Re: Unexpected primitive ack DL_UNITDATA_IND

[Rick Jones <rick.jones2 () hp com>]

Guy Harris wrote:
> On Jun 9, 2004, at 1:27 PM, Rick Jones wrote:
>
> > Does it always happen or just sometimes?  A DL_UNITDATA_IND is 
> > basically saying "Hi there, here is a packet" in DLPI speak.  It looks 
> > like the stream is sending one of those up to the application when 
> > libpcap isn't expecting it.
>
>
> In fact, it's sending one up when, presumably, libpcap is waiting for a 
> DL_INFO_ACK in response to a DL_INFO_REQ it sent downstream to determine 
> either the DLPI provider style or the link-layer type.
>
> I don't have the DLPI spec handy; currently, pcap-dlpi.c makes two info 
> requests, one just after opening the device and one after binding to the 
> PPA (but *before* setting DLIOCRAW or pushing "bufmod") - can a DLPI 
> provider cough up packets at that point?

On the surface I wouldn't think so - simply attaching to a PPA I don't think 
means traffic could arrive - however, if one has attached, and then binds to a 
SAP, then traffic could indeed start arriving.  (Il-informed guesstimation, and 
hopefully I'm recalling the correct distinction between attach and bind)

Although I might wonder about something like a link-down or other housekeeping 
message that the stream might cough-up - although I would have expected those to 
be a PROTO or PCPROTO (if I'm recalling the mnemonics correctly).

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