Hi, Il 13/11/19 09:15, Ilie Halip ha scritto: > The module parameter value, when specified, should be used. > When it is not specified, you probably still want to check the > dmi_table. Hmm, that does not solve the problem mentioned by Valdis: if there is no userspace daemon piloting the fan, the module should not do anything, so that the BIOS still has a chance to regulate the fan (which might not mean much: the BIOS seems to be very buggy, and sometimes it won't speed up the fan even when all the cores are busy compiling C++ for minutes and very hot; but at least that's not the kernel module's fault). Maybe another approach is to only disable the BIOS code when the first SET_FAN command is received, so I know that there is someone in userspace. Of course I am not protected from the possibility that the userspace daemon sends wrong commands or crashes, but I am not protected anyway. I rely on other lower level measures (such as, the CPU powering down then it crosses a critical threshold) to protect the hardware from this scenario. Is this a reasonable solution? > Coincidentally, I have a Dell laptop model which might be > affected by a similar issue. I'd be happy to test your patch. That would be good! Thanks, Giovanni. -- Giovanni Mascellani <g.mascellani@xxxxxxxxx> Postdoc researcher - Université Libre de Bruxelles
Attachment:
signature.asc
Description: OpenPGP digital signature
_______________________________________________ Kernelnewbies mailing list Kernelnewbies@xxxxxxxxxxxxxxxxx https://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies
