uschindler commented on a change in pull request #177: URL: https://github.com/apache/lucene/pull/177#discussion_r647655523
########## File path: lucene/core/src/java/org/apache/lucene/store/MemorySegmentIndexInput.java ########## @@ -0,0 +1,614 @@ +/* + * 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.lucene.store; + +import java.io.EOFException; +import java.io.IOException; +import java.lang.invoke.VarHandle; +import java.nio.ByteOrder; +import java.util.Arrays; +import java.util.Objects; +import jdk.incubator.foreign.MemoryHandles; +import jdk.incubator.foreign.MemorySegment; +import jdk.incubator.foreign.ResourceScope; + +/** + * Base IndexInput implementation that uses an array of MemorySegments to represent a file. + * + * <p>For efficiency, this class requires that the segment size are a power-of-two (<code> + * chunkSizePower</code>). + */ +public abstract class MemorySegmentIndexInput extends IndexInput implements RandomAccessInput { + // We pass 1L as alignment, because currently Lucene file formats are heavy unaligned: :( + static final VarHandle VH_getByte = + MemoryHandles.varHandle(byte.class, 1L, ByteOrder.LITTLE_ENDIAN).withInvokeExactBehavior(); + static final VarHandle VH_getShort = + MemoryHandles.varHandle(short.class, 1L, ByteOrder.LITTLE_ENDIAN).withInvokeExactBehavior(); + static final VarHandle VH_getInt = + MemoryHandles.varHandle(int.class, 1L, ByteOrder.LITTLE_ENDIAN).withInvokeExactBehavior(); + static final VarHandle VH_getLong = + MemoryHandles.varHandle(long.class, 1L, ByteOrder.LITTLE_ENDIAN).withInvokeExactBehavior(); + + static final boolean IS_LITTLE_ENDIAN = (ByteOrder.nativeOrder() == ByteOrder.LITTLE_ENDIAN); + + final boolean isClone; + final long length; + final long chunkSizeMask; + final int chunkSizePower; + final ResourceScope scope; + final MemorySegment[] segments; + + int curSegmentIndex = -1; + MemorySegment + curSegment; // redundant for speed: segments[curSegmentIndex], also marker if closed! + long curPosition; // relative to curSegment, not globally + + public static MemorySegmentIndexInput newInstance( + String resourceDescription, + ResourceScope scope, + MemorySegment[] segments, + long length, + int chunkSizePower) { + if (segments.length == 1) { + return new SingleSegmentImpl( + resourceDescription, scope, segments[0], length, chunkSizePower, false); + } else { + return new MultiSegmentImpl( + resourceDescription, scope, segments, 0, length, chunkSizePower, false); + } + } + + private MemorySegmentIndexInput( + String resourceDescription, + ResourceScope scope, + MemorySegment[] segments, + long length, + int chunkSizePower, + boolean isClone) { + super(resourceDescription); + this.scope = scope; + this.segments = segments; + assert Arrays.stream(segments).map(MemorySegment::scope).allMatch(scope::equals); + this.length = length; + this.chunkSizePower = chunkSizePower; + this.chunkSizeMask = (1L << chunkSizePower) - 1L; + this.isClone = isClone; + this.curSegment = segments[0]; + } + + void ensureOpen() { + if (curSegment == null) { + throw alreadyClosed(); + } + } + + RuntimeException wrapAlreadyClosedException(RuntimeException e) { + if (e instanceof NullPointerException) { Review comment: Thats the same in MMapDirectory using ByteBuffers. We don't do any checks for NULL or OFFSETS anywhere and just catch the exception. The actual check is too expensive. The NullPointerException happens when the IndexInput is closed. We null out the segment on close. Adding some "if stillOpen" is not wanted, as a check on every access is way too expensive. See also Robert's comment below. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org For additional commands, e-mail: issues-h...@lucene.apache.org
