Re: [PATCH v2] add support for Freescale's MMA8653FC 10 bit accelerometer

[Martin Kepplinger <martink@posteo.de>]

Am 2015-03-20 um 13:27 schrieb Benjamin Tissoires:
> On Fri, Mar 20, 2015 at 7:26 AM, Martin Kepplinger
> <martin.kepplinger@theobroma-systems.com> wrote:
>> Am 2015-03-19 um 11:22 schrieb Bastien Nocera:
>>> On Wed, 2015-03-18 at 19:28 +0100, Martin Kepplinger wrote:
>>>> Am 2015-03-18 um 19:05 schrieb Bastien Nocera:
>>>>> On Wed, 2015-03-18 at 19:02 +0100, Martin Kepplinger wrote:
>>>>>> Am 2015-03-18 um 17:59 schrieb Bastien Nocera:
>>>>>>> On Wed, 2015-03-18 at 17:42 +0100, Martin Kepplinger wrote:
>>>>>>>>
>>>>>>> <snip>
>>>>>>>> It could have gone to drivers/iio/accel if it would use an
>>>>>>>> iio   interface, which would make more sense, you are right,
>>>>>>>> but I
>>>>>>>> simply  don't have the time to merge it in to iio.
>>>>>>>>
>>>>>>>> It doesn't use an input interface either but I don't see a
>>>>>>>> good   place for an accelerometer that uses sysfs only.
>>>>>>>>
>>>>>>>> It works well, is a relatively recent chip and a clean
>>>>>>>> dirver.  But  this is all I can provide.
>>>>>>>
>>>>>>> As a person who works on the user-space interaction of those
>>>>>>> with   desktops [1]: Urgh.
>>>>>>>
>>>>>>> I already have 3 (probably 4) types of accelerometers to
>>>>>>> contend  with,  I'm not fond of adding yet another type.
>>>>>>>
>>>>>>> Is there any way to get this hardware working outside the SoCs
>>>>>>> it's  designed for (say, a device with I2C like a Raspberry
>>>>>>> Pi),  so that a  kind soul could handle getting this using the
>>>>>>> right  interfaces?
>>>>>>>
>>>>>>
>>>>>> It works on basically any SoC and is in no way limited in this
>>>>>> regard. Sure, userspace has to expicitely support it and I hear
>>>>>> you.  Using the iio interface would make more sense. I can only
>>>>>> say I'd  love to have the time to move this driver over. I'm
>>>>>> very sorry.
>>>>>
>>>>> How can we get the hardware for somebody to use on their own
>>>>> laptops/embedded boards to implement this driver?
>>>>>
>>>>
>>>> It's connected over I2C. If the included documentation is not clear
>>>> please tell me what exacly. Thanks!
>>>
>>>
>>> I'll ask the question a different way: can you please give the address
>>> of a shop where that hardware is available?
>>>
>>
>> there is
>> http://www.freescale.com/webapp/sps/site/prod_summary.jsp?code=RDMMA865x&lang_cd=
>> and http://linux-sunxi.org/Inet_K970 for example. I think I saw Android
>> devices with it too, and I would guess it would be used more often if it
>> were in linux.
> 
> I am going to say again (in a different way) what Bastien said. In its
> current form, even in drivers/misc, this is a NACK for me (v1, v2, v3
> & v4).
> 
> Putting a driver in Linux means we have to support it forever, and
> definitively, nobody will use an accelerometer in Android if it is in
> drivers/misc.
> Android requires drivers to follow the IIO protocol. Period.
> So having your own will not help android, it will just be a burden.
> 
> The sysfs you are proposing seems simple enough, but we can not afford
> having a 3rd custom way of relying the accelerometer information in
> the Linux tree (they were first handled in input, then IIO, then a
> custom sysfs).
> 
> If you really want to have the driver in the tree, I won't be opposed
> if you put in under staging. This way, you can break it whenever you
> want and people won't rely on it. And then, we can use your driver as
> a base to port it to IIO.

That seems reasonable. I have prepared a (little more cleaned up) v5 of
the patch and moved it to staging, with a TODO file containing also the
current documentation. I hope to be able to do the iio integration
sometime "soon", but this could speed things up.

> 
> Sorry for being rude, but I am starting to get tired of people saying
> that they don't have the time to follow what the reviewers said. You
> obviously spent some time polishing this driver, why not making it
> right from its first inclusion in the tree?

That's totally fine. Of course iio is the way to go. I had the driver
before I really knew iio and this was lazyness (when I found
Documentation/ABI/testing), plus the desire to publish it before the
chip itself is deprecated.

> 
> Cheers,
> Benjamin
> 
>>
>> Please refer to v4 of the patch for different questions. thanks!
>> --
>> To unsubscribe from this list: send the line "unsubscribe linux-input" in
>> the body of a message to majordomo@vger.kernel.org
>> More majordomo info at  http://vger.kernel.org/majordomo-info.html