Re: [PATCH] nl80211: Add notification for dropped Deauth/Disassoc

[Johannes Berg <johannes@sipsolutions.net>]

On Tue, 2010-12-14 at 08:51 +0100, Johannes Berg wrote:
> On Tue, 2010-12-14 at 00:00 +0200, Jouni Malinen wrote:
> 
> > @@ -1539,12 +1539,30 @@ ieee80211_drop_unencrypted_mgmt(struct i
> >  	if (rx->sta && test_sta_flags(rx->sta, WLAN_STA_MFP)) {
> >  		if (unlikely(!ieee80211_has_protected(fc) &&
> >  			     ieee80211_is_unicast_robust_mgmt_frame(rx->skb) &&
> > -			     rx->key))
> > +			     rx->key)) {
> > +			if (ieee80211_is_deauth(fc))
> > +				cfg80211_send_unprot_deauth(rx->sdata->dev,
> > +							    rx->skb->data,
> > +							    rx->skb->len);
> 
> This is the RX path.
> 
> > +void cfg80211_send_unprot_disassoc(struct net_device *dev, const u8 *buf,
> > +				   size_t len)
> > +{
> > +	struct wireless_dev *wdev = dev->ieee80211_ptr;
> > +	struct wiphy *wiphy = wdev->wiphy;
> > +	struct cfg80211_registered_device *rdev = wiphy_to_dev(wiphy);
> > +
> > +	wdev_lock(wdev);
> > +	nl80211_send_unprot_disassoc(rdev, dev, buf, len, GFP_KERNEL);
> 
> And this uses GFP_KERNEL. Surely that can't be right?

And the mutex too, for that matter, which seems unnecessary anyway.

johannes