Newsgroups: comp.archives
Path: utzoo!utgpu!news-server.csri.toronto.edu!math.lsa.umich.edu!math.lsa.umich.edu!emv
From: taylor@THINK.COM (David Taylor)
Subject: [gnu.gcc.bug] Re: Decstaion/Ns32k Cross Compiling
Message-ID: <1990Jun6.054336.8980@math.lsa.umich.edu>
Followup-To: gnu.gcc.bug
Sender: emv@math.lsa.umich.edu (Edward Vielmetti)
Reply-To: taylor@THINK.COM (David Taylor)
Organization: University of Michigan, Department of Mathematics
Date: Wed, 6 Jun 90 05:43:36 GMT
Approved: emv@math.lsa.umich.edu (Edward Vielmetti)
X-Original-Newsgroups: gnu.gcc.bug
Lines: 32

Archive-name: gcc-ns32k/06-Jun-90
Original-posting-by: taylor@THINK.COM (David Taylor)
Original-subject: Re: Decstaion/Ns32k Cross Compiling
Archive-site: augean.ua.oz.au [129.127.4.2]
Reposted-by: emv@math.lsa.umich.edu (Edward Vielmetti)

<Date: 5 Jun 90 19:03:46 GMT
<From: jk3n+@andrew.cmu.edu  (John Stephen Kalucki)
<
<I'm trying to set up GCC to be a cross compiler. The host machine is a
<DECStation 3100 running Ultrix V2.0, Rev 7 and the target is a National
<ns32k. I got the compiler generate NS32k code, but it can't complete
<most compiles because gcc-cc1 calls abort due to some sort of
<internal inconsistancy. Unfortunatly I only have gcc 1.36, and I don't know
<if this bug has been fixed in 1.37. The bug is tickled quite often, it isn't
<subtle at all. I suspect it is an installation problem, but I'm not sure.

There are some problems with gcc generating code for the ns32k series.

I think your best bet is to first grab gcc 1.37.1 from prep and then
grab the diffs ns32k related diffs that Ian Dall has placed in the ftp
area on augean.ua.oz.au.

<Does gcc 1.37 fix this bug? Or did I just install something incorrectly?
<I tracked down the code segment that causes the problem. It is in tm.h,
<which is linked to ./config/tm-ns32k.

I'm not sure.  But, it fails the stage3 test.  Grab 1.37.1 and put the
changes mentioned above in and it'll pass.  You'll also find some ns32k
related changes for gdb and gas in the same directory.

David