All Downloads are FREE. Search and download functionalities are using the official Maven repository.

com.swirlds.merkledb.collections.LongListDisk Maven / Gradle / Ivy

Go to download

Swirlds is a software platform designed to build fully-distributed applications that harness the power of the cloud without servers. Now you can develop applications with fairness in decision making, speed, trust and reliability, at a fraction of the cost of traditional server-based platforms.

There is a newer version: 0.56.6
Show newest version
/*
 * Copyright (C) 2021-2024 Hedera Hashgraph, LLC
 *
 * Licensed 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 com.swirlds.merkledb.collections;

import static java.lang.Math.toIntExact;

import com.swirlds.common.io.utility.LegacyTemporaryFileBuilder;
import com.swirlds.merkledb.utilities.MerkleDbFileUtils;
import edu.umd.cs.findbugs.annotations.NonNull;
import java.io.IOException;
import java.io.RandomAccessFile;
import java.io.UncheckedIOException;
import java.nio.ByteBuffer;
import java.nio.ByteOrder;
import java.nio.channels.FileChannel;
import java.nio.file.Path;
import java.nio.file.StandardOpenOption;
import java.util.Arrays;
import java.util.Deque;
import java.util.concurrent.ConcurrentLinkedDeque;

/**
 *  A direct on disk implementation of LongList. This implementation creates a temporary file to store the data.
 *  If the user provides a file, it will be used to create a copy of the in the temporary file. If the file is absent,
 *  it will take the name of the file provided by the user and create a temporary file with this name.
 * 

* Unlike the "snapshot" file, the temporary files doesn't contain the header, only the body. * */ public class LongListDisk extends AbstractLongList { private static final String STORE_POSTFIX = "longListDisk"; private static final String DEFAULT_FILE_NAME = "LongListDisk.ll"; /** A temp byte buffer for reading and writing longs */ private static final ThreadLocal TEMP_LONG_BUFFER_THREAD_LOCAL; /** This file channel is to work with the temporary file. */ private final FileChannel currentFileChannel; /** * Path to the temporary file used to store the data. * The field is effectively immutable, however it can't be declared * final because in some cases it has to be initialized in {@link LongListDisk#readBodyFromFileChannelOnInit} */ private Path tempFile; /** A temp byte buffer for transferring data between file channels */ private static final ThreadLocal TRANSFER_BUFFER_THREAD_LOCAL; /** * Offsets of the chunks that are free to be used. The offsets are relative to the start of the file. */ private final Deque freeChunks; static { TRANSFER_BUFFER_THREAD_LOCAL = new ThreadLocal<>(); // it's initialized as 8 bytes (Long.BYTES) but likely it's going to be resized later TEMP_LONG_BUFFER_THREAD_LOCAL = ThreadLocal.withInitial(() -> ByteBuffer.allocate(Long.BYTES).order(ByteOrder.nativeOrder())); } /** * Create a {@link LongListDisk} with default parameters. */ public LongListDisk() { this(DEFAULT_NUM_LONGS_PER_CHUNK, DEFAULT_MAX_LONGS_TO_STORE, DEFAULT_RESERVED_BUFFER_LENGTH); } LongListDisk(final int numLongsPerChunk, final long maxLongs, final long reservedBufferLength) { super(numLongsPerChunk, maxLongs, reservedBufferLength); try { currentFileChannel = FileChannel.open( createTempFile(DEFAULT_FILE_NAME), StandardOpenOption.CREATE, StandardOpenOption.READ, StandardOpenOption.WRITE); } catch (IOException e) { throw new UncheckedIOException(e); } freeChunks = new ConcurrentLinkedDeque<>(); fillBufferWithZeroes(initOrGetTransferBuffer()); } /** * Create a {@link LongListDisk} on a file, if the file doesn't exist it will be created. * * @param file The file to read and write to * @throws IOException If there was a problem reading the file */ public LongListDisk(final Path file) throws IOException { this(file, DEFAULT_RESERVED_BUFFER_LENGTH); } LongListDisk(final Path file, final long reservedBufferLength) throws IOException { super(file, reservedBufferLength); freeChunks = new ConcurrentLinkedDeque<>(); // IDE complains that the tempFile is not initialized, but it's initialized in readBodyFromFileChannelOnInit // which is called from the constructor of the parent class //noinspection ConstantValue if (tempFile == null) { throw new IllegalStateException("The temp file is not initialized"); } currentFileChannel = FileChannel.open( tempFile, StandardOpenOption.CREATE, StandardOpenOption.READ, StandardOpenOption.WRITE); } /** * Initializes the file channel to a temporary file. * @param path the path to the source file */ @Override protected void onEmptyOrAbsentSourceFile(final Path path) throws IOException { tempFile = createTempFile(path.toFile().getName()); } /** {@inheritDoc} */ @Override protected void readBodyFromFileChannelOnInit(final String sourceFileName, final FileChannel fileChannel) throws IOException { tempFile = createTempFile(sourceFileName); // create temporary file for writing // the warning is suppressed because the file is not supposed to be closed // as this implementation uses a file channel from it. try (final RandomAccessFile rf = new RandomAccessFile(tempFile.toFile(), "rw")) { // ensure that the amount of disk space is enough // two additional chunks are required to accommodate "compressed" first and last chunks in the original file rf.setLength(fileChannel.size() + 2L * memoryChunkSize); FileChannel tempFileCHannel = rf.getChannel(); final int totalNumberOfChunks = calculateNumberOfChunks(size()); final int firstChunkWithDataIndex = toIntExact(minValidIndex.get() / numLongsPerChunk); final int minValidIndexInChunk = toIntExact(minValidIndex.get() % numLongsPerChunk); // copy the first chunk final ByteBuffer transferBuffer = initOrGetTransferBuffer(); // we need to make sure that the chunk is written in full. // If a value is absent, the list element will have IMPERMISSIBLE_VALUE fillBufferWithZeroes(transferBuffer); transferBuffer.position(minValidIndexInChunk * Long.BYTES); MerkleDbFileUtils.completelyRead(fileChannel, transferBuffer); transferBuffer.flip(); // writing the full chunk, all values before minValidIndexInChunk are zeroes MerkleDbFileUtils.completelyWrite(tempFileCHannel, transferBuffer, 0); chunkList.set(firstChunkWithDataIndex, 0L); // copy everything except for the first chunk and the last chunk MerkleDbFileUtils.completelyTransferFrom( tempFileCHannel, fileChannel, memoryChunkSize, (long) (totalNumberOfChunks - 2) * memoryChunkSize); // copy the last chunk transferBuffer.clear(); MerkleDbFileUtils.completelyRead(fileChannel, transferBuffer); transferBuffer.flip(); MerkleDbFileUtils.completelyWrite( tempFileCHannel, transferBuffer, (long) (totalNumberOfChunks - 1) * memoryChunkSize); for (int i = firstChunkWithDataIndex + 1; i < totalNumberOfChunks; i++) { chunkList.set(i, (long) (i - firstChunkWithDataIndex) * memoryChunkSize); } } } private static void fillBufferWithZeroes(ByteBuffer transferBuffer) { Arrays.fill(transferBuffer.array(), (byte) IMPERMISSIBLE_VALUE); transferBuffer.clear(); } private ByteBuffer initOrGetTransferBuffer() { ByteBuffer buffer = TRANSFER_BUFFER_THREAD_LOCAL.get(); if (buffer == null) { buffer = ByteBuffer.allocate(memoryChunkSize).order(ByteOrder.nativeOrder()); TRANSFER_BUFFER_THREAD_LOCAL.set(buffer); } else { // clean up the buffer buffer.clear(); } return buffer; } static Path createTempFile(final String sourceFileName) throws IOException { return LegacyTemporaryFileBuilder.buildTemporaryDirectory(STORE_POSTFIX).resolve(sourceFileName); } /** {@inheritDoc} */ @Override protected synchronized void putToChunk(final Long chunk, final int subIndex, final long value) { try { final ByteBuffer buf = TEMP_LONG_BUFFER_THREAD_LOCAL.get(); final long offset = chunk + (long) subIndex * Long.BYTES; // write new value to file buf.putLong(0, value); buf.position(0); MerkleDbFileUtils.completelyWrite(currentFileChannel, buf, offset); } catch (final IOException e) { throw new UncheckedIOException(e); } } /** {@inheritDoc} */ @Override protected synchronized boolean putIfEqual( final Long chunk, final int subIndex, final long oldValue, long newValue) { final ByteBuffer buf = TEMP_LONG_BUFFER_THREAD_LOCAL.get(); buf.position(0); try { final long offset = chunk + (long) subIndex * Long.BYTES; MerkleDbFileUtils.completelyRead(currentFileChannel, buf, offset); final long filesOldValue = buf.getLong(0); if (filesOldValue == oldValue) { // write new value to file buf.putLong(0, newValue); buf.position(0); MerkleDbFileUtils.completelyWrite(currentFileChannel, buf, offset); return true; } } catch (final IOException e) { throw new UncheckedIOException(e); } return false; } /** * Calculate the offset in the chunk for the given index. * @param index the index to use * @return the offset in the chunk for the given index */ private long calculateOffsetInChunk(long index) { return (index % numLongsPerChunk) * Long.BYTES; } /** * {@inheritDoc} */ @Override protected void writeLongsData(final FileChannel fc) throws IOException { final ByteBuffer transferBuffer = initOrGetTransferBuffer(); final int totalNumOfChunks = calculateNumberOfChunks(size()); final long currentMinValidIndex = minValidIndex.get(); final int firstChunkWithDataIndex = toIntExact(currentMinValidIndex / numLongsPerChunk); // The following logic sequentially processes chunks. This kind of processing allows to get rid of // non-contiguous memory allocation and gaps that may be present in the current file. // MerkleDbFileUtils.completelyTransferFrom would work faster, it wouldn't allow // the required rearrangement of data. for (int i = firstChunkWithDataIndex; i < totalNumOfChunks; i++) { Long currentChunkStartOffset = chunkList.get(i); // if the chunk is null, we write zeroes to the file. If not, we write the data from the chunk if (currentChunkStartOffset != null) { final long chunkOffset; if (i == firstChunkWithDataIndex) { // writing starts from the first valid index in the first valid chunk final int firstValidIndexInChunk = toIntExact(currentMinValidIndex % numLongsPerChunk); transferBuffer.position(firstValidIndexInChunk * Long.BYTES); chunkOffset = currentChunkStartOffset + calculateOffsetInChunk(currentMinValidIndex); } else { // writing the whole chunk transferBuffer.position(0); chunkOffset = currentChunkStartOffset; } if (i == (totalNumOfChunks - 1)) { // the last array, so set limit to only the data needed final long bytesWrittenSoFar = (long) memoryChunkSize * (long) i; final long remainingBytes = (size() * Long.BYTES) - bytesWrittenSoFar; transferBuffer.limit(toIntExact(remainingBytes)); } else { transferBuffer.limit(transferBuffer.capacity()); } int currentPosition = transferBuffer.position(); MerkleDbFileUtils.completelyRead(currentFileChannel, transferBuffer, chunkOffset); transferBuffer.position(currentPosition); } else { fillBufferWithZeroes(transferBuffer); } MerkleDbFileUtils.completelyWrite(fc, transferBuffer); } } /** * Lookup a long in data * * @param chunkOffset the index of the chunk the long is contained in * @param subIndex The sub index of the long in that chunk * @return The stored long value at given index */ @Override protected long lookupInChunk(@NonNull final Long chunkOffset, final long subIndex) { try { final ByteBuffer buf = TEMP_LONG_BUFFER_THREAD_LOCAL.get(); // if there is nothing to read the buffer will have the default value buf.putLong(0, IMPERMISSIBLE_VALUE); buf.clear(); final long offset = chunkOffset + subIndex * Long.BYTES; MerkleDbFileUtils.completelyRead(currentFileChannel, buf, offset); return buf.getLong(0); } catch (final IOException e) { throw new UncheckedIOException(e); } } /** * Flushes and closes the file chanel and clears the free chunks offset list. */ @Override public void close() { try { // flush if (currentFileChannel.isOpen()) { currentFileChannel.force(false); } // now close currentFileChannel.close(); freeChunks.clear(); } catch (final IOException e) { throw new UncheckedIOException(e); } super.close(); } /** {@inheritDoc} */ @Override protected void releaseChunk(@NonNull final Long chunk) { final ByteBuffer transferBuffer = initOrGetTransferBuffer(); fillBufferWithZeroes(transferBuffer); try { currentFileChannel.write(transferBuffer, chunk); } catch (IOException e) { throw new UncheckedIOException(e); } freeChunks.add(chunk); } /** {@inheritDoc} */ @Override protected void partialChunkCleanup( @NonNull final Long chunkOffset, final boolean leftSide, final long entriesToCleanUp) { final ByteBuffer transferBuffer = initOrGetTransferBuffer(); fillBufferWithZeroes(transferBuffer); transferBuffer.limit(toIntExact(entriesToCleanUp * Long.BYTES)); if (leftSide) { try { currentFileChannel.write(transferBuffer, chunkOffset); } catch (IOException e) { throw new UncheckedIOException(e); } } else { long cleanUpOffset = memoryChunkSize - (entriesToCleanUp * Long.BYTES); try { currentFileChannel.write(transferBuffer, chunkOffset + cleanUpOffset); } catch (IOException e) { throw new UncheckedIOException(e); } } } /** {@inheritDoc} */ @Override protected Long createChunk() { Long chunkOffset = freeChunks.poll(); if (chunkOffset == null) { long maxOffset = -1; for (int i = 0; i < chunkList.length(); i++) { Long currentOffset = chunkList.get(i); maxOffset = Math.max(maxOffset, currentOffset == null ? -1 : currentOffset); } return maxOffset == -1 ? 0 : maxOffset + memoryChunkSize; } else { return chunkOffset; } } // exposed for test purposes only - DO NOT USE IN PROD CODE FileChannel getCurrentFileChannel() { return currentFileChannel; } // exposed for test purposes only - DO NOT USE IN PROD CODE LongListDisk resetTransferBuffer() { TRANSFER_BUFFER_THREAD_LOCAL.remove(); return this; } }





© 2015 - 2024 Weber Informatics LLC | Privacy Policy