referrerpolicy=no-referrer-when-downgrade
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
// Copyright (C) Parity Technologies (UK) Ltd.
// This file is part of Polkadot.

// Polkadot is free software: you can redistribute it and/or modify
// it under the terms of the GNU General Public License as published by
// the Free Software Foundation, either version 3 of the License, or
// (at your option) any later version.

// Polkadot is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
// GNU General Public License for more details.

// You should have received a copy of the GNU General Public License
// along with Polkadot.  If not, see <http://www.gnu.org/licenses/>.

//! # Sending and receiving of `DisputeRequest`s.
//!
//! This subsystem essentially consists of two parts:
//!
//! - a sender
//! - and a receiver
//!
//! The sender is responsible for getting our vote out, see `sender`. The receiver handles
//! incoming [`DisputeRequest`](v1::DisputeRequest)s and offers spam protection, see `receiver`.

use std::time::Duration;

use futures::{channel::mpsc, FutureExt, StreamExt, TryFutureExt};

use polkadot_node_network_protocol::authority_discovery::AuthorityDiscovery;
use polkadot_node_subsystem_util::nesting_sender::NestingSender;
use sp_keystore::KeystorePtr;

use polkadot_node_network_protocol::request_response::{incoming::IncomingRequestReceiver, v1};
use polkadot_node_primitives::DISPUTE_WINDOW;
use polkadot_node_subsystem::{
	messages::DisputeDistributionMessage, overseer, FromOrchestra, OverseerSignal,
	SpawnedSubsystem, SubsystemError,
};
use polkadot_node_subsystem_util::{runtime, runtime::RuntimeInfo};

/// ## The sender [`DisputeSender`]
///
/// The sender (`DisputeSender`) keeps track of live disputes and makes sure our vote gets out for
/// each one of those. The sender is responsible for sending our vote to each validator
/// participating in the dispute and to each authority currently authoring blocks. The sending can
/// be initiated by sending `DisputeDistributionMessage::SendDispute` message to this subsystem.
///
/// In addition the `DisputeSender` will query the coordinator for active disputes on each
/// [`DisputeSender::update_leaves`] call and will initiate sending (start a `SendTask`) for every,
/// to this subsystem, unknown dispute. This is to make sure, we get our vote out, even on
/// restarts.
///
/// The actual work of sending and keeping track of transmission attempts to each validator for a
/// particular dispute are done by [`SendTask`].  The purpose of the `DisputeSender` is to keep
/// track of all ongoing disputes and start and clean up `SendTask`s accordingly.
mod sender;
use self::sender::{DisputeSender, DisputeSenderMessage};

/// ## The receiver [`DisputesReceiver`]
///
/// The receiving side is implemented as `DisputesReceiver` and is run as a separate long running
/// task within this subsystem ([`DisputesReceiver::run`]).
///
/// Conceptually all the receiver has to do, is waiting for incoming requests which are passed in
/// via a dedicated channel and forwarding them to the dispute coordinator via
/// `DisputeCoordinatorMessage::ImportStatements`. Being the interface to the network and untrusted
/// nodes, the reality is not that simple of course. Before importing statements the receiver will
/// batch up imports as well as possible for efficient imports while maintaining timely dispute
/// resolution and handling of spamming validators:
///
/// - Drop all messages from non validator nodes, for this it requires the [`AuthorityDiscovery`]
/// service.
/// - Drop messages from a node, if it sends at a too high rate.
/// - Filter out duplicate messages (over some period of time).
/// - Drop any obviously invalid votes (invalid signatures for example).
/// - Ban peers whose votes were deemed invalid.
///
/// In general dispute-distribution works on limiting the work the dispute-coordinator will have to
/// do, while at the same time making it aware of new disputes as fast as possible.
///
/// For successfully imported votes, we will confirm the receipt of the message back to the sender.
/// This way a received confirmation guarantees, that the vote has been stored to disk by the
/// receiver.
mod receiver;
use self::receiver::DisputesReceiver;

/// Error and [`Result`] type for this subsystem.
mod error;
use error::{log_error, Error, FatalError, FatalResult, Result};

#[cfg(test)]
mod tests;

mod metrics;
//// Prometheus `Metrics` for dispute distribution.
pub use metrics::Metrics;

const LOG_TARGET: &'static str = "parachain::dispute-distribution";

/// Rate limit on the `receiver` side.
///
/// If messages from one peer come in at a higher rate than every `RECEIVE_RATE_LIMIT` on average,
/// we start dropping messages from that peer to enforce that limit.
pub const RECEIVE_RATE_LIMIT: Duration = Duration::from_millis(100);

/// Rate limit on the `sender` side.
///
/// In order to not hit the `RECEIVE_RATE_LIMIT` on the receiving side, we limit out sending rate as
/// well.
///
/// We add 50ms extra, just to have some save margin to the `RECEIVE_RATE_LIMIT`.
pub const SEND_RATE_LIMIT: Duration = RECEIVE_RATE_LIMIT.saturating_add(Duration::from_millis(50));

/// The dispute distribution subsystem.
pub struct DisputeDistributionSubsystem<AD> {
	/// Easy and efficient runtime access for this subsystem.
	runtime: RuntimeInfo,

	/// Sender for our dispute requests.
	disputes_sender: DisputeSender<DisputeSenderMessage>,

	/// Receive messages from `DisputeSender` background tasks.
	sender_rx: mpsc::Receiver<DisputeSenderMessage>,

	/// Receiver for incoming requests.
	req_receiver: Option<IncomingRequestReceiver<v1::DisputeRequest>>,

	/// Authority discovery service.
	authority_discovery: AD,

	/// Metrics for this subsystem.
	metrics: Metrics,
}

#[overseer::subsystem(DisputeDistribution, error = SubsystemError, prefix = self::overseer)]
impl<Context, AD> DisputeDistributionSubsystem<AD>
where
	<Context as overseer::DisputeDistributionContextTrait>::Sender:
		overseer::DisputeDistributionSenderTrait + Sync + Send,
	AD: AuthorityDiscovery + Clone,
{
	fn start(self, ctx: Context) -> SpawnedSubsystem {
		let future = self
			.run(ctx)
			.map_err(|e| SubsystemError::with_origin("dispute-distribution", e))
			.boxed();

		SpawnedSubsystem { name: "dispute-distribution-subsystem", future }
	}
}

#[overseer::contextbounds(DisputeDistribution, prefix = self::overseer)]
impl<AD> DisputeDistributionSubsystem<AD>
where
	AD: AuthorityDiscovery + Clone,
{
	/// Create a new instance of the dispute distribution.
	pub fn new(
		keystore: KeystorePtr,
		req_receiver: IncomingRequestReceiver<v1::DisputeRequest>,
		authority_discovery: AD,
		metrics: Metrics,
	) -> Self {
		let runtime = RuntimeInfo::new_with_config(runtime::Config {
			keystore: Some(keystore),
			session_cache_lru_size: DISPUTE_WINDOW.get(),
		});
		let (tx, sender_rx) = NestingSender::new_root(1);
		let disputes_sender = DisputeSender::new(tx, metrics.clone());
		Self {
			runtime,
			disputes_sender,
			sender_rx,
			req_receiver: Some(req_receiver),
			authority_discovery,
			metrics,
		}
	}

	/// Start processing work as passed on from the Overseer.
	async fn run<Context>(mut self, mut ctx: Context) -> std::result::Result<(), FatalError> {
		let receiver = DisputesReceiver::new(
			ctx.sender().clone(),
			self.req_receiver
				.take()
				.expect("Must be provided on `new` and we take ownership here. qed."),
			self.authority_discovery.clone(),
			self.metrics.clone(),
		);
		ctx.spawn("disputes-receiver", receiver.run().boxed())
			.map_err(FatalError::SpawnTask)?;

		// Process messages for sending side.
		//
		// Note: We want the sender to be rate limited and we are currently taking advantage of the
		// fact that the root task of this subsystem is only concerned with sending: Functions of
		// `DisputeSender` might back pressure if the rate limit is hit, which will slow down this
		// loop. If this fact ever changes, we will likely need another task.
		loop {
			let message = MuxedMessage::receive(&mut ctx, &mut self.sender_rx).await;
			match message {
				MuxedMessage::Subsystem(result) => {
					let result = match result? {
						FromOrchestra::Signal(signal) => {
							match self.handle_signals(&mut ctx, signal).await {
								Ok(SignalResult::Conclude) => return Ok(()),
								Ok(SignalResult::Continue) => Ok(()),
								Err(f) => Err(f),
							}
						},
						FromOrchestra::Communication { msg } =>
							self.handle_subsystem_message(&mut ctx, msg).await,
					};
					log_error(result, "on FromOrchestra")?;
				},
				MuxedMessage::Sender(result) => {
					let result = self
						.disputes_sender
						.on_message(
							&mut ctx,
							&mut self.runtime,
							result.ok_or(FatalError::SenderExhausted)?,
						)
						.await
						.map_err(Error::Sender);
					log_error(result, "on_message")?;
				},
			}
		}
	}

	/// Handle overseer signals.
	async fn handle_signals<Context>(
		&mut self,
		ctx: &mut Context,
		signal: OverseerSignal,
	) -> Result<SignalResult> {
		match signal {
			OverseerSignal::Conclude => return Ok(SignalResult::Conclude),
			OverseerSignal::ActiveLeaves(update) => {
				self.disputes_sender.update_leaves(ctx, &mut self.runtime, update).await?;
			},
			OverseerSignal::BlockFinalized(_, _) => {},
		};
		Ok(SignalResult::Continue)
	}

	/// Handle `DisputeDistributionMessage`s.
	async fn handle_subsystem_message<Context>(
		&mut self,
		ctx: &mut Context,
		msg: DisputeDistributionMessage,
	) -> Result<()> {
		match msg {
			DisputeDistributionMessage::SendDispute(dispute_msg) =>
				self.disputes_sender.start_sender(ctx, &mut self.runtime, dispute_msg).await?,
		}
		Ok(())
	}
}

/// Messages to be handled in this subsystem.
#[derive(Debug)]
enum MuxedMessage {
	/// Messages from other subsystems.
	Subsystem(FatalResult<FromOrchestra<DisputeDistributionMessage>>),
	/// Messages from spawned sender background tasks.
	Sender(Option<DisputeSenderMessage>),
}

#[overseer::contextbounds(DisputeDistribution, prefix = self::overseer)]
impl MuxedMessage {
	async fn receive<Context>(
		ctx: &mut Context,
		from_sender: &mut mpsc::Receiver<DisputeSenderMessage>,
	) -> Self {
		// We are only fusing here to make `select` happy, in reality we will quit if the stream
		// ends.
		let from_overseer = ctx.recv().fuse();
		futures::pin_mut!(from_overseer, from_sender);
		// We select biased to make sure we finish up loose ends, before starting new work.
		futures::select_biased!(
			msg = from_sender.next() => MuxedMessage::Sender(msg),
			msg = from_overseer => MuxedMessage::Subsystem(msg.map_err(FatalError::SubsystemReceive)),
		)
	}
}

/// Result of handling signal from overseer.
enum SignalResult {
	/// Overseer asked us to conclude.
	Conclude,
	/// We can continue processing events.
	Continue,
}