Re: [RFC PATCH v2 2/3] mm: Add PUD level pagetable account

[Baolin Wang]

On 6/22/2022 10:38 PM, Mike Rapoport wrote:
> On Wed, Jun 22, 2022 at 04:58:53PM +0800, Baolin Wang wrote:
> > Now the PUD level ptes are always protected by mm->page_table_lock,
> > which means no split pagetable lock needed. So the generic PUD level
> > pagetable pages allocation will not call pgtable_pte_page_ctor/dtor(),
> > that means we will miss to account PUD level pagetable pages.
> >
> > Adding pagetable account by calling pgtable_set_and_inc() or
> > pgtable_clear_and_dec() when allocating or freeing PUD level pagetable
> > pages to help to get an accurate pagetable accounting.
> >
> > Moreover this patch will also mark the PUD level pagetable with PG_table
> > flag, which will help to do sanity validation in unpoison_memory() and
> > get more accurate pagetable accounting by /proc/kpageflags interface.
> >
> > Meanwhile converting the architectures with using generic PUD pagatable
> > allocation to add corresponding pgtable_set_and_inc() or pgtable_clear_and_dec()
> > to account PUD level pagetable.
> >
> > Signed-off-by: Baolin Wang <baolin.wang@xxxxxxxxxxxxxxxxx>
> > ---
> >   arch/arm64/include/asm/tlb.h         |  5 ++++-
> >   arch/loongarch/include/asm/pgalloc.h | 11 ++++++++---
> >   arch/mips/include/asm/pgalloc.h      | 11 ++++++++---
> >   arch/s390/include/asm/tlb.h          |  1 +
> >   arch/x86/mm/pgtable.c                |  5 ++++-
> >   include/asm-generic/pgalloc.h        | 12 ++++++++++--
> >   6 files changed, 35 insertions(+), 10 deletions(-)
> >
> > diff --git a/arch/arm64/include/asm/tlb.h b/arch/arm64/include/asm/tlb.h
> > index c995d1f..47e0623 100644
> > --- a/arch/arm64/include/asm/tlb.h
> > +++ b/arch/arm64/include/asm/tlb.h
> > @@ -94,7 +94,10 @@ static inline void __pmd_free_tlb(struct mmu_gather *tlb, pmd_t *pmdp,
> >   static inline void __pud_free_tlb(struct mmu_gather *tlb, pud_t *pudp,
> >   				  unsigned long addr)
> >   {
> > -	tlb_remove_table(tlb, virt_to_page(pudp));
> > +	struct page *page = virt_to_page(pudp);
> > +
> > +	pgtable_clear_and_dec(page);
> > +	tlb_remove_table(tlb, page);
> >   }
> >   #endif
> >   
> > diff --git a/arch/loongarch/include/asm/pgalloc.h b/arch/loongarch/include/asm/pgalloc.h
> > index b0a57b2..50a896f 100644
> > --- a/arch/loongarch/include/asm/pgalloc.h
> > +++ b/arch/loongarch/include/asm/pgalloc.h
> > @@ -89,10 +89,15 @@ static inline pmd_t *pmd_alloc_one(struct mm_struct *mm, unsigned long address)
> >   static inline pud_t *pud_alloc_one(struct mm_struct *mm, unsigned long address)
> >   {
> >   	pud_t *pud;
> > +	struct page *pg;
>
> 	struct page *page;
>
> looks better IMO.

Sure.

> > +
> > +	pg = alloc_pages(GFP_KERNEL & ~__GFP_HIGHMEM, PUD_ORDER);
> > +	if (!pg)
> > +		return NULL;
> >   
> > -	pud = (pud_t *) __get_free_pages(GFP_KERNEL, PUD_ORDER);
> > -	if (pud)
> > -		pud_init((unsigned long)pud, (unsigned long)invalid_pmd_table);
> > +	pgtable_set_and_inc(pg);
> > +	pud = (pud_t *)page_address(pg);
>
> I don't think __get_free_pages() should be replaced with alloc_pages()
> here, just call pgtable_set_and_inc() with virt_to_page(pud).
>
> The same applies for the cases below.

Sure. Will do in next version. Thanks.