If someone have console=hvc0 kernel command line, it seems that it needs
to be removed or replaced with console=ttyS0 when migrating the VMs to
F20. With console=hvc0 and latest Fedora kernels, the VM refuses to boot.
Without any console= arg, hvc* consoles, defined by:
.../console/target/@type='virtio' in the libvirt xml defs, are remaining
usable *after* the boot, i.e.:
virsh start vm --console
will show the login prompt eventually, just does not dumps the kernel
messages during the boot. If you need the boot messages, might use
console=ttyS0 (+ .../@type='serial'), like before hvc0 feature instead.
This perhaps have something to do with the recent reconfiguration of
virtio_console as loadable module:
http://pkgs.fedoraproject.org/cgit/kernel.git/commit/?id=d58808f
https://bugzilla.redhat.com/show_bug.cgi?id=1019569
Maybe, the virtio_console module could be included again as builtin for
the Fedora Cloud product kernel?
Kind Regards,
Alek
PS: Some happy day, we might have semantic databases integrating
products changes with our local setups and helping with the faster,
reliable resolutions of these incompatibility introductions. Or at
least nobody can stop us to dream about :-).
PPS: My case:
Before .6-302, I was used with: console=hvc0 command line for direct
kernel boot (my VMs are the simplest possible, provisioned on plain
LVMs/ext4 with just yum+xquery and latter updated with in-house
salt+libvirt scripts):
[*] virsh dumpxml vm:
...
<os>
<type arch='x86_64' machine='pc-i440fx-1.6'>hvm</type>
<kernel>/var/lib/libvirt/boot/vmlinuz-3.11.6-301.fc20.x86_64</kernel>
<initrd>/var/lib/libvirt/boot/initramfs-3.11.6-301.fc20.x86_64.img</initrd>
<cmdline>ro root=/dev/vda console=hvc0</cmdline>
</os>
...
<console type='pty'>
<target type='virtio' port='0'/>
</console>
...
With the latest kernels such configuration hangs, because of hvc0.
_______________________________________________
virt mailing list
virt@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/virt
