Date: Fri, 2 Jul 2004 15:21:06 -0400 From: John Baldwin <jhb@FreeBSD.org> To: "Kevin A. Pieckiel" <pieckiel+freebsd-questions@sdf.lonestar.org> Cc: freebsd-smp@FreeBSD.org Subject: Re: FreeBSD support in SMP platform Message-ID: <200407021521.06272.jhb@FreeBSD.org> In-Reply-To: <20040702173325.GA602@SDF.LONESTAR.ORG> References: <20040604075905.3422.qmail@web16905.mail.tpe.yahoo.com> <200406041450.24062.jhb@FreeBSD.org> <20040702173325.GA602@SDF.LONESTAR.ORG>
next in thread | previous in thread | raw e-mail | index | archive | help
On Friday 02 July 2004 01:33 pm, Kevin A. Pieckiel wrote: > On Fri, Jun 04, 2004 at 02:50:24PM -0400, John Baldwin wrote: > > When a thread is made runnable the idle CPUs contest on sched_lock so > > much that invariable one CPU ends up timing out on sched_lock and > > panic'ing. This will be fixed eventually but not in the near future. > > Obviously, this isn't considered a show stopper bug by the developers, but > it does seem to be quite an egregious error IMO. Afterall, multi-processor > systems seem to be getting more commonplace, and especially with HTT > getting its share of the market, I would think that in not too many years a > system showing eight or more processors--virtual or otherwise--would be not > unheard of. > > Why is fixing this such a low priority? Is it a complicated fix, or simply > a lack of hardware for testing? The fix is not really complicated so much as largely mechanical and time consuming. It also doesn't seem to kick in until at least 8 logical processors or so. However, I don't think the fix will be very long in coming anyway. -- John Baldwin <jhb@FreeBSD.org> <>< http://www.FreeBSD.org/~jhb/ "Power Users Use the Power to Serve" = http://www.FreeBSD.org
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200407021521.06272.jhb>