Date: Wed, 24 Mar 1999 19:50:36 +0100
From: Ronald de Man <deman AT win DOT tue DOT nl>
To: pgcc AT delorie DOT com
Subject: Re: EGCS Optimizations break ZLIB 1.1.3
Message-ID: <19990324195036.A24996@win.tue.nl>
References: <19990324154651 DOT 16303 DOT qmail AT www0j DOT netaddress DOT usa DOT net>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
X-Mailer: Mutt 0.95.4i
In-Reply-To: <19990324154651.16303.qmail@www0j.netaddress.usa.net>; from ADAM SCHROTENBOER on Wed, Mar 24, 1999 at 10:46:51AM -0500
X-Operating-System: Linux localhost 2.2.3 
Reply-To: pgcc AT delorie DOT com
X-Mailing-List: pgcc AT delorie DOT com
X-Unsubscribes-To: listserv AT delorie DOT com
Precedence: bulk

On Wed, Mar 24, 1999 at 10:46:51AM -0500, ADAM SCHROTENBOER wrote:
> When I build zlib 1.1.3 w/ optimzations set to 3 or higher, it breaks zlib.
> 
> The example program crashes w/ a SIGSEGV in __djmove_data+37
> 
> I don't have the full stack trace here, but I do have it at home.
> 
> With 06, the stack trace has 2 entries, w/ 03, it has something like 6, but
> both GPF in the same function.

I remember zlib failing with pgcc (might have been 1.0.2) and -O6.
Adding -fno-risc worked for me. That was probably with older versions
of both zlib and pgcc than you are using, but you might give '-O6 -fno-risc'
a try.

Ronald