[RFC PATCH v3 8/8] tools: gen_pre_load_header.sh: initial import

Simon Glass sjg at chromium.org
Sat Dec 11 12:37:28 CET 2021

Hi Rasmus,

On Fri, 10 Dec 2021 at 00:41, Rasmus Villemoes <rasmus.villemoes at prevas.dk>
> On 10/12/2021 01.14, Simon Glass wrote:
> > Hi,
> >
> > On Wed, 8 Dec 2021 at 11:10, Philippe REYNES
> > <philippe.reynes at softathome.com> wrote:
> >>
> >>
> >> Everybody agrees with this proposal ?
> >
> > So long as this is not a shell script and is done with a binman entry,
> No, no and please no. Binman really isn't as wonderful and magic as you
> think. Having a simple, well-defined format that can be generated _even
> with a simple shell script_ is a good thing, because it means I can go
> generate such a wrapper around any random binary artifact I may have
> somewhere, completely outside the U-Boot repo or the context of the
> U-Boot recipe. That will, in fact, be the most common case. I should not
> have to add random noise to blabla-binman.dtsi to get the U-Boot build
> to wrap my FIT image in the proper format, adding weird inter-recipe
> dependencies between the kernel and U-Boot builds.

That's fine if you want a separate tool - we have mkimage and various
others and we don't want to put all these things in Binman. But the shell
script makes no sense. You should write it in C and add a test. When people
do use Binman to bring things together, it can then still call the tool.


More information about the U-Boot mailing list