Hi, On Sat, 2014-10-04 at 12:51 +0100, Jonathan Cameron wrote: > On 02/10/14 13:08, Ivan T. Ivanov wrote: > > +iio maintainers > > > > On Thu, 2014-10-02 at 12:29 +0300, Ivan T. Ivanov wrote: > > > Hi Stan, > > > > > > > On 09/24/2014 03:56 PM, Stanimir Varbanov wrote: > > > > > > > <snip> > > > > > > > > +static int vadc_get_dt_data(struct vadc_priv *vadc, struct > > > > device_node *node) > > > > +{ > > > > + struct iio_chan_spec *iio_chan = vadc->iio_chans; > > > > + const struct vadc_channels *vadc_chan; > > > > + struct vadc_channel_prop prop; > > > > + struct device_node *child; > > > > + int ret, index = 0; > > > > + > > > > + for_each_available_child_of_node(node, child) { > > > > + ret = vadc_get_dt_channel_data(vadc->dev, &prop, > > > > child); > > > > + if (ret) > > > > + return ret; > > > > + > > > > + vadc->chan_props[index] = prop; > > > > + > > > > + vadc_chan = &vadc_chans[prop.channel]; > > > > + > > > > + iio_chan->channel = prop.channel; > > > > + iio_chan->datasheet_name = vadc_chan->datasheet_name; > > > > + iio_chan->info_mask_separate = vadc_chan->info_mask; > > > > + iio_chan->type = vadc_chan->type; > > > > + iio_chan->indexed = 1; > > > > + iio_chan->scan_type.sign = 's'; > > > > + iio_chan->scan_type.realbits = 15; > > > > + iio_chan->scan_type.storagebits = 16; > > > > + iio_chan->address = index++; > > > > + > > > > + iio_chan++; > > > > + } > > > > + > > > > + return 0; > > > > +} > > > > + > > > > > > > <snip> > > > > > > > > +static int vadc_probe(struct platform_device *pdev) > > > > +{ > > > > > > > <snip> > > > > > > > > + > > > > + indio_dev->dev.parent = dev; > > > > + indio_dev->dev.of_node = node; > > > > + indio_dev->name = pdev->name; > > > > + indio_dev->modes = INDIO_DIRECT_MODE; > > > > + indio_dev->info = &vadc_info; > > > > + indio_dev->channels = vadc->iio_chans; > > > > + indio_dev->num_channels = vadc->nchannels; > > > > > > > This will not work as we thought. Registration will be fine and > > > entries > > > in sysfs are created as expected, but this is breaking > > > numbering scheme > > > for client drivers. For us the problem is that iio framework > > > expect channels > > > numbering to be linear and channel number should not be greater > > > than > > > channel count. For example even if die temperature channel is > > > always 8, if > > > I describe only 5 channels in DTS files (4 for reference points > > > measurement) > > > and 1 for die temperature, clients could not get OF reference to > > > temperature > > > channel using io-channels = <&pm8941_vadc 8>; > > > > > > > Jonathan, please, could you advise us. > I'd like to let the solution you proposed sit on the list a little > longer. > It seems fairly sensible, but lets see what others suggest. It is quiet :-) Just for the record the same aproach is used in PWM, Reset, Gpio and PHY frameworks. Patch in question: https://lkml.org/lkml/2014/10/2/199 Regards, Ivan > > Jonathan -- To unsubscribe from this list: send the line "unsubscribe linux-iio" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html