rotr32
(T2) = rotr32(T2, 16); \
(TMP2) = rotr32((TMP), 8); \
(Y) = (TMP2) ^ rotr32((TMP) ^ (TMP2), 16); \
d = rotr32(d ^ a, 16); \
b = rotr32(b ^ c, 12); \
d = rotr32(d ^ a, 8); \
b = rotr32(b ^ c, 7); \