On Sat, 2007-07-07 at 16:09 -0400, Jon Smirl wrote: > I've fixed the promiscuous problem but I don't know what the cause > was. I noticed that when I captured a lot of packets with libpcap, it > and the adapter were reporting the same number of packets seen. That > implied that nothing was missing which made me suspect wireshark. > Deleting the .wireshark directory made all of the missing packets > appear. So I must have done something to wireshark to make it filter > out packets that weren't directed to me in a way that it remembered > the filter across restarts, but I don't know what I did to get it in > that mode. tcpdump was getting the packets, but I was writing them to > a file and using wireshark to look at them. Weird. I'll try to remember that for next time :) > Something is still wrong with the radio tap headers. Using the bcm4318 > all of my packets say they were captured on channel 3, when they were > captured from channel 1. This may be a problem with wireshark. The > beacons show channel 1 in tcpdump, and channel 3 in wireshark. That can actually happen; getting packets from channel 1 on channel 3 isn't uncommon. Double-check with iwconfig. > Have the structures for radio tap been changed? That could explain > things if wireshark is compiled with the old version. I'm using > 0.99.4-6 from ubuntu. Well, wireshark does in fact have missing definitions for some radiotap things and I think one or two things are actually wrong as well, but nothing that would have this effect. > So now I can get back to my original goal of get zd1211/rt2x00 to run > on an ARM with mac80211 and OLSR. I can say that the zd1211 and rt2x00 > mac80211 drivers appear to be working much better now than they were > six months ago. :) johannes
Attachment:
signature.asc
Description: This is a digitally signed message part
