Re: [PATCH v3 02/11] cfg80211: handle 6 GHz capability of new station

From: Rajkumar Manoharan <rmanohar@codeaurora.org>
To: Johannes Berg <johannes@sipsolutions.net>
Cc: kvalo@codeaurora.org, linux-wireless@vger.kernel.org,
	ath11k@lists.infradead.org
Subject: Re: [PATCH v3 02/11] cfg80211: handle 6 GHz capability of new station
Date: Wed, 27 May 2020 16:24:59 -0700	[thread overview]
Message-ID: <17385b3a2fca16521858bf0956904698@codeaurora.org> (raw)
In-Reply-To: <bf0f6c17c188901477c1d6aa61aab6143c4ac635.camel@sipsolutions.net>

On 2020-05-27 07:00, Johannes Berg wrote:
> On Wed, 2020-05-13 at 12:44 -0700, Rajkumar Manoharan wrote:
>> Handle 6 GHz HE capability while adding new station. It will be used
>> later in mac80211 station processing.
> 
> This doesn't compile without the next patch.
> 
My bad.. I must have overlooked while splitting the patch. :(

>> +       const struct ieee80211_he_6ghz_band_cap *he_6ghz_capa;
> 
> This we made just an __le16, any particular reason to have the struct?
> It does need to be a pointer for the "no changes" case, but the struct
> seems a bit overkill?
> 
Initially I thought of splitting into two u8 for a_mpdu_params and info.
Later changed to __le16 but retained struct. Nothing else :)

>> + * @NL80211_ATTR_HE_6GHZ_CAPABILITY: HE 6 GHz Band Capability element 
>> (from
>> + *	association request when used with NL80211_CMD_NEW_STATION).
> 
> That we have pretty much identically.
> 
>> @@ -2998,6 +3003,7 @@ enum nl80211_attrs {
>>  #define NL80211_HE_MAX_CAPABILITY_LEN           54
>>  #define NL80211_MAX_NR_CIPHER_SUITES		5
>>  #define NL80211_MAX_NR_AKM_SUITES		2
>> +#define NL80211_HE_6GHZ_CAPABILITY_LEN		2
> 
> This not, we defined it just to be a U16.
> 
>> +	[NL80211_ATTR_HE_6GHZ_CAPABILITY] = {
>> +		.type = NLA_EXACT_LEN,
>> +		.len = NL80211_HE_6GHZ_CAPABILITY_LEN,
>> +	},
> 
> This no longer exists, but I guess I'll just take our patch for the U16
> here.
> 
>> +	/* Ensure that HT/VHT capabilities are not set for 6 GHz HE STA */
>> +	if (params.he_6ghz_capa && (params.ht_capa || params.vht_capa))
>> +		return -EINVAL;
> 
> Not sure this makes much sense? We can only check what's being set at
> the same time, so multiple calls here would still be possible ...
> doesn't hurt much though.
> 
> We didn't have this check, and have one additional check:
> 
> @@ -6170,7 +6200,7 @@ static int nl80211_new_station(struct sk_buff
> *skb, struct genl_info *info)
>                 params.vht_capa = NULL;
> 
>                 /* HE requires WME */
> -               if (params.he_capa_len)
> +               if (params.he_capa_len || params.he_6ghz_capa)
>                         return -EINVAL;
>         }
> 
Fine. One more thing. Pradeep found that 6 GHz capability is not filled 
in set_station.
Please handle that in your series. I'm fine with rest of the changes you 
mentioned.

--- a/net/wireless/nl80211.c
+++ b/net/wireless/nl80211.c
@@ -5893,6 +5893,10 @@ static int nl80211_set_station_tdls(struct
genl_info *info,
                         return -EINVAL;
         }

+       if (info->attrs[NL80211_ATTR_HE_6GHZ_CAPABILITY])
+               params->he_6ghz_capa =
+                       
nla_data(info->attrs[NL80211_ATTR_HE_6GHZ_CAPABILITY]);
+

-Rajkumar
