mac80211: Why probe after each scan?

mac80211: Why probe after each scan?

[Paul Stewart]

I found a small bit of behavior related in some vague way to
connection monitoring:  During the scan-completed notification
(ieee80211_mlme_notify_scan_completed) we call
ieee80211_restart_sta_timer() on all our interfaces, which in turn
queues up the &ifmgd->monitor_work, which does a probe request of our
AP.  Why should we do a probe request of your associated AP every time
a scan completes?  Wouldn't the other active monitoring schemes (like
beacon loss) be sufficient for a "scan finished" condition?

--
Paul

Re: mac80211: Why probe after each scan?

[Dan Williams]

On Fri, 2010-09-17 at 20:01 -0700, Paul Stewart wrote:
> I found a small bit of behavior related in some vague way to
> connection monitoring:  During the scan-completed notification
> (ieee80211_mlme_notify_scan_completed) we call
> ieee80211_restart_sta_timer() on all our interfaces, which in turn
> queues up the &ifmgd->monitor_work, which does a probe request of our
> AP.  Why should we do a probe request of your associated AP every time
> a scan completes?  Wouldn't the other active monitoring schemes (like
> beacon loss) be sufficient for a "scan finished" condition?

One reason I can think of is to quickly determine whether your AP is
still around and active given that you may have just been off-channel
for a significant period of time.  Not all drivers background scan yet.
But with any sane driver, you'd have already sent your "no longer in
powersave mode" NULL packet to the AP right after finishing the scan, so
perhaps the probe is redundant.

Dan

Re: mac80211: Why probe after each scan?

[Paul Stewart]

But mac80211 also does beacon loss detection.  I'd much prefer more
aggressive (shorter timeout) beacon loss parameters (perhaps
configured that way only after end-of-scan, although I believe in
general the timeout is rather generous even for the normal case) than
to require round-trips with the AP just to find out if it's there.
For the most part, I think beacons do the job of AP presence detection
fairly well.

--
Paul

On Mon, Sep 20, 2010 at 3:13 PM, Dan Williams <dcbw@redhat.com> wrote:
> On Fri, 2010-09-17 at 20:01 -0700, Paul Stewart wrote:
>> I found a small bit of behavior related in some vague way to
>> connection monitoring:  During the scan-completed notification
>> (ieee80211_mlme_notify_scan_completed) we call
>> ieee80211_restart_sta_timer() on all our interfaces, which in turn
>> queues up the &ifmgd->monitor_work, which does a probe request of our
>> AP.  Why should we do a probe request of your associated AP every time
>> a scan completes?  Wouldn't the other active monitoring schemes (like
>> beacon loss) be sufficient for a "scan finished" condition?
>
> One reason I can think of is to quickly determine whether your AP is
> still around and active given that you may have just been off-channel
> for a significant period of time.  Not all drivers background scan yet.
> But with any sane driver, you'd have already sent your "no longer in
> powersave mode" NULL packet to the AP right after finishing the scan, so
> perhaps the probe is redundant.
>
> Dan
>
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-wireless" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>