Re: [RFC PATCH 01/57] mm: add pte_folio()

[Kiryl Shutsemau]

On Tue, Aug 18, 2026 at 07:09:48PM +0200, David Hildenbrand (Arm) wrote:
> On 8/17/26 00:45, Kiryl Shutsemau wrote:
> > From: "Kiryl Shutsemau (Meta)" <kas@xxxxxxxxxx>
> > 
> > Callers that want the folio behind a present PTE spell it out as
> > page_folio(pte_page(pte)).
> > 
> > Add pte_folio() as the folio companion to pte_page(), and convert the
> > callers in fs/proc/task_mmu.c and mm/hugetlb.c.
> > 
> > Preparation for the anonymous collapse engine, which reads the folio
> > behind a PTE in several places.
> 
> [...]
> 
> >  
> > +/**
> > + * pte_folio - Return the folio mapped by a present PTE.
> > + * @pte: A present page table entry.
> > + *
> > + * The folio companion to pte_page(); only meaningful for a present PTE
> > + * that maps a struct-page-backed folio.
> > + *
> > + * Return: The folio containing the page @pte maps.
> > + */
> > +static inline struct folio *pte_folio(pte_t pte)
> > +{
> > +	return page_folio(pte_page(pte));
> > +}
> 
> There is a reason why most code doesn't need that: because they should be using
> vm_normal_page() / vm_normal_folio(), or need the exact page and handle special
> ptes differently (see gup.c that uses pte_page()).
> 
> And other code that uses pte_page() doesn't really operate on folios AFAIKs.

Fair.  pte_folio() does not fit the abstraction as it stands, so I will
drop it.

I thought that maybe we can rework vm_normal_*() to be a predicate on a
page table entry, so if it returns true the caller can safely use
pxx_folio(). It would be a cleaner interface in my opinion.

vm_ops->find_normal_page() breaks the idea, though.  Xen (being the single
user of the vm_op) wants the special bit to indicate a custom translation
to the page here.  A bool cannot describe this.

Ugh..

-- 
  Kiryl Shutsemau / Kirill A. Shutemov