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, 7 Aug 2020 16:11:18 +0200
From:   peterz@infradead.org
To:     luca abeni <luca.abeni@santannapisa.it>
Cc:     Juri Lelli <juri.lelli@redhat.com>, mingo@redhat.com,
        rostedt@goodmis.org, tglx@linutronix.de,
        linux-kernel@vger.kernel.org, tommaso.cucinotta@santannapisa.it,
        alessio.balsini@gmail.com, bristot@redhat.com,
        dietmar.eggemann@arm.com, linux-rt-users@vger.kernel.org,
        mtosatti@redhat.com, williams@redhat.com,
        valentin.schneider@arm.com
Subject: Re: [RFC PATCH v2 6/6] sched/fair: Implement starvation monitor
Message-ID: <20200807141118.GK2674@hirez.programming.kicks-ass.net>
References: <20200807095051.385985-1-juri.lelli@redhat.com>
 <20200807095604.GO42956@localhost.localdomain>
 <20200807104618.GH2674@hirez.programming.kicks-ass.net>
 <20200807154941.2bb11408@nowhere>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20200807154941.2bb11408@nowhere>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Fri, Aug 07, 2020 at 03:49:41PM +0200, luca abeni wrote:
> Hi Peter,
> 
> peterz@infradead.org wrote:

> > One thing I considerd was scheduling this as a least-laxity entity --
> > such that it runs late, not early
> 
> Are you thinking about scheduling both RT and non-RT tasks through
> deadline servers? If yes,

Maybe, I initially considered this for mixed criticality, where the
'soft' class would run EDF and the 'hard' class would run LLF (or the
other way around, I can't quite remember how I figured it).

If you restrict the hard class to single CPU assignment (IOW the UP
case) and ensure that u_llf + U_gedf/N < 1, it should just work out.

But I shelved all that after I heard about that other balancer idea
Danial was suppose to be working on ;-)))

> then I think that using something like
> laxity-based scheduling for the SCHED_OTHER server can be a good idea
> (but then we need to understand how to combine deadline-based
> scheduling with laxity-based scheduling, etc...)

/me consults notes, EDZL is I think the closest thing there.

> Or are you thinking about keeping the SCHED_OTHER server throttled
> until its laxity is 0 (or until its laxity is lower than some small
> value)? In this second case, the approach would work even if RT tasks
> are not scheduled through a server (but I do not know which kind of
> performance guarantee we could provide).

That would certainly be sufficient for OTHER servers I think.