro/macro_jit_x64.cpp +++ b/src/video_core/macro/macro_jit_x64.cpp @@ -28,15 +28,15 @@ static const std::bitset<32> PERSISTENT_REGISTERS = Common::X64::BuildRegSet({ BRANCH_HOLDER, }); -MacroJITx64::MacroJITx64(Engines::Maxwell3D& maxwell3d) - : MacroEngine::MacroEngine(maxwell3d), maxwell3d(maxwell3d) {} +MacroJITx64::MacroJITx64(Engines::Maxwell3D& maxwell3d_) + : MacroEngine{maxwell3d_}, maxwell3d{maxwell3d_} {} std::unique_ptr MacroJITx64::Compile(const std::vector& code) { return std::make_unique(maxwell3d, code); } -MacroJITx64Impl::MacroJITx64Impl(Engines::Maxwell3D& maxwell3d, const std::vector& code) - : Xbyak::CodeGenerator(MAX_CODE_SIZE), code(code), maxwell3d(maxwell3d) { +MacroJITx64Impl::MacroJITx64Impl(Engines::Maxwell3D& maxwell3d_, const std::vector& code_) + : CodeGenerator{MAX_CODE_SIZE}, code{code_}, maxwell3d{maxwell3d_} { Compile(); } @@ -553,15 +553,15 @@ Xbyak::Reg32 MacroJITx64Impl::Compile_GetRegister(u32 index, Xbyak::Reg32 dst) { } void MacroJITx64Impl::Compile_ProcessResult(Macro::ResultOperation operation, u32 reg) { - const auto SetRegister = [this](u32 reg, const Xbyak::Reg32& result) { + const auto SetRegister = [this](u32 reg_index, const Xbyak::Reg32& result) { // Register 0 is supposed to always return 0. NOP is implemented as a store to the zero // register. - if (reg == 0) { + if (reg_index == 0) { return; } - mov(dword[STATE + offsetof(JITState, registers) + reg * sizeof(u32)], result); + mov(dword[STATE + offsetof(JITState, registers) + reg_index * sizeof(u32)], result); }; - const auto SetMethodAddress = [this](const Xbyak::Reg32& reg) { mov(METHOD_ADDRESS, reg); }; + const auto SetMethodAddress = [this](const Xbyak::Reg32& reg32) { mov(METHOD_ADDRESS, reg32); }; switch (operation) { case Macro::ResultOperation::IgnoreAndFetch: diff --git a/src/video_core/macro/macro_jit_x64.h b/src/video_core/macro/macro_jit_x64.h index a180e7428..7f50ac2f8 100644 --- a/src/video_core/macro/macro_jit_x64.h +++ b/src/video_core/macro/macro_jit_x64.h @@ -23,7 +23,7 @@ constexpr size_t MAX_CODE_SIZE = 0x10000; class MacroJITx64 final : public MacroEngine { public: - explicit MacroJITx64(Engines::Maxwell3D& maxwell3d); + explicit MacroJITx64(Engines::Maxwell3D& maxwell3d_); protected: std::unique_ptr Compile(const std::vector& code) override; @@ -34,7 +34,7 @@ private: class MacroJITx64Impl : public Xbyak::CodeGenerator, public CachedMacro { public: - MacroJITx64Impl(Engines::Maxwell3D& maxwell3d, const std::vector& code); + explicit MacroJITx64Impl(Engines::Maxwell3D& maxwell3d_, const std::vector& code_); ~MacroJITx64Impl(); void Execute(const std::vector& parameters, u32 method) override; diff --git a/src/video_core/memory_manager.h b/src/video_core/memory_manager.h index c078193d9..c35e57689 100644 --- a/src/video_core/memory_manager.h +++ b/src/video_core/memory_manager.h @@ -28,7 +28,7 @@ public: }; constexpr PageEntry() = default; - constexpr PageEntry(State state) : state{state} {} + constexpr PageEntry(State state_) : state{state_} {} constexpr PageEntry(VAddr addr) : state{static_cast(addr >> ShiftBits)} {} [[nodiscard]] constexpr bool IsUnmapped() const { @@ -68,7 +68,7 @@ static_assert(sizeof(PageEntry) == 4, "PageEntry is too large"); class MemoryManager final { public: - explicit MemoryManager(Core::System& system); + explicit MemoryManager(Core::System& system_); ~MemoryManager(); /// Binds a renderer to the memory manager. diff --git a/src/video_core/renderer_opengl/gl_arb_decompiler.cpp b/src/video_core/renderer_opengl/gl_arb_decompiler.cpp index d6120c23e..5378c398e 100644 --- a/src/video_core/renderer_opengl/gl_arb_decompiler.cpp +++ b/src/video_core/renderer_opengl/gl_arb_decompiler.cpp @@ -187,8 +187,8 @@ std::string TextureType(const MetaTexture& meta) { class ARBDecompiler final { public: - explicit ARBDecompiler(const Device& device, const ShaderIR& ir, const Registry& registry, - ShaderType stage, std::string_view identifier); + explicit ARBDecompiler(const Device& device_, const ShaderIR& ir_, const Registry& registry_, + ShaderType stage_, std::string_view identifier); std::string Code() const { return shader_source; @@ -802,9 +802,9 @@ private: }; }; -ARBDecompiler::ARBDecompiler(const Device& device, const ShaderIR& ir, const Registry& registry, - ShaderType stage, std::string_view identifier) - : device{device}, ir{ir}, registry{registry}, stage{stage} { +ARBDecompiler::ARBDecompiler(const Device& device_, const ShaderIR& ir_, const Registry& registry_, + ShaderType stage_, std::string_view identifier) + : device{device_}, ir{ir_}, registry{registry_}, stage{stage_} { DefineGlobalMemory(); AddLine("TEMP RC;"); @@ -1134,44 +1134,44 @@ void ARBDecompiler::VisitAST(const ASTNode& node) { for (ASTNode current = ast->nodes.GetFirst(); current; current = current->GetNext()) { VisitAST(current); } - } else if (const auto ast = std::get_if(&*node->GetInnerData())) { - const std::string condition = VisitExpression(ast->condition); + } else if (const auto if_then = std::get_if(&*node->GetInnerData())) { + const std::string condition = VisitExpression(if_then->condition); ResetTemporaries(); AddLine("MOVC.U RC.x, {};", condition); AddLine("IF NE.x;"); - for (ASTNode current = ast->nodes.GetFirst(); current; current = current->GetNext()) { + for (ASTNode current = if_then->nodes.GetFirst(); current; current = current->GetNext()) { VisitAST(current); } AddLine("ENDIF;"); - } else if (const auto ast = std::get_if(&*node->GetInnerData())) { + } else if (const auto if_else = std::get_if(&*node->GetInnerData())) { AddLine("ELSE;"); - for (ASTNode current = ast->nodes.GetFirst(); current; current = current->GetNext()) { + for (ASTNode current = if_else->nodes.GetFirst(); current; current = current->GetNext()) { VisitAST(current); } - } else if (const auto ast = std::get_if(&*node->GetInnerData())) { - VisitBlock(ast->nodes); - } else if (const auto ast = std::get_if(&*node->GetInnerData())) { - AddLine("MOV.U F{}, {};", ast->index, VisitExpression(ast->condition)); + } else if (const auto decoded = std::get_if(&*node->GetInnerData())) { + VisitBlock(decoded->nodes); + } else if (const auto var_set = std::get_if(&*node->GetInnerData())) { + AddLine("MOV.U F{}, {};", var_set->index, VisitExpression(var_set->condition)); ResetTemporaries(); - } else if (const auto ast = std::get_if(&*node->GetInnerData())) { - const std::string condition = VisitExpression(ast->condition); + } else if (const auto do_while = std::get_if(&*node->GetInnerData())) { + const std::string condition = VisitExpression(do_while->condition); ResetTemporaries(); AddLine("REP;"); - for (ASTNode current = ast->nodes.GetFirst(); current; current = current->GetNext()) { + for (ASTNode current = do_while->nodes.GetFirst(); current; current = current->GetNext()) { VisitAST(current); } AddLine("MOVC.U RC.x, {};", condition); AddLine("BRK (NE.x);"); AddLine("ENDREP;"); - } else if (const auto ast = std::get_if(&*node->GetInnerData())) { - const bool is_true = ExprIsTrue(ast->condition); + } else if (const auto ast_return = std::get_if(&*node->GetInnerData())) { + const bool is_true = ExprIsTrue(ast_return->condition); if (!is_true) { - AddLine("MOVC.U RC.x, {};", VisitExpression(ast->condition)); + AddLine("MOVC.U RC.x, {};", VisitExpression(ast_return->condition)); AddLine("IF NE.x;"); ResetTemporaries(); } - if (ast->kills) { + if (ast_return->kills) { AddLine("KIL TR;"); } else { Exit(); @@ -1179,11 +1179,11 @@ void ARBDecompiler::VisitAST(const ASTNode& node) { if (!is_true) { AddLine("ENDIF;"); } - } else if (const auto ast = std::get_if(&*node->GetInnerData())) { - if (ExprIsTrue(ast->condition)) { + } else if (const auto ast_break = std::get_if(&*node->GetInnerData())) { + if (ExprIsTrue(ast_break->condition)) { AddLine("BRK;"); } else { - AddLine("MOVC.U RC.x, {};", VisitExpression(ast->condition)); + AddLine("MOVC.U RC.x, {};", VisitExpression(ast_break->condition)); AddLine("BRK (NE.x);"); ResetTemporaries(); } diff --git a/src/video_core/renderer_opengl/gl_fence_manager.cpp b/src/video_core/renderer_opengl/gl_fence_manager.cpp index b532fdcc2..6040646cb 100644 --- a/src/video_core/renderer_opengl/gl_fence_manager.cpp +++ b/src/video_core/renderer_opengl/gl_fence_manager.cpp @@ -11,10 +11,10 @@ namespace OpenGL { -GLInnerFence::GLInnerFence(u32 payload, bool is_stubbed) : FenceBase(payload, is_stubbed) {} +GLInnerFence::GLInnerFence(u32 payload_, bool is_stubbed_) : FenceBase{payload_, is_stubbed_} {} -GLInnerFence::GLInnerFence(GPUVAddr address, u32 payload, bool is_stubbed) - : FenceBase(address, payload, is_stubbed) {} +GLInnerFence::GLInnerFence(GPUVAddr address_, u32 payload_, bool is_stubbed_) + : FenceBase{address_, payload_, is_stubbed_} {} GLInnerFence::~GLInnerFence() = default; @@ -45,10 +45,10 @@ void GLInnerFence::Wait() { glClientWaitSync(sync_object.handle, 0, GL_TIMEOUT_IGNORED); } -FenceManagerOpenGL::FenceManagerOpenGL(VideoCore::RasterizerInterface& rasterizer, Tegra::GPU& gpu, - TextureCacheOpenGL& texture_cache, - OGLBufferCache& buffer_cache, QueryCache& query_cache) - : GenericFenceManager{rasterizer, gpu, texture_cache, buffer_cache, query_cache} {} +FenceManagerOpenGL::FenceManagerOpenGL(VideoCore::RasterizerInterface& rasterizer_, + Tegra::GPU& gpu_, TextureCacheOpenGL& texture_cache_, + OGLBufferCache& buffer_cache_, QueryCache& query_cache_) + : GenericFenceManager{rasterizer_, gpu_, texture_cache_, buffer_cache_, query_cache_} {} Fence FenceManagerOpenGL::CreateFence(u32 value, bool is_stubbed) { return std::make_shared(value, is_stubbed); diff --git a/src/video_core/renderer_opengl/gl_fence_manager.h b/src/video_core/renderer_opengl/gl_fence_manager.h index da1dcdace..39ca6125b 100644 --- a/src/video_core/renderer_opengl/gl_fence_manager.h +++ b/src/video_core/renderer_opengl/gl_fence_manager.h @@ -17,8 +17,8 @@ namespace OpenGL { class GLInnerFence : public VideoCommon::FenceBase { public: - GLInnerFence(u32 payload, bool is_stubbed); - GLInnerFence(GPUVAddr address, u32 payload, bool is_stubbed); + explicit GLInnerFence(u32 payload_, bool is_stubbed_); + explicit GLInnerFence(GPUVAddr address_, u32 payload_, bool is_stubbed_); ~GLInnerFence(); void Queue(); @@ -37,9 +37,9 @@ using GenericFenceManager = class FenceManagerOpenGL final : public GenericFenceManager { public: - explicit FenceManagerOpenGL(VideoCore::RasterizerInterface& rasterizer, Tegra::GPU& gpu, - TextureCacheOpenGL& texture_cache, OGLBufferCache& buffer_cache, - QueryCache& query_cache); + explicit FenceManagerOpenGL(VideoCore::RasterizerInterface& rasterizer_, Tegra::GPU& gpu_, + TextureCacheOpenGL& texture_cache_, OGLBufferCache& buffer_cache_, + QueryCache& query_cache_); protected: Fence CreateFence(u32 value, bool is_stubbed) override; diff --git a/src/video_core/renderer_opengl/gl_query_cache.cpp b/src/video_core/renderer_opengl/gl_query_cache.cpp index 1a3d9720e..bcc37471f 100644 --- a/src/video_core/renderer_opengl/gl_query_cache.cpp +++ b/src/video_core/renderer_opengl/gl_query_cache.cpp @@ -59,10 +59,10 @@ bool QueryCache::AnyCommandQueued() const noexcept { return gl_rasterizer.AnyCommandQueued(); } -HostCounter::HostCounter(QueryCache& cache, std::shared_ptr dependency, - VideoCore::QueryType type) - : VideoCommon::HostCounterBase{std::move(dependency)}, cache{cache}, - type{type}, query{cache.AllocateQuery(type)} { +HostCounter::HostCounter(QueryCache& cache_, std::shared_ptr dependency, + VideoCore::QueryType type_) + : HostCounterBase{std::move(dependency)}, cache{cache_}, type{type_}, + query{cache.AllocateQuery(type)} { glBeginQuery(GetTarget(type), query.handle); } @@ -86,13 +86,14 @@ u64 HostCounter::BlockingQuery() const { return static_cast(value); } -CachedQuery::CachedQuery(QueryCache& cache, VideoCore::QueryType type, VAddr cpu_addr, u8* host_ptr) - : VideoCommon::CachedQueryBase{cpu_addr, host_ptr}, cache{&cache}, type{type} {} +CachedQuery::CachedQuery(QueryCache& cache_, VideoCore::QueryType type_, VAddr cpu_addr, + u8* host_ptr) + : CachedQueryBase{cpu_addr, host_ptr}, cache{&cache_}, type{type_} {} CachedQuery::~CachedQuery() = default; CachedQuery::CachedQuery(CachedQuery&& rhs) noexcept - : VideoCommon::CachedQueryBase(std::move(rhs)), cache{rhs.cache}, type{rhs.type} {} + : CachedQueryBase(std::move(rhs)), cache{rhs.cache}, type{rhs.type} {} CachedQuery& CachedQuery::operator=(CachedQuery&& rhs) noexcept { cache = rhs.cache; diff --git a/src/video_core/renderer_opengl/gl_query_cache.h b/src/video_core/renderer_opengl/gl_query_cache.h index 82cac51ee..d9851e880 100644 --- a/src/video_core/renderer_opengl/gl_query_cache.h +++ b/src/video_core/renderer_opengl/gl_query_cache.h @@ -46,8 +46,8 @@ private: class HostCounter final : public VideoCommon::HostCounterBase { public: - explicit HostCounter(QueryCache& cache, std::shared_ptr dependency, - VideoCore::QueryType type); + explicit HostCounter(QueryCache& cache_, std::shared_ptr dependency, + VideoCore::QueryType type_); ~HostCounter(); void EndQuery(); @@ -62,7 +62,7 @@ private: class CachedQuery final : public VideoCommon::CachedQueryBase { public: - explicit CachedQuery(QueryCache& cache, VideoCore::QueryType type, VAddr cpu_addr, + explicit CachedQuery(QueryCache& cache_, VideoCore::QueryType type_, VAddr cpu_addr, u8* host_ptr); ~CachedQuery() override; diff --git a/src/video_core/renderer_opengl/gl_shader_cache.cpp b/src/video_core/renderer_opengl/gl_shader_cache.cpp index bd56bed0c..9f2c0a222 100644 --- a/src/video_core/renderer_opengl/gl_shader_cache.cpp +++ b/src/video_core/renderer_opengl/gl_shader_cache.cpp @@ -198,10 +198,10 @@ ProgramSharedPtr BuildShader(const Device& device, ShaderType shader_type, u64 u return program; } -Shader::Shader(std::shared_ptr registry_, ShaderEntries entries_, - ProgramSharedPtr program_, bool is_built) +Shader::Shader(std::shared_ptr registry_, ShaderEntries entries_, + ProgramSharedPtr program_, bool is_built_) : registry{std::move(registry_)}, entries{std::move(entries_)}, program{std::move(program_)}, - is_built(is_built) { + is_built{is_built_} { handle = program->assembly_program.handle; if (handle == 0) { handle = program->source_program.handle; diff --git a/src/video_core/renderer_opengl/gl_shader_cache.h b/src/video_core/renderer_opengl/gl_shader_cache.h index 1708af06a..ab5374fac 100644 --- a/src/video_core/renderer_opengl/gl_shader_cache.h +++ b/src/video_core/renderer_opengl/gl_shader_cache.h @@ -108,7 +108,7 @@ public: private: explicit Shader(std::shared_ptr registry, ShaderEntries entries, - ProgramSharedPtr program, bool is_built = true); + ProgramSharedPtr program, bool is_built_ = true); std::shared_ptr registry; ShaderEntries entries; diff --git a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp index 95ca96c8e..0940969ba 100644 --- a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp +++ b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp @@ -131,7 +131,7 @@ private: class Expression final { public: - Expression(std::string code, Type type) : code{std::move(code)}, type{type} { + Expression(std::string code_, Type type_) : code{std::move(code_)}, type{type_} { ASSERT(type != Type::Void); } Expression() : type{Type::Void} {} @@ -148,8 +148,8 @@ public: ASSERT(type == Type::Void); } - std::string As(Type type) const { - switch (type) { + std::string As(Type type_) const { + switch (type_) { case Type::Bool: return AsBool(); case Type::Bool2: @@ -418,11 +418,12 @@ struct GenericVaryingDescription { class GLSLDecompiler final { public: - explicit GLSLDecompiler(const Device& device, const ShaderIR& ir, const Registry& registry, - ShaderType stage, std::string_view identifier, std::string_view suffix) - : device{device}, ir{ir}, registry{registry}, stage{stage}, identifier{identifier}, - suffix{suffix}, header{ir.GetHeader()}, use_unified_uniforms{ - UseUnifiedUniforms(device, ir, stage)} { + explicit GLSLDecompiler(const Device& device_, const ShaderIR& ir_, const Registry& registry_, + ShaderType stage_, std::string_view identifier_, + std::string_view suffix_) + : device{device_}, ir{ir_}, registry{registry_}, stage{stage_}, identifier{identifier_}, + suffix{suffix_}, header{ir.GetHeader()}, use_unified_uniforms{ + UseUnifiedUniforms(device_, ir_, stage_)} { if (stage != ShaderType::Compute) { transform_feedback = BuildTransformFeedback(registry.GetGraphicsInfo()); } @@ -777,16 +778,16 @@ private: name = "gs_" + name + "[]"; } - std::string suffix; + std::string suffix_; if (stage == ShaderType::Fragment) { const auto input_mode{header.ps.GetPixelImap(location)}; if (input_mode == PixelImap::Unused) { return; } - suffix = GetInputFlags(input_mode); + suffix_ = GetInputFlags(input_mode); } - code.AddLine("layout (location = {}) {} in vec4 {};", location, suffix, name); + code.AddLine("layout (location = {}) {} in vec4 {};", location, suffix_, name); } void DeclareOutputAttributes() { @@ -2100,13 +2101,13 @@ private: const auto type = meta.sampler.is_shadow ? Type::Float : Type::Int; const bool separate_dc = meta.sampler.is_shadow; - std::vector ir; + std::vector ir_; if (meta.sampler.is_shadow) { - ir = {TextureOffset{}}; + ir_ = {TextureOffset{}}; } else { - ir = {TextureOffset{}, TextureArgument{type, meta.component}}; + ir_ = {TextureOffset{}, TextureArgument{type, meta.component}}; } - return {GenerateTexture(operation, "Gather", ir, separate_dc) + GetSwizzle(meta.element), + return {GenerateTexture(operation, "Gather", ir_, separate_dc) + GetSwizzle(meta.element), Type::Float}; } @@ -2801,7 +2802,7 @@ std::string GetFlowVariable(u32 index) { class ExprDecompiler { public: - explicit ExprDecompiler(GLSLDecompiler& decomp) : decomp{decomp} {} + explicit ExprDecompiler(GLSLDecompiler& decomp_) : decomp{decomp_} {} void operator()(const ExprAnd& expr) { inner += '('; @@ -2856,7 +2857,7 @@ private: class ASTDecompiler { public: - explicit ASTDecompiler(GLSLDecompiler& decomp) : decomp{decomp} {} + explicit ASTDecompiler(GLSLDecompiler& decomp_) : decomp{decomp_} {} void operator()(const ASTProgram& ast) { ASTNode current = ast.nodes.GetFirst(); diff --git a/src/video_core/renderer_opengl/gl_shader_decompiler.h b/src/video_core/renderer_opengl/gl_shader_decompiler.h index 451c9689a..f5a5249f2 100644 --- a/src/video_core/renderer_opengl/gl_shader_decompiler.h +++ b/src/video_core/renderer_opengl/gl_shader_decompiler.h @@ -25,8 +25,8 @@ using ImageEntry = VideoCommon::Shader::Image; class ConstBufferEntry : public VideoCommon::Shader::ConstBuffer { public: - explicit ConstBufferEntry(u32 max_offset, bool is_indirect, u32 index) - : VideoCommon::Shader::ConstBuffer{max_offset, is_indirect}, index{index} {} + explicit ConstBufferEntry(u32 max_offset, bool is_indirect, u32 index_) + : ConstBuffer{max_offset, is_indirect}, index{index_} {} u32 GetIndex() const { return index; @@ -37,10 +37,10 @@ private: }; struct GlobalMemoryEntry { - constexpr explicit GlobalMemoryEntry(u32 cbuf_index, u32 cbuf_offset, bool is_read, - bool is_written) - : cbuf_index{cbuf_index}, cbuf_offset{cbuf_offset}, is_read{is_read}, is_written{ - is_written} {} + constexpr explicit GlobalMemoryEntry(u32 cbuf_index_, u32 cbuf_offset_, bool is_read_, + bool is_written_) + : cbuf_index{cbuf_index_}, cbuf_offset{cbuf_offset_}, is_read{is_read_}, is_written{ + is_written_} {} u32 cbuf_index = 0; u32 cbuf_offset = 0; diff --git a/src/video_core/renderer_opengl/gl_texture_cache.cpp b/src/video_core/renderer_opengl/gl_texture_cache.cpp index a863ef218..a59fe853e 100644 --- a/src/video_core/renderer_opengl/gl_texture_cache.cpp +++ b/src/video_core/renderer_opengl/gl_texture_cache.cpp @@ -258,9 +258,9 @@ constexpr u32 EncodeSwizzle(SwizzleSource x_source, SwizzleSource y_source, Swiz } // Anonymous namespace -CachedSurface::CachedSurface(const GPUVAddr gpu_addr, const SurfaceParams& params, - bool is_astc_supported) - : VideoCommon::SurfaceBase(gpu_addr, params, is_astc_supported) { +CachedSurface::CachedSurface(const GPUVAddr gpu_addr_, const SurfaceParams& params_, + bool is_astc_supported_) + : SurfaceBase{gpu_addr_, params_, is_astc_supported_} { if (is_converted) { internal_format = params.srgb_conversion ? GL_SRGB8_ALPHA8 : GL_RGBA8; format = GL_RGBA; @@ -419,11 +419,11 @@ View CachedSurface::CreateViewInner(const ViewParams& view_key, const bool is_pr return view; } -CachedSurfaceView::CachedSurfaceView(CachedSurface& surface, const ViewParams& params, - bool is_proxy) - : VideoCommon::ViewBase(params), surface{surface}, format{surface.internal_format}, - target{GetTextureTarget(params.target)}, is_proxy{is_proxy} { - if (!is_proxy) { +CachedSurfaceView::CachedSurfaceView(CachedSurface& surface_, const ViewParams& params_, + bool is_proxy_) + : ViewBase{params_}, surface{surface_}, format{surface_.internal_format}, + target{GetTextureTarget(params_.target)}, is_proxy{is_proxy_} { + if (!is_proxy_) { main_view = CreateTextureView(); } } @@ -493,13 +493,13 @@ GLuint CachedSurfaceView::GetTexture(SwizzleSource x_source, SwizzleSource y_sou std::array swizzle{x_source, y_source, z_source, w_source}; - switch (const PixelFormat format = GetSurfaceParams().pixel_format) { + switch (const PixelFormat pixel_format = GetSurfaceParams().pixel_format) { case PixelFormat::D24_UNORM_S8_UINT: case PixelFormat::D32_FLOAT_S8_UINT: case PixelFormat::S8_UINT_D24_UNORM: UNIMPLEMENTED_IF(x_source != SwizzleSource::R && x_source != SwizzleSource::G); glTextureParameteri(view.handle, GL_DEPTH_STENCIL_TEXTURE_MODE, - GetComponent(format, x_source == SwizzleSource::R)); + GetComponent(pixel_format, x_source == SwizzleSource::R)); // Make sure we sample the first component std::transform(swizzle.begin(), swizzle.end(), swizzle.begin(), [](SwizzleSource value) { diff --git a/src/video_core/renderer_opengl/gl_texture_cache.h b/src/video_core/renderer_opengl/gl_texture_cache.h index 7787134fc..76a7b2316 100644 --- a/src/video_core/renderer_opengl/gl_texture_cache.h +++ b/src/video_core/renderer_opengl/gl_texture_cache.h @@ -37,7 +37,8 @@ class CachedSurface final : public VideoCommon::SurfaceBase { friend CachedSurfaceView; public: - explicit CachedSurface(GPUVAddr gpu_addr, const SurfaceParams& params, bool is_astc_supported); + explicit CachedSurface(GPUVAddr gpu_addr_, const SurfaceParams& params_, + bool is_astc_supported_); ~CachedSurface(); void UploadTexture(const std::vector& staging_buffer) override; @@ -77,7 +78,7 @@ private: class CachedSurfaceView final : public VideoCommon::ViewBase { public: - explicit CachedSurfaceView(CachedSurface& surface, const ViewParams& params, bool is_proxy); + explicit CachedSurfaceView(CachedSurface& surface_, const ViewParams& params_, bool is_proxy_); ~CachedSurfaceView(); /// @brief Attaches this texture view to the currently bound fb_target framebuffer diff --git a/src/video_core/renderer_vulkan/vk_command_pool.cpp b/src/video_core/renderer_vulkan/vk_command_pool.cpp index 6339f4fe0..256a39148 100644 --- a/src/video_core/renderer_vulkan/vk_command_pool.cpp +++ b/src/video_core/renderer_vulkan/vk_command_pool.cpp @@ -17,8 +17,8 @@ struct CommandPool::Pool { vk::CommandBuffers cmdbufs; }; -CommandPool::CommandPool(MasterSemaphore& master_semaphore, const VKDevice& device) - : ResourcePool(master_semaphore, COMMAND_BUFFER_POOL_SIZE), device{device} {} +CommandPool::CommandPool(MasterSemaphore& master_semaphore, const VKDevice& device_) + : ResourcePool(master_semaphore, COMMAND_BUFFER_POOL_SIZE), device{device_} {} CommandPool::~CommandPool() = default; diff --git a/src/video_core/renderer_vulkan/vk_command_pool.h b/src/video_core/renderer_vulkan/vk_command_pool.h index b9cb3fb5d..33655eca4 100644 --- a/src/video_core/renderer_vulkan/vk_command_pool.h +++ b/src/video_core/renderer_vulkan/vk_command_pool.h @@ -17,7 +17,7 @@ class VKDevice; class CommandPool final : public ResourcePool { public: - explicit CommandPool(MasterSemaphore& master_semaphore, const VKDevice& device); + explicit CommandPool(MasterSemaphore& master_semaphore, const VKDevice& device_); ~CommandPool() override; void Allocate(size_t begin, size_t end) override; -- cgit v1.2.3 From 2c27127d04a155fe0f893e84263d58f14473785d Mon Sep 17 00:00:00 2001 From: ameerj <52414509+ameerj@users.noreply.github.com> Date: Mon, 28 Dec 2020 01:02:06 -0500 Subject: nvdec syncpt incorporation laying the groundwork for async gpu, although this does not fully implement async nvdec operations --- .../hle/service/nvdrv/devices/nvhost_nvdec.cpp | 5 +++-- src/core/hle/service/nvdrv/devices/nvhost_nvdec.h | 3 ++- .../service/nvdrv/devices/nvhost_nvdec_common.cpp | 26 ++++++++++++++++------ .../service/nvdrv/devices/nvhost_nvdec_common.h | 14 ++++++++---- src/core/hle/service/nvdrv/devices/nvhost_vic.cpp | 5 +++-- src/core/hle/service/nvdrv/devices/nvhost_vic.h | 4 ++-- src/core/hle/service/nvdrv/nvdrv.cpp | 6 +++-- src/video_core/cdma_pusher.cpp | 15 ++++++------- src/video_core/cdma_pusher.h | 10 ++++----- src/video_core/command_classes/host1x.cpp | 6 +++-- src/video_core/command_classes/sync_manager.cpp | 2 +- 11 files changed, 59 insertions(+), 37 deletions(-) (limited to 'src/video_core/cdma_pusher.h') diff --git a/src/core/hle/service/nvdrv/devices/nvhost_nvdec.cpp b/src/core/hle/service/nvdrv/devices/nvhost_nvdec.cpp index d8735491c..36970f828 100644 --- a/src/core/hle/service/nvdrv/devices/nvhost_nvdec.cpp +++ b/src/core/hle/service/nvdrv/devices/nvhost_nvdec.cpp @@ -11,8 +11,9 @@ namespace Service::Nvidia::Devices { -nvhost_nvdec::nvhost_nvdec(Core::System& system, std::shared_ptr nvmap_dev) - : nvhost_nvdec_common(system, std::move(nvmap_dev)) {} +nvhost_nvdec::nvhost_nvdec(Core::System& system, std::shared_ptr nvmap_dev, + SyncpointManager& syncpoint_manager) + : nvhost_nvdec_common(system, std::move(nvmap_dev), syncpoint_manager) {} nvhost_nvdec::~nvhost_nvdec() = default; NvResult nvhost_nvdec::Ioctl1(Ioctl command, const std::vector& input, diff --git a/src/core/hle/service/nvdrv/devices/nvhost_nvdec.h b/src/core/hle/service/nvdrv/devices/nvhost_nvdec.h index 79b8b6de1..77ef53cdd 100644 --- a/src/core/hle/service/nvdrv/devices/nvhost_nvdec.h +++ b/src/core/hle/service/nvdrv/devices/nvhost_nvdec.h @@ -11,7 +11,8 @@ namespace Service::Nvidia::Devices { class nvhost_nvdec final : public nvhost_nvdec_common { public: - explicit nvhost_nvdec(Core::System& system, std::shared_ptr nvmap_dev); + explicit nvhost_nvdec(Core::System& system, std::shared_ptr nvmap_dev, + SyncpointManager& syncpoint_manager); ~nvhost_nvdec() override; NvResult Ioctl1(Ioctl command, const std::vector& input, std::vector& output) override; diff --git a/src/core/hle/service/nvdrv/devices/nvhost_nvdec_common.cpp b/src/core/hle/service/nvdrv/devices/nvhost_nvdec_common.cpp index b49cecb42..64370ad4c 100644 --- a/src/core/hle/service/nvdrv/devices/nvhost_nvdec_common.cpp +++ b/src/core/hle/service/nvdrv/devices/nvhost_nvdec_common.cpp @@ -11,6 +11,7 @@ #include "core/core.h" #include "core/hle/service/nvdrv/devices/nvhost_nvdec_common.h" #include "core/hle/service/nvdrv/devices/nvmap.h" +#include "core/hle/service/nvdrv/syncpoint_manager.h" #include "core/memory.h" #include "video_core/memory_manager.h" #include "video_core/renderer_base.h" @@ -36,8 +37,9 @@ std::size_t WriteVectors(std::vector& dst, const std::vector& src, std::s } } // Anonymous namespace -nvhost_nvdec_common::nvhost_nvdec_common(Core::System& system, std::shared_ptr nvmap_dev) - : nvdevice(system), nvmap_dev(std::move(nvmap_dev)) {} +nvhost_nvdec_common::nvhost_nvdec_common(Core::System& system, std::shared_ptr nvmap_dev, + SyncpointManager& syncpoint_manager) + : nvdevice(system), nvmap_dev(std::move(nvmap_dev)), syncpoint_manager(syncpoint_manager) {} nvhost_nvdec_common::~nvhost_nvdec_common() = default; NvResult nvhost_nvdec_common::SetNVMAPfd(const std::vector& input) { @@ -71,10 +73,14 @@ NvResult nvhost_nvdec_common::Submit(const std::vector& input, std::vectorGetObject(cmd_buffer.memory_id); ASSERT_OR_EXECUTE(object, return NvResult::InvalidState;); @@ -89,6 +95,10 @@ NvResult nvhost_nvdec_common::Submit(const std::vector& input, std::vector& input, std::vector& input, std::ve std::memcpy(¶ms, input.data(), sizeof(IoctlGetSyncpoint)); LOG_DEBUG(Service_NVDRV, "called GetSyncpoint, id={}", params.param); - // We found that implementing this causes deadlocks with async gpu, along with degraded - // performance. TODO: RE the nvdec async implementation - params.value = 0; + if (device_syncpoints[params.param] == 0) { + device_syncpoints[params.param] = syncpoint_manager.AllocateSyncpoint(); + } + params.value = device_syncpoints[params.param]; std::memcpy(output.data(), ¶ms, sizeof(IoctlGetSyncpoint)); return NvResult::Success; 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 d9f95ba58..4c9d4ba41 100644 --- a/src/core/hle/service/nvdrv/devices/nvhost_nvdec_common.h +++ b/src/core/hle/service/nvdrv/devices/nvhost_nvdec_common.h @@ -10,12 +10,16 @@ #include "common/swap.h" #include "core/hle/service/nvdrv/devices/nvdevice.h" -namespace Service::Nvidia::Devices { +namespace Service::Nvidia { +class SyncpointManager; + +namespace Devices { class nvmap; class nvhost_nvdec_common : public nvdevice { public: - explicit nvhost_nvdec_common(Core::System& system, std::shared_ptr nvmap_dev); + explicit nvhost_nvdec_common(Core::System& system, std::shared_ptr nvmap_dev, + SyncpointManager& syncpoint_manager); ~nvhost_nvdec_common() override; protected: @@ -157,8 +161,10 @@ protected: s32_le nvmap_fd{}; u32_le submit_timeout{}; std::shared_ptr nvmap_dev; - + SyncpointManager& syncpoint_manager; + std::array device_syncpoints{}; // This is expected to be ordered, therefore we must use a map, not unordered_map std::map buffer_mappings; }; -}; // namespace Service::Nvidia::Devices +}; // namespace Devices +} // namespace Service::Nvidia diff --git a/src/core/hle/service/nvdrv/devices/nvhost_vic.cpp b/src/core/hle/service/nvdrv/devices/nvhost_vic.cpp index 805fe86ae..72499654c 100644 --- a/src/core/hle/service/nvdrv/devices/nvhost_vic.cpp +++ b/src/core/hle/service/nvdrv/devices/nvhost_vic.cpp @@ -10,8 +10,9 @@ #include "video_core/renderer_base.h" namespace Service::Nvidia::Devices { -nvhost_vic::nvhost_vic(Core::System& system, std::shared_ptr nvmap_dev) - : nvhost_nvdec_common(system, std::move(nvmap_dev)) {} +nvhost_vic::nvhost_vic(Core::System& system, std::shared_ptr nvmap_dev, + SyncpointManager& syncpoint_manager) + : nvhost_nvdec_common(system, std::move(nvmap_dev), syncpoint_manager) {} nvhost_vic::~nvhost_vic() = default; diff --git a/src/core/hle/service/nvdrv/devices/nvhost_vic.h b/src/core/hle/service/nvdrv/devices/nvhost_vic.h index b2e11f4d4..f401c61fa 100644 --- a/src/core/hle/service/nvdrv/devices/nvhost_vic.h +++ b/src/core/hle/service/nvdrv/devices/nvhost_vic.h @@ -7,11 +7,11 @@ #include "core/hle/service/nvdrv/devices/nvhost_nvdec_common.h" namespace Service::Nvidia::Devices { -class nvmap; class nvhost_vic final : public nvhost_nvdec_common { public: - explicit nvhost_vic(Core::System& system, std::shared_ptr nvmap_dev); + explicit nvhost_vic(Core::System& system, std::shared_ptr nvmap_dev, + SyncpointManager& syncpoint_manager); ~nvhost_vic(); NvResult Ioctl1(Ioctl command, const std::vector& input, std::vector& output) override; diff --git a/src/core/hle/service/nvdrv/nvdrv.cpp b/src/core/hle/service/nvdrv/nvdrv.cpp index e03195afe..620c18728 100644 --- a/src/core/hle/service/nvdrv/nvdrv.cpp +++ b/src/core/hle/service/nvdrv/nvdrv.cpp @@ -55,9 +55,11 @@ Module::Module(Core::System& system) : syncpoint_manager{system.GPU()} { devices["/dev/nvdisp_disp0"] = std::make_shared(system, nvmap_dev); devices["/dev/nvhost-ctrl"] = std::make_shared(system, events_interface, syncpoint_manager); - devices["/dev/nvhost-nvdec"] = std::make_shared(system, nvmap_dev); + devices["/dev/nvhost-nvdec"] = + std::make_shared(system, nvmap_dev, syncpoint_manager); devices["/dev/nvhost-nvjpg"] = std::make_shared(system); - devices["/dev/nvhost-vic"] = std::make_shared(system, nvmap_dev); + devices["/dev/nvhost-vic"] = + std::make_shared(system, nvmap_dev, syncpoint_manager); } Module::~Module() = default; diff --git a/src/video_core/cdma_pusher.cpp b/src/video_core/cdma_pusher.cpp index e3e7432f7..94679d5d1 100644 --- a/src/video_core/cdma_pusher.cpp +++ b/src/video_core/cdma_pusher.cpp @@ -33,8 +33,7 @@ CDmaPusher::CDmaPusher(GPU& gpu_) : gpu{gpu_}, nvdec_processor(std::make_shared(gpu)), vic_processor(std::make_unique(gpu, nvdec_processor)), host1x_processor(std::make_unique(gpu)), - nvdec_sync(std::make_unique(gpu)), - vic_sync(std::make_unique(gpu)) {} + sync_manager(std::make_unique(gpu)) {} CDmaPusher::~CDmaPusher() = default; @@ -110,10 +109,10 @@ void CDmaPusher::ExecuteCommand(u32 state_offset, u32 data) { const auto syncpoint_id = static_cast(data & 0xFF); const auto cond = static_cast((data >> 8) & 0xFF); if (cond == 0) { - nvdec_sync->Increment(syncpoint_id); + sync_manager->Increment(syncpoint_id); } else { - nvdec_sync->IncrementWhenDone(static_cast(current_class), syncpoint_id); - nvdec_sync->SignalDone(syncpoint_id); + sync_manager->SignalDone( + sync_manager->IncrementWhenDone(static_cast(current_class), syncpoint_id)); } break; } @@ -135,10 +134,10 @@ void CDmaPusher::ExecuteCommand(u32 state_offset, u32 data) { const auto syncpoint_id = static_cast(data & 0xFF); const auto cond = static_cast((data >> 8) & 0xFF); if (cond == 0) { - vic_sync->Increment(syncpoint_id); + sync_manager->Increment(syncpoint_id); } else { - vic_sync->IncrementWhenDone(static_cast(current_class), syncpoint_id); - vic_sync->SignalDone(syncpoint_id); + sync_manager->SignalDone( + sync_manager->IncrementWhenDone(static_cast(current_class), syncpoint_id)); } break; } diff --git a/src/video_core/cdma_pusher.h b/src/video_core/cdma_pusher.h index 0db1cd646..8ca70b6dd 100644 --- a/src/video_core/cdma_pusher.h +++ b/src/video_core/cdma_pusher.h @@ -116,12 +116,10 @@ private: void ThiStateWrite(ThiRegisters& state, u32 state_offset, const std::vector& arguments); GPU& gpu; - - std::shared_ptr nvdec_processor; - std::unique_ptr vic_processor; - std::unique_ptr host1x_processor; - std::unique_ptr nvdec_sync; - std::unique_ptr vic_sync; + std::shared_ptr nvdec_processor; + std::unique_ptr vic_processor; + std::unique_ptr host1x_processor; + std::unique_ptr sync_manager; ChClassId current_class{}; ThiRegisters vic_thi_state{}; ThiRegisters nvdec_thi_state{}; diff --git a/src/video_core/command_classes/host1x.cpp b/src/video_core/command_classes/host1x.cpp index c4dd4881a..9d0a1b4d9 100644 --- a/src/video_core/command_classes/host1x.cpp +++ b/src/video_core/command_classes/host1x.cpp @@ -34,6 +34,8 @@ void Tegra::Host1x::ProcessMethod(Method method, const std::vector& argumen } void Tegra::Host1x::Execute(u32 data) { - // This method waits on a valid syncpoint. - // TODO: Implement when proper Async is in place + u32 syncpointId = (data & 0xFF); + u32 threshold = state.load_syncpoint_payload32; + + gpu.WaitFence(syncpointId, threshold); } diff --git a/src/video_core/command_classes/sync_manager.cpp b/src/video_core/command_classes/sync_manager.cpp index 19dc9e0ab..579857766 100644 --- a/src/video_core/command_classes/sync_manager.cpp +++ b/src/video_core/command_classes/sync_manager.cpp @@ -38,7 +38,7 @@ u32 SyncptIncrManager::IncrementWhenDone(u32 class_id, u32 id) { } void SyncptIncrManager::SignalDone(u32 handle) { - const auto done_incr = + const auto& done_incr = std::find_if(increments.begin(), increments.end(), [handle](const SyncptIncr& incr) { return incr.id == handle; }); if (done_incr != increments.cend()) { -- cgit v1.2.3