* maximilian attems <m...@stro.at> [20090625 20:28]: > On Tue, 23 Jun 2009, maximilian attems wrote: > > > * maximilian attems <m...@stro.at> [20090622 19:11]: > > > > On Sun, 21 Jun 2009, Michael Prokop wrote:
> > > > > The virtio_pci module is in the 'base' class already (which is
> > > > > relevant for MODULES=netboot together with class 'net').
> > > > > Though not all available and necessary virtio drivers are used -
> > > > > therefore it's not possible to for example boot via PXE in a KVM
> > > > > virtualiziation environment.
> > > > > It would be nice to have the full list of virtio drivers
> > > > > (virtio_blk, virtio_net, virtio_pci and virtio_balloon) included
> > > > > in MODULES=most and MODULES=netboot.
> > > > ok for virtio_blk and virtio_net.
> from a verbose run:
> Adding module
> /lib/modules/2.6.30-trunk-amd64/kernel/drivers/block/virtio_blk.ko
> thought so due to
> copy_modules_dir kernel/drivers/block
> which puts mouse and cats on initramfs, grr for the drivers/block/paride/
> directory.
> so only virtio_net missing.
NACK:
,---- [ mkinitramfs ]
| case "${MODULES}" in
| dep)
| dep_add_modules
| ;;
| most)
| auto_add_modules
| ;;
| netboot)
| auto_add_modules base
| auto_add_modules net
| ;;
| list)
| # nothing to add
| ;;
`----
,---- [ hook-functions ]
| auto_add_modules()
| {
| case "$1" in
| [...]
| block)
| copy_modules_dir kernel/drivers/block
| ;;
`----
So kernel/drivers/block is considered only if using MODULES=most but
not when using MODULES=netboot. The only virtio file that is copied
at all is virtio_pci from the base class. So both virtio_net *and*
virtio_blk are missing.
regards,
-mika-
signature.asc
Description: Digital signature
