Re: [REGRESSION] bitland-mifs-wmi: power-saver write is accepted but profile remains balanced on Redmi Book Pro 14 2022

[top posting, to facilitate processing]

Hmm, nothing happened here.

Mingyou Chen, this seems to be cause by a change of yours. Not totally
sure yet if this qualifies as regression (sounds like one), but could
you maybe share your view on all of this?

Side note: Robert, have you tried if latest 7.2-rc works any better?

Ciao, Thorsten

On 7/13/26 20:35, Robert Munro wrote:
> To: Mingyou Chen <[email protected]>, Hans de Goede
> <[email protected]>, Ilpo Järvinen <[email protected]>
> Cc: [email protected], [email protected]
> Subject: [REGRESSION] bitland-mifs-wmi: power-saver write is accepted
> but profile remains balanced on Redmi Book Pro 14 2022
> 
> Hello,
> 
> Linux 7.1's new bitland-mifs-wmi driver binds on my TIMI Redmi Book Pro
> 14 2022 (TM2107) and exposes platform profiles, but selecting power-
> saver is silently ineffective. The userspace request returns success
> while the sysfs profile never leaves "balanced".
> 
> This reproduces without KDE:
> 
>     $ cat /sys/class/platform-profile/platform-profile-0/name
>     bitland-mifs-wmi
> 
>     $ cat /sys/firmware/acpi/platform_profile_choices
>     low-power balanced balanced-performance performance
> 
>     $ cat /sys/firmware/acpi/platform_profile
>     balanced
> 
>     $ powerprofilesctl set power-saver
>     $ echo $?
>     0
> 
>     $ powerprofilesctl get
>     balanced
> 
>     $ cat /sys/firmware/acpi/platform_profile
>     balanced
> 
>     $ cat /sys/devices/system/cpu/cpufreq/policy0/
> energy_performance_preference
>     balance_performance
> 
> I repeated the request while polling platform_profile every 50 ms. The
> only observed value was "balanced"; no transient "low-power" state
> appeared. The request completed in 162 ms with exit status 0. power-
> profiles-daemon also persisted Profile=balanced afterward. There were no
> relevant kernel or daemon log messages.
> 
> Expected result:
> 
> The advertised low-power profile is selected, platform_profile reads
> "low-power", and the AMD P-State EPP becomes power-oriented.
> 
> Actual result:
> 
> The operation reports success, but firmware, power-profiles-daemon, and
> all AMD CPU policies remain balanced.
> 
> System information:
> 
>     Distribution: CachyOS
>     Kernel: 7.1.3-2-cachyos
>     DMI vendor/product/board: TIMI / Redmi Book Pro 14 2022 / TM2107
>     BIOS: TIMI RMARB4B0P1010, 2023-06-28
>     CPU: AMD Ryzen 5 6600H with Radeon Graphics
>     bitland_mifs_wmi srcversion: 0C0A58A1AE25437010C0B5E
>     power-profiles-daemon: 0.30-3
>     Power source during reproduction: AC online
>     Active profile holds: none
>     Other power managers (TLP, TuneD, auto-cpufreq): none
> 
> powerprofilesctl identifies both drivers:
> 
>     power-saver:
>       CpuDriver: amd_pstate
>       PlatformDriver: platform_profile
> 
> The kernel taint value is 512 because of an unrelated AMDGPU display
> warning during boot (dcn31_program_compbuf_size); bitland_mifs_wmi
> itself is not tainted and emitted no warning during reproduction.
> 
> This began after the 2026-06-24 update from Linux 7.0.12 to 7.1.1, which
> introduced bitland-mifs-wmi (dc1ec4fa86b2). Power Saver worked before
> that update. My installed 6.18 LTS kernel does not contain this driver,
> but I have not yet performed an LTS or vanilla-mainline A/B test, so I
> am explicitly reporting this as reproduced on a CachyOS kernel.
> 
> 
> Regards,
> Robert
> 
> Assisted-by: OpenAI Codex:GPT-5
> 
> Robert Munro
> Director, Executive Team
> [email protected] <mailto:[email protected]>
> ethicsbowl.ca <https://ethicsbowl.ca>
> Ethics Bowl Canada
>