Re: [PATCH] Add dm-userspace to the Xen kernel

[Anthony Liguori <aliguori@us.ibm.com>]

Nivedita Singhvi wrote:
> Anthony Liguori wrote:
>> I think dm-userspace is great, but I don't think it belongs in 
>> patches/.  To quote Chris W.:
>>
>> "Good things there are bits that are ready to go upstream or fixes to 
>> base Linux that are transient by version (backport of a fix that's in 
>> the -rc for the next kernel or somesuch)."
>>
>> Until this ends up in -mm, I really don't think we should add this to 
>> the -xen tree as it has no dependence on Xen.  We should be good 
>> members of the kernel community here and go through the normal 
>> channels.  At any rate, I would think that you would want it to be 
>> useful first before adding it which means bringing in all of the 
>> tools into the tree.
>
> It won't be useful in the kernel unless there are actual
> consumers, of which Xen could/might be one. Having a use
> case flesh out problems and get the performance and design
> architecture examined in detail will be useful to the Linux
> maintainers - they get something more robust and better
> designed, although it's certainly not Xen's obligation to
> provide that, per se. I'm just seeing the possibility of
> mutual benefit, here.
>
> That said, I believe since the patch is in the device-mapper
> tree for review, there is a good chance it will go into
> kernel in any case.

Agreed.  I think it's a good idea to let it go through that review 
process before going into the Xen tree.  As I said, once it's in -mm, I 
would imagine that it the interfaces would be stable enough that there's 
no harm putting it into the Xen tree.

Especially since libdmu is being considered for inclusion in the 
device-mapper userspace.  Things would get pretty ugly if the 
dm-userspace ABI changes between now and when it shows up in -mm as 
there'd be conflicting userspace packages.

Regards,

Anthony Liguori

> thanks,
> Nivedita
>