On Tue, Nov 21, 2017 at 11:19 PM, STeve Andre' <and...@msu.edu> wrote: > On 11/21/17 16:31, Mark Kettenis wrote: >> >> The diff below exposes voltage regulators as sensors. This makes it >> easy to look at the current settings of these regulators. The >> downside is that these aren't really sensors as the voltages are not >> actually measured. >> >> The functionality is optional; callers can pass NULL in the >> regulator_register() if the regulators aren't particularly >> interesting. >> >> This is what it looks like on the rk3399-firefly: >> >> milhaud$ sysctl hw.sensors >> hw.sensors.rktemp0.temp0=23.89 degC (CPU) >> hw.sensors.rktemp0.temp1=28.75 degC (GPU) >> hw.sensors.rkpmic0.volt0=0.90 VDC (vdd_cpu_l) >> hw.sensors.rkpmic0.volt1=1.80 VDC (vcc1v8_dvp) >> hw.sensors.rkpmic0.volt2=1.80 VDC (vcc1v8_pmu) >> hw.sensors.rkpmic0.volt3=3.00 VDC (vcc_sd) >> hw.sensors.rkpmic0.volt4=1.80 VDC (vcca1v8_codec) >> hw.sensors.rkpmic0.volt5=3.00 VDC (vcc_3v0) >> >> thoughts? > > > As someone who does hardware stuff, having easy access to these sensorts > can't hurt, and might be useful in some situations. I've measured voltages > before and found during extreme temperature conditions things changed. So > it's possibly useful and doesn't cost much.
This reply illustrates the problem, and I think it won't be the last time someone misunderstands the feature. They are *not* sensors, so they will not vary under load. They don't reflect the actual voltage, so they are useless for checking if the hardware works as it should. I bet that a lot of people will still assume that they can be used as such, leading people to believe that everything is OK with their hardware when it's not.
