Reviewed-by: NRich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2609)

Reviewed-by: NRich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2609)

This implements the server side KeyUpdate sending capability as well.
Reviewed-by: NRich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2609)

Reviewed-by: NKurt Roeckx <kurt@openssl.org>
Reviewed-by: NAndy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2655)

Fortunately, "openssl verify" makes good use of that API
Reviewed-by: NRich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2652)

Reviewed-by: NRich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2652)

Reviewed-by: NRichard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2650)

Reviewed-by: NRich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2643)

Reviewed-by: NRich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2643)

For client auth call tls_choose_sigalg to select the certificate
and signature algorithm. Use the selected algorithm in
tls_construct_cert_verify.

Remove obsolete tls12_get_sigandhash.
Reviewed-by: NRich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2643)

Select appropriate signature algorithm and certificate for client
authentication using tls_choose_sigalg.

A lot of selection logic is very similar except not finding a
certificate is not a fatal error: we just do not present a
certificate.

For TLS 1.2 and earlier we only check the current certificate
is suitable (for compatibility with previous logic) for TLS 1.3
(where there are no compatibility issues) we support multiple
client certificates for different algorithms.
Reviewed-by: NRich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2643)

Change req_check_len error message, it also accepts 20 bytes, but states 'less than' in the error message
Reviewed-by: NMatt Caswell <matt@openssl.org>
Reviewed-by: NRich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2648)

The intent seems to be that the WIN32 symbol is for things that are a direct
byproduct of being a windows-variant configuration and should be used for
feature en/disablement on windows systems.  Use of the _WIN32 symbol is more
widespread, being used to implement platform portability of more generic code.

We do define WIN32 in some situations in e_os.h, but that is not included
universally.
Reviewed-by: NAndy Polyakov <appro@openssl.org>
Reviewed-by: NRich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2642)

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

Following on from CVE-2017-3733, this removes the OPENSSL_assert() check
that failed and replaces it with a soft assert, and an explicit check of
value with an error return if it fails.
Reviewed-by: NRichard Levitte <levitte@openssl.org>

In 1.1.0 changing the ciphersuite during a renegotiation can result in
a crash leading to a DoS attack. In master this does not occur with TLS
(instead you get an internal error, which is still wrong but not a security
issue) - but the problem still exists in the DTLS code.

The problem is caused by changing the flag indicating whether to use ETM
or not immediately on negotiation of ETM, rather than at CCS. Therefore,
during a renegotiation, if the ETM state is changing (usually due to a
change of ciphersuite), then an error/crash will occur.

Due to the fact that there are separate CCS messages for read and write
we actually now need two flags to determine whether to use ETM or not.

CVE-2017-3733
Reviewed-by: NRichard Levitte <levitte@openssl.org>

In 1.1.0 changing the ciphersuite during a renegotiation can result in
a crash leading to a DoS attack. In master this does not occur with TLS
(instead you get an internal error, which is still wrong but not a security
issue) - but the problem still exists in the DTLS code.

This commit provides a test for the issue.

CVE-2017-3733
Reviewed-by: NRichard Levitte <levitte@openssl.org>

Reviewed-by: NRich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2644)

Reviewed-by: NAndy Polyakov <appro@openssl.org>
Reviewed-by: NRich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2629)

Fix a typo. Probably this has not been found because EVP_CIPHER_CTX is
smaller than EVP_CHACHA_AEAD_CTX and heap overflow does not occur.
Reviewed-by: NRichard Levitte <levitte@openssl.org>
Reviewed-by: NRich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2294)

Capability probing by catching SIGILL appears to be problematic
on iOS. But since Apple universe is "monocultural", it's actually
possible to simply set pre-defined processor capability mask.
Reviewed-by: NRich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2617)

Three modules were left behind in a2859927.
Reviewed-by: NRich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2617)

Reviewed-by: NRichard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2637)

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

Reviewed-by: NMatt Caswell <matt@openssl.org>
Reviewed-by: NRich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2636)

Causes make update to fail.
Reviewed-by: NRichard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2634)

Reviewed-by: NRich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2623)

Reviewed-by: NRich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2623)

Use negotiated signature algorithm and certificate index in
tls_construct_key_exchange instead of recalculating it.
Reviewed-by: NRich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2623)

Now the certificate and signature algorithm is set in one place we
can use it directly insetad of recalculating it. The old functions
ssl_get_server_send_pkey() and ssl_get_server_cert_index() are no
longer required.
Reviewed-by: NRich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2623)

Update tls_choose_sigalg to choose a signature algorithm for all
versions of TLS not just 1.3.

For TLS 1.2 we choose the highest preference signature algorithm
for the chosen ciphersuite.

For TLS 1.1 and earlier the signature algorithm is determined by
the ciphersuite alone. For RSA we use a special MD5+SHA1 signature
algorithm.
Reviewed-by: NRich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2623)

Reviewed-by: NRich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2623)

Add inline function ssl_has_cert which checks to see if a certificate and
private key for a given index are not NULL.
Reviewed-by: NRich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2623)

[skip ci]
Reviewed-by: NRichard Levitte <levitte@openssl.org>
Reviewed-by: NRich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2571)

Reviewed-by: NRichard Levitte <levitte@openssl.org>
Reviewed-by: NRich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1613)

Prevents that OPENSSL_gmtime incorrectly signals success if gmtime_r fails, and that struct* tm result's possibly uninitialized content is used
Reviewed-by: NRichard Levitte <levitte@openssl.org>
Reviewed-by: NRich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1613)

Reviewed-by: NMatt Caswell <matt@openssl.org>
Reviewed-by: NRich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2618)

Prevent allocations of size 0 in sh_init, which are not possible with the default OPENSSL_zalloc, but are possible if the user has installed their own allocator using CRYPTO_set_mem_functions. If the 0-allocations succeeds, the secure heap code will later access (at least) the first byte of that space, which is technically an OOB access. This could lead to problems with some custom allocators that only return a valid pointer for subsequent free()-ing, and do not expect that the pointer is actually dereferenced.
Reviewed-by: NRichard Levitte <levitte@openssl.org>
Reviewed-by: NRich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2605)

Reviewed-by: NAndy Polyakov <appro@openssl.org>
Reviewed-by: NRich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2300)

Reviewed-by: NRichard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2626)