Hi Johannes, On 01/31/2018 03:53 PM, Johannes Berg wrote:
+static int nl80211_tx_control_port(struct sk_buff *skb, struct genl_info *info) +{ + struct cfg80211_registered_device *rdev = info->user_ptr[0]; + struct net_device *dev = info->user_ptr[1]; + struct wireless_dev *wdev = dev->ieee80211_ptr; + const u8 *buf; + size_t len; + u8 *dest; + u16 proto; + bool noencrypt; + int err; + + if (!wiphy_ext_feature_isset(&rdev->wiphy, + NL80211_EXT_FEATURE_CONTROL_PORT_OVER_NL80211)) + return -EOPNOTSUPP; + + if (!rdev->ops->tx_control_port) + return -EOPNOTSUPP;I wonder if maybe we should accept this command only from the socket owner? Is there a use case for something else?
Yes I thought about adding such a check as I think it wouldn't make sense to accept control port frames from any other application besides the socket owner. However the socket owner stuff was a bit controversial, so I left it out.
I would support adding this check...
Actually, then again, that's awkward because doing events and commands on the same socket doesn't mix all _that_ well. Perhaps we just need to fix that problem in libnl or something and be done with it ...
There are no issues on our side.. Regards, -Denis
