* Re: [lm-sensors] lm_sensors, hwmon, conky
2013-11-25 20:02 [lm-sensors] lm_sensors, hwmon, conky Patrick...
@ 2013-11-25 21:15 ` Guenter Roeck
2013-11-25 22:56 ` Guenter Roeck
1 sibling, 0 replies; 3+ messages in thread
From: Guenter Roeck @ 2013-11-25 21:15 UTC (permalink / raw)
To: lm-sensors
On 11/25/2013 12:02 PM, Patrick... wrote:
> Using hwmon to config conky.
> At reboot the info from hwmon appears in different hwmon
> directories. Consequently, conky reads the wrong info into
> it's output.
>
> Is this an unreported bug or ?
>
> Please advise.
>
Hi,
unfortunately, you don't provide enough information to really enable us to help.
From looking into the conky source, I am making the assumption that
you see the problem when booting kernel version 3.13-rc1. Of course,
that is just a wild guess.
conky wrongly assumes that sensor attributes are available in directory
/sys/class/hwmon/hwmonX/device. That assumption is wrong; attributes
can also be in /sys/class/hwmon/hwmonX directly. Several existing drivers
implement this scheme already and thus never worked with conky.
Starting with kernel version 3.13, several drivers have been converted
to attach sysfs attributes to the hwmon device and no longer to its
parent device. Going forward, more drivers will be converted to the new scheme.
Oddly enough, conky also searches through i2c and platform device directories
when looking for hwmon attributes. That is really not a good idea either.
It may work today, but there is no guarantee that it will work tomorrow.
conky should only look in /sys/class/hwmon/hwmonX and /sys/class/hwmon/hwmonX/device
for hwmon attributes. It can select the correct directory by searching for
the mandatory 'name' attribute. Of couse, an even better solution would
be to use libsensors to access hwmon attributes.
Thanks,
Guenter
_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [lm-sensors] lm_sensors, hwmon, conky
2013-11-25 20:02 [lm-sensors] lm_sensors, hwmon, conky Patrick...
2013-11-25 21:15 ` Guenter Roeck
@ 2013-11-25 22:56 ` Guenter Roeck
1 sibling, 0 replies; 3+ messages in thread
From: Guenter Roeck @ 2013-11-25 22:56 UTC (permalink / raw)
To: lm-sensors
On 11/25/2013 02:49 PM, Patrick... wrote:
> I'll try to provide extra info as needed.
>
> The problem is when I set conky to use
> /sys/class/hwmon/hwmon0
> /sys/class/hwmon/hwmon1
> /sys/class/hwmon/hwmon2
>
> These are the only 3 I have.
>
> OK. Sometimes hwmon0 is for the systemp, coretemp,
> or fan speed. It doesn't stay the same. This affects
> hwmon2 and hwmon3 the same way. They constantly
> change after reboot what they are doing in relation to
> systemp, coretemp, or fan speed. hwmon0 isn't always
> the systemp and has to stay for systemp always, and the
> same for hwmon2 and hwmon3. How to fix ? Just started
> this after recent update.
>
> Using kernel 3.4.70
>
You can not depend on hwmonX staying the same across reboots.
To associate hwmonX with a sensor, or rather with a specific
hardware monitoring device, you have to check for the 'name'
attribute, which is available either in /sys/class/hwmon/hwmonX/name
or /sys/class/hwmon/hwmonX/device/name.
Guenter
> thanks for the response.
>
> patrick
>
>
> On Mon, Nov 25, 2013 at 3:15 PM, Guenter Roeck <linux@roeck-us.net <mailto:linux@roeck-us.net>> wrote:
>
> On 11/25/2013 12:02 PM, Patrick... wrote:
>
> Using hwmon to config conky.
> At reboot the info from hwmon appears in different hwmon
> directories. Consequently, conky reads the wrong info into
> it's output.
>
> Is this an unreported bug or ?
>
> Please advise.
>
>
> Hi,
>
> unfortunately, you don't provide enough information to really enable us to help.
>
> >From looking into the conky source, I am making the assumption that
> you see the problem when booting kernel version 3.13-rc1. Of course,
> that is just a wild guess.
>
> conky wrongly assumes that sensor attributes are available in directory
> /sys/class/hwmon/hwmonX/__device. That assumption is wrong; attributes
> can also be in /sys/class/hwmon/hwmonX directly. Several existing drivers
> implement this scheme already and thus never worked with conky.
> Starting with kernel version 3.13, several drivers have been converted
> to attach sysfs attributes to the hwmon device and no longer to its
> parent device. Going forward, more drivers will be converted to the new scheme.
>
> Oddly enough, conky also searches through i2c and platform device directories
> when looking for hwmon attributes. That is really not a good idea either.
> It may work today, but there is no guarantee that it will work tomorrow.
>
> conky should only look in /sys/class/hwmon/hwmonX and /sys/class/hwmon/hwmonX/device
> for hwmon attributes. It can select the correct directory by searching for
> the mandatory 'name' attribute. Of couse, an even better solution would
> be to use libsensors to access hwmon attributes.
>
> Thanks,
> Guenter
>
>
_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
^ permalink raw reply [flat|nested] 3+ messages in thread