Re: [PATCH v4 1/4] mm/rmap: duplicate a PMD mapping as one in folio_dup_file_rmap_pmd()

["Lorenzo Stoakes (ARM)" <ljs@kernel.org>]

On Thu, Oct 22, 2026 at 12:44:40PM +0200, Konstantin Khorenko wrote:
> On 10/22/26 10:07, Lorenzo Stoakes (ARM) wrote:
> >> Fixes: d8ef5e311d7b ("mm/rmap: convert page_dup_file_rmap() to folio_dup_file_rmap_[pte|ptes|pmd]()")
>
> > Was this a mistake in d8ef5e311d7b, or actually intentional for whatever reason?
> > If the latter this shouldn't be a fixes.
>
> commit d8ef5e311d7bfde54b60ab45026f206eff31b2d2
>
>     mm/rmap: convert page_dup_file_rmap() to folio_dup_file_rmap_[pte|ptes|pmd]()
>
> ...
> +static inline void folio_dup_file_rmap_ptes(struct folio *folio,
> +               struct page *page, int nr_pages)
> +{
> +       __folio_dup_file_rmap(folio, page, nr_pages, RMAP_LEVEL_PTE);
> +}
>
> ...
> +static inline void folio_dup_file_rmap_pmd(struct folio *folio,
> +               struct page *page)
> +{
> +#ifdef CONFIG_TRANSPARENT_HUGEPAGE
> +       __folio_dup_file_rmap(folio, page, HPAGE_PMD_NR, RMAP_LEVEL_PTE);
> +#else
> +       WARN_ON_ONCE(true);
> +#endif
> +}

OK so it was just wrong at the time :)

>
> Both functions were introduced in a single commit, i think it's a simple copy-paste issue.
>
> The helper never had a caller, so nobody could hit it. If you prefer, I
> can drop the Fixes tag and only mention d8ef5e311d7b in the changelog, as
> there is nothing to backport.

I mean, it seems like a bug to me? If there's no harm in backporting it I think you
should add Cc: stable@vger.kernel.org in that case actually.

--
Cheers, Lorenzo