Re: DMG Support

Krishna Chaitanya <[email protected]> Fri, 28 Jun 2019 13:32:16 +0530
Newsgroups org.netbsd.radiotap
Message-ID <CABPxzYKAc=J6x0UcC4TpSvgfuymyWS4N8bwCnA_dnDdS1rdgAw@mail.gmail.com>
On Fri, Jun 28, 2019 at 2:12 AM Richard Sharpe
<[email protected]> wrote:
>
> On Thu, Jun 27, 2019 at 12:56 PM Krishna Chaitanya
> <[email protected]> wrote:
> >
> > On Thu, Jun 27, 2019 at 11:46 PM Richard Sharpe
> > <[email protected]> wrote:
> > >
> > > On Thu, Jun 27, 2019 at 10:16 AM Krishna Chaitanya
> > > <[email protected]> wrote:
> > > >
> > > > Hi Guys,
> > > >
> > > > I am looking at support for DMG specific fields to radiotap, the only
> > > > discussion I have found didn't reach any conclusion [1].
> > > >
> > > > Is there a plan to support these fields? It definitely needs a DMG
> > > > Type field with at least below params (from top of my head)
> > > >
> > > > - MCS (and extended MCS) along with data rate mapping
> > > >    - C-PHY
> > > >    - SC-PHY
> > > >    - Low power SC-PHY
> > > >    -OFDM
> > > > - Beam/Sector/Antenna ID Information
> > > > - RCPI
> > >
> > > Is there any kernel capture code or any other device that would support this?
> > Yes, https://elixir.bootlin.com/linux/latest/source/drivers/net/wireless/ath/wil6210/txrx.c#L317.
> > (Only upstream DMG driver) But, it has minimal info due to lack of DMG
> > specific fields.
> > Although, I am not sure whether it supports other PHY info than MCS and RSSI.
> > We could query beams, but beam info is not present in rx_desc.
> >
> > > The general procedure to get there, however, is:
> > >
> > > 1. Propose the new header.
> > > 2. Implement code in Wireshark to handle the header and include useful
> > > info in the pseudo-header
> > > 3. Implement code that can generate sample captures with the new header.
> > > 4. Get someone to add such DMG support to DMG-capable devices to
> > > generate the new header.
> >
> > Yes, that's the norm. Before proposing the fields, just wanted to
> > check if there is an ongoing work or
> > some kind of decision. So, should I go ahead and propose those fields?
>
> I would say yes. Take note that any new headers should use the new TLV format.
>
> Also, there has been a proposal for an S1G header so you might want to
> use the next available type.
Okay, but looks like the TLV and S1G proposals are still WIP, so I will wait for
them to be merged and then start on DMG. Thanks.