Re: [PATCH v10 1/9] mm: Introduce memfd_restricted system call to create restricted user memory

From: Chao Peng
Date: Mon Dec 19 2022 - 02:57:53 EST


On Tue, Dec 13, 2022 at 11:49:13PM +0000, Huang, Kai wrote:
> >
> > memfd_restricted() itself is implemented as a shim layer on top of real
> > memory file systems (currently tmpfs). Pages in restrictedmem are marked
> > as unmovable and unevictable, this is required for current confidential
> > usage. But in future this might be changed.
> >
> >
> I didn't dig full histroy, but I interpret this as we don't support page
> migration and swapping for restricted memfd for now. IMHO "page marked as
> unmovable" can be confused with PageMovable(), which is a different thing from
> this series. It's better to just say something like "those pages cannot be
> migrated and swapped".

Yes, if that helps some clarification.

>
> [...]
>
> > +
> > + /*
> > + * These pages are currently unmovable so don't place them into movable
> > + * pageblocks (e.g. CMA and ZONE_MOVABLE).
> > + */
> > + mapping = memfd->f_mapping;
> > + mapping_set_unevictable(mapping);
> > + mapping_set_gfp_mask(mapping,
> > + mapping_gfp_mask(mapping) & ~__GFP_MOVABLE);
>
> But, IIUC removing __GFP_MOVABLE flag here only makes page allocation from non-
> movable zones, but doesn't necessarily prevent page from being migrated. My
> first glance is you need to implement either a_ops->migrate_folio() or just
> get_page() after faulting in the page to prevent.

The current api restrictedmem_get_page() already does this, after the
caller calling it, it holds a reference to the page. The caller then
decides when to call put_page() appropriately.

>
> So I think the comment also needs improvement -- IMHO we can just call out
> currently those pages cannot be migrated and swapped, which is clearer (and the
> latter justifies mapping_set_unevictable() clearly).

Good to me.

Thanks,
Chao
>
>