Jean Delvare wrote: >Hi Mark, > > > >>since MIC184 has TWO sensors, it should not be added to lm75 IMHO. >>We shouldn't add a second sensor to the lm75 driver. >> >> > >It doesn't really have two sensors. It simply can use an external sensor >*instead of* the internal one. It cannot use both at the same time. It >is certainly possible to design a driver that continuously switches from >internal to external and back, but that would require a clever timing >and I don't think it's worth the effort. The chip is simply not designed >for this use. I invite you to consider all the drawbacks of such a >design (delayed interrupt response to overtemperature, need to have the >same overtemperature and hysteresis temperature for both "channels", >impossibility to know which "channel" caused an alarm/interrupt, and >possibly more bad surprises if we were to really write such a driver). > >Anyway, I don't think that Frank was interested in this - the mic184.c >driver he proposed would only export one set of sysfs files. > >Thanks, > > Jean - I guess I sent one version back that did not have the correct set function. That appears to be the only thing different with my latest version. Thats what happens when you pull from a development machine instead of cvs. Your summary seems to pretty much correct. That's fine. If anybody is looking for this driver in the future, let me know as we will continue to use it ourselves. You did have some good tips that were are going to evaluate. As far as our target, it is a Kontron CPU module on a custom carrier board. The carrier then connects to a PCI based backplane that has i2c, onewire, pci, cplds, fpgas, multiple ps2 keypads, and other fun hardware. Currently we are running a 2.4.25 (Redhat 9 based) kernel and I am migrating to a 2.6.9 (or later, may go to 10 or 11) kernel.
