On Sun, 2008-06-15 at 14:18 +0200, Jean Delvare wrote: > Values above 63 degrees C are also broken due to the cast to s8. You > wrote yourself above that the ADT7461 treats the values as unsigned, > so > this conversion function should really return a u8 not s8. The cast > from u8 to s8 and back should happen only when you write the value to > struct lm90_data and read it back from the structure, respectively > (and > I guess this will be done implicitly so you may not even have to > care.) > I agree that you end up writing the same value to the chip register, > but it's convenient to also be able to use the value inside the driver > (for debugging purposes if nothing else) so it should be correct all > along the way. Point taken -- "broken" sounds a bit harsh since it does actually work at those temperatures as s8/s16 :) The driver didn't appear to use the values in any way that would have resulted in unexpected behavior (though I suppose this could change and could also confuse during debug like you mention). I'll change those to unsigned. Quick question before I address all of your feedback with patch v2: You didn't mention these specifically, but I'm assuming you'd also want the tempX_from_reg_adt7461() functions to use u8 and u16 as well? Thanks for the feedback. -- Nate Case <ncase at xes-inc.com>
