8339576: Speed up raw bytecode processing in ClassFile API #20863

liach · 2024-09-04T22:41:38Z

Currently, raw bytecode access goes through multiple wrappers, include one from ClassFile API and another ByteBuffer for merged big endian value reads. We can merge the ByteBuffer =into the ClassFile API one (RawBytecodeHelper) for safer access.

RawBytecodeHelper is also restructured so we avoid allocating it on the heap. Large rawNext method is now also inlined into the smaller next method.

Current benchmark results show this significantly speeds up jdk.classfile.Write and some degree of speedup for simple lambda startup. The impact on general application workloads is minuscule, but this doesn't seem to bring any regression.

Pinging @wenshao and @cl4es for review.

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-8339576: Speed up raw bytecode processing in ClassFile API (Sub-task - P4)

Reviewers

Adam Sotona (@asotona - Reviewer)
Claes Redestad (@cl4es - Reviewer)

Contributors

Shaojin Wen <swen@openjdk.org>

Reviewing

Using git

Checkout this PR locally:
$ git fetch https://git.openjdk.org/jdk.git pull/20863/head:pull/20863
$ git checkout pull/20863

Update a local copy of the PR:
$ git checkout pull/20863
$ git pull https://git.openjdk.org/jdk.git pull/20863/head

Using Skara CLI tools

Checkout this PR locally:
$ git pr checkout 20863

View PR using the GUI difftool:
$ git pr show -t 20863

Using diff file

Download this PR as a diff file:
https://git.openjdk.org/jdk/pull/20863.diff

Webrev

Link to Webrev Comment

… for escape analysis

… too costly

liach · 2024-09-04T22:41:47Z

/contributor add swen

bridgekeeper · 2024-09-04T22:42:21Z

👋 Welcome back liach! 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 · 2024-09-04T22:43:19Z

@liach 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:

8339576: Speed up raw bytecode processing in ClassFile API

Co-authored-by: Shaojin Wen <swen@openjdk.org>
Reviewed-by: asotona, 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 24 new commits pushed to the master branch:

7db4d46: 8330159: [C2] Remove or clarify Compile::init_start
9e0ccb8: 8339548: GHA: RISC-V: Use Debian snapshot archive for bootstrap
8fb8cd8: 8339347: keytool -importpass insists prompting the user even if there is no terminal
9e1af8c: 8339285: Test fails with assert(depth < max_critical_stack_depth) failed: can't have more than 10 critical frames
48d7943: 8339342: FieldAllocationCount is mostly unused
e203df4: 8338100: C2: assert(!n_loop->is_member(get_loop(lca))) failed: control must not be back in the loop
98020e4: 8338133: Cleanup direct use of new HtmlTree
b895d7c: 8332423: [PPC64] Remove C1_MacroAssembler::call_c_with_frame_resize
59c4649: 8329959: Update DigestMD5Client.java - fix typo in javadoc string
4ffcf89: 8339619: ProblemList runtime/cds/appcds/jvmti/dumpingWithAgent/DumpingWithJavaAgent.java
... and 14 more: https://git.openjdk.org/jdk/compare/1353601dcc8f9ec3e12dea21dc61b3585a154b13...master

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.

openjdk · 2024-09-04T22:43:36Z

@liach
Contributor Shaojin Wen <swen@openjdk.org> successfully added.

openjdk · 2024-09-04T22:43:55Z

@liach 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.

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

…rror

mlbridge · 2024-09-05T04:50:23Z

Webrevs

01: Full - Incremental (a8b9a820)
00: Full (59a95339)

wenshao · 2024-09-05T06:24:46Z

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

+     * {@link #ILLEGAL}, so we can perform value access without bound checks if
+     * we have a valid opcode.
+     */
+    public boolean next() {


In C1, this cannot be inlined. See if you need to add ForceInline

I don't think we should worry too much about making C1 inline more aggressively.

C2 needs 10000 calls to inline this method, so wenshao is worried. However, this method call is almost always followed by a huge switch to handle different opcode, so I doubt how much of a difference inlining brings.

But wenshao discovered that this method has too many field gets; I should indeed convert them to local variable access if possible.

Yes, things like storing endBci() to a local variable can be great if it both reduces code size and improves interpreter/C1 speed - but don't over-do it as it's likely never-ending work for a kind of optimizations leyden might make pointless.

It is not necessary to use RawBytecodeHelper in the StackMapGenerator#detectFrameOffsets method. It may be better to operate directly on the byte[] bytecode.

We could have just tracked the offsets when labels from DirectCodeBuilder are bound; we can explore such a fix in another patch.

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

asotona

From the Class-File API perspective the changes look good to me.

8339576: Speed up raw bytecode processing in ClassFile API

liach · 2024-09-06T11:42:18Z

Thanks for the reviews!
/integrate

openjdk · 2024-09-06T11:42:52Z

Going to push as commit a1eebbd.
Since your change was applied there have been 25 commits pushed to the master branch:

a35fd38: 8339368: Switch targets are not inflated in CodeModel if no StackMap
7db4d46: 8330159: [C2] Remove or clarify Compile::init_start
9e0ccb8: 8339548: GHA: RISC-V: Use Debian snapshot archive for bootstrap
8fb8cd8: 8339347: keytool -importpass insists prompting the user even if there is no terminal
9e1af8c: 8339285: Test fails with assert(depth < max_critical_stack_depth) failed: can't have more than 10 critical frames
48d7943: 8339342: FieldAllocationCount is mostly unused
e203df4: 8338100: C2: assert(!n_loop->is_member(get_loop(lca))) failed: control must not be back in the loop
98020e4: 8338133: Cleanup direct use of new HtmlTree
b895d7c: 8332423: [PPC64] Remove C1_MacroAssembler::call_c_with_frame_resize
59c4649: 8329959: Update DigestMD5Client.java - fix typo in javadoc string
... and 15 more: https://git.openjdk.org/jdk/compare/1353601dcc8f9ec3e12dea21dc61b3585a154b13...master

Your commit was automatically rebased without conflicts.

openjdk · 2024-09-06T11:42:59Z

@liach Pushed as commit a1eebbd.

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

wenshao and others added 7 commits September 4, 2024 15:02

add benchmark

63d30c6

simplify ByteBuffer improve startup performance

8710a7a

Inline ByteBuffer into RawBytecodeHelper, and try to make it eligible…

0bb7a04

… for escape analysis

remove benchmark

e01f730

Encapsulate RawBytecodeHelper, controlled unchecked access

231e909

Bad assertion

ceb3db0

Use inlined representation for class initializer, loop initialization…

3b92745

… too costly

liach mentioned this pull request Sep 4, 2024

ClassFile built-in ByteBuffer speed up startup #20840

Closed

3 tasks

openjdk bot added the core-libs core-libs-dev@openjdk.org label Sep 4, 2024

ExE-Boss reviewed Sep 5, 2024

View reviewed changes

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

Simplify state transition logic, fix dump logical error and comment e…

59a9533

…rror

liach changed the title ~~Speed up raw bytecode processing in ClassFile API~~ 8339576: Speed up raw bytecode processing in ClassFile API Sep 5, 2024

liach marked this pull request as ready for review September 5, 2024 04:46

openjdk bot added the rfr Pull request is ready for review label Sep 5, 2024

wenshao mentioned this pull request Sep 5, 2024

8339168: Optimize ClassFile Util slotSize #20747

Closed

3 tasks

wenshao reviewed Sep 5, 2024

View reviewed changes

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

asotona approved these changes Sep 5, 2024

View reviewed changes

openjdk bot added the ready Pull request is ready to be integrated label Sep 5, 2024

Avoid repeated field usage, eliminate bound checks

a8b9a82

openjdk bot removed the ready Pull request is ready to be integrated label Sep 5, 2024

github-actions bot pushed a commit to dougxc/openjdk-pr-canary that referenced this pull request Sep 5, 2024

test record for openjdk/jdk#20863

114bd39

8339576: Speed up raw bytecode processing in ClassFile API

cl4es approved these changes Sep 5, 2024

View reviewed changes

openjdk bot added the ready Pull request is ready to be integrated label Sep 5, 2024

liach requested a review from asotona September 5, 2024 17:53

asotona approved these changes Sep 6, 2024

View reviewed changes

openjdk bot added the integrated Pull request has been integrated label Sep 6, 2024

openjdk bot closed this Sep 6, 2024

openjdk bot removed ready Pull request is ready to be integrated rfr Pull request is ready for review labels Sep 6, 2024

ExE-Boss mentioned this pull request Sep 11, 2024

8339260: Move rarely used constants out of ClassFile #20773

Closed

4 tasks

liach deleted the cf/bytebuf branch September 24, 2024 17:22

8339576: Speed up raw bytecode processing in ClassFile API #20863

8339576: Speed up raw bytecode processing in ClassFile API #20863

Uh oh!

Conversation

liach commented Sep 4, 2024 • edited by openjdk bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Progress

Issue

Reviewers

Contributors

Reviewing

Webrev

Uh oh!

liach commented Sep 4, 2024

Uh oh!

bridgekeeper bot commented Sep 4, 2024

Uh oh!

openjdk bot commented Sep 4, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

openjdk bot commented Sep 4, 2024

Uh oh!

openjdk bot commented Sep 4, 2024

Uh oh!

Uh oh!

mlbridge bot commented Sep 5, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Webrevs

Uh oh!

wenshao Sep 5, 2024

Choose a reason for hiding this comment

Uh oh!

cl4es Sep 5, 2024

Choose a reason for hiding this comment

Uh oh!

liach Sep 5, 2024

Choose a reason for hiding this comment

Uh oh!

cl4es Sep 5, 2024

Choose a reason for hiding this comment

Uh oh!

wenshao Sep 5, 2024

Choose a reason for hiding this comment

Uh oh!

liach Sep 5, 2024

Choose a reason for hiding this comment

Uh oh!

Uh oh!

asotona left a comment

Choose a reason for hiding this comment

Uh oh!

liach commented Sep 6, 2024

Uh oh!

openjdk bot commented Sep 6, 2024

Uh oh!

openjdk bot commented Sep 6, 2024

Uh oh!

Uh oh!

liach commented Sep 4, 2024 •

edited by openjdk bot

Loading

openjdk bot commented Sep 4, 2024 •

edited

Loading

mlbridge bot commented Sep 5, 2024 •

edited

Loading