You are not logged in.
Hello,
I have an old HP Toshiba laptop that I am using from time to time. I updated it yesterday after few months and noticed that with kernel 5.12.7 the cpu fan is frozen in whatever state it was at boot time. The BIOS does adjust the fans at boot time, though. So for example if the laptop is cold at boot time, the fans will not work and continue so until cpu critical temperature is reached (read from lm_sensors). If then I make a restart, the fans will start at full speed and continue so after reboot. Everything works as it should with the lts-kernel: the fans will start and stop to keep the cpu temperature at about 50°C.
Any ideas where to look to solve this problem?
Last edited by tethys (2021-07-31 04:23:35)
Offline
Something™ takes control over the fan for the OS, compare "lsmod" - hp-wmi has some recent commits reg. "thermal policy", https://github.com/torvalds/linux/commi … 6/hp-wmi.c
You can try blacklisting that (what will likely lose you special keys) or see whether you get userspace control over the fan now, https://wiki.archlinux.org/title/Fan_speed_control
Offline
Thank you for the answer seth. A mistake I made in the first message: it is a Toshiba, not HP laptop. This is what happens when you have too many computers.
I searched (with lsmod) for differences in modules loaded but these are only related to nouveau and sound.
Modules loaded in 5.12 but not in 5.10:
Module Size Used by
snd_intel_sdw_acpi 20480 1 snd_intel_dspcfg
drm_ttm_helper 16384 1 nouveau
Modules loaded in 5.10 but not in 5.12:
Module Size Used by
vfat 20480 1
fat 86016 1 vfat
soundwire_intel 45056 1 snd_intel_dspcfg
soundwire_generic_allocation 16384 1 soundwire_intel
soundwire_cadence 36864 1 soundwire_intel
soundwire_bus 90112 3 soundwire_intel,soundwire_generic_allocation,soundwire_cadence
snd_soc_core 323584 1 soundwire_intel
snd_compress 32768 1 snd_soc_core
ac97_bus 16384 1 snd_soc_core
snd_pcm_dmaengine 16384 1 snd_soc_core
There are however Toshiba-related modules loaded, but the "wmi" module seth suggested does not seem to contain any changes related to thermal policy as in the case of hp-wmi:
$ lsmod | grep -i toshiba
toshiba_acpi 53248 0
sparse_keymap 16384 1 toshiba_acpi
industrialio 90112 1 toshiba_acpi
wmi 36864 3 toshiba_acpi,mxm_wmi,nouveau
toshiba_bluetooth 20480 0
rfkill 28672 4 toshiba_acpi,toshiba_bluetooth,cfg80211
video 53248 2 toshiba_acpi,nouveau
i8042 32768 1 toshiba_acpi
So I left the laptop run hot (with fans not working and kernel 5.12) to see what happens and I noticed that as soon as the "high" temperature, which is the same as the "critical" temperature, was reached the fans started working until the temperature decreased to about 85°C, when they stopped. Now the problem is that this temperature range (85-100°C) is too high, the laptop becomes very hot.
And this is the lts-kernel temperature, which I would like to have:
$ sensors
nouveau-pci-0100
Adapter: PCI adapter
temp1: +45.0°C (high = +95.0°C, hyst = +3.0°C)
(crit = +110.0°C, hyst = +3.0°C)
(emerg = +135.0°C, hyst = +5.0°C)
coretemp-isa-0000
Adapter: ISA adapter
Core 0: +47.0°C (high = +100.0°C, crit = +100.0°C)
Core 1: +48.0°C (high = +100.0°C, crit = +100.0°C)
What part of the 5.12 kernel is responsible for this behavior (given that the lts-kernel works fine)? Is it possible to change the "high" temperature so that the fans are engaged at a lower temperature (without installing AUR tools like NBFC)?
Last edited by tethys (2021-05-29 15:41:02)
Offline
toshiba_acpi is what would provide you fan control.
Which notebook model is that exactly?
Offline
This is a TOSHIBA Satellite A200 from 2007.
Offline
Did you check whether you now have userspace fan control?
Is there a toshiba path under /proc{/acpi}?
Offline
Did you check whether you now have userspace fan control?
I am looking into it. I installed thermald but it said that it does not support my hardware. I'd rather avoid AUR packages and especially NBFC as it has a large dependency (mono). lm_sensors/pwmconfig/fancontrol do not suport my hardware either. I would like to first understand what change between 5.10 and 5.12 caused this problem. Is there any other tool in the official arch repositories that would allow me to control the fan?
Is there a toshiba path under /proc{/acpi}?
Yes, /proc/acpi/toshiba/, and it contains two files:
$ cat /proc/acpi/toshiba/keys
hotkey_ready: 0
hotkey: 0x0000
$ cat /proc/acpi/toshiba/version
driver: 0.24
proc_interface: 1
Offline
The issue does not appear with kernel 5.13.4 anymore, so the problem is solved.
Offline