Use correct signature algorithm list when sending or checking.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2298)

Prepare for 1.0.2l-dev

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

Prepare for 1.0.2k release

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

Update CHANGES and NEWS for new release

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

Better check of DH parameters in TLS data

When the client reads DH parameters from the TLS stream, we only
checked that they all are non-zero.  This change updates the check
as follows:

    check that p is odd
    check that 1 < g < p - 1

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

bn/asm/x86_64-mont5.pl: fix carry bug in bn_sqr8x_internal.

CVE-2017-3732

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

crypto/evp: harden RC4_MD5 cipher.

Originally a crash in 32-bit build was reported CHACHA20-POLY1305
cipher. The crash is triggered by truncated packet and is result
of excessive hashing to the edge of accessible memory (or bogus
MAC value is produced if x86 MD5 assembly module is involved). Since
hash operation is read-only it is not considered to be exploitable
beyond a DoS condition.

Thanks to Robert Święcki for report.

CVE-2017-3731

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

Fix error handling in compute_key, BN_CTX_get can return NULL

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2132)
(cherry picked from commit 7928ee4d685b727619555bc1ec0aee805f6fc8c4)

Fix a ssl session leak due to OOM in lh_SSL_SESSION_insert

- s == NULL can mean c is a new session *or* lh_insert was
  unable to create a hash entry.
- use lh_SSL_SESSION_retrieve to check for this error condition.
- If it happens simply remove the extra reference again.

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2138)
(cherry picked from commit 38088ce9934a90d4aea486edbff864f3935342e6)

Fix SSL_VERIFY_CLIENT_ONCE

The flag SSL_VERIFY_CLIENT_ONCE is documented as follows:

  B<Server mode:> only request a client certificate on the initial TLS/SSL
  handshake. Do not ask for a client certificate again in case of a
  renegotiation. This flag must be used together with SSL_VERIFY_PEER.

  B<Client mode:> ignored

But the implementation actually did nothing. After the server sends its
ServerKeyExchange message, the code was checking s->session->peer to see if
it is NULL. If it was set then it did not ask for another client
certificate. However s->session->peer will only be set in the event of a
resumption, but a ServerKeyExchange message is only sent in the event of a
full handshake (i.e. no resumption).

The documentation suggests that the original intention was for this to
have an effect on renegotiation, and resumption doesn't come into it.

The fix is to properly check for renegotiation, not whether there is already
a client certificate in the session.

As far as I can tell this has been broken for a *long* time.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1984)

Add missing va_end

Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2267)

Fix DSA parameter generation control error

When setting the digest parameter for DSA parameter generation, the
signature MD was set instead of the parameter generation one.
Fortunately, that's also the one that was used for parameter
generation, but it ultimately meant the parameter generator MD and the
signature MD would always be the same.

Fixes github issue #2016

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2250)
(cherry picked from commit 8a05c4d3b5a1bfb9193ea24e71735e11de7168d2)

Clean one unused variable, plus an useless one.

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1906)
(cherry picked from commit 2191dc846a85ce82925cb06b4dd8649da7fc403c)

GH1986: Document -header flag.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2209)

Fix error handling in SSL_CTX_new

Dont free rbuf_freelist here, SSL_CTX_free will do that.

Signed-off-by: Kurt Roeckx <kurt@roeckx.be>
Reviewed-by: Rich Salz <rsalz@openssl.org>
GH: #2129

Fix a memory leak in RSA_padding_add_PKCS1_OAEP_mgf1

Signed-off-by: Kurt Roeckx <kurt@roeckx.be>
Reviewed-by: Matt Caswell <matt@openssl.org>
GH: #2140
(cherry picked from commit c6d215e0d278fcd51ad474a3647b61e1b67290bb)

replace "will lookup up" by "will look up"

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
CLA: trivial
(Merged from https://github.com/openssl/openssl/pull/2145)
(cherry picked from commit 67adf0a7c273a82901ce8705ae8d71ee2f1c959c)

Reformat M_check_autoarg to match our coding style

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2121)
(cherry picked from commit 2629440d42e4d64cd0cb849c1b19fa87a4fcb90f)

M_check_autoarg: sanity check the key

For now, checking that the size is non-zero will suffice.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2120)
(cherry picked from commit d7c8f142ea5953bf260b70a58739c1c9b0f038eb)

Fix typo.

Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
CLA: trivial
(Merged from https://github.com/openssl/openssl/pull/2086)
(cherry picked from commit 0b742f93ea7882a447f6523ac56a6f847d9f8e92)

(cherry picked from commit f7a2da1d584bed2e05774f92d69fee39ce3edda2)

zero pad DHE public key in ServerKeyExchange message for interop

Some versions of the Microsoft TLS stack have problems when the DHE public key
is encoded with fewer bytes than the DHE prime. (Backported from master)

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1350)

Fix ssl_cert_dup: change one 'return NULL' to 'goto err'

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2082)

Make 'err' lable in ssl_cert_dup unconditional

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2082)

Fix a bug in clienthello processing

- Always process ALPN (previously there was an early return in the
  certificate status handling)

1.0.2 did not have the double-alert issue from master, but it seems
cleanest to pull in the structural change to alert handling anyway
and jump to f_err instead of err to send the alert in the caller.

(cherry picked from commit 70c22888c1648fe8652e77107f3c74bf2212de36)

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

perlasm/x86_64-xlate.pl: refine sign extension in ea package.

$1<<32>>32 worked fine with either 32- or 64-bit perl for a good while,
relying on quirk that [pure] 32-bit perl performed it as $1<<0>>0. But
this apparently changed in some version past minimally required 5.10,
and operation result became 0. Yet, it went unnoticed for another while,
because most perl package providers configure their packages with
-Duse64bitint option.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(cherry picked from commit 82e089308bd9a7794a45f0fa3973d7659420fbd8)

UI_OpenSSL()'s session opener fails on MacOS X

If on a non-tty stdin, TTY_get() will fail with errno == ENODEV.
We didn't catch that.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2039)
(cherry picked from commit c901bccec6f747467e1af31473655c8290e32309)

VMS UI_OpenSSL: if the TT device isn't a tty, flag instead of error

On all platforms, if the controlling tty isn't an actual tty, this is
flagged by setting is_a_tty to zero...  except on VMS, where this was
treated as an error.  Change this to behave like the other platforms.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2064)

Check input length to pkey_rsa_verify()

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2065)
(cherry picked from commit 71bbc79b7d3b1195a7a7dd5f547d52ddce32d6f0)

Conflicts:
crypto/rsa/rsa_err.c
include/openssl/rsa.h

Remove extra bang

A bang (!) slipped through in the recent UI cleanup

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2051)
(cherry picked from commit 949320c567811e714216ea987fe24eea1b56da5e)

UI code style cleanup

Mostly condition check changes.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/2047)
(cherry picked from commit 120fb9e43656e1801c75a4fbb7c178ebec9bac18)

Revert "Fix heartbeat_test"

Commit fa4c37457 got reverted, so this one also needs to be reverted as
a result.

This reverts commit ad69a30323cbc6723c2387d6ce546a51b10c42d0.

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

apps/speed.c: Fix crash when config loading fails

Move rsa_key initialization in front of load_config().
If loading the config fails, rsa_key isn't initialized and may
cause invalid free() in the end: cleanup.
Remove superfluous memset.

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

INSTALL: clarify 386 and no-sse2 options.

This is 1.0.2-specific reformat of 5ae5dc96610f0a598dac9d2f267b5c0ddd77b2e4.

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

modes/ctr128.c: fix false carry in counter increment procedure.

GH issue #1916 affects only big-endian platforms. TLS is not affected,
because TLS fragment is never big enough.

Reviewed-by: Matt Caswell <matt@openssl.org>
(cherry picked from commit 76f572ed0469a277d92378848250b7a9705d3071)

Clarify what X509_NAME_online does with the given buffer and size

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1980)
(cherry picked from commit 19cb71ef6e414759d737918bab10be2cc1d8bd99)

(cherry picked from commit 793d9b79033c2fffc8e781dab2fd678661b348cd)

Make SSL_read and SSL_write return the old behaviour and document it.

Backport of beacb0f0c1ae7b0542fe053b95307f515b578eb7, revert of
fa4c374572e94f467900f5820cd1d00af2470a17

Fixes: #1903

Reviewed-by: Matt Caswell <matt@openssl.org>
GH: #1967

Use consistent variable names

In the X509_NAME_get_index_by_NID.pod example, the initialized variable is called
"loc", but the one used in the for loop is called "lastpos". Make the names match.

CLA: trivial
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1949)

(cherry picked from commit 4b9c2669f31da26bfe56f629973fa014083dc2a0)

domd: Preserve Makefile time when it is unchanged

also on systems with makedepend that does not report its version, or that
its version does not contain "gcc" or "clang".

Some versions of makedepends just overwrite Makefile. Preserve the
timestamp of the previous Makefile, and copy it back if it is unchanged.

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1370)

mklink: Do not needlessly overwrite linked files...

... on systems with symlinks.

Creating or overwriting a symlink sets the file ctime to the current time.
This causes needless rebuilds because the time of all the headers is
changed, and apparently make considers the link's time rather than the
time of the target.

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1370)

domd: Do not needlessly overwrite Makefiles

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1370)

mklink: Do not needlessly overwrite linked files...

... on systems without symlinks.

Overwriting all the headers on each Configure causes full rebuild even if
nothing has changed.

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1370)

Configure: Improve incremental build time

When Makefile/opensslconf.h is unchanged, don't write it at all.

Currently every time Configure is executed, these files are overwritten.
Makefile leads to regeneration of buildinf.h, and opensslconf.h is itself
a central header.

As a result, Configure triggers full rebuild, even if nothing is changed.

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1370)

Check return value of some BN functions.

Factorise multiple bn_get_top(group->field) calls
Add missing checks on some conditional BN_copy return value
Add missing checks on some BN_copy return value
Add missing checks on a few bn_wexpand return value

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1626)
(cherry picked from commit 78e09b53a40729f5e99829ccc733b592bd22fea1)

Solution proposal for issue #1647.

Avoid a memory alignment issue.

Signed-off-by: Matthias Kraft <Matthias.Kraft@softwareag.com>
CLA: trivial
Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1650)

Update CHANGES and NEWS

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

bn/asm/x86_64-mont.pl: fix for CVE-2016-7055 (Low severity).

Reviewed-by: Rich Salz <rsalz@openssl.org>
(cherry picked from commit 2fac86d9abeaa643677d1ffd0a139239fdf9406a)

Makefile.org: clear APPS environment variable.

Build failure was reported in GH#1818 if APPS environment was defined.

Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1864)

Missed a mention of RT

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1849)
(cherry picked from commit 1e62cc12f35408508594be254f40bf9b65d2a3a9)

Add a CHANGES entry for the unrecognised record type change

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

Fail if an unrecognised record type is received

TLS1.0 and TLS1.1 say you SHOULD ignore unrecognised record types, but
TLS 1.2 says you MUST send an unexpected message alert. We swap to the
TLS 1.2 behaviour for all protocol versions to prevent issues where no
progress is being made and the peer continually sends unrecognised record
types, using up resources processing them.

Issue reported by 郭志攀

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

Fix heartbeat_test

The heartbeat_test reaches into the internals of libssl and calls some
internal functions. It then checks the return value to check its what it
expected. However commit fa4c37457 changed the return value of these
internal functions, and now the test is failing.

The solution is to update the test to look for the new return value.

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

Secure our notification email.

Forks will have to define their own

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1821)
(cherry picked from commit 5e28b1c1e048eef600dc49820934a5e1531186d4)

Fix grammar-o in CONTRIBUTING

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1625)
(cherry picked from commit e4d94269a5a41594852dc60716500580f1d47cef)

Add $(EX_LIBS) to the LIBDEPS for libgost.so, just as for all other engines

Without this, any linker flag the user gave when configuring are ignored.

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1803)

Implement length checks as a macro

Replace the various length checks in the extension code with a macro to
simplify the logic.

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

Ensure we have length checks for all extensions

The previous commit inspired a review of all the length checks for the
extension adding code. This adds more robust checks and adds checks where
some were missing previously. The real solution for this is to use WPACKET
which is currently in master - but that cannot be applied to release
branches.

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

Fix length check writing status request extension

The status request extension did not correctly check its length, meaning
that writing the extension could go 2 bytes beyond the buffer size. In
practice this makes little difference because, due to logic in buffer.c the
buffer is actually over allocated by approximately 5k!

Issue reported by Guido Vranken.

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

Fix a double free in ca command line

Providing a spkac file with no default section causes a double free.

Thanks to Brian Carpenter for reporting this issue.

Reviewed-by: Kurt Roeckx <kurt@openssl.org>
(cherry picked from commit 229bd12487f8576fc088dc4f641950ac33c62033)

A zero return from BIO_read/BIO_write() could be retryable

A zero return from BIO_read()/BIO_write() could mean that an IO operation
is retryable. A zero return from SSL_read()/SSL_write() means that the
connection has been closed down (either cleanly or not). Therefore we
should not propagate a zero return value from BIO_read()/BIO_write() back
up the stack to SSL_read()/SSL_write(). This could result in a retryable
failure being treated as fatal.

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

Fix typo (reported by Matthias St. Pierre)

Reviewed-by: Tim Hudson <tjh@openssl.org>
(cherry picked from commit 12a7715e3daed439e46cbed461d2a3d9dfd37c0f)

Fix leak of secrecy in ecdh_compute_key()

A temporary buffer containing g^xy was not cleared in ecdh_compute_key()
before freeing it, so the shared secret was leaked in memory.

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

Correctly find all critical CRL extensions

Unhandled critical CRL extensions were not detected if they appeared
after the handled ones.  (GitHub issue 1757).  Thanks to John Chuah
for reporting this.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1769)

remove redundant zero assignments

The structure has already been initialized to zero with memset.
See also commit 64b25758edca688a30f02c260262150f7ad0bc7d (remove 0 assignments)

Signed-off-by: Cristian Stoica <cristian.stoica@nxp.com>
CLA: trivial
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1463)

%p takes void*, so make sure to cast arguments to void*

This avoids failures when configuring with --strict-warnings

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1749)

apps: make setup_engine() and release_engine() available always

This removes some #ifndef clutter.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1644)

If an engine comes up explicitely, it must also come down explicitely

In apps/apps.c, one can set up an engine with setup_engine().
However, we freed the structural reference immediately, which means
that for engines that don't already have a structural reference
somewhere else (because it has registered at least one cipher or digest
algorithm method, and therefore gets a functional reference through the
ENGINE_set_default() call), we end up returning an invalid reference.

Instead, the function release_engine() is added, and called at the end
of the routines that call setup_engine().

Originally, the ENGINE API wasn't designed for this to happen, an
engine had to register at least one algorithm method, and was
especially expected to register the algorithms corresponding to the
key types that could be stored and hidden in hardware.  However, it
turns out that some engines will not register those algorithms with
the ENGINE_set_{algo}, ENGINE_set_cipher or ENGINE_set_digest
functions, as they only want the methods to be used for keys, not as
general crypto accelerator methods.  That may cause ENGINE_set_default()
to do nothing, and no functional reference is therefore made, leading
to a premature deallocation of the engine and it thereby becoming
unavailable when trying to fetch a key.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1644)

Fix no-des

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1748)

Make 'openssl prime ""' not segfault

Fixes RT#4699

Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1668)

Fix strict-warnings build

crypto/evp/e_aes.c: Types of inp and out parameters of AES_xts_en/decrypt
functions need to be changed from char to unsigned char to avoid build
error due to -Werror=incompatible-pointer-types.

crypto/aes/asm/aes-s390x.pl: Comments need to reflect the above change.

Signed-off-by: Patrick Steuer <psteuer@mail.de>
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
CLA: trivial

Fix strict-warnings build

crypto/s390xcap.c: cryptlib.h needs to be included for OPENSSL_cpuid_setup
function prototype is located there to avoid build error due to
-Werror=missing-prototypes.

Signed-off-by: Patrick Steuer <psteuer@mail.de>
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
CLA: trivial

Fix signatures of EVP_Digest{Sign,Verify}Update

These are implemented as macros delegating to `EVP_DigestUpdate`, which
takes a `size_t` as its third argument, not an `unsigned int`.

CLA: trivial

Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(cherry picked from commit 8bdce8d160e29b4e1b80fec31f618d85d8c2b7a8)

Ensure we handle len == 0 in ERR_err_string_n

If len == 0 in a call to ERR_error_string_n() then we can read beyond the
end of the buffer. Really applications should not be calling this function
with len == 0, but we shouldn't be letting it through either!

Thanks to Agostino Sarubbo for reporting this issue. Agostino's blog on
this issue is available here:
https://blogs.gentoo.org/ago/2016/10/14/openssl-libcrypto-stack-based-buffer-overflow-in-err_error_string_n-err-c/

Reviewed-by: Richard Levitte <levitte@openssl.org>
(cherry picked from commit e5c1361580d8de79682958b04a5f0d262e680f8b)

Degrade 3DES to MEDIUM in SSL2

The SWEET32 fix moved 3DES from HIGH to MEDIUM, but omitted SSL2.

CLA: trivial
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1683)

RT is put out to pasture

Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1702)
(cherry picked from commit 7954dced19a7e59e7055eab95a981fa943c7d100)

Add missing error string for SSL_R_TOO_MANY_WARN_ALERTS

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

apps/apps.c: initialize and de-initialize engine around key loading

Before loading a key from an engine, it may need to be initialized.
When done loading the key, we must de-initialize the engine.
(if the engine is already initialized somehow, only the reference
counter will be incremented then decremented)

Reviewed-by: Stephen Henson <steve@openssl.org>
(cherry picked from commit 49e476a5382602d0bad1139d6f1f66ddbc7959d6)

Revert "Call ENGINE_init() before trying to use keys from engine"

This reverts commit 4badd2b3c29c2c6c551c737c07a429a53d9d1a0d.
This fails to call ENGINE_finish; an alternate fix is coming.

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

Call ENGINE_init() before trying to use keys from engine

Things like 'openssl s_client' only ever worked with keys from an engine
which provided a default generic method for some key type — because it
called ENGINE_set_default() and that ended up being an implicit
initialisation and functional refcount.

But an engine which doesn't provide generic methods doesn't get
initialised, and then when you try to use it you get an error:

cannot load client certificate private key file from engine
140688147056384:error:26096075:engine routines:ENGINE_load_private_key:not initialised:crypto/engine/eng_pkey.c:66:
unable to load client certificate private key file

cf. https://github.com/OpenSC/libp11/issues/107 (in which we discover
that engine_pkcs11 *used* to provide generic methods that OpenSSL would
try to use for ephemeral DH keys when negotiating ECDHE cipher suites in
TLS, and that didn't work out very well.)

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1640)

Fix NEWS error

The NEWS file referenced the wrong CVE for 1.0.2

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

Prepare for 1.0.2k-dev

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

Prepare for 1.0.2j release

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

Update CHANGES and NEWS for the new release

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

Add some sanity checks when checking CRL scores

Note: this was accidentally omitted from OpenSSL 1.0.2 branch.
Without this fix any attempt to use CRLs will crash.

CVE-2016-7052

Thanks to Bruce Stephens and Thomas Jakobi for reporting this issue.

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

Fix build with no-nextprotoneg

Add a missing ifdef. Same change is already present in master.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1100)

Fix typo introduced by a03f81f4

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

Prepare for 1.0.2j-dev

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

Prepare for 1.0.2i release

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

Updates CHANGES and NEWS for new release

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

Avoid KCI attack for GOST

Russian GOST ciphersuites are vulnerable to the KCI attack because they use
long-term keys to establish the connection when ssl client authorization is
on. This change brings the GOST implementation into line with the latest
specs in order to avoid the attack. It should not break backwards
compatibility.

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

Fix a mem leak in NPN handling

If a server sent multiple NPN extensions in a single ClientHello then a
mem leak can occur. This will only happen where the client has requested
NPN in the first place. It does not occur during renegotiation. Therefore
the maximum that could be leaked in a single connection with a malicious
server is 64k (the maximum size of the ServerHello extensions section). As
this is client side, only occurs if NPN has been requested and does not
occur during renegotiation this is unlikely to be exploitable.

Issue reported by Shi Lei.

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

Fix OCSP Status Request extension unbounded memory growth

A malicious client can send an excessively large OCSP Status Request
extension. If that client continually requests renegotiation,
sending a large OCSP Status Request extension each time, then there will
be unbounded memory growth on the server. This will eventually lead to a
Denial Of Service attack through memory exhaustion. Servers with a
default configuration are vulnerable even if they do not support OCSP.
Builds using the "no-ocsp" build time option are not affected.

I have also checked other extensions to see if they suffer from a similar
problem but I could not find any other issues.

CVE-2016-6304

Issue reported by Shi Lei.

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

mk1mf.pl: check for no-tls1 here as well

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

Don't allow too many consecutive warning alerts

Certain warning alerts are ignored if they are received. This can mean that
no progress will be made if one peer continually sends those warning alerts.
Implement a count so that we abort the connection if we receive too many.

Issue reported by Shi Lei.

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

Make message buffer slightly larger than message.

Grow TLS/DTLS 16 bytes more than strictly necessary as a precaution against
OOB reads. In most cases this will have no effect because the message buffer
will be large enough already.

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

Use SSL3_HM_HEADER_LENGTH instead of 4.

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

Remove unnecessary check.

The overflow check will never be triggered because the
the n2l3 result is always less than 2^24.

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

Dcoument -alpn flag

Reviewed-by: Richard Levitte <levitte@openssl.org>
(cherry picked from commit 776e15f9393a9e3083bec60a8da376ce2fe1e97e)

GH1555: Don't bump size on realloc failure

Reviewed-by: Richard Levitte <levitte@openssl.org>
(cherry picked from commit 6fcace45bda108ad4d3f95261494dd479720d92c)

apps/apps.c: include sys/socket.h to declare recv()

Reviewed-by: Tim Hudson <tjh@openssl.org>
(cherry picked from commit a19228b7f4fc6fcb49713455b3caedbc24fb0b01)

Fix small OOB reads.

In ssl3_get_client_certificate, ssl3_get_server_certificate and
ssl3_get_certificate_request check we have enough room
before reading a length.

Thanks to Shi Lei (Gear Team, Qihoo 360 Inc.) for reporting these bugs.

CVE-2016-6306

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