HDDS-2784. Read to ByteBuffer uses wrong offset

[adoroszlai]

What changes were proposed in this pull request?

OzoneFSInputStream#read(ByteBuffer) uses the target buffer's position for offsetting into the temporary array, which may result in an IndexOutOfBoundsException. Offset should always be 0, since the temporary array is sized exactly for the length to read, and it has no extra data at the start.

https://issues.apache.org/jira/browse/HDDS-2784

How was this patch tested?

Added unit test.

[INFO] Tests run: 2, Failures: 0, Errors: 0, Skipped: 0, Time elapsed: 0.143 s - in org.apache.hadoop.fs.ozone.contract.TestOzoneFSInputStreamUnit

https://github.com/adoroszlai/hadoop-ozone/runs/370513667

[cxorm]

Thanks @adoroszlai for the report and work.

Overll LGTM.
A little question, could we add new unit test in the existed file ?
(hadoop-ozone/ozonefs/src/test/java/org/apache/hadoop/fs/ozone/TestOzoneFSInputStream.java)

[adoroszlai]

Thanks @cxorm for reviewing this.

A little question, could we add new unit test in the existed file ?
(hadoop-ozone/ozonefs/src/test/java/org/apache/hadoop/fs/ozone/TestOzoneFSInputStream.java)

That file is an integration test (uses mini cluster) and I proposed moving it to integration-test in #383. The new test cases are proper unit tests, and can be run without cluster.

[adoroszlai]

Thanks @cxorm and @bshashikant for review, and @bshashikant for committing it.