Package org.apache.hadoop.io.erasurecode.rawcoder.util

Class RSUtil

java.lang.Object

org.apache.hadoop.io.erasurecode.rawcoder.util.RSUtil

@Private
public final class RSUtil
extends Object

Utilities for implementing Reed-Solomon code, used by RS coder. Some of the codes are borrowed from ISA-L implementation (C or ASM codes).

Field Summary

Fields

Modifier and Type	Field	Description
static GaloisField	GF
static final int	PRIMITIVE_ROOT

Method Summary

Modifier and Type	Method	Description
static void	encodeData(byte[] gfTables, int dataLen, byte[][] inputs, int[] inputOffsets, byte[][] outputs, int[] outputOffsets)	Encode a group of inputs data and generate the outputs.
static void	encodeData(byte[] gfTables, ByteBuffer[] inputs, ByteBuffer[] outputs)	See above.
static void	genCauchyMatrix(byte[] a, int m, int k)	Ported from Intel ISA-L library.
static int[]	getPrimitivePower(int numDataUnits, int numParityUnits)
static void	initTables(int k, int rows, byte[] codingMatrix, int matrixOffset, byte[] gfTables)

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Details

GF

public static GaloisField GF

PRIMITIVE_ROOT

public static final int PRIMITIVE_ROOT

See Also:

• Constant Field Values

Method Details

getPrimitivePower

public static int[] getPrimitivePower(int numDataUnits,
 int numParityUnits)

initTables

public static void initTables(int k,
 int rows,
 byte[] codingMatrix,
 int matrixOffset,
 byte[] gfTables)

genCauchyMatrix

public static void genCauchyMatrix(byte[] a,
 int m,
 int k)

Ported from Intel ISA-L library.

Parameters:

k - k.

a - a.

m - m.

encodeData

public static void encodeData(byte[] gfTables,
 int dataLen,
 byte[][] inputs,
 int[] inputOffsets,
 byte[][] outputs,
 int[] outputOffsets)

Encode a group of inputs data and generate the outputs. It's also used for decoding because, in this implementation, encoding and decoding are unified. The algorithm is ported from Intel ISA-L library for compatible. It leverages Java auto-vectorization support for performance.

Parameters:

gfTables - gfTables.

dataLen - dataLen.

inputs - inputs.

inputOffsets - inputOffsets.

outputs - outputs.

outputOffsets - outputOffsets.

encodeData

public static void encodeData(byte[] gfTables,
 ByteBuffer[] inputs,
 ByteBuffer[] outputs)

See above. Try to use the byte[] version when possible.

Parameters:

gfTables - gfTables.

inputs - inputs.

outputs - outputs.