Switch libfuzzer to use trace-pc-guard

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Andy Polyakov <appro@openssl.org>
GH: #2683

Add -f -r flags to find-unused-errs

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3230)

Reformat evptests.txt

When compiling without EC support the test fails abruptly reading some keys.
Some keys merged in commit db04055 start with
------BEGIN EC PRIVATE KEY-----

this format is not supported without EC support.

This commit reformat those keys with the generic format. After this change the
test simply skips the unsupported EC keys when EC is disabled, without parsing
errors.

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3223)

Remove ecdhtest.c

All tests from ecdhtest.c have been ported to evptests.txt

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3219)

ecdhtest.c: move co-factor ECDH KATs to evptests

move NIST SP800-56A co-factor ECDH KATs from ecdhtest.c to evptests.txt

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3219)

WIP: Convert ui,v3ext,verify_extra_test

verify_extra_test still failing :(

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3194)

Fix unit tests when no-bf configured

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3206)

asn1/a_int.c: don't write result if returning error.

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3192)

asn1/a_int.c: simplify asn1_put_uint64.

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3192)

asn1/a_int.c: remove code duplicate and optimize branches,

i.e. reduce amount of branches and favour likely ones.

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3192)

testlib/OpenSSL/Test.pm: keep default input private.

If $_ is not private, it can wipe caller's one, which proved to be
problematic...

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

Guarantee single argument evaluation for test macros.
Add test case that checks some of them.

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3208)

Update the internal siphash tests to use the framework's output.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3212)

Split the CAST tests up.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3211)

ASN.1: adapt our use of INTxx et al by making them explicitely embedded

Fixes #3191

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3199)

ASN.1: change INTxx, UINTxx and Z variants to be embedable

Fixes #3191

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3199)

ASN.1: extend the possibilities to embed data instead of pointers

Also, when "allocating" or "deallocating" an embedded item, never call
prim_new() or prim_free().  Call prim_clear() instead.

Fixes #3191

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3199)

Test printing of ASN.1 types INTxx et al

[extended tests]

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3190)

OCSP Updates: error codes and multiple certificates

RT3877: Add X509 OCSP error codes and messages
Add additional OCSP error codes for X509 verify usage

RT3867: Support Multiple CA certs in ocsp app
Add the ability to read multiple CA certs from a single file in the
ocsp app.

Update some missing X509 errors in documentation.

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/941)

Correct travis.yml to only build extended tests when explicitely asked to

The check for this was done by checking if $TRAVIS_EVENT_TYPE is
"pull_request".  The trouble is that when new data is pushed to an
already existing pull request, the event type is "push".

Better then to go with another documented variable, $TRAVIS_PULL_REQUEST,
which is "false" in non-PR builds.

Ref: https://docs.travis-ci.com/user/environment-variables/#Default-Environment-Variables

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3205)

CT_POLICY_EVAL_CTX_set_time expects milliseconds, but given seconds

This resulted in the SCT timestamp check always failing, because the
timestamp appeared to be in the future.

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3138)

Add SSL tests for certificates with embedded SCTs

The only SSL tests prior to this tested using certificates with no
embedded Signed Certificate Timestamps (SCTs), which meant they couldn't
confirm whether Certificate Transparency checks in "strict" mode were
working.

These tests reveal a bug in the validation of SCT timestamps, which is
fixed by the next commit.

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3138)

Address review feedback (to be squashed)

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3189)

Convert x509aux, cipherlist, casttest

To new test framework

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3189)

OpenSSL::Test: supported filtered command output

95-test_external_boringssl.t had a specialised run() variant to prefix
the command output so it wouldn't disturb Test::Harness.  This
functionality if now moved to the run() command, using the added
option 'prefix' that can be set to the string to prefix the output
with.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3201)

Remove more stale code in ecdhtest.c

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3187)

ecdhtest.c: move KATs to evptests.txt

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3187)

Remove stale code in ecdhtest.c

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3187)

ecdhtest.c: move NAMED CURVES TESTS to evptests.txt

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3187)

Update destest to use the test infrastructure

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3173)

test/testutil.c: Flush stdout when running tests

Because stdout is usually buffered and stderr isn't, error output
might get printed in one bunch and all the lines saying which test
failed all in one bunch, making it difficult to see exactly what error
output belongs to what test.  Flushing stdout makes sure the runner
output is displayed together with the corresponding error output.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3198)

Remove fprintfs from the poly1305 internal test but keep the test number
information.

The framework will display the non-matching memory.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3156)

Update threadstest to use the test framework

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3196)

Update the internal chacha test to use the framework
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3195)

Add tests of custom negative 1

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3174)

Fix definition of i2d_fn in asn1_encode_test.c

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3174)

In asn1_encode_test.c, add custom DER encoding checks

We're already checking that custom DER decodes to expected values (or
fails to do so), but we didn't check if values encode back to expected
DER.

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3174)

Update sanitytest to use the test infrastructure

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3176)

 Update ideatest to use the test infrastructure

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3172)

Update gmdifftime to use the test infrastructure

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3171)

Update secmemtest and memeleaktest to use the test infrastructure.

It isn't easy to use the test framework since it turns memory debugging
on as well and the CRYPTO_mem_leaks_fp function cannot be called twice.

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3169)

Update d2i_test to use the test infrastructure

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3166)

Remove unused variable. Thanks @pauli-oracle

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3177)

Code review. Split tests up

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3177)

Convert blowfish tests to new framework

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3177)

Remove seed completely...

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3179)

Updates after code review

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3179)

Convert RSA tests to new framework

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3179)

Update rc4test to use the test infrastructure

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3162)

Update rc5test to use the test infrastructure

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3163)

Update mdc2test to use the test infrastructure

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3164)

 Update sha1test to use the test infrastructure

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3167)

 Update sha256t and sha512t to use the test infrastructure

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3168)

Convert exdata tests to new test framework

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3193)

Added error checking for OBJ_create

fixes segmentation fault in case of not enough memory for object creation

CLA: trivial

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3157)

Convert clienthellotest for the new test framework

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3186)

Convert asynciotest for the new test framework

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3182)

Convert bad_dtls_test for the new test framework

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3184)

Fix x_int64.c

Clearing a misunderstanding.  The routines c2i_uint64_int() and
i2c_uint64_int() expect to receive that internal values are absolute
and with a separate sign flag, and the x_int64.c code handles values
that aren't absolute and have the sign bit embedded.  We therefore
need to convert between absolute and non-absolute values for the
encoding of negative values to be correct.

[extended tests]

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3160)

Fix int64 test of t_4bytes_4_neg

{ 0x80, 0x00, 0x00, 0x00 } decoded isn't (positive) 0x80000000,
it's (negative) INT32_MIN.

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3160)

Remove ECDH(E) ciphers from SSLv3

SSLv3 does not support TLS extensions, and thus, cannot provide any
curves for ECDH(E). With the removal of the default (all) list of curves
being used for connections that didn't provide any curves, ECDHE is no
longer possible.

Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3181)

Additional check to handle BAD SSL_write retry

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3124)

Reject decoding of an INT64 with a value >INT64_MAX

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3159)

Update dtlstest to use the test infrastructure

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3170)

Update md2test to use the test infrastructure

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3165)

Update rc2test to use the test infrastructure

Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3161)

Make test marcos for true/false checks reliable for all integral types.

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3154)

Allow an ALPN callback to pretend to not exist

RFC 7301 mandates that the server SHALL respond with a fatal
"no_application_protocol" alert when there is no overlap between
the client's supplied list and the server's list of supported protocols.
In commit 062178678f5374b09f00d70796f6e692e8775aca we changed from
ignoring non-success returns from the supplied alpn_select_cb() to
treating such non-success returns as indicative of non-overlap and
sending the fatal alert.

In effect, this is using the presence of an alpn_select_cb() as a proxy
to attempt to determine whether the application has configured a list
of supported protocols.  However, there may be cases in which an
application's architecture leads it to supply an alpn_select_cb() but
have that callback be configured to take no action on connections that
do not have ALPN configured; returning SSL_TLSEXT_ERR_NOACK from
the callback would be the natural way to do so.  Unfortunately, the
aforementioned behavior change also treated SSL_TLSEXT_ERR_NOACK as
indicative of no overlap and terminated the connection; this change
supplies special handling for SSL_TLSEXT_ERR_NOACK returns from the
callback.  In effect, it provides a way for a callback to obtain the
behavior that would have occurred if no callback was registered at
all, which was not possible prior to this change.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2570)

Fix util/mkdef.pl

The deprecation checking code here didn't work the same way as in
Configure, and used $config{options} to find an --api= option that
was never there.  This is replaced with checking $config{api}, which
is the controlling variable for deprecation.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3126)

Add a note in CHANGES

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3126)

Act on deprecation of LONG and ZLONG, step 3

Teach util/mkdef.pl to recognise these lines:

    #if OPENSSL_API_COMPAT < 0xXXXXXXXXL

    #if OPENSSL_API_COMPAT >= 0xXXXXXXXXL

and add corresponding markers in util/*.num

A final 'make update' sets those markers right for LONG and ZLONG.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3126)

Act on deprecation of LONG and ZLONG, step 2

Replace all remaining uses of LONG and ZLONG with INT32 / ZINT32.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3126)

Act on deprecation of LONG and ZLONG, step 1

Don't compile code that still uses LONG when it's deprecated

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3126)

Discourage the use of LONG and ZLONG, and deprecate it in the future

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3126)

Publish our INT32, UINT32, INT64, UINT64 ASN.1 types and Z variants

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3126)

rand/rand_lib.c: keep fixing no-engine configuration.

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

asn1/x_long.c: remove conditions in inner loops and dependency on BN.

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3152)

appveyor.yml: split {build,test}_scripts to avoid exit code masking.

Last modification effectively masked test failures, so that builds
were reported successful even if they failed.

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

Use 'over 2' for bullet lists.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3117)

Standardize on =over 4 and check for it.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3117)

Add missing =back

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3117)

Check for L<foo|foo>

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3117)

Make default_method mostly compile-time (cont'd)

Forgot to include this commit as requested by review.

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2244)

Make default_method mostly compile-time

Document thread-safety issues
Have RSA_null return NULL (always fails)

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2244)

Fix rand_lib.c for no-engine configuration

When configured no-engine, we still refered to rand_engine_lock.
Rework the lock init code to avoid that.

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3145)

Fix a test failure when configured without TLSv1.3

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3139)

Remove an out of date TODO

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3139)

Create an ENDPOINT enum type for use internally

We need it for the custom extensions API

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3139)

Prefix custom extension API callback types with SSL_

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3139)

Various style tweaks based on feedback

Style updates for the new custom extensions API

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3139)

Add documentation for the new custom extensions API

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3139)

Add some tests for the new custom extensions API

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3139)

Implement a new custom extensions API

The old custom extensions API was not TLSv1.3 aware. Extensions are used
extensively in TLSv1.3 and they can appear in many different types of
messages. Therefore we need a new API to be able to cope with that.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3139)

Move the extensions context codes into the public API

This move prepares for the later addition of the new custom extensions
API. The context codes have an additional "SSL_" added to their name to
ensure we don't have name clashes with other applications.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3139)

Move ssl/t1_ext.c to ssl/statem/extensions_cust.c

Brings all the extensions code together.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3139)

Don't try to clean up RAND from ENGINE

This is especially harmful since OPENSSL_cleanup() has already called
the RAND cleanup function

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3137)

Make getting and setting the RAND default method thread safe

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3137)

e_os2.h: Refine OSSL_SSIZE definition under UEFI environment

Under UEFI build environment, we may encounter the OSSL_SSIZE macro
re-definition error in e_os2.h if any module call OpenSSL API directly
by including "openssl/xxxx.h" (caused by the predefined _WIN32/_WIN64
macro, which should have been un-defined under OPENSSL_SYS_UEFI).

Though it's not one recommended usage, this patch could still eliminate
the possible build issue by refining the OSSL_SSIZE definition under
OPENSSL_SYS_UEFI.

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3121)

Add support for MLOCK_ONFAULT to secure arena

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3115)

In rand_cleanup_int(), don't go creating a default method

If no default method was yet given, RAND_get_rand_method() will set it
up.  Doing so just to clean it away seems pretty silly, so instead,
use the default_RAND_meth variable directly.

This also clears a possible race condition where this will try to init
things, such as ERR or ENGINE when in the middle of a OPENSSL_cleanup.

Fixes #3128

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3136)