Re: [tcpm] Flaw in RFC793 (Fwd: New Version Notification for draft-gont-tcpm-tcp-seq-validation-03.txt)

[Wesley Eddy <[e-mail]>]

On 3/6/2018 4:53 PM, Yoshifumi Nishida wrote:
> Hello,
>
> On Tue, Mar 6, 2018 at 11:05 AM, Eggert, Lars <[e-mail] 
> <mailto:[e-mail]>> wrote:
>
>     Hi,
>
>     isn't this something that could simply go into 793bis, after there
>     is consensus that something needs to be fixed?
>
>
> I think the observation of the bug can go into 793bis (which has been 
> done in the current version).
> However, I think the solution for it needs to be reviewed and discussed.
>


FYI - there is a paragraph on the issue in Appendix A of the current 
revision, which was added after discussion around the Chicago meeting 
(IIRC).  I think if no solution is blessed by the working group, that 
this warning about a "known bug" would be the best we can do in 793bis, 
but is certainly far less desirable than having an actual solution in 
the body of the document.  To work that solution out, discuss the 
options, etc., I'm in favor of advancing it via the draft Fernando and 
David have put together.  They already briefly discuss the Linux 
approach, for instance.


> It will be really great if some implementers speak up if the solution 
> in the draft looks good or they take different approach to address the 
> issue.

Agreed; feedback is useful, though I do think this bug is evident and we 
at least can see what has evidently been done about it already in open 
source stacks, as examples.