cmd/compile/internal/ssa: optimize if control flow rule of loong64

goos: linux
goarch: loong64
pkg: cmd/compile/internal/test
cpu: Loongson-3A6000 @ 2500.00MHz
                                 │     old     │                 new                 │
                                 │   sec/op    │   sec/op     vs base                │
Switch8Predictable                 2.908n ± 0%   2.702n ± 2%   -7.10% (p=0.000 n=10)
Switch8Unpredictable               13.29n ± 0%   12.45n ± 0%   -6.32% (p=0.000 n=10)
Switch32Predictable                3.069n ± 0%   3.137n ± 0%   +2.23% (p=0.000 n=10)
Switch32Unpredictable              15.11n ± 0%   13.21n ± 0%  -12.57% (p=0.000 n=10)
SwitchStringPredictable            5.165n ± 0%   4.447n ± 0%  -13.90% (p=0.000 n=10)
SwitchStringUnpredictable          18.26n ± 0%   16.86n ± 0%   -7.67% (p=0.000 n=10)
SwitchTypePredictable              3.961n ± 0%   3.405n ± 0%  -14.04% (p=0.000 n=10)
SwitchTypeUnpredictable            17.92n ± 0%   16.17n ± 0%   -9.77% (p=0.000 n=10)
SwitchInterfaceTypePredictable     8.463n ± 0%   8.492n ± 0%   +0.34% (p=0.001 n=10)
SwitchInterfaceTypeUnpredictable   23.70n ± 0%   22.48n ± 0%   -5.15% (p=0.000 n=10)
geomean                            8.672n        8.018n        -7.54%

Change-Id: Iefbf8e9e784905779f1ebc9fd6a9b7a28cf2f6f4
Reviewed-on: https://go-review.googlesource.com/c/go/+/620815
Reviewed-by: abner chenc <chenguoqi@loongson.cn>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Michael Pratt <mpratt@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>

src/cmd/compile/internal/ssa/_gen/LOONG64.rules

@@ -453,6 +453,7 @@
 (GetCallerPC ...) => (LoweredGetCallerPC ...)
 
 (If cond yes no) => (NE (MOVBUreg <typ.UInt64> cond) yes no)
+(MOVBUreg x:(XOR (MOVVconst [1]) ((SGT|SGTU) _ _))) => x
 
 // Write barrier.
 (WB ...) => (LoweredWB ...)

src/cmd/compile/internal/ssa/rewriteLOONG64.go

@@ -1907,6 +1907,44 @@ func rewriteValueLOONG64_OpLOONG64MOVBUreg(v *Value) bool {
 		v.AddArg(x)
 		return true
 	}
+	// match: (MOVBUreg x:(XOR (MOVVconst [1]) (SGT _ _)))
+	// result: x
+	for {
+		x := v_0
+		if x.Op != OpLOONG64XOR {
+			break
+		}
+		_ = x.Args[1]
+		x_0 := x.Args[0]
+		x_1 := x.Args[1]
+		for _i0 := 0; _i0 <= 1; _i0, x_0, x_1 = _i0+1, x_1, x_0 {
+			if x_0.Op != OpLOONG64MOVVconst || auxIntToInt64(x_0.AuxInt) != 1 || x_1.Op != OpLOONG64SGT {
+				continue
+			}
+			v.copyOf(x)
+			return true
+		}
+		break
+	}
+	// match: (MOVBUreg x:(XOR (MOVVconst [1]) (SGTU _ _)))
+	// result: x
+	for {
+		x := v_0
+		if x.Op != OpLOONG64XOR {
+			break
+		}
+		_ = x.Args[1]
+		x_0 := x.Args[0]
+		x_1 := x.Args[1]
+		for _i0 := 0; _i0 <= 1; _i0, x_0, x_1 = _i0+1, x_1, x_0 {
+			if x_0.Op != OpLOONG64MOVVconst || auxIntToInt64(x_0.AuxInt) != 1 || x_1.Op != OpLOONG64SGTU {
+				continue
+			}
+			v.copyOf(x)
+			return true
+		}
+		break
+	}
 	// match: (MOVBUreg x:(SGT _ _))
 	// result: x
 	for {