[HIVE-14483] java.lang.ArrayIndexOutOfBoundsException org.apache.orc.impl.TreeReaderFactory$BytesColumnVectorUtil.commonReadByteArrays

• Type: Bug
• Status: Closed
• Priority: Critical
• Resolution: Fixed
• Affects Version/s: 2.1.0
• Fix Version/s: 1.3.0, 2.0.2, 2.1.1, 2.2.0
• Component/s: ORC
• Labels: pull-request-available
• Assignee: Serhii Zadorozhniak
• Reporter: Serhii Zadorozhniak
• Created: 09/Aug/16 14:22
• Updated: 24/Jun/20 00:28
• Resolved: 15/Aug/16 20:59

Description

Error message:

Caused by: java.lang.ArrayIndexOutOfBoundsException: 1024
at org.apache.orc.impl.RunLengthIntegerReaderV2.nextVector(RunLengthIntegerReaderV2.java:369)
at org.apache.orc.impl.TreeReaderFactory$BytesColumnVectorUtil.commonReadByteArrays(TreeReaderFactory.java:1231)
at org.apache.orc.impl.TreeReaderFactory$BytesColumnVectorUtil.readOrcByteArrays(TreeReaderFactory.java:1268)
at org.apache.orc.impl.TreeReaderFactory$StringDirectTreeReader.nextVector(TreeReaderFactory.java:1368)
at org.apache.orc.impl.TreeReaderFactory$StringTreeReader.nextVector(TreeReaderFactory.java:1212)
at org.apache.orc.impl.TreeReaderFactory$ListTreeReader.nextVector(TreeReaderFactory.java:1902)
at org.apache.orc.impl.TreeReaderFactory$StructTreeReader.nextBatch(TreeReaderFactory.java:1737)
at org.apache.orc.impl.RecordReaderImpl.nextBatch(RecordReaderImpl.java:1045)
at org.apache.hadoop.hive.ql.io.orc.RecordReaderImpl.ensureBatch(RecordReaderImpl.java:77)
at org.apache.hadoop.hive.ql.io.orc.RecordReaderImpl.hasNext(RecordReaderImpl.java:89)
at org.apache.hadoop.hive.ql.io.orc.OrcInputFormat$OrcRecordReader.next(OrcInputFormat.java:230)
at org.apache.hadoop.hive.ql.io.orc.OrcInputFormat$OrcRecordReader.next(OrcInputFormat.java:205)
at org.apache.hadoop.hive.ql.io.HiveContextAwareRecordReader.doNext(HiveContextAwareRecordReader.java:350)
... 22 more

How to reproduce?
Configure StringTreeReader which contains StringDirectTreeReader as TreeReader (DIRECT or DIRECT_V2 column encoding)

batchSize = 1026;

invoke method nextVector(ColumnVector previousVector,boolean[] isNull, final int batchSize)

scratchlcv is LongColumnVector with long[] vector (length 1024)

which execute BytesColumnVectorUtil.readOrcByteArrays(stream, lengths, scratchlcv,result, batchSize);

as result in method commonReadByteArrays(stream, lengths, scratchlcv,
result, (int) batchSize) we received ArrayIndexOutOfBoundsException.

If we use StringDictionaryTreeReader, then there is no exception, as we have a verification scratchlcv.ensureSize((int) batchSize, false) before reader.nextVector(scratchlcv, scratchlcv.vector, batchSize);

These changes were made for Hive 2.1.0 by corresponding commit https://github.com/apache/hive/commit/0ac424f0a17b341efe299da167791112e4a953e9#diff-a1cec556fb2db4b69a1a4127a6908177R1467 for task https://issues.apache.org/jira/browse/HIVE-12159 by Owen O'Malley

How to fix?
add only one line :

scratchlcv.ensureSize((int) batchSize, false) ;

in method org.apache.orc.impl.TreeReaderFactory#BytesColumnVectorUtil#commonReadByteArrays(InStream stream, IntegerReader lengths,
LongColumnVector scratchlcv,
BytesColumnVector result, final int batchSize) before invocation lengths.nextVector(scratchlcv, scratchlcv.vector, batchSize);

Issue Links

• is broken by: HIVE-12159 Create vectorized readers for the complex types
• is related to: HIVE-14968 Fix compilation failure on branch-1
• links to: GitHub Pull Request #96
• links to: GitHub Pull Request #96