[Bug tree-optimization/99473] New: redundant conditional zero-initialization not eliminated

[Bug tree-optimization/99473] New: redundant conditional zero-initialization not eliminated

[msebor at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99473

            Bug ID: 99473
           Summary: redundant conditional zero-initialization not
                    eliminated
           Product: gcc
           Version: 11.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: tree-optimization
          Assignee: unassigned at gcc dot gnu.org
          Reporter: msebor at gcc dot gnu.org
  Target Milestone: ---

All three functions below should result in equivalently optimal code (and,
ideally IL) but only g1() and g3() result in the same assembly, and only g1()
result in optimal, branchless GIMPLE.

$ cat x.c && gcc -O2 -S -Wall -o/dev/stdout x.c
void f (int*);

void g1 (int i)
{
  int x;
  if (i)
    x = i;
  else
    x = 0;
  f (&x);
}


void g2 (int i)
{
  int x;
  if (i) {
    x = i;
    f (&x);
  }
  else {
    x = 0;
    f (&x);
  }
}

void g3 (int i)
{
  int x = 0;
  if (i)
    x = i;
  f (&x);
}
        .file   "x.c"
        .text
        .p2align 4
        .globl  g1
        .type   g1, @function
g1:
.LFB0:
        .cfi_startproc
        subq    $24, %rsp
        .cfi_def_cfa_offset 32
        movl    %edi, 12(%rsp)
        leaq    12(%rsp), %rdi
        call    f
        addq    $24, %rsp
        .cfi_def_cfa_offset 8
        ret
        .cfi_endproc
.LFE0:
        .size   g1, .-g1
        .p2align 4
        .globl  g2
        .type   g2, @function
g2:
.LFB1:
        .cfi_startproc
        subq    $24, %rsp
        .cfi_def_cfa_offset 32
        testl   %edi, %edi
        je      .L5
        movl    %edi, 12(%rsp)
        leaq    12(%rsp), %rdi
        call    f
        addq    $24, %rsp
        .cfi_remember_state
        .cfi_def_cfa_offset 8
        ret
        .p2align 4,,10
        .p2align 3
.L5:
        .cfi_restore_state
        leaq    12(%rsp), %rdi
        movl    $0, 12(%rsp)
        call    f
        addq    $24, %rsp
        .cfi_def_cfa_offset 8
        ret
        .cfi_endproc
.LFE1:
        .size   g2, .-g2
        .p2align 4
        .globl  g3
        .type   g3, @function
g3:
.LFB2:
        .cfi_startproc
        subq    $24, %rsp
        .cfi_def_cfa_offset 32
        movl    %edi, 12(%rsp)
        leaq    12(%rsp), %rdi
        call    f
        addq    $24, %rsp
        .cfi_def_cfa_offset 8
        ret
        .cfi_endproc
.LFE2:
        .size   g3, .-g3
        .ident  "GCC: (GNU) 11.0.1 20210308 (experimental)"
        .section        .note.GNU-stack,"",@progbits

[Bug tree-optimization/99473] redundant conditional zero-initialization not eliminated

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99473

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever confirmed|0                           |1
   Last reconfirmed|                            |2021-03-08
           Severity|normal                      |enhancement

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Confirmed, I think there are other bugs dealing with this same thing.
g3 should have been caught by cstore but I don't see why it is not.
Also g2 needs commoning by sinking rather commoning by pulling.

[Bug tree-optimization/99473] redundant conditional zero-initialization not eliminated

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99473

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 #2 from Richard Biener <rguenth at gcc dot gnu.org> ---
g3 and g1 behave differently also because of (there's a dup PR I can't find
right now) sinking happening in a way that the pass store-commoning code
doesn't trigger on the sunk store.

cselim doesn't trigger because

  if ((TREE_CODE (lhs) != MEM_REF
       && TREE_CODE (lhs) != ARRAY_REF
       && TREE_CODE (lhs) != COMPONENT_REF)
      || !is_gimple_reg_type (TREE_TYPE (lhs)))
    return false;

lhs is a VAR_DECL and 'nontrap' only tracks pointers.  There's code to actually
handle auto-vars now but the above still disallows bare decls.  Because
we have the address-taken the transform will also require
-fallow-store-data-races.

diff --git a/gcc/tree-ssa-phiopt.c b/gcc/tree-ssa-phiopt.c
index ddd9d531b13..6f7efa29a1b 100644
--- a/gcc/tree-ssa-phiopt.c
+++ b/gcc/tree-ssa-phiopt.c
@@ -2490,9 +2490,8 @@ cond_store_replacement (basic_block middle_bb,
basic_block join_bb,
   locus = gimple_location (assign);
   lhs = gimple_assign_lhs (assign);
   rhs = gimple_assign_rhs1 (assign);
-  if ((TREE_CODE (lhs) != MEM_REF
-       && TREE_CODE (lhs) != ARRAY_REF
-       && TREE_CODE (lhs) != COMPONENT_REF)
+  if ((!REFERENCE_CLASS_P (lhs)
+       && !DECL_P (lhs))
       || !is_gimple_reg_type (TREE_TYPE (lhs)))
     return false;


fixes g3 this (with -fallow-store-data-races).  Queued for GCC 12.

g2 needs sinking/commoning of f (&x) for which there's yet another PR I think.

[Bug tree-optimization/99473] redundant conditional zero-initialization not eliminated

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99473

--- Comment #3 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:577d05fc914338cd7ddc254f3bee4532331f5c13

commit r12-114-g577d05fc914338cd7ddc254f3bee4532331f5c13
Author: Richard Biener <rguenther@suse.de>
Date:   Tue Mar 9 09:29:29 2021 +0100

    tree-optimization/99473 - more cselim

    This fixes the pre-condition on cselim to include all references
    and decls when they end up as auto-var.

    Bootstrapped/tested on x86_64-linux

    2021-03-09  Richard Biener  <rguenther@suse.de>

            PR tree-optimization/99473
            * tree-ssa-phiopt.c (cond_store_replacement): Handle all
            stores.

            * gcc.dg/tree-ssa/pr99473-1.c: New testcase.

[Bug tree-optimization/99473] redundant conditional zero-initialization not eliminated

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99473

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

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

--- Comment #4 from Richard Biener <rguenth at gcc dot gnu.org> ---
The missed cselim is now fixed.  The other cases are dups, but I don't dare
finding them right now ...

[Bug tree-optimization/99473] redundant conditional zero-initialization not eliminated

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99473

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Depends on|                            |51964

--- Comment #5 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
g2 is PR 51964.


Referenced Bugs:

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=51964
[Bug 51964] Missed tail merging opportunity