If we have an assert then in a debug build we want an abort() to occur.
In a production build we want the function to return an error.
This introduces a new macro to assist with that. The idea is to replace
existing use of OPENSSL_assert() with this new macro. The problem with
OPENSSL_assert() is that it aborts() on an assertion failure in both debug
and production builds. It should never be a library's decision to abort a
process (we don't get to decide when to kill the life support machine or
the nuclear reactor control system). Additionally if an attacker can
cause a reachable assert to be hit then this can be a source of DoS attacks
e.g. see CVE-2017-3733, CVE-2015-0293, CVE-2011-4577 and CVE-2002-1568.
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3496)
Excess bytes, when one block is longer than the other, are not explicitly
highlighted.
The NULL / zero length block output has been cleaned up.
Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3515)
Modify one of the integration builds so that that the
OPENSSL_SMALL_MEMORY option is compiled. There doesn't appear to be an
automatic build with this option set.
I think the options in the modified build are covered elsewhere (without
the small memory) but a new job might still be preferable.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3268)
Fix the small arena test to just check for the symptom of the infinite
loop (i.e. initialized set on failure), rather than the actual infinite
loop. This avoids some valgrind errors.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3512)
The password_callback() function does not necessarily NULL terminate
the password buffer, the caller must use the returned length but the
srp app uses this function as if it was doing NULL termination.
This made the -passin and -passout options of "openssl srp"
fail inexpicably and randomly or even crash.
Fixed by enlarging the buffer by one, so that the maximum password length
remains unchanged, and adding NULL termination upon return.
[Rearrange code for coding style compliance in process.]
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3475)
Switch to TAP::Harness inadvertently masked test failures.
Test::Harness::runtests was terminating with non-zero exit code in case
of failure[s], while TAP::Harness apparently holds caller responsible
for doing so.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Remove assertion when mmap() fails.
Only run the 1<<31 limit test on Linux
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3455)
Import test data from sha1test.c, sha256t.c and sha512t.c which is
from RFC6234 section 8.5
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3451)
Allow multiple "Input" lines to call the update function multiple times.
Add "Ncopy" keyword to copy the input buffer. So for example:
Input = "a"
Ncopy = 1024
Will create a buffer consisting of 1024 "a" characters.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3451)
The return code from tls1_mac is supposed to be a boolean 0 for fail, 1 for
success. In one place we returned -1 on error. This would cause code calling
the mac function to erroneously see this as a success (because a non-zero
value is being treated as success in all call sites).
Fortunately, AFAICT, the place that returns -1 can only happen on an
internal error so is not under attacker control. Additionally this code only
appears in master. In 1.1.0 the return codes are treated differently.
Therefore there are no security implications.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3495)
With the -list option the srp app loops on the main while() endlessly,
whether users were given on the command line or not. The loop should
be stopped when in list mode and there are no more users.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3489)
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)
recocognised -> recognised
CLA: trivial
Signed-off-by: Paul Yang <paulyang.inf@gmail.com>
Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3470)
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)
Notifications can be (and should be) configured on account basis on
the CI web site. This avoids getting emails to openssl-commits for
personal accounts that also build OpenSSL stuff.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3484)
An alert message is 2 bytes long. In theory it is permissible in SSLv3 -
TLSv1.2 to fragment such alerts across multiple records (some of which
could be empty). In practice it make no sense to send an empty alert
record, or to fragment one. TLSv1.3 prohibts this altogether and other
libraries (BoringSSL, NSS) do not support this at all. Supporting it adds
significant complexity to the record layer, and its removal is unlikely
to cause inter-operability issues.
The DTLS code for this never worked anyway and it is not supported at a
protocol level for DTLS. Similarly fragmented DTLS handshake records only
work at a protocol level where at least the handshake message header
exists within the record. DTLS code existed for trying to handle fragmented
handshake records smaller than this size. This code didn't work either so
has also been removed.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3476)
Allow multiple file arguments.
Split bntests.txt into separate files.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3471)
