Re: [External] Re: [PATCH v4 4/6] mm: hugetlb: retry dissolve page when hitting race

[Michal Hocko]

On Wed 13-01-21 19:20:17, Muchun Song wrote:
> On Wed, Jan 13, 2021 at 7:15 PM Oscar Salvador <osalvador@xxxxxxx> wrote:
> >
> > On Wed, Jan 13, 2021 at 07:11:06PM +0800, Muchun Song wrote:
> > > If there is no task to be scheduled. Here is just a while loop.
> > > The cpu_relax is a good thing to insert into busy-wait loops,
> > > right?
> >
> > But if the race window is that small, does it make sense?
> 
> Actually, there is one exception. The race window could
> become larger. If the page is freed via a workqueue (see
> free_huge_page()). In this case, the cpu_relax() can
> make sense. Right?

The system would have to be under serious stress for WQ to clog. I do
not expect there would be nothing runable at that stage. Possible?
Maybe but if that matters than a short sleep would be more preferable
than cpu_relax.
-- 
Michal Hocko
SUSE Labs