Re: [PATCH v3 6/9] drivers: thermal: tsens: Add watchdog support

[Amit Kucheria]

On Fri, Jan 3, 2020 at 1:25 AM Bjorn Andersson
<bjorn.andersson@xxxxxxxxxx> wrote:
>
> On Thu 02 Jan 06:54 PST 2020, Amit Kucheria wrote:
> [..]
> > diff --git a/drivers/thermal/qcom/tsens-common.c b/drivers/thermal/qcom/tsens-common.c
> > index 4cf550766cf6..ecbc722eb348 100644
> > --- a/drivers/thermal/qcom/tsens-common.c
> > +++ b/drivers/thermal/qcom/tsens-common.c
> > @@ -377,6 +377,24 @@ irqreturn_t tsens_critical_irq_thread(int irq, void *data)
> >       struct tsens_irq_data d;
> >       unsigned long flags;
> >       int temp, ret, i;
> > +     u32 wdog_status, wdog_count;
> > +
> > +     if (priv->feat->has_watchdog) {
> > +             /* Watchdog is present only on v2.3+ */
>
> Please omit this comment, you're carrying the motivation for this
> decision when you set has_watchdog already.

Will fix.

>
> > +             ret = regmap_field_read(priv->rf[WDOG_BARK_STATUS], &wdog_status);
> > +             if (ret)
> > +                     return ret;
> > +
> > +             /* Clear WDOG interrupt */
> > +             regmap_field_write(priv->rf[WDOG_BARK_CLEAR], 1);
> > +             regmap_field_write(priv->rf[WDOG_BARK_CLEAR], 0);
>
> Are you sure you need to zero the clear register?

Yes, the corresponding downstream code is as follows:
                       /*
                         * Clear watchdog interrupt and
                         * increment global wd count
                         */
                        writel_relaxed(wd_mask | TSENS_TM_CRITICAL_WD_BARK,
                                (TSENS_TM_CRITICAL_INT_CLEAR
                                (tm->tsens_tm_addr)));
                        writel_relaxed(wd_mask & ~(TSENS_TM_CRITICAL_WD_BARK),
                                (TSENS_TM_CRITICAL_INT_CLEAR
                                (tm->tsens_tm_addr)));


> > +
> > +             ret = regmap_field_read(priv->rf[WDOG_BARK_COUNT], &wdog_count);
> > +             if (ret)
> > +                     return ret;
> > +             if (wdog_count)
> > +                     dev_dbg(priv->dev, "%s: watchdog count: %d\n", __func__, wdog_count);
> > +     }
> >
> [..]
> > @@ -793,6 +815,22 @@ int __init init_common(struct tsens_priv *priv)
> >               }
> >       }
> >
> > +     if (tsens_version(priv) > VER_1_X &&  ver_minor > 2) {
> > +             /* Watchdog is present only on v2.3+ */
> > +             priv->feat->has_watchdog = 1;
> > +             for (i = WDOG_BARK_STATUS; i <= CC_MON_MASK; i++) {
> > +                     priv->rf[i] = devm_regmap_field_alloc(dev, priv->tm_map,
> > +                                                           priv->fields[i]);
> > +                     if (IS_ERR(priv->rf[i])) {
> > +                             ret = PTR_ERR(priv->rf[i]);
> > +                             goto err_put_device;
> > +                     }
> > +             }
> > +             /* Enable WDOG and disable cycle completion monitoring */
>
> Commit message says you're not enabling it. Should this say "WDOG is
> already configured, unmask the bark" or something along those lines?

Will reword

>
> > +             regmap_field_write(priv->rf[WDOG_BARK_MASK], 0);
> > +             regmap_field_write(priv->rf[CC_MON_MASK], 1);
> > +     }
> > +
> >       spin_lock_init(&priv->ul_lock);
> >       tsens_enable_irq(priv);
> >       tsens_debug_init(op);
>
> Regards,
> Bjorn

Thanks for the review.