linux-kernel - Re: PATCH] cpuidle: A new variant of the menu governor to boost IO performance

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

Message-ID: <20090914044510.511d2ac9@infradead.org>
Date:	Mon, 14 Sep 2009 04:45:10 +0200
From:	Arjan van de Ven <arjan@...radead.org>
To:	Andrew Morton <akpm@...ux-foundation.org>
Cc:	linux-kernel@...r.kernel.org, lenb@...nel.org, mingo@...e.hu,
	peterz@...radead.org, yanmin_zhang@...ux.intel.com,
	torvalds@...ux-foundation.org, jens.axboe@...cle.com
Subject: Re: PATCH] cpuidle: A new variant of the menu governor to boost IO
 performance

On Sun, 13 Sep 2009 16:30:07 -0700
Andrew Morton <akpm@...ux-foundation.org> wrote:
> 
> We can't just fix up the existing code?

we could, it'd be just replacing menu.c.
(since the new one started out with menu.c anyway)
I don't mind either way, will replace. 

> > +#define BUCKETS 12
> > +#define RESOLUTION 1024
> > +#define DECAY 4
> > +#define MAX_INTERESTING 50000
> 
> Interesting!  Unobvious what it does though.

at least I documented it:
   * the actual factor is if there is (disk) IO outstanding or not.
   * (as a special twist, we consider every sleep longer than 50 milliseconds
   * as perfect; there is no power gains for sleeping longer than this)

> 
> That's DIV_ROUND_UP().

ok

> > +	new_factor = data->correction_factor[data->bucket]
> > +			* (DECAY - 1) / DECAY;
> 
> This code is available on 32-bit.  I wouldn't be surprised if we see
> some udivdi2-isn't-there problems with some compiler versions.  We'll
> see.

at least not on the ones I tried. DECAY is "4" which means it'll be
shifts anyway (that's not entirely accidental)

> > +
> > +unsigned long this_cpu_load(void)
> > +{
> > +	int this_cpu = smp_processor_id();
> > +	struct rq *this_rq = cpu_rq(this_cpu);
> > +	return this_rq->cpu_load[0];
> > +}
> 
> hm, how come we don't get smp_processor_id-in-preemptible warnings. 
> Are all current callers pinned to a CPU?

the idle loop is pinned by definition, and also runs with irqs off

"all current callers".. this is a new function, with menutng being the
only caller. I'm fine with requiring all callers to be pinned, it's
*this* cpu's load after all ;-)



> 
> 
> some spelling fixlets:

never been my strong point (esp when jetlagged); thanks!


-- 
Arjan van de Ven 	Intel Open Source Technology Centre
For development, discussion and tips for power savings, 
visit http://www.lesswatts.org
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/