The `net.rs` file is not declared as a module in `src/sim/mod.rs` and thus unused.

* fix(http3): always qlog on send_buffer()

`neqo_http3::SendMessage` calls `qlog::h3_data_moved_down()` whenever it moves
data down to the QUIC layer. `SendMessage` moves data down to the QUIC layer
either directly via `self.stream.send_atomic` or indirectly buffered through `self.stream.send_buffer`.

Previously only one of the 3 calls to `self.stream.send_buffer` would thereafter
call `qlog::h3_data_moved_down()`.

This commit moves the `h3_data_moved_down` call into `self.stream.send_buffer`, thus
ensuring the function is always called when data is moved. In addition,
`self.stream.send_atomic` now as well does the qlog call, thus containing all
qlog logic in `buffered_send_stream.rs` instead of `send_message.rs`.

* Trigger benchmark run

* Don't qlog if buffer is empty

* Fix typo

* Use early return to keep indentation short

* Don't qlog if sent is 0

Bumps [codecov/codecov-action](https://github.com/codecov/codecov-action) from 4.3.0 to 4.3.1.
- [Release notes](https://github.com/codecov/codecov-action/releases)
- [Changelog](https://github.com/codecov/codecov-action/blob/main/CHANGELOG.md)
- [Commits](codecov/codecov-action@8450866...5ecb98a)

---
updated-dependencies:
- dependency-name: codecov/codecov-action
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* build(deps): bump actions/download-artifact from 4.1.6 to 4.1.7

Bumps [actions/download-artifact](https://github.com/actions/download-artifact) from 4.1.6 to 4.1.7.
- [Release notes](https://github.com/actions/download-artifact/releases)
- [Commits](actions/download-artifact@9c19ed7...65a9edc)

---
updated-dependencies:
- dependency-name: actions/download-artifact
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>

* Addition

---------

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: Lars Eggert <lars@eggert.org>

* build(deps): bump actions/checkout from 4.1.3 to 4.1.4

Bumps [actions/checkout](https://github.com/actions/checkout) from 4.1.3 to 4.1.4.
- [Release notes](https://github.com/actions/checkout/releases)
- [Changelog](https://github.com/actions/checkout/blob/main/CHANGELOG.md)
- [Commits](actions/checkout@1d96c77...0ad4b8f)

---
updated-dependencies:
- dependency-name: actions/checkout
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>

* Additions

---------

Signed-off-by: dependabot[bot] <support@github.com>
Signed-off-by: Lars Eggert <lars@eggert.org>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: Lars Eggert <lars@eggert.org>

…e is lost (#1882)

It's too complex, and we should instead just fix the interop issues.

* ci: Build Firefox with current neqo

* --application-choice browser

* Allow qlog dupes

* Typo

* Get Firefox source from GitHub

* Again

* Again

* Again

* Again

* Again

* Debug

* Again

* Again

* Again

* Again

* Again

* Again

* qlog

* Fix

* Again

* Again

* Again

* Fingers crossed...

* Try and prevent running out of disk space

* Again

* Again

* Set CARGO_HOME and use sccache

* actionlint

* Use matrix

* Again

* Again

* Upload binaries

* Identify what to upload

* Push

* Naming

* Again

* Again

* Again

* Again

* Again

* Don't disable tests for debug builds

* Work around https://bugzilla.mozilla.org/show_bug.cgi?id=1894031

* Prepare to comment on PR

* Again

* Again

* Undo

* Again

* Again

* Try and comment

* cargo update

* Again

* Again

* Again

* Again

* Again

* Consolidate PR reporting

* Fix actionlint

---------

Signed-off-by: Lars Eggert <lars@eggert.org>

…1838)

* fix(SendMessage): use SendStream::set_writable_event_low_watermark

Previously `SendMessage::send_data` could stall, if less than the minimum
message size is available to be sent. See
#1819 for details.

This commit implements solution (3) proposed in
#1819.

This commit introduces `SendStream::set_writable_event_low_watermark` which is
then used in `SendMessage::send_data` to signal to `SendStream` the minimum
required send space (low watermark) for the next send. Once reached,
`SendStream` emits a `SendStreamWritable` eventually triggering another
`SendMessage::send_data`.

Alternative to #1835. Compared to
#1835, this fix does not utilize the
`SendMessage` buffer, thus does not introduce an indirection to the send path.
In addition, under the assumption that available send space is increased in
larger batches, this fix does not send tiny data frames (2 byte header, 1 byte
goodput). Downside, compared to #1835, is
that it requires both changes in `neqo-transport` and `neqo-http3`.

Secondarily, this fixes #1821 as well.

* Move const

* Add documentation

* Add SendStream test

* Fix intra doc links

* Add neqo-http3 test

* Replace goodput with payload

* Re-trigger benchmarks

Let's see whether the "Download" benchmark is consistent.

* Rename emit_writable_event to maybe_emit_writable_event

* Replace expect with unwrap

* Use NonZeroUsize::get

* Replace expect with unwrap

* %s/Actually sending/Sending

* Typo

* Have update() return available amount

* Document setting once would suffice

* Reduce verbosity

* fix: drop RefCell mutable borrow early

Fixes #1887

* chore: Fix new clippy lints in cargo 1.80.0-nightly (05364cb2f 2024-05-03)

* No need for `#[allow(clippy::mutable_key_type)]`

* No need for `build = "build.rs"`

The QUIC Interop Runner requires an http3 and http09 implementation for both
client and server. The client code is already structured into an http3 and an
http09 implementation since #1727.

This commit does the same for the server side, i.e. splits the http3 and http09
implementation into separate Rust modules.

* fix: Resurrect #1662

Botched the "merge from main" on #1662. Reverted the merge. This PS
has the changes in #1662, rebased to main. Sorry for the mess.

* Fix

* One-liner

* Merge main

* Merge main

* Move back to BTreeMap

* Tweak ordering

* More simplification/performance

* Test expectation fixup

* Update neqo-transport/src/cc/classic_cc.rs

Signed-off-by: Lars Eggert <lars@eggert.org>

* Apply suggestions from code review

Signed-off-by: Lars Eggert <lars@eggert.org>

---------

Signed-off-by: Lars Eggert <lars@eggert.org>
Co-authored-by: Lars Eggert <lars@eggert.org>

* ci: Only build Fx for PRs

And not when landing a PR in `main`.

Signed-off-by: Lars Eggert <lars@eggert.org>

* Add `workflow_dispatch`

Signed-off-by: Lars Eggert <lars@eggert.org>

---------

Signed-off-by: Lars Eggert <lars@eggert.org>

* refactor(bin): introduce server/http3.rs and server/http09.rs

The QUIC Interop Runner requires an http3 and http09 implementation for both
client and server. The client code is already structured into an http3 and an
http09 implementation since #1727.

This commit does the same for the server side, i.e. splits the http3 and http09
implementation into separate Rust modules.

* refactor: merge mozilla-central http3 server into neqo-bin

There are two server implementations based on neqo:

1. https://github.com/mozilla/neqo/tree/main/neqo-bin/src/server
  - http3 and http09 implementation
  - used for manual testing and QUIC Interop

2. https://searchfox.org/mozilla-central/source/netwerk/test/http3server/src/main.rs
  - used to test Firefox

I assume one was once an exact copy of the other. Both implement their own I/O,
event loop, ... Since then, the two implementations diverged significantly.
Especially (1) saw a lot of improvements in recent months:

- #1564
- #1569
- #1578
- #1581
- #1604
- #1612
- #1676
- #1692
- #1707
- #1708
- #1727
- #1753
- #1756
- #1766
- #1772
- #1786
- #1787
- #1788
- #1794
- #1806
- #1808
- #1848
- #1866

At this point, bugs in (2) are hard to fix, see e.g.
#1801.

This commit merges (2) into (1), thus removing all duplicate logic and
having (2) benefit from all the recent improvements to (1).

* Move firefox.rs to mozilla-central

* Reduce HttpServer trait functions

* Extract constructor

* Remove unused deps

* Remove clap color feature

Nice to have. Adds multiple dependencies. Hard to justify for mozilla-central.

* ci: Try and fix CI Fx build

* Again

* Again

* Remove commented code

* Remove more comments

* build(deps): bump actions/checkout from 4.1.3 to 4.1.5

Bumps [actions/checkout](https://github.com/actions/checkout) from 4.1.3 to 4.1.5.
- [Release notes](https://github.com/actions/checkout/releases)
- [Changelog](https://github.com/actions/checkout/blob/main/CHANGELOG.md)
- [Commits](actions/checkout@v4.1.3...44c2b7a)

---
updated-dependencies:
- dependency-name: actions/checkout
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>

* Add more

---------

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: Lars Eggert <lars@eggert.org>

…ns (#1896)

Bumps martenseemann/quic-network-simulator-endpoint from `1259654` to `91b21d4`.

---
updated-dependencies:
- dependency-name: martenseemann/quic-network-simulator-endpoint
  dependency-type: direct:production
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

Bumps [actions/download-artifact](https://github.com/actions/download-artifact) from 4.1.6 to 4.1.7.
- [Release notes](https://github.com/actions/download-artifact/releases)
- [Commits](actions/download-artifact@v4.1.6...65a9edc)

---
updated-dependencies:
- dependency-name: actions/download-artifact
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* feat(bin): use quinn-udp crates.io release instead of git ref

`neqo-bin` has been importing `quinn-udp` as a git reference, in order to
include quinn-rs/quinn#1765. The quinn project has since
released `quinn-udp` `v0.5.0`.

This commit upgrades `neqo-bin` to use `quinn-udp` `v0.5.0`.

`quinn-udp` now takes a data reference (`&[u8]`) instead of owned
data (`bytes::Bytes`) on its send path, thus no longer requiring `neqo-bin` to
convert, but simply pass a reference. See
quinn-rs/quinn#1729 (comment) for details.

`quinn-udp` has dropped `sendmmsg` support in the `v0.5.0`
release (quinn-rs/quinn@ee08826).
`neqo-bin` does not (yet) use `sendmmsg`. This might change in the
future (#1693).

* remove impl From<Datagram> for Vec<u8>

This reverts commit 342e4e7.

With #1878 merged and
https://bugzilla.mozilla.org/show_bug.cgi?id=1895319 available, one can now
reapply the patch removing `Server::timers`.

More specifically, the actual bug fix on mozilla-central side:

``` rust
let output = if self.response_to_send.is_empty() {
    output
} else {
    // In case there are pending responses to send, make sure a reasonable
    // callback is returned.
    const MIN_INTERVAL: Duration = Duration::from_millis(100);

    match output {
        Output::None => Output::Callback(MIN_INTERVAL),
        o @ Output::Datagram(_) => o,
        Output::Callback(d) => Output::Callback(min(d, MIN_INTERVAL)),
    }
};
```

See https://phabricator.services.mozilla.com/D209574.

* make expiry return longer timeout

* fix tests

* again

* address comments

* only compare pto

* address comments

Bumps [codecov/codecov-action](https://github.com/codecov/codecov-action) from 4.3.1 to 4.4.0.
- [Release notes](https://github.com/codecov/codecov-action/releases)
- [Changelog](https://github.com/codecov/codecov-action/blob/main/CHANGELOG.md)
- [Commits](codecov/codecov-action@5ecb98a...6d79887)

---
updated-dependencies:
- dependency-name: codecov/codecov-action
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* ci: Only show performance improvements and regressions details

By default. Makes the perf PR comment a bit more concise.

* Add newline

* build(deps): bump actions/checkout from 4.1.5 to 4.1.6

Bumps [actions/checkout](https://github.com/actions/checkout) from 4.1.5 to 4.1.6.
- [Release notes](https://github.com/actions/checkout/releases)
- [Changelog](https://github.com/actions/checkout/blob/main/CHANGELOG.md)
- [Commits](actions/checkout@44c2b7a...a5ac7e5)

---
updated-dependencies:
- dependency-name: actions/checkout
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>

* Additions

---------

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: Lars Eggert <lars@eggert.org>

Because I've also limited the loopback MTU to 1500.

* ci: Set `RUST_LOG` during coverage tests

So that logging-related code is being included in the coverage check.

* Bump codecov version, to see if that makes it post comments again

* Back to 4.4.0

* 4.3.1

* 4.4.1

* verbose

Each call takes ~1ms.

EcnInfo::on_packet_sent may update the ECN state and can hence change it.

Also throw in some unrelated fixes to make the ECN log output more
quiet, and to test/test.sh.

* Enable Xyber768d00 on servers by default

* Fix clippy

Signed-off-by: Lars Eggert <lars@eggert.org>

* Add xyber connection test

* rustfmt

---------

Signed-off-by: Lars Eggert <lars@eggert.org>
Co-authored-by: Lars Eggert <lars@eggert.org>

Bumps [docker/login-action](https://github.com/docker/login-action) from 3.1.0 to 3.2.0.
- [Release notes](https://github.com/docker/login-action/releases)
- [Commits](docker/login-action@e92390c...0d4c9c5)

---
updated-dependencies:
- dependency-name: docker/login-action
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

Bumps [docker/build-push-action](https://github.com/docker/build-push-action) from 5.3.0 to 5.4.0.
- [Release notes](https://github.com/docker/build-push-action/releases)
- [Commits](docker/build-push-action@2cdde99...ca052bb)

---
updated-dependencies:
- dependency-name: docker/build-push-action
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

Bumps [codecov/codecov-action](https://github.com/codecov/codecov-action) from 4.4.1 to 4.5.0.
- [Release notes](https://github.com/codecov/codecov-action/releases)
- [Changelog](https://github.com/codecov/codecov-action/blob/main/CHANGELOG.md)
- [Commits](codecov/codecov-action@125fc84...e28ff12)

---
updated-dependencies:
- dependency-name: codecov/codecov-action
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* build(deps): bump actions/checkout from 4.1.6 to 4.1.7

Bumps [actions/checkout](https://github.com/actions/checkout) from 4.1.6 to 4.1.7.
- [Release notes](https://github.com/actions/checkout/releases)
- [Changelog](https://github.com/actions/checkout/blob/main/CHANGELOG.md)
- [Commits](actions/checkout@a5ac7e5...692973e)

---
updated-dependencies:
- dependency-name: actions/checkout
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>

* Add actions

* Oops

---------

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: Lars Eggert <lars@eggert.org>

* ci: Hook up Taskcluster

* Move directory

Signed-off-by: Lars Eggert <lars@eggert.org>

* ci: Use cargo-quickinstall instead of cargo-binstall

Since latter just bumped its MSRV way beyond what is reasonable for us.

* Remove flags

Until someone who understands it can debug.

Since binstall is having issues on macOS it seems.

Signed-off-by: Lars Eggert <lars@eggert.org>

…1936)

Bumps [docker/build-push-action](https://github.com/docker/build-push-action) from 5.4.0 to 6.1.0.
- [Release notes](https://github.com/docker/build-push-action/releases)
- [Commits](docker/build-push-action@ca052bb...31159d4)

---
updated-dependencies:
- dependency-name: docker/build-push-action
  dependency-type: direct:production
  update-type: version-update:semver-major
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

#1568 introduced TTL information to
datagrams. On the input path it would take the ttl information, but not act on
it. On the output path it would set only "the default TTL on many OSes".

https://github.com/mozilla/neqo/blob/66504908e5fa070a8a5fa67d8b5a201d2c9a5cc5/neqo-transport/src/path.rs#L576

This commit removes the ttl information from `Datagram`, thus reverting a subset
of #1568.

This is partially motivated by #1920,
introducing `quinn-udp` as the IO library of choice, which does not support
reading and writing TTL.

See also discussion in
#1920 (comment).

They seem to have a race condition that makes them crash.

I wasn't able to figure out why they crash, this seems to happen
inside NSS:
```
Process 47677 stopped
* thread #2, name = 'init_twice_withdb', stop reason = EXC_BAD_ACCESS (code=1, address=0x0)
    frame #0: 0x000000019b4994f8 libsystem_pthread.dylib`pthread_mutex_lock + 12
libsystem_pthread.dylib`pthread_mutex_lock:
->  0x19b4994f8 <+12>: ldr    x8, [x0]
    0x19b4994fc <+16>: mov    w9, #0x545a ; =21594
    0x19b499500 <+20>: movk   w9, #0x4d55, lsl #16
    0x19b499504 <+24>: cmp    x8, x9
(lldb) up
frame #1: 0x0000000100802f58 libnspr4.dylib`PR_Lock + 20
libnspr4.dylib`PR_Lock:
->  0x100802f58 <+20>: bl     0x10080c0c0    ; symbol stub for: pthread_self
    0x100802f5c <+24>: str    x0, [x19, #0xb8]
    0x100802f60 <+28>: mov    w8, #0x1 ; =1
    0x100802f64 <+32>: str    w8, [x19, #0xb0]
(lldb) up
frame #2: 0x00000001007fc188 libnspr4.dylib`PR_CallOnce + 56
libnspr4.dylib`PR_CallOnce:
->  0x1007fc188 <+56>: ldr    w23, [x19]
    0x1007fc18c <+60>: ldr    w20, [x19, #0x8]
    0x1007fc190 <+64>: ldr    x0, [x22, #0x430]
    0x1007fc194 <+68>: bl     0x100802f74    ; PR_Unlock
(lldb) up
frame #3: 0x000000010066f634 libnss3.dylib`nss_Init + 112
libnss3.dylib`nss_Init:
->  0x10066f634 <+112>: cbz    w0, 0x10066f640 ; <+124>
    0x10066f638 <+116>: mov    w19, #-0x1 ; =-1
    0x10066f63c <+120>: b      0x10066f968    ; <+932>
    0x10066f640 <+124>: stp    x21, x22, [x29, #-0x80]
(lldb) up
frame #4: 0x000000010066ff44 libnss3.dylib`NSS_Initialize + 100
libnss3.dylib`NSS_Initialize:
->  0x10066ff44 <+100>: ldp    x29, x30, [sp, #0x40]
    0x10066ff48 <+104>: add    sp, sp, #0x50
    0x10066ff4c <+108>: ret

libnss3.dylib`NSS_InitContext:
    0x10066ff50 <+0>:   sub    sp, sp, #0x60
(lldb) up
frame #5: 0x00000001000035fc init-d4e9f02b33e50e46`init::init_twice_withdb::h34013f715e9b4f6e at init.rs:65:9
   62  	    let pathstr = path.to_str().unwrap();
   63  	    let dircstr = CString::new(pathstr).unwrap();
   64  	    unsafe {
-> 65  	        nss::NSS_Initialize(
   66  	            dircstr.as_ptr(),
   67  	            empty.as_ptr(),
   68  	            empty.as_ptr(),
```

* ci: rename .taskcluster to taskcluster

While we'd like to keep the CI configuration files hidden in this repo,
I came across a bug in Taskgraph triggered by changing the root
Taskgraph directory to a non standard location. This bug will require a
new release to fix (and I suspect there will be other bugs revealed
afterwards).

In the meantime, we can move the files to the standard place while this
is fixed. I filed taskcluster/taskgraph#522 to
track the fix.

* ci: use Decision docker image with Taskgraph included

* ci: re-enable Taskcluster

Bumps [docker/build-push-action](https://github.com/docker/build-push-action) from 6.1.0 to 6.2.0.
- [Release notes](https://github.com/docker/build-push-action/releases)
- [Commits](docker/build-push-action@31159d4...1556069)

---
updated-dependencies:
- dependency-name: docker/build-push-action
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

Until someone who understands it can debug. #1935 didn't fix it.

Signed-off-by: StepSecurity Bot <bot@stepsecurity.io>

* refactor(transport/server): always iterate each connection

Say a `neqo_transport::Server` is managing a single `neqo_transport::Connection`
in `Server::connections`. Assume the following chain of events:

1. A user (e.g. `neqo-http3`) calls `Server::process`.

    ``` rust
    pub fn process(&mut self, dgram: Option<&Datagram>, now: Instant) -> Output {
        if self.wake_at.map_or(false, |c| c <= now) {
            self.wake_at = None;
        }

        dgram
            .and_then(|d| self.process_input(d, now))
            .or_else(|| self.process_next_output(now))
            .map(|d| {
                qtrace!([self], "Send packet: {:?}", d);
                Output::Datagram(d)
            })
            .or_else(|| self.wake_at.take().map(|c| Output::Callback(c - now)))
            .unwrap_or_else(|| {
                qtrace!([self], "Go dormant");
                Output::None
            })
    }
    ```

    https://github.com/mozilla/neqo/blob/6664452e2ba25f028ebf07c404fff3d0193c0ef4/neqo-transport/src/server.rs#L660-L677

    2. `self.wake_at` is `None`.
    3. `dgram` is `None`, thus `self.process_input` is never called.
    4. `self.process_next_output(now)` is called.
        ``` rust
        fn process_next_output(&mut self, now: Instant) -> Option<Datagram> {
            qtrace!([self], "No packet to send, look at waiting connections");
            while let Some(c) = self.waiting.pop_front() {
                if let Some(d) = self.process_connection(&c, None, now) {
                    return Some(d);
                }
            }
        ```

        https://github.com/mozilla/neqo/blob/6664452e2ba25f028ebf07c404fff3d0193c0ef4/neqo-transport/src/server.rs#L636-L642
        1. It attains a reference to the one `Connection` through `self.waiting.pop_front()`.
        2. It calls self.process_connection which in turn calls
        `Connection::process`, returning a `Output::Callback(_)`, which is stored in `self.wake_at`.
  5. `self.wake_at.take()` takes the callback and returns it to the user as `Output::Callback`.
6. The user calls `Server::process` again.
    1. `self.wake_at` is `None`.
    2. `dgram` is `None`, thus `self.process_input` isn't called.
    3. `Server::process` calls `Server::process_next_output`.
        1. `Server::process_next_output` finds no connection reference in
        `self.waiting` and thus returns `None`.
    4. `self.wake_at.take()` is `None`
    5. `Server::process` returns `None`

Result is that the user received an `Output::None` even though the one `Connection` managed
by the `Server` is waiting for a callback. Thus the server stalls.

The single source of truth of whether a `Connection` needs a callback or not is
the `Connection` itself. Instead of duplicating this information in
`Server::wake_at`, `Server::waiting`, `ServerConnectionState::wake_at`, always
ask the single source of truth, i.e. the `Connection`.

More concretely, with this patch `Server::process` always calls
`Connection::process` for each of its `Connection`s. It does not try to be smart
on whether a `Connection` needs `process`ing or not.

* Don't duplicate active state

* Don't duplicate attempt state

* Simplify process_connection call

* Deduplicate active connections

* Attempt fix for zerortt

* clippy

* Attempt fix for same_initial_after_connected

* Address minor TODOs

* Remove Server::add_to_waiting

It is no longer needed as each connection is always processed.

* Allow connected_server with more than one connection

* Remove unnecessary TODO

* Simplify closed connection clean up

* [StepSecurity] Update Dependabot configuration

Signed-off-by: StepSecurity Bot <bot@stepsecurity.io>

* weekly

Signed-off-by: Lars Eggert <lars@eggert.org>

---------

Signed-off-by: StepSecurity Bot <bot@stepsecurity.io>
Signed-off-by: Lars Eggert <lars@eggert.org>
Co-authored-by: Lars Eggert <lars@eggert.org>

This commit adds a basic test, passing a short header packet to a fresh server,
expecting the server to drop the packet for the unknown connection.

So the `<details>` stuff works properly. Hopefully.

See corresponding lints:

- https://rust-lang.github.io/rust-clippy/master/index.html#/mutable_key_type
- https://rust-lang.github.io/rust-clippy/master/index.html#/manual_inspect

See recent CI failure:

https://github.com/mozilla/neqo/actions/runs/9759065486/job/26934981259

```
  error:
     --> neqo-transport/src/path.rs:214:62
      |
  214 |         let old_path = self.primary.replace(Rc::clone(path)).map(|old| {
      |                                                              ^^^
      |
      = help: for further information visit https://rust-lang.github.io/rust-clippy/master/index.html#manual_inspect
      = note: `-D clippy::manual-inspect` implied by `-D warnings`
      = help: to override `-D warnings` add `#[allow(clippy::manual_inspect)]`
  help: try
      |
  214 ~         let old_path = self.primary.replace(Rc::clone(path)).inspect(|old| {
  215 ~             old.borrow_mut().set_primary(false);
      |

  error: mutable key type
     --> neqo-transport/src/server.rs:603:45
      |
  603 |     pub fn active_connections(&mut self) -> HashSet<ActiveConnectionRef> {
      |                                             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
      |
      = help: for further information visit https://rust-lang.github.io/rust-clippy/master/index.html#mutable_key_type
      = note: `-D clippy::mutable-key-type` implied by `-D warnings`
      = help: to override `-D warnings` add `#[allow(clippy::mutable_key_type)]`

  error: could not compile `neqo-transport` (lib) due to 2 previous errors
error: process didn't exit successfully: `/home/runner/.rustup/toolchains/nightly-x86_64-unknown-linux-gnu/bin/cargo clippy --all-targets --manifest-path fuzz/Cargo.toml -- -D warnings` (exit status: 101)
```

* feat(bin): enable Firefox to use quinn-udp through neqo-bin

* Move udp to separate crate

* Remove misleading log

* Reduce tokio feature set

* Use thread_local for receive buffer

* Add windows support

* Cleanup send_inner and recv_inner

* Move WouldBlock handling to Firefox

* Feature flag tests using tokio

* fmt

* Undo unrelated change

* EOL

* Rename socket field to inner

* Use qtrace

* Short to RECV_BUF and move meta into closure

* Udate documentation

* Add TODO to experiment with different buffer sizes

* Remove private doc link

---------

Co-authored-by: Lars Eggert <lars@eggert.org>

* [StepSecurity] Add Dependency review Workflow

Signed-off-by: StepSecurity Bot <bot@stepsecurity.io>

* unharden

Signed-off-by: Lars Eggert <lars@eggert.org>

---------

Signed-off-by: StepSecurity Bot <bot@stepsecurity.io>
Signed-off-by: Lars Eggert <lars@eggert.org>
Co-authored-by: Lars Eggert <lars@eggert.org>

Signed-off-by: StepSecurity Bot <bot@stepsecurity.io>

Set `bench = false`. By default cargo will use `libtest` for benchmarking.
Executing a `cargo bench` with criterion specific command line flags at the
workspace level will thus trigger an error when executing the `neqo-udp`
benchmarks.

Not needed, and install fails on macOS for me.

Otherwise merges stall.

So the `merge_group` trigger works.

See actions/dependency-review-action#456 (comment)

* chore: Turn on more clippy lints, and fix the warnings

* Fixes

* Fixes

* Fixes

* Update neqo-transport/src/addr_valid.rs

Co-authored-by: Martin Thomson <mt@lowentropy.net>
Signed-off-by: Lars Eggert <lars@eggert.org>

* Suggestions from @martinthomson

* More from @martinthomson

* More from @martinthomson

* And more

* Fixes after rebase

* Better Cubic fix

* let _ -> _

* Add reason to `#[allow(clippy::mutable_key_type)]`

* fmt

---------

Signed-off-by: Lars Eggert <lars@eggert.org>
Co-authored-by: Martin Thomson <mt@lowentropy.net>

* build(deps): bump actions/upload-artifact from 4.3.3 to 4.3.4

Bumps [actions/upload-artifact](https://github.com/actions/upload-artifact) from 4.3.3 to 4.3.4.
- [Release notes](https://github.com/actions/upload-artifact/releases)
- [Commits](actions/upload-artifact@6546280...0b2256b)

---
updated-dependencies:
- dependency-name: actions/upload-artifact
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>

* Add actions

---------

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: Lars Eggert <lars@eggert.org>

Bumps [docker/build-push-action](https://github.com/docker/build-push-action) from 6.2.0 to 6.3.0.
- [Release notes](https://github.com/docker/build-push-action/releases)
- [Commits](docker/build-push-action@1556069...1a16264)

---
updated-dependencies:
- dependency-name: docker/build-push-action
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

Bumps [actions/download-artifact](https://github.com/actions/download-artifact) from 4.1.7 to 4.1.8.
- [Release notes](https://github.com/actions/download-artifact/releases)
- [Commits](actions/download-artifact@65a9edc...fa0a91b)

---
updated-dependencies:
- dependency-name: actions/download-artifact
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

Bumps [docker/setup-qemu-action](https://github.com/docker/setup-qemu-action) from 3.0.0 to 3.1.0.
- [Release notes](https://github.com/docker/setup-qemu-action/releases)
- [Commits](docker/setup-qemu-action@6882732...5927c83)

---
updated-dependencies:
- dependency-name: docker/setup-qemu-action
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

Bumps [docker/setup-buildx-action](https://github.com/docker/setup-buildx-action) from 3.3.0 to 3.4.0.
- [Release notes](https://github.com/docker/setup-buildx-action/releases)
- [Commits](docker/setup-buildx-action@d70bba7...4fd8129)

---
updated-dependencies:
- dependency-name: docker/setup-buildx-action
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* WIP

* Fixes

* Minimize diff

* Progress

* Fix clippy

* Reduce diff to main

* Use RefCell

* Make Pacer use PmtudState

* Renamings

* Fix tests broken by changing PATH_MTU_V6

* WIP

* Finalize

* Update neqo-transport/src/path.rs

Co-authored-by: Martin Thomson <mt@lowentropy.net>
Signed-off-by: Lars Eggert <lars@eggert.org>

* Address comments

* Update neqo-transport/src/pmtud.rs

Co-authored-by: Martin Thomson <mt@lowentropy.net>
Signed-off-by: Lars Eggert <lars@eggert.org>

* Remove PmtudRef from ClassicCongestionControl

* Disable PMTUD by default, except for demo client and server, and simulator

* Fix clippy

* Make Pmtud part of Pacer only

* These are now `const_assert`s

* Cleanups

* Address more comments

* Cleanups

* Add some initial tests

* Fix clippy

* Minimize diff

* Probe with non-padding data

* Search table based on TMA paper

* Deal with PMTU reductions

* Fix crypto invocation limits

* Fix comment

* More comments

* Add PMTU_RAISE_TIMER

* Lost PMTUD probes do not elicit a congestion control reaction

* Update pacer when MTU changes

* Better way to update pacer

* Fix last commit

* Potential fix for bench

* Update neqo-transport/src/pmtud.rs

Co-authored-by: Max Inden <mail@max-inden.de>
Signed-off-by: Lars Eggert <lars@eggert.org>

* Update neqo-transport/src/pmtud.rs

Co-authored-by: Max Inden <mail@max-inden.de>
Signed-off-by: Lars Eggert <lars@eggert.org>

* Update neqo-transport/src/pmtud.rs

Co-authored-by: Max Inden <mail@max-inden.de>
Signed-off-by: Lars Eggert <lars@eggert.org>

* Update neqo-transport/src/cc/classic_cc.rs

Co-authored-by: Max Inden <mail@max-inden.de>
Signed-off-by: Lars Eggert <lars@eggert.org>

* Undo

* Simplifications

* rustfmt

* Disarm raise timer when it fired

* test script that triggers the bug

* Revert test.sh (modulo bug fix)

* Increase coverage

* Better test

* Another test

* let Some(...) instead of testing

Co-authored-by: Max Inden <mail@max-inden.de>
Signed-off-by: Martin Thomson <mt@lowentropy.net>

* Fix

* static_assertions to dev-dependencies

* Panic on failure

* Fixes after merge

* Make test-only plpmtu() panic on error

* set_confirmed

* invocations_base -> largest_packet_len

* Simplify

* Set builder limit in output_path()

* fmt

* A bunch of changes based on Martin's review

* Avoid spurious PMTUD restarts better. Possible perf. optimizations.

* Improve coverage

* Enable PMTUD for simulator

* Update neqo-transport/src/connection/mod.rs

Co-authored-by: Max Inden <mail@max-inden.de>
Signed-off-by: Lars Eggert <lars@eggert.org>

* Update neqo-transport/src/connection/mod.rs

Co-authored-by: Max Inden <mail@max-inden.de>
Signed-off-by: Lars Eggert <lars@eggert.org>

* Update neqo-transport/src/connection/mod.rs

Co-authored-by: Max Inden <mail@max-inden.de>
Signed-off-by: Lars Eggert <lars@eggert.org>

* Update neqo-transport/src/packet/mod.rs

Co-authored-by: Max Inden <mail@max-inden.de>
Signed-off-by: Lars Eggert <lars@eggert.org>

* Update neqo-transport/src/stats.rs

Co-authored-by: Max Inden <mail@max-inden.de>
Signed-off-by: Lars Eggert <lars@eggert.org>

* Suggestions from Max

* Update neqo-transport/src/pmtud.rs

Co-authored-by: Martin Thomson <mt@lowentropy.net>
Signed-off-by: Lars Eggert <lars@eggert.org>

* Suggestions from Martin

* More suggestions

* Add TODO

* Use filter fn

Follow-up on #1903 (comment)

* clippy

* Fixes

* doc fix

* refactor(pmtud): implement Copy for Probe

`Probe` is a small simple enum on the stack, thus convention is to implement
`Copy` instead of only `Clone` with a call to `clone()`.

The following helped me in the past:

> When should my type be Copy?
>
> Generally speaking, if your type can implement Copy, it should. Keep in mind,
> though, that implementing Copy is part of the public API of your type. If the
> type might become non-Copy in the future, it could be prudent to omit the Copy
> implementation now, to avoid a breaking API change.

https://doc.rust-lang.org/std/marker/trait.Copy.html#when-should-my-type-be-copy

* Make search_tables identical length, and deal with the fallout

* More

---------

Signed-off-by: Lars Eggert <lars@eggert.org>
Signed-off-by: Martin Thomson <mt@lowentropy.net>
Co-authored-by: Martin Thomson <mt@lowentropy.net>
Co-authored-by: Max Inden <mail@max-inden.de>

* ci: Fix the fuzz targets

And check them during CI, so they keep building.

* Fix

* Fix

* Add cargo-fuzz

* Fix the fuzzers

* Fix two clippy nightly things while I'm here

The use of `Vec` here is unnecessary.  We can use a fixed sized array
instead.

The performance gain here is negligible, but the code becomes
cleaner, so that's a win.

* ci: Use cargo-machete to check for unused dependencies

And remove the ones it found.

* --with-metadata has false positives

* Remove --no-fallback

---------

Signed-off-by: Lars Eggert <lars@eggert.org>

#1983)

Bumps [actions/dependency-review-action](https://github.com/actions/dependency-review-action) from 4.3.3 to 4.3.4.
- [Release notes](https://github.com/actions/dependency-review-action/releases)
- [Commits](actions/dependency-review-action@72eb03d...5a2ce3f)

---
updated-dependencies:
- dependency-name: actions/dependency-review-action
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* ci: Trigger secondary checks off `check.yml`

Might save some trees.

* quotes?

This reverts commit 735be28.

* ci: Bump Linux CI runners to `ubuntu-24.04`

Which is still being rolled out, but has a bunch of advantages for us.

* Try libnss3-dev deb

* See if updating fixes it

* Just upgrade

* NEEDRESTART_MODE

* sudo apt remove needrestart

* Again

* Again

* Omit `python-version`

* 3.8 -> 3.12

* ldd

* ~/.pki

* CHACHA20

* rm test-fixture/db/pkcs11.txt

* Minimize diff

* Print APT info

* Again

* Use libnss3-dev

* Cleanup

* Try and avoid azure mirror

* Again

* Again

* Again

* Again

* Again

* Again

* Again

* Again

* Again

* Hunch

* Try

* Minimize

* 24.04 -> latest

* ci: Cache NSS build

This should speed up CI.

* steps

* !boolean

* Minimize

* path

* arch

* Fix

* Fix

* Fix

* More optimizations

* Again

* Again

* Again

* Again

* ci: Try and drop more permissions

* Again

So we get an indication of which test might be hanging/slow.