drivers - cc110x_ng: Propose new CC110x driver implementation

[fnack]

This PR proposes a new implementation of the cc110x_ng driver and shows example usage for the MSB-IoT board. The new implementation features:

• Usage of the low-level periph driver functionality (making board-specific spi communication implementations obsolete)
• Implementation of the new netdev base interface (still WIP in some parts)

I know this implementation still leaves room for optimization because of a lot of duplicate code between the old and new driver. Proposals to slim the code are very welcome!

This PR depends on #1770

[OlegHahm]

netdev was "invented" in 2014 by @authmillenon, employed by FUB at this time, so I think it's a safe bet to change this accordingly.

[OlegHahm]

I'm definitely no expert on the low-layer stuff and couldn't test, but in general this looks good to me. I'm only a bit worried about the increasing confusion by RIOT now providing three different drivers for the cc110X, but I don't see how this could easily be solved. However, you might want to add some more documentation about how and when this driver can be used - and we should hope that someone's willing to port other boards with this radio chip to the new periph model.

[miri64]

The cleanest solution to reduce the number of cc110x drivers would be to fix #1540 (edit: for the boards in question) first and remove drivers/cc110x then, I guess ;-)

[LudwigKnuepfer]

I know that naming is the hardest problem in CS, but I really dislike the naming scheme of $base, $base_ng, $base_v2. I propose to rename them all in a more sensible manner (e.g. $base_w_mac, $base_periph, ... or even $base_old, $base_older), if the old modules need to be kept.

[haukepetersen]

I agree, that we should be careful with the naming of the drivers to save some confusion. How about something like renaming the old ones into $base_legacy and $base_legacy_w_mac and taking keeping the new one as $base? This way the old boards will still work and could slowly be migrated to the periph driver version of the driver...

[LudwigKnuepfer]

Sounds good to me.

[OlegHahm]

No objections from my side.

[fnack]

Definitely better than the existing names!

[LudwigKnuepfer]

I think the name changing of the older drivers should be done in a fresh PR.

[LudwigKnuepfer]

(and for all other legacy drivers on the go)

[LudwigKnuepfer]

=> #1776

[LudwigKnuepfer]

@fnack rebase?

[LudwigKnuepfer]

(and rename)

[LudwigKnuepfer]

Also, quickly looking over the commits, there appear to be some fixes/extensions of other modules that could be useful to move to separate PRs (the critical commenting error, SPI, possibly more).

[fnack]

I'll do that today! I'll squash and will think about moving some of the commits. The spi change for example was already merged in master with another PR.

[fnack]

What about the cppcheck errors from my travis build? They are definitely false-positives. It looks like cppcheck has problems with bit field struct members like unsigned LL_ACK : 1;

Is this a known issue @N8Fear?

[miri64]

Bit-field struct members are discouraged (see #1609)

[miri64]

For other false-positives, see https://github.com/RIOT-OS/RIOT/wiki/Coding-conventions#travis

[fnack]

Fixed the cppcheck-findings. Finally, Travis is happy :)

[fnack]

Rebased. I would really like to get this PR merged soon. What tasks are left to get an ACK @OlegHahm and @haukepetersen?

[OlegHahm]

broken indentation (I know it was broken before, too)

[fnack]

Could you explain the problem to me? I don't really see what is wrong with the indentation x)

[OlegHahm]

Sorry, is apparently correct - just a browser issue. Never mind!

[OlegHahm]

Apart from minor comments, this PR looks good.

[fnack]

I addressed your comments.

[OlegHahm]

Apart from minor comments, this PR looks good.

...and works (as locally demonstrated). (This is an ACK.)

[fnack]

Rebased, squashed and Travis is happy too. Go...