JDK-8253947: Implementation: JEP 388: Windows AArch64 Support #222

rnkovacs · 2021-08-10T21:43:47Z

This is a more recent version of openjdk/jdk11u#2.

Changes since then:

JDK-8254634 (backport of Build failure with VS 2019 (16.5.0) due to C2039 and C2873) was removed, as it had been committed separately.
JDK-8269391 (backport of AArch64: initialize memory allocated for locals according to Windows AArch64 stack page growth requirement in template interpreter) was added.
JDK-8271095 (backport of C4530 was reported from VS 2019 at access bridge) was added.
JDK-8271002 (backport of AArch64: Fix MacroAssembler::get_thread convention) was added.
JDK-8272181 was added, containing the Windows/AArch64 related part of JDK-8271571 (backport of AArch64: Backtracing broken on PAC enabled systems), necessary for a correct build.
Various small merge conflict resolutions.

Similarly to how it was done on tip, we have incorporated parts of JDK-8253015: Aarch64: Move linux code out from generic CPU feature detection by @AntonKozlov into the JDK-8253947: Implementation: JEP 388: Windows AArch64 Support commit.

Please let me know how I can make the review process easier / faster.

Progress

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

Integration blocker

⚠️ Issue of type Backport is not allowed for integrations

Issue

JDK-8253947: Implementation: JEP 388: Windows AArch64 Support

Contributors

Monica Beckwith <[email protected]>
Ludovic Henry <[email protected]>
Bernhard Urban-Forster <[email protected]>

Reviewing

Using git

Checkout this PR locally:
$ git fetch https://git.openjdk.java.net/jdk11u-dev pull/222/head:pull/222
$ git checkout pull/222

Update a local copy of the PR:
$ git checkout pull/222
$ git pull https://git.openjdk.java.net/jdk11u-dev pull/222/head

Using Skara CLI tools

Checkout this PR locally:
$ git pr checkout 222

View PR using the GUI difftool:
$ git pr show -t 222

Using diff file

Download this PR as a diff file:
https://git.openjdk.java.net/jdk11u-dev/pull/222.diff

bridgekeeper · 2021-08-10T21:45:08Z

Hi @rnkovacs, welcome to this OpenJDK project and thanks for contributing!

We do not recognize you as Contributor and need to ensure you have signed the Oracle Contributor Agreement (OCA). If you have not signed the OCA, please follow the instructions. Please fill in your GitHub username in the "Username" field of the application. Once you have signed the OCA, please let us know by writing /signed in a comment in this pull request.

If you already are an OpenJDK Author, Committer or Reviewer, please click here to open a new issue so that we can record that fact. Please use "Add GitHub user rnkovacs" as summary for the issue.

If you are contributing this work on behalf of your employer and your employer has signed the OCA, please let us know by writing /covered in a comment in this pull request.

rnkovacs · 2021-08-10T21:47:49Z

/covered

bridgekeeper · 2021-08-10T21:48:00Z

Thank you! Please allow for a few business days to verify that your employer has signed the OCA. Also, please note that pull requests that are pending an OCA check will not usually be evaluated, so your patience is appreciated!

rnkovacs · 2021-08-10T21:59:40Z

/contributor add Monica Beckwith [email protected]
/contributor add Ludovic Henry [email protected]
/contributor add Bernhard Urban-Forster [email protected]

openjdk · 2021-08-10T22:01:37Z

@rnkovacs
Contributor Monica Beckwith <[email protected]> successfully added.

openjdk · 2021-08-10T22:02:37Z

@rnkovacs
Contributor Ludovic Henry <[email protected]> successfully added.

openjdk · 2021-08-10T22:04:31Z

@rnkovacs
Contributor Bernhard Urban-Forster <[email protected]> successfully added.

rnkovacs · 2021-08-11T20:55:56Z

Removed tabs and trailing spaces from 8253947 and 8254640.

Reviewed-by: kbarrett, dholmes

Reviewed-by: dholmes, kbarrett, aph, stuefe

Reviewed-by: kbarrett, tschatzl, dholmes

Push changes from JDK-8248817 that were accidentally excluded from the commit. Reviewed-by: kbarrett, dholmes

Reviewed-by: adinn, dholmes

Remove unused variables in the AArch64 backend. Detected by compiling with MSVC, which warns about them. Reviewed-by: aph, dholmes

__thread is gcc-ism, instead rely on compiler independent macro. Reviewed-by: dholmes

MSVC employs min/max as macros Co-authored-by: Ludovic Henry <[email protected]> Reviewed-by: tschatzl, kbarrett

…ultiply Co-authored-by: Monica Beckwith <[email protected]> Co-authored-by: Ludovic Henry <[email protected]> Reviewed-by: aph

The original change missed to update an assert. Co-authored-by: Ludovic Henry <[email protected]> Co-authored-by: Bernhard Urban-Forster <[email protected]> Reviewed-by: dholmes

Co-authored-by: Monica Beckwith <[email protected]> Co-authored-by: Ludovic Henry <[email protected]> Co-authored-by: Bernhard Urban-Forster <[email protected]>

…ARM64 build Reviewed-by: ihse, aph

backed out ZGC bits of original change

Reviewed-by: ihse, never, kvn

... Windows AArch64 stack page growth requirement in template interpreter Reviewed-by: adinn, aph

Reviewed-by: stuefe, serb

Reviewed-by: burban, aph

…enabled systems`

lewurm · 2021-08-13T08:31:03Z

Thanks @rnkovacs for taking this over 🙂 Overall it looks good to me (but I'm not a formal reviewer). Do you have some numbers handy on the tests passed before/after this change on other platforms like Linux/AArch64 and Windows/x86_64 that you could share here?

VladimirKempik · 2021-08-13T11:44:46Z

Hello
for 8248414 which is part of this PR, I have made a similar PR few days ago, I will withdraw it.
You backport seems identical except one thing

ul was replaced with ull few times https://github.com/openjdk/jdk11u-dev/pull/215/files#diff-9c5120cff0dd87fbd1122067c44170768702b61cba7b1faf0c04dcba4c8fdfa3L135
this patch ( the place to patch ) is not present in you tree at all, that's strange.

VladimirKempik · 2021-08-13T12:07:37Z

Also, could you please be more specific, which parts of openjdk/jdk@ec9bee68 were taken and which wasn't ?
it's not trivial to find them in this PR.
Thanks

phohensee · 2021-08-13T14:19:37Z

Copy-paste errors are one of the problems that can occur with a giant squashed backport such as this one. It's difficult to review because it's a lot of code and there's no way to tell which code belongs to which JBS issue, and there's no way to triage the almost inevitable problems that will occur with such a large patch. I'd prefer separate per-JBS-issue backports, such as Vladimir's #215.

phohensee · 2021-08-13T20:34:00Z

I ordered, by resolution date, the parents of the above JBS issues merged with the ones from #2.

phohensee · 2021-08-13T20:35:22Z

I've posted a backport PR for JDK-8248403 at #246.

mlbridge · 2021-08-13T20:39:37Z

Webrevs

00: Full (9dd09a22)

rnkovacs · 2021-08-13T22:54:47Z

@VladimirKempik, changes to

src/hotspot/cpu/aarch64/vm_version_aarch64.cpp,
src/hotspot/cpu/aarch64/vm_version_aarch64.hpp, and
src/hotspot/os_cpu/linux_aarch64/vm_version_linux_aarch64.cpp

were taken over from openjdk/jdk@ec9bee6. However, now that we're breaking up the PR into its elements, it might make sense to backport this whole commit separately too. WDYT?

rnkovacs · 2021-08-13T23:03:00Z

Thank you @phohensee. I understand that reviewing such a huge patch is a pain, and the chance of overlooking something goes up with size. I'll proceed to open PRs for the rest of the issues.

VladimirKempik · 2021-08-15T13:29:16Z

@VladimirKempik, changes to

src/hotspot/cpu/aarch64/vm_version_aarch64.cpp,

src/hotspot/cpu/aarch64/vm_version_aarch64.hpp, and

src/hotspot/os_cpu/linux_aarch64/vm_version_linux_aarch64.cpp

were taken over from openjdk/jdk@ec9bee6. However, now that we're breaking up the PR into its elements, it might make sense to backport this whole commit separately too. WDYT?

Ok, I will do the JDK-8248414 ( it's already reviewed) and its follow-ups: JDK-8250824 and JDK-JDK-8251930.
And Anton's fix for vm_version

rnkovacs · 2021-08-20T07:38:51Z

@VladimirKempik, changes to

src/hotspot/cpu/aarch64/vm_version_aarch64.cpp,

src/hotspot/cpu/aarch64/vm_version_aarch64.hpp, and

src/hotspot/os_cpu/linux_aarch64/vm_version_linux_aarch64.cpp

were taken over from openjdk/jdk@ec9bee6. However, now that we're breaking up the PR into its elements, it might make sense to backport this whole commit separately too. WDYT?

Ok, I will do the JDK-8248414 ( it's already reviewed) and its follow-ups: JDK-8250824 and JDK-JDK-8251930.

Thank you. I assume these are prerequisites to JDK-8253015.

And Anton's fix for vm_version

Which one is that?

rnkovacs · 2021-08-20T07:40:06Z

Following @phohensee's list of JBS issues, links to the individual PRs:

VladimirKempik · 2021-08-20T08:35:49Z

@VladimirKempik, changes to

src/hotspot/cpu/aarch64/vm_version_aarch64.cpp,

src/hotspot/cpu/aarch64/vm_version_aarch64.hpp, and

src/hotspot/os_cpu/linux_aarch64/vm_version_linux_aarch64.cpp

were taken over from openjdk/jdk@ec9bee6. However, now that we're breaking up the PR into its elements, it might make sense to backport this whole commit separately too. WDYT?

Ok, I will do the JDK-8248414 ( it's already reviewed) and its follow-ups: JDK-8250824 and JDK-JDK-8251930.

Thank you. I assume these are prerequisites to JDK-8253015.

And Anton's fix for vm_version

Which one is that?

You can take https://bugs.openjdk.java.net/browse/JDK-8253015

RealCLanger · 2021-08-23T22:28:06Z

Hi,
I have a suggestion: For this series of backports you can use dependent PRs. That is, you create a standard PR for the first change. Then, create the next backport to the branch "pr/#" which will appear in the repo after the first PR is opened. This way, the whole chain is based upon each other and can be reviewed in parallel.

You have to integrate the PRs in order then. Once a PR is merged, its successor will automatically be updated to target the master branch.

openjdk · 2021-08-24T05:02:03Z

@rnkovacs this pull request can not be integrated into master due to one or more merge conflicts. To resolve these merge conflicts and update this pull request you can run the following commands in the local repository for your personal fork:

git checkout win-arm64-port
git fetch https://git.openjdk.java.net/jdk11u-dev master
git merge FETCH_HEAD
# resolve conflicts and follow the instructions given by git merge
git commit -m "Merge master"
git push

rnkovacs · 2021-08-30T09:24:14Z

@RealCLanger thanks for the suggestion. I'm opening a chain of PRs for the last few commits in the following order: #274 - #299 - #301 - 8254827 - 8252114. I liked the simplicity of most of the PRs going in independently, but now in the end chaining makes a lot of sense.

mlbridge · 2021-08-30T13:00:56Z

Mailing list message from Andrew Haley on jdk-updates-dev:

On 8/14/21 12:07 AM, Reka Kovacs wrote:

Thank you @phohensee. I understand that reviewing such a huge patch is a pain, and the chance of overlooking something goes up with size. I'll proceed to open PRs for the rest of the issues.

Thanks. In particular, the integer types patch is pretty much standalone,
and it touches many files. It's good for that one to be reviewed and committed
entirely independently, because while it is a requirement for Windows
support it is in no way Windows specific.

--
Andrew Haley (he/him)
Java Platform Lead Engineer
Red Hat UK Ltd. <https://www.redhat.com>
https://keybase.io/andrewhaley
EAC8 43EB D3EF DB98 CC77 2FAD A5CD 6035 332F A671

RealCLanger · 2021-09-08T21:12:39Z

@RealCLanger thanks for the suggestion. I'm opening a chain of PRs for the last few commits in the following order: #274 - #299 - #301 - 8254827 - 8252114. I liked the simplicity of most of the PRs going in independently, but now in the end chaining makes a lot of sense.

Hi @rnkovacs, maybe you want to close this PR in favor of #301 and the others? Looks like you've opened all required PRs by now and this work is done when all of them are reviewed and integrated...

rnkovacs · 2021-10-04T20:22:19Z

@RealCLanger Sure, I'll close this one. I've been keeping it around to track all the smaller PRs, but now that there's only a few left, we can keep those in mind.

RealCLanger · 2021-10-04T20:33:37Z

@RealCLanger Sure, I'll close this one. I've been keeping it around to track all the smaller PRs, but now that there's only a few left, we can keep those in mind.

Thanks. Do we have a list of what's still left? Maybe add it as a comment to #301?

bridgekeeper bot added the oca Needs verification of OCA signatory status label Aug 10, 2021

bridgekeeper bot added the oca-verify Needs verification of OCA signatory status label Aug 10, 2021

rnkovacs force-pushed the win-arm64-port branch 2 times, most recently from 424bcc9 to 72bfbd3 Compare August 11, 2021 20:54

Andrew Haley and others added 18 commits August 12, 2021 14:05

8254635: AArch64: Remove uses of kernel integer types

c6a4c2f

Reviewed-by: kbarrett, dholmes

8254636: Windows: strengthening in ThreadCritical regarding memory model

901600c

Reviewed-by: dholmes, kbarrett, aph, stuefe

8254638: AArch64: Use ATTRIBUTE_ALIGNED helper

29769dc

Reviewed-by: kbarrett, tschatzl, dholmes

8254639: Push missing parts of JDK-8248817

366c535

Push changes from JDK-8248817 that were accidentally excluded from the commit. Reviewed-by: kbarrett, dholmes

8254640: AArch64: Remove uses of long and unsigned long ints

81d8a5b

Reviewed-by: adinn, dholmes

8254641: AArch64: Remove unused variables

2479719

Remove unused variables in the AArch64 backend. Detected by compiling with MSVC, which warns about them. Reviewed-by: aph, dholmes

8254642: AArch64: Use THREAD_LOCAL instead of __thread

afb9035

__thread is gcc-ism, instead rely on compiler independent macro. Reviewed-by: dholmes

8254643: AArch64: Avoid MIN/MAX macros when using MSVC

294b795

MSVC employs min/max as macros Co-authored-by: Ludovic Henry <[email protected]> Reviewed-by: tschatzl, kbarrett

8254644: C1: Fix signature mismatch of LIRGenerator::strengh_reduce_m…

810170c

…ultiply Co-authored-by: Monica Beckwith <[email protected]> Co-authored-by: Ludovic Henry <[email protected]> Reviewed-by: aph

8254645: AArch64: follow up for JDK-8248414

2635cc8

The original change missed to update an assert. Co-authored-by: Ludovic Henry <[email protected]> Co-authored-by: Bernhard Urban-Forster <[email protected]> Reviewed-by: dholmes

8253947: Implementation: JEP 388: Windows AArch64 Support

d00ae44

Co-authored-by: Monica Beckwith <[email protected]> Co-authored-by: Ludovic Henry <[email protected]> Co-authored-by: Bernhard Urban-Forster <[email protected]>

8264182: AArch64: Get rid of --disable-warnings-as-errors on Windows+…

80438ee

…ARM64 build Reviewed-by: ihse, aph

8264183: Windows-AArch64: Enable and test ZGC and ShenandoahGC

8d3c349

backed out ZGC bits of original change

8264184: JVMCI: Enable it for Windows+AArch64

fabd44b

Reviewed-by: ihse, never, kvn

8269391: AArch64: initialize memory allocated for locals according to...

5ce27cf

... Windows AArch64 stack page growth requirement in template interpreter Reviewed-by: adinn, aph

8263136: C4530 was reported from VS 2019 at access bridge

bdb5df3

Reviewed-by: stuefe, serb

8261072: AArch64: Fix MacroAssembler::get_thread convention

050f92c

Reviewed-by: burban, aph

8272181: Windows-AArch64: Backport fix of `Backtracing broken on PAC …

9dd09a2

…enabled systems`

rnkovacs force-pushed the win-arm64-port branch from 72bfbd3 to 9dd09a2 Compare August 12, 2021 22:20

VladimirKempik mentioned this pull request Aug 13, 2021

8248414: AArch64: Remove uses of long and unsigned long ints #215

Closed

3 tasks

bridgekeeper bot removed oca Needs verification of OCA signatory status oca-verify Needs verification of OCA signatory status labels Aug 13, 2021

openjdk bot added the rfr Pull request is ready for review label Aug 13, 2021

rnkovacs mentioned this pull request Aug 22, 2021

8261072: AArch64: Fix MacroAssembler::get_thread convention #271

Closed

2 tasks

openjdk bot added the merge-conflict Pull request has merge conflict with target branch label Aug 24, 2021

AntonKozlov mentioned this pull request Sep 28, 2021

Review work in jdk11u-dev azul-research/macos-aarch64#4

Open

rnkovacs closed this Oct 4, 2021

rnkovacs mentioned this pull request Oct 4, 2021

8248238: Implementation: JEP 388: Windows AArch64 Support #301

Closed

3 tasks

JDK-8253947: Implementation: JEP 388: Windows AArch64 Support #222

JDK-8253947: Implementation: JEP 388: Windows AArch64 Support #222

Uh oh!

Conversation

rnkovacs commented Aug 10, 2021 • edited by openjdk bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Progress

Integration blocker

Issue

Contributors

Reviewing

Uh oh!

bridgekeeper bot commented Aug 10, 2021

Uh oh!

rnkovacs commented Aug 10, 2021

Uh oh!

bridgekeeper bot commented Aug 10, 2021

Uh oh!

rnkovacs commented Aug 10, 2021

Uh oh!

openjdk bot commented Aug 10, 2021

Uh oh!

openjdk bot commented Aug 10, 2021

Uh oh!

openjdk bot commented Aug 10, 2021

Uh oh!

rnkovacs commented Aug 11, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

lewurm commented Aug 13, 2021

Uh oh!

VladimirKempik commented Aug 13, 2021

Uh oh!

VladimirKempik commented Aug 13, 2021

Uh oh!

phohensee commented Aug 13, 2021

Uh oh!

phohensee commented Aug 13, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

phohensee commented Aug 13, 2021

Uh oh!

mlbridge bot commented Aug 13, 2021

Webrevs

Uh oh!

rnkovacs commented Aug 13, 2021

Uh oh!

rnkovacs commented Aug 13, 2021

Uh oh!

VladimirKempik commented Aug 15, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

rnkovacs commented Aug 20, 2021

Uh oh!

rnkovacs commented Aug 20, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

VladimirKempik commented Aug 20, 2021

Uh oh!

RealCLanger commented Aug 23, 2021

Uh oh!

openjdk bot commented Aug 24, 2021

Uh oh!

rnkovacs commented Aug 30, 2021

Uh oh!

mlbridge bot commented Aug 30, 2021

Uh oh!

RealCLanger commented Sep 8, 2021

Uh oh!

rnkovacs commented Oct 4, 2021

Uh oh!

RealCLanger commented Oct 4, 2021

Uh oh!

Reviewers

Assignees

Labels

Milestone

Development

Uh oh!

9 participants

rnkovacs commented Aug 10, 2021 •

edited by openjdk bot

Loading

rnkovacs commented Aug 11, 2021 •

edited

Loading

phohensee commented Aug 13, 2021 •

edited

Loading

VladimirKempik commented Aug 15, 2021 •

edited

Loading

rnkovacs commented Aug 20, 2021 •

edited

Loading