8334571: Extract control dependency rewiring out of PhaseIdealLoop::dominated_by() into separate method #19806

chhagedorn · 2024-06-20T14:24:23Z

This is another patch split off from the full Assertion Predicates fix.

PhaseIdealLoop::dominated_by() only works for an IfNode that dominates another IfNode. However, at some point, we want to replace the two IfNodes that are generated as Template Assertion Predicate for a hoisted range check with a single dedicated TemplateAssertionPredicateNode. To still be able to rewire the control dependencies in Loop Predication from the hoisted range check to the new TemplateAssertionPredicateNode outside the loop (instead of to the second IfNode projection of the Template Assertion Predicate as done today), we can extract the rewiring part out of dominated_by() into a separate method rewire_safe_outputs_to_dominator() and use that one instead.

We get the following in Loop Predication:

Invariant checks: Still use dominated_by() which calls rewire_safe_outputs_to_dominator()
Range checks: Use eliminate_hoisted_range_check() which calls rewire_safe_outputs_to_dominator().

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-8334571: Extract control dependency rewiring out of PhaseIdealLoop::dominated_by() into separate method (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/19806/head:pull/19806
$ git checkout pull/19806

Update a local copy of the PR:
$ git checkout pull/19806
$ git pull https://git.openjdk.org/jdk.git pull/19806/head

Using Skara CLI tools

Checkout this PR locally:
$ git pr checkout 19806

View PR using the GUI difftool:
$ git pr show -t 19806

Using diff file

Download this PR as a diff file:
https://git.openjdk.org/jdk/pull/19806.diff

Webrev

Link to Webrev Comment

…ominated_by() into separate method

bridgekeeper · 2024-06-20T14:24:57Z

👋 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 · 2024-06-20T14:25:25Z

@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:

8334571: Extract control dependency rewiring out of PhaseIdealLoop::dominated_by() into separate method

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 45 new commits pushed to the master branch:

a4582a8: 8334166: Enable binary check
7baddc2: 8334339: Test java/nio/file/attribute/BasicFileAttributeView/CreationTime.java fails on alinux3
eb110bd: 8334180: NMT gtests introduced with 8312132 should be labeled as NMT
652784c: 8334392: Switch RNG in NMT's treap
72ca7ba: 8334708: FFM: two javadoc problems
7e55ed3: 8333748: javap crash - Fatal error: Unmatched bit position 0x2 for location CLASS
1ff5acd: 8332099: since-checker - Add @ since to package-info in jdk.jsobject
689cee3: 8334509: Cancelling PageDialog does not return the same PageFormat object
8e1d2b0: 8334441: Mark tests in jdk_security_infra group as manual
93d9802: 8334715: [riscv] Mixed use of tab and whitespace in riscv.ad
... and 35 more: https://git.openjdk.org/jdk/compare/50bed6c67b1edd7736bdf79308d135a4e1047ff0...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 · 2024-06-20T14:25:43Z

@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 · 2024-06-20T14:25:49Z

src/hotspot/share/opto/loopPredicate.cpp

+    // Eliminate the old If in the loop body.
+    dominated_by(hoisted_check_predicate_proj, iff, negated);
+
+    C->print_method(PHASE_AFTER_LOOP_PREDICATION_IC, 4, hoisted_check_predicate_proj->in(0));


Additional benefit: This now dumps the graph after eliminating the hoisted check out of the loop. Same for the range checks.

chhagedorn · 2024-06-20T14:28:22Z

src/hotspot/share/opto/loopopts.cpp

+  for (DUIterator_Fast imax, i = source->fast_outs(imax); i < imax; i++) {
+    Node* out = source->fast_out(i); // Control-dependent node
    // Do not rewire Div and Mod nodes which could have a zero divisor to avoid skipping their zero check.
-    if (cd->depends_only_on_test() && _igvn.no_dependent_zero_check(cd)) {
-      assert(cd->in(0) == dp, "");
-      _igvn.replace_input_of(cd, 0, prevdom);
+    if (out->depends_only_on_test() && _igvn.no_dependent_zero_check(out)) {
+      assert(out->in(0) == source, "must be control dependent on source");
+      _igvn.replace_input_of(out, 0, dominator);
      if (pin_array_access_nodes) {
        // Because of Loop Predication, Loads and range check Cast nodes that are control dependent on this range
        // check (that is about to be removed) now depend on multiple dominating Hoisted Check Predicates. After the
        // removal of this range check, these control dependent nodes end up at the lowest/nearest dominating predicate
        // in the graph. To ensure that these Loads/Casts do not float above any of the dominating checks (even when the
        // lowest dominating check is later replaced by yet another dominating check), we need to pin them at the lowest
        // dominating check.
-        Node* clone = cd->pin_array_access_node();
+        Node* clone = out->pin_array_access_node();
        if (clone != nullptr) {
-          clone = _igvn.register_new_node_with_optimizer(clone, cd);
-          _igvn.replace_node(cd, clone);
-          cd = clone;
+          clone = _igvn.register_new_node_with_optimizer(clone, out);
+          _igvn.replace_node(out, clone);
+          out = clone;
        }
      }
-      set_early_ctrl(cd, false);
-      IdealLoopTree* new_loop = get_loop(get_ctrl(cd));
+      set_early_ctrl(out, false);
+      IdealLoopTree* new_loop = get_loop(get_ctrl(out));
      if (old_loop != new_loop) {
        if (!old_loop->_child) {
-          old_loop->_body.yank(cd);
+          old_loop->_body.yank(out);
        }
        if (!new_loop->_child) {
-          new_loop->_body.push(cd);
+          new_loop->_body.push(out);


Renaming only and adding assertion message

chhagedorn · 2024-06-20T14:29:56Z

src/hotspot/share/opto/loopPredicate.cpp

+void PhaseIdealLoop::eliminate_hoisted_range_check(IfTrueNode* hoisted_check_proj,
+                                                   IfTrueNode* template_assertion_predicate_proj) {
+  _igvn.replace_input_of(hoisted_check_proj->in(0), 1, _igvn.intcon(1));
+  rewire_safe_outputs_to_dominator(hoisted_check_proj, template_assertion_predicate_proj, true);


pin_array_access_nodes = true since this is called only for range checks now.

chhagedorn · 2024-06-20T14:37:59Z

src/hotspot/share/opto/loopPredicate.cpp

@@ -1281,24 +1292,21 @@ bool PhaseIdealLoop::loop_predication_impl_helper(IdealLoopTree* loop, IfProjNod
    // with uncommon trap.
    return false;
  }
-  assert(new_predicate_proj != nullptr, "sanity");
-  // Success - attach condition (new_predicate_bol) to predicate if
-  invar.map_ctrl(if_success_proj, new_predicate_proj); // so that invariance test can be appropriate


Moved to invariant checks and range checks separately.

mlbridge · 2024-06-20T14:42:20Z

Webrevs

00: Full (72fe6d38)

rwestrel

Looks reasonable to me.

vnkozlov

Looks good.

chhagedorn · 2024-06-21T08:50:22Z

Thanks Roland and Vladimir for your reviews!

chhagedorn · 2024-06-24T08:57:00Z

/integrate

openjdk · 2024-06-24T08:58:04Z

Going to push as commit ca5a438.
Since your change was applied there have been 50 commits pushed to the master branch:

05ff318: 8334594: Generational ZGC: Deadlock after OopMap rewrites in 8331572
05a63d8: 8334489: Add function os::used_memory
edf7f05: 8334083: C2 SuperWord: TestCompatibleUseDefTypeSize.java fails with -XX:+AlignVector after JDK-8325155
13dce29: 8334560: [PPC64]: postalloc_expand_java_dynamic_call_sched does not copy all fields
863b2a9: 8329994: Zap alignment padding bits for ArrayOops in non-release builds
a4582a8: 8334166: Enable binary check
7baddc2: 8334339: Test java/nio/file/attribute/BasicFileAttributeView/CreationTime.java fails on alinux3
eb110bd: 8334180: NMT gtests introduced with 8312132 should be labeled as NMT
652784c: 8334392: Switch RNG in NMT's treap
72ca7ba: 8334708: FFM: two javadoc problems
... and 40 more: https://git.openjdk.org/jdk/compare/50bed6c67b1edd7736bdf79308d135a4e1047ff0...master

Your commit was automatically rebased without conflicts.

openjdk · 2024-06-24T08:58:12Z

@chhagedorn Pushed as commit ca5a438.

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

8334571: Extract control dependency rewiring out of PhaseIdealLoop::d…

cd6f835

…ominated_by() into separate method

openjdk bot added the hotspot-compiler hotspot-compiler-dev@openjdk.org label Jun 20, 2024

chhagedorn added 2 commits June 20, 2024 16:35

code style

0bdc62b

code style

72fe6d3

chhagedorn commented Jun 20, 2024

View reviewed changes

chhagedorn marked this pull request as ready for review June 20, 2024 14:38

openjdk bot added the rfr Pull request is ready for review label Jun 20, 2024

rwestrel approved these changes Jun 20, 2024

View reviewed changes

openjdk bot added the ready Pull request is ready to be integrated label Jun 20, 2024

vnkozlov approved these changes Jun 20, 2024

View reviewed changes

openjdk bot added the integrated Pull request has been integrated label Jun 24, 2024

openjdk bot closed this Jun 24, 2024

openjdk bot removed ready Pull request is ready to be integrated rfr Pull request is ready for review labels Jun 24, 2024

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

8334571: Extract control dependency rewiring out of PhaseIdealLoop::dominated_by() into separate method #19806

8334571: Extract control dependency rewiring out of PhaseIdealLoop::dominated_by() into separate method #19806

chhagedorn commented Jun 20, 2024 •

edited by openjdk bot

Loading

bridgekeeper bot commented Jun 20, 2024

openjdk bot commented Jun 20, 2024 •

edited

Loading

openjdk bot commented Jun 20, 2024

chhagedorn Jun 20, 2024

chhagedorn Jun 20, 2024

chhagedorn Jun 20, 2024

chhagedorn Jun 20, 2024

mlbridge bot commented Jun 20, 2024

rwestrel left a comment

vnkozlov left a comment

chhagedorn commented Jun 21, 2024

chhagedorn commented Jun 24, 2024

openjdk bot commented Jun 24, 2024

openjdk bot commented Jun 24, 2024

8334571: Extract control dependency rewiring out of PhaseIdealLoop::dominated_by() into separate method #19806

8334571: Extract control dependency rewiring out of PhaseIdealLoop::dominated_by() into separate method #19806

Conversation

chhagedorn commented Jun 20, 2024 • edited by openjdk bot Loading

Progress

Issue

Reviewers

Reviewing

Webrev

bridgekeeper bot commented Jun 20, 2024

openjdk bot commented Jun 20, 2024 • edited Loading

openjdk bot commented Jun 20, 2024

chhagedorn Jun 20, 2024

Choose a reason for hiding this comment

chhagedorn Jun 20, 2024

Choose a reason for hiding this comment

chhagedorn Jun 20, 2024

Choose a reason for hiding this comment

chhagedorn Jun 20, 2024

Choose a reason for hiding this comment

mlbridge bot commented Jun 20, 2024

Webrevs

rwestrel left a comment

Choose a reason for hiding this comment

vnkozlov left a comment

Choose a reason for hiding this comment

chhagedorn commented Jun 21, 2024

chhagedorn commented Jun 24, 2024

openjdk bot commented Jun 24, 2024

openjdk bot commented Jun 24, 2024

chhagedorn commented Jun 20, 2024 •

edited by openjdk bot

Loading

openjdk bot commented Jun 20, 2024 •

edited

Loading