Re: [PATCH 3/6] spl: ram: allow loading FIT images from a supplied buffer

[Carlo Caione <ccaione@baylibre.com>]

On Tue, Nov 03, 2026 at 11:09:31 +0100, Mattijs Korpershoek wrote:
> Hi Carlo,
> 
> Thank you for the patch.
> 
> > RAM FIT loading currently assumes a configured image address. Downloaded
> > images need the same loader with a caller-provided buffer and length so
> > reads cannot run beyond the received data.
> > 
> > Expose a reusable RAM FIT loader with header validation, bounded reads
> > and overlap-safe copying. Keep preloaded images working when their buffer
> > size is unknown. Add tests for embedded and external FIT data.
> > 
> > Signed-off-by: Carlo Caione <ccaione@baylibre.com>
> >
> > diff --git a/MAINTAINERS b/MAINTAINERS
> > index d1f51d9197d9..f464e0bd2bce 100644
> > --- a/MAINTAINERS
> > +++ b/MAINTAINERS
> > @@ -1923,5 +1923,6 @@ Q:	http://patchwork.ozlabs.org/project/uboot/list/
> >  S:	Maintained
> >  T:	git https://git.u-boot-project.org/u-boot/u-boot.git
> >  F:	configs/tools-only_defconfig
> > +F:	test/image/spl_ram.c
> >  F:	*
> >  F:	*/
> > diff --git a/common/spl/spl_ram.c b/common/spl/spl_ram.c
> > index 0c501cf02f28..ee588ef4718b 100644
> > --- a/common/spl/spl_ram.c
> > +++ b/common/spl/spl_ram.c
> > @@ -16,32 +16,39 @@
> >  #include <spl.h>
> >  #include <linux/libfdt.h>
> >  
> > -static ulong spl_ram_load_read(struct spl_load_info *load, ulong sector,
> > +struct spl_ram_image {
> > +	const void *data;
> > +	size_t size;
> > +};
> > +
> > +static ulong spl_ram_load_read(struct spl_load_info *load, ulong offset,
> 
> Why changing sector -> offset? It makes the diff more difficult to read and
> as far as I understand, the callers are still passing the same argument to
> this function.

It was just for the sake of clarity. I'll dump this.
 
> >  			       ulong count, void *buf)
> >  {
> > -	ulong addr = 0;
> > -
> > -	debug("%s: sector %lx, count %lx, buf %lx\n",
> > -	      __func__, sector, count, (ulong)buf);
> 
> Why is this dropped? It's not mentioned in the commit message.

Ooops, I'll restore it.

> Rest looks good, thanks
> 
> -- 
> Mattijs Korpershoek <mkorpershoek@kernel.org>

Thanks,

--
Carlo Caione