Re: [PATCH v3 1/1] mac80211: tell driver when dtim change detected

["Luis R. Rodriguez" <mcgrof@gmail.com>]

On Mon, Jan 25, 2010 at 10:33 AM, Johannes Berg
<johannes@sipsolutions.net> wrote:
> On Mon, 2010-01-25 at 10:18 -0800, Luis R. Rodriguez wrote:
>
>> > Yes, since 0 is nonsensical, we would use 1 instead.
>>
>> OK -- it seems we have a consensus waiting for a beacon would be good,
>> but it would be good to have something in for stable. Is the patch Guy
>> posted acceptable for that purpose? That is, merge this for now until
>> we get beacon wait implemented prior to association.
>
> Not sure we even need it in stable that badly? It seems it only hurts
> power consumption

Good point, we'd just be waking up at every beacon if power save is
enabled since the default is indeed 1.

> somewhat.

Depends how common this is. I am not sure what the usual DTIM value is
either though, but if its not 1 and this is indeed very common that'd
be a bit sloppy for stable.

> And I just implemented waiting today too.

I saw, you're the fucking man.

> Although Bob has a point, we could change the way we pass the DTIM
> period to the driver, only pass it in the powersave callbacks, and then
> we could associate without it and just enable powersave only after
> receiving a beacon. That assumes that drivers don't use it for anything
> else though, of course, but I don't see anything else you could use it
> for really.

Neat -- but that would mean doing the same for other beacon-sync IE
information. What other stuff goes out in beacons that does not go
into probe responses? I think I read a thread the other day about some
WPS information.

  Luis