VPOPCNTD -7(DI)(R8*8), K2, X8                      // 62327d0a5584c7f9ffffff
	VPOPCNTD X12, K2, X26                              // 62427d0a55d4
	VPOPCNTD X16, K2, X26                              // 62227d0a55d0
	VPOPCNTD X23, K2, X26                              // 62227d0a55d7
	VPOPCNTD (R14), K2, X26                            // 62427d0a5516
	VPOPCNTD -7(DI)(R8*8), K2, X26                     // 62227d0a5594c7f9ffffff

	RET

gcm_ghash_p8_4x:
	LVSL     (R8)(R0), T0      // 0x0001..0e0f
	MOVD     $0x70, R8
	LXVD2X   (HTBL)(R9), VH2
	MOVD     $0x80, R9
	VSPLTISB $8, T1            // 0x0808..0808
	MOVD     $0x90, R10
	LXVD2X   (HTBL)(R8), VH3L  // load H^3
	MOVD     $0xa0, R8
	LXVD2X   (HTBL)(R9), VH3
	MOVD     $0xb0, R9
	LXVD2X   (HTBL)(R10), VH3H
	MOVD     $0xc0, R10
	LXVD2X   (HTBL)(R8), VH4L  // load H^4
	MOVD     $0x10, R8

// performing [R4:R12:R11:R10:R9] = R4 + R3 * [R8:R7:R6:R5] + [R12:R11:R10:R9]
// where R4 is carried from the previous iteration, R8:R7:R6:R5 hold the next
// 4 words of x, R3 is y and R12:R11:R10:R9 are part of the result z.
loop:
	CBZ	R0, done

	LDP.P	16(R2), (R5, R6)
	LDP.P	16(R2), (R7, R8)

	LDP	(R1), (R9, R10)
	ADDS	R4, R9
	MUL	R6, R3, R14
	ADCS	R14, R10
	MUL	R7, R3, R15

// func Syscall(trap int64, a1, a2, a3 int64) (r1, r2, err int64);
TEXT ·Syscall(SB),NOSPLIT,$0-56
	JAL	runtime·entersyscall(SB)
	MOVV	a1+8(FP), R4
	MOVV	a2+16(FP), R5
	MOVV	a3+24(FP), R6
	MOVV	R0, R7
	MOVV	R0, R8
	MOVV	R0, R9
	MOVV	trap+0(FP), R2	// syscall entry
	SYSCALL
	BEQ	R7, ok
	MOVV	$-1, R1
	MOVV	R1, r1+32(FP)	// r1
	MOVV	R0, r2+40(FP)	// r2
	MOVV	R2, err+48(FP)	// errno
	JAL	runtime·exitsyscall(SB)
	RET

TEXT runtime·memequal(SB),NOSPLIT|NOFRAME,$0-25
	MOVV	a+0(FP), R1
	MOVV	b+8(FP), R2
	BEQ	R1, R2, eq
	MOVV	size+16(FP), R3
	ADDV	R1, R3, R4

	// chunk size is 16
	SGTU	$16, R3, R8
	BEQ	R0, R8, chunk_entry

byte_loop:
	BNE	R1, R4, byte_test
	MOVV	$1, R1
	MOVB	R1, ret+24(FP)
	RET
byte_test:
	MOVBU	(R1), R6
	ADDV	$1, R1
	MOVBU	(R2), R7
	ADDV	$1, R2
	BEQ	R6, R7, byte_loop

	MOVD	num+0(FP), R1	// syscall entry
	MOVD	a1+8(FP), R2
	MOVD	a2+16(FP), R3
	MOVD	a3+24(FP), R4
	MOVD	a4+32(FP), R5
	MOVD	a5+40(FP), R6
	MOVD	a6+48(FP), R7
	SYSCALL
	MOVD	$0xfffffffffffff001, R8
	CMPUBLT	R2, R8, ok
	MOVD	$-1, r1+56(FP)
	MOVD	$0, r2+64(FP)
	NEG	R2, R2
	MOVD	R2, errno+72(FP)
	RET
ok:
	MOVD	R2, r1+56(FP)
	MOVD	R3, r2+64(FP)
	MOVD	$0, errno+72(FP)

end:
	MOVW	R1, ret+4(FP)
	RET

TEXT	·Store(SB),NOSPLIT,$0-8
	MOVW	addr+0(FP), R1
	MOVW	v+4(FP), R2

	MOVB	runtime·goarm(SB), R8
	CMP	$7, R8
	BGE	native_barrier
	BL	memory_barrier<>(SB)
	B	store
native_barrier:
	DMB	MB_ISH

store:
	MOVW	R2, (R1)

	CMP	$7, R8
	BGE	native_barrier2
	BL	memory_barrier<>(SB)
	RET
native_barrier2:
	DMB	MB_ISH
	RET

TEXT	·Load8(SB),NOSPLIT,$0-5

	BIC	$0xf, R6, R10
	CBZ	R10, small        // length < 16
	ADD	R0, R10           // end of chunk16
	// length >= 16
chunk16_loop:
	LDP.P	16(R0), (R4, R8)
	LDP.P	16(R2), (R5, R9)
	CMP	R4, R5
	BNE	cmp
	CMP	R8, R9
	BNE	cmpnext
	CMP	R10, R0
	BNE	chunk16_loop
	AND	$0xf, R6, R6
	CBZ	R6, samebytes
	SUBS	$8, R6
	BLT	tail
	// the length of tail > 8 bytes
	MOVD.P	8(R0), R4
	MOVD.P	8(R2), R5
	CMP	R4, R5
	BNE	cmp

	MOVD	b_base+0(FP), R0
	MOVD	b_len+8(FP), R2
	MOVBU	c+24(FP), R1
	MOVD	$ret+32(FP), R8
	B	countbytebody<>(SB)

TEXT ·CountString(SB),NOSPLIT,$0-32
	MOVD	s_base+0(FP), R0
	MOVD	s_len+8(FP), R2
	MOVBU	c+16(FP), R1
	MOVD	$ret+24(FP), R8
	B	countbytebody<>(SB)

// input:
//   R0: data
//   R2: data len
//   R1: byte to find
//   R8: address to put result
TEXT countbytebody<>(SB),NOSPLIT,$0

	MOVD	b_base+0(FP), R0
	MOVD	b_len+8(FP), R2
	MOVBU	c+24(FP), R1
	MOVD	$ret+32(FP), R8
	B	indexbytebody<>(SB)

TEXT ·IndexByteString(SB),NOSPLIT,$0-32
	MOVD	s_base+0(FP), R0
	MOVD	s_len+8(FP), R2
	MOVBU	c+16(FP), R1
	MOVD	$ret+24(FP), R8
	B	indexbytebody<>(SB)

// input:
//   R0: data
//   R1: byte to search
//   R2: data len
//   R8: address to put result
TEXT indexbytebody<>(SB),NOSPLIT,$0
	// Core algorithm: