Oolan Zimmer wrote:
In Gcom's T1/E1 driver, a DL_ATTACH_REQ chooses the physical port and a DL_BIND_REQ chooses the logical channel on that port. A logical channel is a collection of one or more timeslots, and its associated SAP is configurable (usually just starts at 1 for the configurations we distribute with the driver).
So, sometimes SAP values have a real meaning associated with them.
Yes - but the question is whether there's any significance to it if DL_PROMISC_SAP is also requested, which is what libpcap does (in HP-UX, it does so only if DL_PROMISC_PHYS is not requested), or if DL_PROMISC_PHYS is also requested (which is done if the "promisc" argument is non-zero).
I.e., libpcap doesn't, at least for LANs, want to request a particular SAP, it wants *all* traffic regardless of the SAP.
If libpcap were to support capturing traffic on logical channels on T1/E1 devices with Gcom's driver, the name of the interface on which the capture was being done might specify a physical port and logical channel. There would presumably have to be some way - whether based on the name of the device, or on something else that could be done before a DL_ATTACH_REQ - to specify whether to do the attach and bind the way it's done for LAN devices or the way it would be done for T1/E1 devices.
-
This is the tcpdump-workers list.
Visit https://lists.sandelman.ca/ to unsubscribe.