2021-01-05 17:02:00

by David Hildenbrand

Subject: Re: [PATCH 1/6] mm: migrate: do not migrate HugeTLB page whose refcount is one

On 04.01.21 07:58, Muchun Song wrote:
> If the refcount is one when it is migrated, it means that the page
> was freed from under us. So we are done and do not need to migrate.
>
> Signed-off-by: Muchun Song <[email protected]>
> ---
> mm/migrate.c | 6 ++++++
> 1 file changed, 6 insertions(+)
>
> diff --git a/mm/migrate.c b/mm/migrate.c
> index 4385f2fb5d18..a6631c4eb6a6 100644
> --- a/mm/migrate.c
> +++ b/mm/migrate.c
> @@ -1279,6 +1279,12 @@ static int unmap_and_move_huge_page(new_page_t get_new_page,
> return -ENOSYS;
> }
>
> + if (page_count(hpage) == 1) {
> + /* page was freed from under us. So we are done. */
> + putback_active_hugepage(hpage);
> + return MIGRATEPAGE_SUCCESS;
> + }
> +
> new_hpage = get_new_page(hpage, private);
> if (!new_hpage)
> return -ENOMEM;
>

This series seems to fix quite some important cases (thanks). Do we want
to cc stable some/all?

--
Thanks,

David / dhildenb

2021-01-05 18:07:24

by Yang Shi

Subject: Re: [PATCH 1/6] mm: migrate: do not migrate HugeTLB page whose refcount is one

On Tue, Jan 5, 2021 at 8:58 AM David Hildenbrand <[email protected]> wrote:
>
> On 04.01.21 07:58, Muchun Song wrote:
> > If the refcount is one when it is migrated, it means that the page
> > was freed from under us. So we are done and do not need to migrate.
> >
> > Signed-off-by: Muchun Song <[email protected]>
> > ---
> > mm/migrate.c | 6 ++++++
> > 1 file changed, 6 insertions(+)
> >
> > diff --git a/mm/migrate.c b/mm/migrate.c
> > index 4385f2fb5d18..a6631c4eb6a6 100644
> > --- a/mm/migrate.c
> > +++ b/mm/migrate.c
> > @@ -1279,6 +1279,12 @@ static int unmap_and_move_huge_page(new_page_t get_new_page,
> > return -ENOSYS;
> > }
> >
> > + if (page_count(hpage) == 1) {
> > + /* page was freed from under us. So we are done. */
> > + putback_active_hugepage(hpage);
> > + return MIGRATEPAGE_SUCCESS;
> > + }
> > +
> > new_hpage = get_new_page(hpage, private);
> > if (!new_hpage)
> > return -ENOMEM;
> >
>
> This series seems to fix quite some important cases (thanks). Do we want
> to cc stable some/all?

For this particular one, I don't think so IMHO. It is an optimization
rather than a bug fix.

>
> --
> Thanks,
>
> David / dhildenb
>
>

2021-01-05 18:09:13

by David Hildenbrand

Subject: Re: [PATCH 1/6] mm: migrate: do not migrate HugeTLB page whose refcount is one

On 05.01.21 19:04, Yang Shi wrote:
> On Tue, Jan 5, 2021 at 8:58 AM David Hildenbrand <[email protected]> wrote:
>>
>> On 04.01.21 07:58, Muchun Song wrote:
>>> If the refcount is one when it is migrated, it means that the page
>>> was freed from under us. So we are done and do not need to migrate.
>>>
>>> Signed-off-by: Muchun Song <[email protected]>
>>> ---
>>> mm/migrate.c | 6 ++++++
>>> 1 file changed, 6 insertions(+)
>>>
>>> diff --git a/mm/migrate.c b/mm/migrate.c
>>> index 4385f2fb5d18..a6631c4eb6a6 100644
>>> --- a/mm/migrate.c
>>> +++ b/mm/migrate.c
>>> @@ -1279,6 +1279,12 @@ static int unmap_and_move_huge_page(new_page_t get_new_page,
>>> return -ENOSYS;
>>> }
>>>
>>> + if (page_count(hpage) == 1) {
>>> + /* page was freed from under us. So we are done. */
>>> + putback_active_hugepage(hpage);
>>> + return MIGRATEPAGE_SUCCESS;
>>> + }
>>> +
>>> new_hpage = get_new_page(hpage, private);
>>> if (!new_hpage)
>>> return -ENOMEM;
>>>
>>
>> This series seems to fix quite some important cases (thanks). Do we want
>> to cc stable some/all?
>
> For this particular one, I don't think so IMHO. It is an optimization
> rather than a bug fix.

I'm rather referring to the actual fixes (I received no cover letter so
I picked the first patch in the series)


--
Thanks,

David / dhildenb