[PATCH 2/9] arm: dts: ls1028a: move devices into /soc

[Vladimir Oltean]

On Tue, Aug 31, 2021 at 11:32:22PM +0200, Michael Walle wrote:
> > Does the Kontron SL28 use SPL?
> 
> Yep :) No BL1/BL2 for TF-A, BL31 is optionally loaded by the
> u-boot SPL. Mainly because NXP doesn't seem to be interested
> in bringing layerscape (besides one architecture) support in
> upstream TF-A. And I don't want to fork yet another software
> part from some vendor supplied tree. Not really needed, so I
> skipped that. Sorry for that rant ;)

No problem, negative feedback is the best feedback!

> > The LS1028A-QDS does not use SPL, it uses TF-A, hence the name:
> > configs/ls1028aqds_tfa_defconfig
> 
> The fsl-ls1028a-qds.dts has "u-boot,dm-pre-reloc" set for the
> i2c0 node, which doesn't work without the same flag for /soc.
> So that means I should remove the u-boot,dm-pre-reloc from the
> i2c0 node instead?

To be clear, the discussion is practically about the pcf2127-rtc.
I'll be honest with you, I've no idea why anyone would need an RTC
driver in SPL, nor how can I even test it if I wanted to.
I've tested just now, and I can run "date" (which calls dm_rtc_get)
in Hush after deleting the "u-boot,dm-pre-reloc" property, and it
behaved absolutely the same (no surprises).

> > Otherwise, the patch itself is pretty much impossible to follow. I guess
> > the only option would be to move nodes one by one to /soc such that it
> > becomes obvious. I don't have a strong preference, just saying that I
> > can't judge it for correctness.
> 
> Sounds like many small patches, but I don't have a strong optinion
> either.

Whichever way you feel like. Just stating that proper review will be
very difficult in this format.