"cell-index" vs. "index" vs. no index in I2C device nodes

[Scott Wood]

Timur Tabi wrote:
> Scott Wood wrote:
> 
>> No, it's not.  It can determine that it's at address 0x4f on the i2c bus
>> at 0xe0003100.  This is exactly how the ethernet phy lookup is done.
> 
> But how does the fabric driver know whether e0003100 is I2C1 or I2C2?

It shouldn't have to care.

> And how does the codec driver, which sees only I2C information, know that it's
> at e0003100?

This is an internal communications failure within the i2c layer, not 
something that warrants expression in the device tree.  Some solutions, 
in increasing order of desirability, are:

1. Assign the I2C bus number based on the adapter's register offset.
2. Let the adapter provide a more helpful bus_id than a Linux-assigned 
index.
3. Create a way to look up an I2C device by its OF node.

-Scott