From 2236f4b2c372e3764d14997f25ae2dee0de6f1ad Mon Sep 17 00:00:00 2001 From: emmauss Date: Wed, 18 Jul 2018 22:05:17 +0300 Subject: [PATCH 01/34] Implement IFileSystem:CleanDirectoryRecursively (#283) * implement ifilesys:cleandirectoryrecursively * clean up Ifilesystem --- .../OsHle/Services/FspSrv/IFileSystem.cs | 51 ++++++++++++------- 1 file changed, 32 insertions(+), 19 deletions(-) diff --git a/Ryujinx.HLE/OsHle/Services/FspSrv/IFileSystem.cs b/Ryujinx.HLE/OsHle/Services/FspSrv/IFileSystem.cs index 441b7e8add..61c6d11507 100644 --- a/Ryujinx.HLE/OsHle/Services/FspSrv/IFileSystem.cs +++ b/Ryujinx.HLE/OsHle/Services/FspSrv/IFileSystem.cs @@ -35,7 +35,7 @@ namespace Ryujinx.HLE.OsHle.Services.FspSrv { 10, Commit }, { 11, GetFreeSpaceSize }, { 12, GetTotalSpaceSize }, - //{ 13, CleanDirectoryRecursively }, + { 13, CleanDirectoryRecursively }, //{ 14, GetFileTimeStampRaw } }; @@ -46,8 +46,6 @@ namespace Ryujinx.HLE.OsHle.Services.FspSrv public long CreateFile(ServiceCtx Context) { - long Position = Context.Request.PtrBuff[0].Position; - string Name = ReadUtf8String(Context); long Mode = Context.RequestData.ReadInt64(); @@ -80,8 +78,6 @@ namespace Ryujinx.HLE.OsHle.Services.FspSrv public long DeleteFile(ServiceCtx Context) { - long Position = Context.Request.PtrBuff[0].Position; - string Name = ReadUtf8String(Context); string FileName = Context.Ns.VFs.GetFullPath(Path, Name); @@ -103,8 +99,6 @@ namespace Ryujinx.HLE.OsHle.Services.FspSrv public long CreateDirectory(ServiceCtx Context) { - long Position = Context.Request.PtrBuff[0].Position; - string Name = ReadUtf8String(Context); string DirName = Context.Ns.VFs.GetFullPath(Path, Name); @@ -141,8 +135,6 @@ namespace Ryujinx.HLE.OsHle.Services.FspSrv private long DeleteDirectory(ServiceCtx Context, bool Recursive) { - long Position = Context.Request.PtrBuff[0].Position; - string Name = ReadUtf8String(Context); string DirName = Context.Ns.VFs.GetFullPath(Path, Name); @@ -220,8 +212,6 @@ namespace Ryujinx.HLE.OsHle.Services.FspSrv public long GetEntryType(ServiceCtx Context) { - long Position = Context.Request.PtrBuff[0].Position; - string Name = ReadUtf8String(Context); string FileName = Context.Ns.VFs.GetFullPath(Path, Name); @@ -246,8 +236,6 @@ namespace Ryujinx.HLE.OsHle.Services.FspSrv public long OpenFile(ServiceCtx Context) { - long Position = Context.Request.PtrBuff[0].Position; - int FilterFlags = Context.RequestData.ReadInt32(); string Name = ReadUtf8String(Context); @@ -282,8 +270,6 @@ namespace Ryujinx.HLE.OsHle.Services.FspSrv public long OpenDirectory(ServiceCtx Context) { - long Position = Context.Request.PtrBuff[0].Position; - int FilterFlags = Context.RequestData.ReadInt32(); string Name = ReadUtf8String(Context); @@ -321,8 +307,6 @@ namespace Ryujinx.HLE.OsHle.Services.FspSrv public long GetFreeSpaceSize(ServiceCtx Context) { - long Position = Context.Request.PtrBuff[0].Position; - string Name = ReadUtf8String(Context); Context.ResponseData.Write(Context.Ns.VFs.GetDrive().AvailableFreeSpace); @@ -332,8 +316,6 @@ namespace Ryujinx.HLE.OsHle.Services.FspSrv public long GetTotalSpaceSize(ServiceCtx Context) { - long Position = Context.Request.PtrBuff[0].Position; - string Name = ReadUtf8String(Context); Context.ResponseData.Write(Context.Ns.VFs.GetDrive().TotalSize); @@ -341,6 +323,37 @@ namespace Ryujinx.HLE.OsHle.Services.FspSrv return 0; } + public long CleanDirectoryRecursively(ServiceCtx Context) + { + string Name = ReadUtf8String(Context); + + string DirName = Context.Ns.VFs.GetFullPath(Path, Name); + + if (!Directory.Exists(DirName)) + { + return MakeError(ErrorModule.Fs, FsErr.PathDoesNotExist); + } + + if (IsPathAlreadyInUse(DirName)) + { + return MakeError(ErrorModule.Fs, FsErr.PathAlreadyInUse); + } + + foreach (string Entry in Directory.EnumerateFileSystemEntries(DirName)) + { + if (Directory.Exists(Entry)) + { + Directory.Delete(Entry, true); + } + else if (File.Exists(Entry)) + { + File.Delete(Entry); + } + } + + return 0; + } + private bool IsPathAlreadyInUse(string Path) { lock (OpenPaths) From 0a13900bc93066dcb453f86d9154d52020255d32 Mon Sep 17 00:00:00 2001 From: greggameplayer <33609333+greggameplayer@users.noreply.github.com> Date: Thu, 19 Jul 2018 01:19:37 +0200 Subject: [PATCH 02/34] Implement Z24S8 TextureFormat (#247) * add Z24S8 TextureFormat * return correct PixelFormat & PixelType * return correct texture size * return correct Bytes Per Pixel * Correct PixelType --- Ryujinx.Graphics/Gal/GalTextureFormat.cs | 1 + Ryujinx.Graphics/Gal/OpenGL/OGLEnumConverter.cs | 1 + Ryujinx.HLE/Gpu/Texture/TextureHelper.cs | 1 + Ryujinx.HLE/Gpu/Texture/TextureReader.cs | 3 ++- 4 files changed, 5 insertions(+), 1 deletion(-) diff --git a/Ryujinx.Graphics/Gal/GalTextureFormat.cs b/Ryujinx.Graphics/Gal/GalTextureFormat.cs index 30f5c22974..f1e025f2a2 100644 --- a/Ryujinx.Graphics/Gal/GalTextureFormat.cs +++ b/Ryujinx.Graphics/Gal/GalTextureFormat.cs @@ -19,6 +19,7 @@ namespace Ryujinx.Graphics.Gal BC3 = 0x26, BC4 = 0x27, BC5 = 0x28, + Z24S8 = 0x29, ZF32 = 0x2f, Astc2D4x4 = 0x40, Astc2D5x5 = 0x41, diff --git a/Ryujinx.Graphics/Gal/OpenGL/OGLEnumConverter.cs b/Ryujinx.Graphics/Gal/OpenGL/OGLEnumConverter.cs index 5d20c931ef..2b0b7403a3 100644 --- a/Ryujinx.Graphics/Gal/OpenGL/OGLEnumConverter.cs +++ b/Ryujinx.Graphics/Gal/OpenGL/OGLEnumConverter.cs @@ -139,6 +139,7 @@ namespace Ryujinx.Graphics.Gal.OpenGL case GalTextureFormat.R16: return (PixelFormat.Red, PixelType.HalfFloat); case GalTextureFormat.R8: return (PixelFormat.Red, PixelType.UnsignedByte); case GalTextureFormat.ZF32: return (PixelFormat.DepthComponent, PixelType.Float); + case GalTextureFormat.Z24S8: return (PixelFormat.DepthStencil, PixelType.UnsignedInt248); } throw new NotImplementedException(Format.ToString()); diff --git a/Ryujinx.HLE/Gpu/Texture/TextureHelper.cs b/Ryujinx.HLE/Gpu/Texture/TextureHelper.cs index dbbc87d409..e934e363d8 100644 --- a/Ryujinx.HLE/Gpu/Texture/TextureHelper.cs +++ b/Ryujinx.HLE/Gpu/Texture/TextureHelper.cs @@ -37,6 +37,7 @@ namespace Ryujinx.HLE.Gpu.Texture case GalTextureFormat.A8B8G8R8: case GalTextureFormat.R32: case GalTextureFormat.ZF32: + case GalTextureFormat.Z24S8: return Texture.Width * Texture.Height * 4; case GalTextureFormat.A1B5G5R5: diff --git a/Ryujinx.HLE/Gpu/Texture/TextureReader.cs b/Ryujinx.HLE/Gpu/Texture/TextureReader.cs index 1d8c8056a6..3a5037bbb7 100644 --- a/Ryujinx.HLE/Gpu/Texture/TextureReader.cs +++ b/Ryujinx.HLE/Gpu/Texture/TextureReader.cs @@ -14,6 +14,7 @@ namespace Ryujinx.HLE.Gpu.Texture case GalTextureFormat.R16G16B16A16: return Read8Bpp (Memory, Texture); case GalTextureFormat.A8B8G8R8: return Read4Bpp (Memory, Texture); case GalTextureFormat.R32: return Read4Bpp (Memory, Texture); + case GalTextureFormat.Z24S8: return Read4Bpp (Memory, Texture); case GalTextureFormat.A1B5G5R5: return Read5551 (Memory, Texture); case GalTextureFormat.B5G6R5: return Read565 (Memory, Texture); case GalTextureFormat.G8R8: return Read2Bpp (Memory, Texture); @@ -42,7 +43,7 @@ namespace Ryujinx.HLE.Gpu.Texture case GalTextureFormat.Astc2D8x5: return Read16BptCompressedTexture(Memory, Texture, 8, 5); case GalTextureFormat.Astc2D10x5: return Read16BptCompressedTexture(Memory, Texture, 10, 5); case GalTextureFormat.Astc2D10x6: return Read16BptCompressedTexture(Memory, Texture, 10, 6); - } + } throw new NotImplementedException(Texture.Format.ToString()); } From 120fe6b74a0d2903471bfaeb25ef8265712fc576 Mon Sep 17 00:00:00 2001 From: greggameplayer <33609333+greggameplayer@users.noreply.github.com> Date: Thu, 19 Jul 2018 01:26:28 +0200 Subject: [PATCH 03/34] Implement BF10GF11RF11 TextureFormat (#246) * add BF10GF11RF11 TextureFormat * return correct PixelFormat & PixelType * return correct texture size * return correct Bytes Per Pixel * correct PixelType --- Ryujinx.Graphics/Gal/GalTextureFormat.cs | 1 + Ryujinx.Graphics/Gal/OpenGL/OGLEnumConverter.cs | 1 + Ryujinx.HLE/Gpu/Texture/TextureHelper.cs | 1 + Ryujinx.HLE/Gpu/Texture/TextureReader.cs | 1 + 4 files changed, 4 insertions(+) diff --git a/Ryujinx.Graphics/Gal/GalTextureFormat.cs b/Ryujinx.Graphics/Gal/GalTextureFormat.cs index f1e025f2a2..c478428325 100644 --- a/Ryujinx.Graphics/Gal/GalTextureFormat.cs +++ b/Ryujinx.Graphics/Gal/GalTextureFormat.cs @@ -14,6 +14,7 @@ namespace Ryujinx.Graphics.Gal G8R8 = 0x18, R16 = 0x1b, R8 = 0x1d, + BF10GF11RF11 = 0x21, BC1 = 0x24, BC2 = 0x25, BC3 = 0x26, diff --git a/Ryujinx.Graphics/Gal/OpenGL/OGLEnumConverter.cs b/Ryujinx.Graphics/Gal/OpenGL/OGLEnumConverter.cs index 2b0b7403a3..9004e9bae2 100644 --- a/Ryujinx.Graphics/Gal/OpenGL/OGLEnumConverter.cs +++ b/Ryujinx.Graphics/Gal/OpenGL/OGLEnumConverter.cs @@ -139,6 +139,7 @@ namespace Ryujinx.Graphics.Gal.OpenGL case GalTextureFormat.R16: return (PixelFormat.Red, PixelType.HalfFloat); case GalTextureFormat.R8: return (PixelFormat.Red, PixelType.UnsignedByte); case GalTextureFormat.ZF32: return (PixelFormat.DepthComponent, PixelType.Float); + case GalTextureFormat.BF10GF11RF11: return (PixelFormat.Rgb, PixelType.UnsignedInt10F11F11FRev); case GalTextureFormat.Z24S8: return (PixelFormat.DepthStencil, PixelType.UnsignedInt248); } diff --git a/Ryujinx.HLE/Gpu/Texture/TextureHelper.cs b/Ryujinx.HLE/Gpu/Texture/TextureHelper.cs index e934e363d8..de26c397d5 100644 --- a/Ryujinx.HLE/Gpu/Texture/TextureHelper.cs +++ b/Ryujinx.HLE/Gpu/Texture/TextureHelper.cs @@ -37,6 +37,7 @@ namespace Ryujinx.HLE.Gpu.Texture case GalTextureFormat.A8B8G8R8: case GalTextureFormat.R32: case GalTextureFormat.ZF32: + case GalTextureFormat.BF10GF11RF11: case GalTextureFormat.Z24S8: return Texture.Width * Texture.Height * 4; diff --git a/Ryujinx.HLE/Gpu/Texture/TextureReader.cs b/Ryujinx.HLE/Gpu/Texture/TextureReader.cs index 3a5037bbb7..26129877ab 100644 --- a/Ryujinx.HLE/Gpu/Texture/TextureReader.cs +++ b/Ryujinx.HLE/Gpu/Texture/TextureReader.cs @@ -14,6 +14,7 @@ namespace Ryujinx.HLE.Gpu.Texture case GalTextureFormat.R16G16B16A16: return Read8Bpp (Memory, Texture); case GalTextureFormat.A8B8G8R8: return Read4Bpp (Memory, Texture); case GalTextureFormat.R32: return Read4Bpp (Memory, Texture); + case GalTextureFormat.BF10GF11RF11: return Read4Bpp (Memory, Texture); case GalTextureFormat.Z24S8: return Read4Bpp (Memory, Texture); case GalTextureFormat.A1B5G5R5: return Read5551 (Memory, Texture); case GalTextureFormat.B5G6R5: return Read565 (Memory, Texture); From fa5545aab80c056fa7e1f8d516a5add79eb30d8b Mon Sep 17 00:00:00 2001 From: LDj3SNuD <35856442+LDj3SNuD@users.noreply.github.com> Date: Thu, 19 Jul 2018 02:06:28 +0200 Subject: [PATCH 04/34] Implement Ssubw_V and Usubw_V instructions. (#287) * Update AOpCodeTable.cs * Update AInstEmitSimdHelper.cs * Update AInstEmitSimdArithmetic.cs * Update AInstEmitSimdMove.cs * Update AInstEmitSimdCmp.cs * Update Instructions.cs * Update CpuTestSimd.cs * Update CpuTestSimdReg.cs --- ChocolArm64/AOpCodeTable.cs | 2 + .../Instruction/AInstEmitSimdArithmetic.cs | 10 + ChocolArm64/Instruction/AInstEmitSimdCmp.cs | 6 +- .../Instruction/AInstEmitSimdHelper.cs | 77 ++++-- ChocolArm64/Instruction/AInstEmitSimdMove.cs | 15 +- Ryujinx.Tests/Cpu/CpuTestSimd.cs | 54 ++++ Ryujinx.Tests/Cpu/CpuTestSimdReg.cs | 248 ++++++++++++++++++ Ryujinx.Tests/Cpu/Tester/Instructions.cs | 243 ++++++++++++++++- 8 files changed, 620 insertions(+), 35 deletions(-) diff --git a/ChocolArm64/AOpCodeTable.cs b/ChocolArm64/AOpCodeTable.cs index 0e979aa44f..c69de38f5a 100644 --- a/ChocolArm64/AOpCodeTable.cs +++ b/ChocolArm64/AOpCodeTable.cs @@ -387,6 +387,7 @@ namespace ChocolArm64 SetA64("0100111101xxxxxx000001xxxxxxxxxx", AInstEmit.Sshr_V, typeof(AOpCodeSimdShImm)); SetA64("0x00111100>>>xxx000101xxxxxxxxxx", AInstEmit.Ssra_V, typeof(AOpCodeSimdShImm)); SetA64("0100111101xxxxxx000101xxxxxxxxxx", AInstEmit.Ssra_V, typeof(AOpCodeSimdShImm)); + SetA64("0x001110<<1xxxxx001100xxxxxxxxxx", AInstEmit.Ssubw_V, typeof(AOpCodeSimdReg)); SetA64("0x00110000000000xxxxxxxxxxxxxxxx", AInstEmit.St__Vms, typeof(AOpCodeSimdMemMs)); SetA64("0x001100100xxxxxxxxxxxxxxxxxxxxx", AInstEmit.St__Vms, typeof(AOpCodeSimdMemMs)); SetA64("0x00110100x00000xxxxxxxxxxxxxxxx", AInstEmit.St__Vss, typeof(AOpCodeSimdMemSs)); @@ -430,6 +431,7 @@ namespace ChocolArm64 SetA64("0110111101xxxxxx000001xxxxxxxxxx", AInstEmit.Ushr_V, typeof(AOpCodeSimdShImm)); SetA64("0x10111100>>>xxx000101xxxxxxxxxx", AInstEmit.Usra_V, typeof(AOpCodeSimdShImm)); SetA64("0110111101xxxxxx000101xxxxxxxxxx", AInstEmit.Usra_V, typeof(AOpCodeSimdShImm)); + SetA64("0x101110<<1xxxxx001100xxxxxxxxxx", AInstEmit.Usubw_V, typeof(AOpCodeSimdReg)); SetA64("0>001110<<0xxxxx000110xxxxxxxxxx", AInstEmit.Uzp1_V, typeof(AOpCodeSimdReg)); SetA64("0>001110<<0xxxxx010110xxxxxxxxxx", AInstEmit.Uzp2_V, typeof(AOpCodeSimdReg)); SetA64("0x001110<<100001001010xxxxxxxxxx", AInstEmit.Xtn_V, typeof(AOpCodeSimd)); diff --git a/ChocolArm64/Instruction/AInstEmitSimdArithmetic.cs b/ChocolArm64/Instruction/AInstEmitSimdArithmetic.cs index 36bb1cbf16..8e7418611e 100644 --- a/ChocolArm64/Instruction/AInstEmitSimdArithmetic.cs +++ b/ChocolArm64/Instruction/AInstEmitSimdArithmetic.cs @@ -1072,6 +1072,11 @@ namespace ChocolArm64.Instruction EmitVectorSaturatingNarrowOpSxZx(Context, () => { }); } + public static void Ssubw_V(AILEmitterCtx Context) + { + EmitVectorWidenRmBinaryOpSx(Context, () => Context.Emit(OpCodes.Sub)); + } + public static void Sub_S(AILEmitterCtx Context) { EmitScalarBinaryOpZx(Context, () => Context.Emit(OpCodes.Sub)); @@ -1225,5 +1230,10 @@ namespace ChocolArm64.Instruction { EmitVectorSaturatingNarrowOpZxZx(Context, () => { }); } + + public static void Usubw_V(AILEmitterCtx Context) + { + EmitVectorWidenRmBinaryOpZx(Context, () => Context.Emit(OpCodes.Sub)); + } } } diff --git a/ChocolArm64/Instruction/AInstEmitSimdCmp.cs b/ChocolArm64/Instruction/AInstEmitSimdCmp.cs index 773d989447..c2d47747ed 100644 --- a/ChocolArm64/Instruction/AInstEmitSimdCmp.cs +++ b/ChocolArm64/Instruction/AInstEmitSimdCmp.cs @@ -364,7 +364,7 @@ namespace ChocolArm64.Instruction AOpCodeSimd Op = (AOpCodeSimd)Context.CurrOp; int Bytes = Op.GetBitsCount() >> 3; - int Elems = (!Scalar ? Bytes >> Op.Size : 1); + int Elems = !Scalar ? Bytes >> Op.Size : 1; ulong SzMask = ulong.MaxValue >> (64 - (8 << Op.Size)); @@ -408,7 +408,7 @@ namespace ChocolArm64.Instruction AOpCodeSimdReg Op = (AOpCodeSimdReg)Context.CurrOp; int Bytes = Op.GetBitsCount() >> 3; - int Elems = (!Scalar ? Bytes >> Op.Size : 1); + int Elems = !Scalar ? Bytes >> Op.Size : 1; ulong SzMask = ulong.MaxValue >> (64 - (8 << Op.Size)); @@ -522,4 +522,4 @@ namespace ChocolArm64.Instruction Context.MarkLabel(LblEnd); } } -} \ No newline at end of file +} diff --git a/ChocolArm64/Instruction/AInstEmitSimdHelper.cs b/ChocolArm64/Instruction/AInstEmitSimdHelper.cs index 7716e2987a..0bd1a62926 100644 --- a/ChocolArm64/Instruction/AInstEmitSimdHelper.cs +++ b/ChocolArm64/Instruction/AInstEmitSimdHelper.cs @@ -419,20 +419,25 @@ namespace ChocolArm64.Instruction int SizeF = Op.Size & 1; int Bytes = Op.GetBitsCount() >> 3; + int Elems = Bytes >> SizeF + 2; - for (int Index = 0; Index < (Bytes >> SizeF + 2); Index++) + bool Rd = (Opers & OperFlags.Rd) != 0; + bool Rn = (Opers & OperFlags.Rn) != 0; + bool Rm = (Opers & OperFlags.Rm) != 0; + + for (int Index = 0; Index < Elems; Index++) { - if (Opers.HasFlag(OperFlags.Rd)) + if (Rd) { EmitVectorExtractF(Context, Op.Rd, Index, SizeF); } - if (Opers.HasFlag(OperFlags.Rn)) + if (Rn) { EmitVectorExtractF(Context, Op.Rn, Index, SizeF); } - if (Opers.HasFlag(OperFlags.Rm)) + if (Rm) { EmitVectorExtractF(Context, ((AOpCodeSimdReg)Op).Rm, Index, SizeF); } @@ -469,8 +474,9 @@ namespace ChocolArm64.Instruction int SizeF = Op.Size & 1; int Bytes = Op.GetBitsCount() >> 3; + int Elems = Bytes >> SizeF + 2; - for (int Index = 0; Index < (Bytes >> SizeF + 2); Index++) + for (int Index = 0; Index < Elems; Index++) { if (Ternary) { @@ -531,19 +537,23 @@ namespace ChocolArm64.Instruction int Bytes = Op.GetBitsCount() >> 3; int Elems = Bytes >> Op.Size; + bool Rd = (Opers & OperFlags.Rd) != 0; + bool Rn = (Opers & OperFlags.Rn) != 0; + bool Rm = (Opers & OperFlags.Rm) != 0; + for (int Index = 0; Index < Elems; Index++) { - if (Opers.HasFlag(OperFlags.Rd)) + if (Rd) { EmitVectorExtract(Context, Op.Rd, Index, Op.Size, Signed); } - if (Opers.HasFlag(OperFlags.Rn)) + if (Rn) { EmitVectorExtract(Context, Op.Rn, Index, Op.Size, Signed); } - if (Opers.HasFlag(OperFlags.Rm)) + if (Rm) { EmitVectorExtract(Context, ((AOpCodeSimdReg)Op).Rm, Index, Op.Size, Signed); } @@ -662,9 +672,6 @@ namespace ChocolArm64.Instruction { AOpCodeSimdReg Op = (AOpCodeSimdReg)Context.CurrOp; - Context.EmitLdvec(Op.Rd); - Context.EmitStvectmp(); - int Elems = 8 >> Op.Size; int Part = Op.RegisterSize == ARegisterSize.SIMD128 ? Elems : 0; @@ -707,9 +714,6 @@ namespace ChocolArm64.Instruction { AOpCodeSimdReg Op = (AOpCodeSimdReg)Context.CurrOp; - Context.EmitLdvec(Op.Rd); - Context.EmitStvectmp(); - int Elems = 8 >> Op.Size; int Part = Op.RegisterSize == ARegisterSize.SIMD128 ? Elems : 0; @@ -747,21 +751,25 @@ namespace ChocolArm64.Instruction { AOpCodeSimdReg Op = (AOpCodeSimdReg)Context.CurrOp; - int Bytes = Op.GetBitsCount() >> 3; + int Words = Op.GetBitsCount() >> 4; + int Pairs = Words >> Op.Size; - int Elems = Bytes >> Op.Size; - int Half = Elems >> 1; - - for (int Index = 0; Index < Elems; Index++) + for (int Index = 0; Index < Pairs; Index++) { - int Elem = (Index & (Half - 1)) << 1; + int Idx = Index << 1; - EmitVectorExtract(Context, Index < Half ? Op.Rn : Op.Rm, Elem + 0, Op.Size, Signed); - EmitVectorExtract(Context, Index < Half ? Op.Rn : Op.Rm, Elem + 1, Op.Size, Signed); + EmitVectorExtract(Context, Op.Rn, Idx, Op.Size, Signed); + EmitVectorExtract(Context, Op.Rn, Idx + 1, Op.Size, Signed); Emit(); - EmitVectorInsertTmp(Context, Index, Op.Size); + EmitVectorExtract(Context, Op.Rm, Idx, Op.Size, Signed); + EmitVectorExtract(Context, Op.Rm, Idx + 1, Op.Size, Signed); + + Emit(); + + EmitVectorInsertTmp(Context, Pairs + Index, Op.Size); + EmitVectorInsertTmp(Context, Index, Op.Size); } Context.EmitLdvectmp(); @@ -818,7 +826,7 @@ namespace ChocolArm64.Instruction int Part = !Scalar && (Op.RegisterSize == ARegisterSize.SIMD128) ? Elems : 0; - long TMaxValue = SignedDst ? (1 << (ESize - 1)) - 1 : (1L << ESize) - 1L; + long TMaxValue = SignedDst ? (1 << (ESize - 1)) - 1 : (long)(~0UL >> (64 - ESize)); long TMinValue = SignedDst ? -((1 << (ESize - 1))) : 0; Context.EmitLdc_I8(0L); @@ -871,7 +879,7 @@ namespace ChocolArm64.Instruction if (Scalar) { - EmitVectorZeroLower(Context, Op.Rd); + EmitVectorZeroLowerTmp(Context); } EmitVectorInsertTmp(Context, Part + Index, Op.Size); @@ -963,6 +971,11 @@ namespace ChocolArm64.Instruction EmitVectorInsert(Context, Rd, 0, 3, 0); } + public static void EmitVectorZeroLowerTmp(AILEmitterCtx Context) + { + EmitVectorInsertTmp(Context, 0, 3, 0); + } + public static void EmitVectorZeroUpper(AILEmitterCtx Context, int Rd) { EmitVectorInsert(Context, Rd, 1, 3, 0); @@ -1008,6 +1021,20 @@ namespace ChocolArm64.Instruction Context.EmitStvec(Reg); } + public static void EmitVectorInsertTmp(AILEmitterCtx Context, int Index, int Size, long Value) + { + ThrowIfInvalid(Index, Size); + + Context.EmitLdc_I8(Value); + Context.EmitLdvectmp(); + Context.EmitLdc_I4(Index); + Context.EmitLdc_I4(Size); + + AVectorHelper.EmitCall(Context, nameof(AVectorHelper.VectorInsertInt)); + + Context.EmitStvectmp(); + } + public static void EmitVectorInsertF(AILEmitterCtx Context, int Reg, int Index, int Size) { ThrowIfInvalidF(Index, Size); diff --git a/ChocolArm64/Instruction/AInstEmitSimdMove.cs b/ChocolArm64/Instruction/AInstEmitSimdMove.cs index 592cab733e..3bf1e4635b 100644 --- a/ChocolArm64/Instruction/AInstEmitSimdMove.cs +++ b/ChocolArm64/Instruction/AInstEmitSimdMove.cs @@ -295,13 +295,22 @@ namespace ChocolArm64.Instruction int Part = Op.RegisterSize == ARegisterSize.SIMD128 ? Elems : 0; + if (Part != 0) + { + Context.EmitLdvec(Op.Rd); + Context.EmitStvectmp(); + } + for (int Index = 0; Index < Elems; Index++) { EmitVectorExtractZx(Context, Op.Rn, Index, Op.Size + 1); - EmitVectorInsert(Context, Op.Rd, Part + Index, Op.Size); + EmitVectorInsertTmp(Context, Part + Index, Op.Size); } + Context.EmitLdvectmp(); + Context.EmitStvec(Op.Rd); + if (Part == 0) { EmitVectorZeroUpper(Context, Op.Rd); @@ -342,7 +351,7 @@ namespace ChocolArm64.Instruction EmitVectorExtractZx(Context, Op.Rm, Idx + Part, Op.Size); EmitVectorInsertTmp(Context, Idx + 1, Op.Size); - EmitVectorInsertTmp(Context, Idx , Op.Size); + EmitVectorInsertTmp(Context, Idx, Op.Size); } Context.EmitLdvectmp(); @@ -398,7 +407,7 @@ namespace ChocolArm64.Instruction EmitVectorExtractZx(Context, Op.Rm, Base + Index, Op.Size); EmitVectorInsertTmp(Context, Idx + 1, Op.Size); - EmitVectorInsertTmp(Context, Idx , Op.Size); + EmitVectorInsertTmp(Context, Idx, Op.Size); } Context.EmitLdvectmp(); diff --git a/Ryujinx.Tests/Cpu/CpuTestSimd.cs b/Ryujinx.Tests/Cpu/CpuTestSimd.cs index b84d29575d..82591edaee 100644 --- a/Ryujinx.Tests/Cpu/CpuTestSimd.cs +++ b/Ryujinx.Tests/Cpu/CpuTestSimd.cs @@ -1377,6 +1377,60 @@ namespace Ryujinx.Tests.Cpu }); Assert.That(((ThreadState.Fpsr >> 27) & 1) != 0, Is.EqualTo(Shared.FPSR[27])); } + + [Test, Description("XTN{2} ., .")] + public void Xtn_V_8H8B_4S4H_2D2S([Values(0u)] uint Rd, + [Values(1u, 0u)] uint Rn, + [ValueSource("_4H2S1D_")] [Random(RndCnt)] ulong Z, + [ValueSource("_4H2S1D_")] [Random(RndCnt)] ulong A, + [Values(0b00u, 0b01u, 0b10u)] uint size) // <8H8B, 4S4H, 2D2S> + { + uint Opcode = 0x0E212800; // XTN V0.8B, V0.8H + Opcode |= ((Rn & 31) << 5) | ((Rd & 31) << 0); + Opcode |= ((size & 3) << 22); + Bits Op = new Bits(Opcode); + + Vector128 V0 = MakeVectorE0E1(Z, Z); + Vector128 V1 = MakeVectorE0E1(A, A); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1); + + AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); + AArch64.Vpart(1, 0, new Bits(A)); AArch64.Vpart(1, 1, new Bits(A)); + SimdFp.Xtn_V(Op[30], Op[23, 22], Op[9, 5], Op[4, 0]); + + Assert.Multiple(() => + { + Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); + Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); + }); + } + + [Test, Description("XTN{2} ., .")] + public void Xtn_V_8H16B_4S8H_2D4S([Values(0u)] uint Rd, + [Values(1u, 0u)] uint Rn, + [ValueSource("_4H2S1D_")] [Random(RndCnt)] ulong Z, + [ValueSource("_4H2S1D_")] [Random(RndCnt)] ulong A, + [Values(0b00u, 0b01u, 0b10u)] uint size) // <8H16B, 4S8H, 2D4S> + { + uint Opcode = 0x4E212800; // XTN2 V0.16B, V0.8H + Opcode |= ((Rn & 31) << 5) | ((Rd & 31) << 0); + Opcode |= ((size & 3) << 22); + Bits Op = new Bits(Opcode); + + Vector128 V0 = MakeVectorE0E1(Z, Z); + Vector128 V1 = MakeVectorE0E1(A, A); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1); + + AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); + AArch64.Vpart(1, 0, new Bits(A)); AArch64.Vpart(1, 1, new Bits(A)); + SimdFp.Xtn_V(Op[30], Op[23, 22], Op[9, 5], Op[4, 0]); + + Assert.Multiple(() => + { + Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); + Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); + }); + } #endif } } diff --git a/Ryujinx.Tests/Cpu/CpuTestSimdReg.cs b/Ryujinx.Tests/Cpu/CpuTestSimdReg.cs index 51db857c31..c67348d1da 100644 --- a/Ryujinx.Tests/Cpu/CpuTestSimdReg.cs +++ b/Ryujinx.Tests/Cpu/CpuTestSimdReg.cs @@ -1659,6 +1659,130 @@ namespace Ryujinx.Tests.Cpu }); } + [Test, Pairwise, Description("SADDW{2} ., ., .")] + public void Saddw_V_8B8H8H_4H4S4S_2S2D2D([Values(0u)] uint Rd, + [Values(1u, 0u)] uint Rn, + [Values(2u, 0u)] uint Rm, + [ValueSource("_8B4H2S1D_")] [Random(RndCnt)] ulong Z, + [ValueSource("_4H2S1D_")] [Random(RndCnt)] ulong A, + [ValueSource("_8B4H2S_")] [Random(RndCnt)] ulong B, + [Values(0b00u, 0b01u, 0b10u)] uint size) // <8B8H8H, 4H4S4S, 2S2D2D> + { + uint Opcode = 0x0E201000; // SADDW V0.8H, V0.8H, V0.8B + Opcode |= ((Rm & 31) << 16) | ((Rn & 31) << 5) | ((Rd & 31) << 0); + Opcode |= ((size & 3) << 22); + Bits Op = new Bits(Opcode); + + Vector128 V0 = MakeVectorE0E1(Z, Z); + Vector128 V1 = MakeVectorE0E1(A, A); + Vector128 V2 = MakeVectorE0(B); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, V2: V2); + + AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); + AArch64.Vpart(1, 0, new Bits(A)); AArch64.Vpart(1, 1, new Bits(A)); + AArch64.Vpart(2, 0, new Bits(B)); + SimdFp.Saddw_V(Op[30], Op[23, 22], Op[20, 16], Op[9, 5], Op[4, 0]); + + Assert.Multiple(() => + { + Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); + Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); + }); + } + + [Test, Pairwise, Description("SADDW{2} ., ., .")] + public void Saddw_V_16B8H8H_8H4S4S_4S2D2D([Values(0u)] uint Rd, + [Values(1u, 0u)] uint Rn, + [Values(2u, 0u)] uint Rm, + [ValueSource("_8B4H2S1D_")] [Random(RndCnt)] ulong Z, + [ValueSource("_4H2S1D_")] [Random(RndCnt)] ulong A, + [ValueSource("_8B4H2S_")] [Random(RndCnt)] ulong B, + [Values(0b00u, 0b01u, 0b10u)] uint size) // <16B8H8H, 8H4S4S, 4S2D2D> + { + uint Opcode = 0x4E201000; // SADDW2 V0.8H, V0.8H, V0.16B + Opcode |= ((Rm & 31) << 16) | ((Rn & 31) << 5) | ((Rd & 31) << 0); + Opcode |= ((size & 3) << 22); + Bits Op = new Bits(Opcode); + + Vector128 V0 = MakeVectorE0E1(Z, Z); + Vector128 V1 = MakeVectorE0E1(A, A); + Vector128 V2 = MakeVectorE1(B); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, V2: V2); + + AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); + AArch64.Vpart(1, 0, new Bits(A)); AArch64.Vpart(1, 1, new Bits(A)); + AArch64.Vpart(2, 1, new Bits(B)); + SimdFp.Saddw_V(Op[30], Op[23, 22], Op[20, 16], Op[9, 5], Op[4, 0]); + + Assert.Multiple(() => + { + Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); + Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); + }); + } + + [Test, Pairwise, Description("SSUBW{2} ., ., .")] + public void Ssubw_V_8B8H8H_4H4S4S_2S2D2D([Values(0u)] uint Rd, + [Values(1u, 0u)] uint Rn, + [Values(2u, 0u)] uint Rm, + [ValueSource("_8B4H2S1D_")] [Random(RndCnt)] ulong Z, + [ValueSource("_4H2S1D_")] [Random(RndCnt)] ulong A, + [ValueSource("_8B4H2S_")] [Random(RndCnt)] ulong B, + [Values(0b00u, 0b01u, 0b10u)] uint size) // <8B8H8H, 4H4S4S, 2S2D2D> + { + uint Opcode = 0x0E203000; // SSUBW V0.8H, V0.8H, V0.8B + Opcode |= ((Rm & 31) << 16) | ((Rn & 31) << 5) | ((Rd & 31) << 0); + Opcode |= ((size & 3) << 22); + Bits Op = new Bits(Opcode); + + Vector128 V0 = MakeVectorE0E1(Z, Z); + Vector128 V1 = MakeVectorE0E1(A, A); + Vector128 V2 = MakeVectorE0(B); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, V2: V2); + + AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); + AArch64.Vpart(1, 0, new Bits(A)); AArch64.Vpart(1, 1, new Bits(A)); + AArch64.Vpart(2, 0, new Bits(B)); + SimdFp.Ssubw_V(Op[30], Op[23, 22], Op[20, 16], Op[9, 5], Op[4, 0]); + + Assert.Multiple(() => + { + Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); + Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); + }); + } + + [Test, Pairwise, Description("SSUBW{2} ., ., .")] + public void Ssubw_V_16B8H8H_8H4S4S_4S2D2D([Values(0u)] uint Rd, + [Values(1u, 0u)] uint Rn, + [Values(2u, 0u)] uint Rm, + [ValueSource("_8B4H2S1D_")] [Random(RndCnt)] ulong Z, + [ValueSource("_4H2S1D_")] [Random(RndCnt)] ulong A, + [ValueSource("_8B4H2S_")] [Random(RndCnt)] ulong B, + [Values(0b00u, 0b01u, 0b10u)] uint size) // <16B8H8H, 8H4S4S, 4S2D2D> + { + uint Opcode = 0x4E203000; // SSUBW2 V0.8H, V0.8H, V0.16B + Opcode |= ((Rm & 31) << 16) | ((Rn & 31) << 5) | ((Rd & 31) << 0); + Opcode |= ((size & 3) << 22); + Bits Op = new Bits(Opcode); + + Vector128 V0 = MakeVectorE0E1(Z, Z); + Vector128 V1 = MakeVectorE0E1(A, A); + Vector128 V2 = MakeVectorE1(B); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, V2: V2); + + AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); + AArch64.Vpart(1, 0, new Bits(A)); AArch64.Vpart(1, 1, new Bits(A)); + AArch64.Vpart(2, 1, new Bits(B)); + SimdFp.Ssubw_V(Op[30], Op[23, 22], Op[20, 16], Op[9, 5], Op[4, 0]); + + Assert.Multiple(() => + { + Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); + Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); + }); + } + [Test, Pairwise, Description("SUB , , ")] public void Sub_S_D([Values(0u)] uint Rd, [Values(1u, 0u)] uint Rn, @@ -2184,6 +2308,130 @@ namespace Ryujinx.Tests.Cpu }); } + [Test, Pairwise, Description("UADDW{2} ., ., .")] + public void Uaddw_V_8B8H8H_4H4S4S_2S2D2D([Values(0u)] uint Rd, + [Values(1u, 0u)] uint Rn, + [Values(2u, 0u)] uint Rm, + [ValueSource("_8B4H2S1D_")] [Random(RndCnt)] ulong Z, + [ValueSource("_4H2S1D_")] [Random(RndCnt)] ulong A, + [ValueSource("_8B4H2S_")] [Random(RndCnt)] ulong B, + [Values(0b00u, 0b01u, 0b10u)] uint size) // <8B8H8H, 4H4S4S, 2S2D2D> + { + uint Opcode = 0x2E201000; // UADDW V0.8H, V0.8H, V0.8B + Opcode |= ((Rm & 31) << 16) | ((Rn & 31) << 5) | ((Rd & 31) << 0); + Opcode |= ((size & 3) << 22); + Bits Op = new Bits(Opcode); + + Vector128 V0 = MakeVectorE0E1(Z, Z); + Vector128 V1 = MakeVectorE0E1(A, A); + Vector128 V2 = MakeVectorE0(B); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, V2: V2); + + AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); + AArch64.Vpart(1, 0, new Bits(A)); AArch64.Vpart(1, 1, new Bits(A)); + AArch64.Vpart(2, 0, new Bits(B)); + SimdFp.Uaddw_V(Op[30], Op[23, 22], Op[20, 16], Op[9, 5], Op[4, 0]); + + Assert.Multiple(() => + { + Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); + Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); + }); + } + + [Test, Pairwise, Description("UADDW{2} ., ., .")] + public void Uaddw_V_16B8H8H_8H4S4S_4S2D2D([Values(0u)] uint Rd, + [Values(1u, 0u)] uint Rn, + [Values(2u, 0u)] uint Rm, + [ValueSource("_8B4H2S1D_")] [Random(RndCnt)] ulong Z, + [ValueSource("_4H2S1D_")] [Random(RndCnt)] ulong A, + [ValueSource("_8B4H2S_")] [Random(RndCnt)] ulong B, + [Values(0b00u, 0b01u, 0b10u)] uint size) // <16B8H8H, 8H4S4S, 4S2D2D> + { + uint Opcode = 0x6E201000; // UADDW2 V0.8H, V0.8H, V0.16B + Opcode |= ((Rm & 31) << 16) | ((Rn & 31) << 5) | ((Rd & 31) << 0); + Opcode |= ((size & 3) << 22); + Bits Op = new Bits(Opcode); + + Vector128 V0 = MakeVectorE0E1(Z, Z); + Vector128 V1 = MakeVectorE0E1(A, A); + Vector128 V2 = MakeVectorE1(B); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, V2: V2); + + AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); + AArch64.Vpart(1, 0, new Bits(A)); AArch64.Vpart(1, 1, new Bits(A)); + AArch64.Vpart(2, 1, new Bits(B)); + SimdFp.Uaddw_V(Op[30], Op[23, 22], Op[20, 16], Op[9, 5], Op[4, 0]); + + Assert.Multiple(() => + { + Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); + Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); + }); + } + + [Test, Pairwise, Description("USUBW{2} ., ., .")] + public void Usubw_V_8B8H8H_4H4S4S_2S2D2D([Values(0u)] uint Rd, + [Values(1u, 0u)] uint Rn, + [Values(2u, 0u)] uint Rm, + [ValueSource("_8B4H2S1D_")] [Random(RndCnt)] ulong Z, + [ValueSource("_4H2S1D_")] [Random(RndCnt)] ulong A, + [ValueSource("_8B4H2S_")] [Random(RndCnt)] ulong B, + [Values(0b00u, 0b01u, 0b10u)] uint size) // <8B8H8H, 4H4S4S, 2S2D2D> + { + uint Opcode = 0x2E203000; // USUBW V0.8H, V0.8H, V0.8B + Opcode |= ((Rm & 31) << 16) | ((Rn & 31) << 5) | ((Rd & 31) << 0); + Opcode |= ((size & 3) << 22); + Bits Op = new Bits(Opcode); + + Vector128 V0 = MakeVectorE0E1(Z, Z); + Vector128 V1 = MakeVectorE0E1(A, A); + Vector128 V2 = MakeVectorE0(B); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, V2: V2); + + AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); + AArch64.Vpart(1, 0, new Bits(A)); AArch64.Vpart(1, 1, new Bits(A)); + AArch64.Vpart(2, 0, new Bits(B)); + SimdFp.Usubw_V(Op[30], Op[23, 22], Op[20, 16], Op[9, 5], Op[4, 0]); + + Assert.Multiple(() => + { + Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); + Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); + }); + } + + [Test, Pairwise, Description("USUBW{2} ., ., .")] + public void Usubw_V_16B8H8H_8H4S4S_4S2D2D([Values(0u)] uint Rd, + [Values(1u, 0u)] uint Rn, + [Values(2u, 0u)] uint Rm, + [ValueSource("_8B4H2S1D_")] [Random(RndCnt)] ulong Z, + [ValueSource("_4H2S1D_")] [Random(RndCnt)] ulong A, + [ValueSource("_8B4H2S_")] [Random(RndCnt)] ulong B, + [Values(0b00u, 0b01u, 0b10u)] uint size) // <16B8H8H, 8H4S4S, 4S2D2D> + { + uint Opcode = 0x6E203000; // USUBW2 V0.8H, V0.8H, V0.16B + Opcode |= ((Rm & 31) << 16) | ((Rn & 31) << 5) | ((Rd & 31) << 0); + Opcode |= ((size & 3) << 22); + Bits Op = new Bits(Opcode); + + Vector128 V0 = MakeVectorE0E1(Z, Z); + Vector128 V1 = MakeVectorE0E1(A, A); + Vector128 V2 = MakeVectorE1(B); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, V2: V2); + + AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); + AArch64.Vpart(1, 0, new Bits(A)); AArch64.Vpart(1, 1, new Bits(A)); + AArch64.Vpart(2, 1, new Bits(B)); + SimdFp.Usubw_V(Op[30], Op[23, 22], Op[20, 16], Op[9, 5], Op[4, 0]); + + Assert.Multiple(() => + { + Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); + Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); + }); + } + [Test, Pairwise, Description("UZP1 ., ., .")] public void Uzp1_V_8B_4H_2S([Values(0u)] uint Rd, [Values(1u, 0u)] uint Rn, diff --git a/Ryujinx.Tests/Cpu/Tester/Instructions.cs b/Ryujinx.Tests/Cpu/Tester/Instructions.cs index 68f83423ba..2f52dcbf55 100644 --- a/Ryujinx.Tests/Cpu/Tester/Instructions.cs +++ b/Ryujinx.Tests/Cpu/Tester/Instructions.cs @@ -3315,6 +3315,37 @@ namespace Ryujinx.Tests.Cpu.Tester Vpart(d, part, result); } + + // xtn_advsimd.html + public static void Xtn_V(bool Q, Bits size, Bits Rn, Bits Rd) + { + /* Decode Vector */ + int d = (int)UInt(Rd); + int n = (int)UInt(Rn); + + /* if size == '11' then ReservedValue(); */ + + int esize = 8 << (int)UInt(size); + int datasize = 64; + int part = (int)UInt(Q); + int elements = datasize / esize; + + /* Operation */ + /* CheckFPAdvSIMDEnabled64(); */ + + Bits result = new Bits(datasize); + Bits operand = V(2 * datasize, n); + Bits element; + + for (int e = 0; e <= elements - 1; e++) + { + element = Elem(operand, e, 2 * esize); + + Elem(result, e, esize, element[esize - 1, 0]); + } + + Vpart(d, part, result); + } #endregion #region "SimdReg" @@ -4395,8 +4426,8 @@ namespace Ryujinx.Tests.Cpu.Tester int part = (int)UInt(Q); int elements = datasize / esize; - bool unsigned = (U == true); bool accumulate = (op == false); + bool unsigned = (U == true); /* Operation */ /* CheckFPAdvSIMDEnabled64(); */ @@ -4484,8 +4515,8 @@ namespace Ryujinx.Tests.Cpu.Tester int part = (int)UInt(Q); int elements = datasize / esize; - bool unsigned = (U == true); bool accumulate = (op == false); + bool unsigned = (U == true); /* Operation */ /* CheckFPAdvSIMDEnabled64(); */ @@ -4511,6 +4542,108 @@ namespace Ryujinx.Tests.Cpu.Tester V(d, result); } + // saddw_advsimd.html + public static void Saddw_V(bool Q, Bits size, Bits Rm, Bits Rn, Bits Rd) + { + const bool U = false; + const bool o1 = false; + + /* Decode */ + int d = (int)UInt(Rd); + int n = (int)UInt(Rn); + int m = (int)UInt(Rm); + + /* if size == '11' then ReservedValue(); */ + + int esize = 8 << (int)UInt(size); + int datasize = 64; + int part = (int)UInt(Q); + int elements = datasize / esize; + + bool sub_op = (o1 == true); + bool unsigned = (U == true); + + /* Operation */ + /* CheckFPAdvSIMDEnabled64(); */ + + Bits result = new Bits(2 * datasize); + Bits operand1 = V(2 * datasize, n); + Bits operand2 = Vpart(datasize, m, part); + BigInteger element1; + BigInteger element2; + BigInteger sum; + + for (int e = 0; e <= elements - 1; e++) + { + element1 = Int(Elem(operand1, e, 2 * esize), unsigned); + element2 = Int(Elem(operand2, e, esize), unsigned); + + if (sub_op) + { + sum = element1 - element2; + } + else + { + sum = element1 + element2; + } + + Elem(result, e, 2 * esize, sum.SubBigInteger(2 * esize - 1, 0)); + } + + V(d, result); + } + + // ssubw_advsimd.html + public static void Ssubw_V(bool Q, Bits size, Bits Rm, Bits Rn, Bits Rd) + { + const bool U = false; + const bool o1 = true; + + /* Decode */ + int d = (int)UInt(Rd); + int n = (int)UInt(Rn); + int m = (int)UInt(Rm); + + /* if size == '11' then ReservedValue(); */ + + int esize = 8 << (int)UInt(size); + int datasize = 64; + int part = (int)UInt(Q); + int elements = datasize / esize; + + bool sub_op = (o1 == true); + bool unsigned = (U == true); + + /* Operation */ + /* CheckFPAdvSIMDEnabled64(); */ + + Bits result = new Bits(2 * datasize); + Bits operand1 = V(2 * datasize, n); + Bits operand2 = Vpart(datasize, m, part); + BigInteger element1; + BigInteger element2; + BigInteger sum; + + for (int e = 0; e <= elements - 1; e++) + { + element1 = Int(Elem(operand1, e, 2 * esize), unsigned); + element2 = Int(Elem(operand2, e, esize), unsigned); + + if (sub_op) + { + sum = element1 - element2; + } + else + { + sum = element1 + element2; + } + + Elem(result, e, 2 * esize, sum.SubBigInteger(2 * esize - 1, 0)); + } + + V(d, result); + } + // sub_advsimd.html#SUB_asisdsame_only public static void Sub_S(Bits size, Bits Rm, Bits Rn, Bits Rd) { @@ -4785,8 +4918,8 @@ namespace Ryujinx.Tests.Cpu.Tester int part = (int)UInt(Q); int elements = datasize / esize; - bool unsigned = (U == true); bool accumulate = (op == false); + bool unsigned = (U == true); /* Operation */ /* CheckFPAdvSIMDEnabled64(); */ @@ -4874,8 +5007,8 @@ namespace Ryujinx.Tests.Cpu.Tester int part = (int)UInt(Q); int elements = datasize / esize; - bool unsigned = (U == true); bool accumulate = (op == false); + bool unsigned = (U == true); /* Operation */ /* CheckFPAdvSIMDEnabled64(); */ @@ -4901,6 +5034,108 @@ namespace Ryujinx.Tests.Cpu.Tester V(d, result); } + // uaddw_advsimd.html + public static void Uaddw_V(bool Q, Bits size, Bits Rm, Bits Rn, Bits Rd) + { + const bool U = true; + const bool o1 = false; + + /* Decode */ + int d = (int)UInt(Rd); + int n = (int)UInt(Rn); + int m = (int)UInt(Rm); + + /* if size == '11' then ReservedValue(); */ + + int esize = 8 << (int)UInt(size); + int datasize = 64; + int part = (int)UInt(Q); + int elements = datasize / esize; + + bool sub_op = (o1 == true); + bool unsigned = (U == true); + + /* Operation */ + /* CheckFPAdvSIMDEnabled64(); */ + + Bits result = new Bits(2 * datasize); + Bits operand1 = V(2 * datasize, n); + Bits operand2 = Vpart(datasize, m, part); + BigInteger element1; + BigInteger element2; + BigInteger sum; + + for (int e = 0; e <= elements - 1; e++) + { + element1 = Int(Elem(operand1, e, 2 * esize), unsigned); + element2 = Int(Elem(operand2, e, esize), unsigned); + + if (sub_op) + { + sum = element1 - element2; + } + else + { + sum = element1 + element2; + } + + Elem(result, e, 2 * esize, sum.SubBigInteger(2 * esize - 1, 0)); + } + + V(d, result); + } + + // usubw_advsimd.html + public static void Usubw_V(bool Q, Bits size, Bits Rm, Bits Rn, Bits Rd) + { + const bool U = true; + const bool o1 = true; + + /* Decode */ + int d = (int)UInt(Rd); + int n = (int)UInt(Rn); + int m = (int)UInt(Rm); + + /* if size == '11' then ReservedValue(); */ + + int esize = 8 << (int)UInt(size); + int datasize = 64; + int part = (int)UInt(Q); + int elements = datasize / esize; + + bool sub_op = (o1 == true); + bool unsigned = (U == true); + + /* Operation */ + /* CheckFPAdvSIMDEnabled64(); */ + + Bits result = new Bits(2 * datasize); + Bits operand1 = V(2 * datasize, n); + Bits operand2 = Vpart(datasize, m, part); + BigInteger element1; + BigInteger element2; + BigInteger sum; + + for (int e = 0; e <= elements - 1; e++) + { + element1 = Int(Elem(operand1, e, 2 * esize), unsigned); + element2 = Int(Elem(operand2, e, esize), unsigned); + + if (sub_op) + { + sum = element1 - element2; + } + else + { + sum = element1 + element2; + } + + Elem(result, e, 2 * esize, sum.SubBigInteger(2 * esize - 1, 0)); + } + + V(d, result); + } + // uzp1_advsimd.html public static void Uzp1_V(bool Q, Bits size, Bits Rm, Bits Rn, Bits Rd) { From 2795af038df06d2f8e0dbbf0fd271bbac5da59a2 Mon Sep 17 00:00:00 2001 From: simonmkwii-dev <40786398+simonmkwii-dev@users.noreply.github.com> Date: Thu, 19 Jul 2018 11:06:45 +1000 Subject: [PATCH 05/34] Implement GetCurrentIpAddress() and stub GetCurrentNetworkProfile() (#271) * Implement GetCurrentIpAddress() ...and stub GetCurrentNetworkProfile() * Update IGeneralService.cs * Actually implement it properly this time... * * Made some requested changes * Added requested changes. * Added more requested changes. * oof * Local > Address * Cyuubumped * Change PrintInfo > PrintDebug * Revert change --- .../OsHle/Services/Nifm/IGeneralService.cs | 29 +++++++++++++++++-- 1 file changed, 27 insertions(+), 2 deletions(-) diff --git a/Ryujinx.HLE/OsHle/Services/Nifm/IGeneralService.cs b/Ryujinx.HLE/OsHle/Services/Nifm/IGeneralService.cs index e289a8db8f..66b1f1b610 100644 --- a/Ryujinx.HLE/OsHle/Services/Nifm/IGeneralService.cs +++ b/Ryujinx.HLE/OsHle/Services/Nifm/IGeneralService.cs @@ -1,6 +1,11 @@ using Ryujinx.HLE.Logging; using Ryujinx.HLE.OsHle.Ipc; +using System; using System.Collections.Generic; +using System.Linq; +using System.Net; +using System.Net.Sockets; +using System.Net.NetworkInformation; namespace Ryujinx.HLE.OsHle.Services.Nifm { @@ -14,10 +19,13 @@ namespace Ryujinx.HLE.OsHle.Services.Nifm { m_Commands = new Dictionary() { - { 4, CreateRequest } + { 4, CreateRequest }, + { 12, GetCurrentIpAddress } }; } + public const int NoInternetConnection = 0x2586e; + //CreateRequest(i32) public long CreateRequest(ServiceCtx Context) { @@ -29,5 +37,22 @@ namespace Ryujinx.HLE.OsHle.Services.Nifm return 0; } + + public long GetCurrentIpAddress(ServiceCtx Context) + { + if (!NetworkInterface.GetIsNetworkAvailable()) + { + return NoInternetConnection; + } + + IPHostEntry Host = Dns.GetHostEntry(Dns.GetHostName()); + IPAddress Address = Host.AddressList.FirstOrDefault(A => A.AddressFamily == AddressFamily.InterNetwork); + + Context.ResponseData.Write(BitConverter.ToUInt32(Address.GetAddressBytes())); + + Context.Ns.Log.PrintInfo(LogClass.ServiceNifm, "Console's local IP is " + Address.ToString()); + + return 0; + } } -} \ No newline at end of file +} From 6a69001aa26fe6f3688fa98901ca40e641743d55 Mon Sep 17 00:00:00 2001 From: Ac_K Date: Thu, 19 Jul 2018 03:10:51 +0200 Subject: [PATCH 06/34] Update IGeneralService.cs Fixed little mistake on the debug string. --- Ryujinx.HLE/OsHle/Services/Nifm/IGeneralService.cs | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/Ryujinx.HLE/OsHle/Services/Nifm/IGeneralService.cs b/Ryujinx.HLE/OsHle/Services/Nifm/IGeneralService.cs index 66b1f1b610..83bb9f370c 100644 --- a/Ryujinx.HLE/OsHle/Services/Nifm/IGeneralService.cs +++ b/Ryujinx.HLE/OsHle/Services/Nifm/IGeneralService.cs @@ -50,7 +50,7 @@ namespace Ryujinx.HLE.OsHle.Services.Nifm Context.ResponseData.Write(BitConverter.ToUInt32(Address.GetAddressBytes())); - Context.Ns.Log.PrintInfo(LogClass.ServiceNifm, "Console's local IP is " + Address.ToString()); + Context.Ns.Log.PrintInfo(LogClass.ServiceNifm, $"Console's local IP is {Address.ToString()}"); return 0; } From 8b685b12f0b7a901139999dff17b24b049b9084b Mon Sep 17 00:00:00 2001 From: Ac_K Date: Thu, 19 Jul 2018 06:03:53 +0200 Subject: [PATCH 07/34] Implement SvcWaitForAddress 0x34 (#289) * Implement SvcWaitForAddress 0x34 Currently needed by Sonic Mania Plus * Fix mistake * read-decrement-write locked --- Ryujinx.HLE/OsHle/Handles/KThread.cs | 6 +- Ryujinx.HLE/OsHle/Kernel/AddressArbiter.cs | 112 +++++++++++++++++++++ Ryujinx.HLE/OsHle/Kernel/KernelErr.cs | 2 +- Ryujinx.HLE/OsHle/Kernel/SvcHandler.cs | 3 +- Ryujinx.HLE/OsHle/Kernel/SvcSystem.cs | 2 +- Ryujinx.HLE/OsHle/Kernel/SvcThreadSync.cs | 51 ++++++++++ 6 files changed, 171 insertions(+), 5 deletions(-) create mode 100644 Ryujinx.HLE/OsHle/Kernel/AddressArbiter.cs diff --git a/Ryujinx.HLE/OsHle/Handles/KThread.cs b/Ryujinx.HLE/OsHle/Handles/KThread.cs index 3db46f3d67..2b980d17b1 100644 --- a/Ryujinx.HLE/OsHle/Handles/KThread.cs +++ b/Ryujinx.HLE/OsHle/Handles/KThread.cs @@ -9,10 +9,12 @@ namespace Ryujinx.HLE.OsHle.Handles public int CoreMask { get; set; } - public long MutexAddress { get; set; } - public long CondVarAddress { get; set; } + public long MutexAddress { get; set; } + public long CondVarAddress { get; set; } + public long ArbiterWaitAddress { get; set; } public bool CondVarSignaled { get; set; } + public bool ArbiterSignaled { get; set; } private Process Process; diff --git a/Ryujinx.HLE/OsHle/Kernel/AddressArbiter.cs b/Ryujinx.HLE/OsHle/Kernel/AddressArbiter.cs new file mode 100644 index 0000000000..ce9ef0cd87 --- /dev/null +++ b/Ryujinx.HLE/OsHle/Kernel/AddressArbiter.cs @@ -0,0 +1,112 @@ +using ChocolArm64.Memory; +using ChocolArm64.State; +using Ryujinx.HLE.OsHle.Handles; + +using static Ryujinx.HLE.OsHle.ErrorCode; + +namespace Ryujinx.HLE.OsHle.Kernel +{ + static class AddressArbiter + { + static ulong WaitForAddress(Process Process, AThreadState ThreadState, long Address, ulong Timeout) + { + KThread CurrentThread = Process.GetThread(ThreadState.Tpidr); + + Process.Scheduler.SetReschedule(CurrentThread.ProcessorId); + + CurrentThread.ArbiterWaitAddress = Address; + CurrentThread.ArbiterSignaled = false; + + Process.Scheduler.EnterWait(CurrentThread, NsTimeConverter.GetTimeMs(Timeout)); + + if (!CurrentThread.ArbiterSignaled) + { + return MakeError(ErrorModule.Kernel, KernelErr.Timeout); + } + + return 0; + } + + public static ulong WaitForAddressIfLessThan(Process Process, + AThreadState ThreadState, + AMemory Memory, + long Address, + int Value, + ulong Timeout, + bool ShouldDecrement) + { + Memory.SetExclusive(ThreadState, Address); + + int CurrentValue = Memory.ReadInt32(Address); + + while (true) + { + if (Memory.TestExclusive(ThreadState, Address)) + { + if (CurrentValue < Value) + { + if (ShouldDecrement) + { + Memory.WriteInt32(Address, CurrentValue - 1); + } + + Memory.ClearExclusiveForStore(ThreadState); + } + else + { + Memory.ClearExclusiveForStore(ThreadState); + + return MakeError(ErrorModule.Kernel, KernelErr.InvalidState); + } + + break; + } + + Memory.SetExclusive(ThreadState, Address); + + CurrentValue = Memory.ReadInt32(Address); + } + + if (Timeout == 0) + { + return MakeError(ErrorModule.Kernel, KernelErr.Timeout); + } + + return WaitForAddress(Process, ThreadState, Address, Timeout); + } + + public static ulong WaitForAddressIfEqual(Process Process, + AThreadState ThreadState, + AMemory Memory, + long Address, + int Value, + ulong Timeout) + { + if (Memory.ReadInt32(Address) != Value) + { + return MakeError(ErrorModule.Kernel, KernelErr.InvalidState); + } + + if (Timeout == 0) + { + return MakeError(ErrorModule.Kernel, KernelErr.Timeout); + } + + return WaitForAddress(Process, ThreadState, Address, Timeout); + } + } + + enum ArbitrationType : int + { + WaitIfLessThan, + DecrementAndWaitIfLessThan, + WaitIfEqual + } + + enum SignalType : int + { + Signal, + IncrementAndSignalIfEqual, + ModifyByWaitingCountAndSignalIfEqual + } +} diff --git a/Ryujinx.HLE/OsHle/Kernel/KernelErr.cs b/Ryujinx.HLE/OsHle/Kernel/KernelErr.cs index ad4fdfb6bd..bbae532555 100644 --- a/Ryujinx.HLE/OsHle/Kernel/KernelErr.cs +++ b/Ryujinx.HLE/OsHle/Kernel/KernelErr.cs @@ -12,7 +12,7 @@ namespace Ryujinx.HLE.OsHle.Kernel public const int Timeout = 117; public const int Canceled = 118; public const int CountOutOfRange = 119; - public const int InvalidInfo = 120; + public const int InvalidEnumValue = 120; public const int InvalidThread = 122; public const int InvalidState = 125; } diff --git a/Ryujinx.HLE/OsHle/Kernel/SvcHandler.cs b/Ryujinx.HLE/OsHle/Kernel/SvcHandler.cs index e05073fda3..e816c44ec3 100644 --- a/Ryujinx.HLE/OsHle/Kernel/SvcHandler.cs +++ b/Ryujinx.HLE/OsHle/Kernel/SvcHandler.cs @@ -73,7 +73,8 @@ namespace Ryujinx.HLE.OsHle.Kernel { 0x2c, SvcMapPhysicalMemory }, { 0x2d, SvcUnmapPhysicalMemory }, { 0x32, SvcSetThreadActivity }, - { 0x33, SvcGetThreadContext3 } + { 0x33, SvcGetThreadContext3 }, + { 0x34, SvcWaitForAddress } }; this.Ns = Ns; diff --git a/Ryujinx.HLE/OsHle/Kernel/SvcSystem.cs b/Ryujinx.HLE/OsHle/Kernel/SvcSystem.cs index a32b2d86fd..08305522f5 100644 --- a/Ryujinx.HLE/OsHle/Kernel/SvcSystem.cs +++ b/Ryujinx.HLE/OsHle/Kernel/SvcSystem.cs @@ -294,7 +294,7 @@ namespace Ryujinx.HLE.OsHle.Kernel InfoType == 19 || InfoType == 20) { - ThreadState.X0 = MakeError(ErrorModule.Kernel, KernelErr.InvalidInfo); + ThreadState.X0 = MakeError(ErrorModule.Kernel, KernelErr.InvalidEnumValue); return; } diff --git a/Ryujinx.HLE/OsHle/Kernel/SvcThreadSync.cs b/Ryujinx.HLE/OsHle/Kernel/SvcThreadSync.cs index ec9c40e08b..9fc426176c 100644 --- a/Ryujinx.HLE/OsHle/Kernel/SvcThreadSync.cs +++ b/Ryujinx.HLE/OsHle/Kernel/SvcThreadSync.cs @@ -197,6 +197,57 @@ namespace Ryujinx.HLE.OsHle.Kernel Process.Scheduler.EnterWait(CurrThread); } + private void SvcWaitForAddress(AThreadState ThreadState) + { + long Address = (long)ThreadState.X0; + ArbitrationType Type = (ArbitrationType)ThreadState.X1; + int Value = (int)ThreadState.X2; + ulong Timeout = ThreadState.X3; + + Ns.Log.PrintDebug(LogClass.KernelSvc, + "Address = " + Address.ToString("x16") + ", " + + "ArbitrationType = " + Type .ToString() + ", " + + "Value = " + Value .ToString("x8") + ", " + + "Timeout = " + Timeout.ToString("x16")); + + if (IsPointingInsideKernel(Address)) + { + Ns.Log.PrintWarning(LogClass.KernelSvc, $"Invalid address 0x{Address:x16}!"); + + ThreadState.X0 = MakeError(ErrorModule.Kernel, KernelErr.InvalidAddress); + + return; + } + + if (IsWordAddressUnaligned(Address)) + { + Ns.Log.PrintWarning(LogClass.KernelSvc, $"Unaligned address 0x{Address:x16}!"); + + ThreadState.X0 = MakeError(ErrorModule.Kernel, KernelErr.InvalidAlignment); + + return; + } + + switch (Type) + { + case ArbitrationType.WaitIfLessThan: + ThreadState.X0 = AddressArbiter.WaitForAddressIfLessThan(Process, ThreadState, Memory, Address, Value, Timeout, false); + break; + + case ArbitrationType.DecrementAndWaitIfLessThan: + ThreadState.X0 = AddressArbiter.WaitForAddressIfLessThan(Process, ThreadState, Memory, Address, Value, Timeout, true); + break; + + case ArbitrationType.WaitIfEqual: + ThreadState.X0 = AddressArbiter.WaitForAddressIfEqual(Process, ThreadState, Memory, Address, Value, Timeout); + break; + + default: + ThreadState.X0 = MakeError(ErrorModule.Kernel, KernelErr.InvalidEnumValue); + break; + } + } + private void MutexUnlock(KThread CurrThread, long MutexAddress) { lock (Process.ThreadSyncLock) From 60f2198a1e8e61fe1cfb8da30a6afcd86a672a85 Mon Sep 17 00:00:00 2001 From: gdkchan Date: Thu, 19 Jul 2018 02:30:21 -0300 Subject: [PATCH 08/34] Support deswizzle of sparse tiled textures and some frame buffer fixes (#275) * Attempt to support deswizzle of sparse tiled textures * Use correct frame buffer and viewport sizes, started to clean up the copy engine * Correct texture width alignment * Use Scale/Translate registers to calculate viewport rect * Allow texture copy between frame buffers --- Ryujinx.Graphics/Gal/IGalFrameBuffer.cs | 19 +++ Ryujinx.Graphics/Gal/OpenGL/OGLFrameBuffer.cs | 82 +++++++++-- Ryujinx.HLE/Gpu/Engines/NvGpuEngine2d.cs | 130 ++++++++++++------ Ryujinx.HLE/Gpu/Engines/NvGpuEngine3d.cs | 26 +++- Ryujinx.HLE/Gpu/Engines/NvGpuEngine3dReg.cs | 14 +- Ryujinx.HLE/Gpu/Texture/TextureFactory.cs | 5 +- Ryujinx.HLE/Gpu/Texture/TextureHelper.cs | 8 +- Ryujinx.HLE/Gpu/Texture/TextureInfo.cs | 19 ++- Ryujinx.HLE/Gpu/Texture/TextureReader.cs | 18 +-- Ryujinx.HLE/Gpu/Texture/TextureWriter.cs | 28 +--- Ryujinx.HLE/OsHle/Services/Vi/NvFlinger.cs | 4 +- 11 files changed, 237 insertions(+), 116 deletions(-) diff --git a/Ryujinx.Graphics/Gal/IGalFrameBuffer.cs b/Ryujinx.Graphics/Gal/IGalFrameBuffer.cs index eaae0a492e..1f62bdb379 100644 --- a/Ryujinx.Graphics/Gal/IGalFrameBuffer.cs +++ b/Ryujinx.Graphics/Gal/IGalFrameBuffer.cs @@ -22,6 +22,25 @@ namespace Ryujinx.Graphics.Gal void Render(); + void Copy( + long SrcKey, + long DstKey, + int SrcX0, + int SrcY0, + int SrcX1, + int SrcY1, + int DstX0, + int DstY0, + int DstX1, + int DstY1); + void GetBufferData(long Key, Action Callback); + + void SetBufferData( + long Key, + int Width, + int Height, + GalTextureFormat Format, + byte[] Buffer); } } \ No newline at end of file diff --git a/Ryujinx.Graphics/Gal/OpenGL/OGLFrameBuffer.cs b/Ryujinx.Graphics/Gal/OpenGL/OGLFrameBuffer.cs index 305fa37d8f..cd52762c79 100644 --- a/Ryujinx.Graphics/Gal/OpenGL/OGLFrameBuffer.cs +++ b/Ryujinx.Graphics/Gal/OpenGL/OGLFrameBuffer.cs @@ -78,11 +78,6 @@ namespace Ryujinx.Graphics.Gal.OpenGL public void Create(long Key, int Width, int Height) { - //TODO: We should either use the original frame buffer size, - //or just remove the Width/Height arguments. - Width = Window.Width; - Height = Window.Height; - if (Fbs.TryGetValue(Key, out FrameBuffer Fb)) { if (Fb.Width != Width || @@ -125,8 +120,6 @@ namespace Ryujinx.Graphics.Gal.OpenGL GL.DrawBuffer(DrawBufferMode.ColorAttachment0); - GL.Viewport(0, 0, Width, Height); - Fbs.Add(Key, Fb); } @@ -230,7 +223,16 @@ namespace Ryujinx.Graphics.Gal.OpenGL { Viewport = new Rect(X, Y, Width, Height); - //TODO + SetViewport(Viewport); + } + + private void SetViewport(Rect Viewport) + { + GL.Viewport( + Viewport.X, + Viewport.Y, + Viewport.Width, + Viewport.Height); } public void Render() @@ -300,10 +302,38 @@ namespace Ryujinx.Graphics.Gal.OpenGL GL.Enable(EnableCap.Blend); } - //GL.Viewport(0, 0, 1280, 720); + SetViewport(Viewport); } } + public void Copy( + long SrcKey, + long DstKey, + int SrcX0, + int SrcY0, + int SrcX1, + int SrcY1, + int DstX0, + int DstY0, + int DstX1, + int DstY1) + { + if (Fbs.TryGetValue(SrcKey, out FrameBuffer SrcFb) && + Fbs.TryGetValue(DstKey, out FrameBuffer DstFb)) + { + GL.BindFramebuffer(FramebufferTarget.ReadFramebuffer, SrcFb.Handle); + GL.BindFramebuffer(FramebufferTarget.DrawFramebuffer, DstFb.Handle); + + GL.Clear(ClearBufferMask.ColorBufferBit); + + GL.BlitFramebuffer( + SrcX0, SrcY0, SrcX1, SrcY1, + DstX0, DstY0, DstX1, DstY1, + ClearBufferMask.ColorBufferBit, + BlitFramebufferFilter.Linear); + } +} + public void GetBufferData(long Key, Action Callback) { if (Fbs.TryGetValue(Key, out FrameBuffer Fb)) @@ -329,13 +359,35 @@ namespace Ryujinx.Graphics.Gal.OpenGL } } - private void SetViewport(Rect Viewport) + public void SetBufferData( + long Key, + int Width, + int Height, + GalTextureFormat Format, + byte[] Buffer) { - GL.Viewport( - Viewport.X, - Viewport.Y, - Viewport.Width, - Viewport.Height); + if (Fbs.TryGetValue(Key, out FrameBuffer Fb)) + { + GL.BindTexture(TextureTarget.Texture2D, Fb.TexHandle); + + const int Level = 0; + const int Border = 0; + + const PixelInternalFormat InternalFmt = PixelInternalFormat.Rgba; + + (PixelFormat GlFormat, PixelType Type) = OGLEnumConverter.GetTextureFormat(Format); + + GL.TexImage2D( + TextureTarget.Texture2D, + Level, + InternalFmt, + Width, + Height, + Border, + GlFormat, + Type, + Buffer); + } } private void EnsureInitialized() diff --git a/Ryujinx.HLE/Gpu/Engines/NvGpuEngine2d.cs b/Ryujinx.HLE/Gpu/Engines/NvGpuEngine2d.cs index f150b3f5e9..d2c5f1262b 100644 --- a/Ryujinx.HLE/Gpu/Engines/NvGpuEngine2d.cs +++ b/Ryujinx.HLE/Gpu/Engines/NvGpuEngine2d.cs @@ -1,6 +1,7 @@ using Ryujinx.Graphics.Gal; using Ryujinx.HLE.Gpu.Memory; using Ryujinx.HLE.Gpu.Texture; +using System; using System.Collections.Generic; namespace Ryujinx.HLE.Gpu.Engines @@ -64,6 +65,8 @@ namespace Ryujinx.HLE.Gpu.Engines bool SrcLinear = ReadRegister(NvGpuEngine2dReg.SrcLinear) != 0; int SrcWidth = ReadRegister(NvGpuEngine2dReg.SrcWidth); int SrcHeight = ReadRegister(NvGpuEngine2dReg.SrcHeight); + int SrcPitch = ReadRegister(NvGpuEngine2dReg.SrcPitch); + int SrcBlkDim = ReadRegister(NvGpuEngine2dReg.SrcBlockDimensions); bool DstLinear = ReadRegister(NvGpuEngine2dReg.DstLinear) != 0; int DstWidth = ReadRegister(NvGpuEngine2dReg.DstWidth); @@ -71,75 +74,114 @@ namespace Ryujinx.HLE.Gpu.Engines int DstPitch = ReadRegister(NvGpuEngine2dReg.DstPitch); int DstBlkDim = ReadRegister(NvGpuEngine2dReg.DstBlockDimensions); + TextureSwizzle SrcSwizzle = SrcLinear + ? TextureSwizzle.Pitch + : TextureSwizzle.BlockLinear; + TextureSwizzle DstSwizzle = DstLinear ? TextureSwizzle.Pitch : TextureSwizzle.BlockLinear; + int SrcBlockHeight = 1 << ((SrcBlkDim >> 4) & 0xf); int DstBlockHeight = 1 << ((DstBlkDim >> 4) & 0xf); - long Key = Vmm.GetPhysicalAddress(MakeInt64From2xInt32(NvGpuEngine2dReg.SrcAddress)); - long SrcAddress = MakeInt64From2xInt32(NvGpuEngine2dReg.SrcAddress); long DstAddress = MakeInt64From2xInt32(NvGpuEngine2dReg.DstAddress); - bool IsFbTexture = Gpu.Engine3d.IsFrameBufferPosition(Key); + long SrcKey = Vmm.GetPhysicalAddress(SrcAddress); + long DstKey = Vmm.GetPhysicalAddress(DstAddress); - if (IsFbTexture && DstLinear) + bool IsSrcFb = Gpu.Engine3d.IsFrameBufferPosition(SrcKey); + bool IsDstFb = Gpu.Engine3d.IsFrameBufferPosition(DstKey); + + TextureInfo SrcTexture() { - DstSwizzle = TextureSwizzle.BlockLinear; + return new TextureInfo( + SrcAddress, + SrcWidth, + SrcHeight, + SrcPitch, + SrcBlockHeight, 1, + SrcSwizzle, + GalTextureFormat.A8B8G8R8); } - TextureInfo DstTexture = new TextureInfo( - DstAddress, - DstWidth, - DstHeight, - DstBlockHeight, - DstBlockHeight, - DstSwizzle, - GalTextureFormat.A8B8G8R8); - - if (IsFbTexture) + TextureInfo DstTexture() { - //TODO: Change this when the correct frame buffer resolution is used. - //Currently, the frame buffer size is hardcoded to 1280x720. - SrcWidth = 1280; - SrcHeight = 720; + return new TextureInfo( + DstAddress, + DstWidth, + DstHeight, + DstPitch, + DstBlockHeight, 1, + DstSwizzle, + GalTextureFormat.A8B8G8R8); + } - Gpu.Renderer.FrameBuffer.GetBufferData(Key, (byte[] Buffer) => + //TODO: fb -> fb copies, tex -> fb copies, formats other than RGBA8, + //make it throw for unimpl stuff (like the copy mode)... + if (IsSrcFb && IsDstFb) + { + //Frame Buffer -> Frame Buffer copy. + Gpu.Renderer.FrameBuffer.Copy( + SrcKey, + DstKey, + 0, + 0, + SrcWidth, + SrcHeight, + 0, + 0, + DstWidth, + DstHeight); + } + if (IsSrcFb) + { + //Frame Buffer -> Texture copy. + Gpu.Renderer.FrameBuffer.GetBufferData(SrcKey, (byte[] Buffer) => { - CopyTexture( - Vmm, - DstTexture, - Buffer, - SrcWidth, - SrcHeight); + TextureInfo Src = SrcTexture(); + TextureInfo Dst = DstTexture(); + + if (Src.Width != Dst.Width || + Src.Height != Dst.Height) + { + throw new NotImplementedException("Texture resizing is not supported"); + } + + TextureWriter.Write(Vmm, Dst, Buffer); }); } + else if (IsDstFb) + { + //Texture -> Frame Buffer copy. + const GalTextureFormat Format = GalTextureFormat.A8B8G8R8; + + byte[] Buffer = TextureReader.Read(Vmm, SrcTexture()); + + Gpu.Renderer.FrameBuffer.SetBufferData( + DstKey, + DstWidth, + DstHeight, + Format, + Buffer); + } else { - long Size = SrcWidth * SrcHeight * 4; + //Texture -> Texture copy. + TextureInfo Src = SrcTexture(); + TextureInfo Dst = DstTexture(); - byte[] Buffer = Vmm.ReadBytes(SrcAddress, Size); + if (Src.Width != Dst.Width || + Src.Height != Dst.Height) + { + throw new NotImplementedException("Texture resizing is not supported"); + } - CopyTexture( - Vmm, - DstTexture, - Buffer, - SrcWidth, - SrcHeight); + TextureWriter.Write(Vmm, Dst, TextureReader.Read(Vmm, Src)); } } - private void CopyTexture( - NvGpuVmm Vmm, - TextureInfo Texture, - byte[] Buffer, - int Width, - int Height) - { - TextureWriter.Write(Vmm, Texture, Buffer, Width, Height); - } - private long MakeInt64From2xInt32(NvGpuEngine2dReg Reg) { return diff --git a/Ryujinx.HLE/Gpu/Engines/NvGpuEngine3d.cs b/Ryujinx.HLE/Gpu/Engines/NvGpuEngine3d.cs index 5c474ab0bc..dce25a5e94 100644 --- a/Ryujinx.HLE/Gpu/Engines/NvGpuEngine3d.cs +++ b/Ryujinx.HLE/Gpu/Engines/NvGpuEngine3d.cs @@ -132,10 +132,22 @@ namespace Ryujinx.HLE.Gpu.Engines int Width = ReadRegister(NvGpuEngine3dReg.FrameBufferNWidth + FbIndex * 0x10); int Height = ReadRegister(NvGpuEngine3dReg.FrameBufferNHeight + FbIndex * 0x10); - //Note: Using the Width/Height results seems to give incorrect results. - //Maybe the size of all frame buffers is hardcoded to screen size? This seems unlikely. - Gpu.Renderer.FrameBuffer.Create(Key, 1280, 720); + float TX = ReadRegisterFloat(NvGpuEngine3dReg.ViewportNTranslateX + FbIndex * 4); + float TY = ReadRegisterFloat(NvGpuEngine3dReg.ViewportNTranslateY + FbIndex * 4); + + float SX = ReadRegisterFloat(NvGpuEngine3dReg.ViewportNScaleX + FbIndex * 4); + float SY = ReadRegisterFloat(NvGpuEngine3dReg.ViewportNScaleY + FbIndex * 4); + + int VpX = (int)MathF.Max(0, TX - MathF.Abs(SX)); + int VpY = (int)MathF.Max(0, TY - MathF.Abs(SY)); + + int VpW = (int)(TX + MathF.Abs(SX)) - VpX; + int VpH = (int)(TY + MathF.Abs(SY)) - VpY; + + Gpu.Renderer.FrameBuffer.Create(Key, Width, Height); Gpu.Renderer.FrameBuffer.Bind(Key); + + Gpu.Renderer.FrameBuffer.SetViewport(VpX, VpY, VpW, VpH); } private long[] UploadShaders(NvGpuVmm Vmm) @@ -195,8 +207,8 @@ namespace Ryujinx.HLE.Gpu.Engines Gpu.Renderer.Shader.Bind(Key); } - float SignX = GetFlipSign(NvGpuEngine3dReg.ViewportScaleX); - float SignY = GetFlipSign(NvGpuEngine3dReg.ViewportScaleY); + float SignX = GetFlipSign(NvGpuEngine3dReg.ViewportNScaleX); + float SignY = GetFlipSign(NvGpuEngine3dReg.ViewportNScaleY); Gpu.Renderer.Shader.SetFlip(SignX, SignY); @@ -220,8 +232,8 @@ namespace Ryujinx.HLE.Gpu.Engines private void SetFrontFace() { - float SignX = GetFlipSign(NvGpuEngine3dReg.ViewportScaleX); - float SignY = GetFlipSign(NvGpuEngine3dReg.ViewportScaleY); + float SignX = GetFlipSign(NvGpuEngine3dReg.ViewportNScaleX); + float SignY = GetFlipSign(NvGpuEngine3dReg.ViewportNScaleY); GalFrontFace FrontFace = (GalFrontFace)ReadRegister(NvGpuEngine3dReg.FrontFace); diff --git a/Ryujinx.HLE/Gpu/Engines/NvGpuEngine3dReg.cs b/Ryujinx.HLE/Gpu/Engines/NvGpuEngine3dReg.cs index 3de2885ef2..e7dabe44a4 100644 --- a/Ryujinx.HLE/Gpu/Engines/NvGpuEngine3dReg.cs +++ b/Ryujinx.HLE/Gpu/Engines/NvGpuEngine3dReg.cs @@ -6,12 +6,14 @@ namespace Ryujinx.HLE.Gpu.Engines FrameBufferNWidth = 0x202, FrameBufferNHeight = 0x203, FrameBufferNFormat = 0x204, - ViewportScaleX = 0x280, - ViewportScaleY = 0x281, - ViewportScaleZ = 0x282, - ViewportTranslateX = 0x283, - ViewportTranslateY = 0x284, - ViewportTranslateZ = 0x285, + ViewportNScaleX = 0x280, + ViewportNScaleY = 0x281, + ViewportNScaleZ = 0x282, + ViewportNTranslateX = 0x283, + ViewportNTranslateY = 0x284, + ViewportNTranslateZ = 0x285, + ViewportNHoriz = 0x300, + ViewportNVert = 0x301, VertexArrayFirst = 0x35d, VertexArrayCount = 0x35e, ClearDepth = 0x364, diff --git a/Ryujinx.HLE/Gpu/Texture/TextureFactory.cs b/Ryujinx.HLE/Gpu/Texture/TextureFactory.cs index 9df0b60002..4db0b6f106 100644 --- a/Ryujinx.HLE/Gpu/Texture/TextureFactory.cs +++ b/Ryujinx.HLE/Gpu/Texture/TextureFactory.cs @@ -55,9 +55,11 @@ namespace Ryujinx.HLE.Gpu.Texture int Pitch = (Tic[3] & 0xffff) << 5; - int BlockHeightLog2 = (Tic[3] >> 3) & 7; + int BlockHeightLog2 = (Tic[3] >> 3) & 7; + int TileWidthLog2 = (Tic[3] >> 10) & 7; int BlockHeight = 1 << BlockHeightLog2; + int TileWidth = 1 << TileWidthLog2; int Width = (Tic[4] & 0xffff) + 1; int Height = (Tic[5] & 0xffff) + 1; @@ -68,6 +70,7 @@ namespace Ryujinx.HLE.Gpu.Texture Height, Pitch, BlockHeight, + TileWidth, Swizzle, Format); diff --git a/Ryujinx.HLE/Gpu/Texture/TextureHelper.cs b/Ryujinx.HLE/Gpu/Texture/TextureHelper.cs index de26c397d5..ecf2b6bf5c 100644 --- a/Ryujinx.HLE/Gpu/Texture/TextureHelper.cs +++ b/Ryujinx.HLE/Gpu/Texture/TextureHelper.cs @@ -7,8 +7,14 @@ namespace Ryujinx.HLE.Gpu.Texture { static class TextureHelper { - public static ISwizzle GetSwizzle(TextureInfo Texture, int Width, int Bpp) + public static ISwizzle GetSwizzle(TextureInfo Texture, int BlockWidth, int Bpp) { + int Width = (Texture.Width + (BlockWidth - 1)) / BlockWidth; + + int AlignMask = Texture.TileWidth * (64 / Bpp) - 1; + + Width = (Width + AlignMask) & ~AlignMask; + switch (Texture.Swizzle) { case TextureSwizzle._1dBuffer: diff --git a/Ryujinx.HLE/Gpu/Texture/TextureInfo.cs b/Ryujinx.HLE/Gpu/Texture/TextureInfo.cs index 31784bbc58..2a98ce00fd 100644 --- a/Ryujinx.HLE/Gpu/Texture/TextureInfo.cs +++ b/Ryujinx.HLE/Gpu/Texture/TextureInfo.cs @@ -11,6 +11,7 @@ namespace Ryujinx.HLE.Gpu.Texture public int Pitch { get; private set; } public int BlockHeight { get; private set; } + public int TileWidth { get; private set; } public TextureSwizzle Swizzle { get; private set; } @@ -29,6 +30,8 @@ namespace Ryujinx.HLE.Gpu.Texture BlockHeight = 16; + TileWidth = 1; + Swizzle = TextureSwizzle.BlockLinear; Format = GalTextureFormat.A8B8G8R8; @@ -40,16 +43,18 @@ namespace Ryujinx.HLE.Gpu.Texture int Height, int Pitch, int BlockHeight, + int TileWidth, TextureSwizzle Swizzle, GalTextureFormat Format) { - this.Position = Position; - this.Width = Width; - this.Height = Height; - this.Pitch = Pitch; - this.BlockHeight = BlockHeight; - this.Swizzle = Swizzle; - this.Format = Format; + this.Position = Position; + this.Width = Width; + this.Height = Height; + this.Pitch = Pitch; + this.BlockHeight = BlockHeight; + this.TileWidth = TileWidth; + this.Swizzle = Swizzle; + this.Format = Format; } } } \ No newline at end of file diff --git a/Ryujinx.HLE/Gpu/Texture/TextureReader.cs b/Ryujinx.HLE/Gpu/Texture/TextureReader.cs index 26129877ab..350ab825f4 100644 --- a/Ryujinx.HLE/Gpu/Texture/TextureReader.cs +++ b/Ryujinx.HLE/Gpu/Texture/TextureReader.cs @@ -56,7 +56,7 @@ namespace Ryujinx.HLE.Gpu.Texture byte[] Output = new byte[Width * Height]; - ISwizzle Swizzle = TextureHelper.GetSwizzle(Texture, Width, 1); + ISwizzle Swizzle = TextureHelper.GetSwizzle(Texture, 1, 1); (AMemory CpuMem, long Position) = TextureHelper.GetMemoryAndPosition( Memory, @@ -89,7 +89,7 @@ namespace Ryujinx.HLE.Gpu.Texture byte[] Output = new byte[Width * Height * 2]; - ISwizzle Swizzle = TextureHelper.GetSwizzle(Texture, Width, 2); + ISwizzle Swizzle = TextureHelper.GetSwizzle(Texture, 1, 2); (AMemory CpuMem, long Position) = TextureHelper.GetMemoryAndPosition( Memory, @@ -127,7 +127,7 @@ namespace Ryujinx.HLE.Gpu.Texture byte[] Output = new byte[Width * Height * 2]; - ISwizzle Swizzle = TextureHelper.GetSwizzle(Texture, Width, 2); + ISwizzle Swizzle = TextureHelper.GetSwizzle(Texture, 1, 2); (AMemory CpuMem, long Position) = TextureHelper.GetMemoryAndPosition( Memory, @@ -164,7 +164,7 @@ namespace Ryujinx.HLE.Gpu.Texture byte[] Output = new byte[Width * Height * 2]; - ISwizzle Swizzle = TextureHelper.GetSwizzle(Texture, Width, 2); + ISwizzle Swizzle = TextureHelper.GetSwizzle(Texture, 1, 2); (AMemory CpuMem, long Position) = TextureHelper.GetMemoryAndPosition( Memory, @@ -197,7 +197,7 @@ namespace Ryujinx.HLE.Gpu.Texture byte[] Output = new byte[Width * Height * 4]; - ISwizzle Swizzle = TextureHelper.GetSwizzle(Texture, Width, 4); + ISwizzle Swizzle = TextureHelper.GetSwizzle(Texture, 1, 4); (AMemory CpuMem, long Position) = TextureHelper.GetMemoryAndPosition( Memory, @@ -230,7 +230,7 @@ namespace Ryujinx.HLE.Gpu.Texture byte[] Output = new byte[Width * Height * 8]; - ISwizzle Swizzle = TextureHelper.GetSwizzle(Texture, Width, 8); + ISwizzle Swizzle = TextureHelper.GetSwizzle(Texture, 1, 8); (AMemory CpuMem, long Position) = TextureHelper.GetMemoryAndPosition( Memory, @@ -263,7 +263,7 @@ namespace Ryujinx.HLE.Gpu.Texture byte[] Output = new byte[Width * Height * 16]; - ISwizzle Swizzle = TextureHelper.GetSwizzle(Texture, Width, 16); + ISwizzle Swizzle = TextureHelper.GetSwizzle(Texture, 1, 16); (AMemory CpuMem, long Position) = TextureHelper.GetMemoryAndPosition( Memory, @@ -298,7 +298,7 @@ namespace Ryujinx.HLE.Gpu.Texture byte[] Output = new byte[Width * Height * 8]; - ISwizzle Swizzle = TextureHelper.GetSwizzle(Texture, Width, 8); + ISwizzle Swizzle = TextureHelper.GetSwizzle(Texture, 4, 8); (AMemory CpuMem, long Position) = TextureHelper.GetMemoryAndPosition( Memory, @@ -331,7 +331,7 @@ namespace Ryujinx.HLE.Gpu.Texture byte[] Output = new byte[Width * Height * 16]; - ISwizzle Swizzle = TextureHelper.GetSwizzle(Texture, Width, 16); + ISwizzle Swizzle = TextureHelper.GetSwizzle(Texture, BlockWidth, 16); (AMemory CpuMem, long Position) = TextureHelper.GetMemoryAndPosition( Memory, diff --git a/Ryujinx.HLE/Gpu/Texture/TextureWriter.cs b/Ryujinx.HLE/Gpu/Texture/TextureWriter.cs index b64302a5ab..a87d4545b0 100644 --- a/Ryujinx.HLE/Gpu/Texture/TextureWriter.cs +++ b/Ryujinx.HLE/Gpu/Texture/TextureWriter.cs @@ -6,29 +6,9 @@ namespace Ryujinx.HLE.Gpu.Texture { static class TextureWriter { - public static void Write( - IAMemory Memory, - TextureInfo Texture, - byte[] Data, - int Width, - int Height) + public unsafe static void Write(IAMemory Memory, TextureInfo Texture, byte[] Data) { - switch (Texture.Format) - { - case GalTextureFormat.A8B8G8R8: Write4Bpp(Memory, Texture, Data, Width, Height); break; - - default: throw new NotImplementedException(Texture.Format.ToString()); - } - } - - private unsafe static void Write4Bpp( - IAMemory Memory, - TextureInfo Texture, - byte[] Data, - int Width, - int Height) - { - ISwizzle Swizzle = TextureHelper.GetSwizzle(Texture, Width, 4); + ISwizzle Swizzle = TextureHelper.GetSwizzle(Texture, 1, 4); (AMemory CpuMem, long Position) = TextureHelper.GetMemoryAndPosition( Memory, @@ -38,8 +18,8 @@ namespace Ryujinx.HLE.Gpu.Texture { long InOffs = 0; - for (int Y = 0; Y < Height; Y++) - for (int X = 0; X < Width; X++) + for (int Y = 0; Y < Texture.Height; Y++) + for (int X = 0; X < Texture.Width; X++) { long Offset = (uint)Swizzle.GetSwizzleOffset(X, Y); diff --git a/Ryujinx.HLE/OsHle/Services/Vi/NvFlinger.cs b/Ryujinx.HLE/OsHle/Services/Vi/NvFlinger.cs index a3ed3ab51c..5307127be4 100644 --- a/Ryujinx.HLE/OsHle/Services/Vi/NvFlinger.cs +++ b/Ryujinx.HLE/OsHle/Services/Vi/NvFlinger.cs @@ -279,8 +279,8 @@ namespace Ryujinx.HLE.OsHle.Services.Android private void SendFrameBuffer(ServiceCtx Context, int Slot) { - int FbWidth = 1280; - int FbHeight = 720; + int FbWidth = BufferQueue[Slot].Data.Width; + int FbHeight = BufferQueue[Slot].Data.Height; int NvMapHandle = BitConverter.ToInt32(BufferQueue[Slot].Data.RawData, 0x4c); int BufferOffset = BitConverter.ToInt32(BufferQueue[Slot].Data.RawData, 0x50); From bdb6cbb43514f6d8eb96847a22b70709ae705827 Mon Sep 17 00:00:00 2001 From: Merry Date: Thu, 19 Jul 2018 06:32:37 +0100 Subject: [PATCH 09/34] AOpCodeTable: Speed up instruction decoding (#284) --- ChocolArm64/AOpCodeTable.cs | 71 +++++++++++++++++++++++++------------ 1 file changed, 49 insertions(+), 22 deletions(-) diff --git a/ChocolArm64/AOpCodeTable.cs b/ChocolArm64/AOpCodeTable.cs index c69de38f5a..689e039237 100644 --- a/ChocolArm64/AOpCodeTable.cs +++ b/ChocolArm64/AOpCodeTable.cs @@ -4,6 +4,7 @@ using ChocolArm64.Instruction; using ChocolArm64.Instruction32; using ChocolArm64.State; using System; +using System.Collections.Generic; namespace ChocolArm64 { @@ -438,18 +439,43 @@ namespace ChocolArm64 SetA64("0>001110<<0xxxxx001110xxxxxxxxxx", AInstEmit.Zip1_V, typeof(AOpCodeSimdReg)); SetA64("0>001110<<0xxxxx011110xxxxxxxxxx", AInstEmit.Zip2_V, typeof(AOpCodeSimdReg)); #endregion + +#region "Generate InstA64FastLookup Table (AArch64)" + var Tmp = new List[FastLookupSize]; + for (int i = 0; i < FastLookupSize; i++) + { + Tmp[i] = new List(); + } + + foreach (var Inst in AllInstA64) + { + int Mask = ToFastLookupIndex(Inst.Mask); + int Value = ToFastLookupIndex(Inst.Value); + + for (int i = 0; i < FastLookupSize; i++) + { + if ((i & Mask) == Value) + { + Tmp[i].Add(Inst); + } + } + } + + for (int i = 0; i < FastLookupSize; i++) + { + InstA64FastLookup[i] = Tmp[i].ToArray(); + } +#endregion } - private class TreeNode + private class InstInfo { public int Mask; public int Value; - public TreeNode Next; - public AInst Inst; - public TreeNode(int Mask, int Value, AInst Inst) + public InstInfo(int Mask, int Value, AInst Inst) { this.Mask = Mask; this.Value = Value; @@ -457,8 +483,11 @@ namespace ChocolArm64 } } - private static TreeNode InstHeadA32; - private static TreeNode InstHeadA64; + private static List AllInstA32 = new List(); + private static List AllInstA64 = new List(); + + private static int FastLookupSize = 0x1000; + private static InstInfo[][] InstA64FastLookup = new InstInfo[FastLookupSize][]; private static void SetA32(string Encoding, AInstInterpreter Interpreter, Type Type) { @@ -519,7 +548,7 @@ namespace ChocolArm64 if (XBits == 0) { - InsertTop(XMask, Value, Inst, Mode); + InsertInst(XMask, Value, Inst, Mode); return; } @@ -535,55 +564,53 @@ namespace ChocolArm64 if (Mask != Blacklisted) { - InsertTop(XMask, Value | Mask, Inst, Mode); + InsertInst(XMask, Value | Mask, Inst, Mode); } } } - private static void InsertTop( + private static void InsertInst( int XMask, int Value, AInst Inst, AExecutionMode Mode) { - TreeNode Node = new TreeNode(XMask, Value, Inst); + InstInfo Info = new InstInfo(XMask, Value, Inst); if (Mode == AExecutionMode.AArch64) { - Node.Next = InstHeadA64; - - InstHeadA64 = Node; + AllInstA64.Add(Info); } else { - Node.Next = InstHeadA32; - - InstHeadA32 = Node; + AllInstA32.Add(Info); } } public static AInst GetInstA32(int OpCode) { - return GetInst(InstHeadA32, OpCode); + return GetInstFromList(AllInstA32, OpCode); } public static AInst GetInstA64(int OpCode) { - return GetInst(InstHeadA64, OpCode); + return GetInstFromList(InstA64FastLookup[ToFastLookupIndex(OpCode)], OpCode); } - private static AInst GetInst(TreeNode Head, int OpCode) + private static int ToFastLookupIndex(int Value) { - TreeNode Node = Head; + return ((Value >> 10) & 0x00F) | ((Value >> 18) & 0xFF0); + } - do + private static AInst GetInstFromList(IEnumerable InstList, int OpCode) + { + foreach (var Node in InstList) { if ((OpCode & Node.Mask) == Node.Value) { return Node.Inst; } } - while ((Node = Node.Next) != null); return AInst.Undefined; } From cd203e98f2bed076798972da1d108bb64b1884ec Mon Sep 17 00:00:00 2001 From: ReinUsesLisp Date: Thu, 19 Jul 2018 02:33:27 -0300 Subject: [PATCH 10/34] Implement Geometry shaders (#280) * Implement Geometry shaders * Add EmitVertex() and EndPrimitive() * Read output geometry data from header * Stub Vmad * Add Iadd_I32 * Stub Mov_S (S2R) * Stub Isberd * Change vertex index to gpr39 in Abuf * Add stub messages for consistency * Do not print input block when there is no attributes * Use GL_ARB_enhanced_layouts * Skip geometry shaders when there's no GL_ARB_enhanced_layouts * Address feedback * Address feedback --- Ryujinx.Graphics/Gal/OpenGL/OGLExtension.cs | 43 +++++ Ryujinx.Graphics/Gal/OpenGL/OGLShader.cs | 22 ++- Ryujinx.Graphics/Gal/Shader/GlslDecl.cs | 7 +- Ryujinx.Graphics/Gal/Shader/GlslDecompiler.cs | 156 ++++++++++++++++-- .../Gal/Shader/ShaderDecodeAlu.cs | 37 +++++ .../Gal/Shader/ShaderDecodeHelper.cs | 8 +- .../Gal/Shader/ShaderDecodeMem.cs | 3 + .../Gal/Shader/ShaderDecodeMove.cs | 20 +++ .../Gal/Shader/ShaderDecodeSpecial.cs | 29 ++++ Ryujinx.Graphics/Gal/Shader/ShaderDecoder.cs | 12 +- Ryujinx.Graphics/Gal/Shader/ShaderHeader.cs | 73 ++++++++ Ryujinx.Graphics/Gal/Shader/ShaderIrInst.cs | 5 +- .../Gal/Shader/ShaderIrOperAbuf.cs | 11 +- .../Gal/Shader/ShaderOpCodeTable.cs | 4 + Ryujinx.Graphics/Gal/ShaderDumper.cs | 44 ++++- 15 files changed, 426 insertions(+), 48 deletions(-) create mode 100644 Ryujinx.Graphics/Gal/OpenGL/OGLExtension.cs create mode 100644 Ryujinx.Graphics/Gal/Shader/ShaderDecodeSpecial.cs create mode 100644 Ryujinx.Graphics/Gal/Shader/ShaderHeader.cs diff --git a/Ryujinx.Graphics/Gal/OpenGL/OGLExtension.cs b/Ryujinx.Graphics/Gal/OpenGL/OGLExtension.cs new file mode 100644 index 0000000000..69fce6d31d --- /dev/null +++ b/Ryujinx.Graphics/Gal/OpenGL/OGLExtension.cs @@ -0,0 +1,43 @@ +using OpenTK.Graphics.OpenGL; + +namespace Ryujinx.Graphics.Gal.OpenGL +{ + static class OGLExtension + { + private static bool Initialized = false; + + private static bool EnhancedLayouts; + + public static bool HasEnhancedLayouts() + { + EnsureInitialized(); + + return EnhancedLayouts; + } + + private static void EnsureInitialized() + { + if (Initialized) + { + return; + } + + EnhancedLayouts = HasExtension("GL_ARB_enhanced_layouts"); + } + + private static bool HasExtension(string Name) + { + int NumExtensions = GL.GetInteger(GetPName.NumExtensions); + + for (int Extension = 0; Extension < NumExtensions; Extension++) + { + if (GL.GetString(StringNameIndexed.Extensions, Extension) == Name) + { + return true; + } + } + + return false; + } + } +} \ No newline at end of file diff --git a/Ryujinx.Graphics/Gal/OpenGL/OGLShader.cs b/Ryujinx.Graphics/Gal/OpenGL/OGLShader.cs index ad71775502..fe98aa0911 100644 --- a/Ryujinx.Graphics/Gal/OpenGL/OGLShader.cs +++ b/Ryujinx.Graphics/Gal/OpenGL/OGLShader.cs @@ -118,20 +118,20 @@ namespace Ryujinx.Graphics.Gal.OpenGL if (IsDualVp) { - ShaderDumper.Dump(Memory, Position + 0x50, Type, "a"); - ShaderDumper.Dump(Memory, PositionB + 0x50, Type, "b"); + ShaderDumper.Dump(Memory, Position, Type, "a"); + ShaderDumper.Dump(Memory, PositionB, Type, "b"); Program = Decompiler.Decompile( Memory, - Position + 0x50, - PositionB + 0x50, + Position, + PositionB, Type); } else { - ShaderDumper.Dump(Memory, Position + 0x50, Type); + ShaderDumper.Dump(Memory, Position, Type); - Program = Decompiler.Decompile(Memory, Position + 0x50, Type); + Program = Decompiler.Decompile(Memory, Position, Type); } return new ShaderStage( @@ -198,6 +198,16 @@ namespace Ryujinx.Graphics.Gal.OpenGL private void Bind(ShaderStage Stage) { + if (Stage.Type == GalShaderType.Geometry) + { + //Enhanced layouts are required for Geometry shaders + //skip this stage if current driver has no ARB_enhanced_layouts + if (!OGLExtension.HasEnhancedLayouts()) + { + return; + } + } + switch (Stage.Type) { case GalShaderType.Vertex: Current.Vertex = Stage; break; diff --git a/Ryujinx.Graphics/Gal/Shader/GlslDecl.cs b/Ryujinx.Graphics/Gal/Shader/GlslDecl.cs index d3284f9f55..7688545c02 100644 --- a/Ryujinx.Graphics/Gal/Shader/GlslDecl.cs +++ b/Ryujinx.Graphics/Gal/Shader/GlslDecl.cs @@ -4,13 +4,13 @@ namespace Ryujinx.Graphics.Gal.Shader { class GlslDecl { + public const int LayerAttr = 0x064; public const int TessCoordAttrX = 0x2f0; public const int TessCoordAttrY = 0x2f4; public const int TessCoordAttrZ = 0x2f8; public const int InstanceIdAttr = 0x2f8; public const int VertexIdAttr = 0x2fc; public const int FaceAttr = 0x3fc; - public const int GlPositionWAttr = 0x7c; public const int MaxUboSize = 1024; @@ -210,7 +210,8 @@ namespace Ryujinx.Graphics.Gal.Shader //This is a built-in input variable. if (Abuf.Offs == VertexIdAttr || Abuf.Offs == InstanceIdAttr || - Abuf.Offs == FaceAttr) + Abuf.Offs == FaceAttr || + Abuf.Offs == LayerAttr) { break; } @@ -254,6 +255,8 @@ namespace Ryujinx.Graphics.Gal.Shader m_Attributes.Add(Index, DeclInfo); } + + Traverse(Abuf, Abuf.Vertex); break; } diff --git a/Ryujinx.Graphics/Gal/Shader/GlslDecompiler.cs b/Ryujinx.Graphics/Gal/Shader/GlslDecompiler.cs index 575fb72f9a..a338f40413 100644 --- a/Ryujinx.Graphics/Gal/Shader/GlslDecompiler.cs +++ b/Ryujinx.Graphics/Gal/Shader/GlslDecompiler.cs @@ -21,10 +21,14 @@ namespace Ryujinx.Graphics.Gal.Shader private const string IdentationStr = " "; + private const int MaxVertexInput = 3; + private static string[] ElemTypes = new string[] { "float", "vec2", "vec3", "vec4" }; private GlslDecl Decl; + private ShaderHeader Header, HeaderB; + private ShaderIrBlock[] Blocks, BlocksB; private StringBuilder SB; @@ -50,6 +54,7 @@ namespace Ryujinx.Graphics.Gal.Shader { ShaderIrInst.Cle, GetCleExpr }, { ShaderIrInst.Clt, GetCltExpr }, { ShaderIrInst.Cne, GetCneExpr }, + { ShaderIrInst.Cut, GetCutExpr }, { ShaderIrInst.Exit, GetExitExpr }, { ShaderIrInst.Fabs, GetAbsExpr }, { ShaderIrInst.Fadd, GetAddExpr }, @@ -110,6 +115,9 @@ namespace Ryujinx.Graphics.Gal.Shader long VpBPosition, GalShaderType ShaderType) { + Header = new ShaderHeader(Memory, VpAPosition); + HeaderB = new ShaderHeader(Memory, VpBPosition); + Blocks = ShaderDecoder.Decode(Memory, VpAPosition); BlocksB = ShaderDecoder.Decode(Memory, VpBPosition); @@ -123,6 +131,9 @@ namespace Ryujinx.Graphics.Gal.Shader public GlslProgram Decompile(IGalMemory Memory, long Position, GalShaderType ShaderType) { + Header = new ShaderHeader(Memory, Position); + HeaderB = null; + Blocks = ShaderDecoder.Decode(Memory, Position); BlocksB = null; @@ -137,6 +148,7 @@ namespace Ryujinx.Graphics.Gal.Shader SB.AppendLine("#version 410 core"); + PrintDeclHeader(); PrintDeclTextures(); PrintDeclUniforms(); PrintDeclAttributes(); @@ -170,6 +182,37 @@ namespace Ryujinx.Graphics.Gal.Shader Decl.Uniforms.Values); } + private void PrintDeclHeader() + { + if (Decl.ShaderType == GalShaderType.Geometry) + { + int MaxVertices = Header.MaxOutputVertexCount; + + string OutputTopology; + + switch (Header.OutputTopology) + { + case ShaderHeader.PointList: OutputTopology = "points"; break; + case ShaderHeader.LineStrip: OutputTopology = "line_strip"; break; + case ShaderHeader.TriangleStrip: OutputTopology = "triangle_strip"; break; + + default: throw new InvalidOperationException(); + } + + SB.AppendLine("#extension GL_ARB_enhanced_layouts : require"); + + SB.AppendLine(); + + SB.AppendLine("// Stubbed. Maxwell geometry shaders don't inform input geometry type"); + + SB.AppendLine("layout(triangles) in;" + Environment.NewLine); + + SB.AppendLine($"layout({OutputTopology}, max_vertices = {MaxVertices}) out;"); + + SB.AppendLine(); + } + } + private void PrintDeclTextures() { PrintDecls(Decl.Textures, "uniform sampler2D"); @@ -201,7 +244,9 @@ namespace Ryujinx.Graphics.Gal.Shader private void PrintDeclAttributes() { - PrintDecls(Decl.Attributes); + string GeometryArray = (Decl.ShaderType == GalShaderType.Geometry) ? "[" + MaxVertexInput + "]" : ""; + + PrintDecls(Decl.Attributes, Suffix: GeometryArray); } private void PrintDeclInAttributes() @@ -211,7 +256,27 @@ namespace Ryujinx.Graphics.Gal.Shader SB.AppendLine("layout (location = " + GlslDecl.PositionOutAttrLocation + ") in vec4 " + GlslDecl.PositionOutAttrName + ";"); } - PrintDeclAttributes(Decl.InAttributes.Values, "in"); + if (Decl.ShaderType == GalShaderType.Geometry) + { + if (Decl.InAttributes.Count > 0) + { + SB.AppendLine("in Vertex {"); + + foreach (ShaderDeclInfo DeclInfo in Decl.InAttributes.Values.OrderBy(DeclKeySelector)) + { + if (DeclInfo.Index >= 0) + { + SB.AppendLine(IdentationStr + "layout (location = " + DeclInfo.Index + ") " + GetDecl(DeclInfo) + "; "); + } + } + + SB.AppendLine("} block_in[];" + Environment.NewLine); + } + } + else + { + PrintDeclAttributes(Decl.InAttributes.Values, "in"); + } } private void PrintDeclOutAttributes() @@ -254,7 +319,7 @@ namespace Ryujinx.Graphics.Gal.Shader PrintDecls(Decl.Preds, "bool"); } - private void PrintDecls(IReadOnlyDictionary Dict, string CustomType = null) + private void PrintDecls(IReadOnlyDictionary Dict, string CustomType = null, string Suffix = "") { foreach (ShaderDeclInfo DeclInfo in Dict.Values.OrderBy(DeclKeySelector)) { @@ -262,15 +327,15 @@ namespace Ryujinx.Graphics.Gal.Shader if (CustomType != null) { - Name = CustomType + " " + DeclInfo.Name + ";"; + Name = CustomType + " " + DeclInfo.Name + Suffix + ";"; } else if (DeclInfo.Name == GlslDecl.FragmentOutputName) { - Name = "layout (location = 0) out " + GetDecl(DeclInfo) + ";" + Environment.NewLine; + Name = "layout (location = 0) out " + GetDecl(DeclInfo) + Suffix + ";" + Environment.NewLine; } else { - Name = GetDecl(DeclInfo) + ";"; + Name = GetDecl(DeclInfo) + Suffix + ";"; } SB.AppendLine(Name); @@ -307,7 +372,21 @@ namespace Ryujinx.Graphics.Gal.Shader string Swizzle = ".xyzw".Substring(0, DeclInfo.Size + 1); - SB.AppendLine(IdentationStr + Attr.Name + Swizzle + " = " + DeclInfo.Name + ";"); + if (Decl.ShaderType == GalShaderType.Geometry) + { + for (int Vertex = 0; Vertex < MaxVertexInput; Vertex++) + { + string Dst = Attr.Name + "[" + Vertex + "]" + Swizzle; + + string Src = "block_in[" + Vertex + "]." + DeclInfo.Name; + + SB.AppendLine(IdentationStr + Dst + " = " + Src + ";"); + } + } + else + { + SB.AppendLine(IdentationStr + Attr.Name + Swizzle + " = " + DeclInfo.Name + ";"); + } } if (BlocksB != null) @@ -320,6 +399,16 @@ namespace Ryujinx.Graphics.Gal.Shader SB.AppendLine(IdentationStr + GlslDecl.ProgramName + "();"); } + if (Decl.ShaderType != GalShaderType.Geometry) + { + PrintAttrToOutput(); + } + + SB.AppendLine("}"); + } + + private void PrintAttrToOutput(string Identation = IdentationStr) + { foreach (KeyValuePair KV in Decl.OutAttributes) { if (!Decl.Attributes.TryGetValue(KV.Key, out ShaderDeclInfo Attr)) @@ -331,21 +420,26 @@ namespace Ryujinx.Graphics.Gal.Shader string Swizzle = ".xyzw".Substring(0, DeclInfo.Size + 1); - SB.AppendLine(IdentationStr + DeclInfo.Name + " = " + Attr.Name + Swizzle + ";"); + string Name = Attr.Name; + + if (Decl.ShaderType == GalShaderType.Geometry) + { + Name += "[0]"; + } + + SB.AppendLine(Identation + DeclInfo.Name + " = " + Name + Swizzle + ";"); } if (Decl.ShaderType == GalShaderType.Vertex) { - SB.AppendLine(IdentationStr + "gl_Position.xy *= " + GlslDecl.FlipUniformName + ";"); + SB.AppendLine(Identation + "gl_Position.xy *= " + GlslDecl.FlipUniformName + ";"); } if (Decl.ShaderType != GalShaderType.Fragment) { - SB.AppendLine(IdentationStr + GlslDecl.PositionOutAttrName + " = gl_Position;"); - SB.AppendLine(IdentationStr + GlslDecl.PositionOutAttrName + ".w = 1;"); + SB.AppendLine(Identation + GlslDecl.PositionOutAttrName + " = gl_Position;"); + SB.AppendLine(Identation + GlslDecl.PositionOutAttrName + ".w = 1;"); } - - SB.AppendLine("}"); } private void PrintBlockScope( @@ -484,11 +578,17 @@ namespace Ryujinx.Graphics.Gal.Shader { SB.AppendLine(Identation + "continue;"); } - - continue; } + else if (Op.Inst == ShaderIrInst.Emit) + { + PrintAttrToOutput(Identation); - SB.AppendLine(Identation + GetSrcExpr(Op, true) + ";"); + SB.AppendLine(Identation + "EmitVertex();"); + } + else + { + SB.AppendLine(Identation + GetSrcExpr(Op, true) + ";"); + } } else if (Node is ShaderIrCmnt Cmnt) { @@ -634,6 +734,14 @@ namespace Ryujinx.Graphics.Gal.Shader private string GetOutAbufName(ShaderIrOperAbuf Abuf) { + if (Decl.ShaderType == GalShaderType.Geometry) + { + switch (Abuf.Offs) + { + case GlslDecl.LayerAttr: return "gl_Layer"; + } + } + return GetAttrTempName(Abuf); } @@ -692,7 +800,16 @@ namespace Ryujinx.Graphics.Gal.Shader throw new InvalidOperationException(); } - return DeclInfo.Name + Swizzle; + if (Decl.ShaderType == GalShaderType.Geometry) + { + string Vertex = "floatBitsToInt(" + GetSrcExpr(Abuf.Vertex) + ")"; + + return DeclInfo.Name + "[" + Vertex + "]" + Swizzle; + } + else + { + return DeclInfo.Name + Swizzle; + } } private string GetName(ShaderIrOperGpr Gpr) @@ -805,6 +922,8 @@ namespace Ryujinx.Graphics.Gal.Shader private string GetCneExpr(ShaderIrOp Op) => GetBinaryExpr(Op, "!="); + private string GetCutExpr(ShaderIrOp Op) => "EndPrimitive()"; + private string GetCneuExpr(ShaderIrOp Op) => GetBinaryExprWithNaN(Op, "!="); private string GetCnumExpr(ShaderIrOp Op) => GetUnaryCall(Op, "!isnan"); @@ -1104,8 +1223,9 @@ namespace Ryujinx.Graphics.Gal.Shader switch (Node) { case ShaderIrOperAbuf Abuf: - return Abuf.Offs == GlslDecl.VertexIdAttr || + return Abuf.Offs == GlslDecl.LayerAttr || Abuf.Offs == GlslDecl.InstanceIdAttr || + Abuf.Offs == GlslDecl.VertexIdAttr || Abuf.Offs == GlslDecl.FaceAttr ? OperType.I32 : OperType.F32; diff --git a/Ryujinx.Graphics/Gal/Shader/ShaderDecodeAlu.cs b/Ryujinx.Graphics/Gal/Shader/ShaderDecodeAlu.cs index a44073513c..00f072f192 100644 --- a/Ryujinx.Graphics/Gal/Shader/ShaderDecodeAlu.cs +++ b/Ryujinx.Graphics/Gal/Shader/ShaderDecodeAlu.cs @@ -442,6 +442,41 @@ namespace Ryujinx.Graphics.Gal.Shader return Signed ? ShaderIrInst.Asr : ShaderIrInst.Lsr; } + public static void Vmad(ShaderIrBlock Block, long OpCode) + { + ShaderIrNode OperA = GetOperGpr8(OpCode); + + ShaderIrNode OperB; + + if (((OpCode >> 50) & 1) != 0) + { + OperB = GetOperGpr20(OpCode); + } + else + { + OperB = GetOperImm19_20(OpCode); + } + + ShaderIrOperGpr OperC = GetOperGpr39(OpCode); + + ShaderIrNode Tmp = new ShaderIrOp(ShaderIrInst.Mul, OperA, OperB); + + ShaderIrNode Final = new ShaderIrOp(ShaderIrInst.Add, Tmp, OperC); + + int Shr = (int)((OpCode >> 51) & 3); + + if (Shr != 0) + { + int Shift = (Shr == 2) ? 15 : 7; + + Final = new ShaderIrOp(ShaderIrInst.Lsr, Final, new ShaderIrOperImm(Shift)); + } + + Block.AddNode(new ShaderIrCmnt("Stubbed. Instruction is reduced to a * b + c")); + + Block.AddNode(GetPredNode(new ShaderIrAsg(GetOperGpr0(OpCode), Final), OpCode)); + } + public static void Xmad_CR(ShaderIrBlock Block, long OpCode) { EmitXmad(Block, OpCode, ShaderOper.CR); @@ -819,6 +854,8 @@ namespace Ryujinx.Graphics.Gal.Shader OperA = GetAluFabsFneg(OperA, AbsA, NegA); + Block.AddNode(new ShaderIrCmnt("Stubbed.")); + Block.AddNode(GetPredNode(new ShaderIrAsg(GetOperGpr0(OpCode), OperA), OpCode)); } diff --git a/Ryujinx.Graphics/Gal/Shader/ShaderDecodeHelper.cs b/Ryujinx.Graphics/Gal/Shader/ShaderDecodeHelper.cs index 1f1b158ef8..7d7b2f6c6d 100644 --- a/Ryujinx.Graphics/Gal/Shader/ShaderDecodeHelper.cs +++ b/Ryujinx.Graphics/Gal/Shader/ShaderDecodeHelper.cs @@ -7,14 +7,15 @@ namespace Ryujinx.Graphics.Gal.Shader public static ShaderIrOperAbuf[] GetOperAbuf20(long OpCode) { int Abuf = (int)(OpCode >> 20) & 0x3ff; - int Reg = (int)(OpCode >> 39) & 0xff; int Size = (int)(OpCode >> 47) & 3; + ShaderIrOperGpr Vertex = GetOperGpr39(OpCode); + ShaderIrOperAbuf[] Opers = new ShaderIrOperAbuf[Size + 1]; for (int Index = 0; Index <= Size; Index++) { - Opers[Index] = new ShaderIrOperAbuf(Abuf + Index * 4, Reg); + Opers[Index] = new ShaderIrOperAbuf(Abuf + Index * 4, Vertex); } return Opers; @@ -23,9 +24,8 @@ namespace Ryujinx.Graphics.Gal.Shader public static ShaderIrOperAbuf GetOperAbuf28(long OpCode) { int Abuf = (int)(OpCode >> 28) & 0x3ff; - int Reg = (int)(OpCode >> 39) & 0xff; - return new ShaderIrOperAbuf(Abuf, Reg); + return new ShaderIrOperAbuf(Abuf, GetOperGpr39(OpCode)); } public static ShaderIrOperCbuf GetOperCbuf34(long OpCode) diff --git a/Ryujinx.Graphics/Gal/Shader/ShaderDecodeMem.cs b/Ryujinx.Graphics/Gal/Shader/ShaderDecodeMem.cs index 083b0c63a1..aea7e744dc 100644 --- a/Ryujinx.Graphics/Gal/Shader/ShaderDecodeMem.cs +++ b/Ryujinx.Graphics/Gal/Shader/ShaderDecodeMem.cs @@ -35,6 +35,9 @@ namespace Ryujinx.Graphics.Gal.Shader { ShaderIrNode[] Opers = GetOperAbuf20(OpCode); + //Used by GS + ShaderIrOperGpr Vertex = GetOperGpr39(OpCode); + int Index = 0; foreach (ShaderIrNode OperA in Opers) diff --git a/Ryujinx.Graphics/Gal/Shader/ShaderDecodeMove.cs b/Ryujinx.Graphics/Gal/Shader/ShaderDecodeMove.cs index 4c9e59cf0c..c6b71fb01a 100644 --- a/Ryujinx.Graphics/Gal/Shader/ShaderDecodeMove.cs +++ b/Ryujinx.Graphics/Gal/Shader/ShaderDecodeMove.cs @@ -85,6 +85,16 @@ namespace Ryujinx.Graphics.Gal.Shader EmitI2i(Block, OpCode, ShaderOper.RR); } + public static void Isberd(ShaderIrBlock Block, long OpCode) + { + //This instruction seems to be used to translate from an address to a vertex index in a GS + //Stub it as such + + Block.AddNode(new ShaderIrCmnt("Stubbed.")); + + Block.AddNode(GetPredNode(new ShaderIrAsg(GetOperGpr0(OpCode), GetOperGpr8(OpCode)), OpCode)); + } + public static void Mov_C(ShaderIrBlock Block, long OpCode) { ShaderIrOperCbuf Cbuf = GetOperCbuf34(OpCode); @@ -128,6 +138,16 @@ namespace Ryujinx.Graphics.Gal.Shader EmitSel(Block, OpCode, ShaderOper.RR); } + public static void Mov_S(ShaderIrBlock Block, long OpCode) + { + Block.AddNode(new ShaderIrCmnt("Stubbed.")); + + //Zero is used as a special number to get a valid "0 * 0 + VertexIndex" in a GS + ShaderIrNode Source = new ShaderIrOperImm(0); + + Block.AddNode(GetPredNode(new ShaderIrAsg(GetOperGpr0(OpCode), Source), OpCode)); + } + private static void EmitF2f(ShaderIrBlock Block, long OpCode, ShaderOper Oper) { bool NegA = ((OpCode >> 45) & 1) != 0; diff --git a/Ryujinx.Graphics/Gal/Shader/ShaderDecodeSpecial.cs b/Ryujinx.Graphics/Gal/Shader/ShaderDecodeSpecial.cs new file mode 100644 index 0000000000..591631ff99 --- /dev/null +++ b/Ryujinx.Graphics/Gal/Shader/ShaderDecodeSpecial.cs @@ -0,0 +1,29 @@ +using System; + +using static Ryujinx.Graphics.Gal.Shader.ShaderDecodeHelper; + +namespace Ryujinx.Graphics.Gal.Shader +{ + static partial class ShaderDecode + { + public static void Out_R(ShaderIrBlock Block, long OpCode) + { + //TODO: Those registers have to be used for something + ShaderIrOperGpr Gpr0 = GetOperGpr0(OpCode); + ShaderIrOperGpr Gpr8 = GetOperGpr8(OpCode); + ShaderIrOperGpr Gpr20 = GetOperGpr20(OpCode); + + int Type = (int)((OpCode >> 39) & 3); + + if ((Type & 1) != 0) + { + Block.AddNode(GetPredNode(new ShaderIrOp(ShaderIrInst.Emit), OpCode)); + } + + if ((Type & 2) != 0) + { + Block.AddNode(GetPredNode(new ShaderIrOp(ShaderIrInst.Cut), OpCode)); + } + } + } +} \ No newline at end of file diff --git a/Ryujinx.Graphics/Gal/Shader/ShaderDecoder.cs b/Ryujinx.Graphics/Gal/Shader/ShaderDecoder.cs index 85522ff950..98f371b573 100644 --- a/Ryujinx.Graphics/Gal/Shader/ShaderDecoder.cs +++ b/Ryujinx.Graphics/Gal/Shader/ShaderDecoder.cs @@ -4,6 +4,8 @@ namespace Ryujinx.Graphics.Gal.Shader { static class ShaderDecoder { + private const long HeaderSize = 0x50; + private const bool AddDbgComments = true; public static ShaderIrBlock[] Decode(IGalMemory Memory, long Start) @@ -32,13 +34,13 @@ namespace Ryujinx.Graphics.Gal.Shader return Output; } - ShaderIrBlock Entry = Enqueue(Start); + ShaderIrBlock Entry = Enqueue(Start + HeaderSize); while (Blocks.Count > 0) { ShaderIrBlock Current = Blocks.Dequeue(); - FillBlock(Memory, Current); + FillBlock(Memory, Current, Start + HeaderSize); //Set child blocks. "Branch" is the block the branch instruction //points to (when taken), "Next" is the block at the next address, @@ -122,14 +124,14 @@ namespace Ryujinx.Graphics.Gal.Shader return Graph; } - private static void FillBlock(IGalMemory Memory, ShaderIrBlock Block) + private static void FillBlock(IGalMemory Memory, ShaderIrBlock Block, long Beginning) { long Position = Block.Position; do { //Ignore scheduling instructions, which are written every 32 bytes. - if ((Position & 0x1f) == 0) + if (((Position - Beginning) & 0x1f) == 0) { Position += 8; @@ -147,7 +149,7 @@ namespace Ryujinx.Graphics.Gal.Shader if (AddDbgComments) { - string DbgOpCode = $"0x{(Position - 8):x16}: 0x{OpCode:x16} "; + string DbgOpCode = $"0x{(Position - Beginning - 8):x16}: 0x{OpCode:x16} "; DbgOpCode += (Decode?.Method.Name ?? "???"); diff --git a/Ryujinx.Graphics/Gal/Shader/ShaderHeader.cs b/Ryujinx.Graphics/Gal/Shader/ShaderHeader.cs new file mode 100644 index 0000000000..8e5057ed9e --- /dev/null +++ b/Ryujinx.Graphics/Gal/Shader/ShaderHeader.cs @@ -0,0 +1,73 @@ +namespace Ryujinx.Graphics.Gal.Shader +{ + class ShaderHeader + { + public const int PointList = 1; + public const int LineStrip = 6; + public const int TriangleStrip = 7; + + public int SphType { get; private set; } + public int Version { get; private set; } + public int ShaderType { get; private set; } + public bool MrtEnable { get; private set; } + public bool KillsPixels { get; private set; } + public bool DoesGlobalStore { get; private set; } + public int SassVersion { get; private set; } + public bool DoesLoadOrStore { get; private set; } + public bool DoesFp64 { get; private set; } + public int StreamOutMask { get; private set; } + + public int ShaderLocalMemoryLowSize { get; private set; } + public int PerPatchAttributeCount { get; private set; } + + public int ShaderLocalMemoryHighSize { get; private set; } + public int ThreadsPerInputPrimitive { get; private set; } + + public int ShaderLocalMemoryCrsSize { get; private set; } + public int OutputTopology { get; private set; } + + public int MaxOutputVertexCount { get; private set; } + public int StoreReqStart { get; private set; } + public int StoreReqEnd { get; private set; } + + public ShaderHeader(IGalMemory Memory, long Position) + { + uint CommonWord0 = (uint)Memory.ReadInt32(Position + 0); + uint CommonWord1 = (uint)Memory.ReadInt32(Position + 4); + uint CommonWord2 = (uint)Memory.ReadInt32(Position + 8); + uint CommonWord3 = (uint)Memory.ReadInt32(Position + 12); + uint CommonWord4 = (uint)Memory.ReadInt32(Position + 16); + + SphType = ReadBits(CommonWord0, 0, 5); + Version = ReadBits(CommonWord0, 5, 5); + ShaderType = ReadBits(CommonWord0, 10, 4); + MrtEnable = ReadBits(CommonWord0, 14, 1) != 0; + KillsPixels = ReadBits(CommonWord0, 15, 1) != 0; + DoesGlobalStore = ReadBits(CommonWord0, 16, 1) != 0; + SassVersion = ReadBits(CommonWord0, 17, 4); + DoesLoadOrStore = ReadBits(CommonWord0, 26, 1) != 0; + DoesFp64 = ReadBits(CommonWord0, 27, 1) != 0; + StreamOutMask = ReadBits(CommonWord0, 28, 4); + + ShaderLocalMemoryLowSize = ReadBits(CommonWord1, 0, 24); + PerPatchAttributeCount = ReadBits(CommonWord1, 24, 8); + + ShaderLocalMemoryHighSize = ReadBits(CommonWord2, 0, 24); + ThreadsPerInputPrimitive = ReadBits(CommonWord2, 24, 8); + + ShaderLocalMemoryCrsSize = ReadBits(CommonWord3, 0, 24); + OutputTopology = ReadBits(CommonWord3, 24, 4); + + MaxOutputVertexCount = ReadBits(CommonWord4, 0, 12); + StoreReqStart = ReadBits(CommonWord4, 12, 8); + StoreReqEnd = ReadBits(CommonWord4, 24, 8); + } + + private static int ReadBits(uint Word, int Offset, int BitWidth) + { + uint Mask = (1u << BitWidth) - 1u; + + return (int)((Word >> Offset) & Mask); + } + } +} \ No newline at end of file diff --git a/Ryujinx.Graphics/Gal/Shader/ShaderIrInst.cs b/Ryujinx.Graphics/Gal/Shader/ShaderIrInst.cs index 9841f58ff0..fd86cadb10 100644 --- a/Ryujinx.Graphics/Gal/Shader/ShaderIrInst.cs +++ b/Ryujinx.Graphics/Gal/Shader/ShaderIrInst.cs @@ -82,6 +82,9 @@ namespace Ryujinx.Graphics.Gal.Shader Bra, Exit, - Kil + Kil, + + Emit, + Cut } } \ No newline at end of file diff --git a/Ryujinx.Graphics/Gal/Shader/ShaderIrOperAbuf.cs b/Ryujinx.Graphics/Gal/Shader/ShaderIrOperAbuf.cs index fa612de76a..f17d9c0e62 100644 --- a/Ryujinx.Graphics/Gal/Shader/ShaderIrOperAbuf.cs +++ b/Ryujinx.Graphics/Gal/Shader/ShaderIrOperAbuf.cs @@ -2,13 +2,14 @@ namespace Ryujinx.Graphics.Gal.Shader { class ShaderIrOperAbuf : ShaderIrNode { - public int Offs { get; private set; } - public int GprIndex { get; private set; } + public int Offs { get; private set; } - public ShaderIrOperAbuf(int Offs, int GprIndex) + public ShaderIrNode Vertex { get; private set; } + + public ShaderIrOperAbuf(int Offs, ShaderIrNode Vertex) { - this.Offs = Offs; - this.GprIndex = GprIndex; + this.Offs = Offs; + this.Vertex = Vertex; } } } \ No newline at end of file diff --git a/Ryujinx.Graphics/Gal/Shader/ShaderOpCodeTable.cs b/Ryujinx.Graphics/Gal/Shader/ShaderOpCodeTable.cs index 1ac1178512..3f20dc4465 100644 --- a/Ryujinx.Graphics/Gal/Shader/ShaderOpCodeTable.cs +++ b/Ryujinx.Graphics/Gal/Shader/ShaderOpCodeTable.cs @@ -74,6 +74,7 @@ namespace Ryujinx.Graphics.Gal.Shader Set("0100110000100x", ShaderDecode.Imnmx_C); Set("0011100x00100x", ShaderDecode.Imnmx_I); Set("0101110000100x", ShaderDecode.Imnmx_R); + Set("1110111111010x", ShaderDecode.Isberd); Set("11100000xxxxxx", ShaderDecode.Ipa); Set("0100110000011x", ShaderDecode.Iscadd_C); Set("0011100x00011x", ShaderDecode.Iscadd_I); @@ -95,7 +96,9 @@ namespace Ryujinx.Graphics.Gal.Shader Set("0011100x10011x", ShaderDecode.Mov_I); Set("000000010000xx", ShaderDecode.Mov_I32); Set("0101110010011x", ShaderDecode.Mov_R); + Set("1111000011001x", ShaderDecode.Mov_S); Set("0101000010000x", ShaderDecode.Mufu); + Set("1111101111100x", ShaderDecode.Out_R); Set("0101000010010x", ShaderDecode.Psetp); Set("0100110010010x", ShaderDecode.Rro_C); Set("0011100x10010x", ShaderDecode.Rro_I); @@ -114,6 +117,7 @@ namespace Ryujinx.Graphics.Gal.Shader Set("1101111101001x", ShaderDecode.Texq); Set("1101100xxxxxxx", ShaderDecode.Texs); Set("1101101xxxxxxx", ShaderDecode.Tlds); + Set("01011111xxxxxx", ShaderDecode.Vmad); Set("0100111xxxxxxx", ShaderDecode.Xmad_CR); Set("0011011x00xxxx", ShaderDecode.Xmad_I); Set("010100010xxxxx", ShaderDecode.Xmad_RC); diff --git a/Ryujinx.Graphics/Gal/ShaderDumper.cs b/Ryujinx.Graphics/Gal/ShaderDumper.cs index 7cd56b21ef..541368e89b 100644 --- a/Ryujinx.Graphics/Gal/ShaderDumper.cs +++ b/Ryujinx.Graphics/Gal/ShaderDumper.cs @@ -18,13 +18,21 @@ namespace Ryujinx.Graphics.Gal string FileName = "Shader" + DumpIndex.ToString("d4") + "." + ShaderExtension(Type) + ExtSuffix + ".bin"; - string FilePath = Path.Combine(DumpDir(), FileName); + string FullPath = Path.Combine(FullDir(), FileName); + string CodePath = Path.Combine(CodeDir(), FileName); DumpIndex++; - using (FileStream Output = File.Create(FilePath)) - using (BinaryWriter Writer = new BinaryWriter(Output)) + using (FileStream FullFile = File.Create(FullPath)) + using (FileStream CodeFile = File.Create(CodePath)) + using (BinaryWriter FullWriter = new BinaryWriter(FullFile)) + using (BinaryWriter CodeWriter = new BinaryWriter(CodeFile)) { + for (long i = 0; i < 0x50; i += 4) + { + FullWriter.Write(Memory.ReadInt32(Position + i)); + } + long Offset = 0; ulong Instruction = 0; @@ -32,8 +40,8 @@ namespace Ryujinx.Graphics.Gal //Dump until a NOP instruction is found while ((Instruction >> 52 & 0xfff8) != 0x50b0) { - uint Word0 = (uint)Memory.ReadInt32(Position + Offset + 0); - uint Word1 = (uint)Memory.ReadInt32(Position + Offset + 4); + uint Word0 = (uint)Memory.ReadInt32(Position + 0x50 + Offset + 0); + uint Word1 = (uint)Memory.ReadInt32(Position + 0x50 + Offset + 4); Instruction = Word0 | (ulong)Word1 << 32; @@ -44,7 +52,8 @@ namespace Ryujinx.Graphics.Gal break; } - Writer.Write(Instruction); + FullWriter.Write(Instruction); + CodeWriter.Write(Instruction); Offset += 8; } @@ -52,13 +61,24 @@ namespace Ryujinx.Graphics.Gal //Align to meet nvdisasm requeriments while (Offset % 0x20 != 0) { - Writer.Write(0); + FullWriter.Write(0); + CodeWriter.Write(0); Offset += 4; } } } + private static string FullDir() + { + return CreateAndReturn(Path.Combine(DumpDir(), "Full")); + } + + private static string CodeDir() + { + return CreateAndReturn(Path.Combine(DumpDir(), "Code")); + } + private static string DumpDir() { if (string.IsNullOrEmpty(RuntimeDir)) @@ -79,6 +99,16 @@ namespace Ryujinx.Graphics.Gal return RuntimeDir; } + private static string CreateAndReturn(string Dir) + { + if (!Directory.Exists(Dir)) + { + Directory.CreateDirectory(Dir); + } + + return Dir; + } + private static string ShaderExtension(GalShaderType Type) { switch (Type) From ee064a2fb8d5e6167122477d5014beae509f92ed Mon Sep 17 00:00:00 2001 From: mailwl Date: Thu, 19 Jul 2018 19:45:50 +0300 Subject: [PATCH 11/34] .gitignore: ignore autogenerated launchSettings.json (#292) thanks to @Cyuubi --- .gitignore | 3 +++ 1 file changed, 3 insertions(+) diff --git a/.gitignore b/.gitignore index 82d9719b59..123f46184d 100644 --- a/.gitignore +++ b/.gitignore @@ -158,3 +158,6 @@ $RECYCLE.BIN/ # Mac desktop service store files .DS_Store + +# VS Launch Settings +launchSettings.json From 8b67297711003dfac432acb3278c5a406617e662 Mon Sep 17 00:00:00 2001 From: emmauss Date: Thu, 19 Jul 2018 20:49:34 +0300 Subject: [PATCH 12/34] Added appveyor configuration file (#277) --- appveyor.yml | 28 ++++++++++++++++++++++++++++ 1 file changed, 28 insertions(+) create mode 100644 appveyor.yml diff --git a/appveyor.yml b/appveyor.yml new file mode 100644 index 0000000000..eced37d77e --- /dev/null +++ b/appveyor.yml @@ -0,0 +1,28 @@ +version: 1.0.{build} +branches: + only: + - master +image: Visual Studio 2017 +configuration: Release +build_script: +- ps: >- + dotnet --version + + dotnet publish -c Release -r win-x64 + + dotnet publish -c Release -r linux-x64 + + dotnet publish -c Release -r osx-x64 + + 7z a ryujinx-$env:APPVEYOR_BUILD_VERSION-win_x64.zip $env:APPVEYOR_BUILD_FOLDER\Ryujinx\bin\Release\netcoreapp2.1\win-x64\publish\ + + 7z a ryujinx-$env:APPVEYOR_BUILD_VERSION-linux_x64.tar $env:APPVEYOR_BUILD_FOLDER\Ryujinx\bin\Release\netcoreapp2.1\linux-x64\publish\ + + 7z a ryujinx-$env:APPVEYOR_BUILD_VERSION-linux_x64.tar.gz ryujinx-$env:APPVEYOR_BUILD_VERSION-linux_x64.tar + + 7z a ryujinx-$env:APPVEYOR_BUILD_VERSION-osx_x64.zip $env:APPVEYOR_BUILD_FOLDER\Ryujinx\bin\Release\netcoreapp2.1\osx-x64\publish\ + +artifacts: +- path: ryujinx-%APPVEYOR_BUILD_VERSION%-win_x64.zip +- path: ryujinx-%APPVEYOR_BUILD_VERSION%-linux_x64.tar.gz +- path: ryujinx-%APPVEYOR_BUILD_VERSION%-osx_x64.zip From c9fc52edb6abc014d5d5671c1634b01ace48de2f Mon Sep 17 00:00:00 2001 From: Thomas Guillemard Date: Thu, 19 Jul 2018 20:44:52 +0200 Subject: [PATCH 13/34] Fix SystemPathToSwitchPath platform issues and make sure to delete temporary NRO after sessions dispose (#293) --- Ryujinx.HLE/OsHle/Horizon.cs | 1 - Ryujinx.HLE/OsHle/Process.cs | 10 +++++----- Ryujinx.HLE/VirtualFileSystem.cs | 4 ++-- 3 files changed, 7 insertions(+), 8 deletions(-) diff --git a/Ryujinx.HLE/OsHle/Horizon.cs b/Ryujinx.HLE/OsHle/Horizon.cs index 9d8a937ff2..70ae24be32 100644 --- a/Ryujinx.HLE/OsHle/Horizon.cs +++ b/Ryujinx.HLE/OsHle/Horizon.cs @@ -96,7 +96,6 @@ namespace Ryujinx.HLE.OsHle if (IsNro && (SwitchFilePath == null || !SwitchFilePath.StartsWith("sdmc:/"))) { - // TODO: avoid copying the file if we are already inside a sdmc directory string SwitchPath = $"sdmc:/switch/{Name}{Homebrew.TemporaryNroSuffix}"; string TempPath = Ns.VFs.SwitchPathToSystemPath(SwitchPath); diff --git a/Ryujinx.HLE/OsHle/Process.cs b/Ryujinx.HLE/OsHle/Process.cs index be27dcc288..c7606dc902 100644 --- a/Ryujinx.HLE/OsHle/Process.cs +++ b/Ryujinx.HLE/OsHle/Process.cs @@ -403,11 +403,6 @@ namespace Ryujinx.HLE.OsHle { if (Disposing && !Disposed) { - if (NeedsHbAbi && Executables[0].FilePath.EndsWith(Homebrew.TemporaryNroSuffix)) - { - File.Delete(Executables[0].FilePath); - } - //If there is still some thread running, disposing the objects is not //safe as the thread may try to access those resources. Instead, we set //the flag to have the Process disposed when all threads finishes. @@ -431,6 +426,11 @@ namespace Ryujinx.HLE.OsHle } } + if (NeedsHbAbi && Executables.Count > 0 && Executables[0].FilePath.EndsWith(Homebrew.TemporaryNroSuffix)) + { + File.Delete(Executables[0].FilePath); + } + INvDrvServices.UnloadProcess(this); AppletState.Dispose(); diff --git a/Ryujinx.HLE/VirtualFileSystem.cs b/Ryujinx.HLE/VirtualFileSystem.cs index 38df81f87d..df1fc9db13 100644 --- a/Ryujinx.HLE/VirtualFileSystem.cs +++ b/Ryujinx.HLE/VirtualFileSystem.cs @@ -57,11 +57,11 @@ namespace Ryujinx.HLE public string SystemPathToSwitchPath(string SystemPath) { - string BaseSystemPath = GetBasePath() + "/"; + string BaseSystemPath = GetBasePath() + Path.DirectorySeparatorChar; if (SystemPath.StartsWith(BaseSystemPath)) { string RawPath = SystemPath.Replace(BaseSystemPath, ""); - int FirstSeparatorOffset = RawPath.IndexOf('/'); + int FirstSeparatorOffset = RawPath.IndexOf(Path.DirectorySeparatorChar); if (FirstSeparatorOffset == -1) { return $"{RawPath}:/"; From 45bb24dbae7b4fb4118036aa74024605995510fd Mon Sep 17 00:00:00 2001 From: emmauss Date: Thu, 19 Jul 2018 18:53:49 +0000 Subject: [PATCH 14/34] fix extra space --- appveyor.yml | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/appveyor.yml b/appveyor.yml index eced37d77e..5392128132 100644 --- a/appveyor.yml +++ b/appveyor.yml @@ -18,7 +18,7 @@ build_script: 7z a ryujinx-$env:APPVEYOR_BUILD_VERSION-linux_x64.tar $env:APPVEYOR_BUILD_FOLDER\Ryujinx\bin\Release\netcoreapp2.1\linux-x64\publish\ - 7z a ryujinx-$env:APPVEYOR_BUILD_VERSION-linux_x64.tar.gz ryujinx-$env:APPVEYOR_BUILD_VERSION-linux_x64.tar + 7z a ryujinx-$env:APPVEYOR_BUILD_VERSION-linux_x64.tar.gz ryujinx-$env:APPVEYOR_BUILD_VERSION-linux_x64.tar 7z a ryujinx-$env:APPVEYOR_BUILD_VERSION-osx_x64.zip $env:APPVEYOR_BUILD_FOLDER\Ryujinx\bin\Release\netcoreapp2.1\osx-x64\publish\ From 5fe0bc584b21c660e083a9cb37aa0a8be4719f95 Mon Sep 17 00:00:00 2001 From: ReinUsesLisp Date: Thu, 19 Jul 2018 16:02:51 -0300 Subject: [PATCH 15/34] Send data to OpenGL host without client-side copies (#285) * Directly send host address to buffer data * Cleanup OGLShader * Directly copy vertex and index data too * Revert shader bind "cache" * Address feedback --- ChocolArm64/Memory/AMemory.cs | 7 ++ Ryujinx.Graphics/Gal/IGalRasterizer.cs | 6 +- Ryujinx.Graphics/Gal/IGalShader.cs | 3 +- Ryujinx.Graphics/Gal/OpenGL/OGLRasterizer.cs | 16 ++-- Ryujinx.Graphics/Gal/OpenGL/OGLShader.cs | 22 ++--- .../Gal/OpenGL/OGLStreamBuffer.cs | 93 +++---------------- Ryujinx.HLE/Gpu/Engines/NvGpuEngine3d.cs | 12 +-- Ryujinx.HLE/Gpu/Memory/NvGpuVmm.cs | 6 ++ 8 files changed, 55 insertions(+), 110 deletions(-) diff --git a/ChocolArm64/Memory/AMemory.cs b/ChocolArm64/Memory/AMemory.cs index da5cf00749..054277b295 100644 --- a/ChocolArm64/Memory/AMemory.cs +++ b/ChocolArm64/Memory/AMemory.cs @@ -204,6 +204,13 @@ namespace ChocolArm64.Memory return Modified; } + public IntPtr GetHostAddress(long Position, long Size) + { + EnsureRangeIsValid(Position, Size, AMemoryPerm.Read); + + return (IntPtr)(RamPtr + (ulong)Position); + } + public sbyte ReadSByte(long Position) { return (sbyte)ReadByte(Position); diff --git a/Ryujinx.Graphics/Gal/IGalRasterizer.cs b/Ryujinx.Graphics/Gal/IGalRasterizer.cs index 0c5d37e40e..a87d36c38f 100644 --- a/Ryujinx.Graphics/Gal/IGalRasterizer.cs +++ b/Ryujinx.Graphics/Gal/IGalRasterizer.cs @@ -1,3 +1,5 @@ +using System; + namespace Ryujinx.Graphics.Gal { public interface IGalRasterizer @@ -45,9 +47,9 @@ namespace Ryujinx.Graphics.Gal void SetPrimitiveRestartIndex(uint Index); - void CreateVbo(long Key, byte[] Buffer); + void CreateVbo(long Key, int DataSize, IntPtr HostAddress); - void CreateIbo(long Key, byte[] Buffer); + void CreateIbo(long Key, int DataSize, IntPtr HostAddress); void SetVertexArray(int Stride, long VboKey, GalVertexAttrib[] Attribs); diff --git a/Ryujinx.Graphics/Gal/IGalShader.cs b/Ryujinx.Graphics/Gal/IGalShader.cs index 9adaceaf50..56235a0709 100644 --- a/Ryujinx.Graphics/Gal/IGalShader.cs +++ b/Ryujinx.Graphics/Gal/IGalShader.cs @@ -1,3 +1,4 @@ +using System; using System.Collections.Generic; namespace Ryujinx.Graphics.Gal @@ -10,7 +11,7 @@ namespace Ryujinx.Graphics.Gal IEnumerable GetTextureUsage(long Key); - void SetConstBuffer(long Key, int Cbuf, byte[] Data); + void SetConstBuffer(long Key, int Cbuf, int DataSize, IntPtr HostAddress); void EnsureTextureBinding(string UniformName, int Value); diff --git a/Ryujinx.Graphics/Gal/OpenGL/OGLRasterizer.cs b/Ryujinx.Graphics/Gal/OpenGL/OGLRasterizer.cs index 0dc56966b3..f2e5859e5e 100644 --- a/Ryujinx.Graphics/Gal/OpenGL/OGLRasterizer.cs +++ b/Ryujinx.Graphics/Gal/OpenGL/OGLRasterizer.cs @@ -211,28 +211,28 @@ namespace Ryujinx.Graphics.Gal.OpenGL GL.PrimitiveRestartIndex(Index); } - public void CreateVbo(long Key, byte[] Buffer) + public void CreateVbo(long Key, int DataSize, IntPtr HostAddress) { int Handle = GL.GenBuffer(); - VboCache.AddOrUpdate(Key, Handle, (uint)Buffer.Length); + VboCache.AddOrUpdate(Key, Handle, (uint)DataSize); - IntPtr Length = new IntPtr(Buffer.Length); + IntPtr Length = new IntPtr(DataSize); GL.BindBuffer(BufferTarget.ArrayBuffer, Handle); - GL.BufferData(BufferTarget.ArrayBuffer, Length, Buffer, BufferUsageHint.StreamDraw); + GL.BufferData(BufferTarget.ArrayBuffer, Length, HostAddress, BufferUsageHint.StreamDraw); } - public void CreateIbo(long Key, byte[] Buffer) + public void CreateIbo(long Key, int DataSize, IntPtr HostAddress) { int Handle = GL.GenBuffer(); - IboCache.AddOrUpdate(Key, Handle, (uint)Buffer.Length); + IboCache.AddOrUpdate(Key, Handle, (uint)DataSize); - IntPtr Length = new IntPtr(Buffer.Length); + IntPtr Length = new IntPtr(DataSize); GL.BindBuffer(BufferTarget.ElementArrayBuffer, Handle); - GL.BufferData(BufferTarget.ElementArrayBuffer, Length, Buffer, BufferUsageHint.StreamDraw); + GL.BufferData(BufferTarget.ElementArrayBuffer, Length, HostAddress, BufferUsageHint.StreamDraw); } public void SetVertexArray(int Stride, long VboKey, GalVertexAttrib[] Attribs) diff --git a/Ryujinx.Graphics/Gal/OpenGL/OGLShader.cs b/Ryujinx.Graphics/Gal/OpenGL/OGLShader.cs index fe98aa0911..37213d8ed1 100644 --- a/Ryujinx.Graphics/Gal/OpenGL/OGLShader.cs +++ b/Ryujinx.Graphics/Gal/OpenGL/OGLShader.cs @@ -5,6 +5,8 @@ using System.Collections.Concurrent; using System.Collections.Generic; using System.Linq; +using Buffer = System.Buffer; + namespace Ryujinx.Graphics.Gal.OpenGL { public class OGLShader : IGalShader @@ -151,7 +153,7 @@ namespace Ryujinx.Graphics.Gal.OpenGL return Enumerable.Empty(); } - public void SetConstBuffer(long Key, int Cbuf, byte[] Data) + public void SetConstBuffer(long Key, int Cbuf, int DataSize, IntPtr HostAddress) { if (Stages.TryGetValue(Key, out ShaderStage Stage)) { @@ -159,13 +161,9 @@ namespace Ryujinx.Graphics.Gal.OpenGL { OGLStreamBuffer Buffer = GetConstBuffer(Stage.Type, Cbuf); - int Size = Math.Min(Data.Length, Buffer.Size); + int Size = Math.Min(DataSize, Buffer.Size); - byte[] Destiny = Buffer.Map(Size); - - Array.Copy(Data, Destiny, Size); - - Buffer.Unmap(Size); + Buffer.SetData(Size, HostAddress); } } } @@ -278,7 +276,7 @@ namespace Ryujinx.Graphics.Gal.OpenGL { int FreeBinding = 0; - int BindUniformBlocksIfNotNull(ShaderStage Stage) + void BindUniformBlocksIfNotNull(ShaderStage Stage) { if (Stage != null) { @@ -297,8 +295,6 @@ namespace Ryujinx.Graphics.Gal.OpenGL FreeBinding++; } } - - return FreeBinding; } BindUniformBlocksIfNotNull(Current.Vertex); @@ -312,7 +308,7 @@ namespace Ryujinx.Graphics.Gal.OpenGL { int FreeBinding = 0; - int BindUniformBuffersIfNotNull(ShaderStage Stage) + void BindUniformBuffersIfNotNull(ShaderStage Stage) { if (Stage != null) { @@ -325,8 +321,6 @@ namespace Ryujinx.Graphics.Gal.OpenGL FreeBinding++; } } - - return FreeBinding; } BindUniformBuffersIfNotNull(Current.Vertex); @@ -347,7 +341,7 @@ namespace Ryujinx.Graphics.Gal.OpenGL //Allocate a maximum of 64 KiB int Size = Math.Min(GL.GetInteger(GetPName.MaxUniformBlockSize), 64 * 1024); - Buffer = OGLStreamBuffer.Create(BufferTarget.UniformBuffer, Size); + Buffer = new OGLStreamBuffer(BufferTarget.UniformBuffer, Size); ConstBuffers[StageIndex][Cbuf] = Buffer; } diff --git a/Ryujinx.Graphics/Gal/OpenGL/OGLStreamBuffer.cs b/Ryujinx.Graphics/Gal/OpenGL/OGLStreamBuffer.cs index 329c5b5df7..0d5dee93f5 100644 --- a/Ryujinx.Graphics/Gal/OpenGL/OGLStreamBuffer.cs +++ b/Ryujinx.Graphics/Gal/OpenGL/OGLStreamBuffer.cs @@ -1,9 +1,9 @@ -using System; using OpenTK.Graphics.OpenGL; +using System; namespace Ryujinx.Graphics.Gal.OpenGL { - abstract class OGLStreamBuffer : IDisposable + class OGLStreamBuffer : IDisposable { public int Handle { get; protected set; } @@ -11,53 +11,25 @@ namespace Ryujinx.Graphics.Gal.OpenGL protected BufferTarget Target { get; private set; } - private bool Mapped = false; - - public OGLStreamBuffer(BufferTarget Target, int MaxSize) + public OGLStreamBuffer(BufferTarget Target, int Size) { - Handle = 0; - Mapped = false; - this.Target = Target; - this.Size = MaxSize; + this.Size = Size; + + Handle = GL.GenBuffer(); + + GL.BindBuffer(Target, Handle); + + GL.BufferData(Target, Size, IntPtr.Zero, BufferUsageHint.StreamDraw); } - public static OGLStreamBuffer Create(BufferTarget Target, int MaxSize) + public void SetData(int Size, IntPtr HostAddress) { - //TODO: Query here for ARB_buffer_storage and use when available - return new SubDataBuffer(Target, MaxSize); + GL.BindBuffer(Target, Handle); + + GL.BufferSubData(Target, IntPtr.Zero, Size, HostAddress); } - public byte[] Map(int Size) - { - if (Handle == 0 || Mapped || Size > this.Size) - { - throw new InvalidOperationException(); - } - - byte[] Memory = InternMap(Size); - - Mapped = true; - - return Memory; - } - - public void Unmap(int UsedSize) - { - if (Handle == 0 || !Mapped) - { - throw new InvalidOperationException(); - } - - InternUnmap(UsedSize); - - Mapped = false; - } - - protected abstract byte[] InternMap(int Size); - - protected abstract void InternUnmap(int UsedSize); - public void Dispose() { Dispose(true); @@ -73,41 +45,4 @@ namespace Ryujinx.Graphics.Gal.OpenGL } } } - - class SubDataBuffer : OGLStreamBuffer - { - private byte[] Memory; - - public SubDataBuffer(BufferTarget Target, int MaxSize) - : base(Target, MaxSize) - { - Memory = new byte[MaxSize]; - - GL.GenBuffers(1, out int Handle); - - GL.BindBuffer(Target, Handle); - - GL.BufferData(Target, Size, IntPtr.Zero, BufferUsageHint.StreamDraw); - - this.Handle = Handle; - } - - protected override byte[] InternMap(int Size) - { - return Memory; - } - - protected override void InternUnmap(int UsedSize) - { - GL.BindBuffer(Target, Handle); - - unsafe - { - fixed (byte* MemoryPtr = Memory) - { - GL.BufferSubData(Target, IntPtr.Zero, UsedSize, (IntPtr)MemoryPtr); - } - } - } - } } diff --git a/Ryujinx.HLE/Gpu/Engines/NvGpuEngine3d.cs b/Ryujinx.HLE/Gpu/Engines/NvGpuEngine3d.cs index dce25a5e94..c3e7a77fcb 100644 --- a/Ryujinx.HLE/Gpu/Engines/NvGpuEngine3d.cs +++ b/Ryujinx.HLE/Gpu/Engines/NvGpuEngine3d.cs @@ -560,9 +560,9 @@ namespace Ryujinx.HLE.Gpu.Engines if (Cb.Enabled) { - byte[] Data = Vmm.ReadBytes(Cb.Position, (uint)Cb.Size); + IntPtr DataAddress = Vmm.GetHostAddress(Cb.Position, Cb.Size); - Gpu.Renderer.Shader.SetConstBuffer(BasePosition + (uint)Offset, Cbuf, Data); + Gpu.Renderer.Shader.SetConstBuffer(BasePosition + (uint)Offset, Cbuf, Cb.Size, DataAddress); } } } @@ -595,9 +595,9 @@ namespace Ryujinx.HLE.Gpu.Engines if (!IboCached || Vmm.IsRegionModified(IboKey, (uint)IbSize, NvGpuBufferType.Index)) { - byte[] Data = Vmm.ReadBytes(IndexPosition, (uint)IbSize); + IntPtr DataAddress = Vmm.GetHostAddress(IndexPosition, IbSize); - Gpu.Renderer.Rasterizer.CreateIbo(IboKey, Data); + Gpu.Renderer.Rasterizer.CreateIbo(IboKey, IbSize, DataAddress); } Gpu.Renderer.Rasterizer.SetIndexArray(IbSize, IndexFormat); @@ -659,9 +659,9 @@ namespace Ryujinx.HLE.Gpu.Engines if (!VboCached || Vmm.IsRegionModified(VboKey, VbSize, NvGpuBufferType.Vertex)) { - byte[] Data = Vmm.ReadBytes(VertexPosition, VbSize); + IntPtr DataAddress = Vmm.GetHostAddress(VertexPosition, VbSize); - Gpu.Renderer.Rasterizer.CreateVbo(VboKey, Data); + Gpu.Renderer.Rasterizer.CreateVbo(VboKey, (int)VbSize, DataAddress); } Gpu.Renderer.Rasterizer.SetVertexArray(Stride, VboKey, Attribs[Index].ToArray()); diff --git a/Ryujinx.HLE/Gpu/Memory/NvGpuVmm.cs b/Ryujinx.HLE/Gpu/Memory/NvGpuVmm.cs index 0c81dd1508..7b23e49fac 100644 --- a/Ryujinx.HLE/Gpu/Memory/NvGpuVmm.cs +++ b/Ryujinx.HLE/Gpu/Memory/NvGpuVmm.cs @@ -1,5 +1,6 @@ using ChocolArm64.Memory; using Ryujinx.Graphics.Gal; +using System; using System.Collections.Concurrent; namespace Ryujinx.HLE.Gpu.Memory @@ -279,6 +280,11 @@ namespace Ryujinx.HLE.Gpu.Memory return Cache.IsRegionModified(Memory, BufferType, PA, Size); } + public IntPtr GetHostAddress(long Position, long Size) + { + return Memory.GetHostAddress(GetPhysicalAddress(Position), Size); + } + public byte ReadByte(long Position) { Position = GetPhysicalAddress(Position); From 241b46540d4e1ab909cc8b1b494189c4800ebcfd Mon Sep 17 00:00:00 2001 From: Rygnus Date: Fri, 20 Jul 2018 00:27:50 +0100 Subject: [PATCH 16/34] Stub AppletOE UnlockExit (#279) * Implement AppletOE UnlockExit --- .../OsHle/Services/Am/ISelfController.cs | 20 ++++++++++++++++++- 1 file changed, 19 insertions(+), 1 deletion(-) diff --git a/Ryujinx.HLE/OsHle/Services/Am/ISelfController.cs b/Ryujinx.HLE/OsHle/Services/Am/ISelfController.cs index ee0fb91569..10d69b9b08 100644 --- a/Ryujinx.HLE/OsHle/Services/Am/ISelfController.cs +++ b/Ryujinx.HLE/OsHle/Services/Am/ISelfController.cs @@ -17,7 +17,9 @@ namespace Ryujinx.HLE.OsHle.Services.Am { m_Commands = new Dictionary() { + { 0, Exit }, { 1, LockExit }, + { 2, UnlockExit }, { 9, GetLibraryAppletLaunchableEvent }, { 10, SetScreenShotPermission }, { 11, SetOperationModeChangedNotification }, @@ -31,8 +33,24 @@ namespace Ryujinx.HLE.OsHle.Services.Am LaunchableEvent = new KEvent(); } + public long Exit(ServiceCtx Context) + { + Context.Ns.Log.PrintStub(LogClass.ServiceAm, "Stubbed."); + + return 0; + } + public long LockExit(ServiceCtx Context) { + Context.Ns.Log.PrintStub(LogClass.ServiceAm, "Stubbed."); + + return 0; + } + + public long UnlockExit(ServiceCtx Context) + { + Context.Ns.Log.PrintStub(LogClass.ServiceAm, "Stubbed."); + return 0; } @@ -114,4 +132,4 @@ namespace Ryujinx.HLE.OsHle.Services.Am return 0; } } -} \ No newline at end of file +} From 4b8a65fd84fcab991f62cf9cfdd8484c2e55501e Mon Sep 17 00:00:00 2001 From: Starlet Date: Fri, 20 Jul 2018 17:53:06 -0400 Subject: [PATCH 17/34] Add Npdm parsing into Ryujinx.HLE --- Ryujinx.HLE/OsHle/Horizon.cs | 20 ++++++++++++++++++++ Ryujinx.HLE/OsHle/SystemStateMgr.cs | 3 +++ 2 files changed, 23 insertions(+) diff --git a/Ryujinx.HLE/OsHle/Horizon.cs b/Ryujinx.HLE/OsHle/Horizon.cs index 70ae24be32..4c117fa478 100644 --- a/Ryujinx.HLE/OsHle/Horizon.cs +++ b/Ryujinx.HLE/OsHle/Horizon.cs @@ -1,4 +1,5 @@ using Ryujinx.HLE.Loaders.Executables; +using Ryujinx.HLE.Loaders.Npdm; using Ryujinx.HLE.Logging; using Ryujinx.HLE.OsHle.Handles; using System; @@ -76,6 +77,25 @@ namespace Ryujinx.HLE.OsHle } } + void LoadNpdm(string FileName) + { + string File = Directory.GetFiles(ExeFsDir, FileName)[0]; + + Ns.Log.PrintInfo(LogClass.Loader, "Loading Title Metadata..."); + + using (FileStream Input = new FileStream(File, FileMode.Open)) + { + SystemStateMgr.TitleMetadata = new Npdm(Input); + } + } + + LoadNpdm("*.npdm"); + + if (!SystemStateMgr.TitleMetadata.Is64Bits) + { + throw new Exception("32-bit titles are unsupported!"); + } + LoadNso("rtld"); MainProcess.SetEmptyArgs(); diff --git a/Ryujinx.HLE/OsHle/SystemStateMgr.cs b/Ryujinx.HLE/OsHle/SystemStateMgr.cs index e78082c45a..32c88994b9 100644 --- a/Ryujinx.HLE/OsHle/SystemStateMgr.cs +++ b/Ryujinx.HLE/OsHle/SystemStateMgr.cs @@ -1,9 +1,12 @@ +using Ryujinx.HLE.Loaders.Npdm; using System; namespace Ryujinx.HLE.OsHle { public class SystemStateMgr { + internal static Npdm TitleMetadata { get; set; } + internal static string[] LanguageCodes = new string[] { "ja", From ed075ae3cdc99163aa910b628bcd9de03dae03dd Mon Sep 17 00:00:00 2001 From: Ac_K Date: Sun, 22 Jul 2018 19:38:29 +0000 Subject: [PATCH 18/34] Update ITimeZoneService.cs Fix typo --- Ryujinx.HLE/OsHle/Services/Time/ITimeZoneService.cs | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/Ryujinx.HLE/OsHle/Services/Time/ITimeZoneService.cs b/Ryujinx.HLE/OsHle/Services/Time/ITimeZoneService.cs index a2206a126e..38440eef90 100644 --- a/Ryujinx.HLE/OsHle/Services/Time/ITimeZoneService.cs +++ b/Ryujinx.HLE/OsHle/Services/Time/ITimeZoneService.cs @@ -117,7 +117,7 @@ namespace Ryujinx.HLE.OsHle.Services.Time TimeZoneInfo Info = TimeZoneInfo.FindSystemTimeZoneById(TzID); byte[] TzData = Encoding.ASCII.GetBytes(Info.Id); - // FIXME: This is not in ANY cases accurate, but the games don't about the content of the buffer, they only pass it. + // FIXME: This is not in ANY cases accurate, but the games don't care about the content of the buffer, they only pass it. // TODO: Reverse the TZif2 conversion in PCV to make this match with real hardware. Context.Memory.WriteBytes(BufferPosition, TzData); } From ed29982f9b0414d42fd75acdcf2182d14174fe90 Mon Sep 17 00:00:00 2001 From: greggameplayer <33609333+greggameplayer@users.noreply.github.com> Date: Mon, 23 Jul 2018 16:20:16 +0200 Subject: [PATCH 19/34] Link BCAT:U & BCAT:A & BCAT:M & BCAT:S (#257) * Link BCAT:U & BCAT:A & BCAT:M & BCAT:S * delete unneeded using * delete unneeded spaces * delete unneeded using * Add comment (1/2) * Add comment (2/2) * delete unneeded using --- .../OsHle/Services/Bcat/IBcatService.cs | 21 ++++++++++ .../Bcat/IDeliveryCacheStorageService.cs | 21 ++++++++++ .../OsHle/Services/Bcat/IServiceCreator.cs | 39 +++++++++++++++++++ Ryujinx.HLE/OsHle/Services/ServiceFactory.cs | 17 +++++++- 4 files changed, 96 insertions(+), 2 deletions(-) create mode 100644 Ryujinx.HLE/OsHle/Services/Bcat/IBcatService.cs create mode 100644 Ryujinx.HLE/OsHle/Services/Bcat/IDeliveryCacheStorageService.cs create mode 100644 Ryujinx.HLE/OsHle/Services/Bcat/IServiceCreator.cs diff --git a/Ryujinx.HLE/OsHle/Services/Bcat/IBcatService.cs b/Ryujinx.HLE/OsHle/Services/Bcat/IBcatService.cs new file mode 100644 index 0000000000..b7754d6b63 --- /dev/null +++ b/Ryujinx.HLE/OsHle/Services/Bcat/IBcatService.cs @@ -0,0 +1,21 @@ +using Ryujinx.HLE.OsHle.Ipc; +using System.Collections.Generic; + +namespace Ryujinx.HLE.OsHle.Services.Bcat +{ + class IBcatService : IpcService + { + private Dictionary m_Commands; + + public override IReadOnlyDictionary Commands => m_Commands; + + public IBcatService() + { + m_Commands = new Dictionary() + { + //... + }; + } + + } +} diff --git a/Ryujinx.HLE/OsHle/Services/Bcat/IDeliveryCacheStorageService.cs b/Ryujinx.HLE/OsHle/Services/Bcat/IDeliveryCacheStorageService.cs new file mode 100644 index 0000000000..0b84d809e4 --- /dev/null +++ b/Ryujinx.HLE/OsHle/Services/Bcat/IDeliveryCacheStorageService.cs @@ -0,0 +1,21 @@ +using Ryujinx.HLE.OsHle.Ipc; +using System.Collections.Generic; + +namespace Ryujinx.HLE.OsHle.Services.Bcat +{ + class IDeliveryCacheStorageService : IpcService + { + private Dictionary m_Commands; + + public override IReadOnlyDictionary Commands => m_Commands; + + public IDeliveryCacheStorageService() + { + m_Commands = new Dictionary() + { + //... + }; + } + + } +} diff --git a/Ryujinx.HLE/OsHle/Services/Bcat/IServiceCreator.cs b/Ryujinx.HLE/OsHle/Services/Bcat/IServiceCreator.cs new file mode 100644 index 0000000000..cc1fc6f8c8 --- /dev/null +++ b/Ryujinx.HLE/OsHle/Services/Bcat/IServiceCreator.cs @@ -0,0 +1,39 @@ +using Ryujinx.HLE.OsHle.Ipc; +using System.Collections.Generic; + +namespace Ryujinx.HLE.OsHle.Services.Bcat +{ + class IServiceCreator : IpcService + { + private Dictionary m_Commands; + + public override IReadOnlyDictionary Commands => m_Commands; + + public IServiceCreator() + { + m_Commands = new Dictionary() + { + { 0, CreateBcatService }, + { 1, CreateDeliveryCacheStorageService } + }; + } + + public long CreateBcatService(ServiceCtx Context) + { + long Id = Context.RequestData.ReadInt64(); + + MakeObject(Context, new IBcatService()); + + return 0; + } + + public long CreateDeliveryCacheStorageService(ServiceCtx Context) + { + long Id = Context.RequestData.ReadInt64(); + + MakeObject(Context, new IDeliveryCacheStorageService()); + + return 0; + } + } +} diff --git a/Ryujinx.HLE/OsHle/Services/ServiceFactory.cs b/Ryujinx.HLE/OsHle/Services/ServiceFactory.cs index b69fc9f884..712698b9af 100644 --- a/Ryujinx.HLE/OsHle/Services/ServiceFactory.cs +++ b/Ryujinx.HLE/OsHle/Services/ServiceFactory.cs @@ -2,6 +2,7 @@ using Ryujinx.HLE.OsHle.Services.Acc; using Ryujinx.HLE.OsHle.Services.Am; using Ryujinx.HLE.OsHle.Services.Apm; using Ryujinx.HLE.OsHle.Services.Aud; +using Ryujinx.HLE.OsHle.Services.Bcat; using Ryujinx.HLE.OsHle.Services.Bsd; using Ryujinx.HLE.OsHle.Services.Caps; using Ryujinx.HLE.OsHle.Services.Friend; @@ -55,6 +56,18 @@ namespace Ryujinx.HLE.OsHle.Services case "audren:u": return new IAudioRendererManager(); + case "bcat:a": + return new Bcat.IServiceCreator(); + + case "bcat:m": + return new Bcat.IServiceCreator(); + + case "bcat:u": + return new Bcat.IServiceCreator(); + + case "bcat:s": + return new Bcat.IServiceCreator(); + case "bsd:s": return new IClient(); @@ -71,10 +84,10 @@ namespace Ryujinx.HLE.OsHle.Services return new IRandomInterface(); case "friend:a": - return new IServiceCreator(); + return new Friend.IServiceCreator(); case "friend:u": - return new IServiceCreator(); + return new Friend.IServiceCreator(); case "fsp-srv": return new IFileSystemProxy(); From 1344a47c774ab2b7491c59e363e04a499fa432f3 Mon Sep 17 00:00:00 2001 From: ReinUsesLisp Date: Mon, 23 Jul 2018 11:21:05 -0300 Subject: [PATCH 20/34] Blit framebuffer without shaders (#229) * Blit framebuffer without shaders * De-hardcode native size values * Adapt to dehardcoded framebuffers and address feedback * Remove framebuffer rebinding --- Ryujinx.Graphics/Gal/IGalFrameBuffer.cs | 2 +- Ryujinx.Graphics/Gal/OpenGL/FbFragShader.glsl | 13 - Ryujinx.Graphics/Gal/OpenGL/FbVtxShader.glsl | 28 -- Ryujinx.Graphics/Gal/OpenGL/OGLFrameBuffer.cs | 284 ++++++------------ Ryujinx.Graphics/Ryujinx.Graphics.csproj | 9 - Ryujinx.HLE/OsHle/Services/Vi/NvFlinger.cs | 53 +--- 6 files changed, 101 insertions(+), 288 deletions(-) delete mode 100644 Ryujinx.Graphics/Gal/OpenGL/FbFragShader.glsl delete mode 100644 Ryujinx.Graphics/Gal/OpenGL/FbVtxShader.glsl diff --git a/Ryujinx.Graphics/Gal/IGalFrameBuffer.cs b/Ryujinx.Graphics/Gal/IGalFrameBuffer.cs index 1f62bdb379..c0287ef8be 100644 --- a/Ryujinx.Graphics/Gal/IGalFrameBuffer.cs +++ b/Ryujinx.Graphics/Gal/IGalFrameBuffer.cs @@ -14,7 +14,7 @@ namespace Ryujinx.Graphics.Gal void Set(byte[] Data, int Width, int Height); - void SetTransform(float SX, float SY, float Rotate, float TX, float TY); + void SetTransform(bool FlipX, bool FlipY, int Top, int Left, int Right, int Bottom); void SetWindowSize(int Width, int Height); diff --git a/Ryujinx.Graphics/Gal/OpenGL/FbFragShader.glsl b/Ryujinx.Graphics/Gal/OpenGL/FbFragShader.glsl deleted file mode 100644 index 74e33bd7c0..0000000000 --- a/Ryujinx.Graphics/Gal/OpenGL/FbFragShader.glsl +++ /dev/null @@ -1,13 +0,0 @@ -#version 330 core - -precision highp float; - -uniform sampler2D tex; - -in vec2 tex_coord; - -out vec4 out_frag_color; - -void main(void) { - out_frag_color = texture(tex, tex_coord); -} \ No newline at end of file diff --git a/Ryujinx.Graphics/Gal/OpenGL/FbVtxShader.glsl b/Ryujinx.Graphics/Gal/OpenGL/FbVtxShader.glsl deleted file mode 100644 index 35d560c09b..0000000000 --- a/Ryujinx.Graphics/Gal/OpenGL/FbVtxShader.glsl +++ /dev/null @@ -1,28 +0,0 @@ -#version 330 core - -precision highp float; - -uniform mat2 transform; -uniform vec2 window_size; -uniform vec2 offset; - -layout(location = 0) in vec2 in_position; -layout(location = 1) in vec2 in_tex_coord; - -out vec2 tex_coord; - -// Have a fixed aspect ratio, fit the image within the available space. -vec2 get_scale_ratio(void) { - vec2 native_size = vec2(1280, 720); - vec2 ratio = vec2( - (window_size.y * native_size.x) / (native_size.y * window_size.x), - (window_size.x * native_size.y) / (native_size.x * window_size.y) - ); - return min(ratio, 1); -} - -void main(void) { - tex_coord = in_tex_coord; - vec2 t_pos = (transform * in_position) + offset; - gl_Position = vec4(t_pos * get_scale_ratio(), 0, 1); -} \ No newline at end of file diff --git a/Ryujinx.Graphics/Gal/OpenGL/OGLFrameBuffer.cs b/Ryujinx.Graphics/Gal/OpenGL/OGLFrameBuffer.cs index cd52762c79..30a3de64a0 100644 --- a/Ryujinx.Graphics/Gal/OpenGL/OGLFrameBuffer.cs +++ b/Ryujinx.Graphics/Gal/OpenGL/OGLFrameBuffer.cs @@ -32,48 +32,45 @@ namespace Ryujinx.Graphics.Gal.OpenGL public int RbHandle { get; private set; } public int TexHandle { get; private set; } - public FrameBuffer(int Width, int Height) + public FrameBuffer(int Width, int Height, bool HasRenderBuffer) { this.Width = Width; this.Height = Height; Handle = GL.GenFramebuffer(); - RbHandle = GL.GenRenderbuffer(); TexHandle = GL.GenTexture(); + + if (HasRenderBuffer) + { + RbHandle = GL.GenRenderbuffer(); + } } } - private struct ShaderProgram - { - public int Handle; - public int VpHandle; - public int FpHandle; - } + private const int NativeWidth = 1280; + private const int NativeHeight = 720; private Dictionary Fbs; - private ShaderProgram Shader; - private Rect Viewport; private Rect Window; - private bool IsInitialized; + private FrameBuffer CurrFb; + private FrameBuffer CurrReadFb; - private int RawFbTexWidth; - private int RawFbTexHeight; - private int RawFbTexHandle; + private FrameBuffer RawFb; - private int CurrFbHandle; - private int CurrTexHandle; + private bool FlipX; + private bool FlipY; - private int VaoHandle; - private int VboHandle; + private int CropTop; + private int CropLeft; + private int CropRight; + private int CropBottom; public OGLFrameBuffer() { Fbs = new Dictionary(); - - Shader = new ShaderProgram(); } public void Create(long Key, int Width, int Height) @@ -92,7 +89,7 @@ namespace Ryujinx.Graphics.Gal.OpenGL return; } - Fb = new FrameBuffer(Width, Height); + Fb = new FrameBuffer(Width, Height, true); SetupTexture(Fb.TexHandle, Width, Height); @@ -129,7 +126,7 @@ namespace Ryujinx.Graphics.Gal.OpenGL { GL.BindFramebuffer(FramebufferTarget.Framebuffer, Fb.Handle); - CurrFbHandle = Fb.Handle; + CurrFb = Fb; } } @@ -147,75 +144,50 @@ namespace Ryujinx.Graphics.Gal.OpenGL { if (Fbs.TryGetValue(Key, out FrameBuffer Fb)) { - CurrTexHandle = Fb.TexHandle; + CurrReadFb = Fb; } } public void Set(byte[] Data, int Width, int Height) { - if (RawFbTexHandle == 0) + if (RawFb == null) { - RawFbTexHandle = GL.GenTexture(); + CreateRawFb(Width, Height); } - if (RawFbTexWidth != Width || - RawFbTexHeight != Height) + if (RawFb.Width != Width || + RawFb.Height != Height) { - SetupTexture(RawFbTexHandle, Width, Height); + SetupTexture(RawFb.TexHandle, Width, Height); - RawFbTexWidth = Width; - RawFbTexHeight = Height; + RawFb.Width = Width; + RawFb.Height = Height; } GL.ActiveTexture(TextureUnit.Texture0); - GL.BindTexture(TextureTarget.Texture2D, RawFbTexHandle); + GL.BindTexture(TextureTarget.Texture2D, RawFb.TexHandle); (PixelFormat Format, PixelType Type) = OGLEnumConverter.GetTextureFormat(GalTextureFormat.A8B8G8R8); GL.TexSubImage2D(TextureTarget.Texture2D, 0, 0, 0, Width, Height, Format, Type, Data); - CurrTexHandle = RawFbTexHandle; + CurrReadFb = RawFb; } - public void SetTransform(float SX, float SY, float Rotate, float TX, float TY) + public void SetTransform(bool FlipX, bool FlipY, int Top, int Left, int Right, int Bottom) { - EnsureInitialized(); + this.FlipX = FlipX; + this.FlipY = FlipY; - Matrix2 Transform; - - Transform = Matrix2.CreateScale(SX, SY); - Transform *= Matrix2.CreateRotation(Rotate); - - Vector2 Offs = new Vector2(TX, TY); - - int CurrentProgram = GL.GetInteger(GetPName.CurrentProgram); - - GL.UseProgram(Shader.Handle); - - int TransformUniformLocation = GL.GetUniformLocation(Shader.Handle, "transform"); - - GL.UniformMatrix2(TransformUniformLocation, false, ref Transform); - - int OffsetUniformLocation = GL.GetUniformLocation(Shader.Handle, "offset"); - - GL.Uniform2(OffsetUniformLocation, ref Offs); - - GL.UseProgram(CurrentProgram); + CropTop = Top; + CropLeft = Left; + CropRight = Right; + CropBottom = Bottom; } public void SetWindowSize(int Width, int Height) { - int CurrentProgram = GL.GetInteger(GetPName.CurrentProgram); - - GL.UseProgram(Shader.Handle); - - int WindowSizeUniformLocation = GL.GetUniformLocation(Shader.Handle, "window_size"); - - GL.Uniform2(WindowSizeUniformLocation, new Vector2(Width, Height)); - - GL.UseProgram(CurrentProgram); - Window = new Rect(0, 0, Width, Height); } @@ -237,72 +209,59 @@ namespace Ryujinx.Graphics.Gal.OpenGL public void Render() { - if (CurrTexHandle != 0) + if (CurrReadFb != null) { - EnsureInitialized(); + int SrcX0, SrcX1, SrcY0, SrcY1; - //bool CullFaceEnable = GL.IsEnabled(EnableCap.CullFace); + if (CropLeft == 0 && CropRight == 0) + { + SrcX0 = 0; + SrcX1 = CurrReadFb.Width; + } + else + { + SrcX0 = CropLeft; + SrcX1 = CropRight; + } - bool DepthTestEnable = GL.IsEnabled(EnableCap.DepthTest); + if (CropTop == 0 && CropBottom == 0) + { + SrcY0 = 0; + SrcY1 = CurrReadFb.Height; + } + else + { + SrcY0 = CropTop; + SrcY1 = CropBottom; + } - bool StencilTestEnable = GL.IsEnabled(EnableCap.StencilTest); + float RatioX = MathF.Min(1f, (Window.Height * (float)NativeWidth) / ((float)NativeHeight * Window.Width)); + float RatioY = MathF.Min(1f, (Window.Width * (float)NativeHeight) / ((float)NativeWidth * Window.Height)); - bool AlphaBlendEnable = GL.IsEnabled(EnableCap.Blend); + int DstWidth = (int)(Window.Width * RatioX); + int DstHeight = (int)(Window.Height * RatioY); - //GL.Disable(EnableCap.CullFace); + int DstPaddingX = (Window.Width - DstWidth) / 2; + int DstPaddingY = (Window.Height - DstHeight) / 2; - GL.Disable(EnableCap.DepthTest); + int DstX0 = FlipX ? Window.Width - DstPaddingX : DstPaddingX; + int DstX1 = FlipX ? DstPaddingX : Window.Width - DstPaddingX; - GL.Disable(EnableCap.StencilTest); - - GL.Disable(EnableCap.Blend); - - GL.ActiveTexture(TextureUnit.Texture0); - - GL.BindTexture(TextureTarget.Texture2D, CurrTexHandle); - - int CurrentProgram = GL.GetInteger(GetPName.CurrentProgram); + int DstY0 = FlipY ? DstPaddingY : Window.Height - DstPaddingY; + int DstY1 = FlipY ? Window.Height - DstPaddingY : DstPaddingY; GL.BindFramebuffer(FramebufferTarget.Framebuffer, 0); - SetViewport(Window); + GL.Viewport(0, 0, Window.Width, Window.Height); - GL.Clear( - ClearBufferMask.ColorBufferBit | - ClearBufferMask.DepthBufferBit); + GL.BindFramebuffer(FramebufferTarget.ReadFramebuffer, CurrReadFb.Handle); - GL.BindVertexArray(VaoHandle); + GL.Clear(ClearBufferMask.ColorBufferBit | ClearBufferMask.DepthBufferBit); - GL.UseProgram(Shader.Handle); - - GL.DrawArrays(PrimitiveType.TriangleStrip, 0, 4); - - //Restore the original state. - GL.BindFramebuffer(FramebufferTarget.Framebuffer, CurrFbHandle); - - GL.UseProgram(CurrentProgram); - - //if (CullFaceEnable) - //{ - // GL.Enable(EnableCap.CullFace); - //} - - if (DepthTestEnable) - { - GL.Enable(EnableCap.DepthTest); - } - - if (StencilTestEnable) - { - GL.Enable(EnableCap.StencilTest); - } - - if (AlphaBlendEnable) - { - GL.Enable(EnableCap.Blend); - } - - SetViewport(Viewport); + GL.BlitFramebuffer( + SrcX0, SrcY0, SrcX1, SrcY1, + DstX0, DstY0, DstX1, DstY1, + ClearBufferMask.ColorBufferBit, BlitFramebufferFilter.Linear); } } @@ -354,8 +313,6 @@ namespace Ryujinx.Graphics.Gal.OpenGL Data); Callback(Data); - - GL.BindFramebuffer(FramebufferTarget.ReadFramebuffer, CurrFbHandle); } } @@ -390,86 +347,29 @@ namespace Ryujinx.Graphics.Gal.OpenGL } } - private void EnsureInitialized() + private void CreateRawFb(int Width, int Height) { - if (!IsInitialized) + if (RawFb == null) { - IsInitialized = true; + RawFb = new FrameBuffer(Width, Height, false); - SetupShader(); - SetupVertex(); + SetupTexture(RawFb.TexHandle, Width, Height); + + RawFb.Width = Width; + RawFb.Height = Height; + + GL.BindFramebuffer(FramebufferTarget.Framebuffer, RawFb.Handle); + + GL.FramebufferTexture( + FramebufferTarget.Framebuffer, + FramebufferAttachment.ColorAttachment0, + RawFb.TexHandle, + 0); + + GL.Viewport(0, 0, Width, Height); } } - private void SetupShader() - { - Shader.VpHandle = GL.CreateShader(ShaderType.VertexShader); - Shader.FpHandle = GL.CreateShader(ShaderType.FragmentShader); - - string VpSource = EmbeddedResource.GetString("GlFbVtxShader"); - string FpSource = EmbeddedResource.GetString("GlFbFragShader"); - - GL.ShaderSource(Shader.VpHandle, VpSource); - GL.ShaderSource(Shader.FpHandle, FpSource); - GL.CompileShader(Shader.VpHandle); - GL.CompileShader(Shader.FpHandle); - - Shader.Handle = GL.CreateProgram(); - - GL.AttachShader(Shader.Handle, Shader.VpHandle); - GL.AttachShader(Shader.Handle, Shader.FpHandle); - GL.LinkProgram(Shader.Handle); - GL.UseProgram(Shader.Handle); - - Matrix2 Transform = Matrix2.Identity; - - int TexUniformLocation = GL.GetUniformLocation(Shader.Handle, "tex"); - - GL.Uniform1(TexUniformLocation, 0); - - int WindowSizeUniformLocation = GL.GetUniformLocation(Shader.Handle, "window_size"); - - GL.Uniform2(WindowSizeUniformLocation, new Vector2(1280.0f, 720.0f)); - - int TransformUniformLocation = GL.GetUniformLocation(Shader.Handle, "transform"); - - GL.UniformMatrix2(TransformUniformLocation, false, ref Transform); - } - - private void SetupVertex() - { - VaoHandle = GL.GenVertexArray(); - VboHandle = GL.GenBuffer(); - - float[] Buffer = new float[] - { - -1, 1, 0, 0, - 1, 1, 1, 0, - -1, -1, 0, 1, - 1, -1, 1, 1 - }; - - IntPtr Length = new IntPtr(Buffer.Length * 4); - - GL.BindBuffer(BufferTarget.ArrayBuffer, VboHandle); - GL.BufferData(BufferTarget.ArrayBuffer, Length, Buffer, BufferUsageHint.StreamDraw); - GL.BindBuffer(BufferTarget.ArrayBuffer, 0); - - GL.BindVertexArray(VaoHandle); - - GL.EnableVertexAttribArray(0); - - GL.BindBuffer(BufferTarget.ArrayBuffer, VboHandle); - - GL.VertexAttribPointer(0, 2, VertexAttribPointerType.Float, false, 16, 0); - - GL.EnableVertexAttribArray(1); - - GL.BindBuffer(BufferTarget.ArrayBuffer, VboHandle); - - GL.VertexAttribPointer(1, 2, VertexAttribPointerType.Float, false, 16, 8); - } - private void SetupTexture(int Handle, int Width, int Height) { GL.BindTexture(TextureTarget.Texture2D, Handle); diff --git a/Ryujinx.Graphics/Ryujinx.Graphics.csproj b/Ryujinx.Graphics/Ryujinx.Graphics.csproj index d0fad10765..7d86cbe134 100644 --- a/Ryujinx.Graphics/Ryujinx.Graphics.csproj +++ b/Ryujinx.Graphics/Ryujinx.Graphics.csproj @@ -21,13 +21,4 @@ - - - GlFbVtxShader - - - GlFbFragShader - - - diff --git a/Ryujinx.HLE/OsHle/Services/Vi/NvFlinger.cs b/Ryujinx.HLE/OsHle/Services/Vi/NvFlinger.cs index 5307127be4..41f2916f4a 100644 --- a/Ryujinx.HLE/OsHle/Services/Vi/NvFlinger.cs +++ b/Ryujinx.HLE/OsHle/Services/Vi/NvFlinger.cs @@ -293,54 +293,17 @@ namespace Ryujinx.HLE.OsHle.Services.Android Rect Crop = BufferQueue[Slot].Crop; - int RealWidth = FbWidth; - int RealHeight = FbHeight; + bool FlipX = BufferQueue[Slot].Transform.HasFlag(HalTransform.FlipX); + bool FlipY = BufferQueue[Slot].Transform.HasFlag(HalTransform.FlipY); - float XSign = BufferQueue[Slot].Transform.HasFlag(HalTransform.FlipX) ? -1 : 1; - float YSign = BufferQueue[Slot].Transform.HasFlag(HalTransform.FlipY) ? -1 : 1; + //Rotation is being ignored - float ScaleX = 1; - float ScaleY = 1; + int Top = Crop.Top; + int Left = Crop.Left; + int Right = Crop.Right; + int Bottom = Crop.Bottom; - float OffsX = 0; - float OffsY = 0; - - if (Crop.Right != 0 && - Crop.Bottom != 0) - { - //Who knows if this is right, I was never good with math... - RealWidth = Crop.Right - Crop.Left; - RealHeight = Crop.Bottom - Crop.Top; - - if (BufferQueue[Slot].Transform.HasFlag(HalTransform.Rotate90)) - { - ScaleY = (float)FbHeight / RealHeight; - ScaleX = (float)FbWidth / RealWidth; - - OffsY = ((-(float)Crop.Left / Crop.Right) + ScaleX - 1) * -XSign; - OffsX = ((-(float)Crop.Top / Crop.Bottom) + ScaleY - 1) * -YSign; - } - else - { - ScaleX = (float)FbWidth / RealWidth; - ScaleY = (float)FbHeight / RealHeight; - - OffsX = ((-(float)Crop.Left / Crop.Right) + ScaleX - 1) * XSign; - OffsY = ((-(float)Crop.Top / Crop.Bottom) + ScaleY - 1) * -YSign; - } - } - - ScaleX *= XSign; - ScaleY *= YSign; - - float Rotate = 0; - - if (BufferQueue[Slot].Transform.HasFlag(HalTransform.Rotate90)) - { - Rotate = -MathF.PI * 0.5f; - } - - Renderer.QueueAction(() => Renderer.FrameBuffer.SetTransform(ScaleX, ScaleY, Rotate, OffsX, OffsY)); + Renderer.QueueAction(() => Renderer.FrameBuffer.SetTransform(FlipX, FlipY, Top, Left, Right, Bottom)); //TODO: Support double buffering here aswell, it is broken for GPU //frame buffers because it seems to be completely out of sync. From 39d36145ba30ce80d7cfeea757014e213f5e5287 Mon Sep 17 00:00:00 2001 From: Arthur Chen Date: Fri, 27 Jul 2018 00:46:05 +0800 Subject: [PATCH 21/34] update encoding for branch instruction (#305) --- ChocolArm64/AOpCodeTable.cs | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/ChocolArm64/AOpCodeTable.cs b/ChocolArm64/AOpCodeTable.cs index 689e039237..aa8220e38d 100644 --- a/ChocolArm64/AOpCodeTable.cs +++ b/ChocolArm64/AOpCodeTable.cs @@ -45,7 +45,7 @@ namespace ChocolArm64 SetA64("11101010xx0xxxxxxxxxxxxxxxxxxxxx", AInstEmit.Ands, typeof(AOpCodeAluRs)); SetA64("x0011010110xxxxx001010xxxxxxxxxx", AInstEmit.Asrv, typeof(AOpCodeAluRs)); SetA64("000101xxxxxxxxxxxxxxxxxxxxxxxxxx", AInstEmit.B, typeof(AOpCodeBImmAl)); - SetA64("01010100xxxxxxxxxxxxxxxxxxxxxxxx", AInstEmit.B_Cond, typeof(AOpCodeBImmCond)); + SetA64("01010100xxxxxxxxxxxxxxxxxxx0xxxx", AInstEmit.B_Cond, typeof(AOpCodeBImmCond)); SetA64("00110011000xxxxx0xxxxxxxxxxxxxxx", AInstEmit.Bfm, typeof(AOpCodeBfm)); SetA64("1011001101xxxxxxxxxxxxxxxxxxxxxx", AInstEmit.Bfm, typeof(AOpCodeBfm)); SetA64("00001010xx1xxxxx0xxxxxxxxxxxxxxx", AInstEmit.Bic, typeof(AOpCodeAluRs)); @@ -53,8 +53,8 @@ namespace ChocolArm64 SetA64("01101010xx1xxxxx0xxxxxxxxxxxxxxx", AInstEmit.Bics, typeof(AOpCodeAluRs)); SetA64("11101010xx1xxxxxxxxxxxxxxxxxxxxx", AInstEmit.Bics, typeof(AOpCodeAluRs)); SetA64("100101xxxxxxxxxxxxxxxxxxxxxxxxxx", AInstEmit.Bl, typeof(AOpCodeBImmAl)); - SetA64("11010110001xxxxx000000xxxxxxxxxx", AInstEmit.Blr, typeof(AOpCodeBReg)); - SetA64("11010110000xxxxx000000xxxxxxxxxx", AInstEmit.Br, typeof(AOpCodeBReg)); + SetA64("1101011000111111000000xxxxx00000", AInstEmit.Blr, typeof(AOpCodeBReg)); + SetA64("1101011000011111000000xxxxx00000", AInstEmit.Br, typeof(AOpCodeBReg)); SetA64("11010100001xxxxxxxxxxxxxxxx00000", AInstEmit.Brk, typeof(AOpCodeException)); SetA64("x0110101xxxxxxxxxxxxxxxxxxxxxxxx", AInstEmit.Cbnz, typeof(AOpCodeBImmCmp)); SetA64("x0110100xxxxxxxxxxxxxxxxxxxxxxxx", AInstEmit.Cbz, typeof(AOpCodeBImmCmp)); @@ -127,7 +127,7 @@ namespace ChocolArm64 SetA64("11111000100xxxxxxxxx00xxxxxxxxxx", AInstEmit.Pfrm, typeof(AOpCodeMemImm)); SetA64("11011000xxxxxxxxxxxxxxxxxxxxxxxx", AInstEmit.Pfrm, typeof(AOpCodeMemLit)); SetA64("x101101011000000000000xxxxxxxxxx", AInstEmit.Rbit, typeof(AOpCodeAlu)); - SetA64("11010110010xxxxx000000xxxxxxxxxx", AInstEmit.Ret, typeof(AOpCodeBReg)); + SetA64("1101011001011111000000xxxxx00000", AInstEmit.Ret, typeof(AOpCodeBReg)); SetA64("x101101011000000000001xxxxxxxxxx", AInstEmit.Rev16, typeof(AOpCodeAlu)); SetA64("x101101011000000000010xxxxxxxxxx", AInstEmit.Rev32, typeof(AOpCodeAlu)); SetA64("1101101011000000000011xxxxxxxxxx", AInstEmit.Rev64, typeof(AOpCodeAlu)); From 51605fafc04e7c34d354bd2dd8aeffc5d912f6a9 Mon Sep 17 00:00:00 2001 From: ReinUsesLisp Date: Thu, 26 Jul 2018 13:49:29 -0300 Subject: [PATCH 22/34] Avoid calling buffer binding when shader didn't change (#295) --- Ryujinx.Graphics/Gal/OpenGL/OGLShader.cs | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/Ryujinx.Graphics/Gal/OpenGL/OGLShader.cs b/Ryujinx.Graphics/Gal/OpenGL/OGLShader.cs index 37213d8ed1..3f3f23b8a2 100644 --- a/Ryujinx.Graphics/Gal/OpenGL/OGLShader.cs +++ b/Ryujinx.Graphics/Gal/OpenGL/OGLShader.cs @@ -257,7 +257,10 @@ namespace Ryujinx.Graphics.Gal.OpenGL GL.UseProgram(Handle); - BindUniformBuffers(Handle); + if (CurrentProgramHandle != Handle) + { + BindUniformBuffers(Handle); + } CurrentProgramHandle = Handle; } From 7a308d9e7305a864504ff9eae32c3927643dbf47 Mon Sep 17 00:00:00 2001 From: ReinUsesLisp Date: Sun, 29 Jul 2018 01:35:36 -0300 Subject: [PATCH 23/34] Window related changes (#308) * Use integer math for touch screen * Sleep polling thread * Rework host input * Add fullscreen with F11 or Alt+Enter * Address feedback --- Ryujinx/Config.cs | 102 +++++++++------ Ryujinx/Ui/GLScreen.cs | 188 +++++++-------------------- Ryujinx/Ui/JoyConController.cs | 226 +++++++++++++++++++++++++++++---- Ryujinx/Ui/JoyConKeyboard.cs | 66 +++++++++- 4 files changed, 376 insertions(+), 206 deletions(-) diff --git a/Ryujinx/Config.cs b/Ryujinx/Config.cs index 0f346122af..4ed35b3d7e 100644 --- a/Ryujinx/Config.cs +++ b/Ryujinx/Config.cs @@ -14,11 +14,6 @@ namespace Ryujinx public static JoyConKeyboard JoyConKeyboard { get; private set; } public static JoyConController JoyConController { get; private set; } - public static float GamePadDeadzone { get; private set; } - public static bool GamePadEnable { get; private set; } - public static int GamePadIndex { get; private set; } - public static float GamePadTriggerThreshold { get; private set; } - public static void Read(Logger Log) { string IniFolder = Path.GetDirectoryName(Assembly.GetEntryAssembly().Location); @@ -37,11 +32,6 @@ namespace Ryujinx Log.SetEnable(LogLevel.Warning, Convert.ToBoolean(Parser.Value("Logging_Enable_Warn"))); Log.SetEnable(LogLevel.Error, Convert.ToBoolean(Parser.Value("Logging_Enable_Error"))); - GamePadEnable = Convert.ToBoolean(Parser.Value("GamePad_Enable")); - GamePadIndex = Convert.ToInt32 (Parser.Value("GamePad_Index")); - GamePadDeadzone = (float)Convert.ToDouble (Parser.Value("GamePad_Deadzone"), CultureInfo.InvariantCulture); - GamePadTriggerThreshold = (float)Convert.ToDouble (Parser.Value("GamePad_Trigger_Threshold"), CultureInfo.InvariantCulture); - string[] FilteredLogClasses = Parser.Value("Logging_Filtered_Classes").Split(',', StringSplitOptions.RemoveEmptyEntries); //When the classes are specified on the list, we only @@ -70,9 +60,9 @@ namespace Ryujinx } } - JoyConKeyboard = new JoyConKeyboard - { - Left = new JoyConKeyboardLeft + JoyConKeyboard = new JoyConKeyboard( + + new JoyConKeyboardLeft { StickUp = Convert.ToInt16(Parser.Value("Controls_Left_JoyConKeyboard_Stick_Up")), StickDown = Convert.ToInt16(Parser.Value("Controls_Left_JoyConKeyboard_Stick_Down")), @@ -88,7 +78,7 @@ namespace Ryujinx ButtonZL = Convert.ToInt16(Parser.Value("Controls_Left_JoyConKeyboard_Button_ZL")) }, - Right = new JoyConKeyboardRight + new JoyConKeyboardRight { StickUp = Convert.ToInt16(Parser.Value("Controls_Right_JoyConKeyboard_Stick_Up")), StickDown = Convert.ToInt16(Parser.Value("Controls_Right_JoyConKeyboard_Stick_Down")), @@ -102,37 +92,69 @@ namespace Ryujinx ButtonPlus = Convert.ToInt16(Parser.Value("Controls_Right_JoyConKeyboard_Button_Plus")), ButtonR = Convert.ToInt16(Parser.Value("Controls_Right_JoyConKeyboard_Button_R")), ButtonZR = Convert.ToInt16(Parser.Value("Controls_Right_JoyConKeyboard_Button_ZR")) - } - }; + }); - JoyConController = new JoyConController - { - Left = new JoyConControllerLeft + JoyConController = new JoyConController( + + Convert.ToBoolean(Parser.Value("GamePad_Enable")), + Convert.ToInt32 (Parser.Value("GamePad_Index")), + (float)Convert.ToDouble (Parser.Value("GamePad_Deadzone"), CultureInfo.InvariantCulture), + (float)Convert.ToDouble (Parser.Value("GamePad_Trigger_Threshold"), CultureInfo.InvariantCulture), + + new JoyConControllerLeft { - Stick = Parser.Value("Controls_Left_JoyConController_Stick"), - StickButton = Parser.Value("Controls_Left_JoyConController_Stick_Button"), - DPadUp = Parser.Value("Controls_Left_JoyConController_DPad_Up"), - DPadDown = Parser.Value("Controls_Left_JoyConController_DPad_Down"), - DPadLeft = Parser.Value("Controls_Left_JoyConController_DPad_Left"), - DPadRight = Parser.Value("Controls_Left_JoyConController_DPad_Right"), - ButtonMinus = Parser.Value("Controls_Left_JoyConController_Button_Minus"), - ButtonL = Parser.Value("Controls_Left_JoyConController_Button_L"), - ButtonZL = Parser.Value("Controls_Left_JoyConController_Button_ZL") + Stick = ToID(Parser.Value("Controls_Left_JoyConController_Stick")), + StickButton = ToID(Parser.Value("Controls_Left_JoyConController_Stick_Button")), + DPadUp = ToID(Parser.Value("Controls_Left_JoyConController_DPad_Up")), + DPadDown = ToID(Parser.Value("Controls_Left_JoyConController_DPad_Down")), + DPadLeft = ToID(Parser.Value("Controls_Left_JoyConController_DPad_Left")), + DPadRight = ToID(Parser.Value("Controls_Left_JoyConController_DPad_Right")), + ButtonMinus = ToID(Parser.Value("Controls_Left_JoyConController_Button_Minus")), + ButtonL = ToID(Parser.Value("Controls_Left_JoyConController_Button_L")), + ButtonZL = ToID(Parser.Value("Controls_Left_JoyConController_Button_ZL")) }, - Right = new JoyConControllerRight + new JoyConControllerRight { - Stick = Parser.Value("Controls_Right_JoyConController_Stick"), - StickButton = Parser.Value("Controls_Right_JoyConController_Stick_Button"), - ButtonA = Parser.Value("Controls_Right_JoyConController_Button_A"), - ButtonB = Parser.Value("Controls_Right_JoyConController_Button_B"), - ButtonX = Parser.Value("Controls_Right_JoyConController_Button_X"), - ButtonY = Parser.Value("Controls_Right_JoyConController_Button_Y"), - ButtonPlus = Parser.Value("Controls_Right_JoyConController_Button_Plus"), - ButtonR = Parser.Value("Controls_Right_JoyConController_Button_R"), - ButtonZR = Parser.Value("Controls_Right_JoyConController_Button_ZR") - } - }; + Stick = ToID(Parser.Value("Controls_Right_JoyConController_Stick")), + StickButton = ToID(Parser.Value("Controls_Right_JoyConController_Stick_Button")), + ButtonA = ToID(Parser.Value("Controls_Right_JoyConController_Button_A")), + ButtonB = ToID(Parser.Value("Controls_Right_JoyConController_Button_B")), + ButtonX = ToID(Parser.Value("Controls_Right_JoyConController_Button_X")), + ButtonY = ToID(Parser.Value("Controls_Right_JoyConController_Button_Y")), + ButtonPlus = ToID(Parser.Value("Controls_Right_JoyConController_Button_Plus")), + ButtonR = ToID(Parser.Value("Controls_Right_JoyConController_Button_R")), + ButtonZR = ToID(Parser.Value("Controls_Right_JoyConController_Button_ZR")) + }); + } + + private static ControllerInputID ToID(string Key) + { + switch (Key.ToUpper()) + { + case "LSTICK": return ControllerInputID.LStick; + case "DPADUP": return ControllerInputID.DPadUp; + case "DPADDOWN": return ControllerInputID.DPadDown; + case "DPADLEFT": return ControllerInputID.DPadLeft; + case "DPADRIGHT": return ControllerInputID.DPadRight; + case "BACK": return ControllerInputID.Back; + case "LSHOULDER": return ControllerInputID.LShoulder; + case "LTRIGGER": return ControllerInputID.LTrigger; + + case "RSTICK": return ControllerInputID.RStick; + case "A": return ControllerInputID.A; + case "B": return ControllerInputID.B; + case "X": return ControllerInputID.X; + case "Y": return ControllerInputID.Y; + case "START": return ControllerInputID.Start; + case "RSHOULDER": return ControllerInputID.RShoulder; + case "RTRIGGER": return ControllerInputID.RTrigger; + + case "LJOYSTICK": return ControllerInputID.LJoystick; + case "RJOYSTICK": return ControllerInputID.RJoystick; + + default: return ControllerInputID.Invalid; + } } } diff --git a/Ryujinx/Ui/GLScreen.cs b/Ryujinx/Ui/GLScreen.cs index 9b5dda4f0c..dfc0b9a418 100644 --- a/Ryujinx/Ui/GLScreen.cs +++ b/Ryujinx/Ui/GLScreen.cs @@ -4,6 +4,7 @@ using OpenTK.Input; using Ryujinx.Graphics.Gal; using Ryujinx.HLE; using Ryujinx.HLE.Input; +using Ryujinx.UI.Input; using System; using System.Threading; @@ -16,9 +17,6 @@ namespace Ryujinx private const int TouchScreenWidth = 1280; private const int TouchScreenHeight = 720; - private const float TouchScreenRatioX = (float)TouchScreenWidth / TouchScreenHeight; - private const float TouchScreenRatioY = (float)TouchScreenHeight / TouchScreenWidth; - private const int TargetFPS = 60; private Switch Ns; @@ -49,10 +47,6 @@ namespace Ryujinx Location = new Point( (DisplayDevice.Default.Width / 2) - (Width / 2), (DisplayDevice.Default.Height / 2) - (Height / 2)); - - ResizeEvent = false; - - TitleEvent = false; } private void RenderLoop() @@ -127,60 +121,9 @@ namespace Ryujinx Title = NewTitle; } } - } - } - - private bool IsGamePadButtonPressedFromString(GamePadState GamePad, string Button) - { - if (Button.ToUpper() == "LTRIGGER" || Button.ToUpper() == "RTRIGGER") - { - return GetGamePadTriggerFromString(GamePad, Button) >= Config.GamePadTriggerThreshold; - } - else - { - return (GetGamePadButtonFromString(GamePad, Button) == ButtonState.Pressed); - } - } - private ButtonState GetGamePadButtonFromString(GamePadState GamePad, string Button) - { - switch (Button.ToUpper()) - { - case "A": return GamePad.Buttons.A; - case "B": return GamePad.Buttons.B; - case "X": return GamePad.Buttons.X; - case "Y": return GamePad.Buttons.Y; - case "LSTICK": return GamePad.Buttons.LeftStick; - case "RSTICK": return GamePad.Buttons.RightStick; - case "LSHOULDER": return GamePad.Buttons.LeftShoulder; - case "RSHOULDER": return GamePad.Buttons.RightShoulder; - case "DPADUP": return GamePad.DPad.Up; - case "DPADDOWN": return GamePad.DPad.Down; - case "DPADLEFT": return GamePad.DPad.Left; - case "DPADRIGHT": return GamePad.DPad.Right; - case "START": return GamePad.Buttons.Start; - case "BACK": return GamePad.Buttons.Back; - default: throw new ArgumentException(); - } - } - - private float GetGamePadTriggerFromString(GamePadState GamePad, string Trigger) - { - switch (Trigger.ToUpper()) - { - case "LTRIGGER": return GamePad.Triggers.Left; - case "RTRIGGER": return GamePad.Triggers.Right; - default: throw new ArgumentException(); - } - } - - private Vector2 GetJoystickAxisFromString(GamePadState GamePad, string Joystick) - { - switch (Joystick.ToUpper()) - { - case "LJOYSTICK": return GamePad.ThumbSticks.Left; - case "RJOYSTICK": return new Vector2(-GamePad.ThumbSticks.Right.Y, -GamePad.ThumbSticks.Right.X); - default: throw new ArgumentException(); + //Polling becomes expensive if it's not slept + Thread.Sleep(1); } } @@ -190,95 +133,37 @@ namespace Ryujinx HidJoystickPosition LeftJoystick; HidJoystickPosition RightJoystick; - int LeftJoystickDX = 0; - int LeftJoystickDY = 0; - int RightJoystickDX = 0; - int RightJoystickDY = 0; - float AnalogStickDeadzone = Config.GamePadDeadzone; + int LeftJoystickDX = 0; + int LeftJoystickDY = 0; + int RightJoystickDX = 0; + int RightJoystickDY = 0; //Keyboard Input if (Keyboard.HasValue) { KeyboardState Keyboard = this.Keyboard.Value; - if (Keyboard[Key.Escape]) this.Exit(); + CurrentButton = Config.JoyConKeyboard.GetButtons(Keyboard); - //LeftJoystick - if (Keyboard[(Key)Config.JoyConKeyboard.Left.StickUp]) LeftJoystickDY = short.MaxValue; - if (Keyboard[(Key)Config.JoyConKeyboard.Left.StickDown]) LeftJoystickDY = -short.MaxValue; - if (Keyboard[(Key)Config.JoyConKeyboard.Left.StickLeft]) LeftJoystickDX = -short.MaxValue; - if (Keyboard[(Key)Config.JoyConKeyboard.Left.StickRight]) LeftJoystickDX = short.MaxValue; + (LeftJoystickDX, LeftJoystickDY) = Config.JoyConKeyboard.GetLeftStick(Keyboard); - //LeftButtons - if (Keyboard[(Key)Config.JoyConKeyboard.Left.StickButton]) CurrentButton |= HidControllerButtons.KEY_LSTICK; - if (Keyboard[(Key)Config.JoyConKeyboard.Left.DPadUp]) CurrentButton |= HidControllerButtons.KEY_DUP; - if (Keyboard[(Key)Config.JoyConKeyboard.Left.DPadDown]) CurrentButton |= HidControllerButtons.KEY_DDOWN; - if (Keyboard[(Key)Config.JoyConKeyboard.Left.DPadLeft]) CurrentButton |= HidControllerButtons.KEY_DLEFT; - if (Keyboard[(Key)Config.JoyConKeyboard.Left.DPadRight]) CurrentButton |= HidControllerButtons.KEY_DRIGHT; - if (Keyboard[(Key)Config.JoyConKeyboard.Left.ButtonMinus]) CurrentButton |= HidControllerButtons.KEY_MINUS; - if (Keyboard[(Key)Config.JoyConKeyboard.Left.ButtonL]) CurrentButton |= HidControllerButtons.KEY_L; - if (Keyboard[(Key)Config.JoyConKeyboard.Left.ButtonZL]) CurrentButton |= HidControllerButtons.KEY_ZL; - - //RightJoystick - if (Keyboard[(Key)Config.JoyConKeyboard.Right.StickUp]) RightJoystickDY = short.MaxValue; - if (Keyboard[(Key)Config.JoyConKeyboard.Right.StickDown]) RightJoystickDY = -short.MaxValue; - if (Keyboard[(Key)Config.JoyConKeyboard.Right.StickLeft]) RightJoystickDX = -short.MaxValue; - if (Keyboard[(Key)Config.JoyConKeyboard.Right.StickRight]) RightJoystickDX = short.MaxValue; - - //RightButtons - if (Keyboard[(Key)Config.JoyConKeyboard.Right.StickButton]) CurrentButton |= HidControllerButtons.KEY_RSTICK; - if (Keyboard[(Key)Config.JoyConKeyboard.Right.ButtonA]) CurrentButton |= HidControllerButtons.KEY_A; - if (Keyboard[(Key)Config.JoyConKeyboard.Right.ButtonB]) CurrentButton |= HidControllerButtons.KEY_B; - if (Keyboard[(Key)Config.JoyConKeyboard.Right.ButtonX]) CurrentButton |= HidControllerButtons.KEY_X; - if (Keyboard[(Key)Config.JoyConKeyboard.Right.ButtonY]) CurrentButton |= HidControllerButtons.KEY_Y; - if (Keyboard[(Key)Config.JoyConKeyboard.Right.ButtonPlus]) CurrentButton |= HidControllerButtons.KEY_PLUS; - if (Keyboard[(Key)Config.JoyConKeyboard.Right.ButtonR]) CurrentButton |= HidControllerButtons.KEY_R; - if (Keyboard[(Key)Config.JoyConKeyboard.Right.ButtonZR]) CurrentButton |= HidControllerButtons.KEY_ZR; + (RightJoystickDX, RightJoystickDY) = Config.JoyConKeyboard.GetRightStick(Keyboard); } //Controller Input - if (Config.GamePadEnable) + CurrentButton |= Config.JoyConController.GetButtons(); + + //Keyboard has priority stick-wise + if (LeftJoystickDX == 0 && LeftJoystickDY == 0) { - GamePadState GamePad = OpenTK.Input.GamePad.GetState(Config.GamePadIndex); - //LeftButtons - if (IsGamePadButtonPressedFromString(GamePad, Config.JoyConController.Left.DPadUp)) CurrentButton |= HidControllerButtons.KEY_DUP; - if (IsGamePadButtonPressedFromString(GamePad, Config.JoyConController.Left.DPadDown)) CurrentButton |= HidControllerButtons.KEY_DDOWN; - if (IsGamePadButtonPressedFromString(GamePad, Config.JoyConController.Left.DPadLeft)) CurrentButton |= HidControllerButtons.KEY_DLEFT; - if (IsGamePadButtonPressedFromString(GamePad, Config.JoyConController.Left.DPadRight)) CurrentButton |= HidControllerButtons.KEY_DRIGHT; - if (IsGamePadButtonPressedFromString(GamePad, Config.JoyConController.Left.StickButton)) CurrentButton |= HidControllerButtons.KEY_LSTICK; - if (IsGamePadButtonPressedFromString(GamePad, Config.JoyConController.Left.ButtonMinus)) CurrentButton |= HidControllerButtons.KEY_MINUS; - if (IsGamePadButtonPressedFromString(GamePad, Config.JoyConController.Left.ButtonL)) CurrentButton |= HidControllerButtons.KEY_L; - if (IsGamePadButtonPressedFromString(GamePad, Config.JoyConController.Left.ButtonZL)) CurrentButton |= HidControllerButtons.KEY_ZL; - - //RightButtons - if (IsGamePadButtonPressedFromString(GamePad, Config.JoyConController.Right.ButtonA)) CurrentButton |= HidControllerButtons.KEY_A; - if (IsGamePadButtonPressedFromString(GamePad, Config.JoyConController.Right.ButtonB)) CurrentButton |= HidControllerButtons.KEY_B; - if (IsGamePadButtonPressedFromString(GamePad, Config.JoyConController.Right.ButtonX)) CurrentButton |= HidControllerButtons.KEY_X; - if (IsGamePadButtonPressedFromString(GamePad, Config.JoyConController.Right.ButtonY)) CurrentButton |= HidControllerButtons.KEY_Y; - if (IsGamePadButtonPressedFromString(GamePad, Config.JoyConController.Right.StickButton)) CurrentButton |= HidControllerButtons.KEY_RSTICK; - if (IsGamePadButtonPressedFromString(GamePad, Config.JoyConController.Right.ButtonPlus)) CurrentButton |= HidControllerButtons.KEY_PLUS; - if (IsGamePadButtonPressedFromString(GamePad, Config.JoyConController.Right.ButtonR)) CurrentButton |= HidControllerButtons.KEY_R; - if (IsGamePadButtonPressedFromString(GamePad, Config.JoyConController.Right.ButtonZR)) CurrentButton |= HidControllerButtons.KEY_ZR; - - //LeftJoystick - if (GetJoystickAxisFromString(GamePad, Config.JoyConController.Left.Stick).X >= AnalogStickDeadzone - || GetJoystickAxisFromString(GamePad, Config.JoyConController.Left.Stick).X <= -AnalogStickDeadzone) - LeftJoystickDX = (int)(GetJoystickAxisFromString(GamePad, Config.JoyConController.Left.Stick).X * short.MaxValue); - - if (GetJoystickAxisFromString(GamePad, Config.JoyConController.Left.Stick).Y >= AnalogStickDeadzone - || GetJoystickAxisFromString(GamePad, Config.JoyConController.Left.Stick).Y <= -AnalogStickDeadzone) - LeftJoystickDY = (int)(GetJoystickAxisFromString(GamePad, Config.JoyConController.Left.Stick).Y * short.MaxValue); - - //RightJoystick - if (GetJoystickAxisFromString(GamePad, Config.JoyConController.Right.Stick).X >= AnalogStickDeadzone - || GetJoystickAxisFromString(GamePad, Config.JoyConController.Right.Stick).X <= -AnalogStickDeadzone) - RightJoystickDX = (int)(GetJoystickAxisFromString(GamePad, Config.JoyConController.Right.Stick).X * short.MaxValue); - - if (GetJoystickAxisFromString(GamePad, Config.JoyConController.Right.Stick).Y >= AnalogStickDeadzone - || GetJoystickAxisFromString(GamePad, Config.JoyConController.Right.Stick).Y <= -AnalogStickDeadzone) - RightJoystickDY = (int)(GetJoystickAxisFromString(GamePad, Config.JoyConController.Right.Stick).Y * short.MaxValue); + (LeftJoystickDX, LeftJoystickDY) = Config.JoyConController.GetLeftStick(); } + if (RightJoystickDX == 0 && RightJoystickDY == 0) + { + (RightJoystickDX, RightJoystickDY) = Config.JoyConController.GetRightStick(); + } + LeftJoystick = new HidJoystickPosition { DX = LeftJoystickDX, @@ -302,13 +187,13 @@ namespace Ryujinx int ScrnWidth = Width; int ScrnHeight = Height; - if (Width > Height * TouchScreenRatioX) + if (Width > (Height * TouchScreenWidth) / TouchScreenHeight) { - ScrnWidth = (int)(Height * TouchScreenRatioX); + ScrnWidth = (Height * TouchScreenWidth) / TouchScreenHeight; } else { - ScrnHeight = (int)(Width * TouchScreenRatioY); + ScrnHeight = (Width * TouchScreenHeight) / TouchScreenWidth; } int StartX = (Width - ScrnWidth) >> 1; @@ -325,8 +210,8 @@ namespace Ryujinx int ScrnMouseX = Mouse.X - StartX; int ScrnMouseY = Mouse.Y - StartY; - int MX = (int)(((float)ScrnMouseX / ScrnWidth) * TouchScreenWidth); - int MY = (int)(((float)ScrnMouseY / ScrnHeight) * TouchScreenHeight); + int MX = (ScrnMouseX * TouchScreenWidth) / ScrnWidth; + int MY = (ScrnMouseY * TouchScreenHeight) / ScrnHeight; HidTouchPoint CurrentPoint = new HidTouchPoint { @@ -397,6 +282,29 @@ namespace Ryujinx protected override void OnKeyDown(KeyboardKeyEventArgs e) { + bool ToggleFullscreen = e.Key == Key.F11 || + (e.Modifiers.HasFlag(KeyModifiers.Alt) && e.Key == Key.Enter); + + if (WindowState == WindowState.Fullscreen) + { + if (e.Key == Key.Escape || ToggleFullscreen) + { + WindowState = WindowState.Normal; + } + } + else + { + if (e.Key == Key.Escape) + { + Exit(); + } + + if (ToggleFullscreen) + { + WindowState = WindowState.Fullscreen; + } + } + Keyboard = e.Keyboard; } diff --git a/Ryujinx/Ui/JoyConController.cs b/Ryujinx/Ui/JoyConController.cs index e525017d3e..aac8efd7c4 100644 --- a/Ryujinx/Ui/JoyConController.cs +++ b/Ryujinx/Ui/JoyConController.cs @@ -1,38 +1,216 @@ -using System; -using System.Collections.Generic; -using System.Text; +using OpenTK; +using OpenTK.Input; +using Ryujinx.HLE.Input; +using System; namespace Ryujinx.UI.Input { + public enum ControllerInputID + { + Invalid, + + LStick, + DPadUp, + DPadDown, + DPadLeft, + DPadRight, + Back, + LShoulder, + + RStick, + A, + B, + X, + Y, + Start, + RShoulder, + + LTrigger, + RTrigger, + + LJoystick, + RJoystick + } + public struct JoyConControllerLeft { - public string Stick; - public string StickButton; - public string DPadUp; - public string DPadDown; - public string DPadLeft; - public string DPadRight; - public string ButtonMinus; - public string ButtonL; - public string ButtonZL; + public ControllerInputID Stick; + public ControllerInputID StickButton; + public ControllerInputID DPadUp; + public ControllerInputID DPadDown; + public ControllerInputID DPadLeft; + public ControllerInputID DPadRight; + public ControllerInputID ButtonMinus; + public ControllerInputID ButtonL; + public ControllerInputID ButtonZL; } public struct JoyConControllerRight { - public string Stick; - public string StickButton; - public string ButtonA; - public string ButtonB; - public string ButtonX; - public string ButtonY; - public string ButtonPlus; - public string ButtonR; - public string ButtonZR; + public ControllerInputID Stick; + public ControllerInputID StickButton; + public ControllerInputID ButtonA; + public ControllerInputID ButtonB; + public ControllerInputID ButtonX; + public ControllerInputID ButtonY; + public ControllerInputID ButtonPlus; + public ControllerInputID ButtonR; + public ControllerInputID ButtonZR; } - public struct JoyConController + public class JoyConController { - public JoyConControllerLeft Left; - public JoyConControllerRight Right; + public bool Enabled { private set; get; } + public int Index { private set; get; } + public float Deadzone { private set; get; } + public float TriggerThreshold { private set; get; } + + public JoyConControllerLeft Left { private set; get; } + public JoyConControllerRight Right { private set; get; } + + public JoyConController( + bool Enabled, + int Index, + float Deadzone, + float TriggerThreshold, + JoyConControllerLeft Left, + JoyConControllerRight Right) + { + this.Enabled = Enabled; + this.Index = Index; + this.Deadzone = Deadzone; + this.TriggerThreshold = TriggerThreshold; + this.Left = Left; + this.Right = Right; + + //Unmapped controllers are problematic, skip them + if (GamePad.GetName(Index) == "Unmapped Controller") + { + this.Enabled = false; + } + } + + public HidControllerButtons GetButtons() + { + if (!Enabled) + { + return 0; + } + + GamePadState GpState = GamePad.GetState(Index); + + HidControllerButtons Buttons = 0; + + if (IsPressed(GpState, Left.DPadUp)) Buttons |= HidControllerButtons.KEY_DUP; + if (IsPressed(GpState, Left.DPadDown)) Buttons |= HidControllerButtons.KEY_DDOWN; + if (IsPressed(GpState, Left.DPadLeft)) Buttons |= HidControllerButtons.KEY_DLEFT; + if (IsPressed(GpState, Left.DPadRight)) Buttons |= HidControllerButtons.KEY_DRIGHT; + if (IsPressed(GpState, Left.StickButton)) Buttons |= HidControllerButtons.KEY_LSTICK; + if (IsPressed(GpState, Left.ButtonMinus)) Buttons |= HidControllerButtons.KEY_MINUS; + if (IsPressed(GpState, Left.ButtonL)) Buttons |= HidControllerButtons.KEY_L; + if (IsPressed(GpState, Left.ButtonZL)) Buttons |= HidControllerButtons.KEY_ZL; + + if (IsPressed(GpState, Right.ButtonA)) Buttons |= HidControllerButtons.KEY_A; + if (IsPressed(GpState, Right.ButtonB)) Buttons |= HidControllerButtons.KEY_B; + if (IsPressed(GpState, Right.ButtonX)) Buttons |= HidControllerButtons.KEY_X; + if (IsPressed(GpState, Right.ButtonY)) Buttons |= HidControllerButtons.KEY_Y; + if (IsPressed(GpState, Right.StickButton)) Buttons |= HidControllerButtons.KEY_RSTICK; + if (IsPressed(GpState, Right.ButtonPlus)) Buttons |= HidControllerButtons.KEY_PLUS; + if (IsPressed(GpState, Right.ButtonR)) Buttons |= HidControllerButtons.KEY_R; + if (IsPressed(GpState, Right.ButtonZR)) Buttons |= HidControllerButtons.KEY_ZR; + + return Buttons; + } + + public (short, short) GetLeftStick() + { + if (!Enabled) + { + return (0, 0); + } + + return GetStick(Left.Stick); + } + + public (short, short) GetRightStick() + { + if (!Enabled) + { + return (0, 0); + } + + return GetStick(Right.Stick); + } + + private (short, short) GetStick(ControllerInputID Joystick) + { + GamePadState GpState = GamePad.GetState(Index); + + switch (Joystick) + { + case ControllerInputID.LJoystick: + return ApplyDeadzone(GpState.ThumbSticks.Left); + + case ControllerInputID.RJoystick: + return ApplyDeadzone(GpState.ThumbSticks.Right); + + default: + return (0, 0); + } + } + + private (short, short) ApplyDeadzone(Vector2 Axis) + { + return (ClampAxis(MathF.Abs(Axis.X) > Deadzone ? Axis.X : 0f), + ClampAxis(MathF.Abs(Axis.Y) > Deadzone ? Axis.Y : 0f)); + } + + private static short ClampAxis(float Value) + { + if (Value <= -short.MaxValue) + { + return -short.MaxValue; + } + else + { + return (short)(Value * short.MaxValue); + } + } + + private bool IsPressed(GamePadState GpState, ControllerInputID Button) + { + switch (Button) + { + case ControllerInputID.A: return GpState.Buttons.A == ButtonState.Pressed; + case ControllerInputID.B: return GpState.Buttons.B == ButtonState.Pressed; + case ControllerInputID.X: return GpState.Buttons.X == ButtonState.Pressed; + case ControllerInputID.Y: return GpState.Buttons.Y == ButtonState.Pressed; + case ControllerInputID.LStick: return GpState.Buttons.LeftStick == ButtonState.Pressed; + case ControllerInputID.RStick: return GpState.Buttons.RightStick == ButtonState.Pressed; + case ControllerInputID.LShoulder: return GpState.Buttons.LeftShoulder == ButtonState.Pressed; + case ControllerInputID.RShoulder: return GpState.Buttons.RightShoulder == ButtonState.Pressed; + case ControllerInputID.DPadUp: return GpState.DPad.Up == ButtonState.Pressed; + case ControllerInputID.DPadDown: return GpState.DPad.Down == ButtonState.Pressed; + case ControllerInputID.DPadLeft: return GpState.DPad.Left == ButtonState.Pressed; + case ControllerInputID.DPadRight: return GpState.DPad.Right == ButtonState.Pressed; + case ControllerInputID.Start: return GpState.Buttons.Start == ButtonState.Pressed; + case ControllerInputID.Back: return GpState.Buttons.Back == ButtonState.Pressed; + + case ControllerInputID.LTrigger: return GpState.Triggers.Left >= TriggerThreshold; + case ControllerInputID.RTrigger: return GpState.Triggers.Right >= TriggerThreshold; + + //Using thumbsticks as buttons is not common, but it would be nice not to ignore them + case ControllerInputID.LJoystick: + return GpState.ThumbSticks.Left.X >= Deadzone || + GpState.ThumbSticks.Left.Y >= Deadzone; + + case ControllerInputID.RJoystick: + return GpState.ThumbSticks.Right.X >= Deadzone || + GpState.ThumbSticks.Right.Y >= Deadzone; + + default: + return false; + } + } } } diff --git a/Ryujinx/Ui/JoyConKeyboard.cs b/Ryujinx/Ui/JoyConKeyboard.cs index b329d9ecd1..ea9645539a 100644 --- a/Ryujinx/Ui/JoyConKeyboard.cs +++ b/Ryujinx/Ui/JoyConKeyboard.cs @@ -1,3 +1,6 @@ +using OpenTK.Input; +using Ryujinx.HLE.Input; + namespace Ryujinx.UI.Input { public struct JoyConKeyboardLeft @@ -32,9 +35,68 @@ namespace Ryujinx.UI.Input public int ButtonZR; } - public struct JoyConKeyboard + public class JoyConKeyboard { - public JoyConKeyboardLeft Left; + public JoyConKeyboardLeft Left; public JoyConKeyboardRight Right; + + public JoyConKeyboard( + JoyConKeyboardLeft Left, + JoyConKeyboardRight Right) + { + this.Left = Left; + this.Right = Right; + } + + public HidControllerButtons GetButtons(KeyboardState Keyboard) + { + HidControllerButtons Buttons = 0; + + if (Keyboard[(Key)Left.StickButton]) Buttons |= HidControllerButtons.KEY_LSTICK; + if (Keyboard[(Key)Left.DPadUp]) Buttons |= HidControllerButtons.KEY_DUP; + if (Keyboard[(Key)Left.DPadDown]) Buttons |= HidControllerButtons.KEY_DDOWN; + if (Keyboard[(Key)Left.DPadLeft]) Buttons |= HidControllerButtons.KEY_DLEFT; + if (Keyboard[(Key)Left.DPadRight]) Buttons |= HidControllerButtons.KEY_DRIGHT; + if (Keyboard[(Key)Left.ButtonMinus]) Buttons |= HidControllerButtons.KEY_MINUS; + if (Keyboard[(Key)Left.ButtonL]) Buttons |= HidControllerButtons.KEY_L; + if (Keyboard[(Key)Left.ButtonZL]) Buttons |= HidControllerButtons.KEY_ZL; + + if (Keyboard[(Key)Right.StickButton]) Buttons |= HidControllerButtons.KEY_RSTICK; + if (Keyboard[(Key)Right.ButtonA]) Buttons |= HidControllerButtons.KEY_A; + if (Keyboard[(Key)Right.ButtonB]) Buttons |= HidControllerButtons.KEY_B; + if (Keyboard[(Key)Right.ButtonX]) Buttons |= HidControllerButtons.KEY_X; + if (Keyboard[(Key)Right.ButtonY]) Buttons |= HidControllerButtons.KEY_Y; + if (Keyboard[(Key)Right.ButtonPlus]) Buttons |= HidControllerButtons.KEY_PLUS; + if (Keyboard[(Key)Right.ButtonR]) Buttons |= HidControllerButtons.KEY_R; + if (Keyboard[(Key)Right.ButtonZR]) Buttons |= HidControllerButtons.KEY_ZR; + + return Buttons; + } + + public (short, short) GetLeftStick(KeyboardState Keyboard) + { + short DX = 0; + short DY = 0; + + if (Keyboard[(Key)Left.StickUp]) DY = short.MaxValue; + if (Keyboard[(Key)Left.StickDown]) DY = -short.MaxValue; + if (Keyboard[(Key)Left.StickLeft]) DX = -short.MaxValue; + if (Keyboard[(Key)Left.StickRight]) DX = short.MaxValue; + + return (DX, DY); + } + + public (short, short) GetRightStick(KeyboardState Keyboard) + { + short DX = 0; + short DY = 0; + + if (Keyboard[(Key)Right.StickUp]) DY = short.MaxValue; + if (Keyboard[(Key)Right.StickDown]) DY = -short.MaxValue; + if (Keyboard[(Key)Right.StickLeft]) DX = -short.MaxValue; + if (Keyboard[(Key)Right.StickRight]) DX = short.MaxValue; + + return (DX, DY); + } } } From fdda67d4762c6a283ff0f1f76c02a02b21d9b119 Mon Sep 17 00:00:00 2001 From: gdkchan Date: Sun, 29 Jul 2018 01:36:29 -0300 Subject: [PATCH 24/34] Some fix to IRequest on NIFM, support sending objects to services (#294) --- Ryujinx.HLE/OsHle/Ipc/IpcHandleDesc.cs | 14 ++++---- Ryujinx.HLE/OsHle/Ipc/IpcMessage.cs | 4 +-- Ryujinx.HLE/OsHle/Services/IpcService.cs | 39 ++++++++++++++++++--- Ryujinx.HLE/OsHle/Services/Nifm/IRequest.cs | 18 +++++----- 4 files changed, 54 insertions(+), 21 deletions(-) diff --git a/Ryujinx.HLE/OsHle/Ipc/IpcHandleDesc.cs b/Ryujinx.HLE/OsHle/Ipc/IpcHandleDesc.cs index 953cac7643..609cc6e05f 100644 --- a/Ryujinx.HLE/OsHle/Ipc/IpcHandleDesc.cs +++ b/Ryujinx.HLE/OsHle/Ipc/IpcHandleDesc.cs @@ -47,13 +47,15 @@ namespace Ryujinx.HLE.OsHle.Ipc HasPId = true; } - public static IpcHandleDesc MakeCopy(int Handle) => new IpcHandleDesc( - new int[] { Handle }, - new int[0]); + public static IpcHandleDesc MakeCopy(params int[] Handles) + { + return new IpcHandleDesc(Handles, new int[0]); + } - public static IpcHandleDesc MakeMove(int Handle) => new IpcHandleDesc( - new int[0], - new int[] { Handle }); + public static IpcHandleDesc MakeMove(params int[] Handles) + { + return new IpcHandleDesc(new int[0], Handles); + } public byte[] GetBytes() { diff --git a/Ryujinx.HLE/OsHle/Ipc/IpcMessage.cs b/Ryujinx.HLE/OsHle/Ipc/IpcMessage.cs index 4e648aec97..0a64a16426 100644 --- a/Ryujinx.HLE/OsHle/Ipc/IpcMessage.cs +++ b/Ryujinx.HLE/OsHle/Ipc/IpcMessage.cs @@ -15,7 +15,7 @@ namespace Ryujinx.HLE.OsHle.Ipc public List ExchangeBuff { get; private set; } public List RecvListBuff { get; private set; } - public List ResponseObjIds { get; private set; } + public List ObjectIds { get; private set; } public byte[] RawData { get; set; } @@ -27,7 +27,7 @@ namespace Ryujinx.HLE.OsHle.Ipc ExchangeBuff = new List(); RecvListBuff = new List(); - ResponseObjIds = new List(); + ObjectIds = new List(); } public IpcMessage(byte[] Data, long CmdPtr) : this() diff --git a/Ryujinx.HLE/OsHle/Services/IpcService.cs b/Ryujinx.HLE/OsHle/Services/IpcService.cs index 25fd56fec0..3c1a136f1b 100644 --- a/Ryujinx.HLE/OsHle/Services/IpcService.cs +++ b/Ryujinx.HLE/OsHle/Services/IpcService.cs @@ -50,9 +50,18 @@ namespace Ryujinx.HLE.OsHle.Services int DomainWord0 = Context.RequestData.ReadInt32(); int DomainObjId = Context.RequestData.ReadInt32(); - long Padding = Context.RequestData.ReadInt64(); + int DomainCmd = (DomainWord0 >> 0) & 0xff; + int InputObjCount = (DomainWord0 >> 8) & 0xff; + int DataPayloadSize = (DomainWord0 >> 16) & 0xffff; - int DomainCmd = DomainWord0 & 0xff; + Context.RequestData.BaseStream.Seek(0x10 + DataPayloadSize, SeekOrigin.Begin); + + for (int Index = 0; Index < InputObjCount; Index++) + { + Context.Request.ObjectIds.Add(Context.RequestData.ReadInt32()); + } + + Context.RequestData.BaseStream.Seek(0x10, SeekOrigin.Begin); if (DomainCmd == 1) { @@ -88,14 +97,14 @@ namespace Ryujinx.HLE.OsHle.Services if (IsDomain) { - foreach (int Id in Context.Response.ResponseObjIds) + foreach (int Id in Context.Response.ObjectIds) { Context.ResponseData.Write(Id); } Context.ResponseData.BaseStream.Seek(0, SeekOrigin.Begin); - Context.ResponseData.Write(Context.Response.ResponseObjIds.Count); + Context.ResponseData.Write(Context.Response.ObjectIds.Count); } Context.ResponseData.BaseStream.Seek(IsDomain ? 0x10 : 0, SeekOrigin.Begin); @@ -117,7 +126,7 @@ namespace Ryujinx.HLE.OsHle.Services if (Service.IsDomain) { - Context.Response.ResponseObjIds.Add(Service.Add(Obj)); + Context.Response.ObjectIds.Add(Service.Add(Obj)); } else { @@ -129,6 +138,26 @@ namespace Ryujinx.HLE.OsHle.Services } } + protected static T GetObject(ServiceCtx Context, int Index) where T : IpcService + { + IpcService Service = Context.Session.Service; + + if (!Service.IsDomain) + { + int Handle = Context.Request.HandleDesc.ToMove[Index]; + + KSession Session = Context.Process.HandleTable.GetData(Handle); + + return Session?.Service is T ? (T)Session.Service : null; + } + + int ObjId = Context.Request.ObjectIds[Index]; + + IIpcService Obj = Service.GetObject(ObjId); + + return Obj is T ? (T)Obj : null; + } + private int Add(IIpcService Obj) { return DomainObjects.Add(Obj); diff --git a/Ryujinx.HLE/OsHle/Services/Nifm/IRequest.cs b/Ryujinx.HLE/OsHle/Services/Nifm/IRequest.cs index c8c679c4c5..2056187ddd 100644 --- a/Ryujinx.HLE/OsHle/Services/Nifm/IRequest.cs +++ b/Ryujinx.HLE/OsHle/Services/Nifm/IRequest.cs @@ -12,7 +12,8 @@ namespace Ryujinx.HLE.OsHle.Services.Nifm public override IReadOnlyDictionary Commands => m_Commands; - private KEvent Event; + private KEvent Event0; + private KEvent Event1; public IRequest() { @@ -26,12 +27,13 @@ namespace Ryujinx.HLE.OsHle.Services.Nifm { 11, SetConnectionConfirmationOption } }; - Event = new KEvent(); + Event0 = new KEvent(); + Event1 = new KEvent(); } public long GetRequestState(ServiceCtx Context) { - Context.ResponseData.Write(0); + Context.ResponseData.Write(1); Context.Ns.Log.PrintStub(LogClass.ServiceNifm, "Stubbed."); @@ -45,13 +47,12 @@ namespace Ryujinx.HLE.OsHle.Services.Nifm return 0; } - //GetSystemEventReadableHandles() -> (KObject, KObject) public long GetSystemEventReadableHandles(ServiceCtx Context) { - //FIXME: Is this supposed to return 2 events? - int Handle = Context.Process.HandleTable.OpenHandle(Event); + int Handle0 = Context.Process.HandleTable.OpenHandle(Event0); + int Handle1 = Context.Process.HandleTable.OpenHandle(Event1); - Context.Response.HandleDesc = IpcHandleDesc.MakeMove(Handle); + Context.Response.HandleDesc = IpcHandleDesc.MakeCopy(Handle0, Handle1); return 0; } @@ -86,7 +87,8 @@ namespace Ryujinx.HLE.OsHle.Services.Nifm { if (Disposing) { - Event.Dispose(); + Event0.Dispose(); + Event1.Dispose(); } } } From 3208173620a0003d09d8e756729ca905ff14c47f Mon Sep 17 00:00:00 2001 From: ReinUsesLisp Date: Sun, 29 Jul 2018 01:39:15 -0300 Subject: [PATCH 25/34] Cache changes (#302) * Skip repeated cache tests between same sync * Skip some checks for regions where just one resource is resident * Dehardcode residency page size * Some cleanup --- ChocolArm64/Memory/AMemory.cs | 10 +-- Ryujinx.HLE/Gpu/Engines/NvGpuEngine3d.cs | 34 ++++++++- Ryujinx.HLE/Gpu/Memory/NvGpuVmmCache.cs | 93 +++++++++++++++++++++++- 3 files changed, 127 insertions(+), 10 deletions(-) diff --git a/ChocolArm64/Memory/AMemory.cs b/ChocolArm64/Memory/AMemory.cs index 054277b295..e969cca5fe 100644 --- a/ChocolArm64/Memory/AMemory.cs +++ b/ChocolArm64/Memory/AMemory.cs @@ -160,23 +160,23 @@ namespace ChocolArm64.Memory return HostPageSize; } - public bool[] IsRegionModified(long Position, long Size) + public (bool[], long) IsRegionModified(long Position, long Size) { if (!RuntimeInformation.IsOSPlatform(OSPlatform.Windows)) { - return null; + return (null, 0); } long EndPos = Position + Size; if ((ulong)EndPos < (ulong)Position) { - return null; + return (null, 0); } if ((ulong)EndPos > AMemoryMgr.RamSize) { - return null; + return (null, 0); } IntPtr MemAddress = new IntPtr(RamPtr + Position); @@ -201,7 +201,7 @@ namespace ChocolArm64.Memory Modified[(VA - Position) / HostPageSize] = true; } - return Modified; + return (Modified, Count); } public IntPtr GetHostAddress(long Position, long Size) diff --git a/Ryujinx.HLE/Gpu/Engines/NvGpuEngine3d.cs b/Ryujinx.HLE/Gpu/Engines/NvGpuEngine3d.cs index c3e7a77fcb..3dd0e98e60 100644 --- a/Ryujinx.HLE/Gpu/Engines/NvGpuEngine3d.cs +++ b/Ryujinx.HLE/Gpu/Engines/NvGpuEngine3d.cs @@ -25,6 +25,8 @@ namespace Ryujinx.HLE.Gpu.Engines private HashSet FrameBuffers; + private List[] UploadedKeys; + public NvGpuEngine3d(NvGpu Gpu) { this.Gpu = Gpu; @@ -57,6 +59,13 @@ namespace Ryujinx.HLE.Gpu.Engines } FrameBuffers = new HashSet(); + + UploadedKeys = new List[(int)NvGpuBufferType.Count]; + + for (int i = 0; i < UploadedKeys.Length; i++) + { + UploadedKeys[i] = new List(); + } } public void CallMethod(NvGpuVmm Vmm, NvGpuPBEntry PBEntry) @@ -516,7 +525,7 @@ namespace Ryujinx.HLE.Gpu.Engines if (Gpu.Renderer.Texture.TryGetCachedTexture(Key, Size, out GalTexture Texture)) { - if (NewTexture.Equals(Texture) && !Vmm.IsRegionModified(Key, Size, NvGpuBufferType.Texture)) + if (NewTexture.Equals(Texture) && !QueryKeyUpload(Vmm, Key, Size, NvGpuBufferType.Texture)) { Gpu.Renderer.Texture.Bind(Key, TexIndex); @@ -593,7 +602,7 @@ namespace Ryujinx.HLE.Gpu.Engines bool IboCached = Gpu.Renderer.Rasterizer.IsIboCached(IboKey, (uint)IbSize); - if (!IboCached || Vmm.IsRegionModified(IboKey, (uint)IbSize, NvGpuBufferType.Index)) + if (!IboCached || QueryKeyUpload(Vmm, IboKey, (uint)IbSize, NvGpuBufferType.Index)) { IntPtr DataAddress = Vmm.GetHostAddress(IndexPosition, IbSize); @@ -657,7 +666,7 @@ namespace Ryujinx.HLE.Gpu.Engines bool VboCached = Gpu.Renderer.Rasterizer.IsVboCached(VboKey, VbSize); - if (!VboCached || Vmm.IsRegionModified(VboKey, VbSize, NvGpuBufferType.Vertex)) + if (!VboCached || QueryKeyUpload(Vmm, VboKey, VbSize, NvGpuBufferType.Vertex)) { IntPtr DataAddress = Vmm.GetHostAddress(VertexPosition, VbSize); @@ -692,6 +701,11 @@ namespace Ryujinx.HLE.Gpu.Engines if (Mode == 0) { + foreach (List Uploaded in UploadedKeys) + { + Uploaded.Clear(); + } + //Write mode. Vmm.WriteInt32(Position, Seq); } @@ -774,5 +788,19 @@ namespace Ryujinx.HLE.Gpu.Engines { return FrameBuffers.Contains(Position); } + + private bool QueryKeyUpload(NvGpuVmm Vmm, long Key, long Size, NvGpuBufferType Type) + { + List Uploaded = UploadedKeys[(int)Type]; + + if (Uploaded.Contains(Key)) + { + return false; + } + + Uploaded.Add(Key); + + return Vmm.IsRegionModified(Key, Size, Type); + } } } \ No newline at end of file diff --git a/Ryujinx.HLE/Gpu/Memory/NvGpuVmmCache.cs b/Ryujinx.HLE/Gpu/Memory/NvGpuVmmCache.cs index ac9bd850e0..3c25604433 100644 --- a/Ryujinx.HLE/Gpu/Memory/NvGpuVmmCache.cs +++ b/Ryujinx.HLE/Gpu/Memory/NvGpuVmmCache.cs @@ -19,12 +19,14 @@ namespace Ryujinx.HLE.Gpu.Memory public Range(long Start, long End) { this.Start = Start; - this.End = End; + this.End = End; } } private List[] Regions; + private HashSet ResidencyKeys; + public LinkedListNode Node { get; set; } public int Timestamp { get; private set; } @@ -37,6 +39,27 @@ namespace Ryujinx.HLE.Gpu.Memory { Regions[Index] = new List(); } + + ResidencyKeys = new HashSet(); + } + + public void AddResidency(long Key) + { + ResidencyKeys.Add(Key); + } + + public void RemoveResidency(HashSet[] Residency, long PageSize) + { + for (int i = 0; i < (int)NvGpuBufferType.Count; i++) + { + foreach (Range Region in Regions[i]) + { + foreach (long Key in ResidencyKeys) + { + Residency[Region.Start / PageSize].Remove(Key); + } + } + } } public bool AddRange(long Start, long End, NvGpuBufferType BufferType) @@ -89,6 +112,10 @@ namespace Ryujinx.HLE.Gpu.Memory private LinkedList SortedCache; + private HashSet[] Residency; + + private long ResidencyPageSize; + private int CpCount; public NvGpuVmmCache() @@ -100,7 +127,7 @@ namespace Ryujinx.HLE.Gpu.Memory public bool IsRegionModified(AMemory Memory, NvGpuBufferType BufferType, long PA, long Size) { - bool[] Modified = Memory.IsRegionModified(PA, Size); + (bool[] Modified, long ModifiedCount) = Memory.IsRegionModified(PA, Size); if (Modified == null) { @@ -111,8 +138,19 @@ namespace Ryujinx.HLE.Gpu.Memory long PageSize = Memory.GetHostPageSize(); + EnsureResidencyInitialized(PageSize); + + bool HasResidents = AddResidency(PA, Size); + + if (!HasResidents && ModifiedCount == 0) + { + return false; + } + long Mask = PageSize - 1; + long ResidencyKey = PA; + long PAEnd = PA + Size; bool RegMod = false; @@ -147,6 +185,8 @@ namespace Ryujinx.HLE.Gpu.Memory Cache[Key] = Cp; } + Cp.AddResidency(ResidencyKey); + Cp.Node = SortedCache.AddLast(Key); RegMod |= Cp.AddRange(PA, PAPgEnd, BufferType); @@ -159,6 +199,53 @@ namespace Ryujinx.HLE.Gpu.Memory return RegMod; } + private bool AddResidency(long PA, long Size) + { + long PageSize = ResidencyPageSize; + + long Mask = PageSize - 1; + + long Key = PA; + + bool ResidentFound = false; + + for (long Cursor = PA & ~Mask; Cursor < ((PA + Size + PageSize - 1) & ~Mask); Cursor += PageSize) + { + long PageIndex = Cursor / PageSize; + + Residency[PageIndex].Add(Key); + + if (Residency[PageIndex].Count > 1) + { + ResidentFound = true; + } + } + + return ResidentFound; + } + + private void EnsureResidencyInitialized(long PageSize) + { + if (Residency == null) + { + Residency = new HashSet[AMemoryMgr.RamSize / PageSize]; + + for (int i = 0; i < Residency.Length; i++) + { + Residency[i] = new HashSet(); + } + + ResidencyPageSize = PageSize; + } + else + { + if (ResidencyPageSize != PageSize) + { + throw new InvalidOperationException("Tried to change residency page size"); + } + } + } + private void ClearCachedPagesIfNeeded() { if (CpCount <= MaxCpCount) @@ -179,6 +266,8 @@ namespace Ryujinx.HLE.Gpu.Memory CachedPage Cp = Cache[Key]; + Cp.RemoveResidency(Residency, ResidencyPageSize); + Cache.Remove(Key); CpCount -= Cp.GetTotalCount(); From e07b0265c6516b15ceccd06ca5bd56e79449f2c9 Mon Sep 17 00:00:00 2001 From: Starlet Date: Sun, 29 Jul 2018 00:40:02 -0400 Subject: [PATCH 26/34] Add 5.0.0 IPCs (#300) * Implement 5.0.0 IPC * Align ControlWithContext --- Ryujinx.HLE/OsHle/Ipc/IpcHandler.cs | 6 ++++-- Ryujinx.HLE/OsHle/Ipc/IpcMessageType.cs | 10 ++++++---- 2 files changed, 10 insertions(+), 6 deletions(-) diff --git a/Ryujinx.HLE/OsHle/Ipc/IpcHandler.cs b/Ryujinx.HLE/OsHle/Ipc/IpcHandler.cs index 9b46cf4bf1..cdb844cfe2 100644 --- a/Ryujinx.HLE/OsHle/Ipc/IpcHandler.cs +++ b/Ryujinx.HLE/OsHle/Ipc/IpcHandler.cs @@ -21,7 +21,8 @@ namespace Ryujinx.HLE.OsHle.Ipc { BinaryReader ReqReader = new BinaryReader(Raw); - if (Request.Type == IpcMessageType.Request) + if (Request.Type == IpcMessageType.Request || + Request.Type == IpcMessageType.RequestWithContext) { Response.Type = IpcMessageType.Response; @@ -44,7 +45,8 @@ namespace Ryujinx.HLE.OsHle.Ipc Response.RawData = ResMS.ToArray(); } } - else if (Request.Type == IpcMessageType.Control) + else if (Request.Type == IpcMessageType.Control || + Request.Type == IpcMessageType.ControlWithContext) { long Magic = ReqReader.ReadInt64(); long CmdId = ReqReader.ReadInt64(); diff --git a/Ryujinx.HLE/OsHle/Ipc/IpcMessageType.cs b/Ryujinx.HLE/OsHle/Ipc/IpcMessageType.cs index f596fea461..3db6844e0a 100644 --- a/Ryujinx.HLE/OsHle/Ipc/IpcMessageType.cs +++ b/Ryujinx.HLE/OsHle/Ipc/IpcMessageType.cs @@ -2,9 +2,11 @@ namespace Ryujinx.HLE.OsHle.Ipc { enum IpcMessageType { - Response = 0, - CloseSession = 2, - Request = 4, - Control = 5 + Response = 0, + CloseSession = 2, + Request = 4, + Control = 5, + RequestWithContext = 6, + ControlWithContext = 7 } } \ No newline at end of file From ce96a45685a9ab9e24d8d604185b14a04f4c4abd Mon Sep 17 00:00:00 2001 From: greggameplayer <33609333+greggameplayer@users.noreply.github.com> Date: Sun, 29 Jul 2018 06:41:02 +0200 Subject: [PATCH 27/34] Implement A2B10G10R10 TextureFormat (#248) * add A2B10G10R10 TextureFormat * return correct PixelFormat & PixelType * return correct texture size * return correct Bytes Per Pixel --- Ryujinx.Graphics/Gal/GalTextureFormat.cs | 1 + Ryujinx.Graphics/Gal/OpenGL/OGLEnumConverter.cs | 1 + Ryujinx.HLE/Gpu/Texture/TextureHelper.cs | 1 + Ryujinx.HLE/Gpu/Texture/TextureReader.cs | 1 + 4 files changed, 4 insertions(+) diff --git a/Ryujinx.Graphics/Gal/GalTextureFormat.cs b/Ryujinx.Graphics/Gal/GalTextureFormat.cs index c478428325..7e3e65e82f 100644 --- a/Ryujinx.Graphics/Gal/GalTextureFormat.cs +++ b/Ryujinx.Graphics/Gal/GalTextureFormat.cs @@ -5,6 +5,7 @@ namespace Ryujinx.Graphics.Gal R32G32B32A32 = 0x1, R16G16B16A16 = 0x3, A8B8G8R8 = 0x8, + A2B10G10R10 = 0x9, R32 = 0xf, BC6H_SF16 = 0x10, BC6H_UF16 = 0x11, diff --git a/Ryujinx.Graphics/Gal/OpenGL/OGLEnumConverter.cs b/Ryujinx.Graphics/Gal/OpenGL/OGLEnumConverter.cs index 9004e9bae2..3c42e5d387 100644 --- a/Ryujinx.Graphics/Gal/OpenGL/OGLEnumConverter.cs +++ b/Ryujinx.Graphics/Gal/OpenGL/OGLEnumConverter.cs @@ -132,6 +132,7 @@ namespace Ryujinx.Graphics.Gal.OpenGL case GalTextureFormat.R32G32B32A32: return (PixelFormat.Rgba, PixelType.Float); case GalTextureFormat.R16G16B16A16: return (PixelFormat.Rgba, PixelType.HalfFloat); case GalTextureFormat.A8B8G8R8: return (PixelFormat.Rgba, PixelType.UnsignedByte); + case GalTextureFormat.A2B10G10R10: return (PixelFormat.Rgba, PixelType.UnsignedInt2101010Reversed); case GalTextureFormat.R32: return (PixelFormat.Red, PixelType.Float); case GalTextureFormat.A1B5G5R5: return (PixelFormat.Rgba, PixelType.UnsignedShort5551); case GalTextureFormat.B5G6R5: return (PixelFormat.Rgb, PixelType.UnsignedShort565); diff --git a/Ryujinx.HLE/Gpu/Texture/TextureHelper.cs b/Ryujinx.HLE/Gpu/Texture/TextureHelper.cs index ecf2b6bf5c..1d621c9226 100644 --- a/Ryujinx.HLE/Gpu/Texture/TextureHelper.cs +++ b/Ryujinx.HLE/Gpu/Texture/TextureHelper.cs @@ -41,6 +41,7 @@ namespace Ryujinx.HLE.Gpu.Texture return Texture.Width * Texture.Height * 8; case GalTextureFormat.A8B8G8R8: + case GalTextureFormat.A2B10G10R10: case GalTextureFormat.R32: case GalTextureFormat.ZF32: case GalTextureFormat.BF10GF11RF11: diff --git a/Ryujinx.HLE/Gpu/Texture/TextureReader.cs b/Ryujinx.HLE/Gpu/Texture/TextureReader.cs index 350ab825f4..f8cd676597 100644 --- a/Ryujinx.HLE/Gpu/Texture/TextureReader.cs +++ b/Ryujinx.HLE/Gpu/Texture/TextureReader.cs @@ -13,6 +13,7 @@ namespace Ryujinx.HLE.Gpu.Texture case GalTextureFormat.R32G32B32A32: return Read16Bpp (Memory, Texture); case GalTextureFormat.R16G16B16A16: return Read8Bpp (Memory, Texture); case GalTextureFormat.A8B8G8R8: return Read4Bpp (Memory, Texture); + case GalTextureFormat.A2B10G10R10: return Read4Bpp (Memory, Texture); case GalTextureFormat.R32: return Read4Bpp (Memory, Texture); case GalTextureFormat.BF10GF11RF11: return Read4Bpp (Memory, Texture); case GalTextureFormat.Z24S8: return Read4Bpp (Memory, Texture); From b92189cb711fcf39255e04d3256c2728f271dbfd Mon Sep 17 00:00:00 2001 From: ReinUsesLisp Date: Wed, 1 Aug 2018 00:48:49 -0300 Subject: [PATCH 28/34] Sleep OpenAL thread (#313) --- Ryujinx.Audio/OpenAL/OpenALAudioOut.cs | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/Ryujinx.Audio/OpenAL/OpenALAudioOut.cs b/Ryujinx.Audio/OpenAL/OpenALAudioOut.cs index 1dd63202ba..85e2d803e9 100644 --- a/Ryujinx.Audio/OpenAL/OpenALAudioOut.cs +++ b/Ryujinx.Audio/OpenAL/OpenALAudioOut.cs @@ -222,7 +222,8 @@ namespace Ryujinx.Audio.OpenAL Td.CallReleaseCallbackIfNeeded(); } - Thread.Yield(); + //If it's not slept it will waste cycles + Thread.Sleep(10); } while (KeepPolling); } From a9a2c0c2f59e1354ca27bdf66d23fcc13e6025d6 Mon Sep 17 00:00:00 2001 From: emmaus Date: Thu, 2 Aug 2018 20:14:49 +0000 Subject: [PATCH 29/34] moved metadata to process --- Ryujinx.HLE/OsHle/Horizon.cs | 4 ++-- Ryujinx.HLE/OsHle/Process.cs | 3 +++ Ryujinx.HLE/OsHle/SystemStateMgr.cs | 2 -- 3 files changed, 5 insertions(+), 4 deletions(-) diff --git a/Ryujinx.HLE/OsHle/Horizon.cs b/Ryujinx.HLE/OsHle/Horizon.cs index 4c117fa478..0318fba56a 100644 --- a/Ryujinx.HLE/OsHle/Horizon.cs +++ b/Ryujinx.HLE/OsHle/Horizon.cs @@ -85,13 +85,13 @@ namespace Ryujinx.HLE.OsHle using (FileStream Input = new FileStream(File, FileMode.Open)) { - SystemStateMgr.TitleMetadata = new Npdm(Input); + MainProcess.Metadata = new Npdm(Input); } } LoadNpdm("*.npdm"); - if (!SystemStateMgr.TitleMetadata.Is64Bits) + if (!MainProcess.Metadata.Is64Bits) { throw new Exception("32-bit titles are unsupported!"); } diff --git a/Ryujinx.HLE/OsHle/Process.cs b/Ryujinx.HLE/OsHle/Process.cs index c7606dc902..ac5ed45602 100644 --- a/Ryujinx.HLE/OsHle/Process.cs +++ b/Ryujinx.HLE/OsHle/Process.cs @@ -4,6 +4,7 @@ using ChocolArm64.Memory; using ChocolArm64.State; using Ryujinx.HLE.Loaders; using Ryujinx.HLE.Loaders.Executables; +using Ryujinx.HLE.Loaders.Npdm; using Ryujinx.HLE.Logging; using Ryujinx.HLE.OsHle.Diagnostics; using Ryujinx.HLE.OsHle.Exceptions; @@ -48,6 +49,8 @@ namespace Ryujinx.HLE.OsHle public AppletStateMgr AppletState { get; private set; } + public Npdm Metadata { get; set; } + private SvcHandler SvcHandler; private ConcurrentDictionary TlsSlots; diff --git a/Ryujinx.HLE/OsHle/SystemStateMgr.cs b/Ryujinx.HLE/OsHle/SystemStateMgr.cs index 32c88994b9..7b6ef7f3cf 100644 --- a/Ryujinx.HLE/OsHle/SystemStateMgr.cs +++ b/Ryujinx.HLE/OsHle/SystemStateMgr.cs @@ -5,8 +5,6 @@ namespace Ryujinx.HLE.OsHle { public class SystemStateMgr { - internal static Npdm TitleMetadata { get; set; } - internal static string[] LanguageCodes = new string[] { "ja", From c68bca5396f8a5e08cd8234ea0c7a8a2174eba91 Mon Sep 17 00:00:00 2001 From: AboodMadridista <41301312+AboodMadridista@users.noreply.github.com> Date: Fri, 3 Aug 2018 08:58:48 +0800 Subject: [PATCH 30/34] Use the more specific NotImplementedException type instead of just Exception (#322) --- Ryujinx.HLE/OsHle/Horizon.cs | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/Ryujinx.HLE/OsHle/Horizon.cs b/Ryujinx.HLE/OsHle/Horizon.cs index 0318fba56a..f420dc0df8 100644 --- a/Ryujinx.HLE/OsHle/Horizon.cs +++ b/Ryujinx.HLE/OsHle/Horizon.cs @@ -93,7 +93,7 @@ namespace Ryujinx.HLE.OsHle if (!MainProcess.Metadata.Is64Bits) { - throw new Exception("32-bit titles are unsupported!"); + throw new NotImplementedException("32-bit titles are unsupported!"); } LoadNso("rtld"); From fa70629fabbab5074640f55cb70f9d7d82cf91cb Mon Sep 17 00:00:00 2001 From: gdkchan Date: Fri, 3 Aug 2018 13:54:34 -0300 Subject: [PATCH 31/34] Fix for integer vertex attributes and iset bf flag (#323) --- Ryujinx.Graphics/Gal/OpenGL/OGLRasterizer.cs | 14 +++++++++++++- Ryujinx.Graphics/Gal/Shader/ShaderDecodeAlu.cs | 11 ++++++----- 2 files changed, 19 insertions(+), 6 deletions(-) diff --git a/Ryujinx.Graphics/Gal/OpenGL/OGLRasterizer.cs b/Ryujinx.Graphics/Gal/OpenGL/OGLRasterizer.cs index f2e5859e5e..c5166b51bf 100644 --- a/Ryujinx.Graphics/Gal/OpenGL/OGLRasterizer.cs +++ b/Ryujinx.Graphics/Gal/OpenGL/OGLRasterizer.cs @@ -278,7 +278,19 @@ namespace Ryujinx.Graphics.Gal.OpenGL int Size = AttribElements[Attrib.Size]; int Offset = Attrib.Offset; - GL.VertexAttribPointer(Attrib.Index, Size, Type, Normalize, Stride, Offset); + if (Attrib.Type == GalVertexAttribType.Sint || + Attrib.Type == GalVertexAttribType.Uint) + { + IntPtr Pointer = new IntPtr(Offset); + + VertexAttribIntegerType IType = (VertexAttribIntegerType)Type; + + GL.VertexAttribIPointer(Attrib.Index, Size, IType, Stride, Pointer); + } + else + { + GL.VertexAttribPointer(Attrib.Index, Size, Type, Normalize, Stride, Offset); + } } } diff --git a/Ryujinx.Graphics/Gal/Shader/ShaderDecodeAlu.cs b/Ryujinx.Graphics/Gal/Shader/ShaderDecodeAlu.cs index 00f072f192..8486d8a7d9 100644 --- a/Ryujinx.Graphics/Gal/Shader/ShaderDecodeAlu.cs +++ b/Ryujinx.Graphics/Gal/Shader/ShaderDecodeAlu.cs @@ -871,11 +871,12 @@ namespace Ryujinx.Graphics.Gal.Shader private static void EmitSet(ShaderIrBlock Block, long OpCode, bool IsFloat, ShaderOper Oper) { - bool NegA = ((OpCode >> 43) & 1) != 0; - bool AbsB = ((OpCode >> 44) & 1) != 0; - bool BoolFloat = ((OpCode >> 52) & 1) != 0; - bool NegB = ((OpCode >> 53) & 1) != 0; - bool AbsA = ((OpCode >> 54) & 1) != 0; + bool NegA = ((OpCode >> 43) & 1) != 0; + bool AbsB = ((OpCode >> 44) & 1) != 0; + bool NegB = ((OpCode >> 53) & 1) != 0; + bool AbsA = ((OpCode >> 54) & 1) != 0; + + bool BoolFloat = ((OpCode >> (IsFloat ? 52 : 44)) & 1) != 0; ShaderIrNode OperA = GetOperGpr8(OpCode), OperB; From 5f34353dce716bca2e3fc1c7e82be6276b95d61a Mon Sep 17 00:00:00 2001 From: LDj3SNuD <35856442+LDj3SNuD@users.noreply.github.com> Date: Sat, 4 Aug 2018 21:58:54 +0200 Subject: [PATCH 32/34] Add SQADD, UQADD, SQSUB, UQSUB, SUQADD, USQADD, SQABS, SQNEG (Scalar, Vector) instructions; add 24 Tests. Most saturation instructions now on ASoftFallback. (#314) * Update AOpCodeTable.cs * Update AInstEmitSimdHelper.cs * Update AInstEmitSimdArithmetic.cs * Update Pseudocode.cs * Update Instructions.cs * Update CpuTestSimd.cs * Update CpuTestSimdReg.cs * Update AInstEmitSimdHelper.cs * Update AInstEmitSimdHelper.cs * Update AInstEmitSimdHelper.cs * Update AInstEmitSimdHelper.cs * Update ASoftFallback.cs * Update AInstEmitSimdHelper.cs * Update ASoftFallback.cs * Update AInstEmitSimdHelper.cs * Update CpuTestSimd.cs * Update CpuTestSimdReg.cs * Update ASoftFallback.cs * Update AInstEmitSimdHelper.cs * Opt. (retest). --- ChocolArm64/AOpCodeTable.cs | 16 + .../Instruction/AInstEmitSimdArithmetic.cs | 80 ++ .../Instruction/AInstEmitSimdHelper.cs | 394 +++++++-- ChocolArm64/Instruction/ASoftFallback.cs | 268 ++++++ Ryujinx.Tests/Cpu/CpuTestSimd.cs | 446 +++++++++- Ryujinx.Tests/Cpu/CpuTestSimdReg.cs | 431 ++++++++++ Ryujinx.Tests/Cpu/Tester/Instructions.cs | 776 ++++++++++++++++++ Ryujinx.Tests/Cpu/Tester/Pseudocode.cs | 4 +- 8 files changed, 2330 insertions(+), 85 deletions(-) diff --git a/ChocolArm64/AOpCodeTable.cs b/ChocolArm64/AOpCodeTable.cs index aa8220e38d..aa8b1be606 100644 --- a/ChocolArm64/AOpCodeTable.cs +++ b/ChocolArm64/AOpCodeTable.cs @@ -374,7 +374,15 @@ namespace ChocolArm64 SetA64("0x001110<<1xxxxx100000xxxxxxxxxx", AInstEmit.Smlal_V, typeof(AOpCodeSimdReg)); SetA64("0x001110<<1xxxxx101000xxxxxxxxxx", AInstEmit.Smlsl_V, typeof(AOpCodeSimdReg)); SetA64("0x001110<<1xxxxx110000xxxxxxxxxx", AInstEmit.Smull_V, typeof(AOpCodeSimdReg)); + SetA64("01011110xx100000011110xxxxxxxxxx", AInstEmit.Sqabs_S, typeof(AOpCodeSimd)); + SetA64("0>001110<<100000011110xxxxxxxxxx", AInstEmit.Sqabs_V, typeof(AOpCodeSimd)); + SetA64("01011110xx1xxxxx000011xxxxxxxxxx", AInstEmit.Sqadd_S, typeof(AOpCodeSimdReg)); + SetA64("0>001110<<1xxxxx000011xxxxxxxxxx", AInstEmit.Sqadd_V, typeof(AOpCodeSimdReg)); + SetA64("01111110xx100000011110xxxxxxxxxx", AInstEmit.Sqneg_S, typeof(AOpCodeSimd)); + SetA64("0>101110<<100000011110xxxxxxxxxx", AInstEmit.Sqneg_V, typeof(AOpCodeSimd)); SetA64("0x00111100>>>xxx100111xxxxxxxxxx", AInstEmit.Sqrshrn_V, typeof(AOpCodeSimdShImm)); + SetA64("01011110xx1xxxxx001011xxxxxxxxxx", AInstEmit.Sqsub_S, typeof(AOpCodeSimdReg)); + SetA64("0>001110<<1xxxxx001011xxxxxxxxxx", AInstEmit.Sqsub_V, typeof(AOpCodeSimdReg)); SetA64("01011110<<100001010010xxxxxxxxxx", AInstEmit.Sqxtn_S, typeof(AOpCodeSimd)); SetA64("0x001110<<100001010010xxxxxxxxxx", AInstEmit.Sqxtn_V, typeof(AOpCodeSimd)); SetA64("01111110<<100001001010xxxxxxxxxx", AInstEmit.Sqxtun_S, typeof(AOpCodeSimd)); @@ -402,6 +410,8 @@ namespace ChocolArm64 SetA64("01111110111xxxxx100001xxxxxxxxxx", AInstEmit.Sub_S, typeof(AOpCodeSimdReg)); SetA64("0>101110<<1xxxxx100001xxxxxxxxxx", AInstEmit.Sub_V, typeof(AOpCodeSimdReg)); SetA64("0x001110<<1xxxxx011000xxxxxxxxxx", AInstEmit.Subhn_V, typeof(AOpCodeSimdReg)); + SetA64("01011110xx100000001110xxxxxxxxxx", AInstEmit.Suqadd_S, typeof(AOpCodeSimd)); + SetA64("0>001110<<100000001110xxxxxxxxxx", AInstEmit.Suqadd_V, typeof(AOpCodeSimd)); SetA64("0x001110000xxxxx0xx000xxxxxxxxxx", AInstEmit.Tbl_V, typeof(AOpCodeSimdTbl)); SetA64("0>001110<<0xxxxx001010xxxxxxxxxx", AInstEmit.Trn1_V, typeof(AOpCodeSimdReg)); SetA64("0>001110<<0xxxxx011010xxxxxxxxxx", AInstEmit.Trn2_V, typeof(AOpCodeSimdReg)); @@ -423,6 +433,10 @@ namespace ChocolArm64 SetA64("0x101110<<1xxxxx101011xxxxxxxxxx", AInstEmit.Uminp_V, typeof(AOpCodeSimdReg)); SetA64("0x001110000xxxxx001111xxxxxxxxxx", AInstEmit.Umov_S, typeof(AOpCodeSimdIns)); SetA64("0x101110<<1xxxxx110000xxxxxxxxxx", AInstEmit.Umull_V, typeof(AOpCodeSimdReg)); + SetA64("01111110xx1xxxxx000011xxxxxxxxxx", AInstEmit.Uqadd_S, typeof(AOpCodeSimdReg)); + SetA64("0>101110<<1xxxxx000011xxxxxxxxxx", AInstEmit.Uqadd_V, typeof(AOpCodeSimdReg)); + SetA64("01111110xx1xxxxx001011xxxxxxxxxx", AInstEmit.Uqsub_S, typeof(AOpCodeSimdReg)); + SetA64("0>101110<<1xxxxx001011xxxxxxxxxx", AInstEmit.Uqsub_V, typeof(AOpCodeSimdReg)); SetA64("01111110<<100001010010xxxxxxxxxx", AInstEmit.Uqxtn_S, typeof(AOpCodeSimd)); SetA64("0x101110<<100001010010xxxxxxxxxx", AInstEmit.Uqxtn_V, typeof(AOpCodeSimd)); SetA64("0>101110<<1xxxxx010001xxxxxxxxxx", AInstEmit.Ushl_V, typeof(AOpCodeSimdReg)); @@ -430,6 +444,8 @@ namespace ChocolArm64 SetA64("0111111101xxxxxx000001xxxxxxxxxx", AInstEmit.Ushr_S, typeof(AOpCodeSimdShImm)); SetA64("0x10111100>>>xxx000001xxxxxxxxxx", AInstEmit.Ushr_V, typeof(AOpCodeSimdShImm)); SetA64("0110111101xxxxxx000001xxxxxxxxxx", AInstEmit.Ushr_V, typeof(AOpCodeSimdShImm)); + SetA64("01111110xx100000001110xxxxxxxxxx", AInstEmit.Usqadd_S, typeof(AOpCodeSimd)); + SetA64("0>101110<<100000001110xxxxxxxxxx", AInstEmit.Usqadd_V, typeof(AOpCodeSimd)); SetA64("0x10111100>>>xxx000101xxxxxxxxxx", AInstEmit.Usra_V, typeof(AOpCodeSimdShImm)); SetA64("0110111101xxxxxx000101xxxxxxxxxx", AInstEmit.Usra_V, typeof(AOpCodeSimdShImm)); SetA64("0x101110<<1xxxxx001100xxxxxxxxxx", AInstEmit.Usubw_V, typeof(AOpCodeSimdReg)); diff --git a/ChocolArm64/Instruction/AInstEmitSimdArithmetic.cs b/ChocolArm64/Instruction/AInstEmitSimdArithmetic.cs index 8e7418611e..6772fe83b7 100644 --- a/ChocolArm64/Instruction/AInstEmitSimdArithmetic.cs +++ b/ChocolArm64/Instruction/AInstEmitSimdArithmetic.cs @@ -1052,6 +1052,46 @@ namespace ChocolArm64.Instruction EmitVectorWidenRnRmBinaryOpSx(Context, () => Context.Emit(OpCodes.Mul)); } + public static void Sqabs_S(AILEmitterCtx Context) + { + EmitScalarSaturatingUnaryOpSx(Context, () => EmitAbs(Context)); + } + + public static void Sqabs_V(AILEmitterCtx Context) + { + EmitVectorSaturatingUnaryOpSx(Context, () => EmitAbs(Context)); + } + + public static void Sqadd_S(AILEmitterCtx Context) + { + EmitScalarSaturatingBinaryOpSx(Context, SaturatingFlags.Add); + } + + public static void Sqadd_V(AILEmitterCtx Context) + { + EmitVectorSaturatingBinaryOpSx(Context, SaturatingFlags.Add); + } + + public static void Sqneg_S(AILEmitterCtx Context) + { + EmitScalarSaturatingUnaryOpSx(Context, () => Context.Emit(OpCodes.Neg)); + } + + public static void Sqneg_V(AILEmitterCtx Context) + { + EmitVectorSaturatingUnaryOpSx(Context, () => Context.Emit(OpCodes.Neg)); + } + + public static void Sqsub_S(AILEmitterCtx Context) + { + EmitScalarSaturatingBinaryOpSx(Context, SaturatingFlags.Sub); + } + + public static void Sqsub_V(AILEmitterCtx Context) + { + EmitVectorSaturatingBinaryOpSx(Context, SaturatingFlags.Sub); + } + public static void Sqxtn_S(AILEmitterCtx Context) { EmitScalarSaturatingNarrowOpSxSx(Context, () => { }); @@ -1099,6 +1139,16 @@ namespace ChocolArm64.Instruction EmitHighNarrow(Context, () => Context.Emit(OpCodes.Sub), Round: false); } + public static void Suqadd_S(AILEmitterCtx Context) + { + EmitScalarSaturatingBinaryOpSx(Context, SaturatingFlags.Accumulate); + } + + public static void Suqadd_V(AILEmitterCtx Context) + { + EmitVectorSaturatingBinaryOpSx(Context, SaturatingFlags.Accumulate); + } + public static void Uaba_V(AILEmitterCtx Context) { EmitVectorTernaryOpZx(Context, () => @@ -1221,6 +1271,26 @@ namespace ChocolArm64.Instruction EmitVectorWidenRnRmBinaryOpZx(Context, () => Context.Emit(OpCodes.Mul)); } + public static void Uqadd_S(AILEmitterCtx Context) + { + EmitScalarSaturatingBinaryOpZx(Context, SaturatingFlags.Add); + } + + public static void Uqadd_V(AILEmitterCtx Context) + { + EmitVectorSaturatingBinaryOpZx(Context, SaturatingFlags.Add); + } + + public static void Uqsub_S(AILEmitterCtx Context) + { + EmitScalarSaturatingBinaryOpZx(Context, SaturatingFlags.Sub); + } + + public static void Uqsub_V(AILEmitterCtx Context) + { + EmitVectorSaturatingBinaryOpZx(Context, SaturatingFlags.Sub); + } + public static void Uqxtn_S(AILEmitterCtx Context) { EmitScalarSaturatingNarrowOpZxZx(Context, () => { }); @@ -1231,6 +1301,16 @@ namespace ChocolArm64.Instruction EmitVectorSaturatingNarrowOpZxZx(Context, () => { }); } + public static void Usqadd_S(AILEmitterCtx Context) + { + EmitScalarSaturatingBinaryOpZx(Context, SaturatingFlags.Accumulate); + } + + public static void Usqadd_V(AILEmitterCtx Context) + { + EmitVectorSaturatingBinaryOpZx(Context, SaturatingFlags.Accumulate); + } + public static void Usubw_V(AILEmitterCtx Context) { EmitVectorWidenRmBinaryOpZx(Context, () => Context.Emit(OpCodes.Sub)); diff --git a/ChocolArm64/Instruction/AInstEmitSimdHelper.cs b/ChocolArm64/Instruction/AInstEmitSimdHelper.cs index 0bd1a62926..161c44ea26 100644 --- a/ChocolArm64/Instruction/AInstEmitSimdHelper.cs +++ b/ChocolArm64/Instruction/AInstEmitSimdHelper.cs @@ -336,17 +336,21 @@ namespace ChocolArm64.Instruction { AOpCodeSimd Op = (AOpCodeSimd)Context.CurrOp; - if (Opers.HasFlag(OperFlags.Rd)) + bool Rd = (Opers & OperFlags.Rd) != 0; + bool Rn = (Opers & OperFlags.Rn) != 0; + bool Rm = (Opers & OperFlags.Rm) != 0; + + if (Rd) { EmitVectorExtract(Context, Op.Rd, 0, Op.Size, Signed); } - if (Opers.HasFlag(OperFlags.Rn)) + if (Rn) { EmitVectorExtract(Context, Op.Rn, 0, Op.Size, Signed); } - if (Opers.HasFlag(OperFlags.Rm)) + if (Rm) { EmitVectorExtract(Context, ((AOpCodeSimdReg)Op).Rm, 0, Op.Size, Signed); } @@ -377,17 +381,21 @@ namespace ChocolArm64.Instruction int SizeF = Op.Size & 1; - if (Opers.HasFlag(OperFlags.Ra)) + bool Ra = (Opers & OperFlags.Ra) != 0; + bool Rn = (Opers & OperFlags.Rn) != 0; + bool Rm = (Opers & OperFlags.Rm) != 0; + + if (Ra) { EmitVectorExtractF(Context, ((AOpCodeSimdReg)Op).Ra, 0, SizeF); } - if (Opers.HasFlag(OperFlags.Rn)) + if (Rn) { EmitVectorExtractF(Context, Op.Rn, 0, SizeF); } - if (Opers.HasFlag(OperFlags.Rm)) + if (Rm) { EmitVectorExtractF(Context, ((AOpCodeSimdReg)Op).Rm, 0, SizeF); } @@ -769,7 +777,7 @@ namespace ChocolArm64.Instruction Emit(); EmitVectorInsertTmp(Context, Pairs + Index, Op.Size); - EmitVectorInsertTmp(Context, Index, Op.Size); + EmitVectorInsertTmp(Context, Index, Op.Size); } Context.EmitLdvectmp(); @@ -781,56 +789,241 @@ namespace ChocolArm64.Instruction } } + [Flags] + public enum SaturatingFlags + { + Scalar = 1 << 0, + Signed = 1 << 1, + + Add = 1 << 2, + Sub = 1 << 3, + + Accumulate = 1 << 4, + + ScalarSx = Scalar | Signed, + ScalarZx = Scalar, + + VectorSx = Signed, + VectorZx = 0, + } + + public static void EmitScalarSaturatingUnaryOpSx(AILEmitterCtx Context, Action Emit) + { + EmitSaturatingUnaryOpSx(Context, Emit, SaturatingFlags.ScalarSx); + } + + public static void EmitVectorSaturatingUnaryOpSx(AILEmitterCtx Context, Action Emit) + { + EmitSaturatingUnaryOpSx(Context, Emit, SaturatingFlags.VectorSx); + } + + public static void EmitSaturatingUnaryOpSx(AILEmitterCtx Context, Action Emit, SaturatingFlags Flags) + { + AOpCodeSimd Op = (AOpCodeSimd)Context.CurrOp; + + bool Scalar = (Flags & SaturatingFlags.Scalar) != 0; + + int Bytes = Op.GetBitsCount() >> 3; + int Elems = !Scalar ? Bytes >> Op.Size : 1; + + if (Scalar) + { + EmitVectorZeroLowerTmp(Context); + } + + for (int Index = 0; Index < Elems; Index++) + { + EmitVectorExtractSx(Context, Op.Rn, Index, Op.Size); + + Emit(); + + EmitUnarySignedSatQAbsOrNeg(Context, Op.Size); + + EmitVectorInsertTmp(Context, Index, Op.Size); + } + + Context.EmitLdvectmp(); + Context.EmitStvec(Op.Rd); + + if ((Op.RegisterSize == ARegisterSize.SIMD64) || Scalar) + { + EmitVectorZeroUpper(Context, Op.Rd); + } + } + + public static void EmitScalarSaturatingBinaryOpSx(AILEmitterCtx Context, SaturatingFlags Flags) + { + EmitSaturatingBinaryOp(Context, SaturatingFlags.ScalarSx | Flags); + } + + public static void EmitScalarSaturatingBinaryOpZx(AILEmitterCtx Context, SaturatingFlags Flags) + { + EmitSaturatingBinaryOp(Context, SaturatingFlags.ScalarZx | Flags); + } + + public static void EmitVectorSaturatingBinaryOpSx(AILEmitterCtx Context, SaturatingFlags Flags) + { + EmitSaturatingBinaryOp(Context, SaturatingFlags.VectorSx | Flags); + } + + public static void EmitVectorSaturatingBinaryOpZx(AILEmitterCtx Context, SaturatingFlags Flags) + { + EmitSaturatingBinaryOp(Context, SaturatingFlags.VectorZx | Flags); + } + + public static void EmitSaturatingBinaryOp(AILEmitterCtx Context, SaturatingFlags Flags) + { + AOpCodeSimd Op = (AOpCodeSimd)Context.CurrOp; + + bool Scalar = (Flags & SaturatingFlags.Scalar) != 0; + bool Signed = (Flags & SaturatingFlags.Signed) != 0; + + bool Add = (Flags & SaturatingFlags.Add) != 0; + bool Sub = (Flags & SaturatingFlags.Sub) != 0; + + bool Accumulate = (Flags & SaturatingFlags.Accumulate) != 0; + + int Bytes = Op.GetBitsCount() >> 3; + int Elems = !Scalar ? Bytes >> Op.Size : 1; + + if (Scalar) + { + EmitVectorZeroLowerTmp(Context); + } + + if (Add || Sub) + { + for (int Index = 0; Index < Elems; Index++) + { + EmitVectorExtract(Context, Op.Rn, Index, Op.Size, Signed); + EmitVectorExtract(Context, ((AOpCodeSimdReg)Op).Rm, Index, Op.Size, Signed); + + if (Op.Size <= 2) + { + Context.Emit(Add ? OpCodes.Add : OpCodes.Sub); + + EmitSatQ(Context, Op.Size, true, Signed); + } + else /* if (Op.Size == 3) */ + { + if (Add) + { + EmitBinarySatQAdd(Context, Signed); + } + else /* if (Sub) */ + { + EmitBinarySatQSub(Context, Signed); + } + } + + EmitVectorInsertTmp(Context, Index, Op.Size); + } + } + else if (Accumulate) + { + for (int Index = 0; Index < Elems; Index++) + { + EmitVectorExtract(Context, Op.Rn, Index, Op.Size, !Signed); + EmitVectorExtract(Context, Op.Rd, Index, Op.Size, Signed); + + if (Op.Size <= 2) + { + Context.Emit(OpCodes.Add); + + EmitSatQ(Context, Op.Size, true, Signed); + } + else /* if (Op.Size == 3) */ + { + EmitBinarySatQAccumulate(Context, Signed); + } + + EmitVectorInsertTmp(Context, Index, Op.Size); + } + } + + Context.EmitLdvectmp(); + Context.EmitStvec(Op.Rd); + + if ((Op.RegisterSize == ARegisterSize.SIMD64) || Scalar) + { + EmitVectorZeroUpper(Context, Op.Rd); + } + } + + [Flags] + public enum SaturatingNarrowFlags + { + Scalar = 1 << 0, + SignedSrc = 1 << 1, + SignedDst = 1 << 2, + + ScalarSxSx = Scalar | SignedSrc | SignedDst, + ScalarSxZx = Scalar | SignedSrc, + ScalarZxSx = Scalar | SignedDst, + ScalarZxZx = Scalar, + + VectorSxSx = SignedSrc | SignedDst, + VectorSxZx = SignedSrc, + VectorZxSx = SignedDst, + VectorZxZx = 0 + } + public static void EmitScalarSaturatingNarrowOpSxSx(AILEmitterCtx Context, Action Emit) { - EmitSaturatingNarrowOp(Context, Emit, true, true, true); + EmitSaturatingNarrowOp(Context, Emit, SaturatingNarrowFlags.ScalarSxSx); } public static void EmitScalarSaturatingNarrowOpSxZx(AILEmitterCtx Context, Action Emit) { - EmitSaturatingNarrowOp(Context, Emit, true, false, true); + EmitSaturatingNarrowOp(Context, Emit, SaturatingNarrowFlags.ScalarSxZx); + } + + public static void EmitScalarSaturatingNarrowOpZxSx(AILEmitterCtx Context, Action Emit) + { + EmitSaturatingNarrowOp(Context, Emit, SaturatingNarrowFlags.ScalarZxSx); } public static void EmitScalarSaturatingNarrowOpZxZx(AILEmitterCtx Context, Action Emit) { - EmitSaturatingNarrowOp(Context, Emit, false, false, true); + EmitSaturatingNarrowOp(Context, Emit, SaturatingNarrowFlags.ScalarZxZx); } public static void EmitVectorSaturatingNarrowOpSxSx(AILEmitterCtx Context, Action Emit) { - EmitSaturatingNarrowOp(Context, Emit, true, true, false); + EmitSaturatingNarrowOp(Context, Emit, SaturatingNarrowFlags.VectorSxSx); } public static void EmitVectorSaturatingNarrowOpSxZx(AILEmitterCtx Context, Action Emit) { - EmitSaturatingNarrowOp(Context, Emit, true, false, false); + EmitSaturatingNarrowOp(Context, Emit, SaturatingNarrowFlags.VectorSxZx); + } + + public static void EmitVectorSaturatingNarrowOpZxSx(AILEmitterCtx Context, Action Emit) + { + EmitSaturatingNarrowOp(Context, Emit, SaturatingNarrowFlags.VectorZxSx); } public static void EmitVectorSaturatingNarrowOpZxZx(AILEmitterCtx Context, Action Emit) { - EmitSaturatingNarrowOp(Context, Emit, false, false, false); + EmitSaturatingNarrowOp(Context, Emit, SaturatingNarrowFlags.VectorZxZx); } - public static void EmitSaturatingNarrowOp( - AILEmitterCtx Context, - Action Emit, - bool SignedSrc, - bool SignedDst, - bool Scalar) + public static void EmitSaturatingNarrowOp(AILEmitterCtx Context, Action Emit, SaturatingNarrowFlags Flags) { AOpCodeSimd Op = (AOpCodeSimd)Context.CurrOp; - int Elems = !Scalar ? 8 >> Op.Size : 1; + bool Scalar = (Flags & SaturatingNarrowFlags.Scalar) != 0; + bool SignedSrc = (Flags & SaturatingNarrowFlags.SignedSrc) != 0; + bool SignedDst = (Flags & SaturatingNarrowFlags.SignedDst) != 0; - int ESize = 8 << Op.Size; + int Elems = !Scalar ? 8 >> Op.Size : 1; int Part = !Scalar && (Op.RegisterSize == ARegisterSize.SIMD128) ? Elems : 0; - long TMaxValue = SignedDst ? (1 << (ESize - 1)) - 1 : (long)(~0UL >> (64 - ESize)); - long TMinValue = SignedDst ? -((1 << (ESize - 1))) : 0; - - Context.EmitLdc_I8(0L); - Context.EmitSttmp(); + if (Scalar) + { + EmitVectorZeroLowerTmp(Context); + } if (Part != 0) { @@ -840,47 +1033,11 @@ namespace ChocolArm64.Instruction for (int Index = 0; Index < Elems; Index++) { - AILLabel LblLe = new AILLabel(); - AILLabel LblGeEnd = new AILLabel(); - EmitVectorExtract(Context, Op.Rn, Index, Op.Size + 1, SignedSrc); Emit(); - Context.Emit(OpCodes.Dup); - - Context.EmitLdc_I8(TMaxValue); - - Context.Emit(SignedSrc ? OpCodes.Ble_S : OpCodes.Ble_Un_S, LblLe); - - Context.Emit(OpCodes.Pop); - - Context.EmitLdc_I8(TMaxValue); - Context.EmitLdc_I8(0x8000000L); - Context.EmitSttmp(); - - Context.Emit(OpCodes.Br_S, LblGeEnd); - - Context.MarkLabel(LblLe); - - Context.Emit(OpCodes.Dup); - - Context.EmitLdc_I8(TMinValue); - - Context.Emit(SignedSrc ? OpCodes.Bge_S : OpCodes.Bge_Un_S, LblGeEnd); - - Context.Emit(OpCodes.Pop); - - Context.EmitLdc_I8(TMinValue); - Context.EmitLdc_I8(0x8000000L); - Context.EmitSttmp(); - - Context.MarkLabel(LblGeEnd); - - if (Scalar) - { - EmitVectorZeroLowerTmp(Context); - } + EmitSatQ(Context, Op.Size, SignedSrc, SignedDst); EmitVectorInsertTmp(Context, Part + Index, Op.Size); } @@ -892,13 +1049,120 @@ namespace ChocolArm64.Instruction { EmitVectorZeroUpper(Context, Op.Rd); } + } + + // TSrc (16bit, 32bit, 64bit; signed, unsigned) > TDst (8bit, 16bit, 32bit; signed, unsigned). + public static void EmitSatQ( + AILEmitterCtx Context, + int SizeDst, + bool SignedSrc, + bool SignedDst) + { + if (SizeDst > 2) + { + throw new ArgumentOutOfRangeException(nameof(SizeDst)); + } + + Context.EmitLdc_I4(SizeDst); + Context.EmitLdarg(ATranslatedSub.StateArgIdx); + + if (SignedSrc) + { + ASoftFallback.EmitCall(Context, SignedDst + ? nameof(ASoftFallback.SignedSrcSignedDstSatQ) + : nameof(ASoftFallback.SignedSrcUnsignedDstSatQ)); + } + else + { + ASoftFallback.EmitCall(Context, SignedDst + ? nameof(ASoftFallback.UnsignedSrcSignedDstSatQ) + : nameof(ASoftFallback.UnsignedSrcUnsignedDstSatQ)); + } + } + + // TSrc (8bit, 16bit, 32bit, 64bit) == TDst (8bit, 16bit, 32bit, 64bit); signed. + public static void EmitUnarySignedSatQAbsOrNeg(AILEmitterCtx Context, int Size) + { + int ESize = 8 << Size; + + long TMaxValue = (1L << (ESize - 1)) - 1L; + long TMinValue = -(1L << (ESize - 1)); + + AILLabel LblFalse = new AILLabel(); + + Context.Emit(OpCodes.Dup); + Context.Emit(OpCodes.Neg); + Context.EmitLdc_I8(TMinValue); + Context.Emit(OpCodes.Ceq); + Context.Emit(OpCodes.Brfalse_S, LblFalse); + + Context.Emit(OpCodes.Pop); + + EmitSetFpsrQCFlag(Context); + + Context.EmitLdc_I8(TMaxValue); + + Context.MarkLabel(LblFalse); + } + + // TSrcs (64bit) == TDst (64bit); signed, unsigned. + public static void EmitBinarySatQAdd(AILEmitterCtx Context, bool Signed) + { + if (((AOpCodeSimdReg)Context.CurrOp).Size < 3) + { + throw new InvalidOperationException(); + } Context.EmitLdarg(ATranslatedSub.StateArgIdx); + + ASoftFallback.EmitCall(Context, Signed + ? nameof(ASoftFallback.BinarySignedSatQAdd) + : nameof(ASoftFallback.BinaryUnsignedSatQAdd)); + } + + // TSrcs (64bit) == TDst (64bit); signed, unsigned. + public static void EmitBinarySatQSub(AILEmitterCtx Context, bool Signed) + { + if (((AOpCodeSimdReg)Context.CurrOp).Size < 3) + { + throw new InvalidOperationException(); + } + + Context.EmitLdarg(ATranslatedSub.StateArgIdx); + + ASoftFallback.EmitCall(Context, Signed + ? nameof(ASoftFallback.BinarySignedSatQSub) + : nameof(ASoftFallback.BinaryUnsignedSatQSub)); + } + + // TSrcs (64bit) == TDst (64bit); signed, unsigned. + public static void EmitBinarySatQAccumulate(AILEmitterCtx Context, bool Signed) + { + if (((AOpCodeSimd)Context.CurrOp).Size < 3) + { + throw new InvalidOperationException(); + } + + Context.EmitLdarg(ATranslatedSub.StateArgIdx); + + ASoftFallback.EmitCall(Context, Signed + ? nameof(ASoftFallback.BinarySignedSatQAcc) + : nameof(ASoftFallback.BinaryUnsignedSatQAcc)); + } + + public static void EmitSetFpsrQCFlag(AILEmitterCtx Context) + { + const int QCFlagBit = 27; + + Context.EmitLdarg(ATranslatedSub.StateArgIdx); + Context.EmitLdarg(ATranslatedSub.StateArgIdx); Context.EmitCallPropGet(typeof(AThreadState), nameof(AThreadState.Fpsr)); - Context.EmitLdtmp(); - Context.Emit(OpCodes.Conv_I4); + + Context.EmitLdc_I4(1 << QCFlagBit); + Context.Emit(OpCodes.Or); + Context.EmitCallPropSet(typeof(AThreadState), nameof(AThreadState.Fpsr)); } diff --git a/ChocolArm64/Instruction/ASoftFallback.cs b/ChocolArm64/Instruction/ASoftFallback.cs index ae3994b029..a4d12dd613 100644 --- a/ChocolArm64/Instruction/ASoftFallback.cs +++ b/ChocolArm64/Instruction/ASoftFallback.cs @@ -1,3 +1,4 @@ +using ChocolArm64.State; using ChocolArm64.Translation; using System; @@ -10,6 +11,273 @@ namespace ChocolArm64.Instruction Context.EmitCall(typeof(ASoftFallback), MthdName); } + public static long BinarySignedSatQAdd(long op1, long op2, AThreadState State) + { + long Add = op1 + op2; + + if ((~(op1 ^ op2) & (op1 ^ Add)) < 0L) + { + SetFpsrQCFlag(State); + + if (op1 < 0L) + { + return long.MinValue; + } + else + { + return long.MaxValue; + } + } + else + { + return Add; + } + } + + public static ulong BinaryUnsignedSatQAdd(ulong op1, ulong op2, AThreadState State) + { + ulong Add = op1 + op2; + + if ((Add < op1) && (Add < op2)) + { + SetFpsrQCFlag(State); + + return ulong.MaxValue; + } + else + { + return Add; + } + } + + public static long BinarySignedSatQSub(long op1, long op2, AThreadState State) + { + long Sub = op1 - op2; + + if (((op1 ^ op2) & (op1 ^ Sub)) < 0L) + { + SetFpsrQCFlag(State); + + if (op1 < 0L) + { + return long.MinValue; + } + else + { + return long.MaxValue; + } + } + else + { + return Sub; + } + } + + public static ulong BinaryUnsignedSatQSub(ulong op1, ulong op2, AThreadState State) + { + ulong Sub = op1 - op2; + + if (op1 < op2) + { + SetFpsrQCFlag(State); + + return ulong.MinValue; + } + else + { + return Sub; + } + } + + public static long BinarySignedSatQAcc(ulong op1, long op2, AThreadState State) + { + if (op1 <= (ulong)long.MaxValue) + { + // op1 from ulong.MinValue to (ulong)long.MaxValue + // op2 from long.MinValue to long.MaxValue + + long Add = (long)op1 + op2; + + if ((~op2 & Add) < 0L) + { + SetFpsrQCFlag(State); + + return long.MaxValue; + } + else + { + return Add; + } + } + else if (op2 >= 0L) + { + // op1 from (ulong)long.MaxValue + 1UL to ulong.MaxValue + // op2 from (long)ulong.MinValue to long.MaxValue + + SetFpsrQCFlag(State); + + return long.MaxValue; + } + else + { + // op1 from (ulong)long.MaxValue + 1UL to ulong.MaxValue + // op2 from long.MinValue to (long)ulong.MinValue - 1L + + ulong Add = op1 + (ulong)op2; + + if (Add > (ulong)long.MaxValue) + { + SetFpsrQCFlag(State); + + return long.MaxValue; + } + else + { + return (long)Add; + } + } + } + + public static ulong BinaryUnsignedSatQAcc(long op1, ulong op2, AThreadState State) + { + if (op1 >= 0L) + { + // op1 from (long)ulong.MinValue to long.MaxValue + // op2 from ulong.MinValue to ulong.MaxValue + + ulong Add = (ulong)op1 + op2; + + if ((Add < (ulong)op1) && (Add < op2)) + { + SetFpsrQCFlag(State); + + return ulong.MaxValue; + } + else + { + return Add; + } + } + else if (op2 > (ulong)long.MaxValue) + { + // op1 from long.MinValue to (long)ulong.MinValue - 1L + // op2 from (ulong)long.MaxValue + 1UL to ulong.MaxValue + + return (ulong)op1 + op2; + } + else + { + // op1 from long.MinValue to (long)ulong.MinValue - 1L + // op2 from ulong.MinValue to (ulong)long.MaxValue + + long Add = op1 + (long)op2; + + if (Add < (long)ulong.MinValue) + { + SetFpsrQCFlag(State); + + return ulong.MinValue; + } + else + { + return (ulong)Add; + } + } + } + + public static long SignedSrcSignedDstSatQ(long op, int Size, AThreadState State) + { + int ESize = 8 << Size; + + long TMaxValue = (1L << (ESize - 1)) - 1L; + long TMinValue = -(1L << (ESize - 1)); + + if (op > TMaxValue) + { + SetFpsrQCFlag(State); + + return TMaxValue; + } + else if (op < TMinValue) + { + SetFpsrQCFlag(State); + + return TMinValue; + } + else + { + return op; + } + } + + public static ulong SignedSrcUnsignedDstSatQ(long op, int Size, AThreadState State) + { + int ESize = 8 << Size; + + ulong TMaxValue = (1UL << ESize) - 1UL; + ulong TMinValue = 0UL; + + if (op > (long)TMaxValue) + { + SetFpsrQCFlag(State); + + return TMaxValue; + } + else if (op < (long)TMinValue) + { + SetFpsrQCFlag(State); + + return TMinValue; + } + else + { + return (ulong)op; + } + } + + public static long UnsignedSrcSignedDstSatQ(ulong op, int Size, AThreadState State) + { + int ESize = 8 << Size; + + long TMaxValue = (1L << (ESize - 1)) - 1L; + + if (op > (ulong)TMaxValue) + { + SetFpsrQCFlag(State); + + return TMaxValue; + } + else + { + return (long)op; + } + } + + public static ulong UnsignedSrcUnsignedDstSatQ(ulong op, int Size, AThreadState State) + { + int ESize = 8 << Size; + + ulong TMaxValue = (1UL << ESize) - 1UL; + + if (op > TMaxValue) + { + SetFpsrQCFlag(State); + + return TMaxValue; + } + else + { + return op; + } + } + + private static void SetFpsrQCFlag(AThreadState State) + { + const int QCFlagBit = 27; + + State.Fpsr |= 1 << QCFlagBit; + } + public static ulong CountLeadingSigns(ulong Value, int Size) { Value ^= Value >> 1; diff --git a/Ryujinx.Tests/Cpu/CpuTestSimd.cs b/Ryujinx.Tests/Cpu/CpuTestSimd.cs index 82591edaee..88c5981cd2 100644 --- a/Ryujinx.Tests/Cpu/CpuTestSimd.cs +++ b/Ryujinx.Tests/Cpu/CpuTestSimd.cs @@ -22,6 +22,17 @@ namespace Ryujinx.Tests.Cpu } #region "ValueSource" + private static ulong[] _1B1H1S1D_() + { + return new ulong[] { 0x0000000000000000ul, 0x000000000000007Ful, + 0x0000000000000080ul, 0x00000000000000FFul, + 0x0000000000007FFFul, 0x0000000000008000ul, + 0x000000000000FFFFul, 0x000000007FFFFFFFul, + 0x0000000080000000ul, 0x00000000FFFFFFFFul, + 0x7FFFFFFFFFFFFFFFul, 0x8000000000000000ul, + 0xFFFFFFFFFFFFFFFFul }; + } + private static ulong[] _1D_() { return new ulong[] { 0x0000000000000000ul, 0x7FFFFFFFFFFFFFFFul, @@ -1126,6 +1137,192 @@ namespace Ryujinx.Tests.Cpu }); } + [Test, Description("SQABS , ")] + public void Sqabs_S_B_H_S_D([Values(0u)] uint Rd, + [Values(1u, 0u)] uint Rn, + [ValueSource("_1B1H1S1D_")] [Random(RndCnt)] ulong Z, + [ValueSource("_1B1H1S1D_")] [Random(RndCnt)] ulong A, + [Values(0b00u, 0b01u, 0b10u, 0b11u)] uint size) // + { + uint Opcode = 0x5E207800; // SQABS B0, B0 + Opcode |= ((Rn & 31) << 5) | ((Rd & 31) << 0); + Opcode |= ((size & 3) << 22); + Bits Op = new Bits(Opcode); + + int Fpsr = (int)TestContext.CurrentContext.Random.NextUInt(); + + Vector128 V0 = MakeVectorE0E1(Z, Z); + Vector128 V1 = MakeVectorE0(A); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, Fpsr: Fpsr); + + AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); + AArch64.V(1, new Bits(A)); + Shared.FPSR = new Bits((uint)Fpsr); + SimdFp.Sqabs_S(Op[23, 22], Op[9, 5], Op[4, 0]); + + Assert.Multiple(() => + { + Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); + Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); + }); + Assert.That(ThreadState.Fpsr, Is.EqualTo((int)Shared.FPSR.ToUInt32())); + } + + [Test, Description("SQABS ., .")] + public void Sqabs_V_8B_4H_2S([Values(0u)] uint Rd, + [Values(1u, 0u)] uint Rn, + [ValueSource("_8B4H2S_")] [Random(RndCnt)] ulong Z, + [ValueSource("_8B4H2S_")] [Random(RndCnt)] ulong A, + [Values(0b00u, 0b01u, 0b10u)] uint size) // <8B, 4H, 2S> + { + uint Opcode = 0x0E207800; // SQABS V0.8B, V0.8B + Opcode |= ((Rn & 31) << 5) | ((Rd & 31) << 0); + Opcode |= ((size & 3) << 22); + Bits Op = new Bits(Opcode); + + int Fpsr = (int)TestContext.CurrentContext.Random.NextUInt(); + + Vector128 V0 = MakeVectorE0E1(Z, Z); + Vector128 V1 = MakeVectorE0(A); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, Fpsr: Fpsr); + + AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); + AArch64.V(1, new Bits(A)); + Shared.FPSR = new Bits((uint)Fpsr); + SimdFp.Sqabs_V(Op[30], Op[23, 22], Op[9, 5], Op[4, 0]); + + Assert.Multiple(() => + { + Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); + Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); + }); + Assert.That(ThreadState.Fpsr, Is.EqualTo((int)Shared.FPSR.ToUInt32())); + } + + [Test, Description("SQABS ., .")] + public void Sqabs_V_16B_8H_4S_2D([Values(0u)] uint Rd, + [Values(1u, 0u)] uint Rn, + [ValueSource("_8B4H2S1D_")] [Random(RndCnt)] ulong Z, + [ValueSource("_8B4H2S1D_")] [Random(RndCnt)] ulong A, + [Values(0b00u, 0b01u, 0b10u, 0b11u)] uint size) // <16B, 8H, 4S, 2D> + { + uint Opcode = 0x4E207800; // SQABS V0.16B, V0.16B + Opcode |= ((Rn & 31) << 5) | ((Rd & 31) << 0); + Opcode |= ((size & 3) << 22); + Bits Op = new Bits(Opcode); + + int Fpsr = (int)TestContext.CurrentContext.Random.NextUInt(); + + Vector128 V0 = MakeVectorE0E1(Z, Z); + Vector128 V1 = MakeVectorE0E1(A, A); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, Fpsr: Fpsr); + + AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); + AArch64.Vpart(1, 0, new Bits(A)); AArch64.Vpart(1, 1, new Bits(A)); + Shared.FPSR = new Bits((uint)Fpsr); + SimdFp.Sqabs_V(Op[30], Op[23, 22], Op[9, 5], Op[4, 0]); + + Assert.Multiple(() => + { + Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); + Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); + }); + Assert.That(ThreadState.Fpsr, Is.EqualTo((int)Shared.FPSR.ToUInt32())); + } + + [Test, Description("SQNEG , ")] + public void Sqneg_S_B_H_S_D([Values(0u)] uint Rd, + [Values(1u, 0u)] uint Rn, + [ValueSource("_1B1H1S1D_")] [Random(RndCnt)] ulong Z, + [ValueSource("_1B1H1S1D_")] [Random(RndCnt)] ulong A, + [Values(0b00u, 0b01u, 0b10u, 0b11u)] uint size) // + { + uint Opcode = 0x7E207800; // SQNEG B0, B0 + Opcode |= ((Rn & 31) << 5) | ((Rd & 31) << 0); + Opcode |= ((size & 3) << 22); + Bits Op = new Bits(Opcode); + + int Fpsr = (int)TestContext.CurrentContext.Random.NextUInt(); + + Vector128 V0 = MakeVectorE0E1(Z, Z); + Vector128 V1 = MakeVectorE0(A); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, Fpsr: Fpsr); + + AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); + AArch64.V(1, new Bits(A)); + Shared.FPSR = new Bits((uint)Fpsr); + SimdFp.Sqneg_S(Op[23, 22], Op[9, 5], Op[4, 0]); + + Assert.Multiple(() => + { + Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); + Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); + }); + Assert.That(ThreadState.Fpsr, Is.EqualTo((int)Shared.FPSR.ToUInt32())); + } + + [Test, Description("SQNEG ., .")] + public void Sqneg_V_8B_4H_2S([Values(0u)] uint Rd, + [Values(1u, 0u)] uint Rn, + [ValueSource("_8B4H2S_")] [Random(RndCnt)] ulong Z, + [ValueSource("_8B4H2S_")] [Random(RndCnt)] ulong A, + [Values(0b00u, 0b01u, 0b10u)] uint size) // <8B, 4H, 2S> + { + uint Opcode = 0x2E207800; // SQNEG V0.8B, V0.8B + Opcode |= ((Rn & 31) << 5) | ((Rd & 31) << 0); + Opcode |= ((size & 3) << 22); + Bits Op = new Bits(Opcode); + + int Fpsr = (int)TestContext.CurrentContext.Random.NextUInt(); + + Vector128 V0 = MakeVectorE0E1(Z, Z); + Vector128 V1 = MakeVectorE0(A); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, Fpsr: Fpsr); + + AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); + AArch64.V(1, new Bits(A)); + Shared.FPSR = new Bits((uint)Fpsr); + SimdFp.Sqneg_V(Op[30], Op[23, 22], Op[9, 5], Op[4, 0]); + + Assert.Multiple(() => + { + Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); + Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); + }); + Assert.That(ThreadState.Fpsr, Is.EqualTo((int)Shared.FPSR.ToUInt32())); + } + + [Test, Description("SQNEG ., .")] + public void Sqneg_V_16B_8H_4S_2D([Values(0u)] uint Rd, + [Values(1u, 0u)] uint Rn, + [ValueSource("_8B4H2S1D_")] [Random(RndCnt)] ulong Z, + [ValueSource("_8B4H2S1D_")] [Random(RndCnt)] ulong A, + [Values(0b00u, 0b01u, 0b10u, 0b11u)] uint size) // <16B, 8H, 4S, 2D> + { + uint Opcode = 0x6E207800; // SQNEG V0.16B, V0.16B + Opcode |= ((Rn & 31) << 5) | ((Rd & 31) << 0); + Opcode |= ((size & 3) << 22); + Bits Op = new Bits(Opcode); + + int Fpsr = (int)TestContext.CurrentContext.Random.NextUInt(); + + Vector128 V0 = MakeVectorE0E1(Z, Z); + Vector128 V1 = MakeVectorE0E1(A, A); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, Fpsr: Fpsr); + + AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); + AArch64.Vpart(1, 0, new Bits(A)); AArch64.Vpart(1, 1, new Bits(A)); + Shared.FPSR = new Bits((uint)Fpsr); + SimdFp.Sqneg_V(Op[30], Op[23, 22], Op[9, 5], Op[4, 0]); + + Assert.Multiple(() => + { + Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); + Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); + }); + Assert.That(ThreadState.Fpsr, Is.EqualTo((int)Shared.FPSR.ToUInt32())); + } + [Test, Description("SQXTN , ")] public void Sqxtn_S_HB_SH_DS([Values(0u)] uint Rd, [Values(1u, 0u)] uint Rn, @@ -1138,12 +1335,15 @@ namespace Ryujinx.Tests.Cpu Opcode |= ((size & 3) << 22); Bits Op = new Bits(Opcode); + int Fpsr = (int)TestContext.CurrentContext.Random.NextUInt(); + Vector128 V0 = MakeVectorE0E1(Z, Z); Vector128 V1 = MakeVectorE0(A); - AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, Fpsr: Fpsr); AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); AArch64.V(1, new Bits(A)); + Shared.FPSR = new Bits((uint)Fpsr); SimdFp.Sqxtn_S(Op[23, 22], Op[9, 5], Op[4, 0]); Assert.Multiple(() => @@ -1151,7 +1351,7 @@ namespace Ryujinx.Tests.Cpu Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); }); - Assert.That(((ThreadState.Fpsr >> 27) & 1) != 0, Is.EqualTo(Shared.FPSR[27])); + Assert.That(ThreadState.Fpsr, Is.EqualTo((int)Shared.FPSR.ToUInt32())); } [Test, Description("SQXTN{2} ., .")] @@ -1166,12 +1366,15 @@ namespace Ryujinx.Tests.Cpu Opcode |= ((size & 3) << 22); Bits Op = new Bits(Opcode); + int Fpsr = (int)TestContext.CurrentContext.Random.NextUInt(); + Vector128 V0 = MakeVectorE0E1(Z, Z); Vector128 V1 = MakeVectorE0E1(A, A); - AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, Fpsr: Fpsr); AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); AArch64.Vpart(1, 0, new Bits(A)); AArch64.Vpart(1, 1, new Bits(A)); + Shared.FPSR = new Bits((uint)Fpsr); SimdFp.Sqxtn_V(Op[30], Op[23, 22], Op[9, 5], Op[4, 0]); Assert.Multiple(() => @@ -1179,7 +1382,7 @@ namespace Ryujinx.Tests.Cpu Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); }); - Assert.That(((ThreadState.Fpsr >> 27) & 1) != 0, Is.EqualTo(Shared.FPSR[27])); + Assert.That(ThreadState.Fpsr, Is.EqualTo((int)Shared.FPSR.ToUInt32())); } [Test, Description("SQXTN{2} ., .")] @@ -1194,12 +1397,15 @@ namespace Ryujinx.Tests.Cpu Opcode |= ((size & 3) << 22); Bits Op = new Bits(Opcode); + int Fpsr = (int)TestContext.CurrentContext.Random.NextUInt(); + Vector128 V0 = MakeVectorE0E1(Z, Z); Vector128 V1 = MakeVectorE0E1(A, A); - AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, Fpsr: Fpsr); AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); AArch64.Vpart(1, 0, new Bits(A)); AArch64.Vpart(1, 1, new Bits(A)); + Shared.FPSR = new Bits((uint)Fpsr); SimdFp.Sqxtn_V(Op[30], Op[23, 22], Op[9, 5], Op[4, 0]); Assert.Multiple(() => @@ -1207,7 +1413,7 @@ namespace Ryujinx.Tests.Cpu Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); }); - Assert.That(((ThreadState.Fpsr >> 27) & 1) != 0, Is.EqualTo(Shared.FPSR[27])); + Assert.That(ThreadState.Fpsr, Is.EqualTo((int)Shared.FPSR.ToUInt32())); } [Test, Description("SQXTUN , ")] @@ -1222,12 +1428,15 @@ namespace Ryujinx.Tests.Cpu Opcode |= ((size & 3) << 22); Bits Op = new Bits(Opcode); + int Fpsr = (int)TestContext.CurrentContext.Random.NextUInt(); + Vector128 V0 = MakeVectorE0E1(Z, Z); Vector128 V1 = MakeVectorE0(A); - AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, Fpsr: Fpsr); AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); AArch64.V(1, new Bits(A)); + Shared.FPSR = new Bits((uint)Fpsr); SimdFp.Sqxtun_S(Op[23, 22], Op[9, 5], Op[4, 0]); Assert.Multiple(() => @@ -1235,7 +1444,7 @@ namespace Ryujinx.Tests.Cpu Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); }); - Assert.That(((ThreadState.Fpsr >> 27) & 1) != 0, Is.EqualTo(Shared.FPSR[27])); + Assert.That(ThreadState.Fpsr, Is.EqualTo((int)Shared.FPSR.ToUInt32())); } [Test, Description("SQXTUN{2} ., .")] @@ -1250,12 +1459,15 @@ namespace Ryujinx.Tests.Cpu Opcode |= ((size & 3) << 22); Bits Op = new Bits(Opcode); + int Fpsr = (int)TestContext.CurrentContext.Random.NextUInt(); + Vector128 V0 = MakeVectorE0E1(Z, Z); Vector128 V1 = MakeVectorE0E1(A, A); - AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, Fpsr: Fpsr); AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); AArch64.Vpart(1, 0, new Bits(A)); AArch64.Vpart(1, 1, new Bits(A)); + Shared.FPSR = new Bits((uint)Fpsr); SimdFp.Sqxtun_V(Op[30], Op[23, 22], Op[9, 5], Op[4, 0]); Assert.Multiple(() => @@ -1263,7 +1475,7 @@ namespace Ryujinx.Tests.Cpu Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); }); - Assert.That(((ThreadState.Fpsr >> 27) & 1) != 0, Is.EqualTo(Shared.FPSR[27])); + Assert.That(ThreadState.Fpsr, Is.EqualTo((int)Shared.FPSR.ToUInt32())); } [Test, Description("SQXTUN{2} ., .")] @@ -1278,12 +1490,15 @@ namespace Ryujinx.Tests.Cpu Opcode |= ((size & 3) << 22); Bits Op = new Bits(Opcode); + int Fpsr = (int)TestContext.CurrentContext.Random.NextUInt(); + Vector128 V0 = MakeVectorE0E1(Z, Z); Vector128 V1 = MakeVectorE0E1(A, A); - AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, Fpsr: Fpsr); AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); AArch64.Vpart(1, 0, new Bits(A)); AArch64.Vpart(1, 1, new Bits(A)); + Shared.FPSR = new Bits((uint)Fpsr); SimdFp.Sqxtun_V(Op[30], Op[23, 22], Op[9, 5], Op[4, 0]); Assert.Multiple(() => @@ -1291,7 +1506,100 @@ namespace Ryujinx.Tests.Cpu Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); }); - Assert.That(((ThreadState.Fpsr >> 27) & 1) != 0, Is.EqualTo(Shared.FPSR[27])); + Assert.That(ThreadState.Fpsr, Is.EqualTo((int)Shared.FPSR.ToUInt32())); + } + + [Test, Description("SUQADD , ")] + public void Suqadd_S_B_H_S_D([Values(0u)] uint Rd, + [Values(1u, 0u)] uint Rn, + [ValueSource("_1B1H1S1D_")] [Random(RndCnt)] ulong Z, + [ValueSource("_1B1H1S1D_")] [Random(RndCnt)] ulong A, + [Values(0b00u, 0b01u, 0b10u, 0b11u)] uint size) // + { + uint Opcode = 0x5E203800; // SUQADD B0, B0 + Opcode |= ((Rn & 31) << 5) | ((Rd & 31) << 0); + Opcode |= ((size & 3) << 22); + Bits Op = new Bits(Opcode); + + int Fpsr = (int)TestContext.CurrentContext.Random.NextUInt(); + + Vector128 V0 = MakeVectorE0E1(Z, Z); + Vector128 V1 = MakeVectorE0(A); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, Fpsr: Fpsr); + + AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); + AArch64.V(1, new Bits(A)); + Shared.FPSR = new Bits((uint)Fpsr); + SimdFp.Suqadd_S(Op[23, 22], Op[9, 5], Op[4, 0]); + + Assert.Multiple(() => + { + Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); + Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); + }); + Assert.That(ThreadState.Fpsr, Is.EqualTo((int)Shared.FPSR.ToUInt32())); + } + + [Test, Description("SUQADD ., .")] + public void Suqadd_V_8B_4H_2S([Values(0u)] uint Rd, + [Values(1u, 0u)] uint Rn, + [ValueSource("_8B4H2S_")] [Random(RndCnt)] ulong Z, + [ValueSource("_8B4H2S_")] [Random(RndCnt)] ulong A, + [Values(0b00u, 0b01u, 0b10u)] uint size) // <8B, 4H, 2S> + { + uint Opcode = 0x0E203800; // SUQADD V0.8B, V0.8B + Opcode |= ((Rn & 31) << 5) | ((Rd & 31) << 0); + Opcode |= ((size & 3) << 22); + Bits Op = new Bits(Opcode); + + int Fpsr = (int)TestContext.CurrentContext.Random.NextUInt(); + + Vector128 V0 = MakeVectorE0E1(Z, Z); + Vector128 V1 = MakeVectorE0(A); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, Fpsr: Fpsr); + + AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); + AArch64.V(1, new Bits(A)); + Shared.FPSR = new Bits((uint)Fpsr); + SimdFp.Suqadd_V(Op[30], Op[23, 22], Op[9, 5], Op[4, 0]); + + Assert.Multiple(() => + { + Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); + Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); + }); + Assert.That(ThreadState.Fpsr, Is.EqualTo((int)Shared.FPSR.ToUInt32())); + } + + [Test, Description("SUQADD ., .")] + public void Suqadd_V_16B_8H_4S_2D([Values(0u)] uint Rd, + [Values(1u, 0u)] uint Rn, + [ValueSource("_8B4H2S1D_")] [Random(RndCnt)] ulong Z, + [ValueSource("_8B4H2S1D_")] [Random(RndCnt)] ulong A, + [Values(0b00u, 0b01u, 0b10u, 0b11u)] uint size) // <16B, 8H, 4S, 2D> + { + uint Opcode = 0x4E203800; // SUQADD V0.16B, V0.16B + Opcode |= ((Rn & 31) << 5) | ((Rd & 31) << 0); + Opcode |= ((size & 3) << 22); + Bits Op = new Bits(Opcode); + + int Fpsr = (int)TestContext.CurrentContext.Random.NextUInt(); + + Vector128 V0 = MakeVectorE0E1(Z, Z); + Vector128 V1 = MakeVectorE0E1(A, A); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, Fpsr: Fpsr); + + AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); + AArch64.Vpart(1, 0, new Bits(A)); AArch64.Vpart(1, 1, new Bits(A)); + Shared.FPSR = new Bits((uint)Fpsr); + SimdFp.Suqadd_V(Op[30], Op[23, 22], Op[9, 5], Op[4, 0]); + + Assert.Multiple(() => + { + Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); + Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); + }); + Assert.That(ThreadState.Fpsr, Is.EqualTo((int)Shared.FPSR.ToUInt32())); } [Test, Description("UQXTN , ")] @@ -1306,12 +1614,15 @@ namespace Ryujinx.Tests.Cpu Opcode |= ((size & 3) << 22); Bits Op = new Bits(Opcode); + int Fpsr = (int)TestContext.CurrentContext.Random.NextUInt(); + Vector128 V0 = MakeVectorE0E1(Z, Z); Vector128 V1 = MakeVectorE0(A); - AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, Fpsr: Fpsr); AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); AArch64.V(1, new Bits(A)); + Shared.FPSR = new Bits((uint)Fpsr); SimdFp.Uqxtn_S(Op[23, 22], Op[9, 5], Op[4, 0]); Assert.Multiple(() => @@ -1319,7 +1630,7 @@ namespace Ryujinx.Tests.Cpu Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); }); - Assert.That(((ThreadState.Fpsr >> 27) & 1) != 0, Is.EqualTo(Shared.FPSR[27])); + Assert.That(ThreadState.Fpsr, Is.EqualTo((int)Shared.FPSR.ToUInt32())); } [Test, Description("UQXTN{2} ., .")] @@ -1334,12 +1645,15 @@ namespace Ryujinx.Tests.Cpu Opcode |= ((size & 3) << 22); Bits Op = new Bits(Opcode); + int Fpsr = (int)TestContext.CurrentContext.Random.NextUInt(); + Vector128 V0 = MakeVectorE0E1(Z, Z); Vector128 V1 = MakeVectorE0E1(A, A); - AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, Fpsr: Fpsr); AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); AArch64.Vpart(1, 0, new Bits(A)); AArch64.Vpart(1, 1, new Bits(A)); + Shared.FPSR = new Bits((uint)Fpsr); SimdFp.Uqxtn_V(Op[30], Op[23, 22], Op[9, 5], Op[4, 0]); Assert.Multiple(() => @@ -1347,7 +1661,7 @@ namespace Ryujinx.Tests.Cpu Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); }); - Assert.That(((ThreadState.Fpsr >> 27) & 1) != 0, Is.EqualTo(Shared.FPSR[27])); + Assert.That(ThreadState.Fpsr, Is.EqualTo((int)Shared.FPSR.ToUInt32())); } [Test, Description("UQXTN{2} ., .")] @@ -1362,12 +1676,15 @@ namespace Ryujinx.Tests.Cpu Opcode |= ((size & 3) << 22); Bits Op = new Bits(Opcode); + int Fpsr = (int)TestContext.CurrentContext.Random.NextUInt(); + Vector128 V0 = MakeVectorE0E1(Z, Z); Vector128 V1 = MakeVectorE0E1(A, A); - AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, Fpsr: Fpsr); AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); AArch64.Vpart(1, 0, new Bits(A)); AArch64.Vpart(1, 1, new Bits(A)); + Shared.FPSR = new Bits((uint)Fpsr); SimdFp.Uqxtn_V(Op[30], Op[23, 22], Op[9, 5], Op[4, 0]); Assert.Multiple(() => @@ -1375,7 +1692,100 @@ namespace Ryujinx.Tests.Cpu Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); }); - Assert.That(((ThreadState.Fpsr >> 27) & 1) != 0, Is.EqualTo(Shared.FPSR[27])); + Assert.That(ThreadState.Fpsr, Is.EqualTo((int)Shared.FPSR.ToUInt32())); + } + + [Test, Description("USQADD , ")] + public void Usqadd_S_B_H_S_D([Values(0u)] uint Rd, + [Values(1u, 0u)] uint Rn, + [ValueSource("_1B1H1S1D_")] [Random(RndCnt)] ulong Z, + [ValueSource("_1B1H1S1D_")] [Random(RndCnt)] ulong A, + [Values(0b00u, 0b01u, 0b10u, 0b11u)] uint size) // + { + uint Opcode = 0x7E203800; // USQADD B0, B0 + Opcode |= ((Rn & 31) << 5) | ((Rd & 31) << 0); + Opcode |= ((size & 3) << 22); + Bits Op = new Bits(Opcode); + + int Fpsr = (int)TestContext.CurrentContext.Random.NextUInt(); + + Vector128 V0 = MakeVectorE0E1(Z, Z); + Vector128 V1 = MakeVectorE0(A); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, Fpsr: Fpsr); + + AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); + AArch64.V(1, new Bits(A)); + Shared.FPSR = new Bits((uint)Fpsr); + SimdFp.Usqadd_S(Op[23, 22], Op[9, 5], Op[4, 0]); + + Assert.Multiple(() => + { + Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); + Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); + }); + Assert.That(ThreadState.Fpsr, Is.EqualTo((int)Shared.FPSR.ToUInt32())); + } + + [Test, Description("USQADD ., .")] + public void Usqadd_V_8B_4H_2S([Values(0u)] uint Rd, + [Values(1u, 0u)] uint Rn, + [ValueSource("_8B4H2S_")] [Random(RndCnt)] ulong Z, + [ValueSource("_8B4H2S_")] [Random(RndCnt)] ulong A, + [Values(0b00u, 0b01u, 0b10u)] uint size) // <8B, 4H, 2S> + { + uint Opcode = 0x2E203800; // USQADD V0.8B, V0.8B + Opcode |= ((Rn & 31) << 5) | ((Rd & 31) << 0); + Opcode |= ((size & 3) << 22); + Bits Op = new Bits(Opcode); + + int Fpsr = (int)TestContext.CurrentContext.Random.NextUInt(); + + Vector128 V0 = MakeVectorE0E1(Z, Z); + Vector128 V1 = MakeVectorE0(A); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, Fpsr: Fpsr); + + AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); + AArch64.V(1, new Bits(A)); + Shared.FPSR = new Bits((uint)Fpsr); + SimdFp.Usqadd_V(Op[30], Op[23, 22], Op[9, 5], Op[4, 0]); + + Assert.Multiple(() => + { + Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); + Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); + }); + Assert.That(ThreadState.Fpsr, Is.EqualTo((int)Shared.FPSR.ToUInt32())); + } + + [Test, Description("USQADD ., .")] + public void Usqadd_V_16B_8H_4S_2D([Values(0u)] uint Rd, + [Values(1u, 0u)] uint Rn, + [ValueSource("_8B4H2S1D_")] [Random(RndCnt)] ulong Z, + [ValueSource("_8B4H2S1D_")] [Random(RndCnt)] ulong A, + [Values(0b00u, 0b01u, 0b10u, 0b11u)] uint size) // <16B, 8H, 4S, 2D> + { + uint Opcode = 0x6E203800; // USQADD V0.16B, V0.16B + Opcode |= ((Rn & 31) << 5) | ((Rd & 31) << 0); + Opcode |= ((size & 3) << 22); + Bits Op = new Bits(Opcode); + + int Fpsr = (int)TestContext.CurrentContext.Random.NextUInt(); + + Vector128 V0 = MakeVectorE0E1(Z, Z); + Vector128 V1 = MakeVectorE0E1(A, A); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, Fpsr: Fpsr); + + AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); + AArch64.Vpart(1, 0, new Bits(A)); AArch64.Vpart(1, 1, new Bits(A)); + Shared.FPSR = new Bits((uint)Fpsr); + SimdFp.Usqadd_V(Op[30], Op[23, 22], Op[9, 5], Op[4, 0]); + + Assert.Multiple(() => + { + Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); + Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); + }); + Assert.That(ThreadState.Fpsr, Is.EqualTo((int)Shared.FPSR.ToUInt32())); } [Test, Description("XTN{2} ., .")] diff --git a/Ryujinx.Tests/Cpu/CpuTestSimdReg.cs b/Ryujinx.Tests/Cpu/CpuTestSimdReg.cs index c67348d1da..e6cfcbde5b 100644 --- a/Ryujinx.Tests/Cpu/CpuTestSimdReg.cs +++ b/Ryujinx.Tests/Cpu/CpuTestSimdReg.cs @@ -22,6 +22,17 @@ namespace Ryujinx.Tests.Cpu } #region "ValueSource" + private static ulong[] _1B1H1S1D_() + { + return new ulong[] { 0x0000000000000000ul, 0x000000000000007Ful, + 0x0000000000000080ul, 0x00000000000000FFul, + 0x0000000000007FFFul, 0x0000000000008000ul, + 0x000000000000FFFFul, 0x000000007FFFFFFFul, + 0x0000000080000000ul, 0x00000000FFFFFFFFul, + 0x7FFFFFFFFFFFFFFFul, 0x8000000000000000ul, + 0xFFFFFFFFFFFFFFFFul }; + } + private static ulong[] _1D_() { return new ulong[] { 0x0000000000000000ul, 0x7FFFFFFFFFFFFFFFul, @@ -1721,6 +1732,216 @@ namespace Ryujinx.Tests.Cpu }); } + [Test, Pairwise, Description("SQADD , , ")] + public void Sqadd_S_B_H_S_D([Values(0u)] uint Rd, + [Values(1u, 0u)] uint Rn, + [Values(2u, 0u)] uint Rm, + [ValueSource("_1B1H1S1D_")] [Random(RndCnt)] ulong Z, + [ValueSource("_1B1H1S1D_")] [Random(RndCnt)] ulong A, + [ValueSource("_1B1H1S1D_")] [Random(RndCnt)] ulong B, + [Values(0b00u, 0b01u, 0b10u, 0b11u)] uint size) // + { + uint Opcode = 0x5E200C00; // SQADD B0, B0, B0 + Opcode |= ((Rm & 31) << 16) | ((Rn & 31) << 5) | ((Rd & 31) << 0); + Opcode |= ((size & 3) << 22); + Bits Op = new Bits(Opcode); + + int Fpsr = (int)TestContext.CurrentContext.Random.NextUInt(); + + Vector128 V0 = MakeVectorE0E1(Z, Z); + Vector128 V1 = MakeVectorE0(A); + Vector128 V2 = MakeVectorE0(B); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, V2: V2, Fpsr: Fpsr); + + AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); + AArch64.V(1, new Bits(A)); + AArch64.V(2, new Bits(B)); + Shared.FPSR = new Bits((uint)Fpsr); + SimdFp.Sqadd_S(Op[23, 22], Op[20, 16], Op[9, 5], Op[4, 0]); + + Assert.Multiple(() => + { + Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); + Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); + }); + Assert.That(ThreadState.Fpsr, Is.EqualTo((int)Shared.FPSR.ToUInt32())); + } + + [Test, Pairwise, Description("SQADD ., ., .")] + public void Sqadd_V_8B_4H_2S([Values(0u)] uint Rd, + [Values(1u, 0u)] uint Rn, + [Values(2u, 0u)] uint Rm, + [ValueSource("_8B4H2S_")] [Random(RndCnt)] ulong Z, + [ValueSource("_8B4H2S_")] [Random(RndCnt)] ulong A, + [ValueSource("_8B4H2S_")] [Random(RndCnt)] ulong B, + [Values(0b00u, 0b01u, 0b10u)] uint size) // <8B, 4H, 2S> + { + uint Opcode = 0x0E200C00; // SQADD V0.8B, V0.8B, V0.8B + Opcode |= ((Rm & 31) << 16) | ((Rn & 31) << 5) | ((Rd & 31) << 0); + Opcode |= ((size & 3) << 22); + Bits Op = new Bits(Opcode); + + int Fpsr = (int)TestContext.CurrentContext.Random.NextUInt(); + + Vector128 V0 = MakeVectorE0E1(Z, Z); + Vector128 V1 = MakeVectorE0(A); + Vector128 V2 = MakeVectorE0(B); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, V2: V2, Fpsr: Fpsr); + + AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); + AArch64.V(1, new Bits(A)); + AArch64.V(2, new Bits(B)); + Shared.FPSR = new Bits((uint)Fpsr); + SimdFp.Sqadd_V(Op[30], Op[23, 22], Op[20, 16], Op[9, 5], Op[4, 0]); + + Assert.Multiple(() => + { + Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); + Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); + }); + Assert.That(ThreadState.Fpsr, Is.EqualTo((int)Shared.FPSR.ToUInt32())); + } + + [Test, Pairwise, Description("SQADD ., ., .")] + public void Sqadd_V_16B_8H_4S_2D([Values(0u)] uint Rd, + [Values(1u, 0u)] uint Rn, + [Values(2u, 0u)] uint Rm, + [ValueSource("_8B4H2S1D_")] [Random(RndCnt)] ulong Z, + [ValueSource("_8B4H2S1D_")] [Random(RndCnt)] ulong A, + [ValueSource("_8B4H2S1D_")] [Random(RndCnt)] ulong B, + [Values(0b00u, 0b01u, 0b10u, 0b11u)] uint size) // <16B, 8H, 4S, 2D> + { + uint Opcode = 0x4E200C00; // SQADD V0.16B, V0.16B, V0.16B + Opcode |= ((Rm & 31) << 16) | ((Rn & 31) << 5) | ((Rd & 31) << 0); + Opcode |= ((size & 3) << 22); + Bits Op = new Bits(Opcode); + + int Fpsr = (int)TestContext.CurrentContext.Random.NextUInt(); + + Vector128 V0 = MakeVectorE0E1(Z, Z); + Vector128 V1 = MakeVectorE0E1(A, A); + Vector128 V2 = MakeVectorE0E1(B, B); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, V2: V2, Fpsr: Fpsr); + + AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); + AArch64.Vpart(1, 0, new Bits(A)); AArch64.Vpart(1, 1, new Bits(A)); + AArch64.Vpart(2, 0, new Bits(B)); AArch64.Vpart(2, 1, new Bits(B)); + Shared.FPSR = new Bits((uint)Fpsr); + SimdFp.Sqadd_V(Op[30], Op[23, 22], Op[20, 16], Op[9, 5], Op[4, 0]); + + Assert.Multiple(() => + { + Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); + Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); + }); + Assert.That(ThreadState.Fpsr, Is.EqualTo((int)Shared.FPSR.ToUInt32())); + } + + [Test, Pairwise, Description("SQSUB , , ")] + public void Sqsub_S_B_H_S_D([Values(0u)] uint Rd, + [Values(1u, 0u)] uint Rn, + [Values(2u, 0u)] uint Rm, + [ValueSource("_1B1H1S1D_")] [Random(RndCnt)] ulong Z, + [ValueSource("_1B1H1S1D_")] [Random(RndCnt)] ulong A, + [ValueSource("_1B1H1S1D_")] [Random(RndCnt)] ulong B, + [Values(0b00u, 0b01u, 0b10u, 0b11u)] uint size) // + { + uint Opcode = 0x5E202C00; // SQSUB B0, B0, B0 + Opcode |= ((Rm & 31) << 16) | ((Rn & 31) << 5) | ((Rd & 31) << 0); + Opcode |= ((size & 3) << 22); + Bits Op = new Bits(Opcode); + + int Fpsr = (int)TestContext.CurrentContext.Random.NextUInt(); + + Vector128 V0 = MakeVectorE0E1(Z, Z); + Vector128 V1 = MakeVectorE0(A); + Vector128 V2 = MakeVectorE0(B); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, V2: V2, Fpsr: Fpsr); + + AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); + AArch64.V(1, new Bits(A)); + AArch64.V(2, new Bits(B)); + Shared.FPSR = new Bits((uint)Fpsr); + SimdFp.Sqsub_S(Op[23, 22], Op[20, 16], Op[9, 5], Op[4, 0]); + + Assert.Multiple(() => + { + Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); + Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); + }); + Assert.That(ThreadState.Fpsr, Is.EqualTo((int)Shared.FPSR.ToUInt32())); + } + + [Test, Pairwise, Description("SQSUB ., ., .")] + public void Sqsub_V_8B_4H_2S([Values(0u)] uint Rd, + [Values(1u, 0u)] uint Rn, + [Values(2u, 0u)] uint Rm, + [ValueSource("_8B4H2S_")] [Random(RndCnt)] ulong Z, + [ValueSource("_8B4H2S_")] [Random(RndCnt)] ulong A, + [ValueSource("_8B4H2S_")] [Random(RndCnt)] ulong B, + [Values(0b00u, 0b01u, 0b10u)] uint size) // <8B, 4H, 2S> + { + uint Opcode = 0x0E202C00; // SQSUB V0.8B, V0.8B, V0.8B + Opcode |= ((Rm & 31) << 16) | ((Rn & 31) << 5) | ((Rd & 31) << 0); + Opcode |= ((size & 3) << 22); + Bits Op = new Bits(Opcode); + + int Fpsr = (int)TestContext.CurrentContext.Random.NextUInt(); + + Vector128 V0 = MakeVectorE0E1(Z, Z); + Vector128 V1 = MakeVectorE0(A); + Vector128 V2 = MakeVectorE0(B); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, V2: V2, Fpsr: Fpsr); + + AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); + AArch64.V(1, new Bits(A)); + AArch64.V(2, new Bits(B)); + Shared.FPSR = new Bits((uint)Fpsr); + SimdFp.Sqsub_V(Op[30], Op[23, 22], Op[20, 16], Op[9, 5], Op[4, 0]); + + Assert.Multiple(() => + { + Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); + Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); + }); + Assert.That(ThreadState.Fpsr, Is.EqualTo((int)Shared.FPSR.ToUInt32())); + } + + [Test, Pairwise, Description("SQSUB ., ., .")] + public void Sqsub_V_16B_8H_4S_2D([Values(0u)] uint Rd, + [Values(1u, 0u)] uint Rn, + [Values(2u, 0u)] uint Rm, + [ValueSource("_8B4H2S1D_")] [Random(RndCnt)] ulong Z, + [ValueSource("_8B4H2S1D_")] [Random(RndCnt)] ulong A, + [ValueSource("_8B4H2S1D_")] [Random(RndCnt)] ulong B, + [Values(0b00u, 0b01u, 0b10u, 0b11u)] uint size) // <16B, 8H, 4S, 2D> + { + uint Opcode = 0x4E202C00; // SQSUB V0.16B, V0.16B, V0.16B + Opcode |= ((Rm & 31) << 16) | ((Rn & 31) << 5) | ((Rd & 31) << 0); + Opcode |= ((size & 3) << 22); + Bits Op = new Bits(Opcode); + + int Fpsr = (int)TestContext.CurrentContext.Random.NextUInt(); + + Vector128 V0 = MakeVectorE0E1(Z, Z); + Vector128 V1 = MakeVectorE0E1(A, A); + Vector128 V2 = MakeVectorE0E1(B, B); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, V2: V2, Fpsr: Fpsr); + + AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); + AArch64.Vpart(1, 0, new Bits(A)); AArch64.Vpart(1, 1, new Bits(A)); + AArch64.Vpart(2, 0, new Bits(B)); AArch64.Vpart(2, 1, new Bits(B)); + Shared.FPSR = new Bits((uint)Fpsr); + SimdFp.Sqsub_V(Op[30], Op[23, 22], Op[20, 16], Op[9, 5], Op[4, 0]); + + Assert.Multiple(() => + { + Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); + Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); + }); + Assert.That(ThreadState.Fpsr, Is.EqualTo((int)Shared.FPSR.ToUInt32())); + } + [Test, Pairwise, Description("SSUBW{2} ., ., .")] public void Ssubw_V_8B8H8H_4H4S4S_2S2D2D([Values(0u)] uint Rd, [Values(1u, 0u)] uint Rn, @@ -2370,6 +2591,216 @@ namespace Ryujinx.Tests.Cpu }); } + [Test, Pairwise, Description("UQADD , , ")] + public void Uqadd_S_B_H_S_D([Values(0u)] uint Rd, + [Values(1u, 0u)] uint Rn, + [Values(2u, 0u)] uint Rm, + [ValueSource("_1B1H1S1D_")] [Random(RndCnt)] ulong Z, + [ValueSource("_1B1H1S1D_")] [Random(RndCnt)] ulong A, + [ValueSource("_1B1H1S1D_")] [Random(RndCnt)] ulong B, + [Values(0b00u, 0b01u, 0b10u, 0b11u)] uint size) // + { + uint Opcode = 0x7E200C00; // UQADD B0, B0, B0 + Opcode |= ((Rm & 31) << 16) | ((Rn & 31) << 5) | ((Rd & 31) << 0); + Opcode |= ((size & 3) << 22); + Bits Op = new Bits(Opcode); + + int Fpsr = (int)TestContext.CurrentContext.Random.NextUInt(); + + Vector128 V0 = MakeVectorE0E1(Z, Z); + Vector128 V1 = MakeVectorE0(A); + Vector128 V2 = MakeVectorE0(B); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, V2: V2, Fpsr: Fpsr); + + AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); + AArch64.V(1, new Bits(A)); + AArch64.V(2, new Bits(B)); + Shared.FPSR = new Bits((uint)Fpsr); + SimdFp.Uqadd_S(Op[23, 22], Op[20, 16], Op[9, 5], Op[4, 0]); + + Assert.Multiple(() => + { + Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); + Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); + }); + Assert.That(ThreadState.Fpsr, Is.EqualTo((int)Shared.FPSR.ToUInt32())); + } + + [Test, Pairwise, Description("UQADD ., ., .")] + public void Uqadd_V_8B_4H_2S([Values(0u)] uint Rd, + [Values(1u, 0u)] uint Rn, + [Values(2u, 0u)] uint Rm, + [ValueSource("_8B4H2S_")] [Random(RndCnt)] ulong Z, + [ValueSource("_8B4H2S_")] [Random(RndCnt)] ulong A, + [ValueSource("_8B4H2S_")] [Random(RndCnt)] ulong B, + [Values(0b00u, 0b01u, 0b10u)] uint size) // <8B, 4H, 2S> + { + uint Opcode = 0x2E200C00; // UQADD V0.8B, V0.8B, V0.8B + Opcode |= ((Rm & 31) << 16) | ((Rn & 31) << 5) | ((Rd & 31) << 0); + Opcode |= ((size & 3) << 22); + Bits Op = new Bits(Opcode); + + int Fpsr = (int)TestContext.CurrentContext.Random.NextUInt(); + + Vector128 V0 = MakeVectorE0E1(Z, Z); + Vector128 V1 = MakeVectorE0(A); + Vector128 V2 = MakeVectorE0(B); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, V2: V2, Fpsr: Fpsr); + + AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); + AArch64.V(1, new Bits(A)); + AArch64.V(2, new Bits(B)); + Shared.FPSR = new Bits((uint)Fpsr); + SimdFp.Uqadd_V(Op[30], Op[23, 22], Op[20, 16], Op[9, 5], Op[4, 0]); + + Assert.Multiple(() => + { + Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); + Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); + }); + Assert.That(ThreadState.Fpsr, Is.EqualTo((int)Shared.FPSR.ToUInt32())); + } + + [Test, Pairwise, Description("UQADD ., ., .")] + public void Uqadd_V_16B_8H_4S_2D([Values(0u)] uint Rd, + [Values(1u, 0u)] uint Rn, + [Values(2u, 0u)] uint Rm, + [ValueSource("_8B4H2S1D_")] [Random(RndCnt)] ulong Z, + [ValueSource("_8B4H2S1D_")] [Random(RndCnt)] ulong A, + [ValueSource("_8B4H2S1D_")] [Random(RndCnt)] ulong B, + [Values(0b00u, 0b01u, 0b10u, 0b11u)] uint size) // <16B, 8H, 4S, 2D> + { + uint Opcode = 0x6E200C00; // UQADD V0.16B, V0.16B, V0.16B + Opcode |= ((Rm & 31) << 16) | ((Rn & 31) << 5) | ((Rd & 31) << 0); + Opcode |= ((size & 3) << 22); + Bits Op = new Bits(Opcode); + + int Fpsr = (int)TestContext.CurrentContext.Random.NextUInt(); + + Vector128 V0 = MakeVectorE0E1(Z, Z); + Vector128 V1 = MakeVectorE0E1(A, A); + Vector128 V2 = MakeVectorE0E1(B, B); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, V2: V2, Fpsr: Fpsr); + + AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); + AArch64.Vpart(1, 0, new Bits(A)); AArch64.Vpart(1, 1, new Bits(A)); + AArch64.Vpart(2, 0, new Bits(B)); AArch64.Vpart(2, 1, new Bits(B)); + Shared.FPSR = new Bits((uint)Fpsr); + SimdFp.Uqadd_V(Op[30], Op[23, 22], Op[20, 16], Op[9, 5], Op[4, 0]); + + Assert.Multiple(() => + { + Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); + Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); + }); + Assert.That(ThreadState.Fpsr, Is.EqualTo((int)Shared.FPSR.ToUInt32())); + } + + [Test, Pairwise, Description("UQSUB , , ")] + public void Uqsub_S_B_H_S_D([Values(0u)] uint Rd, + [Values(1u, 0u)] uint Rn, + [Values(2u, 0u)] uint Rm, + [ValueSource("_1B1H1S1D_")] [Random(RndCnt)] ulong Z, + [ValueSource("_1B1H1S1D_")] [Random(RndCnt)] ulong A, + [ValueSource("_1B1H1S1D_")] [Random(RndCnt)] ulong B, + [Values(0b00u, 0b01u, 0b10u, 0b11u)] uint size) // + { + uint Opcode = 0x7E202C00; // UQSUB B0, B0, B0 + Opcode |= ((Rm & 31) << 16) | ((Rn & 31) << 5) | ((Rd & 31) << 0); + Opcode |= ((size & 3) << 22); + Bits Op = new Bits(Opcode); + + int Fpsr = (int)TestContext.CurrentContext.Random.NextUInt(); + + Vector128 V0 = MakeVectorE0E1(Z, Z); + Vector128 V1 = MakeVectorE0(A); + Vector128 V2 = MakeVectorE0(B); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, V2: V2, Fpsr: Fpsr); + + AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); + AArch64.V(1, new Bits(A)); + AArch64.V(2, new Bits(B)); + Shared.FPSR = new Bits((uint)Fpsr); + SimdFp.Uqsub_S(Op[23, 22], Op[20, 16], Op[9, 5], Op[4, 0]); + + Assert.Multiple(() => + { + Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); + Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); + }); + Assert.That(ThreadState.Fpsr, Is.EqualTo((int)Shared.FPSR.ToUInt32())); + } + + [Test, Pairwise, Description("UQSUB ., ., .")] + public void Uqsub_V_8B_4H_2S([Values(0u)] uint Rd, + [Values(1u, 0u)] uint Rn, + [Values(2u, 0u)] uint Rm, + [ValueSource("_8B4H2S_")] [Random(RndCnt)] ulong Z, + [ValueSource("_8B4H2S_")] [Random(RndCnt)] ulong A, + [ValueSource("_8B4H2S_")] [Random(RndCnt)] ulong B, + [Values(0b00u, 0b01u, 0b10u)] uint size) // <8B, 4H, 2S> + { + uint Opcode = 0x2E202C00; // UQSUB V0.8B, V0.8B, V0.8B + Opcode |= ((Rm & 31) << 16) | ((Rn & 31) << 5) | ((Rd & 31) << 0); + Opcode |= ((size & 3) << 22); + Bits Op = new Bits(Opcode); + + int Fpsr = (int)TestContext.CurrentContext.Random.NextUInt(); + + Vector128 V0 = MakeVectorE0E1(Z, Z); + Vector128 V1 = MakeVectorE0(A); + Vector128 V2 = MakeVectorE0(B); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, V2: V2, Fpsr: Fpsr); + + AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); + AArch64.V(1, new Bits(A)); + AArch64.V(2, new Bits(B)); + Shared.FPSR = new Bits((uint)Fpsr); + SimdFp.Uqsub_V(Op[30], Op[23, 22], Op[20, 16], Op[9, 5], Op[4, 0]); + + Assert.Multiple(() => + { + Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); + Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); + }); + Assert.That(ThreadState.Fpsr, Is.EqualTo((int)Shared.FPSR.ToUInt32())); + } + + [Test, Pairwise, Description("UQSUB ., ., .")] + public void Uqsub_V_16B_8H_4S_2D([Values(0u)] uint Rd, + [Values(1u, 0u)] uint Rn, + [Values(2u, 0u)] uint Rm, + [ValueSource("_8B4H2S1D_")] [Random(RndCnt)] ulong Z, + [ValueSource("_8B4H2S1D_")] [Random(RndCnt)] ulong A, + [ValueSource("_8B4H2S1D_")] [Random(RndCnt)] ulong B, + [Values(0b00u, 0b01u, 0b10u, 0b11u)] uint size) // <16B, 8H, 4S, 2D> + { + uint Opcode = 0x6E202C00; // UQSUB V0.16B, V0.16B, V0.16B + Opcode |= ((Rm & 31) << 16) | ((Rn & 31) << 5) | ((Rd & 31) << 0); + Opcode |= ((size & 3) << 22); + Bits Op = new Bits(Opcode); + + int Fpsr = (int)TestContext.CurrentContext.Random.NextUInt(); + + Vector128 V0 = MakeVectorE0E1(Z, Z); + Vector128 V1 = MakeVectorE0E1(A, A); + Vector128 V2 = MakeVectorE0E1(B, B); + AThreadState ThreadState = SingleOpcode(Opcode, V0: V0, V1: V1, V2: V2, Fpsr: Fpsr); + + AArch64.Vpart(0, 0, new Bits(Z)); AArch64.Vpart(0, 1, new Bits(Z)); + AArch64.Vpart(1, 0, new Bits(A)); AArch64.Vpart(1, 1, new Bits(A)); + AArch64.Vpart(2, 0, new Bits(B)); AArch64.Vpart(2, 1, new Bits(B)); + Shared.FPSR = new Bits((uint)Fpsr); + SimdFp.Uqsub_V(Op[30], Op[23, 22], Op[20, 16], Op[9, 5], Op[4, 0]); + + Assert.Multiple(() => + { + Assert.That(GetVectorE0(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 0).ToUInt64())); + Assert.That(GetVectorE1(ThreadState.V0), Is.EqualTo(AArch64.Vpart(64, 0, 1).ToUInt64())); + }); + Assert.That(ThreadState.Fpsr, Is.EqualTo((int)Shared.FPSR.ToUInt32())); + } + [Test, Pairwise, Description("USUBW{2} ., ., .")] public void Usubw_V_8B8H8H_4H4S4S_2S2D2D([Values(0u)] uint Rd, [Values(1u, 0u)] uint Rn, diff --git a/Ryujinx.Tests/Cpu/Tester/Instructions.cs b/Ryujinx.Tests/Cpu/Tester/Instructions.cs index 2f52dcbf55..8b1b010f65 100644 --- a/Ryujinx.Tests/Cpu/Tester/Instructions.cs +++ b/Ryujinx.Tests/Cpu/Tester/Instructions.cs @@ -3060,6 +3060,210 @@ namespace Ryujinx.Tests.Cpu.Tester V(d, result); } + // sqabs_advsimd.html#SQABS_asisdmisc_R + public static void Sqabs_S(Bits size, Bits Rn, Bits Rd) + { + const bool U = false; + + /* Decode Scalar */ + int d = (int)UInt(Rd); + int n = (int)UInt(Rn); + + int esize = 8 << (int)UInt(size); + int datasize = esize; + int elements = 1; + + bool neg = (U == true); + + /* Operation */ + /* CheckFPAdvSIMDEnabled64(); */ + + Bits result = new Bits(datasize); + Bits operand = V(datasize, n); + BigInteger element; + bool sat; + + for (int e = 0; e <= elements - 1; e++) + { + element = SInt(Elem(operand, e, esize)); + + if (neg) + { + element = -element; + } + else + { + element = Abs(element); + } + + (Bits _result, bool _sat) = SignedSatQ(element, esize); + Elem(result, e, esize, _result); + sat = _sat; + + if (sat) + { + /* FPSR.QC = '1'; */ + FPSR[27] = true; // TODO: Add named fields. + } + } + + V(d, result); + } + + // sqabs_advsimd.html#SQABS_asimdmisc_R + public static void Sqabs_V(bool Q, Bits size, Bits Rn, Bits Rd) + { + const bool U = false; + + /* Decode Vector */ + int d = (int)UInt(Rd); + int n = (int)UInt(Rn); + + /* if size:Q == '110' then ReservedValue(); */ + + int esize = 8 << (int)UInt(size); + int datasize = (Q ? 128 : 64); + int elements = datasize / esize; + + bool neg = (U == true); + + /* Operation */ + /* CheckFPAdvSIMDEnabled64(); */ + + Bits result = new Bits(datasize); + Bits operand = V(datasize, n); + BigInteger element; + bool sat; + + for (int e = 0; e <= elements - 1; e++) + { + element = SInt(Elem(operand, e, esize)); + + if (neg) + { + element = -element; + } + else + { + element = Abs(element); + } + + (Bits _result, bool _sat) = SignedSatQ(element, esize); + Elem(result, e, esize, _result); + sat = _sat; + + if (sat) + { + /* FPSR.QC = '1'; */ + FPSR[27] = true; // TODO: Add named fields. + } + } + + V(d, result); + } + + // sqneg_advsimd.html#SQNEG_asisdmisc_R + public static void Sqneg_S(Bits size, Bits Rn, Bits Rd) + { + const bool U = true; + + /* Decode Scalar */ + int d = (int)UInt(Rd); + int n = (int)UInt(Rn); + + int esize = 8 << (int)UInt(size); + int datasize = esize; + int elements = 1; + + bool neg = (U == true); + + /* Operation */ + /* CheckFPAdvSIMDEnabled64(); */ + + Bits result = new Bits(datasize); + Bits operand = V(datasize, n); + BigInteger element; + bool sat; + + for (int e = 0; e <= elements - 1; e++) + { + element = SInt(Elem(operand, e, esize)); + + if (neg) + { + element = -element; + } + else + { + element = Abs(element); + } + + (Bits _result, bool _sat) = SignedSatQ(element, esize); + Elem(result, e, esize, _result); + sat = _sat; + + if (sat) + { + /* FPSR.QC = '1'; */ + FPSR[27] = true; // TODO: Add named fields. + } + } + + V(d, result); + } + + // sqneg_advsimd.html#SQNEG_asimdmisc_R + public static void Sqneg_V(bool Q, Bits size, Bits Rn, Bits Rd) + { + const bool U = true; + + /* Decode Vector */ + int d = (int)UInt(Rd); + int n = (int)UInt(Rn); + + /* if size:Q == '110' then ReservedValue(); */ + + int esize = 8 << (int)UInt(size); + int datasize = (Q ? 128 : 64); + int elements = datasize / esize; + + bool neg = (U == true); + + /* Operation */ + /* CheckFPAdvSIMDEnabled64(); */ + + Bits result = new Bits(datasize); + Bits operand = V(datasize, n); + BigInteger element; + bool sat; + + for (int e = 0; e <= elements - 1; e++) + { + element = SInt(Elem(operand, e, esize)); + + if (neg) + { + element = -element; + } + else + { + element = Abs(element); + } + + (Bits _result, bool _sat) = SignedSatQ(element, esize); + Elem(result, e, esize, _result); + sat = _sat; + + if (sat) + { + /* FPSR.QC = '1'; */ + FPSR[27] = true; // TODO: Add named fields. + } + } + + V(d, result); + } + // sqxtn_advsimd.html#SQXTN_asisdmisc_N public static void Sqxtn_S(Bits size, Bits Rn, Bits Rd) { @@ -3228,6 +3432,96 @@ namespace Ryujinx.Tests.Cpu.Tester Vpart(d, part, result); } + // suqadd_advsimd.html#SUQADD_asisdmisc_R + public static void Suqadd_S(Bits size, Bits Rn, Bits Rd) + { + const bool U = false; + + /* Decode Scalar */ + int d = (int)UInt(Rd); + int n = (int)UInt(Rn); + + int esize = 8 << (int)UInt(size); + int datasize = esize; + int elements = 1; + + bool unsigned = (U == true); + + /* Operation */ + /* CheckFPAdvSIMDEnabled64(); */ + + Bits result = new Bits(datasize); + Bits operand = V(datasize, n); + Bits operand2 = V(datasize, d); + BigInteger op1; + BigInteger op2; + bool sat; + + for (int e = 0; e <= elements - 1; e++) + { + op1 = Int(Elem(operand, e, esize), !unsigned); + op2 = Int(Elem(operand2, e, esize), unsigned); + + (Bits _result, bool _sat) = SatQ(op1 + op2, esize, unsigned); + Elem(result, e, esize, _result); + sat = _sat; + + if (sat) + { + /* FPSR.QC = '1'; */ + FPSR[27] = true; // TODO: Add named fields. + } + } + + V(d, result); + } + + // suqadd_advsimd.html#SUQADD_asimdmisc_R + public static void Suqadd_V(bool Q, Bits size, Bits Rn, Bits Rd) + { + const bool U = false; + + /* Decode Vector */ + int d = (int)UInt(Rd); + int n = (int)UInt(Rn); + + /* if size:Q == '110' then ReservedValue(); */ + + int esize = 8 << (int)UInt(size); + int datasize = (Q ? 128 : 64); + int elements = datasize / esize; + + bool unsigned = (U == true); + + /* Operation */ + /* CheckFPAdvSIMDEnabled64(); */ + + Bits result = new Bits(datasize); + Bits operand = V(datasize, n); + Bits operand2 = V(datasize, d); + BigInteger op1; + BigInteger op2; + bool sat; + + for (int e = 0; e <= elements - 1; e++) + { + op1 = Int(Elem(operand, e, esize), !unsigned); + op2 = Int(Elem(operand2, e, esize), unsigned); + + (Bits _result, bool _sat) = SatQ(op1 + op2, esize, unsigned); + Elem(result, e, esize, _result); + sat = _sat; + + if (sat) + { + /* FPSR.QC = '1'; */ + FPSR[27] = true; // TODO: Add named fields. + } + } + + V(d, result); + } + // uqxtn_advsimd.html#UQXTN_asisdmisc_N public static void Uqxtn_S(Bits size, Bits Rn, Bits Rd) { @@ -3316,6 +3610,96 @@ namespace Ryujinx.Tests.Cpu.Tester Vpart(d, part, result); } + // usqadd_advsimd.html#USQADD_asisdmisc_R + public static void Usqadd_S(Bits size, Bits Rn, Bits Rd) + { + const bool U = true; + + /* Decode Scalar */ + int d = (int)UInt(Rd); + int n = (int)UInt(Rn); + + int esize = 8 << (int)UInt(size); + int datasize = esize; + int elements = 1; + + bool unsigned = (U == true); + + /* Operation */ + /* CheckFPAdvSIMDEnabled64(); */ + + Bits result = new Bits(datasize); + Bits operand = V(datasize, n); + Bits operand2 = V(datasize, d); + BigInteger op1; + BigInteger op2; + bool sat; + + for (int e = 0; e <= elements - 1; e++) + { + op1 = Int(Elem(operand, e, esize), !unsigned); + op2 = Int(Elem(operand2, e, esize), unsigned); + + (Bits _result, bool _sat) = SatQ(op1 + op2, esize, unsigned); + Elem(result, e, esize, _result); + sat = _sat; + + if (sat) + { + /* FPSR.QC = '1'; */ + FPSR[27] = true; // TODO: Add named fields. + } + } + + V(d, result); + } + + // usqadd_advsimd.html#USQADD_asimdmisc_R + public static void Usqadd_V(bool Q, Bits size, Bits Rn, Bits Rd) + { + const bool U = true; + + /* Decode Vector */ + int d = (int)UInt(Rd); + int n = (int)UInt(Rn); + + /* if size:Q == '110' then ReservedValue(); */ + + int esize = 8 << (int)UInt(size); + int datasize = (Q ? 128 : 64); + int elements = datasize / esize; + + bool unsigned = (U == true); + + /* Operation */ + /* CheckFPAdvSIMDEnabled64(); */ + + Bits result = new Bits(datasize); + Bits operand = V(datasize, n); + Bits operand2 = V(datasize, d); + BigInteger op1; + BigInteger op2; + bool sat; + + for (int e = 0; e <= elements - 1; e++) + { + op1 = Int(Elem(operand, e, esize), !unsigned); + op2 = Int(Elem(operand2, e, esize), unsigned); + + (Bits _result, bool _sat) = SatQ(op1 + op2, esize, unsigned); + Elem(result, e, esize, _result); + sat = _sat; + + if (sat) + { + /* FPSR.QC = '1'; */ + FPSR[27] = true; // TODO: Add named fields. + } + } + + V(d, result); + } + // xtn_advsimd.html public static void Xtn_V(bool Q, Bits size, Bits Rn, Bits Rd) { @@ -4593,6 +4977,202 @@ namespace Ryujinx.Tests.Cpu.Tester V(d, result); } + // sqadd_advsimd.html#SQADD_asisdsame_only + public static void Sqadd_S(Bits size, Bits Rm, Bits Rn, Bits Rd) + { + const bool U = false; + + /* Decode Scalar */ + int d = (int)UInt(Rd); + int n = (int)UInt(Rn); + int m = (int)UInt(Rm); + + int esize = 8 << (int)UInt(size); + int datasize = esize; + int elements = 1; + + bool unsigned = (U == true); + + /* Operation */ + /* CheckFPAdvSIMDEnabled64(); */ + + Bits result = new Bits(datasize); + Bits operand1 = V(datasize, n); + Bits operand2 = V(datasize, m); + BigInteger element1; + BigInteger element2; + BigInteger sum; + bool sat; + + for (int e = 0; e <= elements - 1; e++) + { + element1 = Int(Elem(operand1, e, esize), unsigned); + element2 = Int(Elem(operand2, e, esize), unsigned); + + sum = element1 + element2; + + (Bits _result, bool _sat) = SatQ(sum, esize, unsigned); + Elem(result, e, esize, _result); + sat = _sat; + + if (sat) + { + /* FPSR.QC = '1'; */ + FPSR[27] = true; // TODO: Add named fields. + } + } + + V(d, result); + } + + // sqadd_advsimd.html#SQADD_asimdsame_only + public static void Sqadd_V(bool Q, Bits size, Bits Rm, Bits Rn, Bits Rd) + { + const bool U = false; + + /* Decode Vector */ + int d = (int)UInt(Rd); + int n = (int)UInt(Rn); + int m = (int)UInt(Rm); + + /* if size:Q == '110' then ReservedValue(); */ + + int esize = 8 << (int)UInt(size); + int datasize = (Q ? 128 : 64); + int elements = datasize / esize; + + bool unsigned = (U == true); + + /* Operation */ + /* CheckFPAdvSIMDEnabled64(); */ + + Bits result = new Bits(datasize); + Bits operand1 = V(datasize, n); + Bits operand2 = V(datasize, m); + BigInteger element1; + BigInteger element2; + BigInteger sum; + bool sat; + + for (int e = 0; e <= elements - 1; e++) + { + element1 = Int(Elem(operand1, e, esize), unsigned); + element2 = Int(Elem(operand2, e, esize), unsigned); + + sum = element1 + element2; + + (Bits _result, bool _sat) = SatQ(sum, esize, unsigned); + Elem(result, e, esize, _result); + sat = _sat; + + if (sat) + { + /* FPSR.QC = '1'; */ + FPSR[27] = true; // TODO: Add named fields. + } + } + + V(d, result); + } + + // sqsub_advsimd.html#SQSUB_asisdsame_only + public static void Sqsub_S(Bits size, Bits Rm, Bits Rn, Bits Rd) + { + const bool U = false; + + /* Decode Scalar */ + int d = (int)UInt(Rd); + int n = (int)UInt(Rn); + int m = (int)UInt(Rm); + + int esize = 8 << (int)UInt(size); + int datasize = esize; + int elements = 1; + + bool unsigned = (U == true); + + /* Operation */ + /* CheckFPAdvSIMDEnabled64(); */ + + Bits result = new Bits(datasize); + Bits operand1 = V(datasize, n); + Bits operand2 = V(datasize, m); + BigInteger element1; + BigInteger element2; + BigInteger diff; + bool sat; + + for (int e = 0; e <= elements - 1; e++) + { + element1 = Int(Elem(operand1, e, esize), unsigned); + element2 = Int(Elem(operand2, e, esize), unsigned); + + diff = element1 - element2; + + (Bits _result, bool _sat) = SatQ(diff, esize, unsigned); + Elem(result, e, esize, _result); + sat = _sat; + + if (sat) + { + /* FPSR.QC = '1'; */ + FPSR[27] = true; // TODO: Add named fields. + } + } + + V(d, result); + } + + // sqsub_advsimd.html#SQSUB_asimdsame_only + public static void Sqsub_V(bool Q, Bits size, Bits Rm, Bits Rn, Bits Rd) + { + const bool U = false; + + /* Decode Vector */ + int d = (int)UInt(Rd); + int n = (int)UInt(Rn); + int m = (int)UInt(Rm); + + /* if size:Q == '110' then ReservedValue(); */ + + int esize = 8 << (int)UInt(size); + int datasize = (Q ? 128 : 64); + int elements = datasize / esize; + + bool unsigned = (U == true); + + /* Operation */ + /* CheckFPAdvSIMDEnabled64(); */ + + Bits result = new Bits(datasize); + Bits operand1 = V(datasize, n); + Bits operand2 = V(datasize, m); + BigInteger element1; + BigInteger element2; + BigInteger diff; + bool sat; + + for (int e = 0; e <= elements - 1; e++) + { + element1 = Int(Elem(operand1, e, esize), unsigned); + element2 = Int(Elem(operand2, e, esize), unsigned); + + diff = element1 - element2; + + (Bits _result, bool _sat) = SatQ(diff, esize, unsigned); + Elem(result, e, esize, _result); + sat = _sat; + + if (sat) + { + /* FPSR.QC = '1'; */ + FPSR[27] = true; // TODO: Add named fields. + } + } + + V(d, result); + } + // ssubw_advsimd.html public static void Ssubw_V(bool Q, Bits size, Bits Rm, Bits Rn, Bits Rd) { @@ -5085,6 +5665,202 @@ namespace Ryujinx.Tests.Cpu.Tester V(d, result); } + // uqadd_advsimd.html#UQADD_asisdsame_only + public static void Uqadd_S(Bits size, Bits Rm, Bits Rn, Bits Rd) + { + const bool U = true; + + /* Decode Scalar */ + int d = (int)UInt(Rd); + int n = (int)UInt(Rn); + int m = (int)UInt(Rm); + + int esize = 8 << (int)UInt(size); + int datasize = esize; + int elements = 1; + + bool unsigned = (U == true); + + /* Operation */ + /* CheckFPAdvSIMDEnabled64(); */ + + Bits result = new Bits(datasize); + Bits operand1 = V(datasize, n); + Bits operand2 = V(datasize, m); + BigInteger element1; + BigInteger element2; + BigInteger sum; + bool sat; + + for (int e = 0; e <= elements - 1; e++) + { + element1 = Int(Elem(operand1, e, esize), unsigned); + element2 = Int(Elem(operand2, e, esize), unsigned); + + sum = element1 + element2; + + (Bits _result, bool _sat) = SatQ(sum, esize, unsigned); + Elem(result, e, esize, _result); + sat = _sat; + + if (sat) + { + /* FPSR.QC = '1'; */ + FPSR[27] = true; // TODO: Add named fields. + } + } + + V(d, result); + } + + // uqadd_advsimd.html#UQADD_asimdsame_only + public static void Uqadd_V(bool Q, Bits size, Bits Rm, Bits Rn, Bits Rd) + { + const bool U = true; + + /* Decode Vector */ + int d = (int)UInt(Rd); + int n = (int)UInt(Rn); + int m = (int)UInt(Rm); + + /* if size:Q == '110' then ReservedValue(); */ + + int esize = 8 << (int)UInt(size); + int datasize = (Q ? 128 : 64); + int elements = datasize / esize; + + bool unsigned = (U == true); + + /* Operation */ + /* CheckFPAdvSIMDEnabled64(); */ + + Bits result = new Bits(datasize); + Bits operand1 = V(datasize, n); + Bits operand2 = V(datasize, m); + BigInteger element1; + BigInteger element2; + BigInteger sum; + bool sat; + + for (int e = 0; e <= elements - 1; e++) + { + element1 = Int(Elem(operand1, e, esize), unsigned); + element2 = Int(Elem(operand2, e, esize), unsigned); + + sum = element1 + element2; + + (Bits _result, bool _sat) = SatQ(sum, esize, unsigned); + Elem(result, e, esize, _result); + sat = _sat; + + if (sat) + { + /* FPSR.QC = '1'; */ + FPSR[27] = true; // TODO: Add named fields. + } + } + + V(d, result); + } + + // uqsub_advsimd.html#UQSUB_asisdsame_only + public static void Uqsub_S(Bits size, Bits Rm, Bits Rn, Bits Rd) + { + const bool U = true; + + /* Decode Scalar */ + int d = (int)UInt(Rd); + int n = (int)UInt(Rn); + int m = (int)UInt(Rm); + + int esize = 8 << (int)UInt(size); + int datasize = esize; + int elements = 1; + + bool unsigned = (U == true); + + /* Operation */ + /* CheckFPAdvSIMDEnabled64(); */ + + Bits result = new Bits(datasize); + Bits operand1 = V(datasize, n); + Bits operand2 = V(datasize, m); + BigInteger element1; + BigInteger element2; + BigInteger diff; + bool sat; + + for (int e = 0; e <= elements - 1; e++) + { + element1 = Int(Elem(operand1, e, esize), unsigned); + element2 = Int(Elem(operand2, e, esize), unsigned); + + diff = element1 - element2; + + (Bits _result, bool _sat) = SatQ(diff, esize, unsigned); + Elem(result, e, esize, _result); + sat = _sat; + + if (sat) + { + /* FPSR.QC = '1'; */ + FPSR[27] = true; // TODO: Add named fields. + } + } + + V(d, result); + } + + // uqsub_advsimd.html#UQSUB_asimdsame_only + public static void Uqsub_V(bool Q, Bits size, Bits Rm, Bits Rn, Bits Rd) + { + const bool U = true; + + /* Decode Vector */ + int d = (int)UInt(Rd); + int n = (int)UInt(Rn); + int m = (int)UInt(Rm); + + /* if size:Q == '110' then ReservedValue(); */ + + int esize = 8 << (int)UInt(size); + int datasize = (Q ? 128 : 64); + int elements = datasize / esize; + + bool unsigned = (U == true); + + /* Operation */ + /* CheckFPAdvSIMDEnabled64(); */ + + Bits result = new Bits(datasize); + Bits operand1 = V(datasize, n); + Bits operand2 = V(datasize, m); + BigInteger element1; + BigInteger element2; + BigInteger diff; + bool sat; + + for (int e = 0; e <= elements - 1; e++) + { + element1 = Int(Elem(operand1, e, esize), unsigned); + element2 = Int(Elem(operand2, e, esize), unsigned); + + diff = element1 - element2; + + (Bits _result, bool _sat) = SatQ(diff, esize, unsigned); + Elem(result, e, esize, _result); + sat = _sat; + + if (sat) + { + /* FPSR.QC = '1'; */ + FPSR[27] = true; // TODO: Add named fields. + } + } + + V(d, result); + } + // usubw_advsimd.html public static void Usubw_V(bool Q, Bits size, Bits Rm, Bits Rn, Bits Rd) { diff --git a/Ryujinx.Tests/Cpu/Tester/Pseudocode.cs b/Ryujinx.Tests/Cpu/Tester/Pseudocode.cs index 3a877fb1a2..6c4dfa92ba 100644 --- a/Ryujinx.Tests/Cpu/Tester/Pseudocode.cs +++ b/Ryujinx.Tests/Cpu/Tester/Pseudocode.cs @@ -1193,9 +1193,9 @@ namespace Ryujinx.Tests.Cpu.Tester result = BigInteger.Pow(2, N) - 1; saturated = true; } - else if (i < 0) + else if (i < (BigInteger)0) { - result = 0; + result = (BigInteger)0; saturated = true; } else From eeb626947eac81b9d9d9a90ad6e29035c21d54a7 Mon Sep 17 00:00:00 2001 From: Thomas Guillemard Date: Sat, 4 Aug 2018 23:38:49 +0200 Subject: [PATCH 33/34] Implement Shared Fonts (#215) * Implement Shared Fonts This fully implements shared fonts. This commit is provided without fonts. This commit also add Size to HSharedMem.Positions to be able to add fonts to shared zones when RequestLoad is called. * Require the user to provide fonts in RyuFS/system * Use File.Exits instead of relying ona try/catch and change system resource exception format a bit * Make sure that font sum doesn't exceed 17MB Also rename font data dictionary for coherence. --- Ryujinx.HLE/Font/SharedFontManager.cs | 177 ++++++++++++++++++ .../Services/Pl => Font}/SharedFontType.cs | 4 +- Ryujinx.HLE/Hid/Hid.cs | 10 +- Ryujinx.HLE/Logging/LogClass.cs | 1 + Ryujinx.HLE/OsHle/Handles/HSharedMem.cs | 14 +- Ryujinx.HLE/OsHle/Kernel/SvcHandler.cs | 8 +- Ryujinx.HLE/OsHle/Kernel/SvcMemory.cs | 10 +- Ryujinx.HLE/OsHle/Kernel/SvcSystem.cs | 1 - .../OsHle/Services/Pl/ISharedFontManager.cs | 72 ++++++- .../InvalidSystemResourceException.cs | 13 ++ Ryujinx.HLE/Ryujinx.HLE.csproj | 2 +- Ryujinx.HLE/Switch.cs | 12 +- Ryujinx.HLE/VirtualFileSystem.cs | 3 + 13 files changed, 292 insertions(+), 35 deletions(-) create mode 100644 Ryujinx.HLE/Font/SharedFontManager.cs rename Ryujinx.HLE/{OsHle/Services/Pl => Font}/SharedFontType.cs (76%) create mode 100644 Ryujinx.HLE/Resource/InvalidSystemResourceException.cs diff --git a/Ryujinx.HLE/Font/SharedFontManager.cs b/Ryujinx.HLE/Font/SharedFontManager.cs new file mode 100644 index 0000000000..fce270de86 --- /dev/null +++ b/Ryujinx.HLE/Font/SharedFontManager.cs @@ -0,0 +1,177 @@ +using ChocolArm64.Exceptions; +using ChocolArm64.Memory; +using Ryujinx.HLE.Logging; +using Ryujinx.HLE.OsHle; +using Ryujinx.HLE.OsHle.Handles; +using Ryujinx.HLE.Resource; +using System; +using System.Collections.Generic; +using System.IO; + + +namespace Ryujinx.HLE.Font +{ + public class SharedFontManager + { + private const uint SharedMemorySize = 0x1100000; + private Logger Log; + + private string FontsPath; + + private object ShMemLock; + + private (AMemory, long, long)[] ShMemPositions; + + private Dictionary FontData; + + private uint[] LoadedFonts; + + public SharedFontManager(Logger Log, string SystemPath) + { + this.Log = Log; + this.FontsPath = Path.Combine(SystemPath, "fonts"); + + ShMemLock = new object(); + + ShMemPositions = new(AMemory, long, long)[0]; + + FontData = new Dictionary() + { + { SharedFontType.JapanUsEurope, GetData("FontStandard") }, + { SharedFontType.SimplifiedChinese, GetData("FontChineseSimplified") }, + { SharedFontType.SimplifiedChineseEx, GetData("FontExtendedChineseSimplified") }, + { SharedFontType.TraditionalChinese, GetData("FontChineseTraditional") }, + { SharedFontType.Korean, GetData("FontKorean") }, + { SharedFontType.NintendoEx, GetData("FontNintendoExtended") } + }; + + int FontMemoryUsage = 0; + foreach (byte[] data in FontData.Values) + { + FontMemoryUsage += data.Length; + FontMemoryUsage += 0x8; + } + + if (FontMemoryUsage > SharedMemorySize) + { + throw new InvalidSystemResourceException($"The sum of all fonts size exceed the shared memory size. Please make sure that the fonts don't exceed {SharedMemorySize} bytes in total. (actual size: {FontMemoryUsage} bytes)"); + } + + LoadedFonts = new uint[FontData.Count]; + } + + public byte[] GetData(string FontName) + { + string FontFilePath = Path.Combine(FontsPath, $"{FontName}.ttf"); + if (File.Exists(FontFilePath)) + { + return File.ReadAllBytes(FontFilePath); + } + else + { + throw new InvalidSystemResourceException($"Font \"{FontName}.ttf\" not found. Please provide it in \"{FontsPath}\"."); + } + } + + public void MapFont(SharedFontType FontType, AMemory Memory, long Position) + { + uint SharedMemoryAddressOffset = GetSharedMemoryAddressOffset(FontType); + // TODO: find what are the 8 bytes before the font + Memory.WriteUInt64(Position + SharedMemoryAddressOffset - 8, 0); + Memory.WriteBytes(Position + SharedMemoryAddressOffset, FontData[FontType]); + } + + public void PropagateNewMapFont(SharedFontType Type) + { + lock (ShMemLock) + { + foreach ((AMemory Memory, long Position, long Size) in ShMemPositions) + { + AMemoryMapInfo MemoryInfo = Memory.Manager.GetMapInfo(Position); + + if (MemoryInfo == null) + { + throw new VmmPageFaultException(Position); + } + + // The memory is read only, we need to changes that to add the new font + AMemoryPerm originalPerms = MemoryInfo.Perm; + Memory.Manager.Reprotect(Position, Size, AMemoryPerm.RW); + MapFont(Type, Memory, Position); + Memory.Manager.Reprotect(Position, Size, originalPerms); + } + } + } + + internal void ShMemMap(object sender, EventArgs e) + { + HSharedMem SharedMem = (HSharedMem)sender; + + lock (ShMemLock) + { + ShMemPositions = SharedMem.GetVirtualPositions(); + + (AMemory Memory, long Position, long Size) = ShMemPositions[ShMemPositions.Length - 1]; + + for (int Type = 0; Type < LoadedFonts.Length; Type++) + { + if (LoadedFonts[(int)Type] == 1) + { + MapFont((SharedFontType)Type, Memory, Position); + } + } + } + } + + internal void ShMemUnmap(object sender, EventArgs e) + { + HSharedMem SharedMem = (HSharedMem)sender; + + lock (ShMemLock) + { + ShMemPositions = SharedMem.GetVirtualPositions(); + } + } + + public void Load(SharedFontType FontType) + { + if (LoadedFonts[(int)FontType] == 0) + { + PropagateNewMapFont(FontType); + } + + LoadedFonts[(int)FontType] = 1; + } + + public uint GetLoadState(SharedFontType FontType) + { + if (LoadedFonts[(int)FontType] != 1) + { + // Some games don't request a load, so we need to load it here. + Load(FontType); + return 0; + } + return LoadedFonts[(int)FontType]; + } + + public uint GetFontSize(SharedFontType FontType) + { + return (uint)FontData[FontType].Length; + } + + public uint GetSharedMemoryAddressOffset(SharedFontType FontType) + { + uint Pos = 0x8; + + for (SharedFontType Type = SharedFontType.JapanUsEurope; Type < FontType; Type++) + { + Pos += GetFontSize(Type); + Pos += 0x8; + } + + return Pos; + } + + public int Count => FontData.Count; + } +} diff --git a/Ryujinx.HLE/OsHle/Services/Pl/SharedFontType.cs b/Ryujinx.HLE/Font/SharedFontType.cs similarity index 76% rename from Ryujinx.HLE/OsHle/Services/Pl/SharedFontType.cs rename to Ryujinx.HLE/Font/SharedFontType.cs index 97fd95dc9d..ca8a42b0cd 100644 --- a/Ryujinx.HLE/OsHle/Services/Pl/SharedFontType.cs +++ b/Ryujinx.HLE/Font/SharedFontType.cs @@ -1,6 +1,6 @@ -namespace Ryujinx.HLE.OsHle.Services.Pl +namespace Ryujinx.HLE.Font { - enum SharedFontType + public enum SharedFontType { JapanUsEurope = 0, SimplifiedChinese = 1, diff --git a/Ryujinx.HLE/Hid/Hid.cs b/Ryujinx.HLE/Hid/Hid.cs index 2f007f1fb0..43c040bb5d 100644 --- a/Ryujinx.HLE/Hid/Hid.cs +++ b/Ryujinx.HLE/Hid/Hid.cs @@ -67,7 +67,7 @@ namespace Ryujinx.HLE.Input private object ShMemLock; - private (AMemory, long)[] ShMemPositions; + private (AMemory, long, long)[] ShMemPositions; public Hid(Logger Log) { @@ -75,7 +75,7 @@ namespace Ryujinx.HLE.Input ShMemLock = new object(); - ShMemPositions = new (AMemory, long)[0]; + ShMemPositions = new (AMemory, long, long)[0]; } internal void ShMemMap(object sender, EventArgs e) @@ -86,7 +86,7 @@ namespace Ryujinx.HLE.Input { ShMemPositions = SharedMem.GetVirtualPositions(); - (AMemory Memory, long Position) = ShMemPositions[ShMemPositions.Length - 1]; + (AMemory Memory, long Position, long Size) = ShMemPositions[ShMemPositions.Length - 1]; for (long Offset = 0; Offset < Horizon.HidSize; Offset += 8) { @@ -167,7 +167,7 @@ namespace Ryujinx.HLE.Input { lock (ShMemLock) { - foreach ((AMemory Memory, long Position) in ShMemPositions) + foreach ((AMemory Memory, long Position, long Size) in ShMemPositions) { long ControllerOffset = Position + HidControllersOffset; @@ -218,7 +218,7 @@ namespace Ryujinx.HLE.Input { lock (ShMemLock) { - foreach ((AMemory Memory, long Position) in ShMemPositions) + foreach ((AMemory Memory, long Position, long Size) in ShMemPositions) { long TouchScreenOffset = Position + HidTouchScreenOffset; diff --git a/Ryujinx.HLE/Logging/LogClass.cs b/Ryujinx.HLE/Logging/LogClass.cs index c377ace662..95cae7e0ab 100644 --- a/Ryujinx.HLE/Logging/LogClass.cs +++ b/Ryujinx.HLE/Logging/LogClass.cs @@ -4,6 +4,7 @@ namespace Ryujinx.HLE.Logging { Audio, Cpu, + Font, Gpu, Hid, Kernel, diff --git a/Ryujinx.HLE/OsHle/Handles/HSharedMem.cs b/Ryujinx.HLE/OsHle/Handles/HSharedMem.cs index 6426e585ea..cd3d82237c 100644 --- a/Ryujinx.HLE/OsHle/Handles/HSharedMem.cs +++ b/Ryujinx.HLE/OsHle/Handles/HSharedMem.cs @@ -6,37 +6,37 @@ namespace Ryujinx.HLE.OsHle.Handles { class HSharedMem { - private List<(AMemory, long)> Positions; + private List<(AMemory, long, long)> Positions; public EventHandler MemoryMapped; public EventHandler MemoryUnmapped; public HSharedMem() { - Positions = new List<(AMemory, long)>(); + Positions = new List<(AMemory, long, long)>(); } - public void AddVirtualPosition(AMemory Memory, long Position) + public void AddVirtualPosition(AMemory Memory, long Position, long Size) { lock (Positions) { - Positions.Add((Memory, Position)); + Positions.Add((Memory, Position, Size)); MemoryMapped?.Invoke(this, EventArgs.Empty); } } - public void RemoveVirtualPosition(AMemory Memory, long Position) + public void RemoveVirtualPosition(AMemory Memory, long Position, long Size) { lock (Positions) { - Positions.Remove((Memory, Position)); + Positions.Remove((Memory, Position, Size)); MemoryUnmapped?.Invoke(this, EventArgs.Empty); } } - public (AMemory, long)[] GetVirtualPositions() + public (AMemory, long, long)[] GetVirtualPositions() { return Positions.ToArray(); } diff --git a/Ryujinx.HLE/OsHle/Kernel/SvcHandler.cs b/Ryujinx.HLE/OsHle/Kernel/SvcHandler.cs index e816c44ec3..6f7bc42fd7 100644 --- a/Ryujinx.HLE/OsHle/Kernel/SvcHandler.cs +++ b/Ryujinx.HLE/OsHle/Kernel/SvcHandler.cs @@ -22,7 +22,7 @@ namespace Ryujinx.HLE.OsHle.Kernel private ConcurrentDictionary SyncWaits; - private HashSet<(HSharedMem, long)> MappedSharedMems; + private HashSet<(HSharedMem, long, long)> MappedSharedMems; private ulong CurrentHeapSize; @@ -83,7 +83,7 @@ namespace Ryujinx.HLE.OsHle.Kernel SyncWaits = new ConcurrentDictionary(); - MappedSharedMems = new HashSet<(HSharedMem, long)>(); + MappedSharedMems = new HashSet<(HSharedMem, long, long)>(); } static SvcHandler() @@ -138,9 +138,9 @@ namespace Ryujinx.HLE.OsHle.Kernel { lock (MappedSharedMems) { - foreach ((HSharedMem SharedMem, long Position) in MappedSharedMems) + foreach ((HSharedMem SharedMem, long Position, long Size) in MappedSharedMems) { - SharedMem.RemoveVirtualPosition(Memory, Position); + SharedMem.RemoveVirtualPosition(Memory, Position, Size); } MappedSharedMems.Clear(); diff --git a/Ryujinx.HLE/OsHle/Kernel/SvcMemory.cs b/Ryujinx.HLE/OsHle/Kernel/SvcMemory.cs index bb73f1ea17..f10cad7a20 100644 --- a/Ryujinx.HLE/OsHle/Kernel/SvcMemory.cs +++ b/Ryujinx.HLE/OsHle/Kernel/SvcMemory.cs @@ -174,15 +174,15 @@ namespace Ryujinx.HLE.OsHle.Kernel AMemoryHelper.FillWithZeros(Memory, Src, (int)Size); + SharedMem.AddVirtualPosition(Memory, Src, Size); + Memory.Manager.Reprotect(Src, Size, (AMemoryPerm)Perm); lock (MappedSharedMems) { - MappedSharedMems.Add((SharedMem, Src)); + MappedSharedMems.Add((SharedMem, Src, Size)); } - SharedMem.AddVirtualPosition(Memory, Src); - ThreadState.X0 = 0; } @@ -210,11 +210,11 @@ namespace Ryujinx.HLE.OsHle.Kernel { Memory.Manager.Unmap(Src, Size, (int)MemoryType.SharedMemory); - SharedMem.RemoveVirtualPosition(Memory, Src); + SharedMem.RemoveVirtualPosition(Memory, Src, Size); lock (MappedSharedMems) { - MappedSharedMems.Remove((SharedMem, Src)); + MappedSharedMems.Remove((SharedMem, Src, Size)); } ThreadState.X0 = 0; diff --git a/Ryujinx.HLE/OsHle/Kernel/SvcSystem.cs b/Ryujinx.HLE/OsHle/Kernel/SvcSystem.cs index 08305522f5..a968a1dbc1 100644 --- a/Ryujinx.HLE/OsHle/Kernel/SvcSystem.cs +++ b/Ryujinx.HLE/OsHle/Kernel/SvcSystem.cs @@ -242,7 +242,6 @@ namespace Ryujinx.HLE.OsHle.Kernel Process.Scheduler.Suspend(CurrThread); IpcMessage Cmd = new IpcMessage(CmdData, CmdPtr); - long Result = IpcHandler.IpcCall(Ns, Process, Memory, Session, Cmd, CmdPtr); Thread.Yield(); diff --git a/Ryujinx.HLE/OsHle/Services/Pl/ISharedFontManager.cs b/Ryujinx.HLE/OsHle/Services/Pl/ISharedFontManager.cs index 9f85f3d102..b8447ac65c 100644 --- a/Ryujinx.HLE/OsHle/Services/Pl/ISharedFontManager.cs +++ b/Ryujinx.HLE/OsHle/Services/Pl/ISharedFontManager.cs @@ -1,3 +1,4 @@ +using Ryujinx.HLE.Font; using Ryujinx.HLE.OsHle.Ipc; using System.Collections.Generic; @@ -13,11 +14,12 @@ namespace Ryujinx.HLE.OsHle.Services.Pl { m_Commands = new Dictionary() { - { 0, RequestLoad }, - { 1, GetLoadState }, - { 2, GetFontSize }, - { 3, GetSharedMemoryAddressOffset }, - { 4, GetSharedMemoryNativeHandle } + { 0, RequestLoad }, + { 1, GetLoadState }, + { 2, GetFontSize }, + { 3, GetSharedMemoryAddressOffset }, + { 4, GetSharedMemoryNativeHandle }, + { 5, GetSharedFontInOrderOfPriority } }; } @@ -25,26 +27,34 @@ namespace Ryujinx.HLE.OsHle.Services.Pl { SharedFontType FontType = (SharedFontType)Context.RequestData.ReadInt32(); + Context.Ns.Font.Load(FontType); + return 0; } public long GetLoadState(ServiceCtx Context) { - Context.ResponseData.Write(1); //Loaded + SharedFontType FontType = (SharedFontType)Context.RequestData.ReadInt32(); + + Context.ResponseData.Write(Context.Ns.Font.GetLoadState(FontType)); return 0; } public long GetFontSize(ServiceCtx Context) { - Context.ResponseData.Write(Horizon.FontSize); + SharedFontType FontType = (SharedFontType)Context.RequestData.ReadInt32(); + + Context.ResponseData.Write(Context.Ns.Font.GetFontSize(FontType)); return 0; } public long GetSharedMemoryAddressOffset(ServiceCtx Context) { - Context.ResponseData.Write(0); + SharedFontType FontType = (SharedFontType)Context.RequestData.ReadInt32(); + + Context.ResponseData.Write(Context.Ns.Font.GetSharedMemoryAddressOffset(FontType)); return 0; } @@ -57,5 +67,51 @@ namespace Ryujinx.HLE.OsHle.Services.Pl return 0; } + + private uint AddFontToOrderOfPriorityList(ServiceCtx Context, SharedFontType FontType, uint BufferPos, out uint LoadState) + { + long TypesPosition = Context.Request.ReceiveBuff[0].Position; + long TypesSize = Context.Request.ReceiveBuff[0].Size; + + long OffsetsPosition = Context.Request.ReceiveBuff[1].Position; + long OffsetsSize = Context.Request.ReceiveBuff[1].Size; + + long FontSizeBufferPosition = Context.Request.ReceiveBuff[2].Position; + long FontSizeBufferSize = Context.Request.ReceiveBuff[2].Size; + + LoadState = Context.Ns.Font.GetLoadState(FontType); + + if (BufferPos >= TypesSize || BufferPos >= OffsetsSize || BufferPos >= FontSizeBufferSize) + { + return 0; + } + + Context.Memory.WriteUInt32(TypesPosition + BufferPos, (uint)FontType); + Context.Memory.WriteUInt32(OffsetsPosition + BufferPos, Context.Ns.Font.GetSharedMemoryAddressOffset(FontType)); + Context.Memory.WriteUInt32(FontSizeBufferPosition + BufferPos, Context.Ns.Font.GetFontSize(FontType)); + + BufferPos += 4; + + return BufferPos; + } + + public long GetSharedFontInOrderOfPriority(ServiceCtx Context) + { + ulong LanguageCode = Context.RequestData.ReadUInt64(); + uint LoadedCount = 0; + uint BufferPos = 0; + uint Loaded = 0; + + for (int Type = 0; Type < Context.Ns.Font.Count; Type++) + { + BufferPos = AddFontToOrderOfPriorityList(Context, (SharedFontType)Type, BufferPos, out Loaded); + LoadedCount += Loaded; + } + + Context.ResponseData.Write(LoadedCount); + Context.ResponseData.Write(Context.Ns.Font.Count); + + return 0; + } } } \ No newline at end of file diff --git a/Ryujinx.HLE/Resource/InvalidSystemResourceException.cs b/Ryujinx.HLE/Resource/InvalidSystemResourceException.cs new file mode 100644 index 0000000000..35c4874a34 --- /dev/null +++ b/Ryujinx.HLE/Resource/InvalidSystemResourceException.cs @@ -0,0 +1,13 @@ +using System; + +namespace Ryujinx.HLE.Resource +{ + public class InvalidSystemResourceException : Exception + { + public InvalidSystemResourceException(string message) + : base(message) + { + } + + } +} diff --git a/Ryujinx.HLE/Ryujinx.HLE.csproj b/Ryujinx.HLE/Ryujinx.HLE.csproj index acef4be9e7..f7fb84a588 100644 --- a/Ryujinx.HLE/Ryujinx.HLE.csproj +++ b/Ryujinx.HLE/Ryujinx.HLE.csproj @@ -1,4 +1,4 @@ - + netcoreapp2.1 diff --git a/Ryujinx.HLE/Switch.cs b/Ryujinx.HLE/Switch.cs index 74c0564a9f..a80ca86c19 100644 --- a/Ryujinx.HLE/Switch.cs +++ b/Ryujinx.HLE/Switch.cs @@ -1,5 +1,6 @@ using Ryujinx.Audio; using Ryujinx.Graphics.Gal; +using Ryujinx.HLE.Font; using Ryujinx.HLE.Gpu; using Ryujinx.HLE.Input; using Ryujinx.HLE.Logging; @@ -27,6 +28,8 @@ namespace Ryujinx.HLE public Hid Hid { get; private set; } + public SharedFontManager Font { get; private set; } + public event EventHandler Finish; public Switch(IGalRenderer Renderer, IAalOutput AudioOut) @@ -57,8 +60,13 @@ namespace Ryujinx.HLE Hid = new Hid(Log); - Os.HidSharedMem.MemoryMapped += Hid.ShMemMap; - Os.HidSharedMem.MemoryUnmapped += Hid.ShMemUnmap; + Font = new SharedFontManager(Log, VFs.GetSystemPath()); + + Os.HidSharedMem.MemoryMapped += Hid.ShMemMap; + Os.HidSharedMem.MemoryUnmapped += Hid.ShMemUnmap; + + Os.FontSharedMem.MemoryMapped += Font.ShMemMap; + Os.FontSharedMem.MemoryUnmapped += Font.ShMemUnmap; } public void LoadCart(string ExeFsDir, string RomFsFile = null) diff --git a/Ryujinx.HLE/VirtualFileSystem.cs b/Ryujinx.HLE/VirtualFileSystem.cs index df1fc9db13..31b8e184ce 100644 --- a/Ryujinx.HLE/VirtualFileSystem.cs +++ b/Ryujinx.HLE/VirtualFileSystem.cs @@ -8,6 +8,7 @@ namespace Ryujinx.HLE private const string BasePath = "RyuFs"; private const string NandPath = "nand"; private const string SdCardPath = "sdmc"; + private const string SystemPath = "system"; public Stream RomFs { get; private set; } @@ -45,6 +46,8 @@ namespace Ryujinx.HLE public string GetGameSavesPath() => MakeDirAndGetFullPath(NandPath); + public string GetSystemPath() => MakeDirAndGetFullPath(SystemPath); + public string SwitchPathToSystemPath(string SwitchPath) { string[] Parts = SwitchPath.Split(":"); From 221270db90300a084007d154867be89bb5fddedf Mon Sep 17 00:00:00 2001 From: gdkchan Date: Sun, 5 Aug 2018 02:54:21 -0300 Subject: [PATCH 34/34] More accurate impl of FMINNM/FMAXNM, add vector variants (#296) * More accurate impl of FMINNM/FMAXNM, add vector variants * Optimize for the 0 case when op1 != op2 * Address PR feedback --- ChocolArm64/AOpCodeTable.cs | 6 +- .../Instruction/AInstEmitSimdArithmetic.cs | 116 +++---- ChocolArm64/Instruction/ASoftFloat.cs | 284 +++++++++++++++++- ChocolArm64/Instruction/AVectorHelper.cs | 80 ----- Ryujinx.Tests/Cpu/CpuTestSimdArithmetic.cs | 24 +- 5 files changed, 336 insertions(+), 174 deletions(-) diff --git a/ChocolArm64/AOpCodeTable.cs b/ChocolArm64/AOpCodeTable.cs index aa8b1be606..5ea38b0535 100644 --- a/ChocolArm64/AOpCodeTable.cs +++ b/ChocolArm64/AOpCodeTable.cs @@ -267,11 +267,13 @@ namespace ChocolArm64 SetA64("0>1011100<1xxxxx111111xxxxxxxxxx", AInstEmit.Fdiv_V, typeof(AOpCodeSimdReg)); SetA64("000111110x0xxxxx0xxxxxxxxxxxxxxx", AInstEmit.Fmadd_S, typeof(AOpCodeSimdReg)); SetA64("000111100x1xxxxx010010xxxxxxxxxx", AInstEmit.Fmax_S, typeof(AOpCodeSimdReg)); - SetA64("0x0011100x1xxxxx111101xxxxxxxxxx", AInstEmit.Fmax_V, typeof(AOpCodeSimdReg)); + SetA64("0>0011100<1xxxxx111101xxxxxxxxxx", AInstEmit.Fmax_V, typeof(AOpCodeSimdReg)); SetA64("000111100x1xxxxx011010xxxxxxxxxx", AInstEmit.Fmaxnm_S, typeof(AOpCodeSimdReg)); + SetA64("0>0011100<1xxxxx110001xxxxxxxxxx", AInstEmit.Fmaxnm_V, typeof(AOpCodeSimdReg)); SetA64("000111100x1xxxxx010110xxxxxxxxxx", AInstEmit.Fmin_S, typeof(AOpCodeSimdReg)); - SetA64("0x0011101x1xxxxx111101xxxxxxxxxx", AInstEmit.Fmin_V, typeof(AOpCodeSimdReg)); + SetA64("0>0011101<1xxxxx111101xxxxxxxxxx", AInstEmit.Fmin_V, typeof(AOpCodeSimdReg)); SetA64("000111100x1xxxxx011110xxxxxxxxxx", AInstEmit.Fminnm_S, typeof(AOpCodeSimdReg)); + SetA64("0>0011101<1xxxxx110001xxxxxxxxxx", AInstEmit.Fminnm_V, typeof(AOpCodeSimdReg)); SetA64("010111111<0011100<1xxxxx110011xxxxxxxxxx", AInstEmit.Fmla_V, typeof(AOpCodeSimdReg)); SetA64("0x0011111< { - if (Op.Size == 0) - { - AVectorHelper.EmitCall(Context, nameof(AVectorHelper.MaxF)); - } - else if (Op.Size == 1) - { - AVectorHelper.EmitCall(Context, nameof(AVectorHelper.Max)); - } - else - { - throw new InvalidOperationException(); - } + EmitBinarySoftFloatCall(Context, nameof(ASoftFloat.Max)); }); } public static void Fmax_V(AILEmitterCtx Context) { - AOpCodeSimd Op = (AOpCodeSimd)Context.CurrOp; - EmitVectorBinaryOpF(Context, () => { - if (Op.Size == 0) - { - AVectorHelper.EmitCall(Context, nameof(AVectorHelper.MaxF)); - } - else if (Op.Size == 1) - { - AVectorHelper.EmitCall(Context, nameof(AVectorHelper.Max)); - } - else - { - throw new InvalidOperationException(); - } - }); - } - - public static void Fmin_S(AILEmitterCtx Context) - { - AOpCodeSimd Op = (AOpCodeSimd)Context.CurrOp; - - EmitScalarBinaryOpF(Context, () => - { - if (Op.Size == 0) - { - AVectorHelper.EmitCall(Context, nameof(AVectorHelper.MinF)); - } - else if (Op.Size == 1) - { - AVectorHelper.EmitCall(Context, nameof(AVectorHelper.Min)); - } - else - { - throw new InvalidOperationException(); - } - }); - } - - public static void Fmin_V(AILEmitterCtx Context) - { - AOpCodeSimd Op = (AOpCodeSimd)Context.CurrOp; - - int SizeF = Op.Size & 1; - - EmitVectorBinaryOpF(Context, () => - { - if (SizeF == 0) - { - AVectorHelper.EmitCall(Context, nameof(AVectorHelper.MinF)); - } - else if (SizeF == 1) - { - AVectorHelper.EmitCall(Context, nameof(AVectorHelper.Min)); - } - else - { - throw new InvalidOperationException(); - } + EmitBinarySoftFloatCall(Context, nameof(ASoftFloat.Max)); }); } public static void Fmaxnm_S(AILEmitterCtx Context) { - Fmax_S(Context); + EmitScalarBinaryOpF(Context, () => + { + EmitBinarySoftFloatCall(Context, nameof(ASoftFloat.MaxNum)); + }); + } + + public static void Fmaxnm_V(AILEmitterCtx Context) + { + EmitVectorBinaryOpF(Context, () => + { + EmitBinarySoftFloatCall(Context, nameof(ASoftFloat.MaxNum)); + }); + } + + public static void Fmin_S(AILEmitterCtx Context) + { + EmitScalarBinaryOpF(Context, () => + { + EmitBinarySoftFloatCall(Context, nameof(ASoftFloat.Min)); + }); + } + + public static void Fmin_V(AILEmitterCtx Context) + { + EmitVectorBinaryOpF(Context, () => + { + EmitBinarySoftFloatCall(Context, nameof(ASoftFloat.Min)); + }); } public static void Fminnm_S(AILEmitterCtx Context) { - Fmin_S(Context); + EmitScalarBinaryOpF(Context, () => + { + EmitBinarySoftFloatCall(Context, nameof(ASoftFloat.MinNum)); + }); + } + + public static void Fminnm_V(AILEmitterCtx Context) + { + EmitVectorBinaryOpF(Context, () => + { + EmitBinarySoftFloatCall(Context, nameof(ASoftFloat.MinNum)); + }); } public static void Fmla_Se(AILEmitterCtx Context) diff --git a/ChocolArm64/Instruction/ASoftFloat.cs b/ChocolArm64/Instruction/ASoftFloat.cs index 27f4f7fb4f..8afa4002a8 100644 --- a/ChocolArm64/Instruction/ASoftFloat.cs +++ b/ChocolArm64/Instruction/ASoftFloat.cs @@ -79,7 +79,7 @@ namespace ChocolArm64.Instruction if (scaled == 0) { // Zero -> Infinity - return BitConverter.Int64BitsToDouble((long)(x_sign | 0x7ff0000000000000)); + return BitConverter.Int64BitsToDouble((long)(x_sign | 0x7FF0000000000000)); } // Denormal @@ -94,7 +94,7 @@ namespace ChocolArm64.Instruction if (x_sign != 0) { // Negative -> NaN - return BitConverter.Int64BitsToDouble((long)0x7ff8000000000000); + return BitConverter.Int64BitsToDouble((long)0x7FF8000000000000); } if (x_exp == 0x7ff && scaled == 0) @@ -153,7 +153,7 @@ namespace ChocolArm64.Instruction if (scaled == 0) { // Zero -> Infinity - return BitConverter.Int64BitsToDouble((long)(x_sign | 0x7ff0000000000000)); + return BitConverter.Int64BitsToDouble((long)(x_sign | 0x7FF0000000000000)); } // Denormal @@ -208,8 +208,8 @@ namespace ChocolArm64.Instruction ulong op1_other = op1_bits & 0x7FFFFFFFFFFFFFFF; ulong op2_other = op2_bits & 0x7FFFFFFFFFFFFFFF; - bool inf1 = op1_other == 0x7ff0000000000000; - bool inf2 = op2_other == 0x7ff0000000000000; + bool inf1 = op1_other == 0x7FF0000000000000; + bool inf2 = op2_other == 0x7FF0000000000000; bool zero1 = op1_other == 0; bool zero2 = op2_other == 0; @@ -220,7 +220,7 @@ namespace ChocolArm64.Instruction else if (inf1 || inf2) { // Infinity - return BitConverter.Int64BitsToDouble((long)(0x7ff0000000000000 | (op1_sign ^ op2_sign))); + return BitConverter.Int64BitsToDouble((long)(0x7FF0000000000000 | (op1_sign ^ op2_sign))); } return 2.0 + op1 * op2; @@ -261,5 +261,277 @@ namespace ChocolArm64.Instruction uint new_exp = (uint)((exponent + 127) & 0xFF) << 23; return BitConverter.Int32BitsToSingle((int)((x_sign << 31) | new_exp | (x_mantissa << 13))); } + + public static float MaxNum(float op1, float op2) + { + uint op1_bits = (uint)BitConverter.SingleToInt32Bits(op1); + uint op2_bits = (uint)BitConverter.SingleToInt32Bits(op2); + + if (IsQNaN(op1_bits) && !IsQNaN(op2_bits)) + { + op1 = float.NegativeInfinity; + } + else if (!IsQNaN(op1_bits) && IsQNaN(op2_bits)) + { + op2 = float.NegativeInfinity; + } + + return Max(op1, op2); + } + + public static double MaxNum(double op1, double op2) + { + ulong op1_bits = (ulong)BitConverter.DoubleToInt64Bits(op1); + ulong op2_bits = (ulong)BitConverter.DoubleToInt64Bits(op2); + + if (IsQNaN(op1_bits) && !IsQNaN(op2_bits)) + { + op1 = double.NegativeInfinity; + } + else if (!IsQNaN(op1_bits) && IsQNaN(op2_bits)) + { + op2 = double.NegativeInfinity; + } + + return Max(op1, op2); + } + + public static float Max(float op1, float op2) + { + // Fast path + if (op1 > op2) + { + return op1; + } + + if (op1 < op2 || (op1 == op2 && op2 != 0)) + { + return op2; + } + + uint op1_bits = (uint)BitConverter.SingleToInt32Bits(op1); + uint op2_bits = (uint)BitConverter.SingleToInt32Bits(op2); + + // Handle NaN cases + if (ProcessNaNs(op1_bits, op2_bits, out uint op_bits)) + { + return BitConverter.Int32BitsToSingle((int)op_bits); + } + + // Return the most positive zero + if ((op1_bits & op2_bits) == 0x80000000u) + { + return BitConverter.Int32BitsToSingle(int.MinValue); + } + + return 0; + } + + public static double Max(double op1, double op2) + { + // Fast path + if (op1 > op2) + { + return op1; + } + + if (op1 < op2 || (op1 == op2 && op2 != 0)) + { + return op2; + } + + ulong op1_bits = (ulong)BitConverter.DoubleToInt64Bits(op1); + ulong op2_bits = (ulong)BitConverter.DoubleToInt64Bits(op2); + + // Handle NaN cases + if (ProcessNaNs(op1_bits, op2_bits, out ulong op_bits)) + { + return BitConverter.Int64BitsToDouble((long)op_bits); + } + + // Return the most positive zero + if ((op1_bits & op2_bits) == 0x8000000000000000ul) + { + return BitConverter.Int64BitsToDouble(long.MinValue); + } + + return 0; + } + + public static float MinNum(float op1, float op2) + { + uint op1_bits = (uint)BitConverter.SingleToInt32Bits(op1); + uint op2_bits = (uint)BitConverter.SingleToInt32Bits(op2); + + if (IsQNaN(op1_bits) && !IsQNaN(op2_bits)) + { + op1 = float.PositiveInfinity; + } + else if (!IsQNaN(op1_bits) && IsQNaN(op2_bits)) + { + op2 = float.PositiveInfinity; + } + + return Max(op1, op2); + } + + public static double MinNum(double op1, double op2) + { + ulong op1_bits = (ulong)BitConverter.DoubleToInt64Bits(op1); + ulong op2_bits = (ulong)BitConverter.DoubleToInt64Bits(op2); + + if (IsQNaN(op1_bits) && !IsQNaN(op2_bits)) + { + op1 = double.PositiveInfinity; + } + else if (!IsQNaN(op1_bits) && IsQNaN(op2_bits)) + { + op2 = double.PositiveInfinity; + } + + return Min(op1, op2); + } + + public static float Min(float op1, float op2) + { + // Fast path + if (op1 < op2) + { + return op1; + } + + if (op1 > op2 || (op1 == op2 && op2 != 0)) + { + return op2; + } + + uint op1_bits = (uint)BitConverter.SingleToInt32Bits(op1); + uint op2_bits = (uint)BitConverter.SingleToInt32Bits(op2); + + // Handle NaN cases + if (ProcessNaNs(op1_bits, op2_bits, out uint op_bits)) + { + return BitConverter.Int32BitsToSingle((int)op_bits); + } + + // Return the most negative zero + if ((op1_bits | op2_bits) == 0x80000000u) + { + return BitConverter.Int32BitsToSingle(int.MinValue); + } + + return 0; + } + + public static double Min(double op1, double op2) + { + // Fast path + if (op1 < op2) + { + return op1; + } + + if (op1 > op2 || (op1 == op2 && op2 != 0)) + { + return op2; + } + + ulong op1_bits = (ulong)BitConverter.DoubleToInt64Bits(op1); + ulong op2_bits = (ulong)BitConverter.DoubleToInt64Bits(op2); + + // Handle NaN cases + if (ProcessNaNs(op1_bits, op2_bits, out ulong op_bits)) + { + return BitConverter.Int64BitsToDouble((long)op_bits); + } + + // Return the most negative zero + if ((op1_bits | op2_bits) == 0x8000000000000000ul) + { + return BitConverter.Int64BitsToDouble(long.MinValue); + } + + return 0; + } + + private static bool ProcessNaNs(uint op1_bits, uint op2_bits, out uint op_bits) + { + if (IsSNaN(op1_bits)) + { + op_bits = op1_bits | (1u << 22); // op1 is SNaN, return QNaN op1 + } + else if (IsSNaN(op2_bits)) + { + op_bits = op2_bits | (1u << 22); // op2 is SNaN, return QNaN op2 + } + else if (IsQNaN(op1_bits)) + { + op_bits = op1_bits; // op1 is QNaN, return QNaN op1 + } + else if (IsQNaN(op2_bits)) + { + op_bits = op2_bits; // op2 is QNaN, return QNaN op2 + } + else + { + op_bits = 0; + + return false; + } + + return true; + } + + private static bool ProcessNaNs(ulong op1_bits, ulong op2_bits, out ulong op_bits) + { + if (IsSNaN(op1_bits)) + { + op_bits = op1_bits | (1ul << 51); // op1 is SNaN, return QNaN op1 + } + else if (IsSNaN(op2_bits)) + { + op_bits = op2_bits | (1ul << 51); // op2 is SNaN, return QNaN op2 + } + else if (IsQNaN(op1_bits)) + { + op_bits = op1_bits; // op1 is QNaN, return QNaN op1 + } + else if (IsQNaN(op2_bits)) + { + op_bits = op2_bits; // op2 is QNaN, return QNaN op2 + } + else + { + op_bits = 0; + + return false; + } + + return true; + } + + private static bool IsQNaN(uint op_bits) + { + return (op_bits & 0x007FFFFF) != 0 && + (op_bits & 0x7FC00000) == 0x7FC00000; + } + + private static bool IsQNaN(ulong op_bits) + { + return (op_bits & 0x000FFFFFFFFFFFFF) != 0 && + (op_bits & 0x7FF8000000000000) == 0x7FF8000000000000; + } + + private static bool IsSNaN(uint op_bits) + { + return (op_bits & 0x007FFFFF) != 0 && + (op_bits & 0x7FC00000) == 0x7F800000; + } + + private static bool IsSNaN(ulong op_bits) + { + return (op_bits & 0x000FFFFFFFFFFFFF) != 0 && + (op_bits & 0x7FF8000000000000) == 0x7FF0000000000000; + } } } \ No newline at end of file diff --git a/ChocolArm64/Instruction/AVectorHelper.cs b/ChocolArm64/Instruction/AVectorHelper.cs index a0f887b043..b2d53740eb 100644 --- a/ChocolArm64/Instruction/AVectorHelper.cs +++ b/ChocolArm64/Instruction/AVectorHelper.cs @@ -93,86 +93,6 @@ namespace ChocolArm64.Instruction Value < ulong.MinValue ? ulong.MinValue : (ulong)Value; } - public static double Max(double LHS, double RHS) - { - if (LHS == 0.0 && RHS == 0.0) - { - if (BitConverter.DoubleToInt64Bits(LHS) < 0 && - BitConverter.DoubleToInt64Bits(RHS) < 0) - return -0.0; - - return 0.0; - } - - if (LHS > RHS) - return LHS; - - if (double.IsNaN(LHS)) - return LHS; - - return RHS; - } - - public static float MaxF(float LHS, float RHS) - { - if (LHS == 0.0 && RHS == 0.0) - { - if (BitConverter.SingleToInt32Bits(LHS) < 0 && - BitConverter.SingleToInt32Bits(RHS) < 0) - return -0.0f; - - return 0.0f; - } - - if (LHS > RHS) - return LHS; - - if (float.IsNaN(LHS)) - return LHS; - - return RHS; - } - - public static double Min(double LHS, double RHS) - { - if (LHS == 0.0 && RHS == 0.0) - { - if (BitConverter.DoubleToInt64Bits(LHS) < 0 || - BitConverter.DoubleToInt64Bits(RHS) < 0) - return -0.0; - - return 0.0; - } - - if (LHS < RHS) - return LHS; - - if (double.IsNaN(LHS)) - return LHS; - - return RHS; - } - - public static float MinF(float LHS, float RHS) - { - if (LHS == 0.0 && RHS == 0.0) - { - if (BitConverter.SingleToInt32Bits(LHS) < 0 || - BitConverter.SingleToInt32Bits(RHS) < 0) - return -0.0f; - - return 0.0f; - } - - if (LHS < RHS) - return LHS; - - if (float.IsNaN(LHS)) - return LHS; - - return RHS; - } - public static double Round(double Value, int Fpcr) { switch ((ARoundMode)((Fpcr >> 22) & 3)) diff --git a/Ryujinx.Tests/Cpu/CpuTestSimdArithmetic.cs b/Ryujinx.Tests/Cpu/CpuTestSimdArithmetic.cs index 8e2d9a366e..53e1554291 100644 --- a/Ryujinx.Tests/Cpu/CpuTestSimdArithmetic.cs +++ b/Ryujinx.Tests/Cpu/CpuTestSimdArithmetic.cs @@ -19,9 +19,9 @@ namespace Ryujinx.Tests.Cpu [TestCase(0x1E224820u, 0x000000007F7FFFFFul, 0x00000000807FFFFFul, 0x000000007F7FFFFFul)] [TestCase(0x1E224820u, 0x000000007FC00000ul, 0x000000003F800000ul, 0x000000007FC00000ul)] [TestCase(0x1E224820u, 0x000000003F800000ul, 0x000000007FC00000ul, 0x000000007FC00000ul)] - [TestCase(0x1E224820u, 0x000000007F800001ul, 0x000000007FC00042ul, 0x000000007FC00001ul, Ignore = "NaN test.")] - [TestCase(0x1E224820u, 0x000000007FC00042ul, 0x000000007F800001ul, 0x000000007FC00001ul, Ignore = "NaN test.")] - [TestCase(0x1E224820u, 0x000000007FC0000Aul, 0x000000007FC0000Bul, 0x000000007FC0000Aul, Ignore = "NaN test.")] + [TestCase(0x1E224820u, 0x000000007F800001ul, 0x000000007FC00042ul, 0x000000007FC00001ul)] + [TestCase(0x1E224820u, 0x000000007FC00042ul, 0x000000007F800001ul, 0x000000007FC00001ul)] + [TestCase(0x1E224820u, 0x000000007FC0000Aul, 0x000000007FC0000Bul, 0x000000007FC0000Aul)] [TestCase(0x1E624820u, 0x0000000000000000ul, 0x8000000000000000ul, 0x0000000000000000ul)] [TestCase(0x1E624820u, 0x8000000000000000ul, 0x0000000000000000ul, 0x0000000000000000ul)] [TestCase(0x1E624820u, 0x8000000000000000ul, 0x8000000000000000ul, 0x8000000000000000ul)] @@ -45,9 +45,9 @@ namespace Ryujinx.Tests.Cpu [TestCase(0x7F7FFFFFu, 0x7F7FFFFFu, 0x807FFFFFu, 0x807FFFFFu, 0x7F7FFFFFu, 0x7F7FFFFFu)] [TestCase(0x7FC00000u, 0x7FC00000u, 0x3F800000u, 0x3F800000u, 0x7FC00000u, 0x7FC00000u)] [TestCase(0x3F800000u, 0x3F800000u, 0x7FC00000u, 0x7FC00000u, 0x7FC00000u, 0x7FC00000u)] - [TestCase(0x7F800001u, 0x7F800001u, 0x7FC00042u, 0x7FC00042u, 0x7FC00001u, 0x7FC00001u, Ignore = "NaN test.")] - [TestCase(0x7FC00042u, 0x7FC00042u, 0x7F800001u, 0x7F800001u, 0x7FC00001u, 0x7FC00001u, Ignore = "NaN test.")] - [TestCase(0x7FC0000Au, 0x7FC0000Au, 0x7FC0000Bu, 0x7FC0000Bu, 0x7FC0000Au, 0x7FC0000Au, Ignore = "NaN test.")] + [TestCase(0x7F800001u, 0x7F800001u, 0x7FC00042u, 0x7FC00042u, 0x7FC00001u, 0x7FC00001u)] + [TestCase(0x7FC00042u, 0x7FC00042u, 0x7F800001u, 0x7F800001u, 0x7FC00001u, 0x7FC00001u)] + [TestCase(0x7FC0000Au, 0x7FC0000Au, 0x7FC0000Bu, 0x7FC0000Bu, 0x7FC0000Au, 0x7FC0000Au)] public void Fmax_V(uint A, uint B, uint C, uint D, uint Result0, uint Result1) { uint Opcode = 0x4E22F420; @@ -71,9 +71,9 @@ namespace Ryujinx.Tests.Cpu [TestCase(0x1E225820u, 0x000000007F7FFFFFul, 0x00000000807FFFFFul, 0x00000000807FFFFFul)] [TestCase(0x1E225820u, 0x000000007FC00000ul, 0x000000003F800000ul, 0x000000007FC00000ul)] [TestCase(0x1E225820u, 0x000000003F800000ul, 0x000000007FC00000ul, 0x000000007FC00000ul)] - [TestCase(0x1E225820u, 0x000000007F800001ul, 0x000000007FC00042ul, 0x000000007FC00001ul, Ignore = "NaN test.")] - [TestCase(0x1E225820u, 0x000000007FC00042ul, 0x000000007F800001ul, 0x000000007FC00001ul, Ignore = "NaN test.")] - [TestCase(0x1E225820u, 0x000000007FC0000Aul, 0x000000007FC0000Bul, 0x000000007FC0000Aul, Ignore = "NaN test.")] + [TestCase(0x1E225820u, 0x000000007F800001ul, 0x000000007FC00042ul, 0x000000007FC00001ul)] + [TestCase(0x1E225820u, 0x000000007FC00042ul, 0x000000007F800001ul, 0x000000007FC00001ul)] + [TestCase(0x1E225820u, 0x000000007FC0000Aul, 0x000000007FC0000Bul, 0x000000007FC0000Aul)] [TestCase(0x1E625820u, 0x0000000000000000ul, 0x8000000000000000ul, 0x8000000000000000ul)] [TestCase(0x1E625820u, 0x8000000000000000ul, 0x0000000000000000ul, 0x8000000000000000ul)] [TestCase(0x1E625820u, 0x8000000000000000ul, 0x8000000000000000ul, 0x8000000000000000ul)] @@ -97,9 +97,9 @@ namespace Ryujinx.Tests.Cpu [TestCase(0x7F7FFFFFu, 0x7F7FFFFFu, 0x807FFFFFu, 0x807FFFFFu, 0x807FFFFFu, 0x807FFFFFu)] [TestCase(0x7FC00000u, 0x7FC00000u, 0x3F800000u, 0x3F800000u, 0x7FC00000u, 0x7FC00000u)] [TestCase(0x3F800000u, 0x3F800000u, 0x7FC00000u, 0x7FC00000u, 0x7FC00000u, 0x7FC00000u)] - [TestCase(0x7F800001u, 0x7F800001u, 0x7FC00042u, 0x7FC00042u, 0x7FC00001u, 0x7FC00001u, Ignore = "NaN test.")] - [TestCase(0x7FC00042u, 0x7FC00042u, 0x7F800001u, 0x7F800001u, 0x7FC00001u, 0x7FC00001u, Ignore = "NaN test.")] - [TestCase(0x7FC0000Au, 0x7FC0000Au, 0x7FC0000Bu, 0x7FC0000Bu, 0x7FC0000Au, 0x7FC0000Au, Ignore = "NaN test.")] + [TestCase(0x7F800001u, 0x7F800001u, 0x7FC00042u, 0x7FC00042u, 0x7FC00001u, 0x7FC00001u)] + [TestCase(0x7FC00042u, 0x7FC00042u, 0x7F800001u, 0x7F800001u, 0x7FC00001u, 0x7FC00001u)] + [TestCase(0x7FC0000Au, 0x7FC0000Au, 0x7FC0000Bu, 0x7FC0000Bu, 0x7FC0000Au, 0x7FC0000Au)] public void Fmin_V(uint A, uint B, uint C, uint D, uint Result0, uint Result1) { uint Opcode = 0x4EA2F420;