[PATCH 1/2] ARM: dts: imx6q-cm-fx6: fix the operation points

[Christopher Spinrath]

Hi Fabio,

thanks for your review!

On 06/12/2016 03:43 PM, Fabio Estevam wrote:
> Hi Christopher,
> 
> On Sun, Jun 12, 2016 at 10:20 AM,  <christopher.spinrath at rwth-aachen.de> wrote:
>> From: Valentin Raevsky <valentin at compulab.co.il>
>>
>> Fix the cm-fx6 operation points. Remove settings for 1.2GHz.
>> The current ldo settings do not allow 1.2GHz cpu frequency.
>>
>> Signed-off-by: Valentin Raevsky <valentin at compulab.co.il>
>> [christopher.spinrath at rwth-aachen.de: rebased on upstream and adapted
>>  commit title]
>> Signed-off-by: Christopher Spinrath <christopher.spinrath at rwth-aachen.de>
>> ---
>>  arch/arm/boot/dts/imx6q-cm-fx6.dts | 17 +++++++++++++++++
>>  1 file changed, 17 insertions(+)
>>
>> diff --git a/arch/arm/boot/dts/imx6q-cm-fx6.dts b/arch/arm/boot/dts/imx6q-cm-fx6.dts
>> index 6d8d5e9..c912d58 100644
>> --- a/arch/arm/boot/dts/imx6q-cm-fx6.dts
>> +++ b/arch/arm/boot/dts/imx6q-cm-fx6.dts
>> @@ -91,6 +91,23 @@
>>         };
>>  };
>>
>> +&cpu0 {
>> +       operating-points = <
>> +               /* kHz  uV */
>> +               996000  1250000
>> +               852000  1250000
>> +               792000  1150000
>> +               396000  975000
>> +       >;
>> +       fsl,soc-operating-points = <
>> +               /* ARM kHz      SOC-PU uV */
>> +               996000          1250000
>> +               852000          1250000
>> +               792000          1175000
>> +               396000          1175000
>> +       >;
> 
> These operating points do not match the ones from imx6q.dtsi.
> 
Indeed, the voltage for 792000kHz is not the same. Thanks for pointing
that out.

> In order to avoid 1.2GHz operation we have
> imx6q_opp_check_speed_grading() inside arch/arm/mach-imx/mach-imx6q.c.
> This function checks if the mx6q is capable of running at 1.2GHz or
> not, depending on the fuse reading and then it will skip the 1.2GHz
> operating point if it detects it can't run at such frequency.
> 
> These operating points are defined at the SoC dtsi and we don't have
> any board dts that define them locally like this patch proposes, so I
> recommend we keep using the operating points from imx6q.dtsi.
> 
Well, without this patch the SoC runs at 1.2GHz so it seems like the
fuse indicates that it is possible. But with a fair amount of IPs
enabled and high load the board becomes unstable (display
flickering/blanks, ethernet package drops, ...). I guess CompuLab
allowed 1.2GHz at first and then realized that it is a bad idea (at
least for some boards).

Is there any way to influence the fuse (e.g. from the
firmware/bootloader) or do you have another proposal on how to fix this
issue?

Cheers,
Christopher