/*- * Copyright (c) 2012 The University of Oslo * Copyright (c) 2012-2014 Dag-Erling Smørgrav * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. The name of the author may not be used to endorse or promote * products derived from this software without specific prior written * permission. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include "cryb/impl.h" #include #include #include #include #include #include #include "t.h" #if WITH_OPENSSL #include #define SHA224_DIGEST_LEN SHA224_DIGEST_LENGTH static void t_sha224_complete(const void *msg, size_t msglen, uint8_t *digest) { SHA256_CTX ctx; SHA224_Init(&ctx); SHA224_Update(&ctx, msg, msglen); SHA224_Final(digest, &ctx); } #else #include #define t_sha224_complete(msg, msglen, digest) \ sha224_complete(msg, msglen, digest) #endif /* * Test vectors from NIST CSRC * http://csrc.nist.gov/groups/ST/toolkit/examples.html */ static struct t_vector { const char *desc; const char *msg; const uint8_t digest[SHA224_DIGEST_LEN]; } t_sha224_vectors[] = { { "zero-length message", "", { 0xd1, 0x4a, 0x02, 0x8c, 0x2a, 0x3a, 0x2b, 0xc9, 0x47, 0x61, 0x02, 0xbb, 0x28, 0x82, 0x34, 0xc4, 0x15, 0xa2, 0xb0, 0x1f, 0x82, 0x8e, 0xa6, 0x2a, 0xc5, 0xb3, 0xe4, 0x2f, } }, { "One-block message", "abc", { 0x23, 0x09, 0x7d, 0x22, 0x34, 0x05, 0xd8, 0x22, 0x86, 0x42, 0xa4, 0x77, 0xbd, 0xa2, 0x55, 0xb3, 0x2a, 0xad, 0xbc, 0xe4, 0xbd, 0xa0, 0xb3, 0xf7, 0xe3, 0x6c, 0x9d, 0xa7, } }, { /* * This message is *just* long enough to necessitate a * second block, which consists entirely of padding. */ "Two-block message", "abcdbcdecdefdefgefghfghighijhijkijkljklmklmnlmnomnopnopq", { 0x75, 0x38, 0x8b, 0x16, 0x51, 0x27, 0x76, 0xcc, 0x5d, 0xba, 0x5d, 0xa1, 0xfd, 0x89, 0x01, 0x50, 0xb0, 0xc6, 0x45, 0x5c, 0xb4, 0xf5, 0x8b, 0x19, 0x52, 0x52, 0x25, 0x25, } }, { /* * 1,000,000 x 'a', filled in by t_prepare() */ "Long message", NULL, { 0x20, 0x79, 0x46, 0x55, 0x98, 0x0c, 0x91, 0xd8, 0xbb, 0xb4, 0xc1, 0xea, 0x97, 0x61, 0x8a, 0x4b, 0xf0, 0x3f, 0x42, 0x58, 0x19, 0x48, 0xb2, 0xee, 0x4e, 0xe7, 0xad, 0x67, }, }, { /* * One of the MD5 test vectors, included for the "short * update" test. */ "\"1234567890\"x8", "1234567890123456789012345678901234567890" "1234567890123456789012345678901234567890", { 0xb5, 0x0a, 0xec, 0xbe, 0x4e, 0x9b, 0xb0, 0xb5, 0x7b, 0xc5, 0xf3, 0xae, 0x76, 0x0a, 0x8e, 0x01, 0xdb, 0x24, 0xf2, 0x03, 0xfb, 0x3c, 0xdc, 0xd1, 0x31, 0x48, 0x04, 0x6e, }, }, }; /* * Unit test: compute the SHA224 sum of the specified string and compare it * to the expected result. */ static int t_sha224_vector(char **desc CRYB_UNUSED, void *arg) { struct t_vector *vector = (struct t_vector *)arg; uint8_t digest[SHA224_DIGEST_LEN]; char *msg; if (vector->msg) { t_sha224_complete(vector->msg, strlen(vector->msg), digest); } else { /* special case for FIPS test vector 3 */ if ((msg = malloc(1000000)) == NULL) err(1, "malloc()"); memset(msg, 'a', 1000000); t_sha224_complete(msg, 1000000, digest); free(msg); } return (t_compare_mem(vector->digest, digest, SHA224_DIGEST_LEN)); } #if !defined(WITH_OPENSSL) && !defined(WITH_RSAREF) /* * Various corner cases and error conditions */ static int t_sha224_short_updates(char **desc CRYB_UNUSED, void *arg) { struct t_vector *vector = (struct t_vector *)arg; uint8_t digest[SHA224_DIGEST_LEN]; sha224_ctx ctx; int i, len; sha224_init(&ctx); len = strlen(vector->msg); for (i = 0; i + 5 < len; i += 5) sha224_update(&ctx, vector->msg + i, 5); sha224_update(&ctx, vector->msg + i, len - i); sha224_final(&ctx, digest); return (memcmp(digest, vector->digest, SHA224_DIGEST_LEN) == 0); } #endif /* * Performance test: measure the time spent computing the SHA224 sum of a * message of the specified length. */ #define T_PERF_ITERATIONS 1000 static int t_sha224_perf(char **desc, void *arg) { struct timespec ts, te; unsigned long ns; uint8_t digest[SHA224_DIGEST_LEN]; char *msg, *comment; size_t msglen = *(size_t *)arg; if ((msg = calloc(1, msglen)) == NULL) err(1, "calloc()"); clock_gettime(CLOCK_MONOTONIC, &ts); for (int i = 0; i < T_PERF_ITERATIONS; ++i) t_sha224_complete(msg, msglen, digest); clock_gettime(CLOCK_MONOTONIC, &te); free(msg); ns = te.tv_sec * 1000000000LU + te.tv_nsec; ns -= ts.tv_sec * 1000000000LU + ts.tv_nsec; asprintf(&comment, "%zu bytes: %d iterations in %'lu ns", msglen, T_PERF_ITERATIONS, ns); if (comment == NULL) err(1, "asprintf()"); *desc = comment; return (1); } /*************************************************************************** * Boilerplate */ int t_prepare(int argc, char *argv[]) { int i, n; (void)argc; (void)argv; n = sizeof t_sha224_vectors / sizeof t_sha224_vectors[0]; for (i = 0; i < n; ++i) t_add_test(t_sha224_vector, &t_sha224_vectors[i], t_sha224_vectors[i].desc); #if !defined(WITH_OPENSSL) && !defined(WITH_RSAREF) /* * Run test vector 5 (md5 test vector 7, which is 80 characters * long) 5 characters at a time. This tests a) appending data to * an underfull block and b) appending more data to an underfull * block than it has room for (since 64 % 5 != 0). Test vector 4 * and 5 already exercised the code path for computing a block * directly from source (without copying it in), and all the test * vectors except vector 1 exercised the general case of copying a * small amount of data in without crossing the block boundary. */ t_add_test(t_sha224_short_updates, &t_sha224_vectors[4], "multiple short updates"); #endif if (t_str_is_true(getenv("CRYB_PERFTEST"))) { static size_t one = 1, thousand = 1000, million = 1000000; t_add_test(t_sha224_perf, &one, "performance test (1 byte)"); t_add_test(t_sha224_perf, &thousand, "performance test (1,000 bytes)"); t_add_test(t_sha224_perf, &million, "performance test (1,000,000 bytes)"); } return (0); } void t_cleanup(void) { }