Date: Fri, 5 Dec 2008 14:18:49 -0800 From: "Kip Macy" <kmacy@freebsd.org> To: "Andrew Gallatin" <gallatin@cs.duke.edu> Cc: src-committers@freebsd.org, svn-src-user@freebsd.org Subject: Re: svn commit: r185649 - in user/kmacy/HEAD_fast_multi_xmit/sys: net sys Message-ID: <3c1674c90812051418q1a81ca0an970d410dd034447c@mail.gmail.com> In-Reply-To: <4939A723.80601@cs.duke.edu> References: <200812052116.mB5LGDDr073309@svn.freebsd.org> <4939A723.80601@cs.duke.edu>
next in thread | previous in thread | raw e-mail | index | archive | help
On Fri, Dec 5, 2008 at 2:11 PM, Andrew Gallatin <gallatin@cs.duke.edu> wrote: > Kip Macy wrote: > >> Log: >> add inline helper functions for buf_ring enqueue and free > >> + uint64_t br_drops; > > Do you have any way to make these drops visible in netstat -id, > or should driver writers regularly mirror br_drops > into their ifp->if_snd.ifq_drops? There will be an arbitrary number of buf_rings, so I think that br_drops should be mirrored in to ifq_drops in the drivers callout. Incrementing a shared counter on multiple cpus works but is not best practice from a cache standpoint. Thanks, Kip
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3c1674c90812051418q1a81ca0an970d410dd034447c>