Fix worker <-> teardown by adding separate worker connection teardown semaphore #17033
+84
−23
Conversation
This file contains 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
SeanNijjar
commented
Jan 23, 2025
ttnn/cpp/ttnn/operations/ccl/kernels/edm_fabric/fabric_erisc_datamover_channels.hpp
Outdated
Show resolved
Hide resolved
…atamover_channels.hpp remove dprint
tt-aho
approved these changes
Jan 23, 2025
Comment on lines
179
to
181
| // A proper fix requires for example adding an additional teardown semaphore on the | ||
| // worker side that the EDM writes to to acknowledge teardown. The problem here | ||
| // is that the flow control aliases the teardown. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
Submitting on behalf of @tt-aho who did the work for this fix.
Ticket
#16634
Problem description
See issue for more details. There was a bug in the worker connection teardown logic that would lead to non-deterministic and unsafe worker <-> edm fabric teardown, possibly leading to hangs.
What's changed
Added a dedicated worker teardown semaphore on the worker that removes the (bug) aliasing between flow control logic and worker connection teardown logic. Note that this doesn't appear to resolve all hangs as another hang now appears after running several hundred iterations after applying this fix.
Checklist