[Bug rtl-optimization/66187] New: wrong code at -O1, -O2 and -O3 on x86_64-linux-gnu

[Bug rtl-optimization/66187] New: wrong code at -O1, -O2 and -O3 on x86_64-linux-gnu

[su at cs dot ucdavis.edu]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66187

            Bug ID: 66187
           Summary: wrong code at -O1, -O2 and -O3 on x86_64-linux-gnu
           Product: gcc
           Version: 6.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: rtl-optimization
          Assignee: unassigned at gcc dot gnu.org
          Reporter: su at cs dot ucdavis.edu
  Target Milestone: ---

The current gcc trunk miscompiles the following code on x86_64-linux at -O1,
-O2 and -O3 in both 32-bit and 64-bit modes. 

This is a regression from 5.1.x. 


$ gcc-trunk -v
Using built-in specs.
COLLECT_GCC=gcc-trunk
COLLECT_LTO_WRAPPER=/usr/local/gcc-trunk/libexec/gcc/x86_64-unknown-linux-gnu/6.0.0/lto-wrapper
Target: x86_64-unknown-linux-gnu
Configured with: ../gcc-trunk/configure --prefix=/usr/local/gcc-trunk
--enable-languages=c,c++ --disable-werror --enable-multilib
Thread model: posix
gcc version 6.0.0 20150517 (experimental) [trunk revision 223265] (GCC) 
$ 
$ gcc-trunk -O0 small.c; ./a.out
$ gcc-trunk -Os small.c; ./a.out
$ gcc-5.1 -O1 small.c; ./a.out
$ 
$ gcc-trunk -O1 small.c
$ ./a.out
Aborted (core dumped)
$ 


---------------------------------


struct
{
  short f1;
} b;

int a = 1, e = -1;
short f;

int
main ()
{
  int g;
  f = e;
  g = b.f1 < 0 ? 0 : f + b.f1;
  if ((g & -4) < 0)
    a = 0;

  if (a)
    __builtin_abort ();

  return 0;
}

[Bug rtl-optimization/66187] [6 Regression] wrong code at -O1, -O2 and -O3 on x86_64-linux-gnu

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66187

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2015-05-18
                 CC|                            |jakub at gcc dot gnu.org
   Target Milestone|---                         |6.0
            Summary|wrong code at -O1, -O2 and  |[6 Regression] wrong code
                   |-O3 on x86_64-linux-gnu     |at -O1, -O2 and -O3 on
                   |                            |x86_64-linux-gnu
     Ever confirmed|0                           |1

--- Comment #1 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Started with r222877.
I bet the problem is in:
        && (tree_int_cst_min_precision (@4, UNSIGNED)
            <= TYPE_PRECISION (TREE_TYPE (@0)))
which for signed types IMHO should be < instead.

[Bug rtl-optimization/66187] [6 Regression] wrong code at -O1, -O2 and -O3 on x86_64-linux-gnu

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66187

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED
           Assignee|unassigned at gcc dot gnu.org      |jakub at gcc dot gnu.org

--- Comment #2 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Created attachment 35560
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=35560&action=edit
gcc6-pr66187.patch

Untested fix.  Perhaps for the unsigned type and plus only, we could do even
better, we don't really need to ensure the mask fits into precision, but
perhaps just that the bit immediately above precision is clear (all bits above
that are necessarily clear in the plus).

[Bug rtl-optimization/66187] [6 Regression] wrong code at -O1, -O2 and -O3 on x86_64-linux-gnu

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66187

--- Comment #3 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Author: jakub
Date: Tue May 19 13:54:32 2015
New Revision: 223366

URL: https://gcc.gnu.org/viewcvs?rev=223366&root=gcc&view=rev
Log:
        PR tree-optimization/66187
        * match.pd ((bit_and (plus/minus (convert @0) (convert @1)) mask)):
        Pass TYPE_SIGN to tree_int_cst_min_precision.  If
        !TYPE_OVERFLOW_WRAPS, ensure @4 is non-negative.

        * gcc.c-torture/execute/pr66187.c: New test.
        * gcc.dg/pr66187-1.c: New test.
        * gcc.dg/pr66187-2.c: New test.

Added:
    trunk/gcc/testsuite/gcc.c-torture/execute/pr66187.c
    trunk/gcc/testsuite/gcc.dg/pr66187-1.c
    trunk/gcc/testsuite/gcc.dg/pr66187-2.c
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/match.pd
    trunk/gcc/testsuite/ChangeLog

[Bug rtl-optimization/66187] [6 Regression] wrong code at -O1, -O2 and -O3 on x86_64-linux-gnu

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66187

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

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

--- Comment #4 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Fixed, thanks for the report.