Reduce ProgramCache write lock contention (#1037)

* Reduce ProgramCache write-lock contention * Also consider programs modified by tx * Memoize program cache write lock
anza-xyz · Apr 27, 2024 · 90bea33 · 90bea33
1 parent 3f7b352
commit 90bea33
Show file tree

Hide file tree

Showing 3 changed files with 36 additions and 11 deletions.
diff --git a/program-runtime/src/loaded_programs.rs b/program-runtime/src/loaded_programs.rs
@@ -666,6 +666,8 @@ pub struct ProgramCacheForTxBatch {
     /// The epoch of the last rerooting
     pub latest_root_epoch: Epoch,
     pub hit_max_limit: bool,
+    pub loaded_missing: bool,
+    pub merged_modified: bool,
 }
 
 impl ProgramCacheForTxBatch {
@@ -682,6 +684,8 @@ impl ProgramCacheForTxBatch {
             upcoming_environments,
             latest_root_epoch,
             hit_max_limit: false,
+            loaded_missing: false,
+            merged_modified: false,
         }
     }
 
@@ -697,6 +701,8 @@ impl ProgramCacheForTxBatch {
             upcoming_environments: cache.get_upcoming_environments_for_epoch(epoch),
             latest_root_epoch: cache.latest_root_epoch,
             hit_max_limit: false,
+            loaded_missing: false,
+            merged_modified: false,
         }
     }
 
@@ -750,9 +756,14 @@ impl ProgramCacheForTxBatch {
 
     pub fn merge(&mut self, other: &Self) {
         other.entries.iter().for_each(|(key, entry)| {
+            self.merged_modified = true;
             self.replenish(*key, entry.clone());
         })
     }
+
+    pub fn is_empty(&self) -> bool {
+        self.entries.is_empty()
+    }
 }
 
 pub enum ProgramCacheMatchCriteria {

diff --git a/runtime/src/bank.rs b/runtime/src/bank.rs
@@ -4156,18 +4156,23 @@ impl Bank {
 
         let mut store_executors_which_were_deployed_time =
             Measure::start("store_executors_which_were_deployed_time");
+        let mut cache = None;
         for execution_result in &execution_results {
             if let TransactionExecutionResult::Executed {
                 details,
                 programs_modified_by_tx,
             } = execution_result
             {
-                if details.status.is_ok() {
-                    let mut cache = self.transaction_processor.program_cache.write().unwrap();
-                    cache.merge(programs_modified_by_tx);
+                if details.status.is_ok() && !programs_modified_by_tx.is_empty() {
+                    cache
+                        .get_or_insert_with(|| {
+                            self.transaction_processor.program_cache.write().unwrap()
+                        })
+                        .merge(programs_modified_by_tx);
                 }
             }
         }
+        drop(cache);
         store_executors_which_were_deployed_time.stop();
         saturating_add_assign!(
             timings.execute_accessories.update_executors_us,

diff --git a/svm/src/transaction_processor.rs b/svm/src/transaction_processor.rs
@@ -292,14 +292,22 @@ impl<FG: ForkGraph> TransactionBatchProcessor<FG> {
 
         execution_time.stop();
 
-        const SHRINK_LOADED_PROGRAMS_TO_PERCENTAGE: u8 = 90;
-        self.program_cache
-            .write()
-            .unwrap()
-            .evict_using_2s_random_selection(
-                Percentage::from(SHRINK_LOADED_PROGRAMS_TO_PERCENTAGE),
-                self.slot,
-            );
+        // Skip eviction when there's no chance this particular tx batch has increased the size of
+        // ProgramCache entries. Note that loaded_missing is deliberately defined, so that there's
+        // still at least one other batch, which will evict the program cache, even after the
+        // occurrences of cooperative loading.
+        if programs_loaded_for_tx_batch.borrow().loaded_missing
+            || programs_loaded_for_tx_batch.borrow().merged_modified
+        {
+            const SHRINK_LOADED_PROGRAMS_TO_PERCENTAGE: u8 = 90;
+            self.program_cache
+                .write()
+                .unwrap()
+                .evict_using_2s_random_selection(
+                    Percentage::from(SHRINK_LOADED_PROGRAMS_TO_PERCENTAGE),
+                    self.slot,
+                );
+        }
 
         debug!(
             "load: {}us execute: {}us txs_len={}",
@@ -395,6 +403,7 @@ impl<FG: ForkGraph> TransactionBatchProcessor<FG> {
                 }
                 // Submit our last completed loading task.
                 if let Some((key, program)) = program_to_store.take() {
+                    loaded_programs_for_txs.as_mut().unwrap().loaded_missing = true;
                     if program_cache.finish_cooperative_loading_task(self.slot, key, program)
                         && limit_to_load_programs
                     {