Apps: don't build deprecated DH and DSA apps.

This also means that there doesn't need to be any conditional checks in the .c
files to avoid deprecated builds.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/11067)

Fix coverity issues

CID : 1458169 RESOURCE LEAK ffc_params_generate.c - False positive, but addressed another Leak in adjacent code
CID : 1458171 UNCHECKED RET apps/pkeyutl.c
CID : 1458173 DEAD CODE ffc_params_generate.c
CID : 1458174 RESOURCE LEAK ssl_lib.c

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
(Merged from https://github.com/openssl/openssl/pull/11048)

x86_64: Always generate .note.gnu.property section for ELF outputs

We should always generate .note.gnu.property section in x86_64 assembly
codes for ELF outputs to mark Intel CET support since all input files
must be marked with Intel CET support in order for linker to mark output
with Intel CET support.  Also .note.gnu.property section in x32 should
be aligned to 4 bytes, not 8 bytes and .p2align should be used
consistently.

Verified with

$ CC="gcc -Wl,-z,cet-report=error" ./Configure shared linux-x86_64 -fcf-protection
$ make
$ make test

and

$ CC="gcc -mx32 -Wl,-z,cet-report=error" ./Configure shared linux-x32 -fcf-protection
$ make
$ make test # <<< 90-test_sslapi.t failed because 8-byte pointer size.

Fix #10896

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/10985)

Fix a mem leak in libssl

Make sure we free up any ENGINE references after we have finished using
them.

Fixes #11064

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/11070)

Add Asymmetric RSA cipher tests in FIPS provider

We also add some DSA tests for code that is already in the FIPS provider.

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/10881)

Make the RSA ASYM_CIPHER implementation available inside the FIPS module

RSA ASYM_CIPHER was already available within the default provider. We
now make it also available from inside the FIPS module.

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/10881)

Add assembly config targets for UEFI build

Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/11054)

Fix no-ec build

Don't attempt to build ecx related source files in a "no-ec" build.

Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/11065)

Add OSSL_SERIALIZER_PUBKEY_TO_DER_PQ and friends

It's already used internally, there's no reason the DER serializer
propqueries shouldn't be present alongside the PEM and TEXT ones.

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/11055)

fix build for new HTTP client in case OPENSSL_NO_CMP or OPENSSL_NO_OCSP

fix also formatting nits w.r.t. #if indentations in ocsp.h

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: David von Oheimb <david.von.oheimb@siemens.com>
(Merged from https://github.com/openssl/openssl/pull/11058)

Remove unused ossl_param_bld_to_param_ex() function.

The recently introduced ossl_param_bld_to_param_ex() function is only
called by the unit tests.

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/11053)

Remove unused OSSL_PARAM_construct_from_text() function.

This function is recently introduced and never called by the library or tests.

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/11053)

Add NEWS entry about deprecation of command line public tools

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/10977)

dsa: deprecate applications that depend on the low level DSA functions.

speed is updated to not support DSA instead of being removed.

The dhparam, dsaparam, dsa and gendsa commands are deprecated but still
exist without NO_DEPRECATED defined.

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/10977)

app: add a deprecation warning to all deprecated commands.

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/10977)

test/recipes/80-test_ssl_old.t: Replace 'openssl gendsa'

Use 'openssl genpkey' instead.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/10977)

test_dsa: fix deprecation logic

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/10977)

test/recipes/15-test_dsa.t: Deal with deprecation of 'openssl dsa'

Do not run programs that depend on deprecated APIs when
'no-deprecated' is configured.

We still retain the conversion tests that use 'openssl pkey', and add
the one that's missing.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/10977)

Deprecate the low level DSA functions.

Use of the low level DSA functions has been informally discouraged for a
long time. We now formally deprecate them.

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/10977)

dsa.h: fix preprocessor indentation

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/10977)

DSA: fix the DSA parameter logic in test.

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/10977)

Add S390 support for provider based X25519/X448

Reviewed-by: Patrick Steuer <patrick.steuer@de.ibm.com>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/10964)

Add X25519/X448 Key Exchange to the default provider

Reviewed-by: Patrick Steuer <patrick.steuer@de.ibm.com>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/10964)

Implement Provider side Key Management for X25519 and X448

Reviewed-by: Patrick Steuer <patrick.steuer@de.ibm.com>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/10964)

Implement a stricter ECX_KEY type

Add ref counting and control how we allocate storage for the private key.
We will need this type in following commits where we move the ecx code
to be provider aware.

Reviewed-by: Patrick Steuer <patrick.steuer@de.ibm.com>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/10964)

Check that ed25519 and ed448 are allowed by the security level

Signature algorithms not using an MD weren't checked that they're
allowed by the security level.

Reviewed-by: Matt Caswell <matt@openssl.org>
GH: #10785

Generate new Ed488 certificates

Create a whole chain of Ed488 certificates so that we can use it at security
level 4 (192 bit). We had an 2048 bit RSA (112 bit, level 2) root sign the
Ed488 certificate using SHA256 (128 bit, level 3).

Reviewed-by: Matt Caswell <matt@openssl.org>
GH: #10785

Add a minimal build target for Travis and Appveyor

[extended tests]

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9982)

X509_PUBKEY_set(): Fix memory leak

With the provided method of creating the new X509_PUBKEY, an extra
EVP_PKEY is created and needs to be properly cleaned away.

(note: we could choose to keep it just as well, but there are
consequences, explained in a comment in the code)

Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/11038)

Generalize the HTTP client so far implemented mostly in crypto/ocsp/ocsp_ht.c
The new client has become an independent libcrpyto module in crypto/http/ and
* can handle any types of requests and responses (ASN.1-encoded and plain)
* does not include potentially busy loops when waiting for responses but
* makes use of a new timeout mechanism integrated with socket-based BIO
* supports the use of HTTP proxies and TLS, including HTTPS over proxies
* supports HTTP redirection via codes 301 and 302 for GET requests
* returns more useful diagnostics in various error situations
Also adapts - and strongly simplifies - hitherto uses of HTTP in crypto/ocsp/,
crypto/x509/x_all.c, apps/lib/apps.c, and apps/{ocsp,s_client,s_server}.c

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: David von Oheimb <david.von.oheimb@siemens.com>
(Merged from https://github.com/openssl/openssl/pull/10667)

add BIO_socket_wait(), BIO_wait(), and BIO_connect_retry() improving timeout support

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: David von Oheimb <david.von.oheimb@siemens.com>
(Merged from https://github.com/openssl/openssl/pull/10667)

PROV: Ensure the AlgorithmIdentifier registers in DSA signature impl

When setting up the hash function for DSA signature, the encoded
AlgorithmIdentifier for the DSA+hash combination is queried, but not
stored, which leads to problems when signing ASN.1 items in libcrypto.

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/11037)

Memory allocator code cleanup

Typedefs of CRYPTO malloc, realloc and free.

MEM_CHECK "modes" are used only as a CRYPTO_mem_ctrl() parameter

The CRYPTO_mem_ctrl is defined only if OPENSSL_NO_CRYPTO_MDEBUG is
defined, thus define the MEM_CHECK modes under the same condition.

Maybe the macros can be removed at all since:
1. CRYPTO_mem_ctrl() just returns -1 and ignores the parameter
2. CRYPTO_mem_ctr() is declared as DEPRECATED by 3.0

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/11042)

x86: Always generate .note.gnu.property section for ELF outputs

We should always generate .note.gnu.property section in x86 assembly
codes for ELF outputs to mark Intel CET support since all input files
must be marked with Intel CET support in order for linker to mark output
with Intel CET support.

Verified with

$ CC="gcc -Wl,-z,cet-report=error" ./Configure shared linux-x86 -fcf-protection
$ make
$ make test

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/11044)

Don't compile AESNI code if we're not AESNI capable

Compile failures were occuring on systems that weren't AESNI capable
because the detection wasn't quite right in a couple of files.

This fixes a run-checker build failure for the 386 compile option.

Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/11022)

Fix no-des

Don't attempt to self-test DES in the FIPS provider if we have been built
without FIPS support.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/11034)

Fix no-engine

We don't need to check if an engine has a cipher/digest in a no-engine
build.

Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
(Merged from https://github.com/openssl/openssl/pull/11031)

Fix no-dh

Don't use DH specific macros that might need to be used in a no-dh build.

Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
(Merged from https://github.com/openssl/openssl/pull/11030)

Fix no-ec

Recent SM2 related changes were not properly guarded with OPENSSL_NO_EC

Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
(Merged from https://github.com/openssl/openssl/pull/11029)

Use swapcontext for Intel CET

When Intel CET is enabled, makecontext will create a different shadow
stack for each context.  async_fibre_swapcontext cannot use _longjmp.
It must call swapcontext to swap shadow stack as well as normal stack.

Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/10983)

Adapt all build.info and test recipes to the new $disabled{'deprecated-x.y'}

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/11027)

Configure: Add easy to use disabled deprecated functionality indicators

In C, we have macros like OPENSSL_NO_DEPRECATED_3_0 to check if some
section of code should be disabled to simulate a removal of things
deprecated in the version.

In perl, we had to check $disabled{deprecated} and compare
$config{api} with the proper version code, which is doable but tedious
and error prone.

This change adds $disabled{'deprecated-x.y'} (x.y being a version
number) which directly corresponds to OPENSSL_NO_DEPRECATED_x_y, for
use in build.info conditions, test recipes and other perl stuff.

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/11027)

tests/drbgtest: use new RAND_DRBG callback_data API instead of ex_data

It took me a little while to realize why the test_rand_drbg_reseed test
kept crashing after replacing the RAND_DRBG_{gs}et_ex_data() calls by
RAND_DRBG_{gs}et_callback_data().

The reason was that the ex_data API prohibits modifying the callbacks
or callback data of chained DRBGs and returned an error which was
ignored by the `test_rand_drbg_reseed` test, for good reasons.

The `test_rand_drbg_reseed` test is special in this respect, because
it needs to install callbacks for all DRBGs, in order to intercept
and count the reseeding events.

Since the drbgtest module has access to the internal structures of
the DRBG anyway, the problem could be solved by accessing the members
directly. I added a warning comment in hook_drbg().

[extended tests]

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/10950)

RAND_DRBG: add a callback data for entropy and nonce callbacks

The callback data allows passing context specific data from the
application of the DRBG to to the entropy callbacks.
This a rather specialized feature which is useful for implementing
known answer tests (KATs) or deterministic signatures (RFC6979),
which require passing a specified entropy and nonce for instantiating
the DRBG.

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/10950)

Appveyor: update to Visual Studio 2017.

Default image was currently "Visual Studio 2015"

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
(Merged from https://github.com/openssl/openssl/pull/10327)

KEYMGMT: Require both get_params and gettable_params, or none

The same go for the pairs import + import_types and export + export_types.

This required some additional changes in our KEYMGMT implementations.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/11006)

Adapt existing SERIALIZER implementations to the redesigned interface

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/11006)

Adapt test/keymgmt_internal_test.c to the redesigned interface

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/11006)

Adapt existing KEYMGMT implementations to the redesigned interface

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/11006)

Redesign the KEYMGMT libcrypto <-> provider interface - the basics

The KEYMGMT libcrypto <-> provider interface currently makes a few
assumptions:

1.  provider side domain parameters and key data isn't mutable. In
    other words, as soon as a key has been created in any (loaded,
    imported data, ...), it's set in stone.
2.  provider side domain parameters can be strictly separated from the
    key data.

This does work for the most part, but there are places where that's a
bit too rigid for the functionality that the EVP_PKEY API delivers.
Key data needs to be mutable to allow the flexibility that functions
like EVP_PKEY_copy_parameters promise, as well as to provide the
combinations of data that an EVP_PKEY is generally assumed to be able
to hold:

- domain parameters only
- public key only
- public key + private key
- domain parameters + public key
- domain parameters + public key + private key

To remedy all this, we:

1.  let go of the distinction between domain parameters and key
    material proper in the libcrypto <-> provider interface.

    As a consequence, functions that still need it gain a selection
    argument, which is a set of bits that indicate what parts of the
    key object are to be considered in a specific call.  This allows
    a reduction of very similar functions into one.

2.  Rework the libcrypto <-> provider interface so provider side key
    objects are created and destructed with a separate function, and
    get their data filled and extracted in through import and export.

(future work will see other key object constructors and other
functions to fill them with data)

Fixes #10979

squash! Redesign the KEYMGMT libcrypto <-> provider interface - the basics

Remedy 1 needs a rewrite:

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/11006)

Reorganize the internal evp_keymgmt functions

Some of the evp_keymgmt_ functions are just wrappers around the
EVP_KEYMGMT function pointers.  We move those from keymgmt_lib.c to
keymgmt_meth.c.

Other evp_keymgmt_ functions are utility functions to help the rest of
the EVP functions.  Since their names are easily confused with the
functions that were moved to keymgmt_meth.c, we rename them so they
all start with evp_keymgmt_util_.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/11006)

Temporariy add ASN1_item_verify_ctx() missingcrypto.txt

PR #10942 introduced the new function ASN1_item_verify_ctx(), but did
not document it with the promise that documentation would follow soon.
We temporarily add this function to missingcrypto.txt until it has been
done.

Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/10980)

Don't ignore ASN1 when checking for undocumented symbols

When we run "make doc-nits" (which happens during travis runs) it will
complain if we add any new symbols that aren't documented. However it
was suppressing anything starting with ASN1. There's no reason why we
should allow ASN1 symbols to go undocumented any more than any others.
Therefore we remove that exception.

Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/10980)

Remove unused HAS_LFN_SUPPORT

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/10959)

Fix misspelling errors and typos reported by codespell

Fixes #10998

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/11000)

More accurate doc of -ssl_config option

Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/10972)

Update command list

Add -standard-commands option to list command (documented)
Update standard commands list in openssl.pod

Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/10972)

Add cmd-nits to travis build

Update CHANGES to have a complete and uniform description.

Fixes #9730

Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/10972)

The -hmac option to speed is now #ifdef'd

Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/10972)

Add missing s_client options

Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/10972)

Prevent compiler warning for unused static function.

Prepend missing ossl_unused in front of lh_type_new to make the compiler
happy.

CLA: trivial

Reviewed-by: Richard Levitte <levitte@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/10946)

Removed unused ssl_dane struct declaration.

The actually used structure is named ssl_dane_st.

CLA: trivial

Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/10945)

doc: Fix typo in EVP_DigestSignInit manpage

CLA: trivial

Signed-off-by: Jakub Jelen <jjelen@redhat.com>
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
(Merged from https://github.com/openssl/openssl/pull/10841)

Fix small misspelling in doc for OCSP_response_status

CLA: trivial

Reviewed-by: Paul Yang <kaishen.yy@antfin.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
(Merged from https://github.com/openssl/openssl/pull/10810)

Fix no-sm2

Fix ecdsatest to not run the SM2 test if SM2 has been disabled.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/11015)

Add FFC param/key generation

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/10909)

Fix no-tls1_3

The hostname_cb in sslapitest.c was originally only defined if TLSv1.3
was enabled. A recently added test now uses this unconditionally, so we
move the function implementation earlier in the file, and always compile
it in.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/11014)

Fix builds with no-dh

The various functions in bn_const.c return primes that are
specified for use in DH. However they were not being excluded from
a no-dh build - and was therefore causing the build to fail.

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/10990)

Add a test for SSL_CTX_new_with_libctx()

We test that SSL_CTX_new_with_libctx() can be used to control the libctx
that is in use for SSL operations.

Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/10854)

Use the OPENSSL_CTX and property query string in EVP_PKEY_CTX

When we use an EVP_PKEY_CTX in libssl we should be doing so with the
OPENSSL_CTX and property query string that were specified when the
SSL_CTX object was first created.

Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/10854)

Explicitly fetch ciphers and digests in libssl

We modify libssl to use explicitly fetched ciphers, digests and other
algorithms as required based on the configured library context and
property query string for the SSL_CTX that is being used.

Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/10854)

Params: change UTF8 construct calls to avoid explicit strlen(3) calls.

It is better, safer and smaller to let the library routine handle the
strlen(3) call.

Added a note to the documentation suggesting this.

Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/11019)

Stop accepting certificates signed using SHA1 at security level 1

Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
GH: #10786

Create a new embeddedSCTs1 that's signed using SHA256

Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
GH: #10786

Fix no-multiblock

Minor fixes to resolve compilation errors with the no-multiblock
Configure option.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/11004)

Author: Ross Kinsey <RossIKinsey@gmail.com>
Date:   Wed Jan 29 00:19:40 2020 -0500

Removed unnecessary switch statements from bio/bf_* callback_ctrl functions

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/10967)

x86: Add endbranch to indirect branch targets for Intel CET

To support Intel CET, all indirect branch targets must start with
endbranch.  Here is a patch to add endbranch to all function entries
in x86 assembly codes which are indirect branch targets as discovered
by running openssl testsuite on Intel CET machine and visual inspection.

Since x86 cbc.pl uses indirect branch with a jump table, we also need
to add endbranch to all jump targets.

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/10984)

Make minimum size for secure memory a size_t.

The minimum size argument to CRYPTO_secure_malloc_init() was an int but ought
to be a size_t since it is a size.

From an API perspective, this is a change.  However, the minimum size is
verified as being a positive power of two and it will typically be a small
constant.

Reviewed-by: David von Oheimb <david.von.oheimb@siemens.com>
(Merged from #11003)

EVP_MD_CTX_ctrl(): Remove unnecessary control

A check was present as to what operation is performed with this
context.  It may have been useful at some point, but isn't any more.

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/10947)

PROV: Fix the DSA SIGNATURE implementation for better digests handling

Refactor the DSA SIGNATURE digest setup to be uniform, and to happen
in two places:

1. when given through the digestsign and digestverify inits
2. when given through the set_ctx_params function.

When setting up the digest, we also check that the digest is one of
the officially accepted for DSA.

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/10947)

PROV: Implement padding mode words in the RSA ASYM_CIPHER implementation

Because the libcrypto code has relinquished control of exact words to
express padding mode choices, we re-implement them in the appropriate
provider implementation.

For the sake of legacy controls, we maintain support for the numeric
form of the padding mode, but leave that support otherwise undeclared.

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/10947)

Don't pass a digest-size to signature implementations

It turns out this was never necessary, as the implementation should
always check the default digest size anyway.

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/10947)

Decentralize legacy_ctrl_str_to_param()

This function did a bit too much in terms of central control, actually
more so than the legacy counterpart, where all the string processing
is done in the diverse *_pmeth.c.  Furthermore, there was no room
whatsoever for control keys that libcrypto isn't centrally aware of.

This function is changed to simply translating keys and values to
OSSL_PARAM form and then sent on their merry way to the provider
implementations through EVP_PKEY_CTX_set_params().  It translates
selected well known legacy names to their core name counterpart, and
that's as far as centralized control should extend.

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/10947)

Teach more BIOs how to handle BIO_CTRL_EOF

Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/10907)

Detect EOF while reading in libssl

If we hit an EOF while reading in libssl then we will report an error
back to the application (SSL_ERROR_SYSCALL) but errno will be 0. We add
an error to the stack (which means we instead return SSL_ERROR_SSL) and
therefore give a hint as to what went wrong.

Contains a partial fix for #10880

Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/10907)

Deprecate the ECDSA and EV_KEY_METHOD functions.

Use of the low level ECDSA and EC_KEY_METHOD functions has been informally discouraged for a
long time. We now formally deprecate them.

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/10960)

Deprecate the ECDH functions.

Use of the low level ECDH functions has been informally discouraged for a
long time. We now formally deprecate them.

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/10960)

ec.h: fix preprocessor indentation

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/10960)

Fix coverity issues CID 1457745...1457752, 1457853, 1457854

CID 1457854 - keymgmt_lib.c : OVERRUN
CID 1457853 - self_test_kats.c : UNINT
CID 1457752 - fipsprov.c RESOURCE_LEAK (code change in another PR removed this)
CID 1457751 - apps/pkcs12.c CHECKED_RETURN
CID 1457750 - dsa_ossl.c RESOURCE_LEAK (marked as false positive since tmp can not be NULL)
CID 1457749 - apps/nseq.c : CHECKED_RETURN
CID 1457748 - cipher_aes_cbc_hmac_sha.c : SIZEOF_MISMATCH
CID 1457747 - cipher_aes_cbc_hmac_sha.c : SIZEOF_MISMATCH
CID 1457746 - same as 1457752
CID 1457745 - apps/ocsp  : CHECKED_RETURN

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/10934)

Fix krb5 external test failure

The krb5 test requires the legacy module to be loaded in order to work.
It also seems to be senstive to using relative paths, so we use absolute
ones instead.

[extended tests]

Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/10992)

Missing "obj_mac" header file in "dh_lib"

Usage of `NID_undef` symbol without including its definition was causing
a build fail

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/10996)

Don't complain about documented symbols with find-doc-nits -d -o

find-doc-nits can give a list of symbols that were added since 1.1.1 and
are undocumented (using -o). To do this it uses the missingcrypto111.txt
and missingssl111.txt files which give a snapshot of the undocumented
symbols at the time of the 1.1.1 release. Currently it complains about
symbols that are in those files that have subsequently been documented.
This isn't particularly helpful so we suppress that feature when "-o"
is being used.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/10981)

Fix common test framework options

PR#6975 added the ability to our test framework to have common options to
all tests. For example providing the option "-test 5" to one of our test
programs will just run test number 5. This can be useful when debugging
tests.

Unforuntately this does not work well for a number of tests. In particular
those tests that call test_get_argument() without first skipping over these
common test options will not get the expected value. Some tests did this
correctly but a large number did not.

A helper function is introduced, test_skip_common_options(), to make this
easier for those tests which do not have their own specialised test option
handling, but yet still need to call test_get_argument(). This function
call is then added to all those tests that need it.

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/10975)

Fix no-ec

The cmp_protect_test cert chain tests use some EC certs which breaks in
a no-ec build. The fix is to just skip those tests if no-ec has been
configured.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/10991)

Revert "Legacy digests can have custom control values"

This reverts commit 1f457256ce6a1b2fd7e3f62eee8faa74cd5c835e.

This is causing Travis failures.

[extended tests]

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/10989)

CHANGES: Add note about the refactoring of SM2 EVP_PKEYs

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/10942)

Adapt some 'openssl' commands for SM2 changes.

There's no longer any need to make an EVP_PKEY type change for SM2
keys, so we trim away that code.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/10942)

Adapt tests for SM2 changes.

With test/ecdsatest.c, we test all the curves once for each EC key
type we have, i.e. one round trip with EVP_PKEY_EC and one with
EVP_PKEY_SM2.  This shows that we can use "normal" EC computations on
keys with the SM2 curve (which have the type EVP_PKEY_SM2 by default)
and SM2 computations with any other curve (which have the type
EVP_PKEY_EC by default)

test/evp_test.c, on the other hand, doesn't need to explicitly set the
EVP_PKEY_SM2 alias type, as that now happens automatically.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/10942)

Make SM3 a mandatory hash function for SM2.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/10942)

X509: Refactor X509_verify() and X509_REQ_verify() for better streamlining

The solution to incorporate the SM2 identity processing was an off
the side hack that more or less duplicated the ASN1_item_verify()
code with just a few lines being different.  We replace this with
a new function ASN1_item_verify_ctx(), which takes an EVP_MD_CTX
pointer instead of an EVP_PKEY pointer, just like its sibling
ASN1_item_sign_ctx().

This allows us to refactor X509_verify() and X509_REQ_verify() to
simply create a local EVP_MD_CTX and an attached EVP_PKEY_CTX,
which gets to hold the SM2 identity, if there is one, and then let
ASN1_item_verify_ctx() to its job.

This will also make it easier to adapt ASN1_item_verify_ctx() for
provider based keys.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/10942)