<?xml version="1.0" encoding="utf-8" ?>
<!DOCTYPE chapter SYSTEM "chapter.dtd">
<chapter>
<header>
<copyright>
<year>1999</year><year>2013</year>
<holder>Ericsson AB. All Rights Reserved.</holder>
</copyright>
<legalnotice>
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
</legalnotice>
<title>SSL Release Notes</title>
<file>notes.xml</file>
</header>
<p>This document describes the changes made to the SSL application.</p>
<section><title>SSL 7.3.3.2</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p> An erlang TLS server configured with cipher suites
using rsa key exchange, may be vulnerable to an Adaptive
Chosen Ciphertext attack (AKA Bleichenbacher attack)
against RSA, which when exploited, may result in
plaintext recovery of encrypted messages and/or a
Man-in-the-middle (MiTM) attack, despite the attacker not
having gained access to the server’s private key
itself. <url
href="https://nvd.nist.gov/vuln/detail/CVE-2017-1000385">CVE-2017-1000385</url>
</p> <p> Exploiting this vulnerability to perform
plaintext recovery of encrypted messages will, in most
practical cases, allow an attacker to read the plaintext
only after the session has completed. Only TLS sessions
established using RSA key exchange are vulnerable to this
attack. </p> <p> Exploiting this vulnerability to conduct
a MiTM attack requires the attacker to complete the
initial attack, which may require thousands of server
requests, during the handshake phase of the targeted
session within the window of the configured handshake
timeout. This attack may be conducted against any TLS
session using RSA signatures, but only if cipher suites
using RSA key exchange are also enabled on the server.
The limited window of opportunity, limitations in
bandwidth, and latency make this attack significantly
more difficult to execute. </p> <p> RSA key exchange is
enabled by default although least prioritized if server
order is honored. For such a cipher suite to be chosen it
must also be supported by the client and probably the
only shared cipher suite. </p> <p> Captured TLS sessions
encrypted with ephemeral cipher suites (DHE or ECDHE) are
not at risk for subsequent decryption due to this
vulnerability. </p> <p> As a workaround if default cipher
suite configuration was used you can configure the server
to not use vulnerable suites with the ciphers option like
this: </p> <c> {ciphers, [Suite || Suite <-
ssl:cipher_suites(), element(1,Suite) =/= rsa]} </c> <p>
that is your code will look somethingh like this: </p>
<c> ssl:listen(Port, [{ciphers, [Suite || Suite <-
ssl:cipher_suites(), element(1,S) =/= rsa]} | Options]).
</c> <p> Thanks to Hanno Böck, Juraj Somorovsky and
Craig Young for reporting this vulnerability. </p>
<p>
Own Id: OTP-14748</p>
</item>
</list>
</section>
</section>
<section><title>SSL 7.3.3.1</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
The TLS/SSL protocol version selection for the SSL server
has been corrected to follow RFC 5246 Appendix E.1
especially in case where the list of supported versions
has gaps. Now the server selects the highest protocol
version it supports that is not higher than what the
client supports.</p>
<p>
Own Id: OTP-13753 Aux Id: seq13150 </p>
</item>
</list>
</section>
</section>
<section><title>SSL 7.3.3.0.1</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p> An erlang TLS server configured with cipher suites
using rsa key exchange, may be vulnerable to an Adaptive
Chosen Ciphertext attack (AKA Bleichenbacher attack)
against RSA, which when exploited, may result in
plaintext recovery of encrypted messages and/or a
Man-in-the-middle (MiTM) attack, despite the attacker not
having gained access to the server’s private key
itself. <url
href="https://nvd.nist.gov/vuln/detail/CVE-2017-1000385">CVE-2017-1000385</url>
</p> <p> Exploiting this vulnerability to perform
plaintext recovery of encrypted messages will, in most
practical cases, allow an attacker to read the plaintext
only after the session has completed. Only TLS sessions
established using RSA key exchange are vulnerable to this
attack. </p> <p> Exploiting this vulnerability to conduct
a MiTM attack requires the attacker to complete the
initial attack, which may require thousands of server
requests, during the handshake phase of the targeted
session within the window of the configured handshake
timeout. This attack may be conducted against any TLS
session using RSA signatures, but only if cipher suites
using RSA key exchange are also enabled on the server.
The limited window of opportunity, limitations in
bandwidth, and latency make this attack significantly
more difficult to execute. </p> <p> RSA key exchange is
enabled by default although least prioritized if server
order is honored. For such a cipher suite to be chosen it
must also be supported by the client and probably the
only shared cipher suite. </p> <p> Captured TLS sessions
encrypted with ephemeral cipher suites (DHE or ECDHE) are
not at risk for subsequent decryption due to this
vulnerability. </p> <p> As a workaround if default cipher
suite configuration was used you can configure the server
to not use vulnerable suites with the ciphers option like
this: </p> <c> {ciphers, [Suite || Suite <-
ssl:cipher_suites(), element(1,Suite) =/= rsa]} </c> <p>
that is your code will look somethingh like this: </p>
<c> ssl:listen(Port, [{ciphers, [Suite || Suite <-
ssl:cipher_suites(), element(1,S) =/= rsa]} | Options]).
</c> <p> Thanks to Hanno Böck, Juraj Somorovsky and
Craig Young for reporting this vulnerability. </p>
<p>
Own Id: OTP-14748</p>
</item>
</list>
</section>
</section>
<section><title>SSL 7.3.3</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Correct ssl:prf/5 to use the negotiated cipher suite's
prf function in ssl:prf/5 instead of the default prf.</p>
<p>
Own Id: OTP-13546</p>
</item>
<item>
<p>
Timeouts may have the value 0, guards have been corrected
to allow this</p>
<p>
Own Id: OTP-13635</p>
</item>
<item>
<p>
Change of internal handling of hash sign pairs as the
used one enforced to much restrictions making some valid
combinations unavailable.</p>
<p>
Own Id: OTP-13670</p>
</item>
</list>
</section>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Create a little randomness in sending of session
invalidation messages, to mitigate load when whole table
is invalidated.</p>
<p>
Own Id: OTP-13490</p>
</item>
</list>
</section>
</section>
<section><title>SSL 7.3.2</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Correct cipher suites conversion and gaurd expression.
Caused problems with GCM cipher suites and client side
option to set signature_algorithms extention values.</p>
<p>
Own Id: OTP-13525</p>
</item>
</list>
</section>
</section>
<section><title>SSL 7.3.1</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Corrections to cipher suite handling using the 3 and 4
tuple format in addition to commit
89d7e21cf4ae988c57c8ef047bfe85127875c70c</p>
<p>
Own Id: OTP-13511</p>
</item>
</list>
</section>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Make values for the TLS-1.2 signature_algorithms
extension configurable</p>
<p>
Own Id: OTP-13261</p>
</item>
</list>
</section>
</section>
<section><title>SSL 7.3</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Make sure there is only one poller validator at a time
for validating the session cache.</p>
<p>
Own Id: OTP-13185</p>
</item>
<item>
<p>
A timing related issue could cause ssl to hang,
especially happened with newer versions of OpenSSL in
combination with ECC ciphers.</p>
<p>
Own Id: OTP-13253</p>
</item>
<item>
<p>
Work around a race condition in the TLS distribution
start.</p>
<p>
Own Id: OTP-13268</p>
</item>
<item>
<p>
Big handshake messages are now correctly fragmented in
the TLS record layer.</p>
<p>
Own Id: OTP-13306</p>
</item>
<item>
<p>
Improve portability of ECC tests in Crypto and SSL for
"exotic" OpenSSL versions.</p>
<p>
Own Id: OTP-13311</p>
</item>
<item>
<p>
Certificate extensions marked as critical are ignored
when using verify_none</p>
<p>
Own Id: OTP-13377</p>
</item>
<item>
<p>
If a certificate doesn't contain a CRL Distribution
Points extension, and the relevant CRL is not in the
cache, and the <c>crl_check</c> option is not set to
<c>best_effort</c> , the revocation check should fail.</p>
<p>
Own Id: OTP-13378</p>
</item>
<item>
<p>
Enable TLS distribution over IPv6</p>
<p>
Own Id: OTP-13391</p>
</item>
</list>
</section>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Improve error reporting for TLS distribution</p>
<p>
Own Id: OTP-13219</p>
</item>
<item>
<p>
Include options from connect, listen and accept in
<c>connection_information/1,2</c></p>
<p>
Own Id: OTP-13232</p>
</item>
<item>
<p>
Allow adding extra options for outgoing TLS distribution
connections, as supported for plain TCP connections.</p>
<p>
Own Id: OTP-13285</p>
</item>
<item>
<p>
Use loopback as server option in TLS-distribution module</p>
<p>
Own Id: OTP-13300</p>
</item>
<item>
<p>
Verify certificate signature against original certificate
binary.</p>
<p>
This avoids bugs due to encoding errors when re-encoding
a decode certificate. As there exists several decode step
and using of different ASN.1 specification this is a risk
worth avoiding.</p>
<p>
Own Id: OTP-13334</p>
</item>
<item>
<p>
Use <c>application:ensure_all_started/2</c> instead of
hard-coding dependencies</p>
<p>
Own Id: OTP-13363</p>
</item>
</list>
</section>
</section>
<section><title>SSL 7.2</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Honor distribution port range options</p>
<p>
Own Id: OTP-12838</p>
</item>
<item>
<p>
Correct supervisor specification in TLS distribution.</p>
<p>
Own Id: OTP-13134</p>
</item>
<item>
<p>
Correct cache timeout</p>
<p>
Own Id: OTP-13141</p>
</item>
<item>
<p>
Avoid crash and restart of ssl process when key file does
not exist.</p>
<p>
Own Id: OTP-13144</p>
</item>
<item>
<p>
Enable passing of raw socket options on the format
{raw,_,_,_} to the underlying socket.</p>
<p>
Own Id: OTP-13166</p>
</item>
<item>
<p>
Hibernation with small or a zero timeout will now work as
expected</p>
<p>
Own Id: OTP-13189</p>
</item>
</list>
</section>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Add upper limit for session cache, configurable on ssl
application level.</p>
<p>
If upper limit is reached, invalidate the current cache
entries, e.i the session lifetime is the max time a
session will be keept, but it may be invalidated earlier
if the max limit for the table is reached. This will keep
the ssl manager process well behaved, not exhusting
memeory. Invalidating the entries will incrementally
empty the cache to make room for fresh sessions entries.</p>
<p>
Own Id: OTP-12392</p>
</item>
<item>
<p>
Use new time functions to measure passed time.</p>
<p>
Own Id: OTP-12457</p>
</item>
<item>
<p>
Improved error handling in TLS distribution</p>
<p>
Own Id: OTP-13142</p>
</item>
<item>
<p>
Distribution over TLS now honors the nodelay distribution
flag</p>
<p>
Own Id: OTP-13143</p>
</item>
</list>
</section>
</section>
<section><title>SSL 7.1</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Add DER encoded ECPrivateKey as valid input format for
key option.</p>
<p>
Own Id: OTP-12974</p>
</item>
<item>
<p>
Correct return value of default session callback module</p>
<p>
This error had the symptom that the client check for
unique session would always fail, potentially making the
client session table grow a lot and causing long setup
times.</p>
<p>
Own Id: OTP-12980</p>
</item>
</list>
</section>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Add possibility to downgrade an SSL/TLS connection to a
tcp connection, and give back the socket control to a
user process.</p>
<p>
This also adds the possibility to specify a timeout to
the ssl:close function.</p>
<p>
Own Id: OTP-11397</p>
</item>
<item>
<p>
Add application setting to be able to change fatal alert
shutdown timeout, also shorten the default timeout. The
fatal alert timeout is the number of milliseconds between
sending of a fatal alert and closing the connection.
Waiting a little while improves the peers chances to
properly receiving the alert so it may shutdown
gracefully.</p>
<p>
Own Id: OTP-12832</p>
</item>
</list>
</section>
</section>
<section><title>SSL 7.0</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Ignore signature_algorithm (TLS 1.2 extension) sent to
TLS 1.0 or TLS 1.1 server</p>
<p>
Own Id: OTP-12670</p>
</item>
<item>
<p>
Improve error handling in TLS distribution module to
avoid lingering sockets.</p>
<p>
Own Id: OTP-12799 Aux Id: Tom Briden </p>
</item>
<item>
<p>
Add option {client_renegotiation, boolean()} option to
the server-side of the SSL application.</p>
<p>
Own Id: OTP-12815</p>
</item>
</list>
</section>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Add new API functions to handle CRL-verification</p>
<p>
Own Id: OTP-10362 Aux Id: kunagi-215 [126] </p>
</item>
<item>
<p>
Remove default support for SSL-3.0, due to Poodle
vunrability in protocol specification.</p>
<p>
Add padding check for TLS-1.0 to remove Poodle
vunrability from TLS 1.0, also add the option
padding_check. This option only affects TLS-1.0
connections and if set to false it disables the block
cipher padding check to be able to interoperate with
legacy software.</p>
<p>
Remove default support for RC4 cipher suites, as they are
consider too weak.</p>
<p>
*** POTENTIAL INCOMPATIBILITY ***</p>
<p>
Own Id: OTP-12390</p>
</item>
<item>
<p>
Add support for TLS ALPN (Application-Layer Protocol
Negotiation) extension.</p>
<p>
Own Id: OTP-12580</p>
</item>
<item>
<p>
Add SNI (Server Name Indication) support for the server
side.</p>
<p>
Own Id: OTP-12736</p>
</item>
</list>
</section>
</section>
<section><title>SSL 6.0.1.1</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Gracefully ignore proprietary hash_sign algorithms</p>
<p>
Own Id: OTP-12829</p>
</item>
</list>
</section>
</section>
<section><title>SSL 6.0.1</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Terminate gracefully when receving bad input to premaster
secret calculation</p>
<p>
Own Id: OTP-12783</p>
</item>
</list>
</section>
</section>
<section><title>SSL 6.0</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Exclude self-signed trusted anchor certificates from
certificate prospective certification path according to
RFC 3280.</p>
<p>
This will avoid some unnecessary certificate processing.</p>
<p>
Own Id: OTP-12449</p>
</item>
</list>
</section>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Separate client and server session cache internally.</p>
<p>
Avoid session table growth when client starts many
connections in such a manner that many connections are
started before session reuse is possible. Only save a new
session in client if there is no equivalent session
already stored.</p>
<p>
Own Id: OTP-11365</p>
</item>
<item>
<p>
The PEM cache is now validated by a background process,
instead of always keeping it if it is small enough and
clearing it otherwise. That strategy required that small
caches where cleared by API function if a file changes on
disk.</p>
<p>
However export the API function to clear the cache as it
may still be useful.</p>
<p>
Own Id: OTP-12391</p>
</item>
<item>
<p>
Add padding check for TLS-1.0 to remove Poodle
vulnerability from TLS 1.0, also add the option
padding_check. This option only affects TLS-1.0
connections and if set to false it disables the block
cipher padding check to be able to interoperate with
legacy software.</p>
<p>
*** POTENTIAL INCOMPATIBILITY ***</p>
<p>
Own Id: OTP-12420</p>
</item>
<item>
<p>
Add support for TLS_FALLBACK_SCSV used to prevent
undesired TLS version downgrades. If used by a client
that is vulnerable to the POODLE attack, and the server
also supports TLS_FALLBACK_SCSV, the attack can be
prevented.</p>
<p>
Own Id: OTP-12458</p>
</item>
</list>
</section>
</section>
<section><title>SSL 5.3.8</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Make sure the clean rule for ssh, ssl, eunit and otp_mibs
actually removes generated files.</p>
<p>
Own Id: OTP-12200</p>
</item>
</list>
</section>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Change code to reflect that state data may be secret to
avoid breaking dialyzer contracts.</p>
<p>
Own Id: OTP-12341</p>
</item>
</list>
</section>
</section>
<section><title>SSL 5.3.7</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Handle the fact that servers may send an empty SNI
extension to the client.</p>
<p>
Own Id: OTP-12198</p>
</item>
</list>
</section>
</section>
<section><title>SSL 5.3.6</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Corrected handling of ECC certificates, there where
several small issues with the handling of such
certificates in the ssl and public_key application. Now
ECC signed ECC certificates shall work and not only RSA
signed ECC certificates.</p>
<p>
Own Id: OTP-12026</p>
</item>
<item>
<p>
Check that the certificate chain ends with a trusted ROOT
CA e.i. a self-signed certificate, but provide an option
partial_chain to enable the application to define an
intermediat CA as trusted.</p>
<p>
Own Id: OTP-12149</p>
</item>
</list>
</section>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Add decode functions for SNI (Server Name Indication)</p>
<p>
Own Id: OTP-12048</p>
</item>
</list>
</section>
</section>
<section><title>SSL 5.3.5</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
ssl:recv now returns {error, einval} if applied to a non
passive socket, the same as gen_tcp:recv. </p>
<p>
Thanks to Danil Zagoskin for reporting this issue</p>
<p>
Own Id: OTP-11878</p>
</item>
<item>
<p>
Corrected handling of default values for
signature_algorithms extension in TLS-1.2 and
corresponding values used in previous versions that does
not support this extension. </p>
<p>
Thanks to Danil Zagoskin</p>
<p>
Own Id: OTP-11886</p>
</item>
<item>
<p>
Handle socket option inheritance when pooling of accept
sockets is used</p>
<p>
Own Id: OTP-11897</p>
</item>
<item>
<p>
Make sure that the list of versions, possibly supplied in
the versions option, is not order dependent.</p>
<p>
Thanks to Ransom Richardson for reporting this issue</p>
<p>
Own Id: OTP-11912</p>
</item>
<item>
<p>
Reject connection if the next_protocol message is sent
twice.</p>
<p>
Own Id: OTP-11926</p>
</item>
<item>
<p>
Correct options handling when ssl:ssl_accept/3 is called
with new ssl options after calling ssl:listen/2</p>
<p>
Own Id: OTP-11950</p>
</item>
</list>
</section>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Gracefully handle unknown alerts</p>
<p>
Thanks to Atul Atri for reporting this issue</p>
<p>
Own Id: OTP-11874</p>
</item>
<item>
<p>
Gracefully ignore cipher suites sent by client not
supported by the SSL/TLS version that the client has
negotiated.</p>
<p>
Thanks to Danil Zagoskin for reporting this issue</p>
<p>
Own Id: OTP-11875</p>
</item>
<item>
<p>
Gracefully handle structured garbage, i.e a client sends
some garbage in a ssl record instead of a valid fragment.</p>
<p>
Thanks to Danil Zagoskin</p>
<p>
Own Id: OTP-11880</p>
</item>
<item>
<p>
Gracefully handle invalid alerts</p>
<p>
Own Id: OTP-11890</p>
</item>
<item>
<p>
Generalize handling of default ciphers</p>
<p>
Thanks to Andreas Schultz</p>
<p>
Own Id: OTP-11966</p>
</item>
<item>
<p>
Make sure change cipher spec is correctly handled</p>
<p>
Own Id: OTP-11975</p>
</item>
</list>
</section>
</section>
<section><title>SSL 5.3.4</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Fix incorrect dialyzer spec and types, also enhance
documentation. </p>
<p>
Thanks to Ayaz Tuncer.</p>
<p>
Own Id: OTP-11627</p>
</item>
<item>
<p>
Fix possible mismatch between SSL/TLS version and default
ciphers. Could happen when you specified SSL/TLS-version
in optionlist to listen or accept.</p>
<p>
Own Id: OTP-11712</p>
</item>
<item>
<p>
Application upgrade (appup) files are corrected for the
following applications: </p>
<p>
<c>asn1, common_test, compiler, crypto, debugger,
dialyzer, edoc, eldap, erl_docgen, et, eunit, gs, hipe,
inets, observer, odbc, os_mon, otp_mibs, parsetools,
percept, public_key, reltool, runtime_tools, ssh,
syntax_tools, test_server, tools, typer, webtool, wx,
xmerl</c></p>
<p>
A new test utility for testing appup files is added to
test_server. This is now used by most applications in
OTP.</p>
<p>
(Thanks to Tobias Schlager)</p>
<p>
Own Id: OTP-11744</p>
</item>
</list>
</section>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Moved elliptic curve definition from the crypto
NIF/OpenSSL into Erlang code, adds the RFC-5639 brainpool
curves and makes TLS use them (RFC-7027).</p>
<p>
Thanks to Andreas Schultz</p>
<p>
Own Id: OTP-11578</p>
</item>
<item>
<p>
Unicode adaptations</p>
<p>
Own Id: OTP-11620</p>
</item>
<item>
<p>
Added option honor_cipher_order. This instructs the
server to prefer its own cipher ordering rather than the
client's and can help protect against things like BEAST
while maintaining compatability with clients which only
support older ciphers. </p>
<p>
Thanks to Andrew Thompson for the implementation, and
Andreas Schultz for the test cases.</p>
<p>
Own Id: OTP-11621</p>
</item>
<item>
<p>
Replace boolean checking in validate_option with
is_boolean guard. </p>
<p>
Thanks to Andreas Schultz.</p>
<p>
Own Id: OTP-11634</p>
</item>
<item>
<p>
Some function specs are corrected or moved and some edoc
comments are corrected in order to allow use of edoc.
(Thanks to Pierre Fenoll)</p>
<p>
Own Id: OTP-11702</p>
</item>
<item>
<p>
Correct clean up of certificate database when certs are
inputed in pure DER format.The incorrect code could cause
a memory leek when certs where inputed in DER. Thanks to
Bernard Duggan for reporting this.</p>
<p>
Own Id: OTP-11733</p>
</item>
<item>
<p>
Improved documentation of the cacertfile option</p>
<p>
Own Id: OTP-11759 Aux Id: seq12535 </p>
</item>
<item>
<p>
Avoid next protocol negotiation failure due to incorrect
option format.</p>
<p>
Own Id: OTP-11760</p>
</item>
<item>
<p>
Handle v1 CRLs, with no extensions and fixes issues with
IDP (Issuing Distribution Point) comparison during CRL
validation. </p>
<p>
Thanks to Andrew Thompson</p>
<p>
Own Id: OTP-11761</p>
</item>
<item>
<p>
Server now ignores client ECC curves that it does not
support instead of crashing. </p>
<p>
Thanks to Danil Zagoskin for reporting the issue and
suggesting a solution.</p>
<p>
Own Id: OTP-11780</p>
</item>
<item>
<p>
Handle SNI (Server Name Indication) alert
unrecognized_name and gracefully deal with unexpected
alerts. </p>
<p>
Thanks to Masatake Daimon for reporting this.</p>
<p>
Own Id: OTP-11815</p>
</item>
<item>
<p>
Add possibility to specify ssl options when calling
ssl:ssl_accept</p>
<p>
Own Id: OTP-11837</p>
</item>
</list>
</section>
</section>
<section><title>SSL 5.3.3</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Add missing validation of the server_name_indication
option and test for its explicit use. It was not possible
to set or disable the default server_name_indication as
the validation of the option was missing.</p>
<p>
Own Id: OTP-11567</p>
</item>
<item>
<p>
Elliptic curve selection in server mode now properly
selects a curve suggested by the client, if possible, and
the fallback alternative is changed to a more widely
supported curve.</p>
<p>
Own Id: OTP-11575</p>
</item>
<item>
<p>
Bug in the TLS hello extension handling caused the server
to behave as it did not understand secure renegotiation.</p>
<p>
Own Id: OTP-11595</p>
</item>
</list>
</section>
</section>
<section><title>SSL 5.3.2</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Honors the clients advertised support of elliptic curves
and no longer sends incorrect elliptic curve extension in
server hello.</p>
<p>
Own Id: OTP-11370</p>
</item>
<item>
<p>
Fix initialization of DTLS fragment reassembler, in
previously contributed code, for future support of DTLS .
Thanks to Andreas Schultz.</p>
<p>
Own Id: OTP-11376</p>
</item>
<item>
<p>
Corrected type error in client_preferred_next_protocols
documentation. Thanks to Julien Barbot.</p>
<p>
Own Id: OTP-11457</p>
</item>
</list>
</section>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
TLS code has been refactored to prepare for future DTLS
support. Also some DTLS code is in place but not yet
runnable, some of it contributed by Andreas Schultz and
some of it written by the OTP team. Thanks to to Andreas
for his participation.</p>
<p>
Own Id: OTP-11292</p>
</item>
<item>
<p>
Remove extraneous dev debug code left in the close
function. Thanks to Ken Key.</p>
<p>
Own Id: OTP-11447</p>
</item>
<item>
<p>
Add SSL Server Name Indication (SNI) client support.
Thanks to Julien Barbot.</p>
<p>
Own Id: OTP-11460</p>
</item>
</list>
</section>
</section>
<section><title>SSL 5.3.1</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Setopts during renegotiation caused the renegotiation to
be unsuccessful.</p>
<p>
If calling setopts during a renegotiation the FSM state
might change during the handling of the setopts messages,
this is now handled correctly.</p>
<p>
Own Id: OTP-11228</p>
</item>
<item>
<p>
Now handles signature_algorithm field in digitally_signed
properly with proper defaults. Prior to this change some
elliptic curve cipher suites could fail reporting the
error "bad certificate".</p>
<p>
Own Id: OTP-11229</p>
</item>
<item>
<p>
The code emulating the inet header option was changed in
the belief that it made it inet compatible. However the
testing is a bit hairy as the inet option is actually
broken, now the tests are corrected and the header option
should work in the same broken way as inet again,
preferably use the bitsyntax instead.</p>
<p>
Own Id: OTP-11230</p>
</item>
</list>
</section>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Make the ssl manager name for erlang distribution over
SSL/TLS relative to the module name of the ssl_manager.</p>
<p>
This can be beneficial when making tools that rename
modules for internal processing in the tool.</p>
<p>
Own Id: OTP-11255</p>
</item>
<item>
<p>
Add documentation regarding log_alert option.</p>
<p>
Own Id: OTP-11271</p>
</item>
</list>
</section>
</section>
<section><title>SSL 5.3</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Honor the versions option to ssl:connect and ssl:listen.</p>
<p>
Own Id: OTP-10905</p>
</item>
<item>
<p>
Next protocol negotiation with reused sessions will now
succeed</p>
<p>
Own Id: OTP-10909</p>
</item>
</list>
</section>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Add support for PSK (Pre Shared Key) and SRP (Secure
Remote Password) chipher suits, thanks to Andreas
Schultz.</p>
<p>
Own Id: OTP-10450 Aux Id: kunagi-269 [180] </p>
</item>
<item>
<p>
Fix SSL Next Protocol Negotiation documentation. Thanks
to Julien Barbot.</p>
<p>
Own Id: OTP-10955</p>
</item>
<item>
<p>
Fix ssl_connection to support reading proxy/chain
certificates. Thanks to Valentin Kuznetsov.</p>
<p>
Own Id: OTP-10980</p>
</item>
<item>
<p>
Integrate elliptic curve contribution from Andreas
Schultz </p>
<p>
In order to be able to support elliptic curve cipher
suites in SSL/TLS, additions to handle elliptic curve
infrastructure has been added to public_key and crypto.</p>
<p>
This also has resulted in a rewrite of the crypto API to
gain consistency and remove unnecessary overhead. All OTP
applications using crypto has been updated to use the new
API.</p>
<p>
Impact: Elliptic curve cryptography (ECC) offers
equivalent security with smaller key sizes than other
public key algorithms. Smaller key sizes result in
savings for power, memory, bandwidth, and computational
cost that make ECC especially attractive for constrained
environments.</p>
<p>
Own Id: OTP-11009</p>
</item>
</list>
</section>
</section>
<section><title>SSL 5.2.1</title>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Transport callback handling is changed so that gen_tcp is
treated as a special case where inet will be called
directly for functions such as setopts, as gen_tcp does
not have its own setopts. This will enable users to use
the transport callback for other customizations such as
websockets.</p>
<p>
Own Id: OTP-10847</p>
</item>
<item>
<p>
Follow up to OTP-10451 solved in ssl-5.2 R16A. Make sure
format_error return good strings. Replace confusing
legacy atoms with more descriptive atoms.</p>
<p>
Own Id: OTP-10864</p>
</item>
</list>
</section>
</section>
<section><title>SSL 5.1.2.1</title>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Make log_alert configurable as option in ssl, SSLLogLevel
added as option to inets conf file</p>
<p>
Own Id: OTP-11259</p>
</item>
</list>
</section>
</section>
<section><title>SSL 5.2</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
SSL: TLS 1.2, advertise sha224 support, thanks to Andreas
Schultz.</p>
<p>
Own Id: OTP-10586</p>
</item>
<item>
<p>
If an ssl server is restarted with new options and a
client tries to reuse a session the server must make sure
that it complies to the new options before agreeing to
reuse it.</p>
<p>
Own Id: OTP-10595</p>
</item>
<item>
<p>
Now handles cleaning of CA-certificate database correctly
so that there will be no memory leek, bug was introduced
in ssl- 5.1 when changing implementation to increase
parallel execution.</p>
<p>
Impact: Improved memory usage, especially if you have
many different certificates and upgrade tcp-connections
to TLS-connections.</p>
<p>
Own Id: OTP-10710</p>
</item>
</list>
</section>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Support Next Protocol Negotiation in TLS, thanks to Ben
Murphy for the contribution.</p>
<p>
Impact: Could give performance benefit if used as it
saves a round trip.</p>
<p>
Own Id: OTP-10361 Aux Id: kunagi-214 [125] </p>
</item>
<item>
<p>
TLS 1.2 will now be the default TLS version if sufficient
crypto support is available otherwise TLS 1.1 will be
default.</p>
<p>
Impact: A default TLS connection will have higher
security and hence it may be perceived as slower then
before.</p>
<p>
Own Id: OTP-10425 Aux Id: kunagi-275 [186] </p>
</item>
<item>
<p>
It is now possible to call controlling_process on a
listen socket, same as in gen_tcp.</p>
<p>
Own Id: OTP-10447</p>
</item>
<item>
<p>
Remove filter mechanisms that made error messages
backwards compatible with old ssl but hid information
about what actually happened.</p>
<p>
This does not break the documented API however other
reason terms may be returned, so code that matches on the
reason part of {error, Reason} may fail.</p>
<p>
*** POTENTIAL INCOMPATIBILITY ***</p>
<p>
Own Id: OTP-10451 Aux Id: kunagi-270 [181] </p>
</item>
<item>
<p>
Added missing dependencies to Makefile</p>
<p>
Own Id: OTP-10594</p>
</item>
<item>
<p>
Removed deprecated function ssl:pid/0, it has been
pointless since R14 but has been keep for backwards
compatibility.</p>
<p>
*** POTENTIAL INCOMPATIBILITY ***</p>
<p>
Own Id: OTP-10613 Aux Id: kunagi-331 [242] </p>
</item>
<item>
<p>
Refactor to simplify addition of key exchange methods,
thanks to Andreas Schultz.</p>
<p>
Own Id: OTP-10709</p>
</item>
</list>
</section>
</section>
<section><title>SSL 5.1.2</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
ssl:ssl_accept/2 timeout is no longer ignored</p>
<p>
Own Id: OTP-10600</p>
</item>
</list>
</section>
</section>
<section><title>SSL 5.1.1</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
ssl:recv/3 could "loose" data when the timeout occurs. If
the timout in ssl:connect or ssl:ssl_accept expired the
ssl connection process was not terminated as it should,
this due to gen_fsm:send_all_state_event timout is a
client side time out. These timouts are now handled by
the gen_fsm-procss instead.</p>
<p>
Own Id: OTP-10569</p>
</item>
</list>
</section>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Better termination handling that avoids hanging.</p>
<p>
Own Id: OTP-10574</p>
</item>
</list>
</section>
</section>
<section><title>SSL 5.1</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Sometimes the client process could receive an extra
{error, closed} message after ssl:recv had returned
{error, closed}.</p>
<p>
Own Id: OTP-10118</p>
</item>
<item>
<p>
ssl v3 alert number 41 (no_certificate_RESERVED) is now
recognized</p>
<p>
Own Id: OTP-10196</p>
</item>
</list>
</section>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Experimental support for TLS 1.1 is now available, will
be officially supported from OTP-R16. Thanks to Andreas
Schultz for implementing the first version.</p>
<p>
Own Id: OTP-8871</p>
</item>
<item>
<p>
Experimental support for TLS 1.2 is now available, will
be officially supported from OTP-R16. Thanks to Andreas
Schultz for implementing the first version.</p>
<p>
Own Id: OTP-8872</p>
</item>
<item>
<p>
Removed some bottlenecks increasing the applications
parallelism especially for the client side.</p>
<p>
Own Id: OTP-10113</p>
</item>
<item>
<p>
Workaround for handling certificates that wrongly encode
X509countryname in utf-8 when the actual value is a valid
ASCCI value of length 2. Such certificates are accepted
by many browsers such as Chrome and Fierfox so for
interoperability reasons we will too.</p>
<p>
Own Id: OTP-10222</p>
</item>
</list>
</section>
</section>
<section><title>SSL 5.0.1</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Robustness and improvement to distribution over SSL</p>
<p>
Fix a bug where ssl_tls_dist_proxy would crash at caller
timeout. Fix a bug where a timeout from the SSL layer
would block the distribution indefinately. Run the proxy
exclusively on the loopback interface. (Thanks to Paul
Guyot)</p>
<p>
Own Id: OTP-9915</p>
</item>
<item>
<p>
Fix setup loop of SSL TLS dist proxy</p>
<p>
Fix potential leak of processes waiting indefinately for
data from closed sockets during socket setup phase.
(Thanks to Paul Guyot)</p>
<p>
Own Id: OTP-9916</p>
</item>
<item>
<p>
Correct spelling of registered (Thanks to Richard
Carlsson)</p>
<p>
Own Id: OTP-9925</p>
</item>
<item>
<p>
Added TLS PRF function to the SSL API for generation of
additional key material from a TLS session. (Thanks to
Andreas Schultz)</p>
<p>
Own Id: OTP-10024</p>
</item>
</list>
</section>
</section>
<section><title>SSL 5.0</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Invalidation handling of sessions could cause the
time_stamp field in the session record to be set to
undefined crashing the session clean up process. This did
not affect the connections but would result in that the
session table would grow.</p>
<p>
Own Id: OTP-9696 Aux Id: seq11947 </p>
</item>
<item>
<p>
Changed code to use ets:foldl and throw instead of
ets:next traversal, avoiding the need to explicitly call
ets:safe_fixtable. It was possible to get a badarg-crash
under special circumstances.</p>
<p>
Own Id: OTP-9703 Aux Id: seq11947 </p>
</item>
<item>
<p>
Send ssl_closed notification to active ssl user when a
tcp error occurs.</p>
<p>
Own Id: OTP-9734 Aux Id: seq11946 </p>
</item>
<item>
<p>
If a passive receive was ongoing during a renegotiation
the process evaluating ssl:recv could be left hanging for
ever.</p>
<p>
Own Id: OTP-9744</p>
</item>
</list>
</section>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Support for the old ssl implementation is dropped and the
code is removed.</p>
<p>
Own Id: OTP-7048</p>
</item>
<item>
<p>
The erlang distribution can now be run over the new ssl
implementation. All options can currently not be set but
it is enough to replace to old ssl implementation.</p>
<p>
Own Id: OTP-7053</p>
</item>
<item>
<p>
public_key, ssl and crypto now supports PKCS-8</p>
<p>
Own Id: OTP-9312</p>
</item>
<item>
<p>
Implements a CBC timing attack counter measure. Thanks to
Andreas Schultz for providing the patch.</p>
<p>
Own Id: OTP-9683</p>
</item>
<item>
<p>
Mitigates an SSL/TLS Computational DoS attack by
disallowing the client to renegotiate many times in a row
in a short time interval, thanks to Tuncer Ayaz for
alerting us about this.</p>
<p>
Own Id: OTP-9739</p>
</item>
<item>
<p>
Implements the 1/n-1 splitting countermeasure to the
Rizzo Duong BEAST attack, affects SSL 3.0 and TLS 1.0.
Thanks to Tuncer Ayaz for alerting us about this.</p>
<p>
Own Id: OTP-9750</p>
</item>
</list>
</section>
</section>
<section><title>SSL 4.1.6</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
replace "a ssl" with "an ssl" reindent
pkix_path_validation/3 Trivial documentation fixes
(Thanks to Christian von Roques )</p>
<p>
Own Id: OTP-9464</p>
</item>
</list>
</section>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Adds function clause to avoid denial of service attack.
Thanks to Vinod for reporting this vulnerability.</p>
<p>
Own Id: OTP-9364</p>
</item>
<item>
<p>
Error handling code now takes care of inet:getopts/2 and
inets:setopts/2 crashes. Thanks to Richard Jones for
reporting this.</p>
<p>
Own Id: OTP-9382</p>
</item>
<item>
<p>
Support explicit use of packet option httph and httph_bin</p>
<p>
Own Id: OTP-9461</p>
</item>
<item>
<p>
Decoding of hello extensions could fail to come to the
correct conclusion due to an error in a binary match
pattern. Thanks to Ben Murphy.</p>
<p>
Own Id: OTP-9589</p>
</item>
</list>
</section>
</section>
<section>
<title>SSL 4.1.5</title>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>Calling gen_tcp:connect with option {ip, {127,0,0,1}} results in
an exit with reason badarg. Neither SSL nor INETS This was not
catched, resulting in crashes with incomprehensible reasons.</p>
<p>Own Id: OTP-9289 Aux Id: seq11845</p>
</item>
</list>
</section>
</section>
<section>
<title>SSL 4.1.3</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Fixed error in cache-handling fix from ssl-4.1.2</p>
<p>
Own Id: OTP-9018 Aux Id: seq11739 </p>
</item>
<item>
<p>Verification of a critical extended_key_usage-extension
corrected</p>
<p>Own Id: OTP-9029 Aux Id: seq11541 </p>
</item>
</list>
</section>
</section>
<section>
<title>SSL 4.1.2</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
The ssl application caches certificate files, it will now
invalidate cache entries if the diskfile is changed.</p>
<p>
Own Id: OTP-8965 Aux Id: seq11739 </p>
</item>
<item>
<p>
Now runs the terminate function before returning from the
call made by ssl:close/1, as before the caller of
ssl:close/1 could get problems with the reuseaddr option.</p>
<p>
Own Id: OTP-8992</p>
</item>
</list>
</section>
</section>
<section><title>SSL 4.1.1</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Correct handling of client certificate verify message
When checking the client certificate verify message the
server used the wrong algorithm identifier to determine
the signing algorithm, causing a function clause error in
the public_key application when the key-exchange
algorithm and the public key algorithm of the client
certificate happen to differ.</p>
<p>
Own Id: OTP-8897</p>
</item>
</list>
</section>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
For testing purposes ssl now also support some anonymous
cipher suites when explicitly configured to do so.</p>
<p>
Own Id: OTP-8870</p>
</item>
<item>
<p>
Sends an error alert instead of crashing if a crypto
function for the selected cipher suite fails.</p>
<p>
Own Id: OTP-8930 Aux Id: seq11720 </p>
</item>
</list>
</section>
</section>
<section><title>SSL 4.1</title>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Updated ssl to ignore CA certs that violate the asn1-spec
for a certificate, and updated public key asn1 spec to
handle inherited DSS-params.</p>
<p>
Own Id: OTP-7884</p>
</item>
<item>
<p>
Changed ssl implementation to retain backwards
compatibility for old option {verify, 0} that shall be
equivalent to {verify, verify_none}, also separate the
cases unknown ca and selfsigned peer cert, and restored
return value of deprecated function
public_key:pem_to_der/1.</p>
<p>
Own Id: OTP-8858</p>
</item>
<item>
<p>
Changed the verify fun so that it differentiate between
the peer certificate and CA certificates by using
valid_peer or valid as the second argument to the verify
fun. It may not always be trivial or even possible to
know when the peer certificate is reached otherwise.</p>
<p>
*** POTENTIAL INCOMPATIBILITY ***</p>
<p>
Own Id: OTP-8873</p>
</item>
</list>
</section>
</section>
<section><title>SSL 4.0.1</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
The server now verifies the client certificate verify
message correctly, instead of causing a case-clause.</p>
<p>
Own Id: OTP-8721</p>
</item>
<item>
<p>
The client hello message now always include ALL available
cipher suites (or those specified by the ciphers option).
Previous implementation would filter them based on the
client certificate key usage extension (such filtering
only makes sense for the server certificate).</p>
<p>
Own Id: OTP-8772</p>
</item>
<item>
<p>
Fixed handling of the option {mode, list} that was broken
for some packet types for instance line.</p>
<p>
Own Id: OTP-8785</p>
</item>
<item>
<p>
Empty packets were not delivered to the client.</p>
<p>
Own Id: OTP-8790</p>
</item>
<item>
<p> Building in a source tree without prebuilt platform
independent build results failed on the SSL examples
when: </p> <list><item> cross building. This has been
solved by not building the SSL examples during a cross
build. </item><item> building on Windows. </item></list>
<p>
Own Id: OTP-8791</p>
</item>
<item>
<p>
Fixed a handshake error which occurred on some ssl
implementations.</p>
<p>
Own Id: OTP-8793</p>
</item>
</list>
</section>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Revise the public_key API - Cleaned up and documented the
public_key API to make it useful for general use, also
changed ssl to use the new API.</p>
<p>
Own Id: OTP-8722</p>
</item>
<item>
<p>
Added support for inputing certificates and keys directly
in DER format these options will override the pem-file
options if specified.</p>
<p>
Own Id: OTP-8723</p>
</item>
<item>
<p>
To gain interoperability ssl will not check for padding
errors when using TLS 1.0. It is first in TLS 1.1 that
checking the padding is an requirement.</p>
<p>
Own Id: OTP-8740</p>
</item>
<item>
<p>
Changed the semantics of the verify_fun option in the
ssl-application so that it takes care of both application
handling of path validation errors and verification of
application specific extensions. This means that it is
now possible for the server application in verify_peer
mode to handle path validation errors. This change moved
some functionality earlier in ssl to the public_key
application.</p>
<p>
Own Id: OTP-8770</p>
</item>
<item>
<p>
Added the functionality so that the verification fun will
be called when a certificate is considered valid by the
path validation to allow access to each certificate in
the path to the user application. Also try to verify
subject-AltName, if unable to verify it let the
application verify it.</p>
<p>
Own Id: OTP-8825</p>
</item>
</list>
</section>
</section>
<section><title>SSL 4.0</title>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
New ssl now support client/server-certificates signed by
dsa keys.</p>
<p>
Own Id: OTP-8587</p>
</item>
<item>
<p>
Ssl has now switched default implementation and removed
deprecated certificate handling. All certificate handling
is done by the public_key application.</p>
<p>
Own Id: OTP-8695</p>
</item>
</list>
</section>
</section>
</chapter>