[Bug tree-optimization/103257] New: [12 Regression] Dead Code Elimination Regression at -O3 (trunk vs 11.2.0)

[Bug tree-optimization/103257] New: [12 Regression] Dead Code Elimination Regression at -O3 (trunk vs 11.2.0)

[theodort at inf dot ethz.ch]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103257

            Bug ID: 103257
           Summary: [12 Regression] Dead Code Elimination Regression at
                    -O3 (trunk vs 11.2.0)
           Product: gcc
           Version: 12.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: tree-optimization
          Assignee: unassigned at gcc dot gnu.org
          Reporter: theodort at inf dot ethz.ch
  Target Milestone: ---

cat case.c

void foo(void);
unsigned b, c;
static short a(short e, short f) { return e * f; }
int main() {
  if (a(1  ^ ((0, 0) ^ 1 && b) <= b, c))
    foo();
  c = 0;
}

11.2.0 at -O3 can eliminate the call to foo but trunk at -O3 cannot:

gcc11  case.c -S -O3 -o /dev/stdout
main:
.LFB1:
        .cfi_startproc
        movl    b(%rip), %eax
        xorl    %edx, %edx
        testl   %eax, %eax
        setne   %dl
        cmpl    %edx, %eax
        movl    $0, %eax
        cmovb   c(%rip), %ax
        testw   %ax, %ax
        jne     .L11
        movl    $0, c(%rip)
        xorl    %eax, %eax
        ret
.L11:
        pushq   %rax
        .cfi_def_cfa_offset 16
        call    foo
        xorl    %edx, %edx
        xorl    %eax, %eax
        movl    %edx, c(%rip)
        popq    %rcx
        .cfi_def_cfa_offset 8
        ret
        .cfi_endproc


gcc-trunk -v
Target: x86_64-pc-linux-gnu
Thread model: posix
Supported LTO compression algorithms: zlib zstd
gcc version 12.0.0 20211115 (experimental) (GCC)

It started with
https://gcc.gnu.org/git/?p=gcc.git;a=commit;h=0288527f47cec6698b31ccb3210816415506009e

[Bug tree-optimization/103257] [12 Regression] Dead Code Elimination Regression at -O3 (trunk vs 11.2.0)

[aldyh at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103257

Aldy Hernandez <aldyh at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |amacleod at redhat dot com

--- Comment #1 from Aldy Hernandez <aldyh at gcc dot gnu.org> ---
Looks like something either VRP or the threader could get.

By vrp2 we have:

=========== BB 2 ============
Imports: c.0_1  
Exports: c.0_1  f.6_9  _10  
         f.6_9 : c.0_1(I)  
         _10 : c.0_1(I)  f.6_9  
    <bb 2> [local count: 1073741824]:
    c.0_1 = c;
    b.2_2 = b;
    _11 = b.2_2 != 0;
    iftmp.3_3 = (unsigned int) _11;
    f.6_9 = (unsigned short) c.0_1;
    _10 = b.2_2 < iftmp.3_3 ? f.6_9 : 0;
    if (_10 != 0)
      goto <bb 3>; [33.00%]
    else
      goto <bb 4>; [67.00%]

iftmp.3_3 : unsigned int [0, 1]
2->3  (T) _10 :         unsigned short [1, +INF]
2->4  (F) _10 :         unsigned short [0, 0]

=========== BB 3 ============
    <bb 3> [local count: 354334800]:
    foo ();

[Bug tree-optimization/103257] [12 Regression] Dead Code Elimination Regression at -O3 (trunk vs 11.2.0)

[amacleod at redhat dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103257

--- Comment #2 from Andrew Macleod <amacleod at redhat dot com> ---
At issue is GORI does not yet process COND_EXPRS, so we cannot do any
evaluations to refine ranges on outgoing edges. It is on the list :-)

[Bug tree-optimization/103257] [12 Regression] Dead Code Elimination Regression at -O3 (trunk vs 11.2.0)

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103257

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |12.0
           Keywords|                            |missed-optimization

[Bug tree-optimization/103257] [9/10/11/12 Regression] Dead Code Elimination Regression at -O3 (trunk vs 11.2.0)

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103257

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|                            |2021-11-16
      Known to work|                            |5.1.0, 7.2.0
      Known to fail|                            |10.1.0, 8.1.0
   Target Milestone|12.0                        |9.5
            Summary|[12 Regression] Dead Code   |[9/10/11/12 Regression]
                   |Elimination Regression at   |Dead Code Elimination
                   |-O3 (trunk vs 11.2.0)       |Regression at -O3 (trunk vs
                   |                            |11.2.0)
             Status|UNCONFIRMED                 |NEW
     Ever confirmed|0                           |1

--- Comment #3 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Note this used to be optimized at the gimple level for GCC 7 and before. So
marking it as a regression from that point forward.

[Bug tree-optimization/103257] [9/10/11/12 Regression] Dead Code Elimination Regression at -O3 (trunk vs 11.2.0)

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103257

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |pinskia at gcc dot gnu.org

--- Comment #4 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
r8-2090 caused the issue to show up. Maybe this pattern should be disabled
until canonicalize_math_p () is false.

[Bug tree-optimization/103257] [9/10/11/12 Regression] Dead Code Elimination Regression at -O3 (trunk vs 11.2.0)

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103257

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> ---
(In reply to Andrew Pinski from comment #4)
> r8-2090 caused the issue to show up. Maybe this pattern should be disabled
> until canonicalize_math_p () is false.

I am going to implement that.

[Bug tree-optimization/103257] [9/10/11/12 Regression] Dead Code Elimination Regression at -O3 (trunk vs 11.2.0)

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103257

--- Comment #6 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Created attachment 51804
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=51804&action=edit
untested Patch

This is the patch which I am testing. I checked all of tree-ssa.exp and there
are no failures. I still need to do a full bootstrap/test.

[Bug tree-optimization/103257] [9/10/11/12 Regression] Dead Code Elimination Regression at -O3 (trunk vs 11.2.0)

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103257

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P2

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

[Bug tree-optimization/103257] [9/10/11/12 Regression] Dead Code Elimination Regression at -O3 (trunk vs 11.2.0)

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103257

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |patch
                URL|                            |https://gcc.gnu.org/piperma
                   |                            |il/gcc-patches/2021-Novembe
                   |                            |r/584677.html

--- Comment #8 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Patch submitted:
https://gcc.gnu.org/pipermail/gcc-patches/2021-November/584677.html

[Bug tree-optimization/103257] [9/10/11/12 Regression] Dead Code Elimination Regression at -O3 (trunk vs 11.2.0)

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103257

--- Comment #9 from CVS 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:527e54a431473cc497204226a21f2831d2375e66

commit r12-5392-g527e54a431473cc497204226a21f2831d2375e66
Author: Andrew Pinski <apinski@marvell.com>
Date:   Tue Nov 16 04:46:21 2021 +0000

    Fix tree-optimization/103257: Missed jump threading due too early
conversion of bool*A into bool?A:0

    So like many optimizations on the gimple level, sometimes it makes sense to
do the
    optimization early or later. In this case, creating a cond expression early
causes
    other optimizations to be missed.  So just disable it until
canonicalize_math_p ()
    is false.

    OK? Bootstrapped and tested on x86_64-linux-gnu with no regressions.

            PR tree-optimization/103257

    gcc/ChangeLog:

            * match.pd
            ((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
            Disable until !canonicalize_math_p ().

    gcc/testsuite/ChangeLog:

            * gcc.dg/tree-ssa/vrp116.c: Check optimized instead of vrp1.
            * gcc.dg/tree-ssa/pr103257-1.c: New test.

[Bug tree-optimization/103257] [9/10/11/12 Regression] Dead Code Elimination Regression at -O3 (trunk vs 11.2.0)

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103257

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|---                         |FIXED
             Status|ASSIGNED                    |RESOLVED
   Target Milestone|9.5                         |12.0

--- Comment #10 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Fixed on the trunk, not going to backport for GCC 9 or 10 since it is a small
missed optimization, not found until now.