[Bug tree-optimization/23326] New: [4.0 Regression] Wrong code from forwprop

[Bug tree-optimization/23326] New: [4.0 Regression] Wrong code from forwprop

[rguenth at gcc dot gnu dot org]

The following testcase reduced from Kopete is miscompiled at -O2

struct Bar { bool is() const; int status() const; };

void foobar(void);
void foobar2(void);

void Foo(const Bar& x, const Bar& y)
{
  const bool wasOffline = !x.is();
  const bool isOffline = !y.is();
  if (wasOffline || y.status() == 10)
    foobar();
  if (wasOffline != isOffline)
    foobar2();
}

.t23.mergephi:
<bb 0>:
  D.1589_3 = is (x_2);
  wasOffline_4 = !D.1589_3;
  D.1590_6 = is (y_5);
  isOffline_7 = !D.1590_6;
  if (wasOffline_4) goto <L1>; else goto <L0>;

<L0>:;
  D.1596_10 = status (y_5);
  if (D.1596_10 == 10) goto <L1>; else goto <L5>;

<L1>:;
  foobar ();

<L5>:;
  if (wasOffline_4 != isOffline_7) goto <L6>; else goto <L7>;

<L6>:;
  foobar2 ();

<L7>:;
  return;

.t24.forwprop:
<bb 0>:
  D.1589_3 = is (x_2);
  D.1590_6 = is (y_5);
  isOffline_7 = !D.1590_6;
  if (D.1589_3 == 0) goto <L1>; else goto <L0>;

<L0>:;
  D.1596_10 = status (y_5);
  if (D.1596_10 == 10) goto <L1>; else goto <L5>;

<L1>:;
  foobar ();

<L5>:;
  if (D.1589_3 != 0) goto <L6>; else goto <L7>;

<L6>:;
  foobar2 ();

<L7>:;
  return;

note that the comparison preceding foobar2() is hosed now,
we even don't use isOffline or D.1590 anymore.

-- 
           Summary: [4.0 Regression] Wrong code from forwprop
           Product: gcc
           Version: 4.0.2
            Status: UNCONFIRMED
          Keywords: wrong-code
          Severity: critical
          Priority: P2
         Component: tree-optimization
        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,matz at suse dot de


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23326

[Bug tree-optimization/23326] [4.0 Regression] Wrong code from forwprop

[rguenth at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
      Known to fail|                            |4.0.2
      Known to work|                            |4.1.0
   Target Milestone|---                         |4.0.2


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23326

[Bug tree-optimization/23326] [4.0 Regression] Wrong code from forwprop

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-08-11 15:29 -------
Confirmed.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|                            |1
   Last reconfirmed|0000-00-00 00:00:00         |2005-08-11 15:29:17
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23326

[Bug tree-optimization/23326] [4.0 Regression] Wrong code from forwprop

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-08-11 15:51 -------
Oh, this is truely fixed for 4.1 and not a latent bug there.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23326

[Bug tree-optimization/23326] [4.0 Regression] Wrong code from forwprop

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-08-11 16:03 -------
Created an attachment (id=9474)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=9474&action=view)
Patch which fixes but needs testing

This patch fixes the problem by continuing if we don't have a SSA_NAME or
zero/one in the conditional.

Richard if you could test this for me, that would be nice.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |pinskia at gcc dot gnu dot
                   |dot org                     |org
             Status|NEW                         |ASSIGNED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23326

[Bug tree-optimization/23326] [4.0 Regression] Wrong code from forwprop

[rguenth at tat dot physik dot uni-tuebingen dot de]

------- Additional Comments From rguenth at tat dot physik dot uni-tuebingen dot de  2005-08-11 17:43 -------
I'll do that.  Though

+	      /* If we don't have <NE_EXPR/NE_EXPR x INT_CST>, then we cannot
+	         optimize this case.  */
+	      if ((cond_code == NE_EXPR || cond_code == EQ_EXPR)
+	          && TREE_CODE (TREE_OPERAND (cond, 1)) != INTEGER_CST)
+		continue;

should probably read

+	      /* If we don't have <NE_EXPR/EQ_EXPR x INT_CST>, then we cannot
+	         optimize this case.  */
+	      if (!((cond_code == NE_EXPR || cond_code == EQ_EXPR)
+	            && TREE_CODE (TREE_OPERAND (cond, 1)) == INTEGER_CST))
+		continue;

because else we might get f.i. LE_EXPR passing through?  Maybe the little
context confuses me here, though.

I'll have a look before testing.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23326

[Bug tree-optimization/23326] [4.0 Regression] Wrong code from forwprop

[pinskia at physics dot uc dot edu]

------- Additional Comments From pinskia at physics dot uc dot edu  2005-08-11 17:45 -------
Subject: Re:  [4.0 Regression] Wrong code from forwprop


On Aug 11, 2005, at 1:43 PM, rguenth at tat dot physik dot 
uni-tuebingen dot de wrote:

> because else we might get f.i. LE_EXPR passing through?  Maybe the 
> little
> context confuses me here, though.

We will never get LE_EXPR here as this is always a boolean type.

-- Pinski



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23326

[Bug tree-optimization/23326] [4.0 Regression] Wrong code from forwprop

[belyshev at depni dot sinp dot msu dot ru]

------- Additional Comments From belyshev at depni dot sinp dot msu dot ru  2005-08-11 18:05 -------
// self-contained C++ testcase, compile with -O2

extern "C" void abort (void);

int j;

void foo (bool x, bool y)
{
  if (!x)
    j = 0;
  if (!x == y)
    j = 1;
}

int main (void)
{
  foo (1, 1);
  if (j)
    abort ();
}


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23326

[Bug tree-optimization/23326] [4.0 Regression] Wrong code from forwprop

[rguenth at gcc dot gnu dot org]

------- Additional Comments From rguenth at gcc dot gnu dot org  2005-08-12 10:57 -------
Patch posted.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                URL|                            |http://gcc.gnu.org/ml/gcc-
                   |                            |patches/2005-
                   |                            |08/msg00738.html
           Keywords|                            |patch


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23326

[Bug tree-optimization/23326] [4.0 Regression] Wrong code from forwprop

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-08-22 12:15 -------
*** Bug 23512 has been marked as a duplicate of this bug. ***

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |mec at google dot com


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23326

[Bug tree-optimization/23326] [4.0 Regression] Wrong code from forwprop

[matz at suse dot de]

------- Additional Comments From matz at suse dot de  2005-09-01 15:11 -------
This still isn't in the 4.0 branch.  Perhaps ping it? 

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23326

[Bug tree-optimization/23326] [4.0 Regression] Wrong code from forwprop

[rguenth at gcc dot gnu dot org]

------- Additional Comments From rguenth at gcc dot gnu dot org  2005-09-01 15:18 -------
It's in the tracker and I pinged it once already.  One could commit it as
obvious, though ;)

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23326

[Bug tree-optimization/23326] [4.0 Regression] Wrong code from forwprop

[rguenth at gcc dot gnu dot org]

------- Additional Comments From rguenth at gcc dot gnu dot org  2005-09-03 19:55 -------
Fixed.

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23326

[Bug tree-optimization/23326] [4.0 Regression] Wrong code from forwprop

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2005-09-03 19:55 -------
Subject: Bug 23326

CVSROOT:	/cvs/gcc
Module name:	gcc
Branch: 	gcc-4_0-branch
Changes by:	rguenth@gcc.gnu.org	2005-09-03 19:54:51

Modified files:
	gcc            : ChangeLog tree-ssa-forwprop.c 
	gcc/testsuite  : ChangeLog 
Added files:
	gcc/testsuite/g++.dg/tree-ssa: pr23326.C 

Log message:
	2005-09-03  Richard Guenther  <rguenther@suse.de>
	Andrew Pinski  <pinskia@gcc.gnu.org>
	Serge Belyshev  <belyshev@depni.sinp.msu.ru>
	
	PR tree-optimization/23326
	* tree-ssa-forwprop.c (substitute_single_use_vars): Only
	do transformation if valid.
	
	* g++.dg/tree-ssa/pr23326.C: New testcase.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gcc&only_with_tag=gcc-4_0-branch&r1=2.7592.2.409&r2=2.7592.2.410
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/tree-ssa-forwprop.c.diff?cvsroot=gcc&only_with_tag=gcc-4_0-branch&r1=2.11&r2=2.11.2.1
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/ChangeLog.diff?cvsroot=gcc&only_with_tag=gcc-4_0-branch&r1=1.5084.2.372&r2=1.5084.2.373
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/g++.dg/tree-ssa/pr23326.C.diff?cvsroot=gcc&only_with_tag=gcc-4_0-branch&r1=NONE&r2=1.1.2.1



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23326

[Bug tree-optimization/23326] [4.0 Regression] Wrong code from forwprop

[pinskia at gcc dot gnu dot org]

------- Comment #14 from pinskia at gcc dot gnu dot org  2006-05-17 19:17 -------
*** Bug 27647 has been marked as a duplicate of this bug. ***


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |scharest at druide dot com


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23326

[Bug tree-optimization/23326] [4.0 Regression] Wrong code from forwprop

[pinskia at gcc dot gnu dot org]

------- Comment #13 from pinskia at gcc dot gnu dot org  2005-11-08 16:44 -------
*** Bug 24737 has been marked as a duplicate of this bug. ***


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |fredrik dot littmarck at
                   |                            |propellerheads dot se


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23326