On Tue, Mar 07, 2000 at 07:14:42PM +0000, Nik Clayton wrote:
> On Tue, Mar 07, 2000 at 10:19:57AM -0800, Brooks Davis wrote:
> > On Tue, Mar 07, 2000 at 11:01:09AM +0000, Nik Clayton wrote:
> > > To which the response has been nil.  At this point, you're either all
> > > struck dumb by the staggering simplicity and elegance of this approach,
> > > or you're sat there slowly shaking your head, wondering how I could be
> > > quite so stupid :-)
> > > 
> > > Come on then, which is it?
> > 
> > To be honest it seems like a complete waste of time and a hugh pain in
> > the rear.  All you need to do is build a 4.0 kernel, install it under
> > some random name (say /kernel.current) boot with it to insure that it
> > works.  
> 
> That, at least, was not the case with an upgrade I attempted a few days 
> ago.  On booting with kernel.GENERIC (from -current) it hung mounting the
> disks.  Trying to go back to kernel.stable didn't work, because I'd had 
> to update the /dev entries for -current, and they wouldn't work with
> -stable.  I had to dig out fixit floppies and restore from a backup.
> 
> Ordinarily, you'd make sure that userland, /dev, and the kernel are all
> in sync before you reboot.  However, in this case (and as advised by
> src/UPDATING) you have to reboot with a new kernel after updating /dev,
> but before you update the userland.

You don't need to update /dev though.  As long as you don't change
anything else, a 4.0 kernel will work just fine with a 3.x /dev and
userland (other then top and friends).  You can delay updating /dev
until later as the ata code will deal with wd devices as well as ad
devices.  I built my current laptop configuration by installing 3.3 from
CD, cvsuping to 4.0, installing a new kernel, and then building and
installing world.  I looked at UPDATING.  I think it's wrong.  It should
have you test a new kernel before rebuilding /dev.  You should do it,
but like the scsi disk renaming it isn't mandatory, at least for now.
I think the problem is that UPDATING is a bit to optomistic about
things.  If you have a system that works with the new ata driver it's
correct, but otherwise it's a good way to waste your system. :-(

-- Brooks

-- 
Any statement of the form "X is the one, true Y" is FALSE.


To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message

Reply via email to