Date: Tue, 29 Nov 2016 09:45:16 +0100
From: Michal Hocko <mhocko@kernel.org>
To: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: "linux-mm@kvack.org" <linux-mm@kvack.org>,
        "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>,
        Hugh Dickins <hughd@google.com>,
        Andrew Morton <akpm@linux-foundation.org>,
        Dave Hansen <dave.hansen@intel.com>,
        Andrea Arcangeli <aarcange@redhat.com>,
        Mel Gorman <mgorman@techsingularity.net>,
        Vlastimil Babka <vbabka@suse.cz>,
        Pavel Emelyanov <xemul@parallels.com>, Zi Yan <zi.yan@cs.rutgers.edu>,
        Balbir Singh <bsingharora@gmail.com>,
        "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
        Naoya Horiguchi <nao.horiguchi@gmail.com>
Subject: Re: [PATCH v2 04/12] mm: thp: introduce
 CONFIG_ARCH_ENABLE_THP_MIGRATION
Message-ID: <20161129084516.GB31671@dhcp22.suse.cz>
References: <1478561517-4317-1-git-send-email-n-horiguchi@ah.jp.nec.com>
 <1478561517-4317-5-git-send-email-n-horiguchi@ah.jp.nec.com>
 <20161128142154.GM14788@dhcp22.suse.cz>
 <20161129075006.GA15582@hori1.linux.bs1.fc.nec.co.jp>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20161129075006.GA15582@hori1.linux.bs1.fc.nec.co.jp>
User-Agent: Mutt/1.6.0 (2016-04-01)
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1156
Lines: 25

On Tue 29-11-16 07:50:06, Naoya Horiguchi wrote:
> On Mon, Nov 28, 2016 at 03:21:54PM +0100, Michal Hocko wrote:
> > On Tue 08-11-16 08:31:49, Naoya Horiguchi wrote:
> > > Introduces CONFIG_ARCH_ENABLE_THP_MIGRATION to limit thp migration
> > > functionality to x86_64, which should be safer at the first step.
> > 
> > Please make sure to describe why this has to be arch specific and what
> > are arches supposed to provide in order to enable this option.
> 
> OK, the below will be added in the future version:
> 
>   Thp migration is an arch-specific feature because it depends on the
>   arch-dependent behavior of non-present format of page table entry.
>   What you need to enable this option in other archs are:
>   - to define arch-specific transformation functions like __pmd_to_swp_entry()
>     and __swp_entry_to_pmd(),
>   - to make sure that arch-specific page table walking code can properly handle
>     !pmd_present case (gup_pmd_range() is a good example),
>   - (if your archs enables CONFIG_HAVE_ARCH_SOFT_DIRTY,) to define soft dirty
>     routines like pmd_swp_mksoft_dirty.

Thanks this is _much_ better!
-- 
Michal Hocko
SUSE Labs