Re: [PATCH 1/2] thermal: exynos: Reorder exynos_map_dt_data() function

[Lukasz Majewski <l.majewski@samsung.com>]

Hi Abhilash,

> Hi Lukasz,
> 
> On Fri, Jan 30, 2015 at 8:36 PM, Abhilash Kesavan
> <kesavan.abhilash@gmail.com> wrote:
> > Hi Lukasz,
> >
> > On Fri, Jan 30, 2015 at 1:44 PM, Lukasz Majewski
> > <l.majewski@samsung.com> wrote:
> >> Hi Eduardo, Abhilash,
> >>
> >>> On Thu, Jan 22, 2015 at 06:02:07PM +0530, Abhilash Kesavan wrote:
> >>> > Hi Lukasz,
> >>> >
> >>> > On Thu, Jan 22, 2015 at 2:31 PM, Lukasz Majewski
> >>> > <l.majewski@samsung.com> wrote:
> >>> > > Hi Abhilash,
> >>> > >
> >>> > >> Hi Lukasz,
> >>> > >>
> >>> > >> On Mon, Jan 19, 2015 at 5:14 PM, Lukasz Majewski
> >>> > >> <l.majewski@samsung.com> wrote:
> >>> > >> > The exynos_map_dt_data() function must be called before
> >>> > >> > thermal_zone_of_sensor_register(), and hence provide
> >>> > >> > tmu_read() function, before it is needed.
> >>> > >> >
> >>> > >> > This change is driven by adding support for enabling
> >>> > >> > thermal_zoneX when it is properly initialized.
> >>> > >> >
> >>> > >> > One can read the mode of operation
> >>> > >> > at /sys/class/thermal/thermal_zone0/mode Such
> >>> > >> > functionality was missing in the of-thermal.c code.
> >>> > >> >
> >>> > >> > Reported-by: Abhilash Kesavan <a.kesavan@samsung.com>
> >>> > >> > Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
> >>> > >> > ---
> >>> > >> >  drivers/thermal/samsung/exynos_tmu.c | 7 ++++---
> >>> > >> >  1 file changed, 4 insertions(+), 3 deletions(-)
> >>> > >> >
> >>> > >> > diff --git a/drivers/thermal/samsung/exynos_tmu.c
> >>> > >> > b/drivers/thermal/samsung/exynos_tmu.c index
> >>> > >> > 9d2d685..5d946ab 100644 ---
> >>> > >> > a/drivers/thermal/samsung/exynos_tmu.c +++
> >>> > >> > b/drivers/thermal/samsung/exynos_tmu.c @@ -975,15 +975,16
> >>> > >> > @@ static int exynos_tmu_probe(struct platform_device
> >>> > >> > *pdev) platform_set_drvdata(pdev, data);
> >>> > >> > mutex_init(&data->lock);
> >>> > >> >
> >>> > >> > +       ret = exynos_map_dt_data(pdev);
> >>> > >> > +       if (ret)
> >>> > >> > +               goto err_sensor;
> >>> > >> > +
> >>> > >> >         data->tzd =
> >>> > >> > thermal_zone_of_sensor_register(&pdev->dev, 0, data,
> >>> > >> > &exynos_sensor_ops); if (IS_ERR(data->tzd)) {
> >>> > >> >                 pr_err("thermal: tz: %p ERROR\n",
> >>> > >> > data->tzd); return PTR_ERR(data->tzd);
> >>> > >> >         }
> >>> > >> > -       ret = exynos_map_dt_data(pdev);
> >>> > >> > -       if (ret)
> >>> > >> > -               goto err_sensor;
> >>> > >> >
> >>> > >> >         pdata = data->pdata;
> >>> > >>
> >>> > >> I have been testing this along with your v5 patch set and am
> >>> > >> seeing incorrect temperature being reported at boot-up on
> >>> > >> exynos7.
> >>> > >
> >>> > > Does it show a maximal temperature value (0x1FF)?
> >>> >
> >>> > I did not print the current temperature register, but I
> >>> > remember the message showing ~105C. Will give you the register
> >>> > value when I test with more debug prints tomorrow.
> >>> >
> >>> > >
> >>> > >>  It looks
> >>> > >> like exynos_tmu_read gets called from
> >>> > >> thermal_zone_of_device_update during boot-up, now that we
> >>> > >> have it populated early. However, as the tmu initialization
> >>> > >> function has not been called yet it returns a wrong value.
> >>> > >> Does that sound correct ?
> >>> > >
> >>> > > No, this is a mistake. However, I'm wondering why on Exynos4/5
> >>> > > this error didn't show up...
> >>> >
> >>> > I have been lowering the software trip point temperature in the
> >>> > exynos7 dts file (to 55C) for testing purposes. Hence, when the
> >>> > temperature is read incorrectly as ~105C the board trips at
> >>> > boot-up
> >>
> >>                                         ^^^^ this is a very unusual
> >>                                         value - I had problems with
> >>                                         reading 0xFF values with
> >>                                         similar symptom (but this
> >> was caused by lack of vtmu).
> >>
> >>> > itself. Maybe for exynos4/5 the incorrect value read during
> >>> > boot-up is in the non-tripping range and once the tmu is
> >>> > initialized later it continues to function properly thereafter ?
> >>> >
> >>> > >
> >>> > > The reordering is needed to be able to call set_mode callback
> >>> > > at of-thermal.c to set the mode.
> >>> > >
> >>> > > If this change causes problems, then another solution
> >>> > > (probably not so neat) must be found.
> >>> >
> >>> > Please let me know if you need any further details.
> >>
> >> Abhilash, could you provide more details (like relevant output from
> >> dmesg) and point me a list of patches which shall I apply to test
> >> this issue on Exynos4/5?
> > Sorry, I have not had the time to re-check this and provide you with
> > the current temperature register value. I will definitely do so on
> > Monday and also provide you the dmesg output. I applied the
> > following patches on linux-next:
> >
> > bbf872d thermal: exynos: Add TMU support for Exynos7 SoC
> > b8190ac dts: Documentation: Add documentation for Exynos7 SoC
> > thermal bindings 9330ec1 thermal: exynos: Reorder
> > exynos_map_dt_data() function 4dd41c4 thermal: exynos: dts: Provide
> > device tree bindings identical to the one in exynos_tmu_data.c
> > a7b80b9 thermal: dts: exynos: Trip points and sensor configuration
> > data for Exynos5440
> > 77d072e thermal: exynos: dts: Define default thermal-zones for
> > Exynos4 964dd36 thermal: dts: Default trip points definition for
> > Exynos5420 SoCs c1d2f97 thermal: exynos: dts: Add default
> > definition of the TMU sensor parameter 02a4496 arm: dts: Adding CPU
> > cooling binding for Exynos SoCs bfadff0 arm: dts: odroid: Enable
> > TMU at Exynos4412 based Odroid U3 device 862764c arm: dts: odroid:
> > Add LDO10 regulator node necessary for TMU on Odroid c064731 arm:
> > dts: trats: Enable TMU on the Exynos4210 trats device
> >
> > Along with the above list I have a patch which adds the dt changes
> > required for exynos7 on similar lines as done for exynos4/exynos5.
> > In the exyno7 trip point dts file I have modified the cpu-crit-0
> > temperature to a low value of 55C for testing purposes.
> >
> >>
> >>>
> >>> What is the status of this patch? Is it still required?
> >>
> >> It is strange, since on Exynos4/5 this works and some problems
> >> show up when run on Exynos7.
> >
> > I would have expected the issue to show up on Exynos4/5 too. I will
> > test this on the 5420 based board I have on Monday.
> 
> I tested this on a 5420 based chromebook that I have (Peach Pit) and
> observed similar results as that on Exynos7. Following are the patches
> applied on next-20150130:
> 
> 5b1194d thermal: exynos: Reorder exynos_map_dt_data() function
> 30c6165 thermal: exynos: dts: Provide device tree bindings identical
> to the one in exynos_tmu_data.c
> d94c248 thermal: dts: exynos: Trip points and sensor configuration
> data for Exynos5440
> aac8b3a thermal: exynos: dts: Define default thermal-zones for Exynos4
> 5e8cf52 thermal: dts: Default trip points definition for Exynos5420
> SoCs 17ec9c2 thermal: exynos: dts: Add default definition of the TMU
> sensor parameter 36e247b arm: dts: Adding CPU cooling binding for
> Exynos SoCs 7aa1bb1 arm: dts: odroid: Enable TMU at Exynos4412 based
> Odroid U3 device 8884a76 arm: dts: odroid: Add LDO10 regulator node
> necessary for TMU on Odroid aae2e51 arm: dts: trats: Enable TMU on
> the Exynos4210 trats device 827e3bd Add linux-next specific files for
> 20150130
> 
> On top of these patches, I have the following diff for
> debugging/testing pruposes:
> 
> diff --git a/arch/arm/boot/dts/exynos5420-trip-points.dtsi
> b/arch/arm/boot/dts/exynos5420-trip-points.dtsi
> index 09d6c56..ac8b6a0 100644
> --- a/arch/arm/boot/dts/exynos5420-trip-points.dtsi
> +++ b/arch/arm/boot/dts/exynos5420-trip-points.dtsi
> @@ -13,22 +13,22 @@ polling-delay-passive = <0>;
>  polling-delay = <0>;
>  trips {
>         cpu-alert-0 {
> -               temperature = <85000>; /* millicelsius */
> +               temperature = <55000>; /* millicelsius */
>                 hysteresis = <10000>; /* millicelsius */
>                 type = "active";
>         };
>         cpu-alert-1 {
> -               temperature = <103000>; /* millicelsius */
> +               temperature = <63000>; /* millicelsius */
>                 hysteresis = <10000>; /* millicelsius */
>                 type = "active";
>         };
>         cpu-alert-2 {
> -               temperature = <110000>; /* millicelsius */
> +               temperature = <70000>; /* millicelsius */
>                 hysteresis = <10000>; /* millicelsius */
>                 type = "active";
>         };
>         cpu-crit-0 {
> -               temperature = <1200000>; /* millicelsius */
> +               temperature = <75000>; /* millicelsius */
>                 hysteresis = <0>; /* millicelsius */
>                 type = "critical";
>         };
> diff --git a/drivers/thermal/samsung/exynos_tmu.c
> b/drivers/thermal/samsung/exynos_tmu.c
> index 852e622..7281581 100644
> --- a/drivers/thermal/samsung/exynos_tmu.c
> +++ b/drivers/thermal/samsung/exynos_tmu.c
> @@ -237,6 +237,7 @@ static int code_to_temp(struct exynos_tmu_data
> *data, u8 temp_code)
>                 break;
>         case TYPE_ONE_POINT_TRIMMING:
>                 temp = temp_code - data->temp_error1 +
> pdata->first_point_trim;
> +               printk("temp_code is %d, err1 is %d, trim is %d, temp
> is %d\n", temp_code, data->temp_error1, pdata->first_point_trim,
> temp);
>                 break;
>         default:
>                 temp = temp_code - pdata->default_temp_offset;
> @@ -585,6 +586,7 @@ static int exynos_get_temp(void *p, long *temp)
> 
>         *temp = code_to_temp(data, data->tmu_read(data)) * MCELSIUS;
> 
> +       printk("temperature is %ld\n", *temp);
>         clk_disable(data->clk);
>         mutex_unlock(&data->lock);
> 
> @@ -675,6 +677,7 @@ static int exynos4210_tmu_read(struct
> exynos_tmu_data *data)
> 
>  static int exynos4412_tmu_read(struct exynos_tmu_data *data)
>  {
> +       printk("\n%s: Reading the current temperature register value:
> 0x%x\n\n", __func__, readb(data->base + EXYNOS_TMU_REG_CURRENT_TEMP));
>         return readb(data->base + EXYNOS_TMU_REG_CURRENT_TEMP);
>  }
> 
> diff --git a/drivers/thermal/thermal_core.c
> b/drivers/thermal/thermal_core.c index 48491d1..981fc99 100644
> --- a/drivers/thermal/thermal_core.c
> +++ b/drivers/thermal/thermal_core.c
> @@ -469,7 +469,7 @@ static void update_temperature(struct
> thermal_zone_device *tz)
>         mutex_unlock(&tz->lock);
> 
>         trace_thermal_temperature(tz);
> -       dev_dbg(&tz->device, "last_temperature=%d,
> current_temperature=%d\n",
> +       dev_err(&tz->device, "last_temperature=%d,
> current_temperature=%d\n", tz->last_temperature, tz->temperature);
>  }
> 
> Also, I noticed  a typo (an extra zero) in cpu-crit-0 temperature
> above.
> 
> Following is the relevant boot-up log:
> 
> [    3.160126] TPS65090_RAILSLDO2: supplied by vbat-supply
> [    3.343421] thermal thermal_zone5: last_temperature=0,
> current_temperature=25900
> [    3.349365] sbs-battery 20-000b: sbs-battery: battery gas gauge
> device registered
> [    3.374280] 10060000.tmu supply vtmu not found, using dummy
> regulator [    3.379408]
> [    3.379408] exynos4412_tmu_read: Reading the current temperature
> register value: 0x36
> [    3.379408]
> [    3.390026] temp_code is 54, err1 is 0, trim is 25, temp is 79
> [    3.395827] temperature is 79000
> [    3.399053] thermal thermal_zone0: last_temperature=0,
> current_temperature=79000
> [    3.406429] thermal thermal_zone0: critical temperature reached(79
> C),shutting down
> [    3.414241] reboot: Failed to start orderly shutdown: forcing the
> issue [    3.420690] usb 5-1: new high-speed USB device number 2
> using exynos-ehci [    3.427819] 10064000.tmu supply vtmu not found,
> using dummy regulator [    3.434011] Emergency Sync complete
> [    3.434355]
> [    3.434355] exynos4412_tmu_read: Reading the current temperature
> register value: 0x3
> [    3.434355]
> [    3.434367] temp_code is 3, err1 is 0, trim is 25, temp is 28
> [    3.434373] temperature is 28000
> [    3.434393] thermal thermal_zone1: last_temperature=0,
> current_temperature=28000
> [    3.434744] 10068000.tmu supply vtmu not found, using dummy
> regulator [    3.435069]
> [    3.435069] exynos4412_tmu_read: Reading the current temperature
> register value: 0x2
> [    3.435069]
> [    3.435080] temp_code is 2, err1 is 0, trim is 25, temp is 27
> [    3.435086] temperature is 27000
> [    3.435103] thermal thermal_zone2: last_temperature=0,
> current_temperature=27000
> [    3.435427] 1006c000.tmu supply vtmu not found, using dummy
> regulator [    3.435762]
> [    3.435762] exynos4412_tmu_read: Reading the current temperature
> register value: 0x2
> [    3.435762]
> [    3.435772] temp_code is 2, err1 is 0, trim is 25, temp is 27
> [    3.435778] temperature is 27000
> [    3.435794] thermal thermal_zone3: last_temperature=0,
> current_temperature=27000
> [    3.436112] 100a0000.tmu supply vtmu not found, using dummy
> regulator [    3.436464]
> [    3.436464] exynos4412_tmu_read: Reading the current temperature
> register value: 0x33
> [    3.436464]
> [    3.436475] temp_code is 51, err1 is 0, trim is 25, temp is 76
> [    3.436480] temperature is 76000
> [    3.436496] thermal thermal_zone4: last_temperature=0,
> current_temperature=76000
> [    3.436527] thermal thermal_zone4: critical temperature reached(76
> C),shutting down
> 
> From the log, thermal_zone0 and thermal_zone4 show incorrect
> temperatures. It looks like the first point trim error1 would need to
> be intialized before we do a read ?

Could you check if this error happens when you have default temp
threshold values?

It is strange, that only this exceeded temp happens for thermal zone 0
and 4. Other three show correct temp.

Off topic - where are placed those two misbehaving zones? Isn't thermal
zone 4 the one for GPU? 


> Please let me know if you need any more details.
> 
> Regards,
> Abhilash



-- 
Best regards,

Lukasz Majewski

Samsung R&D Institute Poland (SRPOL) | Linux Platform Group