On Fri, 18 Aug 2017, Darren Hart wrote:
> > before:
> > $ cat /sys/devices/platform/thinkpad_hwmon/{name,fan1_input}
> > after:
> > $ cat /sys/devices/platform/thinkpad_hwmon/hwmon/hwmon1/{name,fan1_input}
> > thinkpad
> > 3478
> > $ cat /sys/class/hwmon/hwmon1/{name,fan1_input}
> > thinkpad
> > 3478
I wonder what's the point of retaining the thinkpad_hwmon separate
device [from the thinkpad_acpi platform device] then... but changing
that might break the userspace API even further or cause other
annoyances down the road, so I guess it is the lesser evil.
> > $ sensors
> > thinkpad-isa-0000
> > Adapter: ISA adapter
> > fan1: 3489 RPM
Yeah, that should cover >90% of the usecases since most people just read
these. It *will* break write accesses using /etc/sysfs.conf and
similar, though (to set fan mode on boot, etc).
It is documented and the userspace ABI is being updated according to the
hwmon subsystem rules *and* the thinkpad-acpi rules... this is enough
for me, but be warned that people might complain.
> This looks very reasonable to me. The lm-sensors user experience is
> effectively unchanged, and the /sys/* changes move from a specific
> implementation to a generic implementation, taking advantage for the
> subsystem.
Yes, which is why I am not against the ABI change.
> This will be 4.14 because we let it sit too long. I'll correct this.
>
> I've queued this to testing for 4.14.
>
> Henrique, please shout if you have any objections here.
No objections.
--
Henrique Holschuh
