Thanks, and sorry for the off-list reply - adding [email protected] to CC now and validating the challenge.
I checked hw.perfpolicy: it was already set to auto,auto here, so that is not the cause. At normal warm idle the fan sits at 3978 RPM with the thermal zone at 51 C - calm thermals, nothing demanding running. So the EC fan curve is running hard regardless of perf policy. The core gap is unchanged: acpithinkpad(4) exposes the fan only as a read-only sensor, so there is no interface to step it down. The same T490 runs quiet under FreeBSD 14.4 and 15, where acpi_ibm(4) writes the fan level via the EC HFSP register - so the hardware supports a write path that acpithinkpad does not implement. That is really the question: whether writable fan-level control in acpithinkpad(4) is feasible. Happy to test patches or run diagnostics on this exact machine. Thanks again for the help getting this onto the list properly. Sent with Proton Mail secure email. On Friday, June 26th, 2026 at 10:30, stephane Tranchemer <[email protected]> wrote: > Hello, > > you sent your answer to me only, include [email protected] in CC, then > validate the challenge, if you want others on the mailing list to see it > too :) >
