Github user tedyu commented on a diff in the pull request:
https://github.com/apache/spark/pull/6444#discussion_r34850864
--- Diff:
core/src/main/java/org/apache/spark/util/collection/unsafe/sort/UnsafeSorterSpillWriter.java
---
@@ -0,0 +1,146 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.spark.util.collection.unsafe.sort;
+
+import java.io.File;
+import java.io.IOException;
+
+import scala.Tuple2;
+
+import org.apache.spark.executor.ShuffleWriteMetrics;
+import org.apache.spark.serializer.DummySerializerInstance;
+import org.apache.spark.storage.BlockId;
+import org.apache.spark.storage.BlockManager;
+import org.apache.spark.storage.BlockObjectWriter;
+import org.apache.spark.storage.TempLocalBlockId;
+import org.apache.spark.unsafe.PlatformDependent;
+
+/**
+ * Spills a list of sorted records to disk. Spill files have the following
format:
+ *
+ * [# of records (int)] [[len (int)][prefix (long)][data (bytes)]...]
+ */
+final class UnsafeSorterSpillWriter {
+
+ static final int DISK_WRITE_BUFFER_SIZE = 1024 * 1024;
+
+ // Small writes to DiskBlockObjectWriter will be fairly inefficient.
Since there doesn't seem to
+ // be an API to directly transfer bytes from managed memory to the disk
writer, we buffer
+ // data through a byte array.
+ private byte[] writeBuffer = new byte[DISK_WRITE_BUFFER_SIZE];
+
+ private final File file;
+ private final BlockId blockId;
+ private final int numRecordsToWrite;
+ private BlockObjectWriter writer;
+ private int numRecordsSpilled = 0;
+
+ public UnsafeSorterSpillWriter(
+ BlockManager blockManager,
+ int fileBufferSize,
+ ShuffleWriteMetrics writeMetrics,
+ int numRecordsToWrite) throws IOException {
+ final Tuple2<TempLocalBlockId, File> spilledFileInfo =
+ blockManager.diskBlockManager().createTempLocalBlock();
+ this.file = spilledFileInfo._2();
+ this.blockId = spilledFileInfo._1();
+ this.numRecordsToWrite = numRecordsToWrite;
+ // Unfortunately, we need a serializer instance in order to construct
a DiskBlockObjectWriter.
+ // Our write path doesn't actually use this serializer (since we end
up calling the `write()`
+ // OutputStream methods), but DiskBlockObjectWriter still calls some
methods on it. To work
+ // around this, we pass a dummy no-op serializer.
+ writer = blockManager.getDiskWriter(
+ blockId, file, DummySerializerInstance.INSTANCE, fileBufferSize,
writeMetrics);
+ // Write the number of records
+ writeIntToBuffer(numRecordsToWrite, 0);
+ writer.write(writeBuffer, 0, 4);
+ }
+
+ // Based on DataOutputStream.writeLong.
+ private void writeLongToBuffer(long v, int offset) throws IOException {
+ writeBuffer[offset + 0] = (byte)(v >>> 56);
+ writeBuffer[offset + 1] = (byte)(v >>> 48);
+ writeBuffer[offset + 2] = (byte)(v >>> 40);
+ writeBuffer[offset + 3] = (byte)(v >>> 32);
+ writeBuffer[offset + 4] = (byte)(v >>> 24);
+ writeBuffer[offset + 5] = (byte)(v >>> 16);
+ writeBuffer[offset + 6] = (byte)(v >>> 8);
+ writeBuffer[offset + 7] = (byte)(v >>> 0);
+ }
+
+ // Based on DataOutputStream.writeInt.
+ private void writeIntToBuffer(int v, int offset) throws IOException {
+ writeBuffer[offset + 0] = (byte)(v >>> 24);
+ writeBuffer[offset + 1] = (byte)(v >>> 16);
+ writeBuffer[offset + 2] = (byte)(v >>> 8);
+ writeBuffer[offset + 3] = (byte)(v >>> 0);
+ }
+
+ /**
+ * Write a record to a spill file.
+ *
+ * @param baseObject the base object / memory page containing the record
+ * @param baseOffset the base offset which points directly to the record
data.
+ * @param recordLength the length of the record.
+ * @param keyPrefix a sort key prefix
+ */
+ public void write(
+ Object baseObject,
+ long baseOffset,
+ int recordLength,
+ long keyPrefix) throws IOException {
+ if (numRecordsSpilled == numRecordsToWrite) {
+ throw new IllegalStateException(
+ "Number of records written exceeded numRecordsToWrite = " +
numRecordsToWrite);
+ } else {
+ numRecordsSpilled++;
+ }
+ writeIntToBuffer(recordLength, 0);
+ writeLongToBuffer(keyPrefix, 4);
+ int dataRemaining = recordLength;
+ int freeSpaceInWriteBuffer = DISK_WRITE_BUFFER_SIZE - 4 - 8; // space
used by prefix + len
+ long recordReadPosition = baseOffset;
+ while (dataRemaining > 0) {
+ final int toTransfer = Math.min(freeSpaceInWriteBuffer,
dataRemaining);
+ PlatformDependent.copyMemory(
+ baseObject,
+ recordReadPosition,
+ writeBuffer,
+ PlatformDependent.BYTE_ARRAY_OFFSET + (DISK_WRITE_BUFFER_SIZE -
freeSpaceInWriteBuffer),
+ toTransfer);
+ writer.write(writeBuffer, 0, (DISK_WRITE_BUFFER_SIZE -
freeSpaceInWriteBuffer) + toTransfer);
+ recordReadPosition += toTransfer;
+ dataRemaining -= toTransfer;
+ freeSpaceInWriteBuffer = DISK_WRITE_BUFFER_SIZE;
+ }
+ if (freeSpaceInWriteBuffer < DISK_WRITE_BUFFER_SIZE) {
--- End diff --
If I understand correctly, this would be true if the above loop is skipped.
In that case we write 12 bytes. Is this intended ?
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
