public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
* [Bug c++/52067] New: force sibling call optimization
@ 2012-01-31 14:51 trashyankes at wp dot pl
  2012-01-31 15:06 ` [Bug c++/52067] " rguenth at gcc dot gnu.org
  2012-01-31 18:28 ` trashyankes at wp dot pl
  0 siblings, 2 replies; 3+ messages in thread
From: trashyankes at wp dot pl @ 2012-01-31 14:51 UTC (permalink / raw)
  To: gcc-bugs

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52067

             Bug #: 52067
           Summary: force sibling call optimization
    Classification: Unclassified
           Product: gcc
           Version: 4.7.0
            Status: UNCONFIRMED
          Severity: enhancement
          Priority: P3
         Component: c++
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: trashyankes@wp.pl


code:
-------------------------------
-------------------------------
template<typename T>
int loop(int a) __attribute__((optimize(3)));

template<typename T>
int loop(int a)
{
    T zz;
    a -= 1;
    if(a>0)
        return loop<T>(a);
    return a;
}

class pod
{
};
class not_pod
{
public:
    not_pod() {}
};

void main()
{
    loop<pod>(10); //use tail call
    loop<not_pod>(10); //dont use tail call
}
-------------------------------
-------------------------------

I want use sibling call optimization and build around this my system.
But because C++ types can break this optimization without any waring, I could
get unexpected stack overflow.

It would be good if GCC have attribute that force function to use sibling call
and if that is impossible, break compilation with error.

something like this:
-------------------------------
-------------------------------
int foo(int a) __attribute__((siblingcall));
int foo(int a)
{
    std::string b = "abc"; //Error, have destructor
    {
        std::string b2 = "zzz"; //OK, destroyed before any tail call
    }
    if(a==1)
        return foo(a+1); //OK
    if(a==2)
        return 1+foo(a+1); //Error, except tail call
    if(a==3)
    {
        int c = foo(a+1); //OK, normal call
        return c;
    }
    return 0; //OK
}
-------------------------------
-------------------------------


^ permalink raw reply	[flat|nested] 3+ messages in thread

* [Bug c++/52067] force sibling call optimization
  2012-01-31 14:51 [Bug c++/52067] New: force sibling call optimization trashyankes at wp dot pl
@ 2012-01-31 15:06 ` rguenth at gcc dot gnu.org
  2012-01-31 18:28 ` trashyankes at wp dot pl
  1 sibling, 0 replies; 3+ messages in thread
From: rguenth at gcc dot gnu.org @ 2012-01-31 15:06 UTC (permalink / raw)
  To: gcc-bugs

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52067

--- Comment #1 from Richard Guenther <rguenth at gcc dot gnu.org> 2012-01-31 14:23:50 UTC ---
I suppose you would want to annotate a specific call, not necessarily _all_
calls to loop (those from main).


^ permalink raw reply	[flat|nested] 3+ messages in thread

* [Bug c++/52067] force sibling call optimization
  2012-01-31 14:51 [Bug c++/52067] New: force sibling call optimization trashyankes at wp dot pl
  2012-01-31 15:06 ` [Bug c++/52067] " rguenth at gcc dot gnu.org
@ 2012-01-31 18:28 ` trashyankes at wp dot pl
  1 sibling, 0 replies; 3+ messages in thread
From: trashyankes at wp dot pl @ 2012-01-31 18:28 UTC (permalink / raw)
  To: gcc-bugs

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52067

--- Comment #2 from trashyankes at wp dot pl 2012-01-31 17:04:10 UTC ---
(In reply to comment #1)
> I suppose you would want to annotate a specific call, not necessarily _all_
> calls to loop (those from main).

Im interested in calls from `loop<T>(int)`. I want guarantee that all
`return something_else();` calls in function to be sibling calls.


^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2012-01-31 17:04 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2012-01-31 14:51 [Bug c++/52067] New: force sibling call optimization trashyankes at wp dot pl
2012-01-31 15:06 ` [Bug c++/52067] " rguenth at gcc dot gnu.org
2012-01-31 18:28 ` trashyankes at wp dot pl

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).