WiMAX TLV value correct in debug but not correct in packet capture

[James Leavitt]

Sorry Alan,

I left that part out since it is coming through ok, here's the whole
thing (you can see the 00 00 60 b5 after the 1a 17):

1a 17 00 00 60 b5 1c 11 00 01 04 45 b7 02 04 cf b7 03 06 cf b7 01 00

1a 17 00 00 60 b5 1c 11 00 01 04 45 b7 02 04 d2 b7 03 06 d0 b7 00 00

Interesting theory though, I did *try* a change in the dictionaries in a
vain attempt solve this issue (tried the included wimax and wichorus),
but I rolled them back. I also compiled 3.0.0 and installed in a new
location, and never touched those dictionaries at all, same bizarre problem.

If the binaries are broken, then I now have two sets of broken binaries
(granted they are on the same platform so perhaps it's a library problem?).

Perhaps I should install a whole new system / os and test on it to see
if a similar problem exists. What I will try now is another TLV and see
how it behaves.

Thanks,

James




On 07/31/2013 01:19 PM, Alan DeKok wrote:
> Re: WiMAX TLV value correct in debug but not correct in packet capture
>
>   See the hex output. The "00 00 60 b5" is the WiMAX forum vendor ID.
> Your debug output has "00 01 04 45" in the same place.  So either the
> dictionaries are broken, or the binaries are broken.
>
>   Either way, this problem doesnt appear in a stock install with the
> stock dictionaries.
>
>   So what changes have you made, and why?
>
> On 2013-07-31, at 10:57 AM, James Leavitt
> <james.leavitt at corp.xplornet.com> wrote:
>
> > HI Alan,
> >
> > Still no dice. I've disabled the database and used the file as suggested
> > (which is something that I had yet to try, but as you recommended I've
> > done so). I have tried with and without the Session-Timeout and
> > Acct-Interim-Interval without any effect.
> >
> > Here's the hex output (from the attached pcap):
> >
> > 1c 11 00 01 04 45 b7 02 04 cf b7 03 06 cf b7 01 00
> >
> > 1c 11 00 01 04 45 b7 02 04 d2 b7 03 06 d0 b7 00 00
> >
> > And the debug snip (this is from a time I removed the other two
> > *working* values):
> >
> > [ttls] Got tunneled reply code 2
> >        WiMAX-Packet-Data-Flow-Id := 14
> >        WiMAX-Service-Data-Flow-Id := 14
> >        WiMAX-Service-Profile-Id := 14
> >        WiMAX-Packet-Data-Flow-Id += 17
> >        WiMAX-Service-Data-Flow-Id += 17
> >        WiMAX-Service-Profile-Id += 17
> >
> > Attached is a pcap of the transaction indicating the TLVs are not
> > consistent with the DB or the file. It has been consistent with
> > radsniff, although I use tcpdump / wireshark when comparing with the
> > working systems.
> >
> > One thing to note is that I am using TTLS and copying the values to the
> > outer tunnel, are you performing the same in your test? I wonder if it's
> > a library somewhere on the OS that's making it go awry. I keep thinking
> > I've set something that would make this happen, but I cannot get over
> > the fact that other values are working fine.
> >
> > Thanks,
> >
> > James
> >
> >
> >
> > On 07/31/2013 10:06 AM, Alan DeKok wrote:
> >> Re: WiMAX TLV value correct in debug but not correct in packet capture
> >>
> >> James Leavitt wrote:
> >>> After some compiling and configuring, I've managed to get version
> 3.0.0
> >>> up and running, and I seem to be having a similar issue:
> >>
> >>  I don't see that on my systems.  radsniff, radclient, and pcap all
> >> show that the WiMAX attributes are correct.
> >>
> >> Data:          1a  17  00 00 60 b5 1c 11 00 01 04 00 0e 02 04 00 0e 03
> >>                        06 00 00 00 0e
> >>                1a  17  00 00 60 b5 1c 11 00 01 04 00 11 02 04 00 11 03
> >>                        06 00 00 00 11
> >>
> >>  Please post a hex dump of the packets.  i.e. put this into the "users"
> >> file:
> >>
> >> bob     Cleartext-Password := "bob"
> >>        WiMAX-Packet-Data-Flow-Id  := 14,
> >>        WiMAX-Service-Data-Flow-Id := 14,
> >>        WiMAX-Service-Profile-Id   := 14,
> >>        WiMAX-Packet-Data-Flow-Id  += 17,
> >>        WiMAX-Service-Data-Flow-Id += 17,
> >>        WiMAX-Service-Profile-Id   += 17
> >>
> >>  And run "radclient -xxxx <args>
> >>
> >>  to do the test.  You will get a hex dump like I posted above.  It
> >> should be identical.
> >>
> >>  My guess is that you have FreeRADIUS using one WiMAX dictionary, and
> >> radsniff, etc. using another.  Some vendors made their own,
> >> incompatible, version of the WiMAX dictionaries.  Which is a stupid
> >> idea, but that's what vendors do.
> >>
> >>  Alan DeKok.
> >> -
> >> List info/subscribe/unsubscribe? See
> >> http://www.freeradius.org/list/users.html
> >>
> >> --
> >> This message has been scanned by MailScanner
> >>
> >
> > --
> >
> >
>
> > <1370_tlv_issue.pcap>
> > -
> > List info/subscribe/unsubscribe? See
> http://www.freeradius.org/list/users.html
> -
> List info/subscribe/unsubscribe? See
> http://www.freeradius.org/list/users.html
>