Bug 107628

First off, I want to state that I'm aware that I am filing a report that concerns a Compaq Presario CQ60 which a) is close to 10 years old, and b) may well have the worst GPU cooling setup in the history of computing. I will therefore completely understand if everybody here concludes they have better ways to spend their time than worrying about this.

That said, the CQ60 has an GeForce 8200M G built in and I'm running Gentoo x86_64 on said laptop with kernel version 4.14.63. 

When I boot the laptop with nouveau either built into the kernel or as a module, as soon as the driver loads, the screen is continuously flooded with messages relating to the GPU temperature either hitting or going below a range of thresholds. The messages have the format: 

[<timestamp>] nouveau 0000:02:00.0: therm: temperature (<x> C) hit/went below the '<threshold>' threshold

The temperatures reported vary wildly between subsequent messages, as in literally jumping from between below 0 and well above 150 degrees (and vice versa).

When this happens I am unable to use Ctrl-Alt-Del to reboot and have to do a hard power-down for it to stop. I can boot successfully provided I pass nouveau.modeset=0 on the kernel command line.

I am currently unable to provide dmesg output as the problem occurs during boot before a command prompt is presented that I could use to secure the log. I do have a mobile phone video recording available of the laptop screen when the problem occurs, in case that helps.

Windows 10 runs fine on this laptop with the nVidia drivers it pulls off the Internet during installation. Under Windows, HWiNFO reports mostly stable GPU temperatures of around 85 C.