We are quite inconsistent about which alerts get sent. Specifically, these
alerts should be used (normally) in the following circumstances:
SSL_AD_DECODE_ERROR = The peer sent a syntactically incorrect message
SSL_AD_ILLEGAL_PARAMETER = The peer sent a message which was syntactically
correct, but a parameter given is invalid for the context
SSL_AD_HANDSHAKE_FAILURE = The peer's messages were syntactically and
semantically correct, but the parameters provided were unacceptable to us
(e.g. because we do not support the requested parameters)
SSL_AD_INTERNAL_ERROR = We messed up (e.g. malloc failure)
The standards themselves aren't always consistent but I think the above
represents the best interpretation.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3480)
add_key_share() is a helper function used during key_share extension
construction. It is expected to be a simple boolean success/fail return.
It shouldn't be using the new EXT_RETURN type but it was partially converted
anyway. This changes it back.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3488)
The current TLSv1.3 spec says:
'If a server is authenticating via a certificate and the client has not
sent a "signature_algorithms" extension, then the server MUST abort the
handshake with a "missing_extension" alert (see Section 8.2).'
If we are resuming then we are not "authenticating via a certificate" but
we were still aborting with the missing_extension alert if sig algs was
missing.
This commit ensures that we only send the alert if we are not resuming.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3478)
We already did this on an ad-hoc per extension basis (for some extensions).
This centralises it and makes sure we do it for all extensions.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3418)
- Mostly missing fall thru comments
- And uninitialized value used in sslapitest.c
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3440)
Add "single part" digest sign and verify functions. These sign and verify
a message in one function. This simplifies some operations and it will later
be used as the API for algorithms which do not support the update/final
mechanism (e.g. PureEdDSA).
Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3409)
The function SSL_set_SSL_CTX() can be used to swap the SSL_CTX used for
a connection as part of an SNI callback. One result of this is that the
s->cert structure is replaced. However this structure contains information
about any custom extensions that have been loaded. In particular flags are
set indicating whether a particular extension has been received in the
ClientHello. By replacing the s->cert structure we lose the custom
extension flag values, and it appears as if a client has not sent those
extensions.
SSL_set_SSL_CTX() should copy any flags for custom extensions that appear
in both the old and the new cert structure.
Fixes#2180
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3425)
It is invalid if we receive an HRR but no change will result in
ClientHello2.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3414)
If an HRR gets sent without a key_share (e.g. cookie only) then the code
fails when it should not.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3414)
It is illegal in a TLSv1.3 ClientHello to send anything other than the
NULL compression method. We should send an alert if we find anything else
there. Previously we were ignoring this error.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3410)
The TLSv1.3 spec says that a server SHOULD send supported_groups in the
EE message if there is a group that it prefers to the one used in the
key_share. Clients MAY act on that. At the moment we don't do anything
with it on the client side, but that may change in the future.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3395)
We already did this for ServerHello and EncryptedExtensions. We should be
doing it for Certificate and HelloRetryRequest as well.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3298)
This fixes a segfault if a NULL parse_cb is passed to
SSL_CTX_add_{client,server}_custom_ext, which was supported in the
pre-1.1.1 implementation.
This behaviour is consistent with the other custom_ext_*_old_cb_wrap
functions, and with the new SSL_CTX_add_custom_ext function.
CLA: trivial
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3310)
Previously, init and finalization function for extensions are called
per extension block, rather than per message. This commit changes
that behaviour, and now they are called per message. The parse
function is still called per extension block.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3244)
Because NST messages arrive post-handshake, the session may have already
gone into the cache. Once in the cache a session must be immutable -
otherwise you could get multi-thread issues.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3008)
Provide a way to test whether the SSL_SESSION object can be used to resume a
sesion or not.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3008)
TLSv1.3 will do the same thing as TLSv1.2 with tickets with regards to session
ids, i.e. it will create a synthetic session id when the session is established,
so it is reasonable to check the session id length, even in TLSv1.3.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3008)
Ensure that there are ciphersuites enabled for the maximum supported
version we will accept in a ClientHello.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3316)
Ensure that there are ciphersuites enabled for the maximum supported
version we are claiming in the ClientHello.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3316)
The function tls_early_post_process_client_hello() was overwriting the
passed "al" parameter even if it was successful. The caller of that
function, tls_post_process_client_hello(), sets "al" to a sensible default
(HANDSHAKE_FAILURE), but this was being overwritten to be INTERNAL_ERROR.
The result is a "no shared cipher" error (and probably other similar errors)
were being reported back to the client with an incorrect INTERNAL_ERROR
alert.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3314)
We were allocating the write buffer based on the size of max_send_fragment,
but ignoring it when writing data. We should fragment handshake messages
if they exceed max_send_fragment and reject application data writes that
are too large.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3286)
There was code existing which attempted to handle the case where application
data is received after a reneg handshake has started in SCTP. In normal DTLS
we just fail the connection if this occurs, so there doesn't seem any reason
to try and work around it for SCTP. In practice it didn't work properly
anyway and is probably a bad idea to start with.
Fixes#3251
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3286)
ECDHE is not properly defined for SSLv3. Commit fe55c4a2 prevented ECDHE
from being selected in that protocol. However, historically, servers do
still select ECDHE anyway so that commit causes interoperability problems.
Clients that previously worked when talking to an SSLv3 server could now
fail.
This commit introduces an exception which enables a client to continue in
SSLv3 if the server selected ECDHE.
[extended tests]
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3204)
doing the pms assignment after log is successful
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3278)
RFC 7301 mandates that the server SHALL respond with a fatal
"no_application_protocol" alert when there is no overlap between
the client's supplied list and the server's list of supported protocols.
In commit 062178678f we changed from
ignoring non-success returns from the supplied alpn_select_cb() to
treating such non-success returns as indicative of non-overlap and
sending the fatal alert.
In effect, this is using the presence of an alpn_select_cb() as a proxy
to attempt to determine whether the application has configured a list
of supported protocols. However, there may be cases in which an
application's architecture leads it to supply an alpn_select_cb() but
have that callback be configured to take no action on connections that
do not have ALPN configured; returning SSL_TLSEXT_ERR_NOACK from
the callback would be the natural way to do so. Unfortunately, the
aforementioned behavior change also treated SSL_TLSEXT_ERR_NOACK as
indicative of no overlap and terminated the connection; this change
supplies special handling for SSL_TLSEXT_ERR_NOACK returns from the
callback. In effect, it provides a way for a callback to obtain the
behavior that would have occurred if no callback was registered at
all, which was not possible prior to this change.
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2570)
The old custom extensions API was not TLSv1.3 aware. Extensions are used
extensively in TLSv1.3 and they can appear in many different types of
messages. Therefore we need a new API to be able to cope with that.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3139)
This move prepares for the later addition of the new custom extensions
API. The context codes have an additional "SSL_" added to their name to
ensure we don't have name clashes with other applications.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3139)
