On Mon, 11 Apr 2022 09:52:19 +0100 Jonathan Cameron <Jonathan.Cameron@xxxxxxxxxx> wrote: > On Sun, 10 Apr 2022 19:37:52 -0300 > Maíra Canal <maira.canal@xxxxxx> wrote: > > > On 04/10, Jonathan Cameron wrote: > > > On Thu, 7 Apr 2022 00:23:29 -0300 > > > Maíra Canal <maira.canal@xxxxxx> wrote: > > > > > > > Hi everyone, I am Maíra Canal an undergrad student at the University > > > > of São Paulo, Brazil, pursuing > > > > computer engineering. I wish to participate in the GSoC 2021 as a part > > > > of the Linux Foundation, IIO Project. > > > > > > Hi Maíra, > > > > > > Nice to 'meet' you ;) > > > > > > > > > > > I have been contributing to the Linux kernel for a couple of months > > > > and have more than 20 > > > > accepted patches in a couple of subsystems. > > > > > > > > I started looking through the catalog of Analog Devices Inc. and I'm > > > > pretty interested in writing a driver for gyroscopes, inertial > > > > measurement units (IMUs), magnetometers, pressure sensors, proximity > > > > sensors, or temperature sensors. But, while looking through the > > > > catalog, I could not figure out a sensor that would be relevant to > > > > Linux Kernel. I mean, I would like to work on a sensor that would be > > > > relevant to the community and to Analog Devices Inc. > > > > > > > > In that sense, I would like to know if anyone in the IIO community > > > > could recommend a sensor that would make sense for the company and the > > > > IIO community. Any suggestion is appreciated! > > > > > > I'm not going to recommend a particular sensor, but more offer some general > > > tips on what 'sort' of device makes a good target for a GSOC. > > > Finding a sensor means trawling datasheets and I'm tight on time today > > > + I've no real insight into what the ADI folk might like to see > > > supported! > > > > > > The nature of a GSOC driver submission is often a little different to > > > how an experienced driver author might go about things, simply because you > > > will / should be looking for feedback at more stages of development and > > > hopefully to upstream things in multiple stages. An old hand at IIO > > > drivers will often just jump directly to a driver supporting all the > > > features they wish to target. As such, the 'perfect' device to target > > > should meet a few requirements that may not be true for the approach of jumping > > > straight to the end goal. Note this is equally true for other people > > > starting out writing drivers - though they can often do very simple > > > devices first and that is not a good plan for a GSOC project where > > > you need to have a progression during the project. > > > > > > Try to find something that offers some advanced features to provide > > > stretch goals but make sure the basic functionality will work with > > > a much simpler driver. So devices that provide straight forward > > > registers to access the latest channel value are great, whereas > > > those that only offer a streaming interfaces / fifo may be less suitable. > > > However if they offer both that is perfect as the fifo make a good > > > later feature for a GSOC project if things are going particularly > > > well! For a real stretch goal, find a device with features that > > > we don't support at all today (perhaps new sensor types, or some > > > other new feature) as they'll give you the experience of defining > > > new ABI + possibly modifying the IIO core to meet some requirements. > > > > > > Another thing to look at it is whether the part is sufficiently > > > different from those supported by existing drivers to justify a > > > separate driver. If not, you may find your GSOC project becomes > > > simply adding an ID! (then rapidly choosing a second device to > > > work on). > > > > > > Hope that provides a few hints on what to look at. Probably the best > > > way around is to suggest one or more parts you think look interesting > > > then we can give feedback on whether we think they'd be a good choice > > > or not. > > > > Hi Jonathan, > > > > I really appreciate the answer. Thank you for your attention and time! > > > > During the week, I ended up picking the ADXL375 accelerometer (although I am > > open to any change proposed by ADI or the IIO community). Based on that device, > > I wrote a proposal and I would appreciate if you provide some feedback on the > > device choice and proposal: https://pt.overleaf.com/read/xsmmdpvzqrhd. > > Unfortunately that part hits the second to last paragraph above. > > It's so nearly compatible with the ADXL345 that the driver already supports it: > https://elixir.bootlin.com/linux/v5.18-rc1/source/drivers/iio/accel/adxl345_i2c.c#L42 > This is very common for these types of devices as there are often many similar > variants, usually with different ranges or with small additional features, or numbers > of interrupt pins etc. Another fun one is parts with ratings for different applications > but identical software interfaces being given different part numbers. > I think in this case the two parts have different scaling, but are otherwise > identical. > > So probably need to find another part. > I took a quick look at your proposal. Main feedback is look to do more steps in submission for upstream. I'd look to post basic read / write via sysfs first without the buffers or events. Key here is that it can take several weeks to get review (particularly during the summer as people can be on vacation) so you want to have something available for review as early as possible. It's absolutely fine to post a v2 with additional features even though v1 hasn't gotten review yet. (just reply to v1 to say you have done so to avoid anyone wasting time reviewing old code). It might make sense to do a project focused on the adxl375 if the aim is to fill in missing features in the driver. However, that's high risk as you may be posing questions that need considerable discussion on list. Long ago we had one GSOC project that ran into such a problem and ended up going in a circle with 3 different solutions proposed before we ended up back where we started. Project was a success but I'd imagine it was very stressful for the person doing the project! Jonathan > Jonathan > > > > > Regards, > > Maíra > > > > > > > > Good luck! > > > > > > Jonathan > > > > > > > > > > > > > > > > > Sincerely, > > > > Maíra Canal > > > >
