Vinod KC created SPARK-58205:
--------------------------------
Summary: Mark JSON/CSV/XML expressions stateful to prevent
shared-evaluator data races
Key: SPARK-58205
URL: https://issues.apache.org/jira/browse/SPARK-58205
Project: Spark
Issue Type: Sub-task
Components: SQL
Affects Versions: 4.2.0, 4.0.0, 4.1.0, 3.5.0, 4.3.0
Reporter: Vinod KC
Seven JSON/CSV/XML expressions (JsonToStructs, GetJsonObject, JsonTuple,
CsvToStructs, StructsToCsv, XmlToStructs, StructsToXml) maintain mutable state
in @transient lazy val evaluator fields but do not override stateful.
As a result, freshCopyIfContainsStatefulExpression() never copies them, so when
two projections share the same expression node (e.g. via ConvertToLocalRelation
or pipelined execution), threads race on the shared evaluator object and
silently corrupt results.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
