Kathleen,

> On May 14, 2015, at 8:32 AM, Kathleen Moriarty 
> <[email protected]> wrote:
> 
> 
> 
> On Wed, May 13, 2015 at 11:59 PM, Suresh Krishnan 
> <[email protected] <mailto:[email protected]>> wrote:
> Hi Ron,
> 
> On 05/13/2015 11:39 PM, Ronald Bonica wrote:
> > Kathleen,
> >
> > AFAIK, most IP stacks include code that detects fragmentation overlap 
> > attacks. (Do I have that right?)
> >
> > So, reassembly attacks shouldn't be effective whether reassembly is 
> > performed at the GRE egress or the ultimate destination.
> >
> > If reassembly is performed at the ultimate destination, the two endpoints 
> > might be alerted. However, if reassembly is performed at the GRE ingress, 
> > the endpoints might never be alerted.
> >
> > Should we add a paragraph about this in Section 5 (Security 
> > Considerations). Or is this just another type of DoS attack, which we have 
> > already mentioned?
> 
> I think it might merit a separate mention since the draft is concerned
> with fragmentation. You can use RFC1858 as a reference for IPv4 and
> RFC5722 as a reference for IPv6 for handling of the overlapping fragment
> problem.
> 
> A separate paragraph would be helpful, thanks.  This attack type could lead 
> to a compromise, so the concern (for me at least) is much higher than a DoS.  
> I'm glad it's addressed in code and it would just be good to mention 
> considerations.
> 

I think that overlapping fragment attacks are orthogonal to whether the packet 
carries GRE or something else. If reassembly happens at the tunnel endpoint, we 
might end up with a rightly dropped packet. This seems covered in Section 4 of 
RFC 1858.

What particulars are you thinking about this attack that are specific to GRE?

Thanks,

— Carlos.

> Thank you,
> Kathleen
> 
> 
> Thanks
> Suresh
> 
> 
> 
> 
> --
> 
> Best regards,
> Kathleen
> _______________________________________________
> Int-area mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/int-area

Attachment: signature.asc
Description: Message signed with OpenPGP using GPGMail

_______________________________________________
Int-area mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/int-area

Reply via email to