Add XTS mode and remove duplicate code

src/LibHac/Common/Buffer.cs

@@ -0,0 +1,68 @@
+using System;
+using System.Diagnostics;
+using System.Runtime.CompilerServices;
+using System.Runtime.InteropServices;
+
+namespace LibHac.Common
+{
+    /// <summary>
+    /// Represents a buffer of 16 bytes.
+    /// Contains functions that assist with common operations on small buffers.
+    /// </summary>
+    [DebuggerDisplay("{ToString()}")]
+    [StructLayout(LayoutKind.Sequential, Size = 16)]
+    public struct Buffer16
+    {
+        [DebuggerBrowsable(DebuggerBrowsableState.Never)] private ulong _dummy0;
+        [DebuggerBrowsable(DebuggerBrowsableState.Never)] private ulong _dummy1;
+
+        public byte this[int i]
+        {
+            get => Bytes[i];
+            set => Bytes[i] = value;
+        }
+
+        public Span<byte> Bytes => SpanHelpers.AsByteSpan(ref this);
+
+        // Prevent a defensive copy by changing the read-only in reference to a reference with Unsafe.AsRef()
+        [MethodImpl(MethodImplOptions.AggressiveInlining)]
+        public static implicit operator Span<byte>(in Buffer16 value)
+        {
+            return SpanHelpers.AsByteSpan(ref Unsafe.AsRef(in value));
+        }
+
+        [MethodImpl(MethodImplOptions.AggressiveInlining)]
+        public static implicit operator ReadOnlySpan<byte>(in Buffer16 value)
+        {
+            return SpanHelpers.AsReadOnlyByteSpan(ref Unsafe.AsRef(in value));
+        }
+
+        [MethodImpl(MethodImplOptions.AggressiveInlining)]
+        public ref T As<T>() where T : unmanaged
+        {
+            if (Unsafe.SizeOf<T>() > (uint)Unsafe.SizeOf<Buffer16>())
+            {
+                throw new ArgumentException();
+            }
+
+            return ref MemoryMarshal.GetReference(AsSpan<T>());
+        }
+
+        [MethodImpl(MethodImplOptions.AggressiveInlining)]
+        public Span<T> AsSpan<T>() where T : unmanaged
+        {
+            return SpanHelpers.AsSpan<Buffer16, T>(ref this);
+        }
+
+        [MethodImpl(MethodImplOptions.AggressiveInlining)]
+        public readonly ReadOnlySpan<T> AsReadOnlySpan<T>() where T : unmanaged
+        {
+            return SpanHelpers.AsReadOnlySpan<Buffer16, T>(ref Unsafe.AsRef(in this));
+        }
+
+        public override string ToString()
+        {
+            return Bytes.ToHexString();
+        }
+    }
+}

src/LibHac/Common/SpanHelpers.cs

@@ -25,11 +25,51 @@ namespace LibHac.Common
             return CreateSpan(ref reference, 1);
         }
 
+        [MethodImpl(MethodImplOptions.AggressiveInlining)]
+        public static Span<TSpan> AsSpan<TStruct, TSpan>(ref TStruct reference)
+            where TStruct : unmanaged where TSpan : unmanaged
+        {
+            return CreateSpan(ref Unsafe.As<TStruct, TSpan>(ref reference),
+                Unsafe.SizeOf<TStruct>() / Unsafe.SizeOf<TSpan>());
+        }
+
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
         public static Span<byte> AsByteSpan<T>(ref T reference) where T : unmanaged
         {
-            Span<T> span = AsSpan(ref reference);
-            return MemoryMarshal.Cast<T, byte>(span);
+            return CreateSpan(ref Unsafe.As<T, byte>(ref reference), Unsafe.SizeOf<T>());
+        }
+
+        [MethodImpl(MethodImplOptions.AggressiveInlining)]
+#if NETCOREAPP
+        public static ReadOnlySpan<T> CreateReadOnlySpan<T>(ref T reference, int length)
+        {
+            return MemoryMarshal.CreateReadOnlySpan(ref reference, length);
+        }
+#else
+        public static unsafe ReadOnlySpan<T> CreateReadOnlySpan<T>(ref T reference, int length)
+        {
+            return new ReadOnlySpan<T>(Unsafe.AsPointer(ref reference), length);
+        }
+#endif
+
+        [MethodImpl(MethodImplOptions.AggressiveInlining)]
+        public static ReadOnlySpan<T> AsReadOnlySpan<T>(ref T reference) where T : unmanaged
+        {
+            return CreateReadOnlySpan(ref reference, 1);
+        }
+
+        [MethodImpl(MethodImplOptions.AggressiveInlining)]
+        public static ReadOnlySpan<TSpan> AsReadOnlySpan<TStruct, TSpan>(ref TStruct reference)
+            where TStruct : unmanaged where TSpan : unmanaged
+        {
+            return CreateReadOnlySpan(ref Unsafe.As<TStruct, TSpan>(ref reference),
+                Unsafe.SizeOf<TStruct>() / Unsafe.SizeOf<TSpan>());
+        }
+
+        [MethodImpl(MethodImplOptions.AggressiveInlining)]
+        public static ReadOnlySpan<byte> AsReadOnlyByteSpan<T>(ref T reference) where T : unmanaged
+        {
+            return CreateReadOnlySpan(ref Unsafe.As<T, byte>(ref reference), Unsafe.SizeOf<T>());
         }
     }
 }

src/LibHac/Crypto2/Aes.cs

@@ -8,6 +8,9 @@ namespace LibHac.Crypto2
 {
     public static class AesCrypto
     {
+        public const int KeySize128 = 0x10;
+        public const int BlockSize = 0x10;
+
         public static bool IsAesNiSupported()
         {
 #if HAS_INTRINSICS
@@ -77,5 +80,29 @@ namespace LibHac.Crypto2
 #endif
             return new AesCtrEncryptor(key, iv);
         }
+
+        public static ICipher CreateXtsDecryptor(ReadOnlySpan<byte> key1, ReadOnlySpan<byte> key2,
+            ReadOnlySpan<byte> iv, bool preferDotNetCrypto = false)
+        {
+#if HAS_INTRINSICS
+            if (IsAesNiSupported() && !preferDotNetCrypto)
+            {
+                return new AesXtsCipherHw(key1, key2, iv, true);
+            }
+#endif
+            return new AesXtsCipher(key1, key2, iv, true);
+        }
+
+        public static ICipher CreateXtsEncryptor(ReadOnlySpan<byte> key1, ReadOnlySpan<byte> key2,
+            ReadOnlySpan<byte> iv, bool preferDotNetCrypto = false)
+        {
+#if HAS_INTRINSICS
+            if (IsAesNiSupported() && !preferDotNetCrypto)
+            {
+                return new AesXtsCipherHw(key1, key2, iv, false);
+            }
+#endif
+            return new AesXtsCipher(key1, key2, iv, false);
+        }
     }
 }

src/LibHac/Crypto2/AesCbcModeHw.cs

@@ -24,32 +24,24 @@ namespace LibHac.Crypto2
         [MethodImpl(MethodImplOptions.AggressiveOptimization)]
         public void Transform(ReadOnlySpan<byte> input, Span<byte> output)
         {
-            ReadOnlySpan<Vector128<byte>> keys = _aesCore.RoundKeys;
-            ReadOnlySpan<Vector128<byte>> inBlocks = MemoryMarshal.Cast<byte, Vector128<byte>>(input);
-            Span<Vector128<byte>> outBlocks = MemoryMarshal.Cast<byte, Vector128<byte>>(output);
+            int blockCount = Math.Min(input.Length, output.Length) >> 4;
 
-            Vector128<byte> b = _iv;
-            
-            for (int i = 0; i < inBlocks.Length; i++)
+            ref Vector128<byte> inBlock = ref Unsafe.As<byte, Vector128<byte>>(ref MemoryMarshal.GetReference(input));
+            ref Vector128<byte> outBlock = ref Unsafe.As<byte, Vector128<byte>>(ref MemoryMarshal.GetReference(output));
+
+            Vector128<byte> iv = _iv;
+
+            for (int i = 0; i < blockCount; i++)
             {
-                b = Sse2.Xor(b, inBlocks[i]);
+                iv = _aesCore.EncryptBlock(Sse2.Xor(iv, inBlock));
 
-                b = Sse2.Xor(b, keys[0]);
-                b = Aes.Encrypt(b, keys[1]);
-                b = Aes.Encrypt(b, keys[2]);
-                b = Aes.Encrypt(b, keys[3]);
-                b = Aes.Encrypt(b, keys[4]);
-                b = Aes.Encrypt(b, keys[5]);
-                b = Aes.Encrypt(b, keys[6]);
-                b = Aes.Encrypt(b, keys[7]);
-                b = Aes.Encrypt(b, keys[8]);
-                b = Aes.Encrypt(b, keys[9]);
-                b = Aes.EncryptLast(b, keys[10]);
+                outBlock = iv;
 
-                outBlocks[i] = b;
+                inBlock = ref Unsafe.Add(ref inBlock, 1);
+                outBlock = ref Unsafe.Add(ref outBlock, 1);
             }
 
-            _iv = b;
+            _iv = iv;
         }
     }
 
@@ -70,32 +62,22 @@ namespace LibHac.Crypto2
         [MethodImpl(MethodImplOptions.AggressiveOptimization)]
         public void Transform(ReadOnlySpan<byte> input, Span<byte> output)
         {
-            ReadOnlySpan<Vector128<byte>> keys = _aesCore.RoundKeys;
-            ReadOnlySpan<Vector128<byte>> inBlocks = MemoryMarshal.Cast<byte, Vector128<byte>>(input);
-            Span<Vector128<byte>> outBlocks = MemoryMarshal.Cast<byte, Vector128<byte>>(output);
+            int blockCount = Math.Min(input.Length, output.Length) >> 4;
+
+            ref Vector128<byte> inBlock = ref Unsafe.As<byte, Vector128<byte>>(ref MemoryMarshal.GetReference(input));
+            ref Vector128<byte> outBlock = ref Unsafe.As<byte, Vector128<byte>>(ref MemoryMarshal.GetReference(output));
 
             Vector128<byte> iv = _iv;
 
-            for (int i = 0; i < inBlocks.Length; i++)
+            for (int i = 0; i < blockCount; i++)
             {
-                Vector128<byte> b = inBlocks[i];
-                Vector128<byte> nextIv = b;
+                Vector128<byte> decBeforeIv = _aesCore.DecryptBlock(inBlock);
+                outBlock = Sse2.Xor(decBeforeIv, iv);
 
-                b = Sse2.Xor(b, keys[10]);
-                b = Aes.Decrypt(b, keys[9]);
-                b = Aes.Decrypt(b, keys[8]);
-                b = Aes.Decrypt(b, keys[7]);
-                b = Aes.Decrypt(b, keys[6]);
-                b = Aes.Decrypt(b, keys[5]);
-                b = Aes.Decrypt(b, keys[4]);
-                b = Aes.Decrypt(b, keys[3]);
-                b = Aes.Decrypt(b, keys[2]);
-                b = Aes.Decrypt(b, keys[1]);
-                b = Aes.DecryptLast(b, keys[0]);
+                iv = inBlock;
 
-                b = Sse2.Xor(b, iv);
-                iv = nextIv;
-                outBlocks[i] = b;
+                inBlock = ref Unsafe.Add(ref inBlock, 1);
+                outBlock = ref Unsafe.Add(ref outBlock, 1);
             }
 
             _iv = iv;

src/LibHac/Crypto2/AesCoreNi.cs

@@ -15,67 +15,90 @@ namespace LibHac.Crypto2
 
         private Vector128<byte> _roundKeys;
 
+        public AesCoreNi(ReadOnlySpan<byte> key, bool isDecrypting)
+        {
+            _roundKeys = default;
+
+            KeyExpansion(key, MemoryMarshal.CreateSpan(ref _roundKeys, RoundKeyCount), isDecrypting);
+        }
+
         public void Initialize(ReadOnlySpan<byte> key, bool isDecrypting)
         {
-            KeyExpansion(key, RoundKeys, isDecrypting);
+            KeyExpansion(key, MemoryMarshal.CreateSpan(ref _roundKeys, RoundKeyCount), isDecrypting);
         }
 
-        public Span<Vector128<byte>> RoundKeys =>
-            MemoryMarshal.CreateSpan(ref _roundKeys, RoundKeyCount);
+        public readonly ReadOnlySpan<Vector128<byte>> RoundKeys =>
+            MemoryMarshal.CreateReadOnlySpan(ref Unsafe.AsRef(in _roundKeys), RoundKeyCount);
 
 
         [MethodImpl(MethodImplOptions.AggressiveOptimization)]
-        public void Encrypt(ReadOnlySpan<byte> input, Span<byte> output)
+        public readonly void Encrypt(ReadOnlySpan<byte> input, Span<byte> output)
         {
-            ReadOnlySpan<Vector128<byte>> keys = RoundKeys;
-            ReadOnlySpan<Vector128<byte>> inBlocks = MemoryMarshal.Cast<byte, Vector128<byte>>(input);
-            Span<Vector128<byte>> outBlocks = MemoryMarshal.Cast<byte, Vector128<byte>>(output);
+            int blockCount = Math.Min(input.Length, output.Length) >> 4;
 
-            for (int i = 0; i < inBlocks.Length; i++)
+            ref Vector128<byte> inBlock = ref Unsafe.As<byte, Vector128<byte>>(ref MemoryMarshal.GetReference(input));
+            ref Vector128<byte> outBlock = ref Unsafe.As<byte, Vector128<byte>>(ref MemoryMarshal.GetReference(output));
+
+            for (int i = 0; i < blockCount; i++)
             {
-                Vector128<byte> b = inBlocks[i];
+                outBlock = EncryptBlock(inBlock);
 
-                b = Sse2.Xor(b, keys[0]);
-                b = Aes.Encrypt(b, keys[1]);
-                b = Aes.Encrypt(b, keys[2]);
-                b = Aes.Encrypt(b, keys[3]);
-                b = Aes.Encrypt(b, keys[4]);
-                b = Aes.Encrypt(b, keys[5]);
-                b = Aes.Encrypt(b, keys[6]);
-                b = Aes.Encrypt(b, keys[7]);
-                b = Aes.Encrypt(b, keys[8]);
-                b = Aes.Encrypt(b, keys[9]);
-                b = Aes.EncryptLast(b, keys[10]);
-
-                outBlocks[i] = b;
+                inBlock = ref Unsafe.Add(ref inBlock, 1);
+                outBlock = ref Unsafe.Add(ref outBlock, 1);
             }
         }
 
         [MethodImpl(MethodImplOptions.AggressiveOptimization)]
-        public void Decrypt(ReadOnlySpan<byte> input, Span<byte> output)
+        public readonly void Decrypt(ReadOnlySpan<byte> input, Span<byte> output)
+        {
+            int blockCount = Math.Min(input.Length, output.Length) >> 4;
+
+            ref Vector128<byte> inBlock = ref Unsafe.As<byte, Vector128<byte>>(ref MemoryMarshal.GetReference(input));
+            ref Vector128<byte> outBlock = ref Unsafe.As<byte, Vector128<byte>>(ref MemoryMarshal.GetReference(output));
+
+            for (int i = 0; i < blockCount; i++)
+            {
+                outBlock = DecryptBlock(inBlock);
+
+                inBlock = ref Unsafe.Add(ref inBlock, 1);
+                outBlock = ref Unsafe.Add(ref outBlock, 1);
+            }
+        }
+
+        [MethodImpl(MethodImplOptions.AggressiveInlining)]
+        public readonly Vector128<byte> EncryptBlock(Vector128<byte> input)
         {
             ReadOnlySpan<Vector128<byte>> keys = RoundKeys;
-            ReadOnlySpan<Vector128<byte>> inBlocks = MemoryMarshal.Cast<byte, Vector128<byte>>(input);
-            Span<Vector128<byte>> outBlocks = MemoryMarshal.Cast<byte, Vector128<byte>>(output);
 
-            for (int i = 0; i < inBlocks.Length; i++)
-            {
-                Vector128<byte> b = inBlocks[i];
+            Vector128<byte> b = Sse2.Xor(input, keys[0]);
+            b = Aes.Encrypt(b, keys[1]);
+            b = Aes.Encrypt(b, keys[2]);
+            b = Aes.Encrypt(b, keys[3]);
+            b = Aes.Encrypt(b, keys[4]);
+            b = Aes.Encrypt(b, keys[5]);
+            b = Aes.Encrypt(b, keys[6]);
+            b = Aes.Encrypt(b, keys[7]);
+            b = Aes.Encrypt(b, keys[8]);
+            b = Aes.Encrypt(b, keys[9]);
+            return Aes.EncryptLast(b, keys[10]);
+        }
 
-                b = Sse2.Xor(b, keys[10]);
-                b = Aes.Decrypt(b, keys[9]);
-                b = Aes.Decrypt(b, keys[8]);
-                b = Aes.Decrypt(b, keys[7]);
-                b = Aes.Decrypt(b, keys[6]);
-                b = Aes.Decrypt(b, keys[5]);
-                b = Aes.Decrypt(b, keys[4]);
-                b = Aes.Decrypt(b, keys[3]);
-                b = Aes.Decrypt(b, keys[2]);
-                b = Aes.Decrypt(b, keys[1]);
-                b = Aes.DecryptLast(b, keys[0]);
+        [MethodImpl(MethodImplOptions.AggressiveInlining)]
+        public readonly Vector128<byte> DecryptBlock(Vector128<byte> input)
+        {
+            ReadOnlySpan<Vector128<byte>> keys = RoundKeys;
 
-                outBlocks[i] = b;
-            }
+            Vector128<byte> b = Sse2.Xor(input, keys[10]);
+            b = Aes.Decrypt(b, keys[9]);
+            b = Aes.Decrypt(b, keys[8]);
+            b = Aes.Decrypt(b, keys[7]);
+            b = Aes.Decrypt(b, keys[6]);
+            b = Aes.Decrypt(b, keys[5]);
+            b = Aes.Decrypt(b, keys[4]);
+            b = Aes.Decrypt(b, keys[3]);
+            b = Aes.Decrypt(b, keys[2]);
+            b = Aes.Decrypt(b, keys[1]);
+            return Aes.DecryptLast(b, keys[0]);
         }
 
         [MethodImpl(MethodImplOptions.AggressiveOptimization)]

src/LibHac/Crypto2/AesCtrModeHw.cs

@@ -23,42 +23,35 @@ namespace LibHac.Crypto2
 
         public void Transform(ReadOnlySpan<byte> input, Span<byte> output)
         {
-            ReadOnlySpan<Vector128<byte>> keys = _aesCore.RoundKeys;
-            ReadOnlySpan<Vector128<byte>> inBlocks = MemoryMarshal.Cast<byte, Vector128<byte>>(input);
-            Span<Vector128<byte>> outBlocks = MemoryMarshal.Cast<byte, Vector128<byte>>(output);
+            int blockCount = Math.Min(input.Length, output.Length) >> 4;
+
+            ref Vector128<byte> inBlock = ref Unsafe.As<byte, Vector128<byte>>(ref MemoryMarshal.GetReference(input));
+            ref Vector128<byte> outBlock = ref Unsafe.As<byte, Vector128<byte>>(ref MemoryMarshal.GetReference(output));
 
             Vector128<byte> byteSwapMask = Vector128.Create((ulong)0x706050403020100, 0x8090A0B0C0D0E0F).AsByte();
             Vector128<ulong> inc = Vector128.Create((ulong)0, 1);
 
-            var iv = _iv;
+            Vector128<byte> iv = _iv;
             Vector128<ulong> bSwappedIv = Ssse3.Shuffle(iv, byteSwapMask).AsUInt64();
 
-            for (int i = 0; i < inBlocks.Length; i++)
+            for (int i = 0; i < blockCount; i++)
             {
-                Vector128<byte> b = Sse2.Xor(iv, keys[0]);
-                b = Aes.Encrypt(b, keys[1]);
-                b = Aes.Encrypt(b, keys[2]);
-                b = Aes.Encrypt(b, keys[3]);
-                b = Aes.Encrypt(b, keys[4]);
-                b = Aes.Encrypt(b, keys[5]);
-                b = Aes.Encrypt(b, keys[6]);
-                b = Aes.Encrypt(b, keys[7]);
-                b = Aes.Encrypt(b, keys[8]);
-                b = Aes.Encrypt(b, keys[9]);
-                b = Aes.EncryptLast(b, keys[10]);
-
-                outBlocks[i] = Sse2.Xor(inBlocks[i], b);
+                Vector128<byte> encIv = _aesCore.EncryptBlock(iv);
+                outBlock = Sse2.Xor(inBlock, encIv);
 
                 // Increase the counter
                 bSwappedIv = Sse2.Add(bSwappedIv, inc);
                 iv = Ssse3.Shuffle(bSwappedIv.AsByte(), byteSwapMask);
+
+                inBlock = ref Unsafe.Add(ref inBlock, 1);
+                outBlock = ref Unsafe.Add(ref outBlock, 1);
             }
 
             _iv = iv;
 
             if ((input.Length & 0xF) != 0)
             {
-                EncryptCtrPartialBlock(input.Slice(inBlocks.Length * 0x10), output.Slice(outBlocks.Length * 0x10));
+                EncryptCtrPartialBlock(input.Slice(blockCount * 0x10), output.Slice(blockCount * 0x10));
             }
         }
 

src/LibHac/Crypto2/AesEcbMode.cs

@@ -1,10 +1,12 @@
 using System;
+using System.Buffers;
 using System.Security.Cryptography;
 
 namespace LibHac.Crypto2
 {
     public class AesEcbEncryptor : ICipher
     {
+        private const int BufferRentThreshold = 1024;
         private ICryptoTransform _encryptor;
 
         public AesEcbEncryptor(ReadOnlySpan<byte> key)
@@ -21,16 +23,34 @@ namespace LibHac.Crypto2
 
         public void Transform(ReadOnlySpan<byte> input, Span<byte> output)
         {
-            var outputBuffer = new byte[input.Length];
+            if (input.Length < BufferRentThreshold)
+            {
+                var outputBuffer = new byte[input.Length];
+                input.CopyTo(outputBuffer);
 
-            _encryptor.TransformBlock(input.ToArray(), 0, input.Length, outputBuffer, 0);
+                _encryptor.TransformBlock(outputBuffer, 0, input.Length, outputBuffer, 0);
 
-            outputBuffer.CopyTo(output);
+                outputBuffer.CopyTo(output);
+            }
+            else
+            {
+                byte[] outputBuffer = ArrayPool<byte>.Shared.Rent(input.Length);
+                try
+                {
+                    input.CopyTo(outputBuffer);
+
+                    _encryptor.TransformBlock(outputBuffer, 0, input.Length, outputBuffer, 0);
+
+                    outputBuffer.CopyTo(output);
+                }
+                finally { ArrayPool<byte>.Shared.Return(outputBuffer); }
+            }
         }
     }
 
     public class AesEcbDecryptor : ICipher
     {
+        private const int BufferRentThreshold = 1024;
         private ICryptoTransform _decryptor;
 
         public AesEcbDecryptor(ReadOnlySpan<byte> key)
@@ -47,11 +67,28 @@ namespace LibHac.Crypto2
 
         public void Transform(ReadOnlySpan<byte> input, Span<byte> output)
         {
-            var outputBuffer = new byte[input.Length];
+            if (input.Length < BufferRentThreshold)
+            {
+                var outputBuffer = new byte[input.Length];
+                input.CopyTo(outputBuffer);
 
-            _decryptor.TransformBlock(input.ToArray(), 0, input.Length, outputBuffer, 0);
+                _decryptor.TransformBlock(outputBuffer, 0, input.Length, outputBuffer, 0);
 
-            outputBuffer.CopyTo(output);
+                outputBuffer.CopyTo(output);
+            }
+            else
+            {
+                byte[] outputBuffer = ArrayPool<byte>.Shared.Rent(input.Length);
+                try
+                {
+                    input.CopyTo(outputBuffer);
+
+                    _decryptor.TransformBlock(outputBuffer, 0, input.Length, outputBuffer, 0);
+
+                    outputBuffer.CopyTo(output);
+                }
+                finally { ArrayPool<byte>.Shared.Return(outputBuffer); }
+            }
         }
     }
 }

src/LibHac/Crypto2/AesEcbModeHw.cs

@@ -1,10 +1,5 @@
-
-
-#if HAS_INTRINSICS
+#if HAS_INTRINSICS
 using System;
-using System.Runtime.Intrinsics;
-using System.Runtime.InteropServices;
-using System.Runtime.Intrinsics.X86;
 
 namespace LibHac.Crypto2
 {
@@ -20,28 +15,7 @@ namespace LibHac.Crypto2
 
         public void Transform(ReadOnlySpan<byte> input, Span<byte> output)
         {
-            ReadOnlySpan<Vector128<byte>> keys = _aesCore.RoundKeys;
-            ReadOnlySpan<Vector128<byte>> inBlocks = MemoryMarshal.Cast<byte, Vector128<byte>>(input);
-            Span<Vector128<byte>> outBlocks = MemoryMarshal.Cast<byte, Vector128<byte>>(output);
-
-            for (int i = 0; i < inBlocks.Length; i++)
-            {
-                Vector128<byte> b = inBlocks[i];
-
-                b = Sse2.Xor(b, keys[0]);
-                b = Aes.Encrypt(b, keys[1]);
-                b = Aes.Encrypt(b, keys[2]);
-                b = Aes.Encrypt(b, keys[3]);
-                b = Aes.Encrypt(b, keys[4]);
-                b = Aes.Encrypt(b, keys[5]);
-                b = Aes.Encrypt(b, keys[6]);
-                b = Aes.Encrypt(b, keys[7]);
-                b = Aes.Encrypt(b, keys[8]);
-                b = Aes.Encrypt(b, keys[9]);
-                b = Aes.EncryptLast(b, keys[10]);
-
-                outBlocks[i] = b;
-            }
+            _aesCore.Encrypt(input, output);
         }
     }
 

src/LibHac/Crypto2/AesXtsMode.cs

@@ -0,0 +1,208 @@
+using System;
+using System.Buffers;
+using System.Diagnostics;
+using System.Runtime.CompilerServices;
+using System.Runtime.InteropServices;
+using LibHac.Common;
+
+namespace LibHac.Crypto2
+{
+    public class AesXtsCipher : ICipher
+    {
+        private ICipher _dataCipher;
+        private ICipher _tweakCipher;
+        private Buffer16 _iv;
+        private bool _decrypting;
+
+        public AesXtsCipher(ReadOnlySpan<byte> key1, ReadOnlySpan<byte> key2, ReadOnlySpan<byte> iv, bool decrypting)
+        {
+            Debug.Assert(key1.Length == AesCrypto.KeySize128);
+            Debug.Assert(key2.Length == AesCrypto.KeySize128);
+            Debug.Assert(iv.Length == AesCrypto.KeySize128);
+
+            if (decrypting)
+            {
+                _dataCipher = new AesEcbDecryptor(key1);
+            }
+            else
+            {
+                _dataCipher = new AesEcbEncryptor(key1);
+            }
+
+            _tweakCipher = new AesEcbEncryptor(key2);
+
+            _iv = new Buffer16();
+            iv.CopyTo(_iv);
+
+            _decrypting = decrypting;
+        }
+
+        public void Encrypt(ReadOnlySpan<byte> input, Span<byte> output)
+        {
+            int length = Math.Min(input.Length, output.Length);
+            int blockCount = length >> 4;
+            int leftover = length & 0xF;
+
+            // Data units must be at least 1 block long.
+            if (length < AesCrypto.BlockSize)
+                throw new ArgumentException();
+
+            var tweak = new Buffer16();
+
+            _tweakCipher.Transform(_iv, tweak);
+
+            byte[] tweakBufferRented = ArrayPool<byte>.Shared.Rent(blockCount * AesCrypto.BlockSize);
+            try
+            {
+                Span<byte> tweakBuffer = tweakBufferRented.AsSpan(0, blockCount * AesCrypto.BlockSize);
+                tweak = FillTweakBuffer(tweak, MemoryMarshal.Cast<byte, Buffer16>(tweakBuffer));
+
+                Util.XorArrays(output, input, tweakBuffer);
+                _dataCipher.Transform(output.Slice(0, blockCount * AesCrypto.BlockSize), output);
+                Util.XorArrays(output, output, tweakBuffer);
+            }
+            finally { ArrayPool<byte>.Shared.Return(tweakBufferRented); }
+
+            if (leftover != 0)
+            {
+                ref Buffer16 inBlock =
+                    ref Unsafe.Add(ref Unsafe.As<byte, Buffer16>(ref MemoryMarshal.GetReference(input)), blockCount);
+
+                ref Buffer16 outBlock =
+                    ref Unsafe.Add(ref Unsafe.As<byte, Buffer16>(ref MemoryMarshal.GetReference(output)), blockCount);
+
+                ref Buffer16 prevOutBlock = ref Unsafe.Subtract(ref outBlock, 1);
+
+                var tmp = new Buffer16();
+
+                for (int i = 0; i < leftover; i++)
+                {
+                    outBlock[i] = prevOutBlock[i];
+                    tmp[i] = inBlock[i];
+                }
+
+                for (int i = leftover; i < AesCrypto.BlockSize; i++)
+                {
+                    tmp[i] = prevOutBlock[i];
+                }
+
+                XorBuffer(ref tmp, ref tmp, ref tweak);
+                _dataCipher.Transform(tmp, tmp);
+                XorBuffer(ref prevOutBlock, ref tmp, ref tweak);
+            }
+        }
+
+        public void Decrypt(ReadOnlySpan<byte> input, Span<byte> output)
+        {
+            int length = Math.Min(input.Length, output.Length);
+            int blockCount = length >> 4;
+            int leftover = length & 0xF;
+
+            // Data units must be at least 1 block long.
+            if (length < AesCrypto.BlockSize)
+                throw new ArgumentException();
+
+            if (leftover != 0) blockCount--;
+
+            var tweak = new Buffer16();
+
+            _tweakCipher.Transform(_iv, tweak);
+
+            if (blockCount > 0)
+            {
+                byte[] tweakBufferRented = ArrayPool<byte>.Shared.Rent(blockCount * AesCrypto.BlockSize);
+                try
+                {
+                    Span<byte> tweakBuffer = tweakBufferRented.AsSpan(0, blockCount * AesCrypto.BlockSize);
+                    tweak = FillTweakBuffer(tweak, MemoryMarshal.Cast<byte, Buffer16>(tweakBuffer));
+
+                    Util.XorArrays(output, input, tweakBuffer);
+                    _dataCipher.Transform(output.Slice(0, blockCount * AesCrypto.BlockSize), output);
+                    Util.XorArrays(output, output, tweakBuffer);
+                }
+                finally { ArrayPool<byte>.Shared.Return(tweakBufferRented); }
+            }
+
+            if (leftover != 0)
+            {
+                Buffer16 finalTweak = tweak;
+                Gf128Mul(ref finalTweak);
+
+                ref Buffer16 inBlock =
+                    ref Unsafe.Add(ref Unsafe.As<byte, Buffer16>(ref MemoryMarshal.GetReference(input)), blockCount);
+
+                ref Buffer16 outBlock =
+                    ref Unsafe.Add(ref Unsafe.As<byte, Buffer16>(ref MemoryMarshal.GetReference(output)), blockCount);
+
+                var tmp = new Buffer16();
+
+                XorBuffer(ref tmp, ref inBlock, ref finalTweak);
+                _dataCipher.Transform(tmp, tmp);
+                XorBuffer(ref outBlock, ref tmp, ref finalTweak);
+
+                ref Buffer16 finalOutBlock = ref Unsafe.Add(ref outBlock, 1);
+                ref Buffer16 finalInBlock = ref Unsafe.Add(ref inBlock, 1);
+
+                for (int i = 0; i < leftover; i++)
+                {
+                    finalOutBlock[i] = outBlock[i];
+                    tmp[i] = finalInBlock[i];
+                }
+
+                for (int i = leftover; i < AesCrypto.BlockSize; i++)
+                {
+                    tmp[i] = outBlock[i];
+                }
+
+                XorBuffer(ref tmp, ref tmp, ref tweak);
+                _dataCipher.Transform(tmp, tmp);
+                XorBuffer(ref outBlock, ref tmp, ref tweak);
+            }
+        }
+
+        public void Transform(ReadOnlySpan<byte> input, Span<byte> output)
+        {
+            if (_decrypting)
+            {
+                Decrypt(input, output);
+            }
+            else
+            {
+                Encrypt(input, output);
+            }
+        }
+
+        private static Buffer16 FillTweakBuffer(Buffer16 initialTweak, Span<Buffer16> tweakBuffer)
+        {
+            for (int i = 0; i < tweakBuffer.Length; i++)
+            {
+                tweakBuffer[i] = initialTweak;
+                Gf128Mul(ref initialTweak);
+            }
+
+            return initialTweak;
+        }
+
+        [MethodImpl(MethodImplOptions.AggressiveInlining)]
+        private static void Gf128Mul(ref Buffer16 buffer)
+        {
+            Span<ulong> b = buffer.AsSpan<ulong>();
+
+            ulong tt = (ulong)((long)b[1] >> 63) & 0x87;
+
+            b[1] = (b[1] << 1) | (b[0] >> 63);
+            b[0] = (b[0] << 1) ^ tt;
+        }
+
+        [MethodImpl(MethodImplOptions.AggressiveInlining)]
+        private static void XorBuffer(ref Buffer16 output, ref Buffer16 input1, ref Buffer16 input2)
+        {
+            Span<ulong> outputS = output.AsSpan<ulong>();
+            Span<ulong> input1S = input1.AsSpan<ulong>();
+            Span<ulong> input2S = input2.AsSpan<ulong>();
+
+            outputS[0] = input1S[0] ^ input2S[0];
+            outputS[1] = input1S[1] ^ input2S[1];
+        }
+    }
+}

src/LibHac/Crypto2/AesXtsModeHw.cs

@@ -0,0 +1,186 @@
+#if HAS_INTRINSICS
+using System;
+using System.Diagnostics;
+using System.Runtime.CompilerServices;
+using System.Runtime.InteropServices;
+using System.Runtime.Intrinsics;
+using System.Runtime.Intrinsics.X86;
+using LibHac.Common;
+
+namespace LibHac.Crypto2
+{
+    public class AesXtsCipherHw : ICipher
+    {
+        private AesCoreNi _dataAesCore;
+        private AesCoreNi _tweakAesCore;
+        private Vector128<byte> _iv;
+        private bool _decrypting;
+
+        public AesXtsCipherHw(ReadOnlySpan<byte> key1, ReadOnlySpan<byte> key2, ReadOnlySpan<byte> iv, bool decrypting)
+        {
+            Debug.Assert(key1.Length == AesCrypto.KeySize128);
+            Debug.Assert(key2.Length == AesCrypto.KeySize128);
+            Debug.Assert(iv.Length == AesCrypto.KeySize128);
+
+            _dataAesCore = new AesCoreNi();
+            _dataAesCore.Initialize(key1, decrypting);
+
+            _tweakAesCore = new AesCoreNi();
+            _tweakAesCore.Initialize(key2, false);
+
+            _iv = Unsafe.ReadUnaligned<Vector128<byte>>(ref MemoryMarshal.GetReference(iv));
+
+            _decrypting = decrypting;
+        }
+
+        public void Encrypt(ReadOnlySpan<byte> input, Span<byte> output)
+        {
+            int length = Math.Min(input.Length, output.Length);
+            int blockCount = length >> 4;
+            int leftover = length & 0xF;
+
+            Debug.Assert(blockCount > 0);
+
+            ref Vector128<byte> inBlock = ref Unsafe.As<byte, Vector128<byte>>(ref MemoryMarshal.GetReference(input));
+            ref Vector128<byte> outBlock = ref Unsafe.As<byte, Vector128<byte>>(ref MemoryMarshal.GetReference(output));
+
+            Vector128<byte> mask = Vector128.Create(0x87, 1).AsByte();
+
+            Vector128<byte> tweak = _tweakAesCore.EncryptBlock(_iv);
+
+            for (int i = 0; i < blockCount; i++)
+            {
+                Vector128<byte> tmp = Sse2.Xor(inBlock, tweak);
+                tmp = _dataAesCore.EncryptBlock(tmp);
+                outBlock = Sse2.Xor(tmp, tweak);
+
+                tweak = Gf128Mul(tweak, mask);
+
+                inBlock = ref Unsafe.Add(ref inBlock, 1);
+                outBlock = ref Unsafe.Add(ref outBlock, 1);
+            }
+
+            if (leftover != 0)
+            {
+                EncryptPartialFinalBlock(ref inBlock, ref outBlock, tweak, leftover);
+            }
+        }
+
+        public void Decrypt(ReadOnlySpan<byte> input, Span<byte> output)
+        {
+            int length = Math.Min(input.Length, output.Length);
+            int blockCount = length >> 4;
+            int leftover = length & 0xF;
+
+            Debug.Assert(blockCount > 0);
+
+            if (leftover != 0) blockCount--;
+
+            ref Vector128<byte> inBlock = ref Unsafe.As<byte, Vector128<byte>>(ref MemoryMarshal.GetReference(input));
+            ref Vector128<byte> outBlock = ref Unsafe.As<byte, Vector128<byte>>(ref MemoryMarshal.GetReference(output));
+
+            Vector128<byte> mask = Vector128.Create(0x87, 1).AsByte();
+
+            Vector128<byte> tweak = _tweakAesCore.EncryptBlock(_iv);
+
+            for (int i = 0; i < blockCount; i++)
+            {
+                Vector128<byte> tmp = Sse2.Xor(inBlock, tweak);
+                tmp = _dataAesCore.DecryptBlock(tmp);
+                outBlock = Sse2.Xor(tmp, tweak);
+
+                tweak = Gf128Mul(tweak, mask);
+
+                inBlock = ref Unsafe.Add(ref inBlock, 1);
+                outBlock = ref Unsafe.Add(ref outBlock, 1);
+            }
+
+            if (leftover != 0)
+            {
+                DecryptPartialFinalBlock(ref inBlock, ref outBlock, tweak, mask, leftover);
+            }
+        }
+
+        // ReSharper disable once RedundantAssignment
+        private void DecryptPartialFinalBlock(ref Vector128<byte> input, ref Vector128<byte> output,
+            Vector128<byte> tweak, Vector128<byte> mask, int finalBlockLength)
+        {
+            Vector128<byte> finalTweak = Gf128Mul(tweak, mask);
+
+            Vector128<byte> tmp = Sse2.Xor(input, finalTweak);
+            tmp = _dataAesCore.DecryptBlock(tmp);
+            output = Sse2.Xor(tmp, finalTweak);
+
+            var x = new Buffer16();
+            ref Buffer16 outBuf = ref Unsafe.As<Vector128<byte>, Buffer16>(ref output);
+            ref Buffer16 nextInBuf = ref Unsafe.As<Vector128<byte>, Buffer16>(ref Unsafe.Add(ref input, 1));
+            ref Buffer16 nextOutBuf = ref Unsafe.As<Vector128<byte>, Buffer16>(ref Unsafe.Add(ref output, 1));
+
+            for (int i = 0; i < finalBlockLength; i++)
+            {
+                nextOutBuf[i] = outBuf[i];
+                x[i] = nextInBuf[i];
+            }
+
+            for (int i = finalBlockLength; i < 16; i++)
+            {
+                x[i] = outBuf[i];
+            }
+
+            tmp = Sse2.Xor(x.As<Vector128<byte>>(), tweak);
+            tmp = _dataAesCore.DecryptBlock(tmp);
+            output = Sse2.Xor(tmp, tweak);
+        }
+
+        private void EncryptPartialFinalBlock(ref Vector128<byte> input, ref Vector128<byte> output,
+            Vector128<byte> tweak, int finalBlockLength)
+        {
+            ref Vector128<byte> prevOutBlock = ref Unsafe.Subtract(ref output, 1);
+
+            var x = new Buffer16();
+            ref Buffer16 outBuf = ref Unsafe.As<Vector128<byte>, Buffer16>(ref output);
+            ref Buffer16 inBuf = ref Unsafe.As<Vector128<byte>, Buffer16>(ref input);
+            ref Buffer16 prevOutBuf = ref Unsafe.As<Vector128<byte>, Buffer16>(ref prevOutBlock);
+
+            for (int i = 0; i < finalBlockLength; i++)
+            {
+                outBuf[i] = prevOutBuf[i];
+                x[i] = inBuf[i];
+            }
+
+            for (int i = finalBlockLength; i < 16; i++)
+            {
+                x[i] = prevOutBuf[i];
+            }
+
+            Vector128<byte> tmp = Sse2.Xor(x.As<Vector128<byte>>(), tweak);
+            tmp = _dataAesCore.EncryptBlock(tmp);
+            prevOutBlock = Sse2.Xor(tmp, tweak);
+        }
+
+        public void Transform(ReadOnlySpan<byte> input, Span<byte> output)
+        {
+            if (_decrypting)
+            {
+                Decrypt(input, output);
+            }
+            else
+            {
+                Encrypt(input, output);
+            }
+        }
+
+        [MethodImpl(MethodImplOptions.AggressiveInlining)]
+        private static Vector128<byte> Gf128Mul(Vector128<byte> iv, Vector128<byte> mask)
+        {
+            Vector128<byte> tmp1 = Sse2.Add(iv.AsUInt64(), iv.AsUInt64()).AsByte();
+
+            Vector128<byte> tmp2 = Sse2.Shuffle(iv.AsInt32(), 0x13).AsByte();
+            tmp2 = Sse2.ShiftRightArithmetic(tmp2.AsInt32(), 31).AsByte();
+            tmp2 = Sse2.And(mask, tmp2);
+
+            return Sse2.Xor(tmp1, tmp2);
+        }
+    }
+}
+#endif

src/LibHac/FsService/SaveDataIndexer.cs

@@ -546,6 +546,7 @@ namespace LibHac.FsService
 
         private void CloseReader(SaveDataInfoReader reader)
         {
+            // ReSharper disable once RedundantAssignment
             bool wasRemoved = OpenedReaders.Remove(reader);
 
             Debug.Assert(wasRemoved);

src/LibHac/Util.cs

@@ -145,13 +145,13 @@ namespace LibHac
             return true;
         }
 
-        public static void XorArrays(Span<byte> transformData, Span<byte> xorData)
+        public static void XorArrays(Span<byte> transformData, ReadOnlySpan<byte> xorData)
         {
             int sisdStart = 0;
             if (Vector.IsHardwareAccelerated)
             {
                 Span<Vector<byte>> dataVec = MemoryMarshal.Cast<byte, Vector<byte>>(transformData);
-                Span<Vector<byte>> xorVec = MemoryMarshal.Cast<byte, Vector<byte>>(xorData);
+                ReadOnlySpan<Vector<byte>> xorVec = MemoryMarshal.Cast<byte, Vector<byte>>(xorData);
                 sisdStart = dataVec.Length * Vector<byte>.Count;
 
                 for (int i = 0; i < dataVec.Length; i++)
@@ -166,6 +166,33 @@ namespace LibHac
             }
         }
 
+        public static void XorArrays(Span<byte> output, ReadOnlySpan<byte> input1, ReadOnlySpan<byte> input2)
+        {
+            int length = Math.Min(input1.Length, input2.Length);
+
+            int sisdStart = 0;
+            if (Vector.IsHardwareAccelerated)
+            {
+                int lengthVec = length / Vector<byte>.Count;
+
+                Span<Vector<byte>> outputVec = MemoryMarshal.Cast<byte, Vector<byte>>(output);
+                ReadOnlySpan<Vector<byte>> input1Vec = MemoryMarshal.Cast<byte, Vector<byte>>(input1);
+                ReadOnlySpan<Vector<byte>> input2Vec = MemoryMarshal.Cast<byte, Vector<byte>>(input2);
+
+                sisdStart = lengthVec * Vector<byte>.Count;
+
+                for (int i = 0; i < lengthVec; i++)
+                {
+                    outputVec[i] = input1Vec[i] ^ input2Vec[i];
+                }
+            }
+
+            for (int i = sisdStart; i < length; i++)
+            {
+                output[i] = (byte)(input1[i] ^ input2[i]);
+            }
+        }
+
         public static void CopyStream(this Stream input, Stream output, long length, IProgressReport progress = null)
         {
             const int bufferSize = 0x8000;

src/hactoolnet/ProcessBench.cs

@@ -116,7 +116,8 @@ namespace hactoolnet
             logger.LogMessage($"{label}{averageRate}/s, fastest run: {fastestRate}/s, slowest run: {slowestRate}/s");
         }
 
-        private static void RunCipherBenchmark(Func<ICipher> cipherNet, Func<ICipher> cipherLibHac, string label, IProgressReport logger)
+        private static void RunCipherBenchmark(Func<ICipher> cipherNet, Func<ICipher> cipherLibHac, bool benchBlocked,
+            string label, IProgressReport logger)
         {
             var srcData = new byte[Size];
 
@@ -131,14 +132,23 @@ namespace hactoolnet
             if (AesCrypto.IsAesNiSupported()) CipherBenchmark(srcData, dstDataLh, cipherLibHac, Iterations, "LibHac impl: ", logger);
             CipherBenchmark(srcData, dstDataNet, cipherNet, Iterations, ".NET impl:   ", logger);
 
-            if (AesCrypto.IsAesNiSupported()) CipherBenchmarkBlocked(srcData, dstDataBlockedLh, cipherLibHac, Iterations / 5, "LibHac impl (blocked): ", logger);
-            CipherBenchmarkBlocked(srcData, dstDataBlockedNet, cipherNet, Iterations / 5, ".NET impl (blocked):   ", logger);
+            if (benchBlocked)
+            {
+                if (AesCrypto.IsAesNiSupported())
+                    CipherBenchmarkBlocked(srcData, dstDataBlockedLh, cipherLibHac, Iterations / 5, "LibHac impl (blocked): ", logger);
+
+                CipherBenchmarkBlocked(srcData, dstDataBlockedNet, cipherNet, Iterations / 5, ".NET impl (blocked):   ", logger);
+            }
 
             if (AesCrypto.IsAesNiSupported())
             {
                 logger.LogMessage($"{dstDataLh.SequenceEqual(dstDataNet)}");
-                logger.LogMessage($"{dstDataLh.SequenceEqual(dstDataBlockedLh)}");
-                logger.LogMessage($"{dstDataLh.SequenceEqual(dstDataBlockedNet)}");
+
+                if (benchBlocked)
+                {
+                    logger.LogMessage($"{dstDataLh.SequenceEqual(dstDataBlockedLh)}");
+                    logger.LogMessage($"{dstDataLh.SequenceEqual(dstDataBlockedNet)}");
+                }
             }
         }
 
@@ -197,12 +207,12 @@ namespace hactoolnet
                     Func<ICipher> encryptorNet = () => AesCrypto.CreateEcbEncryptor(new byte[0x10], true);
                     Func<ICipher> encryptorLh = () => AesCrypto.CreateEcbEncryptor(new byte[0x10]);
 
-                    RunCipherBenchmark(encryptorNet, encryptorLh, "AES-ECB encrypt", ctx.Logger);
+                    RunCipherBenchmark(encryptorNet, encryptorLh, true, "AES-ECB encrypt", ctx.Logger);
 
                     Func<ICipher> decryptorNet = () => AesCrypto.CreateEcbDecryptor(new byte[0x10], true);
                     Func<ICipher> decryptorLh = () => AesCrypto.CreateEcbDecryptor(new byte[0x10]);
 
-                    RunCipherBenchmark(decryptorNet, decryptorLh, "AES-ECB decrypt", ctx.Logger);
+                    RunCipherBenchmark(decryptorNet, decryptorLh, true, "AES-ECB decrypt", ctx.Logger);
 
                     break;
                 }
@@ -211,12 +221,12 @@ namespace hactoolnet
                     Func<ICipher> encryptorNet = () => AesCrypto.CreateCbcEncryptor(new byte[0x10], new byte[0x10], true);
                     Func<ICipher> encryptorLh = () => AesCrypto.CreateCbcEncryptor(new byte[0x10], new byte[0x10]);
 
-                    RunCipherBenchmark(encryptorNet, encryptorLh, "AES-CBC encrypt", ctx.Logger);
+                    RunCipherBenchmark(encryptorNet, encryptorLh, true, "AES-CBC encrypt", ctx.Logger);
 
                     Func<ICipher> decryptorNet = () => AesCrypto.CreateCbcDecryptor(new byte[0x10], new byte[0x10], true);
                     Func<ICipher> decryptorLh = () => AesCrypto.CreateCbcDecryptor(new byte[0x10], new byte[0x10]);
 
-                    RunCipherBenchmark(decryptorNet, decryptorLh, "AES-CBC decrypt", ctx.Logger);
+                    RunCipherBenchmark(decryptorNet, decryptorLh, true, "AES-CBC decrypt", ctx.Logger);
 
                     break;
                 }
@@ -226,7 +236,21 @@ namespace hactoolnet
                     Func<ICipher> encryptorNet = () => AesCrypto.CreateCtrEncryptor(new byte[0x10], new byte[0x10], true);
                     Func<ICipher> encryptorLh = () => AesCrypto.CreateCtrEncryptor(new byte[0x10], new byte[0x10]);
 
-                    RunCipherBenchmark(encryptorNet, encryptorLh, "AES-CTR", ctx.Logger);
+                    RunCipherBenchmark(encryptorNet, encryptorLh, true, "AES-CTR", ctx.Logger);
+
+                    break;
+                }
+                case "aesxtsnew":
+                {
+                    Func<ICipher> encryptorNet = () => AesCrypto.CreateXtsEncryptor(new byte[0x10], new byte[0x10], new byte[0x10], true);
+                    Func<ICipher> encryptorLh = () => AesCrypto.CreateXtsEncryptor(new byte[0x10], new byte[0x10], new byte[0x10]);
+
+                    RunCipherBenchmark(encryptorNet, encryptorLh, false, "AES-XTS encrypt", ctx.Logger);
+
+                    Func<ICipher> decryptorNet = () => AesCrypto.CreateXtsDecryptor(new byte[0x10], new byte[0x10], new byte[0x10], true);
+                    Func<ICipher> decryptorLh = () => AesCrypto.CreateXtsDecryptor(new byte[0x10], new byte[0x10], new byte[0x10]);
+
+                    RunCipherBenchmark(decryptorNet, decryptorLh, false, "AES-XTS decrypt", ctx.Logger);
 
                     break;
                 }

tests/LibHac.Tests/BufferStructTests.cs

@@ -0,0 +1,88 @@
+using System;
+using System.Runtime.CompilerServices;
+using System.Runtime.InteropServices;
+using LibHac.Common;
+using Xunit;
+
+namespace LibHac.Tests
+{
+    public class BufferStructTests
+    {
+        [Fact]
+        public static void BufferIndexer()
+        {
+            var buffer = new Buffer16();
+
+            buffer[0] = 5;
+            buffer[1] = 6;
+
+            Assert.Equal(5, buffer[0]);
+            Assert.Equal(6, buffer[1]);
+        }
+
+        [Fact]
+        public static void CastBufferToByteSpan()
+        {
+            var buffer = new Buffer16();
+
+            Span<byte> byteSpan = buffer.Bytes;
+
+            Assert.Equal(16, byteSpan.Length);
+            Assert.True(Unsafe.AreSame(ref Unsafe.As<Buffer16, byte>(ref buffer), ref byteSpan[0]));
+        }
+
+        [Fact]
+        public static void CastBufferToByteSpanImplicit()
+        {
+            var buffer = new Buffer16();
+
+            Span<byte> byteSpan = buffer;
+
+            Assert.Equal(16, byteSpan.Length);
+            Assert.True(Unsafe.AreSame(ref Unsafe.As<Buffer16, byte>(ref buffer), ref byteSpan[0]));
+        }
+
+        [Fact]
+        public static void CastBufferToReadOnlyByteSpanImplicit()
+        {
+            var buffer = new Buffer16();
+
+            ReadOnlySpan<byte> byteSpan = buffer;
+
+            Assert.Equal(16, byteSpan.Length);
+            Assert.True(Unsafe.AreSame(ref Unsafe.As<Buffer16, byte>(ref buffer), ref Unsafe.AsRef(byteSpan[0])));
+        }
+
+        [Fact]
+        public static void CastBufferToSpan()
+        {
+            var buffer = new Buffer16();
+
+            Span<ulong> ulongSpan = buffer.AsSpan<ulong>();
+
+            Assert.Equal(2, ulongSpan.Length);
+            Assert.True(Unsafe.AreSame(ref Unsafe.As<Buffer16, ulong>(ref buffer), ref ulongSpan[0]));
+        }
+
+        [Fact]
+        public static void CastBufferToStruct()
+        {
+            var buffer = new Buffer16();
+
+            ref ulong ulongSpan = ref buffer.As<ulong>();
+
+            Assert.True(Unsafe.AreSame(ref Unsafe.As<Buffer16, ulong>(ref buffer), ref ulongSpan));
+        }
+
+        [Fact]
+        public static void CastBufferToLargerStruct()
+        {
+            var buffer = new Buffer16();
+
+            Assert.Throws<ArgumentException>(() => buffer.As<Struct32Bytes>());
+        }
+
+        [StructLayout(LayoutKind.Sequential, Size = 32)]
+        private struct Struct32Bytes { }
+    }
+}

tests/LibHac.Tests/CryptoTests/AesXtsTests.cs

@@ -0,0 +1,75 @@
+using System;
+using System.Collections.Generic;
+using System.Linq;
+using LibHac.Crypto2;
+using Xunit;
+
+namespace LibHac.Tests.CryptoTests
+{
+    public class AesXtsTests
+    {
+        public static TheoryData<EncryptionTestVector> EncryptTestVectors =
+            RemovePartialByteTests(RspReader.ReadTestVectors(true, "XTSGenAES128.rsp"));
+
+        public static TheoryData<EncryptionTestVector> DecryptTestVectors =
+            RemovePartialByteTests(RspReader.ReadTestVectors(false, "XTSGenAES128.rsp"));
+
+        // The XTS implementation only supports multiples of whole bytes
+        private static TheoryData<EncryptionTestVector> RemovePartialByteTests(TheoryData<EncryptionTestVector> input)
+        {
+            IEnumerable<EncryptionTestVector> filteredTestVectors = input
+                .Select(x => x[0])
+                .Cast<EncryptionTestVector>()
+                .Where(x => x.DataUnitLength % 8 == 0);
+
+            var output = new TheoryData<EncryptionTestVector>();
+
+            foreach (EncryptionTestVector item in filteredTestVectors)
+            {
+                output.Add(item);
+            }
+
+            return output;
+        }
+
+        [Theory]
+        [MemberData(nameof(EncryptTestVectors))]
+        public static void Encrypt(EncryptionTestVector tv)
+        {
+            Span<byte> key1 = tv.Key.AsSpan(0, 0x10);
+            Span<byte> key2 = tv.Key.AsSpan(0x10, 0x10);
+
+            Common.CipherTestCore(tv.PlainText, tv.CipherText, AesCrypto.CreateXtsEncryptor(key1, key2, tv.Iv, true));
+        }
+
+        [Theory]
+        [MemberData(nameof(DecryptTestVectors))]
+        public static void Decrypt(EncryptionTestVector tv)
+        {
+            Span<byte> key1 = tv.Key.AsSpan(0, 0x10);
+            Span<byte> key2 = tv.Key.AsSpan(0x10, 0x10);
+
+            Common.CipherTestCore(tv.CipherText, tv.PlainText, AesCrypto.CreateXtsDecryptor(key1, key2, tv.Iv, true));
+        }
+
+        [AesIntrinsicsRequiredTheory]
+        [MemberData(nameof(EncryptTestVectors))]
+        public static void EncryptIntrinsics(EncryptionTestVector tv)
+        {
+            Span<byte> key1 = tv.Key.AsSpan(0, 0x10);
+            Span<byte> key2 = tv.Key.AsSpan(0x10, 0x10);
+
+            Common.CipherTestCore(tv.PlainText, tv.CipherText, AesCrypto.CreateXtsEncryptor(key1, key2, tv.Iv));
+        }
+
+        [AesIntrinsicsRequiredTheory]
+        [MemberData(nameof(DecryptTestVectors))]
+        public static void DecryptIntrinsics(EncryptionTestVector tv)
+        {
+            Span<byte> key1 = tv.Key.AsSpan(0, 0x10);
+            Span<byte> key2 = tv.Key.AsSpan(0x10, 0x10);
+
+            Common.CipherTestCore(tv.CipherText, tv.PlainText, AesCrypto.CreateXtsDecryptor(key1, key2, tv.Iv));
+        }
+    }
+}

tests/LibHac.Tests/CryptoTests/RspReader.cs

@@ -56,21 +56,27 @@ namespace LibHac.Tests.CryptoTests
 
                 canOutputVector = true;
 
-                switch (kvp[0])
+                switch (kvp[0].ToUpperInvariant())
                 {
                     case "COUNT":
                         testVector.Count = int.Parse(kvp[1]);
                         break;
+                    case "DATAUNITLEN":
+                        testVector.DataUnitLength = int.Parse(kvp[1]);
+                        break;
                     case "KEY":
                         testVector.Key = kvp[1].ToBytes();
                         break;
                     case "IV":
+                    case "I":
                         testVector.Iv = kvp[1].ToBytes();
                         break;
                     case "PLAINTEXT":
+                    case "PT":
                         testVector.PlainText = kvp[1].ToBytes();
                         break;
                     case "CIPHERTEXT":
+                    case "CT":
                         testVector.CipherText = kvp[1].ToBytes();
                         break;
                 }
@@ -94,7 +100,7 @@ namespace LibHac.Tests.CryptoTests
                     }
                 }
             }
-
+            
             return testVectors;
         }
     }
@@ -103,6 +109,7 @@ namespace LibHac.Tests.CryptoTests
     {
         public bool Encrypt { get; set; }
         public int Count { get; set; }
+        public int DataUnitLength { get; set; }
         public byte[] Key { get; set; }
         public byte[] Iv { get; set; }
         public byte[] PlainText { get; set; }