Also, add the FABSS and FABSD pseudo instructions to the assembler.
The compiler could use FSGNJX[SD] directly but there doesn't seem
to be much advantage to doing so and the pseudo instructions are
easier to understand.
Change-Id: Ie8825b8aa8773c69cc4f07a32ef04abf4061d80d
Reviewed-on: https://go-review.googlesource.com/c/go/+/348989
Trust: Michael Munday <mike.munday@lowrisc.org>
Run-TryBot: Michael Munday <mike.munday@lowrisc.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Joel Sing <joel@sing.id.au>
SNEZ X15, X15 // b337f000
// F extension
+ FABSS F0, F1 // d3200020
FNEGS F0, F1 // d3100020
FNES F0, F1, X7 // d3a300a093c31300
// D extension
+ FABSD F0, F1 // d3200022
FNEGD F0, F1 // d3100022
FNED F0, F1, X5 // d3a200a293c21200
FLTD F0, F1, X5 // d39200a2
ssa.OpRISCV64FADDS, ssa.OpRISCV64FSUBS, ssa.OpRISCV64FMULS, ssa.OpRISCV64FDIVS,
ssa.OpRISCV64FEQS, ssa.OpRISCV64FNES, ssa.OpRISCV64FLTS, ssa.OpRISCV64FLES,
ssa.OpRISCV64FADDD, ssa.OpRISCV64FSUBD, ssa.OpRISCV64FMULD, ssa.OpRISCV64FDIVD,
- ssa.OpRISCV64FEQD, ssa.OpRISCV64FNED, ssa.OpRISCV64FLTD, ssa.OpRISCV64FLED:
+ ssa.OpRISCV64FEQD, ssa.OpRISCV64FNED, ssa.OpRISCV64FLTD, ssa.OpRISCV64FLED,
+ ssa.OpRISCV64FSGNJD:
r := v.Reg()
r1 := v.Args[0].Reg()
r2 := v.Args[1].Reg()
p.SetRestArgs([]obj.Addr{{Type: obj.TYPE_REG, Reg: r3}})
p.To.Type = obj.TYPE_REG
p.To.Reg = r
- case ssa.OpRISCV64FSQRTS, ssa.OpRISCV64FNEGS, ssa.OpRISCV64FSQRTD, ssa.OpRISCV64FNEGD,
+ case ssa.OpRISCV64FSQRTS, ssa.OpRISCV64FNEGS, ssa.OpRISCV64FABSD, ssa.OpRISCV64FSQRTD, ssa.OpRISCV64FNEGD,
ssa.OpRISCV64FMVSX, ssa.OpRISCV64FMVDX,
ssa.OpRISCV64FCVTSW, ssa.OpRISCV64FCVTSL, ssa.OpRISCV64FCVTWS, ssa.OpRISCV64FCVTLS,
ssa.OpRISCV64FCVTDW, ssa.OpRISCV64FCVTDL, ssa.OpRISCV64FCVTWD, ssa.OpRISCV64FCVTLD, ssa.OpRISCV64FCVTDS, ssa.OpRISCV64FCVTSD,
(Sqrt ...) => (FSQRTD ...)
(Sqrt32 ...) => (FSQRTS ...)
+(Copysign ...) => (FSGNJD ...)
+
+(Abs ...) => (FABSD ...)
+
(FMA ...) => (FMADDD ...)
// Sign and zero extension.
{name: "FNMSUBD", argLength: 3, reg: fp31, asm: "FNMSUBD", commutative: true, typ: "Float64"}, // -(arg0 * arg1) - arg2
{name: "FSQRTD", argLength: 1, reg: fp11, asm: "FSQRTD", typ: "Float64"}, // sqrt(arg0)
{name: "FNEGD", argLength: 1, reg: fp11, asm: "FNEGD", typ: "Float64"}, // -arg0
+ {name: "FABSD", argLength: 1, reg: fp11, asm: "FABSD", typ: "Float64"}, // abs(arg0)
+ {name: "FSGNJD", argLength: 2, reg: fp21, asm: "FSGNJD", typ: "Float64"}, // copy sign of arg1 to arg0
{name: "FMVDX", argLength: 1, reg: gpfp, asm: "FMVDX", typ: "Float64"}, // reinterpret arg0 as float
{name: "FCVTDW", argLength: 1, reg: gpfp, asm: "FCVTDW", typ: "Float64"}, // float64(low 32 bits of arg0)
{name: "FCVTDL", argLength: 1, reg: gpfp, asm: "FCVTDL", typ: "Float64"}, // float64(arg0)
OpRISCV64FNMSUBD
OpRISCV64FSQRTD
OpRISCV64FNEGD
+ OpRISCV64FABSD
+ OpRISCV64FSGNJD
OpRISCV64FMVDX
OpRISCV64FCVTDW
OpRISCV64FCVTDL
},
},
},
+ {
+ name: "FABSD",
+ argLen: 1,
+ asm: riscv.AFABSD,
+ reg: regInfo{
+ inputs: []inputInfo{
+ {0, 9223372034707292160}, // F0 F1 F2 F3 F4 F5 F6 F7 F8 F9 F10 F11 F12 F13 F14 F15 F16 F17 F18 F19 F20 F21 F22 F23 F24 F25 F26 F27 F28 F29 F30 F31
+ },
+ outputs: []outputInfo{
+ {0, 9223372034707292160}, // F0 F1 F2 F3 F4 F5 F6 F7 F8 F9 F10 F11 F12 F13 F14 F15 F16 F17 F18 F19 F20 F21 F22 F23 F24 F25 F26 F27 F28 F29 F30 F31
+ },
+ },
+ },
+ {
+ name: "FSGNJD",
+ argLen: 2,
+ asm: riscv.AFSGNJD,
+ reg: regInfo{
+ inputs: []inputInfo{
+ {0, 9223372034707292160}, // F0 F1 F2 F3 F4 F5 F6 F7 F8 F9 F10 F11 F12 F13 F14 F15 F16 F17 F18 F19 F20 F21 F22 F23 F24 F25 F26 F27 F28 F29 F30 F31
+ {1, 9223372034707292160}, // F0 F1 F2 F3 F4 F5 F6 F7 F8 F9 F10 F11 F12 F13 F14 F15 F16 F17 F18 F19 F20 F21 F22 F23 F24 F25 F26 F27 F28 F29 F30 F31
+ },
+ outputs: []outputInfo{
+ {0, 9223372034707292160}, // F0 F1 F2 F3 F4 F5 F6 F7 F8 F9 F10 F11 F12 F13 F14 F15 F16 F17 F18 F19 F20 F21 F22 F23 F24 F25 F26 F27 F28 F29 F30 F31
+ },
+ },
+ },
{
name: "FMVDX",
argLen: 1,
func rewriteValueRISCV64(v *Value) bool {
switch v.Op {
+ case OpAbs:
+ v.Op = OpRISCV64FABSD
+ return true
case OpAdd16:
v.Op = OpRISCV64ADD
return true
case OpConvert:
v.Op = OpRISCV64MOVconvert
return true
+ case OpCopysign:
+ v.Op = OpRISCV64FSGNJD
+ return true
case OpCvt32Fto32:
v.Op = OpRISCV64FCVTWS
return true
func(s *state, n *ir.CallExpr, args []*ssa.Value) *ssa.Value {
return s.newValue1(ssa.OpAbs, types.Types[types.TFLOAT64], args[0])
},
- sys.ARM64, sys.ARM, sys.PPC64, sys.Wasm)
+ sys.ARM64, sys.ARM, sys.PPC64, sys.RISCV64, sys.Wasm)
addF("math", "Copysign",
func(s *state, n *ir.CallExpr, args []*ssa.Value) *ssa.Value {
return s.newValue2(ssa.OpCopysign, types.Types[types.TFLOAT64], args[0], args[1])
},
- sys.PPC64, sys.Wasm)
+ sys.PPC64, sys.RISCV64, sys.Wasm)
addF("math", "FMA",
func(s *state, n *ir.CallExpr, args []*ssa.Value) *ssa.Value {
return s.newValue3(ssa.OpFMA, types.Types[types.TFLOAT64], args[0], args[1], args[2])
"BLEZ",
"BLTZ",
"BNEZ",
+ "FABSD",
+ "FABSS",
"FNEGD",
"FNEGS",
"FNED",
ABLEZ
ABLTZ
ABNEZ
+ AFABSD
+ AFABSS
AFNEGD
AFNEGS
AFNED
ins.as = ASLTU
ins.rs1 = REG_ZERO
+ case AFABSS:
+ // FABSS rs, rd -> FSGNJXS rs, rs, rd
+ ins.as = AFSGNJXS
+ ins.rs1 = uint32(p.From.Reg)
+
+ case AFABSD:
+ // FABSD rs, rd -> FSGNJXD rs, rs, rd
+ ins.as = AFSGNJXD
+ ins.rs1 = uint32(p.From.Reg)
+
case AFNEGS:
// FNEGS rs, rd -> FSGNJNS rs, rs, rd
ins.as = AFSGNJNS
// s390x:"LPDFR\t",-"MOVD\t" (no integer load/store)
// ppc64:"FABS\t"
// ppc64le:"FABS\t"
+ // riscv64:"FABSD\t"
// wasm:"F64Abs"
// arm/6:"ABSD\t"
sink64[0] = math.Abs(x)
// s390x:"CPSDR",-"MOVD" (no integer load/store)
// ppc64:"FCPSGN"
// ppc64le:"FCPSGN"
+ // riscv64:"FSGNJD"
// wasm:"F64Copysign"
sink64[0] = math.Copysign(a, b)
// s390x:"LNDFR\t",-"MOVD\t" (no integer load/store)
// ppc64:"FCPSGN"
// ppc64le:"FCPSGN"
+ // riscv64:"FSGNJD"
// arm64:"ORR", -"AND"
sink64[1] = math.Copysign(c, -1)
// s390x:"CPSDR\t",-"MOVD\t" (no integer load/store)
// ppc64:"FCPSGN"
// ppc64le:"FCPSGN"
+ // riscv64:"FSGNJD"
sink64[3] = math.Copysign(-1, c)
}