Merge commit '6722737ece4b8db3e30b53aef8f981f53db1621e^'

2 files changed, 456 insertions, 512 deletions

diff --git a/libbb/hash_md5_sha.c b/libbb/hash_md5_sha.c
index 3e708ef7e..aeacddef8 100644
--- a/libbb/hash_md5_sha.c
+++ b/libbb/hash_md5_sha.c
@@ -1,31 +1,10 @@
 /* vi: set sw=4 ts=4: */
 /*
- * Based on shasum from http://www.netsw.org/crypto/hash/
- * Majorly hacked up to use Dr Brian Gladman's sha1 code
+ * Utility routines.
  *
- * Copyright (C) 2002 Dr Brian Gladman <brg@gladman.me.uk>, Worcester, UK.
- * Copyright (C) 2003 Glenn L. McGrath
- * Copyright (C) 2003 Erik Andersen
+ * Copyright (C) 2010 Denys Vlasenko
  *
  * Licensed under GPLv2 or later, see file LICENSE in this source tree.
- *
- * ---------------------------------------------------------------------------
- * 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
- *
- * ---------------------------------------------------------------------------
- *
- * SHA256 and SHA512 parts are:
- * Released into the Public Domain by Ulrich Drepper <drepper@redhat.com>.
- * 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"
@@ -53,332 +32,18 @@ static ALWAYS_INLINE uint64_t rotr64(uint64_t x, unsigned n)
 }
 
 
-static void FAST_FUNC sha1_process_block64(sha1_ctx_t *ctx)
-{
-        unsigned t;
-        uint32_t W[80], a, b, c, d, e;
-        const uint32_t *words = (uint32_t*) ctx->wbuffer;
-
-        for (t = 0; t < 16; ++t)
-                W[t] = SWAP_BE32(words[t]);
-        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];
-        b = ctx->hash[1];
-        c = ctx->hash[2];
-        d = ctx->hash[3];
-        e = ctx->hash[4];
-
-#undef ch
-#undef parity
-#undef maj
-#undef rnd
-#define ch(x,y,z)        ((z) ^ ((x) & ((y) ^ (z))))
-#define parity(x,y,z)    ((x) ^ (y) ^ (z))
-#define maj(x,y,z)       (((x) & (y)) | ((z) & ((x) | (y))))
-/* A normal version as set out in the FIPS.  */
-#define rnd(f,k) \
-        do { \
-                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 (t = 0; t < 20; ++t)
-                rnd(ch, 0x5a827999);
-
-        for (/*t = 20*/; t < 40; ++t)
-                rnd(parity, 0x6ed9eba1);
-
-        for (/*t = 40*/; t < 60; ++t)
-                rnd(maj, 0x8f1bbcdc);
-
-        for (/*t = 60*/; t < 80; ++t)
-                rnd(parity, 0xca62c1d6);
-#undef ch
-#undef parity
-#undef maj
-#undef rnd
-
-        ctx->hash[0] += a;
-        ctx->hash[1] += b;
-        ctx->hash[2] += c;
-        ctx->hash[3] += d;
-        ctx->hash[4] += e;
-}
-
-/* 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.
+/* Feed data through a temporary buffer.
+ * The internal buffer remembers previous data until it has 64
+ * bytes worth to pass on.
  */
-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
-};
-
-#undef Ch
-#undef Maj
-#undef S0
-#undef S1
-#undef R0
-#undef R1
-
-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.  */
-#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 S1(x) (rotr32(x, 6) ^ rotr32(x, 11) ^ rotr32(x, 25))
-#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] = SWAP_BE32(words[t]);
-        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->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;
-}
-
-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 S1(x) (rotr64(x, 14) ^ rotr64(x, 18) ^ rotr64(x, 41))
-#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] = SWAP_BE64(words[t]);
-        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->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->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[] = {
-        0x6a09e667,
-        0xbb67ae85,
-        0x3c6ef372,
-        0xa54ff53a,
-        0x510e527f,
-        0x9b05688c,
-        0x1f83d9ab,
-        0x5be0cd19,
-        0,
-        0,
-};
-static const uint32_t init512_lo[] = {
-        0xf3bcc908,
-        0x84caa73b,
-        0xfe94f82b,
-        0x5f1d36f1,
-        0xade682d1,
-        0x2b3e6c1f,
-        0xfb41bd6b,
-        0x137e2179,
-        0,
-        0,
-};
-
-/* Initialize structure containing state of computation.
-   (FIPS 180-2:5.3.2)  */
-void FAST_FUNC sha256_begin(sha256_ctx_t *ctx)
-{
-        memcpy(ctx->hash, init256, sizeof(init256));
-        /*ctx->total64 = 0; - done by extending init256 with two 32-bit zeros */
-        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;
-        /* Two extra iterations zero out ctx->total64[] */
-        for (i = 0; i < 8+2; i++)
-                ctx->hash[i] = ((uint64_t)(init256[i]) << 32) + init512_lo[i];
-        /*ctx->total64[0] = ctx->total64[1] = 0; - already done */
-}
-
-
-/* Used also for sha256 */
-void FAST_FUNC sha1_hash(sha1_ctx_t *ctx, const void *buffer, size_t len)
+static void FAST_FUNC common64_hash(md5_ctx_t *ctx, const void *buffer, size_t len)
 {
         unsigned bufpos = ctx->total64 & 63;
-        unsigned remaining;
 
         ctx->total64 += len;
-#if 0
-        remaining = 64 - bufpos;
-
-        /* Hash whole blocks */
-        while (len >= remaining) {
-                memcpy(ctx->wbuffer + bufpos, buffer, remaining);
-                buffer = (const char *)buffer + remaining;
-                len -= remaining;
-                remaining = 64;
-                bufpos = 0;
-                ctx->process_block(ctx);
-        }
 
-        /* Save last, partial blosk */
-        memcpy(ctx->wbuffer + bufpos, buffer, len);
-#else
-        /* Tiny bit smaller code */
         while (1) {
-                remaining = 64 - bufpos;
+                unsigned remaining = 64 - bufpos;
                 if (remaining > len)
                         remaining = len;
                 /* Copy data into aligned buffer */
@@ -394,62 +59,12 @@ void FAST_FUNC sha1_hash(sha1_ctx_t *ctx, const void *buffer, size_t len)
                 ctx->process_block(ctx);
                 /*bufpos = 0; - already is */
         }
-#endif
 }
 
-void FAST_FUNC sha512_hash(sha512_ctx_t *ctx, const void *buffer, size_t len)
-{
-        unsigned bufpos = ctx->total64[0] & 127;
-        unsigned remaining;
-
-        /* 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]++;
-#if 0
-        remaining = 128 - bufpos;
-
-        /* Hash whole blocks */
-        while (len >= remaining) {
-                memcpy(ctx->wbuffer + bufpos, buffer, remaining);
-                buffer = (const char *)buffer + remaining;
-                len -= remaining;
-                remaining = 128;
-                bufpos = 0;
-                sha512_process_block128(ctx);
-        }
-
-        /* Save last, partial blosk */
-        memcpy(ctx->wbuffer + bufpos, buffer, len);
-#else
-        while (1) {
-                remaining = 128 - bufpos;
-                if (remaining > len)
-                        remaining = len;
-                /* Copy data into aligned buffer */
-                memcpy(ctx->wbuffer + bufpos, buffer, remaining);
-                len -= remaining;
-                buffer = (const char *)buffer + remaining;
-                bufpos += remaining;
-                /* clever way to do "if (bufpos != 128) break; ... ; bufpos = 0;" */
-                bufpos -= 128;
-                if (bufpos != 0)
-                        break;
-                /* Buffer is filled up, process it */
-                sha512_process_block128(ctx);
-                /*bufpos = 0; - already is */
-        }
-#endif
-}
-
-
-/* Used also for sha256 */
-void FAST_FUNC sha1_end(sha1_ctx_t *ctx, void *resbuf)
+/* Process the remaining bytes in the buffer */
+static void FAST_FUNC common64_end(md5_ctx_t *ctx, int swap_needed)
 {
         unsigned bufpos = ctx->total64 & 63;
-
         /* Pad the buffer to the next 64-byte boundary with 0x80,0,0,0... */
         ctx->wbuffer[bufpos++] = 0x80;
 
@@ -459,9 +74,10 @@ void FAST_FUNC sha1_end(sha1_ctx_t *ctx, void *resbuf)
                 memset(ctx->wbuffer + bufpos, 0, remaining);
                 /* Do we have enough space for the length count? */
                 if (remaining >= 8) {
-                        /* Store the 64-bit counter of bits in the buffer in BE format */
+                        /* Store the 64-bit counter of bits in the buffer */
                         uint64_t t = ctx->total64 << 3;
-                        t = SWAP_BE64(t);
+                        if (swap_needed)
+                                t = bb_bswap_64(t);
                         /* wbuffer is suitably aligned for this */
                         *(uint64_t *) (&ctx->wbuffer[64 - 8]) = t;
                 }
@@ -470,49 +86,6 @@ void FAST_FUNC sha1_end(sha1_ctx_t *ctx, void *resbuf)
                         break;
                 bufpos = 0;
         }
-
-        bufpos = (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 < bufpos; ++i)
-                        ctx->hash[i] = SWAP_BE32(ctx->hash[i]);
-        }
-        memcpy(resbuf, ctx->hash, sizeof(ctx->hash[0]) * bufpos);
-}
-
-void FAST_FUNC sha512_end(sha512_ctx_t *ctx, void *resbuf)
-{
-        unsigned bufpos = ctx->total64[0] & 127;
-
-        /* Pad the buffer to the next 128-byte boundary with 0x80,0,0,0... */
-        ctx->wbuffer[bufpos++] = 0x80;
-
-        while (1) {
-                unsigned remaining = 128 - bufpos;
-                memset(ctx->wbuffer + bufpos, 0, remaining);
-                if (remaining >= 16) {
-                        /* Store the 128-bit counter of bits in the buffer in BE format */
-                        uint64_t t;
-                        t = ctx->total64[0] << 3;
-                        t = SWAP_BE64(t);
-                        *(uint64_t *) (&ctx->wbuffer[128 - 8]) = t;
-                        t = (ctx->total64[1] << 3) | (ctx->total64[0] >> 61);
-                        t = SWAP_BE64(t);
-                        *(uint64_t *) (&ctx->wbuffer[128 - 16]) = t;
-                }
-                sha512_process_block128(ctx);
-                if (remaining >= 16)
-                        break;
-                bufpos = 0;
-        }
-
-        if (BB_LITTLE_ENDIAN) {
-                unsigned i;
-                for (i = 0; i < ARRAY_SIZE(ctx->hash); ++i)
-                        ctx->hash[i] = SWAP_BE64(ctx->hash[i]);
-        }
-        memcpy(resbuf, ctx->hash, sizeof(ctx->hash));
 }
 
 
@@ -539,18 +112,6 @@ void FAST_FUNC sha512_end(sha512_ctx_t *ctx, void *resbuf)
 # define MD5_SIZE_VS_SPEED CONFIG_MD5_SIZE_VS_SPEED
 #endif
 
-/* Initialize structure containing state of computation.
- * (RFC 1321, 3.3: Step 3)
- */
-void FAST_FUNC md5_begin(md5_ctx_t *ctx)
-{
-        ctx->A = 0x67452301;
-        ctx->B = 0xefcdab89;
-        ctx->C = 0x98badcfe;
-        ctx->D = 0x10325476;
-        ctx->total64 = 0;
-}
-
 /* These are the four functions used in the four steps of the MD5 algorithm
  * and defined in the RFC 1321.  The first function is a little bit optimized
  * (as found in Colin Plumbs public domain implementation).
@@ -566,7 +127,7 @@ void FAST_FUNC md5_begin(md5_ctx_t *ctx)
 #define FI(b, c, d) (c ^ (b | ~d))
 
 /* Hash a single block, 64 bytes long and 4-byte aligned */
-static void md5_process_block64(md5_ctx_t *ctx)
+static void FAST_FUNC md5_process_block64(md5_ctx_t *ctx)
 {
 #if MD5_SIZE_VS_SPEED > 0
         /* Before we start, one word to the strange constants.
@@ -605,10 +166,10 @@ static void md5_process_block64(md5_ctx_t *ctx)
         };
 #endif
         uint32_t *words = (void*) ctx->wbuffer;
-        uint32_t A = ctx->A;
-        uint32_t B = ctx->B;
-        uint32_t C = ctx->C;
-        uint32_t D = ctx->D;
+        uint32_t A = ctx->hash[0];
+        uint32_t B = ctx->hash[1];
+        uint32_t C = ctx->hash[2];
+        uint32_t D = ctx->hash[3];
 
 #if MD5_SIZE_VS_SPEED >= 2  /* 2 or 3 */
 
@@ -705,10 +266,10 @@ static void md5_process_block64(md5_ctx_t *ctx)
         }
 # endif
         /* Add checksum to the starting values */
-        ctx->A += A;
-        ctx->B += B;
-        ctx->C += C;
-        ctx->D += D;
+        ctx->hash[0] += A;
+        ctx->hash[1] += B;
+        ctx->hash[2] += C;
+        ctx->hash[3] += D;
 
 #else  /* MD5_SIZE_VS_SPEED == 0 or 1 */
 
@@ -860,10 +421,10 @@ static void md5_process_block64(md5_ctx_t *ctx)
 # undef OP
 # endif
         /* Add checksum to the starting values */
-        ctx->A = A_save + A;
-        ctx->B = B_save + B;
-        ctx->C = C_save + C;
-        ctx->D = D_save + D;
+        ctx->hash[0] = A_save + A;
+        ctx->hash[1] = B_save + B;
+        ctx->hash[2] = C_save + C;
+        ctx->hash[3] = D_save + D;
 #endif
 }
 #undef FF
@@ -871,37 +432,404 @@ static void md5_process_block64(md5_ctx_t *ctx)
 #undef FH
 #undef FI
 
-/* Feed data through a temporary buffer to call md5_hash_aligned_block()
- * with chunks of data that are 4-byte aligned and a multiple of 64 bytes.
- * This function's internal buffer remembers previous data until it has 64
- * bytes worth to pass on.  Call md5_end() to flush this buffer. */
+/* Initialize structure containing state of computation.
+ * (RFC 1321, 3.3: Step 3)
+ */
+void FAST_FUNC md5_begin(md5_ctx_t *ctx)
+{
+        ctx->hash[0] = 0x67452301;
+        ctx->hash[1] = 0xefcdab89;
+        ctx->hash[2] = 0x98badcfe;
+        ctx->hash[3] = 0x10325476;
+        ctx->total64 = 0;
+        ctx->process_block = md5_process_block64;
+}
+
+/* Used also for sha1 and sha256 */
 void FAST_FUNC md5_hash(md5_ctx_t *ctx, const void *buffer, size_t len)
 {
-        unsigned bufpos = ctx->total64 & 63;
+        common64_hash(ctx, buffer, len);
+}
+
+/* Process the remaining bytes in the buffer and put result from CTX
+ * in first 16 bytes following RESBUF.  The result is always in little
+ * endian byte order, so that a byte-wise output yields to the wanted
+ * ASCII representation of the message digest.
+ */
+void FAST_FUNC md5_end(md5_ctx_t *ctx, void *resbuf)
+{
+        /* MD5 stores total in LE, need to swap on BE arches: */
+        common64_end(ctx, /*swap_needed:*/ BB_BIG_ENDIAN);
+
+        /* The MD5 result is in little endian byte order */
+#if BB_BIG_ENDIAN
+        ctx->hash[0] = SWAP_LE32(ctx->hash[0]);
+        ctx->hash[1] = SWAP_LE32(ctx->hash[1]);
+        ctx->hash[2] = SWAP_LE32(ctx->hash[2]);
+        ctx->hash[3] = SWAP_LE32(ctx->hash[3]);
+#endif
+        memcpy(resbuf, ctx->hash, sizeof(ctx->hash[0]) * 4);
+}
+
+
+/*
+ * Based on shasum from http://www.netsw.org/crypto/hash/
+ * Majorly hacked up to use Dr Brian Gladman's sha1 code
+ *
+ * Copyright (C) 2002 Dr Brian Gladman <brg@gladman.me.uk>, Worcester, UK.
+ * Copyright (C) 2003 Glenn L. McGrath
+ * Copyright (C) 2003 Erik Andersen
+ *
+ * Licensed under GPLv2 or later, see file LICENSE in this source tree.
+ *
+ * ---------------------------------------------------------------------------
+ * 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
+ *
+ * ---------------------------------------------------------------------------
+ *
+ * SHA256 and SHA512 parts are:
+ * Released into the Public Domain by Ulrich Drepper <drepper@redhat.com>.
+ * 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.
+ */
+
+static void FAST_FUNC sha1_process_block64(sha1_ctx_t *ctx)
+{
+        unsigned t;
+        uint32_t W[80], a, b, c, d, e;
+        const uint32_t *words = (uint32_t*) ctx->wbuffer;
+
+        for (t = 0; t < 16; ++t)
+                W[t] = SWAP_BE32(words[t]);
+        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];
+        b = ctx->hash[1];
+        c = ctx->hash[2];
+        d = ctx->hash[3];
+        e = ctx->hash[4];
+
+#undef ch
+#undef parity
+#undef maj
+#undef rnd
+#define ch(x,y,z)        ((z) ^ ((x) & ((y) ^ (z))))
+#define parity(x,y,z)    ((x) ^ (y) ^ (z))
+#define maj(x,y,z)       (((x) & (y)) | ((z) & ((x) | (y))))
+/* A normal version as set out in the FIPS.  */
+#define rnd(f,k) \
+        do { \
+                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 (t = 0; t < 20; ++t)
+                rnd(ch, 0x5a827999);
+
+        for (/*t = 20*/; t < 40; ++t)
+                rnd(parity, 0x6ed9eba1);
+
+        for (/*t = 40*/; t < 60; ++t)
+                rnd(maj, 0x8f1bbcdc);
+
+        for (/*t = 60*/; t < 80; ++t)
+                rnd(parity, 0xca62c1d6);
+#undef ch
+#undef parity
+#undef maj
+#undef rnd
+
+        ctx->hash[0] += a;
+        ctx->hash[1] += b;
+        ctx->hash[2] += c;
+        ctx->hash[3] += d;
+        ctx->hash[4] += e;
+}
+
+/* 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
+};
+
+#undef Ch
+#undef Maj
+#undef S0
+#undef S1
+#undef R0
+#undef R1
+
+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.  */
+#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 S1(x) (rotr32(x, 6) ^ rotr32(x, 11) ^ rotr32(x, 25))
+#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] = SWAP_BE32(words[t]);
+        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->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;
+}
+
+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 S1(x) (rotr64(x, 14) ^ rotr64(x, 18) ^ rotr64(x, 41))
+#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] = SWAP_BE64(words[t]);
+        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->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->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[] = {
+        0,
+        0,
+        0x6a09e667,
+        0xbb67ae85,
+        0x3c6ef372,
+        0xa54ff53a,
+        0x510e527f,
+        0x9b05688c,
+        0x1f83d9ab,
+        0x5be0cd19,
+};
+static const uint32_t init512_lo[] = {
+        0,
+        0,
+        0xf3bcc908,
+        0x84caa73b,
+        0xfe94f82b,
+        0x5f1d36f1,
+        0xade682d1,
+        0x2b3e6c1f,
+        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->total64, init256, sizeof(init256));
+        /*ctx->total64 = 0; - done by prepending two 32-bit zeros to init256 */
+        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;
+        /* Two extra iterations zero out ctx->total64[2] */
+        uint64_t *tp = ctx->total64;
+        for (i = 0; i < 2+8; i++)
+                tp[i] = ((uint64_t)(init256[i]) << 32) + init512_lo[i];
+        /*ctx->total64[0] = ctx->total64[1] = 0; - already done */
+}
+
+void FAST_FUNC sha512_hash(sha512_ctx_t *ctx, const void *buffer, size_t len)
+{
+        unsigned bufpos = ctx->total64[0] & 127;
         unsigned remaining;
 
-        /* RFC 1321 specifies the possible length of the file up to 2^64 bits.
-         * Here we only track the number of bytes.  */
-        ctx->total64 += len;
+        /* 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]++;
 #if 0
-        remaining = 64 - bufpos;
+        remaining = 128 - bufpos;
 
         /* Hash whole blocks */
         while (len >= remaining) {
                 memcpy(ctx->wbuffer + bufpos, buffer, remaining);
                 buffer = (const char *)buffer + remaining;
                 len -= remaining;
-                remaining = 64;
+                remaining = 128;
                 bufpos = 0;
-                md5_process_block64(ctx);
+                sha512_process_block128(ctx);
         }
 
         /* Save last, partial blosk */
         memcpy(ctx->wbuffer + bufpos, buffer, len);
 #else
-        /* Tiny bit smaller code */
         while (1) {
-                remaining = 64 - bufpos;
+                remaining = 128 - bufpos;
                 if (remaining > len)
                         remaining = len;
                 /* Copy data into aligned buffer */
@@ -909,54 +837,65 @@ void FAST_FUNC md5_hash(md5_ctx_t *ctx, const void *buffer, size_t len)
                 len -= remaining;
                 buffer = (const char *)buffer + remaining;
                 bufpos += remaining;
-                /* clever way to do "if (bufpos != 64) break; ... ; bufpos = 0;" */
-                bufpos -= 64;
+                /* clever way to do "if (bufpos != 128) break; ... ; bufpos = 0;" */
+                bufpos -= 128;
                 if (bufpos != 0)
                         break;
                 /* Buffer is filled up, process it */
-                md5_process_block64(ctx);
+                sha512_process_block128(ctx);
                 /*bufpos = 0; - already is */
         }
 #endif
 }
 
-/* Process the remaining bytes in the buffer and put result from CTX
- * in first 16 bytes following RESBUF.  The result is always in little
- * endian byte order, so that a byte-wise output yields to the wanted
- * ASCII representation of the message digest.
- */
-void FAST_FUNC md5_end(md5_ctx_t *ctx, void *resbuf)
+/* Used also for sha256 */
+void FAST_FUNC sha1_end(sha1_ctx_t *ctx, void *resbuf)
 {
-        unsigned bufpos = ctx->total64 & 63;
-        /* Pad the buffer to the next 64-byte boundary with 0x80,0,0,0... */
+        unsigned hash_size;
+
+        /* SHA stores total in BE, need to swap on LE arches: */
+        common64_end(ctx, /*swap_needed:*/ BB_LITTLE_ENDIAN);
+
+        hash_size = (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 < hash_size; ++i)
+                        ctx->hash[i] = SWAP_BE32(ctx->hash[i]);
+        }
+        memcpy(resbuf, ctx->hash, sizeof(ctx->hash[0]) * hash_size);
+}
+
+void FAST_FUNC sha512_end(sha512_ctx_t *ctx, void *resbuf)
+{
+        unsigned bufpos = ctx->total64[0] & 127;
+
+        /* Pad the buffer to the next 128-byte boundary with 0x80,0,0,0... */
         ctx->wbuffer[bufpos++] = 0x80;
 
-        /* This loop iterates either once or twice, no more, no less */
         while (1) {
-                unsigned remaining = 64 - bufpos;
+                unsigned remaining = 128 - bufpos;
                 memset(ctx->wbuffer + bufpos, 0, remaining);
-                /* Do we have enough space for the length count? */
-                if (remaining >= 8) {
-                        /* Store the 64-bit counter of bits in the buffer in LE format */
-                        uint64_t t = ctx->total64 << 3;
-                        t = SWAP_LE64(t);
-                        /* wbuffer is suitably aligned for this */
-                        *(uint64_t *) (&ctx->wbuffer[64 - 8]) = t;
+                if (remaining >= 16) {
+                        /* Store the 128-bit counter of bits in the buffer in BE format */
+                        uint64_t t;
+                        t = ctx->total64[0] << 3;
+                        t = SWAP_BE64(t);
+                        *(uint64_t *) (&ctx->wbuffer[128 - 8]) = t;
+                        t = (ctx->total64[1] << 3) | (ctx->total64[0] >> 61);
+                        t = SWAP_BE64(t);
+                        *(uint64_t *) (&ctx->wbuffer[128 - 16]) = t;
                 }
-                md5_process_block64(ctx);
-                if (remaining >= 8)
+                sha512_process_block128(ctx);
+                if (remaining >= 16)
                         break;
                 bufpos = 0;
         }
 
-        /* The MD5 result is in little endian byte order.
-         * We (ab)use the fact that A-D are consecutive in memory.
-         */
-#if BB_BIG_ENDIAN
-        ctx->A = SWAP_LE32(ctx->A);
-        ctx->B = SWAP_LE32(ctx->B);
-        ctx->C = SWAP_LE32(ctx->C);
-        ctx->D = SWAP_LE32(ctx->D);
-#endif
-        memcpy(resbuf, &ctx->A, sizeof(ctx->A) * 4);
+        if (BB_LITTLE_ENDIAN) {
+                unsigned i;
+                for (i = 0; i < ARRAY_SIZE(ctx->hash); ++i)
+                        ctx->hash[i] = SWAP_BE64(ctx->hash[i]);
+        }
+        memcpy(resbuf, ctx->hash, sizeof(ctx->hash));
 }
diff --git a/libbb/safe_strncpy.c b/libbb/safe_strncpy.c
index 8eb6a014f..5eb0db0bd 100644
--- a/libbb/safe_strncpy.c
+++ b/libbb/safe_strncpy.c
@@ -20,8 +20,13 @@ char* FAST_FUNC safe_strncpy(char *dst, const char *src, size_t size)
 /* Like strcpy but can copy overlapping strings. */
 void FAST_FUNC overlapping_strcpy(char *dst, const char *src)
 {
-        while ((*dst = *src) != '\0') {
-                dst++;
-                src++;
+        /* Cheap optimization for dst == src case -
+         * better to have it here than in many callers.
+         */
+        if (dst != src) {
+                while ((*dst = *src) != '\0') {
+                        dst++;
+                        src++;
+                }
         }
 }