Re: [PATCH] iio: event_monitor: Enable events before monitoring

[Jonathan Cameron <jic23@kernel.org>]

On Thu, 4 Mar 2021 21:48:25 +0100
Linus Walleij <linus.walleij@linaro.org> wrote:

> On Thu, Mar 4, 2021 at 5:00 PM Bastien Nocera <hadess@hadess.net> wrote:
> 
> > If I'm not mistaken, "-a" does that for the iio_generic_buffer tool.  
> 
> Yeah I implemented that, and I thought about doing the same here
> but ... the name of the tool sort of announce that one want to
> listen to all events so I thought it should just default-enable
> all of them in this case.
> 
> > Maybe moving enable_disable_all_channels() to a common location and
> > using that would cut down on the duplicated code?  
> 
> The event enablement is slightly different, the generic buffer
> turns on various channels, which is conceptually different
> from various events, but let's see what Jonathan says.
> 
> We are sharing most of the code already in the iio-utils.c
> but I can try to break out more if it doesn't get to abstract.

Sadly this doesn't work for many devices.
It is a common thing for hardware to only support a much smaller
set of event monitoring registers / threshold detectors than the
number of channels.  In many cases we handle that by working on
a fifo basis.  So what this will do is enable a bunch of events
which will then be replaced by later events - end result some
random event will be enabled (or maybe 2 of them across N channels)

Not intuitive at all :(

I'm fine with it being controlled by a parameter though if
that works for you.  Docs should explain it doesn't always
result in all events being enabled however if the hardware
is not capable of doing that.

Jonathan


> 
> Thanks Bastien,
> Linus Walleij