From: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
To: Balbir Singh <bsingharora@gmail.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>,
        Michal Hocko <mhocko@kernel.org>, "Vlastimil Babka" <vbabka@suse.cz>,
        Pavel Emelyanov <xemul@parallels.com>,
        "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
        "Naoya Horiguchi" <nao.horiguchi@gmail.com>
Subject: Re: [PATCH v1 02/11] mm: thp: introduce
 CONFIG_ARCH_ENABLE_HUGEPAGE_MIGRATION
Thread-Topic: [PATCH v1 02/11] mm: thp: introduce
 CONFIG_ARCH_ENABLE_HUGEPAGE_MIGRATION
Thread-Index: AQHRdSAzgyBZIsY1k0mVFcl0zH0kqJ9Ml64AgABcboA=
Date: Mon, 7 Mar 2016 06:28:54 +0000
Message-ID: <20160307062853.GA31458@hori1.linux.bs1.fc.nec.co.jp>
References: <1456990918-30906-1-git-send-email-n-horiguchi@ah.jp.nec.com>
 <1456990918-30906-3-git-send-email-n-horiguchi@ah.jp.nec.com>
 <20160307005804.GA25148@cotter.ozlabs.ibm.com>
In-Reply-To: <20160307005804.GA25148@cotter.ozlabs.ibm.com>
Accept-Language: ja-JP, en-US
Content-Language: ja-JP
Content-Type: text/plain; charset="iso-2022-jp"
Content-ID: <D5CEA2EE74625F40B2934D3DA5B24C5E@gisp.nec.co.jp>
Content-Transfer-Encoding: 8BIT
MIME-Version: 1.0
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1091
Lines: 22

On Mon, Mar 07, 2016 at 11:58:04AM +1100, Balbir Singh wrote:
> On Thu, Mar 03, 2016 at 04:41:49PM +0900, Naoya Horiguchi wrote:
> > Introduces CONFIG_ARCH_ENABLE_HUGEPAGE_MIGRATION to limit thp migration
> > functionality to x86_64, which should be safer at the first step.
> >
> 
> The changelog is not helpful. Could you please describe what is
> architecture specific in these changes? What do other arches need to do
> to port these changes over?

The arch specific parts are pmd_present() and swap entry format. Currently
pmd_present() in x86_64 is not simple enough to easily determine pmd's state
(none, normal pmd entry pointing to pte page, pmd for thp, or pmd migration entry ...)
That requires me to assume in this version that pmd migration entry should
have_PAGE_PSE set, which should not be necessary if the complexity is fixed.
So I will mention this pmd_present() problem in the next version.

So if it's fixed, what developers need to do to port this feature to their
architectures is just to enable CONFIG_ARCH_ENABLE_THP_MIGRATION (and test it.)

Thanks,
Naoya Horiguchi