Changes tries build cache (#2933)

* changes tries build cache

added CT build cache test

* fix lines width

* fixed some grumbles

* clear cache when: digests disabled, top-level or skewed digest is built

* cached_changed_keys -> with_cached_changed_keys

substrate/core/client/db/src/lib.rs

@@ -36,7 +36,7 @@ mod utils;
 use std::sync::Arc;
 use std::path::PathBuf;
 use std::io;
-use std::collections::HashMap;
+use std::collections::{HashMap, HashSet};
 
 use client::backend::NewBlockState;
 use client::blockchain::HeaderBackend;
@@ -58,7 +58,7 @@ use sr_primitives::traits::{
 };
 use state_machine::backend::Backend as StateBackend;
 use executor::RuntimeInfo;
-use state_machine::{CodeExecutor, DBValue};
+use state_machine::{CodeExecutor, DBValue, ChangesTrieTransaction, ChangesTrieCacheAction, ChangesTrieBuildCache};
 use crate::utils::{Meta, db_err, meta_keys, read_db, block_id_to_lookup_key, read_meta};
 use client::leaves::{LeafSet, FinalizationDisplaced};
 use client::children;
@@ -405,6 +405,7 @@ pub struct BlockImportOperation<Block: BlockT, H: Hasher> {
 	storage_updates: StorageCollection,
 	child_storage_updates: ChildStorageCollection,
 	changes_trie_updates: MemoryDB<H>,
+	changes_trie_cache_update: Option<ChangesTrieCacheAction<H::Out, NumberFor<Block>>>,
 	pending_block: Option<PendingBlock<Block>>,
 	aux_ops: Vec<(Vec<u8>, Option<Vec<u8>>)>,
 	finalized_blocks: Vec<(BlockId<Block>, Option<Justification>)>,
@@ -487,8 +488,12 @@ where Block: BlockT<Hash=H256>,
 		Ok(root)
 	}
 
-	fn update_changes_trie(&mut self, update: MemoryDB<Blake2Hasher>) -> Result<(), client::error::Error> {
-		self.changes_trie_updates = update;
+	fn update_changes_trie(
+		&mut self,
+		update: ChangesTrieTransaction<Blake2Hasher, NumberFor<Block>>,
+	) -> Result<(), client::error::Error> {
+		self.changes_trie_updates = update.0;
+		self.changes_trie_cache_update = Some(update.1);
 		Ok(())
 	}
 
@@ -565,6 +570,7 @@ pub struct DbChangesTrieStorage<Block: BlockT> {
 	db: Arc<dyn KeyValueDB>,
 	meta: Arc<RwLock<Meta<NumberFor<Block>, Block::Hash>>>,
 	min_blocks_to_keep: Option<u32>,
+	cache: RwLock<ChangesTrieBuildCache<Block::Hash, NumberFor<Block>>>,
 	_phantom: ::std::marker::PhantomData<Block>,
 }
 
@@ -576,6 +582,11 @@ impl<Block: BlockT<Hash=H256>> DbChangesTrieStorage<Block> {
 		}
 	}
 
+	/// Commit changes into changes trie build cache.
+	pub fn commit_cache(&self, cache_update: ChangesTrieCacheAction<Block::Hash, NumberFor<Block>>) {
+		self.cache.write().perform(cache_update);
+	}
+
 	/// Prune obsolete changes tries.
 	pub fn prune(
 		&self,
@@ -699,6 +710,14 @@ where
 		self
 	}
 
+	fn with_cached_changed_keys(
+		&self,
+		root: &H256,
+		functor: &mut dyn FnMut(&HashMap<Option<Vec<u8>>, HashSet<Vec<u8>>>),
+	) -> bool {
+		self.cache.read().with_changed_keys(root, functor)
+	}
+
 	fn get(&self, key: &H256, _prefix: Prefix) -> Result<Option<DBValue>, String> {
 		self.db.get(columns::CHANGES_TRIE, &key[..])
 			.map_err(|err| format!("{}", err))
@@ -783,6 +802,7 @@ impl<Block: BlockT<Hash=H256>> Backend<Block> {
 			db,
 			meta,
 			min_blocks_to_keep: if is_archive_pruning { None } else { Some(MIN_BLOCKS_TO_KEEP_CHANGES_TRIES_FOR) },
+			cache: RwLock::new(ChangesTrieBuildCache::new()),
 			_phantom: Default::default(),
 		};
 
@@ -1098,7 +1118,6 @@ impl<Block: BlockT<Hash=H256>> Backend<Block> {
 			self.changes_tries_storage.commit(&mut transaction, changes_trie_updates);
 			let cache = operation.old_state.release(); // release state reference so that it can be finalized
 
-
 			if finalized {
 				// TODO: ensure best chain contains this block.
 				self.ensure_sequential_finalization(header, Some(last_finalized_hash))?;
@@ -1155,6 +1174,10 @@ impl<Block: BlockT<Hash=H256>> Backend<Block> {
 
 		let write_result = self.storage.db.write(transaction).map_err(db_err);
 
+		if let Some(changes_trie_cache_update) = operation.changes_trie_cache_update {
+			self.changes_tries_storage.commit_cache(changes_trie_cache_update);
+		}
+
 		if let Some((number, hash, enacted, retracted, displaced_leaf, is_best, mut cache)) = imported {
 			if let Err(e) = write_result {
 				let mut leaves = self.blockchain.leaves.write();
@@ -1288,6 +1311,7 @@ impl<Block> client::backend::Backend<Block, Blake2Hasher> for Backend<Block> whe
 			storage_updates: Default::default(),
 			child_storage_updates: Default::default(),
 			changes_trie_updates: MemoryDB::default(),
+			changes_trie_cache_update: None,
 			aux_ops: Vec::new(),
 			finalized_blocks: Vec::new(),
 			set_head: None,
@@ -1515,7 +1539,7 @@ mod tests {
 		let mut op = backend.begin_operation().unwrap();
 		backend.begin_state_operation(&mut op, block_id).unwrap();
 		op.set_block_data(header, None, None, NewBlockState::Best).unwrap();
-		op.update_changes_trie(changes_trie_update).unwrap();
+		op.update_changes_trie((changes_trie_update, ChangesTrieCacheAction::Clear)).unwrap();
 		backend.commit_operation(op).unwrap();
 
 		header_hash

substrate/core/client/src/backend.rs

@@ -24,10 +24,9 @@ use primitives::ChangesTrieConfiguration;
 use sr_primitives::{generic::BlockId, Justification, StorageOverlay, ChildrenStorageOverlay};
 use sr_primitives::traits::{Block as BlockT, NumberFor};
 use state_machine::backend::Backend as StateBackend;
-use state_machine::ChangesTrieStorage as StateChangesTrieStorage;
+use state_machine::{ChangesTrieStorage as StateChangesTrieStorage, ChangesTrieTransaction};
 use consensus::{well_known_cache_keys, BlockOrigin};
 use hash_db::Hasher;
-use trie::MemoryDB;
 use parking_lot::Mutex;
 
 /// In memory array of storage values.
@@ -116,7 +115,7 @@ pub trait BlockImportOperation<Block, H> where
 		child_update: ChildStorageCollection,
 	) -> error::Result<()>;
 	/// Inject changes trie data into the database.
-	fn update_changes_trie(&mut self, update: MemoryDB<H>) -> error::Result<()>;
+	fn update_changes_trie(&mut self, update: ChangesTrieTransaction<H, NumberFor<Block>>) -> error::Result<()>;
 	/// Insert auxiliary keys. Values are `None` if should be deleted.
 	fn insert_aux<I>(&mut self, ops: I) -> error::Result<()>
 		where I: IntoIterator<Item=(Vec<u8>, Option<Vec<u8>>)>;

substrate/core/client/src/call_executor.rs

@@ -17,15 +17,15 @@
 use std::{sync::Arc, cmp::Ord, panic::UnwindSafe, result, cell::RefCell, rc::Rc};
 use codec::{Encode, Decode};
 use sr_primitives::{
-	generic::BlockId, traits::Block as BlockT,
+	generic::BlockId, traits::Block as BlockT, traits::NumberFor,
 };
 use state_machine::{
 	self, OverlayedChanges, Ext, CodeExecutor, ExecutionManager,
 	ExecutionStrategy, NeverOffchainExt, backend::Backend as _,
+	ChangesTrieTransaction,
 };
 use executor::{RuntimeVersion, RuntimeInfo, NativeVersion};
 use hash_db::Hasher;
-use trie::MemoryDB;
 use primitives::{offchain, H256, Blake2Hasher, NativeOrEncoded, NeverNativeValue};
 
 use crate::runtime_api::{ProofRecorder, InitializeBlock};
@@ -111,7 +111,14 @@ where
 		manager: ExecutionManager<F>,
 		native_call: Option<NC>,
 		side_effects_handler: Option<&mut O>,
-	) -> Result<(NativeOrEncoded<R>, (S::Transaction, H::Out), Option<MemoryDB<H>>), error::Error>;
+	) -> Result<
+		(
+			NativeOrEncoded<R>,
+			(S::Transaction, H::Out),
+			Option<ChangesTrieTransaction<Blake2Hasher, NumberFor<B>>>
+		),
+		error::Error,
+	>;
 
 	/// Execute a call to a contract on top of given state, gathering execution proof.
 	///
@@ -345,7 +352,7 @@ where
 	) -> error::Result<(
 		NativeOrEncoded<R>,
 		(S::Transaction, <Blake2Hasher as Hasher>::Out),
-		Option<MemoryDB<Blake2Hasher>>,
+		Option<ChangesTrieTransaction<Blake2Hasher, NumberFor<Block>>>,
 	)> {
 		state_machine::new(
 			state,

substrate/core/client/src/client.rs

@@ -43,7 +43,8 @@ use sr_primitives::{
 use state_machine::{
 	DBValue, Backend as StateBackend, CodeExecutor, ChangesTrieAnchorBlockId,
 	ExecutionStrategy, ExecutionManager, prove_read, prove_child_read,
-	ChangesTrieRootsStorage, ChangesTrieStorage, ChangesTrieConfigurationRange,
+	ChangesTrieRootsStorage, ChangesTrieStorage,
+	ChangesTrieTransaction, ChangesTrieConfigurationRange,
 	key_changes, key_changes_proof, OverlayedChanges, NeverOffchainExt,
 };
 use executor::{RuntimeVersion, RuntimeInfo};
@@ -86,7 +87,7 @@ type StorageUpdate<B, Block> = <
 		<B as backend::Backend<Block, Blake2Hasher>>::BlockImportOperation
 			as BlockImportOperation<Block, Blake2Hasher>
 	>::State as state_machine::Backend<Blake2Hasher>>::Transaction;
-type ChangesUpdate = trie::MemoryDB<Blake2Hasher>;
+type ChangesUpdate<Block> = ChangesTrieTransaction<Blake2Hasher, NumberFor<Block>>;
 
 /// Execution strategies settings.
 #[derive(Debug, Clone)]
@@ -635,6 +636,14 @@ impl<B, E, Block, RA> Client<B, E, Block, RA> where
 				self
 			}
 
+			fn with_cached_changed_keys(
+				&self,
+				root: &H256,
+				functor: &mut dyn FnMut(&HashMap<Option<Vec<u8>>, HashSet<Vec<u8>>>),
+			) -> bool {
+				self.storage.with_cached_changed_keys(root, functor)
+			}
+
 			fn get(&self, key: &H256, prefix: Prefix) -> Result<Option<DBValue>, String> {
 				self.storage.get(key, prefix)
 			}
@@ -1001,7 +1010,7 @@ impl<B, E, Block, RA> Client<B, E, Block, RA> where
 		body: Option<Vec<Block::Extrinsic>>,
 	) -> error::Result<(
 		Option<StorageUpdate<B, Block>>,
-		Option<Option<ChangesUpdate>>,
+		Option<Option<ChangesUpdate<Block>>>,
 		Option<(
 			Vec<(Vec<u8>, Option<Vec<u8>>)>,
 			Vec<(Vec<u8>, Vec<(Vec<u8>, Option<Vec<u8>>)>)>

substrate/core/client/src/in_mem.rs

@@ -16,7 +16,7 @@
 
 //! In memory client backend
 
-use std::collections::HashMap;
+use std::collections::{HashMap, HashSet};
 use std::sync::Arc;
 use parking_lot::{RwLock, Mutex};
 use primitives::{ChangesTrieConfiguration, storage::well_known_keys};
@@ -24,7 +24,7 @@ use sr_primitives::generic::{BlockId, DigestItem};
 use sr_primitives::traits::{Block as BlockT, Header as HeaderT, Zero, NumberFor};
 use sr_primitives::{Justification, StorageOverlay, ChildrenStorageOverlay};
 use state_machine::backend::{Backend as StateBackend, InMemory};
-use state_machine::{self, InMemoryChangesTrieStorage, ChangesTrieAnchorBlockId};
+use state_machine::{self, InMemoryChangesTrieStorage, ChangesTrieAnchorBlockId, ChangesTrieTransaction};
 use hash_db::{Hasher, Prefix};
 use trie::MemoryDB;
 use consensus::well_known_cache_keys::Id as CacheKeyId;
@@ -484,8 +484,8 @@ where
 		Ok(())
 	}
 
-	fn update_changes_trie(&mut self, update: MemoryDB<H>) -> error::Result<()> {
-		self.changes_trie_update = Some(update);
+	fn update_changes_trie(&mut self, update: ChangesTrieTransaction<H, NumberFor<Block>>) -> error::Result<()> {
+		self.changes_trie_update = Some(update.0);
 		Ok(())
 	}
 
@@ -766,6 +766,14 @@ impl<Block, H> state_machine::ChangesTrieStorage<H, NumberFor<Block>> for Change
 		self
 	}
 
+	fn with_cached_changed_keys(
+		&self,
+		_root: &H::Out,
+		_functor: &mut dyn FnMut(&HashMap<Option<Vec<u8>>, HashSet<Vec<u8>>>),
+	) -> bool {
+		false
+	}
+
 	fn get(&self, key: &H::Out, prefix: Prefix) -> Result<Option<state_machine::DBValue>, String> {
 		self.0.get(key, prefix)
 	}

substrate/core/client/src/light/backend.rs

@@ -22,7 +22,7 @@ use std::sync::{Arc, Weak};
 use parking_lot::{RwLock, Mutex};
 
 use sr_primitives::{generic::BlockId, Justification, StorageOverlay, ChildrenStorageOverlay};
-use state_machine::{Backend as StateBackend, TrieBackend, backend::InMemory as InMemoryState};
+use state_machine::{Backend as StateBackend, TrieBackend, backend::InMemory as InMemoryState, ChangesTrieTransaction};
 use sr_primitives::traits::{Block as BlockT, NumberFor, Zero, Header};
 use crate::in_mem::{self, check_genesis_storage};
 use crate::backend::{
@@ -284,7 +284,7 @@ where
 		Ok(())
 	}
 
-	fn update_changes_trie(&mut self, _update: MemoryDB<H>) -> ClientResult<()> {
+	fn update_changes_trie(&mut self, _update: ChangesTrieTransaction<H, NumberFor<Block>>) -> ClientResult<()> {
 		// we're not storing anything locally => ignore changes
 		Ok(())
 	}

substrate/core/client/src/light/call_executor.rs

@@ -25,10 +25,10 @@ use std::{
 use codec::{Encode, Decode};
 use primitives::{offchain, H256, Blake2Hasher, convert_hash, NativeOrEncoded};
 use sr_primitives::generic::BlockId;
-use sr_primitives::traits::{One, Block as BlockT, Header as HeaderT};
+use sr_primitives::traits::{One, Block as BlockT, Header as HeaderT, NumberFor};
 use state_machine::{
 	self, Backend as StateBackend, CodeExecutor, OverlayedChanges,
-	ExecutionStrategy, create_proof_check_backend,
+	ExecutionStrategy, ChangesTrieTransaction, create_proof_check_backend,
 	execution_proof_check_on_trie_backend, ExecutionManager, NeverOffchainExt
 };
 use hash_db::Hasher;
@@ -40,7 +40,6 @@ use crate::call_executor::CallExecutor;
 use crate::error::{Error as ClientError, Result as ClientResult};
 use crate::light::fetcher::{Fetcher, RemoteCallRequest};
 use executor::{RuntimeVersion, NativeVersion};
-use trie::MemoryDB;
 
 /// Call executor that executes methods on remote node, querying execution proof
 /// and checking proof by re-executing locally.
@@ -185,7 +184,7 @@ where
 	) -> ClientResult<(
 		NativeOrEncoded<R>,
 		(S::Transaction, <Blake2Hasher as Hasher>::Out),
-		Option<MemoryDB<Blake2Hasher>>,
+		Option<ChangesTrieTransaction<Blake2Hasher, NumberFor<Block>>>,
 	)> {
 		Err(ClientError::NotAvailableOnLightClient.into())
 	}
@@ -365,7 +364,7 @@ impl<Block, B, Remote, Local> CallExecutor<Block, Blake2Hasher> for
 	) -> ClientResult<(
 		NativeOrEncoded<R>,
 		(S::Transaction, <Blake2Hasher as Hasher>::Out),
-		Option<MemoryDB<Blake2Hasher>>,
+		Option<ChangesTrieTransaction<Blake2Hasher, NumberFor<Block>>>,
 	)> {
 		// there's no actual way/need to specify native/wasm execution strategy on light node
 		// => we can safely ignore passed values

substrate/core/client/src/light/fetcher.rs

@@ -460,7 +460,7 @@ struct RootsStorage<'a, Number: SimpleArithmetic, Hash: 'a> {
 impl<'a, H, Number, Hash> ChangesTrieRootsStorage<H, Number> for RootsStorage<'a, Number, Hash>
 	where
 		H: Hasher,
-		Number: ::std::fmt::Display + Clone + SimpleArithmetic + Encode + Decode + Send + Sync + 'static,
+		Number: ::std::fmt::Display + ::std::hash::Hash + Clone + SimpleArithmetic + Encode + Decode + Send + Sync + 'static,
 		Hash: 'a + Send + Sync + Clone + AsRef<[u8]>,
 {
 	fn build_anchor(