matrix-spec/drafts/erikj_federation.rst

Federation
==========
.. sectnum::
.. contents:: Table of Contents

Authorization
-------------

When receiving new events from remote servers, or creating new events, a server 
must know whether that event is allowed by the authorization rules. These rules
depend solely on the state at that event. The types of state events that affect
authorization are:

- ``m.room.create``
- ``m.room.member``
- ``m.room.join_rules``
- ``m.room.power_levels``

Servers should not create new events that reference unauthorized events. 
However, any event that does reference an unauthorized event is not itself
automatically considered unauthorized. 

Unauthorized events that appear in the event graph do *not* have any effect on 
the state of the graph. 

.. Note:: This is in contrast to redacted events which can still affect the 
          state of the graph. For example, a redacted *"join"* event will still
          result in the user being considered joined.
          

Rules
~~~~~

The following are the rules to determine if an event is authorized (this does
include validation).

**TODO**: What signatures do we expect?

1. If type is ``m.room.create`` allow.
#. If type is ``m.room.member``:
  
   a. If ``membership`` is ``join``:
    
      i. If the previous event is an ``m.room.create``, the depth is 1 and 
         the ``state_key`` is the creator, then allow.
      #. If the ``state_key`` does not match ``sender`` key, reject.
      #. If the current state has ``membership`` set to ``join``.
      #. If the ``sender`` is in the ``m.room.may_join`` list. [Not currently 
         implemented]
      #. If the ``join_rules`` is:
      
         - ``public``:  allow.
         - ``invite``: allow if the current state has ``membership`` set to 
           ``invite``
         - ``knock``: **TODO**.
         - ``private``: Reject.
         
      #. Reject

   #. If ``membership`` is ``invite`` then allow if ``sender`` is in room, 
      otherwise reject.
   #. If ``membership`` is ``leave``:
   
      i. If ``sender`` matches ``state_key`` allow.
      #. If ``sender``'s power level is greater than the the ``kick_level``
         given in the current ``m.room.power_levels`` state (defaults to 50),
         and the ``state_key``'s power level is less than or equal to the
         ``sender``'s power level, then allow.
      #. Reject.
      
   #. If ``membership`` is ``ban``:
   
      i. **TODO**.
   
   #. Reject.

#. Reject the event if the event type's required power level is less that the
   ``sender``'s power level.
#. If the ``sender`` is not in the room, reject.
#. If the type is ``m.room.power_levels``:

   a. **TODO**.

#. Allow.


Definitions
~~~~~~~~~~~

Required Power Level
  A given event type has an associated *required power level*. This is given
  by the current ``m.room.power_levels`` event, it is either listed explicitly
  in the ``events`` section or given by either ``state_default`` or 
  ``events_default`` depending on if the event type is a state event or not.


Auth events
~~~~~~~~~~~

The auth events of an event are the set of events used by the authorization 
algorithm to accept the event. These should be a subset of the current state.

A server is required to store the complete chain of auth events for all events
it serves to remote servers.

.. todo
    We probably should probably give a lower band of how long auth events
    should be kept around for.

Auth chain
~~~~~~~~~~

The *auth chain* for an event is the recursive list of auth events and the auth
chain for those auth events.

The auth chain for event gives all the information a server needs to accept an
event. However, being given an auth chain for an event that appears valid does
not mean that the event might later be rejected. For example if we discover
that the sender had been banned between the join event listed in the auth
events and the event being authed.

**TODO**: Clean the above explanations up a bit.


Auth chain resolution
~~~~~~~~~~~~~~~~~~~~~

**TODO**: If an auth check fails, or if we get told something we accepted
should have been rejected, we need to try and determine who is right.

Both should inform the other of what they think the current auth chain is. If
either are missing auth events that they know are valid (through authorization
and state resolution) they process the missing events as usual.

If either side notice that the other has accepted an auth events we think
should be rejected (for reasons *not* in their auth chain), that server should
inform the other with suitable proof.

The proofs can be:

- An *event chain* that shows an auth event is *not* an ancestor of the event.
  This can be done by giving the full ancestor chains up to the depth of the
  invalid auth event.
- Given an event (and event chain?) showing that authorization had been revoked.

If a server discovers it cannot prove the other side is wrong, then it accepts
that the other is correct; i.e. we always accept that the other side is correct
unless we can prove otherwise.



State Resolution
----------------

    **TODO**

When two branches in the event graph merge, the state of those branches might
differ, so a *state resolution* algorithm must be used to determine the current
state of the resultant merge.

The properties of the state resolution algorithm are:

- Must only depend on the event graph, and not local server state.
- When two state events are comparable, the descendant one should be picked.
- Must not require the full event graph.

The following algorithm satisfies these requirements; given two or more events,
pick the one with the greatest:

#. Depth.
#. Hash of event_id.


This works except in the case of auth events, where we need to mitigate against
the attack where servers artificially netsplit to avoid bans or power level
changes.

We want the following rules to apply:

#. If power levels have been changed on two different branches use the rules
   above, ensuring that the one picked is a valid change from the one not picked.
#. Similarly handle membership changes (e.g. bans, kicks, etc.)
#. If a power level has been changed in a branch, then any state merged from the
   other branch *must* be allowed by the power level event change. Otherwise,
   use the current one from the branch where the power level event changed.



State Conflict Resolution
-------------------------

If a server discovers that it disagrees with another about the current state,
it can follow the same process outlined in *Auth chain resolution* to resolve
these conflicts.

Constructing a new event
------------------------

    **TODO**

When constructing a new event, the server should insert the following fields:

- ``prev_events``: The list of event ids of what the server believes are the
  current leaf nodes of the event graph (i.e., nodes that have been received
  but are yet to be referenced by another event).
- ``depth``: An integer one greater than the maximum depth of the event's
  previous events.
- ``auth_events``: The list of event ids that authorizes this event. This
  should be a subset of the current state.
- ``origin_server_ts``: The time the server created the event.
- ``origin``: The name of the server.


Signing and Hashes
~~~~~~~~~~~~~~~~~~

    **TODO**

Validation
----------

    **TODO**

Domain specific string
    A string of the form ``<prefix><localpart>:<domain>``, where <prefix> is a
    single character, ``<localpart>`` is an arbitrary string that does not
    include a colon, and `<domain>` is a valid server name.

``room_id``
    A domain specific string with prefix ``!`` that is static across all events
    in a graph and uniquely identifies it. The ``domain`` should be that of the
    home server that created the room (i.e., the server that generated the
    first ``m.room.create`` event).

``sender``
    The entity that logically sent the event. This is usually a user id, but
    can also be a server name.

User Id
    A domain specific string with prefix ``@`` representing a user account. The
    ``domain`` is the home server of the user and is the server used to contact
    the user.

Joining a room
--------------

If a user requests to join a room that the server is already in (i.e. the a
user on that server has already joined the room) then the server can simply
generate a join event and send it as normal.

If the server is not already in the room it needs to will need to join via
another server that is already in the room. This is done as a two step process.

First, the local server requests from the remote server a skeleton of a join
event. The remote does this as the local server does not have the event graph
to use to fill out the ``prev_events`` key in the new event. Critically, the
remote server does not process the event it responded with.

Once the local server has this event, it fills it out with any extra data and
signs it. Once ready the local server sends this event to a remote server
(which could be the same or different from the first remote server), this
remote server then processes the event and distributes to all the other
participating servers in that room. The local server is told about the
current state and complete auth chain for the join event. The local server
can then process the join event itself.


.. Note::
   Finding which server to use to join any particular room is not specified.


Inviting a user
---------------

To invite a remote user to a room we need their home server to sign the invite
event. This is done by sending the event to the remote server, which then signs
the event, before distributing the invite to other servers.


Handling incoming events
------------------------

When a server receives an event, it should:

#. Check if it knows about the room. If it doesn't, then it should get the
   current state and auth events to determine whether the server *should* be in
   the room. If so continue, if not drop or reject the event
#. If the server already knew about the room, check the prev events to see if
   it is missing any events. If it is, request them. Servers should limit how
   far back they will walk the event graph for missing events.
#. If the server does not have all the prev events, then it should request the
   current state and auth events from a server.


Failures
--------

A server can notify a remote server about something it thinks it has done
wrong using the failures mechanism. For example, the remote accepted an event
the local think it shouldn't have.

A failure has a severity level depending on the action taken by the local
server. These levels are:

``FATAL``
    The local server could not parse the event, for example due to a missing
    required field.

``ERROR``
    The local server *could* parse the event, but it was rejected. For example,
    the event may have failed an authorization check.

``WARN``
    The local server accepted the event, but something was unexpected about it.
    For example, the event may have referenced another event the local server
    thought should be rejected.

A failure also includes several other fields:

``code``
    A numeric code (to be defined later) indicating a particular type of
    failure.

``reason``
    A short string indicating what was wrong, for diagnosis purposes on the
    remote server.

``affected``
    The event id of the event this failure is responding to. For example, if
    an accepted event referenced a rejected event, this would point to the
    accepted one.

``source``
    The event id of the event that was the source of this unexpected behaviour.
    For example, if an accepted event referenced a rejected event, this would
    point to the rejected one.


Appendix
========

    **TODO**

Example event:

.. code::

    {
        "auth_events": [
            [
                "$14187571482fLeia:localhost:8480",
                {
                    "sha256": "kiZUclzzPetHfy0rVoYKnYXnIv5VxH8a4996zVl8xbw"
                }
            ],
            [
                "$14187571480odWTd:localhost:8480",
                {
                    "sha256": "GqtndjviW9yPGaZ6EJfzuqVCRg5Lhoyo4YYv1NFP7fw"
                }
            ],
            [
                "$14205549830rrMar:localhost:8480",
                {
                    "sha256": "gZmL23QdWjNOmghEZU6YjqgHHrf2fxarKO2z5ZTbkig"
                }
            ]
        ],
        "content": {
            "body": "Test!",
            "msgtype": "m.text"
        },
        "depth": 250,
        "event_id": "$14207181140uTFlx:localhost:8480",
        "hashes": {
            "sha256": "k1nuafFdFvZXzhb5NeTE0Q2Jkqu3E8zkh3uH3mqwIxc"
        },
        "origin": "localhost:8480",
        "origin_server_ts": 1420718114694,
        "prev_events": [
            [
                "$142071809077XNNkP:localhost:8480",
                {
                    "sha256": "xOnU1b+4LOVz5qih0dkNFrdMgUcf35fKx9sdl/gqhjY"
                }
            ]
        ],
        "room_id": "!dwZDafgDEFTtpPKpLy:localhost:8480",
        "sender": "@bob:localhost:8480",
        "signatures": {
            "localhost:8480": {
                "ed25519:auto": "Nzd3D+emFBJJ4LCTzQEZaKO0Sa3sSTR1fGpu8OWXYn+7XUqke9Q1jYUewrEfxb3lPxlYWm/GztVUJizLz1K5Aw"
            }
        },
        "type": "m.room.message",
        "unsigned": {
            "age": 500
        }
    }