2 * SHA1 routine optimized to do word accesses rather than byte accesses,
3 * and to avoid unnecessary copies into the context array.
5 * This was based on the git SHA1 implementation.
8 #include <linux/kernel.h>
9 #include <linux/export.h>
10 #include <linux/bitops.h>
11 #include <linux/cryptohash.h>
12 #include <linux/string.h>
13 #include <asm/unaligned.h>
16 * If you have 32 registers or more, the compiler can (and should)
17 * try to change the array[] accesses into registers. However, on
18 * machines with less than ~25 registers, that won't really work,
19 * and at least gcc will make an unholy mess of it.
21 * So to avoid that mess which just slows things down, we force
22 * the stores to memory to actually happen (we might be better off
23 * with a 'W(t)=(val);asm("":"+m" (W(t))' there instead, as
24 * suggested by Artur Skawina - that will also make gcc unable to
25 * try to do the silly "optimize away loads" part because it won't
26 * see what the value will be).
28 * Ben Herrenschmidt reports that on PPC, the C version comes close
29 * to the optimized asm with this (ie on PPC you don't want that
30 * 'volatile', since there are lots of registers).
32 * On ARM we get the best code generation by forcing a full memory barrier
33 * between each SHA_ROUND, otherwise gcc happily get wild with spilling and
34 * the stack frame size simply explode and performance goes down the drain.
38 #define setW(x, val) (*(volatile __u32 *)&W(x) = (val))
39 #elif defined(CONFIG_ARM)
40 #define setW(x, val) do { W(x) = (val); __asm__("":::"memory"); } while (0)
42 #define setW(x, val) (W(x) = (val))
45 /* This "rolls" over the 512-bit array */
46 #define W(x) (array[(x)&15])
49 * Where do we get the source from? The first 16 iterations get it from
50 * the input data, the next mix it from the 512-bit array.
52 #define SHA_SRC(t) get_unaligned_be32((__u32 *)data + t)
53 #define SHA_MIX(t) rol32(W(t+13) ^ W(t+8) ^ W(t+2) ^ W(t), 1)
55 #define SHA_ROUND(t, input, fn, constant, A, B, C, D, E) do { \
56 __u32 TEMP = input(t); setW(t, TEMP); \
57 E += TEMP + rol32(A,5) + (fn) + (constant); \
59 TEMP = E; E = D; D = C; C = B; B = A; A = TEMP; } while (0)
61 #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 )
62 #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 )
63 #define T_20_39(t, A, B, C, D, E) SHA_ROUND(t, SHA_MIX, (B^C^D) , 0x6ed9eba1, A, B, C, D, E )
64 #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 )
65 #define T_60_79(t, A, B, C, D, E) SHA_ROUND(t, SHA_MIX, (B^C^D) , 0xca62c1d6, A, B, C, D, E )
68 * sha_transform - single block SHA1 transform
70 * @digest: 160 bit digest to update
71 * @data: 512 bits of data to hash
72 * @array: 16 words of workspace (see note)
74 * This function generates a SHA1 digest for a single 512-bit block.
75 * Be warned, it does not handle padding and message digest, do not
76 * confuse it with the full FIPS 180-1 digest algorithm for variable
79 * Note: If the hash is security sensitive, the caller should be sure
80 * to clear the workspace. This is left to the caller to avoid
81 * unnecessary clears between chained hashing operations.
83 void sha_transform(__u32 *digest, const char *data, __u32 *array)
94 /* Round 1 - iterations 0-16 take their input from 'data' */
96 T_0_15(i, A, B, C, D, E);
98 /* Round 1 - tail. Input from 512-bit mixing array */
100 T_16_19(i, A, B, C, D, E);
104 T_20_39(i, A, B, C, D, E);
108 T_40_59(i, A, B, C, D, E);
112 T_60_79(i, A, B, C, D, E);
120 EXPORT_SYMBOL(sha_transform);
123 * sha_init - initialize the vectors for a SHA1 digest
124 * @buf: vector to initialize
126 void sha_init(__u32 *buf)
134 EXPORT_SYMBOL(sha_init);