| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| ... | |
| |\ \
| | |
| | |
| | |
| | | |
tor-netdir: decrease minimum hsdir_interval from 30 to 5
See merge request tpo/core/arti!3042
|
| | | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
This matches the spec change:
<https://gitlab.torproject.org/tpo/core/torspec/-/merge_requests/406>
Progress on:
<https://gitlab.torproject.org/tpo/core/chutney/-/issues/40038>
|
| |\ \ \
| |_|/
|/| |
| | |
| | |
| | |
| | | |
Upgrade dependency to hickory-proto 0.25.2
Closes #1982
See merge request tpo/core/arti!3038
|
| | | |
| | |
| | |
| | | |
Closes #1982.
|
| | | |
| | |
| | |
| | | |
For conflux, we need to test against vegas.
|
| | | | |
|
| | | |
| | |
| | |
| | |
| | | |
I think this is going to be more of a tuning matter, so it's probably
best tackled after we have a working conflux implementation.
|
| | | | |
|
| | | |
| | |
| | |
| | |
| | |
| | |
| | | |
Instead of only shutting down the leg on which the protocol violation
occurred, it seems better to shut down the entire tunnel (because the
other endpoint, i.e. the exit, is the cause of the violation, so we
should probably avoid using it entirely).
|
| | | | |
|
| | | | |
|
| | | | |
|
| | | | |
|
| | | |
| | |
| | |
| | | |
The `UniqId` matters for multipath tunnels.
|
| | | |
| | |
| | |
| | |
| | | |
This preserves the existing behavior of `newcirc()` (which creates
3-hop circuits, where each hop has the same `HopDetail`).
|
| | | | |
|
| | | |
| | |
| | |
| | |
| | | |
For the conflux tests, we need to be able to create circuits with fake
hops that have different peer_ids.
|
| | | |
| | |
| | |
| | |
| | | |
Fixes a bug causing the conflux handshake to always fail with
`ConfluxHandshakeError::Timeout`.
|
| | | | |
|
| | | | |
|
| | | |
| | |
| | |
| | |
| | | |
This was addressed a while ago, in
09a601aed9aac9effa230701b02ba865b5754469
|
| | | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
If we reach this point and the join point is `None`, it means the
conflux set has so far consisted of a single leg. This means we need to
assign the last hop of the (only) leg to the join point. This initial
leg is in `self.circuits`, not in `legs` (`legs` is the list of *new*
legs that are being added to the set).
|
| | | | |
|
| | | |
| | |
| | |
| | | |
This will be needed for logging purposes.
|
| | | | |
|
| | | | |
|
| | | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
`num_legs` keeps track of the number of legs that have an in-progress
conflux handshake.
This updates the calculation to count the "initial" leg of the tunnel
too (because when converting a single-path reactor to a multi-path one,
the existing, "initial" circuit needs to complete the conflux handshake
too).
|
| | | |
| | |
| | |
| | |
| | | |
If we don't make an exception for LINK cells, we'll never be able to
send them, and the circuits will be forever "pending conflux handshake".
|
| | | |
| | |
| | |
| | |
| | |
| | | |
This fixes a bug where we'd fail to set the `ConfluxMsgHandler` for the
initial leg of the `ConfluxSet`, when converting the set from a
single-path set to a multi-path one.
|
| | | |
| | |
| | |
| | |
| | |
| | | |
As per the replacement rules from prop354. Except we can't actually
enforce the replacement rules at this level (they'll have to be enforced
by the caller).
|
| | | | |
|
| | | |
| | |
| | |
| | |
| | | |
We need to eventually tackle all of these, but none of them are
critical, so I propose we downgrade them to `TODO`.
|
| | | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
I really dislike that we're exposing the stream map this way. Ideally
we'd have some way of sharing the stream maps without exposing
the `StreamMap` in `reactor::conflux`.
Closes #2011
|
| | | | |
|
| | | | |
|
| | | | |
|
| | | |
| | |
| | |
| | | |
Most of this is code motion, I recommend reviewing with `--color-moved`.
|
| | | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
This helps hide the `CircHop` internals, and is the first step towards
providing a safer API that aims to reduce contention and prevent
deadlocking on the stream map mutex.
This change is also in preparation for implementing special handling for
the join point of a conflux tunnel (which will involve adding a new
`CircHop` API for sharing the stream map of another `CircHop`).
I recommend reviewing this diff with `--color-moved`.
|
| |\ \ \
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Initial refactoring on CircParameters
Closes #1967 and #1968
See merge request tpo/core/arti!3021
|
| | | | | |
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
When negotiation won't occur, we need to represent the fact by
disabling any settings that would depend on negotiation.
Otherwise we'll wind up with the client thinking everything
was supported, and the relay thinking that nothing is.
|
| | | | | |
|
| | | | | |
|
| | | | | |
|
| | | | | |
|
| | | | | |
|
| | | | | |
|
| | | | |
| | | |
| | | |
| | | |
| | | | |
The fallback CC algorithm is _always_ fixed-window, and we should only
use it when the selected CC algorithm is not supported.
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Now tor-circmgr no longer needs to check which Protover capabilities
are enabled, or construct a separate CircParameters for each hop.
Instead, tor-proto decides whether to use the fallback CC mode,
based on whether the target supports FLOWCTRL_CC.
Closes #1967.
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
We will construct this object based on the circuit parameters _and_
on the target's supported protocol versions, so we need to do so
when we have both pieces of info.
|
