Bug #5529
openInter BSC HO fails due to lack of MSC Preferred Codecs IE
30%
Description
Since https://gerrit.osmocom.org/c/osmo-bsc/+/27405 osmo-bsc requires the MSC Preferred Codec List IE in an incoming handover request as it should according to 3GPP TS 48.008 3.2.1.8
However, osmo-msc does not send it. TTCN3 tests are passing because the test suite generates the IE but does not require it.
It is suggested on the mailing list that the possible (probable?) easier? better? solution to generating this IE in osmo-msc is to work on the neels/codecs branch.
Unfortunately, further to neels' refactoring work on the commits, plus an osmodevcall, it is still maybe representing quite some work load to get that through code review.
A possible workaround (for a locally working system) would be to revert
https://cgit.osmocom.org/osmo-bsc/commit/?id=826ec9ff758c8a40fca2eaf6cca7989ff6471c83
Updated by neels 12 days ago
This sounds really familiar as if i have fixed that in the last year or so...
but i'd need to check from scratch myself now, i don't remember in detail. If
it is solved, then we have a ttcn3 test for it, maybe check there first.
I'm also not sure how the neels/codecs branch relates to it, I expect it
affects/improves the choice of codecs sent to a remote MSC, but the branch
shouldn't affect presence/absence of IEs.
Updated by fixeria 12 days ago
AFAICS, there is no testcase in MSC_Tests verifying presence of the above-mentioned IE. I submitted a few related patches:
https://gerrit.osmocom.org/c/osmo-ttcn3-hacks/+/28919 BSSMAP_Templates: add SpeechCodecList param to tr_BSSMAP_HandoverRequest [NEW]
https://gerrit.osmocom.org/c/osmo-ttcn3-hacks/+/28920 MSC_Tests: SpeechCodecList IE shall be present in BSSMAP HandoverRequest [NEW]
Updated by laforge 11 days ago
neels wrote in #note-4:
This sounds really familiar as if i have fixed that in the last year or so...
I don't know what happens in personal branches. But from irc/mailinglist/redmine/gerrit I only know of the discussion around the time keith reported this very bug 4 months ago.
Updated by fixeria 8 days ago
fixeria wrote in #note-5:
AFAICS, there is no testcase in MSC_Tests verifying presence of the above-mentioned IE. I submitted a few related patches:
https://gerrit.osmocom.org/c/osmo-ttcn3-hacks/+/28919 BSSMAP_Templates: add SpeechCodecList param to tr_BSSMAP_HandoverRequest [NEW]
https://gerrit.osmocom.org/c/osmo-ttcn3-hacks/+/28920 MSC_Tests: SpeechCodecList IE shall be present in BSSMAP HandoverRequest [NEW]
As expected, handover related testcases started to fail in ttcn3-msc-test.
https://jenkins.osmocom.org/jenkins/view/TTCN3/job/ttcn3-msc-test/1732/testReport/ (+10 failures)
Now osmo-msc needs to be modified to include the missing IE.
Updated by neels 7 days ago
- Status changed from New to In Progress
- Assignee changed from msuraev to neels
I've taken a little closer look, and fixing this issue is non-orthogonal to the msc SDP/codecs changes:
- Before the codecs patches, osmo-msc doesn't properly collect all the restrictions on codecs (from MS, RAN, ...)
- After the codecs patches, it becomes a lot easier to compose a "Speech Codec List (MSC Preferred)"
So we can either fix this issue twice: once before and once after merging the codecs patches.
Or we can postpone it until the codecs patches become merged, and fix it then.
Either way it kind of makes sense that I adopt this issue, so that i can easily coordinate the codecs issues ... with myself.
I'm just going to assign it to me now -- msuraev are you fine with that?
Updated by neels 7 days ago
- % Done changed from 0 to 30
Vadim has already made the TTCN3 tests for inter-BSC handover stricter, requiring a Codec List (MSC Preferred) IE to be present.
I have that test passing now on the osmo-msc neels/codecs branch (Ifcf719c5fc9e11749aafbd97bafda7f6f50973ea)
but besides presence, the test does not actually verify that the IE contents are useful or correct.
I also have to properly test the entire neels/codecs branch for real world usability before it can be merged.
Unlike the past years, I am actually working on it and this is actually happening in a reasonable timeframe.
keith, if you have a physical setup where you could build and run the osmo-msc neels/codecs branch
and see whether voice actually works (both before and after inter-BSC HO), your feedback would be very welcome.
I will do the same at some point not too far in the future, but not sure when exactly.
I can also try to fix this issue on the current osmo-msc master branch.
It would be the same patch, except figuring out which codecs to send is an entirely different question on current master.
Let me know if the priority is high enough to fix this for current master as well.
