(bugfix) mlx barrier hung#428
Merged
Merged
Conversation
kawhil-amd
approved these changes
Jun 26, 2026
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Fix GDA barrier hang on MLX5
The barrier deadlocked on MLX5: a rank rang all N atomic-signal doorbells, but
peers saw only one and spun forever in the phase-2 wait.
Cause: the multi-lane path assumed MLX5's per-QP
dbrAddrmade doorbell storesindependent, so lanes rang directly. On this GPU the per-QP doorbell stores
issued in one SIMT step coalesce — only one survives, the rest are dropped, so
those QPs' WQEs are never fetched and peers hang.
Fix: route MLX5 through the same per-lane serialized
ringDoorbellWalkPSD/BNXTalready use.