Bug#616326: oldsys-preseed support for LS-WSGL
X-Debbugs-CC: Martin Michlmayr <email@example.com>
Here is a first try at supporting the Buffalo Linkstation Mini (aka
LS-WSGL) for d-i, in response to advices from Martin Michlmayr.
Le mardi 01 mars 2011 à 18:33 +0000, Martin Michlmayr a écrit :
> Can you please file a wishlist bug on oldsys-preseed to support this
> It would also be great if you could check out oldsys-preseed from git:
> and send a patch.
Here is a first try; I don't know what other file to modify. Is there
something to do in tests/arm/?
diff --git a/oldsys-preseed b/oldsys-preseed
index ac97cd3..9a64f50 100755
@@ -153,14 +153,15 @@ case "`archdetect`" in
umount $path/rootfs || true
rmdir $path/rootfs $path || true
- elif echo "$machine" | grep -q "^Buffalo Linkstation Pro/Live"; then
+ elif echo "$machine" | grep -q "^Buffalo Linkstation Pro/Live" ||
+ echo "$machine" | grep -q "^Buffalo Linkstation Mini"; then
# the default filesystem for the system partition is XFS, which isn't included
# in our startup environment. However, customized boxes might have ext3
# instead, so try to mount anyway.
mkdir -p $path/rootfs
- mount -o ro $rootdev $path/rootfs || true
+ mount -o ro /dev/sda2 $path/rootfs ||
+ mount -o ro /dev/md1 $path/rootfs || true
Furthermore, I don't really see the point with what "unset_matching_var"
does, but the naming scheme for the Mini is indeed the same (i.e. MAC
appended to the hostname).
Please note that I didn't actually tested this code on the device; I
don't know how to test all this d-i stuff.
> Looks at the code that is there for the LS Pro already. Can this be
> adapted for the Mini? Does the Mini use ext2/3 of XFS by default?
> (The LS Pro uses XFS, which we cannot read in Debian.)
The Mini use XFS as rootfs too, but /boot is ext3 (which is not of
interest here for oldsys-preseed, but may help for further support of
the Mini). So, I don't really see a point in preseeding it too, as
people having enough skill customizing their box would rather use some
other parameters, IMHO (I didn't even use the original system on this
device, I directly installed Debian; which means that I may not be able
to test all that "backward checking" stuff)
BTW, the uboot command-line tell the kernel the root is on /dev/sda2,
but I'm quite sure the real root is /dev/md1 which is the software RAID1
device formed by /dev/sda2 and /dev/sdb2. All the informations found on
made me realize that the root fs is /dev/root on the original firmware,
which doesn't help here, but I think the initrd may indeed use /dev/sda2
as a "single" FS (is it even possible if the device is used as a part of
a md array?) before pivot/switching root to the md1 array.
Hope this helps.