On 4/23/2025 9:11 PM, Johannes Berg wrote:
>
>> +++ b/net/mac80211/cfg.c
>> @@ -3080,6 +3080,7 @@ static int ieee80211_set_tx_power(struct wiphy *wiphy,
>> struct ieee80211_local *local = wiphy_priv(wiphy);
>> struct ieee80211_sub_if_data *sdata;
>> enum nl80211_tx_power_setting txp_type = type;
>> + struct ieee80211_chanctx_conf *conf;
>> bool update_txp_type = false;
>> bool has_monitor = false;
>> int user_power_level;
>> @@ -3155,6 +3156,12 @@ static int ieee80211_set_tx_power(struct wiphy *wiphy,
>> if (!link)
>> continue;
>>
>> + if (radio_id >= 0 && radio_id < wiphy->n_radio) {
>> + conf = wiphy_dereference(wiphy, link->conf->chanctx_conf);
>> + if (!conf || conf->radio_idx != radio_id)
>> + continue;
>> + }
>> +
>> link->user_power_level = local->user_power_level;
>> if (txp_type != link->conf->txpower_type)
>> update_txp_type = true;
>> @@ -3175,6 +3182,12 @@ static int ieee80211_set_tx_power(struct wiphy *wiphy,
>> if (!link)
>> continue;
>>
>> + if (radio_id >= 0 && radio_id < wiphy->n_radio) {
>> + conf = wiphy_dereference(wiphy, link->conf->chanctx_conf);
>> + if (!conf || conf->radio_idx != radio_id)
>> + continue;
>> + }
>> +
>>
>
>
> Hmm. Is this really enough? What if the link gets disabled and re-
> enabled on a whole different chanctx on a different radio? Or other
> things like that?
>
> Seems like we may need to change how the TX power is stored in mac80211,
> rather than just paper over it like this?
>
This means that get_tx_power handling should also be changed. So I might have to work
on resigning this. I'll handle this parallelly. Meanwhile, can I send out the first three
patches handling get/set RTS threshold in a separate patch series? I will re-design
tx-power patches and send separately.
> johannes
