Re: [PATCH] mm/hugetlb: two-phase hugepage allocation when reservation is high

[David Hildenbrand]

On 27.08.25 14:33, Li,Rongqing wrote:
> > Not sure if something like the following could fly:
> >
> > diff --git a/mm/hugetlb.c b/mm/hugetlb.c index
> > 1f42186a85ea4..dfb4d717b8a02 100644
> > --- a/mm/hugetlb.c
> > +++ b/mm/hugetlb.c
> > @@ -3595,8 +3595,6 @@ static unsigned long __init
> > hugetlb_pages_alloc_boot(struct hstate *h)
> >           unsigned long jiffies_end;
> >
> >           job.thread_fn   = hugetlb_pages_alloc_boot_node;
> > -       job.start       = 0;
> > -       job.size        = h->max_huge_pages;
> >
> >           /*
> >            * job.max_threads is 25% of the available cpu threads by default.
> > @@ -3620,10 +3618,24 @@ static unsigned long __init
> > hugetlb_pages_alloc_boot(struct hstate *h)
> >           }
> >
> >           job.max_threads = hugepage_allocation_threads;
> > -       job.min_chunk   = h->max_huge_pages /
> > hugepage_allocation_threads;
> >
> >           jiffies_start = jiffies;
> > -       padata_do_multithreaded(&job);
> > +       /* TODO: comment why we retry and how it interacts with
> > vmemmap op. */
> > +       while (h->nr_huge_pages != h->max_huge_pages) {
> > +               unsigned long remaining = h->max_huge_pages -
> > + h->nr_huge_pages;
> > +
> > +               job.start       = h->nr_huge_pages;
> > +               job.size        = remaining;
> > +               job.min_chunk   = remaining /
> > hugepage_allocation_threads;
> > +               padata_do_multithreaded(&job);
> > +
> > +               if (hugetlb_vmemmap_optimizable(h))
> > +                       break;
>
> It should be:
>             if (!hugetlb_vmemmap_optimizable(h))
>                       break;

Very right.

--
Cheers

David / dhildenb