[Bug c++/55635] New: Deallocation function ("operator delete") not called when destructor throws exception

[Bug c++/55635] New: Deallocation function ("operator delete") not called when destructor throws exception

[joedoefawnbuck at googlemail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55635

             Bug #: 55635
           Summary: Deallocation function ("operator delete") not called
                    when destructor throws exception
    Classification: Unclassified
           Product: gcc
           Version: 4.7.2
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: joedoefawnbuck@googlemail.com


According to C++11, 5.3.5/7, the deallocation function of a dynamically
allocated object must be called upon a `delete` expression regardless of
whether the destructor exits with an exception or not.

However, when I run the following code through valgrind, it says that I have
one still-reachable block:

struct Foo { ~Foo() { throw 0; } };

int main()
{
    auto p = new Foo;
    try { delete p; }
    catch (...) { }
}

Am I reading the standard right that the memory should be deallocated
regardless?

(SO discussion: http://stackoverflow.com/q/13792981/596781)

[Bug c++/55635] Deallocation function ("operator delete") not called when destructor throws exception

[redi at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55635

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2012-12-10
     Ever Confirmed|0                           |1

--- Comment #1 from Jonathan Wakely <redi at gcc dot gnu.org> 2012-12-10 00:56:28 UTC ---
Your destructor calls terminate() according to C++11, it should be declared
noexcept(false) to demonstrate the bug.

[Bug c++/55635] Deallocation function ("operator delete") not called when destructor throws exception

[joedoefawnbuck at googlemail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55635

--- Comment #2 from Kerrek SB <joedoefawnbuck at googlemail dot com> 2012-12-10 09:08:46 UTC ---
Interesting, thank you!

The bug remains, however, both with -std=c++11 and noexcept(false), and with
-std=c++98 and throw(int).

Also, with noexcept(false) GCC will *not* call terminate()...

[Bug c++/55635] Deallocation function ("operator delete") not called when destructor throws exception

[joedoefawnbuck at googlemail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55635

--- Comment #3 from Kerrek SB <joedoefawnbuck at googlemail dot com> 2012-12-10 09:10:25 UTC ---
(Sorry, I meant "*without* noexcept(false)".)

[Bug c++/55635] Deallocation function ("operator delete") not called when destructor throws exception

[redi at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55635

--- Comment #4 from Jonathan Wakely <redi at gcc dot gnu.org> 2012-12-10 10:37:29 UTC ---
G++ 4.7 doesn't call terminate because it doesn't implement
http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#1123, but trunk
does.

[Bug c++/55635] Deallocation function ("operator delete") not called when destructor throws exception

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=55635

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |wrong-code
   Target Milestone|---                         |7.0

[Bug c++/55635] Deallocation function ("operator delete") not called when destructor throws exception

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=55635

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |programmerjake at gmail dot com

--- Comment #7 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
*** Bug 70447 has been marked as a duplicate of this bug. ***