> > Nice catching for TEMP_TO_REG, I completely missed that one. BTW, > > now that I took a longer look at it, I really wonder how a negative > > temperature limit could ever been set with this. Looks like it'll > > end up with 0. OTOH, asking for a temperature higher than 127.5 > > degrees will most likely lead to "random" negative values. This is > > probably true for both the 2.4 and the 2.6 driver as of now. Could > > you please test and confirm? Most likely nobody cares about setting > > negative limits, but while we're at cleaning up this part of the > > driver... > > Yep, testing (on 2.6) confirms this, fixed it. Care to provide a similar fix for 2.4? > Btw, "grep SENSORS_LIMIT drivers/i2c/chips/*.c|grep 127" shows that > lm85.c has a rounding error in it's TEMP_TO_REG() macro (and probably > in at least one other place). Correct. BTW, the same grep shows that rounding is probably missing in EXTTEMP_TO_REG and OFFSET_TO_REG of the same driver. Feel free to submit a patch to Greg that fixes this. If a similar fix is required for 2.4 (and most likely it is), a patch will of course be welcome as well. Thanks. -- Jean Delvare http://khali.linux-fr.org/