overlays: sc16is75x: Add generic SPI overlay

[nbuchwitz]

Currently four nearly identical overlays (sc16is750-spi0 / -spi1 and sc16is752-spi0 / -spi1) are provided. Besides the redundant configuration all of them lack support for other SPI interfaces than spi0 and spi1.

Thus refactor the common definitions into a generic sc16is75x-spi overlay which provides support for all known spi / cs combinations. Also choose the chip type via dtparam rather than different overlay.

In order to keep existing setups keep the old overlays, but add a deprecation message via overlay map.

[nbuchwitz]

As talked about yesterday, I will think about some generic solution in order to avoid duplication of the boilerplate spi cs tuples. But this is stuff for another PR (once I found something which will work for most of the overlays)

[pelwell]

I'm looking at this now. The deprecation breaking the overlay check is unfortunate, but I think it would be fixed by actually deleting the original overlays. A more serious problem is the reference to labels that don't exist in the older Pis' dtbs, eg. spi3. This can be avoided by replacing the target property with target-path=spi3 etc., making use of the fact that paths that don't start with a / are assumed to begin with the name of an alias, which is then resolved.

If you don't mind sticking to the old naming scheme - int_pin and xtal - we should be able to replace the deprecated error messages with e.g.:

	sc16is750-spi0 {
		rename = "sc16is75x,sc16is750,spi0-0";
	};

I'm currently getting an error about overlay 'sc16is750-spi0' not being supported on the bcm2835 platform, but I should be able to fix (or at least understand) that.

[pelwell]

Doh, it's obvious - the correct syntax is:

	sc16is750-spi0 {
		renamed = "sc16is75x,sc16is750,spi0-0";
	};

Note the d on renamed.

[nbuchwitz]

Thanks, the changes sound resonable. I will update the branch

[pelwell]

I'd hang on until I've got it all working, especially the automated checks. When I've done that I'll push a patch with my changes.

[pelwell]

Assuming the other tests pass, we can fix up the over-long commit message when we "Squash and merge".

[nbuchwitz]

Looks good so far. Will do some tests with real hardware and then this should be ready

[nbuchwitz]

Tests look good so far. From my side this is ready to be squashed and merged