Re: [PATCH v3 07/14] sched: Introduce restrict_cpus_allowed_ptr() to limit task CPU affinity

From: Will Deacon <will@kernel.org>
To: Valentin Schneider <valentin.schneider@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>,
	linux-arm-kernel@lists.infradead.org, linux-arch@vger.kernel.org,
	linux-kernel@vger.kernel.org,
	Catalin Marinas <catalin.marinas@arm.com>,
	Marc Zyngier <maz@kernel.org>,
	Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
	Morten Rasmussen <morten.rasmussen@arm.com>,
	Qais Yousef <qais.yousef@arm.com>,
	Suren Baghdasaryan <surenb@google.com>,
	Quentin Perret <qperret@google.com>, Tejun Heo <tj@kernel.org>,
	Li Zefan <lizefan@huawei.com>,
	Johannes Weiner <hannes@cmpxchg.org>,
	Ingo Molnar <mingo@redhat.com>,
	Juri Lelli <juri.lelli@redhat.com>,
	Vincent Guittot <vincent.guittot@linaro.org>,
	kernel-team@android.com
Subject: Re: [PATCH v3 07/14] sched: Introduce restrict_cpus_allowed_ptr() to limit task CPU affinity
Date: Thu, 19 Nov 2020 19:25:51 +0000	[thread overview]
Message-ID: <20201119192550.GD4906@willie-the-truck> (raw)
In-Reply-To: <jhj4kllwahv.mognet@arm.com>

On Thu, Nov 19, 2020 at 04:57:22PM +0000, Valentin Schneider wrote:
> 
> On 19/11/20 16:09, Peter Zijlstra wrote:
> > On Thu, Nov 19, 2020 at 01:13:20PM +0000, Will Deacon wrote:
> >
> >> Sure, but I was talking about what userspace sees, and I don't think it ever
> >> sees CPUs that have been hotplugged off, right? That is, sched_getaffinity()
> >> masks its result with the active_mask.
> >
> > # for i in /sys/devices/system/cpu/cpu*/online; do echo -n $i ":"; cat $i; done
> > /sys/devices/system/cpu/cpu1/online :0
> > /sys/devices/system/cpu/cpu2/online :1
> > /sys/devices/system/cpu/cpu3/online :1
> > /sys/devices/system/cpu/cpu4/online :1
> > /sys/devices/system/cpu/cpu5/online :1
> > /sys/devices/system/cpu/cpu6/online :1
> > /sys/devices/system/cpu/cpu7/online :1
> >
> > # grep Cpus_allowed /proc/self/status
> > Cpus_allowed:   ff
> > Cpus_allowed_list:      0-7
> >
> >
> > :-)
> 
> Harumph, so there is that...
> 
> $ while true; do continue; done &
> $ PID=$!
> $ taskset -pc 0-1 $PID
>   pid 849's current affinity list: 0-5
>   pid 849's new affinity list: 0,1
> $ echo 0 > /sys/devices/system/cpu/cpu1/online
>   [12578.545726] CPU1: shutdown
>   [12578.548454] psci: CPU1 killed (polled 0 ms)
> $ taskset -pc $PID
>   pid 849's current affinity list: 0
> $ cat /proc/$PID/status | grep Cpus
>   Cpus_allowed:   03
>   Cpus_allowed_list:      0-1

Yeah, I'm not sure this is worth tackling tbh. sched_getaffinity() does the
right thing, but poking around in /proc and /sys is always going to defeat
the illusion and I don't see what we gain in reporting CPUs on which the
task is _never_ going to run anyway. But I'll revise my stance on it being
identical to hotplug :) (I would've gotten away with it too, if it wasn't
for those pesky hackers).

Will