Re: [TEST/REVIEW]: rootfs mounting code rework

Poul-Henning Kamp wrote:
> In message <p0620073ebdd844240e34_at_[128.113.24.47]>, Garance A Drosihn writes:
> 
>>At 11:08 AM +0100 12/4/04, Poul-Henning Kamp wrote:
>>
>>>http://phk.freebsd.dk/patch/rootfs.patch
>>>
>>>Please test and review.
>>
>>Hmm.  Does this mean that we'll always get the correct partition
>>for '/', even if /etc/fstab is wrong for '/'?  By "correct", I
>>mean "the partition that the kernel was loaded from".
> 
> The loader reads /etc/fstab and picks the entry for '/' out of it
> so I can only pressume that whoever did that disagree with your
> notion of "correct".

Here's a situation that may be worth thinking
about:

   * Working FreeBSD system that uses
     a separate controller.  / is /dev/da4.

   * Controller blows up.

   * Connect hard disk to the on-board controller.

Note: kernel is now being loaded from /dev/da0,
but /etc/fstab still says / is /dev/da4.
With the old rootfs code, this system is
unbootable.

I hope you've considered situations like this.

Tim Kientzle