As far as I can tell most implementations with tcp checksum offloading
specify the checksum as 0 in the apckets captured.
Maybe we should change the tcp checksum code to
1, IF checksum checking is enabled and IF the checksum is wrong and IF
the specified checksum value is 0, THEN mark it as Checksum:0x0000
(Checksum Offloaded)
2, since tcp checksum errors are VERY rare and would result in either
a RST or a retransmission anyway making it easy to spot them, we could
disable checksum checking by default.
On 3/4/06, Guy Harris <gharris@xxxxxxxxx> wrote:
> C.J. Proffitt wrote:
>
> > I've been an ethereal user for several years and have not
> > had one once of trouble out of it up until now. The platform is windows
> > server 2003 SBS and the problem is the server (which is also the machine
> > the software is installed on) generates checksum errors according to
> > ethereal version 10.14. Before I upgraded the server from version 10.12,
> > there were no checksum errors. Now after removing 10.14 and reinstalling
> > 10.12 (including winpccap) I still get checksum errors. This is present
> > on another windows 2003 SBS server I administer. Checksum errors seem to
> > prevail in file and printer sharing protocols (port 445, etc.)
>
> Are the checksum errors only on packets being sent by the machine
> running Ethereal?
>
> If so, see:
>
> http://www.ethereal.com/faq.html#q11.1
> _______________________________________________
> Ethereal-users mailing list
> Ethereal-users@xxxxxxxxxxxx
> http://www.ethereal.com/mailman/listinfo/ethereal-users
>