> Also, consider if other triggers could be used as if not you need to > both document why and add the validation callbacks to stop other triggers > being assigned (once you've added one that can be!) > > Feel free to ask if you have more questions, but your first reference > should be other drivers (and I hope we don't have any that do it this way). > I used this driver as a reference https://github.com/torvalds/linux/blob/master/drivers/iio/adc/vf610_adc.c#L556 > > ouch, not a write 1 to clear register? I can't find docs, but this is really > nasty bit of interface design if you have to toggle the bit. > Actually ST has not provided a register map or any application note for the sensor. So there's no way to cross reference. Hence I kept the original code. But I will try to write 1 to clear register with my sensor. > > + ret = devm_iio_triggered_buffer_setup(&client->dev, > > + indio_dev, > > + &iio_pollfunc_store_time, > > This is odd. You don't seem to have a function to be called to actually store > the data. Note you also need to consider if other triggers might be used. > > I'm not sure what reason we have to do that here though as this is a very > conventional one interrupt per 'scan' of data device. > > So you should be registering a trigger, and a buffer then letting the > trigger drive the buffer. Why do I need to register a trigger? Would it be fine to let the irq fill the buffer with data as it continuously reads it in the poll function? So according to my understanding, I need to move all the data reading and pushing to the poll function and not do it in the irq handler. Then register that poll function here during iio_triggered_buffer_setup. Is there something that I am missing? Regards, Abhash
