fix(trie): Fixes to pruning of MerkleChangeSets

crates/config/src/config.rs

@@ -440,7 +440,7 @@ pub struct PruneConfig {
 
 impl Default for PruneConfig {
     fn default() -> Self {
-        Self { block_interval: DEFAULT_BLOCK_INTERVAL, segments: PruneModes::none() }
+        Self { block_interval: DEFAULT_BLOCK_INTERVAL, segments: PruneModes::default() }
     }
 }
 
@@ -464,6 +464,7 @@ impl PruneConfig {
                     account_history,
                     storage_history,
                     bodies_history,
+                    merkle_changesets,
                     receipts_log_filter,
                 },
         } = other;
@@ -480,6 +481,8 @@ impl PruneConfig {
         self.segments.account_history = self.segments.account_history.or(account_history);
         self.segments.storage_history = self.segments.storage_history.or(storage_history);
         self.segments.bodies_history = self.segments.bodies_history.or(bodies_history);
+        // Merkle changesets is not optional, so we just replace it if provided
+        self.segments.merkle_changesets = merkle_changesets;
 
         if self.segments.receipts_log_filter.0.is_empty() && !receipts_log_filter.0.is_empty() {
             self.segments.receipts_log_filter = receipts_log_filter;
@@ -1001,6 +1004,7 @@ receipts = 'full'
                 account_history: None,
                 storage_history: Some(PruneMode::Before(5000)),
                 bodies_history: None,
+                merkle_changesets: PruneMode::Before(0),
                 receipts_log_filter: ReceiptsLogPruneConfig(BTreeMap::from([(
                     Address::random(),
                     PruneMode::Full,
@@ -1017,6 +1021,7 @@ receipts = 'full'
                 account_history: Some(PruneMode::Distance(2000)),
                 storage_history: Some(PruneMode::Distance(3000)),
                 bodies_history: None,
+                merkle_changesets: PruneMode::Distance(10000),
                 receipts_log_filter: ReceiptsLogPruneConfig(BTreeMap::from([
                     (Address::random(), PruneMode::Distance(1000)),
                     (Address::random(), PruneMode::Before(2000)),
@@ -1035,6 +1040,7 @@ receipts = 'full'
         assert_eq!(config1.segments.receipts, Some(PruneMode::Distance(1000)));
         assert_eq!(config1.segments.account_history, Some(PruneMode::Distance(2000)));
         assert_eq!(config1.segments.storage_history, Some(PruneMode::Before(5000)));
+        assert_eq!(config1.segments.merkle_changesets, PruneMode::Distance(10000));
         assert_eq!(config1.segments.receipts_log_filter, original_filter);
     }
 

crates/exex/exex/src/backfill/factory.rs

@@ -24,7 +24,7 @@ impl<E, P> BackfillJobFactory<E, P> {
         Self {
             evm_config,
             provider,
-            prune_modes: PruneModes::none(),
+            prune_modes: PruneModes::default(),
             thresholds: ExecutionStageThresholds {
                 // Default duration for a database transaction to be considered long-lived is
                 // 60 seconds, so we limit the backfill job to the half of it to be sure we finish

crates/node/core/src/args/pruning.rs

@@ -126,6 +126,7 @@ impl PruningArgs {
                     storage_history: Some(PruneMode::Distance(MINIMUM_PRUNING_DISTANCE)),
                     // TODO: set default to pre-merge block if available
                     bodies_history: None,
+                    merkle_changesets: PruneMode::Distance(MINIMUM_PRUNING_DISTANCE),
                     receipts_log_filter: Default::default(),
                 },
             }

crates/prune/prune/src/builder.rs

@@ -6,8 +6,8 @@ use reth_db_api::{table::Value, transaction::DbTxMut};
 use reth_exex_types::FinishedExExHeight;
 use reth_primitives_traits::NodePrimitives;
 use reth_provider::{
-    providers::StaticFileProvider, BlockReader, DBProvider, DatabaseProviderFactory,
-    NodePrimitivesProvider, PruneCheckpointReader, PruneCheckpointWriter,
+    providers::StaticFileProvider, BlockReader, ChainStateBlockReader, DBProvider,
+    DatabaseProviderFactory, NodePrimitivesProvider, PruneCheckpointReader, PruneCheckpointWriter,
     StaticFileProviderFactory,
 };
 use reth_prune_types::PruneModes;
@@ -83,6 +83,7 @@ impl PrunerBuilder {
                 ProviderRW: PruneCheckpointWriter
                                 + PruneCheckpointReader
                                 + BlockReader<Transaction: Encodable2718>
+                                + ChainStateBlockReader
                                 + StaticFileProviderFactory<
                     Primitives: NodePrimitives<SignedTx: Value, Receipt: Value, BlockHeader: Value>,
                 >,
@@ -113,6 +114,7 @@ impl PrunerBuilder {
                 Primitives: NodePrimitives<SignedTx: Value, Receipt: Value, BlockHeader: Value>,
             > + DBProvider<Tx: DbTxMut>
             + BlockReader<Transaction: Encodable2718>
+            + ChainStateBlockReader
             + PruneCheckpointWriter
             + PruneCheckpointReader,
     {
@@ -132,7 +134,7 @@ impl Default for PrunerBuilder {
     fn default() -> Self {
         Self {
             block_interval: 5,
-            segments: PruneModes::none(),
+            segments: PruneModes::default(),
             delete_limit: MAINNET_PRUNE_DELETE_LIMIT,
             timeout: None,
             finished_exex_height: watch::channel(FinishedExExHeight::NoExExs).1,

crates/prune/prune/src/segments/set.rs

@@ -1,13 +1,13 @@
 use crate::segments::{
-    AccountHistory, ReceiptsByLogs, Segment, SenderRecovery, StorageHistory, TransactionLookup,
-    UserReceipts,
+    AccountHistory, MerkleChangeSets, ReceiptsByLogs, Segment, SenderRecovery, StorageHistory,
+    TransactionLookup, UserReceipts,
 };
 use alloy_eips::eip2718::Encodable2718;
 use reth_db_api::{table::Value, transaction::DbTxMut};
 use reth_primitives_traits::NodePrimitives;
 use reth_provider::{
-    providers::StaticFileProvider, BlockReader, DBProvider, PruneCheckpointReader,
-    PruneCheckpointWriter, StaticFileProviderFactory,
+    providers::StaticFileProvider, BlockReader, ChainStateBlockReader, DBProvider,
+    PruneCheckpointReader, PruneCheckpointWriter, StaticFileProviderFactory,
 };
 use reth_prune_types::PruneModes;
 
@@ -52,7 +52,8 @@ where
         > + DBProvider<Tx: DbTxMut>
         + PruneCheckpointWriter
         + PruneCheckpointReader
-        + BlockReader<Transaction: Encodable2718>,
+        + BlockReader<Transaction: Encodable2718>
+        + ChainStateBlockReader,
 {
     /// Creates a [`SegmentSet`] from an existing components, such as [`StaticFileProvider`] and
     /// [`PruneModes`].
@@ -67,6 +68,7 @@ where
             account_history,
             storage_history,
             bodies_history: _,
+            merkle_changesets,
             receipts_log_filter,
         } = prune_modes;
 
@@ -77,6 +79,8 @@ where
             .segment(StaticFileTransactions::new(static_file_provider.clone()))
             // Static file receipts
             .segment(StaticFileReceipts::new(static_file_provider))
+            // Merkle changesets
+            .segment(MerkleChangeSets::new(merkle_changesets))
             // Account history
             .segment_opt(account_history.map(AccountHistory::new))
             // Storage history

crates/prune/prune/src/segments/user/merkle_change_sets.rs

@@ -3,11 +3,12 @@ use crate::{
     segments::{PruneInput, Segment},
     PrunerError,
 };
-use reth_db_api::{models::BlockNumberAddress, table::Value, tables, transaction::DbTxMut};
+use alloy_primitives::B256;
+use reth_db_api::{models::BlockNumberHashedAddress, table::Value, tables, transaction::DbTxMut};
 use reth_primitives_traits::NodePrimitives;
 use reth_provider::{
-    errors::provider::ProviderResult, BlockReader, DBProvider, NodePrimitivesProvider,
-    PruneCheckpointWriter, TransactionsProvider,
+    errors::provider::ProviderResult, BlockReader, ChainStateBlockReader, DBProvider,
+    NodePrimitivesProvider, PruneCheckpointWriter, TransactionsProvider,
 };
 use reth_prune_types::{
     PruneCheckpoint, PruneMode, PrunePurpose, PruneSegment, SegmentOutput, SegmentOutputCheckpoint,
@@ -31,6 +32,7 @@ where
         + PruneCheckpointWriter
         + TransactionsProvider
         + BlockReader
+        + ChainStateBlockReader
         + NodePrimitivesProvider<Primitives: NodePrimitives<Receipt: Value>>,
 {
     fn segment(&self) -> PruneSegment {
@@ -55,13 +57,20 @@ where
         let block_range_end = *block_range.end();
         let mut limiter = input.limiter;
 
+        // Create range for StoragesTrieChangeSets which uses BlockNumberHashedAddress as key
+        let storage_range_start: BlockNumberHashedAddress =
+            (*block_range.start(), B256::ZERO).into();
+        let storage_range_end: BlockNumberHashedAddress =
+            (*block_range.end() + 1, B256::ZERO).into();
+        let storage_range = storage_range_start..storage_range_end;
+
         let mut last_storages_pruned_block = None;
         let (storages_pruned, done) =
-            provider.tx_ref().prune_table_with_range::<tables::StorageChangeSets>(
-                BlockNumberAddress::range(block_range.clone()),
+            provider.tx_ref().prune_table_with_range::<tables::StoragesTrieChangeSets>(
+                storage_range,
                 &mut limiter,
                 |_| false,
-                |(BlockNumberAddress((block_number, ..)), ..)| {
+                |(BlockNumberHashedAddress((block_number, _)), _)| {
                     last_storages_pruned_block = Some(block_number);
                 },
             )?;

crates/prune/types/src/target.rs

@@ -36,8 +36,13 @@ pub enum HistoryType {
     StorageHistory,
 }
 
+/// Default pruning mode for merkle changesets - aggressively prune to finalized block
+const fn default_merkle_changesets_mode() -> PruneMode {
+    PruneMode::Distance(MINIMUM_PRUNING_DISTANCE)
+}
+
 /// Pruning configuration for every segment of the data that can be pruned.
-#[derive(Debug, Clone, Default, Eq, PartialEq)]
+#[derive(Debug, Clone, Eq, PartialEq)]
 #[cfg_attr(any(test, feature = "serde"), derive(serde::Serialize, serde::Deserialize))]
 #[cfg_attr(any(test, feature = "serde"), serde(default))]
 pub struct PruneModes {
@@ -84,6 +89,16 @@ pub struct PruneModes {
         )
     )]
     pub bodies_history: Option<PruneMode>,
+    /// Merkle Changesets pruning configuration for `AccountsTrieChangeSets` and
+    /// `StoragesTrieChangeSets`.
+    #[cfg_attr(
+        any(test, feature = "serde"),
+        serde(
+            default = "default_merkle_changesets_mode",
+            deserialize_with = "deserialize_prune_mode_with_min_blocks::<MINIMUM_PRUNING_DISTANCE, _>"
+        )
+    )]
+    pub merkle_changesets: PruneMode,
     /// Receipts pruning configuration by retaining only those receipts that contain logs emitted
     /// by the specified addresses, discarding others. This setting is overridden by `receipts`.
     ///
@@ -92,12 +107,22 @@ pub struct PruneModes {
     pub receipts_log_filter: ReceiptsLogPruneConfig,
 }
 
-impl PruneModes {
-    /// Sets pruning to no target.
-    pub fn none() -> Self {
-        Self::default()
+impl Default for PruneModes {
+    fn default() -> Self {
+        Self {
+            sender_recovery: None,
+            transaction_lookup: None,
+            receipts: None,
+            account_history: None,
+            storage_history: None,
+            bodies_history: None,
+            merkle_changesets: default_merkle_changesets_mode(),
+            receipts_log_filter: ReceiptsLogPruneConfig::default(),
+        }
     }
+}
 
+impl PruneModes {
     /// Sets pruning to all targets.
     pub fn all() -> Self {
         Self {
@@ -107,6 +132,7 @@ impl PruneModes {
             account_history: Some(PruneMode::Full),
             storage_history: Some(PruneMode::Full),
             bodies_history: Some(PruneMode::Full),
+            merkle_changesets: PruneMode::Full,
             receipts_log_filter: Default::default(),
         }
     }
@@ -116,11 +142,6 @@ impl PruneModes {
         self.receipts.is_some() || !self.receipts_log_filter.is_empty()
     }
 
-    /// Returns true if all prune modes are set to [`None`].
-    pub fn is_empty(&self) -> bool {
-        self == &Self::none()
-    }
-
     /// Returns an error if we can't unwind to the targeted block because the target block is
     /// outside the range.
     ///
@@ -170,6 +191,28 @@ impl PruneModes {
     }
 }
 
+/// Deserializes [`PruneMode`] and validates that the value is not less than the const
+/// generic parameter `MIN_BLOCKS`. This parameter represents the number of blocks that needs to be
+/// left in database after the pruning.
+///
+/// 1. For [`PruneMode::Full`], it fails if `MIN_BLOCKS > 0`.
+/// 2. For [`PruneMode::Distance`], it fails if `distance < MIN_BLOCKS + 1`. `+ 1` is needed because
+///    `PruneMode::Distance(0)` means that we leave zero blocks from the latest, meaning we have one
+///    block in the database.
+#[cfg(any(test, feature = "serde"))]
+fn deserialize_prune_mode_with_min_blocks<
+    'de,
+    const MIN_BLOCKS: u64,
+    D: serde::Deserializer<'de>,
+>(
+    deserializer: D,
+) -> Result<PruneMode, D::Error> {
+    use serde::Deserialize;
+    let prune_mode = PruneMode::deserialize(deserializer)?;
+    serde_deserialize_validate::<MIN_BLOCKS, D>(&prune_mode)?;
+    Ok(prune_mode)
+}
+
 /// Deserializes [`Option<PruneMode>`] and validates that the value is not less than the const
 /// generic parameter `MIN_BLOCKS`. This parameter represents the number of blocks that needs to be
 /// left in database after the pruning.
@@ -186,28 +229,37 @@ fn deserialize_opt_prune_mode_with_min_blocks<
 >(
     deserializer: D,
 ) -> Result<Option<PruneMode>, D::Error> {
-    use alloc::format;
     use serde::Deserialize;
     let prune_mode = Option::<PruneMode>::deserialize(deserializer)?;
+    if let Some(prune_mode) = prune_mode.as_ref() {
+        serde_deserialize_validate::<MIN_BLOCKS, D>(prune_mode)?;
+    }
+    Ok(prune_mode)
+}
 
+#[cfg(any(test, feature = "serde"))]
+fn serde_deserialize_validate<'a, 'de, const MIN_BLOCKS: u64, D: serde::Deserializer<'de>>(
+    prune_mode: &'a PruneMode,
+) -> Result<(), D::Error> {
+    use alloc::format;
     match prune_mode {
-        Some(PruneMode::Full) if MIN_BLOCKS > 0 => {
+        PruneMode::Full if MIN_BLOCKS > 0 => {
             Err(serde::de::Error::invalid_value(
                 serde::de::Unexpected::Str("full"),
                 // This message should have "expected" wording
                 &format!("prune mode that leaves at least {MIN_BLOCKS} blocks in the database")
                     .as_str(),
             ))
         }
-        Some(PruneMode::Distance(distance)) if distance < MIN_BLOCKS => {
+        PruneMode::Distance(distance) if *distance < MIN_BLOCKS => {
             Err(serde::de::Error::invalid_value(
-                serde::de::Unexpected::Unsigned(distance),
+                serde::de::Unexpected::Unsigned(*distance),
                 // This message should have "expected" wording
                 &format!("prune mode that leaves at least {MIN_BLOCKS} blocks in the database")
                     .as_str(),
             ))
         }
-        _ => Ok(prune_mode),
+        _ => Ok(()),
     }
 }
 
@@ -240,7 +292,7 @@ mod tests {
     #[test]
     fn test_unwind_target_unpruned() {
         // Test case 1: No pruning configured - should always succeed
-        let prune_modes = PruneModes::none();
+        let prune_modes = PruneModes::default();
         assert!(prune_modes.ensure_unwind_target_unpruned(1000, 500, &[]).is_ok());
         assert!(prune_modes.ensure_unwind_target_unpruned(1000, 0, &[]).is_ok());
 

crates/stages/stages/src/sets.rs

@@ -54,7 +54,7 @@ use reth_primitives_traits::{Block, NodePrimitives};
 use reth_provider::HeaderSyncGapProvider;
 use reth_prune_types::PruneModes;
 use reth_stages_api::Stage;
-use std::{ops::Not, sync::Arc};
+use std::sync::Arc;
 use tokio::sync::watch;
 
 /// A set containing all stages to run a fully syncing instance of reth.
@@ -337,12 +337,12 @@ where
                 stages_config: self.stages_config.clone(),
                 prune_modes: self.prune_modes.clone(),
             })
-            // If any prune modes are set, add the prune stage.
-            .add_stage_opt(self.prune_modes.is_empty().not().then(|| {
-                // Prune stage should be added after all hashing stages, because otherwise it will
-                // delete
-                PruneStage::new(self.prune_modes.clone(), self.stages_config.prune.commit_threshold)
-            }))
+            // Prune stage should be added after all hashing stages, because otherwise it will
+            // delete
+            .add_stage(PruneStage::new(
+                self.prune_modes.clone(),
+                self.stages_config.prune.commit_threshold,
+            ))
     }
 }