Date: Fri, 9 Apr 1999 10:57:32 -0500 From: "Richard Seaman, Jr." <dick@tar.com> To: John Birrell <jb@cimlogic.com.au> Cc: Roger Hardiman <roger@cs.strath.ac.uk>, Julian Elischer <julian@whistle.com>, hackers@FreeBSD.ORG Subject: Re: cc -pthread and -kthread switches Message-ID: <19990409105732.Q440@tar.com> In-Reply-To: <199904090919.TAA29901@cimlogic.com.au>; from John Birrell on Fri, Apr 09, 1999 at 07:19:30PM %2B1000 References: <370DC114.E0A81FDE@cs.strath.ac.uk> <199904090919.TAA29901@cimlogic.com.au>
next in thread | previous in thread | raw e-mail | index | archive | help
On Fri, Apr 09, 1999 at 07:19:30PM +1000, John Birrell wrote: > Yuk. Please don't corrupt FreeBSD with Linux. By all means help to make > FreeBSD provide a compatibility mode to run Linux executables, but leave > FreeBSD as FreeBSD. If people want to _develop_ Linux applications, let > the run _Linux_. And for those of us who choose to use _FreeBSD_, > allow us to live in a Linux and Microsoft free zone. The linuxthreads "port" is simply intended to provide native FreeBSD one-to-one kernel threads. It isn't Linux. Its just a port of some Linux code. My position is, and has always been, that the existence of the port is intended as a stop-gap measure until FreeBSD has its own kernel threads code. When that happens, I would asssume the need for the port would evaporate, and certainly my support of it would evaporate. You and I have corresponded privately, and I have offered my assistance in creating a FreeBSD kernel threads implementation. As I have indicated to you, as well as to Julian, I don't think I'm the right person to lead this effort, since there are technical issues involved with "many-to-many" kernel threads that I don't fully understand how to resolve, yet. But, I'd certainly be willing to be an active contributor to this project. On another note, the recent changes to the FreeBSD user pthreads code, which you committed on about March 22, appears to be a major improvement in this code. On my very crude benchmarks, the linuxthreads port used to beat the heck out of the uthread code. This is no longer true. Most tests show them about equal, except that uthread context switches are now much faster than the linuxthreads code, as they should be (before March 22 uthread was slower). It strikes me that the uthread code seems to be in pretty good shape now, and would be a nice foundation for a FreeBSD kernel threads implementation. -- Richard Seaman, Jr. email: dick@tar.com 5182 N. Maple Lane phone: 414-367-5450 Chenequa WI 53058 fax: 414-367-5852 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?19990409105732.Q440>