8344035: Replace predicate walking code in Loop Unswitching with a predicate visitor #22828

chhagedorn · 2024-12-19T13:56:41Z

This patch is a follow up to the clean-ups done with JDK-8342945 and introduces a new predicate visitor for Loop Unswitching to update the last remaining custom predicate cloning code.

This patch includes the following:

New CloneUnswitchedLoopPredicatesVisitor class which delegates the cloning work to a new ClonePredicateToTargetLoop class.
- We walk the predicate chain in the PredicateIterator and call the CloneUnswitchedLoopPredicatesVisitor for each visited predicate. Then we clone the predicate on the fly to the target loop.
New ClonePredicateToTargetLoop class:
- Clones Parse Predicates
- Clones Template Assertion Predicates
  - Includes rewiring of control dependent data nodes
- Rewires the cloned predicates to the target loop with new TargetLoopPredicateChain class:
  - Keeps track of the current chain head, which is the target loop itself when the chain is still empty.
  - Each time a new predicate is inserted at the target loop, the old predicate chain head is set as output of the new predicate.
  - An example is shown as class comment at TargetLoopPredicateChain.
  - I plan to reuse this class later again when also updating CreateAssertionPredicatesVisitor which is done when we tackle the actual still remaining Assertion Predicate bugs.
Removal of custom predicate cloning code found in PhaseIdealLoop.
Changed steps performed in Loop Unswitching from:
1. Clone loop
2. Clone predicates and insert them below the unswitched loop selector If projections
3. Connect the cloned predicates to the unswitched loops
to:
1. Clone loop
2. Connect unswitched loop selector If projections to unswitched loops such that they are now the new loop entries
3. Clone predicates and insert them between the unswitched loop selector If projections and the unswitched loops
Rename/update get_template_assertion_predicates()/TemplateAssertionPredicateCollector to reflect the only use left.

Thanks,
Christian

Progress

Change must be properly reviewed (1 review required, with at least 1 Reviewer)
Change must not contain extraneous whitespace
Commit message must refer to an issue

Issue

JDK-8344035: Replace predicate walking code in Loop Unswitching with a predicate visitor (Sub-task - P4)

Reviewers

Roland Westrelin (@rwestrel - Reviewer)
Vladimir Kozlov (@vnkozlov - Reviewer)

Reviewing

Using git

Checkout this PR locally:
$ git fetch https://git.openjdk.org/jdk.git pull/22828/head:pull/22828
$ git checkout pull/22828

Update a local copy of the PR:
$ git checkout pull/22828
$ git pull https://git.openjdk.org/jdk.git pull/22828/head

Using Skara CLI tools

Checkout this PR locally:
$ git pr checkout 22828

View PR using the GUI difftool:
$ git pr show -t 22828

Using diff file

Download this PR as a diff file:
https://git.openjdk.org/jdk/pull/22828.diff

Using Webrev

Link to Webrev Comment

…edicate visitor

bridgekeeper · 2024-12-19T13:57:12Z

👋 Welcome back chagedorn! 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 · 2024-12-19T13:57:14Z

@chhagedorn 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:

8344035: Replace predicate walking code in Loop Unswitching with a predicate visitor

Reviewed-by: roland, 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 73 new commits pushed to the master branch:

1ef77cd: 8347171: (dc) java/nio/channels/DatagramChannel/InterruptibleOrNot.java fails with virtual thread factory
72f1114: 8346705: SNI not sent with Java 22+ using java.net.http.HttpClient.Builder#sslParameters
cb9a98b: 8347141: Several javac tests compile with an unnecessary -Xlint:-path flag
f9b1133: 8346880: [aix] java/lang/ProcessHandle/InfoTest.java still fails: "reported cputime less than expected"
9c72ded: 8346036: Unnecessary Hashtable usage in javax.swing.text.html.parser.Entity
3024a73: 8345782: Refining the cases that libjsig deprecation warning is issued
2801bc6: 8346460: NotifyFramePop should return JVMTI_ERROR_DUPLICATE
ceae2b9: 8347121: Add missing @serial tags to module java.base
8c87ea2: 8346478: RISC-V: Refactor add/sub assembler routines
765b9e6: 8346706: RISC-V: Add available registers to hs_err
... and 63 more: https://git.openjdk.org/jdk/compare/f1d85ab3e61f923b4e120cf30e16109e04505b53...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 · 2024-12-19T13:57:41Z

@chhagedorn 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.

chhagedorn · 2024-12-19T13:57:07Z

src/hotspot/share/opto/loopPredicate.cpp

-                                                        const int opcode, const bool rewire_uncommon_proj_phi_inputs,
-                                                        AssertionPredicateType assertion_predicate_type) {
+                                                        const int opcode, const bool rewire_uncommon_proj_phi_inputs) {


Squeezed this in here: assertion_predicate_type is dead.

chhagedorn · 2024-12-19T13:57:27Z

src/hotspot/share/opto/loopPredicate.cpp

+void PhaseIdealLoop::get_opaque_template_assertion_predicate_nodes(ParsePredicateSuccessProj* parse_predicate_proj,
+                                                                   Unique_Node_List& list) {


Only applied a renaming here.

chhagedorn · 2024-12-19T13:57:55Z

src/hotspot/share/opto/loopUnswitch.cpp

+  void unswitch(const UnswitchedLoopSelector& unswitched_loop_selector) {
+    const uint first_false_path_loop_node_index = _phase->C->unique();
+    clone_loop(unswitched_loop_selector);
+
+    move_parse_and_template_assertion_predicates_to_unswitched_loops(unswitched_loop_selector,
+                                                                     first_false_path_loop_node_index);
+    DEBUG_ONLY(verify_unswitched_loop_versions(_loop->_head->as_Loop(), unswitched_loop_selector);)
+
+    _phase->recompute_dom_depth();
+    remove_unswitch_candidate_from_loops(unswitched_loop_selector);


New order of operation (see PR description).

chhagedorn · 2024-12-19T13:58:38Z

src/hotspot/share/opto/loopUnswitch.cpp

+    const UnswitchedLoopSelector& unswitched_loop_selector, const uint first_false_path_loop_node_index) const {
+    const NodeInOriginalLoopBody node_in_true_path_loop_body(first_false_path_loop_node_index, _old_new);
+    const NodeInClonedLoopBody node_in_false_path_loop_body(first_false_path_loop_node_index);
+    CloneUnswitchedLoopPredicatesVisitor
+    clone_unswitched_loop_predicates_visitor(_loop_head, old_to_new(_loop_head)->as_Loop(), node_in_true_path_loop_body,
+                                             node_in_false_path_loop_body, _phase);
+    Node* source_loop_entry = unswitched_loop_selector.selector()->in(0);
+    PredicateIterator predicate_iterator(source_loop_entry);
+    predicate_iterator.for_each(clone_unswitched_loop_predicates_visitor);


The main part that is new here: We use visitors to clone the predicates and rewire them correctly.

chhagedorn · 2024-12-19T13:58:45Z

src/hotspot/share/opto/loopUnswitch.cpp

+  Node* old_to_new(const Node* old) const {
+    return _old_new[old->_idx];
+  }


Only moved.

mlbridge · 2024-12-19T14:04:22Z

Webrevs

01: Full - Incremental (acc729f9)
00: Full (9cc2db29)

rwestrel

Looks good to me.

chhagedorn · 2025-01-09T14:29:42Z

Thanks Roland for your review!

vnkozlov

Looks good to me too.

chhagedorn · 2025-01-10T09:17:57Z

Thanks Vladimir for your review!

chhagedorn · 2025-01-13T08:08:41Z

/integrate

openjdk · 2025-01-13T08:09:11Z

Going to push as commit ed0b555.
Since your change was applied there have been 109 commits pushed to the master branch:

b37f123: 8347407: [BACKOUT] C1/C2 don't handle allocation failure properly during initialization (RuntimeStub::new_runtime_stub fatal crash)
1f7925c: 8347270: Remove unix_getParentPidAndTimings, unix_getChildren and unix_getCmdlineAndUserInfo
9fafd63: 8346828: javax/swing/JScrollBar/4865918/bug4865918.java still fails in CI
f04a642: 8346717: serviceability/dcmd/vm/SystemDumpMapTest.java failing on Windows with "Stack base not yet set for thread id"
3145278: 8346727: JvmtiVTMSTransitionDisabler deadlock
761774a: 8346143: add ClearAllFramePops function to speedup debugger single stepping in some cases
6f1f2f2: 8347063: Add comments in ClassFileFormatVersion for class file format evolution history
10f7142: 8347295: Fix WinResourceTest to make it work with WiX v4.0+
01c8ba2: 8347298: Bug in JPackageCommand.ignoreFakeRuntime()
d69463e: 8347300: Don't exclude the "PATH" var from the environment when running app launchers in jpackage tests
... and 99 more: https://git.openjdk.org/jdk/compare/f1d85ab3e61f923b4e120cf30e16109e04505b53...master

Your commit was automatically rebased without conflicts.

openjdk · 2025-01-13T08:09:21Z

@chhagedorn Pushed as commit ed0b555.

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

8344035: Replace predicate walking code in Loop Unswitching with a pr…

9cc2db2

…edicate visitor

openjdk bot added the hotspot-compiler hotspot-compiler-dev@openjdk.org label Dec 19, 2024

chhagedorn commented Dec 19, 2024

View reviewed changes

chhagedorn marked this pull request as ready for review December 19, 2024 14:00

openjdk bot added the rfr Pull request is ready for review label Dec 19, 2024

Merge branch 'master' into JDK-8344035

acc729f

rwestrel approved these changes Jan 9, 2025

View reviewed changes

openjdk bot added the ready Pull request is ready to be integrated label Jan 9, 2025

vnkozlov approved these changes Jan 9, 2025

View reviewed changes

openjdk bot added the integrated Pull request has been integrated label Jan 13, 2025

openjdk bot closed this Jan 13, 2025

openjdk bot removed ready Pull request is ready to be integrated rfr Pull request is ready for review labels Jan 13, 2025

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

8344035: Replace predicate walking code in Loop Unswitching with a predicate visitor #22828

8344035: Replace predicate walking code in Loop Unswitching with a predicate visitor #22828

chhagedorn commented Dec 19, 2024 •

edited by openjdk bot

Loading

bridgekeeper bot commented Dec 19, 2024

openjdk bot commented Dec 19, 2024 •

edited

Loading

openjdk bot commented Dec 19, 2024

chhagedorn Dec 19, 2024

chhagedorn Dec 19, 2024

chhagedorn Dec 19, 2024

chhagedorn Dec 19, 2024

chhagedorn Dec 19, 2024

mlbridge bot commented Dec 19, 2024 •

edited

Loading

rwestrel left a comment

chhagedorn commented Jan 9, 2025

vnkozlov left a comment

chhagedorn commented Jan 10, 2025

chhagedorn commented Jan 13, 2025

openjdk bot commented Jan 13, 2025

openjdk bot commented Jan 13, 2025

		void PhaseIdealLoop::get_opaque_template_assertion_predicate_nodes(ParsePredicateSuccessProj* parse_predicate_proj,
		Unique_Node_List& list) {

8344035: Replace predicate walking code in Loop Unswitching with a predicate visitor #22828

8344035: Replace predicate walking code in Loop Unswitching with a predicate visitor #22828

Conversation

chhagedorn commented Dec 19, 2024 • edited by openjdk bot Loading

Progress

Issue

Reviewers

Reviewing

bridgekeeper bot commented Dec 19, 2024

openjdk bot commented Dec 19, 2024 • edited Loading

openjdk bot commented Dec 19, 2024

chhagedorn Dec 19, 2024

Choose a reason for hiding this comment

chhagedorn Dec 19, 2024

Choose a reason for hiding this comment

chhagedorn Dec 19, 2024

Choose a reason for hiding this comment

chhagedorn Dec 19, 2024

Choose a reason for hiding this comment

chhagedorn Dec 19, 2024

Choose a reason for hiding this comment

mlbridge bot commented Dec 19, 2024 • edited Loading

Webrevs

rwestrel left a comment

Choose a reason for hiding this comment

chhagedorn commented Jan 9, 2025

vnkozlov left a comment

Choose a reason for hiding this comment

chhagedorn commented Jan 10, 2025

chhagedorn commented Jan 13, 2025

openjdk bot commented Jan 13, 2025

openjdk bot commented Jan 13, 2025

chhagedorn commented Dec 19, 2024 •

edited by openjdk bot

Loading

openjdk bot commented Dec 19, 2024 •

edited

Loading

mlbridge bot commented Dec 19, 2024 •

edited

Loading