From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-2.2 required=3.0 tests=HEADER_FROM_DIFFERENT_DOMAINS, MAILING_LIST_MULTI,SPF_HELO_NONE,SPF_PASS,USER_AGENT_SANE_1 autolearn=no autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id 617C3C33CAF for ; Tue, 14 Jan 2020 02:09:26 +0000 (UTC) Received: from kanga.kvack.org (kanga.kvack.org [205.233.56.17]) by mail.kernel.org (Postfix) with ESMTP id 0719824672 for ; Tue, 14 Jan 2020 02:09:25 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 0719824672 Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=linux.intel.com Authentication-Results: mail.kernel.org; spf=pass smtp.mailfrom=owner-linux-mm@kvack.org Received: by kanga.kvack.org (Postfix) id 589618E0005; Mon, 13 Jan 2020 21:09:25 -0500 (EST) Received: by kanga.kvack.org (Postfix, from userid 40) id 53A988E0003; Mon, 13 Jan 2020 21:09:25 -0500 (EST) X-Delivered-To: int-list-linux-mm@kvack.org Received: by kanga.kvack.org (Postfix, from userid 63042) id 4769A8E0005; Mon, 13 Jan 2020 21:09:25 -0500 (EST) X-Delivered-To: linux-mm@kvack.org Received: from forelay.hostedemail.com (smtprelay0160.hostedemail.com [216.40.44.160]) by kanga.kvack.org (Postfix) with ESMTP id 34AB98E0003 for ; Mon, 13 Jan 2020 21:09:25 -0500 (EST) Received: from smtpin25.hostedemail.com (10.5.19.251.rfc1918.com [10.5.19.251]) by forelay03.hostedemail.com (Postfix) with SMTP id DD7F98248068 for ; Tue, 14 Jan 2020 02:09:24 +0000 (UTC) X-FDA: 76374607848.25.bird84_6f32d549cd54e X-HE-Tag: bird84_6f32d549cd54e X-Filterd-Recvd-Size: 7849 Received: from mga04.intel.com (mga04.intel.com [192.55.52.120]) by imf36.hostedemail.com (Postfix) with ESMTP for ; Tue, 14 Jan 2020 02:09:23 +0000 (UTC) X-Amp-Result: UNKNOWN X-Amp-Original-Verdict: FILE UNKNOWN X-Amp-File-Uploaded: False Received: from fmsmga005.fm.intel.com ([10.253.24.32]) by fmsmga104.fm.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 13 Jan 2020 18:09:22 -0800 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.69,431,1571727600"; d="scan'208";a="422998128" Received: from unknown (HELO localhost) ([10.239.159.54]) by fmsmga005.fm.intel.com with ESMTP; 13 Jan 2020 18:09:21 -0800 Date: Tue, 14 Jan 2020 10:09:29 +0800 From: Wei Yang To: Konstantin Khlebnikov Cc: Wei Yang , Li Xinhai , "linux-mm@kvack.org" , akpm , "linux-kernel@vger.kernel.org" , Rik van Riel , "kirill.shutemov" Subject: Re: [PATCH v2 1/2] mm/rmap: fix and simplify reusing mergeable anon_vma as parent when fork Message-ID: <20200114020929.GA6889@richard> Reply-To: Wei Yang References: <20200109025240.GA2000@richard> <20200110023029.GB16823@richard> <20200110112357351531132@gmail.com> <20200110053442.GA27846@richard> <20200111223820.GA15506@richard> <20200113003343.GA27210@richard> <1cf002fa-a3cb-bcef-57dc-ac9c09dcf2eb@yandex-team.ru> MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Disposition: inline In-Reply-To: <1cf002fa-a3cb-bcef-57dc-ac9c09dcf2eb@yandex-team.ru> User-Agent: Mutt/1.9.4 (2018-02-28) Content-Transfer-Encoding: quoted-printable X-Bogosity: Ham, tests=bogofilter, spamicity=0.000000, version=1.2.4 Sender: owner-linux-mm@kvack.org Precedence: bulk X-Loop: owner-majordomo@kvack.org List-ID: On Mon, Jan 13, 2020 at 02:07:18PM +0300, Konstantin Khlebnikov wrote: >On 13/01/2020 03.33, Wei Yang wrote: >> On Sun, Jan 12, 2020 at 12:55:45PM +0300, Konstantin Khlebnikov wrote: >> >=20 >> >=20 >> > On 12/01/2020 01.38, Wei Yang wrote: >> > > On Fri, Jan 10, 2020 at 11:11:23AM +0300, Konstantin Khlebnikov wr= ote: >> > > [...] >> > > > > > > >=20 >> > > > > > > > series of vma in parent with shared AV: >> > > > > > > >=20 >> > > > > > > > SRC1 - AV0 >> > > > > > > > SRC2 - AV0 >> > > > > > > > SRC3 - AV0 >> > > > > > > > ... >> > > > > > > > SRCn - AV0 >> > > > > > > >=20 >> > > > > > > > in child after fork >> > > > > > > >=20 >> > > > > > > > DST1 - AV_OLD_1 (some old vma, picked by anon_vma_clone)= plus DST1 is attached to same AVs as SRC1 >> > > > > > > > DST2 - AV_OLD_2 (other old vma) plus DST1 is attached to= same AVs as SRC2 >> > > > > > > > DST2 - AV1 prev AV parent does not match AV0, no old vma= found for reusing -> allocate new one (child of AV0) >> > > > > > > > DST3 - AV1 - DST2->AV->parent =3D=3D SRC3->AV (AV0) -> s= hare AV with prev >> > > > > > > > DST4 - AV1 - same thing >> > > > > > > > ... >> > > > > > > > DSTn - AV1 >> > > > > > > >=20 >> > >=20 >> > > To focus on the point, I rearranged the order a little. Suppose yo= ur following >> > > comments is explaining the above behavior. >> > >=20 >> > > I've illustrated how two heuristics (reusing-old and sharing-= prev) _could_ work together. >> > > But they both are optional. >> > > At cloning first vma SRC1 -> DST1 there is no prev to share a= non vma, >> > > thus works common code which _could_ reuse old vma because it= have to. >> > > If there is no old anon-vma which have to be reused then DST1= will allocate >> > > new anon-vma (AV1) and it will be used by DST2 and so on like= on your picture. >> > >=20 >> > > I agree with your 3rd paragraph, but confused with 2nd. >> > >=20 >> > > At cloning first vma SRC1 -> DST1, there is no prev so anon_vma_cl= one() would >> > > pick up a reusable anon_vma. Here you named it AV_OLD_1. This look= s good to >> > > me. But I am not sure why you would picked up AV_OLD_2 for DST2? I= n parent, >> > > SRC1 and SRC2 has the same anon_vma, AV0. So in child, DST1 and DS= T2 could >> > > also share the same anon_vma, AV_OLD_1. >> > >=20 >> > > Sorry for my poor understanding, would you mind giving me more hin= t on this >> > > change? >> >=20 >> > For DST2 heuristic "share-with-prev" will not work because if prev (= DST1) >> > uses old AV (AV_OLD_1) and AV_OLD_1->parent isn't SRC2->AV (AV0). >> > So DST2 could only pick another old AV or allocate new. >>=20 >> I know this behavior after your change, my question is why you want to= do so. > >Because I want to keep both heuristics. >This seems most sane way of interaction between them. > I am not sure this is more sane. Still suggest to separate your idea into a new patch, so audience could analysis and notice the change clearly. Otherwise audience would be confu= sed with this behavior. >Unfortunately even this patch is slightly broken. >Condition prev->anon_vma->parent =3D=3D pvma->anon_vma doesn't guarantee= that >prev vma has the same set of anon-vmas like current vma. >I.e. anon_vma_clone(vma, prev) might be not enough for keeping connectiv= ity. >Building such case isn't trivial job but I see nothing that could preven= t it. > >>=20 >> >=20 >> > My patch uses condition dst->prev->anon_vma->parent =3D=3D src->anon= _vma rather >> > than obvious src->prev->anon_vma =3D=3D src->anon_vma because in thi= s way it >> > eliminates all unwanted corner cases and explicitly verifies that we= going to >> > share related anon-vma. >> >=20 >>=20 >> This do eliminates some corner case, but as you showed child and paren= t don't >> share the same AV topology. To keep the same AV topology is the purpos= e of my >> commit. >>=20 >> I agree you found some bug that previous commit doesn't do it is expec= ted. But >> since you change the design a little, I suggest you split this idea to= a >> separate patch so that reviewer and audience in the future could under= stand >> your approach clearly. Otherwise audience would be confused and hard t= o track >> this change. >>=20 >> For example, you describe the behavior after your change. The second v= ma would >> probably have a different AV from first vma. >>=20 >> > Heuristic "reuse-old" uses fact that VMA links and AV parent chain a= re tracked >> > independently: when VMA reuses old AV it still links to all related = AV even >> > if VMA->AV points into some old AV in the middle of inheritance chai= n. >> >=20 >> > >=20 >> > > > > > >=20 >> > > > > > > Yes, your code works for DST3..DSTn. They will pick up AV1= since >> > > > > > > (DST2->AV->parent =3D=3D SRC3->AV). >> > > > > > >=20 >> > > > > > > My question is why DST1 and DST2 has different AV? The pur= pose of my patch >> > > > > > > tries to make child has the same topology and parent. So t= he ideal look of >> > > > > > > child is: >> > > > > > >=20 >> > > > > > > DST1 - AV1 >> > > > > > > DST2 - AV1 >> > > > > > > DST2 - AV1 >> > > > > > > DST3 - AV1 >> > > > > > > DST4 - AV1 >> > > > > > >=20 >> > > > > > > Would you mind putting more words on DST1 and DST2? I didn= 't fully understand >> > > > > > > the logic here. >> > > > > > >=20 >> > > > > > > Thanks >> > > > > > >=20 >> > > > > >=20 >> > > > > > I think that the first version is doing the work as you expe= cted, but been >> > > > > > revised in second version, to limits the number of users of = reused old >> > > > > > anon(which=A0is picked=A0in anon_vma_clone() and keep the tr= ee structure. >> > > > > >=20 >> > > > >=20 >> > > > > Any reason to reduce the reuse? Maybe I lost some point. >> > > >=20 >> > > > >=20 >> > > > > > > -- >> > > > > > > Wei Yang >> > > > > > > Help you, Help me >> > > > >=20 >> > >=20 >>=20 --=20 Wei Yang Help you, Help me