Xref: news-dnh.mv.net comp.os.msdos.djgpp:4074
Path: news-dnh.mv.net!mv!news.sprintlink.net!newsfeed.internetmci.com!in1.uu.net!bird3.i-link.net!usenet
From: royce3 AT i-link DOT net (Royce E. Mitchell III)
Newsgroups: comp.os.msdos.djgpp
Subject: Re: Windows95: ld causes segmentation faults
Date: Sun, 31 Dec 1995 15:32:54 GMT
Organization: I-Link Inc
Lines: 53
Message-ID: <4c63fh$s3g@bird3.i-link.net>
References: <DKD31B DOT 3A5 AT jade DOT mv DOT net>
Reply-To: royce3 AT i-link DOT net
NNTP-Posting-Host: houston-1-2.i-link.net
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp

elzinga AT alumni DOT caltech DOT edu (Dean Elzinga) wrote:

>Sorry for this repeat question, but I was trying to subscribe to
>this list when I first asked about general protection faults--
>unsuccessfully, it turns out, until now.

>I've figured out that gcc works okay in my DOS box, but ld
>causes a segmentation fault.  Usually the error message
>actually calls it a general protection fault.

Usually, if GCC has a GPF, it's because it didn't like something
code, that's the experience I've had.

>I have combed over mail archives at DJ's web page (and one
>other place), and read the FAQ.  Does anyone have any quick
>intuition about what could be the problem or the fix?

>I have tried using the 16-bit real-mode version of GCC, and
>I've tried such things as removing NOEMS from my call to EMM386.
>The DOS box seems to have plenty of memory available to it;
>go32 reports at least 16 MEG of DPMI and 60 meg of swap space.
>Has anyone gotten 1.12M4 working under Windows95.

I've never even touched the 16-bit real-mode version of GCC.
I've been compiling mostly in a DOS box in Win95.  I have 8MB,
and far less that 60MB HD space.

Try this:

put #error's in your code and remove them until you get the GPF
turns up again.  For example...

#error line 4

....some code

#error line 8

If you get the GPF with line 4 removed, but you don't get the GPF
with line 4 there, then the problem is between them.

>Should I get the V2 beta?  Is that more likely to work?

I think the problem is more likely to do with GCC than DJGPP,
but I imagine the V2 is going to work much better than 1.12m4
anyway.  I'd say get it.

>--dean

>elzinga AT alumni DOT caltech DOT edu

hope it helps