On Tue, 23 Apr 2019 17:55:54 +0300, Jarkko Nikula wrote: > On 4/19/19 11:31 PM, Jean Delvare wrote: > > This was only for testing... I'll post clean patches soon. > > > > Thanks for testing so quickly, appreciated. > > Your updated ee1004 driver works here too on 2* and 4* Crucial DIMMs > cases. I.e. decode-dimms is detecting them and outputting sensible > looking data which was possible before only when both 2* Crucial + 2* > Micron DIMMs were plugged in together. Thank you Jarkko and Dreamcat4 for all the tests so far. As I was about to send the patches for upstream inclusion, I noticed that the driver doesn't actually use the SMBus command which is mentioned in the EE1004 specification to select the page. The driver sends one dummy byte after the I2C device address, while the specification says to send two dummy bytes. Well, to be honest I can't see from a technical perspective how 2 dummy bytes could succeed if 1 dummy byte fails - I assume the NACK is sent by the EEPROM after the 1st dummy byte before it has any idea if anything is coming next - but to be on the safe side, I would appreciate if either of you could try the following command with the "failing" setup (2* Crucial): # i2cset <N> 0x36 0x00 0x00 (where <N> is the I2C bus number of the i2c-i801) If it still fails they I'll submit what I have. It it does work then I'll write a different patch. Thanks, -- Jean Delvare SUSE L3 Support
