On Thu, Aug 31, 2023 at 09:39:52AM +0000, Roman Azarenko wrote:
On Mon, 28 Aug 2023 at 02:00, Roman Azarenko
roman.azarenko@iopsys.eu wrote:
On Fri, 2023-08-25 at 12:06 -0600, Simon Glass wrote:
@@ -564,9 +564,13 @@ static int fit_extract_data(struct image_tool_params *params, const char *fname) /* Pack the FDT and place the data after it */ fdt_pack(fdt);
- new_size = fdt_totalsize(fdt); - new_size = ALIGN(new_size, align_size); + unpadded_size = fdt_totalsize(fdt); + new_size = ALIGN(unpadded_size, align_size); fdt_set_totalsize(fdt, new_size);
I didn't know that was allowed...I thought it needed fdt_open_into() ?
The introduction of fdt_set_totalsize() comes from commit ebfe611be91e ("mkimage: fit_image: Add option to make fit header align"). The commit message doesn't describe the choice of this function vs fdt_open_into().
Personally I'm unable to definitively comment on it. I can only blindly guess, that because we're only changing the total length of the fdt struct, and keeping all other fields the same, we don't need to allocate a new fdt struct with a different size.
I am not sure if it would cause problems. I do understand that you didn't write the code. You could copy the people who did (and those that reviewed it) for their input.
@Kever, @Punit, @Tom, could you please comment on this remark by Simon? Thank you!
I assume that we're given an already well-ordered device tree at this point and so that's why we don't use fdt_open_into to move things and just handle it ourselves at that point in the tool.