> > Or at least sensor chips are not detected on them (I experienced > > this a few minutes ago and it took me a while to figure out why). > > What bus driver doesn't specify this? Why was a sensor chip not being > detected? The SMBUS check in the chip driver? My new i2c-parport driver(s) ;) But the older ones (i2c-philips-par, i2c-elv and i2c-velleman) don't specify any class either. The sensor chip not being detected was the ADM1032 of my evaluation board (lm90 driver, port in progress). I copied the I2C_ADAP_CLASS_SMBUS check from lm83, where in turn I had copied it from lm78. All chip drivers but eeprom do have the check. I think I remember that Ky?sti was quite interested in classes because it would let us ignore non-SMBus adapters in sensors. Not sure how great it is however, since actually we should look for specific chips, not specific adapters. I don't see any problem using these classes as long as it is done consistently, and that adapters that are missing their class definition are given one where needed. This could help prevent misdetections (most chip drivers already defend themselves against that, but chips such as the LM75 or the LM80 are rather hard to detect) and help user-space programs as underlined above. OTOH, we definitely can live without them, as we used to. And this will require a little bit more work from us and others. Do we all agree that setting classes for an adapter is more or less arbitrary? To me, it is more of "it was noticed that this adapter usually has this kind of devices attached" than of "this adapter could not have devices of other kinds attached, ever". So it is a hint and a misdetection preventer, not an absolute truth. Additional comments anyone? -- Jean Delvare http://www.ensicaen.ismra.fr/~delvare/
