String(byte[], int, int, Charset) constructor has this check for latin-1 in the latin-1 fast-path:

if ((b1 == (byte)0xc2 || b1 == (byte)0xc3) && ...

Since the two constant bytes differ only on the lowest bit this can be transformed to this:

if ((b1 & 0xfe) == 0xc2 &&

Which makes the code less branchy and produce a small speed-up on a targetted microbenchmark:

Benchmark (charsetName) Mode Cnt Score Error Units
StringDecode.decodeLatin1LongStart UTF-8 avgt 50 2283.591 ± 12.332 ns/op

StringDecode.decodeLatin1LongStart UTF-8 avgt 50 2165.984 ± 13.136 ns/op

(While this minor inefficiency appears to have been introduced in with JEP 254 in JDK 9, the performance of decoding latin-1 strings was much improved thanks to the compactness of latin-1 encoded Strings, so I've not seen a regression caused by this.)