Re: [PATCH 1/1] Add 10EC5280 to bmi160_i2c ACPI IDs to allow binding on some devices

[Jesus Miguel Gonzalez Herrero <jesusmgh@gmail.com>]

Hello Mr. Cameron

First of all thank you for reviewing the patch.

And I most definitely agree with you and Mr. Shevchenko: this absolutely
is a firmware bug that manufacturers should fix. For this reason some
people started talks with affected manufacturers to change it. In my
case it was with GPD, together with some others, including some which
historically had a more direct line with them. This was finally dismissed
as WONTFIX, since their main focus is Windows and their driver supports
the ID, so the end result of those conversations is a lack of a fixed
firmware, and a surplus of frustration.

As far as I know people have been in talks with Aya too, and I do not
know the status of conversations with Lenovo or other manufacturers. I
do not know of any conversation with Realtek, besides what was mentioned
in those emails you linked to from 2021.

I will amend the patch to include a big disclaimer and the reason as
a comment in the code, and send it again in reply to this message. I
don't think I'd go as far as tainting the kernel, but I'm not opposed,
happy anyway if the IMU finally becomes usable, and VERY far from any
expertise whatsoever concerning kernel development!

Here is the relevant extract from the DSDT of my GPD Win Max 2 (AMD
6800U model) with the latest firmware 1.05 installed.

     Scope (_SB.I2CC) {
         Device (BMA2) {
             Name (_ADR, Zero)  // _ADR: Address Name (_HID, "10EC5280")
             // _HID: Hardware ID Name (_CID, "10EC5280")  // _CID:
             Compatible ID Name (_DDN, "Accelerometer")  // _DDN: DOS
             Device Name Name (_UID, One)  // _UID: Unique ID Method
             (_CRS, 0, NotSerialized)  // _CRS: Current Resource Settings {
                 Name (RBUF, ResourceTemplate () {
                     I2cSerialBusV2 (0x0069, ControllerInitiated,
                     0x00061A80,
                         AddressingMode7Bit, "\\_SB.I2CC", 0x00,
                         ResourceConsumer, , Exclusive, )
                 }) Return (RBUF) /* \_SB_.I2CC.BMA2._CRS.RBUF */
             }

             OperationRegion (CMS2, SystemIO, 0x72, 0x02) Field (CMS2,
             ByteAcc, NoLock, Preserve) {
                 IND2,   8, DAT2,   8
             }

             IndexField (IND2, DAT2, ByteAcc, NoLock, Preserve) {
                 Offset (0x74), BACS,   32
             }

             Method (ROMS, 0, NotSerialized) {
                 Name (RBUF, Package (0x03) {
                     "0 -1 0", "-1 0 0", "0 0 1"
                 }) Return (RBUF) /* \_SB_.I2CC.BMA2.ROMS.RBUF */
             }

             Method (CALS, 1, NotSerialized) {
                 Local0 = Arg0 If (((Local0 == Zero) || (Local0 ==
                 Ones))) {
                     Return (Local0)
                 } Else {
                     BACS = Local0
                 }
             }

             Method (_STA, 0, NotSerialized)  // _STA: Status {
                 Return (0x0F)
             }
         }
     }

Thank you for taking this into consideration!

Jesus Gonzalez



On 04/02/2024 15:00, Jonathan Cameron wrote:
> On Fri,  2 Feb 2024 18:30:41 +0100
> Jesus Gonzalez <jesusmgh@gmail.com> wrote:
>
>> "10EC5280" is used by several manufacturers like Lenovo, GPD, or AYA (and
>> probably others) in their ACPI table as the ID for the bmi160 IMU. This
>> means the bmi160_i2c driver won't bind to it, and the IMU is unavailable
>> to the user. Manufacturers have been approached on several occasions to
>> try getting a BIOS with a fixed ID, mostly without actual positive
>> results, and since affected devices are already a few years old, this is
>> not expected to change. This patch enables using the bmi160_i2c driver for
>> the bmi160 IMU on these devices.
> Hi Jesus,
>
> https://lore.kernel.org/lkml/CAHp75Vct-AXnU7QQmdE7nyYZT-=n=p67COPLiiZTet7z7snL-g@mail.gmail.com/
> Lays out what Andy (and for that matter I) consider necessary for such
> a patch.
>
> In short, we want to see devices called out here - with a DSDT section.
> + a clear comment in the code.
>
> The big problem here is this tramples on Realtech's ID space. It's not just
> a made up code (incidentally the BMI0160 isn't valid either),
> it's a valid code but for an entirely different (PCI) device.
>
> So we need as much info as possible in the patch description and the driver
> itself to justify carrying this.   Tempting to add a firmware bug taint on
> it as well but that might scare people :)
>
> Jonathan
>
>
>> Signed-off-by: Jesus Gonzalez <jesusmgh@gmail.com>
>> ---
>> A device-specific transformation matrix can then be provided in a second
>> step through udev hwdb.
>>
>> This has been discussed before in 2021, see here:
>> https://lore.kernel.org/lkml/CACAwPwYQHRcrabw9=0tvenPzAcwwW1pTaR6a+AEWBF9Hqf_wXQ@mail.gmail.com/
>>
>> Lenovo, as an example of a big manufacturer, is also using this ID:
>> https://www.reddit.com/r/linux/comments/r6f9de/comment/hr8bdfs/?context=3
>>
>> At least some discussions with GPD took place on the GPD server Discord,
>> for which I can provide proof on demand via screenshot (if not accessible
>> directly).
>>
>> I have read the patch submission instructions and followed them to the
>> best of my knowledge. Still, this is my first kernel patch submission,
>> so I'd be glad if you could please point out any mistakes. Thank you!
>>
>>
>>   drivers/iio/imu/bmi160/bmi160_spi.c | 1 +
>>   1 file changed, 1 insertion(+)
>>
>> diff --git a/drivers/iio/imu/bmi160/bmi160_spi.c b/drivers/iio/imu/bmi160/bmi160_spi.c
>> index 8b573ea99af2..0874c37c6670 100644
>> --- a/drivers/iio/imu/bmi160/bmi160_spi.c
>> +++ b/drivers/iio/imu/bmi160/bmi160_spi.c
>> @@ -41,6 +41,7 @@ MODULE_DEVICE_TABLE(spi, bmi160_spi_id);
>>   
>>   static const struct acpi_device_id bmi160_acpi_match[] = {
>>   	{"BMI0160", 0},
>> +	{"10EC5280", 0},
>>   	{ },
>>   };
>>   MODULE_DEVICE_TABLE(acpi, bmi160_acpi_match);