[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: [Fwd: Alpha Debian Netinst]



Steve Langasek <vorlon@netexpress.net> writes:

> On Sun, Nov 30, 2003 at 11:27:46PM +0100, Goswin von Brederlow wrote:
> > Steve Langasek <vorlon@netexpress.net> writes:
> 
> > > On Sun, Nov 30, 2003 at 09:27:06PM +0100, Goswin von Brederlow wrote:
> > > > Steve Langasek <vorlon@netexpress.net> writes:
> 
> > > > > On Mon, Nov 17, 2003 at 10:16:16PM +0100, Goswin von Brederlow wrote:
> > > > > > The correct line would be:
> 
> > > > > > 0:boot/linux root=/dev/ram initrd=boot/root.bin ramdisk_size=16384 devfs=mount,dall
> 
> > > > > One thing that seems to be inconsistent here -- the
> > > > > boot/i386/syslinux.cfg settings use 'root=/dev/rd/0' as opposed to
> > > > > 'root=/dev/ram'.  Is there some reason one should be preferred over the
> > > > > other?  I believe that both should work, but assume that if there's a
> > > > > reason to use one over the other, it would apply equally to all
> > > > > platforms.
> 
> > > > /dev/ram is the non-devfs name. Some bootloaders or the kernel
> > > > translate that into major:minor numbers. The devfs name on the other
> > > > hand won't get translated.
> 
> > > > I think root=/dev/rd/0 on i386 is actually wrong but since an initd is
> > > > detected the "root=" is ignored anyway and the initrd is mounted. Could
> > > > be wrong though. You could test using /dev/rd/1 and see if it still
> > > > works.
> 
> > > That would suggest that the root= option is completely superfluous and
> > > should be omitted on all architectures.  I definitely see correct
> > > behavior when booting with root=/dev/rd/0 on my alpha (though I'm not
> > > currently in a position to test root=/dev/rd/1).

I ment testing that on i386. :)

> > Since we dropped /linuxrc and use /sbin/init now on all archs it
> > probably doesn't matter if /dev/ram or /dev/rd/0 is used as long as
> > devfs is used too.
> 
> I don't see how that's particularly relevant.  In all cases, the root
> filesystem must be resolved and mounted before any init program can be
> run, and this was true even with init=/linuxrc.

The kernel wouldn't start /linuxrc with root=/dev/ram. But now looking
at the code again I see that it now matches the major/minor pair so
neither root= should have worked. ARGS, this linuxrc stuff is an ugly
hack.

> At any rate, if there are no other arguments one way or the other, I
> think consistency is most important here and that alpha should follow
> i386's lead on this point.

powerpc has:

root=/dev/ram

2 against 1.

MfG
        Goswin



Reply to: