Improve XmlDictionaryWriter UTF8 encoding performance

src/libraries/System.Private.DataContractSerialization/src/System/Runtime/Serialization/DataContractSerializer.cs

@@ -33,13 +33,27 @@ public sealed class DataContractSerializer : XmlObjectSerializer
         private static SerializationOption s_option = IsReflectionBackupAllowed() ? SerializationOption.ReflectionAsBackup : SerializationOption.CodeGenOnly;
         private static bool s_optionAlreadySet;
 
-        internal static UTF8Encoding UTF8NoBom { get; } = new UTF8Encoding(encoderShouldEmitUTF8Identifier: false, throwOnInvalidBytes: false);
-        internal static UTF8Encoding ValidatingUTF8 { get; } = new UTF8Encoding(encoderShouldEmitUTF8Identifier: false, throwOnInvalidBytes: true);
+        internal sealed class SealedUTF8Encoding : UTF8Encoding
+        {
+            public SealedUTF8Encoding(bool encoderShouldEmitUTF8Identifier, bool throwOnInvalidBytes)
+                : base(encoderShouldEmitUTF8Identifier, throwOnInvalidBytes)
+            { }
+        }
+
+        internal sealed class SealedUnicodeEncoding : UnicodeEncoding
+        {
+            public SealedUnicodeEncoding(bool bigEndian, bool byteOrderMark, bool throwOnInvalidBytes)
+                : base(bigEndian, byteOrderMark, throwOnInvalidBytes)
+            { }
+        }
+
+        internal static SealedUTF8Encoding UTF8NoBom { get; } = new SealedUTF8Encoding(encoderShouldEmitUTF8Identifier: false, throwOnInvalidBytes: false);
+        internal static SealedUTF8Encoding ValidatingUTF8 { get; } = new SealedUTF8Encoding(encoderShouldEmitUTF8Identifier: false, throwOnInvalidBytes: true);
 
-        internal static UnicodeEncoding UTF16NoBom { get; } = new UnicodeEncoding(bigEndian: false, byteOrderMark: false, throwOnInvalidBytes: false);
-        internal static UnicodeEncoding BEUTF16NoBom { get; } = new UnicodeEncoding(bigEndian: true, byteOrderMark: false, throwOnInvalidBytes: false);
-        internal static UnicodeEncoding ValidatingUTF16 { get; } = new UnicodeEncoding(bigEndian: false, byteOrderMark: false, throwOnInvalidBytes: true);
-        internal static UnicodeEncoding ValidatingBEUTF16 { get; } = new UnicodeEncoding(bigEndian: true, byteOrderMark: false, throwOnInvalidBytes: true);
+        internal static SealedUnicodeEncoding UTF16NoBom { get; } = new SealedUnicodeEncoding(bigEndian: false, byteOrderMark: false, throwOnInvalidBytes: false);
+        internal static SealedUnicodeEncoding BEUTF16NoBom { get; } = new SealedUnicodeEncoding(bigEndian: true, byteOrderMark: false, throwOnInvalidBytes: false);
+        internal static SealedUnicodeEncoding ValidatingUTF16 { get; } = new SealedUnicodeEncoding(bigEndian: false, byteOrderMark: false, throwOnInvalidBytes: true);
+        internal static SealedUnicodeEncoding ValidatingBEUTF16 { get; } = new SealedUnicodeEncoding(bigEndian: true, byteOrderMark: false, throwOnInvalidBytes: true);
 
         internal static Base64Encoding Base64Encoding { get; } = new Base64Encoding();
         internal static BinHexEncoding BinHexEncoding { get; } = new BinHexEncoding();

src/libraries/System.Private.DataContractSerialization/src/System/Xml/XmlStreamNodeWriter.cs

@@ -3,6 +3,7 @@
 
 using System.IO;
 using System.Text;
+using System.Runtime.InteropServices;
 using System.Runtime.Serialization;
 using System.Threading.Tasks;
 
@@ -336,32 +337,28 @@ protected unsafe void UnsafeWriteUnicodeChars(char* chars, int charCount)
 
         protected unsafe int UnsafeGetUnicodeChars(char* chars, int charCount, byte[] buffer, int offset)
         {
-            char* charsMax = chars + charCount;
-            while (chars < charsMax)
+            if (BitConverter.IsLittleEndian)
             {
-                char value = *chars++;
-                buffer[offset++] = (byte)value;
-                value >>= 8;
-                buffer[offset++] = (byte)value;
+                new ReadOnlySpan<char>(chars, charCount)
+                    .CopyTo(MemoryMarshal.Cast<byte, char>(buffer.AsSpan(offset)));
             }
+            else
+            {
+                char* charsMax = chars + charCount;
+                while (chars < charsMax)
+                {
+                    char value = *chars++;
+                    buffer[offset++] = (byte)value;
+                    buffer[offset++] = (byte)(value >> 8);
+                }
+            }
+
             return charCount * 2;
         }
 
         protected unsafe int UnsafeGetUTF8Length(char* chars, int charCount)
         {
-            char* charsMax = chars + charCount;
-            while (chars < charsMax)
-            {
-                if (*chars >= 0x80)
-                    break;
-
-                chars++;
-            }
-
-            if (chars == charsMax)
-                return charCount;
-
-            return (int)(chars - (charsMax - charCount)) + (_encoding ?? DataContractSerializer.ValidatingUTF8).GetByteCount(chars, (int)(charsMax - chars));
+            return (_encoding ?? DataContractSerializer.ValidatingUTF8).GetByteCount(chars, charCount);
         }
 
         protected unsafe int UnsafeGetUTF8Chars(char* chars, int charCount, byte[] buffer, int offset)
@@ -370,39 +367,32 @@ protected unsafe int UnsafeGetUTF8Chars(char* chars, int charCount, byte[] buffe
             {
                 fixed (byte* _bytes = &buffer[offset])
                 {
-                    byte* bytes = _bytes;
-                    byte* bytesMax = &bytes[buffer.Length - offset];
-                    char* charsMax = &chars[charCount];
-
-                    while (true)
+                    // Fast path for small strings, use Encoding.GetBytes for larger strings since it is faster when vectorization is possible
+                    if ((uint)charCount < 16)
                     {
+                        byte* bytes = _bytes;
+                        char* charsMax = &chars[charCount];
+
                         while (chars < charsMax)
                         {
                             char t = *chars;
                             if (t >= 0x80)
-                                break;
+                                goto NonAscii;
 
                             *bytes = (byte)t;
                             bytes++;
                             chars++;
                         }
+                        return charCount;
 
-                        if (chars >= charsMax)
-                            break;
-
-                        char* charsStart = chars;
-                        while (chars < charsMax && *chars >= 0x80)
-                        {
-                            chars++;
-                        }
-
-                        bytes += (_encoding ?? DataContractSerializer.ValidatingUTF8).GetBytes(charsStart, (int)(chars - charsStart), bytes, (int)(bytesMax - bytes));
-
-                        if (chars >= charsMax)
-                            break;
+                    NonAscii:
+                        byte* bytesMax = _bytes + buffer.Length - offset;
+                        return (int)(bytes - _bytes) + (_encoding ?? Encoding.UTF8).GetBytes(chars, (int)(charsMax - chars), bytes, (int)(bytesMax - bytes));
+                    }
+                    else
+                    {
+                        return (_encoding ?? Encoding.UTF8).GetBytes(chars, charCount, _bytes, buffer.Length - offset);
                     }
-
-                    return (int)(bytes - _bytes);
                 }
             }
             return 0;

src/libraries/System.Runtime.Serialization.Xml/tests/XmlDictionaryWriterTest.cs

@@ -494,6 +494,71 @@ void AssertBytesWritten(Action<XmlDictionaryWriter> action, XmlBinaryNodeType no
         }
     }
 
+    [Fact]
+    public static void XmlBaseWriter_WriteString()
+    {
+        const byte Chars8Text = 152;
+        const byte Chars16Text = 154;
+        MemoryStream ms = new MemoryStream();
+        XmlDictionaryWriter writer = (XmlDictionaryWriter)XmlDictionaryWriter.CreateBinaryWriter(ms);
+        writer.WriteStartElement("root");
+
+        int[] lengths = new[] { 7, 8, 9, 15, 16, 17, 31, 32, 36, 258 };
+        byte[] buffer = new byte[lengths.Max() + 1];
+
+        foreach (var length in lengths)
+        {
+            string allAscii = string.Create(length, null, (Span<char> chars, object _) =>
+            {
+                for (int i = 0; i < chars.Length; ++i)
+                    chars[i] = (char)(i % 128);
+            });
+            string multiByteLast = string.Create(length, null, (Span<char> chars, object _) =>
+            {
+                for (int i = 0; i < chars.Length; ++i)
+                    chars[i] = (char)(i % 128);
+                chars[^1] = '\u00E4'; // '�' - Latin Small Letter a with Diaeresis. Latin-1 Supplement.
+            });
+
+            int numBytes = Encoding.UTF8.GetBytes(allAscii, buffer);
+            Assert.True(numBytes == length, "Test setup wrong - allAscii");
+            ValidateWriteText(ms, writer, allAscii, expected: buffer.AsSpan(0, numBytes));
+
+            numBytes = Encoding.UTF8.GetBytes(multiByteLast, buffer);
+            Assert.True(numBytes == length + 1, "Test setup wrong - multiByte");
+            ValidateWriteText(ms, writer, multiByteLast, expected: buffer.AsSpan(0, numBytes));
+        }
+
+        static void ValidateWriteText(MemoryStream ms, XmlDictionaryWriter writer, string text, ReadOnlySpan<byte> expected)
+        {
+            writer.Flush();
+            ms.Seek(0, SeekOrigin.Begin);
+            ms.SetLength(0);
+            writer.WriteString(text);
+            writer.Flush();
+
+            ms.TryGetBuffer(out ArraySegment<byte> arraySegment);
+            ReadOnlySpan<byte> buffer = arraySegment;
+
+            if (expected.Length <= byte.MaxValue)
+            {
+                Assert.Equal(Chars8Text, buffer[0]);
+                Assert.Equal(expected.Length, buffer[1]);
+                buffer = buffer.Slice(2);
+            }
+            else if (expected.Length <= ushort.MaxValue)
+            {
+                Assert.Equal(Chars16Text, buffer[0]);
+                Assert.Equal(expected.Length, (int)(buffer[1]) | ((int)buffer[2] << 8));
+                buffer = buffer.Slice(3);
+            }
+            else
+                Assert.Fail("test use to long length");
+
+            AssertExtensions.SequenceEqual(expected, buffer);
+        }
+    }
+
     private static bool ReadTest(MemoryStream ms, Encoding encoding, ReaderWriterFactory.ReaderWriterType rwType, byte[] byteArray)
     {
         ms.Position = 0;