Received-SPF: pass (google.com: domain of linux-kernel+bounces-196764-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:40f1:3f00::1 as permitted sender) client-ip=2604:1380:40f1:3f00::1;
Precedence: bulk
MIME-Version: 1.0
References: <20240531104819.140218-1-21cnbao@gmail.com> <87ac9610-5650-451f-aa54-e634a6310af4@redhat.com>
 <CAGsJ_4zgf8j0oRnVBhe-__=K2RFDHTCo-JnMak95HTvxtMUwnQ@mail.gmail.com> <d4c1a9ad-4945-40d7-9b7a-5b02df805884@redhat.com>
In-Reply-To: <d4c1a9ad-4945-40d7-9b7a-5b02df805884@redhat.com>
From: Barry Song <21cnbao@gmail.com>
Date: Sat, 1 Jun 2024 00:20:13 +1200
Message-ID: <CAGsJ_4xpDwTTbwzMx8ipmpCyNmVmABpRN1f9yfZfFiOaGMKiew@mail.gmail.com>
Subject: Re: [RFC PATCH] mm: swap: reuse exclusive folio directly instead of
 wp page faults
To: David Hildenbrand <david@redhat.com>
Cc: akpm@linux-foundation.org, linux-mm@kvack.org, chrisl@kernel.org, 
	surenb@google.com, kasong@tencent.com, minchan@kernel.org, 
	willy@infradead.org, ryan.roberts@arm.com, linux-kernel@vger.kernel.org, 
	Barry Song <v-songbaohua@oppo.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Sat, Jun 1, 2024 at 12:10=E2=80=AFAM David Hildenbrand <david@redhat.com=
> wrote:
>
> On 31.05.24 13:55, Barry Song wrote:
> > On Fri, May 31, 2024 at 11:08=E2=80=AFPM David Hildenbrand <david@redha=
t.com> wrote:
> >>
> >> On 31.05.24 12:48, Barry Song wrote:
> >>> From: Barry Song <v-songbaohua@oppo.com>
> >>>
> >>> After swapping out, we perform a swap-in operation. If we first read
> >>> and then write, we encounter a major fault in do_swap_page for readin=
g,
> >>> along with additional minor faults in do_wp_page for writing. However=
,
> >>> the latter appears to be unnecessary and inefficient. Instead, we can
> >>> directly reuse in do_swap_page and completely eliminate the need for
> >>> do_wp_page.
> >>>
> >>> This patch achieves that optimization specifically for exclusive foli=
os.
> >>> The following microbenchmark demonstrates the significant reduction i=
n
> >>> minor faults.
> >>>
> >>>    #define DATA_SIZE (2UL * 1024 * 1024)
> >>>    #define PAGE_SIZE (4UL * 1024)
> >>>
> >>>    static void *read_write_data(char *addr)
> >>>    {
> >>>            char tmp;
> >>>
> >>>            for (int i =3D 0; i < DATA_SIZE; i +=3D PAGE_SIZE) {
> >>>                    tmp =3D *(volatile char *)(addr + i);
> >>>                    *(volatile char *)(addr + i) =3D tmp;
> >>>            }
> >>>    }
> >>>
> >>>    int main(int argc, char **argv)
> >>>    {
> >>>            struct rusage ru;
> >>>
> >>>            char *addr =3D mmap(NULL, DATA_SIZE, PROT_READ | PROT_WRIT=
E,
> >>>                            MAP_ANONYMOUS | MAP_PRIVATE, -1, 0);
> >>>            memset(addr, 0x11, DATA_SIZE);
> >>>
> >>>            do {
> >>>                    long old_ru_minflt, old_ru_majflt;
> >>>                    long new_ru_minflt, new_ru_majflt;
> >>>
> >>>                    madvise(addr, DATA_SIZE, MADV_PAGEOUT);
> >>>
> >>>                    getrusage(RUSAGE_SELF, &ru);
> >>>                    old_ru_minflt =3D ru.ru_minflt;
> >>>                    old_ru_majflt =3D ru.ru_majflt;
> >>>
> >>>                    read_write_data(addr);
> >>>                    getrusage(RUSAGE_SELF, &ru);
> >>>                    new_ru_minflt =3D ru.ru_minflt;
> >>>                    new_ru_majflt =3D ru.ru_majflt;
> >>>
> >>>                    printf("minor faults:%ld major faults:%ld\n",
> >>>                            new_ru_minflt - old_ru_minflt,
> >>>                            new_ru_majflt - old_ru_majflt);
> >>>            } while(0);
> >>>
> >>>            return 0;
> >>>    }
> >>>
> >>> w/o patch,
> >>> / # ~/a.out
> >>> minor faults:512 major faults:512
> >>>
> >>> w/ patch,
> >>> / # ~/a.out
> >>> minor faults:0 major faults:512
> >>>
> >>> Minor faults decrease to 0!
> >>>
> >>> Signed-off-by: Barry Song <v-songbaohua@oppo.com>
> >>> ---
> >>>    mm/memory.c | 7 ++++---
> >>>    1 file changed, 4 insertions(+), 3 deletions(-)
> >>>
> >>> diff --git a/mm/memory.c b/mm/memory.c
> >>> index eef4e482c0c2..e1d2e339958e 100644
> >>> --- a/mm/memory.c
> >>> +++ b/mm/memory.c
> >>> @@ -4325,9 +4325,10 @@ vm_fault_t do_swap_page(struct vm_fault *vmf)
> >>>         */
> >>>        if (!folio_test_ksm(folio) &&
> >>>            (exclusive || folio_ref_count(folio) =3D=3D 1)) {
> >>> -             if (vmf->flags & FAULT_FLAG_WRITE) {
> >>> -                     pte =3D maybe_mkwrite(pte_mkdirty(pte), vma);
> >>> -                     vmf->flags &=3D ~FAULT_FLAG_WRITE;
> >>> +             if (vma->vm_flags & VM_WRITE) {
> >>> +                     pte =3D pte_mkwrite(pte_mkdirty(pte), vma);
> >>> +                     if (vmf->flags & FAULT_FLAG_WRITE)
> >>> +                             vmf->flags &=3D ~FAULT_FLAG_WRITE;
> >>
> >> This implies, that even on a read fault, you would mark the pte dirty
> >> and it would have to be written back to swap if still in the swap cach=
e
> >> and only read.
> >>
> >> That is controversial.
> >>
> >> What is less controversial is doing what mprotect() via
> >> change_pte_range()/can_change_pte_writable() would do: mark the PTE
> >> writable but not dirty.
> >>
> >> I suggest setting the pte only dirty if FAULT_FLAG_WRITE is set.
> >
> > Thanks!
> >
> > I assume you mean something as below?
>
> It raises an important point: uffd-wp must be handled accordingly.
>
> >
> > diff --git a/mm/memory.c b/mm/memory.c
> > index eef4e482c0c2..dbf1ba8ccfd6 100644
> > --- a/mm/memory.c
> > +++ b/mm/memory.c
> > @@ -4317,6 +4317,10 @@ vm_fault_t do_swap_page(struct vm_fault *vmf)
> >          add_mm_counter(vma->vm_mm, MM_SWAPENTS, -nr_pages);
> >          pte =3D mk_pte(page, vma->vm_page_prot);
> >
> > +       if (pte_swp_soft_dirty(vmf->orig_pte))
> > +               pte =3D pte_mksoft_dirty(pte);
> > +       if (pte_swp_uffd_wp(vmf->orig_pte))
> > +               pte =3D pte_mkuffd_wp(pte);
> >          /*
> >           * Same logic as in do_wp_page(); however, optimize for pages =
that are
> >           * certainly not shared either because we just allocated them =
without
> > @@ -4325,18 +4329,19 @@ vm_fault_t do_swap_page(struct vm_fault *vmf)
> >           */
> >          if (!folio_test_ksm(folio) &&
> >              (exclusive || folio_ref_count(folio) =3D=3D 1)) {
> > -               if (vmf->flags & FAULT_FLAG_WRITE) {
> > -                       pte =3D maybe_mkwrite(pte_mkdirty(pte), vma);
> > -                       vmf->flags &=3D ~FAULT_FLAG_WRITE;
> > +               if (vma->vm_flags & VM_WRITE) {
> > +                       if (vmf->flags & FAULT_FLAG_WRITE) {
> > +                               pte =3D pte_mkwrite(pte_mkdirty(pte), v=
ma);
> > +                               vmf->flags &=3D ~FAULT_FLAG_WRITE;
> > +                       } else if ((!vma_soft_dirty_enabled(vma) ||
> > pte_soft_dirty(pte))
> > +                                   && !userfaultfd_pte_wp(vma, pte)) {
> > +                                       pte =3D pte_mkwrite(pte, vma);
>
> Even with FAULT_FLAG_WRITE we must respect uffd-wp and *not* do a
> pte_mkwrite(pte). So we have to catch and handle that earlier (I could
> have sworn we handle that somehow).
>
> Note that the existing
>         pte =3D pte_mkuffd_wp(pte);
>
> Will fix that up because it does an implicit pte_wrprotect().

This is exactly what I have missed as I am struggling with why WRITE_FAULT
blindly does mkwrite without checking userfaultfd_pte_wp().

>
>
> So maybe what would work is
>
>
> if ((vma->vm_flags & VM_WRITE) && !userfaultfd_pte_wp(vma, pte) &&
>      !vma_soft_dirty_enabled(vma)) {
>         pte =3D pte_mkwrite(pte);
>
>         /* Only set the PTE dirty on write fault. */
>         if (vmf->flags & FAULT_FLAG_WRITE) {
>                 pte =3D pte_mkdirty(pte);
>                 vmf->flags &=3D ~FAULT_FLAG_WRITE;
>         }
> }
>

looks good!

> --
> Cheers,
>
> David / dhildenb
>

Thanks
Barry