From owner-freebsd-sparc Tue Jan 18 18:47:14 2000 Delivered-To: freebsd-sparc@freebsd.org Received: from www0d.netaddress.usa.net (www0d.netaddress.usa.net [204.68.24.33]) by hub.freebsd.org (Postfix) with SMTP id B606114C99 for ; Tue, 18 Jan 2000 18:47:07 -0800 (PST) (envelope-from lamester@netscape.net) Received: (qmail 1651 invoked by uid 60001); 19 Jan 2000 02:17:00 -0000 Message-ID: <20000119021700.1650.qmail@www0d.netaddress.usa.net> Received: from 204.68.24.33 by www0d for [130.123.128.4] via web-mailer(M3.3.1.96) on Wed Jan 19 02:17:00 GMT 2000 Date: 19 Jan 00 15:17:00 NZDT From: David Ferry To: Jason Evans , Lyndon Griffin Subject: Re: [Re: hello ello llo lo o] Cc: freebsd-sparc@freebsd.org X-Mailer: USANET web-mailer (M3.3.1.96) Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: quoted-printable Sender: owner-freebsd-sparc@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org To see this read the freebsd-alpha mail archives starting jan 1998 ( I th= ink). These first few months give key info on how they did the alpha port. They put as much of the userspace across running under the netbsd kernel.= ie run netbsd, replace binutils+gcc etc with freebsd ones. Kernel work seemed like it started later. I worked with a friend a year and a half ago porting linux to the sun 3/8= 0. We did it the opposite way around. It may help you to read an alternative. That was we started with the kernel boot code assembly. Got it initializi= ng the mmu right which under linux meant that the mm system was in place, th= en supported the console+ serial ports, ethernet. Then configured the kernel= to nfs mount root on another box to get userspace with binutils etc targetin= g a m68k box. Later someone else did scsi. Then we graduated and lost access to the box= en despite the fact they had dead id proms :( Putting my neck out this made far more sense for us than porting user spa= ce when most of the work is going to be kernel space anyway (especially considering you are doing sparc32's just use a cross compiler and target = sparc 32). =46rom bsd documentation I have read the mmu code is multiplatform as lon= g as the pmap layer is implemented on the target platform. This was true of fr= eebsd as it came from 4.4bsd which had the pmap layer (dunno anymore) -David > If you're looking for direction, you might look at how the initial stag= es > of the Alpha port were done. From what I understand, the porters basic= ally > started out with NetBSD and iteratively replaced portions of the system= > with FreeBSD. > = > Jason > = lamester@netscape.net ____________________________________________________________________ Get your own FREE, personal Netscape WebMail account today at http://webm= ail.netscape.com. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-sparc" in the body of the message