Hi Simon, On Sat, Sep 19, 2026 at 4:43 AM Simon Glass <[email protected]> wrote: > > Hi Aristo, > > On 2026-09-12T12:44:46, Aristo Chen <[email protected]> wrote: > > > The 'Shared image data' section of the Flat Image Tree specification [1] > > defines an image-data property [...] > > Thanks for the thorough cover letter and the hardware numbers. Also > for taking this on while I am on a patch holiday :-) > > Please add user-facing documentation under doc/usage/fit/ describing > image-data. A short section (either in multi.rst, since this lands > most naturally in multi-configuration FITs, or a new file) covering > what the property means, the resolved form mkimage produces, the > encrypted-sharing rules and a pointer to the external spec. Right now > someone trying to shrink a multi-board FIT has no way to discover this > from the U-Boot docs. > > Also we have example files and this feels like a case where a new > example would help - see doc/usage/fit/
I will fix this in v2 > > Just to check - patch 4 is closing a validation gap opened by patch 2 > rather than fixing a pre-existing bug, since data-less images did not > exist before image-data was introduced. Would folding it into patch 2 > make things easier to follow? Either way is fine. That is the right reading. I kept it separate mainly for review size: the resolve patch is already the largest change in the series, at 360 added lines, while this is a 45-line edit to a different function, the region collection in fit_import_data(), with its own reasoning. A small follow-on patch seemed easier to follow than growing the big one further. In v2, I will update the commit message to explicitly state that it completes the validation for the new property rather than fixing a pre-existing problem. If you would still rather see it folded into the resolve patch for v2, that is easy to do. Thanks for the review and for the patch-holiday exception! I will also fix other concerns you pointed out for patch 2. > > Regards, > Simon Best regards, Aristo
