8259777: Incorrect predication condition generated by ADLC

[jatin-bhateja]

Please consider following instruction selection pattern:-

match(Set dst (VectorBlend (Binary src1 (AddVB src1 src2)) mask));

Currently, ADLC emits following predication check for src1 in matcher DFA.

/*src1*/(_kids[0]->_kids[0]->_leaf == _kids[1]->_kids[0]->_leaf)

src1 appears twice in the State tree corresponding to this pattern, first as the leaf of left child state of Binary node which is
left child state of VectorBlend and secondly as the leaf of left child of AddVB which is the right child state of Binary node.
Thus following is the correct predication check to test both the occurrences of src1.

/*src1*/(_kids[0]->_kids[0]->_leaf == _kids[0]->_kids[1]->_kids[0]->_leaf)

This patch does a minor re-organizes the predication generation algorithm to fix this issue.

Validated that there is no change in generated DFA for current X86 AD files with and without this patch since such instruction patterns do not exist in AD files currently.

---

Progress

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

Issue

• JDK-8259777: Incorrect predication condition generated by ADLC

Reviewers

• Vladimir Ivanov (@iwanowww - Reviewer)

Download

$ git fetch https://git.openjdk.java.net/jdk pull/2080/head:pull/2080
$ git checkout pull/2080

[bridgekeeper]

👋 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.

[openjdk]

@jatin-bhateja 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]

Webrevs

• 00: Full (b8cbd5d)

[iwanowww]

Looks good.

[openjdk]

@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:

8259777: Incorrect predication condition generated by ADLC

Reviewed-by: vlivanov

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 4 new commits pushed to the master branch:

• ff3e6e4: 8259773: Incorrect encoding of AVX-512 kmovq instruction
• b8ef2ba: 8259563: The CPU model name is printed multiple times when using -Xlog:os+cpu
• b040a3d: 8259631: Reapply pattern match instanceof use in HttpClientImpl
• 3462f7a: 8256955: Move includes of events.hpp out of header files

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.

[jatin-bhateja]

Looks good.

Hi Vladimir, Please suggest if its ok to push this patch with one reviewer approval.

[mlbridge]

Mailing list message from Vladimir Ivanov on hotspot-compiler-dev:

The guideline is to wait around a day to give people a chance speak up.
Otherwise, I don't see a problem having a single (R)eview here.

Best regards,
Vladimir Ivanov

On 14.01.2021 20:31, Jatin Bhateja wrote:

[jatin-bhateja]

/integrate

[openjdk]

@jatin-bhateja Since your change was applied there have been 28 commits pushed to the master branch:

• bbac91a: 8257959: Add gtest run with -XX:+UseLargePages
• 707bce0: 8257212: (bf spec) Clarify byte order of the buffer returned by CharBuffer.subsequence(int,int)
• 0ec2c96: 8259820: JShell does not handle -source 8 properly
• b01a15e: 8258884: [TEST_BUG] Convert applet-based test open/test/jdk/javax/swing/JMenuItem/8031573/bug8031573.java to a regular java test
• 6d4a593: 8259627: Potential memory leaks in JVMTI after JDK-8227745
• 2c8e337: 8259622: TreeMap.computeIfAbsent deviates from spec
• d701bab: Merge
• 4307fa6: 8253505: JFR: onFlush invoked out of order with a sorted event stream
• 0148adf: 8255120: C2: assert(outer->outcnt() >= phis + 2 && outer->outcnt() <= phis + 2 + stores + 1) failed: only phis
• 90960c5: 8252657: JVMTI agent is not unloaded when Agent_OnAttach is failed
• ... and 18 more: https://git.openjdk.java.net/jdk/compare/8b8b1f9a37d8606848e40aede006db8129cdac39...master

Your commit was automatically rebased without conflicts.

Pushed as commit bcf20a0.

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