On Sat, Jul 13, 2024 at 11:53:00AM -0700, Guenter Roeck wrote:
> On 7/13/24 08:22, Tzung-Bi Shih wrote:
> > On Fri, Jul 12, 2024 at 10:35:48AM -0700, Guenter Roeck wrote:
> > > diff --git a/drivers/hwmon/adm1021.c b/drivers/hwmon/adm1021.c
> > [...]
> > > -static const struct i2c_device_id adm1021_id[] = {
> > > - { "adm1021", adm1021 },
> > > - { "adm1023", adm1023 },
> > > - { "max1617", max1617 },
> > > - { "max1617a", max1617a },
> >
> > The device ID "max1617a" only in Documentation/hwmon/lm90.rst but not in
> > drivers/hwmon/lm90.c which results in max1617a is no longer supported after
> > applying the patch.
> >
>
> Thanks for noticing. It is an oversight, really. The chip is the same.
> max1617a is even mentioned in the lm90 documentation.
> The chip is detected and supported as "max1617" in the lm90 driver.
> Here are the notes from that driver.
>
> * Note: Multiple chips with different markings labeled as
> * "MAX1617" (no "A") were observed to report manufacturer ID
> * 0x4d and device ID 0x01. It is unknown if other variants of
> * MAX1617/MAX617A with different behavior exist. The detection
> * code below works for those chips.
Ack.
Reviewed-by: Tzung-Bi Shih <tzungbi@xxxxxxxxxx>
> Either case, all configurations enabling the adm1021 driver also enable the
> lm90 driver, and that in turn prevents the adm1021 driver from being built.
> Effectively that means that the adm1021 driver hasn't been built for a long
> time.
Curious about how this happens: is there a way for telling build system that
we prefer lm90 to adm1021?
