On 31.10.26 15:05, Simon Glass wrote:
> Hi Frieder,
>
> On 2026-10-29T16:23:07+11:00, Frieder Schrempf <[email protected]> wrote:
>
>> Patches 8-10 add sandbox tests for both options. As partition names
>> need GPT, patch 8 adds GPT support to DiskHelper in fs_helper.py, which
>> the RAUC test image now uses.
>
> One thing I'd like to check here. sandbox_defconfig is the only CI
> build that enables BOOTMETH_RAUC, and after patches 9 and 10 it sets
> both new options, so the default behaviour (no name check, MBR image,
> boot script) is no longer built or tested anywhere. Patch 7 drops the
> script-loading code from the build when BOOTCMD is set, so the fix in
> patch 4 is not even compiled in CI, and the 'boot.scr' in the test
> image is now unused. The cover letter says the options are empty by
> default so that existing users see no change, so I think the default
> path is the one that most needs coverage.
>
> Also, the new tests only run if the defconfig string exactly matches
> what the test expects, which is fragile. Please can you find a way to
> cover both configurations? For example the test could override the
> options at runtime (e.g. through a field in the bootmeth's private
> data, defaulting to the Kconfig value), or you could keep a second
> image that stays on MBR with a boot script. What do you think?
Thanks for mentioning this. I think I will try to rewrite this to keep
the options in the code so they can be overridden at runtime for testing.
