Re: [PATCH 3/3] iio: wrapper: unit-converter: new driver

From: Peter Rosin <peda@axentia.se>
To: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Cc: Jonathan Cameron <jic23@kernel.org>,
	linux-kernel@vger.kernel.org, Hartmut Knaack <knaack.h@gmx.de>,
	Lars-Peter Clausen <lars@metafoo.de>,
	Peter Meerwald-Stadler <pmeerw@pmeerw.net>,
	Rob Herring <robh+dt@kernel.org>,
	Mark Rutland <mark.rutland@arm.com>,
	"David S. Miller" <davem@davemloft.net>,
	Mauro Carvalho Chehab <mchehab@kernel.org>,
	Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
	Linus Walleij <linus.walleij@linaro.org>,
	Randy Dunlap <rdunlap@infradead.org>,
	linux-iio@vger.kernel.org, devicetree@vger.kernel.org
Subject: Re: [PATCH 3/3] iio: wrapper: unit-converter: new driver
Date: Tue, 27 Mar 2018 15:32:13 +0200	[thread overview]
Message-ID: <0c9e6fbe-cc05-1a2e-0437-dcde134e6eba@axentia.se> (raw)
In-Reply-To: <20180327142256.00003056@huawei.com>

On 2018-03-27 15:22, Jonathan Cameron wrote:
> On Tue, 27 Mar 2018 09:42:40 +0200
> Peter Rosin <peda@axentia.se> wrote:
>> On 2018-03-24 15:03, Jonathan Cameron wrote:
>>> On Mon, 19 Mar 2018 18:02:46 +0100
>>> Peter Rosin <peda@axentia.se> wrote:
>>>> +	if (iio_channel_has_info(pchan, IIO_CHAN_INFO_RAW))
>>>> +		chan->info_mask_separate |= BIT(IIO_CHAN_INFO_RAW);  
>>> if the parent doesn't support RAW, is there a lot of point in carrying on?  
>>
>> Nope, better to error out I suppose. But I'm not familiar with channels
>> without RAW, what alternatives are there anyway?
> 
> Potentially _PROCESSED though that will need somewhat different handling.
> A nasty trick for that might be to map it to RAW and then have the SCALE
> reflect the divider circuit scale only.

Hmm, I think a lot of things might assume RAW to be a pure integer, and
maybe they are even correct to do so? So yes, that seems nasty indeed...

> It's perfectly possible to have channels with neither _RAW or _PROCESSED
> but I suspect we don't care about them here.
> 
> There might be an application that needs to do buffered data flows in the
> long run, but we can figure out how to do that when one exists.
> 
> It won't be a huge amount more than you have here, though we might need
> a trigger pass through as well to allow you to set the trigger for
> the front end and having it automatically applied to the backend.

Yes, this is the same for the iio-mux. I don't need it, I in fact need
very little bandwidth for these things. Someone with an itch will have
to fill in the buffer/trigger handling...

Cheers,
Peter