Home | History | Annotate | Download | only in gen
      1 // Copyright 2015 The Go Authors. All rights reserved.
      2 // Use of this source code is governed by a BSD-style
      3 // license that can be found in the LICENSE file.
      4 
      5 // +build ignore
      6 
      7 package main
      8 
      9 import "strings"
     10 
     11 // Notes:
     12 //  - Integer types live in the low portion of registers. Upper portions are junk.
     13 //  - Boolean types use the low-order byte of a register. 0=false, 1=true.
     14 //    Upper bytes are junk.
     15 //  - Floating-point types live in the low natural slot of an sse2 register.
     16 //    Unused portions are junk.
     17 //  - We do not use AH,BH,CH,DH registers.
     18 //  - When doing sub-register operations, we try to write the whole
     19 //    destination register to avoid a partial-register write.
     20 //  - Unused portions of AuxInt (or the Val portion of ValAndOff) are
     21 //    filled by sign-extending the used portion.  Users of AuxInt which interpret
     22 //    AuxInt as unsigned (e.g. shifts) must be careful.
     23 
     24 // Suffixes encode the bit width of various instructions.
     25 // Q (quad word) = 64 bit
     26 // L (long word) = 32 bit
     27 // W (word)      = 16 bit
     28 // B (byte)      = 8 bit
     29 
     30 // copied from ../../amd64/reg.go
     31 var regNamesAMD64 = []string{
     32 	"AX",
     33 	"CX",
     34 	"DX",
     35 	"BX",
     36 	"SP",
     37 	"BP",
     38 	"SI",
     39 	"DI",
     40 	"R8",
     41 	"R9",
     42 	"R10",
     43 	"R11",
     44 	"R12",
     45 	"R13",
     46 	"R14",
     47 	"R15",
     48 	"X0",
     49 	"X1",
     50 	"X2",
     51 	"X3",
     52 	"X4",
     53 	"X5",
     54 	"X6",
     55 	"X7",
     56 	"X8",
     57 	"X9",
     58 	"X10",
     59 	"X11",
     60 	"X12",
     61 	"X13",
     62 	"X14",
     63 	"X15",
     64 
     65 	// pseudo-registers
     66 	"SB",
     67 }
     68 
     69 func init() {
     70 	// Make map from reg names to reg integers.
     71 	if len(regNamesAMD64) > 64 {
     72 		panic("too many registers")
     73 	}
     74 	num := map[string]int{}
     75 	for i, name := range regNamesAMD64 {
     76 		num[name] = i
     77 	}
     78 	buildReg := func(s string) regMask {
     79 		m := regMask(0)
     80 		for _, r := range strings.Split(s, " ") {
     81 			if n, ok := num[r]; ok {
     82 				m |= regMask(1) << uint(n)
     83 				continue
     84 			}
     85 			panic("register " + r + " not found")
     86 		}
     87 		return m
     88 	}
     89 
     90 	// Common individual register masks
     91 	var (
     92 		ax         = buildReg("AX")
     93 		cx         = buildReg("CX")
     94 		dx         = buildReg("DX")
     95 		gp         = buildReg("AX CX DX BX BP SI DI R8 R9 R10 R11 R12 R13 R14 R15")
     96 		fp         = buildReg("X0 X1 X2 X3 X4 X5 X6 X7 X8 X9 X10 X11 X12 X13 X14 X15")
     97 		gpsp       = gp | buildReg("SP")
     98 		gpspsb     = gpsp | buildReg("SB")
     99 		callerSave = gp | fp
    100 	)
    101 	// Common slices of register masks
    102 	var (
    103 		gponly = []regMask{gp}
    104 		fponly = []regMask{fp}
    105 	)
    106 
    107 	// Common regInfo
    108 	var (
    109 		gp01      = regInfo{inputs: nil, outputs: gponly}
    110 		gp11      = regInfo{inputs: []regMask{gp}, outputs: gponly}
    111 		gp11sp    = regInfo{inputs: []regMask{gpsp}, outputs: gponly}
    112 		gp11sb    = regInfo{inputs: []regMask{gpspsb}, outputs: gponly}
    113 		gp21      = regInfo{inputs: []regMask{gp, gp}, outputs: gponly}
    114 		gp21sp    = regInfo{inputs: []regMask{gpsp, gp}, outputs: gponly}
    115 		gp21sb    = regInfo{inputs: []regMask{gpspsb, gpsp}, outputs: gponly}
    116 		gp21shift = regInfo{inputs: []regMask{gp, cx}, outputs: []regMask{gp}}
    117 		gp11div   = regInfo{inputs: []regMask{ax, gpsp &^ dx}, outputs: []regMask{ax, dx}}
    118 		gp21hmul  = regInfo{inputs: []regMask{ax, gpsp}, outputs: []regMask{dx}, clobbers: ax}
    119 
    120 		gp2flags = regInfo{inputs: []regMask{gpsp, gpsp}}
    121 		gp1flags = regInfo{inputs: []regMask{gpsp}}
    122 		flagsgp  = regInfo{inputs: nil, outputs: gponly}
    123 
    124 		gp11flags = regInfo{inputs: []regMask{gp}, outputs: []regMask{gp, 0}}
    125 
    126 		readflags = regInfo{inputs: nil, outputs: gponly}
    127 		flagsgpax = regInfo{inputs: nil, clobbers: ax, outputs: []regMask{gp &^ ax}}
    128 
    129 		gpload    = regInfo{inputs: []regMask{gpspsb, 0}, outputs: gponly}
    130 		gploadidx = regInfo{inputs: []regMask{gpspsb, gpsp, 0}, outputs: gponly}
    131 
    132 		gpstore         = regInfo{inputs: []regMask{gpspsb, gpsp, 0}}
    133 		gpstoreconst    = regInfo{inputs: []regMask{gpspsb, 0}}
    134 		gpstoreidx      = regInfo{inputs: []regMask{gpspsb, gpsp, gpsp, 0}}
    135 		gpstoreconstidx = regInfo{inputs: []regMask{gpspsb, gpsp, 0}}
    136 		gpstorexchg     = regInfo{inputs: []regMask{gp, gp, 0}, outputs: []regMask{gp}}
    137 		cmpxchg         = regInfo{inputs: []regMask{gp, ax, gp, 0}, outputs: []regMask{gp, 0}, clobbers: ax}
    138 
    139 		fp01     = regInfo{inputs: nil, outputs: fponly}
    140 		fp21     = regInfo{inputs: []regMask{fp, fp}, outputs: fponly}
    141 		fpgp     = regInfo{inputs: fponly, outputs: gponly}
    142 		gpfp     = regInfo{inputs: gponly, outputs: fponly}
    143 		fp11     = regInfo{inputs: fponly, outputs: fponly}
    144 		fp2flags = regInfo{inputs: []regMask{fp, fp}}
    145 
    146 		fpload    = regInfo{inputs: []regMask{gpspsb, 0}, outputs: fponly}
    147 		fploadidx = regInfo{inputs: []regMask{gpspsb, gpsp, 0}, outputs: fponly}
    148 
    149 		fpstore    = regInfo{inputs: []regMask{gpspsb, fp, 0}}
    150 		fpstoreidx = regInfo{inputs: []regMask{gpspsb, gpsp, fp, 0}}
    151 	)
    152 
    153 	var AMD64ops = []opData{
    154 		// fp ops
    155 		{name: "ADDSS", argLength: 2, reg: fp21, asm: "ADDSS", commutative: true, resultInArg0: true}, // fp32 add
    156 		{name: "ADDSD", argLength: 2, reg: fp21, asm: "ADDSD", commutative: true, resultInArg0: true}, // fp64 add
    157 		{name: "SUBSS", argLength: 2, reg: fp21, asm: "SUBSS", resultInArg0: true},                    // fp32 sub
    158 		{name: "SUBSD", argLength: 2, reg: fp21, asm: "SUBSD", resultInArg0: true},                    // fp64 sub
    159 		{name: "MULSS", argLength: 2, reg: fp21, asm: "MULSS", commutative: true, resultInArg0: true}, // fp32 mul
    160 		{name: "MULSD", argLength: 2, reg: fp21, asm: "MULSD", commutative: true, resultInArg0: true}, // fp64 mul
    161 		{name: "DIVSS", argLength: 2, reg: fp21, asm: "DIVSS", resultInArg0: true},                    // fp32 div
    162 		{name: "DIVSD", argLength: 2, reg: fp21, asm: "DIVSD", resultInArg0: true},                    // fp64 div
    163 
    164 		{name: "MOVSSload", argLength: 2, reg: fpload, asm: "MOVSS", aux: "SymOff", faultOnNilArg0: true}, // fp32 load
    165 		{name: "MOVSDload", argLength: 2, reg: fpload, asm: "MOVSD", aux: "SymOff", faultOnNilArg0: true}, // fp64 load
    166 		{name: "MOVSSconst", reg: fp01, asm: "MOVSS", aux: "Float32", rematerializeable: true},            // fp32 constant
    167 		{name: "MOVSDconst", reg: fp01, asm: "MOVSD", aux: "Float64", rematerializeable: true},            // fp64 constant
    168 		{name: "MOVSSloadidx1", argLength: 3, reg: fploadidx, asm: "MOVSS", aux: "SymOff"},                // fp32 load indexed by i
    169 		{name: "MOVSSloadidx4", argLength: 3, reg: fploadidx, asm: "MOVSS", aux: "SymOff"},                // fp32 load indexed by 4*i
    170 		{name: "MOVSDloadidx1", argLength: 3, reg: fploadidx, asm: "MOVSD", aux: "SymOff"},                // fp64 load indexed by i
    171 		{name: "MOVSDloadidx8", argLength: 3, reg: fploadidx, asm: "MOVSD", aux: "SymOff"},                // fp64 load indexed by 8*i
    172 
    173 		{name: "MOVSSstore", argLength: 3, reg: fpstore, asm: "MOVSS", aux: "SymOff", faultOnNilArg0: true}, // fp32 store
    174 		{name: "MOVSDstore", argLength: 3, reg: fpstore, asm: "MOVSD", aux: "SymOff", faultOnNilArg0: true}, // fp64 store
    175 		{name: "MOVSSstoreidx1", argLength: 4, reg: fpstoreidx, asm: "MOVSS", aux: "SymOff"},                // fp32 indexed by i store
    176 		{name: "MOVSSstoreidx4", argLength: 4, reg: fpstoreidx, asm: "MOVSS", aux: "SymOff"},                // fp32 indexed by 4i store
    177 		{name: "MOVSDstoreidx1", argLength: 4, reg: fpstoreidx, asm: "MOVSD", aux: "SymOff"},                // fp64 indexed by i store
    178 		{name: "MOVSDstoreidx8", argLength: 4, reg: fpstoreidx, asm: "MOVSD", aux: "SymOff"},                // fp64 indexed by 8i store
    179 
    180 		// binary ops
    181 		{name: "ADDQ", argLength: 2, reg: gp21sp, asm: "ADDQ", commutative: true, clobberFlags: true},                // arg0 + arg1
    182 		{name: "ADDL", argLength: 2, reg: gp21sp, asm: "ADDL", commutative: true, clobberFlags: true},                // arg0 + arg1
    183 		{name: "ADDQconst", argLength: 1, reg: gp11sp, asm: "ADDQ", aux: "Int64", typ: "UInt64", clobberFlags: true}, // arg0 + auxint
    184 		{name: "ADDLconst", argLength: 1, reg: gp11sp, asm: "ADDL", aux: "Int32", clobberFlags: true},                // arg0 + auxint
    185 
    186 		{name: "SUBQ", argLength: 2, reg: gp21, asm: "SUBQ", resultInArg0: true, clobberFlags: true},                    // arg0 - arg1
    187 		{name: "SUBL", argLength: 2, reg: gp21, asm: "SUBL", resultInArg0: true, clobberFlags: true},                    // arg0 - arg1
    188 		{name: "SUBQconst", argLength: 1, reg: gp11, asm: "SUBQ", aux: "Int64", resultInArg0: true, clobberFlags: true}, // arg0 - auxint
    189 		{name: "SUBLconst", argLength: 1, reg: gp11, asm: "SUBL", aux: "Int32", resultInArg0: true, clobberFlags: true}, // arg0 - auxint
    190 
    191 		{name: "MULQ", argLength: 2, reg: gp21, asm: "IMULQ", commutative: true, resultInArg0: true, clobberFlags: true}, // arg0 * arg1
    192 		{name: "MULL", argLength: 2, reg: gp21, asm: "IMULL", commutative: true, resultInArg0: true, clobberFlags: true}, // arg0 * arg1
    193 		{name: "MULQconst", argLength: 1, reg: gp11, asm: "IMULQ", aux: "Int64", resultInArg0: true, clobberFlags: true}, // arg0 * auxint
    194 		{name: "MULLconst", argLength: 1, reg: gp11, asm: "IMULL", aux: "Int32", resultInArg0: true, clobberFlags: true}, // arg0 * auxint
    195 
    196 		{name: "HMULQ", argLength: 2, reg: gp21hmul, asm: "IMULQ", clobberFlags: true}, // (arg0 * arg1) >> width
    197 		{name: "HMULL", argLength: 2, reg: gp21hmul, asm: "IMULL", clobberFlags: true}, // (arg0 * arg1) >> width
    198 		{name: "HMULW", argLength: 2, reg: gp21hmul, asm: "IMULW", clobberFlags: true}, // (arg0 * arg1) >> width
    199 		{name: "HMULB", argLength: 2, reg: gp21hmul, asm: "IMULB", clobberFlags: true}, // (arg0 * arg1) >> width
    200 		{name: "HMULQU", argLength: 2, reg: gp21hmul, asm: "MULQ", clobberFlags: true}, // (arg0 * arg1) >> width
    201 		{name: "HMULLU", argLength: 2, reg: gp21hmul, asm: "MULL", clobberFlags: true}, // (arg0 * arg1) >> width
    202 		{name: "HMULWU", argLength: 2, reg: gp21hmul, asm: "MULW", clobberFlags: true}, // (arg0 * arg1) >> width
    203 		{name: "HMULBU", argLength: 2, reg: gp21hmul, asm: "MULB", clobberFlags: true}, // (arg0 * arg1) >> width
    204 
    205 		{name: "AVGQU", argLength: 2, reg: gp21, commutative: true, resultInArg0: true, clobberFlags: true}, // (arg0 + arg1) / 2 as unsigned, all 64 result bits
    206 
    207 		{name: "DIVQ", argLength: 2, reg: gp11div, typ: "(Int64,Int64)", asm: "IDIVQ", clobberFlags: true},   // [arg0 / arg1, arg0 % arg1]
    208 		{name: "DIVL", argLength: 2, reg: gp11div, typ: "(Int32,Int32)", asm: "IDIVL", clobberFlags: true},   // [arg0 / arg1, arg0 % arg1]
    209 		{name: "DIVW", argLength: 2, reg: gp11div, typ: "(Int16,Int16)", asm: "IDIVW", clobberFlags: true},   // [arg0 / arg1, arg0 % arg1]
    210 		{name: "DIVQU", argLength: 2, reg: gp11div, typ: "(UInt64,UInt64)", asm: "DIVQ", clobberFlags: true}, // [arg0 / arg1, arg0 % arg1]
    211 		{name: "DIVLU", argLength: 2, reg: gp11div, typ: "(UInt32,UInt32)", asm: "DIVL", clobberFlags: true}, // [arg0 / arg1, arg0 % arg1]
    212 		{name: "DIVWU", argLength: 2, reg: gp11div, typ: "(UInt16,UInt16)", asm: "DIVW", clobberFlags: true}, // [arg0 / arg1, arg0 % arg1]
    213 
    214 		{name: "MULQU2", argLength: 2, reg: regInfo{inputs: []regMask{ax, gpsp}, outputs: []regMask{dx, ax}}, asm: "MULQ", clobberFlags: true},     // arg0 * arg1, returns (hi, lo)
    215 		{name: "DIVQU2", argLength: 3, reg: regInfo{inputs: []regMask{dx, ax, gpsp}, outputs: []regMask{ax, dx}}, asm: "DIVQ", clobberFlags: true}, // arg0:arg1 / arg2 (128-bit divided by 64-bit), returns (q, r)
    216 
    217 		{name: "ANDQ", argLength: 2, reg: gp21, asm: "ANDQ", commutative: true, resultInArg0: true, clobberFlags: true}, // arg0 & arg1
    218 		{name: "ANDL", argLength: 2, reg: gp21, asm: "ANDL", commutative: true, resultInArg0: true, clobberFlags: true}, // arg0 & arg1
    219 		{name: "ANDQconst", argLength: 1, reg: gp11, asm: "ANDQ", aux: "Int64", resultInArg0: true, clobberFlags: true}, // arg0 & auxint
    220 		{name: "ANDLconst", argLength: 1, reg: gp11, asm: "ANDL", aux: "Int32", resultInArg0: true, clobberFlags: true}, // arg0 & auxint
    221 
    222 		{name: "ORQ", argLength: 2, reg: gp21, asm: "ORQ", commutative: true, resultInArg0: true, clobberFlags: true}, // arg0 | arg1
    223 		{name: "ORL", argLength: 2, reg: gp21, asm: "ORL", commutative: true, resultInArg0: true, clobberFlags: true}, // arg0 | arg1
    224 		{name: "ORQconst", argLength: 1, reg: gp11, asm: "ORQ", aux: "Int64", resultInArg0: true, clobberFlags: true}, // arg0 | auxint
    225 		{name: "ORLconst", argLength: 1, reg: gp11, asm: "ORL", aux: "Int32", resultInArg0: true, clobberFlags: true}, // arg0 | auxint
    226 
    227 		{name: "XORQ", argLength: 2, reg: gp21, asm: "XORQ", commutative: true, resultInArg0: true, clobberFlags: true}, // arg0 ^ arg1
    228 		{name: "XORL", argLength: 2, reg: gp21, asm: "XORL", commutative: true, resultInArg0: true, clobberFlags: true}, // arg0 ^ arg1
    229 		{name: "XORQconst", argLength: 1, reg: gp11, asm: "XORQ", aux: "Int64", resultInArg0: true, clobberFlags: true}, // arg0 ^ auxint
    230 		{name: "XORLconst", argLength: 1, reg: gp11, asm: "XORL", aux: "Int32", resultInArg0: true, clobberFlags: true}, // arg0 ^ auxint
    231 
    232 		{name: "CMPQ", argLength: 2, reg: gp2flags, asm: "CMPQ", typ: "Flags"},                    // arg0 compare to arg1
    233 		{name: "CMPL", argLength: 2, reg: gp2flags, asm: "CMPL", typ: "Flags"},                    // arg0 compare to arg1
    234 		{name: "CMPW", argLength: 2, reg: gp2flags, asm: "CMPW", typ: "Flags"},                    // arg0 compare to arg1
    235 		{name: "CMPB", argLength: 2, reg: gp2flags, asm: "CMPB", typ: "Flags"},                    // arg0 compare to arg1
    236 		{name: "CMPQconst", argLength: 1, reg: gp1flags, asm: "CMPQ", typ: "Flags", aux: "Int64"}, // arg0 compare to auxint
    237 		{name: "CMPLconst", argLength: 1, reg: gp1flags, asm: "CMPL", typ: "Flags", aux: "Int32"}, // arg0 compare to auxint
    238 		{name: "CMPWconst", argLength: 1, reg: gp1flags, asm: "CMPW", typ: "Flags", aux: "Int16"}, // arg0 compare to auxint
    239 		{name: "CMPBconst", argLength: 1, reg: gp1flags, asm: "CMPB", typ: "Flags", aux: "Int8"},  // arg0 compare to auxint
    240 
    241 		{name: "UCOMISS", argLength: 2, reg: fp2flags, asm: "UCOMISS", typ: "Flags"}, // arg0 compare to arg1, f32
    242 		{name: "UCOMISD", argLength: 2, reg: fp2flags, asm: "UCOMISD", typ: "Flags"}, // arg0 compare to arg1, f64
    243 
    244 		{name: "TESTQ", argLength: 2, reg: gp2flags, asm: "TESTQ", typ: "Flags"},                    // (arg0 & arg1) compare to 0
    245 		{name: "TESTL", argLength: 2, reg: gp2flags, asm: "TESTL", typ: "Flags"},                    // (arg0 & arg1) compare to 0
    246 		{name: "TESTW", argLength: 2, reg: gp2flags, asm: "TESTW", typ: "Flags"},                    // (arg0 & arg1) compare to 0
    247 		{name: "TESTB", argLength: 2, reg: gp2flags, asm: "TESTB", typ: "Flags"},                    // (arg0 & arg1) compare to 0
    248 		{name: "TESTQconst", argLength: 1, reg: gp1flags, asm: "TESTQ", typ: "Flags", aux: "Int64"}, // (arg0 & auxint) compare to 0
    249 		{name: "TESTLconst", argLength: 1, reg: gp1flags, asm: "TESTL", typ: "Flags", aux: "Int32"}, // (arg0 & auxint) compare to 0
    250 		{name: "TESTWconst", argLength: 1, reg: gp1flags, asm: "TESTW", typ: "Flags", aux: "Int16"}, // (arg0 & auxint) compare to 0
    251 		{name: "TESTBconst", argLength: 1, reg: gp1flags, asm: "TESTB", typ: "Flags", aux: "Int8"},  // (arg0 & auxint) compare to 0
    252 
    253 		{name: "SHLQ", argLength: 2, reg: gp21shift, asm: "SHLQ", resultInArg0: true, clobberFlags: true},               // arg0 << arg1, shift amount is mod 64
    254 		{name: "SHLL", argLength: 2, reg: gp21shift, asm: "SHLL", resultInArg0: true, clobberFlags: true},               // arg0 << arg1, shift amount is mod 32
    255 		{name: "SHLQconst", argLength: 1, reg: gp11, asm: "SHLQ", aux: "Int64", resultInArg0: true, clobberFlags: true}, // arg0 << auxint, shift amount 0-63
    256 		{name: "SHLLconst", argLength: 1, reg: gp11, asm: "SHLL", aux: "Int32", resultInArg0: true, clobberFlags: true}, // arg0 << auxint, shift amount 0-31
    257 		// Note: x86 is weird, the 16 and 8 byte shifts still use all 5 bits of shift amount!
    258 
    259 		{name: "SHRQ", argLength: 2, reg: gp21shift, asm: "SHRQ", resultInArg0: true, clobberFlags: true},               // unsigned arg0 >> arg1, shift amount is mod 64
    260 		{name: "SHRL", argLength: 2, reg: gp21shift, asm: "SHRL", resultInArg0: true, clobberFlags: true},               // unsigned arg0 >> arg1, shift amount is mod 32
    261 		{name: "SHRW", argLength: 2, reg: gp21shift, asm: "SHRW", resultInArg0: true, clobberFlags: true},               // unsigned arg0 >> arg1, shift amount is mod 32
    262 		{name: "SHRB", argLength: 2, reg: gp21shift, asm: "SHRB", resultInArg0: true, clobberFlags: true},               // unsigned arg0 >> arg1, shift amount is mod 32
    263 		{name: "SHRQconst", argLength: 1, reg: gp11, asm: "SHRQ", aux: "Int64", resultInArg0: true, clobberFlags: true}, // unsigned arg0 >> auxint, shift amount 0-63
    264 		{name: "SHRLconst", argLength: 1, reg: gp11, asm: "SHRL", aux: "Int32", resultInArg0: true, clobberFlags: true}, // unsigned arg0 >> auxint, shift amount 0-31
    265 		{name: "SHRWconst", argLength: 1, reg: gp11, asm: "SHRW", aux: "Int16", resultInArg0: true, clobberFlags: true}, // unsigned arg0 >> auxint, shift amount 0-31
    266 		{name: "SHRBconst", argLength: 1, reg: gp11, asm: "SHRB", aux: "Int8", resultInArg0: true, clobberFlags: true},  // unsigned arg0 >> auxint, shift amount 0-31
    267 
    268 		{name: "SARQ", argLength: 2, reg: gp21shift, asm: "SARQ", resultInArg0: true, clobberFlags: true},               // signed arg0 >> arg1, shift amount is mod 64
    269 		{name: "SARL", argLength: 2, reg: gp21shift, asm: "SARL", resultInArg0: true, clobberFlags: true},               // signed arg0 >> arg1, shift amount is mod 32
    270 		{name: "SARW", argLength: 2, reg: gp21shift, asm: "SARW", resultInArg0: true, clobberFlags: true},               // signed arg0 >> arg1, shift amount is mod 32
    271 		{name: "SARB", argLength: 2, reg: gp21shift, asm: "SARB", resultInArg0: true, clobberFlags: true},               // signed arg0 >> arg1, shift amount is mod 32
    272 		{name: "SARQconst", argLength: 1, reg: gp11, asm: "SARQ", aux: "Int64", resultInArg0: true, clobberFlags: true}, // signed arg0 >> auxint, shift amount 0-63
    273 		{name: "SARLconst", argLength: 1, reg: gp11, asm: "SARL", aux: "Int32", resultInArg0: true, clobberFlags: true}, // signed arg0 >> auxint, shift amount 0-31
    274 		{name: "SARWconst", argLength: 1, reg: gp11, asm: "SARW", aux: "Int16", resultInArg0: true, clobberFlags: true}, // signed arg0 >> auxint, shift amount 0-31
    275 		{name: "SARBconst", argLength: 1, reg: gp11, asm: "SARB", aux: "Int8", resultInArg0: true, clobberFlags: true},  // signed arg0 >> auxint, shift amount 0-31
    276 
    277 		{name: "ROLQconst", argLength: 1, reg: gp11, asm: "ROLQ", aux: "Int64", resultInArg0: true, clobberFlags: true}, // arg0 rotate left auxint, rotate amount 0-63
    278 		{name: "ROLLconst", argLength: 1, reg: gp11, asm: "ROLL", aux: "Int32", resultInArg0: true, clobberFlags: true}, // arg0 rotate left auxint, rotate amount 0-31
    279 		{name: "ROLWconst", argLength: 1, reg: gp11, asm: "ROLW", aux: "Int16", resultInArg0: true, clobberFlags: true}, // arg0 rotate left auxint, rotate amount 0-15
    280 		{name: "ROLBconst", argLength: 1, reg: gp11, asm: "ROLB", aux: "Int8", resultInArg0: true, clobberFlags: true},  // arg0 rotate left auxint, rotate amount 0-7
    281 
    282 		// unary ops
    283 		{name: "NEGQ", argLength: 1, reg: gp11, asm: "NEGQ", resultInArg0: true, clobberFlags: true}, // -arg0
    284 		{name: "NEGL", argLength: 1, reg: gp11, asm: "NEGL", resultInArg0: true, clobberFlags: true}, // -arg0
    285 
    286 		{name: "NOTQ", argLength: 1, reg: gp11, asm: "NOTQ", resultInArg0: true, clobberFlags: true}, // ^arg0
    287 		{name: "NOTL", argLength: 1, reg: gp11, asm: "NOTL", resultInArg0: true, clobberFlags: true}, // ^arg0
    288 
    289 		// BSF{L,Q} returns a tuple [result, flags]
    290 		// result is undefined if the input is zero.
    291 		// flags are set to "equal" if the input is zero, "not equal" otherwise.
    292 		{name: "BSFQ", argLength: 1, reg: gp11flags, asm: "BSFQ", typ: "(UInt64,Flags)"}, // # of low-order zeroes in 64-bit arg
    293 		{name: "BSFL", argLength: 1, reg: gp11flags, asm: "BSFL", typ: "(UInt32,Flags)"}, // # of low-order zeroes in 32-bit arg
    294 
    295 		// Note ASM for ops moves whole register
    296 		//
    297 		{name: "CMOVQEQ", argLength: 3, reg: gp21, asm: "CMOVQEQ", resultInArg0: true}, // if arg2 encodes "equal" return arg1 else arg0
    298 		{name: "CMOVLEQ", argLength: 3, reg: gp21, asm: "CMOVLEQ", resultInArg0: true}, // if arg2 encodes "equal" return arg1 else arg0
    299 
    300 		{name: "BSWAPQ", argLength: 1, reg: gp11, asm: "BSWAPQ", resultInArg0: true, clobberFlags: true}, // arg0 swap bytes
    301 		{name: "BSWAPL", argLength: 1, reg: gp11, asm: "BSWAPL", resultInArg0: true, clobberFlags: true}, // arg0 swap bytes
    302 
    303 		{name: "SQRTSD", argLength: 1, reg: fp11, asm: "SQRTSD"}, // sqrt(arg0)
    304 
    305 		{name: "SBBQcarrymask", argLength: 1, reg: flagsgp, asm: "SBBQ"}, // (int64)(-1) if carry is set, 0 if carry is clear.
    306 		{name: "SBBLcarrymask", argLength: 1, reg: flagsgp, asm: "SBBL"}, // (int32)(-1) if carry is set, 0 if carry is clear.
    307 		// Note: SBBW and SBBB are subsumed by SBBL
    308 
    309 		{name: "SETEQ", argLength: 1, reg: readflags, asm: "SETEQ"}, // extract == condition from arg0
    310 		{name: "SETNE", argLength: 1, reg: readflags, asm: "SETNE"}, // extract != condition from arg0
    311 		{name: "SETL", argLength: 1, reg: readflags, asm: "SETLT"},  // extract signed < condition from arg0
    312 		{name: "SETLE", argLength: 1, reg: readflags, asm: "SETLE"}, // extract signed <= condition from arg0
    313 		{name: "SETG", argLength: 1, reg: readflags, asm: "SETGT"},  // extract signed > condition from arg0
    314 		{name: "SETGE", argLength: 1, reg: readflags, asm: "SETGE"}, // extract signed >= condition from arg0
    315 		{name: "SETB", argLength: 1, reg: readflags, asm: "SETCS"},  // extract unsigned < condition from arg0
    316 		{name: "SETBE", argLength: 1, reg: readflags, asm: "SETLS"}, // extract unsigned <= condition from arg0
    317 		{name: "SETA", argLength: 1, reg: readflags, asm: "SETHI"},  // extract unsigned > condition from arg0
    318 		{name: "SETAE", argLength: 1, reg: readflags, asm: "SETCC"}, // extract unsigned >= condition from arg0
    319 		// Need different opcodes for floating point conditions because
    320 		// any comparison involving a NaN is always FALSE and thus
    321 		// the patterns for inverting conditions cannot be used.
    322 		{name: "SETEQF", argLength: 1, reg: flagsgpax, asm: "SETEQ", clobberFlags: true}, // extract == condition from arg0
    323 		{name: "SETNEF", argLength: 1, reg: flagsgpax, asm: "SETNE", clobberFlags: true}, // extract != condition from arg0
    324 		{name: "SETORD", argLength: 1, reg: flagsgp, asm: "SETPC"},                       // extract "ordered" (No Nan present) condition from arg0
    325 		{name: "SETNAN", argLength: 1, reg: flagsgp, asm: "SETPS"},                       // extract "unordered" (Nan present) condition from arg0
    326 
    327 		{name: "SETGF", argLength: 1, reg: flagsgp, asm: "SETHI"},  // extract floating > condition from arg0
    328 		{name: "SETGEF", argLength: 1, reg: flagsgp, asm: "SETCC"}, // extract floating >= condition from arg0
    329 
    330 		{name: "MOVBQSX", argLength: 1, reg: gp11, asm: "MOVBQSX"}, // sign extend arg0 from int8 to int64
    331 		{name: "MOVBQZX", argLength: 1, reg: gp11, asm: "MOVBLZX"}, // zero extend arg0 from int8 to int64
    332 		{name: "MOVWQSX", argLength: 1, reg: gp11, asm: "MOVWQSX"}, // sign extend arg0 from int16 to int64
    333 		{name: "MOVWQZX", argLength: 1, reg: gp11, asm: "MOVWLZX"}, // zero extend arg0 from int16 to int64
    334 		{name: "MOVLQSX", argLength: 1, reg: gp11, asm: "MOVLQSX"}, // sign extend arg0 from int32 to int64
    335 		{name: "MOVLQZX", argLength: 1, reg: gp11, asm: "MOVL"},    // zero extend arg0 from int32 to int64
    336 
    337 		{name: "MOVLconst", reg: gp01, asm: "MOVL", typ: "UInt32", aux: "Int32", rematerializeable: true}, // 32 low bits of auxint
    338 		{name: "MOVQconst", reg: gp01, asm: "MOVQ", typ: "UInt64", aux: "Int64", rematerializeable: true}, // auxint
    339 
    340 		{name: "CVTTSD2SL", argLength: 1, reg: fpgp, asm: "CVTTSD2SL"}, // convert float64 to int32
    341 		{name: "CVTTSD2SQ", argLength: 1, reg: fpgp, asm: "CVTTSD2SQ"}, // convert float64 to int64
    342 		{name: "CVTTSS2SL", argLength: 1, reg: fpgp, asm: "CVTTSS2SL"}, // convert float32 to int32
    343 		{name: "CVTTSS2SQ", argLength: 1, reg: fpgp, asm: "CVTTSS2SQ"}, // convert float32 to int64
    344 		{name: "CVTSL2SS", argLength: 1, reg: gpfp, asm: "CVTSL2SS"},   // convert int32 to float32
    345 		{name: "CVTSL2SD", argLength: 1, reg: gpfp, asm: "CVTSL2SD"},   // convert int32 to float64
    346 		{name: "CVTSQ2SS", argLength: 1, reg: gpfp, asm: "CVTSQ2SS"},   // convert int64 to float32
    347 		{name: "CVTSQ2SD", argLength: 1, reg: gpfp, asm: "CVTSQ2SD"},   // convert int64 to float64
    348 		{name: "CVTSD2SS", argLength: 1, reg: fp11, asm: "CVTSD2SS"},   // convert float64 to float32
    349 		{name: "CVTSS2SD", argLength: 1, reg: fp11, asm: "CVTSS2SD"},   // convert float32 to float64
    350 
    351 		{name: "PXOR", argLength: 2, reg: fp21, asm: "PXOR", commutative: true, resultInArg0: true}, // exclusive or, applied to X regs for float negation.
    352 
    353 		{name: "LEAQ", argLength: 1, reg: gp11sb, asm: "LEAQ", aux: "SymOff", rematerializeable: true}, // arg0 + auxint + offset encoded in aux
    354 		{name: "LEAQ1", argLength: 2, reg: gp21sb, aux: "SymOff"},                                      // arg0 + arg1 + auxint + aux
    355 		{name: "LEAQ2", argLength: 2, reg: gp21sb, aux: "SymOff"},                                      // arg0 + 2*arg1 + auxint + aux
    356 		{name: "LEAQ4", argLength: 2, reg: gp21sb, aux: "SymOff"},                                      // arg0 + 4*arg1 + auxint + aux
    357 		{name: "LEAQ8", argLength: 2, reg: gp21sb, aux: "SymOff"},                                      // arg0 + 8*arg1 + auxint + aux
    358 		// Note: LEAQ{1,2,4,8} must not have OpSB as either argument.
    359 
    360 		{name: "LEAL", argLength: 1, reg: gp11sb, asm: "LEAL", aux: "SymOff", rematerializeable: true}, // arg0 + auxint + offset encoded in aux
    361 
    362 		// auxint+aux == add auxint and the offset of the symbol in aux (if any) to the effective address
    363 		{name: "MOVBload", argLength: 2, reg: gpload, asm: "MOVBLZX", aux: "SymOff", typ: "UInt8", faultOnNilArg0: true},  // load byte from arg0+auxint+aux. arg1=mem.  Zero extend.
    364 		{name: "MOVBQSXload", argLength: 2, reg: gpload, asm: "MOVBQSX", aux: "SymOff", faultOnNilArg0: true},             // ditto, sign extend to int64
    365 		{name: "MOVWload", argLength: 2, reg: gpload, asm: "MOVWLZX", aux: "SymOff", typ: "UInt16", faultOnNilArg0: true}, // load 2 bytes from arg0+auxint+aux. arg1=mem.  Zero extend.
    366 		{name: "MOVWQSXload", argLength: 2, reg: gpload, asm: "MOVWQSX", aux: "SymOff", faultOnNilArg0: true},             // ditto, sign extend to int64
    367 		{name: "MOVLload", argLength: 2, reg: gpload, asm: "MOVL", aux: "SymOff", typ: "UInt32", faultOnNilArg0: true},    // load 4 bytes from arg0+auxint+aux. arg1=mem.  Zero extend.
    368 		{name: "MOVLQSXload", argLength: 2, reg: gpload, asm: "MOVLQSX", aux: "SymOff", faultOnNilArg0: true},             // ditto, sign extend to int64
    369 		{name: "MOVQload", argLength: 2, reg: gpload, asm: "MOVQ", aux: "SymOff", typ: "UInt64", faultOnNilArg0: true},    // load 8 bytes from arg0+auxint+aux. arg1=mem
    370 		{name: "MOVBstore", argLength: 3, reg: gpstore, asm: "MOVB", aux: "SymOff", typ: "Mem", faultOnNilArg0: true},     // store byte in arg1 to arg0+auxint+aux. arg2=mem
    371 		{name: "MOVWstore", argLength: 3, reg: gpstore, asm: "MOVW", aux: "SymOff", typ: "Mem", faultOnNilArg0: true},     // store 2 bytes in arg1 to arg0+auxint+aux. arg2=mem
    372 		{name: "MOVLstore", argLength: 3, reg: gpstore, asm: "MOVL", aux: "SymOff", typ: "Mem", faultOnNilArg0: true},     // store 4 bytes in arg1 to arg0+auxint+aux. arg2=mem
    373 		{name: "MOVQstore", argLength: 3, reg: gpstore, asm: "MOVQ", aux: "SymOff", typ: "Mem", faultOnNilArg0: true},     // store 8 bytes in arg1 to arg0+auxint+aux. arg2=mem
    374 		{name: "MOVOload", argLength: 2, reg: fpload, asm: "MOVUPS", aux: "SymOff", typ: "Int128", faultOnNilArg0: true},  // load 16 bytes from arg0+auxint+aux. arg1=mem
    375 		{name: "MOVOstore", argLength: 3, reg: fpstore, asm: "MOVUPS", aux: "SymOff", typ: "Mem", faultOnNilArg0: true},   // store 16 bytes in arg1 to arg0+auxint+aux. arg2=mem
    376 
    377 		// indexed loads/stores
    378 		{name: "MOVBloadidx1", argLength: 3, reg: gploadidx, asm: "MOVBLZX", aux: "SymOff"}, // load a byte from arg0+arg1+auxint+aux. arg2=mem
    379 		{name: "MOVWloadidx1", argLength: 3, reg: gploadidx, asm: "MOVWLZX", aux: "SymOff"}, // load 2 bytes from arg0+arg1+auxint+aux. arg2=mem
    380 		{name: "MOVWloadidx2", argLength: 3, reg: gploadidx, asm: "MOVWLZX", aux: "SymOff"}, // load 2 bytes from arg0+2*arg1+auxint+aux. arg2=mem
    381 		{name: "MOVLloadidx1", argLength: 3, reg: gploadidx, asm: "MOVL", aux: "SymOff"},    // load 4 bytes from arg0+arg1+auxint+aux. arg2=mem
    382 		{name: "MOVLloadidx4", argLength: 3, reg: gploadidx, asm: "MOVL", aux: "SymOff"},    // load 4 bytes from arg0+4*arg1+auxint+aux. arg2=mem
    383 		{name: "MOVQloadidx1", argLength: 3, reg: gploadidx, asm: "MOVQ", aux: "SymOff"},    // load 8 bytes from arg0+arg1+auxint+aux. arg2=mem
    384 		{name: "MOVQloadidx8", argLength: 3, reg: gploadidx, asm: "MOVQ", aux: "SymOff"},    // load 8 bytes from arg0+8*arg1+auxint+aux. arg2=mem
    385 		// TODO: sign-extending indexed loads
    386 		{name: "MOVBstoreidx1", argLength: 4, reg: gpstoreidx, asm: "MOVB", aux: "SymOff"}, // store byte in arg2 to arg0+arg1+auxint+aux. arg3=mem
    387 		{name: "MOVWstoreidx1", argLength: 4, reg: gpstoreidx, asm: "MOVW", aux: "SymOff"}, // store 2 bytes in arg2 to arg0+arg1+auxint+aux. arg3=mem
    388 		{name: "MOVWstoreidx2", argLength: 4, reg: gpstoreidx, asm: "MOVW", aux: "SymOff"}, // store 2 bytes in arg2 to arg0+2*arg1+auxint+aux. arg3=mem
    389 		{name: "MOVLstoreidx1", argLength: 4, reg: gpstoreidx, asm: "MOVL", aux: "SymOff"}, // store 4 bytes in arg2 to arg0+arg1+auxint+aux. arg3=mem
    390 		{name: "MOVLstoreidx4", argLength: 4, reg: gpstoreidx, asm: "MOVL", aux: "SymOff"}, // store 4 bytes in arg2 to arg0+4*arg1+auxint+aux. arg3=mem
    391 		{name: "MOVQstoreidx1", argLength: 4, reg: gpstoreidx, asm: "MOVQ", aux: "SymOff"}, // store 8 bytes in arg2 to arg0+arg1+auxint+aux. arg3=mem
    392 		{name: "MOVQstoreidx8", argLength: 4, reg: gpstoreidx, asm: "MOVQ", aux: "SymOff"}, // store 8 bytes in arg2 to arg0+8*arg1+auxint+aux. arg3=mem
    393 		// TODO: add size-mismatched indexed loads, like MOVBstoreidx4.
    394 
    395 		// For storeconst ops, the AuxInt field encodes both
    396 		// the value to store and an address offset of the store.
    397 		// Cast AuxInt to a ValAndOff to extract Val and Off fields.
    398 		{name: "MOVBstoreconst", argLength: 2, reg: gpstoreconst, asm: "MOVB", aux: "SymValAndOff", typ: "Mem", faultOnNilArg0: true}, // store low byte of ValAndOff(AuxInt).Val() to arg0+ValAndOff(AuxInt).Off()+aux.  arg1=mem
    399 		{name: "MOVWstoreconst", argLength: 2, reg: gpstoreconst, asm: "MOVW", aux: "SymValAndOff", typ: "Mem", faultOnNilArg0: true}, // store low 2 bytes of ...
    400 		{name: "MOVLstoreconst", argLength: 2, reg: gpstoreconst, asm: "MOVL", aux: "SymValAndOff", typ: "Mem", faultOnNilArg0: true}, // store low 4 bytes of ...
    401 		{name: "MOVQstoreconst", argLength: 2, reg: gpstoreconst, asm: "MOVQ", aux: "SymValAndOff", typ: "Mem", faultOnNilArg0: true}, // store 8 bytes of ...
    402 
    403 		{name: "MOVBstoreconstidx1", argLength: 3, reg: gpstoreconstidx, asm: "MOVB", aux: "SymValAndOff", typ: "Mem"}, // store low byte of ValAndOff(AuxInt).Val() to arg0+1*arg1+ValAndOff(AuxInt).Off()+aux.  arg2=mem
    404 		{name: "MOVWstoreconstidx1", argLength: 3, reg: gpstoreconstidx, asm: "MOVW", aux: "SymValAndOff", typ: "Mem"}, // store low 2 bytes of ... arg1 ...
    405 		{name: "MOVWstoreconstidx2", argLength: 3, reg: gpstoreconstidx, asm: "MOVW", aux: "SymValAndOff", typ: "Mem"}, // store low 2 bytes of ... 2*arg1 ...
    406 		{name: "MOVLstoreconstidx1", argLength: 3, reg: gpstoreconstidx, asm: "MOVL", aux: "SymValAndOff", typ: "Mem"}, // store low 4 bytes of ... arg1 ...
    407 		{name: "MOVLstoreconstidx4", argLength: 3, reg: gpstoreconstidx, asm: "MOVL", aux: "SymValAndOff", typ: "Mem"}, // store low 4 bytes of ... 4*arg1 ...
    408 		{name: "MOVQstoreconstidx1", argLength: 3, reg: gpstoreconstidx, asm: "MOVQ", aux: "SymValAndOff", typ: "Mem"}, // store 8 bytes of ... arg1 ...
    409 		{name: "MOVQstoreconstidx8", argLength: 3, reg: gpstoreconstidx, asm: "MOVQ", aux: "SymValAndOff", typ: "Mem"}, // store 8 bytes of ... 8*arg1 ...
    410 
    411 		// arg0 = pointer to start of memory to zero
    412 		// arg1 = value to store (will always be zero)
    413 		// arg2 = mem
    414 		// auxint = # of bytes to zero
    415 		// returns mem
    416 		{
    417 			name:      "DUFFZERO",
    418 			aux:       "Int64",
    419 			argLength: 3,
    420 			reg: regInfo{
    421 				inputs:   []regMask{buildReg("DI"), buildReg("X0")},
    422 				clobbers: buildReg("DI"),
    423 			},
    424 			clobberFlags: true,
    425 		},
    426 		{name: "MOVOconst", reg: regInfo{nil, 0, []regMask{fp}}, typ: "Int128", aux: "Int128", rematerializeable: true},
    427 
    428 		// arg0 = address of memory to zero
    429 		// arg1 = # of 8-byte words to zero
    430 		// arg2 = value to store (will always be zero)
    431 		// arg3 = mem
    432 		// returns mem
    433 		{
    434 			name:      "REPSTOSQ",
    435 			argLength: 4,
    436 			reg: regInfo{
    437 				inputs:   []regMask{buildReg("DI"), buildReg("CX"), buildReg("AX")},
    438 				clobbers: buildReg("DI CX"),
    439 			},
    440 		},
    441 
    442 		{name: "CALLstatic", argLength: 1, reg: regInfo{clobbers: callerSave}, aux: "SymOff", clobberFlags: true, call: true},                                             // call static function aux.(*gc.Sym).  arg0=mem, auxint=argsize, returns mem
    443 		{name: "CALLclosure", argLength: 3, reg: regInfo{inputs: []regMask{gpsp, buildReg("DX"), 0}, clobbers: callerSave}, aux: "Int64", clobberFlags: true, call: true}, // call function via closure.  arg0=codeptr, arg1=closure, arg2=mem, auxint=argsize, returns mem
    444 		{name: "CALLdefer", argLength: 1, reg: regInfo{clobbers: callerSave}, aux: "Int64", clobberFlags: true, call: true},                                               // call deferproc.  arg0=mem, auxint=argsize, returns mem
    445 		{name: "CALLgo", argLength: 1, reg: regInfo{clobbers: callerSave}, aux: "Int64", clobberFlags: true, call: true},                                                  // call newproc.  arg0=mem, auxint=argsize, returns mem
    446 		{name: "CALLinter", argLength: 2, reg: regInfo{inputs: []regMask{gp}, clobbers: callerSave}, aux: "Int64", clobberFlags: true, call: true},                        // call fn by pointer.  arg0=codeptr, arg1=mem, auxint=argsize, returns mem
    447 
    448 		// arg0 = destination pointer
    449 		// arg1 = source pointer
    450 		// arg2 = mem
    451 		// auxint = offset from duffcopy symbol to call
    452 		// returns memory
    453 		{
    454 			name:      "DUFFCOPY",
    455 			aux:       "Int64",
    456 			argLength: 3,
    457 			reg: regInfo{
    458 				inputs:   []regMask{buildReg("DI"), buildReg("SI")},
    459 				clobbers: buildReg("DI SI X0"), // uses X0 as a temporary
    460 			},
    461 			clobberFlags: true,
    462 		},
    463 
    464 		// arg0 = destination pointer
    465 		// arg1 = source pointer
    466 		// arg2 = # of 8-byte words to copy
    467 		// arg3 = mem
    468 		// returns memory
    469 		{
    470 			name:      "REPMOVSQ",
    471 			argLength: 4,
    472 			reg: regInfo{
    473 				inputs:   []regMask{buildReg("DI"), buildReg("SI"), buildReg("CX")},
    474 				clobbers: buildReg("DI SI CX"),
    475 			},
    476 		},
    477 
    478 		// (InvertFlags (CMPQ a b)) == (CMPQ b a)
    479 		// So if we want (SETL (CMPQ a b)) but we can't do that because a is a constant,
    480 		// then we do (SETL (InvertFlags (CMPQ b a))) instead.
    481 		// Rewrites will convert this to (SETG (CMPQ b a)).
    482 		// InvertFlags is a pseudo-op which can't appear in assembly output.
    483 		{name: "InvertFlags", argLength: 1}, // reverse direction of arg0
    484 
    485 		// Pseudo-ops
    486 		{name: "LoweredGetG", argLength: 1, reg: gp01}, // arg0=mem
    487 		// Scheduler ensures LoweredGetClosurePtr occurs only in entry block,
    488 		// and sorts it to the very beginning of the block to prevent other
    489 		// use of DX (the closure pointer)
    490 		{name: "LoweredGetClosurePtr", reg: regInfo{outputs: []regMask{buildReg("DX")}}},
    491 		//arg0=ptr,arg1=mem, returns void.  Faults if ptr is nil.
    492 		{name: "LoweredNilCheck", argLength: 2, reg: regInfo{inputs: []regMask{gpsp}}, clobberFlags: true, nilCheck: true, faultOnNilArg0: true},
    493 
    494 		// MOVQconvert converts between pointers and integers.
    495 		// We have a special op for this so as to not confuse GC
    496 		// (particularly stack maps).  It takes a memory arg so it
    497 		// gets correctly ordered with respect to GC safepoints.
    498 		// arg0=ptr/int arg1=mem, output=int/ptr
    499 		{name: "MOVQconvert", argLength: 2, reg: gp11, asm: "MOVQ"},
    500 		{name: "MOVLconvert", argLength: 2, reg: gp11, asm: "MOVL"}, // amd64p32 equivalent
    501 
    502 		// Constant flag values. For any comparison, there are 5 possible
    503 		// outcomes: the three from the signed total order (<,==,>) and the
    504 		// three from the unsigned total order. The == cases overlap.
    505 		// Note: there's a sixth "unordered" outcome for floating-point
    506 		// comparisons, but we don't use such a beast yet.
    507 		// These ops are for temporary use by rewrite rules. They
    508 		// cannot appear in the generated assembly.
    509 		{name: "FlagEQ"},     // equal
    510 		{name: "FlagLT_ULT"}, // signed < and unsigned <
    511 		{name: "FlagLT_UGT"}, // signed < and unsigned >
    512 		{name: "FlagGT_UGT"}, // signed > and unsigned <
    513 		{name: "FlagGT_ULT"}, // signed > and unsigned >
    514 
    515 		// Atomic loads.  These are just normal loads but return <value,memory> tuples
    516 		// so they can be properly ordered with other loads.
    517 		// load from arg0+auxint+aux.  arg1=mem.
    518 		{name: "MOVLatomicload", argLength: 2, reg: gpload, asm: "MOVL", aux: "SymOff", faultOnNilArg0: true},
    519 		{name: "MOVQatomicload", argLength: 2, reg: gpload, asm: "MOVQ", aux: "SymOff", faultOnNilArg0: true},
    520 
    521 		// Atomic stores and exchanges.  Stores use XCHG to get the right memory ordering semantics.
    522 		// store arg0 to arg1+auxint+aux, arg2=mem.
    523 		// These ops return a tuple of <old contents of *(arg1+auxint+aux), memory>.
    524 		// Note: arg0 and arg1 are backwards compared to MOVLstore (to facilitate resultInArg0)!
    525 		{name: "XCHGL", argLength: 3, reg: gpstorexchg, asm: "XCHGL", aux: "SymOff", resultInArg0: true, faultOnNilArg1: true},
    526 		{name: "XCHGQ", argLength: 3, reg: gpstorexchg, asm: "XCHGQ", aux: "SymOff", resultInArg0: true, faultOnNilArg1: true},
    527 
    528 		// Atomic adds.
    529 		// *(arg1+auxint+aux) += arg0.  arg2=mem.
    530 		// Returns a tuple of <old contents of *(arg1+auxint+aux), memory>.
    531 		// Note: arg0 and arg1 are backwards compared to MOVLstore (to facilitate resultInArg0)!
    532 		{name: "XADDLlock", argLength: 3, reg: gpstorexchg, asm: "XADDL", typ: "(UInt32,Mem)", aux: "SymOff", resultInArg0: true, clobberFlags: true, faultOnNilArg1: true},
    533 		{name: "XADDQlock", argLength: 3, reg: gpstorexchg, asm: "XADDQ", typ: "(UInt64,Mem)", aux: "SymOff", resultInArg0: true, clobberFlags: true, faultOnNilArg1: true},
    534 		{name: "AddTupleFirst32", argLength: 2}, // arg0=tuple <x,y>.  Returns <x+arg1,y>.
    535 		{name: "AddTupleFirst64", argLength: 2}, // arg0=tuple <x,y>.  Returns <x+arg1,y>.
    536 
    537 		// Compare and swap.
    538 		// arg0 = pointer, arg1 = old value, arg2 = new value, arg3 = memory.
    539 		// if *(arg0+auxint+aux) == arg1 {
    540 		//   *(arg0+auxint+aux) = arg2
    541 		//   return (true, memory)
    542 		// } else {
    543 		//   return (false, memory)
    544 		// }
    545 		// Note that these instructions also return the old value in AX, but we ignore it.
    546 		// TODO: have these return flags instead of bool.  The current system generates:
    547 		//    CMPXCHGQ ...
    548 		//    SETEQ AX
    549 		//    CMPB  AX, $0
    550 		//    JNE ...
    551 		// instead of just
    552 		//    CMPXCHGQ ...
    553 		//    JEQ ...
    554 		// but we can't do that because memory-using ops can't generate flags yet
    555 		// (flagalloc wants to move flag-generating instructions around).
    556 		{name: "CMPXCHGLlock", argLength: 4, reg: cmpxchg, asm: "CMPXCHGL", aux: "SymOff", clobberFlags: true, faultOnNilArg0: true},
    557 		{name: "CMPXCHGQlock", argLength: 4, reg: cmpxchg, asm: "CMPXCHGQ", aux: "SymOff", clobberFlags: true, faultOnNilArg0: true},
    558 
    559 		// Atomic memory updates.
    560 		{name: "ANDBlock", argLength: 3, reg: gpstore, asm: "ANDB", aux: "SymOff", clobberFlags: true, faultOnNilArg0: true}, // *(arg0+auxint+aux) &= arg1
    561 		{name: "ORBlock", argLength: 3, reg: gpstore, asm: "ORB", aux: "SymOff", clobberFlags: true, faultOnNilArg0: true},   // *(arg0+auxint+aux) |= arg1
    562 	}
    563 
    564 	var AMD64blocks = []blockData{
    565 		{name: "EQ"},
    566 		{name: "NE"},
    567 		{name: "LT"},
    568 		{name: "LE"},
    569 		{name: "GT"},
    570 		{name: "GE"},
    571 		{name: "ULT"},
    572 		{name: "ULE"},
    573 		{name: "UGT"},
    574 		{name: "UGE"},
    575 		{name: "EQF"},
    576 		{name: "NEF"},
    577 		{name: "ORD"}, // FP, ordered comparison (parity zero)
    578 		{name: "NAN"}, // FP, unordered comparison (parity one)
    579 	}
    580 
    581 	archs = append(archs, arch{
    582 		name:            "AMD64",
    583 		pkg:             "cmd/internal/obj/x86",
    584 		genfile:         "../../amd64/ssa.go",
    585 		ops:             AMD64ops,
    586 		blocks:          AMD64blocks,
    587 		regnames:        regNamesAMD64,
    588 		gpregmask:       gp,
    589 		fpregmask:       fp,
    590 		framepointerreg: int8(num["BP"]),
    591 		linkreg:         -1, // not used
    592 	})
    593 }
    594