lkml.org 
[lkml]   [2020]   [Jan]   [16]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH 1/3] iio: amplifiers: hmc425a: Add support for HMC425A step attenuator with gpio interface
On Tue, 14 Jan 2020 07:27:08 +0000
"Ardelean, Alexandru" <alexandru.Ardelean@analog.com> wrote:

> On Mon, 2020-01-13 at 21:57 +0100, Lars-Peter Clausen wrote:
> > [External]
> >
> > On 1/13/20 3:15 PM, Beniamin Bia wrote:
> > [...]
> > > +static int hmc425a_write(struct iio_dev *indio_dev, u32 value)
> > > +{
> > > + struct hmc425a_state *st = iio_priv(indio_dev);
> > > + int i, *values;
> > > +
> > > + values = kmalloc_array(st->chip_info->num_gpios, sizeof(int),
> > > + GFP_KERNEL);
> > > + if (!values)
> > > + return -ENOMEM;
> > > +
> > > + for (i = 0; i < st->chip_info->num_gpios; i++)
> > > + values[i] = (value >> i) & 1;
> > > +
> > > + gpiod_set_array_value_cansleep(st->gpios->ndescs, st->gpios->desc,
> > > + values);
> >
> > This API got changed a while ago in upstream, see
> > https://github.com/analogdevicesinc/linux/commit/b9762bebc6332b40c33e03dea03e30fa12d9e3ed
> >
> > > + kfree(values);
> > > + return 0;
> > > +}
> > [...]
> > > +static int hmc425a_probe(struct platform_device *pdev)
> > > +{
> > [...]
> > > +
> > > + platform_set_drvdata(pdev, indio_dev);
> >
> > drvdata is never accessed, no need to set it.
> >
> > > + mutex_init(&st->lock);
> > > +
> > > + indio_dev->dev.parent = &pdev->dev;
> > > + indio_dev->name = np->name;
> >
> > I know ADI likes to do this in its non upstream drivers, but the above
> > is not IIO ABI compliant. The name is supposed to identify the type of
> > the device, which means for this driver should be static "hmc425a".
> > Maybe consider adding a field to the hmc425a_chip_info for this.
>
> We've actually [recently] had a discussion about this internally regarding
> the 'indio_dev->name'.
>
> Maybe it's a good time to ask here (now).
> A lot of our userspace stuff have been searching IIO devices via the 'name'
> field in sysfs, which is the name assigned here.
> That creates a problem when you have multiple devices with the same driver.
> Which is why, one
>
> So, then some questions would be:
> Is a searching for IIO devices [in userspace] based on IIO device-name not
> recommended? If not, what would be? Or what would be a better idea?
>
> The ABI reads [hopefully I pulled up the right field]:
> What: /sys/bus/iio/devices/iio:deviceX/name
> KernelVersion: 2.6.35
> Contact: linux-iio@vger.kernel.org
> Description:
> Description of the physical chip / device for device X.
> Typically a part number.
>
> The text in description is a bit open to interpretation, so I can't make an
> assessment of what is correct.
> In case there was a discussion about this, sorry for repeating some things
> now.

So I can speak to the 'intent' of that documentation. It's meant to be the part
number.

Now, we have recently added
indio_dev->label which is retrieved from generic device tree property 'label'
to solve the problem of multiple devices of the same type (note that
driver but different device shouldn't matter as name reflects the part number
not the driver). It lets you provide any name you like the DT blob.

I appreciate this is a 'new' feature and so a bit of a problem for old userspace.

For a long time I pushed back against this because it's easy to tell which device
is which, just look at the parent. I got convinced in the end that sometimes
that answer isn't very user friendly :)

Jonathan


>
>
> >
> > > + indio_dev->info = &hmc425a_info;
> > > + indio_dev->modes = INDIO_DIRECT_MODE;
> > > +
> > > + return devm_iio_device_register(&pdev->dev, indio_dev);
> > > +}


\
 
 \ /
  Last update: 2020-01-16 19:29    [W:0.042 / U:5.032 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site