Re: [RFC PATCH 2/7] iio:ABI docs: Avoid repitition of triggerX/sampling_frequency

[Jonathan Cameron <jic23@kernel.org>]

On Sun, 7 Mar 2021 10:40:18 +0100
Lars-Peter Clausen <lars@metafoo.de> wrote:

> On 1/17/21 4:38 PM, Jonathan Cameron wrote:
> > From: Jonathan Cameron <Jonathan.Cameron@huawei.com>
> >
> > The timer-stm32 provided a little more specific information than the main
> > docs about the value of 0 corresponding to stopping sampling.  Given that
> > this makes sense in general, move that statement over to the main docs
> > and drop the version in sysfs-bus-iio-timer-stm32
> >
> > Fixes
> > $ scripts/get_abi.pl validate
> > /sys/bus/iio/devices/triggerX/sampling_frequency is defined 2 times:  ./Documentation/ABI/testing/sysfs-bus-iio-timer-stm32:92  ./Documentation/ABI/testing/sysfs-bus-iio:45
> >
> > Cc: Fabrice Gasnier <fabrice.gasnier@st.com>
> > Reported-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
> > Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
> > ---
> >   Documentation/ABI/testing/sysfs-bus-iio             | 2 ++
> >   Documentation/ABI/testing/sysfs-bus-iio-timer-stm32 | 8 --------
> >   2 files changed, 2 insertions(+), 8 deletions(-)
> >
> > diff --git a/Documentation/ABI/testing/sysfs-bus-iio b/Documentation/ABI/testing/sysfs-bus-iio
> > index d2dd9cc280f9..9b5ceb22363d 100644
> > --- a/Documentation/ABI/testing/sysfs-bus-iio
> > +++ b/Documentation/ABI/testing/sysfs-bus-iio
> > @@ -55,6 +55,8 @@ Description:
> >   		direct access interfaces, it may be found in any of the
> >   		relevant directories.  If it affects all of the above
> >   		then it is to be found in the base device directory.
> > +		Note a value 0 where supported, will correspond to sampling
> > +		stopping.  
> 
> Hm, I don't like this. 0 should be an invalid value for the standard 
> ABI. Enabling/disabling of the trigger should be controlled by whether 
> there are any active trigger consumers.
> 
> The stm32-timer-trigger implements a non-standard ABI and I think we 
> need to document this explicitly. The driver does not have a set_state 
> callback. So the frequency property is used to enable/disable the 
> trigger, but as said above, for standard compliant triggers that should 
> not be the case.
> 
Good point.  Ideally we'd also fix the stm32-timer-trigger to have
an explicit enable / disable (even if that works by setting the frequency
to 0 under the hook)

For now I've dropped the patch until this is resolved.

Thanks,

Jonathan