25 lines
885 B
Markdown
25 lines
885 B
Markdown
|
Ideas on TCP unit testing:
|
||
|
|
<https://www.snellman.net/blog/archive/2015-07-09-unit-testing-a-tcp-stack/>
|
||
|
|
|
||
|
|
Check behaviour around TCP zero-window probing. Is the correct
|
||
|
|
behaviour already a consequence of the way `send-outbound` works?
|
||
|
|
|
||
|
|
Do something smarter with TCP timers and RTT estimation than the
|
||
|
|
nothing that's already being done.
|
||
|
|
|
||
|
|
TCP options negotiation.
|
||
|
|
- SACK
|
||
|
|
- Window scaling
|
||
|
|
|
||
|
|
Check that we handle the situations in figs. 9, 10, 11, pp.33- of RFC 793.
|
||
|
|
|
||
|
|
Bugs:
|
||
|
|
- RST kills a connection even if its sequence number is bogus. Check
|
||
|
|
to make sure it's in the window. (See
|
||
|
|
http://static.googleusercontent.com/media/research.google.com/en//pubs/archive/41848.pdf
|
||
|
|
and RFC 5961)
|
||
|
|
|
||
|
|
Conform better to the rules for reset generation and processing
|
||
|
|
from pp.36- of RFC 793. In particular, do not blindly accept RSTs
|
||
|
|
without checking sequence numbers against windows etc.
|