> I don't know what your firmware/hardware capabilities are :) I was not referring to any specific architecture here. With the 2 choices you had mentioned (mesh must match either one of them / make this apply on mesh and IBSS as well ) we see that this new flag being defined in the interface combinations could better be applicable for all the beaconing interfaces. If we had restricted this to AP / P2P GO and have Mesh match either of them , a new request ( in future ? ) to have a different interval for Mesh, IMO would ask a need for new flag , isn't ? Our intention here was to avoid this by ensuring that this flag is applicable for all the beaconing interface combinations.
>I'd argue that instead of having the interface combinations flag, that >nl80211 attribute could carry the GCD? If your say here is to just use the wiphy parameter rather than having the flag in the interface combinations, signifying the support for different beacon intervals , Yes we too agree to your point. This is considering the fact that the current implementation restricts the validation of the beacon interval (cfg80211_validate_beacon_int ) to only AP / P2P GO. Neither the Mesh / IBSS ( join ) have this restriction . Not sure why this is not imposed for Mesh / IBSS though. After introducing this flag in the interface combinations, we do feel that currently it is not going to apply for Mesh / IBSS , unless there is a future need to do so , which we are not aware of . With the current restriction of validating the beacon interval only for AP / P2P GO , we do agree to have a single parameter advertised by the drivers intending both - support the different beacon intervals and also signify the GCD among them (a non-zero value for "min_diff_beacon_interval_multiplier" should address this , isn't ). Regards, Sunil -----Original Message----- From: Johannes Berg [mailto:johan...@sipsolutions.net] Sent: Tuesday, August 9, 2016 1:27 PM To: Undekari, Sunil Dutt <usd...@qti.qualcomm.com>; Kushwaha, Purushottam <pkush...@qti.qualcomm.com> Cc: linux-wireless@vger.kernel.org; Malinen, Jouni <jo...@qca.qualcomm.com>; Kondabattini, Ganesh <gane...@qti.qualcomm.com>; Kalikot Veetil, Mahesh Kumar <mkali...@qca.qualcomm.com>; Hullur Subramanyam, Amarnath <amarn...@qca.qualcomm.com>; Kumar, Deepak (QCA) <djin...@qti.qualcomm.com> Subject: Re: [PATCH v4] cfg80211: Provision to allow the support for different beacon intervals > I guess , we can extend this to mesh and IBSS as well. I don't know what your firmware/hardware capabilities are :) > > It seems to me that if I were to specify beacon intervals which have > > a very small GCD, you'll run into trouble when actually sending > > beacons. > > Perhaps there should be a requirement on the GCD? > Can we have this published by the host drivers through a new wiphy > parameter , say "min_diff_beacon_interval_multiplier". This set's the > expectation that any different beacon intervals on the wiphy shall be > a multiple of this parameter which is advertised by the host driver , > isn't ? > I'd argue that instead of having the interface combinations flag, that nl80211 attribute could carry the GCD? johannes
