8283935: Parallel: Crash during pretouch after large pages allocation failure #8090

tschatzl · 2022-04-04T11:07:09Z

Hi all,

can I have reviews for this change that fixes a crash with Parallel GC, AlwaysPretouch and misconfigured large pages?

The AlwaysPreTouch code for Parallel GC does not use the actual used large page size for the heap, but the one passed in by the user. If that page size is not actually available, and the heap is not aligned to that used page size, there will be an out-of-bounds access by the pretouch code.

The change simply makes the code use the correct page size (which is passed in as alignment into MutableSpace - but that is another issue).

Testing: local testing of failing test case (see CR), tier1-3

Thanks,
Thomas

Progress

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

Issue

JDK-8283935: Parallel: Crash during pretouch after large pages allocation failure

Reviewers

Kim Barrett (@kimbarrett - Reviewer)
Albert Mingkun Yang (@albertnetymk - Reviewer)

Reviewing

Using git

Checkout this PR locally:
$ git fetch https://git.openjdk.java.net/jdk pull/8090/head:pull/8090
$ git checkout pull/8090

Update a local copy of the PR:
$ git checkout pull/8090
$ git pull https://git.openjdk.java.net/jdk pull/8090/head

Using Skara CLI tools

Checkout this PR locally:
$ git pr checkout 8090

View PR using the GUI difftool:
$ git pr show -t 8090

Using diff file

Download this PR as a diff file:
https://git.openjdk.java.net/jdk/pull/8090.diff

bridgekeeper · 2022-04-04T11:07:44Z

👋 Welcome back tschatzl! 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 · 2022-04-04T11:09:34Z

@tschatzl 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 · 2022-04-04T11:15:04Z

Webrevs

src/hotspot/share/gc/parallel/mutableSpace.hpp

kimbarrett · 2022-04-04T20:08:38Z

src/hotspot/share/gc/parallel/mutableSpace.cpp

    }
    assert(mr.contains(head) && mr.contains(tail), "Sanity");

+    size_t page_size = UseLargePages ? alignment() : os::vm_page_size();


I think just using alignment should be sufficient here, and there's no need to conditionalize on UseLargePages. But maybe alignment should be renamed to page_size?

I think just using alignment should be sufficient here

I did not want to change too much, but yes, I'll try that.

But maybe alignment should be renamed to page_size?

I wanted to keep the patch small, but I'll look into this too.

I think it is actively wrong to be checking UseLargePages here. That is done when the space is reserved, and the reservation succeeds or fails, with the actual page size determined by that. Once we're here the decision of whether to use large pages is over with, and might have been in the negative, but without changing the flag.

Changed. I agree.

I was too quick changing the setter name from alignment() to page_size() - the subclass MutableNUMASpace already contains such a method, and that has been another reason to not do this. Please let me undo and look at this again.

I pushed a change now that just fixes the issues, without renaming alignment(). I would like to keep this out of this change because MutableNUMASpace has its own page_size() already, which it may modify (and reallocate everything apparently)

Bleh! It seems like there's a bunch of confused / confusing code in MutableNUMASpace. It looks like page_size() always equals alignment() and alignment() is set at construction time and never changes. There are a number of places in MutableNUMASpace that are checking for changes in the page size that seemingly can never happen. And there are places in MutableNUMASpace that are checking UseLargePages inappropriately for the same reasons as discussed above. And even when the value is accurate, some of those places look (at first glance) like they are doing unnecessary extra work.

So I'm going to call your changes good, in so far as they are fixing the crashing bug. But there really should be some followup in this area.

This reverts commit b75a785.

kimbarrett

Change looks good. There needs to be some followup work.

openjdk · 2022-04-05T15:18:32Z

@tschatzl 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:

8283935: Parallel: Crash during pretouch after large pages allocation failure

Reviewed-by: kbarrett, ayang

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 26 new commits pushed to the master branch:

0a67d68: 8284294: Create an automated regression test for RFE 4138746
955d61d: 8284369: TestFailedAllocationBadGraph fails with -XX:TieredStopAtLevel < 4
4ffe96a: 8282506: Clean up remaining references to TwoStacksPlain*SocketImpl
741be46: 8183390: Fix and re-enable post loop vectorization
500f9a5: 8283396: Null pointer dereference in loopnode.cpp:2851
e9e3aa7: 8283730: Improve discussion of modeling of packages and modules
20acea4: 8284362: Remove the "unsupported API" warning from jdk.javadoc
ff1daa8: 8284380: ProblemList jdk/jshell/HighlightUITest.java on more platforms
526e734: 8282819: Deprecate Locale class constructors
648efd7: 8284275: Remove unused sun.nio.fs.Reflect
... and 16 more: https://git.openjdk.java.net/jdk/compare/1012d59e64327817b9a2e01b94c1ad1b962c070f...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.

tschatzl · 2022-04-06T08:00:06Z

Thanks @albertnetymk @kimbarrett for your reviews.

/integrate

openjdk · 2022-04-06T08:02:00Z

Going to push as commit b56df28.
Since your change was applied there have been 26 commits pushed to the master branch:

0a67d68: 8284294: Create an automated regression test for RFE 4138746
955d61d: 8284369: TestFailedAllocationBadGraph fails with -XX:TieredStopAtLevel < 4
4ffe96a: 8282506: Clean up remaining references to TwoStacksPlain*SocketImpl
741be46: 8183390: Fix and re-enable post loop vectorization
500f9a5: 8283396: Null pointer dereference in loopnode.cpp:2851
e9e3aa7: 8283730: Improve discussion of modeling of packages and modules
20acea4: 8284362: Remove the "unsupported API" warning from jdk.javadoc
ff1daa8: 8284380: ProblemList jdk/jshell/HighlightUITest.java on more platforms
526e734: 8282819: Deprecate Locale class constructors
648efd7: 8284275: Remove unused sun.nio.fs.Reflect
... and 16 more: https://git.openjdk.java.net/jdk/compare/1012d59e64327817b9a2e01b94c1ad1b962c070f...master

Your commit was automatically rebased without conflicts.

openjdk · 2022-04-06T08:02:15Z

@tschatzl Pushed as commit b56df28.

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

Initial version

5a6b1f1

openjdk bot changed the title ~~8283935~~ 8283935: Parallel: Crash during pretouch after large pages allocation failure Apr 4, 2022

openjdk bot added the hotspot-gc hotspot-gc-dev@openjdk.org label Apr 4, 2022

tschatzl marked this pull request as ready for review April 4, 2022 11:10

openjdk bot added the rfr Pull request is ready for review label Apr 4, 2022

kimbarrett suggested changes Apr 4, 2022

View reviewed changes

tschatzl added 3 commits April 5, 2022 11:45

kbarrett review

b75a785

Revert "kbarrett review"

a3bb7dc

This reverts commit b75a785.

kbarrett review - remove unused introduced method, page size calculation

0195027

kimbarrett approved these changes Apr 5, 2022

View reviewed changes

openjdk bot added the ready Pull request is ready to be integrated label Apr 5, 2022

albertnetymk approved these changes Apr 6, 2022

View reviewed changes

openjdk bot added the integrated Pull request has been integrated label Apr 6, 2022

openjdk bot closed this Apr 6, 2022

openjdk bot removed ready Pull request is ready to be integrated rfr Pull request is ready for review labels Apr 6, 2022

tschatzl deleted the submit/8283935-parallel-gc-wrong-page-size-for-pretouch branch April 6, 2022 10:11

sandlerwang mentioned this pull request Jan 30, 2024

8324817: Parallel GC does not pre-touch all heap pages when AlwaysPreTouch enabled and large page disabled #17610

Closed

3 tasks

8283935: Parallel: Crash during pretouch after large pages allocation failure #8090

8283935: Parallel: Crash during pretouch after large pages allocation failure #8090

Uh oh!

Conversation

tschatzl commented Apr 4, 2022 • edited by openjdk bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Progress

Issue

Reviewers

Reviewing

Uh oh!

bridgekeeper bot commented Apr 4, 2022

Uh oh!

openjdk bot commented Apr 4, 2022

Uh oh!

mlbridge bot commented Apr 4, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Webrevs

Uh oh!

Uh oh!

kimbarrett Apr 4, 2022

Choose a reason for hiding this comment

Uh oh!

tschatzl Apr 5, 2022

Choose a reason for hiding this comment

Uh oh!

kimbarrett Apr 5, 2022

Choose a reason for hiding this comment

Uh oh!

tschatzl Apr 5, 2022

Choose a reason for hiding this comment

Uh oh!

tschatzl Apr 5, 2022

Choose a reason for hiding this comment

Uh oh!

tschatzl Apr 5, 2022

Choose a reason for hiding this comment

Uh oh!

kimbarrett Apr 5, 2022

Choose a reason for hiding this comment

Uh oh!

kimbarrett left a comment

Choose a reason for hiding this comment

Uh oh!

openjdk bot commented Apr 5, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

tschatzl commented Apr 6, 2022

Uh oh!

openjdk bot commented Apr 6, 2022

Uh oh!

openjdk bot commented Apr 6, 2022

Uh oh!

Reviewers

Assignees

Labels

Milestone

Development

Uh oh!

3 participants

tschatzl commented Apr 4, 2022 •

edited by openjdk bot

Loading

mlbridge bot commented Apr 4, 2022 •

edited

Loading

openjdk bot commented Apr 5, 2022 •

edited

Loading