Implement SSL_read_ex() and SSL_write_ex() as documented.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
GH: #1964

Add an sslapitest for early callback

Make sure that we can stop handshake processing and resume it later.
Also check that the cipher list and compression methods are sane.
Unfortunately, we don't have the client-side APIs needed to force
a specific (known) session ID to be sent in the ClientHello, so
that accessor cannot be tested here.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2279)

Let test handshakes stop on certain errors

Certain callback APIs allow the callback to request async processing
by trickling a particular error value up the stack to the application
as an error return from the handshake function.  In those cases,
SSL_want() returns a code specific to the type of async processing
needed.

The create_ssl_connection() helper function for the tests is very
helpful for several things, including creating API tests.  However,
it does not currently let us test the async processing functionality
of these callback interfaces, because the special SSL error codes
are treated as generic errors and the helper continues to loop until
it reaches its maximum iteration count.

Add a new parameter, 'want', that indicates an expected/desired
special SSL error code, so that the helper will terminate when
either side reports that error, giving control back to the calling
function and allowing the test to proceed.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2279)

Use correct variable in test diagnostic

create_ssl_connection() prints out the results if SSL_accept() and/or
SSL_connect() fail, but was reusing the client return value when printing
about SSL_accept() failures.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2279)

Adopt test to changed behavior

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2279)

Tests for SSL early callback

Plumb things through in the same place as the SNI callback, since
we recommend that the early callback replace (and supplement) the
SNI callback, and add a few test cases.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2279)

Add SSL_CTX early callback

Provide a callback interface that gives the application the ability
to adjust the nascent SSL object at the earliest stage of ClientHello
processing, immediately after extensions have been collected but
before they have been processed.

This is akin to BoringSSL's "select_certificate_cb" (though it is not
API compatible), and as the name indicates, one major use is to examine
the supplied server name indication and select what certificate to
present to the client.  However, it can also be used to make more
sweeping configuration changes to the SSL object according to the
selected server identity and configuration.  That may include adjusting
the permitted TLS versions, swapping out the SSL_CTX object (as is
traditionally done in a tlsext_servername_callback), changing the
server's cipher list, and more.

We also wish to allow an early callback to indicate that it needs to perform
additional work asynchronously and resume processing later.  To that effect,
refactor the second half of tls_process_client_hello() into a subroutine to be
called at the post-processing stage (including the early callback itself), to
allow the callback to result in remaining in the same work stage for a later
call to succeed.  This requires allocating for and storing the CLIENTHELLO_MSG
in the SSL object to be preserved across such calls, but the storage is
reclaimed after ClientHello processing finishes.

Information about the CliehtHello is available to the callback by means of
accessor functions that can only be used from the early callback.  This allows
extensions to make use of the existing internal parsing machinery without
exposing structure internals (e.g., of PACKET), so that applications do not
have to write fragile parsing code.

Applications are encouraged to utilize an early callback and not use
a servername_callback, in order to avoid unexpected behavior that
occurs due to the relative order of processing between things like
session resumption and the historical servername callback.

Also tidy up nearby style by removing unnecessary braces around one-line
conditional bodies.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2279)

Prepare for WORK_MORE_C

Add the new enum value and case statements as appropriate.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2279)

Tests for SSL_bytes_to_cipher_list()

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2279)

Refactor SSL_bytes_to_cipher_list()

Split off the portions that mutate the SSL object into a separate
function that the state machine calls, so that the public API can
be a pure function.  (It still needs the SSL parameter in order
to determine what SSL_METHOD's get_cipher_by_char() routine to use,
though.)

Instead of returning the stack of ciphers (functionality that was
not used internally), require using the output parameter, and add
a separate output parameter for the SCSVs contained in the supplied
octets, if desired.  This lets us move to the standard return value
convention.  Also make both output stacks optional parameters.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2279)

Export SSL_bytes_to_cipher_list()

Move ssl_bytes_to_cipher_list() to ssl_lib.c and create a public
wrapper around it.  This lets application early callbacks easily get
SSL_CIPHER objects from the raw ciphers bytes without having to
reimplement the parsing code.  In particular, they do not need to
know the details of the sslv2 format ClientHello's ciphersuite
specifications.

Document the new public function, including the arguably buggy behavior
of modifying the supplied SSL object.  On the face of it, such a function
should be able to be pure, just a direct translation of wire octets to
internal data structures.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2279)

Let ssl_get_cipher_by_char yield not-valid ciphers

Now that we have made SCSVs into more of a first-class object, provide
a way for the bytes-to-SSL_CIPHER conversion to actually return them.
Add a flag 'all' to ssl_get_cipher_by_char to indicate that we want
all the known ciphers, not just the ones valid for encryption.  This will,
in practice, let the caller retrieve the SCSVs.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2279)

Add more first-class support for SCSVS

Just as we have a table of ssl3_ciphers, add a table of ssl3_scsvs, to contain
SSL_CIPHER objects for these non-valid ciphers.  This will allow for unified
handling of such indicators, especially as we are preparing to pass them around
between functions.

Since the 'valid' field is not set for the SCSVs, they should not be used
for anything requiring a cryptographic cipher (as opposed to something
being stuck in a cipher-shaped hole in the TLS wire protocol).

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2279)

Move CLIENTHELLO_MSG up in the header

We'll be adding a field of this type to struct ssl_st in a subsequent
commit, and need the type definition to be in scope already.
Also move up the RAW_EXTENSION definition that it depends on.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2279)

Store the number of extensions in CLIENTHELLO_MSG

Keep track of the length of the pre_proc_exts array.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2279)

output number of exts from tls_collect_extensions()

Modify the API of tls_collect_extensions() to be able to output the number of
extensions that are known (i.e., the length of its 'res' output).  This number
can never be zero on a successful return due to the builtin extensions list,
but use a separate output variable so as to not overload the return value
semantics.

Having this value easily available will give consumers a way to avoid repeating
the calculation.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2279)

Check for the presence of _WIN32 rather than its value.

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2727)

In apps/rehash.c, decorate the inclusion of internal/o_dir.h for VMS

The library files are built with symbol names as is, while the
application is built with the default uppercase-all-symbols mode.
That's fine for public APIs, because we have __DECC_INCLUDE_PROLOGUE.H
and __DECC_INCLUDE_EPILOGUE.H automatically telling the compiler how
to treat the public header files.  However, we don't have the same
setup for internal library APIs, since they are usually only used by
the libraries.

Because apps/rehash.c uses a library internal header file, we have to
surround that inclusion with the same kind of pragmas found in
__DECC_INCLUDE_PROLOGUE.H and __DECC_INCLUDE_EPILOGUE.H, or we get
unresolved symbols when building no-shared.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2725)

Add a test case that tests more of the cipher modes.

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2715)

On VMS, massage the fetch file names to remove the generation number

The generation number is ';nnn' at the end of the file name fetched
with readdir().  Because rehash checks for specific extensions and
doesn't expect an additional generation number, the easiest is to
massage the received file name early by simply removing the generation
number.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2717)

Let the output from 'openssl enc -ciphers' go to stdout

Also, don't exit with an error code

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2716)

Fix typo, should be && rather than &

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2689)

Remove -Wno-parentheses-equality from gcc --strict-warnings options.
There has never been any gcc option of that kind.

Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2705)

Fix typo, missing ||

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2707)

Iterate over EC_GROUP's poly array in a safe way

Prevent that memory beyond the last element is accessed if every element
of group->poly[] is non-zero

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2689)

Make "openssl rehash" work on VMS 8.3 and up

A spelling error prevented it from building correctly.
Furthermore, we need to be more careful when to add a / at the end
of the dirname and when not.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2706)

Have the directory reader use the Unix API on VMS

opendir(), readdir() and closedir() have been available on VMS since
version 7.0.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2707)

apps/req.c: flag "-new" is implied by "-precert"

Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/843)

Change CA.pl flag from --newprecert to --precert

Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/843)

Documentation for the -precert flag for "openssl req"

Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/843)

Basic test for "openssl req -precert" via apps/CA.pl

TODO(robpercival): Should actually test that the output certificate
contains the poison extension.

Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/843)

Adds a "-precert" flag to "openssl req" for creating pre-certificates

This makes it a little easier to create a pre-certificate.

Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/843)

Add some more consistency checks in tls_decrypt_ticket.

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2704)

Correct the no-dh and no-dsa fix

The condition wasn't quite right

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2702)

Fix i2d_SSL_SESSION pp output parameter should point to end of asn1 data.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2607)

appveyor.yml: engage VC-WIN64A-masm.

One of the reasons for why masm/ml64 is not [fully] supported is that
it's problematic to support multiple versions. But latest one usually
works and/or it's lesser problem to make it work. So idea here is to
have a "whistle" when it breaks, so that problems can be evaluated as
they emerge. It's kind of "best effort" thing, as opposite to "full
support".

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

Fix memory leak in pkcs12 -export

Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2676)

Fix some more memory leaks with TXT_DB_insert.

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2684)

Fix a few memleaks in TXT_DB.

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2684)

Provided support for the -nameopt flag in s_client, s_server and s_time
commands.

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2695)

Prevent OOB in SRP base64 code.

Change size comparison from > (GT) to >= (GTE) to ensure an additional
byte of output buffer, to prevent OOB reads/writes later in the function
Reject input strings larger than 2GB
Detect invalid output buffer size and return early

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2672)

Update client, server and x509 fuzz corpus

Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
GH: #2682

Check validity, not just signing for all certificates

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2679)

Add no siglags test for ECDSA certificate

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2679)

Set default validity flags.

Set default validity flags if signature algorithms extension
is not present. Preserve flags when checking chains.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2679)

remove md array: it is not used any more.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2679)

Removed ugly size_t less than zero check.

CLA: trivial.

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2674)

Ensure minsize >= sizeof(SH_LIST)

The sh_add_to_list function will overwrite subsequent slots in the free list
for small allocations.  This causes a segmentation fault if the writes goes
off the end of the secure memory.  I've not investigated if this problem
can overwrite memory without the segmentation fault, but it seems likely.

This fix limits the minsize to the sizeof of the SH_LIST structure (which
also has a side effect of properly aligning the pointers).

The alternative would be to return an error if minsize is too small.

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2657)

fix spelling of Camellia in comment

Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2337)

Implementation of the ARIA cipher as described in RFC 5794.

This implementation is written in endian agnostic C code. No attempt
at providing machine specific assembly code has been made. This
implementation expands the evptests by including the test cases from
RFC 5794 and ARIA official site rather than providing an individual
test case. Support for ARIA has been integrated into the command line
applications, but not TLS. Implemented modes are CBC, CFB1, CFB8,
CFB128, CTR, ECB and OFB128.

Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2337)

Added '-nameopt' option to the verify command.

It makes possible to print the certificate's DN correctly in case of verification errors.

Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2656)

Don't call memcpy if len is zero.

Prevent undefined behavior in CRYPTO_cbc128_encrypt: calling this function
with the 'len' parameter being 0 would result in a memcpy where the source
and destination parameters are the same, which is undefined behavior.
Do same for AES_ige_encrypt.

Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2671)

Revert "Use memcmp() instead of CRYPTO_memcmp() when fuzzing"

This reverts commit 3aad8e18707bccaabee5f111de2db0696b45781c.

Reviewed-by: Andy Polyakov <appro@openssl.org>
GH: #2686

.travis.yml: remove osx from build matrix.

Travis OS X utilization and backlog statistics suggest that it became
bottleneck for our integration builds with requests piling up for days
during working days of the week. Suggestion is to remove osx till
capacity is lesser issue.

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

Use memcmp() instead of CRYPTO_memcmp() when fuzzing

Reviewed-by: Andy Polyakov <appro@openssl.org>
GH: #2633

Internal siphash tests are not run.

Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2649)

VMS fix of test/recipes/80-test_ssl_new.t

On VMS, file names with more than one period get all but the last get
escaped with a ^, so 21-key-update.conf.in becomes 21-key-update^.conf.in
That means that %conf_dependent_tests and %skip become useless unless
we massage the file names that are used as indexes.

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2678)

Fix no-dh and no-dsa

Since 20-cert-select.conf will vary depending in no-dh and no-dsa,
don't check it against original when those options are selected

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2680)

If all versions of a proto are disabled, disabled the proto as well

For example, 'no-dtls1 no-dtls1_2' will imply 'no-dtls'

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2670)

test/README: clarify last test number group

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

Explicitly disallow DSA for TLS 1.3

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2667)

add DSA cert tests

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2667)

Add DH parameters, DSA cert and key

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2667)

Add DSA support to mkcert.sh

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2667)

Add and use function test_pem to work out test filenames.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2667)

Fix a slightly confusing if condition in a2i_ASN1_INTEGER.

Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2640)

Fix a merge error

Fix an error code clash due to a merge

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2665)

Fix test_x509_store

Don't run this test unless 'openssl rehash' works properly.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2664)

Fix symbol shadow

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2663)

Updates following review feedback

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2609)

Don't use an enum in the return type for a public API function

We use an int instead. That means SSL_key_update() also should use an int.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2609)

Add some KeyUpdate tests

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2609)

Add documentation for SSL_key_update() and SSL_get_key_update_type()

This also adds documentation for the pre-existing and related
SSL_renegotiate*() functions.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2609)

Fix a shadowed global variable warning

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2609)

Updates following review feedback

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2609)

Limit the number of KeyUpdate messages we can process

Too many KeyUpdate message could be inicative of a problem (e.g. an
infinite KeyUpdate loop if the peer always responds to a KeyUpdate message
with an "update_requested" KeyUpdate response), or (conceivably) an attack.
Either way we limit the number of KeyUpdate messages we are prepared to
handle.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2609)

Actually update the keys when a KeyUpdate message is sent or received

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2609)

If we receive an "update_requested" KeyUpdate then respond with a KeyUpdate

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2609)

Add the ability for a server to receive a KeyUpdate message

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2609)

Add the ability for s_client to send a KeyUpdate message

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2609)

Add a SSL_get_key_update_type() function

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2609)

Add the ability for a client to send a KeyUpdate message

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2609)

Add the ability for a client to receive a KeyUpdate message

This just receives the message. It doesn't actually update any keys yet.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2609)

Add SSL_trace() support for KeyUpdate messages

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2609)

Add a capability to s_server to send KeyUpdate messages

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2609)

Provide a function to send a KeyUpdate message

This implements the server side KeyUpdate sending capability as well.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2609)

Fix typo in x86_64-mont5.pl CFI directives

Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2655)

Add a test of the X509_STORE / X509_LOOKUP API

Fortunately, "openssl verify" makes good use of that API

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2652)

test/README: clarify test number groups

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2652)

Fix a mem leak in ssl_test_ctx.c

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2650)

Add client auth TLS 1.3 certificate selection tests

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2643)

Add ECDSA client certificates

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2643)

Use tls_choose_sigalg for client auth.

For client auth call tls_choose_sigalg to select the certificate
and signature algorithm. Use the selected algorithm in
tls_construct_cert_verify.

Remove obsolete tls12_get_sigandhash.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2643)

Add client side support to tls_choose_sigalg.

Select appropriate signature algorithm and certificate for client
authentication using tls_choose_sigalg.

A lot of selection logic is very similar except not finding a
certificate is not a fatal error: we just do not present a
certificate.

For TLS 1.2 and earlier we only check the current certificate
is suitable (for compatibility with previous logic) for TLS 1.3
(where there are no compatibility issues) we support multiple
client certificates for different algorithms.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2643)

Change req_check_len error message, it also accepts 20 bytes, but states 'less than' in the error message

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2648)

Use _WIN32 over WIN32 for preprocessor conditional

The intent seems to be that the WIN32 symbol is for things that are a direct
byproduct of being a windows-variant configuration and should be used for
feature en/disablement on windows systems.  Use of the _WIN32 symbol is more
widespread, being used to implement platform portability of more generic code.

We do define WIN32 in some situations in e_os.h, but that is not included
universally.

Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2642)

Update CHANGES and NEWS for new release

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

Remove an OPENSSL_assert() and replace with a soft assert and check

Following on from CVE-2017-3733, this removes the OPENSSL_assert() check
that failed and replaces it with a soft assert, and an explicit check of
value with an error return if it fails.

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

Don't change the state of the ETM flags until CCS processing

In 1.1.0 changing the ciphersuite during a renegotiation can result in
a crash leading to a DoS attack. In master this does not occur with TLS
(instead you get an internal error, which is still wrong but not a security
issue) - but the problem still exists in the DTLS code.

The problem is caused by changing the flag indicating whether to use ETM
or not immediately on negotiation of ETM, rather than at CCS. Therefore,
during a renegotiation, if the ETM state is changing (usually due to a
change of ciphersuite), then an error/crash will occur.

Due to the fact that there are separate CCS messages for read and write
we actually now need two flags to determine whether to use ETM or not.

CVE-2017-3733

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

Provide a test for the Encrypt-Then-Mac renegotiation crash

In 1.1.0 changing the ciphersuite during a renegotiation can result in
a crash leading to a DoS attack. In master this does not occur with TLS
(instead you get an internal error, which is still wrong but not a security
issue) - but the problem still exists in the DTLS code.

This commit provides a test for the issue.

CVE-2017-3733

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