sasdf

Your patch also fix the issue when using x64 relative addressing (e.g. `lea rdi, [rip + label]`) with symbol resolver. Thanks a lot!

Thank you for liking this challenge. Playing with smaller p (e.g. 7, 17, 31) could help you find it. Bruteforce all possibilities is very fast even with unoptimized python code....

Nice analysis. I think you also need to prove that the swapping gadget will left other numbers as it was to say that we can build a solution using these...

@cfrantz @pamaury Could you please provide a brief review of the proposed approach? I'll continue migrating the remaining tests if this method is acceptable.

Rebased to resolve merge conflict with 28bfbc1

Hi, Could I get some feedback on this PR? I apologize for its size, but this is the minimum amount that can still be tested end-to-end with `./bazelisk.sh coverage`. Thanks!

> Does the newer OTBN binary even need the patch? This is a patch to OTBN binary in earlgrey ROM, which won't have newer OTBN binary. Though, I can't speak...

> If we change this patching code to accomodate the changed OTBN boot services program, it won't work anymore for A2. I'm a bit confused. This PR is trying to...

The design document about the constant byte array approach and MSb is available [here](https://docs.google.com/document/d/1fYsukm2aqi3Dzai3Y4JJnEAhbbPc5OFJcfiHf-r803w/edit?tab=t.0). Regarding merging the cdi_0 and cdi_1 builders: Since CDI 0 will be produced by immutable RomExt,...