Re: [PATCH v3 3/3] iio: imu: st_lsm6dsx: add support for rotation sensor

[Francesco Lavra <flavra@baylibre.com>]

[-- Attachment #1: Type: text/plain, Size: 2020 bytes --]

On Tue, 2026-01-20 at 12:25 +0200, Andy Shevchenko wrote:
> On Tue, Jan 20, 2026 at 12:03 PM Francesco Lavra <flavra@baylibre.com>
> wrote:
> > On Tue, 2026-01-20 at 11:36 +0200, Andy Shevchenko wrote:
> > > On Tue, Jan 20, 2026 at 10:28:15AM +0100, Francesco Lavra wrote:
> > > > On Mon, 2026-01-19 at 12:33 +0200, Andy Shevchenko wrote:
> > > > > On Mon, Jan 19, 2026 at 11:04:49AM +0100, Francesco Lavra wrote:
> 
> ...
> 
> > > > > > +       snprintf(sensor->name, sizeof(sensor->name), "%s_sf",
> > > > > > name);
> > > > > 
> > > > > Does GCC complain on this (`make W=1` build)?
> > > > > Since this can cut the string and we don't check the return
> > > > > value,
> > > > > the Q
> > > > > is:
> > > > > is this okay to have a reduced string?
> > > > 
> > > > gcc does not complain with W=1. sensor->name is appropriately sized
> > > > to
> > > > accommodate the longest possible name; if it wasn't, the string
> > > > would
> > > > be
> > > > cut in the accel and gyro IIO devices too (which use a longer
> > > > suffix
> > > > than
> > > > "_sf").
> > > 
> > > Right, the question is if compiler can prove that or not.
> > > 
> > > We have several patches in input subsystem to hide the warning by
> > > switching
> > > to scnprintf(), which I consider not the best approach, but still it
> > > depends
> > > if we care about cut or not. If we do, we should check for overflow.
> > 
> > A truncated string would result in a corrupted value in
> > /sys/bus/iio/devices/iio:deviceX/name, so I would say we do care if
> > it's
> > truncated. But I think the most appropriate check we could add in the
> > code
> > is a BUG_ON(); would that be acceptable?
> 
> No. BUG_ON() == panic == reboot (with "reboot_on_panic" or so). Mere
> drivers like this one should (almost) never use BUG() nor WARN(). Just
> add an appropriate check.

I'm struggling to find a suitable error code to be returned in case of
failure. What do you suggest? -ENOSPC? -EOVERFLOW?

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 659 bytes --]