linux-kernel - Re: [PATCH 4/5] mm: compaction: Determine if dirty pages can be migreated without blocking within ->migratepage

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20111121111931.GB19415@suse.de>
Date:	Mon, 21 Nov 2011 11:19:31 +0000
From:	Mel Gorman <mgorman@...e.de>
To:	Nai Xia <nai.xia@...il.com>
Cc:	Linux-MM <linux-mm@...ck.org>,
	Andrea Arcangeli <aarcange@...hat.com>,
	Minchan Kim <minchan.kim@...il.com>, Jan Kara <jack@...e.cz>,
	Andy Isaacson <adi@...apodia.org>,
	Johannes Weiner <jweiner@...hat.com>,
	LKML <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH 4/5] mm: compaction: Determine if dirty pages can be
 migreated without blocking within ->migratepage

On Sat, Nov 19, 2011 at 04:59:10PM +0800, Nai Xia wrote:
> > <SNIP>
> > @@ -453,13 +494,18 @@ int buffer_migrate_page(struct address_space *mapping,
> >        if (rc)
> >                return rc;
> >
> > -       bh = head;
> > -       do {
> > -               get_bh(bh);
> > -               lock_buffer(bh);
> > -               bh = bh->b_this_page;
> > -
> > -       } while (bh != head);
> > +       if (!buffer_migrate_lock_buffers(head, sync)) {
> > +               /*
> > +                * We have to revert the radix tree update. If this returns
> > +                * non-zero, it either means that the page count changed
> > +                * which "can't happen" or the slot changed from underneath
> > +                * us in which case someone operated on a page that did not
> > +                * have buffers fully migrated which is alarming so warn
> > +                * that it happened.
> > +                */
> > +               WARN_ON(migrate_page_move_mapping(mapping, page, newpage));
> > +               return -EBUSY;
> 
> If this migrate_page_move_mapping() really fails, seems disk IO will be needed
> to bring the previously already cached page back,

Aside from that, I couldn't see a way of handling the case where the
page had an elevated count due to a speculative lookup.

> I wonder if we should make the
> double check for the two conditions of "page refs is ok " and "all bh
> trylocked"
> before doing radix_tree_replace_slot() ? which I think does not
> involve IO on the
> error path.
> 

I reached the same conclusion when figuring out how to backout of the
the elevated page count case. In an updated patch,
migrate_page_move_mapping() returns with buffers locked in the async
case.

-- 
Mel Gorman
SUSE Labs
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/