Re: [PATCH 3/3] of: kexec: Always use FDT_PROP_INITRD_START and FDT_PROP_INITRD_END

[Geert Uytterhoeven <geert@linux-m68k.org>]

Hi Rob,

On Wed, Jun 16, 2021 at 7:14 PM Rob Herring <robh+dt@kernel.org> wrote:
> On Wed, Jun 16, 2021 at 3:27 AM Geert Uytterhoeven
> <geert+renesas@glider.be> wrote:
> > Commit b30be4dc733e5067 ("of: Add a common kexec FDT setup function")
> > introduced macros FDT_PROP_INITRD_* to refer to initrd properties, but
> > didn't use them everywhere.  Convert the remaining users from string
> > literals to macros.
>
> I'm not really a fan of the defines, so if anything I'd get rid of

Oh, as you authored that patch, I thought you liked them ;-)
And I was thinking of moving them to a header file, so they can be
used by other .c files, too...

Upon closer inspection, I see you just copied them from arm64, which
was not that visible due to commit ac10be5cdbfa8521 ("arm64: Use
common of_kexec_alloc_and_setup_fdt()") being a separate commit...

> them. But the bigger problem is what you brought to light with the
> variable size. As I mentioned, we should refactor this and the fdt.c

The number of cells to use for the initrd properties doesn't seem to
be well-defined.
drivers/of/fdt.c derives it from the length of the property, which
more or less always works ("be strict when sending, be liberal when
receiving").  Some code hardcodes it to 1 or 2.  I suspect (didn't
check) there's also code out there that uses the root number of cells?

> code to have a common function to read the initrd start and end.

What with code that needs to set the start and end?
It needs to use what the receiving end will expect...

Gr{oetje,eeting}s,

                        Geert


--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds