<?xml version="1.0" encoding="utf-8" ?>
<!DOCTYPE chapter SYSTEM "chapter.dtd" [
<!ENTITY % also SYSTEM "seealso.ent" >
<!ENTITY % here SYSTEM "seehere.ent" >
%also;
%here;
]>
<chapter>
<header>
<copyright>
<year>2011</year>
<year>2017</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>Release Notes</title>
<prepared></prepared>
<docno></docno>
<date></date>
<rev></rev>
<file>notes.xml</file>
</header>
<p>
Releases are listed in reverse chronological order, most recent
first.</p>
<!-- ===================================================================== -->
<section><title>diameter 2.1.2</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
A fault introduced in diameter 2.1 could cause decode
errors to be ignored in AVPs following the header of a
Grouped AVP.</p>
<p>
Own Id: OTP-14684 Aux Id: ERIERL-85 </p>
</item>
</list>
</section>
</section>
<section><title>diameter 2.1.1</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
An inadvertently removed monitor in diameter 2.1 caused
the ets table diameter_reg to leak entries, and caused
service restart and more to fail.</p>
<p>
Own Id: OTP-14668 Aux Id: ERIERL-83 </p>
</item>
</list>
</section>
</section>
<section><title>diameter 2.1</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Fix handling of Proxy-Info in answer messages setting the
E-bit.</p>
<p>
RFC 6733 requires that Proxy-Info AVPs in an incoming
request be echoed in an outgoing answer. This was not
done in answers formulated by diameter; for example, as a
result of a handle_request callback having returned an
'answer-message' or protocol_error tuple.</p>
<p>
Own Id: OTP-9869</p>
</item>
<item>
<p>
React to nodeup/nodedown when sharing peer connections.</p>
<p>
Service configuration share_peers and use_shared_peers
did not respond to the coming and going of remote nodes.</p>
<p>
Own Id: OTP-14011</p>
</item>
<item>
<p>
Fix inappropriate message callbacks.</p>
<p>
An incoming CER or DPR was regarded as discarded,
resulting in a corresponding message callback (if
configured) in diameter_tcp/sctp.</p>
<p>
Own Id: OTP-14486</p>
</item>
<item>
<p>
Fix handling of 5009 errors (DIAMETER_AVP_OCCURS_TOO_MANY
TIMES).</p>
<p>
RFC 6733 says that the first AVP that exceeds the bound
should be reported, but the suggestions in the errors
field of a diameter_packet record counted AVPs from the
rear of the message, not the front. Additionally,
diameter 2.0 in OTP 20.0 broke the counting by accepting
one more AVP than the message grammar in question
allowed.</p>
<p>
Own Id: OTP-14512</p>
</item>
<item>
<p>
Match case insensitively in diameter_tcp/sctp accept
tuple.</p>
<p>
Matching of remote addresses when accepting connections
in a listening transport was case-sensitive, causing the
semantics to change as a consequence of (kernel)
OTP-13006.</p>
<p>
Own Id: OTP-14535 Aux Id: OTP-13006 </p>
</item>
<item>
<p>
Fix backwards incompatibility of remote send when sharing
transports.</p>
<p>
The sending of requests over a transport connection on a
remote node running an older version of diameter was
broken by diameter 2.0 in OTP 20.0.</p>
<p>
Own Id: OTP-14552</p>
</item>
<item>
<p>
Fix diameter_packet.avps decode of Grouped AVP errors in
Failed-AVP.</p>
<p>
Decode didn't produce a list of diameter_avp records, so
information about faulty component AVPs was lost.</p>
<p>
Own Id: OTP-14607</p>
</item>
</list>
</section>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Let unordered delivery be configured in diameter_sctp.</p>
<p>
With option {unordered, boolean() | pos_integer()}, with
false the default, and N equivalent to OS =< N, where
OS is the number of outbound streams negotiated on the
association in question. If configured, unordered sending
commences upon reception of a second message, outgoing
messages being sent on stream 0 before this.</p>
<p>
The default false is for backwards compatibility, but
false or 1 should be set to follow RFC 6733's
recommendation on the use of unordered sending to avoid
head-of-line blocking. There is typically no meaningful
order to preserve, since the order in which outgoing
messages are received by a transport process isn't known
to the sender.</p>
<p>
Own Id: OTP-10889</p>
</item>
<item>
<p>
Complete/simplify Standards Compliance in User's Guide.</p>
<p>
Own Id: OTP-10927</p>
</item>
<item>
<p>
Add service option decode_format.</p>
<p>
To allow incoming messages to be decoded into maps or
lists instead of records. Messages can be presented in
any of the formats for encode.</p>
<p>
Decode performance has also been improved.</p>
<p>
Own Id: OTP-14511 Aux Id: OTP-14343 </p>
</item>
<item>
<p>
Add service option traffic_counters.</p>
<p>
To let message-related counters be disabled, which can be
a performance improvement in some usecases.</p>
<p>
Own Id: OTP-14521</p>
</item>
<item>
<p>
Allow loopback/any as local addresses in
diameter_tcp/sctp.</p>
<p>
The atoms were implied by documentation, but not handled
in code.</p>
<p>
Own Id: OTP-14544</p>
</item>
<item>
<p>
Add transport option strict_capx.</p>
<p>
To allow the RFC 6733 requirement that a transport
connection be closed if a message is received before
capabilities exchange to be relaxed.</p>
<p>
Own Id: OTP-14546</p>
</item>
<item>
<p>
Be consistent with service/transport configuration.</p>
<p>
For options for which it's meaningful, defaults values
for transport options can now be configured on a service.
This was previously the case only for an arbitrary subset
of options.</p>
<p>
Own Id: OTP-14555</p>
</item>
<item>
<p>
Add service/transport option avp_dictionaries.</p>
<p>
To provide better support for AVPs that are not defined
in the application dictionary: configuring additional
dictionaries in an avp_dictionaries tuple allows their
AVPs to be encoded/decoded in much the same fashion as
application AVPs.</p>
<p>
The motivation is RFC 7683 Diameter Overload, Indicator
Conveyance (DOIC), that defines AVPs intended to be
piggybacked onto arbitrary messages. A DOIC dictionary
has been included in the installation, in module
diameter_gen_doic_rfc7683.</p>
<p>
Own Id: OTP-14588</p>
</item>
<item>
<p>
Decode application AVPs in answers setting the E-bit.</p>
<p>
AVPs defined in the application of the message being sent
were previously not decoded, only those in the common
application that defines the answer-message grammar.</p>
<p>
Own Id: OTP-14596</p>
</item>
</list>
</section>
</section>
<section><title>diameter 2.0</title>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Let candidate peers be passed to diameter:call/4</p>
<p>
With call option peer, to allow a request to be sent to a
peer that hasn't advertised support for the application
in question.</p>
<p>
RFC 6733 2.4 requires a node to send the application
identifiers of all locally supported applications at
capabilities exchange, but not all nodes respect this for
the common application, and diameter itself will send
D[WP][RA] without the common application having been
explicitly advertised. Regarding the common application
as implicit renders Result-Code 5010
(DIAMETER_NO_COMMON_APPLICATION) meaningless however, so
allow any request to be sent as long as there is a
configured dictionary to support it.</p>
<p>
Own Id: OTP-14338</p>
</item>
<item>
<p>
Improve performance of message encode/decode and related
handling.</p>
<p>
Dictionaries using @custom_types or @codecs will need to
adapt the corresponding functions to accept an additional
argument that is now passed through encode/decode, which
was required to remove various process dictionary-based
workarounds that have been used to solve problems in the
past.</p>
<p>
*** POTENTIAL INCOMPATIBILITY ***</p>
<p>
Own Id: OTP-14343</p>
</item>
<item>
<p>
Add transport options to avoid deadlock and allow for
load regulation.</p>
<p>
Both diameter_tcp and diameter_sctp now accept two new
configuration options: sender and message_cb. The former
causes outgoing sends to take place in a dedicated
process, to avoid the possibility of deadlock when both
the transport process and its peer block in send. The
latter allows a callback to control the reading of
messages on the socket, to allow for backpressure towards
peers when the rate of incoming traffic is greater than
can otherwise be handled.</p>
<p>
Neither of these options are yet documented, but are
unlikely to change unless problems are discovered. The
sender option is not the default since it should probably
always be used in combination with message_cb, to prevent
incoming requests from being read at a higher rate than a
peer allows outgoing answers to be sent.</p>
<p>
Own Id: OTP-14455 Aux Id: ERL-332 </p>
</item>
</list>
</section>
</section>
<section><title>diameter 1.12.2</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
An improvement in the handling of peer failover in
diameter 1.12.1 adversely affected performance when
sending requests. Further, the inefficient use of a
public table to route incoming answers has been removed.</p>
<p>
Own Id: OTP-14206</p>
</item>
<item>
<p>
Fixed xml issues in old release notes</p>
<p>
Own Id: OTP-14269</p>
</item>
</list>
</section>
</section>
<section><title>diameter 1.12.1</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Close diameter_tcp/sctp listening sockets at
diameter:stop_service/1.</p>
<p>
Broken by OTP-13611.</p>
<p>
Own Id: OTP-13787 Aux Id: OTP-13611 </p>
</item>
<item>
<p>
Update build scripts to not make assumtions about where
env, cp and perl are located.</p>
<p>
Own Id: OTP-13800</p>
</item>
</list>
</section>
</section>
<section><title>diameter 1.12</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Ensure listening socket is closed at transport removal.</p>
<p>
Transport removal did not immediately close a
<c>diameter_tcp/sctp</c> listening socket, and a
subsequent peer connection caused it to remain open.</p>
<p>
Own Id: OTP-13611</p>
</item>
</list>
</section>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Add <c>diameter:peer_info/1</c>.</p>
<p>
That retrieves information in the style of
<c>diameter:service_info/2</c>, but for a single peer
connection.</p>
<p>
Own Id: OTP-13508</p>
</item>
</list>
</section>
</section>
<section><title>diameter 1.11.2</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Make peer handling more efficient.</p>
<p>
Inefficient lookup and manipulation of peer lists could
result in poor performance when many outgoing requests
were sent simultaneously, or when many peers connected
simultaneously. Filtering peer lists on realm/host is now
also more efficient in many cases.</p>
<p>
Own Id: OTP-13164</p>
</item>
<item>
<p>
Fix handling of shared peer connections in watchdog state
SUSPECT.</p>
<p>
A peer connection shared from a remote node was regarded
as being up for the lifetime of the connection, ignoring
watchdog transitions into state SUSPECT.</p>
<p>
Own Id: OTP-13342</p>
</item>
</list>
</section>
</section>
<section><title>diameter 1.11.1</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Fix request table leaks</p>
<p>
The End-to-End and Hop-by-Hop identifiers of outgoing
Diameter requests are stored in a table in order for the
caller to be located when the corresponding answer
message is received. Entries were orphaned if the handler
was terminated by an exit signal as a consequence of
actions taken by callback functions, or if callbacks
modified identifiers in retransmission cases.</p>
<p>
Own Id: OTP-13137</p>
</item>
</list>
</section>
</section>
<section><title>diameter 1.11</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Fix relay encode of nested, Grouped AVPs.</p>
<p>
A fault in OTP-12475 caused encode to fail if the first
AVP in a Grouped AVP was itself Grouped.</p>
<p>
Own Id: OTP-12879 Aux Id: OTP-12475 </p>
</item>
<item>
<p>
Match acceptable peer addresses case insensitively.</p>
<p>
Regular expressions passed in an 'accept' tuple to
diameter_tcp or diameter_sctp inappropriately matched
case.</p>
<p>
Own Id: OTP-12902</p>
</item>
<item>
<p>
Fix diameter_watchdog function clause.</p>
<p>
OTP-12912 introduced an error with accepting transports
setting <c>{restrict_connections, false}</c>, causing
processes to fail when peer connections were terminated.</p>
<p>
Own Id: OTP-12969</p>
</item>
</list>
</section>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Don't report 5005 (DIAMETER_AVP_MISSING) errors
unnecessarily.</p>
<p>
An AVP whose decode failed was reported as missing,
despite having been reported with another error as a
consequence of the failure.</p>
<p>
Own Id: OTP-12871</p>
</item>
<item>
<p>
Improve decode performance.</p>
<p>
The time required to decode a message increased
quadratically with the number of AVPs in the worst case,
leading to extremely long execution times.</p>
<p>
Own Id: OTP-12891</p>
</item>
<item>
<p>
Improve watchdog and statistics performance.</p>
<p>
Inefficient use of timers contributed to poor performance
at high load, as did ordering of the table statistics are
written to.</p>
<p>
Own Id: OTP-12912</p>
</item>
<item>
<p>
Add service_opt() strict_mbit.</p>
<p>
There are differing opinions on whether or not reception
of an arbitrary AVP setting the M-bit is an error. The
default interpretation is strict: if a command grammar
doesn't explicitly allow an AVP setting the M-bit then
reception of such an AVP is regarded as an error. Setting
<c>{strict_mbit, false}</c> disables this check.</p>
<p>
Own Id: OTP-12947</p>
</item>
</list>
</section>
</section>
<section><title>diameter 1.10</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Fix decode of Grouped AVPs containing errors.</p>
<p>
RFC 6733 says this of Failed-AVP in 7.5:</p>
<taglist><tag></tag><item><p><c> In the case where the offending AVP
is embedded within a Grouped AVP, the Failed-AVP MAY
contain the grouped AVP, which in turn contains the
single offending AVP. The same method MAY be employed if
the grouped AVP itself is embedded in yet another grouped
AVP and so on. In this case, the Failed-AVP MAY contain
the grouped AVP hierarchy up to the single offending AVP.
This enables the recipient to detect the location of the
offending AVP when embedded in a
group.</c></p></item></taglist>
<p>
It says this of DIAMETER_INVALID_AVP_LENGTH in 7.1.5:</p>
<taglist><tag></tag><item><p><c> The request contained an AVP with
an invalid length. A Diameter message indicating this
error MUST include the offending AVPs within a Failed-AVP
AVP. In cases where the erroneous AVP length value
exceeds the message length or is less than the minimum
AVP header length, it is sufficient to include the
offending AVP header and a zero filled payload of the
minimum required length for the payloads data type. If
the AVP is a Grouped AVP, the Grouped AVP header with an
empty payload would be sufficient to indicate the
offending AVP. In the case where the offending AVP header
cannot be fully decoded when the AVP length is less than
the minimum AVP header length, it is sufficient to
include an offending AVP header that is formulated by
padding the incomplete AVP header with zero up to the
minimum AVP header length.</c></p></item></taglist>
<p>
The AVPs placed in the errors field of a diameter_packet
record are intended to be appropriate for inclusion in a
Failed-AVP, but neither of the above paragraphs has been
followed in the Grouped case: the entire faulty AVP
(non-faulty components and all) has been included. This
made it difficult to identify the actual faulty AVP in
all but simple cases.</p>
<p>
The decode is now adapted to the RFC, and implements the
suggested single faulty AVP, nested in as many Grouped
containers as required.</p>
<p>
Own Id: OTP-12721</p>
</item>
<item>
<p>
Fix SCTP problems on Solaris.</p>
<p>
The allocation of association ids in Solaris was in
conflict with an assumption made in diameter_sctp,
resulting in failures when accepting multiple peer
connections.</p>
<p>
Own Id: OTP-12768</p>
</item>
<item>
<p>
Fix start order of alternate transports.</p>
<p>
A transport configured with diameter:add_transport/2 can
be passed multiple transport_module/transport_config
tuples in order to specify alternate configuration,
modules being attempted in order until one succeeds. This
is primarily for the connecting case; for example, to
allow a transport to be configured to first attempt
connection over SCTP, and then TCP in case SCTP fails.
Multiple module tuples can be paired with a single config
tuple, but in this case the start order was reversed
relative to the order in which the modules were specifed.</p>
<p>
Own Id: OTP-12851</p>
</item>
</list>
</section>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Change license text from Erlang Public License to Apache
Public License v2.</p>
<p>
Own Id: OTP-12845</p>
</item>
</list>
</section>
</section>
<section><title>diameter 1.9.2</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Fix broken relay counters.</p>
<p>
OTP-12654 in OTP 17.5.3 broke counters in the case of
answer messages received in the relay application.
Counters were accumulated as unknown messages or
no_result_code instead of as relayed messages on the
intended Result-Code and 'Experimental-Result' tuples.</p>
<p>
Own Id: OTP-12741</p>
</item>
<item>
<p>
Fix diameter_sctp listener race.</p>
<p>
An oversight in OTP-12428 made it possible to start a
transport process that could not establish associations.</p>
<p>
Own Id: OTP-12744</p>
</item>
</list>
</section>
</section>
<section><title>diameter 1.9.1</title>
<section><title>Known Bugs and Problems</title>
<list>
<item>
<p>
Don't leave extra bit in decoded AVP data.</p>
<p>
OTP-12074 in OTP 17.3 missed one case: a length error on
a trailing AVP unknown to the dictionary in question.</p>
<p>
Own Id: OTP-12642</p>
</item>
<item>
<p>
Don't confuse Result-Code and Experimental-Result.</p>
<p>
The errors field of a decoded diameter_packet record was
populated with a Result-Code AVP when an
Experimental-Result containing a 3xxx Result-Code was
received in an answer not setting the E-bit. The correct
AVP is now extracted from the incoming message.</p>
<p>
Own Id: OTP-12654</p>
</item>
<item>
<p>
Don't count on unknown Application Id.</p>
<p>
OTP-11721 in OTP 17.1 missed the case of an Application
Id not agreeing with that of the dictionary in question,
causing counters to be accumulated on keys containing the
unknown id.</p>
<p>
Own Id: OTP-12701</p>
</item>
</list>
</section>
</section>
<section><title>diameter 1.9</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Don't discard outgoing answers unnecessarily.</p>
<p>
Answers missing a Result-Code AVP or setting an E-bit
inappropriately were discarded even if encode was
successful.</p>
<p>
Own Id: OTP-11492</p>
</item>
<item>
<p>
Increase supervision timeouts.</p>
<p>
At diameter application shutdown, DPR could be omitted on
open peer connections because of short supervision
timeouts.</p>
<p>
Own Id: OTP-12412</p>
</item>
<item>
<p>
Fix retransmission of messages sent as header/avps list.</p>
<p>
Extracting End-to-End and Hop-by-Hop Identifiers resulted
in a function clause error, resulting in a handle_error
callback.</p>
<p>
Own Id: OTP-12415</p>
</item>
<item>
<p>
Fix diameter_avp decode of Grouped AVPs having decode
errors.</p>
<p>
Components of such an AVP were not extracted, causing it
to be represented by a single diameter_avp record instead
of the intended list.</p>
<p>
Dictionary files must be recompiled for the fix to have
effect.</p>
<p>
Own Id: OTP-12475</p>
</item>
<item>
<p>
Fix ordering of AVPs in relayed messages.</p>
<p>
The order was reversed relative to the received order,
with a Route-Record AVP prepended.</p>
<p>
Thanks to Andrzej Trawiński.</p>
<p>
Own Id: OTP-12551</p>
</item>
<item>
<p>
Fix issues with DiameterURI encode/decode.</p>
<p>
RFC 6773 changed the default port and transport, but the
RFC 3588 defaults were used even if the RFC 6733 common
dictionary was in use. The RFC 3588 defaults are now only
used when the common dictionary is
diameter_gen_base_rfc3588.</p>
<p>
Both RFC 3588 and 6733 disallow
transport=udp;protocol=diameter. Encode of the
combination now fails.</p>
<p>
Decode of ports numbers outside the range 0-65535 and
fully qualified domain names longer than 255 octets now
fails.</p>
<p>
Note that RFC 3588 is obsolete, and that there is a
diameter_gen_base_rfc6733. The change in defaults is a
potential interoperability problem when moving to RFC
6733 with peers that do not send all URI components. The
fact that 6733 allows 5xxx result codes in answer
messages setting the E-bit, which RFC 3588 doesn't, is
another.</p>
<p>
Own Id: OTP-12589</p>
</item>
</list>
</section>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Add service_opt() string_decode.</p>
<p>
To disable the decode of potentially large binaries to
string. This prevents large strings from being copied
when incoming Diameter messages are passed between
processes, a vulnerability that can lead to memory being
exhausted given sufficiently malicious peers.</p>
<p>
The value is a boolean(), true being the default for
backwards compatibility. Setting false causes both
diameter_caps records and decoded messages to contain
binary() in relevant places that previously had string():
diameter_app(3) callbacks need to be prepared for the
change.</p>
<p>
The Diameter types affected are OctetString and the
derived types UTF8String, DiameterIdentity, DiameterURI,
IPFilterRule, and QoSFilterRule. Time and Address are
unaffected.</p>
<p>
Own Id: OTP-11952</p>
</item>
<item>
<p>
Add transport_opt() pool_size.</p>
<p>
To allow for pools of accepting transport processes,
which can better service multiple simultaneous peer
connections. The option can also be used with connecting
transports, to establish multiple connections to the same
peer without having to configure multiple transports.</p>
<p>
Own Id: OTP-12428</p>
</item>
<item>
<p>
Allow DPR to be sent with diameter:call/4.</p>
<p>
It has been possible to send, but the answer was regarded
as unsolicited and discarded. DPA now causes the
transport process in question to be terminated, as for
DPR that diameter itself sends.</p>
<p>
Own Id: OTP-12542</p>
</item>
<item>
<p>
Discard requests after DPR.</p>
<p>
RFC 6733 is imprecise, but the tone is that messages
received after DPR are an exception to be dealt with only
because of the possibility of unordered delivery over
SCTP. As a consequence, and because a request following
DPR is unlikely to be answered due to the impending loss
of the peer connection, discard outgoing requests
following an outgoing or incoming DPR. Incoming requests
are also discarded, with the exception of DPR itself.
Answers are sent and received as usual.</p>
<p>
Own Id: OTP-12543</p>
</item>
<item>
<p>
Add transport_opt() dpr_timeout.</p>
<p>
To cause a peer connection to be closed following an
outgoing DPA when the peer fails to do so. It is the
recipient of DPA that should close the connection
according to RFC 6733.</p>
<p>
Own Id: OTP-12609</p>
</item>
<item>
<p>
Add service_opt() incoming_maxlen.</p>
<p>
To bound the expected size of incoming Diameter messages.
Messages larger than the specified number of bytes are
discarded, to prevent a malicious peer from generating
excessive load.</p>
<p>
Own Id: OTP-12628</p>
</item>
</list>
</section>
</section>
<section><title>diameter 1.8</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Fix remote diameter_request table leak.</p>
<p>
An outgoing request whose pick_peer callback selected a
transport on another node resulted in an orphaned table
entry on that node.</p>
<p>
Own Id: OTP-12196</p>
</item>
<item>
<p>
Fix handling of 3xxx Result-Code without E-bit.</p>
<p>
OTP-12233 broke the population of the errors field of the
diameter_packet record when an incoming request with an
E-bit/Result-Code mismatch was detected, causing a
4-tuple to be inserted as Result-Code in a diameter_avp
record.</p>
<p>
Own Id: OTP-12233</p>
</item>
<item>
<p>
Fix ignored connect timer.</p>
<p>
There are two timers governing the establishment of peer
connections: connect_timer and watchdog_timer. The former
is the RFC 6733 Tc timer, and is used at initial
connection establishment. The latter is RFC 3539 TwInit,
and is used for connection reestablishment. A connecting
transport erroneously used watchdog_timer in both cases.</p>
<p>
Own Id: OTP-12281 Aux Id: seq12728 </p>
</item>
</list>
</section>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Order candidate peers in pick_peer callbacks.</p>
<p>
The order of candidate peers presented to a
diameter_app(3) pick_peer callback has previously not
been documented, but there are use cases that are
simplified by an ordering. The order is now determined by
the filter.</p>
<p>
Own Id: OTP-12308</p>
</item>
</list>
</section>
</section>
<section><title>diameter 1.7.1</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Don't leave extra bit in decoded AVP data.</p>
<p>
An extra bit could be communicated in the data field of a
diameter_avp record in the case of length errors. Of no
consequence for code using the record encoding of
Diameter messages, but code examining diameter_avp
records would see this bit.</p>
<p>
Dictionary files must be recompiled for the fix to have
effect.</p>
<p>
Own Id: OTP-12074</p>
</item>
<item>
<p>
Fix counting of outgoing requests and answers setting the
E-bit.</p>
<p>
OTP-11721 broke these counters for all outgoing requests
except DWR, and caused answers setting the E-bit to be
counted as unknown messages.</p>
<p>
Own Id: OTP-12080</p>
</item>
<item>
<p>
Fix Failed-AVP decode.</p>
<p>
The best-effort decode only worked for AVPs in the common
dictionary, not for those in the dictionary of the
application identified in the Diameter Header of the
answer message in question.</p>
<p>
Failed-AVP in an answer decoded with the RFC 3588 common
dictionary (diameter_gen_base_rfc3588) was regarded as an
error. The RFC 6733 dictionary was unaffected.</p>
<p>
Dictionary files must be recompiled for the fix to have
effect.</p>
<p>
Own Id: OTP-12094</p>
</item>
</list>
</section>
</section>
<section><title>diameter 1.7</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Improve robustness.</p>
<p>
Counters returned by diameter:service_info/2 now only
count messages known to the dictionary in question, so
that an attacker cannot cause arbitrarily many counters
to be created.</p>
<p>
Messages to the Erlang log have been minimized, and those
related to traffic have been removed entirely since an
attacker could cause a node to be logged to death.
Consequently, the default answer_errors configuration has
been changed from report to discard. A service needs to
be restarted for the change in default to take effect.</p>
<p>
Own Id: OTP-11721</p>
</item>
<item>
<p>
Fix request table leak.</p>
<p>
Outgoing Diameter requests are stored in a table until an
answer is received or times out. Calling
diameter:stop_service/1 before this took place would
orphan the entries, resulting in a memory leak.</p>
<p>
Own Id: OTP-11893</p>
</item>
<item>
<p>
Fix broken SCTP transport.</p>
<p>
OTP-11593 caused the sending of answer messages over SCTP
to fail.</p>
<p>
Own Id: OTP-11901 Aux Id: OTP-11593 </p>
</item>
<item>
<p>
Fix watchdog process leak.</p>
<p>
A failed capabilities exchange on a listening transport
would orphan a process, causing a memory leak.</p>
<p>
Own Id: OTP-11934</p>
</item>
<item>
<p>
Fix incorrect handling of incoming DPR.</p>
<p>
In the case of a listening transport, a reconnection by a
peer following DPR could transition the watchdog state to
REOPEN instead of OKAY.</p>
<p>
Own Id: OTP-11938</p>
</item>
<item>
<p>
Fix handling of AVP length errors on unknown AVPs.</p>
<p>
An AVP (Header) length that pointed past the end of the
message was not flagged as a 5014 error in this case.
Moreover, encoding such an AVP in the Failed-AVP of an
answer message as a consequence of other errors (eg.
M-bit, resulting in 5001) failed if the AVP contained a
complete header.</p>
<p>
Dictionary files must be recompiled for the fix to have
effect.</p>
<p>
Own Id: OTP-11946</p>
</item>
<item>
<p>
Fix broken check in dictionary compilation.</p>
<p>
That an AVP specified in the content of a @codecs or
@custom_types section was undefined went undetected,
causing compilation to fail when attempting to lookup the
AVP's type.</p>
<p>
Own Id: OTP-11958</p>
</item>
</list>
</section>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Add result code counters for CEA, DWA, and DPA.</p>
<p>
In addition to the existing result code counters on other
answer messages.</p>
<p>
Own Id: OTP-11891</p>
</item>
<item>
<p>
Add best-effort decode of AVPs within Failed-AVP.</p>
<p>
OTP-11007 disabled the decode of AVPs in Failed-AVP since
errors could cause the decode of Failed-AVP itself to
fail. Component AVPs are now decoded if possible,
otherwise not. AVPs of type Grouped are decoded as much
as possible, as deeply as possible.</p>
<p>
Dictionary files must be recompiled for the fix to have
effect.</p>
<p>
Own Id: OTP-11936 Aux Id: OTP-11007 </p>
</item>
<item>
<p>
Add counters for encode errors in outgoing Diameter
messages.</p>
<p>
In addition to the existing counters on decode errors.
The latter now count independently of result codes in
answer messages since decode errors do not preclude the
presence of a result code.</p>
<p>
Own Id: OTP-11937</p>
</item>
</list>
</section>
</section>
<section><title>diameter 1.6</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Add missing check at dictionary compilation.</p>
<p>
In particular, that an AVP defined as having type Grouped
in an @avp_types section has a corresponding definition
in a @grouped section.</p>
<p>
Own Id: OTP-11561</p>
</item>
<item>
<p>
Correct documentation on the setting of Origin-State-Id</p>
<p>
It was incorrectly stated that the AVP would be set in an
outgoing DPR/DPA.</p>
<p>
Own Id: OTP-11583</p>
</item>
<item>
<p>
Change interface for communicating outbound stream id to
diameter_sctp</p>
<p>
The module uses the transport_data field of record
diameter_packet to communicate the stream on which the an
incoming message is received and on which an outgoing
message should be sent, the previous interface being that
both are communicated as a tuple of the form {stream,
Id}. However, since diameter retains the value of an
incoming request's transport_data unless the
corresponding answer message specifies otherwise, the
behaviour in this case is to send an answer on the
outbound stream with the same identifier as the that of
the inbound stream on which the request was received. If
the inbound stream id is greater than or equal to the
number of outbound streams then this is guaranteed to
fail, causing the transport process in question to
terminate. There is no relationship between inbound and
outbound stream identifiers so diameter_sctp's imposition
of one is simply wrong.</p>
<p>
Outbound stream ids are now communicated with a different
tuple: {outstream, Id}, interpreted modulo the number of
outbound streams. Thus, retention of an inbound request's
transport_data has no effect on the selection of an
outbound stream.</p>
<p>
The change in interface is not strictly backwards
compatible because of the new atom for the outbound
stream. However, as there is currently no documented way
of obtaining the available number of outbound streams for
a peer connection, there is no way for a client to have
known the range of ids from which it could reliably have
chosen with the previous interface, so any setting of the
outbound stream has probably been unintentional. Not
explicitly specifying an outbound stream now results in a
round-robin selection.</p>
<p>
Thanks to Sharmila Pillai for reporting the problem.</p>
<p>
*** POTENTIAL INCOMPATIBILITY ***</p>
<p>
Own Id: OTP-11593</p>
</item>
<item>
<p>
Fix unicode path failure in diameter_make:codec/2.</p>
<p>
A dictionary path containing a unicode codepoint > 255
caused the function to fail.</p>
<p>
Own Id: OTP-11655</p>
</item>
<item>
<p>
Fix 'accept' config to diameter_sctp.</p>
<p>
OTP-10893 added support for {accept, Match} tuples to
specify addresses or regexps that should be matched
against peer addresses to decide whether or not a newly
established association should be retained, but this
hasn't been functional in the SCTP case because of
missing support in inet(3).</p>
<p>
The display of both local and peer addresses in
diameter:service_info/2 output has also been corrected.</p>
<p>
Own Id: OTP-11661 Aux Id: OTP-10229 </p>
</item>
<item>
<p>
Be lenient with the M-bit in Grouped AVPs.</p>
<p>
RFC 6733 says this, in 4.4:</p>
<taglist><tag></tag><item><p><c>Receivers of a Grouped AVP that does
not have the 'M' (mandatory) bit set and one or more of
the encapsulated AVPs within the group has the 'M'
(mandatory) bit set MAY simply be ignored if the Grouped
AVP itself is unrecognized. The rule applies even if the
encapsulated AVP with its 'M' (mandatory) bit set is
further encapsulated within other sub-groups, i.e., other
Grouped AVPs embedded within the Grouped
AVP.</c></p></item></taglist>
<p>
The first sentence is mangled but take it to mean this:</p>
<taglist><tag></tag><item><p><c>An unrecognized AVP of type Grouped
that does not set the 'M' bit MAY be ignored even if one
of its encapsulated AVPs sets the 'M'
bit.</c></p></item></taglist>
<p>
This is a bit of a non-statement since if the AVP is
unrecognized then its type is unknown. We therefore don't
know that its data bytes contain encapsulated AVPs, so
can't but ignore any of those that set the M-bit. Doing
anything else when the type *is* known would be
inconsistent.</p>
<p>
OTP-11087 (R16B03) caused the M-bit on any unrecognized
AVP to be regarded as an error, unrecognized being taken
to mean "not explicitly defined as a member of its
container". (That is, an AVP that can't be packed into a
dedicated record field, which is slightly stronger than
"not defined".) This fixed the original intention for
top-level AVPs but broke the required leniency for
Grouped AVPs whose type is known. This leniency is now
restored.</p>
<p>
Note that dictionary files need to be recompiled for the
change to have effect.</p>
<p>
Thanks to Rory McKeown for reporting the problem.</p>
<p>
Own Id: OTP-11675 Aux Id: OTP-11087 </p>
</item>
<item>
<p>
Fix pick_peer case clause failure.</p>
<p>
In the case of {call_mutates_state, true} configuration
on the service in question, any peer selection that
failed to select a peer resulted in a case clause
failure. This was noticed in the case of a peer failover
in which an alternate peer wasn't available.</p>
<p>
Own Id: OTP-11789</p>
</item>
</list>
</section>
</section>
<section><title>diameter 1.5</title>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Rename reconnect_timer to connect_timer.</p>
<p>
The former is still accepted for backwards compatibility,
but the name is misleading given the semantics of the
timer.</p>
<p>
Own Id: OTP-11168</p>
</item>
<item>
<p>
Extend diameter_make(3).</p>
<p>
Dictionaries can now be compiled from strings, not just
filesystem paths, and results can be returned instead of
written to the filesystem.</p>
<p>
Own Id: OTP-11348</p>
</item>
<item>
<p>
Remove hardcoding of diameter_base as @prefix on
dictionaries for application id 0.</p>
<p>
Own Id: OTP-11361</p>
</item>
</list>
</section>
</section>
<section><title>diameter 1.4.4</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Fix setting of End-to-End and Hop-by-Hop Identifiers in
outgoing DWA.</p>
<p>
Broken by OTP-11184, which caused the identifiers to be
set anew, discarding the values from the incoming DWR.</p>
<p>
Own Id: OTP-11367</p>
</item>
<item>
<p>
Fix handling of 5014, DIAMETER_INVALID_AVP_LENGTH.</p>
<p>
The error was detected as 5004,
DIAMETER_INVALID_AVP_VALUE, for some Diameter types, in
which case an AVP length that pointed past the end of a
message resulted in encode failure.</p>
<p>
Own Id: OTP-11395</p>
</item>
</list>
</section>
</section>
<section><title>diameter 1.4.3</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Fix UTF8String encode.</p>
<p>
Encode now accepts any nested list of codepoints and
binaries. A list containing a binary was previously
misinterpreted and the documentation was incomplete.</p>
<p>
Own Id: OTP-11172</p>
</item>
<item>
<p>
Ensure DWR isn't sent immediately after DWA.</p>
<p>
This was possible if the timing was unfortunate. An
incoming DWR now properly resets the watchdog timer.</p>
<p>
Own Id: OTP-11184</p>
</item>
<item>
<p>
Fix faulty encode of Failed-AVP</p>
<p>
Reception of a CER, DWR or DPR that has decode failures
caused encode of the corresponding answer message to
fail.</p>
<p>
Own Id: OTP-11293</p>
</item>
<item>
<p>
Fix broken service_opt() spawn_opt.</p>
<p>
The option was ignored.</p>
<p>
Own Id: OTP-11299</p>
</item>
</list>
</section>
</section>
<section><title>diameter 1.4.2</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Fix handling of 5014 (INVALID_AVP_LENGTH) errors.</p>
<p>
This was in some cases reported as 3009
(INVALID_AVP_BITS).</p>
<p>
Note that the correction is partially implemented in
modules generated by diameterc(1): a dictionary file must
be recompiled for the correction to apply to any messages
it defines.</p>
<p>
Own Id: OTP-11007</p>
</item>
<item>
<p>
Fix faulty capitalization in release notes.</p>
<p>
Diameter = the protocol.<br/> diameter = the Erlang
application.</p>
<p>
Own Id: OTP-11014</p>
</item>
<item>
<p>
Fix watchdog memory leak.</p>
<p>
Entries were not removed from a service-specific ets
table, causing them to be orphaned at connection
reestablishment for listening transports, and
diameter:remove_transport/2 for both listening and
connecting transports.</p>
<p>
The fault was introduced by OTP-10692 in diameter-1.4.1
(R16B).</p>
<p>
Own Id: OTP-11019 Aux Id: OTP-10692 </p>
</item>
<item>
<p>
Fix decode failure on AVP Length < 8.</p>
<p>
The failure caused the message in question to be
discarded.</p>
<p>
Own Id: OTP-11026</p>
</item>
<item>
<p>
Respect Host-IP-Address configuration.</p>
<p>
Addresses returned from a transport module were always
used to populate Host-IP-Address AVP's in an outgoing
CER/CEA, which precluded the sending of a VIP address.
Transport addresses are now only used if Host-IP-Address
is unspecified.</p>
<p>
Own Id: OTP-11045</p>
</item>
<item>
<p>
Fix mkdir race.</p>
<p>
Install could fail if examples/code and examples/dict
were created in parallel. Noticed on FreeBSD.</p>
<p>
Own Id: OTP-11051</p>
</item>
<item>
<p>
Fix recognition of 5001 on mandatory AVP's.</p>
<p>
An AVP setting the M-bit was not regarded as erroneous if
it was defined in the dictionary in question and its
container (message or Grouped AVP) had an 'AVP' field.
It's now regarded as a 5001 error (AVP_UNSUPPORTED), as
in the case that the AVP is not defined.</p>
<p>
Note that the correction is partially implemented in
modules generated by diameterc(1): a dictionary file must
be recompiled for the correction to apply to any messages
it defines.</p>
<p>
Own Id: OTP-11087</p>
</item>
<item>
<p>
Fix setting of Failed-AVP on handle_request
{answer_message, 5xxx} return.</p>
<p>
Failed-AVP was never in the outgoing answer-message. It
is now set with the AVP from the first entry with the
specified Result-Code in the errors field of the incoming
diameter_packet, if found.</p>
<p>
Own Id: OTP-11092</p>
</item>
<item>
<p>
Fix watchdog function_clause</p>
<p>
A listening transport on a service that allowed multiple
connections to the same peer could result in a
function_clause error in module diameter_watchdog. The
resulting crash was harmless but unseemly.</p>
<p>
Thanks to Aleksander Nycz.</p>
<p>
Own Id: OTP-11115</p>
</item>
<item>
<p>
Fix population of Failed-AVP.</p>
<p>
In cases in which diameter populated this AVP, many
values were sent instead of one as suggested by RFC 6733.
This was partially corrected by OTP-11007.</p>
<p>
Own Id: OTP-11127 Aux Id: OTP-11007 </p>
</item>
<item>
<p>
Fix list-valued Vendor-Specific-Application-Id config</p>
<p>
R16B (specifically, OTP-10760) broke the handling of such
configuration, resulting in a function clause error if
the list was not of length 3, and faulty interpretation
of the list's contents otherwise. Only record-valued
configuration was properly interpreted.</p>
<p>
Own Id: OTP-11165</p>
</item>
</list>
</section>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Allow peer connections to be shared between Erlang nodes
for the purpose of sending outgoing requests.</p>
<p>
A diameter_app(3) pick_peer/4 callback gets a list of
remote candidates as argument, allowing a callback on one
node to select a transport connection established on
another node. The service_opt() share_peers controls the
extent to which local connections are shared with remote
nodes. The service_opt() use_shared_peers controls the
extent to which connections shared from remote nodes are
utilized on the local node.</p>
<p>
Own Id: OTP-9610</p>
</item>
<item>
<p>
Allow listening diameter_{tcp,sctp} transports to be
configured with remote addresses.</p>
<p>
Option 'accept' allows remote addresses to be configured
as tuples or regular expressions. Remote addresses are
matched against the configured values at connection
establishment, any non-matching address causing the
connection to be aborted.</p>
<p>
Own Id: OTP-10893</p>
</item>
<item>
<p>
Detect more transport_opt() configuration errors at
diameter:add_transport/2.</p>
<p>
Many errors would previously not be detected until
transport start, diameter:add_transport/2 returning 'ok'
but transport connections failing to be established. An
error tuple is now returned.</p>
<p>
Own Id: OTP-10972</p>
</item>
<item>
<p>
Make explicit local address configuration optional in
diameter_tcp:start/3.</p>
<p>
The default address (as determined by gen_tcp) is now
used when a local address is not explicitly configured.</p>
<p>
Own Id: OTP-10986</p>
</item>
<item>
<p>
Improve handling of unrecognized service options.</p>
<p>
Such options were silently ignored by
diameter:start_service/2. An error tuple is now returned.</p>
<p>
Own Id: OTP-11017</p>
</item>
<item>
<p>
Don't send default Inband-Security-Id in CER/CEA.</p>
<p>
RFC 6733 recommends against the use of
Inband-Security-Id. Only send a value that differs from
the default, NO_INBAND_SECURITY = 0.</p>
<p>
Own Id: OTP-11050</p>
</item>
<item>
<p>
Make spawn options for request processes configurable.</p>
<p>
Own Id: OTP-11060</p>
</item>
</list>
</section>
</section>
<section><title>diameter 1.4.1.1</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Fix broken Vendor-Specific-Application-Id configuration.</p>
<p>
RFC 6733 changed the definition of this Grouped AVP,
changing the arity of Vendor-Id from 1* to 1. A component
Vendor-Id can now be either list- or integer-valued in
service and transport configuration, allowing it to be
used with both RFC 3588 and RFC 6733 dictionaries.</p>
<p>
Own Id: OTP-10942</p>
</item>
</list>
</section>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Add transport_opt() watchdog_config to allow non-standard
behaviour of the watchdog state machine.</p>
<p>
This can be useful during test but should not be used on
nodes that must conform to RFC 3539.</p>
<p>
Own Id: OTP-10898</p>
</item>
</list>
</section>
</section>
<section><title>diameter 1.4.1</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Fix erroneous watchdog transition from DOWN to INITIAL.</p>
<p>
This transition took place when a peer connection was
reestablished following a failed capabilities exchange.
RFC 3539 requires DOWN to transition into REOPEN.</p>
<p>
Own Id: OTP-10692</p>
</item>
</list>
</section>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Add application_opt() request_errors to make the handling
of incoming requests containing decode errors
configurable.</p>
<p>
The value 'callback' ensures that a handle_request
callback takes place for all such requests, the default
being for diameter to answer 3xxx series errors itself.</p>
<p>
Own Id: OTP-10686</p>
</item>
<item>
<p>
Add transport_opt() length_errors.</p>
<p>
The value determines how messages received over the
transport interface with an incorrect Message Length are
dealt with.</p>
<p>
Own Id: OTP-10687</p>
</item>
<item>
<p>
Add commentary on RFC 6733 to Standards Compliance
chapter of the User's Guide.</p>
<p>
Own Id: OTP-10688</p>
</item>
<item>
<p>
Allow a 5xxx result code in an answer-message on peer
connections using the RFC 6733 common dictionary.</p>
<p>
RFC 6733 allows this while RFC 3588 does not. A
handle_request callback can return {answer_message,
3000..3999|5000..5999} in the simplest case.</p>
<p>
Own Id: OTP-10759</p>
</item>
<item>
<p>
Add dictionaries for RFC 6733.</p>
<p>
Both the common and accounting dictionaries differ from
their RFC 3588 counterparts, which is reflected in
generated record definitions. Application configuration
on a service or transport determines the dictionary that
will be used on a given peer connection.</p>
<p>
Own Id: OTP-10760</p>
</item>
<item>
<p>
Allow a handle_request callback to control diameter's
setting of Result-Code and Failed-AVP.</p>
<p>
Setting errors = false in a returned #diameter_packet{}
disables the setting.</p>
<p>
Own Id: OTP-10761</p>
</item>
</list>
</section>
</section>
<section><title>diameter 1.4</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Add registered server names to the app file.</p>
<p>
Own Id: OTP-10442</p>
</item>
<item>
<p>
Fix #diameter_header{} handling broken by OTP-10445.</p>
<p>
The fault caused the the header of a [Header | Avps]
request to be ignored if both end_to_end_id and
hop_by_hop_id were undefined.</p>
<p>
Own Id: OTP-10609</p>
</item>
<item>
<p>
Fix error handling for handle_request callback.</p>
<p>
A callback that returned a #diameter_packet{} would fail
if the incoming request had decode errors.</p>
<p>
Own Id: OTP-10614</p>
</item>
<item>
<p>
Fix timing of service start event.</p>
<p>
The event did not necessarily precede other events as
documented.</p>
<p>
Own Id: OTP-10618</p>
</item>
<item>
<p>
Fix setting of header T flag at peer failover.</p>
<p>
The flag is now set in the diameter_header record passed
to a prepare_retransmit callback.</p>
<p>
Own Id: OTP-10619</p>
</item>
<item>
<p>
Fix sending of CER/CEA timeout event at capx_timeout.</p>
<p>
The event was not sent as documented.</p>
<p>
Own Id: OTP-10628</p>
</item>
<item>
<p>
Fix improper setting of Application-ID in the Diameter
header of an answer message whose E flag is set.</p>
<p>
The value should be that of the request in question. The
fault caused it always to be 0.</p>
<p>
Own Id: OTP-10655</p>
</item>
<item>
<p>
Fix faulty handling of AVP length errors.</p>
<p>
An incorrect AVP length but no other errors caused an
incoming request to fail.</p>
<p>
Own Id: OTP-10693</p>
</item>
</list>
</section>
</section>
<section><title>diameter 1.3.1</title>
<section><title>Known Bugs and Problems</title>
<list>
<item>
<p>
Fix function clause resulting from use of an eval
callback.</p>
<p>
Own Id: OTP-10685</p>
</item>
</list>
</section>
</section>
<section><title>diameter 1.3</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Fix faulty handling of Origin-State-Id and faulty config
values.</p>
<p>
The former was expected in a list despite the
documentation requiring (correctly) an integer. A bare
value for a list-valued capability was not handled.</p>
<p>
Own Id: OTP-10440</p>
</item>
<item>
<p>
Fix timing of up/down events.</p>
<p>
Previously, a call to diameter:call/4 following a peer_up
callback might incorrectly return {error, no_connection},
depending on timing. Both events now follow the
corresponding callbacks.</p>
<p>
Own Id: OTP-10459</p>
</item>
<item>
<p>
Make diameter:service_info/2 usable in peer_up, peer_down
and pick_peer callbacks.</p>
<p>
Except for in pick_peer when {call_mutates_state, false},
it would previously hang indefinitely.</p>
<p>
Own Id: OTP-10460</p>
</item>
<item>
<p>
Verify that End-to-End and Hop-by-Hop Identifiers in an
incoming CEA/DPA match those sent in the corresponding
CER/DPR.</p>
<p>
The values were previously ignored. Answers whose
identifiers do not match are handled as unexpected.</p>
<p>
Own Id: OTP-10565</p>
</item>
<item>
<p>
Fix formatting problems in PDF documentation.</p>
<p>
In particular, text corresponding to links in HTML was
omitted in preformatted blocks. There are still issues
with indentation but this is not diameter-specific.</p>
<p>
Own Id: OTP-10583</p>
</item>
</list>
</section>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Let prepare_request, prepare_retransmit and
handle_request callbacks return a function to be invoked
on outgoing messages after encode.</p>
<p>
This allows encoded messages to be logged for example.</p>
<p>
Own Id: OTP-10441</p>
</item>
<item>
<p>
Add service_opt() 'restrict_connections' to allow
multiple transport connections with the same peer.</p>
<p>
Own Id: OTP-10443</p>
</item>
<item>
<p>
Add service_opt() 'sequence' to allow the masking of a
constant onto the topmost bits of End-to-End and
Hop-by-Hop identifiers.</p>
<p>
This allows the same service on different nodes to use
distinct values in outgoing request messages.</p>
<p>
Own Id: OTP-10445</p>
</item>
<item>
<p>
Add diameter:service_info(PeerRef) to return the
transport_ref() and transport_opt() list of the
corresponding transport.</p>
<p>
This allows easy access to these from diameter_app
callbacks that only get peer_ref() as an argument.</p>
<p>
Own Id: OTP-10470</p>
</item>
<item>
<p>
Add reference pages diameter_codec(3) and
diameter_make(3).</p>
<p>
Own Id: OTP-10471</p>
</item>
<item>
<p>
Add events for service start and stop.</p>
<p>
Own Id: OTP-10492</p>
</item>
<item>
<p>
Add transport_opt() 'disconnect_cb' to make the sending
of DPR configurable.</p>
<p>
Whether or not DPR should be sent at application stop,
service stop or transport removal is determined by the
value returned by the callback, as is the
Disconnect-Cause and timeout if DPA is not received.</p>
<p>
Own Id: OTP-10493</p>
</item>
<item>
<p>
Add transport_opt() 'capx_timeout' for the timeout
associated with non-reception of CER/CEA.</p>
<p>
Own Id: OTP-10554</p>
</item>
<item>
<p>
Allow a handle_request callback to return a
#diameter_packet{}.</p>
<p>
This allows an answer to set transport_data and header
fields.</p>
<p>
Own Id: OTP-10566</p>
</item>
<item>
<p>
Update documentation for RFC 6733.</p>
<p>
RFC 3588 is now obsolete.</p>
<p>
Own Id: OTP-10568</p>
</item>
</list>
</section>
</section>
<section><title>diameter 1.2</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Fix broken Result-Code setting and Destination-Host/Realm
extraction.</p>
<p>
Result-Code was assumed to have arity 1 when setting this
value in an answer to a request containing AVP decode
errors. Destination-Host/Realm were only correctly
extracted from messages in the common application.</p>
<p>
Own Id: OTP-10202</p>
</item>
<item>
<p>
Handle insufficient capabilities configuration more
gracefully.</p>
<p>
A transport that does not have sufficient capabilities
configuration in order to encode CER/CEA will now emit an
error report noting the configuration error and exit
instead of failing. The error is not detected at
diameter:add_transport/2 since there is no requirement
that a service be configured before its transports.</p>
<p>
Own Id: OTP-10203</p>
</item>
<item>
<p>
Ensure a failing peer_up/down callback does not affect
transport connections to other peers.</p>
<p>
Such a failure would previously have taken down all of a
service's connections.</p>
<p>
Own Id: OTP-10215</p>
</item>
</list>
</section>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Statistics related to Diameter messages can be retrieved
using diameter:service_info/2.</p>
<p>
Both Diameter and socket-level statistics are available,
for both incoming and outgoing messages.</p>
<p>
Own Id: OTP-9608</p>
</item>
<item>
<p>
Allow multiple transport_module/config to
diameter:add_transport/2.</p>
<p>
Multiple values are attempted in sequence until one
results in an established connection. This provides a way
for a connecting transport to specify configuration in
order of preference. (For example, SCTP before TCP.)</p>
<p>
Own Id: OTP-9885</p>
</item>
<item>
<p>
Add events for state transitions in the RFC 3539 watchdog
state machine.</p>
<p>
The watchdog state is also available through
diameter:service_info/2.</p>
<p>
Own Id: OTP-10212</p>
</item>
<item>
<p>
Add diameter:service_info(SvcName, connections).</p>
<p>
This provides an alternative to
diameter:service_info(SvcName, transport) that presents
information per established connection instead of per
transport reference.</p>
<p>
Own Id: OTP-10213</p>
</item>
<item>
<p>
Assorted documentation corrections/improvements.</p>
<p>
Own Id: OTP-10216</p>
</item>
</list>
</section>
</section>
<section><title>diameter 1.1</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Fix fault in sending of 'closed' events.</p>
<p>
The fault made it possible for the 'closed' event not to
be sent following a failed capabilities exchange.</p>
<p>
Own Id: OTP-9824</p>
</item>
<item>
<p>
Fix faulty diameterc -name/-prefix.</p>
<p>
A minor blunder when introducing the new dictionary
parser in diameter-1.0 broke these options.</p>
<p>
Own Id: OTP-9826</p>
</item>
</list>
</section>
</section>
<section><title>diameter 1.0</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Fix faulty cleanup after diameter:remove_transport/2.</p>
<p>
Removing a transport removed the configuration but did
not prevent the transport process from being restarted.</p>
<p>
Own Id: OTP-9756</p>
</item>
</list>
</section>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
Add support for TLS over TCP.</p>
<p>
RFC 3588 requires that a Diameter server support TLS. In
practice this seems to mean TLS over SCTP since there are
limitations with running over SCTP: see RFC 6083 (DTLS
over SCTP), which is a response to RFC 3436 (TLS over
SCTP). The current RFC 3588 draft acknowledges this by
equating TLS with TLS/TCP and DTLS/SCTP.</p>
<p>
TLS handshaking can take place either following a CER/CEA
that negotiates TLS using the Inband-Security-Id AVP (the
method documented in RFC 3588) or immediately following
connection establishment (the method added to the current
draft).</p>
<p>
Own Id: OTP-9605</p>
</item>
<item>
<p>
Improvements to the dictionary parser.</p>
<p>
The dictionary parser now emits useful error messages in
case of faults in the input file, also identifying the
line number at which the fault was detected. There are
semantic checks that were missing in the previous parser,
a fault in the interpretation of vendor id's in
combination with @inherits has been fixed and @end can be
used to terminate parsing explicitly instead of always
parsing to end of file.</p>
<p>
Own Id: OTP-9639</p>
</item>
<item>
<p>
Improve dictionary reusability.</p>
<p>
Reusing a dictionary just to get a different generated
module name or prefix previously required taking a copy
of the source, which may consist of several files if
inheritance is used, just to edit a couple of lines which
don't affect the semantics of the Diameter application
being defined. Options --name, --prefix and --inherits
have been added to diameterc to allow corresponding
values to be set at compile time.</p>
<p>
Own Id: OTP-9641</p>
</item>
<item>
<p>
Add capabilities_cb transport option.</p>
<p>
Its value is a function that's applied to the transport
reference and capabilities record after capabilities
exchange. If a callback returns anything but 'ok' then
the connection is closed. In the case of an incoming CER,
the callback can return a result code with which to
answer. Multiple callbacks can be specified and are
applied until either all return 'ok' or one doesn't.</p>
<p>
This provides a way to reject a peer connection.</p>
<p>
Own Id: OTP-9654</p>
</item>
<item>
<p>
Add @codecs to dictionary format.</p>
<p>
The semantics are similar to @custom_types but results in
codec functions of the form TypeName(encode|decode,
AvpName, Data) rather than AvpName(encode|decode,
TypeName, Data). That is, the role of the AVP name and
Diameter type name are reversed. This eliminates the need
for exporting one function for each AVP sharing a common
specialized encode/decode.</p>
<p>
Own Id: OTP-9708 Aux Id: OTP-9639 </p>
</item>
<item>
<p>
Add #diameter_callback{} for more flexible callback
configuration.</p>
<p>
The record allows individual functions to be configured
for each of the diameter_app(3) callbacks, as well as a
default callback.</p>
<p>
Own Id: OTP-9777</p>
</item>
</list>
</section>
</section>
<section><title>diameter 0.10</title>
<section><title>Fixed Bugs and Malfunctions</title>
<list>
<item>
<p>
Handle #sctp_paddr_change and #sctp_pdapi_event from
gen_sctp.</p>
<p>
The events are enabled by default but diameter_sctp
neither disabled nor dealt with them. Reception of such
an event caused a transport process to crash.</p>
<p>
Own Id: OTP-9538</p>
</item>
<item>
<p>
Fix header folding bug.</p>
<p>
A prepare_request callback from diameter can return a
diameter_header record in order to set values in the
header of an outgoing request. A fault in
diameter_lib:fold_tuple/3 caused the subsequent encode of
the outgoing request to fail.</p>
<p>
Own Id: OTP-9577</p>
</item>
<item>
<p>
Fix bugs in sending of answer-message replies.</p>
<p>
3001 (DIAMETER_COMMAND_UNSUPPORTED) was not sent since
the decode placed the AVP list in the wrong field of the
diameter_packet, causing the subsequent encode to fail.
Session-Id was also set improperly, causing encode to
fail even in this case.</p>
<p>
Own Id: OTP-9578</p>
</item>
<item>
<p>
Fix improper use of error_logger:info_report/2.</p>
<p>
Function doesn't take a format string and arguments as it
was called. Instead use error_logger:info_report/1 and
use the same report format as used for warning and error
reports.</p>
<p>
Own Id: OTP-9579</p>
</item>
<item>
<p>
Fix and clarify semantics of peer filters.</p>
<p>
An eval filter returning a non-true value caused the call
process to fail and the doc was vague on how an exception
was treated. Clarify that the non-tuple host/realm
filters assume messages of a certain form.</p>
<p>
Own Id: OTP-9580</p>
</item>
<item>
<p>
Fix and clarify relay behaviour.</p>
<p>
Implicit filtering of the sending peer in relaying a
request could cause loop detection to be preempted in a
manner not specified by RFC3588. Reply with 3002
(DIAMETER_UNABLE_TO_DELIVER) on anything but an answer to
a relayed request.</p>
<p>
Own Id: OTP-9583</p>
</item>
</list>
</section>
<section><title>Improvements and New Features</title>
<list>
<item>
<p>
@id required in dictionary files only when @messages is
specified.</p>
<p>
@id defines an application identifier and this is used
only when sending or receiving messages. A dictionary can
define only AVP's however, to be included by other
dictionaries using @inherits, in which case it makes no
sense to require @id.</p>
<p>
Note that message definitions are not inherited with
@inherits, only AVP's</p>
<p>
Own Id: OTP-9467</p>
</item>
<item>
<p>
Allow @enum when AVP is defined in an inherited
dictionary.</p>
<p>
3GPP standards (for one) extend the values allowed for
RFC 3588 AVP's of type Enumerated. Previously, extending
an AVP was only possible by completely redefining the
AVP.</p>
<p>
Own Id: OTP-9469</p>
</item>
<item>
<p>
Migrate testsuites to pure common test and add both
suites and testcases.</p>
<p>
Own Id: OTP-9553</p>
</item>
<item>
<p>
Requests of arbitrary form.</p>
<p>
diameter:call/4 can be passed anything, as long as the
subsequent prepare_request callback returns a term that
can be encoded.</p>
<p>
Own Id: OTP-9581</p>
</item>
</list>
</section>
</section>
<section>
<title>diameter 0.9</title>
<p>
Initial release of the diameter application.</p>
<p>
Known issues or limitations:</p>
<list>
<item>
<p>
Some agent-related functionality is not entirely complete.
In particular, support for proxy agents, that advertise specific
Diameter applications but otherwise relay messages in much the same
way as relay agents (for which a handle_request
callback can return a <c>relay</c> tuple), will be completed in an
upcoming release.
There may also be more explicit support for redirect agents, although
redirect behaviour can be implemented with the current
functionality.</p>
</item>
<item>
<p>
There is some asymmetry in the treatment of messages sent as
<c>diameter_header/avp</c> records and those sent in the "normal"
fashion, and not all of this is documented.
This is related to the previous point since this form of sending a
message was introduced specifically to handle relay agent behaviour
using the same callback interface as for client/server behaviour.</p>
</item>
<item>
<p>
The User's Guide is currently quite thin.
The introductory chapter followed by the examples (in the application
<c>examples</c> subdirectory) may be sufficient
for those having some familiarity with the Diameter protocol but the
intention is to provide more introductory text.
The reference documentation is quite complete, although some points
could likely be expanded upon.</p>
</item>
<item>
<p>
The function diameter:service_info/2
can be used to retrieve information about a started service
(statistics, information about connected peers, etc) but
this is not yet documented and both the input and output may change
in the next release.</p>
</item>
</list>
<p>
See <seealso marker="diameter_soc">Standards Compliance</seealso> for
standards-related issues.</p>
</section>
</chapter>
