On Thu, 2008-10-23 at 14:27 +0530, Sujith wrote: > > Yeah, this is something we definitely need to work out. We haven't even > > managed to decide yet how aggregated packets are given to the driver :) > > I think for atheros and bcom a model where mac80211 decides pretty much > > everything and hands the driver a list of skbs to aggregate would work > > best, but then I think that wouldn't work with your hw at all. > > > > Retransmission of unacked frames is done within the driver, in ath9k. > In case of retry exhaustion (failed transmission), we do what Intel does, > i.e, request for a BAR to be sent out. We have this "ampdu_ack_map" parameter in the TX status which could be set by the driver. But that assumes we only want a single TX status for all A-MPDU frames, do we really? Or do we want to split them up and handle it all in mac80211? johannes
Attachment:
signature.asc
Description: This is a digitally signed message part
