feat(geyser): add completed data set shred range to deshred txn

Author: leafaar

core/src/completed_data_sets_service.rs

@@ -1,5 +1,8 @@
-//! [`CompletedDataSetsService`] is a hub, that runs different operations when a "completed data
-//! set", also known as a [`Vec<Entry>`], is received by the validator.
+//! [`CompletedDataSetsService`] is a hub that runs different operations when a completed data set
+//! is received by the validator.
+//!
+//! A completed data set is a contiguous range of data shreds whose combined payload deserializes
+//! to a single [`Vec<Entry>`].
 //!
 //! Currently, `WindowService` sends [`CompletedDataSetInfo`]s via a `completed_sets_receiver`
 //! provided to the [`CompletedDataSetsService`].
@@ -148,6 +151,8 @@ impl CompletedDataSetsService {
                                                total_data_sets: &mut u64,
                                                lut_transactions: &mut u64| {
             let CompletedDataSetInfo { slot, indices } = completed_data_set_info;
+            let completed_data_set_starting_shred_index = indices.start;
+            let completed_data_set_ending_shred_index_exclusive = indices.end;
             match blockstore.get_entries_in_data_block(slot, indices, /*slot_meta:*/ None) {
                 Ok(entries) => {
                     *total_data_sets += 1;
@@ -177,6 +182,8 @@ impl CompletedDataSetsService {
                                     let mut notify_measure = Measure::start("notify_deshred");
                                     notifier.notify_deshred_transaction(
                                         slot,
+                                        completed_data_set_starting_shred_index,
+                                        completed_data_set_ending_shred_index_exclusive,
                                         signature,
                                         is_vote,
                                         tx,

geyser-plugin-interface/src/geyser_plugin_interface.rs

@@ -213,10 +213,40 @@ pub struct ReplicaDeshredTransactionInfo<'a> {
     pub loaded_addresses: Option<&'a LoadedAddresses>,
 }
 
+/// Extends ReplicaDeshredTransactionInfo with metadata about the completed data set that
+/// produced the transaction.
+///
+/// A completed data set is a contiguous range of data shreds whose combined payload deserializes
+/// to a single `Vec<Entry>`. Multiple transactions can share the same completed-data-set range,
+/// and completed data sets for the same slot may be observed out of order. These fields describe
+/// the data-set container; they are not a block-wide transaction index.
+#[derive(Clone, Debug)]
+#[repr(C)]
+pub struct ReplicaDeshredTransactionInfoV2<'a> {
+    /// The transaction signature, used for identifying the transaction.
+    pub signature: &'a Signature,
+
+    /// Indicates if the transaction is a simple vote transaction.
+    pub is_vote: bool,
+
+    /// The versioned transaction.
+    pub transaction: &'a VersionedTransaction,
+
+    /// Addresses loaded from address lookup tables for V0 transactions.
+    pub loaded_addresses: Option<&'a LoadedAddresses>,
+
+    /// The inclusive starting shred index of the completed data set containing this transaction.
+    pub completed_data_set_starting_shred_index: u32,
+
+    /// The exclusive ending shred index of the completed data set containing this transaction.
+    pub completed_data_set_ending_shred_index_exclusive: u32,
+}
+
 /// A wrapper to future-proof ReplicaDeshredTransactionInfo handling.
 #[repr(u32)]
 pub enum ReplicaDeshredTransactionInfoVersions<'a> {
     V0_0_1(&'a ReplicaDeshredTransactionInfo<'a>),
+    V0_0_2(&'a ReplicaDeshredTransactionInfoV2<'a>),
 }
 
 #[derive(Clone, Debug)]

geyser-plugin-manager/src/deshred_transaction_notifier.rs

@@ -2,7 +2,7 @@
 use {
     crate::geyser_plugin_manager::GeyserPluginManager,
     agave_geyser_plugin_interface::geyser_plugin_interface::{
-        ReplicaDeshredTransactionInfo, ReplicaDeshredTransactionInfoVersions,
+        ReplicaDeshredTransactionInfoV2, ReplicaDeshredTransactionInfoVersions,
     },
     log::*,
     solana_clock::Slot,
@@ -28,32 +28,36 @@ impl DeshredTransactionNotifier for DeshredTransactionNotifierImpl {
     fn notify_deshred_transaction(
         &self,
         slot: Slot,
+        completed_data_set_starting_shred_index: u32,
+        completed_data_set_ending_shred_index_exclusive: u32,
         signature: &Signature,
         is_vote: bool,
         transaction: &VersionedTransaction,
         loaded_addresses: Option<&LoadedAddresses>,
     ) {
+        let plugin_manager = self.plugin_manager.read().unwrap();
+
+        if plugin_manager.plugins.is_empty() {
+            return;
+        }
+
         let mut measure =
             Measure::start("geyser-plugin-notify_plugins_of_deshred_transaction_info");
-        let transaction_info = Self::build_replica_deshred_transaction_info(
+        let transaction_info = ReplicaDeshredTransactionInfoV2 {
             signature,
             is_vote,
             transaction,
             loaded_addresses,
-        );
-
-        let plugin_manager = self.plugin_manager.read().unwrap();
-
-        if plugin_manager.plugins.is_empty() {
-            return;
-        }
+            completed_data_set_starting_shred_index,
+            completed_data_set_ending_shred_index_exclusive,
+        };
 
         for plugin in plugin_manager.plugins.iter() {
             if !plugin.deshred_transaction_notifications_enabled() {
                 continue;
             }
             match plugin.notify_deshred_transaction(
-                ReplicaDeshredTransactionInfoVersions::V0_0_1(&transaction_info),
+                ReplicaDeshredTransactionInfoVersions::V0_0_2(&transaction_info),
                 slot,
             ) {
                 Err(err) => {
@@ -85,18 +89,4 @@ impl DeshredTransactionNotifierImpl {
     pub fn new(plugin_manager: Arc<RwLock<GeyserPluginManager>>) -> Self {
         Self { plugin_manager }
     }
-
-    fn build_replica_deshred_transaction_info<'a>(
-        signature: &'a Signature,
-        is_vote: bool,
-        transaction: &'a VersionedTransaction,
-        loaded_addresses: Option<&'a LoadedAddresses>,
-    ) -> ReplicaDeshredTransactionInfo<'a> {
-        ReplicaDeshredTransactionInfo {
-            signature,
-            is_vote,
-            transaction,
-            loaded_addresses,
-        }
-    }
 }

ledger/src/deshred_transaction_notifier_interface.rs

@@ -8,10 +8,16 @@ use {
 
 /// Trait for notifying about transactions when they are deshredded.
 /// This is called when entries are formed from shreds, before any execution occurs.
+///
+/// The completed-data-set shred range identifies the contiguous range of data shreds whose
+/// combined payload deserializes to a single `Vec<Entry>`. All transactions reconstructed from
+/// that same completed data set share the same shred-range metadata.
 pub trait DeshredTransactionNotifier {
     fn notify_deshred_transaction(
         &self,
         slot: Slot,
+        completed_data_set_starting_shred_index: u32,
+        completed_data_set_ending_shred_index_exclusive: u32,
         signature: &Signature,
         is_vote: bool,
         transaction: &VersionedTransaction,