Hi Mike!
Some comments:
* it would have been good to Cc the mailinglist when you re-posted the patch,
this way more eyes are on the code
* I've looked at it but unfortunately there's a number of issues. That's
why I've committed it to a new 'ussd' branch in git.
* I've pushed an incremental patch as commit that cleans up the coding style
issues that I had seen with the code
There are also issues with the actual code that I'd like to see addressed,
preferrably by incremental patches to my ussd branch:
* the use of static global variables like ussd_string_buf in gsm_04_80.c
Code like this will prove very difficult to ever use concurrently at some
later point. global variables in the main program (or similar) are fine,
but in actual 'library' style code, we should not keep code in static global
variables across multiple calls. In such a case, the state needs to be
associated with some of the data structures we have in memory anyway.
ussd_string_buff looks like something the caller should allocate and pass into
gsm0480_rcv_ussd(). Also, since rcv_ussd() only decodes, it should probably be
called gsm0480_decode_ussd().
Things like the static last_transaction_id also will cause race conditions once
multiple people send USSD requests at the same time. This data needs to be per
subscriber, or per lchan, I presume.
* the assumption that extension numbers are not longer than 5 characters
in ussd.c is really not good. Please check the specs what is the maximum
length of the phone number in this context, add a #define and then make sure
the string buffer is large enough for it. Also, I prefer snprintf() to
having XXXX that then get overwritten by memcopy :)
If you could send patches for those two issues (one patch for each logical
change), I would appreciate that and in the end merge the ussd branch into
master.
Thanks.
--
- 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)
Hi everyone,
As I mentionned in the earlier post, I think it'd be great if we could
use FOSDEM to get known more. These projects covers a wide area, from
quite low level RF stuff, signal processing to higher level GSM
protocols ...so the work is certainly not missing :)
So what I'd like to know is who's planning on coming (yes/no/maybe) to
FOSDEM and if you'd be ready to help out or have equipment you could
bring.
Exactly _what_ we could do will most likely depend on the number of
answer to this mail :)
I've included OpenBTS, OpenBSC and airprobe here because they're the
three projects I know that deal with the 'network' part of GSM (as
opposed to the MS side for things like OpenMoko, Android, ...). And
even if they currently don't share that much code, the GSM knowledge
you gain by contributing to one of these can be directly applicable to
help out in the others.
Cheers !
Sylvain
(Hopefully) last patch attached - now sends the correct op-code in
response, works with all 6 phones I have tried. Interesting that 5 out
of 6 were happy with a completely different op-code and payload...
Mike H.
For info:
Ok, I called the IBPT and I got the procedure to follow to request frequencies.
The 900 MHz is entirely allocated to commercial operators but in the
1800 spectrum there are possibilities to get experimental licenses for
a small fee so I'm filling up the paper work ASAP.
For OpenBTS to work in the 1800 band we may need more stable clock
source (which I don't have yet).
For OpenBSC, I have a 1800 nanoBTS.
Sylvain
On Mon, Oct 12, 2009 at 8:38 PM, Sylvain Munaut <246tnt(a)gmail.com> wrote:
> On Mon, Oct 12, 2009 at 7:20 PM, Michel Daggelinckx <03taxi(a)gmail.com> wrote:
>>
>>
>> FOSDEM [1], the most developer-oriented Free and Opensource conference in
>> Europe, is taking place in Brussels, Belgium [2] on *Saturday 6 and Sunday 7
>> February 2010*. Apart from having many invited speakers, the conference
>> offers developer rooms, stands and lightning talks to projects from the Free
>> and Opensource community. This results in a staggering number of 250+
>> lectures!
>>
>> We hereby welcome proposals from speakers and projects to talk in a main
>> track, man a stand, or hold a lightning talk. /Information on developer
>> rooms
>> will become available later, as we are slightly reworking the concept/.
>
> In relation to that, I'm trying to get an experimental license from the IBPT to
> operate on the GSM spectrum during the conference. I didn't find much in the
> legislation about this kind of licence in Belgium, but I must call
> back the guy tomorrow ...
>
> I didn't take any contact with the fosdem yet to have a spot or
> anything, I wanted to
> hear back from the ibpt first.
>
> If other people are motivated, there is definitely something to do there.
> I have equipment to run both OpenBTS and OpenBSC and I'm located in Belgium,
> so I'll definitely be there.
>
>
> Sylvain
>
Hi!
I finally finished the report from the GSM test network that we operated at
HAR2009 using OpenBSC and two BS-11.
The report is available for download from
http://openbsc.gnumonks.org/trac/attachment/wiki/HAR2009/har2009-gsm-report…
Regards,
--
- 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)
>I believe in the positive impact of OpenBTS and OpenBSC
interoperability for
>the both communities and as encouraging factor for possible future
OpenBTS
>test sites.
>
>We're ready to work with you beforehand to minimize on-site efforts.
>
>PS Where can we find agenda? Official 26C3 site is nearly empty.
hi all,
on the congress i use linux-call-router to route isdn traffic from
versatel to berlin and vice versa for free as part of versatel
sponsoring. i would like to join the POC (phone operation center) in the
upper hall again this year. there we sit close to the "eventphone" guys,
so we can easier communicate and do cabeling of our hardware. it would
be nice to have also the openbsc and openbts projects close to us. we
could easier interact and build one test networks. since eventphone uses
an own linux-call-router hardware, i can use my test machine without
disturbing the external isdn connection to versatel. it is equipped with
two E1 cards, one for interconnection with eventphone, the second one
can be used for BS11. eventphone has a big DECT pbx running and provides
mobile access.
what do you think?
andreas
Hello again,
More patches - these two (combined) add support for mobile-originated
USSD. The demonstration application, implemented in ussd.c, is that
sending *#100# to the network will display your 5-digit extension. It's
not 100% finished - I've tested it on 6 handsets, 5 of them work
perfectly, but the 6th (Samsung i520) doesn't seem to receive the
response, so more experimenting to be done...
Best regards to all,
Mike H.
Hi folks.
I can't get my setup working. Everything works, the BS11 boots up well,
openBSC loads well. At the software side, everything looks perfect.
But the setup does not work. I have testet my BSC now with 3 different
BS11. One of the BS11 is brandnew (Has been set in standalone directly
after installing the firmware.)!
I found this and i think that this is an explaination of the problem:
http://lists.gnumonks.org/pipermail/openbsc/2009-August/000783.html
Finally my question is if anone of you ever observed a semilar effect.
regards.
Philipp