Re: [RFC 0/4] Intel Integrated Sensor Hub Support (ISH)

[Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>]

On Fri, 2016-06-10 at 16:26 +0200, Bastien Nocera wrote:
> On Fri, 2016-06-10 at 06:55 -0700, Srinivas Pandruvada wrote:
> > On Fri, 2016-06-10 at 10:44 +0100, Grant Likely wrote:
> > [...]
> > > 
> > > Yes, I do see the in_*_raw_* files in sysfs, and I'm able to get
> > > data
> > > out of them. Gnome also seems to recognize that the
> > > accelerometers
> > > are
> > > there because it adds a rotation lock button to the system menu.
> > > 
> > > However, the events are getting through yet. It may be that I'm
> > > missing something in my kernel config. I'm rebuilding the kernel
> > > with
> > > Debian's config for the 4.5.5 kernel as a quick sanity test. I'll
> > > report back when I've tried.
> > > 
> > You need to add in /usr/lib/system/systemd/iio-sensor-*
> > 
> > After=multi-user.target
> > 
> > 
> > This service starts too early.
> 
> Nobody has been able to explain to me why that would be a problem.
> 
> Can you back this up with some data? Does iio-sensor-proxy exit too
> early? Does that leave time to the IIO sub-system to do some init
> work
> it should already have done?
The driver's  trigger callback doesn't even get called to take any
action or provide any data. So to get called need to restart the
service or do suspend/resume.
After=multi-user.target, is not the ideal solution as it is too late.
So need further debug.

> 
> See also https://github.com/hadess/iio-sensor-proxy/issues/82
> 
> > > I'll also take some time and review the patch series later this
> > > afternoon.
> > 
> > Atri Bhattacharya reported issues with suspend/resume. With my test
> > patches it is fixed for him
> > 
> > If you want to hold on, I will repost with fixes for susend/resume.
>