On Mon, 29 Mar 2004, David S. Miller wrote:

On Mon, 29 Mar 2004 22:54:14 +0530 (IST)
Nagendra Singh Tomar [off-list ref] wrote:

quoted

	Thats right. But what about the other cases of retransmission 
failures for which we are having a negative return (-ENOMEM, -EAGAIN, 
-EHOSTUNREACH etc). Even for these cases its not a good idea to 
artificially increment tp->retransmits, lest in some extreme case we might 
timeout a connection without a single packet going on the wire.

That's just like the packet getting dropped at the next hop,
and not the case this branch of code intends to deal with.

I understand your point, but we should try our best to retransmit uptill 
the "max retransmission count". Packets can be dropped at any hop, 
but thats excatly why TCP doess a large number of retransmissions, 
before giving up. Whats wrong in having the check as 

if (tcp_retransmit_skb(sk, skb_peek(&sk->write_queue)) != 0)

so that we take care of both the cases. Does it have any bad effects ?



Thanx,
tomar


-- You have moved the mouse. Windows must be restarted for the 
   changes to take effect.