Patch Set 1:

Patch Set 1: Code-Review+1

I'm wondering why call_leg_local_bridge() is even called before all sides are ready.
I think placing a safeguard here to avoid a crash is a good thing (fix DoS).

But if the RAB Assignment didn't complete, it should not even be reaching this stage, right?

Agree, I also find this a bit weird that it reaches this point. I guess on the 3g side (MO side) the call proceeds due to CM Service Request, and is not currently waiting for RAB Assignment-Response?

I think the point is that the RAB-AssignmentRequest does complete, but the RAB-AssignmentResponse wasn't forwarded by the hnbgw for some reason. The call then proceeds. (Connect, Connect Acknowledge...) Thats at least how I can trigger the problem. I just uncomment the line where the modified RAB-AssignmentResponse is forwarded.

After all the HNB sends the RAB-AssignmentResponse and then the Connect. Due to the MGW handling it is now possible that the Connect overtakes the RAB-AssignmentResponse, if the operations on the MGW take very long. This is probably not a real world problem but maybe we have to synchronize the Connect message. I am not sure. At least from the trace I can see there are about 5 sec. between RAB-AssignmentResponse and Connect.

View Change

To view, visit change 26917. To unsubscribe, or for help writing mail filters, visit settings.