[Bug c/111472] New: Wrong code at -Os on x86_64-linux-gnu since

[Bug c/111472] New: Wrong code at -Os on x86_64-linux-gnu since

[shaohua.li at inf dot ethz.ch]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111472

            Bug ID: 111472
           Summary: Wrong code at -Os on x86_64-linux-gnu since
           Product: gcc
           Version: 14.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c
          Assignee: unassigned at gcc dot gnu.org
          Reporter: shaohua.li at inf dot ethz.ch
                CC: amacleod at redhat dot com
  Target Milestone: ---

gcc at -Os produced the wrong code.

Bisected to r11-4563-gd0d8b5d836

Compiler explorer: https://godbolt.org/z/EPznGbaje

With -fwrapv the issue goes away.

$ cat a.c
int printf(const char *, ...);
int a, b, e, f;
short c = -1, i;
char g;
int h[4][1] = {8, 5, 8, 5};
int main() {
  int *j = &e, *k = &f;
  unsigned short d = c;
  i = d >> b;
  *j = i;
  g = 0;
  for (; g <= 3; g++)
    *k = h[(unsigned char)(1 + e) + g][0];
  printf("%d\n", f);
}
$
$ gcc -O0 a.c && ./a.out
5
$ gcc -Os a.c && ./a.out
Segmentation fault
$
$ gcc -fsanitize=undefined a.c && ./a.out
5
$

[Bug tree-optimization/111472] Wrong code at -Os on x86_64-linux-gnu since r11-4563-gd0d8b5d836

[xry111 at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111472

Xi Ruoyao <xry111 at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|c                           |tree-optimization

--- Comment #1 from Xi Ruoyao <xry111 at gcc dot gnu.org> ---
Another ivopts issue.

Note that if something is different with different -O levels it's almost never
a "c" bug.  The "c" in component only covers C frontend which only contains
very straightforward optimizations.

[Bug tree-optimization/111472] Wrong code at -Os on x86_64-linux-gnu since r11-4563-gd0d8b5d836

[shaohua.li at inf dot ethz.ch]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111472

--- Comment #2 from Shaohua Li <shaohua.li at inf dot ethz.ch> ---
(In reply to Xi Ruoyao from comment #1)
> Another ivopts issue.
> 
> Note that if something is different with different -O levels it's almost
> never a "c" bug.  The "c" in component only covers C frontend which only
> contains very straightforward optimizations.

Thanks for the explanation!

[Bug tree-optimization/111472] [11/12/13/14 Regression] Wrong code at -Os on x86_64-linux-gnu since r11-4563-gd0d8b5d836

[xry111 at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111472

Xi Ruoyao <xry111 at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
     Ever confirmed|0                           |1
            Summary|Wrong code at -Os on        |[11/12/13/14 Regression]
                   |x86_64-linux-gnu since      |Wrong code at -Os on
                   |r11-4563-gd0d8b5d836        |x86_64-linux-gnu since
                   |                            |r11-4563-gd0d8b5d836
                 CC|                            |xry111 at gcc dot gnu.org
   Last reconfirmed|                            |2023-09-19
             Status|UNCONFIRMED                 |NEW

[Bug tree-optimization/111472] [11/12/13/14 Regression] Wrong code at -Os on x86_64-linux-gnu since r11-4563-gd0d8b5d836

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111472

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |wrong-code
   Target Milestone|---                         |11.5

[Bug tree-optimization/111472] [11/12/13/14 Regression] Wrong code at -Os on x86_64-linux-gnu since r11-4563-gd0d8b5d836

[amacleod at redhat dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111472

--- Comment #3 from Andrew Macleod <amacleod at redhat dot com> ---
I'm not sure that this didn't uncover something elsewhere, possibly ivopts
since that pass seems to be generating something different and I think there
were some fixes put in there on trunk?.

Regardless, this currently works on trunk. Can we run a regression on trunk and
see what patch fixed it? 

Andrew

[Bug tree-optimization/111472] [11/12/13/14 Regression] Wrong code at -Os on x86_64-linux-gnu since r11-4563-gd0d8b5d836

[amacleod at redhat dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111472

--- Comment #4 from Andrew Macleod <amacleod at redhat dot com> ---
(In reply to Andrew Macleod from comment #3)
> I'm not sure that this didn't uncover something elsewhere, possibly ivopts
> since that pass seems to be generating something different and I think there
> were some fixes put in there on trunk?.
> 
> Regardless, this currently works on trunk. Can we run a regression on trunk
> and see what patch fixed it? 
> 
> Andrew

And still fails on GCC13, so it appears to have been a trunk patch that fixed
it.

[Bug tree-optimization/111472] [11/12/13/14 Regression] Wrong code at -Os on x86_64-linux-gnu since r11-4563-gd0d8b5d836

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111472

--- Comment #5 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
(In reply to Andrew Macleod from comment #3)
> I'm not sure that this didn't uncover something elsewhere, possibly ivopts
> since that pass seems to be generating something different and I think there
> were some fixes put in there on trunk?.
> 
> Regardless, this currently works on trunk. Can we run a regression on trunk
> and see what patch fixed it? 

Most likely  r14-4789-g44e7e4498c (aka PR 110243).

[Bug tree-optimization/111472] [11/12/13 Regression] Wrong code at -Os on x86_64-linux-gnu since r11-4563-gd0d8b5d836

[law at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111472

Jeffrey A. Law <law at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |law at gcc dot gnu.org
            Summary|[11/12/13/14 Regression]    |[11/12/13 Regression] Wrong
                   |Wrong code at -Os on        |code at -Os on
                   |x86_64-linux-gnu since      |x86_64-linux-gnu since
                   |r11-4563-gd0d8b5d836        |r11-4563-gd0d8b5d836

--- Comment #6 from Jeffrey A. Law <law at gcc dot gnu.org> ---
Close.  It was the next revision my bisection landed on as the fix:

commit 9692309ed6b625f0fb358c0e230404b5603f69a6 (refs/bisect/bad)
Author: Richard Biener <rguenther@suse.de>
Date:   Fri Oct 20 15:08:49 2023 +0200

    tree-optimization/111445 - simple_iv simplification fault

    The following fixes a missed check in the simple_iv attempt
    to simplify (signed T)((unsigned T) base + step) where it
    allows a truncating inner conversion leading to wrong code.

            PR tree-optimization/111445
            * tree-scalar-evolution.cc (simple_iv_with_niters):
            Add missing check for a sign-conversion.

            * gcc.dg/torture/pr111445.c: New testcase.

[Bug tree-optimization/111472] [11 Regression] Wrong code at -Os on x86_64-linux-gnu since r11-4563-gd0d8b5d836

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111472

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
      Known to work|                            |12.3.1, 13.2.1
      Known to fail|                            |12.3.0, 13.2.0
           Assignee|unassigned at gcc dot gnu.org      |rguenth at gcc dot gnu.org
             Status|NEW                         |ASSIGNED
   Last reconfirmed|2023-09-19 00:00:00         |2024-5-8
           Priority|P3                          |P2
            Summary|[11/12/13 Regression] Wrong |[11 Regression] Wrong code
                   |code at -Os on              |at -Os on x86_64-linux-gnu
                   |x86_64-linux-gnu since      |since r11-4563-gd0d8b5d836
                   |r11-4563-gd0d8b5d836        |

--- Comment #7 from Richard Biener <rguenth at gcc dot gnu.org> ---
Re-confirmed on the 11 branch, fixed/avoided on newer branches by the backport
of referenced bug.