This requires casts, which is not entirely desirable; however, this way
at least no *more* pointer sign issues can creep in.
Output object files are unchanged on amd64.
consttime_memequal is the same as the old consttime_bcmp.
explicit_memset is to memset as explicit_bzero was to bcmp.
Passes amd64 release and i386/ALL, but I'm sure I missed some spots,
so please let me know.
abd userland, as proposed on tech-security, with explicit_bzero using
a volatile function pointer as suggested by Alan Barrett.
Both do what the name says. For userland, both are prefixed by "__"
to keep them out of the user namespace.
Change some memset/memcmp uses to the new functions where it makes
sense -- these are just some examples, more to come.
because this is not a public function and crypt-sha1.c is the only
place where it is used (It must have been that way formerly, because
there was a stale prototype in crypt.h all the time.)
Being here, do some RCSID and prototype cleanup, and add as much "const"
as needed to compile these files with WARNS=3.
The algorithm used is essentially PBKDF1 from RFC 2898 but using
hmac_sha1 rather than SHA1 directly (suggested by smb@research.att.com).
* The format of the encrypted password is:
* $<tag>$<iterations>$<salt>$<digest>
*
* where:
* <tag> is "sha1"
* <iterations> is an unsigned int identifying how many rounds
* have been applied to <digest>. The number
* should vary slightly for each password to make
* it harder to generate a dictionary of
* pre-computed hashes. See crypt_sha1_iterations.
* <salt> up to 64 bytes of random data, 8 bytes is
* currently considered more than enough.
* <digest> the hashed password.
hmac.c implementes HMAC as defined in RFC 2104 and includes a unit
test for both hmac_sha1 and hmac_sha1 using a selection of the Known
Answer Tests from RFC 2202.
It is worth noting that to be FIPS compliant the hmac key (password)
should be 10-20 chars.
riastradh