🚀 go-pugleaf

Forwarded message:

>From ctshp!cam Sat Mar  7 01:43:36 1992
Message-Id: <m0lMqRu-0007XEC@ctshp.cts.celtech.com>
From: cam@ctshp.cts.celtech.com (Cameron S. Elliott)
Subject: GCC 2.0 differences between stage1 compiler output and stage2 compiler output (fwd)
To: cam@ctsx
Date: Fri, 6 Mar 92 17:43:46 PST
Mailer: Elm [revision: 66.25]


Hi Tim and GNU, 

I have gotten the hp800 compiler generating programs on the HP9000/837
HP/UX 8.02

I have compiled stage1, stage2, and stage3 compilers.
Unfortunatly stage2/cccp.o and ./cccp.o (stage3) have differences
so I did a make with -S (compile to .s) to generate the stage2 cccp.s from
the stage1 compiler, and did the same with the stage2 compiler to generate the
stage3 cccp.s.  
with the following commands:
374     make CC=stage1/gcc CFLAGS="-O -Bstage1/ -S" cccp.o ; mv cccp.s cccp1.s
375     make CC=stage2/gcc CFLAGS="-O -Bstage2/ -S" cccp.o ; mv cccp.s cccp2.s
377     diff cccp1.s cccp2.s >/tmp/camdiff
You will find /tmp/camdiff at the end of the message.

The stage1 generated stuff looks like it does the functional same thing,
but just not as optimized, what could this all mean?
Thanks.

PS GNU How can I help make my changes to GCC2.0 for the HP9000/837 (HPUX 8.02)
become part of the next distribution, if you are interested.
According to the HP Response Center HP/UX 8.02 is the most recent release
for the HP9000/8xx series so it would seem like there might be interest in this.
The changes are pretty minor.  Two assembly generation differences, 
no .stabs, etc.  Any information on how to do this is great.




/tmp/camdiff:
18812,18815c18812,18814
< 	ldo R'hashtab-$global$(1),21
< 	add 19,21,20
< 	stw 20,8(0,28)
< 	add 19,21,19
---
> 	ldo R'hashtab-$global$(1),20
> 	add 19,20,19
> 	stw 19,8(0,28)

--
Cameron Elliott (206)727-6372			cam@ctsx.cts.celtech.com


--
Cameron Elliott (206)727-6372			cam@ctsx.cts.celtech.com