On 19.11.2023 09:39, Jean Delvare wrote: > Hi Heiner, > > On Wed, 15 Nov 2023 12:00:18 +0100, Heiner Kallweit wrote: >> After thinking once more about it I think we have to do it from the >> ee1004 driver for DDR4. For DDR3 from at24. Only this way we can read >> the "temp sensor present" flag from SPD. E.g. for ee1004 the SPD EEPROM >> may be switched to the second page and we have to switch it to the first >> page first. > > We indeed need to read the EEPROM data at some point. My initial > thinking was to instantiate the at24 or ee1004 device first (from > i2c-smbus), then read the value from the EEPROM and instantiate the > jc42 device (still from i2c-smbus). This requires an internal read > function. I think we already have that in at24, because it uses the > nvmem framework, but ee1004 lacks it as far as I can see. > I sent RFC patches for both, ee1004 and at24. > But anyway, if you have another approach which works, that's equally > fine with me. > >> I'll send a RFC patch for this. Unfortunately I have no RAM with temp >> sensor to test it. > > Neither do I :-( > I organized a DDR4 module with temp sensor and the ee1004 RFC patch properly instantiated the temp sensor. Surprisingly I found that my old DDR4 modules also have a temp sensor, it's just not advertised in SPD.
