Booting OS on PowerPC

[Milton Miller]

On Dec 12, 2005, at 4:14 AM, Segher Boessenkool wrote:

>> Except I am guessing that our "don't overwrite the initrd with the 
>> device tree" logic might break.  In other words, prom_init.c might 
>> need updating.
>
> Yeah -- there needs to be a transition period where prom_init.c
> accepts both ways, with a warning to update yaboot if needed.
>
> On the other hand, do people find external initrd's useful at
> all still, or is everyone using built-in initramfs?  If so, we
> could just deprecate initrd caompletely.

Perhaps you didn't realize that external initramfs pieces are passed to 
the kernel via the initrd mechanizm?

One of the key features of initramfs is that it can be built from 
smaller pieces.  The kernel puts them all together, then looks at the 
result.

I use this feature daily to piece together a base layout with 
customizations for the use of the machine.  I even include a rc file 
that is specific to a given machine.


Perhaps a better question is; does anyone build a custom initramfs into 
their kernel?  Beyond those replacing the built-in functionality 
(prepare_namespae) with userspace?

milton