Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Date:   Fri, 30 Apr 2021 13:14:38 +0000
From:   Quentin Perret <qperret@google.com>
To:     Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc:     Vincent Guittot <vincent.guittot@linaro.org>,
        Ingo Molnar <mingo@redhat.com>,
        Peter Zijlstra <peterz@infradead.org>,
        Juri Lelli <juri.lelli@redhat.com>,
        Steven Rostedt <rostedt@goodmis.org>,
        Ben Segall <bsegall@google.com>, Mel Gorman <mgorman@suse.de>,
        Daniel Bristot de Oliveira <bristot@redhat.com>,
        Qais Yousef <qais.yousef@arm.com>,
        Android Kernel Team <kernel-team@android.com>,
        linux-kernel <linux-kernel@vger.kernel.org>,
        Patrick Bellasi <patrick.bellasi@matbug.net>
Subject: Re: [PATCH v2] sched: Fix out-of-bound access in uclamp
Message-ID: <YIwCvkfrPGkyk17d@google.com>
References: <20210429152656.4118460-1-qperret@google.com>
 <CAKfTPtAfgVnMYkY_M+SDmNQDb_EsxSctQw-fkt2WJhczZakOjQ@mail.gmail.com>
 <YIu/EjZ8QsEl9sum@google.com>
 <CAKfTPtDOuB0UzZM5OtNzCUABftz_A40Z4W1z9OAORzFitfPDEQ@mail.gmail.com>
 <YIvQoG52Vk96DJQ/@google.com>
 <CAKfTPtB-0N7ogk0snCgj78zDrXRWgFUL=G7MoRDn+Bg8Hnk-PQ@mail.gmail.com>
 <b6737247-ca02-e197-70c7-020952d95c1b@arm.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <b6737247-ca02-e197-70c7-020952d95c1b@arm.com>
Precedence: bulk

On Friday 30 Apr 2021 at 15:00:00 (+0200), Dietmar Eggemann wrote:
> On 30/04/2021 14:03, Vincent Guittot wrote:
> IMHO, this asks for
> 
> min_t(unsigned int, clamp_value/UCLAMP_BUCKET_DELTA, UCLAMP_BUCKETS-1);

Yep, that's what I have locally.

> >>  }
> >>
> >>  static inline unsigned int uclamp_none(enum uclamp_id clamp_id)
> 
> Looks like this will fix a lot of possible configs:
> 
> nbr buckets 1-4, 7-8, 10-12, 14-17, *20*, 26, 29-32 ...
> 
> We would still introduce larger last buckets, right?

Indeed. The only better alternative I could see was to 'spread' the
error accross multiple buckets (e.g. make the last few buckets a bit
bigger instead of having all of it accumulated on the last one), but not
sure it is worth the overhead.

Suggestions are welcome though.

> Examples:
> 
> nbr_buckets 	delta	last bucket size
> 
> 20 		51	 +5 = 56
>
> 26		39	+10 = 49
> 
> 29		35	 +9 = 44

Yes the error can become worse in the relative sense with a large number
of buckets, but again the max is 20 so we should be fine?

Thanks,
Quentin