Subject: Heads up: Maybe working 68LC040/68LC060 fpu emulation in -current
To: None <port-m68k@netbsd.org>
From: Ignatios Souvatzis <is@beverly.kleinbus.org>
List: current-users
Date: 01/05/2001 01:00:06
--3MwIy2ne0vdjdPXF
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: quoted-printable
Hi,
please test netbsd-current on 68LC040 and 68040V machines (well, and on
FPU-less 68020/68030 machines, but I can do that myself) and report.
"Should work now as well as on 020/030".
Long story:
I was working on adding the twoliner needed to add 68060 with disabled FPU/
68LC060 support.
I don't really believe it worked at all on LC040 (although my 68060 seems=
=20
to behave strangely with disabled FPU sometimes (sometimes defined as "I
couldn't track down the exact condition yet")). I found (and removed) at
least one bug when using the frame-provided next instruction address in the=
=20
040/060 case. Using the frame-provided effective address still was seriously
buggy, so I removed all 040/060 special handling (which, alas, should have
been faster!) with the exception of restoring the "this instruction" pc to
the stack frame. Now it seems to work as good as on the 020/030 for me.
Remaining todo:
a) 68LC040/68040V/68LC060-specific:=20
a1) use the frame-provided effective address when possible. This should
work, dammit!
a2) maybe use the frame provided next instruction address, instead of
calculating it ourselves (but we still need to calculate it for=20
branches, so it might not be worth the effort).
b) think really hard about signal return.
c) complete the emulation.
Regards,
Ignatios
--3MwIy2ne0vdjdPXF
Content-Type: application/pgp-signature
-----BEGIN PGP SIGNATURE-----
Version: 2.6.3ia
iQCVAwUBOlUOhO4WPQce5FrdAQFDnAQAkJWvWB1yEfOTaoWd72VOoSJVENUq42me
O7nfcI8e+O2mMsbk1jZ5nv7v7OLGBAghuLMjwIaP0Nxq6G4hseRur2snqVmGbahm
AfTHEI0K2U/yNPk7ngqvMOEXf2RdRu82R/JJR+HnwF5BunD3BXSqsscXIJhmeSpw
CiL+JAYlYgA=
=Oakv
-----END PGP SIGNATURE-----
--3MwIy2ne0vdjdPXF--