crypto: sha512 - use standard ror64()

Use standard ror64() instead of hand-written. There is no standard ror64, so create it. The difference is shift value being "unsigned int" instead of uint64_t (for which there is no reason). gcc starts to emit native ROR instructions which it doesn't do for some reason currently. This should make the code faster. Patch survives in-tree crypto test and ping flood with hmac(sha512) on. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

crypto: sha512 - use standard ror64()
Use standard ror64() instead of hand-written. There is no standard ror64, so create it. The difference is shift value being "unsigned int" instead of uint64_t (for which there is no reason). gcc starts to emit native ROR instructions which it doesn't do for some reason currently. This should make the code faster. Patch survives in-tree crypto test and ping flood with hmac(sha512) on. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Alexey Dobriyan · Herbert Xu
1 parent 3a92d687c8
Showing 2 changed files with 24 additions and 9 deletions Side-by-side Diff
crypto/sha512_generic.c
include/linux/bitops.h
@@ -31,11 +31,6 @@
         return (x & y) | (z & (x | y));
 }
  
-static inline u64 RORu64(u64 x, u64 y)
-{
-        return (x >> y) | (x << (64 - y));
-}
-
 static const u64 sha512_K[80] = {
         0x428a2f98d728ae22ULL, 0x7137449123ef65cdULL, 0xb5c0fbcfec4d3b2fULL,
         0xe9b5dba58189dbbcULL, 0x3956c25bf348b538ULL, 0x59f111f1b605d019ULL,
@@ -66,10 +61,10 @@
         0x5fcb6fab3ad6faecULL, 0x6c44198c4a475817ULL,
 };
  
-#define e0(x)       (RORu64(x,28) ^ RORu64(x,34) ^ RORu64(x,39))
-#define e1(x)       (RORu64(x,14) ^ RORu64(x,18) ^ RORu64(x,41))
-#define s0(x)       (RORu64(x, 1) ^ RORu64(x, 8) ^ (x >> 7))
-#define s1(x)       (RORu64(x,19) ^ RORu64(x,61) ^ (x >> 6))
+#define e0(x)       (ror64(x,28) ^ ror64(x,34) ^ ror64(x,39))
+#define e1(x)       (ror64(x,14) ^ ror64(x,18) ^ ror64(x,41))
+#define s0(x)       (ror64(x, 1) ^ ror64(x, 8) ^ (x >> 7))
+#define s1(x)       (ror64(x,19) ^ ror64(x,61) ^ (x >> 6))
  
 static inline void LOAD_OP(int I, u64 *W, const u8 *input)
 {
@@ -50,6 +50,26 @@
 }
  
 /**
+ * rol64 - rotate a 64-bit value left
+ * @word: value to rotate
+ * @shift: bits to roll
+ */
+static inline __u64 rol64(__u64 word, unsigned int shift)
+{
+	return (word << shift) | (word >> (64 - shift));
+}
+
+/**
+ * ror64 - rotate a 64-bit value right
+ * @word: value to rotate
+ * @shift: bits to roll
+ */
+static inline __u64 ror64(__u64 word, unsigned int shift)
+{
+	return (word >> shift) | (word << (64 - shift));
+}
+
+/**
  * rol32 - rotate a 32-bit value left
  * @word: value to rotate
  * @shift: bits to roll
...	...	@@ -31,11 +31,6 @@
31	31	return (x & y) \| (z & (x \| y));
32	32	}
33	33
34		-static inline u64 RORu64(u64 x, u64 y)
35		-{
36		- return (x >> y) \| (x << (64 - y));
37		-}
38		-
39	34	static const u64 sha512_K[80] = {
40	35	0x428a2f98d728ae22ULL, 0x7137449123ef65cdULL, 0xb5c0fbcfec4d3b2fULL,
41	36	0xe9b5dba58189dbbcULL, 0x3956c25bf348b538ULL, 0x59f111f1b605d019ULL,
...	...	@@ -66,10 +61,10 @@
66	61	0x5fcb6fab3ad6faecULL, 0x6c44198c4a475817ULL,
67	62	};
68	63
69		-#define e0(x) (RORu64(x,28) ^ RORu64(x,34) ^ RORu64(x,39))
70		-#define e1(x) (RORu64(x,14) ^ RORu64(x,18) ^ RORu64(x,41))
71		-#define s0(x) (RORu64(x, 1) ^ RORu64(x, 8) ^ (x >> 7))
72		-#define s1(x) (RORu64(x,19) ^ RORu64(x,61) ^ (x >> 6))
	64	+#define e0(x) (ror64(x,28) ^ ror64(x,34) ^ ror64(x,39))
	65	+#define e1(x) (ror64(x,14) ^ ror64(x,18) ^ ror64(x,41))
	66	+#define s0(x) (ror64(x, 1) ^ ror64(x, 8) ^ (x >> 7))
	67	+#define s1(x) (ror64(x,19) ^ ror64(x,61) ^ (x >> 6))
73	68
74	69	static inline void LOAD_OP(int I, u64 W, const u8 input)
75	70	{
...	...	@@ -50,6 +50,26 @@
50	50	}
51	51
52	52	/**
	53	+ * rol64 - rotate a 64-bit value left
	54	+ * @word: value to rotate
	55	+ * @shift: bits to roll
	56	+ */
	57	+static inline __u64 rol64(__u64 word, unsigned int shift)
	58	+{
	59	+ return (word << shift) \| (word >> (64 - shift));
	60	+}
	61	+
	62	+/**
	63	+ * ror64 - rotate a 64-bit value right
	64	+ * @word: value to rotate
	65	+ * @shift: bits to roll
	66	+ */
	67	+static inline __u64 ror64(__u64 word, unsigned int shift)
	68	+{
	69	+ return (word >> shift) \| (word << (64 - shift));
	70	+}
	71	+
	72	+/**
53	73	* rol32 - rotate a 32-bit value left
54	74	* @word: value to rotate
55	75	* @shift: bits to roll