Received-SPF: pass (google.com: domain of linux-kernel+bounces-160842-linux.lists.archive=gmail.com@vger.kernel.org designates 139.178.88.99 as permitted sender) client-ip=139.178.88.99;
Precedence: bulk
MIME-Version: 1.0
References: <20240427003733.3898961-1-shakeel.butt@linux.dev>
 <20240427003733.3898961-4-shakeel.butt@linux.dev> <CAJD7tkY2EwtOHh3w=e0+OC3gOPb9xCC7C36BhM0RHYPv184U+Q@mail.gmail.com>
 <uewheydidpkiqywtc4m6dcmvlfj7raocpngtbqocksxvf4nyz4@srxnazpqxvzy>
In-Reply-To: <uewheydidpkiqywtc4m6dcmvlfj7raocpngtbqocksxvf4nyz4@srxnazpqxvzy>
From: Yosry Ahmed <yosryahmed@google.com>
Date: Fri, 26 Apr 2024 18:18:50 -0700
Message-ID: <CAJD7tkbj1hdka1wJiRZA+nOXFb3YW3adN=CENp2F+FACm7N+Tg@mail.gmail.com>
Subject: Re: [PATCH v2 3/7] memcg: reduce memory for the lruvec and memcg stats
To: Shakeel Butt <shakeel.butt@linux.dev>
Cc: Andrew Morton <akpm@linux-foundation.org>, Johannes Weiner <hannes@cmpxchg.org>, 
	Michal Hocko <mhocko@kernel.org>, Roman Gushchin <roman.gushchin@linux.dev>, 
	Muchun Song <muchun.song@linux.dev>, linux-mm@kvack.org, linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Fri, Apr 26, 2024 at 6:16=E2=80=AFPM Shakeel Butt <shakeel.butt@linux.de=
v> wrote:
>
> On Fri, Apr 26, 2024 at 05:51:23PM -0700, Yosry Ahmed wrote:
> > On Fri, Apr 26, 2024 at 5:37=E2=80=AFPM Shakeel Butt <shakeel.butt@linu=
x.dev> wrote:
> > >
> > > At the moment, the amount of memory allocated for stats related struc=
ts
> > > in the mem_cgroup corresponds to the size of enum node_stat_item.
> > > However not all fields in enum node_stat_item has corresponding memcg
> > > stats. So, let's use indirection mechanism similar to the one used fo=
r
> > > memcg vmstats management.
> >
> > Just curious, does the indirection table cause any noticeable
> > performance impact? My guess is no, but the update paths are usually
> > very performance sensitive.
> >
> > I guess lkp will shout at us if there are any noticeable regressions.
>
> Yeah and that is the reason I made the indirection table smaller (i.e.
> int8_t) to keep more entries in a single cacheline.

Thanks for pointing this out, probably worth including in the commit log :)