Date: Sat, 29 Jul 2006 21:05:29 +0200 From: =?ISO-8859-1?Q?Sten_Daniel_S=F8rsdal?= <lists@wm-access.no> To: Peter Jeremy <peterjeremy@optushome.com.au> Cc: freebsd-current@freebsd.org Subject: Re: vmstat's entries type Message-ID: <44CBB179.6070904@wm-access.no> In-Reply-To: <20060728210154.GC748@turion.vk2pj.dyndns.org> References: <200607251254.k6PCsBef092737@lurza.secnetix.de> <200607271058.13055.jhb@freebsd.org> <20060728121525.GA44917@uk.tiscali.com> <200607280928.36573.jhb@freebsd.org> <20060728134701.GA45273@uk.tiscali.com> <20060728210154.GC748@turion.vk2pj.dyndns.org>
next in thread | previous in thread | raw e-mail | index | archive | help
Peter Jeremy wrote:
> On Fri, 2006-Jul-28 14:47:01 +0100, Brian Candler wrote:
>> On Fri, Jul 28, 2006 at 09:28:36AM -0400, John Baldwin wrote:
>>> lock incl counter
>>> jnc 1f
>>> lock incl counter+4
>>> 1:
>=20
> This approach still requires the reader to loop with something like
> do {
> a.lo =3D counter.lo;
> a.hi =3D counter.hi;
> b.lo =3D counter.lo;
> b.hi =3D counter.hi;
> } while (a.hi !=3D b.hi || a.lo > b.lo);
> to ensure that the reader doesn't read the middle of an update.
>=20
>> The 'polling' argument says just do
>> lock incl counter
>> and poll all counters every 5 minutes, looking for a wrap. I think tha=
t's
>> almost certainly going to be cheaper, as long as you can keep track of=
where
>> all these counters are located.
>=20
> lock prefixes are always going to be extremely expensive on a MP
> system because they require physical bus cycles. RISC architectures
> usually only have TAS lock primitives (because "inc mem" doesn't
> exist) and so require a spinlock to perform an atomic update.
>=20
> In a MP configuration where it doesn't particularly matter if a
> particular update gets counted this time or next time, I think the
> cheapest option is to have per-CPU 32-bit counters (so no locks are
> needed to update the counters) with a polling function to accumulate
> all the individual counters into a 64-bit total. This pushes the cost
> from the update (very frequent) into the read (which is relatively
> infrequent), for a lower overall cost.
>=20
> This turns the update into something like:
> PCPU_SET(counter, PCPU_GET(counter)+1);
> or
> incl %fs:counter
> (no locks or atomic operations)
>=20
> Whilst the poll/read pseudo code looks something like
> lock counter
> foreach cpu {
> uint32 a =3D cpu->counter;
> uint32 b =3D cpu->last_counter;
> uint32 c =3D counter.lo;
> if (b > a)
> counter.hi++;
> counter.lo +=3D a - b;
> if (counter.lo < c)
> counter.hi++;
> cpu->last_counter =3D a;
> }
> unlock counter;
> (the lock prevents multiple readers updating counter simultaneously).
>=20
> You execute this whenever a reader wants the counter value (eg via
> SYSCTL_PROC), as well as a rate sufficient to prevent missing wraps
> (eg every 2 seconds for a 10g byte counter). This rate is sufficiently=
> lower than the update rate to make the whole exercise worthwhile.
>=20
Is caching necessary somewhere or can the function return the value
directly without storing the global accumulated counter?
( trying to get an understanding )
--=20
Sten Daniel S=F8rsdal
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?44CBB179.6070904>