Re: [PATCH next] iio: imu: inv_icm45600: Add a missing return statement in probe()

[Jonathan Cameron <jic23@kernel.org>]

On Fri, 31 Oct 2025 09:56:45 +0000
Remi Buisson <Remi.Buisson@tdk.com> wrote:

> >
> >
> >From: Jonathan Cameron <jic23@kernel.org> 
> >Sent: Monday, October 27, 2025 3:24 PM
> >To: Dan Carpenter <dan.carpenter@linaro.org>
> >Cc: Remi Buisson <Remi.Buisson@tdk.com>; David Lechner <dlechner@baylibre.com>; Nuno Sá <nuno.sa@analog.com>; Andy Shevchenko <andy@kernel.org>; linux-iio@vger.kernel.org; linux-kernel@vger.kernel.org; kernel-janitors@vger.kernel.org
> >Subject: Re: [PATCH next] iio: imu: inv_icm45600: Add a missing return statement in probe()
> >
> >On Wed, 22 Oct 2025 14: 02: 20 +0300 Dan Carpenter <dan. carpenter@ linaro. org> wrote: > The intention here was clearly to return -ENODEV but the return statement > was missing. It would result in an off by one read in i3c_chip_info[]
> >On Wed, 22 Oct 2025 14:02:20 +0300
> >Dan Carpenter <dan.carpenter@linaro.org> wrote:
> >  
> >> The intention here was clearly to return -ENODEV but the return statement
> >> was missing.  It would result in an off by one read in i3c_chip_info[] on
> >> the next line.  Add the return statement.
> >> 
> >> Fixes: 1bef24e9007e ("iio: imu: inv_icm45600: add I3C driver for inv_icm45600 driver")
> >> Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org>
> >> ---
> >>  drivers/iio/imu/inv_icm45600/inv_icm45600_i3c.c | 3 ++-
> >>  1 file changed, 2 insertions(+), 1 deletion(-)
> >> 
> >> diff --git a/drivers/iio/imu/inv_icm45600/inv_icm45600_i3c.c b/drivers/iio/imu/inv_icm45600/inv_icm45600_i3c.c
> >> index b5df06b97d44..9247eae9b3e2 100644
> >> --- a/drivers/iio/imu/inv_icm45600/inv_icm45600_i3c.c
> >> +++ b/drivers/iio/imu/inv_icm45600/inv_icm45600_i3c.c
> >> @@ -57,7 +57,8 @@ static int inv_icm45600_i3c_probe(struct i3c_device *i3cdev)
> >>  	}
> >>  
> >>  	if (chip == nb_chip)
> >> -		dev_err_probe(&i3cdev->dev, -ENODEV, "Failed to match part id %d\n", whoami);
> >> +		return dev_err_probe(&i3cdev->dev, -ENODEV,
> >> +				     "Failed to match part id %d\n", whoami);
> >>  
> >>  	return inv_icm45600_core_probe(regmap, i3c_chip_info[chip], false, NULL);
> >>  }  
> >
> >I'm going to apply this but the resulting code is still wrong (even if not
> >a true bug after this fix).
> >
> >A hard ID match like this breaks use of dt fallback compatibles.
> >What this should do is 'give it a go' on matching, but if there no match it should
> >carry on as if the match was to whatever the compatible that was supplied was.
> >When that happens a dev_info() is appropriate but not error out as this does.
> >
> >Remi, if possible could you look at adding such a patch on top of this?
> >
> >Thanks,
> >
> >Jonathan
> >  
> Thanks Jonathan, the fix is correct.
> The problem is that I3C don't specify any device in the device tree,
> and these sensors cannot be identified by their I3C IDs neither.
> So, the driver cannot fallbacks to any compatible device, other than picking one, more or less, at random.
> Do you see any way to work around this limitation?

Ah.  I'd missed that. Fair enough, sounds like nothing we can do
if the driver doesn't know the ID.

Jonathan