F3 initrd strangeness

[T. Horsnell (tsh)]

>T. Horsnell (tsh) wrote:
>> I'm having problems building a diskless net-bootable FC3 kernel,
>> and decided to take a look inside the vanilla initrd file
>> initrd-2.6.9-1.667.img. However, if I gunzip this and try
>> to mount it, the mount fails:
>> 
>> [root at fw1 ~]$ gunzip -c initrd-2.6.9-1.667.img > initrd-2.6.9-1.667
>> [root at fw1 ~]$ ls -l initrd-2.6.9-1.667*
>> -rw-r--r--  1 root root 1215488 Dec  6 15:29 initrd-2.6.9-1.667
>> -rw-r--r--  1 root root  530308 Dec  6 15:29 initrd-2.6.9-1.667.img
>> [root at fw1 ~]$ mount -o loop -t ext2 /root/initrd-2.6.9-1.667 /mnt
>> mount: wrong fs type, bad option, bad superblock on /dev/loop0,
>>        or too many mounted file systems
>>        (could this be the IDE device where you in fact use
>>        ide-scsi so that sr0 or sda or so is needed?)
>
>$ file /root/initrd-2.6.9-1.667
>/root/initrd-2.6.9-1.667: ASCII cpio archive (SVR4 with no CRC)
>
>It's not a filesystem any more, it's just a cpioball.
>

Ah - so it is. Thank you. But how does it then get used as a
boot-time initrd? To see its contents I've had to:

[root at fw1 ~]$ dd if=/dev/zero of=initrd bs=1300000 count=1
[root at fw1 ~]$ mke2fs -F -m0 initrd
[root at fw1 ~]$ mount -t ext2 -o loop initrd /mnt
[root at fw1 ~]$ cd /mnt
[root at fw1 ~]$ cpio -i < /root/initrd-2.6.9-1.667

How does a just-booted kernel do it on FC3?
Also, there's no linuxrc on it.
Just when I think I'm beginning to understand things, they change...

Terry.