This repository has been archived by the owner on Feb 20, 2025. It is now read-only.
Fix unpacker race condition by putting stall after MMIO #45
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
added 2 commits
November 5, 2024 20:09
…structions in reduce_revert_delta function" This reverts commit fefda7c.
|
Post commit tests : https://github.com/tenstorrent/tt-metal/actions/runs/11692311526 |
rdjogoTT
approved these changes
Nov 6, 2024
rtawfik01
approved these changes
Nov 6, 2024
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
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Outdated suggestions cannot be applied.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 the race condition by introducing a stall. Brought down the semaphore_post right before the stall as this is supposed to have no functional change. Since you cannot write to the address which is currently being used by the unpacker since there is a wait_for_next_context(2) right above it. Context can start from 0, then go to 1 and then come back to 0 and reach the MMIO writes when wait_for_next_context(2) passes, and that happens when the code that was previously working in context 0 has finished. Same argument for context 1.
Reverted the previous workaround change.