Re: [lwip-users] Fast retransmit

[Leon Woestenberg]

Hello Sam, Kieran,

Sam Jansen wrote:

> Yep, it makes it segfault. I tracked down the problem to a small 
> logical error introduced in the latest updates. Attached is a diff to 
> fix it (just a one-liner!). Once I fixed this lwip works as expected 
> without the previous funky retransmit behaviour.
>
> +  pcb->unsent = pcb->unacked;
>  
Committed in HEAD. BTW, did you test against HEAD CVS, or 0.7.2 with 
Kieran fix and your one-liner fix?

Thanks to both of you guys for the fixes!

BTW, I assume I can merge this back into 0.7.x as a bug fix soon, unless 
I am told otherwise, as the TCP code
is similar (minus another small seemingly irrelevant fix in HEAD from 
Kieran).

Regards,

Leon.