[ On Mon, November 23, 1998 at 13:36:36 (-0800), Paul Goyette wrote: ]
> Subject: Re: This PMTU thread 
>
> On Mon, 23 Nov 1998, Greg A. Woods wrote:
> >
> > So far as I can tell "DF" can't be set directly by an application.
> 
> Well, why not?  There's at least one "application" I know of that _does_
> set the DF bit:  GateD.  If GateD can set the bit, any other application
> could also set it.  (I haven't looked at the source, so I don't know
> _how_ it gets set, but I've seen the results in a tcpdump...)

I should have been more specific since this was taken slightly out of
context:

Since the "DF" bit can't be set on UDP and TCP packets directly by an
application, in any implementation so far as I can tell, and since so
far as I know the only use of the "DF" bit in TCP packets is for PMTU-D,
it should be safe to ignore that bit in an instance which appears to be
caused by failing PMTU-D.

-- 
							Greg A. Woods

+1 416 218-0098      VE3TCP      <gwoods@acm.org>      <robohack!woods>
Planix, Inc. <woods@planix.com>; Secrets of the Weird <woods@weird.com>