8262901: [macos_aarch64] NativeCallTest expected:<-3.8194101E18> but was:<3.02668882E10> #6641

KaperD · 2021-12-01T16:17:21Z

This is the fix of aarch64 jvmci calling convention.

On MacOS/aarch64 "Function arguments may consume slots on the stack that are not multiples of 8 bytes" [1], but current approach uses only wordsize or bigger slots, which is incorrect (that is why tests were failing [4]). Now arguments consume the right amount of bytes.

Another problem is that current approach don't make 16-byte alignment of Stack Pointer [1][2][3]. However, tests not fail on Linux/aarch64 and Windows/aarch64. They pass because in this tests all functions have even number of argumets, that is why 16-byte alignment comes automatically. But if you try to add or delete one argumets, tests will fail with SIGBUS.

I've tested this patch on MacOS/aarch64 and Linux/aarch64, all tests have passed.

Also I don't understand, why current tests (NativeCallTest) use only int, long, float and double as arguments types. Is it possible to add functions with another types like byte or short? I tried, but it fails on every platform.

[1] https://developer.apple.com/documentation/xcode/writing-arm64-code-for-apple-platforms
[2] https://github.com/ARM-software/abi-aa/blob/main/aapcs64/aapcs64.rst#the-stack
[3] https://docs.microsoft.com/en-us/cpp/build/arm64-windows-abi-conventions?view=msvc-160#stack
[4] https://bugs.openjdk.java.net/browse/JDK-8262901

Progress

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

Issue

JDK-8262901: [macos_aarch64] NativeCallTest expected:<-3.8194101E18> but was:<3.02668882E10>

Reviewers

Anton Kozlov (@AntonKozlov - Committer)
Andrew Haley (@theRealAph - Reviewer)

Reviewing

Using git

Checkout this PR locally:
$ git fetch https://git.openjdk.java.net/jdk pull/6641/head:pull/6641
$ git checkout pull/6641

Update a local copy of the PR:
$ git checkout pull/6641
$ git pull https://git.openjdk.java.net/jdk pull/6641/head

Using Skara CLI tools

Checkout this PR locally:
$ git pr checkout 6641

View PR using the GUI difftool:
$ git pr show -t 6641

Using diff file

Download this PR as a diff file:
https://git.openjdk.java.net/jdk/pull/6641.diff

bridgekeeper · 2021-12-01T16:18:08Z

👋 Welcome back KaperD! 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 · 2021-12-01T16:20:05Z

@KaperD 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 · 2021-12-01T16:25:40Z

Webrevs

01: Full - Incremental (2ff8c037)
00: Full (c76331bc)

theRealAph · 2021-12-01T17:01:13Z

Thanks for this contribution. Which set of tests have you been using?

KaperD · 2021-12-01T18:02:34Z

Thanks for this contribution. Which set of tests have you been using?

Tier1 and tests from compiler/jvmci/jdk.vm.ci.code.test (maybe they are included to Tier1, but I also run them separately)

theRealAph · 2021-12-01T18:17:41Z

Please add some test with an odd number of arguments.

KaperD · 2021-12-02T07:30:26Z

Please add some test with an odd number of arguments.

Done. I took test I32SDILDS, copied it and change last 6 arguments for 1 int argument, also changed return type to int.
Without stack alignment from this patch this test fails with SIGBUS on Linux/aarch64 and MacOS/aarch64 (on MacOS/aarch64 even existing F32SDILDS fails)

KaperD · 2021-12-08T06:14:58Z

I don't know why checks were stoped. What can I do to fix it?

KaperD · 2021-12-08T07:11:32Z

I think that patch is ready to be viewed

KaperD · 2021-12-20T07:49:51Z

@theRealAph Can you review changes?

bridgekeeper · 2022-01-17T13:56:03Z

@KaperD 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!

AntonKozlov

Could someone look at this? The change looks good to me, but I'm not a reviewer. Danil does not have JBS account (yet), so the issue is assigned to me.

dean-long · 2022-01-18T17:21:09Z

Looks OK to me. @dougxc or @tkrodriguez, could you look at this too?

openjdk · 2022-01-18T17:28:22Z

@KaperD 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:

8262901: [macos_aarch64] NativeCallTest expected:<-3.8194101E18> but was:<3.02668882E10>

Reviewed-by: akozlov, aph

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 1232 new commits pushed to the master branch:

a5ebcc0: 8282072: G1: Rename CardSetPtr to CardSetContainerPtr
3da5204: 8283229: compiler/arguments/TestCodeEntryAlignment.java fails with release VMs
3e39304: 8281631: HashMap copy constructor and putAll can over-allocate table
0cf291b: 8283234: Improve @jls usage in java.base
9b8afce: 8283260: gcc is not supported on mac
08cadb4: 8271195: Use largest available large page size smaller than LargePageSizeInBytes when available
de4f04c: 8253495: CDS generates non-deterministic output
4df24c5: 8283230: Improve @jls usage in ElementType
27fe3d7: 8240756: [macos] SwingSet2:TableDemo:Printed Japanese characters were garbled
bacfaa3: 8282414: x86: Enhance the assembler to generate more compact instructions
... and 1222 more: https://git.openjdk.java.net/jdk/compare/e9b36a83160d3c1fa79841692e9fadf336bf7954...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.

As you do not have Committer status in this project an existing Committer must agree to sponsor your change. Possible candidates are the reviewers of this PR (@AntonKozlov, @dean-long, @dougxc, @theRealAph) but any other Committer may sponsor as well.

➡️ To flag this PR as ready for integration with the above commit message, type /integrate in a new comment. (Afterwards, your sponsor types /sponsor in a new comment to perform the integration).

dougxc

Changing my review based on input from @teshull. He will leave comments on further changes that are needed.

teshull · 2022-01-19T08:14:17Z

...es/jdk.vm.ci.hotspot.aarch64/src/jdk/vm/ci/hotspot/aarch64/AArch64HotSpotRegisterConfig.java

@@ -265,11 +268,25 @@ private CallingConvention callingConvention(RegisterArray generalParameterRegist

            if (locations[i] == null) {
                ValueKind<?> valueKind = valueKindFactory.getValueKind(kind);
+                int kindSize = valueKind.getPlatformKind().getSizeInBytes();
+                if (macOS && currentStackOffset % kindSize != 0) {


Shouldn't this logic only be included for HotSpotCallingConventionType.NativeCall, but not JavaCall/JavaCallee?

In native-image, the equivalent logic is here: https://github.com/oracle/graal/blob/cf81d2a7c3cbaa2574d0ecbbc2e76879b7719b20/substratevm/src/com.oracle.svm.core.graal.aarch64/src/com/oracle/svm/core/graal/aarch64/SubstrateAArch64RegisterConfig.java#L333

teshull · 2022-01-19T08:18:04Z

...es/jdk.vm.ci.hotspot.aarch64/src/jdk/vm/ci/hotspot/aarch64/AArch64HotSpotRegisterConfig.java

            }
        }
-
+        if (currentStackOffset % target.stackAlignment != 0) {


I think this must be taken care of somewhere else. Otherwise, amd64 would also have this same problem, as its stack alignment is also 16 bytes

The SIGBUS issue looks to be more a problem with AArch64TestAssembler.java rather than an issue here. It would be best to align stack size at emitGrowStack(cc.getStackSize()); and emitGrowStack(-cc.getStackSize());

The same thing could be done to AMD64TestAssembler.

At least for graal, this more closely matches the logic of AArchFrameMap and AMD64FrameMap

dean-long

Following Doug's lead and removing approval.

teshull · 2022-02-16T14:58:05Z

@KaperD are you planning to update this PR?

bridgekeeper · 2022-03-16T21:07:12Z

@KaperD 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!

bridgekeeper · 2022-04-14T09:14:46Z

@KaperD This pull request has been inactive for more than 8 weeks and will now be automatically closed. If you would like to continue working on this pull request in the future, feel free to reopen it! This can be done using the /open pull request command.

KaperD added 2 commits November 28, 2021 23:17

Add stack alignment and MacOS specific code

1f743ee

Remove NativeCallTest from ProblemList for macosx-aarch64

c76331b

openjdk bot added the rfr Pull request is ready for review label Dec 1, 2021

openjdk bot added the hotspot-compiler hotspot-compiler-dev@openjdk.org label Dec 1, 2021

KaperD changed the title ~~8262901: Fix aarch64 jvmci calling convention~~ 8262901: [macos_aarch64] NativeCallTest expected:<-3.8194101E18> but was:<3.02668882E10> Dec 1, 2021

Add test with an odd number of arguments

2ff8c03

AntonKozlov approved these changes Jan 18, 2022

View reviewed changes

dean-long approved these changes Jan 18, 2022

View reviewed changes

openjdk bot added the ready Pull request is ready to be integrated label Jan 18, 2022

dougxc approved these changes Jan 18, 2022

View reviewed changes

dougxc suggested changes Jan 19, 2022

View reviewed changes

teshull suggested changes Jan 19, 2022

View reviewed changes

dean-long suggested changes Jan 19, 2022

View reviewed changes

openjdk bot removed the ready Pull request is ready to be integrated label Jan 19, 2022

theRealAph approved these changes Mar 17, 2022

View reviewed changes

openjdk bot added the ready Pull request is ready to be integrated label Mar 17, 2022

bridgekeeper bot closed this Apr 14, 2022

omikhaltsova mentioned this pull request Sep 12, 2022

8262901: [macos_aarch64] NativeCallTest expected:<-3.8194101E18> but was:<3.02668882E10> #10238

Closed

3 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

8262901: [macos_aarch64] NativeCallTest expected:<-3.8194101E18> but was:<3.02668882E10> #6641

8262901: [macos_aarch64] NativeCallTest expected:<-3.8194101E18> but was:<3.02668882E10> #6641

KaperD commented Dec 1, 2021 •

edited by openjdk bot

Loading

bridgekeeper bot commented Dec 1, 2021

openjdk bot commented Dec 1, 2021

mlbridge bot commented Dec 1, 2021 •

edited

Loading

theRealAph commented Dec 1, 2021

KaperD commented Dec 1, 2021

theRealAph commented Dec 1, 2021

KaperD commented Dec 2, 2021

KaperD commented Dec 8, 2021

KaperD commented Dec 8, 2021

KaperD commented Dec 20, 2021

bridgekeeper bot commented Jan 17, 2022

AntonKozlov left a comment

dean-long commented Jan 18, 2022

openjdk bot commented Jan 18, 2022 •

edited

Loading

dougxc left a comment

teshull Jan 19, 2022 •

edited

Loading

teshull Jan 19, 2022 •

edited by bridgekeeper bot

Loading

teshull Jan 19, 2022

dean-long left a comment

teshull commented Feb 16, 2022

bridgekeeper bot commented Mar 16, 2022

bridgekeeper bot commented Apr 14, 2022

8262901: [macos_aarch64] NativeCallTest expected:<-3.8194101E18> but was:<3.02668882E10> #6641

8262901: [macos_aarch64] NativeCallTest expected:<-3.8194101E18> but was:<3.02668882E10> #6641

Conversation

KaperD commented Dec 1, 2021 • edited by openjdk bot Loading

Progress

Issue

Reviewers

Reviewing

bridgekeeper bot commented Dec 1, 2021

openjdk bot commented Dec 1, 2021

mlbridge bot commented Dec 1, 2021 • edited Loading

Webrevs

theRealAph commented Dec 1, 2021

KaperD commented Dec 1, 2021

theRealAph commented Dec 1, 2021

KaperD commented Dec 2, 2021

KaperD commented Dec 8, 2021

KaperD commented Dec 8, 2021

KaperD commented Dec 20, 2021

bridgekeeper bot commented Jan 17, 2022

AntonKozlov left a comment

Choose a reason for hiding this comment

dean-long commented Jan 18, 2022

openjdk bot commented Jan 18, 2022 • edited Loading

dougxc left a comment

Choose a reason for hiding this comment

teshull Jan 19, 2022 • edited Loading

Choose a reason for hiding this comment

teshull Jan 19, 2022 • edited by bridgekeeper bot Loading

Choose a reason for hiding this comment

teshull Jan 19, 2022

Choose a reason for hiding this comment

dean-long left a comment

Choose a reason for hiding this comment

teshull commented Feb 16, 2022

bridgekeeper bot commented Mar 16, 2022

bridgekeeper bot commented Apr 14, 2022

KaperD commented Dec 1, 2021 •

edited by openjdk bot

Loading

mlbridge bot commented Dec 1, 2021 •

edited

Loading

openjdk bot commented Jan 18, 2022 •

edited

Loading

teshull Jan 19, 2022 •

edited

Loading

teshull Jan 19, 2022 •

edited by bridgekeeper bot

Loading