[Bug tree-optimization/108392] New: Unexpected optimization in presence of '__attribute__((noipa))'

[Bug tree-optimization/108392] New: Unexpected optimization in presence of '__attribute__((noipa))'

[tschwinge at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108392

            Bug ID: 108392
           Summary: Unexpected optimization in presence of
                    '__attribute__((noipa))'
           Product: gcc
           Version: 13.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: tree-optimization
          Assignee: unassigned at gcc dot gnu.org
          Reporter: tschwinge at gcc dot gnu.org
  Target Milestone: ---

Consider:

    __attribute__((noipa))
    int f(int n)
    {
      if (n < 0)
        n = f(-n);
      return n;
    }

For x86_64-pc-linux-gnu at recent commit
de99049f6fe5341024d4d939ac50d063280f90db, for '-O2' I see:

    --- 1.c.044t.phiopt1   2023-01-13 12:26:20.196625109 +0100
    +++ 1.c.045t.tailr1    2023-01-13 12:26:20.196625109 +0100
    [...]
     __attribute__((noipa, noinline, noclone, no_icf))
     int f (int n)
     {
       int _1;

       <bb 2> :
    -  gimple_cond <lt_expr, n_4(D), 0, NULL, NULL>
    +  # gimple_phi <n_4, n_8(D)(0), _1(3)>
    +  gimple_cond <lt_expr, n_4, 0, NULL, NULL>
         goto <bb 3>; [INV]
       else
         goto <bb 4>; [INV]

       <bb 3> :
    -  gimple_assign <negate_expr, _1, n_4(D), NULL, NULL>
    -  gimple_call <f, n_7, _1>
    +  gimple_assign <negate_expr, _1, n_4, NULL, NULL>
    +  goto <bb 2>; [INV]

       <bb 4> :
    -  # gimple_phi <n_2, n_4(D)(2), n_7(3)>
    +  # gimple_phi <n_2, n_4(2)>
       gimple_return <n_2>

     }

I was assuming that '__attribute__((noipa))' would inhibit optimizations such
as replacing a 'gimple_call' by knowledge of the body of the callee?

(This caused confusion in a different test case, where I didn't expect the
recursive function call to disappear.)

[Bug tree-optimization/108392] Unexpected optimization in presence of '__attribute__((noipa))'

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108392

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|---                         |INVALID

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
No noipa does not turn off sibcalling optimization. Or turns off the local
transformation of tail call. 
Rather noipa turns off all of the ipa pass knowledge of that function.

Tail call is a local transformation is not considered ever an ipa based one.