2019-10-01 16:52:20

by Patrick Williams

[permalink] [raw]
Subject: [PATCH] hwmon: (pmbus) add VR12/VR13 mode support write paths

pmbus_core supported VR11/VR12/VR13 modes when reading
VID-formatted registers, but the write path only supported
VR11 translations. Add support for VR12 and VR13 to
'data2reg_vid' for translating these formats. This is the
inverse of 'reg2data_vid'.

Signed-off-by: Patrick Williams <[email protected]>
---
drivers/hwmon/pmbus/pmbus_core.c | 14 ++++++++++++--
1 file changed, 12 insertions(+), 2 deletions(-)

diff --git a/drivers/hwmon/pmbus/pmbus_core.c b/drivers/hwmon/pmbus/pmbus_core.c
index 8470097907bc..f0d696552142 100644
--- a/drivers/hwmon/pmbus/pmbus_core.c
+++ b/drivers/hwmon/pmbus/pmbus_core.c
@@ -845,9 +845,19 @@ static u16 pmbus_data2reg_direct(struct pmbus_data *data,
static u16 pmbus_data2reg_vid(struct pmbus_data *data,
struct pmbus_sensor *sensor, long val)
{
- val = clamp_val(val, 500, 1600);
+ switch (data->info->vrm_version) {
+ case vr11:
+ val = clamp_val(val, 500, 1600);
+ return 2 + DIV_ROUND_CLOSEST((1600 - val) * 100, 625);
+ case vr12:
+ val = clamp_val(val, 0, 1520);
+ return ((val - 250) / 5) + 1;
+ case vr13:
+ val = clamp_val(val, 0, 2500);
+ return ((val - 500) / 10) + 1;
+ }

- return 2 + DIV_ROUND_CLOSEST((1600 - val) * 100, 625);
+ return 0;
}

static u16 pmbus_data2reg(struct pmbus_data *data,
--
2.17.2 (Apple Git-113)


2019-10-02 06:32:16

by Patrick Williams

[permalink] [raw]
Subject: Re: [PATCH] hwmon: (pmbus) add VR12/VR13 mode support write paths

---- On Tue, 01 Oct 2019 16:02:10 -0700 Guenter Roeck <[email protected]> wrote ----

> > - val = clamp_val(val, 500, 1600);
> > + switch (data->info->vrm_version) {
> > + case vr11:
> > + val = clamp_val(val, 500, 1600);
> > + return 2 + DIV_ROUND_CLOSEST((1600 - val) * 100, 625);
> > + case vr12:
> > + val = clamp_val(val, 0, 1520);
> > + return ((val - 250) / 5) + 1;
> > + case vr13:
> > + val = clamp_val(val, 0, 2500);
> > + return ((val - 500) / 10) + 1;
>
> Both vr12 and vr13 converts low values into negative values,
> which are then converted into more or less random register
> values. That can not be correct. The resulting register values
> must always be valid.
>
> Guenter

Thanks for catching this. It may be as simple as me adjusting the lower
bound on the clamp_val. I’ll check with one of the device specs and
confirm the appropriate behavior on the lower bounds.