lm_sensors
lm_sensors (Linux monitoring sensors) is a free and open-source application that provides tools and drivers for monitoring temperatures, voltage, and fans. This document explains how to install, configure, and use lm_sensors.
Installation
Install the lm_sensors package.
Setup
Use sensors-detect as root to detect and generate a list of kernel modules:
Enter
), unless you know exactly what you are doing. See #Laptop screen issues after running sensors-detect.# sensors-detect
It will ask to probe for various hardware. The "safe" answers are the defaults, so just hitting Enter
to all the questions will generally not cause any problems. This will create the /etc/conf.d/lm_sensors
configuration file which is used by to automatically load kernel modules on boot.
When the detection is finished, a summary of the probes is presented.
Example:
Running sensors
Example running :
$ sensors
coretemp-isa-0000 Adapter: ISA adapter Core 0: +35.0°C (crit = +105.0°C) Core 1: +32.0°C (crit = +105.0°C) w83l771-i2c-0-4c Adapter: SMBus nForce2 adapter at 4d00 temp1: +28.0°C (low = -40.0°C, high = +70.0°C) (crit = +85.0°C, hyst = +75.0°C) temp2: +37.4°C (low = -40.0°C, high = +70.0°C) (crit = +110.0°C, hyst = +100.0°C)
Adding DIMM Temperature sensors
To find the temperature sensors of DIMMs, install the package. Once installed, load the kernel module.
# modprobe i2c_dev
To show all the columns, use i2cdetect as root:
Otherwise, its output will appear as follows:
i2c-2 unknown SMBus PIIX4 adapter port 2 at 0b00 N/A i2c-2 unknown SMBus PIIX4 adapter port 1 at 0b20 N/A i2c-0 unknown SMBus PIIX4 adapter port 0 at 0b00 N/A
In my system, RAM sticks connected to the bus is SMBus 0. The i2cdetect command will show the devices that connected to the bus. The argument means use i2c-0 smbus. You can check other buses if needed.
RAM SPD's are start from address 0x50 and RAM temp sensors start from 0x18 at same bus. In my system, there are 2 DIMMs available. So address of 0x18 and 0x19 are DIMMs temp sensors.
After found this info, to read temperatures of RAM sticks, we need kernel module loaded. After that you need to tell to module that which addresses are need to used. This process consists of writing and to smbus_path
. For example:
# modprobe jc42 # echo jc42 0x18 > /sys/bus/i2c/devices/i2c-0/new_device # echo jc42 0x19 > /sys/bus/i2c/devices/i2c-0/new_device
After that your ram sticks temperature will be visible on command:
jc42-i2c-0-19 Adapter: SMBus PIIX4 adapter port 0 at 0b00 temp1: +50.7°C (low = +0.0°C) ALARM (HIGH, CRIT) (high = +0.0°C, hyst = +0.0°C) (crit = +0.0°C, hyst = +0.0°C) jc42-i2c-0-18 Adapter: SMBus PIIX4 adapter port 0 at 0b00 temp1: +51.8°C (low = +0.0°C) ALARM (HIGH, CRIT) (high = +0.0°C, hyst = +0.0°C) (crit = +0.0°C, hyst = +0.0°C)
Reading SPD values from memory modules (optional)
To read the SPD timing values from memory modules, install the package. Once installed, load the eeprom
kernel module.
# modprobe eeprom
Finally, view memory information with .
Here is partial output from one machine:
Using sensor data
Graphical front-ends
There are a variety of front-ends for sensors data.
For specific Desktop environments:
- Sensors (Xfce4 panel plugin) — Hardware sensors plugin for the Xfce panel.
sensord
There is an optional daemon called sensord (included with the lm_sensors package) which can log data to a round robin database (rrd) and later visualize graphically. See the sensord(8) man page for details.
Tips and tricks
Adjusting values
In some cases, the data displayed might be incorrect or users may wish to rename the output. Use cases include:
- Incorrect temperature values due to a wrong offset (i.e. temps are reported 20 °C higher than actual).
- Users wish to rename the output of some sensors.
- The cores might be displayed in an incorrect order.
All of the above (and more) can be adjusted by overriding the package provides settings in by creating wherein any number of tweaks will override the default values. It is recommended to rename 'foo' to the motherboard brand and model but this naming nomenclature is optional.
Example 1. Adjusting temperature offsets
This is a real example on a Zotac ION-ITX-A-U motherboard. The coretemp values are off by 20 °C (too high) and are adjusted down to Intel specs.
Run with the switch to see what options are available for each physical chip (raw mode). If some of the raw labels you are presented seem not to be configurable, look at the directory tree. Each device mentioned there has a file, which can be used to match the device it is referring to. And then try the labels referred to by that directory.
$ sensors -u
coretemp-isa-0000 Adapter: ISA adapter Core 0: temp2_input: 57.000 temp2_crit: 125.000 temp2_crit_alarm: 0.000 Core 1: temp3_input: 55.000 temp3_crit: 125.000 temp3_crit_alarm: 0.000 ...
Create the following file overriding the default values:
Now invoking shows the adjust values:
Example 2. Renaming labels
This is a real example on an Asus A7M266. The user wishes more verbose names for the temperature labels temp1
and :
Create the following file to override the default values:
Now invoking shows the adjust values:
Example 3. Renumbering cores for multi-CPU systems
This is a real example on an HP Z600 workstation with dual Xeons. The actual numbering of physical cores is incorrect: numbered 0, 1, 9, 10 which is repeated into the second CPU. Most users expect the core temperatures to report out in sequential order, i.e. 0,1,2,3,4,5,6,7.
Again, run with the switch to see what options are available for each physical chip:
Create the following file overriding the default values:
/etc/sensors.d/HP_Z600
chip "coretemp-isa-0000" label temp2 "Core 0" label temp3 "Core 1" label temp11 "Core 2" label temp12 "Core 3" chip "coretemp-isa-0004" label temp2 "Core 4" label temp3 "Core 5" label temp11 "Core 6" label temp12 "Core 7"
Now invoking shows the adjust values:
Automatic lm_sensors deployment
Users wishing to deploy lm_sensors on multiple machines can use the following to accept the defaults to all questions:
# sensors-detect --auto
S.M.A.R.T. drive temperature
Since kernel 5.6 the module will report SATA/SAS temperature through hwmon, but does not automatically detect this so the module must be manually loaded.
# modprobe drivetemp
You should now see entries similar to this in your output:
You can now load the module at boot. Alternatively, manually add it to the line of /etc/conf.d/lm-sensors
. Do note it will not be added automatically when will be allowed to write this file again.
Troubleshooting
K10Temp module
Some K10 processors have issues with their temperature sensor. See the k10temp documentation for more information.
On affected machines the module will report "unreliable CPU thermal sensor; monitoring disabled". To force monitoring anyway, you can run the following:
# rmmod k10temp # modprobe k10temp force=1
Confirm that the sensor is in fact valid and reliable. If it is, can edit and add:
options k10temp force=1
This will allow the module to load at boot.
Asus B450M-A/A320M-K/A320M-K-BR motherboards
These motherboards use a IT8655E chip, which is not supported by the it87 kernel driver, as of Nov 2020 . However, it is supported by the upstream version of the kernel driver . The DKMS variant is contained in .
Asus B450/X399/X470 motherboards with AM4 Socket
Some recent Asus motherboards use a ITE IT8665E chip, accessing the temperature, fan and voltage sensors may require the module. It is part of the mainline kernel since 5.17: load the kernel module which uses the UEFI interface and may require a BIOS update on some boards .
Alternatively, the module reads the values from the chip directly, install and load the kernel module.
Asus H97/Z97/Z170/Z370i/X570/B550 motherboards
With some recent Asus motherboards, fan and voltage sensor access may require the nct6775
kernel module to be loaded.
You may also need to add the following kernel parameter:
acpi_enforce_resources=lax
See https://bugzilla.kernel.org/show_bug.cgi?id=204807 for more information.
Note: Starting with Kernel 5.16 , the above kernel parameter is no longer be required for most boards and should be avoided.
Asrock Deskmini H470
The STX board of the Deskmini H470 uses a NCT6683 chip, for accessing the temperature, fan and voltage sensors the loading of module is required.
For proper values of the module have a module config file created:
Gigabyte B250/Z370/B450M/B560M/Z690 motherboards
Some Gigabyte motherboards use the ITE IT8686E, ITE8689 (for B560) or ITE8689E (for Z690) chip, which is not supported by the it87 kernel driver, as of May 2019 . However, it is supported by the upstream version of the kernel driver . The DKMS variant is contained in . As with #Asus H97/Z97/Z170/Z370i/X570/B550 motherboards, a kernel parameter is required before attempting to install the module:
acpi_enforce_resources=lax
Furthermore, supply the id of the chip when loading the module as follows:
# modprobe it87 force_id=0x8686 or # modprobe it87 force_id=0x8689 # for B560 # modprobe it87 force_id=0x8628 # for Z690
Or you can load the module during boot process by creating the following two files:
For Z690
For others
Once the module is loaded you can use the sensors tool to probe the chip. Now you can also use fancontrol to control the speed step of your case fan.
Optionally installation of may allow greater fine tuning of the motherboard's cooling system. However, it does disable the default k10temp module.
Gigabyte GA-J1900N-D3V
This motherboard uses the ITE IT8620E chip (useful also to read voltages, mainboard temp, fan speed). As of October 2014, lm_sensors has no driver support for chip ITE IT8620E . lm_sensors developers had a report that the chip is somewhat compatible with the IT8728F for the hardware monitoring part. However, as of August 2016, lists the IT8620E as supported.
You can load the module at runtime with modprobe:
$ modprobe it87 force_id=0x8728
Or you can load the modules during boot process by creating the following two files:
/etc/modules-load.d/it87.conf
it87
Once the module is loaded you can use the sensors tool to probe the chip.
Now you can also use fancontrol to control the speedsteps of your case fan.
Laptop screen issues after running sensors-detect
This is caused by lm-sensors messing with the Vcom values of the screen while probing for sensors. It has been discussed and solved at the forums already: https://bbs.archlinux.org/viewtopic.php?id=193048. However, make sure to read through the thread carefully before running any of the suggested commands.
i2c bus errors on AMD Navi 2 GPUs
There is currently a bug in the way the kernel handles reading the i2c bus on AMD Navi 2 GPUs. The bus currently can only be used with EEPROMs and trying to use it with other devices will cause it to fail. This can cause crashes, black screens, and even cause the card to behave oddly like unable to switch power states. Its currently advised not to scan the i2c bus if you have a Navi 2 based card. You can read more here: https://gitlab.freedesktop.org/drm/amd/-/issues/1470