[Bug tree-optimization/103063] New: Wrong code while using -O3

[Bug tree-optimization/103063] New: Wrong code while using -O3

[stefansf at linux dot ibm.com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103063

            Bug ID: 103063
           Summary: Wrong code while using -O3
           Product: gcc
           Version: 12.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: tree-optimization
          Assignee: unassigned at gcc dot gnu.org
          Reporter: stefansf at linux dot ibm.com
  Target Milestone: ---

int a = 0;
unsigned char b = 0;

int main() {
  a - 6;
  for (; a >= -13; a = a - 8)
    while((unsigned char)(b-- * 6))
      ;
  if (b != 127)
    __builtin_abort();
  return 0;
}

Running the example while compiled with -O{0,1,2} works fine whereas it fails
with -O{3,fast} using gcc-12-4860-g73658e70d9e. Couldn't find a good commit so
far. Fails on IBM Z as well as x64.

Still not sure where it fails. ifcvt looks good to me.

[Bug tree-optimization/103063] Wrong code while using -O3

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103063

Martin Liška <marxin at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |amker at gcc dot gnu.org,
                   |                            |marxin at gcc dot gnu.org
   Last reconfirmed|                            |2021-11-03
           See Also|                            |https://gcc.gnu.org/bugzill
                   |                            |a/show_bug.cgi?id=100499
     Ever confirmed|0                           |1
             Status|UNCONFIRMED                 |NEW
           Keywords|                            |wrong-code

--- Comment #1 from Martin Liška <marxin at gcc dot gnu.org> ---
Started with r7-2373-g69b806f6a60efcf1.

[Bug tree-optimization/103063] [9/10/11/12 Regression] Wrong code while using -O3 since r7-2373-g69b806f6a60efcf1

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103063

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |9.5
            Summary|Wrong code while using -O3  |[9/10/11/12 Regression]
                   |since                       |Wrong code while using -O3
                   |r7-2373-g69b806f6a60efcf1   |since
                   |                            |r7-2373-g69b806f6a60efcf1

[Bug tree-optimization/103063] [9/10/11/12 Regression] Wrong code while using -O3 since r7-2373-g69b806f6a60efcf1

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103063

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P2
                 CC|                            |rguenth at gcc dot gnu.org

--- Comment #2 from Richard Biener <rguenth at gcc dot gnu.org> ---
-fdisable-tree-cunroll or -fno-tree-vectorize avoid the miscompile, the key is
probably one of

t.c:7:11: optimized: loop with 14 iterations completely unrolled (header
execution count 21807684)
t.c:4:5: optimized: loop with 7 iterations completely unrolled (header
execution count 29666503)

[Bug tree-optimization/103063] [9/10/11/12 Regression] Wrong code while using -O3 since r7-2373-g69b806f6a60efcf1

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103063

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> ---
Unless I'm missing something, the outer loop iterates with a = -6,
b is 0, so (unsigned char) (b-- * 6) is 0 (so the inner loop doesn't iterate
further) and b is decremented to 255.
Next a = a - 8 makes a = -14 out of that, so the outer loop doesn't iterate any
longer, so abort is what the test should do (unless changed to b != 255).

[Bug tree-optimization/103063] [9/10/11/12 Regression] Wrong code while using -O3 since r7-2373-g69b806f6a60efcf1

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103063

--- Comment #4 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Oops, no, the a - 6 is there just to confuse me, it does nothing and can be
safely dropped (I didn't like it before the loop, misread it and made it for (a
= -6; a >= -13; a = a - 8)).

[Bug tree-optimization/103063] [9/10/11/12 Regression] Wrong code while using -O3 since r7-2373-g69b806f6a60efcf1

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103063

--- Comment #5 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
ifcvt dump looks correct to me, in vect it seems the vectorizer thinks it can
compute the number of iterations of the inner loop using:
  _33 = b_lsm.8_5 * 6;
  _34 = _33 / 6;
  niters.11_32 = _34 + 1;
  _42 = b_lsm.8_5 * 6;
  if (_42 <= 89)

For b_lsm.8_5 == 0 niters.11_32 is correctly 1, but for b_lsm.8_5 == 255 it
gives
42 (i.e. (0x5fa & 0xff) / 6 + 1), while it should be 128, because the only
[0,255] values for which (v*6) & 0xff == 0 are 0 and 128.

[Bug tree-optimization/103063] [9/10/11/12 Regression] Wrong code while using -O3 since r7-2373-g69b806f6a60efcf1

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103063

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

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

--- Comment #6 from Richard Biener <rguenth at gcc dot gnu.org> ---
I'm going to look at this and the duplicates.  There have been patches proposed
by Jiufu Guo but I need to thoroughly look and understand the code and the
failure mode to make sense of them (and most of them complicated the code
further).

[Bug tree-optimization/103063] [9/10/11/12 Regression] Wrong code while using -O3 since r7-2373-g69b806f6a60efcf1

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103063

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

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

--- Comment #7 from Richard Biener <rguenth at gcc dot gnu.org> ---
Duplicate btw.

*** This bug has been marked as a duplicate of bug 100499 ***