On Mon, Apr 01, 2019 at 03:37:57PM +0200, Simon Horman wrote:
> On Fri, Mar 22, 2019 at 05:47:53AM +0900, Yoshihiro Kaneko wrote:
> > From: Dien Pham <[email protected]>
> >
> > 1/ As evaluation of hardware team, temperature calculation formula
> > of M3-W is difference from all other SoCs as below:
> > - M3-W: Tj_1: 116 (so Tj_1 - Tj_3 = 157)
> > - Others: Tj_1: 126 (so Tj_1 - Tj_3 = 167)
> >
> > 2/ Update the formula to calculate CTEMP:
> > Currently, the CTEMP is average of val1 (is calculated by
> > formula 1) and val2 (is calculated by formula 2). But,
> > as description in HWM (chapter 10A.3.1.1 Setting of Normal Mode)
> >
> > If (STEMP < Tj_T) CTEMP value should be val1.
> > If (STEMP > Tj_T) CTEMP value should be val2.
> >
> > 3/ Update the formula to calculate temperature:
> > Currently, current TEMP is calculated as
> > average of val1 (is calculated by formula 1)
> > and val2 (is calculated by formula 2). But,
> > as description in HWM (chapter 10A.3.1.2 Normal Mode.)
> >
> > If (TEMP_CODE < THCODE2[11:0]) CTEMP value should be val1.
> > If (TEMP_CODE > THCODE2[11:0]) CTEMP value should be val2.
> >
> > Signed-off-by: Dien Pham <[email protected]>
> > [[email protected]: use the data field of the of_device_id for Tj_1]
> > [[email protected]: revise a description of case 1 of the commit log]
> > Signed-off-by: Yoshihiro Kaneko <[email protected]>
>
> I'm wondering if we could split this up into three patches,
> one for each problem it solves? Partly because I think it is good to fix
> one problem per patch. And partly because am having trouble
> verifying this patch - the new if statement in rcar_gen3_thermal_get_temp()
> seems to result in 0 temperature readings when the else case is used.
>
> > ---
> > drivers/thermal/rcar_gen3_thermal.c | 86
> > +++++++++++++++++++++++++------------
> > 1 file changed, 58 insertions(+), 28 deletions(-)
> >
> > This patch is based on the master branch of Linus Torvalds's linux tree.
> >
> > diff --git a/drivers/thermal/rcar_gen3_thermal.c
> > b/drivers/thermal/rcar_gen3_thermal.c
> > index 88fa41c..de6f244 100644
> > --- a/drivers/thermal/rcar_gen3_thermal.c
> > +++ b/drivers/thermal/rcar_gen3_thermal.c
> > @@ -63,6 +63,15 @@
> >
> > #define TSC_MAX_NUM 3
> >
> > +static int tj_2;
>
> We need to avoid global variables. There can be multiple
> instances of this driver. (Although they will all get the same
> value for tj_2.
>
> Perhaps it can be added to struct rcar_gen3_thermal_tsc?
>
> ...
Also, from just a little further down the patch
> > +/* default THCODE values if FUSEs are missing */
> > +static int thcode[TSC_MAX_NUM][3] = {
> > + { 3397, 2800, 2221 },
> > + { 3393, 2795, 2216 },
> > + { 3389, 2805, 2237 },
> > +};
I think thcode can be marked const.
