MOVD	R2, 8(R15)	// fn unsafe.Pointer
	MOVD	R3, 16(R15)	// a unsafe.Pointer
	// Skip R4 = n uint32
	MOVD	R5, 24(R15)	// ctxt uintptr
	BL	runtime·cgocallback(SB)

	FMOVD	32(R15), F8
	FMOVD	40(R15), F9
	FMOVD	48(R15), F10
	FMOVD	56(R15), F11
	FMOVD	64(R15), F12
	FMOVD	72(R15), F13
	FMOVD	80(R15), F14
	FMOVD	88(R15), F15

	// De-allocate stack frame.
	MOVD	$96(R15), R15

	// Restore R6-R15.

	MOVD	R1, 8(R15)                 // fn
	MOVD	$runtime·newosproc(SB), R1
	BL	R1

restore:
	// Restore F8-F15 from our stack frame.
	FMOVD	16(R15), F8
	FMOVD	24(R15), F9
	FMOVD	32(R15), F10
	FMOVD	40(R15), F11
	FMOVD	48(R15), F12
	FMOVD	56(R15), F13
	FMOVD	64(R15), F14
	FMOVD	72(R15), F15
	MOVD	$80(R15), R15

	// Restore R6-R15.
	LMG	48(R15), R6, R15
	RET

// _rt0_s390x_lib_go initializes the Go runtime.

	SHA256ROUND0(6, 0x923f82a4, R10, R11, R12, R13, R14, R15, R8, R9)
	SHA256ROUND0(7, 0xab1c5ed5, R9, R10, R11, R12, R13, R14, R15, R8)
	SHA256ROUND0(8, 0xd807aa98, R8, R9, R10, R11, R12, R13, R14, R15)
	SHA256ROUND0(9, 0x12835b01, R15, R8, R9, R10, R11, R12, R13, R14)
	SHA256ROUND0(10, 0x243185be, R14, R15, R8, R9, R10, R11, R12, R13)
	SHA256ROUND0(11, 0x550c7dc3, R13, R14, R15, R8, R9, R10, R11, R12)

	SHA512ROUND0(5, 0x59f111f1b605d019, R11, R12, R13, R14, R15, R8, R9, R10)
	SHA512ROUND0(6, 0x923f82a4af194f9b, R10, R11, R12, R13, R14, R15, R8, R9)
	SHA512ROUND0(7, 0xab1c5ed5da6d8118, R9, R10, R11, R12, R13, R14, R15, R8)
	SHA512ROUND0(8, 0xd807aa98a3030242, R8, R9, R10, R11, R12, R13, R14, R15)
	SHA512ROUND0(9, 0x12835b0145706fbe, R15, R8, R9, R10, R11, R12, R13, R14)

	XORQ R15, R15
	RET
TEXT ·a5(SB), 0, $0-0
	CMPL runtime·writeBarrier(SB), $0
	XORL R15, R15
	RET
TEXT ·a6(SB), 0, $0-0
	CMPL runtime·writeBarrier(SB), $0
	POPQ R15
	PUSHQ R15
	RET
TEXT ·a7(SB), 0, $0-0
	CMPL runtime·writeBarrier(SB), $0
	MOVQ R15, AX // ERROR "when dynamic linking, R15 is clobbered by a global variable access and is used here"
	RET
TEXT ·a8(SB), 0, $0-0
	CMPL runtime·writeBarrier(SB), $0

	std     %f9, 8(%r15)
	std     %f10, 16(%r15)
	std     %f11, 24(%r15)
	std     %f12, 32(%r15)
	std     %f13, 40(%r15)
	std     %f14, 48(%r15)
	std     %f15, 56(%r15)

	/* restore g pointer */
	lgr     %r13, %r3

	/* call fn */
	basr    %r14, %r2

	/* restore floating point registers */
	ld      %f8, 0(%r15)
	ld      %f9, 8(%r15)
	ld      %f10, 16(%r15)
	ld      %f11, 24(%r15)
	ld      %f12, 32(%r15)

				{1, 49151}, // AX CX DX BX SP BP SI DI R8 R9 R10 R11 R12 R13 R15
			},
		},
	},
	{
		name:   "CMPW",
		argLen: 2,
		asm:    x86.ACMPW,
		reg: regInfo{
			inputs: []inputInfo{
				{0, 49151}, // AX CX DX BX SP BP SI DI R8 R9 R10 R11 R12 R13 R15
				{1, 49151}, // AX CX DX BX SP BP SI DI R8 R9 R10 R11 R12 R13 R15
			},
		},
	},
	{

	// Case for 8 ≤ n < 16 bytes
	MOVD	(R4)(R8), R14   // R14 = a[i,...,i+7]
	MOVD	(R5)(R8), R15   // R15 = b[i,...,i+7]
	XOR	R14, R15, R16   // R16 = a[] ^ b[]
	SUB	$8, R6          // n = n - 8
	MOVD	R16, (R3)(R8)   // Store to dst
	ADD	$8, R8
xor4:
	CMP	R6, $4
	BLT	xor2
	MOVWZ	(R4)(R8), R14
	MOVWZ	(R5)(R8), R15
	XOR	R14, R15, R16
	MOVW	R16, (R3)(R8)
	ADD	$4,R8
	ADD	$-4,R6
xor2:
	CMP	R6, $2
	BLT	xor1

	// r4 = a0×b4
	MOVQ (CX), AX
	MULQ 32(BX)
	MOVQ AX, R15
	MOVQ DX, R14

	// r4 += a1×b3
	MOVQ 8(CX), AX
	MULQ 24(BX)
	ADDQ AX, R15
	ADCQ DX, R14

	// r4 += a2×b2
	MOVQ 16(CX), AX
	MULQ 16(BX)
	ADDQ AX, R15
	ADCQ DX, R14

	// r4 += a3×b1
	MOVQ 24(CX), AX
	MULQ 8(BX)
	ADDQ AX, R15
	ADCQ DX, R14

	// r4 += a4×b0
	MOVQ 32(CX), AX
	MULQ (BX)

	MOVQ msg_base+8(FP), SI
	MOVQ msg_len+16(FP), R15

	MOVQ 0(DI), R8   // h0
	MOVQ 8(DI), R9   // h1
	MOVQ 16(DI), R10 // h2
	MOVQ 24(DI), R11 // r0
	MOVQ 32(DI), R12 // r1

	CMPQ R15, $16
	JB   bytes_between_0_and_15

loop:
	POLY1305_ADD(SI, R8, R9, R10)

multiply:
	POLY1305_MUL(R8, R9, R10, R11, R12, BX, CX, R13, R14)
	SUBQ $16, R15
	CMPQ R15, $16
	JAE  loop

bytes_between_0_and_15: