Nouveau - Mar 2021 - [PATCH v7 5/8] mm: Device exclusive memory access

On Fri, Mar 26, 2021 at 11:08:02AM +1100, Alistair Popple
wrote:
> diff --git a/mm/memory.c b/mm/memory.c
> index 3a5705cfc891..33d11527ef77 100644
> +++ b/mm/memory.c
> @@ -781,6 +781,27 @@ copy_nonpresent_pte(struct mm_struct *dst_mm, struct
mm_struct *src_mm,
>  				pte = pte_swp_mkuffd_wp(pte);
>  			set_pte_at(src_mm, addr, src_pte, pte);
>  		}
> +	} else if (is_device_exclusive_entry(entry)) {
> +		page = pfn_swap_entry_to_page(entry);
> +
> +		get_page(page);
> +		rss[mm_counter(page)]++;
> +
> +		if (is_writable_device_exclusive_entry(entry) &&
> +		    is_cow_mapping(vm_flags)) {
> +			/*
> +			 * COW mappings require pages in both
> +			 * parent and child to be set to read.
> +			 */
> +			entry = make_readable_device_exclusive_entry(
> +							swp_offset(entry));
> +			pte = swp_entry_to_pte(entry);
> +			if (pte_swp_soft_dirty(*src_pte))
> +				pte = pte_swp_mksoft_dirty(pte);
> +			if (pte_swp_uffd_wp(*src_pte))
> +				pte = pte_swp_mkuffd_wp(pte);
> +			set_pte_at(src_mm, addr, src_pte, pte);
> +		}
This needs to have the same logic as we now have in
copy_present_page(). The page *is* present and we can't copy the PTE
value hidden in a swap entry if we can't copy the PTE normally.

The code should be shared because nobody is going to remember about
this corner case.

Jason

On Wednesday, 31 March 2021 6:32:34 AM AEDT Jason Gunthorpe
wrote:
> On Fri, Mar 26, 2021 at 11:08:02AM +1100, Alistair Popple wrote:
> > diff --git a/mm/memory.c b/mm/memory.c
> > index 3a5705cfc891..33d11527ef77 100644
> > +++ b/mm/memory.c
> > @@ -781,6 +781,27 @@ copy_nonpresent_pte(struct mm_struct *dst_mm,
struct
mm_struct *src_mm,
> >  				pte = pte_swp_mkuffd_wp(pte);
> >  			set_pte_at(src_mm, addr, src_pte, pte);
> >  		}
> > +	} else if (is_device_exclusive_entry(entry)) {
> > +		page = pfn_swap_entry_to_page(entry);
> > +
> > +		get_page(page);
> > +		rss[mm_counter(page)]++;
> > +
> > +		if (is_writable_device_exclusive_entry(entry) &&
> > +		    is_cow_mapping(vm_flags)) {
> > +			/*
> > +			 * COW mappings require pages in both
> > +			 * parent and child to be set to read.
> > +			 */
> > +			entry = make_readable_device_exclusive_entry(
> > +							swp_offset(entry));
> > +			pte = swp_entry_to_pte(entry);
> > +			if (pte_swp_soft_dirty(*src_pte))
> > +				pte = pte_swp_mksoft_dirty(pte);
> > +			if (pte_swp_uffd_wp(*src_pte))
> > +				pte = pte_swp_mkuffd_wp(pte);
> > +			set_pte_at(src_mm, addr, src_pte, pte);
> > +		}
> 
> This needs to have the same logic as we now have in
> copy_present_page(). The page *is* present and we can't copy the PTE
> value hidden in a swap entry if we can't copy the PTE normally.
You're saying we need to use copy_present_page() to make sure the split goes
the right way for pinned pages? I guess that makes sense as the split could go 
either way at the moment but I should add a check to make sure this isn't
used
with pinned pages anyway.

 - Alistair
> The code should be shared because nobody is going to remember about
> this corner case.
> 
> Jason
>