[REGRESSION] platform/x86: bitland-mifs-wmi: platform profiles cannot be changed on USB-C-only RedmiBook Pro 15 (TM2017)

[REGRESSION] platform/x86: bitland-mifs-wmi: platform profiles cannot be changed on USB-C-only RedmiBook Pro 15 (TM2017)

[Serhiy Shliapuhin]

Hi Mingyou,

On my RedmiBook Pro 15 (2021) none of the platform profiles exposed
by bitland-mifs-wmi can actually be selected. "performance" and
"balanced-performance" are always rejected with -EOPNOTSUPP, even
while the laptop is charging, and "low-power" is accepted but has no
effect. As a result power-profiles-daemon can only stay in Balanced.

This is a regression. With Ubuntu kernel 7.0.0-38-generic the
Performance and Power Saver profiles could be selected via
power-profiles-daemon without errors (only the intel_pstate CPU
driver was used, as no platform_profile handler was bound on this
machine). After upgrading to 7.3.0-9-generic (Ubuntu 26.10),
bitland-mifs-wmi binds on this laptop, power-profiles-daemon also
uses the platform_profile driver, and profile switching no longer
works.

Last good: 7.0.0-38-generic (Ubuntu)
First bad: 7.3.0-9-generic (Ubuntu)

#regzbot introduced: dc1ec4fa86b2

Hardware / software:
  DMI: TIMI RedmiBook Pro 15/TM2017, BIOS RMATG5B0P0909 03/31/2021
  Kernel: 7.3.0-9-generic (Ubuntu 26.10)

1) DC-only modes are never available

This model charges exclusively over USB-C; it has no dedicated DC
barrel jack. The driver's DC power check (as documented in
Documentation/wmi/devices/bitland-mifs-wmi.rst, "Performance" and
"Full-speed" require a DC power source, not USB-C/PD) can therefore
never be satisfied on this machine.

Platform profile handler:
  $ 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

Power supplies while the charger is connected (no ACPI AC device):
  $ ls /sys/class/power_supply/
  BAT0  ucsi-source-psy-USBC000:001  ucsi-source-psy-USBC000:002
  $ cat /sys/class/power_supply/ucsi-source-psy-USBC000:002/online
  1

  $ echo performance | sudo tee /sys/firmware/acpi/platform_profile
  tee: /sys/firmware/acpi/platform_profile: Operation not supported

  $ powerprofilesctl set performance
  Failed to activate platform driver 'platform_profile': Failed to
  write to acpi_platform_profile: Error writing
  '/sys/firmware/acpi/platform_profile': Operation not supported (15)

I have not been able to verify the behaviour under Windows.

2) "low-power" is silently ignored

Writing "low-power" is accepted without error, but reading the
attribute back immediately returns "balanced":

  $ echo low-power | sudo tee /sys/firmware/acpi/platform_profile
  low-power
  $ cat /sys/firmware/acpi/platform_profile
  balanced

power-profiles-daemon then sees "balanced" and reverts its own
profile, so Power Saver cannot be selected either. It looks like on
this firmware the mode writes are either ignored or the current mode
is read back incorrectly.

Summary of all writable modes on this machine (charger connected):
  low-power             -> accepted, reads back "balanced"
  balanced-performance  -> -EOPNOTSUPP
  performance           -> -EOPNOTSUPP

I am ready to provide acpidump or any other information you need.

Thanks,
Serhiy Shliapuhin

Re: [REGRESSION] platform/x86: bitland-mifs-wmi: platform profiles cannot be changed on USB-C-only RedmiBook Pro 15 (TM2017)

[Thorsten Leemhuis]

On 10/20/26 07:12, Serhiy Shliapuhin wrote:
> 
> On my RedmiBook Pro 15 (2021) none of the platform profiles exposed
> by bitland-mifs-wmi can actually be selected. "performance" and
> "balanced-performance" are always rejected with -EOPNOTSUPP, even
> while the laptop is charging, and "low-power" is accepted but has no
> effect. As a result power-profiles-daemon can only stay in Balanced.
> 
> This is a regression. With Ubuntu kernel 7.0.0-38-generic [...]

Can you avoid it by blacklisting bitland-mifs-wmi? And does it still
happen with vanilla 7.3? 7.0 is long EOL upstream (and 7.2 soon, too).

Ciao, Thorsten

Re: [REGRESSION] platform/x86: bitland-mifs-wmi: platform profiles cannot be changed on USB-C-only RedmiBook Pro 15 (TM2017)

[Serhiy Shliapuhin]

> Can you avoid it by blacklisting bitland-mifs-wmi?

Yes. With bitland-mifs-wmi blacklisted no platform_profile handler is
registered, and power-profiles-daemon switches all three profiles via
intel_pstate EPP again, same as before the driver was added.

The same can be achieved by running power-profiles-daemon
with --block-driver=platform_profile, which is what I use now.

> And does it still happen with vanilla 7.3? 7.0 is long EOL upstream
> (and 7.2 soon, too).

Yes, it is reproducible with vanilla 7.3-rc3 (Ubuntu mainline build,
no distro patches; newer mainline builds are not available yet -
bitland-mifs-wmi.c has not changed since 2026-07-03 anyway).

  $ uname -r
  7.3.0-070300rc3-generic
  $ cat /sys/class/platform-profile/platform-profile-0/name
  bitland-mifs-wmi

  low-power             -> accepted, reads back "balanced"
  balanced-performance  -> -EOPNOTSUPP, stays "balanced"
  performance           -> -EOPNOTSUPP, stays "balanced"

Understood regarding 7.0; I only mentioned it as the last kernel
where profile switching worked on this machine, before
bitland-mifs-wmi was added.

Thanks,
Serhiy Shliapuhin

On Wed, Oct 21, 2026 at 5:56 PM Thorsten Leemhuis
<regressions@leemhuis.info> wrote:
>
> On 10/20/26 07:12, Serhiy Shliapuhin wrote:
> >
> > On my RedmiBook Pro 15 (2021) none of the platform profiles exposed
> > by bitland-mifs-wmi can actually be selected. "performance" and
> > "balanced-performance" are always rejected with -EOPNOTSUPP, even
> > while the laptop is charging, and "low-power" is accepted but has no
> > effect. As a result power-profiles-daemon can only stay in Balanced.
> >
> > This is a regression. With Ubuntu kernel 7.0.0-38-generic [...]
>
> Can you avoid it by blacklisting bitland-mifs-wmi? And does it still
> happen with vanilla 7.3? 7.0 is long EOL upstream (and 7.2 soon, too).
>
> Ciao, Thorsten
>
>

Re: [REGRESSION] platform/x86: bitland-mifs-wmi: platform profiles cannot be changed on USB-C-only RedmiBook Pro 15 (TM2017)

[Mingyou Chen]

On 10/22/26 16:08, Serhiy Shliapuhin wrote:
>> Can you avoid it by blacklisting bitland-mifs-wmi?
> Yes. With bitland-mifs-wmi blacklisted no platform_profile handler is
> registered, and power-profiles-daemon switches all three profiles via
> intel_pstate EPP again, same as before the driver was added.
>
> The same can be achieved by running power-profiles-daemon
> with --block-driver=platform_profile, which is what I use now.
>
>> And does it still happen with vanilla 7.3? 7.0 is long EOL upstream
>> (and 7.2 soon, too).
> Yes, it is reproducible with vanilla 7.3-rc3 (Ubuntu mainline build,
> no distro patches; newer mainline builds are not available yet -
> bitland-mifs-wmi.c has not changed since 2026-07-03 anyway).
>
>    $ uname -r
>    7.3.0-070300rc3-generic
>    $ cat /sys/class/platform-profile/platform-profile-0/name
>    bitland-mifs-wmi
>
>    low-power             -> accepted, reads back "balanced"
>    balanced-performance  -> -EOPNOTSUPP, stays "balanced"
>    performance           -> -EOPNOTSUPP, stays "balanced"
>
> Understood regarding 7.0; I only mentioned it as the last kernel
> where profile switching worked on this machine, before
> bitland-mifs-wmi was added.
>
> Thanks,
> Serhiy Shliapuhin
>
> On Wed, Oct 21, 2026 at 5:56 PM Thorsten Leemhuis
> <regressions@leemhuis.info> wrote:
>> On 10/20/26 07:12, Serhiy Shliapuhin wrote:
>>> On my RedmiBook Pro 15 (2021) none of the platform profiles exposed
>>> by bitland-mifs-wmi can actually be selected. "performance" and
>>> "balanced-performance" are always rejected with -EOPNOTSUPP, even
>>> while the laptop is charging, and "low-power" is accepted but has no
>>> effect. As a result power-profiles-daemon can only stay in Balanced.
>>>
>>> This is a regression. With Ubuntu kernel 7.0.0-38-generic [...]
>> Can you avoid it by blacklisting bitland-mifs-wmi? And does it still
>> happen with vanilla 7.3? 7.0 is long EOL upstream (and 7.2 soon, too).
>>
>> Ciao, Thorsten
>>
Hi, We're tracking this issue in the recent patch series.

https://lore.kernel.org/platform-driver-x86/20261021123414.25516-1-qby140326@gmail.com/T/#t