Hi Greg,
On Fri, Oct 18, 2024 at 1:29 AM Greg KH <gregkh@xxxxxxxxxxxxxxxxxxx> wrote:
>
> On Thu, Oct 17, 2024 at 02:58:03PM -0700, chrisl@xxxxxxxxxx wrote:
> > From: Yu Zhao <yuzhao@xxxxxxxxxx>
> >
> > [ Upstream commit 95599ef684d01136a8b77c16a7c853496786e173 ]
> >
> > The current assumption is that a large folio can only be split into
> > order-0 folios. That is not the case for hugeTLB demotion, nor for THP
> > split: see commit c010d47f107f ("mm: thp: split huge page to any lower
> > order pages").
> >
> > When a large folio is split into ones of a lower non-zero order, only the
> > new head pages should be tagged. Tagging tail pages can cause imbalanced
> > "calls" counters, since only head pages are untagged by pgalloc_tag_sub()
> > and the "calls" counts on tail pages are leaked, e.g.,
> >
> > # echo 2048kB >/sys/kernel/mm/hugepages/hugepages-1048576kB/demote_size
> > # echo 700 >/sys/kernel/mm/hugepages/hugepages-1048576kB/nr_hugepages
> > # time echo 700 >/sys/kernel/mm/hugepages/hugepages-1048576kB/demote
> > # echo 0 >/sys/kernel/mm/hugepages/hugepages-2048kB/nr_hugepages
> > # grep alloc_gigantic_folio /proc/allocinfo
> >
> > Before this patch:
> > 0 549427200 mm/hugetlb.c:1549 func:alloc_gigantic_folio
> >
> > real 0m2.057s
> > user 0m0.000s
> > sys 0m2.051s
> >
> > After this patch:
> > 0 0 mm/hugetlb.c:1549 func:alloc_gigantic_folio
> >
> > real 0m1.711s
> > user 0m0.000s
> > sys 0m1.704s
> >
> > Not tagging tail pages also improves the splitting time, e.g., by about
> > 15% when demoting 1GB hugeTLB folios to 2MB ones, as shown above.
> >
> > Link: https://lkml.kernel.org/r/20240906042108.1150526-2-yuzhao@xxxxxxxxxx
> > Fixes: be25d1d4e822 ("mm: create new codetag references during page splitting")
> > Signed-off-by: Yu Zhao <yuzhao@xxxxxxxxxx>
> > Acked-by: Suren Baghdasaryan <surenb@xxxxxxxxxx>
> > Cc: Kent Overstreet <kent.overstreet@xxxxxxxxx>
> > Cc: Muchun Song <muchun.song@xxxxxxxxx>
> > Cc: <stable@xxxxxxxxxxxxxxx>
> > Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
>
> You did not sign off on this backport, so even if I wanted to take it, I
> couldn't :(
My bad, sorry my first attempt back port some patches for someone else.
> Please fix this, and patch 3/3 up, and just send those.
I will add sign off to all 3 patches and send out the v2. I haven't
found an easy way to skip the cover letter in b4 yet. It might be
easier just to refresh the whole series.
Chris
