Re: [xmpp] IQ Handling vulnerabilities

[Alexander Holler <[e-mail]>]

Am 10.02.2014 20:25, schrieb Thijs Alkemade:
>
> On 10 feb. 2014, at 18:15, Joe Hildebrand (jhildebr) <[e-mail]> wrote:
>
>> On 2/7/14 8:13 AM, "Joe Hildebrand (jhildebr)" <[e-mail]> wrote:
>>
>>> I have a couple of ludicrous s2s attacks on mind, but more important I
>>> think is doing what
>>>
>>> Mobile/terse. DYAC.
>>
>> Yes, well that was even more terse than normal.  I was going to say
>> something about generating less-guessable IDs that don't eat up much
>> entropy.  For example:
>>
>> start = sha1(crytpo_rand())
>> start+1 = sha1(start)
>>
>> would probably do nicely.
>
> Uhm. Maybe this email is missing a line again, but if you use those values as
> 'id's directly, they will not be unpredictable at all, as anyone who receives
> an <iq/> can generate the rest of the chain.
>
> However, if you make sure the 'id' values are only half the hashes, it should
> be unpredictable unless an attacker is willing to spend an insane amount of
> work.
>
> So:
>
> start = sha1(crytpo_rand())
> start+1 = sha1(start)
> ...
>
> id1 = start[0:10]
> id2 = start+1[0:10]
> …

Hmm, in all these mails it was never be mentioned that IDs still have to 
be unique over some time for one session. I'm not sure if such is given 
with the above constructs. It might be very unlikely that the same ID 
will appear twice, but someone has to take a deeper look at it when 
using such constructs like above. Of course, in reality the window in 
time IDs must be unique is rather small, but ...

Regards,

Alexander Holler