Use curve_id not the nid

Found by OSS-Fuzz and the tests

Reviewed-by: Andy Polyakov <appro@openssl.org>
GH: #4410

Remove dhparam from SSL_CONF list.

Avoid duplicate assertion by removing dhparam from SSL_CONF parameter list:
dhparam is handled manually by s_server.

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4408)

Fix 'key' option in s_server can be in ENGINE keyform

Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Stephen Henson <steve@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4405)

Store groups as uint16_t

Instead of storing supported groups in on-the-wire format store
them as parsed uint16_t values. This simplifies handling of groups
as the values can be directly used instead of being converted.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4406)

Configure: add -Wmisleading-indentation to strict warnings flags.

The warning flag in question was added in GCC version 6, hence
addition has to be conditional.

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/4401)

Guard against DoS in name constraints handling.

This guards against the name constraints check consuming large amounts
of CPU time when certificates in the presented chain contain an
excessive number of names (specifically subject email names or subject
alternative DNS names) and/or name constraints.

Name constraints checking compares the names presented in a certificate
against the name constraints included in a certificate higher up in the
chain using two nested for loops.

Move the name constraints check so that it happens after signature
verification so peers cannot exploit this using a chain with invalid
signatures. Also impose a hard limit on the number of name constraints
check loop iterations to further mitigate the issue.

Thanks to NCC for finding this issue. Fix written by Martin Kreichgauer.

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4393)

Cleanup whitespace in ssl_lib.c (tabs to spaces)

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4383)

Fix strict-warnings build

Compilation failed due to -Werror=misleading-indentation.

Signed-off-by: Patrick Steuer <patrick.steuer@de.ibm.com>
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/4395)

Configure: unify clang's -Qunused-arguments option treatment.

Detect clang even if it's disguised, e.g. cross-compiler or invoked by
explicit path name, and add the option based on that.

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/4383)

Reenable s_server -dhparam option

This option was lost when converting to a table-driven option parser
in commit 7e1b7485706c2b11091b5fa897fe496a2faa56cc.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4398)

Add RSA-PSS certificate type TLS tests

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/4368)

Add RSA-PSS test certificates

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/4368)

Allow use of RSA-PSS certificates in TLS 1.2

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/4368)

Allow RSA certificates to be used for RSA-PSS

Allo RSA certificate to be used for RSA-PSS signatures: this needs
to be explicit because RSA and RSA-PSS certificates are now distinct
types.

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/4368)

Add RSA-PSS key certificate type.

Recognise RSA-PSS certificate algorithm and add a new certificate
type.

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/4368)

Fix overflow in c2i_ASN1_BIT_STRING.

c2i_ASN1_BIT_STRING takes length as a long but uses it as an int.  Check
bounds before doing so. Previously, excessively large inputs to the
function could write a single byte outside the target buffer. (This is
unreachable as asn1_ex_c2i already uses int for the length.)

Thanks to NCC for finding this issue. Fix written by Martin Kreichgauer.

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4385)

Stack sorting safety

Use the defined typechecking stack method to sort the compression methods stack
rather than using the generic function and apply type casts.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4382)

Null pointer used.
Address coverity report of null pointer being dereferenced.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4381)

Provide getters for min/max proto version

OpenSSL 1.1.0 made SSL_CTX and SSL structs opaque and introduced a new
API to set the minimum and maximum protocol version for SSL_CTX with
TLS_method(). Add getters to introspect the configured versions:

  int SSL_CTX_get_min_proto_version(SSL_CTX *ctx);
  int SSL_CTX_get_max_proto_version(SSL_CTX *ctx);
  int SSL_get_min_proto_version(SSL *ssl);
  int SSL_get_max_proto_version(SSL *ssl);

NOTE: The getters do not resolv the version in case when the minimum or
maxium version are configured as '0' (meaning auto-select lowest and
highst version number).

Signed-off-by: Christian Heimes <christian@python.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/4364)

Avoid signed vs unsigned comparison error.
Introduced by #4372

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4374)

Add explanatory comment about fitting into a size_t.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4373)

Remote unrequited casts
Also use strndup instead of a malloc/memcpy pair.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4372)

Replace malloc+strcpy with strdup

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4371)

Fix function name in ECerr call

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4371)

Fix no-ec no-dh build

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4369)

Fix doc nits

[to be squashed]
[skip ci]

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4356)

Support EVP_PKEY_meth_remove and pmeth internal cleanup

1. make app pkey methods cleanup internal
2. add EVP_PKEY_meth_remove

Fixes travis-ci failure in #4337

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4356)

Ensure that the requested memory size cannot exceed the limit imposed by a
size_t variable.

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4357)

Manually revert "Ensure allocation size fits into size_t"

This reverts commit cc9c56894606fdf324933cd8090d9a54d967bf5b for the file
pbe_scrypt.c instead of scrypt.c

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4357)

Revert "Reuse strndup(), simplify code"

This reverts commit 1caaea133873d549fa52fbf265298d2d35442477.

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4357)

Revert "GH614: Use memcpy()/strdup() when possible"

This reverts commit a89c9a0d855bce735116acfe147b24e386f566ba.

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4357)

Fix typo in comment

The peek result is now called 'err', not just 'e'.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4362)

Support key check in EVP interface

A new method is added to EVP_PKEY_METH as:

    int (*check) (EVP_PKEY_CTX *ctx);

and to EVP_PKEY_ASN1_METHOD as:

    int (*pkey_check) (EVP_PKEY_CTX *ctx);

This is used to check the validity of a specific key.

The order of calls is:
EVP_PKEY_check -> pmeth.check -> ameth.pkey_check.

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4337)

Always use $ as shell prompt in example

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4359)

Document the recursive option

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4336)

Add a recursive option to 'openssl storeutl'

Simply put, any NAME type OSS_STORE_INTO is a new object that can be
looked into, and potentially lead to a whole tree of data to dive
into.  The recursive option allows someone to view the whole tree and
its data in one go.

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4336)

Disable the EGD seeding meachanism when stdio is disabled

crypto/rand/rand_egd.c makes extensive use of stdio functions.  When
they are disabled, it makes sense to disable egd as well.

Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4358)

rsa/rsa_pmeth.c: recognize SHA3 as supported RSA digest.

Reviewed-by: Richard Levitte <levitte@openssl.org>

evp/m_sha3.c: wire SHA3 to rsaEncryption.

Reviewed-by: Richard Levitte <levitte@openssl.org>

objects/obj_xref.txt: cross-reference SHA3 and rsaEncryption.

Reviewed-by: Richard Levitte <levitte@openssl.org>

objects/objects.txt: fix RSA-SHA3-284 typo.

Reviewed-by: Richard Levitte <levitte@openssl.org>

Update rsautl.pod for typo

Fixes the typo
CLA: trivial

Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4354)

SSL Trace improvements

A fix formatting fixes.

SSL Trace: internal constification
-  static trace tables
-  trace methods arguments

SSL Trace: enhance error message when tracing an invalid extension packet
... instead of just "Message length parse error!".

SSL trace: add Maximum-Fragment-Length TLS extension log support

Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4353)

sha/asm/keccak1600-armv8.pl: fix return value buglet and ...

... 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>

Introduce named constants for the ClientHello callback.

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)

Fix test documentation.

The instructions don't work.

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/4349)

Rename SSL_CTX_set_early_cb to SSL_CTX_set_client_hello_cb.

"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)

Add UI functions to set result with explicit length and to retrieve the length

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)

Document default section and library configuration

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)

Allow an endpoint to read the alert data before closing the socket

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)

Fix error handling/cleanup

Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/4326)

Fix test_client_hello with no-tls1_2_method

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)

Restore historical behavior for absent ServerHello extensions

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
7da160b0f46d832dbf285cb0b48ae56d4a8b884d, 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)

Add checks for alloc failing.

Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/4341)

add callback handler for setting DTLS timer interval

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
(Merged from https://github.com/openssl/openssl/pull/4011)

Fix OSSL_STORE's 'file' loader: make sure peekbuf is initialised

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)

Don't use ciphersuites for inflating the ClientHello in clienthellotest

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)

Don't attempt to add a zero length padding extension

The padding extension should always be at least 1 byte long

Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/4331)

Fix 90-test_store.t: using config() requires OpenSSL::Test::Utils

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4332)

Fix an include location problem in the extrended tests.
[extended tests]

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4327)

OSSL_STORE: Avoid testing with URIs on the mingw command line

Some URIs get "mistreated" (converted) by the MSYS run-time.
Unfortunately, avoiding this conversion doesn't help either.

    http://www.mingw.org/wiki/Posix_path_conversion

Fixes #4314

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4322)

Fix doc-nits from previous commit

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4325)

config: get "stty technique" working again on MacOS X.

Addresses GH#2167.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4311)

test/run_tests.pl: don't use Module::Load::Conditional.

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)

Less documentation for deprecated API

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)

Fix OpenSSL::Test::Utils::config to actualy load the config data

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4319)

Fix long SNI lengths in test/handshake_helper.c

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)

Fixup include path in ossl_shim test after e_os.h work

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)

Tighten up SSL_get1_supported_ciphers() docs

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)

Fix Proxy where a timeout occurs waiting for both client and server

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4305)

crypto/cryptlib.c: mask more capability bits upon FXSR bit flip.

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)

ssl/statem/extensions_clnt.c: fix return code buglet.

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4297)

ssl/statem/*.c: address "enum mixed with another type" warnings.

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)

struct timeval include guards

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)

Address feedback

Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/4294)

Add CRYPTO_thread_glock_new

Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/4294)

Fix potential null problem.

Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4313)

Avoid possible uninitialized variable.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4307)

OpenSSL::Test::__fixup_prg: don't check program existence

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)

util/mkdef.pl: handle line terminators correctly

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)

Various review fixes for PSK early_data support

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/3926)

Test for late client side detection of ALPN inconsistenties

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/3926)

Client side sanity check of ALPN after server has accepted early_data

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/3926)

Add some fixes for Travis failures

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/3926)

Add PSK early_data tests

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/3926)

Add server side sanity checks of SNI/ALPN for use with early_data

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/3926)

Make sure we save ALPN data in the session

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/3926)

Complain if we are writing early data but SNI or ALPN is incorrect

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)

Add functions for getting/setting SNI/ALPN info in SSL_SESSION

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/3926)

Show the error stack if there was an error writing early data in s_client

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/3926)

Update the tests for SNI changes

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 no SNI has been explicitly set use the one from the session

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)

Make sure we use the correct cipher when using the early_secret

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/3926)

Add HISTORY and SEE ALSO sections for the new TLSv1.3 PSK functions

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/3926)

Add documentation for SSL_SESSION_set_max_early_data()

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/3926)

Add some PSK early_data tests

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/3926)

Add SSL_SESSION_set_max_early_data()

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/3926)

Enable the ability to use an external PSK for sending early_data

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/3926)

Fixed address family test error for AF_UNIX in BIO_ADDR_make

CLA: trivial

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4298)

Add documentation for ARIA GCM modes.

Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4300)