[PATCH 1/3] mm: Move arch_do_swap_page() call to before swap_free()

[Peter Collingbourne]

On Wed, May 17, 2023 at 1:30 AM David Hildenbrand <david at redhat.com> wrote:
>
> >> Would the idea be to fail swap_readpage() on the one that comes last,
> >> simply retrying to lookup the page?
> >
> > The idea would be that T2's arch_swap_readpage() could potentially not
> > find tags if it ran after swap_free(), so T2 would produce a page
> > without restored tags. But that wouldn't matter, because T1 reaching
> > swap_free() means that T2 will follow the goto at [1] after waiting
> > for T1 to unlock at [2], and T2's page will be discarded.
>
> Ah, right.
>
> >
> >> This might be a naive question, but how does MTE play along with shared
> >> anonymous pages?
> >
> > It should work fine. shmem_writepage() calls swap_writepage() which
> > calls arch_prepare_to_swap() to write the tags. And
> > shmem_swapin_folio() has a call to arch_swap_restore() to restore
> > them.
>
> Sorry, I meant actual anonymous memory pages, not shmem. Like, anonymous
> pages that are COW-shared due to fork() or KSM.
>
> How does MTE, in general, interact with that? Assume one process ends up
> modifying the tags ... and the page is COW-shared with a different
> process that should not observe these tag modifications.

Tag modifications cause write faults if the page is read-only, so for
COW shared pages we would end up copying the page in the usual way,
which on arm64 would copy the tags as well via the copy_highpage hook
(see arch/arm64/mm/copypage.c).

Peter