8259947: (fs) Optimize UnixPath.encode implementation #2135

cl4es · 2021-01-19T00:35:51Z

This patch improves UnixPath.encode by reusing JLA.getBytesNoRepl (which has fast-paths for common encoding) and avoiding a toCharArray call on the input by refactoring the normalizeNativePath code to operate on String. This might have a cost on files on Mac that need additional native normalization.

This removes another ThreadLocal and a source of SoftReferences. Together with the UTF-8 fast-path my UTF-8 encoded file system see substantial speed-ups in a trivial new File(str).toPath() microbenchmark.

Progress

Change must not contain extraneous whitespace
Commit message must refer to an issue
Change must be properly reviewed

Issue

JDK-8259947: (fs) Optimize UnixPath.encode implementation

Reviewers

Chris Hegarty (@ChrisHegarty - Reviewer) ⚠️ Review applies to 27a55ee
Aleksey Shipilev (@shipilev - Reviewer)
Alan Bateman (@AlanBateman - Reviewer) ⚠️ Review applies to b023c0a

Download

$ git fetch https://git.openjdk.java.net/jdk pull/2135/head:pull/2135
$ git checkout pull/2135

…caching - use FileOpen as a baseline

cl4es · 2021-01-19T00:36:48Z

Microbenchmark results, baseline:

Benchmark                 Mode  Cnt     Score    Error  Units
FileToPath.mix            avgt   15  1669.996 ± 76.308  ns/op
FileToPath.normalized     avgt   15   349.300 ± 16.851  ns/op
FileToPath.notNormalized  avgt   15   553.013 ± 28.736  ns/op
FileToPath.trailingSlash  avgt   15   415.107 ± 18.322  ns/op

Target:

Benchmark                 Mode  Cnt    Score    Error  Units
FileToPath.mix            avgt   15  887.191 ± 34.167  ns/op
FileToPath.normalized     avgt   15  132.653 ±  2.907  ns/op
FileToPath.notNormalized  avgt   15  333.678 ± 17.665  ns/op
FileToPath.trailingSlash  avgt   15  192.272 ±  7.644  ns/op

bridgekeeper · 2021-01-19T00:36:56Z

👋 Welcome back redestad! A progress list of the required criteria for merging this PR into master will be added to the body of your pull request. There are additional pull request commands available for use with this pull request.

openjdk · 2021-01-19T00:38:35Z

@cl4es The following labels will be automatically applied to this pull request:

core-libs
nio

When this pull request is ready to be reviewed, an "RFR" email will be sent to the corresponding mailing lists. If you would like to change these labels, use the /label pull request command.

mlbridge · 2021-01-19T00:51:08Z

Webrevs

ChrisHegarty

I think that this looks good ( I had a similar thought when looking through this code recently, for a separate issue ).

openjdk · 2021-01-19T10:13:15Z

@cl4es This change now passes all automated pre-integration checks.

ℹ️ This project also has non-automated pre-integration requirements. Please see the file CONTRIBUTING.md for details.

After integration, the commit message for the final commit will be:

8259947: (fs) Optimize UnixPath.encode implementation

Reviewed-by: chegar, shade, alanb

You can use pull request commands such as /summary, /contributor and /issue to adjust it as needed.

At the time when this comment was updated there had been 1 new commit pushed to the master branch:

52ed2aa: 8259786: initialize last parameter of getpwuid_r

Please see this link for an up-to-date comparison between the source branch of this pull request and the master branch.
As there are no conflicts, your changes will automatically be rebased on top of these commits when integrating. If you prefer to avoid this automatic rebasing, please check the documentation for the /integrate command for further details.

➡️ To integrate this PR with the above commit message to the master branch, type /integrate in a new comment.

shipilev · 2021-01-19T10:45:56Z

test/micro/org/openjdk/bench/java/io/FileToPath.java

+    public void mix(Blackhole bh)  {
+        bh.consume(new File(normalFile).toPath());
+        bh.consume(new File(trailingSlash).toPath());
+        bh.consume(new File(root).toPath());


No singular test for new File(root), but here it is in the mix anyway? What would probably be not comparable.

test/micro/org/openjdk/bench/java/io/FileToPath.java

src/java.base/unix/classes/sun/nio/fs/UnixPath.java

AlanBateman · 2021-01-19T12:14:12Z

src/java.base/unix/classes/sun/nio/fs/UnixPath.java

+        input = fs.normalizeNativePath(input);
+        try {
+            return JLA.getBytesNoRepl(input, Util.jnuEncoding());
+        } catch (CharacterCodingException cce) {


The encode method pre-dates JLA.getBytesNoRepl and the recent optimisations so this is a good cleanup.

shipilev · 2021-01-20T13:33:30Z

test/micro/org/openjdk/bench/java/io/FileOpen.java

+    @BenchmarkMode(Mode.AverageTime)
+    @OutputTimeUnit(TimeUnit.NANOSECONDS)
+    @State(Scope.Thread)
+    @Warmup(time=2, iterations=5)
+    @Measurement(time=3, iterations=5)
+    @Fork(value=2, jvmArgs="-Xmx1g")


I thought this can be inherited from the enclosing benchmark.

Maybe it's only @State that doesn't inherit.

shipilev · 2021-01-20T14:06:10Z

test/micro/org/openjdk/bench/java/io/FileOpen.java

+    @Warmup(time=2, iterations=5)
+    @Measurement(time=3, iterations=5)
+    @Fork(value=2, jvmArgs="-Xmx1g")
+    public static class ToPath {


Do we really have to nest this test? It seems to gain nothing per se: the annotations and fields are still pretty much duplicated.

Yeah, no. Nesting was to pre-emptively decouple the micros, but it makes sense to squash them into one for now.

cl4es · 2021-01-20T15:14:06Z

/integrate

openjdk · 2021-01-20T15:15:18Z

@cl4es Since your change was applied there have been 2 commits pushed to the master branch:

69f90b5: 8259843: initialize dli_fname array before calling dll_address_to_library_name
52ed2aa: 8259786: initialize last parameter of getpwuid_r

Your commit was automatically rebased without conflicts.

Pushed as commit 5891509.

💡 You may see a message that your pull request was closed with unmerged commits. This can be safely ignored.

cl4es added 2 commits January 19, 2021 01:24

Optimize UnixPath.encode

6363d04

Add micro. To properly examine cost of toPath() needs a new File due …

27a55ee

…caching - use FileOpen as a baseline

openjdk bot added nio nio-dev@openjdk.org core-libs core-libs-dev@openjdk.org labels Jan 19, 2021

cl4es marked this pull request as ready for review January 19, 2021 00:46

openjdk bot added the rfr Pull request is ready for review label Jan 19, 2021

ChrisHegarty approved these changes Jan 19, 2021

View reviewed changes

cl4es changed the title ~~8259947: Optimize UnixPath.encode~~ 8259947: (fs) Optimize UnixPath.encode implementation Jan 19, 2021

openjdk bot added the ready Pull request is ready to be integrated label Jan 19, 2021

shipilev suggested changes Jan 19, 2021

View reviewed changes

Fold ToPath into FileOpen, add root benchmarks to keep mix comparable

18c3105

AlanBateman reviewed Jan 19, 2021

View reviewed changes

src/java.base/unix/classes/sun/nio/fs/UnixPath.java Outdated Show resolved Hide resolved

AlanBateman reviewed Jan 19, 2021

View reviewed changes

Move JLA to top, add imports

b023c0a

AlanBateman approved these changes Jan 19, 2021

View reviewed changes

shipilev reviewed Jan 20, 2021

View reviewed changes

cl4es added 2 commits January 20, 2021 16:05

Move FileOpen.ToPath micros into top class

698cf75

Merge branch 'master' into unix_encode

98c7619

shipilev approved these changes Jan 20, 2021

View reviewed changes

openjdk bot closed this Jan 20, 2021

openjdk bot added integrated Pull request has been integrated and removed ready Pull request is ready to be integrated rfr Pull request is ready for review labels Jan 20, 2021

8259947: (fs) Optimize UnixPath.encode implementation #2135

8259947: (fs) Optimize UnixPath.encode implementation #2135

Uh oh!

Conversation

cl4es commented Jan 19, 2021 • edited by openjdk bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Progress

Issue

Reviewers

Download

Uh oh!

cl4es commented Jan 19, 2021

Uh oh!

bridgekeeper bot commented Jan 19, 2021

Uh oh!

openjdk bot commented Jan 19, 2021

Uh oh!

mlbridge bot commented Jan 19, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Webrevs

Uh oh!

ChrisHegarty left a comment

Choose a reason for hiding this comment

Uh oh!

openjdk bot commented Jan 19, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

shipilev Jan 19, 2021

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

AlanBateman Jan 19, 2021

Choose a reason for hiding this comment

Uh oh!

shipilev Jan 20, 2021

Choose a reason for hiding this comment

Uh oh!

cl4es Jan 20, 2021

Choose a reason for hiding this comment

Uh oh!

shipilev Jan 20, 2021

Choose a reason for hiding this comment

Uh oh!

cl4es Jan 20, 2021

Choose a reason for hiding this comment

Uh oh!

cl4es commented Jan 20, 2021

Uh oh!

openjdk bot commented Jan 20, 2021

Uh oh!

Reviewers

Assignees

Labels

Milestone

Development

Uh oh!

4 participants

cl4es commented Jan 19, 2021 •

edited by openjdk bot

Loading

mlbridge bot commented Jan 19, 2021 •

edited

Loading

openjdk bot commented Jan 19, 2021 •

edited

Loading