[591] in Zephyr_Bugs
Re: Changes fro zephyr on the alpha
daemon@ATHENA.MIT.EDU (Greg Hudson)
Mon Jun 12 01:51:59 1995
Date: Mon, 12 Jun 1995 01:56:47 -0400
From: Greg Hudson <ghudson@MIT.EDU>
To: epeisach@MIT.EDU
Cc: bug-zephyr@MIT.EDU, rel-eng@MIT.EDU
In-Reply-To: "[585] in Zephyr_Bugs"
> Many of the changes were
> a) 32 vs 64 bit ints
> b) for some reason I needed Code_t to be an u_int instead of int.
> (b) required many protottypes to be added and this caused code
> cleanups...
> I will reverify if the Code_t really need to be u_int anymore, but
> the other changes are still valid.
I have integrated the changes you submitted into my current
development tree. I made a lot of the changes differently, since I'm
using autoconf now and can figure out what a 32-bit integral type is
without resorting to "#ifdef __alpha".
I left Code_t as a signed integer for all platforms for now. Tell me
if you find a reason why it needs to be unsigned.
This change puzzled me:
> + extern int des_quad_cksum();
All versions of the DES library I could find (in the Athena source
tree, in the Zephyr source tree, in the CNS Kerberos release) define
des_quad_cksum() as returning unsigned long. It probably *should*
have returned u_int32, but since Zephyr links against external
Kerberos and DES libraries, I have to be compatible with what's
distributed out there. I don't think it's a problem that
des_quad_cksum() might return a 64-bit value.
