RORXL    $11, e, y1;                  \ // y1 = e >> 11				// S1B
	;                                     \
	ADDL     (disp + 0*4)(SP)(SRND*1), h; \ // h = k + w + h        // disp = k + w
	ORL      c, y3;                       \ // y3 = a|c				// MAJA
	VPALIGNR $4, XDWORD2, XDWORD3, XTMP0; \ // XTMP0 = W[-7]
	MOVL     f, y2;                       \ // y2 = f				// CH

	s *= x[i+3]
	// 386: `ANDL\t16\([A-Z]+\)\([A-Z]+\*4\), [A-Z]+`
	// amd64: `ANDL\t16\([A-Z]+[0-9]*\)\([A-Z]+[0-9]*\*4\), [A-Z]+[0-9]*`
	s &= x[i+4]
	// 386: `ORL\t20\([A-Z]+\)\([A-Z]+\*4\), [A-Z]+`
	// amd64: `ORL\t20\([A-Z]+[0-9]*\)\([A-Z]+[0-9]*\*4\), [A-Z]+[0-9]*`
	s |= x[i+5]
	// 386: `XORL\t24\([A-Z]+\)\([A-Z]+\*4\), [A-Z]+`
	// amd64: `XORL\t24\([A-Z]+[0-9]*\)\([A-Z]+[0-9]*\*4\), [A-Z]+[0-9]*`

		{name: "ANDLconst", argLength: 1, reg: gp11, asm: "ANDL", aux: "Int32", resultInArg0: true, clobberFlags: true}, // arg0 & auxint

		{name: "ORL", argLength: 2, reg: gp21, asm: "ORL", commutative: true, resultInArg0: true, clobberFlags: true}, // arg0 | arg1
		{name: "ORLconst", argLength: 1, reg: gp11, asm: "ORL", aux: "Int32", resultInArg0: true, clobberFlags: true}, // arg0 | auxint

	XORL	c, R9; \
	ANDL	b, R9; \
	XORL	d, R9

#define FUNC2(a, b, c, d, e) \
	MOVL	b, R9; \
	XORL	c, R9; \
	XORL	d, R9

#define FUNC3(a, b, c, d, e) \
	MOVL	b, R8; \
	ORL	c, R8; \
	ANDL	d, R8; \
	MOVL	b, R9; \
	ANDL	c, R9; \
	ORL	R8, R9

#define FUNC4 FUNC2

#define MIX(a, b, c, d, e, const) \
	ROLL	$30, b; \
	ADDL	R9, e; \
	MOVL	a, R8; \
	ROLL	$5, R8; \
	LEAL	const(e)(R10*1), e; \
	ADDL	R8, e

	// s390x:"FLOGR","MOVWZ"
	// ppc64x/power8:"ANDN","POPCNTW"
	// ppc64x/power9: "CNTTZW"
	// wasm:"I64Ctz"
	return bits.TrailingZeros32(n)
}

func TrailingZeros16(n uint16) int {
	// amd64:"BSFL","ORL\\t\\$65536"
	// 386:"BSFL\t"
	// arm:"ORR\t\\$65536","CLZ",-"MOVHU\tR"
	// arm64:"ORR\t\\$65536","RBITW","CLZW",-"MOVHU\tR",-"RBIT\t",-"CLZ\t"
	// s390x:"FLOGR","OR\t\\$65536"
	// ppc64x/power8:"POPCNTD","ORIS\\t\\$1"

	// 386:`MOVWLZX\s\([A-Z]+\)\([A-Z]+`,-`ORL`,-`MOVB`
	// amd64:`MOVWLZX\s\([A-Z]+\)\([A-Z]+`,-`MOVB`,-`OR`
	// ppc64le:`MOVHZ`,-`MOVBZ`
	// ppc64:`MOVHBR`,-`MOVBZ`
	return uint16(s[idx]) | uint16(s[idx+1])<<8
}

func load_le_byte2_uint16_idx_inv(s []byte, idx int) uint16 {
	// arm64:`MOVHU\s\(R[0-9]+\)\(R[0-9]+\)`,-`ORR`,-`MOVB`
	// 386:`MOVWLZX\s\([A-Z]+\)\([A-Z]+`,-`ORL`,-`MOVB`

	"MULSD",
	"MULSS",
	"MULW",
	"MULXL",
	"MULXQ",
	"MWAIT",
	"NEGB",
	"NEGL",
	"NEGQ",
	"NEGW",
	"NOPL",
	"NOPW",
	"NOTB",
	"NOTL",
	"NOTQ",
	"NOTW",
	"ORB",
	"ORL",
	"ORPD",
	"ORPS",
	"ORQ",
	"ORW",
	"OUTB",
	"OUTL",
	"OUTSB",
	"OUTSL",
	"OUTSW",
	"OUTW",
	"PABSB",
	"PABSD",
	"PABSW",
	"PACKSSLW",
	"PACKSSWB",
	"PACKUSDW",

GLOBL shifts<>(SB),RODATA,$256

TEXT ·checkASM(SB),NOSPLIT,$0-1
	// check that masks<>(SB) and shifts<>(SB) are aligned to 16-byte
	MOVL	$masks<>(SB), AX
	MOVL	$shifts<>(SB), BX
	ORL	BX, AX
	TESTL	$15, AX
	SETEQ	ret+0(FP)
	RET

TEXT runtime·return0(SB), NOSPLIT, $0
	MOVL	$0, AX
	RET

// Called from cgo wrappers, this function returns g->m->curg.stack.hi.