As part of getting a handle on RSA state, I would like for RSA keys
created from parsing to come pre-"frozen". This reduces some contention
on first use. But that potentially breaks an existing use case: today,
you're allowed to parse a private key and then override one field
without problems, because none of the cached state has materialized yet.
If the caller modified the RSA key by reaching into the struct, it's
hopeless. If they used the setters, we actually can handle it correctly,
so go ahead and make this work.
DH and DSA aren't of particular interest to us, but fix them while I'm
at it.
This also avoids having to later document that something doesn't work,
just that it's a terrible API.
Bug: 316
Change-Id: Idf02c777b932a62df9396e21de459381455950e0
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/59385
Auto-Submit: David Benjamin <davidben@google.com>
Commit-Queue: David Benjamin <davidben@google.com>
Reviewed-by: Adam Langley <agl@google.com>
David Benjamin
9939e14cff