public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
* [Bug libstdc++/56613] New: [4.8 Regression] map::operator[](key_type&&) fails with custom allocator
@ 2013-03-13 13:33 redi at gcc dot gnu.org
  2013-03-13 13:35 ` [Bug libstdc++/56613] " redi at gcc dot gnu.org
                   ` (4 more replies)
  0 siblings, 5 replies; 6+ messages in thread
From: redi at gcc dot gnu.org @ 2013-03-13 13:33 UTC (permalink / raw)
  To: gcc-bugs


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56613

             Bug #: 56613
           Summary: [4.8 Regression] map::operator[](key_type&&) fails
                    with custom allocator
    Classification: Unclassified
           Product: gcc
           Version: 4.8.0
            Status: UNCONFIRMED
          Keywords: rejects-valid
          Severity: normal
          Priority: P3
         Component: libstdc++
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: redi@gcc.gnu.org


This program is valid in C++03 and C++11 but can't be compiled with trunk using
-std=c++11

#include <map>

template<typename T>
struct alloc
{
    typedef T value_type;
    typedef T* pointer;
    typedef const T* const_pointer;
    typedef T& reference;
    typedef const T& const_reference;
    typedef unsigned size_type;
    typedef int difference_type;

    template<typename U>
        struct rebind {
            typedef alloc<U> other;
        };

    alloc() { }
    template<typename U>
        alloc(const alloc<U>&) { }

    pointer allocate(size_type n, const void* = 0) { return
std::allocator<T>().allocate(n); }
    void deallocate(pointer p, size_type n) { std::allocator<T>().deallocate(p,
n); }

    size_type max_size() const { return -1; }

    void construct(pointer p, const T& t) { new ((void*) p) T(t); }
    void destroy(pointer p) { p->~T(); }
};

template<typename T, typename U>
bool operator==(alloc<T>, alloc<U>) { return true; }

template<typename T, typename U>
bool operator!=(alloc<T>, alloc<U>) { return false; }

int main()
{
    std::map<int, int, std::less<int>, alloc<int> > m;
    m[1];
}


$ g++-4.8 -std=c++11 map.cc
In file included from /home/redi/gcc/4.x/include/c++/4.8.0/map:60:0,
                 from map.cc:1:
/home/redi/gcc/4.x/include/c++/4.8.0/bits/stl_tree.h: In instantiation of
'std::_Rb_tree_node<_Val>* std::_Rb_tree<_Key, _Val, _KeyOfValue, _Compare,
_Alloc>::_M_create_node(_Args&& ...) [with _Args = {const
std::piecewise_construct_t&, std::tuple<int&&>, std::tuple<>}; _Key = int; _Val
= std::pair<const int, int>; _KeyOfValue = std::_Select1st<std::pair<const int,
int> >; _Compare = std::less<int>; _Alloc = alloc<std::pair<const int, int> >;
std::_Rb_tree<_Key, _Val, _KeyOfValue, _Compare, _Alloc>::_Link_type =
std::_Rb_tree_node<std::pair<const int, int> >*]':
/home/redi/gcc/4.x/include/c++/4.8.0/bits/stl_tree.h:1662:64:   required from
'std::_Rb_tree<_Key, _Val, _KeyOfValue, _Compare, _Alloc>::iterator
std::_Rb_tree<_Key, _Val, _KeyOfValue, _Compare,
_Alloc>::_M_emplace_hint_unique(std::_Rb_tree<_Key, _Val, _KeyOfValue,
_Compare, _Alloc>::const_iterator, _Args&& ...) [with _Args = {const
std::piecewise_construct_t&, std::tuple<int&&>, std::tuple<>}; _Key = int; _Val
= std::pair<const int, int>; _KeyOfValue = std::_Select1st<std::pair<const int,
int> >; _Compare = std::less<int>; _Alloc = alloc<std::pair<const int, int> >;
std::_Rb_tree<_Key, _Val, _KeyOfValue, _Compare, _Alloc>::iterator =
std::_Rb_tree_iterator<std::pair<const int, int> >; std::_Rb_tree<_Key, _Val,
_KeyOfValue, _Compare, _Alloc>::const_iterator =
std::_Rb_tree_const_iterator<std::pair<const int, int> >]'
/home/redi/gcc/4.x/include/c++/4.8.0/bits/stl_map.h:484:8:   required from
'std::map<_Key, _Tp, _Compare, _Alloc>::mapped_type& std::map<_Key, _Tp,
_Compare, _Alloc>::operator[](std::map<_Key, _Tp, _Compare,
_Alloc>::key_type&&) [with _Key = int; _Tp = int; _Compare = std::less<int>;
_Alloc = alloc<int>; std::map<_Key, _Tp, _Compare, _Alloc>::mapped_type = int;
std::map<_Key, _Tp, _Compare, _Alloc>::key_type = int]'
map.cc:41:8:   required from here
/home/redi/gcc/4.x/include/c++/4.8.0/bits/stl_tree.h:403:8: error: no matching
function for call to 'alloc<std::_Rb_tree_node<std::pair<const int, int> >
>::construct(std::_Rb_tree_node<std::pair<const int, int> >*&, const
std::piecewise_construct_t&, std::tuple<int&&>, std::tuple<>)'
        _M_get_Node_allocator().construct(__tmp,
        ^
/home/redi/gcc/4.x/include/c++/4.8.0/bits/stl_tree.h:403:8: note: candidate is:
map.cc:28:10: note: void alloc<T>::construct(alloc<T>::pointer, const T&) [with
T = std::_Rb_tree_node<std::pair<const int, int> >; alloc<T>::pointer =
std::_Rb_tree_node<std::pair<const int, int> >*]
     void construct(pointer p, const T& t) { new ((void*) p) T(t); }
          ^
map.cc:28:10: note:   candidate expects 2 arguments, 4 provided


The problem is that stl_map.h uses piecewise construction for the pair object,
which calls alloc::construct with four arguments, incorrectly assuming that all
allocators in C++11 have a variadic construct() member.

It's easily fixed by changing the C++11 version of _M_create_node to use
allocator_traits::construct:

                std::allocator_traits<_Node_allocator>::
                    construct(_M_get_Node_allocator(), __tmp,
                              std::forward<_Args>(__args)...);


(For 4.9 I was going to make all the RB-trees allocator-aware, but this
regression needs to be fixed for 4.8 without the rest of the allocator changes)


^ permalink raw reply	[flat|nested] 6+ messages in thread

* [Bug libstdc++/56613] [4.8 Regression] map::operator[](key_type&&) fails with custom allocator
  2013-03-13 13:33 [Bug libstdc++/56613] New: [4.8 Regression] map::operator[](key_type&&) fails with custom allocator redi at gcc dot gnu.org
@ 2013-03-13 13:35 ` redi at gcc dot gnu.org
  2013-03-14  1:44 ` redi at gcc dot gnu.org
                   ` (3 subsequent siblings)
  4 siblings, 0 replies; 6+ messages in thread
From: redi at gcc dot gnu.org @ 2013-03-13 13:35 UTC (permalink / raw)
  To: gcc-bugs


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56613

--- Comment #1 from Jonathan Wakely <redi at gcc dot gnu.org> 2013-03-13 13:35:14 UTC ---
Oops, the program isn't quite valid C++03, I forgot these members of the
allocator:

  pointer       address(reference x) const throw()       { return &x; }
  const_pointer address(const_reference x) const throw() { return &x; }

That doesn't affect the issue though.


^ permalink raw reply	[flat|nested] 6+ messages in thread

* [Bug libstdc++/56613] [4.8 Regression] map::operator[](key_type&&) fails with custom allocator
  2013-03-13 13:33 [Bug libstdc++/56613] New: [4.8 Regression] map::operator[](key_type&&) fails with custom allocator redi at gcc dot gnu.org
  2013-03-13 13:35 ` [Bug libstdc++/56613] " redi at gcc dot gnu.org
@ 2013-03-14  1:44 ` redi at gcc dot gnu.org
  2013-03-14  9:14 ` rguenth at gcc dot gnu.org
                   ` (2 subsequent siblings)
  4 siblings, 0 replies; 6+ messages in thread
From: redi at gcc dot gnu.org @ 2013-03-14  1:44 UTC (permalink / raw)
  To: gcc-bugs


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56613

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |ASSIGNED
   Last reconfirmed|                            |2013-03-14
         AssignedTo|unassigned at gcc dot       |redi at gcc dot gnu.org
                   |gnu.org                     |
     Ever Confirmed|0                           |1

--- Comment #2 from Jonathan Wakely <redi at gcc dot gnu.org> 2013-03-14 01:43:53 UTC ---
Created attachment 29664
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=29664
fix the bug by using allocator_traits


^ permalink raw reply	[flat|nested] 6+ messages in thread

* [Bug libstdc++/56613] [4.8 Regression] map::operator[](key_type&&) fails with custom allocator
  2013-03-13 13:33 [Bug libstdc++/56613] New: [4.8 Regression] map::operator[](key_type&&) fails with custom allocator redi at gcc dot gnu.org
  2013-03-13 13:35 ` [Bug libstdc++/56613] " redi at gcc dot gnu.org
  2013-03-14  1:44 ` redi at gcc dot gnu.org
@ 2013-03-14  9:14 ` rguenth at gcc dot gnu.org
  2013-03-14 23:28 ` redi at gcc dot gnu.org
  2013-03-14 23:31 ` redi at gcc dot gnu.org
  4 siblings, 0 replies; 6+ messages in thread
From: rguenth at gcc dot gnu.org @ 2013-03-14  9:14 UTC (permalink / raw)
  To: gcc-bugs


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56613

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P1
      Known to work|                            |4.7.2
   Target Milestone|---                         |4.8.0


^ permalink raw reply	[flat|nested] 6+ messages in thread

* [Bug libstdc++/56613] [4.8 Regression] map::operator[](key_type&&) fails with custom allocator
  2013-03-13 13:33 [Bug libstdc++/56613] New: [4.8 Regression] map::operator[](key_type&&) fails with custom allocator redi at gcc dot gnu.org
                   ` (2 preceding siblings ...)
  2013-03-14  9:14 ` rguenth at gcc dot gnu.org
@ 2013-03-14 23:28 ` redi at gcc dot gnu.org
  2013-03-14 23:31 ` redi at gcc dot gnu.org
  4 siblings, 0 replies; 6+ messages in thread
From: redi at gcc dot gnu.org @ 2013-03-14 23:28 UTC (permalink / raw)
  To: gcc-bugs


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56613

--- Comment #3 from Jonathan Wakely <redi at gcc dot gnu.org> 2013-03-14 23:28:18 UTC ---
Author: redi
Date: Thu Mar 14 23:28:11 2013
New Revision: 196666

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=196666
Log:
    PR libstdc++/56613
    * include/bits/stl_tree.h (_Rb_tree::_M_create_node): Use
    allocator_traits instead of calling construct directly.
    * testsuite/23_containers/map/56613.cc: New.

Added:
    trunk/libstdc++-v3/testsuite/23_containers/map/56613.cc
Modified:
    trunk/libstdc++-v3/ChangeLog
    trunk/libstdc++-v3/include/bits/stl_tree.h


^ permalink raw reply	[flat|nested] 6+ messages in thread

* [Bug libstdc++/56613] [4.8 Regression] map::operator[](key_type&&) fails with custom allocator
  2013-03-13 13:33 [Bug libstdc++/56613] New: [4.8 Regression] map::operator[](key_type&&) fails with custom allocator redi at gcc dot gnu.org
                   ` (3 preceding siblings ...)
  2013-03-14 23:28 ` redi at gcc dot gnu.org
@ 2013-03-14 23:31 ` redi at gcc dot gnu.org
  4 siblings, 0 replies; 6+ messages in thread
From: redi at gcc dot gnu.org @ 2013-03-14 23:31 UTC (permalink / raw)
  To: gcc-bugs


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56613

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

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

--- Comment #4 from Jonathan Wakely <redi at gcc dot gnu.org> 2013-03-14 23:30:53 UTC ---
fixed


^ permalink raw reply	[flat|nested] 6+ messages in thread

end of thread, other threads:[~2013-03-14 23:31 UTC | newest]

Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2013-03-13 13:33 [Bug libstdc++/56613] New: [4.8 Regression] map::operator[](key_type&&) fails with custom allocator redi at gcc dot gnu.org
2013-03-13 13:35 ` [Bug libstdc++/56613] " redi at gcc dot gnu.org
2013-03-14  1:44 ` redi at gcc dot gnu.org
2013-03-14  9:14 ` rguenth at gcc dot gnu.org
2013-03-14 23:28 ` redi at gcc dot gnu.org
2013-03-14 23:31 ` redi at gcc dot gnu.org

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).