8323183: ClassFile API performance improvements #17306

asotona · 2024-01-08T13:38:16Z

ClassFile API performance related improvements have been separated from #17121 into this PR.

These improvements are important to minimize performance regression of
8294961: Convert java.base/java.lang.reflect.ProxyGenerator to use the Classfile API to generate proxy classes #17121

Please review.

Thanks,
Adam

Progress

Change must be properly reviewed (1 review required, with at least 1 Reviewer)
Change must not contain extraneous whitespace
Commit message must refer to an issue

Issue

JDK-8323183: ClassFile API performance improvements (Enhancement - P4)

Reviewers

Claes Redestad (@cl4es - Reviewer)

Reviewing

Using git

Checkout this PR locally:
$ git fetch https://git.openjdk.org/jdk.git pull/17306/head:pull/17306
$ git checkout pull/17306

Update a local copy of the PR:
$ git checkout pull/17306
$ git pull https://git.openjdk.org/jdk.git pull/17306/head

Using Skara CLI tools

Checkout this PR locally:
$ git pr checkout 17306

View PR using the GUI difftool:
$ git pr show -t 17306

Using diff file

Download this PR as a diff file:
https://git.openjdk.org/jdk/pull/17306.diff

Webrev

Link to Webrev Comment

…e Classfile API to generate proxy classes

…ounter.java Co-authored-by: ExE Boss <[email protected]>

This reverts commit 75e31e3.

This reverts commit a21fb26.

This reverts commit 3080d24.

…o use the Classfile API to generate proxy classes" This reverts commit 0267d65.

bridgekeeper · 2024-01-08T13:39:46Z

👋 Welcome back asotona! 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 · 2024-01-08T13:40:28Z

@asotona The following label will be automatically applied to this pull request:

core-libs

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

mlbridge · 2024-01-08T13:43:57Z

Webrevs

src/java.base/share/classes/jdk/internal/classfile/impl/StackCounter.java

liach · 2024-01-08T14:43:58Z

You need to update the slot counting from DirectCodeBuilder and StackMapDecoder to fully avoid creating any MethodTypeDesc.

asotona · 2024-01-08T16:34:48Z

You need to update the slot counting from DirectCodeBuilder and StackMapDecoder to fully avoid creating any MethodTypeDesc.

It would be good to avoid all bottlenecks, however not all of them have equal effect.
This patch avoids MTD construction for every invocation instruction. We may cache the symbol in the relevant Utf8Entry, however it still means to create each MTD at least once. "Hairy" methods doing a lot of various invocations are the most affected.

src/java.base/share/classes/jdk/internal/classfile/impl/StackCounter.java

…ounter.java Co-authored-by: liach <[email protected]>

…eAttributeTools

asotona · 2024-01-09T13:37:45Z

StackMapGenerator benchmark is heavily benefiting from symbols caching for repeated execution.
I've created new CodeAttributeTools::benchmarkStackMapsGenerator and CodeAttributeTools::benchmarkStackCounter with focus on single use of each parsed model to eliminate background caching effect.
Then I've compared version with custom method parameter slots counting and version with symbols cached in NaT entries.
There is insignificant difference between these two approaches.
Explanation may by that repeated uses of the same NaT entry (of the called method) compensate the initial parsing cost.

Based on these results I've decided to revert custom parameter slots counting method and apply the previously proposed solution by @liach .

src/java.base/share/classes/jdk/internal/classfile/impl/StackMapDecoder.java

…apDecoder.java Co-authored-by: Andrey Turbanov <[email protected]>

bridgekeeper · 2024-02-12T14:56:20Z

@asotona This pull request has been inactive for more than 4 weeks and will be automatically closed if another 4 weeks passes without any activity. To avoid this, simply add a new comment to the pull request. Feel free to ask for assistance if you need help with progressing this pull request towards integration!

asotona · 2024-03-04T09:16:26Z

Please review.

Thank you!

cl4es

Improvement looks good, but the microbenchmark needs to sink new instances into a blackhole to make sure we measure the right thing.

cl4es · 2024-03-04T10:08:27Z

src/java.base/share/classes/jdk/internal/classfile/impl/StackCounter.java

 import java.lang.classfile.constantpool.DynamicConstantPoolEntry;
 import java.lang.classfile.constantpool.MemberRefEntry;
 import static java.lang.classfile.ClassFile.*;
+import java.lang.constant.MethodTypeDesc;


Imports are strangely out-of-order and imports in the classfile package look haphazard. Is there some system to it that I don't see? I don't know if we have an applicable style guide to lean on, but alphabetically sorted and static imports split out at the end seem like the convention in the OpenJDK sources.

Fixed, thanks.

cl4es · 2024-03-04T10:15:28Z

src/java.base/share/classes/jdk/internal/classfile/impl/StackCounter.java

        this.methodDesc = methodDesc;
        this.cp = cp;
-        map = new LinkedHashMap<>();
+        targets = new ArrayDeque<>();


Seeing ArrayDeque being put to good use always makes me happy.

cl4es · 2024-03-04T10:25:06Z

test/micro/org/openjdk/bench/jdk/classfile/CodeAttributeTools.java

+    public void benchmarkStackMapsGenerator() {
+        for (var d : data) new StackMapGenerator(


When we return something from a benchmark JMH makes sure to sink that something into a blackhole to avoid JIT assuming the result is unused and eliminating it, wholly or in parts. When it's impractical to return a single thing that captures all the computation in the benchmark it's good practice to let a org.openjdk.jmh.infra.Blackhole consume effects to attain the same effect:

public void benchmarkStackMapsGenerator(Blackhole bh) { for (var d : data) bh.consume(new StackMapGenerator(

Fixed, thanks.

…mance-fixes

openjdk · 2024-03-04T12:26:54Z

@asotona 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:

8323183: ClassFile API performance improvements

Reviewed-by: redestad

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 no new commits pushed to the master branch. If another commit should be pushed before you perform the /integrate command, your PR will be automatically rebased. If you prefer to avoid any potential 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.

asotona · 2024-03-04T15:24:56Z

/integrate

openjdk · 2024-03-04T15:26:28Z

Going to push as commit 0583f73.
Since your change was applied there has been 1 commit pushed to the master branch:

b69d1b5: 8327042: G1: Parallelism used for redirty logged cards needs better control.

Your commit was automatically rebased without conflicts.

openjdk · 2024-03-04T15:26:38Z

@asotona Pushed as commit 0583f73.

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

asotona and others added 14 commits December 15, 2023 13:52

8294961: Convert java.base/java.lang.reflect.ProxyGenerator to use th…

0267d65

…e Classfile API to generate proxy classes

performance improvements

3080d24

StackCounter performance boost

0dc63d4

performance improvements

a21fb26

SplitConstantPool performance fix

b7a60ae

performance improvements

75e31e3

applied the recommended changes

7d0da2c

minor StackCounter fix

41e8793

Update src/java.base/share/classes/jdk/internal/classfile/impl/StackC…

c8f1d30

…ounter.java Co-authored-by: ExE Boss <[email protected]>

StackCounter fix

c6b761a

Revert "performance improvements"

ef71aff

This reverts commit 75e31e3.

Revert "performance improvements"

11ba1ba

This reverts commit a21fb26.

Revert "performance improvements"

f0fae9e

This reverts commit 3080d24.

Revert "8294961: Convert java.base/java.lang.reflect.ProxyGenerator t…

bb2cf21

…o use the Classfile API to generate proxy classes" This reverts commit 0267d65.

asotona mentioned this pull request Jan 8, 2024

8294961: Convert java.base/java.lang.reflect.ProxyGenerator to use the Classfile API to generate proxy classes #17121

Closed

3 tasks

openjdk bot added the rfr Pull request is ready for review label Jan 8, 2024

openjdk bot added the core-libs [email protected] label Jan 8, 2024

mcimadamore reviewed Jan 8, 2024

View reviewed changes

src/java.base/share/classes/jdk/internal/classfile/impl/StackCounter.java Outdated Show resolved Hide resolved

mcimadamore reviewed Jan 8, 2024

View reviewed changes

src/java.base/share/classes/jdk/internal/classfile/impl/StackCounter.java Outdated Show resolved Hide resolved

mcimadamore reviewed Jan 8, 2024

View reviewed changes

src/java.base/share/classes/jdk/internal/classfile/impl/StackCounter.java Outdated Show resolved Hide resolved

asotona added 3 commits January 8, 2024 17:01

improved StackMapDescoder::initFrameLocals performance

c523c7b

applied suggested changes

718af50

applied suggested changes

bbbbbde

liach reviewed Jan 8, 2024

View reviewed changes

src/java.base/share/classes/jdk/internal/classfile/impl/StackCounter.java Outdated Show resolved Hide resolved

Update src/java.base/share/classes/jdk/internal/classfile/impl/StackC…

7939417

…ounter.java Co-authored-by: liach <[email protected]>

asotona added 3 commits January 9, 2024 14:20

improved and extended GenerateStackMaps benchmarks and renamed to Cod…

f281e0a

…eAttributeTools

reverted custom method slots counting in StackCounter

e567029

updated copyright year

5f67f8c

turbanoff reviewed Jan 15, 2024

View reviewed changes

src/java.base/share/classes/jdk/internal/classfile/impl/StackMapDecoder.java Outdated Show resolved Hide resolved

Update src/java.base/share/classes/jdk/internal/classfile/impl/StackM…

5a04ec5

…apDecoder.java Co-authored-by: Andrey Turbanov <[email protected]>

cl4es suggested changes Mar 4, 2024

View reviewed changes

asotona added 2 commits March 4, 2024 12:23

applied suggested changes

2d5ddf6

Merge remote-tracking branch 'openjdk/master' into JDK-8323184-perfor…

747fce8

…mance-fixes

cl4es approved these changes Mar 4, 2024

View reviewed changes

openjdk bot added the ready Pull request is ready to be integrated label Mar 4, 2024

openjdk bot added the integrated Pull request has been integrated label Mar 4, 2024

openjdk bot closed this Mar 4, 2024

openjdk bot removed ready Pull request is ready to be integrated rfr Pull request is ready for review labels Mar 4, 2024

		public void benchmarkStackMapsGenerator() {
		for (var d : data) new StackMapGenerator(

8323183: ClassFile API performance improvements #17306

8323183: ClassFile API performance improvements #17306

Uh oh!

Conversation

asotona commented Jan 8, 2024 • edited by openjdk bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Progress

Issue

Reviewers

Reviewing

Webrev

Uh oh!

bridgekeeper bot commented Jan 8, 2024

Uh oh!

openjdk bot commented Jan 8, 2024

Uh oh!

mlbridge bot commented Jan 8, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Webrevs

Uh oh!

Uh oh!

Uh oh!

Uh oh!

liach commented Jan 8, 2024

Uh oh!

asotona commented Jan 8, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

asotona commented Jan 9, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

bridgekeeper bot commented Feb 12, 2024

Uh oh!

asotona commented Mar 4, 2024

Uh oh!

cl4es left a comment

Choose a reason for hiding this comment

Uh oh!

cl4es Mar 4, 2024

Choose a reason for hiding this comment

Uh oh!

asotona Mar 4, 2024

Choose a reason for hiding this comment

Uh oh!

cl4es Mar 4, 2024

Choose a reason for hiding this comment

Uh oh!

cl4es Mar 4, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

asotona Mar 4, 2024

Choose a reason for hiding this comment

Uh oh!

openjdk bot commented Mar 4, 2024

Uh oh!

asotona commented Mar 4, 2024

Uh oh!

openjdk bot commented Mar 4, 2024

Uh oh!

openjdk bot commented Mar 4, 2024

Uh oh!

Reviewers

Assignees

Labels

Milestone

Development

Uh oh!

5 participants

asotona commented Jan 8, 2024 •

edited by openjdk bot

Loading

mlbridge bot commented Jan 8, 2024 •

edited

Loading

asotona commented Jan 8, 2024 •

edited

Loading

asotona commented Jan 9, 2024 •

edited

Loading

cl4es Mar 4, 2024 •

edited

Loading