On Thu, 2013-05-16 at 20:11 +0300, Jouni Malinen wrote: > - if (!info->attrs[NL80211_ATTR_STA_AID]) > + if (!info->attrs[NL80211_ATTR_STA_AID] && > + !info->attrs[NL80211_ATTR_PEER_AID]) > return -EINVAL; Technically here I think you could check that this attribute isn't used with non-TDLS stations, since in new_station you know what it's going to be, right? > @@ -3985,7 +3989,10 @@ static int nl80211_new_station(struct sk_buff *skb, struct genl_info *info) > params.listen_interval = > nla_get_u16(info->attrs[NL80211_ATTR_STA_LISTEN_INTERVAL]); > > - params.aid = nla_get_u16(info->attrs[NL80211_ATTR_STA_AID]); > + if (info->attrs[NL80211_ATTR_STA_AID]) > + params.aid = nla_get_u16(info->attrs[NL80211_ATTR_STA_AID]); > + else > + params.aid = nla_get_u16(info->attrs[NL80211_ATTR_PEER_AID]); This is a bit strange? Wouldn't you want to prefer the new attribute, so you can put some possibly invalid value into the old attribute for old kernels? Otherwise what's the reason for even reading the new attribute in this code at all, if new userspace code must set the old attribute for old kernels anyway? johannes -- To unsubscribe from this list: send the line "unsubscribe linux-wireless" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
