Quoting Amit Kucheria (2019-08-29 05:30:59)
> On Thu, Aug 29, 2019 at 3:12 AM Stephen Boyd <swboyd@xxxxxxxxxxxx> wrote:
> >
> > Quoting Amit Kucheria (2019-08-27 05:14:11)
> > > + thermal_zone_device_update(priv->sensor[i].tzd,
> > > + THERMAL_EVENT_UNSPECIFIED);
> > > + } else {
> > > + dev_dbg(priv->dev, "[%u] %s: no violation: %d\n",
> > > + hw_id, __func__, temp);
> > > + }
> > > + }
> > > +
> > > + return IRQ_HANDLED;
> >
> > Should we return IRQ_NONE in the case that the above for loop didn't
> > find anything in those if/else-ifs?
>
> The upper/lower interrupts are non-sticky, level-triggered. So if the
> temp returns to within the thresholds in the time that a IRQ was
> triggered and the handler scheduled, we might not see any threshold
> violations/interrupt bits set.
>
> It feels to me that this is a case of the IRQ being handled
> (automagically) instead of IRQ_NONE. The definition of IRQ_NONE[1]
> also seems to suggest that it should be used when the IRQ wasn't
> handled. But it was handled in this case (although, automatically),
> wasn't it?
Ok I see. Sounds fine then to always return IRQ_HANDLED. Maybe you can
add a comment to this effect right above the return statement.
