> http://cr.opensolaris.org/~darrenm/6626962/webrev/
Nit: A number of the Copyright notices are out of date.
The top level Makefiles seem to be missing a "package" target
that the other lib/crypt_modules Makefiles contain. Is this
intentional?
Relative to CRYPT_MAXCIPHERTEXTLEN size change, IIRC, you
exposed that for my benefit in making some code I was doing
in passwdutil.so easier, though 2000/510 does expose it as
Evolving and this would be an incompatible change. I guess
this is an evolution ;-)
crypt_sha.c:
163, 169, Why is the length of B MIXCHARS rather than
DIGEST_LEN?
200, perhaps ... memcpy(Sp, DS, DIGEST_LEN)) + MIXCHARS;
208-215, 225-232, based Ulrich's description, perhaps
reverse the logic, e.g., if ((1& 1) == 0)
212, 214, 227, 229 DIGEST_LEN rather than MIXCHARS
Step 21 c), f), I don't understand the nomenclature
add digest A/C and how it translated to
if round 0 add digest A and all other rounds
add plaintext (DP).
> I haven't yet implemented the crypt_gensalt_impl entry point but the
> bulk of the review is for crypt_genhash_impl. I'll update the webrev
> with that soon.
Why is salt using urandom? Why not random?
Gary..
