for the host byte order is when we copy the final digest (and that
is just an optimization).

man help jmc@

Use sizeof() in the Init functions where it makes sense.
Use weak aliases instead of wrapper functions.
Probably should have gone in before the major bump but as these are only
used internally by the sha2 functions themselves there should be no problem.

It was defined before and there is no need to for this knob...

OK otto@

I missed earlier.

The Pad function does padding like in Final but does not finish processing.
The FileChunk function creates a digest from a portion of a file.
Also made the length parameters consistent (and size_t).

Also make the open functions take a mode_t, not int.  OK deraadt@

ok deraadt@ millert@

the tree was stable. No point in code supporting a company which screws
its employees being in the OpenBSD tree.

handle a few subtle details caught by the regression tests: correct
termination, non copying if buffer length == 0, errno setting.
let all former users of __strerror go through strerror_r.
Work by Todd Miller and I. Okay millert@.

okay millert@

okay millert@.
- atoi was bogus.
- atoi was not reentrant.
- there's no reason to cast stuff to unsigned int.

to /usr/libexec/mail.local;
from millert@

ok xsa@ henning@

ok cedric@ mcbride@

o Make MD5Update() more readable
o Simplify padding in MD5Final()
o Do copy + byte flip in MD5Transform() instead of in the caller
The result is a faster md5 and the byte order foo is kept to a minimum.

functions can take a NULL buf pointer.  They will malloc memory as
needed in this case.

aligned so copy it in a way that a) is endian indepenent and b) does
not rely on alignment.  Problem found and solution tested by hshoexer@