Asynchronous-backing compatible Aura, not plugged in (#2573)

* rough draft of potential parent search * get things compiling * fmt * add new function to all RelayChainInterface implementations * fix compilation * set slot and timestamp based on relay parent, prepare for find-parent * skeleton of new aura logic * fmt * introduce a collator module in the Aura crate * extract different implementations into own modules * make interface more convenient * docs and todos for lookahead * refactor basic collator to use new collator utility * some more refactoring * finish most of the control flow for new aura * introduce backend as parameter * fix compilation * fix a couple more TODOs * add an `announce_block` function to collator service * announce with barrier * rename block announcement validator to be more specific * fmt * clean up unused import errors * update references to BlockAnnounceValidator * rename unstable_reimpl * add AuraUnincludedSegmentApi * finish rename * integrate AuraUnincludedSegmentApi * add a new block announcement validator for backwards compatibility * add some naive equivocation defenses * rustfmt * clean up remaining TODO [now]s * fmt * try to fix inprocess-interface * actually fix compilation * ignored -> rejected rephrase * fix test compilation * fmt * clippy
2026-05-30 09:21:04 +00:00 · 2023-07-11 13:56:09 +02:00
parent cc5e0ae9ab
commit 406f92ad90
20 changed files with 1588 additions and 548 deletions
@@ -0,0 +1,254 @@
+// Copyright 2023 Parity Technologies (UK) Ltd.
+// This file is part of Cumulus.
+
+// Cumulus 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.
+
+// Cumulus 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 Cumulus.  If not, see <http://www.gnu.org/licenses/>.
+
+/// An import queue which provides some equivocation resistance with lenient trait bounds.
+///
+/// Equivocation resistance in general is a hard problem, as different nodes in the network
+/// may see equivocations in a different order, and therefore may not agree on which blocks
+/// should be thrown out and which ones should be kept.
+use codec::{Decode, Encode};
+use cumulus_client_consensus_common::ParachainBlockImportMarker;
+use lru::LruCache;
+
+use sc_consensus::{
+	import_queue::{BasicQueue, Verifier as VerifierT},
+	BlockImport, BlockImportParams, ForkChoiceStrategy,
+};
+use sc_consensus_aura::standalone as aura_internal;
+use sc_telemetry::{telemetry, TelemetryHandle, CONSENSUS_DEBUG, CONSENSUS_TRACE};
+use sp_api::ProvideRuntimeApi;
+use sp_block_builder::BlockBuilder as BlockBuilderApi;
+use sp_consensus::error::Error as ConsensusError;
+use sp_consensus_aura::{AuraApi, Slot, SlotDuration};
+use sp_core::crypto::Pair;
+use sp_inherents::{CreateInherentDataProviders, InherentDataProvider};
+use sp_runtime::traits::{Block as BlockT, Header as HeaderT};
+use std::{fmt::Debug, num::NonZeroUsize, sync::Arc};
+
+const LRU_WINDOW: usize = 256;
+const EQUIVOCATION_LIMIT: usize = 16;
+
+struct NaiveEquivocationDefender {
+	cache: LruCache<u64, usize>,
+}
+
+impl Default for NaiveEquivocationDefender {
+	fn default() -> Self {
+		NaiveEquivocationDefender {
+			cache: LruCache::new(NonZeroUsize::new(LRU_WINDOW).expect("window > 0; qed")),
+		}
+	}
+}
+
+impl NaiveEquivocationDefender {
+	// return `true` if equivocation is beyond the limit.
+	fn insert_and_check(&mut self, slot: Slot) -> bool {
+		let val = self.cache.get_or_insert_mut(*slot, || 0);
+		if *val == EQUIVOCATION_LIMIT {
+			true
+		} else {
+			*val += 1;
+			false
+		}
+	}
+}
+
+struct Verifier<P, Client, Block, CIDP> {
+	client: Arc<Client>,
+	create_inherent_data_providers: CIDP,
+	slot_duration: SlotDuration,
+	defender: NaiveEquivocationDefender,
+	telemetry: Option<TelemetryHandle>,
+	_marker: std::marker::PhantomData<(Block, P)>,
+}
+
+#[async_trait::async_trait]
+impl<P, Client, Block, CIDP> VerifierT<Block> for Verifier<P, Client, Block, CIDP>
+where
+	P: Pair,
+	P::Signature: Encode + Decode,
+	P::Public: Encode + Decode + PartialEq + Clone + Debug,
+	Block: BlockT,
+	Client: ProvideRuntimeApi<Block> + Send + Sync,
+	<Client as ProvideRuntimeApi<Block>>::Api: BlockBuilderApi<Block> + AuraApi<Block, P::Public>,
+
+	CIDP: CreateInherentDataProviders<Block, ()>,
+{
+	async fn verify(
+		&mut self,
+		mut block_params: BlockImportParams<Block, ()>,
+	) -> Result<BlockImportParams<Block, ()>, String> {
+		// Skip checks that include execution, if being told so, or when importing only state.
+		//
+		// This is done for example when gap syncing and it is expected that the block after the gap
+		// was checked/chosen properly, e.g. by warp syncing to this block using a finality proof.
+		if block_params.state_action.skip_execution_checks() || block_params.with_state() {
+			return Ok(block_params)
+		}
+
+		let post_hash = block_params.header.hash();
+		let parent_hash = *block_params.header.parent_hash();
+
+		// check seal and update pre-hash/post-hash
+		{
+			let authorities = aura_internal::fetch_authorities(self.client.as_ref(), parent_hash)
+				.map_err(|e| {
+				format!("Could not fetch authorities at {:?}: {}", parent_hash, e)
+			})?;
+
+			let slot_now = slot_now(self.slot_duration);
+			let res = aura_internal::check_header_slot_and_seal::<Block, P>(
+				slot_now,
+				block_params.header,
+				&authorities,
+			);
+
+			match res {
+				Ok((pre_header, slot, seal_digest)) => {
+					telemetry!(
+						self.telemetry;
+						CONSENSUS_TRACE;
+						"aura.checked_and_importing";
+						"pre_header" => ?pre_header,
+					);
+
+					block_params.header = pre_header;
+					block_params.post_digests.push(seal_digest);
+					block_params.fork_choice = Some(ForkChoiceStrategy::LongestChain);
+					block_params.post_hash = Some(post_hash);
+
+					// Check for and reject egregious amounts of equivocations.
+					if self.defender.insert_and_check(slot) {
+						return Err(format!(
+							"Rejecting block {:?} due to excessive equivocations at slot",
+							post_hash,
+						))
+					}
+				},
+				Err(aura_internal::SealVerificationError::Deferred(hdr, slot)) => {
+					telemetry!(
+						self.telemetry;
+						CONSENSUS_DEBUG;
+						"aura.header_too_far_in_future";
+						"hash" => ?post_hash,
+						"a" => ?hdr,
+						"b" => ?slot,
+					);
+
+					return Err(format!(
+						"Rejecting block ({:?}) from future slot {:?}",
+						post_hash, slot
+					))
+				},
+				Err(e) =>
+					return Err(format!(
+						"Rejecting block ({:?}) with invalid seal ({:?})",
+						post_hash, e
+					)),
+			}
+		}
+
+		// check inherents.
+		if let Some(body) = block_params.body.clone() {
+			let block = Block::new(block_params.header.clone(), body);
+			let create_inherent_data_providers = self
+				.create_inherent_data_providers
+				.create_inherent_data_providers(parent_hash, ())
+				.await
+				.map_err(|e| format!("Could not create inherent data {:?}", e))?;
+
+			let inherent_data = create_inherent_data_providers
+				.create_inherent_data()
+				.await
+				.map_err(|e| format!("Could not create inherent data {:?}", e))?;
+
+			let inherent_res = self
+				.client
+				.runtime_api()
+				.check_inherents_with_context(
+					parent_hash,
+					block_params.origin.into(),
+					block,
+					inherent_data,
+				)
+				.map_err(|e| format!("Unable to check block inherents {:?}", e))?;
+
+			if !inherent_res.ok() {
+				for (i, e) in inherent_res.into_errors() {
+					match create_inherent_data_providers.try_handle_error(&i, &e).await {
+						Some(res) => res.map_err(|e| format!("Inherent Error {:?}", e))?,
+						None =>
+							return Err(format!(
+								"Unknown inherent error, source {:?}",
+								String::from_utf8_lossy(&i[..])
+							)),
+					}
+				}
+			}
+		}
+
+		Ok(block_params)
+	}
+}
+
+fn slot_now(slot_duration: SlotDuration) -> Slot {
+	let timestamp = sp_timestamp::InherentDataProvider::from_system_time().timestamp();
+	Slot::from_timestamp(timestamp, slot_duration)
+}
+
+/// Start an import queue for a Cumulus node which checks blocks' seals and inherent data.
+///
+/// Pass in only inherent data providers which don't include aura or parachain consensus inherents,
+/// e.g. things like timestamp and custom inherents for the runtime.
+///
+/// The others are generated explicitly internally.
+///
+/// This should only be used for runtimes where the runtime does not check all inherents and
+/// seals in `execute_block` (see <https://github.com/paritytech/cumulus/issues/2436>)
+pub fn fully_verifying_import_queue<P, Client, Block: BlockT, I, CIDP>(
+	client: Arc<Client>,
+	block_import: I,
+	create_inherent_data_providers: CIDP,
+	slot_duration: SlotDuration,
+	spawner: &impl sp_core::traits::SpawnEssentialNamed,
+	registry: Option<&substrate_prometheus_endpoint::Registry>,
+	telemetry: Option<TelemetryHandle>,
+) -> BasicQueue<Block, I::Transaction>
+where
+	P: Pair,
+	P::Signature: Encode + Decode,
+	P::Public: Encode + Decode + PartialEq + Clone + Debug,
+	I: BlockImport<Block, Error = ConsensusError>
+		+ ParachainBlockImportMarker
+		+ Send
+		+ Sync
+		+ 'static,
+	I::Transaction: Send,
+	Client: ProvideRuntimeApi<Block> + Send + Sync + 'static,
+	<Client as ProvideRuntimeApi<Block>>::Api: BlockBuilderApi<Block> + AuraApi<Block, P::Public>,
+	CIDP: CreateInherentDataProviders<Block, ()> + 'static,
+{
+	let verifier = Verifier::<P, _, _, _> {
+		client,
+		create_inherent_data_providers,
+		defender: NaiveEquivocationDefender::default(),
+		slot_duration,
+		telemetry,
+		_marker: std::marker::PhantomData,
+	};
+
+	BasicQueue::new(verifier, Box::new(block_import), None, spawner, registry)
+}