Date: Mon, 30 Mar 2009 19:20:56 +0000 (UTC) From: Jeff Roberson <jeff@FreeBSD.org> To: src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-stable@freebsd.org, svn-src-stable-7@freebsd.org Subject: svn commit: r190570 - stable/7/sys/kern Message-ID: <200903301920.n2UJKuuB075965@svn.freebsd.org>
next in thread | raw e-mail | index | archive | help
Author: jeff Date: Mon Mar 30 19:20:56 2009 New Revision: 190570 URL: http://svn.freebsd.org/changeset/base/190570 Log: MFC SVN rev 189787. - Fix steal_thresh calculation with odd numbers of cpus and sched_affinity() for threads on runqueues. Approved by: re Modified: stable/7/sys/kern/sched_ule.c Modified: stable/7/sys/kern/sched_ule.c ============================================================================== --- stable/7/sys/kern/sched_ule.c Mon Mar 30 18:47:13 2009 (r190569) +++ stable/7/sys/kern/sched_ule.c Mon Mar 30 19:20:56 2009 (r190570) @@ -1395,11 +1395,11 @@ sched_initticks(void *dummy) */ balance_interval = realstathz; /* - * Set steal thresh to log2(mp_ncpu) but no greater than 4. This - * prevents excess thrashing on large machines and excess idle on - * smaller machines. + * Set steal thresh to roughly log2(mp_ncpu) but no greater than 4. + * This prevents excess thrashing on large machines and excess idle + * on smaller machines. */ - steal_thresh = min(ffs(mp_ncpus) - 1, 4); + steal_thresh = min(fls(mp_ncpus) - 1, 3); affinity = SCHED_AFFINITY_DEFAULT; #endif } @@ -2549,6 +2549,11 @@ sched_affinity(struct thread *td) ts = td->td_sched; if (THREAD_CAN_SCHED(td, ts->ts_cpu)) return; + if (TD_ON_RUNQ(td)) { + sched_rem(td); + sched_add(td, SRQ_BORING); + return; + } if (!TD_IS_RUNNING(td)) return; td->td_flags |= TDF_NEEDRESCHED;
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200903301920.n2UJKuuB075965>