On Wed, Aug 02, 2006 at 09:20:54PM -0400, Thor Lancelot Simon wrote:
> On Wed, Aug 02, 2006 at 01:55:12PM -0700, Darrin Chandler wrote:
> > On Wed, Aug 02, 2006 at 04:28:41PM -0400, Thor Lancelot Simon wrote:
> > > 
> > > How do you know the data were not corrupted after the TCP checksum was
> > > performed?
> > 
> > If that were the case then it would have been dropped and retransmitted.
> > That's the whole reason for a TCP checksum.
> 
> Uh, no.  What you wrote above _would_ be true _if_ I'd written "after the
> TCP checksum was performed by the source host" -- but I didn't.
> 
> Data corruption after the TCP checksum is validated on receive happens all
> the time; there is a whole family of rather expensive gigabit Ethernet
> controllers in which about every 3rd member has bugs that cause it to
> corrupt data _after_ checking the TCP checksum!

Seems like communication is harder than I thought ;) I get your meaning
now.

-- 
Darrin Chandler            |  Phoenix BSD Users Group
dwchandler@stilyagin.com   |  http://bsd.phoenix.az.us/
http://www.stilyagin.com/  |