Date: Thu, 27 Jul 2006 10:58:12 -0400 From: John Baldwin <jhb@freebsd.org> To: freebsd-current@freebsd.org Cc: Oliver Fromme <olli@lurza.secnetix.de> Subject: Re: vmstat's entries type Message-ID: <200607271058.13055.jhb@freebsd.org> In-Reply-To: <200607251254.k6PCsBef092737@lurza.secnetix.de> References: <200607251254.k6PCsBef092737@lurza.secnetix.de>
next in thread | previous in thread | raw e-mail | index | archive | help
On Tuesday 25 July 2006 08:54, Oliver Fromme wrote: > John Baldwin <jhb@freebsd.org> wrote: > > On Sunday 23 July 2006 20:03, Sten Daniel S=F8rsdal wrote: > > > sthaug@nethelp.no wrote: > > > > > > One approach that we could use for 64-bit counters would be to= =20 just > > > > > > use 32-bits one, and poll them for overflow and bump an overfl= ow > > > > > > count. This assumes that the 32-bit counters overflow much le= ss=20 often > > > > > > than the polling interval, and easily triples the amount of=20 storage > > > > > > for each of them... It is ugly :-( > > > > > >=20 > > > > > What's wrong with the add+adc (asm) approach found on any i386? > > > >=20 > > > > Presumably the fact that add + adc isn't an atomic operation. So if > > > > you want to guarantee 64 bit consistency, you need locking or=20 similar. > > > >=20 > > >=20 > > > Would it not be necessary to do this locking anyway? > > > I don't see how polling for overflow would help this consistency. > > > Are both suggestions insufficient? > >=20 > > I actually think that add + adc is ok for the case of incrementing sim= ple=20 > > counters. You can even do 'inc ; addc $0' >=20 > (I'm familiar with asm programming, but I'm not a low-level > threading or SMP expert, so please excuse me if this is a > dumb question ...) >=20 > If you just do add+adc (or inc+adc) and another thread (on > the same or different processor, I don't know) happens to > read the counter value at the same time (i.e. after the > lower 32bit have overflowed, but before the upper 32bit get > incremented), then that other thread would get a value > that's off by 2^32. >=20 > What am I missing? That these counters are for stats. :) You always have a race when reading = the=20 amount, so you can choose what is "good enough" to satisfy the conflicting= =20 requirements of "cheap" and "accurate". To me, the cheapness of add+adc=20 (compared to say, a cmpxchg8b loop with a branch, etc.) is worth it if you= =20 have this rare race. =2D-=20 John Baldwin
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200607271058.13055.jhb>