On Wed, 2011-02-23 at 10:40 -0800, Adrian Chadd wrote: > The trouble is, I don't know if this macbook pro NIC (some broadcom > 11n thing) is handing over garbage frames or not. I know it sometimes > shows frames from ath9k with incorrect FCS, but that could be in-air > garbage. So it's entirely possible there -was- a frame in there, but > it was garbled to the point where the NIC couldn't decode it.
Yeah.. That was kind of the reason I wanted to see a capture trace, so that I could look at the timing of the frames. > I've also got a wireless capture file showing this exchange (which > feels a lot more "validly" broken) : > > > Ath9k -> STA : QoS Data (non-aggregate) > STA->AP : ACK > Ath9k -> STA : same frame > STA->AP : ACK > .. > .. > .. > Ath9k -> STA: QoS data (same frame, again) > STA-> AP : ACK > You can find that at > http://people.freebsd.org/~adrian/pcap4-sn-512-retrans.pcap Which device is the ath9k device? The one with MAC address ending a2:2a? There were some innovative uses of sequence number 512 in some frames, but the payload was actually different even if the seq# did not change. I don't think that those frames would be duplicated by ath9k (they could be duplicated by mac80211). Which kernel/driver/mac80211 was used here? > I also see the FreeBSD STA (AR9280) send multiple CTS-to-self before > not sending an actual data frame. Are you sure the TX descriptor for the frame that triggered this is set correctly? > Is there any way to stick an ath9k hostap interface in actual monitor > mode, and get =all= the frames to show up on mon.wlan0 ? (So I can > compare what ath9k thinks its sending/receiving, to what FreeBSD STA > thinks its sending/receiving, to what's going on in the air) ? Or > should I just hack the code to enable all the RX filter bits? I don't think you would want all the frames on mon.wlan0, but adding another monitor interface without using cooked mode could do more or less this. Not that all frames would be seen since TX frames generated in hardware (like ack) would not be indicated. - Jouni _______________________________________________ ath9k-devel mailing list ath9k-devel@lists.ath9k.org https://lists.ath9k.org/mailman/listinfo/ath9k-devel
