According to the Vector API doc, the LSHR operator computes a>>>(n&(ESIZE*8-1)).
However, the current implementation is incorrect for negative bytes/shorts.

Reproduce:

$ java --add-modules jdk.incubator.vector ByteShift.java

Expected output:

Done

Actual output:

(...)
i: 128
a[i]: -128
scalar: -16
vector: 16