Re: [PATCH] iio: imu: inv_mpu6050: fix magnetometer scale of Y and Z axes
From: Jonathan Cameron
Date: Sat Oct 17 2026 - 19:05:52 EST
On Mon, 12 Oct 2026 07:50:52 +0000
Jean-Baptiste Maneyrol <Jean-Baptiste.Maneyrol@xxxxxxx> wrote:
> >inv_mpu_magn_get_scale() returns st->magn_raw_to_gauss[chan->address],
> >where magn_raw_to_gauss[] holds one coefficient per axis, computed in
> >inv_magn_init() from the per-axis factory sensitivity adjustment values
> >(ASAX, ASAY, ASAZ) of the AK8975/AK8963 magnetometer.
> >
> >However INV_MPU9X50_MAGN_CHAN() never sets .address, so it is 0 for all
> >three magnetometer channels. As a result the Y and Z channels report the
> >scale computed from the X axis adjustment value, and userspace scales
> >the Y and Z readings with the wrong per-axis calibration.
> >
> >Set .address to the axis index (0 for X, 1 for Y, 2 for Z) in the
> >channel macro, matching the order used to fill magn_raw_to_gauss[].
> >
> >Fixes: b1392de06184 ("iio: imu: inv_mpu6050: add MPU925x magnetometer support")
> >Assisted-by: LLM
> >Signed-off-by: Arnav Kapoor <kapoorarnav43@xxxxxxxxx>
Applied to the fixes-togreg branch of iio.git and marked for stable.
Note I'll be rebasing that tree on rc1 before sending a pull request.
Thanks,
Jonathan
> >---
> >Compile-tested only (W=1); I don't have MPU9150/MPU925x hardware. I
> >checked the built object: before this patch the X, Y and Z magnetometer
> >channels all have .address == 0, after it they have 0, 1 and 2.
> >
> > drivers/iio/imu/inv_mpu6050/inv_mpu_core.c | 2 ++
> > 1 file changed, 2 insertions(+)
> >
> >diff --git a/drivers/iio/imu/inv_mpu6050/inv_mpu_core.c b/drivers/iio/imu/inv_mpu6050/inv_mpu_core.c
> >index 38c3f3b398d4..1614f49be66c 100644
> >--- a/drivers/iio/imu/inv_mpu6050/inv_mpu_core.c
> >+++ b/drivers/iio/imu/inv_mpu6050/inv_mpu_core.c
> >@@ -1576,6 +1576,8 @@ static const unsigned long inv_mpu_scan_masks[] = {
> > .type = IIO_MAGN, \
> > .modified = 1, \
> > .channel2 = _chan2, \
> >+ /* index of the axis in magn_raw_to_gauss[] */ \
> >+ .address = (_chan2) - IIO_MOD_X, \
> > .info_mask_separate = BIT(IIO_CHAN_INFO_SCALE) | \
> > BIT(IIO_CHAN_INFO_RAW), \
> > .scan_index = _index, \
> >--
> >2.53.0
> >
>
> Hello Arnav,
>
> good catch indeed, thanks for the patch.
>
> Acked-by: Jean-Baptiste Maneyrol <jean-baptiste.maneyrol@xxxxxxx>
>
> Thanks,
> JB