* Merlijn Wajer <merlijn@xxxxxxxxxx> [211216 11:34]:
> Hi,
>
> On 13/12/2021 14:50, Tony Lindgren wrote:
> > * Merlijn Wajer <merlijn@xxxxxxxxxx> [211213 10:45]:
> >
> >> and I am no longer seeing the oopses or resets when running "sleep 30;
> >> blkid" from minimal userspace.
> >
> > If disabling the cpu_thermal makes things behave, chances are we have wrong
> > thermal values in the dts for n900 and possible all 34xx devices. The 36xx
> > values are behaving AFAIK.
>
> Just disabling the bandgap doesn't seem to be enough. I also tried
> disabling just the bandgap and thermal_zones, but that is also isn't
> enough. However, *just* disabling cpu_thermal also isn't enough.
>
> Disabling both cpu_thermal and bandgap is enough to stop the oopses [1].
> So it sounds like there might potentially be some problems in
> cpu_thermal as well then?
Not sure what's wrong..
But meanwhile, looks like the patch below produces thermal values, so
maybe check if it allows idling with thermal enabled on n900. It might
just leave out all the extra polling.
Regards,
Tony
diff --git a/drivers/thermal/ti-soc-thermal/omap3-thermal-data.c b/drivers/thermal/ti-soc-thermal/omap3-thermal-data.c
--- a/drivers/thermal/ti-soc-thermal/omap3-thermal-data.c
+++ b/drivers/thermal/ti-soc-thermal/omap3-thermal-data.c
@@ -66,7 +66,8 @@ omap34xx_adc_to_temp[128] = {
/* OMAP34XX data */
const struct ti_bandgap_data omap34xx_data = {
- .features = TI_BANDGAP_FEATURE_CLK_CTRL | TI_BANDGAP_FEATURE_UNRELIABLE,
+ .features = TI_BANDGAP_FEATURE_CLK_CTRL | TI_BANDGAP_FEATURE_UNRELIABLE |
+ TI_BANDGAP_FEATURE_CONT_MODE_ONLY,
.fclock_name = "ts_fck",
.div_ck_name = "ts_fck",
.conv_table = omap34xx_adc_to_temp,
