--- Begin Message ---
Just two comments inline.
On Mon, Dec 5, 2022 at 3:45 PM Denis Ovsienko via tcpdump-workers <
tcpdump-workers@lists.tcpdump.org> wrote:
>
>
>
> ---------- Forwarded message ----------
> From: Denis Ovsienko <de...@ovsienko.info>
> To: tcpdump-workers@lists.tcpdump.org
> Cc:
> Bcc:
> Date: Mon, 5 Dec 2022 23:44:46 +0000
> Subject: Re: [tcpdump-workers] Public Release of Z-Wave G.9959 TAP
> Specification
> On Mon, 5 Dec 2022 11:53:11 -0800
> Chris Brandson via tcpdump-workers <tcpdump-workers@lists.tcpdump.org>
> wrote:
>
> > Hello everyone,
> >
> > We are pleased to publish the draft Z-Wave G.9959 TAP Specification
> [...]
>
> Hello Chris.
>
> Thank you for preparing the document. I am not familiar with the
> standard to make more substantial comments, but a few things seem to
> require some attention:
>
> * Page 7: "Length is a minimum of 4 octets and must be a multiple of 4.
> The addition of new TLVs does not and must not require incrementing
> the version number." -- this definition automatically creates a
> problem space on the receiving/reading end of this encoding because
> slightly more than 75% of all possible length values are invalid by
> definition. The classic solution would be not to include the first 4
> octets into the length, and to count in multiples of 4, this way any
> length value would be valid, free of underflows and better consumable
> by simple parsers such as BPF bytecode.
>
> Note that the TAP header length field differs from the TLV header length
field. I'm not opposed to omitting the header bytes from the TAP header
length but just wanted to make sure we all understood the difference in
this part of the document. For a little background: I know this
specification was based on IEEE802.15.4-TAP (
https://github.com/jkcko/ieee802.15.4-tap) and that was based on
IEEE802.11-RADIOTAP (https://www.radiotap.org/). This was also likely
because IP header length and total length fields also includes the header
bytes.
* Page 8: "Z-Wave PHY Payload" -- is this the PPDU from Figure A.7 of
> G.9959 2015/01, starting from "preamble sequence"?
>
> * Page 9: "length - number of octets for type in value field (not
> including padding)" -- this definition looks better, but in the next
> three TLV diagrams the length does include the T and the L.
>
> 4.1 FCS length should be 1 rather than 4 here. Padding not included.
Regards,
James
--- End Message ---
_______________________________________________
tcpdump-workers mailing list
tcpdump-workers@lists.tcpdump.org
https://lists.sandelman.ca/mailman/listinfo/tcpdump-workers