assertEqualCharNodes(traverser.depthFirstPostOrder(charactersOf("bd")), "adcb");
    assertEqualCharNodes(traverser.depthFirstPostOrder(charactersOf("dc")), "cbad");
    assertEqualCharNodes(traverser.depthFirstPostOrder(charactersOf("bc")), "adcb");
  }

  @Test
  public void forGraph_depthFirstPostOrder_twoCycles() {

    assertEqualCharNodes(traverser.depthFirstPostOrder(charactersOf("bd")), "adcb");
    assertEqualCharNodes(traverser.depthFirstPostOrder(charactersOf("dc")), "cbad");
    assertEqualCharNodes(traverser.depthFirstPostOrder(charactersOf("bc")), "adcb");
  }

  @Test
  public void forGraph_depthFirstPostOrder_twoCycles() {

	ASUBMEVCC
	ASUBMEV
	ASUBV
	ASUBE
	ASUBECC
	ASUBEV
	ASUBEVCC
	ASUBZE
	ASUBZECC
	ASUBZEVCC
	ASUBZEV
	ASYNC
	AXOR
	AXORCC
	AXORIS

	ADCBF
	ADCBI
	ADCBST
	ADCBT
	ADCBTST
	ADCBZ
	AEIEIO
	AICBI
	AISYNC
	APTESYNC
	ATLBIE
	ATLBIEL
	ATLBSYNC
	ATW

	ASYSCALL
	AWORD

	ARFCI

	AFCPSGN
	AFCPSGNCC

	ADCS	acc2, acc2
	ADCS	acc3, acc3
	ADCS	acc4, acc4
	ADCS	acc5, acc5
	ADCS	acc6, acc6
	ADC	$0, acc7
	// Missing products
	MUL	x0, x0, acc0
	UMULH	x0, x0, t0
	ADDS	t0, acc1, acc1

	MUL	x1, x1, t0
	ADCS	t0, acc2, acc2
	UMULH	x1, x1, t1
	ADCS	t1, acc3, acc3

	MUL	x2, x2, t0
	ADCS	t0, acc4, acc4
	UMULH	x2, x2, t1
	ADCS	t1, acc5, acc5

	MUL	x3, x3, t0
	ADCS	t0, acc6, acc6

	ADDQ t1, acc3
	ADCQ $0, DX
	ADDQ AX, acc3
	ADCQ $0, DX
	MOVQ DX, t1

	MOVQ (8*3)(x_ptr), AX
	MULQ t0
	ADDQ t1, acc4
	ADCQ $0, DX
	ADDQ AX, acc4
	ADCQ DX, acc5
	ADCQ $0, acc0
	// Second reduction step
	MOVQ acc1, AX
	MOVQ acc1, t1
	SHLQ $32, acc1
	MULQ p256const1<>(SB)
	SHRQ $32, t1
	ADDQ acc1, acc2
	ADCQ t1, acc3
	ADCQ AX, acc4
	ADCQ DX, acc5
	ADCQ $0, acc0

U1:	// n >= 0
	// regular loop body unrolled 4x
	ADDQ CX, CX		// restore CF
	MOVQ 0(R8)(SI*8), R11
	MOVQ 8(R8)(SI*8), R12
	MOVQ 16(R8)(SI*8), R13
	MOVQ 24(R8)(SI*8), R14
	ADCQ 0(R9)(SI*8), R11
	ADCQ 8(R9)(SI*8), R12
	ADCQ 16(R9)(SI*8), R13
	ADCQ 24(R9)(SI*8), R14
	MOVQ R11, 0(R10)(SI*8)
	MOVQ R12, 8(R10)(SI*8)
	MOVQ R13, 16(R10)(SI*8)
	MOVQ R14, 24(R10)(SI*8)
	SBBQ CX, CX		// save CF

	ADDQ $4, SI		// i += 4

	ANDQ  $0xFFFFFFFFFFFFFFFC, t0; \
	ADDQ  t0, h0;                  \
	ADCQ  t3, h1;                  \
	ADCQ  $0, h2;                  \
	SHRQ  $2, t3, t2;              \
	SHRQ  $2, t3;                  \
	ADDQ  t2, h0;                  \
	ADCQ  t3, h1;                  \
	ADCQ  $0, h2

// func update(state *[7]uint64, msg []byte)
TEXT ·update(SB), $0-32
	MOVQ state+0(FP), DI
	MOVQ msg_base+8(FP), SI
	MOVQ msg_len+16(FP), R15

	ADDS	R4, R9
	MUL	R6, R3, R14
	ADCS	R14, R10
	MUL	R7, R3, R15
	LDP	16(R1), (R11, R12)
	ADCS	R15, R11
	MUL	R8, R3, R16
	ADCS	R16, R12
	UMULH	R8, R3, R20
	ADC	$0, R20

	MUL	R5, R3, R13
	ADDS	R13, R9
	UMULH	R5, R3, R17
	ADCS	R17, R10
	UMULH	R6, R3, R21
	STP.P	(R9, R10), 16(R1)
	ADCS	R21, R11
	UMULH	R7, R3, R19
	ADCS	R19, R12
	STP.P	(R11, R12), 16(R1)
	ADC	$0, R20, R4

	testenv.ParallelOn64Bit(t)
	h := newHashSet()
	var b [3]byte
	for i := 0; i < 256; i++ {
		b[0] = byte(i)
		h.addB(b[:1])
		for j := 0; j < 256; j++ {
			b[1] = byte(j)
			h.addB(b[:2])
			if !testing.Short() {
				for k := 0; k < 256; k++ {
					b[2] = byte(k)
					h.addB(b[:3])
				}
			}
		}
	}
	h.check(t)
}

// Different length strings of all zeros have distinct hashes.

	JMP E1

L1:	MOVL (SI)(BX*4), AX
	ADDL DX, DX		// restore CF
	ADCL (CX)(BX*4), AX
	SBBL DX, DX		// save CF
	MOVL AX, (DI)(BX*4)
	ADDL $1, BX		// i++

E1:	CMPL BX, BP		// i < n
	JL L1

	NEGL DX
	MOVL DX, c+36(FP)
	RET


// func subVV(z, x, y []Word) (c Word)
// (same as addVV except for SBBL instead of ADCL and label names)
TEXT ·subVV(SB),NOSPLIT,$0
	MOVL z+0(FP), DI
	MOVL x+12(FP), SI