From bcb03cdb0ebedff8ea18c4428a659850df5ea037 Mon Sep 17 00:00:00 2001
From: Orri Erling <oerling@fb.com>
Date: Tue, 12 Dec 2023 07:51:59 -0800
Subject: [PATCH] Faster hash modes test (#7973)

Summary:
AggregationTest.hashmodes covers the required cases with less data in about half the time. Less timeouts. Checks that the end state is kHash.

Pull Request resolved: https://github.com/facebookincubator/velox/pull/7973

Reviewed By: kevinwilfong

Differential Revision: D52055200

Pulled By: oerling

fbshipit-source-id: b1af9ff0e21ceb76d09f67560034bdc382420aff
---
 velox/exec/HashTable.cpp             |  1 +
 velox/exec/tests/AggregationTest.cpp | 11 +++++++++--
 2 files changed, 10 insertions(+), 2 deletions(-)

diff --git a/velox/exec/HashTable.cpp b/velox/exec/HashTable.cpp
index d58be2bb30f7..7bfcfecc6dfc 100644
--- a/velox/exec/HashTable.cpp
+++ b/velox/exec/HashTable.cpp
@@ -1231,6 +1231,7 @@ void HashTable<ignoreNullKeys>::rehash(bool initNormalizedKeys) {
 template <bool ignoreNullKeys>
 void HashTable<ignoreNullKeys>::setHashMode(HashMode mode, int32_t numNew) {
   VELOX_CHECK_NE(hashMode_, HashMode::kHash);
+  TestValue::adjust("facebook::velox::exec::HashTable::setHashMode", &mode);
   if (mode == HashMode::kArray) {
     const auto bytes = capacity_ * tableSlotSize();
     const auto numPages = memory::AllocationTraits::numPages(bytes);
diff --git a/velox/exec/tests/AggregationTest.cpp b/velox/exec/tests/AggregationTest.cpp
index 01d2c3b65eba..342f1967e133 100644
--- a/velox/exec/tests/AggregationTest.cpp
+++ b/velox/exec/tests/AggregationTest.cpp
@@ -697,10 +697,10 @@ TEST_F(AggregationTest, hashmodes) {
   makeModeTestKeys(rowType, 20000, 2, 2, 2, 4, 4, 4, batches);
   // 20K rows with all at slightly higher cardinality, still in array range.
   makeModeTestKeys(rowType, 20000, 2, 2, 2, 4, 16, 4, batches);
-  // 100K rows with cardinality outside of array range. We transit to
+  // 25K rows with cardinality outside of array range. We transit to
   // generic hash table from normalized keys when running out of quota
   // for distinct string storage for the sixth key.
-  makeModeTestKeys(rowType, 100000, 1000000, 2, 2, 4, 4, 1000000, batches);
+  makeModeTestKeys(rowType, 25000, 1000000, 2, 2, 4, 4, 1000000, batches);
   createDuckDbTable(batches);
   auto op =
       PlanBuilder()
@@ -708,10 +708,17 @@ TEST_F(AggregationTest, hashmodes) {
           .singleAggregation({"c0", "c1", "c2", "c3", "c4", "c5"}, {"sum(1)"})
           .planNode();
 
+  std::atomic<BaseHashTable::HashMode> mode{BaseHashTable::HashMode::kArray};
+  SCOPED_TESTVALUE_SET(
+      "facebook::velox::exec::HashTable::setHashMode",
+      std::function<void(void*)>([&](void* newMode) {
+        mode = *reinterpret_cast<BaseHashTable::HashMode*>(newMode);
+      }));
   assertQuery(
       op,
       "SELECT c0, c1, C2, C3, C4, C5, sum(1) FROM tmp "
       " GROUP BY c0, c1, c2, c3, c4, c5");
+  EXPECT_EQ(mode, BaseHashTable::HashMode::kHash);
 }
 
 TEST_F(AggregationTest, rangeToDistinct) {