From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gcc-bugs-return-424819-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Received: (qmail 7274 invoked by alias); 22 Jun 2013 12:22:38 -0000
Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Id: <gcc-bugs.gcc.gnu.org>
List-Archive: <http://gcc.gnu.org/ml/gcc-bugs/>
List-Post: <mailto:gcc-bugs@gcc.gnu.org>
List-Help: <mailto:gcc-bugs-help@gcc.gnu.org>
Sender: gcc-bugs-owner@gcc.gnu.org
Received: (qmail 7209 invoked by uid 48); 22 Jun 2013 12:22:31 -0000
From: "olegendo at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug rtl-optimization/50749] Auto-inc-dec does not find subsequent contiguous mem accesses
Date: Sat, 22 Jun 2013 12:22:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: rtl-optimization
X-Bugzilla-Version: 4.7.0
X-Bugzilla-Keywords:
X-Bugzilla-Severity: enhancement
X-Bugzilla-Who: olegendo at gcc dot gnu.org
X-Bugzilla-Status: UNCONFIRMED
X-Bugzilla-Priority: P3
X-Bugzilla-Assigned-To: unassigned at gcc dot gnu.org
X-Bugzilla-Target-Milestone: ---
X-Bugzilla-Flags:
X-Bugzilla-Changed-Fields:
Message-ID: <bug-50749-4-wlXAYGpi0C@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-50749-4@http.gcc.gnu.org/bugzilla/>
References: <bug-50749-4@http.gcc.gnu.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 7bit
X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
X-SW-Source: 2013-06/txt/msg01198.txt.bz2

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50749

--- Comment #14 from Oleg Endo <olegendo at gcc dot gnu.org> ---
Just wanted to clarify the reason why the examples in the description have
'-fno-ivopts', as it caused some confusion on the mailing list:

int test_0 (char* p, int c)
{
  int r = 0;
  r += *p++;
  r += *p++;
  r += *p++;
  return r;
}

compiled with -m4 -O2:

        mov.b   @(1,r4),r0
        mov.b   @r4,r1
        add     r0,r1
        mov.b   @(2,r4),r0
        rts
        add     r1,r0

compiled with -m4 -Os:

        mov.b   @(1,r4),r0
        mov.b   @r4,r2
        add     r0,r2
        mov.b   @(2,r4),r0
        add     r0,r2
        rts
        mov     r2,r0

As seen above, the memory accesses are turned into reg+disp addressing,
although on SH using post-inc addressing would be cheaper (as it is also
reported 'sh_address_cost').
Turning off ivopts leaves the post-inc addressing and points out the original
problem of this PR.  I believe that the proper fix for this would be having PR
56590 fixed.