[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20131016093308.GD21963@ulmo.nvidia.com>
Date: Wed, 16 Oct 2013 11:33:09 +0200
From: Thierry Reding <thierry.reding@...il.com>
To: Andrew Morton <akpm@...ux-foundation.org>
Cc: Naoya Horiguchi <n-horiguchi@...jp.nec.com>, linux-mm@...ck.org,
Matt Mackall <mpm@...enic.com>, Cliff Wickman <cpw@....com>,
KOSAKI Motohiro <kosaki.motohiro@...fujitsu.com>,
Johannes Weiner <hannes@...xchg.org>,
KAMEZAWA Hiroyuki <kamezawa.hiroyu@...fujitsu.com>,
Michal Hocko <mhocko@...e.cz>,
"Aneesh Kumar K.V" <aneesh.kumar@...ux.vnet.ibm.com>,
Pavel Emelyanov <xemul@...allels.com>,
linux-kernel@...r.kernel.org, Mark Brown <broonie@...nel.org>
Subject: Re: [PATCH 0/11] update page table walker
On Tue, Oct 15, 2013 at 01:43:17PM -0700, Andrew Morton wrote:
> On Mon, 14 Oct 2013 13:36:59 -0400 Naoya Horiguchi <n-horiguchi@...jp.nec.com> wrote:
>
> > Page table walker is widely used when you want to traverse page table
> > tree and do some work for the entries (and pages pointed to by them.)
> > This is a common operation, and keep the code clean and maintainable
> > is important. Moreover this patchset introduces caller-specific walk
> > control function which is helpful for us to newly introduce page table
> > walker to some other users. Core change comes from patch 1, so please
> > see it for how it's supposed to work.
> >
> > This patchset changes core code in mm/pagewalk.c at first in patch 1 and 2,
> > and then updates all of current users to make the code cleaner in patch
> > 3-9. Patch 10 changes the interface of hugetlb_entry(), I put it here to
> > keep bisectability of the whole patchset. Patch 11 applies page table walker
> > to a new user queue_pages_range().
>
> Unfortunately this is very incompatible with pending changes in
> fs/proc/task_mmu.c. Especially Kirill's "mm, thp: change
> pmd_trans_huge_lock() to return taken lock".
>
> Stephen will be away for a couple more weeks so I'll get an mmotm
> released and hopefully Thierry and Mark will scoop it up(?).
> Alternatively, http://git.cmpxchg.org/?p=linux-mmots.git;a=summary is
> up to date.
>
> Please take a look, decide what you think we should do?
Hi Andrew,
I haven't had the time to look at writing up the scripts to import the
mmotm into the linux-next trees that I create. From what I understand,
it might be unwise to just pull linux-mmots into linux-next because it
isn't very well tested. Then again, increasing test coverage is one of
the goals of linux-next. If you think it's safe to include linux-mmots
in linux-next I can easily do that.
Otherwise I'll see if I can resume work on the scripts I started to
import the mmotm. Stephen has also provided the scripts that he used,
but I haven't had much time to look at them in detail yet because of
other things that have been keeping me busy.
Thierry
Content of type "application/pgp-signature" skipped
Powered by blists - more mailing lists