- tor-dev - lists.torproject.org

TOR socket for P2P in Python
by Martin Neuenhofen 16 Aug '22

16 Aug '22

Dear Tor Developers, in my application, a client connects to a server via: *client_socket = socket.socket(socket.AF_INET, socket.SOCK_STREAM)client_socket.connect(ip,port)* I want to replace these two lines to create a client_socket whose IP address cannot be seen by the server. The application is e-polling. I looked into - torpy (too many timeouts), - socks (requires too much external configuration), - and I have tor installed (unknown how to use from within python). I just need a replacement for the two-liner code above. *The ideal solution:* is device & OS agnostic, portable, uses no peripherals or configuration (i.e., just one pip package to install), and works reliably. Thus, the below torpy solution would have been ideal if it had worked reliably: *from torpy import TorClient # pip install torpy* *class Torsocket:* * def __init__(self,ip,port): self.mgr1 = TorClient() self.tor = type(self.mgr1).__enter__(self.mgr1) self.mgr2 = self.tor.create_circuit(3) self.circuit = type(self.mgr2).__enter__(self.mgr2) self.mgr3 = self.circuit.create_stream((ip,port)) self.socket = type(self.mgr3).__enter__(self.mgr3) def send(self,data): self.socket.send(data) def recv(self,size): return self.socket.recv(size) def __del__(self): for bla in [self.mgr3,self.mgr2,self.mgr1]: type(bla).__exit__(bla, None, None, None)* It replaces the two-liner from the beginning with the below one-liner *client_socket = Torsocket(*server_data.server_address)* My questions/requests: - Is torpy connected to the real tor network (or is it a little toy twin)? If it is the real tor network then I will close this issue and open another one. - Are practical solutions available to my problem already?, and which of them works best for user-friendliness? - Can a pip package for a torsocket be made, as described above? i.e., something like torpy, but connecting to the real network? - Can I offer anything so that the ideal solution would be crafted and packaged?* *) If it helps, I could tell a lot about my project and how I believe it is significant and will be for the greater good of people. It is a fully implemented 1000LOC LSAG+WOT decentral P2P e-polling program with GUI that shall return trust and sovereignty into the people's hands. The only missing bit is a reliable tor connection.

2 3

bridge:// URI and QR codes
by meskio 10 Aug '22

10 Aug '22

Hello, We have a messy situation to share bridges between devices. BridgeDB and Tor Browser have QR codes but they use different format for them. There has being some attempts to standarize a bridge URI[0][1], but it looks like we never finished that work. Pier has produced an RFC of a bridge URI proposal: https://gitlab.torproject.org/tpo/applications/team/-/wikis/Design-Document… If you have an application that uses (or plan to use) bridge URIs or QR codes I'm interested on your take on that proposal. Do your application already have any kind of support for bridge URIs or QR codes? If so how do they look like? Will it break things if we change them? What do you think of the proposal? How can we improve it? Thank you. [0] https://gitlab.torproject.org/legacy/trac/-/issues/15035 [1] https://gitlab.torproject.org/tpo/applications/fenix/-/issues/40189 -- meskio | https://meskio.net/ -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- My contact info: https://meskio.net/crypto.txt -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- Nos vamos a Croatan.

5 8

Proposal 341: A better algorithm for out-of-sockets eviction
by Nick Mathewson 27 Jul '22

27 Jul '22

``` Filename: 341-better-oos.md Title: A better algorithm for out-of-sockets eviction Author: Nick Mathewson Created: 25 July 2022 Status: Open ``` # Introduction Our existing algorithm for handling an out-of-sockets condition needs improvement. It only handles sockets used for OR connections, and prioritizes those with more circuits. Because of these weaknesses, the algorithm is trivial to circumvent, and it's disabled by default with `DisableOOSCheck`. Here we propose a new algorithm for choosing which connections to close when we're out of sockets. In summary, the new algorithm works by deciding which kinds of connections we have "too many" of, and then by closing excess connections of each kind. The algorithm for selecting connections of each kind is different. # Intuitions behind the algorithm below We want to keep a healthy mix of connections running; favoring one kind of connection over another gives the attacker a fine way to starve the disfavored connections by making a bunch of the favored kind. The correct mix of connections depends on the type of service we are providing. Everywhere _except_ authorities, for example, inbound directory connections are perfectly fine to close, since nothing in our protocol actually generates them. In general, we would prefer to close DirPort connections, then Exit connections, then OR connections. The priority with which to close connections is different depending on the connection type. "Age of connection" or "number of circuits" may be a fine metric for how truly used an OR connection is, but for a DirPort connection, high age is suspicious. # The algorithm Define a "candidate" connection as one that has a socket, and is either an exit stream, an _inbound_ directory stream, or an OR connection. (Note that OR connections can be from clients, relays, or bridges. Note that ordinary relays should not get directory streams that use sockets, since clients always use `BEGIN_DIR` to create tunneled directory streams.) In all of the following, treat subtraction as saturating at zero. In other words, when you see "A - B" below, read it as "MAX(A-B, 0)". ## Phase 1: Deciding how many connections to close When we are find that we are low on sockets, we pick a number of sockets that we want to close according to our existing algorithm. (That is, we try to close 1/4 of our maximum sockets if we have reached our upper limit, or 1/10 of our maximum sockets if we have encountered a failure from socket(2).) Call this `N_CLOSE`. Then we decide which sockets to target based on this algorithm. 1. Consider the total number of sockets used for exit streams (`N_EXIT`), the total number used for _inbound_ directory streams (`N_DIR`), and the total number used for OR connections (`N_OR`). (In these calculations, we exclude connections that are already marked to be closed.) Call the total `N_CONN = N_DIR + N_OR + N_EXIT`. Define `N_RETAIN = N_CONN - N_CLOSE`. 2. Compute how many connections of each type are "in excess". First, calculate our target proportions: * If we are an authority, let `T_DIR` = 1. Otherwise set `T_DIR = 0.1`. * If we are an exit or we are running an onion service, let `T_EXIT = 2`. Otherwise let `T_EXIT = 0.1`. * Let `T_OR` = 1. > TODO: Should those numbers be consensus parameters? These numbers define the relative proportions of connections that we would be willing to retain retain in our final mix. Compute a number of _excess_ connections of each type by calculating. ``` T_TOTAL = T_OR + T_DIR + T_EXIT. EXCESS_DIR = N_DIR - N_RETAIN * (T_DIR / T_TOTAL) EXCESS_EXIT = N_EXIT - N_RETAIN * (T_EXIT / T_TOTAL) EXCESS_OR = N_OR - N_RETAIN * (T_OR / T_TOTAL) ``` 3. Finally, divide N_CLOSE among the different types of excess connections, assigning first to excess directory connections, then excess exit connections, and finally to excess OR connections. ``` CLOSE_DIR = MIN(EXCESS_DIR, N_CLOSE) N_CLOSE := N_CLOSE - CLOSE_DIR CLOSE_EXIT = MIN(EXCESS_EXIT, N_CLOSE) N_CLOSE := N_CLOSE - CLOSE_EXIT CLOSE_OR = MIN(EXCESS_OR, N_CLOSE) ``` We will try to close `CLOSE_DIR` directory connections, `CLOSE_EXIT` exit connections, and `CLOSE_OR` OR connections. ## Phase 2: Closing directory connections We want to close a certain number of directory connections. To select our targets, we sort first by the number of directory connections from a similar address (see "similar address" below) and then by their age, preferring to close the oldest ones first. > This approach defeats "many requests from the same address" and "Open > a connection and hold it open, and do so from many addresses". It > doesn't do such a great job with defeating "open and close frequently > and do so on many addresses." > Note that fallback directories do not typically use sockets for > handling directory connections: theirs are usually created with > BEGIN_DIR. ## Phase 3: Closing exit connections. We want to close a certain number of exit connections. To do this, we pick an exit connection at random, then close its circuit _along with all the other exit connections on the same circuit_. Then we repeat until we have closed at least our target number of exit connections. > This approach probabilistically favors closing circuits with a large > number of sockets open, regardless of how long those sockets have been > open. This defeats the easiest way of opening a large number of exit > streams ("open them all on once circuit") without making the > counter-approach ("open each exit stream on a its own circuit") much > more attractive. ## Phase 3: Closing OR connections. We want to close a certain number of OR connections, to clients, bridges, or relays. To do this, we first close OR connections with zero circuits. Then we close all OR connections but the most recent 2 from each "similar address". Then we close OR connections at random from among those _not_ to a recognized relay in the latest directory. Finally, we close OR connections at random. > We used to unconditionally prefer to close connections with fewer > circuits. That's trivial for an adversary to circumvent, though: they > can just open a bunch of circuits on their bogus OR connections, and > force us to preferentially close circuits from real clients, bridges, > and relays. > Note that some connections that seem like client connections ("not > from relays in the latest directory") are actually those created by > bridges. ## What is "A similar address"? We define two connections as having a similar address if they are in the same IPv4 /30, or if they are in the same IPv6 /90. # Acknowledgments This proposal was inspired by a set of [OOS improvements](https://gitlab.torproject.org/tpo/core/tor/-/issues/32794 ) from `starlight`.

2 1

Shrink Tor binary size
by Sergey Ponomarev 17 Jul '22

17 Jul '22

I’m working on a firmware for routers based on OpenWrt and it needs Tor out of the box for NAT punching i.e. SSH and Web admin access. It will expose a Single Onion Service i.e. not "hidden" with just 3 hop for a better performance. In fact it just needs to connect to some random relay to have a tunnel and encrypt/decrypt traffic. The full hidden service functionality with IP/RP dances is not needed: I'm not trying to hide the location of the router. In fact most users even don't need anonymity with the 3 hop to access their router but Tor requires 3. So this is probably a kind of abuse of the Tor network but this is the only free network of proxies. At the same time a user will have an encryption while accessing the router's admin panel. I may enable HTTPS with a self signed cert but users must accept it manually which is hard for inexperienced users. The Tor is already ported to OpenWrt but its binary is more than 2mb. For 16mb routers this is not critical but a lot and I need to keep space. So is it possible to compile Tor without some features? As far as I understood from autoconf I can disable Relay functionality. But maybe I can also disable some admin api parts and SOCKS proxy. Also OpenWrt out of the box has WolfSSL so is it possible to compile tor with it instead of OpenSSL? Another problem is files sizes inside of /var/lib/tor/: 20,442 cached-certs 2,303,443 cached-microdesc-consensus 8,725,598 cached-microdescs.new 3,531 state I think that all these files may be minimized. E.g. cached-certs contains keys in PEM format which is just Base64 form of DER. So switching to plain DER certs can significantly reduce size. As a side question: is it possible to make a small relay proxy that can work part time when I sleep? Imagine that each router already has a Tor. This is potentially thousands of relay nodes and all of them have a motivation to support the network which they are using themselves. But users don’t want to lose bandwidth. I see that I can set some traffic or bandwidth limits. But maybe I can write some script that will enable or disable the relay by schedule in the evening. As far I know relays must be always online so this will make the relay unstable and it won’t be used by Tor. Is it technically possible? Regards, Sergey Ponomarev, stokito.com

2 1

Want to build circuit via unpublished relay, can't add microdesc using control port
by ValdikSS 11 Jul '22

11 Jul '22

Hello everyone. I'm experimenting with Tor network and wanted to build a circuit via unpublished relay (PublishServerDescriptor 0). To do so, I set up a relay, got its authority descriptor, imported it with +POSTDESCRIPTOR in the client using control port and tried to build a circuit using its fingerprint with EXTENDCIRCUIT. Everything works fine with UseMicrodescriptors 0, however with MicroDescriptors activated, the client does not detect imported (full) descriptor, returning: 552 No descriptor for "fingerprint here" I found no way to import microdescriptor using control port and no other means to use unpublished nodes. How can I add microdesc which is not in consensus into router? Or how can I build circuit with a relay not in consensus otherwise?

3 2

Tor can't read HiddenServicePort unix socket through group permissions when starting as root and using setgid?
by keyandthegate 06 Jul '22

06 Jul '22

Permissions are set so tor should be able to access through the `postfix-test-queue` user: > $ sudo ls -l /var/spool/postfix-test/public/smtpd > srw-rw-rw- 1 postfix-test postfix-test 0 █████ /var/spool/postfix-test/public/smtpd > $ sudo ls -l /var/spool/postfix-test > # ... > drwx--x--- 2 postfix-test postfix-test-queue 4096 █████ public > $ sudo -u _tor-test id > uid=130(_tor-test) gid=141(_tor-test) groups=141(_tor-test),1006(postfix-test-queue) > $ cat /etc/tor/instances/test/torrc | grep HiddenServicePort > HiddenServicePort 25 unix:/var/spool/postfix-test/public/smtpd > $ cat /run/tor-instances/test.defaults | grep User > User _tor-test Running `tor@test` via the default systemctl config shows: > $ ps -ax -o uid,gid,supgid,command | grep /usr/bin/tor > 130 141 141 /usr/bin/tor --defaults-torrc /run/tor-instances/test.defaults -f /etc/tor/instances/test/torrc Which is missing the `postfix-test-queue` `1006` user which, for example shows up if I do: > $ sudo -u _tor-test sleep 1000 & ps -ax -o uid,gid,supgid,command | grep sleep > [1] 132314 > 0 141 141,1006 sudo -u _tor-test sleep 1000 Connecting using `sudo -u` works (the message indicates successful connection): > $ sudo -u _tor-test curl --unix-socket /var/spool/postfix-test/public/smtpd http://localhost > curl: (1) Received HTTP/0.9 when not allowed But connecting via tor does not: > $ torsocks --ipv6 curl http://█████.onion:25 > █████ ERROR torsocks[134873]: Host unreachable (in socks5_recv_connect_reply() at socks5.c:539) > curl: (7) Couldn't connect to server But does if I allow access to the socket to everyone: > $ sudo chmod "o+x" /var/spool/postfix-test/public/ > $ torsocks --ipv6 curl http://█████.onion:25 > curl: (1) Received HTTP/0.9 when not allowed Tor's relevant source code: [Tor: lib/process/setuid.c Source File](https://tpo.pages.torproject.net/core/doc/tor/setuid_8c_source.html)

1 1

Bootstrapping with a very inaccurate clock
by Michael Rogers 28 Jun '22

28 Jun '22

Hello again, Apologies for starting two threads in quick succession. First of all I want to say thank you for all the improvements in handling clock skew between 0.3.5 and 0.4.5. We recently migrated Briar to 0.4.5, and some situations that are common on mobile, such as the system clock being configured with the wrong timezone, are now much less likely to affect Tor connectivity (except when using obfs4). When the system clock is very far behind real time, Tor's clock skew events enable us to warn the user to check their time and date settings. When the clock is ahead of real time by up to a day, there are no clock skew events but Tor can bootstrap. But when the clock is several days ahead of real time, there are no clock skew events and Tor gets stuck boostrapping at 14%. At the moment we're having trouble distinguishing this situation from other situations that might cause bootstrapping to get stuck, such as access to the Tor network being blocked. So ideally we'd like to be able to detect that the clock is the problem and ask the user to fix it. I should say that this is probably a rare situation. Old devices that haven't been turned on for a while sometimes reset their clocks to a point in the distant past, but I've never seen a device spontaneously change its clock to a point in the distant future. Still, people do mess around with the time and date settings so we'd like to detect and fix this issue if possible. Is there any way for Tor to tell, during bootstrapping, that the system clock appears to be far ahead of real time? Thanks, Michael

1 0

Metrics: Estimating fraction of reported directory-request statistics
by David Fifield 27 Jun '22

27 Jun '22

I am trying to reproduce the "frac" computation from the Reproducible Metrics instructions: https://metrics.torproject.org/reproducible-metrics.html#relay-users Which is also Section 3 in the tech report on counting bridge users: https://research.torproject.org/techreports/counting-daily-bridge-users-201… h(R^H) * n(H) + h(H) * n(R\H) frac = ----------------------------- h(H) * n(N) My minor goal is to reproduce the "frac" column from the Metrics web site (which I assume is the same as the frac above, expressed as a percentage): https://metrics.torproject.org/userstats-relay-country.csv?start=2022-04-01… date,country,users,lower,upper,frac 2022-04-01,,2262557,,,92 2022-04-02,,2181639,,,92 2022-04-03,,2179544,,,93 2022-04-04,,2350360,,,93 2022-04-05,,2388772,,,93 2022-04-06,,2356170,,,93 2022-04-07,,2323184,,,93 2022-04-08,,2310170,,,91 I'm having trouble with the computation of n(R\H) and h(R∧H). I understand that R is the subset of relays that report directory request counts (i.e. that have dirreq-stats-end in their extra-info descriptors) and H is the subset of relays that report directory request byte counts (i.e. that have dirreq-write-history in their extra-info descriptors). R and H partially overlap: there are relays that are in R but not H, others that are in H but not R, and others that are in both. The computations depend on some values that are directly from descriptors: n(R) = sum of hours, for relays with directory request counts n(H) = sum of hours, for relays with directory write histories h(H) = sum of written bytes, for relays with directory write histories > Compute n(R\H) as the number of hours for which responses have been > reported but no written directory bytes. This fraction is determined > by summing up all interval lengths and then subtracting the written > directory bytes interval length from the directory response interval > length. Negative results are discarded. I interpret this to mean: add up all the dirrect-stats-end intervals (this is n(R)), add up all the dirreq-write-history intervals (this is n(H)), and compute n(R\H) as n(R) − n(H). This seems wrong: it would only be true when H is a subset of R. > Compute h(R∧H) as the number of written directory bytes for the > fraction of time when a server was reporting both written directory > bytes and directory responses. As above, this fraction is determined > by first summing up all interval lengths and then computing the > minimum of both sums divided by the sum of reported written directory > bytes. This seems to be saying to compute h(R∧H) (a count of bytes) as min(n(R), n(H)) / h(H). This is dimensionally wrong: the units are hours / bytes. What would be more natural to me is min(n(R), n(H)) / max(n(R), n(H)) × h(H); i.e., divide the smaller of n(R) and n(R) by the larger, then multiply this ratio by the observable byte count. But this, too, only works when H is a subset of R. Where is this computation done in the metrics code? I would like to refer to it, but I could not find it. Using the formulas and assumptions above, here's my attempt at computing recent "frac" values: date `n(N)` `n(H)` `h(H)` `n(R)` `n(R\H)` `h(R∧H)` frac 2022-04-01 166584 177638. 2.24e13 125491. 0 1.59e13 0.753 2022-04-02 166951 177466. 2.18e13 125686. 0 1.54e13 0.753 2022-04-03 167100 177718. 2.27e13 127008. 0 1.62e13 0.760 2022-04-04 166970 177559. 2.43e13 126412. 0 1.73e13 0.757 2022-04-05 166729 177585. 2.44e13 125389. 0 1.72e13 0.752 2022-04-06 166832 177470. 2.39e13 127077. 0 1.71e13 0.762 2022-04-07 166532 177210. 2.48e13 127815. 0 1.79e13 0.768 2022-04-08 167695 176879. 2.52e13 127697. 0 1.82e13 0.761 The "frac" column does not match the CSV. Also notice that n(N) < n(H), which should be impossible because H is supposed to be a subset of N (N is the set of all relays). But this is what I get when I estimate n(N) from a network-status-consensus-3 and n(H) from extra-info documents. Also notice that n(R) < n(H), which means that H cannot be a subset of R, contrary to the observations above.

2 4

tor-browser-build : any way to get it to use system packages for building?
by Jason Vas Dias 15 Jun '22

15 Jun '22

Good day - I recently attempted to check-out and build tor-browser-build - MUST it download ALL the gits, if system packages are found to satisfy version dependencies ? I just don't have enough storage / network bandwidth . Why am I downloading the llvm source when I already have llvm and clang 13 (-devel packages also) on my Fedora 36 system ? Is there any interest in producing a version of tor-browser-build to use installed system packages if they meet version requirements instead of checking out GITs & building for ALL dependencies ? It could be enabled with a USE_SYSTEM_PKGS=1 make option or something. I'm going to have to develop this if I want to work on tor-browser, which I do want to do. Best Regards, Jason

3 3

Proposal 340: Packed and fragmented relay messages
by Nick Mathewson 31 May '22

31 May '22

``` Filename: 340-packed-and-fragmented.md Title: Packed and fragmented relay messages Author: Nick Mathewson Created: 31 May 2022 Status: Open ``` # Introduction Tor sends long-distance messages on circuits via _relay cells_. The current relay cell format allows one _relay message_ (e.g., "BEGIN" or "DATA" or "END") per relay cell. We want to relax this 1:1 requirement, between messages and cells, for two reasons: * To support relay messages that are longer than the current 498-byte limit. Applications would include wider handshake messages for postquantum crypto, UDP messages, and SNIP transfer in walking onions. * To transmit small messages more efficiently. Several message types (notably `SENDME`, `XON`, `XOFF`, and several types from [proposal 329](./329-traffic-splitting.txt)) are much smaller than the relay cell size, and could be sent comparatively often. In this proposal, we describe a way to decouple relay cells from relay messages. Relay messages can now be packed into multiple cells or split across multiple cells. This proposal combines ideas from [proposal 319](./319-wide-everything.md) (fragmentation) and [proposal 325](./325-packed-relay-cells.md) (packed cells). It requires [ntor v3](./332-ntor-v3-with-extra-data.md) and prepares for [next-generation relay cryptography](./308-counter-galois-onion). ## A preliminary change: Relay encryption, version 1.5 We are fairly sure that, whatever we do for our next batch of relay cryptography, we will want to increase the size of the data used to authenticate relay cells to 128 bits. (Currently it uses a 4-byte tag plus 2 bytes of zeros.) To avoid proliferating formats, I'm going to suggest that we make the other changes in this proposal changes concurrently with a change in our relay cryptography, so that we do not have too many incompatible cell formats going on at the same time. The new format for a decrypted relay _cell_ will be: recognized [2 bytes] digest [14 bytes] body [509 - 16 = 493 bytes] "Digest" and "recognized" are computed as before; the only difference is that they occur _before_ the rest of the cell, and that "digest" is truncated to 14 bytes instead of 4. If we are lucky, we won't have to build this encryption at all, and we can just move to some version of GCM-UIV or other RPRP that reserves 16 bytes for an authentication tag or similar cryptographic object. ## New relay message packing Relay _messages_ now follow the following format: Header command u8 length u16 stream_id u16 Body data u8[length] We require that "command" is never 0. Messages can be split across relay cells; multiple messages can occur in a single relay cell. We enforce the following rules: * Headers may not be split across cells. * If a 0 byte follows any relay message, there are no more messages in that cell. * A relay cell may not be "empty": it must have at least some part of some message. Unless specified elsewhere, **all** message types may be packed, and **all** message types may be fragmented. Every command has an associated maximum length for its messages. If not specified elsewhere, the maximum length for every message is 498 bytes (for legacy reasons). Receivers MUST validate that headers are well-formed and have valid lengths while handling the cell in which the header is encoded. If the header is invalid, the receiver must destroy the circuit. ### Some examples ## Negotiation This message format requires a new `Relay` subprotocol version to indicate support. If a client wants to indicate support for this format, it sends the following extension as part of its `ntor3` handshake: RELAY_PROTOCOL version u8 The `version` field is the `Relay` subprotocol version that the client wants to use. The relay must send back the same extension in its ntor3 handshake to acknowledge support. ## Migration We add a consensus parameter, "streamed-relay-messages", with default value 0, minimum value 0, and maximum value 1. If this value is 0, then clients will not (by default) negotiate this relay protocol. If it is 1, then clients will negotiate it when relays support it. For testing, clients can override this setting. Once enough relays support this proposal, we'll change the consensus parameter to 1. Later, we'll change the default to 1 as well. ## Packing decisions We specify the following greedy algorithm for making decisions about fragmentation and packing. Other algorithms are possible, but this one is fairly simple, and using it will help avoid distinguishability issues: Whenever a client or relay is about to send a cell that would leave at least 32 bytes unused in a relay cell, it checks to see whether there is any pending data to be sent in the same circuit (in a data cell). If there is, then it adds a DATA message to the end of the current cell, with as much data as possible. Otherwise, the client sends the cell with no packed data. ## Onion services Negotiating this for onion services will happen in a separate proposal; it is not a current priority, since there is nothing sent over rendezvous circuits that we currently need to fragment or pack. ## Miscellany ### Handling `RELAY_EARLY` The `RELAY_EARLY` status for a command is determined based on the relay cell in which the command's _header_ appeared. ### Handling `SENDME`s SENDME messages may not be fragmented; the body and the command must appear in the same cell. (This is necessary so authenticated sendmes can have a reasonable implementation.) ### An exception for `DATA`. Data messages may not be fragmented. (There is never a reason to do this.) ### Extending message-length maxima For now, the maximum length for every message is 498 bytes, except as follows: - `DATAGRAM` messages (see proposal 339) have a maximum body length of 1967 bytes. (This works out to four relay cells, and accommodates most reasonable MTU choices) Any increase in maximum length for any other message type requires a new Relay subprotocol version. (For example, if we later want to allow EXTEND2 messages to be 2000 bytes long, we need to add a new proposal saying so, and reserving a new subprotocol version.) # Appendix: Example cells Here is an example of the simplest case: one message, sent in one relay cell. Here it is a BEGIN message. ``` Cell 1: Cell header circid .. [4 bytes] command RELAY [1 byte] relay cell header recognized 0 [2 bytes] digest (...) [14 bytes] message header: command BEGIN [1 byte] length 23 [2 bytes] stream_id (...) [2 bytes] message body "www.torproject.org:443\0" [23 bytes] end-of-messages marker 0 [1 byte] padding up to end of cell random [464 bytes] ``` Here's an example with fragmentation only: a large EXTEND2 message split across two relay cells. ``` Cell 1: Cell header circid .. [4 bytes] command RELAY_EARLY [1 byte] relay cell header recognized 0 [2 bytes] digest (...) [14 bytes] message header: command EXTEND [1 byte] length 800 [2 bytes] stream_id 0 [2 bytes] message body (extend body, part 1) [488 bytes] Cell 2: Cell header circid .. [4 bytes] command RELAY [1 byte] relay cell header recognized 0 [2 bytes] digest (...) [14 bytes] message body, continued (extend body, part 2) [312 bytes] end-of-messages marker 0 [1 byte] padding up to end of cell random [180 bytes] ``` Here is an example with packing only: A begin_dir message and a data message in the same cell. ``` Cell 1: Cell header circid .. [4 bytes] command RELAY [1 byte] relay cell header recognized 0 [2 bytes] digest (...) [14 bytes] message header: command BEGIN_DIR [1 byte] length 0 [2 bytes] stream_id 32 [2 bytes] message body: (empty) --- [0 bytes] message header: command DATA [1 byte] length 25 [2 bytes] stream_id 32 [2 bytes] message body: "HTTP/1.0 GET /tor/foo\r\n\r\n" [25 bytes] end-of-messages marker 0 [1 byte] padding up to end of cell random [457 bytes] ``` Here is an example with packing and fragmentation: a large DATAGRAM cell, a SENDME cell, and an XON cell. (Note that this sequence of cells would not actually be generated by the algorithm described in "Packing decisions" above; this is only an example of what parties need to accept.) ``` Cell 1: Cell header circid .. [4 bytes] command RELAY [1 byte] relay cell header recognized 0 [2 bytes] digest (...) [14 bytes] message header: command DATAGRAM [1 byte] length 1200 [2 bytes] stream_id 99 [2 bytes] message body (datagram body, part 1) [488 bytes] Cell 2: Cell header circid .. [4 bytes] command RELAY [1 byte] relay cell header recognized 0 [2 bytes] digest (...) [14 bytes] message body, continued (datagram body, part 2) [493 bytes] Cell 3: Cell header circid .. [4 bytes] command RELAY [1 byte] relay cell header recognized 0 [2 bytes] digest (...) [14 bytes] message body, continued (datagram body, part 3) [219 bytes] (488+493+219=1200) message header: command SENDME [1 byte] length 23 [2 bytes] stream_id 0 [2 bytes] message body: version 1 [1 byte] datalen 20 [2 bytes] data (digest to ack) [20 bytes] message header: command XON [1 byte] length 1 [2 bytes] stream_id 50 [2 bytes] message body: version 1 [1 byte] end-of-messages marker 0 [1 byte] padding up to end of cell random [239 bytes] ```

1 0