Hm, just noticed... Looks like the LM76 is 100% compatible with the LM92, and we have a driver for the LM92 (lm92.c). Since the MAX6635 is obviously a clone of the LM76, it should fit well in the lm92 driver as well. I admit that the lm92 driver isn't our best driver, there are several places where it does the things in a very different way when compared with the other drivers. Still it would be much better to extend and fix this existing driver that to duplicate the exact same functionality in a different driver. Since the LM92 has a manufacturer ID register, you can use that to distinguish it from Maxim's clones. Also, I noticed that the five chips (LM76, LM92, MAX6633, MAX6634 and MAX6635) have different address ranges. That's something you can use to distinguish between the four of them that do not have a manufacturer's ID register (although it will of course not be perfect). You would also add the 4-bit cycle trick as mentioned in an earlier, and the unused bits in the temperature limits registers. The lm92.c driver has a very simple detection function at the moment because it could rely on the manufacturer's ID. Now that you can't, you have to use all the tricks we've discussed before. You are welcome to clean whatever you fell like cleaning in the lm92.c driver, I don't like it much (although it's definitely better than no driver at all, I don't intend to denigrate Abraham's work, it's just that it is different from the other drivers, which makes it harder to maintain). Sorry for not checking before, that would have saved us much time and work :/ -- Jean Delvare http://www.ensicaen.ismra.fr/~delvare/
