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

Re: [libvirt] [PATCH 2/5] Add virNodeDeviceAttach()/ReAttach()/Reset() APIs

On Wed, 2009-02-25 at 19:58 +0000, Daniel P. Berrange wrote:

> > > While I can see a point in providing public APIs to attach/detach
> > > drivers to devices - because we need this for Xen driver PCI 
> > > passthrough, I'm not sure theres a compelling need for exposing
> > > a reset function, because both Xen & your KVM impl are quite
> > > happy doing the resets themselves.
> > 
> > The idea with the reset function is that calling reset is a way for the
> > app to query whether this is an assignable device - e.g. if the user
> > chooses a given NIC to pass through in one of the early screens in
> > virt-manager, we can give a "you can't assign that device" error at that
> > point rather than just having the guest fail to start up much later on.
> I wonder if we should generalize that beyond just host device, to cover 
> VM device hotplug in general, passing in the same XML doc you'd
> use for the subsequent virDomainAttachDevice() call.
>    virDomainCanAttachDevice(virDomainPtr dom, const char *xml);
> Though, obviously that won't help for scenarios before a virDomainPtr
> exists. So might also want a way to ask whether a device is likely to
> be supported before creating a VM
>    virConnectSupportsDevice(virConnectPtr conn, const char *xml)
> But then maybe this is getting into overkill, and just rename the reset
> function to
>    int virNodeDeviceAssignable(virNodeDevicePtr dev);
> so we're not explicitly saying 'reset' is the test we're doing, even though
> that may be the internal impl

I see what you're getting at - we may in future want have an "is
assignable" API that goes beyond just "does reset" work and it would be
a pity to have to add yet another API.

However, "is assignable" in the context of a hotplugging into a running
guest is an operation on a given domain - e.g. if you're hotplugging a
device on the same bus as a device previously assigned to the same
guest, then that's allowed.

"reset" seems to me to be a pretty sane operation to expose - especially
for what we're now calling "non-managed" assigned devices. It doesn't
cover all cases, but it's an step in the right direction.


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