Re: [RFC 1/1] net: move IFF_LIVE_ADDR_CHANGE to public flag

From: Johannes Berg <johannes@sipsolutions.net>
To: James Prestwood <prestwoj@gmail.com>, Jakub Kicinski <kuba@kernel.org>
Cc: netdev@vger.kernel.org
Subject: Re: [RFC 1/1] net: move IFF_LIVE_ADDR_CHANGE to public flag
Date: Wed, 10 Aug 2022 19:17:19 +0200	[thread overview]
Message-ID: <9ec77cf1ffaa29aedd57c29ac77b525d0e700acf.camel@sipsolutions.net> (raw)
In-Reply-To: <d585f719af13d7a7194e7cb734c5a7446954bf01.camel@gmail.com>

On Wed, 2022-08-10 at 09:26 -0700, James Prestwood wrote:
> 
> Ok, so this is how I originally did it in those old patches:
> 
> https://lore.kernel.org/linux-wireless/20190913195908.7871-2-prestwoj@gmail.com/
> 
> i.e. remove_interface, change the mac, add_interface. 

Hah, I didn't even remember that ... sorry.

> But before I revive those I want to make sure a flag can be advertised
> to userspace e.g. NL80211_EXT_FEATURE_LIVE_ADDRESS_CHANGE. (or
> POWERED). Since this was the reason the patches got dropped in the
> first place.
> 

Well it seems that my objection then was basically that you have a
feature flag in nl80211, but it affects an RT netlink operation ...
which is a bit strange.

Thinking about that now, maybe it's not _that_ bad? Especially given
that "live" can mean different things (as discussed here), and for
wireless that doesn't necessarily mean IFF_UP, but rather something like
"IFF_UP + not active".

Jakub, what do you think?

(I'll also note you also have error handling problems in your patch, so
if/when you revive, please take a look at handling errors from add and
remove interface. Also indentation, and a comment on station/p2p-client
might be good, and the scanning check is wrong, can check scan_sdata
regardless of the iftype.)

johannes