#ifdef GOAMD64_v2
#define ROL16(R, T) PSHUFB ·rol16<>(SB), R
#else
#define ROL16(R, T) ROL(16, R, T)
#endif

// ROL8 rotates the uint32s in register R left by 8, using temporary T if needed.
#ifdef GOAMD64_v2
#define ROL8(R, T) PSHUFB ·rol8<>(SB), R
#else
#define ROL8(R, T) ROL(8, R, T)
#endif

// QR is the ChaCha quarter-round on A, B, C, and D. T is an available temporary.
#define QR(A, B, C, D, T) \

DATA ·rol16<>+0x10(SB)/8, $0x0504070601000302
DATA ·rol16<>+0x18(SB)/8, $0x0D0C0F0E09080B0A
// <<< 8 with PSHUFB
DATA ·rol8<>+0x00(SB)/8, $0x0605040702010003
DATA ·rol8<>+0x08(SB)/8, $0x0E0D0C0F0A09080B
DATA ·rol8<>+0x10(SB)/8, $0x0605040702010003
DATA ·rol8<>+0x18(SB)/8, $0x0E0D0C0F0A09080B

DATA ·avx2InitMask<>+0x00(SB)/8, $0x0
DATA ·avx2InitMask<>+0x08(SB)/8, $0x0
DATA ·avx2InitMask<>+0x10(SB)/8, $0x1