diff options
Diffstat (limited to 'src')
46 files changed, 1613 insertions, 1179 deletions
diff --git a/src/common/CMakeLists.txt b/src/common/CMakeLists.txt index 207c7a0a6..d20e6c3b5 100644 --- a/src/common/CMakeLists.txt +++ b/src/common/CMakeLists.txt @@ -102,6 +102,7 @@ add_library(common STATIC atomic_ops.h detached_tasks.cpp detached_tasks.h + bit_cast.h bit_field.h bit_util.h cityhash.cpp diff --git a/src/common/bit_cast.h b/src/common/bit_cast.h new file mode 100644 index 000000000..a32a063d1 --- /dev/null +++ b/src/common/bit_cast.h @@ -0,0 +1,22 @@ +// Copyright 2020 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <cstring> +#include <type_traits> + +namespace Common { + +template <typename To, typename From> +[[nodiscard]] std::enable_if_t<sizeof(To) == sizeof(From) && std::is_trivially_copyable_v<From> && + std::is_trivially_copyable_v<To>, + To> +BitCast(const From& src) noexcept { + To dst; + std::memcpy(&dst, &src, sizeof(To)); + return dst; +} + +} // namespace Common diff --git a/src/common/logging/backend.cpp b/src/common/logging/backend.cpp index 90dfa22ca..7859344b9 100644 --- a/src/common/logging/backend.cpp +++ b/src/common/logging/backend.cpp @@ -222,6 +222,7 @@ void DebuggerBackend::Write(const Entry& entry) { SUB(Service, NPNS) \ SUB(Service, NS) \ SUB(Service, NVDRV) \ + SUB(Service, OLSC) \ SUB(Service, PCIE) \ SUB(Service, PCTL) \ SUB(Service, PCV) \ diff --git a/src/common/logging/log.h b/src/common/logging/log.h index 13a4f1e30..835894918 100644 --- a/src/common/logging/log.h +++ b/src/common/logging/log.h @@ -95,6 +95,7 @@ enum class Class : ClassType { Service_NPNS, ///< The NPNS service Service_NS, ///< The NS services Service_NVDRV, ///< The NVDRV (Nvidia driver) service + Service_OLSC, ///< The OLSC service Service_PCIE, ///< The PCIe service Service_PCTL, ///< The PCTL (Parental control) service Service_PCV, ///< The PCV service diff --git a/src/common/page_table.cpp b/src/common/page_table.cpp index e5d3090d5..bccea0894 100644 --- a/src/common/page_table.cpp +++ b/src/common/page_table.cpp @@ -8,7 +8,7 @@ namespace Common { PageTable::PageTable() = default; -PageTable::~PageTable() = default; +PageTable::~PageTable() noexcept = default; void PageTable::Resize(std::size_t address_space_width_in_bits, std::size_t page_size_in_bits, bool has_attribute) { diff --git a/src/common/page_table.h b/src/common/page_table.h index cf5eed780..9754fabf9 100644 --- a/src/common/page_table.h +++ b/src/common/page_table.h @@ -4,9 +4,7 @@ #pragma once -#include <vector> - -#include <boost/icl/interval_map.hpp> +#include <tuple> #include "common/common_types.h" #include "common/memory_hook.h" @@ -51,13 +49,21 @@ struct SpecialRegion { */ struct PageTable { PageTable(); - ~PageTable(); + ~PageTable() noexcept; + + PageTable(const PageTable&) = delete; + PageTable& operator=(const PageTable&) = delete; + + PageTable(PageTable&&) noexcept = default; + PageTable& operator=(PageTable&&) noexcept = default; /** * Resizes the page table to be able to accomodate enough pages within * a given address space. * * @param address_space_width_in_bits The address size width in bits. + * @param page_size_in_bits The page size in bits. + * @param has_attribute Whether or not this page has any backing attributes. */ void Resize(std::size_t address_space_width_in_bits, std::size_t page_size_in_bits, bool has_attribute); diff --git a/src/common/virtual_buffer.cpp b/src/common/virtual_buffer.cpp index b009cb500..e3ca29258 100644 --- a/src/common/virtual_buffer.cpp +++ b/src/common/virtual_buffer.cpp @@ -13,7 +13,7 @@ namespace Common { -void* AllocateMemoryPages(std::size_t size) { +void* AllocateMemoryPages(std::size_t size) noexcept { #ifdef _WIN32 void* base{VirtualAlloc(nullptr, size, MEM_COMMIT, PAGE_READWRITE)}; #else @@ -29,7 +29,7 @@ void* AllocateMemoryPages(std::size_t size) { return base; } -void FreeMemoryPages(void* base, [[maybe_unused]] std::size_t size) { +void FreeMemoryPages(void* base, [[maybe_unused]] std::size_t size) noexcept { if (!base) { return; } diff --git a/src/common/virtual_buffer.h b/src/common/virtual_buffer.h index 125cb42f0..91d430036 100644 --- a/src/common/virtual_buffer.h +++ b/src/common/virtual_buffer.h @@ -4,29 +4,53 @@ #pragma once -#include "common/common_funcs.h" +#include <type_traits> +#include <utility> namespace Common { -void* AllocateMemoryPages(std::size_t size); -void FreeMemoryPages(void* base, std::size_t size); +void* AllocateMemoryPages(std::size_t size) noexcept; +void FreeMemoryPages(void* base, std::size_t size) noexcept; template <typename T> -class VirtualBuffer final : NonCopyable { +class VirtualBuffer final { public: + static_assert( + std::is_trivially_constructible_v<T>, + "T must be trivially constructible, as non-trivial constructors will not be executed " + "with the current allocator"); + constexpr VirtualBuffer() = default; explicit VirtualBuffer(std::size_t count) : alloc_size{count * sizeof(T)} { base_ptr = reinterpret_cast<T*>(AllocateMemoryPages(alloc_size)); } - ~VirtualBuffer() { + ~VirtualBuffer() noexcept { FreeMemoryPages(base_ptr, alloc_size); } + VirtualBuffer(const VirtualBuffer&) = delete; + VirtualBuffer& operator=(const VirtualBuffer&) = delete; + + VirtualBuffer(VirtualBuffer&& other) noexcept + : alloc_size{std::exchange(other.alloc_size, 0)}, base_ptr{std::exchange(other.base_ptr), + nullptr} {} + + VirtualBuffer& operator=(VirtualBuffer&& other) noexcept { + alloc_size = std::exchange(other.alloc_size, 0); + base_ptr = std::exchange(other.base_ptr, nullptr); + return *this; + } + void resize(std::size_t count) { + const auto new_size = count * sizeof(T); + if (new_size == alloc_size) { + return; + } + FreeMemoryPages(base_ptr, alloc_size); - alloc_size = count * sizeof(T); + alloc_size = new_size; base_ptr = reinterpret_cast<T*>(AllocateMemoryPages(alloc_size)); } diff --git a/src/core/CMakeLists.txt b/src/core/CMakeLists.txt index a1d8dcfa5..e370fd225 100644 --- a/src/core/CMakeLists.txt +++ b/src/core/CMakeLists.txt @@ -458,6 +458,8 @@ add_library(core STATIC hle/service/nvflinger/buffer_queue.h hle/service/nvflinger/nvflinger.cpp hle/service/nvflinger/nvflinger.h + hle/service/olsc/olsc.cpp + hle/service/olsc/olsc.h hle/service/pcie/pcie.cpp hle/service/pcie/pcie.h hle/service/pctl/module.cpp diff --git a/src/core/hle/service/nvdrv/devices/nvdevice.h b/src/core/hle/service/nvdrv/devices/nvdevice.h index 0240d6643..5681599ba 100644 --- a/src/core/hle/service/nvdrv/devices/nvdevice.h +++ b/src/core/hle/service/nvdrv/devices/nvdevice.h @@ -24,25 +24,37 @@ public: explicit nvdevice(Core::System& system) : system{system} {} virtual ~nvdevice() = default; - union Ioctl { - u32_le raw; - BitField<0, 8, u32> cmd; - BitField<8, 8, u32> group; - BitField<16, 14, u32> length; - BitField<30, 1, u32> is_in; - BitField<31, 1, u32> is_out; - }; + /** + * Handles an ioctl1 request. + * @param command The ioctl command id. + * @param input A buffer containing the input data for the ioctl. + * @param output A buffer where the output data will be written to. + * @returns The result code of the ioctl. + */ + virtual NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, + std::vector<u8>& output) = 0; + + /** + * Handles an ioctl2 request. + * @param command The ioctl command id. + * @param input A buffer containing the input data for the ioctl. + * @param inline_input A buffer containing the input data for the ioctl which has been inlined. + * @param output A buffer where the output data will be written to. + * @returns The result code of the ioctl. + */ + virtual NvResult Ioctl2(Ioctl command, const std::vector<u8>& input, + const std::vector<u8>& inline_input, std::vector<u8>& output) = 0; /** - * Handles an ioctl request. + * Handles an ioctl3 request. * @param command The ioctl command id. * @param input A buffer containing the input data for the ioctl. * @param output A buffer where the output data will be written to. + * @param inline_output A buffer where the inlined output data will be written to. * @returns The result code of the ioctl. */ - virtual u32 ioctl(Ioctl command, const std::vector<u8>& input, const std::vector<u8>& input2, - std::vector<u8>& output, std::vector<u8>& output2, IoctlCtrl& ctrl, - IoctlVersion version) = 0; + virtual NvResult Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output, + std::vector<u8>& inline_output) = 0; protected: Core::System& system; diff --git a/src/core/hle/service/nvdrv/devices/nvdisp_disp0.cpp b/src/core/hle/service/nvdrv/devices/nvdisp_disp0.cpp index 3f7b8e670..ce615c758 100644 --- a/src/core/hle/service/nvdrv/devices/nvdisp_disp0.cpp +++ b/src/core/hle/service/nvdrv/devices/nvdisp_disp0.cpp @@ -18,11 +18,22 @@ nvdisp_disp0::nvdisp_disp0(Core::System& system, std::shared_ptr<nvmap> nvmap_de : nvdevice(system), nvmap_dev(std::move(nvmap_dev)) {} nvdisp_disp0 ::~nvdisp_disp0() = default; -u32 nvdisp_disp0::ioctl(Ioctl command, const std::vector<u8>& input, const std::vector<u8>& input2, - std::vector<u8>& output, std::vector<u8>& output2, IoctlCtrl& ctrl, - IoctlVersion version) { - UNIMPLEMENTED_MSG("Unimplemented ioctl"); - return 0; +NvResult nvdisp_disp0::Ioctl1(Ioctl command, const std::vector<u8>& input, + std::vector<u8>& output) { + UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw); + return NvResult::NotImplemented; +} + +NvResult nvdisp_disp0::Ioctl2(Ioctl command, const std::vector<u8>& input, + const std::vector<u8>& inline_input, std::vector<u8>& output) { + UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw); + return NvResult::NotImplemented; +} + +NvResult nvdisp_disp0::Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output, + std::vector<u8>& inline_output) { + UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw); + return NvResult::NotImplemented; } void nvdisp_disp0::flip(u32 buffer_handle, u32 offset, u32 format, u32 width, u32 height, diff --git a/src/core/hle/service/nvdrv/devices/nvdisp_disp0.h b/src/core/hle/service/nvdrv/devices/nvdisp_disp0.h index 6fcdeee84..55a33b7e4 100644 --- a/src/core/hle/service/nvdrv/devices/nvdisp_disp0.h +++ b/src/core/hle/service/nvdrv/devices/nvdisp_disp0.h @@ -20,9 +20,11 @@ public: explicit nvdisp_disp0(Core::System& system, std::shared_ptr<nvmap> nvmap_dev); ~nvdisp_disp0() override; - u32 ioctl(Ioctl command, const std::vector<u8>& input, const std::vector<u8>& input2, - std::vector<u8>& output, std::vector<u8>& output2, IoctlCtrl& ctrl, - IoctlVersion version) override; + NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) override; + NvResult Ioctl2(Ioctl command, const std::vector<u8>& input, + const std::vector<u8>& inline_input, std::vector<u8>& output) override; + NvResult Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output, + std::vector<u8>& inline_output) override; /// Performs a screen flip, drawing the buffer pointed to by the handle. void flip(u32 buffer_handle, u32 offset, u32 format, u32 width, u32 height, u32 stride, diff --git a/src/core/hle/service/nvdrv/devices/nvhost_as_gpu.cpp b/src/core/hle/service/nvdrv/devices/nvhost_as_gpu.cpp index f2529a12e..6b062e10e 100644 --- a/src/core/hle/service/nvdrv/devices/nvhost_as_gpu.cpp +++ b/src/core/hle/service/nvdrv/devices/nvhost_as_gpu.cpp @@ -17,59 +17,77 @@ namespace Service::Nvidia::Devices { -namespace NvErrCodes { -constexpr u32 Success{}; -constexpr u32 OutOfMemory{static_cast<u32>(-12)}; -constexpr u32 InvalidInput{static_cast<u32>(-22)}; -} // namespace NvErrCodes - nvhost_as_gpu::nvhost_as_gpu(Core::System& system, std::shared_ptr<nvmap> nvmap_dev) : nvdevice(system), nvmap_dev(std::move(nvmap_dev)) {} nvhost_as_gpu::~nvhost_as_gpu() = default; -u32 nvhost_as_gpu::ioctl(Ioctl command, const std::vector<u8>& input, const std::vector<u8>& input2, - std::vector<u8>& output, std::vector<u8>& output2, IoctlCtrl& ctrl, - IoctlVersion version) { - LOG_DEBUG(Service_NVDRV, "called, command=0x{:08X}, input_size=0x{:X}, output_size=0x{:X}", - command.raw, input.size(), output.size()); - - switch (static_cast<IoctlCommand>(command.raw)) { - case IoctlCommand::IocInitalizeExCommand: - return InitalizeEx(input, output); - case IoctlCommand::IocAllocateSpaceCommand: - return AllocateSpace(input, output); - case IoctlCommand::IocMapBufferExCommand: - return MapBufferEx(input, output); - case IoctlCommand::IocBindChannelCommand: - return BindChannel(input, output); - case IoctlCommand::IocGetVaRegionsCommand: - return GetVARegions(input, output); - case IoctlCommand::IocUnmapBufferCommand: - return UnmapBuffer(input, output); - case IoctlCommand::IocFreeSpaceCommand: - return FreeSpace(input, output); +NvResult nvhost_as_gpu::Ioctl1(Ioctl command, const std::vector<u8>& input, + std::vector<u8>& output) { + switch (command.group) { + case 'A': + switch (command.cmd) { + case 0x1: + return BindChannel(input, output); + case 0x2: + return AllocateSpace(input, output); + case 0x3: + return FreeSpace(input, output); + case 0x5: + return UnmapBuffer(input, output); + case 0x6: + return MapBufferEx(input, output); + case 0x8: + return GetVARegions(input, output); + case 0x9: + return InitalizeEx(input, output); + case 0x14: + return Remap(input, output); + default: + break; + } + break; default: break; } - if (static_cast<IoctlCommand>(command.cmd.Value()) == IoctlCommand::IocRemapCommand) { - return Remap(input, output); - } + UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw); + return NvResult::NotImplemented; +} + +NvResult nvhost_as_gpu::Ioctl2(Ioctl command, const std::vector<u8>& input, + const std::vector<u8>& inline_input, std::vector<u8>& output) { + UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw); + return NvResult::NotImplemented; +} - UNIMPLEMENTED_MSG("Unimplemented ioctl command"); - return 0; +NvResult nvhost_as_gpu::Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output, + std::vector<u8>& inline_output) { + switch (command.group) { + case 'A': + switch (command.cmd) { + case 0x8: + return GetVARegions(input, output, inline_output); + default: + break; + } + break; + default: + break; + } + UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw); + return NvResult::NotImplemented; } -u32 nvhost_as_gpu::InitalizeEx(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_as_gpu::InitalizeEx(const std::vector<u8>& input, std::vector<u8>& output) { IoctlInitalizeEx params{}; std::memcpy(¶ms, input.data(), input.size()); LOG_WARNING(Service_NVDRV, "(STUBBED) called, big_page_size=0x{:X}", params.big_page_size); - return 0; + return NvResult::Success; } -u32 nvhost_as_gpu::AllocateSpace(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_as_gpu::AllocateSpace(const std::vector<u8>& input, std::vector<u8>& output) { IoctlAllocSpace params{}; std::memcpy(¶ms, input.data(), input.size()); @@ -83,17 +101,17 @@ u32 nvhost_as_gpu::AllocateSpace(const std::vector<u8>& input, std::vector<u8>& params.offset = system.GPU().MemoryManager().Allocate(size, params.align); } - auto result{NvErrCodes::Success}; + auto result = NvResult::Success; if (!params.offset) { LOG_CRITICAL(Service_NVDRV, "allocation failed for size {}", size); - result = NvErrCodes::OutOfMemory; + result = NvResult::InsufficientMemory; } std::memcpy(output.data(), ¶ms, output.size()); return result; } -u32 nvhost_as_gpu::FreeSpace(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_as_gpu::FreeSpace(const std::vector<u8>& input, std::vector<u8>& output) { IoctlFreeSpace params{}; std::memcpy(¶ms, input.data(), input.size()); @@ -104,15 +122,15 @@ u32 nvhost_as_gpu::FreeSpace(const std::vector<u8>& input, std::vector<u8>& outp static_cast<std::size_t>(params.pages) * params.page_size); std::memcpy(output.data(), ¶ms, output.size()); - return NvErrCodes::Success; + return NvResult::Success; } -u32 nvhost_as_gpu::Remap(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_as_gpu::Remap(const std::vector<u8>& input, std::vector<u8>& output) { const auto num_entries = input.size() / sizeof(IoctlRemapEntry); LOG_DEBUG(Service_NVDRV, "called, num_entries=0x{:X}", num_entries); - auto result{NvErrCodes::Success}; + auto result = NvResult::Success; std::vector<IoctlRemapEntry> entries(num_entries); std::memcpy(entries.data(), input.data(), input.size()); @@ -123,7 +141,7 @@ u32 nvhost_as_gpu::Remap(const std::vector<u8>& input, std::vector<u8>& output) const auto object{nvmap_dev->GetObject(entry.nvmap_handle)}; if (!object) { LOG_CRITICAL(Service_NVDRV, "invalid nvmap_handle={:X}", entry.nvmap_handle); - result = NvErrCodes::InvalidInput; + result = NvResult::InvalidState; break; } @@ -134,7 +152,7 @@ u32 nvhost_as_gpu::Remap(const std::vector<u8>& input, std::vector<u8>& output) if (!addr) { LOG_CRITICAL(Service_NVDRV, "map returned an invalid address!"); - result = NvErrCodes::InvalidInput; + result = NvResult::InvalidState; break; } } @@ -143,7 +161,7 @@ u32 nvhost_as_gpu::Remap(const std::vector<u8>& input, std::vector<u8>& output) return result; } -u32 nvhost_as_gpu::MapBufferEx(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_as_gpu::MapBufferEx(const std::vector<u8>& input, std::vector<u8>& output) { IoctlMapBufferEx params{}; std::memcpy(¶ms, input.data(), input.size()); @@ -157,7 +175,7 @@ u32 nvhost_as_gpu::MapBufferEx(const std::vector<u8>& input, std::vector<u8>& ou if (!object) { LOG_CRITICAL(Service_NVDRV, "invalid nvmap_handle={:X}", params.nvmap_handle); std::memcpy(output.data(), ¶ms, output.size()); - return NvErrCodes::InvalidInput; + return NvResult::InvalidState; } // The real nvservices doesn't make a distinction between handles and ids, and @@ -184,16 +202,16 @@ u32 nvhost_as_gpu::MapBufferEx(const std::vector<u8>& input, std::vector<u8>& ou params.mapping_size, params.offset); std::memcpy(output.data(), ¶ms, output.size()); - return NvErrCodes::InvalidInput; + return NvResult::InvalidState; } std::memcpy(output.data(), ¶ms, output.size()); - return NvErrCodes::Success; + return NvResult::Success; } else { LOG_CRITICAL(Service_NVDRV, "address not mapped offset={}", params.offset); std::memcpy(output.data(), ¶ms, output.size()); - return NvErrCodes::InvalidInput; + return NvResult::InvalidState; } } @@ -213,10 +231,10 @@ u32 nvhost_as_gpu::MapBufferEx(const std::vector<u8>& input, std::vector<u8>& ou params.offset = gpu.MemoryManager().Map(physical_address, params.offset, size); } - auto result{NvErrCodes::Success}; + auto result = NvResult::Success; if (!params.offset) { LOG_CRITICAL(Service_NVDRV, "failed to map size={}", size); - result = NvErrCodes::InvalidInput; + result = NvResult::InvalidState; } else { AddBufferMap(params.offset, size, physical_address, is_alloc); } @@ -225,7 +243,7 @@ u32 nvhost_as_gpu::MapBufferEx(const std::vector<u8>& input, std::vector<u8>& ou return result; } -u32 nvhost_as_gpu::UnmapBuffer(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_as_gpu::UnmapBuffer(const std::vector<u8>& input, std::vector<u8>& output) { IoctlUnmapBuffer params{}; std::memcpy(¶ms, input.data(), input.size()); @@ -238,20 +256,42 @@ u32 nvhost_as_gpu::UnmapBuffer(const std::vector<u8>& input, std::vector<u8>& ou } std::memcpy(output.data(), ¶ms, output.size()); - return NvErrCodes::Success; + return NvResult::Success; } -u32 nvhost_as_gpu::BindChannel(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_as_gpu::BindChannel(const std::vector<u8>& input, std::vector<u8>& output) { IoctlBindChannel params{}; std::memcpy(¶ms, input.data(), input.size()); - - LOG_DEBUG(Service_NVDRV, "called, fd={:X}", params.fd); + LOG_WARNING(Service_NVDRV, "(STUBBED) called, fd={:X}", params.fd); channel = params.fd; - return 0; + return NvResult::Success; +} + +NvResult nvhost_as_gpu::GetVARegions(const std::vector<u8>& input, std::vector<u8>& output) { + IoctlGetVaRegions params{}; + std::memcpy(¶ms, input.data(), input.size()); + + LOG_WARNING(Service_NVDRV, "(STUBBED) called, buf_addr={:X}, buf_size={:X}", params.buf_addr, + params.buf_size); + + params.buf_size = 0x30; + params.regions[0].offset = 0x04000000; + params.regions[0].page_size = 0x1000; + params.regions[0].pages = 0x3fbfff; + + params.regions[1].offset = 0x04000000; + params.regions[1].page_size = 0x10000; + params.regions[1].pages = 0x1bffff; + + // TODO(ogniK): This probably can stay stubbed but should add support way way later + + std::memcpy(output.data(), ¶ms, output.size()); + return NvResult::Success; } -u32 nvhost_as_gpu::GetVARegions(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_as_gpu::GetVARegions(const std::vector<u8>& input, std::vector<u8>& output, + std::vector<u8>& inline_output) { IoctlGetVaRegions params{}; std::memcpy(¶ms, input.data(), input.size()); @@ -270,7 +310,8 @@ u32 nvhost_as_gpu::GetVARegions(const std::vector<u8>& input, std::vector<u8>& o // TODO(ogniK): This probably can stay stubbed but should add support way way later std::memcpy(output.data(), ¶ms, output.size()); - return 0; + std::memcpy(inline_output.data(), ¶ms.regions, inline_output.size()); + return NvResult::Success; } std::optional<nvhost_as_gpu::BufferMap> nvhost_as_gpu::FindBufferMap(GPUVAddr gpu_addr) const { diff --git a/src/core/hle/service/nvdrv/devices/nvhost_as_gpu.h b/src/core/hle/service/nvdrv/devices/nvhost_as_gpu.h index fcdb40d93..08035fa0e 100644 --- a/src/core/hle/service/nvdrv/devices/nvhost_as_gpu.h +++ b/src/core/hle/service/nvdrv/devices/nvhost_as_gpu.h @@ -30,9 +30,11 @@ public: explicit nvhost_as_gpu(Core::System& system, std::shared_ptr<nvmap> nvmap_dev); ~nvhost_as_gpu() override; - u32 ioctl(Ioctl command, const std::vector<u8>& input, const std::vector<u8>& input2, - std::vector<u8>& output, std::vector<u8>& output2, IoctlCtrl& ctrl, - IoctlVersion version) override; + NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) override; + NvResult Ioctl2(Ioctl command, const std::vector<u8>& input, + const std::vector<u8>& inline_input, std::vector<u8>& output) override; + NvResult Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output, + std::vector<u8>& inline_output) override; private: class BufferMap final { @@ -74,32 +76,21 @@ private: bool is_allocated{}; }; - enum class IoctlCommand : u32_le { - IocInitalizeExCommand = 0x40284109, - IocAllocateSpaceCommand = 0xC0184102, - IocRemapCommand = 0x00000014, - IocMapBufferExCommand = 0xC0284106, - IocBindChannelCommand = 0x40044101, - IocGetVaRegionsCommand = 0xC0404108, - IocUnmapBufferCommand = 0xC0084105, - IocFreeSpaceCommand = 0xC0104103, - }; - struct IoctlInitalizeEx { - u32_le big_page_size; // depends on GPU's available_big_page_sizes; 0=default - s32_le as_fd; // ignored; passes 0 - u32_le flags; // passes 0 - u32_le reserved; // ignored; passes 0 - u64_le unk0; - u64_le unk1; - u64_le unk2; + u32_le big_page_size{}; // depends on GPU's available_big_page_sizes; 0=default + s32_le as_fd{}; // ignored; passes 0 + u32_le flags{}; // passes 0 + u32_le reserved{}; // ignored; passes 0 + u64_le unk0{}; + u64_le unk1{}; + u64_le unk2{}; }; static_assert(sizeof(IoctlInitalizeEx) == 40, "IoctlInitalizeEx is incorrect size"); struct IoctlAllocSpace { - u32_le pages; - u32_le page_size; - AddressSpaceFlags flags; + u32_le pages{}; + u32_le page_size{}; + AddressSpaceFlags flags{}; INSERT_PADDING_WORDS(1); union { u64_le offset; @@ -109,70 +100,73 @@ private: static_assert(sizeof(IoctlAllocSpace) == 24, "IoctlInitalizeEx is incorrect size"); struct IoctlFreeSpace { - u64_le offset; - u32_le pages; - u32_le page_size; + u64_le offset{}; + u32_le pages{}; + u32_le page_size{}; }; static_assert(sizeof(IoctlFreeSpace) == 16, "IoctlFreeSpace is incorrect size"); struct IoctlRemapEntry { - u16_le flags; - u16_le kind; - u32_le nvmap_handle; - u32_le map_offset; - u32_le offset; - u32_le pages; + u16_le flags{}; + u16_le kind{}; + u32_le nvmap_handle{}; + u32_le map_offset{}; + u32_le offset{}; + u32_le pages{}; }; static_assert(sizeof(IoctlRemapEntry) == 20, "IoctlRemapEntry is incorrect size"); struct IoctlMapBufferEx { - AddressSpaceFlags flags; // bit0: fixed_offset, bit2: cacheable - u32_le kind; // -1 is default - u32_le nvmap_handle; - u32_le page_size; // 0 means don't care - s64_le buffer_offset; - u64_le mapping_size; - s64_le offset; + AddressSpaceFlags flags{}; // bit0: fixed_offset, bit2: cacheable + u32_le kind{}; // -1 is default + u32_le nvmap_handle{}; + u32_le page_size{}; // 0 means don't care + s64_le buffer_offset{}; + u64_le mapping_size{}; + s64_le offset{}; }; static_assert(sizeof(IoctlMapBufferEx) == 40, "IoctlMapBufferEx is incorrect size"); struct IoctlUnmapBuffer { - s64_le offset; + s64_le offset{}; }; static_assert(sizeof(IoctlUnmapBuffer) == 8, "IoctlUnmapBuffer is incorrect size"); struct IoctlBindChannel { - u32_le fd; + s32_le fd{}; }; static_assert(sizeof(IoctlBindChannel) == 4, "IoctlBindChannel is incorrect size"); struct IoctlVaRegion { - u64_le offset; - u32_le page_size; + u64_le offset{}; + u32_le page_size{}; INSERT_PADDING_WORDS(1); - u64_le pages; + u64_le pages{}; }; static_assert(sizeof(IoctlVaRegion) == 24, "IoctlVaRegion is incorrect size"); struct IoctlGetVaRegions { - u64_le buf_addr; // (contained output user ptr on linux, ignored) - u32_le buf_size; // forced to 2*sizeof(struct va_region) - u32_le reserved; - IoctlVaRegion regions[2]; + u64_le buf_addr{}; // (contained output user ptr on linux, ignored) + u32_le buf_size{}; // forced to 2*sizeof(struct va_region) + u32_le reserved{}; + IoctlVaRegion regions[2]{}; }; static_assert(sizeof(IoctlGetVaRegions) == 16 + sizeof(IoctlVaRegion) * 2, "IoctlGetVaRegions is incorrect size"); - u32 channel{}; + s32 channel{}; + + NvResult InitalizeEx(const std::vector<u8>& input, std::vector<u8>& output); + NvResult AllocateSpace(const std::vector<u8>& input, std::vector<u8>& output); + NvResult Remap(const std::vector<u8>& input, std::vector<u8>& output); + NvResult MapBufferEx(const std::vector<u8>& input, std::vector<u8>& output); + NvResult UnmapBuffer(const std::vector<u8>& input, std::vector<u8>& output); + NvResult FreeSpace(const std::vector<u8>& input, std::vector<u8>& output); + NvResult BindChannel(const std::vector<u8>& input, std::vector<u8>& output); - u32 InitalizeEx(const std::vector<u8>& input, std::vector<u8>& output); - u32 AllocateSpace(const std::vector<u8>& input, std::vector<u8>& output); - u32 Remap(const std::vector<u8>& input, std::vector<u8>& output); - u32 MapBufferEx(const std::vector<u8>& input, std::vector<u8>& output); - u32 UnmapBuffer(const std::vector<u8>& input, std::vector<u8>& output); - u32 FreeSpace(const std::vector<u8>& input, std::vector<u8>& output); - u32 BindChannel(const std::vector<u8>& input, std::vector<u8>& output); - u32 GetVARegions(const std::vector<u8>& input, std::vector<u8>& output); + NvResult GetVARegions(const std::vector<u8>& input, std::vector<u8>& output); + NvResult GetVARegions(const std::vector<u8>& input, std::vector<u8>& output, + std::vector<u8>& inline_output); std::optional<BufferMap> FindBufferMap(GPUVAddr gpu_addr) const; void AddBufferMap(GPUVAddr gpu_addr, std::size_t size, VAddr cpu_addr, bool is_allocated); diff --git a/src/core/hle/service/nvdrv/devices/nvhost_ctrl.cpp b/src/core/hle/service/nvdrv/devices/nvhost_ctrl.cpp index 8356a8139..d90cf90a8 100644 --- a/src/core/hle/service/nvdrv/devices/nvhost_ctrl.cpp +++ b/src/core/hle/service/nvdrv/devices/nvhost_ctrl.cpp @@ -20,41 +20,54 @@ nvhost_ctrl::nvhost_ctrl(Core::System& system, EventInterface& events_interface, : nvdevice(system), events_interface{events_interface}, syncpoint_manager{syncpoint_manager} {} nvhost_ctrl::~nvhost_ctrl() = default; -u32 nvhost_ctrl::ioctl(Ioctl command, const std::vector<u8>& input, const std::vector<u8>& input2, - std::vector<u8>& output, std::vector<u8>& output2, IoctlCtrl& ctrl, - IoctlVersion version) { - LOG_DEBUG(Service_NVDRV, "called, command=0x{:08X}, input_size=0x{:X}, output_size=0x{:X}", - command.raw, input.size(), output.size()); - - switch (static_cast<IoctlCommand>(command.raw)) { - case IoctlCommand::IocGetConfigCommand: - return NvOsGetConfigU32(input, output); - case IoctlCommand::IocCtrlEventWaitCommand: - return IocCtrlEventWait(input, output, false, ctrl); - case IoctlCommand::IocCtrlEventWaitAsyncCommand: - return IocCtrlEventWait(input, output, true, ctrl); - case IoctlCommand::IocCtrlEventRegisterCommand: - return IocCtrlEventRegister(input, output); - case IoctlCommand::IocCtrlEventUnregisterCommand: - return IocCtrlEventUnregister(input, output); - case IoctlCommand::IocCtrlClearEventWaitCommand: - return IocCtrlClearEventWait(input, output); +NvResult nvhost_ctrl::Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) { + switch (command.group) { + case 0x0: + switch (command.cmd) { + case 0x1b: + return NvOsGetConfigU32(input, output); + case 0x1c: + return IocCtrlClearEventWait(input, output); + case 0x1d: + return IocCtrlEventWait(input, output, false); + case 0x1e: + return IocCtrlEventWait(input, output, true); + case 0x1f: + return IocCtrlEventRegister(input, output); + case 0x20: + return IocCtrlEventUnregister(input, output); + } + break; default: - UNIMPLEMENTED_MSG("Unimplemented ioctl"); - return 0; + break; } + + UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw); + return NvResult::NotImplemented; +} + +NvResult nvhost_ctrl::Ioctl2(Ioctl command, const std::vector<u8>& input, + const std::vector<u8>& inline_input, std::vector<u8>& output) { + UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw); + return NvResult::NotImplemented; +} + +NvResult nvhost_ctrl::Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output, + std::vector<u8>& inline_output) { + UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw); + return NvResult::NotImplemented; } -u32 nvhost_ctrl::NvOsGetConfigU32(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_ctrl::NvOsGetConfigU32(const std::vector<u8>& input, std::vector<u8>& output) { IocGetConfigParams params{}; std::memcpy(¶ms, input.data(), sizeof(params)); LOG_TRACE(Service_NVDRV, "called, setting={}!{}", params.domain_str.data(), params.param_str.data()); - return 0x30006; // Returns error on production mode + return NvResult::ConfigVarNotFound; // Returns error on production mode } -u32 nvhost_ctrl::IocCtrlEventWait(const std::vector<u8>& input, std::vector<u8>& output, - bool is_async, IoctlCtrl& ctrl) { +NvResult nvhost_ctrl::IocCtrlEventWait(const std::vector<u8>& input, std::vector<u8>& output, + bool is_async) { IocCtrlEventWaitParams params{}; std::memcpy(¶ms, input.data(), sizeof(params)); LOG_DEBUG(Service_NVDRV, "syncpt_id={}, threshold={}, timeout={}, is_async={}", @@ -126,10 +139,7 @@ u32 nvhost_ctrl::IocCtrlEventWait(const std::vector<u8>& input, std::vector<u8>& params.value |= event_id; event.event.writable->Clear(); gpu.RegisterSyncptInterrupt(params.syncpt_id, target_value); - if (!is_async && ctrl.fresh_call) { - ctrl.must_delay = true; - ctrl.timeout = params.timeout; - ctrl.event_id = event_id; + if (!is_async) { return NvResult::Timeout; } std::memcpy(output.data(), ¶ms, sizeof(params)); @@ -139,7 +149,7 @@ u32 nvhost_ctrl::IocCtrlEventWait(const std::vector<u8>& input, std::vector<u8>& return NvResult::BadParameter; } -u32 nvhost_ctrl::IocCtrlEventRegister(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_ctrl::IocCtrlEventRegister(const std::vector<u8>& input, std::vector<u8>& output) { IocCtrlEventRegisterParams params{}; std::memcpy(¶ms, input.data(), sizeof(params)); const u32 event_id = params.user_event_id & 0x00FF; @@ -154,7 +164,8 @@ u32 nvhost_ctrl::IocCtrlEventRegister(const std::vector<u8>& input, std::vector< return NvResult::Success; } -u32 nvhost_ctrl::IocCtrlEventUnregister(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_ctrl::IocCtrlEventUnregister(const std::vector<u8>& input, + std::vector<u8>& output) { IocCtrlEventUnregisterParams params{}; std::memcpy(¶ms, input.data(), sizeof(params)); const u32 event_id = params.user_event_id & 0x00FF; @@ -169,7 +180,7 @@ u32 nvhost_ctrl::IocCtrlEventUnregister(const std::vector<u8>& input, std::vecto return NvResult::Success; } -u32 nvhost_ctrl::IocCtrlClearEventWait(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_ctrl::IocCtrlClearEventWait(const std::vector<u8>& input, std::vector<u8>& output) { IocCtrlEventSignalParams params{}; std::memcpy(¶ms, input.data(), sizeof(params)); diff --git a/src/core/hle/service/nvdrv/devices/nvhost_ctrl.h b/src/core/hle/service/nvdrv/devices/nvhost_ctrl.h index 24ad96cb9..c5aa1362a 100644 --- a/src/core/hle/service/nvdrv/devices/nvhost_ctrl.h +++ b/src/core/hle/service/nvdrv/devices/nvhost_ctrl.h @@ -18,132 +18,113 @@ public: SyncpointManager& syncpoint_manager); ~nvhost_ctrl() override; - u32 ioctl(Ioctl command, const std::vector<u8>& input, const std::vector<u8>& input2, - std::vector<u8>& output, std::vector<u8>& output2, IoctlCtrl& ctrl, - IoctlVersion version) override; + NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) override; + NvResult Ioctl2(Ioctl command, const std::vector<u8>& input, + const std::vector<u8>& inline_input, std::vector<u8>& output) override; + NvResult Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output, + std::vector<u8>& inline_output) override; private: - enum class IoctlCommand : u32_le { - IocSyncptReadCommand = 0xC0080014, - IocSyncptIncrCommand = 0x40040015, - IocSyncptWaitCommand = 0xC00C0016, - IocModuleMutexCommand = 0x40080017, - IocModuleRegRDWRCommand = 0xC0180018, - IocSyncptWaitexCommand = 0xC0100019, - IocSyncptReadMaxCommand = 0xC008001A, - IocGetConfigCommand = 0xC183001B, - IocCtrlClearEventWaitCommand = 0xC004001C, - IocCtrlEventWaitCommand = 0xC010001D, - IocCtrlEventWaitAsyncCommand = 0xC010001E, - IocCtrlEventRegisterCommand = 0xC004001F, - IocCtrlEventUnregisterCommand = 0xC0040020, - IocCtrlEventKillCommand = 0x40080021, - }; struct IocSyncptReadParams { - u32_le id; - u32_le value; + u32_le id{}; + u32_le value{}; }; static_assert(sizeof(IocSyncptReadParams) == 8, "IocSyncptReadParams is incorrect size"); struct IocSyncptIncrParams { - u32_le id; + u32_le id{}; }; static_assert(sizeof(IocSyncptIncrParams) == 4, "IocSyncptIncrParams is incorrect size"); struct IocSyncptWaitParams { - u32_le id; - u32_le thresh; - s32_le timeout; + u32_le id{}; + u32_le thresh{}; + s32_le timeout{}; }; static_assert(sizeof(IocSyncptWaitParams) == 12, "IocSyncptWaitParams is incorrect size"); struct IocModuleMutexParams { - u32_le id; - u32_le lock; // (0 = unlock and 1 = lock) + u32_le id{}; + u32_le lock{}; // (0 = unlock and 1 = lock) }; static_assert(sizeof(IocModuleMutexParams) == 8, "IocModuleMutexParams is incorrect size"); struct IocModuleRegRDWRParams { - u32_le id; - u32_le num_offsets; - u32_le block_size; - u32_le offsets; - u32_le values; - u32_le write; + u32_le id{}; + u32_le num_offsets{}; + u32_le block_size{}; + u32_le offsets{}; + u32_le values{}; + u32_le write{}; }; static_assert(sizeof(IocModuleRegRDWRParams) == 24, "IocModuleRegRDWRParams is incorrect size"); struct IocSyncptWaitexParams { - u32_le id; - u32_le thresh; - s32_le timeout; - u32_le value; + u32_le id{}; + u32_le thresh{}; + s32_le timeout{}; + u32_le value{}; }; static_assert(sizeof(IocSyncptWaitexParams) == 16, "IocSyncptWaitexParams is incorrect size"); struct IocSyncptReadMaxParams { - u32_le id; - u32_le value; + u32_le id{}; + u32_le value{}; }; static_assert(sizeof(IocSyncptReadMaxParams) == 8, "IocSyncptReadMaxParams is incorrect size"); struct IocGetConfigParams { - std::array<char, 0x41> domain_str; - std::array<char, 0x41> param_str; - std::array<char, 0x101> config_str; + std::array<char, 0x41> domain_str{}; + std::array<char, 0x41> param_str{}; + std::array<char, 0x101> config_str{}; }; static_assert(sizeof(IocGetConfigParams) == 387, "IocGetConfigParams is incorrect size"); struct IocCtrlEventSignalParams { - u32_le event_id; + u32_le event_id{}; }; static_assert(sizeof(IocCtrlEventSignalParams) == 4, "IocCtrlEventSignalParams is incorrect size"); struct IocCtrlEventWaitParams { - u32_le syncpt_id; - u32_le threshold; - s32_le timeout; - u32_le value; + u32_le syncpt_id{}; + u32_le threshold{}; + s32_le timeout{}; + u32_le value{}; }; static_assert(sizeof(IocCtrlEventWaitParams) == 16, "IocCtrlEventWaitParams is incorrect size"); struct IocCtrlEventWaitAsyncParams { - u32_le syncpt_id; - u32_le threshold; - u32_le timeout; - u32_le value; + u32_le syncpt_id{}; + u32_le threshold{}; + u32_le timeout{}; + u32_le value{}; }; static_assert(sizeof(IocCtrlEventWaitAsyncParams) == 16, "IocCtrlEventWaitAsyncParams is incorrect size"); struct IocCtrlEventRegisterParams { - u32_le user_event_id; + u32_le user_event_id{}; }; static_assert(sizeof(IocCtrlEventRegisterParams) == 4, "IocCtrlEventRegisterParams is incorrect size"); struct IocCtrlEventUnregisterParams { - u32_le user_event_id; + u32_le user_event_id{}; }; static_assert(sizeof(IocCtrlEventUnregisterParams) == 4, "IocCtrlEventUnregisterParams is incorrect size"); struct IocCtrlEventKill { - u64_le user_events; + u64_le user_events{}; }; static_assert(sizeof(IocCtrlEventKill) == 8, "IocCtrlEventKill is incorrect size"); - u32 NvOsGetConfigU32(const std::vector<u8>& input, std::vector<u8>& output); - - u32 IocCtrlEventWait(const std::vector<u8>& input, std::vector<u8>& output, bool is_async, - IoctlCtrl& ctrl); - - u32 IocCtrlEventRegister(const std::vector<u8>& input, std::vector<u8>& output); - - u32 IocCtrlEventUnregister(const std::vector<u8>& input, std::vector<u8>& output); - - u32 IocCtrlClearEventWait(const std::vector<u8>& input, std::vector<u8>& output); + NvResult NvOsGetConfigU32(const std::vector<u8>& input, std::vector<u8>& output); + NvResult IocCtrlEventWait(const std::vector<u8>& input, std::vector<u8>& output, bool is_async); + NvResult IocCtrlEventRegister(const std::vector<u8>& input, std::vector<u8>& output); + NvResult IocCtrlEventUnregister(const std::vector<u8>& input, std::vector<u8>& output); + NvResult IocCtrlClearEventWait(const std::vector<u8>& input, std::vector<u8>& output); EventInterface& events_interface; SyncpointManager& syncpoint_manager; diff --git a/src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.cpp b/src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.cpp index fba89e7a6..2d7ea433c 100644 --- a/src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.cpp +++ b/src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.cpp @@ -15,39 +15,66 @@ namespace Service::Nvidia::Devices { nvhost_ctrl_gpu::nvhost_ctrl_gpu(Core::System& system) : nvdevice(system) {} nvhost_ctrl_gpu::~nvhost_ctrl_gpu() = default; -u32 nvhost_ctrl_gpu::ioctl(Ioctl command, const std::vector<u8>& input, - const std::vector<u8>& input2, std::vector<u8>& output, - std::vector<u8>& output2, IoctlCtrl& ctrl, IoctlVersion version) { - LOG_DEBUG(Service_NVDRV, "called, command=0x{:08X}, input_size=0x{:X}, output_size=0x{:X}", - command.raw, input.size(), output.size()); - - switch (static_cast<IoctlCommand>(command.raw)) { - case IoctlCommand::IocGetCharacteristicsCommand: - return GetCharacteristics(input, output, output2, version); - case IoctlCommand::IocGetTPCMasksCommand: - return GetTPCMasks(input, output, output2, version); - case IoctlCommand::IocGetActiveSlotMaskCommand: - return GetActiveSlotMask(input, output); - case IoctlCommand::IocZcullGetCtxSizeCommand: - return ZCullGetCtxSize(input, output); - case IoctlCommand::IocZcullGetInfo: - return ZCullGetInfo(input, output); - case IoctlCommand::IocZbcSetTable: - return ZBCSetTable(input, output); - case IoctlCommand::IocZbcQueryTable: - return ZBCQueryTable(input, output); - case IoctlCommand::IocFlushL2: - return FlushL2(input, output); - case IoctlCommand::IocGetGpuTime: - return GetGpuTime(input, output); +NvResult nvhost_ctrl_gpu::Ioctl1(Ioctl command, const std::vector<u8>& input, + std::vector<u8>& output) { + switch (command.group) { + case 'G': + switch (command.cmd) { + case 0x1: + return ZCullGetCtxSize(input, output); + case 0x2: + return ZCullGetInfo(input, output); + case 0x3: + return ZBCSetTable(input, output); + case 0x4: + return ZBCQueryTable(input, output); + case 0x5: + return GetCharacteristics(input, output); + case 0x6: + return GetTPCMasks(input, output); + case 0x7: + return FlushL2(input, output); + case 0x14: + return GetActiveSlotMask(input, output); + case 0x1c: + return GetGpuTime(input, output); + default: + break; + } + break; + } + UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw); + return NvResult::NotImplemented; +} + +NvResult nvhost_ctrl_gpu::Ioctl2(Ioctl command, const std::vector<u8>& input, + const std::vector<u8>& inline_input, std::vector<u8>& output) { + UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw); + return NvResult::NotImplemented; +} + +NvResult nvhost_ctrl_gpu::Ioctl3(Ioctl command, const std::vector<u8>& input, + std::vector<u8>& output, std::vector<u8>& inline_output) { + switch (command.group) { + case 'G': + switch (command.cmd) { + case 0x5: + return GetCharacteristics(input, output, inline_output); + case 0x6: + return GetTPCMasks(input, output, inline_output); + default: + break; + } + break; default: - UNIMPLEMENTED_MSG("Unimplemented ioctl"); - return 0; + break; } + UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw); + return NvResult::NotImplemented; } -u32 nvhost_ctrl_gpu::GetCharacteristics(const std::vector<u8>& input, std::vector<u8>& output, - std::vector<u8>& output2, IoctlVersion version) { +NvResult nvhost_ctrl_gpu::GetCharacteristics(const std::vector<u8>& input, + std::vector<u8>& output) { LOG_DEBUG(Service_NVDRV, "called"); IoctlCharacteristics params{}; std::memcpy(¶ms, input.data(), input.size()); @@ -88,36 +115,83 @@ u32 nvhost_ctrl_gpu::GetCharacteristics(const std::vector<u8>& input, std::vecto params.gc.gr_compbit_store_base_hw = 0x0; params.gpu_characteristics_buf_size = 0xA0; params.gpu_characteristics_buf_addr = 0xdeadbeef; // Cannot be 0 (UNUSED) + std::memcpy(output.data(), ¶ms, output.size()); + return NvResult::Success; +} - if (version == IoctlVersion::Version3) { - std::memcpy(output.data(), input.data(), output.size()); - std::memcpy(output2.data(), ¶ms.gc, output2.size()); - } else { - std::memcpy(output.data(), ¶ms, output.size()); - } - return 0; +NvResult nvhost_ctrl_gpu::GetCharacteristics(const std::vector<u8>& input, std::vector<u8>& output, + std::vector<u8>& inline_output) { + LOG_DEBUG(Service_NVDRV, "called"); + IoctlCharacteristics params{}; + std::memcpy(¶ms, input.data(), input.size()); + params.gc.arch = 0x120; + params.gc.impl = 0xb; + params.gc.rev = 0xa1; + params.gc.num_gpc = 0x1; + params.gc.l2_cache_size = 0x40000; + params.gc.on_board_video_memory_size = 0x0; + params.gc.num_tpc_per_gpc = 0x2; + params.gc.bus_type = 0x20; + params.gc.big_page_size = 0x20000; + params.gc.compression_page_size = 0x20000; + params.gc.pde_coverage_bit_count = 0x1B; + params.gc.available_big_page_sizes = 0x30000; + params.gc.gpc_mask = 0x1; + params.gc.sm_arch_sm_version = 0x503; + params.gc.sm_arch_spa_version = 0x503; + params.gc.sm_arch_warp_count = 0x80; + params.gc.gpu_va_bit_count = 0x28; + params.gc.reserved = 0x0; + params.gc.flags = 0x55; + params.gc.twod_class = 0x902D; + params.gc.threed_class = 0xB197; + params.gc.compute_class = 0xB1C0; + params.gc.gpfifo_class = 0xB06F; + params.gc.inline_to_memory_class = 0xA140; + params.gc.dma_copy_class = 0xB0B5; + params.gc.max_fbps_count = 0x1; + params.gc.fbp_en_mask = 0x0; + params.gc.max_ltc_per_fbp = 0x2; + params.gc.max_lts_per_ltc = 0x1; + params.gc.max_tex_per_tpc = 0x0; + params.gc.max_gpc_count = 0x1; + params.gc.rop_l2_en_mask_0 = 0x21D70; + params.gc.rop_l2_en_mask_1 = 0x0; + params.gc.chipname = 0x6230326D67; + params.gc.gr_compbit_store_base_hw = 0x0; + params.gpu_characteristics_buf_size = 0xA0; + params.gpu_characteristics_buf_addr = 0xdeadbeef; // Cannot be 0 (UNUSED) + + std::memcpy(output.data(), input.data(), output.size()); + std::memcpy(inline_output.data(), ¶ms.gc, inline_output.size()); + return NvResult::Success; } -u32 nvhost_ctrl_gpu::GetTPCMasks(const std::vector<u8>& input, std::vector<u8>& output, - std::vector<u8>& output2, IoctlVersion version) { +NvResult nvhost_ctrl_gpu::GetTPCMasks(const std::vector<u8>& input, std::vector<u8>& output) { IoctlGpuGetTpcMasksArgs params{}; std::memcpy(¶ms, input.data(), input.size()); LOG_DEBUG(Service_NVDRV, "called, mask_buffer_size=0x{:X}", params.mask_buffer_size); if (params.mask_buffer_size != 0) { params.tcp_mask = 3; } + std::memcpy(output.data(), ¶ms, output.size()); + return NvResult::Success; +} - if (version == IoctlVersion::Version3) { - std::memcpy(output.data(), input.data(), output.size()); - std::memcpy(output2.data(), ¶ms.tcp_mask, output2.size()); - } else { - std::memcpy(output.data(), ¶ms, output.size()); +NvResult nvhost_ctrl_gpu::GetTPCMasks(const std::vector<u8>& input, std::vector<u8>& output, + std::vector<u8>& inline_output) { + IoctlGpuGetTpcMasksArgs params{}; + std::memcpy(¶ms, input.data(), input.size()); + LOG_DEBUG(Service_NVDRV, "called, mask_buffer_size=0x{:X}", params.mask_buffer_size); + if (params.mask_buffer_size != 0) { + params.tcp_mask = 3; } - - return 0; + std::memcpy(output.data(), ¶ms, output.size()); + std::memcpy(inline_output.data(), ¶ms.tcp_mask, inline_output.size()); + return NvResult::Success; } -u32 nvhost_ctrl_gpu::GetActiveSlotMask(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_ctrl_gpu::GetActiveSlotMask(const std::vector<u8>& input, std::vector<u8>& output) { LOG_DEBUG(Service_NVDRV, "called"); IoctlActiveSlotMask params{}; @@ -127,10 +201,10 @@ u32 nvhost_ctrl_gpu::GetActiveSlotMask(const std::vector<u8>& input, std::vector params.slot = 0x07; params.mask = 0x01; std::memcpy(output.data(), ¶ms, output.size()); - return 0; + return NvResult::Success; } -u32 nvhost_ctrl_gpu::ZCullGetCtxSize(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_ctrl_gpu::ZCullGetCtxSize(const std::vector<u8>& input, std::vector<u8>& output) { LOG_DEBUG(Service_NVDRV, "called"); IoctlZcullGetCtxSize params{}; @@ -139,10 +213,10 @@ u32 nvhost_ctrl_gpu::ZCullGetCtxSize(const std::vector<u8>& input, std::vector<u } params.size = 0x1; std::memcpy(output.data(), ¶ms, output.size()); - return 0; + return NvResult::Success; } -u32 nvhost_ctrl_gpu::ZCullGetInfo(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_ctrl_gpu::ZCullGetInfo(const std::vector<u8>& input, std::vector<u8>& output) { LOG_DEBUG(Service_NVDRV, "called"); IoctlNvgpuGpuZcullGetInfoArgs params{}; @@ -162,47 +236,47 @@ u32 nvhost_ctrl_gpu::ZCullGetInfo(const std::vector<u8>& input, std::vector<u8>& params.subregion_height_align_pixels = 0x40; params.subregion_count = 0x10; std::memcpy(output.data(), ¶ms, output.size()); - return 0; + return NvResult::Success; } -u32 nvhost_ctrl_gpu::ZBCSetTable(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_ctrl_gpu::ZBCSetTable(const std::vector<u8>& input, std::vector<u8>& output) { LOG_WARNING(Service_NVDRV, "(STUBBED) called"); IoctlZbcSetTable params{}; std::memcpy(¶ms, input.data(), input.size()); // TODO(ogniK): What does this even actually do? std::memcpy(output.data(), ¶ms, output.size()); - return 0; + return NvResult::Success; } -u32 nvhost_ctrl_gpu::ZBCQueryTable(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_ctrl_gpu::ZBCQueryTable(const std::vector<u8>& input, std::vector<u8>& output) { LOG_WARNING(Service_NVDRV, "(STUBBED) called"); IoctlZbcQueryTable params{}; std::memcpy(¶ms, input.data(), input.size()); // TODO : To implement properly std::memcpy(output.data(), ¶ms, output.size()); - return 0; + return NvResult::Success; } -u32 nvhost_ctrl_gpu::FlushL2(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_ctrl_gpu::FlushL2(const std::vector<u8>& input, std::vector<u8>& output) { LOG_WARNING(Service_NVDRV, "(STUBBED) called"); IoctlFlushL2 params{}; std::memcpy(¶ms, input.data(), input.size()); // TODO : To implement properly std::memcpy(output.data(), ¶ms, output.size()); - return 0; + return NvResult::Success; } -u32 nvhost_ctrl_gpu::GetGpuTime(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_ctrl_gpu::GetGpuTime(const std::vector<u8>& input, std::vector<u8>& output) { LOG_DEBUG(Service_NVDRV, "called"); IoctlGetGpuTime params{}; std::memcpy(¶ms, input.data(), input.size()); params.gpu_time = static_cast<u64_le>(system.CoreTiming().GetGlobalTimeNs().count()); std::memcpy(output.data(), ¶ms, output.size()); - return 0; + return NvResult::Success; } } // namespace Service::Nvidia::Devices diff --git a/src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.h b/src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.h index ef60f72ce..137b88238 100644 --- a/src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.h +++ b/src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.h @@ -16,32 +16,13 @@ public: explicit nvhost_ctrl_gpu(Core::System& system); ~nvhost_ctrl_gpu() override; - u32 ioctl(Ioctl command, const std::vector<u8>& input, const std::vector<u8>& input2, - std::vector<u8>& output, std::vector<u8>& output2, IoctlCtrl& ctrl, - IoctlVersion version) override; + NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) override; + NvResult Ioctl2(Ioctl command, const std::vector<u8>& input, + const std::vector<u8>& inline_input, std::vector<u8>& output) override; + NvResult Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output, + std::vector<u8>& inline_output) override; private: - enum class IoctlCommand : u32_le { - IocGetCharacteristicsCommand = 0xC0B04705, - IocGetTPCMasksCommand = 0xC0184706, - IocGetActiveSlotMaskCommand = 0x80084714, - IocZcullGetCtxSizeCommand = 0x80044701, - IocZcullGetInfo = 0x80284702, - IocZbcSetTable = 0x402C4703, - IocZbcQueryTable = 0xC0344704, - IocFlushL2 = 0x40084707, - IocInvalICache = 0x4008470D, - IocSetMmudebugMode = 0x4008470E, - IocSetSmDebugMode = 0x4010470F, - IocWaitForPause = 0xC0084710, - IocGetTcpExceptionEnStatus = 0x80084711, - IocNumVsms = 0x80084712, - IocVsmsMapping = 0xC0044713, - IocGetErrorChannelUserData = 0xC008471B, - IocGetGpuTime = 0xC010471C, - IocGetCpuTimeCorrelationInfo = 0xC108471D, - }; - struct IoctlGpuCharacteristics { u32_le arch; // 0x120 (NVGPU_GPU_ARCH_GM200) u32_le impl; // 0xB (NVGPU_GPU_IMPL_GM20B) @@ -159,17 +140,21 @@ private: }; static_assert(sizeof(IoctlGetGpuTime) == 0x10, "IoctlGetGpuTime is incorrect size"); - u32 GetCharacteristics(const std::vector<u8>& input, std::vector<u8>& output, - std::vector<u8>& output2, IoctlVersion version); - u32 GetTPCMasks(const std::vector<u8>& input, std::vector<u8>& output, std::vector<u8>& output2, - IoctlVersion version); - u32 GetActiveSlotMask(const std::vector<u8>& input, std::vector<u8>& output); - u32 ZCullGetCtxSize(const std::vector<u8>& input, std::vector<u8>& output); - u32 ZCullGetInfo(const std::vector<u8>& input, std::vector<u8>& output); - u32 ZBCSetTable(const std::vector<u8>& input, std::vector<u8>& output); - u32 ZBCQueryTable(const std::vector<u8>& input, std::vector<u8>& output); - u32 FlushL2(const std::vector<u8>& input, std::vector<u8>& output); - u32 GetGpuTime(const std::vector<u8>& input, std::vector<u8>& output); + NvResult GetCharacteristics(const std::vector<u8>& input, std::vector<u8>& output); + NvResult GetCharacteristics(const std::vector<u8>& input, std::vector<u8>& output, + std::vector<u8>& inline_output); + + NvResult GetTPCMasks(const std::vector<u8>& input, std::vector<u8>& output); + NvResult GetTPCMasks(const std::vector<u8>& input, std::vector<u8>& output, + std::vector<u8>& inline_output); + + NvResult GetActiveSlotMask(const std::vector<u8>& input, std::vector<u8>& output); + NvResult ZCullGetCtxSize(const std::vector<u8>& input, std::vector<u8>& output); + NvResult ZCullGetInfo(const std::vector<u8>& input, std::vector<u8>& output); + NvResult ZBCSetTable(const std::vector<u8>& input, std::vector<u8>& output); + NvResult ZBCQueryTable(const std::vector<u8>& input, std::vector<u8>& output); + NvResult FlushL2(const std::vector<u8>& input, std::vector<u8>& output); + NvResult GetGpuTime(const std::vector<u8>& input, std::vector<u8>& output); }; } // namespace Service::Nvidia::Devices diff --git a/src/core/hle/service/nvdrv/devices/nvhost_gpu.cpp b/src/core/hle/service/nvdrv/devices/nvhost_gpu.cpp index b1d9d55b5..af8b3d9f1 100644 --- a/src/core/hle/service/nvdrv/devices/nvhost_gpu.cpp +++ b/src/core/hle/service/nvdrv/devices/nvhost_gpu.cpp @@ -23,107 +23,132 @@ nvhost_gpu::nvhost_gpu(Core::System& system, std::shared_ptr<nvmap> nvmap_dev, nvhost_gpu::~nvhost_gpu() = default; -u32 nvhost_gpu::ioctl(Ioctl command, const std::vector<u8>& input, const std::vector<u8>& input2, - std::vector<u8>& output, std::vector<u8>& output2, IoctlCtrl& ctrl, - IoctlVersion version) { - LOG_DEBUG(Service_NVDRV, "called, command=0x{:08X}, input_size=0x{:X}, output_size=0x{:X}", - command.raw, input.size(), output.size()); - - switch (static_cast<IoctlCommand>(command.raw)) { - case IoctlCommand::IocSetNVMAPfdCommand: - return SetNVMAPfd(input, output); - case IoctlCommand::IocSetClientDataCommand: - return SetClientData(input, output); - case IoctlCommand::IocGetClientDataCommand: - return GetClientData(input, output); - case IoctlCommand::IocZCullBind: - return ZCullBind(input, output); - case IoctlCommand::IocSetErrorNotifierCommand: - return SetErrorNotifier(input, output); - case IoctlCommand::IocChannelSetPriorityCommand: - return SetChannelPriority(input, output); - case IoctlCommand::IocAllocGPFIFOEx2Command: - return AllocGPFIFOEx2(input, output); - case IoctlCommand::IocAllocObjCtxCommand: - return AllocateObjectContext(input, output); - case IoctlCommand::IocChannelGetWaitbaseCommand: - return GetWaitbase(input, output); - case IoctlCommand::IocChannelSetTimeoutCommand: - return ChannelSetTimeout(input, output); - case IoctlCommand::IocChannelSetTimeslice: - return ChannelSetTimeslice(input, output); - default: +NvResult nvhost_gpu::Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) { + switch (command.group) { + case 0x0: + switch (command.cmd) { + case 0x3: + return GetWaitbase(input, output); + default: + break; + } + break; + case 'H': + switch (command.cmd) { + case 0x1: + return SetNVMAPfd(input, output); + case 0x3: + return ChannelSetTimeout(input, output); + case 0x8: + return SubmitGPFIFOBase(input, output, false); + case 0x9: + return AllocateObjectContext(input, output); + case 0xb: + return ZCullBind(input, output); + case 0xc: + return SetErrorNotifier(input, output); + case 0xd: + return SetChannelPriority(input, output); + case 0x1a: + return AllocGPFIFOEx2(input, output); + case 0x1b: + return SubmitGPFIFOBase(input, output, true); + case 0x1d: + return ChannelSetTimeslice(input, output); + default: + break; + } + break; + case 'G': + switch (command.cmd) { + case 0x14: + return SetClientData(input, output); + case 0x15: + return GetClientData(input, output); + default: + break; + } break; } + UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw); + return NvResult::NotImplemented; +}; - if (command.group == NVGPU_IOCTL_MAGIC) { - if (command.cmd == NVGPU_IOCTL_CHANNEL_SUBMIT_GPFIFO) { - return SubmitGPFIFO(input, output); - } - if (command.cmd == NVGPU_IOCTL_CHANNEL_KICKOFF_PB) { - return KickoffPB(input, output, input2, version); +NvResult nvhost_gpu::Ioctl2(Ioctl command, const std::vector<u8>& input, + const std::vector<u8>& inline_input, std::vector<u8>& output) { + switch (command.group) { + case 'H': + switch (command.cmd) { + case 0x1b: + return SubmitGPFIFOBase(input, inline_input, output); } + break; } + UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw); + return NvResult::NotImplemented; +} - UNIMPLEMENTED_MSG("Unimplemented ioctl"); - return 0; -}; +NvResult nvhost_gpu::Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output, + std::vector<u8>& inline_output) { + UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw); + return NvResult::NotImplemented; +} -u32 nvhost_gpu::SetNVMAPfd(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_gpu::SetNVMAPfd(const std::vector<u8>& input, std::vector<u8>& output) { IoctlSetNvmapFD params{}; std::memcpy(¶ms, input.data(), input.size()); LOG_DEBUG(Service_NVDRV, "called, fd={}", params.nvmap_fd); nvmap_fd = params.nvmap_fd; - return 0; + return NvResult::Success; } -u32 nvhost_gpu::SetClientData(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_gpu::SetClientData(const std::vector<u8>& input, std::vector<u8>& output) { LOG_DEBUG(Service_NVDRV, "called"); IoctlClientData params{}; std::memcpy(¶ms, input.data(), input.size()); user_data = params.data; - return 0; + return NvResult::Success; } -u32 nvhost_gpu::GetClientData(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_gpu::GetClientData(const std::vector<u8>& input, std::vector<u8>& output) { LOG_DEBUG(Service_NVDRV, "called"); IoctlClientData params{}; std::memcpy(¶ms, input.data(), input.size()); params.data = user_data; std::memcpy(output.data(), ¶ms, output.size()); - return 0; + return NvResult::Success; } -u32 nvhost_gpu::ZCullBind(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_gpu::ZCullBind(const std::vector<u8>& input, std::vector<u8>& output) { std::memcpy(&zcull_params, input.data(), input.size()); LOG_DEBUG(Service_NVDRV, "called, gpu_va={:X}, mode={:X}", zcull_params.gpu_va, zcull_params.mode); std::memcpy(output.data(), &zcull_params, output.size()); - return 0; + return NvResult::Success; } -u32 nvhost_gpu::SetErrorNotifier(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_gpu::SetErrorNotifier(const std::vector<u8>& input, std::vector<u8>& output) { IoctlSetErrorNotifier params{}; std::memcpy(¶ms, input.data(), input.size()); LOG_WARNING(Service_NVDRV, "(STUBBED) called, offset={:X}, size={:X}, mem={:X}", params.offset, params.size, params.mem); std::memcpy(output.data(), ¶ms, output.size()); - return 0; + return NvResult::Success; } -u32 nvhost_gpu::SetChannelPriority(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_gpu::SetChannelPriority(const std::vector<u8>& input, std::vector<u8>& output) { std::memcpy(&channel_priority, input.data(), input.size()); LOG_DEBUG(Service_NVDRV, "(STUBBED) called, priority={:X}", channel_priority); - return 0; + return NvResult::Success; } -u32 nvhost_gpu::AllocGPFIFOEx2(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_gpu::AllocGPFIFOEx2(const std::vector<u8>& input, std::vector<u8>& output) { IoctlAllocGpfifoEx2 params{}; std::memcpy(¶ms, input.data(), input.size()); LOG_WARNING(Service_NVDRV, @@ -137,10 +162,10 @@ u32 nvhost_gpu::AllocGPFIFOEx2(const std::vector<u8>& input, std::vector<u8>& ou params.fence_out = channel_fence; std::memcpy(output.data(), ¶ms, output.size()); - return 0; + return NvResult::Success; } -u32 nvhost_gpu::AllocateObjectContext(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_gpu::AllocateObjectContext(const std::vector<u8>& input, std::vector<u8>& output) { IoctlAllocObjCtx params{}; std::memcpy(¶ms, input.data(), input.size()); LOG_WARNING(Service_NVDRV, "(STUBBED) called, class_num={:X}, flags={:X}", params.class_num, @@ -148,7 +173,7 @@ u32 nvhost_gpu::AllocateObjectContext(const std::vector<u8>& input, std::vector< params.obj_id = 0x0; std::memcpy(output.data(), ¶ms, output.size()); - return 0; + return NvResult::Success; } static std::vector<Tegra::CommandHeader> BuildWaitCommandList(Fence fence) { @@ -192,8 +217,8 @@ static std::vector<Tegra::CommandHeader> BuildIncrementWithWfiCommandList(Fence return result; } -u32 nvhost_gpu::SubmitGPFIFOImpl(IoctlSubmitGpfifo& params, std::vector<u8>& output, - Tegra::CommandList&& entries) { +NvResult nvhost_gpu::SubmitGPFIFOImpl(IoctlSubmitGpfifo& params, std::vector<u8>& output, + Tegra::CommandList&& entries) { LOG_TRACE(Service_NVDRV, "called, gpfifo={:X}, num_entries={:X}, flags={:X}", params.address, params.num_entries, params.flags.raw); @@ -227,69 +252,70 @@ u32 nvhost_gpu::SubmitGPFIFOImpl(IoctlSubmitGpfifo& params, std::vector<u8>& out } std::memcpy(output.data(), ¶ms, sizeof(IoctlSubmitGpfifo)); - return 0; + return NvResult::Success; } -u32 nvhost_gpu::SubmitGPFIFO(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_gpu::SubmitGPFIFOBase(const std::vector<u8>& input, std::vector<u8>& output, + bool kickoff) { if (input.size() < sizeof(IoctlSubmitGpfifo)) { UNIMPLEMENTED(); + return NvResult::InvalidSize; } IoctlSubmitGpfifo params{}; std::memcpy(¶ms, input.data(), sizeof(IoctlSubmitGpfifo)); - Tegra::CommandList entries(params.num_entries); - std::memcpy(entries.command_lists.data(), &input[sizeof(IoctlSubmitGpfifo)], - params.num_entries * sizeof(Tegra::CommandListHeader)); + + if (kickoff) { + system.Memory().ReadBlock(params.address, entries.command_lists.data(), + params.num_entries * sizeof(Tegra::CommandListHeader)); + } else { + std::memcpy(entries.command_lists.data(), &input[sizeof(IoctlSubmitGpfifo)], + params.num_entries * sizeof(Tegra::CommandListHeader)); + } return SubmitGPFIFOImpl(params, output, std::move(entries)); } -u32 nvhost_gpu::KickoffPB(const std::vector<u8>& input, std::vector<u8>& output, - const std::vector<u8>& input2, IoctlVersion version) { +NvResult nvhost_gpu::SubmitGPFIFOBase(const std::vector<u8>& input, + const std::vector<u8>& input_inline, + std::vector<u8>& output) { if (input.size() < sizeof(IoctlSubmitGpfifo)) { UNIMPLEMENTED(); + return NvResult::InvalidSize; } IoctlSubmitGpfifo params{}; std::memcpy(¶ms, input.data(), sizeof(IoctlSubmitGpfifo)); - Tegra::CommandList entries(params.num_entries); - if (version == IoctlVersion::Version2) { - std::memcpy(entries.command_lists.data(), input2.data(), - params.num_entries * sizeof(Tegra::CommandListHeader)); - } else { - system.Memory().ReadBlock(params.address, entries.command_lists.data(), - params.num_entries * sizeof(Tegra::CommandListHeader)); - } - + std::memcpy(entries.command_lists.data(), input_inline.data(), input_inline.size()); return SubmitGPFIFOImpl(params, output, std::move(entries)); } -u32 nvhost_gpu::GetWaitbase(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_gpu::GetWaitbase(const std::vector<u8>& input, std::vector<u8>& output) { IoctlGetWaitbase params{}; std::memcpy(¶ms, input.data(), sizeof(IoctlGetWaitbase)); LOG_INFO(Service_NVDRV, "called, unknown=0x{:X}", params.unknown); params.value = 0; // Seems to be hard coded at 0 std::memcpy(output.data(), ¶ms, output.size()); - return 0; + return NvResult::Success; } -u32 nvhost_gpu::ChannelSetTimeout(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_gpu::ChannelSetTimeout(const std::vector<u8>& input, std::vector<u8>& output) { IoctlChannelSetTimeout params{}; std::memcpy(¶ms, input.data(), sizeof(IoctlChannelSetTimeout)); LOG_INFO(Service_NVDRV, "called, timeout=0x{:X}", params.timeout); - return 0; + return NvResult::Success; } -u32 nvhost_gpu::ChannelSetTimeslice(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_gpu::ChannelSetTimeslice(const std::vector<u8>& input, std::vector<u8>& output) { IoctlSetTimeslice params{}; std::memcpy(¶ms, input.data(), sizeof(IoctlSetTimeslice)); LOG_INFO(Service_NVDRV, "called, timeslice=0x{:X}", params.timeslice); channel_timeslice = params.timeslice; - return 0; + return NvResult::Success; } } // namespace Service::Nvidia::Devices diff --git a/src/core/hle/service/nvdrv/devices/nvhost_gpu.h b/src/core/hle/service/nvdrv/devices/nvhost_gpu.h index a252fc06d..e0298b4fe 100644 --- a/src/core/hle/service/nvdrv/devices/nvhost_gpu.h +++ b/src/core/hle/service/nvdrv/devices/nvhost_gpu.h @@ -20,43 +20,19 @@ class SyncpointManager; namespace Service::Nvidia::Devices { class nvmap; -constexpr u32 NVGPU_IOCTL_MAGIC('H'); -constexpr u32 NVGPU_IOCTL_CHANNEL_SUBMIT_GPFIFO(0x8); -constexpr u32 NVGPU_IOCTL_CHANNEL_KICKOFF_PB(0x1b); - class nvhost_gpu final : public nvdevice { public: explicit nvhost_gpu(Core::System& system, std::shared_ptr<nvmap> nvmap_dev, SyncpointManager& syncpoint_manager); ~nvhost_gpu() override; - u32 ioctl(Ioctl command, const std::vector<u8>& input, const std::vector<u8>& input2, - std::vector<u8>& output, std::vector<u8>& output2, IoctlCtrl& ctrl, - IoctlVersion version) override; + NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) override; + NvResult Ioctl2(Ioctl command, const std::vector<u8>& input, + const std::vector<u8>& inline_input, std::vector<u8>& output) override; + NvResult Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output, + std::vector<u8>& inline_output) override; private: - enum class IoctlCommand : u32_le { - IocSetNVMAPfdCommand = 0x40044801, - IocAllocGPFIFOCommand = 0x40084805, - IocSetClientDataCommand = 0x40084714, - IocGetClientDataCommand = 0x80084715, - IocZCullBind = 0xc010480b, - IocSetErrorNotifierCommand = 0xC018480C, - IocChannelSetPriorityCommand = 0x4004480D, - IocEnableCommand = 0x0000480E, - IocDisableCommand = 0x0000480F, - IocPreemptCommand = 0x00004810, - IocForceResetCommand = 0x00004811, - IocEventIdControlCommand = 0x40084812, - IocGetErrorNotificationCommand = 0xC0104817, - IocAllocGPFIFOExCommand = 0x40204818, - IocAllocGPFIFOEx2Command = 0xC020481A, - IocAllocObjCtxCommand = 0xC0104809, - IocChannelGetWaitbaseCommand = 0xC0080003, - IocChannelSetTimeoutCommand = 0x40044803, - IocChannelSetTimeslice = 0xC004481D, - }; - enum class CtxObjects : u32_le { Ctx2D = 0x902D, Ctx3D = 0xB197, @@ -67,63 +43,63 @@ private: }; struct IoctlSetNvmapFD { - u32_le nvmap_fd; + s32_le nvmap_fd{}; }; static_assert(sizeof(IoctlSetNvmapFD) == 4, "IoctlSetNvmapFD is incorrect size"); struct IoctlChannelSetTimeout { - u32_le timeout; + u32_le timeout{}; }; static_assert(sizeof(IoctlChannelSetTimeout) == 4, "IoctlChannelSetTimeout is incorrect size"); struct IoctlAllocGPFIFO { - u32_le num_entries; - u32_le flags; + u32_le num_entries{}; + u32_le flags{}; }; static_assert(sizeof(IoctlAllocGPFIFO) == 8, "IoctlAllocGPFIFO is incorrect size"); struct IoctlClientData { - u64_le data; + u64_le data{}; }; static_assert(sizeof(IoctlClientData) == 8, "IoctlClientData is incorrect size"); struct IoctlZCullBind { - u64_le gpu_va; - u32_le mode; // 0=global, 1=no_ctxsw, 2=separate_buffer, 3=part_of_regular_buf + u64_le gpu_va{}; + u32_le mode{}; // 0=global, 1=no_ctxsw, 2=separate_buffer, 3=part_of_regular_buf INSERT_PADDING_WORDS(1); }; static_assert(sizeof(IoctlZCullBind) == 16, "IoctlZCullBind is incorrect size"); struct IoctlSetErrorNotifier { - u64_le offset; - u64_le size; - u32_le mem; // nvmap object handle + u64_le offset{}; + u64_le size{}; + u32_le mem{}; // nvmap object handle INSERT_PADDING_WORDS(1); }; static_assert(sizeof(IoctlSetErrorNotifier) == 24, "IoctlSetErrorNotifier is incorrect size"); struct IoctlChannelSetPriority { - u32_le priority; + u32_le priority{}; }; static_assert(sizeof(IoctlChannelSetPriority) == 4, "IoctlChannelSetPriority is incorrect size"); struct IoctlSetTimeslice { - u32_le timeslice; + u32_le timeslice{}; }; static_assert(sizeof(IoctlSetTimeslice) == 4, "IoctlSetTimeslice is incorrect size"); struct IoctlEventIdControl { - u32_le cmd; // 0=disable, 1=enable, 2=clear - u32_le id; + u32_le cmd{}; // 0=disable, 1=enable, 2=clear + u32_le id{}; }; static_assert(sizeof(IoctlEventIdControl) == 8, "IoctlEventIdControl is incorrect size"); struct IoctlGetErrorNotification { - u64_le timestamp; - u32_le info32; - u16_le info16; - u16_le status; // always 0xFFFF + u64_le timestamp{}; + u32_le info32{}; + u16_le info16{}; + u16_le status{}; // always 0xFFFF }; static_assert(sizeof(IoctlGetErrorNotification) == 16, "IoctlGetErrorNotification is incorrect size"); @@ -131,39 +107,39 @@ private: static_assert(sizeof(Fence) == 8, "Fence is incorrect size"); struct IoctlAllocGpfifoEx { - u32_le num_entries; - u32_le flags; - u32_le unk0; - u32_le unk1; - u32_le unk2; - u32_le unk3; - u32_le unk4; - u32_le unk5; + u32_le num_entries{}; + u32_le flags{}; + u32_le unk0{}; + u32_le unk1{}; + u32_le unk2{}; + u32_le unk3{}; + u32_le unk4{}; + u32_le unk5{}; }; static_assert(sizeof(IoctlAllocGpfifoEx) == 32, "IoctlAllocGpfifoEx is incorrect size"); struct IoctlAllocGpfifoEx2 { - u32_le num_entries; // in - u32_le flags; // in - u32_le unk0; // in (1 works) - Fence fence_out; // out - u32_le unk1; // in - u32_le unk2; // in - u32_le unk3; // in + u32_le num_entries{}; // in + u32_le flags{}; // in + u32_le unk0{}; // in (1 works) + Fence fence_out{}; // out + u32_le unk1{}; // in + u32_le unk2{}; // in + u32_le unk3{}; // in }; static_assert(sizeof(IoctlAllocGpfifoEx2) == 32, "IoctlAllocGpfifoEx2 is incorrect size"); struct IoctlAllocObjCtx { - u32_le class_num; // 0x902D=2d, 0xB197=3d, 0xB1C0=compute, 0xA140=kepler, 0xB0B5=DMA, - // 0xB06F=channel_gpfifo - u32_le flags; - u64_le obj_id; // (ignored) used for FREE_OBJ_CTX ioctl, which is not supported + u32_le class_num{}; // 0x902D=2d, 0xB197=3d, 0xB1C0=compute, 0xA140=kepler, 0xB0B5=DMA, + // 0xB06F=channel_gpfifo + u32_le flags{}; + u64_le obj_id{}; // (ignored) used for FREE_OBJ_CTX ioctl, which is not supported }; static_assert(sizeof(IoctlAllocObjCtx) == 16, "IoctlAllocObjCtx is incorrect size"); struct IoctlSubmitGpfifo { - u64_le address; // pointer to gpfifo entry structs - u32_le num_entries; // number of fence objects being submitted + u64_le address{}; // pointer to gpfifo entry structs + u32_le num_entries{}; // number of fence objects being submitted union { u32_le raw; BitField<0, 1, u32_le> add_wait; // append a wait sync_point to the list @@ -172,7 +148,7 @@ private: BitField<4, 1, u32_le> suppress_wfi; // suppress wait for interrupt BitField<8, 1, u32_le> increment; // increment the returned fence } flags; - Fence fence_out; // returned new fence object for others to wait on + Fence fence_out{}; // returned new fence object for others to wait on u32 AddIncrementValue() const { return flags.add_increment.Value() << 1; @@ -182,33 +158,34 @@ private: "IoctlSubmitGpfifo is incorrect size"); struct IoctlGetWaitbase { - u32 unknown; // seems to be ignored? Nintendo added this - u32 value; + u32 unknown{}; // seems to be ignored? Nintendo added this + u32 value{}; }; static_assert(sizeof(IoctlGetWaitbase) == 8, "IoctlGetWaitbase is incorrect size"); - u32_le nvmap_fd{}; + s32_le nvmap_fd{}; u64_le user_data{}; IoctlZCullBind zcull_params{}; u32_le channel_priority{}; u32_le channel_timeslice{}; - u32 SetNVMAPfd(const std::vector<u8>& input, std::vector<u8>& output); - u32 SetClientData(const std::vector<u8>& input, std::vector<u8>& output); - u32 GetClientData(const std::vector<u8>& input, std::vector<u8>& output); - u32 ZCullBind(const std::vector<u8>& input, std::vector<u8>& output); - u32 SetErrorNotifier(const std::vector<u8>& input, std::vector<u8>& output); - u32 SetChannelPriority(const std::vector<u8>& input, std::vector<u8>& output); - u32 AllocGPFIFOEx2(const std::vector<u8>& input, std::vector<u8>& output); - u32 AllocateObjectContext(const std::vector<u8>& input, std::vector<u8>& output); - u32 SubmitGPFIFOImpl(IoctlSubmitGpfifo& params, std::vector<u8>& output, - Tegra::CommandList&& entries); - u32 SubmitGPFIFO(const std::vector<u8>& input, std::vector<u8>& output); - u32 KickoffPB(const std::vector<u8>& input, std::vector<u8>& output, - const std::vector<u8>& input2, IoctlVersion version); - u32 GetWaitbase(const std::vector<u8>& input, std::vector<u8>& output); - u32 ChannelSetTimeout(const std::vector<u8>& input, std::vector<u8>& output); - u32 ChannelSetTimeslice(const std::vector<u8>& input, std::vector<u8>& output); + NvResult SetNVMAPfd(const std::vector<u8>& input, std::vector<u8>& output); + NvResult SetClientData(const std::vector<u8>& input, std::vector<u8>& output); + NvResult GetClientData(const std::vector<u8>& input, std::vector<u8>& output); + NvResult ZCullBind(const std::vector<u8>& input, std::vector<u8>& output); + NvResult SetErrorNotifier(const std::vector<u8>& input, std::vector<u8>& output); + NvResult SetChannelPriority(const std::vector<u8>& input, std::vector<u8>& output); + NvResult AllocGPFIFOEx2(const std::vector<u8>& input, std::vector<u8>& output); + NvResult AllocateObjectContext(const std::vector<u8>& input, std::vector<u8>& output); + NvResult SubmitGPFIFOImpl(IoctlSubmitGpfifo& params, std::vector<u8>& output, + Tegra::CommandList&& entries); + NvResult SubmitGPFIFOBase(const std::vector<u8>& input, std::vector<u8>& output, + bool kickoff = false); + NvResult SubmitGPFIFOBase(const std::vector<u8>& input, const std::vector<u8>& input_inline, + std::vector<u8>& output); + NvResult GetWaitbase(const std::vector<u8>& input, std::vector<u8>& output); + NvResult ChannelSetTimeout(const std::vector<u8>& input, std::vector<u8>& output); + NvResult ChannelSetTimeslice(const std::vector<u8>& input, std::vector<u8>& output); std::shared_ptr<nvmap> nvmap_dev; SyncpointManager& syncpoint_manager; diff --git a/src/core/hle/service/nvdrv/devices/nvhost_nvdec.cpp b/src/core/hle/service/nvdrv/devices/nvhost_nvdec.cpp index b6df48360..d8735491c 100644 --- a/src/core/hle/service/nvdrv/devices/nvhost_nvdec.cpp +++ b/src/core/hle/service/nvdrv/devices/nvhost_nvdec.cpp @@ -15,46 +15,58 @@ nvhost_nvdec::nvhost_nvdec(Core::System& system, std::shared_ptr<nvmap> nvmap_de : nvhost_nvdec_common(system, std::move(nvmap_dev)) {} nvhost_nvdec::~nvhost_nvdec() = default; -u32 nvhost_nvdec::ioctl(Ioctl command, const std::vector<u8>& input, const std::vector<u8>& input2, - std::vector<u8>& output, std::vector<u8>& output2, IoctlCtrl& ctrl, - IoctlVersion version) { - LOG_DEBUG(Service_NVDRV, "called, command=0x{:08X}, input_size=0x{:X}, output_size=0x{:X}", - command.raw, input.size(), output.size()); - - switch (static_cast<IoctlCommand>(command.raw)) { - case IoctlCommand::IocSetNVMAPfdCommand: - return SetNVMAPfd(input); - case IoctlCommand::IocSubmit: - return Submit(input, output); - case IoctlCommand::IocGetSyncpoint: - return GetSyncpoint(input, output); - case IoctlCommand::IocGetWaitbase: - return GetWaitbase(input, output); - case IoctlCommand::IocMapBuffer: - case IoctlCommand::IocMapBuffer2: - case IoctlCommand::IocMapBuffer3: - case IoctlCommand::IocMapBufferEx: - return MapBuffer(input, output); - case IoctlCommand::IocUnmapBufferEx: { - // This command is sent when the video stream has ended, flush all video contexts - // This is usually sent in the folowing order: vic, nvdec, vic. - // Inform the GPU to clear any remaining nvdec buffers when this is detected. - LOG_INFO(Service_NVDRV, "NVDEC video stream ended"); - Tegra::ChCommandHeaderList cmdlist(1); - cmdlist[0] = Tegra::ChCommandHeader{0xDEADB33F}; - system.GPU().PushCommandBuffer(cmdlist); - [[fallthrough]]; // fallthrough to unmap buffers - }; - case IoctlCommand::IocUnmapBuffer: - case IoctlCommand::IocUnmapBuffer2: - case IoctlCommand::IocUnmapBuffer3: - return UnmapBuffer(input, output); - case IoctlCommand::IocSetSubmitTimeout: - return SetSubmitTimeout(input, output); +NvResult nvhost_nvdec::Ioctl1(Ioctl command, const std::vector<u8>& input, + std::vector<u8>& output) { + switch (command.group) { + case 0x0: + switch (command.cmd) { + case 0x1: + return Submit(input, output); + case 0x2: + return GetSyncpoint(input, output); + case 0x3: + return GetWaitbase(input, output); + case 0x7: + return SetSubmitTimeout(input, output); + case 0x9: + return MapBuffer(input, output); + case 0xa: { + if (command.length == 0x1c) { + LOG_INFO(Service_NVDRV, "NVDEC video stream ended"); + Tegra::ChCommandHeaderList cmdlist(1); + cmdlist[0] = Tegra::ChCommandHeader{0xDEADB33F}; + system.GPU().PushCommandBuffer(cmdlist); + } + return UnmapBuffer(input, output); + } + default: + break; + } + break; + case 'H': + switch (command.cmd) { + case 0x1: + return SetNVMAPfd(input); + default: + break; + } + break; } - UNIMPLEMENTED_MSG("Unimplemented ioctl 0x{:X}", command.raw); - return 0; + UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw); + return NvResult::NotImplemented; +} + +NvResult nvhost_nvdec::Ioctl2(Ioctl command, const std::vector<u8>& input, + const std::vector<u8>& inline_input, std::vector<u8>& output) { + UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw); + return NvResult::NotImplemented; +} + +NvResult nvhost_nvdec::Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output, + std::vector<u8>& inline_output) { + UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw); + return NvResult::NotImplemented; } } // namespace Service::Nvidia::Devices diff --git a/src/core/hle/service/nvdrv/devices/nvhost_nvdec.h b/src/core/hle/service/nvdrv/devices/nvhost_nvdec.h index 102777ddd..79b8b6de1 100644 --- a/src/core/hle/service/nvdrv/devices/nvhost_nvdec.h +++ b/src/core/hle/service/nvdrv/devices/nvhost_nvdec.h @@ -14,26 +14,11 @@ public: explicit nvhost_nvdec(Core::System& system, std::shared_ptr<nvmap> nvmap_dev); ~nvhost_nvdec() override; - u32 ioctl(Ioctl command, const std::vector<u8>& input, const std::vector<u8>& input2, - std::vector<u8>& output, std::vector<u8>& output2, IoctlCtrl& ctrl, - IoctlVersion version) override; - -private: - enum class IoctlCommand : u32_le { - IocSetNVMAPfdCommand = 0x40044801, - IocSubmit = 0xC0400001, - IocGetSyncpoint = 0xC0080002, - IocGetWaitbase = 0xC0080003, - IocMapBuffer = 0xC01C0009, - IocMapBuffer2 = 0xC16C0009, - IocMapBuffer3 = 0xC15C0009, - IocMapBufferEx = 0xC0A40009, - IocUnmapBuffer = 0xC0A4000A, - IocUnmapBuffer2 = 0xC16C000A, - IocUnmapBufferEx = 0xC01C000A, - IocUnmapBuffer3 = 0xC15C000A, - IocSetSubmitTimeout = 0x40040007, - }; + NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) override; + NvResult Ioctl2(Ioctl command, const std::vector<u8>& input, + const std::vector<u8>& inline_input, std::vector<u8>& output) override; + NvResult Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output, + std::vector<u8>& inline_output) override; }; } // namespace Service::Nvidia::Devices 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 30f03f845..b49cecb42 100644 --- a/src/core/hle/service/nvdrv/devices/nvhost_nvdec_common.cpp +++ b/src/core/hle/service/nvdrv/devices/nvhost_nvdec_common.cpp @@ -36,26 +36,20 @@ std::size_t WriteVectors(std::vector<u8>& dst, const std::vector<T>& src, std::s } } // Anonymous namespace -namespace NvErrCodes { -constexpr u32 Success{}; -[[maybe_unused]] constexpr u32 OutOfMemory{static_cast<u32>(-12)}; -constexpr u32 InvalidInput{static_cast<u32>(-22)}; -} // namespace NvErrCodes - nvhost_nvdec_common::nvhost_nvdec_common(Core::System& system, std::shared_ptr<nvmap> nvmap_dev) : nvdevice(system), nvmap_dev(std::move(nvmap_dev)) {} nvhost_nvdec_common::~nvhost_nvdec_common() = default; -u32 nvhost_nvdec_common::SetNVMAPfd(const std::vector<u8>& input) { +NvResult nvhost_nvdec_common::SetNVMAPfd(const std::vector<u8>& input) { IoctlSetNvmapFD params{}; std::memcpy(¶ms, input.data(), sizeof(IoctlSetNvmapFD)); LOG_DEBUG(Service_NVDRV, "called, fd={}", params.nvmap_fd); nvmap_fd = params.nvmap_fd; - return 0; + return NvResult::Success; } -u32 nvhost_nvdec_common::Submit(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_nvdec_common::Submit(const std::vector<u8>& input, std::vector<u8>& output) { IoctlSubmit params{}; std::memcpy(¶ms, input.data(), sizeof(IoctlSubmit)); LOG_DEBUG(Service_NVDRV, "called NVDEC Submit, cmd_buffer_count={}", params.cmd_buffer_count); @@ -83,12 +77,12 @@ u32 nvhost_nvdec_common::Submit(const std::vector<u8>& input, std::vector<u8>& o for (const auto& cmd_buffer : command_buffers) { auto object = nvmap_dev->GetObject(cmd_buffer.memory_id); - ASSERT_OR_EXECUTE(object, return NvErrCodes::InvalidInput;); + ASSERT_OR_EXECUTE(object, return NvResult::InvalidState;); const auto map = FindBufferMap(object->dma_map_addr); if (!map) { LOG_ERROR(Service_NVDRV, "Tried to submit an invalid offset 0x{:X} dma 0x{:X}", object->addr, object->dma_map_addr); - return 0; + return NvResult::Success; } Tegra::ChCommandHeaderList cmdlist(cmd_buffer.word_count); gpu.MemoryManager().ReadBlock(map->StartAddr() + cmd_buffer.offset, cmdlist.data(), @@ -105,10 +99,10 @@ u32 nvhost_nvdec_common::Submit(const std::vector<u8>& input, std::vector<u8>& o offset = WriteVectors(output, syncpt_increments, offset); offset = WriteVectors(output, wait_checks, offset); - return NvErrCodes::Success; + return NvResult::Success; } -u32 nvhost_nvdec_common::GetSyncpoint(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_nvdec_common::GetSyncpoint(const std::vector<u8>& input, std::vector<u8>& output) { IoctlGetSyncpoint params{}; std::memcpy(¶ms, input.data(), sizeof(IoctlGetSyncpoint)); LOG_DEBUG(Service_NVDRV, "called GetSyncpoint, id={}", params.param); @@ -118,18 +112,18 @@ u32 nvhost_nvdec_common::GetSyncpoint(const std::vector<u8>& input, std::vector< params.value = 0; std::memcpy(output.data(), ¶ms, sizeof(IoctlGetSyncpoint)); - return NvErrCodes::Success; + return NvResult::Success; } -u32 nvhost_nvdec_common::GetWaitbase(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_nvdec_common::GetWaitbase(const std::vector<u8>& input, std::vector<u8>& output) { IoctlGetWaitbase params{}; std::memcpy(¶ms, input.data(), sizeof(IoctlGetWaitbase)); params.value = 0; // Seems to be hard coded at 0 std::memcpy(output.data(), ¶ms, sizeof(IoctlGetWaitbase)); - return 0; + return NvResult::Success; } -u32 nvhost_nvdec_common::MapBuffer(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_nvdec_common::MapBuffer(const std::vector<u8>& input, std::vector<u8>& output) { IoctlMapBuffer params{}; std::memcpy(¶ms, input.data(), sizeof(IoctlMapBuffer)); std::vector<MapBufferEntry> cmd_buffer_handles(params.num_entries); @@ -143,7 +137,7 @@ u32 nvhost_nvdec_common::MapBuffer(const std::vector<u8>& input, std::vector<u8> if (!object) { LOG_ERROR(Service_NVDRV, "invalid cmd_buffer nvmap_handle={:X}", cmf_buff.map_handle); std::memcpy(output.data(), ¶ms, output.size()); - return NvErrCodes::InvalidInput; + return NvResult::InvalidState; } if (object->dma_map_addr == 0) { // NVDEC and VIC memory is in the 32-bit address space @@ -165,10 +159,10 @@ u32 nvhost_nvdec_common::MapBuffer(const std::vector<u8>& input, std::vector<u8> std::memcpy(output.data() + sizeof(IoctlMapBuffer), cmd_buffer_handles.data(), cmd_buffer_handles.size() * sizeof(MapBufferEntry)); - return NvErrCodes::Success; + return NvResult::Success; } -u32 nvhost_nvdec_common::UnmapBuffer(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_nvdec_common::UnmapBuffer(const std::vector<u8>& input, std::vector<u8>& output) { IoctlMapBuffer params{}; std::memcpy(¶ms, input.data(), sizeof(IoctlMapBuffer)); std::vector<MapBufferEntry> cmd_buffer_handles(params.num_entries); @@ -181,7 +175,7 @@ u32 nvhost_nvdec_common::UnmapBuffer(const std::vector<u8>& input, std::vector<u if (!object) { LOG_ERROR(Service_NVDRV, "invalid cmd_buffer nvmap_handle={:X}", cmf_buff.map_handle); std::memcpy(output.data(), ¶ms, output.size()); - return NvErrCodes::InvalidInput; + return NvResult::InvalidState; } if (const auto size{RemoveBufferMap(object->dma_map_addr)}; size) { gpu.MemoryManager().Unmap(object->dma_map_addr, *size); @@ -193,13 +187,14 @@ u32 nvhost_nvdec_common::UnmapBuffer(const std::vector<u8>& input, std::vector<u object->dma_map_addr = 0; } std::memset(output.data(), 0, output.size()); - return NvErrCodes::Success; + return NvResult::Success; } -u32 nvhost_nvdec_common::SetSubmitTimeout(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_nvdec_common::SetSubmitTimeout(const std::vector<u8>& input, + std::vector<u8>& output) { std::memcpy(&submit_timeout, input.data(), input.size()); LOG_WARNING(Service_NVDRV, "(STUBBED) called"); - return NvErrCodes::Success; + return NvResult::Success; } std::optional<nvhost_nvdec_common::BufferMap> nvhost_nvdec_common::FindBufferMap( 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 c249c5349..86ba3a4d1 100644 --- a/src/core/hle/service/nvdrv/devices/nvhost_nvdec_common.h +++ b/src/core/hle/service/nvdrv/devices/nvhost_nvdec_common.h @@ -18,9 +18,37 @@ public: explicit nvhost_nvdec_common(Core::System& system, std::shared_ptr<nvmap> nvmap_dev); ~nvhost_nvdec_common() override; - virtual u32 ioctl(Ioctl command, const std::vector<u8>& input, const std::vector<u8>& input2, - std::vector<u8>& output, std::vector<u8>& output2, IoctlCtrl& ctrl, - IoctlVersion version) = 0; + /** + * Handles an ioctl1 request. + * @param command The ioctl command id. + * @param input A buffer containing the input data for the ioctl. + * @param output A buffer where the output data will be written to. + * @returns The result code of the ioctl. + */ + virtual NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, + std::vector<u8>& output) = 0; + + /** + * Handles an ioctl2 request. + * @param command The ioctl command id. + * @param input A buffer containing the input data for the ioctl. + * @param inline_input A buffer containing the input data for the ioctl which has been inlined. + * @param output A buffer where the output data will be written to. + * @returns The result code of the ioctl. + */ + virtual NvResult Ioctl2(Ioctl command, const std::vector<u8>& input, + const std::vector<u8>& inline_input, std::vector<u8>& output) = 0; + + /** + * Handles an ioctl3 request. + * @param command The ioctl command id. + * @param input A buffer containing the input data for the ioctl. + * @param output A buffer where the output data will be written to. + * @param inline_output A buffer where the inlined output data will be written to. + * @returns The result code of the ioctl. + */ + virtual NvResult Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output, + std::vector<u8>& inline_output) = 0; protected: class BufferMap final { @@ -63,102 +91,102 @@ protected: }; struct IoctlSetNvmapFD { - u32_le nvmap_fd; + s32_le nvmap_fd{}; }; static_assert(sizeof(IoctlSetNvmapFD) == 4, "IoctlSetNvmapFD is incorrect size"); struct IoctlSubmitCommandBuffer { - u32_le id; - u32_le offset; - u32_le count; + u32_le id{}; + u32_le offset{}; + u32_le count{}; }; static_assert(sizeof(IoctlSubmitCommandBuffer) == 0xC, "IoctlSubmitCommandBuffer is incorrect size"); struct IoctlSubmit { - u32_le cmd_buffer_count; - u32_le relocation_count; - u32_le syncpoint_count; - u32_le fence_count; + u32_le cmd_buffer_count{}; + u32_le relocation_count{}; + u32_le syncpoint_count{}; + u32_le fence_count{}; }; static_assert(sizeof(IoctlSubmit) == 0x10, "IoctlSubmit has incorrect size"); struct CommandBuffer { - s32 memory_id; - u32 offset; - s32 word_count; + s32 memory_id{}; + u32 offset{}; + s32 word_count{}; }; static_assert(sizeof(CommandBuffer) == 0xC, "CommandBuffer has incorrect size"); struct Reloc { - s32 cmdbuffer_memory; - s32 cmdbuffer_offset; - s32 target; - s32 target_offset; + s32 cmdbuffer_memory{}; + s32 cmdbuffer_offset{}; + s32 target{}; + s32 target_offset{}; }; static_assert(sizeof(Reloc) == 0x10, "CommandBuffer has incorrect size"); struct SyncptIncr { - u32 id; - u32 increments; + u32 id{}; + u32 increments{}; }; static_assert(sizeof(SyncptIncr) == 0x8, "CommandBuffer has incorrect size"); struct Fence { - u32 id; - u32 value; + u32 id{}; + u32 value{}; }; static_assert(sizeof(Fence) == 0x8, "CommandBuffer has incorrect size"); struct IoctlGetSyncpoint { // Input - u32_le param; + u32_le param{}; // Output - u32_le value; + u32_le value{}; }; static_assert(sizeof(IoctlGetSyncpoint) == 8, "IocGetIdParams has wrong size"); struct IoctlGetWaitbase { - u32_le unknown; // seems to be ignored? Nintendo added this - u32_le value; + u32_le unknown{}; // seems to be ignored? Nintendo added this + u32_le value{}; }; static_assert(sizeof(IoctlGetWaitbase) == 0x8, "IoctlGetWaitbase is incorrect size"); struct IoctlMapBuffer { - u32_le num_entries; - u32_le data_address; // Ignored by the driver. - u32_le attach_host_ch_das; + u32_le num_entries{}; + u32_le data_address{}; // Ignored by the driver. + u32_le attach_host_ch_das{}; }; static_assert(sizeof(IoctlMapBuffer) == 0x0C, "IoctlMapBuffer is incorrect size"); struct IocGetIdParams { // Input - u32_le param; + u32_le param{}; // Output - u32_le value; + u32_le value{}; }; static_assert(sizeof(IocGetIdParams) == 8, "IocGetIdParams has wrong size"); // Used for mapping and unmapping command buffers struct MapBufferEntry { - u32_le map_handle; - u32_le map_address; + u32_le map_handle{}; + u32_le map_address{}; }; static_assert(sizeof(IoctlMapBuffer) == 0x0C, "IoctlMapBuffer is incorrect size"); /// Ioctl command implementations - u32 SetNVMAPfd(const std::vector<u8>& input); - u32 Submit(const std::vector<u8>& input, std::vector<u8>& output); - u32 GetSyncpoint(const std::vector<u8>& input, std::vector<u8>& output); - u32 GetWaitbase(const std::vector<u8>& input, std::vector<u8>& output); - u32 MapBuffer(const std::vector<u8>& input, std::vector<u8>& output); - u32 UnmapBuffer(const std::vector<u8>& input, std::vector<u8>& output); - u32 SetSubmitTimeout(const std::vector<u8>& input, std::vector<u8>& output); + NvResult SetNVMAPfd(const std::vector<u8>& input); + NvResult Submit(const std::vector<u8>& input, std::vector<u8>& output); + NvResult GetSyncpoint(const std::vector<u8>& input, std::vector<u8>& output); + NvResult GetWaitbase(const std::vector<u8>& input, std::vector<u8>& output); + NvResult MapBuffer(const std::vector<u8>& input, std::vector<u8>& output); + NvResult UnmapBuffer(const std::vector<u8>& input, std::vector<u8>& output); + NvResult SetSubmitTimeout(const std::vector<u8>& input, std::vector<u8>& output); std::optional<BufferMap> FindBufferMap(GPUVAddr gpu_addr) const; void AddBufferMap(GPUVAddr gpu_addr, std::size_t size, VAddr cpu_addr, bool is_allocated); std::optional<std::size_t> RemoveBufferMap(GPUVAddr gpu_addr); - u32_le nvmap_fd{}; + s32_le nvmap_fd{}; u32_le submit_timeout{}; std::shared_ptr<nvmap> nvmap_dev; diff --git a/src/core/hle/service/nvdrv/devices/nvhost_nvjpg.cpp b/src/core/hle/service/nvdrv/devices/nvhost_nvjpg.cpp index 96e7b7dab..2d06955c0 100644 --- a/src/core/hle/service/nvdrv/devices/nvhost_nvjpg.cpp +++ b/src/core/hle/service/nvdrv/devices/nvhost_nvjpg.cpp @@ -13,28 +13,44 @@ namespace Service::Nvidia::Devices { nvhost_nvjpg::nvhost_nvjpg(Core::System& system) : nvdevice(system) {} nvhost_nvjpg::~nvhost_nvjpg() = default; -u32 nvhost_nvjpg::ioctl(Ioctl command, const std::vector<u8>& input, const std::vector<u8>& input2, - std::vector<u8>& output, std::vector<u8>& output2, IoctlCtrl& ctrl, - IoctlVersion version) { - LOG_DEBUG(Service_NVDRV, "called, command=0x{:08X}, input_size=0x{:X}, output_size=0x{:X}", - command.raw, input.size(), output.size()); - - switch (static_cast<IoctlCommand>(command.raw)) { - case IoctlCommand::IocSetNVMAPfdCommand: - return SetNVMAPfd(input, output); +NvResult nvhost_nvjpg::Ioctl1(Ioctl command, const std::vector<u8>& input, + std::vector<u8>& output) { + switch (command.group) { + case 'H': + switch (command.cmd) { + case 0x1: + return SetNVMAPfd(input, output); + default: + break; + } + break; + default: + break; } - UNIMPLEMENTED_MSG("Unimplemented ioctl"); - return 0; + UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw); + return NvResult::NotImplemented; } -u32 nvhost_nvjpg::SetNVMAPfd(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvhost_nvjpg::Ioctl2(Ioctl command, const std::vector<u8>& input, + const std::vector<u8>& inline_input, std::vector<u8>& output) { + UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw); + return NvResult::NotImplemented; +} + +NvResult nvhost_nvjpg::Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output, + std::vector<u8>& inline_output) { + UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw); + return NvResult::NotImplemented; +} + +NvResult nvhost_nvjpg::SetNVMAPfd(const std::vector<u8>& input, std::vector<u8>& output) { IoctlSetNvmapFD params{}; std::memcpy(¶ms, input.data(), input.size()); LOG_DEBUG(Service_NVDRV, "called, fd={}", params.nvmap_fd); nvmap_fd = params.nvmap_fd; - return 0; + return NvResult::Success; } } // namespace Service::Nvidia::Devices diff --git a/src/core/hle/service/nvdrv/devices/nvhost_nvjpg.h b/src/core/hle/service/nvdrv/devices/nvhost_nvjpg.h index 98dcac52f..43948d18d 100644 --- a/src/core/hle/service/nvdrv/devices/nvhost_nvjpg.h +++ b/src/core/hle/service/nvdrv/devices/nvhost_nvjpg.h @@ -16,23 +16,21 @@ public: explicit nvhost_nvjpg(Core::System& system); ~nvhost_nvjpg() override; - u32 ioctl(Ioctl command, const std::vector<u8>& input, const std::vector<u8>& input2, - std::vector<u8>& output, std::vector<u8>& output2, IoctlCtrl& ctrl, - IoctlVersion version) override; + NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) override; + NvResult Ioctl2(Ioctl command, const std::vector<u8>& input, + const std::vector<u8>& inline_input, std::vector<u8>& output) override; + NvResult Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output, + std::vector<u8>& inline_output) override; private: - enum class IoctlCommand : u32_le { - IocSetNVMAPfdCommand = 0x40044801, - }; - struct IoctlSetNvmapFD { - u32_le nvmap_fd; + s32_le nvmap_fd{}; }; static_assert(sizeof(IoctlSetNvmapFD) == 4, "IoctlSetNvmapFD is incorrect size"); - u32_le nvmap_fd{}; + s32_le nvmap_fd{}; - u32 SetNVMAPfd(const std::vector<u8>& input, std::vector<u8>& output); + NvResult SetNVMAPfd(const std::vector<u8>& input, std::vector<u8>& output); }; } // namespace Service::Nvidia::Devices diff --git a/src/core/hle/service/nvdrv/devices/nvhost_vic.cpp b/src/core/hle/service/nvdrv/devices/nvhost_vic.cpp index 60db54d00..805fe86ae 100644 --- a/src/core/hle/service/nvdrv/devices/nvhost_vic.cpp +++ b/src/core/hle/service/nvdrv/devices/nvhost_vic.cpp @@ -15,36 +15,50 @@ nvhost_vic::nvhost_vic(Core::System& system, std::shared_ptr<nvmap> nvmap_dev) nvhost_vic::~nvhost_vic() = default; -u32 nvhost_vic::ioctl(Ioctl command, const std::vector<u8>& input, const std::vector<u8>& input2, - std::vector<u8>& output, std::vector<u8>& output2, IoctlCtrl& ctrl, - IoctlVersion version) { - LOG_DEBUG(Service_NVDRV, "called, command=0x{:08X}, input_size=0x{:X}, output_size=0x{:X}", - command.raw, input.size(), output.size()); - - switch (static_cast<IoctlCommand>(command.raw)) { - case IoctlCommand::IocSetNVMAPfdCommand: - return SetNVMAPfd(input); - case IoctlCommand::IocSubmit: - return Submit(input, output); - case IoctlCommand::IocGetSyncpoint: - return GetSyncpoint(input, output); - case IoctlCommand::IocGetWaitbase: - return GetWaitbase(input, output); - case IoctlCommand::IocMapBuffer: - case IoctlCommand::IocMapBuffer2: - case IoctlCommand::IocMapBuffer3: - case IoctlCommand::IocMapBuffer4: - case IoctlCommand::IocMapBufferEx: - return MapBuffer(input, output); - case IoctlCommand::IocUnmapBuffer: - case IoctlCommand::IocUnmapBuffer2: - case IoctlCommand::IocUnmapBuffer3: - case IoctlCommand::IocUnmapBufferEx: - return UnmapBuffer(input, output); +NvResult nvhost_vic::Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) { + switch (command.group) { + case 0x0: + switch (command.cmd) { + case 0x1: + return Submit(input, output); + case 0x2: + return GetSyncpoint(input, output); + case 0x3: + return GetWaitbase(input, output); + case 0x9: + return MapBuffer(input, output); + case 0xa: + return UnmapBuffer(input, output); + default: + break; + } + break; + case 'H': + switch (command.cmd) { + case 0x1: + return SetNVMAPfd(input); + default: + break; + } + break; + default: + break; } - UNIMPLEMENTED_MSG("Unimplemented ioctl 0x{:X}", command.raw); - return 0; + UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw); + return NvResult::NotImplemented; +} + +NvResult nvhost_vic::Ioctl2(Ioctl command, const std::vector<u8>& input, + const std::vector<u8>& inline_input, std::vector<u8>& output) { + UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw); + return NvResult::NotImplemented; +} + +NvResult nvhost_vic::Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output, + std::vector<u8>& inline_output) { + UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw); + return NvResult::NotImplemented; } } // namespace Service::Nvidia::Devices diff --git a/src/core/hle/service/nvdrv/devices/nvhost_vic.h b/src/core/hle/service/nvdrv/devices/nvhost_vic.h index f975b190c..b2e11f4d4 100644 --- a/src/core/hle/service/nvdrv/devices/nvhost_vic.h +++ b/src/core/hle/service/nvdrv/devices/nvhost_vic.h @@ -13,25 +13,11 @@ class nvhost_vic final : public nvhost_nvdec_common { public: explicit nvhost_vic(Core::System& system, std::shared_ptr<nvmap> nvmap_dev); ~nvhost_vic(); - u32 ioctl(Ioctl command, const std::vector<u8>& input, const std::vector<u8>& input2, - std::vector<u8>& output, std::vector<u8>& output2, IoctlCtrl& ctrl, - IoctlVersion version) override; -private: - enum class IoctlCommand : u32_le { - IocSetNVMAPfdCommand = 0x40044801, - IocSubmit = 0xC0400001, - IocGetSyncpoint = 0xC0080002, - IocGetWaitbase = 0xC0080003, - IocMapBuffer = 0xC01C0009, - IocMapBuffer2 = 0xC0340009, - IocMapBuffer3 = 0xC0140009, - IocMapBuffer4 = 0xC00C0009, - IocMapBufferEx = 0xC03C0009, - IocUnmapBuffer = 0xC03C000A, - IocUnmapBuffer2 = 0xC034000A, - IocUnmapBuffer3 = 0xC00C000A, - IocUnmapBufferEx = 0xC01C000A, - }; + NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) override; + NvResult Ioctl2(Ioctl command, const std::vector<u8>& input, + const std::vector<u8>& inline_input, std::vector<u8>& output) override; + NvResult Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output, + std::vector<u8>& inline_output) override; }; } // namespace Service::Nvidia::Devices diff --git a/src/core/hle/service/nvdrv/devices/nvmap.cpp b/src/core/hle/service/nvdrv/devices/nvmap.cpp index 9436e16ad..4015a2740 100644 --- a/src/core/hle/service/nvdrv/devices/nvmap.cpp +++ b/src/core/hle/service/nvdrv/devices/nvmap.cpp @@ -11,13 +11,6 @@ namespace Service::Nvidia::Devices { -namespace NvErrCodes { -enum { - OperationNotPermitted = -1, - InvalidValue = -22, -}; -} - nvmap::nvmap(Core::System& system) : nvdevice(system) { // Handle 0 appears to be used when remapping, so we create a placeholder empty nvmap object to // represent this. @@ -26,6 +19,46 @@ nvmap::nvmap(Core::System& system) : nvdevice(system) { nvmap::~nvmap() = default; +NvResult nvmap::Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) { + switch (command.group) { + case 0x1: + switch (command.cmd) { + case 0x1: + return IocCreate(input, output); + case 0x3: + return IocFromId(input, output); + case 0x4: + return IocAlloc(input, output); + case 0x5: + return IocFree(input, output); + case 0x9: + return IocParam(input, output); + case 0xe: + return IocGetId(input, output); + default: + break; + } + break; + default: + break; + } + + UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw); + return NvResult::NotImplemented; +} + +NvResult nvmap::Ioctl2(Ioctl command, const std::vector<u8>& input, + const std::vector<u8>& inline_input, std::vector<u8>& output) { + UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw); + return NvResult::NotImplemented; +} + +NvResult nvmap::Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output, + std::vector<u8>& inline_output) { + UNIMPLEMENTED_MSG("Unimplemented ioctl={:08X}", command.raw); + return NvResult::NotImplemented; +} + VAddr nvmap::GetObjectAddress(u32 handle) const { auto object = GetObject(handle); ASSERT(object); @@ -33,28 +66,6 @@ VAddr nvmap::GetObjectAddress(u32 handle) const { return object->addr; } -u32 nvmap::ioctl(Ioctl command, const std::vector<u8>& input, const std::vector<u8>& input2, - std::vector<u8>& output, std::vector<u8>& output2, IoctlCtrl& ctrl, - IoctlVersion version) { - switch (static_cast<IoctlCommand>(command.raw)) { - case IoctlCommand::Create: - return IocCreate(input, output); - case IoctlCommand::Alloc: - return IocAlloc(input, output); - case IoctlCommand::GetId: - return IocGetId(input, output); - case IoctlCommand::FromId: - return IocFromId(input, output); - case IoctlCommand::Param: - return IocParam(input, output); - case IoctlCommand::Free: - return IocFree(input, output); - } - - UNIMPLEMENTED_MSG("Unimplemented ioctl"); - return 0; -} - u32 nvmap::CreateObject(u32 size) { // Create a new nvmap object and obtain a handle to it. auto object = std::make_shared<Object>(); @@ -70,35 +81,35 @@ u32 nvmap::CreateObject(u32 size) { return handle; } -u32 nvmap::IocCreate(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvmap::IocCreate(const std::vector<u8>& input, std::vector<u8>& output) { IocCreateParams params; std::memcpy(¶ms, input.data(), sizeof(params)); LOG_DEBUG(Service_NVDRV, "size=0x{:08X}", params.size); if (!params.size) { LOG_ERROR(Service_NVDRV, "Size is 0"); - return static_cast<u32>(NvErrCodes::InvalidValue); + return NvResult::BadValue; } params.handle = CreateObject(params.size); std::memcpy(output.data(), ¶ms, sizeof(params)); - return 0; + return NvResult::Success; } -u32 nvmap::IocAlloc(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvmap::IocAlloc(const std::vector<u8>& input, std::vector<u8>& output) { IocAllocParams params; std::memcpy(¶ms, input.data(), sizeof(params)); LOG_DEBUG(Service_NVDRV, "called, addr={:X}", params.addr); if (!params.handle) { LOG_ERROR(Service_NVDRV, "Handle is 0"); - return static_cast<u32>(NvErrCodes::InvalidValue); + return NvResult::BadValue; } if ((params.align - 1) & params.align) { LOG_ERROR(Service_NVDRV, "Incorrect alignment used, alignment={:08X}", params.align); - return static_cast<u32>(NvErrCodes::InvalidValue); + return NvResult::BadValue; } const u32 min_alignment = 0x1000; @@ -109,12 +120,12 @@ u32 nvmap::IocAlloc(const std::vector<u8>& input, std::vector<u8>& output) { auto object = GetObject(params.handle); if (!object) { LOG_ERROR(Service_NVDRV, "Object does not exist, handle={:08X}", params.handle); - return static_cast<u32>(NvErrCodes::InvalidValue); + return NvResult::BadValue; } if (object->status == Object::Status::Allocated) { LOG_ERROR(Service_NVDRV, "Object is already allocated, handle={:08X}", params.handle); - return static_cast<u32>(NvErrCodes::OperationNotPermitted); + return NvResult::InsufficientMemory; } object->flags = params.flags; @@ -124,10 +135,10 @@ u32 nvmap::IocAlloc(const std::vector<u8>& input, std::vector<u8>& output) { object->status = Object::Status::Allocated; std::memcpy(output.data(), ¶ms, sizeof(params)); - return 0; + return NvResult::Success; } -u32 nvmap::IocGetId(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvmap::IocGetId(const std::vector<u8>& input, std::vector<u8>& output) { IocGetIdParams params; std::memcpy(¶ms, input.data(), sizeof(params)); @@ -135,22 +146,22 @@ u32 nvmap::IocGetId(const std::vector<u8>& input, std::vector<u8>& output) { if (!params.handle) { LOG_ERROR(Service_NVDRV, "Handle is zero"); - return static_cast<u32>(NvErrCodes::InvalidValue); + return NvResult::BadValue; } auto object = GetObject(params.handle); if (!object) { LOG_ERROR(Service_NVDRV, "Object does not exist, handle={:08X}", params.handle); - return static_cast<u32>(NvErrCodes::OperationNotPermitted); + return NvResult::BadValue; } params.id = object->id; std::memcpy(output.data(), ¶ms, sizeof(params)); - return 0; + return NvResult::Success; } -u32 nvmap::IocFromId(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvmap::IocFromId(const std::vector<u8>& input, std::vector<u8>& output) { IocFromIdParams params; std::memcpy(¶ms, input.data(), sizeof(params)); @@ -160,13 +171,13 @@ u32 nvmap::IocFromId(const std::vector<u8>& input, std::vector<u8>& output) { [&](const auto& entry) { return entry.second->id == params.id; }); if (itr == handles.end()) { LOG_ERROR(Service_NVDRV, "Object does not exist, handle={:08X}", params.handle); - return static_cast<u32>(NvErrCodes::InvalidValue); + return NvResult::BadValue; } auto& object = itr->second; if (object->status != Object::Status::Allocated) { LOG_ERROR(Service_NVDRV, "Object is not allocated, handle={:08X}", params.handle); - return static_cast<u32>(NvErrCodes::InvalidValue); + return NvResult::BadValue; } itr->second->refcount++; @@ -175,10 +186,10 @@ u32 nvmap::IocFromId(const std::vector<u8>& input, std::vector<u8>& output) { params.handle = itr->first; std::memcpy(output.data(), ¶ms, sizeof(params)); - return 0; + return NvResult::Success; } -u32 nvmap::IocParam(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvmap::IocParam(const std::vector<u8>& input, std::vector<u8>& output) { enum class ParamTypes { Size = 1, Alignment = 2, Base = 3, Heap = 4, Kind = 5, Compr = 6 }; IocParamParams params; @@ -189,12 +200,12 @@ u32 nvmap::IocParam(const std::vector<u8>& input, std::vector<u8>& output) { auto object = GetObject(params.handle); if (!object) { LOG_ERROR(Service_NVDRV, "Object does not exist, handle={:08X}", params.handle); - return static_cast<u32>(NvErrCodes::InvalidValue); + return NvResult::BadValue; } if (object->status != Object::Status::Allocated) { LOG_ERROR(Service_NVDRV, "Object is not allocated, handle={:08X}", params.handle); - return static_cast<u32>(NvErrCodes::OperationNotPermitted); + return NvResult::BadValue; } switch (static_cast<ParamTypes>(params.param)) { @@ -216,10 +227,10 @@ u32 nvmap::IocParam(const std::vector<u8>& input, std::vector<u8>& output) { } std::memcpy(output.data(), ¶ms, sizeof(params)); - return 0; + return NvResult::Success; } -u32 nvmap::IocFree(const std::vector<u8>& input, std::vector<u8>& output) { +NvResult nvmap::IocFree(const std::vector<u8>& input, std::vector<u8>& output) { // TODO(Subv): These flags are unconfirmed. enum FreeFlags { Freed = 0, @@ -234,14 +245,14 @@ u32 nvmap::IocFree(const std::vector<u8>& input, std::vector<u8>& output) { auto itr = handles.find(params.handle); if (itr == handles.end()) { LOG_ERROR(Service_NVDRV, "Object does not exist, handle={:08X}", params.handle); - return static_cast<u32>(NvErrCodes::InvalidValue); + return NvResult::BadValue; } if (!itr->second->refcount) { LOG_ERROR( Service_NVDRV, "There is no references to this object. The object is already freed. handle={:08X}", params.handle); - return static_cast<u32>(NvErrCodes::InvalidValue); + return NvResult::BadValue; } itr->second->refcount--; @@ -261,7 +272,7 @@ u32 nvmap::IocFree(const std::vector<u8>& input, std::vector<u8>& output) { handles.erase(params.handle); std::memcpy(output.data(), ¶ms, sizeof(params)); - return 0; + return NvResult::Success; } } // namespace Service::Nvidia::Devices diff --git a/src/core/hle/service/nvdrv/devices/nvmap.h b/src/core/hle/service/nvdrv/devices/nvmap.h index 04b9ef540..4484bd79f 100644 --- a/src/core/hle/service/nvdrv/devices/nvmap.h +++ b/src/core/hle/service/nvdrv/devices/nvmap.h @@ -19,13 +19,15 @@ public: explicit nvmap(Core::System& system); ~nvmap() override; + NvResult Ioctl1(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output) override; + NvResult Ioctl2(Ioctl command, const std::vector<u8>& input, + const std::vector<u8>& inline_input, std::vector<u8>& output) override; + NvResult Ioctl3(Ioctl command, const std::vector<u8>& input, std::vector<u8>& output, + std::vector<u8>& inline_output) override; + /// Returns the allocated address of an nvmap object given its handle. VAddr GetObjectAddress(u32 handle) const; - u32 ioctl(Ioctl command, const std::vector<u8>& input, const std::vector<u8>& input2, - std::vector<u8>& output, std::vector<u8>& output2, IoctlCtrl& ctrl, - IoctlVersion version) override; - /// Represents an nvmap object. struct Object { enum class Status { Created, Allocated }; @@ -58,76 +60,68 @@ private: /// Mapping of currently allocated handles to the objects they represent. std::unordered_map<u32, std::shared_ptr<Object>> handles; - enum class IoctlCommand : u32 { - Create = 0xC0080101, - FromId = 0xC0080103, - Alloc = 0xC0200104, - Free = 0xC0180105, - Param = 0xC00C0109, - GetId = 0xC008010E, - }; struct IocCreateParams { // Input - u32_le size; + u32_le size{}; // Output - u32_le handle; + u32_le handle{}; }; static_assert(sizeof(IocCreateParams) == 8, "IocCreateParams has wrong size"); struct IocFromIdParams { // Input - u32_le id; + u32_le id{}; // Output - u32_le handle; + u32_le handle{}; }; static_assert(sizeof(IocFromIdParams) == 8, "IocFromIdParams has wrong size"); struct IocAllocParams { // Input - u32_le handle; - u32_le heap_mask; - u32_le flags; - u32_le align; - u8 kind; + u32_le handle{}; + u32_le heap_mask{}; + u32_le flags{}; + u32_le align{}; + u8 kind{}; INSERT_PADDING_BYTES(7); - u64_le addr; + u64_le addr{}; }; static_assert(sizeof(IocAllocParams) == 32, "IocAllocParams has wrong size"); struct IocFreeParams { - u32_le handle; + u32_le handle{}; INSERT_PADDING_BYTES(4); - u64_le address; - u32_le size; - u32_le flags; + u64_le address{}; + u32_le size{}; + u32_le flags{}; }; static_assert(sizeof(IocFreeParams) == 24, "IocFreeParams has wrong size"); struct IocParamParams { // Input - u32_le handle; - u32_le param; + u32_le handle{}; + u32_le param{}; // Output - u32_le result; + u32_le result{}; }; static_assert(sizeof(IocParamParams) == 12, "IocParamParams has wrong size"); struct IocGetIdParams { // Output - u32_le id; + u32_le id{}; // Input - u32_le handle; + u32_le handle{}; }; static_assert(sizeof(IocGetIdParams) == 8, "IocGetIdParams has wrong size"); u32 CreateObject(u32 size); - u32 IocCreate(const std::vector<u8>& input, std::vector<u8>& output); - u32 IocAlloc(const std::vector<u8>& input, std::vector<u8>& output); - u32 IocGetId(const std::vector<u8>& input, std::vector<u8>& output); - u32 IocFromId(const std::vector<u8>& input, std::vector<u8>& output); - u32 IocParam(const std::vector<u8>& input, std::vector<u8>& output); - u32 IocFree(const std::vector<u8>& input, std::vector<u8>& output); + NvResult IocCreate(const std::vector<u8>& input, std::vector<u8>& output); + NvResult IocAlloc(const std::vector<u8>& input, std::vector<u8>& output); + NvResult IocGetId(const std::vector<u8>& input, std::vector<u8>& output); + NvResult IocFromId(const std::vector<u8>& input, std::vector<u8>& output); + NvResult IocParam(const std::vector<u8>& input, std::vector<u8>& output); + NvResult IocFree(const std::vector<u8>& input, std::vector<u8>& output); }; } // namespace Service::Nvidia::Devices diff --git a/src/core/hle/service/nvdrv/interface.cpp b/src/core/hle/service/nvdrv/interface.cpp index 88fbfa9b0..f6c38e853 100644 --- a/src/core/hle/service/nvdrv/interface.cpp +++ b/src/core/hle/service/nvdrv/interface.cpp @@ -23,124 +23,170 @@ void NVDRV::SignalGPUInterruptSyncpt(const u32 syncpoint_id, const u32 value) { void NVDRV::Open(Kernel::HLERequestContext& ctx) { LOG_DEBUG(Service_NVDRV, "called"); + if (!is_initialized) { + ServiceError(ctx, NvResult::NotInitialized); + LOG_ERROR(Service_NVDRV, "NvServices is not initalized!"); + return; + } + const auto& buffer = ctx.ReadBuffer(); - std::string device_name(buffer.begin(), buffer.end()); + const std::string device_name(buffer.begin(), buffer.end()); + DeviceFD fd = nvdrv->Open(device_name); - u32 fd = nvdrv->Open(device_name); IPC::ResponseBuilder rb{ctx, 4}; rb.Push(RESULT_SUCCESS); - rb.Push<u32>(fd); - rb.Push<u32>(0); + rb.Push<DeviceFD>(fd); + rb.PushEnum(fd != INVALID_NVDRV_FD ? NvResult::Success : NvResult::FileOperationFailed); +} + +void NVDRV::ServiceError(Kernel::HLERequestContext& ctx, NvResult result) { + IPC::ResponseBuilder rb{ctx, 3}; + rb.Push(RESULT_SUCCESS); + rb.PushEnum(result); } -void NVDRV::IoctlBase(Kernel::HLERequestContext& ctx, IoctlVersion version) { +void NVDRV::Ioctl1(Kernel::HLERequestContext& ctx) { IPC::RequestParser rp{ctx}; - u32 fd = rp.Pop<u32>(); - u32 command = rp.Pop<u32>(); - - /// Ioctl 3 has 2 outputs, first in the input params, second is the result - std::vector<u8> output(ctx.GetWriteBufferSize(0)); - std::vector<u8> output2; - if (version == IoctlVersion::Version3) { - output2.resize((ctx.GetWriteBufferSize(1))); + const auto fd = rp.Pop<DeviceFD>(); + const auto command = rp.PopRaw<Ioctl>(); + LOG_DEBUG(Service_NVDRV, "called fd={}, ioctl=0x{:08X}", fd, command.raw); + + if (!is_initialized) { + ServiceError(ctx, NvResult::NotInitialized); + LOG_ERROR(Service_NVDRV, "NvServices is not initalized!"); + return; } - /// Ioctl2 has 2 inputs. It's used to pass data directly instead of providing a pointer. - /// KickOfPB uses this - auto input = ctx.ReadBuffer(0); + // Check device + std::vector<u8> output_buffer(ctx.GetWriteBufferSize(0)); + const auto input_buffer = ctx.ReadBuffer(0); - std::vector<u8> input2; - if (version == IoctlVersion::Version2) { - input2 = ctx.ReadBuffer(1); - } + const auto nv_result = nvdrv->Ioctl1(fd, command, input_buffer, output_buffer); - IoctlCtrl ctrl{}; - - u32 result = nvdrv->Ioctl(fd, command, input, input2, output, output2, ctrl, version); - - if (ctrl.must_delay) { - ctrl.fresh_call = false; - ctx.SleepClientThread( - "NVServices::DelayedResponse", ctrl.timeout, - [=, this](std::shared_ptr<Kernel::Thread> thread, Kernel::HLERequestContext& ctx_, - Kernel::ThreadWakeupReason reason) { - IoctlCtrl ctrl2{ctrl}; - std::vector<u8> tmp_output = output; - std::vector<u8> tmp_output2 = output2; - const u32 ioctl_result = nvdrv->Ioctl(fd, command, input, input2, tmp_output, - tmp_output2, ctrl2, version); - ctx_.WriteBuffer(tmp_output, 0); - if (version == IoctlVersion::Version3) { - ctx_.WriteBuffer(tmp_output2, 1); - } - IPC::ResponseBuilder rb{ctx_, 3}; - rb.Push(RESULT_SUCCESS); - rb.Push(ioctl_result); - }, - nvdrv->GetEventWriteable(ctrl.event_id)); - } else { - ctx.WriteBuffer(output); - if (version == IoctlVersion::Version3) { - ctx.WriteBuffer(output2, 1); - } + if (command.is_out != 0) { + ctx.WriteBuffer(output_buffer); } + IPC::ResponseBuilder rb{ctx, 3}; rb.Push(RESULT_SUCCESS); - rb.Push(result); -} - -void NVDRV::Ioctl(Kernel::HLERequestContext& ctx) { - LOG_DEBUG(Service_NVDRV, "called"); - IoctlBase(ctx, IoctlVersion::Version1); + rb.PushEnum(nv_result); } void NVDRV::Ioctl2(Kernel::HLERequestContext& ctx) { - LOG_DEBUG(Service_NVDRV, "called"); - IoctlBase(ctx, IoctlVersion::Version2); + IPC::RequestParser rp{ctx}; + const auto fd = rp.Pop<DeviceFD>(); + const auto command = rp.PopRaw<Ioctl>(); + LOG_DEBUG(Service_NVDRV, "called fd={}, ioctl=0x{:08X}", fd, command.raw); + + if (!is_initialized) { + ServiceError(ctx, NvResult::NotInitialized); + LOG_ERROR(Service_NVDRV, "NvServices is not initalized!"); + return; + } + + const auto input_buffer = ctx.ReadBuffer(0); + const auto input_inlined_buffer = ctx.ReadBuffer(1); + std::vector<u8> output_buffer(ctx.GetWriteBufferSize(0)); + + const auto nv_result = + nvdrv->Ioctl2(fd, command, input_buffer, input_inlined_buffer, output_buffer); + + if (command.is_out != 0) { + ctx.WriteBuffer(output_buffer); + } + + IPC::ResponseBuilder rb{ctx, 3}; + rb.Push(RESULT_SUCCESS); + rb.PushEnum(nv_result); } void NVDRV::Ioctl3(Kernel::HLERequestContext& ctx) { - LOG_DEBUG(Service_NVDRV, "called"); - IoctlBase(ctx, IoctlVersion::Version3); + IPC::RequestParser rp{ctx}; + const auto fd = rp.Pop<DeviceFD>(); + const auto command = rp.PopRaw<Ioctl>(); + LOG_DEBUG(Service_NVDRV, "called fd={}, ioctl=0x{:08X}", fd, command.raw); + + if (!is_initialized) { + ServiceError(ctx, NvResult::NotInitialized); + LOG_ERROR(Service_NVDRV, "NvServices is not initalized!"); + return; + } + + const auto input_buffer = ctx.ReadBuffer(0); + std::vector<u8> output_buffer(ctx.GetWriteBufferSize(0)); + std::vector<u8> output_buffer_inline(ctx.GetWriteBufferSize(1)); + + const auto nv_result = + nvdrv->Ioctl3(fd, command, input_buffer, output_buffer, output_buffer_inline); + + if (command.is_out != 0) { + ctx.WriteBuffer(output_buffer, 0); + ctx.WriteBuffer(output_buffer_inline, 1); + } + + IPC::ResponseBuilder rb{ctx, 3}; + rb.Push(RESULT_SUCCESS); + rb.PushEnum(nv_result); } void NVDRV::Close(Kernel::HLERequestContext& ctx) { LOG_DEBUG(Service_NVDRV, "called"); - IPC::RequestParser rp{ctx}; - u32 fd = rp.Pop<u32>(); + if (!is_initialized) { + ServiceError(ctx, NvResult::NotInitialized); + LOG_ERROR(Service_NVDRV, "NvServices is not initalized!"); + return; + } - auto result = nvdrv->Close(fd); + IPC::RequestParser rp{ctx}; + const auto fd = rp.Pop<DeviceFD>(); + const auto result = nvdrv->Close(fd); - IPC::ResponseBuilder rb{ctx, 2}; - rb.Push(result); + IPC::ResponseBuilder rb{ctx, 3}; + rb.Push(RESULT_SUCCESS); + rb.PushEnum(result); } void NVDRV::Initialize(Kernel::HLERequestContext& ctx) { LOG_WARNING(Service_NVDRV, "(STUBBED) called"); + is_initialized = true; + IPC::ResponseBuilder rb{ctx, 3}; rb.Push(RESULT_SUCCESS); - rb.Push<u32>(0); + rb.PushEnum(NvResult::Success); } void NVDRV::QueryEvent(Kernel::HLERequestContext& ctx) { IPC::RequestParser rp{ctx}; - u32 fd = rp.Pop<u32>(); - // TODO(Blinkhawk): Figure the meaning of the flag at bit 16 - u32 event_id = rp.Pop<u32>() & 0x000000FF; + const auto fd = rp.Pop<DeviceFD>(); + const auto event_id = rp.Pop<u32>() & 0x00FF; LOG_WARNING(Service_NVDRV, "(STUBBED) called, fd={:X}, event_id={:X}", fd, event_id); - IPC::ResponseBuilder rb{ctx, 3, 1}; - rb.Push(RESULT_SUCCESS); + if (!is_initialized) { + ServiceError(ctx, NvResult::NotInitialized); + LOG_ERROR(Service_NVDRV, "NvServices is not initalized!"); + return; + } + + const auto nv_result = nvdrv->VerifyFD(fd); + if (nv_result != NvResult::Success) { + LOG_ERROR(Service_NVDRV, "Invalid FD specified DeviceFD={}!", fd); + ServiceError(ctx, nv_result); + return; + } + if (event_id < MaxNvEvents) { + IPC::ResponseBuilder rb{ctx, 3, 1}; + rb.Push(RESULT_SUCCESS); auto event = nvdrv->GetEvent(event_id); event->Clear(); rb.PushCopyObjects(event); - rb.Push<u32>(NvResult::Success); + rb.PushEnum(NvResult::Success); } else { - rb.Push<u32>(0); - rb.Push<u32>(NvResult::BadParameter); + IPC::ResponseBuilder rb{ctx, 3}; + rb.Push(RESULT_SUCCESS); + rb.PushEnum(NvResult::BadParameter); } } @@ -151,7 +197,7 @@ void NVDRV::SetAruid(Kernel::HLERequestContext& ctx) { IPC::ResponseBuilder rb{ctx, 3}; rb.Push(RESULT_SUCCESS); - rb.Push<u32>(0); + rb.PushEnum(NvResult::Success); } void NVDRV::SetGraphicsFirmwareMemoryMarginEnabled(Kernel::HLERequestContext& ctx) { @@ -164,8 +210,9 @@ void NVDRV::SetGraphicsFirmwareMemoryMarginEnabled(Kernel::HLERequestContext& ct void NVDRV::GetStatus(Kernel::HLERequestContext& ctx) { LOG_WARNING(Service_NVDRV, "(STUBBED) called"); - IPC::ResponseBuilder rb{ctx, 2}; + IPC::ResponseBuilder rb{ctx, 3}; rb.Push(RESULT_SUCCESS); + rb.PushEnum(NvResult::Success); } void NVDRV::DumpGraphicsMemoryInfo(Kernel::HLERequestContext& ctx) { @@ -181,7 +228,7 @@ NVDRV::NVDRV(std::shared_ptr<Module> nvdrv, const char* name) : ServiceFramework(name), nvdrv(std::move(nvdrv)) { static const FunctionInfo functions[] = { {0, &NVDRV::Open, "Open"}, - {1, &NVDRV::Ioctl, "Ioctl"}, + {1, &NVDRV::Ioctl1, "Ioctl"}, {2, &NVDRV::Close, "Close"}, {3, &NVDRV::Initialize, "Initialize"}, {4, &NVDRV::QueryEvent, "QueryEvent"}, diff --git a/src/core/hle/service/nvdrv/interface.h b/src/core/hle/service/nvdrv/interface.h index 72e17a728..e05f905ae 100644 --- a/src/core/hle/service/nvdrv/interface.h +++ b/src/core/hle/service/nvdrv/interface.h @@ -23,7 +23,7 @@ public: private: void Open(Kernel::HLERequestContext& ctx); - void Ioctl(Kernel::HLERequestContext& ctx); + void Ioctl1(Kernel::HLERequestContext& ctx); void Ioctl2(Kernel::HLERequestContext& ctx); void Ioctl3(Kernel::HLERequestContext& ctx); void Close(Kernel::HLERequestContext& ctx); @@ -33,11 +33,13 @@ private: void SetGraphicsFirmwareMemoryMarginEnabled(Kernel::HLERequestContext& ctx); void GetStatus(Kernel::HLERequestContext& ctx); void DumpGraphicsMemoryInfo(Kernel::HLERequestContext& ctx); - void IoctlBase(Kernel::HLERequestContext& ctx, IoctlVersion version); + + void ServiceError(Kernel::HLERequestContext& ctx, NvResult result); std::shared_ptr<Module> nvdrv; u64 pid{}; + bool is_initialized{}; }; } // namespace Service::Nvidia diff --git a/src/core/hle/service/nvdrv/nvdata.h b/src/core/hle/service/nvdrv/nvdata.h index 529b03471..3294bc0e7 100644 --- a/src/core/hle/service/nvdrv/nvdata.h +++ b/src/core/hle/service/nvdrv/nvdata.h @@ -1,12 +1,16 @@ #pragma once #include <array> +#include "common/bit_field.h" #include "common/common_types.h" namespace Service::Nvidia { constexpr u32 MaxSyncPoints = 192; constexpr u32 MaxNvEvents = 64; +using DeviceFD = s32; + +constexpr DeviceFD INVALID_NVDRV_FD = -1; struct Fence { s32 id; @@ -20,11 +24,61 @@ struct MultiFence { std::array<Fence, 4> fences; }; -enum NvResult : u32 { - Success = 0, - BadParameter = 4, - Timeout = 5, - ResourceError = 15, +enum class NvResult : u32 { + Success = 0x0, + NotImplemented = 0x1, + NotSupported = 0x2, + NotInitialized = 0x3, + BadParameter = 0x4, + Timeout = 0x5, + InsufficientMemory = 0x6, + ReadOnlyAttribute = 0x7, + InvalidState = 0x8, + InvalidAddress = 0x9, + InvalidSize = 0xA, + BadValue = 0xB, + AlreadyAllocated = 0xD, + Busy = 0xE, + ResourceError = 0xF, + CountMismatch = 0x10, + OverFlow = 0x11, + InsufficientTransferMemory = 0x1000, + InsufficientVideoMemory = 0x10000, + BadSurfaceColorScheme = 0x10001, + InvalidSurface = 0x10002, + SurfaceNotSupported = 0x10003, + DispInitFailed = 0x20000, + DispAlreadyAttached = 0x20001, + DispTooManyDisplays = 0x20002, + DispNoDisplaysAttached = 0x20003, + DispModeNotSupported = 0x20004, + DispNotFound = 0x20005, + DispAttachDissallowed = 0x20006, + DispTypeNotSupported = 0x20007, + DispAuthenticationFailed = 0x20008, + DispNotAttached = 0x20009, + DispSamePwrState = 0x2000A, + DispEdidFailure = 0x2000B, + DispDsiReadAckError = 0x2000C, + DispDsiReadInvalidResp = 0x2000D, + FileWriteFailed = 0x30000, + FileReadFailed = 0x30001, + EndOfFile = 0x30002, + FileOperationFailed = 0x30003, + DirOperationFailed = 0x30004, + EndOfDirList = 0x30005, + ConfigVarNotFound = 0x30006, + InvalidConfigVar = 0x30007, + LibraryNotFound = 0x30008, + SymbolNotFound = 0x30009, + MemoryMapFailed = 0x3000A, + IoctlFailed = 0x3000F, + AccessDenied = 0x30010, + DeviceNotFound = 0x30011, + KernelDriverNotFound = 0x30012, + FileNotFound = 0x30013, + PathAlreadyExists = 0x30014, + ModuleNotPresent = 0xA000E, }; enum class EventState { @@ -34,21 +88,13 @@ enum class EventState { Busy = 3, }; -enum class IoctlVersion : u32 { - Version1, - Version2, - Version3, -}; - -struct IoctlCtrl { - // First call done to the servioce for services that call itself again after a call. - bool fresh_call{true}; - // Tells the Ioctl Wrapper that it must delay the IPC response and send the thread to sleep - bool must_delay{}; - // Timeout for the delay - s64 timeout{}; - // NV Event Id - s32 event_id{-1}; +union Ioctl { + u32_le raw; + BitField<0, 8, u32> cmd; + BitField<8, 8, u32> group; + BitField<16, 14, u32> length; + BitField<30, 1, u32> is_in; + BitField<31, 1, u32> is_out; }; } // namespace Service::Nvidia diff --git a/src/core/hle/service/nvdrv/nvdrv.cpp b/src/core/hle/service/nvdrv/nvdrv.cpp index 046a1f28c..bdbbedd0d 100644 --- a/src/core/hle/service/nvdrv/nvdrv.cpp +++ b/src/core/hle/service/nvdrv/nvdrv.cpp @@ -62,36 +62,101 @@ Module::Module(Core::System& system) : syncpoint_manager{system.GPU()} { Module::~Module() = default; -u32 Module::Open(const std::string& device_name) { - ASSERT_MSG(devices.find(device_name) != devices.end(), "Trying to open unknown device {}", - device_name); +NvResult Module::VerifyFD(DeviceFD fd) const { + if (fd < 0) { + LOG_ERROR(Service_NVDRV, "Invalid DeviceFD={}!", fd); + return NvResult::InvalidState; + } + + if (open_files.find(fd) == open_files.end()) { + LOG_ERROR(Service_NVDRV, "Could not find DeviceFD={}!", fd); + return NvResult::NotImplemented; + } + + return NvResult::Success; +} + +DeviceFD Module::Open(const std::string& device_name) { + if (devices.find(device_name) == devices.end()) { + LOG_ERROR(Service_NVDRV, "Trying to open unknown device {}", device_name); + return INVALID_NVDRV_FD; + } auto device = devices[device_name]; - const u32 fd = next_fd++; + const DeviceFD fd = next_fd++; open_files[fd] = std::move(device); return fd; } -u32 Module::Ioctl(u32 fd, u32 command, const std::vector<u8>& input, const std::vector<u8>& input2, - std::vector<u8>& output, std::vector<u8>& output2, IoctlCtrl& ctrl, - IoctlVersion version) { - auto itr = open_files.find(fd); - ASSERT_MSG(itr != open_files.end(), "Tried to talk to an invalid device"); +NvResult Module::Ioctl1(DeviceFD fd, Ioctl command, const std::vector<u8>& input, + std::vector<u8>& output) { + if (fd < 0) { + LOG_ERROR(Service_NVDRV, "Invalid DeviceFD={}!", fd); + return NvResult::InvalidState; + } - auto& device = itr->second; - return device->ioctl({command}, input, input2, output, output2, ctrl, version); + const auto itr = open_files.find(fd); + + if (itr == open_files.end()) { + LOG_ERROR(Service_NVDRV, "Could not find DeviceFD={}!", fd); + return NvResult::NotImplemented; + } + + return itr->second->Ioctl1(command, input, output); } -ResultCode Module::Close(u32 fd) { - auto itr = open_files.find(fd); - ASSERT_MSG(itr != open_files.end(), "Tried to talk to an invalid device"); +NvResult Module::Ioctl2(DeviceFD fd, Ioctl command, const std::vector<u8>& input, + const std::vector<u8>& inline_input, std::vector<u8>& output) { + if (fd < 0) { + LOG_ERROR(Service_NVDRV, "Invalid DeviceFD={}!", fd); + return NvResult::InvalidState; + } + + const auto itr = open_files.find(fd); + + if (itr == open_files.end()) { + LOG_ERROR(Service_NVDRV, "Could not find DeviceFD={}!", fd); + return NvResult::NotImplemented; + } + + return itr->second->Ioctl2(command, input, inline_input, output); +} + +NvResult Module::Ioctl3(DeviceFD fd, Ioctl command, const std::vector<u8>& input, + std::vector<u8>& output, std::vector<u8>& inline_output) { + if (fd < 0) { + LOG_ERROR(Service_NVDRV, "Invalid DeviceFD={}!", fd); + return NvResult::InvalidState; + } + + const auto itr = open_files.find(fd); + + if (itr == open_files.end()) { + LOG_ERROR(Service_NVDRV, "Could not find DeviceFD={}!", fd); + return NvResult::NotImplemented; + } + + return itr->second->Ioctl3(command, input, output, inline_output); +} + +NvResult Module::Close(DeviceFD fd) { + if (fd < 0) { + LOG_ERROR(Service_NVDRV, "Invalid DeviceFD={}!", fd); + return NvResult::InvalidState; + } + + const auto itr = open_files.find(fd); + + if (itr == open_files.end()) { + LOG_ERROR(Service_NVDRV, "Could not find DeviceFD={}!", fd); + return NvResult::NotImplemented; + } open_files.erase(itr); - // TODO(flerovium): return correct result code if operation failed. - return RESULT_SUCCESS; + return NvResult::Success; } void Module::SignalSyncpt(const u32 syncpoint_id, const u32 value) { diff --git a/src/core/hle/service/nvdrv/nvdrv.h b/src/core/hle/service/nvdrv/nvdrv.h index f3d863dac..7654bb026 100644 --- a/src/core/hle/service/nvdrv/nvdrv.h +++ b/src/core/hle/service/nvdrv/nvdrv.h @@ -112,14 +112,23 @@ public: return std::static_pointer_cast<T>(itr->second); } + NvResult VerifyFD(DeviceFD fd) const; + /// Opens a device node and returns a file descriptor to it. - u32 Open(const std::string& device_name); + DeviceFD Open(const std::string& device_name); + /// Sends an ioctl command to the specified file descriptor. - u32 Ioctl(u32 fd, u32 command, const std::vector<u8>& input, const std::vector<u8>& input2, - std::vector<u8>& output, std::vector<u8>& output2, IoctlCtrl& ctrl, - IoctlVersion version); + NvResult Ioctl1(DeviceFD fd, Ioctl command, const std::vector<u8>& input, + std::vector<u8>& output); + + NvResult Ioctl2(DeviceFD fd, Ioctl command, const std::vector<u8>& input, + const std::vector<u8>& inline_input, std::vector<u8>& output); + + NvResult Ioctl3(DeviceFD fd, Ioctl command, const std::vector<u8>& input, + std::vector<u8>& output, std::vector<u8>& inline_output); + /// Closes a device file descriptor and returns operation success. - ResultCode Close(u32 fd); + NvResult Close(DeviceFD fd); void SignalSyncpt(const u32 syncpoint_id, const u32 value); @@ -132,10 +141,10 @@ private: SyncpointManager syncpoint_manager; /// Id to use for the next open file descriptor. - u32 next_fd = 1; + DeviceFD next_fd = 1; /// Mapping of file descriptors to the devices they reference. - std::unordered_map<u32, std::shared_ptr<Devices::nvdevice>> open_files; + std::unordered_map<DeviceFD, std::shared_ptr<Devices::nvdevice>> open_files; /// Mapping of device node names to their implementation. std::unordered_map<std::string, std::shared_ptr<Devices::nvdevice>> devices; diff --git a/src/core/hle/service/olsc/olsc.cpp b/src/core/hle/service/olsc/olsc.cpp new file mode 100644 index 000000000..aad4ca706 --- /dev/null +++ b/src/core/hle/service/olsc/olsc.cpp @@ -0,0 +1,69 @@ +// Copyright 2020 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include "core/hle/ipc_helpers.h" +#include "core/hle/kernel/hle_ipc.h" +#include "core/hle/service/olsc/olsc.h" +#include "core/hle/service/service.h" +#include "core/hle/service/sm/sm.h" + +namespace Service::OLSC { + +class OLSC final : public ServiceFramework<OLSC> { +public: + explicit OLSC() : ServiceFramework{"olsc:u"} { + // clang-format off + static const FunctionInfo functions[] = { + {0, &OLSC::Initialize, "Initialize"}, + {10, nullptr, "VerifySaveDataBackupLicenseAsync"}, + {13, nullptr, "GetSaveDataBackupSetting"}, + {14, &OLSC::SetSaveDataBackupSettingEnabled, "SetSaveDataBackupSettingEnabled"}, + {15, nullptr, "SetCustomData"}, + {16, nullptr, "DeleteSaveDataBackupSetting"}, + {18, nullptr, "GetSaveDataBackupInfoCache"}, + {19, nullptr, "UpdateSaveDataBackupInfoCacheAsync"}, + {22, nullptr, "DeleteSaveDataBackupAsync"}, + {25, nullptr, "ListDownloadableSaveDataBackupInfoAsync"}, + {26, nullptr, "DownloadSaveDataBackupAsync"}, + {9010, nullptr, "VerifySaveDataBackupLicenseAsyncForDebug"}, + {9013, nullptr, "GetSaveDataBackupSettingForDebug"}, + {9014, nullptr, "SetSaveDataBackupSettingEnabledForDebug"}, + {9015, nullptr, "SetCustomDataForDebug"}, + {9016, nullptr, "DeleteSaveDataBackupSettingForDebug"}, + {9018, nullptr, "GetSaveDataBackupInfoCacheForDebug"}, + {9019, nullptr, "UpdateSaveDataBackupInfoCacheAsyncForDebug"}, + {9022, nullptr, "DeleteSaveDataBackupAsyncForDebug"}, + {9025, nullptr, "ListDownloadableSaveDataBackupInfoAsyncForDebug"}, + {9026, nullptr, "DownloadSaveDataBackupAsyncForDebug"}, + }; + // clang-format on + + RegisterHandlers(functions); + } + +private: + void Initialize(Kernel::HLERequestContext& ctx) { + LOG_WARNING(Service_OLSC, "(STUBBED) called"); + + initialized = true; + + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(RESULT_SUCCESS); + } + + void SetSaveDataBackupSettingEnabled(Kernel::HLERequestContext& ctx) { + LOG_WARNING(Service_OLSC, "(STUBBED) called"); + + IPC::ResponseBuilder rb{ctx, 2}; + rb.Push(RESULT_SUCCESS); + } + + bool initialized{}; +}; + +void InstallInterfaces(SM::ServiceManager& service_manager) { + std::make_shared<OLSC>()->InstallAsService(service_manager); +} + +} // namespace Service::OLSC diff --git a/src/core/hle/service/olsc/olsc.h b/src/core/hle/service/olsc/olsc.h new file mode 100644 index 000000000..edee4376b --- /dev/null +++ b/src/core/hle/service/olsc/olsc.h @@ -0,0 +1,16 @@ +// Copyright 2020 yuzu emulator team +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +namespace Service::SM { +class ServiceManager; +} + +namespace Service::OLSC { + +/// Registers all SSL services with the specified service manager. +void InstallInterfaces(SM::ServiceManager& service_manager); + +} // namespace Service::OLSC diff --git a/src/core/hle/service/service.cpp b/src/core/hle/service/service.cpp index ba9159ee0..fbfda2d5b 100644 --- a/src/core/hle/service/service.cpp +++ b/src/core/hle/service/service.cpp @@ -51,6 +51,7 @@ #include "core/hle/service/ns/ns.h" #include "core/hle/service/nvdrv/nvdrv.h" #include "core/hle/service/nvflinger/nvflinger.h" +#include "core/hle/service/olsc/olsc.h" #include "core/hle/service/pcie/pcie.h" #include "core/hle/service/pctl/module.h" #include "core/hle/service/pcv/pcv.h" @@ -231,6 +232,7 @@ void Init(std::shared_ptr<SM::ServiceManager>& sm, Core::System& system) { NPNS::InstallInterfaces(*sm); NS::InstallInterfaces(*sm, system); Nvidia::InstallInterfaces(*sm, *nv_flinger, system); + OLSC::InstallInterfaces(*sm); PCIe::InstallInterfaces(*sm); PCTL::InstallInterfaces(*sm); PCV::InstallInterfaces(*sm); diff --git a/src/input_common/gcadapter/gc_poller.cpp b/src/input_common/gcadapter/gc_poller.cpp index fe57c13a5..d95574bb5 100644 --- a/src/input_common/gcadapter/gc_poller.cpp +++ b/src/input_common/gcadapter/gc_poller.cpp @@ -302,8 +302,8 @@ public: bool SetRumblePlay(f32 amp_low, f32 freq_low, f32 amp_high, f32 freq_high) const override { const auto mean_amplitude = (amp_low + amp_high) * 0.5f; - const auto processed_amplitude = static_cast<u8>( - pow(mean_amplitude, 0.5f) * (3.0f - 2.0f * pow(mean_amplitude, 0.15f)) * 0x8); + const auto processed_amplitude = + static_cast<u8>((mean_amplitude + std::pow(mean_amplitude, 0.3f)) * 0.5f * 0x8); return gcadapter->RumblePlay(port, processed_amplitude); } diff --git a/src/input_common/sdl/sdl_impl.cpp b/src/input_common/sdl/sdl_impl.cpp index 8c48bb861..c395d96cf 100644 --- a/src/input_common/sdl/sdl_impl.cpp +++ b/src/input_common/sdl/sdl_impl.cpp @@ -402,8 +402,7 @@ public: bool SetRumblePlay(f32 amp_low, f32 freq_low, f32 amp_high, f32 freq_high) const override { const auto process_amplitude = [](f32 amplitude) { - return static_cast<u16>(std::pow(amplitude, 0.5f) * - (3.0f - 2.0f * std::pow(amplitude, 0.15f)) * 0xFFFF); + return static_cast<u16>((amplitude + std::pow(amplitude, 0.3f)) * 0.5f * 0xFFFF); }; const auto processed_amp_low = process_amplitude(amp_low); diff --git a/src/video_core/engines/maxwell_3d.cpp b/src/video_core/engines/maxwell_3d.cpp index 57ebc785f..6287df633 100644 --- a/src/video_core/engines/maxwell_3d.cpp +++ b/src/video_core/engines/maxwell_3d.cpp @@ -124,6 +124,112 @@ void Maxwell3D::InitializeRegisterDefaults() { mme_inline[MAXWELL3D_REG_INDEX(index_array.count)] = true; } +void Maxwell3D::ProcessMacro(u32 method, const u32* base_start, u32 amount, bool is_last_call) { + if (executing_macro == 0) { + // A macro call must begin by writing the macro method's register, not its argument. + ASSERT_MSG((method % 2) == 0, + "Can't start macro execution by writing to the ARGS register"); + executing_macro = method; + } + + macro_params.insert(macro_params.end(), base_start, base_start + amount); + + // Call the macro when there are no more parameters in the command buffer + if (is_last_call) { + CallMacroMethod(executing_macro, macro_params); + macro_params.clear(); + } +} + +u32 Maxwell3D::ProcessShadowRam(u32 method, u32 argument) { + // Keep track of the register value in shadow_state when requested. + const auto control = shadow_state.shadow_ram_control; + if (control == Regs::ShadowRamControl::Track || + control == Regs::ShadowRamControl::TrackWithFilter) { + shadow_state.reg_array[method] = argument; + return argument; + } + if (control == Regs::ShadowRamControl::Replay) { + return shadow_state.reg_array[method]; + } + return argument; +} + +void Maxwell3D::ProcessDirtyRegisters(u32 method, u32 argument) { + if (regs.reg_array[method] == argument) { + return; + } + regs.reg_array[method] = argument; + + for (const auto& table : dirty.tables) { + dirty.flags[table[method]] = true; + } +} + +void Maxwell3D::ProcessMethodCall(u32 method, u32 argument, u32 nonshadow_argument, + bool is_last_call) { + switch (method) { + case MAXWELL3D_REG_INDEX(wait_for_idle): + return rasterizer->WaitForIdle(); + case MAXWELL3D_REG_INDEX(shadow_ram_control): + shadow_state.shadow_ram_control = static_cast<Regs::ShadowRamControl>(nonshadow_argument); + return; + case MAXWELL3D_REG_INDEX(macros.data): + return macro_engine->AddCode(regs.macros.upload_address, argument); + case MAXWELL3D_REG_INDEX(macros.bind): + return ProcessMacroBind(argument); + case MAXWELL3D_REG_INDEX(firmware[4]): + return ProcessFirmwareCall4(); + case MAXWELL3D_REG_INDEX(const_buffer.cb_data[0]): + case MAXWELL3D_REG_INDEX(const_buffer.cb_data[1]): + case MAXWELL3D_REG_INDEX(const_buffer.cb_data[2]): + case MAXWELL3D_REG_INDEX(const_buffer.cb_data[3]): + case MAXWELL3D_REG_INDEX(const_buffer.cb_data[4]): + case MAXWELL3D_REG_INDEX(const_buffer.cb_data[5]): + case MAXWELL3D_REG_INDEX(const_buffer.cb_data[6]): + case MAXWELL3D_REG_INDEX(const_buffer.cb_data[7]): + case MAXWELL3D_REG_INDEX(const_buffer.cb_data[8]): + case MAXWELL3D_REG_INDEX(const_buffer.cb_data[9]): + case MAXWELL3D_REG_INDEX(const_buffer.cb_data[10]): + case MAXWELL3D_REG_INDEX(const_buffer.cb_data[11]): + case MAXWELL3D_REG_INDEX(const_buffer.cb_data[12]): + case MAXWELL3D_REG_INDEX(const_buffer.cb_data[13]): + case MAXWELL3D_REG_INDEX(const_buffer.cb_data[14]): + case MAXWELL3D_REG_INDEX(const_buffer.cb_data[15]): + return StartCBData(method); + case MAXWELL3D_REG_INDEX(cb_bind[0]): + return ProcessCBBind(0); + case MAXWELL3D_REG_INDEX(cb_bind[1]): + return ProcessCBBind(1); + case MAXWELL3D_REG_INDEX(cb_bind[2]): + return ProcessCBBind(2); + case MAXWELL3D_REG_INDEX(cb_bind[3]): + return ProcessCBBind(3); + case MAXWELL3D_REG_INDEX(cb_bind[4]): + return ProcessCBBind(4); + case MAXWELL3D_REG_INDEX(draw.vertex_end_gl): + return DrawArrays(); + case MAXWELL3D_REG_INDEX(clear_buffers): + return ProcessClearBuffers(); + case MAXWELL3D_REG_INDEX(query.query_get): + return ProcessQueryGet(); + case MAXWELL3D_REG_INDEX(condition.mode): + return ProcessQueryCondition(); + case MAXWELL3D_REG_INDEX(counter_reset): + return ProcessCounterReset(); + case MAXWELL3D_REG_INDEX(sync_info): + return ProcessSyncPoint(); + case MAXWELL3D_REG_INDEX(exec_upload): + return upload_state.ProcessExec(regs.exec_upload.linear != 0); + case MAXWELL3D_REG_INDEX(data_upload): + upload_state.ProcessData(argument, is_last_call); + if (is_last_call) { + OnMemoryWrite(); + } + return; + } +} + void Maxwell3D::CallMacroMethod(u32 method, const std::vector<u32>& parameters) { // Reset the current macro. executing_macro = 0; @@ -157,142 +263,16 @@ void Maxwell3D::CallMethod(u32 method, u32 method_argument, bool is_last_call) { // Methods after 0xE00 are special, they're actually triggers for some microcode that was // uploaded to the GPU during initialization. if (method >= MacroRegistersStart) { - // We're trying to execute a macro - if (executing_macro == 0) { - // A macro call must begin by writing the macro method's register, not its argument. - ASSERT_MSG((method % 2) == 0, - "Can't start macro execution by writing to the ARGS register"); - executing_macro = method; - } - - macro_params.push_back(method_argument); - - // Call the macro when there are no more parameters in the command buffer - if (is_last_call) { - CallMacroMethod(executing_macro, macro_params); - macro_params.clear(); - } + ProcessMacro(method, &method_argument, 1, is_last_call); return; } ASSERT_MSG(method < Regs::NUM_REGS, "Invalid Maxwell3D register, increase the size of the Regs structure"); - u32 arg = method_argument; - // Keep track of the register value in shadow_state when requested. - if (shadow_state.shadow_ram_control == Regs::ShadowRamControl::Track || - shadow_state.shadow_ram_control == Regs::ShadowRamControl::TrackWithFilter) { - shadow_state.reg_array[method] = arg; - } else if (shadow_state.shadow_ram_control == Regs::ShadowRamControl::Replay) { - arg = shadow_state.reg_array[method]; - } - - if (regs.reg_array[method] != arg) { - regs.reg_array[method] = arg; - - for (const auto& table : dirty.tables) { - dirty.flags[table[method]] = true; - } - } - - switch (method) { - case MAXWELL3D_REG_INDEX(wait_for_idle): { - rasterizer->WaitForIdle(); - break; - } - case MAXWELL3D_REG_INDEX(shadow_ram_control): { - shadow_state.shadow_ram_control = static_cast<Regs::ShadowRamControl>(method_argument); - break; - } - case MAXWELL3D_REG_INDEX(macros.data): { - macro_engine->AddCode(regs.macros.upload_address, arg); - break; - } - case MAXWELL3D_REG_INDEX(macros.bind): { - ProcessMacroBind(arg); - break; - } - case MAXWELL3D_REG_INDEX(firmware[4]): { - ProcessFirmwareCall4(); - break; - } - case MAXWELL3D_REG_INDEX(const_buffer.cb_data[0]): - case MAXWELL3D_REG_INDEX(const_buffer.cb_data[1]): - case MAXWELL3D_REG_INDEX(const_buffer.cb_data[2]): - case MAXWELL3D_REG_INDEX(const_buffer.cb_data[3]): - case MAXWELL3D_REG_INDEX(const_buffer.cb_data[4]): - case MAXWELL3D_REG_INDEX(const_buffer.cb_data[5]): - case MAXWELL3D_REG_INDEX(const_buffer.cb_data[6]): - case MAXWELL3D_REG_INDEX(const_buffer.cb_data[7]): - case MAXWELL3D_REG_INDEX(const_buffer.cb_data[8]): - case MAXWELL3D_REG_INDEX(const_buffer.cb_data[9]): - case MAXWELL3D_REG_INDEX(const_buffer.cb_data[10]): - case MAXWELL3D_REG_INDEX(const_buffer.cb_data[11]): - case MAXWELL3D_REG_INDEX(const_buffer.cb_data[12]): - case MAXWELL3D_REG_INDEX(const_buffer.cb_data[13]): - case MAXWELL3D_REG_INDEX(const_buffer.cb_data[14]): - case MAXWELL3D_REG_INDEX(const_buffer.cb_data[15]): { - StartCBData(method); - break; - } - case MAXWELL3D_REG_INDEX(cb_bind[0]): { - ProcessCBBind(0); - break; - } - case MAXWELL3D_REG_INDEX(cb_bind[1]): { - ProcessCBBind(1); - break; - } - case MAXWELL3D_REG_INDEX(cb_bind[2]): { - ProcessCBBind(2); - break; - } - case MAXWELL3D_REG_INDEX(cb_bind[3]): { - ProcessCBBind(3); - break; - } - case MAXWELL3D_REG_INDEX(cb_bind[4]): { - ProcessCBBind(4); - break; - } - case MAXWELL3D_REG_INDEX(draw.vertex_end_gl): { - DrawArrays(); - break; - } - case MAXWELL3D_REG_INDEX(clear_buffers): { - ProcessClearBuffers(); - break; - } - case MAXWELL3D_REG_INDEX(query.query_get): { - ProcessQueryGet(); - break; - } - case MAXWELL3D_REG_INDEX(condition.mode): { - ProcessQueryCondition(); - break; - } - case MAXWELL3D_REG_INDEX(counter_reset): { - ProcessCounterReset(); - break; - } - case MAXWELL3D_REG_INDEX(sync_info): { - ProcessSyncPoint(); - break; - } - case MAXWELL3D_REG_INDEX(exec_upload): { - upload_state.ProcessExec(regs.exec_upload.linear != 0); - break; - } - case MAXWELL3D_REG_INDEX(data_upload): { - upload_state.ProcessData(arg, is_last_call); - if (is_last_call) { - OnMemoryWrite(); - } - break; - } - default: - break; - } + const u32 argument = ProcessShadowRam(method, method_argument); + ProcessDirtyRegisters(method, argument); + ProcessMethodCall(method, argument, method_argument, is_last_call); } void Maxwell3D::CallMultiMethod(u32 method, const u32* base_start, u32 amount, @@ -300,23 +280,7 @@ void Maxwell3D::CallMultiMethod(u32 method, const u32* base_start, u32 amount, // Methods after 0xE00 are special, they're actually triggers for some microcode that was // uploaded to the GPU during initialization. if (method >= MacroRegistersStart) { - // We're trying to execute a macro - if (executing_macro == 0) { - // A macro call must begin by writing the macro method's register, not its argument. - ASSERT_MSG((method % 2) == 0, - "Can't start macro execution by writing to the ARGS register"); - executing_macro = method; - } - - for (std::size_t i = 0; i < amount; i++) { - macro_params.push_back(base_start[i]); - } - - // Call the macro when there are no more parameters in the command buffer - if (amount == methods_pending) { - CallMacroMethod(executing_macro, macro_params); - macro_params.clear(); - } + ProcessMacro(method, base_start, amount, amount == methods_pending); return; } switch (method) { @@ -335,15 +299,14 @@ void Maxwell3D::CallMultiMethod(u32 method, const u32* base_start, u32 amount, case MAXWELL3D_REG_INDEX(const_buffer.cb_data[12]): case MAXWELL3D_REG_INDEX(const_buffer.cb_data[13]): case MAXWELL3D_REG_INDEX(const_buffer.cb_data[14]): - case MAXWELL3D_REG_INDEX(const_buffer.cb_data[15]): { + case MAXWELL3D_REG_INDEX(const_buffer.cb_data[15]): ProcessCBMultiData(method, base_start, amount); break; - } - default: { + default: for (std::size_t i = 0; i < amount; i++) { CallMethod(method, base_start[i], methods_pending - static_cast<u32>(i) <= 1); } - } + break; } } diff --git a/src/video_core/engines/maxwell_3d.h b/src/video_core/engines/maxwell_3d.h index bc289c55d..1cbe8fe67 100644 --- a/src/video_core/engines/maxwell_3d.h +++ b/src/video_core/engines/maxwell_3d.h @@ -1461,6 +1461,14 @@ public: private: void InitializeRegisterDefaults(); + void ProcessMacro(u32 method, const u32* base_start, u32 amount, bool is_last_call); + + u32 ProcessShadowRam(u32 method, u32 argument); + + void ProcessDirtyRegisters(u32 method, u32 argument); + + void ProcessMethodCall(u32 method, u32 argument, u32 nonshadow_argument, bool is_last_call); + Core::System& system; MemoryManager& memory_manager; diff --git a/src/video_core/engines/shader_bytecode.h b/src/video_core/engines/shader_bytecode.h index a3c05d1b0..37d17efdc 100644 --- a/src/video_core/engines/shader_bytecode.h +++ b/src/video_core/engines/shader_bytecode.h @@ -32,31 +32,31 @@ struct Register { constexpr Register() = default; - constexpr Register(u64 value) : value(value) {} + constexpr Register(u64 value_) : value(value_) {} - constexpr operator u64() const { + [[nodiscard]] constexpr operator u64() const { return value; } template <typename T> - constexpr u64 operator-(const T& oth) const { + [[nodiscard]] constexpr u64 operator-(const T& oth) const { return value - oth; } template <typename T> - constexpr u64 operator&(const T& oth) const { + [[nodiscard]] constexpr u64 operator&(const T& oth) const { return value & oth; } - constexpr u64 operator&(const Register& oth) const { + [[nodiscard]] constexpr u64 operator&(const Register& oth) const { return value & oth.value; } - constexpr u64 operator~() const { + [[nodiscard]] constexpr u64 operator~() const { return ~value; } - u64 GetSwizzledIndex(u64 elem) const { + [[nodiscard]] u64 GetSwizzledIndex(u64 elem) const { elem = (value + elem) & 3; return (value & ~3) + elem; } @@ -75,7 +75,7 @@ enum class AttributeSize : u64 { union Attribute { Attribute() = default; - constexpr explicit Attribute(u64 value) : value(value) {} + constexpr explicit Attribute(u64 value_) : value(value_) {} enum class Index : u64 { LayerViewportPointSize = 6, @@ -107,7 +107,7 @@ union Attribute { BitField<31, 1, u64> patch; BitField<47, 3, AttributeSize> size; - bool IsPhysical() const { + [[nodiscard]] bool IsPhysical() const { return patch == 0 && element == 0 && static_cast<u64>(index.Value()) == 0; } } fmt20; @@ -124,7 +124,7 @@ union Attribute { union Sampler { Sampler() = default; - constexpr explicit Sampler(u64 value) : value(value) {} + constexpr explicit Sampler(u64 value_) : value(value_) {} enum class Index : u64 { Sampler_0 = 8, @@ -137,7 +137,7 @@ union Sampler { union Image { Image() = default; - constexpr explicit Image(u64 value) : value{value} {} + constexpr explicit Image(u64 value_) : value{value_} {} BitField<36, 13, u64> index; u64 value; @@ -505,14 +505,14 @@ struct IpaMode { IpaInterpMode interpolation_mode; IpaSampleMode sampling_mode; - bool operator==(const IpaMode& a) const { + [[nodiscard]] bool operator==(const IpaMode& a) const { return std::tie(interpolation_mode, sampling_mode) == std::tie(a.interpolation_mode, a.sampling_mode); } - bool operator!=(const IpaMode& a) const { + [[nodiscard]] bool operator!=(const IpaMode& a) const { return !operator==(a); } - bool operator<(const IpaMode& a) const { + [[nodiscard]] bool operator<(const IpaMode& a) const { return std::tie(interpolation_mode, sampling_mode) < std::tie(a.interpolation_mode, a.sampling_mode); } @@ -658,10 +658,10 @@ union Instruction { return *this; } - constexpr Instruction(u64 value) : value{value} {} + constexpr Instruction(u64 value_) : value{value_} {} constexpr Instruction(const Instruction& instr) : value(instr.value) {} - constexpr bool Bit(u64 offset) const { + [[nodiscard]] constexpr bool Bit(u64 offset) const { return ((value >> offset) & 1) != 0; } @@ -746,34 +746,34 @@ union Instruction { BitField<28, 8, u64> imm_lut28; BitField<48, 8, u64> imm_lut48; - u32 GetImmLut28() const { + [[nodiscard]] u32 GetImmLut28() const { return static_cast<u32>(imm_lut28); } - u32 GetImmLut48() const { + [[nodiscard]] u32 GetImmLut48() const { return static_cast<u32>(imm_lut48); } } lop3; - u16 GetImm20_16() const { + [[nodiscard]] u16 GetImm20_16() const { return static_cast<u16>(imm20_16); } - u32 GetImm20_19() const { + [[nodiscard]] u32 GetImm20_19() const { u32 imm{static_cast<u32>(imm20_19)}; imm <<= 12; imm |= negate_imm ? 0x80000000 : 0; return imm; } - u32 GetImm20_32() const { + [[nodiscard]] u32 GetImm20_32() const { return static_cast<u32>(imm20_32); } - s32 GetSignedImm20_20() const { - u32 immediate = static_cast<u32>(imm20_19 | (negate_imm << 19)); + [[nodiscard]] s32 GetSignedImm20_20() const { + const auto immediate = static_cast<u32>(imm20_19 | (negate_imm << 19)); // Sign extend the 20-bit value. - u32 mask = 1U << (20 - 1); + const auto mask = 1U << (20 - 1); return static_cast<s32>((immediate ^ mask) - mask); } } alu; @@ -857,7 +857,7 @@ union Instruction { BitField<56, 1, u64> second_negate; BitField<30, 9, u64> second; - u32 PackImmediates() const { + [[nodiscard]] u32 PackImmediates() const { // Immediates are half floats shifted. constexpr u32 imm_shift = 6; return static_cast<u32>((first << imm_shift) | (second << (16 + imm_shift))); @@ -1033,7 +1033,7 @@ union Instruction { BitField<28, 2, AtomicType> type; BitField<30, 22, s64> offset; - s32 GetImmediateOffset() const { + [[nodiscard]] s32 GetImmediateOffset() const { return static_cast<s32>(offset << 2); } } atoms; @@ -1215,7 +1215,7 @@ union Instruction { BitField<39, 4, u64> rounding; // H0, H1 extract for F16 missing BitField<41, 1, u64> selector; // Guessed as some games set it, TODO: reverse this value - F2fRoundingOp GetRoundingMode() const { + [[nodiscard]] F2fRoundingOp GetRoundingMode() const { constexpr u64 rounding_mask = 0x0B; return static_cast<F2fRoundingOp>(rounding.Value() & rounding_mask); } @@ -1239,15 +1239,15 @@ union Instruction { BitField<54, 1, u64> aoffi_flag; BitField<55, 3, TextureProcessMode> process_mode; - bool IsComponentEnabled(std::size_t component) const { - return ((1ull << component) & component_mask) != 0; + [[nodiscard]] bool IsComponentEnabled(std::size_t component) const { + return ((1ULL << component) & component_mask) != 0; } - TextureProcessMode GetTextureProcessMode() const { + [[nodiscard]] TextureProcessMode GetTextureProcessMode() const { return process_mode; } - bool UsesMiscMode(TextureMiscMode mode) const { + [[nodiscard]] bool UsesMiscMode(TextureMiscMode mode) const { switch (mode) { case TextureMiscMode::DC: return dc_flag != 0; @@ -1271,15 +1271,15 @@ union Instruction { BitField<36, 1, u64> aoffi_flag; BitField<37, 3, TextureProcessMode> process_mode; - bool IsComponentEnabled(std::size_t component) const { + [[nodiscard]] bool IsComponentEnabled(std::size_t component) const { return ((1ULL << component) & component_mask) != 0; } - TextureProcessMode GetTextureProcessMode() const { + [[nodiscard]] TextureProcessMode GetTextureProcessMode() const { return process_mode; } - bool UsesMiscMode(TextureMiscMode mode) const { + [[nodiscard]] bool UsesMiscMode(TextureMiscMode mode) const { switch (mode) { case TextureMiscMode::DC: return dc_flag != 0; @@ -1299,7 +1299,7 @@ union Instruction { BitField<31, 4, u64> component_mask; BitField<49, 1, u64> nodep_flag; - bool UsesMiscMode(TextureMiscMode mode) const { + [[nodiscard]] bool UsesMiscMode(TextureMiscMode mode) const { switch (mode) { case TextureMiscMode::NODEP: return nodep_flag != 0; @@ -1309,7 +1309,7 @@ union Instruction { return false; } - bool IsComponentEnabled(std::size_t component) const { + [[nodiscard]] bool IsComponentEnabled(std::size_t component) const { return ((1ULL << component) & component_mask) != 0; } } txq; @@ -1321,11 +1321,11 @@ union Instruction { BitField<35, 1, u64> ndv_flag; BitField<49, 1, u64> nodep_flag; - bool IsComponentEnabled(std::size_t component) const { - return ((1ull << component) & component_mask) != 0; + [[nodiscard]] bool IsComponentEnabled(std::size_t component) const { + return ((1ULL << component) & component_mask) != 0; } - bool UsesMiscMode(TextureMiscMode mode) const { + [[nodiscard]] bool UsesMiscMode(TextureMiscMode mode) const { switch (mode) { case TextureMiscMode::NDV: return (ndv_flag != 0); @@ -1347,7 +1347,7 @@ union Instruction { BitField<54, 2, u64> offset_mode; BitField<56, 2, u64> component; - bool UsesMiscMode(TextureMiscMode mode) const { + [[nodiscard]] bool UsesMiscMode(TextureMiscMode mode) const { switch (mode) { case TextureMiscMode::NDV: return ndv_flag != 0; @@ -1373,7 +1373,7 @@ union Instruction { BitField<33, 2, u64> offset_mode; BitField<37, 2, u64> component; - bool UsesMiscMode(TextureMiscMode mode) const { + [[nodiscard]] bool UsesMiscMode(TextureMiscMode mode) const { switch (mode) { case TextureMiscMode::NDV: return ndv_flag != 0; @@ -1399,7 +1399,7 @@ union Instruction { BitField<52, 2, u64> component; BitField<55, 1, u64> fp16_flag; - bool UsesMiscMode(TextureMiscMode mode) const { + [[nodiscard]] bool UsesMiscMode(TextureMiscMode mode) const { switch (mode) { case TextureMiscMode::DC: return dc_flag != 0; @@ -1422,16 +1422,20 @@ union Instruction { BitField<53, 4, u64> texture_info; BitField<59, 1, u64> fp32_flag; - TextureType GetTextureType() const { + [[nodiscard]] TextureType GetTextureType() const { // The TEXS instruction has a weird encoding for the texture type. - if (texture_info == 0) + if (texture_info == 0) { return TextureType::Texture1D; - if (texture_info >= 1 && texture_info <= 9) + } + if (texture_info >= 1 && texture_info <= 9) { return TextureType::Texture2D; - if (texture_info >= 10 && texture_info <= 11) + } + if (texture_info >= 10 && texture_info <= 11) { return TextureType::Texture3D; - if (texture_info >= 12 && texture_info <= 13) + } + if (texture_info >= 12 && texture_info <= 13) { return TextureType::TextureCube; + } LOG_CRITICAL(HW_GPU, "Unhandled texture_info: {}", static_cast<u32>(texture_info.Value())); @@ -1439,7 +1443,7 @@ union Instruction { return TextureType::Texture1D; } - TextureProcessMode GetTextureProcessMode() const { + [[nodiscard]] TextureProcessMode GetTextureProcessMode() const { switch (texture_info) { case 0: case 2: @@ -1458,7 +1462,7 @@ union Instruction { return TextureProcessMode::None; } - bool UsesMiscMode(TextureMiscMode mode) const { + [[nodiscard]] bool UsesMiscMode(TextureMiscMode mode) const { switch (mode) { case TextureMiscMode::DC: return (texture_info >= 4 && texture_info <= 6) || texture_info == 9; @@ -1470,16 +1474,16 @@ union Instruction { return false; } - bool IsArrayTexture() const { + [[nodiscard]] bool IsArrayTexture() const { // TEXS only supports Texture2D arrays. return texture_info >= 7 && texture_info <= 9; } - bool HasTwoDestinations() const { + [[nodiscard]] bool HasTwoDestinations() const { return gpr28.Value() != Register::ZeroIndex; } - bool IsComponentEnabled(std::size_t component) const { + [[nodiscard]] bool IsComponentEnabled(std::size_t component) const { static constexpr std::array<std::array<u32, 8>, 4> mask_lut{{ {}, {0x1, 0x2, 0x4, 0x8, 0x3, 0x9, 0xa, 0xc}, @@ -1506,7 +1510,7 @@ union Instruction { BitField<54, 1, u64> cl; BitField<55, 1, u64> process_mode; - TextureProcessMode GetTextureProcessMode() const { + [[nodiscard]] TextureProcessMode GetTextureProcessMode() const { return process_mode == 0 ? TextureProcessMode::LZ : TextureProcessMode::LL; } } tld; @@ -1516,7 +1520,7 @@ union Instruction { BitField<53, 4, u64> texture_info; BitField<59, 1, u64> fp32_flag; - TextureType GetTextureType() const { + [[nodiscard]] TextureType GetTextureType() const { // The TLDS instruction has a weird encoding for the texture type. if (texture_info <= 1) { return TextureType::Texture1D; @@ -1535,13 +1539,14 @@ union Instruction { return TextureType::Texture1D; } - TextureProcessMode GetTextureProcessMode() const { - if (texture_info == 1 || texture_info == 5 || texture_info == 12) + [[nodiscard]] TextureProcessMode GetTextureProcessMode() const { + if (texture_info == 1 || texture_info == 5 || texture_info == 12) { return TextureProcessMode::LL; + } return TextureProcessMode::LZ; } - bool UsesMiscMode(TextureMiscMode mode) const { + [[nodiscard]] bool UsesMiscMode(TextureMiscMode mode) const { switch (mode) { case TextureMiscMode::AOFFI: return texture_info == 12 || texture_info == 4; @@ -1555,7 +1560,7 @@ union Instruction { return false; } - bool IsArrayTexture() const { + [[nodiscard]] bool IsArrayTexture() const { // TEXS only supports Texture2D arrays. return texture_info == 8; } @@ -1567,7 +1572,7 @@ union Instruction { BitField<35, 1, u64> aoffi_flag; BitField<49, 1, u64> nodep_flag; - bool UsesMiscMode(TextureMiscMode mode) const { + [[nodiscard]] bool UsesMiscMode(TextureMiscMode mode) const { switch (mode) { case TextureMiscMode::AOFFI: return aoffi_flag != 0; @@ -1591,7 +1596,7 @@ union Instruction { BitField<20, 3, StoreType> store_data_layout; BitField<20, 4, u64> component_mask_selector; - bool IsComponentEnabled(std::size_t component) const { + [[nodiscard]] bool IsComponentEnabled(std::size_t component) const { ASSERT(mode == SurfaceDataMode::P); constexpr u8 R = 0b0001; constexpr u8 G = 0b0010; @@ -1604,7 +1609,7 @@ union Instruction { return std::bitset<4>{mask.at(component_mask_selector)}.test(component); } - StoreType GetStoreDataLayout() const { + [[nodiscard]] StoreType GetStoreDataLayout() const { ASSERT(mode == SurfaceDataMode::D_BA); return store_data_layout; } @@ -1622,14 +1627,15 @@ union Instruction { BitField<20, 24, u64> target; BitField<5, 1, u64> constant_buffer; - s32 GetBranchTarget() const { + [[nodiscard]] s32 GetBranchTarget() const { // Sign extend the branch target offset - u32 mask = 1U << (24 - 1); - u32 value = static_cast<u32>(target); + const auto mask = 1U << (24 - 1); + const auto target_value = static_cast<u32>(target); + constexpr auto instruction_size = static_cast<s32>(sizeof(Instruction)); + // The branch offset is relative to the next instruction and is stored in bytes, so // divide it by the size of an instruction and add 1 to it. - return static_cast<s32>((value ^ mask) - mask) / static_cast<s32>(sizeof(Instruction)) + - 1; + return static_cast<s32>((target_value ^ mask) - mask) / instruction_size + 1; } } bra; @@ -1637,14 +1643,15 @@ union Instruction { BitField<20, 24, u64> target; BitField<5, 1, u64> constant_buffer; - s32 GetBranchExtend() const { + [[nodiscard]] s32 GetBranchExtend() const { // Sign extend the branch target offset - u32 mask = 1U << (24 - 1); - u32 value = static_cast<u32>(target); + const auto mask = 1U << (24 - 1); + const auto target_value = static_cast<u32>(target); + constexpr auto instruction_size = static_cast<s32>(sizeof(Instruction)); + // The branch offset is relative to the next instruction and is stored in bytes, so // divide it by the size of an instruction and add 1 to it. - return static_cast<s32>((value ^ mask) - mask) / static_cast<s32>(sizeof(Instruction)) + - 1; + return static_cast<s32>((target_value ^ mask) - mask) / instruction_size + 1; } } brx; @@ -1697,7 +1704,7 @@ union Instruction { BitField<50, 1, u64> is_op_b_register; BitField<51, 3, VmnmxOperation> operation; - VmnmxType SourceFormatA() const { + [[nodiscard]] VmnmxType SourceFormatA() const { switch (src_format_a) { case 0b11: return VmnmxType::Bits32; @@ -1708,7 +1715,7 @@ union Instruction { } } - VmnmxType SourceFormatB() const { + [[nodiscard]] VmnmxType SourceFormatB() const { switch (src_format_b) { case 0b11: return VmnmxType::Bits32; @@ -1739,7 +1746,7 @@ union Instruction { BitField<20, 14, u64> shifted_offset; BitField<34, 5, u64> index; - u64 GetOffset() const { + [[nodiscard]] u64 GetOffset() const { return shifted_offset * 4; } } cbuf34; @@ -1748,7 +1755,7 @@ union Instruction { BitField<20, 16, s64> offset; BitField<36, 5, u64> index; - s64 GetOffset() const { + [[nodiscard]] s64 GetOffset() const { return offset; } } cbuf36; @@ -1997,29 +2004,29 @@ public: /// Returns whether an opcode has an execution predicate field or not (ie, whether it can be /// conditionally executed). - static bool IsPredicatedInstruction(Id opcode) { + [[nodiscard]] static bool IsPredicatedInstruction(Id opcode) { // TODO(Subv): Add the rest of unpredicated instructions. return opcode != Id::SSY && opcode != Id::PBK; } class Matcher { public: - constexpr Matcher(const char* const name, u16 mask, u16 expected, Id id, Type type) - : name{name}, mask{mask}, expected{expected}, id{id}, type{type} {} + constexpr Matcher(const char* const name_, u16 mask_, u16 expected_, Id id_, Type type_) + : name{name_}, mask{mask_}, expected{expected_}, id{id_}, type{type_} {} - constexpr const char* GetName() const { + [[nodiscard]] constexpr const char* GetName() const { return name; } - constexpr u16 GetMask() const { + [[nodiscard]] constexpr u16 GetMask() const { return mask; } - constexpr Id GetId() const { + [[nodiscard]] constexpr Id GetId() const { return id; } - constexpr Type GetType() const { + [[nodiscard]] constexpr Type GetType() const { return type; } @@ -2028,7 +2035,7 @@ public: * @param instruction The instruction to test * @returns true if the given instruction matches. */ - constexpr bool Matches(u16 instruction) const { + [[nodiscard]] constexpr bool Matches(u16 instruction) const { return (instruction & mask) == expected; } @@ -2040,7 +2047,8 @@ public: Type type; }; - static std::optional<std::reference_wrapper<const Matcher>> Decode(Instruction instr) { + using DecodeResult = std::optional<std::reference_wrapper<const Matcher>>; + [[nodiscard]] static DecodeResult Decode(Instruction instr) { static const auto table{GetDecodeTable()}; const auto matches_instruction = [instr](const auto& matcher) { @@ -2062,7 +2070,7 @@ private: * A '0' in a bitstring indicates that a zero must be present at that bit position. * A '1' in a bitstring indicates that a one must be present at that bit position. */ - static constexpr auto GetMaskAndExpect(const char* const bitstring) { + [[nodiscard]] static constexpr auto GetMaskAndExpect(const char* const bitstring) { u16 mask = 0, expect = 0; for (std::size_t i = 0; i < opcode_bitsize; i++) { const std::size_t bit_position = opcode_bitsize - i - 1; @@ -2084,14 +2092,14 @@ private: public: /// Creates a matcher that can match and parse instructions based on bitstring. - static constexpr auto GetMatcher(const char* const bitstring, Id op, Type type, - const char* const name) { + [[nodiscard]] static constexpr auto GetMatcher(const char* const bitstring, Id op, + Type type, const char* const name) { const auto [mask, expected] = GetMaskAndExpect(bitstring); return Matcher(name, mask, expected, op, type); } }; - static std::vector<Matcher> GetDecodeTable() { + [[nodiscard]] static std::vector<Matcher> GetDecodeTable() { std::vector<Matcher> table = { #define INST(bitstring, op, type, name) Detail::GetMatcher(bitstring, op, type, name) INST("111000110011----", Id::KIL, Type::Flow, "KIL"), diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp index 36bf92808..cfddbde5d 100644 --- a/src/video_core/renderer_opengl/gl_rasterizer.cpp +++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp @@ -1156,7 +1156,7 @@ void RasterizerOpenGL::SyncViewport() { flags[Dirty::ClipControl] = false; bool flip_y = false; - if (regs.viewport_transform[0].scale_y < 0.0) { + if (regs.viewport_transform[0].scale_y < 0.0f) { flip_y = !flip_y; } if (regs.screen_y_control.y_negate != 0) { @@ -1579,10 +1579,6 @@ void RasterizerOpenGL::SyncAlphaTest() { flags[Dirty::AlphaTest] = false; const auto& regs = maxwell3d.regs; - if (regs.alpha_test_enabled && regs.rt_control.count > 1) { - LOG_WARNING(Render_OpenGL, "Alpha testing with more than one render target is not tested"); - } - if (regs.alpha_test_enabled) { glEnable(GL_ALPHA_TEST); glAlphaFunc(MaxwellToGL::ComparisonOp(regs.alpha_test_func), regs.alpha_test_ref); diff --git a/src/video_core/shader/async_shaders.cpp b/src/video_core/shader/async_shaders.cpp index 39cc3b869..6920afdf2 100644 --- a/src/video_core/shader/async_shaders.cpp +++ b/src/video_core/shader/async_shaders.cpp @@ -43,8 +43,8 @@ void AsyncShaders::AllocateWorkers() { // Create workers for (std::size_t i = 0; i < num_workers; i++) { context_list.push_back(emu_window.CreateSharedContext()); - worker_threads.push_back( - std::thread(&AsyncShaders::ShaderCompilerThread, this, context_list[i].get())); + worker_threads.emplace_back(&AsyncShaders::ShaderCompilerThread, this, + context_list[i].get()); } } @@ -106,8 +106,7 @@ std::vector<AsyncShaders::Result> AsyncShaders::GetCompletedWork() { std::vector<Result> results; { std::unique_lock lock{completed_mutex}; - results.assign(std::make_move_iterator(finished_work.begin()), - std::make_move_iterator(finished_work.end())); + results = std::move(finished_work); finished_work.clear(); } return results; @@ -116,11 +115,10 @@ std::vector<AsyncShaders::Result> AsyncShaders::GetCompletedWork() { void AsyncShaders::QueueOpenGLShader(const OpenGL::Device& device, Tegra::Engines::ShaderType shader_type, u64 uid, std::vector<u64> code, std::vector<u64> code_b, - u32 main_offset, - VideoCommon::Shader::CompilerSettings compiler_settings, - const VideoCommon::Shader::Registry& registry, - VAddr cpu_addr) { - WorkerParams params{ + u32 main_offset, CompilerSettings compiler_settings, + const Registry& registry, VAddr cpu_addr) { + std::unique_lock lock(queue_mutex); + pending_queue.push({ .backend = device.UseAssemblyShaders() ? Backend::GLASM : Backend::OpenGL, .device = &device, .shader_type = shader_type, @@ -131,9 +129,7 @@ void AsyncShaders::QueueOpenGLShader(const OpenGL::Device& device, .compiler_settings = compiler_settings, .registry = registry, .cpu_address = cpu_addr, - }; - std::unique_lock lock(queue_mutex); - pending_queue.push(std::move(params)); + }); cv.notify_one(); } @@ -145,7 +141,8 @@ void AsyncShaders::QueueVulkanShader(Vulkan::VKPipelineCache* pp_cache, std::vector<VkDescriptorSetLayoutBinding> bindings, Vulkan::SPIRVProgram program, Vulkan::GraphicsPipelineCacheKey key) { - WorkerParams params{ + std::unique_lock lock(queue_mutex); + pending_queue.push({ .backend = Backend::Vulkan, .pp_cache = pp_cache, .vk_device = &device, @@ -153,13 +150,10 @@ void AsyncShaders::QueueVulkanShader(Vulkan::VKPipelineCache* pp_cache, .descriptor_pool = &descriptor_pool, .update_descriptor_queue = &update_descriptor_queue, .renderpass_cache = &renderpass_cache, - .bindings = bindings, - .program = program, + .bindings = std::move(bindings), + .program = std::move(program), .key = key, - }; - - std::unique_lock lock(queue_mutex); - pending_queue.push(std::move(params)); + }); cv.notify_one(); } diff --git a/src/yuzu/configuration/configure_input_player.cpp b/src/yuzu/configuration/configure_input_player.cpp index 72640f5e7..56ab32a35 100644 --- a/src/yuzu/configuration/configure_input_player.cpp +++ b/src/yuzu/configuration/configure_input_player.cpp @@ -1137,7 +1137,7 @@ void ConfigureInputPlayer::CreateProfile() { return; } - if (!profiles->IsProfileNameValid(profile_name.toStdString())) { + if (!InputProfiles::IsProfileNameValid(profile_name.toStdString())) { QMessageBox::critical(this, tr("Create Input Profile"), tr("The given profile name is not valid!")); return; |