[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#475783: closed by Daniel Baumann <daniel@debian.org> (reply to daniel@debian.org) (Re: generated initrd fails mounting loopback device)



* Daniel Baumann <daniel at debian.org> [20080421 23:52]:
> Michael Prokop wrote:

> > Well, the combination "live-initramfs from sid, [...] busybox, klibc
> > from sid" with a kernel which is known to be working just fine (and
> > worked without any problems regarding this issue until now) is
> > broken now and there seems to be a regression somewhere *you* don't
> > encounter but *me* doing so.

> you are saying that you can reproduce it with the /current/ packages in
> debian (whereas current means the current ones, and live-initramfs 1.130
> with, optionally, your patches you've done to your 1.110 version in grml)?

Yes, when using live-initramfs 1.130.1-2 (without any grml patches)
I notice that the squashfs kernel module is missing in the
initramfs.

The kernel module is present in the chroot:

  /lib/modules/2.6.23-grml/kernel/fs/squashfs/squashfs.ko

but it's not present in the initramfs, even though I see:

# grep -r 'manual_add_modules squashfs' /usr/share/initramfs-tools
/usr/share/initramfs-tools/hooks/live:manual_add_modules squashfs

Running manual_add_modules() of
/usr/share/initramfs-tools/hook-functions in debugging mode (using
'set -x') it does not give me a single line mentioning squashfs.

When running /usr/share/initramfs-tools/hooks/live in debugging mode
as well I see:

  cp: `/lib/libdevmapper.so.1.02.1' and `/tmp/mkinitramfs_f26814/lib/libdevmapper.so.1.02.1' are the same file

The reason why the 'set -e' of the hooks/live-script is being
triggered when executing the code is:

for hidden_klibc_dep_library in libacl libblkid libuuid libdevmapper libattr
do
       cp -a /lib/${hidden_klibc_dep_library}.so.* "${DESTDIR}"/lib
done

Because of the /lib/libdevmapper.so.1.02.1 error the initramfs isn't
"ready to go" for live-setup. When running
/usr/share/initramfs-tools/hooks/live without 'set -e' it works
again. AFAICS that's the trigger of my problem, because booting with
the "workaround" initramfs works again (seeing "mount: you must
specify the filesystem type" during initramfs sequence three times,
not sure whether that's relevant though).

The software packages being involved in the live-initramfs setup:

ii  busybox                 1:1.9.2-3               Tiny utilities for small and embedded systems
ii  cryptsetup              2:1.0.6-1               configures encrypted block devices
ii  dmraid                  1.0.0.rc13-2.1          Device-Mapper Software RAID support tool
ii  initramfs-tools         0.92                    tools for generating an initramfs
ii  klibc-utils             1.5.9-2                 small utilities built with klibc for early boot
ii  libdevmapper1.02.1      2:1.02.24-4             The Linux Kernel Device Mapper userspace library
ii  live-initramfs          1.130.1-2               Debian Live initramfs hook
ii  lvm2                    2.02.33-2               The Linux Logical Volume Manager
ii  mdadm                   2.6.4-1                 tool to administer Linux MD arrays (software RAID)
ii  udev                    0.114-2                 /dev/ and hotplug management daemon

regards,
-mika-
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : http://lists.alioth.debian.org/pipermail/debian-live-devel/attachments/20080422/480e1699/attachment.pgp 


Reply to: