Re: [PATCH 03/11] mm/hugetlb: introduce alloc_control structure to simplify migration target allocation APIs

[Joonsoo Kim]

2020ë 5ì 22ì (ê) ìì 3:57, Mike Kravetz <mike.kravetz@xxxxxxxxxx>ëì ìì:
>
> On 5/17/20 6:20 PM, js1304@xxxxxxxxx wrote:
> > From: Joonsoo Kim <iamjoonsoo.kim@xxxxxxx>
> >
> > Currently, page allocation functions for migration requires some arguments.
> > More worse, in the following patch, more argument will be needed to unify
> > the similar functions. To simplify them, in this patch, unified data
> > structure that controls allocation behaviour is introduced.
>
> As a followup to Roman's question and your answer about adding a suffix/prefix
> to the new structure.  It 'may' be a bit confusing as alloc_context is already
> defined and *ac is passsed around for page allocations.  Perhaps, this new
> structure could somehow have migrate in the name as it is all about allocating
> migrate targets?

I have considered that but I cannot find appropriate prefix. In hugetlb code,
struct alloc_control is passed to the internal function which is not
fully dedicated
to the migration so 'migrate' would not be appropriate prefix.

alloc_context is used by page allocation core and alloc_control would be used by
outside of it so I think that we can endure it. If there is a good
suggestion, I will change
the name happily.

> >
> > For clean-up, function declarations are re-ordered.
> >
> > Note that, gfp_mask handling on alloc_huge_page_(node|nodemask) is
> > slightly changed, from ASSIGN to OR. It's safe since caller of these
> > functions doesn't pass extra gfp_mask except htlb_alloc_mask().
> >
> > Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@xxxxxxx>
>
> Patch makes sense.

Thanks!

> > diff --git a/mm/migrate.c b/mm/migrate.c
> > index a298a8c..94d2386 100644
> > --- a/mm/migrate.c
> > +++ b/mm/migrate.c
> > @@ -1526,10 +1526,15 @@ struct page *new_page_nodemask(struct page *page,
> >       unsigned int order = 0;
> >       struct page *new_page = NULL;
> >
> > -     if (PageHuge(page))
> > -             return alloc_huge_page_nodemask(
> > -                             page_hstate(compound_head(page)),
> > -                             preferred_nid, nodemask);
> > +     if (PageHuge(page)) {
> > +             struct hstate *h = page_hstate(page);
>
> I assume the removal of compound_head(page) was intentional?  Just asking
> because PageHuge will look at head page while page_hstate will not.  So,
> if passed a non-head page things could go bad.

I was thinking that page_hstate() can handle the tail page but it seems that
it's not. Thanks for correction. I will change it on next version.

Thanks.