Home | History | Annotate | Download | only in gen
      1 // Copyright 2016 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 // L (long word) = 32 bit
     26 // W (word)      = 16 bit
     27 // B (byte)      = 8 bit
     28 
     29 // copied from ../../x86/reg.go
     30 var regNames386 = []string{
     31 	"AX",
     32 	"CX",
     33 	"DX",
     34 	"BX",
     35 	"SP",
     36 	"BP",
     37 	"SI",
     38 	"DI",
     39 	"X0",
     40 	"X1",
     41 	"X2",
     42 	"X3",
     43 	"X4",
     44 	"X5",
     45 	"X6",
     46 	"X7",
     47 
     48 	// pseudo-registers
     49 	"SB",
     50 }
     51 
     52 // Notes on 387 support.
     53 //  - The 387 has a weird stack-register setup for floating-point registers.
     54 //    We use these registers when SSE registers are not available (when GO386=387).
     55 //  - We use the same register names (X0-X7) but they refer to the 387
     56 //    floating-point registers. That way, most of the SSA backend is unchanged.
     57 //  - The instruction generation pass maintains an SSE->387 register mapping.
     58 //    This mapping is updated whenever the FP stack is pushed or popped so that
     59 //    we can always find a given SSE register even when the TOS pointer has changed.
     60 //  - To facilitate the mapping from SSE to 387, we enforce that
     61 //    every basic block starts and ends with an empty floating-point stack.
     62 
     63 func init() {
     64 	// Make map from reg names to reg integers.
     65 	if len(regNames386) > 64 {
     66 		panic("too many registers")
     67 	}
     68 	num := map[string]int{}
     69 	for i, name := range regNames386 {
     70 		num[name] = i
     71 	}
     72 	buildReg := func(s string) regMask {
     73 		m := regMask(0)
     74 		for _, r := range strings.Split(s, " ") {
     75 			if n, ok := num[r]; ok {
     76 				m |= regMask(1) << uint(n)
     77 				continue
     78 			}
     79 			panic("register " + r + " not found")
     80 		}
     81 		return m
     82 	}
     83 
     84 	// Common individual register masks
     85 	var (
     86 		ax         = buildReg("AX")
     87 		cx         = buildReg("CX")
     88 		dx         = buildReg("DX")
     89 		gp         = buildReg("AX CX DX BX BP SI DI")
     90 		fp         = buildReg("X0 X1 X2 X3 X4 X5 X6 X7")
     91 		gpsp       = gp | buildReg("SP")
     92 		gpspsb     = gpsp | buildReg("SB")
     93 		callerSave = gp | fp
     94 	)
     95 	// Common slices of register masks
     96 	var (
     97 		gponly = []regMask{gp}
     98 		fponly = []regMask{fp}
     99 	)
    100 
    101 	// Common regInfo
    102 	var (
    103 		gp01      = regInfo{inputs: nil, outputs: gponly}
    104 		gp11      = regInfo{inputs: []regMask{gp}, outputs: gponly}
    105 		gp11sp    = regInfo{inputs: []regMask{gpsp}, outputs: gponly}
    106 		gp11sb    = regInfo{inputs: []regMask{gpspsb}, outputs: gponly}
    107 		gp21      = regInfo{inputs: []regMask{gp, gp}, outputs: gponly}
    108 		gp11carry = regInfo{inputs: []regMask{gp}, outputs: []regMask{gp, 0}}
    109 		gp21carry = regInfo{inputs: []regMask{gp, gp}, outputs: []regMask{gp, 0}}
    110 		gp1carry1 = regInfo{inputs: []regMask{gp}, outputs: gponly}
    111 		gp2carry1 = regInfo{inputs: []regMask{gp, gp}, outputs: gponly}
    112 		gp21sp    = regInfo{inputs: []regMask{gpsp, gp}, outputs: gponly}
    113 		gp21sb    = regInfo{inputs: []regMask{gpspsb, gpsp}, outputs: gponly}
    114 		gp21shift = regInfo{inputs: []regMask{gp, cx}, outputs: []regMask{gp}}
    115 		gp11div   = regInfo{inputs: []regMask{ax, gpsp &^ dx}, outputs: []regMask{ax}, clobbers: dx}
    116 		gp21hmul  = regInfo{inputs: []regMask{ax, gpsp}, outputs: []regMask{dx}, clobbers: ax}
    117 		gp11mod   = regInfo{inputs: []regMask{ax, gpsp &^ dx}, outputs: []regMask{dx}, clobbers: ax}
    118 		gp21mul   = regInfo{inputs: []regMask{ax, gpsp}, outputs: []regMask{dx, ax}}
    119 
    120 		gp2flags = regInfo{inputs: []regMask{gpsp, gpsp}}
    121 		gp1flags = regInfo{inputs: []regMask{gpsp}}
    122 		flagsgp  = regInfo{inputs: nil, outputs: gponly}
    123 
    124 		readflags = regInfo{inputs: nil, outputs: gponly}
    125 		flagsgpax = regInfo{inputs: nil, clobbers: ax, outputs: []regMask{gp &^ ax}}
    126 
    127 		gpload    = regInfo{inputs: []regMask{gpspsb, 0}, outputs: gponly}
    128 		gploadidx = regInfo{inputs: []regMask{gpspsb, gpsp, 0}, outputs: gponly}
    129 
    130 		gpstore         = regInfo{inputs: []regMask{gpspsb, gpsp, 0}}
    131 		gpstoreconst    = regInfo{inputs: []regMask{gpspsb, 0}}
    132 		gpstoreidx      = regInfo{inputs: []regMask{gpspsb, gpsp, gpsp, 0}}
    133 		gpstoreconstidx = regInfo{inputs: []regMask{gpspsb, gpsp, 0}}
    134 
    135 		fp01     = regInfo{inputs: nil, outputs: fponly}
    136 		fp21     = regInfo{inputs: []regMask{fp, fp}, outputs: fponly}
    137 		fpgp     = regInfo{inputs: fponly, outputs: gponly}
    138 		gpfp     = regInfo{inputs: gponly, outputs: fponly}
    139 		fp11     = regInfo{inputs: fponly, outputs: fponly}
    140 		fp2flags = regInfo{inputs: []regMask{fp, fp}}
    141 
    142 		fpload    = regInfo{inputs: []regMask{gpspsb, 0}, outputs: fponly}
    143 		fploadidx = regInfo{inputs: []regMask{gpspsb, gpsp, 0}, outputs: fponly}
    144 
    145 		fpstore    = regInfo{inputs: []regMask{gpspsb, fp, 0}}
    146 		fpstoreidx = regInfo{inputs: []regMask{gpspsb, gpsp, fp, 0}}
    147 	)
    148 
    149 	var _386ops = []opData{
    150 		// fp ops
    151 		{name: "ADDSS", argLength: 2, reg: fp21, asm: "ADDSS", commutative: true, resultInArg0: true, usesScratch: true}, // fp32 add
    152 		{name: "ADDSD", argLength: 2, reg: fp21, asm: "ADDSD", commutative: true, resultInArg0: true},                    // fp64 add
    153 		{name: "SUBSS", argLength: 2, reg: fp21, asm: "SUBSS", resultInArg0: true, usesScratch: true},                    // fp32 sub
    154 		{name: "SUBSD", argLength: 2, reg: fp21, asm: "SUBSD", resultInArg0: true},                                       // fp64 sub
    155 		{name: "MULSS", argLength: 2, reg: fp21, asm: "MULSS", commutative: true, resultInArg0: true, usesScratch: true}, // fp32 mul
    156 		{name: "MULSD", argLength: 2, reg: fp21, asm: "MULSD", commutative: true, resultInArg0: true},                    // fp64 mul
    157 		{name: "DIVSS", argLength: 2, reg: fp21, asm: "DIVSS", resultInArg0: true, usesScratch: true},                    // fp32 div
    158 		{name: "DIVSD", argLength: 2, reg: fp21, asm: "DIVSD", resultInArg0: true},                                       // fp64 div
    159 
    160 		{name: "MOVSSload", argLength: 2, reg: fpload, asm: "MOVSS", aux: "SymOff", faultOnNilArg0: true}, // fp32 load
    161 		{name: "MOVSDload", argLength: 2, reg: fpload, asm: "MOVSD", aux: "SymOff", faultOnNilArg0: true}, // fp64 load
    162 		{name: "MOVSSconst", reg: fp01, asm: "MOVSS", aux: "Float32", rematerializeable: true},            // fp32 constant
    163 		{name: "MOVSDconst", reg: fp01, asm: "MOVSD", aux: "Float64", rematerializeable: true},            // fp64 constant
    164 		{name: "MOVSSloadidx1", argLength: 3, reg: fploadidx, asm: "MOVSS", aux: "SymOff"},                // fp32 load indexed by i
    165 		{name: "MOVSSloadidx4", argLength: 3, reg: fploadidx, asm: "MOVSS", aux: "SymOff"},                // fp32 load indexed by 4*i
    166 		{name: "MOVSDloadidx1", argLength: 3, reg: fploadidx, asm: "MOVSD", aux: "SymOff"},                // fp64 load indexed by i
    167 		{name: "MOVSDloadidx8", argLength: 3, reg: fploadidx, asm: "MOVSD", aux: "SymOff"},                // fp64 load indexed by 8*i
    168 
    169 		{name: "MOVSSstore", argLength: 3, reg: fpstore, asm: "MOVSS", aux: "SymOff", faultOnNilArg0: true}, // fp32 store
    170 		{name: "MOVSDstore", argLength: 3, reg: fpstore, asm: "MOVSD", aux: "SymOff", faultOnNilArg0: true}, // fp64 store
    171 		{name: "MOVSSstoreidx1", argLength: 4, reg: fpstoreidx, asm: "MOVSS", aux: "SymOff"},                // fp32 indexed by i store
    172 		{name: "MOVSSstoreidx4", argLength: 4, reg: fpstoreidx, asm: "MOVSS", aux: "SymOff"},                // fp32 indexed by 4i store
    173 		{name: "MOVSDstoreidx1", argLength: 4, reg: fpstoreidx, asm: "MOVSD", aux: "SymOff"},                // fp64 indexed by i store
    174 		{name: "MOVSDstoreidx8", argLength: 4, reg: fpstoreidx, asm: "MOVSD", aux: "SymOff"},                // fp64 indexed by 8i store
    175 
    176 		// binary ops
    177 		{name: "ADDL", argLength: 2, reg: gp21sp, asm: "ADDL", commutative: true, clobberFlags: true},                // arg0 + arg1
    178 		{name: "ADDLconst", argLength: 1, reg: gp11sp, asm: "ADDL", aux: "Int32", typ: "UInt32", clobberFlags: true}, // arg0 + auxint
    179 
    180 		{name: "ADDLcarry", argLength: 2, reg: gp21carry, asm: "ADDL", commutative: true, resultInArg0: true},                // arg0 + arg1, generates <carry,result> pair
    181 		{name: "ADDLconstcarry", argLength: 1, reg: gp11carry, asm: "ADDL", aux: "Int32", resultInArg0: true},                // arg0 + auxint, generates <carry,result> pair
    182 		{name: "ADCL", argLength: 3, reg: gp2carry1, asm: "ADCL", commutative: true, resultInArg0: true, clobberFlags: true}, // arg0+arg1+carry(arg2), where arg2 is flags
    183 		{name: "ADCLconst", argLength: 2, reg: gp1carry1, asm: "ADCL", aux: "Int32", resultInArg0: true, clobberFlags: true}, // arg0+auxint+carry(arg1), where arg1 is flags
    184 
    185 		{name: "SUBL", argLength: 2, reg: gp21, asm: "SUBL", resultInArg0: true, clobberFlags: true},                    // arg0 - arg1
    186 		{name: "SUBLconst", argLength: 1, reg: gp11, asm: "SUBL", aux: "Int32", resultInArg0: true, clobberFlags: true}, // arg0 - auxint
    187 
    188 		{name: "SUBLcarry", argLength: 2, reg: gp21carry, asm: "SUBL", resultInArg0: true},                                   // arg0-arg1, generates <borrow,result> pair
    189 		{name: "SUBLconstcarry", argLength: 1, reg: gp11carry, asm: "SUBL", aux: "Int32", resultInArg0: true},                // arg0-auxint, generates <borrow,result> pair
    190 		{name: "SBBL", argLength: 3, reg: gp2carry1, asm: "SBBL", resultInArg0: true, clobberFlags: true},                    // arg0-arg1-borrow(arg2), where arg2 is flags
    191 		{name: "SBBLconst", argLength: 2, reg: gp1carry1, asm: "SBBL", aux: "Int32", resultInArg0: true, clobberFlags: true}, // arg0-auxint-borrow(arg1), where arg1 is flags
    192 
    193 		{name: "MULL", argLength: 2, reg: gp21, asm: "IMULL", commutative: true, resultInArg0: true, clobberFlags: true}, // arg0 * arg1
    194 		{name: "MULLconst", argLength: 1, reg: gp11, asm: "IMULL", aux: "Int32", resultInArg0: true, clobberFlags: true}, // arg0 * auxint
    195 
    196 		{name: "HMULL", argLength: 2, reg: gp21hmul, asm: "IMULL", clobberFlags: true}, // (arg0 * arg1) >> width
    197 		{name: "HMULLU", argLength: 2, reg: gp21hmul, asm: "MULL", 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: "HMULWU", argLength: 2, reg: gp21hmul, asm: "MULW", clobberFlags: true}, // (arg0 * arg1) >> width
    201 		{name: "HMULBU", argLength: 2, reg: gp21hmul, asm: "MULB", clobberFlags: true}, // (arg0 * arg1) >> width
    202 
    203 		{name: "MULLQU", argLength: 2, reg: gp21mul, asm: "MULL", clobberFlags: true}, // arg0 * arg1, high 32 in result[0], low 32 in result[1]
    204 
    205 		{name: "DIVL", argLength: 2, reg: gp11div, asm: "IDIVL", clobberFlags: true}, // arg0 / arg1
    206 		{name: "DIVW", argLength: 2, reg: gp11div, asm: "IDIVW", clobberFlags: true}, // arg0 / arg1
    207 		{name: "DIVLU", argLength: 2, reg: gp11div, asm: "DIVL", clobberFlags: true}, // arg0 / arg1
    208 		{name: "DIVWU", argLength: 2, reg: gp11div, asm: "DIVW", clobberFlags: true}, // arg0 / arg1
    209 
    210 		{name: "MODL", argLength: 2, reg: gp11mod, asm: "IDIVL", clobberFlags: true}, // arg0 % arg1
    211 		{name: "MODW", argLength: 2, reg: gp11mod, asm: "IDIVW", clobberFlags: true}, // arg0 % arg1
    212 		{name: "MODLU", argLength: 2, reg: gp11mod, asm: "DIVL", clobberFlags: true}, // arg0 % arg1
    213 		{name: "MODWU", argLength: 2, reg: gp11mod, asm: "DIVW", clobberFlags: true}, // arg0 % arg1
    214 
    215 		{name: "ANDL", argLength: 2, reg: gp21, asm: "ANDL", commutative: true, resultInArg0: true, clobberFlags: true}, // arg0 & arg1
    216 		{name: "ANDLconst", argLength: 1, reg: gp11, asm: "ANDL", aux: "Int32", resultInArg0: true, clobberFlags: true}, // arg0 & auxint
    217 
    218 		{name: "ORL", argLength: 2, reg: gp21, asm: "ORL", commutative: true, resultInArg0: true, clobberFlags: true}, // arg0 | arg1
    219 		{name: "ORLconst", argLength: 1, reg: gp11, asm: "ORL", aux: "Int32", resultInArg0: true, clobberFlags: true}, // arg0 | auxint
    220 
    221 		{name: "XORL", argLength: 2, reg: gp21, asm: "XORL", commutative: true, resultInArg0: true, clobberFlags: true}, // arg0 ^ arg1
    222 		{name: "XORLconst", argLength: 1, reg: gp11, asm: "XORL", aux: "Int32", resultInArg0: true, clobberFlags: true}, // arg0 ^ auxint
    223 
    224 		{name: "CMPL", argLength: 2, reg: gp2flags, asm: "CMPL", typ: "Flags"},                    // arg0 compare to arg1
    225 		{name: "CMPW", argLength: 2, reg: gp2flags, asm: "CMPW", typ: "Flags"},                    // arg0 compare to arg1
    226 		{name: "CMPB", argLength: 2, reg: gp2flags, asm: "CMPB", typ: "Flags"},                    // arg0 compare to arg1
    227 		{name: "CMPLconst", argLength: 1, reg: gp1flags, asm: "CMPL", typ: "Flags", aux: "Int32"}, // arg0 compare to auxint
    228 		{name: "CMPWconst", argLength: 1, reg: gp1flags, asm: "CMPW", typ: "Flags", aux: "Int16"}, // arg0 compare to auxint
    229 		{name: "CMPBconst", argLength: 1, reg: gp1flags, asm: "CMPB", typ: "Flags", aux: "Int8"},  // arg0 compare to auxint
    230 
    231 		{name: "UCOMISS", argLength: 2, reg: fp2flags, asm: "UCOMISS", typ: "Flags", usesScratch: true}, // arg0 compare to arg1, f32
    232 		{name: "UCOMISD", argLength: 2, reg: fp2flags, asm: "UCOMISD", typ: "Flags", usesScratch: true}, // arg0 compare to arg1, f64
    233 
    234 		{name: "TESTL", argLength: 2, reg: gp2flags, asm: "TESTL", typ: "Flags"},                    // (arg0 & arg1) compare to 0
    235 		{name: "TESTW", argLength: 2, reg: gp2flags, asm: "TESTW", typ: "Flags"},                    // (arg0 & arg1) compare to 0
    236 		{name: "TESTB", argLength: 2, reg: gp2flags, asm: "TESTB", typ: "Flags"},                    // (arg0 & arg1) compare to 0
    237 		{name: "TESTLconst", argLength: 1, reg: gp1flags, asm: "TESTL", typ: "Flags", aux: "Int32"}, // (arg0 & auxint) compare to 0
    238 		{name: "TESTWconst", argLength: 1, reg: gp1flags, asm: "TESTW", typ: "Flags", aux: "Int16"}, // (arg0 & auxint) compare to 0
    239 		{name: "TESTBconst", argLength: 1, reg: gp1flags, asm: "TESTB", typ: "Flags", aux: "Int8"},  // (arg0 & auxint) compare to 0
    240 
    241 		{name: "SHLL", argLength: 2, reg: gp21shift, asm: "SHLL", resultInArg0: true, clobberFlags: true},               // arg0 << arg1, shift amount is mod 32
    242 		{name: "SHLLconst", argLength: 1, reg: gp11, asm: "SHLL", aux: "Int32", resultInArg0: true, clobberFlags: true}, // arg0 << auxint, shift amount 0-31
    243 		// Note: x86 is weird, the 16 and 8 byte shifts still use all 5 bits of shift amount!
    244 
    245 		{name: "SHRL", argLength: 2, reg: gp21shift, asm: "SHRL", resultInArg0: true, clobberFlags: true},               // unsigned arg0 >> arg1, shift amount is mod 32
    246 		{name: "SHRW", argLength: 2, reg: gp21shift, asm: "SHRW", resultInArg0: true, clobberFlags: true},               // unsigned arg0 >> arg1, shift amount is mod 32
    247 		{name: "SHRB", argLength: 2, reg: gp21shift, asm: "SHRB", resultInArg0: true, clobberFlags: true},               // unsigned arg0 >> arg1, shift amount is mod 32
    248 		{name: "SHRLconst", argLength: 1, reg: gp11, asm: "SHRL", aux: "Int32", resultInArg0: true, clobberFlags: true}, // unsigned arg0 >> auxint, shift amount 0-31
    249 		{name: "SHRWconst", argLength: 1, reg: gp11, asm: "SHRW", aux: "Int16", resultInArg0: true, clobberFlags: true}, // unsigned arg0 >> auxint, shift amount 0-31
    250 		{name: "SHRBconst", argLength: 1, reg: gp11, asm: "SHRB", aux: "Int8", resultInArg0: true, clobberFlags: true},  // unsigned arg0 >> auxint, shift amount 0-31
    251 
    252 		{name: "SARL", argLength: 2, reg: gp21shift, asm: "SARL", resultInArg0: true, clobberFlags: true},               // signed arg0 >> arg1, shift amount is mod 32
    253 		{name: "SARW", argLength: 2, reg: gp21shift, asm: "SARW", resultInArg0: true, clobberFlags: true},               // signed arg0 >> arg1, shift amount is mod 32
    254 		{name: "SARB", argLength: 2, reg: gp21shift, asm: "SARB", resultInArg0: true, clobberFlags: true},               // signed arg0 >> arg1, shift amount is mod 32
    255 		{name: "SARLconst", argLength: 1, reg: gp11, asm: "SARL", aux: "Int32", resultInArg0: true, clobberFlags: true}, // signed arg0 >> auxint, shift amount 0-31
    256 		{name: "SARWconst", argLength: 1, reg: gp11, asm: "SARW", aux: "Int16", resultInArg0: true, clobberFlags: true}, // signed arg0 >> auxint, shift amount 0-31
    257 		{name: "SARBconst", argLength: 1, reg: gp11, asm: "SARB", aux: "Int8", resultInArg0: true, clobberFlags: true},  // signed arg0 >> auxint, shift amount 0-31
    258 
    259 		{name: "ROLLconst", argLength: 1, reg: gp11, asm: "ROLL", aux: "Int32", resultInArg0: true, clobberFlags: true}, // arg0 rotate left auxint, rotate amount 0-31
    260 		{name: "ROLWconst", argLength: 1, reg: gp11, asm: "ROLW", aux: "Int16", resultInArg0: true, clobberFlags: true}, // arg0 rotate left auxint, rotate amount 0-15
    261 		{name: "ROLBconst", argLength: 1, reg: gp11, asm: "ROLB", aux: "Int8", resultInArg0: true, clobberFlags: true},  // arg0 rotate left auxint, rotate amount 0-7
    262 
    263 		// unary ops
    264 		{name: "NEGL", argLength: 1, reg: gp11, asm: "NEGL", resultInArg0: true, clobberFlags: true}, // -arg0
    265 
    266 		{name: "NOTL", argLength: 1, reg: gp11, asm: "NOTL", resultInArg0: true, clobberFlags: true}, // ^arg0
    267 
    268 		{name: "BSFL", argLength: 1, reg: gp11, asm: "BSFL", clobberFlags: true}, // arg0 # of low-order zeroes ; undef if zero
    269 		{name: "BSFW", argLength: 1, reg: gp11, asm: "BSFW", clobberFlags: true}, // arg0 # of low-order zeroes ; undef if zero
    270 
    271 		{name: "BSRL", argLength: 1, reg: gp11, asm: "BSRL", clobberFlags: true}, // arg0 # of high-order zeroes ; undef if zero
    272 		{name: "BSRW", argLength: 1, reg: gp11, asm: "BSRW", clobberFlags: true}, // arg0 # of high-order zeroes ; undef if zero
    273 
    274 		{name: "BSWAPL", argLength: 1, reg: gp11, asm: "BSWAPL", resultInArg0: true, clobberFlags: true}, // arg0 swap bytes
    275 
    276 		{name: "SQRTSD", argLength: 1, reg: fp11, asm: "SQRTSD"}, // sqrt(arg0)
    277 
    278 		{name: "SBBLcarrymask", argLength: 1, reg: flagsgp, asm: "SBBL"}, // (int32)(-1) if carry is set, 0 if carry is clear.
    279 		// Note: SBBW and SBBB are subsumed by SBBL
    280 
    281 		{name: "SETEQ", argLength: 1, reg: readflags, asm: "SETEQ"}, // extract == condition from arg0
    282 		{name: "SETNE", argLength: 1, reg: readflags, asm: "SETNE"}, // extract != condition from arg0
    283 		{name: "SETL", argLength: 1, reg: readflags, asm: "SETLT"},  // extract signed < condition from arg0
    284 		{name: "SETLE", argLength: 1, reg: readflags, asm: "SETLE"}, // extract signed <= condition from arg0
    285 		{name: "SETG", argLength: 1, reg: readflags, asm: "SETGT"},  // extract signed > condition from arg0
    286 		{name: "SETGE", argLength: 1, reg: readflags, asm: "SETGE"}, // extract signed >= condition from arg0
    287 		{name: "SETB", argLength: 1, reg: readflags, asm: "SETCS"},  // extract unsigned < condition from arg0
    288 		{name: "SETBE", argLength: 1, reg: readflags, asm: "SETLS"}, // extract unsigned <= condition from arg0
    289 		{name: "SETA", argLength: 1, reg: readflags, asm: "SETHI"},  // extract unsigned > condition from arg0
    290 		{name: "SETAE", argLength: 1, reg: readflags, asm: "SETCC"}, // extract unsigned >= condition from arg0
    291 		// Need different opcodes for floating point conditions because
    292 		// any comparison involving a NaN is always FALSE and thus
    293 		// the patterns for inverting conditions cannot be used.
    294 		{name: "SETEQF", argLength: 1, reg: flagsgpax, asm: "SETEQ", clobberFlags: true}, // extract == condition from arg0
    295 		{name: "SETNEF", argLength: 1, reg: flagsgpax, asm: "SETNE", clobberFlags: true}, // extract != condition from arg0
    296 		{name: "SETORD", argLength: 1, reg: flagsgp, asm: "SETPC"},                       // extract "ordered" (No Nan present) condition from arg0
    297 		{name: "SETNAN", argLength: 1, reg: flagsgp, asm: "SETPS"},                       // extract "unordered" (Nan present) condition from arg0
    298 
    299 		{name: "SETGF", argLength: 1, reg: flagsgp, asm: "SETHI"},  // extract floating > condition from arg0
    300 		{name: "SETGEF", argLength: 1, reg: flagsgp, asm: "SETCC"}, // extract floating >= condition from arg0
    301 
    302 		{name: "MOVBLSX", argLength: 1, reg: gp11, asm: "MOVBLSX"}, // sign extend arg0 from int8 to int32
    303 		{name: "MOVBLZX", argLength: 1, reg: gp11, asm: "MOVBLZX"}, // zero extend arg0 from int8 to int32
    304 		{name: "MOVWLSX", argLength: 1, reg: gp11, asm: "MOVWLSX"}, // sign extend arg0 from int16 to int32
    305 		{name: "MOVWLZX", argLength: 1, reg: gp11, asm: "MOVWLZX"}, // zero extend arg0 from int16 to int32
    306 
    307 		{name: "MOVLconst", reg: gp01, asm: "MOVL", typ: "UInt32", aux: "Int32", rematerializeable: true}, // 32 low bits of auxint
    308 
    309 		{name: "CVTTSD2SL", argLength: 1, reg: fpgp, asm: "CVTTSD2SL", usesScratch: true}, // convert float64 to int32
    310 		{name: "CVTTSS2SL", argLength: 1, reg: fpgp, asm: "CVTTSS2SL", usesScratch: true}, // convert float32 to int32
    311 		{name: "CVTSL2SS", argLength: 1, reg: gpfp, asm: "CVTSL2SS", usesScratch: true},   // convert int32 to float32
    312 		{name: "CVTSL2SD", argLength: 1, reg: gpfp, asm: "CVTSL2SD", usesScratch: true},   // convert int32 to float64
    313 		{name: "CVTSD2SS", argLength: 1, reg: fp11, asm: "CVTSD2SS", usesScratch: true},   // convert float64 to float32
    314 		{name: "CVTSS2SD", argLength: 1, reg: fp11, asm: "CVTSS2SD"},                      // convert float32 to float64
    315 
    316 		{name: "PXOR", argLength: 2, reg: fp21, asm: "PXOR", commutative: true, resultInArg0: true}, // exclusive or, applied to X regs for float negation.
    317 
    318 		{name: "LEAL", argLength: 1, reg: gp11sb, aux: "SymOff", rematerializeable: true}, // arg0 + auxint + offset encoded in aux
    319 		{name: "LEAL1", argLength: 2, reg: gp21sb, aux: "SymOff"},                         // arg0 + arg1 + auxint + aux
    320 		{name: "LEAL2", argLength: 2, reg: gp21sb, aux: "SymOff"},                         // arg0 + 2*arg1 + auxint + aux
    321 		{name: "LEAL4", argLength: 2, reg: gp21sb, aux: "SymOff"},                         // arg0 + 4*arg1 + auxint + aux
    322 		{name: "LEAL8", argLength: 2, reg: gp21sb, aux: "SymOff"},                         // arg0 + 8*arg1 + auxint + aux
    323 		// Note: LEAL{1,2,4,8} must not have OpSB as either argument.
    324 
    325 		// auxint+aux == add auxint and the offset of the symbol in aux (if any) to the effective address
    326 		{name: "MOVBload", argLength: 2, reg: gpload, asm: "MOVBLZX", aux: "SymOff", typ: "UInt8", faultOnNilArg0: true},  // load byte from arg0+auxint+aux. arg1=mem.  Zero extend.
    327 		{name: "MOVBLSXload", argLength: 2, reg: gpload, asm: "MOVBLSX", aux: "SymOff", faultOnNilArg0: true},             // ditto, sign extend to int32
    328 		{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.
    329 		{name: "MOVWLSXload", argLength: 2, reg: gpload, asm: "MOVWLSX", aux: "SymOff", faultOnNilArg0: true},             // ditto, sign extend to int32
    330 		{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.
    331 		{name: "MOVBstore", argLength: 3, reg: gpstore, asm: "MOVB", aux: "SymOff", typ: "Mem", faultOnNilArg0: true},     // store byte in arg1 to arg0+auxint+aux. arg2=mem
    332 		{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
    333 		{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
    334 
    335 		// indexed loads/stores
    336 		{name: "MOVBloadidx1", argLength: 3, reg: gploadidx, asm: "MOVBLZX", aux: "SymOff"}, // load a byte from arg0+arg1+auxint+aux. arg2=mem
    337 		{name: "MOVWloadidx1", argLength: 3, reg: gploadidx, asm: "MOVWLZX", aux: "SymOff"}, // load 2 bytes from arg0+arg1+auxint+aux. arg2=mem
    338 		{name: "MOVWloadidx2", argLength: 3, reg: gploadidx, asm: "MOVWLZX", aux: "SymOff"}, // load 2 bytes from arg0+2*arg1+auxint+aux. arg2=mem
    339 		{name: "MOVLloadidx1", argLength: 3, reg: gploadidx, asm: "MOVL", aux: "SymOff"},    // load 4 bytes from arg0+arg1+auxint+aux. arg2=mem
    340 		{name: "MOVLloadidx4", argLength: 3, reg: gploadidx, asm: "MOVL", aux: "SymOff"},    // load 4 bytes from arg0+4*arg1+auxint+aux. arg2=mem
    341 		// TODO: sign-extending indexed loads
    342 		{name: "MOVBstoreidx1", argLength: 4, reg: gpstoreidx, asm: "MOVB", aux: "SymOff"}, // store byte in arg2 to arg0+arg1+auxint+aux. arg3=mem
    343 		{name: "MOVWstoreidx1", argLength: 4, reg: gpstoreidx, asm: "MOVW", aux: "SymOff"}, // store 2 bytes in arg2 to arg0+arg1+auxint+aux. arg3=mem
    344 		{name: "MOVWstoreidx2", argLength: 4, reg: gpstoreidx, asm: "MOVW", aux: "SymOff"}, // store 2 bytes in arg2 to arg0+2*arg1+auxint+aux. arg3=mem
    345 		{name: "MOVLstoreidx1", argLength: 4, reg: gpstoreidx, asm: "MOVL", aux: "SymOff"}, // store 4 bytes in arg2 to arg0+arg1+auxint+aux. arg3=mem
    346 		{name: "MOVLstoreidx4", argLength: 4, reg: gpstoreidx, asm: "MOVL", aux: "SymOff"}, // store 4 bytes in arg2 to arg0+4*arg1+auxint+aux. arg3=mem
    347 		// TODO: add size-mismatched indexed loads, like MOVBstoreidx4.
    348 
    349 		// For storeconst ops, the AuxInt field encodes both
    350 		// the value to store and an address offset of the store.
    351 		// Cast AuxInt to a ValAndOff to extract Val and Off fields.
    352 		{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
    353 		{name: "MOVWstoreconst", argLength: 2, reg: gpstoreconst, asm: "MOVW", aux: "SymValAndOff", typ: "Mem", faultOnNilArg0: true}, // store low 2 bytes of ...
    354 		{name: "MOVLstoreconst", argLength: 2, reg: gpstoreconst, asm: "MOVL", aux: "SymValAndOff", typ: "Mem", faultOnNilArg0: true}, // store low 4 bytes of ...
    355 
    356 		{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
    357 		{name: "MOVWstoreconstidx1", argLength: 3, reg: gpstoreconstidx, asm: "MOVW", aux: "SymValAndOff", typ: "Mem"}, // store low 2 bytes of ... arg1 ...
    358 		{name: "MOVWstoreconstidx2", argLength: 3, reg: gpstoreconstidx, asm: "MOVW", aux: "SymValAndOff", typ: "Mem"}, // store low 2 bytes of ... 2*arg1 ...
    359 		{name: "MOVLstoreconstidx1", argLength: 3, reg: gpstoreconstidx, asm: "MOVL", aux: "SymValAndOff", typ: "Mem"}, // store low 4 bytes of ... arg1 ...
    360 		{name: "MOVLstoreconstidx4", argLength: 3, reg: gpstoreconstidx, asm: "MOVL", aux: "SymValAndOff", typ: "Mem"}, // store low 4 bytes of ... 4*arg1 ...
    361 
    362 		// arg0 = pointer to start of memory to zero
    363 		// arg1 = value to store (will always be zero)
    364 		// arg2 = mem
    365 		// auxint = offset into duffzero code to start executing
    366 		// returns mem
    367 		{
    368 			name:      "DUFFZERO",
    369 			aux:       "Int64",
    370 			argLength: 3,
    371 			reg: regInfo{
    372 				inputs:   []regMask{buildReg("DI"), buildReg("AX")},
    373 				clobbers: buildReg("DI CX"),
    374 				// Note: CX is only clobbered when dynamic linking.
    375 			},
    376 		},
    377 
    378 		// arg0 = address of memory to zero
    379 		// arg1 = # of 4-byte words to zero
    380 		// arg2 = value to store (will always be zero)
    381 		// arg3 = mem
    382 		// returns mem
    383 		{
    384 			name:      "REPSTOSL",
    385 			argLength: 4,
    386 			reg: regInfo{
    387 				inputs:   []regMask{buildReg("DI"), buildReg("CX"), buildReg("AX")},
    388 				clobbers: buildReg("DI CX"),
    389 			},
    390 		},
    391 
    392 		{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
    393 		{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
    394 		{name: "CALLdefer", argLength: 1, reg: regInfo{clobbers: callerSave}, aux: "Int64", clobberFlags: true, call: true},                                               // call deferproc.  arg0=mem, auxint=argsize, returns mem
    395 		{name: "CALLgo", argLength: 1, reg: regInfo{clobbers: callerSave}, aux: "Int64", clobberFlags: true, call: true},                                                  // call newproc.  arg0=mem, auxint=argsize, returns mem
    396 		{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
    397 
    398 		// arg0 = destination pointer
    399 		// arg1 = source pointer
    400 		// arg2 = mem
    401 		// auxint = offset from duffcopy symbol to call
    402 		// returns memory
    403 		{
    404 			name:      "DUFFCOPY",
    405 			aux:       "Int64",
    406 			argLength: 3,
    407 			reg: regInfo{
    408 				inputs:   []regMask{buildReg("DI"), buildReg("SI")},
    409 				clobbers: buildReg("DI SI CX"), // uses CX as a temporary
    410 			},
    411 			clobberFlags: true,
    412 		},
    413 
    414 		// arg0 = destination pointer
    415 		// arg1 = source pointer
    416 		// arg2 = # of 8-byte words to copy
    417 		// arg3 = mem
    418 		// returns memory
    419 		{
    420 			name:      "REPMOVSL",
    421 			argLength: 4,
    422 			reg: regInfo{
    423 				inputs:   []regMask{buildReg("DI"), buildReg("SI"), buildReg("CX")},
    424 				clobbers: buildReg("DI SI CX"),
    425 			},
    426 		},
    427 
    428 		// (InvertFlags (CMPL a b)) == (CMPL b a)
    429 		// So if we want (SETL (CMPL a b)) but we can't do that because a is a constant,
    430 		// then we do (SETL (InvertFlags (CMPL b a))) instead.
    431 		// Rewrites will convert this to (SETG (CMPL b a)).
    432 		// InvertFlags is a pseudo-op which can't appear in assembly output.
    433 		{name: "InvertFlags", argLength: 1}, // reverse direction of arg0
    434 
    435 		// Pseudo-ops
    436 		{name: "LoweredGetG", argLength: 1, reg: gp01}, // arg0=mem
    437 		// Scheduler ensures LoweredGetClosurePtr occurs only in entry block,
    438 		// and sorts it to the very beginning of the block to prevent other
    439 		// use of DX (the closure pointer)
    440 		{name: "LoweredGetClosurePtr", reg: regInfo{outputs: []regMask{buildReg("DX")}}},
    441 		//arg0=ptr,arg1=mem, returns void.  Faults if ptr is nil.
    442 		{name: "LoweredNilCheck", argLength: 2, reg: regInfo{inputs: []regMask{gpsp}}, clobberFlags: true, nilCheck: true, faultOnNilArg0: true},
    443 
    444 		// MOVLconvert converts between pointers and integers.
    445 		// We have a special op for this so as to not confuse GC
    446 		// (particularly stack maps).  It takes a memory arg so it
    447 		// gets correctly ordered with respect to GC safepoints.
    448 		// arg0=ptr/int arg1=mem, output=int/ptr
    449 		{name: "MOVLconvert", argLength: 2, reg: gp11, asm: "MOVL"},
    450 
    451 		// Constant flag values. For any comparison, there are 5 possible
    452 		// outcomes: the three from the signed total order (<,==,>) and the
    453 		// three from the unsigned total order. The == cases overlap.
    454 		// Note: there's a sixth "unordered" outcome for floating-point
    455 		// comparisons, but we don't use such a beast yet.
    456 		// These ops are for temporary use by rewrite rules. They
    457 		// cannot appear in the generated assembly.
    458 		{name: "FlagEQ"},     // equal
    459 		{name: "FlagLT_ULT"}, // signed < and unsigned <
    460 		{name: "FlagLT_UGT"}, // signed < and unsigned >
    461 		{name: "FlagGT_UGT"}, // signed > and unsigned <
    462 		{name: "FlagGT_ULT"}, // signed > and unsigned >
    463 
    464 		// Special op for -x on 387
    465 		{name: "FCHS", argLength: 1, reg: fp11},
    466 
    467 		// Special ops for PIC floating-point constants.
    468 		// MOVSXconst1 loads the address of the constant-pool entry into a register.
    469 		// MOVSXconst2 loads the constant from that address.
    470 		// MOVSXconst1 returns a pointer, but we type it as uint32 because it can never point to the Go heap.
    471 		{name: "MOVSSconst1", reg: gp01, typ: "UInt32", aux: "Float32"},
    472 		{name: "MOVSDconst1", reg: gp01, typ: "UInt32", aux: "Float64"},
    473 		{name: "MOVSSconst2", argLength: 1, reg: gpfp, asm: "MOVSS"},
    474 		{name: "MOVSDconst2", argLength: 1, reg: gpfp, asm: "MOVSD"},
    475 	}
    476 
    477 	var _386blocks = []blockData{
    478 		{name: "EQ"},
    479 		{name: "NE"},
    480 		{name: "LT"},
    481 		{name: "LE"},
    482 		{name: "GT"},
    483 		{name: "GE"},
    484 		{name: "ULT"},
    485 		{name: "ULE"},
    486 		{name: "UGT"},
    487 		{name: "UGE"},
    488 		{name: "EQF"},
    489 		{name: "NEF"},
    490 		{name: "ORD"}, // FP, ordered comparison (parity zero)
    491 		{name: "NAN"}, // FP, unordered comparison (parity one)
    492 	}
    493 
    494 	archs = append(archs, arch{
    495 		name:            "386",
    496 		pkg:             "cmd/internal/obj/x86",
    497 		genfile:         "../../x86/ssa.go",
    498 		ops:             _386ops,
    499 		blocks:          _386blocks,
    500 		regnames:        regNames386,
    501 		gpregmask:       gp,
    502 		fpregmask:       fp,
    503 		framepointerreg: int8(num["BP"]),
    504 		linkreg:         -1, // not used
    505 	})
    506 }
    507