Allow NULL for some _free routines.

Based on the description in https://github.com/openssl/openssl/pull/5757,
this re-implements the "allow NULL to be passed" behavior of a number of
xxx_free routines.  I also fixed up some egregious formatting errors
that were nearby.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5761)

o_time.c: use gmtime_s with MSVC
ts/ts_rsp_sign.c: change to OPENSSL_gmtime.

Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5720)

doc/man1/x509.pod: corrected "S/MIME signing" requirements

When the "certificate purpose" is checked and KeyUsage extension is present,
either 'digitalSignature' or 'nonRepudiation' is accepted.

Manual page corrected to reflect the above.

Signed-off-by: Konstantin Shemyak <konstantin@shemyak.com>
Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5727)

statem/statem_clnt.c: omit redundant check in tls_construct_client_hello.

Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5729)

rand/randfile.c: permit non-regular files in RAND_load_file.

Apparently applications rely on RAND_load_file's ability to work with
non-regular files, customarily with /dev/urandom, so that the ban was
not exactly appropriate.

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5737)

Configure: make LIST command work with dynamic 15-android.conf.

This is quick-n-dirty ad-hoc solution, the problem asks for more
elegant one...

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5755)

Temporarily disable some tests that hang

The previous commit causes some tests to hang so we temporarily disable them.

Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
(Merged from https://github.com/openssl/openssl/pull/5757)

Remove some code for a contributor that we cannot find

This removes some code because we cannot trace the original contributor
to get their agreement for the licence change (original commit e03ddfae).

After this change there will be numerous failures in the test cases until
someone rewrites the missing code.

All *_free functions should accept a NULL parameter. After this change
the following *_free functions will fail if a NULL parameter is passed:

BIO_ACCEPT_free()
BIO_CONNECT_free()
BN_BLINDING_free()
BN_CTX_free()
BN_MONT_CTX_free()
BN_RECP_CTX_free()
BUF_MEM_free()
COMP_CTX_free()
ERR_STATE_free()
TXT_DB_free()
X509_STORE_free()
ssl3_free()
ssl_cert_free()
SSL_SESSION_free()
SSL_free()

[skip ci]

Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
(Merged from https://github.com/openssl/openssl/pull/5757)

Assert that alpn_selected is NULL before we assign it

The alpn_selected value in the session should be NULL before we first
populate it if this is a new session. We assert to make sure it is.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5715)

Don't write out a bad OID

If we don't have OID data for an object then we should fail if we
are asked to encode the ASN.1 for that OID.

Fixes #5723

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5725)

Update CHANGES and NEWS for the new release

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

test/test_test.c: add CRYPTO_memcmp regression test.

Reviewed-by: Matt Caswell <matt@openssl.org>

pariscid.pl: fix nasty typo in CRYPTO_memcmp.

Comparison was effectively reduced to least significant bits.

CVE-2018-0733

Reviewed-by: Matt Caswell <matt@openssl.org>

Add fuzz corpora file that found the ASN.1 stack depth issue

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

Limit ASN.1 constructed types recursive definition depth

Constructed types with a recursive definition (such as can be found in
PKCS7) could eventually exceed the stack given malicious input with
excessive recursion. Therefore we limit the stack depth.

CVE-2018-0739

Credit to OSSFuzz for finding this issue.

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

Remove QNX support

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5756)

Configurations/README: mention Libs.private in ex_libs description.

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

Previously this x509 command line was working, restore that

openssl x509 -in server.pem -signkey privkey.pem -out server.pem

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5746)

Fix dsaparam -genkey with DER outform

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5744)

Fix ecparam -genkey with point compression or DER outform

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5741)

Adds multiple checks to avoid buffer over reads

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
(Merged from https://github.com/openssl/openssl/pull/5675)

Configurations/windows-makefile.tmpl: refine clean targets.

'nmake clean' was leaving some artefacts behind.

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

test/shlibloadtest.c: make some variables block local

myDSO_dsobyaddr and myDSO_free are only used in a narrow block of
code, and can therefore be made local to that block.  Otherwise, some
compilers may warn that they are unused.

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5733)

test/shlibloadtest.c: fix various errors

These errors were hidden because compiling this file didn't get the
macros derived from the dso_scheme attribute, and therefore, some code
never got compiled.

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5733)

Include "internal/dso_conf.h" where needed and appropriate

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5733)

Move the handling of dso_scheme to dso_conf.h

The macros resulting from the dso_scheme attribute were defined for
libraries only, but there's a test program that uses the macros as
well.  The easier way is to move the handling of this macro to
crypto/include/internal/dso_conf.h and having the modules that need it
include it.

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5733)

Configurations/unix-Makefile.tmpl: fix inconsistencies with resource files

.res files weren't treated consistently.  They weren't included at all
in the build of a shared library, and were named inconsistently; .res
sometimes, and .res.o otherwise.  Now we standardise it to .res.o,
which is the recommended way with GNU tools.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5730)

Configurations/10-main.conf: correct AIX targets.

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5713)

Handle partial messages in TLSProxy

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5412)

Fix resource files

Add it to apps as well as libraries.
Fix the copyright year generation.
Thanks to user RTT for pointing this out.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5704)

include/openssl/rand.h: omit intermediate typedef.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4159)

openssl/rand.h: fix formatting.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4159)

Remove presumably temporary left-over file.

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

man3/OPENSSL_ia32cap.pod: update assembler version requirements.

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

curve448/field.h: relax alignment, as it doesn't work universally.

Some platforms, cough-DJGPP, fail to compile claiming that requested
alignment is greater than maximum possible. Supposedly original
alignment was result of an attempt to utilize AVX2...

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5708)

bio/bss_log.c: on DJGPP syslog facility is part of sockets library.

In other words no-sock DJGPP build should suppress syslogging.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5708)

o_fopen.c: compensate for e_os.h omission.

At earlier point e_os.h was omitted from a number of headers (in order
to emphasize OS neutrality), but this affected o_fopen.c, which is not
OS-neutral, and contains some DJGPP-specific code.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5708)

Don't call strsignal, just print the signal number.
The strsignal call is not supported by some machines, so avoid its use.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5658)

aes ctr_drbg: add cavs tests

Signed-off-by: Patrick Steuer <patrick.steuer@de.ibm.com>
Reviewed-by: Kurt Roeckx <kurt@roeckx.be>
Reviewed-by: Rich Salz <rsalz@openssl.org>
GH: #5580

DRBG: Use the EVP layer to do AES encryption

Reviewed-by: Rich Salz <rsalz@openssl.org>
GH: #5580

Unify s_client/s_server srtp profiles option handling

Add missing guards around STRP-related fields
Remove two unneeded global variables: my 2'cents to #4679
Merge definition and instantiation of srpsrvparm global.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4908)

The default conv_form is uncompressed

Fixes #5711

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5712)

Fixes integer underflow with SSL_trace support

Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5706)

Document EC_POINT_get_affine_coordinates_*.

In particular, x and y may be NULL, as used in ecdsa_ossl.c. Make use of
this in ecdh_ossl.c as well, to save an otherwise unnecessary temporary.

Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5532)

Cleanup the s_time command.

Various code-cleanups.
Use SSL_CTX_set_mode(ctx, SSL_MODE_AUTO_RETRY) insead of handling
SSL_ERROR_WANT_READ everywhere.
Turn off the linger option on connected sockets to avoid failure.
Add BIO_set_conn_mode(conn, BIO_SOCK_NODELAY) to improve thruput.

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3952)

Convert _meth_get_ functions to const getters

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2181)

Don't wait for dry at the end of a handshake

For DTLS/SCTP we were waiting for a dry event during the call to
tls_finish_handshake(). This function just tidies up various internal
things, and after it completes the handshake is over. I can find no good
reason for waiting for a dry event here, and nothing in RFC6083 suggests
to me that we should need to. More importantly though it seems to be
wrong. It is perfectly possible for a peer to send app data/alerts/new
handshake while we are still cleaning up our handshake. If this happens
then we will never get the dry event and so we cannot continue.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5084)

Check for alerts while waiting for a dry event

At a couple of points in a DTLS/SCTP handshake we need to wait for a dry
event before continuing. However if an alert has been sent by the peer
then we will never receive that dry event and an infinite loop results.

This commit changes things so that we attempt to read a message if we
are waiting for a dry event but haven't got one yet. This should never
succeed, but any alerts will be processed.

Fixes #4763

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5084)

Fix stack-use-after-scope

Fixes regression from #5667.

Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5701)

o_fopen.c,rand/randfile.c: compensate for e_os.h omission.

At earlier point e_os.h was omitted from a number of headers (in order
to emphasize OS neutrality), but this affected o_fopen.c and randfile.c
which are not OS-neutral, and contain some Win32-specific code.

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5676)

Add dladdr() for AIX

Although it deviates from the actual prototype of DSO_dsobyaddr(), this
is now ISO C compliant and gcc -Wpedantic accepts the code.

Added DATA segment checking to catch ptrgl virtual addresses. Avoid
memleaks with every AIX/dladdr() call. Removed debug-fprintf()s.
Added test case for DSO_dsobyaddr(), which will eventually call dladdr().
Removed unecessary AIX ifdefs again.

The implementation can only lookup function symbols, no data symbols.
Added PIC-flag to aix*-cc build targets.

As AIX is missing a dladdr() implementation it is currently uncertain our
exit()-handlers can still be called when the application exits. After
dlclose() the whole library might have been unloaded already.

Signed-off-by: Matthias Kraft <makr@gmx.eu>
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5668)

Do not cache sessions with zero sid_ctx_length when SSL_VERIFY_PEER

The sid_ctx is something of a "certificate request context" or a
"session ID context" -- something from the application that gives
extra indication of what sort of thing this session is/was for/from.
Without a sid_ctx, we only know that there is a session that we
issued, but it could have come from a number of things, especially
with an external (shared) session cache.  Accordingly, when resuming,
we will hard-error the handshake when presented with a session with
zero-length sid_ctx and SSL_VERIFY_PEER is set -- we simply have no
information about the peer to verify, so the verification must fail.

In order to prevent these future handshake failures, proactively
decline to add the problematic sessions to the session cache.

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5175)

Remove mention of link between message digests and public key algorithms.

The comment in EVP_DigestInit.pod is:

> Returns the NID of the public key signing algorithm associated with this
digest. For example EVP_sha1() is associated with RSA so this will return
B<NID_sha1WithRSAEncryption>. Since digests and signature algorithms are no
longer linked this function is only retained for compatibility reasons.

I.e. there is no link anymore.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5682)

aix compat fixes for ocsp.c

WCOREDUMP and vsyslog are not portable

Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5657)

Make pkeyutl a bit more user-friendly

Give meaningful error messages when the user incorrectly uses pkeyutl.

Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3987)

Reduce the verbosity of test_store

The travis logs are going above 4Mb causing the builds to fail. One
test creates excessive output. This change reduces that output by approx
180k.

[extended tests]

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5694)

Fix the OCSP responder mode

Broken by commit 3e3c7c36.

Fixes #5681

Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
(Merged from https://github.com/openssl/openssl/pull/5688)

Prepare for 1.1.1-pre4-dev

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

Prepare for 1.1.1-pre3 release

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

Update copyright year

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5689)

crypto/rand/rand_vms.c: include "internal/rand_int.h"

Without it, the RAND_POOL typedef is missing

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5685)

Fix: drbgtest fails when tests are executed in random order

[extended tests]

The test_rand_reseed assumed that the global DRBGs were not used
previously. This assumption is false when the tests are executed
in random order (OPENSSL_TEST_RAND_ORDER). So we uninstantiate
them first and add a test for the first instantiation.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5680)

s_client, s_server: do generic SSL configuration first, specialization after

We did the SSL_CONF_cmd() pass last of all things that could affect
the SSL ctx.  However, the results of this, for example:

    -max_protocol TLSv1.3 -tls1_2

... would mean that the protocol min got set to TLSv1.2 and the
protocol max to TLSv1.3, when they should clearly both be TLSv1.2.

However, if we see the SSL_CONF_cmd() switches as generic and those
internal to s_client and s_server as specialisations, we get something
that makes a little more sense.

Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5679)

Fix no-sm3/no-sm2 (with strict-warnings)

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5677)

Fix no-sm3 (and no-sm2)

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5677)

Don't generate buildtest_*err.c

The error string header files aren't supposed to be included directly,
so there's no point testing that they can.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5678)

Enhance ssltestlib's create_ssl_ctx_pair to take min and max proto version

Have all test programs using that function specify those versions.
Additionally, have the remaining test programs that use SSL_CTX_new
directly specify at least the maximum protocol version.

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5663)

In TLSProxy::Proxy, specify TLSv1.3 as maximum allowable protocol

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5663)

Support "-min_protocol" and "-max_protocol" in s_server and s_client

If for nothing else, they are needed when doing a regression test

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5663)

Fix no-ec

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5673)

Fix no-sm2

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5673)

Fix no-posix-io compile failure

The fix in conf_include_test.c  seems to be required because some
compilers give an error if you give an empty string for the second
argument to strpbrk(). It doesn't really make sense to send an empty
string for this argument anyway, so make sure it has at least one character
in it.

Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
(Merged from https://github.com/openssl/openssl/pull/5666)

Place ticket keys into secure memory

Place the session ticket AES and HMAC keys into secure memory.

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2351)

Fix no-psk

Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5667)

Revise and cleanup; use strict,warnings

Use shorter names for some defines, so also had to change the .c file
that used them.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5669)

Apply system_default configuration on SSL_CTX_new().

When SSL_CTX is created preinitialize it with system default
configuration from system_default section.

Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4848)

Add a multithread rand test

Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
(Merged from https://github.com/openssl/openssl/pull/5547)

Don't use a ssl specific DRBG anymore

Since the public and private DRBG are per thread we don't need one
per ssl object anymore. It could also try to get entropy from a DRBG
that's really from an other thread because the SSL object moved to an
other thread.

Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
(Merged from https://github.com/openssl/openssl/pull/5547)

Make the public and private DRBG thread local

This avoids lock contention.

Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
(Merged from https://github.com/openssl/openssl/pull/5547)

Handle evp_tests assumption of EVP_PKEY_FLAG_AUTOARGLEN

Without actually using EVP_PKEY_FLAG_AUTOARGLEN

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4793)

Support SM2 ECIES scheme via EVP

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4793)

Add SM2 signature and ECIES schemes

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4793)

Configurations/15-android.conf: detect clang by PATH, not by CC.

Since they intend to omit gcc, it's more appropriate to simply detect
if there is NDK's clang on PATH, as opposite to requiring to specify it
with CC=clang (and looking for it on PATH).

Also detect NDK version and default to armv7-a for NDK>16.

Address failure to recognize -D__ADNDROID_API__=N in CPPFLAGS.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5613)

Add NOTES.ANDROID.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5613)

Configurations/15-android.conf: default to RC4_CHAR whenever possible.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5613)

MIPS assembly pack: default heuristic detection to little-endian.

Current endianness detection is somewhat opportunistic and can fail
in cross-compile scenario. Since we are more likely to cross-compile
for little-endian now, adjust the default accordingly.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5613)

Configurations/15-android.conf: refine clang support.

Adjusting ARM default broke clang support, and x86[_64] needed
path adjustment.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5613)

Configure: pass -no-integrated-as.

Occasionally you have to pass -no-integrated-as to clang, but we
consider any -no-option as no-option. Don't touch -no-integrated-as.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5613)

Fix bio callback backward compatibility

Don't pass a pointer to uninitialized processed value
for BIO_CB_READ and BIO_CB_WRITE

Check the correct cmd code in BIO_callback_ctrl

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5516)

Fix a crash in SSLfatal due to invalid enc_write_ctx

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5645)

Fix no-cmac

Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5665)

Fix no-ec

Raw private/public key loading may fail for X25519/X448 if ec has been
disabled.

Also fixed a missing blank line in evppkey.txt resulting in a warning in
the test output.

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5664)

Add a CHANGES entry to mention the replay protection capabilities

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5644)

Document the replay protection capabilities

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5644)

Add a test for 0RTT replay protection

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5644)

Add an anti-replay mechanism

If the server is configured to allow early data then we check if the PSK
session presented by the client is available in the cache or not. If it
isn't then this may be a replay and we disallow it. If it is then we allow
it and remove the session from the cache. Note: the anti-replay protection
is not used for externally established PSKs.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5644)

Don't update the session cache when processing a client certificate in TLSv1.3

We should only update the session cache when we issue a NewSessionTicket.
These are issued automatically after processing a client certificate.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5644)

Always call the new_session_cb when issuing a NewSessionTicket in TLSv1.3

Conceptually in TLSv1.3 there can be multiple sessions associated with a
single connection. Each NewSessionTicket issued can be considered a
separate session. We can end up issuing multiple NewSessionTickets on a
single connection at the moment (e.g. in a post-handshake auth scenario).
Each of those issued tickets should have the new_session_cb called, it
should go into the session cache separately and it should have a unique
id associated with it (so that they can be found individually in the
cache).

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5644)

engines/asm/e_padlock*: add support for Zhaoxin's x86 platform

VIA and Shanghai United Investment Co.,Ltd. found Shanghai ZhaoXin,
which is a fabless x86 CPU IC design company. ZhaoXin has issued
ZX-C, ZX-D x86 processors, which have 'Shanghai' CPU vendor id.

Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5640)

Clarify a couple of details around "make variables"

Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5660)