Re: port-macppc/59673: PMG5 11,2: multiple channel-0 node confuses ki2c(4) driver

[Jason Thorpe <thorpej%me.com@localhost>]

> On Sep 28, 2025, at 5:55 AM, rokuyama.rk%gmail.com@localhost wrote:
> 
> Power Mac G5 11,2 has multiple channel-0 nodes for ki2c1, which confuses the driver.
> 
> Full dmesg (patched kernel) and `ofwctl -pv` log are provided here:
> 
> https://gist.github.com/rokuyama/1d90c9af42d97a84870b93bf0daad682
> https://gist.github.com/rokuyama/079eebc318ad9417665f2c023527dac4
> 
> Here, i2c@18000 has two child node i2c-bus@0 (channel 0): The first one
> has two audio codec's, whereas the second one has only one.

Wow, that really does look like an error in the OFW device tree, although which branch of the tree is the wrong one remains to be seen.  It would be really fantastic if, at some point, you could visually confirm that there’s only one TI / Burr-Brown PCM3052 chip on the board (for that matter, visual confirmation that a cs8416 *is* present might also be instructive!).

I’m 99.999% sure that prior to my changes in this area, that second instance would have been silently ignored for other reasons.

> If the second one is skipped with this patch:

Thanks, I’m going to check in a slightly tweaked version of this (adds a debug log message).

-- thorpej