Re: [PATCH v5] cfg80211: save power spectral density(psd) of regulatory rule

[Wen Gong <wgong@codeaurora.org>]

Hi Johannes,

do you have comments about my description for PSD?

On 2021-10-26 19:26, Wen Gong wrote:
> On 2021-10-26 04:09, Johannes Berg wrote:
>> On Mon, 2021-10-11 at 15:48 +0800, Wen Gong wrote:
>>> 
>>> > IMO, Only power rules and PSD info might vary for AP and STATION. Rest
>>> > of the rules will remains same right?
>>> >
>>> The freq_range may also be different for AP and STATION.
>>> and reg_rules number also may also be different for AP and STATION.
>>> 
>>> for example:
>>> SUBORDINATE CLIENT of STANDARD POWER reg rules number 2
>>> reg rule 1: (5945 - 6425 @ 160) (0, 30) (FLAGS 0) (psd flag 1 EIRP 17
>>> dB/MHz)
>>> reg rule 2: (6525 - 6885 @ 160) (0, 30) (FLAGS 0) (psd flag 1 EIRP 17
>>> dB/MHz)
>>> 
>>> INDOOR AP reg rules number 1
>>> reg rule 1: (5945 - 7125 @ 160) (0, 24) (FLAGS 0) (psd flag 0 EIRP 0
>>> dB/MHz)
>> 
>> That seems right, but isn't that an orthogonal question?
>> 
>> Here, on this patch, we're discussing what data we should have in the
>> channel information, and it would seem that if it's different for
>> AP/client, then we do need both information stored, so that we can 
>> cope
>> with concurrency between AP and client?
>> 
>> If we additionally need to have different data for the regulatory 
>> rules
>> for AP and client, that might mean we need to go back and actually
>> change the code there *as well*, and then fill in the right fields in
>> this patch?
>> 
>> Unless somehow we're convinced that for this feature we don't need to
>> worry about concurrently using AP and client modes?
>> 
>> johannes
> 
> Currently these patches of mac80211/cfg80211/ieee80211 for LPI/SP/VLP 
> is
> the base patches, to enable the feature of LPI/SP/VLP, it still need 
> other
> patches of lower drivers such as ath11k to enable it. It will not have
> LPI/SP/VLP without patches of ath11k, it means all these patches will
> not take effect.
> 
> When lower driver such as ath11k set max_interfaces of
> ieee80211_iface_combination
> to 1, then it can not start more than 1 interface on the same
> ieee80211_hw/wiphy.
> When STATION interface is up, then AP interface can not start up. AP 
> interface
> can start up after STATION interfacedown. Also when AP interface is up,
> STATION interface can not start up. STATION interface can start up 
> after
> AP interface down.
> 
> I have sent out my ath11k
> patches(https://lore.kernel.org/linux-wireless/20211026111913.7346-1-quic_wgong@quicinc.com/),
> it will allow only one interface
> up simultaneously for the chip which enable LPI/SP/VLP feature in this
> patch: "ath11k: allow only one interface up simultaneously for WCN6855"
> https://lore.kernel.org/linux-wireless/20211026111913.7346-5-quic_wgong@quicinc.com/
> It means it will not have both AP/STA together and these patches of 
> mac80211/
> cfg80211/ieee80211 not need changes and it will not have bugs.
> 
> If there are some chip want to both enable LPI/SP/VLP feature and
> enable AP/STA simultaneously in same ieee80211_hw/wiphy in future,
> then he/she need to refine reg rules and channels of mac80211/cfg80211/
> ieee80211, but at that moment, this patch "cfg80211: save power
> spectral density(psd) of regulatory rule" still not need change.
> Because this patch is change in each reg rule/each channel in a
> low layer, the refine reg rules and channels is a high layer, they
> have no intersection.