svn commit: r237269 - in head: etc lib/libutil

[Alexey Dokuchaev]

On Tue, Jun 19, 2012 at 05:21:13PM +0100, Chris Rees wrote:
> On Jun 19, 2012 5:15 PM, "Alexey Dokuchaev" <danfe at freebsd.org> wrote:
> > Pardon my possible unawareness, but was this change discussed anywhere?
> 
> http://lists.freebsd.org/pipermail/freebsd-security/2012-June/006271.html

Thanks for the link, I didn't check -security@ for some reason.

> > I understand the rationale to move away from MD5, but reasons for SHA512
> > seem moot.  I've personally had been using Blowfish for password hashes
> > since OpenBSD switched to it, for example, as fast and apparently reliable
> > hash.  Is there anything wrong with it?  Why SHA512 is clear winner here?
> > FWIW, ports use SHA256 for now.  Could it be that switch to SHA512 will
> > impose performance problems?
> 
> Why would you want password matching to be fast?  That makes brute-forcing
> easier.

Maybe I don't.  I just want to know if I should switch from Blowfish to
SHA512.  It seems that the former is quite popular judging from discussion
link given above.  It also seems that des@' rationale for the switch boils
down to "I vastly prefer sha512 to blf, as that is what the rest of the
world uses."  If there's nothing wrong with Blowfish, I guess I'll stick to
it as I prefer compatibility among *BSD to some weird Unix clones.  :-)

./danfe