3e7c8fda96
Pass proto-layer connection events internally
...
Groundwork for replacing generic message-passing with lighter
event-representation-aware communication strategies.
2023-12-18 11:31:22 -08:00
253eba94d1
Replace dummy message with direct wake to trigger endpoint events
2023-12-18 11:31:21 -08:00
91bc3a49be
Switch to AtomicWaker to wake the endpoint driver
...
Groundwork for direct connection -> driver wake-ups, and yields one
less piece of state inside a monolithic mutex.
2023-12-18 11:31:21 -08:00
5a8483baec
Pass proto-layer endpoint events internally
...
Groundwork for replacing generic message-passing with lighter
event-representation-aware communication strategies.
2023-12-18 11:28:46 -08:00
302b3369bf
Don't write out Default manually
2023-12-17 13:14:23 -08:00
5eac310a5f
Don't plumb current time through endpoint events
...
This was only needed to determine the base time for the next
CID retirement timer, for which the marginal delay caused by fetching
a new `Instant::now()` on demand is harmless.
2023-12-17 12:44:27 -08:00
7723cbc03f
Validate 0-RTT frames based on packet type, not handshake progress
...
0-RTT packets might be received after we derive 1-RTT keys, e.g. if we
received a complete ClientHello. That shouldn't cause us to relax
0-RTT frame type restrictions.
2023-12-08 23:29:39 +01:00
106d44b1f7
Take the whole Packet structure in process_payload
2023-12-08 23:29:39 +01:00
b5d23a8647
Remove "max_tlps" configuration variable
...
It is not used since
<https://github.com/quinn-rs/quinn/pull/134 >
(commit bdc9ac76a4
2023-11-30 16:34:12 -08:00
d1bb2b8b13
Update windows-sys requirement from 0.48.0 to 0.52.0
...
Updates the requirements on [windows-sys](https://github.com/microsoft/windows-rs ) to permit the latest version.
- [Release notes](https://github.com/microsoft/windows-rs/releases )
- [Commits](https://github.com/microsoft/windows-rs/compare/0.48.0...0.52.0 )
---
updated-dependencies:
- dependency-name: windows-sys
dependency-type: direct:production
...
Signed-off-by: dependabot[bot] <support@github.com >
2023-11-20 14:28:51 +01:00
9bb40be8fd
Run FreeBSD workflow on Ubuntu
2023-11-14 14:11:28 -08:00
6cad76a3b7
Bump vmactions/freebsd-vm from 0 to 1
...
Bumps [vmactions/freebsd-vm](https://github.com/vmactions/freebsd-vm ) from 0 to 1.
- [Release notes](https://github.com/vmactions/freebsd-vm/releases )
- [Commits](https://github.com/vmactions/freebsd-vm/compare/v0...v1 )
---
updated-dependencies:
- dependency-name: vmactions/freebsd-vm
dependency-type: direct:production
update-type: version-update:semver-major
...
Signed-off-by: dependabot[bot] <support@github.com >
2023-11-14 20:32:48 +01:00
e77cd3e33d
Updates docs for Connection::accept_bi() and Connection::open_bi()
2023-11-10 21:49:56 +01:00
ca0fae8896
Downgrade duplicate packet message to debug level
2023-11-07 09:17:54 +01:00
ab280e467a
proto: add a way to configure the minimum MTU change needed to stop
...
the binary search used during the MTU discovery phase.
2023-11-06 10:41:49 -08:00
49aa4b61e0
#729 : proto: write outgoing packets to caller-supplied memory ( #1697 )
2023-11-01 12:17:10 -07:00
10f08e0652
ensure LICENSE files are present in all published crates
2023-10-28 17:26:10 +02:00
6d80efeeae
Update async-io requirement from 1.6 to 2.0
...
Updates the requirements on [async-io](https://github.com/smol-rs/async-io ) to permit the latest version.
- [Release notes](https://github.com/smol-rs/async-io/releases )
- [Changelog](https://github.com/smol-rs/async-io/blob/master/CHANGELOG.md )
- [Commits](https://github.com/smol-rs/async-io/compare/v1.6.0...v2.0.0 )
---
updated-dependencies:
- dependency-name: async-io
dependency-type: direct:production
...
Signed-off-by: dependabot[bot] <support@github.com >
2023-10-23 15:58:29 +02:00
15222f2f5d
Implement Debug for IdleTimeout.
...
Fix typo.
2023-10-23 13:42:40 +02:00
dd34d57a0a
Make quinn-proto::{Connection, Endpoint} deterministic ( #1691 )
...
`quinn-proto::{Connection, Endpoint}` structs call
`StdRng::from_entropy()` in their implementation. It makes it hard to
reproduce errors in an otherwise deterministic test case.
This PR addresses the issue by adding a `StdRng` argument to the
respective `new` functions. For other functions that may create an
endpoint/connection, an argument of type `impl FnOnce() -> StdRng` is
added. This avoids eager calls to generate fresh entropy. e.g. the
`Endpoint::handle` function is frequently called but we'd only need the
rng when handling a new incoming connection.
2023-10-23 11:46:15 +02:00
7bd92f65aa
Avoid redundant link target in docs
2023-10-23 11:42:54 +02:00
24212da394
Use or_default() as suggested by clippy
2023-10-23 11:42:54 +02:00
9f437c0da7
proto: move IterErr below users
2023-10-23 11:42:54 +02:00
99ef446331
Yield protocol violation for packets without frames
2023-10-23 11:42:54 +02:00
e1e1e6e392
Remove explicit rust-version requirement from unpublished crates
2023-09-24 14:22:14 -07:00
81d83a93be
Reduce rightward drift
2023-09-24 14:22:14 -07:00
e51f848296
Bump MSRV to 1.65 for let-else
2023-09-24 14:22:14 -07:00
fc5e689b88
Fix ACK frequency extension corner case
...
Reflects the logic change in
https://github.com/quicwg/ack-frequency/pull/222 . See also discussion
in https://github.com/quicwg/ack-frequency/issues/183 .
2023-09-24 14:22:14 -07:00
e5bf98c3d7
Pass byte read count to FinishEarly
...
If read_exact() finishes early (end of stream), report the number of
bytes actually read.
Fixes #1663
2023-09-20 12:18:42 -07:00
f81c2fafa7
More principled error handling for invalid frames
2023-09-20 12:16:58 -07:00
0af891e0b8
Don't send IMMEDIATE_ACK frames every RTT
...
This behavior was redundant to intelligent transmission of
ACK_FREQUENCY frames when RTT change is observed.
2023-09-09 10:21:24 -07:00
b83d5c483c
Test automatic adjustment of max ACK delay
2023-09-09 10:21:24 -07:00
1f70570c65
Don't clamp max ACK delay to RTTs below 25ms
2023-09-09 10:21:24 -07:00
2d26d9363a
Adjust max ACK delay dynamically based on RTT
2023-09-09 10:21:24 -07:00
86ea03ec86
Push max ACK delay clamping down into ACK frequency logic
2023-09-09 10:21:24 -07:00
d30d5479a3
Intelligible time logging in tests
...
Replaces "advancing to Instant { ... }" with e.g. "advancing to 10ms"
2023-09-09 10:21:24 -07:00
8b7cea22ca
Trace-log transmitted max_ack_delay values
2023-09-09 10:21:24 -07:00
5d4d58387d
Bump actions/checkout from 3 to 4
...
Bumps [actions/checkout](https://github.com/actions/checkout ) from 3 to 4.
- [Release notes](https://github.com/actions/checkout/releases )
- [Changelog](https://github.com/actions/checkout/blob/main/CHANGELOG.md )
- [Commits](https://github.com/actions/checkout/compare/v3...v4 )
---
updated-dependencies:
- dependency-name: actions/checkout
dependency-type: direct:production
update-type: version-update:semver-major
...
Signed-off-by: dependabot[bot] <support@github.com >
2023-09-04 16:24:17 +02:00
d51b630fcd
Add client/server spans to bulk benchmark
2023-09-02 22:30:11 -07:00
7442da16f9
Enable ACK frequency extension in bulk benchmark
2023-09-02 22:30:11 -07:00
089e213fba
Generalize UdpStats::transmits to ios
...
Since this field exists for both tx and rx, we might as well make use
of it.
2023-09-02 21:22:12 +02:00
a516c2b836
Silence an info message related to ack frequency
2023-08-31 17:33:09 +00:00
a7a3c4da6a
Add missing export for proto::AckFrequencyConfig.
2023-08-29 16:25:32 +02:00
307d80b939
proto: make now explicit for Endpoint::connect()
2023-08-25 11:55:43 -07:00
c2a54b9acf
Set workspace resolver explicitly
2023-08-25 11:55:43 -07:00
2d9bdaecc8
Silence clippy false positive
2023-08-25 11:55:43 -07:00
e0f81593f7
Fix formatting for Rust 1.72
2023-08-25 11:55:43 -07:00
7f26029284
proto: detect stateless resets in authed and unprotected packets
...
RFC9000 says: endpoints MUST treat any packet ending in a valid
stateless reset token as a Stateless Reset
Previously, we did not detect stateless resets that appeared to be
unprotected packets (e.g. Retry or Version Negotiation) or which were
successfully decrypted (astronomically unlikely with TLS, but possible
with custom cryptographic layers).
2023-08-23 14:19:02 -07:00
394ac8c2b8
proto: don't panic when draining a unknown connection
...
A panic here was reported by a user when testing under very scarce CPU
conditions. The root cause is not yet well understood (perhaps a
connection emitting `Drained` twice for some reason?), but we can
still reduce the blast radius.
2023-08-23 13:32:12 -07:00
47678a1a62
Replace Gitter link with Discord
2023-08-22 13:29:59 -07:00
Benjamin Saunders