Received-SPF: pass (google.com: domain of linux-wireless-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <9e37eb24f63dbac0b13ae8b36393c8f393599d1e.camel@sipsolutions.net>
Subject: Re: [RFC 1/2] wifi: mac80211: add a new field in
 ieee80211_rx_status for link id
From:   Johannes Berg <johannes@sipsolutions.net>
To:     Vasanthakumar Thiagarajan <quic_vthiagar@quicinc.com>,
        Jeff Johnson <quic_jjohnson@quicinc.com>
Cc:     linux-wireless@vger.kernel.org
Date:   Tue, 09 Aug 2022 20:12:38 +0200
In-Reply-To: <733b6f1a-8b07-4b23-2b10-af1ce68b6227@quicinc.com>
References: <20220802065019.20791-1-quic_vthiagar@quicinc.com>
         <20220802065019.20791-2-quic_vthiagar@quicinc.com>
         <29694008-63ef-4cd3-a6c6-f82fc8377be6@quicinc.com>
         <733b6f1a-8b07-4b23-2b10-af1ce68b6227@quicinc.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
User-Agent: Evolution 3.44.4 (3.44.4-1.fc36) 
MIME-Version: 1.0
Precedence: bulk

On Wed, 2022-08-03 at 22:16 +0530, Vasanthakumar Thiagarajan wrote:
>=20
> > > +=C2=A0=C2=A0=C2=A0 u8 link_id;
> > > =C2=A0 };
> > > =C2=A0 static inline u32
> >=20
> > in other parts of the MLO code the link_id is defined as int and a valu=
e=20
> > of -1 is used for a non-MLO link. but I don't know if that is currently=
=20
> > universally true.
> >=20
> > if that is curently universally true, do we want to now have divergent=
=20
> > definitions of a link_id?
>=20
> Good point, i see link_id is used both as unsigned and int based on
> their usage. The reason I preferred unsigned is that we can make use of=
=20
> the remaining 4-bits for some other purpose in future,=20
> ieee80211_rx_status has size limitation.
>=20

It's a bit tricky though - do we want to have 0 for all the drivers that
don't support MLO? Might not really be an issue, but OTOH not
initializing it should probably not result in a valid value otherwise
you might test something, think it's fine, and it really isn't.

I think we should spend a bit to have a validity indication in this
case. Even -1 wouldn't address it since you'd have to initialize to
that.

It works better the other way around in APIs etc. where we have a few
producers and many consumers (cfg80211 down to drivers), but less well
this way where we'd have to change drivers...

johannes