On 23.08.22 15:26, Wang, Haiyue wrote:
>> -----Original Message-----
>> From: David Hildenbrand <david@xxxxxxxxxx>
>> Sent: Tuesday, August 23, 2022 18:07
>> To: Miaohe Lin <linmiaohe@xxxxxxxxxx>; Wang, Haiyue <haiyue.wang@xxxxxxxxx>; linux-mm@xxxxxxxxx;
>> linux-kernel@xxxxxxxxxxxxxxx
>> Cc: akpm@xxxxxxxxxxxxxxxxxxxx; apopple@xxxxxxxxxx; Huang, Ying <ying.huang@xxxxxxxxx>;
>> songmuchun@xxxxxxxxxxxxx; naoya.horiguchi@xxxxxxxxx; alex.sierra@xxxxxxx; Felix Kuehling
>> <Felix.Kuehling@xxxxxxx>
>> Subject: Re: [PATCH v6 2/2] mm: fix the handling Non-LRU pages returned by follow_page
>>
>> On 17.08.22 04:34, Miaohe Lin wrote:
>>> On 2022/8/16 10:21, Haiyue Wang wrote:
>>>> The handling Non-LRU pages returned by follow_page() jumps directly, it
>>>> doesn't call put_page() to handle the reference count, since 'FOLL_GET'
>>>> flag for follow_page() has get_page() called. Fix the zone device page
>>>> check by handling the page reference count correctly before returning.
>>>>
>>>> And as David reviewed, "device pages are never PageKsm pages". Drop this
>>>> zone device page check for break_ksm().
>>>>
>>>> Fixes: 3218f8712d6b ("mm: handling Non-LRU pages returned by vm_normal_pages")
>>>> Signed-off-by: Haiyue Wang <haiyue.wang@xxxxxxxxx>
>>>> Reviewed-by: "Huang, Ying" <ying.huang@xxxxxxxxx>
>>>> Reviewed-by: Felix Kuehling <Felix.Kuehling@xxxxxxx>
>>>
>>> Thanks for your fixing. LGTM with one nit below. But I have no strong opinion on it.
>>> So with or without fixing below nit:
>>>
>>> Reviewed-by: Miaohe Lin <linmiaohe@xxxxxxxxxx>
>>>
>>>> ---
>>>> mm/huge_memory.c | 4 ++--
>>>> mm/ksm.c | 12 +++++++++---
>>>> mm/migrate.c | 19 ++++++++++++-------
>>>> 3 files changed, 23 insertions(+), 12 deletions(-)
>>>>
>>>> diff --git a/mm/huge_memory.c b/mm/huge_memory.c
>>>> index 8a7c1b344abe..b2ba17c3dcd7 100644
>>>> --- a/mm/huge_memory.c
>>>> +++ b/mm/huge_memory.c
>>>> @@ -2963,10 +2963,10 @@ static int split_huge_pages_pid(int pid, unsigned long vaddr_start,
>>>> /* FOLL_DUMP to ignore special (like zero) pages */
>>>> page = follow_page(vma, addr, FOLL_GET | FOLL_DUMP);
>>>>
>>>> - if (IS_ERR_OR_NULL(page) || is_zone_device_page(page))
>>>> + if (IS_ERR_OR_NULL(page))
>>>> continue;
>>>>
>>>> - if (!is_transparent_hugepage(page))
>>>> + if (is_zone_device_page(page) || !is_transparent_hugepage(page))
>>>
>>> !is_transparent_hugepage should already do the work here? IIRC, zone_device_page can't be
>>> a transhuge page anyway. And only transparent_hugepage is cared here.
>>
>> I agree.
>
> OK, will remove it in next version.
>
>>
>> Can we avoid sending a new version of a patch series as reply to another
>> patch series (previous version)?
>
> Don't use '--in-reply-to=' when running 'git send-email' ?
Yes. That makes it easier top identify versions of patch series, without
having to dig down into an ever-growing thread.
--
Thanks,
David / dhildenb
