8259276: C2: Empty expression stack when reexecuting tableswitch/lookupswitch instructions after deoptimization

[iwanowww]

During parsing of lookupswitch and tableswitch instructions C2 may insert a safepoint. Corresponding JVM state has always re-execute bit set since the interpreter will unconditionally re-execute the instruction after deoptimization is over (see AbstractInterpreter::bytecode_should_reexecute for the full list of instructions).

But Parse::do_tableswitch()/Parse::do_lookupswitch() attach wrong JVM state: it describes the state after the instruction since the first thing they do is they pop the operand from the expression stack. Instead, the JVM state before the instruction should be used to respect the re-execution in the interpreter.

The bug manifests as a stack corruption after deoptimization at a broken safepoint.

The fix is to keep the initial JVM state (before the instruction) intact until all the safepoints at the particular instruction are inserted.

Testing:

• hs-tier1 - hs-tier7 (in progress)
• Kitchensink24h w/ -XX:+DeoptimizeALot (in progress)

---

Progress

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

Issue

• JDK-8259276: C2: Empty expression stack when reexecuting tableswitch/lookupswitch instructions after deoptimization

Reviewers

• Dean Long (@dean-long - Reviewer)
• Vladimir Kozlov (@vnkozlov - Reviewer)
• Tobias Hartmann (@TobiHartmann - Reviewer)

Download

$ git fetch https://git.openjdk.java.net/jdk16 pull/130/head:pull/130
$ git checkout pull/130

[bridgekeeper]

👋 Welcome back vlivanov! 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]

@iwanowww 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]

Webrevs

• 00: Full (2045467)

[openjdk]

@iwanowww 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:

8259276: C2: Empty expression stack when reexecuting tableswitch/lookupswitch instructions after deoptimization

Reviewed-by: dlong, kvn, thartmann

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 no new commits pushed to the master branch. If another commit should be pushed before you perform the /integrate command, your PR will be automatically rebased. If you prefer to avoid any potential 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.

[vnkozlov]

Make sense.

[vnkozlov]

Add Fix request for JDK 16 to bug report.

[TobiHartmann]

Looks good. Just wondering, although this triggers with stress testing, should we still add a targeted regression test? (Can be done later, since this is for JDK 16).

[iwanowww]

Thanks for the reviews, Dean, Vladimir, and Tobias.

Just wondering, although this triggers with stress testing, should we still add a targeted regression test? (Can be done later, since this is for JDK 16).

The fix introduces an assert which reliably fires during compilation with the problematic test case (Kitchensink). So, I see less value in a dedicated regression test now.

[TobiHartmann]

The fix introduces an assert which reliably fires during compilation with the problematic test case (Kitchensink). So, I see less value in a dedicated regression test now.

Okay, fair enough.

[iwanowww]

/integrate

[openjdk]

@iwanowww Since your change was applied there has been 1 commit pushed to the master branch:

• c5ab7c3: 8260284: C2: assert(_base == Int) failed: Not an Int

Your commit was automatically rebased without conflicts.

Pushed as commit 81e730e.

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