[lm-sensors] Using lm_sensors for accessing SMSC EMC6W201

Philip Pokorny ppokorny at penguincomputing.com
Wed Jun 14 02:53:37 CEST 2006


Do you have a URL for the datasheet?  Perhaps someone else on this list 
can look at the datasheet and help you find a compatible driver.

Do you know what address the EMC6W201 device is at?

Can you post the output of 'i2cdetect -l' and 'i2cdetect 0'

If you know what the I2C address should be (it should be in the 
datasheet what the default I2C address is), then you should also run 
'i2cdump 0 0xYY' where 0xYY is the I2C address of the device.

Thanks,
:v)

Ric Johnson wrote:

>Hi:
>
>I need to monitor temperatures on my system in real time. ( big surprise,
>huh?, :) )
>The mobo is a Gigabyte(GBT) GA-2CEWH dual Opteron based system which includes
>an ITE 8712F Super I/O as well as the SMSC EMC6W201 monitoring device.  
>The SMSC is similar in function to the SMSC EMC6D* devices but that is pretty
>much all -only "similar":  lm_sensors does not recognize the 6W201 and trying
>to use the EMC6D* modules does not work. 
>
>According to GBT the 8712F is strictly used for I/O and all monitoring is done
>by the SMSC device.  I am not positive that is entirely true but it is mostly
>so.  The installation of lm_sensors (SUSE 10.x) provides pretty wacky results
>for everything except the 12VDC line and, possibly, one fan.  ...
>
>I have read most of the 6W201 datasheet and found, I hope, all pertinent info
>necessary to get the monitoring done. Of course, I do not know how to modify
>lm_sensors to get that accomplished. After reading various man pages, I still
>do not see a direct way to get it done.
>
>Cutting to the chase, I first need to be able to WRITE a "START" bit(01h) to
>bit 0 of register 40h on the $5000 SMBus(i.e., 0000 0001). That starts the
>monitoring and fills the EMC6W201 registers with valuable info, and, according
>to the datasheet, makes reading of the registers possible.  PRIOR to writing
>the START bit the registers are not accessible and any value in them is not
>good. ...
>
>After that, I need to read the registers.  First the temps, then later
>voltages and fan speeds; finally, control w/b most excellent to have(hey,
>dreams _are_ good, :) ). 
>THE readings needed most are in ($5000) SMBus registers(hex):
>
>26, diode 1 temp reading,
>27, diode 2 temp reading,
>28, diode 3 temp reading,
>29, diode 4 temp reading,
>2A, diode 5 temp reading, and
>2B, Ambient
>-- All 6 are needed because which is for what is unknown... 
>The value in those registers are signed, 8-bit, 2's compliment of the
>temperature in centigrade ranging from -127*c to 127*c.  E.g., a few:
>0*c = 0000 0000
>1*c = 0000 0001
>2*c = 0000 0010
>3*c = 0000 0011
>...
>25*c = 0001 1001
>... 
>and so forth.
>
>Of course I would like to read register 40h to make sure that it is something
>I want lm_sensors to write a 01h to before actually doing that.  ...sometimes
>datasheets and implementations are not the same.  GBT assured me that they
>followed the SMSC rec's so it should be okay but I would like to proceed
>cautiously.  I.e., the sensors-detect does find EEPROMs and I do not want to
>write there.
>
>I am not capable of providing an official driver but could possibly tweak and
>test some already established code.  Curently, I am using openSUSE 10.1-x86_64
>with kernel 2.6.16.13-4-smp.  
>There is a huge hole in my experience and knowledge about reading and writing
>binary/hex to devices on a mainboard.  That's probably obvious but I mention
>it to emphasize that I know there are steps missing from what I described
>above but unless someone takes the time to assist me, I won't know what they
>are and could ruin my mobo. So, please, at least, warn me of the shark
>infested waters before I go swimming, so to speak.
>
>HOW can I modify lm_sensors to do those things?
>
>If lm_sensors cannot be modified please let me know.  
>
>Thanks, :)
>
>Ric Johnson
>
>
>
>_______________________________________________
>lm-sensors mailing list
>lm-sensors at lm-sensors.org
>http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
>  
>





More information about the lm-sensors mailing list