Received-SPF: pass (google.com: domain of linux-kernel+bounces-14063-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.80.249 as permitted sender) client-ip=147.75.80.249;
Precedence: bulk
MIME-Version: 1.0
References: <20231016071245.2865233-1-zhaoyang.huang@unisoc.com>
 <20231107172820.GA3745089@cmpxchg.org> <CAGWkznGbGciTocrqMowepCc=aStYFHPt8RPGYjWxQ4TRG1ZqXQ@mail.gmail.com>
 <20231229114032.64d5320a97a7c776e10e8acf@linux-foundation.org>
In-Reply-To: <20231229114032.64d5320a97a7c776e10e8acf@linux-foundation.org>
From: Zhaoyang Huang <huangzhaoyang@gmail.com>
Date: Tue, 2 Jan 2024 13:50:16 +0800
Message-ID: <CAGWkznFLP7ACf0mxkSoLefewzaoRL38RJ13PXvFmZCgcvx+63A@mail.gmail.com>
Subject: Re: [PATCHv6 1/1] mm: optimization on page allocation when CMA enabled
To: Andrew Morton <akpm@linux-foundation.org>
Cc: Johannes Weiner <hannes@cmpxchg.org>, "zhaoyang.huang" <zhaoyang.huang@unisoc.com>, 
	Roman Gushchin <roman.gushchin@linux.dev>, linux-mm@kvack.org, linux-kernel@vger.kernel.org, 
	steve.kang@unisoc.com, Vlastimil Babka <vbabka@suse.cz>, 
	Mel Gorman <mgorman@techsingularity.net>, Joonsoo Kim <iamjoonsoo.kim@lge.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Sat, Dec 30, 2023 at 3:40=E2=80=AFAM Andrew Morton <akpm@linux-foundatio=
n.org> wrote:
>
> On Wed, 8 Nov 2023 16:55:19 +0800 Zhaoyang Huang <huangzhaoyang@gmail.com=
> wrote:
>
> > > > +static bool use_cma_first(struct zone *zone, unsigned int order, u=
nsigned int alloc_flags)
> > > > +{
> > > > +     unsigned long watermark;
> > > > +     bool cma_first =3D false;
> > > > +
> > > > +     watermark =3D wmark_pages(zone, alloc_flags & ALLOC_WMARK_MAS=
K);
> > > > +     /* check if GFP_MOVABLE pass previous zone_watermark_ok via t=
he help of CMA */
> > > > +     if (zone_watermark_ok(zone, order, watermark, 0, alloc_flags =
& (~ALLOC_CMA))) {
> > > > +             /*
> > > > +              * Balance movable allocations between regular and CM=
A areas by
> > > > +              * allocating from CMA when over half of the zone's f=
ree memory
> > > > +              * is in the CMA area.
> > > > +              */
> > ok, thanks for point out.
> > Could we simple it like this, which will mis-judge kmalloc within
> > ioctl as GFP_USER. I think it is ok as it is rare
> >              if (current_is_kswapd() || !current->mm)
> >                  gfp_flags =3D GFP_KERNEL;
> >              else
> >                  gfp_flags =3D GFP_USER;
> >             free_pages =3D zone_page_state(zone, NR_FREE_PAGES);
> >             free_pages -=3D zone->lowmem_reserve[gfp_zone(gfp_flags)];
> >             free_pages -=3D wmark_pages(zone, alloc_flags & ALLOC_WMARK=
_MASK);
> >             cma_first =3D free_pages > zone_page_state(zone, NR_FREE_PA=
GES) / 2);
> >
>
> This went all quiet.  Do we feel that "mm: optimization on page
> allocation when CMA enabled" should be merged as-is, or dropped in the
> expectation that something based on Johannes's suggestion will be
> developed?
I just establish a v6.6 environment and will provide comparison
results with and without the patch
>