> On 16 Jan 2016, at 7:17 AM, HU, Jun (Jun) <[email protected]> wrote:
> 
> 
> 
>> -----Original Message-----
>> From: IPsec [mailto:[email protected]] On Behalf Of EXT Yoav Nir
>> Sent: Friday, January 15, 2016 8:15 PM
>> To: Valery Smyslov
>> Cc: [email protected]; Paul Wouters; Scott Fluhrer (sfluhrer)
>> Subject: Re: [IPsec] SLOTH & IKEv2
>> 
>> 
>>> On 15 Jan 2016, at 10:32 PM, Valery Smyslov <[email protected]> wrote:
>>> 
>>> 
>>>>> What about the responder - he doesn't see any cookie in this attack
>>>>> - the attacker sends the crafted cookie only to the initiator and
>>>>> sends a crafted IKE_SA_INIT message w/o cookie to the responder (as
>> far as I understand the attack).
>>>> 
>>>> There is a cookie. See Figure 12 in Paul’s blog post:
>>>> https://securityblog.redhat.com/2016/01/13/the-sloth-attack-and-ikeip
>>>> sec/
>>> 
>>> Ah, you are right. I missed that in a quick read.
>>> 
>>> After second read it seems to me that there is one more  obstacle to
>> that attack in real world.
>>> It seems that attacker appends original initiator's SAi, KEi, Ni
>>> payloads to its message sent to responder (as info`). So, this message
>>> would contain two SA payloads, two KE payloads etc. I believe the
>> responder must return INVALID_SYNTAX in this case.
>> 
>> IIUC there are no two SA payloads and two KE payloads. All of those are
>> part of the “cookie” sent to the real Responder. That is why it’s so
>> large.
> 
> [HJ] according to this 
> figure(https://securityblog.redhat.com/wp-content/uploads/2016/01/sloth-ike-2.png):
> The IKE_INIT request(m1') send to real responder contain infoi' at the end, 
> which equals=SAi|g^x|Ni|infoi, so the actual 
> m1'=HDR|C2|SAi'|g^x'|ni|SAi|g^x|ni|infoi; thus two SA, tw KE, two Ni 
> payloads; C2 is the cookie payload in m1', it doesn't contain any payload. 
> while the cookie payload in m1(IKE_INIT request from release initiator) does 
> contain C1|SAi'|g^x’|ni

OK, but if those extra payloads are disguised as some notification (there is no 
payload actually called “info”), then responders do tend to ignore 
notifications they don’t recognize. 

Yoav

_______________________________________________
IPsec mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/ipsec

Reply via email to