Re: DLT value for IP over IB (Infiniband)

[Darren Reed <darren.reed () oracle com>]

Guy Harris wrote:
> On Jul 14, 2011, at 5:23 AM, Darren Reed wrote:
>
>   
> > Some more follow up on this...
> >
> > Looks are deceiving - there is no RFC 4391/4392 header being prepended to the IP packet:
> > /*
> > * In order to transmit the datagram to correct destination, an extra
> > * header including destination address is required. IB does not provide an
> > * interface for sending a link layer header directly to the IB link and the
> > * link layer header received from the IB link is missing information that
> > * GLDv3 requires. So mac_ib plugin defines a "soft" header as below.
> > */
> > (From a header file on Solaris)
> > The above probably explains this output on Linux:
> >
> > # tcpdump -i ib0 -vv -e -c 5
> > tcpdump: WARNING: arptype 32 not supported by libpcap - falling back to cooked socket
> >     
>
> What explains that part of the output is that libpcap doesn't have any code to map ARPHRD_INFINIBAND to any DLT_ value, because 
> nobody's contributed any such code.  Perhaps that's because there's no DLT_ value that corresponds to whatever the link-layer 
> header looks like on ARPHRD_INFINIBAND devices on Linux; it's not because of anything specific to Solaris.
>
>   
> > ... and which ultimately means there is not likely to ever be a real link layer header presented for outbound packets. 
> > For inbound, it would seem that it is implementation dependent.
> >     
>
> If "implementation-dependent" means "it's different on Solaris and Linux", the solution is simple - different LINKTYPE_ values for Solaris 
> and Linux; there's already precedent for LINKTYPE_ values for particular OS's flavor of link-layer headers: there are both LINKTYPE_ARCNET_BSD and 
> LINKTYPE_ARCNET_LINUX, there's LINKTYPE_APPLE_IP_OVER_IEEE1394 for Apple's headers for IP-over-FireWire, there are a bunch of other LINKTYPE_ values with 
> LINUX in the name, etc.-
>   

In that case, I'd be willing to go with LINKTYPE_IPOIB_SOLARIS and 
LINKTYPE_IPOIB_LINUX, although I can't contribute anything for the 
latter. With that, would you see keeping DLT_IPOIB as a Solaris private 
thing or creating DLT_
IPOIB_SOLARIS and DLT_IPOIB_LINUX or just DLT_IPOIB and map it to the 
specific LINKTYPE_IPOIB_ depending on where tcpdump is running?

Although ARPHRD_INFINIBAND is a correct name, LINKTYPE_INFINIBAND for 
the packet data that I've seen thus far would be inaccurate.

Darren

-
This is the tcpdump-workers list.
Visit https://cod.sandelman.ca/ to unsubscribe.