SHA384_DIGEST_LENGTH in sha2_mac_atomic(), but not ..._verify_atomic()

[Dan McDonald]

On Wed, Feb 13, 2008 at 03:33:42PM -0800, Anthony Scarpino wrote:
>> without any SHA384 exception.  Is this a bug?
>>
>> Dan
>
> It does look a bit interesting, but I remember there 384 uses a lot of 64 
> byte lengths in it's operations..

Which is why the fix is localized to that bit, not the big wad I posted
yesterday.

> Did this fix your problem with sparc vs x86 & x86 vs x86? Or was that 

Funny thing, I couldn't get sparc vs. sparc to *work* without this fix -
makes me wonder if I had used 256 or 512 for that batch of testing
mistakenly.

Anyway, with my last patch in place, IT DOES WORK on x86 vs. sparc now, as
well as sparc vs. sparc.  I'll verify x86 vs. x86 later this evening, but
that shouldn't be a problem.

I'll file the bug shortly, what cat/subcat should I use?  I'll put it back
with the IPsec/SHA2 wad.

Dan