Le ven. 23 déc. 2022 à 09:29, Matti Vaittinen <mazziesaccount@xxxxxxxxx> a écrit : > > Hi Mehdi, > > On 12/22/22 16:06, Mehdi Djait wrote: > > Hello Matti, > > > > First of all I apologize for bothering you out of the blue and I hope it > > is not a bad time to send you an email with a couple of questions. > > No problem. But I suggest you CC the IIO mail list when sending this > kind of questions. It may be the messages where the list is not in CC > are regarded as spam by my mail filters... Besides, others with more > insight to IIO may be able to further help you when the list is CC'd ;) > First of all, thank you for the quick response! I will CC the IIO mailing list next time :) > > My name is Mehdi Djait, I am a student living in Germany currently, > > REALLY interested in linux kernel development and still trying to find > > something to contribute to the kernel. The IIO subsystem seems like a > > good place to start for a kernel Newbie and am currently investing my > > time in learning the subsystem (reading driver implementations, the > > IIO-core, lurking in the mailing list ...) I have been looking lately > > everywhere for sensors without a kernel driver and I stumbled upon the > > KX132/134-1211 accelerometers and that's why I decided to send you an > > email with these questions: > > Oh, thanks for contacting me :) So, you have the sensor hardware at your > hands? I think this is mandatory when writing a driver. > Yes of course! > > - Am I correct in assuming that the accelerometer does not have a kernel > > driver ? I looked in the RohmSemiconductor Linux-Kernel-Input-Drivers > > and Linux-Kernel-Sensor-Drivers GitHub repos and grepped inside the iio, > > staging and linux-next trees > > Yes and no :) There is a kernel driver which supports many of the kx* > accelerometers in the ROHM downstream Linux-Kernel-Input-Drivers, and I > believe it also supports the KX132/KX134. Yet, as the name suggests - > these drivers are not exposing the accelerometer using IIO but the INPUT > subsystem. Reason is that quite a few device manufacturers still use the > accelerometers via HAL written on top of the INPUT. Heikki Haikola (in > CC) has authored many of these drivers. The INPUT is not the preferred > subsystem for acceleromerers though - so you're correct when planning to > write an IIO driver for upstream. > > > - Is anyone working on a driver now ? If not, I am really interested in > > trying to write one for some basic functionality. I am fully __AWARE__ > > how difficult the task is (I did read your LinkedIn posts) but I am > > truly motivated and I will use every example I can find as reference: > > the drivers under iio/accel, the GitHub repos from RohmSemiconductor and > > your KX02AA driver that recently landed in linux-next (Congrats :) ) > > Upstream support for (ROHM or any other) ICs is very welcome. It enables > more and more people to benefit from the ICs. Companies can rarely > support small customers and open source drivers make it possible for the > people to work without this support :) This work has been on my TODO > list since I added KX022A driver - but it has had low priority and I > seem to always have some other tasks. So, Your work on these ICs is more > than welcome! > I am glad to hear that! > > I’m eager to receive your feedback! (and sorry for the long text) > > It was not too long :) > > If you wish to proceed with the driver, then I would have couple of > suggestions to you. As far as I remember the KX132 is not really _that_ > different from the KX022A. Sure the resolution is different as may be > the G-ranges. But the logic of the register interface is largely same. I > suggest you to study the possibility of extending the KX022A driver to > support the other variants. I would differentiate the IC types based on > device-tree and fill in the register offsets/required quirks in the > driver private data. > > The other suggestion from me is that you could check the IC's logic from > the Linux-Kernel-Input-Drivers. I think there were some minor > differencies in the behaviour for example when the BUFE or PC1 bits are > toggled. > > Feel free to ask us if you hit to problems with the work, but please > note that most of us in Finland are having a holiday season right now - > so many of us will be offline until January. > I will probably hit many problems ;) But I will try my best Kind regards Mehdi Djait > Yours, > -- Matti > > > Kind regards > > Mehdi Djait > > -- > Matti Vaittinen > Linux kernel developer at ROHM Semiconductors > Oulu Finland > > ~~ When things go utterly wrong vim users can always type :help! ~~ > >
