Re: [PATCH 2/2] hwmon: (pmbus/max20830) Add support for MAX20810, MAX20815, LTM4739, LTM4740

[Guenter Roeck]

Hi Carlos,

On 10/30/26 03:03, Carlos Jones wrote:
> From: Carlos Jones Jr <carlosjr.jones@xxxxxxxxxx>
>
> Hi Guenter,
>
> Thanks for the review.
>
> > According to the datasheets, MAX20810 reports "MAX20810", and MAX20815
> > reports "MAX20815". I find no reference to the "B" variants on the
> > Analog Web page.
> >
> > LTM4739, according to its datasheet, reports "LTM4739", and LTM4740
> > reports "LTM4740".
> >
> > Please explain the discrepancy.
>
> The discrepancy is between the datasheets and what the parts actually store
> in IC_DEVICE_ID (0xAD).
>
> The LTM4739 and LTM4740 are uModule products built around the MAX20810 and
> MAX20815 dies, respectively. On the hardware I tested, reading IC_DEVICE_ID
> returns the internal MAX die identifier, not the LTM part name:
>
>      LTM4739 -> "MAX20810B"
>      LTM4740 -> "MAX20815B"
>
> So the modules report the die's ID (with a "B" die/stepping suffix), which is
> not the string the LTM datasheets currently document. This is a datasheet vs
> silicon mismatch, not a driver choice: the driver matches what the silicon
> actually returns so that probe succeeds on real hardware.
>
> I only added the IDs I could confirm on hardware (MAX20810B / MAX20815B, as
> reported by the LTM4739 / LTM4740 modules). I deliberately did not add the
> "MAX20810" / "MAX20815" / "LTM4739" / "LTM4740" strings, since no part I have
> reports them. The driver keeps the LTM4739/LTM4740 naming and device-tree
> compatibles for the modules while matching on the MAX die IDs they report.
>
> I raised this with the product line and a datasheet correction has been
> initiated to document the MAX20810B / MAX20815B IC_DEVICE_ID values.
>
> If you'd prefer the discrepancy spelled out in-tree, I'm happy to add a note to
> the commit message and a code comment in a v2.

Please do. At the very least we should know what we get into. _If_ parts
do exist which match the current datasheets, it should at least be documented
why those are not matched.

Thanks,
Guenter