Re: [PATCH v3 1/4] platform/x86: bitland-mifs-wmi: Add Xiaomi Book Pro 14 quirk

[Mingyou Chen <qby140326@gmail.com>]

On 11/2/26 00:18, Chris Taraszka wrote:
> The Xiaomi Book Pro 14 (board TM2424) implements SystemPerMode with
> its own values. Its firmware ignores 0x00 and 0x01, and GET returns one
> of 0x02, 0x03, 0x04, 0x09 or 0x0a, so with the default table balanced
> and balanced-performance are no-ops and reading the profile fails with
> -EINVAL whenever the firmware is in 0x04, 0x09 or 0x0a. The failed read
> also aborts system suspend.
>
> The firmware passes the mode to Intel DTT as an OEM design variable.
> The DTT tables in its data vault name the modes and set the sustained
> package power limit (PL1) for each, by power source. For the H484 CPU
> variant (H404 has the same layout with lower limits):
>
>    value  AC                 battery
>    0x02   Silent  15-25 W    Silent  15-25 W
>    0x0a   Silent  15-25 W    Silent  15-25 W
>    0x09   Balance 35-45 W    Balance 32-40 W
>    0x03   Smart   36-37 W    Turbo   50-58 W
>    0x04   Crazy   50-58 W    no entry
>
> Map low-power, balanced and performance to 0x02, 0x09 and 0x04. 0x0a
> selects the same tables as 0x02, and 0x03 gives on battery what 0x04
> gives on AC, so report them as low-power and performance without
> offering them for SET. There is no balanced-performance mode.
>
> The machine is USB-C only and WMI_FN_SYSTEM_AC_TYPE is not implemented,
> so skip the AC type check.
>
> Signed-off-by: Chris Taraszka <chris@miget.com>
> ---
>
> Notes:
>      Tested on the Xiaomi Book Pro 14 (TM2424, BIOS XMAPT4B0P0A0A, H484)
>      with this series on top of v15, loaded on v7.3. Firmware mode (EC 0x68),
>      DTT OEM variable 1 and PL1 a few seconds after each set:
>      
>        AC       low-power    0x02  ODV1 2  25 W
>                 balanced     0x09  ODV1 0  45 W
>                 performance  0x04  ODV1 4  58 W
>        battery  low-power    0x02  ODV1 2  25 W
>                 performance  0x03  ODV1 1  58 W  (with patch 3)
>      
>      With v2 on battery, balanced set 0x03 and PL1 went to 54 W.
>
>   drivers/platform/x86/bitland-mifs-wmi.c | 21 +++++++++++++++++++++
>   1 file changed, 21 insertions(+)
>
> diff --git a/drivers/platform/x86/bitland-mifs-wmi.c b/drivers/platform/x86/bitland-mifs-wmi.c
> index a431d3fef774..63fae12ebd83 100644
> --- a/drivers/platform/x86/bitland-mifs-wmi.c
> +++ b/drivers/platform/x86/bitland-mifs-wmi.c
> @@ -113,7 +113,28 @@ static const struct bitland_mifs_quirks default_quirks = {
>   	.num_profile_entries = ARRAY_SIZE(default_profile_entries),
>   };
>   
> +static const struct bitland_profile_entry tm2424_profile_entries[] = {
> +	{ PLATFORM_PROFILE_LOW_POWER,   0x02 },
> +	{ PLATFORM_PROFILE_LOW_POWER,   0x0a, true },
> +	{ PLATFORM_PROFILE_BALANCED,    0x09 },
> +	{ PLATFORM_PROFILE_PERFORMANCE, 0x04 },
> +	{ PLATFORM_PROFILE_PERFORMANCE, 0x03, true },
> +};
> +
> +static const struct bitland_mifs_quirks tm2424_quirks = {
> +	.profile_entries = tm2424_profile_entries,
> +	.num_profile_entries = ARRAY_SIZE(tm2424_profile_entries),
> +	.skip_ac_type_check = true,
> +};
> +
>   static const struct dmi_system_id bitland_mifs_dmi_table[] = {
> +	{
> +		.matches = {
> +			DMI_EXACT_MATCH(DMI_SYS_VENDOR, "XIAOMI"),
> +			DMI_EXACT_MATCH(DMI_BOARD_NAME, "TM2424"),
> +		},
> +		.driver_data = (void *)&tm2424_quirks,
> +	},
>   	{}
>   };
>   
I'd prefer to have this applied on top of patch 3/4 instead, so the 
infrastructure is in place first.