> The problem is that redhat included a patch to libpcap that allows you to > sniff on all interfaces at the same time. Well the also modified the pcap > output to now contain an interface field for each packet. This is the > source of the problem. How would that cause the capture to hang? I'd expect our capture code to just ignore that extra information (given that we call "pcap_open_live()" rather than "pcap_open_live_new()", we *shouldn't* The hang sounds more like the lack of the patch to cause the "to_ms" argument to "pcap_open_live()" actually do something; at least in the copy I grabbed of the RH 6.1 source, the "to_ms" argument doesn't do anything in either "pcap_open_live()" or "pcap_open_live_new()". > We could have ethereal check for that type of pcap > output and automajically switch, but to get it working, I just installed > to old tcpdump source rpm, applied the linux libpcap patch, Applying the Linux libpcap patch may have been what fixed the problem for you.
Powered by MHonArc 2.6.10