Re: [PATCH] Provide more details of exactly how configuration signatures are calculated

Hi Martin, On Wed, 20 Jul 2022 at 02:46, Martin Bonner wrote: > > OK. I've just realized there is an important warning missing from > this (if you verify a configuration signature outside u-boot, you > _must_ verify the hashes for the kernel, fdt, and ramdisk images match > too). What is the pr

Re: [PATCH] Provide more details of exactly how configuration signatures are calculated

OK. I've just realized there is an important warning missing from this (if you verify a configuration signature outside u-boot, you _must_ verify the hashes for the kernel, fdt, and ramdisk images match too). What is the protocol for handling that? Submit a new patch with title "[Patch v2] ..."?

Re: [PATCH] Provide more details of exactly how configuration signatures are calculated

There are a couple of uncertainties still remaining: - The "hashed-nodes" property is created by mkimage in image order. What happens if somebody manipulates a signed image to change the order? Does it make any difference? Do u-boot and fit_check_sign handle it in the same way? (My preference wo

[PATCH] Provide more details of exactly how configuration signatures are calculated

I have just spent a week reverse-engineering this, so I thought I'd contribute the work back to the community diff --git a/doc/uImage.FIT/signature.txt b/doc/uImage.FIT/signature.txt index 61a72db3c7..61cdd55e40 100644 --- a/doc/uImage.FIT/signature.txt +++ b/doc/uImage.FIT/signature.txt @@ -382,6