Battery information in sysfs for userspace tools

Hi,

Evgeni Golov (CC:ed) is currently taking care of fixing acpitool wrt. sysfs support (which is badly broken).

Using the files in /sys/class/power_supply/BAT*/ we have some wondering about units.

For the current rate, for example, we have:

corsac@hidalgo: grep . /sys/class/power_supply/BAT0/*now /sys/class/power_supply/BAT0/current_now:14415000 /sys/class/power_supply/BAT0/energy_now:44170000 /sys/class/power_supply/BAT0/voltage_now:12481000 corsac@hidalgo: grep . /sys/devices/platform/smapi/BAT0/*now /sys/devices/platform/smapi/BAT0/current_now:1151 /sys/devices/platform/smapi/BAT0/power_now:14365

I'm a bit puzzled with the differences in current_now (the battery is charging at that time), and don't really know if we can trust the units.

I would assume (in /sys/class/power_supply):
current_now to be in uA
energy_now to be in uWh
voltage_now to be in uV

as said in include/linux/power_supply.h

/*
* All voltages, currents, charges, energies, time and temperatures in uV,
* µA, µAh, µWh, seconds and tenths of degree Celsius unless otherwise
* stated. It's driver's job to convert its raw values to units in which
* this class operates.
*/

and the current (dis)charging rate in W to be (current_now/1.000.000)*(energy_now/1.000.000). But that gives, for an end of charge, a charging rate of:

(14415000 / 1000000) * (12481000 / 1000000) = approx. 179.91362

which seems definitely wrong.

Where am I mistaken?

Btw this is on 2.6.29.2 on a Thinkpad T61, and tp-smapi is loaded.

Cheers,

Yves-Alexis

--
To unsubscribe from this list: send the line "unsubscribe linux-acpi" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Currently, sysfs interface repeats brokenness of /proc/acpi/battery interface.
it is possible to have battery info in energy units and charge units.
If you see 'energy_now' variable, then 'current_now' and more appropriate 'power_now'
will report in uW units.
But if you have 'charge_now', then 'current_now' will report in uA units, and you will not see 'power_now'.
So, in short, 'charge_now' reports in same units (sans *h) as capacity, either 'charge_now' or 'energy_now'.
The proper way to handle it is to check for 'energy_now' and if it is present, look for 'power_now'. It should be
present in 2.6.30 and later kernels (remaining time energy_now/power_now). If it is not present (as in 2.6.29), then you should assume that 'current_now' has value in uW, and remaining time is energy_now/current_now. If you have 'charge_now', remaining time is charge_now/current_now without any complications.

Regards,
Alex.
--
To unsubscribe from this list: send the line "unsubscribe linux-acpi" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

Hmmh, ok, but it seems to have change recently (since powertop and acpitools are definitely wrong when it comes to that)

[…]

Wow. Thanks for the info, that's definitely not really easily guessable :) It'd be nice to have that documented somewhere until things settle down a bit and the situation becomes unbroken.

Cheers,

Yves-Alexis

--
To unsubscribe from this list: send the line "unsubscribe linux-acpi" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html