Date: Thu, 19 Apr 2001 18:30:20 -0700 (PDT) From: John Baldwin <jhb@FreeBSD.org> To: Greg Lehey <grog@lemis.com> Cc: Rik van Riel <riel@conectiva.com.br>, Alfred Perlstein <bright@wintelcom.net>, Kris Kennaway <kris@obsecurity.org>, freebsd-hackers@FreeBSD.org, Dennis <dennis@etinc.com> Subject: Re: SMP in 2.4 (fwd) Message-ID: <XFMail.010419183020.jhb@FreeBSD.org> In-Reply-To: <20010420104614.C72002@wantadilla.lemis.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On 20-Apr-01 Greg Lehey wrote: > On Thursday, 19 April 2001 at 10:10:51 -0700, John Baldwin wrote: >> >> On 19-Apr-01 Dennis wrote: >>> I understand there is a language thing, but I went out of my way to say >>> that i wasnt saying that SMP shouldnt be supported. It already is, and its >>> been done very cleanly in a way that doesnt compromise the integrity of the >>> OS internals. >> >> Actually, it's done in about the most inefficient manner possible, to be >> brutally honest. The first stage of the SMP support focused more on getting >> the machine to run than on getting it to perform well. You really should go >> do >> some actual research on SMP before spouting off. I highly recommend Curt >> Schimmel's _Unix Systems for Modern Architectures_: Caching and SMP for >> Kernel >> Programmers. If you read it, you will find that our current implementation >> is >> actually worse than a master/slave kernel setup, which is the slowest one >> mentioned in the book. :( > > Well, no, it does mention our approach as being the slowest, even > slower than master-slave :-) Ah, so it does. :) In fact, the last few lines read: "This situation is made worse by the fact that the processor already holding the lock continues to hold it if it context-switches to another kernel-mode process, making it possible to lock other processors out of the kernel for indefinite periods of time. Situations like these must be avoided in any practical MP kernel implementation." Oops. :) Just to clarify, however, I'm not trying to say that the initial SMP effort was a bad thing. We didn't run at all on SMP hardware before the Giant lock. The implementation can stand some improvement is all. I doubt that SMPng will be perfect as well. -- John Baldwin <jhb@FreeBSD.org> -- http://www.FreeBSD.org/~jhb/ PGP Key: http://www.Baldwin.cx/~john/pgpkey.asc "Power Users Use the Power to Serve!" - http://www.FreeBSD.org/ To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?XFMail.010419183020.jhb>