Good news, everyone [tm]
GPRS is by now in a fairly useful state. I've spent the better part of the last 3 days to fix all the remaining known bugs. It is working fine from a variety of phones including a G1 (MSM7200) , K800i (unkown baseband), TYTN2 (also Qualcomm MSM but with windows mobile), E680 (Neptune LTE), ...
The following parts are still missing: * Integration with the HLR, so every phone is accepted on GPRS * IP Header compression (ROHC) * Data compression (v.42bis) * Authentication (depends on HLR integration) * GEA3 Encryption (though most of the infrastructure is there) * Ability to route APNs to different GGSNs
So everyone who has access to a nanoBTS: Try it now. The build/config instructions haven't changed from what I described last time when I wrote a GPRS status update to this list.
Happy hacking, Harald
Hi,
So everyone who has access to a nanoBTS: Try it now.
I just tried it again, updating all the components.
And unfortunately it seems it doesn't work for me anymore.
osmo-sgsn says :
<0016> gprs_bssgp.c:343 BSSGP TLLI=0xe7e2fc84 UPLINK-UNITDATA <0017> gprs_llc.c:410 LLC SAPI=1 C FCS=0xea1c55(WRONG) CMD=UI DATA
several time when a phones tries to connect. I never used to have FCS errors before.
VTY shows there is a mm-context but no pdp-context. Also several LLC entities.
Not quite sure what's going on yet ... (my gprs knowledge is quite limited ATM)
The build/config instructions haven't changed from what I described last time when I wrote a GPRS status update to this list.
There is also a howto here: http://openbsc.osmocom.org/trac/wiki/OpenBSC_GPRS
Cheers,
Sylvain
Hi Sylvain,
On Sat, Jul 03, 2010 at 08:52:21PM +0200, Sylvain Munaut wrote:
So everyone who has access to a nanoBTS: Try it now.
I just tried it again, updating all the components.
And unfortunately it seems it doesn't work for me anymore.
Just an update for the list: it has been fixed. I was using some length values for certain IEs to do input validation... and those length values came from 04.08 where 24.008 already has much longer lengths for those IEs.
It's fixed now in master.
osmo-sgsn says :
<0016> gprs_bssgp.c:343 BSSGP TLLI=0xe7e2fc84 UPLINK-UNITDATA <0017> gprs_llc.c:410 LLC SAPI=1 C FCS=0xea1c55(WRONG) CMD=UI DATA
this is a problem with the debug statement not with the actual code. will fix it soon.
VTY shows there is a mm-context but no pdp-context. Also several LLC entities.
Not quite sure what's going on yet ... (my gprs knowledge is quite limited ATM)
The build/config instructions haven't changed from what I described last time when I wrote a GPRS status update to this list.
There is also a howto here: http://openbsc.osmocom.org/trac/wiki/OpenBSC_GPRS
good. Now we only need a actual wiki page for OsmoSGSN and its VTY commands.
-
Harald Welte -
Sylvain Munaut