Still not sure where the problem lies. I thought I found an issue, but it didn't fix anything and caused a different issue.

At first, after plugging in a bunch of printfs, I thought it was a simple issue in the dts (it appears not to be the issue): ***** Getting a little back into it, I find an issue in: "./lib/libfdt/fdt_ro.c" "fdt_path_offset()" where "-FDT_ERR_BADPATH" is returned for "path" = "sdhci@12530000" when: "./lib/fdtdec.c" "fdtdec_add_aliases_for_id()" is examining "mmc2"

"fdt_path_offset()" "_fdt_nodename_eq()" finds "aliases"

But the issue comes from "./lib/libfdt/fdt_ro.c" "_fdt_string_eq()" where "p" = "mmc2" <> "s" = "sdhci@12530000"

What I'm noticing different in the "exynos4412-odroid.dts" (odroid u2) and "exynos54xx.dtsi" is the "/" for the alias paths "mmc2" and "mmc4". *****

The problem with that "finding" was: "fdtdec_add_aliases_for_id: warning: alias 'mmc2' points to a node 'sdhci@12530000' that is missing or is not compatible with 'samsung,exynos-dwmmc'" Not sure why it would even mention dwmmc for mmc2 when it is specified "exynos-mmc" in "exynos4.dtsi"

A common issue with or without the slash is: "_gpio_request_by_name_nodev: Node 'sdhci@12530000', property 'pwr-gpios', failed to request GPIO index 0: -2"

I think the kicker is "No match for node 'sdhci@12530000'". I even manually set/forced the "status=okay" in the dts, but it still says it's disabled.

I shall continue to pursue this, working my way back from "lists_bind_fdt()" in "lists.c"