Ollie Lho wrote:
>
> Nikolai Vladychevski wrote:
> >
> > Hi again,
> >
> > now I am trying to burn to my DoC a compressed root filesystem (just
> > after the kernel image) and I want the kernel to boot this filesystem
> > decompressing it to ram. I made my ramdisk and , burned it after the
> > kernel image previously doing "rdev vmlinux /dev/mtdblock0" and "rdev -r
> > vmlinux 960 (my offset)" on it but the kernel seems to ignore it (I get
> > "I have not root and I want to scream"). I also added /dev/mtdblock0
> > major and minor data to the array in init/main.c but it is not a big
> > help. I am now hacking into the kernel and looking the internals and
> > after 2 days a question picked up in my mind.... Has anybody done this
> > before? And if so, how? I feel like I am loosing completely .....
> >
> > Thanks in advance.
> >
> > Nikolai
>
> I don't think you can do that
why not? if the root device is /dev/mtdblock0 and it is recognized by
the kernel before root is mounting (and I achieved that already) it
could be used just like /dev/fd0 to load ramdisk (in theory of course)
>. There was somebody working on using
> ramdisk
I looked at linuxbios source and there is code that handles some stuff
with initrd, but since no docs talks about initrd support I guess it is
not supported yet? I went through the list archives and found somebody
talked about ramdisk, maybe I will contact that person directly to ask
how he did it.
> but I did not hear any good new form him. The usually practice here is
> use /dev/nftla to host a CRAMfs image.
yes, this worked fine for me. It is just I don't want to lose a single
byte of the DoC, it is already too small , if the kernel has already
everything to decompress and load into ram, why not reuse it? Using
/dev/nftla to host compressed ram image would require to store the code
to decompress it and load into ram, and this code has to be stored on
/dev/ntfla too. I am correct?
>
> Ollie
Nikolai
