Date: Sun, 12 Dec 1999 23:15:20 -0800 From: Mike Smith <msmith@freebsd.org> To: remy@synx.com Cc: smp@freebsd.org Subject: Re: Fwd: Idle loop in SMP. Message-ID: <199912130715.XAA05688@mass.cdrom.com> In-Reply-To: Your message of "Sun, 12 Dec 1999 15:11:08 %2B0100." <199912121411.PAA39511@gw0.boostworks.com>
next in thread | previous in thread | raw e-mail | index | archive | help
> (Forwarded to -current, due to lack of audience in -smp. Sorry for > bothering you). Wrong course of action. This should have stayed on -smp. > While investigating a temperature problem, I seen that the default_halt > entry called for an idle processor do not really halt the processor. I > found the reason on the CVS logs and it is intended to react to changes > made on the run queue by the other processor. (i386/i386/swtch.s, 1.61). > > Since this is dated Sept 97, can we expect a better solution regarding > the progress made in the SMP area ? Such as what? Any other solution adds overhead (eg. sending an IPI to a halted processor), and the rationale for SMP would seem to be performance, not economy. Naturally, patches that actually halted the idle CPUs and woke them in an efficient fashion would be accepted, but I don't think that anyone actively working on SMP currently considers this a major problem. (Hence the lack of response to your original post.) -- \\ Give a man a fish, and you feed him for a day. \\ Mike Smith \\ Tell him he should learn how to fish himself, \\ msmith@freebsd.org \\ and he'll hate you for a lifetime. \\ msmith@cdrom.com To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-smp" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199912130715.XAA05688>