From: Miaohe Lin <linmiaohe@huawei.com>
To: "HORIGUCHI NAOYA(堀口 直也)" <naoya.horiguchi@nec.com>
Cc: "akpm@linux-foundation.org" <akpm@linux-foundation.org>,
"linux-mm@kvack.org" <linux-mm@kvack.org>,
"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH 2/6] mm, hwpoison: use __PageMovable() to detect non-lru movable pages
Date: Mon, 5 Sep 2022 14:53:41 +0800 [thread overview]
Message-ID: <1f7ee86e-7d28-0d8c-e0de-b7a5a94519e8@huawei.com> (raw)
In-Reply-To: <20220905052243.GA1355682@hori.linux.bs1.fc.nec.co.jp>
On 2022/9/5 13:22, HORIGUCHI NAOYA(堀口 直也) wrote:
> Hi Miaohe,
>
> On Tue, Aug 30, 2022 at 08:36:00PM +0800, Miaohe Lin wrote:
>> It's more recommended to use __PageMovable() to detect non-lru movable
>> pages. We can avoid bumping page refcnt via isolate_movable_page() for
>> the isolated lru pages. Also if pages become PageLRU just after they're
>> checked but before trying to isolate them, isolate_lru_page() will be
>> called to do the right work.
>
> Good point, non-lru movable page is currently handled by isolate_lru_page(),
> which always fails. This means that we lost the chance of soft-offlining
> for any non-lru movable page. So this patch improves the situation.
Non-lru movable page will still be handled by isolate_movable_page() before the code change
as they don't have PageLRU set. The current situation is that the isolated LRU pages are
passed to isolate_movable_page() uncorrectly. This might not hurt. But the chance that pages
become un-isolated and thus available just after checking could be seized with this patch.
>
>>
>> Signed-off-by: Miaohe Lin <linmiaohe@huawei.com>
>> ---
>> mm/memory-failure.c | 2 +-
>> 1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/mm/memory-failure.c b/mm/memory-failure.c
>> index a923a6dde871..3966fa6abe03 100644
>> --- a/mm/memory-failure.c
>> +++ b/mm/memory-failure.c
>> @@ -2404,7 +2404,7 @@ EXPORT_SYMBOL(unpoison_memory);
>> static bool isolate_page(struct page *page, struct list_head *pagelist)
>> {
>> bool isolated = false;
>> - bool lru = PageLRU(page);
>> + bool lru = !__PageMovable(page);
>
> It seems that PAGE_MAPPING_MOVABLE is not set for hugetlb pages, so
> lru becomes true for them. Then, if isolate_hugetlb() succeeds,
> inc_node_page_state() is called for hugetlb pages, maybe that's not expected.
Yes, that's unexpected. Thanks for pointing this out.
>
>>
>> if (PageHuge(page)) {
>> isolated = !isolate_hugetlb(page, pagelist);
> } else {
> if (lru)
> isolated = !isolate_lru_page(page);
> else
> isolated = !isolate_movable_page(page, ISOLATE_UNEVICTABLE);
>
> if (isolated)
> list_add(&page->lru, pagelist);
> }
>
> if (isolated && lru)
> inc_node_page_state(page, NR_ISOLATED_ANON +
> page_is_file_lru(page));
>
> so, how about moving this if block into the above else block?
> Then, the automatic variable lru can be moved into the else block.
Do you mean something like below?
diff --git a/mm/memory-failure.c b/mm/memory-failure.c
index df3bf266eebf..48780f3a61d3 100644
--- a/mm/memory-failure.c
+++ b/mm/memory-failure.c
@@ -2404,24 +2404,25 @@ EXPORT_SYMBOL(unpoison_memory);
static bool isolate_page(struct page *page, struct list_head *pagelist)
{
bool isolated = false;
- bool lru = !__PageMovable(page);
if (PageHuge(page)) {
isolated = !isolate_hugetlb(page, pagelist);
} else {
+ bool lru = !__PageMovable(page);
+
if (lru)
isolated = !isolate_lru_page(page);
else
isolated = !isolate_movable_page(page, ISOLATE_UNEVICTABLE);
- if (isolated)
+ if (isolated) {
list_add(&page->lru, pagelist);
+ if (lru)
+ inc_node_page_state(page, NR_ISOLATED_ANON +
+ page_is_file_lru(page));
+ }
}
- if (isolated && lru)
- inc_node_page_state(page, NR_ISOLATED_ANON +
- page_is_file_lru(page));
-
/*
* If we succeed to isolate the page, we grabbed another refcount on
* the page, so we can safely drop the one we got from get_any_pages().
>
> Thanks,
> Naoya Horiguchi
Thanks a lot for your review and comment on this series, Naoya.
Thanks,
Miaohe Lin.
next prev parent reply other threads:[~2022-09-05 6:53 UTC|newest]
Thread overview: 18+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-08-30 12:35 [PATCH 0/6] A few cleanup patches for memory-failure Miaohe Lin
2022-08-30 12:35 ` [PATCH 1/6] mm, hwpoison: use ClearPageHWPoison() in memory_failure() Miaohe Lin
2022-09-05 5:23 ` HORIGUCHI NAOYA(堀口 直也)
2022-08-30 12:36 ` [PATCH 2/6] mm, hwpoison: use __PageMovable() to detect non-lru movable pages Miaohe Lin
2022-09-05 5:22 ` HORIGUCHI NAOYA(堀口 直也)
2022-09-05 6:53 ` Miaohe Lin [this message]
2022-09-05 7:15 ` HORIGUCHI NAOYA(堀口 直也)
2022-09-05 7:29 ` Miaohe Lin
2022-09-05 21:53 ` Andrew Morton
2022-09-06 1:07 ` Miaohe Lin
2022-08-30 12:36 ` [PATCH 3/6] mm, hwpoison: use num_poisoned_pages_sub() to decrease num_poisoned_pages Miaohe Lin
2022-09-05 5:24 ` HORIGUCHI NAOYA(堀口 直也)
2022-08-30 12:36 ` [PATCH 4/6] mm, hwpoison: avoid unneeded page_mapped_in_vma() overhead in collect_procs_anon() Miaohe Lin
2022-09-05 5:24 ` HORIGUCHI NAOYA(堀口 直也)
2022-08-30 12:36 ` [PATCH 5/6] mm, hwpoison: check PageTable() explicitly in hwpoison_user_mappings() Miaohe Lin
2022-09-05 5:25 ` HORIGUCHI NAOYA(堀口 直也)
2022-08-30 12:36 ` [PATCH 6/6] mm, hwpoison: cleanup some obsolete comments Miaohe Lin
2022-09-05 5:25 ` HORIGUCHI NAOYA(堀口 直也)
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=1f7ee86e-7d28-0d8c-e0de-b7a5a94519e8@huawei.com \
--to=linmiaohe@huawei.com \
--cc=akpm@linux-foundation.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=naoya.horiguchi@nec.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;
as well as URLs for NNTP newsgroup(s).