Greg Troxel wrote:
> It seems that contents of 'struct ieee80211_key' can correspond to not
> doing any crypto and not having a key, but I'm not sure.  Should there
> be some test (at least one of xmit/recv/group flags being set?) in the
> iterator that writes all keys, or should all 4 be written
> unconditionally?
> 

net80211 keys that have wk_keyix != IEEE80211_KEYIX_NONE should be 
written regardless of the flags.  It doesn't appear the split tx/rx 
hardware key indices has made it over from freebsd yet; when that 
happens you'll need to look at both to hit all the h/w key cache entries.

However given that we're talking about 4 keys here...

	Sam