Make crypto.c compile/link with OpenSSL 3

[ngie-eign]

• Initialize libssl using non-deprecated APIs

OpenSSL 3 deprecated SSL_library_init and SSL_load_error_strings in
favor of OPENSSL_init_ssl. Use OPENSSL_init_ssl when dealing with
OpenSSL 1.1 and newer to unbreak the build with OpenSSL 3.

• Move MD5 APIs to EVP_MD APIs

OpenSSL 3 deprecated all of the MD5_* APIs. Move to the equivalent
EVP_MD* APIs so the code doesn't need to be pinned down to 1.1
compatible APIs and uplifted at a later date.

Co-authored-by: Pierre Pronchery [email protected]
Co-authored-by: Ed Maste [email protected]
Signed-off-by: Enji Cooper [email protected]
Sponsored by: The FreeBSD Foundation

[ngie-eign]

@emaste , @khorben: FYI

[ngie-eign]

More work's required in order to get the program to compile cleanly with OpenSSL 3. I'll move this PR back to draft mode and fix those issues.

[emaste]

Is -DOPENSSL_API_COMPAT=0x10100000L sufficient to build with OpenSSL 3.0 at least initially?

[ngie-eign]

Is -DOPENSSL_API_COMPAT=0x10100000L sufficient to build with OpenSSL 3.0 at least initially?

Yes, but I want to clean this up so the folks in this project don't have to do it later.

[emaste]

Yes, but I want to clean this up so the folks in this project don't have to do it later.

Sounds good!

[emaste]

\o/

[ngie-eign]

I've confirmed that the newest revision is functionally correct and produces the same values as it did previously -- I also compared the logic in the python hmac module to confirm that it was sane:

$ make check
./hmac-md5-test1 --short
key: this is a passcode
text: abcdefghijklmnop
digest: 369b855bac0edb31107b180d2b76ed31
./hmac-md5-test1 --long
key: this is a SUPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPER LONG passcode
text: abcdefghijklmnop
digest: 54ecf5f3274c
./hmac-md5-test3 --short
key: this is a passcode
text: abcdefghijklmnop
digest: 369b855bac0edb31107b180d2b76ed31
./hmac-md5-test3 --long
key: this is a SUPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPER LONG passcode
text: abcdefghijklmnop
digest: 54ecf5f3274c
./hmac-md5-test.py --short
key: this is a passcode
text: abcdefghijklmnop
digest: 369b855bac0e
./hmac-md5-test.py --long
key: this is a SUPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPER LONG passcode
text: abcdefghijklmnop
digest: 54ecf5f3274c

[corecode]

@emaste can you check and merge this PR?

[emaste]

@corecode it might be a little while, but I will put it on the list.

I presume we'd like @ngie-eign to squash the fixups into one commit?

[ngie-eign]

@corecode it might be a little while, but I will put it on the list.

I presume we'd like @ngie-eign to squash the fixups into one commit?

Yes, I was planning on doing that. I left them as fixup commits to aid with reviewing.