cmd/compile: remove superfluous signed right shift used for signed division by 2

A signed right shift before an unsigned right shift by register width-1
(extracts the sign bit) is superflous.

trigger counts during ./make.bash
 0   (Rsh8U  (Rsh8  x _) 7  ) -> (Rsh8U  x 7 )
 0   (Rsh16U (Rsh16 x _) 15 ) -> (Rsh16U x 15)
 2   (Rsh32U (Rsh32 x _) 31 ) -> (Rsh32U x 31)
 251 (Rsh64U (Rsh64 x _) 63 ) -> (Rsh64U x 63)

Changes the instructions generated on AMD64 for x / 2 where
x is a signed integer from:

 MOVQ    AX, CX
 SARQ    $63, AX
 SHRQ    $63, AX
 ADDQ    CX, AX
 SARQ    $1, AX

to:

 MOVQ    AX, CX
 SHRQ    $63, AX
 ADDQ    CX, AX
 SARQ    $1, AX

Change-Id: I86321ae8fc9dc24b8fa9eb80aa5c7299eff8c9dc
Reviewed-on: https://go-review.googlesource.com/115956
Run-TryBot: Martin Möhrmann <moehrmann@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>

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

@@ -371,6 +371,12 @@
 (Rsh16Ux64 <t> (Rsh16Ux64 x (Const64 [c])) (Const64 [d])) && !uaddOvf(c,d) -> (Rsh16Ux64 x (Const64 <t> [c+d]))
 (Rsh8Ux64  <t> (Rsh8Ux64  x (Const64 [c])) (Const64 [d])) && !uaddOvf(c,d) -> (Rsh8Ux64  x (Const64 <t> [c+d]))
 
+// Remove signed right shift before an unsigned right shift that extracts the sign bit.
+(Rsh8Ux64  (Rsh8x64  x _) (Const64 <t> [7] )) -> (Rsh8Ux64  x (Const64 <t> [7] ))
+(Rsh16Ux64 (Rsh16x64 x _) (Const64 <t> [15])) -> (Rsh16Ux64 x (Const64 <t> [15]))
+(Rsh32Ux64 (Rsh32x64 x _) (Const64 <t> [31])) -> (Rsh32Ux64 x (Const64 <t> [31]))
+(Rsh64Ux64 (Rsh64x64 x _) (Const64 <t> [63])) -> (Rsh64Ux64 x (Const64 <t> [63]))
+
 // ((x >> c1) << c2) >> c3
 (Rsh(64|32|16|8)Ux64 (Lsh(64|32|16|8)x64 (Rsh(64|32|16|8)Ux64 x (Const64 [c1])) (Const64 [c2])) (Const64 [c3]))
   && uint64(c1) >= uint64(c2) && uint64(c3) >= uint64(c2) && !uaddOvf(c1-c2, c3)

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

@@ -24905,6 +24905,32 @@ func rewriteValuegeneric_OpRsh16Ux64_0(v *Value) bool {
 		v.AddArg(v0)
 		return true
 	}
+	// match: (Rsh16Ux64 (Rsh16x64 x _) (Const64 <t> [15]))
+	// cond:
+	// result: (Rsh16Ux64 x (Const64 <t> [15]))
+	for {
+		_ = v.Args[1]
+		v_0 := v.Args[0]
+		if v_0.Op != OpRsh16x64 {
+			break
+		}
+		_ = v_0.Args[1]
+		x := v_0.Args[0]
+		v_1 := v.Args[1]
+		if v_1.Op != OpConst64 {
+			break
+		}
+		t := v_1.Type
+		if v_1.AuxInt != 15 {
+			break
+		}
+		v.reset(OpRsh16Ux64)
+		v.AddArg(x)
+		v0 := b.NewValue0(v.Pos, OpConst64, t)
+		v0.AuxInt = 15
+		v.AddArg(v0)
+		return true
+	}
 	// match: (Rsh16Ux64 (Lsh16x64 (Rsh16Ux64 x (Const64 [c1])) (Const64 [c2])) (Const64 [c3]))
 	// cond: uint64(c1) >= uint64(c2) && uint64(c3) >= uint64(c2) && !uaddOvf(c1-c2, c3)
 	// result: (Rsh16Ux64 x (Const64 <typ.UInt64> [c1-c2+c3]))
@@ -25449,6 +25475,32 @@ func rewriteValuegeneric_OpRsh32Ux64_0(v *Value) bool {
 		v.AddArg(v0)
 		return true
 	}
+	// match: (Rsh32Ux64 (Rsh32x64 x _) (Const64 <t> [31]))
+	// cond:
+	// result: (Rsh32Ux64 x (Const64 <t> [31]))
+	for {
+		_ = v.Args[1]
+		v_0 := v.Args[0]
+		if v_0.Op != OpRsh32x64 {
+			break
+		}
+		_ = v_0.Args[1]
+		x := v_0.Args[0]
+		v_1 := v.Args[1]
+		if v_1.Op != OpConst64 {
+			break
+		}
+		t := v_1.Type
+		if v_1.AuxInt != 31 {
+			break
+		}
+		v.reset(OpRsh32Ux64)
+		v.AddArg(x)
+		v0 := b.NewValue0(v.Pos, OpConst64, t)
+		v0.AuxInt = 31
+		v.AddArg(v0)
+		return true
+	}
 	// match: (Rsh32Ux64 (Lsh32x64 (Rsh32Ux64 x (Const64 [c1])) (Const64 [c2])) (Const64 [c3]))
 	// cond: uint64(c1) >= uint64(c2) && uint64(c3) >= uint64(c2) && !uaddOvf(c1-c2, c3)
 	// result: (Rsh32Ux64 x (Const64 <typ.UInt64> [c1-c2+c3]))
@@ -26055,6 +26107,32 @@ func rewriteValuegeneric_OpRsh64Ux64_0(v *Value) bool {
 		v.AddArg(v0)
 		return true
 	}
+	// match: (Rsh64Ux64 (Rsh64x64 x _) (Const64 <t> [63]))
+	// cond:
+	// result: (Rsh64Ux64 x (Const64 <t> [63]))
+	for {
+		_ = v.Args[1]
+		v_0 := v.Args[0]
+		if v_0.Op != OpRsh64x64 {
+			break
+		}
+		_ = v_0.Args[1]
+		x := v_0.Args[0]
+		v_1 := v.Args[1]
+		if v_1.Op != OpConst64 {
+			break
+		}
+		t := v_1.Type
+		if v_1.AuxInt != 63 {
+			break
+		}
+		v.reset(OpRsh64Ux64)
+		v.AddArg(x)
+		v0 := b.NewValue0(v.Pos, OpConst64, t)
+		v0.AuxInt = 63
+		v.AddArg(v0)
+		return true
+	}
 	// match: (Rsh64Ux64 (Lsh64x64 (Rsh64Ux64 x (Const64 [c1])) (Const64 [c2])) (Const64 [c3]))
 	// cond: uint64(c1) >= uint64(c2) && uint64(c3) >= uint64(c2) && !uaddOvf(c1-c2, c3)
 	// result: (Rsh64Ux64 x (Const64 <typ.UInt64> [c1-c2+c3]))
@@ -26723,6 +26801,32 @@ func rewriteValuegeneric_OpRsh8Ux64_0(v *Value) bool {
 		v.AddArg(v0)
 		return true
 	}
+	// match: (Rsh8Ux64 (Rsh8x64 x _) (Const64 <t> [7]))
+	// cond:
+	// result: (Rsh8Ux64 x (Const64 <t> [7] ))
+	for {
+		_ = v.Args[1]
+		v_0 := v.Args[0]
+		if v_0.Op != OpRsh8x64 {
+			break
+		}
+		_ = v_0.Args[1]
+		x := v_0.Args[0]
+		v_1 := v.Args[1]
+		if v_1.Op != OpConst64 {
+			break
+		}
+		t := v_1.Type
+		if v_1.AuxInt != 7 {
+			break
+		}
+		v.reset(OpRsh8Ux64)
+		v.AddArg(x)
+		v0 := b.NewValue0(v.Pos, OpConst64, t)
+		v0.AuxInt = 7
+		v.AddArg(v0)
+		return true
+	}
 	// match: (Rsh8Ux64 (Lsh8x64 (Rsh8Ux64 x (Const64 [c1])) (Const64 [c2])) (Const64 [c3]))
 	// cond: uint64(c1) >= uint64(c2) && uint64(c3) >= uint64(c2) && !uaddOvf(c1-c2, c3)
 	// result: (Rsh8Ux64 x (Const64 <typ.UInt64> [c1-c2+c3]))