Re: "dnet: Failed to open device" errors on AIX, as root

[David Fifield <david () bamsoftware com>]

On Sat, Dec 08, 2012 at 11:43:43AM -0500, Ben Lentz wrote:
> > It might help if you can try to compile and run this example code from
> > IBM. If its bind works, but that of eth-ndd.c doesn't, then we can
> > identify what the important differences are.
> >
> > http://pic.dhe.ibm.com/infocenter/aix/v6r1/topic/com.ibm.aix.progcomm/doc/progcomc/skt_sndother_ex.htm
>
> I was able to compile the "Sending Packets Over Ethernet Example
> Program" and I was able to get it to run (I'm not totally clear on
> what it's supposed to do) against both my local MAC address and the
> MAC address of another system on the same network (10.0.17.0/24) so
> long as I specified the ethernet type as 0x0, and not 0x800. As with
> the problem that Peter reported, if 0x800 is used on AIX, it gives the
> bind EEXIST error.
>
> $ sudo ./sample en2 7e:ef:bb:08:f1:0c 0x800 10 10
> src addr = 7EEFBB08F10C
> dst addr = 7EEFBB08F10C
> Ethertype: 800
> Count: 10
> Size: 60
> bind: File exists
>
> $ sudo ./sample en2 7e:ef:bb:08:f1:0c 0x0 10 10
> src addr = 7EEFBB08F10C
> dst addr = 7EEFBB08F10C
> Ethertype: 0
> Count: 10
> Size: 60
> $ sudo ./sample en2 9e:8b:f7:e6:00:0b 0x0 10 10
> src addr = 7EEFBB08F10C
> dst addr = 9E8BF7E6000B
> Ethertype: 0
> Count: 10
> Size: 60
>
> If I update libdnet-stripped/src/eth-ndd.c as follows:
>
> $ /opt/local/diffutils/bin/diff -u libdnet-stripped/src/eth-ndd.c.orig
> libdnet-stripped/src/eth-ndd.c
> --- libdnet-stripped/src/eth-ndd.c.orig 2012-12-08 11:41:06.000000000 -0500
> +++ libdnet-stripped/src/eth-ndd.c      2012-12-08 11:41:29.000000000 -0500
> @@ -42,9 +42,9 @@
>         sa.sndd_8022_family = AF_NDD;
>          sa.sndd_8022_len = sizeof(sa);
>         sa.sndd_8022_filtertype = NS_ETHERTYPE;
> -       sa.sndd_8022_ethertype = ETH_TYPE_IP;
> +       sa.sndd_8022_ethertype = 0;
>         sa.sndd_8022_filterlen = sizeof(struct ns_8022);
> -       strlcpy(sa.sndd_8022_nddname, device, sizeof(sa.sndd_8022_nddname));
> +       strlcpy(e->device, device, sizeof(e->device));
>
>         if (bind(e->fd, (struct sockaddr *)&sa, sizeof(sa)) < 0)
>                 return (eth_close(e));

Did you also try this patch without the second change? I think it's
wrong not to assign sndd_8022_nddname. It is probably also wrong not to
assign e->device, but I don't think that would actually have an effect.

The only other thing I can think of is that maybe the payload needs to
be padded to 60 bytes, as reported here:
        http://seclists.org/nmap-dev/2012/q1/96
You could try setting the --data-length option to test if this is the
case. I think it is unlikely, though, as the earlier report had EMSGSIZE
for this error. I think it's more likely that you need to restore the
assignment to sndd_8022_nddname.

David Fifield
_______________________________________________
Sent through the dev mailing list
http://nmap.org/mailman/listinfo/dev
Archived at http://seclists.org/nmap-dev/