[Bug c/111779] New: Fail to vectorize the struct include struct

[Bug c/111779] New: Fail to vectorize the struct include struct

[juzhe.zhong at rivai dot ai]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111779

            Bug ID: 111779
           Summary: Fail to vectorize the struct include struct
           Product: gcc
           Version: 14.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c
          Assignee: unassigned at gcc dot gnu.org
          Reporter: juzhe.zhong at rivai dot ai
  Target Milestone: ---

#include <stdbool.h>
#include <assert.h>
struct C
{
    int c;
    int d;
    bool f :1;
    float e;
};

struct A
{
  unsigned int a;
  unsigned char c1, c2;
  bool b1 : 1;
  bool b2 : 1;
  bool b3 : 1;
  struct C b4;
};

void
foo (const struct A * __restrict x, int y)
{
  int s = 0, i = 0;
  for (i = 0; i < y; ++i)
    {
      const struct A a = x[i];
      s += a.b4.f ? 1 : 0;
    }
  assert (s == 0);
    //__builtin_abort ();
}

int
main ()
{
  struct A x[100];
  int i;
  __builtin_memset (x, -1, sizeof (x));
  for (i = 0; i < 100; i++)
    {
      x[i].b1 = false;
      x[i].b2 = false;
      x[i].b3 = false;
      x[i].b4.f = false;
    }
  foo (x, 100);
  return 0;
}

https://godbolt.org/z/KWb7c1n5h

Both SVE GCC and RVV GCC failed to vectorize it.
But Clang succeed on vectorization.

[Bug tree-optimization/111779] Fail to vectorize the struct include struct

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111779

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|c                           |tree-optimization
             Status|UNCONFIRMED                 |NEW
     Ever confirmed|0                           |1
                 CC|                            |jamborm at gcc dot gnu.org
   Last reconfirmed|                            |2023-10-12

--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
The issue is the aggregate copy:

t.c:26:22: missed:  not vectorized: more than one data ref in stmt: a = *_3;

which SRA fails to scalarize:

  <bb 3> [local count: 955630224]:
  # s_23 = PHI <s_18(3), 0(2)>
  # i_25 = PHI <i_20(3), 0(2)>
  _1 = (long unsigned int) i_25;
  _2 = _1 * 24;
  _3 = x_16(D) + _2;
  a = *_3;
  _4 = BIT_FIELD_REF <a.b4, 8, 64>;
  _12 = _4 & 1;
  _6 = (int) _12;
  s_18 = _6 + s_23;
  a ={v} {CLOBBER(eol)};
  i_20 = i_25 + 1;
  if (y_14(D) > i_20)

Candidate (2778): a
...
! Disqualifying a - No scalar replacements to be created.

the BIT_FIELD_REF is already created by early folding in
optimize_bit_field_compare folding (int) a.b4.f != 0

    s = ((int) NON_LVALUE_EXPR <BIT_FIELD_REF <a.b4, 8, 64>> & 1) + s;

SRA could handle BIT_FIELD_REFs just fine - esp. quantities with
a byte size.  And then this folding is just premature...

Removing the folding that handles BF != CST fixes it.  I know removing
all of it, esp. BF != BF will regress some stuff.  I'll put this half-way
patch through testing.

diff --git a/gcc/fold-const.cc b/gcc/fold-const.cc
index 82299bb7f1d..3db383360d6 100644
--- a/gcc/fold-const.cc
+++ b/gcc/fold-const.cc
@@ -4695,7 +4695,7 @@ optimize_bit_field_compare (location_t loc, enum
tree_code code,
     return 0;

   if (const_p)
-    rreversep = lreversep;
+    return 0;
   else
    {
      /* If this is not a constant, we can only do something if bit positions,

[Bug tree-optimization/111779] Fail to vectorize the struct include struct

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111779

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           See Also|                            |https://gcc.gnu.org/bugzill
                   |                            |a/show_bug.cgi?id=108070

--- Comment #2 from Richard Biener <rguenth at gcc dot gnu.org> ---
Ah, did that already, it regressed, I filed PR108070 for it.

[Bug tree-optimization/111779] Fail to vectorize the struct include struct

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111779

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 #3 from Richard Biener <rguenth at gcc dot gnu.org> ---
I posted a patch for SRA.

[Bug tree-optimization/111779] Fail to vectorize the struct include struct

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111779

--- Comment #4 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:6decda1a35be5764101987c210b5693a0d914e58

commit r14-4612-g6decda1a35be5764101987c210b5693a0d914e58
Author: Richard Biener <rguenther@suse.de>
Date:   Thu Oct 12 11:34:57 2023 +0200

    tree-optimization/111779 - Handle some BIT_FIELD_REFs in SRA

    The following handles byte-aligned, power-of-two and byte-multiple
    sized BIT_FIELD_REF reads in SRA.  In particular this should cover
    BIT_FIELD_REFs created by optimize_bit_field_compare.

    For gcc.dg/tree-ssa/ssa-dse-26.c we now SRA the BIT_FIELD_REF
    appearing there leading to more DSE, fully eliding the aggregates.

    This results in the same false positive -Wuninitialized as the
    older attempt to remove the folding from optimize_bit_field_compare,
    fixed by initializing part of the aggregate unconditionally.

            PR tree-optimization/111779
    gcc/
            * tree-sra.cc (sra_handled_bf_read_p): New function.
            (build_access_from_expr_1): Handle some BIT_FIELD_REFs.
            (sra_modify_expr): Likewise.
            (make_fancy_name_1): Skip over BIT_FIELD_REF.

    gcc/fortran/
            * trans-expr.cc (gfc_trans_assignment_1): Initialize
            lhs_caf_attr and rhs_caf_attr codimension flag to avoid
            false positive -Wuninitialized.

    gcc/testsuite/
            * gcc.dg/tree-ssa/ssa-dse-26.c: Adjust for more DSE.
            * gcc.dg/vect/vect-pr111779.c: New testcase.

[Bug tree-optimization/111779] Fail to vectorize the struct include struct

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111779

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

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

--- Comment #5 from Richard Biener <rguenth at gcc dot gnu.org> ---
This should be fixed now.