Re: [PATCH v2] hugetlbfs: Disable softIRQ when taking hugetlb_lock

[Davidlohr Bueso]

On Thu, 12 Dec 2019, Andi Kleen wrote:

> Davidlohr Bueso <dave@xxxxxxxxxxxx> writes:
> > +void free_huge_page(struct page *page)
> > +{
> > +	struct hugetlb_free_page_work work;
> > +
> > +	work.page = page;
> > +	INIT_WORK_ONSTACK(&work.work, free_huge_page_workfn);
> > +	queue_work(hugetlb_free_page_wq, &work.work);
> > +
> > +	/*
> > +	 * Wait until free_huge_page is done.
> > +	 */
> > +	flush_work(&work.work);
> > +	destroy_work_on_stack(&work.work);
>
> Does flushing really work in softirq context?
>
> Anyways, waiting seems inefficient over fire'n'forget

Yep. I was only thinking about the workerfn not blocking
and therefore we could wait safely, but flush_work has no
such guarantees.

Thanks,
Davidlohr