kern/110698: nat rule of pf without "on" clause causes invalid packed chksum

[Daniel Hartmeier]

The following reply was made to PR kern/110698; it has been noted by GNATS.

From: Daniel Hartmeier <daniel at benzedrine.cx>
To: "Vladimir V. Kalashnikov" <hw at ksue.edu.ua>
Cc: FreeBSD-gnats-submit at freebsd.org
Subject: Re: kern/110698: nat rule of pf without "on" clause causes invalid packed chksum
Date: Fri, 24 Aug 2007 10:45:44 +0200

 On Fri, Mar 23, 2007 at 10:13:48AM +0200, Vladimir V. Kalashnikov wrote:
 
 > ###### S, cksum 0xee62 (incorrect (-> 0xfcc5), 3464239052:3464239052(0)
 > ###### here the effect ^^^^^^^^^^^^^^^^^^^^^^^
 
 Can you make sure that packets with invalid checksums really go out on
 the wire, by tcpdumping from another host on the network?
 
 When tcpdumping on the endpoint, it's possible to see misleading
 'incorrect' checksums due to hardware checksumming, i.e. the checksums
 get correctly set by the NIC and the packets on the wire have correct
 checksums.
 
 Daniel