On Fri, Feb 18, 2011 at 1:36 PM, Pierre Lorenzon <[email protected]
> wrote:

>
> Hi Nathan,
>
> From: Nathan Coulson <[email protected]>
> Subject: Re: kernel options
> Date: Thu, 17 Feb 2011 16:32:41 -0800
>
> > On Wed, Feb 16, 2011 at 12:11 PM, Pierre Lorenzon <
> > [email protected]> wrote:
>
>   Thanks for your answer. But I am not sure to completely
>  understand the situation : Is the root= option on the line
>  kernel ABSOLUTELY MANDATORY if there's no initrd or initramfs
>  ?
>

the default it chooses, is whatever was mounted as / when the kernel was
compiled.

ex:/ if /dev/sda2 was mounted as /, that's what it uses.  [I think this
behavior may have been taken out somewhere in the last 10 kernels, as they
considered this behavior obsolete.  but I can confirm it does not
autodetect]

you can change this with rlimit,  (if that's still around) to another
partition (but it's still sd##, or hd##)


> >
> >
> > you may have to resort to building a initrd or a initramfs.  Without one,
> > you need to specify a partition.
> >
> > just consider it like a "mini" linux system that can run a small bash
> > script, which could find your usb stick, mount it, and continue.  there
> is a
> > command called switch_root that is very useful for this.  At that stage
> > too,  you can use UUID's if you have a linux environment to work with.
> >
> > Documentation/kernel-parameters.txt in the linux kernel may have other
>
>   I find this documentation very minimal. Indeed I didn't see
>  any answer to my question above about the fact that an option
>  SHOULD have a value or might be leaved empty and in which
>  circumstances. Anyway if you have a good suggestion for a
>  documentation to learn these things I'd greatly appreciate !
>
>
> > options that I am unaware of, but afaik, root= can only be given devices.
> >
> > there was a way to determine what your boot device was in /sys, but I
> > believe you still need a initrd/initramfs to take advantage of this.
>
>   This is a paradox indeed ! Information is available in the
>  kernel since it will be exported to the sys by the kernel
>  itself (I suspect at leat !) but kernel cannot export this
>  information to itself ! I know that things might be more
>  complicated than I simply think, but anyway ...
>
>  Regards
>
>  Pierre
>

In most cases, they have always preferred to have the logic in a
initrd/initramfs, instead of bloating their kernel code.  I cannot speak for
them though.

If the kernel could use the UUID to mount it, that would be the easiest
solution (again, another solution you'd have to make a initrd/initramfs for
though)


>
> >
> > --
> > Nathan Coulson (conathan)
> > ------
> > Location: British Columbia, Canada
> > Timezone: PST (-8)
> > Webpage: http://www.nathancoulson.com
> --
> http://linuxfromscratch.org/mailman/listinfo/blfs-support
> FAQ: http://www.linuxfromscratch.org/blfs/faq.html
> Unsubscribe: See the above information page
>



-- 
Nathan Coulson (conathan)
------
Location: British Columbia, Canada
Timezone: PST (-8)
Webpage: http://www.nathancoulson.com
-- 
http://linuxfromscratch.org/mailman/listinfo/blfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page

Reply via email to