8360219: [AIX] assert(locals_base >= l2) failed: bad placement

[reinrich]

Weaken assertion because it is too strict. While the interpreted caller sometimes has a frame::top_ijava_frame_abi it is sufficient to assert that the locals don't overlap with the smaller frame::parent_ijava_frame_abi because only that's reserved for non-top frames (aka parent frames).

Tested on AIX and Linux ppc: Tier 1-4 of hotspot and jdk. All of langtools and jaxp. Renaissance Suite and SAP specific tests.

Details:

It cannot be assumed that the interpreted caller of the bottom interpreted frame (from a compiled deoptee frame) has a large frame::top_ijava_frame_abi. In an ordinary i2c call it would keep its frame::top_ijava_frame_abi (see call_from_interpreter) but when it was thawed then it'll have only a frame::java_abi (alias for parent_ijava_frame_abi).

There are diagrams commenting ThawBase::new_stack_frame() that show this.

It's not easy to see it in the code though. Note that the frame size is calculated relative to hf's unextended_sp which includes frame::metadata_words which is the size of java_abi.

---

Progress

• Change must be properly reviewed (1 review required, with at least 1 Reviewer)
• Change must not contain extraneous whitespace
• Commit message must refer to an issue

Issue

• JDK-8360219: [AIX] assert(locals_base >= l2) failed: bad placement (Bug - P4)

Reviewers

• Dean Long (@dean-long - Reviewer)

Reviewing

Using git

Checkout this PR locally:
$ git fetch https://git.openjdk.org/jdk.git pull/26643/head:pull/26643
$ git checkout pull/26643

Update a local copy of the PR:
$ git checkout pull/26643
$ git pull https://git.openjdk.org/jdk.git pull/26643/head

Using Skara CLI tools

Checkout this PR locally:
$ git pr checkout 26643

View PR using the GUI difftool:
$ git pr show -t 26643

Using diff file

Download this PR as a diff file:
https://git.openjdk.org/jdk/pull/26643.diff

Using Webrev

Link to Webrev Comment

[bridgekeeper]

👋 Welcome back rrich! 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]

@reinrich 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:

8360219: [AIX] assert(locals_base >= l2) failed: bad placement

Reviewed-by: dlong

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 17 new commits pushed to the master branch:

• c56fb0b: 8364503: gc/g1/TestCodeCacheUnloadDuringConcCycle.java fails because of race printing to stdout
• 487cc3c: 8359690: New test TestCPUTimeSampleThrottling still fails intermittently
• 078d0d4: 8364235: Fix for JDK-8361447 breaks the alignment requirements for GuardedMemory
• ... and 14 more: https://git.openjdk.org/jdk/compare/d25b9befe0a462b9785502806ad14e0a5f6b4320...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.

➡️ To integrate this PR with the above commit message to the master branch, type /integrate in a new comment.

[openjdk]

@reinrich The following label will be automatically applied to this pull request:

• hotspot

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 (dbff1108)

[dean-long]

It seems like relaxing the assert would allow us to silently overwrite part of a frame::top_ijava_frame_abi, which is only harmless if we are guaranteed to always treat that area as the smaller "parent" ABI past this point. Is there any way to determine if the ABI frame flavor is "top" or "parent" without adding something like a new "abi_frame_type" slot?

[reinrich]

It seems like relaxing the assert would allow us to silently overwrite part of a frame::top_ijava_frame_abi, which is only harmless if we are guaranteed to always treat that area as the smaller "parent" ABI past this point.

It is harmless. The interpreted and nmethod calling conventions only use the smaller frame::java_abi (alias for parent_ijava_frame_abi). Calling the VM or other native code requires the full ABI.

Is there any way to determine if the ABI frame flavor is "top" or "parent" without adding something like a new "abi_frame_type" slot?

I would not be aware of any way to determine the ABI type easily. I'm sure many people were looking for it since bugs caused by using the wrong type are painful. Even with a dedicated slot in dbg builds it would be hard to keep the correct type since there are many places where frames get resized.

[dean-long]

Thanks @reinrich . It looks good.