[Bug middle-end/19807] New: fold does not fold &a[4]-1

[Bug middle-end/19807] New: fold does not fold &a[4]-1

[rguenth at gcc dot gnu dot org]

PR19639 is pessimized by (?)

extern void link_error(void);
int main(void)
{
  int a[4];
  if (&a[2]-1 != &a[1])
    link_error();
  return 0;
}

where for C++ (!!) we do not fold the addition of ADDR_EXPR (ARRAY_REF (..))
 and INTEGER_CST into the ARRAY_REF offset.  For C the frontend avoids
passing ARRAY_REFs to fold.

-- 
           Summary: fold does not fold &a[4]-1
           Product: gcc
           Version: 4.0.0
            Status: UNCONFIRMED
          Keywords: missed-optimization
          Severity: enhancement
          Priority: P2
         Component: middle-end
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: rguenth at gcc dot gnu dot org
                CC: gcc-bugs at gcc dot gnu dot org
OtherBugsDependingO 19639
             nThis:


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19807

[Bug middle-end/19807] fold does not fold &a[4]-1

[rguenth at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
OtherBugsDependingO|19639                       |
              nThis|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19807

[Bug middle-end/19807] fold does not fold &a[4]-1

[rguenth at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
OtherBugsDependingO|                            |19639
              nThis|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19807

[Bug middle-end/19807] fold does not fold &a[4]-1

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-02-07 19:30 -------
Confirmed.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |pinskia at gcc dot gnu dot
                   |                            |org
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|                            |1
   Last reconfirmed|0000-00-00 00:00:00         |2005-02-07 19:30:41
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19807

[Bug middle-end/19807] fold does not fold &a[4]-1

[rguenth at gcc dot gnu dot org]

------- Additional Comments From rguenth at gcc dot gnu dot org  2005-02-09 11:15 -------
Another testcase is the following:

xtern void link_error(void);
void foo(int i)
{
        int a[4];
        if ((char*)&a[1] + 4*i + 4 != (char*)&a[i+2])
                link_error();
        if (&a[1] + i + 1 != &a[i+2])
                link_error();
}

which I have a patch for that works for C++ by extending the existing
try_move_mult_to_index and trying to avoid inserting (unsinged) casts
there.

Unfortunately, for C (whose initial trees do not contain ARRAY_REF exprs),
the frontend already inserts casts to unsigned (.03.generic):

foo (i)
{
  charD.1 * a.0D.1129;
  intD.0 D.1130;
  charD.1 * D.1131;
  charD.1 * D.1132;
  charD.1 * D.1133;
  unsigned intD.3 i.1D.1134;
  unsigned intD.3 D.1135;
  intD.0 * D.1136;
  intD.0 * D.1137;
  intD.0 * D.1138;
  charD.1 * D.1139;
  intD.0 aD.1128[4];

  a.0D.1129 = (charD.1 *) &aD.1128;
  D.1130 = iD.1125 * 4;
  D.1131 = (charD.1 *) D.1130;
  D.1132 = a.0D.1129 + D.1131;
  D.1133 = D.1132 + 8B;
  i.1D.1134 = (unsigned intD.3) iD.1125;
  D.1135 = i.1D.1134 * 4;
  D.1136 = (intD.0 *) D.1135;
  D.1137 = &aD.1128 + D.1136;
  D.1138 = D.1137 + 8B;
  D.1139 = (charD.1 *) D.1138;
  if (D.1133 != D.1139)
    {
      link_error ();
    }
  else
    {

    }
  if (0)
    {
      link_error ();
    }
  else
    {

    }
}

which maybe even wrong, if not hindering optimizations at least.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |rguenth at gcc dot gnu dot
                   |dot org                     |org
             Status|NEW                         |ASSIGNED
   Last reconfirmed|2005-02-07 19:30:41         |2005-02-09 11:15:29
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19807

[Bug middle-end/19807] fold does not fold &a[4]-1

[rguenth at gcc dot gnu dot org]

-- 
Bug 19807 depends on bug 21082, which changed state.

Bug 21082 Summary: &a[b] - &a[c] is not folded to b - c
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21082

           What    |Old Value                   |New Value
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |FIXED

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19807

[Bug middle-end/19807] fold does not fold &a[4]-1

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-04-25 14:49 -------
Patch posted: <http://gcc.gnu.org/ml/gcc-patches/2005-04/msg02592.html>.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                URL|                            |http://gcc.gnu.org/ml/gcc-
                   |                            |patches/2005-
                   |                            |04/msg02592.html
           Keywords|                            |patch


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19807

[Bug middle-end/19807] fold does not fold &a[4]-1

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2005-05-11 08:15 -------
Subject: Bug 19807

CVSROOT:	/cvs/gcc
Module name:	gcc
Changes by:	rguenth@gcc.gnu.org	2005-05-11 08:14:45

Modified files:
	gcc            : ChangeLog fold-const.c tree-ssa-propagate.c 
	gcc/testsuite  : ChangeLog 
Added files:
	gcc/testsuite/g++.dg/tree-ssa: pr19807.C 

Log message:
	2005-05-11  Richard Guenther  <rguenth@gcc.gnu.org>
	
	PR middle-end/19807
	PR tree-optimization/19639
	* fold-const.c (try_move_mult_to_index): Handle INTEGER_CST
	and generic summands for char* as s * delta, too, folding &a[i]
	CODE x to &a[i CODE x/s].  Use tree_int_cst_equal
	for comparison of steps.  Convert types for index addition.
	(fold_binary): Adjust the callers to always dispatch to
	try_move_mult_to_index.
	* tree-ssa-propagate.c (set_rhs): Avoid setting rhs to
	expr with non-gimple ARRAY_REF offset.
	
	* g++.dg/tree-ssa/pr19807.C: New testcase.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gcc&r1=2.8725&r2=2.8726
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/fold-const.c.diff?cvsroot=gcc&r1=1.576&r2=1.577
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/tree-ssa-propagate.c.diff?cvsroot=gcc&r1=2.18&r2=2.19
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/ChangeLog.diff?cvsroot=gcc&r1=1.5463&r2=1.5464
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/g++.dg/tree-ssa/pr19807.C.diff?cvsroot=gcc&r1=NONE&r2=1.1



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19807

[Bug middle-end/19807] fold does not fold &a[4]-1

[rguenth at gcc dot gnu dot org]

------- Additional Comments From rguenth at gcc dot gnu dot org  2005-05-11 09:07 -------
Fixed.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |FIXED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19807

[Bug middle-end/19807] fold does not fold &a[4]-1

[rguenth at gcc dot gnu dot org]

------- Additional Comments From rguenth at gcc dot gnu dot org  2005-05-11 09:08 -------
Always forget to set target milestone.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |4.1.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19807