Mateusz Viste wrote:
> I believe you wanted to say "frame" instead of packet and "64" instead of 
> "60"? As far as I know, an IP packet can transport no payload, which gives a 
> 20-bytes packet in result. :-P
> The limitation is about the minimum size of the frame, which (on ethernet) is 
> of 64 bytes, to allow proper collision detection... (I hoped that such 
> low-layer stuff would be handled by the packet driver... seems that it is 
> not...)
> All in all, mTCP sounds like a great project. It's really nice to see that 
> someone still cares to write some good stuff for DOS! :-)
> Best regards,
> Mateusz Viste


Yes, you caught me being sloppy with the terminology.  You are correct, 
it was the Ethernet frames that I was working with, not IP.  
Specifically causing the problems were ARP packets, which are far 
smaller than the minimum.  The symptoms were that ARP would never get a 
response from the outside world because the packet driver was 'eating' 
the non-compliant frame/packet.

I used '60' instead of '64' because from the user/stack perspective the 
four bytes of CRC are not part of the payload.  So if one assumes and 
uses a minimum payload size of 60 and the driver/hardware add the CRC, 
then you get to 64 bytes.


Start uncovering the many advantages of virtual appliances
and start using them to simplify application deployment and
accelerate your shift to cloud computing.
Freedos-user mailing list

Reply via email to