[Bug c++/61213] New: std::forward_as_tuple and std::tuple<T&&> and any rvalue loose data

[Bug c++/61213] New: std::forward_as_tuple and std::tuple<T&&> and any rvalue loose data

[tower120 at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61213

            Bug ID: 61213
           Summary: std::forward_as_tuple and std::tuple<T&&> and any
                    rvalue loose data
           Product: gcc
           Version: 4.8.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: tower120 at gmail dot com

std::forward_as_tuple elements loose data on -O2.
Works fine with -O0


Consider the following code:
http://coliru.stacked-crooked.com/a/67499e052283bd5a

#include <iostream>
#include <tuple>

class Widget {
public:
  Widget(int h) : h(h){}
  int h;
};

template<typename T>
struct RvalueTest{
    RvalueTest(T&& value) : value( std::forward<T>(value) ){}
    T&& value;
};

int main() {
    auto t = std::forward_as_tuple(Widget(12));
    //auto t = std::tuple<Widget&&>(Widget(12));  // this not work also
    std::cout << std::get<0>(t).h;                // <-- some random value here

    /// This one not work, either:    
    /*RvalueTest<Widget> r(Widget(12));
    std::cout << r.value.h;*/
}


Code from above works fine with clang compiler.
I suspect that rvalue object life time is wrong in -O2 mode.

[Bug c++/61213] [4.7/4.8/4.9/4.10 Regression] std::forward_as_tuple and std::tuple<T&&> and any rvalue loose data

[redi at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61213

Jonathan Wakely <redi at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2014-05-17
      Known to work|                            |4.6.4
            Summary|std::forward_as_tuple and   |[4.7/4.8/4.9/4.10
                   |std::tuple<T&&> and any     |Regression]
                   |rvalue loose data           |std::forward_as_tuple and
                   |                            |std::tuple<T&&> and any
                   |                            |rvalue loose data
     Ever confirmed|0                           |1
      Known to fail|                            |4.10.0, 4.7.3, 4.8.2, 4.9.0

--- Comment #1 from Jonathan Wakely <redi at gcc dot gnu.org> ---
4.6 gave the right answer

Using -fno-indirect-inlining gives the correct result

Also reproducable with -O1 -findirect-inlining -finline-small-functions

[Bug c++/61213] [4.7/4.8/4.9/4.10 Regression] std::forward_as_tuple and std::tuple<T&&> and any rvalue loose data

[tower120 at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61213

--- Comment #2 from tower120 <tower120 at gmail dot com> ---
(In reply to Jonathan Wakely from comment #1)
> Using -fno-indirect-inlining gives the correct result

I not see this with
-std=c++11 -O2 -fno-indirect-inlining -Wall -pedantic -pthread

http://coliru.stacked-crooked.com/a/e7c9408daef01448

[Bug c++/61213] [4.7/4.8/4.9/4.10 Regression] std::forward_as_tuple and std::tuple<T&&> and any rvalue loose data

[glisse at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61213

--- Comment #3 from Marc Glisse <glisse at gcc dot gnu.org> ---
Are you sure the code is valid? It doesn't seem so to me.

gcc's optimizers have become good enough to recycle the space from dead
temporaries.

[Bug c++/61213] [4.7/4.8/4.9/4.10 Regression] std::forward_as_tuple and std::tuple<T&&> and any rvalue loose data

[redi at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61213

Jonathan Wakely <redi at gcc dot gnu.org> changed:

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

--- Comment #4 from Jonathan Wakely <redi at gcc dot gnu.org> ---
(In reply to Marc Glisse from comment #3)
> Are you sure the code is valid? It doesn't seem so to me.

Good point, the Widget temporary goes out of scope at the end of the expression
and the tuple holds a dangling reference.