On Wed, Mar 27, 2024 at 10:07:23PM +0800, Chuyi Zhou wrote: > > One approach we can take is pushing the cpuset_migrate_mm_wq flushing to > > task_work so that it happens after cpuset mutex is dropped. That way we > > maintain the operation synchronicity for the issuer while avoiding bothering > > anyone else. > > > > Can you see whether the following patch fixes the issue for you? Thanks. > > > Thanks for the reply! I think it would help to fix this issue. > > BTW, will you merge this patch to the mainline? Sure but I'd really like to see that it actually solves the issues you're experiencing before merging it. Thanks. -- tejun
