On 12/20/2011 04:08 AM, Avi Kivity wrote:
On 12/20/2011 12:03 PM, Avi Kivity wrote:
On 12/20/2011 04:46 AM, Anthony Liguori wrote:
I would hope that you would agree that when designing the device
model, we should aim to do what makes sense independent of migration.
If we cannot achieve a certain feature with migration given the
logical modeling of devices, it probably suggests that we need to
improve our migration infrastructure.
I assume that given the above, we all agree that separate devices is
what makes the most sense ignoring migration.
I don't agree with this.
The problem with having two devices, is that now you have to identify
the common code, put them somewhere, and use them as necessary.
"apic" and "kvm-apic" both is-a (are-a?) "apic". This suggests either a
base class (containing the common code) and derived classes, or (like
Jan's implementation), just one class, that defers part of the
implementation to an interface implemented by two other classes.
Yes, a base-class is what I'm suggesting since this is what qdev is capable of
today.
The other approach to this is to have an APICFrontend has-a APICBackend and then
UserspaceAPIC is-a APICBackend and KernelAPIC is-a APICBackend.
You still now have three visible devices in the device model. This is
essentially what Jan's patches do today.
I think a simple base-class + subclass inheritance scheme makes the most sense here.
Regards,
Anthony Liguori
Two unrelated classes which happen to implement exactly the same
interface (vmstate fields) except one (visible name) and share some code
are a strange solution to this problem.
--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
