Daiki Ueno [Mon, 5 Oct 2020 15:44:30 +0000 (17:44 +0200)]
fips: add self-tests for TLS-PRF
FIPS140-2 IG D.8 mandates self-tests on approved KDF algorithms. As
the guidance only requires to run a single instance of each KDF
mechanism, this only exercises TLS1.2 PRF with HMAC-SHA-256 as the
underlying MAC algorithm.
Daiki Ueno [Tue, 6 Oct 2020 09:54:21 +0000 (11:54 +0200)]
fips: use larger prime for DH self-tests
According to FIPS140-2 IG 7.5, the minimum key size of FFC through
2030 is defined as 2048 bits. This updates the relevant self-test
using ffdhe3072 defined in RFC 7919.
Daiki Ueno [Mon, 5 Oct 2020 14:59:50 +0000 (16:59 +0200)]
fips: add self-tests for PBKDF2
FIPS140-2 IG D.8 mandates self-tests on approved KDF algorithms. As
the guidance only requires running a single instance of each KDF
mechanism, this only exercises PBKDF2 with HMAC-SHA-256 as the
underlying MAC algorithm.
Daiki Ueno [Mon, 5 Oct 2020 14:12:46 +0000 (16:12 +0200)]
fips: add self-tests for HKDF
FIPS140-2 IG D.8 mandates self-test on approved KDF algorithms. As
the guidance only requires running a single instance of each KDF
mechanism, this only exercises HKDF-Extract and HKDF-Expand operations
with HMAC-SHA-256 as the underlying MAC.
Although HKDF is non-approved, it would be sensible to do that as it
will be approved in FIPS140-3.
Modify the variables _gnutls_sha_padlock and _gnutls_sha_padlock_nano.
Add a comment for detecting CPU. Modify the indentation. Delete
initialization etc.
x86:add detection of instruction set on Zhaoxin CPU
Add detection of extended instruction set on Zhaoxin cpu,e.g:ssse3,sha,
etc. Set the priority of the algorithm according to the benchmark
test result on Zhaoxin cpu.
In the case of setting environment variables, AVX cannot be detected
correctly. Because only MOVBE is added to variable _gnutls_x86_cpuid_s,
there is no OSXSAVE. And according to the intel manual, using AVX
does not need to detect FMA.
This avoids -fanalyzer false-positive in GCC 10:
https://bugzilla.redhat.com/show_bug.cgi?id=1878600
as well as the cppcheck warning:
"variableScope:lib/inih/ini.c:99,style,The scope of the variable 'start' can be reduced."
The function was not really useful because _gnutls_free_datum()
has a NULL check as in free(). This also makes GCC 10 happy if
-Warray-bounds=2 is specified:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=96984
The nanosleep wrapper in glibc has changed the implementation using
the clock_nanosleep syscall:
https://sourceware.org/git/?p=glibc.git;a=commit;h=3537ecb49cf7177274607004c562d6f9ecc99474
Modifies P_hash() to hash the seed and label separately
Thereby not restricting the implementation of prf to MAX_SEED_SIZE
MAX_SEED_SIZE is not used anymore
This brings in the build fixes of parse-datetime module:
https://lists.gnu.org/archive/html/bug-gnulib/2020-07/msg00178.html
https://lists.gnu.org/archive/html/bug-gnulib/2020-08/msg00001.html
https://lists.gnu.org/archive/html/bug-gnulib/2020-09/msg00046.html
Daiki Ueno [Sat, 22 Aug 2020 15:19:39 +0000 (17:19 +0200)]
handshake: reject no_renegotiation alert if handshake is incomplete
If the initial handshake is incomplete and the server sends a
no_renegotiation alert, the client should treat it as a fatal error
even if its level is warning. Otherwise the same handshake
state (e.g., DHE parameters) are reused in the next gnutls_handshake
call, if it is called in the loop idiom:
do {
ret = gnutls_handshake(session);
} while (ret < 0 && gnutls_error_is_fatal(ret) == 0);
src/cli: adds new option '--ca-auto-retrieve' that can be used with gnutls-cli to
automatically download missing intermediate CAs in a certificate chain
lib/cred-cert.c : adds set and get APIs to get user data in the
gnutls_x509_trust_list_set_getissuer_function() callback.
Nikolay Sivov [Tue, 1 Sep 2020 16:29:20 +0000 (19:29 +0300)]
Fix optional parameters counter when importing RSA private keys.
Currently gnutls_privkey_import_rsa_raw() allows 3 last arguments to be omitted,
key fixup logic however checks for 3 missing arguments when updating coefficient 'u'
but then asserts when updating exponents 'e1' and 'e2' assuming only 2 parameters
are missing at that point.
Signed-off-by: Nikolay Sivov <nsivov@codeweavers.com>
Daiki Ueno [Thu, 13 Aug 2020 16:17:08 +0000 (18:17 +0200)]
gnutls_aead_cipher_decrypt: check output buffer size before writing
While the documentation of gnutls_aead_cipher_decrypt indicates that
the inout argument ptext_len initially holds the size that
sufficiently fits the expected output size, there was no runtime check
on that. This makes the interface robuster against misuses.
Daiki Ueno [Sun, 16 Aug 2020 09:43:35 +0000 (11:43 +0200)]
handshake: check TLS version against modified server priorities
The server needs to take into account of multiple factors when
determining the TLS protocol version actually being used:
- the legacy version
- "supported_versions" extension
- user_hello_func that may modify the server's priorities
Only after that it can check whether the TLS version is enabled in the
server's priorities.
Fiona Klute [Sun, 16 Aug 2020 12:05:44 +0000 (14:05 +0200)]
Fix parser output in tests/cert-tests/data/gost-cert-nogost.pem
When building without GOST support parsing a GOST certificate must
return an "error importing public key" message instead of key
details. This change makes tests/cert-tests/pem-decoding pass for
builds with --disable-gost.
Daiki Ueno [Fri, 14 Aug 2020 05:27:40 +0000 (07:27 +0200)]
cert-session: check OCSP error responses
If the OCSP responder returns an error code, such as tryLater, we
can't proceed to examine the response bytes. In that case, just skip
the check unless the stapling is mandatory on this certificate.
Daiki Ueno [Thu, 4 Jun 2020 14:42:07 +0000 (16:42 +0200)]
_gnutls_fips_mode_enabled: treat selftest failure as FIPS disabled
Previously gnutls_fips140_mode_enabled() returned true, even after
selftests have failed and the library state has switched to error.
While later calls to crypto operations fails, it would be more
convenient to have a function to detect that state.
Daiki Ueno [Wed, 12 Aug 2020 05:29:30 +0000 (07:29 +0200)]
serv, cli: ensure that invalid flag is always set
According to the documentation, the GNUTLS_CERT_INVALID flag must
always be set in case of verification failure, together with the flag
indicating the actual error cause.
projects / thirdparty / gnutls.git / log
