[Bug c/52286] New: wrong code bug

[Bug c/52286] New: wrong code bug

[regehr at cs dot utah.edu]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52286

             Bug #: 52286
           Summary: wrong code bug
    Classification: Unclassified
           Product: gcc
           Version: 4.7.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: regehr@cs.utah.edu

[Bug c/52286] wrong code bug

[regehr at cs dot utah.edu]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52286

John Regehr <regehr at cs dot utah.edu> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |chenyang at cs dot utah.edu

--- Comment #1 from John Regehr <regehr at cs dot utah.edu> 2012-02-16 18:48:18 UTC ---
[regehr@gamow 1]$ current-gcc -O0 small.c ; ./a.out 
1
[regehr@gamow 1]$ current-gcc -O1 small.c ; ./a.out 
0
[regehr@gamow 1]$ current-gcc -v
Using built-in specs.
COLLECT_GCC=current-gcc
COLLECT_LTO_WRAPPER=/uusoc/exports/scratch/regehr/z/compiler-install/gcc-r184311-install/bin/../libexec/gcc/x86_64-unknown-linux-gnu/4.7.0/lto-wrapper
Target: x86_64-unknown-linux-gnu
Configured with: ../configure --with-libelf=/usr/local --enable-lto
--prefix=/home/regehr/z/compiler-install/gcc-r184311-install
--program-prefix=r184311- --enable-languages=c,c++
Thread model: posix
gcc version 4.7.0 20120216 (experimental) (GCC) 
[regehr@gamow 1]$ cat small.c
int printf ( const char *, ... );

int c,
  a,
  b;

int
fn1 ( void )
{
  return c < 0;
}

int
main (  )
{
  b = ( ~a | 0 >= 0 ) & 0x98685255F;
  printf ( "%d\n", ( ( b ) < 0 ) );
  return 0;
}

[Bug c/52286] wrong code bug

[regehr at cs dot utah.edu]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52286

--- Comment #2 from John Regehr <regehr at cs dot utah.edu> 2012-02-16 18:51:50 UTC ---
Sorry, previous one wasn't quite reduced.


int printf ( const char *, ... );

int a, b;

int main (void)
{
  b = (~a | 0 >= 0) & 0x98685255F;
  printf ("%d\n", b < 0);
  return 0;
}

[Bug c/52286] wrong code bug

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52286

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jakub at gcc dot gnu.org

--- Comment #3 from Jakub Jelinek <jakub at gcc dot gnu.org> 2012-02-16 19:16:28 UTC ---
Started with http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=162943

Slightly adjusted testcase:
extern void abort (void);

int
main ()
{
  int a, b;
  asm volatile ("" : "=r" (a) : "0" (0));
  b = (~a | 1) & -2038094497;
  if (b >= 0)
    abort ();
  return 0;
}

[Bug c/52286] [4.6/4.7 Regression] wrong code bug

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52286

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |ASSIGNED
   Last reconfirmed|                            |2012-02-16
         AssignedTo|unassigned at gcc dot       |jakub at gcc dot gnu.org
                   |gnu.org                     |
   Target Milestone|---                         |4.6.3
            Summary|wrong code bug              |[4.6/4.7 Regression] wrong
                   |                            |code bug
     Ever Confirmed|0                           |1

[Bug tree-optimization/52286] [4.6/4.7 Regression] wrong code bug

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52286

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |wrong-code
          Component|c                           |tree-optimization
   Target Milestone|4.6.3                       |---

--- Comment #4 from Andrew Pinski <pinskia at gcc dot gnu.org> 2012-02-17 04:48:25 UTC ---
Confirmed.
D.1711_3 = D.1710_2 & 2256872798;
which is likely CONSTANT
Lattice value changed to CONSTANT Lattice value changed to CONSTANT
0x00000000000000000 (0x0000000008685255e).  Adding SSA edges to worklist.


The problem is obvious, 0x0000000008685255e should have its sign bit extended
to all the way to the end.

[Bug tree-optimization/52286] [4.6/4.7 Regression] wrong code bug

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52286

--- Comment #5 from Andrew Pinski <pinskia at gcc dot gnu.org> 2012-02-17 04:50:20 UTC ---
double_int_negative_p is returning true because of that.

[Bug tree-optimization/52286] [4.6/4.7 Regression] wrong code bug

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52286

--- Comment #6 from Andrew Pinski <pinskia at gcc dot gnu.org> 2012-02-17 04:52:36 UTC ---
Note VRP has the same issue.

[Bug tree-optimization/52286] [4.6/4.7 Regression] wrong code bug

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52286

--- Comment #7 from Jakub Jelinek <jakub at gcc dot gnu.org> 2012-02-17 10:31:14 UTC ---
Yeah, the bug is in fold-const.c, working on a fix already.

[Bug tree-optimization/52286] [4.6/4.7 Regression] wrong code bug

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52286

--- Comment #8 from Jakub Jelinek <jakub at gcc dot gnu.org> 2012-02-17 11:17:12 UTC ---
Created attachment 26694
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=26694
gcc47-pr52286.patch

Untested fix.

[Bug tree-optimization/52286] [4.6/4.7 Regression] wrong code bug

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52286

Richard Guenther <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |4.6.3

[Bug tree-optimization/52286] [4.6/4.7 Regression] wrong code bug

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52286

--- Comment #9 from Jakub Jelinek <jakub at gcc dot gnu.org> 2012-02-20 12:19:54 UTC ---
Author: jakub
Date: Mon Feb 20 12:19:47 2012
New Revision: 184391

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=184391
Log:
    PR tree-optimization/52286
    * fold-const.c (fold_binary_loc): For (X & C1) | C2
    optimization use double_int_to_tree instead of build_int_cst_wide,
    rewrite to use double_int vars.

    * gcc.c-torture/execute/pr52286.c: New test.

Added:
    trunk/gcc/testsuite/gcc.c-torture/execute/pr52286.c
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/fold-const.c
    trunk/gcc/testsuite/ChangeLog

[Bug tree-optimization/52286] [4.6 Regression] wrong code bug

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52286

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|[4.6/4.7 Regression] wrong  |[4.6 Regression] wrong code
                   |code bug                    |bug

--- Comment #10 from Jakub Jelinek <jakub at gcc dot gnu.org> 2012-02-20 12:52:41 UTC ---
Fixed for 4.7 so far.

[Bug tree-optimization/52286] [4.6 Regression] wrong code bug

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52286

--- Comment #11 from Jakub Jelinek <jakub at gcc dot gnu.org> 2012-02-23 18:20:26 UTC ---
Author: jakub
Date: Thu Feb 23 18:20:19 2012
New Revision: 184520

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=184520
Log:
    Backported from trunk
    2012-02-20  Georg-Johann Lay  <avr@gjlay.de>

    * gcc.c-torture/execute/pr52286.c: Fix FAIL on 16-bit int platforms.

    2012-02-20  Jakub Jelinek  <jakub@redhat.com>

    PR tree-optimization/52286
    * fold-const.c (fold_binary_loc): For (X & C1) | C2
    optimization use double_int_to_tree instead of build_int_cst_wide,
    rewrite to use double_int vars.

    * gcc.c-torture/execute/pr52286.c: New test.

Added:
    branches/gcc-4_6-branch/gcc/testsuite/gcc.c-torture/execute/pr52286.c
Modified:
    branches/gcc-4_6-branch/gcc/ChangeLog
    branches/gcc-4_6-branch/gcc/fold-const.c
    branches/gcc-4_6-branch/gcc/testsuite/ChangeLog

[Bug tree-optimization/52286] [4.6 Regression] wrong code bug

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52286

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

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

--- Comment #12 from Jakub Jelinek <jakub at gcc dot gnu.org> 2012-02-23 21:29:10 UTC ---
Fixed.