TCP conntrack queries

TCP conntrack queries

[Kiran Kumar]

Hi,
  As I went through ip_conntrack_proto_tcp.c, I had
these couple of queries.

- Why is there a distinction b/w states NONE and
CLOSED?

- The array of timeouts has been replaced (as I get
from the diff b/w my RH9 kernel sources and 2.6
sources) by an array of pointers to variables. Why was
this done?

=====
Regards,
Kiran Kumar Immidi

__________________________________
Do you Yahoo!?
Yahoo! Finance Tax Center - File online. File on time.
http://taxes.yahoo.com/filing.html

Re: TCP conntrack queries

[Pablo Neira]

Hi!

Kiran Kumar wrote:

>Hi,
>  As I went through ip_conntrack_proto_tcp.c, I had
>these couple of queries.
>
>- Why is there a distinction b/w states NONE and
>CLOSED?
>  
>
None is the initial state where every connection starts when a conntrack 
is created and Close is when closed connection (I know, quite obvious!).

Anyway have a look at tcp-window-tracking patch because Jozsef spent its 
time well-documenting all the possible transitions to let us understand :).

>- The array of timeouts has been replaced (as I get
>from the diff b/w my RH9 kernel sources and 2.6
>sources) by an array of pointers to variables. Why was
>this done?
>  
>
because now you can set your own timeouts via sysctl.

regards,
Pablo