8306706: Support out-of-line code generation for MachNodes #13602

merykitty · 2023-04-23T18:22:35Z

Hi,

This patch adds supports for MachNodes to emit an out-of-line piece of code in the stub section of the compiled method. This allows the separation of the uncommon path from the common one, which speeds up the common path a little bit and increases compiled code density. Please take a look and leave reviews.

Thanks a lot.

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-8306706: Support out-of-line code generation for MachNodes

Reviewers

Tobias Hartmann (@TobiHartmann - Reviewer) ⚠️ Review applies to a17bcb76
Vladimir Kozlov (@vnkozlov - Reviewer)

Reviewing

Using git

Checkout this PR locally:
$ git fetch https://git.openjdk.org/jdk.git pull/13602/head:pull/13602
$ git checkout pull/13602

Update a local copy of the PR:
$ git checkout pull/13602
$ git pull https://git.openjdk.org/jdk.git pull/13602/head

Using Skara CLI tools

Checkout this PR locally:
$ git pr checkout 13602

View PR using the GUI difftool:
$ git pr show -t 13602

Using diff file

Download this PR as a diff file:
https://git.openjdk.org/jdk/pull/13602.diff

Webrev

Link to Webrev Comment

bridgekeeper · 2023-04-23T18:23:19Z

👋 Welcome back qamai! 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 · 2023-04-23T18:25:40Z

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

hotspot-compiler

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 · 2023-04-23T18:29:35Z

Webrevs

merykitty · 2023-04-23T18:47:02Z

With this patch, the compiled code for a float-to-int conversion is changed:

Before:

    vcvttss2si %xmm1,%eax
    cmp    $0x80000000,%eax
    jne    DONE
    sub    $0x8,%rsp
    vmovss %xmm1,(%rsp)
    call   Stub::f2i_fixup              ;   {runtime_call StubRoutines (initial stubs)}
    pop    %rax
DONE:


After:

    vcvttss2si %xmm1,%eax
    cmp    $0x80000000,%eax
    je     STUB
CONTINUE:

STUB:
    sub    $0x8,%rsp
    vmovss %xmm1,(%rsp)
    call   Stub::f2i_fixup              ;   {runtime_call StubRoutines (initial stubs)}
    pop    %rax
    jmp    CONTINUE

And there are slight improvements shown in microbenchmarks, although the result differs run-to-run, the patched version seems to be generally more performant:

                                      Before             After
Benchmark             Mode  Cnt    Score    Error    Score    Error  Units   Change
ConvertF2I.d2iArray   avgt    5  266.890 ±  3.277  260.720 ±  1.382  ns/op   -2.31%
ConvertF2I.d2iSingle  avgt    5    0.378 ±  0.005    0.317 ±  0.013  ns/op  -16.14%
ConvertF2I.d2lArray   avgt    5  273.999 ± 12.571  267.862 ±  4.806  ns/op   -2.24%
ConvertF2I.d2lSingle  avgt    5    0.379 ±  0.005    0.348 ±  0.044  ns/op   -8.18%
ConvertF2I.f2iArray   avgt    5  261.549 ±  1.391  255.522 ± 15.133  ns/op   -2.30%
ConvertF2I.f2iSingle  avgt    5    0.378 ±  0.005    0.311 ±  0.007  ns/op  -17.72%
ConvertF2I.f2lArray   avgt    5  272.745 ±  1.661  267.770 ±  7.033  ns/op   -1.82%
ConvertF2I.f2lSingle  avgt    5    0.379 ±  0.007    0.350 ±  0.022  ns/op   -7.65%

merykitty · 2023-04-23T18:54:02Z

The generated node for the stub looks like this:

class convF2I_reg_regStub : public C2CodeStub {
private:
    const convF2I_reg_regNode* _node;
    PhaseRegAlloc* ra_;
    MachOper* opnd_array(uint index) const { return _node->opnd_array(index); }
public:
    convF2I_reg_regStub(const convF2I_reg_regNode* node, PhaseRegAlloc* ra) : _node(node), ra_(ra) {}
    int max_size() const { return 23; }
    void emit(C2_MacroAssembler& masm);
};

And the corresponding node's emit method has an additional section:

void convF2I_reg_regNode::emit(CodeBuffer& cbuf, PhaseRegAlloc* ra_) const {
    cbuf.set_insts_mark();
    convF2I_reg_regStub* stub = new (Compile::current()->comp_arena()) convF2I_reg_regStub(this, ra_);
    if (!Compile::current()->output()->in_scratch_emit_size()) {
        Compile::current()->output()->add_stub(stub);
    }
    ...
}

TobiHartmann

Great work. I'm just wondering if the extra complexity is justified for optimizing only the floating point conversions. Do you plan to use this for other optimizations?

merykitty · 2023-05-17T14:20:36Z

@TobiHartmann Thanks for taking a look, I think this can be used for the vectorized version of these nodes, as well as the max, min nodes for floating point numbers. I also see compact header uses out-of-line code to slow path LoadNKlass.

vnkozlov

Is it possible to do this in c2_MacroAssembler_x86 instead (as for verified_entry)?
We are trying to move complex coding from .ad files to macroassembler.

merykitty · 2023-05-17T18:26:26Z

@vnkozlov Yes we can explicitly define a stub without relying on code generation, it may be more preferable since it avoids adding complexity to adlc generation. The only downside is that there is some boilerplate for each usage but I think the boilerplate is not too terrible.

vnkozlov · 2023-05-17T22:10:38Z

@vnkozlov Yes we can explicitly define a stub without relying on code generation, it may be more preferable since it avoids adding complexity to adlc generation. The only downside is that there is some boilerplate for each usage but I think the boilerplate is not too terrible.

Can you look on that? There could be other cases in Macroassembler which can use this

openjdk · 2023-05-18T13:13:55Z

@merykitty Please do not rebase or force-push to an active PR as it invalidates existing review comments. Note for future reference, the bots always squash all changes into a single commit automatically as part of the integration. See OpenJDK Developers’ Guide for more information.

merykitty · 2023-05-18T14:16:09Z

@TobiHartmann @vnkozlov I have reworked the patch, now it relies on template instead of adlc generation to achieve the desired behaviours, I think this is a much more preferable approach.

vnkozlov

Clever. Let me test it.

vnkozlov

My testing passed. Good.

openjdk · 2023-05-18T23:10:30Z

@merykitty 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:

8306706: Support out-of-line code generation for MachNodes

Reviewed-by: thartmann, kvn

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 83 new commits pushed to the master branch:

710453c: 8308016: Use snippets in java.io package
e9320f3: 8308116: jdk.test.lib.compiler.InMemoryJavaCompiler.compile does not close files
97d3b27: 8307523: [vectorapi] Optimize MaskFromLongBenchmark.java
bb0ff48: 8305091: Change ChaCha20 cipher init behavior to match AES-GCM
c0c4d77: 8308544: Fix compilation regression from JDK-8306983 on musl libc
9e196b3: 8308565: HttpClient: Sanitize logging while stopping
582ddeb: 8308545: java/net/httpclient/ShutdownNow.java fails with "stream 1 cancelled"
1cfb265: 8307814: In the case of two methods with Record Patterns, the second one contains a line number from the first method
eb11508: 8308281: Java snippets in the FFM API need to be updated
26227a6: 8305073: Fix VerifyLoopOptimizations - step 2 - verify idom
... and 73 more: https://git.openjdk.org/jdk/compare/902585bec1d4d5681208213bea180302d1b52df9...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.

TobiHartmann

Nice! The new version looks good to me.

src/hotspot/share/opto/c2_CodeStubs.hpp

merykitty · 2023-05-23T16:07:35Z

@vnkozlov Thanks for your reviews and testing
@TobiHartmann Thanks for your suggestion, I have added comments to describe the purpose of C2GeneralStub

vnkozlov

Good.

merykitty · 2023-05-23T17:04:11Z

/integrate

openjdk · 2023-05-23T17:06:34Z

Going to push as commit ab241b3.
Since your change was applied there have been 83 commits pushed to the master branch:

710453c: 8308016: Use snippets in java.io package
e9320f3: 8308116: jdk.test.lib.compiler.InMemoryJavaCompiler.compile does not close files
97d3b27: 8307523: [vectorapi] Optimize MaskFromLongBenchmark.java
bb0ff48: 8305091: Change ChaCha20 cipher init behavior to match AES-GCM
c0c4d77: 8308544: Fix compilation regression from JDK-8306983 on musl libc
9e196b3: 8308565: HttpClient: Sanitize logging while stopping
582ddeb: 8308545: java/net/httpclient/ShutdownNow.java fails with "stream 1 cancelled"
1cfb265: 8307814: In the case of two methods with Record Patterns, the second one contains a line number from the first method
eb11508: 8308281: Java snippets in the FFM API need to be updated
26227a6: 8305073: Fix VerifyLoopOptimizations - step 2 - verify idom
... and 73 more: https://git.openjdk.org/jdk/compare/902585bec1d4d5681208213bea180302d1b52df9...master

Your commit was automatically rebased without conflicts.

openjdk · 2023-05-23T17:06:44Z

@merykitty Pushed as commit ab241b3.

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

vamsi-parasa · 2025-06-16T22:03:04Z

src/hotspot/cpu/x86/c2_MacroAssembler_x86.cpp

+    }
+  }
+
+  auto stub = C2CodeStub::make<Register, XMMRegister, address>(dst, src, slowpath_target, 23, convertF2I_slowpath);


Hi @merykitty, could you please explain how the size 23 was computed? This value does not work with APX and I created a PR (#25787) for that.

@vamsi-parasa Hi, I just manually assembled the snippet and see its size, for such a small snippet it is easy to see that the size is indeed the largest possible.

openjdk bot added the rfr Pull request is ready for review label Apr 23, 2023

openjdk bot added the hotspot-compiler [email protected] label Apr 23, 2023

TobiHartmann reviewed May 17, 2023

View reviewed changes

vnkozlov reviewed May 17, 2023

View reviewed changes

merykitty added 3 commits May 18, 2023 20:06

refactor

d3fb0bb

add microbenchmark

2e5f573

trailing new lines

262e7a3

merykitty force-pushed the inststub branch from 3b13e9e to 262e7a3 Compare May 18, 2023 13:13

openjdk bot removed the rfr Pull request is ready for review label May 18, 2023

whitespace

a17bcb7

openjdk bot added the rfr Pull request is ready for review label May 18, 2023

vnkozlov reviewed May 18, 2023

View reviewed changes

vnkozlov approved these changes May 18, 2023

View reviewed changes

openjdk bot added the ready Pull request is ready to be integrated label May 18, 2023

TobiHartmann approved these changes May 23, 2023

View reviewed changes

src/hotspot/share/opto/c2_CodeStubs.hpp Show resolved Hide resolved

comments describe C2GeneralStub

76d181b

vnkozlov approved these changes May 23, 2023

View reviewed changes

openjdk bot added the integrated Pull request has been integrated label May 23, 2023

openjdk bot closed this May 23, 2023

openjdk bot removed ready Pull request is ready to be integrated rfr Pull request is ready for review labels May 23, 2023

vamsi-parasa reviewed Jun 16, 2025

View reviewed changes

8306706: Support out-of-line code generation for MachNodes #13602

8306706: Support out-of-line code generation for MachNodes #13602

Uh oh!

Conversation

merykitty commented Apr 23, 2023 • 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 Apr 23, 2023

Uh oh!

openjdk bot commented Apr 23, 2023

Uh oh!

mlbridge bot commented Apr 23, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Webrevs

Uh oh!

merykitty commented Apr 23, 2023

Uh oh!

merykitty commented Apr 23, 2023

Uh oh!

TobiHartmann left a comment

Choose a reason for hiding this comment

Uh oh!

merykitty commented May 17, 2023

Uh oh!

vnkozlov left a comment

Choose a reason for hiding this comment

Uh oh!

merykitty commented May 17, 2023

Uh oh!

vnkozlov commented May 17, 2023

Uh oh!

openjdk bot commented May 18, 2023

Uh oh!

merykitty commented May 18, 2023

Uh oh!

vnkozlov left a comment

Choose a reason for hiding this comment

Uh oh!

vnkozlov left a comment

Choose a reason for hiding this comment

Uh oh!

openjdk bot commented May 18, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

TobiHartmann left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

merykitty commented May 23, 2023

Uh oh!

vnkozlov left a comment

Choose a reason for hiding this comment

Uh oh!

merykitty commented May 23, 2023

Uh oh!

openjdk bot commented May 23, 2023

Uh oh!

openjdk bot commented May 23, 2023

Uh oh!

vamsi-parasa Jun 16, 2025

Choose a reason for hiding this comment

Uh oh!

merykitty Jun 19, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

merykitty commented Apr 23, 2023 •

edited by openjdk bot

Loading

mlbridge bot commented Apr 23, 2023 •

edited

Loading

openjdk bot commented May 18, 2023 •

edited

Loading