[ell-i-developers] Re: IP/UDP/CoAP status

  • From: Pekka Nikander <pekka.nikander@xxxxxx>
  • To: ell-i-developers@xxxxxxxxxxxxx
  • Date: Mon, 3 Mar 2014 08:49:22 +0200

UDP and CoAP work now in the emulator, at least the one very simple test case I 
have.  Next I will write a simple ARP.  After that we will need to integrate 
with Ivan, and thereafter start to clean up the code.  At the moment it is 
quite messy.  As before, the messy code is available in the feature-coap-wip 
branch in Runtime.git.


On 2014–03–02, at 22:58 , Pekka Nikander <pekka.nikander@xxxxxx> wrote:

> I implemented pcap support to a mockup test platform, so that when running 
> under the emulator the stack is now able to receive real packets from a file 
> and send packets to standard output, in pcap format.  I also implemented a 
> very simple ICMP echo support.
> With this test platform, I have now been able to debug and fix the quite a 
> few bugs in the Ethernet, IP and ICMP layers.  The code now replies to ping.  
> UDP and CoAP are still to be tested.  ARP and DHCP remain to be written.  
> Hence, I believe we are in good shape for having working CoAP on Tuesday or 
> Wednesday latest.  DHCP most probably quite much later, as it requires timers 
> and other state requiring functionality.
> While I'm not able to produce real binaries yet (due to some missing stuff 
> between ENC28J60 and Ethernet layer), the approximate binary size at the 
> moment is 3.5kB.  Hence, I have all the reason believe that the final working 
> version with a working ARP/IP/ICMP/UDP/CoAP will take less than 4k.  DHCP 
> most probably fits into less than 1k, and therefore the final CoAP-capable 
> runtime is likely to fit into 5k.  Hence, in theory at least we should be 
> able to support the lowest end Cortex-M0 chips with just 2-4k of RAM and 8k 
> of FLASH, e.g. NXP LCP1100XL.  Of course, at the moment we don't have drivers 
> for any NXP peripherals, but if we want to go there that is not such a big 
> deal.  At the STM side we should be able to use STM32F030, which should cost 
> around 30 cents in mass quantities.
> --Pekka

Other related posts: