tor-parallel-relay-conn/doc/tor-spec.txt

$Id$ 

Tor Spec

Note: This is an attempt to specify Tor as it exists as implemented in
early June, 2003.  It is not recommended that others implement this
design as it stands; future versions of Tor will implement improved
protocols.

TODO: (very soon)
      - Specify truncate/truncated
      - Sendme w/stream0 is circuit sendme
      - Integrate -NM and -RD comments
      - EXTEND cells should have hostnames or nicknames, so that OPs never
        resolve OR hostnames.  Else DNS servers can give different answers to
        different OPs, and compromise their anonymity.

EVEN LATER:
      - Do TCP-style sequencing and ACKing of DATA cells so that we can afford
        to lose some data cells.
      - 

0. Notation:

   PK -- a public key.
   SK -- a private key 
   K  -- a key for a symmetric cypher

   a|b -- concatenation of 'a' with 'b'.

   All numeric values are encoded in network (big-endian) order.

   Unless otherwise specified, all symmetric ciphers are AES in counter
   mode, with an IV of all 0 bytes.  Asymmetric ciphers are either RSA
   with 1024-bit keys and exponents of 65537, or DH with the safe prime
   from rfc2409, section 6.2, whose hex representation is:

     "FFFFFFFFFFFFFFFFC90FDAA22168C234C4C6628B80DC1CD129024E08"
     "8A67CC74020BBEA63B139B22514A08798E3404DDEF9519B3CD3A431B"
     "302B0A6DF25F14374FE1356D6D51C245E485B576625E7EC6F44C42E9"
     "A637ED6B0BFF5CB6F406B7EDEE386BFB5A899FA5AE9F24117C4B1FE6"
     "49286651ECE65381FFFFFFFFFFFFFFFF"


1. System overview

Tor is a connection-oriented anonymizing communication service. Users
build a path known as a "virtual circuit" through the network, in which
each node knows its predecessor and successor, but no others. Traffic
flowing down the circuit is unwrapped by a symmetric key at each node,
which reveals the downstream node.

2. Connections

   There are two ways to connect to an onion router (OR). The first is
   as an onion proxy (OP), which allows the OP to authenticate the OR
   without authenticating itself.  The second is as another OR, which
   allows mutual authentication.

   Tor uses TLS for link encryption, using the cipher suite
   "TLS_DHE_RSA_WITH_AES_128_CBC_SHA".  An OR always sends a
   self-signed X.509 certificate whose commonName is the server's
   nickname, and whose public key is in the server directory.
   
   All parties receiving certificates must confirm that the public
   key is as it appears in the server directory, and close the
   connection if it is not.

   Once a TLS connection is established, the two sides send cells
   (specified below) to one another.  Cells are sent serially.  All
   cells are 256 bytes long.  Cells may be sent embedded in TLS
   records of any size or divided across TLS records, but the framing
   of TLS records should not leak information about the type or
   contents of the cells.

   OR-to-OR connections are never deliberately closed.  An OP should
   close a connection to an OR if there are no circuits running over
   the connection, and an amount of time (KeepalivePeriod, defaults to
   5 minutes) has passed.

3. Cell Packet format

   The basic unit of communication for onion routers and onion
   proxies is a fixed-width "cell".  Each cell contains the following
   fields:

        ACI (anonymous circuit identifier)    [2 bytes]
        Command                               [1 byte]
        Length                                [1 byte]
        Sequence number (unused, set to 0)    [4 bytes]
        Payload (padded with 0 bytes)         [248 bytes]
                                         [Total size: 256 bytes]

   The 'Command' field holds one of the following values:
         0 -- PADDING     (Padding)                 (See Sec 6.2)
         1 -- CREATE      (Create a circuit)        (See Sec 4)
         2 -- CREATED     (Acknowledge create)      (See Sec 4)
         3 -- RELAY       (End-to-end data)         (See Sec 5)
         4 -- DESTROY     (Stop using a circuit)    (See Sec 4)

   The interpretation of 'Length' and 'Payload' depend on the type of
   the cell.
      PADDING: Neither field is used.
      CREATE: Length is 144; the payload contains the first phase of the
        DH handshake.
      CREATED: Length is 128; the payload contains the second phase of
        the DH handshake.
      RELAY: Length is a value between 8 and 248; the first 'length'
        bytes of payload contain useful data.
      DESTROY: Neither field is used.

   Unused fields are filled with 0 bytes.  The payload is padded with
   0 bytes.

   PADDING cells are currently used to implement connection
   keepalive.  ORs and OPs send one another a PADDING cell every few
   minutes.

   CREATE and DESTROY cells are used to manage circuits; see section
   4 below.

   RELAY cells are used to send commands and data along a circuit; see
   section 5 below.


4. Circuit management

4.1. CREATE and CREATED cells

   Users set up circuits incrementally, one hop at a time. To create
   a new circuit, users send a CREATE cell to the first node, with the
   first half of the DH handshake; that node responds with a CREATED cell
   with the second half of the DH handshake. To extend a circuit past
   the first hop, the user sends an EXTEND relay cell (see section 5)
   which instructs the last node in the circuit to send a CREATE cell
   to extend the circuit.

   The payload for a CREATE cell is an 'onion skin', consisting of:
         RSA-encrypted data            [128 bytes]
         Symmetrically-encrypted data  [16 bytes]
   The RSA-encrypted portion contains:
         Symmetric key                 [16 bytes]
         First part of DH data (g^x)   [112 bytes]
   The symmetrically encrypted portion contains:
         Second part of DH data (g^x)  [16 bytes] 

   The two parts of the DH data, once decrypted and concatenated, form
   g^x as calculated by the client.

   The relay payload for an EXTEND relay cell consists of:
         Address                       [4 bytes]
         Port                          [2 bytes]
         Onion skin                    [144 bytes]

   The port and address field denote the IPV4 address and port of the
   next onion router in the circuit.  

4.2. Setting circuit keys

   Once the handshake between the OP and an OR is completed, both
   servers can now calculate g^xy with ordinary DH.  From the base key
   material g^xy, they compute two 16 byte keys, called Kf and Kb as
   follows.  First, the server represents g^xy as a big-endian
   unsigned integer.  Next, the server computes 40 bytes of key data
   as K = SHA1(g^xy | [00]) | SHA1(g^xy | [01]) where "00" is a single
   octet whose value is zero, and "01" is a single octet whose value
   is one.  The first 16 bytes of K form Kf, and the next 16 bytes of
   K form Kb.  

   Kf is used to encrypt the stream of data going from the OP to the
   OR, whereas Kb is used to encrypt the stream of data going from the
   OR to the OP.

4.3. Creating circuits

   When creating a circuit through the network, the circuit creator
   performs the following steps:

      1. Choose a chain of N onion routers (R_1...R_N) to constitute
         the path, such that no router appears in the path twice.
         [this is wrong, see October 2003 discussion on or-dev]

      2. If not already connected to the first router in the chain,
         open a new connection to that router.

      3. Choose an ACI not already in use on the connection with the
         first router in the chain.  If we are an onion router and our
         nickname is lexicographically greater than the nickname of the
         other side, then let the high bit of the ACI be 1, else 0.

      4. Send a CREATE cell along the connection, to be received by
         the first onion router.

      5. Wait until a CREATED cell is received; finish the handshake
         and extract the forward key Kf_1 and the back key Kb_1.

      6. For each subsequent onion router R (R_2 through R_N), extend
         the circuit to R.

   To extend the circuit by a single onion router R_M, the circuit
   creator performs these steps:

      1. Create an onion skin, encrypting the RSA-encrypted part with
         R's public key.

      2. Encrypt and send the onion skin in a relay EXTEND cell along
         the circuit (see section 5).

      3. When a relay EXTENDED cell is received, calculate the shared
         keys.  The circuit is now extended.

   When an onion router receives an EXTEND relay cell, it sends a
   CREATE cell to the next onion router, with the enclosed onion skin
   as its payload.  The initiating onion router chooses some ACI not
   yet used on the connection between the two onion routers.  (But see
   section 4.3. above, concerning choosing ACIs.)

   As an extension (called router twins), if the desired next onion
   router R in the circuit is down, and some other onion router R'
   has the same key as R, then it's ok to extend to R' rather than R.

   When an onion router receives a CREATE cell, if it already has a
   circuit on the given connection with the given ACI, it drops the
   cell.  Otherwise, sometime after receiving the CREATE cell, it completes
   the DH handshake, and replies with a CREATED cell, containing g^y
   as its [128 byte] payload.  Upon receiving a CREATED cell, an onion
   router packs it payload into an EXTENDED relay cell (see section 5),
   and sends that cell up the circuit.  Upon receiving the EXTENDED
   relay cell, the OP can retrieve g^y.

   (As an optimization, OR implementations may delay processing onions
   until a break in traffic allows time to do so without harming
   network latency too greatly.)

4.4. Tearing down circuits

   Circuits are torn down when an unrecoverable error occurs along
   the circuit, or when all streams on a circuit are closed and the
   circuit's intended lifetime is over.  Circuits may be torn down
   either completely or hop-by-hop.

   To tear down a circuit completely, an OR or OP sends a DESTROY
   cell to the adjacent nodes on that circuit, using the appropriate
   direction's ACI.

   Upon receiving an outgoing DESTROY cell, an OR frees resources
   associated with the corresponding circuit. If it's not the end of
   the circuit, it sends a DESTROY cell for that circuit to the next OR
   in the circuit. If the node is the end of the circuit, then it tears
   down any associated edge connections (see section 5.1).

   After a DESTROY cell has been processed, an OR ignores all data or
   destroy cells for the corresponding circuit.

   To tear down part of a circuit, the OP sends a RELAY_TRUNCATE cell
   signaling a given OR (Stream ID zero).  That OR sends a DESTROY
   cell to the next node in the circuit, and replies to the OP with a
   RELAY_TRUNCATED cell.

   When an unrecoverable error occurs along one connection in a
   circuit, the nodes on either side of the connection should, if they
   are able, act as follows:  the node closer to the OP should send a
   RELAY_TRUNCATED cell towards the OP; the node farther from the OP
   should send a DESTROY cell down the circuit.

   [We'll have to reevaluate this section once we figure out cleaner
    circuit/connection killing conventions. -RD]

4.5. Routing data cells

   When an OR receives a RELAY cell, it checks the cell's ACI and
   determines whether it has a corresponding circuit along that
   connection.  If not, the OR drops the RELAY cell.

   Otherwise, if the OR is not at the OP edge of the circuit (that is,
   either an 'exit node' or a non-edge node), it de/encrypts the length
   field and the payload with AES/CTR, as follows:
        'Forward' relay cell (same direction as CREATE):
            Use Kf as key; encrypt.
        'Back' relay cell (opposite direction from CREATE):
            Use Kb as key; decrypt.
   If the OR recognizes the stream ID on the cell (it is either the ID
   of an open stream or the signaling (zero) ID), the OR processes the
   contents of the relay cell.  Otherwise, it passes the decrypted
   relay cell along the circuit if the circuit continues, or drops the
   cell if it's the end of the circuit. [Getting an unrecognized
   relay cell at the end of the circuit must be allowed for now;
   we can reexamine this once we've designed full tcp-style close
   handshakes. -RD]

   Otherwise, if the data cell is coming from the OP edge of the
   circuit, the OP decrypts the length and payload fields with AES/CTR as
   follows:
         OP sends data cell to node R_M:
            For I=1...M, decrypt with Kf_I.

   Otherwise, if the data cell is arriving at the OP edge if the
   circuit, the OP encrypts the length and payload fields with AES/CTR as
   follows:
         OP receives data cell:
            For I=N...1, 
                Encrypt with Kb_I.  If the stream ID is a recognized
                stream for R_I, or if the stream ID is the signaling
                ID (zero), then stop and process the payload.

   For more information, see section 5 below.

5. Application connections and stream management

5.1. Streams

   Within a circuit, the OP and the exit node use the contents of
   RELAY packets to tunnel end-to-end commands and TCP connections
   ("Streams") across circuits.  End-to-end commands can be initiated
   by either edge; streams are initiated by the OP.

   The first 8 bytes of each relay cell are reserved as follows:
         Relay command           [1 byte]
         Stream ID               [7 bytes]

   The recognized relay commands are:
         1 -- RELAY_BEGIN
         2 -- RELAY_DATA
         3 -- RELAY_END
         4 -- RELAY_CONNECTED
         5 -- RELAY_SENDME
         6 -- RELAY_EXTEND
         7 -- RELAY_EXTENDED
         8 -- RELAY_TRUNCATE
         9 -- RELAY_TRUNCATED
        10 -- RELAY_DROP

   All RELAY cells pertaining to the same tunneled stream have the
   same stream ID.  Stream ID's are chosen randomly by the OP.  A
   stream ID is considered "recognized" on a circuit C by an OP or an
   OR if it already has an existing stream established on that
   circuit, or if the stream ID is equal to the signaling stream ID,
   which is all zero: [00 00 00 00 00 00 00]

   To create a new anonymized TCP connection, the OP sends a
   RELAY_BEGIN data cell with a payload encoding the address and port
   of the destination host.  The stream ID is zero.  The payload format is:
         NEWSTREAMID | ADDRESS | ':' | PORT | '\000'
   where NEWSTREAMID is the newly generated Stream ID to use for
   this stream, ADDRESS may be a DNS hostname, or an IPv4 address in
   dotted-quad format; and where PORT is encoded in decimal.

   Upon receiving this packet, the exit node resolves the address as
   necessary, and opens a new TCP connection to the target port.  If
   the address cannot be resolved, or a connection can't be
   established, the exit node replies with a RELAY_END cell.
   Otherwise, the exit node replies with a RELAY_CONNECTED cell.

   The OP waits for a RELAY_CONNECTED cell before sending any data.
   Once a connection has been established, the OP and exit node
   package stream data in RELAY_DATA cells, and upon receiving such
   cells, echo their contents to the corresponding TCP stream.  

   Relay RELAY_DROP cells are long-range dummies; upon receiving such
   a cell, the OR or OP must drop it.

5.2. Closing streams

   [Note -- TCP streams can only be half-closed for reading.  Our
   Bickford's conversation was incorrect. -NM]

   Because TCP connections can be half-open, we follow an equivalent
   to TCP's FIN/FIN-ACK/ACK protocol to close streams.

   An exit connection can have a TCP stream in one of three states:
   'OPEN', 'DONE_PACKAGING', and 'DONE_DELIVERING'.  For the purposes
   of modeling transitions, we treat 'CLOSED' as a fourth state,
   although connections in this state are not, in fact, tracked by the
   onion router.

   A stream begins in the 'OPEN' state.  Upon receiving a 'FIN' from
   the corresponding TCP connection, the edge node sends a 'RELAY_END'
   cell along the circuit and changes its state to 'DONE_PACKAGING'.
   Upon receiving a 'RELAY_END' cell, an edge node sends a 'FIN' to
   the corresponding TCP connection (e.g., by calling
   shutdown(SHUT_WR)) and changing its state to 'DONE_DELIVERING'.

   When a stream in already in 'DONE_DELIVERING' receives a 'FIN', it
   also sends a 'RELAY_END' along the circuit, and changes its state
   to 'CLOSED'.  When a stream already in 'DONE_PACKAGING' receives a
   'RELAY_END' cell, it sends a 'FIN' and changes its state to
   'CLOSED'.

   [Note: Please rename 'RELAY_END2'. :) -NM ]

   If an edge node encounters an error on any stram, it sends a
   'RELAY_END2' cell along the circuit (if possible) and closes the
   TCP connection immediately.  If an edge node receives a
   'RELAY_END2' cell for any stream, it closes the TCP connection
   completely, and sends nothing along the circuit.

6. Flow control

6.1. Link throttling

   Each node should do appropriate bandwidth throttling to keep its
   user happy.

   Communicants rely on TCP's default flow control to push back when they
   stop reading.

6.2. Link padding

   Currently nodes are not required to do any sort of link padding or
   dummy traffic. Because strong attacks exist even with link padding,
   and because link padding greatly increases the bandwidth requirements
   for running a node, we plan to leave out link padding until this
   tradeoff is better understood.

6.3. Circuit-level flow control

   To control a circuit's bandwidth usage, each OR keeps track of
   two 'windows', consisting of how many RELAY_DATA cells it is
   allowed to package for transmission, and how many RELAY_DATA cells
   it is willing to deliver to streams outside the network.  
   Each 'window' value is initially set to 1000 data cells
   in each direction (cells that are not data cells do not affect
   the window).  When an OR is willing to deliver more cells, it sends a
   RELAY_SENDME cell towards the OP, with Stream ID zero.  When an OR 
   receives a RELAY_SENDME cell with stream ID zero, it increments its
   packaging window.

   Each of these cells increments the corresponding window by 100.

   The OP behaves identically, except that it must track a packaging
   window and a delivery window for every OR in the circuit.
 
   An OR or OP sends cells to increment its delivery window when the
   corresponding window value falls under some threshold (900).

   If a packaging window reaches 0, the OR or OP stops reading from
   TCP connections for all streams on the corresponding circuit, and
   sends no more RELAY_DATA cells until receiving a RELAY_SENDME cell.
[this stuff is badly worded; copy in the tor-design section -RD]

6.4. Stream-level flow control

   Edge nodes use RELAY_SENDME cells to implement end-to-end flow
   control for individual connections across circuits. Similarly to
   circuit-level flow control, edge nodes begin with a window of cells
   (500) per stream, and increment the window by a fixed value (50)
   upon receiving a RELAY_SENDME cell. Edge nodes initiate RELAY_SENDME
   cells when both a) the window is <= 450, and b) there are less than
   ten cell payloads remaining to be flushed at that edge.


7. Directories and routers

7.1. Router descriptor format.

(Unless otherwise noted, tokens on the same line are space-separated.)

Router ::= Router-Line  Date-Line Onion-Key Link-Key Signing-Key  Exit-Policy Router-Signature NL
Router-Line ::= "router" nickname address ORPort SocksPort DirPort bandwidth NL
Date-Line ::= "published" YYYY-MM-DD HH:MM:SS NL
Onion-key ::= "onion-key"  NL  a public key in PEM format   NL
Link-key ::= "link-key"  NL  a public key in PEM format  NL
Signing-Key ::= "signing-key"  NL  a public key in PEM format   NL
Exit-Policy ::= Exit-Line*
Exit-Line ::= ("accept"|"reject")  string  NL
Router-Signature ::= "router-signature"  NL  Signature
Signature ::= "-----BEGIN SIGNATURE-----" NL
              Base-64-encoded-signature NL "-----END SIGNATURE-----" NL

ORport ::= port where the router listens for routers/proxies (speaking cells)
SocksPort ::=  where the router listens for applications (speaking socks)
DirPort ::= where the router listens for directory download requests
bandwidth ::= maximum bandwidth, in bytes/s

nickname ::= between 1 and 32 alphanumeric characters.  case-insensitive.

Example:
router moria1 moria.mit.edu 9001 9021 9031 100000
published 2003-09-24 19:36:05
-----BEGIN RSA PUBLIC KEY-----
MIGJAoGBAMBBuk1sYxEg5jLAJy86U3GGJ7EGMSV7yoA6mmcsEVU3pwTUrpbpCmwS
7BvovoY3z4zk63NZVBErgKQUDkn3pp8n83xZgEf4GI27gdWIIwaBjEimuJlEY+7K
nZ7kVMRoiXCbjL6VAtNa4Zy1Af/GOm0iCIDpholeujQ95xew7rQnAgMA//8=
-----END RSA PUBLIC KEY-----
signing-key
-----BEGIN RSA PUBLIC KEY-----
7BvovoY3z4zk63NZVBErgKQUDkn3pp8n83xZgEf4GI27gdWIIwaBjEimuJlEY+7K
MIGJAoGBAMBBuk1sYxEg5jLAJy86U3GGJ7EGMSV7yoA6mmcsEVU3pwTUrpbpCmwS
f/GOm0iCIDpholeujQ95xew7rnZ7kVMRoiXCbjL6VAtNa4Zy1AQnAgMA//8=
-----END RSA PUBLIC KEY-----
reject 18.0.0.0/24

Note: The extra newline at the end of the router block is intentional.

7.2. Directory format

Directory ::= Directory-Header  Directory-Router  Router*  Signature
Directory-Header ::= "signed-directory" NL Software-Line NL
Software-Line: "recommended-software"  comma-separated-version-list
Directory-Router ::= Router
Directory-Signature ::= "directory-signature"  NL  Signature
Signature ::= "-----BEGIN SIGNATURE-----" NL
              Base-64-encoded-signature NL "-----END SIGNATURE-----" NL

Note:  The router block for the directory server must appear first.
The signature is computed by computing the SHA-1 hash of the
directory, from the characters "signed-directory", through the newline
after "directory-signature".  This digest is then padded with PKCS.1,
and signed with the directory server's signing key.

7.3. Behavior of a directory server

lists nodes that are connected currently
speaks http on a socket, spits out directory on request

-----------
(for emacs)
  Local Variables:
  mode:text
  indent-tabs-mode:nil
  fill-column:77
  End: