[GLUTEN-11088][VL] Fix NativeBenchmarkPlanGenerator (#11144)

Author: marin-ma

.github/workflows/velox_backend_x86.yml

@@ -1257,7 +1257,7 @@ jobs:
           cd ./cpp/build && ctest -V
       - name: Run CPP benchmark test
         run: |
-          $MVN_CMD test -Pspark-3.5 -Pbackends-velox -pl backends-velox -am \
+          $MVN_CMD clean test -Pspark-3.5 -Pbackends-velox -pl backends-velox -am \
           -DtagsToInclude="org.apache.gluten.tags.GenerateExample" -Dtest=none -DfailIfNoTests=false -Dexec.skip
           # This test depends on files generated by the above mvn test.
           ./cpp/build/velox/benchmarks/generic_benchmark --with-shuffle --partitioning hash --threads 1 --iterations 1 \
@@ -1266,14 +1266,27 @@ jobs:
           --data $(realpath backends-velox/generated-native-benchmark/data_12_0_*_0.parquet),$(realpath backends-velox/generated-native-benchmark/data_12_0_*_1.parquet)
       - name: Run UDF test
         run: |
-          # Depends on --build_example=ON.
           yum install -y java-17-openjdk-devel
           export JAVA_HOME=/usr/lib/jvm/java-17-openjdk
           export PATH=$JAVA_HOME/bin:$PATH
           java -version
+          # Depends on --build_example=ON.
           $MVN_CMD test -Pspark-3.5 -Pbackends-velox -Pjava-17 -Piceberg -Pdelta -Ppaimon -DtagsToExclude=org.apache.gluten.tags.EnhancedFeaturesTest \
           -DtagsToInclude=org.apache.gluten.tags.UDFTest \
           -DargLine="-Dspark.test.home=/opt/shims/spark35/spark_home/"
+      - name: Run CPP benchmark test for Spark 4.0
+        run: |
+          export SPARK_SCALA_VERSION=2.13
+          yum install -y java-17-openjdk-devel
+          export JAVA_HOME=/usr/lib/jvm/java-17-openjdk
+          export PATH=$JAVA_HOME/bin:$PATH
+          $MVN_CMD clean test -Pspark-4.0 -Pjava-17 -Pscala-2.13 -Pbackends-velox -pl backends-velox -am \
+          -DtagsToInclude="org.apache.gluten.tags.GenerateExample" -Dtest=none -DfailIfNoTests=false -Dexec.skip
+          # This test depends on files generated by the above mvn test.
+          ./cpp/build/velox/benchmarks/generic_benchmark --with-shuffle --partitioning hash --threads 1 --iterations 1 \
+          --conf $(realpath backends-velox/generated-native-benchmark/conf_12_0_*.ini) \
+          --plan $(realpath backends-velox/generated-native-benchmark/plan_12_0_*.json) \
+          --data $(realpath backends-velox/generated-native-benchmark/data_12_0_*_0.parquet),$(realpath backends-velox/generated-native-benchmark/data_12_0_*_1.parquet)
       - name: Upload test report
         uses: actions/upload-artifact@v4
         with:

backends-velox/src/test/scala/org/apache/gluten/benchmarks/NativeBenchmarkPlanGenerator.scala

@@ -71,8 +71,7 @@ class NativeBenchmarkPlanGenerator extends VeloxWholeStageTransformerSuite {
     }
   }
 
-  // TODO: fix on spark-4.0
-  testWithMaxSparkVersion("Test plan json non-empty - AQE on", "3.5") {
+  test("Test plan json non-empty - AQE on") {
     withSQLConf(
       SQLConf.ADAPTIVE_EXECUTION_ENABLED.key -> "true",
       GlutenConfig.CACHE_WHOLE_STAGE_TRANSFORMER_CONTEXT.key -> "true") {
@@ -85,7 +84,7 @@ class NativeBenchmarkPlanGenerator extends VeloxWholeStageTransformerSuite {
       executedPlan.execute()
 
       val finalPlan = executedPlan.asInstanceOf[AdaptiveSparkPlanExec].executedPlan
-      val lastStageTransformer = finalPlan.find(_.isInstanceOf[WholeStageTransformer])
+      val lastStageTransformer = stripAQEPlan(finalPlan).find(_.isInstanceOf[WholeStageTransformer])
       assert(lastStageTransformer.nonEmpty)
       val planJson = lastStageTransformer.get.asInstanceOf[WholeStageTransformer].substraitPlanJson
       assert(planJson.nonEmpty)