RFC: add built-in support for kvm devices to -virtual kernels

[Thierry Carrez]

Scott Moser wrote:
> If we decide that CONFIG_SCSI_SYM53C8XX_2=y is too risky, then our options
> are
> 1.) drop the ramdiskless boot from uec
> 2.) move '-virtual' from being a 'sub-flavour' to a full 'flavour' build.

Could kernel experts precise what type of risk we are talking about here ?

My concern is that this could create incompatibilities with specific
hardware, that would:
- have worked very well for all the past releases
- only be discovered when people start to exercise that specific
hardware after release

If that's the case, it sounds like a bad move for a LTS release... The
net gain of having a ramdiskless UEC image would be clearly outweighed
by the risk that would introduce.

About the "moving from sub-flavour to full flavour" option, how costly
would this be ? That sounds like the best option if the -virtual kernel
is set to diverge more and more from the regular server kernels. As we
try to be host-friendly to an increasing number of cloud environments,
that might well be the case...

-- 
Thierry Carrez
Ubuntu Server team