If s->hit is set, s->session corresponds to a session created on
a previous connection, and is a data structure that is potentially
shared across other SSL objects. As such, there are thread-safety
issues with modifying the structure without taking its lock (and
of course all corresponding read accesses would also need to take
the lock as well), which have been observed to cause double-frees.
Regardless of thread-safety, the resumed session object is intended
to reflect parameters of the connection that created the session,
and modifying it to reflect the parameters from the current connection
is confusing. So, modifications to the session object during
ClientHello processing should only be performed on new connections,
i.e., those where s->hit is not set.
The code mostly got this right, providing such checks when processing
SNI and EC point formats, but the supported groups (formerly
supported curves) extension was missing it, which is fixed by this commit.
However, TLS 1.3 makes the suppported_groups extension mandatory
(when using (EC)DHE, which is the normal case), checking for the group
list in the key_share extension processing. But, TLS 1.3 only [0] supports
session tickets for session resumption, so the session object in question
is the output of d2i_SSL_SESSION(), and will not be shared across SSL
objects. Thus, it is safe to modify s->session for TLS 1.3 connections.
[0] A psk_find_session callback can also be used, but the restriction that
each callback execution must produce a distinct SSL_SESSION structure
can be documented when the psk_find_session callback documentation is
completed.
Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4123)
When we are using the internal cache we have to make a copy of the
session before removing it from the parent context's cache, since
we want our copy to still be resumable. However, SSL_CTX_remove_session()
just detaches the session from the SSL_CTX; it does not free the session.
So, we must call SSL_SESSION_free() ourselves before overwriting the
variable that we dup'd from.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4126)
In the generated HTML document, the `<pre>` tag is not closed. This patch also has a trivial code-style improvement, unrelated to the bug fix.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Kurt Roeckx <kurt@roeckx.be>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4088)
Code Health (Tuesday?): Parameters' names are not correct.
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4117)
This is a 'code health' commit to respond to this round of code health
Tuesday...
Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4099)
Building without the scrypt KDF is now possible, the OPENSSL_NO_SCRYPT
define is honored in code. Previous this lead to undefined references.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Stephen Henson <steve@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4116)
Added manpage for the new scrypt EVP_PKEY_METHOD KDF interface.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Stephen Henson <steve@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4026)
Added the pkey_meth_kdf_test tests which test the PKEY_METHOD macros (at
the moment, of HKDF and scrypt).
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Stephen Henson <steve@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4026)
Add an interface that allows accessing the scrypt KDF as a PKEY_METHOD.
This fixes#4021 (at least for the scrypt portion of the issue).
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Stephen Henson <steve@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4026)
Try to put DRBG and rand_bytes buffers in secure heap
Read the TSC fewer times (but it's still not enabled).
Short-circuit return in win RAND_poll_ex; other minor tweaks and
format-fixes.
Use the _bytes version of rdrand/rdseed
Fix ia32cap checks.
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
(Merged from https://github.com/openssl/openssl/pull/4100)
Use atfork to count child forks, and reseed DRBG when the counts don't
match.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/4101)
reference not by value. This allows an error return from the setup function.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4083)
The test ENGINE effectively used a predictable PRNG because it supplied
a bogus implementation of SHA256 which the old version of OpenSSL's PRNG
used. The new DRBG does not use SHA256 so it is no longer predictable
if the SHA256 implementation is replaced. Use an explicit predictable
PRNG instead.
Reviewed-by: Kurt Roeckx <kurt@roeckx.be>
(Merged from https://github.com/openssl/openssl/pull/4098)
Out-of-bounds array access in the example documentation of
EVP_PKEY_CTX_set_hkdf_md fixed.
Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Stephen Henson <steve@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4081)
Add missing ASN1_TIME functions
Do some cleanup of the ASN1_TIME code.
Add ASN1_TIME_normalize() to normalize ASN1_TIME structures.
Add ASN1_TIME_compare() to compare two ASN1_TIME structures.
Add ASN1_TIME_cmp_time_t() to compare an ASN1_TIME to time_t
(generic version of ASN1_UTCTIME_cmp_time_t()).
Replace '0' .. '9' compares with isdigit()
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2753)
Change the fixture types to pointers to structures that are heap allocated in the tests that use SETUP_TEST_FIXTURE. This will permit error returns from the setup function and allow for future running tests in parallel.
Also removed a call of `exit(2)` which allows the remaining tests to run if one fails to initialise.
Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4071)
Add a new global DRBG for private keys used by RAND_priv_bytes.
Add BN_priv_rand() and BN_priv_rand_range() which use RAND_priv_bytes().
Change callers to use the appropriate BN_priv... function.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/4076)
Give each SSL object it's own DRBG, chained to the parent global
DRBG which is used only as a source of randomness into the per-SSL
DRBG. This is used for all session, ticket, and pre-master secret keys.
It is NOT used for ECDH key generation which use only the global
DRBG. (Doing that without changing the API is tricky, if not impossible.)
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/4050)
If RAND_add wraps around, XOR with existing. Add test to drbgtest that
does the wrap-around.
Re-order seeding and stop after first success.
Add RAND_poll_ex()
Use the DF and therefore lower RANDOMNESS_NEEDED. Also, for child DRBG's,
mix in the address as the personalization bits.
Centralize the entropy callbacks, from drbg_lib to rand_lib.
(Conceptually, entropy is part of the enclosing application.)
Thanks to Dr. Matthias St Pierre for the suggestion.
Various code cleanups:
-Make state an enum; inline RANDerr calls.
-Add RAND_POLL_RETRIES (thanks Pauli for the idea)
-Remove most RAND_seed calls from rest of library
-Rename DRBG_CTX to RAND_DRBG, etc.
-Move some code from drbg_lib to drbg_rand; drbg_lib is now only the
implementation of NIST DRBG.
-Remove blocklength
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/4019)
Move the definition of ossl_assert() out of e_os.h which is intended for OS
specific things. Instead it is moved into internal/cryptlib.h.
This also changes the definition to remove the (int) cast.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4073)
The comment "The following should not return 1, otherwise, things
are very strange" is from the very first commit of OpenSSL. The
really meaning of the comment is if the identical session can be
found from internal cache after calling get_session_cb but not
found before calling get_session_cb, it is just strange.
The value 1 was originated from the old doc of SSLeay, reversed
from the actual return value of SSL_CTX_add_session().
Anyway either return value of SSL_CTX_add_session() should not
interrupt the session resumption process. So the checking of
return value of SSL_CTX_add_session() is not necessary.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4014)
Conform to coding guidelines.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Stephen Henson <steve@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3989)
This was previously mistakenly handled as a single error code.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Stephen Henson <steve@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3989)
Changed HKDF to use EVP_PKEY_CTX_md() (review comment of @snhenson) and
introduced more specific error codes (not only indicating *that* some
parameter is missing, but actually *which* one it is).
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Stephen Henson <steve@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3989)
HKDF now handles an invalid digest like TLS1-PRF does (i.e., returns
KDF_R_INVALID_DIGEST if the passed digest is not known). Both KDFs now
set the error code KDF_R_UNKNOWN_PARAMETER_TYPE if a type was passed
that is not recognized. This will have the effect of improving debugging
output in case a user uses "openssl pkeyutl -kdf ..." in a wrong way and
result in an actual error code (instead of just "failure" and an empty
error stack).
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Stephen Henson <steve@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3989)
Introduce KDF_F_PKEY_HKDF_DERIVE and return the KDF_R_MISSING_PARAMETER
error code when required parameters have not been set. This will make
"openssl pkeyutl -kdf HKDF" return a meaningful error message instead of
simply "Public Key operation error".
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Stephen Henson <steve@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3989)
This is achieved mostly by ~10% reduction of amount of instructions
per round thanks to a) switch to KECCAK_2X variant; b) merge of
almost 1/2 rotations with logical instructions. Performance is
improved on all observed processors except on Cortex-A15. This is
because it's capable of exploiting more parallelism and can execute
original code for same amount of time.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
(Merged from https://github.com/openssl/openssl/pull/4057)
Removing the use of SETUP_TEST_FIXTURE reduces complxity in those tests that
used it.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4066)
The OID for {1 3 6 1 5 5 8 1 2} HMAC-SHA1 (NID_hmac_sha1) is explicitly
referenced by RFC 2510, RFC 3370, and RFC 4210. This is essential for the
common implementations of CMP (Certificate Managing Protocol, RFC4210).
HMAC-MD5's OID {1 3 6 1 5 5 8 1 1} (NID_hmac_md5) is in the same branch and
it seems to generally exist (-> Internet search), but it is unclear where it is
actually defined as it appears not to be referenced by RFCs and practically
rather unused.
Those OIDs are both duplicates to OIDs from an RSA OID branch, which are already
included in builtin_pbe[]:
HMAC-SHA1 also has another OID defined in PKCS#5/RFC2898 (NID_hmacWithSHA1).
It is also unclear where the other OID for HMAC-MD5 (NID_hmacWithMD5) from the
RSA branch is officially specified, as only HMAC-SHA1 from PKCS#5 was found to be
defined. Anyway, HMAC-MD5 likely only plays a neglectable role in the future.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/3811)
Simplify the only test that uses this macro so it doesn't need it anymore.
Clean up the formatting a little.
Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4034)
"More" refers to the fact that we make active BIT_INTERLEAVE choice
in some specific cases. Update commentary correspondingly.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Even though tm->length >= 15 && v[14] == '.' works in practice,
[because "YYYYMMDDHHMMSS." would be rejected as invalid by
asn1_time_to_tm,] formal correctness with respect to buffer
overstep in few lines vicinity improves readability.
[Also fold one if condition and improve expression readability.]
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4058)
