Date: Sun, 14 Jan 2001 03:15:11 -0800 From: Peter Wemm <peter@netplex.com.au> To: Poul-Henning Kamp <phk@critter.freebsd.dk> Cc: Wilko Bulte <wkb@freebie.demon.nl>, John Baldwin <jhb@FreeBSD.org>, cvs-committers@FreeBSD.org, cvs-all@FreeBSD.org Subject: Re: cvs commit: src/sys/i386/conf GENERIC Message-ID: <200101141115.f0EBFBQ89810@mobile.wemm.org> In-Reply-To: <28586.979469127@critter>
next in thread | previous in thread | raw e-mail | index | archive | help
Poul-Henning Kamp wrote: > In message <20010114114418.A46703@freebie.demon.nl>, Wilko Bulte writes: > >On Sun, Jan 14, 2001 at 02:11:11AM -0800, John Baldwin wrote: > >> jhb 2001/01/14 02:11:10 PST > >> > >> Modified files: > >> sys/i386/conf GENERIC > >> Log: > >> Remove I386_CPU from GENERIC. Support for the 386 seriously pessimizes > >> performance on other x86 processors. Custom kernels can still be built > >> that will run on the 386. > > > >Does this mean installation won't run on 386 anymore? > > It would be trivial to add i386 to the install kernel, and > probably worthwhile. No, because simply doing that leaves you with an unbootable machine. Anyway, I defy anybody to do a standard CDROM or boot floppy install on a 386 and stay sane. Everybody that I know of that does this sort of thing does one of the following type things: 1: cross builds from a fast machine to a small image and dd's it to disks. Remember, 99% of 386's cannot handle more than 528MB IDE disks. 2: install on a fast machine using sysinstall, and strip the hell out of the kernel, world etc. Then transport the disk to a 386. Building a 5.0 kernel on a *486* takes forever these days, let alone a 386. Remember, 386's were essentially ISA-only. I think we should send people to 2.2.x if they want to run on an ISA-only i386. 5.0 will *seriously suck* on typical 386 hardware. My personal experience is that it *seriously sucks* on a 486 right now (yes, I have one running right now, and a 486DX33 w/ 64M of ram is **painful**). In fact, it was near impossible to run on a 486-33 w/ 12MB ram when I tried it about 12 months ago on what was then -current. I was eventually able to tune things down enough (maxusers 5, 2 gettys, etc) to get it to the point that it didn't lock up the VM system during a compile. i386 *pc* hardware that supports more than 16MB was pretty rare if I recall. Embedded systems are a different issue, but I doubt many people use the FreeBSD cdrom and sysinstall to set up an embedded micro-OS install... Heck, even picobsd uses its own kernel configs. Yes, we could make an alternate 386 kernel, and a 386 boot disk etc. But I really dont think it is worth while. The user experience would be rather uninspiring - I think we'd be far better pointing them to 2.2.x. In fact, another net-only point release of 2.2.x to fix the known security holes would probably be less cumulative effort than it would take to keep i386 a viable 'GENERIC' option for the SMPng kernel over the next 6-12 months. The bottom line is that I feel the time is just about right to yank i386 entirely, not just taking it out of GENERIC. But I wont push for that (yet :-). But ending the expensive runtime cost of i386 support in GENERIC is well overdue I feel. The cost of slowing down copyin()/copyout() etc is just not worth it. Cheers, -Peter -- Peter Wemm - peter@FreeBSD.org; peter@yahoo-inc.com; peter@netplex.com.au "All of this is for nothing if we don't go to the stars" - JMS/B5 To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe cvs-all" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200101141115.f0EBFBQ89810>