De-tile GOB when DMA copying from block linear to pitch kind memory regions (#3207)

* De-tile GOB when DMA copying from block linear to pitch kind memory regions

* XML docs + nits

* Remove using

* No flush for regular buffer copies

* Add back ulong casts, fix regression due to oversight

ARMeilleure/Decoders/Decoder.cs

@@ -18,7 +18,7 @@ namespace ARMeilleure.Decoders
         // For lower code quality translation, we set a lower limit since we're blocking execution.
         private const int MaxInstsPerFunctionLowCq = 500;
 
-        public static Block[] Decode(IMemoryManager memory, ulong address, ExecutionMode mode, bool highCq, bool singleBlock)
+        public static Block[] Decode(IMemoryManager memory, ulong address, ExecutionMode mode, bool highCq, DecoderMode dMode)
         {
             List<Block> blocks = new List<Block>();
 
@@ -38,7 +38,7 @@ namespace ARMeilleure.Decoders
                 {
                     block = new Block(blkAddress);
 
-                    if ((singleBlock && visited.Count >= 1) || opsCount > instructionLimit || !memory.IsMapped(blkAddress))
+                    if ((dMode != DecoderMode.MultipleBlocks && visited.Count >= 1) || opsCount > instructionLimit || !memory.IsMapped(blkAddress))
                     {
                         block.Exit = true;
                         block.EndAddress = blkAddress;
@@ -96,6 +96,12 @@ namespace ARMeilleure.Decoders
                         }
                     }
 
+                    if (dMode == DecoderMode.SingleInstruction)
+                    {
+                        // Only read at most one instruction
+                        limitAddress = currBlock.Address + 1;
+                    }
+
                     FillBlock(memory, mode, currBlock, limitAddress);
 
                     opsCount += currBlock.OpCodes.Count;
@@ -115,7 +121,7 @@ namespace ARMeilleure.Decoders
                             currBlock.Branch = GetBlock((ulong)op.Immediate);
                         }
 
-                        if (!IsUnconditionalBranch(lastOp) || isCall)
+                        if (isCall || !(IsUnconditionalBranch(lastOp) || IsTrap(lastOp)))
                         {
                             currBlock.Next = GetBlock(currBlock.EndAddress);
                         }
@@ -143,7 +149,7 @@ namespace ARMeilleure.Decoders
                 throw new InvalidOperationException($"Decoded a single empty exit block. Entry point = 0x{address:X}.");
             }
 
-            if (!singleBlock)
+            if (dMode == DecoderMode.MultipleBlocks)
             {
                 return TailCallRemover.RunPass(address, blocks);
             }
@@ -257,6 +263,11 @@ namespace ARMeilleure.Decoders
             // so we must consider such operations as a branch in potential aswell.
             if (opCode is IOpCode32Alu opAlu && opAlu.Rd == RegisterAlias.Aarch32Pc)
             {
+                if (opCode is OpCodeT32)
+                {
+                    return opCode.Instruction.Name != InstName.Tst && opCode.Instruction.Name != InstName.Teq &&
+                           opCode.Instruction.Name != InstName.Cmp && opCode.Instruction.Name != InstName.Cmn;
+                }
                 return true;
             }
 
@@ -318,9 +329,13 @@ namespace ARMeilleure.Decoders
         }
 
         private static bool IsException(OpCode opCode)
+        {
+            return IsTrap(opCode) || opCode.Instruction.Name == InstName.Svc;
+        }
+
+        private static bool IsTrap(OpCode opCode)
         {
             return opCode.Instruction.Name == InstName.Brk ||
-                   opCode.Instruction.Name == InstName.Svc ||
                    opCode.Instruction.Name == InstName.Trap ||
                    opCode.Instruction.Name == InstName.Und;
         }

ARMeilleure/Decoders/DecoderMode.cs

@@ -0,0 +1,9 @@
+namespace ARMeilleure.Decoders
+{
+    enum DecoderMode
+    {
+        MultipleBlocks,
+        SingleBlock,
+        SingleInstruction,
+    }
+}

ARMeilleure/Decoders/OpCode32.cs

@@ -13,11 +13,25 @@ namespace ARMeilleure.Decoders
             Cond = (Condition)((uint)opCode >> 28);
         }
 
+        public bool IsThumb()
+        {
+            return this is OpCodeT16 || this is OpCodeT32;
+        }
+
         public uint GetPc()
         {
             // Due to backwards compatibility and legacy behavior of ARMv4 CPUs pipeline,
             // the PC actually points 2 instructions ahead.
-            return (uint)Address + (uint)OpCodeSizeInBytes * 2;
+            if (IsThumb())
+            {
+                // PC is ahead by 4 in thumb mode whether or not the current instruction
+                // is 16 or 32 bit.
+                return (uint)Address + 4u;
+            }
+            else
+            {
+                return (uint)Address + 8u;
+            }
         }
     }
 }

ARMeilleure/Decoders/OpCodeT16BImm11.cs

@@ -1,6 +1,6 @@
 namespace ARMeilleure.Decoders
 {
-    class OpCodeT16BImm11 : OpCode32, IOpCode32BImm
+    class OpCodeT16BImm11 : OpCodeT16, IOpCode32BImm
     {
         public long Immediate { get; }
 

ARMeilleure/Decoders/OpCodeT16BImm8.cs

@@ -1,6 +1,6 @@
 namespace ARMeilleure.Decoders
 {
-    class OpCodeT16BImm8 : OpCode32, IOpCode32BImm
+    class OpCodeT16BImm8 : OpCodeT16, IOpCode32BImm
     {
         public long Immediate { get; }
 

ARMeilleure/Decoders/OpCodeT32.cs

@@ -0,0 +1,14 @@
+namespace ARMeilleure.Decoders
+{
+    class OpCodeT32 : OpCode32
+    {
+        public new static OpCode Create(InstDescriptor inst, ulong address, int opCode) => new OpCodeT32(inst, address, opCode);
+
+        public OpCodeT32(InstDescriptor inst, ulong address, int opCode) : base(inst, address, opCode)
+        {
+            Cond = Condition.Al;
+
+            OpCodeSizeInBytes = 4;
+        }
+    }
+}

ARMeilleure/Decoders/OpCodeT32Alu.cs

@@ -0,0 +1,20 @@
+namespace ARMeilleure.Decoders
+{
+    class OpCodeT32Alu : OpCodeT32, IOpCode32Alu
+    {
+        public int Rd { get; }
+        public int Rn { get; }
+
+        public bool? SetFlags { get; }
+
+        public new static OpCode Create(InstDescriptor inst, ulong address, int opCode) => new OpCodeT32Alu(inst, address, opCode);
+
+        public OpCodeT32Alu(InstDescriptor inst, ulong address, int opCode) : base(inst, address, opCode)
+        {
+            Rd = (opCode >> 8) & 0xf;
+            Rn = (opCode >> 16) & 0xf;
+
+            SetFlags = ((opCode >> 20) & 1) != 0;
+        }
+    }
+}

ARMeilleure/Decoders/OpCodeT32AluImm.cs

@@ -0,0 +1,38 @@
+using ARMeilleure.Common;
+using System.Runtime.Intrinsics;
+
+namespace ARMeilleure.Decoders
+{
+    class OpCodeT32AluImm : OpCodeT32Alu, IOpCode32AluImm
+    {
+        public int Immediate { get; }
+
+        public bool IsRotated { get; }
+
+        public new static OpCode Create(InstDescriptor inst, ulong address, int opCode) => new OpCodeT32AluImm(inst, address, opCode);
+
+        private static readonly Vector128<int> _factor = Vector128.Create(1, 0x00010001, 0x01000100, 0x01010101);
+
+        public OpCodeT32AluImm(InstDescriptor inst, ulong address, int opCode) : base(inst, address, opCode)
+        {
+            int imm8 = (opCode >> 0) & 0xff;
+            int imm3 = (opCode >> 12) & 7;
+            int imm1 = (opCode >> 26) & 1;
+
+            int imm12 = imm8 | (imm3 << 8) | (imm1 << 11);
+
+            if ((imm12 >> 10) == 0)
+            {
+                Immediate = imm8 * _factor.GetElement((imm12 >> 8) & 3);
+                IsRotated = false;
+            }
+            else
+            {
+                int shift = imm12 >> 7;
+
+                Immediate = BitUtils.RotateRight(0x80 | (imm12 & 0x7f), shift, 32);
+                IsRotated = shift != 0;
+            }
+        }
+    }
+}

ARMeilleure/Decoders/OpCodeT32AluRsImm.cs

@@ -0,0 +1,20 @@
+namespace ARMeilleure.Decoders
+{
+    class OpCodeT32AluRsImm : OpCodeT32Alu, IOpCode32AluRsImm
+    {
+        public int Rm        { get; }
+        public int Immediate { get; }
+
+        public ShiftType ShiftType { get; }
+
+        public new static OpCode Create(InstDescriptor inst, ulong address, int opCode) => new OpCodeT32AluRsImm(inst, address, opCode);
+
+        public OpCodeT32AluRsImm(InstDescriptor inst, ulong address, int opCode) : base(inst, address, opCode)
+        {
+            Rm        = (opCode >> 0) & 0xf;
+            Immediate = ((opCode >> 6) & 3) | ((opCode >> 10) & 0x1c);
+
+            ShiftType = (ShiftType)((opCode >> 4) & 3);
+        }
+    }
+}

ARMeilleure/Decoders/OpCodeT32BImm20.cs

@@ -0,0 +1,29 @@
+using ARMeilleure.Instructions;
+
+namespace ARMeilleure.Decoders
+{
+    class OpCodeT32BImm20 : OpCodeT32, IOpCode32BImm
+    {
+        public long Immediate { get; }
+
+        public new static OpCode Create(InstDescriptor inst, ulong address, int opCode) => new OpCodeT32BImm20(inst, address, opCode);
+
+        public OpCodeT32BImm20(InstDescriptor inst, ulong address, int opCode) : base(inst, address, opCode)
+        {
+            uint pc = GetPc();
+
+            int imm11 = (opCode >> 0) & 0x7ff;
+            int j2 = (opCode >> 11) & 1;
+            int j1 = (opCode >> 13) & 1;
+            int imm6 = (opCode >> 16) & 0x3f;
+            int s = (opCode >> 26) & 1;
+
+            int imm32 = imm11 | (imm6 << 11) | (j1 << 17) | (j2 << 18) | (s << 19);
+            imm32 = (imm32 << 13) >> 12;
+
+            Immediate = pc + imm32;
+
+            Cond = (Condition)((opCode >> 22) & 0xf);
+        }
+    }
+}

ARMeilleure/Decoders/OpCodeT32BImm24.cs

@@ -0,0 +1,35 @@
+using ARMeilleure.Instructions;
+
+namespace ARMeilleure.Decoders
+{
+    class OpCodeT32BImm24 : OpCodeT32, IOpCode32BImm
+    {
+        public long Immediate { get; }
+
+        public new static OpCode Create(InstDescriptor inst, ulong address, int opCode) => new OpCodeT32BImm24(inst, address, opCode);
+
+        public OpCodeT32BImm24(InstDescriptor inst, ulong address, int opCode) : base(inst, address, opCode)
+        {
+            uint pc = GetPc();
+
+            if (inst.Name == InstName.Blx)
+            {
+                pc &= ~3u;
+            }
+
+            int imm11 = (opCode >> 0) & 0x7ff;
+            int j2 = (opCode >> 11) & 1;
+            int j1 = (opCode >> 13) & 1;
+            int imm10 = (opCode >> 16) & 0x3ff;
+            int s = (opCode >> 26) & 1;
+
+            int i1 = j1 ^ s ^ 1;
+            int i2 = j2 ^ s ^ 1;
+
+            int imm32 = imm11 | (imm10 << 11) | (i2 << 21) | (i1 << 22) | (s << 23);
+            imm32 = (imm32 << 9) >> 8;
+
+            Immediate = pc + imm32;
+        }
+    }
+}

ARMeilleure/Decoders/OpCodeTable.cs

@@ -1,6 +1,7 @@
 using ARMeilleure.Instructions;
 using System;
 using System.Collections.Generic;
+using System.Numerics;
 
 namespace ARMeilleure.Decoders
 {
@@ -972,8 +973,7 @@ namespace ARMeilleure.Decoders
             SetA32("111100111x11<<10xxxx00011xx0xxxx", InstName.Vzip,     InstEmit32.Vzip,     OpCode32SimdCmpZ.Create);
 #endregion
 
-#region "OpCode Table (AArch32, T16/T32)"
-            // T16
+#region "OpCode Table (AArch32, T16)"
             SetT16("000<<xxxxxxxxxxx", InstName.Mov,    InstEmit32.Mov,     OpCodeT16ShiftImm.Create);
             SetT16("0001100xxxxxxxxx", InstName.Add,    InstEmit32.Add,     OpCodeT16AddSubReg.Create);
             SetT16("0001101xxxxxxxxx", InstName.Sub,    InstEmit32.Sub,     OpCodeT16AddSubReg.Create);
@@ -1045,6 +1045,46 @@ namespace ARMeilleure.Decoders
             SetT16("11100xxxxxxxxxxx", InstName.B,      InstEmit32.B,       OpCodeT16BImm11.Create);
 #endregion
 
+#region "OpCode Table (AArch32, T32)"
+            // Base
+            SetT32("11101011010xxxxx0xxxxxxxxxxxxxxx", InstName.Adc,      InstEmit32.Adc,      OpCodeT32AluRsImm.Create);
+            SetT32("11110x01010xxxxx0xxxxxxxxxxxxxxx", InstName.Adc,      InstEmit32.Adc,      OpCodeT32AluImm.Create);
+            SetT32("11101011000<xxxx0xxx<<<<xxxxxxxx", InstName.Add,      InstEmit32.Add,      OpCodeT32AluRsImm.Create);
+            SetT32("11110x01000<xxxx0xxx<<<<xxxxxxxx", InstName.Add,      InstEmit32.Add,      OpCodeT32AluImm.Create);
+            SetT32("11101010000<xxxx0xxx<<<<xxxxxxxx", InstName.And,      InstEmit32.And,      OpCodeT32AluRsImm.Create);
+            SetT32("11110x00000<xxxx0xxx<<<<xxxxxxxx", InstName.And,      InstEmit32.And,      OpCodeT32AluImm.Create);
+            SetT32("11110x<<<xxxxxxx10x0xxxxxxxxxxxx", InstName.B,        InstEmit32.B,        OpCodeT32BImm20.Create);
+            SetT32("11110xxxxxxxxxxx10x1xxxxxxxxxxxx", InstName.B,        InstEmit32.B,        OpCodeT32BImm24.Create);
+            SetT32("11101010001xxxxx0xxxxxxxxxxxxxxx", InstName.Bic,      InstEmit32.Bic,      OpCodeT32AluRsImm.Create);
+            SetT32("11110x00001xxxxx0xxxxxxxxxxxxxxx", InstName.Bic,      InstEmit32.Bic,      OpCodeT32AluImm.Create);
+            SetT32("11110xxxxxxxxxxx11x1xxxxxxxxxxxx", InstName.Bl,       InstEmit32.Bl,       OpCodeT32BImm24.Create);
+            SetT32("11110xxxxxxxxxxx11x0xxxxxxxxxxx0", InstName.Blx,      InstEmit32.Blx,      OpCodeT32BImm24.Create);
+            SetT32("111010110001xxxx0xxx1111xxxxxxxx", InstName.Cmn,      InstEmit32.Cmn,      OpCodeT32AluRsImm.Create);
+            SetT32("11110x010001xxxx0xxx1111xxxxxxxx", InstName.Cmn,      InstEmit32.Cmn,      OpCodeT32AluImm.Create);
+            SetT32("111010111011xxxx0xxx1111xxxxxxxx", InstName.Cmp,      InstEmit32.Cmp,      OpCodeT32AluRsImm.Create);
+            SetT32("11110x011011xxxx0xxx1111xxxxxxxx", InstName.Cmp,      InstEmit32.Cmp,      OpCodeT32AluImm.Create);
+            SetT32("11101010100<xxxx0xxx<<<<xxxxxxxx", InstName.Eor,      InstEmit32.Eor,      OpCodeT32AluRsImm.Create);
+            SetT32("11110x00100<xxxx0xxx<<<<xxxxxxxx", InstName.Eor,      InstEmit32.Eor,      OpCodeT32AluImm.Create);
+            SetT32("11101010010x11110xxxxxxxxxxxxxxx", InstName.Mov,      InstEmit32.Mov,      OpCodeT32AluRsImm.Create);
+            SetT32("11110x00010x11110xxxxxxxxxxxxxxx", InstName.Mov,      InstEmit32.Mov,      OpCodeT32AluImm.Create);
+            SetT32("11101010011x11110xxxxxxxxxxxxxxx", InstName.Mvn,      InstEmit32.Mvn,      OpCodeT32AluRsImm.Create);
+            SetT32("11110x00011x11110xxxxxxxxxxxxxxx", InstName.Mvn,      InstEmit32.Mvn,      OpCodeT32AluImm.Create);
+            SetT32("11101010011x<<<<0xxxxxxxxxxxxxxx", InstName.Orn,      InstEmit32.Orn,      OpCodeT32AluRsImm.Create);
+            SetT32("11110x00011x<<<<0xxxxxxxxxxxxxxx", InstName.Orn,      InstEmit32.Orn,      OpCodeT32AluImm.Create);
+            SetT32("11101010010x<<<<0xxxxxxxxxxxxxxx", InstName.Orr,      InstEmit32.Orr,      OpCodeT32AluRsImm.Create);
+            SetT32("11110x00010x<<<<0xxxxxxxxxxxxxxx", InstName.Orr,      InstEmit32.Orr,      OpCodeT32AluImm.Create);
+            SetT32("11101011110xxxxx0xxxxxxxxxxxxxxx", InstName.Rsb,      InstEmit32.Rsb,      OpCodeT32AluRsImm.Create);
+            SetT32("11110x01110xxxxx0xxxxxxxxxxxxxxx", InstName.Rsb,      InstEmit32.Rsb,      OpCodeT32AluImm.Create);
+            SetT32("11101011011xxxxx0xxxxxxxxxxxxxxx", InstName.Sbc,      InstEmit32.Sbc,      OpCodeT32AluRsImm.Create);
+            SetT32("11110x01011xxxxx0xxxxxxxxxxxxxxx", InstName.Sbc,      InstEmit32.Sbc,      OpCodeT32AluImm.Create);
+            SetT32("11101011101<xxxx0xxx<<<<xxxxxxxx", InstName.Sub,      InstEmit32.Sub,      OpCodeT32AluRsImm.Create);
+            SetT32("11110x01101<xxxx0xxx<<<<xxxxxxxx", InstName.Sub,      InstEmit32.Sub,      OpCodeT32AluImm.Create);
+            SetT32("111010101001xxxx0xxx1111xxxxxxxx", InstName.Teq,      InstEmit32.Teq,      OpCodeT32AluRsImm.Create);
+            SetT32("11110x001001xxxx0xxx1111xxxxxxxx", InstName.Teq,      InstEmit32.Teq,      OpCodeT32AluImm.Create);
+            SetT32("111010100001xxxx0xxx1111xxxxxxxx", InstName.Tst,      InstEmit32.Tst,      OpCodeT32AluRsImm.Create);
+            SetT32("11110x000001xxxx0xxx1111xxxxxxxx", InstName.Tst,      InstEmit32.Tst,      OpCodeT32AluImm.Create);
+#endregion
+
             FillFastLookupTable(InstA32FastLookup, AllInstA32, ToFastLookupIndexA);
             FillFastLookupTable(InstT32FastLookup, AllInstT32, ToFastLookupIndexT);
             FillFastLookupTable(InstA64FastLookup, AllInstA64, ToFastLookupIndexA);
@@ -1092,8 +1132,11 @@ namespace ARMeilleure.Decoders
 
         private static void SetT32(string encoding, InstName name, InstEmitter emitter, MakeOp makeOp)
         {
-            encoding = encoding.Substring(16) + encoding.Substring(0, 16);
-            Set(encoding, AllInstT32, new InstDescriptor(name, emitter), makeOp);
+            string reversedEncoding = encoding.Substring(16) + encoding.Substring(0, 16);
+            MakeOp reversedMakeOp =
+                (InstDescriptor inst, ulong address, int opCode)
+                    => makeOp(inst, address, (int)BitOperations.RotateRight((uint)opCode, 16));
+            Set(reversedEncoding, AllInstT32, new InstDescriptor(name, emitter), reversedMakeOp);
         }
 
         private static void SetA64(string encoding, InstName name, InstEmitter emitter, MakeOp makeOp)

ARMeilleure/Instructions/InstEmitAlu32.cs

@@ -244,6 +244,23 @@ namespace ARMeilleure.Instructions
             EmitAluStore(context, res);
         }
 
+        public static void Orn(ArmEmitterContext context)
+        {
+            IOpCode32Alu op = (IOpCode32Alu)context.CurrOp;
+
+            Operand n = GetAluN(context);
+            Operand m = GetAluM(context);
+
+            Operand res = context.BitwiseOr(n, context.BitwiseNot(m));
+
+            if (ShouldSetFlags(context))
+            {
+                EmitNZFlagsCheck(context, res);
+            }
+
+            EmitAluStore(context, res);
+        }
+
         public static void Pkh(ArmEmitterContext context)
         {
             OpCode32AluRsImm op = (OpCode32AluRsImm)context.CurrOp;

ARMeilleure/Instructions/InstEmitAluHelper.cs

@@ -128,7 +128,7 @@ namespace ARMeilleure.Instructions
         {
             Debug.Assert(value.Type == OperandType.I32);
 
-            if (IsThumb(context.CurrOp))
+            if (((OpCode32)context.CurrOp).IsThumb())
             {
                 bool isReturn = IsA32Return(context);
                 if (!isReturn)
@@ -197,7 +197,7 @@ namespace ARMeilleure.Instructions
                 // ARM32.
                 case IOpCode32AluImm op:
                 {
-                    if (ShouldSetFlags(context) && op.IsRotated)
+                    if (ShouldSetFlags(context) && op.IsRotated && setCarry)
                     {
                         SetFlag(context, PState.CFlag, Const((uint)op.Immediate >> 31));
                     }

ARMeilleure/Instructions/InstEmitException.cs

@@ -9,18 +9,25 @@ namespace ARMeilleure.Instructions
     {
         public static void Brk(ArmEmitterContext context)
         {
-            EmitExceptionCall(context, nameof(NativeInterface.Break));
+            OpCodeException op = (OpCodeException)context.CurrOp;
+
+            string name = nameof(NativeInterface.Break);
+
+            context.StoreToContext();
+
+            context.Call(typeof(NativeInterface).GetMethod(name), Const(op.Address), Const(op.Id));
+
+            context.LoadFromContext();
+
+            context.Return(Const(op.Address));
         }
 
         public static void Svc(ArmEmitterContext context)
-        {
-            EmitExceptionCall(context, nameof(NativeInterface.SupervisorCall));
-        }
-
-        private static void EmitExceptionCall(ArmEmitterContext context, string name)
         {
             OpCodeException op = (OpCodeException)context.CurrOp;
 
+            string name = nameof(NativeInterface.SupervisorCall);
+
             context.StoreToContext();
 
             context.Call(typeof(NativeInterface).GetMethod(name), Const(op.Address), Const(op.Id));
@@ -41,6 +48,8 @@ namespace ARMeilleure.Instructions
             context.Call(typeof(NativeInterface).GetMethod(name), Const(op.Address), Const(op.RawOpCode));
 
             context.LoadFromContext();
+
+            context.Return(Const(op.Address));
         }
     }
 }

ARMeilleure/Instructions/InstEmitException32.cs

@@ -9,19 +9,11 @@ namespace ARMeilleure.Instructions
     static partial class InstEmit32
     {
         public static void Svc(ArmEmitterContext context)
-        {
-            EmitExceptionCall(context, nameof(NativeInterface.SupervisorCall));
-        }
-
-        public static void Trap(ArmEmitterContext context)
-        {
-            EmitExceptionCall(context, nameof(NativeInterface.Break));
-        }
-
-        private static void EmitExceptionCall(ArmEmitterContext context, string name)
         {
             IOpCode32Exception op = (IOpCode32Exception)context.CurrOp;
 
+            string name = nameof(NativeInterface.SupervisorCall);
+
             context.StoreToContext();
 
             context.Call(typeof(NativeInterface).GetMethod(name), Const(((IOpCode)op).Address), Const(op.Id));
@@ -30,5 +22,20 @@ namespace ARMeilleure.Instructions
 
             Translator.EmitSynchronization(context);
         }
+
+        public static void Trap(ArmEmitterContext context)
+        {
+            IOpCode32Exception op = (IOpCode32Exception)context.CurrOp;
+
+            string name = nameof(NativeInterface.Break);
+
+            context.StoreToContext();
+
+            context.Call(typeof(NativeInterface).GetMethod(name), Const(((IOpCode)op).Address), Const(op.Id));
+
+            context.LoadFromContext();
+
+            context.Return(Const(context.CurrOp.Address));
+        }
     }
 }

ARMeilleure/Instructions/InstEmitFlow32.cs

@@ -34,7 +34,7 @@ namespace ARMeilleure.Instructions
 
             uint pc = op.GetPc();
 
-            bool isThumb = IsThumb(context.CurrOp);
+            bool isThumb = ((OpCode32)context.CurrOp).IsThumb();
 
             uint currentPc = isThumb
                 ? pc | 1
@@ -61,7 +61,7 @@ namespace ARMeilleure.Instructions
             Operand addr = context.Copy(GetIntA32(context, op.Rm));
             Operand bitOne = context.BitwiseAnd(addr, Const(1));
 
-            bool isThumb = IsThumb(context.CurrOp);
+            bool isThumb = ((OpCode32)context.CurrOp).IsThumb();
 
             uint currentPc = isThumb
                 ? (pc - 2) | 1
@@ -71,7 +71,7 @@ namespace ARMeilleure.Instructions
 
             SetFlag(context, PState.TFlag, bitOne);
 
-            EmitVirtualCall(context, addr);
+            EmitBxWritePc(context, addr);
         }
 
         public static void Bx(ArmEmitterContext context)

ARMeilleure/Instructions/InstEmitHelper.cs

@@ -10,11 +10,6 @@ namespace ARMeilleure.Instructions
 {
     static class InstEmitHelper
     {
-        public static bool IsThumb(OpCode op)
-        {
-            return op is OpCodeT16;
-        }
-
         public static Operand GetExtendedM(ArmEmitterContext context, int rm, IntType type)
         {
             Operand value = GetIntOrZR(context, rm);
@@ -186,7 +181,7 @@ namespace ARMeilleure.Instructions
 
             SetFlag(context, PState.TFlag, mode);
 
-            Operand addr = context.ConditionalSelect(mode, pc, context.BitwiseAnd(pc, Const(~3)));
+            Operand addr = context.ConditionalSelect(mode, context.BitwiseAnd(pc, Const(~1)), context.BitwiseAnd(pc, Const(~3)));
 
             InstEmitFlowHelper.EmitVirtualJump(context, addr, isReturn);
         }

ARMeilleure/Instructions/InstEmitMemoryEx.cs

@@ -130,11 +130,6 @@ namespace ARMeilleure.Instructions
             bool ordered   = (accType & AccessType.Ordered)   != 0;
             bool exclusive = (accType & AccessType.Exclusive) != 0;
 
-            if (ordered)
-            {
-                EmitBarrier(context);
-            }
-
             Operand address = context.Copy(GetIntOrSP(context, op.Rn));
 
             Operand t = GetIntOrZR(context, op.Rt);
@@ -163,6 +158,11 @@ namespace ARMeilleure.Instructions
             {
                 EmitStoreExclusive(context, address, t, exclusive, op.Size, op.Rs, a32: false);
             }
+
+            if (ordered)
+            {
+                EmitBarrier(context);
+            }
         }
 
         private static void EmitBarrier(ArmEmitterContext context)

ARMeilleure/Instructions/InstEmitMemoryEx32.cs

@@ -146,13 +146,13 @@ namespace ARMeilleure.Instructions
             var exclusive = (accType & AccessType.Exclusive) != 0;
             var ordered = (accType & AccessType.Ordered) != 0;
 
-            if (ordered)
-            {
-                EmitBarrier(context);
-            }
-
             if ((accType & AccessType.Load) != 0)
             {
+                if (ordered)
+                {
+                    EmitBarrier(context);
+                }
+
                 if (size == DWordSizeLog2)
                 {
                     // Keep loads atomic - make the call to get the whole region and then decompose it into parts
@@ -219,6 +219,11 @@ namespace ARMeilleure.Instructions
                     Operand value = context.ZeroExtend32(OperandType.I64, GetIntA32(context, op.Rt));
                     EmitStoreExclusive(context, address, value, exclusive, size, op.Rd, a32: true);
                 }
+
+                if (ordered)
+                {
+                    EmitBarrier(context);
+                }
             }
         }
 

ARMeilleure/Instructions/InstEmitSystem.cs

@@ -12,7 +12,8 @@ namespace ARMeilleure.Instructions
 {
     static partial class InstEmit
     {
-        private const int DczSizeLog2 = 4;
+        private const int DczSizeLog2 = 4; // Log2 size in words
+        public const int DczSizeInBytes = 4 << DczSizeLog2;
 
         public static void Hint(ArmEmitterContext context)
         {
@@ -87,7 +88,7 @@ namespace ARMeilleure.Instructions
                     // DC ZVA
                     Operand t = GetIntOrZR(context, op.Rt);
 
-                    for (long offset = 0; offset < (4 << DczSizeLog2); offset += 8)
+                    for (long offset = 0; offset < DczSizeInBytes; offset += 8)
                     {
                         Operand address = context.Add(t, Const(offset));
 
@@ -98,7 +99,12 @@ namespace ARMeilleure.Instructions
                 }
 
                 // No-op
-                case 0b11_011_0111_1110_001: //DC CIVAC
+                case 0b11_011_0111_1110_001: // DC CIVAC
+                    break;
+
+                case 0b11_011_0111_0101_001: // IC IVAU
+                    Operand target = Register(op.Rt, RegisterType.Integer, OperandType.I64);
+                    context.Call(typeof(NativeInterface).GetMethod(nameof(NativeInterface.InvalidateCacheLine)), target);
                     break;
             }
         }

ARMeilleure/Instructions/NativeInterface.cs

@@ -242,6 +242,11 @@ namespace ARMeilleure.Instructions
             return (ulong)function.FuncPtr.ToInt64();
         }
 
+        public static void InvalidateCacheLine(ulong address)
+        {
+            Context.Translator.InvalidateJitCacheRegion(address, InstEmit.DczSizeInBytes);
+        }
+
         public static bool CheckSynchronization()
         {
             Statistics.PauseTimer();

ARMeilleure/State/ExecutionContext.cs

@@ -43,6 +43,12 @@ namespace ARMeilleure.State
         public long TpidrEl0 { get; set; }
         public long Tpidr    { get; set; }
 
+        public uint Pstate
+        {
+            get => _nativeContext.GetPstate();
+            set => _nativeContext.SetPstate(value);
+        }
+
         public FPCR Fpcr { get; set; }
         public FPSR Fpsr { get; set; }
         public FPCR StandardFpcrValue => (Fpcr & (FPCR.Ahp)) | FPCR.Dn | FPCR.Fz;

ARMeilleure/State/NativeContext.cs

@@ -95,6 +95,25 @@ namespace ARMeilleure.State
             GetStorage().Flags[(int)flag] = value ? 1u : 0u;
         }
 
+        public unsafe uint GetPstate()
+        {
+            uint value = 0;
+            for (int flag = 0; flag < RegisterConsts.FlagsCount; flag++)
+            {
+                value |= GetStorage().Flags[flag] != 0 ? 1u << flag : 0u;
+            }
+            return value;
+        }
+
+        public unsafe void SetPstate(uint value)
+        {
+            for (int flag = 0; flag < RegisterConsts.FlagsCount; flag++)
+            {
+                uint bit = 1u << flag;
+                GetStorage().Flags[flag] = (value & bit) == bit ? 1u : 0u;
+            }
+        }
+
         public unsafe bool GetFPStateFlag(FPState flag)
         {
             if ((uint)flag >= RegisterConsts.FpFlagsCount)

ARMeilleure/Translation/Delegates.cs

@@ -114,6 +114,7 @@ namespace ARMeilleure.Translation
             SetDelegateInfo(typeof(NativeInterface).GetMethod(nameof(NativeInterface.GetFpscr))); // A32 only.
             SetDelegateInfo(typeof(NativeInterface).GetMethod(nameof(NativeInterface.GetFpsr)));
             SetDelegateInfo(typeof(NativeInterface).GetMethod(nameof(NativeInterface.GetFunctionAddress)));
+            SetDelegateInfo(typeof(NativeInterface).GetMethod(nameof(NativeInterface.InvalidateCacheLine)));
             SetDelegateInfo(typeof(NativeInterface).GetMethod(nameof(NativeInterface.GetTpidr)));
             SetDelegateInfo(typeof(NativeInterface).GetMethod(nameof(NativeInterface.GetTpidr32))); // A32 only.
             SetDelegateInfo(typeof(NativeInterface).GetMethod(nameof(NativeInterface.GetTpidrEl0)));

ARMeilleure/Translation/IntervalTree.cs

@@ -0,0 +1,756 @@
+using System;
+using System.Collections.Generic;
+
+namespace ARMeilleure.Translation
+{
+    /// <summary>
+    /// An Augmented Interval Tree based off of the "TreeDictionary"'s Red-Black Tree. Allows fast overlap checking of ranges.
+    /// </summary>
+    /// <typeparam name="K">Key</typeparam>
+    /// <typeparam name="V">Value</typeparam>
+    public class IntervalTree<K, V> where K : IComparable<K>
+    {
+        private const int ArrayGrowthSize = 32;
+
+        private const bool Black = true;
+        private const bool Red = false;
+        private IntervalTreeNode<K, V> _root = null;
+        private int _count = 0;
+
+        public int Count => _count;
+
+        public IntervalTree() { }
+
+        #region Public Methods
+
+        /// <summary>
+        /// Gets the values of the interval whose key is <paramref name="key"/>.
+        /// </summary>
+        /// <param name="key">Key of the node value to get</param>
+        /// <param name="value">Value with the given <paramref name="key"/></param>
+        /// <returns>True if the key is on the dictionary, false otherwise</returns>
+        public bool TryGet(K key, out V value)
+        {
+            IntervalTreeNode<K, V> node = GetNode(key);
+
+            if (node == null)
+            {
+                value = default;
+                return false;
+            }
+
+            value = node.Value;
+            return true;
+        }
+
+        /// <summary>
+        /// Returns the start addresses of the intervals whose start and end keys overlap the given range.
+        /// </summary>
+        /// <param name="start">Start of the range</param>
+        /// <param name="end">End of the range</param>
+        /// <param name="overlaps">Overlaps array to place results in</param>
+        /// <param name="overlapCount">Index to start writing results into the array. Defaults to 0</param>
+        /// <returns>Number of intervals found</returns>
+        public int Get(K start, K end, ref K[] overlaps, int overlapCount = 0)
+        {
+            GetValues(_root, start, end, ref overlaps, ref overlapCount);
+
+            return overlapCount;
+        }
+
+        /// <summary>
+        /// Adds a new interval into the tree whose start is <paramref name="start"/>, end is <paramref name="end"/> and value is <paramref name="value"/>.
+        /// </summary>
+        /// <param name="start">Start of the range to add</param>
+        /// <param name="end">End of the range to insert</param>
+        /// <param name="value">Value to add</param>
+        /// <param name="updateFactoryCallback">Optional factory used to create a new value if <paramref name="start"/> is already on the tree</param>
+        /// <exception cref="ArgumentNullException"><paramref name="value"/> is null</exception>
+        /// <returns>True if the value was added, false if the start key was already in the dictionary</returns>
+        public bool AddOrUpdate(K start, K end, V value, Func<K, V, V> updateFactoryCallback)
+        {
+            if (value == null)
+            {
+                throw new ArgumentNullException(nameof(value));
+            }
+
+            return BSTInsert(start, end, value, updateFactoryCallback, out IntervalTreeNode<K, V> node);
+        }
+
+        /// <summary>
+        /// Gets an existing or adds a new interval into the tree whose start is <paramref name="start"/>, end is <paramref name="end"/> and value is <paramref name="value"/>.
+        /// </summary>
+        /// <param name="start">Start of the range to add</param>
+        /// <param name="end">End of the range to insert</param>
+        /// <param name="value">Value to add</param>
+        /// <exception cref="ArgumentNullException"><paramref name="value"/> is null</exception>
+        /// <returns><paramref name="value"/> if <paramref name="start"/> is not yet on the tree, or the existing value otherwise</returns>
+        public V GetOrAdd(K start, K end, V value)
+        {
+            if (value == null)
+            {
+                throw new ArgumentNullException(nameof(value));
+            }
+
+            BSTInsert(start, end, value, null, out IntervalTreeNode<K, V> node);
+            return node.Value;
+        }
+
+        /// <summary>
+        /// Removes a value from the tree, searching for it with <paramref name="key"/>.
+        /// </summary>
+        /// <param name="key">Key of the node to remove</param>
+        /// <returns>Number of deleted values</returns>
+        public int Remove(K key)
+        {
+            int removed = Delete(key);
+
+            _count -= removed;
+
+            return removed;
+        }
+
+        /// <summary>
+        /// Adds all the nodes in the dictionary into <paramref name="list"/>.
+        /// </summary>
+        /// <returns>A list of all values sorted by Key Order</returns>
+        public List<V> AsList()
+        {
+            List<V> list = new List<V>();
+
+            AddToList(_root, list);
+
+            return list;
+        }
+
+        #endregion
+
+        #region Private Methods (BST)
+
+        /// <summary>
+        /// Adds all values that are children of or contained within <paramref name="node"/> into <paramref name="list"/>, in Key Order.
+        /// </summary>
+        /// <param name="node">The node to search for values within</param>
+        /// <param name="list">The list to add values to</param>
+        private void AddToList(IntervalTreeNode<K, V> node, List<V> list)
+        {
+            if (node == null)
+            {
+                return;
+            }
+
+            AddToList(node.Left, list);
+
+            list.Add(node.Value);
+
+            AddToList(node.Right, list);
+        }
+
+        /// <summary>
+        /// Retrieve the node reference whose key is <paramref name="key"/>, or null if no such node exists.
+        /// </summary>
+        /// <param name="key">Key of the node to get</param>
+        /// <exception cref="ArgumentNullException"><paramref name="key"/> is null</exception>
+        /// <returns>Node reference in the tree</returns>
+        private IntervalTreeNode<K, V> GetNode(K key)
+        {
+            if (key == null)
+            {
+                throw new ArgumentNullException(nameof(key));
+            }
+
+            IntervalTreeNode<K, V> node = _root;
+            while (node != null)
+            {
+                int cmp = key.CompareTo(node.Start);
+                if (cmp < 0)
+                {
+                    node = node.Left;
+                }
+                else if (cmp > 0)
+                {
+                    node = node.Right;
+                }
+                else
+                {
+                    return node;
+                }
+            }
+            return null;
+        }
+
+        /// <summary>
+        /// Retrieve all values that overlap the given start and end keys.
+        /// </summary>
+        /// <param name="start">Start of the range</param>
+        /// <param name="end">End of the range</param>
+        /// <param name="overlaps">Overlaps array to place results in</param>
+        /// <param name="overlapCount">Overlaps count to update</param>
+        private void GetValues(IntervalTreeNode<K, V> node, K start, K end, ref K[] overlaps, ref int overlapCount)
+        {
+            if (node == null || start.CompareTo(node.Max) >= 0)
+            {
+                return;
+            }
+
+            GetValues(node.Left, start, end, ref overlaps, ref overlapCount);
+
+            bool endsOnRight = end.CompareTo(node.Start) > 0;
+            if (endsOnRight)
+            {
+                if (start.CompareTo(node.End) < 0)
+                {
+                    if (overlaps.Length >= overlapCount)
+                    {
+                        Array.Resize(ref overlaps, overlapCount + ArrayGrowthSize);
+                    }
+
+                    overlaps[overlapCount++] = node.Start;
+                }
+
+                GetValues(node.Right, start, end, ref overlaps, ref overlapCount);
+            }
+        }
+
+        /// <summary>
+        /// Propagate an increase in max value starting at the given node, heading up the tree.
+        /// This should only be called if the max increases - not for rebalancing or removals.
+        /// </summary>
+        /// <param name="node">The node to start propagating from</param>
+        private void PropagateIncrease(IntervalTreeNode<K, V> node)
+        {
+            K max = node.Max;
+            IntervalTreeNode<K, V> ptr = node;
+
+            while ((ptr = ptr.Parent) != null)
+            {
+                if (max.CompareTo(ptr.Max) > 0)
+                {
+                    ptr.Max = max;
+                }
+                else
+                {
+                    break;
+                }
+            }
+        }
+
+        /// <summary>
+        /// Propagate recalculating max value starting at the given node, heading up the tree.
+        /// This fully recalculates the max value from all children when there is potential for it to decrease.
+        /// </summary>
+        /// <param name="node">The node to start propagating from</param>
+        private void PropagateFull(IntervalTreeNode<K, V> node)
+        {
+            IntervalTreeNode<K, V> ptr = node;
+
+            do
+            {
+                K max = ptr.End;
+
+                if (ptr.Left != null && ptr.Left.Max.CompareTo(max) > 0)
+                {
+                    max = ptr.Left.Max;
+                }
+
+                if (ptr.Right != null && ptr.Right.Max.CompareTo(max) > 0)
+                {
+                    max = ptr.Right.Max;
+                }
+
+                ptr.Max = max;
+            } while ((ptr = ptr.Parent) != null);
+        }
+
+        /// <summary>
+        /// Insertion Mechanism for the interval tree. Similar to a BST insert, with the start of the range as the key.
+        /// Iterates the tree starting from the root and inserts a new node where all children in the left subtree are less than <paramref name="start"/>, and all children in the right subtree are greater than <paramref name="start"/>.
+        /// Each node can contain multiple values, and has an end address which is the maximum of all those values.
+        /// Post insertion, the "max" value of the node and all parents are updated.
+        /// </summary>
+        /// <param name="start">Start of the range to insert</param>
+        /// <param name="end">End of the range to insert</param>
+        /// <param name="value">Value to insert</param>
+        /// <param name="updateFactoryCallback">Optional factory used to create a new value if <paramref name="start"/> is already on the tree</param>
+        /// <param name="outNode">Node that was inserted or modified</param>
+        /// <returns>True if <paramref name="start"/> was not yet on the tree, false otherwise</returns>
+        private bool BSTInsert(K start, K end, V value, Func<K, V, V> updateFactoryCallback, out IntervalTreeNode<K, V> outNode)
+        {
+            IntervalTreeNode<K, V> parent = null;
+            IntervalTreeNode<K, V> node = _root;
+
+            while (node != null)
+            {
+                parent = node;
+                int cmp = start.CompareTo(node.Start);
+                if (cmp < 0)
+                {
+                    node = node.Left;
+                }
+                else if (cmp > 0)
+                {
+                    node = node.Right;
+                }
+                else
+                {
+                    outNode = node;
+
+                    if (updateFactoryCallback != null)
+                    {
+                        // Replace
+                        node.Value = updateFactoryCallback(start, node.Value);
+
+                        int endCmp = end.CompareTo(node.End);
+
+                        if (endCmp > 0)
+                        {
+                            node.End = end;
+                            if (end.CompareTo(node.Max) > 0)
+                            {
+                                node.Max = end;
+                                PropagateIncrease(node);
+                                RestoreBalanceAfterInsertion(node);
+                            }
+                        }
+                        else if (endCmp < 0)
+                        {
+                            node.End = end;
+                            PropagateFull(node);
+                        }
+                    }
+
+                    return false;
+                }
+            }
+            IntervalTreeNode<K, V> newNode = new IntervalTreeNode<K, V>(start, end, value, parent);
+            if (newNode.Parent == null)
+            {
+                _root = newNode;
+            }
+            else if (start.CompareTo(parent.Start) < 0)
+            {
+                parent.Left = newNode;
+            }
+            else
+            {
+                parent.Right = newNode;
+            }
+
+            PropagateIncrease(newNode);
+            _count++;
+            RestoreBalanceAfterInsertion(newNode);
+            outNode = newNode;
+            return true;
+        }
+
+        /// <summary>
+        /// Removes the value from the dictionary after searching for it with <paramref name="key">.
+        /// </summary>
+        /// <param name="key">Key to search for</param>
+        /// <returns>Number of deleted values</returns>
+        private int Delete(K key)
+        {
+            IntervalTreeNode<K, V> nodeToDelete = GetNode(key);
+
+            if (nodeToDelete == null)
+            {
+                return 0;
+            }
+
+            IntervalTreeNode<K, V> replacementNode;
+
+            if (LeftOf(nodeToDelete) == null || RightOf(nodeToDelete) == null)
+            {
+                replacementNode = nodeToDelete;
+            }
+            else
+            {
+                replacementNode = PredecessorOf(nodeToDelete);
+            }
+
+            IntervalTreeNode<K, V> tmp = LeftOf(replacementNode) ?? RightOf(replacementNode);
+
+            if (tmp != null)
+            {
+                tmp.Parent = ParentOf(replacementNode);
+            }
+
+            if (ParentOf(replacementNode) == null)
+            {
+                _root = tmp;
+            }
+            else if (replacementNode == LeftOf(ParentOf(replacementNode)))
+            {
+                ParentOf(replacementNode).Left = tmp;
+            }
+            else
+            {
+                ParentOf(replacementNode).Right = tmp;
+            }
+
+            if (replacementNode != nodeToDelete)
+            {
+                nodeToDelete.Start = replacementNode.Start;
+                nodeToDelete.Value = replacementNode.Value;
+                nodeToDelete.End = replacementNode.End;
+                nodeToDelete.Max = replacementNode.Max;
+            }
+
+            PropagateFull(replacementNode);
+
+            if (tmp != null && ColorOf(replacementNode) == Black)
+            {
+                RestoreBalanceAfterRemoval(tmp);
+            }
+
+            return 1;
+        }
+
+        /// <summary>
+        /// Returns the node with the largest key where <paramref name="node"/> is considered the root node.
+        /// </summary>
+        /// <param name="node">Root Node</param>
+        /// <returns>Node with the maximum key in the tree of <paramref name="node"/></returns>
+        private static IntervalTreeNode<K, V> Maximum(IntervalTreeNode<K, V> node)
+        {
+            IntervalTreeNode<K, V> tmp = node;
+            while (tmp.Right != null)
+            {
+                tmp = tmp.Right;
+            }
+
+            return tmp;
+        }
+
+        /// <summary>
+        /// Finds the node whose key is immediately less than <paramref name="node"/>.
+        /// </summary>
+        /// <param name="node">Node to find the predecessor of</param>
+        /// <returns>Predecessor of <paramref name="node"/></returns>
+        private static IntervalTreeNode<K, V> PredecessorOf(IntervalTreeNode<K, V> node)
+        {
+            if (node.Left != null)
+            {
+                return Maximum(node.Left);
+            }
+            IntervalTreeNode<K, V> parent = node.Parent;
+            while (parent != null && node == parent.Left)
+            {
+                node = parent;
+                parent = parent.Parent;
+            }
+            return parent;
+        }
+
+        #endregion
+
+        #region Private Methods (RBL)
+
+        private void RestoreBalanceAfterRemoval(IntervalTreeNode<K, V> balanceNode)
+        {
+            IntervalTreeNode<K, V> ptr = balanceNode;
+
+            while (ptr != _root && ColorOf(ptr) == Black)
+            {
+                if (ptr == LeftOf(ParentOf(ptr)))
+                {
+                    IntervalTreeNode<K, V> sibling = RightOf(ParentOf(ptr));
+
+                    if (ColorOf(sibling) == Red)
+                    {
+                        SetColor(sibling, Black);
+                        SetColor(ParentOf(ptr), Red);
+                        RotateLeft(ParentOf(ptr));
+                        sibling = RightOf(ParentOf(ptr));
+                    }
+                    if (ColorOf(LeftOf(sibling)) == Black && ColorOf(RightOf(sibling)) == Black)
+                    {
+                        SetColor(sibling, Red);
+                        ptr = ParentOf(ptr);
+                    }
+                    else
+                    {
+                        if (ColorOf(RightOf(sibling)) == Black)
+                        {
+                            SetColor(LeftOf(sibling), Black);
+                            SetColor(sibling, Red);
+                            RotateRight(sibling);
+                            sibling = RightOf(ParentOf(ptr));
+                        }
+                        SetColor(sibling, ColorOf(ParentOf(ptr)));
+                        SetColor(ParentOf(ptr), Black);
+                        SetColor(RightOf(sibling), Black);
+                        RotateLeft(ParentOf(ptr));
+                        ptr = _root;
+                    }
+                }
+                else
+                {
+                    IntervalTreeNode<K, V> sibling = LeftOf(ParentOf(ptr));
+
+                    if (ColorOf(sibling) == Red)
+                    {
+                        SetColor(sibling, Black);
+                        SetColor(ParentOf(ptr), Red);
+                        RotateRight(ParentOf(ptr));
+                        sibling = LeftOf(ParentOf(ptr));
+                    }
+                    if (ColorOf(RightOf(sibling)) == Black && ColorOf(LeftOf(sibling)) == Black)
+                    {
+                        SetColor(sibling, Red);
+                        ptr = ParentOf(ptr);
+                    }
+                    else
+                    {
+                        if (ColorOf(LeftOf(sibling)) == Black)
+                        {
+                            SetColor(RightOf(sibling), Black);
+                            SetColor(sibling, Red);
+                            RotateLeft(sibling);
+                            sibling = LeftOf(ParentOf(ptr));
+                        }
+                        SetColor(sibling, ColorOf(ParentOf(ptr)));
+                        SetColor(ParentOf(ptr), Black);
+                        SetColor(LeftOf(sibling), Black);
+                        RotateRight(ParentOf(ptr));
+                        ptr = _root;
+                    }
+                }
+            }
+            SetColor(ptr, Black);
+        }
+
+        private void RestoreBalanceAfterInsertion(IntervalTreeNode<K, V> balanceNode)
+        {
+            SetColor(balanceNode, Red);
+            while (balanceNode != null && balanceNode != _root && ColorOf(ParentOf(balanceNode)) == Red)
+            {
+                if (ParentOf(balanceNode) == LeftOf(ParentOf(ParentOf(balanceNode))))
+                {
+                    IntervalTreeNode<K, V> sibling = RightOf(ParentOf(ParentOf(balanceNode)));
+
+                    if (ColorOf(sibling) == Red)
+                    {
+                        SetColor(ParentOf(balanceNode), Black);
+                        SetColor(sibling, Black);
+                        SetColor(ParentOf(ParentOf(balanceNode)), Red);
+                        balanceNode = ParentOf(ParentOf(balanceNode));
+                    }
+                    else
+                    {
+                        if (balanceNode == RightOf(ParentOf(balanceNode)))
+                        {
+                            balanceNode = ParentOf(balanceNode);
+                            RotateLeft(balanceNode);
+                        }
+                        SetColor(ParentOf(balanceNode), Black);
+                        SetColor(ParentOf(ParentOf(balanceNode)), Red);
+                        RotateRight(ParentOf(ParentOf(balanceNode)));
+                    }
+                }
+                else
+                {
+                    IntervalTreeNode<K, V> sibling = LeftOf(ParentOf(ParentOf(balanceNode)));
+
+                    if (ColorOf(sibling) == Red)
+                    {
+                        SetColor(ParentOf(balanceNode), Black);
+                        SetColor(sibling, Black);
+                        SetColor(ParentOf(ParentOf(balanceNode)), Red);
+                        balanceNode = ParentOf(ParentOf(balanceNode));
+                    }
+                    else
+                    {
+                        if (balanceNode == LeftOf(ParentOf(balanceNode)))
+                        {
+                            balanceNode = ParentOf(balanceNode);
+                            RotateRight(balanceNode);
+                        }
+                        SetColor(ParentOf(balanceNode), Black);
+                        SetColor(ParentOf(ParentOf(balanceNode)), Red);
+                        RotateLeft(ParentOf(ParentOf(balanceNode)));
+                    }
+                }
+            }
+            SetColor(_root, Black);
+        }
+
+        private void RotateLeft(IntervalTreeNode<K, V> node)
+        {
+            if (node != null)
+            {
+                IntervalTreeNode<K, V> right = RightOf(node);
+                node.Right = LeftOf(right);
+                if (node.Right != null)
+                {
+                    node.Right.Parent = node;
+                }
+                IntervalTreeNode<K, V> nodeParent = ParentOf(node);
+                right.Parent = nodeParent;
+                if (nodeParent == null)
+                {
+                    _root = right;
+                }
+                else if (node == LeftOf(nodeParent))
+                {
+                    nodeParent.Left = right;
+                }
+                else
+                {
+                    nodeParent.Right = right;
+                }
+                right.Left = node;
+                node.Parent = right;
+
+                PropagateFull(node);
+            }
+        }
+
+        private void RotateRight(IntervalTreeNode<K, V> node)
+        {
+            if (node != null)
+            {
+                IntervalTreeNode<K, V> left = LeftOf(node);
+                node.Left = RightOf(left);
+                if (node.Left != null)
+                {
+                    node.Left.Parent = node;
+                }
+                IntervalTreeNode<K, V> nodeParent = ParentOf(node);
+                left.Parent = nodeParent;
+                if (nodeParent == null)
+                {
+                    _root = left;
+                }
+                else if (node == RightOf(nodeParent))
+                {
+                    nodeParent.Right = left;
+                }
+                else
+                {
+                    nodeParent.Left = left;
+                }
+                left.Right = node;
+                node.Parent = left;
+
+                PropagateFull(node);
+            }
+        }
+
+        #endregion
+
+        #region Safety-Methods
+
+        // These methods save memory by allowing us to forego sentinel nil nodes, as well as serve as protection against NullReferenceExceptions.
+
+        /// <summary>
+        /// Returns the color of <paramref name="node"/>, or Black if it is null.
+        /// </summary>
+        /// <param name="node">Node</param>
+        /// <returns>The boolean color of <paramref name="node"/>, or black if null</returns>
+        private static bool ColorOf(IntervalTreeNode<K, V> node)
+        {
+            return node == null || node.Color;
+        }
+
+        /// <summary>
+        /// Sets the color of <paramref name="node"/> node to <paramref name="color"/>.
+        /// <br></br>
+        /// This method does nothing if <paramref name="node"/> is null.
+        /// </summary>
+        /// <param name="node">Node to set the color of</param>
+        /// <param name="color">Color (Boolean)</param>
+        private static void SetColor(IntervalTreeNode<K, V> node, bool color)
+        {
+            if (node != null)
+            {
+                node.Color = color;
+            }
+        }
+
+        /// <summary>
+        /// This method returns the left node of <paramref name="node"/>, or null if <paramref name="node"/> is null.
+        /// </summary>
+        /// <param name="node">Node to retrieve the left child from</param>
+        /// <returns>Left child of <paramref name="node"/></returns>
+        private static IntervalTreeNode<K, V> LeftOf(IntervalTreeNode<K, V> node)
+        {
+            return node?.Left;
+        }
+
+        /// <summary>
+        /// This method returns the right node of <paramref name="node"/>, or null if <paramref name="node"/> is null.
+        /// </summary>
+        /// <param name="node">Node to retrieve the right child from</param>
+        /// <returns>Right child of <paramref name="node"/></returns>
+        private static IntervalTreeNode<K, V> RightOf(IntervalTreeNode<K, V> node)
+        {
+            return node?.Right;
+        }
+
+        /// <summary>
+        /// Returns the parent node of <paramref name="node"/>, or null if <paramref name="node"/> is null.
+        /// </summary>
+        /// <param name="node">Node to retrieve the parent from</param>
+        /// <returns>Parent of <paramref name="node"/></returns>
+        private static IntervalTreeNode<K, V> ParentOf(IntervalTreeNode<K, V> node)
+        {
+            return node?.Parent;
+        }
+
+        #endregion
+
+        public bool ContainsKey(K key)
+        {
+            return GetNode(key) != null;
+        }
+
+        public void Clear()
+        {
+            _root = null;
+            _count = 0;
+        }
+    }
+
+    /// <summary>
+    /// Represents a node in the IntervalTree which contains start and end keys of type K, and a value of generic type V.
+    /// </summary>
+    /// <typeparam name="K">Key type of the node</typeparam>
+    /// <typeparam name="V">Value type of the node</typeparam>
+    internal class IntervalTreeNode<K, V>
+    {
+        internal bool Color = true;
+        internal IntervalTreeNode<K, V> Left = null;
+        internal IntervalTreeNode<K, V> Right = null;
+        internal IntervalTreeNode<K, V> Parent = null;
+
+        /// <summary>
+        /// The start of the range.
+        /// </summary>
+        internal K Start;
+
+        /// <summary>
+        /// The end of the range.
+        /// </summary>
+        internal K End;
+
+        /// <summary>
+        /// The maximum end value of this node and all its children.
+        /// </summary>
+        internal K Max;
+
+        /// <summary>
+        /// Value stored on this node.
+        /// </summary>
+        internal V Value;
+
+        public IntervalTreeNode(K start, K end, V value, IntervalTreeNode<K, V> parent)
+        {
+            this.Start = start;
+            this.End = end;
+            this.Max = end;
+            this.Value = value;
+            this.Parent = parent;
+        }
+    }
+}

ARMeilleure/Translation/PTC/Ptc.cs

@@ -27,7 +27,7 @@ namespace ARMeilleure.Translation.PTC
         private const string OuterHeaderMagicString = "PTCohd\0\0";
         private const string InnerHeaderMagicString = "PTCihd\0\0";
 
-        private const uint InternalVersion = 3061; //! To be incremented manually for each change to the ARMeilleure project.
+        private const uint InternalVersion = 3193; //! To be incremented manually for each change to the ARMeilleure project.
 
         private const string ActualDir = "0";
         private const string BackupDir = "1";
@@ -585,7 +585,7 @@ namespace ARMeilleure.Translation.PTC
 
                     translator.RegisterFunction(infoEntry.Address, func);
 
-                    bool isAddressUnique = translator.Functions.TryAdd(infoEntry.Address, func);
+                    bool isAddressUnique = translator.Functions.TryAdd(infoEntry.Address, infoEntry.GuestSize, func);
 
                     Debug.Assert(isAddressUnique, $"The address 0x{infoEntry.Address:X16} is not unique.");
                 }
@@ -815,7 +815,7 @@ namespace ARMeilleure.Translation.PTC
 
                     TranslatedFunction func = translator.Translate(address, item.funcProfile.Mode, item.funcProfile.HighCq);
 
-                    bool isAddressUnique = translator.Functions.TryAdd(address, func);
+                    bool isAddressUnique = translator.Functions.TryAdd(address, func.GuestSize, func);
 
                     Debug.Assert(isAddressUnique, $"The address 0x{address:X16} is not unique.");
 

ARMeilleure/Translation/PTC/PtcProfiler.cs

@@ -96,7 +96,7 @@ namespace ARMeilleure.Translation.PTC
             return address >= StaticCodeStart && address < StaticCodeStart + StaticCodeSize;
         }
 
-        internal static ConcurrentQueue<(ulong address, FuncProfile funcProfile)> GetProfiledFuncsToTranslate(ConcurrentDictionary<ulong, TranslatedFunction> funcs)
+        internal static ConcurrentQueue<(ulong address, FuncProfile funcProfile)> GetProfiledFuncsToTranslate(TranslatorCache<TranslatedFunction> funcs)
         {
             var profiledFuncsToTranslate = new ConcurrentQueue<(ulong address, FuncProfile funcProfile)>();
 

ARMeilleure/Translation/SsaConstruction.cs

@@ -113,7 +113,7 @@ namespace ARMeilleure.Translation
                     }
                 }
 
-                Array.Clear(localDefs, 0, localDefs.Length);
+                Array.Clear(localDefs);
             }
 
             // Second pass, rename variables with definitions on different blocks.

ARMeilleure/Translation/Translator.cs

@@ -49,7 +49,7 @@ namespace ARMeilleure.Translation
         private readonly AutoResetEvent _backgroundTranslatorEvent;
         private readonly ReaderWriterLock _backgroundTranslatorLock;
 
-        internal ConcurrentDictionary<ulong, TranslatedFunction> Functions { get; }
+        internal TranslatorCache<TranslatedFunction> Functions { get; }
         internal AddressTable<ulong> FunctionTable { get; }
         internal EntryTable<uint> CountTable { get; }
         internal TranslatorStubs Stubs { get; }
@@ -75,7 +75,7 @@ namespace ARMeilleure.Translation
             JitCache.Initialize(allocator);
 
             CountTable = new EntryTable<uint>();
-            Functions = new ConcurrentDictionary<ulong, TranslatedFunction>();
+            Functions = new TranslatorCache<TranslatedFunction>();
             FunctionTable = new AddressTable<ulong>(for64Bits ? Levels64Bit : Levels32Bit);
             Stubs = new TranslatorStubs(this);
 
@@ -93,12 +93,12 @@ namespace ARMeilleure.Translation
             {
                 _backgroundTranslatorLock.AcquireReaderLock(Timeout.Infinite);
 
-                if (_backgroundStack.TryPop(out RejitRequest request) && 
+                if (_backgroundStack.TryPop(out RejitRequest request) &&
                     _backgroundSet.TryRemove(request.Address, out _))
                 {
                     TranslatedFunction func = Translate(request.Address, request.Mode, highCq: true);
 
-                    Functions.AddOrUpdate(request.Address, func, (key, oldFunc) =>
+                    Functions.AddOrUpdate(request.Address, func.GuestSize, func, (key, oldFunc) =>
                     {
                         EnqueueForDeletion(key, oldFunc);
                         return func;
@@ -196,7 +196,7 @@ namespace ARMeilleure.Translation
             }
         }
 
-        public ulong ExecuteSingle(State.ExecutionContext context, ulong address)
+        private ulong ExecuteSingle(State.ExecutionContext context, ulong address)
         {
             TranslatedFunction func = GetOrTranslate(address, context.ExecutionMode);
 
@@ -209,13 +209,24 @@ namespace ARMeilleure.Translation
             return nextAddr;
         }
 
+        public ulong Step(State.ExecutionContext context, ulong address)
+        {
+            TranslatedFunction func = Translate(address, context.ExecutionMode, highCq: false, singleStep: true);
+
+            address = func.Execute(context);
+
+            EnqueueForDeletion(address, func);
+
+            return address;
+        }
+
         internal TranslatedFunction GetOrTranslate(ulong address, ExecutionMode mode)
         {
             if (!Functions.TryGetValue(address, out TranslatedFunction func))
             {
                 func = Translate(address, mode, highCq: false);
 
-                TranslatedFunction oldFunc = Functions.GetOrAdd(address, func);
+                TranslatedFunction oldFunc = Functions.GetOrAdd(address, func.GuestSize, func);
 
                 if (oldFunc != func)
                 {
@@ -242,7 +253,7 @@ namespace ARMeilleure.Translation
             }
         }
 
-        internal TranslatedFunction Translate(ulong address, ExecutionMode mode, bool highCq)
+        internal TranslatedFunction Translate(ulong address, ExecutionMode mode, bool highCq, bool singleStep = false)
         {
             var context = new ArmEmitterContext(
                 Memory,
@@ -255,7 +266,7 @@ namespace ARMeilleure.Translation
 
             Logger.StartPass(PassName.Decoding);
 
-            Block[] blocks = Decoder.Decode(Memory, address, mode, highCq, singleBlock: false);
+            Block[] blocks = Decoder.Decode(Memory, address, mode, highCq, singleStep ? DecoderMode.SingleInstruction : DecoderMode.MultipleBlocks);
 
             Logger.EndPass(PassName.Decoding);
 
@@ -285,14 +296,14 @@ namespace ARMeilleure.Translation
 
             var options = highCq ? CompilerOptions.HighCq : CompilerOptions.None;
 
-            if (context.HasPtc)
+            if (context.HasPtc && !singleStep)
             {
                 options |= CompilerOptions.Relocatable;
             }
 
             CompiledFunction compiledFunc = Compiler.Compile(cfg, argTypes, retType, options);
 
-            if (context.HasPtc)
+            if (context.HasPtc && !singleStep)
             {
                 Hash128 hash = Ptc.ComputeHash(Memory, address, funcSize);
 
@@ -471,7 +482,24 @@ namespace ARMeilleure.Translation
             // If rejit is running, stop it as it may be trying to rejit a function on the invalidated region.
             ClearRejitQueue(allowRequeue: true);
 
-            // TODO: Completely remove functions overlapping the specified range from the cache.
+            ulong[] overlapAddresses = Array.Empty<ulong>();
+
+            int overlapsCount = Functions.GetOverlaps(address, size, ref overlapAddresses);
+
+            for (int index = 0; index < overlapsCount; index++)
+            {
+                ulong overlapAddress = overlapAddresses[index];
+
+                if (Functions.TryGetValue(overlapAddress, out TranslatedFunction overlap))
+                {
+                    Functions.Remove(overlapAddress);
+                    Volatile.Write(ref FunctionTable.GetValue(overlapAddress), FunctionTable.Fill);
+                    EnqueueForDeletion(overlapAddress, overlap);
+                }
+            }
+
+            // TODO: Remove overlapping functions from the JitCache aswell.
+            // This should be done safely, with a mechanism to ensure the function is not being executed.
         }
 
         internal void EnqueueForRejit(ulong guestAddress, ExecutionMode mode)
@@ -493,7 +521,9 @@ namespace ARMeilleure.Translation
             // Ensure no attempt will be made to compile new functions due to rejit.
             ClearRejitQueue(allowRequeue: false);
 
-            foreach (var func in Functions.Values)
+            List<TranslatedFunction> functions = Functions.AsList();
+
+            foreach (var func in functions)
             {
                 JitCache.Unmap(func.FuncPtr);
 

ARMeilleure/Translation/TranslatorCache.cs

@@ -0,0 +1,95 @@
+using System;
+using System.Collections.Generic;
+using System.Threading;
+
+namespace ARMeilleure.Translation
+{
+    internal class TranslatorCache<T>
+    {
+        private readonly IntervalTree<ulong, T> _tree;
+        private readonly ReaderWriterLock _treeLock;
+
+        public int Count => _tree.Count;
+
+        public TranslatorCache()
+        {
+            _tree = new IntervalTree<ulong, T>();
+            _treeLock = new ReaderWriterLock();
+        }
+
+        public bool TryAdd(ulong address, ulong size, T value)
+        {
+            return AddOrUpdate(address, size, value, null);
+        }
+
+        public bool AddOrUpdate(ulong address, ulong size, T value, Func<ulong, T, T> updateFactoryCallback)
+        {
+            _treeLock.AcquireWriterLock(Timeout.Infinite);
+            bool result = _tree.AddOrUpdate(address, address + size, value, updateFactoryCallback);
+            _treeLock.ReleaseWriterLock();
+
+            return result;
+        }
+
+        public T GetOrAdd(ulong address, ulong size, T value)
+        {
+            _treeLock.AcquireWriterLock(Timeout.Infinite);
+            value = _tree.GetOrAdd(address, address + size, value);
+            _treeLock.ReleaseWriterLock();
+
+            return value;
+        }
+
+        public bool Remove(ulong address)
+        {
+            _treeLock.AcquireWriterLock(Timeout.Infinite);
+            bool removed = _tree.Remove(address) != 0;
+            _treeLock.ReleaseWriterLock();
+
+            return removed;
+        }
+
+        public void Clear()
+        {
+            _treeLock.AcquireWriterLock(Timeout.Infinite);
+            _tree.Clear();
+            _treeLock.ReleaseWriterLock();
+        }
+
+        public bool ContainsKey(ulong address)
+        {
+            _treeLock.AcquireReaderLock(Timeout.Infinite);
+            bool result = _tree.ContainsKey(address);
+            _treeLock.ReleaseReaderLock();
+
+            return result;
+        }
+
+        public bool TryGetValue(ulong address, out T value)
+        {
+            _treeLock.AcquireReaderLock(Timeout.Infinite);
+            bool result = _tree.TryGet(address, out value);
+            _treeLock.ReleaseReaderLock();
+
+            return result;
+        }
+
+        public int GetOverlaps(ulong address, ulong size, ref ulong[] overlaps)
+        {
+            _treeLock.AcquireReaderLock(Timeout.Infinite);
+            int count = _tree.Get(address, address + size, ref overlaps);
+            _treeLock.ReleaseReaderLock();
+
+            return count;
+        }
+
+        public List<T> AsList()
+        {
+            _treeLock.AcquireReaderLock(Timeout.Infinite);
+            List<T> list = _tree.AsList();
+            _treeLock.ReleaseReaderLock();
+
+            return list;
+        }
+    }
+}

Ryujinx.Audio/Renderer/Dsp/State/ReverbState.cs

@@ -112,12 +112,12 @@ namespace Ryujinx.Audio.Renderer.Dsp.State
 
         private ReadOnlySpan<float> GetFdnDelayTimesByLateMode(ReverbLateMode lateMode)
         {
-            return FdnDelayTimes.AsSpan().Slice((int)lateMode * 4, 4);
+            return FdnDelayTimes.AsSpan((int)lateMode * 4, 4);
         }
 
         private ReadOnlySpan<float> GetDecayDelayTimesByLateMode(ReverbLateMode lateMode)
         {
-            return DecayDelayTimes.AsSpan().Slice((int)lateMode * 4, 4);
+            return DecayDelayTimes.AsSpan((int)lateMode * 4, 4);
         }
 
         public ReverbState(ref ReverbParameter parameter, ulong workBuffer, bool isLongSizePreDelaySupported)

Ryujinx.Audio/Renderer/Server/Performance/PerformanceManagerGeneric.cs

@@ -149,11 +149,21 @@ namespace Ryujinx.Audio.Renderer.Server.Performance
 
                 Span<byte> targetSpan = performanceOutput.Slice(nextOffset);
 
+                // NOTE: We check for the space for two headers for the final blank header.
+                int requiredSpace = Unsafe.SizeOf<THeader>() + Unsafe.SizeOf<TEntry>() * inputHeader.GetEntryCount()
+                                                             + Unsafe.SizeOf<TEntryDetail>() * inputHeader.GetEntryDetailCount()
+                                                             + Unsafe.SizeOf<THeader>();
+
+                if (targetSpan.Length < requiredSpace)
+                {
+                    break;
+                }
+
                 ref THeader outputHeader = ref MemoryMarshal.Cast<byte, THeader>(targetSpan)[0];
 
                 nextOffset += Unsafe.SizeOf<THeader>();
 
-                Span<TEntry> outputEntries = MemoryMarshal.Cast<byte, TEntry>(targetSpan.Slice(nextOffset));
+                Span<TEntry> outputEntries = MemoryMarshal.Cast<byte, TEntry>(performanceOutput.Slice(nextOffset));
 
                 int totalProcessingTime = 0;
 
@@ -175,7 +185,7 @@ namespace Ryujinx.Audio.Renderer.Server.Performance
                     }
                 }
 
-                Span<TEntryDetail> outputEntriesDetail = MemoryMarshal.Cast<byte, TEntryDetail>(targetSpan.Slice(nextOffset));
+                Span<TEntryDetail> outputEntriesDetail = MemoryMarshal.Cast<byte, TEntryDetail>(performanceOutput.Slice(nextOffset));
 
                 int effectiveEntryDetailCount = 0;
 

Ryujinx.Audio/Renderer/Server/Voice/VoiceState.cs

@@ -459,7 +459,7 @@ namespace Ryujinx.Audio.Renderer.Server.Voice
 
             for (int i = 0; i < Constants.VoiceWaveBufferCount; i++)
             {
-                UpdateWaveBuffer(errorInfos.AsSpan().Slice(i * 2, 2), ref WaveBuffers[i], ref parameter.WaveBuffers[i], parameter.SampleFormat, voiceUpdateState.IsWaveBufferValid[i], ref mapper, ref behaviourContext);
+                UpdateWaveBuffer(errorInfos.AsSpan(i * 2, 2), ref WaveBuffers[i], ref parameter.WaveBuffers[i], parameter.SampleFormat, voiceUpdateState.IsWaveBufferValid[i], ref mapper, ref behaviourContext);
             }
         }
 

Ryujinx.Common/Configuration/AppDataManager.cs

@@ -34,6 +34,7 @@ namespace Ryujinx.Common.Configuration
         private const string DefaultModsDir = "mods";
 
         public static string CustomModsPath { get; set; }
+        public static string CustomSdModsPath {get; set; }
         public static string CustomNandPath { get; set; } // TODO: Actually implement this into VFS
         public static string CustomSdCardPath { get; set; } // TODO: Actually implement this into VFS
 
@@ -84,6 +85,7 @@ namespace Ryujinx.Common.Configuration
             Directory.CreateDirectory(KeysDirPath = Path.Combine(BaseDirPath, KeysDir));
         }
 
-        public static string GetModsPath() => CustomModsPath ?? Directory.CreateDirectory(Path.Combine(BaseDirPath, DefaultModsDir)).FullName;
+        public static string GetModsPath()   => CustomModsPath ?? Directory.CreateDirectory(Path.Combine(BaseDirPath, DefaultModsDir)).FullName;
+        public static string GetSdModsPath() => CustomSdModsPath ?? Directory.CreateDirectory(Path.Combine(BaseDirPath, DefaultSdcardDir, "atmosphere")).FullName;
     }
 }

Ryujinx.Common/Configuration/Hid/Controller/JoyconConfigControllerStick.cs

@@ -5,6 +5,7 @@
         public Stick Joystick { get; set; }
         public bool InvertStickX { get; set; }
         public bool InvertStickY { get; set; }
+        public bool Rotate90CW { get; set; }
         public Button StickButton { get; set; }
     }
 }

Ryujinx.Common/Logging/LogClass.cs

@@ -47,6 +47,7 @@ namespace Ryujinx.Common.Logging
         ServiceNim,
         ServiceNs,
         ServiceNsd,
+        ServiceNtc,
         ServiceNv,
         ServiceOlsc,
         ServicePctl,

Ryujinx.Common/ReleaseInformations.cs

@@ -1,10 +1,14 @@
-using System.Reflection;
+using Ryujinx.Common.Configuration;
+using System;
+using System.Reflection;
 
 namespace Ryujinx.Common
 {
     // DO NOT EDIT, filled by CI
     public static class ReleaseInformations
     {
+        private const string FlatHubChannelOwner = "flathub";
+
         public static string BuildVersion = "%%RYUJINX_BUILD_VERSION%%";
         public static string BuildGitHash = "%%RYUJINX_BUILD_GIT_HASH%%";
         public static string ReleaseChannelName = "%%RYUJINX_TARGET_RELEASE_CHANNEL_NAME%%";
@@ -19,6 +23,11 @@ namespace Ryujinx.Common
                    !ReleaseChannelRepo.StartsWith("%%");
         }
 
+        public static bool IsFlatHubBuild()
+        {
+            return IsValid() && ReleaseChannelOwner.Equals(FlatHubChannelOwner);
+        }
+
         public static string GetVersion()
         {
             if (IsValid())
@@ -30,5 +39,15 @@ namespace Ryujinx.Common
                 return Assembly.GetEntryAssembly().GetCustomAttribute<AssemblyInformationalVersionAttribute>().InformationalVersion;
             }
         }
+
+        public static string GetBaseApplicationDirectory()
+        {
+            if (IsFlatHubBuild())
+            {
+                return AppDataManager.BaseDirPath;
+            }
+
+            return AppDomain.CurrentDomain.BaseDirectory;
+        }
     }
 }

Ryujinx.Cpu/CpuContext.cs

@@ -28,5 +28,10 @@ namespace Ryujinx.Cpu
         {
             _translator.Execute(context, address);
         }
+
+        public void InvalidateCacheRegion(ulong address, ulong size)
+        {
+            _translator.InvalidateJitCacheRegion(address, size);
+        }
     }
 }

Ryujinx.Graphics.GAL/Format.cs

@@ -52,7 +52,7 @@ namespace Ryujinx.Graphics.GAL
         R32G32B32A32Sint,
         S8Uint,
         D16Unorm,
-        D24X8Unorm,
+        S8UintD24Unorm,
         D32Float,
         D24UnormS8Uint,
         D32FloatS8Uint,
@@ -266,7 +266,7 @@ namespace Ryujinx.Graphics.GAL
             {
                 case Format.D16Unorm:
                 case Format.D24UnormS8Uint:
-                case Format.D24X8Unorm:
+                case Format.S8UintD24Unorm:
                 case Format.D32Float:
                 case Format.D32FloatS8Uint:
                 case Format.S8Uint:

Ryujinx.Graphics.Gpu/Engine/Dma/DmaClass.cs

@@ -1,7 +1,7 @@
 using Ryujinx.Common;
-using Ryujinx.Common.Logging;
 using Ryujinx.Graphics.Device;
 using Ryujinx.Graphics.Gpu.Engine.Threed;
+using Ryujinx.Graphics.Gpu.Memory;
 using Ryujinx.Graphics.Texture;
 using System;
 using System.Collections.Generic;
@@ -330,11 +330,95 @@ namespace Ryujinx.Graphics.Gpu.Engine.Dma
                 {
                     // TODO: Implement remap functionality.
                     // Buffer to buffer copy.
-                    memoryManager.Physical.BufferCache.CopyBuffer(memoryManager, srcGpuVa, dstGpuVa, size);
+
+                    bool srcIsPitchKind = memoryManager.GetKind(srcGpuVa).IsPitch();
+                    bool dstIsPitchKind = memoryManager.GetKind(dstGpuVa).IsPitch();
+
+                    if (!srcIsPitchKind && dstIsPitchKind)
+                    {
+                        CopyGobBlockLinearToLinear(memoryManager, srcGpuVa, dstGpuVa, size);
+                    }
+                    else if (srcIsPitchKind && !dstIsPitchKind)
+                    {
+                        CopyGobLinearToBlockLinear(memoryManager, srcGpuVa, dstGpuVa, size);
+                    }
+                    else
+                    {
+                        memoryManager.Physical.BufferCache.CopyBuffer(memoryManager, srcGpuVa, dstGpuVa, size);
+                    }
                 }
             }
         }
 
+        /// <summary>
+        /// Copies block linear data with block linear GOBs to a block linear destination with linear GOBs.
+        /// </summary>
+        /// <param name="memoryManager">GPU memory manager</param>
+        /// <param name="srcGpuVa">Source GPU virtual address</param>
+        /// <param name="dstGpuVa">Destination GPU virtual address</param>
+        /// <param name="size">Size in bytes of the copy</param>
+        private static void CopyGobBlockLinearToLinear(MemoryManager memoryManager, ulong srcGpuVa, ulong dstGpuVa, ulong size)
+        {
+            if (((srcGpuVa | dstGpuVa | size) & 0xf) == 0)
+            {
+                for (ulong offset = 0; offset < size; offset += 16)
+                {
+                    Vector128<byte> data = memoryManager.Read<Vector128<byte>>(ConvertGobLinearToBlockLinearAddress(srcGpuVa + offset), true);
+                    memoryManager.Write(dstGpuVa + offset, data);
+                }
+            }
+            else
+            {
+                for (ulong offset = 0; offset < size; offset++)
+                {
+                    byte data = memoryManager.Read<byte>(ConvertGobLinearToBlockLinearAddress(srcGpuVa + offset), true);
+                    memoryManager.Write(dstGpuVa + offset, data);
+                }
+            }
+        }
+
+        /// <summary>
+        /// Copies block linear data with linear GOBs to a block linear destination with block linear GOBs.
+        /// </summary>
+        /// <param name="memoryManager">GPU memory manager</param>
+        /// <param name="srcGpuVa">Source GPU virtual address</param>
+        /// <param name="dstGpuVa">Destination GPU virtual address</param>
+        /// <param name="size">Size in bytes of the copy</param>
+        private static void CopyGobLinearToBlockLinear(MemoryManager memoryManager, ulong srcGpuVa, ulong dstGpuVa, ulong size)
+        {
+            if (((srcGpuVa | dstGpuVa | size) & 0xf) == 0)
+            {
+                for (ulong offset = 0; offset < size; offset += 16)
+                {
+                    Vector128<byte> data = memoryManager.Read<Vector128<byte>>(srcGpuVa + offset, true);
+                    memoryManager.Write(ConvertGobLinearToBlockLinearAddress(dstGpuVa + offset), data);
+                }
+            }
+            else
+            {
+                for (ulong offset = 0; offset < size; offset++)
+                {
+                    byte data = memoryManager.Read<byte>(srcGpuVa + offset, true);
+                    memoryManager.Write(ConvertGobLinearToBlockLinearAddress(dstGpuVa + offset), data);
+                }
+            }
+        }
+
+        /// <summary>
+        /// Calculates the GOB block linear address from a linear address.
+        /// </summary>
+        /// <param name="address">Linear address</param>
+        /// <returns>Block linear address</returns>
+        private static ulong ConvertGobLinearToBlockLinearAddress(ulong address)
+        {
+            // y2 y1 y0 x5 x4 x3 x2 x1 x0 -> x5 y2 y1 x4 y0 x3 x2 x1 x0
+            return (address & ~0x1f0UL) |
+                ((address & 0x40) >> 2) |
+                ((address & 0x10) << 1) |
+                ((address & 0x180) >> 1) |
+                ((address & 0x20) << 3);
+        }
+
         /// <summary>
         /// Performs a buffer to buffer, or buffer to texture copy, then optionally releases a semaphore.
         /// </summary>

Ryujinx.Graphics.Gpu/Engine/Twod/TwodClass.cs

@@ -1,11 +1,15 @@
-using Ryujinx.Graphics.Device;
+using Ryujinx.Common;
+using Ryujinx.Graphics.Device;
 using Ryujinx.Graphics.GAL;
 using Ryujinx.Graphics.Gpu.Engine.Types;
 using Ryujinx.Graphics.Gpu.Image;
 using Ryujinx.Graphics.Texture;
+using Ryujinx.Memory;
 using System;
 using System.Collections.Generic;
 using System.Runtime.CompilerServices;
+using System.Runtime.InteropServices;
+using System.Runtime.Intrinsics;
 
 namespace Ryujinx.Graphics.Gpu.Engine.Twod
 {
@@ -44,6 +48,180 @@ namespace Ryujinx.Graphics.Gpu.Engine.Twod
         /// <param name="data">Data to be written</param>
         public void Write(int offset, int data) => _state.Write(offset, data);
 
+        /// <summary>
+        /// Determines if data is compatible between the source and destination texture.
+        /// The two textures must have the same size, layout, and bytes per pixel.
+        /// </summary>
+        /// <param name="lhs">Info for the first texture</param>
+        /// <param name="rhs">Info for the second texture</param>
+        /// <param name="lhsFormat">Format of the first texture</param>
+        /// <param name="rhsFormat">Format of the second texture</param>
+        /// <returns>True if the data is compatible, false otherwise</returns>
+        private bool IsDataCompatible(TwodTexture lhs, TwodTexture rhs, FormatInfo lhsFormat, FormatInfo rhsFormat)
+        {
+            if (lhsFormat.BytesPerPixel != rhsFormat.BytesPerPixel ||
+                lhs.Height != rhs.Height ||
+                lhs.Depth != rhs.Depth ||
+                lhs.LinearLayout != rhs.LinearLayout ||
+                lhs.MemoryLayout.Packed != rhs.MemoryLayout.Packed)
+            {
+                return false;
+            }
+
+            if (lhs.LinearLayout)
+            {
+                return lhs.Stride == rhs.Stride;
+            }
+            else
+            {
+                return lhs.Width == rhs.Width;
+            }
+        }
+
+        /// <summary>
+        /// Determine if the given region covers the full texture, also considering width alignment.
+        /// </summary>
+        /// <param name="texture">The texture to check</param>
+        /// <param name="formatInfo"></param>
+        /// <param name="x1">Region start x</param>
+        /// <param name="y1">Region start y</param>
+        /// <param name="x2">Region end x</param>
+        /// <param name="y2">Region end y</param>
+        /// <returns>True if the region covers the full texture, false otherwise</returns>
+        private bool IsCopyRegionComplete(TwodTexture texture, FormatInfo formatInfo, int x1, int y1, int x2, int y2)
+        {
+            if (x1 != 0 || y1 != 0 || y2 != texture.Height)
+            {
+                return false;
+            }
+
+            int width;
+            int widthAlignment;
+
+            if (texture.LinearLayout)
+            {
+                widthAlignment = 1;
+                width = texture.Stride / formatInfo.BytesPerPixel;
+            }
+            else
+            {
+                widthAlignment = Constants.GobAlignment / formatInfo.BytesPerPixel;
+                width = texture.Width;
+            }
+
+            return width == BitUtils.AlignUp(x2, widthAlignment);
+        }
+
+        /// <summary>
+        /// Performs a full data copy between two textures, reading and writing guest memory directly.
+        /// The textures must have a matching layout, size, and bytes per pixel.
+        /// </summary>
+        /// <param name="src">The source texture</param>
+        /// <param name="dst">The destination texture</param>
+        /// <param name="w">Copy width</param>
+        /// <param name="h">Copy height</param>
+        /// <param name="bpp">Bytes per pixel</param>
+        private void UnscaledFullCopy(TwodTexture src, TwodTexture dst, int w, int h, int bpp)
+        {
+            var srcCalculator = new OffsetCalculator(
+                w,
+                h,
+                src.Stride,
+                src.LinearLayout,
+                src.MemoryLayout.UnpackGobBlocksInY(),
+                src.MemoryLayout.UnpackGobBlocksInZ(),
+                bpp);
+
+            (int _, int srcSize) = srcCalculator.GetRectangleRange(0, 0, w, h);
+
+            var memoryManager = _channel.MemoryManager;
+
+            ulong srcGpuVa = src.Address.Pack();
+            ulong dstGpuVa = dst.Address.Pack();
+
+            ReadOnlySpan<byte> srcSpan = memoryManager.GetSpan(srcGpuVa, srcSize, true);
+
+            int width;
+            int height = src.Height;
+            if (src.LinearLayout)
+            {
+                width = src.Stride / bpp;
+            }
+            else
+            {
+                width = src.Width;
+            }
+
+            // If the copy is not equal to the width and height of the texture, we will need to copy partially.
+            // It's worth noting that it has already been established that the src and dst are the same size.
+
+            if (w == width && h == height)
+            {
+                memoryManager.Write(dstGpuVa, srcSpan);
+            }
+            else
+            {
+                using WritableRegion dstRegion = memoryManager.GetWritableRegion(dstGpuVa, srcSize, true);
+                Span<byte> dstSpan = dstRegion.Memory.Span;
+
+                if (src.LinearLayout)
+                {
+                    int stride = src.Stride;
+                    int offset = 0;
+                    int lineSize = width * bpp;
+
+                    for (int y = 0; y < height; y++)
+                    {
+                        srcSpan.Slice(offset, lineSize).CopyTo(dstSpan.Slice(offset));
+
+                        offset += stride;
+                    }
+                }
+                else
+                {
+                    // Copy with the block linear layout in mind.
+                    // Recreate the offset calculate with bpp 1 for copy.
+
+                    int stride = w * bpp;
+
+                    srcCalculator = new OffsetCalculator(
+                        stride,
+                        h,
+                        0,
+                        false,
+                        src.MemoryLayout.UnpackGobBlocksInY(),
+                        src.MemoryLayout.UnpackGobBlocksInZ(),
+                        1);
+
+                    int strideTrunc = BitUtils.AlignDown(stride, 16);
+
+                    ReadOnlySpan<Vector128<byte>> srcVec = MemoryMarshal.Cast<byte, Vector128<byte>>(srcSpan);
+                    Span<Vector128<byte>> dstVec = MemoryMarshal.Cast<byte, Vector128<byte>>(dstSpan);
+
+                    for (int y = 0; y < h; y++)
+                    {
+                        int x = 0;
+
+                        srcCalculator.SetY(y);
+
+                        for (; x < strideTrunc; x += 16)
+                        {
+                            int offset = srcCalculator.GetOffset(x) >> 4;
+
+                            dstVec[offset] = srcVec[offset];
+                        }
+
+                        for (; x < stride; x++)
+                        {
+                            int offset = srcCalculator.GetOffset(x);
+
+                            dstSpan[offset] = srcSpan[offset];
+                        }
+                    }
+                }
+            }
+        }
+
         /// <summary>
         /// Performs the blit operation, triggered by the register write.
         /// </summary>
@@ -114,16 +292,31 @@ namespace Ryujinx.Graphics.Gpu.Engine.Twod
                 srcX1 = 0;
             }
 
+            FormatInfo dstCopyTextureFormat = dstCopyTexture.Format.Convert();
+
+            bool canDirectCopy = GraphicsConfig.Fast2DCopy &&
+                srcX2 == dstX2 && srcY2 == dstY2 &&
+                IsDataCompatible(srcCopyTexture, dstCopyTexture, srcCopyTextureFormat, dstCopyTextureFormat) &&
+                IsCopyRegionComplete(srcCopyTexture, srcCopyTextureFormat, srcX1, srcY1, srcX2, srcY2) &&
+                IsCopyRegionComplete(dstCopyTexture, dstCopyTextureFormat, dstX1, dstY1, dstX2, dstY2);
+
             var srcTexture = memoryManager.Physical.TextureCache.FindOrCreateTexture(
                 memoryManager,
                 srcCopyTexture,
                 offset,
                 srcCopyTextureFormat,
+                !canDirectCopy,
                 false,
                 srcHint);
 
             if (srcTexture == null)
             {
+                if (canDirectCopy)
+                {
+                    // Directly copy the data on CPU.
+                    UnscaledFullCopy(srcCopyTexture, dstCopyTexture, srcX2, srcY2, srcCopyTextureFormat.BytesPerPixel);
+                }
+
                 return;
             }
 
@@ -132,7 +325,6 @@ namespace Ryujinx.Graphics.Gpu.Engine.Twod
             // When the source texture that was found has a depth format,
             // we must enforce the target texture also has a depth format,
             // as copies between depth and color formats are not allowed.
-            FormatInfo dstCopyTextureFormat;
 
             if (srcTexture.Format.IsDepthOrStencil())
             {
@@ -148,6 +340,7 @@ namespace Ryujinx.Graphics.Gpu.Engine.Twod
                 dstCopyTexture,
                 0,
                 dstCopyTextureFormat,
+                true,
                 srcTexture.ScaleMode == TextureScaleMode.Scaled,
                 dstHint);
 

Ryujinx.Graphics.Gpu/Engine/Types/ZetaFormat.cs

@@ -28,13 +28,13 @@ namespace Ryujinx.Graphics.Gpu.Engine.Types
         {
             return format switch
             {
-                ZetaFormat.D32Float          => new FormatInfo(Format.D32Float,          1, 1, 4,  1),
-                ZetaFormat.D16Unorm          => new FormatInfo(Format.D16Unorm,          1, 1, 2,  1),
-                ZetaFormat.D24UnormS8Uint    => new FormatInfo(Format.D24UnormS8Uint,    1, 1, 4,  2),
-                ZetaFormat.D24Unorm          => new FormatInfo(Format.D24UnormS8Uint,    1, 1, 4,  1),
-                ZetaFormat.S8UintD24Unorm    => new FormatInfo(Format.D24UnormS8Uint,    1, 1, 4,  2),
-                ZetaFormat.S8Uint            => new FormatInfo(Format.S8Uint,            1, 1, 1,  1),
-                ZetaFormat.D32FloatS8Uint    => new FormatInfo(Format.D32FloatS8Uint,    1, 1, 8,  2),
+                ZetaFormat.D32Float          => new FormatInfo(Format.D32Float,       1, 1, 4, 1),
+                ZetaFormat.D16Unorm          => new FormatInfo(Format.D16Unorm,       1, 1, 2, 1),
+                ZetaFormat.D24UnormS8Uint    => new FormatInfo(Format.D24UnormS8Uint, 1, 1, 4, 2),
+                ZetaFormat.D24Unorm          => new FormatInfo(Format.D24UnormS8Uint, 1, 1, 4, 1),
+                ZetaFormat.S8UintD24Unorm    => new FormatInfo(Format.S8UintD24Unorm, 1, 1, 4, 2),
+                ZetaFormat.S8Uint            => new FormatInfo(Format.S8Uint,         1, 1, 1, 1),
+                ZetaFormat.D32FloatS8Uint    => new FormatInfo(Format.D32FloatS8Uint, 1, 1, 8, 2),
                 _                            => FormatInfo.Default
             };
         }

Ryujinx.Graphics.Gpu/GraphicsConfig.cs

@@ -28,6 +28,14 @@ namespace Ryujinx.Graphics.Gpu
         /// </summary>
         public static bool FastGpuTime = true;
 
+        /// <summary>
+        /// Enables or disables fast 2d engine texture copies entirely on CPU when possible.
+        /// Reduces stuttering and # of textures in games that copy textures around for streaming, 
+        /// as textures will not need to be created for the copy, and the data does not need to be 
+        /// flushed from GPU.
+        /// </summary>
+        public static bool Fast2DCopy = true;
+
         /// <summary>
         /// Enables or disables the Just-in-Time compiler for GPU Macro code.
         /// </summary>

Ryujinx.Graphics.Gpu/Image/FormatTable.cs

@@ -55,6 +55,7 @@ namespace Ryujinx.Graphics.Gpu.Image
             { 0x24a0e, new FormatInfo(Format.D24UnormS8Uint,    1,  1,  4,  2) },
             { 0x24a29, new FormatInfo(Format.D24UnormS8Uint,    1,  1,  4,  2) },
             { 0x48a29, new FormatInfo(Format.D24UnormS8Uint,    1,  1,  4,  2) },
+            { 0x4912b, new FormatInfo(Format.S8UintD24Unorm,    1,  1,  4,  2) },
             { 0x25385, new FormatInfo(Format.D32FloatS8Uint,    1,  1,  8,  2) },
             { 0x253b0, new FormatInfo(Format.D32FloatS8Uint,    1,  1,  8,  2) },
             { 0xa4908, new FormatInfo(Format.R8G8B8A8Srgb,      1,  1,  4,  4) },

Ryujinx.Graphics.Gpu/Image/TextureCache.cs

@@ -7,7 +7,6 @@ using Ryujinx.Graphics.Gpu.Engine.Types;
 using Ryujinx.Graphics.Gpu.Image;
 using Ryujinx.Graphics.Gpu.Memory;
 using Ryujinx.Graphics.Texture;
-using Ryujinx.Memory;
 using Ryujinx.Memory.Range;
 using System;
 using System.Collections.Generic;
@@ -40,6 +39,7 @@ namespace Ryujinx.Graphics.Gpu.Image
         private readonly PhysicalMemory _physicalMemory;
 
         private readonly MultiRangeList<Texture> _textures;
+        private readonly HashSet<Texture> _partiallyMappedTextures;
 
         private Texture[] _textureOverlaps;
         private OverlapInfo[] _overlapInfo;
@@ -57,6 +57,7 @@ namespace Ryujinx.Graphics.Gpu.Image
             _physicalMemory = physicalMemory;
 
             _textures = new MultiRangeList<Texture>();
+            _partiallyMappedTextures = new HashSet<Texture>();
 
             _textureOverlaps = new Texture[OverlapsBufferInitialCapacity];
             _overlapInfo = new OverlapInfo[OverlapsBufferInitialCapacity];
@@ -74,17 +75,7 @@ namespace Ryujinx.Graphics.Gpu.Image
             Texture[] overlaps = new Texture[10];
             int overlapCount;
 
-            MultiRange unmapped;
-
-            try
-            {
-                unmapped = ((MemoryManager)sender).GetPhysicalRegions(e.Address, e.Size);
-            }
-            catch (InvalidMemoryRegionException)
-            {
-                // This event fires on Map in case any mappings are overwritten. In that case, there may not be an existing mapping.
-                return;
-            }
+            MultiRange unmapped = ((MemoryManager)sender).GetPhysicalRegions(e.Address, e.Size);
 
             lock (_textures)
             {
@@ -95,6 +86,24 @@ namespace Ryujinx.Graphics.Gpu.Image
             {
                 overlaps[i].Unmapped(unmapped);
             }
+
+            // If any range was previously unmapped, we also need to purge
+            // all partially mapped texture, as they might be fully mapped now.
+            for (int i = 0; i < unmapped.Count; i++)
+            {
+                if (unmapped.GetSubRange(i).Address == MemoryManager.PteUnmapped)
+                {
+                    lock (_partiallyMappedTextures)
+                    {
+                        foreach (var texture in _partiallyMappedTextures)
+                        {
+                            texture.Unmapped(unmapped);
+                        }
+                    }
+
+                    break;
+                }
+            }
         }
 
         /// <summary>
@@ -194,6 +203,7 @@ namespace Ryujinx.Graphics.Gpu.Image
             TwodTexture copyTexture,
             ulong offset,
             FormatInfo formatInfo,
+            bool shouldCreate,
             bool preferScaling = true,
             Size? sizeHint = null)
         {
@@ -234,6 +244,11 @@ namespace Ryujinx.Graphics.Gpu.Image
                 flags |= TextureSearchFlags.WithUpscale;
             }
 
+            if (!shouldCreate)
+            {
+                flags |= TextureSearchFlags.NoCreate;
+            }
+
             Texture texture = FindOrCreateTexture(memoryManager, flags, info, 0, sizeHint);
 
             texture?.SynchronizeMemory();
@@ -480,15 +495,29 @@ namespace Ryujinx.Graphics.Gpu.Image
 
                 return texture;
             }
+            else if (flags.HasFlag(TextureSearchFlags.NoCreate))
+            {
+                return null;
+            }
 
             // Calculate texture sizes, used to find all overlapping textures.
             SizeInfo sizeInfo = info.CalculateSizeInfo(layerSize);
 
             ulong size = (ulong)sizeInfo.TotalSize;
+            bool partiallyMapped = false;
 
             if (range == null)
             {
                 range = memoryManager.GetPhysicalRegions(info.GpuAddress, size);
+
+                for (int i = 0; i < range.Value.Count; i++)
+                {
+                    if (range.Value.GetSubRange(i).Address == MemoryManager.PteUnmapped)
+                    {
+                        partiallyMapped = true;
+                        break;
+                    }
+                }
             }
 
             // Find view compatible matches.
@@ -658,7 +687,7 @@ namespace Ryujinx.Graphics.Gpu.Image
                     else
                     {
                         bool dataOverlaps = texture.DataOverlaps(overlap, compatibility);
-                        
+
                         if (!overlap.IsView && dataOverlaps && !incompatibleOverlaps.Exists(incompatible => incompatible.Group == overlap.Group))
                         {
                             incompatibleOverlaps.Add(new TextureIncompatibleOverlap(overlap.Group, compatibility));
@@ -774,6 +803,14 @@ namespace Ryujinx.Graphics.Gpu.Image
                 _textures.Add(texture);
             }
 
+            if (partiallyMapped)
+            {
+                lock (_partiallyMappedTextures)
+                {
+                    _partiallyMappedTextures.Add(texture);
+                }
+            }
+
             ShrinkOverlapsBufferIfNeeded();
 
             for (int i = 0; i < overlapsCount; i++)
@@ -1069,6 +1106,11 @@ namespace Ryujinx.Graphics.Gpu.Image
             {
                 _textures.Remove(texture);
             }
+
+            lock (_partiallyMappedTextures)
+            {
+                _partiallyMappedTextures.Remove(texture);
+            }
         }
 
         /// <summary>

Ryujinx.Graphics.Gpu/Image/TextureCompatibility.cs

@@ -203,7 +203,7 @@ namespace Ryujinx.Graphics.Gpu.Image
                 }
 
                 if ((lhs.FormatInfo.Format == Format.D24UnormS8Uint ||
-                     lhs.FormatInfo.Format == Format.D24X8Unorm) && rhs.FormatInfo.Format == Format.B8G8R8A8Unorm)
+                     lhs.FormatInfo.Format == Format.S8UintD24Unorm) && rhs.FormatInfo.Format == Format.B8G8R8A8Unorm)
                 {
                     return TextureMatchQuality.FormatAlias;
                 }

Ryujinx.Graphics.Gpu/Image/TextureGroup.cs

@@ -236,7 +236,7 @@ namespace Ryujinx.Graphics.Gpu.Image
         }
 
         /// <summary>
-        /// Synchronize memory for a given texture. 
+        /// Synchronize memory for a given texture.
         /// If overlapping tracking handles are dirty, fully or partially synchronize the texture data.
         /// </summary>
         /// <param name="texture">The texture being used</param>
@@ -280,7 +280,7 @@ namespace Ryujinx.Graphics.Gpu.Image
 
                     // Evaluate if any copy dependencies need to be fulfilled. A few rules:
                     // If the copy handle needs to be synchronized, prefer our own state.
-                    // If we need to be synchronized and there is a copy present, prefer the copy. 
+                    // If we need to be synchronized and there is a copy present, prefer the copy.
 
                     if (group.NeedsCopy && group.Copy(_context))
                     {
@@ -618,7 +618,7 @@ namespace Ryujinx.Graphics.Gpu.Image
         }
 
         /// <summary>
-        /// Evaluate the range of tracking handles which a view texture overlaps with, 
+        /// Evaluate the range of tracking handles which a view texture overlaps with,
         /// using the view's position and slice/level counts.
         /// </summary>
         /// <param name="firstLayer">The first layer of the texture</param>
@@ -879,7 +879,7 @@ namespace Ryujinx.Graphics.Gpu.Image
                 int sliceStart = Math.Clamp(offset, 0, subRangeSize);
                 int sliceEnd = Math.Clamp(endOffset, 0, subRangeSize);
 
-                if (sliceStart != sliceEnd)
+                if (sliceStart != sliceEnd && item.Address != MemoryManager.PteUnmapped)
                 {
                     result.Add(GenerateHandle(item.Address + (ulong)sliceStart, (ulong)(sliceEnd - sliceStart)));
                 }
@@ -1097,11 +1097,20 @@ namespace Ryujinx.Graphics.Gpu.Image
             {
                 // Single dirty region.
                 var cpuRegionHandles = new CpuRegionHandle[TextureRange.Count];
+                int count = 0;
 
                 for (int i = 0; i < TextureRange.Count; i++)
                 {
                     var currentRange = TextureRange.GetSubRange(i);
-                    cpuRegionHandles[i] = GenerateHandle(currentRange.Address, currentRange.Size);
+                    if (currentRange.Address != MemoryManager.PteUnmapped)
+                    {
+                        cpuRegionHandles[count++] = GenerateHandle(currentRange.Address, currentRange.Size);
+                    }
+                }
+
+                if (count != TextureRange.Count)
+                {
+                    Array.Resize(ref cpuRegionHandles, count);
                 }
 
                 var groupHandle = new TextureGroupHandle(this, 0, Storage.Size, _views, 0, 0, 0, _allOffsets.Length, cpuRegionHandles);
@@ -1277,7 +1286,7 @@ namespace Ryujinx.Graphics.Gpu.Image
                         TextureInfo info = Storage.Info;
                         TextureInfo otherInfo = other.Storage.Info;
 
-                        if (TextureCompatibility.ViewLayoutCompatible(info, otherInfo, level, otherLevel) && 
+                        if (TextureCompatibility.ViewLayoutCompatible(info, otherInfo, level, otherLevel) &&
                             TextureCompatibility.CopySizeMatches(info, otherInfo, level, otherLevel))
                         {
                             // These textures are copy compatible. Create the dependency.

Ryujinx.Graphics.Gpu/Image/TexturePool.cs

@@ -362,7 +362,7 @@ namespace Ryujinx.Graphics.Gpu.Image
                 return DepthStencilMode.Depth;
             }
 
-            if (format == Format.D24X8Unorm || format == Format.D24UnormS8Uint)
+            if (format == Format.D24UnormS8Uint)
             {
                 return component == SwizzleComponent.Red
                     ? DepthStencilMode.Stencil

Ryujinx.Graphics.Gpu/Image/TextureSearchFlags.cs

@@ -12,6 +12,7 @@ namespace Ryujinx.Graphics.Gpu.Image
         Strict      = 1 << 0,
         ForSampler  = 1 << 1,
         ForCopy     = 1 << 2,
-        WithUpscale = 1 << 3
+        WithUpscale = 1 << 3,
+        NoCreate    = 1 << 4
     }
 }

Ryujinx.Graphics.Gpu/Memory/BufferCache.cs

@@ -17,6 +17,8 @@ namespace Ryujinx.Graphics.Gpu.Memory
         private const ulong BufferAlignmentSize = 0x1000;
         private const ulong BufferAlignmentMask = BufferAlignmentSize - 1;
 
+        private const ulong MaxDynamicGrowthSize = 0x100000;
+
         private readonly GpuContext _context;
         private readonly PhysicalMemory _physicalMemory;
 
@@ -166,10 +168,35 @@ namespace Ryujinx.Graphics.Gpu.Memory
                 // Otherwise, we must delete the overlapping buffers and create a bigger buffer
                 // that fits all the data we need. We also need to copy the contents from the
                 // old buffer(s) to the new buffer.
+
                 ulong endAddress = address + size;
 
                 if (_bufferOverlaps[0].Address > address || _bufferOverlaps[0].EndAddress < endAddress)
                 {
+                    // Check if the following conditions are met:
+                    // - We have a single overlap.
+                    // - The overlap starts at or before the requested range. That is, the overlap happens at the end.
+                    // - The size delta between the new, merged buffer and the old one is of at most 2 pages.
+                    // In this case, we attempt to extend the buffer further than the requested range,
+                    // this can potentially avoid future resizes if the application keeps using overlapping
+                    // sequential memory.
+                    // Allowing for 2 pages (rather than just one) is necessary to catch cases where the
+                    // range crosses a page, and after alignment, ends having a size of 2 pages.
+                    if (overlapsCount == 1 &&
+                        address >= _bufferOverlaps[0].Address &&
+                        endAddress - _bufferOverlaps[0].EndAddress <= BufferAlignmentSize * 2)
+                    {
+                        // Try to grow the buffer by 1.5x of its current size.
+                        // This improves performance in the cases where the buffer is resized often by small amounts.
+                        ulong existingSize = _bufferOverlaps[0].Size;
+                        ulong growthSize = (existingSize + Math.Min(existingSize >> 1, MaxDynamicGrowthSize)) & ~BufferAlignmentMask;
+
+                        size = Math.Max(size, growthSize);
+                        endAddress = address + size;
+
+                        overlapsCount = _buffers.FindOverlapsNonOverlapping(address, size, ref _bufferOverlaps);
+                    }
+
                     for (int index = 0; index < overlapsCount; index++)
                     {
                         Buffer buffer = _bufferOverlaps[index];
@@ -183,7 +210,9 @@ namespace Ryujinx.Graphics.Gpu.Memory
                         }
                     }
 
-                    Buffer newBuffer = new Buffer(_context, _physicalMemory, address, endAddress - address, _bufferOverlaps.Take(overlapsCount));
+                    ulong newSize = endAddress - address;
+
+                    Buffer newBuffer = new Buffer(_context, _physicalMemory, address, newSize, _bufferOverlaps.Take(overlapsCount));
 
                     lock (_buffers)
                     {
@@ -202,7 +231,7 @@ namespace Ryujinx.Graphics.Gpu.Memory
                         buffer.DisposeData();
                     }
 
-                    newBuffer.SynchronizeMemory(address, endAddress - address);
+                    newBuffer.SynchronizeMemory(address, newSize);
 
                     // Existing buffers were modified, we need to rebind everything.
                     NotifyBuffersModified?.Invoke();

Ryujinx.Graphics.Gpu/Memory/MemoryManager.cs

@@ -28,7 +28,7 @@ namespace Ryujinx.Graphics.Gpu.Memory
         private const int PtLvl1Bit = PtPageBits;
         private const int AddressSpaceBits = PtPageBits + PtLvl1Bits + PtLvl0Bits;
 
-        public const ulong PteUnmapped = 0xffffffff_ffffffff;
+        public const ulong PteUnmapped = ulong.MaxValue;
 
         private readonly ulong[][] _pageTable;
 
@@ -154,14 +154,15 @@ namespace Ryujinx.Graphics.Gpu.Memory
         /// <summary>
         /// Gets a writable region from GPU mapped memory.
         /// </summary>
-        /// <param name="address">Start address of the range</param>
+        /// <param name="va">Start address of the range</param>
         /// <param name="size">Size in bytes to be range</param>
+        /// <param name="tracked">True if write tracking is triggered on the span</param>
         /// <returns>A writable region with the data at the specified memory location</returns>
-        public WritableRegion GetWritableRegion(ulong va, int size)
+        public WritableRegion GetWritableRegion(ulong va, int size, bool tracked = false)
         {
             if (IsContiguous(va, size))
             {
-                return Physical.GetWritableRegion(Translate(va), size);
+                return Physical.GetWritableRegion(Translate(va), size, tracked);
             }
             else
             {
@@ -169,7 +170,7 @@ namespace Ryujinx.Graphics.Gpu.Memory
 
                 GetSpan(va, size).CopyTo(memory.Span);
 
-                return new WritableRegion(this, va, memory);
+                return new WritableRegion(this, va, memory, tracked);
             }
         }
 
@@ -263,7 +264,8 @@ namespace Ryujinx.Graphics.Gpu.Memory
         /// <param name="pa">CPU virtual address to map into</param>
         /// <param name="va">GPU virtual address to be mapped</param>
         /// <param name="size">Size in bytes of the mapping</param>
-        public void Map(ulong pa, ulong va, ulong size)
+        /// <param name="kind">Kind of the resource located at the mapping</param>
+        public void Map(ulong pa, ulong va, ulong size, PteKind kind)
         {
             lock (_pageTable)
             {
@@ -271,7 +273,7 @@ namespace Ryujinx.Graphics.Gpu.Memory
 
                 for (ulong offset = 0; offset < size; offset += PageSize)
                 {
-                    SetPte(va + offset, pa + offset);
+                    SetPte(va + offset, PackPte(pa + offset, kind));
                 }
             }
         }
@@ -339,7 +341,6 @@ namespace Ryujinx.Graphics.Gpu.Memory
         /// <param name="va">Virtual address of the range</param>
         /// <param name="size">Size of the range</param>
         /// <returns>Multi-range with the physical regions</returns>
-        /// <exception cref="InvalidMemoryRegionException">The memory region specified by <paramref name="va"/> and <paramref name="size"/> is not fully mapped</exception>
         public MultiRange GetPhysicalRegions(ulong va, ulong size)
         {
             if (IsContiguous(va, (int)size))
@@ -347,11 +348,6 @@ namespace Ryujinx.Graphics.Gpu.Memory
                 return new MultiRange(Translate(va), size);
             }
 
-            if (!IsMapped(va))
-            {
-                throw new InvalidMemoryRegionException($"The specified GPU virtual address 0x{va:X} is not mapped.");
-            }
-
             ulong regionStart = Translate(va);
             ulong regionSize = Math.Min(size, PageSize - (va & PageMask));
 
@@ -366,14 +362,10 @@ namespace Ryujinx.Graphics.Gpu.Memory
 
             for (int page = 0; page < pages - 1; page++)
             {
-                if (!IsMapped(va + PageSize))
-                {
-                    throw new InvalidMemoryRegionException($"The specified GPU virtual memory range 0x{va:X}..0x{(va + size):X} is not fully mapped.");
-                }
-
+                ulong currPa = Translate(va);
                 ulong newPa = Translate(va + PageSize);
 
-                if (Translate(va) + PageSize != newPa)
+                if ((currPa != PteUnmapped || newPa != PteUnmapped) && currPa + PageSize != newPa)
                 {
                     regions.Add(new MemoryRange(regionStart, regionSize));
                     regionStart = newPa;
@@ -404,18 +396,35 @@ namespace Ryujinx.Graphics.Gpu.Memory
             {
                 MemoryRange currentRange = range.GetSubRange(i);
 
-                ulong address = currentRange.Address & ~PageMask;
-                ulong endAddress = (currentRange.EndAddress + PageMask) & ~PageMask;
-
-                while (address < endAddress)
+                if (currentRange.Address != PteUnmapped)
                 {
-                    if (Translate(va) != address)
-                    {
-                        return false;
-                    }
+                    ulong address = currentRange.Address & ~PageMask;
+                    ulong endAddress = (currentRange.EndAddress + PageMask) & ~PageMask;
 
-                    va += PageSize;
-                    address += PageSize;
+                    while (address < endAddress)
+                    {
+                        if (Translate(va) != address)
+                        {
+                            return false;
+                        }
+
+                        va += PageSize;
+                        address += PageSize;
+                    }
+                }
+                else
+                {
+                    ulong endVa = va + (((currentRange.Size) + PageMask) & ~PageMask);
+
+                    while (va < endVa)
+                    {
+                        if (Translate(va) != PteUnmapped)
+                        {
+                            return false;
+                        }
+
+                        va += PageSize;
+                    }
                 }
             }
 
@@ -454,14 +463,37 @@ namespace Ryujinx.Graphics.Gpu.Memory
                 return PteUnmapped;
             }
 
-            ulong baseAddress = GetPte(va);
+            ulong pte = GetPte(va);
 
-            if (baseAddress == PteUnmapped)
+            if (pte == PteUnmapped)
             {
                 return PteUnmapped;
             }
 
-            return baseAddress + (va & PageMask);
+            return UnpackPaFromPte(pte) + (va & PageMask);
+        }
+
+        /// <summary>
+        /// Gets the kind of a given memory page.
+        /// This might indicate the type of resource that can be allocated on the page, and also texture tiling.
+        /// </summary>
+        /// <param name="va">GPU virtual address</param>
+        /// <returns>Kind of the memory page</returns>
+        public PteKind GetKind(ulong va)
+        {
+            if (!ValidateAddress(va))
+            {
+                return PteKind.Invalid;
+            }
+
+            ulong pte = GetPte(va);
+
+            if (pte == PteUnmapped)
+            {
+                return PteKind.Invalid;
+            }
+
+            return UnpackKindFromPte(pte);
         }
 
         /// <summary>
@@ -504,5 +536,36 @@ namespace Ryujinx.Graphics.Gpu.Memory
 
             _pageTable[l0][l1] = pte;
         }
+
+        /// <summary>
+        /// Creates a page table entry from a physical address and kind.
+        /// </summary>
+        /// <param name="pa">Physical address</param>
+        /// <param name="kind">Kind</param>
+        /// <returns>Page table entry</returns>
+        private static ulong PackPte(ulong pa, PteKind kind)
+        {
+            return pa | ((ulong)kind << 56);
+        }
+
+        /// <summary>
+        /// Unpacks kind from a page table entry.
+        /// </summary>
+        /// <param name="pte">Page table entry</param>
+        /// <returns>Kind</returns>
+        private static PteKind UnpackKindFromPte(ulong pte)
+        {
+            return (PteKind)(pte >> 56);
+        }
+
+        /// <summary>
+        /// Unpacks physical address from a page table entry.
+        /// </summary>
+        /// <param name="pte">Page table entry</param>
+        /// <returns>Physical address</returns>
+        private static ulong UnpackPaFromPte(ulong pte)
+        {
+            return pte & 0xffffffffffffffUL;
+        }
     }
 }

Ryujinx.Graphics.Gpu/Memory/PhysicalMemory.cs

@@ -7,8 +7,6 @@ using Ryujinx.Memory.Range;
 using Ryujinx.Memory.Tracking;
 using System;
 using System.Collections.Generic;
-using System.Runtime.CompilerServices;
-using System.Runtime.InteropServices;
 using System.Threading;
 
 namespace Ryujinx.Graphics.Gpu.Memory
@@ -19,8 +17,6 @@ namespace Ryujinx.Graphics.Gpu.Memory
     /// </summary>
     class PhysicalMemory : IDisposable
     {
-        public const int PageSize = 0x1000;
-
         private readonly GpuContext _context;
         private IVirtualMemoryManagerTracked _cpuMemory;
         private int _referenceCount;
@@ -103,24 +99,28 @@ namespace Ryujinx.Graphics.Gpu.Memory
             if (range.Count == 1)
             {
                 var singleRange = range.GetSubRange(0);
-                return _cpuMemory.GetSpan(singleRange.Address, (int)singleRange.Size, tracked);
-            }
-            else
-            {
-                Span<byte> data = new byte[range.GetSize()];
-
-                int offset = 0;
-
-                for (int i = 0; i < range.Count; i++)
+                if (singleRange.Address != MemoryManager.PteUnmapped)
                 {
-                    var currentRange = range.GetSubRange(i);
-                    int size = (int)currentRange.Size;
-                    _cpuMemory.GetSpan(currentRange.Address, size, tracked).CopyTo(data.Slice(offset, size));
-                    offset += size;
+                    return _cpuMemory.GetSpan(singleRange.Address, (int)singleRange.Size, tracked);
                 }
-
-                return data;
             }
+
+            Span<byte> data = new byte[range.GetSize()];
+
+            int offset = 0;
+
+            for (int i = 0; i < range.Count; i++)
+            {
+                var currentRange = range.GetSubRange(i);
+                int size = (int)currentRange.Size;
+                if (currentRange.Address != MemoryManager.PteUnmapped)
+                {
+                    _cpuMemory.GetSpan(currentRange.Address, size, tracked).CopyTo(data.Slice(offset, size));
+                }
+                offset += size;
+            }
+
+            return data;
         }
 
         /// <summary>
@@ -156,11 +156,13 @@ namespace Ryujinx.Graphics.Gpu.Memory
                 int offset = 0;
                 for (int i = 0; i < range.Count; i++)
                 {
-                    MemoryRange subrange = range.GetSubRange(i);
-
-                    GetSpan(subrange.Address, (int)subrange.Size).CopyTo(memory.Span.Slice(offset, (int)subrange.Size));
-
-                    offset += (int)subrange.Size;
+                    var currentRange = range.GetSubRange(i);
+                    int size = (int)currentRange.Size;
+                    if (currentRange.Address != MemoryManager.PteUnmapped)
+                    {
+                        GetSpan(currentRange.Address, size).CopyTo(memory.Span.Slice(offset, size));
+                    }
+                    offset += size;
                 }
 
                 return new WritableRegion(new MultiRangeWritableBlock(range, this), 0, memory, tracked);
@@ -253,7 +255,10 @@ namespace Ryujinx.Graphics.Gpu.Memory
             if (range.Count == 1)
             {
                 var singleRange = range.GetSubRange(0);
-                writeCallback(singleRange.Address, data);
+                if (singleRange.Address != MemoryManager.PteUnmapped)
+                {
+                    writeCallback(singleRange.Address, data);
+                }
             }
             else
             {
@@ -263,7 +268,10 @@ namespace Ryujinx.Graphics.Gpu.Memory
                 {
                     var currentRange = range.GetSubRange(i);
                     int size = (int)currentRange.Size;
-                    writeCallback(currentRange.Address, data.Slice(offset, size));
+                    if (currentRange.Address != MemoryManager.PteUnmapped)
+                    {
+                        writeCallback(currentRange.Address, data.Slice(offset, size));
+                    }
                     offset += size;
                 }
             }
@@ -288,11 +296,20 @@ namespace Ryujinx.Graphics.Gpu.Memory
         public GpuRegionHandle BeginTracking(MultiRange range)
         {
             var cpuRegionHandles = new CpuRegionHandle[range.Count];
+            int count = 0;
 
             for (int i = 0; i < range.Count; i++)
             {
                 var currentRange = range.GetSubRange(i);
-                cpuRegionHandles[i] = _cpuMemory.BeginTracking(currentRange.Address, currentRange.Size);
+                if (currentRange.Address != MemoryManager.PteUnmapped)
+                {
+                    cpuRegionHandles[count++] = _cpuMemory.BeginTracking(currentRange.Address, currentRange.Size);
+                }
+            }
+
+            if (count != range.Count)
+            {
+                Array.Resize(ref cpuRegionHandles, count);
             }
 
             return new GpuRegionHandle(cpuRegionHandles);

Ryujinx.Graphics.Gpu/Memory/PteKind.cs

@@ -0,0 +1,268 @@
+namespace Ryujinx.Graphics.Gpu.Memory
+{
+    /// <summary>
+    /// Kind of the resource at the given memory mapping.
+    /// </summary>
+    public enum PteKind : byte
+    {
+        Invalid = 0xff,
+        Pitch = 0x00,
+        Z16 = 0x01,
+        Z162C = 0x02,
+        Z16MS22C = 0x03,
+        Z16MS42C = 0x04,
+        Z16MS82C = 0x05,
+        Z16MS162C = 0x06,
+        Z162Z = 0x07,
+        Z16MS22Z = 0x08,
+        Z16MS42Z = 0x09,
+        Z16MS82Z = 0x0a,
+        Z16MS162Z = 0x0b,
+        Z162CZ = 0x36,
+        Z16MS22CZ = 0x37,
+        Z16MS42CZ = 0x38,
+        Z16MS82CZ = 0x39,
+        Z16MS162CZ = 0x5f,
+        Z164CZ = 0x0c,
+        Z16MS24CZ = 0x0d,
+        Z16MS44CZ = 0x0e,
+        Z16MS84CZ = 0x0f,
+        Z16MS164CZ = 0x10,
+        S8Z24 = 0x11,
+        S8Z241Z = 0x12,
+        S8Z24MS21Z = 0x13,
+        S8Z24MS41Z = 0x14,
+        S8Z24MS81Z = 0x15,
+        S8Z24MS161Z = 0x16,
+        S8Z242CZ = 0x17,
+        S8Z24MS22CZ = 0x18,
+        S8Z24MS42CZ = 0x19,
+        S8Z24MS82CZ = 0x1a,
+        S8Z24MS162CZ = 0x1b,
+        S8Z242CS = 0x1c,
+        S8Z24MS22CS = 0x1d,
+        S8Z24MS42CS = 0x1e,
+        S8Z24MS82CS = 0x1f,
+        S8Z24MS162CS = 0x20,
+        S8Z244CSZV = 0x21,
+        S8Z24MS24CSZV = 0x22,
+        S8Z24MS44CSZV = 0x23,
+        S8Z24MS84CSZV = 0x24,
+        S8Z24MS164CSZV = 0x25,
+        V8Z24MS4VC12 = 0x26,
+        V8Z24MS4VC4 = 0x27,
+        V8Z24MS8VC8 = 0x28,
+        V8Z24MS8VC24 = 0x29,
+        V8Z24MS4VC121ZV = 0x2e,
+        V8Z24MS4VC41ZV = 0x2f,
+        V8Z24MS8VC81ZV = 0x30,
+        V8Z24MS8VC241ZV = 0x31,
+        V8Z24MS4VC122CS = 0x32,
+        V8Z24MS4VC42CS = 0x33,
+        V8Z24MS8VC82CS = 0x34,
+        V8Z24MS8VC242CS = 0x35,
+        V8Z24MS4VC122CZV = 0x3a,
+        V8Z24MS4VC42CZV = 0x3b,
+        V8Z24MS8VC82CZV = 0x3c,
+        V8Z24MS8VC242CZV = 0x3d,
+        V8Z24MS4VC122ZV = 0x3e,
+        V8Z24MS4VC42ZV = 0x3f,
+        V8Z24MS8VC82ZV = 0x40,
+        V8Z24MS8VC242ZV = 0x41,
+        V8Z24MS4VC124CSZV = 0x42,
+        V8Z24MS4VC44CSZV = 0x43,
+        V8Z24MS8VC84CSZV = 0x44,
+        V8Z24MS8VC244CSZV = 0x45,
+        Z24S8 = 0x46,
+        Z24S81Z = 0x47,
+        Z24S8MS21Z = 0x48,
+        Z24S8MS41Z = 0x49,
+        Z24S8MS81Z = 0x4a,
+        Z24S8MS161Z = 0x4b,
+        Z24S82CS = 0x4c,
+        Z24S8MS22CS = 0x4d,
+        Z24S8MS42CS = 0x4e,
+        Z24S8MS82CS = 0x4f,
+        Z24S8MS162CS = 0x50,
+        Z24S82CZ = 0x51,
+        Z24S8MS22CZ = 0x52,
+        Z24S8MS42CZ = 0x53,
+        Z24S8MS82CZ = 0x54,
+        Z24S8MS162CZ = 0x55,
+        Z24S84CSZV = 0x56,
+        Z24S8MS24CSZV = 0x57,
+        Z24S8MS44CSZV = 0x58,
+        Z24S8MS84CSZV = 0x59,
+        Z24S8MS164CSZV = 0x5a,
+        Z24V8MS4VC12 = 0x5b,
+        Z24V8MS4VC4 = 0x5c,
+        Z24V8MS8VC8 = 0x5d,
+        Z24V8MS8VC24 = 0x5e,
+        YUVB8C12Y = 0x60,
+        YUVB8C22Y = 0x61,
+        YUVB10C12Y = 0x62,
+        YUVB10C22Y = 0x6b,
+        YUVB12C12Y = 0x6c,
+        YUVB12C22Y = 0x6d,
+        Z24V8MS4VC121ZV = 0x63,
+        Z24V8MS4VC41ZV = 0x64,
+        Z24V8MS8VC81ZV = 0x65,
+        Z24V8MS8VC241ZV = 0x66,
+        Z24V8MS4VC122CS = 0x67,
+        Z24V8MS4VC42CS = 0x68,
+        Z24V8MS8VC82CS = 0x69,
+        Z24V8MS8VC242CS = 0x6a,
+        Z24V8MS4VC122CZV = 0x6f,
+        Z24V8MS4VC42CZV = 0x70,
+        Z24V8MS8VC82CZV = 0x71,
+        Z24V8MS8VC242CZV = 0x72,
+        Z24V8MS4VC122ZV = 0x73,
+        Z24V8MS4VC42ZV = 0x74,
+        Z24V8MS8VC82ZV = 0x75,
+        Z24V8MS8VC242ZV = 0x76,
+        Z24V8MS4VC124CSZV = 0x77,
+        Z24V8MS4VC44CSZV = 0x78,
+        Z24V8MS8VC84CSZV = 0x79,
+        Z24V8MS8VC244CSZV = 0x7a,
+        ZF32 = 0x7b,
+        ZF321Z = 0x7c,
+        ZF32MS21Z = 0x7d,
+        ZF32MS41Z = 0x7e,
+        ZF32MS81Z = 0x7f,
+        ZF32MS161Z = 0x80,
+        ZF322CS = 0x81,
+        ZF32MS22CS = 0x82,
+        ZF32MS42CS = 0x83,
+        ZF32MS82CS = 0x84,
+        ZF32MS162CS = 0x85,
+        ZF322CZ = 0x86,
+        ZF32MS22CZ = 0x87,
+        ZF32MS42CZ = 0x88,
+        ZF32MS82CZ = 0x89,
+        ZF32MS162CZ = 0x8a,
+        X8Z24X16V8S8MS4VC12 = 0x8b,
+        X8Z24X16V8S8MS4VC4 = 0x8c,
+        X8Z24X16V8S8MS8VC8 = 0x8d,
+        X8Z24X16V8S8MS8VC24 = 0x8e,
+        X8Z24X16V8S8MS4VC121CS = 0x8f,
+        X8Z24X16V8S8MS4VC41CS = 0x90,
+        X8Z24X16V8S8MS8VC81CS = 0x91,
+        X8Z24X16V8S8MS8VC241CS = 0x92,
+        X8Z24X16V8S8MS4VC121ZV = 0x97,
+        X8Z24X16V8S8MS4VC41ZV = 0x98,
+        X8Z24X16V8S8MS8VC81ZV = 0x99,
+        X8Z24X16V8S8MS8VC241ZV = 0x9a,
+        X8Z24X16V8S8MS4VC121CZV = 0x9b,
+        X8Z24X16V8S8MS4VC41CZV = 0x9c,
+        X8Z24X16V8S8MS8VC81CZV = 0x9d,
+        X8Z24X16V8S8MS8VC241CZV = 0x9e,
+        X8Z24X16V8S8MS4VC122CS = 0x9f,
+        X8Z24X16V8S8MS4VC42CS = 0xa0,
+        X8Z24X16V8S8MS8VC82CS = 0xa1,
+        X8Z24X16V8S8MS8VC242CS = 0xa2,
+        X8Z24X16V8S8MS4VC122CSZV = 0xa3,
+        X8Z24X16V8S8MS4VC42CSZV = 0xa4,
+        X8Z24X16V8S8MS8VC82CSZV = 0xa5,
+        X8Z24X16V8S8MS8VC242CSZV = 0xa6,
+        ZF32X16V8S8MS4VC12 = 0xa7,
+        ZF32X16V8S8MS4VC4 = 0xa8,
+        ZF32X16V8S8MS8VC8 = 0xa9,
+        ZF32X16V8S8MS8VC24 = 0xaa,
+        ZF32X16V8S8MS4VC121CS = 0xab,
+        ZF32X16V8S8MS4VC41CS = 0xac,
+        ZF32X16V8S8MS8VC81CS = 0xad,
+        ZF32X16V8S8MS8VC241CS = 0xae,
+        ZF32X16V8S8MS4VC121ZV = 0xb3,
+        ZF32X16V8S8MS4VC41ZV = 0xb4,
+        ZF32X16V8S8MS8VC81ZV = 0xb5,
+        ZF32X16V8S8MS8VC241ZV = 0xb6,
+        ZF32X16V8S8MS4VC121CZV = 0xb7,
+        ZF32X16V8S8MS4VC41CZV = 0xb8,
+        ZF32X16V8S8MS8VC81CZV = 0xb9,
+        ZF32X16V8S8MS8VC241CZV = 0xba,
+        ZF32X16V8S8MS4VC122CS = 0xbb,
+        ZF32X16V8S8MS4VC42CS = 0xbc,
+        ZF32X16V8S8MS8VC82CS = 0xbd,
+        ZF32X16V8S8MS8VC242CS = 0xbe,
+        ZF32X16V8S8MS4VC122CSZV = 0xbf,
+        ZF32X16V8S8MS4VC42CSZV = 0xc0,
+        ZF32X16V8S8MS8VC82CSZV = 0xc1,
+        ZF32X16V8S8MS8VC242CSZV = 0xc2,
+        ZF32X24S8 = 0xc3,
+        ZF32X24S81CS = 0xc4,
+        ZF32X24S8MS21CS = 0xc5,
+        ZF32X24S8MS41CS = 0xc6,
+        ZF32X24S8MS81CS = 0xc7,
+        ZF32X24S8MS161CS = 0xc8,
+        ZF32X24S82CSZV = 0xce,
+        ZF32X24S8MS22CSZV = 0xcf,
+        ZF32X24S8MS42CSZV = 0xd0,
+        ZF32X24S8MS82CSZV = 0xd1,
+        ZF32X24S8MS162CSZV = 0xd2,
+        ZF32X24S82CS = 0xd3,
+        ZF32X24S8MS22CS = 0xd4,
+        ZF32X24S8MS42CS = 0xd5,
+        ZF32X24S8MS82CS = 0xd6,
+        ZF32X24S8MS162CS = 0xd7,
+        S8 = 0x2a,
+        S82S = 0x2b,
+        Generic16Bx2 = 0xfe,
+        C322C = 0xd8,
+        C322CBR = 0xd9,
+        C322CBA = 0xda,
+        C322CRA = 0xdb,
+        C322BRA = 0xdc,
+        C32MS22C = 0xdd,
+        C32MS22CBR = 0xde,
+        C32MS24CBRA = 0xcc,
+        C32MS42C = 0xdf,
+        C32MS42CBR = 0xe0,
+        C32MS42CBA = 0xe1,
+        C32MS42CRA = 0xe2,
+        C32MS42BRA = 0xe3,
+        C32MS44CBRA = 0x2c,
+        C32MS8MS162C = 0xe4,
+        C32MS8MS162CRA = 0xe5,
+        C642C = 0xe6,
+        C642CBR = 0xe7,
+        C642CBA = 0xe8,
+        C642CRA = 0xe9,
+        C642BRA = 0xea,
+        C64MS22C = 0xeb,
+        C64MS22CBR = 0xec,
+        C64MS24CBRA = 0xcd,
+        C64MS42C = 0xed,
+        C64MS42CBR = 0xee,
+        C64MS42CBA = 0xef,
+        C64MS42CRA = 0xf0,
+        C64MS42BRA = 0xf1,
+        C64MS44CBRA = 0x2d,
+        C64MS8MS162C = 0xf2,
+        C64MS8MS162CRA = 0xf3,
+        C1282C = 0xf4,
+        C1282CR = 0xf5,
+        C128MS22C = 0xf6,
+        C128MS22CR = 0xf7,
+        C128MS42C = 0xf8,
+        C128MS42CR = 0xf9,
+        C128MS8MS162C = 0xfa,
+        C128MS8MS162CR = 0xfb,
+        X8C24 = 0xfc,
+        PitchNoSwizzle = 0xfd,
+        SmSkedMessage = 0xca,
+        SmHostMessage = 0xcb
+    }
+
+    static class PteKindExtensions
+    {
+        /// <summary>
+        /// Checks if the kind is pitch.
+        /// </summary>
+        /// <param name="kind">Kind to check</param>
+        /// <returns>True if pitch, false otherwise</returns>
+        public static bool IsPitch(this PteKind kind)
+        {
+            return kind == PteKind.Pitch || kind == PteKind.PitchNoSwizzle;
+        }
+    }
+}

Ryujinx.Graphics.Gpu/Memory/ResourceName.cs

@@ -1,13 +0,0 @@
-namespace Ryujinx.Graphics.Gpu.Memory
-{
-    /// <summary>
-    /// Name of a GPU resource.
-    /// </summary>
-    public enum ResourceName
-    {
-        Buffer,
-        Texture,
-        TexturePool,
-        SamplerPool
-    }
-}

Ryujinx.Graphics.Gpu/Shader/Cache/CacheHelper.cs

@@ -113,7 +113,7 @@ namespace Ryujinx.Graphics.Gpu.Shader.Cache
                 dataSpan[i++] = hash;
             }
 
-            manifestHeader.UpdateChecksum(data.AsSpan().Slice(Unsafe.SizeOf<CacheManifestHeader>()));
+            manifestHeader.UpdateChecksum(data.AsSpan(Unsafe.SizeOf<CacheManifestHeader>()));
 
             MemoryMarshal.Write(data, ref manifestHeader);
 
@@ -447,12 +447,12 @@ namespace Ryujinx.Graphics.Gpu.Shader.Cache
 
                 if (cb1DataAddress != 0 && cb1DataSize != 0)
                 {
-                    memoryManager.Physical.GetSpan(cb1DataAddress, cb1DataSize).CopyTo(code.AsSpan().Slice(size, cb1DataSize));
+                    memoryManager.Physical.GetSpan(cb1DataAddress, cb1DataSize).CopyTo(code.AsSpan(size, cb1DataSize));
                 }
 
                 if (translatorContext2 != null)
                 {
-                    memoryManager.GetSpan(translatorContext2.Address, sizeA).CopyTo(code.AsSpan().Slice(size + cb1DataSize, sizeA));
+                    memoryManager.GetSpan(translatorContext2.Address, sizeA).CopyTo(code.AsSpan(size + cb1DataSize, sizeA));
                 }
 
                 GuestGpuAccessorHeader gpuAccessorHeader = CreateGuestGpuAccessorCache(context.GpuAccessor);

Ryujinx.Graphics.Gpu/Shader/ShaderCache.cs

@@ -40,7 +40,7 @@ namespace Ryujinx.Graphics.Gpu.Shader
         /// <summary>
         /// Version of the codegen (to be changed when codegen or guest format change).
         /// </summary>
-        private const ulong ShaderCodeGenVersion = 3132;
+        private const ulong ShaderCodeGenVersion = 3054;
 
         // Progress reporting helpers
         private volatile int _shaderCount;
@@ -206,7 +206,7 @@ namespace Ryujinx.Graphics.Gpu.Shader
                                     program = new ShaderProgram(entry.Header.Stage, "");
                                     shaderProgramInfo = hostShaderEntries[0].ToShaderProgramInfo();
 
-                                    byte[] code = entry.Code.AsSpan().Slice(0, entry.Header.Size - entry.Header.Cb1DataSize).ToArray();
+                                    byte[] code = entry.Code.AsSpan(0, entry.Header.Size - entry.Header.Cb1DataSize).ToArray();
 
                                     ShaderCodeHolder shader = new ShaderCodeHolder(program, shaderProgramInfo, code);
 
@@ -244,7 +244,7 @@ namespace Ryujinx.Graphics.Gpu.Shader
                                             return true; // Exit early, the decoding step failed.
                                         }
 
-                                        byte[] code = entry.Code.AsSpan().Slice(0, entry.Header.Size - entry.Header.Cb1DataSize).ToArray();
+                                        byte[] code = entry.Code.AsSpan(0, entry.Header.Size - entry.Header.Cb1DataSize).ToArray();
 
                                         ShaderCodeHolder shader = new ShaderCodeHolder(program, shaderProgramInfo, code);
 
@@ -394,8 +394,8 @@ namespace Ryujinx.Graphics.Gpu.Shader
                                         }
 
                                         // NOTE: Vertex B comes first in the shader cache.
-                                        byte[] code = entry.Code.AsSpan().Slice(0, entry.Header.Size - entry.Header.Cb1DataSize).ToArray();
-                                        byte[] code2 = entry.Header.SizeA != 0 ? entry.Code.AsSpan().Slice(entry.Header.Size, entry.Header.SizeA).ToArray() : null;
+                                        byte[] code = entry.Code.AsSpan(0, entry.Header.Size - entry.Header.Cb1DataSize).ToArray();
+                                        byte[] code2 = entry.Header.SizeA != 0 ? entry.Code.AsSpan(entry.Header.Size, entry.Header.SizeA).ToArray() : null;
 
                                         shaders[i] = new ShaderCodeHolder(program, shaderProgramInfo, code, code2);
 

Ryujinx.Graphics.Nvdec/Image/SurfaceReader.cs

@@ -16,7 +16,7 @@ namespace Ryujinx.Graphics.Nvdec.Image
             int width = surface.Width;
             int height = surface.Height;
             int stride = surface.Stride;
- 
+
             ReadOnlySpan<byte> luma = gmm.DeviceGetSpan(lumaOffset, GetBlockLinearSize(width, height, 1));
 
             ReadLuma(surface.YPlane.AsSpan(), luma, stride, width, height);

Ryujinx.Graphics.OpenGL/FormatTable.cs

@@ -6,15 +6,15 @@ namespace Ryujinx.Graphics.OpenGL
 {
     struct FormatTable
     {
-        private static FormatInfo[] Table;
-        private static SizedInternalFormat[] TableImage;
+        private static FormatInfo[] _table;
+        private static SizedInternalFormat[] _tableImage;
 
         static FormatTable()
         {
             int tableSize = Enum.GetNames<Format>().Length;
 
-            Table = new FormatInfo[tableSize];
-            TableImage = new SizedInternalFormat[tableSize];
+            _table = new FormatInfo[tableSize];
+            _tableImage = new SizedInternalFormat[tableSize];
 
             Add(Format.R8Unorm,             new FormatInfo(1, true,  false, All.R8,                PixelFormat.Red,            PixelType.UnsignedByte));
             Add(Format.R8Snorm,             new FormatInfo(1, true,  false, All.R8Snorm,           PixelFormat.Red,            PixelType.Byte));
@@ -66,7 +66,7 @@ namespace Ryujinx.Graphics.OpenGL
             Add(Format.R32G32B32A32Sint,    new FormatInfo(4, false, false, All.Rgba32i,           PixelFormat.RgbaInteger,    PixelType.Int));
             Add(Format.S8Uint,              new FormatInfo(1, false, false, All.StencilIndex8,     PixelFormat.StencilIndex,   PixelType.UnsignedByte));
             Add(Format.D16Unorm,            new FormatInfo(1, false, false, All.DepthComponent16,  PixelFormat.DepthComponent, PixelType.UnsignedShort));
-            Add(Format.D24X8Unorm,          new FormatInfo(1, false, false, All.DepthComponent24,  PixelFormat.DepthComponent, PixelType.UnsignedInt));
+            Add(Format.S8UintD24Unorm,      new FormatInfo(1, false, false, All.Depth24Stencil8,   PixelFormat.DepthStencil,   PixelType.UnsignedInt248));
             Add(Format.D32Float,            new FormatInfo(1, false, false, All.DepthComponent32f, PixelFormat.DepthComponent, PixelType.Float));
             Add(Format.D24UnormS8Uint,      new FormatInfo(1, false, false, All.Depth24Stencil8,   PixelFormat.DepthStencil,   PixelType.UnsignedInt248));
             Add(Format.D32FloatS8Uint,      new FormatInfo(1, false, false, All.Depth32fStencil8,  PixelFormat.DepthStencil,   PixelType.Float32UnsignedInt248Rev));
@@ -218,22 +218,22 @@ namespace Ryujinx.Graphics.OpenGL
 
         private static void Add(Format format, FormatInfo info)
         {
-            Table[(int)format] = info;
+            _table[(int)format] = info;
         }
 
         private static void Add(Format format, SizedInternalFormat sif)
         {
-            TableImage[(int)format] = sif;
+            _tableImage[(int)format] = sif;
         }
 
         public static FormatInfo GetFormatInfo(Format format)
         {
-            return Table[(int)format];
+            return _table[(int)format];
         }
 
         public static SizedInternalFormat GetImageFormat(Format format)
         {
-            return TableImage[(int)format];
+            return _tableImage[(int)format];
         }
     }
 }

Ryujinx.Graphics.OpenGL/Framebuffer.cs

@@ -127,14 +127,13 @@ namespace Ryujinx.Graphics.OpenGL
         private static bool IsPackedDepthStencilFormat(Format format)
         {
             return format == Format.D24UnormS8Uint ||
-                   format == Format.D32FloatS8Uint;
+                   format == Format.D32FloatS8Uint ||
+                   format == Format.S8UintD24Unorm;
         }
 
         private static bool IsDepthOnlyFormat(Format format)
         {
-            return format == Format.D16Unorm ||
-                   format == Format.D24X8Unorm ||
-                   format == Format.D32Float;
+            return format == Format.D16Unorm || format == Format.D32Float;
         }
 
         public void Dispose()

Ryujinx.Graphics.OpenGL/Image/FormatConverter.cs

@@ -0,0 +1,149 @@
+using System;
+using System.Numerics;
+using System.Runtime.InteropServices;
+using System.Runtime.Intrinsics;
+using System.Runtime.Intrinsics.X86;
+
+namespace Ryujinx.Graphics.OpenGL.Image
+{
+    static class FormatConverter
+    {
+        public unsafe static byte[] ConvertS8D24ToD24S8(ReadOnlySpan<byte> data)
+        {
+            byte[] output = new byte[data.Length];
+
+            int start = 0;
+
+            if (Avx2.IsSupported)
+            {
+                var mask = Vector256.Create(
+                    (byte)3, (byte)0, (byte)1, (byte)2,
+                    (byte)7, (byte)4, (byte)5, (byte)6,
+                    (byte)11, (byte)8, (byte)9, (byte)10,
+                    (byte)15, (byte)12, (byte)13, (byte)14,
+                    (byte)19, (byte)16, (byte)17, (byte)18,
+                    (byte)23, (byte)20, (byte)21, (byte)22,
+                    (byte)27, (byte)24, (byte)25, (byte)26,
+                    (byte)31, (byte)28, (byte)29, (byte)30);
+
+                int sizeAligned = data.Length & ~31;
+
+                fixed (byte* pInput = data, pOutput = output)
+                {
+                    for (uint i = 0; i < sizeAligned; i += 32)
+                    {
+                        var dataVec = Avx.LoadVector256(pInput + i);
+
+                        dataVec = Avx2.Shuffle(dataVec, mask);
+
+                        Avx.Store(pOutput + i, dataVec);
+                    }
+                }
+
+                start = sizeAligned;
+            }
+            else if (Ssse3.IsSupported)
+            {
+                var mask = Vector128.Create(
+                    (byte)3, (byte)0, (byte)1, (byte)2,
+                    (byte)7, (byte)4, (byte)5, (byte)6,
+                    (byte)11, (byte)8, (byte)9, (byte)10,
+                    (byte)15, (byte)12, (byte)13, (byte)14);
+
+                int sizeAligned = data.Length & ~15;
+
+                fixed (byte* pInput = data, pOutput = output)
+                {
+                    for (uint i = 0; i < sizeAligned; i += 16)
+                    {
+                        var dataVec = Sse2.LoadVector128(pInput + i);
+
+                        dataVec = Ssse3.Shuffle(dataVec, mask);
+
+                        Sse2.Store(pOutput + i, dataVec);
+                    }
+                }
+
+                start = sizeAligned;
+            }
+
+            var outSpan = MemoryMarshal.Cast<byte, uint>(output);
+            var dataSpan = MemoryMarshal.Cast<byte, uint>(data);
+            for (int i = start / sizeof(uint); i < dataSpan.Length; i++)
+            {
+                outSpan[i] = BitOperations.RotateLeft(dataSpan[i], 8);
+            }
+
+            return output;
+        }
+
+        public unsafe static byte[] ConvertD24S8ToS8D24(ReadOnlySpan<byte> data)
+        {
+            byte[] output = new byte[data.Length];
+
+            int start = 0;
+
+            if (Avx2.IsSupported)
+            {
+                var mask = Vector256.Create(
+                    (byte)1, (byte)2, (byte)3, (byte)0,
+                    (byte)5, (byte)6, (byte)7, (byte)4,
+                    (byte)9, (byte)10, (byte)11, (byte)8,
+                    (byte)13, (byte)14, (byte)15, (byte)12,
+                    (byte)17, (byte)18, (byte)19, (byte)16,
+                    (byte)21, (byte)22, (byte)23, (byte)20,
+                    (byte)25, (byte)26, (byte)27, (byte)24,
+                    (byte)29, (byte)30, (byte)31, (byte)28);
+
+                int sizeAligned = data.Length & ~31;
+
+                fixed (byte* pInput = data, pOutput = output)
+                {
+                    for (uint i = 0; i < sizeAligned; i += 32)
+                    {
+                        var dataVec = Avx.LoadVector256(pInput + i);
+
+                        dataVec = Avx2.Shuffle(dataVec, mask);
+
+                        Avx.Store(pOutput + i, dataVec);
+                    }
+                }
+
+                start = sizeAligned;
+            }
+            else if (Ssse3.IsSupported)
+            {
+                var mask = Vector128.Create(
+                    (byte)1, (byte)2, (byte)3, (byte)0,
+                    (byte)5, (byte)6, (byte)7, (byte)4,
+                    (byte)9, (byte)10, (byte)11, (byte)8,
+                    (byte)13, (byte)14, (byte)15, (byte)12);
+
+                int sizeAligned = data.Length & ~15;
+
+                fixed (byte* pInput = data, pOutput = output)
+                {
+                    for (uint i = 0; i < sizeAligned; i += 16)
+                    {
+                        var dataVec = Sse2.LoadVector128(pInput + i);
+
+                        dataVec = Ssse3.Shuffle(dataVec, mask);
+
+                        Sse2.Store(pOutput + i, dataVec);
+                    }
+                }
+
+                start = sizeAligned;
+            }
+
+            var outSpan = MemoryMarshal.Cast<byte, uint>(output);
+            var dataSpan = MemoryMarshal.Cast<byte, uint>(data);
+            for (int i = start / sizeof(uint); i < dataSpan.Length; i++)
+            {
+                outSpan[i] = BitOperations.RotateRight(dataSpan[i], 8);
+            }
+
+            return output;
+        }
+    }
+}

Ryujinx.Graphics.OpenGL/Image/TextureCopy.cs

@@ -291,7 +291,7 @@ namespace Ryujinx.Graphics.OpenGL.Image
 
         private static ClearBufferMask GetMask(Format format)
         {
-            if (format == Format.D24UnormS8Uint || format == Format.D32FloatS8Uint)
+            if (format == Format.D24UnormS8Uint || format == Format.D32FloatS8Uint || format == Format.S8UintD24Unorm)
             {
                 return ClearBufferMask.DepthBufferBit | ClearBufferMask.StencilBufferBit;
             }
@@ -311,9 +311,7 @@ namespace Ryujinx.Graphics.OpenGL.Image
 
         private static bool IsDepthOnly(Format format)
         {
-            return format == Format.D16Unorm   ||
-                   format == Format.D24X8Unorm ||
-                   format == Format.D32Float;
+            return format == Format.D16Unorm || format == Format.D32Float;
         }
 
         public TextureView BgraSwap(TextureView from)

Ryujinx.Graphics.OpenGL/Image/TextureView.cs

@@ -140,9 +140,11 @@ namespace Ryujinx.Graphics.OpenGL.Image
                 size += Info.GetMipSize(level);
             }
 
+            ReadOnlySpan<byte> data;
+
             if (HwCapabilities.UsePersistentBufferForFlush)
             {
-                return _renderer.PersistentBuffers.Default.GetTextureData(this, size);
+                data = _renderer.PersistentBuffers.Default.GetTextureData(this, size);
             }
             else
             {
@@ -150,8 +152,15 @@ namespace Ryujinx.Graphics.OpenGL.Image
 
                 WriteTo(target);
 
-                return new ReadOnlySpan<byte>(target.ToPointer(), size);
+                data = new ReadOnlySpan<byte>(target.ToPointer(), size);
             }
+
+            if (Format == Format.S8UintD24Unorm)
+            {
+                data = FormatConverter.ConvertD24S8ToS8D24(data);
+            }
+
+            return data;
         }
 
         public unsafe ReadOnlySpan<byte> GetData(int layer, int level)
@@ -285,6 +294,11 @@ namespace Ryujinx.Graphics.OpenGL.Image
 
         public void SetData(ReadOnlySpan<byte> data)
         {
+            if (Format == Format.S8UintD24Unorm)
+            {
+                data = FormatConverter.ConvertS8D24ToD24S8(data);
+            }
+
             unsafe
             {
                 fixed (byte* ptr = data)
@@ -296,6 +310,11 @@ namespace Ryujinx.Graphics.OpenGL.Image
 
         public void SetData(ReadOnlySpan<byte> data, int layer, int level)
         {
+            if (Format == Format.S8UintD24Unorm)
+            {
+                data = FormatConverter.ConvertS8D24ToD24S8(data);
+            }
+
             unsafe
             {
                 fixed (byte* ptr = data)

Ryujinx.Graphics.OpenGL/Pipeline.cs

@@ -1478,6 +1478,11 @@ namespace Ryujinx.Graphics.OpenGL
             _currentComponentMasks |= componentMaskAtIndex;
         }
 
+        public void RestoreClipControl()
+        {
+            GL.ClipControl(_clipOrigin, _clipDepthMode);
+        }
+
         public void RestoreScissor0Enable()
         {
             if ((_scissorEnables & 1u) != 0)
@@ -1494,6 +1499,11 @@ namespace Ryujinx.Graphics.OpenGL
             }
         }
 
+        public void RestoreViewport0()
+        {
+            GL.ViewportArray(0, 1, _viewportArray);
+        }
+
         public bool TryHostConditionalRendering(ICounterEvent value, ulong compare, bool isEqual)
         {
             if (value is CounterQueueEvent)

Ryujinx.Graphics.OpenGL/Program.cs

@@ -124,7 +124,7 @@ namespace Ryujinx.Graphics.OpenGL
 
             GL.GetProgramBinary(Handle, size, out _, out BinaryFormat binFormat, data);
 
-            BinaryPrimitives.WriteInt32LittleEndian(data.AsSpan().Slice(size, 4), (int)binFormat);
+            BinaryPrimitives.WriteInt32LittleEndian(data.AsSpan(size, 4), (int)binFormat);
 
             return data;
         }

Ryujinx.Graphics.OpenGL/Window.cs

@@ -27,11 +27,12 @@ namespace Ryujinx.Graphics.OpenGL
         {
             GL.Disable(EnableCap.FramebufferSrgb);
 
-            CopyTextureToFrameBufferRGB(0, GetCopyFramebufferHandleLazy(), (TextureView)texture, crop);
+            CopyTextureToFrameBufferRGB(0, GetCopyFramebufferHandleLazy(), (TextureView)texture, crop, swapBuffersCallback);
 
             GL.Enable(EnableCap.FramebufferSrgb);
 
-            swapBuffersCallback();
+            // Restore unpack alignment to 4, as performance overlays such as RTSS may change this to load their resources.
+            GL.PixelStore(PixelStoreParameter.UnpackAlignment, 4);
         }
 
         public void SetSize(int width, int height)
@@ -40,7 +41,7 @@ namespace Ryujinx.Graphics.OpenGL
             _height = height;
         }
 
-        private void CopyTextureToFrameBufferRGB(int drawFramebuffer, int readFramebuffer, TextureView view, ImageCrop crop)
+        private void CopyTextureToFrameBufferRGB(int drawFramebuffer, int readFramebuffer, TextureView view, ImageCrop crop, Action swapBuffersCallback)
         {
             (int oldDrawFramebufferHandle, int oldReadFramebufferHandle) = ((Pipeline)_renderer.Pipeline).GetBoundFramebuffers();
 
@@ -139,11 +140,20 @@ namespace Ryujinx.Graphics.OpenGL
                 ((Pipeline)_renderer.Pipeline).RestoreComponentMask(i);
             }
 
+            // Set clip control, viewport and the framebuffer to the output to placate overlays and OBS capture.
+            GL.ClipControl(ClipOrigin.LowerLeft, ClipDepthMode.NegativeOneToOne);
+            GL.Viewport(0, 0, _width, _height);
+            GL.BindFramebuffer(FramebufferTarget.Framebuffer, drawFramebuffer);
+
+            swapBuffersCallback();
+
             GL.BindFramebuffer(FramebufferTarget.ReadFramebuffer, oldReadFramebufferHandle);
             GL.BindFramebuffer(FramebufferTarget.DrawFramebuffer, oldDrawFramebufferHandle);
 
+            ((Pipeline)_renderer.Pipeline).RestoreClipControl();
             ((Pipeline)_renderer.Pipeline).RestoreScissor0Enable();
             ((Pipeline)_renderer.Pipeline).RestoreRasterizerDiscard();
+            ((Pipeline)_renderer.Pipeline).RestoreViewport0();
 
             if (viewConverted != view)
             {

Ryujinx.Graphics.Shader/Decoders/Decoder.cs

@@ -308,7 +308,8 @@ namespace Ryujinx.Graphics.Shader.Decoders
                     int attr = offset + elemIndex * 4;
                     if (attr >= AttributeConsts.UserAttributeBase && attr < AttributeConsts.UserAttributeEnd)
                     {
-                        int index = (attr - AttributeConsts.UserAttributeBase) / 16;
+                        int userAttr = attr - AttributeConsts.UserAttributeBase;
+                        int index = userAttr / 16;
 
                         if (isStore)
                         {
@@ -316,7 +317,7 @@ namespace Ryujinx.Graphics.Shader.Decoders
                         }
                         else
                         {
-                            config.SetInputUserAttribute(index, perPatch);
+                            config.SetInputUserAttribute(index, (userAttr >> 2) & 3, perPatch);
                         }
                     }
 

Ryujinx.Graphics.Shader/Translation/ShaderConfig.cs

@@ -54,6 +54,11 @@ namespace Ryujinx.Graphics.Shader.Translation
         private int _nextUsedInputAttributes;
         private int _thisUsedInputAttributes;
 
+        public UInt128 NextInputAttributesComponents { get; private set; }
+        public UInt128 ThisInputAttributesComponents { get; private set; }
+        public UInt128 NextInputAttributesPerPatchComponents { get; private set; }
+        public UInt128 ThisInputAttributesPerPatchComponents { get; private set; }
+
         private int _usedConstantBuffers;
         private int _usedStorageBuffers;
         private int _usedStorageBuffersWrite;
@@ -227,11 +232,12 @@ namespace Ryujinx.Graphics.Shader.Translation
             UsedOutputAttributes |= 1 << index;
         }
 
-        public void SetInputUserAttribute(int index, bool perPatch)
+        public void SetInputUserAttribute(int index, int component, bool perPatch)
         {
             if (perPatch)
             {
                 UsedInputAttributesPerPatch |= 1 << index;
+                ThisInputAttributesPerPatchComponents |= UInt128.Pow2(index * 4 + component);
             }
             else
             {
@@ -239,6 +245,7 @@ namespace Ryujinx.Graphics.Shader.Translation
 
                 UsedInputAttributes |= mask;
                 _thisUsedInputAttributes |= mask;
+                ThisInputAttributesComponents |= UInt128.Pow2(index * 4 + component);
             }
         }
 
@@ -256,6 +263,8 @@ namespace Ryujinx.Graphics.Shader.Translation
 
         public void MergeFromtNextStage(ShaderConfig config)
         {
+            NextInputAttributesComponents = config.ThisInputAttributesComponents;
+            NextInputAttributesPerPatchComponents = config.ThisInputAttributesPerPatchComponents;
             NextUsesFixedFuncAttributes = config.UsedFeatures.HasFlag(FeatureFlags.FixedFuncAttr);
             MergeOutputUserAttributes(config.UsedInputAttributes, config.UsedInputAttributesPerPatch);
         }
@@ -319,6 +328,7 @@ namespace Ryujinx.Graphics.Shader.Translation
         public void SetAllInputUserAttributes()
         {
             UsedInputAttributes |= Constants.AllAttributesMask;
+            ThisInputAttributesComponents |= ~UInt128.Zero >> (128 - Constants.MaxAttributes * 4);
         }
 
         public void SetAllOutputUserAttributes()

Ryujinx.Graphics.Shader/Translation/Translator.cs

@@ -214,24 +214,31 @@ namespace Ryujinx.Graphics.Shader.Translation
                 InitializeOutput(context, AttributeConsts.PositionX, perPatch: false);
             }
 
-            int usedAttributes = context.Config.UsedOutputAttributes;
-            while (usedAttributes != 0)
+            UInt128 usedAttributes = context.Config.NextInputAttributesComponents;
+            while (usedAttributes != UInt128.Zero)
             {
-                int index = BitOperations.TrailingZeroCount(usedAttributes);
+                int index = usedAttributes.TrailingZeroCount();
+                int vecIndex = index / 4;
 
-                InitializeOutput(context, AttributeConsts.UserAttributeBase + index * 16, perPatch: false);
+                usedAttributes &= ~UInt128.Pow2(index);
 
-                usedAttributes &= ~(1 << index);
+                // We don't need to initialize passthrough attributes.
+                if ((context.Config.PassthroughAttributes & (1 << vecIndex)) != 0)
+                {
+                    continue;
+                }
+
+                InitializeOutputComponent(context, AttributeConsts.UserAttributeBase + index * 4, perPatch: false);
             }
 
-            int usedAttributesPerPatch = context.Config.UsedOutputAttributesPerPatch;
-            while (usedAttributesPerPatch != 0)
+            UInt128 usedAttributesPerPatch = context.Config.NextInputAttributesPerPatchComponents;
+            while (usedAttributesPerPatch != UInt128.Zero)
             {
-                int index = BitOperations.TrailingZeroCount(usedAttributesPerPatch);
+                int index = usedAttributesPerPatch.TrailingZeroCount();
 
-                InitializeOutput(context, AttributeConsts.UserAttributeBase + index * 16, perPatch: true);
+                InitializeOutputComponent(context, AttributeConsts.UserAttributeBase + index * 4, perPatch: true);
 
-                usedAttributesPerPatch &= ~(1 << index);
+                usedAttributesPerPatch &= ~UInt128.Pow2(index);
             }
 
             if (config.NextUsesFixedFuncAttributes)
@@ -260,6 +267,12 @@ namespace Ryujinx.Graphics.Shader.Translation
             }
         }
 
+        private static void InitializeOutputComponent(EmitterContext context, int attrOffset, bool perPatch)
+        {
+            int c = (attrOffset >> 2) & 3;
+            context.Copy(perPatch ? AttributePerPatch(attrOffset) : Attribute(attrOffset), ConstF(c == 3 ? 1f : 0f));
+        }
+
         private static void EmitOps(EmitterContext context, Block block)
         {
             for (int opIndex = 0; opIndex < block.OpCodes.Count; opIndex++)

Ryujinx.Graphics.Shader/Translation/UInt128.cs

@@ -0,0 +1,112 @@
+using System;
+using System.Numerics;
+
+namespace Ryujinx.Graphics.Shader.Translation
+{
+    struct UInt128 : IEquatable<UInt128>
+    {
+        public static UInt128 Zero => new UInt128() { _v0 = 0, _v1 = 0 };
+
+        private ulong _v0;
+        private ulong _v1;
+
+        public UInt128(ulong low, ulong high)
+        {
+            _v0 = low;
+            _v1 = high;
+        }
+
+        public int TrailingZeroCount()
+        {
+            int count = BitOperations.TrailingZeroCount(_v0);
+            if (count == 64)
+            {
+                count += BitOperations.TrailingZeroCount(_v1);
+            }
+
+            return count;
+        }
+
+        public static UInt128 Pow2(int x)
+        {
+            if (x >= 64)
+            {
+                return new UInt128(0, 1UL << (x - 64));
+            }
+
+            return new UInt128(1UL << x, 0);
+        }
+
+        public static UInt128 operator ~(UInt128 x)
+        {
+            return new UInt128(~x._v0, ~x._v1);
+        }
+
+        public static UInt128 operator &(UInt128 x, UInt128 y)
+        {
+            return new UInt128(x._v0 & y._v0, x._v1 & y._v1);
+        }
+
+        public static UInt128 operator |(UInt128 x, UInt128 y)
+        {
+            return new UInt128(x._v0 | y._v0, x._v1 | y._v1);
+        }
+
+        public static UInt128 operator <<(UInt128 x, int shift)
+        {
+            if (shift == 0)
+            {
+                return new UInt128(x._v0, x._v1);
+            }
+            else if (shift >= 64)
+            {
+                return new UInt128(0, x._v0 << (shift - 64));
+            }
+
+            ulong shiftOut = x._v0 >> (64 - shift);
+
+            return new UInt128(x._v0 << shift, (x._v1 << shift) | shiftOut);
+        }
+
+        public static UInt128 operator >>(UInt128 x, int shift)
+        {
+            if (shift == 0)
+            {
+                return new UInt128(x._v0, x._v1);
+            }
+            else if (shift >= 64)
+            {
+                return new UInt128(x._v1 >> (shift - 64), 0);
+            }
+
+            ulong shiftOut = x._v1 & ((1UL << shift) - 1);
+
+            return new UInt128((x._v0 >> shift) | (shiftOut << (64 - shift)), x._v1 >> shift);
+        }
+
+        public static bool operator ==(UInt128 x, UInt128 y)
+        {
+            return x.Equals(y);
+        }
+
+        public static bool operator !=(UInt128 x, UInt128 y)
+        {
+            return !x.Equals(y);
+        }
+
+        public override bool Equals(object obj)
+        {
+            return obj is UInt128 other && Equals(other);
+        }
+
+        public bool Equals(UInt128 other)
+        {
+            return _v0 == other._v0 && _v1 == other._v1;
+        }
+
+        public override int GetHashCode()
+        {
+            return HashCode.Combine(_v0, _v1);
+        }
+    }
+}

Ryujinx.HLE/HOS/ApplicationLoader.cs

@@ -84,7 +84,10 @@ namespace Ryujinx.HLE.HOS
 
             MetaLoader metaData = ReadNpdm(codeFs);
 
-            _device.Configuration.VirtualFileSystem.ModLoader.CollectMods(new[] { TitleId }, _device.Configuration.VirtualFileSystem.ModLoader.GetModsBasePath());
+            _device.Configuration.VirtualFileSystem.ModLoader.CollectMods(
+                new[] { TitleId }, 
+                _device.Configuration.VirtualFileSystem.ModLoader.GetModsBasePath(), 
+                _device.Configuration.VirtualFileSystem.ModLoader.GetSdModsBasePath());
 
             if (TitleId != 0)
             {
@@ -388,7 +391,10 @@ namespace Ryujinx.HLE.HOS
 
             MetaLoader metaData = ReadNpdm(codeFs);
 
-            _device.Configuration.VirtualFileSystem.ModLoader.CollectMods(_device.Configuration.ContentManager.GetAocTitleIds().Prepend(TitleId), _device.Configuration.VirtualFileSystem.ModLoader.GetModsBasePath());
+            _device.Configuration.VirtualFileSystem.ModLoader.CollectMods(
+                _device.Configuration.ContentManager.GetAocTitleIds().Prepend(TitleId), 
+                _device.Configuration.VirtualFileSystem.ModLoader.GetModsBasePath(), 
+                _device.Configuration.VirtualFileSystem.ModLoader.GetSdModsBasePath());
 
             if (controlNca != null)
             {
@@ -481,14 +487,14 @@ namespace Ryujinx.HLE.HOS
 
                 if (result.IsSuccess() && bytesRead == controlData.ByteSpan.Length)
                 {
-                    titleName = controlData.Value.Titles[(int)device.System.State.DesiredTitleLanguage].Name.ToString();
+                    titleName = controlData.Value.Title[(int)device.System.State.DesiredTitleLanguage].NameString.ToString();
 
                     if (string.IsNullOrWhiteSpace(titleName))
                     {
-                        titleName = controlData.Value.Titles.ToArray().FirstOrDefault(x => x.Name[0] != 0).Name.ToString();
+                        titleName = controlData.Value.Title.ItemsRo.ToArray().FirstOrDefault(x => x.Name[0] != 0).NameString.ToString();
                     }
 
-                    displayVersion = controlData.Value.DisplayVersion.ToString();
+                    displayVersion = controlData.Value.DisplayVersionString.ToString();
                 }
             }
             else
@@ -615,20 +621,20 @@ namespace Ryujinx.HLE.HOS
 
                                 ref ApplicationControlProperty nacp = ref ControlData.Value;
 
-                                programInfo.Name = nacp.Titles[(int)_device.System.State.DesiredTitleLanguage].Name.ToString();
+                                programInfo.Name = nacp.Title[(int)_device.System.State.DesiredTitleLanguage].NameString.ToString();
 
                                 if (string.IsNullOrWhiteSpace(programInfo.Name))
                                 {
-                                    programInfo.Name = nacp.Titles.ToArray().FirstOrDefault(x => x.Name[0] != 0).Name.ToString();
+                                    programInfo.Name = nacp.Title.ItemsRo.ToArray().FirstOrDefault(x => x.Name[0] != 0).NameString.ToString();
                                 }
 
                                 if (nacp.PresenceGroupId != 0)
                                 {
                                     programInfo.ProgramId = nacp.PresenceGroupId;
                                 }
-                                else if (nacp.SaveDataOwnerId.Value != 0)
+                                else if (nacp.SaveDataOwnerId != 0)
                                 {
-                                    programInfo.ProgramId = nacp.SaveDataOwnerId.Value;
+                                    programInfo.ProgramId = nacp.SaveDataOwnerId;
                                 }
                                 else if (nacp.AddOnContentBaseId != 0)
                                 {
@@ -776,14 +782,14 @@ namespace Ryujinx.HLE.HOS
                 // The set sizes don't actually matter as long as they're non-zero because we use directory savedata.
                 control.UserAccountSaveDataSize = 0x4000;
                 control.UserAccountSaveDataJournalSize = 0x4000;
-                control.SaveDataOwnerId = applicationId;
+                control.SaveDataOwnerId = applicationId.Value;
 
                 Logger.Warning?.Print(LogClass.Application,
                     "No control file was found for this game. Using a dummy one instead. This may cause inaccuracies in some games.");
             }
 
             HorizonClient hos = _device.System.LibHacHorizonManager.RyujinxClient;
-            Result resultCode = hos.Fs.EnsureApplicationCacheStorage(out _, out _, applicationId, ref control);
+            Result resultCode = hos.Fs.EnsureApplicationCacheStorage(out _, out _, applicationId, in control);
 
             if (resultCode.IsFailure())
             {
@@ -792,7 +798,7 @@ namespace Ryujinx.HLE.HOS
                 return resultCode;
             }
 
-            resultCode = EnsureApplicationSaveData(hos.Fs, out _, applicationId, ref control, ref user);
+            resultCode = hos.Fs.EnsureApplicationSaveData(out _, applicationId, in control, in user);
 
             if (resultCode.IsFailure())
             {

Ryujinx.HLE/HOS/ArmProcessContext.cs

@@ -36,6 +36,11 @@ namespace Ryujinx.HLE.HOS
             _cpuContext.Execute(context, codeAddress);
         }
 
+        public void InvalidateCacheRegion(ulong address, ulong size)
+        {
+            _cpuContext.InvalidateCacheRegion(address, size);
+        }
+
         public void Dispose()
         {
             if (_memoryManager is IRefCounted rc)

Ryujinx.HLE/HOS/Horizon.cs

@@ -76,6 +76,7 @@ namespace Ryujinx.HLE.HOS
         internal ServerBase BsdServer { get; private set; }
         internal ServerBase AudRenServer { get; private set; }
         internal ServerBase AudOutServer { get; private set; }
+        internal ServerBase FsServer { get; private set; }
         internal ServerBase HidServer { get; private set; }
         internal ServerBase NvDrvServer { get; private set; }
         internal ServerBase TimeServer { get; private set; }
@@ -298,6 +299,7 @@ namespace Ryujinx.HLE.HOS
             BsdServer = new ServerBase(KernelContext, "BsdServer");
             AudRenServer = new ServerBase(KernelContext, "AudioRendererServer");
             AudOutServer = new ServerBase(KernelContext, "AudioOutServer");
+            FsServer = new ServerBase(KernelContext, "FsServer");
             HidServer = new ServerBase(KernelContext, "HidServer");
             NvDrvServer = new ServerBase(KernelContext, "NvservicesServer");
             TimeServer = new ServerBase(KernelContext, "TimeServer");
@@ -464,7 +466,7 @@ namespace Ryujinx.HLE.HOS
 
                 AudioRendererManager.Dispose();
 
-                LibHacHorizonManager.AmClient.Fs.UnregisterProgram(LibHacHorizonManager.ApplicationClient.Os.GetCurrentProcessId().Value);
+                LibHacHorizonManager.PmClient.Fs.UnregisterProgram(LibHacHorizonManager.ApplicationClient.Os.GetCurrentProcessId().Value).ThrowIfFailure();
                 
                 KernelContext.Dispose();
             }

Ryujinx.HLE/HOS/Kernel/Process/IProcessContext.cs

@@ -9,5 +9,6 @@ namespace Ryujinx.HLE.HOS.Kernel.Process
         IVirtualMemoryManager AddressSpace { get; }
 
         void Execute(ExecutionContext context, ulong codeAddress);
+        void InvalidateCacheRegion(ulong address, ulong size);
     }
 }

Ryujinx.HLE/HOS/Kernel/Process/KProcess.cs

@@ -751,7 +751,8 @@ namespace Ryujinx.HLE.HOS.Kernel.Process
         {
             KThread currentThread = KernelStatic.GetCurrentThread();
 
-            if (currentThread.Owner != null &&
+            if (currentThread.Context.Running &&
+                currentThread.Owner != null &&
                 currentThread.GetUserDisableCount() != 0 &&
                 currentThread.Owner.PinnedThreads[currentThread.CurrentCore] == null)
             {

Ryujinx.HLE/HOS/Kernel/Process/ProcessContext.cs

@@ -18,6 +18,10 @@ namespace Ryujinx.HLE.HOS.Kernel.Process
             throw new NotSupportedException();
         }
 
+        public void InvalidateCacheRegion(ulong address, ulong size)
+        {
+        }
+
         public void Dispose()
         {
         }

Ryujinx.HLE/HOS/Kernel/Threading/KThread.cs

@@ -658,10 +658,7 @@ namespace Ryujinx.HLE.HOS.Kernel.Threading
 
         private static uint GetPsr(ARMeilleure.State.ExecutionContext context)
         {
-            return (context.GetPstateFlag(ARMeilleure.State.PState.NFlag) ? (1U << (int)ARMeilleure.State.PState.NFlag) : 0U) |
-                   (context.GetPstateFlag(ARMeilleure.State.PState.ZFlag) ? (1U << (int)ARMeilleure.State.PState.ZFlag) : 0U) |
-                   (context.GetPstateFlag(ARMeilleure.State.PState.CFlag) ? (1U << (int)ARMeilleure.State.PState.CFlag) : 0U) |
-                   (context.GetPstateFlag(ARMeilleure.State.PState.VFlag) ? (1U << (int)ARMeilleure.State.PState.VFlag) : 0U);
+            return context.Pstate & 0xFF0FFE20;
         }
 
         private ThreadContext GetCurrentContext()
@@ -1371,7 +1368,7 @@ namespace Ryujinx.HLE.HOS.Kernel.Threading
 
             PreferredCore = _originalPreferredCore;
             AffinityMask = _originalAffinityMask;
-            
+
             if (AffinityMask != affinityMask)
             {
                 if ((AffinityMask & 1UL << ActiveCore) != 0)

Ryujinx.HLE/HOS/LibHacHorizonManager.cs

@@ -24,6 +24,7 @@ namespace Ryujinx.HLE.HOS
         public HorizonClient BcatClient        { get; private set; }
         public HorizonClient FsClient          { get; private set; }
         public HorizonClient NsClient          { get; private set; }
+        public HorizonClient PmClient          { get; private set; }
         public HorizonClient SdbClient         { get; private set; }
 
         private SharedRef<LibHacIReader> _arpIReader;
@@ -65,6 +66,7 @@ namespace Ryujinx.HLE.HOS
 
         public void InitializeSystemClients()
         {
+            PmClient      = Server.CreatePrivilegedHorizonClient();
             AccountClient = Server.CreateHorizonClient(new ProgramLocation(SystemProgramId.Account, StorageId.BuiltInSystem), AccountFsPermissions);
             AmClient      = Server.CreateHorizonClient(new ProgramLocation(SystemProgramId.Am,      StorageId.BuiltInSystem), AmFsPermissions);
             NsClient      = Server.CreateHorizonClient(new ProgramLocation(SystemProgramId.Ns,      StorageId.BuiltInSystem), NsFsPermissions);

Ryujinx.HLE/HOS/ModLoader.cs

@@ -136,7 +136,8 @@ namespace Ryujinx.HLE.HOS
 
         private static bool StrEquals(string s1, string s2) => string.Equals(s1, s2, StringComparison.OrdinalIgnoreCase);
 
-        public string GetModsBasePath() => EnsureBaseDirStructure(AppDataManager.GetModsPath());
+        public string GetModsBasePath()   => EnsureBaseDirStructure(AppDataManager.GetModsPath());
+        public string GetSdModsBasePath() => EnsureBaseDirStructure(AppDataManager.GetSdModsPath());
 
         private string EnsureBaseDirStructure(string modsBasePath)
         {
@@ -695,7 +696,7 @@ namespace Ryujinx.HLE.HOS
 
             var buildIds = programs.Select(p => p switch
             {
-                NsoExecutable nso => BitConverter.ToString(nso.BuildId.Bytes.ToArray()).Replace("-", "").TrimEnd('0'),
+                NsoExecutable nso => BitConverter.ToString(nso.BuildId.ItemsRo.ToArray()).Replace("-", "").TrimEnd('0'),
                 NroExecutable nro => BitConverter.ToString(nro.Header.BuildId).Replace("-", "").TrimEnd('0'),
                 _ => string.Empty
             }).ToList();

Ryujinx.HLE/HOS/ProgramLoader.cs

@@ -160,7 +160,7 @@ namespace Ryujinx.HLE.HOS
 
             var buildIds = executables.Select(e => (e switch
             {
-                NsoExecutable nso => BitConverter.ToString(nso.BuildId.Bytes.ToArray()),
+                NsoExecutable nso => BitConverter.ToString(nso.BuildId.ItemsRo.ToArray()),
                 NroExecutable nro => BitConverter.ToString(nro.Header.BuildId),
                 _ => ""
             }).Replace("-", "").ToUpper());

Ryujinx.HLE/HOS/Services/Account/Acc/AccountManager.cs

@@ -3,6 +3,7 @@ using LibHac.Common;
 using LibHac.Fs;
 using LibHac.Fs.Shim;
 using Ryujinx.Common;
+using Ryujinx.Common.Logging;
 using System;
 using System.Collections.Concurrent;
 using System.Collections.Generic;
@@ -25,7 +26,7 @@ namespace Ryujinx.HLE.HOS.Services.Account.Acc
 
         public UserProfile LastOpenedUser { get; private set; }
 
-        public AccountManager(HorizonClient horizonClient)
+        public AccountManager(HorizonClient horizonClient, string initialProfileName = null)
         {
             _horizonClient = horizonClient;
 
@@ -43,7 +44,14 @@ namespace Ryujinx.HLE.HOS.Services.Account.Acc
             }
             else
             {
-                OpenUser(_accountSaveDataManager.LastOpened);
+                UserId commandLineUserProfileOverride = default; 
+                if (!string.IsNullOrEmpty(initialProfileName))
+                { 
+                    commandLineUserProfileOverride = _profiles.Values.FirstOrDefault(x => x.Name == initialProfileName)?.UserId ?? default;
+                    if (commandLineUserProfileOverride.IsNull)
+                        Logger.Warning?.Print(LogClass.Application, $"The command line specified profile named '{initialProfileName}' was not found");
+                }
+                OpenUser(commandLineUserProfileOverride.IsNull ? _accountSaveDataManager.LastOpened : commandLineUserProfileOverride);
             }
         }
 
@@ -168,8 +176,8 @@ namespace Ryujinx.HLE.HOS.Services.Account.Acc
 
         private void DeleteSaveData(UserId userId)
         {
-            SaveDataFilter saveDataFilter = new SaveDataFilter();
-            saveDataFilter.SetUserId(new LibHac.Fs.UserId((ulong)userId.High, (ulong)userId.Low));
+            var saveDataFilter = SaveDataFilter.Make(programId: default, saveType: default,
+                new LibHac.Fs.UserId((ulong)userId.High, (ulong)userId.Low), saveDataId: default, index: default);
 
             using var saveDataIterator = new UniqueRef<SaveDataIterator>();
 

Ryujinx.HLE/HOS/Services/Account/Acc/IAccountServiceForApplication.cs

@@ -169,7 +169,7 @@ namespace Ryujinx.HLE.HOS.Services.Account.Acc
             // TODO: Account actually calls nn::arp::detail::IReader::GetApplicationControlProperty() with the current Pid and store the result (NACP file) internally.
             //       But since we use LibHac and we load one Application at a time, it's not necessary.
 
-            context.ResponseData.Write(context.Device.Application.ControlData.Value.UserAccountSwitchLock);
+            context.ResponseData.Write((byte)context.Device.Application.ControlData.Value.UserAccountSwitchLock);
 
             Logger.Stub?.PrintStub(LogClass.ServiceAcc);
 

Ryujinx.HLE/HOS/Services/Am/AppletOE/ApplicationProxyService/ApplicationProxy/IApplicationFunctions.cs

@@ -131,7 +131,7 @@ namespace Ryujinx.HLE.HOS.Services.Am.AppletOE.ApplicationProxyService.Applicati
             }
 
             HorizonClient hos = context.Device.System.LibHacHorizonManager.AmClient;
-            Result result = EnsureApplicationSaveData(hos.Fs, out long requiredSize, applicationId, ref control, ref userId);
+            Result result = hos.Fs.EnsureApplicationSaveData(out long requiredSize, applicationId, in control, in userId);
 
             context.ResponseData.Write(requiredSize);
 
@@ -148,7 +148,7 @@ namespace Ryujinx.HLE.HOS.Services.Am.AppletOE.ApplicationProxyService.Applicati
             // TODO: When above calls are implemented, switch to using ns:am
 
             long desiredLanguageCode = context.Device.System.State.DesiredLanguageCode;
-            int  supportedLanguages  = (int)context.Device.Application.ControlData.Value.SupportedLanguages;
+            int  supportedLanguages  = (int)context.Device.Application.ControlData.Value.SupportedLanguageFlag;
             int  firstSupported      = BitOperations.TrailingZeroCount(supportedLanguages);
 
             if (firstSupported > (int)SystemState.TitleLanguage.BrazilianPortuguese)
@@ -190,7 +190,6 @@ namespace Ryujinx.HLE.HOS.Services.Am.AppletOE.ApplicationProxyService.Applicati
         // GetDisplayVersion() -> nn::oe::DisplayVersion
         public ResultCode GetDisplayVersion(ServiceCtx context)
         {
-            // This should work as DisplayVersion U8Span always gives a 0x10 size byte array.
             // If an NACP isn't found, the buffer will be all '\0' which seems to be the correct implementation.
             context.ResponseData.Write(context.Device.Application.ControlData.Value.DisplayVersion);
 
@@ -252,7 +251,7 @@ namespace Ryujinx.HLE.HOS.Services.Am.AppletOE.ApplicationProxyService.Applicati
             BlitStruct<ApplicationControlProperty> controlHolder = context.Device.Application.ControlData;
 
             Result result = _horizon.Fs.CreateApplicationCacheStorage(out long requiredSize,
-                out CacheStorageTargetMedia storageTarget, applicationId, ref controlHolder.Value, index, saveSize,
+                out CacheStorageTargetMedia storageTarget, applicationId, in controlHolder.Value, index, saveSize,
                 journalSize);
 
             if (result.IsFailure()) return (ResultCode)result.Value;

Ryujinx.HLE/HOS/Services/Arp/LibHacIReader.cs

@@ -16,7 +16,7 @@ namespace Ryujinx.HLE.HOS.Services.Arp
         {
             launchProperty = new LibHac.Arp.ApplicationLaunchProperty
             {
-                BaseStorageId = StorageId.BuiltInUser,
+                StorageId = StorageId.BuiltInUser,
                 ApplicationId = ApplicationId
             };
 
@@ -29,7 +29,7 @@ namespace Ryujinx.HLE.HOS.Services.Arp
         {
             launchProperty = new LibHac.Arp.ApplicationLaunchProperty
             {
-                BaseStorageId = StorageId.BuiltInUser,
+                StorageId = StorageId.BuiltInUser,
                 ApplicationId = applicationId
             };
 

Ryujinx.HLE/HOS/Services/DisposableIpcService.cs

@@ -7,6 +7,8 @@ namespace Ryujinx.HLE.HOS.Services
     {
         private int _disposeState;
 
+        public DisposableIpcService(ServerBase server = null) : base(server) { }
+
         protected abstract void Dispose(bool isDisposing);
 
         public void Dispose()

Ryujinx.HLE/HOS/Services/Fs/FileSystemProxy/FileSystemProxyHelper.cs

@@ -132,22 +132,10 @@ namespace Ryujinx.HLE.HOS.Services.Fs.FileSystemProxy
             }
         }
 
-        public static Result ReadFsPath(out FsPath path, ServiceCtx context, int index = 0)
-        {
-            ulong position = context.Request.PtrBuff[index].Position;
-            ulong size     = context.Request.PtrBuff[index].Size;
-
-            byte[] pathBytes = new byte[size];
-
-            context.Memory.Read(position, pathBytes);
-
-            return FsPath.FromSpan(out path, pathBytes);
-        }
-
         public static ref readonly FspPath GetFspPath(ServiceCtx context, int index = 0)
         {
-            ulong position = (ulong)context.Request.PtrBuff[index].Position;
-            ulong size = (ulong)context.Request.PtrBuff[index].Size;
+            ulong position = context.Request.PtrBuff[index].Position;
+            ulong size = context.Request.PtrBuff[index].Size;
 
             ReadOnlySpan<byte> buffer = context.Memory.GetSpan(position, (int)size);
             ReadOnlySpan<FspPath> fspBuffer = MemoryMarshal.Cast<byte, FspPath>(buffer);
@@ -157,8 +145,8 @@ namespace Ryujinx.HLE.HOS.Services.Fs.FileSystemProxy
 
         public static ref readonly LibHac.FsSrv.Sf.Path GetSfPath(ServiceCtx context, int index = 0)
         {
-            ulong position = (ulong)context.Request.PtrBuff[index].Position;
-            ulong size = (ulong)context.Request.PtrBuff[index].Size;
+            ulong position = context.Request.PtrBuff[index].Position;
+            ulong size = context.Request.PtrBuff[index].Size;
 
             ReadOnlySpan<byte> buffer = context.Memory.GetSpan(position, (int)size);
             ReadOnlySpan<LibHac.FsSrv.Sf.Path> pathBuffer = MemoryMarshal.Cast<byte, LibHac.FsSrv.Sf.Path>(buffer);

Ryujinx.HLE/HOS/Services/Fs/IDeviceOperator.cs

@@ -1,6 +1,6 @@
 using LibHac;
 using LibHac.Common;
-using LibHac.FsSrv;
+using LibHac.Fs;
 
 namespace Ryujinx.HLE.HOS.Services.Fs
 {

Ryujinx.HLE/HOS/Services/Fs/IFileSystemProxy.cs

@@ -28,7 +28,7 @@ namespace Ryujinx.HLE.HOS.Services.Fs
     {
         private SharedRef<LibHac.FsSrv.Sf.IFileSystemProxy> _baseFileSystemProxy;
 
-        public IFileSystemProxy(ServiceCtx context)
+        public IFileSystemProxy(ServiceCtx context) : base(context.Device.System.FsServer)
         {
             var applicationClient = context.Device.System.LibHacHorizonManager.ApplicationClient;
             _baseFileSystemProxy = applicationClient.Fs.Impl.GetFileSystemProxyServiceObject();

Ryujinx.HLE/HOS/Services/Nifm/StaticService/IGeneralService.cs

@@ -14,6 +14,9 @@ namespace Ryujinx.HLE.HOS.Services.Nifm.StaticService
     {
         private GeneralServiceDetail _generalServiceDetail;
 
+        private IPInterfaceProperties _targetPropertiesCache = null;
+        private UnicastIPAddressInformation _targetAddressInfoCache = null;
+
         public IGeneralService()
         {
             _generalServiceDetail = new GeneralServiceDetail
@@ -22,6 +25,8 @@ namespace Ryujinx.HLE.HOS.Services.Nifm.StaticService
                 IsAnyInternetRequestAccepted = true // NOTE: Why not accept any internet request?
             };
 
+            NetworkChange.NetworkAddressChanged += new NetworkAddressChangedEventHandler(LocalInterfaceCacheHandler);
+
             GeneralServiceManager.Add(_generalServiceDetail);
         }
 
@@ -165,6 +170,11 @@ namespace Ryujinx.HLE.HOS.Services.Nifm.StaticService
                 return (null, null);
             }
 
+            if (_targetPropertiesCache != null && _targetAddressInfoCache != null)
+            {
+                return (_targetPropertiesCache, _targetAddressInfoCache);
+            }
+
             IPInterfaceProperties       targetProperties  = null;
             UnicastIPAddressInformation targetAddressInfo = null;
 
@@ -194,13 +204,26 @@ namespace Ryujinx.HLE.HOS.Services.Nifm.StaticService
                 }
             }
 
+            _targetPropertiesCache  = targetProperties;
+            _targetAddressInfoCache = targetAddressInfo;
+
             return (targetProperties, targetAddressInfo);
         }
 
+        private void LocalInterfaceCacheHandler(object sender, EventArgs e)
+        {
+            Logger.Info?.Print(LogClass.ServiceNifm, $"NetworkAddress changed, invalidating cached data.");
+
+            _targetPropertiesCache  = null;
+            _targetAddressInfoCache = null;
+        }
+
         protected override void Dispose(bool isDisposing)
         {
             if (isDisposing)
             {
+                NetworkChange.NetworkAddressChanged -= LocalInterfaceCacheHandler;
+
                 GeneralServiceManager.Remove(_generalServiceDetail.ClientId);
             }
         }

Ryujinx.HLE/HOS/Services/Nim/Ntc/IStaticService.cs

@@ -1,8 +1,24 @@
-namespace Ryujinx.HLE.HOS.Services.Nim.Ntc
+using Ryujinx.Common.Logging;
+using Ryujinx.HLE.HOS.Services.Nim.Ntc.StaticService;
+
+namespace Ryujinx.HLE.HOS.Services.Nim.Ntc
 {
     [Service("ntc")]
     class IStaticService : IpcService
     {
         public IStaticService(ServiceCtx context) { }
+
+        [CommandHipc(0)]
+        // OpenEnsureNetworkClockAvailabilityService(u64) -> object<nn::ntc::detail::service::IEnsureNetworkClockAvailabilityService>
+        public ResultCode CreateAsyncInterface(ServiceCtx context)
+        {
+            ulong unknown = context.RequestData.ReadUInt64();
+
+            MakeObject(context, new IEnsureNetworkClockAvailabilityService(context));
+
+            Logger.Stub?.PrintStub(LogClass.ServiceNtc, new { unknown });
+
+            return ResultCode.Success;
+        }
     }
 }

Ryujinx.HLE/HOS/Services/Nim/Ntc/StaticService/IEnsureNetworkClockAvailabilityService.cs

@@ -0,0 +1,77 @@
+using Ryujinx.Common.Logging;
+using Ryujinx.HLE.HOS.Ipc;
+using Ryujinx.HLE.HOS.Kernel.Common;
+using Ryujinx.HLE.HOS.Kernel.Threading;
+using System;
+
+namespace Ryujinx.HLE.HOS.Services.Nim.Ntc.StaticService
+{
+    class IEnsureNetworkClockAvailabilityService : IpcService 
+    {
+        private KEvent     _finishNotificationEvent;
+        private ResultCode _taskResultCode;
+
+        public IEnsureNetworkClockAvailabilityService(ServiceCtx context)
+        {
+            _finishNotificationEvent = new KEvent(context.Device.System.KernelContext);
+            _taskResultCode          = ResultCode.Success;
+
+            // NOTE: The service starts a thread that polls Nintendo NTP server and syncs the time with it.
+            //       Additionnally it gets and uses some settings too:
+            //       autonomic_correction_interval_seconds, autonomic_correction_failed_retry_interval_seconds,
+            //       autonomic_correction_immediate_try_count_max, autonomic_correction_immediate_try_interval_milliseconds
+        }
+
+        [CommandHipc(0)]
+        // StartTask()
+        public ResultCode StartTask(ServiceCtx context)
+        {
+            if (!context.Device.Configuration.EnableInternetAccess)
+            {
+                return (ResultCode)Time.ResultCode.NetworkTimeNotAvailable;
+            }
+
+            // NOTE: Since we don't support the Nintendo NTP server, we can signal the event now to confirm the update task is done.
+            _finishNotificationEvent.ReadableEvent.Signal();
+
+            Logger.Stub?.PrintStub(LogClass.ServiceNtc);
+
+            return ResultCode.Success;
+        }
+
+        [CommandHipc(1)]
+        // GetFinishNotificationEvent() -> handle<copy>
+        public ResultCode GetFinishNotificationEvent(ServiceCtx context)
+        {
+            if (context.Process.HandleTable.GenerateHandle(_finishNotificationEvent.ReadableEvent, out int finishNotificationEventHandle) != KernelResult.Success)
+            {
+                throw new InvalidOperationException("Out of handles!");
+            }
+
+            context.Response.HandleDesc = IpcHandleDesc.MakeCopy(finishNotificationEventHandle);
+
+            return ResultCode.Success;
+        }
+
+        [CommandHipc(2)]
+        // GetResult()
+        public ResultCode GetResult(ServiceCtx context)
+        {
+            return _taskResultCode;
+        }
+
+        [CommandHipc(3)]
+        // Cancel()
+        public ResultCode Cancel(ServiceCtx context)
+        {
+            // NOTE: The update task should be canceled here.
+            _finishNotificationEvent.ReadableEvent.Signal();
+
+            _taskResultCode = (ResultCode)Time.ResultCode.NetworkTimeTaskCanceled;
+
+            Logger.Stub?.PrintStub(LogClass.ServiceNtc);
+
+            return ResultCode.Success;
+        }
+    }
+}

Ryujinx.HLE/HOS/Services/Nv/NvDrvServices/NvHostAsGpu/NvHostAsGpuDeviceFile.cs

@@ -232,7 +232,7 @@ namespace Ryujinx.HLE.HOS.Services.Nv.NvDrvServices.NvHostAsGpu
                         ulong virtualAddress = arguments.Offset + arguments.BufferOffset;
 
                         physicalAddress += arguments.BufferOffset;
-                        _asContext.Gmm.Map(physicalAddress, virtualAddress, arguments.MappingSize);
+                        _asContext.Gmm.Map(physicalAddress, virtualAddress, arguments.MappingSize, (PteKind)arguments.Kind);
 
                         return NvInternalResult.Success;
                     }
@@ -282,7 +282,7 @@ namespace Ryujinx.HLE.HOS.Services.Nv.NvDrvServices.NvHostAsGpu
                 {
                     if (_asContext.ValidateFixedBuffer(arguments.Offset, size, pageSize))
                     {
-                        _asContext.Gmm.Map(physicalAddress, arguments.Offset, size);
+                        _asContext.Gmm.Map(physicalAddress, arguments.Offset, size, (PteKind)arguments.Kind);
                     }
                     else
                     {
@@ -301,7 +301,7 @@ namespace Ryujinx.HLE.HOS.Services.Nv.NvDrvServices.NvHostAsGpu
                         _memoryAllocator.AllocateRange(va, size, freeAddressStartPosition);
                     }
 
-                    _asContext.Gmm.Map(physicalAddress, va, size);
+                    _asContext.Gmm.Map(physicalAddress, va, size, (PteKind)arguments.Kind);
                     arguments.Offset = va;
                 }
 
@@ -366,26 +366,30 @@ namespace Ryujinx.HLE.HOS.Services.Nv.NvDrvServices.NvHostAsGpu
 
             for (int index = 0; index < arguments.Length; index++)
             {
-                ulong mapOffs = (ulong)arguments[index].MapOffset << 16;
-                ulong gpuVa = (ulong)arguments[index].GpuOffset << 16;
-                ulong size = (ulong)arguments[index].Pages << 16;
+                ref RemapArguments argument = ref arguments[index];
+                ulong gpuVa = (ulong)argument.GpuOffset << 16;
+                ulong size = (ulong)argument.Pages << 16;
+                int nvmapHandle = argument.NvMapHandle;
 
-                if (arguments[index].NvMapHandle == 0)
+                if (nvmapHandle == 0)
                 {
                     gmm.Unmap(gpuVa, size);
                 }
                 else
                 {
-                    NvMapHandle map = NvMapDeviceFile.GetMapFromHandle(Owner, arguments[index].NvMapHandle);
+                    ulong mapOffs = (ulong)argument.MapOffset << 16;
+                    PteKind kind = (PteKind)argument.Kind;
+
+                    NvMapHandle map = NvMapDeviceFile.GetMapFromHandle(Owner, nvmapHandle);
 
                     if (map == null)
                     {
-                        Logger.Warning?.Print(LogClass.ServiceNv, $"Invalid NvMap handle 0x{arguments[index].NvMapHandle:x8}!");
+                        Logger.Warning?.Print(LogClass.ServiceNv, $"Invalid NvMap handle 0x{nvmapHandle:x8}!");
 
                         return NvInternalResult.InvalidInput;
                     }
 
-                    gmm.Map(mapOffs + map.Address, gpuVa, size);
+                    gmm.Map(mapOffs + map.Address, gpuVa, size, kind);
                 }
             }
 

Ryujinx.HLE/HOS/Services/Nv/NvDrvServices/NvHostChannel/NvHostChannelDeviceFile.cs

@@ -251,7 +251,7 @@ namespace Ryujinx.HLE.HOS.Services.Nv.NvDrvServices.NvHostChannel
                         if (va != NvMemoryAllocator.PteUnmapped && va <= uint.MaxValue && (va + (uint)map.Size) <= uint.MaxValue)
                         {
                             _host1xContext.MemoryAllocator.AllocateRange(va, (uint)map.Size, freeAddressStartPosition);
-                            _host1xContext.Smmu.Map(map.Address, va, (uint)map.Size);
+                            _host1xContext.Smmu.Map(map.Address, va, (uint)map.Size, PteKind.Pitch); // FIXME: This should not use the GMMU.
                             map.DmaMapAddress = va;
                         }
                         else

Ryujinx.HLE/HOS/Services/Olsc/IOlscServiceForApplication.cs

@@ -1,13 +1,15 @@
 using Ryujinx.Common;
 using Ryujinx.Common.Logging;
 using Ryujinx.HLE.HOS.Services.Account.Acc;
+using System.Collections.Generic;
 
 namespace Ryujinx.HLE.HOS.Services.Olsc
 {
     [Service("olsc:u")] // 10.0.0+
     class IOlscServiceForApplication : IpcService
     {
-        private bool _initialized;
+        private bool                     _initialized;
+        private Dictionary<UserId, bool> _saveDataBackupSettingDatabase;
 
         public IOlscServiceForApplication(ServiceCtx context) { }
 
@@ -16,7 +18,9 @@ namespace Ryujinx.HLE.HOS.Services.Olsc
         public ResultCode Initialize(ServiceCtx context)
         {
             // NOTE: Service call arp:r GetApplicationInstanceUnregistrationNotifier with the pid and initialize some internal struct.
-            //       Since we will not support online savedata backup. It's fine to stub it for now.
+            //       Since we will not support online savedata backup, it's fine to stub it for now.
+
+            _saveDataBackupSettingDatabase = new Dictionary<UserId, bool>();
 
             _initialized = true;
 
@@ -25,12 +29,11 @@ namespace Ryujinx.HLE.HOS.Services.Olsc
             return ResultCode.Success;
         }
 
-        [CommandHipc(14)]
-        // SetSaveDataBackupSettingEnabled(nn::account::Uid, bool)
-        public ResultCode SetSaveDataBackupSettingEnabled(ServiceCtx context)
+        [CommandHipc(13)]
+        // GetSaveDataBackupSetting(nn::account::Uid) -> u8
+        public ResultCode GetSaveDataBackupSetting(ServiceCtx context)
         {
-            UserId userId                       = context.RequestData.ReadStruct<UserId>();
-            ulong  saveDataBackupSettingEnabled = context.RequestData.ReadUInt64();
+            UserId userId = context.RequestData.ReadStruct<UserId>();
 
             if (!_initialized)
             {
@@ -42,8 +45,42 @@ namespace Ryujinx.HLE.HOS.Services.Olsc
                 return ResultCode.NullArgument;
             }
 
-            // NOTE: Service store the UserId and the boolean in an internal SaveDataBackupSettingDatabase object.
-            //       Since we will not support online savedata backup. It's fine to stub it for now.
+            if (_saveDataBackupSettingDatabase.TryGetValue(userId, out bool enabled) && enabled)
+            {
+                context.ResponseData.Write((byte)1); // TODO: Determine value.
+            }
+            else
+            {
+                context.ResponseData.Write((byte)2); // TODO: Determine value.
+            }
+
+            // NOTE: Since we will not support online savedata backup, it's fine to stub it for now.
+
+            Logger.Stub?.PrintStub(LogClass.ServiceOlsc, new { userId });
+
+            return ResultCode.Success;
+        }
+
+        [CommandHipc(14)]
+        // SetSaveDataBackupSettingEnabled(nn::account::Uid, bool)
+        public ResultCode SetSaveDataBackupSettingEnabled(ServiceCtx context)
+        {
+            bool   saveDataBackupSettingEnabled = context.RequestData.ReadUInt64() != 0;
+            UserId userId                       = context.RequestData.ReadStruct<UserId>();
+
+            if (!_initialized)
+            {
+                return ResultCode.NotInitialized;
+            }
+
+            if (userId.IsNull)
+            {
+                return ResultCode.NullArgument;
+            }
+
+            _saveDataBackupSettingDatabase[userId] = saveDataBackupSettingEnabled;
+
+            // NOTE: Since we will not support online savedata backup, it's fine to stub it for now.
 
             Logger.Stub?.PrintStub(LogClass.ServiceOlsc, new { userId, saveDataBackupSettingEnabled });
 

Ryujinx.HLE/HOS/Services/Pctl/ParentalControlServiceFactory/IParentalControlService.cs

@@ -1,8 +1,9 @@
-using LibHac.Ns;
 using Ryujinx.Common.Logging;
 using Ryujinx.HLE.HOS.Services.Arp;
 using System;
 
+using static LibHac.Ns.ApplicationControlProperty;
+
 namespace Ryujinx.HLE.HOS.Services.Pctl.ParentalControlServiceFactory
 {
     class IParentalControlService : IpcService
@@ -52,8 +53,8 @@ namespace Ryujinx.HLE.HOS.Services.Pctl.ParentalControlServiceFactory
                         _titleId = titleId;
 
                         // TODO: Call nn::arp::GetApplicationControlProperty here when implemented, if it return ResultCode.Success we assign fields.
-                        _ratingAge           = Array.ConvertAll(context.Device.Application.ControlData.Value.RatingAge.ToArray(), Convert.ToInt32);
-                        _parentalControlFlag = context.Device.Application.ControlData.Value.ParentalControl;
+                        _ratingAge           = Array.ConvertAll(context.Device.Application.ControlData.Value.RatingAge.ItemsRo.ToArray(), Convert.ToInt32);
+                        _parentalControlFlag = context.Device.Application.ControlData.Value.ParentalControlFlag;
                     }
                 }
 
@@ -224,7 +225,7 @@ namespace Ryujinx.HLE.HOS.Services.Pctl.ParentalControlServiceFactory
         private ResultCode IsStereoVisionPermittedImpl()
         {
             /*
-                // TODO: Application Exemptions are readed from file "appExemptions.dat" in the service savedata.
+                // TODO: Application Exemptions are read from file "appExemptions.dat" in the service savedata.
                 //       Since we don't support the pctl savedata for now, this can be implemented later.
 
                 if (appExemption)

Ryujinx.HLE/HOS/Services/Sdb/Pdm/QueryService/QueryPlayStatisticsManager.cs

@@ -15,6 +15,8 @@ namespace Ryujinx.HLE.HOS.Services.Sdb.Pdm.QueryService
 
         internal static ResultCode GetPlayStatistics(ServiceCtx context, bool byUserId = false)
         {
+            ref readonly var controlProperty = ref context.Device.Application.ControlData.Value;
+
             ulong inputPosition = context.Request.SendBuff[0].Position;
             ulong inputSize     = context.Request.SendBuff[0].Size;
 
@@ -31,7 +33,7 @@ namespace Ryujinx.HLE.HOS.Services.Sdb.Pdm.QueryService
                 }
             }
 
-            PlayLogQueryCapability queryCapability = (PlayLogQueryCapability)context.Device.Application.ControlData.Value.PlayLogQueryCapability;
+            PlayLogQueryCapability queryCapability = (PlayLogQueryCapability)controlProperty.PlayLogQueryCapability;
 
             List<ulong> titleIds = new List<ulong>();
 
@@ -45,7 +47,7 @@ namespace Ryujinx.HLE.HOS.Services.Sdb.Pdm.QueryService
                 // Check if input title ids are in the whitelist.
                 foreach (ulong titleId in titleIds)
                 {
-                    if (!context.Device.Application.ControlData.Value.PlayLogQueryableApplicationId.Contains(titleId))
+                    if (!controlProperty.PlayLogQueryableApplicationId.ItemsRo.Contains(titleId))
                     {
                         return (ResultCode)Am.ResultCode.ObjectInvalid;
                     }

Ryujinx.HLE/HOS/Services/ServerBase.cs

@@ -185,7 +185,7 @@ namespace Ryujinx.HLE.HOS.Services
             {
                 for (int i = 0; i < request.RecvListBuff.Count; i++)
                 {
-                    ulong size = (ulong)BinaryPrimitives.ReadInt16LittleEndian(request.RawData.AsSpan().Slice(sizesOffset + i * 2, 2));
+                    ulong size = (ulong)BinaryPrimitives.ReadInt16LittleEndian(request.RawData.AsSpan(sizesOffset + i * 2, 2));
 
                     response.PtrBuff.Add(new IpcPtrBuffDesc(tempAddr, (uint)i, size));
 

Ryujinx.HLE/HOS/Services/Time/ResultCode.cs

@@ -7,16 +7,18 @@
 
         Success = 0,
 
-        TimeServiceNotInitialized = (0   << ErrorCodeShift) | ModuleId,
-        PermissionDenied          = (1   << ErrorCodeShift) | ModuleId,
-        TimeMismatch              = (102 << ErrorCodeShift) | ModuleId,
-        UninitializedClock        = (103 << ErrorCodeShift) | ModuleId,
-        TimeNotFound              = (200 << ErrorCodeShift) | ModuleId,
-        Overflow                  = (201 << ErrorCodeShift) | ModuleId,
-        LocationNameTooLong       = (801 << ErrorCodeShift) | ModuleId,
-        OutOfRange                = (902 << ErrorCodeShift) | ModuleId,
-        TimeZoneConversionFailed  = (903 << ErrorCodeShift) | ModuleId,
-        TimeZoneNotFound          = (989 << ErrorCodeShift) | ModuleId,
-        NotImplemented            = (990 << ErrorCodeShift) | ModuleId,
+        TimeServiceNotInitialized = (0    << ErrorCodeShift) | ModuleId,
+        PermissionDenied          = (1    << ErrorCodeShift) | ModuleId,
+        TimeMismatch              = (102  << ErrorCodeShift) | ModuleId,
+        UninitializedClock        = (103  << ErrorCodeShift) | ModuleId,
+        TimeNotFound              = (200  << ErrorCodeShift) | ModuleId,
+        Overflow                  = (201  << ErrorCodeShift) | ModuleId,
+        LocationNameTooLong       = (801  << ErrorCodeShift) | ModuleId,
+        OutOfRange                = (902  << ErrorCodeShift) | ModuleId,
+        TimeZoneConversionFailed  = (903  << ErrorCodeShift) | ModuleId,
+        TimeZoneNotFound          = (989  << ErrorCodeShift) | ModuleId,
+        NotImplemented            = (990  << ErrorCodeShift) | ModuleId,
+        NetworkTimeNotAvailable   = (1000 << ErrorCodeShift) | ModuleId,
+        NetworkTimeTaskCanceled   = (1003 << ErrorCodeShift) | ModuleId,
     }
 }

Ryujinx.HLE/HOS/Services/Vi/IApplicationRootService.cs

@@ -16,7 +16,7 @@ namespace Ryujinx.HLE.HOS.Services.Vi
 
             if (serviceType != ViServiceType.Application)
             {
-                return ResultCode.InvalidRange;
+                return ResultCode.PermissionDenied;
             }
 
             MakeObject(context, new IApplicationDisplayService(serviceType));

Ryujinx.HLE/HOS/Services/Vi/IManagerRootService.cs

@@ -17,7 +17,7 @@ namespace Ryujinx.HLE.HOS.Services.Vi
 
             if (serviceType != ViServiceType.Manager)
             {
-                return ResultCode.InvalidRange;
+                return ResultCode.PermissionDenied;
             }
 
             MakeObject(context, new IApplicationDisplayService(serviceType));