On Wed, 2011-10-12 at 11:47 -0400, Jean Delvare wrote: > On Wed, 12 Oct 2011 11:36:58 +0200, michael.hennerich@xxxxxxxxxx wrote: > > From: Michael Hennerich <michael.hennerich@xxxxxxxxxx> > > > > The Analog Devices ADT75 has an additional register at 0x04 > > for initiating oneshot captures. It is not actively used in > > this driver but we must avoid assuming it will behave as > > 0x05-0x07 do and return the last read value. In fact register 0x04 > > reads the same as register 0x00. And all registers are repetitive > > mirrored around register 0x04. This fact is used to detect the ADT75. > > I object. The detection of the original LM75 is there for historical > reasons (that chip was very popular on PC motherboards at the end of > the 90's. Other compatible chips were never so popular on PC > motherboards and this is the reason why they are _not_ detected by the > lm75 driver. If you have any of these chips, you have to instantiate > them explicitly. > > Same goes for the ADT75. I don't expect it on PC boards, but rather on > embedded designs where it should simply be instantiated, rather than > detected. > > It's still fine to document the ADT75 as being supported by the driver, > and adding an lm75_type enum value for it. But no detection, please. > Good point. I am fine with that. Turns out the chip I referred to in my other e-mail is a TI TMP75. It won't be detected either, but works fine in our design because I instantiate it with i2c_register_board_info() during board initialization. Which is why I never noticed that auto-detection does not work for it. Thanks, Guenter _______________________________________________ lm-sensors mailing list lm-sensors@xxxxxxxxxxxxxx http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
