Received-SPF: pass (google.com: best guess record for domain of linux-wireless-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Message-ID: <cedf5016a59f313c073e1549961317b8fcf0ceab.camel@gmail.com>
Subject: Re: [PATCH 1/2] nl80211: Add LIVE_ADDR_CHANGE feature
From:   James Prestwood <prestwoj@gmail.com>
To:     Johannes Berg <johannes@sipsolutions.net>,
        linux-wireless@vger.kernel.org
Date:   Tue, 08 Oct 2019 15:13:30 -0700
In-Reply-To: <83bde7f521e636903b6b386faaa31b73479add8b.camel@sipsolutions.net>
References: <20190913195908.7871-1-prestwoj@gmail.com>
         (sfid-20190913_220113_419728_36032050) <6c85d9b108825939afb84ebc8a708a211f7e2b98.camel@sipsolutions.net>
         <38c222129dd182de4c55e536a89eada7f58a1bf1.camel@gmail.com>
         (sfid-20190913_225826_111097_E5ED9AB1) <47aaaccf7d8fb58587f0cfbe38acd18e1420dc9a.camel@sipsolutions.net>
         <b43f737c06a392957e6f651d59282a3f36fd05e5.camel@gmail.com>
         (sfid-20190913_231625_014352_843A9715) <83bde7f521e636903b6b386faaa31b73479add8b.camel@sipsolutions.net>
Content-Type: text/plain; charset="UTF-8"
Mime-Version: 1.0
Content-Transfer-Encoding: 7bit
Sender: linux-wireless-owner@vger.kernel.org
Precedence: bulk

Hi Johannes,

On Tue, 2019-10-01 at 11:14 +0200, Johannes Berg wrote:
> Hi,
> 
> > > > Because userspace needs to know if this is supported?
> > > > IFF_LIVE_ADDR_CHANGE is a private flag... AFAIK userspace has
> > > > no
> > > > way of
> > > > obtaining this.
> > > 
> > > Oh, annoying.
> > > 
> > > But that doesn't really mean that nl80211 is an appropriate place
> > > to
> > > advertise it, IMHO?
> > 
> > The intention of the flag was not soley related to
> > CMD_CONNECT/CMD_AUTHENTICATE. Its an indication that the
> > hardware/driver supports a live address change. If you don't want
> > it
> > here could you suggest a better location for it?
> 
> I guess RTNL would be the right place? This can hardly be specific to
> wireless, the flag comes from elsewhere.

I do see where your coming from, and maybe the answer is both RTNL and
NL80211?

In this context a NL80211 flag would mean something different than only
putting a flag into RTNL.

The NL80211 flag means this device supports changing the MAC while
running and not offchannel/scanning. Which is what -EBUSY would mean in
this context. Managing the offchannel/scanning work is only possible
with a NL80211 application.

An RTNL only application has no idea about scanning/offchannel work, so
it has no way of knowing what -EBUSY means in the wireless device
context, or a way to prevent this from happening (stopping scanning or
offchannel before changing the MAC).

Now, another example would be an RTNL application that only cares about
non-wireless devices (like ethernet), and in this case it could benefit
from some type of RTNL specific flag and not care about a NL80211 flag.

Since there is no such application that cares about this RTNL flag
(since it doesn't exist) I would say that adding NL80211 flag is the
way to go. If someone wants to add an RTNL flag I would say that is
appropriate too. But in terms of this feature, a NL80211 flag is really
what fits best since the changes are wireless specific.

Thanks,
James

> 
> johannes
>