On Sunday, February 1, 2015, Davide Italiano <dav...@freebsd.org> wrote:
> On Mon, Feb 2, 2015 at 1:17 AM, Steven Hartland <s...@freebsd.org > <javascript:;>> wrote: > > Author: smh > > Date: Mon Feb 2 00:17:36 2015 > > New Revision: 278040 > > URL: https://svnweb.freebsd.org/changeset/base/278040 > > > > Log: > > Prevent inlining txg_quiesce > > > > This allows dtrace to monitor the calls to txg_quiesce which can be > really > > helpful. > > > > Also standardise __noinline order for arc_kmem_reap_now. > > > > Sponsored by: Multiplay > > > > > > I found this while I was at iXsystems, trying to port Richard Elling's > arc statistics utility to FreeBSD. > I do not disagree with the change in theory, but I never committed > this because I was worried of performance penalty. > Last time I checked Solaris/Illumos actually do not inline this > function without any particular directive (I guess this is mostly due > to difference in toolchain, in particular in the compiler), so it > shouldn't be a big deal, but if you have time, do you mind to check > the penalty introduced, if any, is negligible? > It should be negligible. The function is only called once per txg, which generally is only supposed to happen about 1-0.2Hz. --Will. _______________________________________________ svn-src-head@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/svn-src-head To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"