crypto/subtle: simplify and speed up constant-time primitives

This changes improves the ConstantTimeByteEq and ConstantTimeEq
primitives to both simplify them and improve their performance.

Also, since there were no benchmarks for this package before,
this change adds benchmarks for ConstantTimeByteEq,
ConstantTimeEq, and ConstantTimeLessOrEq.

benchmarks on darwin/amd64, 10 runs on old vs new code:

name                    old time/op  new time/op  delta
ConstantTimeByteEq-4    2.28ns ±16%  1.53ns ± 2%  -33.09%  (p=0.000 n=10+9)
ConstantTimeEq-4        2.77ns ±10%  1.51ns ± 2%  -45.59%  (p=0.000 n=10+9)
ConstantTimeLessOrEq-4  1.52ns ± 8%  1.50ns ± 2%     ~     (p=0.866 n=9+9)

Change-Id: I29b8cbcf158e1f30411720db82d38b4ecd166b15
Reviewed-on: https://go-review.googlesource.com/45310
Reviewed-by: Adam Langley <agl@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Run-TryBot: Adam Langley <agl@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

src/crypto/subtle/constant_time.go

@@ -29,24 +29,12 @@ func ConstantTimeSelect(v, x, y int) int { return ^(v-1)&x | (v-1)&y }
 
 // ConstantTimeByteEq returns 1 if x == y and 0 otherwise.
 func ConstantTimeByteEq(x, y uint8) int {
-	z := ^(x ^ y)
+	return int((uint32(x^y) - 1) >> 31)
-	z &= z >> 4
-	z &= z >> 2
-	z &= z >> 1
-
-	return int(z)
 }
 
 // ConstantTimeEq returns 1 if x == y and 0 otherwise.
 func ConstantTimeEq(x, y int32) int {
-	z := ^(x ^ y)
+	return int((uint64(uint32(x^y)) - 1) >> 63)
-	z &= z >> 16
-	z &= z >> 8
-	z &= z >> 4
-	z &= z >> 2
-	z &= z >> 1
-
-	return int(z & 1)
 }
 
 // ConstantTimeCopy copies the contents of y into x (a slice of equal length)

src/crypto/subtle/constant_time_test.go

@@ -125,3 +125,35 @@ func TestConstantTimeLessOrEq(t *testing.T) {
 		}
 	}
 }
+
+var benchmarkGlobal uint8
+
+func BenchmarkConstantTimeByteEq(b *testing.B) {
+	var x, y uint8
+
+	for i := 0; i < b.N; i++ {
+		x, y = uint8(ConstantTimeByteEq(x, y)), x
+	}
+
+	benchmarkGlobal = x
+}
+
+func BenchmarkConstantTimeEq(b *testing.B) {
+	var x, y int
+
+	for i := 0; i < b.N; i++ {
+		x, y = ConstantTimeEq(int32(x), int32(y)), x
+	}
+
+	benchmarkGlobal = uint8(x)
+}
+
+func BenchmarkConstantTimeLessOrEq(b *testing.B) {
+	var x, y int
+
+	for i := 0; i < b.N; i++ {
+		x, y = ConstantTimeLessOrEq(x, y), x
+	}
+
+	benchmarkGlobal = uint8(x)
+}