🚀 go-pugleaf

In article <323@fe2o3.UUCP> michael@fe2o3.UUCP (Michael Katzmann) writes:
>Does any one have the GNU C++ compiler (and GCC and GDB for that matter)
>working on an NS32000 system?  

Depends on what operation system you are useing. If you have SysV.2
from NS, then I have GCC 1.36 and GDB 3.5 working and quite stable on
my icm3216. There *are* a significant number of problems. If you want
I will send you patches.

>I am very interested in getting C++ and the gnu debugger going on my
>ICM3216 system but I dont want to have to reinvent the wheel.

G++ I haven't touched yet. I am under the impression that one really needs
the gnu loader to make g++ work properly, and that means installing all the
coff-encap stuff. That means gas as well. I am currently working on gas.

Speaking of gas. What is the "one true assembler syntax" for the 32k
series.  I have taken the instruction set reference manual as being
the most authorative. That means '@' prefixes absolute addressing mode
and there is no prefix for immediate mode. Most of the questions I have
are with how the default should be interpreted. Eg

foo=5
label:
	addd foo, r0		; add 5 to r0
	addd @foo, r0		; add the contents of address 5 to r0
	br label		; goto label. Should this be @label?
	addr foo, r1		; what does this do

The branch and the addr cases are the most tricky. The branch operand must
be always converted to a pc relative addressing mode, but what exactly is
converted? In the case of the addr instruction, an immediate operand is
presumably illegal. So should the assembler default to absolute mode or
insist on the instruction being written "addr @foo, r1".
-- 
Ian Dall     life (n). A sexually transmitted disease which afflicts
                       some people more severely than others.