public inbox for linux-mm@kvack.org
 help / color / mirror / Atom feed
From: Kefeng Wang <wangkefeng.wang@huawei.com>
To: Baolin Wang <baolin.wang@linux.alibaba.com>,
	Andrew Morton <akpm@linux-foundation.org>
Cc: David Hildenbrand <david@redhat.com>,
	Matthew Wilcox <willy@infradead.org>, Zi Yan <ziy@nvidia.com>,
	<linux-mm@kvack.org>
Subject: Re: [PATCH 2/4] mm: compaction: convert to folio_isolate_movable()
Date: Wed, 28 Aug 2024 20:36:03 +0800	[thread overview]
Message-ID: <606adde7-44ff-476a-bd01-2141eb28943e@huawei.com> (raw)
In-Reply-To: <3f8300d9-1c21-46ad-a311-e97dc94eda08@linux.alibaba.com>



On 2024/8/28 18:04, Baolin Wang wrote:
> Hi Kefeng,
> 
> On 2024/8/26 12:01, Kefeng Wang wrote:
>> The tail page will always fail to isolate for non-lru-movable and
>> LRU page in isolate_migratepages_block(), so move PageTail check
>> ahead, then convert to use folio_isolate_movable() helper and more
>> folios.
>>
>> Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
>> ---
>>   mm/compaction.c | 32 +++++++++++++++++---------------
>>   1 file changed, 17 insertions(+), 15 deletions(-)
>>
>> diff --git a/mm/compaction.c b/mm/compaction.c
>> index a2b16b08cbbf..aa2e8bb9fa58 100644
>> --- a/mm/compaction.c
>> +++ b/mm/compaction.c
>> @@ -1074,39 +1074,41 @@ isolate_migratepages_block(struct 
>> compact_control *cc, unsigned long low_pfn,
>>               }
>>           }
>> +        if (PageTail(page))
>> +            goto isolate_fail;
>> +
>> +        folio = page_folio(page);
> 
> I wonder if this is stable. Before page_folio(), it does not hold a 
> reference on the page, so seems we should re-check the folio still 
> contains this page after gaining a reference on the folio?

Oh, you are right, so two way to avoid this,

1) re-check 'page_folio(page)  == folio', but this need change a little 
more.

2)  directly use folio_get_nontail_page(page) here, and folio_put in the 
following path, this will try to get for any pages, but it should be 
accepted ?

I'd prefer 2) but any other suggestion?

Thanks.

> 
>>           /*
>>            * Check may be lockless but that's ok as we recheck later.
>> -         * It's possible to migrate LRU and non-lru movable pages.
>> -         * Skip any other type of page
>> +         * It's possible to migrate LRU and non-lru movable folioss.
>> +         * Skip any other type of folios.
>>            */
>> -        if (!PageLRU(page)) {
>> +        if (!folio_test_lru(folio)) {
>>               /*
>> -             * __PageMovable can return false positive so we need
>> -             * to verify it under page_lock.
>> +             * __folio_test_movable can return false positive so
>> +             * we need to verify it under page_lock.
>>                */
>> -            if (unlikely(__PageMovable(page)) &&
>> -                    !PageIsolated(page)) {
>> +            if (unlikely(__folio_test_movable(folio)) &&
>> +                    !folio_test_isolated(folio)) {
>>                   if (locked) {
>>                       unlock_page_lruvec_irqrestore(locked, flags);
>>                       locked = NULL;
>>                   }
>> -                if (isolate_movable_page(page, mode)) {
>> -                    folio = page_folio(page);
>> +                if (folio_isolate_movable(folio, mode))
>>                       goto isolate_success;
>> -                }
>>               }
>>               goto isolate_fail;
>>           }
>>           /*
>> -         * Be careful not to clear PageLRU until after we're
>> -         * sure the page is not being freed elsewhere -- the
>> -         * page release code relies on it.
>> +         * Be careful not to clear lru flag until after we're
>> +         * sure the folio is not being freed elsewhere -- the
>> +         * folio release code relies on it.
>>            */
>> -        folio = folio_get_nontail_page(page);
>> -        if (unlikely(!folio))
>> +        if (unlikely(folio_try_get(folio)))
>>               goto isolate_fail;
>>           /*


  reply	other threads:[~2024-08-28 12:36 UTC|newest]

Thread overview: 13+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-08-26  4:01 [PATCH 0/4] mm: convert to folio_isolate_movable() Kefeng Wang
2024-08-26  4:01 ` [PATCH 1/4] mm: migrate: add folio_isolate_movable() Kefeng Wang
2024-08-28  1:14   ` Andrew Morton
2024-08-28  6:37     ` Kefeng Wang
2024-08-26  4:01 ` [PATCH 2/4] mm: compaction: convert to folio_isolate_movable() Kefeng Wang
2024-08-28 10:04   ` Baolin Wang
2024-08-28 12:36     ` Kefeng Wang [this message]
2024-08-28 19:54       ` Andrew Morton
2024-08-28 23:37       ` Vishal Moola
2024-08-29  1:11         ` Baolin Wang
2024-08-29  1:34           ` Kefeng Wang
2024-08-26  4:01 ` [PATCH 3/4] mm: migrate: " Kefeng Wang
2024-08-26  4:01 ` [PATCH 4/4] mm: migrate: remove isolate_movable_page() Kefeng Wang

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=606adde7-44ff-476a-bd01-2141eb28943e@huawei.com \
    --to=wangkefeng.wang@huawei.com \
    --cc=akpm@linux-foundation.org \
    --cc=baolin.wang@linux.alibaba.com \
    --cc=david@redhat.com \
    --cc=linux-mm@kvack.org \
    --cc=willy@infradead.org \
    --cc=ziy@nvidia.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox