8265448: (zipfs): Reduce read contention in ZipFileSystem

Reviewed-by: alanb, lancea
This commit is contained in:
Jason Zaugg 2021-05-11 18:06:37 +00:00 committed by Lance Andersen
parent acf02ed553
commit 0a12605df8
3 changed files with 101 additions and 13 deletions

View File

@ -1227,8 +1227,12 @@ class ZipFileSystem extends FileSystem {
}
private long readFullyAt(ByteBuffer bb, long pos) throws IOException {
synchronized(ch) {
return ch.position(pos).read(bb);
if (ch instanceof FileChannel fch) {
return fch.read(bb, pos);
} else {
synchronized(ch) {
return ch.position(pos).read(bb);
}
}
}
@ -2116,7 +2120,7 @@ class ZipFileSystem extends FileSystem {
// streams.add(eis);
return eis;
} else { // untouched CEN or COPY
eis = new EntryInputStream(e, ch);
eis = new EntryInputStream(e);
}
if (e.method == METHOD_DEFLATED) {
// MORE: Compute good size for inflater stream:
@ -2173,15 +2177,12 @@ class ZipFileSystem extends FileSystem {
// Inner class implementing the input stream used to read
// a (possibly compressed) zip file entry.
private class EntryInputStream extends InputStream {
private final SeekableByteChannel zfch; // local ref to zipfs's "ch". zipfs.ch might
// point to a new channel after sync()
private long pos; // current position within entry data
private long rem; // number of remaining bytes within entry
EntryInputStream(Entry e, SeekableByteChannel zfch)
EntryInputStream(Entry e)
throws IOException
{
this.zfch = zfch;
rem = e.csize;
pos = e.locoff;
if (pos == -1) {
@ -2206,14 +2207,10 @@ class ZipFileSystem extends FileSystem {
if (len > rem) {
len = (int) rem;
}
// readFullyAt()
long n;
ByteBuffer bb = ByteBuffer.wrap(b);
bb.position(off);
bb.limit(off + len);
synchronized(zfch) {
n = zfch.position(pos).read(bb);
}
long n = readFullyAt(bb, pos);
if (n > 0) {
pos += n;
rem -= n;

View File

@ -499,10 +499,12 @@ public class ZipFSTester {
}
// System.out.printf(" --> %d, %d%n", pos, len);
bb.position(0).limit(len); // bb.flip().limit(len);
int expectedReadResult = sbc.size() == 0 ? -1 : len;
if (sbc.position(pos).position() != pos ||
sbc.read(bb) != len ||
sbc.read(bb) != expectedReadResult ||
!Arrays.equals(buf, 0, bb.position(), expected, pos, pos + len)) {
System.out.printf("read()/position() failed%n");
throw new RuntimeException("CHECK FAILED!");
}
}
} catch (IOException x) {

View File

@ -0,0 +1,89 @@
/*
* Copyright (c) 2020, Oracle and/or its affiliates. All rights reserved.
* DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
*
* This code is free software; you can redistribute it and/or modify it
* under the terms of the GNU General Public License version 2 only, as
* published by the Free Software Foundation.
*
* This code is distributed in the hope that it will be useful, but WITHOUT
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
* FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
* version 2 for more details (a copy is included in the LICENSE file that
* accompanied this code).
*
* You should have received a copy of the GNU General Public License version
* 2 along with this work; if not, write to the Free Software Foundation,
* Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
*
* Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
* or visit www.oracle.com if you need additional information or have any
* questions.
*/
package org.openjdk.bench.jdk.nio.zipfs;
import org.openjdk.jmh.annotations.*;
import org.openjdk.jmh.infra.Blackhole;
import java.io.IOException;
import java.io.InputStream;
import java.nio.file.FileSystem;
import java.nio.file.Files;
import java.nio.file.Path;
import java.nio.file.spi.FileSystemProvider;
import java.util.Map;
import java.util.Random;
import java.util.concurrent.TimeUnit;
@BenchmarkMode(Mode.AverageTime)
@Warmup(iterations = 5, time = 500, timeUnit = TimeUnit.MILLISECONDS)
@Measurement(iterations = 10, time = 500, timeUnit = TimeUnit.MILLISECONDS)
@State(Scope.Benchmark)
@Threads(Threads.MAX)
@OutputTimeUnit(TimeUnit.MILLISECONDS)
@Fork(value = 3)
public class ZipFileSystemBenchmark {
private static final String FILE_NAME = "filename";
private FileSystemProvider jarFsProvider;
private Path readPath;
private FileSystem fileSystem;
private Path zip;
@Setup(Level.Trial) public void setup() throws IOException {
jarFsProvider = FileSystemProvider.installedProviders().stream().filter(x -> x.getScheme().equals("jar")).findFirst().get();
zip = Files.createTempFile("zipfs-benchmark", ".jar");
createTestZip();
fileSystem = jarFsProvider.newFileSystem(zip, Map.of());
Path rootRead = fileSystem.getRootDirectories().iterator().next();
readPath = rootRead.resolve(FILE_NAME);
}
private void createTestZip() throws IOException {
Files.delete(zip);
FileSystem writableFileSystem = jarFsProvider.newFileSystem(zip, Map.of("create", "true"));
byte[] data = new byte[16 * 1024 * 1024];
new Random(31).nextBytes(data);
Path root = writableFileSystem.getRootDirectories().iterator().next();
Files.write(root.resolve(FILE_NAME), data);
writableFileSystem.close();
}
@TearDown public void tearDown() throws IOException {
if (fileSystem != null) {
fileSystem.close();
}
Files.deleteIfExists(zip);
}
// Performance should remain constant when varying the number of threads up to the
// number of physical cores if the NIO implementation on the platform supports
// concurrent reads to a single FileChannel instance. At the time of writing, NIO on Windows
// serializes access.
@Benchmark public void read(Blackhole bh) throws IOException {
InputStream inputStream = Files.newInputStream(readPath);
byte[] buffer = new byte[8192];
while (inputStream.read(buffer) != -1) {
bh.consume(buffer);
}
}
}