> We pretend to support the LM77 for years, using our lm75 driver. This > isn't the absolute truth. The address range is shorter, the > configuration register isn't exactly compatible, and the LM77 has two > more registers. A LM77 would never be detected by our detection > routine.(We use the fact that the LM75's registers cycle over 4-byte > (well, sort of [1]) boundaries, which of course won't work with the > LM77 since it as 6 registers.) And the lm75 module does the same check > on load so it wouldn't load for a LM77 without the force parameter. > > Suggestions: > > 1* Add a different detection routine for the LM77 to sensors-detect. I > think we can assume that the registers cycle over 8-byte [1] > boundaries. Since cycling over 4-byte boundaries implies cycling over > 8-byte boundaries, we would have to give it a lower confidence value > than the LM75. > > 2* Update the docs to explain the force parameter has to be used. I guess the collective silence here means "we don't care", so I'll do just as described above. > > 3* Really support the LM77. Not sure anyone is using that chip, so > maybe we don't have to do this part. No support planned for now. If someone asks for it, we'll see. -- Jean Delvare http://www.ensicaen.ismra.fr/~delvare/
