Matt Caswell [Thu, 24 Nov 2016 16:59:48 +0000 (16:59 +0000)]
Refactor ClientHello extension parsing
This builds on the work started in 1ab3836b3 and extends is so that
each extension has its own identified parsing functions, as well as an
allowed context identifying which messages and protocols it is relevant for.
Subsequent commits will do a similar job for the ServerHello extensions.
This will enable us to have common functions for processing extension blocks
no matter which of the multiple messages they are received from. In TLSv1.3
a number of different messages have extension blocks, and some extensions
have moved from one message to another when compared to TLSv1.2.
Perl changes reviewed by Richard Levitte. Non-perl changes reviewed by Rich
Salz
Reviewed-by: Rich Salz <rsalz@openssl.org> Reviewed-by: Richard Levitte <levitte@openssl.org>
Richard Levitte [Thu, 8 Dec 2016 10:16:37 +0000 (11:16 +0100)]
In UI_OpenSSL's open(), generate an error on unknown errno
TTY_get() sometimes surprises us with new errno values to determine if
we have a controling terminal or not. This generated error is a
helpful tool to figure out that this was what happened and what the
unknown value is.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2043)
Richard Levitte [Wed, 7 Dec 2016 21:44:47 +0000 (22:44 +0100)]
Add a test for the UI API
The best way to test the UI interface is currently by using an openssl
command that uses password_callback. The only one that does this is
'genrsa'.
Since password_callback uses a UI method derived from UI_OpenSSL(), it
ensures that one gets tested well enough as well.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2040)
Richard Levitte [Wed, 7 Dec 2016 15:36:44 +0000 (16:36 +0100)]
UI_process() didn't generate errors
Since there are many parts of UI_process() that can go wrong, it isn't
very helpful to only return -1 with no further explanation. With this
change, the error message will at least show which part went wrong.
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2037)
Matt Caswell [Mon, 21 Nov 2016 17:11:51 +0000 (17:11 +0000)]
Add a TLS1.3 TODO for the msg callback
At the moment the msg callback only received the record header with the
outer record type in it. We never pass the inner record type - we probably
need to at some point.
Kurt Roeckx [Fri, 2 Dec 2016 18:49:34 +0000 (19:49 +0100)]
Run a some tests with -DFUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
The fuzzers use -DFUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION, and actually
get different results based on that. We should have at least some
targets that actually fully use the fuzz corpora.
Kurt Roeckx [Fri, 2 Dec 2016 18:34:54 +0000 (19:34 +0100)]
Make the fuzzers more reproducible
We want to be in the same global state each time we come in
FuzzerTestOneInput(). There are various reasons why we might not be that
include:
- Initialization that happens on first use. This is mostly the
RUN_ONCE() things, or loading of error strings.
- Results that get cached. For instance a stack that is sorted, RSA
blinding that has been set up, ...
So I try to trigger as much as possible in FuzzerInitialize(), and for
things I didn't find out how to trigger this it needs to happen in
FuzzerTestOneInput().
Matt Caswell [Thu, 17 Nov 2016 18:00:17 +0000 (18:00 +0000)]
Use the TLSv1.3 nonce construction
This updates the record layer to use the TLSv1.3 style nonce construciton.
It also updates TLSProxy and ossltest to be able to recognise the new
layout.
Matt Caswell [Tue, 29 Nov 2016 09:49:04 +0000 (09:49 +0000)]
Fix mac-then-encrypt test with enable-tls1_3
Commit b3618f44 added a test for mac-then-encrypt. However the test fails
when running with "enable-tls1_3". The problem is that the test creates a
connection, which ends up being TLSv1.3. However it also restricts the
ciphers to a single mac-then-encrypt ciphersuite that is not TLSv1.3
compatible so the connection aborts and the test fails. Mac-then-encrypt
is not relevant to TLSv1.3, so the test should disable that protocol
version.
Matt Caswell [Wed, 23 Nov 2016 23:03:13 +0000 (23:03 +0000)]
Ensure we are in accept state in DTLSv1_listen
Calling SSL_set_accept_state() after DTLSv1_listen() clears the state, so
SSL_accept() no longer works. In 1.0.2 calling DTLSv1_listen() would set
the accept state automatically. We should still do that.
Rich Salz [Mon, 28 Nov 2016 17:26:05 +0000 (12:26 -0500)]
Make bntest be (mostly) file-based.
Test suite used from boring, written by David Benjamin.
Test driver converted from C++ to C.
Added a Perl program to check the testsuite file.
Extensive review feedback incorporated (thanks folks).
Reviewed-by: Emilia Käsper <emilia@openssl.org> Reviewed-by: Richard Levitte <levitte@openssl.org>
Emilia Kasper [Fri, 25 Nov 2016 16:05:30 +0000 (17:05 +0100)]
Test mac-then-encrypt
Verify that the encrypt-then-mac negotiation is handled
correctly. Additionally, when compiled with no-asm, this test ensures
coverage for the constant-time MAC copying code in
ssl3_cbc_copy_mac. The proxy-based CBC padding test covers that as
well but it's nevertheless better to have an explicit handshake test
for mac-then-encrypt.
Andy Polyakov [Mon, 21 Nov 2016 14:15:52 +0000 (15:15 +0100)]
bn/asm/ppc-mont.pl: signal no-op in 32-bit bit build.
The bug was introduced in 80d27cdb84985c697f8fabb7649abf1f54714d13,
one too many instructions was removed. It went unnoticed, because
new subroutine introduced in previous commit is called in real-life
RSA/DSA/DH cases, while original code is called only in rare tests.
The bug was caught in test_fuzz.
Matt Caswell [Mon, 21 Nov 2016 13:24:50 +0000 (13:24 +0000)]
Fix SSL_IS_TLS13(s)
The SSL_IS_TLS13() macro wasn't quite right. It would come back with true
in the case where we haven't yet negotiated TLSv1.3, but it could be
negotiated.
Matt Caswell [Tue, 15 Nov 2016 14:22:29 +0000 (14:22 +0000)]
Add an s_server capability to read an OCSP Response from a file
Current s_server can only get an OCSP Response from an OCSP responder. This
provides the capability to instead get the OCSP Response from a DER encoded
file.
Matt Caswell [Tue, 15 Nov 2016 11:09:25 +0000 (11:09 +0000)]
Fix some TLSProxy warnings
After the client processes the server's initial flight in TLS1.3 it may
respond with either an encrypted, or an unencrypted alert. We needed to
teach TLSProxy about this so that it didn't issue spurious warnings.
Matt Caswell [Tue, 15 Nov 2016 10:13:09 +0000 (10:13 +0000)]
Ensure the end of first server flight processing is done
There is a set of miscellaneous processing for OCSP, CT etc at the end of
the ServerDone processing. In TLS1.3 we don't have a ServerDone, so this
needs to move elsewhere.
Matt Caswell [Wed, 9 Nov 2016 14:06:12 +0000 (14:06 +0000)]
Update state machine to be closer to TLS1.3
This is a major overhaul of the TLSv1.3 state machine. Currently it still
looks like TLSv1.2. This commit changes things around so that it starts
to look a bit less like TLSv1.2 and bit more like TLSv1.3.
Key differences between this intermediate position and the final TLSv1.3
position are:
- No EncryptedExtensions message yet
- No server side CertificateVerify message yet
- CertificateStatus still exists as a separate message
- A number of the messages are still in the TLSv1.2 format
- Still running on the TLSv1.2 record layer
Matt Caswell [Tue, 8 Nov 2016 16:10:21 +0000 (16:10 +0000)]
Remove old style NewSessionTicket from TLSv1.3
TLSv1.3 has a NewSessionTicket message, but it is *completely* different to
the TLSv1.2 one and may as well have been called something else. This commit
removes the old style NewSessionTicket from TLSv1.3. We will have to add the
new style one back in later.
Beat Bolli [Fri, 18 Nov 2016 08:47:50 +0000 (09:47 +0100)]
Use consistent variable names in example
In the X509_NAME_get_index_by_NID.pod example, the initialized variable is called
"loc", but the one used in the for loop is called "lastpos". Make the names match.
CLA: trivial Reviewed-by: Richard Levitte <levitte@openssl.org> Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1949)
Beat Bolli [Fri, 18 Nov 2016 09:17:28 +0000 (10:17 +0100)]
Update the location of the pod files
CLA: trivial
Since 99d63d4 ("Move manpages to man[1357] structure.", 2016-10-26), the location
of the pod files has changed. Reviewed-by: Richard Levitte <levitte@openssl.org> Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1951)
Andy Polyakov [Sun, 13 Nov 2016 21:32:09 +0000 (22:32 +0100)]
sha/asm/sha512-armv8.pl: fix big-endian support in __KERNEL__ case.
In non-__KERNEL__ context 32-bit-style __ARMEB__/__ARMEL__ macros were
set in arm_arch.h, which is shared between 32- and 64-bit builds. Since
it's not included in __KERNEL__ case, we have to adhere to official
64-bit pre-defines, __AARCH64EB__/__AARCH64EL__.
[If we are to share more code, it would need similar adjustment.]
Reviewed-by: Rich Salz <rsalz@openssl.org> Reviewed-by: Richard Levitte <levitte@openssl.org>
Nicola Tuveri [Fri, 7 Oct 2016 14:23:17 +0000 (17:23 +0300)]
Handle "int_ctx_new:unsupported algorithm" error
Calling EVP_PKEY_CTX_new_id(curve_NID, NULL) causes an error for most
curves that are implemented through the EC low-level API, and in the
last commit we call it for every curve to avoid treating X25519 as a
special case.
Last commit code already handles correctly this failure, but does not
remove these events from the thread error queue, thus some
false-positive warnings are printed at the end of execution.
This commit ensures that the error queue is clean, without flushing
other errors.
Reviewed-by: Richard Levitte <levitte@openssl.org> Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1658)