aboutsummaryrefslogtreecommitdiff
path: root/crates/tor-proto/src
Commit message (Collapse)AuthorAgeFilesLines
...
| * Define and implement a backend padding API, based on maybenot.Nick Mathewson2025-09-025-0/+1336
| | | | | | | | | | | | | | | | There are incomplete pieces, marked with "TODO circpad". There is no integration into the circuit reactor code yet. Part of #63
* | Merge branch 'chan-handshake-refactor' into 'main'David Goulet2025-09-021-121/+262
|\ \ | |/ |/| | | | | proto: Refactor channel handshake code for upcoming relay handshake See merge request tpo/core/arti!3182
| * proto: Move is_authenticating() into the base initiator handshake traitDavid Goulet2025-09-021-7/+10
| | | | | | | | | | | | | | | | Initiator always know if they will authenticate or not. Responder is different as a relay doesn't know until the end of the handshake if it is responding to a relay or a client. Signed-off-by: David Goulet <[email protected]>
| * proto: Various channel handshake fixesDavid Goulet2025-09-021-21/+38
| | | | | | | | | | | | | | | | | | | | These are following the review of MR 3182. They are put in a single commit because the git absorb has a large amount of conflicts on rebase and this commit allows the reviewers to see what happened. The base branch was rebased on main due to the need for 3184. Signed-off-by: David Goulet <[email protected]>
| * proto: Fix unit tests after changesDavid Goulet2025-09-021-5/+5
| | | | | | | | Signed-off-by: David Goulet <[email protected]>
| * proto: Add a helper to calculate the handshake clock skewDavid Goulet2025-09-021-12/+33
| | | | | | | | | | | | In the spirit of avoidin code duplication. Signed-off-by: David Goulet <[email protected]>
| * proto: Add a channel initiator handshake base traitDavid Goulet2025-09-021-76/+118
| | | | | | | | | | | | | | | | | | All initiator handshake will implement this in order to get access to the helper function to receive the relay responder cells. Relay will implement this in future commits. Signed-off-by: David Goulet <[email protected]>
| * proto: Introduce a ChannelBaseHandshake traitDavid Goulet2025-09-021-37/+96
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Client and relay handhsake share a lot of code because they both send/recv the same cells, just handles them differently for verification. This is the base trait for all handshake implementing basic getters and VERSIONS cell handling. This will allow the RelayInitiatorHandshake and RelayResponderHandshake to use this common code. See, traits are fun. Win-win-win. Signed-off-by: David Goulet <[email protected]>
| * proto: Put in a ChannelFrame<T> into the client handshakeDavid Goulet2025-09-021-10/+9
| | | | | | | | | | | | | | | | | | | | | | | | The ClientInitiatorHandshake holds a "tls" sink but the very first thing we do is transform it to a ChannelFrame<T>. Instead, just store the frame to the object directly so we can then use a channel frame uniformily accross its lifetime. This will be useful for the future refactoring paving the way for relay channel authentication. Signed-off-by: David Goulet <[email protected]>
* | tor-proto: Switched to "assert in const" patternhashcatHitman2025-09-011-2/+1
|/ | | | | | | Replaced use of [`static_assertions::const_assert`] with the newly available "assert in const" pattern. Signed-off-by: hashcatHitman <[email protected]>
* proto: Use allow(unused) instead of expect(unused).Gabriela Moldovan2025-08-291-1/+1
| | | | | | | | | | | With rustc 1.85, this `expect(unused)` triggers a "lint expectation unfulfilled" error, because rust thinks the struct *is* used (because of the d-d `TryFrom` implementation), despite the fact that it's never actually constructed. This isn't a problem on newer versions (it's correctly identified as unused/dead code on 1.89, for example). Using `allow` instead of `expect` is slightly less nice, but it makes `clippy` pass on 1.85 too.
* proto: Move TunnelId to a separate, shared module (fmt).Gabriela Moldovan2025-08-286-6/+6
|
* proto: Move TunnelId to a separate, shared module.Gabriela Moldovan2025-08-2810-58/+65
| | | | | | | | | | | | | The `TunnelId*` types will be reused in the relay reactor (exit relays need to have the concept of a "tunnel ID" because of conflux). Now the `relay::reactor` module only has a single import from `client` (for the `unwrap_or_shutdown` helper, which we should be able to remove soon). From now, we will avoid importing anything from `client` in the `relay` module, and instead prefer refactoring the code as needed (to pull the implementation-agnostic parts outside of `client`). This commit has no functional changes, just code motion.
* proto: Add a circuit module shared between client and relay impls.Gabriela Moldovan2025-08-2818-33/+39
| | | | | | | This is just code motion (I suggest reviewing with `--color-moved`). This also moves the implementation-agnostic parts from `tor_proto::client::circuit` to a new `tor_proto::circuit` module.
* Merge branch 'relay-chan-msg' into 'main'gabi-2502025-08-281-31/+72
|\ | | | | | | | | proto: Add a new RelayCircChanMsg message subclass. See merge request tpo/core/arti!3198
| * proto: Avoid referring to restricted ChanMsgs as "subclasses".Gabriela Moldovan2025-08-281-5/+5
| |
| * proto: Add a new RelayCircChanMsg message subclass.Gabriela Moldovan2025-08-281-0/+43
| | | | | | | | | | This will be used to restrict the types of messages that can be sent on the relay-specific channels.
| * proto: Derive ChanMsgSubclass for CreateResponse, ClientCircChanMsg.Gabriela Moldovan2025-08-271-32/+5
| | | | | | | | | | This enables us to remove the open-coded implementations in favor of the derived version.
| * proto: Add d-d macro for creating AnyChanMsg subclasses.Gabriela Moldovan2025-08-271-1/+26
| | | | | | | | | | | | The code for generating these is repetitive (see `CreateResponse` and `ClientChanMsg`), and we will soon need a `RelayChanMsg` type too, so now is a good time to introduce a helper for generating the boilerplate.
* | tor-proto: don't allow consecutive XOFF messagesSteven Engler2025-08-271-1/+14
|/
* proto: Only allow VERSIONS cell for the new handshake stateDavid Goulet2025-08-272-51/+20
| | | | | | | | | Due to this, it is not possible to get a VPADDING before because it requires a link protocol version to decideon the encoding: https://gitlab.torproject.org/tpo/core/torspec/-/issues/366 Signed-off-by: David Goulet <[email protected]>
* Merge branch 'arti-p112-docs' into 'main'David Goulet2025-08-213-0/+40
|\ | | | | | | | | proto: tweak docs to say where the prop349 checks are implemented See merge request tpo/core/arti!3171
| * proto: Indent line in docs to satisfy clippy.Gabriela Moldovan2025-08-211-1/+1
| |
| * proto: Add prop349 note about the resolve stream handler.Gabriela Moldovan2025-08-201-0/+14
| |
| * proto: Document how handle_meta_cell can cause circuit teardown.Gabriela Moldovan2025-08-201-0/+14
| |
| * proto: Add docs about the lifecycle of MetaCellHandlers.Gabriela Moldovan2025-08-201-0/+12
| |
* | proto: Remove the AUTHORIZE as a parsable cellDavid Goulet2025-08-213-24/+8
| | | | | | | | | | | | | | | | | | | | | | | | | | The AUTHORIZE cell command is simply reserved but not defined. The tor specification, at this point in time, is allowing such cell before the handshake starts but it is very unclear on what ordering is allowed nor how many can are allowed. C-tor silents drop them like VPADDING and so clearly unused. Instead of dealing with it, simply remove its support but keeping its reserved number. Signed-off-by: David Goulet <[email protected]>
* | proto: Add a channel handler comment and a fixDavid Goulet2025-08-211-3/+8
| | | | | | | | Signed-off-by: David Goulet <[email protected]>
* | proto: Return AUTHORIZE, VPADDING and VERSIONS at handshakeDavid Goulet2025-08-212-24/+65
| | | | | | | | | | | | | | | | When starting a handshake, we were only expecting a VERSIONS which is not what the protocol say. An AUTHORIZE and VPADDING can arrive before a VERSIONS. Signed-off-by: David Goulet <[email protected]>
* | proto: Allow padding in all channel message setsDavid Goulet2025-08-211-3/+13
| | | | | | | | | | | | | | | | | | | | | | | | | | | | First of all, VPADDING has been added in link protocol version 3 so it was missing from v4. Second, after closely looking at C-tor and the spec, it appears that we allow VPADDING at any point on a channel which should simply be silently dropped. Any number in any order. Third, couple sets were missing the PADDING cell which is only allowed on an open channel. Signed-off-by: David Goulet <[email protected]>
* | tor-proto: simplify some match statementsSteven Engler2025-08-201-46/+26
| |
* | proto: Change (crate) to (super) for all objects in msg.rsDavid Goulet2025-08-201-25/+25
| | | | | | | | Signed-off-by: David Goulet <[email protected]>
* | proto: Cleanup allow(unused)David Goulet2025-08-202-4/+0
| | | | | | | | Signed-off-by: David Goulet <[email protected]>
* | proto: Unit tests for channel handlerDavid Goulet2025-08-201-1/+107
| | | | | | | | Signed-off-by: David Goulet <[email protected]>
* | proto: Rename OutboundClientHandshakeDavid Goulet2025-08-202-11/+11
| | | | | | | | | | | | | | | | | | Use the specification terminology which is also the same for ChannelType. Part of #1597 Signed-off-by: David Goulet <[email protected]>
* | proto: Make the OutboundClientHandshake use new cell handlerDavid Goulet2025-08-206-211/+97
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Use the ChannelFrame<> for the entirety of the outbound client handshake that is the ClientInitiator channel type. With this change, the codec.rs code is not needed anymore along its CodecError as well which has been normalized onto the crate::Error instead in order to simplify error handling and avoid duplication of error types. Unit tests have been modified to reflect this change of what can be done with a channel frame. Also renamed to focus on client behavior. Part of #1597 Signed-off-by: David Goulet <[email protected]>
* | proto: Add helper functions/type for cell handlingDavid Goulet2025-08-201-0/+20
| | | | | | | | | | | | | | | | | | This type and functions will be used in the handshake process in future commits. Part of #1597 Signed-off-by: David Goulet <[email protected]>
* | proto: Add channel cell handlerDavid Goulet2025-08-202-0/+562
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The handler.rs file contains a generic "ChannelCellHandler" which is split into three different handler depending of the channel state (new, handshaking or open). These handlers implement Encoder/Decoder so we can give a ChannelCellHandler to a asynchronous_codec::Framed along a TLS stream. That cell handler is also in charge of tracking the CLOG/SLOG (see tor-spec), running digest of cells seen, which is used to authenticate a channel for the Relay <-> Relay case. This ChannelCellHandler auto transitions as the setters function are used. The handshake code will use this to advance the handler. Each handler uses a MessageFilter from msg.rs in order to allow or not to return the message. A keen eye will notice that we can avoid encoding a message if we don't need but we will decode all possible messages and only then allow it or not. The channel cell handler is not used at this commit. Part of #1597 Signed-off-by: David Goulet <[email protected]>
* | proto: Add message filtering to restricted message setsDavid Goulet2025-08-201-2/+361
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This adds the code in msg.rs to be able to filter an inbound or outbound message on a channel. Each link protocol version implement a "is_allowed()" which is quite verbose and tests each possibilities for human readability. Then, we have several small struct/enum that are used to describe how a message is filtered. It is still unused at this commit. Part of #1597 Signed-off-by: David Goulet <[email protected]>
* | proto: Implement a From<std::io::Error> for ErrorDavid Goulet2025-08-201-0/+6
| | | | | | | | | | | | | | | | | | To be able to return a crate::Error from the Decoded/Encoder trait, it needs to implement this conversion. Part of #1597 Signed-off-by: David Goulet <[email protected]>
* | proto: Add restricted channel message setsDavid Goulet2025-08-202-3/+211
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Add the msg.rs file containing all the allowed message sets based on the channel type and direction. They are also namespaced by link protocol version. Unused at this commit. They will be used by the channel reactor along the channel type and link protocol version in order to know if the message is allowed or not. See is_allowed() helper function in this commit. Part of #1597 Signed-off-by: David Goulet <[email protected]>
* | proto: Add ChannelType enumDavid Goulet2025-08-203-10/+65
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The ChannelType indicates the type of channel in order to dictate which message is allowed on it. The value use the Initiator and Responder terminology from tor-spec documents. At this commit, we only have client channel meaning the "ClientInitiator" type. In future commits, the channel type will be used by the channel reactor to restrict which message is allowed or not. Part of #1597 Signed-off-by: David Goulet <[email protected]>
* | chan: Rename channel launch to launch_clientDavid Goulet2025-08-202-2/+3
|/ | | | Signed-off-by: David Goulet <[email protected]>
* tor-proto: add comments to `CC_XOFF_CLIENT`Steven Engler2025-08-181-0/+29
| | | | | | This tries to explain that the amount of incoming data we choose to buffer on an arti stream doesn't really matter for arti's socks proxy, since the amount of data buffered by the kernel is significantly higher.
* misc: cleanup now that `_report!` macros support fieldsSteven Engler2025-08-181-3/+2
|
* tor-proto: report tunnel/channel id as a fieldSteven Engler2025-08-182-4/+4
| | | | This restores the pre-374889d34aa0 behaviour.
* Merge branch 'relay-reactor-placeholder' into 'main'David Goulet2025-08-1839-138/+464
|\ | | | | | | | | proto: Add a placeholder for the relay reactor. See merge request tpo/core/arti!3162
| * proto: Fix tests post-crate reorg (fmt).Gabriela Moldovan2025-08-182-2/+2
| |
| * proto: Fix tests post-crate reorg.Gabriela Moldovan2025-08-182-2/+2
| |
| * proto: Avoid calling the relay reactor a tunnel reactor.Gabriela Moldovan2025-08-181-3/+3
| | | | | | | | | | A "tunnel" is a higher level concept we'll want to avoid using from now on when talking about the proto implementation.
generated by cgit v1.2.3 (git 2.46.0) at 2026-08-27 22:05:10 +0000