diff mbox series

[v5,5/6] mm: vmscan: Avoid split during shrink_folio_list()

Message ID 20240327144537.4165578-6-ryan.roberts@arm.com (mailing list archive)
State New
Headers show
Series Swap-out mTHP without splitting | expand

Commit Message

Ryan Roberts March 27, 2024, 2:45 p.m. UTC
Now that swap supports storing all mTHP sizes, avoid splitting large
folios before swap-out. This benefits performance of the swap-out path
by eliding split_folio_to_list(), which is expensive, and also sets us
up for swapping in large folios in a future series.

If the folio is partially mapped, we continue to split it since we want
to avoid the extra IO overhead and storage of writing out pages
uneccessarily.

Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Barry Song <v-songbaohua@oppo.com>
Signed-off-by: Ryan Roberts <ryan.roberts@arm.com>
---
 mm/vmscan.c | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

Comments

Barry Song March 28, 2024, 8:18 a.m. UTC | #1
On Thu, Mar 28, 2024 at 3:45 AM Ryan Roberts <ryan.roberts@arm.com> wrote:
>
> Now that swap supports storing all mTHP sizes, avoid splitting large
> folios before swap-out. This benefits performance of the swap-out path
> by eliding split_folio_to_list(), which is expensive, and also sets us
> up for swapping in large folios in a future series.
>
> If the folio is partially mapped, we continue to split it since we want
> to avoid the extra IO overhead and storage of writing out pages
> uneccessarily.
>
> Reviewed-by: David Hildenbrand <david@redhat.com>
> Reviewed-by: Barry Song <v-songbaohua@oppo.com>
> Signed-off-by: Ryan Roberts <ryan.roberts@arm.com>
> ---
>  mm/vmscan.c | 9 +++++----
>  1 file changed, 5 insertions(+), 4 deletions(-)
>
> diff --git a/mm/vmscan.c b/mm/vmscan.c
> index 00adaf1cb2c3..293120fe54f3 100644
> --- a/mm/vmscan.c
> +++ b/mm/vmscan.c
> @@ -1223,11 +1223,12 @@ static unsigned int shrink_folio_list(struct list_head *folio_list,
>                                         if (!can_split_folio(folio, NULL))
>                                                 goto activate_locked;
>                                         /*
> -                                        * Split folios without a PMD map right
> -                                        * away. Chances are some or all of the
> -                                        * tail pages can be freed without IO.
> +                                        * Split partially mapped folios right
> +                                        * away. We can free the unmapped pages
> +                                        * without IO.
>                                          */
> -                                       if (!folio_entire_mapcount(folio) &&
> +                                       if (data_race(!list_empty(
> +                                               &folio->_deferred_list)) &&
>                                             split_folio_to_list(folio,
>                                                                 folio_list))
>                                                 goto activate_locked;

Hi Ryan,

Sorry for bringing up another minor issue at this late stage.

During the debugging of thp counter patch v2, I noticed the discrepancy between
THP_SWPOUT_FALLBACK and THP_SWPOUT.

Should we make adjustments to the counter?

diff --git a/mm/vmscan.c b/mm/vmscan.c
index 293120fe54f3..d7856603f689 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -1241,8 +1241,10 @@ static unsigned int shrink_folio_list(struct
list_head *folio_list,
                                                                folio_list))
                                                goto activate_locked;
 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
-
count_memcg_folio_events(folio, THP_SWPOUT_FALLBACK, 1);
-                                       count_vm_event(THP_SWPOUT_FALLBACK);
+                                       if (folio_test_pmd_mappable(folio)) {
+
count_memcg_folio_events(folio, THP_SWPOUT_FALLBACK, 1);
+
count_vm_event(THP_SWPOUT_FALLBACK);
+                                       }
 #endif
                                        if (!add_to_swap(folio))
                                                goto activate_locked_split;


Because THP_SWPOUT is only for pmd:

static inline void count_swpout_vm_event(struct folio *folio)
{
#ifdef CONFIG_TRANSPARENT_HUGEPAGE
        if (unlikely(folio_test_pmd_mappable(folio))) {
                count_memcg_folio_events(folio, THP_SWPOUT, 1);
                count_vm_event(THP_SWPOUT);
        }
#endif
        count_vm_events(PSWPOUT, folio_nr_pages(folio));
}

I can provide per-order counters for this in my THP counter patch.

> --
> 2.25.1
>

Thanks
Barry
Ryan Roberts March 28, 2024, 8:48 a.m. UTC | #2
On 28/03/2024 08:18, Barry Song wrote:
> On Thu, Mar 28, 2024 at 3:45 AM Ryan Roberts <ryan.roberts@arm.com> wrote:
>>
>> Now that swap supports storing all mTHP sizes, avoid splitting large
>> folios before swap-out. This benefits performance of the swap-out path
>> by eliding split_folio_to_list(), which is expensive, and also sets us
>> up for swapping in large folios in a future series.
>>
>> If the folio is partially mapped, we continue to split it since we want
>> to avoid the extra IO overhead and storage of writing out pages
>> uneccessarily.
>>
>> Reviewed-by: David Hildenbrand <david@redhat.com>
>> Reviewed-by: Barry Song <v-songbaohua@oppo.com>
>> Signed-off-by: Ryan Roberts <ryan.roberts@arm.com>
>> ---
>>  mm/vmscan.c | 9 +++++----
>>  1 file changed, 5 insertions(+), 4 deletions(-)
>>
>> diff --git a/mm/vmscan.c b/mm/vmscan.c
>> index 00adaf1cb2c3..293120fe54f3 100644
>> --- a/mm/vmscan.c
>> +++ b/mm/vmscan.c
>> @@ -1223,11 +1223,12 @@ static unsigned int shrink_folio_list(struct list_head *folio_list,
>>                                         if (!can_split_folio(folio, NULL))
>>                                                 goto activate_locked;
>>                                         /*
>> -                                        * Split folios without a PMD map right
>> -                                        * away. Chances are some or all of the
>> -                                        * tail pages can be freed without IO.
>> +                                        * Split partially mapped folios right
>> +                                        * away. We can free the unmapped pages
>> +                                        * without IO.
>>                                          */
>> -                                       if (!folio_entire_mapcount(folio) &&
>> +                                       if (data_race(!list_empty(
>> +                                               &folio->_deferred_list)) &&
>>                                             split_folio_to_list(folio,
>>                                                                 folio_list))
>>                                                 goto activate_locked;
> 
> Hi Ryan,
> 
> Sorry for bringing up another minor issue at this late stage.
> 
> During the debugging of thp counter patch v2, I noticed the discrepancy between
> THP_SWPOUT_FALLBACK and THP_SWPOUT.

Ahh good spot! I had noticed this previously and clearly forgot all about it.

I'm on holiday today and over the long weekend in the UK. I'll take a proper
look next week and send a fix.

> 
> Should we make adjustments to the counter?
> 
> diff --git a/mm/vmscan.c b/mm/vmscan.c
> index 293120fe54f3..d7856603f689 100644
> --- a/mm/vmscan.c
> +++ b/mm/vmscan.c
> @@ -1241,8 +1241,10 @@ static unsigned int shrink_folio_list(struct
> list_head *folio_list,
>                                                                 folio_list))
>                                                 goto activate_locked;
>  #ifdef CONFIG_TRANSPARENT_HUGEPAGE
> -
> count_memcg_folio_events(folio, THP_SWPOUT_FALLBACK, 1);
> -                                       count_vm_event(THP_SWPOUT_FALLBACK);
> +                                       if (folio_test_pmd_mappable(folio)) {
> +
> count_memcg_folio_events(folio, THP_SWPOUT_FALLBACK, 1);
> +
> count_vm_event(THP_SWPOUT_FALLBACK);
> +                                       }
>  #endif
>                                         if (!add_to_swap(folio))
>                                                 goto activate_locked_split;
> 
> 
> Because THP_SWPOUT is only for pmd:
> 
> static inline void count_swpout_vm_event(struct folio *folio)
> {
> #ifdef CONFIG_TRANSPARENT_HUGEPAGE
>         if (unlikely(folio_test_pmd_mappable(folio))) {
>                 count_memcg_folio_events(folio, THP_SWPOUT, 1);
>                 count_vm_event(THP_SWPOUT);
>         }
> #endif
>         count_vm_events(PSWPOUT, folio_nr_pages(folio));
> }
> 
> I can provide per-order counters for this in my THP counter patch.
> 
>> --
>> 2.25.1
>>
> 
> Thanks
> Barry
Ryan Roberts April 2, 2024, 1:10 p.m. UTC | #3
On 28/03/2024 08:18, Barry Song wrote:
> On Thu, Mar 28, 2024 at 3:45 AM Ryan Roberts <ryan.roberts@arm.com> wrote:
>>
>> Now that swap supports storing all mTHP sizes, avoid splitting large
>> folios before swap-out. This benefits performance of the swap-out path
>> by eliding split_folio_to_list(), which is expensive, and also sets us
>> up for swapping in large folios in a future series.
>>
>> If the folio is partially mapped, we continue to split it since we want
>> to avoid the extra IO overhead and storage of writing out pages
>> uneccessarily.
>>
>> Reviewed-by: David Hildenbrand <david@redhat.com>
>> Reviewed-by: Barry Song <v-songbaohua@oppo.com>
>> Signed-off-by: Ryan Roberts <ryan.roberts@arm.com>
>> ---
>>  mm/vmscan.c | 9 +++++----
>>  1 file changed, 5 insertions(+), 4 deletions(-)
>>
>> diff --git a/mm/vmscan.c b/mm/vmscan.c
>> index 00adaf1cb2c3..293120fe54f3 100644
>> --- a/mm/vmscan.c
>> +++ b/mm/vmscan.c
>> @@ -1223,11 +1223,12 @@ static unsigned int shrink_folio_list(struct list_head *folio_list,
>>                                         if (!can_split_folio(folio, NULL))
>>                                                 goto activate_locked;
>>                                         /*
>> -                                        * Split folios without a PMD map right
>> -                                        * away. Chances are some or all of the
>> -                                        * tail pages can be freed without IO.
>> +                                        * Split partially mapped folios right
>> +                                        * away. We can free the unmapped pages
>> +                                        * without IO.
>>                                          */
>> -                                       if (!folio_entire_mapcount(folio) &&
>> +                                       if (data_race(!list_empty(
>> +                                               &folio->_deferred_list)) &&
>>                                             split_folio_to_list(folio,
>>                                                                 folio_list))
>>                                                 goto activate_locked;
> 
> Hi Ryan,
> 
> Sorry for bringing up another minor issue at this late stage.

No problem - I'd rather take a bit longer and get it right, rather than rush it
and get it wrong!

> 
> During the debugging of thp counter patch v2, I noticed the discrepancy between
> THP_SWPOUT_FALLBACK and THP_SWPOUT.
> 
> Should we make adjustments to the counter?

Yes, agreed; we want to be consistent here with all the other existing THP
counters; they only refer to PMD-sized THP. I'll make the change for the next
version.

I guess we will eventually want equivalent counters for per-size mTHP using the
framework you are adding.

> 
> diff --git a/mm/vmscan.c b/mm/vmscan.c
> index 293120fe54f3..d7856603f689 100644
> --- a/mm/vmscan.c
> +++ b/mm/vmscan.c
> @@ -1241,8 +1241,10 @@ static unsigned int shrink_folio_list(struct
> list_head *folio_list,
>                                                                 folio_list))
>                                                 goto activate_locked;
>  #ifdef CONFIG_TRANSPARENT_HUGEPAGE
> -
> count_memcg_folio_events(folio, THP_SWPOUT_FALLBACK, 1);
> -                                       count_vm_event(THP_SWPOUT_FALLBACK);
> +                                       if (folio_test_pmd_mappable(folio)) {
> +
> count_memcg_folio_events(folio, THP_SWPOUT_FALLBACK, 1);
> +
> count_vm_event(THP_SWPOUT_FALLBACK);
> +                                       }
>  #endif
>                                         if (!add_to_swap(folio))
>                                                 goto activate_locked_split;
> 
> 
> Because THP_SWPOUT is only for pmd:
> 
> static inline void count_swpout_vm_event(struct folio *folio)
> {
> #ifdef CONFIG_TRANSPARENT_HUGEPAGE
>         if (unlikely(folio_test_pmd_mappable(folio))) {
>                 count_memcg_folio_events(folio, THP_SWPOUT, 1);
>                 count_vm_event(THP_SWPOUT);
>         }
> #endif
>         count_vm_events(PSWPOUT, folio_nr_pages(folio));
> }
> 
> I can provide per-order counters for this in my THP counter patch.
> 
>> --
>> 2.25.1
>>
> 
> Thanks
> Barry
Lance Yang April 2, 2024, 1:22 p.m. UTC | #4
Baolin Wang's patch[1] has avoided confusion with PMD mapped THP
related statistics.

So, these three counters (THP_SPLIT_PAGE, THP_SPLIT_PAGE_FAILED,
and THP_DEFERRED_SPLIT_PAGE) no longer include mTHP.

[1] https://lore.kernel.org/linux-mm/a5341defeef27c9ac7b85c97f030f93e4368bbc1.1711694852.git.baolin.wang@linux.alibaba.com/

On Tue, Apr 2, 2024 at 9:10 PM Ryan Roberts <ryan.roberts@arm.com> wrote:
>
> On 28/03/2024 08:18, Barry Song wrote:
> > On Thu, Mar 28, 2024 at 3:45 AM Ryan Roberts <ryan.roberts@arm.com> wrote:
> >>
> >> Now that swap supports storing all mTHP sizes, avoid splitting large
> >> folios before swap-out. This benefits performance of the swap-out path
> >> by eliding split_folio_to_list(), which is expensive, and also sets us
> >> up for swapping in large folios in a future series.
> >>
> >> If the folio is partially mapped, we continue to split it since we want
> >> to avoid the extra IO overhead and storage of writing out pages
> >> uneccessarily.
> >>
> >> Reviewed-by: David Hildenbrand <david@redhat.com>
> >> Reviewed-by: Barry Song <v-songbaohua@oppo.com>
> >> Signed-off-by: Ryan Roberts <ryan.roberts@arm.com>
> >> ---
> >>  mm/vmscan.c | 9 +++++----
> >>  1 file changed, 5 insertions(+), 4 deletions(-)
> >>
> >> diff --git a/mm/vmscan.c b/mm/vmscan.c
> >> index 00adaf1cb2c3..293120fe54f3 100644
> >> --- a/mm/vmscan.c
> >> +++ b/mm/vmscan.c
> >> @@ -1223,11 +1223,12 @@ static unsigned int shrink_folio_list(struct list_head *folio_list,
> >>                                         if (!can_split_folio(folio, NULL))
> >>                                                 goto activate_locked;
> >>                                         /*
> >> -                                        * Split folios without a PMD map right
> >> -                                        * away. Chances are some or all of the
> >> -                                        * tail pages can be freed without IO.
> >> +                                        * Split partially mapped folios right
> >> +                                        * away. We can free the unmapped pages
> >> +                                        * without IO.
> >>                                          */
> >> -                                       if (!folio_entire_mapcount(folio) &&
> >> +                                       if (data_race(!list_empty(
> >> +                                               &folio->_deferred_list)) &&
> >>                                             split_folio_to_list(folio,
> >>                                                                 folio_list))
> >>                                                 goto activate_locked;
> >
> > Hi Ryan,
> >
> > Sorry for bringing up another minor issue at this late stage.
>
> No problem - I'd rather take a bit longer and get it right, rather than rush it
> and get it wrong!
>
> >
> > During the debugging of thp counter patch v2, I noticed the discrepancy between
> > THP_SWPOUT_FALLBACK and THP_SWPOUT.
> >
> > Should we make adjustments to the counter?
>
> Yes, agreed; we want to be consistent here with all the other existing THP
> counters; they only refer to PMD-sized THP. I'll make the change for the next
> version.
>
> I guess we will eventually want equivalent counters for per-size mTHP using the
> framework you are adding.
>
> >
> > diff --git a/mm/vmscan.c b/mm/vmscan.c
> > index 293120fe54f3..d7856603f689 100644
> > --- a/mm/vmscan.c
> > +++ b/mm/vmscan.c
> > @@ -1241,8 +1241,10 @@ static unsigned int shrink_folio_list(struct
> > list_head *folio_list,
> >                                                                 folio_list))
> >                                                 goto activate_locked;
> >  #ifdef CONFIG_TRANSPARENT_HUGEPAGE
> > -
> > count_memcg_folio_events(folio, THP_SWPOUT_FALLBACK, 1);
> > -                                       count_vm_event(THP_SWPOUT_FALLBACK);
> > +                                       if (folio_test_pmd_mappable(folio)) {
> > +
> > count_memcg_folio_events(folio, THP_SWPOUT_FALLBACK, 1);
> > +
> > count_vm_event(THP_SWPOUT_FALLBACK);
> > +                                       }
> >  #endif
> >                                         if (!add_to_swap(folio))
> >                                                 goto activate_locked_split;
> >
> >
> > Because THP_SWPOUT is only for pmd:
> >
> > static inline void count_swpout_vm_event(struct folio *folio)
> > {
> > #ifdef CONFIG_TRANSPARENT_HUGEPAGE
> >         if (unlikely(folio_test_pmd_mappable(folio))) {
> >                 count_memcg_folio_events(folio, THP_SWPOUT, 1);
> >                 count_vm_event(THP_SWPOUT);
> >         }
> > #endif
> >         count_vm_events(PSWPOUT, folio_nr_pages(folio));
> > }
> >
> > I can provide per-order counters for this in my THP counter patch.
> >
> >> --
> >> 2.25.1
> >>
> >
> > Thanks
> > Barry
>
Ryan Roberts April 2, 2024, 1:22 p.m. UTC | #5
On 02/04/2024 14:10, Ryan Roberts wrote:
> On 28/03/2024 08:18, Barry Song wrote:
>> On Thu, Mar 28, 2024 at 3:45 AM Ryan Roberts <ryan.roberts@arm.com> wrote:
>>>
>>> Now that swap supports storing all mTHP sizes, avoid splitting large
>>> folios before swap-out. This benefits performance of the swap-out path
>>> by eliding split_folio_to_list(), which is expensive, and also sets us
>>> up for swapping in large folios in a future series.
>>>
>>> If the folio is partially mapped, we continue to split it since we want
>>> to avoid the extra IO overhead and storage of writing out pages
>>> uneccessarily.
>>>
>>> Reviewed-by: David Hildenbrand <david@redhat.com>
>>> Reviewed-by: Barry Song <v-songbaohua@oppo.com>
>>> Signed-off-by: Ryan Roberts <ryan.roberts@arm.com>
>>> ---
>>>  mm/vmscan.c | 9 +++++----
>>>  1 file changed, 5 insertions(+), 4 deletions(-)
>>>
>>> diff --git a/mm/vmscan.c b/mm/vmscan.c
>>> index 00adaf1cb2c3..293120fe54f3 100644
>>> --- a/mm/vmscan.c
>>> +++ b/mm/vmscan.c
>>> @@ -1223,11 +1223,12 @@ static unsigned int shrink_folio_list(struct list_head *folio_list,
>>>                                         if (!can_split_folio(folio, NULL))
>>>                                                 goto activate_locked;
>>>                                         /*
>>> -                                        * Split folios without a PMD map right
>>> -                                        * away. Chances are some or all of the
>>> -                                        * tail pages can be freed without IO.
>>> +                                        * Split partially mapped folios right
>>> +                                        * away. We can free the unmapped pages
>>> +                                        * without IO.
>>>                                          */
>>> -                                       if (!folio_entire_mapcount(folio) &&
>>> +                                       if (data_race(!list_empty(
>>> +                                               &folio->_deferred_list)) &&
>>>                                             split_folio_to_list(folio,
>>>                                                                 folio_list))
>>>                                                 goto activate_locked;
>>
>> Hi Ryan,
>>
>> Sorry for bringing up another minor issue at this late stage.
> 
> No problem - I'd rather take a bit longer and get it right, rather than rush it
> and get it wrong!
> 
>>
>> During the debugging of thp counter patch v2, I noticed the discrepancy between
>> THP_SWPOUT_FALLBACK and THP_SWPOUT.
>>
>> Should we make adjustments to the counter?
> 
> Yes, agreed; we want to be consistent here with all the other existing THP
> counters; they only refer to PMD-sized THP. I'll make the change for the next
> version.
> 
> I guess we will eventually want equivalent counters for per-size mTHP using the
> framework you are adding.
> 
>>
>> diff --git a/mm/vmscan.c b/mm/vmscan.c
>> index 293120fe54f3..d7856603f689 100644
>> --- a/mm/vmscan.c
>> +++ b/mm/vmscan.c
>> @@ -1241,8 +1241,10 @@ static unsigned int shrink_folio_list(struct
>> list_head *folio_list,
>>                                                                 folio_list))
>>                                                 goto activate_locked;
>>  #ifdef CONFIG_TRANSPARENT_HUGEPAGE
>> -
>> count_memcg_folio_events(folio, THP_SWPOUT_FALLBACK, 1);
>> -                                       count_vm_event(THP_SWPOUT_FALLBACK);
>> +                                       if (folio_test_pmd_mappable(folio)) {

This doesn't quite work because we have already split the folio here, so this
will always return false. I've changed it to:

if (nr_pages >= HPAGE_PMD_NR) {


>> +
>> count_memcg_folio_events(folio, THP_SWPOUT_FALLBACK, 1);
>> +
>> count_vm_event(THP_SWPOUT_FALLBACK);
>> +                                       }
>>  #endif
>>                                         if (!add_to_swap(folio))
>>                                                 goto activate_locked_split;
>>
>>
>> Because THP_SWPOUT is only for pmd:
>>
>> static inline void count_swpout_vm_event(struct folio *folio)
>> {
>> #ifdef CONFIG_TRANSPARENT_HUGEPAGE
>>         if (unlikely(folio_test_pmd_mappable(folio))) {
>>                 count_memcg_folio_events(folio, THP_SWPOUT, 1);
>>                 count_vm_event(THP_SWPOUT);
>>         }
>> #endif
>>         count_vm_events(PSWPOUT, folio_nr_pages(folio));
>> }
>>
>> I can provide per-order counters for this in my THP counter patch.
>>
>>> --
>>> 2.25.1
>>>
>>
>> Thanks
>> Barry
>
Barry Song April 2, 2024, 10:54 p.m. UTC | #6
On Wed, Apr 3, 2024 at 2:22 AM Ryan Roberts <ryan.roberts@arm.com> wrote:
>
> On 02/04/2024 14:10, Ryan Roberts wrote:
> > On 28/03/2024 08:18, Barry Song wrote:
> >> On Thu, Mar 28, 2024 at 3:45 AM Ryan Roberts <ryan.roberts@arm.com> wrote:
> >>>
> >>> Now that swap supports storing all mTHP sizes, avoid splitting large
> >>> folios before swap-out. This benefits performance of the swap-out path
> >>> by eliding split_folio_to_list(), which is expensive, and also sets us
> >>> up for swapping in large folios in a future series.
> >>>
> >>> If the folio is partially mapped, we continue to split it since we want
> >>> to avoid the extra IO overhead and storage of writing out pages
> >>> uneccessarily.
> >>>
> >>> Reviewed-by: David Hildenbrand <david@redhat.com>
> >>> Reviewed-by: Barry Song <v-songbaohua@oppo.com>
> >>> Signed-off-by: Ryan Roberts <ryan.roberts@arm.com>
> >>> ---
> >>>  mm/vmscan.c | 9 +++++----
> >>>  1 file changed, 5 insertions(+), 4 deletions(-)
> >>>
> >>> diff --git a/mm/vmscan.c b/mm/vmscan.c
> >>> index 00adaf1cb2c3..293120fe54f3 100644
> >>> --- a/mm/vmscan.c
> >>> +++ b/mm/vmscan.c
> >>> @@ -1223,11 +1223,12 @@ static unsigned int shrink_folio_list(struct list_head *folio_list,
> >>>                                         if (!can_split_folio(folio, NULL))
> >>>                                                 goto activate_locked;
> >>>                                         /*
> >>> -                                        * Split folios without a PMD map right
> >>> -                                        * away. Chances are some or all of the
> >>> -                                        * tail pages can be freed without IO.
> >>> +                                        * Split partially mapped folios right
> >>> +                                        * away. We can free the unmapped pages
> >>> +                                        * without IO.
> >>>                                          */
> >>> -                                       if (!folio_entire_mapcount(folio) &&
> >>> +                                       if (data_race(!list_empty(
> >>> +                                               &folio->_deferred_list)) &&
> >>>                                             split_folio_to_list(folio,
> >>>                                                                 folio_list))
> >>>                                                 goto activate_locked;
> >>
> >> Hi Ryan,
> >>
> >> Sorry for bringing up another minor issue at this late stage.
> >
> > No problem - I'd rather take a bit longer and get it right, rather than rush it
> > and get it wrong!
> >
> >>
> >> During the debugging of thp counter patch v2, I noticed the discrepancy between
> >> THP_SWPOUT_FALLBACK and THP_SWPOUT.
> >>
> >> Should we make adjustments to the counter?
> >
> > Yes, agreed; we want to be consistent here with all the other existing THP
> > counters; they only refer to PMD-sized THP. I'll make the change for the next
> > version.
> >
> > I guess we will eventually want equivalent counters for per-size mTHP using the
> > framework you are adding.
> >
> >>
> >> diff --git a/mm/vmscan.c b/mm/vmscan.c
> >> index 293120fe54f3..d7856603f689 100644
> >> --- a/mm/vmscan.c
> >> +++ b/mm/vmscan.c
> >> @@ -1241,8 +1241,10 @@ static unsigned int shrink_folio_list(struct
> >> list_head *folio_list,
> >>                                                                 folio_list))
> >>                                                 goto activate_locked;
> >>  #ifdef CONFIG_TRANSPARENT_HUGEPAGE
> >> -
> >> count_memcg_folio_events(folio, THP_SWPOUT_FALLBACK, 1);
> >> -                                       count_vm_event(THP_SWPOUT_FALLBACK);
> >> +                                       if (folio_test_pmd_mappable(folio)) {
>
> This doesn't quite work because we have already split the folio here, so this
> will always return false. I've changed it to:
>
> if (nr_pages >= HPAGE_PMD_NR) {

make sense to me.

>
>
> >> +
> >> count_memcg_folio_events(folio, THP_SWPOUT_FALLBACK, 1);
> >> +
> >> count_vm_event(THP_SWPOUT_FALLBACK);
> >> +                                       }
> >>  #endif
> >>                                         if (!add_to_swap(folio))
> >>                                                 goto activate_locked_split;
> >>
> >>
> >> Because THP_SWPOUT is only for pmd:
> >>
> >> static inline void count_swpout_vm_event(struct folio *folio)
> >> {
> >> #ifdef CONFIG_TRANSPARENT_HUGEPAGE
> >>         if (unlikely(folio_test_pmd_mappable(folio))) {
> >>                 count_memcg_folio_events(folio, THP_SWPOUT, 1);
> >>                 count_vm_event(THP_SWPOUT);
> >>         }
> >> #endif
> >>         count_vm_events(PSWPOUT, folio_nr_pages(folio));
> >> }
> >>
> >> I can provide per-order counters for this in my THP counter patch.
> >>
> >>> --
> >>> 2.25.1
> >>>
> >>

Thanks
Barry
Barry Song April 5, 2024, 4:06 a.m. UTC | #7
On Wed, Apr 3, 2024 at 2:10 AM Ryan Roberts <ryan.roberts@arm.com> wrote:
>
> On 28/03/2024 08:18, Barry Song wrote:
> > On Thu, Mar 28, 2024 at 3:45 AM Ryan Roberts <ryan.roberts@arm.com> wrote:
> >>
> >> Now that swap supports storing all mTHP sizes, avoid splitting large
> >> folios before swap-out. This benefits performance of the swap-out path
> >> by eliding split_folio_to_list(), which is expensive, and also sets us
> >> up for swapping in large folios in a future series.
> >>
> >> If the folio is partially mapped, we continue to split it since we want
> >> to avoid the extra IO overhead and storage of writing out pages
> >> uneccessarily.
> >>
> >> Reviewed-by: David Hildenbrand <david@redhat.com>
> >> Reviewed-by: Barry Song <v-songbaohua@oppo.com>
> >> Signed-off-by: Ryan Roberts <ryan.roberts@arm.com>
> >> ---
> >>  mm/vmscan.c | 9 +++++----
> >>  1 file changed, 5 insertions(+), 4 deletions(-)
> >>
> >> diff --git a/mm/vmscan.c b/mm/vmscan.c
> >> index 00adaf1cb2c3..293120fe54f3 100644
> >> --- a/mm/vmscan.c
> >> +++ b/mm/vmscan.c
> >> @@ -1223,11 +1223,12 @@ static unsigned int shrink_folio_list(struct list_head *folio_list,
> >>                                         if (!can_split_folio(folio, NULL))
> >>                                                 goto activate_locked;
> >>                                         /*
> >> -                                        * Split folios without a PMD map right
> >> -                                        * away. Chances are some or all of the
> >> -                                        * tail pages can be freed without IO.
> >> +                                        * Split partially mapped folios right
> >> +                                        * away. We can free the unmapped pages
> >> +                                        * without IO.
> >>                                          */
> >> -                                       if (!folio_entire_mapcount(folio) &&
> >> +                                       if (data_race(!list_empty(
> >> +                                               &folio->_deferred_list)) &&
> >>                                             split_folio_to_list(folio,
> >>                                                                 folio_list))
> >>                                                 goto activate_locked;
> >
> > Hi Ryan,
> >
> > Sorry for bringing up another minor issue at this late stage.
>
> No problem - I'd rather take a bit longer and get it right, rather than rush it
> and get it wrong!
>
> >
> > During the debugging of thp counter patch v2, I noticed the discrepancy between
> > THP_SWPOUT_FALLBACK and THP_SWPOUT.
> >
> > Should we make adjustments to the counter?
>
> Yes, agreed; we want to be consistent here with all the other existing THP
> counters; they only refer to PMD-sized THP. I'll make the change for the next
> version.
>
> I guess we will eventually want equivalent counters for per-size mTHP using the
> framework you are adding.

Hi Ryan,

Today, I created counters for per-order SWPOUT and SWPOUT_FALLBACK.
I'd appreciate any
suggestions you might have before I submit this as patch 2/2 of my
mTHP counters series.

diff --git a/include/linux/huge_mm.h b/include/linux/huge_mm.h
index cc13fa14aa32..762a6d8759b9 100644
--- a/include/linux/huge_mm.h
+++ b/include/linux/huge_mm.h
@@ -267,6 +267,8 @@ unsigned long thp_vma_allowable_orders(struct
vm_area_struct *vma,
 enum thp_stat_item {
        THP_STAT_ANON_ALLOC,
        THP_STAT_ANON_ALLOC_FALLBACK,
+       THP_STAT_ANON_SWPOUT,
+       THP_STAT_ANON_SWPOUT_FALLBACK,
        __THP_STAT_COUNT
 };

diff --git a/mm/huge_memory.c b/mm/huge_memory.c
index e704b4408181..7f2b5d2852cc 100644
--- a/mm/huge_memory.c
+++ b/mm/huge_memory.c
@@ -554,10 +554,14 @@ static struct kobj_attribute _name##_attr =
__ATTR_RO(_name)

 THP_STATE_ATTR(anon_alloc, THP_STAT_ANON_ALLOC);
 THP_STATE_ATTR(anon_alloc_fallback, THP_STAT_ANON_ALLOC_FALLBACK);
+THP_STATE_ATTR(anon_swpout, THP_STAT_ANON_SWPOUT);
+THP_STATE_ATTR(anon_swpout_fallback, THP_STAT_ANON_SWPOUT_FALLBACK);

 static struct attribute *stats_attrs[] = {
        &anon_alloc_attr.attr,
        &anon_alloc_fallback_attr.attr,
+       &anon_swpout_attr.attr,
+       &anon_swpout_fallback_attr.attr,
        NULL,
 };

diff --git a/mm/page_io.c b/mm/page_io.c
index a9a7c236aecc..be4f822b39f8 100644
--- a/mm/page_io.c
+++ b/mm/page_io.c
@@ -212,13 +212,16 @@ int swap_writepage(struct page *page, struct
writeback_control *wbc)

 static inline void count_swpout_vm_event(struct folio *folio)
 {
+       long nr_pages = folio_nr_pages(folio);
 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
        if (unlikely(folio_test_pmd_mappable(folio))) {
                count_memcg_folio_events(folio, THP_SWPOUT, 1);
                count_vm_event(THP_SWPOUT);
        }
+       if (nr_pages > 0 && nr_pages <= HPAGE_PMD_NR)
+               count_thp_state(folio_order(folio), THP_STAT_ANON_SWPOUT);
 #endif
-       count_vm_events(PSWPOUT, folio_nr_pages(folio));
+       count_vm_events(PSWPOUT, nr_pages);
 }

 #if defined(CONFIG_MEMCG) && defined(CONFIG_BLK_CGROUP)
diff --git a/mm/vmscan.c b/mm/vmscan.c
index ffc4553c8615..b7c5fbd830b6 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -1247,6 +1247,10 @@ static unsigned int shrink_folio_list(struct
list_head *folio_list,
                                                count_vm_event(
                                                        THP_SWPOUT_FALLBACK);
                                        }
+                                       if (nr_pages > 0 && nr_pages
<= HPAGE_PMD_NR)
+
count_thp_state(folio_order(folio),
+
THP_STAT_ANON_SWPOUT_FALLBACK);
+
 #endif
                                        if (!add_to_swap(folio))
                                                goto activate_locked_split;


Thanks
Barry
Ryan Roberts April 5, 2024, 7:28 a.m. UTC | #8
On 05/04/2024 05:06, Barry Song wrote:
> On Wed, Apr 3, 2024 at 2:10 AM Ryan Roberts <ryan.roberts@arm.com> wrote:
>>
>> On 28/03/2024 08:18, Barry Song wrote:
>>> On Thu, Mar 28, 2024 at 3:45 AM Ryan Roberts <ryan.roberts@arm.com> wrote:
>>>>
>>>> Now that swap supports storing all mTHP sizes, avoid splitting large
>>>> folios before swap-out. This benefits performance of the swap-out path
>>>> by eliding split_folio_to_list(), which is expensive, and also sets us
>>>> up for swapping in large folios in a future series.
>>>>
>>>> If the folio is partially mapped, we continue to split it since we want
>>>> to avoid the extra IO overhead and storage of writing out pages
>>>> uneccessarily.
>>>>
>>>> Reviewed-by: David Hildenbrand <david@redhat.com>
>>>> Reviewed-by: Barry Song <v-songbaohua@oppo.com>
>>>> Signed-off-by: Ryan Roberts <ryan.roberts@arm.com>
>>>> ---
>>>>  mm/vmscan.c | 9 +++++----
>>>>  1 file changed, 5 insertions(+), 4 deletions(-)
>>>>
>>>> diff --git a/mm/vmscan.c b/mm/vmscan.c
>>>> index 00adaf1cb2c3..293120fe54f3 100644
>>>> --- a/mm/vmscan.c
>>>> +++ b/mm/vmscan.c
>>>> @@ -1223,11 +1223,12 @@ static unsigned int shrink_folio_list(struct list_head *folio_list,
>>>>                                         if (!can_split_folio(folio, NULL))
>>>>                                                 goto activate_locked;
>>>>                                         /*
>>>> -                                        * Split folios without a PMD map right
>>>> -                                        * away. Chances are some or all of the
>>>> -                                        * tail pages can be freed without IO.
>>>> +                                        * Split partially mapped folios right
>>>> +                                        * away. We can free the unmapped pages
>>>> +                                        * without IO.
>>>>                                          */
>>>> -                                       if (!folio_entire_mapcount(folio) &&
>>>> +                                       if (data_race(!list_empty(
>>>> +                                               &folio->_deferred_list)) &&
>>>>                                             split_folio_to_list(folio,
>>>>                                                                 folio_list))
>>>>                                                 goto activate_locked;
>>>
>>> Hi Ryan,
>>>
>>> Sorry for bringing up another minor issue at this late stage.
>>
>> No problem - I'd rather take a bit longer and get it right, rather than rush it
>> and get it wrong!
>>
>>>
>>> During the debugging of thp counter patch v2, I noticed the discrepancy between
>>> THP_SWPOUT_FALLBACK and THP_SWPOUT.
>>>
>>> Should we make adjustments to the counter?
>>
>> Yes, agreed; we want to be consistent here with all the other existing THP
>> counters; they only refer to PMD-sized THP. I'll make the change for the next
>> version.
>>
>> I guess we will eventually want equivalent counters for per-size mTHP using the
>> framework you are adding.
> 
> Hi Ryan,
> 
> Today, I created counters for per-order SWPOUT and SWPOUT_FALLBACK.
> I'd appreciate any
> suggestions you might have before I submit this as patch 2/2 of my
> mTHP counters series.

Amazing - this is going to be very useful!

> 
> diff --git a/include/linux/huge_mm.h b/include/linux/huge_mm.h
> index cc13fa14aa32..762a6d8759b9 100644
> --- a/include/linux/huge_mm.h
> +++ b/include/linux/huge_mm.h
> @@ -267,6 +267,8 @@ unsigned long thp_vma_allowable_orders(struct
> vm_area_struct *vma,
>  enum thp_stat_item {
>         THP_STAT_ANON_ALLOC,
>         THP_STAT_ANON_ALLOC_FALLBACK,
> +       THP_STAT_ANON_SWPOUT,
> +       THP_STAT_ANON_SWPOUT_FALLBACK,
>         __THP_STAT_COUNT
>  };
> 
> diff --git a/mm/huge_memory.c b/mm/huge_memory.c
> index e704b4408181..7f2b5d2852cc 100644
> --- a/mm/huge_memory.c
> +++ b/mm/huge_memory.c
> @@ -554,10 +554,14 @@ static struct kobj_attribute _name##_attr =
> __ATTR_RO(_name)
> 
>  THP_STATE_ATTR(anon_alloc, THP_STAT_ANON_ALLOC);
>  THP_STATE_ATTR(anon_alloc_fallback, THP_STAT_ANON_ALLOC_FALLBACK);
> +THP_STATE_ATTR(anon_swpout, THP_STAT_ANON_SWPOUT);
> +THP_STATE_ATTR(anon_swpout_fallback, THP_STAT_ANON_SWPOUT_FALLBACK);
> 
>  static struct attribute *stats_attrs[] = {
>         &anon_alloc_attr.attr,
>         &anon_alloc_fallback_attr.attr,
> +       &anon_swpout_attr.attr,
> +       &anon_swpout_fallback_attr.attr,
>         NULL,
>  };
> 
> diff --git a/mm/page_io.c b/mm/page_io.c
> index a9a7c236aecc..be4f822b39f8 100644
> --- a/mm/page_io.c
> +++ b/mm/page_io.c
> @@ -212,13 +212,16 @@ int swap_writepage(struct page *page, struct
> writeback_control *wbc)
> 
>  static inline void count_swpout_vm_event(struct folio *folio)
>  {
> +       long nr_pages = folio_nr_pages(folio);
>  #ifdef CONFIG_TRANSPARENT_HUGEPAGE
>         if (unlikely(folio_test_pmd_mappable(folio))) {
>                 count_memcg_folio_events(folio, THP_SWPOUT, 1);
>                 count_vm_event(THP_SWPOUT);
>         }
> +       if (nr_pages > 0 && nr_pages <= HPAGE_PMD_NR)

The guard is a bit ugly; I wonder if we should at least check that order is in
bounds in count_thp_state(), since all callers could benefit? Then we only have
to care about the nr_pages > 0 condition here. Just a thought...

> +               count_thp_state(folio_order(folio), THP_STAT_ANON_SWPOUT);

So you're counting THPs, not pages; I agree with that approach.

>  #endif
> -       count_vm_events(PSWPOUT, folio_nr_pages(folio));
> +       count_vm_events(PSWPOUT, nr_pages);
>  }
> 
>  #if defined(CONFIG_MEMCG) && defined(CONFIG_BLK_CGROUP)
> diff --git a/mm/vmscan.c b/mm/vmscan.c
> index ffc4553c8615..b7c5fbd830b6 100644
> --- a/mm/vmscan.c
> +++ b/mm/vmscan.c
> @@ -1247,6 +1247,10 @@ static unsigned int shrink_folio_list(struct
> list_head *folio_list,
>                                                 count_vm_event(
>                                                         THP_SWPOUT_FALLBACK);
>                                         }
> +                                       if (nr_pages > 0 && nr_pages
> <= HPAGE_PMD_NR)
> +
> count_thp_state(folio_order(folio),
> +
> THP_STAT_ANON_SWPOUT_FALLBACK);
> +
>  #endif
>                                         if (!add_to_swap(folio))
>                                                 goto activate_locked_split;
> 
> 
> Thanks
> Barry
diff mbox series

Patch

diff --git a/mm/vmscan.c b/mm/vmscan.c
index 00adaf1cb2c3..293120fe54f3 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -1223,11 +1223,12 @@  static unsigned int shrink_folio_list(struct list_head *folio_list,
 					if (!can_split_folio(folio, NULL))
 						goto activate_locked;
 					/*
-					 * Split folios without a PMD map right
-					 * away. Chances are some or all of the
-					 * tail pages can be freed without IO.
+					 * Split partially mapped folios right
+					 * away. We can free the unmapped pages
+					 * without IO.
 					 */
-					if (!folio_entire_mapcount(folio) &&
+					if (data_race(!list_empty(
+						&folio->_deferred_list)) &&
 					    split_folio_to_list(folio,
 								folio_list))
 						goto activate_locked;