Received-SPF: pass (google.com: domain of linux-wireless-owner@vger.kernel.org designates 2620:137:e000::1:18 as permitted sender) client-ip=2620:137:e000::1:18;
Date:   Tue, 17 May 2022 10:37:58 -0700
From:   Jakub Kicinski <kuba@kernel.org>
To:     Johannes Berg <johannes@sipsolutions.net>
Cc:     Florian Fainelli <f.fainelli@gmail.com>, davem@davemloft.net,
        netdev@vger.kernel.org, edumazet@google.com, pabeni@redhat.com,
        alex.aring@gmail.com, stefan@datenfreihafen.org,
        mareklindner@neomailbox.ch, sw@simonwunderlich.de, a@unstable.cc,
        sven@narfation.org, linux-wireless@vger.kernel.org,
        linux-wpan@vger.kernel.org
Subject: Re: [PATCH net-next] net: ifdefy the wireless pointers in struct
 net_device
Message-ID: <20220517103758.353c2476@kernel.org>
In-Reply-To: <74bdbec0580ed05d0f18533eae9af50bc0a4a0ef.camel@sipsolutions.net>
References: <20220516215638.1787257-1-kuba@kernel.org>
        <8e9f1b04-d17b-2812-22bb-e62b5560aa6e@gmail.com>
        <74bdbec0580ed05d0f18533eae9af50bc0a4a0ef.camel@sipsolutions.net>
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Precedence: bulk

On Tue, 17 May 2022 09:48:24 +0200 Johannes Berg wrote:
> On Mon, 2022-05-16 at 19:12 -0700, Florian Fainelli wrote:
> > 
> > On 5/16/2022 2:56 PM, Jakub Kicinski wrote:  
> > > Most protocol-specific pointers in struct net_device are under
> > > a respective ifdef. Wireless is the notable exception. Since
> > > there's a sizable number of custom-built kernels for datacenter
> > > workloads which don't build wireless it seems reasonable to
> > > ifdefy those pointers as well.
> > > 
> > > While at it move IPv4 and IPv6 pointers up, those are special
> > > for obvious reasons.
> > > 
> > > Signed-off-by: Jakub Kicinski <kuba@kernel.org>  
> > 
> > Could not we move to an union of pointers in the future since in many 
> > cases a network device can only have one of those pointers at any given 
> > time?  
> 
> Then at the very least we'd need some kind of type that we can assign to
> disambiguate, because today e.g. we have a netdev notifier (and other
> code) that could get a non-wireless netdev and check like this:
> 
> static int cfg80211_netdev_notifier_call(struct notifier_block *nb,
>                                          unsigned long state, void *ptr)
> {
>         struct net_device *dev = netdev_notifier_info_to_dev(ptr);
>         struct wireless_dev *wdev = dev->ieee80211_ptr;
> [...]
>         if (!wdev)
>                 return NOTIFY_DONE;

Can we use enum netdev_ml_priv_type netdev::ml_priv and
netdev::ml_priv_type for this?