[ath9k-devel] Bringing up a virtual wiphy on AR9160 brings down the node

[Ben Greear <greearb@candelatech.com>]

On 07/06/2010 09:57 AM, Jouni Malinen wrote:
> On Tue, 2010-07-06 at 09:42 -0700, Javier Cardona wrote:
>
>> We have a dedicated target for this, so we can quickly test patches or
>> diffent kernel versions.
>
> May I ask what are you trying to achieve with virtual wiphys? Are you
> trying to do multi-channel operations? If not, some other option may be
> more appropriate design.
>
>> Does anyone know if virtual wiphys should work on the AR9160 ?
>
> I have not tested this in months, so cannot really say what the current
> state is. Anyway, please note that the virtual wiphy design is unlikely
> to be supported in the future. It was an experimental change and it
> seems likely that other designs (moving more of this into mac80211) are
> the way to go in the future. As such, I think the virtual wiphy design
> will be removed once mac80211 can handle multi-channel operations with
> concurrent virtual interfaces. For single-channel cases, mac80211 vifs
> should already be used instead of trying to get virtual wiphy design
> working in ath9k.

I'm curious what kind of limitations we should expect for multi-channel
vifs on ath9k (and perhaps other?) hardware.

For instance, can the chip somehow listen on all or multiple channels
and then quickly flip to the right channel for transmitting for each
VIF?  Or, would it be somewhat like the v-phy logic where it sequentially
flips to new channels to briefly listen and do work?

Thanks,
Ben

-- 
Ben Greear <greearb@candelatech.com>
Candela Technologies Inc  http://www.candelatech.com