Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

8317507: C2 compilation fails with "Exceeded _node_regs array" #2039

Closed

Conversation

shipilev
Copy link
Member

@shipilev shipilev commented Dec 11, 2023

Clean backport to fix the C2 corner case.

Additional testing:

  • New regression test fails without the fix, passes with it
  • Linux x86_64 server fastdebug,tier{1,2,3,4}
  • Linux AArch64 server fastdebug,tier{1,2,3}

Progress

  • Change must not contain extraneous whitespace
  • Commit message must refer to an issue
  • JDK-8317507 needs maintainer approval

Issue

  • JDK-8317507: C2 compilation fails with "Exceeded _node_regs array" (Bug - P2 - Approved)

Reviewing

Using git

Checkout this PR locally:
$ git fetch https://git.openjdk.org/jdk17u-dev.git pull/2039/head:pull/2039
$ git checkout pull/2039

Update a local copy of the PR:
$ git checkout pull/2039
$ git pull https://git.openjdk.org/jdk17u-dev.git pull/2039/head

Using Skara CLI tools

Checkout this PR locally:
$ git pr checkout 2039

View PR using the GUI difftool:
$ git pr show -t 2039

Using diff file

Download this PR as a diff file:
https://git.openjdk.org/jdk17u-dev/pull/2039.diff

Webrev

Link to Webrev Comment

@bridgekeeper
Copy link

bridgekeeper bot commented Dec 11, 2023

👋 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 openjdk bot changed the title Backport a5818972c16bd883d768ff2fb23a8aa9e0142c65 8317507: C2 compilation fails with "Exceeded _node_regs array" Dec 11, 2023
@openjdk
Copy link

openjdk bot commented Dec 11, 2023

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

@shipilev shipilev marked this pull request as ready for review December 12, 2023 10:18
@shipilev
Copy link
Member Author

/approval request Clean backport to fix the C2 corner case. Applies cleanly. Tests pass.

@openjdk
Copy link

openjdk bot commented Dec 12, 2023

⚠️ @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.

@openjdk openjdk bot added the rfr Pull request is ready for review label Dec 12, 2023
@openjdk
Copy link

openjdk bot commented Dec 12, 2023

@shipilev
8317507: The approval request has been created successfully.

@openjdk openjdk bot added the approval label Dec 12, 2023
@mlbridge
Copy link

mlbridge bot commented Dec 12, 2023

Webrevs

@GoeLin
Copy link
Member

GoeLin commented Dec 13, 2023

Won't this increase the memory consumption of the compiler significantly? As I understand it can push more nodes on the arenas now.

@shipilev
Copy link
Member Author

Won't this increase the memory consumption of the compiler significantly? As I understand it can push more nodes on the arenas now.

The footprint increase effect might be in that we would be able to carry more nodes in these degenerate situations before bailing. Currently it is capped by PhaseRegAlloc::alloc_node_regs, which is a small number: original report mentions 1118 nodes for the test case in question. But AFAIU, it is basically a choice what do you do if compiler/user over-unrolled: before this patch, C2 would eventually just bail; after this patch, C2 would still compile the thing.

@GoeLin
Copy link
Member

GoeLin commented Dec 13, 2023

Yes, it will now compile the method. But it might need twice as many memory as before. This might kill existing installations. Compiler memory is often a bottle neck.

@shipilev
Copy link
Member Author

Yeah, accepting more nodes is a disadvantage of the fix. But if we look closely at the way it fails in fastdebug for the given test cases, I think assert catches the OOB access in PhaseCFG::insert_goto_at calling to PhaseRegAlloc::set_bad, which suggests that we are actually dealing with memory stomp, not a clean bail. This explains why bug is P2. So, on balance, accepting more compiler footprint to avoid compiler memory stomp is good?

@openjdk
Copy link

openjdk bot commented Dec 14, 2023

@shipilev 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:

8317507: C2 compilation fails with "Exceeded _node_regs array"

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 11 new commits pushed to the master branch:

  • 78839e9: 8054022: HttpURLConnection timeouts with Expect: 100-Continue and no chunking
  • 6493102: 7167356: (javac) investigate failing tests in JavacParserTest
  • 71471c1: 8294254: [macOS] javax/swing/plaf/aqua/CustomComboBoxFocusTest.java failure
  • aba80d1: 8274634: Use String.equals instead of String.compareTo in java.desktop
  • 519fd89: 8295124: Atomic::add to pointer type may return wrong value
  • 5bc2302: 8271616: oddPart in MutableBigInteger::mutableModInverse contains info on final result
  • fb950fa: 8316415: Parallelize sun/security/rsa/SignedObjectChain.java subtests
  • 3a2bf8e: 8225377: type annotations are not visible to javac plugins across compilation boundaries
  • 9596656: 8271118: C2: StressGCM should have higher priority than frequency-based policy
  • f8e54a9: 8316929: Shenandoah: Shenandoah degenerated GC and full GC need to cleanup old OopMapCache entries
  • ... and 1 more: https://git.openjdk.org/jdk17u-dev/compare/c67144742b6f4777da941f5158021884b7ea7768...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 openjdk bot added ready Pull request is ready to be integrated and removed approval labels Dec 14, 2023
@shipilev
Copy link
Member Author

@GoeLin, I see jdk17u-fix-yes now. I take it as the signal that you are okay with the explanation above, and think this makes sense for jdk17u-dev? Just checking.

@GoeLin
Copy link
Member

GoeLin commented Dec 19, 2023

yes :)

@shipilev
Copy link
Member Author

All right then!

/integrate

@openjdk
Copy link

openjdk bot commented Dec 19, 2023

Going to push as commit 5ed1926.
Since your change was applied there have been 24 commits pushed to the master branch:

  • 7973bfe: 8321215: Incorrect x86 instruction encoding for VSIB addressing mode
  • c555fe3: 8315680: java/lang/ref/ReachabilityFenceTest.java should run with -Xbatch
  • 3809b71: 8293117: Add atomic bitset functions
  • a032136: 8295068: SSLEngine throws NPE parsing CertificateRequests
  • f033de0: 8274632: Possible pointer overflow in PretouchTask chunk claiming
  • c44bd2b: 8320001: javac crashes while adding type annotations to the return type of a constructor
  • 97d8ed2: 8309733: [macOS, Accessibility] VoiceOver: Incorrect announcements of JRadioButton
  • 48c8478: 8320798: Console read line with zero out should zero out underlying buffer
  • 5653d2d: 8320597: RSA signature verification fails on signed data that does not encode params correctly
  • dcba538: 8283400: [macos] a11y : Screen magnifier does not reflect JRadioButton value change
  • ... and 14 more: https://git.openjdk.org/jdk17u-dev/compare/c67144742b6f4777da941f5158021884b7ea7768...master

Your commit was automatically rebased without conflicts.

@openjdk openjdk bot added the integrated Pull request has been integrated label Dec 19, 2023
@openjdk openjdk bot closed this Dec 19, 2023
@openjdk openjdk bot removed ready Pull request is ready to be integrated rfr Pull request is ready for review labels Dec 19, 2023
@openjdk
Copy link

openjdk bot commented Dec 19, 2023

@shipilev Pushed as commit 5ed1926.

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

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
backport clean integrated Pull request has been integrated
2 participants