Skip to contentSkip to conversation

Bug#1029850: linux: Driver not loaded for ST Microelectronics LSM6DS3TR-C accelerometer (acpi:SMO8B30:SMO8B30:)

DISCUSSION Jonathan Cameron 5 participants 29 messages Jan 29, 2023 – Feb 5, 2023 4 messages not in the archive

Subsystems (2)
lore
Share & tools
lore

A message that is not in the archive (sent off-list or not yet received): 167493679618.4533.12181720504943588640.reportbug@debian-duet

A message that is not in the archive (sent off-list or not yet received): [email protected]

A message that is not in the archive (sent off-list or not yet received): CAMxBKG1670TFuV3nHP7Yk8s6H+oBF7iiyiB-b=PvKv9hcH22xQ@mail.gmail.com

Jonathan Cameron

author maintainer
#3 lore raw
Quoted 5 lines · On Sun, 29 Jan 2023 17:03:51 +0000 Darrell Kavanagh…

Hi,

I raised this bug in Debian, and have been asked to raise it upstream and was given your addresses to do so. Will this email be OK, or should I raise it in a bug tracking system somewhere?

Email is the right option.

Quoted 27 lines

Many thanks,
Darrell

---------- Forwarded message ---------
From: Salvatore Bonaccorso <[email protected]>
Date: Sat, 28 Jan 2023 at 20:33
Subject: Re: Bug#1029850: linux: Driver not loaded for ST Microelectronics LSM6DS3TR-C accelerometer (acpi:SMO8B30:SMO8B30:)
To: Darrell Kavanagh <[email protected]>, <[email protected]>

Hi Darrell,

Quoted 9 lines · Darrell Kavanagh

Package: src:linux
Version: 6.1.4-1
Severity: normal
File: linux
X-Debbugs-Cc: [email protected]

Dear Maintainer,

This is a convertable touchscreen tablet/laptop. The rotation sensor

Quoted 9 lines

device

Quoted 1 line

ST Microelectronics LSM6DS3TR-C does not work. It is detected via ACPI

and the

Quoted 1 line

sysfs trees are created at

devices/pci0000:00/0000:00:17.1/i2c_designware.3/i2c-4/i2c-SMO8B30:00

Quoted 2 lines

and devices/LNXSYSTM:00/LNXSYBUS:00/PNP0A08:00/device:3c/SMO8B30:00 with symlinks bus/acpi/devices/SMO8B30:00 and bus/i2c/devices/i2c-SMO8B30:00,

but

Quoted 1 line

no driver is loaded.

At least this is using the ST PNP ID which is better than average (long story!)

The driver in question (ultimately drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c does not currently have an ACPI support. It should be straight forwards to add though the driver first needs converting to use device_get_match_data() with appropriate fallback so that it will match on ACPI, OF or original spi_device_id tables

Completely untested but something like the following
(the offset in the enum is needed to allow us to tell if we got a result when calling device_get_match_data() as it returns NULL on failure IIRC)

I'm not sure how sucessful the driver will be at finding any interrupts etc, but it may get you basic functionality.

Good luck and others more familiar with the driver may well tell me what I forgot when hacking the below ;)

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx.h +1 -1
index 499fcf8875b4..2617ce236ddc 100644
@@ -39,7 +39,7 @@
 #define ST_ISM330IS_DEV_NAME   "ism330is"
 
 enum st_lsm6dsx_hw_id {
-       ST_LSM6DS3_ID,
+       ST_LSM6DS3_ID = 1,
        ST_LSM6DS3H_ID,
        ST_LSM6DSL_ID,
        ST_LSM6DSM_ID,
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c +12 -2
index df5f60925260..ecfceb2fb3db 100644
@@ -23,10 +23,15 @@ static const struct regmap_config st_lsm6dsx_i2c_regmap_config = {
 
 static int st_lsm6dsx_i2c_probe(struct i2c_client *client)
 {
-       const struct i2c_device_id *id = i2c_client_get_device_id(client);
-       int hw_id = id->driver_data;
+       int hw_id;
        struct regmap *regmap;
 
+       hw_id = (kernel_ulong_t)device_get_match_data(&client->dev);
+       if (!hw_id)
+               hw_id = i2c_client_get_device_id(client)->driver_data;
+       if (!hw_id)
+               return -EINVAL;
+
        regmap = devm_regmap_init_i2c(client, &st_lsm6dsx_i2c_regmap_config);
        if (IS_ERR(regmap)) {
                dev_err(&client->dev, "Failed to register i2c regmap %ld\n", PTR_ERR(regmap));
@@ -129,6 +134,10 @@ static const struct of_device_id st_lsm6dsx_i2c_of_match[] = {
 };
 MODULE_DEVICE_TABLE(of, st_lsm6dsx_i2c_of_match);
 
+static const struct acpi_device_id st_lsm6dsx_i2c_acpi_match[] = {
+       { "SMO8B30", ST_LSM6DS3TRC_ID, },
+};
+
 static const struct i2c_device_id st_lsm6dsx_i2c_id_table[] = {
        { ST_LSM6DS3_DEV_NAME, ST_LSM6DS3_ID },
        { ST_LSM6DS3H_DEV_NAME, ST_LSM6DS3H_ID },
@@ -161,6 +170,7 @@ static struct i2c_driver st_lsm6dsx_driver = {
                .name = "st_lsm6dsx_i2c",
                .pm = pm_sleep_ptr(&st_lsm6dsx_pm_ops),
                .of_match_table = st_lsm6dsx_i2c_of_match,
+               .acpi_match_table = st_lsm6dsx_i2c_acpi_match,
        },
        .probe_new = st_lsm6dsx_i2c_probe,
        .id_table = st_lsm6dsx_i2c_id_table,
Quoted 33 lines
Quoted 15 lines

The device is identifying itself to the kernel with PNP id SMO8B30:
physical_node:

      modalias=acpi:SMO8B30:SMO8B30:
      name=SMO8B30:00
      uevent=MODALIAS=acpi:SMO8B30:SMO8B30:
      waiting_for_supplier=0
firmware_node:
      hid=SMO8B30
      modalias=acpi:SMO8B30:SMO8B30:
      path=\_SB_.PCI0.I2C5.DEV_
      status=15
      uevent=MODALIAS=acpi:SMO8B30:SMO8B30:
      uid=0
Quoted 1 line

The kernel module for the appropriate driver (st_lsm6dsx_i2c) is not

loaded on boot.

Quoted 2 lines

Modprobing it does not associate it with the device, as I would expect as the module does not provide an alias for the above acpi/pnp id.

Can you report this issue upstream? Gues to reach out are according to get_maintainers.pl script:

Lorenzo Bianconi <[email protected]> (maintainer:ST LSM6DSx IMU IIO DRIVER) Jonathan Cameron <[email protected]> (maintainer:IIO SUBSYSTEM AND DRIVERS) Lars-Peter Clausen <[email protected]> (reviewer:IIO SUBSYSTEM AND DRIVERS) [email protected] (open list:ST LSM6DSx IMU IIO DRIVER) [email protected] (open list)

Please keep us in the loop.

Regards,
Salvatore

A message that is not in the archive (sent off-list or not yet received): CAMxBKG0tyLSpaDPGBXsJbqgHSG9rH6owtSJsLw_ekmTA3Kyvdw@mail.gmail.com

#5 lore raw

Forwarding because original html messages were rejected by the server...

---------- Forwarded message ---------
From: Darrell Kavanagh <[email protected]>
Date: Mon, 30 Jan 2023 at 02:52
Subject: Re: Bug#1029850: linux: Driver not loaded for ST Microelectronics LSM6DS3TR-C accelerometer (acpi:SMO8B30:SMO8B30:)
To: Jonathan Cameron <[email protected]>
Cc: <[email protected]>, <[email protected]>, <[email protected]>, <[email protected]>, <[email protected]>

Hi Jonathan,

Thank you. The driver has evolved quite a bit in 6.2 (I read somewhere that 6.2 includes some i2c enhancements), but I adapted your changes to fit my Debian 6.1 kernel and it works. Two IIO devices are created in sysfs, iio-sensor-proxy.service starts up and automatic screen rotation in Gnome just works.

To get the modules to load on boot, I made a small change to your code in st_lsm6dsx_i2c to add the acpi alias to modules.alias:

adding a null element to st_lsm6dsx_i2c_acpi_match:

    static const struct acpi_device_id st_lsm6dsx_i2c_acpi_match[] = {
         { "SMO8B30", ST_LSM6DS3TRC_ID, },
         { },
    };

then:
MODULE_DEVICE_TABLE(acpi, st_lsm6dsx_i2c_acpi_match);

dmesg shows:

[ 7366.120208] st_lsm6dsx_i2c i2c-SMO8B30:00: supply vdd not found,
using dummy regulator
[ 7366.120260] st_lsm6dsx_i2c i2c-SMO8B30:00: supply vddio not found,
using dummy regulator
[ 7366.650839] st_lsm6dsx_i2c i2c-SMO8B30:00: mounting matrix not

found: using identity...

Is this a problem?

Thanks again.

Darrell

Quoted 158 lines · Jonathan Cameron
Quoted 5 lines · On Sun, 29 Jan 2023 17:03:51 +0000 Darrell Kavanagh…

Hi,

I raised this bug in Debian, and have been asked to raise it upstream and was given your addresses to do so. Will this email be OK, or should I raise it in a bug tracking system somewhere?

Email is the right option.

Quoted 27 lines

Many thanks,
Darrell

---------- Forwarded message ---------
From: Salvatore Bonaccorso <[email protected]>
Date: Sat, 28 Jan 2023 at 20:33
Subject: Re: Bug#1029850: linux: Driver not loaded for ST Microelectronics LSM6DS3TR-C accelerometer (acpi:SMO8B30:SMO8B30:)
To: Darrell Kavanagh <[email protected]>, <[email protected]>

Hi Darrell,

Quoted 9 lines · Darrell Kavanagh

Package: src:linux
Version: 6.1.4-1
Severity: normal
File: linux
X-Debbugs-Cc: [email protected]

Dear Maintainer,

This is a convertable touchscreen tablet/laptop. The rotation sensor

Quoted 9 lines

device

Quoted 1 line

ST Microelectronics LSM6DS3TR-C does not work. It is detected via ACPI

and the

Quoted 1 line

sysfs trees are created at

devices/pci0000:00/0000:00:17.1/i2c_designware.3/i2c-4/i2c-SMO8B30:00

Quoted 2 lines

and devices/LNXSYSTM:00/LNXSYBUS:00/PNP0A08:00/device:3c/SMO8B30:00 with symlinks bus/acpi/devices/SMO8B30:00 and bus/i2c/devices/i2c-SMO8B30:00,

but

Quoted 1 line

no driver is loaded.

At least this is using the ST PNP ID which is better than average (long story!)

The driver in question (ultimately drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c does not currently have an ACPI support. It should be straight forwards to add though the driver first needs converting to use device_get_match_data() with appropriate fallback so that it will match on ACPI, OF or original spi_device_id tables

Completely untested but something like the following
(the offset in the enum is needed to allow us to tell if we got a result when calling device_get_match_data() as it returns NULL on failure IIRC)

I'm not sure how sucessful the driver will be at finding any interrupts etc, but it may get you basic functionality.

Good luck and others more familiar with the driver may well tell me what I forgot when hacking the below ;)

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx.h +1 -1
index 499fcf8875b4..2617ce236ddc 100644
@@ -39,7 +39,7 @@
 #define ST_ISM330IS_DEV_NAME   "ism330is"
 
 enum st_lsm6dsx_hw_id {
-       ST_LSM6DS3_ID,
+       ST_LSM6DS3_ID = 1,
        ST_LSM6DS3H_ID,
        ST_LSM6DSL_ID,
        ST_LSM6DSM_ID,
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c +12 -2
index df5f60925260..ecfceb2fb3db 100644
@@ -23,10 +23,15 @@ static const struct regmap_config st_lsm6dsx_i2c_regmap_config = {
 
 static int st_lsm6dsx_i2c_probe(struct i2c_client *client)
 {
-       const struct i2c_device_id *id = i2c_client_get_device_id(client);
-       int hw_id = id->driver_data;
+       int hw_id;
        struct regmap *regmap;
 
+       hw_id = (kernel_ulong_t)device_get_match_data(&client->dev);
+       if (!hw_id)
+               hw_id = i2c_client_get_device_id(client)->driver_data;
+       if (!hw_id)
+               return -EINVAL;
+
        regmap = devm_regmap_init_i2c(client, &st_lsm6dsx_i2c_regmap_config);
        if (IS_ERR(regmap)) {
                dev_err(&client->dev, "Failed to register i2c regmap %ld\n", PTR_ERR(regmap));
@@ -129,6 +134,10 @@ static const struct of_device_id st_lsm6dsx_i2c_of_match[] = {
 };
 MODULE_DEVICE_TABLE(of, st_lsm6dsx_i2c_of_match);
 
+static const struct acpi_device_id st_lsm6dsx_i2c_acpi_match[] = {
+       { "SMO8B30", ST_LSM6DS3TRC_ID, },
+};
+
 static const struct i2c_device_id st_lsm6dsx_i2c_id_table[] = {
        { ST_LSM6DS3_DEV_NAME, ST_LSM6DS3_ID },
        { ST_LSM6DS3H_DEV_NAME, ST_LSM6DS3H_ID },
@@ -161,6 +170,7 @@ static struct i2c_driver st_lsm6dsx_driver = {
                .name = "st_lsm6dsx_i2c",
                .pm = pm_sleep_ptr(&st_lsm6dsx_pm_ops),
                .of_match_table = st_lsm6dsx_i2c_of_match,
+               .acpi_match_table = st_lsm6dsx_i2c_acpi_match,
        },
        .probe_new = st_lsm6dsx_i2c_probe,
        .id_table = st_lsm6dsx_i2c_id_table,
Quoted 35 lines
Quoted 17 lines

The device is identifying itself to the kernel with PNP id SMO8B30:
physical_node:

      modalias=acpi:SMO8B30:SMO8B30:
      name=SMO8B30:00
      uevent=MODALIAS=acpi:SMO8B30:SMO8B30:
      waiting_for_supplier=0
firmware_node:
      hid=SMO8B30
      modalias=acpi:SMO8B30:SMO8B30:
      path=\_SB_.PCI0.I2C5.DEV_
      status=15
      uevent=MODALIAS=acpi:SMO8B30:SMO8B30:
      uid=0
Quoted 1 line

The kernel module for the appropriate driver (st_lsm6dsx_i2c) is not

loaded on boot.

Quoted 2 lines

Modprobing it does not associate it with the device, as I would expect as the module does not provide an alias for the above acpi/pnp id.

Can you report this issue upstream? Gues to reach out are according to get_maintainers.pl script:

Lorenzo Bianconi <[email protected]> (maintainer:ST LSM6DSx IMU IIO DRIVER) Jonathan Cameron <[email protected]> (maintainer:IIO SUBSYSTEM AND DRIVERS) Lars-Peter Clausen <[email protected]> (reviewer:IIO SUBSYSTEM AND DRIVERS) [email protected] (open list:ST LSM6DSx IMU IIO DRIVER) [email protected] (open list)

Please keep us in the loop.

Regards,
Salvatore

#6 lore raw
Quoted 24 lines · On Mon, 30 Jan 2023 03:37:23 +0000 Darrell Kavanagh…

Forwarding because original html messages were rejected by the server...

---------- Forwarded message ---------
From: Darrell Kavanagh <[email protected]>
Date: Mon, 30 Jan 2023 at 02:52
Subject: Re: Bug#1029850: linux: Driver not loaded for ST Microelectronics LSM6DS3TR-C accelerometer (acpi:SMO8B30:SMO8B30:)
To: Jonathan Cameron <[email protected]>
Cc: <[email protected]>, <[email protected]>, <[email protected]>, <[email protected]>, <[email protected]>

Hi Jonathan,

Thank you. The driver has evolved quite a bit in 6.2 (I read somewhere that 6.2 includes some i2c enhancements), but I adapted your changes to fit my Debian 6.1 kernel and it works. Two IIO devices are created in sysfs, iio-sensor-proxy.service starts up and automatic screen rotation in Gnome just works.

To get the modules to load on boot, I made a small change to your code in st_lsm6dsx_i2c to add the acpi alias to modules.alias:

Quoted 7 lines

adding a null element to st_lsm6dsx_i2c_acpi_match:

    static const struct acpi_device_id st_lsm6dsx_i2c_acpi_match[] = {
         { "SMO8B30", ST_LSM6DS3TRC_ID, },
         { },
    };
then:
   MODULE_DEVICE_TABLE(acpi, st_lsm6dsx_i2c_acpi_match);

doh! That was indeed sloppy of me to miss even for an untested hack.

Quoted 4 lines

dmesg shows:

Quoted 8 lines
[ 7366.120208] st_lsm6dsx_i2c i2c-SMO8B30:00: supply vdd not found,
using dummy regulator
[ 7366.120260] st_lsm6dsx_i2c i2c-SMO8B30:00: supply vddio not found,
using dummy regulator
[ 7366.650839] st_lsm6dsx_i2c i2c-SMO8B30:00: mounting matrix not

found: using identity...

Is this a problem?

Those are all fine. For regulators that's expected on ACPI and should be harmless as it's up to the firmware to manage power (in DT it may be up to the kernel).
For the mounting matrix, there is often something in ACPI DSDT (non standard though). Could you
cat /sys/firmware/acpi/tables/DSDT > ~/dsdt
then run through iasl from acpitools
iasl -d ~/dsdt
and find the bit related to this device.

If you can then share that there may be a _DSM or similar in there that is effectively the mounting matrix. If we are lucky it will look like some existing versions we have code to handle and can add that support as well.

Either way - I'll spin a formal patch with your fixes above and we can get this upstream for future kernels. Mounting matrix can follow later if needed.

Thanks,

Jonathan

Quoted 165 lines

Thanks again.

Darrell

Quoted 158 lines · Jonathan Cameron
Quoted 5 lines · On Sun, 29 Jan 2023 17:03:51 +0000 Darrell Kavanagh…

Hi,

I raised this bug in Debian, and have been asked to raise it upstream and was given your addresses to do so. Will this email be OK, or should I raise it in a bug tracking system somewhere?

Email is the right option.

Quoted 27 lines

Many thanks,
Darrell

---------- Forwarded message ---------
From: Salvatore Bonaccorso <[email protected]>
Date: Sat, 28 Jan 2023 at 20:33
Subject: Re: Bug#1029850: linux: Driver not loaded for ST Microelectronics LSM6DS3TR-C accelerometer (acpi:SMO8B30:SMO8B30:)
To: Darrell Kavanagh <[email protected]>, <[email protected]>

Hi Darrell,

Quoted 9 lines · Darrell Kavanagh

Package: src:linux
Version: 6.1.4-1
Severity: normal
File: linux
X-Debbugs-Cc: [email protected]

Dear Maintainer,

This is a convertable touchscreen tablet/laptop. The rotation sensor

Quoted 9 lines

device

Quoted 1 line

ST Microelectronics LSM6DS3TR-C does not work. It is detected via ACPI

and the

Quoted 1 line

sysfs trees are created at

devices/pci0000:00/0000:00:17.1/i2c_designware.3/i2c-4/i2c-SMO8B30:00

Quoted 2 lines

and devices/LNXSYSTM:00/LNXSYBUS:00/PNP0A08:00/device:3c/SMO8B30:00 with symlinks bus/acpi/devices/SMO8B30:00 and bus/i2c/devices/i2c-SMO8B30:00,

but

Quoted 1 line

no driver is loaded.

At least this is using the ST PNP ID which is better than average (long story!)

The driver in question (ultimately drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c does not currently have an ACPI support. It should be straight forwards to add though the driver first needs converting to use device_get_match_data() with appropriate fallback so that it will match on ACPI, OF or original spi_device_id tables

Completely untested but something like the following
(the offset in the enum is needed to allow us to tell if we got a result when calling device_get_match_data() as it returns NULL on failure IIRC)

I'm not sure how sucessful the driver will be at finding any interrupts etc, but it may get you basic functionality.

Good luck and others more familiar with the driver may well tell me what I forgot when hacking the below ;)

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx.h +1 -1
index 499fcf8875b4..2617ce236ddc 100644
@@ -39,7 +39,7 @@
 #define ST_ISM330IS_DEV_NAME   "ism330is"
 
 enum st_lsm6dsx_hw_id {
-       ST_LSM6DS3_ID,
+       ST_LSM6DS3_ID = 1,
        ST_LSM6DS3H_ID,
        ST_LSM6DSL_ID,
        ST_LSM6DSM_ID,
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c +12 -2
index df5f60925260..ecfceb2fb3db 100644
@@ -23,10 +23,15 @@ static const struct regmap_config st_lsm6dsx_i2c_regmap_config = {
 
 static int st_lsm6dsx_i2c_probe(struct i2c_client *client)
 {
-       const struct i2c_device_id *id = i2c_client_get_device_id(client);
-       int hw_id = id->driver_data;
+       int hw_id;
        struct regmap *regmap;
 
+       hw_id = (kernel_ulong_t)device_get_match_data(&client->dev);
+       if (!hw_id)
+               hw_id = i2c_client_get_device_id(client)->driver_data;
+       if (!hw_id)
+               return -EINVAL;
+
        regmap = devm_regmap_init_i2c(client, &st_lsm6dsx_i2c_regmap_config);
        if (IS_ERR(regmap)) {
                dev_err(&client->dev, "Failed to register i2c regmap %ld\n", PTR_ERR(regmap));
@@ -129,6 +134,10 @@ static const struct of_device_id st_lsm6dsx_i2c_of_match[] = {
 };
 MODULE_DEVICE_TABLE(of, st_lsm6dsx_i2c_of_match);
 
+static const struct acpi_device_id st_lsm6dsx_i2c_acpi_match[] = {
+       { "SMO8B30", ST_LSM6DS3TRC_ID, },
+};
+
 static const struct i2c_device_id st_lsm6dsx_i2c_id_table[] = {
        { ST_LSM6DS3_DEV_NAME, ST_LSM6DS3_ID },
        { ST_LSM6DS3H_DEV_NAME, ST_LSM6DS3H_ID },
@@ -161,6 +170,7 @@ static struct i2c_driver st_lsm6dsx_driver = {
                .name = "st_lsm6dsx_i2c",
                .pm = pm_sleep_ptr(&st_lsm6dsx_pm_ops),
                .of_match_table = st_lsm6dsx_i2c_of_match,
+               .acpi_match_table = st_lsm6dsx_i2c_acpi_match,
        },
        .probe_new = st_lsm6dsx_i2c_probe,
        .id_table = st_lsm6dsx_i2c_id_table,
Quoted 35 lines
Quoted 17 lines

The device is identifying itself to the kernel with PNP id SMO8B30:
physical_node:

      modalias=acpi:SMO8B30:SMO8B30:
      name=SMO8B30:00
      uevent=MODALIAS=acpi:SMO8B30:SMO8B30:
      waiting_for_supplier=0
firmware_node:
      hid=SMO8B30
      modalias=acpi:SMO8B30:SMO8B30:
      path=\_SB_.PCI0.I2C5.DEV_
      status=15
      uevent=MODALIAS=acpi:SMO8B30:SMO8B30:
      uid=0
Quoted 1 line

The kernel module for the appropriate driver (st_lsm6dsx_i2c) is not

loaded on boot.

Quoted 2 lines

Modprobing it does not associate it with the device, as I would expect as the module does not provide an alias for the above acpi/pnp id.

Can you report this issue upstream? Gues to reach out are according to get_maintainers.pl script:

Lorenzo Bianconi <[email protected]> (maintainer:ST LSM6DSx IMU IIO DRIVER) Jonathan Cameron <[email protected]> (maintainer:IIO SUBSYSTEM AND DRIVERS) Lars-Peter Clausen <[email protected]> (reviewer:IIO SUBSYSTEM AND DRIVERS) [email protected] (open list:ST LSM6DSx IMU IIO DRIVER) [email protected] (open list)

Please keep us in the loop.

Regards,
Salvatore

#7 lore raw

Here is what I hope is the relevant section from the DSDT:

    Scope (_SB.PCI0.I2C5)
    {
        Device (DEV)
        {
            Name (_HID, EisaId ("SMO8B30"))  // _HID: Hardware ID
            Name (_CID, EisaId ("SMO8B30"))  // _CID: Compatible ID
            Name (_UID, Zero)  // _UID: Unique ID
            Method (_CRS, 0, NotSerialized)  // _CRS: Current Resource Settings
            {
                Name (RBUF, ResourceTemplate ()
                {
                    I2cSerialBusV2 (0x006A, ControllerInitiated, 0x00061A80,
                        AddressingMode7Bit, "\\_SB.PCI0.I2C5",
                        0x00, ResourceConsumer, , Exclusive,
                        )
                })
                Return (RBUF) /* \_SB_.PCI0.I2C5.DEV_._CRS.RBUF */
            }

            Method (ROTM, 0, NotSerialized)
            {
                Name (RBUF, Package (0x03)
                {
                    "0 -1 0",
                    "1 0 0",
                    "0 0 1"
                })
                Return (RBUF) /* \_SB_.PCI0.I2C5.DEV_.ROTM.RBUF */
            }

            Method (PRIM, 0, NotSerialized)
            {
                Name (RBUF, Buffer (One)
                {
                     0x01                                             // .
                })
                Return (RBUF) /* \_SB_.PCI0.I2C5.DEV_.PRIM.RBUF */
            }

            Method (_STA, 0, NotSerialized)  // _STA: Status
            {
                If ((GAVT == 0x6A))
                {
                    Return (0x0F)
                }
                Else
                {
                    Return (Zero)
                }
            }

            Method (CALS, 1, NotSerialized)
            {
                Local0 = Arg0
                If (((Local0 == Zero) || (Local0 == Ones)))
                {
                    Local0 = BAC1 /* \BAC1 */
                    Return (Local0)
                }
                Else
                {
                    BAC1 = Local0
                    BACS = Local0
                    BSCA (0xB0)
                }
            }
        }
    }

Thanks,
Darrell

Quoted 243 lines · Jonathan Cameron
Quoted 24 lines · On Mon, 30 Jan 2023 03:37:23 +0000 Darrell Kavanagh…

Forwarding because original html messages were rejected by the server...

---------- Forwarded message ---------
From: Darrell Kavanagh <[email protected]>
Date: Mon, 30 Jan 2023 at 02:52
Subject: Re: Bug#1029850: linux: Driver not loaded for ST Microelectronics LSM6DS3TR-C accelerometer (acpi:SMO8B30:SMO8B30:)
To: Jonathan Cameron <[email protected]>
Cc: <[email protected]>, <[email protected]>, <[email protected]>, <[email protected]>, <[email protected]>

Hi Jonathan,

Thank you. The driver has evolved quite a bit in 6.2 (I read somewhere that 6.2 includes some i2c enhancements), but I adapted your changes to fit my Debian 6.1 kernel and it works. Two IIO devices are created in sysfs, iio-sensor-proxy.service starts up and automatic screen rotation in Gnome just works.

To get the modules to load on boot, I made a small change to your code in st_lsm6dsx_i2c to add the acpi alias to modules.alias:

Quoted 7 lines

adding a null element to st_lsm6dsx_i2c_acpi_match:

    static const struct acpi_device_id st_lsm6dsx_i2c_acpi_match[] = {
         { "SMO8B30", ST_LSM6DS3TRC_ID, },
         { },
    };
then:
   MODULE_DEVICE_TABLE(acpi, st_lsm6dsx_i2c_acpi_match);

doh! That was indeed sloppy of me to miss even for an untested hack.

Quoted 4 lines

dmesg shows:

Quoted 8 lines
[ 7366.120208] st_lsm6dsx_i2c i2c-SMO8B30:00: supply vdd not found,
using dummy regulator
[ 7366.120260] st_lsm6dsx_i2c i2c-SMO8B30:00: supply vddio not found,
using dummy regulator
[ 7366.650839] st_lsm6dsx_i2c i2c-SMO8B30:00: mounting matrix not

found: using identity...

Is this a problem?

Those are all fine. For regulators that's expected on ACPI and should be harmless as it's up to the firmware to manage power (in DT it may be up to the kernel).
For the mounting matrix, there is often something in ACPI DSDT (non standard though). Could you
cat /sys/firmware/acpi/tables/DSDT > ~/dsdt
then run through iasl from acpitools
iasl -d ~/dsdt
and find the bit related to this device.

If you can then share that there may be a _DSM or similar in there that is effectively the mounting matrix. If we are lucky it will look like some existing versions we have code to handle and can add that support as well.

Either way - I'll spin a formal patch with your fixes above and we can get this upstream for future kernels. Mounting matrix can follow later if needed.

Thanks,

Jonathan

Quoted 165 lines

Thanks again.

Darrell

Quoted 158 lines · Jonathan Cameron
Quoted 5 lines · On Sun, 29 Jan 2023 17:03:51 +0000 Darrell Kavanagh…

Hi,

I raised this bug in Debian, and have been asked to raise it upstream and was given your addresses to do so. Will this email be OK, or should I raise it in a bug tracking system somewhere?

Email is the right option.

Quoted 27 lines

Many thanks,
Darrell

---------- Forwarded message ---------
From: Salvatore Bonaccorso <[email protected]>
Date: Sat, 28 Jan 2023 at 20:33
Subject: Re: Bug#1029850: linux: Driver not loaded for ST Microelectronics LSM6DS3TR-C accelerometer (acpi:SMO8B30:SMO8B30:)
To: Darrell Kavanagh <[email protected]>, <[email protected]>

Hi Darrell,

Quoted 9 lines · Darrell Kavanagh

Package: src:linux
Version: 6.1.4-1
Severity: normal
File: linux
X-Debbugs-Cc: [email protected]

Dear Maintainer,

This is a convertable touchscreen tablet/laptop. The rotation sensor

Quoted 9 lines

device

Quoted 1 line

ST Microelectronics LSM6DS3TR-C does not work. It is detected via ACPI

and the

Quoted 1 line

sysfs trees are created at

devices/pci0000:00/0000:00:17.1/i2c_designware.3/i2c-4/i2c-SMO8B30:00

Quoted 2 lines

and devices/LNXSYSTM:00/LNXSYBUS:00/PNP0A08:00/device:3c/SMO8B30:00 with symlinks bus/acpi/devices/SMO8B30:00 and bus/i2c/devices/i2c-SMO8B30:00,

but

Quoted 1 line

no driver is loaded.

At least this is using the ST PNP ID which is better than average (long story!)

The driver in question (ultimately drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c does not currently have an ACPI support. It should be straight forwards to add though the driver first needs converting to use device_get_match_data() with appropriate fallback so that it will match on ACPI, OF or original spi_device_id tables

Completely untested but something like the following
(the offset in the enum is needed to allow us to tell if we got a result when calling device_get_match_data() as it returns NULL on failure IIRC)

I'm not sure how sucessful the driver will be at finding any interrupts etc, but it may get you basic functionality.

Good luck and others more familiar with the driver may well tell me what I forgot when hacking the below ;)

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx.h +1 -1
index 499fcf8875b4..2617ce236ddc 100644
@@ -39,7 +39,7 @@
 #define ST_ISM330IS_DEV_NAME   "ism330is"
 
 enum st_lsm6dsx_hw_id {
-       ST_LSM6DS3_ID,
+       ST_LSM6DS3_ID = 1,
        ST_LSM6DS3H_ID,
        ST_LSM6DSL_ID,
        ST_LSM6DSM_ID,
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c +12 -2
index df5f60925260..ecfceb2fb3db 100644
@@ -23,10 +23,15 @@ static const struct regmap_config st_lsm6dsx_i2c_regmap_config = {
 
 static int st_lsm6dsx_i2c_probe(struct i2c_client *client)
 {
-       const struct i2c_device_id *id = i2c_client_get_device_id(client);
-       int hw_id = id->driver_data;
+       int hw_id;
        struct regmap *regmap;
 
+       hw_id = (kernel_ulong_t)device_get_match_data(&client->dev);
+       if (!hw_id)
+               hw_id = i2c_client_get_device_id(client)->driver_data;
+       if (!hw_id)
+               return -EINVAL;
+
        regmap = devm_regmap_init_i2c(client, &st_lsm6dsx_i2c_regmap_config);
        if (IS_ERR(regmap)) {
                dev_err(&client->dev, "Failed to register i2c regmap %ld\n", PTR_ERR(regmap));
@@ -129,6 +134,10 @@ static const struct of_device_id st_lsm6dsx_i2c_of_match[] = {
 };
 MODULE_DEVICE_TABLE(of, st_lsm6dsx_i2c_of_match);
 
+static const struct acpi_device_id st_lsm6dsx_i2c_acpi_match[] = {
+       { "SMO8B30", ST_LSM6DS3TRC_ID, },
+};
+
 static const struct i2c_device_id st_lsm6dsx_i2c_id_table[] = {
        { ST_LSM6DS3_DEV_NAME, ST_LSM6DS3_ID },
        { ST_LSM6DS3H_DEV_NAME, ST_LSM6DS3H_ID },
@@ -161,6 +170,7 @@ static struct i2c_driver st_lsm6dsx_driver = {
                .name = "st_lsm6dsx_i2c",
                .pm = pm_sleep_ptr(&st_lsm6dsx_pm_ops),
                .of_match_table = st_lsm6dsx_i2c_of_match,
+               .acpi_match_table = st_lsm6dsx_i2c_acpi_match,
        },
        .probe_new = st_lsm6dsx_i2c_probe,
        .id_table = st_lsm6dsx_i2c_id_table,
Quoted 35 lines
Quoted 17 lines

The device is identifying itself to the kernel with PNP id SMO8B30:
physical_node:

      modalias=acpi:SMO8B30:SMO8B30:
      name=SMO8B30:00
      uevent=MODALIAS=acpi:SMO8B30:SMO8B30:
      waiting_for_supplier=0
firmware_node:
      hid=SMO8B30
      modalias=acpi:SMO8B30:SMO8B30:
      path=\_SB_.PCI0.I2C5.DEV_
      status=15
      uevent=MODALIAS=acpi:SMO8B30:SMO8B30:
      uid=0
Quoted 1 line

The kernel module for the appropriate driver (st_lsm6dsx_i2c) is not

loaded on boot.

Quoted 2 lines

Modprobing it does not associate it with the device, as I would expect as the module does not provide an alias for the above acpi/pnp id.

Can you report this issue upstream? Gues to reach out are according to get_maintainers.pl script:

Lorenzo Bianconi <[email protected]> (maintainer:ST LSM6DSx IMU IIO DRIVER) Jonathan Cameron <[email protected]> (maintainer:IIO SUBSYSTEM AND DRIVERS) Lars-Peter Clausen <[email protected]> (reviewer:IIO SUBSYSTEM AND DRIVERS) [email protected] (open list:ST LSM6DSx IMU IIO DRIVER) [email protected] (open list)

Please keep us in the loop.

Regards,
Salvatore

#8 lore raw
Quoted 22 lines · On Mon, 30 Jan 2023 13:08:47 +0000 Darrell Kavanagh…

Here is what I hope is the relevant section from the DSDT:

    Scope (_SB.PCI0.I2C5)
    {
        Device (DEV)
        {
            Name (_HID, EisaId ("SMO8B30"))  // _HID: Hardware ID
            Name (_CID, EisaId ("SMO8B30"))  // _CID: Compatible ID
            Name (_UID, Zero)  // _UID: Unique ID
            Method (_CRS, 0, NotSerialized)  // _CRS: Current Resource Settings
            {
                Name (RBUF, ResourceTemplate ()
                {
                    I2cSerialBusV2 (0x006A, ControllerInitiated, 0x00061A80,
                        AddressingMode7Bit, "\\_SB.PCI0.I2C5",
                        0x00, ResourceConsumer, , Exclusive,
                        )
                })
                Return (RBUF) /* \_SB_.PCI0.I2C5.DEV_._CRS.RBUF */
            }

            Method (ROTM, 0, NotSerialized)
Quoted 8 lines
            {
                Name (RBUF, Package (0x03)
                {
                    "0 -1 0",
                    "1 0 0",
                    "0 0 1"
                })
                Return (RBUF) /* \_SB_.PCI0.I2C5.DEV_.ROTM.RBUF */

That certainly looks like suitable matrix.

If we are lucky it matches the handling in bmc150-accel-core.c for an identically named method. That is going to swap the x and y axis which is I'd have thought would be rather bad if what you have is currently working well.

Quoted 290 lines
            }

            Method (PRIM, 0, NotSerialized)
            {
                Name (RBUF, Buffer (One)
                {
                     0x01                                             // .
                })
                Return (RBUF) /* \_SB_.PCI0.I2C5.DEV_.PRIM.RBUF */
            }

            Method (_STA, 0, NotSerialized)  // _STA: Status
            {
                If ((GAVT == 0x6A))
                {
                    Return (0x0F)
                }
                Else
                {
                    Return (Zero)
                }
            }

            Method (CALS, 1, NotSerialized)
            {
                Local0 = Arg0
                If (((Local0 == Zero) || (Local0 == Ones)))
                {
                    Local0 = BAC1 /* \BAC1 */
                    Return (Local0)
                }
                Else
                {
                    BAC1 = Local0
                    BACS = Local0
                    BSCA (0xB0)
                }
            }
        }
    }

Thanks,
Darrell

Quoted 243 lines · Jonathan Cameron
Quoted 24 lines · On Mon, 30 Jan 2023 03:37:23 +0000 Darrell Kavanagh…

Forwarding because original html messages were rejected by the server...

---------- Forwarded message ---------
From: Darrell Kavanagh <[email protected]>
Date: Mon, 30 Jan 2023 at 02:52
Subject: Re: Bug#1029850: linux: Driver not loaded for ST Microelectronics LSM6DS3TR-C accelerometer (acpi:SMO8B30:SMO8B30:)
To: Jonathan Cameron <[email protected]>
Cc: <[email protected]>, <[email protected]>, <[email protected]>, <[email protected]>, <[email protected]>

Hi Jonathan,

Thank you. The driver has evolved quite a bit in 6.2 (I read somewhere that 6.2 includes some i2c enhancements), but I adapted your changes to fit my Debian 6.1 kernel and it works. Two IIO devices are created in sysfs, iio-sensor-proxy.service starts up and automatic screen rotation in Gnome just works.

To get the modules to load on boot, I made a small change to your code in st_lsm6dsx_i2c to add the acpi alias to modules.alias:

Quoted 7 lines

adding a null element to st_lsm6dsx_i2c_acpi_match:

    static const struct acpi_device_id st_lsm6dsx_i2c_acpi_match[] = {
         { "SMO8B30", ST_LSM6DS3TRC_ID, },
         { },
    };
then:
   MODULE_DEVICE_TABLE(acpi, st_lsm6dsx_i2c_acpi_match);  

doh! That was indeed sloppy of me to miss even for an untested hack.

Quoted 4 lines

dmesg shows:

Quoted 8 lines
[ 7366.120208] st_lsm6dsx_i2c i2c-SMO8B30:00: supply vdd not found,
using dummy regulator
[ 7366.120260] st_lsm6dsx_i2c i2c-SMO8B30:00: supply vddio not found,
using dummy regulator
[ 7366.650839] st_lsm6dsx_i2c i2c-SMO8B30:00: mounting matrix not

found: using identity...

Is this a problem?

Those are all fine. For regulators that's expected on ACPI and should be harmless as it's up to the firmware to manage power (in DT it may be up to the kernel).
For the mounting matrix, there is often something in ACPI DSDT (non standard though). Could you
cat /sys/firmware/acpi/tables/DSDT > ~/dsdt
then run through iasl from acpitools
iasl -d ~/dsdt
and find the bit related to this device.

If you can then share that there may be a _DSM or similar in there that is effectively the mounting matrix. If we are lucky it will look like some existing versions we have code to handle and can add that support as well.

Either way - I'll spin a formal patch with your fixes above and we can get this upstream for future kernels. Mounting matrix can follow later if needed.

Thanks,

Jonathan

Quoted 165 lines

Thanks again.

Darrell

Quoted 158 lines · Jonathan Cameron
Quoted 5 lines · On Sun, 29 Jan 2023 17:03:51 +0000 Darrell Kavanagh…

Hi,

I raised this bug in Debian, and have been asked to raise it upstream and was given your addresses to do so. Will this email be OK, or should I raise it in a bug tracking system somewhere?

Email is the right option.

Quoted 27 lines

Many thanks,
Darrell

---------- Forwarded message ---------
From: Salvatore Bonaccorso <[email protected]>
Date: Sat, 28 Jan 2023 at 20:33
Subject: Re: Bug#1029850: linux: Driver not loaded for ST Microelectronics LSM6DS3TR-C accelerometer (acpi:SMO8B30:SMO8B30:)
To: Darrell Kavanagh <[email protected]>, <[email protected]>

Hi Darrell,

Quoted 9 lines · Darrell Kavanagh

Package: src:linux
Version: 6.1.4-1
Severity: normal
File: linux
X-Debbugs-Cc: [email protected]

Dear Maintainer,

This is a convertable touchscreen tablet/laptop. The rotation sensor

Quoted 9 lines

device

Quoted 1 line

ST Microelectronics LSM6DS3TR-C does not work. It is detected via ACPI

and the

Quoted 1 line

sysfs trees are created at

devices/pci0000:00/0000:00:17.1/i2c_designware.3/i2c-4/i2c-SMO8B30:00

Quoted 2 lines

and devices/LNXSYSTM:00/LNXSYBUS:00/PNP0A08:00/device:3c/SMO8B30:00 with symlinks bus/acpi/devices/SMO8B30:00 and bus/i2c/devices/i2c-SMO8B30:00,

but

Quoted 1 line

no driver is loaded.

At least this is using the ST PNP ID which is better than average (long story!)

The driver in question (ultimately drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c does not currently have an ACPI support. It should be straight forwards to add though the driver first needs converting to use device_get_match_data() with appropriate fallback so that it will match on ACPI, OF or original spi_device_id tables

Completely untested but something like the following
(the offset in the enum is needed to allow us to tell if we got a result when calling device_get_match_data() as it returns NULL on failure IIRC)

I'm not sure how sucessful the driver will be at finding any interrupts etc, but it may get you basic functionality.

Good luck and others more familiar with the driver may well tell me what I forgot when hacking the below ;)

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx.h +1 -1
index 499fcf8875b4..2617ce236ddc 100644
@@ -39,7 +39,7 @@
 #define ST_ISM330IS_DEV_NAME   "ism330is"
 
 enum st_lsm6dsx_hw_id {
-       ST_LSM6DS3_ID,
+       ST_LSM6DS3_ID = 1,
        ST_LSM6DS3H_ID,
        ST_LSM6DSL_ID,
        ST_LSM6DSM_ID,
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c +12 -2
index df5f60925260..ecfceb2fb3db 100644
@@ -23,10 +23,15 @@ static const struct regmap_config st_lsm6dsx_i2c_regmap_config = {
 
 static int st_lsm6dsx_i2c_probe(struct i2c_client *client)
 {
-       const struct i2c_device_id *id = i2c_client_get_device_id(client);
-       int hw_id = id->driver_data;
+       int hw_id;
        struct regmap *regmap;
 
+       hw_id = (kernel_ulong_t)device_get_match_data(&client->dev);
+       if (!hw_id)
+               hw_id = i2c_client_get_device_id(client)->driver_data;
+       if (!hw_id)
+               return -EINVAL;
+
        regmap = devm_regmap_init_i2c(client, &st_lsm6dsx_i2c_regmap_config);
        if (IS_ERR(regmap)) {
                dev_err(&client->dev, "Failed to register i2c regmap %ld\n", PTR_ERR(regmap));
@@ -129,6 +134,10 @@ static const struct of_device_id st_lsm6dsx_i2c_of_match[] = {
 };
 MODULE_DEVICE_TABLE(of, st_lsm6dsx_i2c_of_match);
 
+static const struct acpi_device_id st_lsm6dsx_i2c_acpi_match[] = {
+       { "SMO8B30", ST_LSM6DS3TRC_ID, },
+};
+
 static const struct i2c_device_id st_lsm6dsx_i2c_id_table[] = {
        { ST_LSM6DS3_DEV_NAME, ST_LSM6DS3_ID },
        { ST_LSM6DS3H_DEV_NAME, ST_LSM6DS3H_ID },
@@ -161,6 +170,7 @@ static struct i2c_driver st_lsm6dsx_driver = {
                .name = "st_lsm6dsx_i2c",
                .pm = pm_sleep_ptr(&st_lsm6dsx_pm_ops),
                .of_match_table = st_lsm6dsx_i2c_of_match,
+               .acpi_match_table = st_lsm6dsx_i2c_acpi_match,
        },
        .probe_new = st_lsm6dsx_i2c_probe,
        .id_table = st_lsm6dsx_i2c_id_table,
Quoted 35 lines
Quoted 17 lines

The device is identifying itself to the kernel with PNP id SMO8B30:
physical_node:

      modalias=acpi:SMO8B30:SMO8B30:
      name=SMO8B30:00
      uevent=MODALIAS=acpi:SMO8B30:SMO8B30:
      waiting_for_supplier=0
firmware_node:
      hid=SMO8B30
      modalias=acpi:SMO8B30:SMO8B30:
      path=\_SB_.PCI0.I2C5.DEV_
      status=15
      uevent=MODALIAS=acpi:SMO8B30:SMO8B30:
      uid=0
Quoted 1 line

The kernel module for the appropriate driver (st_lsm6dsx_i2c) is not

loaded on boot.

Quoted 2 lines

Modprobing it does not associate it with the device, as I would expect as the module does not provide an alias for the above acpi/pnp id.

Can you report this issue upstream? Gues to reach out are according to get_maintainers.pl script:

Lorenzo Bianconi <[email protected]> (maintainer:ST LSM6DSx IMU IIO DRIVER) Jonathan Cameron <[email protected]> (maintainer:IIO SUBSYSTEM AND DRIVERS) Lars-Peter Clausen <[email protected]> (reviewer:IIO SUBSYSTEM AND DRIVERS) [email protected] (open list:ST LSM6DSx IMU IIO DRIVER) [email protected] (open list)

Please keep us in the loop.

Regards,
Salvatore

#9 lore raw
Quoted 6 lines · Jonathan Cameron

That certainly looks like suitable matrix.

If we are lucky it matches the handling in bmc150-accel-core.c for an identically named method. That is going to swap the x and y axis which is I'd have thought would be rather bad if what you have is currently working well.

Actually, that would be good, because at present I have to rotate 90 degrees in my grub command line.

Darrell

Quoted 291 lines
Quoted 290 lines
            }

            Method (PRIM, 0, NotSerialized)
            {
                Name (RBUF, Buffer (One)
                {
                     0x01                                             // .
                })
                Return (RBUF) /* \_SB_.PCI0.I2C5.DEV_.PRIM.RBUF */
            }

            Method (_STA, 0, NotSerialized)  // _STA: Status
            {
                If ((GAVT == 0x6A))
                {
                    Return (0x0F)
                }
                Else
                {
                    Return (Zero)
                }
            }

            Method (CALS, 1, NotSerialized)
            {
                Local0 = Arg0
                If (((Local0 == Zero) || (Local0 == Ones)))
                {
                    Local0 = BAC1 /* \BAC1 */
                    Return (Local0)
                }
                Else
                {
                    BAC1 = Local0
                    BACS = Local0
                    BSCA (0xB0)
                }
            }
        }
    }

Thanks,
Darrell

Quoted 243 lines · Jonathan Cameron
Quoted 24 lines · On Mon, 30 Jan 2023 03:37:23 +0000 Darrell Kavanagh…

Forwarding because original html messages were rejected by the server...

---------- Forwarded message ---------
From: Darrell Kavanagh <[email protected]>
Date: Mon, 30 Jan 2023 at 02:52
Subject: Re: Bug#1029850: linux: Driver not loaded for ST Microelectronics LSM6DS3TR-C accelerometer (acpi:SMO8B30:SMO8B30:)
To: Jonathan Cameron <[email protected]>
Cc: <[email protected]>, <[email protected]>, <[email protected]>, <[email protected]>, <[email protected]>

Hi Jonathan,

Thank you. The driver has evolved quite a bit in 6.2 (I read somewhere that 6.2 includes some i2c enhancements), but I adapted your changes to fit my Debian 6.1 kernel and it works. Two IIO devices are created in sysfs, iio-sensor-proxy.service starts up and automatic screen rotation in Gnome just works.

To get the modules to load on boot, I made a small change to your code in st_lsm6dsx_i2c to add the acpi alias to modules.alias:

Quoted 7 lines

adding a null element to st_lsm6dsx_i2c_acpi_match:

    static const struct acpi_device_id st_lsm6dsx_i2c_acpi_match[] = {
         { "SMO8B30", ST_LSM6DS3TRC_ID, },
         { },
    };
then:
   MODULE_DEVICE_TABLE(acpi, st_lsm6dsx_i2c_acpi_match);

doh! That was indeed sloppy of me to miss even for an untested hack.

Quoted 4 lines

dmesg shows:

Quoted 8 lines
[ 7366.120208] st_lsm6dsx_i2c i2c-SMO8B30:00: supply vdd not found,
using dummy regulator
[ 7366.120260] st_lsm6dsx_i2c i2c-SMO8B30:00: supply vddio not found,
using dummy regulator
[ 7366.650839] st_lsm6dsx_i2c i2c-SMO8B30:00: mounting matrix not

found: using identity...

Is this a problem?

Those are all fine. For regulators that's expected on ACPI and should be harmless as it's up to the firmware to manage power (in DT it may be up to the kernel).
For the mounting matrix, there is often something in ACPI DSDT (non standard though). Could you
cat /sys/firmware/acpi/tables/DSDT > ~/dsdt
then run through iasl from acpitools
iasl -d ~/dsdt
and find the bit related to this device.

If you can then share that there may be a _DSM or similar in there that is effectively the mounting matrix. If we are lucky it will look like some existing versions we have code to handle and can add that support as well.

Either way - I'll spin a formal patch with your fixes above and we can get this upstream for future kernels. Mounting matrix can follow later if needed.

Thanks,

Jonathan

Quoted 165 lines

Thanks again.

Darrell

Quoted 158 lines · Jonathan Cameron
Quoted 5 lines · On Sun, 29 Jan 2023 17:03:51 +0000 Darrell Kavanagh…

Hi,

I raised this bug in Debian, and have been asked to raise it upstream and was given your addresses to do so. Will this email be OK, or should I raise it in a bug tracking system somewhere?

Email is the right option.

Quoted 27 lines

Many thanks,
Darrell

---------- Forwarded message ---------
From: Salvatore Bonaccorso <[email protected]>
Date: Sat, 28 Jan 2023 at 20:33
Subject: Re: Bug#1029850: linux: Driver not loaded for ST Microelectronics LSM6DS3TR-C accelerometer (acpi:SMO8B30:SMO8B30:)
To: Darrell Kavanagh <[email protected]>, <[email protected]>

Hi Darrell,

Quoted 9 lines · Darrell Kavanagh

Package: src:linux
Version: 6.1.4-1
Severity: normal
File: linux
X-Debbugs-Cc: [email protected]

Dear Maintainer,

This is a convertable touchscreen tablet/laptop. The rotation sensor

Quoted 9 lines

device

Quoted 1 line

ST Microelectronics LSM6DS3TR-C does not work. It is detected via ACPI

and the

Quoted 1 line

sysfs trees are created at

devices/pci0000:00/0000:00:17.1/i2c_designware.3/i2c-4/i2c-SMO8B30:00

Quoted 2 lines

and devices/LNXSYSTM:00/LNXSYBUS:00/PNP0A08:00/device:3c/SMO8B30:00 with symlinks bus/acpi/devices/SMO8B30:00 and bus/i2c/devices/i2c-SMO8B30:00,

but

Quoted 1 line

no driver is loaded.

At least this is using the ST PNP ID which is better than average (long story!)

The driver in question (ultimately drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c does not currently have an ACPI support. It should be straight forwards to add though the driver first needs converting to use device_get_match_data() with appropriate fallback so that it will match on ACPI, OF or original spi_device_id tables

Completely untested but something like the following
(the offset in the enum is needed to allow us to tell if we got a result when calling device_get_match_data() as it returns NULL on failure IIRC)

I'm not sure how sucessful the driver will be at finding any interrupts etc, but it may get you basic functionality.

Good luck and others more familiar with the driver may well tell me what I forgot when hacking the below ;)

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx.h +1 -1
index 499fcf8875b4..2617ce236ddc 100644
@@ -39,7 +39,7 @@
 #define ST_ISM330IS_DEV_NAME   "ism330is"
 
 enum st_lsm6dsx_hw_id {
-       ST_LSM6DS3_ID,
+       ST_LSM6DS3_ID = 1,
        ST_LSM6DS3H_ID,
        ST_LSM6DSL_ID,
        ST_LSM6DSM_ID,
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c +12 -2
index df5f60925260..ecfceb2fb3db 100644
@@ -23,10 +23,15 @@ static const struct regmap_config st_lsm6dsx_i2c_regmap_config = {
 
 static int st_lsm6dsx_i2c_probe(struct i2c_client *client)
 {
-       const struct i2c_device_id *id = i2c_client_get_device_id(client);
-       int hw_id = id->driver_data;
+       int hw_id;
        struct regmap *regmap;
 
+       hw_id = (kernel_ulong_t)device_get_match_data(&client->dev);
+       if (!hw_id)
+               hw_id = i2c_client_get_device_id(client)->driver_data;
+       if (!hw_id)
+               return -EINVAL;
+
        regmap = devm_regmap_init_i2c(client, &st_lsm6dsx_i2c_regmap_config);
        if (IS_ERR(regmap)) {
                dev_err(&client->dev, "Failed to register i2c regmap %ld\n", PTR_ERR(regmap));
@@ -129,6 +134,10 @@ static const struct of_device_id st_lsm6dsx_i2c_of_match[] = {
 };
 MODULE_DEVICE_TABLE(of, st_lsm6dsx_i2c_of_match);
 
+static const struct acpi_device_id st_lsm6dsx_i2c_acpi_match[] = {
+       { "SMO8B30", ST_LSM6DS3TRC_ID, },
+};
+
 static const struct i2c_device_id st_lsm6dsx_i2c_id_table[] = {
        { ST_LSM6DS3_DEV_NAME, ST_LSM6DS3_ID },
        { ST_LSM6DS3H_DEV_NAME, ST_LSM6DS3H_ID },
@@ -161,6 +170,7 @@ static struct i2c_driver st_lsm6dsx_driver = {
                .name = "st_lsm6dsx_i2c",
                .pm = pm_sleep_ptr(&st_lsm6dsx_pm_ops),
                .of_match_table = st_lsm6dsx_i2c_of_match,
+               .acpi_match_table = st_lsm6dsx_i2c_acpi_match,
        },
        .probe_new = st_lsm6dsx_i2c_probe,
        .id_table = st_lsm6dsx_i2c_id_table,
Quoted 35 lines
Quoted 17 lines

The device is identifying itself to the kernel with PNP id SMO8B30:
physical_node:

      modalias=acpi:SMO8B30:SMO8B30:
      name=SMO8B30:00
      uevent=MODALIAS=acpi:SMO8B30:SMO8B30:
      waiting_for_supplier=0
firmware_node:
      hid=SMO8B30
      modalias=acpi:SMO8B30:SMO8B30:
      path=\_SB_.PCI0.I2C5.DEV_
      status=15
      uevent=MODALIAS=acpi:SMO8B30:SMO8B30:
      uid=0
Quoted 1 line

The kernel module for the appropriate driver (st_lsm6dsx_i2c) is not

loaded on boot.

Quoted 2 lines

Modprobing it does not associate it with the device, as I would expect as the module does not provide an alias for the above acpi/pnp id.

Can you report this issue upstream? Gues to reach out are according to get_maintainers.pl script:

Lorenzo Bianconi <[email protected]> (maintainer:ST LSM6DSx IMU IIO DRIVER) Jonathan Cameron <[email protected]> (maintainer:IIO SUBSYSTEM AND DRIVERS) Lars-Peter Clausen <[email protected]> (reviewer:IIO SUBSYSTEM AND DRIVERS) [email protected] (open list:ST LSM6DSx IMU IIO DRIVER) [email protected] (open list)

Please keep us in the loop.

Regards,
Salvatore

author maintainer
#10 lore raw
Quoted 5 lines · On Mon, 30 Jan 2023 18:32:02 +0000 Darrell Kavanagh…
Quoted 2 lines · Jonathan Cameron

That certainly looks like suitable matrix.

Quoted 7 lines
Quoted 4 lines

If we are lucky it matches the handling in bmc150-accel-core.c for an identically named method. That is going to swap the x and y axis which is I'd have thought would be rather bad if what you have is currently working well.

Actually, that would be good, because at present I have to rotate 90 degrees in my grub command line.

:)

I'll see if I can roll a suitable patch.

Jonathan

Quoted 294 lines

Darrell

Quoted 291 lines
Quoted 290 lines
            }

            Method (PRIM, 0, NotSerialized)
            {
                Name (RBUF, Buffer (One)
                {
                     0x01                                             // .
                })
                Return (RBUF) /* \_SB_.PCI0.I2C5.DEV_.PRIM.RBUF */
            }

            Method (_STA, 0, NotSerialized)  // _STA: Status
            {
                If ((GAVT == 0x6A))
                {
                    Return (0x0F)
                }
                Else
                {
                    Return (Zero)
                }
            }

            Method (CALS, 1, NotSerialized)
            {
                Local0 = Arg0
                If (((Local0 == Zero) || (Local0 == Ones)))
                {
                    Local0 = BAC1 /* \BAC1 */
                    Return (Local0)
                }
                Else
                {
                    BAC1 = Local0
                    BACS = Local0
                    BSCA (0xB0)
                }
            }
        }
    }

Thanks,
Darrell

Quoted 243 lines · Jonathan Cameron
Quoted 24 lines · On Mon, 30 Jan 2023 03:37:23 +0000 Darrell Kavanagh…

Forwarding because original html messages were rejected by the server...

---------- Forwarded message ---------
From: Darrell Kavanagh <[email protected]>
Date: Mon, 30 Jan 2023 at 02:52
Subject: Re: Bug#1029850: linux: Driver not loaded for ST Microelectronics LSM6DS3TR-C accelerometer (acpi:SMO8B30:SMO8B30:)
To: Jonathan Cameron <[email protected]>
Cc: <[email protected]>, <[email protected]>, <[email protected]>, <[email protected]>, <[email protected]>

Hi Jonathan,

Thank you. The driver has evolved quite a bit in 6.2 (I read somewhere that 6.2 includes some i2c enhancements), but I adapted your changes to fit my Debian 6.1 kernel and it works. Two IIO devices are created in sysfs, iio-sensor-proxy.service starts up and automatic screen rotation in Gnome just works.

To get the modules to load on boot, I made a small change to your code in st_lsm6dsx_i2c to add the acpi alias to modules.alias:

Quoted 7 lines

adding a null element to st_lsm6dsx_i2c_acpi_match:

    static const struct acpi_device_id st_lsm6dsx_i2c_acpi_match[] = {
         { "SMO8B30", ST_LSM6DS3TRC_ID, },
         { },
    };
then:
   MODULE_DEVICE_TABLE(acpi, st_lsm6dsx_i2c_acpi_match);  

doh! That was indeed sloppy of me to miss even for an untested hack.

Quoted 4 lines

dmesg shows:

Quoted 8 lines
[ 7366.120208] st_lsm6dsx_i2c i2c-SMO8B30:00: supply vdd not found,
using dummy regulator
[ 7366.120260] st_lsm6dsx_i2c i2c-SMO8B30:00: supply vddio not found,
using dummy regulator
[ 7366.650839] st_lsm6dsx_i2c i2c-SMO8B30:00: mounting matrix not

found: using identity...

Is this a problem?

Those are all fine. For regulators that's expected on ACPI and should be harmless as it's up to the firmware to manage power (in DT it may be up to the kernel).
For the mounting matrix, there is often something in ACPI DSDT (non standard though). Could you
cat /sys/firmware/acpi/tables/DSDT > ~/dsdt
then run through iasl from acpitools
iasl -d ~/dsdt
and find the bit related to this device.

If you can then share that there may be a _DSM or similar in there that is effectively the mounting matrix. If we are lucky it will look like some existing versions we have code to handle and can add that support as well.

Either way - I'll spin a formal patch with your fixes above and we can get this upstream for future kernels. Mounting matrix can follow later if needed.

Thanks,

Jonathan

Quoted 165 lines

Thanks again.

Darrell

Quoted 158 lines · Jonathan Cameron
Quoted 5 lines · On Sun, 29 Jan 2023 17:03:51 +0000 Darrell Kavanagh…

Hi,

I raised this bug in Debian, and have been asked to raise it upstream and was given your addresses to do so. Will this email be OK, or should I raise it in a bug tracking system somewhere?

Email is the right option.

Quoted 27 lines

Many thanks,
Darrell

---------- Forwarded message ---------
From: Salvatore Bonaccorso <[email protected]>
Date: Sat, 28 Jan 2023 at 20:33
Subject: Re: Bug#1029850: linux: Driver not loaded for ST Microelectronics LSM6DS3TR-C accelerometer (acpi:SMO8B30:SMO8B30:)
To: Darrell Kavanagh <[email protected]>, <[email protected]>

Hi Darrell,

Quoted 9 lines · Darrell Kavanagh

Package: src:linux
Version: 6.1.4-1
Severity: normal
File: linux
X-Debbugs-Cc: [email protected]

Dear Maintainer,

This is a convertable touchscreen tablet/laptop. The rotation sensor

Quoted 9 lines

device

Quoted 1 line

ST Microelectronics LSM6DS3TR-C does not work. It is detected via ACPI

and the

Quoted 1 line

sysfs trees are created at

devices/pci0000:00/0000:00:17.1/i2c_designware.3/i2c-4/i2c-SMO8B30:00

Quoted 2 lines

and devices/LNXSYSTM:00/LNXSYBUS:00/PNP0A08:00/device:3c/SMO8B30:00 with symlinks bus/acpi/devices/SMO8B30:00 and bus/i2c/devices/i2c-SMO8B30:00,

but

Quoted 1 line

no driver is loaded.

At least this is using the ST PNP ID which is better than average (long story!)

The driver in question (ultimately drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c does not currently have an ACPI support. It should be straight forwards to add though the driver first needs converting to use device_get_match_data() with appropriate fallback so that it will match on ACPI, OF or original spi_device_id tables

Completely untested but something like the following
(the offset in the enum is needed to allow us to tell if we got a result when calling device_get_match_data() as it returns NULL on failure IIRC)

I'm not sure how sucessful the driver will be at finding any interrupts etc, but it may get you basic functionality.

Good luck and others more familiar with the driver may well tell me what I forgot when hacking the below ;)

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx.h +1 -1
index 499fcf8875b4..2617ce236ddc 100644
@@ -39,7 +39,7 @@
 #define ST_ISM330IS_DEV_NAME   "ism330is"
 
 enum st_lsm6dsx_hw_id {
-       ST_LSM6DS3_ID,
+       ST_LSM6DS3_ID = 1,
        ST_LSM6DS3H_ID,
        ST_LSM6DSL_ID,
        ST_LSM6DSM_ID,
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c +12 -2
index df5f60925260..ecfceb2fb3db 100644
@@ -23,10 +23,15 @@ static const struct regmap_config st_lsm6dsx_i2c_regmap_config = {
 
 static int st_lsm6dsx_i2c_probe(struct i2c_client *client)
 {
-       const struct i2c_device_id *id = i2c_client_get_device_id(client);
-       int hw_id = id->driver_data;
+       int hw_id;
        struct regmap *regmap;
 
+       hw_id = (kernel_ulong_t)device_get_match_data(&client->dev);
+       if (!hw_id)
+               hw_id = i2c_client_get_device_id(client)->driver_data;
+       if (!hw_id)
+               return -EINVAL;
+
        regmap = devm_regmap_init_i2c(client, &st_lsm6dsx_i2c_regmap_config);
        if (IS_ERR(regmap)) {
                dev_err(&client->dev, "Failed to register i2c regmap %ld\n", PTR_ERR(regmap));
@@ -129,6 +134,10 @@ static const struct of_device_id st_lsm6dsx_i2c_of_match[] = {
 };
 MODULE_DEVICE_TABLE(of, st_lsm6dsx_i2c_of_match);
 
+static const struct acpi_device_id st_lsm6dsx_i2c_acpi_match[] = {
+       { "SMO8B30", ST_LSM6DS3TRC_ID, },
+};
+
 static const struct i2c_device_id st_lsm6dsx_i2c_id_table[] = {
        { ST_LSM6DS3_DEV_NAME, ST_LSM6DS3_ID },
        { ST_LSM6DS3H_DEV_NAME, ST_LSM6DS3H_ID },
@@ -161,6 +170,7 @@ static struct i2c_driver st_lsm6dsx_driver = {
                .name = "st_lsm6dsx_i2c",
                .pm = pm_sleep_ptr(&st_lsm6dsx_pm_ops),
                .of_match_table = st_lsm6dsx_i2c_of_match,
+               .acpi_match_table = st_lsm6dsx_i2c_acpi_match,
        },
        .probe_new = st_lsm6dsx_i2c_probe,
        .id_table = st_lsm6dsx_i2c_id_table,
Quoted 35 lines
Quoted 17 lines

The device is identifying itself to the kernel with PNP id SMO8B30:
physical_node:

      modalias=acpi:SMO8B30:SMO8B30:
      name=SMO8B30:00
      uevent=MODALIAS=acpi:SMO8B30:SMO8B30:
      waiting_for_supplier=0
firmware_node:
      hid=SMO8B30
      modalias=acpi:SMO8B30:SMO8B30:
      path=\_SB_.PCI0.I2C5.DEV_
      status=15
      uevent=MODALIAS=acpi:SMO8B30:SMO8B30:
      uid=0
Quoted 1 line

The kernel module for the appropriate driver (st_lsm6dsx_i2c) is not

loaded on boot.

Quoted 2 lines

Modprobing it does not associate it with the device, as I would expect as the module does not provide an alias for the above acpi/pnp id.

Can you report this issue upstream? Gues to reach out are according to get_maintainers.pl script:

Lorenzo Bianconi <[email protected]> (maintainer:ST LSM6DSx IMU IIO DRIVER) Jonathan Cameron <[email protected]> (maintainer:IIO SUBSYSTEM AND DRIVERS) Lars-Peter Clausen <[email protected]> (reviewer:IIO SUBSYSTEM AND DRIVERS) [email protected] (open list:ST LSM6DSx IMU IIO DRIVER) [email protected] (open list)

Please keep us in the loop.

Regards,
Salvatore

#11 lore raw

Thanks. To be clear, before I changed the grub command line, the system always booted up "sideways", even when the sensor was not being detected. This was true for everything, not just Gnome, except grub itself.

I added:

GRUB_CMDLINE_LINUX_DEFAULT="fbcon=rotate:1 video=DSI-1:panel_orientation=right_side_up quiet splash"

This fixed the orientation for pre-splash boot messages, the splash screen and the desktop environment. But not, for example (as I saw after adding my own module signing key for testing your fixes), the MOK validation screens.

Does this make sense?

Does what you are proposing act at a lower level than changing the systemd hwdb orientation matrix?

Thanks,
Darrell

Quoted 317 lines · Jonathan Cameron
Quoted 5 lines · On Mon, 30 Jan 2023 18:32:02 +0000 Darrell Kavanagh…
Quoted 2 lines · Jonathan Cameron

That certainly looks like suitable matrix.

Quoted 7 lines
Quoted 4 lines

If we are lucky it matches the handling in bmc150-accel-core.c for an identically named method. That is going to swap the x and y axis which is I'd have thought would be rather bad if what you have is currently working well.

Actually, that would be good, because at present I have to rotate 90 degrees in my grub command line.

:)

I'll see if I can roll a suitable patch.

Jonathan

Quoted 294 lines

Darrell

Quoted 291 lines
Quoted 290 lines
            }

            Method (PRIM, 0, NotSerialized)
            {
                Name (RBUF, Buffer (One)
                {
                     0x01                                             // .
                })
                Return (RBUF) /* \_SB_.PCI0.I2C5.DEV_.PRIM.RBUF */
            }

            Method (_STA, 0, NotSerialized)  // _STA: Status
            {
                If ((GAVT == 0x6A))
                {
                    Return (0x0F)
                }
                Else
                {
                    Return (Zero)
                }
            }

            Method (CALS, 1, NotSerialized)
            {
                Local0 = Arg0
                If (((Local0 == Zero) || (Local0 == Ones)))
                {
                    Local0 = BAC1 /* \BAC1 */
                    Return (Local0)
                }
                Else
                {
                    BAC1 = Local0
                    BACS = Local0
                    BSCA (0xB0)
                }
            }
        }
    }

Thanks,
Darrell

Quoted 243 lines · Jonathan Cameron
Quoted 24 lines · On Mon, 30 Jan 2023 03:37:23 +0000 Darrell Kavanagh…

Forwarding because original html messages were rejected by the server...

---------- Forwarded message ---------
From: Darrell Kavanagh <[email protected]>
Date: Mon, 30 Jan 2023 at 02:52
Subject: Re: Bug#1029850: linux: Driver not loaded for ST Microelectronics LSM6DS3TR-C accelerometer (acpi:SMO8B30:SMO8B30:)
To: Jonathan Cameron <[email protected]>
Cc: <[email protected]>, <[email protected]>, <[email protected]>, <[email protected]>, <[email protected]>

Hi Jonathan,

Thank you. The driver has evolved quite a bit in 6.2 (I read somewhere that 6.2 includes some i2c enhancements), but I adapted your changes to fit my Debian 6.1 kernel and it works. Two IIO devices are created in sysfs, iio-sensor-proxy.service starts up and automatic screen rotation in Gnome just works.

To get the modules to load on boot, I made a small change to your code in st_lsm6dsx_i2c to add the acpi alias to modules.alias:

Quoted 7 lines

adding a null element to st_lsm6dsx_i2c_acpi_match:

    static const struct acpi_device_id st_lsm6dsx_i2c_acpi_match[] = {
         { "SMO8B30", ST_LSM6DS3TRC_ID, },
         { },
    };
then:
   MODULE_DEVICE_TABLE(acpi, st_lsm6dsx_i2c_acpi_match);

doh! That was indeed sloppy of me to miss even for an untested hack.

Quoted 4 lines

dmesg shows:

Quoted 8 lines
[ 7366.120208] st_lsm6dsx_i2c i2c-SMO8B30:00: supply vdd not found,
using dummy regulator
[ 7366.120260] st_lsm6dsx_i2c i2c-SMO8B30:00: supply vddio not found,
using dummy regulator
[ 7366.650839] st_lsm6dsx_i2c i2c-SMO8B30:00: mounting matrix not

found: using identity...

Is this a problem?

Those are all fine. For regulators that's expected on ACPI and should be harmless as it's up to the firmware to manage power (in DT it may be up to the kernel).
For the mounting matrix, there is often something in ACPI DSDT (non standard though). Could you
cat /sys/firmware/acpi/tables/DSDT > ~/dsdt
then run through iasl from acpitools
iasl -d ~/dsdt
and find the bit related to this device.

If you can then share that there may be a _DSM or similar in there that is effectively the mounting matrix. If we are lucky it will look like some existing versions we have code to handle and can add that support as well.

Either way - I'll spin a formal patch with your fixes above and we can get this upstream for future kernels. Mounting matrix can follow later if needed.

Thanks,

Jonathan

Quoted 165 lines

Thanks again.

Darrell

Quoted 158 lines · Jonathan Cameron
Quoted 5 lines · On Sun, 29 Jan 2023 17:03:51 +0000 Darrell Kavanagh…

Hi,

I raised this bug in Debian, and have been asked to raise it upstream and was given your addresses to do so. Will this email be OK, or should I raise it in a bug tracking system somewhere?

Email is the right option.

Quoted 27 lines

Many thanks,
Darrell

---------- Forwarded message ---------
From: Salvatore Bonaccorso <[email protected]>
Date: Sat, 28 Jan 2023 at 20:33
Subject: Re: Bug#1029850: linux: Driver not loaded for ST Microelectronics LSM6DS3TR-C accelerometer (acpi:SMO8B30:SMO8B30:)
To: Darrell Kavanagh <[email protected]>, <[email protected]>

Hi Darrell,

Quoted 9 lines · Darrell Kavanagh

Package: src:linux
Version: 6.1.4-1
Severity: normal
File: linux
X-Debbugs-Cc: [email protected]

Dear Maintainer,

This is a convertable touchscreen tablet/laptop. The rotation sensor

Quoted 9 lines

device

Quoted 1 line

ST Microelectronics LSM6DS3TR-C does not work. It is detected via ACPI

and the

Quoted 1 line

sysfs trees are created at

devices/pci0000:00/0000:00:17.1/i2c_designware.3/i2c-4/i2c-SMO8B30:00

Quoted 2 lines

and devices/LNXSYSTM:00/LNXSYBUS:00/PNP0A08:00/device:3c/SMO8B30:00 with symlinks bus/acpi/devices/SMO8B30:00 and bus/i2c/devices/i2c-SMO8B30:00,

but

Quoted 1 line

no driver is loaded.

At least this is using the ST PNP ID which is better than average (long story!)

The driver in question (ultimately drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c does not currently have an ACPI support. It should be straight forwards to add though the driver first needs converting to use device_get_match_data() with appropriate fallback so that it will match on ACPI, OF or original spi_device_id tables

Completely untested but something like the following
(the offset in the enum is needed to allow us to tell if we got a result when calling device_get_match_data() as it returns NULL on failure IIRC)

I'm not sure how sucessful the driver will be at finding any interrupts etc, but it may get you basic functionality.

Good luck and others more familiar with the driver may well tell me what I forgot when hacking the below ;)

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx.h +1 -1
index 499fcf8875b4..2617ce236ddc 100644
@@ -39,7 +39,7 @@
 #define ST_ISM330IS_DEV_NAME   "ism330is"
 
 enum st_lsm6dsx_hw_id {
-       ST_LSM6DS3_ID,
+       ST_LSM6DS3_ID = 1,
        ST_LSM6DS3H_ID,
        ST_LSM6DSL_ID,
        ST_LSM6DSM_ID,
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c +12 -2
index df5f60925260..ecfceb2fb3db 100644
@@ -23,10 +23,15 @@ static const struct regmap_config st_lsm6dsx_i2c_regmap_config = {
 
 static int st_lsm6dsx_i2c_probe(struct i2c_client *client)
 {
-       const struct i2c_device_id *id = i2c_client_get_device_id(client);
-       int hw_id = id->driver_data;
+       int hw_id;
        struct regmap *regmap;
 
+       hw_id = (kernel_ulong_t)device_get_match_data(&client->dev);
+       if (!hw_id)
+               hw_id = i2c_client_get_device_id(client)->driver_data;
+       if (!hw_id)
+               return -EINVAL;
+
        regmap = devm_regmap_init_i2c(client, &st_lsm6dsx_i2c_regmap_config);
        if (IS_ERR(regmap)) {
                dev_err(&client->dev, "Failed to register i2c regmap %ld\n", PTR_ERR(regmap));
@@ -129,6 +134,10 @@ static const struct of_device_id st_lsm6dsx_i2c_of_match[] = {
 };
 MODULE_DEVICE_TABLE(of, st_lsm6dsx_i2c_of_match);
 
+static const struct acpi_device_id st_lsm6dsx_i2c_acpi_match[] = {
+       { "SMO8B30", ST_LSM6DS3TRC_ID, },
+};
+
 static const struct i2c_device_id st_lsm6dsx_i2c_id_table[] = {
        { ST_LSM6DS3_DEV_NAME, ST_LSM6DS3_ID },
        { ST_LSM6DS3H_DEV_NAME, ST_LSM6DS3H_ID },
@@ -161,6 +170,7 @@ static struct i2c_driver st_lsm6dsx_driver = {
                .name = "st_lsm6dsx_i2c",
                .pm = pm_sleep_ptr(&st_lsm6dsx_pm_ops),
                .of_match_table = st_lsm6dsx_i2c_of_match,
+               .acpi_match_table = st_lsm6dsx_i2c_acpi_match,
        },
        .probe_new = st_lsm6dsx_i2c_probe,
        .id_table = st_lsm6dsx_i2c_id_table,
Quoted 35 lines
Quoted 17 lines

The device is identifying itself to the kernel with PNP id SMO8B30:
physical_node:

      modalias=acpi:SMO8B30:SMO8B30:
      name=SMO8B30:00
      uevent=MODALIAS=acpi:SMO8B30:SMO8B30:
      waiting_for_supplier=0
firmware_node:
      hid=SMO8B30
      modalias=acpi:SMO8B30:SMO8B30:
      path=\_SB_.PCI0.I2C5.DEV_
      status=15
      uevent=MODALIAS=acpi:SMO8B30:SMO8B30:
      uid=0
Quoted 1 line

The kernel module for the appropriate driver (st_lsm6dsx_i2c) is not

loaded on boot.

Quoted 2 lines

Modprobing it does not associate it with the device, as I would expect as the module does not provide an alias for the above acpi/pnp id.

Can you report this issue upstream? Gues to reach out are according to get_maintainers.pl script:

Lorenzo Bianconi <[email protected]> (maintainer:ST LSM6DSx IMU IIO DRIVER) Jonathan Cameron <[email protected]> (maintainer:IIO SUBSYSTEM AND DRIVERS) Lars-Peter Clausen <[email protected]> (reviewer:IIO SUBSYSTEM AND DRIVERS) [email protected] (open list:ST LSM6DSx IMU IIO DRIVER) [email protected] (open list)

Please keep us in the loop.

Regards,
Salvatore

author maintainer
#12 lore raw
Quoted 15 lines · On Mon, 30 Jan 2023 20:02:31 +0000 Darrell Kavanagh…

Thanks. To be clear, before I changed the grub command line, the system always booted up "sideways", even when the sensor was not being detected. This was true for everything, not just Gnome, except grub itself.

I added:

GRUB_CMDLINE_LINUX_DEFAULT="fbcon=rotate:1 video=DSI-1:panel_orientation=right_side_up quiet splash"

This fixed the orientation for pre-splash boot messages, the splash screen and the desktop environment. But not, for example (as I saw after adding my own module signing key for testing your fixes), the MOK validation screens.

Quoted 4 lines

Does this make sense?

Does what you are proposing act at a lower level than changing the systemd hwdb orientation matrix?

I'm not sure on the userspace side of things, but intent is that it will provide the orientation data to any users - though only after the kernel boots and software needs to be aware of it. Give it a go, and if not Bastien (IIRC wrote iio-sensor-proxy) may be able to advise.

For Bastien - patches for kernel side are:
https://lore.kernel.org/linux-iio/[email protected]/T/#t

Darrell is going to test them after back porting to 6.1. With the first patch he gets the right result in gnome but we weren't picking up the rotation matrix at that point (ROTM in ACPI).

Thanks,

Jonathan

Quoted 322 lines

Thanks,
Darrell

Quoted 317 lines · Jonathan Cameron
Quoted 5 lines · On Mon, 30 Jan 2023 18:32:02 +0000 Darrell Kavanagh…
Quoted 2 lines · Jonathan Cameron

That certainly looks like suitable matrix.

Quoted 7 lines
Quoted 4 lines

If we are lucky it matches the handling in bmc150-accel-core.c for an identically named method. That is going to swap the x and y axis which is I'd have thought would be rather bad if what you have is currently working well.

Actually, that would be good, because at present I have to rotate 90 degrees in my grub command line.

:)

I'll see if I can roll a suitable patch.

Jonathan

Quoted 294 lines

Darrell

Quoted 291 lines
Quoted 290 lines
            }

            Method (PRIM, 0, NotSerialized)
            {
                Name (RBUF, Buffer (One)
                {
                     0x01                                             // .
                })
                Return (RBUF) /* \_SB_.PCI0.I2C5.DEV_.PRIM.RBUF */
            }

            Method (_STA, 0, NotSerialized)  // _STA: Status
            {
                If ((GAVT == 0x6A))
                {
                    Return (0x0F)
                }
                Else
                {
                    Return (Zero)
                }
            }

            Method (CALS, 1, NotSerialized)
            {
                Local0 = Arg0
                If (((Local0 == Zero) || (Local0 == Ones)))
                {
                    Local0 = BAC1 /* \BAC1 */
                    Return (Local0)
                }
                Else
                {
                    BAC1 = Local0
                    BACS = Local0
                    BSCA (0xB0)
                }
            }
        }
    }

Thanks,
Darrell

Quoted 243 lines · Jonathan Cameron
Quoted 24 lines · On Mon, 30 Jan 2023 03:37:23 +0000 Darrell Kavanagh…

Forwarding because original html messages were rejected by the server...

---------- Forwarded message ---------
From: Darrell Kavanagh <[email protected]>
Date: Mon, 30 Jan 2023 at 02:52
Subject: Re: Bug#1029850: linux: Driver not loaded for ST Microelectronics LSM6DS3TR-C accelerometer (acpi:SMO8B30:SMO8B30:)
To: Jonathan Cameron <[email protected]>
Cc: <[email protected]>, <[email protected]>, <[email protected]>, <[email protected]>, <[email protected]>

Hi Jonathan,

Thank you. The driver has evolved quite a bit in 6.2 (I read somewhere that 6.2 includes some i2c enhancements), but I adapted your changes to fit my Debian 6.1 kernel and it works. Two IIO devices are created in sysfs, iio-sensor-proxy.service starts up and automatic screen rotation in Gnome just works.

To get the modules to load on boot, I made a small change to your code in st_lsm6dsx_i2c to add the acpi alias to modules.alias:

Quoted 7 lines

adding a null element to st_lsm6dsx_i2c_acpi_match:

    static const struct acpi_device_id st_lsm6dsx_i2c_acpi_match[] = {
         { "SMO8B30", ST_LSM6DS3TRC_ID, },
         { },
    };
then:
   MODULE_DEVICE_TABLE(acpi, st_lsm6dsx_i2c_acpi_match);  

doh! That was indeed sloppy of me to miss even for an untested hack.

Quoted 4 lines

dmesg shows:

Quoted 8 lines
[ 7366.120208] st_lsm6dsx_i2c i2c-SMO8B30:00: supply vdd not found,
using dummy regulator
[ 7366.120260] st_lsm6dsx_i2c i2c-SMO8B30:00: supply vddio not found,
using dummy regulator
[ 7366.650839] st_lsm6dsx_i2c i2c-SMO8B30:00: mounting matrix not

found: using identity...

Is this a problem?

Those are all fine. For regulators that's expected on ACPI and should be harmless as it's up to the firmware to manage power (in DT it may be up to the kernel).
For the mounting matrix, there is often something in ACPI DSDT (non standard though). Could you
cat /sys/firmware/acpi/tables/DSDT > ~/dsdt
then run through iasl from acpitools
iasl -d ~/dsdt
and find the bit related to this device.

If you can then share that there may be a _DSM or similar in there that is effectively the mounting matrix. If we are lucky it will look like some existing versions we have code to handle and can add that support as well.

Either way - I'll spin a formal patch with your fixes above and we can get this upstream for future kernels. Mounting matrix can follow later if needed.

Thanks,

Jonathan

Quoted 165 lines

Thanks again.

Darrell

Quoted 158 lines · Jonathan Cameron
Quoted 5 lines · On Sun, 29 Jan 2023 17:03:51 +0000 Darrell Kavanagh…

Hi,

I raised this bug in Debian, and have been asked to raise it upstream and was given your addresses to do so. Will this email be OK, or should I raise it in a bug tracking system somewhere?

Email is the right option.

Quoted 27 lines

Many thanks,
Darrell

---------- Forwarded message ---------
From: Salvatore Bonaccorso <[email protected]>
Date: Sat, 28 Jan 2023 at 20:33
Subject: Re: Bug#1029850: linux: Driver not loaded for ST Microelectronics LSM6DS3TR-C accelerometer (acpi:SMO8B30:SMO8B30:)
To: Darrell Kavanagh <[email protected]>, <[email protected]>

Hi Darrell,

Quoted 9 lines · Darrell Kavanagh

Package: src:linux
Version: 6.1.4-1
Severity: normal
File: linux
X-Debbugs-Cc: [email protected]

Dear Maintainer,

This is a convertable touchscreen tablet/laptop. The rotation sensor

Quoted 9 lines

device

Quoted 1 line

ST Microelectronics LSM6DS3TR-C does not work. It is detected via ACPI

and the

Quoted 1 line

sysfs trees are created at

devices/pci0000:00/0000:00:17.1/i2c_designware.3/i2c-4/i2c-SMO8B30:00

Quoted 2 lines

and devices/LNXSYSTM:00/LNXSYBUS:00/PNP0A08:00/device:3c/SMO8B30:00 with symlinks bus/acpi/devices/SMO8B30:00 and bus/i2c/devices/i2c-SMO8B30:00,

but

Quoted 1 line

no driver is loaded.

At least this is using the ST PNP ID which is better than average (long story!)

The driver in question (ultimately drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c does not currently have an ACPI support. It should be straight forwards to add though the driver first needs converting to use device_get_match_data() with appropriate fallback so that it will match on ACPI, OF or original spi_device_id tables

Completely untested but something like the following
(the offset in the enum is needed to allow us to tell if we got a result when calling device_get_match_data() as it returns NULL on failure IIRC)

I'm not sure how sucessful the driver will be at finding any interrupts etc, but it may get you basic functionality.

Good luck and others more familiar with the driver may well tell me what I forgot when hacking the below ;)

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx.h +1 -1
index 499fcf8875b4..2617ce236ddc 100644
@@ -39,7 +39,7 @@
 #define ST_ISM330IS_DEV_NAME   "ism330is"
 
 enum st_lsm6dsx_hw_id {
-       ST_LSM6DS3_ID,
+       ST_LSM6DS3_ID = 1,
        ST_LSM6DS3H_ID,
        ST_LSM6DSL_ID,
        ST_LSM6DSM_ID,
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c +12 -2
index df5f60925260..ecfceb2fb3db 100644
@@ -23,10 +23,15 @@ static const struct regmap_config st_lsm6dsx_i2c_regmap_config = {
 
 static int st_lsm6dsx_i2c_probe(struct i2c_client *client)
 {
-       const struct i2c_device_id *id = i2c_client_get_device_id(client);
-       int hw_id = id->driver_data;
+       int hw_id;
        struct regmap *regmap;
 
+       hw_id = (kernel_ulong_t)device_get_match_data(&client->dev);
+       if (!hw_id)
+               hw_id = i2c_client_get_device_id(client)->driver_data;
+       if (!hw_id)
+               return -EINVAL;
+
        regmap = devm_regmap_init_i2c(client, &st_lsm6dsx_i2c_regmap_config);
        if (IS_ERR(regmap)) {
                dev_err(&client->dev, "Failed to register i2c regmap %ld\n", PTR_ERR(regmap));
@@ -129,6 +134,10 @@ static const struct of_device_id st_lsm6dsx_i2c_of_match[] = {
 };
 MODULE_DEVICE_TABLE(of, st_lsm6dsx_i2c_of_match);
 
+static const struct acpi_device_id st_lsm6dsx_i2c_acpi_match[] = {
+       { "SMO8B30", ST_LSM6DS3TRC_ID, },
+};
+
 static const struct i2c_device_id st_lsm6dsx_i2c_id_table[] = {
        { ST_LSM6DS3_DEV_NAME, ST_LSM6DS3_ID },
        { ST_LSM6DS3H_DEV_NAME, ST_LSM6DS3H_ID },
@@ -161,6 +170,7 @@ static struct i2c_driver st_lsm6dsx_driver = {
                .name = "st_lsm6dsx_i2c",
                .pm = pm_sleep_ptr(&st_lsm6dsx_pm_ops),
                .of_match_table = st_lsm6dsx_i2c_of_match,
+               .acpi_match_table = st_lsm6dsx_i2c_acpi_match,
        },
        .probe_new = st_lsm6dsx_i2c_probe,
        .id_table = st_lsm6dsx_i2c_id_table,
Quoted 35 lines
Quoted 17 lines

The device is identifying itself to the kernel with PNP id SMO8B30:
physical_node:

      modalias=acpi:SMO8B30:SMO8B30:
      name=SMO8B30:00
      uevent=MODALIAS=acpi:SMO8B30:SMO8B30:
      waiting_for_supplier=0
firmware_node:
      hid=SMO8B30
      modalias=acpi:SMO8B30:SMO8B30:
      path=\_SB_.PCI0.I2C5.DEV_
      status=15
      uevent=MODALIAS=acpi:SMO8B30:SMO8B30:
      uid=0
Quoted 1 line

The kernel module for the appropriate driver (st_lsm6dsx_i2c) is not

loaded on boot.

Quoted 2 lines

Modprobing it does not associate it with the device, as I would expect as the module does not provide an alias for the above acpi/pnp id.

Can you report this issue upstream? Gues to reach out are according to get_maintainers.pl script:

Lorenzo Bianconi <[email protected]> (maintainer:ST LSM6DSx IMU IIO DRIVER) Jonathan Cameron <[email protected]> (maintainer:IIO SUBSYSTEM AND DRIVERS) Lars-Peter Clausen <[email protected]> (reviewer:IIO SUBSYSTEM AND DRIVERS) [email protected] (open list:ST LSM6DSx IMU IIO DRIVER) [email protected] (open list)

Please keep us in the loop.

Regards,
Salvatore

#13 lore raw

Just a quick update on testing: I'm finding what look like interactions between the kernel parm video=DSI-1:panel_orientation=right_side_up and the sensor, which also seem to differ depending on whether I'm in an X or Wayland session. I'm documenting the effects of the various combinations and will post my results when complete.

Thanks,
Darrell

Quoted 363 lines · Jonathan Cameron
Quoted 15 lines · On Mon, 30 Jan 2023 20:02:31 +0000 Darrell Kavanagh…

Thanks. To be clear, before I changed the grub command line, the system always booted up "sideways", even when the sensor was not being detected. This was true for everything, not just Gnome, except grub itself.

I added:

GRUB_CMDLINE_LINUX_DEFAULT="fbcon=rotate:1 video=DSI-1:panel_orientation=right_side_up quiet splash"

This fixed the orientation for pre-splash boot messages, the splash screen and the desktop environment. But not, for example (as I saw after adding my own module signing key for testing your fixes), the MOK validation screens.

Quoted 4 lines

Does this make sense?

Does what you are proposing act at a lower level than changing the systemd hwdb orientation matrix?

I'm not sure on the userspace side of things, but intent is that it will provide the orientation data to any users - though only after the kernel boots and software needs to be aware of it. Give it a go, and if not Bastien (IIRC wrote iio-sensor-proxy) may be able to advise.

For Bastien - patches for kernel side are:
https://lore.kernel.org/linux-iio/[email protected]/T/#t

Darrell is going to test them after back porting to 6.1. With the first patch he gets the right result in gnome but we weren't picking up the rotation matrix at that point (ROTM in ACPI).

Thanks,

Jonathan

Quoted 322 lines

Thanks,
Darrell

Quoted 317 lines · Jonathan Cameron
Quoted 5 lines · On Mon, 30 Jan 2023 18:32:02 +0000 Darrell Kavanagh…
Quoted 2 lines · Jonathan Cameron

That certainly looks like suitable matrix.

Quoted 7 lines
Quoted 4 lines

If we are lucky it matches the handling in bmc150-accel-core.c for an identically named method. That is going to swap the x and y axis which is I'd have thought would be rather bad if what you have is currently working well.

Actually, that would be good, because at present I have to rotate 90 degrees in my grub command line.

:)

I'll see if I can roll a suitable patch.

Jonathan

Quoted 294 lines

Darrell

Quoted 291 lines
Quoted 290 lines
            }

            Method (PRIM, 0, NotSerialized)
            {
                Name (RBUF, Buffer (One)
                {
                     0x01                                             // .
                })
                Return (RBUF) /* \_SB_.PCI0.I2C5.DEV_.PRIM.RBUF */
            }

            Method (_STA, 0, NotSerialized)  // _STA: Status
            {
                If ((GAVT == 0x6A))
                {
                    Return (0x0F)
                }
                Else
                {
                    Return (Zero)
                }
            }

            Method (CALS, 1, NotSerialized)
            {
                Local0 = Arg0
                If (((Local0 == Zero) || (Local0 == Ones)))
                {
                    Local0 = BAC1 /* \BAC1 */
                    Return (Local0)
                }
                Else
                {
                    BAC1 = Local0
                    BACS = Local0
                    BSCA (0xB0)
                }
            }
        }
    }

Thanks,
Darrell

Quoted 243 lines · Jonathan Cameron
Quoted 24 lines · On Mon, 30 Jan 2023 03:37:23 +0000 Darrell Kavanagh…

Forwarding because original html messages were rejected by the server...

---------- Forwarded message ---------
From: Darrell Kavanagh <[email protected]>
Date: Mon, 30 Jan 2023 at 02:52
Subject: Re: Bug#1029850: linux: Driver not loaded for ST Microelectronics LSM6DS3TR-C accelerometer (acpi:SMO8B30:SMO8B30:)
To: Jonathan Cameron <[email protected]>
Cc: <[email protected]>, <[email protected]>, <[email protected]>, <[email protected]>, <[email protected]>

Hi Jonathan,

Thank you. The driver has evolved quite a bit in 6.2 (I read somewhere that 6.2 includes some i2c enhancements), but I adapted your changes to fit my Debian 6.1 kernel and it works. Two IIO devices are created in sysfs, iio-sensor-proxy.service starts up and automatic screen rotation in Gnome just works.

To get the modules to load on boot, I made a small change to your code in st_lsm6dsx_i2c to add the acpi alias to modules.alias:

Quoted 7 lines

adding a null element to st_lsm6dsx_i2c_acpi_match:

    static const struct acpi_device_id st_lsm6dsx_i2c_acpi_match[] = {
         { "SMO8B30", ST_LSM6DS3TRC_ID, },
         { },
    };
then:
   MODULE_DEVICE_TABLE(acpi, st_lsm6dsx_i2c_acpi_match);

doh! That was indeed sloppy of me to miss even for an untested hack.

Quoted 4 lines

dmesg shows:

Quoted 8 lines
[ 7366.120208] st_lsm6dsx_i2c i2c-SMO8B30:00: supply vdd not found,
using dummy regulator
[ 7366.120260] st_lsm6dsx_i2c i2c-SMO8B30:00: supply vddio not found,
using dummy regulator
[ 7366.650839] st_lsm6dsx_i2c i2c-SMO8B30:00: mounting matrix not

found: using identity...

Is this a problem?

Those are all fine. For regulators that's expected on ACPI and should be harmless as it's up to the firmware to manage power (in DT it may be up to the kernel).
For the mounting matrix, there is often something in ACPI DSDT (non standard though). Could you
cat /sys/firmware/acpi/tables/DSDT > ~/dsdt
then run through iasl from acpitools
iasl -d ~/dsdt
and find the bit related to this device.

If you can then share that there may be a _DSM or similar in there that is effectively the mounting matrix. If we are lucky it will look like some existing versions we have code to handle and can add that support as well.

Either way - I'll spin a formal patch with your fixes above and we can get this upstream for future kernels. Mounting matrix can follow later if needed.

Thanks,

Jonathan

Quoted 165 lines

Thanks again.

Darrell

Quoted 158 lines · Jonathan Cameron
Quoted 5 lines · On Sun, 29 Jan 2023 17:03:51 +0000 Darrell Kavanagh…

Hi,

I raised this bug in Debian, and have been asked to raise it upstream and was given your addresses to do so. Will this email be OK, or should I raise it in a bug tracking system somewhere?

Email is the right option.

Quoted 27 lines

Many thanks,
Darrell

---------- Forwarded message ---------
From: Salvatore Bonaccorso <[email protected]>
Date: Sat, 28 Jan 2023 at 20:33
Subject: Re: Bug#1029850: linux: Driver not loaded for ST Microelectronics LSM6DS3TR-C accelerometer (acpi:SMO8B30:SMO8B30:)
To: Darrell Kavanagh <[email protected]>, <[email protected]>

Hi Darrell,

Quoted 9 lines · Darrell Kavanagh

Package: src:linux
Version: 6.1.4-1
Severity: normal
File: linux
X-Debbugs-Cc: [email protected]

Dear Maintainer,

This is a convertable touchscreen tablet/laptop. The rotation sensor

Quoted 9 lines

device

Quoted 1 line

ST Microelectronics LSM6DS3TR-C does not work. It is detected via ACPI

and the

Quoted 1 line

sysfs trees are created at

devices/pci0000:00/0000:00:17.1/i2c_designware.3/i2c-4/i2c-SMO8B30:00

Quoted 2 lines

and devices/LNXSYSTM:00/LNXSYBUS:00/PNP0A08:00/device:3c/SMO8B30:00 with symlinks bus/acpi/devices/SMO8B30:00 and bus/i2c/devices/i2c-SMO8B30:00,

but

Quoted 1 line

no driver is loaded.

At least this is using the ST PNP ID which is better than average (long story!)

The driver in question (ultimately drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c does not currently have an ACPI support. It should be straight forwards to add though the driver first needs converting to use device_get_match_data() with appropriate fallback so that it will match on ACPI, OF or original spi_device_id tables

Completely untested but something like the following
(the offset in the enum is needed to allow us to tell if we got a result when calling device_get_match_data() as it returns NULL on failure IIRC)

I'm not sure how sucessful the driver will be at finding any interrupts etc, but it may get you basic functionality.

Good luck and others more familiar with the driver may well tell me what I forgot when hacking the below ;)

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx.h +1 -1
index 499fcf8875b4..2617ce236ddc 100644
@@ -39,7 +39,7 @@
 #define ST_ISM330IS_DEV_NAME   "ism330is"
 
 enum st_lsm6dsx_hw_id {
-       ST_LSM6DS3_ID,
+       ST_LSM6DS3_ID = 1,
        ST_LSM6DS3H_ID,
        ST_LSM6DSL_ID,
        ST_LSM6DSM_ID,
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c +12 -2
index df5f60925260..ecfceb2fb3db 100644
@@ -23,10 +23,15 @@ static const struct regmap_config st_lsm6dsx_i2c_regmap_config = {
 
 static int st_lsm6dsx_i2c_probe(struct i2c_client *client)
 {
-       const struct i2c_device_id *id = i2c_client_get_device_id(client);
-       int hw_id = id->driver_data;
+       int hw_id;
        struct regmap *regmap;
 
+       hw_id = (kernel_ulong_t)device_get_match_data(&client->dev);
+       if (!hw_id)
+               hw_id = i2c_client_get_device_id(client)->driver_data;
+       if (!hw_id)
+               return -EINVAL;
+
        regmap = devm_regmap_init_i2c(client, &st_lsm6dsx_i2c_regmap_config);
        if (IS_ERR(regmap)) {
                dev_err(&client->dev, "Failed to register i2c regmap %ld\n", PTR_ERR(regmap));
@@ -129,6 +134,10 @@ static const struct of_device_id st_lsm6dsx_i2c_of_match[] = {
 };
 MODULE_DEVICE_TABLE(of, st_lsm6dsx_i2c_of_match);
 
+static const struct acpi_device_id st_lsm6dsx_i2c_acpi_match[] = {
+       { "SMO8B30", ST_LSM6DS3TRC_ID, },
+};
+
 static const struct i2c_device_id st_lsm6dsx_i2c_id_table[] = {
        { ST_LSM6DS3_DEV_NAME, ST_LSM6DS3_ID },
        { ST_LSM6DS3H_DEV_NAME, ST_LSM6DS3H_ID },
@@ -161,6 +170,7 @@ static struct i2c_driver st_lsm6dsx_driver = {
                .name = "st_lsm6dsx_i2c",
                .pm = pm_sleep_ptr(&st_lsm6dsx_pm_ops),
                .of_match_table = st_lsm6dsx_i2c_of_match,
+               .acpi_match_table = st_lsm6dsx_i2c_acpi_match,
        },
        .probe_new = st_lsm6dsx_i2c_probe,
        .id_table = st_lsm6dsx_i2c_id_table,
Quoted 35 lines
Quoted 17 lines

The device is identifying itself to the kernel with PNP id SMO8B30:
physical_node:

      modalias=acpi:SMO8B30:SMO8B30:
      name=SMO8B30:00
      uevent=MODALIAS=acpi:SMO8B30:SMO8B30:
      waiting_for_supplier=0
firmware_node:
      hid=SMO8B30
      modalias=acpi:SMO8B30:SMO8B30:
      path=\_SB_.PCI0.I2C5.DEV_
      status=15
      uevent=MODALIAS=acpi:SMO8B30:SMO8B30:
      uid=0
Quoted 1 line

The kernel module for the appropriate driver (st_lsm6dsx_i2c) is not

loaded on boot.

Quoted 2 lines

Modprobing it does not associate it with the device, as I would expect as the module does not provide an alias for the above acpi/pnp id.

Can you report this issue upstream? Gues to reach out are according to get_maintainers.pl script:

Lorenzo Bianconi <[email protected]> (maintainer:ST LSM6DSx IMU IIO DRIVER) Jonathan Cameron <[email protected]> (maintainer:IIO SUBSYSTEM AND DRIVERS) Lars-Peter Clausen <[email protected]> (reviewer:IIO SUBSYSTEM AND DRIVERS) [email protected] (open list:ST LSM6DSx IMU IIO DRIVER) [email protected] (open list)

Please keep us in the loop.

Regards,
Salvatore

#14 lore raw

Hello, all.

I've finally reached a conclusion on this, after testing all the combinations of the patches (with and without reading the acpi mounting matrix), window managers (wayland, xorg) and the presence or not of my custom kernel parms.

What works well is the full set of patches with the custom kernel parms and a new hwdb entry for the sensor:

sensor:modalias:acpi:SMO8B30*:dmi:*:svnLENOVO*:pn82AT:*
ACCEL_MOUNT_MATRIX=0, 1, 0; -1, 0, 0; 0, 0, 1

The autorotate then works correctly in wayland and xorg, but for xorg, the settings say the screen is "portrait left" when in actual fact it is in standard laptop landscape orientation. Wayland does not have this problem (I guess because wayland's view of the screen is straight from the kernel).

Without the hwdb entry, the orientation is 90 degrees out without using the acpi matrix and 180 degrees out when using it. I could have gone either way here with appropriate hwdb entries, but my view is that we *should* be using the matrix.

BTW, when I reported before that it was working correctly in Gnome, I was using a third party shell extension to force auto-rotation rather than actually detaching the keyboard. I take the view that if an extension works differently to the built-in tablet mode, it's the extension's problem!

The only thing that concerns me is the need for custom kernel parms. It would be better if there was a way to avoid this, so that the user didn't have to mess around with their grub config. Though having said that, the sensors fix as we have it doesn't make things worse - under currently released kernels the screen always starts up sideways unless custom parms are added in grub.

Without the custom parms, the patches and the new hwdb entry do ensure that things end up the right way around, but it's a bit clunky. The boot messages/splash is sideways and you can see the sensors kicking in on the gdm screen: it shows momentarily sideways before righting itself. This can also cause the scaling to be set to 200%, which can carry through to the user's session.

So thanks for your work on this - I now have a fully functioning device. I hope the changes are robust enough to be pushed into a future kernel so that others can benefit.

Regards,
Darrell

Quoted 379 lines · Darrell Kavanagh

Just a quick update on testing: I'm finding what look like interactions between the kernel parm video=DSI-1:panel_orientation=right_side_up and the sensor, which also seem to differ depending on whether I'm in an X or Wayland session. I'm documenting the effects of the various combinations and will post my results when complete.

Thanks,
Darrell

Quoted 363 lines · Jonathan Cameron
Quoted 15 lines · On Mon, 30 Jan 2023 20:02:31 +0000 Darrell Kavanagh…

Thanks. To be clear, before I changed the grub command line, the system always booted up "sideways", even when the sensor was not being detected. This was true for everything, not just Gnome, except grub itself.

I added:

GRUB_CMDLINE_LINUX_DEFAULT="fbcon=rotate:1 video=DSI-1:panel_orientation=right_side_up quiet splash"

This fixed the orientation for pre-splash boot messages, the splash screen and the desktop environment. But not, for example (as I saw after adding my own module signing key for testing your fixes), the MOK validation screens.

Quoted 4 lines

Does this make sense?

Does what you are proposing act at a lower level than changing the systemd hwdb orientation matrix?

I'm not sure on the userspace side of things, but intent is that it will provide the orientation data to any users - though only after the kernel boots and software needs to be aware of it. Give it a go, and if not Bastien (IIRC wrote iio-sensor-proxy) may be able to advise.

For Bastien - patches for kernel side are:
https://lore.kernel.org/linux-iio/[email protected]/T/#t

Darrell is going to test them after back porting to 6.1. With the first patch he gets the right result in gnome but we weren't picking up the rotation matrix at that point (ROTM in ACPI).

Thanks,

Jonathan

Quoted 322 lines

Thanks,
Darrell

Quoted 317 lines · Jonathan Cameron
Quoted 5 lines · On Mon, 30 Jan 2023 18:32:02 +0000 Darrell Kavanagh…
Quoted 2 lines · Jonathan Cameron

That certainly looks like suitable matrix.

Quoted 7 lines
Quoted 4 lines

If we are lucky it matches the handling in bmc150-accel-core.c for an identically named method. That is going to swap the x and y axis which is I'd have thought would be rather bad if what you have is currently working well.

Actually, that would be good, because at present I have to rotate 90 degrees in my grub command line.

:)

I'll see if I can roll a suitable patch.

Jonathan

Quoted 294 lines

Darrell

Quoted 291 lines
Quoted 290 lines
            }

            Method (PRIM, 0, NotSerialized)
            {
                Name (RBUF, Buffer (One)
                {
                     0x01                                             // .
                })
                Return (RBUF) /* \_SB_.PCI0.I2C5.DEV_.PRIM.RBUF */
            }

            Method (_STA, 0, NotSerialized)  // _STA: Status
            {
                If ((GAVT == 0x6A))
                {
                    Return (0x0F)
                }
                Else
                {
                    Return (Zero)
                }
            }

            Method (CALS, 1, NotSerialized)
            {
                Local0 = Arg0
                If (((Local0 == Zero) || (Local0 == Ones)))
                {
                    Local0 = BAC1 /* \BAC1 */
                    Return (Local0)
                }
                Else
                {
                    BAC1 = Local0
                    BACS = Local0
                    BSCA (0xB0)
                }
            }
        }
    }

Thanks,
Darrell

Quoted 243 lines · Jonathan Cameron
Quoted 24 lines · On Mon, 30 Jan 2023 03:37:23 +0000 Darrell Kavanagh…

Forwarding because original html messages were rejected by the server...

---------- Forwarded message ---------
From: Darrell Kavanagh <[email protected]>
Date: Mon, 30 Jan 2023 at 02:52
Subject: Re: Bug#1029850: linux: Driver not loaded for ST Microelectronics LSM6DS3TR-C accelerometer (acpi:SMO8B30:SMO8B30:)
To: Jonathan Cameron <[email protected]>
Cc: <[email protected]>, <[email protected]>, <[email protected]>, <[email protected]>, <[email protected]>

Hi Jonathan,

Thank you. The driver has evolved quite a bit in 6.2 (I read somewhere that 6.2 includes some i2c enhancements), but I adapted your changes to fit my Debian 6.1 kernel and it works. Two IIO devices are created in sysfs, iio-sensor-proxy.service starts up and automatic screen rotation in Gnome just works.

To get the modules to load on boot, I made a small change to your code in st_lsm6dsx_i2c to add the acpi alias to modules.alias:

Quoted 7 lines

adding a null element to st_lsm6dsx_i2c_acpi_match:

    static const struct acpi_device_id st_lsm6dsx_i2c_acpi_match[] = {
         { "SMO8B30", ST_LSM6DS3TRC_ID, },
         { },
    };
then:
   MODULE_DEVICE_TABLE(acpi, st_lsm6dsx_i2c_acpi_match);

doh! That was indeed sloppy of me to miss even for an untested hack.

Quoted 4 lines

dmesg shows:

Quoted 8 lines
[ 7366.120208] st_lsm6dsx_i2c i2c-SMO8B30:00: supply vdd not found,
using dummy regulator
[ 7366.120260] st_lsm6dsx_i2c i2c-SMO8B30:00: supply vddio not found,
using dummy regulator
[ 7366.650839] st_lsm6dsx_i2c i2c-SMO8B30:00: mounting matrix not

found: using identity...

Is this a problem?

Those are all fine. For regulators that's expected on ACPI and should be harmless as it's up to the firmware to manage power (in DT it may be up to the kernel).
For the mounting matrix, there is often something in ACPI DSDT (non standard though). Could you
cat /sys/firmware/acpi/tables/DSDT > ~/dsdt
then run through iasl from acpitools
iasl -d ~/dsdt
and find the bit related to this device.

If you can then share that there may be a _DSM or similar in there that is effectively the mounting matrix. If we are lucky it will look like some existing versions we have code to handle and can add that support as well.

Either way - I'll spin a formal patch with your fixes above and we can get this upstream for future kernels. Mounting matrix can follow later if needed.

Thanks,

Jonathan

Quoted 165 lines

Thanks again.

Darrell

Quoted 158 lines · Jonathan Cameron
Quoted 5 lines · On Sun, 29 Jan 2023 17:03:51 +0000 Darrell Kavanagh…

Hi,

I raised this bug in Debian, and have been asked to raise it upstream and was given your addresses to do so. Will this email be OK, or should I raise it in a bug tracking system somewhere?

Email is the right option.

Quoted 27 lines

Many thanks,
Darrell

---------- Forwarded message ---------
From: Salvatore Bonaccorso <[email protected]>
Date: Sat, 28 Jan 2023 at 20:33
Subject: Re: Bug#1029850: linux: Driver not loaded for ST Microelectronics LSM6DS3TR-C accelerometer (acpi:SMO8B30:SMO8B30:)
To: Darrell Kavanagh <[email protected]>, <[email protected]>

Hi Darrell,

Quoted 9 lines · Darrell Kavanagh

Package: src:linux
Version: 6.1.4-1
Severity: normal
File: linux
X-Debbugs-Cc: [email protected]

Dear Maintainer,

This is a convertable touchscreen tablet/laptop. The rotation sensor

Quoted 9 lines

device

Quoted 1 line

ST Microelectronics LSM6DS3TR-C does not work. It is detected via ACPI

and the

Quoted 1 line

sysfs trees are created at

devices/pci0000:00/0000:00:17.1/i2c_designware.3/i2c-4/i2c-SMO8B30:00

Quoted 2 lines

and devices/LNXSYSTM:00/LNXSYBUS:00/PNP0A08:00/device:3c/SMO8B30:00 with symlinks bus/acpi/devices/SMO8B30:00 and bus/i2c/devices/i2c-SMO8B30:00,

but

Quoted 1 line

no driver is loaded.

At least this is using the ST PNP ID which is better than average (long story!)

The driver in question (ultimately drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c does not currently have an ACPI support. It should be straight forwards to add though the driver first needs converting to use device_get_match_data() with appropriate fallback so that it will match on ACPI, OF or original spi_device_id tables

Completely untested but something like the following
(the offset in the enum is needed to allow us to tell if we got a result when calling device_get_match_data() as it returns NULL on failure IIRC)

I'm not sure how sucessful the driver will be at finding any interrupts etc, but it may get you basic functionality.

Good luck and others more familiar with the driver may well tell me what I forgot when hacking the below ;)

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx.h +1 -1
index 499fcf8875b4..2617ce236ddc 100644
@@ -39,7 +39,7 @@
 #define ST_ISM330IS_DEV_NAME   "ism330is"
 
 enum st_lsm6dsx_hw_id {
-       ST_LSM6DS3_ID,
+       ST_LSM6DS3_ID = 1,
        ST_LSM6DS3H_ID,
        ST_LSM6DSL_ID,
        ST_LSM6DSM_ID,
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c +12 -2
index df5f60925260..ecfceb2fb3db 100644
@@ -23,10 +23,15 @@ static const struct regmap_config st_lsm6dsx_i2c_regmap_config = {
 
 static int st_lsm6dsx_i2c_probe(struct i2c_client *client)
 {
-       const struct i2c_device_id *id = i2c_client_get_device_id(client);
-       int hw_id = id->driver_data;
+       int hw_id;
        struct regmap *regmap;
 
+       hw_id = (kernel_ulong_t)device_get_match_data(&client->dev);
+       if (!hw_id)
+               hw_id = i2c_client_get_device_id(client)->driver_data;
+       if (!hw_id)
+               return -EINVAL;
+
        regmap = devm_regmap_init_i2c(client, &st_lsm6dsx_i2c_regmap_config);
        if (IS_ERR(regmap)) {
                dev_err(&client->dev, "Failed to register i2c regmap %ld\n", PTR_ERR(regmap));
@@ -129,6 +134,10 @@ static const struct of_device_id st_lsm6dsx_i2c_of_match[] = {
 };
 MODULE_DEVICE_TABLE(of, st_lsm6dsx_i2c_of_match);
 
+static const struct acpi_device_id st_lsm6dsx_i2c_acpi_match[] = {
+       { "SMO8B30", ST_LSM6DS3TRC_ID, },
+};
+
 static const struct i2c_device_id st_lsm6dsx_i2c_id_table[] = {
        { ST_LSM6DS3_DEV_NAME, ST_LSM6DS3_ID },
        { ST_LSM6DS3H_DEV_NAME, ST_LSM6DS3H_ID },
@@ -161,6 +170,7 @@ static struct i2c_driver st_lsm6dsx_driver = {
                .name = "st_lsm6dsx_i2c",
                .pm = pm_sleep_ptr(&st_lsm6dsx_pm_ops),
                .of_match_table = st_lsm6dsx_i2c_of_match,
+               .acpi_match_table = st_lsm6dsx_i2c_acpi_match,
        },
        .probe_new = st_lsm6dsx_i2c_probe,
        .id_table = st_lsm6dsx_i2c_id_table,
Quoted 35 lines
Quoted 17 lines

The device is identifying itself to the kernel with PNP id SMO8B30:
physical_node:

      modalias=acpi:SMO8B30:SMO8B30:
      name=SMO8B30:00
      uevent=MODALIAS=acpi:SMO8B30:SMO8B30:
      waiting_for_supplier=0
firmware_node:
      hid=SMO8B30
      modalias=acpi:SMO8B30:SMO8B30:
      path=\_SB_.PCI0.I2C5.DEV_
      status=15
      uevent=MODALIAS=acpi:SMO8B30:SMO8B30:
      uid=0
Quoted 1 line

The kernel module for the appropriate driver (st_lsm6dsx_i2c) is not

loaded on boot.

Quoted 2 lines

Modprobing it does not associate it with the device, as I would expect as the module does not provide an alias for the above acpi/pnp id.

Can you report this issue upstream? Gues to reach out are according to get_maintainers.pl script:

Lorenzo Bianconi <[email protected]> (maintainer:ST LSM6DSx IMU IIO DRIVER) Jonathan Cameron <[email protected]> (maintainer:IIO SUBSYSTEM AND DRIVERS) Lars-Peter Clausen <[email protected]> (reviewer:IIO SUBSYSTEM AND DRIVERS) [email protected] (open list:ST LSM6DSx IMU IIO DRIVER) [email protected] (open list)

Please keep us in the loop.

Regards,
Salvatore

#15 lore raw

I should also add that I'm more than willing to help in testing any further changes for you, whether for this bug or future ones.

Darrell

Quoted 434 lines · Darrell Kavanagh

Hello, all.

I've finally reached a conclusion on this, after testing all the combinations of the patches (with and without reading the acpi mounting matrix), window managers (wayland, xorg) and the presence or not of my custom kernel parms.

What works well is the full set of patches with the custom kernel parms and a new hwdb entry for the sensor:

sensor:modalias:acpi:SMO8B30*:dmi:*:svnLENOVO*:pn82AT:*
ACCEL_MOUNT_MATRIX=0, 1, 0; -1, 0, 0; 0, 0, 1

The autorotate then works correctly in wayland and xorg, but for xorg, the settings say the screen is "portrait left" when in actual fact it is in standard laptop landscape orientation. Wayland does not have this problem (I guess because wayland's view of the screen is straight from the kernel).

Without the hwdb entry, the orientation is 90 degrees out without using the acpi matrix and 180 degrees out when using it. I could have gone either way here with appropriate hwdb entries, but my view is that we *should* be using the matrix.

BTW, when I reported before that it was working correctly in Gnome, I was using a third party shell extension to force auto-rotation rather than actually detaching the keyboard. I take the view that if an extension works differently to the built-in tablet mode, it's the extension's problem!

The only thing that concerns me is the need for custom kernel parms. It would be better if there was a way to avoid this, so that the user didn't have to mess around with their grub config. Though having said that, the sensors fix as we have it doesn't make things worse - under currently released kernels the screen always starts up sideways unless custom parms are added in grub.

Without the custom parms, the patches and the new hwdb entry do ensure that things end up the right way around, but it's a bit clunky. The boot messages/splash is sideways and you can see the sensors kicking in on the gdm screen: it shows momentarily sideways before righting itself. This can also cause the scaling to be set to 200%, which can carry through to the user's session.

So thanks for your work on this - I now have a fully functioning device. I hope the changes are robust enough to be pushed into a future kernel so that others can benefit.

Regards,
Darrell

Quoted 379 lines · Darrell Kavanagh

Just a quick update on testing: I'm finding what look like interactions between the kernel parm video=DSI-1:panel_orientation=right_side_up and the sensor, which also seem to differ depending on whether I'm in an X or Wayland session. I'm documenting the effects of the various combinations and will post my results when complete.

Thanks,
Darrell

Quoted 363 lines · Jonathan Cameron
Quoted 15 lines · On Mon, 30 Jan 2023 20:02:31 +0000 Darrell Kavanagh…

Thanks. To be clear, before I changed the grub command line, the system always booted up "sideways", even when the sensor was not being detected. This was true for everything, not just Gnome, except grub itself.

I added:

GRUB_CMDLINE_LINUX_DEFAULT="fbcon=rotate:1 video=DSI-1:panel_orientation=right_side_up quiet splash"

This fixed the orientation for pre-splash boot messages, the splash screen and the desktop environment. But not, for example (as I saw after adding my own module signing key for testing your fixes), the MOK validation screens.

Quoted 4 lines

Does this make sense?

Does what you are proposing act at a lower level than changing the systemd hwdb orientation matrix?

I'm not sure on the userspace side of things, but intent is that it will provide the orientation data to any users - though only after the kernel boots and software needs to be aware of it. Give it a go, and if not Bastien (IIRC wrote iio-sensor-proxy) may be able to advise.

For Bastien - patches for kernel side are:
https://lore.kernel.org/linux-iio/[email protected]/T/#t

Darrell is going to test them after back porting to 6.1. With the first patch he gets the right result in gnome but we weren't picking up the rotation matrix at that point (ROTM in ACPI).

Thanks,

Jonathan

Quoted 322 lines

Thanks,
Darrell

Quoted 317 lines · Jonathan Cameron
Quoted 5 lines · On Mon, 30 Jan 2023 18:32:02 +0000 Darrell Kavanagh…
Quoted 2 lines · Jonathan Cameron

That certainly looks like suitable matrix.

Quoted 7 lines
Quoted 4 lines

If we are lucky it matches the handling in bmc150-accel-core.c for an identically named method. That is going to swap the x and y axis which is I'd have thought would be rather bad if what you have is currently working well.

Actually, that would be good, because at present I have to rotate 90 degrees in my grub command line.

:)

I'll see if I can roll a suitable patch.

Jonathan

Quoted 294 lines

Darrell

Quoted 291 lines
Quoted 290 lines
            }

            Method (PRIM, 0, NotSerialized)
            {
                Name (RBUF, Buffer (One)
                {
                     0x01                                             // .
                })
                Return (RBUF) /* \_SB_.PCI0.I2C5.DEV_.PRIM.RBUF */
            }

            Method (_STA, 0, NotSerialized)  // _STA: Status
            {
                If ((GAVT == 0x6A))
                {
                    Return (0x0F)
                }
                Else
                {
                    Return (Zero)
                }
            }

            Method (CALS, 1, NotSerialized)
            {
                Local0 = Arg0
                If (((Local0 == Zero) || (Local0 == Ones)))
                {
                    Local0 = BAC1 /* \BAC1 */
                    Return (Local0)
                }
                Else
                {
                    BAC1 = Local0
                    BACS = Local0
                    BSCA (0xB0)
                }
            }
        }
    }

Thanks,
Darrell

Quoted 243 lines · Jonathan Cameron
Quoted 24 lines · On Mon, 30 Jan 2023 03:37:23 +0000 Darrell Kavanagh…

Forwarding because original html messages were rejected by the server...

---------- Forwarded message ---------
From: Darrell Kavanagh <[email protected]>
Date: Mon, 30 Jan 2023 at 02:52
Subject: Re: Bug#1029850: linux: Driver not loaded for ST Microelectronics LSM6DS3TR-C accelerometer (acpi:SMO8B30:SMO8B30:)
To: Jonathan Cameron <[email protected]>
Cc: <[email protected]>, <[email protected]>, <[email protected]>, <[email protected]>, <[email protected]>

Hi Jonathan,

Thank you. The driver has evolved quite a bit in 6.2 (I read somewhere that 6.2 includes some i2c enhancements), but I adapted your changes to fit my Debian 6.1 kernel and it works. Two IIO devices are created in sysfs, iio-sensor-proxy.service starts up and automatic screen rotation in Gnome just works.

To get the modules to load on boot, I made a small change to your code in st_lsm6dsx_i2c to add the acpi alias to modules.alias:

Quoted 7 lines

adding a null element to st_lsm6dsx_i2c_acpi_match:

    static const struct acpi_device_id st_lsm6dsx_i2c_acpi_match[] = {
         { "SMO8B30", ST_LSM6DS3TRC_ID, },
         { },
    };
then:
   MODULE_DEVICE_TABLE(acpi, st_lsm6dsx_i2c_acpi_match);

doh! That was indeed sloppy of me to miss even for an untested hack.

Quoted 4 lines

dmesg shows:

Quoted 8 lines
[ 7366.120208] st_lsm6dsx_i2c i2c-SMO8B30:00: supply vdd not found,
using dummy regulator
[ 7366.120260] st_lsm6dsx_i2c i2c-SMO8B30:00: supply vddio not found,
using dummy regulator
[ 7366.650839] st_lsm6dsx_i2c i2c-SMO8B30:00: mounting matrix not

found: using identity...

Is this a problem?

Those are all fine. For regulators that's expected on ACPI and should be harmless as it's up to the firmware to manage power (in DT it may be up to the kernel).
For the mounting matrix, there is often something in ACPI DSDT (non standard though). Could you
cat /sys/firmware/acpi/tables/DSDT > ~/dsdt
then run through iasl from acpitools
iasl -d ~/dsdt
and find the bit related to this device.

If you can then share that there may be a _DSM or similar in there that is effectively the mounting matrix. If we are lucky it will look like some existing versions we have code to handle and can add that support as well.

Either way - I'll spin a formal patch with your fixes above and we can get this upstream for future kernels. Mounting matrix can follow later if needed.

Thanks,

Jonathan

Quoted 165 lines

Thanks again.

Darrell

Quoted 158 lines · Jonathan Cameron
Quoted 5 lines · On Sun, 29 Jan 2023 17:03:51 +0000 Darrell Kavanagh…

Hi,

I raised this bug in Debian, and have been asked to raise it upstream and was given your addresses to do so. Will this email be OK, or should I raise it in a bug tracking system somewhere?

Email is the right option.

Quoted 27 lines

Many thanks,
Darrell

---------- Forwarded message ---------
From: Salvatore Bonaccorso <[email protected]>
Date: Sat, 28 Jan 2023 at 20:33
Subject: Re: Bug#1029850: linux: Driver not loaded for ST Microelectronics LSM6DS3TR-C accelerometer (acpi:SMO8B30:SMO8B30:)
To: Darrell Kavanagh <[email protected]>, <[email protected]>

Hi Darrell,

Quoted 9 lines · Darrell Kavanagh

Package: src:linux
Version: 6.1.4-1
Severity: normal
File: linux
X-Debbugs-Cc: [email protected]

Dear Maintainer,

This is a convertable touchscreen tablet/laptop. The rotation sensor

Quoted 9 lines

device

Quoted 1 line

ST Microelectronics LSM6DS3TR-C does not work. It is detected via ACPI

and the

Quoted 1 line

sysfs trees are created at

devices/pci0000:00/0000:00:17.1/i2c_designware.3/i2c-4/i2c-SMO8B30:00

Quoted 2 lines

and devices/LNXSYSTM:00/LNXSYBUS:00/PNP0A08:00/device:3c/SMO8B30:00 with
symlinks bus/acpi/devices/SMO8B30:00 and bus/i2c/devices/i2c-SMO8B30:00,

but

Quoted 1 line

no driver is loaded.

At least this is using the ST PNP ID which is better than average (long story!)

The driver in question (ultimately drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c does not currently have an ACPI support. It should be straight forwards to add though the driver first needs converting to use device_get_match_data() with appropriate fallback so that it will match on ACPI, OF or original spi_device_id tables

Completely untested but something like the following
(the offset in the enum is needed to allow us to tell if we got a result when calling device_get_match_data() as it returns NULL on failure IIRC)

I'm not sure how sucessful the driver will be at finding any interrupts etc, but it may get you basic functionality.

Good luck and others more familiar with the driver may well tell me what I forgot when hacking the below ;)

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx.h +1 -1
index 499fcf8875b4..2617ce236ddc 100644
@@ -39,7 +39,7 @@
 #define ST_ISM330IS_DEV_NAME   "ism330is"
 
 enum st_lsm6dsx_hw_id {
-       ST_LSM6DS3_ID,
+       ST_LSM6DS3_ID = 1,
        ST_LSM6DS3H_ID,
        ST_LSM6DSL_ID,
        ST_LSM6DSM_ID,
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c +12 -2
index df5f60925260..ecfceb2fb3db 100644
@@ -23,10 +23,15 @@ static const struct regmap_config st_lsm6dsx_i2c_regmap_config = {
 
 static int st_lsm6dsx_i2c_probe(struct i2c_client *client)
 {
-       const struct i2c_device_id *id = i2c_client_get_device_id(client);
-       int hw_id = id->driver_data;
+       int hw_id;
        struct regmap *regmap;
 
+       hw_id = (kernel_ulong_t)device_get_match_data(&client->dev);
+       if (!hw_id)
+               hw_id = i2c_client_get_device_id(client)->driver_data;
+       if (!hw_id)
+               return -EINVAL;
+
        regmap = devm_regmap_init_i2c(client, &st_lsm6dsx_i2c_regmap_config);
        if (IS_ERR(regmap)) {
                dev_err(&client->dev, "Failed to register i2c regmap %ld\n", PTR_ERR(regmap));
@@ -129,6 +134,10 @@ static const struct of_device_id st_lsm6dsx_i2c_of_match[] = {
 };
 MODULE_DEVICE_TABLE(of, st_lsm6dsx_i2c_of_match);
 
+static const struct acpi_device_id st_lsm6dsx_i2c_acpi_match[] = {
+       { "SMO8B30", ST_LSM6DS3TRC_ID, },
+};
+
 static const struct i2c_device_id st_lsm6dsx_i2c_id_table[] = {
        { ST_LSM6DS3_DEV_NAME, ST_LSM6DS3_ID },
        { ST_LSM6DS3H_DEV_NAME, ST_LSM6DS3H_ID },
@@ -161,6 +170,7 @@ static struct i2c_driver st_lsm6dsx_driver = {
                .name = "st_lsm6dsx_i2c",
                .pm = pm_sleep_ptr(&st_lsm6dsx_pm_ops),
                .of_match_table = st_lsm6dsx_i2c_of_match,
+               .acpi_match_table = st_lsm6dsx_i2c_acpi_match,
        },
        .probe_new = st_lsm6dsx_i2c_probe,
        .id_table = st_lsm6dsx_i2c_id_table,
Quoted 35 lines
Quoted 18 lines

The device is identifying itself to the kernel with PNP id SMO8B30:
physical_node:
modalias=acpi:SMO8B30:SMO8B30:
name=SMO8B30:00
uevent=MODALIAS=acpi:SMO8B30:SMO8B30:
waiting_for_supplier=0
firmware_node:
hid=SMO8B30
modalias=acpi:SMO8B30:SMO8B30:
path=\_SB_.PCI0.I2C5.DEV_
status=15
uevent=MODALIAS=acpi:SMO8B30:SMO8B30:
uid=0

The kernel module for the appropriate driver (st_lsm6dsx_i2c) is not

loaded on boot.

Quoted 2 lines

Modprobing it does not associate it with the device, as I would expect as
the module does not provide an alias for the above acpi/pnp id.

Can you report this issue upstream? Gues to reach out are according to get_maintainers.pl script:

Lorenzo Bianconi <[email protected]> (maintainer:ST LSM6DSx IMU IIO DRIVER) Jonathan Cameron <[email protected]> (maintainer:IIO SUBSYSTEM AND DRIVERS) Lars-Peter Clausen <[email protected]> (reviewer:IIO SUBSYSTEM AND DRIVERS) [email protected] (open list:ST LSM6DSx IMU IIO DRIVER) [email protected] (open list)

Please keep us in the loop.

Regards,
Salvatore

#16 lore raw
Quoted 3 lines · On Wed, 1 Feb 2023 01:46:11 +0000 Darrell Kavanagh…

I should also add that I'm more than willing to help in testing any further changes for you, whether for this bug or future ones.

Thanks for the excellent reporting and testing! I'll definitely keep you in mind for this driver and in particular ACPI side of things.

Jonathan

Quoted 438 lines

Darrell

Quoted 434 lines · Darrell Kavanagh

Hello, all.

I've finally reached a conclusion on this, after testing all the combinations of the patches (with and without reading the acpi mounting matrix), window managers (wayland, xorg) and the presence or not of my custom kernel parms.

What works well is the full set of patches with the custom kernel parms and a new hwdb entry for the sensor:

sensor:modalias:acpi:SMO8B30*:dmi:*:svnLENOVO*:pn82AT:*
ACCEL_MOUNT_MATRIX=0, 1, 0; -1, 0, 0; 0, 0, 1

The autorotate then works correctly in wayland and xorg, but for xorg, the settings say the screen is "portrait left" when in actual fact it is in standard laptop landscape orientation. Wayland does not have this problem (I guess because wayland's view of the screen is straight from the kernel).

Without the hwdb entry, the orientation is 90 degrees out without using the acpi matrix and 180 degrees out when using it. I could have gone either way here with appropriate hwdb entries, but my view is that we *should* be using the matrix.

BTW, when I reported before that it was working correctly in Gnome, I was using a third party shell extension to force auto-rotation rather than actually detaching the keyboard. I take the view that if an extension works differently to the built-in tablet mode, it's the extension's problem!

The only thing that concerns me is the need for custom kernel parms. It would be better if there was a way to avoid this, so that the user didn't have to mess around with their grub config. Though having said that, the sensors fix as we have it doesn't make things worse - under currently released kernels the screen always starts up sideways unless custom parms are added in grub.

Without the custom parms, the patches and the new hwdb entry do ensure that things end up the right way around, but it's a bit clunky. The boot messages/splash is sideways and you can see the sensors kicking in on the gdm screen: it shows momentarily sideways before righting itself. This can also cause the scaling to be set to 200%, which can carry through to the user's session.

So thanks for your work on this - I now have a fully functioning device. I hope the changes are robust enough to be pushed into a future kernel so that others can benefit.

Regards,
Darrell

Quoted 379 lines · Darrell Kavanagh

Just a quick update on testing: I'm finding what look like interactions between the kernel parm video=DSI-1:panel_orientation=right_side_up and the sensor, which also seem to differ depending on whether I'm in an X or Wayland session. I'm documenting the effects of the various combinations and will post my results when complete.

Thanks,
Darrell

Quoted 363 lines · Jonathan Cameron
Quoted 15 lines · On Mon, 30 Jan 2023 20:02:31 +0000 Darrell Kavanagh…

Thanks. To be clear, before I changed the grub command line, the system always booted up "sideways", even when the sensor was not being detected. This was true for everything, not just Gnome, except grub itself.

I added:

GRUB_CMDLINE_LINUX_DEFAULT="fbcon=rotate:1 video=DSI-1:panel_orientation=right_side_up quiet splash"

This fixed the orientation for pre-splash boot messages, the splash screen and the desktop environment. But not, for example (as I saw after adding my own module signing key for testing your fixes), the MOK validation screens.

Quoted 4 lines

Does this make sense?

Does what you are proposing act at a lower level than changing the systemd hwdb orientation matrix?

I'm not sure on the userspace side of things, but intent is that it will provide the orientation data to any users - though only after the kernel boots and software needs to be aware of it. Give it a go, and if not Bastien (IIRC wrote iio-sensor-proxy) may be able to advise.

For Bastien - patches for kernel side are:
https://lore.kernel.org/linux-iio/[email protected]/T/#t

Darrell is going to test them after back porting to 6.1. With the first patch he gets the right result in gnome but we weren't picking up the rotation matrix at that point (ROTM in ACPI).

Thanks,

Jonathan

Quoted 322 lines

Thanks,
Darrell

Quoted 317 lines · Jonathan Cameron
Quoted 5 lines · On Mon, 30 Jan 2023 18:32:02 +0000 Darrell Kavanagh…
Quoted 2 lines · Jonathan Cameron

That certainly looks like suitable matrix.

Quoted 7 lines
Quoted 4 lines

If we are lucky it matches the handling in bmc150-accel-core.c for an identically named method. That is going to swap the x and y axis which is I'd have thought would be rather bad if what you have is currently working well.

Actually, that would be good, because at present I have to rotate 90 degrees in my grub command line.

:)

I'll see if I can roll a suitable patch.

Jonathan

Quoted 294 lines

Darrell

Quoted 291 lines
Quoted 290 lines
            }

            Method (PRIM, 0, NotSerialized)
            {
                Name (RBUF, Buffer (One)
                {
                     0x01                                             // .
                })
                Return (RBUF) /* \_SB_.PCI0.I2C5.DEV_.PRIM.RBUF */
            }

            Method (_STA, 0, NotSerialized)  // _STA: Status
            {
                If ((GAVT == 0x6A))
                {
                    Return (0x0F)
                }
                Else
                {
                    Return (Zero)
                }
            }

            Method (CALS, 1, NotSerialized)
            {
                Local0 = Arg0
                If (((Local0 == Zero) || (Local0 == Ones)))
                {
                    Local0 = BAC1 /* \BAC1 */
                    Return (Local0)
                }
                Else
                {
                    BAC1 = Local0
                    BACS = Local0
                    BSCA (0xB0)
                }
            }
        }
    }

Thanks,
Darrell

Quoted 243 lines · Jonathan Cameron
Quoted 24 lines · On Mon, 30 Jan 2023 03:37:23 +0000 Darrell Kavanagh…

Forwarding because original html messages were rejected by the server...

---------- Forwarded message ---------
From: Darrell Kavanagh <[email protected]>
Date: Mon, 30 Jan 2023 at 02:52
Subject: Re: Bug#1029850: linux: Driver not loaded for ST Microelectronics LSM6DS3TR-C accelerometer (acpi:SMO8B30:SMO8B30:)
To: Jonathan Cameron <[email protected]>
Cc: <[email protected]>, <[email protected]>, <[email protected]>, <[email protected]>, <[email protected]>

Hi Jonathan,

Thank you. The driver has evolved quite a bit in 6.2 (I read somewhere that 6.2 includes some i2c enhancements), but I adapted your changes to fit my Debian 6.1 kernel and it works. Two IIO devices are created in sysfs, iio-sensor-proxy.service starts up and automatic screen rotation in Gnome just works.

To get the modules to load on boot, I made a small change to your code in st_lsm6dsx_i2c to add the acpi alias to modules.alias:

Quoted 7 lines

adding a null element to st_lsm6dsx_i2c_acpi_match:

    static const struct acpi_device_id st_lsm6dsx_i2c_acpi_match[] = {
         { "SMO8B30", ST_LSM6DS3TRC_ID, },
         { },
    };
then:
   MODULE_DEVICE_TABLE(acpi, st_lsm6dsx_i2c_acpi_match);  

doh! That was indeed sloppy of me to miss even for an untested hack.

Quoted 4 lines

dmesg shows:

Quoted 8 lines
[ 7366.120208] st_lsm6dsx_i2c i2c-SMO8B30:00: supply vdd not found,
using dummy regulator
[ 7366.120260] st_lsm6dsx_i2c i2c-SMO8B30:00: supply vddio not found,
using dummy regulator
[ 7366.650839] st_lsm6dsx_i2c i2c-SMO8B30:00: mounting matrix not

found: using identity...

Is this a problem?

Those are all fine. For regulators that's expected on ACPI and should be harmless as it's up to the firmware to manage power (in DT it may be up to the kernel).
For the mounting matrix, there is often something in ACPI DSDT (non standard though). Could you
cat /sys/firmware/acpi/tables/DSDT > ~/dsdt
then run through iasl from acpitools
iasl -d ~/dsdt
and find the bit related to this device.

If you can then share that there may be a _DSM or similar in there that is effectively the mounting matrix. If we are lucky it will look like some existing versions we have code to handle and can add that support as well.

Either way - I'll spin a formal patch with your fixes above and we can get this upstream for future kernels. Mounting matrix can follow later if needed.

Thanks,

Jonathan

Quoted 165 lines

Thanks again.

Darrell

Quoted 158 lines · Jonathan Cameron
Quoted 5 lines · On Sun, 29 Jan 2023 17:03:51 +0000 Darrell Kavanagh…

Hi,

I raised this bug in Debian, and have been asked to raise it upstream and
was given your addresses to do so. Will this email be OK, or should I raise
it in a bug tracking system somewhere?

Email is the right option.

Quoted 36 lines

Many thanks,
Darrell

---------- Forwarded message ---------
From: Salvatore Bonaccorso <[email protected]>
Date: Sat, 28 Jan 2023 at 20:33
Subject: Re: Bug#1029850: linux: Driver not loaded for ST Microelectronics
LSM6DS3TR-C accelerometer (acpi:SMO8B30:SMO8B30:)
To: Darrell Kavanagh <[email protected]>, <[email protected]>

Hi Darrell,

On Sat, Jan 28, 2023 at 08:13:16PM +0000, Darrell Kavanagh wrote:
Package: src:linux
Version: 6.1.4-1
Severity: normal
File: linux
X-Debbugs-Cc: [email protected]

Dear Maintainer,

This is a convertable touchscreen tablet/laptop. The rotation sensor
device
ST Microelectronics LSM6DS3TR-C does not work. It is detected via ACPI
and the
sysfs trees are created at
devices/pci0000:00/0000:00:17.1/i2c_designware.3/i2c-4/i2c-SMO8B30:00
and devices/LNXSYSTM:00/LNXSYBUS:00/PNP0A08:00/device:3c/SMO8B30:00 with
symlinks bus/acpi/devices/SMO8B30:00 and bus/i2c/devices/i2c-SMO8B30:00,
but
no driver is loaded.

At least this is using the ST PNP ID which is better than average (long story!)

The driver in question (ultimately drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c does not currently have an ACPI support. It should be straight forwards to add though the driver first needs converting to use device_get_match_data() with appropriate fallback so that it will match on ACPI, OF or original spi_device_id tables

Completely untested but something like the following
(the offset in the enum is needed to allow us to tell if we got a result when calling device_get_match_data() as it returns NULL on failure IIRC)

I'm not sure how sucessful the driver will be at finding any interrupts etc, but it may get you basic functionality.

Good luck and others more familiar with the driver may well tell me what I forgot when hacking the below ;)

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx.h +1 -1
index 499fcf8875b4..2617ce236ddc 100644
@@ -39,7 +39,7 @@
 #define ST_ISM330IS_DEV_NAME   "ism330is"
 
 enum st_lsm6dsx_hw_id {
-       ST_LSM6DS3_ID,
+       ST_LSM6DS3_ID = 1,
        ST_LSM6DS3H_ID,
        ST_LSM6DSL_ID,
        ST_LSM6DSM_ID,
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c +12 -2
index df5f60925260..ecfceb2fb3db 100644
@@ -23,10 +23,15 @@ static const struct regmap_config st_lsm6dsx_i2c_regmap_config = {
 
 static int st_lsm6dsx_i2c_probe(struct i2c_client *client)
 {
-       const struct i2c_device_id *id = i2c_client_get_device_id(client);
-       int hw_id = id->driver_data;
+       int hw_id;
        struct regmap *regmap;
 
+       hw_id = (kernel_ulong_t)device_get_match_data(&client->dev);
+       if (!hw_id)
+               hw_id = i2c_client_get_device_id(client)->driver_data;
+       if (!hw_id)
+               return -EINVAL;
+
        regmap = devm_regmap_init_i2c(client, &st_lsm6dsx_i2c_regmap_config);
        if (IS_ERR(regmap)) {
                dev_err(&client->dev, "Failed to register i2c regmap %ld\n", PTR_ERR(regmap));
@@ -129,6 +134,10 @@ static const struct of_device_id st_lsm6dsx_i2c_of_match[] = {
 };
 MODULE_DEVICE_TABLE(of, st_lsm6dsx_i2c_of_match);
 
+static const struct acpi_device_id st_lsm6dsx_i2c_acpi_match[] = {
+       { "SMO8B30", ST_LSM6DS3TRC_ID, },
+};
+
 static const struct i2c_device_id st_lsm6dsx_i2c_id_table[] = {
        { ST_LSM6DS3_DEV_NAME, ST_LSM6DS3_ID },
        { ST_LSM6DS3H_DEV_NAME, ST_LSM6DS3H_ID },
@@ -161,6 +170,7 @@ static struct i2c_driver st_lsm6dsx_driver = {
                .name = "st_lsm6dsx_i2c",
                .pm = pm_sleep_ptr(&st_lsm6dsx_pm_ops),
                .of_match_table = st_lsm6dsx_i2c_of_match,
+               .acpi_match_table = st_lsm6dsx_i2c_acpi_match,
        },
        .probe_new = st_lsm6dsx_i2c_probe,
        .id_table = st_lsm6dsx_i2c_id_table,
Quoted 35 lines

The device is identifying itself to the kernel with PNP id SMO8B30:
physical_node:
modalias=acpi:SMO8B30:SMO8B30:
name=SMO8B30:00
uevent=MODALIAS=acpi:SMO8B30:SMO8B30:
waiting_for_supplier=0
firmware_node:
hid=SMO8B30
modalias=acpi:SMO8B30:SMO8B30:
path=\_SB_.PCI0.I2C5.DEV_
status=15
uevent=MODALIAS=acpi:SMO8B30:SMO8B30:
uid=0

The kernel module for the appropriate driver (st_lsm6dsx_i2c) is not
loaded on boot.
Modprobing it does not associate it with the device, as I would expect as
the module does not provide an alias for the above acpi/pnp id.

Can you report this issue upstream? Gues to reach out are according to
get_maintainers.pl script:

Lorenzo Bianconi <[email protected]> (maintainer:ST LSM6DSx IMU IIO DRIVER)
Jonathan Cameron <[email protected]> (maintainer:IIO SUBSYSTEM AND DRIVERS)
Lars-Peter Clausen <[email protected]> (reviewer:IIO SUBSYSTEM AND DRIVERS)
[email protected] (open list:ST LSM6DSx IMU IIO DRIVER)
[email protected] (open list)

Please keep us in the loop.

Regards,
Salvatore

#17 lore raw
In reply to Darrell Kavanagh
Quoted 19 lines · On Wed, 1 Feb 2023 01:40:49 +0000 Darrell Kavanagh…

Hello, all.

I've finally reached a conclusion on this, after testing all the combinations of the patches (with and without reading the acpi mounting matrix), window managers (wayland, xorg) and the presence or not of my custom kernel parms.

What works well is the full set of patches with the custom kernel parms and a new hwdb entry for the sensor:

sensor:modalias:acpi:SMO8B30*:dmi:*:svnLENOVO*:pn82AT:*
ACCEL_MOUNT_MATRIX=0, 1, 0; -1, 0, 0; 0, 0, 1

The autorotate then works correctly in wayland and xorg, but for xorg, the settings say the screen is "portrait left" when in actual fact it is in standard laptop landscape orientation. Wayland does not have this problem (I guess because wayland's view of the screen is straight from the kernel).

Quoted 4 lines

Without the hwdb entry, the orientation is 90 degrees out without using the acpi matrix and 180 degrees out when using it. I could have gone either way here with appropriate hwdb entries, but my view is that we *should* be using the matrix.

Added Hans de Goede as he has probably run into more of this mess than anyone else. Hans, any thoughts on if we are doing something wrong on kernel side? Or is the matrix just wrong *sigh* I think 'ROTM' is defined by MS.
https://learn.microsoft.com/en-us/windows-hardware/drivers/sensors/sensors-acpi-entries

Quoted 21 lines

BTW, when I reported before that it was working correctly in Gnome, I was using a third party shell extension to force auto-rotation rather than actually detaching the keyboard. I take the view that if an extension works differently to the built-in tablet mode, it's the extension's problem!

The only thing that concerns me is the need for custom kernel parms. It would be better if there was a way to avoid this, so that the user didn't have to mess around with their grub config. Though having said that, the sensors fix as we have it doesn't make things worse - under currently released kernels the screen always starts up sideways unless custom parms are added in grub.

Without the custom parms, the patches and the new hwdb entry do ensure that things end up the right way around, but it's a bit clunky. The boot messages/splash is sideways and you can see the sensors kicking in on the gdm screen: it shows momentarily sideways before righting itself. This can also cause the scaling to be set to 200%, which can carry through to the user's session.

Quoted 3 lines

So thanks for your work on this - I now have a fully functioning device. I hope the changes are robust enough to be pushed into a future kernel so that others can benefit.

I'll certainly pick up the ID patch, but I'm dubious about the mount matrix unless we manage to establish why it isn't directly giving us the right answer (and what the solution should be).

Jonathan

Quoted 386 lines

Regards,
Darrell

Quoted 379 lines · Darrell Kavanagh

Just a quick update on testing: I'm finding what look like interactions between the kernel parm video=DSI-1:panel_orientation=right_side_up and the sensor, which also seem to differ depending on whether I'm in an X or Wayland session. I'm documenting the effects of the various combinations and will post my results when complete.

Thanks,
Darrell

Quoted 363 lines · Jonathan Cameron
Quoted 15 lines · On Mon, 30 Jan 2023 20:02:31 +0000 Darrell Kavanagh…

Thanks. To be clear, before I changed the grub command line, the system always booted up "sideways", even when the sensor was not being detected. This was true for everything, not just Gnome, except grub itself.

I added:

GRUB_CMDLINE_LINUX_DEFAULT="fbcon=rotate:1 video=DSI-1:panel_orientation=right_side_up quiet splash"

This fixed the orientation for pre-splash boot messages, the splash screen and the desktop environment. But not, for example (as I saw after adding my own module signing key for testing your fixes), the MOK validation screens.

Quoted 4 lines

Does this make sense?

Does what you are proposing act at a lower level than changing the systemd hwdb orientation matrix?

I'm not sure on the userspace side of things, but intent is that it will provide the orientation data to any users - though only after the kernel boots and software needs to be aware of it. Give it a go, and if not Bastien (IIRC wrote iio-sensor-proxy) may be able to advise.

For Bastien - patches for kernel side are:
https://lore.kernel.org/linux-iio/[email protected]/T/#t

Darrell is going to test them after back porting to 6.1. With the first patch he gets the right result in gnome but we weren't picking up the rotation matrix at that point (ROTM in ACPI).

Thanks,

Jonathan

Quoted 322 lines

Thanks,
Darrell

Quoted 317 lines · Jonathan Cameron
Quoted 5 lines · On Mon, 30 Jan 2023 18:32:02 +0000 Darrell Kavanagh…
Quoted 2 lines · Jonathan Cameron

That certainly looks like suitable matrix.

Quoted 7 lines
Quoted 4 lines

If we are lucky it matches the handling in bmc150-accel-core.c for an identically named method. That is going to swap the x and y axis which is I'd have thought would be rather bad if what you have is currently working well.

Actually, that would be good, because at present I have to rotate 90 degrees in my grub command line.

:)

I'll see if I can roll a suitable patch.

Jonathan

Quoted 294 lines

Darrell

Quoted 291 lines
Quoted 290 lines
            }

            Method (PRIM, 0, NotSerialized)
            {
                Name (RBUF, Buffer (One)
                {
                     0x01                                             // .
                })
                Return (RBUF) /* \_SB_.PCI0.I2C5.DEV_.PRIM.RBUF */
            }

            Method (_STA, 0, NotSerialized)  // _STA: Status
            {
                If ((GAVT == 0x6A))
                {
                    Return (0x0F)
                }
                Else
                {
                    Return (Zero)
                }
            }

            Method (CALS, 1, NotSerialized)
            {
                Local0 = Arg0
                If (((Local0 == Zero) || (Local0 == Ones)))
                {
                    Local0 = BAC1 /* \BAC1 */
                    Return (Local0)
                }
                Else
                {
                    BAC1 = Local0
                    BACS = Local0
                    BSCA (0xB0)
                }
            }
        }
    }

Thanks,
Darrell

Quoted 243 lines · Jonathan Cameron
Quoted 24 lines · On Mon, 30 Jan 2023 03:37:23 +0000 Darrell Kavanagh…

Forwarding because original html messages were rejected by the server...

---------- Forwarded message ---------
From: Darrell Kavanagh <[email protected]>
Date: Mon, 30 Jan 2023 at 02:52
Subject: Re: Bug#1029850: linux: Driver not loaded for ST Microelectronics LSM6DS3TR-C accelerometer (acpi:SMO8B30:SMO8B30:)
To: Jonathan Cameron <[email protected]>
Cc: <[email protected]>, <[email protected]>, <[email protected]>, <[email protected]>, <[email protected]>

Hi Jonathan,

Thank you. The driver has evolved quite a bit in 6.2 (I read somewhere that 6.2 includes some i2c enhancements), but I adapted your changes to fit my Debian 6.1 kernel and it works. Two IIO devices are created in sysfs, iio-sensor-proxy.service starts up and automatic screen rotation in Gnome just works.

To get the modules to load on boot, I made a small change to your code in st_lsm6dsx_i2c to add the acpi alias to modules.alias:

Quoted 7 lines

adding a null element to st_lsm6dsx_i2c_acpi_match:

    static const struct acpi_device_id st_lsm6dsx_i2c_acpi_match[] = {
         { "SMO8B30", ST_LSM6DS3TRC_ID, },
         { },
    };
then:
   MODULE_DEVICE_TABLE(acpi, st_lsm6dsx_i2c_acpi_match);  

doh! That was indeed sloppy of me to miss even for an untested hack.

Quoted 4 lines

dmesg shows:

Quoted 8 lines
[ 7366.120208] st_lsm6dsx_i2c i2c-SMO8B30:00: supply vdd not found,
using dummy regulator
[ 7366.120260] st_lsm6dsx_i2c i2c-SMO8B30:00: supply vddio not found,
using dummy regulator
[ 7366.650839] st_lsm6dsx_i2c i2c-SMO8B30:00: mounting matrix not

found: using identity...

Is this a problem?

Those are all fine. For regulators that's expected on ACPI and should be harmless as it's up to the firmware to manage power (in DT it may be up to the kernel).
For the mounting matrix, there is often something in ACPI DSDT (non standard though). Could you
cat /sys/firmware/acpi/tables/DSDT > ~/dsdt
then run through iasl from acpitools
iasl -d ~/dsdt
and find the bit related to this device.

If you can then share that there may be a _DSM or similar in there that is effectively the mounting matrix. If we are lucky it will look like some existing versions we have code to handle and can add that support as well.

Either way - I'll spin a formal patch with your fixes above and we can get this upstream for future kernels. Mounting matrix can follow later if needed.

Thanks,

Jonathan

Quoted 165 lines

Thanks again.

Darrell

Quoted 158 lines · Jonathan Cameron
Quoted 5 lines · On Sun, 29 Jan 2023 17:03:51 +0000 Darrell Kavanagh…

Hi,

I raised this bug in Debian, and have been asked to raise it upstream and was given your addresses to do so. Will this email be OK, or should I raise it in a bug tracking system somewhere?

Email is the right option.

Quoted 27 lines

Many thanks,
Darrell

---------- Forwarded message ---------
From: Salvatore Bonaccorso <[email protected]>
Date: Sat, 28 Jan 2023 at 20:33
Subject: Re: Bug#1029850: linux: Driver not loaded for ST Microelectronics LSM6DS3TR-C accelerometer (acpi:SMO8B30:SMO8B30:)
To: Darrell Kavanagh <[email protected]>, <[email protected]>

Hi Darrell,

Quoted 9 lines · Darrell Kavanagh

Package: src:linux
Version: 6.1.4-1
Severity: normal
File: linux
X-Debbugs-Cc: [email protected]

Dear Maintainer,

This is a convertable touchscreen tablet/laptop. The rotation sensor

Quoted 9 lines

device

Quoted 1 line

ST Microelectronics LSM6DS3TR-C does not work. It is detected via ACPI

and the

Quoted 1 line

sysfs trees are created at

devices/pci0000:00/0000:00:17.1/i2c_designware.3/i2c-4/i2c-SMO8B30:00

Quoted 2 lines

and devices/LNXSYSTM:00/LNXSYBUS:00/PNP0A08:00/device:3c/SMO8B30:00 with
symlinks bus/acpi/devices/SMO8B30:00 and bus/i2c/devices/i2c-SMO8B30:00,

but

Quoted 1 line

no driver is loaded.

At least this is using the ST PNP ID which is better than average (long story!)

The driver in question (ultimately drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c does not currently have an ACPI support. It should be straight forwards to add though the driver first needs converting to use device_get_match_data() with appropriate fallback so that it will match on ACPI, OF or original spi_device_id tables

Completely untested but something like the following
(the offset in the enum is needed to allow us to tell if we got a result when calling device_get_match_data() as it returns NULL on failure IIRC)

I'm not sure how sucessful the driver will be at finding any interrupts etc, but it may get you basic functionality.

Good luck and others more familiar with the driver may well tell me what I forgot when hacking the below ;)

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx.h +1 -1
index 499fcf8875b4..2617ce236ddc 100644
@@ -39,7 +39,7 @@
 #define ST_ISM330IS_DEV_NAME   "ism330is"
 
 enum st_lsm6dsx_hw_id {
-       ST_LSM6DS3_ID,
+       ST_LSM6DS3_ID = 1,
        ST_LSM6DS3H_ID,
        ST_LSM6DSL_ID,
        ST_LSM6DSM_ID,
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c +12 -2
index df5f60925260..ecfceb2fb3db 100644
@@ -23,10 +23,15 @@ static const struct regmap_config st_lsm6dsx_i2c_regmap_config = {
 
 static int st_lsm6dsx_i2c_probe(struct i2c_client *client)
 {
-       const struct i2c_device_id *id = i2c_client_get_device_id(client);
-       int hw_id = id->driver_data;
+       int hw_id;
        struct regmap *regmap;
 
+       hw_id = (kernel_ulong_t)device_get_match_data(&client->dev);
+       if (!hw_id)
+               hw_id = i2c_client_get_device_id(client)->driver_data;
+       if (!hw_id)
+               return -EINVAL;
+
        regmap = devm_regmap_init_i2c(client, &st_lsm6dsx_i2c_regmap_config);
        if (IS_ERR(regmap)) {
                dev_err(&client->dev, "Failed to register i2c regmap %ld\n", PTR_ERR(regmap));
@@ -129,6 +134,10 @@ static const struct of_device_id st_lsm6dsx_i2c_of_match[] = {
 };
 MODULE_DEVICE_TABLE(of, st_lsm6dsx_i2c_of_match);
 
+static const struct acpi_device_id st_lsm6dsx_i2c_acpi_match[] = {
+       { "SMO8B30", ST_LSM6DS3TRC_ID, },
+};
+
 static const struct i2c_device_id st_lsm6dsx_i2c_id_table[] = {
        { ST_LSM6DS3_DEV_NAME, ST_LSM6DS3_ID },
        { ST_LSM6DS3H_DEV_NAME, ST_LSM6DS3H_ID },
@@ -161,6 +170,7 @@ static struct i2c_driver st_lsm6dsx_driver = {
                .name = "st_lsm6dsx_i2c",
                .pm = pm_sleep_ptr(&st_lsm6dsx_pm_ops),
                .of_match_table = st_lsm6dsx_i2c_of_match,
+               .acpi_match_table = st_lsm6dsx_i2c_acpi_match,
        },
        .probe_new = st_lsm6dsx_i2c_probe,
        .id_table = st_lsm6dsx_i2c_id_table,
Quoted 35 lines
Quoted 18 lines

The device is identifying itself to the kernel with PNP id SMO8B30:
physical_node:
modalias=acpi:SMO8B30:SMO8B30:
name=SMO8B30:00
uevent=MODALIAS=acpi:SMO8B30:SMO8B30:
waiting_for_supplier=0
firmware_node:
hid=SMO8B30
modalias=acpi:SMO8B30:SMO8B30:
path=\_SB_.PCI0.I2C5.DEV_
status=15
uevent=MODALIAS=acpi:SMO8B30:SMO8B30:
uid=0

The kernel module for the appropriate driver (st_lsm6dsx_i2c) is not

loaded on boot.

Quoted 2 lines

Modprobing it does not associate it with the device, as I would expect as
the module does not provide an alias for the above acpi/pnp id.

Can you report this issue upstream? Gues to reach out are according to get_maintainers.pl script:

Lorenzo Bianconi <[email protected]> (maintainer:ST LSM6DSx IMU IIO DRIVER) Jonathan Cameron <[email protected]> (maintainer:IIO SUBSYSTEM AND DRIVERS) Lars-Peter Clausen <[email protected]> (reviewer:IIO SUBSYSTEM AND DRIVERS) [email protected] (open list:ST LSM6DSx IMU IIO DRIVER) [email protected] (open list)

Please keep us in the loop.

Regards,
Salvatore

#18 lore raw

Hi,

Quoted 22 lines · Jonathan Cameron
Quoted 19 lines · On Wed, 1 Feb 2023 01:40:49 +0000 Darrell Kavanagh…

Hello, all.

I've finally reached a conclusion on this, after testing all the combinations of the patches (with and without reading the acpi mounting matrix), window managers (wayland, xorg) and the presence or not of my custom kernel parms.

What works well is the full set of patches with the custom kernel parms and a new hwdb entry for the sensor:

sensor:modalias:acpi:SMO8B30*:dmi:*:svnLENOVO*:pn82AT:*
ACCEL_MOUNT_MATRIX=0, 1, 0; -1, 0, 0; 0, 0, 1

The autorotate then works correctly in wayland and xorg, but for xorg, the settings say the screen is "portrait left" when in actual fact it is in standard laptop landscape orientation. Wayland does not have this problem (I guess because wayland's view of the screen is straight from the kernel).

Quoted 8 lines
Quoted 4 lines

Without the hwdb entry, the orientation is 90 degrees out without using the acpi matrix and 180 degrees out when using it. I could have gone either way here with appropriate hwdb entries, but my view is that we *should* be using the matrix.

Added Hans de Goede as he has probably run into more of this mess than anyone else. Hans, any thoughts on if we are doing something wrong on kernel side? Or is the matrix just wrong *sigh*

I see below that this laptop has a panel which is mounted 90 degrees rotated, that likely explains why the ACPI matrix does not work. So the best thing to do here is to just override it with a hwdb entries.

IIRC there are already 1 or 2 other hwdb entries which actually override the ACPI provided matrix because of similar issues.

Linux userspace expects the matrix in this case to be set so that it causes e.g. gnome's auto-rotation to put the image upright even with older gnome versions / mate / xfce which don't know about the panel being mounted 90 degrees.

So e.g. "monitor-sensor" will report left-side-up or right-side-up while the device is actually in normal clamshell mode with the display up-right.

This reporting of left-side-up or right-side-up is actually "correct" looking from the native LCD panel orientation and as mentioned is done for backward compatibility. This is documented here:

https://github.com/systemd/systemd/blob/main/hwdb.d/60-sensor.hwdb#L54

The way we are handling this is likely incompatible with how Windows handles this special case of 90° rotated screen + ROTM. Or the matrix in the ACPI tables could be just wrong...

Quoted 2 lines

I think 'ROTM' is defined by MS.
https://learn.microsoft.com/en-us/windows-hardware/drivers/sensors/sensors-acpi-entries

Right and as such it would be good if we can still add support to it to the sensor driver in question. Because the ROTM info usually is correct and avoids the need for adding more and more hwdb entries.

Note there already is existing support in some other sensor drivers.

So we probably need to factor out some helper code for this and share that between sensor drivers.

Quoted 6 lines
Quoted 6 lines

The only thing that concerns me is the need for custom kernel parms. It would be better if there was a way to avoid this, so that the user didn't have to mess around with their grub config. Though having said that, the sensors fix as we have it doesn't make things worse - under currently released kernels the screen always starts up sideways unless custom parms are added in grub.

We actually have a quirk mechanism in the kernel for specifying the need for: video=DSI-1:panel_orientation=right_side_up and this will also automatically fix the fbcon orientation, see:

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/gpu/drm/drm_panel_orientation_quirks.c

If you submit a patch for this upstream please Cc me.

Regards,

Hans

Quoted 365 lines
Quoted 364 lines
Quoted 364 lines · Jonathan Cameron
Quoted 363 lines
Quoted 15 lines · On Mon, 30 Jan 2023 20:02:31 +0000 Darrell Kavanagh…

Thanks. To be clear, before I changed the grub command line, the system always booted up "sideways", even when the sensor was not being detected. This was true for everything, not just Gnome, except grub itself.

I added:

GRUB_CMDLINE_LINUX_DEFAULT="fbcon=rotate:1 video=DSI-1:panel_orientation=right_side_up quiet splash"

This fixed the orientation for pre-splash boot messages, the splash screen and the desktop environment. But not, for example (as I saw after adding my own module signing key for testing your fixes), the MOK validation screens.

Quoted 4 lines

Does this make sense?

Does what you are proposing act at a lower level than changing the systemd hwdb orientation matrix?

I'm not sure on the userspace side of things, but intent is that it will provide the orientation data to any users - though only after the kernel boots and software needs to be aware of it. Give it a go, and if not Bastien (IIRC wrote iio-sensor-proxy) may be able to advise.

For Bastien - patches for kernel side are:
https://lore.kernel.org/linux-iio/[email protected]/T/#t

Darrell is going to test them after back porting to 6.1. With the first patch he gets the right result in gnome but we weren't picking up the rotation matrix at that point (ROTM in ACPI).

Thanks,

Jonathan

Quoted 322 lines

Thanks,
Darrell

Quoted 317 lines · Jonathan Cameron
Quoted 5 lines · On Mon, 30 Jan 2023 18:32:02 +0000 Darrell Kavanagh…
Quoted 2 lines · Jonathan Cameron

That certainly looks like suitable matrix.

Quoted 7 lines
Quoted 4 lines

If we are lucky it matches the handling in bmc150-accel-core.c for an identically named method. That is going to swap the x and y axis which is I'd have thought would be rather bad if what you have is currently working well.

Actually, that would be good, because at present I have to rotate 90 degrees in my grub command line.

:)

I'll see if I can roll a suitable patch.

Jonathan

Quoted 294 lines

Darrell

Quoted 291 lines
Quoted 290 lines
            }

            Method (PRIM, 0, NotSerialized)
            {
                Name (RBUF, Buffer (One)
                {
                     0x01                                             // .
                })
                Return (RBUF) /* \_SB_.PCI0.I2C5.DEV_.PRIM.RBUF */
            }

            Method (_STA, 0, NotSerialized)  // _STA: Status
            {
                If ((GAVT == 0x6A))
                {
                    Return (0x0F)
                }
                Else
                {
                    Return (Zero)
                }
            }

            Method (CALS, 1, NotSerialized)
            {
                Local0 = Arg0
                If (((Local0 == Zero) || (Local0 == Ones)))
                {
                    Local0 = BAC1 /* \BAC1 */
                    Return (Local0)
                }
                Else
                {
                    BAC1 = Local0
                    BACS = Local0
                    BSCA (0xB0)
                }
            }
        }
    }

Thanks,
Darrell

Quoted 243 lines · Jonathan Cameron
Quoted 24 lines · On Mon, 30 Jan 2023 03:37:23 +0000 Darrell Kavanagh…

Forwarding because original html messages were rejected by the server...

---------- Forwarded message ---------
From: Darrell Kavanagh <[email protected]>
Date: Mon, 30 Jan 2023 at 02:52
Subject: Re: Bug#1029850: linux: Driver not loaded for ST Microelectronics LSM6DS3TR-C accelerometer (acpi:SMO8B30:SMO8B30:)
To: Jonathan Cameron <[email protected]>
Cc: <[email protected]>, <[email protected]>, <[email protected]>, <[email protected]>, <[email protected]>

Hi Jonathan,

Thank you. The driver has evolved quite a bit in 6.2 (I read somewhere that 6.2 includes some i2c enhancements), but I adapted your changes to fit my Debian 6.1 kernel and it works. Two IIO devices are created in sysfs, iio-sensor-proxy.service starts up and automatic screen rotation in Gnome just works.

To get the modules to load on boot, I made a small change to your code in st_lsm6dsx_i2c to add the acpi alias to modules.alias:

Quoted 7 lines

adding a null element to st_lsm6dsx_i2c_acpi_match:

    static const struct acpi_device_id st_lsm6dsx_i2c_acpi_match[] = {
         { "SMO8B30", ST_LSM6DS3TRC_ID, },
         { },
    };
then:
   MODULE_DEVICE_TABLE(acpi, st_lsm6dsx_i2c_acpi_match);  

doh! That was indeed sloppy of me to miss even for an untested hack.

Quoted 4 lines

dmesg shows:

Quoted 8 lines
[ 7366.120208] st_lsm6dsx_i2c i2c-SMO8B30:00: supply vdd not found,
using dummy regulator
[ 7366.120260] st_lsm6dsx_i2c i2c-SMO8B30:00: supply vddio not found,
using dummy regulator
[ 7366.650839] st_lsm6dsx_i2c i2c-SMO8B30:00: mounting matrix not

found: using identity...

Is this a problem?

Those are all fine. For regulators that's expected on ACPI and should be harmless as it's up to the firmware to manage power (in DT it may be up to the kernel).
For the mounting matrix, there is often something in ACPI DSDT (non standard though). Could you
cat /sys/firmware/acpi/tables/DSDT > ~/dsdt
then run through iasl from acpitools
iasl -d ~/dsdt
and find the bit related to this device.

If you can then share that there may be a _DSM or similar in there that is effectively the mounting matrix. If we are lucky it will look like some existing versions we have code to handle and can add that support as well.

Either way - I'll spin a formal patch with your fixes above and we can get this upstream for future kernels. Mounting matrix can follow later if needed.

Thanks,

Jonathan

Quoted 165 lines

Thanks again.

Darrell

Quoted 158 lines · Jonathan Cameron
Quoted 5 lines · On Sun, 29 Jan 2023 17:03:51 +0000 Darrell Kavanagh…

Hi,

I raised this bug in Debian, and have been asked to raise it upstream and
was given your addresses to do so. Will this email be OK, or should I raise
it in a bug tracking system somewhere?

Email is the right option.

Quoted 36 lines

Many thanks,
Darrell

---------- Forwarded message ---------
From: Salvatore Bonaccorso <[email protected]>
Date: Sat, 28 Jan 2023 at 20:33
Subject: Re: Bug#1029850: linux: Driver not loaded for ST Microelectronics
LSM6DS3TR-C accelerometer (acpi:SMO8B30:SMO8B30:)
To: Darrell Kavanagh <[email protected]>, <[email protected]>

Hi Darrell,

On Sat, Jan 28, 2023 at 08:13:16PM +0000, Darrell Kavanagh wrote:
Package: src:linux
Version: 6.1.4-1
Severity: normal
File: linux
X-Debbugs-Cc: [email protected]

Dear Maintainer,

This is a convertable touchscreen tablet/laptop. The rotation sensor
device
ST Microelectronics LSM6DS3TR-C does not work. It is detected via ACPI
and the
sysfs trees are created at
devices/pci0000:00/0000:00:17.1/i2c_designware.3/i2c-4/i2c-SMO8B30:00
and devices/LNXSYSTM:00/LNXSYBUS:00/PNP0A08:00/device:3c/SMO8B30:00 with
symlinks bus/acpi/devices/SMO8B30:00 and bus/i2c/devices/i2c-SMO8B30:00,
but
no driver is loaded.

At least this is using the ST PNP ID which is better than average (long story!)

The driver in question (ultimately drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c does not currently have an ACPI support. It should be straight forwards to add though the driver first needs converting to use device_get_match_data() with appropriate fallback so that it will match on ACPI, OF or original spi_device_id tables

Completely untested but something like the following
(the offset in the enum is needed to allow us to tell if we got a result when calling device_get_match_data() as it returns NULL on failure IIRC)

I'm not sure how sucessful the driver will be at finding any interrupts etc, but it may get you basic functionality.

Good luck and others more familiar with the driver may well tell me what I forgot when hacking the below ;)

drivers/iio/imu/st_lsm6dsx/st_lsm6dsx.h +1 -1
index 499fcf8875b4..2617ce236ddc 100644
@@ -39,7 +39,7 @@
 #define ST_ISM330IS_DEV_NAME   "ism330is"
 
 enum st_lsm6dsx_hw_id {
-       ST_LSM6DS3_ID,
+       ST_LSM6DS3_ID = 1,
        ST_LSM6DS3H_ID,
        ST_LSM6DSL_ID,
        ST_LSM6DSM_ID,
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c +12 -2
index df5f60925260..ecfceb2fb3db 100644
@@ -23,10 +23,15 @@ static const struct regmap_config st_lsm6dsx_i2c_regmap_config = {
 
 static int st_lsm6dsx_i2c_probe(struct i2c_client *client)
 {
-       const struct i2c_device_id *id = i2c_client_get_device_id(client);
-       int hw_id = id->driver_data;
+       int hw_id;
        struct regmap *regmap;
 
+       hw_id = (kernel_ulong_t)device_get_match_data(&client->dev);
+       if (!hw_id)
+               hw_id = i2c_client_get_device_id(client)->driver_data;
+       if (!hw_id)
+               return -EINVAL;
+
        regmap = devm_regmap_init_i2c(client, &st_lsm6dsx_i2c_regmap_config);
        if (IS_ERR(regmap)) {
                dev_err(&client->dev, "Failed to register i2c regmap %ld\n", PTR_ERR(regmap));
@@ -129,6 +134,10 @@ static const struct of_device_id st_lsm6dsx_i2c_of_match[] = {
 };
 MODULE_DEVICE_TABLE(of, st_lsm6dsx_i2c_of_match);
 
+static const struct acpi_device_id st_lsm6dsx_i2c_acpi_match[] = {
+       { "SMO8B30", ST_LSM6DS3TRC_ID, },
+};
+
 static const struct i2c_device_id st_lsm6dsx_i2c_id_table[] = {
        { ST_LSM6DS3_DEV_NAME, ST_LSM6DS3_ID },
        { ST_LSM6DS3H_DEV_NAME, ST_LSM6DS3H_ID },
@@ -161,6 +170,7 @@ static struct i2c_driver st_lsm6dsx_driver = {
                .name = "st_lsm6dsx_i2c",
                .pm = pm_sleep_ptr(&st_lsm6dsx_pm_ops),
                .of_match_table = st_lsm6dsx_i2c_of_match,
+               .acpi_match_table = st_lsm6dsx_i2c_acpi_match,
        },
        .probe_new = st_lsm6dsx_i2c_probe,
        .id_table = st_lsm6dsx_i2c_id_table,
Quoted 35 lines

The device is identifying itself to the kernel with PNP id SMO8B30:
physical_node:
modalias=acpi:SMO8B30:SMO8B30:
name=SMO8B30:00
uevent=MODALIAS=acpi:SMO8B30:SMO8B30:
waiting_for_supplier=0
firmware_node:
hid=SMO8B30
modalias=acpi:SMO8B30:SMO8B30:
path=\_SB_.PCI0.I2C5.DEV_
status=15
uevent=MODALIAS=acpi:SMO8B30:SMO8B30:
uid=0

The kernel module for the appropriate driver (st_lsm6dsx_i2c) is not
loaded on boot.
Modprobing it does not associate it with the device, as I would expect as
the module does not provide an alias for the above acpi/pnp id.

Can you report this issue upstream? Gues to reach out are according to
get_maintainers.pl script:

Lorenzo Bianconi <[email protected]> (maintainer:ST LSM6DSx IMU IIO DRIVER)
Jonathan Cameron <[email protected]> (maintainer:IIO SUBSYSTEM AND DRIVERS)
Lars-Peter Clausen <[email protected]> (reviewer:IIO SUBSYSTEM AND DRIVERS)
[email protected] (open list:ST LSM6DSx IMU IIO DRIVER)
[email protected] (open list)

Please keep us in the loop.

Regards,
Salvatore

#19 lore raw

Thanks for the info, Hans. I was hoping there'd be something like that. Compiling the whole kernel on this little machine will be interesting, though...

Quoted 464 lines · Hans de Goede

Hi,

Quoted 22 lines · Jonathan Cameron
Quoted 19 lines · On Wed, 1 Feb 2023 01:40:49 +0000 Darrell Kavanagh…

Hello, all.

I've finally reached a conclusion on this, after testing all the combinations of the patches (with and without reading the acpi mounting matrix), window managers (wayland, xorg) and the presence or not of my custom kernel parms.

What works well is the full set of patches with the custom kernel parms and a new hwdb entry for the sensor:

sensor:modalias:acpi:SMO8B30*:dmi:*:svnLENOVO*:pn82AT:*
ACCEL_MOUNT_MATRIX=0, 1, 0; -1, 0, 0; 0, 0, 1

The autorotate then works correctly in wayland and xorg, but for xorg, the settings say the screen is "portrait left" when in actual fact it is in standard laptop landscape orientation. Wayland does not have this problem (I guess because wayland's view of the screen is straight from the kernel).

Quoted 8 lines
Quoted 4 lines

Without the hwdb entry, the orientation is 90 degrees out without using the acpi matrix and 180 degrees out when using it. I could have gone either way here with appropriate hwdb entries, but my view is that we *should* be using the matrix.

Added Hans de Goede as he has probably run into more of this mess than anyone else. Hans, any thoughts on if we are doing something wrong on kernel side? Or is the matrix just wrong *sigh*

I see below that this laptop has a panel which is mounted 90 degrees rotated, that likely explains why the ACPI matrix does not work. So the best thing to do here is to just override it with a hwdb entries.

IIRC there are already 1 or 2 other hwdb entries which actually override the ACPI provided matrix because of similar issues.

Linux userspace expects the matrix in this case to be set so that it causes e.g. gnome's auto-rotation to put the image upright even with older gnome versions / mate / xfce which don't know about the panel being mounted 90 degrees.

So e.g. "monitor-sensor" will report left-side-up or right-side-up while the device is actually in normal clamshell mode with the display up-right.

This reporting of left-side-up or right-side-up is actually "correct" looking from the native LCD panel orientation and as mentioned is done for backward compatibility. This is documented here:

https://github.com/systemd/systemd/blob/main/hwdb.d/60-sensor.hwdb#L54

The way we are handling this is likely incompatible with how Windows handles this special case of 90° rotated screen + ROTM. Or the matrix in the ACPI tables could be just wrong...

Quoted 2 lines

I think 'ROTM' is defined by MS.
https://learn.microsoft.com/en-us/windows-hardware/drivers/sensors/sensors-acpi-entries

Right and as such it would be good if we can still add support to it to the sensor driver in question. Because the ROTM info usually is correct and avoids the need for adding more and more hwdb entries.

Note there already is existing support in some other sensor drivers.

So we probably need to factor out some helper code for this and share that between sensor drivers.

Quoted 6 lines
Quoted 6 lines

The only thing that concerns me is the need for custom kernel parms. It would be better if there was a way to avoid this, so that the user didn't have to mess around with their grub config. Though having said that, the sensors fix as we have it doesn't make things worse - under currently released kernels the screen always starts up sideways unless custom parms are added in grub.

We actually have a quirk mechanism in the kernel for specifying the need for: video=DSI-1:panel_orientation=right_side_up and this will also automatically fix the fbcon orientation, see:

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/gpu/drm/drm_panel_orientation_quirks.c

If you submit a patch for this upstream please Cc me.

Regards,

Hans

Quoted 365 lines
Quoted 364 lines
Quoted 364 lines · Jonathan Cameron
Quoted 363 lines
Quoted 15 lines · On Mon, 30 Jan 2023 20:02:31 +0000 Darrell Kavanagh…

Thanks. To be clear, before I changed the grub command line, the system always booted up "sideways", even when the sensor was not being detected. This was true for everything, not just Gnome, except grub itself.

I added:

GRUB_CMDLINE_LINUX_DEFAULT="fbcon=rotate:1 video=DSI-1:panel_orientation=right_side_up quiet splash"

This fixed the orientation for pre-splash boot messages, the splash screen and the desktop environment. But not, for example (as I saw after adding my own module signing key for testing your fixes), the MOK validation screens.

Quoted 4 lines

Does this make sense?

Does what you are proposing act at a lower level than changing the systemd hwdb orientation matrix?

I'm not sure on the userspace side of things, but intent is that it will provide the orientation data to any users - though only after the kernel boots and software needs to be aware of it. Give it a go, and if not Bastien (IIRC wrote iio-sensor-proxy) may be able to advise.

For Bastien - patches for kernel side are:
https://lore.kernel.org/linux-iio/[email protected]/T/#t

Darrell is going to test them after back porting to 6.1. With the first patch he gets the right result in gnome but we weren't picking up the rotation matrix at that point (ROTM in ACPI).

Thanks,

Jonathan

Quoted 322 lines

Thanks,
Darrell

Quoted 317 lines · Jonathan Cameron
Quoted 5 lines · On Mon, 30 Jan 2023 18:32:02 +0000 Darrell Kavanagh…
Quoted 2 lines · Jonathan Cameron

That certainly looks like suitable matrix.

Quoted 7 lines
Quoted 4 lines

If we are lucky it matches the handling in bmc150-accel-core.c for an identically named method. That is going to swap the x and y axis which is I'd have thought would be rather bad if what you have is currently working well.

Actually, that would be good, because at present I have to rotate 90 degrees in my grub command line.

:)

I'll see if I can roll a suitable patch.

Jonathan

Quoted 294 lines

Darrell

Quoted 291 lines
Quoted 290 lines
            }

            Method (PRIM, 0, NotSerialized)
            {
                Name (RBUF, Buffer (One)
                {
                     0x01                                             // .
                })
                Return (RBUF) /* \_SB_.PCI0.I2C5.DEV_.PRIM.RBUF */
            }

            Method (_STA, 0, NotSerialized)  // _STA: Status
            {
                If ((GAVT == 0x6A))
                {
                    Return (0x0F)
                }
                Else
                {
                    Return (Zero)
                }
            }

            Method (CALS, 1, NotSerialized)
            {
                Local0 = Arg0
                If (((Local0 == Zero) || (Local0 == Ones)))
                {
                    Local0 = BAC1 /* \BAC1 */
                    Return (Local0)
                }
                Else
                {
                    BAC1 = Local0
                    BACS = Local0
                    BSCA (0xB0)
                }
            }
        }
    }

Thanks,
Darrell

Quoted 243 lines · Jonathan Cameron
Quoted 24 lines · On Mon, 30 Jan 2023 03:37:23 +0000 Darrell Kavanagh…

Forwarding because original html messages were rejected by the server...

---------- Forwarded message ---------
From: Darrell Kavanagh <[email protected]>
Date: Mon, 30 Jan 2023 at 02:52
Subject: Re: Bug#1029850: linux: Driver not loaded for ST Microelectronics LSM6DS3TR-C accelerometer (acpi:SMO8B30:SMO8B30:)
To: Jonathan Cameron <[email protected]>
Cc: <[email protected]>, <[email protected]>, <[email protected]>, <[email protected]>, <[email protected]>

Hi Jonathan,

Thank you. The driver has evolved quite a bit in 6.2 (I read somewhere that 6.2 includes some i2c enhancements), but I adapted your changes to fit my Debian 6.1 kernel and it works. Two IIO devices are created in sysfs, iio-sensor-proxy.service starts up and automatic screen rotation in Gnome just works.

To get the modules to load on boot, I made a small change to your code in st_lsm6dsx_i2c to add the acpi alias to modules.alias:

Quoted 7 lines

adding a null element to st_lsm6dsx_i2c_acpi_match:

    static const struct acpi_device_id st_lsm6dsx_i2c_acpi_match[] = {
         { "SMO8B30", ST_LSM6DS3TRC_ID, },
         { },
    };
then:
   MODULE_DEVICE_TABLE(acpi, st_lsm6dsx_i2c_acpi_match);

doh! That was indeed sloppy of me to miss even for an untested hack.

Quoted 4 lines

dmesg shows:

Quoted 8 lines
[ 7366.120208] st_lsm6dsx_i2c i2c-SMO8B30:00: supply vdd not found,
using dummy regulator
[ 7366.120260] st_lsm6dsx_i2c i2c-SMO8B30:00: supply vddio not found,
using dummy regulator
[ 7366.650839] st_lsm6dsx_i2c i2c-SMO8B30:00: mounting matrix not

found: using identity...

Is this a problem?

Those are all fine. For regulators that's expected on ACPI and should be harmless as it's up to the firmware to manage power (in DT it may be up to the kernel).
For the mounting matrix, there is often something in ACPI DSDT (non standard though). Could you
cat /sys/firmware/acpi/tables/DSDT > ~/dsdt
then run through iasl from acpitools
iasl -d ~/dsdt
and find the bit related to this device.

If you can then share that there may be a _DSM or similar in there that is effectively the mounting matrix. If we are lucky it will look like some existing versions we have code to handle and can add that support as well.

Either way - I'll spin a formal patch with your fixes above and we can get this upstream for future kernels. Mounting matrix can follow later if needed.

Thanks,

Jonathan

Quoted 165 lines

Thanks again.

Darrell

Quoted 158 lines · Jonathan Cameron

On Sun, 29 Jan 2023 17:03:51 +0000
Darrell Kavanagh <[email protected]> wrote:

Hi,

I raised this bug in Debian, and have been asked to raise it upstream and
was given your addresses to do so. Will this email be OK, or should I raise
it in a bug tracking system somewhere?

Email is the right option.

Many thanks,
Darrell

---------- Forwarded message ---------
From: Salvatore Bonaccorso <[email protected]>
Date: Sat, 28 Jan 2023 at 20:33
Subject: Re: Bug#1029850: linux: Driver not loaded for ST Microelectronics
LSM6DS3TR-C accelerometer (acpi:SMO8B30:SMO8B30:)
To: Darrell Kavanagh <[email protected]>, <[email protected]>

Hi Darrell,

On Sat, Jan 28, 2023 at 08:13:16PM +0000, Darrell Kavanagh wrote:
Package: src:linux
Version: 6.1.4-1
Severity: normal
File: linux
X-Debbugs-Cc: [email protected]

Dear Maintainer,

This is a convertable touchscreen tablet/laptop. The rotation sensor
device
ST Microelectronics LSM6DS3TR-C does not work. It is detected via ACPI
and the
sysfs trees are created at
devices/pci0000:00/0000:00:17.1/i2c_designware.3/i2c-4/i2c-SMO8B30:00
and devices/LNXSYSTM:00/LNXSYBUS:00/PNP0A08:00/device:3c/SMO8B30:00 with
symlinks bus/acpi/devices/SMO8B30:00 and bus/i2c/devices/i2c-SMO8B30:00,
but
no driver is loaded.

At least this is using the ST PNP ID which is better than average
(long story!)

The driver in question (ultimately drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c
does not currently have an ACPI support. It should be straight forwards
to add though the driver first needs converting to use
device_get_match_data() with appropriate fallback so that it will match on
ACPI, OF or original spi_device_id tables

Completely untested but something like the following
(the offset in the enum is needed to allow us to tell if we got a result when
calling device_get_match_data() as it returns NULL on failure IIRC)

I'm not sure how sucessful the driver will be at finding any interrupts etc, but
it may get you basic functionality.

Good luck and others more familiar with the driver may well tell me what I forgot
when hacking the below ;)

diff --git a/drivers/iio/imu/st_lsm6dsx/st_lsm6dsx.h b/drivers/iio/imu/st_lsm6dsx/st_lsm6dsx.h
index 499fcf8875b4..2617ce236ddc 100644
--- a/drivers/iio/imu/st_lsm6dsx/st_lsm6dsx.h
+++ b/drivers/iio/imu/st_lsm6dsx/st_lsm6dsx.h
@@ -39,7 +39,7 @@
#define ST_ISM330IS_DEV_NAME "ism330is"

enum st_lsm6dsx_hw_id {
- ST_LSM6DS3_ID,
+ ST_LSM6DS3_ID = 1,
ST_LSM6DS3H_ID,
ST_LSM6DSL_ID,
ST_LSM6DSM_ID,
diff --git a/drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c b/drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c
index df5f60925260..ecfceb2fb3db 100644
--- a/drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c
+++ b/drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c
@@ -23,10 +23,15 @@ static const struct regmap_config st_lsm6dsx_i2c_regmap_config = {

static int st_lsm6dsx_i2c_probe(struct i2c_client *client)
{
- const struct i2c_device_id *id = i2c_client_get_device_id(client);
- int hw_id = id->driver_data;
+ int hw_id;
struct regmap *regmap;

+ hw_id = (kernel_ulong_t)device_get_match_data(&client->dev);
+ if (!hw_id)
+ hw_id = i2c_client_get_device_id(client)->driver_data;
+ if (!hw_id)
+ return -EINVAL;
+
regmap = devm_regmap_init_i2c(client, &st_lsm6dsx_i2c_regmap_config);
if (IS_ERR(regmap)) {
dev_err(&client->dev, "Failed to register i2c regmap %ld\n", PTR_ERR(regmap));
@@ -129,6 +134,10 @@ static const struct of_device_id st_lsm6dsx_i2c_of_match[] = {
};
MODULE_DEVICE_TABLE(of, st_lsm6dsx_i2c_of_match);

+static const struct acpi_device_id st_lsm6dsx_i2c_acpi_match[] = {
+ { "SMO8B30", ST_LSM6DS3TRC_ID, },
+};
+
static const struct i2c_device_id st_lsm6dsx_i2c_id_table[] = {
{ ST_LSM6DS3_DEV_NAME, ST_LSM6DS3_ID },
{ ST_LSM6DS3H_DEV_NAME, ST_LSM6DS3H_ID },
@@ -161,6 +170,7 @@ static struct i2c_driver st_lsm6dsx_driver = {
.name = "st_lsm6dsx_i2c",
.pm = pm_sleep_ptr(&st_lsm6dsx_pm_ops),
.of_match_table = st_lsm6dsx_i2c_of_match,
+ .acpi_match_table = st_lsm6dsx_i2c_acpi_match,
},
.probe_new = st_lsm6dsx_i2c_probe,
.id_table = st_lsm6dsx_i2c_id_table,

The device is identifying itself to the kernel with PNP id SMO8B30:
physical_node:
modalias=acpi:SMO8B30:SMO8B30:
name=SMO8B30:00
uevent=MODALIAS=acpi:SMO8B30:SMO8B30:
waiting_for_supplier=0
firmware_node:
hid=SMO8B30
modalias=acpi:SMO8B30:SMO8B30:
path=\_SB_.PCI0.I2C5.DEV_
status=15
uevent=MODALIAS=acpi:SMO8B30:SMO8B30:
uid=0

The kernel module for the appropriate driver (st_lsm6dsx_i2c) is not
loaded on boot.
Modprobing it does not associate it with the device, as I would expect as
the module does not provide an alias for the above acpi/pnp id.

Can you report this issue upstream? Gues to reach out are according to
get_maintainers.pl script:

Lorenzo Bianconi <[email protected]> (maintainer:ST LSM6DSx IMU IIO DRIVER)
Jonathan Cameron <[email protected]> (maintainer:IIO SUBSYSTEM AND DRIVERS)
Lars-Peter Clausen <[email protected]> (reviewer:IIO SUBSYSTEM AND DRIVERS)
[email protected] (open list:ST LSM6DSx IMU IIO DRIVER)
[email protected] (open list)

Please keep us in the loop.

Regards,
Salvatore

#20 lore raw

Hi,

Quoted 3 lines · Darrell Kavanagh

Thanks for the info, Hans. I was hoping there'd be something like that. Compiling the whole kernel on this little machine will be interesting, though...

I usually build on a big machine and then also run "sudo make modules_install" on the big machine (which normally will not conflict with distor kernels since they have extra info in their version number) and then I do e.g.:

# on big machine:
make -j15 bzImage && make -j15 modules && sudo make modules_install
# on target, as root:
mkdir /lib/modules/6.2.0-rc6+
# on big machine
cd /lib/modules/6.2.0-rc6+/
scp -pr kernel modules.* [email protected]:/lib/modules/6.2.0-rc6+/
cd ~/projects/linux
cp arch/x86/boot/bzImage [email protected]:/boot/vmlinuz-6.2.0-rc6+
# on target, as root:
kernel-install add 6.2.0-rc6+ /boot/vmlinuz-6.2.0-rc6+

I believe you can also pass an alternate install location for the modules, then you also don't need the sudo for the make modules_install, but I tend to dogfood my own kernel tree on my $big-desktop-machine, so I actually want the modules in /lib/modules/$version/ ...

Regards,

Hans

Quoted 466 lines · Hans de Goede
Quoted 464 lines

Hi,

Quoted 22 lines · Jonathan Cameron
Quoted 19 lines · On Wed, 1 Feb 2023 01:40:49 +0000 Darrell Kavanagh…

Hello, all.

I've finally reached a conclusion on this, after testing all the combinations of the patches (with and without reading the acpi mounting matrix), window managers (wayland, xorg) and the presence or not of my custom kernel parms.

What works well is the full set of patches with the custom kernel parms and a new hwdb entry for the sensor:

sensor:modalias:acpi:SMO8B30*:dmi:*:svnLENOVO*:pn82AT:*
ACCEL_MOUNT_MATRIX=0, 1, 0; -1, 0, 0; 0, 0, 1

The autorotate then works correctly in wayland and xorg, but for xorg, the settings say the screen is "portrait left" when in actual fact it is in standard laptop landscape orientation. Wayland does not have this problem (I guess because wayland's view of the screen is straight from the kernel).

Quoted 8 lines
Quoted 4 lines

Without the hwdb entry, the orientation is 90 degrees out without using the acpi matrix and 180 degrees out when using it. I could have gone either way here with appropriate hwdb entries, but my view is that we *should* be using the matrix.

Added Hans de Goede as he has probably run into more of this mess than anyone else. Hans, any thoughts on if we are doing something wrong on kernel side? Or is the matrix just wrong *sigh*

I see below that this laptop has a panel which is mounted 90 degrees rotated, that likely explains why the ACPI matrix does not work. So the best thing to do here is to just override it with a hwdb entries.

IIRC there are already 1 or 2 other hwdb entries which actually override the ACPI provided matrix because of similar issues.

Linux userspace expects the matrix in this case to be set so that it causes e.g. gnome's auto-rotation to put the image upright even with older gnome versions / mate / xfce which don't know about the panel being mounted 90 degrees.

So e.g. "monitor-sensor" will report left-side-up or right-side-up while the device is actually in normal clamshell mode with the display up-right.

This reporting of left-side-up or right-side-up is actually "correct" looking from the native LCD panel orientation and as mentioned is done for backward compatibility. This is documented here:

https://github.com/systemd/systemd/blob/main/hwdb.d/60-sensor.hwdb#L54

The way we are handling this is likely incompatible with how Windows handles this special case of 90° rotated screen + ROTM. Or the matrix in the ACPI tables could be just wrong...

Quoted 2 lines

I think 'ROTM' is defined by MS.
https://learn.microsoft.com/en-us/windows-hardware/drivers/sensors/sensors-acpi-entries

Right and as such it would be good if we can still add support to it to the sensor driver in question. Because the ROTM info usually is correct and avoids the need for adding more and more hwdb entries.

Note there already is existing support in some other sensor drivers.

So we probably need to factor out some helper code for this and share that between sensor drivers.

Quoted 6 lines
Quoted 6 lines

The only thing that concerns me is the need for custom kernel parms. It would be better if there was a way to avoid this, so that the user didn't have to mess around with their grub config. Though having said that, the sensors fix as we have it doesn't make things worse - under currently released kernels the screen always starts up sideways unless custom parms are added in grub.

We actually have a quirk mechanism in the kernel for specifying the need for: video=DSI-1:panel_orientation=right_side_up and this will also automatically fix the fbcon orientation, see:

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/gpu/drm/drm_panel_orientation_quirks.c

If you submit a patch for this upstream please Cc me.

Regards,

Hans

Quoted 365 lines
Quoted 364 lines
Quoted 364 lines · Jonathan Cameron
Quoted 363 lines
Quoted 15 lines · On Mon, 30 Jan 2023 20:02:31 +0000 Darrell Kavanagh…

Thanks. To be clear, before I changed the grub command line, the system always booted up "sideways", even when the sensor was not being detected. This was true for everything, not just Gnome, except grub itself.

I added:

GRUB_CMDLINE_LINUX_DEFAULT="fbcon=rotate:1 video=DSI-1:panel_orientation=right_side_up quiet splash"

This fixed the orientation for pre-splash boot messages, the splash screen and the desktop environment. But not, for example (as I saw after adding my own module signing key for testing your fixes), the MOK validation screens.

Quoted 4 lines

Does this make sense?

Does what you are proposing act at a lower level than changing the systemd hwdb orientation matrix?

I'm not sure on the userspace side of things, but intent is that it will provide the orientation data to any users - though only after the kernel boots and software needs to be aware of it. Give it a go, and if not Bastien (IIRC wrote iio-sensor-proxy) may be able to advise.

For Bastien - patches for kernel side are:
https://lore.kernel.org/linux-iio/[email protected]/T/#t

Darrell is going to test them after back porting to 6.1. With the first patch he gets the right result in gnome but we weren't picking up the rotation matrix at that point (ROTM in ACPI).

Thanks,

Jonathan

Quoted 322 lines

Thanks,
Darrell

Quoted 317 lines · Jonathan Cameron
Quoted 5 lines · On Mon, 30 Jan 2023 18:32:02 +0000 Darrell Kavanagh…
Quoted 2 lines · Jonathan Cameron

That certainly looks like suitable matrix.

Quoted 7 lines
Quoted 4 lines

If we are lucky it matches the handling in bmc150-accel-core.c for an identically named method. That is going to swap the x and y axis which is I'd have thought would be rather bad if what you have is currently working well.

Actually, that would be good, because at present I have to rotate 90 degrees in my grub command line.

:)

I'll see if I can roll a suitable patch.

Jonathan

Quoted 294 lines

Darrell

Quoted 291 lines
Quoted 290 lines
            }

            Method (PRIM, 0, NotSerialized)
            {
                Name (RBUF, Buffer (One)
                {
                     0x01                                             // .
                })
                Return (RBUF) /* \_SB_.PCI0.I2C5.DEV_.PRIM.RBUF */
            }

            Method (_STA, 0, NotSerialized)  // _STA: Status
            {
                If ((GAVT == 0x6A))
                {
                    Return (0x0F)
                }
                Else
                {
                    Return (Zero)
                }
            }

            Method (CALS, 1, NotSerialized)
            {
                Local0 = Arg0
                If (((Local0 == Zero) || (Local0 == Ones)))
                {
                    Local0 = BAC1 /* \BAC1 */
                    Return (Local0)
                }
                Else
                {
                    BAC1 = Local0
                    BACS = Local0
                    BSCA (0xB0)
                }
            }
        }
    }

Thanks,
Darrell

Quoted 243 lines · Jonathan Cameron
Quoted 31 lines · On Mon, 30 Jan 2023 03:37:23 +0000 Darrell Kavanagh…

Forwarding because original html messages were rejected by the server...

---------- Forwarded message ---------
From: Darrell Kavanagh <[email protected]>
Date: Mon, 30 Jan 2023 at 02:52
Subject: Re: Bug#1029850: linux: Driver not loaded for ST
Microelectronics LSM6DS3TR-C accelerometer (acpi:SMO8B30:SMO8B30:)
To: Jonathan Cameron <[email protected]>
Cc: <[email protected]>, <[email protected]>,
<[email protected]>, <[email protected]>,
<[email protected]>

Hi Jonathan,

Thank you. The driver has evolved quite a bit in 6.2 (I read somewhere
that 6.2 includes some i2c enhancements), but I adapted your changes
to fit my Debian 6.1 kernel and it works. Two IIO devices are created
in sysfs, iio-sensor-proxy.service starts up and automatic screen
rotation in Gnome just works.

To get the modules to load on boot, I made a small change to your code
in st_lsm6dsx_i2c to add the acpi alias to modules.alias:

adding a null element to st_lsm6dsx_i2c_acpi_match:
static const struct acpi_device_id st_lsm6dsx_i2c_acpi_match[] = {
{ "SMO8B30", ST_LSM6DS3TRC_ID, },
{ },
};
then:
MODULE_DEVICE_TABLE(acpi, st_lsm6dsx_i2c_acpi_match);

doh! That was indeed sloppy of me to miss even for an untested hack.

Quoted 12 lines

dmesg shows:

[ 7366.120208] st_lsm6dsx_i2c i2c-SMO8B30:00: supply vdd not found,
using dummy regulator
[ 7366.120260] st_lsm6dsx_i2c i2c-SMO8B30:00: supply vddio not found,
using dummy regulator
[ 7366.650839] st_lsm6dsx_i2c i2c-SMO8B30:00: mounting matrix not
found: using identity...

Is this a problem?

Those are all fine. For regulators that's expected on ACPI and should be harmless as it's up to the firmware to manage power (in DT it may be up to the kernel).
For the mounting matrix, there is often something in ACPI DSDT (non standard though). Could you
cat /sys/firmware/acpi/tables/DSDT > ~/dsdt
then run through iasl from acpitools
iasl -d ~/dsdt
and find the bit related to this device.

If you can then share that there may be a _DSM or similar in there that is effectively the mounting matrix. If we are lucky it will look like some existing versions we have code to handle and can add that support as well.

Either way - I'll spin a formal patch with your fixes above and we can get this upstream for future kernels. Mounting matrix can follow later if needed.

Thanks,

Jonathan

Quoted 165 lines

Thanks again.

Darrell

On Sun, 29 Jan 2023 at 18:10, Jonathan Cameron <[email protected]> wrote:

On Sun, 29 Jan 2023 17:03:51 +0000
Darrell Kavanagh <[email protected]> wrote:

Hi,

I raised this bug in Debian, and have been asked to raise it upstream and
was given your addresses to do so. Will this email be OK, or should I raise
it in a bug tracking system somewhere?

Email is the right option.

Many thanks,
Darrell

---------- Forwarded message ---------
From: Salvatore Bonaccorso <[email protected]>
Date: Sat, 28 Jan 2023 at 20:33
Subject: Re: Bug#1029850: linux: Driver not loaded for ST Microelectronics
LSM6DS3TR-C accelerometer (acpi:SMO8B30:SMO8B30:)
To: Darrell Kavanagh <[email protected]>, <[email protected]>

Hi Darrell,

On Sat, Jan 28, 2023 at 08:13:16PM +0000, Darrell Kavanagh wrote:
Package: src:linux
Version: 6.1.4-1
Severity: normal
File: linux
X-Debbugs-Cc: [email protected]

Dear Maintainer,

This is a convertable touchscreen tablet/laptop. The rotation sensor
device
ST Microelectronics LSM6DS3TR-C does not work. It is detected via ACPI
and the
sysfs trees are created at
devices/pci0000:00/0000:00:17.1/i2c_designware.3/i2c-4/i2c-SMO8B30:00
and devices/LNXSYSTM:00/LNXSYBUS:00/PNP0A08:00/device:3c/SMO8B30:00 with
symlinks bus/acpi/devices/SMO8B30:00 and bus/i2c/devices/i2c-SMO8B30:00,
but
no driver is loaded.

At least this is using the ST PNP ID which is better than average
(long story!)

The driver in question (ultimately drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c
does not currently have an ACPI support. It should be straight forwards
to add though the driver first needs converting to use
device_get_match_data() with appropriate fallback so that it will match on
ACPI, OF or original spi_device_id tables

Completely untested but something like the following
(the offset in the enum is needed to allow us to tell if we got a result when
calling device_get_match_data() as it returns NULL on failure IIRC)

I'm not sure how sucessful the driver will be at finding any interrupts etc, but
it may get you basic functionality.

Good luck and others more familiar with the driver may well tell me what I forgot
when hacking the below ;)

diff --git a/drivers/iio/imu/st_lsm6dsx/st_lsm6dsx.h b/drivers/iio/imu/st_lsm6dsx/st_lsm6dsx.h
index 499fcf8875b4..2617ce236ddc 100644
--- a/drivers/iio/imu/st_lsm6dsx/st_lsm6dsx.h
+++ b/drivers/iio/imu/st_lsm6dsx/st_lsm6dsx.h
@@ -39,7 +39,7 @@
#define ST_ISM330IS_DEV_NAME "ism330is"

enum st_lsm6dsx_hw_id {
- ST_LSM6DS3_ID,
+ ST_LSM6DS3_ID = 1,
ST_LSM6DS3H_ID,
ST_LSM6DSL_ID,
ST_LSM6DSM_ID,
diff --git a/drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c b/drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c
index df5f60925260..ecfceb2fb3db 100644
--- a/drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c
+++ b/drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_i2c.c
@@ -23,10 +23,15 @@ static const struct regmap_config st_lsm6dsx_i2c_regmap_config = {

static int st_lsm6dsx_i2c_probe(struct i2c_client *client)
{
- const struct i2c_device_id *id = i2c_client_get_device_id(client);
- int hw_id = id->driver_data;
+ int hw_id;
struct regmap *regmap;

+ hw_id = (kernel_ulong_t)device_get_match_data(&client->dev);
+ if (!hw_id)
+ hw_id = i2c_client_get_device_id(client)->driver_data;
+ if (!hw_id)
+ return -EINVAL;
+
regmap = devm_regmap_init_i2c(client, &st_lsm6dsx_i2c_regmap_config);
if (IS_ERR(regmap)) {
dev_err(&client->dev, "Failed to register i2c regmap %ld\n", PTR_ERR(regmap));
@@ -129,6 +134,10 @@ static const struct of_device_id st_lsm6dsx_i2c_of_match[] = {
};
MODULE_DEVICE_TABLE(of, st_lsm6dsx_i2c_of_match);

+static const struct acpi_device_id st_lsm6dsx_i2c_acpi_match[] = {
+ { "SMO8B30", ST_LSM6DS3TRC_ID, },
+};
+
static const struct i2c_device_id st_lsm6dsx_i2c_id_table[] = {
{ ST_LSM6DS3_DEV_NAME, ST_LSM6DS3_ID },
{ ST_LSM6DS3H_DEV_NAME, ST_LSM6DS3H_ID },
@@ -161,6 +170,7 @@ static struct i2c_driver st_lsm6dsx_driver = {
.name = "st_lsm6dsx_i2c",
.pm = pm_sleep_ptr(&st_lsm6dsx_pm_ops),
.of_match_table = st_lsm6dsx_i2c_of_match,
+ .acpi_match_table = st_lsm6dsx_i2c_acpi_match,
},
.probe_new = st_lsm6dsx_i2c_probe,
.id_table = st_lsm6dsx_i2c_id_table,

The device is identifying itself to the kernel with PNP id SMO8B30:
physical_node:
modalias=acpi:SMO8B30:SMO8B30:
name=SMO8B30:00
uevent=MODALIAS=acpi:SMO8B30:SMO8B30:
waiting_for_supplier=0
firmware_node:
hid=SMO8B30
modalias=acpi:SMO8B30:SMO8B30:
path=\_SB_.PCI0.I2C5.DEV_
status=15
uevent=MODALIAS=acpi:SMO8B30:SMO8B30:
uid=0

The kernel module for the appropriate driver (st_lsm6dsx_i2c) is not
loaded on boot.
Modprobing it does not associate it with the device, as I would expect as
the module does not provide an alias for the above acpi/pnp id.

Can you report this issue upstream? Gues to reach out are according to
get_maintainers.pl script:

Lorenzo Bianconi <[email protected]> (maintainer:ST LSM6DSx IMU IIO DRIVER)
Jonathan Cameron <[email protected]> (maintainer:IIO SUBSYSTEM AND DRIVERS)
Lars-Peter Clausen <[email protected]> (reviewer:IIO SUBSYSTEM AND DRIVERS)
[email protected] (open list:ST LSM6DSx IMU IIO DRIVER)
[email protected] (open list)

Please keep us in the loop.

Regards,
Salvatore

#21 lore raw
In reply to Hans de Goede
Quoted 93 lines · Hans de Goede

Hi,

Quoted 34 lines · Jonathan Cameron
Quoted 24 lines · On Wed, 1 Feb 2023 01:40:49 +0000 Darrell Kavanagh…

Hello, all.

I've finally reached a conclusion on this, after testing all the combinations of the patches (with and without reading the acpi mounting matrix), window managers (wayland, xorg) and the presence or
not of my custom kernel parms.

What works well is the full set of patches with the custom kernel parms and a new hwdb entry for the sensor:

sensor:modalias:acpi:SMO8B30*:dmi:*:svnLENOVO*:pn82AT:* ACCEL_MOUNT_MATRIX=0, 1, 0; -1, 0, 0; 0, 0, 1

The autorotate then works correctly in wayland and xorg, but for xorg,
the settings say the screen is "portrait left" when in actual fact it
is in standard laptop landscape orientation. Wayland does not have
this problem (I guess because wayland's view of the screen is straight
from the kernel).

Quoted 6 lines

Without the hwdb entry, the orientation is 90 degrees out without using the acpi matrix and 180 degrees out when using it. I could have
gone either way here with appropriate hwdb entries, but my view is
that we *should* be using the matrix.

Quoted 3 lines

Added Hans de Goede as he has probably run into more of this mess than anyone else.  Hans, any thoughts on if we are doing something wrong on kernel side?  Or is the matrix just wrong *sigh*

I see below that this laptop has a panel which is mounted 90 degrees rotated, that likely explains why the ACPI matrix does not work. So the best thing to do here is to just override it with a hwdb entries.

IIRC there are already 1 or 2 other hwdb entries which actually override the ACPI provided matrix because of similar issues.

Linux userspace expects the matrix in this case to be set so that it causes e.g. gnome's auto-rotation to put the image upright even with older gnome versions / mate / xfce which don't know about the panel being mounted 90 degrees.

So e.g. "monitor-sensor" will report left-side-up or right-side-up while the device is actually in normal clamshell mode with the display up-right.

This reporting of left-side-up or right-side-up is actually "correct" looking from the native LCD panel orientation and as mentioned is done for backward compatibility. This is documented here:

https://github.com/systemd/systemd/blob/main/hwdb.d/60-sensor.hwdb#L54

The way we are handling this is likely incompatible with how Windows handles this special case of 90° rotated screen + ROTM. Or the matrix in the ACPI tables could be just wrong...

Quoted 2 lines

I think 'ROTM' is defined by MS.
https://learn.microsoft.com/en-us/windows-hardware/drivers/sensors/sensors-acpi-entries

Right and as such it would be good if we can still add support to it to the sensor driver in question. Because the ROTM info usually is correct and avoids the need for adding more and more hwdb entries.

Note there already is existing support in some other sensor drivers.

So we probably need to factor out some helper code for this and share that between sensor drivers.

Quoted 11 lines
Quoted 11 lines

The only thing that concerns me is the need for custom kernel parms.
It would be better if there was a way to avoid this, so that the user
didn't have to mess around with their grub config. Though having said
that, the sensors fix as we have it doesn't make things worse - under
currently released kernels the screen always starts up sideways unless
custom parms are added in grub.

Quoted 7 lines

We actually have a quirk mechanism in the kernel for specifying the need for: video=DSI-1:panel_orientation=right_side_up  and this will also automatically fix the fbcon orientation, see:

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/gpu/drm/drm_panel_orientation_quirks.c

If you submit a patch for this upstream please Cc me.

And if after that change, and copy/pasting the orientation from the DSDT into hwdb the sensor and screen move in the expected ways, then maybe stealing the BMC150 driver's bmc150_apply_bosc0200_acpi_orientation() might be a good idea.

Once exported through "mount_matrix", iio-sensor-proxy should see it and read it without the need for a hwdb entry.

Cheers

#22 lore raw

Thank you. I don't have anything that could be called a big machine. The fastest processor I have access to is a Core m3-8100Y - that's in a Chromebook with 4GB memory - it can run Linux in a chroot or officially in Google's VM. I also have an ancient gen 2 core i5-2410M machine which is slower than the m3 in theory, but that has 6GB of memory.

Is the kernel build more processor or memory bound?

Thanks,
Darrell

Quoted 112 lines · Bastien Nocera
Quoted 93 lines · Hans de Goede

Hi,

Quoted 34 lines · Jonathan Cameron
Quoted 24 lines · On Wed, 1 Feb 2023 01:40:49 +0000 Darrell Kavanagh…

Hello, all.

I've finally reached a conclusion on this, after testing all the combinations of the patches (with and without reading the acpi mounting matrix), window managers (wayland, xorg) and the presence or
not of my custom kernel parms.

What works well is the full set of patches with the custom kernel parms and a new hwdb entry for the sensor:

sensor:modalias:acpi:SMO8B30*:dmi:*:svnLENOVO*:pn82AT:*
ACCEL_MOUNT_MATRIX=0, 1, 0; -1, 0, 0; 0, 0, 1

The autorotate then works correctly in wayland and xorg, but for xorg,
the settings say the screen is "portrait left" when in actual fact it
is in standard laptop landscape orientation. Wayland does not have
this problem (I guess because wayland's view of the screen is straight
from the kernel).

Quoted 6 lines

Without the hwdb entry, the orientation is 90 degrees out without using the acpi matrix and 180 degrees out when using it. I could have
gone either way here with appropriate hwdb entries, but my view is
that we *should* be using the matrix.

Quoted 3 lines

Added Hans de Goede as he has probably run into more of this mess than anyone else. Hans, any thoughts on if we are doing something wrong on kernel side? Or is the matrix just wrong *sigh*

I see below that this laptop has a panel which is mounted 90 degrees rotated, that likely explains why the ACPI matrix does not work. So the best thing to do here is to just override it with a hwdb entries.

IIRC there are already 1 or 2 other hwdb entries which actually override the ACPI provided matrix because of similar issues.

Linux userspace expects the matrix in this case to be set so that it causes e.g. gnome's auto-rotation to put the image upright even with older gnome versions / mate / xfce which don't know about the panel being mounted 90 degrees.

So e.g. "monitor-sensor" will report left-side-up or right-side-up while the device is actually in normal clamshell mode with the display up-right.

This reporting of left-side-up or right-side-up is actually "correct" looking from the native LCD panel orientation and as mentioned is done for backward compatibility. This is documented here:

https://github.com/systemd/systemd/blob/main/hwdb.d/60-sensor.hwdb#L54

The way we are handling this is likely incompatible with how Windows handles this special case of 90° rotated screen + ROTM. Or the matrix in the ACPI tables could be just wrong...

Quoted 2 lines

I think 'ROTM' is defined by MS.
https://learn.microsoft.com/en-us/windows-hardware/drivers/sensors/sensors-acpi-entries

Right and as such it would be good if we can still add support to it to the sensor driver in question. Because the ROTM info usually is correct and avoids the need for adding more and more hwdb entries.

Note there already is existing support in some other sensor drivers.

So we probably need to factor out some helper code for this and share that between sensor drivers.

Quoted 11 lines
Quoted 11 lines

The only thing that concerns me is the need for custom kernel parms.
It would be better if there was a way to avoid this, so that the user
didn't have to mess around with their grub config. Though having said
that, the sensors fix as we have it doesn't make things worse - under
currently released kernels the screen always starts up sideways unless
custom parms are added in grub.

Quoted 7 lines

We actually have a quirk mechanism in the kernel for specifying the need for: video=DSI-1:panel_orientation=right_side_up and this will also automatically fix the fbcon orientation, see:

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/gpu/drm/drm_panel_orientation_quirks.c

If you submit a patch for this upstream please Cc me.

And if after that change, and copy/pasting the orientation from the DSDT into hwdb the sensor and screen move in the expected ways, then maybe stealing the BMC150 driver's bmc150_apply_bosc0200_acpi_orientation() might be a good idea.

Once exported through "mount_matrix", iio-sensor-proxy should see it and read it without the need for a hwdb entry.

Cheers

#23 lore raw

Hi,

Quoted 8 lines · Darrell Kavanagh

Thank you. I don't have anything that could be called a big machine. The fastest processor I have access to is a Core m3-8100Y - that's in a Chromebook with 4GB memory - it can run Linux in a chroot or officially in Google's VM. I also have an ancient gen 2 core i5-2410M machine which is slower than the m3 in theory, but that has 6GB of memory.

Is the kernel build more processor or memory bound?

It is mostly processor bound, esp. wtih something like make -j4, make -j16 will start taking some RAM, but with make -j4 I expect you to be fully CPU bound.

Regards,

Hans

Quoted 114 lines · Bastien Nocera
Quoted 112 lines · Hans de Goede
Quoted 93 lines

Hi,

Quoted 34 lines · Jonathan Cameron
Quoted 24 lines · On Wed, 1 Feb 2023 01:40:49 +0000 Darrell Kavanagh…

Hello, all.

I've finally reached a conclusion on this, after testing all the combinations of the patches (with and without reading the acpi mounting matrix), window managers (wayland, xorg) and the presence or
not of my custom kernel parms.

What works well is the full set of patches with the custom kernel parms and a new hwdb entry for the sensor:

sensor:modalias:acpi:SMO8B30*:dmi:*:svnLENOVO*:pn82AT:*
ACCEL_MOUNT_MATRIX=0, 1, 0; -1, 0, 0; 0, 0, 1

The autorotate then works correctly in wayland and xorg, but for xorg,
the settings say the screen is "portrait left" when in actual fact it
is in standard laptop landscape orientation. Wayland does not have
this problem (I guess because wayland's view of the screen is straight
from the kernel).

Quoted 6 lines

Without the hwdb entry, the orientation is 90 degrees out without using the acpi matrix and 180 degrees out when using it. I could have
gone either way here with appropriate hwdb entries, but my view is
that we *should* be using the matrix.

Quoted 3 lines

Added Hans de Goede as he has probably run into more of this mess than anyone else. Hans, any thoughts on if we are doing something wrong on kernel side? Or is the matrix just wrong *sigh*

I see below that this laptop has a panel which is mounted 90 degrees rotated, that likely explains why the ACPI matrix does not work. So the best thing to do here is to just override it with a hwdb entries.

IIRC there are already 1 or 2 other hwdb entries which actually override the ACPI provided matrix because of similar issues.

Linux userspace expects the matrix in this case to be set so that it causes e.g. gnome's auto-rotation to put the image upright even with older gnome versions / mate / xfce which don't know about the panel being mounted 90 degrees.

So e.g. "monitor-sensor" will report left-side-up or right-side-up while the device is actually in normal clamshell mode with the display up-right.

This reporting of left-side-up or right-side-up is actually "correct" looking from the native LCD panel orientation and as mentioned is done for backward compatibility. This is documented here:

https://github.com/systemd/systemd/blob/main/hwdb.d/60-sensor.hwdb#L54

The way we are handling this is likely incompatible with how Windows handles this special case of 90° rotated screen + ROTM. Or the matrix in the ACPI tables could be just wrong...

Quoted 2 lines

I think 'ROTM' is defined by MS.
https://learn.microsoft.com/en-us/windows-hardware/drivers/sensors/sensors-acpi-entries

Right and as such it would be good if we can still add support to it to the sensor driver in question. Because the ROTM info usually is correct and avoids the need for adding more and more hwdb entries.

Note there already is existing support in some other sensor drivers.

So we probably need to factor out some helper code for this and share that between sensor drivers.

Quoted 11 lines
Quoted 11 lines

The only thing that concerns me is the need for custom kernel parms.
It would be better if there was a way to avoid this, so that the user
didn't have to mess around with their grub config. Though having said
that, the sensors fix as we have it doesn't make things worse - under
currently released kernels the screen always starts up sideways unless
custom parms are added in grub.

Quoted 7 lines

We actually have a quirk mechanism in the kernel for specifying the need for: video=DSI-1:panel_orientation=right_side_up and this will also automatically fix the fbcon orientation, see:

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/gpu/drm/drm_panel_orientation_quirks.c

If you submit a patch for this upstream please Cc me.

And if after that change, and copy/pasting the orientation from the DSDT into hwdb the sensor and screen move in the expected ways, then maybe stealing the BMC150 driver's bmc150_apply_bosc0200_acpi_orientation() might be a good idea.

Once exported through "mount_matrix", iio-sensor-proxy should see it and read it without the need for a hwdb entry.

Cheers

#24 lore raw

Finally got a 6.2.0-rc6 kernel built and installed, with the following patch, and everything is working as expected.

Moving on now to look at Bastien's suggestion.

Thanks,
Darrell

(quoted diff) +0 -0

b/kernel/linux-6.2-rc6/drivers/gpu/drm/drm_panel_orientation_quirks.c index 3659f04..590bb7b 100644

kernel/linux-6.2-rc6/drivers/gpu/drm/drm_panel_orientation_quirks.c renamed from kernel/drm_panel_orientation_quirks.c +0 -0
@@ -304,6 +304,12 @@ static const struct dmi_system_id orientation_data[] = {
                  DMI_EXACT_MATCH(DMI_PRODUCT_VERSION, "Lenovo ideapad

D330-10IGM"),

                },
                .driver_data = (void *)&lcd1200x1920_rightside_up,

+ }, { /* Lenovo IdeaPad Duet 3 10IGL5 */
+ .matches = {
+ DMI_EXACT_MATCH(DMI_SYS_VENDOR, "LENOVO"),
+ DMI_EXACT_MATCH(DMI_PRODUCT_VERSION, "IdeaPad Duet 3 10IGL5"),

+               },
+               .driver_data = (void *)&lcd1200x1920_rightside_up,
        }, {    /* Lenovo Ideapad D330-10IGL (HD) */
                .matches = {
                  DMI_EXACT_MATCH(DMI_SYS_VENDOR, "LENOVO"),
Quoted 145 lines · Hans de Goede

Hi,

Quoted 8 lines · Darrell Kavanagh

Thank you. I don't have anything that could be called a big machine. The fastest processor I have access to is a Core m3-8100Y - that's in a Chromebook with 4GB memory - it can run Linux in a chroot or officially in Google's VM. I also have an ancient gen 2 core i5-2410M machine which is slower than the m3 in theory, but that has 6GB of memory.

Is the kernel build more processor or memory bound?

It is mostly processor bound, esp. wtih something like make -j4, make -j16 will start taking some RAM, but with make -j4 I expect you to be fully CPU bound.

Regards,

Hans

Quoted 114 lines · Bastien Nocera
Quoted 112 lines · Hans de Goede
Quoted 93 lines

Hi,

Quoted 34 lines · Jonathan Cameron
Quoted 24 lines · On Wed, 1 Feb 2023 01:40:49 +0000 Darrell Kavanagh…

Hello, all.

I've finally reached a conclusion on this, after testing all the combinations of the patches (with and without reading the acpi mounting matrix), window managers (wayland, xorg) and the presence or
not of my custom kernel parms.

What works well is the full set of patches with the custom kernel parms and a new hwdb entry for the sensor:

sensor:modalias:acpi:SMO8B30*:dmi:*:svnLENOVO*:pn82AT:*
ACCEL_MOUNT_MATRIX=0, 1, 0; -1, 0, 0; 0, 0, 1

The autorotate then works correctly in wayland and xorg, but for xorg,
the settings say the screen is "portrait left" when in actual fact it
is in standard laptop landscape orientation. Wayland does not have
this problem (I guess because wayland's view of the screen is straight
from the kernel).

Quoted 6 lines

Without the hwdb entry, the orientation is 90 degrees out without using the acpi matrix and 180 degrees out when using it. I could have
gone either way here with appropriate hwdb entries, but my view is
that we *should* be using the matrix.

Quoted 3 lines

Added Hans de Goede as he has probably run into more of this mess than anyone else. Hans, any thoughts on if we are doing something wrong on kernel side? Or is the matrix just wrong *sigh*

I see below that this laptop has a panel which is mounted 90 degrees rotated, that likely explains why the ACPI matrix does not work. So the best thing to do here is to just override it with a hwdb entries.

IIRC there are already 1 or 2 other hwdb entries which actually override the ACPI provided matrix because of similar issues.

Linux userspace expects the matrix in this case to be set so that it causes e.g. gnome's auto-rotation to put the image upright even with older gnome versions / mate / xfce which don't know about the panel being mounted 90 degrees.

So e.g. "monitor-sensor" will report left-side-up or right-side-up while the device is actually in normal clamshell mode with the display up-right.

This reporting of left-side-up or right-side-up is actually "correct" looking from the native LCD panel orientation and as mentioned is done for backward compatibility. This is documented here:

https://github.com/systemd/systemd/blob/main/hwdb.d/60-sensor.hwdb#L54

The way we are handling this is likely incompatible with how Windows handles this special case of 90° rotated screen + ROTM. Or the matrix in the ACPI tables could be just wrong...

Quoted 2 lines

I think 'ROTM' is defined by MS.
https://learn.microsoft.com/en-us/windows-hardware/drivers/sensors/sensors-acpi-entries

Right and as such it would be good if we can still add support to it to the sensor driver in question. Because the ROTM info usually is correct and avoids the need for adding more and more hwdb entries.

Note there already is existing support in some other sensor drivers.

So we probably need to factor out some helper code for this and share that between sensor drivers.

Quoted 11 lines
Quoted 11 lines

The only thing that concerns me is the need for custom kernel parms.
It would be better if there was a way to avoid this, so that the user
didn't have to mess around with their grub config. Though having said
that, the sensors fix as we have it doesn't make things worse - under
currently released kernels the screen always starts up sideways unless
custom parms are added in grub.

Quoted 7 lines

We actually have a quirk mechanism in the kernel for specifying the need for: video=DSI-1:panel_orientation=right_side_up and this will also automatically fix the fbcon orientation, see:

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/gpu/drm/drm_panel_orientation_quirks.c

If you submit a patch for this upstream please Cc me.

And if after that change, and copy/pasting the orientation from the DSDT into hwdb the sensor and screen move in the expected ways, then maybe stealing the BMC150 driver's bmc150_apply_bosc0200_acpi_orientation() might be a good idea.

Once exported through "mount_matrix", iio-sensor-proxy should see it and read it without the need for a hwdb entry.

Cheers

#25 lore raw

I've just noticed that the working mount matrix that I added to my hwdb is the matrix retrieved from the ACPI ROTM call in the amended driver, transposed.

Useful information or coincidence?

Darrell

Quoted 175 lines · Darrell Kavanagh

Finally got a 6.2.0-rc6 kernel built and installed, with the following patch, and everything is working as expected.

Moving on now to look at Bastien's suggestion.

Thanks,
Darrell

kernel/linux-6.2-rc6/drivers/gpu/drm/drm_panel_orientation_quirks.c renamed from kernel/drm_panel_orientation_quirks.c +6 -0
index 3659f04..590bb7b 100644
@@ -304,6 +304,12 @@ static const struct dmi_system_id orientation_data[] = {
                  DMI_EXACT_MATCH(DMI_PRODUCT_VERSION, "Lenovo ideapad
 D330-10IGM"),
                },
                .driver_data = (void *)&lcd1200x1920_rightside_up,
+       }, {    /* Lenovo IdeaPad Duet 3 10IGL5 */
+               .matches = {
+                 DMI_EXACT_MATCH(DMI_SYS_VENDOR, "LENOVO"),
+                 DMI_EXACT_MATCH(DMI_PRODUCT_VERSION, "IdeaPad Duet 3 10IGL5"),
+               },
+               .driver_data = (void *)&lcd1200x1920_rightside_up,
        }, {    /* Lenovo Ideapad D330-10IGL (HD) */
                .matches = {

DMI_EXACT_MATCH(DMI_SYS_VENDOR, "LENOVO"),

Quoted 145 lines · Hans de Goede

Hi,

Quoted 8 lines · Darrell Kavanagh

Thank you. I don't have anything that could be called a big machine. The fastest processor I have access to is a Core m3-8100Y - that's in a Chromebook with 4GB memory - it can run Linux in a chroot or officially in Google's VM. I also have an ancient gen 2 core i5-2410M machine which is slower than the m3 in theory, but that has 6GB of memory.

Is the kernel build more processor or memory bound?

It is mostly processor bound, esp. wtih something like make -j4, make -j16 will start taking some RAM, but with make -j4 I expect you to be fully CPU bound.

Regards,

Hans

Quoted 114 lines · Bastien Nocera
Quoted 112 lines · Hans de Goede
Quoted 93 lines

Hi,

Quoted 34 lines · Jonathan Cameron
Quoted 24 lines · On Wed, 1 Feb 2023 01:40:49 +0000 Darrell Kavanagh…

Hello, all.

I've finally reached a conclusion on this, after testing all the combinations of the patches (with and without reading the acpi mounting matrix), window managers (wayland, xorg) and the presence or
not of my custom kernel parms.

What works well is the full set of patches with the custom kernel parms and a new hwdb entry for the sensor:

sensor:modalias:acpi:SMO8B30*:dmi:*:svnLENOVO*:pn82AT:*
ACCEL_MOUNT_MATRIX=0, 1, 0; -1, 0, 0; 0, 0, 1

The autorotate then works correctly in wayland and xorg, but for xorg,
the settings say the screen is "portrait left" when in actual fact it
is in standard laptop landscape orientation. Wayland does not have
this problem (I guess because wayland's view of the screen is straight
from the kernel).

Quoted 6 lines

Without the hwdb entry, the orientation is 90 degrees out without using the acpi matrix and 180 degrees out when using it. I could have
gone either way here with appropriate hwdb entries, but my view is
that we *should* be using the matrix.

Quoted 3 lines

Added Hans de Goede as he has probably run into more of this mess than anyone else. Hans, any thoughts on if we are doing something wrong on kernel side? Or is the matrix just wrong *sigh*

I see below that this laptop has a panel which is mounted 90 degrees rotated, that likely explains why the ACPI matrix does not work. So the best thing to do here is to just override it with a hwdb entries.

IIRC there are already 1 or 2 other hwdb entries which actually override the ACPI provided matrix because of similar issues.

Linux userspace expects the matrix in this case to be set so that it causes e.g. gnome's auto-rotation to put the image upright even with older gnome versions / mate / xfce which don't know about the panel being mounted 90 degrees.

So e.g. "monitor-sensor" will report left-side-up or right-side-up while the device is actually in normal clamshell mode with the display up-right.

This reporting of left-side-up or right-side-up is actually "correct" looking from the native LCD panel orientation and as mentioned is done for backward compatibility. This is documented here:

https://github.com/systemd/systemd/blob/main/hwdb.d/60-sensor.hwdb#L54

The way we are handling this is likely incompatible with how Windows handles this special case of 90° rotated screen + ROTM. Or the matrix in the ACPI tables could be just wrong...

Quoted 2 lines

I think 'ROTM' is defined by MS.
https://learn.microsoft.com/en-us/windows-hardware/drivers/sensors/sensors-acpi-entries

Right and as such it would be good if we can still add support to it to the sensor driver in question. Because the ROTM info usually is correct and avoids the need for adding more and more hwdb entries.

Note there already is existing support in some other sensor drivers.

So we probably need to factor out some helper code for this and share that between sensor drivers.

Quoted 11 lines
Quoted 11 lines

The only thing that concerns me is the need for custom kernel parms.
It would be better if there was a way to avoid this, so that the user
didn't have to mess around with their grub config. Though having said
that, the sensors fix as we have it doesn't make things worse - under
currently released kernels the screen always starts up sideways unless
custom parms are added in grub.

Quoted 7 lines

We actually have a quirk mechanism in the kernel for specifying the need for: video=DSI-1:panel_orientation=right_side_up and this will also automatically fix the fbcon orientation, see:

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/gpu/drm/drm_panel_orientation_quirks.c

If you submit a patch for this upstream please Cc me.

And if after that change, and copy/pasting the orientation from the DSDT into hwdb the sensor and screen move in the expected ways, then maybe stealing the BMC150 driver's bmc150_apply_bosc0200_acpi_orientation() might be a good idea.

Once exported through "mount_matrix", iio-sensor-proxy should see it and read it without the need for a hwdb entry.

Cheers

#26 lore raw

Hi,

Quoted 3 lines · Darrell Kavanagh

I've just noticed that the working mount matrix that I added to my hwdb is the matrix retrieved from the ACPI ROTM call in the amended driver, transposed.

An other word for the mount matrix would be a rotation matrix, since it defines how the physical sensor is mounted on the PCB in a rotated fashion compared to its standard orientation.

The x, y, z axis relationship underling of course does not change by the rotation, so yes all mount matrices are a transposition of the standard:

1, 0, 0 : 0, 1, 0 : 0, 0, 1

matrix, that is expected. Where that to not be the case then there would be a bug in the accelerometer driver itself where the driver itself is swapping or inverting axis.

Regards,

Hans

Quoted 178 lines
Quoted 175 lines · Darrell Kavanagh

Finally got a 6.2.0-rc6 kernel built and installed, with the following patch, and everything is working as expected.

Moving on now to look at Bastien's suggestion.

Thanks,
Darrell

kernel/linux-6.2-rc6/drivers/gpu/drm/drm_panel_orientation_quirks.c renamed from kernel/drm_panel_orientation_quirks.c +6 -0
index 3659f04..590bb7b 100644
@@ -304,6 +304,12 @@ static const struct dmi_system_id orientation_data[] = {
                  DMI_EXACT_MATCH(DMI_PRODUCT_VERSION, "Lenovo ideapad
 D330-10IGM"),
                },
                .driver_data = (void *)&lcd1200x1920_rightside_up,
+       }, {    /* Lenovo IdeaPad Duet 3 10IGL5 */
+               .matches = {
+                 DMI_EXACT_MATCH(DMI_SYS_VENDOR, "LENOVO"),
+                 DMI_EXACT_MATCH(DMI_PRODUCT_VERSION, "IdeaPad Duet 3 10IGL5"),
+               },
+               .driver_data = (void *)&lcd1200x1920_rightside_up,
        }, {    /* Lenovo Ideapad D330-10IGL (HD) */
                .matches = {

DMI_EXACT_MATCH(DMI_SYS_VENDOR, "LENOVO"),

Quoted 145 lines · Hans de Goede

Hi,

Quoted 8 lines · Darrell Kavanagh

Thank you. I don't have anything that could be called a big machine. The fastest processor I have access to is a Core m3-8100Y - that's in a Chromebook with 4GB memory - it can run Linux in a chroot or officially in Google's VM. I also have an ancient gen 2 core i5-2410M machine which is slower than the m3 in theory, but that has 6GB of memory.

Is the kernel build more processor or memory bound?

It is mostly processor bound, esp. wtih something like make -j4, make -j16 will start taking some RAM, but with make -j4 I expect you to be fully CPU bound.

Regards,

Hans

Quoted 114 lines · Bastien Nocera
Quoted 112 lines · Hans de Goede
Quoted 93 lines

Hi,

Quoted 34 lines · Jonathan Cameron
Quoted 24 lines · On Wed, 1 Feb 2023 01:40:49 +0000 Darrell Kavanagh…

Hello, all.

I've finally reached a conclusion on this, after testing all the combinations of the patches (with and without reading the acpi mounting matrix), window managers (wayland, xorg) and the presence or
not of my custom kernel parms.

What works well is the full set of patches with the custom kernel parms and a new hwdb entry for the sensor:

sensor:modalias:acpi:SMO8B30*:dmi:*:svnLENOVO*:pn82AT:*
ACCEL_MOUNT_MATRIX=0, 1, 0; -1, 0, 0; 0, 0, 1

The autorotate then works correctly in wayland and xorg, but for xorg,
the settings say the screen is "portrait left" when in actual fact it
is in standard laptop landscape orientation. Wayland does not have
this problem (I guess because wayland's view of the screen is straight
from the kernel).

Quoted 6 lines

Without the hwdb entry, the orientation is 90 degrees out without using the acpi matrix and 180 degrees out when using it. I could have
gone either way here with appropriate hwdb entries, but my view is
that we *should* be using the matrix.

Quoted 3 lines

Added Hans de Goede as he has probably run into more of this mess than anyone else. Hans, any thoughts on if we are doing something wrong on kernel side? Or is the matrix just wrong *sigh*

I see below that this laptop has a panel which is mounted 90 degrees rotated, that likely explains why the ACPI matrix does not work. So the best thing to do here is to just override it with a hwdb entries.

IIRC there are already 1 or 2 other hwdb entries which actually override the ACPI provided matrix because of similar issues.

Linux userspace expects the matrix in this case to be set so that it causes e.g. gnome's auto-rotation to put the image upright even with older gnome versions / mate / xfce which don't know about the panel being mounted 90 degrees.

So e.g. "monitor-sensor" will report left-side-up or right-side-up while the device is actually in normal clamshell mode with the display up-right.

This reporting of left-side-up or right-side-up is actually "correct" looking from the native LCD panel orientation and as mentioned is done for backward compatibility. This is documented here:

https://github.com/systemd/systemd/blob/main/hwdb.d/60-sensor.hwdb#L54

The way we are handling this is likely incompatible with how Windows handles this special case of 90° rotated screen + ROTM. Or the matrix in the ACPI tables could be just wrong...

Quoted 2 lines

I think 'ROTM' is defined by MS.
https://learn.microsoft.com/en-us/windows-hardware/drivers/sensors/sensors-acpi-entries

Right and as such it would be good if we can still add support to it to the sensor driver in question. Because the ROTM info usually is correct and avoids the need for adding more and more hwdb entries.

Note there already is existing support in some other sensor drivers.

So we probably need to factor out some helper code for this and share that between sensor drivers.

Quoted 11 lines
Quoted 11 lines

The only thing that concerns me is the need for custom kernel parms.
It would be better if there was a way to avoid this, so that the user
didn't have to mess around with their grub config. Though having said
that, the sensors fix as we have it doesn't make things worse - under
currently released kernels the screen always starts up sideways unless
custom parms are added in grub.

Quoted 7 lines

We actually have a quirk mechanism in the kernel for specifying the need for: video=DSI-1:panel_orientation=right_side_up and this will also automatically fix the fbcon orientation, see:

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/gpu/drm/drm_panel_orientation_quirks.c

If you submit a patch for this upstream please Cc me.

And if after that change, and copy/pasting the orientation from the DSDT into hwdb the sensor and screen move in the expected ways, then maybe stealing the BMC150 driver's bmc150_apply_bosc0200_acpi_orientation() might be a good idea.

Once exported through "mount_matrix", iio-sensor-proxy should see it and read it without the need for a hwdb entry.

Cheers

#27 lore raw

Yes, I understand that.

What I mean is that the matrix read from the DSDT by Jonathan's amended driver is

 0 -1  0
 1  0  0
 0  0  1

and the (correct) matrix created with my new hwdb entry is

 0  1  0
-1  0  0
 0  0  1

which is the algebraic transposition (ie reflection in the diagonal) of the DSDT one.

In other words, though the DST matrix is wrong, it is wrong in a specific way - the rows should be the columns, and vv. I was just wondering if this was a DSDT bug that might have been seen elsewhere before.

BTW, there is another matrix in the DSTD, but I can't find the associated HID (10EC5280) anywhere (Linux sysfs or Windows Powershell system data extract). It's not a correct matrix, though - could it be just a bit of redundant code in the DST?

Darrell

Quoted 208 lines · Hans de Goede

Hi,

Quoted 3 lines · Darrell Kavanagh

I've just noticed that the working mount matrix that I added to my hwdb is the matrix retrieved from the ACPI ROTM call in the amended driver, transposed.

An other word for the mount matrix would be a rotation matrix, since it defines how the physical sensor is mounted on the PCB in a rotated fashion compared to its standard orientation.

The x, y, z axis relationship underling of course does not change by the rotation, so yes all mount matrices are a transposition of the standard:

1, 0, 0 : 0, 1, 0 : 0, 0, 1

matrix, that is expected. Where that to not be the case then there would be a bug in the accelerometer driver itself where the driver itself is swapping or inverting axis.

Regards,

Hans

Quoted 178 lines
Quoted 175 lines · Darrell Kavanagh

Finally got a 6.2.0-rc6 kernel built and installed, with the following patch, and everything is working as expected.

Moving on now to look at Bastien's suggestion.

Thanks,
Darrell

kernel/linux-6.2-rc6/drivers/gpu/drm/drm_panel_orientation_quirks.c renamed from kernel/drm_panel_orientation_quirks.c +6 -0
index 3659f04..590bb7b 100644
@@ -304,6 +304,12 @@ static const struct dmi_system_id orientation_data[] = {
                  DMI_EXACT_MATCH(DMI_PRODUCT_VERSION, "Lenovo ideapad
 D330-10IGM"),
                },
                .driver_data = (void *)&lcd1200x1920_rightside_up,
+       }, {    /* Lenovo IdeaPad Duet 3 10IGL5 */
+               .matches = {
+                 DMI_EXACT_MATCH(DMI_SYS_VENDOR, "LENOVO"),
+                 DMI_EXACT_MATCH(DMI_PRODUCT_VERSION, "IdeaPad Duet 3 10IGL5"),
+               },
+               .driver_data = (void *)&lcd1200x1920_rightside_up,
        }, {    /* Lenovo Ideapad D330-10IGL (HD) */
                .matches = {

DMI_EXACT_MATCH(DMI_SYS_VENDOR, "LENOVO"),

Quoted 145 lines · Hans de Goede

Hi,

Quoted 8 lines · Darrell Kavanagh

Thank you. I don't have anything that could be called a big machine. The fastest processor I have access to is a Core m3-8100Y - that's in a Chromebook with 4GB memory - it can run Linux in a chroot or officially in Google's VM. I also have an ancient gen 2 core i5-2410M machine which is slower than the m3 in theory, but that has 6GB of memory.

Is the kernel build more processor or memory bound?

It is mostly processor bound, esp. wtih something like make -j4, make -j16 will start taking some RAM, but with make -j4 I expect you to be fully CPU bound.

Regards,

Hans

Quoted 114 lines · Bastien Nocera
Quoted 112 lines · Hans de Goede
Quoted 93 lines

Hi,

Quoted 34 lines · Jonathan Cameron
Quoted 24 lines · On Wed, 1 Feb 2023 01:40:49 +0000 Darrell Kavanagh…

Hello, all.

I've finally reached a conclusion on this, after testing all the combinations of the patches (with and without reading the acpi mounting matrix), window managers (wayland, xorg) and the presence or
not of my custom kernel parms.

What works well is the full set of patches with the custom kernel parms and a new hwdb entry for the sensor:

sensor:modalias:acpi:SMO8B30*:dmi:*:svnLENOVO*:pn82AT:*
ACCEL_MOUNT_MATRIX=0, 1, 0; -1, 0, 0; 0, 0, 1

The autorotate then works correctly in wayland and xorg, but for xorg,
the settings say the screen is "portrait left" when in actual fact it
is in standard laptop landscape orientation. Wayland does not have
this problem (I guess because wayland's view of the screen is straight
from the kernel).

Quoted 6 lines

Without the hwdb entry, the orientation is 90 degrees out without using the acpi matrix and 180 degrees out when using it. I could have
gone either way here with appropriate hwdb entries, but my view is
that we *should* be using the matrix.

Quoted 3 lines

Added Hans de Goede as he has probably run into more of this mess than anyone else. Hans, any thoughts on if we are doing something wrong on kernel side? Or is the matrix just wrong *sigh*

I see below that this laptop has a panel which is mounted 90 degrees rotated, that likely explains why the ACPI matrix does not work. So the best thing to do here is to just override it with a hwdb entries.

IIRC there are already 1 or 2 other hwdb entries which actually override the ACPI provided matrix because of similar issues.

Linux userspace expects the matrix in this case to be set so that it causes e.g. gnome's auto-rotation to put the image upright even with older gnome versions / mate / xfce which don't know about the panel being mounted 90 degrees.

So e.g. "monitor-sensor" will report left-side-up or right-side-up while the device is actually in normal clamshell mode with the display up-right.

This reporting of left-side-up or right-side-up is actually "correct" looking from the native LCD panel orientation and as mentioned is done for backward compatibility. This is documented here:

https://github.com/systemd/systemd/blob/main/hwdb.d/60-sensor.hwdb#L54

The way we are handling this is likely incompatible with how Windows handles this special case of 90° rotated screen + ROTM. Or the matrix in the ACPI tables could be just wrong...

Quoted 2 lines

I think 'ROTM' is defined by MS.
https://learn.microsoft.com/en-us/windows-hardware/drivers/sensors/sensors-acpi-entries

Right and as such it would be good if we can still add support to it to the sensor driver in question. Because the ROTM info usually is correct and avoids the need for adding more and more hwdb entries.

Note there already is existing support in some other sensor drivers.

So we probably need to factor out some helper code for this and share that between sensor drivers.

Quoted 11 lines
Quoted 11 lines

The only thing that concerns me is the need for custom kernel parms.
It would be better if there was a way to avoid this, so that the user
didn't have to mess around with their grub config. Though having said
that, the sensors fix as we have it doesn't make things worse - under
currently released kernels the screen always starts up sideways unless
custom parms are added in grub.

Quoted 7 lines

We actually have a quirk mechanism in the kernel for specifying the need for: video=DSI-1:panel_orientation=right_side_up and this will also automatically fix the fbcon orientation, see:

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/gpu/drm/drm_panel_orientation_quirks.c

If you submit a patch for this upstream please Cc me.

And if after that change, and copy/pasting the orientation from the DSDT into hwdb the sensor and screen move in the expected ways, then maybe stealing the BMC150 driver's bmc150_apply_bosc0200_acpi_orientation() might be a good idea.

Once exported through "mount_matrix", iio-sensor-proxy should see it and read it without the need for a hwdb entry.

Cheers

#28 lore raw

Hi,

Quoted 15 lines · Darrell Kavanagh

Yes, I understand that.

What I mean is that the matrix read from the DSDT by Jonathan's amended driver is

 0 -1  0
 1  0  0
 0  0  1

and the (correct) matrix created with my new hwdb entry is

 0  1  0
-1  0  0
 0  0  1
Quoted 7 lines

which is the algebraic transposition (ie reflection in the diagonal) of the DSDT one.

In other words, though the DST matrix is wrong, it is wrong in a specific way - the rows should be the columns, and vv. I was just wondering if this was a DSDT bug that might have been seen elsewhere before.

No this does not ring a bell, but the x and y axis being swapped does seem related to the LCD panel being 90° rotated.

Quoted 4 lines

BTW, there is another matrix in the DSTD, but I can't find the associated HID (10EC5280) anywhere (Linux sysfs or Windows Powershell system data extract). It's not a correct matrix, though - could it be just a bit of redundant code in the DST?

Yes that is likely there often is a bunch of dead stuff DSDT leftover from other device models.

Regards,

Hans

Quoted 213 lines

Darrell

Quoted 208 lines · Hans de Goede

Hi,

Quoted 3 lines · Darrell Kavanagh

I've just noticed that the working mount matrix that I added to my hwdb is the matrix retrieved from the ACPI ROTM call in the amended driver, transposed.

An other word for the mount matrix would be a rotation matrix, since it defines how the physical sensor is mounted on the PCB in a rotated fashion compared to its standard orientation.

The x, y, z axis relationship underling of course does not change by the rotation, so yes all mount matrices are a transposition of the standard:

1, 0, 0 : 0, 1, 0 : 0, 0, 1

matrix, that is expected. Where that to not be the case then there would be a bug in the accelerometer driver itself where the driver itself is swapping or inverting axis.

Regards,

Hans

Quoted 178 lines
Quoted 175 lines · Darrell Kavanagh

Finally got a 6.2.0-rc6 kernel built and installed, with the following patch, and everything is working as expected.

Moving on now to look at Bastien's suggestion.

Thanks,
Darrell

kernel/linux-6.2-rc6/drivers/gpu/drm/drm_panel_orientation_quirks.c renamed from kernel/drm_panel_orientation_quirks.c +6 -0
index 3659f04..590bb7b 100644
@@ -304,6 +304,12 @@ static const struct dmi_system_id orientation_data[] = {
                  DMI_EXACT_MATCH(DMI_PRODUCT_VERSION, "Lenovo ideapad
 D330-10IGM"),
                },
                .driver_data = (void *)&lcd1200x1920_rightside_up,
+       }, {    /* Lenovo IdeaPad Duet 3 10IGL5 */
+               .matches = {
+                 DMI_EXACT_MATCH(DMI_SYS_VENDOR, "LENOVO"),
+                 DMI_EXACT_MATCH(DMI_PRODUCT_VERSION, "IdeaPad Duet 3 10IGL5"),
+               },
+               .driver_data = (void *)&lcd1200x1920_rightside_up,
        }, {    /* Lenovo Ideapad D330-10IGL (HD) */
                .matches = {

DMI_EXACT_MATCH(DMI_SYS_VENDOR, "LENOVO"),

Quoted 145 lines · Hans de Goede

Hi,

Quoted 8 lines · Darrell Kavanagh

Thank you. I don't have anything that could be called a big machine. The fastest processor I have access to is a Core m3-8100Y - that's in a Chromebook with 4GB memory - it can run Linux in a chroot or officially in Google's VM. I also have an ancient gen 2 core i5-2410M machine which is slower than the m3 in theory, but that has 6GB of memory.

Is the kernel build more processor or memory bound?

It is mostly processor bound, esp. wtih something like make -j4, make -j16 will start taking some RAM, but with make -j4 I expect you to be fully CPU bound.

Regards,

Hans

Quoted 114 lines · Bastien Nocera
Quoted 112 lines · Hans de Goede
Quoted 93 lines

Hi,

Quoted 34 lines · Jonathan Cameron
Quoted 24 lines · On Wed, 1 Feb 2023 01:40:49 +0000 Darrell Kavanagh…

Hello, all.

I've finally reached a conclusion on this, after testing all the combinations of the patches (with and without reading the acpi mounting matrix), window managers (wayland, xorg) and the presence or
not of my custom kernel parms.

What works well is the full set of patches with the custom kernel parms and a new hwdb entry for the sensor:

sensor:modalias:acpi:SMO8B30*:dmi:*:svnLENOVO*:pn82AT:*
ACCEL_MOUNT_MATRIX=0, 1, 0; -1, 0, 0; 0, 0, 1

The autorotate then works correctly in wayland and xorg, but for xorg,
the settings say the screen is "portrait left" when in actual fact it
is in standard laptop landscape orientation. Wayland does not have
this problem (I guess because wayland's view of the screen is straight
from the kernel).

Quoted 6 lines

Without the hwdb entry, the orientation is 90 degrees out without using the acpi matrix and 180 degrees out when using it. I could have
gone either way here with appropriate hwdb entries, but my view is
that we *should* be using the matrix.

Quoted 3 lines

Added Hans de Goede as he has probably run into more of this mess than anyone else. Hans, any thoughts on if we are doing something wrong on kernel side? Or is the matrix just wrong *sigh*

I see below that this laptop has a panel which is mounted 90 degrees rotated, that likely explains why the ACPI matrix does not work. So the best thing to do here is to just override it with a hwdb entries.

IIRC there are already 1 or 2 other hwdb entries which actually override the ACPI provided matrix because of similar issues.

Linux userspace expects the matrix in this case to be set so that it causes e.g. gnome's auto-rotation to put the image upright even with older gnome versions / mate / xfce which don't know about the panel being mounted 90 degrees.

So e.g. "monitor-sensor" will report left-side-up or right-side-up while the device is actually in normal clamshell mode with the display up-right.

This reporting of left-side-up or right-side-up is actually "correct" looking from the native LCD panel orientation and as mentioned is done for backward compatibility. This is documented here:

https://github.com/systemd/systemd/blob/main/hwdb.d/60-sensor.hwdb#L54

The way we are handling this is likely incompatible with how Windows handles this special case of 90° rotated screen + ROTM. Or the matrix in the ACPI tables could be just wrong...

Quoted 2 lines

I think 'ROTM' is defined by MS.
https://learn.microsoft.com/en-us/windows-hardware/drivers/sensors/sensors-acpi-entries

Right and as such it would be good if we can still add support to it to the sensor driver in question. Because the ROTM info usually is correct and avoids the need for adding more and more hwdb entries.

Note there already is existing support in some other sensor drivers.

So we probably need to factor out some helper code for this and share that between sensor drivers.

Quoted 11 lines
Quoted 11 lines

The only thing that concerns me is the need for custom kernel parms.
It would be better if there was a way to avoid this, so that the user
didn't have to mess around with their grub config. Though having said
that, the sensors fix as we have it doesn't make things worse - under
currently released kernels the screen always starts up sideways unless
custom parms are added in grub.

Quoted 7 lines

We actually have a quirk mechanism in the kernel for specifying the need for: video=DSI-1:panel_orientation=right_side_up and this will also automatically fix the fbcon orientation, see:

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/gpu/drm/drm_panel_orientation_quirks.c

If you submit a patch for this upstream please Cc me.

And if after that change, and copy/pasting the orientation from the DSDT into hwdb the sensor and screen move in the expected ways, then maybe stealing the BMC150 driver's bmc150_apply_bosc0200_acpi_orientation() might be a good idea.

Once exported through "mount_matrix", iio-sensor-proxy should see it and read it without the need for a hwdb entry.

Cheers

author maintainer
#29 lore raw
Quoted 17 lines · On Sun, 5 Feb 2023 09:50:51 +0100 Hans de Goede <[email protected]> wrote:

Hi,

Quoted 14 lines · Darrell Kavanagh

Yes, I understand that.

What I mean is that the matrix read from the DSDT by Jonathan's amended driver is

 0 -1  0
 1  0  0
 0  0  1

and the (correct) matrix created with my new hwdb entry is

 0  1  0
-1  0  0
 0  0  1

May be concidence, but I think that's the inverse of the one we are reading from ROTM - so represents the transform in the other direction.

The way ROTM is defined is that first row represents the direction of the x axis in device coordinates - so it's the transform from sensor to device space.

I wonder if the hwdb matrix is defined from world space to sensor? Seems unlikely.

The IIO ABI docs describe mount matrix as being what you apply to data to tranform into device space (oh for a diagram in the docs). Anyhow my reading is that matches with ROTM definition but maybe I'm reading that wrong...

For extra annoyance, the ROTM matrix on this device isn't a rotation matrix. It's flipping the handedness of the sensor. Determinant isn't -1 which it should be. I guess the sensor itself might have an axis backwards from windows convention though *sigh* I think windows uses left handed convention and looks like sensor is using right handed (which I think is what Android and similar use).

Quoted 242 lines
Quoted 8 lines

which is the algebraic transposition (ie reflection in the diagonal) of the DSDT one.

In other words, though the DST matrix is wrong, it is wrong in a specific way - the rows should be the columns, and vv. I was just wondering if this was a DSDT bug that might have been seen elsewhere before.

No this does not ring a bell, but the x and y axis being swapped does seem related to the LCD panel being 90° rotated.

Quoted 4 lines

BTW, there is another matrix in the DSTD, but I can't find the associated HID (10EC5280) anywhere (Linux sysfs or Windows Powershell system data extract). It's not a correct matrix, though - could it be just a bit of redundant code in the DST?

Yes that is likely there often is a bunch of dead stuff DSDT leftover from other device models.

Regards,

Hans

Quoted 213 lines

Darrell

Quoted 208 lines · Hans de Goede

Hi,

Quoted 3 lines · Darrell Kavanagh

I've just noticed that the working mount matrix that I added to my hwdb is the matrix retrieved from the ACPI ROTM call in the amended driver, transposed.

An other word for the mount matrix would be a rotation matrix, since it defines how the physical sensor is mounted on the PCB in a rotated fashion compared to its standard orientation.

The x, y, z axis relationship underling of course does not change by the rotation, so yes all mount matrices are a transposition of the standard:

1, 0, 0 : 0, 1, 0 : 0, 0, 1

matrix, that is expected. Where that to not be the case then there would be a bug in the accelerometer driver itself where the driver itself is swapping or inverting axis.

Regards,

Hans

Quoted 178 lines
Quoted 175 lines · Darrell Kavanagh

Finally got a 6.2.0-rc6 kernel built and installed, with the following patch, and everything is working as expected.

Moving on now to look at Bastien's suggestion.

Thanks,
Darrell

kernel/linux-6.2-rc6/drivers/gpu/drm/drm_panel_orientation_quirks.c renamed from kernel/drm_panel_orientation_quirks.c +6 -0
index 3659f04..590bb7b 100644
@@ -304,6 +304,12 @@ static const struct dmi_system_id orientation_data[] = {
                  DMI_EXACT_MATCH(DMI_PRODUCT_VERSION, "Lenovo ideapad
 D330-10IGM"),
                },
                .driver_data = (void *)&lcd1200x1920_rightside_up,
+       }, {    /* Lenovo IdeaPad Duet 3 10IGL5 */
+               .matches = {
+                 DMI_EXACT_MATCH(DMI_SYS_VENDOR, "LENOVO"),
+                 DMI_EXACT_MATCH(DMI_PRODUCT_VERSION, "IdeaPad Duet 3 10IGL5"),
+               },
+               .driver_data = (void *)&lcd1200x1920_rightside_up,
        }, {    /* Lenovo Ideapad D330-10IGL (HD) */
                .matches = {

DMI_EXACT_MATCH(DMI_SYS_VENDOR, "LENOVO"),

Quoted 145 lines · Hans de Goede

Hi,

Quoted 8 lines · Darrell Kavanagh

Thank you. I don't have anything that could be called a big machine. The fastest processor I have access to is a Core m3-8100Y - that's in a Chromebook with 4GB memory - it can run Linux in a chroot or officially in Google's VM. I also have an ancient gen 2 core i5-2410M machine which is slower than the m3 in theory, but that has 6GB of memory.

Is the kernel build more processor or memory bound?

It is mostly processor bound, esp. wtih something like make -j4, make -j16 will start taking some RAM, but with make -j4 I expect you to be fully CPU bound.

Regards,

Hans

Quoted 114 lines · Bastien Nocera
Quoted 112 lines · Hans de Goede
Quoted 93 lines

Hi,

Quoted 34 lines · Jonathan Cameron
Quoted 24 lines · On Wed, 1 Feb 2023 01:40:49 +0000 Darrell Kavanagh…

Hello, all.

I've finally reached a conclusion on this, after testing all the combinations of the patches (with and without reading the acpi mounting matrix), window managers (wayland, xorg) and the presence or
not of my custom kernel parms.

What works well is the full set of patches with the custom kernel parms and a new hwdb entry for the sensor:

sensor:modalias:acpi:SMO8B30*:dmi:*:svnLENOVO*:pn82AT:*
ACCEL_MOUNT_MATRIX=0, 1, 0; -1, 0, 0; 0, 0, 1

The autorotate then works correctly in wayland and xorg, but for xorg,
the settings say the screen is "portrait left" when in actual fact it
is in standard laptop landscape orientation. Wayland does not have
this problem (I guess because wayland's view of the screen is straight
from the kernel).

Quoted 6 lines

Without the hwdb entry, the orientation is 90 degrees out without using the acpi matrix and 180 degrees out when using it. I could have
gone either way here with appropriate hwdb entries, but my view is
that we *should* be using the matrix.

Quoted 3 lines

Added Hans de Goede as he has probably run into more of this mess than anyone else. Hans, any thoughts on if we are doing something wrong on kernel side? Or is the matrix just wrong *sigh*

I see below that this laptop has a panel which is mounted 90 degrees rotated, that likely explains why the ACPI matrix does not work. So the best thing to do here is to just override it with a hwdb entries.

IIRC there are already 1 or 2 other hwdb entries which actually override the ACPI provided matrix because of similar issues.

Linux userspace expects the matrix in this case to be set so that it causes e.g. gnome's auto-rotation to put the image upright even with older gnome versions / mate / xfce which don't know about the panel being mounted 90 degrees.

So e.g. "monitor-sensor" will report left-side-up or right-side-up while the device is actually in normal clamshell mode with the display up-right.

This reporting of left-side-up or right-side-up is actually "correct" looking from the native LCD panel orientation and as mentioned is done for backward compatibility. This is documented here:

https://github.com/systemd/systemd/blob/main/hwdb.d/60-sensor.hwdb#L54

The way we are handling this is likely incompatible with how Windows handles this special case of 90° rotated screen + ROTM. Or the matrix in the ACPI tables could be just wrong...

Quoted 2 lines

I think 'ROTM' is defined by MS.
https://learn.microsoft.com/en-us/windows-hardware/drivers/sensors/sensors-acpi-entries

Right and as such it would be good if we can still add support to it to the sensor driver in question. Because the ROTM info usually is correct and avoids the need for adding more and more hwdb entries.

Note there already is existing support in some other sensor drivers.

So we probably need to factor out some helper code for this and share that between sensor drivers.

Quoted 11 lines
Quoted 11 lines

The only thing that concerns me is the need for custom kernel parms.
It would be better if there was a way to avoid this, so that the user
didn't have to mess around with their grub config. Though having said
that, the sensors fix as we have it doesn't make things worse - under
currently released kernels the screen always starts up sideways unless
custom parms are added in grub.

Quoted 7 lines

We actually have a quirk mechanism in the kernel for specifying the need for: video=DSI-1:panel_orientation=right_side_up and this will also automatically fix the fbcon orientation, see:

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/gpu/drm/drm_panel_orientation_quirks.c

If you submit a patch for this upstream please Cc me.

And if after that change, and copy/pasting the orientation from the DSDT into hwdb the sensor and screen move in the expected ways, then maybe stealing the BMC150 driver's bmc150_apply_bosc0200_acpi_orientation() might be a good idea.

Once exported through "mount_matrix", iio-sensor-proxy should see it and read it without the need for a hwdb entry.

Cheers

#30 lore raw

So does this mean that the least worst (only?) option is to get my hwdb mount matrix entry added to systemd? I can raise a bug as suggested in hwdb.d/60-sensor.hwdb if so.

Darrell

Quoted 286 lines · Jonathan Cameron
Quoted 17 lines · On Sun, 5 Feb 2023 09:50:51 +0100 Hans de Goede <[email protected]> wrote:

Hi,

Quoted 14 lines · Darrell Kavanagh

Yes, I understand that.

What I mean is that the matrix read from the DSDT by Jonathan's amended driver is

 0 -1  0
 1  0  0
 0  0  1

and the (correct) matrix created with my new hwdb entry is

 0  1  0
-1  0  0
 0  0  1

May be concidence, but I think that's the inverse of the one we are reading from ROTM - so represents the transform in the other direction.

The way ROTM is defined is that first row represents the direction of the x axis in device coordinates - so it's the transform from sensor to device space.

I wonder if the hwdb matrix is defined from world space to sensor? Seems unlikely.

The IIO ABI docs describe mount matrix as being what you apply to data to tranform into device space (oh for a diagram in the docs). Anyhow my reading is that matches with ROTM definition but maybe I'm reading that wrong...

For extra annoyance, the ROTM matrix on this device isn't a rotation matrix. It's flipping the handedness of the sensor. Determinant isn't -1 which it should be. I guess the sensor itself might have an axis backwards from windows convention though *sigh* I think windows uses left handed convention and looks like sensor is using right handed (which I think is what Android and similar use).

Quoted 242 lines
Quoted 8 lines

which is the algebraic transposition (ie reflection in the diagonal) of the DSDT one.

In other words, though the DST matrix is wrong, it is wrong in a specific way - the rows should be the columns, and vv. I was just wondering if this was a DSDT bug that might have been seen elsewhere before.

No this does not ring a bell, but the x and y axis being swapped does seem related to the LCD panel being 90° rotated.

Quoted 4 lines

BTW, there is another matrix in the DSTD, but I can't find the associated HID (10EC5280) anywhere (Linux sysfs or Windows Powershell system data extract). It's not a correct matrix, though - could it be just a bit of redundant code in the DST?

Yes that is likely there often is a bunch of dead stuff DSDT leftover from other device models.

Regards,

Hans

Quoted 213 lines

Darrell

Quoted 208 lines · Hans de Goede

Hi,

Quoted 3 lines · Darrell Kavanagh

I've just noticed that the working mount matrix that I added to my hwdb is the matrix retrieved from the ACPI ROTM call in the amended driver, transposed.

An other word for the mount matrix would be a rotation matrix, since it defines how the physical sensor is mounted on the PCB in a rotated fashion compared to its standard orientation.

The x, y, z axis relationship underling of course does not change by the rotation, so yes all mount matrices are a transposition of the standard:

1, 0, 0 : 0, 1, 0 : 0, 0, 1

matrix, that is expected. Where that to not be the case then there would be a bug in the accelerometer driver itself where the driver itself is swapping or inverting axis.

Regards,

Hans

Quoted 178 lines
Quoted 175 lines · Darrell Kavanagh

Finally got a 6.2.0-rc6 kernel built and installed, with the following patch, and everything is working as expected.

Moving on now to look at Bastien's suggestion.

Thanks,
Darrell

kernel/linux-6.2-rc6/drivers/gpu/drm/drm_panel_orientation_quirks.c renamed from kernel/drm_panel_orientation_quirks.c +6 -0
index 3659f04..590bb7b 100644
@@ -304,6 +304,12 @@ static const struct dmi_system_id orientation_data[] = {
                  DMI_EXACT_MATCH(DMI_PRODUCT_VERSION, "Lenovo ideapad
 D330-10IGM"),
                },
                .driver_data = (void *)&lcd1200x1920_rightside_up,
+       }, {    /* Lenovo IdeaPad Duet 3 10IGL5 */
+               .matches = {
+                 DMI_EXACT_MATCH(DMI_SYS_VENDOR, "LENOVO"),
+                 DMI_EXACT_MATCH(DMI_PRODUCT_VERSION, "IdeaPad Duet 3 10IGL5"),
+               },
+               .driver_data = (void *)&lcd1200x1920_rightside_up,
        }, {    /* Lenovo Ideapad D330-10IGL (HD) */
                .matches = {

DMI_EXACT_MATCH(DMI_SYS_VENDOR, "LENOVO"),

Quoted 145 lines · Hans de Goede

Hi,

Quoted 8 lines · Darrell Kavanagh

Thank you. I don't have anything that could be called a big machine. The fastest processor I have access to is a Core m3-8100Y - that's in a Chromebook with 4GB memory - it can run Linux in a chroot or officially in Google's VM. I also have an ancient gen 2 core i5-2410M machine which is slower than the m3 in theory, but that has 6GB of memory.

Is the kernel build more processor or memory bound?

It is mostly processor bound, esp. wtih something like make -j4, make -j16 will start taking some RAM, but with make -j4 I expect you to be fully CPU bound.

Regards,

Hans

Quoted 114 lines · Bastien Nocera
Quoted 112 lines · Hans de Goede
Quoted 93 lines

Hi,

Quoted 34 lines · Jonathan Cameron
Quoted 24 lines · On Wed, 1 Feb 2023 01:40:49 +0000 Darrell Kavanagh…

Hello, all.

I've finally reached a conclusion on this, after testing all the combinations of the patches (with and without reading the acpi mounting matrix), window managers (wayland, xorg) and the presence or
not of my custom kernel parms.

What works well is the full set of patches with the custom kernel parms and a new hwdb entry for the sensor:

sensor:modalias:acpi:SMO8B30*:dmi:*:svnLENOVO*:pn82AT:*
ACCEL_MOUNT_MATRIX=0, 1, 0; -1, 0, 0; 0, 0, 1

The autorotate then works correctly in wayland and xorg, but for xorg,
the settings say the screen is "portrait left" when in actual fact it
is in standard laptop landscape orientation. Wayland does not have
this problem (I guess because wayland's view of the screen is straight
from the kernel).

Quoted 6 lines

Without the hwdb entry, the orientation is 90 degrees out without using the acpi matrix and 180 degrees out when using it. I could have
gone either way here with appropriate hwdb entries, but my view is
that we *should* be using the matrix.

Quoted 3 lines

Added Hans de Goede as he has probably run into more of this mess than anyone else. Hans, any thoughts on if we are doing something wrong on kernel side? Or is the matrix just wrong *sigh*

I see below that this laptop has a panel which is mounted 90 degrees rotated, that likely explains why the ACPI matrix does not work. So the best thing to do here is to just override it with a hwdb entries.

IIRC there are already 1 or 2 other hwdb entries which actually override the ACPI provided matrix because of similar issues.

Linux userspace expects the matrix in this case to be set so that it causes e.g. gnome's auto-rotation to put the image upright even with older gnome versions / mate / xfce which don't know about the panel being mounted 90 degrees.

So e.g. "monitor-sensor" will report left-side-up or right-side-up while the device is actually in normal clamshell mode with the display up-right.

This reporting of left-side-up or right-side-up is actually "correct" looking from the native LCD panel orientation and as mentioned is done for backward compatibility. This is documented here:

https://github.com/systemd/systemd/blob/main/hwdb.d/60-sensor.hwdb#L54

The way we are handling this is likely incompatible with how Windows handles this special case of 90° rotated screen + ROTM. Or the matrix in the ACPI tables could be just wrong...

Quoted 2 lines

I think 'ROTM' is defined by MS.
https://learn.microsoft.com/en-us/windows-hardware/drivers/sensors/sensors-acpi-entries

Right and as such it would be good if we can still add support to it to the sensor driver in question. Because the ROTM info usually is correct and avoids the need for adding more and more hwdb entries.

Note there already is existing support in some other sensor drivers.

So we probably need to factor out some helper code for this and share that between sensor drivers.

Quoted 11 lines
Quoted 11 lines

The only thing that concerns me is the need for custom kernel parms.
It would be better if there was a way to avoid this, so that the user
didn't have to mess around with their grub config. Though having said
that, the sensors fix as we have it doesn't make things worse - under
currently released kernels the screen always starts up sideways unless
custom parms are added in grub.

Quoted 7 lines

We actually have a quirk mechanism in the kernel for specifying the need for: video=DSI-1:panel_orientation=right_side_up and this will also automatically fix the fbcon orientation, see:

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/gpu/drm/drm_panel_orientation_quirks.c

If you submit a patch for this upstream please Cc me.

And if after that change, and copy/pasting the orientation from the DSDT into hwdb the sensor and screen move in the expected ways, then maybe stealing the BMC150 driver's bmc150_apply_bosc0200_acpi_orientation() might be a good idea.

Once exported through "mount_matrix", iio-sensor-proxy should see it and read it without the need for a hwdb entry.

Cheers

#31 lore raw

Hi,

Quoted 3 lines · Darrell Kavanagh

So does this mean that the least worst (only?) option is to get my hwdb mount matrix entry added to systemd? I can raise a bug as suggested in hwdb.d/60-sensor.hwdb if so.

Yes you should add a hwdb entry for this, note just submitting a pull-req with the fix is better then filing an issue for this.

Regards,

Hans

Quoted 288 lines · Jonathan Cameron
Quoted 286 lines
Quoted 17 lines · On Sun, 5 Feb 2023 09:50:51 +0100 Hans de Goede <[email protected]> wrote:

Hi,

Quoted 14 lines · Darrell Kavanagh

Yes, I understand that.

What I mean is that the matrix read from the DSDT by Jonathan's amended driver is

 0 -1  0
 1  0  0
 0  0  1

and the (correct) matrix created with my new hwdb entry is

 0  1  0
-1  0  0
 0  0  1

May be concidence, but I think that's the inverse of the one we are reading from ROTM - so represents the transform in the other direction.

The way ROTM is defined is that first row represents the direction of the x axis in device coordinates - so it's the transform from sensor to device space.

I wonder if the hwdb matrix is defined from world space to sensor? Seems unlikely.

The IIO ABI docs describe mount matrix as being what you apply to data to tranform into device space (oh for a diagram in the docs). Anyhow my reading is that matches with ROTM definition but maybe I'm reading that wrong...

For extra annoyance, the ROTM matrix on this device isn't a rotation matrix. It's flipping the handedness of the sensor. Determinant isn't -1 which it should be. I guess the sensor itself might have an axis backwards from windows convention though *sigh* I think windows uses left handed convention and looks like sensor is using right handed (which I think is what Android and similar use).

Quoted 242 lines
Quoted 8 lines

which is the algebraic transposition (ie reflection in the diagonal) of the DSDT one.

In other words, though the DST matrix is wrong, it is wrong in a specific way - the rows should be the columns, and vv. I was just wondering if this was a DSDT bug that might have been seen elsewhere before.

No this does not ring a bell, but the x and y axis being swapped does seem related to the LCD panel being 90° rotated.

Quoted 4 lines

BTW, there is another matrix in the DSTD, but I can't find the associated HID (10EC5280) anywhere (Linux sysfs or Windows Powershell system data extract). It's not a correct matrix, though - could it be just a bit of redundant code in the DST?

Yes that is likely there often is a bunch of dead stuff DSDT leftover from other device models.

Regards,

Hans

Quoted 213 lines

Darrell

Quoted 208 lines · Hans de Goede

Hi,

Quoted 3 lines · Darrell Kavanagh

I've just noticed that the working mount matrix that I added to my hwdb is the matrix retrieved from the ACPI ROTM call in the amended driver, transposed.

An other word for the mount matrix would be a rotation matrix, since it defines how the physical sensor is mounted on the PCB in a rotated fashion compared to its standard orientation.

The x, y, z axis relationship underling of course does not change by the rotation, so yes all mount matrices are a transposition of the standard:

1, 0, 0 : 0, 1, 0 : 0, 0, 1

matrix, that is expected. Where that to not be the case then there would be a bug in the accelerometer driver itself where the driver itself is swapping or inverting axis.

Regards,

Hans

Quoted 178 lines
Quoted 175 lines · Darrell Kavanagh

Finally got a 6.2.0-rc6 kernel built and installed, with the following patch, and everything is working as expected.

Moving on now to look at Bastien's suggestion.

Thanks,
Darrell

kernel/linux-6.2-rc6/drivers/gpu/drm/drm_panel_orientation_quirks.c renamed from kernel/drm_panel_orientation_quirks.c +6 -0
index 3659f04..590bb7b 100644
@@ -304,6 +304,12 @@ static const struct dmi_system_id orientation_data[] = {
                  DMI_EXACT_MATCH(DMI_PRODUCT_VERSION, "Lenovo ideapad
 D330-10IGM"),
                },
                .driver_data = (void *)&lcd1200x1920_rightside_up,
+       }, {    /* Lenovo IdeaPad Duet 3 10IGL5 */
+               .matches = {
+                 DMI_EXACT_MATCH(DMI_SYS_VENDOR, "LENOVO"),
+                 DMI_EXACT_MATCH(DMI_PRODUCT_VERSION, "IdeaPad Duet 3 10IGL5"),
+               },
+               .driver_data = (void *)&lcd1200x1920_rightside_up,
        }, {    /* Lenovo Ideapad D330-10IGL (HD) */
                .matches = {

DMI_EXACT_MATCH(DMI_SYS_VENDOR, "LENOVO"),

Quoted 145 lines · Hans de Goede

Hi,

Quoted 8 lines · Darrell Kavanagh

Thank you. I don't have anything that could be called a big machine. The fastest processor I have access to is a Core m3-8100Y - that's in a Chromebook with 4GB memory - it can run Linux in a chroot or officially in Google's VM. I also have an ancient gen 2 core i5-2410M machine which is slower than the m3 in theory, but that has 6GB of memory.

Is the kernel build more processor or memory bound?

It is mostly processor bound, esp. wtih something like make -j4, make -j16 will start taking some RAM, but with make -j4 I expect you to be fully CPU bound.

Regards,

Hans

Quoted 114 lines · Bastien Nocera
Quoted 112 lines · Hans de Goede
Quoted 93 lines

Hi,

Quoted 34 lines · Jonathan Cameron
Quoted 30 lines · On Wed, 1 Feb 2023 01:40:49 +0000 Darrell Kavanagh…

Hello, all.

I've finally reached a conclusion on this, after testing all the
combinations of the patches (with and without reading the acpi
mounting matrix), window managers (wayland, xorg) and the
presence or
not of my custom kernel parms.

What works well is the full set of patches with the custom kernel
parms and a new hwdb entry for the sensor:

sensor:modalias:acpi:SMO8B30*:dmi:*:svnLENOVO*:pn82AT:*
ACCEL_MOUNT_MATRIX=0, 1, 0; -1, 0, 0; 0, 0, 1

The autorotate then works correctly in wayland and xorg, but for
xorg,
the settings say the screen is "portrait left" when in actual
fact it
is in standard laptop landscape orientation. Wayland does not
have
this problem (I guess because wayland's view of the screen is
straight
from the kernel).

Without the hwdb entry, the orientation is 90 degrees out without
using the acpi matrix and 180 degrees out when using it. I could
have
gone either way here with appropriate hwdb entries, but my view
is
that we *should* be using the matrix.

Quoted 3 lines

Added Hans de Goede as he has probably run into more of this mess than anyone else. Hans, any thoughts on if we are doing something wrong on kernel side? Or is the matrix just wrong *sigh*

I see below that this laptop has a panel which is mounted 90 degrees rotated, that likely explains why the ACPI matrix does not work. So the best thing to do here is to just override it with a hwdb entries.

IIRC there are already 1 or 2 other hwdb entries which actually override the ACPI provided matrix because of similar issues.

Linux userspace expects the matrix in this case to be set so that it causes e.g. gnome's auto-rotation to put the image upright even with older gnome versions / mate / xfce which don't know about the panel being mounted 90 degrees.

So e.g. "monitor-sensor" will report left-side-up or right-side-up while the device is actually in normal clamshell mode with the display up-right.

This reporting of left-side-up or right-side-up is actually "correct" looking from the native LCD panel orientation and as mentioned is done for backward compatibility. This is documented here:

https://github.com/systemd/systemd/blob/main/hwdb.d/60-sensor.hwdb#L54

The way we are handling this is likely incompatible with how Windows handles this special case of 90° rotated screen + ROTM. Or the matrix in the ACPI tables could be just wrong...

Quoted 2 lines

I think 'ROTM' is defined by MS.
https://learn.microsoft.com/en-us/windows-hardware/drivers/sensors/sensors-acpi-entries

Right and as such it would be good if we can still add support to it to the sensor driver in question. Because the ROTM info usually is correct and avoids the need for adding more and more hwdb entries.

Note there already is existing support in some other sensor drivers.

So we probably need to factor out some helper code for this and share that between sensor drivers.

Quoted 11 lines
Quoted 11 lines

The only thing that concerns me is the need for custom kernel
parms.
It would be better if there was a way to avoid this, so that the
user
didn't have to mess around with their grub config. Though having
said
that, the sensors fix as we have it doesn't make things worse -
under
currently released kernels the screen always starts up sideways
unless
custom parms are added in grub.

Quoted 7 lines

We actually have a quirk mechanism in the kernel for specifying the need for: video=DSI-1:panel_orientation=right_side_up and this will also automatically fix the fbcon orientation, see:

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/gpu/drm/drm_panel_orientation_quirks.c

If you submit a patch for this upstream please Cc me.

And if after that change, and copy/pasting the orientation from the DSDT into hwdb the sensor and screen move in the expected ways, then maybe stealing the BMC150 driver's bmc150_apply_bosc0200_acpi_orientation() might be a good idea.

Once exported through "mount_matrix", iio-sensor-proxy should see it and read it without the need for a hwdb entry.

Cheers

#32 lore raw

OK, pull request has been submitted.
https://github.com/systemd/systemd/pull/26317

Quoted 306 lines · Hans de Goede

Hi,

Quoted 3 lines · Darrell Kavanagh

So does this mean that the least worst (only?) option is to get my hwdb mount matrix entry added to systemd? I can raise a bug as suggested in hwdb.d/60-sensor.hwdb if so.

Yes you should add a hwdb entry for this, note just submitting a pull-req with the fix is better then filing an issue for this.

Regards,

Hans

Quoted 288 lines · Jonathan Cameron
Quoted 286 lines
Quoted 17 lines · On Sun, 5 Feb 2023 09:50:51 +0100 Hans de Goede <[email protected]> wrote:

Hi,

Quoted 14 lines · Darrell Kavanagh

Yes, I understand that.

What I mean is that the matrix read from the DSDT by Jonathan's amended driver is

 0 -1  0
 1  0  0
 0  0  1

and the (correct) matrix created with my new hwdb entry is

 0  1  0
-1  0  0
 0  0  1

May be concidence, but I think that's the inverse of the one we are reading from ROTM - so represents the transform in the other direction.

The way ROTM is defined is that first row represents the direction of the x axis in device coordinates - so it's the transform from sensor to device space.

I wonder if the hwdb matrix is defined from world space to sensor? Seems unlikely.

The IIO ABI docs describe mount matrix as being what you apply to data to tranform into device space (oh for a diagram in the docs). Anyhow my reading is that matches with ROTM definition but maybe I'm reading that wrong...

For extra annoyance, the ROTM matrix on this device isn't a rotation matrix. It's flipping the handedness of the sensor. Determinant isn't -1 which it should be. I guess the sensor itself might have an axis backwards from windows convention though *sigh* I think windows uses left handed convention and looks like sensor is using right handed (which I think is what Android and similar use).

Quoted 242 lines
Quoted 8 lines

which is the algebraic transposition (ie reflection in the diagonal) of the DSDT one.

In other words, though the DST matrix is wrong, it is wrong in a specific way - the rows should be the columns, and vv. I was just wondering if this was a DSDT bug that might have been seen elsewhere before.

No this does not ring a bell, but the x and y axis being swapped does seem related to the LCD panel being 90° rotated.

Quoted 4 lines

BTW, there is another matrix in the DSTD, but I can't find the associated HID (10EC5280) anywhere (Linux sysfs or Windows Powershell system data extract). It's not a correct matrix, though - could it be just a bit of redundant code in the DST?

Yes that is likely there often is a bunch of dead stuff DSDT leftover from other device models.

Regards,

Hans

Quoted 213 lines

Darrell

Quoted 208 lines · Hans de Goede

Hi,

Quoted 3 lines · Darrell Kavanagh

I've just noticed that the working mount matrix that I added to my hwdb is the matrix retrieved from the ACPI ROTM call in the amended driver, transposed.

An other word for the mount matrix would be a rotation matrix, since it defines how the physical sensor is mounted on the PCB in a rotated fashion compared to its standard orientation.

The x, y, z axis relationship underling of course does not change by the rotation, so yes all mount matrices are a transposition of the standard:

1, 0, 0 : 0, 1, 0 : 0, 0, 1

matrix, that is expected. Where that to not be the case then there would be a bug in the accelerometer driver itself where the driver itself is swapping or inverting axis.

Regards,

Hans

Quoted 178 lines
Quoted 175 lines · Darrell Kavanagh

Finally got a 6.2.0-rc6 kernel built and installed, with the following patch, and everything is working as expected.

Moving on now to look at Bastien's suggestion.

Thanks,
Darrell

kernel/linux-6.2-rc6/drivers/gpu/drm/drm_panel_orientation_quirks.c renamed from kernel/drm_panel_orientation_quirks.c +6 -0
index 3659f04..590bb7b 100644
@@ -304,6 +304,12 @@ static const struct dmi_system_id orientation_data[] = {
                  DMI_EXACT_MATCH(DMI_PRODUCT_VERSION, "Lenovo ideapad
 D330-10IGM"),
                },
                .driver_data = (void *)&lcd1200x1920_rightside_up,
+       }, {    /* Lenovo IdeaPad Duet 3 10IGL5 */
+               .matches = {
+                 DMI_EXACT_MATCH(DMI_SYS_VENDOR, "LENOVO"),
+                 DMI_EXACT_MATCH(DMI_PRODUCT_VERSION, "IdeaPad Duet 3 10IGL5"),
+               },
+               .driver_data = (void *)&lcd1200x1920_rightside_up,
        }, {    /* Lenovo Ideapad D330-10IGL (HD) */
                .matches = {

DMI_EXACT_MATCH(DMI_SYS_VENDOR, "LENOVO"),

Quoted 145 lines · Hans de Goede

Hi,

Quoted 8 lines · Darrell Kavanagh

Thank you. I don't have anything that could be called a big machine. The fastest processor I have access to is a Core m3-8100Y - that's in a Chromebook with 4GB memory - it can run Linux in a chroot or officially in Google's VM. I also have an ancient gen 2 core i5-2410M machine which is slower than the m3 in theory, but that has 6GB of memory.

Is the kernel build more processor or memory bound?

It is mostly processor bound, esp. wtih something like make -j4, make -j16 will start taking some RAM, but with make -j4 I expect you to be fully CPU bound.

Regards,

Hans

Quoted 114 lines · Bastien Nocera
Quoted 112 lines · Hans de Goede
Quoted 93 lines

Hi,

Quoted 37 lines · Jonathan Cameron

On Wed, 1 Feb 2023 01:40:49 +0000
Darrell Kavanagh <[email protected]> wrote:

Hello, all.

I've finally reached a conclusion on this, after testing all the
combinations of the patches (with and without reading the acpi
mounting matrix), window managers (wayland, xorg) and the
presence or
not of my custom kernel parms.

What works well is the full set of patches with the custom kernel
parms and a new hwdb entry for the sensor:

sensor:modalias:acpi:SMO8B30*:dmi:*:svnLENOVO*:pn82AT:*
ACCEL_MOUNT_MATRIX=0, 1, 0; -1, 0, 0; 0, 0, 1

The autorotate then works correctly in wayland and xorg, but for
xorg,
the settings say the screen is "portrait left" when in actual
fact it
is in standard laptop landscape orientation. Wayland does not
have
this problem (I guess because wayland's view of the screen is
straight
from the kernel).

Without the hwdb entry, the orientation is 90 degrees out without
using the acpi matrix and 180 degrees out when using it. I could
have
gone either way here with appropriate hwdb entries, but my view
is
that we *should* be using the matrix.

Added Hans de Goede as he has probably run into more of this mess
than anyone else. Hans, any thoughts on if we are doing something
wrong on kernel side? Or is the matrix just wrong *sigh*

I see below that this laptop has a panel which is mounted 90 degrees rotated, that likely explains why the ACPI matrix does not work. So the best thing to do here is to just override it with a hwdb entries.

IIRC there are already 1 or 2 other hwdb entries which actually override the ACPI provided matrix because of similar issues.

Linux userspace expects the matrix in this case to be set so that it causes e.g. gnome's auto-rotation to put the image upright even with older gnome versions / mate / xfce which don't know about the panel being mounted 90 degrees.

So e.g. "monitor-sensor" will report left-side-up or right-side-up while the device is actually in normal clamshell mode with the display up-right.

This reporting of left-side-up or right-side-up is actually "correct" looking from the native LCD panel orientation and as mentioned is done for backward compatibility. This is documented here:

https://github.com/systemd/systemd/blob/main/hwdb.d/60-sensor.hwdb#L54

The way we are handling this is likely incompatible with how Windows handles this special case of 90° rotated screen + ROTM. Or the matrix in the ACPI tables could be just wrong...

Quoted 2 lines

I think 'ROTM' is defined by MS.
https://learn.microsoft.com/en-us/windows-hardware/drivers/sensors/sensors-acpi-entries

Right and as such it would be good if we can still add support to it to the sensor driver in question. Because the ROTM info usually is correct and avoids the need for adding more and more hwdb entries.

Note there already is existing support in some other sensor drivers.

So we probably need to factor out some helper code for this and share that between sensor drivers.

Quoted 11 lines

The only thing that concerns me is the need for custom kernel
parms.
It would be better if there was a way to avoid this, so that the
user
didn't have to mess around with their grub config. Though having
said
that, the sensors fix as we have it doesn't make things worse -
under
currently released kernels the screen always starts up sideways
unless
custom parms are added in grub.

Quoted 7 lines

We actually have a quirk mechanism in the kernel for specifying the need for: video=DSI-1:panel_orientation=right_side_up and this will also automatically fix the fbcon orientation, see:

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/gpu/drm/drm_panel_orientation_quirks.c

If you submit a patch for this upstream please Cc me.

And if after that change, and copy/pasting the orientation from the DSDT into hwdb the sensor and screen move in the expected ways, then maybe stealing the BMC150 driver's bmc150_apply_bosc0200_acpi_orientation() might be a good idea.

Once exported through "mount_matrix", iio-sensor-proxy should see it and read it without the need for a hwdb entry.

Cheers