cmd/compile: replace ANDL with MOV?ZX

According to "Intel 64 and IA-32 Architectures Optimization Reference
Manual" Section: "3.5.1.13 Zero-Latency MOV Instructions"
MOV?ZX instructions have zero latency on newer processors.

during make.bash:
(ANDLconst [0xFF] x) -> (MOVBQZX x)
applies 422 times
(ANDLconst [0xFFFF] x) -> (MOVWQZX x)
applies 114 times

Updates #15105

Change-Id: I10933af599de3c26449c52f4b5cd859331028f39
Reviewed-on: https://go-review.googlesource.com/31639
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
Run-TryBot: David Chase <drchase@google.com>

src/cmd/compile/internal/ssa/gen/AMD64.rules

@@ -631,7 +631,9 @@
 (CMPB x (MOVLconst [c])) -> (CMPBconst x [int64(int8(c))])
 (CMPB (MOVLconst [c]) x) -> (InvertFlags (CMPBconst x [int64(int8(c))]))
 
-// Using MOVBQZX instead of ANDQ is cheaper.
+// Using MOVZX instead of AND is cheaper.
+(ANDLconst [0xFF] x) -> (MOVBQZX x)
+(ANDLconst [0xFFFF] x) -> (MOVWQZX x)
 (ANDQconst [0xFF] x) -> (MOVBQZX x)
 (ANDQconst [0xFFFF] x) -> (MOVWQZX x)
 (ANDQconst [0xFFFFFFFF] x) -> (MOVLQZX x)

src/cmd/compile/internal/ssa/rewriteAMD64.go

@@ -1409,6 +1409,30 @@ func rewriteValueAMD64_OpAMD64ANDLconst(v *Value, config *Config) bool {
 		v.AddArg(x)
 		return true
 	}
+	// match: (ANDLconst [0xFF] x)
+	// cond:
+	// result: (MOVBQZX x)
+	for {
+		if v.AuxInt != 0xFF {
+			break
+		}
+		x := v.Args[0]
+		v.reset(OpAMD64MOVBQZX)
+		v.AddArg(x)
+		return true
+	}
+	// match: (ANDLconst [0xFFFF] x)
+	// cond:
+	// result: (MOVWQZX x)
+	for {
+		if v.AuxInt != 0xFFFF {
+			break
+		}
+		x := v.Args[0]
+		v.reset(OpAMD64MOVWQZX)
+		v.AddArg(x)
+		return true
+	}
 	// match: (ANDLconst [c] _)
 	// cond: int32(c)==0
 	// result: (MOVLconst [0])