8271589: fatal error with variable shift count integer rotate operation. #4956

jatin-bhateja · 2021-08-02T12:48:10Z

Problem seen on targets which do not support variable vector shifts, in such a case vector rotation node inferred by auto-vectorizer are not disintegrable into LeftShift/RightShift and Or operations.

Progress

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

Issue

JDK-8271589: fatal error with variable shift count integer rotate operation.

Reviewers

Vladimir Kozlov (@vnkozlov - Reviewer) ⚠️ Review applies to ef3316b
Sandhya Viswanathan (@sviswa7 - Reviewer)

Reviewing

Using git

Checkout this PR locally:
$ git fetch https://git.openjdk.java.net/jdk pull/4956/head:pull/4956
$ git checkout pull/4956

Update a local copy of the PR:
$ git checkout pull/4956
$ git pull https://git.openjdk.java.net/jdk pull/4956/head

Using Skara CLI tools

Checkout this PR locally:
$ git pr checkout 4956

View PR using the GUI difftool:
$ git pr show -t 4956

Using diff file

Download this PR as a diff file:
https://git.openjdk.java.net/jdk/pull/4956.diff

bridgekeeper · 2021-08-02T12:49:30Z

👋 Welcome back jbhateja! 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.

jatin-bhateja · 2021-08-02T12:50:20Z

/label hotspot-compiler-dev

openjdk · 2021-08-02T12:50:56Z

@jatin-bhateja
The hotspot-compiler label was successfully added.

mlbridge · 2021-08-02T12:54:45Z

Webrevs

vnkozlov

I assume TestLongVectRotate.java should be also updated.

And you need add cases for Long type vector shifts if they can be generated.

vnkozlov · 2021-08-02T16:17:10Z

src/hotspot/share/opto/superword.cpp

@@ -2488,7 +2488,6 @@ void SuperWord::output() {
      } else if (VectorNode::is_scalar_rotate(n)) {
        Node* in1 = low_adr->in(1);
        Node* in2 = p->at(0)->in(2);
-        assert(in2->bottom_type()->isa_int(), "Shift must always be an int value");


Why you removed the assert? What other types can be seen here?
The following code (specifically in2->get_int()) will fail if it is not Integer type.

At this point in code, the shift is either an int value or a vector of an int value. If it's a vector !in2->is_Con() will make the condition true and no need to check for get_int due to short-circuiting.

Okay. IT should be fine because get_int() has own assert.

vnkozlov · 2021-08-02T16:17:31Z

src/hotspot/share/opto/vectornode.cpp

-  } else {
-    // Variable shift case.
+  } else if (VectorNode::is_invariant_vector(cnt)) {
+    // Scalar variable shift, handle replicates generated by auto vectorizer.
    assert(VectorNode::is_invariant_vector(cnt), "Broadcast expected");


This assert is not needed anymore since you already checked for this case.

Will remove this before checking in the final version of the patch.

vnkozlov · 2021-08-02T16:23:33Z

src/hotspot/share/opto/vectornode.cpp

+         cnt = cnt->in(1);
+      } else {
+         assert(cnt->bottom_type()->isa_long() &&
+                cnt->bottom_type()->is_long()->is_con(), "Long constant expected");


Is this case from Vector API? I think in Java we can have only Int type for shifts.

Is this case from Vector API? I think in Java we can have only Int type for shifts.

This is to handle long rotates, SLP will insert a ConvI2L before broadcasting shift value into a long vector. In case of constant int value, ConvI2L::Value will create a constant of TypeLong and which is what is being handled here.

jatin-bhateja · 2021-08-02T18:17:48Z

I assume TestLongVectRotate.java should be also updated.

And you need add cases for Long type vector shifts if they can be generated.

Long rotates patterns with variable shifts i.e. Long.rotateLeft/Right(a[i], int_b[i]/ (int) long_b[i]) are not auto-vectorized currently.
Problem is only seen with rotation over integers. I have kept the changes minimal as suggested earlier.

jatin-bhateja · 2021-08-02T18:22:43Z

Hi @vnkozlov, I have responded to your individual comments, please let me know if there are more comments.

vnkozlov · 2021-08-02T19:39:39Z

I assume TestLongVectRotate.java should be also updated.
And you need add cases for Long type vector shifts if they can be generated.

Long rotates patterns with variable shifts i.e. Long.rotateLeft/Right(a[i], int_b[i]/ (int) long_b[i]) are not auto-vectorized currently.
Problem is only seen with rotation over integers. I have kept the changes minimal as suggested earlier.

You said in an other comment:
"This is to handle long rotates, SLP will insert a ConvI2L before broadcasting shift value into a long vector."

Is it really a big change for test to cover this case? This is new code in your changes which needs test coverage I think.

I will start testing current changes (with removed assert(is_invariant_vector()) we agreed on).

vnkozlov · 2021-08-02T23:02:14Z

Testing passed clean.

jatin-bhateja · 2021-08-03T01:52:43Z

I assume TestLongVectRotate.java should be also updated.
And you need add cases for Long type vector shifts if they can be generated.

Long rotates patterns with variable shifts i.e. Long.rotateLeft/Right(a[i], int_b[i]/ (int) long_b[i]) are not auto-vectorized currently.
Problem is only seen with rotation over integers. I have kept the changes minimal as suggested earlier.

You said in an other comment:
"This is to handle long rotates, SLP will insert a ConvI2L before broadcasting shift value into a long vector."

Yes this comment is valid for Long.rotateLeft/Right(a[i], SHIFT/IMM_SHIFT) patterns.

Is it really a big change for test to cover this case? This is new code in your changes which needs test coverage I think.

I will start testing current changes (with removed assert(is_invariant_vector()) we agreed on).

vnkozlov · 2021-08-03T02:40:35Z

@jatin-bhateja I don't understand why you don't want to add testing for Long.rotateLeft/Right(a[i], SHIFT/IMM_SHIFT) .
Yes, this bug failed only for Int vectors. But you added code for Longs which have to be verified.

And you need second review.

jatin-bhateja · 2021-08-03T05:00:12Z

@jatin-bhateja I don't understand why you don't want to add testing for Long.rotateLeft/Right(a[i], SHIFT/IMM_SHIFT) .
Yes, this bug failed only for Int vectors. But you added code for Longs which have to be verified.

And you need second review.

Hi @vnkozlov ,
My apologies for not mentioning earlier that such test patterns (Long.rotateLeft/Right(a[i], scalar_shift/imm_shift) are already present in jdk/test/hotspot/jtreg/compiler/c2/cr6340864/TestLongVectRotate.java

openjdk · 2021-08-03T05:38:47Z

@jatin-bhateja 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:

8271589: fatal error with variable shift count integer rotate operation.

Reviewed-by: kvn, sviswanathan

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 44 new commits pushed to the master branch:

452f7d7: 8271217: Fix race between G1PeriodicGCTask checks and GC request
221e4b9: 8270797: ShortECDSA.java test is not complete
0a27f26: 8265057: G1: Investigate removal of maintenance of two BOT thresholds
eec64f5: 8256844: Make NMT late-initializable
4df1bc4: 6350025: API documentation for JOptionPane using deprecated methods.
efcdcc7: 8270893: IndexOutOfBoundsException while reading large TIFF file
977b8c4: 8271836: runtime/ErrorHandling/ClassPathEnvVar.java fails with release VMs
04134fc: 8264543: Cross modify fence optimization for x86
9e76909: 8271824: mark hotspot runtime/CompressedOops tests which ignore external VM flags
e49b7d9: 8271828: mark hotspot runtime/classFileParserBug tests which ignore external VM flags
... and 34 more: https://git.openjdk.java.net/jdk/compare/7cc1eb3e571e00f1cbfd62eb843df96ba8e88199...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.

sviswa7 · 2021-08-03T17:20:36Z

src/hotspot/share/opto/vectornode.cpp

    }
+    shiftRCnt = cnt;


This seems redundant. shiftRCnt is being overwritten in the very next statement.

Other than this redundant statement, the patch looks good to me.

jatin-bhateja · 2021-08-04T17:48:16Z

/integrate

openjdk · 2021-08-04T17:49:27Z

Going to push as commit 392fcc9.
Since your change was applied there have been 45 commits pushed to the master branch:

9f1edaf: 8271599: Javadoc of floorDiv() and floorMod() families is inaccurate in some places
452f7d7: 8271217: Fix race between G1PeriodicGCTask checks and GC request
221e4b9: 8270797: ShortECDSA.java test is not complete
0a27f26: 8265057: G1: Investigate removal of maintenance of two BOT thresholds
eec64f5: 8256844: Make NMT late-initializable
4df1bc4: 6350025: API documentation for JOptionPane using deprecated methods.
efcdcc7: 8270893: IndexOutOfBoundsException while reading large TIFF file
977b8c4: 8271836: runtime/ErrorHandling/ClassPathEnvVar.java fails with release VMs
04134fc: 8264543: Cross modify fence optimization for x86
9e76909: 8271824: mark hotspot runtime/CompressedOops tests which ignore external VM flags
... and 35 more: https://git.openjdk.java.net/jdk/compare/7cc1eb3e571e00f1cbfd62eb843df96ba8e88199...master

Your commit was automatically rebased without conflicts.

openjdk · 2021-08-04T17:49:52Z

@jatin-bhateja Pushed as commit 392fcc9.

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

8271589: fatal error with variable shift count integer rotate operation.

6243cf9

openjdk bot added rfr Pull request is ready for review hotspot-compiler hotspot-compiler-dev@openjdk.org labels Aug 2, 2021

8271589: Removing redundant assert.

ef3316b

vnkozlov suggested changes Aug 2, 2021

View reviewed changes

vnkozlov approved these changes Aug 3, 2021

View reviewed changes

openjdk bot added the ready Pull request is ready to be integrated label Aug 3, 2021

sviswa7 reviewed Aug 3, 2021

View reviewed changes

8271589: Review comment resolution.

aacc06e

sviswa7 approved these changes Aug 4, 2021

View reviewed changes

openjdk bot closed this Aug 4, 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 Aug 4, 2021

jatin-bhateja deleted the JDK-8271589 branch December 6, 2021 04:34

8271589: fatal error with variable shift count integer rotate operation. #4956

8271589: fatal error with variable shift count integer rotate operation. #4956

Uh oh!

Conversation

jatin-bhateja commented Aug 2, 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 Aug 2, 2021

Uh oh!

jatin-bhateja commented Aug 2, 2021

Uh oh!

openjdk bot commented Aug 2, 2021

Uh oh!

mlbridge bot commented Aug 2, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Webrevs

Uh oh!

vnkozlov left a comment

Choose a reason for hiding this comment

Uh oh!

vnkozlov Aug 2, 2021

Choose a reason for hiding this comment

Uh oh!

jatin-bhateja Aug 2, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

vnkozlov Aug 2, 2021

Choose a reason for hiding this comment

Uh oh!

vnkozlov Aug 2, 2021

Choose a reason for hiding this comment

Uh oh!

jatin-bhateja Aug 2, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

vnkozlov Aug 2, 2021

Choose a reason for hiding this comment

Uh oh!

jatin-bhateja Aug 2, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

vnkozlov Aug 2, 2021

Choose a reason for hiding this comment

Uh oh!

jatin-bhateja commented Aug 2, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

jatin-bhateja commented Aug 2, 2021

Uh oh!

vnkozlov commented Aug 2, 2021

Uh oh!

vnkozlov commented Aug 2, 2021

Uh oh!

jatin-bhateja commented Aug 3, 2021

Uh oh!

vnkozlov commented Aug 3, 2021

Uh oh!

jatin-bhateja commented Aug 3, 2021

Uh oh!

openjdk bot commented Aug 3, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

sviswa7 Aug 3, 2021

Choose a reason for hiding this comment

Uh oh!

sviswa7 Aug 3, 2021

Choose a reason for hiding this comment

Uh oh!

jatin-bhateja commented Aug 4, 2021

Uh oh!

openjdk bot commented Aug 4, 2021

Uh oh!

openjdk bot commented Aug 4, 2021

jatin-bhateja commented Aug 2, 2021 •

edited by openjdk bot

Loading

mlbridge bot commented Aug 2, 2021 •

edited

Loading

jatin-bhateja Aug 2, 2021 •

edited

Loading

jatin-bhateja Aug 2, 2021 •

edited

Loading

jatin-bhateja Aug 2, 2021 •

edited

Loading

jatin-bhateja commented Aug 2, 2021 •

edited

Loading

openjdk bot commented Aug 3, 2021 •

edited

Loading