Compatibles for i2c muxes nxp,pca954x and ti,tca954x

Compatibles for i2c muxes nxp,pca954x and ti,tca954x

[Peter Rosin]

Hi!

Texas has apparently made copies for some of the NXP devices
handled by the drivers/i2c/muxes/i2c-mux-pca954x.c driver.

How is that best handled?

I see e.g. that arch/arm/boot/dts/vf610-zii-dev-rev-{b,c}.dts has
this snippet:

&i2c2 {
	tca9548@70 {
		compatible = "nxp,pca9548";

Which indicates that it really is a Texas chip sitting there, but that
someone did not bother to state that explicitly.

As I see it, there are two levels to this.

1. Update the above to be

		compatible = "ti,tca9548a", "nxp,pca9548";

   and rely on the secondary compatible to be matched with the driver.

2. Also update the driver to recognize the "ti,tca9548a" compatible.

There are also "ti,tca9543a", "ti,tca9544a", "ti,tca9545a" and
"ti,tca9546a" to consider (but no "ti,tca9542a" or "ti,tca9547a").

AFAIU, the Texas chips are completely compatible from the driver
point of view.

Step 1 above is probably a bugfix, should someone find some
unexpected incompatibility in the future. Without differentiating
the compatibles now, there's just no way to handle that in a future
driver.

But is step 2 needed/desired until such a time that differences between
the chips are found? How is this situation handled elsewhere?

Cheers,
peda

Re: Compatibles for i2c muxes nxp,pca954x and ti,tca954x

[Andrew Lunn]

On Thu, Aug 03, 2017 at 12:00:27PM +0200, Peter Rosin wrote:
> Hi!
> 
> Texas has apparently made copies for some of the NXP devices
> handled by the drivers/i2c/muxes/i2c-mux-pca954x.c driver.

This happens a lot in the i2c space. Look at all the at24 EEPROM
clones. Yet they all use the same compatible string.

	Andrew

Re: Compatibles for i2c muxes nxp,pca954x and ti,tca954x

[Peter Rosin]

On 2017-08-03 15:44, Andrew Lunn wrote:
> On Thu, Aug 03, 2017 at 12:00:27PM +0200, Peter Rosin wrote:
>> Hi!
>>
>> Texas has apparently made copies for some of the NXP devices
>> handled by the drivers/i2c/muxes/i2c-mux-pca954x.c driver.
> 
> This happens a lot in the i2c space. Look at all the at24 EEPROM
> clones. Yet they all use the same compatible string.

Errhm, no they don't. Hint, try

$ git grep 'compatible.*,24c'

There are all sorts of manufacturers in use.

atmel, microchip, onsemi, st, etc.

There are also "bad" manufacturers such as 'at,24c64' and "bad"
parts such as 'atmel,at24c16'. It's a mess.

Anyway, the only reason that ever works is because of loose
matching by the i2c core (it falls back to ignoring the
manufacturer or something like that, but I'm not sure about
the exact details). IIUC there is ongoing work to get rid of
that loose matching, and adding more of it is not the way
forward.

Cheers,
peda
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html