RE: [RFC net-next 1/2] if_link : add support for VF privileges

-----Original Message-----
From: David Miller [mailto:davem@davemloft.net]
Sent: Tuesday, February 21, 2012 4:05 PM
To: Khaparde, Ajit
Cc: bhutchings@solarflare.com; shemminger@linux-foundation.org;
netdev@vger.kernel.org
Subject: Re: [RFC net-next 1/2] if_link : add support for VF privileges

From: <redacted>
Date: Tue, 21 Feb 2012 14:02:27 -0800

quoted

quoted

-----Original Message-----
From: Ben Hutchings [mailto:bhutchings@solarflare.com]
Sent: Tuesday, February 14, 2012 5:25 PM
To: Khaparde, Ajit
Cc: davem@davemloft.net; shemminger@linux-foundation.org;
netdev@vger.kernel.org
Subject: Re: [RFC net-next 1/2] if_link : add support for VF privileges

On Tue, 2012-02-14 at 13:26 -0600, Ajit Khaparde wrote:

quoted

+enum {
+	IFLA_VF_PRIVILEGE_DEFAULT = 1,	/* Default privileges */

What are the default privileges?  Should existing drivers report that
their VFs have this?

Yes. Vendors can decide what privileges they want to grant for VFs by default.

That's terrible and a very bad interface for users.  It means every system
can have different defaults, from which we'll derive zero consistency.

Since different hardware devices can have different capabilities this possibility exists.
Do you suggest indicating the individual capabilities in the bit mask rather than just a single bit?
That way every VF driver will get to indicate its capabilities.