On Fri, Dec 10, 2010 at 11:00:23AM -0500, Jean Delvare wrote: > On Fri, 10 Dec 2010 07:12:09 -0800, Guenter Roeck wrote: > > On Fri, Dec 10, 2010 at 09:36:46AM -0500, Jean Delvare wrote: > > > On Thu, 9 Dec 2010 16:07:08 -0800, Ira W. Snyder wrote: > > > > I would like to keep the power1_input sysfs file. > > > > > > I have no problem with this, at least as long as libsensors doesn't > > > offer a way to bind current sensors to voltage sensors. > > > > > > > I do not have any > > > > objections to changing power1_alarm to in2_alarm or in2_min_alarm. > > > > > > in2_* doesn't seem right for a voltage output alarm. I'd say such a > > > feature doesn't belong to the hwmon ABI in the first place. > > > > Why not ? You lost me there. It optionally monitors the voltage it controls, > > and provides the monitored value(s) to the user. Many of the other recent chips > > do the same. ltc4215, ltc4245, ltc4261, smm655, and pretty much all PMBus devices. > > The ltc42xx devices act as voltage switch (on/off) and don't otherwise affect > > the controlled voltage. > > For ltc4215, in2_input is still connected to a a sensor input (AD2IN or so), as is > > the FB pin (which causes the alarm if the voltage connected to it drops below a > > certain level). It doesn't make sense to state that the monitored voltage can not > > be reported through hwmon just because it may be a voltage it controls. > > Thanks for clarifying. I didn't read the datasheet, and what Ira wrote > confused me. If the alarm is related to the value of in2_input, then of > course in2_alarm, in2_min_alarm (assuming in2_min is present) or > in2_max_alarm (assuming in2_max is present) makes sense. > Is that an implied requirement for supporting min_alarm ? Reason for asking is that many chips have limit alarms but no matching (readable/settable) limit values. ltc4215, ltc4245, and ltc4261 are examples. Alarm status is determined by a voltage on a pin exceeding a hardcoded limit, and the actual voltage on a pin is determined by external resistor arrays. The ltc4245 and ltc4261 drivers already provide min_alarm and/or max_alarm but not min and max. > BTW, powerX_alarm makes no sense at all if powerX_input is computed, as > is the case for the ltc4215 according to Ira. > Agreed. It needs to be changed to in2_alarm or in2_min_alarm. Guenter _______________________________________________ lm-sensors mailing list lm-sensors@xxxxxxxxxxxxxx http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
