*
* SHA256 and SHA512 parts are:
* Released into the Public Domain by Ulrich Drepper <drepper@redhat.com>.
- * TODO: shrink them.
+ * Shrank by Denys Vlasenko.
+ *
+ * ---------------------------------------------------------------------------
+ *
+ * The best way to test random blocksizes is to go to coreutils/md5_sha1_sum.c
+ * and replace "4096" with something like "2000 + time(NULL) % 2097",
+ * then rebuild and compare "shaNNNsum bigfile" results.
*/
#include "libbb.h"
#endif
-#define SHA1_BLOCK_SIZE 64
-#define SHA1_DIGEST_SIZE 20
-#define SHA1_HASH_SIZE SHA1_DIGEST_SIZE
-#define SHA1_MASK (SHA1_BLOCK_SIZE - 1)
+/* Some arch headers have conflicting defines */
+#undef ch
+#undef parity
+#undef maj
+#undef rnd
-static void sha1_process_block64(sha1_ctx_t *ctx)
+static void FAST_FUNC sha1_process_block64(sha1_ctx_t *ctx)
{
- uint32_t w[80], i, a, b, c, d, e, t;
+ unsigned t;
+ uint32_t W[80], a, b, c, d, e;
+ const uint32_t *words = (uint32_t*) ctx->wbuffer;
- /* 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] = ntohl(ctx->wbuffer[i]);
+ for (t = 0; t < 16; ++t) {
+ W[t] = ntohl(*words);
+ words++;
+ }
- for (/*i = SHA1_BLOCK_SIZE / 4*/; i < 80; ++i) {
- t = w[i - 3] ^ w[i - 8] ^ w[i - 14] ^ w[i - 16];
- w[i] = rotl32(t, 1);
+ for (/*t = 16*/; t < 80; ++t) {
+ uint32_t T = W[t - 3] ^ W[t - 8] ^ W[t - 14] ^ W[t - 16];
+ W[t] = rotl32(T, 1);
}
a = ctx->hash[0];
/* partial loop unrolling and is optimised for the Pentium 4 */
#define rnd(f,k) \
do { \
- t = a; a = rotl32(a,5) + f(b,c,d) + e + k + w[i]; \
- e = d; d = c; c = rotl32(b, 30); b = t; \
+ uint32_t T = a; \
+ a = rotl32(a, 5) + f(b, c, d) + e + k + W[t]; \
+ e = d; \
+ d = c; \
+ c = rotl32(b, 30); \
+ b = T; \
} while (0)
- for (i = 0; i < 20; ++i)
+ for (t = 0; t < 20; ++t)
rnd(ch, 0x5a827999);
- for (i = 20; i < 40; ++i)
+ for (/*t = 20*/; t < 40; ++t)
rnd(parity, 0x6ed9eba1);
- for (i = 40; i < 60; ++i)
+ for (/*t = 40*/; t < 60; ++t)
rnd(maj, 0x8f1bbcdc);
- for (i = 60; i < 80; ++i)
+ for (/*t = 60*/; t < 80; ++t)
rnd(parity, 0xca62c1d6);
#undef ch
#undef parity
ctx->hash[4] += e;
}
-/* Constants for SHA256 from FIPS 180-2:4.2.2. */
-static const uint32_t K256[80] = {
- 0x428a2f98, 0x71374491,
- 0xb5c0fbcf, 0xe9b5dba5,
- 0x3956c25b, 0x59f111f1,
- 0x923f82a4, 0xab1c5ed5,
- 0xd807aa98, 0x12835b01,
- 0x243185be, 0x550c7dc3,
- 0x72be5d74, 0x80deb1fe,
- 0x9bdc06a7, 0xc19bf174,
- 0xe49b69c1, 0xefbe4786,
- 0x0fc19dc6, 0x240ca1cc,
- 0x2de92c6f, 0x4a7484aa,
- 0x5cb0a9dc, 0x76f988da,
- 0x983e5152, 0xa831c66d,
- 0xb00327c8, 0xbf597fc7,
- 0xc6e00bf3, 0xd5a79147,
- 0x06ca6351, 0x14292967,
- 0x27b70a85, 0x2e1b2138,
- 0x4d2c6dfc, 0x53380d13,
- 0x650a7354, 0x766a0abb,
- 0x81c2c92e, 0x92722c85,
- 0xa2bfe8a1, 0xa81a664b,
- 0xc24b8b70, 0xc76c51a3,
- 0xd192e819, 0xd6990624,
- 0xf40e3585, 0x106aa070,
- 0x19a4c116, 0x1e376c08,
- 0x2748774c, 0x34b0bcb5,
- 0x391c0cb3, 0x4ed8aa4a,
- 0x5b9cca4f, 0x682e6ff3,
- 0x748f82ee, 0x78a5636f,
- 0x84c87814, 0x8cc70208,
- 0x90befffa, 0xa4506ceb,
- 0xbef9a3f7, 0xc67178f2,
- 0xca273ece, 0xd186b8c7, /* [64]+ are used for sha512 only */
- 0xeada7dd6, 0xf57d4f7f,
- 0x06f067aa, 0x0a637dc5,
- 0x113f9804, 0x1b710b35,
- 0x28db77f5, 0x32caab7b,
- 0x3c9ebe0a, 0x431d67c4,
- 0x4cc5d4be, 0x597f299c,
- 0x5fcb6fab, 0x6c44198c
-};
-/* Constants for SHA512 from FIPS 180-2:4.2.3. */
-static const uint32_t K512_lo[80] = {
- 0xd728ae22, 0x23ef65cd,
- 0xec4d3b2f, 0x8189dbbc,
- 0xf348b538, 0xb605d019,
- 0xaf194f9b, 0xda6d8118,
- 0xa3030242, 0x45706fbe,
- 0x4ee4b28c, 0xd5ffb4e2,
- 0xf27b896f, 0x3b1696b1,
- 0x25c71235, 0xcf692694,
- 0x9ef14ad2, 0x384f25e3,
- 0x8b8cd5b5, 0x77ac9c65,
- 0x592b0275, 0x6ea6e483,
- 0xbd41fbd4, 0x831153b5,
- 0xee66dfab, 0x2db43210,
- 0x98fb213f, 0xbeef0ee4,
- 0x3da88fc2, 0x930aa725,
- 0xe003826f, 0x0a0e6e70,
- 0x46d22ffc, 0x5c26c926,
- 0x5ac42aed, 0x9d95b3df,
- 0x8baf63de, 0x3c77b2a8,
- 0x47edaee6, 0x1482353b,
- 0x4cf10364, 0xbc423001,
- 0xd0f89791, 0x0654be30,
- 0xd6ef5218, 0x5565a910,
- 0x5771202a, 0x32bbd1b8,
- 0xb8d2d0c8, 0x5141ab53,
- 0xdf8eeb99, 0xe19b48a8,
- 0xc5c95a63, 0xe3418acb,
- 0x7763e373, 0xd6b2b8a3,
- 0x5defb2fc, 0x43172f60,
- 0xa1f0ab72, 0x1a6439ec,
- 0x23631e28, 0xde82bde9,
- 0xb2c67915, 0xe372532b,
- 0xea26619c, 0x21c0c207,
- 0xcde0eb1e, 0xee6ed178,
- 0x72176fba, 0xa2c898a6,
- 0xbef90dae, 0x131c471b,
- 0x23047d84, 0x40c72493,
- 0x15c9bebc, 0x9c100d4c,
- 0xcb3e42b6, 0xfc657e2a,
- 0x3ad6faec, 0x4a475817
+/* Constants for SHA512 from FIPS 180-2:4.2.3.
+ * SHA256 constants from FIPS 180-2:4.2.2
+ * are the most significant half of first 64 elements
+ * of the same array.
+ */
+static const uint64_t sha_K[80] = {
+ 0x428a2f98d728ae22ULL, 0x7137449123ef65cdULL,
+ 0xb5c0fbcfec4d3b2fULL, 0xe9b5dba58189dbbcULL,
+ 0x3956c25bf348b538ULL, 0x59f111f1b605d019ULL,
+ 0x923f82a4af194f9bULL, 0xab1c5ed5da6d8118ULL,
+ 0xd807aa98a3030242ULL, 0x12835b0145706fbeULL,
+ 0x243185be4ee4b28cULL, 0x550c7dc3d5ffb4e2ULL,
+ 0x72be5d74f27b896fULL, 0x80deb1fe3b1696b1ULL,
+ 0x9bdc06a725c71235ULL, 0xc19bf174cf692694ULL,
+ 0xe49b69c19ef14ad2ULL, 0xefbe4786384f25e3ULL,
+ 0x0fc19dc68b8cd5b5ULL, 0x240ca1cc77ac9c65ULL,
+ 0x2de92c6f592b0275ULL, 0x4a7484aa6ea6e483ULL,
+ 0x5cb0a9dcbd41fbd4ULL, 0x76f988da831153b5ULL,
+ 0x983e5152ee66dfabULL, 0xa831c66d2db43210ULL,
+ 0xb00327c898fb213fULL, 0xbf597fc7beef0ee4ULL,
+ 0xc6e00bf33da88fc2ULL, 0xd5a79147930aa725ULL,
+ 0x06ca6351e003826fULL, 0x142929670a0e6e70ULL,
+ 0x27b70a8546d22ffcULL, 0x2e1b21385c26c926ULL,
+ 0x4d2c6dfc5ac42aedULL, 0x53380d139d95b3dfULL,
+ 0x650a73548baf63deULL, 0x766a0abb3c77b2a8ULL,
+ 0x81c2c92e47edaee6ULL, 0x92722c851482353bULL,
+ 0xa2bfe8a14cf10364ULL, 0xa81a664bbc423001ULL,
+ 0xc24b8b70d0f89791ULL, 0xc76c51a30654be30ULL,
+ 0xd192e819d6ef5218ULL, 0xd69906245565a910ULL,
+ 0xf40e35855771202aULL, 0x106aa07032bbd1b8ULL,
+ 0x19a4c116b8d2d0c8ULL, 0x1e376c085141ab53ULL,
+ 0x2748774cdf8eeb99ULL, 0x34b0bcb5e19b48a8ULL,
+ 0x391c0cb3c5c95a63ULL, 0x4ed8aa4ae3418acbULL,
+ 0x5b9cca4f7763e373ULL, 0x682e6ff3d6b2b8a3ULL,
+ 0x748f82ee5defb2fcULL, 0x78a5636f43172f60ULL,
+ 0x84c87814a1f0ab72ULL, 0x8cc702081a6439ecULL,
+ 0x90befffa23631e28ULL, 0xa4506cebde82bde9ULL,
+ 0xbef9a3f7b2c67915ULL, 0xc67178f2e372532bULL,
+ 0xca273eceea26619cULL, 0xd186b8c721c0c207ULL, /* [64]+ are used for sha512 only */
+ 0xeada7dd6cde0eb1eULL, 0xf57d4f7fee6ed178ULL,
+ 0x06f067aa72176fbaULL, 0x0a637dc5a2c898a6ULL,
+ 0x113f9804bef90daeULL, 0x1b710b35131c471bULL,
+ 0x28db77f523047d84ULL, 0x32caab7b40c72493ULL,
+ 0x3c9ebe0a15c9bebcULL, 0x431d67c49c100d4cULL,
+ 0x4cc5d4becb3e42b6ULL, 0x597f299cfc657e2aULL,
+ 0x5fcb6fab3ad6faecULL, 0x6c44198c4a475817ULL
};
-/* Process LEN bytes of BUFFER, accumulating context into CTX.
- LEN is rounded _down_ to 64. */
-static void sha256_process_block64(const void *buffer, size_t len, sha256_ctx_t *ctx)
-{
- const uint32_t *words = buffer;
- uint32_t a = ctx->H[0];
- uint32_t b = ctx->H[1];
- uint32_t c = ctx->H[2];
- uint32_t d = ctx->H[3];
- uint32_t e = ctx->H[4];
- uint32_t f = ctx->H[5];
- uint32_t g = ctx->H[6];
- uint32_t h = ctx->H[7];
-
- /* First increment the byte count. FIPS 180-2 specifies the possible
- length of the file up to 2^64 _bits_.
- We compute the number of _bytes_ and convert to bits later. */
- len &= ~(size_t)(sizeof(uint32_t) * 16 - 1);
- ctx->total64 += len;
+#undef Ch
+#undef Maj
+#undef S0
+#undef S1
+#undef R0
+#undef R1
- /* Process all bytes in the buffer with 64 bytes in each round of
- the loop. */
- len /= (sizeof(uint32_t) * 16);
- while (len) {
- unsigned t;
- uint32_t W[64];
+static void FAST_FUNC sha256_process_block64(sha256_ctx_t *ctx)
+{
+ unsigned t;
+ uint32_t W[64], a, b, c, d, e, f, g, h;
+ const uint32_t *words = (uint32_t*) ctx->wbuffer;
- /* Operators defined in FIPS 180-2:4.1.2. */
+ /* Operators defined in FIPS 180-2:4.1.2. */
#define Ch(x, y, z) ((x & y) ^ (~x & z))
#define Maj(x, y, z) ((x & y) ^ (x & z) ^ (y & z))
#define S0(x) (rotr32(x, 2) ^ rotr32(x, 13) ^ rotr32(x, 22))
#define R0(x) (rotr32(x, 7) ^ rotr32(x, 18) ^ (x >> 3))
#define R1(x) (rotr32(x, 17) ^ rotr32(x, 19) ^ (x >> 10))
- /* Compute the message schedule according to FIPS 180-2:6.2.2 step 2. */
- for (t = 0; t < 16; ++t) {
- W[t] = ntohl(*words);
- ++words;
- }
+ /* Compute the message schedule according to FIPS 180-2:6.2.2 step 2. */
+ for (t = 0; t < 16; ++t) {
+ W[t] = ntohl(*words);
+ words++;
+ }
- for (/*t = 16*/; t < 64; ++t)
- W[t] = R1(W[t - 2]) + W[t - 7] + R0(W[t - 15]) + W[t - 16];
-
- /* The actual computation according to FIPS 180-2:6.2.2 step 3. */
- for (t = 0; t < 64; ++t) {
- uint32_t T1 = h + S1(e) + Ch(e, f, g) + K256[t] + W[t];
- uint32_t T2 = S0(a) + Maj(a, b, c);
- h = g;
- g = f;
- f = e;
- e = d + T1;
- d = c;
- c = b;
- b = a;
- a = T1 + T2;
- }
+ for (/*t = 16*/; t < 64; ++t)
+ W[t] = R1(W[t - 2]) + W[t - 7] + R0(W[t - 15]) + W[t - 16];
+
+ a = ctx->hash[0];
+ b = ctx->hash[1];
+ c = ctx->hash[2];
+ d = ctx->hash[3];
+ e = ctx->hash[4];
+ f = ctx->hash[5];
+ g = ctx->hash[6];
+ h = ctx->hash[7];
+
+ /* The actual computation according to FIPS 180-2:6.2.2 step 3. */
+ for (t = 0; t < 64; ++t) {
+ /* Need to fetch upper half of sha_K[t]
+ * (I hope compiler is clever enough to just fetch
+ * upper half)
+ */
+ uint32_t K_t = sha_K[t] >> 32;
+ uint32_t T1 = h + S1(e) + Ch(e, f, g) + K_t + W[t];
+ uint32_t T2 = S0(a) + Maj(a, b, c);
+ h = g;
+ g = f;
+ f = e;
+ e = d + T1;
+ d = c;
+ c = b;
+ b = a;
+ a = T1 + T2;
+ }
#undef Ch
#undef Maj
#undef S0
#undef S1
#undef R0
#undef R1
- /* Add the starting values of the context according to FIPS 180-2:6.2.2
- step 4. */
- ctx->H[0] = a += ctx->H[0];
- ctx->H[1] = b += ctx->H[1];
- ctx->H[2] = c += ctx->H[2];
- ctx->H[3] = d += ctx->H[3];
- ctx->H[4] = e += ctx->H[4];
- ctx->H[5] = f += ctx->H[5];
- ctx->H[6] = g += ctx->H[6];
- ctx->H[7] = h += ctx->H[7];
-
- /* Prepare for the next round. */
- len--;
- }
+ /* Add the starting values of the context according to FIPS 180-2:6.2.2
+ step 4. */
+ ctx->hash[0] += a;
+ ctx->hash[1] += b;
+ ctx->hash[2] += c;
+ ctx->hash[3] += d;
+ ctx->hash[4] += e;
+ ctx->hash[5] += f;
+ ctx->hash[6] += g;
+ ctx->hash[7] += h;
}
-/* Process LEN bytes of BUFFER, accumulating context into CTX.
- LEN is rounded _down_ to 128. */
-static void sha512_process_block128(const void *buffer, size_t len, sha512_ctx_t *ctx)
-{
- const uint64_t *words = buffer;
- uint64_t a = ctx->H[0];
- uint64_t b = ctx->H[1];
- uint64_t c = ctx->H[2];
- uint64_t d = ctx->H[3];
- uint64_t e = ctx->H[4];
- uint64_t f = ctx->H[5];
- uint64_t g = ctx->H[6];
- uint64_t h = ctx->H[7];
- /* First increment the byte count. FIPS 180-2 specifies the possible
- length of the file up to 2^128 _bits_.
- We compute the number of _bytes_ and convert to bits later. */
- len &= ~(size_t)(sizeof(uint64_t) * 16 - 1);
- ctx->total64[0] += len;
- if (ctx->total64[0] < len)
- ctx->total64[1]++;
-
- len /= (sizeof(uint64_t) * 16);
- while (len) {
- unsigned t;
- uint64_t W[80];
-
- /* Operators defined in FIPS 180-2:4.1.2. */
+static void FAST_FUNC sha512_process_block128(sha512_ctx_t *ctx)
+{
+ unsigned t;
+ uint64_t W[80];
+ /* On i386, having assignments here (not later as sha256 does)
+ * produces 99 bytes smaller code with gcc 4.3.1
+ */
+ uint64_t a = ctx->hash[0];
+ uint64_t b = ctx->hash[1];
+ uint64_t c = ctx->hash[2];
+ uint64_t d = ctx->hash[3];
+ uint64_t e = ctx->hash[4];
+ uint64_t f = ctx->hash[5];
+ uint64_t g = ctx->hash[6];
+ uint64_t h = ctx->hash[7];
+ const uint64_t *words = (uint64_t*) ctx->wbuffer;
+
+ /* Operators defined in FIPS 180-2:4.1.2. */
#define Ch(x, y, z) ((x & y) ^ (~x & z))
#define Maj(x, y, z) ((x & y) ^ (x & z) ^ (y & z))
#define S0(x) (rotr64(x, 28) ^ rotr64(x, 34) ^ rotr64(x, 39))
#define R0(x) (rotr64(x, 1) ^ rotr64(x, 8) ^ (x >> 7))
#define R1(x) (rotr64(x, 19) ^ rotr64(x, 61) ^ (x >> 6))
- /* Compute the message schedule according to FIPS 180-2:6.3.2 step 2. */
- for (t = 0; t < 16; ++t) {
- W[t] = ntoh64(*words);
- ++words;
- }
- for (/*t = 16*/; t < 80; ++t)
- W[t] = R1(W[t - 2]) + W[t - 7] + R0(W[t - 15]) + W[t - 16];
-
- /* The actual computation according to FIPS 180-2:6.3.2 step 3. */
- for (t = 0; t < 80; ++t) {
- uint64_t K512_t = ((uint64_t)(K256[t]) << 32) + K512_lo[t];
- uint64_t T1 = h + S1(e) + Ch(e, f, g) + K512_t + W[t];
- uint64_t T2 = S0(a) + Maj(a, b, c);
- h = g;
- g = f;
- f = e;
- e = d + T1;
- d = c;
- c = b;
- b = a;
- a = T1 + T2;
- }
+ /* Compute the message schedule according to FIPS 180-2:6.3.2 step 2. */
+ for (t = 0; t < 16; ++t) {
+ W[t] = ntoh64(*words);
+ words++;
+ }
+ for (/*t = 16*/; t < 80; ++t)
+ W[t] = R1(W[t - 2]) + W[t - 7] + R0(W[t - 15]) + W[t - 16];
+
+ /* The actual computation according to FIPS 180-2:6.3.2 step 3. */
+ for (t = 0; t < 80; ++t) {
+ uint64_t T1 = h + S1(e) + Ch(e, f, g) + sha_K[t] + W[t];
+ uint64_t T2 = S0(a) + Maj(a, b, c);
+ h = g;
+ g = f;
+ f = e;
+ e = d + T1;
+ d = c;
+ c = b;
+ b = a;
+ a = T1 + T2;
+ }
#undef Ch
#undef Maj
#undef S0
#undef S1
#undef R0
#undef R1
- /* Add the starting values of the context according to FIPS 180-2:6.3.2
- step 4. */
- ctx->H[0] = a += ctx->H[0];
- ctx->H[1] = b += ctx->H[1];
- ctx->H[2] = c += ctx->H[2];
- ctx->H[3] = d += ctx->H[3];
- ctx->H[4] = e += ctx->H[4];
- ctx->H[5] = f += ctx->H[5];
- ctx->H[6] = g += ctx->H[6];
- ctx->H[7] = h += ctx->H[7];
-
- len--;
- }
+ /* Add the starting values of the context according to FIPS 180-2:6.3.2
+ step 4. */
+ ctx->hash[0] += a;
+ ctx->hash[1] += b;
+ ctx->hash[2] += c;
+ ctx->hash[3] += d;
+ ctx->hash[4] += e;
+ ctx->hash[5] += f;
+ ctx->hash[6] += g;
+ ctx->hash[7] += h;
}
void FAST_FUNC sha1_begin(sha1_ctx_t *ctx)
{
- ctx->total64 = 0;
ctx->hash[0] = 0x67452301;
ctx->hash[1] = 0xefcdab89;
ctx->hash[2] = 0x98badcfe;
ctx->hash[3] = 0x10325476;
ctx->hash[4] = 0xc3d2e1f0;
+ ctx->total64 = 0;
+ ctx->process_block = sha1_process_block64;
}
static const uint32_t init256[] = {
0xfb41bd6b,
0x137e2179
};
+
/* Initialize structure containing state of computation.
(FIPS 180-2:5.3.2) */
void FAST_FUNC sha256_begin(sha256_ctx_t *ctx)
{
- memcpy(ctx->H, init256, sizeof(init256));
+ memcpy(ctx->hash, init256, sizeof(init256));
ctx->total64 = 0;
- ctx->wbuflen = 0;
+ ctx->process_block = sha256_process_block64;
}
+
/* Initialize structure containing state of computation.
(FIPS 180-2:5.3.3) */
void FAST_FUNC sha512_begin(sha512_ctx_t *ctx)
{
int i;
for (i = 0; i < 8; i++)
- ctx->H[i] = ((uint64_t)(init256[i]) << 32) + init512_lo[i];
+ ctx->hash[i] = ((uint64_t)(init256[i]) << 32) + init512_lo[i];
ctx->total64[0] = ctx->total64[1] = 0;
- ctx->wbuflen = 0;
}
-/* SHA1 hash data in an array of bytes into hash buffer and call the */
-/* hash_compile function as required. */
+/* Used also for sha256 */
void FAST_FUNC sha1_hash(const void *buffer, size_t len, sha1_ctx_t *ctx)
{
- uint32_t pos = (uint32_t) (ctx->total64 & SHA1_MASK);
- uint32_t freeb = SHA1_BLOCK_SIZE - pos;
- const unsigned char *sp = buffer;
+ unsigned in_buf = ctx->total64 & 63;
+ unsigned add = 64 - in_buf;
ctx->total64 += len;
- while (len >= freeb) { /* transfer whole blocks while possible */
- memcpy(((unsigned char *) ctx->wbuffer) + pos, sp, freeb);
- sp += freeb;
- len -= freeb;
- freeb = SHA1_BLOCK_SIZE;
- pos = 0;
- sha1_process_block64(ctx);
- }
-
- memcpy(((unsigned char *) ctx->wbuffer) + pos, sp, len);
-}
-
-void FAST_FUNC sha256_hash(const void *buffer, size_t len, sha256_ctx_t *ctx)
-{
- /* When we already have some bits in our internal buffer concatenate
- both inputs first. */
- if (ctx->wbuflen != 0) {
- unsigned add;
-
- /* NB: 1/2 of wbuffer is used only in sha256_end
- * when length field is added and hashed.
- * With buffer twice as small, it may happen that
- * we have it almost full and can't add length field. */
-
- add = sizeof(ctx->wbuffer)/2 - ctx->wbuflen;
- if (add > len)
- add = len;
- memcpy(&ctx->wbuffer[ctx->wbuflen], buffer, add);
- ctx->wbuflen += add;
-
- /* If we still didn't collect full wbuffer, bail out */
- if (ctx->wbuflen < sizeof(ctx->wbuffer)/2)
- return;
-
- sha256_process_block64(ctx->wbuffer, 64, ctx);
- ctx->wbuflen = 0;
+ while (len >= add) { /* transfer whole blocks while possible */
+ memcpy(ctx->wbuffer + in_buf, buffer, add);
buffer = (const char *)buffer + add;
len -= add;
+ add = 64;
+ in_buf = 0;
+ ctx->process_block(ctx);
}
- /* Process available complete blocks. */
- if (len >= 64) {
- if (UNALIGNED_P(buffer, uint32_t)) {
- while (len > 64) {
- sha256_process_block64(memcpy(ctx->wbuffer, buffer, 64), 64, ctx);
- buffer = (const char *)buffer + 64;
- len -= 64;
- }
- } else {
- sha256_process_block64(buffer, len /*& ~63*/, ctx);
- buffer = (const char *)buffer + (len & ~63);
- len &= 63;
- }
- }
-
- /* Move remaining bytes into internal buffer. */
- if (len > 0) {
- memcpy(ctx->wbuffer, buffer, len);
- ctx->wbuflen = len;
- }
+ memcpy(ctx->wbuffer + in_buf, buffer, len);
}
void FAST_FUNC sha512_hash(const void *buffer, size_t len, sha512_ctx_t *ctx)
{
- if (ctx->wbuflen != 0) {
- unsigned add;
+ unsigned in_buf = ctx->total64[0] & 127;
+ unsigned add = 128 - in_buf;
- add = sizeof(ctx->wbuffer)/2 - ctx->wbuflen;
- if (add > len)
- add = len;
- memcpy(&ctx->wbuffer[ctx->wbuflen], buffer, add);
- ctx->wbuflen += add;
-
- if (ctx->wbuflen < sizeof(ctx->wbuffer)/2)
- return;
+ /* First increment the byte count. FIPS 180-2 specifies the possible
+ length of the file up to 2^128 _bits_.
+ We compute the number of _bytes_ and convert to bits later. */
+ ctx->total64[0] += len;
+ if (ctx->total64[0] < len)
+ ctx->total64[1]++;
- sha512_process_block128(ctx->wbuffer, 128, ctx);
- ctx->wbuflen = 0;
+ while (len >= add) { /* transfer whole blocks while possible */
+ memcpy(ctx->wbuffer + in_buf, buffer, add);
buffer = (const char *)buffer + add;
len -= add;
+ add = 128;
+ in_buf = 0;
+ sha512_process_block128(ctx);
}
- if (len >= 128) {
- if (UNALIGNED_P(buffer, uint64_t)) {
- while (len > 128) {
- sha512_process_block128(memcpy(ctx->wbuffer, buffer, 128), 128, ctx);
- buffer = (const char *)buffer + 128;
- len -= 128;
- }
- } else {
- sha512_process_block128(buffer, len /*& ~127*/, ctx);
- buffer = (const char *)buffer + (len & ~127);
- len &= 127;
- }
- }
-
- if (len > 0) {
- memcpy(ctx->wbuffer, buffer, len);
- ctx->wbuflen = len;
- }
+ memcpy(ctx->wbuffer + in_buf, buffer, len);
}
+/* Used also for sha256 */
void FAST_FUNC sha1_end(void *resbuf, sha1_ctx_t *ctx)
{
- /* SHA1 Final padding and digest calculation */
-#if BB_BIG_ENDIAN
- static const uint32_t mask[4] = { 0x00000000, 0xff000000, 0xffff0000, 0xffffff00 };
- static const uint32_t bits[4] = { 0x80000000, 0x00800000, 0x00008000, 0x00000080 };
-#else
- static const uint32_t mask[4] = { 0x00000000, 0x000000ff, 0x0000ffff, 0x00ffffff };
- static const uint32_t bits[4] = { 0x00000080, 0x00008000, 0x00800000, 0x80000000 };
-#endif
-
- uint8_t *hval = resbuf;
- uint32_t i, cnt = (uint32_t) (ctx->total64 & 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->wbuffer[cnt >> 2] =
- (ctx->wbuffer[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->wbuffer[15] = 0;
- sha1_process_block64(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->wbuffer[cnt++] = 0;
-
- /* assemble the 64-bit counter of bits in the buffer in BE */
- /* format */
- {
- uint64_t t = ctx->total64 << 3;
- t = hton64(t);
- /* wbuffer is suitably aligned for this */
- *(uint64_t *) &ctx->wbuffer[14] = t;
+ unsigned pad, in_buf;
+
+ in_buf = ctx->total64 & 63;
+ /* Pad the buffer to the next 64-byte boundary with 0x80,0,0,0... */
+ ctx->wbuffer[in_buf++] = 0x80;
+
+ /* This loop iterates either once or twice, no more, no less */
+ while (1) {
+ pad = 64 - in_buf;
+ memset(ctx->wbuffer + in_buf, 0, pad);
+ in_buf = 0;
+ /* Do we have enough space for the length count? */
+ if (pad >= 8) {
+ /* Store the 64-bit counter of bits in the buffer in BE format */
+ uint64_t t = ctx->total64 << 3;
+ t = hton64(t);
+ /* wbuffer is suitably aligned for this */
+ *(uint64_t *) (&ctx->wbuffer[64 - 8]) = t;
+ }
+ ctx->process_block(ctx);
+ if (pad >= 8)
+ break;
}
- sha1_process_block64(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));
-}
-
-
-/* Process the remaining bytes in the internal buffer and the usual
- prolog according to the standard and write the result to RESBUF.
-
- IMPORTANT: On some systems it is required that RESBUF is correctly
- aligned for a 32 bits value. */
-void FAST_FUNC sha256_end(void *resbuf, sha256_ctx_t *ctx)
-{
- /* Take yet unprocessed bytes into account. */
- unsigned bytes = ctx->wbuflen;
- unsigned pad;
-
- /* Now count remaining bytes. */
- ctx->total64 += bytes;
-
- /* Pad the buffer to the next 64-byte boundary with 0x80,0,0,0...
- (FIPS 180-2:5.1.1) */
- pad = (bytes >= 56 ? 64 + 56 - bytes : 56 - bytes);
- memset(&ctx->wbuffer[bytes], 0, pad);
- ctx->wbuffer[bytes] = 0x80;
-
- /* Put the 64-bit file length in *bits* at the end of the buffer. */
- {
- uint64_t t = ctx->total64 << 3;
- t = hton64(t);
- /* wbuffer is suitably aligned for this */
- *(uint64_t *) &ctx->wbuffer[bytes + pad] = t;
+ in_buf = (ctx->process_block == sha1_process_block64) ? 5 : 8;
+ /* This way we do not impose alignment constraints on resbuf: */
+ if (BB_LITTLE_ENDIAN) {
+ unsigned i;
+ for (i = 0; i < in_buf; ++i)
+ ctx->hash[i] = htonl(ctx->hash[i]);
}
-
- /* Process last bytes. */
- sha256_process_block64(ctx->wbuffer, bytes + pad + 8, ctx);
-
- for (unsigned i = 0; i < 8; ++i)
- ((uint32_t *) resbuf)[i] = ntohl(ctx->H[i]);
+ memcpy(resbuf, ctx->hash, sizeof(ctx->hash[0]) * in_buf);
}
-/* Process the remaining bytes in the internal buffer and the usual
- prolog according to the standard and write the result to RESBUF.
-
- IMPORTANT: On some systems it is required that RESBUF is correctly
- aligned for a 64 bits value. */
void FAST_FUNC sha512_end(void *resbuf, sha512_ctx_t *ctx)
{
- unsigned bytes = ctx->wbuflen;
- unsigned pad;
-
- ctx->total64[0] += bytes;
- if (ctx->total64[0] < bytes)
- ctx->total64[1]++;
+ unsigned pad, in_buf;
+ in_buf = ctx->total64[0] & 127;
/* Pad the buffer to the next 128-byte boundary with 0x80,0,0,0...
- (FIPS 180-2:5.1.2) */
- pad = bytes >= 112 ? 128 + 112 - bytes : 112 - bytes;
- memset(&ctx->wbuffer[bytes], 0, pad);
- ctx->wbuffer[bytes] = 0x80;
-
- *(uint64_t *) &ctx->wbuffer[bytes + pad + 8] = hton64(ctx->total64[0] << 3);
- *(uint64_t *) &ctx->wbuffer[bytes + pad] = hton64((ctx->total64[1] << 3) | (ctx->total64[0] >> 61));
-
- sha512_process_block128(ctx->wbuffer, bytes + pad + 16, ctx);
+ * (FIPS 180-2:5.1.2)
+ */
+ ctx->wbuffer[in_buf++] = 0x80;
+
+ while (1) {
+ pad = 128 - in_buf;
+ memset(ctx->wbuffer + in_buf, 0, pad);
+ in_buf = 0;
+ if (pad >= 16) {
+ /* Store the 128-bit counter of bits in the buffer in BE format */
+ uint64_t t;
+ t = ctx->total64[0] << 3;
+ t = hton64(t);
+ *(uint64_t *) (&ctx->wbuffer[128 - 8]) = t;
+ t = (ctx->total64[1] << 3) | (ctx->total64[0] >> 61);
+ t = hton64(t);
+ *(uint64_t *) (&ctx->wbuffer[128 - 16]) = t;
+ }
+ sha512_process_block128(ctx);
+ if (pad >= 16)
+ break;
+ }
- for (unsigned i = 0; i < 8; ++i)
- ((uint64_t *) resbuf)[i] = hton64(ctx->H[i]);
+ if (BB_LITTLE_ENDIAN) {
+ unsigned i;
+ for (i = 0; i < ARRAY_SIZE(ctx->hash); ++i)
+ ctx->hash[i] = hton64(ctx->hash[i]);
+ }
+ memcpy(resbuf, ctx->hash, sizeof(ctx->hash));
}
projects / oweals / busybox.git / blobdiff