On Fri, Jan 22, 2021 at 08:40:49AM +0530, Viresh Kumar wrote:

On 22-01-21, 10:39, David Gibson wrote:

quoted

No, it definitely will not work in general.  It might kinda work in a
few trivial cases, but it absolutely will not do the neccessary
handling in some cases.

quoted

I
did inspect the output dtb (made by merging two overlays) using
fdtdump and it looked okay.

Ok.. but if you're using these bizarre messed up "dtbs" that this test
code seems to be, I don't really trust that tells you much.

I only looked if the changes from the second overlay were present in
the merge and they were. And so I assumed that it must have worked.

What about checking the base dtb for /plugin/; in fdtoverlay and fail
the whole thing in case it is present ? I think it is possible for
people to get confused otherwise, like I did.

/plugin/ doesn't exist in the dtb, only in the dts.  From the dtb
encoding point of view, there's no difference between a dtb and a
dtbo, a dtbo is just a dtb that follows some conventions for its
content.

If we were doing this from scratch, it would be better for dtbos to
have a different magic number from regular dtbs.  I think I actually
suggested that sometime in the past, but by the time that came up,
dtbos were already in pretty widespread use with the existing format.

-- 
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson