I've been automating the process in doc/README.chromium-chainload and while experimenting with whether a "kernel" image with u-boot-spl and u-boot would work, noticed I couldn't align/offset/pad the two parts.
E.g. in something like the following, binman doesn't add the necessary padding to place the "u-boot" to the correct offset within the "kernel-1" data:
fit { description = "example";
images { kernel-1 { description = "U-Boot with SPL"; type = "kernel"; arch = "arm64"; os = "linux"; compression = "none";
u-boot-spl { }; u-boot { offset = <CONFIG_SPL_PAD_TO>; }; }; }; };
Not sure if that'll ever be really necessary besides my experiment, but it doesn't seem like skipping the padding was a deliberate choice, so here are some fixes I wrote for that.
Alper Nebi Yasak (3): binman: Ignore hash*, signature* nodes in sections binman: Respect pad-before property of section subentries binman: Build FIT image subentries with the section etype
tools/binman/etype/fit.py | 22 +++---- tools/binman/etype/section.py | 4 +- tools/binman/ftest.py | 32 +++++++++++ tools/binman/test/165_pad_in_sections.dts | 26 +++++++++ .../test/166_fit_image_subentry_alignment.dts | 57 +++++++++++++++++++ 5 files changed, 129 insertions(+), 12 deletions(-) create mode 100644 tools/binman/test/165_pad_in_sections.dts create mode 100644 tools/binman/test/166_fit_image_subentry_alignment.dts