TLS 1.3 allows for the "psk_ke" and "psk_dhe_ke" key-exchange modes.
Only the latter mode introduces a new ephemeral (Diffie-Hellman)
key exchange, with the PSK being the only key material used in the
former case.
It's a compliance requirement of RFC 8446 that the server MUST NOT
send a KeyShareEntry when using the "psk_ke" mode, but prior to
this commit we would send a key-share based solely on whether the
client sent one. This bug goes unnoticed in our internal test suite
since openssl communicating with openssl can never negotiate the
PSK-only key-exchange mode. However, we should still be compliant
with the spec, so check whether the DHE mode was offered and don't
send a key-share if it wasn't.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/14749)
It's a MUST-level requirement that if the client sends a pre_shared_key
extension not accompanied by a psk_key_exchange_modes extension, the
server must abort the handshake. Prior to this commit the server
would continue on.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/14749)
In TLS we process received messages like this:
1) Read Message Header
2) Validate and transition state based on received message type
3) Read Message Body
4) Process Message
In DTLS we read messages like this:
1) Read Message Header and Body
2) Validate and transition state based on received message type
3) Process Message
The difference is because of the stream vs datagram semantics of the
underlying transport.
In both TLS and DTLS we were doing finished MAC processing as part of
reading the message body. This means that in DTLS this was occurring
*before* the state transition has been validated. A crash was occurring
in DTLS if a Finished message was sent in an invalid state due to
assumptions in the code that certain variables would have been setup by
the time a Finished message arrives.
To avoid this problem we shift the finished MAC processing to be after
the state transition in DTLS.
Thanks to github user @bathooman for reporting this issue.
Fixes#14906
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/14930)
Add a "where did this EVP_{CIPHER,MD} come from" flag: global, via fetch,
or via EVP_{CIPHER,MD}_meth_new. Update EVP_{CIPHER,MD}_free to handle all
three origins. The flag is deliberately right before some function pointers,
so that compile-time failures (int/pointer) will occur, as opposed to
taking a bit in the existing "flags" field. The "global variable" flag
is non-zero, so the default case of using OPENSSL_zalloc (for provider
ciphers), will do the right thing. Ref-counting is a no-op for
Make up_ref no-op for global MD and CIPHER objects
Deprecate EVP_MD_CTX_md(). Added EVP_MD_CTX_get0_md() (same semantics as
the deprecated function) and EVP_MD_CTX_get1_md(). Likewise, deprecate
EVP_CIPHER_CTX_cipher() in favor of EVP_CIPHER_CTX_get0_cipher(), and add
EVP_CIPHER_CTX_get1_CIPHER().
Refactor EVP_MD_free() and EVP_MD_meth_free() to call new common
evp_md_free_int() function.
Refactor EVP_CIPHER_free() and EVP_CIPHER_meth_free() to call new common
evp_cipher_free_int() function.
Also change some flags tests to explicit test == or != zero. E.g.,
if (flags & x) --> if ((flags & x) != 0)
if (!(flags & x)) --> if ((flags & x) == 0)
Only done for those lines where "get0_cipher" calls were made.
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/14193)
This change includes swapping the PUT and SPT configuration,
includes of sys/stat.h and sys/types.h in the correct scope
to be picked up by SPT definitions.
Fixes: #14698Fixes: #14734
CLA: The author has the permission to grant the OpenSSL Team the right to use this change.
Signed-off-by: Randall S. Becker <rsbecker@nexbridge.com>
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/14736)
Following on from CVE-2021-3449 which was caused by a non-zero length
associated with a NULL buffer, other buffer/length pairs are updated to
ensure that they too are always in sync.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
As the variable peer_sigalgslen is not cleared on ssl rehandshake, it's
possible to crash an openssl tls secured server remotely by sending a
manipulated hello message in a rehandshake.
On such a manipulated rehandshake, tls1_set_shared_sigalgs() calls
tls12_shared_sigalgs() with the peer_sigalgslen of the previous
handshake, while the peer_sigalgs has been freed.
As a result tls12_shared_sigalgs() walks over the available
peer_sigalgs and tries to access data of a NULL pointer.
This issue was introduced by c589c34e61 (Add support for the TLS 1.3
signature_algorithms_cert extension, 2018-01-11).
Signed-off-by: Peter Kästle <peter.kaestle@nokia.com>
Signed-off-by: Samuel Sapalski <samuel.sapalski@nokia.com>
CVE-2021-3449
CLA: trivial
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
If we have TLSv1.3 enabled then we must have at least one TLSv1.3 capable
group available. This check was not always working
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/14430)
If the EVP_MD_CTX_ctrl is deprecated the code will
generate deprecation warnings. So there is no point in marking
all EVP_MD_CTX_ctrl() calls with TODOs.
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/14367)
The low level DH API has two functions for checking parameters:
DH_check_ex() and DH_check_params_ex(). The former does a "full" check,
while the latter does a "quick" check. Most importantly it skips the
check for a safe prime. We're ok without using safe primes here because
we're doing ephemeral DH.
Now that libssl is fully using the EVP API, we need a way to specify that
we want a quick check instead of a full check. Therefore we introduce
EVP_PKEY_param_check_quick() and use it.
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/14146)
The low level SRP implementation has been deprecated with no replacement.
Therefore the libssl level APIs need to be similarly deprecated.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/14132)
This field has not been used since #3858 was merged in 2017 when we
moved to a table-based lookup for certificate type properties instead of
an index-based one.
Reviewed-by: Kurt Roeckx <kurt@roeckx.be>
(Merged from https://github.com/openssl/openssl/pull/13991)
The existing names such as EVP_PKEY_param_fromdata_settable were a bit
confusing since the 'param' referred to key params not OSSL_PARAM. To simplify
the interface a 'selection' parameter will be passed instead. The
changes are:
(1) EVP_PKEY_fromdata_init() replaces both EVP_PKEY_key_fromdata_init() and EVP_PKEY_param_fromdata_init().
(2) EVP_PKEY_fromdata() has an additional selection parameter.
(3) EVP_PKEY_fromdata_settable() replaces EVP_PKEY_key_fromdata_settable() and EVP_PKEY_param_fromdata_settable().
EVP_PKEY_fromdata_settable() also uses a selection parameter.
Fixes#12989
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/14076)
The supported groups code was checking the OPENSSL_NO_EC and
OPENSSL_NO_DH guards in order to work, and the list of default groups was
based on those guards. However we now need it to work even in a no-ec
and no-dh build, because new groups might be added from providers.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13916)
The default supported groups code was disabled in the event of a build
with no-ec and no-dh. However now that providers can add there own
groups (which might not fit into either of these categories), this is
no longer appropriate.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13916)
This removes man unnecessary OPENSSL_NO_DH guards from libssl. Now that
libssl is entirely using the EVP APIs and implementations can be plugged
in via providers it is no longer needed to disable DH at compile time in
libssl. Instead it should detect at runtime whether DH is available from
the loaded providers.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13916)
The client-side cert verification callback function may not only return
as usual for success or 0 for failure, but also -1,
typically on failure verifying the server certificate.
This makes the handshake suspend and return control to the calling application
with SSL_ERROR_WANT_RETRY_VERIFY.
The app can for instance fetch further certificates or cert status information
needed for the verification.
Calling SSL_connect() again resumes the connection attempt
by retrying the server certificate verification step.
This process may even be repeated if need be.
The core implementation of the feature is in ssl/statem/statem_clnt.c,
splitting tls_process_server_certificate() into a preparation step
that just copies the certificates received from the server to s->session->peer_chain
(rather than having them in a local variable at first) and returns to the state machine,
and a post-processing step in tls_post_process_server_certificate() that can be repeated:
Try verifying the current contents of s->session->peer_chain basically as before,
but give the verification callback function the chance to pause connecting and
make the TLS state machine later call tls_post_process_server_certificate() again.
Otherwise processing continues as usual.
The documentation of the new feature is added to SSL_CTX_set_cert_verify_callback.pod
and SSL_want.pod.
This adds two tests:
* A generic test in test/helpers/handshake.c
on the usability of the new server cert verification retry feature.
It is triggered via test/ssl-tests/03-custom_verify.cnf.in (while the bulky auto-
generated changes to test/ssl-tests/03-custom_verify.cnf can be basically ignored).
* A test in test/sslapitest.c that demonstrates the effectiveness of the approach
for augmenting the cert chain provided by the server in between SSL_connect() calls.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13906)
Update constant to maximum permitted by RFC 8446
Fixes#13868
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13874)
Fixes#13183
From the original issue report, before this commit, on master and on
1.1.1, the issue can be detected with the following steps:
- Start with a default SSL_CTX, initiate a TLS 1.3 connection with SNI,
"Accept" count of default context gets incremented
- After servername lookup, "Accept" count of default context gets
decremented and that of SNI context is incremented
- Server sends a "Hello Retry Request"
- Client sends the second "Client Hello", now again "Accept" count of
default context is decremented. Hence giving a negative value.
This commit fixes it by adding a check on `s->hello_retry_request` in
addition to `SSL_IS_FIRST_HANDSHAKE(s)`, to ensure the counter is moved
only on the first ClientHello.
CLA: trivial
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/13297)
The configuration option 'no-rsa' was dropped with OpenSSL 1.1.0, so
this is simply a cleanup of the remains.
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/13700)
A servername cb may change the available certificates, so if we have one
set then we cannot rely on the configured certificates to determine if we
are capable of negotiating TLSv1.3 or not.
Fixes#13291
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/13304)
For the moment, we translate the result to a NID, because that's still
used in several locations in libssl. Future development should change
all the internals to be name based instead.
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/13436)
no-dh disables the low level API for DH. However, since we're now using
the high level EVP API in most places we don't need to disable quite so
much.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13368)
We instead set the encoded public key directly in the EVP_PKEY object.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13368)
We get DH related parameters directly from the EVP_PKEY instead of
downgrading to a DH object first.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13368)
There is no need for us to downgrade the EVP_PKEY into a DH object
for this function so we rewrite things to avoid it.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13368)
These ctrls pass around a DH object which is now deprecated, so we
deprecate the ctrls themselves.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13368)
Previously we were constructing a DH object and then assigning it to an
EVP_PKEY. Instead we construct an EVP_PKEY directly.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13368)
Previously a DH object was constructed and then assigned to an EVP_PKEY.
Instead we now construct the EVP_PKEY directly instead.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13368)
Since SSLfatal() doesn't take a function code any more, we drop that
argument everywhere. Also, we convert all combinations of SSLfatal()
and ERR_add_data() to an SSLfatal_data() call.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/13316)
ossl_statem_fatal() is refactored to be an extended ERR_set_error(),
and SSLfatal() is refactored to work like ERR_raise(). We also add
SSLfatal_data() to work like ERR_raise_data().
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/13316)
Use SSL_R_NO_PROTOCOLS_AVAILABLE instead of ERR_R_INTERNAL_ERROR,
to match what the BoringSSL tests expect for this case.
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13251)
We were missing a call to SSLfatal. A comment claimed that we had already
called it - but that is incorrect.
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/13229)
We do the same thing for the "get1" version. In reality this has broader
use than just TLS (it can also be used in CMS), and "encodedpoint" only
makes sense when you are talking about EC based algorithms.
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/13105)
The security operation SSL_SECOP_TMP_DH is defined to take an EVP_PKEY
in the "other" parameter:
/* Temporary DH key */
# define SSL_SECOP_TMP_DH (7 | SSL_SECOP_OTHER_PKEY)
In most places this is what is passed. All these places occur server side.
However there is one client side call of this security operation and it
passes a DH object instead. This is incorrect according to the
definition of SSL_SECOP_TMP_DH, and is inconsistent with all of the other
locations.
Our own default security callback, and the debug callback in the apps,
never look at this value and therefore this issue was never noticed
previously. In theory a client side application could be relying on this
behaviour and could be broken by this change. This is probably fairly
unlikely but can't be ruled out.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/13136)
We create a new file ssl/tls_depr.c to contain functions that need to call
deprecated APIs in libssl. This enables us to remove
OPENSSL_SUPPRESS_DEPRECATED from a number of other libssl files.
The deprecated API usage is either related to ENGINEs and is needed to
continue to support applications that use such ENGINEs. Or they are needed
to support some deprecated public libssl APIs.
One other file remains in libssl that still uses deprecated APIs: s3_cbc.c
This is needed to support the deprecated SSLv3.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/13135)
Automatically rename all instances of _with_libctx() to _ex() as per
our coding style.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/12970)
The check is applied only with X509_V_FLAG_X509_STRICT.
Fixes#12139
Reviewed-by: David von Oheimb <david.von.oheimb@siemens.com>
Reviewed-by: Nicola Tuveri <nic.tuv@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/12683)
We fix 3 problems with safestack:
- Including an openssl header file without linking against libcrypto
can cause compilation failures (even if the app does not otherwise need
to link against libcrypto). See issue #8102
- Recent changes means that applications in no-deprecated builds will need
to include additional macro calls in the source code for all stacks that
they need to use - which is an API break. This changes avoids that
necessity.
- It is not possible to write code using stacks that works in both a
no-deprecated and a normal build of OpenSSL. See issue #12707.
Fixes#12707
Contains a partial fix for #8102. A similar PR will be needed for hash to
fully fix.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/12781)
If we don't have MD5-SHA1 then we must be use (D)TLSv1.2 or above. We
check that this is consistent with the way we've been configured. We also
ensure that we never attempt to negotiate <(D)TLSv1.2 if MD5-SHA1 is not
available.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/12733)
If a digest is not available we just get an "internal error" error
message - which isn't very helpful for diagnosing problems. Instead we
explicitly state that we couldn't find a suitable digest.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/12733)
Convert various mac key creation function calls to use the _with_libctx
variants.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/12637)
-Added EVP_SignFinal_with_libctx() and EVP_VerifyFinal_with_libctx()
-Renamed EVP_DigestSignInit_ex() and EVP_DigestVerifyInit_with_libctx() to
EVP_DigestSignInit_with_libctx() and EVP_DigestVerifyInit_with_libctx()
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/11884)
- In order to not add many X509_XXXX_with_libctx() functions the libctx and propq may be stored in the X509 object via a call to X509_new_with_libctx().
- Loading via PEM_read_bio_X509() or d2i_X509() should pass in a created cert using X509_new_with_libctx().
- Renamed some XXXX_ex() to XXX_with_libctx() for X509 API's.
- Removed the extra parameters in check_purpose..
- X509_digest() has been modified so that it expects a const EVP_MD object() and then internally it does the fetch when it needs to (via ASN1_item_digest_with_libctx()).
- Added API's that set the libctx when they load such as X509_STORE_new_with_libctx() so that the cert chains can be verified.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/12153)
The 'MinProtocol' and 'MaxProtocol' configuration commands now silently
ignore TLS protocol version bounds when configurign DTLS-based contexts,
and conversely, silently ignore DTLS protocol version bounds when
configuring TLS-based contexts. The commands can be repeated to set
bounds of both types. The same applies with the corresponding
"min_protocol" and "max_protocol" command-line switches, in case some
application uses both TLS and DTLS.
SSL_CTX instances that are created for a fixed protocol version (e.g.
TLSv1_server_method()) also silently ignore version bounds. Previously
attempts to apply bounds to these protocol versions would result in an
error. Now only the "version-flexible" SSL_CTX instances are subject to
limits in configuration files in command-line options.
Expected to resolve#12394
Reviewed-by: Paul Dale <paul.dale@oracle.com>
GH: #12472
Deprecate SSL_get_peer_certificte() and replace with
SSL_get1_peer_certificate().
Add SSL_get0_peer_certificate.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/8730)
The DTLS1_COOKIE_LENGTH value was incorrect in the header files. We
couldn't change it before due to ABI concerns. However 3.0 is not ABI
compatible so we can now fix it.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/12179)
The supported_groups extension only supported EC groups in DTLS.
Therefore we shouldn't send it in a no-ec build.
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/11914)
If EVP_PKEY_copy_parameters() failed in libssl we did not provide a very
helpful error message. We provide a better one.
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/11914)
Now that we have added the TLS-GROUP capability to the default provider
we can use that to discover the supported group list based on the loaded
providers.
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/11914)
Abort renegotiation if server receives client hello with Extended Master
Secret extension dropped in comparison to the initial session.
Fixes#9754
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/12045)
We were downgrading to legacy keys at various points in libssl in
order to get or set an encoded point. Now that the encoded point
functions work with provided keys this is no longer necessary.
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/11898)
This API requests that the TLS stack generate a (TLS 1.3)
NewSessionTicket message the next time it is safe to do so (i.e., we do
not have other data pending write, which could be mid-record). For
efficiency, defer actually generating/writing the ticket until there
is other data to write, to avoid producing server-to-client traffic when
not needed.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/11416)
An 'if' clause was nestled against a previous closing brace as it if was
an 'else if', but should properly stand on its own line.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/11416)
... and only *define* them in the source files that need them.
Use DEFINE_OR_DECLARE which is set appropriately for internal builds
and not non-deprecated builds.
Deprecate stack-of-block
Better documentation
Move some ASN1 struct typedefs to types.h
Update ParseC to handle this. Most of all, ParseC needed to be more
consistent. The handlers are "recursive", in so far that they are called
again and again until they terminate, which depends entirely on what the
"massager" returns. There's a comment at the beginning of ParseC that
explains how that works. {Richard Levtte}
Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/10669)
We had a spot where a fatal error was occurring but we hadn't sent an
alert. This results in a later assertion failure.
Reviewed-by: Paul Yang <kaishen.yy@antfin.com>
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/11537)
Libssl is OPENSSL_CTX aware so we should use it when creating an
X509_STORE_CTX.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/11457)
The transfer of TLS encodedpoint to backends isn't yet fully supported
in provider implementations. This is a temporary measure so as not to
get stuck in other development.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/11358)
libssl code uses EVP_PKEY_get0_EC_KEY() to extract certain basic data
from the EC_KEY. We replace that with internal EVP_PKEY functions.
This may or may not be refactored later on.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/11358)
Having created a DH object and assigned it to an EVP_PKEY - we should
not free both the EVP_PKEY and the original DH. This will lead to a
double free occurring.
This issue was discovered and reported by GitHub Security Lab team member
Agustin Gianni.
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/11441)
We use AES-256-CBC to encrypt stateless session tickets. We should
ensure that the implementation is fetched from the appropriate provider.
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/11356)
We use the SHA256 digest of the ticket as a "fake" session id. We should
ensure that the SHA256 implementation is fetched from the appropriate
provider.
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/11356)
When constructing an RSA ClientKeyExchange make sure we construct our
EVP_PKEY_CTX using the correct libctx and properties
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/11357)
The server-side ChangeCipherState processing stores the new cipher
in the SSL_SESSION object, so that the new state can be used if
this session gets resumed. However, writing to the session is only
thread-safe for initial handshakes, as at other times the session
object may be in a shared cache and in use by another thread at the
same time. Reflect this invariant in the code by only writing to
s->session->cipher when it is currently NULL (we do not cache sessions
with no cipher). The code prior to this change would never actually
change the (non-NULL) cipher value in a session object, since our
server enforces that (pre-TLS-1.3) resumptions use the exact same
cipher as the initial connection, and non-abbreviated renegotiations
have produced a new session object before we get to this point.
Regardless, include logic to detect such a condition and abort the
handshake if it occurs, to avoid any risk of inadvertently using
the wrong cipher on a connection.
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/10943)
TLS 1.3 maintains a separate keys chedule in the SSL object, but
was writing to the 'master_key_length' field in the SSL_SESSION
when generating the per-SSL master_secret. (The generate_master_secret
SSL3_ENC_METHOD function needs an output variable for the master secret
length, but the TLS 1.3 implementation just uses the output size of
the handshake hash function to get the lengths, so the only natural-looking
thing to use as the output length was the field in the session.
This would potentially involve writing to a SSL_SESSION object that was
in the cache (i.e., resumed) and shared with other threads, though.
The thread-safety impact should be minimal, since TLS 1.3 requires the
hash from the original handshake to be associated with the resumption
PSK and used for the subsequent connection. This means that (in the
resumption case) the value being written would be the same value that was
previously there, so the only risk would be on architectures that can
produce torn writes/reads for aligned size_t values.
Since the value is essentially ignored anyway, just provide the
address of a local dummy variable to generate_master_secret() instead.
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/10943)
Use of the low level DH functions has been informally discouraged for a
long time. We now formally deprecate them.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/11024)
When we use an EVP_PKEY_CTX in libssl we should be doing so with the
OPENSSL_CTX and property query string that were specified when the
SSL_CTX object was first created.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/10854)
We modify libssl to use explicitly fetched ciphers, digests and other
algorithms as required based on the configured library context and
property query string for the SSL_CTX that is being used.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/10854)
If the servername cb decides to send back a warning alert then the
handshake continues, but we should not signal to the client that the
servername has been accepted.
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/10018)
The SNI behaviour for TLSv1.3 and the behaviour of SSL_get_servername()
was not quite right, and not entirely consistent with the RFC.
The TLSv1.3 RFC explicitly says that SNI is negotiated on each handshake
and the server is not required to associate it with the session. This was
not quite reflected in the code so we fix that.
Additionally there were some additional checks around early_data checking
that the SNI between the original session and this session were
consistent. In fact the RFC does not require any such checks, so they are
removed.
Finally the behaviour of SSL_get_servername() was not quite right. The
behaviour was not consistent between resumption and normal handshakes,
and also not quite consistent with historical behaviour. We clarify the
behaviour in various scenarios and also attempt to make it match historical
behaviour as closely as possible.
Fixes#8822
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/10018)
Backwards compatibility with the old ticket key call back is maintained.
This will be removed when the low level HMAC APIs are finally removed.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/10836)
Now that libssl knows about libctx we should use it wherever we generate
a random number.
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/10927)
Large GOST ClientKeyExchange messages are sent by VipNet CSP, one of
Russian certified products implementing GOST TLS, when a server
certificate contains 512-bit keys.
This behaviour was present in 1.0.2 branch and needs to be restored.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/10376)
Server side RSA key transport code in a Client Key Exchange message
currently uses constant time code to check that the RSA decrypt is
correctly formatted. The previous commit taught the underlying RSA
implementation how to do this instead, so we use that implementation and
remove this code from libssl.
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/10411)
As was done for ciphers, supported groups, and EC point formats in
https://github.com/openssl/openssl/pull/9162, only write the negotiated
SNI hostname value to the session object when not resuming, even for
TLS 1.3 resumptions. Otherwise, when using a stateful session cache
(as is done by default when 0-RTT data is enabled), we can have multiple
SSLs active using the same in-memory session object, which leads to
double-frees and similar race conditions in the SNI handler prior
to this commit.
Fortunately, since draft-ietf-tls-tls13-22, there is no requirement
that the SNI hostname be preserved across TLS 1.3 resumption, and thus
not a need to continually update the session object with the "current"
value (to be used when producing session tickets, so that the subsequent
resumption can be checked against the current value). So we can just
relax the logic and only write to the session object for initial handshakes.
This still leaves us in a somewhat inconsistent state, since if the SNI value
does change across handshakes, the session object will continue to record
the initial handshake's value, even if that bears no relation to the
current handshake. The current SSL_get_servername() implementation
prefers the value from the session if s->hit, but a more complete fix
for that and related issues is underway in
https://github.com/openssl/openssl/pull/10018; there is no need to wait
for the complete fix for SNI name handling in order to close the
race condition and avoid runtime crashes.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/10441)
The resumption_label variable when CHARSET_EBCDIC was enabled, was misspelled.
Instead of evaluating to 'res binder' as expected, it evaluated to 'red binder'.
CLA: trivial
Reviewed-by: Kurt Roeckx <kurt@roeckx.be>
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
(Merged from https://github.com/openssl/openssl/pull/10396)
Fixes#9869
CLA:trivial
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/9878)
(cherry picked from commit d8e8ed0220)
Also added blanks lines after declarations in a couple of places.
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9916)
Apart from public and internal header files, there is a third type called
local header files, which are located next to source files in the source
directory. Currently, they have different suffixes like
'*_lcl.h', '*_local.h', or '*_int.h'
This commit changes the different suffixes to '*_local.h' uniformly.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9333)
If a TLSv1.3 server configured to respond to the status_request extension
also attempted to send a CertificateRequest then it was incorrectly
inserting a non zero length status_request extension into that message.
The TLSv1.3 RFC does allow that extension in that message but it must
always be zero length.
In fact we should not be sending the extension at all in that message
because we don't support it.
Fixes#9767
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/9780)
Fix a few places where calling ossl_isdigit does the wrong thing on
EBCDIC based systems.
Replaced with ascii_isdigit.
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9556)
If compiled with 'no-deprecated', ERR_put_error() is undefined. We
had one spot where we were using it directly, because the file and
line information was passed from elsewhere.
Fortunately, it's possible to use ERR_raise() for that situation, and
call ERR_set_debug() immediately after and thereby override the
information that ERR_raise() stored in the error record.
util/mkerr.pl needed a small adjustment to not generate code that
won't compile in a 'no-deprecated' configuration.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/9452)
tls_parse_stoc_key_share was generating a new EVP_PKEY public/private
keypair and then overrides it with the server public key, so the
generation was a waste anyway. Instead, it should create a
parameters-only EVP_PKEY.
(This is a consequence of OpenSSL using the same type for empty key,
empty key with key type, empty key with key type + parameters, public
key, and private key. As a result, it's easy to mistakenly mix such
things up, as happened here.)
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Kurt Roeckx <kurt@roeckx.be>
(Merged from https://github.com/openssl/openssl/pull/9445)
Deprecate all xxx_F_ defines.
Removed some places that tested for a specific function.
Use empty field for the function names in output.
Update documentation.
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9058)
This reverts commit 524006dd1b.
While this change did prevent the sigalgs from getting inadvertently
clobbered by SSL_set_SSL_CTX(), it also caused the sigalgs to not be
set when the cert_cb runs. This, in turn, caused significant breakage,
such as SSL_check_chain() failing to find any valid chain. An alternate
approach to fixing the issue from #7244 will follow.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9157)
The previous 2 commits moved supported groups and ciphers out of the
session object to avoid race conditions. We now also move ecpointformats
for consistency. There does not seem to be a race condition with access
to this data since it is only ever set in a non-resumption handshake.
However, there is no reason for it to be in the session.
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/9162)
Similarly to the previous commit we were storing the peer offered list
of ciphers in the session. In practice there is no need for this
information to be avilable from one resumption to the next since this
list is specific to a particular handshake. Since the session object is
supposed to be immutable we should not be updating it once we have decided
to resume. The solution is to remove the session list out of the session
object.
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/9162)
In TLSv1.3 the supported groups can be negotiated each time a handshake
occurs, regardless of whether we are resuming or not. We should not store
the supported groups information in the session because session objects
can be shared between multiple threads and we can end up with race
conditions. For most users this won't be seen because, by default, we
use stateless tickets in TLSv1.3 which don't get shared. However if you
use SSL_OP_NO_TICKET (to get stateful tickets in TLSv1.3) then this can
happen.
The answer is to move the supported the supported group information into
the SSL object instead.
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/9162)
Make sure that the combination of no-ec with no-dh builds successfully.
If neither ec or dh are available then TLSv1.3 is not possible.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9156)
Now that we have TLSv1.3 FFDHE support there is no reason why we should
not allow TLSv1.3 to be used in a no-ec build. This commit enables that
to happen.
It also fixes no-ec which was previously broken.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9156)
This is still required currently by engines and digestsign/digestverify.
This PR contains merged in code from Richard Levitte's PR #9126.
[extended tests]
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9103)
Move digest code into the relevant providers (fips, default, legacy).
The headers are temporarily moved to be internal, and will be moved
into providers after all external references are resolved. The deprecated
digest code can not be removed until EVP_PKEY (signing) is supported by
providers. EVP_MD data can also not yet be cleaned up for the same reasons.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8763)
If we receive a KeyUpdate message (update requested) from the peer while
we are in the middle of a write, we should defer sending the responding
KeyUpdate message until after the current write is complete. We do this
by waiting to send the KeyUpdate until the next time we write and there is
no pending write data.
This does imply a subtle change in behaviour. Firstly the responding
KeyUpdate message won't be sent straight away as it is now. Secondly if
the peer sends multiple KeyUpdates without us doing any writing then we
will only send one response, as opposed to previously where we sent a
response for each KeyUpdate received.
Fixes#8677
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/8773)
With the removal of SSLv2, the s3 structure is always allocated, so
there is little point in having it be an allocated pointer. Collapse
the ssl3_state_st structure into ssl_st and fixup any references.
This should be faster than going through an indirection and due to
fewer allocations, but I'm not seeing any significant performance
improvement; it seems to be within the margin of error in timing.
Reviewed-by: Paul Yang <yang.yang@baishancloud.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7888)
Sessions must be immutable once they can be shared with multiple threads.
We were breaking that rule by writing the ticket index into it during the
handshake. This can lead to incorrect behaviour, including failed
connections in multi-threaded environments.
Reported by David Benjamin.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8383)
Prior to this commit we were keeping a count of how many KeyUpdates we
have processed and failing if we had had too many. This simplistic approach
is not sufficient for long running connections. Since many KeyUpdates
would not be a particular good DoS route anyway, the simplest solution is
to simply remove the key update count.
Fixes#8068
Reviewed-by: Kurt Roeckx <kurt@roeckx.be>
(Merged from https://github.com/openssl/openssl/pull/8299)
Add SSL_OP64_NO_EXTENDED_MASTER_SECRET, that can be set on either
an SSL or an SSL_CTX. When processing a ClientHello, if this flag
is set, do not indicate that the EMS TLS extension was received in
either the ssl3 object or the SSL_SESSION. Retain most of the
sanity checks between the previous and current session during
session resumption, but weaken the check when the current SSL
object is configured to not use EMS.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3910)
The original 1.1.1 design was to use SSL_CB_HANDSHAKE_START and
SSL_CB_HANDSHAKE_DONE to signal start/end of a post-handshake message
exchange in TLSv1.3. Unfortunately experience has shown that this confuses
some applications who mistake it for a TLSv1.2 renegotiation. This means
that KeyUpdate messages are not handled properly.
This commit removes the use of SSL_CB_HANDSHAKE_START and
SSL_CB_HANDSHAKE_DONE to signal the start/end of a post-handshake
message exchange. Individual post-handshake messages are still signalled in
the normal way.
This is a potentially breaking change if there are any applications already
written that expect to see these TLSv1.3 events. However, without it,
KeyUpdate is not currently usable for many applications.
Fixes#8069
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8096)
When computing the end-point shared secret, don't take the
terminating NULL character into account.
Please note that this fix breaks interoperability with older
versions of OpenSSL, which are not fixed.
Fixes#7956
Reviewed-by: Kurt Roeckx <kurt@roeckx.be>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7957)
This commit erroneously kept the DTLS timer running after the end of the
handshake. This is not correct behaviour and shold be reverted.
This reverts commit f7506416b1.
Fixes#7998
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8047)
We were setting a limit of SSL3_RT_MAX_PLAIN_LENGTH on the size of the
ClientHello. AFAIK there is nothing in the standards that requires this
limit.
The limit goes all the way back to when support for extensions was first
added for TLSv1.0. It got converted into a WPACKET max size in 1.1.1. Most
likely it was originally added to avoid the complexity of having to grow
the init_buf in the middle of adding extensions. With WPACKET this is
irrelevant since it will grow automatically.
This issue came up when an attempt was made to send a very large
certificate_authorities extension in the ClientHello.
We should just remove the limit.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7424)
The cryptopro extension is supposed to be unsolicited and appears in the
ServerHello only. Additionally it is unofficial and unregistered - therefore
we should really treat it like any other unknown extension if we see it in
the ClientHello.
Fixes#7747
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/7984)
Fix some issues in tls13_hkdf_expand() which impact the above function
for TLSv1.3. In particular test that we can use the maximum label length
in TLSv1.3.
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7755)
If compile OpenSSL with SSL_DEBUG macro, some test cases will cause the
process crashed in the debug code.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/7707)
SSL(_CTX)?_set_client_CA_list() was a server side only function in 1.1.0.
If it was called on the client side then it was ignored. In 1.1.1 it now
makes sense to have a CA list defined for both client and server (the
client now sends it the the TLSv1.3 certificate_authorities extension).
Unfortunately some applications were using the same SSL_CTX for both
clients and servers and this resulted in some client ClientHellos being
excessively large due to the number of certificate authorities being sent.
This commit seperates out the CA list updated by
SSL(_CTX)?_set_client_CA_list() and the more generic
SSL(_CTX)?_set0_CA_list(). This means that SSL(_CTX)?_set_client_CA_list()
still has no effect on the client side. If both CA lists are set then
SSL(_CTX)?_set_client_CA_list() takes priority.
Fixes#7411
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7503)
TLSv1.3 is more restrictive about the curve used. There must be a matching
sig alg defined for that curve. Therefore if we are using some other curve
in our certificate then we should not negotiate TLSv1.3.
Fixes#7435
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7442)
use_ecc() was always returning 1 because there are default (TLSv1.3)
ciphersuites that use ECC - even if those ciphersuites are disabled by
other options.
Fixes#7471
Reviewed-by: Kurt Roeckx <kurt@roeckx.be>
(Merged from https://github.com/openssl/openssl/pull/7479)
Commit 9ef9088c15 switched the SSL/SSL_CTX
statistics counters to using Thread-Sanitizer-friendly primitives.
However, it erroneously converted an addition of -1
(for s->session_ctx->stats.sess_accept) to an addition of +1, since that
is the only counter API provided by the internal tsan_assist.h header
until the previous commit. This means that for each accepted (initial)
connection, the session_ctx's counter would get doubly incremented, and the
(switched) ctx's counter would also get incremented.
Restore the counter decrement so that each accepted connection increments
exactly one counter exactly once (in net effect).
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/7464)
