Comment # 5 on bug 1181890 from Jean Delvare

My initial fix to the pwmconfig script was incomplete, as the error message was
still printed (despite the error itself being handled properly). I'll post an
update soon.

The script fails to spot some of the correlations because it considers speeds
as not changing if they don't drop below 75% of the maximum speed. In your
case, the fan speed changes from 4504 RPM to 3579 RPM during the test, which is
79%, so above the detection threshold. Maybe we need to increase the threshold,
however there is then a risk of false positive for fan inputs with too much
noise.

From your logs, it appears that the dell_smm and f71882fg drivers are operating
on the same fan inputs. Dell's SMM is an abstraction layer on top of the
physical hardware monitoring device. It's generally a bad idea to use both the
abstraction layer and the physical device directly at the same time. The two
access methods could race with each other and that would lead to possibly bad
consequences. I advise that you load either dell_smm or f71882fg but not both.

As a matter of fact, I see inconsistencies in your results. dell_smm's fan1 is
reported as being controlled by both pwm1 (4504 -> 1492) and pwm2 (4504 ->
3579). That's unusual and does not make much sense, unless maybe these controls
use different methods (PWM vs DC, or 4th fan pin vs legacy 3-pin fan control).
But this could also be the result of a bad interaction between the two drivers.

The fact that you need to use weird PWM values to get proper results is not a
problem of pwmconfig. It is either a bug in the kernel drivers, or the
consequence of board-specific quirks the drivers aren't aware of.