Ethereal-users: Re: [Ethereal-users] Bad TCP checksums from local host?

Note: This archive is from the project's previous web site, ethereal.com. This list is no longer active.

Date Index Thread Index Other Months All Mailing Lists
Date Prev Date Next Thread Prev Thread Next
From: "Martin Regner" <martin.regner@xxxxxxxxx>
Date: Sun, 15 Aug 2004 18:58:10 +0200
Mike Weber:
> I'm running ethereal 0.10.6 with winpcap 3.1 beta 3 under winXP with all 
> the patches windows update wanted to install except service pack 2. I'm 
> running an Athlon64 3400+ on an NVidia nForce3 250Gb - based motherboard.
> 
> While going through an analysis I noticed that probably ~90% of the 
> segments generated from my machine are reported as having bad TCP 
> checksums. There doesn't seem to be a problem transferring data; I've 
> transferred gigabytes of data over this link over the last few days and 
> haven't had trouble on a scale that correlates with the number of bad 
> checksums I'm seeing. (I have had a problem where certain email message 
> bodies never get ACK'd, but I don't know if that's a checksum problem or 
> something else. It is the reason I broke out ethereal, though.)
> 
> Is this something to worry about? Is it a bug in winpcap or ethereal? Or is 
> the checksumming code in my NVidia NForce3 chipset (does tcp/ip offloading) 
> broken?
> 
> Thanks for any information pointers or advice!
> 

Don't worry ...

It is most likely due to tcp checksum calculation offloading

http://www.ethereal.com/faq.html#q5.14

The packets with correct checksum is probably packets that are shorter than a certain length.