[Bug c/21985] New: miscompiled or wrong code snippet?

[Bug c/21985] New: miscompiled or wrong code snippet?

[marcus at jet dot franken dot de]

Hi,  
  
the attached code compiles fine and does calculate the offset between  
the current stackpointer and the passed point in gcc versions before 4.0.  
  
In 4.0 the expression is reduced to -16384 even in the t03.generic dump  
which makes me suspect a parser problem.  
 
Or it might just be wrong code... I am not sure.

-- 
           Summary: miscompiled or wrong code snippet?
           Product: gcc
           Version: 4.0.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: c
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: marcus at jet dot franken dot de
                CC: gcc-bugs at gcc dot gnu dot org
 GCC build triplet: i686-pc-linux-gnu
  GCC host triplet: i686-pc-linux-gnu
GCC target triplet: i686-pc-linux-gnu


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21985

[Bug c/21985] miscompiled or wrong code snippet?

[marcus at jet dot franken dot de]

------- Additional Comments From marcus at jet dot franken dot de  2005-06-09 16:14 -------
Created an attachment (id=9055)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=9055&action=view)
xx.c

gcc -c -O2 xx.c

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21985

[Bug middle-end/21985] [4.0/4.1 Regression] miscompiled or wrong code snippet?

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-06-09 16:26 -------
Confirmed, caused by:
2004-11-10  Zdenek Dvorak  <dvorakz@suse.cz>
        * fold-const.c (fold): Attempt to use ptr_difference_const whenever
        one of the arguments of MINUS_EXPR is an address.
        (split_address_to_core_and_offset): New function.
        (ptr_difference_const): Handle case when one of the operands is a
        pointer.


-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |rakdver at gcc dot gnu dot
                   |                            |org, pinskia at gcc dot gnu
                   |                            |dot org
             Status|UNCONFIRMED                 |NEW
          Component|c                           |middle-end
     Ever Confirmed|                            |1
           Keywords|                            |wrong-code
   Last reconfirmed|0000-00-00 00:00:00         |2005-06-09 16:26:32
               date|                            |
            Summary|miscompiled or wrong code   |[4.0/4.1 Regression]
                   |snippet?                    |miscompiled or wrong code
                   |                            |snippet?
   Target Milestone|---                         |4.0.2


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21985

[Bug middle-end/21985] [4.0/4.1 Regression] miscompiled or wrong code snippet?

[pinskia at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.0.2                       |4.0.1


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21985

[Bug middle-end/21985] [4.0/4.1 Regression] miscompiled or wrong code snippet?

[aj at gcc dot gnu dot org]

------- Additional Comments From aj at gcc dot gnu dot org  2005-06-09 17:01 -------
Let me just add the following comment so that searches for "grub miscompilation"
will find this bug:
This snippet is based on code in the grub bootloader which does not work if
compiled by GCC 4.0.0.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |aj at gcc dot gnu dot org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21985

[Bug middle-end/21985] [4.0/4.1 Regression] miscompiled or wrong code snippet?

[falk at debian dot org]

------- Additional Comments From falk at debian dot org  2005-06-09 21:48 -------
Pointer subtraction is only defined for pointers to the same array object.
So this code has undefined behaviour.


-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |INVALID


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21985

[Bug middle-end/21985] [4.0/4.1 Regression] miscompiled or wrong code snippet?

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-06-09 21:55 -------
(In reply to comment #4)
> Pointer subtraction is only defined for pointers to the same array object.
> So this code has undefined behaviour.

Yes that might be true but we can get a testcase where it fails even though it is valid code:
void abort (void);
int
main (void)
{
  unsigned char *b;
  unsigned size;

  b = ((unsigned char *)&b)+1;

  /* location of pointer on stack - pointer - constvar */
  size = ((unsigned char *) &b) - b;
  if (size!=-1)
    abort ();
  return 0;
}

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
         Resolution|INVALID                     |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21985

[Bug middle-end/21985] [4.0/4.1 Regression] miscompiled or wrong code snippet?

[steven at gcc dot gnu dot org]

------- Additional Comments From steven at gcc dot gnu dot org  2005-06-10 23:41 -------
Must be a fold bug, caused by this: 
http://gcc.gnu.org/ml/gcc-patches/2004-11/msg00715.html 
 
My .gimple dump already looks wrong: 
 
main () 
{ 
  unsigned char * b.0; 
  unsigned char * D.1575; 
  int D.1576; 
  unsigned char * b; 
  unsigned int size; 
 
  b.0 = (unsigned char *) &b; 
  D.1575 = b.0 + 1B; 
  b = D.1575; 
  size = 0; 
  if (size != 4294967295) 
    { 
      abort (); 
    } 
  else 
    { 
 
    } 
  D.1576 = 0; 
  return D.1576; 
} 
 
Note "size = 0" which is BS. 
 

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|2005-06-09 16:26:32         |2005-06-10 23:41:31
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21985

[Bug middle-end/21985] [4.0/4.1 Regression] miscompiled or wrong code snippet?

[steven at gcc dot gnu dot org]

------- Additional Comments From steven at gcc dot gnu dot org  2005-06-10 23:45 -------
This is broken: 
 
*************** fold (tree expr) 
*** 7074,7084 **** 
        { 
        HOST_WIDE_INT diff; 
   
!       if (TREE_CODE (arg0) == ADDR_EXPR 
!           && TREE_CODE (arg1) == ADDR_EXPR 
!           && ptr_difference_const (TREE_OPERAND (arg0, 0), 
!                                    TREE_OPERAND (arg1, 0), 
!                                    &diff)) 
          return build_int_cst_type (type, diff); 
        } 
           
--- 7074,7082 ---- 
        { 
        HOST_WIDE_INT diff; 
   
!       if ((TREE_CODE (arg0) == ADDR_EXPR 
!            || TREE_CODE (arg1) == ADDR_EXPR) 
!           && ptr_difference_const (arg0, arg1, &diff)) 
          return build_int_cst_type (type, diff); 
        } 
      
ptr_difference_const returns true for: 
#1  0x00000000006d5593 in fold_binary (code=MINUS_EXPR, type=0x2a95896c30, 
op0=0x2a95a3d240, 
    op1=0x2a95a3d200) at fold-const.c:7958 
7958            if ((TREE_CODE (arg0) == ADDR_EXPR 
(gdb) p ptr_difference_const (arg0, arg1, &diff) 
$3 = 1 '\001' 
(gdb) p debug_generic_expr(op1) 
(long intD.2) bD.1572 
$4 = void 
(gdb) p debug_generic_expr(op1) 
(long intD.2) bD.1572 
$5 = void 
(gdb)  
 

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21985

[Bug middle-end/21985] [4.0/4.1 Regression] miscompiled or wrong code snippet?

[steven at gcc dot gnu dot org]

------- Additional Comments From steven at gcc dot gnu dot org  2005-06-10 23:46 -------
Make that: 
(gdb) p debug_generic_expr(arg0) 
&bD.1572 
$7 = void 
(gdb) p debug_generic_expr(arg1) 
bD.1572 
$8 = void 
(gdb)     

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21985

[Bug middle-end/21985] [4.0/4.1 Regression] miscompiled or wrong code snippet?

[pinskia at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|REOPENED                    |NEW


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21985

[Bug middle-end/21985] [4.0/4.1 Regression] miscompiled or wrong code snippet?

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2005-06-13 14:59 -------
Subject: Bug 21985

CVSROOT:	/cvs/gcc
Module name:	gcc
Changes by:	rakdver@gcc.gnu.org	2005-06-13 14:59:40

Modified files:
	gcc            : ChangeLog fold-const.c 
	gcc/testsuite  : ChangeLog 
Added files:
	gcc/testsuite/gcc.dg/tree-ssa: pr21985.c 

Log message:
	PR middle-end/21985
	* fold-const.c (split_address_to_core_and_offset): Always return
	the address of the base object.
	
	* gcc.dg/tree-ssa/pr21985.c: New test.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gcc&r1=2.9135&r2=2.9136
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/fold-const.c.diff?cvsroot=gcc&r1=1.594&r2=1.595
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/ChangeLog.diff?cvsroot=gcc&r1=1.5629&r2=1.5630
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/gcc.dg/tree-ssa/pr21985.c.diff?cvsroot=gcc&r1=NONE&r2=1.1



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21985

[Bug middle-end/21985] [4.0/4.1 Regression] miscompiled or wrong code snippet?

[jsm28 at gcc dot gnu dot org]

------- Additional Comments From jsm28 at gcc dot gnu dot org  2005-06-14 20:26 -------
The test pr21985.c is FAILing on hppa64-hpux and ia64-hpux -mlp64.  It looks
like it is simply broken with regard to 64-bit platforms.

FAIL: gcc.dg/tree-ssa/pr21985.c (test for excess errors)
FAIL: gcc.dg/tree-ssa/pr21985.c scan-tree-dump-times foo \([0-9]*\) 2

/home/gcc/nightlies/src-mainline-2005-06-14/gcc-mainline/gcc/testsuite/gcc.dg/tree-ssa/pr21985.c:
In function 'main':
/home/gcc/nightlies/src-mainline-2005-06-14/gcc-mainline/gcc/testsuite/gcc.dg/tree-ssa/pr21985.c:26:
warning: cast from pointer to integer of different size
/home/gcc/nightlies/src-mainline-2005-06-14/gcc-mainline/gcc/testsuite/gcc.dg/tree-ssa/pr21985.c:26:
warning: cast from pointer to integer of different size


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21985

[Bug middle-end/21985] [4.0/4.1 Regression] miscompiled or wrong code snippet?

[hp at gcc dot gnu dot org]

------- Additional Comments From hp at gcc dot gnu dot org  2005-06-18 17:09 -------
Regarding comment #10, I can confirm that the test fails on mmix-knuth-mmixware
too (a 64-bit target), ever since it was committed; but not on cris-elf,
cris-axis-linux-gnu (32-bit targets).

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21985

[Bug middle-end/21985] [4.0 Regression] miscompiled or wrong code snippet?

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-06-20 00:03 -------
(In reply to comment #11)
> Regarding comment #10, I can confirm that the test fails on mmix-knuth-mmixware
> too (a 64-bit target), ever since it was committed; but not on cris-elf,
> cris-axis-linux-gnu (32-bit targets).

It should be passing now.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|[4.0/4.1 Regression]        |[4.0 Regression] miscompiled
                   |miscompiled or wrong code   |or wrong code snippet?
                   |snippet?                    |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21985

[Bug middle-end/21985] [4.0 Regression] miscompiled or wrong code snippet?

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2005-06-24 03:44 -------
Subject: Bug 21985

CVSROOT:	/cvs/gcc
Module name:	gcc
Branch: 	gcc-4_0-branch
Changes by:	mmitchel@gcc.gnu.org	2005-06-24 03:44:51

Modified files:
	gcc            : ChangeLog fold-const.c 
	gcc/testsuite  : ChangeLog 
Added files:
	gcc/testsuite/gcc.dg/tree-ssa: pr21985.c 

Log message:
	Backport:
	2005-06-13  Zdenek Dvorak  <dvorakz@suse.cz>
	PR middle-end/21985
	* fold-const.c (split_address_to_core_and_offset): Always return
	the address of the base object.
	
	Backport:
	2005-06-19  Richard Henderson  <rth@redhat.com>
	* gcc.dg/tree-ssa/pr21985.c: Use size_t.
	2005-06-13  Zdenek Dvorak  <dvorakz@suse.cz>
	PR middle-end/21985
	* gcc.dg/tree-ssa/pr21985.c: New test.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gcc&only_with_tag=gcc-4_0-branch&r1=2.7592.2.291&r2=2.7592.2.292
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/fold-const.c.diff?cvsroot=gcc&only_with_tag=gcc-4_0-branch&r1=1.517.2.10&r2=1.517.2.11
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/gcc.dg/tree-ssa/pr21985.c.diff?cvsroot=gcc&only_with_tag=gcc-4_0-branch&r1=NONE&r2=1.2.2.1
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/ChangeLog.diff?cvsroot=gcc&only_with_tag=gcc-4_0-branch&r1=1.5084.2.246&r2=1.5084.2.247



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21985

[Bug middle-end/21985] [4.0 Regression] miscompiled or wrong code snippet?

[mmitchel at gcc dot gnu dot org]

------- Additional Comments From mmitchel at gcc dot gnu dot org  2005-06-24 03:53 -------
Fixed in 4.0.1.

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21985