cachepc-linux

Fork of AMDESE/linux with modifications for CachePC side-channel attack
git clone https://git.sinitax.com/sinitax/cachepc-linux
Log | Files | Refs | README | LICENSE | sfeed.txt

sha1.c (4468B)

      1// SPDX-License-Identifier: GPL-2.0
      2/*
      3 * SHA1 routine optimized to do word accesses rather than byte accesses,
      4 * and to avoid unnecessary copies into the context array.
      5 *
      6 * This was based on the git SHA1 implementation.
      7 */
      8
      9#include <linux/kernel.h>
     10#include <linux/export.h>
     11#include <linux/bitops.h>
     12#include <linux/string.h>
     13#include <crypto/sha1.h>
     14#include <asm/unaligned.h>
     15
     16/*
     17 * If you have 32 registers or more, the compiler can (and should)
     18 * try to change the array[] accesses into registers. However, on
     19 * machines with less than ~25 registers, that won't really work,
     20 * and at least gcc will make an unholy mess of it.
     21 *
     22 * So to avoid that mess which just slows things down, we force
     23 * the stores to memory to actually happen (we might be better off
     24 * with a 'W(t)=(val);asm("":"+m" (W(t))' there instead, as
     25 * suggested by Artur Skawina - that will also make gcc unable to
     26 * try to do the silly "optimize away loads" part because it won't
     27 * see what the value will be).
     28 *
     29 * Ben Herrenschmidt reports that on PPC, the C version comes close
     30 * to the optimized asm with this (ie on PPC you don't want that
     31 * 'volatile', since there are lots of registers).
     32 *
     33 * On ARM we get the best code generation by forcing a full memory barrier
     34 * between each SHA_ROUND, otherwise gcc happily get wild with spilling and
     35 * the stack frame size simply explode and performance goes down the drain.
     36 */
     37
     38#ifdef CONFIG_X86
     39  #define setW(x, val) (*(volatile __u32 *)&W(x) = (val))
     40#elif defined(CONFIG_ARM)
     41  #define setW(x, val) do { W(x) = (val); __asm__("":::"memory"); } while (0)
     42#else
     43  #define setW(x, val) (W(x) = (val))
     44#endif
     45
     46/* This "rolls" over the 512-bit array */
     47#define W(x) (array[(x)&15])
     48
     49/*
     50 * Where do we get the source from? The first 16 iterations get it from
     51 * the input data, the next mix it from the 512-bit array.
     52 */
     53#define SHA_SRC(t) get_unaligned_be32((__u32 *)data + t)
     54#define SHA_MIX(t) rol32(W(t+13) ^ W(t+8) ^ W(t+2) ^ W(t), 1)
     55
     56#define SHA_ROUND(t, input, fn, constant, A, B, C, D, E) do { \
     57	__u32 TEMP = input(t); setW(t, TEMP); \
     58	E += TEMP + rol32(A,5) + (fn) + (constant); \
     59	B = ror32(B, 2); \
     60	TEMP = E; E = D; D = C; C = B; B = A; A = TEMP; } while (0)
     61
     62#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 )
     63#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 )
     64#define T_20_39(t, A, B, C, D, E) SHA_ROUND(t, SHA_MIX, (B^C^D) , 0x6ed9eba1, A, B, C, D, E )
     65#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 )
     66#define T_60_79(t, A, B, C, D, E) SHA_ROUND(t, SHA_MIX, (B^C^D) ,  0xca62c1d6, A, B, C, D, E )
     67
     68/**
     69 * sha1_transform - single block SHA1 transform (deprecated)
     70 *
     71 * @digest: 160 bit digest to update
     72 * @data:   512 bits of data to hash
     73 * @array:  16 words of workspace (see note)
     74 *
     75 * This function executes SHA-1's internal compression function.  It updates the
     76 * 160-bit internal state (@digest) with a single 512-bit data block (@data).
     77 *
     78 * Don't use this function.  SHA-1 is no longer considered secure.  And even if
     79 * you do have to use SHA-1, this isn't the correct way to hash something with
     80 * SHA-1 as this doesn't handle padding and finalization.
     81 *
     82 * Note: If the hash is security sensitive, the caller should be sure
     83 * to clear the workspace. This is left to the caller to avoid
     84 * unnecessary clears between chained hashing operations.
     85 */
     86void sha1_transform(__u32 *digest, const char *data, __u32 *array)
     87{
     88	__u32 A, B, C, D, E;
     89	unsigned int i = 0;
     90
     91	A = digest[0];
     92	B = digest[1];
     93	C = digest[2];
     94	D = digest[3];
     95	E = digest[4];
     96
     97	/* Round 1 - iterations 0-16 take their input from 'data' */
     98	for (; i < 16; ++i)
     99		T_0_15(i, A, B, C, D, E);
    100
    101	/* Round 1 - tail. Input from 512-bit mixing array */
    102	for (; i < 20; ++i)
    103		T_16_19(i, A, B, C, D, E);
    104
    105	/* Round 2 */
    106	for (; i < 40; ++i)
    107		T_20_39(i, A, B, C, D, E);
    108
    109	/* Round 3 */
    110	for (; i < 60; ++i)
    111		T_40_59(i, A, B, C, D, E);
    112
    113	/* Round 4 */
    114	for (; i < 80; ++i)
    115		T_60_79(i, A, B, C, D, E);
    116
    117	digest[0] += A;
    118	digest[1] += B;
    119	digest[2] += C;
    120	digest[3] += D;
    121	digest[4] += E;
    122}
    123EXPORT_SYMBOL(sha1_transform);
    124
    125/**
    126 * sha1_init - initialize the vectors for a SHA1 digest
    127 * @buf: vector to initialize
    128 */
    129void sha1_init(__u32 *buf)
    130{
    131	buf[0] = 0x67452301;
    132	buf[1] = 0xefcdab89;
    133	buf[2] = 0x98badcfe;
    134	buf[3] = 0x10325476;
    135	buf[4] = 0xc3d2e1f0;
    136}
    137EXPORT_SYMBOL(sha1_init);