[PATCH 3/4] Fix connection to AP mode

[Kalle Valo]

Eugene Krasnikov <k.eugene.e at gmail.com> writes:

> Set correct AID in config_sta and set stakey.
> Also announce HW caps otherwise hostapd will
> not bind to interface.
>
> Signed-off-by: Eugene Krasnikov <k.eugene.e at gmail.com>

[...]

> @@ -608,7 +612,12 @@ static struct ieee80211_supported_band wcn_band_2ghz = {
>  	.bitrates	= wcn_legacy_rates,
>  	.n_bitrates	= ARRAY_SIZE(wcn_legacy_rates),
>  	.ht_cap		= {
> -		.cap = IEEE80211_HT_CAP_GRN_FLD | IEEE80211_HT_CAP_SGI_20 |
> +		.cap = IEEE80211_HT_CAP_GRN_FLD |
> +		       IEEE80211_HT_CAP_SGI_20 |
> +		       IEEE80211_HT_CAP_DSSSCCK40 |
> +		       IEEE80211_HT_CAP_LSIG_TXOP_PROT|
> +		       IEEE80211_HT_CAP_SGI_40 |
> +		       IEEE80211_HT_CAP_SUP_WIDTH_20_40 |

Do we really support and enable all those in firmware now? I'm worried
that enabling all sort of features without testing them creates bugs
which are not that easy to find.

>  			(1 << IEEE80211_HT_CAP_RX_STBC_SHIFT),
>  		.ht_supported = true,
>  		.ampdu_factor = IEEE80211_HT_MAX_AMPDU_8K,
> @@ -666,7 +675,8 @@ static int wcn36xx_init_ieee80211(struct wcn36xx *wcn)
>  	};
>  
>  	wcn->hw->flags = IEEE80211_HW_SIGNAL_DBM |
> -		IEEE80211_HW_HAS_RATE_CONTROL;
> +		IEEE80211_HW_HAS_RATE_CONTROL |
> +		IEEE80211_HW_REPORTS_TX_ACK_STATUS;

This would be good to have in patch 4 (the workaround patch).

>  	wcn->hw->wiphy->interface_modes = BIT(NL80211_IFTYPE_STATION) |
>  		BIT(NL80211_IFTYPE_AP) |
> @@ -683,6 +693,11 @@ static int wcn36xx_init_ieee80211(struct wcn36xx *wcn)
>  
>  	wcn->hw->wiphy->cipher_suites = cipher_suites;
>  	wcn->hw->wiphy->n_cipher_suites = ARRAY_SIZE(cipher_suites);
> +	wcn->hw->wiphy->flags |= WIPHY_FLAG_AP_PROBE_RESP_OFFLOAD;

Ah, you had that here. But I still think the extra call to template
update is not needed.

> +	wcn->hw->wiphy->probe_resp_offload =
> +		NL80211_PROBE_RESP_OFFLOAD_SUPPORT_WPS |
> +		NL80211_PROBE_RESP_OFFLOAD_SUPPORT_WPS2 |
> +		NL80211_PROBE_RESP_OFFLOAD_SUPPORT_P2P;

Do we really support all these? I would prefer to enable only after we
have verified that the offload is working.

-- 
Kalle Valo