Repository navigation
Conversation
The lock and flag an asynchronous task synchronises on are plain volatile ints. The fences added for them order nothing on their own: a release fence pairs with an acquire fence only through atomic accesses, so a weakly ordered CPU may e.g. see a lock released before the data it guards. Access every lock and flag with __atomic acquire loads and release stores, which are valid in C and C++ and need no declaration change, and drop the now redundant ThreadFence.
This branch has not been deployed
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Draft on top of #3055 (base
fix-wait-timing); to be rebased ontomainonce #3055 merges.The handshake lock/flag are plain
volatile ints; the separate__atomic_thread_fences order nothing without atomic accesses, so a weakly ordered CPU may see a lock released before the data it guards. Every lock/flag access inside #3055'sBusyWait/BusyWaitCallnodes is now an__atomicacquire load or release store (valid C and C++, no declaration change), andThreadFenceis removed.Tests:
test_iet.py::test_async_handshake_atomics[save|prefetch](fail on #3055 alone);TestStreamingexpectations updated. On an H200 node (nvc 25.9) the OpenACC and OpenMP suites give failure sets identical to #3055.🤖 Generated with Claude Code