[SPARK-22838] Avoid unnecessary copying of data

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Invalid
• Affects Version/s: 2.2.1
• Component/s: Spark Core
• Assignee: Unassigned
• Reporter: Xianyang Liu
• Created: 20/Dec/17 03:59
• Updated: 22/Jan/18 04:33
• Resolved: 22/Jan/18 04:33

Description

If we read data from FileChannel to HeapByteBuffer, there is a need to copy the data from the off-heap to the on-heap, you can see the follow code:
```java
static int read(FileDescriptor var0, ByteBuffer var1, long var2, NativeDispatcher var4) throws IOException {
if(var1.isReadOnly())

{ throw new IllegalArgumentException("Read-only buffer"); }

else if(var1 instanceof DirectBuffer)

{ return readIntoNativeBuffer(var0, var1, var2, var4); }

else {
ByteBuffer var5 = Util.getTemporaryDirectBuffer(var1.remaining());

int var7;
try {
int var6 = readIntoNativeBuffer(var0, var5, var2, var4);
var5.flip();
if(var6 > 0)

{ var1.put(var5); }

var7 = var6;
} finally

{ Util.offerFirstTemporaryDirectBuffer(var5); }

return var7;
}
}
```

Issue Links

• links to: [Github] Pull Request #20026 (ConeyLiu)