RIOT port for the MKW22D512 SiP and Phytec PBA-D-01 PhyWave Evaluations-Board

[jfischer-no]

This PR add support for Freescale MKW22D512 SiP and Phytec PBA-D-01 PhyWave Evaluations-Board.

[LudwigKnuepfer]

Please sync your repos master with RIOT's master and rebase on it.
(Feel free to ask for details if this doesn't make sense to you.)

[jfischer-no]

@LudwigOrtmann: I do not know if I did it right, scolds me :-)

[haukepetersen]

Maybe have a look at the comment from BytesGalore on the bottom of the comments in #2018. I think he explained the rebasing quite nicely...

And by the way: Welcome to RIOT!

[LudwigKnuepfer]

OK, step by step. I assume git://github.com/RIOT-OS/RIOT.git is the remote called upstream and git://github.com/jfischer-phytec-iot/RIOT.git is called origin.

# get updates, reset master
git fetch --all
git checkout master
git reset --hard upstream/master

# rebase this branch
git checkout wip@mkw2xd
git checkout -b backup # create a backup, this also checks it out
git checkout wip@mkw2xd
git rebase -i master
# throw out all unrelated commits (there is an on-line help in the view that opens)

# in case of conflicts, you might need
git mergetool # to resolve conflicts
git rebase --continue # after resolving conflicts
git rebase --abort # if you messed up

# push changes
git checkout master
git push -f origin master
git checkout wip@mkw2xd
git push -f origin wip@mkw2xd

[jfischer-no]

@LudwigOrtmann : done, but travis has a problem with unknown license header: 'cpu/mkw2x/include/MKW22D5.h'
@haukepetersen : thanks, we already know us (libopencm3, nrf51 :-) )

[LudwigKnuepfer]

@jfischer-phytec-iot Would be nice to use your fresh first-hand experience to improve this wiki entry if necessary:
https://github.com/RIOT-OS/RIOT/wiki/Git-cheatsheet#how-to-rebase-your-master-on-a-current-riot-master

[LudwigKnuepfer]

Regarding the license - I hope @haukepetersen can help.

[haukepetersen]

@jfischer-phytec-iot Yes, I remembered by now... Got confused by your new github name... Nice port by the way! Hope I will get to reviewing it soon...

[haukepetersen]

As to the license, as far as I see it the file doesn't have one... So I guess it should be safe to include it in the license checker?!

[jfischer-no]

@haukepetersen I have created a new account for work. Can we add the description for our board in RIOT-wiki? What do you mean with "include it in the license checker"?

[haukepetersen]

Of course, just go ahead! By the way, is the board freely available/buyable?

With license checker i meant the script that is run by Travis which compares every single license header found in RIOTs source code against a set of pre-defined headers. These can be found in RIOT/dist/tools/licenses/patterns/*. Just add a pattern for the Freescale header file included in your PR.

[jfischer-no]

@haukepetersen We have presented our Modules and Board on Electronica last week. The Board should be available at the Embedded World, but we can provide some developer samples, specially for RIOT-developers :-).

[haukepetersen]

Sounds great! And also one more reason to go to the Embedded World next year... I take it there is no public information (i.e. datasheets/manuals) available so far? If I had the right board on your website it looked quite interesting!

[LudwigKnuepfer]

@haukepetersen The file says it is a CMSIS file - I thought there were replacements for that?
And either way - the file needs to have a license. We can't add a 'no license is fine' rule to the checker or whitelist individual files.

[jfischer-no]

@haukepetersen I will clear the license type with Freescale.

[haukepetersen]

perfect!

[PeterKietzmann]

Tested i2c with #2119, #2121, #2123 and #2148. Worked fine

[OlegHahm]

transceiver is not implemented yet.

[jfischer-no]

We will completely rework it in the next few days. Transceiver will also appear soon.

[OlegHahm]

So, should we mark this as WIP for now?

[jfischer-no]

Yes

[OlegHahm]

Do I see it correctly, that the license issue has been resolved?

@jfischer-phytec-iot, what's left on the Todo list?

[jfischer-no]

@OlegHahm It is a CMSIS header with 3c-BSD license now.
Todo list? You mean #2188?
I have moved all peripheral driver in kinetis_common. I'm also porting RIOT to KW01Z128 and will make a PR soon. Internally, we already use kinetis_common for two MCUs (KW22D512 and KW01Z128). I think kinetis_common should include initially only the peripheral (uart, spi, i2c, hwtimer) drivers.
Currently I do not know what I can improve on that branch :-).

[OlegHahm]

For this PR, please go on and squash and merge at will.

[jfischer-no]

fccbb18 makes travis unhappy (failed: driver_at86rf2xx, driver_xbee, unittests)
I will remove it and squash. After #2756 has merged, we can open a PR for fccbb18.

[OlegHahm]

agreed

[jfischer-no]

travis was canceled 😕 ?

[OlegHahm]

I killed some of the outdated builds in Travis - if I accidentally killed the most current one, too, I'm sorry, but apparently it's running again.

[jonas-rem]

As https://www.traviscistatus.com/ states, there was an issue today. Maybe it was related to that somehow.

[jonas-rem]

Travis failed because the testapp for the isl29125 driver uses GPIO5, which is per default disabled in our board.

[OlegHahm]

How about either modifying the Makefile with a special case for you board or (what I would prefer) blacklist the application?

[jonas-rem]

I think we might not even need a special case for our board. Since GPIO_5 is just the default value we could change it to GPIO_0? This would also be consistent with the ng_at86rf2xx driver that counts the defaults from GPIO_0. GPIO_0 is defined by default and should be defined on all boards. As the comment in the Makefile (tests/driver_isl29125) says, this seems to be just a random value "set random default".

[OlegHahm]

Sounds good, too. @LudwigOrtmann, it's "your" driver. Any reason against changing this?

[jfischer-no]

I have reorganized the GPIOs, again, a few more can not hurt.

[jonas-rem]

Oh, the changes in periph_conf.h that were neccessary to build the kw2xrf app are included in the postponed fccbb18 commit. However, if you comment-in GPIO 24 conf everything should be fine.

/* GPIO channel 24 config */
/*
#define GPIO_24_DEV         KW2XDRF_GPIO
#define GPIO_24_PORT        KW2XDRF_PORT
#define GPIO_24_PORT_BASE   KW2XDRF_PORT_BASE
#define GPIO_24_PIN         KW2XDRF_PCS0_PIN
#define GPIO_24_CLKEN()     KW2XDRF_PORT_CLKEN()
#define GPIO_24_IRQ         KW2XDRF_PORT_IRQn
#define KW2XRF_CS_GPIO      GPIO_24
*/

[jfischer-no]

my fault, i will update 470781b. i will revert 470781b

[jfischer-no]

@jremmert-phytec-iot
jonas-rem@3287e71#diff-26037c789932d7c8d7bea2a82565588dR8
I do not want to enable GPIO_24 because we are using hardware CS per default (KW2XRF_SHARED_SPI=1).

[jonas-rem]

We have (KW2XRF_SHARED_SPI=0) as default. But yes, that means hardware CS.

I have also thought about that, but didn´t saw an alternative. How should we handle the parameter KWRF_CS then? We whould have to introduce two different parameter set of kw2xrf_init; i) without cs_pin and ii) with cs_pin and distinguish and let the preprocessor decide what function to use. I wanted to avoid that.

 54 int kw2xrf_spi_init(spi_t spi, spi_speed_t spi_speed,
 55                     gpio_t cs_pin)

Do you have other solutions in mind?

[LudwigKnuepfer]

Just define a specific GPIO for your board in the drivers Makefile.

[jfischer-no]

@OlegHahm ack & squash ?

[OlegHahm]

superfluous space

[OlegHahm]

Yes, please fix the tiny indentation thing and squash. ACK

[jfischer-no]

GO?

[OlegHahm]

Push the button!