8309675: Generational ZGC: compiler/gcbarriers/UnsafeIntrinsicsTest.java fails in nmt_commit

[stefank]

The failure happens because we have extra verification code that is incorrect to use when calling the NMT tracking code.

The fix is to simply inline the address calculation code without the verification code.

This fixes a tier3 failure. I'm going to run this through more extensive testing but it would be good to get this reviewed.

---

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-8309675: Generational ZGC: compiler/gcbarriers/UnsafeIntrinsicsTest.java fails in nmt_commit (Bug - P2)

Reviewers

• Axel Boldt-Christmas (@xmas92 - Committer)
• Daniel D. Daugherty (@dcubed-ojdk - Reviewer)

Reviewing

Using git

Checkout this PR locally:
$ git fetch https://git.openjdk.org/jdk.git pull/14382/head:pull/14382
$ git checkout pull/14382

Update a local copy of the PR:
$ git checkout pull/14382
$ git pull https://git.openjdk.org/jdk.git pull/14382/head

Using Skara CLI tools

Checkout this PR locally:
$ git pr checkout 14382

View PR using the GUI difftool:
$ git pr show -t 14382

Using diff file

Download this PR as a diff file:
https://git.openjdk.org/jdk/pull/14382.diff

Webrev

Link to Webrev Comment

[bridgekeeper]

👋 Welcome back stefank! 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]

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

• hotspot-gc

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 (d9ff5f98)

[stefank]

The reason why this only happens on this test is that -XX:+ZVerifyOops is only turned on in this test.

[stefank]

To understand the fix see the function I'm replacing:

inline zaddress ZOffset::address(zoffset offset) {
  return to_zaddress(untype(offset) | ZAddressHeapBase);
}

and the verification code in this function:

inline zaddress to_zaddress(uintptr_t value) {
  const zaddress addr = zaddress(value);
  assert_is_valid(addr);
  DEBUG_ONLY(dereferenceable_test(addr));
  return addr;
}

The crash happens because of the dereferenceable_test, which checks that the memory have been committed and is accessible. As described in nmt_commit, we fake the address we report to NMT, so we can't try to dereference the contents of that address.

[xmas92]

Guess this hit the dereference test.
Could probably use zaddress_unsafe as well?
Either solution works.

[stefank]

Yes, that's a good suggestion, though I think I'd prefer to stick with the current solution because then it is more clear that the code matches the comment in nmt_commit.

Thanks for reviewing!

[dcubed-ojdk]

Thumbs up. Yet another disconnect between NMT virtual memory
modeling versus Z virtual memory modeling?

[dcubed-ojdk]

Did you run the revised test on all Tier3 platforms?

[openjdk]

@stefank 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:

8309675: Generational ZGC: compiler/gcbarriers/UnsafeIntrinsicsTest.java fails in nmt_commit

Reviewed-by: aboldtch, dcubed

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 2 new commits pushed to the master branch:

• 34f0a6e: 8305593: Add @SPEC tags in java.desktop
• 0be3905: 8309595: Allow javadoc to process unnamed classes

Please see this link for an up-to-date comparison between the source branch of this pull request and the master branch.
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.

[stefank]

Thumbs up. Yet another disconnect between NMT virtual memory modeling versus Z virtual memory modeling?

Thanks for reviewing. Yes, in away. It's the workaround for this disconnect that triggered our verification code, which in turn found our workaround with the fake addresses. :)

[stefank]

Did you run the revised test on all Tier3 platforms?

I'm running tier1-7 with Generational ZGC. There's just a few more tests running, so I'm on the verge of integrating this.

[stefank]

/integrate

[openjdk]

Going to push as commit c4e6542.
Since your change was applied there have been 2 commits pushed to the master branch:

• 34f0a6e: 8305593: Add @SPEC tags in java.desktop
• 0be3905: 8309595: Allow javadoc to process unnamed classes

Your commit was automatically rebased without conflicts.

[openjdk]

@stefank Pushed as commit c4e6542.

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