[PATCH v2 0/2] acpi: battery: Query BCT method and steamdeck

[PATCH v2 0/2] acpi: battery: Query BCT method and steamdeck

[Umang Jain]

Short series to query Battery charge time ("_BCT") from ACPI tables
(1/2).

Patch 2/2 handles the case where steamdeck devices can report "Charging"
even when low power PSU is connected which is discharging the device.

Changes in v2:
- comment improvements
- split off BCT method handling into separate patch

v1:
https://lore.kernel.org/all/20260918135628.2322188-1-uajain@igalia.com/

Umang Jain (2):
  acpi: battery: Query Battery charge time ("_BCT") method
  acpi: battery: Handle Valve steamdeck devices quirk

 drivers/acpi/battery.c | 45 +++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 44 insertions(+), 1 deletion(-)

-- 
2.55.0

[PATCH v2 1/2] acpi: battery: Query Battery charge time ("_BCT") method

[Umang Jain]

Add plumbing for querying the Battery Charge Time ("_BCT") method.
If present, set ACPI_BATTERY_BCT_PRESENT in battery flags to denote
that the method is supported in ACPI.

Signed-off-by: Umang Jain <uajain@igalia.com>
---
 drivers/acpi/battery.c | 17 +++++++++++++++++
 1 file changed, 17 insertions(+)

diff --git a/drivers/acpi/battery.c b/drivers/acpi/battery.c
index 670853ec3a4d..a1b7a821abaf 100644
--- a/drivers/acpi/battery.c
+++ b/drivers/acpi/battery.c
@@ -76,6 +76,7 @@ MODULE_DEVICE_TABLE(acpi, battery_device_ids);
 enum {
 	ACPI_BATTERY_ALARM_PRESENT,
 	ACPI_BATTERY_XINFO_PRESENT,
+	ACPI_BATTERY_BCT_PRESENT,
 	ACPI_BATTERY_QUIRK_PERCENTAGE_CAPACITY,
 	/* On Lenovo Thinkpad models from 2010 and 2011, the power unit
 	 * switches between mWh and mAh depending on whether the system
@@ -110,6 +111,7 @@ struct acpi_battery {
 
 	struct mutex property_lock; /* Protects properties below. */
 	unsigned long update_time;
+	unsigned long long charge_time;
 	int revision;
 	int rate_now;
 	int capacity_now;
@@ -613,6 +615,10 @@ static int acpi_battery_get_state(struct acpi_battery *battery)
 	int result = 0;
 	acpi_status status = 0;
 	struct acpi_buffer buffer = { ACPI_ALLOCATE_BUFFER, NULL };
+	union acpi_object arg0 = { ACPI_TYPE_INTEGER };
+	struct acpi_object_list args = { 1, &arg0 };
+
+	arg0.integer.value = 100;
 
 	lockdep_assert_held(&battery->property_lock);
 
@@ -638,6 +644,15 @@ static int acpi_battery_get_state(struct acpi_battery *battery)
 	battery->update_time = jiffies;
 	kfree(buffer.pointer);
 
+	if (test_bit(ACPI_BATTERY_BCT_PRESENT, &battery->flags)) {
+		status = acpi_evaluate_integer(battery->device->handle, "_BCT",
+					       &args, &battery->charge_time);
+	if (ACPI_FAILURE(status))
+		acpi_handle_info(battery->device->handle,
+				 "_BCT evaluation failed: %s",
+				 acpi_format_exception(status));
+	}
+
 	/* For buggy DSDTs that report negative 16-bit values for either
 	 * charging or discharging current and/or report 0 as 65536
 	 * due to bad math.
@@ -1083,6 +1098,8 @@ static int acpi_battery_update(struct acpi_battery *battery, bool resume)
 			if (result)
 				return result;
 			acpi_battery_init_alarm(battery);
+			if (acpi_has_method(battery->device->handle, "_BCT"))
+				set_bit(ACPI_BATTERY_BCT_PRESENT, &battery->flags);
 		}
 
 		result = acpi_battery_get_state(battery);
-- 
2.55.0

[PATCH v2 2/2] acpi: battery: Handle Valve steamdeck devices quirk

[Umang Jain]

Valve steamdeck devices can report battery status as
charging, even when low-powered power supply is connected
to the device. The battery status is reported as charging
through sysfs, even though in reality, it is discharging.

To handle this case, we introduce a battery quirk for steamdeck
devices ACPI_BATTERY_QUIRK_VALVE_STEAMDECK. Additionally, it has been
found that on such devices, Battery Charge Time ("_BCT") is reported
accurately, hence use that information for report charging vs
discharging status. Battery Charge Time("_BCT") with an eligible power
supply unit will have valid values whereas with a low-powered supply
unit, it shall report ACPI_BATTERY_VALUE_UNKNOWN. Use this as a
differentiator to determine, whether the battery of steamdeck devices
is really charging or not.

Signed-off-by: Umang Jain <uajain@igalia.com>
---
 drivers/acpi/battery.c | 28 +++++++++++++++++++++++++++-
 1 file changed, 27 insertions(+), 1 deletion(-)

diff --git a/drivers/acpi/battery.c b/drivers/acpi/battery.c
index a1b7a821abaf..8defafd2b7c7 100644
--- a/drivers/acpi/battery.c
+++ b/drivers/acpi/battery.c
@@ -95,6 +95,14 @@ enum {
 	 * on a full charge, but showing degradation in full charge cap.
 	 */
 	ACPI_BATTERY_QUIRK_DEGRADED_FULL_CHARGE,
+	/*
+	 * Valve steamdeck devices can get discharged even when battery status
+	 * is charging. This happens typically when low power supply unit is
+	 * connected. To accurately determine this scenario, we shall read
+	 * battery charge time ("_BCT"). This is the key differentiator to
+	 * determine whether the steamdeck is actually charging or not.
+	 */
+	ACPI_BATTERY_QUIRK_VALVE_STEAMDECK,
 };
 
 struct acpi_battery {
@@ -236,7 +244,7 @@ static int acpi_battery_get_property(struct power_supply *psy,
 					: POWER_SUPPLY_STATUS_DISCHARGING;
 				return 0;
 			}
-		else if (battery->state & ACPI_BATTERY_STATE_CHARGING)
+		else if (battery->state & ACPI_BATTERY_STATE_CHARGING) {
 			/* Check the rate and capacity to validate the status. */
 			if (!acpi_battery_is_full(battery) ||
 			    (battery->rate_now != ACPI_BATTERY_VALUE_UNKNOWN &&
@@ -246,6 +254,12 @@ static int acpi_battery_get_property(struct power_supply *psy,
 				/* Full and zero rate. */
 				val->intval = POWER_SUPPLY_STATUS_NOT_CHARGING;
 			}
+
+			if (test_bit(ACPI_BATTERY_QUIRK_VALVE_STEAMDECK, &battery->flags) &&
+			    test_bit(ACPI_BATTERY_BCT_PRESENT, &battery->flags) &&
+			    battery->charge_time == ACPI_BATTERY_VALUE_UNKNOWN)
+				val->intval = POWER_SUPPLY_STATUS_DISCHARGING;
+		}
 		else if (battery->state & ACPI_BATTERY_STATE_CHARGE_LIMITING)
 			val->intval = POWER_SUPPLY_STATUS_NOT_CHARGING;
 		else if (acpi_battery_is_charged(battery))
@@ -1072,6 +1086,18 @@ static void acpi_battery_quirks(struct acpi_battery *battery)
 		set_bit(ACPI_BATTERY_QUIRK_DEGRADED_FULL_CHARGE, &battery->flags);
 		battery->capacity_now = battery->full_charge_capacity;
 	}
+
+	if (test_bit(ACPI_BATTERY_QUIRK_VALVE_STEAMDECK, &battery->flags))
+		return;
+
+	if (dmi_name_in_vendors("Valve")) {
+		const char *name;
+
+		name = dmi_get_system_info(DMI_PRODUCT_NAME);
+		if ((name && (!strncmp("Jupiter", name, 7))) ||
+		    (name && (!strncmp("Galileo", name, 7))))
+			set_bit(ACPI_BATTERY_QUIRK_VALVE_STEAMDECK, &battery->flags);
+	}
 }
 
 static int acpi_battery_update(struct acpi_battery *battery, bool resume)
-- 
2.55.0

Re: [PATCH v2 0/2] acpi: battery: Query BCT method and steamdeck

[Umang Jain]

Hello,

Gentle ping for some review?

Thanks,
Umang

On 2026-09-23 10:29, Umang Jain wrote:
> Short series to query Battery charge time ("_BCT") from ACPI tables
> (1/2).
> 
> Patch 2/2 handles the case where steamdeck devices can report "Charging"
> even when low power PSU is connected which is discharging the device.
> 
> Changes in v2:
> - comment improvements
> - split off BCT method handling into separate patch
> 
> v1:
> https://lore.kernel.org/all/20260918135628.2322188-1-uajain@igalia.com/
> 
> Umang Jain (2):
>   acpi: battery: Query Battery charge time ("_BCT") method
>   acpi: battery: Handle Valve steamdeck devices quirk
> 
>  drivers/acpi/battery.c | 45 +++++++++++++++++++++++++++++++++++++++++-
>  1 file changed, 44 insertions(+), 1 deletion(-)

Re: [PATCH v2 1/2] acpi: battery: Query Battery charge time ("_BCT") method

[Rafael J. Wysocki (Intel)]

On Wed, Sep 23, 2026 at 10:28 AM Umang Jain <uajain@igalia.com> wrote:
>
> Add plumbing for querying the Battery Charge Time ("_BCT") method.
> If present, set ACPI_BATTERY_BCT_PRESENT in battery flags to denote
> that the method is supported in ACPI.
>
> Signed-off-by: Umang Jain <uajain@igalia.com>
> ---
>  drivers/acpi/battery.c | 17 +++++++++++++++++
>  1 file changed, 17 insertions(+)
>
> diff --git a/drivers/acpi/battery.c b/drivers/acpi/battery.c
> index 670853ec3a4d..a1b7a821abaf 100644
> --- a/drivers/acpi/battery.c
> +++ b/drivers/acpi/battery.c
> @@ -76,6 +76,7 @@ MODULE_DEVICE_TABLE(acpi, battery_device_ids);
>  enum {
>         ACPI_BATTERY_ALARM_PRESENT,
>         ACPI_BATTERY_XINFO_PRESENT,
> +       ACPI_BATTERY_BCT_PRESENT,
>         ACPI_BATTERY_QUIRK_PERCENTAGE_CAPACITY,
>         /* On Lenovo Thinkpad models from 2010 and 2011, the power unit
>          * switches between mWh and mAh depending on whether the system
> @@ -110,6 +111,7 @@ struct acpi_battery {
>
>         struct mutex property_lock; /* Protects properties below. */
>         unsigned long update_time;
> +       unsigned long long charge_time;

This doesn't need to be unsigned long long.  As per the spec, unsigned
int should be sufficient.

>         int revision;
>         int rate_now;
>         int capacity_now;
> @@ -613,6 +615,10 @@ static int acpi_battery_get_state(struct acpi_battery *battery)
>         int result = 0;
>         acpi_status status = 0;
>         struct acpi_buffer buffer = { ACPI_ALLOCATE_BUFFER, NULL };
> +       union acpi_object arg0 = { ACPI_TYPE_INTEGER };
> +       struct acpi_object_list args = { 1, &arg0 };
> +
> +       arg0.integer.value = 100;

Please move the definitions of the variables related to _BCT
evaluation to the code block where it gets evaluated.

>
>         lockdep_assert_held(&battery->property_lock);
>
> @@ -638,6 +644,15 @@ static int acpi_battery_get_state(struct acpi_battery *battery)
>         battery->update_time = jiffies;
>         kfree(buffer.pointer);
>
> +       if (test_bit(ACPI_BATTERY_BCT_PRESENT, &battery->flags)) {
> +               status = acpi_evaluate_integer(battery->device->handle, "_BCT",
> +                                              &args, &battery->charge_time);

Currently, you only need to evaluate _BCT to implement the quirk in
the second patch AFAICS, so it is not useful to evaluate _BCT if the
quirk is not going to be used.  Any chance to rearrange the code to
avoid evaluating _BCT when it is not needed?

> +       if (ACPI_FAILURE(status))
> +               acpi_handle_info(battery->device->handle,
> +                                "_BCT evaluation failed: %s",
> +                                acpi_format_exception(status));

Missing indentation.

> +       }
> +
>         /* For buggy DSDTs that report negative 16-bit values for either
>          * charging or discharging current and/or report 0 as 65536
>          * due to bad math.
> @@ -1083,6 +1098,8 @@ static int acpi_battery_update(struct acpi_battery *battery, bool resume)
>                         if (result)
>                                 return result;
>                         acpi_battery_init_alarm(battery);
> +                       if (acpi_has_method(battery->device->handle, "_BCT"))
> +                               set_bit(ACPI_BATTERY_BCT_PRESENT, &battery->flags);
>                 }
>
>                 result = acpi_battery_get_state(battery);
> --

Re: [PATCH v2 2/2] acpi: battery: Handle Valve steamdeck devices quirk

[Rafael J. Wysocki (Intel)]

On Wed, Sep 23, 2026 at 10:28 AM Umang Jain <uajain@igalia.com> wrote:
>
> Valve steamdeck devices can report battery status as
> charging, even when low-powered power supply is connected
> to the device. The battery status is reported as charging
> through sysfs, even though in reality, it is discharging.
>
> To handle this case, we introduce a battery quirk for steamdeck
> devices ACPI_BATTERY_QUIRK_VALVE_STEAMDECK. Additionally, it has been
> found that on such devices, Battery Charge Time ("_BCT") is reported
> accurately, hence use that information for report charging vs
> discharging status. Battery Charge Time("_BCT") with an eligible power
> supply unit will have valid values whereas with a low-powered supply
> unit, it shall report ACPI_BATTERY_VALUE_UNKNOWN. Use this as a
> differentiator to determine, whether the battery of steamdeck devices
> is really charging or not.
>
> Signed-off-by: Umang Jain <uajain@igalia.com>
> ---
>  drivers/acpi/battery.c | 28 +++++++++++++++++++++++++++-
>  1 file changed, 27 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/acpi/battery.c b/drivers/acpi/battery.c
> index a1b7a821abaf..8defafd2b7c7 100644
> --- a/drivers/acpi/battery.c
> +++ b/drivers/acpi/battery.c
> @@ -95,6 +95,14 @@ enum {
>          * on a full charge, but showing degradation in full charge cap.
>          */
>         ACPI_BATTERY_QUIRK_DEGRADED_FULL_CHARGE,
> +       /*
> +        * Valve steamdeck devices can get discharged even when battery status
> +        * is charging. This happens typically when low power supply unit is
> +        * connected. To accurately determine this scenario, we shall read
> +        * battery charge time ("_BCT"). This is the key differentiator to
> +        * determine whether the steamdeck is actually charging or not.
> +        */
> +       ACPI_BATTERY_QUIRK_VALVE_STEAMDECK,
>  };
>
>  struct acpi_battery {
> @@ -236,7 +244,7 @@ static int acpi_battery_get_property(struct power_supply *psy,
>                                         : POWER_SUPPLY_STATUS_DISCHARGING;
>                                 return 0;
>                         }
> -               else if (battery->state & ACPI_BATTERY_STATE_CHARGING)
> +               else if (battery->state & ACPI_BATTERY_STATE_CHARGING) {
>                         /* Check the rate and capacity to validate the status. */
>                         if (!acpi_battery_is_full(battery) ||
>                             (battery->rate_now != ACPI_BATTERY_VALUE_UNKNOWN &&
> @@ -246,6 +254,12 @@ static int acpi_battery_get_property(struct power_supply *psy,
>                                 /* Full and zero rate. */
>                                 val->intval = POWER_SUPPLY_STATUS_NOT_CHARGING;
>                         }
> +
> +                       if (test_bit(ACPI_BATTERY_QUIRK_VALVE_STEAMDECK, &battery->flags) &&
> +                           test_bit(ACPI_BATTERY_BCT_PRESENT, &battery->flags) &&
> +                           battery->charge_time == ACPI_BATTERY_VALUE_UNKNOWN)

Would it be practical to evaluate _BCT here without storing
charge_time in the battery object?

> +                               val->intval = POWER_SUPPLY_STATUS_DISCHARGING;
> +               }
>                 else if (battery->state & ACPI_BATTERY_STATE_CHARGE_LIMITING)
>                         val->intval = POWER_SUPPLY_STATUS_NOT_CHARGING;
>                 else if (acpi_battery_is_charged(battery))
> @@ -1072,6 +1086,18 @@ static void acpi_battery_quirks(struct acpi_battery *battery)
>                 set_bit(ACPI_BATTERY_QUIRK_DEGRADED_FULL_CHARGE, &battery->flags);
>                 battery->capacity_now = battery->full_charge_capacity;
>         }
> +
> +       if (test_bit(ACPI_BATTERY_QUIRK_VALVE_STEAMDECK, &battery->flags))
> +               return;
> +
> +       if (dmi_name_in_vendors("Valve")) {
> +               const char *name;
> +
> +               name = dmi_get_system_info(DMI_PRODUCT_NAME);
> +               if ((name && (!strncmp("Jupiter", name, 7))) ||
> +                   (name && (!strncmp("Galileo", name, 7))))
> +                       set_bit(ACPI_BATTERY_QUIRK_VALVE_STEAMDECK, &battery->flags);
> +       }
>  }
>
>  static int acpi_battery_update(struct acpi_battery *battery, bool resume)
> --

Re: [PATCH v2 1/2] acpi: battery: Query Battery charge time ("_BCT") method

[Umang Jain]

Hi Rafael,

Thanks for the review.

On 2026-10-12 17:15, Rafael J. Wysocki (Intel) wrote:
> On Wed, Sep 23, 2026 at 10:28 AM Umang Jain <uajain@igalia.com> wrote:
>>
>> Add plumbing for querying the Battery Charge Time ("_BCT") method.
>> If present, set ACPI_BATTERY_BCT_PRESENT in battery flags to denote
>> that the method is supported in ACPI.
>>
>> Signed-off-by: Umang Jain <uajain@igalia.com>
>> ---
>>  drivers/acpi/battery.c | 17 +++++++++++++++++
>>  1 file changed, 17 insertions(+)
>>
>> diff --git a/drivers/acpi/battery.c b/drivers/acpi/battery.c
>> index 670853ec3a4d..a1b7a821abaf 100644
>> --- a/drivers/acpi/battery.c
>> +++ b/drivers/acpi/battery.c
>> @@ -76,6 +76,7 @@ MODULE_DEVICE_TABLE(acpi, battery_device_ids);
>>  enum {
>>         ACPI_BATTERY_ALARM_PRESENT,
>>         ACPI_BATTERY_XINFO_PRESENT,
>> +       ACPI_BATTERY_BCT_PRESENT,
>>         ACPI_BATTERY_QUIRK_PERCENTAGE_CAPACITY,
>>         /* On Lenovo Thinkpad models from 2010 and 2011, the power unit
>>          * switches between mWh and mAh depending on whether the system
>> @@ -110,6 +111,7 @@ struct acpi_battery {
>>
>>         struct mutex property_lock; /* Protects properties below. */
>>         unsigned long update_time;
>> +       unsigned long long charge_time;
> 
> This doesn't need to be unsigned long long.  As per the spec, unsigned
> int should be sufficient.
> 
>>         int revision;
>>         int rate_now;
>>         int capacity_now;
>> @@ -613,6 +615,10 @@ static int acpi_battery_get_state(struct acpi_battery *battery)
>>         int result = 0;
>>         acpi_status status = 0;
>>         struct acpi_buffer buffer = { ACPI_ALLOCATE_BUFFER, NULL };
>> +       union acpi_object arg0 = { ACPI_TYPE_INTEGER };
>> +       struct acpi_object_list args = { 1, &arg0 };
>> +
>> +       arg0.integer.value = 100;
> 
> Please move the definitions of the variables related to _BCT
> evaluation to the code block where it gets evaluated.
> 
>>
>>         lockdep_assert_held(&battery->property_lock);
>>
>> @@ -638,6 +644,15 @@ static int acpi_battery_get_state(struct acpi_battery *battery)
>>         battery->update_time = jiffies;
>>         kfree(buffer.pointer);
>>
>> +       if (test_bit(ACPI_BATTERY_BCT_PRESENT, &battery->flags)) {
>> +               status = acpi_evaluate_integer(battery->device->handle, "_BCT",
>> +                                              &args, &battery->charge_time);
> 
> Currently, you only need to evaluate _BCT to implement the quirk in
> the second patch AFAICS, so it is not useful to evaluate _BCT if the
> quirk is not going to be used.  Any chance to rearrange the code to
> avoid evaluating _BCT when it is not needed?

Currently yes, _BCT is not useful in all cases, but since it is a ACPI
specification component, I didn't keep it behind the quirk and read as a
global param to read out.

I've re-arranged the code to evaluate _BCT if the quirk is present, but
want to double sure if that's the design we need to go ahead with?

> 
>> +       if (ACPI_FAILURE(status))
>> +               acpi_handle_info(battery->device->handle,
>> +                                "_BCT evaluation failed: %s",
>> +                                acpi_format_exception(status));
> 
> Missing indentation.

Looks fine to me, what's wrong? The patches sent are sanitized with
./scripts/checkpatch.pl, unless I am missing something?
> 
>> +       }
>> +
>>         /* For buggy DSDTs that report negative 16-bit values for either
>>          * charging or discharging current and/or report 0 as 65536
>>          * due to bad math.
>> @@ -1083,6 +1098,8 @@ static int acpi_battery_update(struct acpi_battery *battery, bool resume)
>>                         if (result)
>>                                 return result;
>>                         acpi_battery_init_alarm(battery);
>> +                       if (acpi_has_method(battery->device->handle, "_BCT"))
>> +                               set_bit(ACPI_BATTERY_BCT_PRESENT, &battery->flags);
>>                 }
>>
>>                 result = acpi_battery_get_state(battery);
>> --