[PATCH v2 0/2] hwmon: (gpd-fan) Fix the second fan on duo boards, split Win 5 fans

[PATCH v2 0/2] hwmon: (gpd-fan) Fix the second fan on duo boards, split Win 5 fans

[Alexei Solcanu via B4 Relay]

Patch 1 fixes pwm1_enable=2 leaving the second fan at its last manual
duty cycle on boards using gpd_duo_drvdata. It is unchanged from v1 and
is still needed on the GPD Duo, and on kernels where the Win 5 uses the
Duo settings.

Patch 2 gives the GPD Win 5 its own board data, with one fan and pwm
channel per fan. The Win 5 had one channel because it was added using
the Duo's settings, which treat both fans as one.

Tested on a GPD Win 5 (G1618-05). Not tested on the GPD Duo.

---
Changes in v2:
- Add patch 2: register both fans on the GPD Win 5 (Guenter)
- Link to v1: https://patch.msgid.link/20260925-gpd-fan-duo-auto-v1-1-29bb670b3331@pm.me

---
Alexei Solcanu (2):
      hwmon: (gpd-fan) Reset both fans to automatic on duo boards
      hwmon: (gpd-fan) Register both fans on the GPD Win 5

 Documentation/hwmon/gpd-fan.rst |  14 ++--
 drivers/hwmon/gpd-fan.c         | 147 +++++++++++++++++++++++++++-------------
 2 files changed, 108 insertions(+), 53 deletions(-)
---
base-commit: 3cbb5f15e464583e59c8501a7085ec12b53e23ac
change-id: 20260925-gpd-fan-duo-auto-871de9e006dc

Best regards,
--  
Alexei Solcanu <alexei.solcanu@pm.me>

[PATCH v2 1/2] hwmon: (gpd-fan) Reset both fans to automatic on duo boards

[Alexei Solcanu via B4 Relay]

From: Alexei Solcanu <alexei.solcanu@pm.me>

On boards using gpd_duo_drvdata (GPD Duo, GPD Win 5), writing
pwm1_enable=2 only returns the first fan to EC control. The second fan
keeps its last manual duty cycle, e.g. full speed after pwm1=255, so a
fan control tool that exits leaves it running at that speed.

gpd_duo_write_pwm() writes both fan registers (pwm_write and
pwm_write + 1), but switching to automatic only clears pwm_write.
Also clear pwm_write + 1, so both fans return to EC control.

Tested on a GPD Win 5 (G1618-05). Not tested on the GPD Duo.

Fixes: 0ab88e239439 ("hwmon: add GPD devices sensor driver")
Assisted-by: LLM
Signed-off-by: Alexei Solcanu <alexei.solcanu@pm.me>
---
 drivers/hwmon/gpd-fan.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/hwmon/gpd-fan.c b/drivers/hwmon/gpd-fan.c
index ed212615b2f..b3b75f99f58 100644
--- a/drivers/hwmon/gpd-fan.c
+++ b/drivers/hwmon/gpd-fan.c
@@ -444,6 +444,7 @@ static void gpd_duo_set_pwm_enable(struct gpd_fan_data *data, enum FAN_PWM_ENABL
 		break;
 	case AUTOMATIC:
 		gpd_ecram_write(data->drvdata, data->drvdata->pwm_write, 0);
+		gpd_ecram_write(data->drvdata, data->drvdata->pwm_write + 1, 0);
 		break;
 	}
 }

-- 
2.55.0

Re: [PATCH v2 1/2] hwmon: (gpd-fan) Reset both fans to automatic on duo boards

[Antheas Kapenekakis]

On Sat, 26 Sept 2026 at 22:12, Alexei Solcanu via B4 Relay
<devnull+alexei.solcanu.pm.me@kernel.org> wrote:
>
> From: Alexei Solcanu <alexei.solcanu@pm.me>
>
> On boards using gpd_duo_drvdata (GPD Duo, GPD Win 5), writing
> pwm1_enable=2 only returns the first fan to EC control. The second fan
> keeps its last manual duty cycle, e.g. full speed after pwm1=255, so a
> fan control tool that exits leaves it running at that speed.
>
> gpd_duo_write_pwm() writes both fan registers (pwm_write and
> pwm_write + 1), but switching to automatic only clears pwm_write.
> Also clear pwm_write + 1, so both fans return to EC control.
>
> Tested on a GPD Win 5 (G1618-05). Not tested on the GPD Duo.

We expect the Win Max 3 to release with the same register layout. Can
you verify it has two fans? If yes this is good to go.

I will try to pull this patch for testing. Second one will take a bit
longer because it needs (my) userspace changes.

Thanks,
Antheas

Reviewed-by: Antheas Kapenekakis <lkml@antheas.dev>

> Fixes: 0ab88e239439 ("hwmon: add GPD devices sensor driver")
> Assisted-by: LLM
> Signed-off-by: Alexei Solcanu <alexei.solcanu@pm.me>
> ---
>  drivers/hwmon/gpd-fan.c | 1 +
>  1 file changed, 1 insertion(+)
>
> diff --git a/drivers/hwmon/gpd-fan.c b/drivers/hwmon/gpd-fan.c
> index ed212615b2f..b3b75f99f58 100644
> --- a/drivers/hwmon/gpd-fan.c
> +++ b/drivers/hwmon/gpd-fan.c
> @@ -444,6 +444,7 @@ static void gpd_duo_set_pwm_enable(struct gpd_fan_data *data, enum FAN_PWM_ENABL
>                 break;
>         case AUTOMATIC:
>                 gpd_ecram_write(data->drvdata, data->drvdata->pwm_write, 0);
> +               gpd_ecram_write(data->drvdata, data->drvdata->pwm_write + 1, 0);
>                 break;
>         }
>  }
>
> --
> 2.55.0
>
>
>

[PATCH v2 2/2] hwmon: (gpd-fan) Register both fans on the GPD Win 5

[Alexei Solcanu via B4 Relay]

From: Alexei Solcanu <alexei.solcanu@pm.me>

The GPD Win 5 uses gpd_duo_drvdata, which exposes a single fan and pwm
channel: pwm1 writes the same duty cycle to both fans, and fan1_input
only reads the first fan's RPM.

The two fans have independent registers. Fan 1 uses duty cycle 0x047A
and RPM 0x0478/0x0479, fan 2 uses duty cycle 0x047B and RPM
0x0476/0x0477. A duty cycle of 0 returns that fan to EC control.

Add separate board data for the Win 5 and register fan1/pwm1 and
fan2/pwm2, one channel per fan. Boards with a single fan are unchanged.

This changes behaviour on the Win 5: pwm1 and pwm1_enable now control
only the first fan. The second fan stays under EC control unless pwm2
and pwm2_enable are used.

The GPD Duo keeps a single channel, since it is untested and its
second fan's RPM register is unknown.

Tested on a GPD Win 5 (G1618-05).

Suggested-by: Guenter Roeck <linux@roeck-us.net>
Assisted-by: LLM
Signed-off-by: Alexei Solcanu <alexei.solcanu@pm.me>
---
 Documentation/hwmon/gpd-fan.rst |  14 ++--
 drivers/hwmon/gpd-fan.c         | 146 +++++++++++++++++++++++++++-------------
 2 files changed, 107 insertions(+), 53 deletions(-)

diff --git a/Documentation/hwmon/gpd-fan.rst b/Documentation/hwmon/gpd-fan.rst
index b27657d3305..6c8569c5f64 100644
--- a/Documentation/hwmon/gpd-fan.rst
+++ b/Documentation/hwmon/gpd-fan.rst
@@ -28,6 +28,7 @@ Currently the driver supports the following handhelds:
  - GPD Win Max 2 2025 (HX370)
  - GPD Win 4 (6800U)
  - GPD Win 4 (7840U)
+ - GPD Win 5
  - GPD Micro PC 2
 
 Module parameters
@@ -59,23 +60,26 @@ Sysfs entries
 
 The following attributes are supported:
 
-fan1_input
+fan[1-2]_input
   Read Only. Reads current fan RPM.
 
-pwm1_enable
+pwm[1-2]_enable
   Read/Write. Enable manual fan control. Write "0" to disable control and run
   at full speed. Write "1" to set to manual, write "2" to let the EC control
   decide fan speed. Read this attribute to see current status.
 
   NB: In consideration of the safety of the device, when setting to manual mode,
   the pwm speed will be set to the maximum value (255) by default. You can set
-  a different value by writing pwm1 later.
+  a different value by writing pwm[1-2] later.
 
-pwm1
+pwm[1-2]
   Read/Write. Read this attribute to see current duty cycle in the range
-  [0-255]. When pwm1_enable is set to "1" (manual) write any value in the
+  [0-255]. When pwm[1-2]_enable is set to "1" (manual) write any value in the
   range [0-255] to set fan speed.
 
   NB: Many boards (except listed under wm2 above) don't support reading the
   current pwm value in auto mode. That will just return EOPNOTSUPP. In manual
   mode it will always return the real value.
+
+fan2_input, pwm2_enable and pwm2 only exist on devices with two fans
+(GPD Win 5).
diff --git a/drivers/hwmon/gpd-fan.c b/drivers/hwmon/gpd-fan.c
index b3b75f99f58..e03b0dde66f 100644
--- a/drivers/hwmon/gpd-fan.c
+++ b/drivers/hwmon/gpd-fan.c
@@ -31,6 +31,7 @@ enum gpd_board {
 	win4_6800u,
 	win_max_2,
 	duo,
+	win5,
 	mpc2,
 };
 
@@ -40,9 +41,11 @@ enum FAN_PWM_ENABLE {
 	AUTOMATIC	= 2,
 };
 
+#define GPD_MAX_FANS	2
+
 struct gpd_fan_data {
-	enum FAN_PWM_ENABLE pwm_enable;
-	u8 pwm_value;
+	enum FAN_PWM_ENABLE pwm_enable[GPD_MAX_FANS];
+	u8 pwm_value[GPD_MAX_FANS];
 	const struct gpd_fan_drvdata *drvdata;
 };
 
@@ -55,6 +58,8 @@ struct gpd_fan_drvdata {
 	const u16 manual_control_enable;
 	const u16 rpm_read;
 	const u16 pwm_write;
+	const u16 rpm2_read; // Second fan, 0 if there is none
+	const u16 pwm2_write;
 	const u16 pwm_max;
 };
 
@@ -82,6 +87,20 @@ static struct gpd_fan_drvdata gpd_duo_drvdata = {
 	.pwm_max		= 244,
 };
 
+static struct gpd_fan_drvdata gpd_win5_drvdata = {
+	.board_name		= "win5",
+	.board			= win5,
+
+	.addr_port		= 0x4E,
+	.data_port		= 0x4F,
+	.manual_control_enable	= 0x047A,
+	.rpm_read		= 0x0478,
+	.pwm_write		= 0x047A,
+	.rpm2_read		= 0x0476,
+	.pwm2_write		= 0x047B,
+	.pwm_max		= 244,
+};
+
 static struct gpd_fan_drvdata gpd_win4_drvdata = {
 	.board_name		= "win4",
 	.board			= win4_6800u,
@@ -214,7 +233,7 @@ static const struct dmi_system_id dmi_table[] = {
 			DMI_MATCH(DMI_SYS_VENDOR, "GPD"),
 			DMI_MATCH(DMI_PRODUCT_NAME, "G1618-05"),
 		},
-		.driver_data = &gpd_duo_drvdata,
+		.driver_data = &gpd_win5_drvdata,
 	},
 	{
 		// GPD Pocket 4
@@ -290,13 +309,29 @@ static void gpd_ecram_write(const struct gpd_fan_drvdata *drvdata, u16 offset, u
 	outb(value, data_port);
 }
 
-static int gpd_generic_read_rpm(struct gpd_fan_data *data)
+static int gpd_num_fans(const struct gpd_fan_drvdata *drvdata)
+{
+	return drvdata->pwm2_write ? 2 : 1;
+}
+
+static u16 gpd_rpm_reg(const struct gpd_fan_drvdata *drvdata, int channel)
+{
+	return channel ? drvdata->rpm2_read : drvdata->rpm_read;
+}
+
+static u16 gpd_pwm_reg(const struct gpd_fan_drvdata *drvdata, int channel)
+{
+	return channel ? drvdata->pwm2_write : drvdata->pwm_write;
+}
+
+static int gpd_generic_read_rpm(struct gpd_fan_data *data, int channel)
 {
 	const struct gpd_fan_drvdata *drvdata = data->drvdata;
+	u16 reg = gpd_rpm_reg(drvdata, channel);
 	u8 high, low;
 
-	gpd_ecram_read(drvdata, drvdata->rpm_read, &high);
-	gpd_ecram_read(drvdata, drvdata->rpm_read + 1, &low);
+	gpd_ecram_read(drvdata, reg, &high);
+	gpd_ecram_read(drvdata, reg + 1, &low);
 
 	return (u16)high << 8 | low;
 }
@@ -315,18 +350,19 @@ static int gpd_wm2_read_rpm(struct gpd_fan_data *data)
 			gpd_ecram_write(drvdata, pwm_ctr_offset, 0xB8);
 	}
 
-	return gpd_generic_read_rpm(data);
+	return gpd_generic_read_rpm(data, 0);
 }
 
-// Read value for fan1_input
-static int gpd_read_rpm(struct gpd_fan_data *data)
+// Read value for fanN_input
+static int gpd_read_rpm(struct gpd_fan_data *data, int channel)
 {
 	switch (data->drvdata->board) {
 	case win4_6800u:
 	case win_mini:
 	case duo:
+	case win5:
 	case mpc2:
-		return gpd_generic_read_rpm(data);
+		return gpd_generic_read_rpm(data, channel);
 	case win_max_2:
 		return gpd_wm2_read_rpm(data);
 	}
@@ -349,19 +385,20 @@ static int gpd_wm2_read_pwm(struct gpd_fan_data *data)
 	return DIV_ROUND_CLOSEST((var - 1) * 255, (drvdata->pwm_max - 1));
 }
 
-// Read value for pwm1
-static int gpd_read_pwm(struct gpd_fan_data *data)
+// Read value for pwmN
+static int gpd_read_pwm(struct gpd_fan_data *data, int channel)
 {
 	switch (data->drvdata->board) {
 	case win_mini:
 	case duo:
+	case win5:
 	case win4_6800u:
 	case mpc2:
-		switch (data->pwm_enable) {
+		switch (data->pwm_enable[channel]) {
 		case DISABLE:
 			return 255;
 		case MANUAL:
-			return data->pwm_value;
+			return data->pwm_value[channel];
 		case AUTOMATIC:
 			return -EOPNOTSUPP;
 		}
@@ -378,13 +415,13 @@ static inline u8 gpd_cast_pwm_range(const struct gpd_fan_drvdata *drvdata, u8 va
 	return DIV_ROUND_CLOSEST(val * (drvdata->pwm_max - 1), 255) + 1;
 }
 
-static void gpd_generic_write_pwm(struct gpd_fan_data *data, u8 val)
+static void gpd_generic_write_pwm(struct gpd_fan_data *data, int channel, u8 val)
 {
 	const struct gpd_fan_drvdata *drvdata = data->drvdata;
 	u8 pwm_reg;
 
 	pwm_reg = gpd_cast_pwm_range(drvdata, val);
-	gpd_ecram_write(drvdata, drvdata->pwm_write, pwm_reg);
+	gpd_ecram_write(drvdata, gpd_pwm_reg(drvdata, channel), pwm_reg);
 }
 
 static void gpd_duo_write_pwm(struct gpd_fan_data *data, u8 val)
@@ -397,10 +434,10 @@ static void gpd_duo_write_pwm(struct gpd_fan_data *data, u8 val)
 	gpd_ecram_write(drvdata, drvdata->pwm_write + 1, pwm_reg);
 }
 
-// Write value for pwm1
-static int gpd_write_pwm(struct gpd_fan_data *data, u8 val)
+// Write value for pwmN
+static int gpd_write_pwm(struct gpd_fan_data *data, int channel, u8 val)
 {
-	if (data->pwm_enable != MANUAL)
+	if (data->pwm_enable[channel] != MANUAL)
 		return -EPERM;
 
 	switch (data->drvdata->board) {
@@ -410,25 +447,27 @@ static int gpd_write_pwm(struct gpd_fan_data *data, u8 val)
 	case win_mini:
 	case win4_6800u:
 	case win_max_2:
+	case win5:
 	case mpc2:
-		gpd_generic_write_pwm(data, val);
+		gpd_generic_write_pwm(data, channel, val);
 		break;
 	}
 
 	return 0;
 }
 
-static void gpd_win_mini_set_pwm_enable(struct gpd_fan_data *data, enum FAN_PWM_ENABLE pwm_enable)
+static void gpd_win_mini_set_pwm_enable(struct gpd_fan_data *data, int channel,
+					enum FAN_PWM_ENABLE pwm_enable)
 {
 	switch (pwm_enable) {
 	case DISABLE:
-		gpd_generic_write_pwm(data, 255);
+		gpd_generic_write_pwm(data, channel, 255);
 		break;
 	case MANUAL:
-		gpd_generic_write_pwm(data, data->pwm_value);
+		gpd_generic_write_pwm(data, channel, data->pwm_value[channel]);
 		break;
 	case AUTOMATIC:
-		gpd_ecram_write(data->drvdata, data->drvdata->pwm_write, 0);
+		gpd_ecram_write(data->drvdata, gpd_pwm_reg(data->drvdata, channel), 0);
 		break;
 	}
 }
@@ -440,7 +479,7 @@ static void gpd_duo_set_pwm_enable(struct gpd_fan_data *data, enum FAN_PWM_ENABL
 		gpd_duo_write_pwm(data, 255);
 		break;
 	case MANUAL:
-		gpd_duo_write_pwm(data, data->pwm_value);
+		gpd_duo_write_pwm(data, data->pwm_value[0]);
 		break;
 	case AUTOMATIC:
 		gpd_ecram_write(data->drvdata, data->drvdata->pwm_write, 0);
@@ -455,11 +494,11 @@ static void gpd_wm2_set_pwm_enable(struct gpd_fan_data *data, enum FAN_PWM_ENABL
 
 	switch (enable) {
 	case DISABLE:
-		gpd_generic_write_pwm(data, 255);
+		gpd_generic_write_pwm(data, 0, 255);
 		gpd_ecram_write(drvdata, drvdata->manual_control_enable, 1);
 		break;
 	case MANUAL:
-		gpd_generic_write_pwm(data, data->pwm_value);
+		gpd_generic_write_pwm(data, 0, data->pwm_value[0]);
 		gpd_ecram_write(drvdata, drvdata->manual_control_enable, 1);
 		break;
 	case AUTOMATIC:
@@ -468,19 +507,20 @@ static void gpd_wm2_set_pwm_enable(struct gpd_fan_data *data, enum FAN_PWM_ENABL
 	}
 }
 
-// Write value for pwm1_enable
-static void gpd_set_pwm_enable(struct gpd_fan_data *data, enum FAN_PWM_ENABLE enable)
+// Write value for pwmN_enable
+static void gpd_set_pwm_enable(struct gpd_fan_data *data, int channel, enum FAN_PWM_ENABLE enable)
 {
 	if (enable == MANUAL)
 		// Set pwm_value to max firstly when switching to manual mode, in
 		// consideration of device safety.
-		data->pwm_value = 255;
+		data->pwm_value[channel] = 255;
 
 	switch (data->drvdata->board) {
 	case win_mini:
 	case win4_6800u:
+	case win5:
 	case mpc2:
-		gpd_win_mini_set_pwm_enable(data, enable);
+		gpd_win_mini_set_pwm_enable(data, channel, enable);
 		break;
 	case duo:
 		gpd_duo_set_pwm_enable(data, enable);
@@ -491,10 +531,15 @@ static void gpd_set_pwm_enable(struct gpd_fan_data *data, enum FAN_PWM_ENABLE en
 	}
 }
 
-static umode_t gpd_fan_hwmon_is_visible(__always_unused const void *drvdata,
+static umode_t gpd_fan_hwmon_is_visible(const void *hwmon_data,
 					enum hwmon_sensor_types type, u32 attr,
-					__always_unused int channel)
+					int channel)
 {
+	const struct gpd_fan_data *data = hwmon_data;
+
+	if (channel >= gpd_num_fans(data->drvdata))
+		return 0;
+
 	if (type == hwmon_fan && attr == hwmon_fan_input) {
 		return 0444;
 	} else if (type == hwmon_pwm) {
@@ -511,14 +556,14 @@ static umode_t gpd_fan_hwmon_is_visible(__always_unused const void *drvdata,
 
 static int gpd_fan_hwmon_read(struct device *dev,
 			      enum hwmon_sensor_types type, u32 attr,
-			      __always_unused int channel, long *val)
+			      int channel, long *val)
 {
 	struct gpd_fan_data *data = dev_get_drvdata(dev);
 	int ret;
 
 	if (type == hwmon_fan) {
 		if (attr == hwmon_fan_input) {
-			ret = gpd_read_rpm(data);
+			ret = gpd_read_rpm(data, channel);
 
 			if (ret < 0)
 				return ret;
@@ -529,10 +574,10 @@ static int gpd_fan_hwmon_read(struct device *dev,
 	} else if (type == hwmon_pwm) {
 		switch (attr) {
 		case hwmon_pwm_enable:
-			*val = data->pwm_enable;
+			*val = data->pwm_enable[channel];
 			return 0;
 		case hwmon_pwm_input:
-			ret = gpd_read_pwm(data);
+			ret = gpd_read_pwm(data, channel);
 
 			if (ret < 0)
 				return ret;
@@ -547,7 +592,7 @@ static int gpd_fan_hwmon_read(struct device *dev,
 
 static int gpd_fan_hwmon_write(struct device *dev,
 			       enum hwmon_sensor_types type, u32 attr,
-			       __always_unused int channel, long val)
+			       int channel, long val)
 {
 	struct gpd_fan_data *data = dev_get_drvdata(dev);
 
@@ -557,17 +602,17 @@ static int gpd_fan_hwmon_write(struct device *dev,
 			if (!in_range(val, 0, 3))
 				return -EINVAL;
 
-			data->pwm_enable = val;
+			data->pwm_enable[channel] = val;
 
-			gpd_set_pwm_enable(data, data->pwm_enable);
+			gpd_set_pwm_enable(data, channel, data->pwm_enable[channel]);
 			return 0;
 		case hwmon_pwm_input:
 			if (!in_range(val, 0, 256))
 				return -EINVAL;
 
-			data->pwm_value = val;
+			data->pwm_value[channel] = val;
 
-			return gpd_write_pwm(data, val);
+			return gpd_write_pwm(data, channel, val);
 		}
 	}
 
@@ -581,8 +626,9 @@ static const struct hwmon_ops gpd_fan_ops = {
 };
 
 static const struct hwmon_channel_info *gpd_fan_hwmon_channel_info[] = {
-	HWMON_CHANNEL_INFO(fan, HWMON_F_INPUT),
-	HWMON_CHANNEL_INFO(pwm, HWMON_PWM_INPUT | HWMON_PWM_ENABLE),
+	HWMON_CHANNEL_INFO(fan, HWMON_F_INPUT, HWMON_F_INPUT),
+	HWMON_CHANNEL_INFO(pwm, HWMON_PWM_INPUT | HWMON_PWM_ENABLE,
+			   HWMON_PWM_INPUT | HWMON_PWM_ENABLE),
 	NULL
 };
 
@@ -619,8 +665,10 @@ static void gpd_fan_reset_hardware(void *pdata)
 	struct gpd_fan_data *data = pdata;
 
 	if (data) {
-		data->pwm_enable = AUTOMATIC;
-		gpd_set_pwm_enable(data, AUTOMATIC);
+		for (int i = 0; i < gpd_num_fans(data->drvdata); i++) {
+			data->pwm_enable[i] = AUTOMATIC;
+			gpd_set_pwm_enable(data, i, AUTOMATIC);
+		}
 	}
 }
 
@@ -654,8 +702,10 @@ static int gpd_fan_probe(struct platform_device *pdev)
 		return -EINVAL;
 
 	data->drvdata = match;
-	data->pwm_enable = AUTOMATIC;
-	data->pwm_value = 255;
+	for (int i = 0; i < gpd_num_fans(match); i++) {
+		data->pwm_enable[i] = AUTOMATIC;
+		data->pwm_value[i] = 255;
+	}
 
 	dev_set_drvdata(dev, data);
 

-- 
2.55.0

Re: [PATCH v2 2/2] hwmon: (gpd-fan) Register both fans on the GPD Win 5

[Antheas Kapenekakis]

On Sat, 26 Sept 2026 at 22:12, Alexei Solcanu via B4 Relay
<devnull+alexei.solcanu.pm.me@kernel.org> wrote:
>
> From: Alexei Solcanu <alexei.solcanu@pm.me>
>
> The GPD Win 5 uses gpd_duo_drvdata, which exposes a single fan and pwm
> channel: pwm1 writes the same duty cycle to both fans, and fan1_input
> only reads the first fan's RPM.
>
> The two fans have independent registers. Fan 1 uses duty cycle 0x047A
> and RPM 0x0478/0x0479, fan 2 uses duty cycle 0x047B and RPM
> 0x0476/0x0477. A duty cycle of 0 returns that fan to EC control.
>
> Add separate board data for the Win 5 and register fan1/pwm1 and
> fan2/pwm2, one channel per fan. Boards with a single fan are unchanged.
>
> This changes behaviour on the Win 5: pwm1 and pwm1_enable now control
> only the first fan. The second fan stays under EC control unless pwm2
> and pwm2_enable are used.
>
> The GPD Duo keeps a single channel, since it is untested and its
> second fan's RPM register is unknown.
>
> Tested on a GPD Win 5 (G1618-05).
>
> Suggested-by: Guenter Roeck <linux@roeck-us.net>
> Assisted-by: LLM
> Signed-off-by: Alexei Solcanu <alexei.solcanu@pm.me>
> ---
>  Documentation/hwmon/gpd-fan.rst |  14 ++--
>  drivers/hwmon/gpd-fan.c         | 146 +++++++++++++++++++++++++++-------------
>  2 files changed, 107 insertions(+), 53 deletions(-)
>
> diff --git a/Documentation/hwmon/gpd-fan.rst b/Documentation/hwmon/gpd-fan.rst
> index b27657d3305..6c8569c5f64 100644
> --- a/Documentation/hwmon/gpd-fan.rst
> +++ b/Documentation/hwmon/gpd-fan.rst
> @@ -28,6 +28,7 @@ Currently the driver supports the following handhelds:
>   - GPD Win Max 2 2025 (HX370)
>   - GPD Win 4 (6800U)
>   - GPD Win 4 (7840U)
> + - GPD Win 5
>   - GPD Micro PC 2
>
>  Module parameters
> @@ -59,23 +60,26 @@ Sysfs entries
>
>  The following attributes are supported:
>
> -fan1_input
> +fan[1-2]_input
>    Read Only. Reads current fan RPM.
>
> -pwm1_enable
> +pwm[1-2]_enable
>    Read/Write. Enable manual fan control. Write "0" to disable control and run
>    at full speed. Write "1" to set to manual, write "2" to let the EC control
>    decide fan speed. Read this attribute to see current status.
>
>    NB: In consideration of the safety of the device, when setting to manual mode,
>    the pwm speed will be set to the maximum value (255) by default. You can set
> -  a different value by writing pwm1 later.
> +  a different value by writing pwm[1-2] later.
>
> -pwm1
> +pwm[1-2]
>    Read/Write. Read this attribute to see current duty cycle in the range
> -  [0-255]. When pwm1_enable is set to "1" (manual) write any value in the
> +  [0-255]. When pwm[1-2]_enable is set to "1" (manual) write any value in the
>    range [0-255] to set fan speed.
>
>    NB: Many boards (except listed under wm2 above) don't support reading the
>    current pwm value in auto mode. That will just return EOPNOTSUPP. In manual
>    mode it will always return the real value.
> +
> +fan2_input, pwm2_enable and pwm2 only exist on devices with two fans
> +(GPD Win 5).
> diff --git a/drivers/hwmon/gpd-fan.c b/drivers/hwmon/gpd-fan.c
> index b3b75f99f58..e03b0dde66f 100644
> --- a/drivers/hwmon/gpd-fan.c
> +++ b/drivers/hwmon/gpd-fan.c
> @@ -31,6 +31,7 @@ enum gpd_board {
>         win4_6800u,
>         win_max_2,
>         duo,
> +       win5,
>         mpc2,
>  };
>
> @@ -40,9 +41,11 @@ enum FAN_PWM_ENABLE {
>         AUTOMATIC       = 2,
>  };
>
> +#define GPD_MAX_FANS   2
> +
>  struct gpd_fan_data {
> -       enum FAN_PWM_ENABLE pwm_enable;
> -       u8 pwm_value;
> +       enum FAN_PWM_ENABLE pwm_enable[GPD_MAX_FANS];
> +       u8 pwm_value[GPD_MAX_FANS];
>         const struct gpd_fan_drvdata *drvdata;
>  };
>
> @@ -55,6 +58,8 @@ struct gpd_fan_drvdata {
>         const u16 manual_control_enable;
>         const u16 rpm_read;
>         const u16 pwm_write;
> +       const u16 rpm2_read; // Second fan, 0 if there is none
> +       const u16 pwm2_write;
>         const u16 pwm_max;
>  };
>
> @@ -82,6 +87,20 @@ static struct gpd_fan_drvdata gpd_duo_drvdata = {
>         .pwm_max                = 244,
>  };
>
> +static struct gpd_fan_drvdata gpd_win5_drvdata = {
> +       .board_name             = "win5",
> +       .board                  = win5,
> +
> +       .addr_port              = 0x4E,
> +       .data_port              = 0x4F,
> +       .manual_control_enable  = 0x047A,
> +       .rpm_read               = 0x0478,
> +       .pwm_write              = 0x047A,
> +       .rpm2_read              = 0x0476,
> +       .pwm2_write             = 0x047B,
> +       .pwm_max                = 244,
> +};
> +
>  static struct gpd_fan_drvdata gpd_win4_drvdata = {
>         .board_name             = "win4",
>         .board                  = win4_6800u,
> @@ -214,7 +233,7 @@ static const struct dmi_system_id dmi_table[] = {
>                         DMI_MATCH(DMI_SYS_VENDOR, "GPD"),
>                         DMI_MATCH(DMI_PRODUCT_NAME, "G1618-05"),
>                 },
> -               .driver_data = &gpd_duo_drvdata,
> +               .driver_data = &gpd_win5_drvdata,
>         },
>         {
>                 // GPD Pocket 4
> @@ -290,13 +309,29 @@ static void gpd_ecram_write(const struct gpd_fan_drvdata *drvdata, u16 offset, u
>         outb(value, data_port);
>  }
>
> -static int gpd_generic_read_rpm(struct gpd_fan_data *data)
> +static int gpd_num_fans(const struct gpd_fan_drvdata *drvdata)
> +{
> +       return drvdata->pwm2_write ? 2 : 1;
> +}
> +
> +static u16 gpd_rpm_reg(const struct gpd_fan_drvdata *drvdata, int channel)
> +{
> +       return channel ? drvdata->rpm2_read : drvdata->rpm_read;
> +}
> +
> +static u16 gpd_pwm_reg(const struct gpd_fan_drvdata *drvdata, int channel)
> +{
> +       return channel ? drvdata->pwm2_write : drvdata->pwm_write;
> +}

in my view adding all these functions is unneeded complexity. There
should be some simpler ternary way.

> +
> +static int gpd_generic_read_rpm(struct gpd_fan_data *data, int channel)
>  {
>         const struct gpd_fan_drvdata *drvdata = data->drvdata;
> +       u16 reg = gpd_rpm_reg(drvdata, channel);
>         u8 high, low;
>
> -       gpd_ecram_read(drvdata, drvdata->rpm_read, &high);
> -       gpd_ecram_read(drvdata, drvdata->rpm_read + 1, &low);
> +       gpd_ecram_read(drvdata, reg, &high);
> +       gpd_ecram_read(drvdata, reg + 1, &low);
>
>         return (u16)high << 8 | low;
>  }
> @@ -315,18 +350,19 @@ static int gpd_wm2_read_rpm(struct gpd_fan_data *data)
>                         gpd_ecram_write(drvdata, pwm_ctr_offset, 0xB8);
>         }
>
> -       return gpd_generic_read_rpm(data);
> +       return gpd_generic_read_rpm(data, 0);
>  }
>
> -// Read value for fan1_input
> -static int gpd_read_rpm(struct gpd_fan_data *data)
> +// Read value for fanN_input
> +static int gpd_read_rpm(struct gpd_fan_data *data, int channel)
>  {
>         switch (data->drvdata->board) {
>         case win4_6800u:
>         case win_mini:
>         case duo:
> +       case win5:
>         case mpc2:
> -               return gpd_generic_read_rpm(data);
> +               return gpd_generic_read_rpm(data, channel);
>         case win_max_2:
>                 return gpd_wm2_read_rpm(data);
>         }
> @@ -349,19 +385,20 @@ static int gpd_wm2_read_pwm(struct gpd_fan_data *data)
>         return DIV_ROUND_CLOSEST((var - 1) * 255, (drvdata->pwm_max - 1));
>  }
>
> -// Read value for pwm1
> -static int gpd_read_pwm(struct gpd_fan_data *data)
> +// Read value for pwmN
> +static int gpd_read_pwm(struct gpd_fan_data *data, int channel)
>  {
>         switch (data->drvdata->board) {
>         case win_mini:
>         case duo:
> +       case win5:
>         case win4_6800u:
>         case mpc2:
> -               switch (data->pwm_enable) {
> +               switch (data->pwm_enable[channel]) {
>                 case DISABLE:
>                         return 255;
>                 case MANUAL:
> -                       return data->pwm_value;
> +                       return data->pwm_value[channel];
>                 case AUTOMATIC:
>                         return -EOPNOTSUPP;
>                 }
> @@ -378,13 +415,13 @@ static inline u8 gpd_cast_pwm_range(const struct gpd_fan_drvdata *drvdata, u8 va
>         return DIV_ROUND_CLOSEST(val * (drvdata->pwm_max - 1), 255) + 1;
>  }
>
> -static void gpd_generic_write_pwm(struct gpd_fan_data *data, u8 val)
> +static void gpd_generic_write_pwm(struct gpd_fan_data *data, int channel, u8 val)
>  {
>         const struct gpd_fan_drvdata *drvdata = data->drvdata;
>         u8 pwm_reg;
>
>         pwm_reg = gpd_cast_pwm_range(drvdata, val);
> -       gpd_ecram_write(drvdata, drvdata->pwm_write, pwm_reg);
> +       gpd_ecram_write(drvdata, gpd_pwm_reg(drvdata, channel), pwm_reg);
>  }
>
>  static void gpd_duo_write_pwm(struct gpd_fan_data *data, u8 val)
> @@ -397,10 +434,10 @@ static void gpd_duo_write_pwm(struct gpd_fan_data *data, u8 val)
>         gpd_ecram_write(drvdata, drvdata->pwm_write + 1, pwm_reg);
>  }
>
> -// Write value for pwm1
> -static int gpd_write_pwm(struct gpd_fan_data *data, u8 val)
> +// Write value for pwmN
> +static int gpd_write_pwm(struct gpd_fan_data *data, int channel, u8 val)
>  {
> -       if (data->pwm_enable != MANUAL)
> +       if (data->pwm_enable[channel] != MANUAL)
>                 return -EPERM;
>
>         switch (data->drvdata->board) {
> @@ -410,25 +447,27 @@ static int gpd_write_pwm(struct gpd_fan_data *data, u8 val)
>         case win_mini:
>         case win4_6800u:
>         case win_max_2:
> +       case win5:
>         case mpc2:
> -               gpd_generic_write_pwm(data, val);
> +               gpd_generic_write_pwm(data, channel, val);
>                 break;
>         }
>
>         return 0;
>  }
>
> -static void gpd_win_mini_set_pwm_enable(struct gpd_fan_data *data, enum FAN_PWM_ENABLE pwm_enable)
> +static void gpd_win_mini_set_pwm_enable(struct gpd_fan_data *data, int channel,
> +                                       enum FAN_PWM_ENABLE pwm_enable)
>  {
>         switch (pwm_enable) {
>         case DISABLE:
> -               gpd_generic_write_pwm(data, 255);
> +               gpd_generic_write_pwm(data, channel, 255);
>                 break;
>         case MANUAL:
> -               gpd_generic_write_pwm(data, data->pwm_value);
> +               gpd_generic_write_pwm(data, channel, data->pwm_value[channel]);
>                 break;
>         case AUTOMATIC:
> -               gpd_ecram_write(data->drvdata, data->drvdata->pwm_write, 0);
> +               gpd_ecram_write(data->drvdata, gpd_pwm_reg(data->drvdata, channel), 0);
>                 break;
>         }
>  }
> @@ -440,7 +479,7 @@ static void gpd_duo_set_pwm_enable(struct gpd_fan_data *data, enum FAN_PWM_ENABL
>                 gpd_duo_write_pwm(data, 255);
>                 break;
>         case MANUAL:
> -               gpd_duo_write_pwm(data, data->pwm_value);
> +               gpd_duo_write_pwm(data, data->pwm_value[0]);
>                 break;
>         case AUTOMATIC:
>                 gpd_ecram_write(data->drvdata, data->drvdata->pwm_write, 0);
> @@ -455,11 +494,11 @@ static void gpd_wm2_set_pwm_enable(struct gpd_fan_data *data, enum FAN_PWM_ENABL
>
>         switch (enable) {
>         case DISABLE:
> -               gpd_generic_write_pwm(data, 255);
> +               gpd_generic_write_pwm(data, 0, 255);
>                 gpd_ecram_write(drvdata, drvdata->manual_control_enable, 1);
>                 break;
>         case MANUAL:
> -               gpd_generic_write_pwm(data, data->pwm_value);
> +               gpd_generic_write_pwm(data, 0, data->pwm_value[0]);
>                 gpd_ecram_write(drvdata, drvdata->manual_control_enable, 1);
>                 break;
>         case AUTOMATIC:
> @@ -468,19 +507,20 @@ static void gpd_wm2_set_pwm_enable(struct gpd_fan_data *data, enum FAN_PWM_ENABL
>         }
>  }
>
> -// Write value for pwm1_enable
> -static void gpd_set_pwm_enable(struct gpd_fan_data *data, enum FAN_PWM_ENABLE enable)
> +// Write value for pwmN_enable
> +static void gpd_set_pwm_enable(struct gpd_fan_data *data, int channel, enum FAN_PWM_ENABLE enable)
>  {
>         if (enable == MANUAL)
>                 // Set pwm_value to max firstly when switching to manual mode, in
>                 // consideration of device safety.
> -               data->pwm_value = 255;
> +               data->pwm_value[channel] = 255;
>
>         switch (data->drvdata->board) {
>         case win_mini:
>         case win4_6800u:
> +       case win5:
>         case mpc2:
> -               gpd_win_mini_set_pwm_enable(data, enable);
> +               gpd_win_mini_set_pwm_enable(data, channel, enable);
>                 break;
>         case duo:
>                 gpd_duo_set_pwm_enable(data, enable);
> @@ -491,10 +531,15 @@ static void gpd_set_pwm_enable(struct gpd_fan_data *data, enum FAN_PWM_ENABLE en
>         }
>  }
>
> -static umode_t gpd_fan_hwmon_is_visible(__always_unused const void *drvdata,
> +static umode_t gpd_fan_hwmon_is_visible(const void *hwmon_data,
>                                         enum hwmon_sensor_types type, u32 attr,
> -                                       __always_unused int channel)
> +                                       int channel)
>  {
> +       const struct gpd_fan_data *data = hwmon_data;
> +
> +       if (channel >= gpd_num_fans(data->drvdata))
> +               return 0;
> +
>         if (type == hwmon_fan && attr == hwmon_fan_input) {
>                 return 0444;
>         } else if (type == hwmon_pwm) {
> @@ -511,14 +556,14 @@ static umode_t gpd_fan_hwmon_is_visible(__always_unused const void *drvdata,
>
>  static int gpd_fan_hwmon_read(struct device *dev,
>                               enum hwmon_sensor_types type, u32 attr,
> -                             __always_unused int channel, long *val)
> +                             int channel, long *val)
>  {
>         struct gpd_fan_data *data = dev_get_drvdata(dev);
>         int ret;
>
>         if (type == hwmon_fan) {
>                 if (attr == hwmon_fan_input) {
> -                       ret = gpd_read_rpm(data);
> +                       ret = gpd_read_rpm(data, channel);
>
>                         if (ret < 0)
>                                 return ret;
> @@ -529,10 +574,10 @@ static int gpd_fan_hwmon_read(struct device *dev,
>         } else if (type == hwmon_pwm) {
>                 switch (attr) {
>                 case hwmon_pwm_enable:
> -                       *val = data->pwm_enable;
> +                       *val = data->pwm_enable[channel];
>                         return 0;
>                 case hwmon_pwm_input:
> -                       ret = gpd_read_pwm(data);
> +                       ret = gpd_read_pwm(data, channel);
>
>                         if (ret < 0)
>                                 return ret;
> @@ -547,7 +592,7 @@ static int gpd_fan_hwmon_read(struct device *dev,
>
>  static int gpd_fan_hwmon_write(struct device *dev,
>                                enum hwmon_sensor_types type, u32 attr,
> -                              __always_unused int channel, long val)
> +                              int channel, long val)
>  {
>         struct gpd_fan_data *data = dev_get_drvdata(dev);
>
> @@ -557,17 +602,17 @@ static int gpd_fan_hwmon_write(struct device *dev,
>                         if (!in_range(val, 0, 3))
>                                 return -EINVAL;
>
> -                       data->pwm_enable = val;
> +                       data->pwm_enable[channel] = val;
>
> -                       gpd_set_pwm_enable(data, data->pwm_enable);
> +                       gpd_set_pwm_enable(data, channel, data->pwm_enable[channel]);
>                         return 0;
>                 case hwmon_pwm_input:
>                         if (!in_range(val, 0, 256))
>                                 return -EINVAL;
>
> -                       data->pwm_value = val;
> +                       data->pwm_value[channel] = val;
>
> -                       return gpd_write_pwm(data, val);
> +                       return gpd_write_pwm(data, channel, val);
>                 }
>         }
>
> @@ -581,8 +626,9 @@ static const struct hwmon_ops gpd_fan_ops = {
>  };
>
>  static const struct hwmon_channel_info *gpd_fan_hwmon_channel_info[] = {
> -       HWMON_CHANNEL_INFO(fan, HWMON_F_INPUT),
> -       HWMON_CHANNEL_INFO(pwm, HWMON_PWM_INPUT | HWMON_PWM_ENABLE),
> +       HWMON_CHANNEL_INFO(fan, HWMON_F_INPUT, HWMON_F_INPUT),
> +       HWMON_CHANNEL_INFO(pwm, HWMON_PWM_INPUT | HWMON_PWM_ENABLE,
> +                          HWMON_PWM_INPUT | HWMON_PWM_ENABLE),

Are you sure the correct number of fans are populated for each device
because of the visible entry? If yes, it is fine for this hunk. If no,
needs dynamic population for the struct in probe.

You also add a board match for win5, but doesnt the duo have two fans?
I think we had this discussion last year and I said two fans more
complexity lets leave it. But if we do it properly this time lets do
it properly.

Verify which duo-like devices implement two fans, it should be all of
them. I think Cryolitia has a Duo. Win Max 3 according to my GPD
contact has a duo layout. Unfortunately, I have no users so no DMI. So
it is at least Duo, Win Max 3, and Win5. Then make it so only these 3
devices (which should need no separation) report two fans and two fan
speeds throughout the driver.

Best,
Antheas

>         NULL
>  };
>
> @@ -619,8 +665,10 @@ static void gpd_fan_reset_hardware(void *pdata)
>         struct gpd_fan_data *data = pdata;
>
>         if (data) {
> -               data->pwm_enable = AUTOMATIC;
> -               gpd_set_pwm_enable(data, AUTOMATIC);
> +               for (int i = 0; i < gpd_num_fans(data->drvdata); i++) {
> +                       data->pwm_enable[i] = AUTOMATIC;
> +                       gpd_set_pwm_enable(data, i, AUTOMATIC);
> +               }
>         }
>  }
>
> @@ -654,8 +702,10 @@ static int gpd_fan_probe(struct platform_device *pdev)
>                 return -EINVAL;
>
>         data->drvdata = match;
> -       data->pwm_enable = AUTOMATIC;
> -       data->pwm_value = 255;
> +       for (int i = 0; i < gpd_num_fans(match); i++) {
> +               data->pwm_enable[i] = AUTOMATIC;
> +               data->pwm_value[i] = 255;
> +       }
>
>         dev_set_drvdata(dev, data);
>
>
> --
> 2.55.0
>
>
>

Re: [PATCH v2 2/2] hwmon: (gpd-fan) Register both fans on the GPD Win 5

[Alexei Solcanu]

On Sat, 26 Sept 2026 at 22:32, Antheas Kapenekakis <lkml@antheas.dev> wrote:
> in my view adding all these functions is unneeded complexity. There
> should be some simpler ternary way.

Ok, I'll inline them as ternaries in v3.

> Are you sure the correct number of fans are populated for each device
> because of the visible entry? If yes, it is fine for this hunk. If no,
> needs dynamic population for the struct in probe.

Yes. The hwmon core only creates the attributes for which is_visible()
returns non-zero. On the Win 5 both channels show up and work
independently. Loading the driver on it with gpd_fan_board=win_mini
works as expected: only fan1_input, pwm1 and pwm1_enable are created.

> You also add a board match for win5, but doesnt the duo have two fans?
[...]
> Verify which duo-like devices implement two fans, it should be all of
> them. I think Cryolitia has a Duo. Win Max 3 according to my GPD
> contact has a duo layout. Unfortunately, I have no users so no DMI. So
> it is at least Duo, Win Max 3, and Win5. Then make it so only these 3
> devices (which should need no separation) report two fans and two fan
> speeds throughout the driver.

Agreed. In v3 I'll drop the separate win5 board and give the duo board
data two channels instead, so the Duo and the Win 5 share it.

I don't have a Win Max 3, so I can't add its DMI string, but I'll add
the duo board data to gpd_module_drvdata, so gpd_fan_board=duo can be
used on it until it has a DMI entry.

I only have a Win 5, so I can't verify the Duo. On the Win 5 the
second fan's RPM is at 0x0476/0x0477. Cryolitia, could you confirm
that the Duo uses the same register?

Re: [PATCH v2 2/2] hwmon: (gpd-fan) Register both fans on the GPD Win 5

[Antheas Kapenekakis]

On Sat, 26 Sept 2026 at 23:36, Alexei Solcanu <alexei.solcanu@pm.me> wrote:
>
> On Sat, 26 Sept 2026 at 22:32, Antheas Kapenekakis <lkml@antheas.dev> wrote:
> > in my view adding all these functions is unneeded complexity. There
> > should be some simpler ternary way.
>
> Ok, I'll inline them as ternaries in v3.
>
> > Are you sure the correct number of fans are populated for each device
> > because of the visible entry? If yes, it is fine for this hunk. If no,
> > needs dynamic population for the struct in probe.
>
> Yes. The hwmon core only creates the attributes for which is_visible()
> returns non-zero. On the Win 5 both channels show up and work
> independently. Loading the driver on it with gpd_fan_board=win_mini
> works as expected: only fan1_input, pwm1 and pwm1_enable are created.
>
> > You also add a board match for win5, but doesnt the duo have two fans?
> [...]
> > Verify which duo-like devices implement two fans, it should be all of
> > them. I think Cryolitia has a Duo. Win Max 3 according to my GPD
> > contact has a duo layout. Unfortunately, I have no users so no DMI. So
> > it is at least Duo, Win Max 3, and Win5. Then make it so only these 3
> > devices (which should need no separation) report two fans and two fan
> > speeds throughout the driver.
>
> Agreed. In v3 I'll drop the separate win5 board and give the duo board
> data two channels instead, so the Duo and the Win 5 share it.
>
> I don't have a Win Max 3, so I can't add its DMI string, but I'll add
> the duo board data to gpd_module_drvdata, so gpd_fan_board=duo can be
> used on it until it has a DMI entry.
>
> I only have a Win 5, so I can't verify the Duo. On the Win 5 the
> second fan's RPM is at 0x0476/0x0477. Cryolitia, could you confirm
> that the Duo uses the same register?

Actually, recalling what we said last year duo had the new layout but
only one fan if I recall. Cryolitia can confirm. Then, only Max 3
needs confirming but even then you still need the new quirk so most of
this patch is good as is

Best,
Antheas

>

Re: [PATCH v2 2/2] hwmon: (gpd-fan) Register both fans on the GPD Win 5

[Alexei Solcanu]

On Sat, 26 Sept 2026 at 23:52, Antheas Kapenekakis <lkml@antheas.dev> wrote:
> Actually, recalling what we said last year duo had the new layout but
> only one fan if I recall. Cryolitia can confirm. Then, only Max 3
> needs confirming but even then you still need the new quirk so most of
> this patch is good as is

Cryolitia confirmed the register layout last year in the out-of-tree
driver [1]: the Duo has two duty cycle registers but only one RPM
register, while the Win 5 has two of each.

So v3 keeps the separate win5 board data, inlines the helpers as
ternaries, and adds win5 to gpd_module_drvdata so gpd_fan_board=win5
can be used on the Win Max 3 until it has a DMI entry.

[1] https://github.com/Cryolitia/gpd-fan-driver/pull/21

Re: [PATCH v2 2/2] hwmon: (gpd-fan) Register both fans on the GPD Win 5

[Guenter Roeck]

On Sat, Sep 26, 2026 at 04:11:21PM -0400, Alexei Solcanu via B4 Relay wrote:
> From: Alexei Solcanu <alexei.solcanu@pm.me>
> 
> The GPD Win 5 uses gpd_duo_drvdata, which exposes a single fan and pwm
> channel: pwm1 writes the same duty cycle to both fans, and fan1_input
> only reads the first fan's RPM.
> 
> The two fans have independent registers. Fan 1 uses duty cycle 0x047A
> and RPM 0x0478/0x0479, fan 2 uses duty cycle 0x047B and RPM
> 0x0476/0x0477. A duty cycle of 0 returns that fan to EC control.
> 
> Add separate board data for the Win 5 and register fan1/pwm1 and
> fan2/pwm2, one channel per fan. Boards with a single fan are unchanged.
> 
> This changes behaviour on the Win 5: pwm1 and pwm1_enable now control
> only the first fan. The second fan stays under EC control unless pwm2
> and pwm2_enable are used.
> 
> The GPD Duo keeps a single channel, since it is untested and its
> second fan's RPM register is unknown.
> 
> Tested on a GPD Win 5 (G1618-05).
> 
> Suggested-by: Guenter Roeck <linux@roeck-us.net>
> Assisted-by: LLM
> Signed-off-by: Alexei Solcanu <alexei.solcanu@pm.me>

Another patch new revision sent as reply of a previous one.

Where does that idea come from ?  Can _someone_ please let me know ?
It seems like a contagious disease, but no one is willing to admit
where they caught it.

Guenter

Re: [PATCH v2 2/2] hwmon: (gpd-fan) Register both fans on the GPD Win 5

[Antheas Kapenekakis]

On Mon, 28 Sept 2026 at 13:05, Guenter Roeck <linux@roeck-us.net> wrote:
>
> On Sat, Sep 26, 2026 at 04:11:21PM -0400, Alexei Solcanu via B4 Relay wrote:
> > From: Alexei Solcanu <alexei.solcanu@pm.me>
> >
> > The GPD Win 5 uses gpd_duo_drvdata, which exposes a single fan and pwm
> > channel: pwm1 writes the same duty cycle to both fans, and fan1_input
> > only reads the first fan's RPM.
> >
> > The two fans have independent registers. Fan 1 uses duty cycle 0x047A
> > and RPM 0x0478/0x0479, fan 2 uses duty cycle 0x047B and RPM
> > 0x0476/0x0477. A duty cycle of 0 returns that fan to EC control.
> >
> > Add separate board data for the Win 5 and register fan1/pwm1 and
> > fan2/pwm2, one channel per fan. Boards with a single fan are unchanged.
> >
> > This changes behaviour on the Win 5: pwm1 and pwm1_enable now control
> > only the first fan. The second fan stays under EC control unless pwm2
> > and pwm2_enable are used.
> >
> > The GPD Duo keeps a single channel, since it is untested and its
> > second fan's RPM register is unknown.
> >
> > Tested on a GPD Win 5 (G1618-05).
> >
> > Suggested-by: Guenter Roeck <linux@roeck-us.net>
> > Assisted-by: LLM
> > Signed-off-by: Alexei Solcanu <alexei.solcanu@pm.me>
>
> Another patch new revision sent as reply of a previous one.
>
> Where does that idea come from ?  Can _someone_ please let me know ?
> It seems like a contagious disease, but no one is willing to admit
> where they caught it.

Maybe the following?
https://en.andros.dev/blog/75beece9/sending-patches-via-email-with-git/
https://nickdesaulniers.github.io/blog/2017/05/16/submitting-your-first-patch-to-the-linux-kernel-and-responding-to-feedback/

I faintly recall something like that when i was sending my first patch
a few years ago.

I checked the kernel docs and they are correct now and explicitly say
do not do this. Perhaps that was a recent change though.

Antheas

>
> Guenter
>

Re: [PATCH v2 2/2] hwmon: (gpd-fan) Register both fans on the GPD Win 5

[Guenter Roeck]

On Mon, Sep 28, 2026 at 02:03:57PM +0200, Antheas Kapenekakis wrote:
> On Mon, 28 Sept 2026 at 13:05, Guenter Roeck <linux@roeck-us.net> wrote:
> >
> > On Sat, Sep 26, 2026 at 04:11:21PM -0400, Alexei Solcanu via B4 Relay wrote:
> > > From: Alexei Solcanu <alexei.solcanu@pm.me>
> > >
> > > The GPD Win 5 uses gpd_duo_drvdata, which exposes a single fan and pwm
> > > channel: pwm1 writes the same duty cycle to both fans, and fan1_input
> > > only reads the first fan's RPM.
> > >
> > > The two fans have independent registers. Fan 1 uses duty cycle 0x047A
> > > and RPM 0x0478/0x0479, fan 2 uses duty cycle 0x047B and RPM
> > > 0x0476/0x0477. A duty cycle of 0 returns that fan to EC control.
> > >
> > > Add separate board data for the Win 5 and register fan1/pwm1 and
> > > fan2/pwm2, one channel per fan. Boards with a single fan are unchanged.
> > >
> > > This changes behaviour on the Win 5: pwm1 and pwm1_enable now control
> > > only the first fan. The second fan stays under EC control unless pwm2
> > > and pwm2_enable are used.
> > >
> > > The GPD Duo keeps a single channel, since it is untested and its
> > > second fan's RPM register is unknown.
> > >
> > > Tested on a GPD Win 5 (G1618-05).
> > >
> > > Suggested-by: Guenter Roeck <linux@roeck-us.net>
> > > Assisted-by: LLM
> > > Signed-off-by: Alexei Solcanu <alexei.solcanu@pm.me>
> >
> > Another patch new revision sent as reply of a previous one.
> >
> > Where does that idea come from ?  Can _someone_ please let me know ?
> > It seems like a contagious disease, but no one is willing to admit
> > where they caught it.
> 
> Maybe the following?
> https://en.andros.dev/blog/75beece9/sending-patches-via-email-with-git/
> https://nickdesaulniers.github.io/blog/2017/05/16/submitting-your-first-patch-to-the-linux-kernel-and-responding-to-feedback/
> 

Sigh. That "keeping together" does cause me to miss new patch revisions
on a regular basis. Oh well. At some point I may just create an
auto-responder telling people that I am not responding to nested
revisions, and copy the authors of those articles.

> I faintly recall something like that when i was sending my first patch
> a few years ago.
> 
> I checked the kernel docs and they are correct now and explicitly say
> do not do this. Perhaps that was a recent change though.

No, that was not a recent change, unless you count commit d7ac8d85d39f4
("Documentation/SubmittingPatches: discuss In-Reply-To") from 2015 as a
"recent change".

Guenter

Re: [PATCH v2 2/2] hwmon: (gpd-fan) Register both fans on the GPD Win 5

[Alexei Solcanu]

On Mon, Sep 28, 2026 at 11:05 UTC, Guenter Roeck wrote:
 > Another patch new revision sent as reply of a previous one.

I sent v2 as a new thread with b4, it's not in reply to v1.
v2 includes the original commit from v1 as patch 1/2.
This is my first patch so maybe I'm just reading this wrong.

v1: 
https://lore.kernel.org/linux-hwmon/20260925-gpd-fan-duo-auto-v1-1-29bb670b3331@pm.me/T/
v2: 
https://lore.kernel.org/linux-hwmon/20260926-gpd-fan-duo-auto-v2-0-0a2ea41ee3b5@pm.me/T/

Alexei