> I've compiled and installed tcpdump from tcpdump.org. It is running on
> AIX 4.3.3 maintenance release 08.
>
> The basic functionality is ok, but if I do 'tcpdump host hostname'
> tcpdump reports:
>
> tcpdump: /dev/en0: No such file or directory
>
> AIX does not use special device files in /dev to define these
> interfaces.
No, it appears to use special device files in the "dlpi" subdirectory of
"/dev", instead.
For some reason, a number of people who have sent in fixes to the DLPI
support in libpcap for AIX haven't reported that problem, and haven't
sent in fixes for it, perhaps because they used
tcpdump -i /dev/dlpi/en0
or something such as that, explicitly telling tcpdump to use
"/dev/dlpi/en" rather than "/dev/en". (Yes, without the 0.)
> Does anyone have a list of the known bugs with AIX?
I don't know of all the problems - I don't have an AIX box handy on
which to try stuff.
The libpcap bugs I think exist in the current CVS version and 0.7 beta
version are
1) the AIX BPF is very annoyingly non-standard (BIOCGDLT returns
SNMP ifType values rather than DLT_ values, and the time
stamps on packets are in seconds and nanoseconds rather than
seconds and microseconds) so you may get bizarre results if
you make libpcap use BPF (it'll think Ethernet devices are
Token Ring devices, and the time stamps will be bogus);
2) the DLPI code needs an explicit pathname for interfaces
(unless there are versions of AIX where the DLPI device *is*
"/dev/{en,tr,fi,...}" rather than "/dev/dlpi/{en,tr.fi,...}".
There may be others. There *are* others in 0.6.2, e.g. capturing on
Token Ring devices might not work.
(Also, our tcpdump probably won't work very well with the libpcap that
comes with AIX 5.x, as that libpcap cheerfully exposes to libpcap
programs the AIX BPF's non-standard nature - but that's a separate
issue.)
-
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
