Re: [PATCH net-next] net: ifdefy the wireless pointers in struct net_device

[Jakub Kicinski <kuba@kernel.org>]

On Tue, 17 May 2022 09:51:31 +0200 Johannes Berg wrote:
> On Mon, 2022-05-16 at 14:56 -0700, Jakub Kicinski wrote:
> > 
> > +#if IS_ENABLED(CONFIG_WIRELESS)
> >  	struct wireless_dev	*ieee80211_ptr;
> > +#endif  
> 
> Technically, you should be able to use CONFIG_CFG80211 here, but in
> practice I'd really hope nobody enables WIRELESS without CFG80211 :)

ack

> > +++ b/include/net/cfg80211.h
> > @@ -8004,10 +8004,7 @@ int cfg80211_register_netdevice(struct net_device *dev);
> >   *
> >   * Requires the RTNL and wiphy mutex to be held.
> >   */
> > -static inline void cfg80211_unregister_netdevice(struct net_device *dev)
> > -{
> > -	cfg80211_unregister_wdev(dev->ieee80211_ptr);
> > -}
> > +void cfg80211_unregister_netdevice(struct net_device *dev);  
> 
> Exported functions aren't free either - I think in this case I'd
> (slightly) prefer the extra ifdef.

fine

> Anyway, we can do this, but I also like Florian's suggestion about the
> union, and sent an attempt at a disambiguation patch there.

Would you be willing to do that as a follow up? Are you talking about
wifi only or all the proto pointers?

As a netdev maintainer I'd like to reduce the divergence in whether 
the proto pointers are ifdef'd or not.