DRILL-8200: Update Hadoop libs to ≥ 3.2.3 for CVE-2022-26612 (#2525)
* Remove pointless Buffer casts.
Compiling Drill with JDK > 8 will still result in ByteBuffer <-> Buffer cast
exceptions at runtime when running on JDK 8 even though maven.target.version
is set to 8. Setting maven.compiler.release to 8 solves the Buffer casts
but raises a compilation error of package sun.security.jgss does not exist
for JDK 8. There were a few handwritten casts to avoid the Buffer casting
issue but many instances are not covered so the few reverted in this commit
achieve nothing.
* Update Hadoop to 3.2.3.
diff --git a/contrib/storage-phoenix/README.md b/contrib/storage-phoenix/README.md
index 01278b5..fa8467f 100644
--- a/contrib/storage-phoenix/README.md
+++ b/contrib/storage-phoenix/README.md
@@ -103,7 +103,7 @@
1. Download HBase 2.4.2 sources and rebuild with Hadoop 3.
- ```mvn clean install -DskipTests -Dhadoop.profile=3.0 -Dhadoop-three.version=3.2.2```
+ ```mvn clean install -DskipTests -Dhadoop.profile=3.0 -Dhadoop-three.version=3.2.3```
2. Remove the `Ignore` annotation in `PhoenixTestSuite.java`.
diff --git a/distribution/src/main/resources/winutils/hadoop.dll b/distribution/src/main/resources/winutils/hadoop.dll
index 441d3ed..763c40a 100644
--- a/distribution/src/main/resources/winutils/hadoop.dll
+++ b/distribution/src/main/resources/winutils/hadoop.dll
Binary files differ
diff --git a/distribution/src/main/resources/winutils/winutils.exe b/distribution/src/main/resources/winutils/winutils.exe
index 75be699..b2c4819 100644
--- a/distribution/src/main/resources/winutils/winutils.exe
+++ b/distribution/src/main/resources/winutils/winutils.exe
Binary files differ
diff --git a/docs/dev/HadoopWinutils.md b/docs/dev/HadoopWinutils.md
index a9ead27..aab93fa 100644
--- a/docs/dev/HadoopWinutils.md
+++ b/docs/dev/HadoopWinutils.md
@@ -3,7 +3,7 @@
Hadoop Winutils native libraries are required to run Drill on Windows. The last version present in maven repository is 2.7.1 and is not updated anymore.
That's why Winutils version matching Hadoop version used in Drill is located in distribution/src/main/resources.
-Current Winutils version: *3.2.2.*
+Current Winutils version: *3.2.3.*
## References
- Official wiki: [Windows Problems](https://cwiki.apache.org/confluence/display/HADOOP2/WindowsProblems).
diff --git a/exec/java-exec/src/main/java/org/apache/parquet/hadoop/ColumnChunkIncReadStore.java b/exec/java-exec/src/main/java/org/apache/parquet/hadoop/ColumnChunkIncReadStore.java
index 3ad0a7a..773a861 100644
--- a/exec/java-exec/src/main/java/org/apache/parquet/hadoop/ColumnChunkIncReadStore.java
+++ b/exec/java-exec/src/main/java/org/apache/parquet/hadoop/ColumnChunkIncReadStore.java
@@ -18,7 +18,6 @@
package org.apache.parquet.hadoop;
import java.io.IOException;
-import java.nio.Buffer;
import java.nio.ByteBuffer;
import java.util.ArrayList;
import java.util.HashMap;
@@ -216,20 +215,20 @@
// Note that the repetition and definition levels are stored uncompressed in
// the v2 page format.
int pageBufOffset = 0;
- ByteBuffer bb = (ByteBuffer) ((Buffer)pageBuf).position(pageBufOffset);
+ ByteBuffer bb = (ByteBuffer) pageBuf.position(pageBufOffset);
BytesInput repLevelBytes = BytesInput.from(
(ByteBuffer) bb.slice().limit(pageBufOffset + repLevelSize)
);
pageBufOffset += repLevelSize;
- bb = (ByteBuffer) ((Buffer)pageBuf).position(pageBufOffset);
+ bb = (ByteBuffer) pageBuf.position(pageBufOffset);
final BytesInput defLevelBytes = BytesInput.from(
(ByteBuffer) bb.slice().limit(pageBufOffset + defLevelSize)
);
pageBufOffset += defLevelSize;
// we've now reached the beginning of compressed column data
- bb = (ByteBuffer) ((Buffer)pageBuf).position(pageBufOffset);
+ bb = (ByteBuffer) pageBuf.position(pageBufOffset);
final BytesInput colDataBytes = decompressor.decompress(
BytesInput.from((ByteBuffer) bb.slice()),
pageSize - repLevelSize - defLevelSize
