[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE : [lwip-users] TCP retransmissions and some general questions
From: |
Frédéric BERNON |
Subject: |
RE : [lwip-users] TCP retransmissions and some general questions |
Date: |
Wed, 20 Jun 2007 13:07:43 +0200 |
(sorry if you receive twice this post...)
http://lists.gnu.org/archive/html/lwip-users/2006-11/msg00007.html // tune some
TCP options (read all the thread)
http://lists.gnu.org/archive/html/lwip-users/2007-04/msg00056.html // Reduce
footprint (read all the thread)
http://savannah.nongnu.org/task/?1549 // See the Simon Goldschmidt's big job to
document lwIP functions (I have attach a RTF doxygen doc.)
http://savannah.nongnu.org/task/?6683 // Some tips about lwIP thread safety
http://savannah.nongnu.org/patch/index.php?5960 // Some tips about lwIP thread
safety (more recent)
Perhaps it will be good to put a "news" on Savannah with these links waiting to
document the next release?
====================================
Frédéric BERNON
HYMATOM SA
Chef de projet informatique
Microsoft Certified Professional
Tél. : +33 (0)4-67-87-61-10
Fax. : +33 (0)4-67-70-85-44
Email : address@hidden
Web Site : http://www.hymatom.fr
====================================
P Avant d'imprimer, penser à l'environnement
-----Message d'origine-----
De : address@hidden [mailto:address@hidden De la part de Kieran Mansley Envoyé
: mercredi 20 juin 2007 12:15 À : Mailing list for lwIP users Objet : Re:
[lwip-users] TCP retransmissions and some general questions
On Wed, 2007-06-20 at 11:52 +0200, Matthias Weisser wrote:
> Hi
>
> Today I found some time to do some deeper testing of my lwIP port. I
> used wireshark and tested around a bit.
>
> The first thing I saw that there are TCP retransmissions from time to
> time. These retransmissions happens about 5-10ms after the original
> packet was transmitted but not for every packet. That seems a little
> to early to me.
Were there any ACKs from the other side between the sending of the packet and
its retransmission? If so, it could be fast retransmission and not a problem.
If not, it does seem a bit early. Perhaps you could send a wireshark/ethereal
capture file to illustrate the problem?
> Another question is about the TCP window size. Which value makes sense
> together with all the other options in the lwipopts.h? Maybe its worth
> to write a little bit more about these options and how they depend on
> each other.
You're right that this the links between many options are not obvious, and it's
been a frequently asked question/problem on the mailing list. I agree that more
documentation would be good. Any volunteers? There was a useful post on the
mailing list a few months ago documenting some "good" settings for everything,
but I don't have a link to hand at the moment.
As for the window size, apart from making sure it's at least twice the MSS, and
that you have sufficient buffer space to accept a window worth of data in one
go, there's not a lot more to it. Of course, bigger is better in terms of
performance, but will require more memory for buffers.
> Another point is about mailboxes. I implemented them as FIFOs but how
> deep should this FIFOs be? I use 32 entries now but is this enough?
Not sure.
Kieran
_______________________________________________
lwip-users mailing list
address@hidden http://lists.nongnu.org/mailman/listinfo/lwip-users
Frédéric BERNON.vcf
Description: Frédéric BERNON.vcf
lwip-doc.CVS_HEAD.zip
Description: lwip-doc.CVS_HEAD.zip