Fix a few if(, for(, while( inside code.

Fix some indentation at the same time

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1292)

Remove reduntant X509_STORE_CTX_set_verify_cb declaration

f0e0fd51fd8307f6eae64862ad9aaea113f1177a added X509_STORE_CTX_set_verify_cb
with a typedef'd argument, making the original one redundant.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>

Simplify buffer limit checking, and reuse BIO_snprintf returned value.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1284)

Code factorisation and simplification

Fix some code indentation

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1284)

Fix double calls to strlen

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1284)

Simplify code related to tmp_email_dn.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1284)

Use more X509_REQ_get0_pubkey & X509_get0_pubkey

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1284)

OCSP_request_add0_id() inconsistent error return

There are two failure cases for OCSP_request_add_id():
1. OCSP_ONEREQ_new() failure, where |cid| is not freed
2. sk_OCSP_ONEREQ_push() failure, where |cid| is freed

This changes makes the error behavior consistent, such that |cid| is
not freed when sk_OCSP_ONEREQ_push() fails. OpenSSL only takes
ownership of |cid| when the function succeeds.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1289)

Sanity check in ssl_get_algorithm2().

RT#4600

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

Send alert on CKE error.

RT#4610

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

Resolve over command syntax error which causes 'make install' to fail

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1312)

Document the slight change in CRYPTO_mem_ctrl()

Reviewed-by: Kurt Roeckx <kurt@openssl.org>

Document the slight change in ERR_get_next_error_library()

Reviewed-by: Kurt Roeckx <kurt@openssl.org>

make update

Reviewed-by: Kurt Roeckx <kurt@openssl.org>

Change all our uses of CRYPTO_THREAD_run_once to use RUN_ONCE instead

That way, we have a way to check if the init function was successful
or not.

Reviewed-by: Kurt Roeckx <kurt@openssl.org>

Define a few internal macros for easy use of run_once functions

Because pthread_once() takes a function taking no argument and
returning nothing, and we want to be able to check if they're
successful, we define a few internal macros to get around the issue.

Reviewed-by: Kurt Roeckx <kurt@openssl.org>

RT4593: Add space after comma (doc nits)

Update find-doc-nits to find errors in SYNOPSIS (the most common
place where they were missing).

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

Fix forgotten goto

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

Fix building with no-cms

The new fuzzing code broke no-cms

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

Convert the last uses of sockaddr in apps/* to use BIO_ADDR instead

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

Fix two bugs in clienthello processing

- Always process ALPN (previously there was an early return in the
  certificate status handling)
- Don't send a duplicate alert. Previously, both
  ssl_check_clienthello_tlsext_late and its caller would send an
  alert. Consolidate alert sending code in the caller.

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

SSL test framework: port NPN and ALPN tests

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

Cleanup after sk_push fail

Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1281)

Update error codes following tls_process_key_exchange() refactor

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

Tidy up tls_process_key_exchange()

After the refactor of tls_process_key_exchange(), this commit tidies up
some loose ends.

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

Split out ECDHE from tls_process_key_exchange()

Continuing from the previous commit. Refactor tls_process_key_exchange() to
split out into a separate function the ECDHE aspects.

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

Split out DHE from tls_process_key_exchange()

Continuing from the previous commit. Refactor tls_process_key_exchange() to
split out into a separate function the DHE aspects.

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

Split out SRP from tls_process_key_exchange()

Continuing from the previous commit. Refactor tls_process_key_exchange() to
split out into a separate function the SRP aspects.

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

Split out the PSK preamble from tls_process_key_exchange()

The tls_process_key_exchange() function is too long. This commit starts
the process of splitting it up by moving the PSK preamble code to a
separate function.

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

Move the PSK preamble for tls_process_key_exchange()

The function tls_process_key_exchange() is too long. This commit moves
the PSK preamble processing out to a separate function.

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

Narrow scope of locals vars in tls_process_key_exchange()

Narrow the scope of the local vars in preparation for split up this
function.

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

Add more session tests

Add some more tests for sessions following on from the previous commit
to ensure the callbacks are called when appropriate.

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

Remove sessions from external cache, even if internal cache not used.

If the SSL_SESS_CACHE_NO_INTERNAL_STORE cache mode is used then we weren't
removing sessions from the external cache, e.g. if an alert occurs the
session is supposed to be automatically removed.

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

Have the Travis builds do a "make update"

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

make update

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

Fixup a few SSLerr calls in ssl/statem/

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

Fixup collision between SSL_F_TLS_PROCESS_SKE and SSL_F_TLS_PROCESS_CKE macros

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

Check and print out boolean type properly.

If underlying type is boolean don't check field is NULL.

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

Refactor Identity Hint handling

Don't call strncpy with strlen of the source as the length. Don't call
strlen multiple times. Eventually we will want to replace this with a proper
PACKET style handling (but for construction of PACKETs instead of just
reading them as it is now). For now though this is safe because
PSK_MAX_IDENTITY_LEN will always fit into the destination buffer.

This addresses an OCAP Audit issue.

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

Fix up error codes after splitting up tls_construct_key_exchange()

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

Some tidy ups after the CKE construction refactor

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

Split out SRP CKE construction into a separate function

Continuing previous commit to break up the
tls_construct_client_key_exchange() function. This splits out the SRP
code.

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

Split out GOST CKE construction into a separate function

Continuing previous commit to break up the
tls_construct_client_key_exchange() function. This splits out the GOST
code.

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

Split out DHE CKE construction into a separate function

Continuing previous commit to break up the
tls_construct_client_key_exchange() function. This splits out the ECDHE
code.

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

Split out DHE CKE construction into a separate function

Continuing previous commit to break up the
tls_construct_client_key_exchange() function. This splits out the DHE
code.

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

Split out CKE construction PSK pre-amble and RSA into a separate function

The tls_construct_client_key_exchange() function is too long. This splits
out the construction of the PSK pre-amble into a separate function as well
as the RSA construction.

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

Narrow the scope of local variables in tls_construct_client_key_exchange()

This is in preparation for splitting up this over long function.

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

Fix bug with s2n et al macros

The parameters should have parens around them when used.

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

Errors fix up following break up of CKE processing

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

Remove the f_err lable from tls_process_client_key_exchange()

The f_err label is no longer needed so it can be removed.

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

Split out GOST from process CKE code

Continuing from the previous commits, this splits out the GOST code into
a separate function from the process CKE code.

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

Split out ECDHE from process CKE code

Continuing from the previous commits, this splits out the ECDHE code into
a separate function from the process CKE code.

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

Split out DHE from process CKE code

Continuing from the previous commit, this splits out the DHE code into
a separate function from the process CKE code.

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

Split out PSK preamble and RSA from process CKE code

The tls_process_client_key_exchange() function is far too long. This
splits out the PSK preamble processing, and the RSA processing into
separate functions.

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

Reduce the scope of some variables in tls_process_client_key_exchange()

In preparation for splitting this function up into smaller functions this
commit reduces the scope of some of the variables to only be in scope for
the algorithm specific parts. In some cases that makes the error handling
more verbose than it needs to be - but we'll clean that up in a later
commit.

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

Don't make a difference between building test programs and other programs

This adds a new target 'build_programs' and makes 'build_apps' and
'build_tests' aliases for it, for backward compatibility.

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

Use _NO_INST in some build.info files

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

Adapt the build files to the new "install" hash table

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

Document the _NO_INST variants

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

build.info: implement PROGRAM_NO_INST, and dito for ENGINES, SCRIPTS, LIBS

PROGRAM_NO_INST, ENGINES_NO_INST, SCRIPTS_NO_INST and LIBS_NO_INST are
to be used to specify program, engines, scripts and libraries that are
not to be installed in the system.  Fuzzers, test programs, that sort
of things are of the _NO_INST type, for example.

For the benefit of build file templates and other templates that use
data from configdata.pm, a new hash table $unified_info{install} is
created.  It contains a set of subhashes, one for each type of
installable, each having an array of file names as values.  For
example, it can look like this:

    "install" =>
        {
            "engines" =>
                [
                    "engines/afalg/afalg",
                    "engines/capi",
                    "engines/dasync",
                    "engines/padlock",
                ],
            "libraries" =>
                [
                    "libcrypto",
                    "libssl",
                ],
            "programs" =>
                [
                    "apps/openssl",
                ],
            "scripts" =>
                [
                    "apps/CA.pl",
                    "apps/tsget",
                    "tools/c_rehash",
                ],
        },

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

Fix print of ASN.1 BIGNUM type.

The ASN.1 BIGNUM type needs to be handled in a custom way as it is
not a generic ASN1_STRING type.

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

Fix formatting in statem_srvr.c based on review feedback

Also elaborated a comment based on feedback.

Reviewed-by: Emilia Käsper <emilia@openssl.org>

Ensure Travis tests SSLv3

Switch on Travis testing of SSLv3.

Reviewed-by: Emilia Käsper <emilia@openssl.org>

Try and make the transition tests for CKE message clearer

The logic testing whether a CKE message is allowed or not was a little
difficult to follow. This tries to clean it up.

Reviewed-by: Emilia Käsper <emilia@openssl.org>

Simplify key_exchange_expected() logic

The static function key_exchange_expected() used to return -1 on error.
Commit 361a119127 changed that so that it can never fail. This means that
some tidy up can be done to simplify error handling in callers of that
function.

Reviewed-by: Emilia Käsper <emilia@openssl.org>

Fix client auth test_ssl_new failures when enabling/disabling protocols

If configuring for anything other than the default TLS protocols then
test failures were occuring.

Reviewed-by: Emilia Käsper <emilia@openssl.org>

Make sure we call ssl3_digest_cached_records() when necessary

Having received a ClientKeyExchange message instead of a Certificate we
know that we are not going to receive a CertificateVerify message. This
means we can free up the handshake_buffer. However we better call
ssl3_digest_cached_records() instead of just freeing it up, otherwise we
later try and use it anyway and a core dump results. This could happen,
for example, in SSLv3 where we send a CertificateRequest but the client
sends no Certificate message at all. This is valid in SSLv3 (in TLS
clients are required to send an empty Certificate message).

Found using the BoringSSL test suite.

Reviewed-by: Emilia Käsper <emilia@openssl.org>

Fix SSLv3 alert if no Client Ceritifcate sent after a request for one

In TLS if the server sends a CertificateRequest and the client does not
provide one, if the server cannot continue it should send a
HandshakeFailure alert. In SSLv3 the same should happen, but instead we
were sending an UnexpectedMessage alert. This is incorrect - the message
isn't unexpected - it is valid for the client not to send one - its just
that we cannot continue without one.

Reviewed-by: Emilia Käsper <emilia@openssl.org>

Fix SSLv3 ClientAuth alert checking

In TLS during ClientAuth if the CA is not recognised you should get an
UnknownCA alert. In SSLv3 this does not exist and you should get a
BadCertificate alert.

Reviewed-by: Emilia Käsper <emilia@openssl.org>

Fix Client Auth tests

The Client Auth tests were not correctly setting the Protocol, so that this
aspect had no effect. It was testing the same thing lots of times for
TLSv1.2 every time.

Reviewed-by: Emilia Käsper <emilia@openssl.org>

Prepare the client certificate earlier

Move the preparation of the client certificate to be post processing work
after reading the CertificateRequest message rather than pre processing
work prior to writing the Certificate message. As part of preparing the
client certificate we may discover that we do not have one available. If
we are also talking SSLv3 then we won't send the Certificate message at
all. However, if we don't discover this until we are about to send the
Certificate message it is too late and we send an empty one anyway. This
is wrong for SSLv3.

Reviewed-by: Emilia Käsper <emilia@openssl.org>

Fix mingw build

Mingw builds on Travis were failing because INT_MAX was undeclared.

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

Fix strict-warnings build

The i2d_SCT_LIST function is declared as __owur, therefore we need to check
the result or a --strict-warnings build will fail.

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

aes/asm/aesfx-sparcv9.pl: switch to fshiftorx to improve single-block
and short-input performance.

[Fix bug in misaligned output handling.]

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

SPARC assembly pack: enforce V8+ ABI constraints.

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

evp/e_aes.c: wire new CBC and CTR subroutines from aesfx-sparcv9.

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

aes/asm/aesfx-sparcv9.pl: add "teaser" CBC and CTR subroutines.

[Also optimize aligaddr usage in single-block subroutines.]

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

fuzzers: print and convert it back

Reviewed-by: Rich Salz <rsalz@openssl.org>
GH: #1323

Return error when trying to print invalid ASN1 integer

Reviewed-by: Rich Salz <rsalz@openssl.org>
GH: #1322

INSTALL: clarify --cross-compile-prefix.

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

crypto/LPdir_win.c: rationalize temporary allocations.

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

crypto/LPdir_win.c: harmonize with o_fopen.c.

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

Make fuzzer and fuzz tester builds less magic

Instead of having fuzz/build.info.fuzz magically and conditionally
included along with the other build.info files, incorporate it in
fuzz/build.info and add the conditions there instead.

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

fix memory leaks

Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1313)

Remove the silly CVS markers from LPdir_*.c

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

Add back lost copyright and license text in LPdir_win.c

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

Fix: dummy definition of rand_hw_seed() should also return int

Reviewed-by: Stephen Henson <steve@openssl.org>

Fix ASN.1 private encode of EC_KEY to not change the input key

RT#4611

Reviewed-by: Stephen Henson <steve@openssl.org>

check return values for EVP_Digest*() APIs

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

Windows: allow input and output flags to end with a space, or not

With a number of tools, especially those coming with Visual Studio,
some command options are separated from their argument with a space,
others with a space.  Since we parametrise them, we can't know
beforehand which it will be, so we must allow the input and output
options to have either.

However, spaces at the end of nmake macro values are trimmed, so allow
spaces to exist by adding a reference to an undefined macro at the end.

Reviewed-by: Andy Polyakov <appro@openssl.org>

crypto/x86[_64]cpuid.pl: add OPENSSL_ia32_rd[rand|seed]_bytes.

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

Small fixes in unix-Makefile.tmpl

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

Install applink.c with the public header files.

This is only done for the platforms where 'OPENSSL_USE_APPLINK' is defined.

Also, change the docs of OPENSSL_Applink to say where to find applink.c
in the installation directory.

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

Windows: take care of manifest files

The easiest way to take care of manifest files is to integrate them
into the associated binary (.exe or .dll).  MT (the Manifest Tool) is
the utility to use for this.

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

make update

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

Add OCSP accessors.

RT#4605

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

Fix DH error-handling in tls_process_key_exchange.

The set0 setters take ownership of their arguments, so the values should
be set to NULL to avoid a double-free in the cleanup block should
ssl_security(SSL_SECOP_TMP_DH) fail. Found by BoringSSL's WeakDH test.

Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1299)

Make update

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

Perform DANE-EE(3) name checks by default

In light of potential UKS (unknown key share) attacks on some
applications, primarily browsers, despite RFC761, name checks are
by default applied with DANE-EE(3) TLSA records.  Applications for
which UKS is not a problem can optionally disable DANE-EE(3) name
checks via the new SSL_CTX_dane_set_flags() and friends.

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

Extend mkcert.sh to support nameConstraints generation and more complex
subject alternate names.

Add nameConstraints tests incluing DNS, IP and email tests both in
subject alt name extension and subject name.

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