From: Andrea Arcangeli <aarcange@redhat.com>
To: Michel Lespinasse <walken@google.com>
Cc: linux-mm <linux-mm@kvack.org>
Subject: Re: get_page() vs __split_huge_page_refcount()
Date: Fri, 25 Mar 2011 17:48:47 +0100 [thread overview]
Message-ID: <20110325164847.GD431@random.random> (raw)
In-Reply-To: <AANLkTinHBouEU2pAVOfuakxYqA_QFVLz=qY-f8ZW6fTG@mail.gmail.com>
Hi Michel,
On Thu, Mar 24, 2011 at 10:00:16PM -0700, Michel Lespinasse wrote:
> My question is this: After someone obtains a page reference using
> get_user_pages(), what prevents them from getting additional
> references with get_page() ? I always thought it was legal to
> duplicate references that way, but now I don't see how it'd be safe
> doing so on anon pages with THP enabled.
It's not legal anymore as you noticed, but I'm not aware of anything
doing that. I don't see an useful case where a driver could need to
take one extra refcount after GUP returned. The normal API is
GUP/put_page. We could make it legal again by taking the compound_lock
after a PageCompound check though. I hope it's not needed though. It's
unavoidable in put_page because put_page will run out of order with
regard to __split_huge_page_refcount. But serializing get_page in GUP
against __split_huge_page_refcount is automatic through the
pmd_trans_splitting bit and needed for all page table walkers anyway.
Maybe it's good idea to add a comment to transhuge.txt about that? I
don't think I added it.
Grepping for get_page in drivers doesn't show too many, they mostly
run through the vm_ops->fault handler. Most important I can't see how
possibly it could be useful to run a get_page after
get_user_pages(FOLL_GET) returns.
--
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>
prev parent reply other threads:[~2011-03-25 16:48 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2011-03-25 5:00 get_page() vs __split_huge_page_refcount() Michel Lespinasse
2011-03-25 16:48 ` Andrea Arcangeli [this message]
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=20110325164847.GD431@random.random \
--to=aarcange@redhat.com \
--cc=linux-mm@kvack.org \
--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).