Re: [PATCH v2 4/5] mm: Refector release_pages()

[Matthew Wilcox <willy@xxxxxxxxxxxxx>]

On Wed, Aug 30, 2023 at 10:50:10AM +0100, Ryan Roberts wrote:
> In preparation for implementing folios_put_refs() in the next patch,
> refactor release_pages() into a set of helper functions, which can be
> reused. The primary difference between release_pages() and
> folios_put_refs() is how they iterate over the set of folios. The
> per-folio actions are identical.

As you noted, we have colliding patchsets.  I'm not hugely happy with
how patch 4 turned out, so I thought I'd send some addendum patches to
my RFC series that implement pfn_range_put() (maybe should have been
pfn_ranges_put()?) on top of my patch series.  I think it's a bit nicer,
but not quite as nice as it could be.

I'm thinking about doing ...

void release_unref_folios(struct folio_batch *folios)
{
	struct lruvec *lruvec = NULL;
	unsigned long flags = 0;
	int i;

	for (i = 0; i < folios->nr; i++) {
		struct folio *folio = folios->folios[i];
		free_swap_cache(folio);
		__page_cache_release(folio, &lruvec, &flags);
	}
	mem_cgroup_uncharge_folios(folios);
	free_unref_folios(folios);
}

then this becomes:

void folios_put(struct folio_batch *folios)
{
        int i, j;

        for (i = 0, j = 0; i < folios->nr; i++) {
                struct folio *folio = folios->folios[i];

                if (is_huge_zero_page(&folio->page))
                        continue;
                if (folio_is_zone_device(folio)) {
                       if (put_devmap_managed_page(&folio->page))
                                continue;
                        if (folio_put_testzero(folio))
                                free_zone_device_page(&folio->page);
                        continue;
                }

                if (!folio_put_testzero(folio))
                        continue;
                if (folio_test_hugetlb(folio)) {
                       free_huge_folio(folio);
                        continue;
                }

                if (j != i)
                        folios->folios[j] = folio;
                j++;
        }

        folios->nr = j;
        if (!j)
                return;
	release_unref_folios(folios);
}

and pfn_range_put() also becomes shorter and loses all the lruvec work.

Thoughts?