lib/sha1: use the git implementation of SHA-1

For ChromiumOS, we use SHA-1 to verify the integrity of the root
filesystem.  The speed of the kernel sha-1 implementation has a major
impact on our boot performance.

To improve boot performance, we investigated using the heavily optimized
sha-1 implementation used in git.  With the git sha-1 implementation, we
see a 11.7% improvement in boot time.

10 reboots, remove slowest/fastest.

Before:

  Mean: 6.58 seconds Stdev: 0.14

After (with git sha-1, this patch):

  Mean: 5.89 seconds Stdev: 0.07

The other cool thing about the git SHA-1 implementation is that it only
needs 64 bytes of stack for the workspace while the original kernel
implementation needed 320 bytes.
Signed-off-by: Mandeep Singh Baines <msb@chromium.org>
Cc: Ramsay Jones <ramsay@ramsay1.demon.co.uk>
Cc: Nicolas Pitre <nico@cam.org>
Cc: Herbert Xu <herbert@gondor.apana.org.au>
Cc: David S. Miller <davem@davemloft.net>
Cc: linux-crypto@vger.kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

include/linux/cryptohash.h

@@ -3,7 +3,7 @@
 
 #define SHA_DIGEST_WORDS 5
 #define SHA_MESSAGE_BYTES (512 /*bits*/ / 8)
-#define SHA_WORKSPACE_WORDS 80
+#define SHA_WORKSPACE_WORDS 16
 
 void sha_init(__u32 *buf);
 void sha_transform(__u32 *digest, const char *data, __u32 *W);

lib/sha1.c

 /*
- * SHA transform algorithm, originally taken from code written by
- * Peter Gutmann, and placed in the public domain.
+ * SHA1 routine optimized to do word accesses rather than byte accesses,
+ * and to avoid unnecessary copies into the context array.
+ *
+ * This was based on the git SHA1 implementation.
  */
 
 #include <linux/kernel.h>
 #include <linux/module.h>
-#include <linux/cryptohash.h>
+#include <linux/bitops.h>
+#include <asm/unaligned.h>
 
-/* The SHA f()-functions.  */
+/*
+ * If you have 32 registers or more, the compiler can (and should)
+ * try to change the array[] accesses into registers. However, on
+ * machines with less than ~25 registers, that won't really work,
+ * and at least gcc will make an unholy mess of it.
+ *
+ * So to avoid that mess which just slows things down, we force
+ * the stores to memory to actually happen (we might be better off
+ * with a 'W(t)=(val);asm("":"+m" (W(t))' there instead, as
+ * suggested by Artur Skawina - that will also make gcc unable to
+ * try to do the silly "optimize away loads" part because it won't
+ * see what the value will be).
+ *
+ * Ben Herrenschmidt reports that on PPC, the C version comes close
+ * to the optimized asm with this (ie on PPC you don't want that
+ * 'volatile', since there are lots of registers).
+ *
+ * On ARM we get the best code generation by forcing a full memory barrier
+ * between each SHA_ROUND, otherwise gcc happily get wild with spilling and
+ * the stack frame size simply explode and performance goes down the drain.
+ */
 
-#define f1(x,y,z)   (z ^ (x & (y ^ z)))		/* x ? y : z */
-#define f2(x,y,z)   (x ^ y ^ z)			/* XOR */
-#define f3(x,y,z)   ((x & y) + (z & (x ^ y)))	/* majority */
+#ifdef CONFIG_X86
+  #define setW(x, val) (*(volatile __u32 *)&W(x) = (val))
+#elif defined(CONFIG_ARM)
+  #define setW(x, val) do { W(x) = (val); __asm__("":::"memory"); } while (0)
+#else
+  #define setW(x, val) (W(x) = (val))
+#endif
 
-/* The SHA Mysterious Constants */
+/* This "rolls" over the 512-bit array */
+#define W(x) (array[(x)&15])
 
-#define K1  0x5A827999L			/* Rounds  0-19: sqrt(2) * 2^30 */
-#define K2  0x6ED9EBA1L			/* Rounds 20-39: sqrt(3) * 2^30 */
-#define K3  0x8F1BBCDCL			/* Rounds 40-59: sqrt(5) * 2^30 */
-#define K4  0xCA62C1D6L			/* Rounds 60-79: sqrt(10) * 2^30 */
+/*
+ * Where do we get the source from? The first 16 iterations get it from
+ * the input data, the next mix it from the 512-bit array.
+ */
+#define SHA_SRC(t) get_unaligned_be32((__u32 *)data + t)
+#define SHA_MIX(t) rol32(W(t+13) ^ W(t+8) ^ W(t+2) ^ W(t), 1)
+
+#define SHA_ROUND(t, input, fn, constant, A, B, C, D, E) do { \
+	__u32 TEMP = input(t); setW(t, TEMP); \
+	E += TEMP + rol32(A,5) + (fn) + (constant); \
+	B = ror32(B, 2); } while (0)
+
+#define T_0_15(t, A, B, C, D, E)  SHA_ROUND(t, SHA_SRC, (((C^D)&B)^D) , 0x5a827999, A, B, C, D, E )
+#define T_16_19(t, A, B, C, D, E) SHA_ROUND(t, SHA_MIX, (((C^D)&B)^D) , 0x5a827999, A, B, C, D, E )
+#define T_20_39(t, A, B, C, D, E) SHA_ROUND(t, SHA_MIX, (B^C^D) , 0x6ed9eba1, A, B, C, D, E )
+#define T_40_59(t, A, B, C, D, E) SHA_ROUND(t, SHA_MIX, ((B&C)+(D&(B^C))) , 0x8f1bbcdc, A, B, C, D, E )
+#define T_60_79(t, A, B, C, D, E) SHA_ROUND(t, SHA_MIX, (B^C^D) ,  0xca62c1d6, A, B, C, D, E )
 
 /**
  * sha_transform - single block SHA1 transform
  *
  * @digest: 160 bit digest to update
  * @data:   512 bits of data to hash
- * @W:      80 words of workspace (see note)
+ * @array:  16 words of workspace (see note)
  *
  * This function generates a SHA1 digest for a single 512-bit block.
  * Be warned, it does not handle padding and message digest, do not
@@ -36,47 +77,111 @@
  * to clear the workspace. This is left to the caller to avoid
  * unnecessary clears between chained hashing operations.
  */
-void sha_transform(__u32 *digest, const char *in, __u32 *W)
+void sha_transform(__u32 *digest, const char *data, __u32 *array)
 {
-	__u32 a, b, c, d, e, t, i;
-
-	for (i = 0; i < 16; i++)
-		W[i] = be32_to_cpu(((const __be32 *)in)[i]);
-
-	for (i = 0; i < 64; i++)
-		W[i+16] = rol32(W[i+13] ^ W[i+8] ^ W[i+2] ^ W[i], 1);
-
-	a = digest[0];
-	b = digest[1];
-	c = digest[2];
-	d = digest[3];
-	e = digest[4];
-
-	for (i = 0; i < 20; i++) {
-		t = f1(b, c, d) + K1 + rol32(a, 5) + e + W[i];
-		e = d; d = c; c = rol32(b, 30); b = a; a = t;
-	}
-
-	for (; i < 40; i ++) {
-		t = f2(b, c, d) + K2 + rol32(a, 5) + e + W[i];
-		e = d; d = c; c = rol32(b, 30); b = a; a = t;
-	}
-
-	for (; i < 60; i ++) {
-		t = f3(b, c, d) + K3 + rol32(a, 5) + e + W[i];
-		e = d; d = c; c = rol32(b, 30); b = a; a = t;
-	}
-
-	for (; i < 80; i ++) {
-		t = f2(b, c, d) + K4 + rol32(a, 5) + e + W[i];
-		e = d; d = c; c = rol32(b, 30); b = a; a = t;
-	}
-
-	digest[0] += a;
-	digest[1] += b;
-	digest[2] += c;
-	digest[3] += d;
-	digest[4] += e;
+	__u32 A, B, C, D, E;
+
+	A = digest[0];
+	B = digest[1];
+	C = digest[2];
+	D = digest[3];
+	E = digest[4];
+
+	/* Round 1 - iterations 0-16 take their input from 'data' */
+	T_0_15( 0, A, B, C, D, E);
+	T_0_15( 1, E, A, B, C, D);
+	T_0_15( 2, D, E, A, B, C);
+	T_0_15( 3, C, D, E, A, B);
+	T_0_15( 4, B, C, D, E, A);
+	T_0_15( 5, A, B, C, D, E);
+	T_0_15( 6, E, A, B, C, D);
+	T_0_15( 7, D, E, A, B, C);
+	T_0_15( 8, C, D, E, A, B);
+	T_0_15( 9, B, C, D, E, A);
+	T_0_15(10, A, B, C, D, E);
+	T_0_15(11, E, A, B, C, D);
+	T_0_15(12, D, E, A, B, C);
+	T_0_15(13, C, D, E, A, B);
+	T_0_15(14, B, C, D, E, A);
+	T_0_15(15, A, B, C, D, E);
+
+	/* Round 1 - tail. Input from 512-bit mixing array */
+	T_16_19(16, E, A, B, C, D);
+	T_16_19(17, D, E, A, B, C);
+	T_16_19(18, C, D, E, A, B);
+	T_16_19(19, B, C, D, E, A);
+
+	/* Round 2 */
+	T_20_39(20, A, B, C, D, E);
+	T_20_39(21, E, A, B, C, D);
+	T_20_39(22, D, E, A, B, C);
+	T_20_39(23, C, D, E, A, B);
+	T_20_39(24, B, C, D, E, A);
+	T_20_39(25, A, B, C, D, E);
+	T_20_39(26, E, A, B, C, D);
+	T_20_39(27, D, E, A, B, C);
+	T_20_39(28, C, D, E, A, B);
+	T_20_39(29, B, C, D, E, A);
+	T_20_39(30, A, B, C, D, E);
+	T_20_39(31, E, A, B, C, D);
+	T_20_39(32, D, E, A, B, C);
+	T_20_39(33, C, D, E, A, B);
+	T_20_39(34, B, C, D, E, A);
+	T_20_39(35, A, B, C, D, E);
+	T_20_39(36, E, A, B, C, D);
+	T_20_39(37, D, E, A, B, C);
+	T_20_39(38, C, D, E, A, B);
+	T_20_39(39, B, C, D, E, A);
+
+	/* Round 3 */
+	T_40_59(40, A, B, C, D, E);
+	T_40_59(41, E, A, B, C, D);
+	T_40_59(42, D, E, A, B, C);
+	T_40_59(43, C, D, E, A, B);
+	T_40_59(44, B, C, D, E, A);
+	T_40_59(45, A, B, C, D, E);
+	T_40_59(46, E, A, B, C, D);
+	T_40_59(47, D, E, A, B, C);
+	T_40_59(48, C, D, E, A, B);
+	T_40_59(49, B, C, D, E, A);
+	T_40_59(50, A, B, C, D, E);
+	T_40_59(51, E, A, B, C, D);
+	T_40_59(52, D, E, A, B, C);
+	T_40_59(53, C, D, E, A, B);
+	T_40_59(54, B, C, D, E, A);
+	T_40_59(55, A, B, C, D, E);
+	T_40_59(56, E, A, B, C, D);
+	T_40_59(57, D, E, A, B, C);
+	T_40_59(58, C, D, E, A, B);
+	T_40_59(59, B, C, D, E, A);
+
+	/* Round 4 */
+	T_60_79(60, A, B, C, D, E);
+	T_60_79(61, E, A, B, C, D);
+	T_60_79(62, D, E, A, B, C);
+	T_60_79(63, C, D, E, A, B);
+	T_60_79(64, B, C, D, E, A);
+	T_60_79(65, A, B, C, D, E);
+	T_60_79(66, E, A, B, C, D);
+	T_60_79(67, D, E, A, B, C);
+	T_60_79(68, C, D, E, A, B);
+	T_60_79(69, B, C, D, E, A);
+	T_60_79(70, A, B, C, D, E);
+	T_60_79(71, E, A, B, C, D);
+	T_60_79(72, D, E, A, B, C);
+	T_60_79(73, C, D, E, A, B);
+	T_60_79(74, B, C, D, E, A);
+	T_60_79(75, A, B, C, D, E);
+	T_60_79(76, E, A, B, C, D);
+	T_60_79(77, D, E, A, B, C);
+	T_60_79(78, C, D, E, A, B);
+	T_60_79(79, B, C, D, E, A);
+
+	digest[0] += A;
+	digest[1] += B;
+	digest[2] += C;
+	digest[3] += D;
+	digest[4] += E;
 }
 EXPORT_SYMBOL(sha_transform);
 
@@ -92,4 +197,3 @@ void sha_init(__u32 *buf)
 	buf[3] = 0x10325476;
 	buf[4] = 0xc3d2e1f0;
 }
-