Move certificate table to header file so it can be tested.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3858)

Add additional ECDSA/Ed25519 selection tests.

Add two tests with ECDSA+SHA256 preferred over Ed25519, the second also
excludes P-256 from the supported curves extension which will force the
use of Ed25519 in TLS 1.2, but not TLS 1.3: this would fail before the
certificate table updates.

Add TLS 1.3 test also with P-256 exclude from the groups extension: this
should have no effect as the groups extension is not used for signature
selection in TLS 1.3

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3858)

Use cert tables in ssl_set_sig_mask

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3858)

make errors

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3858)

Use cert tables instead of X509_certificate_type

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3858)

Use certificate tables instead of ssl_cert_type

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3858)

Use certificate tables instead of ssl_cipher_get_cert_index.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3858)

Replace tls12_get_pkey_idx

The functiontls12_get_pkey_idx is only used to see if a certificate index is
enabled: call ssl_cert_is_disabled instead.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3858)

Add SSL_aCERT: this is used for any ciphersuite with a certificate.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3858)

Add certificate properties table.

Add certificate table giving properties of each certificate index:
specifically the NID associated with the index and the the auth mask
value for any cipher the certificate can be used with.

This will be used to generalise certificate handling instead of hard coding
algorithm specific cases.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3858)

Demo style fixes and modernisation.

Address some style issues in the demos and modernise the C.
Fix the exit/return from main handling.

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3914)

coding style: remove extra whitespace charactor

CLA: trivial

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/3913)

Rewrite RAND_egd

Use stdio and its buffering.
Limit to 255 bytes (could remove that if neceessary).

Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/3888)

Fix error handling in get_header_and_data.

Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/3896)

Fix crash in BUF_MEM_grow_clean.

Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/3896)

Optimize sha/asm/keccak1600-avx2.pl.

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

Add sha/asm/keccak1600-avx2.pl.

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

Trivial bounds checking.

Bounds checking strpy, strcat and sprintf.
These are the remaining easy ones to cover a recently removed commit.
Some are trivial, some have been modified and a couple left as they are because the reverted change didn't bounds check properly.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3871)

Fix cipher_compare

Unsigned overflow.  Found by Brian Carpenter

Fixes #3889

Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3890)

Update various RAND podpages

Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3883)

Fix s_client crash where the hostname is provided as a positional arg

If the hostname is provided as a positional arg then s_client crashes.
The crash occurs as s_client exits (after either a successful or
unsuccessful connection attempt).

This issue was introduced by commit 729ef85611.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3881)

Some SSL_OP_ values can't be used in 1.1.x

SSL_OP_ALL was set in 0x0BFF so reusing some of these bits would cause
ABI compatibility issues.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3833)

Choose a safer value for SSL_OP_ALLOW_NO_DHE_KEX

1.1.0 included the previous value for SSL_OP_ALLOW_NO_DHE_KEX in
SSL_OP_ALL. This might cause binary compatibility issues. We should choose
a value that is not in SSL_OP_ALL.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3833)

Update the documentation for the new SSL_OP_ALLOW_NO_DHE_KEX option

Also the associated configuration parameters and command line switches.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3833)

Do not allow non-dhe kex_modes by default

Allow that mode to be configured if desired.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3833)

Updates following review feedback of TLSv1.3 draft-21 code

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/3852)

Update SSL_trace() to know about ticket_nonce

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/3852)

Update the early_secret generation to use the new ticket_nonce field

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/3852)

Update the test/session.pem to have a tick_nonce value

Otherwise the ClientHello test fails

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/3852)

Update tls13_hkdf_expand() to take the length of the data

In most scenarios the length of the input data is the hashsize, or 0 if
the data is NULL. However with the new ticket_nonce changes the length can
be different.

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/3852)

The correct key length for a TLSv1.3 SHA384 ciphersuite is 48

Our test was using 32. The latest ticket nonce changes now validate this
value and so sslapitest was failing.

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/3852)

Send and receive the ticket_nonce field in a NewSessionTicket

This just adds the processing for sending and receiving the newly added
ticket_nonce field. It doesn't actually use it yet.

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/3852)

Update the version number for TLSv1.3 draft 21

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/3852)

Fix memory leak when using PSK session files

We were not freeing the session created when loading a PSK session file.

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
(Merged from https://github.com/openssl/openssl/pull/3855)

test/run_tests.pl: Make sure to exit with a code that's understood universally

TAP::Parser::Aggregator::has_errors may return any number, not just 0
and 1.  With Perl on VMS, any number from 2 and on is interpreted as a
VMS status, the 3 lower bits are the encoded severity (1 = SUCCESS,
for example), so depending on what has_errors returns, a test failure
might be interpreted as a success.  Therefore, it's better to make
sure the exit code is 0 or 1, nothing else (they are special on VMS,
and mean SUCCESS or FAILURE, to match Unix conventions).

Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3880)

test/recipes/90-test_shlibload.t: Make sure to handle library renames

VMS renames our libraries to fit VMS conventions.  This must be accounted
for when we want to load them.

Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3880)

VMS: When running a sub-MMS, make sure to give it the main MMS' qualifiers

Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3880)

Add sha/asm/keccak1600-avx512.pl.

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
(Merged from https://github.com/openssl/openssl/pull/3861)

Address potential buffer overflows.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3878)

change return (x) to return x

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3878)

Rearrange link line so the libraries come after the source.
Some linkers like it this way.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3879)

Remove some now-unneeded VMS controls

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3875)

BIO range checking.

Add length limits to avoid problems with sprintf, strcpy and strcat.  This replaces recently removed code but also guards some previously missing function calls (for DOS & Windows).

Reworked the BIO_dump_indent_cb code to reduce temporary storage.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3870)

Cleanup RAND_load_file,RAND_write_file

Document an internal assumption that these are only for use with files,
and return an error if not. That made the code much simpler.
Leave it as writing 1024 bytes, even though we don't need more than 256
from a security perspective.  But the amount isn't specified, now, so we
can change it later if we want.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3864)

Memory bounds checking in asn1 code.

Check that sprint, strcpy don't overflow.

Avoid some strlen operations when the previous sprintf return value can be used.

Also fix the undefined behaviour `*(long *)x = y` when x isn't a long or character pointer.
ISO/IEC 9899:1999 6.5/7 for the details.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3869)

Bounds check string functions in apps.
This includes strcat, strcpy and sprintf.

In the x509 app, the code has been cleaned up as well.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3868)

Add two trivial fixes from old commits

Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3865)

Rework the append_buf function

It won't overflow the buffer and will allocate new buffers sufficiently large to
hold new strings longer than the expansion factor.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3847)

Avoid buffer overruns in the req command line utility.

Clean up some of the formatting "return x" instead of "return (x)" mostly.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3848)

Fix crash

[extended tests]

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3700)

Undo commit dc00fb9

Original text:
    Document openssl dgst -hmac option

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3700)

Undo commit cd359b2

Original text:
    Clarify use of |$end0| in stitched x86-64 AES-GCM code.

    There was some uncertainty about what the code is doing with |$end0|
    and whether it was necessary for |$len| to be a multiple of 16 or 96.
    Hopefully these added comments make it clear that the code is correct
    except for the caveat regarding low memory addresses.

    Change-Id: Iea546a59dc7aeb400f50ac5d2d7b9cb88ace9027
    Reviewed-on: https://boringssl-review.googlesource.com/7194
Reviewed-by: Adam Langley <agl@google.com>
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3700)

Undo commit 40720ce

Comment in the commit:
    /* Ignore NULLs, thanks to Bob Beck <beck@obtuse.com> */

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3700)

Undo commit de02ec2

Original text:
    Check if a random "file" is really a device file, and treat it
    specially if it is.
    Add a few OpenBSD-specific cases.
    This is part of a large change submitted by Markus Friedl <markus@openbsd.or

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3700)

Undo commit 0755217

Original text:
    Fix Perl problems on sparc64.
    This is part of a large change submitted by Markus Friedl
    <markus@openbsd.org>

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3700)

STORE 'file' scheme loader: fix try_decode_params() to check ambiguity

The way try_decode_params works in raw more, it would take the first ASN1
that could decode and return a STORE_INFO with the resulting EVP_PKEY.
This change has it go through all the matching ASN1 methods and properly
check if there's more than one match, i.e. an ambiguity.

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/3863)

Error out when forcing an unsupported TLS version

If the result of a SSL_{CTX_,}set_{min,max}_proto_version() call
leaves the min and max version identical, and support for that version
is compiled out of the library, return an error.  Such an object has
no hope of successfully completing a handshake, and this error may
be easier to decipher than the resulting handshake failure.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3422)

Improve BN_CTX documentation

Since BN_CTX_init() is gone, all calls use BN_CTX_new().  Also,
essentially all consumers will use BN_CTX_start()/BN_CTX_end(),
so make that more clear from the BN_CTX_new() man page.

Document the thread-unsafety of individual BN_CTX objects.

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3682)

Fix small UI issues

- in EVP_read_pw_string_min(), the return value from UI_add_* wasn't
  properly checked
- in UI_process(), |state| was never made NULL, which means an error
  when closing the session wouldn't be accurately reported.

Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/3849)

Avoid possible memleak in X509_policy_check()

When tree_calculate_user_set() fails, a jump to error failed to
deallocate a possibly allocated |auth_nodes|.

Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/3850)

Fix compiler warnings

Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/3839)

Undo commit d420ac2

[extended tests]

Original text:
    Use BUF_strlcpy() instead of strcpy().
    Use BUF_strlcat() instead of strcat().
    Use BIO_snprintf() instead of sprintf().
    In some cases, keep better track of buffer lengths.
    This is part of a large change submitted by Markus Friedl <markus@openbsd.org>

Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/3701)

Remove the TEST_check macro.

This macro aborted the process which stopped any later tests from running.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3838)

Add echo for end of each build phase

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3842)

STORE: fix possible memory leak

If scheme is NULL, the allocated res is leaked

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3841)

Use the return value from write(2)
This prevents a warning when building with crypto-mdebug.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3836)

BN_pseudo_rand is really BN_rand

And BN_pseudo_rand_range is really BN_rand_range.
Document that we might deprecate those functions.

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3743)

'make update' after objects/objects.txt update.

Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
(Merged from https://github.com/openssl/openssl/pull/3643)

objects/objects.txt: add SHA3 OIDs.

Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
(Merged from https://github.com/openssl/openssl/pull/3643)

Correct documentation for UI_get0_result_string

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/3824)

sha/keccak1600.c: internalize KeccakF1600 and simplify SHA3_absorb.

Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>

sha/asm/keccak1600-x86_64.pl: close gap with Keccak Code Package.

[Also typo and readability fixes. Ryzen result is added.]

Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>

sha/asm/keccak1600-s390x.pl: typo and readability, minor size optimization.

Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>

x86_64 assembly pack: fill some blanks in Ryzen results.

Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>

Various doc fixes

Fix a =head1 section name
Fix a typo in POD label
Remove a spurious =back
Add a missing blank line
Avoid 'legacy' -- use 'deprecated' if still needed if we cannot just reword.
Always do strict checking
Do not warn about missing "RETURN VALUES" unless -s is set.
Change OpenSSL version 1.1 -> 1.1.0

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3830)

Fix potential crash in tls_construct_finished.

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3667)

Fix a crash in tls_construct_client_certificate.

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3667)

Fix TLSv1.3 exporter

We need to use the hashsize in generating the exportsecret not 0! Otherwise
we end up with random garbage for the secret.

It was pure chance that this passed the tests previously. It so happens
that, because we call SSL_export_keying_material() repeatedly for different
scenarios in the test, we end up in the tls13_export_keying_material() at
exactly the same position in the stack each time and therefore end up using
the same random garbage secret each time!

Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3810)

Remove the possibility to disable the UI module entirely

Instead, make it possible to disable the console reader that's part of
the UI module.  This makes it possible to use the UI API and other UI
methods in environments where the console reader isn't useful.

To disable the console reader, configure with 'no-ui-console' /
'disable-ui-console'.

'no-ui' / 'disable-ui' is now an alias for  'no-ui-console' /
'disable-ui-console'.

Fixes #3806

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3820)

Make sure OSSL_STORE_load() isn't caught in an endless loop

The post process callback might potentially say "no" to everything (by
constantly returning NULL) and thereby cause an endless loop.  Ensure
that we stop all processing when "eof" is reached.

Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/3823)

Fix copyright date for the ARIA evp file.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3831)

Update fuzz corpora

Reviewed-by: Rich Salz <rsalz@openssl.org>
GH: #3829

STORE 'file' scheme loader: DNS name in URI is case insensitive

...  so compare accordingly with "//localhost"

Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
(Merged from https://github.com/openssl/openssl/pull/3827)

When apps_startup() fails, exit with a failure code and a message

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3816)

Cleanup some copyright stuff

Remove some incorrect copyright references.
Move copyright to standard place
Add OpenSSL copyright where missing.
Remove copyrighted file that we don't use any more
Remove Itanium assembler for RC4 and MD5 (assembler versions of old and
weak algorithms for an old chip)
Standardize apps/rehash copyright comment; approved by Timo
Put dual-copyright notice on mkcert

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3691)

Fix build with no-threads no-ec

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3817)

Fix atfork flag.  Avoid double-negatives :)

Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
(Merged from https://github.com/openssl/openssl/pull/3815)

Merge Intel copyright notice into standard

This is done with the kind permission of Intel.

Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3813)

Add ECHO to makefiles for real silence

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3801)

Drop support for OPENSSL_NO_TLS1_3_METHOD

There are no public TLSv1_3_*method() functions so
OPENSSL_NO_TLS1_3_METHOD doesn't make any sense and should be removed.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3800)

Add fork handlers, based on pthread_atfork

Only for Unix platforms

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3754)

STORE: Make sure the loader to be registered is complete

Most of the loader function pointers are crucial, they must be defined
unconditionally.  Therefore, let's make sure OSSL_STORE_register_loader
refuses to register incomplete loaders

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/3805)

STORE: simplify store_loader_cmp()

We have already made sure that the loader scheme isn't NULL, so
checking if they are NULL or not when comparing registered loaders
is redundant.  We still soft assert it, just to be entirely sure.

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/3805)

util/mkdef.pl: Make symbol version processing Linux only

For Windows, we care which way it is, the resulting file is just a pile
of symbols.  For VMS, we really need to care about the numeric ordering,
and getting the symbols sorted by symbol version too didn't agree with
that.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3804)

util/mkdef.pl: Add UNIX as a platform

This allows us to guard Unix specific functions with
#ifndef / #ifdef OPENSSL_SYS_UNIX

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3804)

Add sha/asm/keccak1600-s390x.pl.

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

sha/asm/keccak1600-x86_64.pl: add CFI directives.

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

sha/asm/keccak1600-x86_64.pl: optimize by re-ordering instructions.

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

sha/asm/keccak1600-x86_64.pl: remove redundant moves.

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

Add sha/asm/keccak1600-x86_64.pl.

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