From 61f6db130e9c2b524d99a34d623c6d05732c6c21 Mon Sep 17 00:00:00 2001 From: Eric Andersen Date: Thu, 3 Jul 2003 09:41:41 +0000 Subject: [PATCH] As implemented, sha1sum would sometimes give the wrong answer. This fixes it and uses faster sha1 code from Dr. Gladman. -Erik --- coreutils/sha1sum.c | 401 ++++++++++++++++++++++++++++++++------------ 1 file changed, 295 insertions(+), 106 deletions(-) diff --git a/coreutils/sha1sum.c b/coreutils/sha1sum.c index 7071fb6b7..1148aac19 100644 --- a/coreutils/sha1sum.c +++ b/coreutils/sha1sum.c @@ -1,13 +1,10 @@ /* * Based on shasum from http://www.netsw.org/crypto/hash/ - * - * shasum fixed with reference to coreutils and the nist fip180-1 document - * which is incorrect, in section 5 - * - ft(B,C,D) = (B AND C) OR ((NOT B) AND D) ( 0 <= t <= 19) - * + ft(B,C,D) = (D XOR (B AND (C XOR D))) ( 0 <= t <= 19) + * Majorly hacked up to use Dr Brian Gladman's sha1 code * * Copyright (C) 1999 Scott G. Miller * Copyright (C) 2003 Glenn L. McGrath + * Copyright (C) 2003 Erik Andersen * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -29,137 +26,325 @@ #include #include #include +#include +#include #include +#include #include "busybox.h" + +/* + --------------------------------------------------------------------------- + Begin Dr. Gladman's sha1 code + --------------------------------------------------------------------------- +*/ + +/* + --------------------------------------------------------------------------- + Copyright (c) 2002, Dr Brian Gladman , Worcester, UK. + All rights reserved. + + LICENSE TERMS + + The free distribution and use of this software in both source and binary + form is allowed (with or without changes) provided that: + + 1. distributions of this source code include the above copyright + notice, this list of conditions and the following disclaimer; + + 2. distributions in binary form include the above copyright + notice, this list of conditions and the following disclaimer + in the documentation and/or other associated materials; + + 3. the copyright holder's name is not used to endorse products + built using this software without specific written permission. + + ALTERNATIVELY, provided that this notice is retained in full, this product + may be distributed under the terms of the GNU General Public License (GPL), + in which case the provisions of the GPL apply INSTEAD OF those given above. + + DISCLAIMER + + This software is provided 'as is' with no explicit or implied warranties + in respect of its properties, including, but not limited to, correctness + and/or fitness for purpose. + --------------------------------------------------------------------------- + Issue Date: 10/11/2002 + + This is a byte oriented version of SHA1 that operates on arrays of bytes + stored in memory. It runs at 22 cycles per byte on a Pentium P4 processor +*/ + +#define SHA1_BLOCK_SIZE 64 +#define SHA1_DIGEST_SIZE 20 +#define SHA1_HASH_SIZE SHA1_DIGEST_SIZE +#define SHA2_GOOD 0 +#define SHA2_BAD 1 + +/* type to hold the SHA1 context */ +typedef struct +{ uint32_t count[2]; + uint32_t hash[5]; + uint32_t wbuf[16]; +} sha1_ctx; + +#define rotl32(x,n) (((x) << n) | ((x) >> (32 - n))) + #if __BYTE_ORDER == __BIG_ENDIAN -# define SWAP(n) (n) +# define swap_b32(x) (x) +#elif defined(bswap_32) +# define swap_b32(x) bswap_32(x) #else -# define SWAP(n) \ - (((n) << 24) | (((n) & 0xff00) << 8) | (((n) >> 8) & 0xff00) | ((n) >> 24)) +# define swap_b32(x) ((rotl32((x), 8) & 0x00ff00ff) | (rotl32((x), 24) & 0xff00ff00)) #endif -#define f1(X,Y,Z) (Z ^ (X & (Y ^ Z))) -#define f2(X,Y,Z) (X ^ Y ^ Z) -#define f3(X,Y,Z) ((X & Y) | (Z & (X | Y))) +#define SHA1_MASK (SHA1_BLOCK_SIZE - 1) -#define rol1(x) (x<<1) | ((x>>31) & 1) -#define rol5(x) ((x<<5) | ((x>>27) & 0x1f)) -#define rol30(x) (x<<30) | ((x>>2) & 0x3fffffff) +/* reverse byte order in 32-bit words */ +#define ch(x,y,z) (((x) & (y)) ^ (~(x) & (z))) +#define parity(x,y,z) ((x) ^ (y) ^ (z)) +#define maj(x,y,z) (((x) & (y)) ^ ((x) & (z)) ^ ((y) & (z))) -static void sha_hash(unsigned int *data, int *hash) +/* A normal version as set out in the FIPS. This version uses */ +/* partial loop unrolling and is optimised for the Pentium 4 */ +#define rnd(f,k) \ + t = a; a = rotl32(a,5) + f(b,c,d) + e + k + w[i]; \ + e = d; d = c; c = rotl32(b, 30); b = t + +void sha1_compile(sha1_ctx ctx[1]) { - RESERVE_CONFIG_BUFFER(word, 80 * sizeof(unsigned int)); - int *W = (unsigned int *) &word; - int a = hash[0]; - int b = hash[1]; - int c = hash[2]; - int d = hash[3]; - int e = hash[4]; - int t; - int TEMP; + uint32_t w[80], i, a, b, c, d, e, t; - for (t = 0; t < 16; t++) { - W[t] = SWAP(data[t]); - } + /* note that words are compiled from the buffer into 32-bit */ + /* words in big-endian order so an order reversal is needed */ + /* here on little endian machines */ + for(i = 0; i < SHA1_BLOCK_SIZE / 4; ++i) + w[i] = swap_b32(ctx->wbuf[i]); - /** Data expansion from 16 to 80 blocks **/ - for (t = 16; t < 80; t++) { - int x = W[t - 3] ^ W[t - 8] ^ W[t - 14] ^ W[t - 16]; - W[t] = rol1(x); - } + for(i = SHA1_BLOCK_SIZE / 4; i < 80; ++i) + w[i] = rotl32(w[i - 3] ^ w[i - 8] ^ w[i - 14] ^ w[i - 16], 1); - /** Main loops **/ - for (t = 0; t < 20; t++) { - TEMP = rol5(a) + f1(b, c, d) + e + W[t] + 0x5a827999; - e = d; - d = c; - c = rol30(b); - b = a; - a = TEMP; - } - for (; t < 40; t++) { - TEMP = rol5(a) + f2(b, c, d) + e + W[t] + 0x6ed9eba1; - e = d; - d = c; - c = rol30(b); - b = a; - a = TEMP; - } - for (; t < 60; t++) { - TEMP = rol5(a) + f3(b, c, d) + e + W[t] + 0x8f1bbcdc; - e = d; - d = c; - c = rol30(b); - b = a; - a = TEMP; - } - for (; t < 80; t++) { - TEMP = rol5(a) + f2(b, c, d) + e + W[t] + 0xca62c1d6; - e = d; - d = c; - c = rol30(b); - b = a; - a = TEMP; - } + a = ctx->hash[0]; + b = ctx->hash[1]; + c = ctx->hash[2]; + d = ctx->hash[3]; + e = ctx->hash[4]; - RELEASE_CONFIG_BUFFER(word); + for(i = 0; i < 20; ++i) + { + rnd(ch, 0x5a827999); + } - hash[0] += a; - hash[1] += b; - hash[2] += c; - hash[3] += d; - hash[4] += e; + for(i = 20; i < 40; ++i) + { + rnd(parity, 0x6ed9eba1); + } + + for(i = 40; i < 60; ++i) + { + rnd(maj, 0x8f1bbcdc); + } + + for(i = 60; i < 80; ++i) + { + rnd(parity, 0xca62c1d6); + } + + ctx->hash[0] += a; + ctx->hash[1] += b; + ctx->hash[2] += c; + ctx->hash[3] += d; + ctx->hash[4] += e; } -static char sha1sum_stream(FILE *fd, unsigned int *hashval) +void sha1_begin(sha1_ctx ctx[1]) { - RESERVE_CONFIG_BUFFER(buffer, 64); - int length = 0; + ctx->count[0] = ctx->count[1] = 0; + ctx->hash[0] = 0x67452301; + ctx->hash[1] = 0xefcdab89; + ctx->hash[2] = 0x98badcfe; + ctx->hash[3] = 0x10325476; + ctx->hash[4] = 0xc3d2e1f0; +} - hashval[0] = 0x67452301; - hashval[1] = 0xefcdab89; - hashval[2] = 0x98badcfe; - hashval[3] = 0x10325476; - hashval[4] = 0xc3d2e1f0; +/* SHA1 hash data in an array of bytes into hash buffer and call the */ +/* hash_compile function as required. */ +void sha1_hash(const unsigned char data[], unsigned int len, sha1_ctx ctx[1]) +{ + uint32_t pos = (uint32_t)(ctx->count[0] & SHA1_MASK), + freeb = SHA1_BLOCK_SIZE - pos; + const unsigned char *sp = data; - while (!feof(fd) && !ferror(fd)) { - int c = fread(&buffer, 1, 64, fd); - length += c; - if (feof(fd) || ferror(fd)) { - int i; - for (i = c; i < 61; i++) { - if (i == c) { - buffer[i] = 0x80; - } - else if (i == 60) { - /* This ends up being swaped twice */ - ((unsigned int *) &buffer)[15] = SWAP(length * 8); - } else { - buffer[i] = 0; - } - } + if((ctx->count[0] += len) < len) + ++(ctx->count[1]); + + while(len >= freeb) /* tranfer whole blocks while possible */ + { + memcpy(((unsigned char*)ctx->wbuf) + pos, sp, freeb); + sp += freeb; len -= freeb; freeb = SHA1_BLOCK_SIZE; pos = 0; + sha1_compile(ctx); + } + + memcpy(((unsigned char*)ctx->wbuf) + pos, sp, len); +} + +/* SHA1 Final padding and digest calculation */ +#if __BYTE_ORDER == __LITTLE_ENDIAN +static uint32_t mask[4] = + { 0x00000000, 0x000000ff, 0x0000ffff, 0x00ffffff }; +static uint32_t bits[4] = + { 0x00000080, 0x00008000, 0x00800000, 0x80000000 }; +#else +static uint32_t mask[4] = + { 0x00000000, 0xff000000, 0xffff0000, 0xffffff00 }; +static uint32_t bits[4] = + { 0x80000000, 0x00800000, 0x00008000, 0x00000080 }; +#endif + +void sha1_end(unsigned char hval[], sha1_ctx ctx[1]) +{ + uint32_t i, cnt = (uint32_t)(ctx->count[0] & SHA1_MASK); + + /* mask out the rest of any partial 32-bit word and then set */ + /* the next byte to 0x80. On big-endian machines any bytes in */ + /* the buffer will be at the top end of 32 bit words, on little */ + /* endian machines they will be at the bottom. Hence the AND */ + /* and OR masks above are reversed for little endian systems */ + ctx->wbuf[cnt >> 2] = (ctx->wbuf[cnt >> 2] & mask[cnt & 3]) | bits[cnt & 3]; + + /* we need 9 or more empty positions, one for the padding byte */ + /* (above) and eight for the length count. If there is not */ + /* enough space pad and empty the buffer */ + if(cnt > SHA1_BLOCK_SIZE - 9) + { + if(cnt < 60) ctx->wbuf[15] = 0; + sha1_compile(ctx); + cnt = 0; + } + else /* compute a word index for the empty buffer positions */ + cnt = (cnt >> 2) + 1; + + while(cnt < 14) /* and zero pad all but last two positions */ + ctx->wbuf[cnt++] = 0; + + /* assemble the eight byte counter in the buffer in big-endian */ + /* format */ + + ctx->wbuf[14] = swap_b32((ctx->count[1] << 3) | (ctx->count[0] >> 29)); + ctx->wbuf[15] = swap_b32(ctx->count[0] << 3); + + sha1_compile(ctx); + + /* extract the hash value as bytes in case the hash buffer is */ + /* misaligned for 32-bit words */ + + for(i = 0; i < SHA1_DIGEST_SIZE; ++i) + hval[i] = (unsigned char)(ctx->hash[i >> 2] >> 8 * (~i & 3)); +} + +#if 0 +void sha1(unsigned char hval[], const unsigned char data[], unsigned int len) +{ sha1_ctx cx[1]; + + sha1_begin(cx); sha1_hash(data, len, cx); sha1_end(hval, cx); +} +#endif + +/* + --------------------------------------------------------------------------- + End of Dr. Gladman's sha1 code + --------------------------------------------------------------------------- +*/ + +/* Using a larger blocksize can make things _much_ faster + * by avoiding a zillion tiny little reads */ +#define BLOCKSIZE 65536 +/* Ensure that BLOCKSIZE is a multiple of 64. */ +#if BLOCKSIZE % SHA1_BLOCK_SIZE != 0 +# error "BLOCKSIZE not a multiple of 64" +#endif + +static int sha1sum_stream(FILE *stream, unsigned char *hashval) +{ + int result = 0; + sha1_ctx cx[1]; + size_t sum, n; + RESERVE_CONFIG_BUFFER(buffer, BLOCKSIZE + 72); + + /* Initialize the computation context. */ + sha1_begin(cx); + + /* Iterate over full file contents. */ + while (1) + { + /* We read the file in blocks of BLOCKSIZE bytes. One call of the + computation function processes the whole buffer so that with the + next round of the loop another block can be read. */ + sum = 0; + + /* Read block. Take care for partial reads. */ + while (1) + { + n = fread (buffer + sum, 1, BLOCKSIZE - sum, stream); + sum += n; + + if (sum == BLOCKSIZE) + break; + + if (n == 0) { + /* Check for the error flag IFF N == 0, so that we don't + exit the loop after a partial read due to e.g., EAGAIN + or EWOULDBLOCK. */ + if (feof (stream)) { + sum = 0; + goto process_partial_block; } - sha_hash((unsigned int *) &buffer, hashval); + if (ferror (stream)) { + result++; + goto all_done; + } + goto process_partial_block; + } + + /* We've read at least one byte, so ignore errors. But always + check for EOF, since feof may be true even though N > 0. + Otherwise, we could end up calling fread after EOF. */ + if (feof (stream)) + goto process_partial_block; } - RELEASE_CONFIG_BUFFER(buffer); + /* Process buffer */ + sha1_hash(buffer, BLOCKSIZE, cx); + } - return(EXIT_SUCCESS); +process_partial_block: + + /* Process any remaining bytes. */ + if (sum > 0) + sha1_hash(buffer, sum, cx); + + /* Finalize and write the hash into our buffer. */ + sha1_end(hashval, cx); + +all_done: + + RELEASE_CONFIG_BUFFER(buffer); + return result; } #define FLAG_SILENT 1 #define FLAG_CHECK 2 #define FLAG_WARN 4 -static unsigned char *hash_bin_to_hex(unsigned int *hash_value, unsigned char hash_length) +static unsigned char *hash_bin_to_hex(unsigned char *hash_value, unsigned char hash_length) { - unsigned char x; + int x, len, max; unsigned char *hex_value; - hex_value = xmalloc(hash_length * 8); - for (x = 0; x < hash_length; x++) { - sprintf(&hex_value[x * 8], "%08x", hash_value[x]); + max = (hash_length * 2) + 2; + hex_value = xmalloc(max); + for (x = len = 0; x < hash_length; x++) { + len += snprintf(hex_value+len, max-len, "%02x", hash_value[x]); } return(hex_value); } @@ -178,9 +363,11 @@ FILE *wfopen_file_or_stdin(const char *file_ptr) } /* This could become a common function for md5 as well, by using md5_stream */ -extern int authenticate(int argc, char **argv, char (*hash_ptr)(FILE *stream, unsigned int *hashval), const unsigned char hash_length) +extern int authenticate(int argc, char **argv, + int (*hash_ptr)(FILE *stream, unsigned char *hashval), + const unsigned char hash_length) { - unsigned int hash_value[hash_length]; + unsigned char hash_value[hash_length]; unsigned int flags; int return_value = EXIT_SUCCESS; @@ -245,6 +432,7 @@ extern int authenticate(int argc, char **argv, char (*hash_ptr)(FILE *stream, un line_ptr++; stream = bb_wfopen(line_ptr, "r"); if (hash_ptr(stream, hash_value) == EXIT_FAILURE) { + bb_perror_msg("%s", file_ptr); return_value = EXIT_FAILURE; } if (fclose(stream) == EOF) { @@ -280,6 +468,7 @@ extern int authenticate(int argc, char **argv, char (*hash_ptr)(FILE *stream, un continue; } if (hash_ptr(stream, hash_value) == EXIT_FAILURE) { + bb_perror_msg("%s", file_ptr); return_value = EXIT_FAILURE; } else if (!flags & FLAG_SILENT) { @@ -299,5 +488,5 @@ extern int authenticate(int argc, char **argv, char (*hash_ptr)(FILE *stream, un extern int sha1sum_main(int argc, char **argv) { - return (authenticate(argc, argv, sha1sum_stream, 5)); + return (authenticate(argc, argv, sha1sum_stream, SHA1_HASH_SIZE)); }