This commit is contained in:
Jarek Syrylak
2018-06-06 21:12:27 +01:00
14 changed files with 186 additions and 30 deletions

View File

@@ -144,6 +144,7 @@ static u32 latest_signal = 0;
static bool memory_break = false;
static Kernel::Thread* current_thread = nullptr;
static int thread_id = -1;
// Binding to a port within the reserved ports range (0-1023) requires root permissions,
// so default to a port outside of that range.
@@ -562,11 +563,9 @@ static void HandleQuery() {
/// Handle set thread command from gdb client.
static void HandleSetThread() {
if (memcmp(command_buffer, "Hc", 2) == 0 || memcmp(command_buffer, "Hg", 2) == 0) {
int thread_id = -1;
thread_id = -1;
if (command_buffer[2] != '-') {
thread_id = static_cast<int>(HexToInt(
command_buffer + 2,
command_length - 2 /*strlen(reinterpret_cast<char*>(command_buffer) + 2)*/));
thread_id = static_cast<int>(HexToInt(command_buffer + 2, command_length - 2));
}
if (thread_id >= 1) {
current_thread = FindThreadById(thread_id);
@@ -585,9 +584,7 @@ static void HandleSetThread() {
/// Handle thread alive command from gdb client.
static void HandleThreadAlive() {
int thread_id = static_cast<int>(
HexToInt(command_buffer + 1,
command_length - 1 /*strlen(reinterpret_cast<char*>(command_buffer) + 1)*/));
int thread_id = static_cast<int>(HexToInt(command_buffer + 1, command_length - 1));
if (thread_id == 0) {
thread_id = 1;
}
@@ -1192,7 +1189,9 @@ void SetCpuStepFlag(bool is_step) {
}
void SendTrap(Kernel::Thread* thread, int trap) {
if (send_trap) {
//NGLOG_ERROR(Debug_GDBStub, "SendTrap {} {} {} {}", thread->GetThreadId(), thread_id, trap,
// send_trap);
if (send_trap && (thread == current_thread)) {
send_trap = false;
SendSignal(thread, trap);
}

View File

@@ -561,7 +561,7 @@ IApplicationFunctions::IApplicationFunctions() : ServiceFramework("IApplicationF
{32, nullptr, "BeginBlockingHomeButton"},
{33, nullptr, "EndBlockingHomeButton"},
{40, &IApplicationFunctions::NotifyRunning, "NotifyRunning"},
{50, nullptr, "GetPseudoDeviceId"},
{50, &IApplicationFunctions::GetPseudoDeviceId, "GetPseudoDeviceId"},
{60, nullptr, "SetMediaPlaybackStateForApplication"},
{65, nullptr, "IsGamePlayRecordingSupported"},
{66, &IApplicationFunctions::InitializeGamePlayRecording, "InitializeGamePlayRecording"},
@@ -684,6 +684,17 @@ void IApplicationFunctions::NotifyRunning(Kernel::HLERequestContext& ctx) {
NGLOG_WARNING(Service_AM, "(STUBBED) called");
}
void IApplicationFunctions::GetPseudoDeviceId(Kernel::HLERequestContext& ctx) {
IPC::ResponseBuilder rb{ctx, 6};
rb.Push(RESULT_SUCCESS);
// Returns a 128-bit UUID
rb.Push<u64>(0);
rb.Push<u64>(0);
NGLOG_WARNING(Service_AM, "(STUBBED) called");
}
void InstallInterfaces(SM::ServiceManager& service_manager,
std::shared_ptr<NVFlinger::NVFlinger> nvflinger) {
std::make_shared<AppletAE>(nvflinger)->InstallAsService(service_manager);

View File

@@ -138,6 +138,7 @@ private:
void InitializeGamePlayRecording(Kernel::HLERequestContext& ctx);
void SetGamePlayRecordingState(Kernel::HLERequestContext& ctx);
void NotifyRunning(Kernel::HLERequestContext& ctx);
void GetPseudoDeviceId(Kernel::HLERequestContext& ctx);
};
class IHomeMenuFunctions final : public ServiceFramework<IHomeMenuFunctions> {

View File

@@ -38,7 +38,7 @@ public:
{8, nullptr, "SetPriority"},
{9, nullptr, "SetNetworkProfileId"},
{10, nullptr, "SetRejectable"},
{11, nullptr, "SetConnectionConfirmationOption"},
{11, &IRequest::SetConnectionConfirmationOption, "SetConnectionConfirmationOption"},
{12, nullptr, "SetPersistent"},
{13, nullptr, "SetInstant"},
{14, nullptr, "SetSustainable"},
@@ -67,23 +67,32 @@ private:
rb.Push(RESULT_SUCCESS);
rb.Push<u32>(0);
}
void GetResult(Kernel::HLERequestContext& ctx) {
NGLOG_WARNING(Service_NIFM, "(STUBBED) called");
IPC::ResponseBuilder rb{ctx, 2};
rb.Push(RESULT_SUCCESS);
}
void GetSystemEventReadableHandles(Kernel::HLERequestContext& ctx) {
NGLOG_WARNING(Service_NIFM, "(STUBBED) called");
IPC::ResponseBuilder rb{ctx, 2, 2};
rb.Push(RESULT_SUCCESS);
rb.PushCopyObjects(event1, event2);
}
void Cancel(Kernel::HLERequestContext& ctx) {
NGLOG_WARNING(Service_NIFM, "(STUBBED) called");
IPC::ResponseBuilder rb{ctx, 2};
rb.Push(RESULT_SUCCESS);
}
void SetConnectionConfirmationOption(Kernel::HLERequestContext& ctx) {
NGLOG_WARNING(Service_NIFM, "(STUBBED) called");
IPC::ResponseBuilder rb{ctx, 2};
rb.Push(RESULT_SUCCESS);
}
Kernel::SharedPtr<Kernel::Event> event1, event2;
};

View File

@@ -26,6 +26,10 @@ u32 nvhost_ctrl_gpu::ioctl(Ioctl command, const std::vector<u8>& input, std::vec
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);
}
UNIMPLEMENTED_MSG("Unimplemented ioctl");
return 0;
@@ -136,4 +140,22 @@ u32 nvhost_ctrl_gpu::ZBCSetTable(const std::vector<u8>& input, std::vector<u8>&
return 0;
}
u32 nvhost_ctrl_gpu::ZBCQueryTable(const std::vector<u8>& input, std::vector<u8>& output) {
NGLOG_WARNING(Service_NVDRV, "(STUBBED) called");
IoctlZbcQueryTable params{};
std::memcpy(&params, input.data(), input.size());
// TODO : To implement properly
std::memcpy(output.data(), &params, output.size());
return 0;
}
u32 nvhost_ctrl_gpu::FlushL2(const std::vector<u8>& input, std::vector<u8>& output) {
NGLOG_WARNING(Service_NVDRV, "(STUBBED) called");
IoctlFlushL2 params{};
std::memcpy(&params, input.data(), input.size());
// TODO : To implement properly
std::memcpy(output.data(), &params, output.size());
return 0;
}
} // namespace Service::Nvidia::Devices

View File

@@ -26,6 +26,18 @@ private:
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 {
@@ -127,12 +139,31 @@ private:
};
static_assert(sizeof(IoctlZbcSetTable) == 44, "IoctlZbcSetTable is incorrect size");
struct IoctlZbcQueryTable {
u32_le color_ds[4];
u32_le color_l2[4];
u32_le depth;
u32_le ref_cnt;
u32_le format;
u32_le type;
u32_le index_size;
};
static_assert(sizeof(IoctlZbcQueryTable) == 52, "IoctlZbcQueryTable is incorrect size");
struct IoctlFlushL2 {
u32_le flush; // l2_flush | l2_invalidate << 1 | fb_flush << 2
u32_le reserved;
};
static_assert(sizeof(IoctlFlushL2) == 8, "IoctlFlushL2 is incorrect size");
u32 GetCharacteristics(const std::vector<u8>& input, std::vector<u8>& output);
u32 GetTPCMasks(const std::vector<u8>& input, std::vector<u8>& output);
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);
};
} // namespace Service::Nvidia::Devices

View File

@@ -173,6 +173,13 @@ enum class SubOp : u64 {
Min = 0x8,
};
enum class FloatRoundingOp : u64 {
None = 0,
Floor = 1,
Ceil = 2,
Trunc = 3,
};
union Instruction {
Instruction& operator=(const Instruction& instr) {
value = instr.value;
@@ -290,11 +297,20 @@ union Instruction {
union {
BitField<10, 2, Register::Size> size;
BitField<13, 1, u64> is_signed;
BitField<12, 1, u64> is_output_signed;
BitField<13, 1, u64> is_input_signed;
BitField<41, 2, u64> selector;
BitField<45, 1, u64> negate_a;
BitField<49, 1, u64> abs_a;
BitField<50, 1, u64> saturate_a;
union {
BitField<39, 2, FloatRoundingOp> rounding;
} f2i;
union {
BitField<39, 4, u64> rounding;
} f2f;
} conversion;
union {
@@ -560,9 +576,9 @@ private:
INST("0100110010101---", Id::F2F_C, Type::Conversion, "F2F_C"),
INST("0101110010101---", Id::F2F_R, Type::Conversion, "F2F_R"),
INST("0011100-10101---", Id::F2F_IMM, Type::Conversion, "F2F_IMM"),
INST("0100110010110---", Id::F2I_C, Type::Arithmetic, "F2I_C"),
INST("0101110010110---", Id::F2I_R, Type::Arithmetic, "F2I_R"),
INST("0011100-10110---", Id::F2I_IMM, Type::Arithmetic, "F2I_IMM"),
INST("0100110010110---", Id::F2I_C, Type::Conversion, "F2I_C"),
INST("0101110010110---", Id::F2I_R, Type::Conversion, "F2I_R"),
INST("0011100-10110---", Id::F2I_IMM, Type::Conversion, "F2I_IMM"),
INST("0100110010011---", Id::MOV_C, Type::Arithmetic, "MOV_C"),
INST("0101110010011---", Id::MOV_R, Type::Arithmetic, "MOV_R"),
INST("0011100-10011---", Id::MOV_IMM, Type::Arithmetic, "MOV_IMM"),

View File

@@ -26,6 +26,10 @@ u32 RenderTargetBytesPerPixel(RenderTargetFormat format) {
ASSERT(format != RenderTargetFormat::NONE);
switch (format) {
case RenderTargetFormat::RGBA32_FLOAT:
return 16;
case RenderTargetFormat::RGBA16_FLOAT:
return 8;
case RenderTargetFormat::RGBA8_UNORM:
case RenderTargetFormat::RGB10_A2_UNORM:
return 4;

View File

@@ -15,10 +15,12 @@ namespace Tegra {
enum class RenderTargetFormat : u32 {
NONE = 0x0,
RGBA32_FLOAT = 0xC0,
RGBA16_FLOAT = 0xCA,
RGB10_A2_UNORM = 0xD1,
RGBA8_UNORM = 0xD5,
RGBA8_SRGB = 0xD6,
R11G11B10_FLOAT = 0xE0,
};
/// Returns the number of bytes per pixel of each rendertarget format.

View File

@@ -50,6 +50,7 @@ static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_form
{GL_RGB5_A1, GL_RGBA, GL_UNSIGNED_SHORT_1_5_5_5_REV, false}, // A1B5G5R5
{GL_R8, GL_RED, GL_UNSIGNED_BYTE, false}, // R8
{GL_RGBA16F, GL_RGBA, GL_HALF_FLOAT, false}, // RGBA16F
{GL_R11F_G11F_B10F, GL_RGB, GL_UNSIGNED_INT_10F_11F_11F_REV, false}, // R11FG11FB10F
{GL_COMPRESSED_RGB_S3TC_DXT1_EXT, GL_RGB, GL_UNSIGNED_INT_8_8_8_8, true}, // DXT1
{GL_COMPRESSED_RGBA_S3TC_DXT3_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, true}, // DXT23
{GL_COMPRESSED_RGBA_S3TC_DXT5_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, true}, // DXT45
@@ -60,8 +61,10 @@ static const FormatTuple& GetFormatTuple(PixelFormat pixel_format, ComponentType
const SurfaceType type = SurfaceParams::GetFormatType(pixel_format);
if (type == SurfaceType::ColorTexture) {
ASSERT(static_cast<size_t>(pixel_format) < tex_format_tuples.size());
// For now only UNORM components are supported, or RGBA16F which is type FLOAT
ASSERT(component_type == ComponentType::UNorm || pixel_format == PixelFormat::RGBA16F);
// For now only UNORM components are supported, or either R11FG11FB10F or RGBA16F which are
// type FLOAT
ASSERT(component_type == ComponentType::UNorm || pixel_format == PixelFormat::RGBA16F ||
pixel_format == PixelFormat::R11FG11FB10F);
return tex_format_tuples[static_cast<unsigned int>(pixel_format)];
} else if (type == SurfaceType::Depth || type == SurfaceType::DepthStencil) {
// TODO(Subv): Implement depth formats
@@ -110,11 +113,12 @@ static constexpr std::array<void (*)(u32, u32, u32, u8*, Tegra::GPUVAddr, Tegra:
Tegra::GPUVAddr),
SurfaceParams::MaxPixelFormat>
morton_to_gl_fns = {
MortonCopy<true, PixelFormat::ABGR8>, MortonCopy<true, PixelFormat::B5G6R5>,
MortonCopy<true, PixelFormat::A2B10G10R10>, MortonCopy<true, PixelFormat::A1B5G5R5>,
MortonCopy<true, PixelFormat::R8>, MortonCopy<true, PixelFormat::RGBA16F>,
MortonCopy<true, PixelFormat::DXT1>, MortonCopy<true, PixelFormat::DXT23>,
MortonCopy<true, PixelFormat::DXT45>, MortonCopy<true, PixelFormat::DXN1>,
MortonCopy<true, PixelFormat::ABGR8>, MortonCopy<true, PixelFormat::B5G6R5>,
MortonCopy<true, PixelFormat::A2B10G10R10>, MortonCopy<true, PixelFormat::A1B5G5R5>,
MortonCopy<true, PixelFormat::R8>, MortonCopy<true, PixelFormat::RGBA16F>,
MortonCopy<true, PixelFormat::R11FG11FB10F>, MortonCopy<true, PixelFormat::DXT1>,
MortonCopy<true, PixelFormat::DXT23>, MortonCopy<true, PixelFormat::DXT45>,
MortonCopy<true, PixelFormat::DXN1>,
};
static constexpr std::array<void (*)(u32, u32, u32, u8*, Tegra::GPUVAddr, Tegra::GPUVAddr,
@@ -127,6 +131,7 @@ static constexpr std::array<void (*)(u32, u32, u32, u8*, Tegra::GPUVAddr, Tegra:
MortonCopy<false, PixelFormat::A1B5G5R5>,
MortonCopy<false, PixelFormat::R8>,
MortonCopy<false, PixelFormat::RGBA16F>,
MortonCopy<false, PixelFormat::R11FG11FB10F>,
// TODO(Subv): Swizzling the DXT1/DXT23/DXT45/DXN1 formats is not yet supported
nullptr,
nullptr,

View File

@@ -60,10 +60,11 @@ struct SurfaceParams {
A1B5G5R5 = 3,
R8 = 4,
RGBA16F = 5,
DXT1 = 6,
DXT23 = 7,
DXT45 = 8,
DXN1 = 9, // This is also known as BC4
R11FG11FB10F = 6,
DXT1 = 7,
DXT23 = 8,
DXT45 = 9,
DXN1 = 10, // This is also known as BC4
Max,
Invalid = 255,
@@ -104,7 +105,8 @@ struct SurfaceParams {
1, // A2B10G10R10
1, // A1B5G5R5
1, // R8
2, // RGBA16F
1, // RGBA16F
1, // R11FG11FB10F
4, // DXT1
4, // DXT23
4, // DXT45
@@ -129,6 +131,7 @@ struct SurfaceParams {
16, // A1B5G5R5
8, // R8
64, // RGBA16F
32, // R11FG11FB10F
64, // DXT1
128, // DXT23
128, // DXT45
@@ -151,6 +154,8 @@ struct SurfaceParams {
return PixelFormat::A2B10G10R10;
case Tegra::RenderTargetFormat::RGBA16_FLOAT:
return PixelFormat::RGBA16F;
case Tegra::RenderTargetFormat::R11G11B10_FLOAT:
return PixelFormat::R11FG11FB10F;
default:
NGLOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format));
UNREACHABLE();
@@ -182,6 +187,8 @@ struct SurfaceParams {
return PixelFormat::R8;
case Tegra::Texture::TextureFormat::R16_G16_B16_A16:
return PixelFormat::RGBA16F;
case Tegra::Texture::TextureFormat::BF10GF11RF11:
return PixelFormat::R11FG11FB10F;
case Tegra::Texture::TextureFormat::DXT1:
return PixelFormat::DXT1;
case Tegra::Texture::TextureFormat::DXT23:
@@ -211,6 +218,8 @@ struct SurfaceParams {
return Tegra::Texture::TextureFormat::R8;
case PixelFormat::RGBA16F:
return Tegra::Texture::TextureFormat::R16_G16_B16_A16;
case PixelFormat::R11FG11FB10F:
return Tegra::Texture::TextureFormat::BF10GF11RF11;
case PixelFormat::DXT1:
return Tegra::Texture::TextureFormat::DXT1;
case PixelFormat::DXT23:
@@ -243,6 +252,7 @@ struct SurfaceParams {
case Tegra::RenderTargetFormat::RGB10_A2_UNORM:
return ComponentType::UNorm;
case Tegra::RenderTargetFormat::RGBA16_FLOAT:
case Tegra::RenderTargetFormat::R11G11B10_FLOAT:
return ComponentType::Float;
default:
NGLOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format));

View File

@@ -895,7 +895,7 @@ private:
}
case OpCode::Type::Shift: {
std::string op_a = regs.GetRegisterAsInteger(instr.gpr8, 0, false);
std::string op_a = regs.GetRegisterAsInteger(instr.gpr8, 0, true);
std::string op_b;
if (instr.is_b_imm) {
@@ -991,18 +991,20 @@ private:
ASSERT_MSG(!instr.conversion.selector, "Unimplemented");
std::string op_a =
regs.GetRegisterAsInteger(instr.gpr20, 0, instr.conversion.is_signed);
regs.GetRegisterAsInteger(instr.gpr20, 0, instr.conversion.is_input_signed);
if (instr.conversion.abs_a) {
op_a = "abs(" + op_a + ')';
}
regs.SetRegisterToInteger(instr.gpr0, instr.conversion.is_signed, 0, op_a, 1, 1);
regs.SetRegisterToInteger(instr.gpr0, instr.conversion.is_output_signed, 0, op_a, 1,
1);
break;
}
case OpCode::Id::I2F_R: {
ASSERT_MSG(!instr.conversion.selector, "Unimplemented");
std::string op_a =
regs.GetRegisterAsInteger(instr.gpr20, 0, instr.conversion.is_signed);
regs.GetRegisterAsInteger(instr.gpr20, 0, instr.conversion.is_input_signed);
if (instr.conversion.abs_a) {
op_a = "abs(" + op_a + ')';
@@ -1012,6 +1014,8 @@ private:
break;
}
case OpCode::Id::F2F_R: {
// TODO(Subv): Implement rounding operations.
ASSERT_MSG(instr.conversion.f2f.rounding == 0, "Unimplemented rounding operation");
std::string op_a = regs.GetRegisterAsFloat(instr.gpr20);
if (instr.conversion.abs_a) {
@@ -1021,6 +1025,43 @@ private:
regs.SetRegisterToFloat(instr.gpr0, 0, op_a, 1, 1);
break;
}
case OpCode::Id::F2I_R: {
std::string op_a = regs.GetRegisterAsFloat(instr.gpr20);
if (instr.conversion.abs_a) {
op_a = "abs(" + op_a + ')';
}
using Tegra::Shader::FloatRoundingOp;
switch (instr.conversion.f2i.rounding) {
case FloatRoundingOp::None:
break;
case FloatRoundingOp::Floor:
op_a = "floor(" + op_a + ')';
break;
case FloatRoundingOp::Ceil:
op_a = "ceil(" + op_a + ')';
break;
case FloatRoundingOp::Trunc:
op_a = "trunc(" + op_a + ')';
break;
default:
NGLOG_CRITICAL(HW_GPU, "Unimplemented f2i rounding mode {}",
static_cast<u32>(instr.conversion.f2i.rounding.Value()));
UNREACHABLE();
break;
}
if (instr.conversion.is_output_signed) {
op_a = "int(" + op_a + ')';
} else {
op_a = "uint(" + op_a + ')';
}
regs.SetRegisterToInteger(instr.gpr0, instr.conversion.is_output_signed, 0, op_a, 1,
1);
break;
}
default: {
NGLOG_CRITICAL(HW_GPU, "Unhandled conversion instruction: {}", opcode->GetName());
UNREACHABLE();

View File

@@ -100,6 +100,8 @@ inline GLenum WrapMode(Tegra::Texture::WrapMode wrap_mode) {
switch (wrap_mode) {
case Tegra::Texture::WrapMode::Wrap:
return GL_REPEAT;
case Tegra::Texture::WrapMode::Mirror:
return GL_MIRRORED_REPEAT;
case Tegra::Texture::WrapMode::ClampToEdge:
return GL_CLAMP_TO_EDGE;
case Tegra::Texture::WrapMode::ClampOGL:

View File

@@ -55,6 +55,7 @@ u32 BytesPerPixel(TextureFormat format) {
return 16;
case TextureFormat::A8R8G8B8:
case TextureFormat::A2B10G10R10:
case TextureFormat::BF10GF11RF11:
return 4;
case TextureFormat::A1B5G5R5:
case TextureFormat::B5G6R5:
@@ -92,6 +93,7 @@ std::vector<u8> UnswizzleTexture(VAddr address, TextureFormat format, u32 width,
case TextureFormat::B5G6R5:
case TextureFormat::R8:
case TextureFormat::R16_G16_B16_A16:
case TextureFormat::BF10GF11RF11:
CopySwizzledData(width, height, bytes_per_pixel, bytes_per_pixel, data,
unswizzled_data.data(), true, block_height);
break;
@@ -118,6 +120,7 @@ std::vector<u8> DecodeTexture(const std::vector<u8>& texture_data, TextureFormat
case TextureFormat::A1B5G5R5:
case TextureFormat::B5G6R5:
case TextureFormat::R8:
case TextureFormat::BF10GF11RF11:
// TODO(Subv): For the time being just forward the same data without any decoding.
rgba_data = texture_data;
break;