Hi Jonas,
On 6/17/24 8:53 PM, Jonas Karlman wrote:
[...]
I'm wondering if we shouldn't do the same for U-Boot proper DTB as well?
I think that would be good, and should also remove the need for the
recursive pinctrl lookup for bootph props during pre-reloc.
i.e.
cmd_fdt_rm_props = cat $< | $(objtree)/tools/fdtgrep -r -O dtb - -o $@
to
cmd_fdt_rm_props = $(objtree)/tools/fdtgrep $(fdtgrep_props) -u $< |
$(objtree)/tools/fdtgrep -r -O dtb - -o $@
I assume we need to add a bunch more options though.
But that would make sense to me, the fact that there's a pre-reloc stage
in U-Boot proper is already a bit odd (to me), but if we have something
that behaves differently than the pre-reloc stage in earlier stages...
I'm not sure this is a good idea?
Fully agree that part of the pre-reloc stage make little sense at least
on Rockchip platform where full memory is available before pre-reloc.
Will try to come up with something.
[...]
We can/should probably also extend fdtgrep to remove disabled nodes for
the VPL/SPL/TPL.
I didn't find any disabled node in SPL DTB for RK3588 Tiger and RK3399
Puma, so maybe not worth the added complexity to fdtgrep right now. But
that's a good idea :)
Cheers,
Quentin