Re: Heimdal / MS kpasswd differences?

["Henry B. Hotz" <hotz@jpl.nasa.gov>]

On Sep 4, 2006, at 12:38 AM, Love Hörnquist Åstrand wrote:

>
> 4 sep 2006 kl. 01.55 skrev Michael B Allen:
>
>> On Mon, 4 Sep 2006 08:51:04 +1000
>> "ronnie sahlberg" <ronniesahlberg@gmail.com> wrote:
>>
>>> what client are you using?  I have never seen a client before  
>>> that reuses
>>> the same clientside port between both krb5 and kpasswd
>>
>> I see. So you just need to look a little deeper before picking a  
>> decoder.
>>
>> The client is stock Heimdal kpasswd 0.7.2. I'm using mechglue branch
>> with some minor modifications but I don't think there's anything that
>> could affect what UDP port is being used. I'd say that's normal  
>> Heimdal
>> behavior unless Love says otherwise.
>
> I think that is not normal behavior, I just tried it out on my mac  
> and tcpdump tells me this.
>
> 09:34:13.075186 IP hummel.it.su.se.65065 > pal.su.se.kerberos:  v5
> 09:34:13.078695 IP pal.su.se.kerberos > hummel.it.su.se.65065:  v5
> 09:34:26.126523 IP hummel.it.su.se.65068 > pal.su.se.kpasswd: UDP,  
> length: 610
> 09:34:26.168186 IP pal.su.se.kpasswd > hummel.it.su.se.65068: UDP,  
> length: 274
> 09:34:26.273385 IP hummel.it.su.se.65071 > pal.su.se.kpasswd: UDP,  
> length: 576
> 09:34:26.310696 IP pal.su.se.kpasswd > hummel.it.su.se.65071: UDP,  
> length: 274
>
> Love

Reusing ephemeral ports would be OS dependent and probably also  
depends on the time interval between the two exchanges, *and* *also*  
might depend on whether it's the same program/task doing the two  
exchanges.  Don't think it's recommended to *depend* on ephemeral  
port numbers always being different.
------------------------------------------------------------------------ 
----
The opinions expressed in this message are mine,
not those of Caltech, JPL, NASA, or the US Government.
Henry.B.Hotz@jpl.nasa.gov, or hbhotz@oxy.edu