MOVW R3, 12(R29)
	MOVW R4, 16(R29)
	MOVW R5, 20(R29)
	MOVW R6, 24(R29)
	MOVW R7, 28(R29)
	MOVW R8, 32(R29)
	MOVW R9, 36(R29)
	MOVW R10, 40(R29)
	MOVW R11, 44(R29)
	MOVW R12, 48(R29)
	MOVW R13, 52(R29)
	MOVW R14, 56(R29)
	MOVW R15, 60(R29)
	MOVW R16, 64(R29)
	MOVW R17, 68(R29)
	MOVW R18, 72(R29)
	MOVW R19, 76(R29)
	MOVW R20, 80(R29)
	MOVW R21, 84(R29)
	MOVW R22, 88(R29)
	MOVW R24, 92(R29)

	print("r8      ", hex(c.r8()), "\n")
	print("r9      ", hex(c.r9()), "\n")
	print("r10     ", hex(c.r10()), "\n")
	print("r11     ", hex(c.r11()), "\n")
	print("r12     ", hex(c.r12()), "\n")
	print("r13     ", hex(c.r13()), "\n")
	print("r14     ", hex(c.r14()), "\n")
	print("r15     ", hex(c.r15()), "\n")
	print("r16     ", hex(c.r16()), "\n")
	print("r17     ", hex(c.r17()), "\n")
	print("r18     ", hex(c.r18()), "\n")

	MOVL $495, AX		// lwp_create
	SYSCALL
	MOVL	AX, ret+8(FP)
	RET

TEXT runtime·lwp_start(SB),NOSPLIT,$0
	MOVQ	DI, R13 // m

	// set up FS to point at m->tls
	LEAQ	m_tls(R13), DI
	CALL	runtime·settls(SB)	// smashes DI

	// set up m, g
	get_tls(CX)
	MOVQ	m_g0(R13), DI
	MOVQ	R13, g_m(DI)
	MOVQ	DI, g(CX)

	CALL	runtime·stackcheck(SB)
	CALL	runtime·mstart(SB)

	MOVQ 0, AX			// crash (not reached)

	return ptracePtr(PTRACE_SETREGS, pid, 0, unsafe.Pointer(regs))
}

// PtraceRegsAmd64 is the registers used by amd64 binaries.
type PtraceRegsAmd64 struct {
	R15      uint64
	R14      uint64
	R13      uint64
	R12      uint64
	Rbp      uint64
	Rbx      uint64
	R11      uint64
	R10      uint64
	R9       uint64
	R8       uint64
	Rax      uint64
	Rcx      uint64
	Rdx      uint64
	Rsi      uint64

	print("r8   ", hex(c.r8()), "\t")
	print("r9   ", hex(c.r9()), "\n")
	print("r10  ", hex(c.r10()), "\t")
	print("r11  ", hex(c.r11()), "\n")
	print("r12  ", hex(c.r12()), "\t")
	print("r13  ", hex(c.r13()), "\n")
	print("r14  ", hex(c.r14()), "\t")
	print("r15  ", hex(c.r15()), "\n")
	print("r16  ", hex(c.r16()), "\t")
	print("r17  ", hex(c.r17()), "\n")
	print("r18  ", hex(c.r18()), "\t")

	p.R8 = (<- chan <- chan chan int)(nil)
	p.R9 = (<- chan chan <- chan int)(nil)
	p.R10 = (chan <- <- chan chan int)(nil)
	p.R11 = (chan <- chan <- chan int)(nil)
	p.R12 = (chan chan <- <- chan int)(nil)
	p.R13 = (chan chan <- chan <- int)(nil)

	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) \
	ADDW	$const, a; \
	ADDW	R8, a; \
	MOVW	(index*4)(R1), R8; \
	EORW	c, R9; \

	MOVQ    $0, R8
	// Copy mp, gp, fn off parent stack for use by child.
	// Careful: Linux system call clobbers CX and R11.
	MOVQ	mp+16(FP), R13
	MOVQ	gp+24(FP), R9
	MOVQ	fn+32(FP), R12
	CMPQ	R13, $0    // m
	JEQ	nog1
	CMPQ	R9, $0    // g
	JEQ	nog1
	LEAQ	m_tls(R13), R8
#ifdef GOOS_android
	// Android stores the TLS offset in runtime·tls_g.
	SUBQ	runtime·tls_g(SB), R8
#else

	POPCNTL DX, DX
	MOVQ	DX, (R8)
	RET

avx2:
#ifndef hasAVX2
	CMPB   internal∕cpu·X86+const_offsetX86HasAVX2(SB), $1
	JNE sse
#endif
	MOVD AX, X0
	LEAQ -64(SI)(BX*1), R11
	LEAQ (SI)(BX*1), R13
	VPBROADCASTB  X0, Y1
	PCALIGN $32
avx2_loop:
	VMOVDQU (DI), Y2
	VMOVDQU 32(DI), Y4
	VPCMPEQB Y1, Y2, Y3
	VPCMPEQB Y1, Y4, Y5
	VPMOVMSKB Y3, DX
	VPMOVMSKB Y5, CX
	POPCNTL DX, DX

	// R4 = a_base
	// R5 = a_len
	// R6 = b_base
	// R7 = b_len
	JMP	cmpbody<>(SB)

// On entry:
// R5 length of a
// R7 length of b
// R4 points to the start of a
// R6 points to the start of b
// R13 points to the return value (-1/0/1)
TEXT cmpbody<>(SB),NOSPLIT|NOFRAME,$0
	BEQ	R4, R6, samebytes // same start of a and b

	SGTU	R5, R7, R9
	BNE	R0, R9, r2_lt_r1
	MOVV	R5, R14
	JMP	entry
r2_lt_r1: