Re: [PATCH 13/16] mm/migration: return errno when isolate_huge_page failed

[Miaohe Lin]

On 2022/3/8 9:32, Baolin Wang wrote:
> 
> 
> On 3/7/2022 8:20 PM, Miaohe Lin wrote:
>> On 2022/3/7 10:14, Baolin Wang wrote:
>>> Hi Miaohe,
>>>
>>> On 3/4/2022 5:34 PM, Miaohe Lin wrote:
>>>> We should return errno (-EBUSY here) when failed to isolate the huge page
>>>> rather than always return 1 which could confuse the user.
>>>>
>>>> Signed-off-by: Miaohe Lin <linmiaohe@xxxxxxxxxx>
>>>> ---
>>>>    mm/migrate.c | 6 ++----
>>>>    1 file changed, 2 insertions(+), 4 deletions(-)
>>>>
>>>> diff --git a/mm/migrate.c b/mm/migrate.c
>>>> index 6c2dfed2ddb8..279940c0c064 100644
>>>> --- a/mm/migrate.c
>>>> +++ b/mm/migrate.c
>>>> @@ -1618,10 +1618,8 @@ static int add_page_for_migration(struct mm_struct *mm, unsigned long addr,
>>>>            goto out_putpage;
>>>>          if (PageHuge(page)) {
>>>> -        if (PageHead(page)) {
>>>> -            isolate_huge_page(page, pagelist);
>>>> -            err = 1;
>>>> -        }
>>>> +        if (PageHead(page))
>>>> +            err = isolate_huge_page(page, pagelist) ? 1 : -EBUSY;
>>>
>>> Could you elaborate on which case the huge page isolation can be failed in this case? Or you met a real problem? Cause now we've got this head huge page refcnt, I can not see why we'll fail to isolate this huge page.
>>
>> IIUC, this could happen when hugepage is under migration which cleared HPageMigratable. Page refcnt cannot
>> prevent isolate_huge_page from happening. Or am I miss something?
> 
> Yes, that's possible. Thanks for your explanation. It will be better if you can copy the possible scenario description to the commit log to help other understand the issue.
> 

Sounds reasonable. Will do. Many thanks for review.

> Reviewed-by: Baolin Wang <baolin.wang@xxxxxxxxxxxxxxxxx>
> .