Re: [RFC] VHT fields

[Johannes Berg <johannes-cdvu00un1VgdHxzADdlk8Q@public.gmane.org>]

On Fri, 2012-08-24 at 07:46 +0000, Alwin Beukers wrote:
> > We're just discussing for Linux how to specify the channel parameters
> > and we need to consider 80+80 as well. Maybe these discussions should
> > cross-pollinate.
> 
> Yes. However, this issue isn't directly related to or blocking the VHT field 
> proposal, is it? Can I go ahead and post an adaptation request for that?

Right.

> > I see two basic options:
> >  * continue what we do now, and specify the control channel & type
> >    around it (40+/40+,...)
> >    + this is how linux works now
> >    - many many channel types (4 for 80 MHz, 8 for 160 MHz)
> >    - doesn't handle 80+80
> 
> When it comes to radiotap, this might do as an interim solution for chips
> that don't support 80+80 yet. The new channel types would fit in the
> XChannel field.

We could even do both unless the channel is 80+80 for older parsers like
existing wireshark since they'd then be able to parse at least
something.

> >  * specify the center frequency, width & control channel offset
> >    + more aligned with the spec (?)
> >    - requires backward compat handling
> >    - doesn't handle 80+80
> 
> I feel this is the best long-term solution. The 80+80 case could be handled 
> By adding a secondary center frequency.

Yeah, something like that. We need to work that out in Linux too, not
even sure there are chips that will support 80+80 though?

johannes