8361699: C2: assert(can_reduce_phi(n->as_Phi())) failed: Sanity: previous reducible Phi is no longer reducible before SUT

[shipilev]

Fixes C2 EA bug, regression in JDK 22. This looks to be a correctness bug, as we have to handle another case in propagating non-scalar-replaceable state.

Additional testing:

• Linux AArch64 server fastdebug, new regression test fails without the fix, passes with it
• Linux AArch64 server fastdebug, all

---

Progress

• Change must not contain extraneous whitespace
• Commit message must refer to an issue
• JDK-8361699 needs maintainer approval

Issue

• JDK-8361699: C2: assert(can_reduce_phi(n->as_Phi())) failed: Sanity: previous reducible Phi is no longer reducible before SUT (Bug - P3 - Requested)

Reviewers

• Cesar Soares Lucas (@JohnTortugo - no project role)

Reviewing

Using git

Checkout this PR locally:
$ git fetch https://git.openjdk.org/jdk25u-dev.git pull/376/head:pull/376
$ git checkout pull/376

Update a local copy of the PR:
$ git checkout pull/376
$ git pull https://git.openjdk.org/jdk25u-dev.git pull/376/head

Using Skara CLI tools

Checkout this PR locally:
$ git pr checkout 376

View PR using the GUI difftool:
$ git pr show -t 376

Using diff file

Download this PR as a diff file:
https://git.openjdk.org/jdk25u-dev/pull/376.diff

Using Webrev

Link to Webrev Comment

[bridgekeeper]

👋 Welcome back shade! 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]

⚠️ @shipilev 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 (81ba80e4)

[JohnTortugo]

Thanks, LGTM

[shipilev]

/approval request Fixes C2 EA bug, regression in JDK 22, so it affects JDK 25. This is bug in EA, which shows that some EA invariants do not hold in corner cases, meaning it can be a correctness issue. Applies cleanly; reproducer fails without the fix and passes with it. All other tests also pass. The usual risk of C2 changes, this one looks to be fairly straightforward.

[openjdk]

@shipilev
8361699: The approval request has been created successfully.