On 2/12/06, H. Peter Anvin <hpa@zytor.com> wrote:> I have just pushed out a git tree which actually has the root-mounting
> code removed from the kernel. Since there are any number of strange
> boot configurations out there, I would really appreciate any help in
> testing this stuff out.
>
> I have *NOT* implemented support for the following, which I'm hoping
has
> fallen out of use by now:
>
> -> Setting boot flags via rdev, as opposed to on the command
> line
> -> When loading the ramdisk from a block device, stop and
> ask the user for a second disk. This one could probably
> be implemented without too much trouble if people actually
> care.
>
> git://git.kernel.org/pub/scm/linux/kernel/git/hpa/linux-2.6-klibc.git
>
> -hpa
In a matter of days I will try this, and push an experimental kernel
for Archlinux, allowing users with obscure configurations to test
this. This way, I'd hope to get a decent chunk of testing in there
for you (no, I won't forward you errors on my part, heh).
Just a note, this image will contain modules needed to access the root
device, which will be loaded via sysfs cruft before calling kinit. I
have been planning on replacing the default Archlinux kernel with a
klibc enabled version for a bit, but haven't had the time to
completely script everything. So, with the advent of a fully
functional kinit, my job becomes trivial.
Thanks alot for all your work, and I'll try to get you as much testing
data as I can.
Aaron Griffin