size_t-ification of those algorithms in future version of OpenSSL...

And couple of bug-fixes in fips/rand code [return without lock release and
incorrect return value in fips_rand_bytes].

key.

isn't a FIPS-approved algorithm.

Note: this means the user needs to keep track of this, and we need to
add support for that...

can't use the variable in libcrypto, since it's supposedly unknown.

Note: currently only supported in MONOLITH mode.

mk1mf.pl, when building in FIPS mode.

Note: UNTESTED!

PR: 851

don't crash and burn.

EVP_*_cfb128.

generated from the Unixly configuration file.

FIPS branch into this.  It's needed at least for certain OpenVMS
versions, and should really be used in a more general way.

let's keep internal variables static.

use it.

Notified by Frédéric L. W. Meunier <0@pervalidus.tk> in PR 713

well.

Notified by Frank Kardel <kardel@acm.org> in PR 879.

PR:870

Due to the changes in CA.pl in 0.9.8 (use of -self_sign) a slightly different
technique is used to ensure that 'ca' uses the next serial number. It
now initializes the serial number using 'openssl x509 -next_serial'.

some similar code elsewhere.

Thanks to Francesco Petruzzi for bringing this to my attention.