Re: [PATCH v2 2/2] iio: adc: Add the NXP SAR ADC support for the s32g2/3 platforms

[Jonathan Cameron <jonathan.cameron@huawei.com>]

On Fri, 12 Sep 2025 09:19:43 +0100
Nuno Sá <noname.nuno@gmail.com> wrote:

> On Fri, 2025-09-12 at 08:38 +0300, Andy Shevchenko wrote:
> > On Fri, Sep 12, 2025 at 2:03 AM Daniel Lezcano
> > <daniel.lezcano@linaro.org> wrote:  
> > > On 11/09/2025 22:10, David Lechner wrote:  
> > > > On 9/10/25 10:57 AM, Daniel Lezcano wrote:  
> > 
> > [ ... ]
> >   
> > > > > +            /* iio_push_to_buffers_with_timestamp should not be called
> > > > > +             * with dma_samples as parameter. The samples will be
> > > > > smashed
> > > > > +             * if timestamp is enabled.
> > > > > +             */  
> > 
> > /*
> >  * Btw, comment style for multi-line
> >  * comments is wrong for this subsystem.
> >  * Use this as an example, Also, refer to
> >  * the function as func(), i.e. mind the parentheses.
> >  */
> >   
> > > > > +            timestamp = iio_get_time_ns(indio_dev);
> > > > > +            ret = iio_push_to_buffers_with_timestamp(indio_dev,
> > > > > +                                                     info->buffer,
> > > > > +                                                     timestamp);  
> > > > 
> > > > Is it OK to call this with spinlock held? It looks like it can call
> > > > devm_krealloc() which may sleep.  
> > > 
> > > It should be ok, devm_krealloc is in the code path of
> > > iio_push_to_buffers_with_ts_unaligned(), not in
> > > iio_push_to_buffers_with_timestamp()  
> > 
> > This is a good observation, can we document this in the respective
> > kernel-doc:s please? Also add might_sleep().might_sleep_if() in the
> > appropriate functions.  
> 
> That's a good idea!

Agreed. I'd forgotten that hidden allocation was there in the unaligned()
variant.  A might_sleep() seems wise.

> 
> - Nuno Sá