Restore the display of options with 'openssl version -a'

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5572)

Display the library building flags

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5560)

Configure et al: Move the definition of library only flags

We're currently using the attributes 'defines', 'cppflags', 'cflags'
etc quite liberally, with no regard for where that ends up.  Quite a
few of those flags are actually only relevant for the libraries
(mostly libcrypto), so it's safe to say that those could be applied to
the libraries only.

So, we move some of those flags to 'lib_defines', 'lib_cppflags',
'lib_cflags', etc.

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5560)

speed: add ecdhx448 to ecdh choices

CLA: trivial
Signed-off-by: Steven Noonan <steven@uplinklabs.net>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5551)

Fixed a spelling mistake in ASN1_TIME_set.pod

Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5569)

Give more information in the SSL_stateless return code

Allow users to distinguish between an error occurring and an HRR being
issued.

Fixes #5549

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5562)

Test the new PSK behaviour

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5554)

Update documentation for the new PSK behaviour

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5554)

Fallback on old style PSK callbacks if the new style ones aren't present

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5554)

Tolerate TLSv1.3 PSKs that are a different size to the hash size

We also default to SHA256 as per the spec if we do not have an explicit
digest defined.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5554)

Make OCSP "multi" compatible with "no-sock" builds.

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

Remove useless -D_ENDIAN from MPE/iX-gcc config

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5564)

Reduce travis-ci log output

Travis-ci log output is huge and overflows internal travis-ci view,
which makes it hard to find errors.
Redirect some output to a file and dump it only if it fails.
Remove "v" option from tar that builds and extracts the srcdist.
While running the tests manually, some non-POSIX (bashisms) with ==
vs = came to light.

Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5555)

Configure: correct the check of env vars vs command line flags

The check to detect if env / make variables conflicted with compiler /
linker flags on the configure command line went a little too far, and
would stop the configuration process if any such command line flag was
given, with no regard for the env / make variables at all.

This change refines the check so the stop only gets triggered if any
such flags were given AND any of the corresponding variables are set.

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5561)

Make a few more asm modules conform: last argument is output file

Fixes #5310

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5315)

Configure: disallow the mixture of compiling flags and env / make variables

Note that this might give surprising results if someone forgets an environment
variable that has been set previously.

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5534)

Configurations/README: update documentation on flags

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5534)

Configure et al: get rid of last traces of plib_lflags / PLIB_LDFLAGS

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5534)

config: Pass diverse flags to Configure via the environment

Passing flags "discovered" by 'config' on the command line to
'Configure' mixes them up with flags given by the user.  That is
contrary to their intention, so they need to be passed in a different
manner.

Enter the environment variables __CNF_CPPDEFINES, __CNF_CPPINCLUDES,
__CNF_CPPFLAGS, __CNF_CFLAGS, __CNF_CXXFLAGS, __CNF_LDFLAGS, and
__CNF_LDLIBS, initialised by 'config', and then used by Configure to
initialise these %config values: defines, includes, cppflags, cflags,
cxxflags, lflags, and ex_libs.

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5534)

Configurations/10-main.conf: Duplicate cflags to cxxflags

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5534)

Add space to asoutflag value where appropriate

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5534)

Make "make variables" config attributes for overridable flags

With the support of "make variables" comes the possibility for the
user to override them.  However, we need to make a difference between
defaults that we use (and that should be overridable by the user) and
flags that are crucial for building OpenSSL (should not be
overridable).

Typically, overridable flags are those setting optimization levels,
warnings levels, that kind of thing, while non-overridable flags are,
for example, macros that indicate aspects of how the config target
should be treated, such as L_ENDIAN and B_ENDIAN.

We do that differentiation by allowing upper case attributes in the
config targets, named exactly like the "make variables" we support,
and reserving the lower case attributes for non-overridable project
flags.

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5534)

Configure et al: rename all dso_* to module_* in shared-info.pl

Because there are already attributes with the dso_ prefix that are
used instead of the corresponding lib_ attributes rather than in
addition to them, it gets confusing to have similar or exactly the
same attributes working with different semantics on Unix.

So we rename those by changing the prefix dso_ to module_, and having
those work just like the shared_ attributes, but for DSOs.

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5534)

Duplicate entries ssl_handshake_tbl trace entries...

... and add some missing known values.
Sort ssl/tls extension array list

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5304)

Fix issues in ia32 RDRAND asm leading to reduced entropy

This patch fixes two issues in the ia32 RDRAND assembly code that result in a
(possibly significant) loss of entropy.

The first, less significant, issue is that, by returning success as 0 from
OPENSSL_ia32_rdrand() and OPENSSL_ia32_rdseed(), a subtle bias was introduced.
Specifically, because the assembly routine copied the remaining number of
retries over the result when RDRAND/RDSEED returned 'successful but zero', a
bias towards values 1-8 (primarily 8) was introduced.

The second, more worrying issue was that, due to a mixup in registers, when a
buffer that was not size 0 or 1 mod 8 was passed to OPENSSL_ia32_rdrand_bytes
or OPENSSL_ia32_rdseed_bytes, the last (n mod 8) bytes were all the same value.
This issue impacts only the 64-bit variant of the assembly.

This change fixes both issues by first eliminating the only use of
OPENSSL_ia32_rdrand, replacing it with OPENSSL_ia32_rdrand_bytes, and fixes the
register mixup in OPENSSL_ia32_rdrand_bytes. It also adds a sanity test for
OPENSSL_ia32_rdrand_bytes and OPENSSL_ia32_rdseed_bytes to help catch problems
of this nature in the future.

Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5342)

Fixed several readability issues in DH_generate_parameters.pod

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/5553)

Corrected two typos in a man page

Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5542)

Add GOST OIDs for Edwards parameter sets

Add OIDs for parameter sets of Edwards elliptic curves.

CLA: trivial

Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5380)

Remove OPENSSL_USE_NODELETE guards in shlibloadtest

PR #3399 converted shlibloadtest to the new test framework. It also
seemed to add some `OPENSSL_USE_NODELETE` guards to the library
unloading part of the test. This part was added in a commit with this
description:

    Review feedback; use single main, #ifdef ADD_TEST

    Suppose OPENSSL_USE_NODELETE (via Nick Reilly)

Strangely though there doesn't seem to be any relevant review feedback
in that PR that could justify the addition of those guards. The guards
do not appear in 1.1.0.

Having the guards changes the nature of the test, so that we only test
library unloading on platforms where OPENSSL_USE_NODELETE is set (Linux
and Windows). I can't think of any good reason for this and as it doesn't
seem to be necessary in 1.1.0 so I think we should remove them.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5530)

Implement multi-process OCSP responder.

With "-multi" the OCSP responder forks multiple child processes,
and respawns them as needed.  This can be used as a long-running
service, not just a demo program.  Therefore the index file is
automatically re-read when changed.  The responder also now optionally
times out client requests.

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

Prepare to detect index changes in OCSP responder.

Retain open file handle and previous stat data for the CA index
file, enabling detection and index reload (upcoming commit).

Check requirements before entering accept loop.

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

Configurations/unix-Makefile.tmpl: remove assignment of AS and ASFLAGS

We have never used these variables with the Unix Makefile, and there's
no reason for us to change this, so to avoid confusion, we remove them.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5545)

BIO_s_mem.pod: fix indirection for out parameter **pp

BIO_get_mem_data() and BIO_get_mem_ptr() assign to *pp, not pp

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5544)

crypto/armcap.c: mask SHA512 hardware detection on iOS.

When running iOS application from command line it's impossible to
get past the failing capability detection. This is because it's
executed under debugger and iOS debugger is impossible to deal with.
[If Apple implements SHA512 in silicon, it would have to be detected
with sysctlbyname.]

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

Configurations/10-main.conf: add -fno-common back to darwin-ppc-cc.

-fno-common was removed for all Darwin targets in
0c8734198d4282f6997965a03cd2e0ceaf207549 with rationale "it's either
'ranlib -c' or '-fno-common'." However, it's still absolutely required
in 32-bit darwin-ppc-cc. And when trying things out I didn't quite
see why it was formulated as one-or-another choice, as 'ranlib -c'
shouldn't [and doesn't] have problems with object modules without
commons. [Well, to be frank, I didn't manage to reproduce the problem
the modification was meaning to resolve either...]

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

util/dofile.pl: only quote stuff that actually needs quoting

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5533)

opensslconf.h.in: Use all the "openssl_api_defines"

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5533)

Configure: fix small spelling error

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5533)

Avoid unconditional store in CRYPTO_malloc.

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5372)

Fix a typo in the s_client man page

Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5525)

Fix a bunch of gcc warnings in packettest.c

Had been observed with recent gcc-8 snapshot and
$ ./config --strict-warnings enable-asan

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5519)

Check the parent DRBG's strength

We currently don't support the algorithm from NIST SP 800-90C
10.1.2 to use a weaker DRBG as source

Reviewed-by: Dr. Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
GH: #5506

bnrand_range: Always call bnrand() with the correct flag

It was calling the BN_rand() when it should have call BN_priv_rand()

Reviewed-by: Tim Hudson <tjh@openssl.org>
GH: #5514

Add support for .include directive in config files

Either files or directories of *.cnf or *.conf files
can be included.
Recursive inclusion of directories is not supported.

Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5351)

If not sending key_share (no TLSv1.3), return appropriately.

This fixes an issue raised in PR #4964 by kaduk.

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5491)

Fix status_request and SCT extensions

They are valid for use in a CertificateRequest message, but we did not
allow it. If a server sent such a message using either of those two
extensions then the handshake would abort.

This corrects that error, but does not add support for actually processing
the extensions. They are simply ignored, and a TODO is inserted to add
support at a later time.

This was found during interoperability testing with btls:
https://gitlab.com/ilari_l/btls

Prompted by these errors I reviewed the complete list of extensions and
compared them with the latest table in draft-24 to confirm there were no
other errors of a similar type. I did not find any.

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/5490)

Update CHANGES for X448/Ed448 support in libssl

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Kurt Roeckx <kurt@roeckx.be>
(Merged from https://github.com/openssl/openssl/pull/5470)

Update tests for TLS Ed448

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Kurt Roeckx <kurt@roeckx.be>
(Merged from https://github.com/openssl/openssl/pull/5470)

Add X448/Ed448 support to libssl

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Kurt Roeckx <kurt@roeckx.be>
(Merged from https://github.com/openssl/openssl/pull/5470)

Fixed a typo in a man page

Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5508)

Fix credit for SRP code

Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5504)

Windows makefile: Don't quote generator arguments

Rely on the build.info constructor to do the right thing.

Fixes #5500

Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5501)

test/ctype_internal_test.c: portability fixup.

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/5493)

mem_sec.c: portability fixup.

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/5493)

ec/curve448/f_generic.c: fix VC-WIN32 debug build failure.

Debugging asserts had implicit casts that triggered the warnings.
However, instead of making the casts explicit it's more appropriate
to perform checks that ensure that implicit casts were safe.

ec/curve448/scalar.c: size_t-fy scalar_decode_short.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5494)

ec/curve448/curve448.c: fix undefined behaviour sanitizer failure.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5494)

Fix a possible memory leak in engine_table_register

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5489)

Fixed a typo in a man page

Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5498)

Update CHANGES for X448 and Ed448

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Kurt Roeckx <kurt@roeckx.be>
(Merged from https://github.com/openssl/openssl/pull/5481)

Add test vectors for X448 and Ed448

This adds the Ed448 test vectors from RFC8032 and the X448 test vectors
from RFC7748.

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Kurt Roeckx <kurt@roeckx.be>
(Merged from https://github.com/openssl/openssl/pull/5481)

Update some documentation for X448/Ed448

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Kurt Roeckx <kurt@roeckx.be>
(Merged from https://github.com/openssl/openssl/pull/5481)

Integrate X448 and Ed448 into libcrypto

This adds all of the relevant EVP plumbing required to make
X448 and Ed448 work.

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Kurt Roeckx <kurt@roeckx.be>
(Merged from https://github.com/openssl/openssl/pull/5481)

Add pkey types for curve448

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Kurt Roeckx <kurt@roeckx.be>
(Merged from https://github.com/openssl/openssl/pull/5481)

Fix typo in ASN1_STRING_length doc

CLA: trivial

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
(Merged from https://github.com/openssl/openssl/pull/5458)

Test the result of CMS_RecipientInfo_ktri_get0_algs() before using its output in rsa_cms_encrypt().

CLA: trivial

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/4966)

ec/asm/x25519-x86_64.pl: remove redundant carry chain.

Why is it redundant? We're looking at carry from addition of small,
11-bit number to 256-bit one. And carry would mean only one thing,
resulting first limb being small number and remaing ones - zeros.
Hence adding 38 to first limb can't carry.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5476)

Do not set a nonzero default max_early_data

When early data support was first added, this seemed like a good
idea, as it would allow applications to just add SSL_read_early_data()
calls as needed and have things "Just Work".  However, for applications
that do not use TLS 1.3 early data, there is a negative side effect.
Having a nonzero max_early_data in a SSL_CTX (and thus, SSL objects
derived from it) means that when generating a session ticket,
tls_construct_stoc_early_data() will indicate to the client that
the server supports early data.  This is true, in that the implementation
of TLS 1.3 (i.e., OpenSSL) does support early data, but does not
necessarily indicate that the server application supports early data,
when the default value is nonzero.  In this case a well-intentioned
client would send early data along with its resumption attempt, which
would then be ignored by the server application, a waste of network
bandwidth.

Since, in order to successfully use TLS 1.3 early data, the application
must introduce calls to SSL_read_early_data(), it is not much additional
burden to require that the application also calls
SSL_{CTX_,}set_max_early_data() in order to enable the feature; doing
so closes this scenario where early data packets would be sent on
the wire but ignored.

Update SSL_read_early_data.pod accordingly, and make s_server and
our test programs into applications that are compliant with the new
requirements on applications that use early data.

Fixes #4725

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5483)

Tell the ciphers which DRBG to use for generating random bytes.

Reviewed-by: Richard Levitte <levitte@openssl.org>
GH: #4672

Fixed typo in description of EVP_CIPHER_meth_set_iv_length().

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
(Merged from https://github.com/openssl/openssl/pull/5477)

Adapt 15-test_out_option.t for more than just Unix

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4008)

Enable the -out option test on VMS as well

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4008)

Check on VMS as well

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4008)

Add VMS version of app_dirname()

Related to #3709

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/4008)

Fix the type of -out option

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3709)

Add test cases for this -out option check

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3709)

Check directory is able to create files for various -out option

This is to address issue #3404, only works in Unix-like platforms

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/3709)

storeutl: make sure s2i_ASN1_INTEGER is correctly declared

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/5482)

Always use adr with __thumb2__.

Thumb2 addresses are a bit a mess, depending on whether a label is
interpreted as a function pointer value (for use with BX and BLX) or as
a program counter value (for use with PC-relative addressing). Clang's
integrated assembler mis-assembles this code. See
https://crbug.com/124610#c54 for details.

Instead, use the ADR pseudo-instruction which has clear semantics and
should be supported by every assembler that handles the OpenSSL Thumb2
code. (In other files, the ADR vs SUB conditionals are based on
__thumb2__ already. For some reason, this one is based on __APPLE__, I'm
guessing to deal with an older version of clang assembler.)

It's unclear to me which of clang or binutils is "correct" or if this is
even a well-defined notion beyond "whatever binutils does". But I will
note that https://github.com/openssl/openssl/pull/4669 suggests binutils
has also changed behavior around this before.

Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5431)

Fix documentation link to reference man3

CLA: trivial

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/5473)

Make some perl scripts output to stdout

And only generate one output "file" at a time for objects.pl

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5432)

crypto/ec/asm/x25519-x86_64.pl: close STDOUT, not $STDOUT

Fixes #5471

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5472)

configdata.pm: ensure $prefix is defined

Fixes #5471

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5472)

bio_b64.c: prevent base64 filter BIO from decoding out-of-bound data

Fixes #5405, #1381

The base64 filter BIO reads its input in chunks of B64_BLOCK_SIZE bytes.
When processing input in PEM format it can happen in rare cases that

- the trailing PEM marker crosses the boundary of a chunk, and
- the beginning of the following chunk contains valid base64 encoded data.

This happened in issue #5405, where the PEM marker was split into
"-----END CER" and "TIFICATE-----" at the end of the first chunk.

The decoding of the first chunk terminated correctly at the '-' character,
which is treated as an EOF marker, and b64_read() returned. However,
when called the second time, b64_read() read the next chunk and interpreted
the string "TIFICATE" as valid base64 encoded data, adding 6 extra bytes
'4c 81 48 08 04 c4'.

This patch restores the assignment of the error code to 'ctx->cont', which
was deleted accidentally in commit 5562cfaca4f3 and which prevents b64_read()
from reading additional data on subsequent calls.

This issue was observed and reported by Annie Yousar.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5422)

Prepare for 1.1.1-pre3-dev

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

Prepare for 1.1.1-pre2 release

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

Update copyright year

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

Clear some sslapitest global variables after use

Otherwise we get a use after free if the test order is randomised.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5467)

mem_sec.c: relax POSIX requirement.

Even though mlock(2) was standardized in POSIX.1-2001, vendors did
implement it prior that point.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5460)

test/ct_test.c: remove dependency on -lm.

fabs(3m) is customarily inlined, but it's not, one has to link with -lm.
Since fabs(3m) is the only reference, it makes more sense to avoid it.

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

Export keying material using early exporter master secret

This commit adds SSL_export_keying_material_early() which exports
keying material using early exporter master secret.

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5252)

Make it possible to give --libdir an absolute path

With this, we introduce the make variable 'libdir', which differs from
'LIBDIR' not only in casing, but also by being the absolute path to
the library installation directory.  This variable is intentionally
compatible with the GNU coding standards.

When --libdir is given an absolute path, it is considered as a value
according to GNU coding standards, and the variables LIBDIR and libdir
will be this:

    LIBDIR=
    libdir=/absolute/path

When --libdir is given a relative path (just the name of the desired
library directory), or not given at all, it is considered as a
"traditional" OpenSSL value, and the variables LIBDIR and libdir will
be this:

    LIBDIR=relativepath
    libdir=$(INSTALLTOP)/$(LIBDIR)

Fixes #5398

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5446)

appveyor.yml: omit makedepend step.

makedepend makes lesser sense in a throw-away build like CI, but
it spares some computational time, because with MSVC it takes
separate per-file compiler invocation.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5452)

ec/ecp_nistp{224,256,521}.c: harmonize usage of __uint128_t.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5449)

{ec/curve25519,poly1305/poly1305}.c: relax pedantic constraint.

As it turns out gcc -pedantic doesn't seem to consider __uint128_t
as non-standard, unlike __int128 that is.

Fix even MSVC warnings in curve25519.c.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5449)

ec/curve448: portability fixups.

SPARC condition in __SIZEOF_INT128__==16 is rather performance thing
than portability. Even though compiler advertises int128 capability,
corresponding operations are inefficient, because they are not
directly backed by instruction set.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5449)

ec/asm/x25519-x86_64.pl: fix up ADCX/ADOX fallback.

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5447)

Remove OSSLzu, and fix the one place that used it.

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5439)

Fix some bugs with the cfb1 bitsize handling

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5426)

Remove executable bit from test/recipes/03-test_internal_sm4.t

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5434)

Add a note in CHANGES

[extended tests]

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2688)