From 196b6f0c05d6fb837273ac83621cdcbdbbd066b0 Mon Sep 17 00:00:00 2001
From: Xiaoxuan Meng <xiaoxmeng@meta.com>
Date: Wed, 18 Dec 2024 13:49:38 -0800
Subject: [PATCH] fix: Fix flaky tsan failure in async data cache

Summary:
Mark refresh stats method to avoid tsan check to see if it can prevent tsan failure. The refresh stats will access the
entries which are under initialization and it is read only has not pointer chasing so it is safe to do so.

Differential Revision: D67378594
---
 velox/common/caching/AsyncDataCache.cpp | 6 +++++-
 velox/common/caching/AsyncDataCache.h   | 8 +++++++-
 2 files changed, 12 insertions(+), 2 deletions(-)

diff --git a/velox/common/caching/AsyncDataCache.cpp b/velox/common/caching/AsyncDataCache.cpp
index 17a626bf5474..fabd275d9b02 100644
--- a/velox/common/caching/AsyncDataCache.cpp
+++ b/velox/common/caching/AsyncDataCache.cpp
@@ -518,7 +518,9 @@ void CacheShard::updateStats(CacheStats& stats) {
     if (!entry || !entry->key_.fileNum.hasValue()) {
       ++stats.numEmptyEntries;
       continue;
-    } else if (entry->isExclusive()) {
+    }
+
+    if (entry->isExclusive()) {
       stats.exclusivePinnedBytes +=
           entry->data().byteSize() + entry->tinyData_.capacity();
       ++stats.numExclusive;
@@ -527,10 +529,12 @@ void CacheShard::updateStats(CacheStats& stats) {
           entry->data().byteSize() + entry->tinyData_.capacity();
       ++stats.numShared;
     }
+
     if (entry->isPrefetch_) {
       ++stats.numPrefetch;
       stats.prefetchBytes += entry->size();
     }
+
     ++stats.numEntries;
     stats.tinySize += entry->tinyData_.size();
     stats.tinyPadding += entry->tinyData_.capacity() - entry->tinyData_.size();
diff --git a/velox/common/caching/AsyncDataCache.h b/velox/common/caching/AsyncDataCache.h
index 173bb2a4e481..5f0bf7bf6088 100644
--- a/velox/common/caching/AsyncDataCache.h
+++ b/velox/common/caching/AsyncDataCache.h
@@ -787,9 +787,15 @@ class AsyncDataCache : public memory::Cache {
   /// Returns true if there is an entry for 'key'. Updates access time.
   bool exists(RawFileCacheKey key) const;
 
+#if defined(__has_feature)
+#if __has_feature(thread_sanitizer)
+  __attribute__((__no_sanitize__("thread")))
+#endif
+#endif
   /// Returns snapshot of the aggregated stats from all shards and the stats of
   /// SSD cache if used.
-  virtual CacheStats refreshStats() const;
+  virtual CacheStats
+  refreshStats() const;
 
   /// If 'details' is true, returns the stats of the backing memory allocator
   /// and ssd cache. Otherwise, only returns the cache stats.