From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gcc-return-41706-listarch-gcc=gcc.gnu.org@gcc.gnu.org>
Received: (qmail 22593 invoked by alias); 27 Nov 2001 08:02:37 -0000
Mailing-List: contact gcc-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Archive: <http://gcc.gnu.org/ml/gcc/>
List-Post: <mailto:gcc@gcc.gnu.org>
List-Help: <http://gcc.gnu.org/ml/>
Sender: gcc-owner@gcc.gnu.org
Received: (qmail 22423 invoked from network); 27 Nov 2001 08:02:31 -0000
Received: from unknown (HELO dot.cygnus.com) (205.180.230.224)
  by hostedprojects.ges.redhat.com with SMTP; 27 Nov 2001 08:02:31 -0000
Received: (from rth@localhost)
	by dot.cygnus.com (8.11.2/8.11.2) id fAR81os29747;
	Tue, 27 Nov 2001 00:01:50 -0800
X-Authentication-Warning: dot.cygnus.com: rth set sender to rth@redhat.com using -f
Date: Sun, 18 Nov 2001 09:18:00 -0000
From: Richard Henderson <rth@redhat.com>
To: Andreas Schwab <schwab@suse.de>
Cc: Corey Minyard <minyard@acm.org>, gcc@gcc.gnu.org
Subject: Re: Loop optimization bug with Ada front end on PPC (and probably Alpha)
Message-ID: <20011127000150.A29737@redhat.com>
Mail-Followup-To: Richard Henderson <rth@redhat.com>,
	Andreas Schwab <schwab@suse.de>, Corey Minyard <minyard@acm.org>,
	gcc@gcc.gnu.org
References: <3BFCA770.5070304@acm.org> <20011124124100.A2485@redhat.com> <jelmgu4wm4.fsf@sykes.suse.de>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
User-Agent: Mutt/1.2.5i
In-Reply-To: <jelmgu4wm4.fsf@sykes.suse.de>; from schwab@suse.de on Sun, Nov 25, 2001 at 05:57:07PM +0100
X-SW-Source: 2001-11/txt/msg00825.txt.bz2

I've committed the following simplification as execute/20011126-2.c.


r~



/* Problem originally visible on ia64.

   There is a partial redundancy of "in + 1" that makes GCSE want to
   transform the final while loop to 

     p = in + 1;
     tmp = p;
     ...
     goto start;
   top:
     tmp = tmp + 1;
   start:
     in = tmp;
     if (in < p) goto top;

   We miscalculate the number of loop iterations as (p - tmp) = 0
   instead of (p - in) = 1, which results in overflow in the doloop
   optimization.  */

static const char *
test (const char *in, char *out)
{
  while (1)
    {
      if (*in == 'a')
	{
	  const char *p = in + 1;
	  while (*p == 'x')
	    ++p;
	  if (*p == 'b')
	    return p;
	  while (in < p)
	    *out++ = *in++;
	}
    }
}

int main ()
{
  char out[4];
  test ("aab", out);
  return 0;
}

From mboxrd@z Thu Jan  1 00:00:00 1970
From: Richard Henderson <rth@redhat.com>
To: Andreas Schwab <schwab@suse.de>
Cc: Corey Minyard <minyard@acm.org>, gcc@gcc.gnu.org
Subject: Re: Loop optimization bug with Ada front end on PPC (and probably Alpha)
Date: Tue, 27 Nov 2001 00:02:00 -0000
Message-ID: <20011127000150.A29737@redhat.com>
References: <3BFCA770.5070304@acm.org> <20011124124100.A2485@redhat.com> <jelmgu4wm4.fsf@sykes.suse.de>
X-SW-Source: 2001-11/msg01329.html
Message-ID: <20011127000200.MK3w2Fb_qKJng9ZLAT8RHX27zCmXHNPILkR-lSOFVcM@z>

I've committed the following simplification as execute/20011126-2.c.


r~



/* Problem originally visible on ia64.

   There is a partial redundancy of "in + 1" that makes GCSE want to
   transform the final while loop to 

     p = in + 1;
     tmp = p;
     ...
     goto start;
   top:
     tmp = tmp + 1;
   start:
     in = tmp;
     if (in < p) goto top;

   We miscalculate the number of loop iterations as (p - tmp) = 0
   instead of (p - in) = 1, which results in overflow in the doloop
   optimization.  */

static const char *
test (const char *in, char *out)
{
  while (1)
    {
      if (*in == 'a')
	{
	  const char *p = in + 1;
	  while (*p == 'x')
	    ++p;
	  if (*p == 'b')
	    return p;
	  while (in < p)
	    *out++ = *in++;
	}
    }
}

int main ()
{
  char out[4];
  test ("aab", out);
  return 0;
}