8275330: C2: assert(n->is_Root() || n->is_Region() || n->is_Phi() || n->is_MachMerge() || def_block->dominates(block)) failed: uses must be dominated by definitions #6429

rwestrel · 2021-11-17T12:16:32Z

This is similar to previous bugs where:

a cast/conv node captures a narrow type in a loop body because of a
range check,
the range check is optimized out of the loop, pre/main/post loop are
created
overunrolling causes the main loop to become unreachable (the range
check, if still in the main loop, would fail), the cast transforms to
top but c2 can't optimize the loop out

This was fixed by adding predicates above the main loop. With this
particular bug, the cast node is in the post loop. The fix I propose
is to also add predicates above the post loop. There are a few
locations in the code that cause a post loop to be added: either the
initial post loop or some other post loops for vectorization
support. I think the new predicates are needed in a all cases. To be
able to add predicates at these different points in the optimization
process, the new predicates are copied from the main loop predicates.

I also delayed folding of Opaque4 nodes to macro expansion rather than
post loop opts igvn. The reason for that is that I believe there's a
risk that an Opaque4 is removed (that is replaced by its input 2)
before its input 1 has a chance to constant fold. That wouldn't happen
with a debug build because we leave the tests in (that is replace the
Opaque4 node by its input 1) so that corner case is not properly
tested currently. The reason for leaving the tests in was to sanity
check that the tests are indeed correct.

Progress

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

Issue

JDK-8275330: C2: assert(n->is_Root() || n->is_Region() || n->is_Phi() || n->is_MachMerge() || def_block->dominates(block)) failed: uses must be dominated by definitions

Reviewers

Tobias Hartmann (@TobiHartmann - Reviewer)
Christian Hagedorn (@chhagedorn - Reviewer)

Reviewing

Using git

Checkout this PR locally:
$ git fetch https://git.openjdk.java.net/jdk pull/6429/head:pull/6429
$ git checkout pull/6429

Update a local copy of the PR:
$ git checkout pull/6429
$ git pull https://git.openjdk.java.net/jdk pull/6429/head

Using Skara CLI tools

Checkout this PR locally:
$ git pr checkout 6429

View PR using the GUI difftool:
$ git pr show -t 6429

Using diff file

Download this PR as a diff file:
https://git.openjdk.java.net/jdk/pull/6429.diff

bridgekeeper · 2021-11-17T12:16:39Z

👋 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 · 2021-11-17T12:19:02Z

@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 · 2021-11-17T12:23:01Z

Webrevs

01: Full - Incremental (b194700e)
00: Full (50c8c797)

TobiHartmann

That looks good to me.

TobiHartmann · 2021-11-23T07:34:58Z

src/hotspot/share/opto/loopTransform.cpp

+        prev_proj = clone_skeleton_predicate_for_main_or_post_loop(iff, init, stride, ctrl, proj, post_loop_entry,
+                                                                   post_loop, prev_proj);
+        assert(!skeleton_predicate_has_opaque(prev_proj->in(0)->as_If()), "unexpected");
+      }


Intendation is wrong.

openjdk · 2021-11-23T07:36:26Z

@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:

8275330: C2:  assert(n->is_Root() || n->is_Region() || n->is_Phi() || n->is_MachMerge() || def_block->dominates(block)) failed: uses must be dominated by definitions

Reviewed-by: thartmann, chagedorn

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 6 new commits pushed to the master branch:

00a6238: 8277508: need to check has_predicated_vectors before calling scalable_predicate_reg_slots
3383c0d: 8277659: [TESTBUG] Microbenchmark ThreadOnSpinWaitProducerConsumer.java hangs
7e54d06: 8277165: jdeps --multi-release --print-module-deps fails if module-info.class in different versioned directories
eb4d886: 8277504: Use String.stripTrailing instead of hand-crafted method in SwingUtilities2
9879920: 8277825: Remove unused ReferenceProcessorPhaseTimes::_sub_phases_total_time_ms
f788834: 8277786: G1: Rename log2_card_region_per_heap_region used in G1CardSet

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.

chhagedorn

Otherwise, the fix looks reasonable to me! When I was fixing related bugs before, I found myself wondering if the post loop does not need these predicates as well - turns out now it does.

chhagedorn · 2021-11-23T11:46:55Z

src/hotspot/share/opto/loopTransform.cpp

@@ -1544,7 +1545,8 @@ void PhaseIdealLoop::insert_pre_post_loops(IdealLoopTree *loop, Node_List &old_n
  // Add the post loop
  const uint idx_before_pre_post = Compile::current()->unique();
  CountedLoopNode *post_head = NULL;
-  Node *main_exit = insert_post_loop(loop, old_new, main_head, main_end, incr, limit, post_head);
+  Node* post_incr = incr;
+  Node *main_exit = insert_post_loop(loop, old_new, main_head, main_end, post_incr, limit, post_head);


Could also be updated: Node *main_exit -> Node* main_exit.

chhagedorn · 2021-11-23T11:46:55Z

src/hotspot/share/opto/loopTransform.cpp

@@ -1968,6 +1975,34 @@ void PhaseIdealLoop::update_main_loop_skeleton_predicates(Node* ctrl, CountedLoo
  }
 }

+void PhaseIdealLoop::insert_post_loop_skeleton_predicates(LoopNode* main_loop_head, CountedLoopNode* post_loop_head, Node* init, Node* stride) {


Maybe this could be renamed to copy_skeleton_predicates_to_post_loop() to be consistent with the method copy_skeleton_predicates_to_main_loop()?

chhagedorn · 2021-11-23T11:46:55Z

src/hotspot/share/opto/loopTransform.cpp

@@ -1915,6 +1920,7 @@ Node *PhaseIdealLoop::insert_post_loop(IdealLoopTree *loop, Node_List &old_new,
  // CastII for the new post loop:
  Node* castii = cast_incr_before_loop(zer_opaq->in(1), zer_taken, post_head);
  assert(castii != NULL, "no castII inserted");
+  incr = castii;


You could directly assign incr on L1921 instead of using the additional castii variable.

rwestrel · 2021-11-26T07:48:45Z

@TobiHartmann @chhagedorn Thanks for the reviews. All comment should be addressed in the updated change.

chhagedorn

Thanks for doing the updates, looks good!

TobiHartmann

Looks good. All testing passed.

rwestrel · 2021-11-26T09:47:00Z

/integrate

openjdk · 2021-11-26T09:48:20Z

Going to push as commit 3e798dd.
Since your change was applied there have been 7 commits pushed to the master branch:

99e4bda: 8277417: C1 LIR instruction for load-klass
00a6238: 8277508: need to check has_predicated_vectors before calling scalable_predicate_reg_slots
3383c0d: 8277659: [TESTBUG] Microbenchmark ThreadOnSpinWaitProducerConsumer.java hangs
7e54d06: 8277165: jdeps --multi-release --print-module-deps fails if module-info.class in different versioned directories
eb4d886: 8277504: Use String.stripTrailing instead of hand-crafted method in SwingUtilities2
9879920: 8277825: Remove unused ReferenceProcessorPhaseTimes::_sub_phases_total_time_ms
f788834: 8277786: G1: Rename log2_card_region_per_heap_region used in G1CardSet

Your commit was automatically rebased without conflicts.

openjdk · 2021-11-26T09:48:32Z

@rwestrel Pushed as commit 3e798dd.

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

fix

50c8c79

openjdk bot added the rfr label Nov 17, 2021

openjdk bot added the hotspot-compiler label Nov 17, 2021

TobiHartmann approved these changes Nov 23, 2021

View changes

openjdk bot added the ready label Nov 23, 2021

chhagedorn approved these changes Nov 23, 2021

View changes

rwestrel added 2 commits Nov 25, 2021

reviews

417dc88

Merge branch 'master' into JDK-8275330

b194700

chhagedorn approved these changes Nov 26, 2021

View changes

TobiHartmann approved these changes Nov 26, 2021

View changes

openjdk bot closed this Nov 26, 2021

openjdk bot added integrated and removed ready rfr labels Nov 26, 2021

pfustc mentioned this pull request Jul 26, 2022

8289954: C2: Assert failed in PhaseCFG::verify() after JDK-8183390 openjdk/jdk19#130

Closed

3 tasks

8275330: C2: assert(n->is_Root() || n->is_Region() || n->is_Phi() || n->is_MachMerge() || def_block->dominates(block)) failed: uses must be dominated by definitions #6429

8275330: C2: assert(n->is_Root() || n->is_Region() || n->is_Phi() || n->is_MachMerge() || def_block->dominates(block)) failed: uses must be dominated by definitions #6429

rwestrel commented Nov 17, 2021 •

edited by openjdk <span class="Label Label--secondary">bot</span>

bridgekeeper bot commented Nov 17, 2021

openjdk bot commented Nov 17, 2021

mlbridge bot commented Nov 17, 2021 •

edited

TobiHartmann left a comment

TobiHartmann Nov 23, 2021

openjdk bot commented Nov 23, 2021 •

edited

chhagedorn left a comment

chhagedorn Nov 23, 2021

chhagedorn Nov 23, 2021

chhagedorn Nov 23, 2021

rwestrel commented Nov 26, 2021

chhagedorn left a comment

TobiHartmann left a comment

rwestrel commented Nov 26, 2021

openjdk bot commented Nov 26, 2021

openjdk bot commented Nov 26, 2021

8275330: C2: assert(n->is_Root() || n->is_Region() || n->is_Phi() || n->is_MachMerge() || def_block->dominates(block)) failed: uses must be dominated by definitions #6429

8275330: C2: assert(n->is_Root() || n->is_Region() || n->is_Phi() || n->is_MachMerge() || def_block->dominates(block)) failed: uses must be dominated by definitions #6429

Conversation

rwestrel commented Nov 17, 2021 • edited by openjdk <span class="Label Label--secondary">bot</span>

Progress

Issue

Reviewers

Reviewing

bridgekeeper bot commented Nov 17, 2021

openjdk bot commented Nov 17, 2021

mlbridge bot commented Nov 17, 2021 • edited

Webrevs

TobiHartmann left a comment

TobiHartmann Nov 23, 2021

Choose a reason for hiding this comment

openjdk bot commented Nov 23, 2021 • edited

chhagedorn left a comment

chhagedorn Nov 23, 2021

Choose a reason for hiding this comment

chhagedorn Nov 23, 2021

Choose a reason for hiding this comment

chhagedorn Nov 23, 2021

Choose a reason for hiding this comment

rwestrel commented Nov 26, 2021

chhagedorn left a comment

TobiHartmann left a comment

rwestrel commented Nov 26, 2021

openjdk bot commented Nov 26, 2021

openjdk bot commented Nov 26, 2021

rwestrel commented Nov 17, 2021 •

edited by openjdk <span class="Label Label--secondary">bot</span>

mlbridge bot commented Nov 17, 2021 •

edited

openjdk bot commented Nov 23, 2021 •

edited