gnrc/rpl/control_messages: refactor gnrc_rpl_recv_DIO

[elenaf9]

Contribution description

The gnrc_rpl_recv_DIO function in gnrc_rpl_control_messages is > 150 LOC that are quite hard to understand, especially as someone that is new to RIOT.

This PR refactors the function into smaller sub-functions and extracts logic that is common between the two main branches (creating a new DODAG and updating an existing DODAG).

I split the refactoring into separate commits that are easier to review one-by-one; can squash them together at the end.

I am not super happy with the resulting code, but I think it is a small improvement.
Further feedback on how to improve it would be appreciated.

Open TODOs

• Document new sub-functions?

Testing procedure

This PR shouldn't change any logic apart from the order in which individual DODAG properties are updated (see a8bc3ad).

I tested it manually:

• examples/networking/gnrc_networking + added USEMODULE += nimble_rpble
• 9 nodes (nrf52dk), 1 node A set as root (ifconfig 8 add 2001:db8:1::1/64 && rpl root 1 2001:db8:1::1)
• => RPL tree was created as expected with A as root

Issues/PRs references

Preparation for future work on #21574.

[riot-ci]

Murdock results

✔️ PASSED

a461e5c gnrc/rpl/control_messages: doc gnrc_rpl_recv_DIO

Success	Failures	Total	Runtime
10515	0	10516	12m:20s

Artifacts

• Documentation preview

[benpicco]

Thank you for taking up the task of breathing some new life into RPL! A working mesh protocol in RIOT would be highly appreciated.

[elenaf9]

Friendly ping :)

I have a few other changes in RPL that I'd like to PR but they depend on this branch.

[benpicco]

Looks good to me - sorry for the delay.
I suppose you ran e.g. tests/net/gnrc_rpl (unfortunately we disabled that on CI 😕)

[elenaf9]

Last commits just add docs, and fix one copy-paste error I noticed while self reviewing.

One more thing I noticed just now is that there is actually a minor logic change in this PR:

We now don't remove the whole RPL instance anymore if our preferred parent sent a DIO message with invalid options. Instead we just early return after the parent was updated (which doesn't change any dodag opts).
I could change this back by returning different error codes from _update_dodag_from_DIO and handling them appropriately, but I was wondering if we even want to remove the whole dodag if there is a (potentially one-time) error in the opts?

---

I suppose you ran e.g. tests/net/gnrc_rpl (unfortunately we disabled that on CI 😕)

Yes, I ran all RPL tests, and did some manual testing described in the PR description.

---

Edit: Should I squash now? (will fix the typos while squashing)