... script data load.
On related note an attempt was made to merge rotations with logical
operations. I mean as we know, ARM ISA has merged rotate-n-logical
instructions which can be used here. And they were used to improve
keccak1600-armv4 performance. But not here. Even though this approach
resulted in improvement on Cortex-A53 proportional to reduction of
amount of instructions, ~8%, it didn't exactly worked out on
non-Cortex cores. Presumably because they break merged instructions
to separate μ-ops, which results in higher *operations* count. X-Gene
and Denver went ~20% slower and Apple A7 - 40%. The optimization was
therefore dismissed.
Reviewed-by: Rich Salz <rsalz@openssl.org>
It is otherwise unclear what all the magic numbers mean.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/4349)
"Early callback" is a little ambiguous now that early data exists.
Perhaps "ClientHello callback"?
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/4349)
This allows completely arbitrary passphrases to be entered, including
NUL bytes.
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/3821)
It is talked around but not explicitly stated in one part of the
documentation that you should put library configuration lines at the
start of the configuration file.
CLA: trivial
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/3959)
If an alert gets sent and then we close the connection immediately with
data still in the input buffer then a TCP-RST gets sent. Some OSs
immediately abandon data in their input buffer if a TCP-RST is received -
meaning the alert data itself gets ditched. Sending a TCP-FIN before the
TCP-RST seems to avoid this.
This was causing test failures in MSYS2 builds.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/4333)
The extensions not sent when TLS 1.2 is not used caused the message
length to be 109, which is less than the 127 threshold needed
to activate the F5 workaround. Add another 20 bytes of dummy ALPN
data do push it over the threshold.
Also, fix the definition of the (unused) local macro indicating
the threshold.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4346)
In OpenSSL 1.1.0, when there were no extensions added to the ServerHello,
we did not write the extension data length bytes to the end of the
ServerHello; this is needed for compatibility with old client implementations
that do not support TLS extensions (such as the default configuration of
OpenSSL 0.9.8). When ServerHello extension construction was converted
to the new extensions framework in commit
7da160b0f4, this behavior was inadvertently
limited to cases when SSLv3 was negotiated (and similarly for ClientHellos),
presumably since extensions are not defined at all for SSLv3. However,
extensions for TLS prior to TLS 1.3 have been defined in separate
RFCs (6066, 4366, and 3546) from the TLS protocol specifications, and as such
should be considered an optional protocol feature in those cases.
Accordingly, be conservative in what we send, and skip the extensions block
when there are no extensions to be sent, regardless of the TLS/SSL version.
(TLS 1.3 requires extensions and can safely be treated differently.)
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/4296)
This quiets down complaints about the use of uninitialised memory
[extended tests]
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4340)
clienthellotest tries to fill out the size of the ClientHello by adding
extra ciphersuites in order to test the padding extension. This is
unreliable because they are very dependent on configuration options. If we
add too much data the test will fail! We were already also adding some dummy
ALPN protocols to pad out the size, and it turns out that this is sufficient
just in itself, so drop the extra ciphersuites.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/4331)
Ironically enough not all installations get Module::Load::Conditional
installed by default... [It's a bit half-hearted, because such
installations are likely to lack more stuffi that is needed, but
nevertheless, it proved to be helpful.]
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4323)
Deprecated functions are still documented.
Put HISTORY after SEE ALSO; add HISTORY to BN_zero
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/3746)
If the server_name extension is long enough to require two bytes to
hold the length of either field, the test suite would not decode
the length properly. Using the PACKET_ APIs would have avoided this,
but it was desired to avoid using private APIs in this part of the
test suite, to keep ourselves honest.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4318)
The include search path was not picking up files in the root of
the tree.
[extended tests]
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4316)
This function is really emulating what would happen in client mode,
and does not necessarily reflect what is usable for a server SSL.
Make this a bit more explicit, and do some wordsmithing while here.
Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4284)
OPENSSL_ia32cap.pod discusses possibility to disable operations on
XMM register bank. This formally means that this flag has to be checked
in combination with other flags. But it customarily isn't. But instead
of chasing all the cases we can flip more bits together with FXSR one.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4303)
This is actually not all warnings, only return values.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4297)
Move struct timeval includes into e_os.h (where the Windows ones were).
Enaure that the include is guarded canonically.
Refer #4271
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4312)
The program will fail to run if it doesn't exist anyway, no need to
check its existence here.
Fixes#4306
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4309)
When parsing the header files, mkdef.pl didn't clear the line
terminator properly. In most cases, this didn't matter, but there
were moments when this caused parsing errors (such as CRLFs in certain
cases).
Fixes#4267
Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4304)
SNI and ALPN must be set to be consistent with the PSK. Otherwise this is
an error.
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/3926)
If there is no SNI in the session then s_client no longer sends the SNI
extension. Update the tests to take account of that
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/3926)
If we have not decided on an SNI value yet, but we are attempting to reuse
a session, and SNI is set in that, then we should use that value by
default.
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/3926)
