Use the enc_flags field to determine whether we should use explicit IV,
signature algorithms or SHA256 default PRF instead of hard coding which
versions support each requirement.
(cherry picked from commit cbd64894)

Conflicts:

	ssl/ssl_locl.h

(cherry picked from commit 6de2649a)

Revise DTLS code. There was a *lot* of code duplication in the
DTLS code that generates records. This makes it harder to maintain and
sometimes a TLS update is omitted by accident from the DTLS code.

Specifically almost all of the record generation functions have code like
this:

some_pointer = buffer + HANDSHAKE_HEADER_LENGTH;
... Record creation stuff ...
set_handshake_header(ssl, SSL_MT_SOMETHING, message_len);

...

write_handshake_message(ssl);

Where the "Record creation stuff" is identical between SSL/TLS and DTLS or
in some cases has very minor differences.

By adding a few fields to SSL3_ENC to include the header length, some flags
and function pointers for handshake header setting and handshake writing the
code can cope with both cases.

(cherry picked from commit 173e72e6)

Add various functions to allocate and set the fields of an ECDSA_METHOD
structure.
(cherry picked from commit 94c2f77a)

Force no SSL2 when custom extensions in use.
Don't clear extension state when cert is set.
Clear on renegotiate.

Conflicts:
	ssl/t1_lib.c

OS X 10.8..10.8.3 has broken support for ECDHE-ECDSA ciphers.

Merge branch 'OpenSSL_1_0_2-stable' of /usr/local/google/home/bmoeller/openssl/openssl into OpenSSL_1_0_2-stable

    - EC_GROUP_cmp shouldn't consider curves equal just because
      the curve name is the same. (They really *should* be the same
      in this case, but there's an EC_GROUP_set_curve_name API,
      which could be misused.)

    - EC_POINT_cmp shouldn't return 0 for ERR_R_SHOULD_NOT_HAVE_BEEN_CALLED
      or EC_R_INCOMPATIBLE_OBJECTS errors because in a cmp API, 0 indicates
      equality (not an error).

    Reported by: king cope

(cherry picked from commit 312a46791ab465cfa3bf26764361faed0e5df014)

    - EC_GROUP_cmp shouldn't consider curves equal just because
      the curve name is the same. (They really *should* be the same
      in this case, but there's an EC_GROUP_set_curve_name API,
      which could be misused.)

    - EC_POINT_cmp shouldn't return 0 for ERR_R_SHOULD_NOT_HAVE_BEEN_CALLED
      or EC_R_INCOMPATIBLE_OBJECTS errors because in a cmp API, 0 indicates
      equality (not an error).

    Reported by: king cope

(cherry picked from commit 312a46791ab465cfa3bf26764361faed0e5df014)

PR: 3125
Submitted by: Kyle McMartin
(cherry picked from commit 8e52a906)

(cherry picked from commit 139cd16c)

Avoid occasional up to 8% performance drops.
(cherry picked from commit 7a1a1223)

(cherry picked from commit 72a15870)

This reverts commit 514f1a78.

(cherry picked from commit ce455596)

When verifying a partial path always check to see if the EE certificate
is explicitly trusted: the path could contain other untrusted certificates.
(cherry picked from commit 52073b76)

(cherry picked from commit c3eb3376)

(cherry picked from commit b093a06866bf632a97a9a0286e2d08f69c3cf7dd)

(cherry picked from commit 3a918ea2bbf4175d9461f81be1403d3781b2c0dc)

(cherry picked from commit f7ac0ec8)

Backport of ASN1_TIME_diff and OPENSSL_gmtime_diff functions from master
branch.

(cherry picked from commit 51b9115b)

is needed to test some profiles/protocols which reject certificates
with unsupported versions.
(cherry picked from commit df316fd4)

(cherry picked from commit 96cfba0f)

(cherry picked from commit 7c8ac505)

(cherry picked from commit b5cadfb5)

(cherry picked from commit 68575593)

just like a "real" server making it easier to trace any problems.

(manually applied from commit 35b0ea4e)

(cherry picked from commit 14536c8c)