3 Tor Protocol Specification
8 Note: This is an attempt to specify Tor as currently implemented. Future
9 versions of Tor will implement improved protocols, and compatibility is not
12 This is not a design document; most design criteria are not examined. For
13 more information on why Tor acts as it does, see tor-design.pdf.
16 - REASON_CONNECTFAILED should include an IP.
17 - Copy prose from tor-design to make everything more readable.
23 K -- a key for a symmetric cypher
25 a|b -- concatenation of 'a' and 'b'.
27 [A0 B1 C2] -- a three-byte sequence, containing the bytes with
28 hexadecimal values A0, B1, and C2, in that order.
30 All numeric values are encoded in network (big-endian) order.
32 Unless otherwise specified, all symmetric ciphers are AES in counter
33 mode, with an IV of all 0 bytes. Asymmetric ciphers are either RSA
34 with 1024-bit keys and exponents of 65537, or DH with the safe prime
35 from rfc2409, section 6.2, whose hex representation is:
37 "FFFFFFFFFFFFFFFFC90FDAA22168C234C4C6628B80DC1CD129024E08"
38 "8A67CC74020BBEA63B139B22514A08798E3404DDEF9519B3CD3A431B"
39 "302B0A6DF25F14374FE1356D6D51C245E485B576625E7EC6F44C42E9"
40 "A637ED6B0BFF5CB6F406B7EDEE386BFB5A899FA5AE9F24117C4B1FE6"
41 "49286651ECE65381FFFFFFFFFFFFFFFF"
43 All "hashes" are 20-byte SHA1 cryptographic digests.
45 When we refer to "the hash of a public key", we mean the SHA1 hash of the
46 ASN.1 encoding of an RSA public key (as specified in PKCS.1).
50 Onion Routing is a distributed overlay network designed to anonymize
51 low-latency TCP-based applications such as web browsing, secure shell,
52 and instant messaging. Clients choose a path through the network and
53 build a ``circuit'', in which each node (or ``onion router'' or ``OR'')
54 in the path knows its predecessor and successor, but no other nodes in
55 the circuit. Traffic flowing down the circuit is sent in fixed-size
56 ``cells'', which are unwrapped by a symmetric key at each node (like
57 the layers of an onion) and relayed downstream.
61 There are two ways to connect to an onion router (OR). The first is
62 as an onion proxy (OP), which allows the OP to authenticate the OR
63 without authenticating itself. The second is as another OR, which
64 allows mutual authentication.
66 Tor uses TLS for link encryption. All implementations MUST support
67 the TLS ciphersuite "TLS_EDH_RSA_WITH_DES_192_CBC3_SHA", and SHOULD
68 support "TLS_DHE_RSA_WITH_AES_128_CBC_SHA" if it is available.
69 Implementations MAY support other ciphersuites, but MUST NOT
70 support any suite without ephemeral keys, symmetric keys of at
71 least 128 bits, and digests of at least 160 bits.
73 An OP or OR always sends a two-certificate chain, consisting of a
74 self-signed certificate containing the OR's identity key, and a second
75 certificate using a short-term connection key. The commonName of the
76 second certificate is the OR's nickname, and the commonName of the first
77 certificate is the OR's nickname, followed by a space and the string
80 All parties receiving certificates must confirm that the identity key is
81 as expected. (When initiating a connection, the expected identity key is
82 the one given in the directory; when creating a connection because of an
83 EXTEND cell, the expected identity key is the one given in the cell.) If
84 the key is not as expected, the party must close the connection.
86 All parties SHOULD reject connections to or from ORs that have malformed
87 or missing certificates. ORs MAY accept or reject connections from OPs
88 with malformed or missing certificates.
90 Once a TLS connection is established, the two sides send cells
91 (specified below) to one another. Cells are sent serially. All
92 cells are 512 bytes long. Cells may be sent embedded in TLS
93 records of any size or divided across TLS records, but the framing
94 of TLS records MUST NOT leak information about the type or contents
97 OR-to-OR connections are never deliberately closed. When an OR
98 starts or receives a new directory, it tries to open new
99 connections to any OR it is not already connected to.
100 [not true, unused OR conns close after 5 mins too -RD]
102 OR-to-OP connections are not permanent. An OP should close a
103 connection to an OR if there are no circuits running over the
104 connection, and an amount of time (KeepalivePeriod, defaults to 5
107 3. Cell Packet format
109 The basic unit of communication for onion routers and onion
110 proxies is a fixed-width "cell". Each cell contains the following
115 Payload (padded with 0 bytes) [509 bytes]
116 [Total size: 512 bytes]
118 The CircID field determines which circuit, if any, the cell is
121 The 'Command' field holds one of the following values:
122 0 -- PADDING (Padding) (See Sec 6.2)
123 1 -- CREATE (Create a circuit) (See Sec 4)
124 2 -- CREATED (Acknowledge create) (See Sec 4)
125 3 -- RELAY (End-to-end data) (See Sec 5)
126 4 -- DESTROY (Stop using a circuit) (See Sec 4)
128 The interpretation of 'Payload' depends on the type of the cell.
129 PADDING: Payload is unused.
130 CREATE: Payload contains the handshake challenge.
131 CREATED: Payload contains the handshake response.
132 RELAY: Payload contains the relay header and relay body.
133 DESTROY: Payload is unused.
134 Upon receiving any other value for the command field, an OR must
137 The payload is padded with 0 bytes.
139 PADDING cells are currently used to implement connection keepalive.
140 If there is no other traffic, ORs and OPs send one another a PADDING
141 cell every few minutes.
143 CREATE, CREATED, and DESTROY cells are used to manage circuits;
146 RELAY cells are used to send commands and data along a circuit; see
149 4. Circuit management
151 4.1. CREATE and CREATED cells
153 Users set up circuits incrementally, one hop at a time. To create a
154 new circuit, OPs send a CREATE cell to the first node, with the
155 first half of the DH handshake; that node responds with a CREATED
156 cell with the second half of the DH handshake plus the first 20 bytes
157 of derivative key data (see section 4.2). To extend a circuit past
158 the first hop, the OP sends an EXTEND relay cell (see section 5)
159 which instructs the last node in the circuit to send a CREATE cell
160 to extend the circuit.
162 The payload for a CREATE cell is an 'onion skin', which consists
163 of the first step of the DH handshake data (also known as g^x).
165 The data is encrypted to Bob's PK as follows: Suppose Bob's PK is
166 L octets long. If the data to be encrypted is shorter than L-42,
167 then it is encrypted directly (with OAEP padding). If the data is at
168 least as long as L-42, then a randomly generated 16-byte symmetric
169 key is prepended to the data, after which the first L-16-42 bytes
170 of the data are encrypted with Bob's PK; and the rest of the data is
171 encrypted with the symmetric key.
173 So in this case, the onion skin on the wire looks like:
175 OAEP padding [42 bytes]
176 Symmetric key [16 bytes]
177 First part of g^x [70 bytes]
178 Symmetrically encrypted:
179 Second part of g^x [58 bytes]
181 The relay payload for an EXTEND relay cell consists of:
184 Onion skin [186 bytes]
185 Public key hash [20 bytes]
187 The port and address field denote the IPV4 address and port of the next
188 onion router in the circuit; the public key hash is the SHA1 hash of the
189 PKCS#1 ASN1 encoding of the next onion router's identity (signing) key.
191 [XXXX Before 0.0.8, EXTEND cells did not include the public key hash.
192 Servers running 0.0.8 distinguish the old-style cells based on the
193 length of payloads. (Servers running 0.0.7 blindly pass on the extend
194 cell regardless of length.) In a future release, old-style EXTEND
195 cells will not be supported.]
197 The payload for a CREATED cell, or the relay payload for an
198 EXTENDED cell, contains:
199 DH data (g^y) [128 bytes]
200 Derivative key data (KH) [20 bytes] <see 4.2 below>
202 The CircID for a CREATE cell is an arbitrarily chosen 2-byte integer,
203 selected by the node (OP or OR) that sends the CREATE cell. To prevent
204 CircID collisions, when one OR sends a CREATE cell to another, it chooses
205 from only one half of the possible values based on the ORs' public
206 identity keys: if the sending OR has a lower key, it chooses a CircID with
207 an MSB of 0; otherwise, it chooses a CircID with an MSB of 1.
209 Public keys are compared numerically by modulus.
211 (Older versions of Tor compared OR nicknames, and did it in a broken and
212 unreliable way. To support versions of Tor earlier than 0.0.9pre6,
213 implementations should notice when the other side of a connection is
214 sending CREATE cells with the "wrong" MSG, and switch accordingly.)
216 4.2. Setting circuit keys
218 Once the handshake between the OP and an OR is completed, both
219 servers can now calculate g^xy with ordinary DH. From the base key
220 material g^xy, they compute derivative key material as follows.
221 First, the server represents g^xy as a big-endian unsigned integer.
222 Next, the server computes 100 bytes of key data as K = SHA1(g^xy |
223 [00]) | SHA1(g^xy | [01]) | ... SHA1(g^xy | [04]) where "00" is
224 a single octet whose value is zero, [01] is a single octet whose
225 value is one, etc. The first 20 bytes of K form KH, bytes 21-40 form
226 the forward digest Df, 41-60 form the backward digest Db, 61-76 form
227 Kf, and 77-92 form Kb.
229 KH is used in the handshake response to demonstrate knowledge of the
230 computed shared key. Df is used to seed the integrity-checking hash
231 for the stream of data going from the OP to the OR, and Db seeds the
232 integrity-checking hash for the data stream from the OR to the OP. Kf
233 is used to encrypt the stream of data going from the OP to the OR, and
234 Kb is used to encrypt the stream of data going from the OR to the OP.
236 4.3. Creating circuits
238 When creating a circuit through the network, the circuit creator
239 (OP) performs the following steps:
241 1. Choose an onion router as an exit node (R_N), such that the onion
242 router's exit policy does not exclude all pending streams
245 2. Choose a chain of (N-1) chain of N onion routers
246 (R_1...R_N-1) to constitute the path, such that no router
247 appears in the path twice.
249 3. If not already connected to the first router in the chain,
250 open a new connection to that router.
252 4. Choose a circID not already in use on the connection with the
253 first router in the chain; send a CREATE cell along the
254 connection, to be received by the first onion router.
256 5. Wait until a CREATED cell is received; finish the handshake
257 and extract the forward key Kf_1 and the backward key Kb_1.
259 6. For each subsequent onion router R (R_2 through R_N), extend
262 To extend the circuit by a single onion router R_M, the OP performs
265 1. Create an onion skin, encrypted to R_M's public key.
267 2. Send the onion skin in a relay EXTEND cell along
268 the circuit (see section 5).
270 3. When a relay EXTENDED cell is received, verify KH, and
271 calculate the shared keys. The circuit is now extended.
273 When an onion router receives an EXTEND relay cell, it sends a CREATE
274 cell to the next onion router, with the enclosed onion skin as its
275 payload. The initiating onion router chooses some circID not yet
276 used on the connection between the two onion routers. (But see
277 section 4.1. above, concerning choosing circIDs based on
278 lexicographic order of nicknames.)
280 As an extension (called router twins), if the desired next onion
281 router R in the circuit is down, and some other onion router R'
282 has the same public keys as R, then it's ok to extend to R' rather than R.
284 When an onion router receives a CREATE cell, if it already has a
285 circuit on the given connection with the given circID, it drops the
286 cell. Otherwise, after receiving the CREATE cell, it completes the
287 DH handshake, and replies with a CREATED cell. Upon receiving a
288 CREATED cell, an onion router packs it payload into an EXTENDED relay
289 cell (see section 5), and sends that cell up the circuit. Upon
290 receiving the EXTENDED relay cell, the OP can retrieve g^y.
292 (As an optimization, OR implementations may delay processing onions
293 until a break in traffic allows time to do so without harming
294 network latency too greatly.)
296 4.4. Tearing down circuits
298 Circuits are torn down when an unrecoverable error occurs along
299 the circuit, or when all streams on a circuit are closed and the
300 circuit's intended lifetime is over. Circuits may be torn down
301 either completely or hop-by-hop.
303 To tear down a circuit completely, an OR or OP sends a DESTROY
304 cell to the adjacent nodes on that circuit, using the appropriate
307 Upon receiving an outgoing DESTROY cell, an OR frees resources
308 associated with the corresponding circuit. If it's not the end of
309 the circuit, it sends a DESTROY cell for that circuit to the next OR
310 in the circuit. If the node is the end of the circuit, then it tears
311 down any associated edge connections (see section 5.1).
313 After a DESTROY cell has been processed, an OR ignores all data or
314 destroy cells for the corresponding circuit.
316 (The rest of this section is not currently used; on errors, circuits
317 are destroyed, not truncated.)
319 To tear down part of a circuit, the OP may send a RELAY_TRUNCATE cell
320 signaling a given OR (Stream ID zero). That OR sends a DESTROY
321 cell to the next node in the circuit, and replies to the OP with a
322 RELAY_TRUNCATED cell.
324 When an unrecoverable error occurs along one connection in a
325 circuit, the nodes on either side of the connection should, if they
326 are able, act as follows: the node closer to the OP should send a
327 RELAY_TRUNCATED cell towards the OP; the node farther from the OP
328 should send a DESTROY cell down the circuit.
330 4.5. Routing relay cells
332 When an OR receives a RELAY cell, it checks the cell's circID and
333 determines whether it has a corresponding circuit along that
334 connection. If not, the OR drops the RELAY cell.
336 Otherwise, if the OR is not at the OP edge of the circuit (that is,
337 either an 'exit node' or a non-edge node), it de/encrypts the payload
338 with AES/CTR, as follows:
339 'Forward' relay cell (same direction as CREATE):
340 Use Kf as key; decrypt.
341 'Back' relay cell (opposite direction from CREATE):
342 Use Kb as key; encrypt.
344 The OR then decides whether it recognizes the relay cell, by
345 inspecting the payload as described in section 5.1 below. If the OR
346 recognizes the cell, it processes the contents of the relay cell.
347 Otherwise, it passes the decrypted relay cell along the circuit if
348 the circuit continues. If the OR at the end of the circuit
349 encounters an unrecognized relay cell, an error has occurred: the OR
350 sends a DESTROY cell to tear down the circuit.
352 When a relay cell arrives at an OP, the OP decrypts the payload
353 with AES/CTR as follows:
354 OP receives data cell:
356 Decrypt with Kb_I. If the payload is recognized (see
357 section 5.1), then stop and process the payload.
359 For more information, see section 5 below.
361 5. Application connections and stream management
365 Within a circuit, the OP and the exit node use the contents of
366 RELAY packets to tunnel end-to-end commands and TCP connections
367 ("Streams") across circuits. End-to-end commands can be initiated
368 by either edge; streams are initiated by the OP.
370 The payload of each unencrypted RELAY cell consists of:
371 Relay command [1 byte]
372 'Recognized' [2 bytes]
378 The relay commands are:
392 The 'Recognized' field in any unencrypted relay payload is always
393 set to zero; the 'digest' field is computed as the first four bytes
394 of the running SHA-1 digest of all the bytes that have travelled
395 over this circuit, seeded from Df or Db respectively (obtained in
396 section 4.2 above), and including this RELAY cell's entire payload
397 (taken with the digest field set to zero).
399 When the 'recognized' field of a RELAY cell is zero, and the digest
400 is correct, the cell is considered "recognized" for the purposes of
401 decryption (see section 4.5 above).
403 All RELAY cells pertaining to the same tunneled stream have the
404 same stream ID. StreamIDs are chosen randomly by the OP. RELAY
405 cells that affect the entire circuit rather than a particular
406 stream use a StreamID of zero.
408 The 'Length' field of a relay cell contains the number of bytes in
409 the relay payload which contain real payload data. The remainder of
410 the payload is padded with NUL bytes.
412 5.2. Opening streams and transferring data
414 To open a new anonymized TCP connection, the OP chooses an open
415 circuit to an exit that may be able to connect to the destination
416 address, selects an arbitrary StreamID not yet used on that circuit,
417 and constructs a RELAY_BEGIN cell with a payload encoding the address
418 and port of the destination host. The payload format is:
420 ADDRESS | ':' | PORT | [00]
422 where ADDRESS can be a DNS hostname, or an IPv4 address in
423 dotted-quad format, or an IPv6 address surrounded by square brackets;
424 and where PORT is encoded in decimal.
426 [What is the [00] for? -NM]
427 [It's so the payload is easy to parse out with string funcs -RD]
429 Upon receiving this cell, the exit node resolves the address as
430 necessary, and opens a new TCP connection to the target port. If the
431 address cannot be resolved, or a connection can't be established, the
432 exit node replies with a RELAY_END cell. (See 5.4 below.)
433 Otherwise, the exit node replies with a RELAY_CONNECTED cell, whose
434 payload is the 4-byte IPv4 address or the 16-byte IPv6 address to which
435 the connection was made.
437 The OP waits for a RELAY_CONNECTED cell before sending any data.
438 Once a connection has been established, the OP and exit node
439 package stream data in RELAY_DATA cells, and upon receiving such
440 cells, echo their contents to the corresponding TCP stream.
441 RELAY_DATA cells sent to unrecognized streams are dropped.
443 Relay RELAY_DROP cells are long-range dummies; upon receiving such
444 a cell, the OR or OP must drop it.
448 When an anonymized TCP connection is closed, or an edge node
449 encounters error on any stream, it sends a 'RELAY_END' cell along the
450 circuit (if possible) and closes the TCP connection immediately. If
451 an edge node receives a 'RELAY_END' cell for any stream, it closes
452 the TCP connection completely, and sends nothing more along the
453 circuit for that stream.
455 The payload of a RELAY_END cell begins with a single 'reason' byte to
456 describe why the stream is closing, plus optional data (depending on
457 the reason.) The values are:
459 1 -- REASON_MISC (catch-all for unlisted reasons)
460 2 -- REASON_RESOLVEFAILED (couldn't look up hostname)
461 3 -- REASON_CONNECTREFUSED (remote host refused connection) [*]
462 4 -- REASON_EXITPOLICY (OR refuses to connect to host or port)
463 5 -- REASON_DESTROY (Circuit is being destroyed)
464 6 -- REASON_DONE (Anonymized TCP connection was closed)
465 7 -- REASON_TIMEOUT (Connection timed out, or OR timed out
467 8 -- (unallocated) [**]
468 9 -- REASON_HIBERNATING (OR is temporarily hibernating)
469 10 -- REASON_INTERNAL (Internal error at the OR)
470 11 -- REASON_RESOURCELIMIT (OR has no resources to fulfill request)
471 12 -- REASON_CONNRESET (Connection was unexpectedly reset)
472 13 -- REASON_TORPROTOCOL (Sent when closing connection because of
473 Tor protocol violations.)
475 (With REASON_EXITPOLICY, the 4-byte IPv4 address or 16-byte IPv6 address
476 forms the optional data; no other reason currently has extra data.)
478 OPs and ORs MUST accept reasons not on the above list, since future
479 versions of Tor may provide more fine-grained reasons.
481 [*] Older versions of Tor also send this reason when connections are
483 [**] Due to a bug in versions of Tor through 0095, error reason 8 must
484 remain allocated until that version is obsolete.
486 --- [The rest of this section describes unimplemented functionality.]
488 Because TCP connections can be half-open, we follow an equivalent
489 to TCP's FIN/FIN-ACK/ACK protocol to close streams.
491 An exit connection can have a TCP stream in one of three states:
492 'OPEN', 'DONE_PACKAGING', and 'DONE_DELIVERING'. For the purposes
493 of modeling transitions, we treat 'CLOSED' as a fourth state,
494 although connections in this state are not, in fact, tracked by the
497 A stream begins in the 'OPEN' state. Upon receiving a 'FIN' from
498 the corresponding TCP connection, the edge node sends a 'RELAY_FIN'
499 cell along the circuit and changes its state to 'DONE_PACKAGING'.
500 Upon receiving a 'RELAY_FIN' cell, an edge node sends a 'FIN' to
501 the corresponding TCP connection (e.g., by calling
502 shutdown(SHUT_WR)) and changing its state to 'DONE_DELIVERING'.
504 When a stream in already in 'DONE_DELIVERING' receives a 'FIN', it
505 also sends a 'RELAY_FIN' along the circuit, and changes its state
506 to 'CLOSED'. When a stream already in 'DONE_PACKAGING' receives a
507 'RELAY_FIN' cell, it sends a 'FIN' and changes its state to
510 If an edge node encounters an error on any stream, it sends a
511 'RELAY_END' cell (if possible) and closes the stream immediately.
513 5.4. Remote hostname lookup
515 To find the address associated with a hostname, the OP sends a
516 RELAY_RESOLVE cell containing the hostname to be resolved. (For a reverse
517 lookup, the OP sends a RELAY_RESOLVE cell containing an in-addr.arpa
518 address.) The OR replies with a RELAY_RESOLVED cell containing a status
519 byte, and any number of answers. Each answer is of the form:
522 Value (variable-width)
523 "Length" is the length of the Value field.
528 0xF0 -- Error, transient
529 0xF1 -- Error, nontransient
531 If any answer has a type of 'Error', then no other answer may be given.
533 The RELAY_RESOLVE cell must use a nonzero, distinct streamID; the
534 corresponding RELAY_RESOLVED cell must use the same streamID. No stream
535 is actually created by the OR when resolving the name.
541 Each node should do appropriate bandwidth throttling to keep its
544 Communicants rely on TCP's default flow control to push back when they
549 Currently nodes are not required to do any sort of link padding or
550 dummy traffic. Because strong attacks exist even with link padding,
551 and because link padding greatly increases the bandwidth requirements
552 for running a node, we plan to leave out link padding until this
553 tradeoff is better understood.
555 6.3. Circuit-level flow control
557 To control a circuit's bandwidth usage, each OR keeps track of
558 two 'windows', consisting of how many RELAY_DATA cells it is
559 allowed to package for transmission, and how many RELAY_DATA cells
560 it is willing to deliver to streams outside the network.
561 Each 'window' value is initially set to 1000 data cells
562 in each direction (cells that are not data cells do not affect
563 the window). When an OR is willing to deliver more cells, it sends a
564 RELAY_SENDME cell towards the OP, with Stream ID zero. When an OR
565 receives a RELAY_SENDME cell with stream ID zero, it increments its
568 Each of these cells increments the corresponding window by 100.
570 The OP behaves identically, except that it must track a packaging
571 window and a delivery window for every OR in the circuit.
573 An OR or OP sends cells to increment its delivery window when the
574 corresponding window value falls under some threshold (900).
576 If a packaging window reaches 0, the OR or OP stops reading from
577 TCP connections for all streams on the corresponding circuit, and
578 sends no more RELAY_DATA cells until receiving a RELAY_SENDME cell.
579 [this stuff is badly worded; copy in the tor-design section -RD]
581 6.4. Stream-level flow control
583 Edge nodes use RELAY_SENDME cells to implement end-to-end flow
584 control for individual connections across circuits. Similarly to
585 circuit-level flow control, edge nodes begin with a window of cells
586 (500) per stream, and increment the window by a fixed value (50)
587 upon receiving a RELAY_SENDME cell. Edge nodes initiate RELAY_SENDME
588 cells when both a) the window is <= 450, and b) there are less than
589 ten cell payloads remaining to be flushed at that edge.
591 7. Directories and routers
593 7.1. Extensible information format
595 Router descriptors and directories both obey the following lightweight
596 extensible information format.
598 The highest level object is a Document, which consists of one or more Items.
599 Every Item begins with a KeywordLine, followed by one or more Objects. A
600 KeywordLine begins with a Keyword, optionally followed by a space and more
601 non-newline characters, and ends with a newline. A Keyword is a sequence of
602 one or more characters in the set [A-Za-z0-9-]. An Object is a block of
603 encoded data in pseudo-Open-PGP-style armor. (cf. RFC 2440)
607 Document ::= (Item | NL)+
608 Item ::= KeywordLine Object*
609 KeywordLine ::= Keyword NL | Keyword SP ArgumentsChar+ NL
610 Keyword = KeywordChar+
611 KeywordChar ::= 'A' ... 'Z' | 'a' ... 'z' | '0' ... '9' | '-'
612 ArgumentChar ::= any printing ASCII character except NL.
613 Object ::= BeginLine Base-64-encoded-data EndLine
614 BeginLine ::= "-----BEGIN " Keyword "-----" NL
615 EndLine ::= "-----END " Keyword "-----" NL
617 The BeginLine and EndLine of an Object must use the same keyword.
619 When interpreting a Document, software MUST reject any document containing a
620 KeywordLine that starts with a keyword it doesn't recognize.
622 The "opt" keyword is reserved for non-critical future extensions. All
623 implementations MUST ignore any item of the form "opt keyword ....." when
624 they would not recognize "keyword ....."; and MUST treat "opt keyword ....."
625 as synonymous with "keyword ......" when keyword is recognized.
627 7.1. Router descriptor format.
629 Every router descriptor MUST start with a "router" Item; MUST end with a
630 "router-signature" Item and an extra NL; and MUST contain exactly one
631 instance of each of the following Items: "published" "onion-key" "link-key"
632 "signing-key" "bandwidth". Additionally, a router descriptor MAY contain any
633 number of "accept", "reject", "fingerprint", "uptime", and "opt" Items.
634 Other than "router" and "router-signature", the items may appear in any
637 The items' formats are as follows:
638 "router" nickname address (ORPort SocksPort DirPort)?
640 Indicates the beginning of a router descriptor. "address" must be an
641 IPv4 address in dotted-quad format. The Port values will soon be
642 deprecated; using them here is equivalent to using them in a "ports"
645 "ports" ORPort SocksPort DirPort
647 Indicates the TCP ports at which this OR exposes functionality.
648 ORPort is a port at which this OR accepts TLS connections for the main
649 OR protocol; SocksPort is the port at which this OR accepts SOCKS
650 connections; and DirPort is the port at which this OR accepts
651 directory-related HTTP connections. If any port is not supported, the
652 value 0 is given instead of a port number.
654 "bandwidth" bandwidth-avg bandwidth-burst bandwidth-observed
656 Estimated bandwidth for this router, in bytes per second. The
657 "average" bandwidth is the volume per second that the OR is willing
658 to sustain over long periods; the "burst" bandwidth is the volume
659 that the OR is willing to sustain in very short intervals. The
660 "observed" value is an estimate of the capacity this server can
661 handle. The server remembers the max bandwidth sustained output
662 over any ten second period in the past day, and another sustained
663 input. The "observed" value is the lesser of these two numbers.
665 [bandwidth-observed was not present before 0.0.8.]
669 A human-readable string describing the system on which this OR is
670 running. This MAY include the operating system, and SHOULD include
671 the name and version of the software implementing the Tor protocol.
673 "published" YYYY-MM-DD HH:MM:SS
675 The time, in GMT, when this descriptor was generated.
679 A fingerprint (20 byte SHA1 hash of asn1 encoded public key, encoded
680 in hex, with spaces after every 4 characters) for this router's
685 The number of seconds that this OR process has been running.
687 "onion-key" NL a public key in PEM format
689 This key is used to encrypt EXTEND cells for this OR. The key MUST
690 be accepted for at least XXXX hours after any new key is published in
691 a subsequent descriptor.
693 "signing-key" NL a public key in PEM format
695 The OR's long-term identity key.
700 These lines, in order, describe the rules that an OR follows when
701 deciding whether to allow a new stream to a given address. The
702 'exitpattern' syntax is described below.
704 "router-signature" NL Signature NL
706 The "SIGNATURE" object contains a signature of the PKCS1-padded SHA1
707 hash of the entire router descriptor, taken from the beginning of the
708 "router" line, through the newline after the "router-signature" line.
709 The router descriptor is invalid unless the signature is performed
710 with the router's identity key.
712 "dircacheport" port NL
714 Same as declaring "port" as this OR's directory port in the 'router'
715 line. At most one of dircacheport and the directory port in the router
716 line may be non-zero.
718 [Obsolete; will go away once 0.0.8 is dead. Older versions of Tor
719 did poorly when non-authoritative directories had a non-zero directory
720 port. To transition, Tor 0.0.8 used dircacheport for
721 nonauthoritative directories.]
725 Describes a way to contact the server's administrator, preferably
726 including an email address and a PGP key fingerprint.
730 'Names' is a space-separated list of server nicknames. If two ORs
731 list one another in their "family" entries, then OPs should treat
732 them as a single OR for the purpose of path selection.
734 For example, if node A's descriptor contains "family B", and node B's
735 descriptor contains "family A", then node A and node B should never
736 be used on the same circuit.
738 "read-history" YYYY-MM-DD HH:MM:SS (NSEC s) NUM,NUM,NUM,NUM,NUM... NL
739 "write-history" YYYY-MM-DD HH:MM:SS (NSEC s) NUM,NUM,NUM,NUM,NUM... NL
741 Declare how much bandwidth the OR has used recently. Usage is divided
742 into intervals of NSEC seconds. The YYYY-MM-DD HH:MM:SS field defines
743 the end of the most recent interval. The numbers are the number of
744 bytes used in the most recent intervals, ordered from oldest to newest.
746 nickname ::= between 1 and 19 alphanumeric characters, case-insensitive.
748 exitpattern ::= addrspec ":" portspec
749 portspec ::= "*" | port | port "-" port
750 port ::= an integer between 1 and 65535, inclusive.
751 addrspec ::= "*" | ip4spec | ip6spec
752 ipv4spec ::= ip4 | ip4 "/" num_ip4_bits | ip4 "/" ip4mask
753 ip4 ::= an IPv4 address in dotted-quad format
754 ip4mask ::= an IPv4 mask in dotted-quad format
755 num_ip4_bits ::= an integer between 0 and 32
756 ip6spec ::= ip6 | ip6 "/" num_ip6_bits
757 ip6 ::= an IPv6 address, surrounded by square brackets.
758 num_ip6_bits ::= an integer between 0 and 128
760 Ports are required; if they are not included in the router
761 line, they must appear in the "ports" lines.
763 7.2. Directory format
765 A Directory begins with a "signed-directory" item, followed by one each of
766 the following, in any order: "recommended-software", "published",
767 "router-status", "directory-signing-key". It may include any number of "opt"
768 items. After these items, a directory includes any number of router
769 descriptors, and a single "directory-signature" item.
773 Indicates the start of a directory.
775 "published" YYYY-MM-DD HH:MM:SS
777 The time at which this directory was generated and signed, in GMT.
779 "directory-signing-key"
781 The key used to sign this directory; see "signing-key" for format.
783 "recommended-software" comma-separated-version-list
785 A list of which versions of which implementations are currently
786 believed to be secure and compatible with the network.
788 "running-routers" space-separated-list
790 A description of which routers are currently believed to be up or
791 down. Every entry consists of an optional "!", followed by either an
792 OR's nickname, or "$" followed by a hexadecimal encoding of the hash
793 of an OR's identity key. If the "!" is included, the router is
794 believed not to be running; otherwise, it is believed to be running.
795 If a router's nickname is given, exactly one router of that nickname
796 will appear in the directory, and that router is "approved" by the
797 directory server. If a hashed identity key is given, that OR is not
798 "approved". [XXXX The 'running-routers' line is only provided for
799 backward compatibility. New code should parse 'router-status'
802 "router-status" space-separated-list
804 A description of which routers are currently believed to be up or
805 down, and which are verified or unverified. Contains one entry for
806 every router that the directory server knows. Each entry is of the
809 !name=$digest [Verified router, currently not live.]
810 name=$digest [Verified router, currently live.]
811 !$digest [Unverified router, currently not live.]
812 or $digest [Unverified router, currently live.]
814 (where 'name' is the router's nickname and 'digest' is a hexadecimal
815 encoding of the hash of the routers' identity key).
817 When parsing this line, clients should only mark a router as
818 'verified' if its nickname AND digest match the one provided.
819 [XXXX 'router-status' was added in 0.0.9pre5; older directory code
820 uses 'running-routers' instead.]
822 "directory-signature" nickname-of-dirserver NL Signature
824 Note: The router descriptor for the directory server MUST appear first.
825 The signature is computed by computing the SHA-1 hash of the
826 directory, from the characters "signed-directory", through the newline
827 after "directory-signature". This digest is then padded with PKCS.1,
828 and signed with the directory server's signing key.
830 If software encounters an unrecognized keyword in a single router descriptor,
831 it should reject only that router descriptor, and continue using the
832 others. If it encounters an unrecognized keyword in the directory header,
833 it should reject the entire directory.
835 7.3. Network-status descriptor
837 A "network-status" (a.k.a "running-routers") document is a truncated
838 directory that contains only the current status of a list of nodes, not
839 their actual descriptors. It contains exactly one of each of the following
846 "published" YYYY-MM-DD HH:MM:SS
854 "directory-signature" NL signature
858 7.4. Behavior of a directory server
860 lists nodes that are connected currently
861 speaks HTTP on a socket, spits out directory on request
863 Directory servers listen on a certain port (the DirPort), and speak a
864 limited version of HTTP 1.0. Clients send either GET or POST commands.
865 The basic interactions are:
866 "%s %s HTTP/1.0\r\nContent-Length: %lu\r\nHost: %s\r\n\r\n",
867 command, url, content-length, host.
868 Get "/tor/" to fetch a full directory.
869 Get "/tor/dir.z" to fetch a compressed full directory.
870 Get "/tor/running-routers" to fetch a network-status descriptor.
871 Post "/tor/" to post a server descriptor, with the body of the
872 request containing the descriptor.
874 "host" is used to specify the address:port of the dirserver, so
875 the request can survive going through HTTP proxies.
877 A.1. Differences between spec and implementation
879 - The current specification requires all ORs to have IPv4 addresses, but
880 allows servers to exit and resolve to IPv6 addresses, and to declare IPv6
881 addresses in their exit policies. The current codebase has no IPv6