Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
References: <1594789529-6206-1-git-send-email-iamjoonsoo.kim@lge.com>
 <332d620b-bfe3-3b69-931b-77e3a74edbfd@suse.cz> <CAAmzW4NbG0fCtU2mV83pRamUeOEqKKxGTpQK2zuDxzmoF2FVrg@mail.gmail.com>
 <6f18d999-4518-31ce-4cea-9b5b89a577ad@suse.cz> <CAAmzW4MLc8bmkYW1q1fL_WRFQHksX-oy9tS-s9Kb-A=ZEeGETQ@mail.gmail.com>
 <5a8b13d5-da40-7b1b-2968-e6701001cc0e@suse.cz> <c2edfd71-eee5-3d73-a9ee-0c384a3440be@suse.cz>
In-Reply-To: <c2edfd71-eee5-3d73-a9ee-0c384a3440be@suse.cz>
From:   Joonsoo Kim <js1304@gmail.com>
Date:   Fri, 17 Jul 2020 18:12:15 +0900
Message-ID: <CAAmzW4PSUsuK4G4mSReksnrj7wq1do=yY=VfvZVMQd-ghwcVKQ@mail.gmail.com>
Subject: Re: [PATCH 1/4] mm/page_alloc: fix non cma alloc context
To:     Vlastimil Babka <vbabka@suse.cz>
Cc:     Andrew Morton <akpm@linux-foundation.org>,
        Linux Memory Management List <linux-mm@kvack.org>,
        LKML <linux-kernel@vger.kernel.org>, kernel-team@lge.com,
        Christoph Hellwig <hch@infradead.org>,
        Roman Gushchin <guro@fb.com>,
        Mike Kravetz <mike.kravetz@oracle.com>,
        Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>,
        Michal Hocko <mhocko@suse.com>,
        "Aneesh Kumar K . V" <aneesh.kumar@linux.ibm.com>,
        Joonsoo Kim <iamjoonsoo.kim@lge.com>, stable@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

2020=EB=85=84 7=EC=9B=94 17=EC=9D=BC (=EA=B8=88) =EC=98=A4=ED=9B=84 5:15, V=
lastimil Babka <vbabka@suse.cz>=EB=8B=98=EC=9D=B4 =EC=9E=91=EC=84=B1:
>
> On 7/17/20 10:10 AM, Vlastimil Babka wrote:
> > On 7/17/20 9:29 AM, Joonsoo Kim wrote:
> >> 2020=EB=85=84 7=EC=9B=94 16=EC=9D=BC (=EB=AA=A9) =EC=98=A4=ED=9B=84 4:=
45, Vlastimil Babka <vbabka@suse.cz>=EB=8B=98=EC=9D=B4 =EC=9E=91=EC=84=B1:
> >>>
> >>> On 7/16/20 9:27 AM, Joonsoo Kim wrote:
> >>> > 2020=EB=85=84 7=EC=9B=94 15=EC=9D=BC (=EC=88=98) =EC=98=A4=ED=9B=84=
 5:24, Vlastimil Babka <vbabka@suse.cz>=EB=8B=98=EC=9D=B4 =EC=9E=91=EC=84=
=B1:
> >>> >> >  /*
> >>> >> >   * get_page_from_freelist goes through the zonelist trying to a=
llocate
> >>> >> >   * a page.
> >>> >> > @@ -3706,6 +3714,8 @@ get_page_from_freelist(gfp_t gfp_mask, uns=
igned int order, int alloc_flags,
> >>> >> >       struct pglist_data *last_pgdat_dirty_limit =3D NULL;
> >>> >> >       bool no_fallback;
> >>> >> >
> >>> >> > +     current_alloc_flags(gfp_mask, &alloc_flags);
> >>> >>
> >>> >> I don't see why to move the test here? It will still be executed i=
n the
> >>> >> fastpath, if that's what you wanted to avoid.
> >>> >
> >>> > I want to execute it on the fastpath, too. Reason that I moved it h=
ere
> >>> > is that alloc_flags could be reset on slowpath. See the code where
> >>> > __gfp_pfmemalloc_flags() is on. This is the only place that I can a=
pply
> >>> > this option to all the allocation paths at once.
> >>>
> >>> But get_page_from_freelist() might be called multiple times in the sl=
owpath, and
> >>> also anyone looking for gfp and alloc flags setup will likely not exa=
mine this
> >>> function. I don't see a problem in having it in two places that alrea=
dy deal
> >>> with alloc_flags setup, as it is now.
> >>
> >> I agree that anyone looking alloc flags will miss that function easily=
. Okay.
> >> I will place it on its original place, although we now need to add one
> >> more place.
> >> *Three places* are gfp_to_alloc_flags(), prepare_alloc_pages() and
> >> __gfp_pfmemalloc_flags().
> >
> > Hm the check below should also work for ALLOC_OOM|ALLOC_NOCMA then.
> >
> > /* Avoid allocations with no watermarks from looping endlessly */
> >    if (tsk_is_oom_victim(current) &&
> >         (alloc_flags =3D=3D ALLOC_OOM ||
> >          (gfp_mask & __GFP_NOMEMALLOC)))
> >             goto nopage;
> >
> > Maybe it's simpler to change get_page_from_freelist() then. But documen=
t well.
>
> But then we have e.g. should_reclaim_retry() which calls __zone_watermark=
_ok()
> where ALLOC_CMA plays a role too, so that means we should have alloc_mask=
 set up
> correctly wrt ALLOC_CMA at the __alloc_pages_slowpath() level...

Good catch! Hmm... Okay. It would be necessarily handled in three places.
I will fix it on the next version. Anyway, we need some clean-up about
alloc_flags
handling since it looks not good for maintenance.

Thanks.