Re: Kernel 2.4.x woes
Am 08. Sep, 2001 schwäzte dman so:
> If you compile your kernel without initrd support then you don't need
> an initrd. The stock kernels need the initrd AFAIK. (Well, I didn't
> get it to work without it)
We keep saying stock kernels. It should be pointed out that we mean the
kernel-image* packages, not necessarily the standard setup when grabbing the
kernel source.
> It is a file that has a cramfs filesystem it. This filesystem is used
> as the root filesystem during startup until the real root filesystem
> is mounted. It allows for loading modules and stuff, if it is setup
> right. I have no idea what benefit I get from it, but there it is
> anyways :-).
It keeps the initial kernel size down. That makes a difference in x86 land,
especially with older boxen. It's a way of lying and convincing the firmware
that you're gonna load something small.
An initrd can be even more important when running the same image on multiple
boxen that all have different boot configs, e.g. scsi vs. ide root devices,
different filesystems for root, etc.
BTW, it's been working flawlessly with the kernel-image packages for me
except for the not updating the boot loader config problem. I'm using 2.4.x
kernel-image packages on most of my workstation boxen as well as servers at
home. We'll be moving to the packages for most of the servers at work as
well.
ciao,
der.hans
--
# der.hans@LuftHans.com home.pages.de/~lufthans/ www.DevelopOnline.com
# Keine Ahnung, was ich dir sagen soll,
# keine Ahnung und keinen (.)plan. -- die Toten Hosen
Reply to: