View previous topic :: View next topic |
Author |
Message |
sunox Tux's lil' helper
Joined: 26 Jan 2022 Posts: 136
|
Posted: Mon Jan 01, 2024 6:16 pm Post subject: thinkpad_acpi thermal readout issues |
|
|
I have 'thinkpad_acpi' enabled in the kernel which populates /proc/acpi/ibm. There is a thinkpad hwmon entry as well which seems to come from the same source.
There appear to be 8 sensors. In hwmon these are labelled: CPU, GPU, and 6 for which I can't find a label. For the 6 unlabelled, when I cat the file I get "no such device or address".
The CPU sensor shows a correct temperature. The rest of the sensors either show 0 or "no such device or address".
Last night while I was messing around trying to get TPM enabled I noticed that my fans stopped coming on when compiling, and the CPU sensor was now giving the "no such device or address" error. I disabled thinkfan, and then the fans were just always on full-speed. I rebooted a few times trying different things, and then the sensor seemingly randomly came back. Right before this I had disabled TPM in the bios which caused the system to fully power off, which may have done the trick.
So to summarize there are two possibly related problems:
1. Most of my sensors do not show anything
2. The CPU sensor 'randomly' stopped working and came back.
Does anyone know what might be going on with either of these? For 2., is it possible that the sensor just got bugged and a power cycle fixed it? |
|
Back to top |
|
|
spica Apprentice
Joined: 04 Jun 2021 Posts: 287
|
Posted: Tue Jan 02, 2024 12:42 am Post subject: Re: thinkpad_acpi thermal readout issues |
|
|
sunox wrote: | There appear to be 8 sensors. In hwmon these are labelled: CPU, GPU, and 6 for which I can't find a label. For the 6 unlabelled, when I cat the file I get "no such device or address". |
How can other people find this "label" on their thinkpads? I ran find /sys/class/hwmon/hwmon*/ -name label and I do not see a node named "label"
Show us your /etc/thinkfan.conf |
|
Back to top |
|
|
pietinger Moderator
Joined: 17 Oct 2006 Posts: 4158 Location: Bavaria
|
Posted: Tue Jan 02, 2024 12:58 am Post subject: |
|
|
sunox,
reading sensors data in Linux can be tricky because you will need the appropriate kernel module(-s) first. Some sensors are connected via i2c/smbus, others use IO ports ... and if you have a very new machine it could be possible there exists no kernel module right now ...
My recommendation for sensors is:
1. Boot with the newest UbuntuLiveCD and do a "sensors-detect"
(In Gentoo you will find it in package "lm_sensors"; see more: https://wiki.gentoo.org/wiki/Lm_sensors )
2. Check also with "lsmod" which modules are loaded under Ubuntu.
3. Try these settings now with your kernel and do a "sensors-detect" again. _________________ https://wiki.gentoo.org/wiki/User:Pietinger |
|
Back to top |
|
|
sunox Tux's lil' helper
Joined: 26 Jan 2022 Posts: 136
|
Posted: Tue Jan 02, 2024 1:03 am Post subject: Re: thinkpad_acpi thermal readout issues |
|
|
spica wrote: | sunox wrote: | There appear to be 8 sensors. In hwmon these are labelled: CPU, GPU, and 6 for which I can't find a label. For the 6 unlabelled, when I cat the file I get "no such device or address". |
How can other people find this "label" on their thinkpads? I ran find /sys/class/hwmon/hwmon*/ -name label and I do not see a node named "label"
Show us your /etc/thinkfan.conf |
The 'name' in hwmon4 is 'thinkpad'. For each temp sensor I get an "*_input" and "*_label" (e.g. temp1_input and temp1_label). This is a standard hwmon thing, no?
I don't think my thinkfan is the issue but I could be wrong. It stopped working because it depends on the temp readouts in /proc/acpi/ibm to decide how fast to spin the fan.
Code: |
sensors:
- tpacpi: /proc/acpi/ibm/thermal
indices: [0]
fans:
- tpacpi: /proc/acpi/ibm/fan
levels:
- [0, 0, 55]
- [1, 55, 60]
- [2, 60, 65]
- [3, 65, 70]
- [4, 70, 75]
- [5, 75, 80]
- [7, 80, 32767]
|
Should maybe also note that I have a fairly aggressive undervolt going on, and I had just bumped it down a bit more before the CPU temp sensor stopped working.
Last edited by sunox on Tue Jan 02, 2024 1:12 am; edited 3 times in total |
|
Back to top |
|
|
sunox Tux's lil' helper
Joined: 26 Jan 2022 Posts: 136
|
Posted: Tue Jan 02, 2024 1:06 am Post subject: |
|
|
pietinger wrote: | sunox,
reading sensors data in Linux can be tricky because you will need the appropriate kernel module(-s) first. Some sensors are connected via i2c/smbus, others use IO ports ... and if you have a very new machine it could be possible there exists no kernel module right now ...
My recommendation for sensors is:
1. Boot with the newest UbuntuLiveCD and do a "sensors-detect"
(In Gentoo you will find it in package "lm_sensors"; see more: https://wiki.gentoo.org/wiki/Lm_sensors )
2. Check also with "lsmod" which modules are loaded under Ubuntu.
3. Try these settings now with your kernel and do a "sensors-detect" again. |
Thanks, I will give this a try at some point. The laptop is from 2018 so pretty old now. There's a good chance I didn't include the appropriate model. I am okay at making sure I have support for all the 'big' pieces (CPU GPU etc) but when it gets to more nitty gritty motherboard stuff I honestly don't know what hardware I have or what it requires. If there is an easy way to find this out please let me know! Is lspci good enough? |
|
Back to top |
|
|
pietinger Moderator
Joined: 17 Oct 2006 Posts: 4158 Location: Bavaria
|
Posted: Tue Jan 02, 2024 1:15 am Post subject: |
|
|
sunox wrote: | [...] If there is an easy way to find this out please let me know! Is lspci good enough? |
No, "lspci" shows to less for this; you need "lsmod" (from ubuntu). Just try "sensors-detect" to get a feeling what it all could be (and say "no" at the end when it wants create a modules-list ) _________________ https://wiki.gentoo.org/wiki/User:Pietinger |
|
Back to top |
|
|
sunox Tux's lil' helper
Joined: 26 Jan 2022 Posts: 136
|
Posted: Tue Jan 02, 2024 1:30 am Post subject: |
|
|
pietinger wrote: | sunox wrote: | [...] If there is an easy way to find this out please let me know! Is lspci good enough? |
No, "lspci" shows to less for this; you need "lsmod" (from ubuntu). Just try "sensors-detect" to get a feeling what it all could be (and say "no" at the end when it wants create a modules-list ) |
I actually did this yesterday, and said "yes" to most of the probes but skipped the last few i2c ones, and all it found was something called "coretemp". I'll give the ubuntu thing a try.
And my lspci question was about trying to figure out which hardware I have. I feel like if I knew my hardware better I would be in better position to find out what modules to include.Or is it always better to just see what modules load when running a kitchen sink distro? |
|
Back to top |
|
|
pietinger Moderator
Joined: 17 Oct 2006 Posts: 4158 Location: Bavaria
|
Posted: Tue Jan 02, 2024 1:41 am Post subject: |
|
|
sunox wrote: | [...] but skipped the last few i2c ones, and all it found was something called "coretemp". I'll give the ubuntu thing a try. |
Then you miss at least CONFIG_SENSORS_DRIVETEMP=y to get the temperature of your harddisk
Do you have "I2C device interface" and the correct PINCTRL module enabled in your kernel config ? (Do you have an AMD or Intel machine?)
sunox wrote: | Or is it always better to just see what modules load when running a kitchen sink distro? |
When I did my last installation last May (2023) I used our GentooMinimalInstallationCD and also UbuntuLive to check both "lsmod" s ... I choosed Ubuntu because there almost every kernel module is enabled. _________________ https://wiki.gentoo.org/wiki/User:Pietinger |
|
Back to top |
|
|
sunox Tux's lil' helper
Joined: 26 Jan 2022 Posts: 136
|
Posted: Tue Jan 02, 2024 1:59 am Post subject: |
|
|
My laptop is intel. I'll look for the intel i2c device interface.
My 'base' config was the one used by the Gentoo live cd but it definitely leaves out a few things. I'll be sure to do the ubuntu thing at some point, maybe tomorrow. If a module loads is it guaranteed that it is 'correct' for my system? |
|
Back to top |
|
|
sunox Tux's lil' helper
Joined: 26 Jan 2022 Posts: 136
|
Posted: Tue Jan 02, 2024 2:11 am Post subject: |
|
|
One strange and maybe related thing is I get error messages when booting related to smbus:
Code: | [ 10.926311] i801_smbus 0000:00:1f.4: Transaction timeout
[ 10.928391] i801_smbus 0000:00:1f.4: Failed terminating the transaction
[ 10.928414] i801_smbus 0000:00:1f.4: SMBus is busy, can't use it! |
I looked into this a bit when I first installed, and this is what 'lspci -k' shows for that device:
Code: | 00:1f.4 SMBus: Intel Corporation Cannon Point-LP SMBus Controller (rev 11)
Subsystem: Lenovo Cannon Point-LP SMBus Controller
Kernel driver in use: i801_smbus
Kernel modules: i2c_i801 |
I google the device and on the page it seems to say that it's supposed to use i2c_i801 module.
https://linux-hardware.org/?id=pci:8086-9da3-1025-128d
I tried blacklisting the i801_smbus module and I forget exactly what happened but I remember things got worse somehow. |
|
Back to top |
|
|
sunox Tux's lil' helper
Joined: 26 Jan 2022 Posts: 136
|
Posted: Tue Jan 02, 2024 3:32 am Post subject: |
|
|
I booted into xubuntu and found the same issues with hwmon 'thinkpad' thermal readouts. Could this mean the sensors are damaged or something? I think I've heard they can be flaky.
I found a few extra kernel modules to load so far, including one for an SPI device that was totally inactive, so thanks again for the tip. |
|
Back to top |
|
|
pietinger Moderator
Joined: 17 Oct 2006 Posts: 4158 Location: Bavaria
|
|
Back to top |
|
|
pietinger Moderator
Joined: 17 Oct 2006 Posts: 4158 Location: Bavaria
|
Posted: Tue Jan 02, 2024 12:29 pm Post subject: |
|
|
sunox wrote: | I booted into xubuntu and found the same issues with hwmon 'thinkpad' thermal readouts. Could this mean the sensors are damaged or something? I think I've heard they can be flaky. |
Hmm ... I always think that if something doesn't work on Ubuntu, it's unlikely to work on Gentoo. But you should never say never. Sometimes it helped to wait for a new kernel. _________________ https://wiki.gentoo.org/wiki/User:Pietinger |
|
Back to top |
|
|
sunox Tux's lil' helper
Joined: 26 Jan 2022 Posts: 136
|
Posted: Tue Jan 02, 2024 5:32 pm Post subject: |
|
|
Thanks, I do have LPSS enabled. I found something which suggests it might be a problem with my specific model of laptop. I could maybe try to find a way around it in hopes it would solve the thermal sensor issue, but things work well enough that I guess I'm happy to leave it alone.
I also enabled a few more kernel options after reading the post you linked. |
|
Back to top |
|
|
spica Apprentice
Joined: 04 Jun 2021 Posts: 287
|
Posted: Tue Jan 02, 2024 6:44 pm Post subject: Re: thinkpad_acpi thermal readout issues |
|
|
sunox wrote: | The 'name' in hwmon4 is 'thinkpad'. For each temp sensor I get an "*_input" and "*_label" (e.g. temp1_input and temp1_label). This is a standard hwmon thing, no? |
It is not mandatory, it is implementation specific.
Code: | temp[1-*]_label Suggested temperature channel label.
Text string
Should only be created if the driver has hints about what
this temperature channel is being used for, and user-space
doesn't. In all other cases, the label is provided by
user-space.
RO |
|
|
Back to top |
|
|
|