User-space API for accelerometer(s)?

User-space API for accelerometer(s)?

[Bastien Nocera]

Hey,

I'm trying to integrate the accelerometer of the Lenovo Yoga into GNOME
and a modern desktop. The accelerometer is exported through an IIO
device as per:
https://github.com/pfps/yoga-laptop/blob/master/sensors/orientation.c

We already have some integration for accelerometers in udev/systemd and
GNOME, and they rely on the accelerometer being an input device, being
tagged with the ID_INPUT_ACCELEROMETER and sending out a kevent/uevent
when a major orientation change took place. This is handled by:
http://cgit.freedesktop.org/systemd/systemd/tree/src/udev/accelerometer/accelerometer.c
for the WeTab/Pegatron devices for example.

So, my question regarding the IIO user-space API is:
is it possible to make the IIO accelerometer send out a kevent when the
orientation changes in a major way (using triggers?) or does user-space
need to poll the device instead?

If the former, I intend on writing a small helper to set up the
accelerometer, and a helper similar to the one already in the udev tree.

If the latter, I'll probably write a long-running helper, which would
offer the same interface as the one used by the Pegatron accelerometer
driver, so that we don't need to make any more changes to user-space.

Cheers

Re: User-space API for accelerometer(s)?

[Reyad Attiyat]

Hello Bastien Nocera,

I'm not the best person to answer your questions but I will try and
help. The best people to talk to would be Jonathan Cameron, the IIO
maintainer, and Srinivas Pandruvada, the author of the hid-sensor-hub
device drivers. I have CC'ed them for you to hopefully get a better
response than what I can offer.

Your device uses the same sensor hub, over usb, as my device does. I
have a Microsoft Surface and it also uses many of the sensors found on
the Lenovo Yoga.

> So, my question regarding the IIO user-space API is:
> is it possible to make the IIO accelerometer send out a kevent when the
> orientation changes in a major way (using triggers?) or does user-space
> need to poll the device instead?

Of course it's possible but this does not occur in the current hid
sensor hub drivers, from my understanding of the code. You should
probably check out the current documentation on the IIO sysfs user
interface:
https://www.kernel.org/doc/Documentation/ABI/testing/sysfs-bus-iio

These outline the possible attributes sysfs offers. As of now the only
ones that are found on the hid sensor hub devices is _raw as of 3.16
and _scale, _offset, _sampling_frequency, _hysteresis. The raw
attribute lets you read the data that is at the top of the iio buffer.
As you can see in the documentation there are many events support by
other drivers that could possible be implemented by the hid-sensor-hub
devices.

IIO devices use a buffer that have a fixed size. It can be enabled
with the sysfs interface. An example of this can be found in the
kernel (drivers/staging/iio/iio_simple_dummy_buffer.c) there is also
additional documentation there as well as dummy drivers.

To use these interfaces in user space, or in the context of GNOME
desktop, I think you would need to poll the iio buffer for new data
and calculate the changes that way. The iio subsystem does support
events but you would need to add these to the hid-sensor-hub devices
and do the calculations for each event. This would allow you to poll
for these iio events instead of polling the buffer.

Hope this helps,
Reyad Attiyat

Re: User-space API for accelerometer(s)?

[Srinivas Pandruvada]

On 06/18/2014 04:31 PM, Reyad Attiyat wrote:
> Hello Bastien Nocera,
>
> I'm not the best person to answer your questions but I will try and
> help. The best people to talk to would be Jonathan Cameron, the IIO
> maintainer, and Srinivas Pandruvada, the author of the hid-sensor-hub
> device drivers. I have CC'ed them for you to hopefully get a better
> response than what I can offer.
>
> Your device uses the same sensor hub, over usb, as my device does. I
> have a Microsoft Surface and it also uses many of the sensors found on
> the Lenovo Yoga.
>
>> So, my question regarding the IIO user-space API is:
>> is it possible to make the IIO accelerometer send out a kevent when the
>> orientation changes in a major way (using triggers?) or does user-space
>> need to poll the device instead?

You can check a program called generic_buffer.c in 
"drivers/staging/iio/Documentation". I have used this as a reference to 
port to Android.
You don't need to poll, you can also check 
"https://github.com/pfps/yoga-laptop/sensors " developed by Peter F. 
Patel-Schneider.

Thanks,
Srinivas

>
> Of course it's possible but this does not occur in the current hid
> sensor hub drivers, from my understanding of the code. You should
> probably check out the current documentation on the IIO sysfs user
> interface:
> https://www.kernel.org/doc/Documentation/ABI/testing/sysfs-bus-iio
>
> These outline the possible attributes sysfs offers. As of now the only
> ones that are found on the hid sensor hub devices is _raw as of 3.16
> and _scale, _offset, _sampling_frequency, _hysteresis. The raw
> attribute lets you read the data that is at the top of the iio buffer.
> As you can see in the documentation there are many events support by
> other drivers that could possible be implemented by the hid-sensor-hub
> devices.
>
> IIO devices use a buffer that have a fixed size. It can be enabled
> with the sysfs interface. An example of this can be found in the
> kernel (drivers/staging/iio/iio_simple_dummy_buffer.c) there is also
> additional documentation there as well as dummy drivers.
>
> To use these interfaces in user space, or in the context of GNOME
> desktop, I think you would need to poll the iio buffer for new data
> and calculate the changes that way. The iio subsystem does support
> events but you would need to add these to the hid-sensor-hub devices
> and do the calculations for each event. This would allow you to poll
> for these iio events instead of polling the buffer.
>
> Hope this helps,
> Reyad Attiyat
>

Re: User-space API for accelerometer(s)?

[Bastien Nocera]

On Wed, 2014-06-18 at 16:45 -0700, Srinivas Pandruvada wrote:
> On 06/18/2014 04:31 PM, Reyad Attiyat wrote:
> > Hello Bastien Nocera,
> >
> > I'm not the best person to answer your questions but I will try and
> > help. The best people to talk to would be Jonathan Cameron, the IIO
> > maintainer, and Srinivas Pandruvada, the author of the hid-sensor-hub
> > device drivers. I have CC'ed them for you to hopefully get a better
> > response than what I can offer.
> >
> > Your device uses the same sensor hub, over usb, as my device does. I
> > have a Microsoft Surface and it also uses many of the sensors found on
> > the Lenovo Yoga.
> >
> >> So, my question regarding the IIO user-space API is:
> >> is it possible to make the IIO accelerometer send out a kevent when the
> >> orientation changes in a major way (using triggers?) or does user-space
> >> need to poll the device instead?
> 
> You can check a program called generic_buffer.c in 
> "drivers/staging/iio/Documentation". I have used this as a reference to 
> port to Android.
> You don't need to poll, you can also check 
> "https://github.com/pfps/yoga-laptop/sensors " developed by Peter F. 
> Patel-Schneider.

I've read this code, and in fact, I mentioned it in my original mail. I
fail to see how this code isn't polling. It's also much more complicated
than doing the same thing for a evdev accelerometer.

Re: User-space API for accelerometer(s)?

[Jonathan Cameron]

On 19/06/14 12:20, Bastien Nocera wrote:
> On Wed, 2014-06-18 at 16:45 -0700, Srinivas Pandruvada wrote:
>> On 06/18/2014 04:31 PM, Reyad Attiyat wrote:
>>> Hello Bastien Nocera,
>>>
>>> I'm not the best person to answer your questions but I will try and
>>> help. The best people to talk to would be Jonathan Cameron, the IIO
>>> maintainer, and Srinivas Pandruvada, the author of the hid-sensor-hub
>>> device drivers. I have CC'ed them for you to hopefully get a better
>>> response than what I can offer.
>>>
>>> Your device uses the same sensor hub, over usb, as my device does. I
>>> have a Microsoft Surface and it also uses many of the sensors found on
>>> the Lenovo Yoga.
>>>
>>>> So, my question regarding the IIO user-space API is:
>>>> is it possible to make the IIO accelerometer send out a kevent when the
>>>> orientation changes in a major way (using triggers?) or does user-space
>>>> need to poll the device instead?
>>
>> You can check a program called generic_buffer.c in
>> "drivers/staging/iio/Documentation". I have used this as a reference to
>> port to Android.
>> You don't need to poll, you can also check
>> "https://github.com/pfps/yoga-laptop/sensors " developed by Peter F.
>> Patel-Schneider.
>
> I've read this code, and in fact, I mentioned it in my original mail. I
> fail to see how this code isn't polling. It's also much more complicated
> than doing the same thing for a evdev accelerometer.
>
Just to throw it in there.  There is an out of tree bridge driver from
IIO to input.  It's only out of tree because I haven't had a chance to
tidy it up (anyone else is welcome to take this on if they like!)
Google for iio_input.c to find it.

The intent of that was to allow general accelerometer drivers and similar
in IIO to work in conjunction with iio-input to provide input style interfaces.
This came about after previous debates on where the 'right' place for
accelerometers was in the kernel. I believe that at least in principle,
Dmitry was happy with this concept.

Jonathan

Re: User-space API for accelerometer(s)?

[Bastien Nocera]

On Sat, 2014-06-21 at 13:01 +0100, Jonathan Cameron wrote:
> On 19/06/14 12:20, Bastien Nocera wrote:
> > On Wed, 2014-06-18 at 16:45 -0700, Srinivas Pandruvada wrote:
> >> On 06/18/2014 04:31 PM, Reyad Attiyat wrote:
> >>> Hello Bastien Nocera,
> >>>
> >>> I'm not the best person to answer your questions but I will try and
> >>> help. The best people to talk to would be Jonathan Cameron, the IIO
> >>> maintainer, and Srinivas Pandruvada, the author of the hid-sensor-hub
> >>> device drivers. I have CC'ed them for you to hopefully get a better
> >>> response than what I can offer.
> >>>
> >>> Your device uses the same sensor hub, over usb, as my device does. I
> >>> have a Microsoft Surface and it also uses many of the sensors found on
> >>> the Lenovo Yoga.
> >>>
> >>>> So, my question regarding the IIO user-space API is:
> >>>> is it possible to make the IIO accelerometer send out a kevent when the
> >>>> orientation changes in a major way (using triggers?) or does user-space
> >>>> need to poll the device instead?
> >>
> >> You can check a program called generic_buffer.c in
> >> "drivers/staging/iio/Documentation". I have used this as a reference to
> >> port to Android.
> >> You don't need to poll, you can also check
> >> "https://github.com/pfps/yoga-laptop/sensors " developed by Peter F.
> >> Patel-Schneider.
> >
> > I've read this code, and in fact, I mentioned it in my original mail. I
> > fail to see how this code isn't polling. It's also much more complicated
> > than doing the same thing for a evdev accelerometer.
> >
> Just to throw it in there.  There is an out of tree bridge driver from
> IIO to input.  It's only out of tree because I haven't had a chance to
> tidy it up (anyone else is welcome to take this on if they like!)
> Google for iio_input.c to find it.

Found it, thanks :)

http://thread.gmane.org/gmane.linux.kernel.iio/4464/

(By the way, "iio_input", not "iio_snoop")

> The intent of that was to allow general accelerometer drivers and similar
> in IIO to work in conjunction with iio-input to provide input style interfaces.
> This came about after previous debates on where the 'right' place for
> accelerometers was in the kernel. I believe that at least in principle,
> Dmitry was happy with this concept.

Cool. The only question left would be whether this could throw a kevent
when the orientation changes in a major way. This is a firmware feature
in the WeTab/Pegatron machine, and is useful because it avoids the need
to have the accelerometer constantly opened when not needed (saving
battery by avoiding wakeups).

Would it be worth adding such a feature inside the iio-input driver? I
would have preferred it if the hardware sent those events, but I'd be
happy just the same if the driver did it. The backing IIO device is
always opened and processing data, it seems, whether or not the backing
input device is opened in user-space or not.

Cheers

Re: User-space API for accelerometer(s)?

[Srinivas Pandruvada]

On 06/21/2014 05:37 AM, Bastien Nocera wrote:
> On Sat, 2014-06-21 at 13:01 +0100, Jonathan Cameron wrote:
>> On 19/06/14 12:20, Bastien Nocera wrote:
>>> On Wed, 2014-06-18 at 16:45 -0700, Srinivas Pandruvada wrote:
>>>> On 06/18/2014 04:31 PM, Reyad Attiyat wrote:
>>>>> Hello Bastien Nocera,
>>>>>
>>>>> I'm not the best person to answer your questions but I will try and
>>>>> help. The best people to talk to would be Jonathan Cameron, the IIO
>>>>> maintainer, and Srinivas Pandruvada, the author of the hid-sensor-hub
>>>>> device drivers. I have CC'ed them for you to hopefully get a better
>>>>> response than what I can offer.
>>>>>
>>>>> Your device uses the same sensor hub, over usb, as my device does. I
>>>>> have a Microsoft Surface and it also uses many of the sensors found on
>>>>> the Lenovo Yoga.
>>>>>
>>>>>> So, my question regarding the IIO user-space API is:
>>>>>> is it possible to make the IIO accelerometer send out a kevent when the
>>>>>> orientation changes in a major way (using triggers?) or does user-space
>>>>>> need to poll the device instead?
>>>> You can check a program called generic_buffer.c in
>>>> "drivers/staging/iio/Documentation". I have used this as a reference to
>>>> port to Android.
>>>> You don't need to poll, you can also check
>>>> "https://github.com/pfps/yoga-laptop/sensors " developed by Peter F.
>>>> Patel-Schneider.
>>> I've read this code, and in fact, I mentioned it in my original mail. I
>>> fail to see how this code isn't polling. It's also much more complicated
>>> than doing the same thing for a evdev accelerometer.
>>>
>> Just to throw it in there.  There is an out of tree bridge driver from
>> IIO to input.  It's only out of tree because I haven't had a chance to
>> tidy it up (anyone else is welcome to take this on if they like!)
>> Google for iio_input.c to find it.
> Found it, thanks :)
>
> http://thread.gmane.org/gmane.linux.kernel.iio/4464/
>
> (By the way, "iio_input", not "iio_snoop")
>
>> The intent of that was to allow general accelerometer drivers and similar
>> in IIO to work in conjunction with iio-input to provide input style interfaces.
>> This came about after previous debates on where the 'right' place for
>> accelerometers was in the kernel. I believe that at least in principle,
>> Dmitry was happy with this concept.
> Cool. The only question left would be whether this could throw a kevent
> when the orientation changes in a major way. This is a firmware feature
> in the WeTab/Pegatron machine, and is useful because it avoids the need
> to have the accelerometer constantly opened when not needed (saving
> battery by avoiding wakeups).
That will happen with sensor hubs, You can set the hysteresis to not to get
awakened unless the change is significant.

Thanks,
Srinivas

> Would it be worth adding such a feature inside the iio-input driver? I
> would have preferred it if the hardware sent those events, but I'd be
> happy just the same if the driver did it. The backing IIO device is
> always opened and processing data, it seems, whether or not the backing
> input device is opened in user-space or not.
>
> Cheers
>
>

Re: User-space API for accelerometer(s)?

[Bastien Nocera]

Hey again Jonathan,

On Sat, 2014-06-21 at 13:01 +0100, Jonathan Cameron wrote:
<snip>
> Just to throw it in there.  There is an out of tree bridge driver from
> IIO to input.  It's only out of tree because I haven't had a chance to
> tidy it up (anyone else is welcome to take this on if they like!)
> Google for iio_input.c to find it.
> 
> The intent of that was to allow general accelerometer drivers and similar
> in IIO to work in conjunction with iio-input to provide input style interfaces.
> This came about after previous debates on where the 'right' place for
> accelerometers was in the kernel. I believe that at least in principle,
> Dmitry was happy with this concept.

After updating forward-porting the driver so that it runs on a more
recent version of the kernel, I tried to get it running.

I'm guessing that you expected the iio_input driver to be instantiated
by a board specific file. Is there any way to have it generically try
out all the IIO devices, similarly to pci_register_driver()? Or should I
do that in the hid-sensor-accel driver?

Cheers

Re: User-space API for accelerometer(s)?

[Jonathan Cameron]

On 01/07/14 13:10, Bastien Nocera wrote:
> Hey again Jonathan,
>
> On Sat, 2014-06-21 at 13:01 +0100, Jonathan Cameron wrote:
> <snip>
>> Just to throw it in there.  There is an out of tree bridge driver from
>> IIO to input.  It's only out of tree because I haven't had a chance to
>> tidy it up (anyone else is welcome to take this on if they like!)
>> Google for iio_input.c to find it.
>>
>> The intent of that was to allow general accelerometer drivers and similar
>> in IIO to work in conjunction with iio-input to provide input style interfaces.
>> This came about after previous debates on where the 'right' place for
>> accelerometers was in the kernel. I believe that at least in principle,
>> Dmitry was happy with this concept.
>
> After updating forward-porting the driver so that it runs on a more
> recent version of the kernel, I tried to get it running.
>
> I'm guessing that you expected the iio_input driver to be instantiated
> by a board specific file. Is there any way to have it generically try
> out all the IIO devices, similarly to pci_register_driver()? Or should I
> do that in the hid-sensor-accel driver?
At the moment we only have support to instantiate it via such a board specific
file. We probably want to be a little careful about how to add more generic
means for creating such bindings..

Simply generically trying all IIO drivers isn't the way to go as there is
no obvious way of deciding what it makes sense to bind to on a given machine.

Perhaps something closer to the way you can instantiate i2c devices from
userspace, or perhaps we need to consider something else such as configfs for
creating such bindings... (cc'd Lars for comments on whether that makes sense...)

We did also have a uinput based approach at one point but it was fairly clunky.
That took data from iio buffers and pushed it back into the kernel via inputs
userspace driver support. Not particularly nice but I thought I'd best mention
it!

J
> Cheers
>

Re: User-space API for accelerometer(s)?

[Lars-Peter Clausen]

On 07/03/2014 07:51 PM, Jonathan Cameron wrote:
> On 01/07/14 13:10, Bastien Nocera wrote:
>> Hey again Jonathan,
>>
>> On Sat, 2014-06-21 at 13:01 +0100, Jonathan Cameron wrote:
>> <snip>
>>> Just to throw it in there.  There is an out of tree bridge driver from
>>> IIO to input.  It's only out of tree because I haven't had a chance to
>>> tidy it up (anyone else is welcome to take this on if they like!)
>>> Google for iio_input.c to find it.
>>>
>>> The intent of that was to allow general accelerometer drivers and similar
>>> in IIO to work in conjunction with iio-input to provide input style interfaces.
>>> This came about after previous debates on where the 'right' place for
>>> accelerometers was in the kernel. I believe that at least in principle,
>>> Dmitry was happy with this concept.
>>
>> After updating forward-porting the driver so that it runs on a more
>> recent version of the kernel, I tried to get it running.
>>
>> I'm guessing that you expected the iio_input driver to be instantiated
>> by a board specific file. Is there any way to have it generically try
>> out all the IIO devices, similarly to pci_register_driver()? Or should I
>> do that in the hid-sensor-accel driver?
> At the moment we only have support to instantiate it via such a board specific
> file. We probably want to be a little careful about how to add more generic
> means for creating such bindings..
>
> Simply generically trying all IIO drivers isn't the way to go as there is
> no obvious way of deciding what it makes sense to bind to on a given machine.
>
> Perhaps something closer to the way you can instantiate i2c devices from
> userspace, or perhaps we need to consider something else such as configfs for
> creating such bindings... (cc'd Lars for comments on whether that makes sense...)

Yes, configfs is what came to my mind after reading the first few lines of 
Bastien's mail. Exposing a IIO device as something other than a IIO device is a 
policy decision. And policy decisions should preferably be made by userspace. 
But yea, we have to be careful with the ABI.

- Lars

Re: User-space API for accelerometer(s)?

[Bastien Nocera]

On Thu, 2014-07-03 at 18:51 +0100, Jonathan Cameron wrote:
> On 01/07/14 13:10, Bastien Nocera wrote:
> > Hey again Jonathan,
> >
> > On Sat, 2014-06-21 at 13:01 +0100, Jonathan Cameron wrote:
> > <snip>
> >> Just to throw it in there.  There is an out of tree bridge driver from
> >> IIO to input.  It's only out of tree because I haven't had a chance to
> >> tidy it up (anyone else is welcome to take this on if they like!)
> >> Google for iio_input.c to find it.
> >>
> >> The intent of that was to allow general accelerometer drivers and similar
> >> in IIO to work in conjunction with iio-input to provide input style interfaces.
> >> This came about after previous debates on where the 'right' place for
> >> accelerometers was in the kernel. I believe that at least in principle,
> >> Dmitry was happy with this concept.
> >
> > After updating forward-porting the driver so that it runs on a more
> > recent version of the kernel, I tried to get it running.
> >
> > I'm guessing that you expected the iio_input driver to be instantiated
> > by a board specific file. Is there any way to have it generically try
> > out all the IIO devices, similarly to pci_register_driver()? Or should I
> > do that in the hid-sensor-accel driver?
> At the moment we only have support to instantiate it via such a board specific
> file. We probably want to be a little careful about how to add more generic
> means for creating such bindings..
> 
> Simply generically trying all IIO drivers isn't the way to go as there is
> no obvious way of deciding what it makes sense to bind to on a given machine.
> 
> Perhaps something closer to the way you can instantiate i2c devices from
> userspace, or perhaps we need to consider something else such as configfs for
> creating such bindings... (cc'd Lars for comments on whether that makes sense...)

Are there cases where we have a hid-sensor that we wouldn't want to
expose through the input layer?

> We did also have a uinput based approach at one point but it was fairly clunky.
> That took data from iio buffers and pushed it back into the kernel via inputs
> userspace driver support. Not particularly nice but I thought I'd best mention
> it!

That's actually something that I looked at and discussed with Benjamin.
And it currently looks like the best possible solution in the short-term
(I will only have the machine for a couple more weeks...).

Cheers

Re: User-space API for accelerometer(s)?

[Bastien Nocera]

On Thu, 2014-07-03 at 18:51 +0100, Jonathan Cameron wrote:
<snip>
> We did also have a uinput based approach at one point but it was fairly clunky.
> That took data from iio buffers and pushed it back into the kernel via inputs
> userspace driver support. Not particularly nice but I thought I'd best mention
> it!

This is how I went in the end:
https://github.com/hadess/iio-sensor-proxy/

It's still far too resource hungry compared to the amount of work it's
doing (1 full percent of CPU!), and the rotation is too sensitive.

But, along with this systemd patch:
http://cgit.freedesktop.org/systemd/systemd/commit/?id=a545c6e1aa31b4d7e80c9d3609d9fc4fc9921498

It works out of the box with GNOME's orientation plugin.

I think that the code could also do with a bit more cleanup.

Please drop me a mail privately or file an issue if you can test this on
various systems (other models of Yoga and the Surface would be nice to
test), and I'll add it to the list of tested devices.

Cheers

Re: User-space API for accelerometer(s)?

[Peter Meerwald]

Hello Bastian,

> > We did also have a uinput based approach at one point but it was fairly clunky.
> > That took data from iio buffers and pushed it back into the kernel via inputs
> > userspace driver support. Not particularly nice but I thought I'd best mention
> > it!
> 
> This is how I went in the end:
> https://github.com/hadess/iio-sensor-proxy/
> 
> It's still far too resource hungry compared to the amount of work it's
> doing (1 full percent of CPU!), and the rotation is too sensitive.

if I read the code correctly, prepare_output() configures the trigger, 
enables the buffer, then performs one read, before undoing everything FOR 
EACH SAMPLE

this is probably not the way it should be done; I'd suggest to set up the 
IIO buffer and then poll() or block on /dev/iio:deviceX -- however, this 
uses the IIO device exclusively (might be an issue)

regards, p.

-- 

Peter Meerwald
+43-664-2444418 (mobile)

Re: User-space API for accelerometer(s)?

[Bastien Nocera]

Hey Peter,

On Wed, 2014-07-09 at 16:54 +0200, Peter Meerwald wrote:
> Hello Bastian,
> 
> > > We did also have a uinput based approach at one point but it was fairly clunky.
> > > That took data from iio buffers and pushed it back into the kernel via inputs
> > > userspace driver support. Not particularly nice but I thought I'd best mention
> > > it!
> > 
> > This is how I went in the end:
> > https://github.com/hadess/iio-sensor-proxy/
> > 
> > It's still far too resource hungry compared to the amount of work it's
> > doing (1 full percent of CPU!), and the rotation is too sensitive.
> 
> if I read the code correctly, prepare_output() configures the trigger, 
> enables the buffer, then performs one read, before undoing everything FOR 
> EACH SAMPLE

I based that code off Peter F. Patel-Schneider, and his yoga utilities,
and he probably got the idea from the generic_buffer.c example. It's not
clear what needs to be done there.

> this is probably not the way it should be done; I'd suggest to set up the 
> IIO buffer and then poll() or block on /dev/iio:deviceX -- however, this 
> uses the IIO device exclusively (might be an issue)

I've now pushed a version that will do the setup once, and open/close
the iio device when needed, in the timeout.
 The daemon still shows up in top, but I couldn't make it register
enough activity to show up in sysprof, so I'm guessing that top just
isn't measuring the CPU usage accurately.

I still have to fix the overeagerness to switch orientations, and I'll
be done for now I think.

Cheers

Re: User-space API for accelerometer(s)?

[Peter F. Patel-Schneider]

Yes, my code derives from generic-buffer, at least in part, as I first had to 
fix problems in generic-buffer, and then get problems fixed in the sensor hub 
drivers.  For quite some time the only way to get the code to work at all was 
to do everything each time, so I just kept doing it that way.

There appears to be lots of improvements in the drivers so a better solution 
is certainly possible now.

peter


On 07/09/2014 03:16 PM, Bastien Nocera wrote:
> Hey Peter,
>
> On Wed, 2014-07-09 at 16:54 +0200, Peter Meerwald wrote:
>> Hello Bastian,
>>
>>>> We did also have a uinput based approach at one point but it was fairly clunky.
>>>> That took data from iio buffers and pushed it back into the kernel via inputs
>>>> userspace driver support. Not particularly nice but I thought I'd best mention
>>>> it!
>>>
>>> This is how I went in the end:
>>> https://github.com/hadess/iio-sensor-proxy/
>>>
>>> It's still far too resource hungry compared to the amount of work it's
>>> doing (1 full percent of CPU!), and the rotation is too sensitive.
>>
>> if I read the code correctly, prepare_output() configures the trigger,
>> enables the buffer, then performs one read, before undoing everything FOR
>> EACH SAMPLE
>
> I based that code off Peter F. Patel-Schneider, and his yoga utilities,
> and he probably got the idea from the generic_buffer.c example. It's not
> clear what needs to be done there.
>
>> this is probably not the way it should be done; I'd suggest to set up the
>> IIO buffer and then poll() or block on /dev/iio:deviceX -- however, this
>> uses the IIO device exclusively (might be an issue)
>
> I've now pushed a version that will do the setup once, and open/close
> the iio device when needed, in the timeout.
>   The daemon still shows up in top, but I couldn't make it register
> enough activity to show up in sysprof, so I'm guessing that top just
> isn't measuring the CPU usage accurately.
>
> I still have to fix the overeagerness to switch orientations, and I'll
> be done for now I think.
>
> Cheers
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-iio" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>

Re: User-space API for accelerometer(s)?

[Srinivas Pandruvada]

I suggest to look at https://01.org/android-ia/downloads. Under 
device/intel, you can see user space HAL for sensor hubs.

Thanks,
Srinivas


On 07/09/2014 06:38 PM, Peter F. Patel-Schneider wrote:
> Yes, my code derives from generic-buffer, at least in part, as I first
> had to fix problems in generic-buffer, and then get problems fixed in
> the sensor hub drivers.  For quite some time the only way to get the
> code to work at all was to do everything each time, so I just kept doing
> it that way.
>
> There appears to be lots of improvements in the drivers so a better
> solution is certainly possible now.
>
> peter
>
>
> On 07/09/2014 03:16 PM, Bastien Nocera wrote:
>> Hey Peter,
>>
>> On Wed, 2014-07-09 at 16:54 +0200, Peter Meerwald wrote:
>>> Hello Bastian,
>>>
>>>>> We did also have a uinput based approach at one point but it was
>>>>> fairly clunky.
>>>>> That took data from iio buffers and pushed it back into the kernel
>>>>> via inputs
>>>>> userspace driver support. Not particularly nice but I thought I'd
>>>>> best mention
>>>>> it!
>>>>
>>>> This is how I went in the end:
>>>> https://github.com/hadess/iio-sensor-proxy/
>>>>
>>>> It's still far too resource hungry compared to the amount of work it's
>>>> doing (1 full percent of CPU!), and the rotation is too sensitive.
>>>
>>> if I read the code correctly, prepare_output() configures the trigger,
>>> enables the buffer, then performs one read, before undoing everything
>>> FOR
>>> EACH SAMPLE
>>
>> I based that code off Peter F. Patel-Schneider, and his yoga utilities,
>> and he probably got the idea from the generic_buffer.c example. It's not
>> clear what needs to be done there.
>>
>>> this is probably not the way it should be done; I'd suggest to set up
>>> the
>>> IIO buffer and then poll() or block on /dev/iio:deviceX -- however, this
>>> uses the IIO device exclusively (might be an issue)
>>
>> I've now pushed a version that will do the setup once, and open/close
>> the iio device when needed, in the timeout.
>>   The daemon still shows up in top, but I couldn't make it register
>> enough activity to show up in sysprof, so I'm guessing that top just
>> isn't measuring the CPU usage accurately.
>>
>> I still have to fix the overeagerness to switch orientations, and I'll
>> be done for now I think.
>>
>> Cheers
>>
>> --
>> To unsubscribe from this list: send the line "unsubscribe linux-iio" in
>> the body of a message to majordomo@vger.kernel.org
>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>>
>

Re: User-space API for accelerometer(s)?

[Bastien Nocera]

On Thu, 2014-07-10 at 08:04 -0700, Srinivas Pandruvada wrote:
> I suggest to look at https://01.org/android-ia/downloads. Under 
> device/intel, you can see user space HAL for sensor hubs.

I can't find it anywhere on those pages. There only seems to be
installable images, with no source code.

Re: User-space API for accelerometer(s)?

[Srinivas Pandruvada]

On 07/23/2014 05:19 AM, Bastien Nocera wrote:
> On Thu, 2014-07-10 at 08:04 -0700, Srinivas Pandruvada wrote:
>> I suggest to look at https://01.org/android-ia/downloads. Under
>> device/intel, you can see user space HAL for sensor hubs.
> I can't find it anywhere on those pages. There only seems to be
> installable images, with no source code.
>

You can see instruction here to get source code. You can see source in 
device/intel folder.

https://01.org/android-ia/downloads/android-4.4.2r1-ia1

Thanks,
Srinivas

Re: User-space API for accelerometer(s)?

[Bastien Nocera]

On Wed, 2014-07-23 at 06:02 -0700, Srinivas Pandruvada wrote:
> On 07/23/2014 05:19 AM, Bastien Nocera wrote:
> > On Thu, 2014-07-10 at 08:04 -0700, Srinivas Pandruvada wrote:
> >> I suggest to look at https://01.org/android-ia/downloads. Under
> >> device/intel, you can see user space HAL for sensor hubs.
> > I can't find it anywhere on those pages. There only seems to be
> > installable images, with no source code.
> >
> 
> You can see instruction here to get source code. You can see source in 
> device/intel folder.
> 
> https://01.org/android-ia/downloads/android-4.4.2r1-ia1

Right, I found this:
https://github.com/android-ia/device_intel/blob/release/android-4.1.1_r1-ia0/pc_std/libsensors/SensorConfig.h

Which seems to have a number of bugs (defines "GRAVITY" but actually
uses "GRAVITY_EARTH" in some of the formulas), and doesn't make it clear
which one of the sensors each of the sections are for.

In any case, my problem was simply one of threshold. For reference, I've
also downloaded Microsoft's documentation for hardware manufacturers:
http://msdn.microsoft.com/en-us/library/windows/hardware/dn613934(v=vs.85).aspx

Cheers