linux-kernel - Re: [PATCH] sched_ext: Provide a sysfs enable

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <e93dca67d070beb548c771a8836110cd015b7c48@linux.dev>
Date: Mon, 23 Sep 2024 17:16:16 +0000
From: andrea.righi@...ux.dev
To: "Phil Auld" <pauld@...hat.com>, "Tejun Heo" <tj@...nel.org>
Cc: "David Vernet" <void@...ifault.com>, "Giovanni Gherdovich"
 <giovanni.gherdovich@...e.com>, "Kleber Sacilotto de Souza"
 <kleber.souza@...onical.com>, "Marcelo Henrique Cerri"
 <marcelo.cerri@...onical.com>, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] sched_ext: Provide a sysfs enable_seq counter

September 23, 2024 at 6:57 PM, "Phil Auld" <pauld@...hat.com> wrote:
> 
> On Mon, Sep 23, 2024 at 06:47:12AM -1000 Tejun Heo wrote:
> 
> > 
> > Hello,
> > 
> >  
> > 
> >  On Mon, Sep 23, 2024 at 06:34:20PM +0200, Andrea Righi wrote:
> > 
> >  ...
> > 
> >  > > Yes, the load sequence number should stay persistent across all schedulers,
> > 
> >  > > but each scheduler should report the sequence number at which *it* was
> > 
> >  > > loaded. Note that this doesn't really change anything now. If you only care
> > 
> >  > > whether any SCX scheduler has ever been loaded, you'd always look under
> > 
> >  > > root.
> > 
> >  > >
> > 
> >  > 
> > 
> >  > In my testing root is not there is nothing is loaded. 
> > 
> >  
> > 
> >  Ah, right.
> > 
> >  
> > 
> >  Right, there's no root if no sched_ext scheduler is loaded. Maybe we
> > 
> >  should always keep root present, or have a global counter and one
> > 
> >  per-sched?
> > 
> >  
> > 
> >  I'll apply as-is. Let's add per-scheduler load seq separately.
> > 
> 
> Thanks! I was thinking that per-scheduler you could just snapshot the
> 
> global counter (either before or after increment) on load. Then you
> 
> could easily tell when each was loaded relative to each other etc.
> 
> Especially while "per-scheduler" is defined by string comparison I'd
> 
> prefer not to rely on that for this use case. 

Right, and it should be a trivial change, I'll work on that as soon as I find a stable spot (still on the road for conferences) :)

-Andrea