Re: initrd/cramfs/etc (was Re: srm nightmare, milo & kernel future questions)
- To: Nick Ntarmos <email@example.com>, firstname.lastname@example.org
- Subject: Re: initrd/cramfs/etc (was Re: srm nightmare, milo & kernel future questions)
- From: Chris Hecker <email@example.com>
- Date: Fri, 20 Jun 2003 01:44:04 -0700
- Message-id: <188.8.131.52.2.20030620010650.043e1ed8@localhost>
- In-reply-to: <20030619014800.GA13844@Noth.ceid.upatras.gr>
- References: <184.108.40.206.2.20030618174817.03c0d558@localhost> <220.127.116.11.2.20030615004500.032abf98@localhost> <18.104.22.168.2.20030614150002.02e1adf8@localhost> <22.214.171.124.2.20030615004500.032abf98@localhost> <126.96.36.199.2.20030618174817.03c0d558@localhost>
That's not true. Any decent linux kernel can support an initrd.
But not a cramfs initrd, which is what mkinitrd builds by default. Also, I
had all of the CONFIG_ variables you mentioned set already. However, your
pointer to the Documentation/initrd.txt file was the key...it walked me
through making an ext2 initrd, and that finally worked. I didn't
understand that mkinitrd was a Debian-only tool, and that's why it wasn't
documented very well. I still have no idea why the "cramfs: wrong magic"
error was happening on both romfs and cramfs (cramfs was compiled in, the
images mounted fine, etc.), but it'd be nice to see the Debian cramfs
initrd patch available somewhere.
So, I booted 2.4.21 built specifically for pc164 with SRM by gcc-3.2 and
everything, and of course it still doesn't save the SRM variables
correctly, or reboot/halt consistently. So I lose.
I think I'm going to go back to MILO unless somebody can come up with
another idea here for making SRM act right on reboot/halt.