gabrieldetoni

Subject/context: IT8689E rev 2 — PWM writes accepted but have no effect (fix available in upstream PR #114) Hi ich777, first of all thanks for all your driver plugins! System: Unraid 7.3.1 · Gigabyte B760I AORUS PRO DDR4 · IT8689E revision 2 (detected via your IT87 plugin: it87: Found IT8689E chip at 0xa40 [MMIO at 0x00000000fc000000], revision 2) Problem: Sensors and fan RPM readings work perfectly, but manual PWM control has no effect: echo 1 > pwmN_enable succeeds and persists echo <duty> > pwmN succeeds and persists (value reads back correctly) Fan RPM never changes — on any of the 5 channels Tested with mmio=on (default) and mmio=off ignore_resource_conflict=1 — same result BIOS Smart Fan curves set to flat 40% per the CoolerControl docs workaround — the EC respects the flat curve (fans dropped from 100% to 40%), but sysfs writes still have no effect on top of it This is exactly the symptom tracked in frankcrawford/it87 issue #96 (IT8689E: PWM writes accepted, zero effect, works fine on Windows via LibreHardwareMonitor). The good news: there is now a minimal tested fix upstream — PR #114 (https://github.com/frankcrawford/it87/pull/114). Root cause: on these FEAT_BRIDGE_MMIO chips the internal "vector" fan controller only releases the channel in manual mode when all temp-map-select bits are set; LibreHardwareMonitor writes 0x7f to the base PWM control register on Windows, which is why it works there. The PR is a one-line change in set_pwm_enable() and a reviewer confirmed this method is specifically needed on the IT8689 (the IT8696 is unaffected). Request: would it be possible to get a test build of the IT87 plugin with the PR #114 patch applied (kernel 6.18.33-Unraid / Unraid 7.3.1)? Happy to test and report back — I have a reproducible setup and can validate manual control + enable=2 auto-restore on all channels. Thanks a lot!