[HADOOP-11526] Memory leak in Bzip2Compressor and Bzip2Decompressor

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.6.0
• Fix Version/s: 2.7.0
• Component/s: io, native
• Target Version/s: 2.7.0
• Hadoop Flags: Reviewed
• Assignee: Anu Engineer
• Reporter: Ian Rogers
• Created: 30/Jan/15 03:57
• Updated: 10/Apr/15 20:04
• Resolved: 06/Feb/15 01:05

Description

The use of JNI's GetStringUTFChars should be paired with ReleaseStringUTFChars or else the utf-8 char* created by Java's JNI implementation is leaked. It isn't in Bzip2Decompressor.c:

https://apache.googlesource.com/hadoop-common/+/refs/heads/trunk/hadoop-common-project/hadoop-common/src/main/native/src/org/apache/hadoop/io/compress/bzip2/Bzip2Decompressor.c#45

A less error-prone way of handling JNI resources like local references and UTF strings is to use a smart pointer like the Apache licensed code in Android's ScopedLocalRef and ScopedUtfChars:

https://android.googlesource.com/platform/libnativehelper/+/jb-mr1.1-dev-plus-aosp/include/nativehelper/ScopedLocalRef.h
https://android.googlesource.com/platform/libnativehelper/+/jb-mr1.1-dev-plus-aosp/include/nativehelper/ScopedUtfChars.h