Optimize varint frame decoding again

Use ByteProcessor in a controlled matter in one specific case. Performance measurements with my Ryzen 5 3600 indicate a 25-35% improvement in time spent framing incoming packets.

proxy/src/main/java/com/velocitypowered/proxy/protocol/netty/MinecraftVarintFrameDecoder.java

@@ -1,41 +1,63 @@
 package com.velocitypowered.proxy.protocol.netty;
 
-import com.velocitypowered.proxy.protocol.ProtocolUtils;
+import com.velocitypowered.proxy.util.except.QuietException;
 import io.netty.buffer.ByteBuf;
 import io.netty.channel.ChannelHandlerContext;
 import io.netty.handler.codec.ByteToMessageDecoder;
-import io.netty.handler.codec.CorruptedFrameException;
+import io.netty.util.ByteProcessor;
 import java.util.List;
 
 public class MinecraftVarintFrameDecoder extends ByteToMessageDecoder {
 
+  private static final QuietException BAD_LENGTH_CACHED = new QuietException("Bad packet length");
+  private final VarintByteDecoder reader = new VarintByteDecoder();
+
   @Override
-  protected void decode(ChannelHandlerContext ctx, ByteBuf in, List<Object> out) throws Exception {
-    read_lens: while (in.isReadable()) {
-      int origReaderIndex = in.readerIndex();
-      for (int i = 0; i < 3; i++) {
-        if (!in.isReadable()) {
-          in.readerIndex(origReaderIndex);
-          return;
-        }
-
-        byte read = in.readByte();
-        if (read >= 0) {
-          // Make sure reader index of length buffer is returned to the beginning
-          in.readerIndex(origReaderIndex);
-          int packetLength = ProtocolUtils.readVarInt(in);
-
-          if (in.readableBytes() >= packetLength) {
-            out.add(in.readBytes(packetLength));
-            continue read_lens;
-          } else {
-            in.readerIndex(origReaderIndex);
-            return;
-          }
-        }
+  protected void decode(ChannelHandlerContext ctx, ByteBuf in, List<Object> out) {
+    while (in.isReadable()) {
+      int varintEnd = in.forEachByte(reader);
+      if (varintEnd == -1) {
+        return;
       }
 
-      throw new CorruptedFrameException("VarInt too big");
+      if (!reader.successfulDecode) {
+        throw BAD_LENGTH_CACHED;
+      }
+
+      int minimumRead = reader.bytesRead + reader.readVarint;
+      if (in.isReadable(minimumRead)) {
+        out.add(in.retainedSlice(varintEnd + 1, reader.readVarint));
+        in.skipBytes(minimumRead);
+        reader.reset();
+      } else {
+        reader.reset();
+        return;
+      }
+    }
+  }
+
+  private static class VarintByteDecoder implements ByteProcessor {
+    private int readVarint;
+    private int bytesRead;
+    private boolean successfulDecode;
+
+    @Override
+    public boolean process(byte k) {
+      readVarint |= (k & 0x7F) << bytesRead++ * 7;
+      if (bytesRead > 3) {
+        return false;
+      }
+      if ((k & 0x80) != 128) {
+        successfulDecode = true;
+        return false;
+      }
+      return true;
+    }
+
+    void reset() {
+      readVarint = 0;
+      bytesRead = 0;
+      successfulDecode = false;
     }
   }
 }