Il giorno Tue, 12 Dec 2006 17:35:30 +0100 David Härdeman <[EMAIL PROTECTED]> ha scritto:
> Thanks for the patches, I think I prefer the second one...
me too :-)
> one
> question though:
>
> >--- /usr/share/initramfs-tools/scripts/local-top/cryptroot.old
> >2006-12-12 16:09:21.000000000 +0100
> >+++ /usr/share/initramfs-tools/scripts/local-top/cryptroot
> >2006-12-12 16:10:28.000000000 +0100 @@ -114,6 +114,23 @@
> > return $?
> > }
> >
> >+activate_evms()
> >+{
> >+ local module
> >+
> >+ # Sanity checks
> >+ if [ ! -x /sbin/evms_activate ] || [ "${1#/dev/evms/}" =
> >"$1" ]; then
> >+ return 1
> >+ fi
> >+
> >+ # activating evms
> >+ for module in dm-mod linear raid0 raid1 raid10 raid5
> >raid6 ; do
>
> How can you know that these modules are included? What adds them
> to the initramfs image?
The evms hook includes them in initrd. The piece of code is the same
that is executed by the evms script. If they are non present then evms
cannot work. In any case if they are not present modprobe exits without
errors, so this approach is good even if they are statically compiled
into the kernel.
> Also, in the inverse scenario...i.e. if you have evms on top of
> crypto instead of the other way around, would it perhaps be a good
> idea to also run evms_activate after the crypto volume has been setup?
I really don't know, it's a configuration that I have never used. I'm
also not sure if it's possible, AFAIK this configuration requires a
patched version of evms (not the one present in debian), and at the
moment this solution isn't possible without deep changes in debian boot
scripts. But maybe it's better if you contact evms mantainer and ask
his opinion, his opinion is surely better than mine.
signature.asc
Description: PGP signature
