From: Minchan Kim <minchan.kim@gmail.com>
To: Michel Lespinasse <walken@google.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>,
Rik van Riel <riel@redhat.com>, Hugh Dickins <hughd@google.com>,
Andrew Morton <akpm@linux-foundation.org>,
linux-mm <linux-mm@kvack.org>
Subject: Re: THP, rmap and page_referenced_one()
Date: Mon, 7 Mar 2011 17:35:37 +0900 [thread overview]
Message-ID: <AANLkTinncv11r3cJnOr0HWZyaSu5NQMz6pEYThMkmFd0@mail.gmail.com> (raw)
In-Reply-To: <AANLkTikJpr9H2NJHyw_uajL=Ef_p16L3QYgmJSfFynSZ@mail.gmail.com>
On Mon, Mar 7, 2011 at 3:50 PM, Michel Lespinasse <walken@google.com> wrote:
> Hi,
>
> I have been wondering about the following:
>
> Before the THP work, the if (vma->vm_flags & VM_LOCKED) test in
> page_referenced_one() was placed after the page_check_address() call,
> but now it is placed above it. Could this be a problem ?
>
> My understanding is that the page_check_address() check may return
> false positives - for example, if an anon page was created before a
> process forked, rmap will indicate that the page could be mapped in
> both of the processes, even though one of them might have since broken
> COW. What would happen if the child process mlocks the corresponding
> VMA ? my understanding is that this would break COW, but not cause
> rmap to be updated, so the parent's page would still be marked in rmap
> as being possibly mapped in the children's VM_LOCKED vma. With the
> VM_LOCKED check now placed above the page_check_address() call, this
> would cause vmscan to see both the parent's and the child's pages as
> being unevictable.
I agree.
There are two processes called P_A, P_B.
P_B is child of P_A.
A page "page A" is share between V_A(A's VMA)and V_B(B's VMA) since
P_B is created by forking from P_A. When P_B calls mlock the V_B, P_B
allocates new page B instead of reusing page A by COW and mapped P_B's
page table but rmap of page A still indicates page A is mapped by V_A
and V_B.
The page_check_address can filter this situation that V_B doesn't
include page A any more.
So page_check_address should be placed before checking the VM_LOCKED.
I think it's valuable to add the comment why we need
page_check_address should be placed before the checking VM_LOCKED.
--
Kind regards,
Minchan Kim
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
next prev parent reply other threads:[~2011-03-07 8:35 UTC|newest]
Thread overview: 11+ messages / expand[flat|nested] mbox.gz Atom feed top
2011-03-07 6:50 THP, rmap and page_referenced_one() Michel Lespinasse
2011-03-07 7:33 ` KOSAKI Motohiro
2011-03-07 8:35 ` Minchan Kim [this message]
2011-03-07 10:35 ` Michel Lespinasse
2011-03-08 11:32 ` Andrea Arcangeli
2011-03-08 12:21 ` Michel Lespinasse
2011-03-08 12:58 ` Andrea Arcangeli
2011-03-08 22:48 ` Minchan Kim
2011-03-09 9:18 ` Johannes Weiner
2011-03-08 13:57 ` Rik van Riel
2011-03-09 7:11 ` KOSAKI Motohiro
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=AANLkTinncv11r3cJnOr0HWZyaSu5NQMz6pEYThMkmFd0@mail.gmail.com \
--to=minchan.kim@gmail.com \
--cc=aarcange@redhat.com \
--cc=akpm@linux-foundation.org \
--cc=hughd@google.com \
--cc=linux-mm@kvack.org \
--cc=riel@redhat.com \
--cc=walken@google.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).