Attention is currently required from: daniel, laforge, pespin.
fixeria has posted comments on this change. ( https://gerrit.osmocom.org/c/osmo-ttcn3-hacks/+/36601?usp=email )
Change subject: sgsn: TC_sgsn_context_req_in: match PDP Context IE ......................................................................
Patch Set 6:
(1 comment)
File sgsn/SGSN_Tests.ttcn:
https://gerrit.osmocom.org/c/osmo-ttcn3-hacks/+/36601/comment/be34a9b2_441b8... PS6, Line 3748: f_pdp_ctx_act(apars);
@vyanitskiy@sysmocom. […]
I don't think it really matters for this specific test scenario whether the requested IP address matches the assigned one. Likewise, I don't think it matters here whether we allocate IP address statically or dynamically.
Just to clarify: I was not the one implementing `f_pdp_ctx_act()`, so I am simply trying to use the existing code base. What you suggest makes sense for those testcases specifically testing the PDP Context Activation procedure. And AFAICS, the existing testcases do not cover these scenarios, unfortunately.
Here I just want to make sure that osmo-sgsn provides the `PDP Context IE` in `SGSN Context Response`. The IP address assignment approach is a less significant detail preceding the "core" phase of the test scenario.
I have two options with the existing code base:
* a) leave `apars.addr.addressInfo` empty (`omit`) - this is the default, and this results in a PDP Context activated with no address at all. The testcase is failing because osmo-sgsn does not include the `PDP Context IE`. * b) assign a valid address to `apars.addr.addressInfo` - this is what I did in the last patch revision and got the testcase passing.
Regarding a), @dwillmann@sysmocom.de told me that it's impossible to encode a `PDP Context IE` in `SGSN Context Response` without an IP address. I just checked 3GPP TS 29.060, section 7.7.29 myself, and AFAICS it **is** possible: we can simply set `PDP Address Length` to 0.
So all in all, I can:
* stick to b) merging the current patch revision; * fix osmo-sgsn to handle PCP contexts with empty IP address correctly (is this even a valid scenario on practice?) and revert back to the previous patch revision a); * spend more hours on improving the testsuite, implementing additional testcases for static vs dynamic + requested != assigned scenarios (this goes beyond the scope of OS#6294).