[PATCH v4 1/4] nl80211: set NL80211_SCAN_FLAG_COLOCATED_6GHZ in scan

[Peer, Ilan]

Hi,

> -----Original Message-----
> From: Jouni Malinen <j at w1.fi>
> Sent: Monday, April 18, 2022 18:16
> To: Peer, Ilan <ilan.peer at intel.com>
> Cc: hostap at lists.infradead.org; Mussai, Tova <tova.mussai at intel.com>;
> Otcheretianski, Andrei <andrei.otcheretianski at intel.com>; Stern, Avraham
> <avraham.stern at intel.com>
> Subject: Re: [PATCH v4 1/4] nl80211: set
> NL80211_SCAN_FLAG_COLOCATED_6GHZ in scan
> 
> On Mon, Apr 11, 2022 at 05:21:58PM +0300, Ilan Peer wrote:
> > Set NL80211_SCAN_FLAG_COLOCATED_6GHZ in the scan params to enable
> > scanning for co-located AP's found in 2.4/5 GHz bands when not
> > scanning passively. Do so only when collocated scanning is not
> > disabled by higher layer logic.
> 
> Where does this "not scanning passively" part comes from? nl80211.h
> documents this flag in the following way:
> 
> "NL80211_SCAN_FLAG_COLOCATED_6GHZ: scan for colocated APs reported
> by
> 2.4/5 GHz APs"
> 
> Is that incorrect or just incomplete? They way I read that nl80211.h comment
> is that this flag is or adding some additional steps to scanning, but the
> comment in this proposed hostap.git change says
> following:
> 
> > diff --git a/src/drivers/driver_nl80211_scan.c
> > b/src/drivers/driver_nl80211_scan.c
> > @@ -203,6 +203,17 @@ nl80211_scan_common(struct i802_bss *bss, u8
> cmd,
> > +		/*
> > +		 * If allowed, scan for 6GHz APs that are reported by other
> > +		 * APs. If the flag is not set and 6GHz channels are to be
> > +		 * scanned, they will be scanned passively.
> > +		 */
> > +		wpa_printf(MSG_DEBUG, "nl80211: non_coloc_6ghz=%u",
> > +			   params->non_coloc_6ghz);
> > +
> > +		if (!params->non_coloc_6ghz)
> > +			scan_flags |=
> NL80211_SCAN_FLAG_COLOCATED_6GHZ;
> 
> This seems to imply that if the flag is there, there would actually be less
> scanning rather than the other way around..
> 
> What exactly is the expected behavior here? What is that "6 GHz channels
> are to be scanned" referring to?
> 
> Should the nl80211.h documentation be extended to cover all the details that
> this flag is really doing? I cannot really figure out whether this proposed
> hostap.git patch is doing what the flag is supposed to do..
> 
> Is this somehow connected to the preferred scanning channels (PSC)?
> 

The scan procedures defined in Draft P802.11ax_D8.0 section 27. 17.2.3, limit the
transmissions of probe requests on the 6GHz channels such that a non-AP station
is allowed to transmit a probe requests only if there is information that there is
an AP operating on the channel. For non PSC channels, this information is expected
to be available based on information obtained from RNR elements (found in
beacons/probe responses received on the 2/5GHz channels) which indicate the presence
of collocated APs on 6GHz channels.

Without enabling collocated scan logic, i.e., using the information obtained from the scan
on the 2/5GHz bands, a non-AP station is very limited in the active scanning it can perform
on non-PSC channels which would mean that in practice it would need to passively scan
these channels.

With collocated scanning enabled, the non-AP station would both limit the
scan for channels on which APs are expected to be operating and also be able to transmit
probe requests, thus shortening the overall scan time.

I think that the nl80211 API documentation can be extended to describe the expected outcome
of setting (and not setting) the NL80211_SCAN_FLAG_COLOCATED_6GHZ flag, so the implications
on the scan would be clear. I can also further clarify this patch.

Hope this clarifies things.

Regards,

Ilan.