lm-sensors does not read anything other than Tctl: k10temp-pci-00c3 Adapter: PCI adapter Tctl: +37.0°C in https://github.com/torvalds/linux/blob/master/drivers/hwmon/k10temp.c#L451-L477 there is no check for model 0x8 # cat /sys/devices/system/cpu/modalias cpu:type:x86,ven0002fam0019mod0008:feature:,0000,<lots of features> My processor is a 5955WX but I believe model 0x8 stands for all TR 5000WX series (for example a 5975WX: https://linux-hardware.org/?probe=45a8669840&log=cpuid).
k10temp is quite limited and that's exactly how it works now. A totally different CPU: k10temp-pci-00c3 Adapter: PCI adapter Tctl: +35.6°C
What is that "totally different" (family, model? see the cat command in my original post) CPU? Perhaps it is not supported either? Anyway, I recompiled k10temp by simply adding a case for 0x8 along the rest of the Zen3 stuff: --- linux-6.5.11/drivers/hwmon/k10temp.c 2023-11-08 15:09:07.000000000 +0200 +++ k10linux-6.5.11/drivers/hwmon/k10temp.c 2023-12-09 18:44:44.945290498 +0200 @@ -455,6 +455,7 @@ static int k10temp_probe(struct pci_dev switch (boot_cpu_data.x86_model) { case 0x0 ... 0x1: /* Zen3 SP3/TR */ + case 0x8: /* Zen3 TR (5000WX) */ case 0x21: /* Zen3 Ryzen Desktop */ case 0x50 ... 0x5f: /* Green Sardine */ data->ccd_offset = 0x154; Now I can see this, which is probably correct: k10temp-pci-00c3 Adapter: PCI adapter Tctl: +45.0°C Tccd3: +36.0°C Tccd5: +38.2°C
Tctl is the highest temp for the entire die, other temp sensors are quite inconsequential. Please send the patch via email to the maintainer of the hw sensors subsystem.
Actually, Tctl is used for fan control and it might have offsets for consistent fan policies across the platform. Tccd is the actual, *real* temperature. I tried submitting a patch but failed miserably (never done it before, did not get accepted) and I'd rather not embarass myself any further, nor should I waste the hwmon guys time with my attempts. So, if someone could submit a patch with the 0x8 case (as seen in my earlier post), that would be neat. Here's how it looks under all-core stress-ng test, just to confirm the values look sane both in idle and under load: k10temp-pci-00c3 Adapter: PCI adapter Tctl: +74.8°C Tccd3: +71.5°C Tccd5: +75.8°C
No worries, its not embarrassing have a patch getting rejected, this happens regularly. I will resubmit the patch for you.
Created attachment 305586 [details] Modified patch
Can you check if the modified patch works?
(In reply to Armin Wolf from comment #5) > No worries, its not embarrassing have a patch getting rejected, this happens > regularly. > I will resubmit the patch for you. Thank you very much! (In reply to Armin Wolf from comment #7) > Can you check if the modified patch works? Yes, I can confirm it works. (I actually did compile the modules, instead of just looking at the patch and saying "yep that'll work").
The patch has been accepted, your CPU should be supported by kernel 6.8.
Thanks Armin. Here's a link to the git commit for reference: https://git.kernel.org/pub/scm/linux/kernel/git/groeck/linux-staging.git/commit/?h=hwmon-next&id=62a991b3fcc9529c1a905ee42a4d860c31ca606a