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)
libssl at the moment downgrades an EVP_PKEY to an EC_KEY object in order
to get the conv form and field type. Instead we provide EVP_PKEY level
functions to do this.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13139)
Co-author: Richard Levitte <levitte@openssl.org>
Co-author: Tomas Mraz <tmraz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13139)
Whenever we set a private key in libssl, we first found the certificate
that matched the key algorithm. Then we copied the key parameters from the
private key into the public key for the certficate before finally checking
that the private key matched the public key in the certificate. This makes
no sense! Part of checking the private key is to make sure that the
parameters match. It seems that this code has been present since SSLeay.
Perhaps at some point it made sense to do this - but it doesn't any more.
We remove that piece of code altogether. The previous code also had the
undocumented side effect of removing the certificate if the key didn't
match. This makes sense if you've just overwritten the parameters in the
certificate with bad values - but doesn't seem to otherwise. I've also
removed that error logic.
Due to issue #13893, the public key associated with the certificate is
always a legacy key. EVP_PKEY_copy_parameters will downgrade the "from"
key to legacy if the target is legacy, so this means that in libssl all
private keys were always downgraded to legacy when they are first set
in the SSL/SSL_CTX. Removing the EVP_PKEY_copy_parameters code has the
added benefit of removing that downgrade.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13899)
Linux kernel is going to support ChaCha20-Poly1305 in TLS offload.
Add support for this cipher.
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13475)
To clarify the purpose of these two calls rename them to
EVP_CIPHER_CTX_get_original_iv and EVP_CIPHER_CTX_get_updated_iv.
Also rename the OSSL_CIPHER_PARAM_IV_STATE to OSSL_CIPHER_PARAM_UPDATED_IV
to better align with the function name.
Fixes#13411
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13870)
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 openssl code base has only a few occurrences of 'unsigned const char'
(15 occurrences), compared to the more common 'const unsigned char' (4420
occurrences).
While the former is not illegal C, mixing the 'const' keyword (a 'type
qualifier') in between 'unsigned' and 'char' (both 'type specifiers') is a
bit odd.
The background for writing this patch is not to be pedantic, but because
the 'opmock' program (used to mock headers for unit tests) does not accept
the 'unsigned const char' construct. While this definitely is a bug in
opmock or one of its dependencies, openssl is the only piece of software we
are using in combination with opmock that has this construct.
CLA: trivial
Reviewed-by: Nicola Tuveri <nic.tuv@gmail.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
(Merged from https://github.com/openssl/openssl/pull/13722)
Function SSL_group_to_name() added, together with documentation and tests.
This now permits displaying names of internal and external
provider-implemented groups.
Partial fix of #13767
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
Reviewed-by: Nicola Tuveri <nic.tuv@gmail.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13785)
Our free functions should be able to deal with the case where the object
being freed is NULL. This turns out to not be quite the case for DTLS
related objects.
Fixes#13649
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/13655)
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)
This code started off as a copy of ssl3_write_bytes(), and the comment
was not updated with the implementation.
Reported by yangyangtiantianlonglong in #13518
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13566)
Various sections of code assumed that at least one of dh or ec would be
available. We also now also need to handle cases where a provider has
a key exchange algorithm and TLS-GROUP that we don't know about.
Fixes#13536
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/13549)
These two bodies should be grouped together anyway as the reason for
the call to BIO_flush() is to permit using BIO_set_ktls_ctrl_msg().
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/13090)
When using KTLS, empty fragments sent as a mitigation for known-IV
weakenesses in TLS 1.0 are sent as writes of 0 bytes. The TLS header
and trailer are added to the empty fragment by the kernel.
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/13090)
This deprecates all the ERR_load_ functions, and moves their definition to
separate C source files that can easily be removed when those functions are
finally removed.
This also reduces include/openssl/kdferr.h to include cryptoerr_legacy.h,
moves the declaration of ERR_load_ERR_strings() from include/openssl/err.h
to include/openssl/cryptoerr_legacy.h, and finally removes the declaration
of ERR_load_DSO_strings(), which was entirely internal anyway.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13390)
disabled_enc_mask et al were global. Now that cipher loading is done
individually for each SSL_CTX, based on the libctx configured for that
SSL_CTX this means that some things will be disabled for one SSL_CTX but
not for another. The global variables set up the potential for different
SSL_CTXs to trample on each other. We move these variables into the SSL_CTX
structure.
Fixes#12040
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/13465)
This was probably due to a merge
Fixes#13449
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: David von Oheimb <david.von.oheimb@siemens.com>
(Merged from https://github.com/openssl/openssl/pull/13450)
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)
The old function took a DH as a parameter. In the new version we pass
an EVP_PKEY instead. Similarly for the SSL_CTX version of this function.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13368)
This option calls SSL_set_tmp_dh() which does not exist in a no-deprecated
build. We need to implement an alternative.
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)
DTLS by design ignores records/packets with bad MAC or failed AEAD tag
validation. However, recent changes to have provided cipher
implementations caused tls1_enc() to leave an entry on the error queue
for invalid GCM tags, e.g.:
800BEAEF487F0000:error::Provider routines:gcm_stream_update:cipher operation failed:providers/implementations/ciphers/ciphercommon_gcm.c:306
The BoringSSL tests check for entries on the error queue with
SSL_get_error() and so we were seeing spurious test failures
due to the additional item on the error queue. To avoid leaving
such spurious entries on the error queue, set a mark before calling
the ssl3_enc 'enc' method, and pop to that mark before ignoring
invalid packets.
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13251)
The handling for the SCSVs was the same as for regular ciphers;
just merge them into the same table-driven handler.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/13010)
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)
