Re: [PATCH -v2] Add MCE support to KVM

[Avi Kivity <avi@redhat.com>]

Anthony Liguori wrote:
> Avi Kivity wrote:
>> No argument.  But some kernel features will require major rework in 
>> qemu before we can support them properly.  We'll still want to bring 
>> them to users quickly so users can enjoy them (and we can enjoy users 
>> testing them).  Waiting for a qemu rework may take a while, and we'll 
>> lose valuable feedback.
>
> Then we're failing.  If a particular implementation of a feature is 
> acceptable for kvm-userspace users, and we don't take it in QEMU 
> without requiring a huge amount of different work, then it suggests 
> something is broken about the QEMU process.
>

Example: SMP.
Example: vlan API.

>> For this kind of work, we can provide kvm-userspace.git as a kind of 
>> playground where these experiments may be made.  kvm-userspace.git 
>> exists to support kvm.git; while qemu.git has a life of its own and 
>> more stringent barriers to acceptance.
>
> As long as people are using kvm-userspace.git instead of qemu.git, 
> we're failing IMHO.  If kvm-userspace.git is basically the equivalent 
> of the x86 git kernel tree, linux-next, or something like that, then 
> that's a good thing.

That's definitely a long term goal, but qemu is not yet at a point where 
it is easy to implement new features efficiently.  Once it reaches that 
state, kvm-userspace will become a simple staging ground (or even 
disappear entirely).

-- 
Do not meddle in the internals of kernels, for they are subtle and quick to panic.