f1eba83ebd
Fix a few clippy warnings
2020-12-17 15:23:55 -05:00
e4594810fc
Handle an error in circuit.rs without crashing.
2020-12-17 21:16:49 +02:00
52afa87ecd
Simplify RelayCellBody::recognized() method.
...
self is no longer mutable.
Co-authored-by: David Goulet <dgoulet@torproject.org>
2020-12-17 21:16:43 +02:00
cae16665cb
Basic support for IPv6 and begin flags.
...
We now have a way to tell a circuitmgr whether we require ipv4/ipv6
support for a target address in an exit policy, and we use similar
logic to set begin flags.
2020-12-15 14:41:03 -05:00
90fc798646
Fix some rustdoc warnings.
2020-12-09 09:46:39 -05:00
26a0654b05
Make RSAIdentity implement Copy.
2020-12-08 17:07:30 -05:00
a1f1dac4eb
Resolve a few nightly-only clippy issues
2020-12-07 16:14:36 -05:00
a52585d9d6
Mark some XXXX issues as XXXX-A1.
2020-12-07 16:07:59 -05:00
01de0cafc5
Run cargo fix --edition-idioms
2020-12-02 15:14:29 -05:00
efa6dd708b
Expose UniqId types for channels and circuits
2020-12-01 14:00:07 -05:00
ea0cb1d2cc
Rename LogId -> UniqId.
2020-12-01 13:44:25 -05:00
3fe97baa2a
Derive more types for circuit and channel LogIds
2020-12-01 13:21:46 -05:00
31773e0251
Enforce uniqueness on circuit and channel logids
2020-12-01 13:20:41 -05:00
e18b4e07cb
Upgrade a couple of dependencies
2020-11-28 14:17:42 -05:00
5188263429
Refactor first-hop handling types a bit
...
This lets us have the notion of "get the first hop of a path as some
kind of a chantarget", which will make it easier to write other path
types.
2020-11-17 16:46:23 -05:00
cb59ae3933
Move closed-check for circuits into an atomicbool.
2020-11-13 09:08:15 -05:00
4a83f21ae9
Light refactoring on remaining ChannelImpl functions
2020-11-13 08:39:01 -05:00
4b1ed6c39c
Turn the channel "closed" method into an AtomicBool
...
This lets us make the test for closed channels non-async.
2020-11-13 08:34:30 -05:00
11b257e1ae
Move a few immutable fields from ChannelImpl outside the lock.
...
Since these don't change while the channel is alive, we don't need
the lock to protect them.
2020-11-13 08:17:36 -05:00
6e5db3a0e7
Make Arc<>-ness of ClientCirc explicit.
2020-11-12 12:58:33 -05:00
ac174c7666
Make the Arc-ness of channels more explicit.
...
Previously every channel was a secret Arc<>, which I think is bad
style, and which stopped us from using weak references in other
places.
2020-11-12 12:34:50 -05:00
af16ec18a3
update some dependencies
2020-11-11 09:14:26 -05:00
4d24fff9c0
Upgrade dependencies
2020-11-10 12:27:10 -05:00
9c7ed1fdde
upgrade a few packages.
2020-11-05 16:56:29 -05:00
c0f9d94b34
Refactor DataStream to be splittable, and have internal locking.
...
This lets me fix the deadlock in the proxy code.
2020-11-04 12:10:09 -05:00
954108e0ee
stream: Update to enable more fine-grained locking
...
Not sure this is quite right, or quite deadlock free, but it lets
us read/write cells on a stream we haven't locked.
2020-11-04 11:50:23 -05:00
80b70881af
Add a tiny bit of cleanup logic to tor-circmgr
2020-11-02 14:21:14 -05:00
fbbc14d4fb
Tests for tor-chanmgr.
2020-10-30 15:45:53 -04:00
8d8ce482c9
chanmgr: don't return closing channels.
2020-10-30 09:51:52 -04:00
f399851bcf
Implement a channel-manager type to get or launch channels on request.
2020-10-30 09:51:31 -04:00
67e39450fb
Remove duplication of ipaddr argument in handshake.
2020-10-29 17:03:12 -04:00
f8d526a85a
Function to check whether a channel matches a target.
2020-10-29 17:03:12 -04:00
4cf972d69b
tor-proto API: Use new_ref, not clone.
...
Semantically, cloning a Channel or Circuit is actually increasing an
internal reference count. We should name the functions to reflect
that, to be more explicit about what we're doing.
2020-10-29 12:11:36 -04:00
5989b843cc
Add a simple test for stream creation.
2020-10-27 14:40:23 -04:00
705d9c061f
Add a test for extending a circuit with ntor.
2020-10-27 14:26:22 -04:00
911185ec38
circuit: test meta-cell receiving
2020-10-27 13:41:31 -04:00
b90ca2acd7
Downgrade an XXXXM3 message.
2020-10-27 13:11:57 -04:00
f8e1e6976f
Unify interface for detecting stream close.
2020-10-27 13:09:17 -04:00
4cded7a05a
Add a test for sending a cell.
2020-10-27 13:03:39 -04:00
8f5d239c21
Expand test to handle building circuit with ntor.
2020-10-27 12:23:18 -04:00
483425ba4b
Make a test for create_firsthop_fast. WOW that was hard.
2020-10-27 11:55:24 -04:00
dcf0936396
Make circuit reactors also have a run_once() method for testing.
2020-10-27 11:54:11 -04:00
5bd85e57ec
Add a fake, reactorless channel, for use in testing circuits
2020-10-27 10:16:05 -04:00
e2f7fe3827
Remove use of Cell<> in tor_proto.
...
I was confused about what this was for, I guess.
2020-10-27 10:14:28 -04:00
94042b60b8
Rename CircIDRange to CircIdRange
2020-10-27 09:51:28 -04:00
ee9ad7f59d
Run "cargo upgrade".
2020-10-26 14:56:06 -04:00
b93e4b76c8
Remove a bunch of dbg! calls.
2020-10-26 14:33:14 -04:00
a796f7bf6a
Use event_listener to find out about sendme replenishment.
...
Previously we would fail if a sendme window was exhausted and two
tasks were waiting for it to be replenished at the same time.
2020-10-26 14:25:46 -04:00
4348faff3c
Resolve a bunch of XXXM3 comments in tor-proto.
2020-10-26 14:03:25 -04:00
4d9c08669e
Improve documentation in tor-proto
2020-10-26 13:08:46 -04:00
db94abadcc
rudimentary test for channelbuilder.
2020-10-26 09:29:59 -04:00
1b86818085
Tests for sending bad/good cell types on a channel
2020-10-26 09:23:42 -04:00
498bc3633d
Clean up some XXXXs in channel::reactor.rs
2020-10-26 09:23:10 -04:00
6d6cff5176
Tests for receving incoming DESTROY cell.
2020-10-26 09:06:52 -04:00
937d924b0a
channel::reactor: Tests for handling incoming RELAY cells
2020-10-26 09:06:52 -04:00
46c991a087
Test for rejecting CREATED cells
2020-10-26 09:06:52 -04:00
494c4679b5
Add a CircEnt::DestroySent circuit state.
...
A circuit enters DestroySent after it's sent a DESTROY cell. We use
this state to make sure that we don't receive too many cells on a
destroyed circuit, _and_ so that we don't block such cells
completely.
2020-10-26 09:06:52 -04:00
32b025de1d
Use better return type in advance_from_opening
2020-10-25 23:37:25 -04:00
8c09981427
Use from_bytes name for Ed25519Identity
2020-10-25 20:32:47 -04:00
fa01a50f32
Rename CircID and StreamID to end with Id instead, for consistency
2020-10-25 20:29:01 -04:00
fbf200ae6f
Add some tests for channel reactor and related code.
2020-10-25 15:09:06 -04:00
0e91d97f76
Add a test for running and shutting down reactors.
...
This took a long time since I needed to learn about futures and
executors, but I think it'll work out okay.
2020-10-25 14:12:51 -04:00
e20bfaab0f
channel: Refactor the point when we split the framed codec.
...
We used to do this during creating the channel in Channel::new, but
now we do it one step before, so that it's easier to construct
channels for testing.
2020-10-25 12:42:53 -04:00
510a27f2d3
channel: Move looping part of run_impl() into run().
...
This lets us turn run-impl() into a run_once() function for easier
testing.
2020-10-25 12:06:36 -04:00
d1e5d54c24
Move cell logging into ChannelImpl
2020-10-23 10:19:56 -04:00
e1b741f0fc
circuit: Remove the needlessly separated ReactorCore type.
2020-10-23 09:49:33 -04:00
913db08384
channel: Remove the needlessly separated ReactorCore type.
2020-10-23 09:47:40 -04:00
9ef63a01e3
Test for finish-handshake function
2020-10-23 09:04:10 -04:00
b80b0b7019
handshake: tests for invalid signatures
2020-10-23 08:52:21 -04:00
9b46f04f58
Add tests for handshakes certifying the wrong thing
2020-10-23 08:45:40 -04:00
04f22a122e
Tests for handshakes with missing certs
...
Also handle tor_cell::Error::ChanProto better.
2020-10-23 08:34:35 -04:00
f13c759f0f
Use an enum for the return value of StreamMap::terminate
2020-10-22 13:42:45 -04:00
2de91b1c16
Initial tests for tor_proto::circuit::streammap
2020-10-22 13:35:18 -04:00
b5f019ccf1
Tests for tor_proto::circuit::halfstream
2020-10-22 12:36:39 -04:00
60e7d3f3c1
tor-proto: Add tests for circuit::sendme
2020-10-22 12:07:28 -04:00
19c48a2de5
Add tests for tor_proto::circuit::celltypes
2020-10-21 17:59:58 -04:00
9d1c279f65
Test a trivial succeeding case of link cert validation
2020-10-21 15:40:58 -04:00
11cd138c74
Fix a security issue (!) in link handshake validation.
...
When making sure that the peer had the right RSA identity, we
were comparing the RSA identity with itself, not with the RSA
identity we expected.
Found via unit testing (!).
2020-10-21 15:37:55 -04:00
53c27f7956
Infrastructure for testing handshake cert-validation code
2020-10-21 14:48:19 -04:00
48d0984d2f
More tests for handshake connect function
2020-10-21 14:09:20 -04:00
e0344dd78b
Start a basic test for the first part of the handshake.
2020-10-21 14:09:20 -04:00
a94dc1c7f1
LogId test: tolerate concurrent tests.
2020-10-21 14:09:20 -04:00
b528c914ff
Start on some tests for pieces of tor-proto::channel
2020-10-21 12:45:36 -04:00
3f3e98d190
Follow API convention about mutable getters.
...
They're supposed to be called field_mut().
2020-10-21 08:12:46 -04:00
743c3167e8
Rename as_message to into_message.
...
According to the API guidelines, "as_" is only for
borrowed->borrowed conversions.
2020-10-21 08:05:33 -04:00
486693115a
Add a close() method for streams
2020-10-20 14:46:59 -04:00
425ee8376b
Add new `terminate()` method to circuits and channels.
...
These aren't called "close" because they're more destructive than
that: they can be called even if other parties are using the circuit
or channel.
This is for arti#21.
2020-10-20 14:28:44 -04:00
c554db9881
Missing file and docs
2020-10-20 13:46:43 -04:00
6e5d248418
Add a TODO for register_meta_handler API
2020-10-20 13:41:34 -04:00
4f567e4a94
Work on closing flooding-based side-channels
...
We need to make sure that we're dropping cells that we don't
recognize or want, so that we can't be flooded with bogus junk.
2020-10-20 13:22:16 -04:00
b0cdd025a1
Add some helpful logs in circuit code.
2020-10-20 12:40:08 -04:00
7ce158ed1e
Add reasonable logging (I hope!) to channel
2020-10-20 11:45:53 -04:00
1dc3fc214a
Implement From, not Into.
...
These traits are inverses of one another, but implementing From is
always preferred since rust 1.41 relaxed the "orphan rules".
2020-10-19 17:40:43 -04:00
c7a2efe72b
Add and use remaining CertType values in tor-cert
2020-10-19 14:17:25 -04:00
378aa6d639
Add an enumeration for destroy reasons.
2020-10-19 13:13:55 -04:00
295a791a82
Improvements to Relay type in tor-netdir.
...
Now, a Relay is always valid. This required some changes to the
API: all_relays() has to return a new UncheckedRelay type that might
or might not be valid, and the functions on Relay and ChanTarget
that return ed25519 identities need to return an Ed25519Identity,
not an ed25519::PublicKey.
This change required some new encoding/decoding/conversion functions
on Ed25519Identity.
2020-10-19 09:22:49 -04:00
5523134607
Mark must-resolve XXXX issues with "XXXXM3".
...
"M3" is for "milestone 3" -- my target to fix the technical debt
that I think will be bad if we ship even a pre-alpha with it.
These aren't necessarily _all_ must-resolve, but they're all
must-look-at.
Closes #15
2020-10-18 17:38:59 -04:00
5173ee960f
tor-proto: Add a ClientCircChanMsg type.
...
This wraps exactly the ChanMsg values that are valid on open client
circuits, so that we can be sure that only those cells are sent to a
ClientCirc's reactor.
2020-10-15 14:23:12 -04:00
34962c3c9f
tor-proto: Add a "CreateResponse" type.
...
CreateResponse includes exactly those cells that are a correct
response to a CREATE2/CREATE_FAST, so we can be sure that only those
cells are actually passed to a PendingClientCirc.
2020-10-15 14:09:19 -04:00
3408c9ad26
Run "cargo upgrade".
2020-10-15 13:32:02 -04:00
Nick Mathewson
George Kadianakis