lkml.org 
[lkml]   [2024]   [Apr]   [18]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
Date
SubjectRe: [PATCH v9 4/4] mm/madvise: optimize lazyfreeing with mTHP in madvise_free
Hey David,

Thanks a lot for taking time to review!

On Thu, Apr 18, 2024 at 8:03 PM David Hildenbrand <david@redhat.com> wrote:
>
> On 18.04.24 12:57, Lance Yang wrote:
> > This patch optimizes lazyfreeing with PTE-mapped mTHP[1]
> > (Inspired by David Hildenbrand[2]). We aim to avoid unnecessary folio
> > splitting if the large folio is fully mapped within the target range.
> >
> > If a large folio is locked or shared, or if we fail to split it, we just
> > leave it in place and advance to the next PTE in the range. But note that
> > the behavior is changed; previously, any failure of this sort would cause
> > the entire operation to give up. As large folios become more common,
> > sticking to the old way could result in wasted opportunities.
> >
> > On an Intel I5 CPU, lazyfreeing a 1GiB VMA backed by PTE-mapped folios of
> > the same size results in the following runtimes for madvise(MADV_FREE) in
> > seconds (shorter is better):
> >
> > Folio Size | Old | New | Change
> > ------------------------------------------
> > 4KiB | 0.590251 | 0.590259 | 0%
> > 16KiB | 2.990447 | 0.185655 | -94%
> > 32KiB | 2.547831 | 0.104870 | -95%
> > 64KiB | 2.457796 | 0.052812 | -97%
> > 128KiB | 2.281034 | 0.032777 | -99%
> > 256KiB | 2.230387 | 0.017496 | -99%
> > 512KiB | 2.189106 | 0.010781 | -99%
> > 1024KiB | 2.183949 | 0.007753 | -99%
> > 2048KiB | 0.002799 | 0.002804 | 0%
> >
> > [1] https://lkml.kernel.org/r/20231207161211.2374093-5-ryan.roberts@armcom
> > [2] https://lore.kernel.org/linux-mm/20240214204435.167852-1-david@redhat.com
> >
> > Reviewed-by: Ryan Roberts <ryan.roberts@arm.com>
> > Signed-off-by: Lance Yang <ioworker0@gmail.com>
> > ---
> > mm/madvise.c | 85 +++++++++++++++++++++++++++-------------------------
> > 1 file changed, 44 insertions(+), 41 deletions(-)
> >
> > diff --git a/mm/madvise.c b/mm/madvise.c
> > index 4597a3568e7e..375ab3234603 100644
> > --- a/mm/madvise.c
> > +++ b/mm/madvise.c
> > @@ -643,6 +643,7 @@ static int madvise_free_pte_range(pmd_t *pmd, unsigned long addr,
> > unsigned long end, struct mm_walk *walk)
> >
> > {
> > + const cydp_t cydp_flags = CYDP_CLEAR_YOUNG | CYDP_CLEAR_DIRTY;
> > struct mmu_gather *tlb = walk->private;
> > struct mm_struct *mm = tlb->mm;
> > struct vm_area_struct *vma = walk->vma;
> > @@ -697,44 +698,57 @@ static int madvise_free_pte_range(pmd_t *pmd, unsigned long addr,
> > continue;
> >
> > /*
> > - * If pmd isn't transhuge but the folio is large and
> > - * is owned by only this process, split it and
> > - * deactivate all pages.
> > + * If we encounter a large folio, only split it if it is not
> > + * fully mapped within the range we are operating on. Otherwise
> > + * leave it as is so that it can be marked as lazyfree. If we
> > + * fail to split a folio, leave it in place and advance to the
> > + * next pte in the range.
> > */
> > if (folio_test_large(folio)) {
> > - int err;
> > + bool any_young, any_dirty;
> >
> > - if (folio_likely_mapped_shared(folio))
> > - break;
> > - if (!folio_trylock(folio))
> > - break;
> > - folio_get(folio);
> > - arch_leave_lazy_mmu_mode();
> > - pte_unmap_unlock(start_pte, ptl);
> > - start_pte = NULL;
> > - err = split_folio(folio);
> > - folio_unlock(folio);
> > - folio_put(folio);
> > - if (err)
> > - break;
> > - start_pte = pte =
> > - pte_offset_map_lock(mm, pmd, addr, &ptl);
> > - if (!start_pte)
> > - break;
> > - arch_enter_lazy_mmu_mode();
> > - pte--;
> > - addr -= PAGE_SIZE;
> > - continue;
> > + nr = madvise_folio_pte_batch(addr, end, folio, pte,
> > + ptent, &any_young, NULL);
> > +
> > + if (nr < folio_nr_pages(folio)) {
> > + int err;
> > +
> > + if (folio_likely_mapped_shared(folio))
> > + continue;
> > + if (!folio_trylock(folio))
> > + continue;
> > + folio_get(folio);
> > + arch_leave_lazy_mmu_mode();
> > + pte_unmap_unlock(start_pte, ptl);
> > + start_pte = NULL;
> > + err = split_folio(folio);
> > + folio_unlock(folio);
> > + folio_put(folio);
> > + start_pte = pte =
> > + pte_offset_map_lock(mm, pmd, addr, &ptl);
>
> I'd just put it on a single line.

Got it.

>
> > + if (!start_pte)
> > + break;
> > + arch_enter_lazy_mmu_mode();
> > + if (!err)
> > + nr = 0;
> > + continue;
> > + }
> > +
> > + if (any_young)
> > + ptent = pte_mkyoung(ptent);
> > + if (any_dirty)
>
> any_dirty is never set, likely missed to pass it to
> madvise_folio_pte_batch().

Good spot! My bad for the mistake - sorry :(

>
> Apart from that LGTM and this patch is much easier to review now!
>
>
> With above:
>
> Acked-by: David Hildenbrand <david@redhat.com>

Thanks again for the review!
Lance

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

\
 
 \ /
  Last update: 2024-04-18 14:17    [W:0.048 / U:0.388 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site