1
0
mirror of https://github.com/golang/go synced 2024-11-17 02:24:42 -07:00

cmd/compile: add PPC64 ssa ops to support carry chain arithmetic

These are the opcodes required to lower math/bits.Add64 and
math/bits.Sub64 directly into ssa form. Likewise, opcodes which
clobber CA are marked.

This does not alter code generation. It prepares for future changes
to support scheduling carry chaining ops more effectively, and then
changes to lower into PPC64 opcodes.

Change-Id: I2723deee4a98b3c365f691857512df53280ae40f
Reviewed-on: https://go-review.googlesource.com/c/go/+/394594
Trust: Keith Randall <khr@golang.org>
Run-TryBot: Paul Murphy <murp@ibm.com>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
This commit is contained in:
Paul E. Murphy 2022-03-22 08:41:34 -05:00 committed by Paul Murphy
parent dac0220841
commit 3d7ca9f7c0
3 changed files with 183 additions and 0 deletions

View File

@ -733,6 +733,41 @@ func ssaGenValue(s *ssagen.State, v *ssa.Value) {
p.To.Type = obj.TYPE_REG
p.To.Reg = v.Reg()
case ssa.OpPPC64ADDC, ssa.OpPPC64ADDE, ssa.OpPPC64SUBC, ssa.OpPPC64SUBE:
r := v.Reg0() // CA is the first, implied argument.
r1 := v.Args[0].Reg()
r2 := v.Args[1].Reg()
p := s.Prog(v.Op.Asm())
p.From.Type = obj.TYPE_REG
p.From.Reg = r2
p.Reg = r1
p.To.Type = obj.TYPE_REG
p.To.Reg = r
case ssa.OpPPC64ADDZEzero, ssa.OpPPC64SUBZEzero:
p := s.Prog(v.Op.Asm())
p.From.Type = obj.TYPE_REG
p.From.Reg = ppc64.REG_R0
p.To.Type = obj.TYPE_REG
p.To.Reg = v.Reg()
case ssa.OpPPC64ADDCconst:
p := s.Prog(v.Op.Asm())
p.Reg = v.Args[0].Reg()
p.From.Type = obj.TYPE_CONST
p.From.Offset = v.AuxInt
p.To.Type = obj.TYPE_REG
// Output is a pair, the second is the CA, which is implied.
p.To.Reg = v.Reg0()
case ssa.OpPPC64SUBCconst:
p := s.Prog(v.Op.Asm())
p.SetFrom3Const(v.AuxInt)
p.From.Type = obj.TYPE_REG
p.From.Reg = v.Args[0].Reg()
p.To.Type = obj.TYPE_REG
p.To.Reg = v.Reg0()
case ssa.OpPPC64SUBFCconst:
p := s.Prog(v.Op.Asm())
p.SetFrom3Const(v.AuxInt)

View File

@ -139,10 +139,14 @@ func init() {
// tls = buildReg("R13")
gp01 = regInfo{inputs: nil, outputs: []regMask{gp}}
gp11 = regInfo{inputs: []regMask{gp | sp | sb}, outputs: []regMask{gp}}
xergp = regInfo{inputs: []regMask{xer}, outputs: []regMask{gp}, clobbers: xer}
gp11cxer = regInfo{inputs: []regMask{gp | sp | sb}, outputs: []regMask{gp}, clobbers: xer}
gp11xer = regInfo{inputs: []regMask{gp | sp | sb}, outputs: []regMask{gp, xer}}
gp21 = regInfo{inputs: []regMask{gp | sp | sb, gp | sp | sb}, outputs: []regMask{gp}}
gp21a0 = regInfo{inputs: []regMask{gp, gp | sp | sb}, outputs: []regMask{gp}}
gp21cxer = regInfo{inputs: []regMask{gp | sp | sb, gp | sp | sb}, outputs: []regMask{gp}, clobbers: xer}
gp21xer = regInfo{inputs: []regMask{gp | sp | sb, gp | sp | sb}, outputs: []regMask{gp, xer}, clobbers: xer}
gp2xer1xer = regInfo{inputs: []regMask{gp | sp | sb, gp | sp | sb, xer}, outputs: []regMask{gp, xer}, clobbers: xer}
gp31 = regInfo{inputs: []regMask{gp | sp | sb, gp | sp | sb, gp | sp | sb}, outputs: []regMask{gp}}
gp22 = regInfo{inputs: []regMask{gp | sp | sb, gp | sp | sb}, outputs: []regMask{gp, gp}}
gp32 = regInfo{inputs: []regMask{gp | sp | sb, gp | sp | sb, gp | sp | sb}, outputs: []regMask{gp, gp}}
@ -223,6 +227,16 @@ func init() {
{name: "LoweredAdd64Carry", argLength: 3, reg: gp32, resultNotInArgs: true}, // arg0 + arg1 + carry, returns (sum, carry)
// Operations which consume or generate the CA (xer)
{name: "ADDC", argLength: 2, reg: gp21xer, asm: "ADDC", commutative: true, typ: "(UInt64, UInt64)"}, // arg0 + arg1 -> out, CA
{name: "SUBC", argLength: 2, reg: gp21xer, asm: "SUBC", typ: "(UInt64, UInt64)"}, // arg0 - arg1 -> out, CA
{name: "ADDCconst", argLength: 1, reg: gp11xer, asm: "ADDC", typ: "(UInt64, UInt64)", aux: "Int64"}, // arg0 + imm16 -> out, CA
{name: "SUBCconst", argLength: 1, reg: gp11xer, asm: "SUBC", typ: "(UInt64, UInt64)", aux: "Int64"}, // imm16 - arg0 -> out, CA
{name: "ADDE", argLength: 3, reg: gp2xer1xer, asm: "ADDE", typ: "(UInt64, UInt64)", commutative: true}, // arg0 + arg1 + CA (arg2) -> out, CA
{name: "SUBE", argLength: 3, reg: gp2xer1xer, asm: "SUBE", typ: "(UInt64, UInt64)"}, // arg0 - arg1 - CA (arg2) -> out, CA
{name: "ADDZEzero", argLength: 1, reg: xergp, asm: "ADDZE", typ: "UInt64"}, // CA (arg0) + $0 -> out
{name: "SUBZEzero", argLength: 1, reg: xergp, asm: "SUBZE", typ: "UInt64"}, // $0 - CA (arg0) -> out
{name: "SRADconst", argLength: 1, reg: gp11cxer, asm: "SRAD", aux: "Int64"}, // signed arg0 >> auxInt, 0 <= auxInt < 64, 64 bit width
{name: "SRAWconst", argLength: 1, reg: gp11cxer, asm: "SRAW", aux: "Int64"}, // signed arg0 >> auxInt, 0 <= auxInt < 32, 32 bit width
{name: "SRDconst", argLength: 1, reg: gp11, asm: "SRD", aux: "Int64"}, // unsigned arg0 >> auxInt, 0 <= auxInt < 64, 64 bit width

View File

@ -1911,6 +1911,14 @@ const (
OpPPC64CLRLSLWI
OpPPC64CLRLSLDI
OpPPC64LoweredAdd64Carry
OpPPC64ADDC
OpPPC64SUBC
OpPPC64ADDCconst
OpPPC64SUBCconst
OpPPC64ADDE
OpPPC64SUBE
OpPPC64ADDZEzero
OpPPC64SUBZEzero
OpPPC64SRADconst
OpPPC64SRAWconst
OpPPC64SRDconst
@ -25578,6 +25586,132 @@ var opcodeTable = [...]opInfo{
},
},
},
{
name: "ADDC",
argLen: 2,
commutative: true,
asm: ppc64.AADDC,
reg: regInfo{
inputs: []inputInfo{
{0, 1073733630}, // SP SB R3 R4 R5 R6 R7 R8 R9 R10 R11 R12 R14 R15 R16 R17 R18 R19 R20 R21 R22 R23 R24 R25 R26 R27 R28 R29
{1, 1073733630}, // SP SB R3 R4 R5 R6 R7 R8 R9 R10 R11 R12 R14 R15 R16 R17 R18 R19 R20 R21 R22 R23 R24 R25 R26 R27 R28 R29
},
clobbers: 9223372036854775808, // XER
outputs: []outputInfo{
{1, 9223372036854775808}, // XER
{0, 1073733624}, // R3 R4 R5 R6 R7 R8 R9 R10 R11 R12 R14 R15 R16 R17 R18 R19 R20 R21 R22 R23 R24 R25 R26 R27 R28 R29
},
},
},
{
name: "SUBC",
argLen: 2,
asm: ppc64.ASUBC,
reg: regInfo{
inputs: []inputInfo{
{0, 1073733630}, // SP SB R3 R4 R5 R6 R7 R8 R9 R10 R11 R12 R14 R15 R16 R17 R18 R19 R20 R21 R22 R23 R24 R25 R26 R27 R28 R29
{1, 1073733630}, // SP SB R3 R4 R5 R6 R7 R8 R9 R10 R11 R12 R14 R15 R16 R17 R18 R19 R20 R21 R22 R23 R24 R25 R26 R27 R28 R29
},
clobbers: 9223372036854775808, // XER
outputs: []outputInfo{
{1, 9223372036854775808}, // XER
{0, 1073733624}, // R3 R4 R5 R6 R7 R8 R9 R10 R11 R12 R14 R15 R16 R17 R18 R19 R20 R21 R22 R23 R24 R25 R26 R27 R28 R29
},
},
},
{
name: "ADDCconst",
auxType: auxInt64,
argLen: 1,
asm: ppc64.AADDC,
reg: regInfo{
inputs: []inputInfo{
{0, 1073733630}, // SP SB R3 R4 R5 R6 R7 R8 R9 R10 R11 R12 R14 R15 R16 R17 R18 R19 R20 R21 R22 R23 R24 R25 R26 R27 R28 R29
},
outputs: []outputInfo{
{1, 9223372036854775808}, // XER
{0, 1073733624}, // R3 R4 R5 R6 R7 R8 R9 R10 R11 R12 R14 R15 R16 R17 R18 R19 R20 R21 R22 R23 R24 R25 R26 R27 R28 R29
},
},
},
{
name: "SUBCconst",
auxType: auxInt64,
argLen: 1,
asm: ppc64.ASUBC,
reg: regInfo{
inputs: []inputInfo{
{0, 1073733630}, // SP SB R3 R4 R5 R6 R7 R8 R9 R10 R11 R12 R14 R15 R16 R17 R18 R19 R20 R21 R22 R23 R24 R25 R26 R27 R28 R29
},
outputs: []outputInfo{
{1, 9223372036854775808}, // XER
{0, 1073733624}, // R3 R4 R5 R6 R7 R8 R9 R10 R11 R12 R14 R15 R16 R17 R18 R19 R20 R21 R22 R23 R24 R25 R26 R27 R28 R29
},
},
},
{
name: "ADDE",
argLen: 3,
commutative: true,
asm: ppc64.AADDE,
reg: regInfo{
inputs: []inputInfo{
{2, 9223372036854775808}, // XER
{0, 1073733630}, // SP SB R3 R4 R5 R6 R7 R8 R9 R10 R11 R12 R14 R15 R16 R17 R18 R19 R20 R21 R22 R23 R24 R25 R26 R27 R28 R29
{1, 1073733630}, // SP SB R3 R4 R5 R6 R7 R8 R9 R10 R11 R12 R14 R15 R16 R17 R18 R19 R20 R21 R22 R23 R24 R25 R26 R27 R28 R29
},
clobbers: 9223372036854775808, // XER
outputs: []outputInfo{
{1, 9223372036854775808}, // XER
{0, 1073733624}, // R3 R4 R5 R6 R7 R8 R9 R10 R11 R12 R14 R15 R16 R17 R18 R19 R20 R21 R22 R23 R24 R25 R26 R27 R28 R29
},
},
},
{
name: "SUBE",
argLen: 3,
asm: ppc64.ASUBE,
reg: regInfo{
inputs: []inputInfo{
{2, 9223372036854775808}, // XER
{0, 1073733630}, // SP SB R3 R4 R5 R6 R7 R8 R9 R10 R11 R12 R14 R15 R16 R17 R18 R19 R20 R21 R22 R23 R24 R25 R26 R27 R28 R29
{1, 1073733630}, // SP SB R3 R4 R5 R6 R7 R8 R9 R10 R11 R12 R14 R15 R16 R17 R18 R19 R20 R21 R22 R23 R24 R25 R26 R27 R28 R29
},
clobbers: 9223372036854775808, // XER
outputs: []outputInfo{
{1, 9223372036854775808}, // XER
{0, 1073733624}, // R3 R4 R5 R6 R7 R8 R9 R10 R11 R12 R14 R15 R16 R17 R18 R19 R20 R21 R22 R23 R24 R25 R26 R27 R28 R29
},
},
},
{
name: "ADDZEzero",
argLen: 1,
asm: ppc64.AADDZE,
reg: regInfo{
inputs: []inputInfo{
{0, 9223372036854775808}, // XER
},
clobbers: 9223372036854775808, // XER
outputs: []outputInfo{
{0, 1073733624}, // R3 R4 R5 R6 R7 R8 R9 R10 R11 R12 R14 R15 R16 R17 R18 R19 R20 R21 R22 R23 R24 R25 R26 R27 R28 R29
},
},
},
{
name: "SUBZEzero",
argLen: 1,
asm: ppc64.ASUBZE,
reg: regInfo{
inputs: []inputInfo{
{0, 9223372036854775808}, // XER
},
clobbers: 9223372036854775808, // XER
outputs: []outputInfo{
{0, 1073733624}, // R3 R4 R5 R6 R7 R8 R9 R10 R11 R12 R14 R15 R16 R17 R18 R19 R20 R21 R22 R23 R24 R25 R26 R27 R28 R29
},
},
},
{
name: "SRADconst",
auxType: auxInt64,