Hi all, On Sat, 7 Nov 2009 15:46:21 +0100, Jean Delvare wrote: > Do you have a list of differences between the ADT7473 and ADT7475? > Looking briefly at the datasheets and drivers, I am under the > impression that they are essentially compatible and that a combined > driver for both chips would make a lot of sense. What do you think? The more I look at these datasheets, the more obvious is becomes that we really want a single driver for both chips. It is pointless to have 35 kB of duplicated code. Now there's the question of which driver to keep. I've gathered the following evidences: Difference between the ADT7473 and ADT7475 devices: * The ADT7475 has a fixed I2C address (0x2e) while some ADT7473 variants have a selectable address (0x2c, 0x2d or 0x2e). * The ADT7475 has two additional configuration registers (Configuration 6 at 0x10 and Configuration 7 at 0x11), but the adt7475 driver doesn't make use of them. * The ADT7473 has five additional registers (Temperature operating points at 0x33-0x35 and Dynamic Tmin control at 0x36 and 0x37) but the adt7473 driver doesn't make use of them. Difference between the adt7473 and adt7475 drivers: * The adt7473 driver is older (kernel 2.6.25, while the adt7475 driver was added in kernel 2.6.29) but the adt7475 driver is probably more popular. * The adt7475 driver reads the extra resolution bits in registers 0x76 and 0x77, while the adt7473 driver does not. * The adt7475 driver has support for temperature offsets, critical limits and their hysteresis, and fault flag, and PWM frequency selection, while the adt7473 driver does not support any of these. * The adt7473 implements the non-standard attribute pwm_use_point2_pwm_at_crit, the adt7475 driver doesn't support that. I couldn't see any obvious difference between the chips for all the registers which are used by the drivers. But maybe there are, only a careful datasheet review would tell. Bugs found in the drivers: * The adt7475 driver creates pwm#_auto_channel_temp attributes instead of the standard pwm#_auto_channels_temp. * The adt7475 driver refreshed limit values every 2 seconds even though the comment in the code says every 60 seconds. 60 seconds would make more sense. * One coding style error in driver adt7473. At the light of all this, I think I'd keep the adt7475 driver and merge ADT7473 support therein. If anyone can think of good reasons to do it the other way around, please speak up. Thanks, -- Jean Delvare _______________________________________________ lm-sensors mailing list lm-sensors@xxxxxxxxxxxxxx http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
