On Mon, 2008-10-20 at 15:37 +0530, Sujith wrote: > Johannes Berg wrote: > > On Mon, 2008-10-20 at 02:14 -0700, Luis R. Rodriguez wrote: > > > > > Indeed I agree with this. I think we can do this in three steps: > > > > > > 1. resolve aggregation for now for ath9k > > > > That's what Sujith's patch does, basically. It just leaves the Intel > > hooks in place. > > > > > 2. resolve aggregation for amdpu_queue case. My suggestion here is to > > > use skb_buf_head (after Jouni's suggestion for our driver in fact), > > > and do not make use of a qdisc at all for this > > > > I agree on both, but this is more of an implementation detail than an > > API description, which we really want to have first. > > > Would be nice if an interim fix is accepted temporarily, > until the architectural details of proper mac80211 aggregation > support for the 2 classes of HW are defined, and implemented. Yeah I think that if you clean up this patch a bit wrt. that flag and double-check it won't cause problems on hardware that cannot do aggregation at all I'd be ok with it. Like I said, a later patch can remove all the if (ampdu_queues) stuff johannes
Attachment:
signature.asc
Description: This is a digitally signed message part