Windows compile fix #1
Conversation
Line 54 in context_primitive.c seems to be a VLA, although I'm not sure that is intended. The change makes this a fixed array and MSVC will now compile it. Also note that there is another issue related to latest Rust, so downgraded to v1.88 for a clean compile.
The KID was not being included in set request. The problem seems to be twofold. a) the k-flag was being overwritten. b) There was confusion over whether the AAD was written with sender of receiver KID. KID now included in OSCORE option (RFC 8613 Section 5.4) ✅ AAD now calculated correctly with sender's KID
|
This version has been used in all the FENCE PoC work to v0.1.10 with no problems. |
|
Thanks for reporting this. This'll need a bit of clean-up, likely to happen at the actual repository https://gitlab.com/oscore/liboscore (GitHub is just a mirror, and unlike issues, PRs can't be disabled). Actually, this mirror is even a bit lagging, you will find more recent fixes (including for current Rust compilers) there. The array being accidentally VLA is indeed a bug; still looking into whether a defined or an assigned-value enum is the right fix here. This PR also contains a KID issue, which I'll copy into a separate PR to review there. Do you have an account on GitLab which I can CC when doing that? |
|
Thanks for that.
I’ve just opened a GitLab account, my user name is @WayneHowell.
Thanks for the pointer to the root repo.
And apologies – I should have made the Kid issue a separate PR.
Regards
Wayne
|
2 participants
The original code would not compile on Windows and there was a problem with KID being overwritten. Both fixed.