Am 01.07.2010 09:27, schrieb Nathan Coulson:
> On Wed, Jun 30, 2010 at 11:24 PM, splotz90<[email protected]> wrote:
>> Am 01.07.2010 05:17, schrieb Bruce Dubbs:
>>> I'll ask the question again. How is search useful in an LFS
>>> environment where we don't have initrd available?
>> As I already said:
>>
>> The search command is only useful if we're using a seperate boot
>> partition ...
>>
>> With help of the search line, GRUB can find the boot partition, even if
>> the device of the boot partition has changed (for example /dev/sda3 -->
>> /dev/sda4).
>> After that, GRUB will boot the kernel. The kernel can mount the root
>> partition.
>>
>> If the device of the root partition has changed (for example /dev/sda1
>> --> /dev/sda2) the boot process will fail (the kernel won't be able to
>> boot the root partition).
>>
>> But we have to use a LABEL or UUID entry for the boot partition in
>> /etc/fstab (if we're using something like /dev/sda1 if fstab, the mount
>> of the boot partition will fail).
> no you dont. / is already mounted. If you say it's in /dev/sdz65,
> it'll be fine.
>
> After it boots (when it hits the bootscripts),
> /etc/rc.d/init.d/checkfs has a "mount / -o remount,ro'. It does not
> parse /etc/fstab as / is already mounted
>
> Next time, fstab is used is in mountfs. This is where LABEL or UUID
> is useful for all partitions "except" /.
>
> Infact, if you boot with root=/bin/bash, / is still mounted, and
> /etc/fstab will never have been read
>
>> This means: If the LFS-User changes the device for the boot partition,
>> the system will still boot (with help of the search line).
>>
>> So I suggest that we are writing the following:
>>
>>
>> "The search lines are only meaningful for LFS systems if a separate boot
>> partition and a LABEL or UUID entry for this partition in /etc/fstab is
>> used."
> /etc/fstab is only useful "after" / is mounted. so I would remove the
> part that says "separate boot partition and a LABEL or UUID entry for
> this partition in /etc/fstab is used." (the rest of the sentence does
> not stand well on it's own though).
>
> the linux kernel has to mount something as / and all it has to work
> with are (a) root=, or (b) the drive/partition# embedded at
> compiletime.
>
>> I think, I've started a big discussion with just one small ticket ;-)
>
>
I think you've understood me wrong ...
Here is an example:
/dev/sda1 --> LFS-system ( / partition)
/dev/sda2 --> unformatted partition
/dev/sda3 --> boot partition ( /boot)
the fstab:
...
/dev/sda1 / ext3 defaults 1 1
/dev/sda3 /boot ext3 defaults 1 1
...
If the LFS-user decides to delete the unused partition (/dev/sda2) we
will have the following:
/dev/sda1 --> LFS-system ( / partition)
/dev/sda2 --> boot partition ( /boot)
But if we are still using /dev/sda3 for /boot in fstab, the mount of the
boot partition will fail (/dev/sda3 doesn't exist anymore).
This fstab would be better:
...
/dev/sda1
/ ext3 defaults 1 1
# boot partition:
UUID=03b77228-ed4c-4218-910e-11b9f77c4b46 /boot ext3
defaults 1 1
...
I don't want to use a UUID for / in the fstab ...
Sebastian
--
http://linuxfromscratch.org/mailman/listinfo/lfs-dev
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page
