Fix use --pika:print-bind
, --pika:bind=none
, and a non-full process mask on the process together
#1082
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Outdated suggestions cannot be applied.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.
When using
--pika:print-bind
the runtime checks that the mask reported by hwloc and the mask supposedly used by pika are the same. When--pika:bind=none
is used we store an empty mask for worker threads. In this case we also don't set bindings for the worker threads, because we assume that the process has no binding set. However, in the case that the full process already has a mask set (e.g. withhwloc-bind
or slurm) the worker threads will inherit that mask even with--pika:bind=none
. The worker threads then end up with a non-full process mask even if the user asked for a full process mask (no binding).This fixes the issue by always setting a mask, even if the mask is empty/full. This also adds a simple
hello_world
test to the CircleCI configuration. I'm not adding this a CTest test since 1. I'm not quite sure how to do that cleanly and 2. I'd like to avoid making thehwloc-bind
executable a dependency for tests. Ideas for better integration into the tests is welcome though.