Received-SPF: pass (google.com: domain of linux-kernel+bounces-201484-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.199.223 as permitted sender) client-ip=147.75.199.223;
Message-ID: <acb5d2789f0a2fa81f95c81d4c21d122cfddb244.camel@linux.intel.com>
Subject: Re: [PATCH] sched/balance: Skip unnecessary updates to idle load
 balancer's flags
From: Tim Chen <tim.c.chen@linux.intel.com>
To: Chen Yu <yu.c.chen@intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>, Ingo Molnar <mingo@redhat.com>, 
 Vincent Guittot <vincent.guittot@linaro.org>, linux-kernel@vger.kernel.org,
 Vinicius Gomes <vinicius.gomes@intel.com>
Date: Tue, 04 Jun 2024 15:32:11 -0700
In-Reply-To: <Zl53lXjEmnWhlW8p@chenyu5-mobl2>
References: <20240531205452.65781-1-tim.c.chen@linux.intel.com>
	 <Zlygeqy+SVs1KZYW@chenyu5-mobl2>
	 <c3d2a49e580cea9ae86e692f72094119310adc8f.camel@linux.intel.com>
	 <Zl53lXjEmnWhlW8p@chenyu5-mobl2>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
User-Agent: Evolution 3.44.4 (3.44.4-3.fc36) 
Precedence: bulk
MIME-Version: 1.0

On Tue, 2024-06-04 at 10:10 +0800, Chen Yu wrote:
>=20
> >=20
> > Not sure I follow your comments about return being safe.  Let me explai=
n
> > in details.
> >=20
> > We will return directly if and only if the bits set in flags are also s=
et
> > in nohz_flags(ilb_cpu). =20
> >=20
> > The comment's intention is to say that if the above statement is true, =
then
> > the later operation of=20
> >=20
> > 	atomic_fetch_or(flags, nohz_flags(ilb_cpu))
> >=20
> > will be useless and not result in any change to nohz_flags(ilb_cpu), si=
nce all the set bits
> > in flags are already set in nohz_flags(ilb_cpu).
>=20
> Understand. My previous thought was that, what if the current nohz_flags(=
ilb_cpu) is 0 or
> NOHZ_NEWILB_KICK. If yes, return directly might miss one ipi to the ilb_c=
pu(because
> the current code checks flags & NOHZ_KICK_MASK to return directly). But f=
rom the current
> logic when we reach kick_ilb(), the flag is not 0, and the flag passed by=
 nohz_balancer_kick()
> satisfy (flags & NOHZ_KICK_MASK), so returns here is ok.

Should not have issues of missing an IPI.  kick_ilb(flags) is always called=
 with non-zero flags
value.  And the first guy who update nohz_flags(ilb_cpu) successfully with =
bits
in the NOHZ_KICK_MASK will send the IPI.

Tim