lwip-users
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [lwip-users] Lwip 1.3.1 TCP Retransmission problem


From: Kieran Mansley
Subject: Re: [lwip-users] Lwip 1.3.1 TCP Retransmission problem
Date: Thu, 03 Mar 2011 10:37:45 +0000

On Tue, 2011-03-01 at 13:31 -0800, theflasher1986 wrote:
> 
> (if i say send i mean the prefix + message send together)
> The problem is, between the eight and ninth send from the server to
> client i
> get a lot of TCP-retransmissions.It takes about 2 secondes before the
> ninth
> send can be done. The same problem occurs after the 17. send or the
> 26.
> send. (8-9, 17-18, 26-27, 35-36 ...) It is periodic.  

That does look odd.  You're getting lots of very similar packets (they
only seem to differ in the IP ID field) being sent by the stack in a
very short space of time.  I wonder if there's a problem with the way
TCP timers are being called in your port of lwIP?  Can you give any
details about how that is being done?

1.3.1 is now a few years old and many bug fixes out of date.  It will be
much easier to offer support if you're able to upgrade to something more
recent, although I appreciate that you're probably working with a port
someone else has done and so not in a position to do that.

> Below i have posted my source code, so maybe sobody can discover a bug
> or
> anything.

It looks OK to me, although I don't understand how you exit the while()
loop for doing netconn_write()?  Should there be a break or return in
there after doing the send?  You'll also need to be careful that you
don't free inbuf twice.

Kieran




reply via email to

[Prev in Thread] Current Thread [Next in Thread]