iio: accel: kxcjk1013: Fix kbd/touchpad not working on some 2-in-1s
cover iio: accel: kxcjk1013: Fix kbd/touchpad not working on some 2-in-1s
Hi All,
This series fixes a somewhat special bug where sometimes the kbd/touchpad do not work on certain 2-in-1s with 360 degree (yoga) hinges, which use 2 accelerometers to allow figuring out the angle between the display and the base.
The display accelerometer ACPI device / fwnode has a special ACPI method which allows suppressing kbd + touchpads events when folded into tablet-mode and sometimes the event suppression gets enabled by default, this series calls the ACPI method at probe time to disable the suppression, fixing the kbd/touchpad not working.
A web-search shows various reports about this issue in forums and such, so it would be good if we can get these 2 patches added to a 5.10-rc# as a bugfix, as I indicated in the Subject prefix.
Regards,
Hans
PATCH 1/2 iio: accel: kxcjk1013: Replace is_smo8500_device with an acpi_type enum
Replace the boolean is_smo8500_device variable with an acpi_type enum.
For now this can be either ACPI_GENERIC or ACPI_SMO8500, this is a preparation patch for adding special handling for the KIOX010A ACPI HID, which will add a ACPI_KIOX010A acpi_type to the introduced enum.
--- drivers/iio/accel/kxcjk-1013.c | 15 ++++++++++----- 1 file changed, 10 insertions(+), 5 deletions(-)
drivers/iio/accel/kxcjk-1013.c +10 -5
@@ -126,6 +126,11 @@ enum kx_chipset {KX_MAX_CHIPS /* this must be last */ }; +enum kx_acpi_type { +ACPI_GENERIC , +ACPI_SMO8500 , +}; +struct kxcjk1013_data {struct i2c_client *client;struct iio_trigger *dready_trig;
@@ -143,7 +148,7 @@ struct kxcjk1013_data {bool motion_trigger_on;int64_t timestamp;enum kx_chipset chipset; -bool is_smo8500_device; +enum kx_acpi_type acpi_type; };enum kxcjk1013_axis {
@@ -1247,7 +1252,7 @@ static irqreturn_t kxcjk1013_data_rdy_trig_poll(int irq, void *private)static const char *kxcjk1013_match_acpi_device (struct device *dev,enum kx_chipset *chipset, -bool *is_smo8500_device) +enum kx_acpi_type *acpi_type) {const struct acpi_device_id *id;
@@ -1256,7 +1261,7 @@ static const char *kxcjk1013_match_acpi_device(struct device *dev,return NULL ;if (strcmp (id->id,"SMO8500" ) ==0 ) - *is_smo8500_device = true; + *acpi_type =ACPI_SMO8500 ; *chipset = (enum kx_chipset )id->driver_data;
@@ -1299,7 +1304,7 @@ static int kxcjk1013_probe(struct i2c_client *client, }else if (ACPI_HANDLE (&client->dev)) { name =kxcjk1013_match_acpi_device (&client->dev, &data->chipset, - &data->is_smo8500_device); + &data->acpi_type); }else return -ENODEV ;
@@ -1316,7 +1321,7 @@ static int kxcjk1013_probe(struct i2c_client *client, indio_dev->modes =INDIO_DIRECT_MODE ; indio_dev->info = &kxcjk1013_info; -if (client->irq >0 && !data->is_smo8500_device) { +if (client->irq >0 && data->acpi_type !=ACPI_SMO8500 ) { ret = devm_request_threaded_irq(&client->dev, client->irq, kxcjk1013_data_rdy_trig_poll, kxcjk1013_event_handler,
Signature · 1 line
2.28.0
PATCH 2/2 iio: accel: kxcjk1013: Add support for KIOX010A ACPI DSM for setting tablet-mode
In reply to Hans de GoedeSome 360 degree hinges (yoga) style 2-in-1 devices use 2 KXCJ91008-s to allow the OS to determine the angle between the display and the base of the device, so that the OS can determine if the 2-in-1 is in laptop or in tablet-mode.
On Windows both accelerometers are read by a special HingeAngleService process; and this process calls a DSM (Device Specific Method) on the ACPI KIOX010A device node for the sensor in the display, to let the embedded-controller (EC) know about the mode so that it can disable the kbd and touchpad to avoid spurious input while folded into tablet-mode.
This notifying of the EC is problematic because sometimes the EC comes up thinking that device is in tablet-mode and the kbd and touchpad do not work. This happens for example on Irbis NB111 devices after a suspend / resume cycle (after a complete battery drain / hard reset without having booted Windows at least once). Other 2-in-1s which are likely affected too are e.g. the Teclast F5 and F6 series.
The kxcjk-1013 driver may seem like a strange place to deal with this, but since it is *the* driver for the ACPI KIOX010A device, it is also the driver which has access to the ACPI handle needed by the DSM.
Add support for calling the DSM and on probe unconditionally tell the EC that the device is laptop mode, fixing the kbd and touchpad sometimes not working.
--- drivers/iio/accel/kxcjk-1013.c | 36 ++++++++++++++++++++++++++++++++++ 1 file changed, 36 insertions(+)
drivers/iio/accel/kxcjk-1013.c +36 -0
@@ -129,6 +129,7 @@ enum kx_chipset {enum kx_acpi_type {ACPI_GENERIC ,ACPI_SMO8500 , +ACPI_KIOX010A , };struct kxcjk1013_data {
@@ -275,6 +276,32 @@ static const struct { {19163, 1, 0}, {38326, 0, 1} }; +#ifdef CONFIG_ACPI +enum kiox010a_fn_index { +KIOX010A_SET_LAPTOP_MODE =1 , +KIOX010A_SET_TABLET_MODE =2 , +}; + +static int kiox010a_dsm (struct device *dev,int fn_index) +{ +acpi_handle handle =ACPI_HANDLE (dev); +guid_t kiox010a_dsm_guid; +union acpi_object *obj; + +if (!handle) +return -ENODEV ; + +guid_parse ("1f339696-d475-4e26-8cad-2e9f8e6d7a91" , &kiox010a_dsm_guid); + + obj =acpi_evaluate_dsm (handle, &kiox010a_dsm_guid,1 , fn_index,NULL ); +if (!obj) +return -EIO ; + +ACPI_FREE (obj); +return 0 ; +} +#endif +static int kxcjk1013_set_mode (struct kxcjk1013_data *data,enum kxcjk1013_mode mode) {
@@ -352,6 +379,13 @@ static int kxcjk1013_chip_init(struct kxcjk1013_data *data) {int ret; +#ifdef CONFIG_ACPI +if (data->acpi_type ==ACPI_KIOX010A ) { +/* Make sure the kbd and touchpad on 2-in-1s using 2 KXCJ91008-s work */ +kiox010a_dsm (&data->client->dev,KIOX010A_SET_LAPTOP_MODE ); + } +#endif + ret =i2c_smbus_read_byte_data (data->client,KXCJK1013_REG_WHO_AM_I );if (ret <0 ) {dev_err (&data->client->dev,"Error reading who_am_i\n" );
@@ -1262,6 +1296,8 @@ static const char *kxcjk1013_match_acpi_device(struct device *dev,if (strcmp (id->id,"SMO8500" ) ==0 ) *acpi_type =ACPI_SMO8500 ; +else if (strcmp (id->id,"KIOX010A" ) ==0 ) + *acpi_type =ACPI_KIOX010A ; *chipset = (enum kx_chipset )id->driver_data;
Signature · 1 line
2.28.0
Jonathan Cameron
maintainerRe: [PATCH bugfix for 5.10 2/2] iio: accel: kxcjk1013: Add support for KIOX010A ACPI DSM for setting tablet-mode
Quoted 22 lines · On Tue, 10 Nov 2020 14:38:35 +0100 Hans de Goede <[email protected]> wrote:
Some 360 degree hinges (yoga) style 2-in-1 devices use 2 KXCJ91008-s to allow the OS to determine the angle between the display and the base of the device, so that the OS can determine if the 2-in-1 is in laptop or in tablet-mode.
On Windows both accelerometers are read by a special HingeAngleService process; and this process calls a DSM (Device Specific Method) on the ACPI KIOX010A device node for the sensor in the display, to let the embedded-controller (EC) know about the mode so that it can disable the kbd and touchpad to avoid spurious input while folded into tablet-mode.
This notifying of the EC is problematic because sometimes the EC comes up thinking that device is in tablet-mode and the kbd and touchpad do not work. This happens for example on Irbis NB111 devices after a suspend / resume cycle (after a complete battery drain / hard reset without having booted Windows at least once). Other 2-in-1s which are likely affected too are e.g. the Teclast F5 and F6 series.
The kxcjk-1013 driver may seem like a strange place to deal with this, but since it is *the* driver for the ACPI KIOX010A device, it is also the driver which has access to the ACPI handle needed by the DSM.
Quoted 6 lines
Add support for calling the DSM and on probe unconditionally tell the EC that the device is laptop mode, fixing the kbd and touchpad sometimes not working.
Reported-and-tested-by russianneuromancer <[email protected]>Signed-off-by Hans de Goede <[email protected]>
Hi Hans,
*Mutters darkly about crazy firmware hacks*
I'm fine taking this but I assume we want to backport and for that I'm after a fixes tag.
Thanks,
Jonathan
Quoted 72 lines
--- drivers/iio/accel/kxcjk-1013.c | 36 ++++++++++++++++++++++++++++++++++ 1 file changed, 36 insertions(+)drivers/iio/accel/kxcjk-1013.c +36
-0@@ -129,6 +129,7 @@ enum kx_chipset {enum kx_acpi_type {ACPI_GENERIC ,ACPI_SMO8500 , +ACPI_KIOX010A , };struct kxcjk1013_data {@@ -275,6 +276,32 @@ static const struct { {19163, 1, 0}, {38326, 0, 1} }; +#ifdef CONFIG_ACPI +enum kiox010a_fn_index { +KIOX010A_SET_LAPTOP_MODE =1 , +KIOX010A_SET_TABLET_MODE =2 , +}; + +static int kiox010a_dsm (struct device *dev,int fn_index) +{ +acpi_handle handle =ACPI_HANDLE (dev); +guid_t kiox010a_dsm_guid; +union acpi_object *obj; + +if (!handle) +return -ENODEV ; + +guid_parse ("1f339696-d475-4e26-8cad-2e9f8e6d7a91" , &kiox010a_dsm_guid); + + obj =acpi_evaluate_dsm (handle, &kiox010a_dsm_guid,1 , fn_index,NULL ); +if (!obj) +return -EIO ; + +ACPI_FREE (obj); +return 0 ; +} +#endif +static int kxcjk1013_set_mode (struct kxcjk1013_data *data,enum kxcjk1013_mode mode) {@@ -352,6 +379,13 @@ static int kxcjk1013_chip_init(struct kxcjk1013_data *data) {int ret; +#ifdef CONFIG_ACPI +if (data->acpi_type ==ACPI_KIOX010A ) { +/* Make sure the kbd and touchpad on 2-in-1s using 2 KXCJ91008-s work */ +kiox010a_dsm (&data->client->dev,KIOX010A_SET_LAPTOP_MODE ); + } +#endif + ret =i2c_smbus_read_byte_data (data->client,KXCJK1013_REG_WHO_AM_I );if (ret <0 ) {dev_err (&data->client->dev,"Error reading who_am_i\n" );@@ -1262,6 +1296,8 @@ static const char *kxcjk1013_match_acpi_device(struct device *dev,if (strcmp (id->id,"SMO8500" ) ==0 ) *acpi_type =ACPI_SMO8500 ; +else if (strcmp (id->id,"KIOX010A" ) ==0 ) + *acpi_type =ACPI_KIOX010A ; *chipset = (enum kx_chipset )id->driver_data;
Re: [PATCH bugfix for 5.10 2/2] iio: accel: kxcjk1013: Add support for KIOX010A ACPI DSM for setting tablet-mode
Hi,
Quoted 29 lines · Jonathan Cameron
Quoted 26 lines · On Tue, 10 Nov 2020 14:38:35 +0100 Hans de Goede <[email protected]> wrote:
Some 360 degree hinges (yoga) style 2-in-1 devices use 2 KXCJ91008-s to allow the OS to determine the angle between the display and the base of the device, so that the OS can determine if the 2-in-1 is in laptop or in tablet-mode.
On Windows both accelerometers are read by a special HingeAngleService process; and this process calls a DSM (Device Specific Method) on the ACPI KIOX010A device node for the sensor in the display, to let the embedded-controller (EC) know about the mode so that it can disable the kbd and touchpad to avoid spurious input while folded into tablet-mode.
This notifying of the EC is problematic because sometimes the EC comes up thinking that device is in tablet-mode and the kbd and touchpad do not work. This happens for example on Irbis NB111 devices after a suspend / resume cycle (after a complete battery drain / hard reset without having booted Windows at least once). Other 2-in-1s which are likely affected too are e.g. the Teclast F5 and F6 series.
The kxcjk-1013 driver may seem like a strange place to deal with this, but since it is *the* driver for the ACPI KIOX010A device, it is also the driver which has access to the ACPI handle needed by the DSM.
Add support for calling the DSM and on probe unconditionally tell the EC that the device is laptop mode, fixing the kbd and touchpad sometimes not working.
Quoted 8 lines
Quoted 2 lines
Reported-and-tested-by russianneuromancer <[email protected]>Signed-off-by Hans de Goede <[email protected]>Hi Hans,
*Mutters darkly about crazy firmware hacks*
I'm fine taking this but I assume we want to backport and for that I'm after a fixes tag.
Good point, I guess taking the commit which originally added the KIOX010A ACPI Hardware-ID makes the most sense:
Regards,
Hans
Quoted 73 lines
Quoted 72 lines
--- drivers/iio/accel/kxcjk-1013.c | 36 ++++++++++++++++++++++++++++++++++ 1 file changed, 36 insertions(+)drivers/iio/accel/kxcjk-1013.c +36
-0@@ -129,6 +129,7 @@ enum kx_chipset {enum kx_acpi_type {ACPI_GENERIC ,ACPI_SMO8500 , +ACPI_KIOX010A , };struct kxcjk1013_data {@@ -275,6 +276,32 @@ static const struct { {19163, 1, 0}, {38326, 0, 1} }; +#ifdef CONFIG_ACPI +enum kiox010a_fn_index { +KIOX010A_SET_LAPTOP_MODE =1 , +KIOX010A_SET_TABLET_MODE =2 , +}; + +static int kiox010a_dsm (struct device *dev,int fn_index) +{ +acpi_handle handle =ACPI_HANDLE (dev); +guid_t kiox010a_dsm_guid; +union acpi_object *obj; + +if (!handle) +return -ENODEV ; + +guid_parse ("1f339696-d475-4e26-8cad-2e9f8e6d7a91" , &kiox010a_dsm_guid); + + obj =acpi_evaluate_dsm (handle, &kiox010a_dsm_guid,1 , fn_index,NULL ); +if (!obj) +return -EIO ; + +ACPI_FREE (obj); +return 0 ; +} +#endif +static int kxcjk1013_set_mode (struct kxcjk1013_data *data,enum kxcjk1013_mode mode) {@@ -352,6 +379,13 @@ static int kxcjk1013_chip_init(struct kxcjk1013_data *data) {int ret; +#ifdef CONFIG_ACPI +if (data->acpi_type ==ACPI_KIOX010A ) { +/* Make sure the kbd and touchpad on 2-in-1s using 2 KXCJ91008-s work */ +kiox010a_dsm (&data->client->dev,KIOX010A_SET_LAPTOP_MODE ); + } +#endif + ret =i2c_smbus_read_byte_data (data->client,KXCJK1013_REG_WHO_AM_I );if (ret <0 ) {dev_err (&data->client->dev,"Error reading who_am_i\n" );@@ -1262,6 +1296,8 @@ static const char *kxcjk1013_match_acpi_device(struct device *dev,if (strcmp (id->id,"SMO8500" ) ==0 ) *acpi_type =ACPI_SMO8500 ; +else if (strcmp (id->id,"KIOX010A" ) ==0 ) + *acpi_type =ACPI_KIOX010A ; *chipset = (enum kx_chipset )id->driver_data;
Re: [PATCH bugfix for 5.10 2/2] iio: accel: kxcjk1013: Add support for KIOX010A ACPI DSM for setting tablet-mode
Quoted 38 lines · On Sat, 14 Nov 2020 18:07:38 +0100 Hans de Goede <[email protected]> wrote:
Hi,
Quoted 35 lines · Jonathan Cameron
Quoted 22 lines · On Tue, 10 Nov 2020 14:38:35 +0100 Hans de Goede <[email protected]> wrote:
Some 360 degree hinges (yoga) style 2-in-1 devices use 2 KXCJ91008-s to allow the OS to determine the angle between the display and the base of the device, so that the OS can determine if the 2-in-1 is in laptop or in tablet-mode.
On Windows both accelerometers are read by a special HingeAngleService process; and this process calls a DSM (Device Specific Method) on the ACPI KIOX010A device node for the sensor in the display, to let the embedded-controller (EC) know about the mode so that it can disable the kbd and touchpad to avoid spurious input while folded into tablet-mode.
This notifying of the EC is problematic because sometimes the EC comes up thinking that device is in tablet-mode and the kbd and touchpad do not work. This happens for example on Irbis NB111 devices after a suspend / resume cycle (after a complete battery drain / hard reset without having booted Windows at least once). Other 2-in-1s which are likely affected too are e.g. the Teclast F5 and F6 series.
The kxcjk-1013 driver may seem like a strange place to deal with this, but since it is *the* driver for the ACPI KIOX010A device, it is also the driver which has access to the ACPI handle needed by the DSM.
Quoted 6 lines
Add support for calling the DSM and on probe unconditionally tell the EC that the device is laptop mode, fixing the kbd and touchpad sometimes not working.
Reported-and-tested-by russianneuromancer <[email protected]>Signed-off-by Hans de Goede <[email protected]>Hi Hans,
*Mutters darkly about crazy firmware hacks*
Quoted 7 lines
Quoted 2 lines
I'm fine taking this but I assume we want to backport and for that I'm after a fixes tag.
Good point, I guess taking the commit which originally added the KIOX010A ACPI Hardware-ID makes the most sense:
Fixes 7f6232e69539 ("iio: accel: kxcjk1013: Add KIOX010A ACPI Hardware-ID")
Agreed.
Applied to the fixes-togreg branch of iio.git and marked for stable.
Thanks,
Jonathan
Quoted 81 lines
Regards,
Hans
Quoted 73 lines
Quoted 72 lines
--- drivers/iio/accel/kxcjk-1013.c | 36 ++++++++++++++++++++++++++++++++++ 1 file changed, 36 insertions(+)drivers/iio/accel/kxcjk-1013.c +36
-0@@ -129,6 +129,7 @@ enum kx_chipset {enum kx_acpi_type {ACPI_GENERIC ,ACPI_SMO8500 , +ACPI_KIOX010A , };struct kxcjk1013_data {@@ -275,6 +276,32 @@ static const struct { {19163, 1, 0}, {38326, 0, 1} }; +#ifdef CONFIG_ACPI +enum kiox010a_fn_index { +KIOX010A_SET_LAPTOP_MODE =1 , +KIOX010A_SET_TABLET_MODE =2 , +}; + +static int kiox010a_dsm (struct device *dev,int fn_index) +{ +acpi_handle handle =ACPI_HANDLE (dev); +guid_t kiox010a_dsm_guid; +union acpi_object *obj; + +if (!handle) +return -ENODEV ; + +guid_parse ("1f339696-d475-4e26-8cad-2e9f8e6d7a91" , &kiox010a_dsm_guid); + + obj =acpi_evaluate_dsm (handle, &kiox010a_dsm_guid,1 , fn_index,NULL ); +if (!obj) +return -EIO ; + +ACPI_FREE (obj); +return 0 ; +} +#endif +static int kxcjk1013_set_mode (struct kxcjk1013_data *data,enum kxcjk1013_mode mode) {@@ -352,6 +379,13 @@ static int kxcjk1013_chip_init(struct kxcjk1013_data *data) {int ret; +#ifdef CONFIG_ACPI +if (data->acpi_type ==ACPI_KIOX010A ) { +/* Make sure the kbd and touchpad on 2-in-1s using 2 KXCJ91008-s work */ +kiox010a_dsm (&data->client->dev,KIOX010A_SET_LAPTOP_MODE ); + } +#endif + ret =i2c_smbus_read_byte_data (data->client,KXCJK1013_REG_WHO_AM_I );if (ret <0 ) {dev_err (&data->client->dev,"Error reading who_am_i\n" );@@ -1262,6 +1296,8 @@ static const char *kxcjk1013_match_acpi_device(struct device *dev,if (strcmp (id->id,"SMO8500" ) ==0 ) *acpi_type =ACPI_SMO8500 ; +else if (strcmp (id->id,"KIOX010A" ) ==0 ) + *acpi_type =ACPI_KIOX010A ; *chipset = (enum kx_chipset )id->driver_data;