8383262: Implement optimized aarch64_get_thread_helper() for Windows/AArch64

[raneashay]

Required not just for the performance optimization from the patch but also because without this patch, several AOT tests fail:

• runtime/cds/MetaspaceAllocGaps
• runtime/cds/appcds/aotCache/HelloAOTCache
• runtime/cds/appcds/aotCode/AOTCodeCompressedOopsTest
• runtime/cds/appcds/aotCode/AOTCodeFlags#Z
• runtime/cds/appcds/aotCode/AOTCodeFlags#default_gc
• runtime/cds/appcds/aotCode/AOTCodeFlags#parallel
• runtime/cds/appcds/aotCode/AOTCodeFlags#shenandoah

The reason for the AOT tests failing without this patch is that without this patch, the MSVC compiler applies an optimization called Identical COMDAT Folding (ICF) that results in both Thread::current() and aarch64_get_thread_helper() being mapped to the same address (since before this patch, aarch64_get_thread_helper() simply called Thread::current()). In FastDebug builds, this results in the assertion failure _hash_table->get(addr) == nullptr. In Release builds, since assertions are dropped, the duplicate insert silently overwrites the table entry, so Release builds don't see the test failure. Slow debug builds also don't see test failures because in slo debug builds, MSVC does not apply ICF.

Patch applies cleanly from JDK tip.

---

• I confirm that I make this contribution in accordance with the OpenJDK Interim AI Policy.

---

Progress

• Change must not contain extraneous whitespace
• Commit message must refer to an issue
• JDK-8383262 needs maintainer approval

Issue

• JDK-8383262: Implement optimized aarch64_get_thread_helper() for Windows/AArch64 (Enhancement - P4 - Requested)

Reviewing

Using git

Checkout this PR locally:
$ git fetch https://git.openjdk.org/jdk25u-dev.git pull/605/head:pull/605
$ git checkout pull/605

Update a local copy of the PR:
$ git checkout pull/605
$ git pull https://git.openjdk.org/jdk25u-dev.git pull/605/head

Using Skara CLI tools

Checkout this PR locally:
$ git pr checkout 605

View PR using the GUI difftool:
$ git pr show -t 605

Using diff file

Download this PR as a diff file:
https://git.openjdk.org/jdk25u-dev/pull/605.diff

Using Webrev

Link to Webrev Comment

[bridgekeeper]

👋 Welcome back arane! 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]

❗ This change is not yet ready to be integrated.
See the Progress checklist in the description for automated requirements.

[openjdk]

This backport pull request has now been updated with issue from the original commit.

[openjdk]

⚠️ @raneashay This change is now ready for you to apply for maintainer approval. This can be done directly in each associated issue or by using the /approval command.

[mlbridge]

Webrevs

• 00: Full (349a2f5a)

[raneashay]

/approval request Several AOT tests fail without this patch because the MSVC compiler applies an optimization called Identical COMDAT Folding (ICF) that results in both Thread::current() and aarch64_get_thread_helper() being mapped to the same address (since before this patch, aarch64_get_thread_helper() simply called Thread::current()). In FastDebug builds, this results in the assertion failure _hash_table->get(addr) == nullptr. In Release builds, since assertions are dropped, the duplicate insert silently overwrites the table entry, so Release builds don't see the test failure. Slow debug builds also don't see test failures because in slo debug builds, MSVC does not apply ICF.

[openjdk]

@raneashay
8383262: The approval request has been created successfully.

[GoeLin]

Hi @raneashay
Please check the failed tests. Next time, please do this before asking for approval.

When was this bug introduced? Is it caused by a new compiler that implements the optimization you mention?

How did you test?

[theRealAph]

In FastDebug builds, this results in the assertion failure _hash_table->get(addr) == nullptr

Where is this assertion?

[theRealAph]

Suggested change

	// block. Unlike ELF, which provides `tlsdesc` relocations that lets
	// block. Unlike ELF, which provides `tlsdesc` relocations that let