[libvirt] [PATCH 20/24] hostdev: Save netdev configuration of actual device

[Andrea Bolognani]

On Wed, 2016-03-16 at 15:05 -0400, John Ferlan wrote:
> > > And yes, I'm still curious about
virHostdevPCINodeDeviceDetach and how
> > > that plays into things.
> > > 
> > What exactly is confusing you about that
function?
> 
> It's somewhere that puts devices on the inactiveList via
> virPCIDeviceDetach and my eyes/brain kept telling
me, don't go there,
> you don't want to confuse yourself, stay away, it's a trap.
> 
> I think it was more of how does it
play in the whole scheme of things...
> Then something written in 21 or 22 perhaps made me think that part of
> the
process of "self managing" whether the device is attached to the
> host or not is something telling libvirt via that
function that the
> device was detached.  It doesn't care about the domain, just the device.

Yeah, basically that function is used when the user is willing
to take on part of the responsibility himself. Or needs to.

If you recall the hostdev series I posted before this one, it
was all about making sure that "detach from host" and "reattach
to host" do exactly the same thing regardless of whether
they're performed explicitly by the user or implicitly as part
of working with managed devices. That's the end goal[1].

Cheers.


[1] Well, the end goal is really fixing #1372300, but we can't
    do that in a sane way before we've gotten rid of this code
    duplication :)
-- 
Andrea Bolognani
Software Engineer - Virtualization Team