[lm-sensors] SE95 patch - some questions

Tonu Samuel tonu at jes.ee
Fri Oct 7 14:27:53 CEST 2005


I am writing support for Philips SE95 temperature sensor which is very similar 
to LM75 but has much higher precision. Because it is my first try with 
lm_sensors I came to different dead ends and now again in doubt.

Because lm75 and se95 are very similar I ended up in modifying lm75.c code 
instead of making one more copy of it. Patch against 2.6.14-pre3 is there:


This patch actually works for SE95 and I hope I did not broke LM75 either. I 
have LM75 in house, so I later check it anyway.

Main difference between LM75 and SE95 is precision (9 vs 13 bits of data). I 
want to reuse as much possible of lm75 code but still avoid floating point 
arithmetics and such. Please take a look and tell what way to go. My 
questions (pretty related to each other):

1. Should I introduce value "precision" in bits in code or just have separate 
   hardcoded formulas for lm75 and se95?
2. Any more sensors which may be added here?
3. Maybe I should go back and implement se95 code in separate file at all?
4. I haven't searched hard but can I use value "kind" somehow to distinguish
    between different sensors? How? Probably I just grep around and find it
    myself too :)

Some output:

Adapter: SMBus PIIX4 adapter at 5000

temp:      +19.4°C  (high = +80.0°C, hyst = +75.0°C)

next second:

Adapter: SMBus PIIX4 adapter at 5000

temp:      +19.5°C  (high = +80.0°C, hyst = +75.0°C)

Ouch, were I can increase decimals? We need few more for that sensor :)


More information about the lm-sensors mailing list