Hi Rasmus, On Mon, 25 Sept 2023 at 02:47, Rasmus Villemoes <rasmus.villem...@prevas.dk> wrote: > > On 22/09/2023 17.26, Simon Glass wrote: > > >>> Shouldn't this be fdt_open_into()? > >> > >> I honestly just copy-pasted fit_extract_data() and shaved it down to the > >> part that does the "align the FDT part of the file". > >> > >> I don't really understand your question. Are you saying this doesn't > >> work (or maybe doesn't work in some cases), or are you saying that > >> there's a simpler way to do this? For the latter, sure, one doesn't > >> really need to parse the whole FDT; we could just > >> > >> open() > >> pread() length from FDT header, convert to cpu-endianness > >> length = ALIGN(length) > >> pwrite() the new length in fdt-endianness > >> ftruncate() > >> close() > >> > >> but I thought it was better to stay closer to how fit_extract_data() was > >> done. > > > > I mean that fdt_open_into() does more than just set the size (from > > what I can tell). But looking further I see other code which calls > > fdt_set_totalsize() so perhaps it is fine. > > Yes, I think it's as it should be - as a I said, this is really just a > trimmed-down copy of the function which moves the data externally, and > also needs to make the size of the base fdt structure aligned. > > Since patches 2,3,4 touch binman code, could you take all four?
Yes, will do. I didn't pick them up in the most recent PR as I try to have things sit for a week before applying. Regards, Simon
