Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Mon, 11 Apr 2022 07:08:46 -0700
From:   "Paul E. McKenney" <paulmck@kernel.org>
To:     Uladzislau Rezki <urezki@gmail.com>
Cc:     Joel Fernandes <joel@joelfernandes.org>,
        LKML <linux-kernel@vger.kernel.org>,
        Josh Triplett <josh@joshtriplett.org>,
        Lai Jiangshan <jiangshanlai@gmail.com>,
        Mathieu Desnoyers <mathieu.desnoyers@efficios.com>,
        rcu <rcu@vger.kernel.org>, Steven Rostedt <rostedt@goodmis.org>,
        frederic@kernel.org
Subject: Re: [PATCH RFC] rcu/nocb: Provide default all-CPUs mask for
 RCU_NOCB_CPU=y
Message-ID: <20220411140846.GA84069@paulmck-ThinkPad-P17-Gen-1>
Reply-To: paulmck@kernel.org
References: <20220407210734.2548973-1-joel@joelfernandes.org>
 <20220408142232.GA4285@paulmck-ThinkPad-P17-Gen-1>
 <CAEXW_YQWeqfcKdAKmCn4fFGyWXjOGd=29wvi6bL3k7s2bGkDJw@mail.gmail.com>
 <20220408155002.GF4285@paulmck-ThinkPad-P17-Gen-1>
 <CAEXW_YQDgSO2XkkVhN3RBBz3vwYdAtTuPz-xYYsAPnwEnbYZPA@mail.gmail.com>
 <20220408174908.GK4285@paulmck-ThinkPad-P17-Gen-1>
 <CAEXW_YQ+oE3xQ0tLnBMFxRXLqKZkT5UfjF+CULxnhf9F-dEA2g@mail.gmail.com>
 <CAEXW_YRK2t2JO4RyBTd8cR9sTVpgP7Z5Ywhb1g7CRz3HJ_kNQA@mail.gmail.com>
 <20220408205440.GL4285@paulmck-ThinkPad-P17-Gen-1>
 <YlCtJzlJDj1DBHQB@pc638.lan>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <YlCtJzlJDj1DBHQB@pc638.lan>
Precedence: bulk

On Fri, Apr 08, 2022 at 11:46:15PM +0200, Uladzislau Rezki wrote:
> > 
> > Here is what I believe you are proposing:
> > 
> > 
> > 				---	rcu_nocbs	rcu_nocbs=???
> > 
> > CONFIG_RCU_NOCB_CPU_ALL=n	[1]	[2]		[3]
> > 
> > CONFIG_RCU_NOCB_CPU_ALL=y	[4]	[4]		[3]
> > 
> > 
> > [1]	No CPUs are offloaded at boot.	CPUs cannot be offloaded at
> > 	runtime.
> > 
> > [2]	No CPUs are offloaded at boot, but any CPU can be offloaded
> > 	(and later de-offloaded) at runtime.
> > 
> > [3]	The set of CPUs that are offloaded at boot are specified by the
> > 	mask, represented above with "???".  The CPUs that are offloaded
> > 	at boot can be de-offloaded and offloaded at runtime.  The CPUs
> > 	not offloaded at boot cannot be offloaded at runtime.
> > 
> > [4]	All CPUs are offloaded at boot, and any CPU can be de-offloaded
> > 	and offloaded at runtime.  This is the same behavior that
> > 	you would currently get with CONFIG_RCU_NOCB_CPU_ALL=n and
> > 	rcu_nocbs=0-N.
> > 
> > 
> > I am adding Frederic on CC, who will not be shy about correcting any
> > confusion I be suffering from have with respect to the current code.
> > 
> > Either way, if this is not what you had in mind, what are you suggesting
> > instead?
> > 
> > I believe that Steve Rostedt's review would carry weight for ChromeOS,
> > however, I am suffering a senior moment on the right person for Android.
> > 
> We(in Sony) mark all CPUs as offloaded ones because of power reasons. The
> energy aware scheduler has a better knowledge where to place an rcuop/x
> task to invoke callbacks. The decision is taken based on many reason and
> the most important is to drain less power as a result of task placement.
> For example, power table, if OPP becomes higher or not, CPU is idle, etc.
> 
> What Joel does in this patch sounds natural to me at least from the first
> glance. I mean converting the RCU_NOCB_CPU=y to make all CPUs to do offloading.

Just to be very clear, given appropriate acks/reviews, adding something
like CONFIG_RCU_NOCB_CPU_ALL to get default rcu_nocbs=0-N is fine.
However, Joel's original patch would not be good for the enterprise
distros, which rely on the current default.

							Thanx, Paul