[Bug c++/53875] New: calls to const functions are eliminated at -O0

[Bug c++/53875] New: calls to const functions are eliminated at -O0

[James.H.McKim at nasa dot gov]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53875

             Bug #: 53875
           Summary: calls to const functions are eliminated at -O0
    Classification: Unclassified
           Product: gcc
           Version: 4.6.3
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: James.H.McKim@nasa.gov


Compiler version 4.6.3.
System: ubuntu 12.1

When source containing a call to a const function is compiled, that function's
invocation is always (regardless of -O level) eliminated if the returned result
is discarded. That action is an optimization, and when no optimization is
specified, having it happen is surprising.

If the compiler is intended to always eliminate, it would be helpful if there
was a diagnostic indicating the elimination. For instance, instead of the
warning "statement has no effect", maybe "statement has no effect,
eliminating".


The code was compiled with the command:

g++ -O0 -S -o t1b.lst t1b.cxx


The example code (t1b.cxx):

extern int pthread_self (void) __attribute__ ((__const__));

int main() {

  // Demonstrate the compiler doesn't emit this call.
  pthread_self();

  // A workaround.
  // volatile int x = pthread_self();

  return 0;
}

[Bug c++/53875] calls to const functions are eliminated at -O0

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53875

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2012-07-09
     Ever Confirmed|0                           |1

--- Comment #1 from Richard Guenther <rguenth at gcc dot gnu.org> 2012-07-09 08:42:26 UTC ---
Confirmed.  We generally do very many "small" optimizations at -O0 that would
surprise users.  we probably should not.

[Bug middle-end/53875] calls to const functions are eliminated at -O0

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=53875

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|c++                         |middle-end

--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
This even happens in C.
It happens during expand phase:
;; Generating RTL for gimple basic block 2

;; pthread_self ();

(nil)

Maybe at -O0 or -Og, during expand we still expand the call ...

[Bug middle-end/53875] calls to const functions are eliminated at -O0

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=53875

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
      Known to work|                            |3.3.3
      Known to fail|                            |3.4.0

--- Comment #3 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
The code which introduced this was g:8c6a82695b85f8ed74cdc67f2cf74c5a62d0d91d .

[Bug middle-end/53875] calls to const functions are eliminated at -O0

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=53875

--- Comment #4 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
(In reply to Andrew Pinski from comment #3)
> The code which introduced this was
> g:8c6a82695b85f8ed74cdc67f2cf74c5a62d0d91d .

https://gcc.gnu.org/pipermail/gcc-patches/2003-May/104797.html

[Bug middle-end/53875] calls to const functions are eliminated at -O0

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=53875

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |roger at nextmovesoftware dot com

--- Comment #5 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
diff --git a/gcc/calls.c b/gcc/calls.c
index 27b59f26ad3..f23dde58671 100644
--- a/gcc/calls.c
+++ b/gcc/calls.c
@@ -2735,7 +2735,8 @@ expand_call (tree exp, rtx target, int ignore)
       && (!(flags & ECF_LOOPING_CONST_OR_PURE))
       && (flags & ECF_NOTHROW)
       && (ignore || target == const0_rtx
-         || TYPE_MODE (rettype) == VOIDmode))
+         || TYPE_MODE (rettype) == VOIDmode)
+      && !optimize)
     {
       bool volatilep = false;
       tree arg;

[Bug middle-end/53875] calls to const functions are eliminated at -O0

[roger at nextmovesoftware dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=53875

--- Comment #6 from Roger Sayle <roger at nextmovesoftware dot com> ---
I believe this should be "... && optimize != 0", i.e. we want to eliminate
these calls (only) when optimizing.  But this seems like a reasonable
fix/feature request (to identify breakpoints at -O0).