VST1	[V4.B16, V5.B16, V6.B16, V7.B16], (R11)
	B	ksDone
ks256:
	LDP	invSRows<>(SB), (R0, R1)
	VMOV	R0, V4.D[0]
	VMOV	R1, V4.D[1]
	LDPW	(R9), (R0, R1)
	LDPW	8(R9), (R2, R3)
	LDPW	16(R9), (R4, R5)
	LDPW	24(R9), (R6, R7)
	STPW.P	(R0, R1), 8(R10)
	STPW.P	(R2, R3), 8(R10)
	SUB	$7, R8
ks256Loop:
		STPW.P	(R4, R5), 8(R10)
		STPW.P	(R6, R7), 8(R10)
		VMOV	R7, V2.S[0]

//go:build !purego

#include "textflag.h"

TEXT	·block(SB),NOSPLIT,$0-32
	MOVD	dig+0(FP), R0
	MOVD	p+8(FP), R1
	MOVD	p_len+16(FP), R2
	AND	$~63, R2
	CBZ	R2, zero

	ADD	R1, R2, R21
	LDPW	(0*8)(R0), (R4, R5)
	LDPW	(1*8)(R0), (R6, R7)

loop:
	MOVW	R4, R12
	MOVW	R5, R13
	MOVW	R6, R14
	MOVW	R7, R15

	MOVW	(0*4)(R1), R8
	MOVW	R7, R9

#define ROUND1(a, b, c, d, index, const, shift) \

	LDP	8(R1), (ZR, R2)     // 3f8840a9
	LDPW	-5(R0), (R1, R2)    // 1b1400d1610b4029
	LDPW	(R0), (R1, R2)      // 01084029
	LDPW	4(R0), (R1, R2)     // 01884029
	LDPW	-4(R0), (R1, R2)    // 01887f29
	LDPW.W	4(R0), (R1, R2)     // 0188c029
	LDPW.P	4(R0), (R1, R2)     // 0188c028
	LDPW	11(R0), (R1, R2)    // 1b2c0091610b4029
	LDPW	1024(R0), (R1, R2)  // 1b001091610b4029
	LDPW	(RSP), (R1, R2)     // e10b4029

	"LDORALB",
	"LDORALD",
	"LDORALH",
	"LDORALW",
	"LDORAW",
	"LDORB",
	"LDORD",
	"LDORH",
	"LDORLB",
	"LDORLD",
	"LDORLH",
	"LDORLW",
	"LDORW",
	"LDP",
	"LDPSW",
	"LDPW",
	"LDXP",
	"LDXPW",
	"LDXR",
	"LDXRB",
	"LDXRH",
	"LDXRW",
	"LSL",
	"LSLW",
	"LSR",
	"LSRW",
	"MADD",
	"MADDW",
	"MNEG",
	"MNEGW",
	"MOVB",
	"MOVBU",
	"MOVD",