[PATCH 18/19] P2PS: add a wildcard with other advertised service info

[Jouni Malinen]

On Tue, Jun 23, 2015 at 02:46:31PM +0000, Stepanov, Max wrote:
> >From: Jouni Malinen [mailto:j at w1.fi]
> >P2PS: Fix P2P_FIND seek parameter parsing
> 
> Please see P2PS: Fix p2p_find last parameter handling
> 
> 
> >P2PS: Fix Probe Response frame building in error cases
> 
> Please see P2PS: Fix attribute addition in p2p_buf_add_service_instance()
> It fixes a small enumeration issue (with my credits to Ilan who noticed it)

Thanks!

> >P2PS: Fix service hash matching for org.wi-fi.wfds
> >P2PS: Fix org.wi-fi.wfds matching when building the response
> 
> IMHO, these two patches are tricky. I tend to think that the original implementation was correct.
> In P2PS 1.1 spec section 3.4.1: 
> "The ASP?at the?Service Advertiser shall respond positively if a Service Seeker sends the hash value for ?org.wi-fi.wfds? and has at least one?advertised?P2Ps?service."

These patches were based on this note in 3.4.3.2:

'If the ASP finds a match with the Service Hash value of
"org.wi-fi.wfds" (i.e. advertises one of the Wi-Fi Alliance defined
services listed in 3.2), then in the Probe Response it shall set the
value of its Service Name field to "org.wi-fi.wfds".'

The bullet item that you are referring from 3.4.1 is also starting with
"If any one of Wi-Fi Alliance defined services..". While that is a
separate sentence just before the one you quoted above, I would be
tempted to interpret the "P2Ps service" in that second sentence to be
constrained by the "Wi-Fi Alliance defined".

> Also P2PS test verifying discovery of multiple services using Prefix Search assumes that advertiser publishes ?org.wi-fi.wfds" with adv_id 0 together with ?test.abc.xyz?.
> If to assume that ASP doesn't add ?org.wi-fi.wfds? explicitly, it looks like the right behavior is to reply with ?org.wi-fi.wfds? if at least one service advertisement is present.

I'm not completely sure I understood what you mean here (and this is
unlikely to be a suitable item for this mailing list anyway). I'll ask
someone who worked with the specification to provide an opinion on how
this is supposed to work and how those two locations in the spec are
supposed to be interpreted (and whether they should be clarified to make
it obvious what they mean).

-- 
Jouni Malinen                                            PGP id EFC895FA