MOVL	((index-2)*4)(BP), AX; \
	MOVL	AX, CX; \
	RORL	$17, AX; \
	MOVL	CX, DX; \
	RORL	$19, CX; \
	SHRL	$10, DX; \
	MOVL	((index-15)*4)(BP), BX; \
	XORL	CX, AX; \
	MOVL	BX, CX; \
	XORL	DX, AX; \
	RORL	$7, BX; \
	MOVL	CX, DX; \
	SHRL	$3, DX; \
	RORL	$18, CX; \
	ADDL	((index-7)*4)(BP), AX; \
	XORL	CX, BX; \
	XORL	DX, BX; \
	ADDL	((index-16)*4)(BP), BX; \
	ADDL	BX, AX; \
	MOVL	AX, ((index)*4)(BP)

	//	func f(x uint64) (uint64, uint64) { return x/1000000000, x%1000000000 }
	// adapted to reduce duplication
	MOVQ	AX, CX
	MOVQ	$1360296554856532783, AX
	MULQ	CX
	ADDQ	CX, DX
	RCRQ	$1, DX
	SHRQ	$29, DX
	MOVQ	DX, sec+0(FP)
	IMULQ	$1000000000, DX
	SUBQ	DX, CX
	MOVL	CX, nsec+8(FP)

	SBBL $(delta >> 32), DX

	// nano/100 = DX:AX
	// split into two decimal halves by div 1e9.
	// (decimal point is two spots over from correct place,
	// but we avoid overflow in the high word.)
	MOVL	$1000000000, CX
	DIVL	CX
	MOVL	AX, DI
	MOVL	DX, SI

	// DI = nano/100/1e9 = nano/1e11 = sec/100, DX = SI = nano/100%1e9
	// split DX into seconds and nanoseconds by div 1e7 magic multiply.
	MOVL	DX, AX
	MOVL	$1801439851, CX

	MOVQ	((index-2)*8)(BP), AX; \
	MOVQ	AX, CX; \
	RORQ	$19, AX; \
	MOVQ	CX, DX; \
	RORQ	$61, CX; \
	SHRQ	$6, DX; \
	MOVQ	((index-15)*8)(BP), BX; \
	XORQ	CX, AX; \
	MOVQ	BX, CX; \
	XORQ	DX, AX; \
	RORQ	$1, BX; \
	MOVQ	CX, DX; \
	SHRQ	$7, DX; \
	RORQ	$8, CX; \
	ADDQ	((index-7)*8)(BP), AX; \
	XORQ	CX, BX; \
	XORQ	DX, BX; \
	ADDQ	((index-16)*8)(BP), BX; \
	ADDQ	BX, AX; \
	MOVQ	AX, ((index)*8)(BP)

	// so this function is reentrant.
	MOVQ	m_vdsoPC(BX), CX
	MOVQ	m_vdsoSP(BX), DX
	MOVQ	CX, 0(SP)
	MOVQ	DX, 8(SP)

	LEAQ	ret+0(FP), DX
	MOVQ	-8(DX), CX
	MOVQ	CX, m_vdsoPC(BX)
	MOVQ	DX, m_vdsoSP(BX)

	CMPQ	R14, m_curg(BX)	// Only switch if on curg.
	JNE	noswitch

	MOVQ	m_g0(BX), DX
	MOVQ	(g_sched+gobuf_sp)(DX), SP	// Set SP to g0 stack

noswitch:
	SUBQ	$16, SP		// Space for results

	MOVQ	0(BX), DI		// arg 1 addr
	MOVQ	8(BX), SI		// arg 2 len
	MOVL	16(BX), DX		// arg 3 prot
	MOVL	20(BX), CX		// arg 4 flags
	MOVL	24(BX), R8		// arg 5 fid
	MOVL	28(BX), R9		// arg 6 offset
	CALL	libc_mmap(SB)
	XORL	DX, DX
	CMPQ	AX, $-1
	JNE	ok
	CALL	libc_error(SB)
	MOVLQSX	(AX), DX		// errno
	XORL	AX, AX
ok:
	MOVQ	AX, 32(BX)
	MOVQ	DX, 40(BX)
	RET

TEXT runtime·munmap_trampoline(SB),NOSPLIT,$0

	MOVQ	new+8(FP), SI
	MOVQ	old+16(FP), DX
	MOVL	$83, AX
	SYSCALL
	RET

// func walltime() (sec int64, nsec int32)
TEXT runtime·walltime(SB), NOSPLIT, $32
	MOVL	$232, AX // clock_gettime
	MOVQ	$0, DI  	// CLOCK_REALTIME
	LEAQ	8(SP), SI
	SYSCALL
	MOVQ	8(SP), AX	// sec
	MOVQ	16(SP), DX	// nsec

	// sec is in AX, nsec in DX
	MOVQ	AX, sec+0(FP)
	MOVL	DX, nsec+8(FP)
	RET

	ADDL	$const_stackGuard, AX
	MOVL	AX, g_stackguard0(DX)
	MOVL	AX, g_stackguard1(DX)

	// Set up tls.
	LEAL	m_tls(CX), DI
	MOVL	CX, g_m(DX)
	MOVL	DX, g(DI)
	MOVL	DI, 4(SP)
	CALL	runtime·setldt(SB) // clobbers CX and DX

	// Someday the convention will be D is always cleared.
	CLD

	CALL	runtime·stackcheck(SB)	// clobbers AX,CX
	CALL	runtime·mstart(SB)

	RET

// ---------------------------
// num | AX          | AX
// a1  | BX          | DI
// a2  | CX          | SI
// a3  | DI          | DX
// a4  | SI          | R10
// a5  | R8          | R8
// a6  | R9          | R9
//
// r1  | AX          | AX
// r2  | BX          | DX
// err | CX          | part of AX
//
// Note that this differs from "standard" ABI convention, which would pass 4th
// arg in CX, not R10.

	MOVL	DX, nsec+8(FP)
	RET

TEXT runtime·nanotime1(SB),NOSPLIT,$32
	MOVQ	$CLOCK_MONOTONIC, DI	// arg 1 - clock_id
	LEAQ	8(SP), SI		// arg 2 - tp
	MOVL	$SYS___clock_gettime50, AX
	SYSCALL
	MOVQ	8(SP), AX		// sec
	MOVQ	16(SP), DX		// nsec

	// sec is in AX, nsec in DX
	// return nsec in AX
	IMULQ	$1000000000, AX
	ADDQ	DX, AX
	MOVQ	AX, ret+0(FP)
	RET

TEXT runtime·getcontext(SB),NOSPLIT,$-8