nettle-bugs

nettle-bugs@lists.lysator.liu.se

1 participants
2197 discussions

[PATCH] Add benchmarking for HMAC functions
by Dmitry Eremin-Solenikov 30 Jul '18

30 Jul '18

In preparation of changing internal HMAC interface add benchmarking for HMAC functions. Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov(a)gmail.com> --- examples/nettle-benchmark.c | 154 ++++++++++++++++++++++++++++++++++++ 1 file changed, 154 insertions(+) diff --git a/examples/nettle-benchmark.c b/examples/nettle-benchmark.c index 5a88e72dd678..101ba7d4abcc 100644 --- a/examples/nettle-benchmark.c +++ b/examples/nettle-benchmark.c @@ -67,6 +67,7 @@ #include "umac.h" #include "cmac.h" #include "poly1305.h" +#include "hmac.h" #include "nettle-meta.h" #include "nettle-internal.h" @@ -476,6 +477,147 @@ time_poly1305_aes(void) time_function(bench_hash, &info)); } +struct bench_hmac_info +{ + void *ctx; + nettle_hash_update_func *update; + nettle_hash_digest_func *digest; + size_t length; + size_t digest_length; + const uint8_t *data; +}; + +static void +bench_hmac(void *arg) +{ + struct bench_hmac_info *info = arg; + uint8_t digest[NETTLE_MAX_HASH_DIGEST_SIZE]; + size_t pos, length; + + length = info->length; + for (pos = 0; pos < BENCH_BLOCK; pos += length) + { + size_t single = pos + length < BENCH_BLOCK ? + length : + BENCH_BLOCK - pos; + info->update(info->ctx, single, info->data + pos); + info->digest(info->ctx, info->digest_length, digest); + } +} + +static const struct +{ + size_t length; + const char *msg; +} hmac_tests[] = { + { 64, "64 bytes" }, + { 256, "256 bytes" }, + { 1024, "1024 bytes" }, + { 4096, "4096 bytes" }, + { BENCH_BLOCK, "single msg" }, + { 0, NULL }, +}; + +static void +time_hmac_md5(void) +{ + static uint8_t data[BENCH_BLOCK]; + struct bench_hmac_info info; + struct hmac_md5_ctx md5_ctx; + unsigned int pos; + + init_data(data); + info.data = data; + + hmac_md5_set_key(&md5_ctx, MD5_BLOCK_SIZE, data); + info.ctx = &md5_ctx; + info.update = (nettle_hash_update_func *) hmac_md5_update; + info.digest = (nettle_hash_digest_func *) hmac_md5_digest; + info.digest_length = MD5_DIGEST_SIZE; + + for (pos = 0; hmac_tests[pos].length != 0; pos++) + { + info.length = hmac_tests[pos].length; + display("hmac-md5", hmac_tests[pos].msg, MD5_BLOCK_SIZE, + time_function(bench_hmac, &info)); + } +} + +static void +time_hmac_sha1(void) +{ + static uint8_t data[BENCH_BLOCK]; + struct bench_hmac_info info; + struct hmac_sha1_ctx sha1_ctx; + unsigned int pos; + + init_data(data); + info.data = data; + + hmac_sha1_set_key(&sha1_ctx, SHA1_BLOCK_SIZE, data); + info.ctx = &sha1_ctx; + info.update = (nettle_hash_update_func *) hmac_sha1_update; + info.digest = (nettle_hash_digest_func *) hmac_sha1_digest; + info.digest_length = SHA1_DIGEST_SIZE; + + for (pos = 0; hmac_tests[pos].length != 0; pos++) + { + info.length = hmac_tests[pos].length; + display("hmac-sha1", hmac_tests[pos].msg, SHA1_BLOCK_SIZE, + time_function(bench_hmac, &info)); + } +} + +static void +time_hmac_sha256(void) +{ + static uint8_t data[BENCH_BLOCK]; + struct bench_hmac_info info; + struct hmac_sha256_ctx sha256_ctx; + unsigned int pos; + + init_data(data); + info.data = data; + + hmac_sha256_set_key(&sha256_ctx, SHA256_BLOCK_SIZE, data); + info.ctx = &sha256_ctx; + info.update = (nettle_hash_update_func *) hmac_sha256_update; + info.digest = (nettle_hash_digest_func *) hmac_sha256_digest; + info.digest_length = SHA256_DIGEST_SIZE; + + for (pos = 0; hmac_tests[pos].length != 0; pos++) + { + info.length = hmac_tests[pos].length; + display("hmac-sha256", hmac_tests[pos].msg, SHA256_BLOCK_SIZE, + time_function(bench_hmac, &info)); + } +} + +static void +time_hmac_sha512(void) +{ + static uint8_t data[BENCH_BLOCK]; + struct bench_hmac_info info; + struct hmac_sha512_ctx sha512_ctx; + unsigned int pos; + + init_data(data); + info.data = data; + + hmac_sha512_set_key(&sha512_ctx, SHA512_BLOCK_SIZE, data); + info.ctx = &sha512_ctx; + info.update = (nettle_hash_update_func *) hmac_sha512_update; + info.digest = (nettle_hash_digest_func *) hmac_sha512_digest; + info.digest_length = SHA512_DIGEST_SIZE; + + for (pos = 0; hmac_tests[pos].length != 0; pos++) + { + info.length = hmac_tests[pos].length; + display("hmac-sha512", hmac_tests[pos].msg, SHA512_BLOCK_SIZE, + time_function(bench_hmac, &info)); + } +} + static int prefix_p(const char *prefix, const char *s) { @@ -883,6 +1025,18 @@ main(int argc, char **argv) if (!alg || strstr(aeads[i]->name, alg)) time_aead(aeads[i]); + if (!alg || strstr ("hmac-md5", alg)) + time_hmac_md5(); + + if (!alg || strstr ("hmac-sha1", alg)) + time_hmac_sha1(); + + if (!alg || strstr ("hmac-sha256", alg)) + time_hmac_sha256(); + + if (!alg || strstr ("hmac-sha512", alg)) + time_hmac_sha512(); + optind++; } while (alg && argv[optind]); -- 2.18.0

2 1

[PATCH 0/2] Rework HMAC interface
by Dmitry Eremin-Solenikov 25 Jul '18

25 Jul '18

Hello colleagues, I gave a little thought to Niels' idea: 5. Revamp hmac and underlying hash functions with a separate state struct. Probably low priority, but it is a bit silly that, e.g., hmac_sha512_ctx includes three 128-byte large block buffers. I've implemented new approach using hmac2 prefix, but if you like this approach I can switch hmac2 prefix to just hmac and drop older API. -- With best wishes Dmitry

3 6

nettle symbols: improve map files
by Nikos Mavrogiannopoulos 16 Jul '18

16 Jul '18

Hi, This patch makes the exported symbols explicit in the map file. Furthermore it moves symbols only listed in internal headers in a special section which makes them valid only for testing. I've tested it with abi-compliance-checker and it detects the following missing from nettle: nettle_aeads, nettle_armors, nettle_ciphers, nettle_hashes and from hogweed the nettle_secp_*r1, which seem expected. Niels, I'm not sure however if that was your intention. Didn't you want to deprecate some of the _nettle symbols as well like _nettle_secp_256r1? These are however still present in the exported headers and unless we remove them, they seem to be part of the API and ABI. I also attach the script which generated the files in case you'd like to play with it. regards, Nikos

3 26

Xcode/CLT 10+ Patch
by Dominyk Tiller 18 Jun '18

18 Jun '18

Hey folks, With the obvious enough disclaimer that Xcode/CLT 10 are very much beta releases at this point and will remain so for several months yet, I've attached a diff to fix the only current build issue with the new developer tools available for macOS 10.13/10.14. The new Apple LLVM (10.0.0) ships with a new stack probing function, which embeds the symbol `____chkstk_darwin` into seemingly the vast majority of libraries & binaries, including both libnettle.a & libhogweed.a, which in turn currently blows up nettle's `make check` stage on the symbol check: ---------------------- PASS: sexp-conv PASS: pkcs1-conv PASS: nettle-pbkdf2 Exported symbols in libnettle.a, lacking the nettle prefix: 00000000000001b0 T ____chkstk_darwin 0000000000000230 T ____chkstk_darwin 0000000000000350 T ____chkstk_darwin 0000000000000360 T ____chkstk_darwin 0000000000000380 T ____chkstk_darwin FAIL: symbols PASS: dlopen ==================== 1 of 94 tests failed ==================== make[1]: *** [check] Error 1 make: *** [check] Error 2 ---------------------- The attached diff simply scopes out an exception to the check, with a short explanatory note. I appreciate fully y'all may decide it's too soon to be accepting patches for prerelease developer tools, but the chances of Apple removing the new function or disabling it by default seem pretty low. Let me know if anything further is needed. Thanks, Dominyk -- Sent from macOS. If you wish to communicate more securely my PGP Public Key is at: https://pgp.mit.edu/pks/lookup?search=0xE5F21DD98E4DF470&op=index

2 2

[PATCH] Add support for reading/writing integers in Little Endian format
by Dmitry Eremin-Solenikov 17 Jun '18

17 Jun '18

Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov(a)gmail.com> --- bignum.c | 35 +++++++++++++++++++++++++++++++++++ bignum.h | 17 +++++++++++++++++ testsuite/bignum-test.c | 29 +++++++++++++++++++++++++++++ 3 files changed, 81 insertions(+) diff --git a/bignum.c b/bignum.c index 335252876544..413a630cb3aa 100644 --- a/bignum.c +++ b/bignum.c @@ -133,12 +133,32 @@ nettle_mpz_get_str_256(size_t length, uint8_t *s, const mpz_t x) } } +void +nettle_mpz_get_str_256_u_le(size_t length, uint8_t *s, const mpz_t x) +{ + if (!length) + { + /* x must be zero */ + assert(!mpz_sgn(x)); + return; + } + + size_t count; + + assert(nettle_mpz_sizeinbase_256_u(x) <= length); + mpz_export(s, &count, -1, 1, 0, 0, x); + memset(s + count, 0, length - count); +} + /* Converting from strings */ /* mpz_import was introduced in GMP-4.1 */ #define nettle_mpz_from_octets(x, length, s) \ mpz_import((x), (length), 1, 1, 0, 0, (s)) +#define nettle_mpz_from_octets_le(x, length, s) \ + mpz_import((x), (length), -1, 1, 0, 0, (s)) + void nettle_mpz_set_str_256_u(mpz_t x, size_t length, const uint8_t *s) @@ -154,6 +174,21 @@ nettle_mpz_init_set_str_256_u(mpz_t x, nettle_mpz_from_octets(x, length, s); } +void +nettle_mpz_set_str_256_u_le(mpz_t x, + size_t length, const uint8_t *s) +{ + nettle_mpz_from_octets_le(x, length, s); +} + +void +nettle_mpz_init_set_str_256_u_le(mpz_t x, + size_t length, const uint8_t *s) +{ + mpz_init(x); + nettle_mpz_from_octets_le(x, length, s); +} + void nettle_mpz_set_str_256_s(mpz_t x, size_t length, const uint8_t *s) diff --git a/bignum.h b/bignum.h index 7871d843168c..42a97e63d6b6 100644 --- a/bignum.h +++ b/bignum.h @@ -68,11 +68,19 @@ nettle_mpz_sizeinbase_256_s(const mpz_t x); size_t nettle_mpz_sizeinbase_256_u(const mpz_t x); +#define nettle_mpz_sizeinbase_256_u_le nettle_mpz_sizeinbase_256_u + /* Writes an integer as length octets, using big endian byte order, * and two's complement for negative numbers. */ void nettle_mpz_get_str_256(size_t length, uint8_t *s, const mpz_t x); +/* Writes an integer as length octets, using big endian byte order, + * and unsigned number format. */ +void +nettle_mpz_get_str_256_u_le(size_t length, uint8_t *s, const mpz_t x); + + /* Reads a big endian, two's complement, integer. */ void nettle_mpz_set_str_256_s(mpz_t x, @@ -92,6 +100,15 @@ void nettle_mpz_init_set_str_256_u(mpz_t x, size_t length, const uint8_t *s); +/* Similar, but for little endian byte order. */ +void +nettle_mpz_set_str_256_u_le(mpz_t x, + size_t length, const uint8_t *s); + +void +nettle_mpz_init_set_str_256_u_le(mpz_t x, + size_t length, const uint8_t *s); + /* Returns a uniformly distributed random number 0 <= x < 2^n */ void nettle_mpz_random_size(mpz_t x, diff --git a/testsuite/bignum-test.c b/testsuite/bignum-test.c index 602554b53941..79fd92e2d37d 100644 --- a/testsuite/bignum-test.c +++ b/testsuite/bignum-test.c @@ -34,6 +34,30 @@ test_bignum(const char *hex, const struct tstring *base256) free(buf); } +static void +test_bignum_le(const char *hex, const struct tstring *base256) +{ + mpz_t a; + mpz_t b; + uint8_t *buf; + + mpz_init_set_str(a, hex, 16); + nettle_mpz_init_set_str_256_u_le(b, base256->length, base256->data); + + ASSERT(mpz_cmp(a, b) == 0); + + buf = xalloc(base256->length + 1); + memset(buf, 17, base256->length + 1); + + nettle_mpz_get_str_256_u_le(base256->length, buf, a); + ASSERT(MEMEQ(base256->length, buf, base256->data)); + + ASSERT(buf[base256->length] == 17); + + mpz_clear(a); mpz_clear(b); + free(buf); +} + static void test_size(long x, unsigned size) { @@ -86,6 +110,11 @@ test_main(void) test_bignum("-7fff", SHEX( "8001")); test_bignum("-8000", SHEX( "8000")); test_bignum("-8001", SHEX("ff7fff")); + + test_bignum_le("0", SHEX("00")); + test_bignum_le("010203040506", SHEX("060504030201")); + test_bignum_le("80010203040506", SHEX("06050403020180")); + test_bignum_le("01020304050680", SHEX("80060504030201")); #else /* !WITH_HOGWEED */ SKIP(); -- 2.17.1

2 2

[PATCH] Fix quoting in autoconf ifunc test
by Dmitry Eremin-Solenikov 17 Jun '18

17 Jun '18

* aclocal.m4 (NETTLE_CHECK_IFUNC): fix quoting so that AC_LINK_IFELSE/AC_TRY_LINK is defined outside of this test. Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov(a)gmail.com> --- aclocal.m4 | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/aclocal.m4 b/aclocal.m4 index 783dbc492b5c..772f68e8eb9d 100644 --- a/aclocal.m4 +++ b/aclocal.m4 @@ -650,7 +650,7 @@ AC_DEFUN([NETTLE_CHECK_IFUNC], [AC_REQUIRE([AC_PROG_CC]) AC_CACHE_CHECK([for ifunc support], nettle_cv_link_ifunc, - AC_LINK_IFELSE([AC_LANG_PROGRAM([ + [AC_LINK_IFELSE([AC_LANG_PROGRAM([ static int foo_imp(int x) { @@ -671,7 +671,7 @@ int foo (int x) __attribute__ ((ifunc("foo_resolv"))); ])], [nettle_cv_link_ifunc=yes], -[nettle_cv_link_ifunc=no])) +[nettle_cv_link_ifunc=no])]) AH_TEMPLATE([HAVE_LINK_IFUNC], [Define if compiler and linker supports __attribute__ ifunc]) if test "x$nettle_cv_link_ifunc" = xyes ; then AC_DEFINE(HAVE_LINK_IFUNC) -- 2.17.1

2 2

[Dominyk Tiller] Fwd: Xcode/CLT 10+ Patch
by nisse＠lysator.liu.se 14 Jun '18

14 Jun '18

-- Niels Möller. PGP-encrypted email is preferred. Keyid 368C6677. Internet email is subject to wholesale government surveillance.

1 0

API for new AEAD modes
by Nikos Mavrogiannopoulos 31 May '18

31 May '18

Hi, As it is now AEAD ciphers in nettle are supported with their own API. AES-CCM provides: ccm_aes128_set_key ccm_aes128_set_nonce ccm_aes128_update ccm_aes128_encrypt ccm_aes128_decrypt ccm_aes128_digest ccm_aes128_encrypt_message ccm_aes128_decrypt_message AES-GCM: gcm_aes128_set_key gcm_aes128_update gcm_aes128_set_iv gcm_aes128_encrypt gcm_aes128_decrypt gcm_aes128_digest chacha-poly1305: chacha_poly1305_set_key chacha_poly1305_set_nonce chacha_poly1305_update chacha_poly1305_encrypt chacha_poly1305_decrypt chacha_poly1305_digest ccm_aes128_set_nonce takes different parameters from the chacha and gcm equivalent function. Furthermore the AEAD modes for block ciphers include an abstract layer which can be re-used of other ciphers. For example the GCM mode abstraction is re-used for camellia as well. So overall there is a kind of "abstract" API followed with the set_key, set_nonce, update, encrypt/decrypt, however these have not always the same parameters as in the CCM case. I'm checking the AES-SIV-CMAC mode which cannot even be put in that abstraction as the MAC in SIV is the IV and thus the MAC-style API doesn't really suit it. My understanding is that AEAD modes according to RFC5116 need to have 4 inputs: key, nonce, associated data, plaintext so the CCM encrypt and decrypt message seem to be the closer to that API. Would it make sense to standardize on these, and only provide that API for AEAD modes? The reason I'm asking is because SIV could benefit of a very custom API as well because it can take advantage of multiple associated data, but in the end I believe AEAD is about simplicity. Providing a unique API per AEAD cipher seems to me quite contradictory to that goal. regards, Nikos

4 6

https://git.lysator.liu.se/ seems to be offline...
by Tim Rühsen 14 Apr '18

14 Apr '18

1 0

symbol versioning update
by Nikos Mavrogiannopoulos 03 Apr '18

03 Apr '18

Hi, When we initially introduced symbol versioning in nettle we bundled all symbols from the library in a single version. That means that new symbols added to a release like nettle_get_hashes() may cause issues like this: https://bugzilla.redhat.com/show_bug.cgi?id=1549190 The underlying issue is in rpm-based systems which detect which symbols exist in which library version based on the version. That way, when a dependency is tracked, the version with the right symbol will be retrieved. Debian-based systems work differently so that shouldn't affect it. To address these problems in gnutls, I follow these additional rules in to having a map file: https://gitlab.com/gnutls/gnutls/blob/master/CONTRIBUTING.md#symbol-and-lib… which in turn comes from libvirt: https://www.berrange.com/posts/2011/01/13/versioning-in-the-libvirt-library/ That of course adds overhead on releases. nettle doesn't necessarily need to follow that process, I send that issue in order to spark discussion on whether that's desirable to have. regards, Nikos

3 6

Jump to page:

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

nettle-bugs