JIT: Improve precision of tailcall "has GC safe point" assert #123489
+55
−14
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Since tailcalls can form tight loops that do not allow for return address hijacking we take care to introduce GC safe points in those scenarios. We had an assert for this, but the assert can result in false positives. In the test case we have two tailcalls via JIT helpers, both in blocks with GC safe points in them. Later tail merging merges them into a new block without a GC safe point, and that makes the assert in lowering hit, despite the fact that both preds are GC safe points. Fix the assert by introducing a graph walk. Fast tailcalls can use the same assert now, fixing a TODO-Cleanup at the same time.
github-actions
bot
added
the
area-CodeGen-coreclr
Contributor
|
Tagging subscribers to this area: @JulieLeeMSFT, @dotnet/jit-contrib |
Member
Author
|
/azp run runtime-coreclr jitstress |
|
Azure Pipelines successfully started running 1 pipeline(s). |
Member
Author
|
cc @dotnet/jit-contrib PTAL @EgorBo |
Contributor
There was a problem hiding this comment.
Pull request overview
This PR updates the JIT’s debug-only tailcall “GC safe point” assertion logic to avoid false negatives caused by tail-merge creating a new block without a GC safe point, despite all predecessors being GC safe points.
Changes:
- Replace the old tailcall-via-helper “current block OR entry block is a GC safe point” assert with a reachability-based check.
- Apply the same reachability-based assert to fast tailcalls, removing the prior TODO-based workaround.
- Introduce a DEBUG-only CFG walk helper (
IsBlockReachableWithoutGCSafePoint) used by these asserts.
Reviewed changes
Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.
| File | Description |
|---|---|
| src/coreclr/jit/lower.h | Declares a new helper used by debug asserts to validate GC-safe-point reachability. |
| src/coreclr/jit/lower.cpp | Updates tailcall asserts and adds a DEBUG-only CFG walk to avoid false negatives after tail merging. |
jakobbotsch
commented
Jan 23, 2026
EgorBo
approved these changes
Jan 23, 2026
Member
Author
|
/azp run runtime-coreclr jitstress, runtime-coreclr libraries-jitstress |
|
Azure Pipelines successfully started running 2 pipeline(s). |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Since tailcalls can form tight loops that do not allow for return address hijacking we take care to introduce GC safe points in those scenarios. We had an assert for this, but the assert can result in false negatives.
In the test case we have two tailcalls via JIT helpers, both in blocks with GC safe points in them. Later tail merging merges them into a new block without a GC safe point, and that makes the assert in lowering hit, despite the fact that both preds are GC safe points.
Fix the assert by introducing a graph walk. Fast tailcalls can use the same assert now, fixing a TODO-Cleanup at the same time.
Fix #122481