Preloading IP stack arp table

gregory.menke at gsfc.nasa.gov gregory.menke at gsfc.nasa.gov
Thu Oct 13 19:05:32 UTC 2005



They don't take much, RTEMS "introduces" itself by arp when the
interfaces initialize and arp packets are exchanged occasionally in
operation.  

Our application involves the use of a TDMA ethernet driver & rewriting
ethernet frame headers on the fly to help implement redundancy and other
features.

As such, the physical layer seen by the IP stack is "virtualized" and
all phy and IP addresses are known ahead of time anyhow- so arp packets
must be suppressed, and consequently the arp table has to be
preconfigured.

Greg


Bruce Robinson writes:
 > Hi Greg,
 > 
 > I'm working on the development of an embedded controller that will be 
 > tethered to a PC running Windows or Linux. I'm curious, with ethernet do 
 > the ARP's take up a significant amount of time? I assumed they only 
 > occurred on the initial communication.
 > 
 > Bruce
 > 
 > 
 > At 07:24 AM 10/13/2005, you wrote:
 > 
 > >Hi,
 > >
 > >We have an application where we want to avoid all arp traffic- easy
 > >enough to suppress it at the driver, but it is necessary to load static
 > >entries into the arp table.
 > 




More information about the users mailing list