Hi Dieter,
On Tue, Jul 13, 2010 at 11:09:18PM +0200, Dieter Spaar wrote:
Hello Harald,
On Tue, 13 Jul 2010 16:06:55 +0200, "Harald Welte" <laforge(a)gnumonks.org>
wrote:
The Netzing NE110 (as well as other MTK phones) seem to have a bug in
their MO-SMS code. Dieter has told me he is looking at a workaround to this
(by sending dummy authentication request/response), and he will e-mail this
list once he has some results.
This was a tough one. I tried Authentication and Ciphering commands in
various combinations without any success. Finally I changed the allocated
channel to SDCCH instead of TCH. And voila, it worked, even without
Authentication and/or Ciphering.
Ah, ok. This of course is an interesting issue.
So it seems that there is a problem with MTK phones
(and maybe others)
and MO-SMS on a TCH channel. I am not sure yet if there is a workaround
to solve this and still use the TCH.
The problem is that I could not see a very well-defined / unified behavior
among all the various phoen models on what kind of "establishment cause"
they send during the RACH request.
So instead of assignign an SDCCH and later having to re-assing to a TCH
after learning the phone wants to make a voice call, we always assing a TCH
to the phone on any RACH request.
Technically, according to the specification, a phone has no right to make any
assumptions on the type of channel it will be assigned. So it is clearly a bug
in the MTK GSM stack.
However, the only real solution to this problem probably is to implement "late
assignment" of the TCH, i.e. first send the IMM ASS to a SDCCH, and then do an
additional ASSIGNMENT CMD to a TCH once we know the phone wants to make a
call (and not send SMS).
Regards,
Harald
--
- Harald Welte <laforge(a)gnumonks.org>
http://laforge.gnumonks.org/
============================================================================
"Privacy in residential applications is a desirable marketing option."
(ETSI EN 300 175-7 Ch. A6)