CPU overheating with recent 14.2 kernel updates
by astrogeek from LinuxQuestions.org on (#4V86P)
With recent kernel updates I have experienced a strange CPU overheating condition on a 32-bit laptop which I use daily.
It began with the 199 kernel, and continues with the 201 kernel. I have not yet installed the 202 which was updated a little while ago.
My first thought was fan problem or heatsink thermal grease, so I opened it up this evening and cleaned the fan and cleaned and reinstalled the heatsink with new thermal compound. But no change.
The reason I say it is a strange problem is because the fan is running and the CPU temperature climbs steadily even with a fairly quiescent system - that is, there is no runaway process eating the CPU.
CPU is Centrino dual core:
Code:processor : 0
vendor_id : GenuineIntel
cpu family : 6
model : 14
model name : Genuine Intel(R) CPU T2050 @ 1.60GHz
Linux 4.4.201-smp #2 SMP Tue Nov 12 20:34:52 CST 2019 i686 Genuine Intel(R) CPU T2050 @ 1.60GHz GenuineIntel GNU/LinuxTypically it runs 20C to 40C with my normal usage patterns.
Beginning with the 199 kernel it is 40C to 50C when I login immediately after boot. I hear the fan and can feel the heat through the keyboard as it climbs through 60C, then 70C over the next few minutes. Both cores always within a few degrees of each other.
Starting a browser or music player will drive it towards 90C, and closing those will slowly cool it back into the 50's or 60's, but not lower.
I have booted to another 14.2 instance on the same system which is using a previous kernel, 4.4.182, and it runs cool as is normal for this machine, so I would rule out a silicon fault for the time being.
But with fan running and no high use processes, I am somewhat at a loss about where to begin looking. I need the machine so may simply revert to the 189 kernel if 202 does not change anything.
Any thoughts...
UPDATE: I normally use the generic-smp kernel with initrd, and load Intel microcode in the initrd. I just tried an initrd without including the microcode, 4.4.201 kernel, and it appears to run cool after a quick test. I am not sure what that tells us but I wonder if the latest mitigations in the kernel conflict with previous microcode changes on this old CPU...
It began with the 199 kernel, and continues with the 201 kernel. I have not yet installed the 202 which was updated a little while ago.
My first thought was fan problem or heatsink thermal grease, so I opened it up this evening and cleaned the fan and cleaned and reinstalled the heatsink with new thermal compound. But no change.
The reason I say it is a strange problem is because the fan is running and the CPU temperature climbs steadily even with a fairly quiescent system - that is, there is no runaway process eating the CPU.
CPU is Centrino dual core:
Code:processor : 0
vendor_id : GenuineIntel
cpu family : 6
model : 14
model name : Genuine Intel(R) CPU T2050 @ 1.60GHz
Linux 4.4.201-smp #2 SMP Tue Nov 12 20:34:52 CST 2019 i686 Genuine Intel(R) CPU T2050 @ 1.60GHz GenuineIntel GNU/LinuxTypically it runs 20C to 40C with my normal usage patterns.
Beginning with the 199 kernel it is 40C to 50C when I login immediately after boot. I hear the fan and can feel the heat through the keyboard as it climbs through 60C, then 70C over the next few minutes. Both cores always within a few degrees of each other.
Starting a browser or music player will drive it towards 90C, and closing those will slowly cool it back into the 50's or 60's, but not lower.
I have booted to another 14.2 instance on the same system which is using a previous kernel, 4.4.182, and it runs cool as is normal for this machine, so I would rule out a silicon fault for the time being.
But with fan running and no high use processes, I am somewhat at a loss about where to begin looking. I need the machine so may simply revert to the 189 kernel if 202 does not change anything.
Any thoughts...
UPDATE: I normally use the generic-smp kernel with initrd, and load Intel microcode in the initrd. I just tried an initrd without including the microcode, 4.4.201 kernel, and it appears to run cool after a quick test. I am not sure what that tells us but I wonder if the latest mitigations in the kernel conflict with previous microcode changes on this old CPU...