On Tue, 2008-10-21 at 12:44 +0530, Sujith wrote: > Wireless HW without any dedicated queues for aggregation > do not need the ampdu_queues mechanism present right now > in mac80211. Since mac80211 is still incomplete wrt TX MQ > changes, do not allow aggregation sessions for drivers that > set ampdu_queues. > > This is only an interim hack until Intel fixes the requeue issue. Looks ok to me, but I'd feel safer if you introduced a HW capability flag that said "I can handle aggregation", and check it in ieee80211_start_tx_ba_session. Right now, that will refuse to work when the driver announces no ampdu queues, but you're changing that, so it would do something even if the driver isn't prepared to handle that. Not that it would ever be called, but it seems safer to actually encode that capability somehow rather than relying on the RC algorithm not doing it. > Remove is_part_ampdu from struct sk_buff, and set > IEEE80211_TX_CTL_AMPDU > in __ieee80211_tx_prepare() - which I think is a better place than > ieee80211_tx_h_sequence(). agreed. johannes
Attachment:
signature.asc
Description: This is a digitally signed message part
