[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] [PATCH (V9) 0/2] Add V4V to Xen

On Fri, 2013-05-31 at 08:36 +0100, Vincent Hanquez wrote:
> On 05/30/2013 12:57 PM, Ian Campbell wrote:
> > On Tue, 2013-05-28 at 15:43 -0400, Ross Philipson wrote:
> >> [PATCH (V9) 1/2] xen: events, exposes evtchn_alloc_unbound_domain
> >> [PATCH (V9) 2/2] xen: Add V4V implementation
> > What is happening on the tools/client side?
> I'm not sure what kind of tool you want to see here; the only thing i 
> can think of is a way to manipulate the v4v firewall in dom0.
>  From a client side PoV, this is up to the guest kernel to expose the 
> capability.

Or a userspace library, as I clearly said in my original mail. That
means libvchan.

libvchan is the current interface for guest to guest communication in
upstream Xen and I am not in favour of just throwing it away because
something new has come along. IOW I think v4v should be exposed as an
incremental improvement to the existing interfaces. (Whether it is also
exposed via other means such as kernel plugins is a different question).

> For example on linux, the idea is to exposes it is using a new socket 
> family, so that
> everything is handled by usual syscalls (socket(2), recvmsg(3), 
> sendmsg(3), connect(2), listen(2), ioctl(2), ...), and the userspace 
> doens't need any library,
> and specially just a trivial change to already written code that use ip 
> socket without changing any semantics.

So where is this code? Has it been posted upstream? Is anyone working on
getting it in a state such that it acceptable upstream? Would it even be
acceptable to upstream?

So far I haven't seen any realistic attempt at actually making this v4v
hypervisor patch useful to general members of the Xen community.


Xen-devel mailing list



Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.