[Bug tree-optimization/110941] New: [14 Regression] Dead Code Elimination Regression at -O3 since r14-2379-gc496d15954c

[Bug tree-optimization/110941] New: [14 Regression] Dead Code Elimination Regression at -O3 since r14-2379-gc496d15954c

[scherrer.sv at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110941

            Bug ID: 110941
           Summary: [14 Regression] Dead Code Elimination Regression at
                    -O3 since r14-2379-gc496d15954c
           Product: gcc
           Version: 14.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: tree-optimization
          Assignee: unassigned at gcc dot gnu.org
          Reporter: scherrer.sv at gmail dot com
  Target Milestone: ---

static int a;
void foo(void);
void bar349_(void);
void bar363_(void);
void bar275_(void);
int main() {
  {
    {
      short b = 26;
      for (; b >= 1; b = b - 4) {
        if (b >= 2 && b <= 26)
          bar275_();
        if (a)
          bar363_();
        if (a)
          bar349_();
        int c = b;
        if (!(c >= 2 && c <= 26))
          foo();
      }
    }
    a = 0;
  }
}

gcc-25c4b1620eb (trunk) -O3 cannot eliminate the call to foo but
gcc-releases/gcc-13.1.0 -O3 can.
-----------------------------------------------------------------------
gcc-25c4b1620ebc10fceabd86a34fdbbaf8037e7e82 -O3 case.c -S -o case.s
--------- OUTPUT ---------
main:
.LFB0:
        .cfi_startproc
        pushq   %rbx
        .cfi_def_cfa_offset 16
        .cfi_offset 3, -16
        movl    $24, %ebx
        .p2align 4,,10
        .p2align 3
.L8:
        cmpw    $24, %bx
        jbe     .L2
        movl    a(%rip), %edx
        testl   %edx, %edx
        jne     .L3
.L7:
        call    foo
        xorl    %eax, %eax
        popq    %rbx
        .cfi_remember_state
        .cfi_def_cfa_offset 8
        movl    $0, a(%rip)
        ret
        .p2align 4,,10
        .p2align 3
.L3:
        .cfi_restore_state
        call    bar363_
        movl    a(%rip), %eax
        testl   %eax, %eax
        je      .L7
        call    bar349_
        jmp     .L7
.L2:
        call    bar275_
        cmpl    $0, a(%rip)
        jne     .L15
        .p2align 4,,10
        .p2align 3
.L6:
        subl    $4, %ebx
        cmpl    $-4, %ebx
        jne     .L8
        movl    $0, a(%rip)
        xorl    %eax, %eax
        popq    %rbx
        .cfi_remember_state
        .cfi_def_cfa_offset 8
        ret
.L15:
        .cfi_restore_state
        call    bar363_
        cmpl    $0, a(%rip)
        je      .L6
        call    bar349_
        cmpl    $24, %ebx
        jbe     .L6
        jmp     .L7
---------- END OUTPUT ---------

-----------------------------------------------------------------------
gcc-2b98cc24d6af0432a74f6dad1c722ce21c1f7458 -O3 case.c -S -o case.s
--------- OUTPUT ---------
main:
.LFB0:
        .cfi_startproc
        pushq   %rbx
        .cfi_def_cfa_offset 16
        .cfi_offset 3, -16
        movl    $24, %ebx
.L10:
        cmpw    $24, %bx
        jbe     .L2
        movl    a(%rip), %edx
        testl   %edx, %edx
        jne     .L17
.L3:
        movl    $0, a(%rip)
        xorl    %eax, %eax
        popq    %rbx
        .cfi_remember_state
        .cfi_def_cfa_offset 8
        ret
        .p2align 4,,10
        .p2align 3
.L17:
        .cfi_restore_state
        call    bar363_
        movl    a(%rip), %eax
        testl   %eax, %eax
        jne     .L4
.L15:
        subl    $4, %ebx
        jmp     .L10
        .p2align 4,,10
        .p2align 3
.L4:
        call    bar349_
        jmp     .L15
.L2:
        call    bar275_
        cmpl    $0, a(%rip)
        jne     .L18
        .p2align 4,,10
        .p2align 3
.L9:
        subl    $4, %ebx
        cmpw    $-4, %bx
        jne     .L10
        jmp     .L3
.L18:
        call    bar363_
        cmpl    $0, a(%rip)
        je      .L9
        call    bar349_
        jmp     .L9
---------- END OUTPUT ---------

-----------------------------------------------------------------------
Bisects to r14-2379-gc496d15954c

[Bug tree-optimization/110941] [14 Regression] Dead Code Elimination Regression at -O3 since r14-2379-gc496d15954c

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110941

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |14.0

[Bug tree-optimization/110941] [14 Regression] Dead Code Elimination Regression at -O3 since r14-2379-gc496d15954c

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110941

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|                            |2023-08-08
     Ever confirmed|0                           |1
             Status|UNCONFIRMED                 |NEW

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
  # RANGE [irange] unsigned int [0, 16][20, 20][24, 24][65532,
4294901776][4294967293, +INF] MASK 0xfffffffc VALUE 0x0
  # ivtmp.18_37 = PHI <ivtmp.18_38(8), 24(2)>
  # RANGE [irange] unsigned short [0, +INF] MASK 0xfffc VALUE 0xfffc
  _21 = (unsigned short) ivtmp.18_37;
  if (_21 <= 24)


Confirmed. The range for _21 is way to conserative.
It should have been: `[0, 16][20, 20][24, 24]`.

[Bug tree-optimization/110941] [14 Regression] Dead Code Elimination Regression at -O3 since r14-2379-gc496d15954c

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110941

--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
GCC 13:
Global Exported: _6 = [irange] unsigned int [0, 24] NONZERO 0x1e

trunk:
Global Exported: _6 = [irange] unsigned int [0, 24][+INF, +INF] MASK 0x1c VALUE
0x0


And then:
Folding predicate _6 > 24 to 0

does not happen.
But what is interesting is with
https://gcc.gnu.org/pipermail/gcc-patches/2023-September/629128.html ,
We get:
```
Simplified relational if (_6 > 24)
 into if (_6 == 4294967295)

Folded into: if (_6 == 4294967295)
```

And then during cfgcleanup the match pattern:
/* X == C (or X & Z == Y | C) is impossible if ~nonzero(X) & C != 0.  */
Hits which is totally funny because that means the range and nonzero
information are out of sync ...

[Bug tree-optimization/110941] [14 Regression] Dead Code Elimination Regression at -O3 since r14-2379-gc496d15954c

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110941

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |needs-bisection

--- Comment #3 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
This seems to be fixed now. We are able to optimize out the call to foo even in
evrp now too.

[Bug tree-optimization/110941] [14 Regression] Dead Code Elimination Regression at -O3 since r14-2379-gc496d15954c

[jamborm at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110941

Martin Jambor <jamborm at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jamborm at gcc dot gnu.org
           Keywords|needs-bisection             |

--- Comment #4 from Martin Jambor <jamborm at gcc dot gnu.org> ---
This has been fixed with r14-5109-ga291237b628f41 (Andrew MacLeod: Remove
simple ranges from trailing zero bitmasks).

[Bug tree-optimization/110941] [14 Regression] Dead Code Elimination Regression at -O3 since r14-2379-gc496d15954c

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110941

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

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

--- Comment #5 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Will commit a testcase for this.
r14-5109-ga291237b628f41 is definitely the fix for this based on my analysis in
comment #2:
> GCC 13:
> Global Exported: _6 = [irange] unsigned int [0, 24] NONZERO 0x1e

> trunk:
> Global Exported: _6 = [irange] unsigned int [0, 24][+INF, +INF] MASK 0x1c VALUE 0x0

The removal of `[+INF, +INF]` here fixes the issue earlier

[Bug tree-optimization/110941] [14 Regression] Dead Code Elimination Regression at -O3 since r14-2379-gc496d15954c

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110941

--- Comment #6 from GCC Commits <cvs-commit at gcc dot gnu.org> ---
The trunk branch has been updated by Andrew Pinski <pinskia@gcc.gnu.org>:

https://gcc.gnu.org/g:34a827039fabcf24ce78da25984a1cc8be7ca2c3

commit r14-7221-g34a827039fabcf24ce78da25984a1cc8be7ca2c3
Author: Andrew Pinski <quic_apinski@quicinc.com>
Date:   Fri Jan 12 20:24:34 2024 -0800

    Add a few testcases for fix missed optimization regressions

    Adds a few new testcases for some missed optimization regressions.
    The analysis on how each should be optimized is in the testcases
    themselves (and in the bug report).

    Committed as obvious after running the testsuite to make sure they pass.

            PR tree-optimization/107823
            PR tree-optimization/110768
            PR tree-optimization/110941
            PR tree-optimization/110450
            PR tree-optimization/110841

    gcc/testsuite/ChangeLog:

            * gcc.dg/tree-ssa/ssa-thread-22.c: New test.
            * gcc.dg/tree-ssa/vrp-loop-1.c: New test.
            * gcc.dg/tree-ssa/vrp-loop-2.c: New test.
            * gcc.dg/tree-ssa/vrp-unreachable-1.c: New test.
            * gcc.dg/tree-ssa/vrp-unreachable-2.c: New test.

    Signed-off-by: Andrew Pinski <quic_apinski@quicinc.com>

[Bug tree-optimization/110941] [14 Regression] Dead Code Elimination Regression at -O3 since r14-2379-gc496d15954c

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110941

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

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

--- Comment #7 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Fixed.