1 // SPDX-License-Identifier: GPL-2.0
3 * SHA1 routine optimized to do word accesses rather than byte accesses,
4 * and to avoid unnecessary copies into the context array.
6 * This was based on the git SHA1 implementation.
9 #include <linux/kernel.h>
10 #include <linux/export.h>
11 #include <linux/module.h>
12 #include <linux/bitops.h>
13 #include <linux/string.h>
14 #include <crypto/sha1.h>
15 #include <asm/unaligned.h>
18 * If you have 32 registers or more, the compiler can (and should)
19 * try to change the array[] accesses into registers. However, on
20 * machines with less than ~25 registers, that won't really work,
21 * and at least gcc will make an unholy mess of it.
23 * So to avoid that mess which just slows things down, we force
24 * the stores to memory to actually happen (we might be better off
25 * with a 'W(t)=(val);asm("":"+m" (W(t))' there instead, as
26 * suggested by Artur Skawina - that will also make gcc unable to
27 * try to do the silly "optimize away loads" part because it won't
28 * see what the value will be).
30 * Ben Herrenschmidt reports that on PPC, the C version comes close
31 * to the optimized asm with this (ie on PPC you don't want that
32 * 'volatile', since there are lots of registers).
34 * On ARM we get the best code generation by forcing a full memory barrier
35 * between each SHA_ROUND, otherwise gcc happily get wild with spilling and
36 * the stack frame size simply explode and performance goes down the drain.
40 #define setW(x, val) (*(volatile __u32 *)&W(x) = (val))
41 #elif defined(CONFIG_ARM)
42 #define setW(x, val) do { W(x) = (val); __asm__("":::"memory"); } while (0)
44 #define setW(x, val) (W(x) = (val))
47 /* This "rolls" over the 512-bit array */
48 #define W(x) (array[(x)&15])
51 * Where do we get the source from? The first 16 iterations get it from
52 * the input data, the next mix it from the 512-bit array.
54 #define SHA_SRC(t) get_unaligned_be32((__u32 *)data + t)
55 #define SHA_MIX(t) rol32(W(t+13) ^ W(t+8) ^ W(t+2) ^ W(t), 1)
57 #define SHA_ROUND(t, input, fn, constant, A, B, C, D, E) do { \
58 __u32 TEMP = input(t); setW(t, TEMP); \
59 E += TEMP + rol32(A,5) + (fn) + (constant); \
61 TEMP = E; E = D; D = C; C = B; B = A; A = TEMP; } while (0)
63 #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 )
64 #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 )
65 #define T_20_39(t, A, B, C, D, E) SHA_ROUND(t, SHA_MIX, (B^C^D) , 0x6ed9eba1, A, B, C, D, E )
66 #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 )
67 #define T_60_79(t, A, B, C, D, E) SHA_ROUND(t, SHA_MIX, (B^C^D) , 0xca62c1d6, A, B, C, D, E )
70 * sha1_transform - single block SHA1 transform (deprecated)
72 * @digest: 160 bit digest to update
73 * @data: 512 bits of data to hash
74 * @array: 16 words of workspace (see note)
76 * This function executes SHA-1's internal compression function. It updates the
77 * 160-bit internal state (@digest) with a single 512-bit data block (@data).
79 * Don't use this function. SHA-1 is no longer considered secure. And even if
80 * you do have to use SHA-1, this isn't the correct way to hash something with
81 * SHA-1 as this doesn't handle padding and finalization.
83 * Note: If the hash is security sensitive, the caller should be sure
84 * to clear the workspace. This is left to the caller to avoid
85 * unnecessary clears between chained hashing operations.
87 void sha1_transform(__u32 *digest, const char *data, __u32 *array)
98 /* Round 1 - iterations 0-16 take their input from 'data' */
100 T_0_15(i, A, B, C, D, E);
102 /* Round 1 - tail. Input from 512-bit mixing array */
104 T_16_19(i, A, B, C, D, E);
108 T_20_39(i, A, B, C, D, E);
112 T_40_59(i, A, B, C, D, E);
116 T_60_79(i, A, B, C, D, E);
124 EXPORT_SYMBOL(sha1_transform);
127 * sha1_init - initialize the vectors for a SHA1 digest
128 * @buf: vector to initialize
130 void sha1_init(__u32 *buf)
138 EXPORT_SYMBOL(sha1_init);
140 MODULE_LICENSE("GPL");