From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gcc-bugs-return-427115-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Received: (qmail 7997 invoked by alias); 1 Aug 2013 16:48:38 -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 7923 invoked by uid 48); 1 Aug 2013 16:48:36 -0000
From: "jamborm at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug middle-end/58041] Unaligned access to arrays in packed structure
Date: Thu, 01 Aug 2013 16:48:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: middle-end
X-Bugzilla-Version: 4.9.0
X-Bugzilla-Keywords:
X-Bugzilla-Severity: normal
X-Bugzilla-Who: jamborm at gcc dot gnu.org
X-Bugzilla-Status: ASSIGNED
X-Bugzilla-Priority: P3
X-Bugzilla-Assigned-To: wschmidt at gcc dot gnu.org
X-Bugzilla-Target-Milestone: ---
X-Bugzilla-Flags:
X-Bugzilla-Changed-Fields:
Message-ID: <bug-58041-4-f2WrWwDZZ2@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-58041-4@http.gcc.gnu.org/bugzilla/>
References: <bug-58041-4@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: 2013-08/txt/msg00039.txt.bz2

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58041

--- Comment #9 from Martin Jambor <jamborm at gcc dot gnu.org> ---
More specifically, if I am correct assuming that the MEM_REF
replace_ref builds always accesses exactly the same memory as the
previous access *expr does (and only the address is computed better)
and that *expr is how the function accessed that memory before, then I
think what you need is the following (untested, except that I know it
seems to fix the testcase, at least the assembly looks very different
:-)

*** /tmp/heQHTs_gimple-ssa-strength-reduction.c Thu Aug  1 18:43:45 2013
--- gcc/gimple-ssa-strength-reduction.c Thu Aug  1 18:38:31 2013
*************** dump_incr_vec (void)
*** 1728,1738 ****
  static void
  replace_ref (tree *expr, slsr_cand_t c)
  {
!   tree add_expr = fold_build2 (POINTER_PLUS_EXPR, TREE_TYPE (c->base_expr),
!                              c->base_expr, c->stride);
!   tree mem_ref = fold_build2 (MEM_REF, TREE_TYPE (*expr), add_expr,
!                             double_int_to_tree (c->cand_type, c->index));
!   
    /* Gimplify the base addressing expression for the new MEM_REF tree.  */
    gimple_stmt_iterator gsi = gsi_for_stmt (c->cand_stmt);
    TREE_OPERAND (mem_ref, 0)
--- 1728,1748 ----
  static void
  replace_ref (tree *expr, slsr_cand_t c)
  {
!   tree add_expr, mem_ref, acc_type = TREE_TYPE (*expr);
!   unsigned HOST_WIDE_INT misalign;
!   unsigned align;
! 
!   get_object_alignment_1 (*expr, &align, &misalign);
!   if (misalign != 0)
!     align = (misalign & -misalign);
!   if (align < TYPE_ALIGN (acc_type))
!     acc_type = build_aligned_type (acc_type, align);
! 
!   add_expr = fold_build2 (POINTER_PLUS_EXPR, TREE_TYPE (c->base_expr),
!                         c->base_expr, c->stride);
!   mem_ref = fold_build2 (MEM_REF, acc_type, add_expr,
!                        double_int_to_tree (c->cand_type, c->index));
! 
    /* Gimplify the base addressing expression for the new MEM_REF tree.  */
    gimple_stmt_iterator gsi = gsi_for_stmt (c->cand_stmt);
    TREE_OPERAND (mem_ref, 0)