8263303: C2 compilation fails with assert(found_sfpt) failed: no node in loop that's not input to safepoint #4278

rwestrel · 2021-06-01T08:30:48Z

The loop strip mining verification code catches a node which is pinned
in the outer strip mined loop but not referenced from the safepoint
node.

The test case is an example of how this could happen. The array[i]
load is referenced from the safepoint initially through:

(j - 10) * array[i]

It is sunk out of the inner loop and pinned in the outer loop. A
following loop opts round causes j - 10 to constant fold to 0 and as a
consequence the array load is no longer referenced from the safepoint
but pinned in the outer strip mined loop.

I tried to find a way to preserve the invariant that all nodes in the
outer strip mined loop are referenced from the safepoint but found no
robust way for that. So the fix removes that part of the verification
code.

This requires loop cloning code to be adjusted for nodes pinned in the
outer strip mined loop but not referenced from the safepoint.

Progress

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

Issue

JDK-8263303: C2 compilation fails with assert(found_sfpt) failed: no node in loop that's not input to safepoint

Reviewers

Christian Hagedorn (@chhagedorn - Reviewer)
Vladimir Kozlov (@vnkozlov - Reviewer)

Reviewing

Using git

Checkout this PR locally:
$ git fetch https://git.openjdk.java.net/jdk pull/4278/head:pull/4278
$ git checkout pull/4278

Update a local copy of the PR:
$ git checkout pull/4278
$ git pull https://git.openjdk.java.net/jdk pull/4278/head

Using Skara CLI tools

Checkout this PR locally:
$ git pr checkout 4278

View PR using the GUI difftool:
$ git pr show -t 4278

Using diff file

Download this PR as a diff file:
https://git.openjdk.java.net/jdk/pull/4278.diff

bridgekeeper · 2021-06-01T08:31:50Z

👋 Welcome back roland! 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 · 2021-06-01T08:33:55Z

@rwestrel 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 · 2021-06-01T08:38:36Z

Webrevs

01: Full - Incremental (474d374b)
00: Full (92264e39)

chhagedorn

That's reasonable. Looks good to me!

chhagedorn · 2021-06-09T07:50:16Z

test/hotspot/jtreg/compiler/loopstripmining/TestPinnedUseInOuterLSMUnusedBySfpt.java

+// rr record ~/jdk-jdk/build/linux-x86_64-server-fastdebug/images/jdk/bin/java -XX:-TieredCompilation -XX:-BackgroundCompilation -XX:-UseOnStackReplacement -XX:+PrintCompilation -XX:CompileOnly=TestPinnedUseInOuterLSM::test -XX:CompileCommand=quiet -XX:PrintIdealGraphFile=graph.xml -XX:PrintIdealGraphLevel=2 -XX:+UseG1GC -XX:+TraceLoopOpts -XX:LoopUnrollLimit=0 TestPinnedUseInOuterLSM
+


Probably a leftover, can be removed

openjdk · 2021-06-09T08:48:56Z

@rwestrel 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:

8263303: C2 compilation fails with assert(found_sfpt) failed: no node in loop that's not input to safepoint

Reviewed-by: chagedorn, kvn

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 175 new commits pushed to the master branch:

bb3d226: 8238213: Method resolution should stop on static error
81fdeb5: 8268417: Add test from JDK-8268360
caf7f49: 8268122: Add specific gc cause for G1 full collections
43e38a1: 8268377: Windows 32bit build fails after JDK-8268174
5fbb62c: 8268163: Change the order of fallback full GCs in G1
7b1e402: 8266598: Exception values for AnnotationTypeMismatchException are not always informative
13d6180: 8264859: Implement Context-Specific Deserialization Filters
dd34a4c: 8268372: ZGC: dynamically select the number of concurrent GC threads used
4388959: 8268056: Update java.net and java.nio to use switch expressions
9cfd560: 8267663: [vector] Add unsigned comparison operators on AArch64
... and 165 more: https://git.openjdk.java.net/jdk/compare/ae2f37f868bfdcb3d46098e91ed537fb199d7dbe...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.

rwestrel · 2021-06-09T09:06:12Z

@chhagedorn thanks for the review.

vnkozlov · 2021-06-09T15:32:44Z

Yes, we can't assume that all node in outer loop are referenced by Safepoint node. SP nodes reference only live values in current BC. In the test case it is value 'v'. After constant fold `v' become 0 and that is what SF should see (MulNode::Value() optimizes multiplication by 0). array[i] load should be dead at this point since nothing should be referencing it (except GC barriers, may be).

And I don't get how array[i] is moved from inner loop - it depends on i - loop's index.

rwestrel · 2021-06-09T16:02:39Z

Thanks for looking at this @vnkozlov

Yes, we can't assume that all node in outer loop are referenced by Safepoint node. SP nodes reference only live values in current BC. In the test case it is value 'v'. After constant fold `v' become 0 and that is what SF should see (MulNode::Value() optimizes multiplication by 0). array[i] load should be dead at this point since nothing should be referencing it (except GC barriers, may be).

The test case has: return v + array[i-1] + f;
array[i-1] is after i is incremented so commons with array[i] in the loop body. That keeps the array[i] load alive.

And I don't get how array[i] is moved from inner loop - it depends on i - loop's index.

It's sunk out of loop because it only has uses out of loop. One of the uses is still indirectly the safepoint at the time sinking happens so it's pinned in the outer loop.

vnkozlov

Let me run tests before you push.

vnkozlov · 2021-06-09T16:22:01Z

Thanks for looking at this @vnkozlov

Yes, we can't assume that all node in outer loop are referenced by Safepoint node. SP nodes reference only live values in current BC. In the test case it is value 'v'. After constant fold `v' become 0 and that is what SF should see (MulNode::Value() optimizes multiplication by 0). array[i] load should be dead at this point since nothing should be referencing it (except GC barriers, may be).

The test case has: return v + array[i-1] + f;
array[i-1] is after i is incremented so commons with array[i] in the loop body. That keeps the array[i] load alive.

Got it. I assume it happens after constant folding j - 10 to 0.

And I don't get how array[i] is moved from inner loop - it depends on i - loop's index.

It's sunk out of loop because it only has uses out of loop. One of the uses is still indirectly the safepoint at the time sinking happens so it's pinned in the outer loop.

Okay.

vnkozlov · 2021-06-09T22:41:03Z

Testing results are good.

rwestrel · 2021-06-10T06:54:19Z

@vnkozlov thanks for the review & testing

rwestrel · 2021-06-10T06:54:36Z

/integrate

openjdk · 2021-06-10T06:56:37Z

@rwestrel Since your change was applied there have been 189 commits pushed to the master branch:

dd1cbad: 8268285: vmTestbase/nsk/jvmti/GetThreadState/thrstat002 failed with "Wrong thread "thr1" (...) state after SuspendThread"
ece3ae3: 8268388: Update large pages information in Java manpage
2623b0b: 8268475: execute runtime/InvocationTests w/ -UseVtableBasedCHA
f839308: 8268407: ProblemList sun/tools/jstat/jstatLineCountsX.sh on linux-aarch64 due to JDK-8268211
58ba48b: 8268192: LambdaMetafactory with invokespecial causes VerificationError
b41f3f8: 8268478: JVMCI tests failing after JDK-8268052
7ff6e7b: 8267954: Shared classes that failed to load should not be loaded again
991ca14: 8267430: GraphicsDevice.setDisplayMode(REFRESH_RATE_UNKNOWN) throws IAE: Unable to set display mode!
bf29a01: 8228343: JCMD and attach fail to work across Linux Container boundary
408e0a9: 8255148: Confusing log output: SSLSocket duplex close failed
... and 179 more: https://git.openjdk.java.net/jdk/compare/ae2f37f868bfdcb3d46098e91ed537fb199d7dbe...master

Your commit was automatically rebased without conflicts.

Pushed as commit d4377af.

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

rwestrel added 2 commits June 1, 2021 09:58

fix

463d243

test

92264e3

openjdk bot added the rfr Pull request is ready for review label Jun 1, 2021

openjdk bot added the hotspot-compiler hotspot-compiler-dev@openjdk.org label Jun 1, 2021

chhagedorn approved these changes Jun 9, 2021

View reviewed changes

openjdk bot added the ready Pull request is ready to be integrated label Jun 9, 2021

remove useless line from test

474d374

chhagedorn approved these changes Jun 9, 2021

View reviewed changes

vnkozlov approved these changes Jun 9, 2021

View reviewed changes

openjdk bot closed this Jun 10, 2021

openjdk bot added integrated Pull request has been integrated and removed ready Pull request is ready to be integrated rfr Pull request is ready for review labels Jun 10, 2021

		// rr record ~/jdk-jdk/build/linux-x86_64-server-fastdebug/images/jdk/bin/java -XX:-TieredCompilation -XX:-BackgroundCompilation -XX:-UseOnStackReplacement -XX:+PrintCompilation -XX:CompileOnly=TestPinnedUseInOuterLSM::test -XX:CompileCommand=quiet -XX:PrintIdealGraphFile=graph.xml -XX:PrintIdealGraphLevel=2 -XX:+UseG1GC -XX:+TraceLoopOpts -XX:LoopUnrollLimit=0 TestPinnedUseInOuterLSM

8263303: C2 compilation fails with assert(found_sfpt) failed: no node in loop that's not input to safepoint #4278

8263303: C2 compilation fails with assert(found_sfpt) failed: no node in loop that's not input to safepoint #4278

Uh oh!

Conversation

rwestrel commented Jun 1, 2021 • 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 Jun 1, 2021

Uh oh!

openjdk bot commented Jun 1, 2021

Uh oh!

mlbridge bot commented Jun 1, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Webrevs

Uh oh!

chhagedorn left a comment

Choose a reason for hiding this comment

Uh oh!

chhagedorn Jun 9, 2021

Choose a reason for hiding this comment

Uh oh!

rwestrel Jun 9, 2021

Choose a reason for hiding this comment

Uh oh!

openjdk bot commented Jun 9, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

rwestrel commented Jun 9, 2021

Uh oh!

vnkozlov commented Jun 9, 2021

Uh oh!

rwestrel commented Jun 9, 2021

Uh oh!

vnkozlov left a comment

Choose a reason for hiding this comment

Uh oh!

vnkozlov commented Jun 9, 2021

Uh oh!

vnkozlov commented Jun 9, 2021

Uh oh!

rwestrel commented Jun 10, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

rwestrel commented Jun 10, 2021

Uh oh!

openjdk bot commented Jun 10, 2021

Uh oh!

Uh oh!

rwestrel commented Jun 1, 2021 •

edited by openjdk bot

Loading

mlbridge bot commented Jun 1, 2021 •

edited

Loading

openjdk bot commented Jun 9, 2021 •

edited

Loading

rwestrel commented Jun 10, 2021 •

edited

Loading