fix

velox/common/memory/tests/SharedArbitratorTest.cpp

@@ -338,7 +338,8 @@ class SharedArbitrationTest : public exec::test::HiveConnectorTestBase {
   }
 
   RowVectorPtr newVector() {
-    VectorFuzzer fuzzer(fuzzerOpts_, pool());
+    auto seed = folly::Random::rand32();
+    VectorFuzzer fuzzer(fuzzerOpts_, pool(), seed);
     return fuzzer.fuzzRow(rowType_);
   }
 
@@ -1291,7 +1292,7 @@ DEBUG_ONLY_TEST_F(SharedArbitrationTest, reclaimFromJoinBuilder) {
     std::atomic_bool taskPauseDone{false};
     folly::EventCount taskPauseWait;
 
-    const auto joinMemoryUsage = 32L << 20;
+    const auto joinMemoryUsage = 2L << 20;
     const auto fakeAllocationSize = kMemoryCapacity - joinMemoryUsage / 2;
 
     std::atomic<bool> injectAllocationOnce{true};

velox/exec/HashBuild.cpp

@@ -330,7 +330,7 @@ void HashBuild::addInput(RowVectorPtr input) {
   TestValue::adjust("facebook::velox::exec::HashBuild::addInput", this);
 
   // Prevents the memory arbitrator to reclaim memory from this operator during
-  // the execution belowg.
+  // the execution below.
   NonReclaimableSection guard(this);
 
   activeRows_.resize(input->size());
@@ -848,16 +848,10 @@ bool HashBuild::finishHashBuild() {
   // https://github.com/facebookincubator/velox/issues/3567 is fixed.
   const bool allowParallelJoinBuild =
       !otherTables.empty() && spillPartitions.empty();
-  // For left semi and anti join, in case when HashBuild operator runs
-  // single-threaded there is no need to build hash table again after
-  // processing all input.
-  const bool needPrepareJoinTable = !(dropDuplicates_ && otherTables.empty());
-  if (needPrepareJoinTable) {
-    table_->prepareJoinTable(
-        std::move(otherTables),
-        allowParallelJoinBuild ? operatorCtx_->task()->queryCtx()->executor()
-                               : nullptr);
-  }
+  table_->prepareJoinTable(
+      std::move(otherTables),
+      allowParallelJoinBuild ? operatorCtx_->task()->queryCtx()->executor()
+                             : nullptr);
   addRuntimeStats();
   if (joinBridge_->setHashTable(
           std::move(table_), std::move(spillPartitions), joinHasNullKeys_)) {
@@ -1168,7 +1162,9 @@ void HashBuild::reclaim(
     // TODO: support fine-grain disk spilling based on 'targetBytes' after
     // having row container memory compaction support later.
     buildOp->spiller_->spill();
-    VELOX_CHECK_EQ(buildOp->table_->numDistinct(), 0);
+    if (!dropDuplicates_) {
+      VELOX_CHECK_EQ(buildOp->table_->numDistinct(), 0);
+    }
     buildOp->table_->clear();
     // Release the minimum reserved memory.
     op->pool()->release();