9 Feb
2017
9 Feb
'17
3:19 p.m.
On 9 Feb 2017, at 4:02 , Nick Downing wrote:
As to moving to a V7 kernel and then adding TCP/IP
I'm not sure if
this is adviseable, as I was saying earlier I think it might be best
to keep that functionality outboard from the kernel.
That approach was taken in various early implementations, with varying
degrees of success.
The best one seems to have been the 3Com stack, which puts IP in the
kernel and TCP in a daemon. By the way, this implementation is also
where SLIP seems to have originated.
http://bitsavers.informatik.uni-stuttgart.de/pdf/3Com/3Com_UNET_Nov80.pdf
and (2) does it have to respond to ping or incoming
connections at any time.
I'm not sure economizing on ICMP support is the best approach. Perhaps
economizing on fragmentation and and window management is better. For
example, the 1979 Wingfield implementation discards out of order packets
and simply waits for retransmission of the expected packet; this
simplifies the code and greatly reduces the need for buffer space.
Arguably, the handling of incoming connections does not require much code
or data.
---
A TCP connection spends most of its time in the 'established' state. I
wonder if just putting the code for this state in the kernel and handling
only the state changes and other states in a daemon is perhaps the best
split on constrained hardware. I'm not aware of any historical
implementation having that design, so perhaps it is flawed thinking.
Paul