On Mon, 2008-08-18 at 22:10 +0300, Jouni Malinen wrote: > + * @drop_unencrypted: Whether to drop unencrypted frames > + * (0 = no, 1 = yes, -1 = do not change) Should that be part of some encryption settings instead? Not really sure. > + * @userspace_mlme: Whether to use userspace MLME > + * (0 = no, 1 = yes, -1 = do not change) I don't really like this much, it feels like a specific hack to get wpa_supplicant and mac80211 play together, rather than an API to do this generically. What I've been thinking and talking about for quite a long time is to make cfg80211 more agnostic as to where the MLME lives, and let nl80211 still be able to configure the interface when a userspace MLME is active. That way, all tools could continue to work regardless of where the MLME is. That would, of course, require to finally put wext into cfg80211, make the wext code capable of sending nl80211 messages to the userspace MLME and also have nl80211 divert messages to the MLME back to the userspace MLME unless they are coming from the userspace MLME. Ultimately, that would allow the userspace MLME to have much more control while users are still able to use the regular tools for many things. Unless, of course, those guys really put d-bus into the kernel and we can just have the MLME service for a specific interface bound anywhere... johannes
Attachment:
signature.asc
Description: This is a digitally signed message part