Fix a race condition in supported groups handling

In TLSv1.3 the supported groups can be negotiated each time a handshake
occurs, regardless of whether we are resuming or not. We should not store
the supported groups information in the session because session objects
can be shared between multiple threads and we can end up with race
conditions. For most users this won't be seen because, by default, we
use stateless tickets in TLSv1.3 which don't get shared. However if you
use SSL_OP_NO_TICKET (to get stateful tickets in TLSv1.3) then this can
happen.

The answer is to move the supported the supported group information into
the SSL object instead.

Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/9162)

Fix error handling at openssl_strerror_r

When bufsize == 0, openssl_strerror_r should return 0 (if _GNU_SOURCE is defined),
to be consistent with non-_GNU_SOURCE variants, which exhibit the same behavior.
Fix a few cases, where the return value of openssl_strerror_r was ignored.

Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/9163)

 Swap #if blocks in uid.c so target platform gets checked before host

This avoids the case where a UEFI build on FreeBSD tries to call the system
issetugid function instead of returning 0 as it should do.

CLA: trivial

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from #9158)

Standardise the function naming conventions in initthread.c

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9040)

Document OPENSSL_thread_stop_ex()

This new function works in the same way as OPENSSL_thread_stop() but
for a specified OPENSSL_CTX.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9040)

Add the function OPENSSL_thread_stop_ex()

This adds the ability to clean up a thread on a per OPENSSL_CTX basis.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9040)

Tell the FIPS provider about thread stop events

The RAND code needs to know about threads stopping in order to cleanup
local thread data. Therefore we add a callback for libcrypto to tell
providers about such events.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9040)

Provide a version of ossl_init_thread_start that works in FIPS mode

This will need to be hooked up in a later commit with an event sent to
the FIPS provider informing it of thread stop events.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9040)

Split thread intialisation and handling out of init.c

We're going to need some of these functions in the FIPS module, but most
of the rest of the code in init.c is not needed. Therefore we split it out.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9040)

Convert thread stop handling into a publish/subscribe model

In later commits this will allow providers to subscribe to thread stop
events. We will need this in the FIPS module. We also make thread stop
handling OPENSSL_CTX aware (different OPENSSL_CTXs may have different
thread local data that needs cleaning up).

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9040)

Add the function openssl_ctx_get_concrete()

This adds the ability to take an OPENSSL_CTX parameter and either return it
as is (unchanged), or if it is NULL return a pointer to the default ctx.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9040)

Configure: final cleanup of asm related things

Remove the *_asm templates in Configurations/00-base-templates.conf,
all attempts to inherit them, and the asm() perl function.

[extended tests]

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9166)

Move padlock_asm_src file information to build.info files

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9166)

Move keccak1600_asm_src file information to build.info files

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9166)

Move poly1305_asm_src file information to build.info files

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9166)

Move chacha_asm_src file information to build.info files

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9166)

Move modes_asm_src file information to build.info files

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9166)

Move cmll_asm_src file information to build.info files

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9166)

Move wp_asm_src file information to build.info files

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9166)

Move rc5_asm_src file information to build.info files

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9166)

Move rmd160_asm_src file information to build.info files

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9166)

Move rc4_asm_src file information to build.info files

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9166)

Move cast_asm_src file information to build.info files

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9166)

Move sha1_asm_src file information to build.info files

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9166)

Move md5_asm_src file information to build.info files

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9166)

Move bf_asm_src file information to build.info files

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9166)

Move aes_asm_src file information to build.info files

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9166)

Move des_asm_src file information to build.info files

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9166)

Move ec_asm_src file information to build.info files

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9166)

Move cpuid_asm_src file information to build.info files

Also took away the internal 'debug-linux-ia32-aes' config target, as
it's broken (refers to files that no longer exist).

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9166)

Move bn_asm_src file information to build.info files

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9166)

Configure: add mechanism to specify asm target architecture

As preparation for moving asm file specs to build.info files, we must
make sure there is still some base information to help select the
correct files.

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9166)

Move uplink file information to build.info files

This file information was hidden in config target files, when they
should really be part of build.info like any other file we build
from.  With build.info variables, the task became much easier.

We take the opportunity to move apps_init_src and apps_aux_src to
apps/build.info as well, and to clean up apps/build.info.

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9166)

Configure: add mechanism to specify uplink target architecture

As preparation for moving uplink file specs to build.info files, we
must make sure there is still some base information to help select the
correct files.

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9166)

Configure: allow conditions and variable values to have variable references

This will allow building variables on other variables, and to have
conditions based on variable contents.

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9166)

Configure: AIX needs a specific DSO extension function

DSO extensions are normally derived from platform->shlibextsimple() on
Unix.  This isn't the case for AIX, so it needs to define its own DSO
extension specifically.

Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9005)

Fix no-ec with no-dh

Make sure that the combination of no-ec with no-dh builds successfully.
If neither ec or dh are available then TLSv1.3 is not possible.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9156)

Allow TLSv1.3 in a no-ec build

Now that we have TLSv1.3 FFDHE support there is no reason why we should
not allow TLSv1.3 to be used in a no-ec build. This commit enables that
to happen.

It also fixes no-ec which was previously broken.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9156)

Replumbing: Adapt the default and legacy providers to use library context upcall

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9160)

Replumbing: Adapt the FIPS module to use the library context upcall

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9160)

Replumbing: offer a core upcall to get the provider object's library context

The FIPS module currently has "magic" support to have the library
context become the provider context within the core code, for the FIPS
module's inner provider.

We replace that with a core upcall that returns the library context
associated with a provider object.  That way, the FIPS module can
handle the assignment of the inner provider context itself.  This
allows the FIPS module (and any other provider module that wishes to
use a similar mechanism) to define for itself what the provider
context is.  It's currently simply a pointer to a library context,
but may contain other stuff as well in the future.

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9160)

crypto/ppccap.c: Fix FIPS build on PPC

Some code was temporarly disabled in the FIPS module because SHA other
SHA1 hadn't been ported.  Now that they have, we must enable this code
again.

Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/9168)

Better document specific OSSL_PARAM symbols.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9114)

Add documentation for X509_cmp and related APIs

Fixes: #9088
Functions documented in this commit: X509_cmp, X509_NAME_cmp,
X509_issuer_and_serial_cmp, X509_issuer_name_cmp, X509_subject_name_cmp,
X509_CRL_cmp, X509_CRL_match

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9091)

Use variables in build.info files where it's worth the while

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/9144)

Configure: Add support for variables in build.info files

Variables have the syntax defined with this regular expression:

    \$([[:alpha:]_][[:alnum:]_]*)

They are always local to the build.info they are defined in, and are
defined like this:

    $VAR=text

Expansion is done very simply, any reference to the variable (with the
exact same variable syntax) is replaced with its defined value.

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/9144)

Fix building with enable-trace

Tracing doesn't work in the FIPS module. Ensure we switch it off there.

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/9159)

Fix no-dh

The recent TLSv1.3 FFDHE support missed a few OPENSSL_NO_DH guards.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9153)

Update the d2i docs to reflect reality

The d2i docs state that if an error occurs then |*a| is not freed. This
is not correct. On error it is freed and set to NULL. We update the docs
to say this, and also discuss the fact that this behaviour was inconsistent
prior to OpenSSL 1.1.0.

Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/9146)

ts: Fix awkward sentences in the documentation and the default digest

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9143)

ts: Use sha256 as default digest for TS query

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/7900)

Fix typo mistake on calls to SSL_ctrl in ssl.h

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/9140)

Update docs for TLS1.3 FFDHE

Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8178)

Add testcase for TLS1.3 FFDHE

Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8178)

TLS1.3 FFDHE Support

Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8178)

Enable find-doc-nits to find undocumented symbols since 1.1.1

A previous commit added the ability to find newly undocumented symbols.
We extend this capability to check anything that was newly added since
1.1.1 which is undocumented. A new option -o is added to find-doc-nits
to amend the behaviour of -v or -e to check symbols that were newly
added since the release of 1.1.1.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9094)

Fix an incorrect macro

A macro was missing a space which was confusing find-doc-nits

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9094)

i2d_PublicKey was listed in 2 different man pages

find-doc-nits complains if a symbol is documented in more than one
location.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9094)

Make find-doc-nits check for newly added undocumented symbols

We create lists of undocumented functions and macros as they are now so
that find-doc-nits can check for newly introduced functions/macros that
are undocumented.

This works in a similar way to the -u and -d options to find-doc-nits.
These count undocumented symbols and print a detailed list of undocumented
symbols repsectively. This commit adds the -v and -e options to restrict
the count/detailed list to newly added undocumented symbols only.

There is also a new -s option that does the same as -e except that it
produces no output if there are no newly undocumented symbols.

We also amend "make doc-nits" to add the -s option which should cause
travis to fail if a PR adds undocumented symbols.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9094)

Configure: count basenames for all library sources

Make sure that each basename only appears once.  This is due to the
static library archiver on Unix, that indexes archived object files by
base name only, thereby making base name clashes...  interesting.

This is a safety net for OpenSSL developer!

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/9133)

Ensure code is compiled with correct BIGNUM assembler defines

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/9130)

Add some dummy BIGNUM calls from inside 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/9130)

Make BIGNUM code available from within 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/9130)

Temporarily disable RAND call in FIPS_MODE until RAND is available

Other commits will enable the RAND code in FIPS_MODE. Until those commits
are in place we temporarily disable making RAND calls while in FIPS_MODE.

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/9130)

Convert bn_rand.c to use EVP

Replace the low level SHA512_* function calls with EVP calls.

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/9130)

Create BN_CTX_new_ex() and BN_CTX_secure_new_ex()

These variants of BN_CTX_new() and BN_CTX_secure_new() enable passing
an OPENSSL_CTX so that we can access this where needed throughout the
BIGNUM sub library.

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/9130)

Remove redundant include

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/9137)

OSSL_PARAM example code bug fix.

Technically not a bug since the code worked but the array index shouldn't have
been constant after searching for the field.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9136)

Move a fall through comment

When compiling with --strict-warnings using gcc 7.4.0 the compiler
complains that a case falls through, even though there is an explicit
comment stating this. Moving the comment outside of the conditional
compilation section resolves this.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9131)

Fix a URL to the NMBRTHRY list archive

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/9123)

Fix app opt compile failure due to missing <inttypes.h>

opt.c uses functions that are only available if inttypes.h exists.
It now checks a define which is unavailable if
inttypes.h is included. The include is done automagically inside e_os2.h.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8986)

Fix Windows Compile failure due to missing <inttypes.h>

Including <inttypes.h> caused a windows build failure.
The test is now skipped if strtoimax & strtoumax are not supported.
It does this by checking for a define which is only available if
inttypes.h is not included. The include is done automagically inside e_os2.h.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8986)

Add defines to indicate if intypes.h and stdint.h are unavailable

Use the defines OPENSSL_NO_INTTYPES_H & OPENSSL_NO_STDINT_H to determine
if the headers are unavailable for a platform.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8986)

PBKDF2 updates to conform to SP800-132

The existing code used PKCS5 specifications.
SP800-132 adds the following additional constraints for:
  - the range of the key length.
  - the minimum iteration count (1000 recommended).
  - salt length (at least 128 bits).
These additional constraints may cause errors (in scrypt, and
some PKCS5 related test vectors). To disable the new
constraints use the new ctrl string "pkcs5".
For backwards compatability, the checks are only enabled by
default for fips mode.

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8868)

Make EVP_MD_CTX_ctrl() work for legacy use cases (ssl3).

This is still required currently by engines and digestsign/digestverify.
This PR contains merged in code from Richard Levitte's PR #9126.

[extended tests]

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9103)

Fix Keccak structure name reference in S390 legacy code

Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9120)

Fix doc to remove const arg in GEN_SESSION_CB

ae3947de095 changed the callback arg not to have a const parameter.

CLA: trivial

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9104)

Change cipher default strings to a function

Making the default cipher strings a function gives the library more
control over the defaults. Potentially allowing a change in the
future as ciphers become deprecated or dangerous.
Also allows third party distributors to change the defaults for their
installations.

Reviewed-by: Paul Yang <yang.yang@baishancloud.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8686)

BIO_lookup_ex: Retry with AI_ADDRCONFIG cleared if getaddrinfo fails

The lookup for ::1 with getaddrinfo() might return error even if
the ::1 would work if AI_ADDRCONFIG flag is used.

Fixes: #9053
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9108)

Improve the Windows OneCore target support. (Add targets for building libraries for Windows Store apps.)

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8917)

doc/internal/man3/ossl_method_construct.pod: follow common conventions

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/9109)

EVP fetching: make operation_id part of the method identity

Because the operation identity wasn't integrated with the created
methods, the following code would give unexpected results:

    EVP_MD *md = EVP_MD_fetch(NULL, "MD5", NULL);
    EVP_CIPHER *cipher = EVP_CIPHER_fetch(NULL, "MD5", NULL);

    if (md != NULL)
        printf("MD5 is a digest\n");
    if (cipher != NULL)
        printf("MD5 is a cipher\n");

The message is that MD5 is both a digest and a cipher.

Partially fixes #9106

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/9109)

Revert the DEVRANDOM_WAIT feature

The DEVRANDOM_WAIT feature added a select() call to wait for the
`/dev/random` device to become readable before reading from the
`/dev/urandom` device. It was introduced in commit 38023b87f037
in order to mitigate the fact that the `/dev/urandom` device
does not block until the initial seeding of the kernel CSPRNG
has completed, contrary to the behaviour of the `getrandom()`
system call.

It turned out that this change had negative side effects on
performance which were not acceptable. After some discussion it
was decided to revert this feature and leave it up to the OS
resp. the platform maintainer to ensure a proper initialization
during early boot time.

Fixes #9078

This partially reverts commit 38023b87f037.

Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9084)

Add documentation for new functions taking an OPENSSL_CTX parameter

Various functions have been added that take an OPENSSL_CTX parameter as
a result of moving the RAND code into the FIPS module. We document all of
those functions.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9039)

Move the rand_nonce_lock code into drbg_lib.c

It was previously rand_lib but it makes more sense in drbg_lib.c since
all the functions that use this lock are only ever called from drbg_lib.c

We add some FIPS_MODE defines in preparation for later moving this code
into the FIPS module.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9039)

Make the rand_crng code OPENSSL_CTX aware

This is in preparation for moving this code inside the FIPS module.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9039)

Convert drbg_lib to use OPENSSL_CTX for its global data

In preparation for moving the RAND code into the FIPS module we make
drbg_lib.c OPENSSL_CTX aware.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9039)

Replace BIO_printf with ASN1_STRING_print in GENERAL_NAME_print

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/9101)

Fixed typo in code comment.
[skip ci]

Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9102)

Change EVP_MAC method from copy to dup

Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
GH: #7651

Replace EVP_MAC_CTX_copy() by EVP_MAC_CTX_dup()

Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
GH: #7651

Fix typo in macro argument of SSL_set1_client_sigalgs_list()

Fixes #9092

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/9093)

Address property documentation concerns.

Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
(Merged from https://github.com/openssl/openssl/pull/9090)

rename the digest provider files to avoid any name clashes with other folders

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9083)

Coverity fixes

covID 1445689 Resource leak (in error path)
covID 1445318 Resource leak (in test - minor)
covID 1443705 Unchecked return value (Needed if CRYPTO_atomic_add() was used)
covID 1443691 Resource leak (in app - minor)

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9071)

man: fix doc-nit in property.pod

This causes travis build failures on master

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9087)

Document property hierarchy.

Add documentation to indicate the interaction between global (context level)
property queries and local (passed to fetch) ones.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9042)

EVP_DigestInit_ex(): drop previous context engine earlier

If a EVP_MD_CTX holds a reference to a previously given engine, and
the type of its digest isn't the same as the one given in the new
call, drop that engine reference, allowing providers or other engines
to provide the new algorithm on an equal basis.

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9077)

Document deprecation of version-specific SSL/TLS methods

In commit 2b8fa1d56cd3 the version-specific SSL/TLS methods were
deprecated. This patch improves the documentation of that change
by stating the deprecation more prominently in the manual page
and explaining the reason for the deprecation.

Fixes #8989

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/9001)

build_SYS_str_reasons: Fix a crash caused by overlong locales

The 4 kB SPACE_SYS_STR_REASONS in crypto/err/err.c isn't enough for some locales.
The Russian locales consume 6856 bytes, Ukrainian even 7000.

build_SYS_str_reasons() contains an overflow check:

if (cnt > sizeof(strerror_pool))
    cnt = sizeof(strerror_pool);

But since commit 9f15e5b911ba6053e09578f190354568e01c07d7 it no longer
works as cnt is incremented once more after the condition.

cnt greater than sizeof(strerror_pool) results in an unbounded
OPENSSL_strlcpy() in openssl_strerror_r(), eventually causing a crash.

When the first received error string was empty or contained only
spaces, cur would move in front of the start of the strerror_pool.

Also don't call openssl_strerror_r when the pool is full.

Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/8966)