msokolov commented on a change in pull request #177:
URL: https://github.com/apache/lucene/pull/177#discussion_r647429868
##########
File path: lucene/core/src/java/org/apache/lucene/store/MMapDirectory.java
##########
@@ -231,55 +224,60 @@ public IndexInput openInput(String name, IOContext
context) throws IOException {
ensureOpen();
ensureCanRead(name);
Path path = directory.resolve(name);
- try (FileChannel c = FileChannel.open(path, StandardOpenOption.READ)) {
- final String resourceDescription = "MMapIndexInput(path=\"" +
path.toString() + "\")";
- final boolean useUnmap = getUseUnmap();
- return ByteBufferIndexInput.newInstance(
- resourceDescription,
- map(resourceDescription, c, 0, c.size()),
- c.size(),
- chunkSizePower,
- new ByteBufferGuard(resourceDescription, useUnmap ? CLEANER : null));
+ final String resourceDescription = "MemorySegmentIndexInput(path=\"" +
path.toString() + "\")";
+ final long fileSize = Files.size(path);
+
+ boolean success = false;
+ final ResourceScope scope = ResourceScope.newSharedScope();
+ try {
+ final MemorySegment[] segments = map(scope, resourceDescription, path,
fileSize);
+ final IndexInput in =
+ MemorySegmentIndexInput.newInstance(
+ resourceDescription, scope, segments, fileSize, chunkSizePower);
+ success = true;
+ return in;
Review comment:
So in this case the `ResourceScope` remains open. Don't we need to close
it when closing the `MMapDirectory`?
##########
File path: lucene/core/src/java/org/apache/lucene/store/MMapDirectory.java
##########
@@ -16,28 +16,17 @@
*/
package org.apache.lucene.store;
-import static java.lang.invoke.MethodHandles.*;
-import static java.lang.invoke.MethodType.methodType;
-
import java.io.IOException;
-import java.lang.invoke.MethodHandle;
-import java.lang.reflect.Field;
-import java.nio.ByteBuffer;
-import java.nio.ByteOrder;
-import java.nio.MappedByteBuffer;
import java.nio.channels.ClosedChannelException; // javadoc @link
-import java.nio.channels.FileChannel;
import java.nio.channels.FileChannel.MapMode;
+import java.nio.file.Files;
import java.nio.file.Path;
-import java.nio.file.StandardOpenOption;
-import java.security.AccessController;
-import java.security.PrivilegedAction;
import java.util.Locale;
-import java.util.Objects;
import java.util.concurrent.Future;
-import org.apache.lucene.store.ByteBufferGuard.BufferCleaner;
+import jdk.incubator.foreign.MemorySegment;
+import jdk.incubator.foreign.ResourceScope;
import org.apache.lucene.util.Constants;
-import org.apache.lucene.util.SuppressForbidden;
+import org.apache.lucene.util.Unwrapable;
Review comment:
Should it be `Unwrappable`? Or maybe ... `Wrapped`?
##########
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:
scary - why do we have this? Who is throwing NPE? do we have null
segments in the array?
##########
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) {
+ return alreadyClosed();
+ }
+ // TODO: maybe open a JDK issue to have a separate, more
+ // meaningful exception for unmapped segments:
+ if (e.getMessage() != null && e.getMessage().contains("closed")) {
+ return alreadyClosed();
+ }
+ return e;
+ }
+
+ RuntimeException handlePositionalIOOBE(String action, long pos) throws
IOException {
+ if (pos < 0L) {
+ return new IllegalArgumentException(action + " negative position: " +
this);
+ } else {
+ throw new EOFException(action + " past EOF: " + this);
+ }
+ }
+
+ private AlreadyClosedException alreadyClosed() {
+ return new AlreadyClosedException("Already closed: " + this);
+ }
+
+ @Override
+ public final byte readByte() throws IOException {
+ try {
+ final byte v = (byte) VH_getByte.get(curSegment, curPosition);
+ curPosition++;
+ return v;
+ } catch (
+ @SuppressWarnings("unused")
+ IndexOutOfBoundsException e) {
+ do {
+ curSegmentIndex++;
+ if (curSegmentIndex >= segments.length) {
+ throw new EOFException("read past EOF: " + this);
+ }
+ curSegment = segments[curSegmentIndex];
+ curPosition = 0L;
+ } while (curSegment.byteSize() == 0L);
+ final byte v = (byte) VH_getByte.get(curSegment, curPosition);
+ curPosition++;
+ return v;
+ } catch (NullPointerException | IllegalStateException e) {
+ throw wrapAlreadyClosedException(e);
+ }
+ }
+
+ @Override
+ public final void readBytes(byte[] b, int offset, int len) throws
IOException {
+ final MemorySegment arraySegment = MemorySegment.ofArray(b);
+ try {
+ arraySegment.asSlice(offset).copyFrom(curSegment.asSlice(curPosition,
len));
+ curPosition += len;
+ } catch (
+ @SuppressWarnings("unused")
+ IndexOutOfBoundsException e) {
+ readBytesBoundary(arraySegment, offset, len);
+ } catch (NullPointerException | IllegalStateException e) {
+ throw wrapAlreadyClosedException(e);
+ }
+ }
+
+ private void readBytesBoundary(final MemorySegment arraySegment, int offset,
int len)
+ throws IOException {
+ try {
+ long curAvail = curSegment.byteSize() - curPosition;
+ while (len > curAvail) {
+ arraySegment.asSlice(offset).copyFrom(curSegment.asSlice(curPosition,
curAvail));
+ len -= curAvail;
+ offset += curAvail;
+ curSegmentIndex++;
+ if (curSegmentIndex >= segments.length) {
+ throw new EOFException("read past EOF: " + this);
+ }
+ curSegment = segments[curSegmentIndex];
+ curPosition = 0L;
+ curAvail = curSegment.byteSize();
+ }
+ arraySegment.asSlice(offset).copyFrom(curSegment.asSlice(curPosition,
len));
Review comment:
Is it OK to call this if len == 0?
##########
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) {
+ return alreadyClosed();
+ }
+ // TODO: maybe open a JDK issue to have a separate, more
+ // meaningful exception for unmapped segments:
+ if (e.getMessage() != null && e.getMessage().contains("closed")) {
+ return alreadyClosed();
+ }
+ return e;
+ }
+
+ RuntimeException handlePositionalIOOBE(String action, long pos) throws
IOException {
+ if (pos < 0L) {
+ return new IllegalArgumentException(action + " negative position: " +
this);
+ } else {
+ throw new EOFException(action + " past EOF: " + this);
+ }
+ }
+
+ private AlreadyClosedException alreadyClosed() {
+ return new AlreadyClosedException("Already closed: " + this);
+ }
+
+ @Override
+ public final byte readByte() throws IOException {
+ try {
+ final byte v = (byte) VH_getByte.get(curSegment, curPosition);
+ curPosition++;
+ return v;
+ } catch (
+ @SuppressWarnings("unused")
+ IndexOutOfBoundsException e) {
+ do {
+ curSegmentIndex++;
+ if (curSegmentIndex >= segments.length) {
+ throw new EOFException("read past EOF: " + this);
+ }
+ curSegment = segments[curSegmentIndex];
+ curPosition = 0L;
+ } while (curSegment.byteSize() == 0L);
+ final byte v = (byte) VH_getByte.get(curSegment, curPosition);
+ curPosition++;
+ return v;
+ } catch (NullPointerException | IllegalStateException e) {
+ throw wrapAlreadyClosedException(e);
+ }
+ }
+
+ @Override
+ public final void readBytes(byte[] b, int offset, int len) throws
IOException {
+ final MemorySegment arraySegment = MemorySegment.ofArray(b);
+ try {
+ arraySegment.asSlice(offset).copyFrom(curSegment.asSlice(curPosition,
len));
+ curPosition += len;
+ } catch (
+ @SuppressWarnings("unused")
+ IndexOutOfBoundsException e) {
+ readBytesBoundary(arraySegment, offset, len);
+ } catch (NullPointerException | IllegalStateException e) {
+ throw wrapAlreadyClosedException(e);
+ }
+ }
+
+ private void readBytesBoundary(final MemorySegment arraySegment, int offset,
int len)
+ throws IOException {
+ try {
+ long curAvail = curSegment.byteSize() - curPosition;
+ while (len > curAvail) {
+ arraySegment.asSlice(offset).copyFrom(curSegment.asSlice(curPosition,
curAvail));
+ len -= curAvail;
+ offset += curAvail;
+ curSegmentIndex++;
+ if (curSegmentIndex >= segments.length) {
+ throw new EOFException("read past EOF: " + this);
+ }
+ curSegment = segments[curSegmentIndex];
+ curPosition = 0L;
+ curAvail = curSegment.byteSize();
+ }
+ arraySegment.asSlice(offset).copyFrom(curSegment.asSlice(curPosition,
len));
+ curPosition += len;
+ } catch (NullPointerException | IllegalStateException e) {
+ throw wrapAlreadyClosedException(e);
+ }
+ }
+
+ @Override
+ public void readLongs(long[] dst, int offset, int length) throws IOException
{
+ if (IS_LITTLE_ENDIAN) {
+ final MemorySegment targetSlice =
+ MemorySegment.ofArray(dst).asSlice((long) offset << 3, (long) length
<< 3);
+ try {
+ targetSlice.copyFrom(curSegment.asSlice(curPosition,
targetSlice.byteSize()));
+ curPosition += targetSlice.byteSize();
+ } catch (
+ @SuppressWarnings("unused")
+ IndexOutOfBoundsException iobe) {
+ super.readLongs(dst, offset, length);
Review comment:
I'm confused about these calls to `super` methods. Is the idea -- we use
the MemorySegment methods if the read is within a single segment, otherwise we
fall back to one-at-a-time decoding? If so, should we eventually handle
multiple bulk reads here, and/or do the bounds-checking arithmetic ourselves
rather than rely on handling IOOBE?
--
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
