FULL PRODUCT VERSION :
java version "1.6.0_03"
Java(TM) SE Runtime Environment (build 1.6.0_03-b05)
Java HotSpot(TM) Client VM (build 1.6.0_03-b05, mixed mode)


ADDITIONAL OS VERSION INFORMATION :
Microsoft Windows XP [version 5.1.2600]
(and all others OSes...)

A DESCRIPTION OF THE PROBLEM :
nextBytes(byte[]) in java.util.Random could be faster, by avoiding divisions. The proposed implementation generates exactly the same sequence, but it generates 4 bytes per loop without using any division, with a minor unrolling of the generation loop which also simplifies the loop condition.


REPRODUCIBILITY :
This bug can be reproduced always.

CUSTOMER SUBMITTED WORKAROUND :
Change java.util.Random implemenation to use:

    public void nextBytes(final byte[] bytes) {
        int i = 0, n;
        final int length;
        n = (length = bytes.length) >> 2;
        while (--n >= 0) {
            final int rnd;
            bytes[i] = (byte)(rnd = nextInt());
            bytes[i + 1] = (byte)(rnd >> 8);
            bytes[i + 2] = (byte)(rnd >> 16);
            bytes[i + 3] = (byte)(rnd >> 24);
            i += 4;
        }
        if ((n = length & 3) > 0) {
            final int rnd;
            bytes[i] = (byte)(rnd = nextInt());
            if (n > 1) {
                bytes[i + 1] = (byte)(rnd >> 8);
                if (n > 2)
                    bytes[i + 2] = (byte)(rnd >> 16);
            }
        }
    }