dongjoon-hyun commented on a change in pull request #26471: [SPARK-29850][SQL]
sort-merge-join an empty table should not memory leak
URL: https://github.com/apache/spark/pull/26471#discussion_r344909743
##########
File path:
sql/core/src/main/scala/org/apache/spark/sql/execution/aggregate/HashAggregateExec.scala
##########
@@ -695,34 +695,39 @@ case class HashAggregateExec(
ctx.addInnerClass(generatedMap)
// Inline mutable state since not many aggregation operations in a task
- fastHashMapTerm = ctx.addMutableState(fastHashMapClassName,
"vectorizedHastHashMap",
- v => s"$v = new $fastHashMapClassName();", forceInline = true)
- ctx.addMutableState(s"java.util.Iterator<InternalRow>",
"vectorizedFastHashMapIter",
+ fastHashMapTerm = ctx.addMutableState(
+ fastHashMapClassName, "vectorizedHastHashMap", forceInline = true)
+ val iter = ctx.addMutableState(
+ s"java.util.Iterator<InternalRow>",
+ "vectorizedFastHashMapIter",
forceInline = true)
+ val create = s"$fastHashMapTerm = new $fastHashMapClassName();"
+ (iter, create)
} else {
val generatedMap = new RowBasedHashMapGenerator(ctx,
aggregateExpressions,
fastHashMapClassName, groupingKeySchema, bufferSchema,
bitMaxCapacity).generate()
ctx.addInnerClass(generatedMap)
// Inline mutable state since not many aggregation operations in a task
- fastHashMapTerm = ctx.addMutableState(fastHashMapClassName,
"fastHashMap",
- v => s"$v = new $fastHashMapClassName(" +
- s"$thisPlan.getTaskMemoryManager(),
$thisPlan.getEmptyAggregationBuffer());",
- forceInline = true)
- ctx.addMutableState(
+ fastHashMapTerm = ctx.addMutableState(
+ fastHashMapClassName, "fastHashMap", forceInline = true)
+ val iter = ctx.addMutableState(
"org.apache.spark.unsafe.KVIterator<UnsafeRow, UnsafeRow>",
"fastHashMapIter", forceInline = true)
+ val create = s"$fastHashMapTerm = new $fastHashMapClassName(" +
+ s"$thisPlan.getTaskMemoryManager(),
$thisPlan.getEmptyAggregationBuffer());"
+ (iter, create)
}
- }
+ } else ("", "")
// Create a name for the iterator from the regular hash map.
// Inline mutable state since not many aggregation operations in a task
val iterTerm = ctx.addMutableState(classOf[KVIterator[UnsafeRow,
UnsafeRow]].getName,
"mapIter", forceInline = true)
// create hashMap
val hashMapClassName = classOf[UnsafeFixedWidthAggregationMap].getName
- hashMapTerm = ctx.addMutableState(hashMapClassName, "hashMap",
- v => s"$v = $thisPlan.createHashMap();", forceInline = true)
+ hashMapTerm = ctx.addMutableState(
+ hashMapClassName, "hashMap", forceInline = true)
Review comment:
nit. One line?
```scala
hashMapTerm = ctx.addMutableState(hashMapClassName, "hashMap", forceInline =
true)
```
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
