> :) Well, the first problem with libpcap 0.7 versus 0.6 has surfaced
> for me, on Linux 2.0.39 sniffing ppp0 will make tcpdump crash (+core
> sometimes), giving a notice that it couldn't set SIOC*-flags back to
> their original value...

Do you have a stack trace from the crash?

> After it has crashed once (note: it sets promisc *and* multicast on a
> ppp device !),

If you ask for promiscuous mode on a network interface, libpcap will set
promiscuous mode regardless of the type of the device.

> libpcap will fail telling that it doesn't support the ppp-device...

What message does it print?

> Libpcap 0.6 works without a hitch, in combination with tcpdump 3.6 ...
> Although I still see some weirdness when sniffing 'ippp'-devices
> (syncppp devices created by ISDN4Linux), especially with fragments...

SOCK_PACKET sockets, and PF_PACKET/SOCK_RAW sockets, don't work very
well with PPP devices in any version of Linux; the ways in which they
don't work very well differ between PPP devices.

PF_PACKET/SOCK_DGRAM sockets hide those problems; unfortunately,
PF_PACKET sockets require 2.2 or later kernels.

I'm not sure there's much libpcap can do about the problem; if *ALL* PPP
drivers on Linux, *and* the drivers on top of which they run (serial
port, synchronous device, ISDN, etc.) were to arrange that, for incoming
*and* outgoing packets, the "mac.raw" pointer in the socket buffer
pointed to the beginning of the PPP header, things could be made to
work, but I suspect that might be difficult, especially for ISDN.
-
This is the TCPDUMP workers list. It is archived at
http://www.tcpdump.org/lists/workers/index.html
To unsubscribe use mailto:[EMAIL PROTECTED]?body=unsubscribe

Reply via email to