2023-08-09 08:49:37

by Charan Teja Kalla

Subject: [PATCH] Multi-gen LRU: skip CMA pages when they are not eligible

This patch is based on the commit 5da226dbfce3("mm: skip CMA pages when
they are not available") which skips cma pages reclaim when they are not
eligible for the current allocation context. In mglru, such pages are
added to the tail of the immediate generation to maintain better LRU
order, which is unlike the case of conventional LRU where such pages are
directly added to the head of the LRU list(akin to adding to head of the
youngest generation in mglru).

No observable issue without this patch on MGLRU, but logically it make
sense to skip the CMA page reclaim when those pages can't be satisfied
for the current allocation context.

Signed-off-by: Charan Teja Kalla <[email protected]>
---
mm/vmscan.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/mm/vmscan.c b/mm/vmscan.c
index b4329f9..6cbe921 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -4943,7 +4943,7 @@ static bool sort_folio(struct lruvec *lruvec, struct folio *folio, struct scan_c
}

/* ineligible */
- if (zone > sc->reclaim_idx) {
+ if (zone > sc->reclaim_idx || skip_cma(folio, sc)) {
gen = folio_inc_gen(lruvec, folio, false);
list_move_tail(&folio->lru, &lrugen->folios[gen][type][zone]);
return true;
--
2.7.4

2023-08-09 15:27:19

by David Hildenbrand

Subject: Re: [PATCH] Multi-gen LRU: skip CMA pages when they are not eligible

On 09.08.23 10:05, Charan Teja Kalla wrote:
> This patch is based on the commit 5da226dbfce3("mm: skip CMA pages when
> they are not available") which skips cma pages reclaim when they are not
> eligible for the current allocation context. In mglru, such pages are
> added to the tail of the immediate generation to maintain better LRU
> order, which is unlike the case of conventional LRU where such pages are
> directly added to the head of the LRU list(akin to adding to head of the
> youngest generation in mglru).
>
> No observable issue without this patch on MGLRU, but logically it make
> sense to skip the CMA page reclaim when those pages can't be satisfied
> for the current allocation context.
>
> Signed-off-by: Charan Teja Kalla <[email protected]>
> ---
> mm/vmscan.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/mm/vmscan.c b/mm/vmscan.c
> index b4329f9..6cbe921 100644
> --- a/mm/vmscan.c
> +++ b/mm/vmscan.c
> @@ -4943,7 +4943,7 @@ static bool sort_folio(struct lruvec *lruvec, struct folio *folio, struct scan_c
> }
>
> /* ineligible */
> - if (zone > sc->reclaim_idx) {
> + if (zone > sc->reclaim_idx || skip_cma(folio, sc)) {
> gen = folio_inc_gen(lruvec, folio, false);
> list_move_tail(&folio->lru, &lrugen->folios[gen][type][zone]);
> return true;

Makes sense to me.

--
Cheers,

David / dhildenb

2023-08-09 16:28:30

by Kalesh Singh

Subject: Re: [PATCH] Multi-gen LRU: skip CMA pages when they are not eligible

On Wed, Aug 9, 2023 at 7:00 AM David Hildenbrand <[email protected]> wrote:
>
> On 09.08.23 10:05, Charan Teja Kalla wrote:
> > This patch is based on the commit 5da226dbfce3("mm: skip CMA pages when
> > they are not available") which skips cma pages reclaim when they are not
> > eligible for the current allocation context. In mglru, such pages are
> > added to the tail of the immediate generation to maintain better LRU
> > order, which is unlike the case of conventional LRU where such pages are
> > directly added to the head of the LRU list(akin to adding to head of the
> > youngest generation in mglru).
> >
> > No observable issue without this patch on MGLRU, but logically it make
> > sense to skip the CMA page reclaim when those pages can't be satisfied
> > for the current allocation context.
> >
> > Signed-off-by: Charan Teja Kalla <[email protected]>

Reviewed-by: Kalesh Singh <[email protected]>

Thanks,
Kalesh

> > ---
> > mm/vmscan.c | 2 +-
> > 1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/mm/vmscan.c b/mm/vmscan.c
> > index b4329f9..6cbe921 100644
> > --- a/mm/vmscan.c
> > +++ b/mm/vmscan.c
> > @@ -4943,7 +4943,7 @@ static bool sort_folio(struct lruvec *lruvec, struct folio *folio, struct scan_c
> > }
> >
> > /* ineligible */
> > - if (zone > sc->reclaim_idx) {
> > + if (zone > sc->reclaim_idx || skip_cma(folio, sc)) {
> > gen = folio_inc_gen(lruvec, folio, false);
> > list_move_tail(&folio->lru, &lrugen->folios[gen][type][zone]);
> > return true;
>
> Makes sense to me.
>
> --
> Cheers,
>
> David / dhildenb
>

2023-08-09 19:12:34

by Yu Zhao

Subject: Re: [PATCH] Multi-gen LRU: skip CMA pages when they are not eligible

On Wed, Aug 9, 2023 at 9:54 AM Kalesh Singh <[email protected]> wrote:
>
> On Wed, Aug 9, 2023 at 7:00 AM David Hildenbrand <[email protected]> wrote:
> >
> > On 09.08.23 10:05, Charan Teja Kalla wrote:
> > > This patch is based on the commit 5da226dbfce3("mm: skip CMA pages when
> > > they are not available") which skips cma pages reclaim when they are not
> > > eligible for the current allocation context. In mglru, such pages are
> > > added to the tail of the immediate generation to maintain better LRU
> > > order, which is unlike the case of conventional LRU where such pages are
> > > directly added to the head of the LRU list(akin to adding to head of the
> > > youngest generation in mglru).
> > >
> > > No observable issue without this patch on MGLRU, but logically it make
> > > sense to skip the CMA page reclaim when those pages can't be satisfied
> > > for the current allocation context.
> > >
> > > Signed-off-by: Charan Teja Kalla <[email protected]>
>
> Reviewed-by: Kalesh Singh <[email protected]>

Thanks, Charan! Do we need a "Fixes" tag?

> > > ---
> > > mm/vmscan.c | 2 +-
> > > 1 file changed, 1 insertion(+), 1 deletion(-)
> > >
> > > diff --git a/mm/vmscan.c b/mm/vmscan.c
> > > index b4329f9..6cbe921 100644
> > > --- a/mm/vmscan.c
> > > +++ b/mm/vmscan.c
> > > @@ -4943,7 +4943,7 @@ static bool sort_folio(struct lruvec *lruvec, struct folio *folio, struct scan_c
> > > }
> > >
> > > /* ineligible */
> > > - if (zone > sc->reclaim_idx) {
> > > + if (zone > sc->reclaim_idx || skip_cma(folio, sc)) {
> > > gen = folio_inc_gen(lruvec, folio, false);
> > > list_move_tail(&folio->lru, &lrugen->folios[gen][type][zone]);
> > > return true;
> >
> > Makes sense to me.

2023-08-12 05:38:29

by Charan Teja Kalla

Subject: Re: [PATCH] Multi-gen LRU: skip CMA pages when they are not eligible

On 8/9/2023 10:39 PM, Yu Zhao wrote:
>> On Wed, Aug 9, 2023 at 7:00 AM David Hildenbrand <[email protected]> wrote:
>>> On 09.08.23 10:05, Charan Teja Kalla wrote:
>>>> This patch is based on the commit 5da226dbfce3("mm: skip CMA pages when
>>>> they are not available") which skips cma pages reclaim when they are not
>>>> eligible for the current allocation context. In mglru, such pages are
>>>> added to the tail of the immediate generation to maintain better LRU
>>>> order, which is unlike the case of conventional LRU where such pages are
>>>> directly added to the head of the LRU list(akin to adding to head of the
>>>> youngest generation in mglru).
>>>>
>>>> No observable issue without this patch on MGLRU, but logically it make
>>>> sense to skip the CMA page reclaim when those pages can't be satisfied
>>>> for the current allocation context.
>>>>
>>>> Signed-off-by: Charan Teja Kalla <[email protected]>
>> Reviewed-by: Kalesh Singh <[email protected]>
> Thanks, Charan! Do we need a "Fixes" tag?
I think the below tag matches.

Fixes: ac35a4902374 ("mm: multi-gen LRU: minimal implementation")
>