On 27 May 2012 20:40, David Brownlee<abs%netbsd.org@localhost> wrote:
On 27 May 2012 20:24, Christos Zoulas<christos%astron.com@localhost> wrote:
In
article<CAGN_6pY=QoOZ-J87U_kr0=40=7VrMJHrYbJ8UpCj8enC-C-65g%mail.gmail.com@localhost>,
David Brownlee<abs%absd.org@localhost> wrote:
The POLY instruction was dropped in Rigel and later CPU variants,
which is a problem for NetBSD/vax's use of them in
lib/libm/arch/vax/{n_argred.S,n_atan2.S}>
c) emulate it.
Aha!
On 03 Mar 2005 09:32, ragge%ludd.luth.se@localhost> wrote:
The emulation code for polyd is in place; I wrote it 5 years ago and
Brandon Creighton did a bunch of cleanup and added emodd, see
vax/vax/unimpl_emul.S. So I think it's some kind of bug in the
emulation code that should be searched for.
So now its down to debugging that! :)
Apologies for following up and spamming the list - but does anyone
have a netbsd-5 or earlier Rigel, Mariah, or (ideally) NVAX baxed
VAX on which they could run the original sintest.c (which just
prints out the value of sin(1))
Given NetBSD should be emulating polyd, and all that looks to be
enabled in NetBSD-6, we're down to potentially something that did
work and now does not...