On 12/12/2011 11:23 PM, Bhatia, Manav (Manav) wrote:
> Ok, let me put it this way. I was trying to propose extending NTP to
support having an arbitrary MAC size. I was also trying to push
HMAC-SHA-1, instead of the regular SHA-1.

That's one possibility. Another is SHA-2. That's why we need to revisit
the question.

Danny

> 
> Cheers, Manav
> 
>> -----Original Message-----
>> From: Danny Mayer [mailto:[email protected]] 
>> Sent: Tuesday, December 13, 2011 9:49 AM
>> To: Bhatia, Manav (Manav)
>> Cc: David L. Mills; [email protected]; [email protected]
>> Subject: Re: [TICTOC] [ntpwg] NTP Extension Field without 
>> Authentication
>>
>> On 12/12/2011 9:50 PM, Bhatia, Manav (Manav) wrote:
>>> Hi Danny,
>>>
>>> I had raised a similar point way back in 2008 that we cant 
>> assume that the MAC in NTP will always be just 16 bytes.
>>>
>>
>> It already isn't. The reference implementation supports both MD5 and
>> SHA-1 with different MAC sizes. The reference implementation 
>> assumes one or the other depending on the size and that is 
>> obviously not sustainable.
>>
>> Danny
>>
>>> http://lists.ntp.org/pipermail/ntpwg/2008-December/001435.html
>>>
>>> Cheers, Manav
>>>
>>>> -----Original Message-----
>>>> From: [email protected]
>>>> [mailto:[email protected]] On Behalf Of Danny Mayer
>>>> Sent: Tuesday, December 13, 2011 7:07 AM
>>>> To: David L. Mills
>>>> Cc: [email protected]; [email protected]
>>>> Subject: Re: [TICTOC] [ntpwg] NTP Extension Field without 
>>>> Authentication
>>>>
>>>> Dave,
>>>>
>>>> You are talking about the current specification. Tal is 
>> talking about 
>>>> changing the specification.
>>>>
>>>> I see no reason why it cannot be done, provided we can 
>> find a way for 
>>>> it to be specified in a backward compatible way. The real 
>> question is 
>>>> whether or not it should be made optional.
>>>> The parsing rules can always be changed, not necessarily 
>> easily, but 
>>>> it can be done.
>>>>
>>>> RFC 5906 (autokey) should require it whatever is decided.
>>>>
>>>> Can you remind everyone why we might like to make it not 
>> mandatory? 
>>>> Also we need to have a MAC can we specify the 
>> algorithm/size for the 
>>>> MAC since we shouldn't make assumptions about this as has 
>> been done 
>>>> so far.
>>>>
>>>> Danny
>>>>
>>>> On 12/12/2011 1:53 PM, David L. Mills wrote:
>>>>> Tai,
>>>>>
>>>>> All I can say is read my message again. Doing without the 
>> MAC is a 
>>>>> very special case and unintended by the specification.
>>>>>
>>>>> Dave
>>>>>
>>>>> Tal Mizrahi wrote:
>>>>>>
>>>>>> Hi Dave,
>>>>>>
>>>>>>  
>>>>>>
>>>>>> So you are saying that according to the current spec it is
>>>> possible
>>>>>> in some configurations to have an extension field without the 
>>>>>> existence of a MAC?
>>>>>>
>>>>>>  
>>>>>>
>>>>>> Tal.
>>>>>>
>>>>>>  
>>>>>>
>>>>>> *From:*David L. Mills [mailto:[email protected]]
>>>>>> *Sent:* Monday, December 12, 2011 6:01 AM
>>>>>> *To:* Tal Mizrahi
>>>>>> *Cc:* [email protected]; [email protected]
>>>>>> *Subject:* Re: [ntpwg] [TICTOC] NTP Extension Field without 
>>>>>> Authentication
>>>>>>
>>>>>>  
>>>>>>
>>>>>> Tal ,
>>>>>>
>>>>>> It's a little more complicated than it seems. The parsing rules 
>>>>>> assume that a message digest is always present if an
>>>> extension field
>>>>>> is present. The NT$ packet header, extension fields and MAC are 
>>>>>> multiples of 32-bit words. The minimum MAC length is 5 words and 
>>>>>> maximum length is 6 words. The minimum extension field 
>> length is 2 
>>>>>> words. If the remaining number of words during the parse
>>>> is less than
>>>>>> 7, the remainder is the MAC. If not, an extension field is
>>>> present. 
>>>>>> The parser updates the parser pointer folloowing the
>>>> extension field
>>>>>> and tries again.
>>>>>>
>>>>>> Thus, if there are at least 7 words remaining and the
>>>> extension field
>>>>>> eats up all those words, the MAC could be assumed absent . 
>>>> This is a
>>>>>> rather hokey design, but would in principle work.
>>>>>>
>>>>>> Dave
>>>>>>
>>>>>> Tal Mizrahi wrote:
>>>>>>
>>>>>> Hi,
>>>>>>
>>>>>>  
>>>>>>
>>>>>> Revisiting an issue that was raised a few months ago and
>>>> is yet to be
>>>>>> resolved:
>>>>>>
>>>>>> RFC 5905 defines an extension field. The RFC states that a
>>>> MAC must
>>>>>> be present when there is an extension field.
>>>>>>
>>>>>>  
>>>>>>
>>>>>> Obviously, it would be beneficial for various purposes to allow 
>>>>>> Extension Fields independent of whether the MAC is present.
>>>>>>
>>>>>>  
>>>>>>
>>>>>> Some people thought this is a mistake in the spec, and
>>>> that it should
>>>>>> be included in the errata. Others thought that Extension Fields 
>>>>>> without MAC are something new that needs to be defined in
>>>> a new document.
>>>>>>
>>>>>> This was discussed in IETF 81, and then revisited in the ad-hoc 
>>>>>> meeting in October, but no conclusion was reached.
>>>>>>
>>>>>>  
>>>>>>
>>>>>> It would be great to hear the opinion of the WG and the
>>>> chairs about
>>>>>> how to proceed with this.
>>>>>>
>>>>>>  
>>>>>>
>>>>>> Thanks,
>>>>>>
>>>>>> Tal.

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

Reply via email to