hopr_network_types/
udp.rs

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
use crate::utils::SocketAddrStr;
use futures::{pin_mut, ready, FutureExt, Sink, SinkExt};
use std::fmt::Debug;
use std::io::ErrorKind;
use std::num::NonZeroUsize;
use std::pin::Pin;
use std::sync::{Arc, OnceLock};
use std::task::{Context, Poll};
use tokio::net::UdpSocket;
use tracing::{debug, error, trace, warn};

type BoxIoSink<T> = Box<dyn Sink<T, Error = std::io::Error> + Send + Unpin>;

#[cfg(all(feature = "prometheus", not(test)))]
lazy_static::lazy_static! {
    static ref METRIC_UDP_INGRESS_LEN: hopr_metrics::MultiHistogram =
        hopr_metrics::MultiHistogram::new(
            "hopr_udp_ingress_packet_len",
            "UDP packet lengths on ingress per counterparty",
            vec![20.0, 40.0, 80.0, 160.0, 320.0, 640.0, 1280.0, 2560.0, 5120.0],
            &["counterparty"]
    ).unwrap();
    static ref METRIC_UDP_EGRESS_LEN: hopr_metrics::MultiHistogram =
        hopr_metrics::MultiHistogram::new(
            "hopr_udp_egress_packet_len",
            "UDP packet lengths on egress per counterparty",
            vec![20.0, 40.0, 80.0, 160.0, 320.0, 640.0, 1280.0, 2560.0, 5120.0],
            &["counterparty"]
    ).unwrap();
}

/// Mimics TCP-like stream functionality on a UDP socket by restricting it to a single
/// counterparty and implements [`tokio::io::AsyncRead`] and [`tokio::io::AsyncWrite`].
/// The instance is always constructed using a [`UdpStreamBuilder`].
///
/// To set a counterparty, one of the following must happen:
/// 1) setting it during build via [`UdpStreamBuilder::with_counterparty`]
/// 2) receiving some data from the other side.
///
/// Whatever of the above happens first, sets the counterparty.
/// Once the counterparty is set, all data sent and received will be sent or filtered by this
/// counterparty address.
///
/// If data from another party is received, an error is raised, unless the object has been constructed
/// with [`ForeignDataMode::Discard`] or [`ForeignDataMode::Accept`] setting.
///
/// This object is also capable of parallel processing on a UDP socket.
/// If [parallelism](UdpStreamBuilder::with_receiver_parallelism) is set, the instance will create
/// multiple sockets with `SO_REUSEADDR` and spin parallel tasks that will coordinate data and
/// transmission retrieval using these sockets. This is driven by RX/TX MPMC queues, which are
/// per-default unbounded (see [queue size](UdpStreamBuilder::with_queue_size) for details).
pub struct ConnectedUdpStream {
    socket_handles: Vec<tokio::task::JoinHandle<()>>,
    ingress_rx: Box<dyn tokio::io::AsyncRead + Send + Unpin>,
    egress_tx: Option<BoxIoSink<Box<[u8]>>>,
    counterparty: Arc<OnceLock<SocketAddrStr>>,
    bound_to: std::net::SocketAddr,
}

/// Defines what happens when data from another [`SocketAddr`](std::net::SocketAddr) arrives
/// into the [`ConnectedUdpStream`] (other than the one that is considered a counterparty for that
/// instance).
#[derive(Clone, Copy, Debug, PartialEq, Eq, Default)]
pub enum ForeignDataMode {
    /// Foreign data are simply discarded.
    Discard,
    /// Foreign data are accepted as if they arrived from the set counterparty.
    Accept,
    /// Error is raised on the `poll_read` attempt.
    #[default]
    Error,
}

/// Determines how many parallel readers or writer sockets should be bound in [`ConnectedUdpStream`].
///
/// Each UDP socket is bound with `SO_REUSEADDR` and `SO_REUSEPORT` to facilitate parallel processing
/// of send and/or receive operations.
///
/// **NOTE**: This is a Linux-specific optimization, and it will have no effect on other systems.
///
/// - If some [`Specific`](UdpStreamParallelism::Specific) value `n` > 0 is given, the [`ConnectedUdpStream`] will bind `n` sockets.
/// - If [`Auto`](UdpStreamParallelism::Auto) is given, the number of sockets bound by [`ConnectedUdpStream`] is determined by [`std::thread::available_parallelism`].
///
/// The default is `Specific(1)`.
///
/// Always use [`into_num_tasks`](UdpStreamParallelism::into_num_tasks) or [`split_evenly_with`](UdpStreamParallelism::split_evenly_with)
/// to determine the correct number of sockets to spawn.
#[derive(Copy, Clone, Debug, PartialEq, Eq)]
pub enum UdpStreamParallelism {
    /// Bind as many sender or receiver sockets as given by [`std::thread::available_parallelism`].
    Auto,
    /// Bind specific number of sender or receiver sockets.
    Specific(NonZeroUsize),
}

impl Default for UdpStreamParallelism {
    fn default() -> Self {
        Self::Specific(NonZeroUsize::MIN)
    }
}

impl UdpStreamParallelism {
    fn avail_parallelism() -> usize {
        // On non-Linux system this will always default to 1, since the
        // multiple UDP socket optimization is not possible for those platforms.
        std::thread::available_parallelism()
            .map(|n| {
                if cfg!(target_os = "linux") {
                    n
                } else {
                    NonZeroUsize::MIN
                }
            })
            .unwrap_or_else(|e| {
                warn!(error = %e, "failed to determine available parallelism, defaulting to 1.");
                NonZeroUsize::MIN
            })
            .into()
    }

    /// Returns the number of sockets for this and the `other` instance
    /// when they evenly split the available CPU parallelism.
    pub fn split_evenly_with(self, other: UdpStreamParallelism) -> (usize, usize) {
        let cpu_half = (Self::avail_parallelism() / 2).max(1);

        match (self, other) {
            (UdpStreamParallelism::Auto, UdpStreamParallelism::Auto) => (cpu_half, cpu_half),
            (UdpStreamParallelism::Specific(a), UdpStreamParallelism::Auto) => {
                let a = cpu_half.min(a.into());
                (a, cpu_half * 2 - a)
            }
            (UdpStreamParallelism::Auto, UdpStreamParallelism::Specific(b)) => {
                let b = cpu_half.min(b.into());
                (cpu_half * 2 - b, b)
            }
            (UdpStreamParallelism::Specific(a), UdpStreamParallelism::Specific(b)) => {
                (cpu_half.min(a.into()), cpu_half.min(b.into()))
            }
        }
    }

    /// Calculates the actual number of tasks for this instance.
    ///
    /// The returned value is never more than the maximum available CPU parallelism.
    pub fn into_num_tasks(self) -> usize {
        let avail_parallelism = Self::avail_parallelism();
        match self {
            UdpStreamParallelism::Auto => avail_parallelism,
            UdpStreamParallelism::Specific(n) => usize::from(n).min(avail_parallelism),
        }
    }
}

impl From<usize> for UdpStreamParallelism {
    fn from(value: usize) -> Self {
        NonZeroUsize::new(value).map(Self::Specific).unwrap_or_default()
    }
}

impl From<Option<usize>> for UdpStreamParallelism {
    fn from(value: Option<usize>) -> Self {
        value.map(UdpStreamParallelism::from).unwrap_or_default()
    }
}

/// Builder object for the [`ConnectedUdpStream`].
///
/// If you wish to use defaults, do `UdpStreamBuilder::default().build(addr)`.
#[derive(Debug, Clone)]
pub struct UdpStreamBuilder {
    foreign_data_mode: ForeignDataMode,
    buffer_size: usize,
    queue_size: Option<usize>,
    receiver_parallelism: UdpStreamParallelism,
    sender_parallelism: UdpStreamParallelism,
    counterparty: Option<std::net::SocketAddr>,
}

impl Default for UdpStreamBuilder {
    fn default() -> Self {
        Self {
            buffer_size: 2048,
            foreign_data_mode: Default::default(),
            queue_size: None,
            receiver_parallelism: Default::default(),
            sender_parallelism: Default::default(),
            counterparty: None,
        }
    }
}

impl UdpStreamBuilder {
    /// Defines the behavior when data from an unexpected source arrive into the socket.
    /// See [`ForeignDataMode`] for details.
    ///
    /// Default is [`ForeignDataMode::Error`].
    pub fn with_foreign_data_mode(mut self, mode: ForeignDataMode) -> Self {
        self.foreign_data_mode = mode;
        self
    }

    /// The size of the UDP receive buffer.
    ///
    /// This size must be at least the size of the MTU, otherwise the unread UDP data that
    /// does not fit this buffer will be discarded.
    ///
    /// Default is 2048.
    pub fn with_buffer_size(mut self, buffer_size: usize) -> Self {
        self.buffer_size = buffer_size;
        self
    }

    /// Size of the TX/RX queue that dispatches data of reads from/writings to
    /// the sockets.
    ///
    /// This an important back-pressure mechanism when dispatching received data from
    /// fast senders.
    /// Reduces the maximum memory consumed by the object, which is given by:
    /// [`buffer_size`](UdpStreamBuilder::with_buffer_size) *
    /// [`queue_size`](UdpStreamBuilder::with_queue_size)
    ///
    /// Default is unbounded.
    pub fn with_queue_size(mut self, queue_size: usize) -> Self {
        self.queue_size = Some(queue_size);
        self
    }

    /// Sets how many parallel receiving sockets should be bound.
    ///
    /// Has no effect on non-Linux machines. See [`UdpStreamParallelism`] for details.
    ///
    /// Default is `1`.
    pub fn with_receiver_parallelism<T: Into<UdpStreamParallelism>>(mut self, parallelism: T) -> Self {
        self.receiver_parallelism = parallelism.into();
        self
    }

    /// Sets how many parallel sending sockets should be bound.
    ///
    /// Has no effect on non-Linux machines. See [`UdpStreamParallelism`] for details.
    ///
    /// Default is `1`.
    pub fn with_sender_parallelism<T: Into<UdpStreamParallelism>>(mut self, parallelism: T) -> Self {
        self.sender_parallelism = parallelism.into();
        self
    }

    /// Sets the expected counterparty for data sent/received by the UDP sockets.
    ///
    /// If not specified, the counterparty is determined from the first packet received.
    /// However, no data can be sent up until this point.
    /// Therefore, the value must be set if data are sent first rather than received.
    /// If data is expected to be received first, the value does not need to be set.
    ///
    /// See [`ConnectedUdpStream`] and [`ForeignDataMode`] for details.
    ///
    /// Default is none.
    pub fn with_counterparty(mut self, counterparty: std::net::SocketAddr) -> Self {
        self.counterparty = Some(counterparty);
        self
    }

    /// Builds the [`ConnectedUdpStream`] with UDP socket(s) bound to `bind_addr`.
    ///
    /// The number of RX sockets bound is determined by [receiver parallelism](UdpStreamBuilder::with_receiver_parallelism),
    /// and similarly, the number of TX sockets bound is determined by [sender parallelism](UdpStreamBuilder::with_sender_parallelism).
    /// On non-Linux platforms, only a single receiver and sender will be bound, regardless of the above.
    ///
    /// The returned instance is always ready to receive data.
    /// It is also ready to send data
    /// if the [counterparty](UdpStreamBuilder::with_counterparty) has been set.
    ///
    /// If `bind_addr` yields multiple addresses, binding will be attempted with each of the addresses
    /// until one succeeds. If none of the addresses succeed in binding the socket(s),
    /// the `AddrNotAvailable` error is returned.
    ///
    /// Note that wildcard addresses (such as `0.0.0.0`) are *not* considered as multiple addresses,
    /// and such socket(s) will bind to all available interfaces at the system level.
    pub fn build<A: std::net::ToSocketAddrs>(self, bind_addr: A) -> std::io::Result<ConnectedUdpStream> {
        let (num_rx_socks, num_tx_socks) = self.receiver_parallelism.split_evenly_with(self.sender_parallelism);

        let counterparty = Arc::new(
            self.counterparty
                .map(|s| OnceLock::from(SocketAddrStr::from(s)))
                .unwrap_or_default(),
        );
        let ((ingress_tx, ingress_rx), (egress_tx, egress_rx)) = if let Some(q) = self.queue_size {
            (flume::bounded(q), flume::bounded(q))
        } else {
            (flume::unbounded(), flume::unbounded())
        };

        let num_socks_to_bind = num_rx_socks.max(num_tx_socks);
        let mut socket_handles = Vec::with_capacity(num_socks_to_bind);
        let mut bound_addr: Option<std::net::SocketAddr> = None;

        // Try binding on all network addresses in `bind_addr`
        for binding_to in bind_addr.to_socket_addrs()? {
            debug!(
                %binding_to,
                num_socks_to_bind, num_rx_socks, num_tx_socks, "binding UDP stream"
            );

            // TODO: split bound sockets into a separate cloneable object

            // Try to bind sockets on the current network interface address
            (0..num_socks_to_bind)
                .map(|sock_id| {
                    let domain = match &binding_to {
                        std::net::SocketAddr::V4(_) => socket2::Domain::IPV4,
                        std::net::SocketAddr::V6(_) => socket2::Domain::IPV6,
                    };

                    // Bind a new non-blocking UDP socket
                    let sock = socket2::Socket::new(domain, socket2::Type::DGRAM, None)?;
                    if num_socks_to_bind > 1 {
                        sock.set_reuse_address(true)?; // Needed for every next socket with non-wildcard IP
                        sock.set_reuse_port(true)?; // Needed on Linux to evenly distribute datagrams
                    }
                    sock.set_nonblocking(true)?;
                    sock.bind(&bound_addr.unwrap_or(binding_to).into())?;

                    // Determine the address we bound this socket to, so we can also bind the others
                    let socket_bound_addr = sock
                        .local_addr()?
                        .as_socket()
                        .ok_or(std::io::Error::other("invalid socket type"))?;

                    match bound_addr {
                        None => bound_addr = Some(socket_bound_addr),
                        Some(addr) if addr != socket_bound_addr => {
                            return Err(std::io::Error::other(format!(
                                "inconsistent binding address {addr} != {socket_bound_addr} on socket id {sock_id}"
                            )))
                        }
                        _ => {}
                    }

                    let sock = Arc::new(UdpSocket::from_std(sock.into())?);
                    debug!(
                        socket_id = sock_id,
                        addr = %socket_bound_addr,
                        "bound UDP socket"
                    );

                    Ok((sock_id, sock))
                })
                .filter_map(|result| match result {
                    Ok(bound) => Some(bound),
                    Err(e) => {
                        error!(
                            %binding_to,
                            "failed to bind udp socket: {e}"
                        );
                        None
                    }
                })
                .for_each(|(sock_id, sock)| {
                    if sock_id < num_tx_socks {
                        socket_handles.push(tokio::task::spawn(ConnectedUdpStream::setup_tx_queue(
                            sock_id,
                            sock.clone(),
                            egress_rx.clone(),
                            counterparty.clone(),
                        )));
                    }
                    if sock_id < num_rx_socks {
                        socket_handles.push(tokio::task::spawn(ConnectedUdpStream::setup_rx_queue(
                            sock_id,
                            sock.clone(),
                            ingress_tx.clone(),
                            counterparty.clone(),
                            self.foreign_data_mode,
                            self.buffer_size,
                        )));
                    }
                });
        }

        Ok(ConnectedUdpStream {
            ingress_rx: Box::new(tokio_util::io::StreamReader::new(ingress_rx.into_stream())),
            egress_tx: Some(Box::new(
                egress_tx
                    .into_sink()
                    .sink_map_err(|e| std::io::Error::other(e.to_string())),
            )),
            socket_handles,
            counterparty,
            bound_to: bound_addr.ok_or(ErrorKind::AddrNotAvailable)?,
        })
    }
}

impl ConnectedUdpStream {
    /// Creates a receiver queue for the UDP stream.
    fn setup_rx_queue(
        socket_id: usize,
        sock_rx: Arc<UdpSocket>,
        ingress_tx: flume::Sender<std::io::Result<tokio_util::bytes::Bytes>>,
        counterparty: Arc<OnceLock<SocketAddrStr>>,
        foreign_data_mode: ForeignDataMode,
        buf_size: usize,
    ) -> futures::future::BoxFuture<'static, ()> {
        let counterparty_rx = counterparty.clone();
        async move {
            let mut buffer = vec![0u8; buf_size];
            let mut done = false;
            loop {
                // Read data from the socket
                let out_res = match sock_rx.recv_from(&mut buffer).await {
                    Ok((read, read_addr)) if read > 0 => {
                        trace!(
                            socket_id,
                            udp_bound_addr = ?sock_rx.local_addr(),
                            bytes = read,
                            from = %read_addr,
                            "received data from"
                        );

                        let addr = counterparty_rx.get_or_init(|| read_addr.into());

                        #[cfg(all(feature = "prometheus", not(test)))]
                        METRIC_UDP_INGRESS_LEN.observe(&[addr.as_str()], read as f64);

                        // If the data is from a counterparty, or we accept anything, pass it
                        if read_addr.eq(addr) || foreign_data_mode == ForeignDataMode::Accept {
                            let out_buffer = tokio_util::bytes::Bytes::copy_from_slice(&buffer[..read]);
                            Some(Ok(out_buffer))
                        } else {
                            match foreign_data_mode {
                                ForeignDataMode::Discard => {
                                    // Don't even bother sending an error about discarded stuff
                                    warn!(
                                        socket_id,
                                        udp_bound_addr = ?sock_rx.local_addr(),
                                        ?read_addr,
                                        expected_addr = ?addr,
                                        "discarded data, which didn't come from the expected address"
                                    );
                                    None
                                }
                                ForeignDataMode::Error => {
                                    // Terminate here, the ingress_tx gets dropped
                                    done = true;
                                    Some(Err(std::io::Error::new(
                                        ErrorKind::ConnectionRefused,
                                        "data from foreign client not allowed",
                                    )))
                                }
                                // ForeignDataMode::Accept has been handled above
                                _ => unreachable!(),
                            }
                        }
                    }
                    Ok(_) => {
                        // Read EOF, terminate here, the ingress_tx gets dropped
                        trace!(
                            socket_id,
                            udp_bound_addr = ?sock_rx.local_addr(),
                            "read EOF on socket"
                        );
                        done = true;
                        None
                    }
                    Err(e) => {
                        // Forward the error
                        debug!(
                            socket_id,
                            udp_bound_addr = ?sock_rx.local_addr(),
                            error = %e,
                            "forwarded error from socket"
                        );
                        done = true;
                        Some(Err(e))
                    }
                };

                // Dispatch the received data to the queue.
                // If the underlying queue is bounded, it will wait until there is space.
                if let Some(out_res) = out_res {
                    if let Err(err) = ingress_tx.send_async(out_res).await {
                        error!(
                            socket_id,
                            udp_bound_addr = ?sock_rx.local_addr(),
                            error = %err,
                            "failed to dispatch received data"
                        );
                        done = true;
                    }
                }

                if done {
                    trace!(
                        socket_id,
                        udp_bound_addr = ?sock_rx.local_addr(),
                        "rx queue done"
                    );
                    break;
                }
            }
        }
        .boxed()
    }

    /// Creates a transmission queue for the UDP stream.
    fn setup_tx_queue(
        socket_id: usize,
        sock_tx: Arc<UdpSocket>,
        egress_rx: flume::Receiver<Box<[u8]>>,
        counterparty: Arc<OnceLock<SocketAddrStr>>,
    ) -> futures::future::BoxFuture<'static, ()> {
        let counterparty_tx = counterparty.clone();
        async move {
            loop {
                match egress_rx.recv_async().await {
                    Ok(data) => {
                        if let Some(target) = counterparty_tx.get() {
                            if let Err(e) = sock_tx.send_to(&data, target.as_ref()).await {
                                error!(
                                    ?socket_id,
                                    udp_bound_addr = ?sock_tx.local_addr(),
                                    ?target,
                                    error = %e,
                                    "failed to send data"
                                );
                            }
                            trace!(socket_id, bytes = data.len(), ?target, "sent bytes to");

                            #[cfg(all(feature = "prometheus", not(test)))]
                            METRIC_UDP_EGRESS_LEN.observe(&[target.as_str()], data.len() as f64);
                        } else {
                            error!(
                                ?socket_id,
                                udp_bound_addr = ?sock_tx.local_addr(),
                                "cannot send data, counterparty not set"
                            );
                            break;
                        }
                    }
                    Err(e) => {
                        error!(
                            ?socket_id,
                            udp_bound_addr = ?sock_tx.local_addr(),
                            error = %e,
                            "cannot receive more data from egress channel"
                        );
                        break;
                    }
                }
                trace!(
                    ?socket_id,
                    udp_bound_addr = tracing::field::debug(sock_tx.local_addr()),
                    "tx queue done"
                );
            }
        }
        .boxed()
    }

    /// Local address that all UDP sockets in this instance are bound to.
    pub fn bound_address(&self) -> &std::net::SocketAddr {
        &self.bound_to
    }

    /// Creates a new [builder](UdpStreamBuilder).
    pub fn builder() -> UdpStreamBuilder {
        UdpStreamBuilder::default()
    }
}

impl Drop for ConnectedUdpStream {
    fn drop(&mut self) {
        self.socket_handles.iter().for_each(|handle| {
            handle.abort();
        })
    }
}

impl tokio::io::AsyncRead for ConnectedUdpStream {
    fn poll_read(
        mut self: Pin<&mut Self>,
        cx: &mut Context<'_>,
        buf: &mut tokio::io::ReadBuf<'_>,
    ) -> Poll<std::io::Result<()>> {
        trace!(
            remaining_bytes = buf.remaining(),
            counterparty = ?self.counterparty.get(),
            "polling read of from udp stream",
        );
        match Pin::new(&mut self.ingress_rx).poll_read(cx, buf) {
            Poll::Ready(Ok(())) => {
                let read = buf.filled().len();
                trace!(bytes = read, "read bytes");
                Poll::Ready(Ok(()))
            }
            Poll::Ready(Err(err)) => Poll::Ready(Err(err)),
            Poll::Pending => Poll::Pending,
        }
    }
}

impl tokio::io::AsyncWrite for ConnectedUdpStream {
    fn poll_write(mut self: Pin<&mut Self>, cx: &mut Context<'_>, buf: &[u8]) -> Poll<std::io::Result<usize>> {
        trace!(
            bytes = buf.len(),
            counterparty = ?self.counterparty.get(),
            "polling write to udp stream",
        );
        if let Some(sender) = &mut self.egress_tx {
            if let Err(e) = ready!(sender.poll_ready_unpin(cx)) {
                return Poll::Ready(Err(e));
            }

            let len = buf.iter().len();
            if let Err(e) = sender.start_send_unpin(Box::from(buf)) {
                return Poll::Ready(Err(e));
            }

            // Explicitly flush after each data sent
            pin_mut!(sender);
            sender.poll_flush(cx).map_ok(|_| len)
        } else {
            Poll::Ready(Err(std::io::Error::new(
                ErrorKind::NotConnected,
                "udp stream is closed",
            )))
        }
    }

    fn poll_flush(mut self: Pin<&mut Self>, cx: &mut Context<'_>) -> Poll<std::io::Result<()>> {
        trace!(
            counterparty = ?self.counterparty.get(),
            "polling flush to udp stream"
        );
        if let Some(sender) = &mut self.egress_tx {
            pin_mut!(sender);
            sender
                .poll_flush(cx)
                .map_err(|err| std::io::Error::other(err.to_string()))
        } else {
            Poll::Ready(Err(std::io::Error::new(
                ErrorKind::NotConnected,
                "udp stream is closed",
            )))
        }
    }

    fn poll_shutdown(mut self: Pin<&mut Self>, cx: &mut Context<'_>) -> Poll<std::io::Result<()>> {
        trace!(
            counterparty = ?self.counterparty.get(),
            "polling close on udp stream"
        );
        // Take the sender to make sure it gets dropped
        let mut taken_sender = self.egress_tx.take();
        if let Some(sender) = &mut taken_sender {
            pin_mut!(sender);
            sender
                .poll_close(cx)
                .map_err(|err| std::io::Error::other(err.to_string()))
        } else {
            Poll::Ready(Err(std::io::Error::new(
                ErrorKind::NotConnected,
                "udp stream is closed",
            )))
        }
    }
}

#[cfg(test)]
mod tests {
    use super::*;
    use anyhow::Context;
    use futures::future::Either;
    use parameterized::parameterized;
    use tokio::io::{AsyncReadExt, AsyncWriteExt};
    use tokio::net::UdpSocket;

    #[parameterized(parallelism = {None, Some(2), Some(0)})]
    #[parameterized_macro(tokio::test)]
    //#[parameterized_macro(test_log::test(tokio::test))]
    async fn basic_udp_stream_tests(parallelism: Option<usize>) -> anyhow::Result<()> {
        const DATA_SIZE: usize = 200;

        let listener = UdpSocket::bind("127.0.0.1:0").await.context("bind listener")?;
        let listen_addr = listener.local_addr()?;

        // Simple echo UDP server
        tokio::task::spawn(async move {
            loop {
                let mut buf = [0u8; DATA_SIZE];
                let (read, addr) = listener.recv_from(&mut buf).await.expect("recv must not fail");
                if read > 0 {
                    assert_eq!(DATA_SIZE, read, "read size must be exactly {DATA_SIZE}");
                    listener.send_to(&buf, addr).await.expect("send must not fail");
                }
            }
        });

        let mut builder = ConnectedUdpStream::builder()
            .with_buffer_size(1024)
            .with_queue_size(512)
            .with_counterparty(listen_addr);

        if let Some(parallelism) = parallelism {
            builder = builder.with_receiver_parallelism(parallelism);
        }

        let mut stream = builder.build(("127.0.0.1", 0)).context("connection")?;

        for _ in 1..1000 {
            let mut w_buf = [0u8; DATA_SIZE];
            hopr_crypto_random::random_fill(&mut w_buf);
            let written = stream.write(&w_buf).await?;
            assert_eq!(written, DATA_SIZE);

            let mut r_buf = [0u8; DATA_SIZE];
            let read = stream.read_exact(&mut r_buf).await?;
            assert_eq!(read, DATA_SIZE);

            assert_eq!(w_buf, r_buf);
        }

        stream.shutdown().await?;

        Ok(())
    }

    #[tokio::test]
    async fn udp_stream_should_process_sequential_writes() -> anyhow::Result<()> {
        const BUF_SIZE: usize = 1024;
        const EXPECTED_DATA_LEN: usize = BUF_SIZE + 500;

        let mut listener = ConnectedUdpStream::builder()
            .with_buffer_size(BUF_SIZE)
            .with_queue_size(512)
            .build(("127.0.0.1", 0))
            .context("bind listener")?;

        let bound_addr = *listener.bound_address();

        let jh = tokio::task::spawn(async move {
            let mut buf = [0u8; BUF_SIZE / 4];
            let mut vec = Vec::<u8>::new();
            loop {
                let sz = listener.read(&mut buf).await.unwrap();
                if sz > 0 {
                    vec.extend_from_slice(&buf[..sz]);
                    if vec.len() >= EXPECTED_DATA_LEN {
                        return vec;
                    }
                } else {
                    return vec;
                }
            }
        });

        let msg = [1u8; EXPECTED_DATA_LEN];
        let sender = UdpSocket::bind(("127.0.0.1", 0)).await.context("bind")?;

        sender.send_to(&msg[..BUF_SIZE], bound_addr).await?;
        sender.send_to(&msg[BUF_SIZE..], bound_addr).await?;

        let timeout = tokio::time::sleep(std::time::Duration::from_millis(1000));
        pin_mut!(timeout);
        pin_mut!(jh);

        match futures::future::select(jh, timeout).await {
            Either::Left((Ok(v), _)) => {
                assert_eq!(v.len(), EXPECTED_DATA_LEN);
                assert_eq!(v.as_slice(), &msg);
                Ok(())
            }
            _ => Err(anyhow::anyhow!("timeout or invalid data")),
        }
    }
}