CALL	4(SP)
	CALL	(AX)
	CALL	(SP)
//	CALL	(AX*4) // TODO: This line is silently dropped on the floor!
	JCS	2(PC)
	JMP	(AX)(AX*4)

// LTYPEN spec4	{ outcode(int($1), &$2); }
	NOP
	NOP	AX
	NOP	foo+4(SB)

// LTYPES spec5	{ outcode(int($1), &$2); }
	SHLL	$4, BX
	SHLL	$4, foo+4(SB)
	SHLL	$4, foo+4(SB):AX // SHLL $4, AX, foo+4(SB)

// LTYPEM spec6	{ outcode(int($1), &$2); }
	MOVL	AX, BX
	MOVL	$4, BX

	VADDPD X30, X29, X28        // 6201950058e6
	VADDPD (AX), X29, X0        // 62f195005800
	VADDPD (AX), X1, X28        // 6261f5085820
	VADDPD (AX), X29, X28       // 626195005820
	VADDPD (AX), X29, X10       // 627195005810
	VADDPD (AX), X10, X28       // 6261ad085820
	VADDPD (CX)(AX*1), X29, X0  // 62f19500580401
	VADDPD (CX)(AX*1), X1, X28  // 6261f508582401
	VADDPD (CX)(AX*1), X29, X28 // 62619500582401

	RET
TEXT ·a23(SB), 0, $0-0
	CMPL runtime·writeBarrier(SB), $0
	LEAQ (AX)(CX*1), R15
	RET
TEXT ·a24(SB), 0, $0-0
	CMPL runtime·writeBarrier(SB), $0
	LEAQ (R15)(AX*1), AX // ERROR "when dynamic linking, R15 is clobbered by a global variable access and is used here"
	RET
TEXT ·a25(SB), 0, $0-0
	CMPL runtime·writeBarrier(SB), $0

JB Ldec128 MOVUPS (AX), X1 AESDEC X1, X0 MOVUPS 16(AX), X1 AESDEC X1, X0 ADDQ $0x20, AX Ldec192: MOVUPS (AX), X1 AESDEC X1, X0 MOVUPS 16(AX), X1 AESDEC X1, X0 ADDQ $0x20, AX Ldec128: MOVUPS (AX), X1 AESDEC X1, X0 MOVUPS 16(AX), X1 AESDEC X1, X0 MOVUPS 32(AX), X1 AESDEC X1, X0 MOVUPS 48(AX), X1 AESDEC X1, X0 MOVUPS 64(AX), X1 AESDEC X1, X0 MOVUPS 80(AX), X1 AESDEC X1, X0 MOVUPS 96(AX), X1 AESDEC X1, X0 MOVUPS 112(AX), X1 AESDEC X1, X0 MOVUPS 128(AX), X1 AESDEC X1, X0 MOVUPS 144(AX), X1 AESDECLAST X1,...

	MOVQ 2147483647+1(AX), AX       // ERROR "offset too large"
	MOVQ 3395469782(R10), R8        // ERROR "offset too large"
	LEAQ 3395469782(AX), AX         // ERROR "offset too large"
	ADDQ 3395469782(AX), AX         // ERROR "offset too large"
	ADDL 3395469782(AX), AX         // ERROR "offset too large"
	ADDW 3395469782(AX), AX         // ERROR "offset too large"
	LEAQ 433954697820(AX), AX       // ERROR "offset too large"

//	JMP	*(R12*4) // TODO: This line is silently dropped on the floor!
	JCS	2(PC)
	JMP	*(R12)(R13*4) // JMP (R12)(R13*4)
	JCS	2(PC)
	JMP	*(AX) // JMP (AX)
	JCS	2(PC)
	JMP	*(SP) // JMP (SP)
	JCS	2(PC)
//	JMP	*(AX*4) // TODO: This line is silently dropped on the floor!
	JCS	2(PC)
	JMP	*(AX)(AX*4) // JMP (AX)(AX*4)
	JCS	2(PC)
	JMP	4(SP)
	JCS	2(PC)
	JMP	(R12)
	JCS	2(PC)
//	JMP	(R12*4) // TODO: This line is silently dropped on the floor!

	MOVQ (AX), M0  // 0f6f00
	MOVQ M0, 8(SP) // 0f7f442408
	MOVQ 8(SP), M0 // 0f6f442408
	MOVQ M0, (AX)  // 0f7f00
	MOVQ M0, (BX)  // 0f7f03
	// On non-64bit arch, Go asm allowed uint32 offsets instead of int32.
	// These tests check that property for backwards-compatibility.
	MOVL 2147483648(AX), AX  // 8b8000000080
	MOVL -2147483648(AX), AX // 8b8000000080
	ADDL 2147483648(AX), AX  // 038000000080

	{"$~(1<<63)", "$9223372036854775807"},
	{"$~0x3F", "$-64"},
	{"$~15", "$-16"},
	{"(((8)&0xf)*4)(SP)", "32(SP)"},
	{"(((8-14)&0xf)*4)(SP)", "40(SP)"},
	{"(6+8)(AX)", "14(AX)"},
	{"(8*4)(BP)", "32(BP)"},
	{"(AX)", "(AX)"},
	{"(AX)(CX*8)", "(AX)(CX*8)"},
	{"(BP)(CX*4)", "(BP)(CX*4)"},
	{"(BP)(DX*4)", "(BP)(DX*4)"},
	{"(BP)(R8*4)", "(BP)(R8*4)"},
	{"(BX)", "(BX)"},
	{"(DI)", "(DI)"},

JB Ldec128 MOVUPS (AX), X1 AESDEC X1, X0 MOVUPS 16(AX), X1 AESDEC X1, X0 ADDQ $0x20, AX Ldec192: MOVUPS (AX), X1 AESDEC X1, X0 MOVUPS 16(AX), X1 AESDEC X1, X0 ADDQ $0x20, AX Ldec128: MOVUPS (AX), X1 AESDEC X1, X0 MOVUPS 16(AX), X1 AESDEC X1, X0 MOVUPS 32(AX), X1 AESDEC X1, X0 MOVUPS 48(AX), X1 AESDEC X1, X0 MOVUPS 64(AX), X1 AESDEC X1, X0 MOVUPS 80(AX), X1 AESDEC X1, X0 MOVUPS 96(AX), X1 AESDEC X1, X0 MOVUPS 112(AX), X1 AESDEC X1, X0 MOVUPS 128(AX), X1 AESDEC X1, X0 MOVUPS 144(AX), X1 AESDECLAST X1,...

	},
	{
		"LOAD macro",
		lines(
			"#define LOAD(off, reg) \\",
			"\tMOVBLZX	(off*4)(R12),	reg \\",
			"\tADDB	reg,		DX",
			"",
			"LOAD(8, AX)",
		),
		"\n.\n.MOVBLZX.(.8.*.4.).(.R12.).,.AX.\n.ADDB.AX.,.DX.\n",
	},
	{
		"nested multiline macro",
		lines(
			"#define KEYROUND(xmm, load, off, r1, r2, index) \\",
			"\tMOVBLZX	(BP)(DX*4),	R8 \\",
			"\tload((off+1), r2) \\",