Re: Tcl socket checksums

On 22 May 2007 05:24:34 -0700, derek.philip@xxxxxxx wrote:

On 22 May, 12:15, Mark Janssen <mpc.jans...@xxxxxxxxx> wrote:
On May 22, 12:46 pm, Mark Janssen <mpc.jans...@xxxxxxxxx> wrote:



On May 22, 12:25 pm, derek.phi...@xxxxxxx wrote:

Hi All

I had a reason to observe the communication between two scripts,
client, server, that use tcl sockets over a network.
I used ethereal to capture packets on the server

I was supprised to see that all the packets from the server were
marked as bad, Ethereal indicates that the TCP checksum is incorrect.

I have had a look in the bug list but can't find any mention of this.

I'm not overly concerned, at the moment, as the scripts have been
working for quite some time.

I havn't looked on any other platform to see if it exists elsewhere.

Any one else seen this?
What information would need to be captured to get a better
understanding of the problem?

Setup details:

Ubuntu Linux 6.06 - with current updates
Tcl 8.4.12 - from the Ubuntu repositories.

I have observed this behaviour with Ethereal before (on Sun Hardware),
it might indicate that the checksums on your packets are checked in
hardware and set to bogus values at the point ethereal gets to look at
them.
You could look at the checksums of the sent and received packages and
compare them to see if they are equal. If they are not, this is
probably what's happening.
If your packets are otherwise correct, I wouldn't worry about this too
much though.

Mark

See also:http://www.ethereal.com/faq.html#q11.1

Mark
hi

Thanks for the info, why didn't I checkout the Ethereal page before
hand???

derek

Don't feel too bad, I never checked out the Ethereal FAQ until today,
my explanation of what was happening was based on guesswork and some
tracing.

Mark (Who never looks in the obvious places (like FAQs))
.

Relevant Pages

  • Re: tap interface and locally generated packets
    ... Yes, ifconfig -txcsum fixes the problem, so somewhere packets are not ... getting marked to be summed if the hardware checksum is turned on, ... >> Packet pings from the tap interface works to any ip address, ... >> I have found this is due to tcp checksums not being generated, ...
    (freebsd-net)
  • Re: TCP checksum error
    ... I remember a time when I was trying to track a network problem down and ... I had bad checksums, it worked out the bad checksums was being caused by ... an application injecting packets into the stack, ... error by wireshark, I found if I wiresharked from a different box the ...
    (Debian-User)
  • Re: [PATCH 0/4][RFC] lro: Generic Large Receive Offload for TCP traffic
    ... when modified packets are flushed, else the stack will see bad ... These drivers could set this ip_summed field to CHECKSUM_NONE, and thus leave the checksum check to the stack. ... CHECKSUM_PARTIAL drivers using the skb interface by allowing them ... and saw tcpdump complain about bad checksums. ...
    (Linux-Kernel)
  • RE: Filter, repackaging NET_BUFFERs and offload checksums
    ... are being transmitted out of your NIC without checksums? ... Problem is that in repackaged buffers is missing some info, because packets ... I dont want to switch off offload tasks on NIC, ... only used data space is copied and not data from 0 to ...
    (microsoft.public.development.device.drivers)
  • Re: Damn you, FEDEX! or Nikon D40 lost in Springfield, MO blackhole.
    ... the 2 mp Mavica he had been using with a Nikon D40. ... After shopping around, he got me to order one for him. ... The shipper had it insured, but from what I have read it could take weeks to sort this crap out. ... You may get your insurance from FedEx and a couple weeks later they find it and deliver it. ...
    (alt.photography)