From a329e21ccdc0a3df9630ccab2096ea658e17201b Mon Sep 17 00:00:00 2001
From: Keith Randall <khr@golang.org>
Date: Sat, 12 Sep 2015 13:26:57 -0700
Subject: [PATCH] [dev.ssa] cmd/compile/internal/ssa: implement OSQRT

Change-Id: Iec61ca1bdc064c29ceca6d47f600d5643d0a64dd
Reviewed-on: https://go-review.googlesource.com/14533
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
---
 src/cmd/compile/internal/gc/ssa.go             | 10 +++++++++-
 src/cmd/compile/internal/ssa/gen/AMD64.rules   |  2 ++
 src/cmd/compile/internal/ssa/gen/AMD64Ops.go   |  2 ++
 src/cmd/compile/internal/ssa/gen/genericOps.go |  2 ++
 src/cmd/compile/internal/ssa/opGen.go          | 18 ++++++++++++++++++
 src/cmd/compile/internal/ssa/rewriteAMD64.go   | 16 ++++++++++++++++
 6 files changed, 49 insertions(+), 1 deletion(-)

diff --git a/src/cmd/compile/internal/gc/ssa.go b/src/cmd/compile/internal/gc/ssa.go
index 7086a4f84a..a0bd07d681 100644
--- a/src/cmd/compile/internal/gc/ssa.go
+++ b/src/cmd/compile/internal/gc/ssa.go
@@ -988,6 +988,8 @@ var opToSSA = map[opAndType]ssa.Op{
 	opAndType{OLROT, TUINT16}: ssa.OpLrot16,
 	opAndType{OLROT, TUINT32}: ssa.OpLrot32,
 	opAndType{OLROT, TUINT64}: ssa.OpLrot64,
+
+	opAndType{OSQRT, TFLOAT64}: ssa.OpSqrt,
 }
 
 func (s *state) concreteEtype(t *Type) uint8 {
@@ -1643,7 +1645,7 @@ func (s *state) expr(n *Node) *ssa.Value {
 				s.newValue1(negop, tp, s.newValue1(ssa.OpComplexImag, tp, a)))
 		}
 		return s.newValue1(s.ssaOp(n.Op, n.Type), a.Type, a)
-	case ONOT, OCOM:
+	case ONOT, OCOM, OSQRT:
 		a := s.expr(n.Left)
 		return s.newValue1(s.ssaOp(n.Op, n.Type), a.Type, a)
 	case OIMAG, OREAL:
@@ -3325,6 +3327,12 @@ func (s *genState) genValue(v *ssa.Value) {
 		p := Prog(v.Op.Asm())
 		p.To.Type = obj.TYPE_REG
 		p.To.Reg = r
+	case ssa.OpAMD64SQRTSD:
+		p := Prog(v.Op.Asm())
+		p.From.Type = obj.TYPE_REG
+		p.From.Reg = regnum(v.Args[0])
+		p.To.Type = obj.TYPE_REG
+		p.To.Reg = regnum(v)
 	case ssa.OpSP, ssa.OpSB:
 		// nothing to do
 	case ssa.OpAMD64SETEQ, ssa.OpAMD64SETNE,
diff --git a/src/cmd/compile/internal/ssa/gen/AMD64.rules b/src/cmd/compile/internal/ssa/gen/AMD64.rules
index cba16eadc7..0591e8f8ef 100644
--- a/src/cmd/compile/internal/ssa/gen/AMD64.rules
+++ b/src/cmd/compile/internal/ssa/gen/AMD64.rules
@@ -89,6 +89,8 @@
 (Com16 x) -> (NOTW x)
 (Com8 x) -> (NOTB x)
 
+(Sqrt x) -> (SQRTSD x)
+
 // Note: we always extend to 64 bits even though some ops don't need that many result bits.
 (SignExt8to16 x) -> (MOVBQSX x)
 (SignExt8to32 x) -> (MOVBQSX x)
diff --git a/src/cmd/compile/internal/ssa/gen/AMD64Ops.go b/src/cmd/compile/internal/ssa/gen/AMD64Ops.go
index 0eee551f32..d5bd5b38e0 100644
--- a/src/cmd/compile/internal/ssa/gen/AMD64Ops.go
+++ b/src/cmd/compile/internal/ssa/gen/AMD64Ops.go
@@ -307,6 +307,8 @@ func init() {
 		{name: "NOTW", reg: gp11, asm: "NOTW"}, // ^arg0
 		{name: "NOTB", reg: gp11, asm: "NOTB"}, // ^arg0
 
+		{name: "SQRTSD", reg: fp11, asm: "SQRTSD"}, // sqrt(arg0)
+
 		{name: "SBBQcarrymask", reg: flagsgp, asm: "SBBQ"}, // (int64)(-1) if carry is set, 0 if carry is clear.
 		{name: "SBBLcarrymask", reg: flagsgp, asm: "SBBL"}, // (int32)(-1) if carry is set, 0 if carry is clear.
 		// Note: SBBW and SBBB are subsumed by SBBL
diff --git a/src/cmd/compile/internal/ssa/gen/genericOps.go b/src/cmd/compile/internal/ssa/gen/genericOps.go
index a9497644e8..dcaff959c3 100644
--- a/src/cmd/compile/internal/ssa/gen/genericOps.go
+++ b/src/cmd/compile/internal/ssa/gen/genericOps.go
@@ -232,6 +232,8 @@ var genericOps = []opData{
 	{name: "Com32"},
 	{name: "Com64"},
 
+	{name: "Sqrt"}, // sqrt(arg0), float64 only
+
 	// Data movement
 	{name: "Phi"},  // select an argument based on which predecessor block we came from
 	{name: "Copy"}, // output = arg0
diff --git a/src/cmd/compile/internal/ssa/opGen.go b/src/cmd/compile/internal/ssa/opGen.go
index 7918c209a4..f7f1ca3068 100644
--- a/src/cmd/compile/internal/ssa/opGen.go
+++ b/src/cmd/compile/internal/ssa/opGen.go
@@ -203,6 +203,7 @@ const (
 	OpAMD64NOTL
 	OpAMD64NOTW
 	OpAMD64NOTB
+	OpAMD64SQRTSD
 	OpAMD64SBBQcarrymask
 	OpAMD64SBBLcarrymask
 	OpAMD64SETEQ
@@ -448,6 +449,7 @@ const (
 	OpCom16
 	OpCom32
 	OpCom64
+	OpSqrt
 	OpPhi
 	OpCopy
 	OpConstBool
@@ -2361,6 +2363,18 @@ var opcodeTable = [...]opInfo{
 			},
 		},
 	},
+	{
+		name: "SQRTSD",
+		asm:  x86.ASQRTSD,
+		reg: regInfo{
+			inputs: []inputInfo{
+				{0, 4294901760}, // .X0 .X1 .X2 .X3 .X4 .X5 .X6 .X7 .X8 .X9 .X10 .X11 .X12 .X13 .X14 .X15
+			},
+			outputs: []regMask{
+				4294901760, // .X0 .X1 .X2 .X3 .X4 .X5 .X6 .X7 .X8 .X9 .X10 .X11 .X12 .X13 .X14 .X15
+			},
+		},
+	},
 	{
 		name: "SBBQcarrymask",
 		asm:  x86.ASBBQ,
@@ -3809,6 +3823,10 @@ var opcodeTable = [...]opInfo{
 		name:    "Com64",
 		generic: true,
 	},
+	{
+		name:    "Sqrt",
+		generic: true,
+	},
 	{
 		name:    "Phi",
 		generic: true,
diff --git a/src/cmd/compile/internal/ssa/rewriteAMD64.go b/src/cmd/compile/internal/ssa/rewriteAMD64.go
index 7917d8d971..cb6405d44d 100644
--- a/src/cmd/compile/internal/ssa/rewriteAMD64.go
+++ b/src/cmd/compile/internal/ssa/rewriteAMD64.go
@@ -9391,6 +9391,22 @@ func rewriteValueAMD64(v *Value, config *Config) bool {
 		goto endcef6d6001d3f25cf5dacee11a46e5c8c
 	endcef6d6001d3f25cf5dacee11a46e5c8c:
 		;
+	case OpSqrt:
+		// match: (Sqrt x)
+		// cond:
+		// result: (SQRTSD x)
+		{
+			x := v.Args[0]
+			v.Op = OpAMD64SQRTSD
+			v.AuxInt = 0
+			v.Aux = nil
+			v.resetArgs()
+			v.AddArg(x)
+			return true
+		}
+		goto end72f79ca9ec139e15856aaa03338cf543
+	end72f79ca9ec139e15856aaa03338cf543:
+		;
 	case OpStaticCall:
 		// match: (StaticCall [argwid] {target} mem)
 		// cond: