[Bug tree-optimization/112585] New: wrong code at -O3 on x86_64-linux-gnu

[Bug tree-optimization/112585] New: wrong code at -O3 on x86_64-linux-gnu

[zhendong.su at inf dot ethz.ch]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=112585

            Bug ID: 112585
           Summary: wrong code at -O3 on x86_64-linux-gnu
           Product: gcc
           Version: unknown
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: tree-optimization
          Assignee: unassigned at gcc dot gnu.org
          Reporter: zhendong.su at inf dot ethz.ch
  Target Milestone: ---

This appears to be a recent regression as it does not reproduce with 13.2. 

Compiler Explorer: https://godbolt.org/z/ec94Yo6To


[563] % gcctk -v
Using built-in specs.
COLLECT_GCC=gcctk
COLLECT_LTO_WRAPPER=/local/suz-local/software/local/gcc-trunk/libexec/gcc/x86_64-pc-linux-gnu/14.0.0/lto-wrapper
Target: x86_64-pc-linux-gnu
Configured with: ../gcc-trunk/configure --disable-bootstrap
--enable-checking=yes --prefix=/local/suz-local/software/local/gcc-trunk
--enable-sanitizers --enable-languages=c,c++ --disable-werror --enable-multilib
Thread model: posix
Supported LTO compression algorithms: zlib
gcc version 14.0.0 20231117 (experimental) (GCC) 
[564] % 
[564] % gcctk -O2 small.c; ./a.out
[565] % 
[565] % gcctk -O3 small.c
[566] % ./a.out
Aborted
[567] % 
[567] % cat small.c
static int a[10], b;
char c, *d = &c, e;
int main() {
  int f = 0;
  for (; f < 9; f++) {
    a[f] = 1;
    a[f + 1] = e;
    for (b = 0; b < 8; b++)
      *d = 0;
  }
  if (a[2] != 1)
    __builtin_abort();
  return 0;
}

[Bug tree-optimization/112585] wrong code at -O3 on x86_64-linux-gnu since r14-5444

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=112585

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
            Summary|wrong code at -O3 on        |wrong code at -O3 on
                   |x86_64-linux-gnu            |x86_64-linux-gnu since
                   |                            |r14-5444
                 CC|                            |jakub at gcc dot gnu.org,
                   |                            |rguenth at gcc dot gnu.org
   Last reconfirmed|                            |2023-11-17
   Target Milestone|---                         |14.0
     Ever confirmed|0                           |1
           Priority|P3                          |P1

--- Comment #1 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Started with r14-5444-g5ea2965b499f9e491e45db19fedbccfccb75076a

[Bug tree-optimization/112585] wrong code at -O3 on x86_64-linux-gnu since r14-5444

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=112585

--- Comment #2 from Richard Biener <rguenth at gcc dot gnu.org> ---
Confirmed.  In .optimized all loops are unrolled and we have

  e.0_103 = e;
  _105 = (int) e.0_103;
  a[2] = _105;
...
  _7 = a[2];
  if (_7 != 1)
    goto <bb 3>; [0.00%]
  else
    goto <bb 4>; [100.00%]

  <bb 3> [count: 0]:
  __builtin_abort ();

loop distribution splits the loop, putting a loop with just a[f] = 1 first.
I think that's wrong.  Related to PR112281, here we have a zero distance
for the inner loop.

[Bug tree-optimization/112585] wrong code at -O3 on x86_64-linux-gnu since r14-5444

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=112585

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Blocks|                            |112281
            Version|unknown                     |14.0
             Status|NEW                         |ASSIGNED
           Assignee|unassigned at gcc dot gnu.org      |rguenth at gcc dot gnu.org

--- Comment #3 from Richard Biener <rguenth at gcc dot gnu.org> ---
Mine.


Referenced Bugs:

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=112281
[Bug 112281] [12/13 Regression] wrong code at -O3 on x86_64-linux-gnu since
r12-2097-g9f34b780b0461e

[Bug tree-optimization/112585] [14 Regression] wrong code at -O3 on x86_64-linux-gnu since r14-5444

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=112585

--- Comment #4 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Richard Biener <rguenth@gcc.gnu.org>:

https://gcc.gnu.org/g:04abafe9831f6867af1211ecae853ff373235b2d

commit r14-5555-g04abafe9831f6867af1211ecae853ff373235b2d
Author: Richard Biener <rguenther@suse.de>
Date:   Fri Nov 17 14:49:06 2023 +0100

    tree-optimization/112585 - new testcase

    The offending commit r14-5444-g5ea2965b499f9e was reverted.  The
    following adds a testcase.

            PR tree-optimization/112585
            * gcc.dg/torture/pr112585.c: New testcase.

[Bug tree-optimization/112585] [14 Regression] wrong code at -O3 on x86_64-linux-gnu since r14-5444

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=112585

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

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

--- Comment #5 from Richard Biener <rguenth at gcc dot gnu.org> ---
Fixed.  r14-5444-g5ea2965b499f9e reverted.