[Bug c++/58063] New: default arguments evaluated twice per call

[Bug c++/58063] New: default arguments evaluated twice per call

[plokinom at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58063

            Bug ID: 58063
           Summary: default arguments evaluated twice per call
           Product: gcc
           Version: 4.8.1
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: plokinom at gmail dot com

% cat try.cc
#include <iostream>

void f(bool x = !(std::cout << "hi!\n")) {
    std::cout << x << '\n';
}

int main() {
    f();
}

% g++ try.cc
% ./a.out 
hi!
hi!
0

My question: Why is the default argument evaluated twice for each call to f()?

Is this allowed by the standard or a bug in g++?


% g++ -v
Using built-in specs.
COLLECT_GCC=g++
COLLECT_LTO_WRAPPER=/home/mauke/usr/local/libexec/gcc/i686-pc-linux-gnu/4.8.1/lto-wrapper
Target: i686-pc-linux-gnu
Configured with: ../gcc-4.8.1/configure --prefix=/home/mauke/usr/local
--with-arch=native --with-tune=native --enable-__cxa_atexit
--enable-languages=c,c++,go --disable-nls
Thread model: posix
gcc version 4.8.1 (GCC)

[Bug c++/58063] default arguments evaluated twice per call

[paolo.carlini at oracle dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58063

--- Comment #2 from Paolo Carlini <paolo.carlini at oracle dot com> ---
The standard streams are indeed special, being constructed once and never
destroyed, see libstdc++-v3/src/c++98/ios_init.cc. I suppose a minimal
reproducer could involve a file scope static of some sort...

[Bug c++/58063] default arguments evaluated twice per call

[daniel.kruegler at googlemail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58063

--- Comment #3 from Daniel Krügler <daniel.kruegler at googlemail dot com> ---
(In reply to Paolo Carlini from comment #2)
> I suppose a minimal reproducer could involve a file scope static of some sort...

I'm a bit confused by your reply, Paolo: Isn't my_cout also a "file scope
static"? (Even, if I declare it to have static linkage, it still behaves
differently compared to std::cout)
>From gcc-bugs-return-427291-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Sat Aug 03 20:50:21 2013
Return-Path: <gcc-bugs-return-427291-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 20305 invoked by alias); 3 Aug 2013 20:50:21 -0000
Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Id: <gcc-bugs.gcc.gnu.org>
List-Archive: <http://gcc.gnu.org/ml/gcc-bugs/>
List-Post: <mailto:gcc-bugs@gcc.gnu.org>
List-Help: <mailto:gcc-bugs-help@gcc.gnu.org>
Sender: gcc-bugs-owner@gcc.gnu.org
Delivered-To: mailing list gcc-bugs@gcc.gnu.org
Received: (qmail 20222 invoked by uid 48); 3 Aug 2013 20:50:18 -0000
From: "paolo.carlini at oracle dot com" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug c++/58063] default arguments evaluated twice per call
Date: Sat, 03 Aug 2013 20:50:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: c++
X-Bugzilla-Version: 4.8.1
X-Bugzilla-Keywords:
X-Bugzilla-Severity: normal
X-Bugzilla-Who: paolo.carlini at oracle dot com
X-Bugzilla-Status: UNCONFIRMED
X-Bugzilla-Priority: P3
X-Bugzilla-Assigned-To: unassigned at gcc dot gnu.org
X-Bugzilla-Target-Milestone: ---
X-Bugzilla-Flags:
X-Bugzilla-Changed-Fields:
Message-ID: <bug-58063-4-qNUSnIWMHP@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-58063-4@http.gcc.gnu.org/bugzilla/>
References: <bug-58063-4@http.gcc.gnu.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 7bit
X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
X-SW-Source: 2013-08/txt/msg00215.txt.bz2
Content-length: 256

http://gcc.gnu.org/bugzilla/show_bug.cgi?idX063

--- Comment #4 from Paolo Carlini <paolo.carlini at oracle dot com> ---
Sorry, I didn't study it in sufficient detail. Anyway, no mysteries, this is
free software: libstdc++-v3/src/c++98/ios_init.cc etc.

[Bug c++/58063] default arguments evaluated twice per call

[paolo.carlini at oracle dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58063

--- Comment #5 from Paolo Carlini <paolo.carlini at oracle dot com> ---
Ah, in case isn't obvious already: it only happens when the "I/O expression"
has the ! operator in front.

[Bug c++/58063] default arguments evaluated twice per call

[daniel.kruegler at googlemail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58063

--- Comment #6 from Daniel Krügler <daniel.kruegler at googlemail dot com> ---
(In reply to Paolo Carlini from comment #5)
> Ah, in case isn't obvious already: it only happens when the "I/O expression"
> has the ! operator in front.

I suspected that and ensured that I added a similar operator! to my ostream
model type, but this hadn't any impact on the outcome for that type. Further, I
don't understand how it is related to ostream initialization, because the issue
also occurs when we invert the order of the function calls:

int main() {
 f2();
 std::cout << "------------------\n";
 f();
}

gives me:

<quote>
hi!
0
------------------
hi!
hi!
0
</quote>

Are you sure that this is not due to improper code generation?
>From gcc-bugs-return-427294-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Sat Aug 03 21:31:00 2013
Return-Path: <gcc-bugs-return-427294-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 13827 invoked by alias); 3 Aug 2013 21:31:00 -0000
Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Id: <gcc-bugs.gcc.gnu.org>
List-Archive: <http://gcc.gnu.org/ml/gcc-bugs/>
List-Post: <mailto:gcc-bugs@gcc.gnu.org>
List-Help: <mailto:gcc-bugs-help@gcc.gnu.org>
Sender: gcc-bugs-owner@gcc.gnu.org
Delivered-To: mailing list gcc-bugs@gcc.gnu.org
Received: (qmail 13732 invoked by uid 48); 3 Aug 2013 21:30:58 -0000
From: "paolo.carlini at oracle dot com" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug c++/58063] default arguments evaluated twice per call
Date: Sat, 03 Aug 2013 21:31:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: c++
X-Bugzilla-Version: 4.8.1
X-Bugzilla-Keywords:
X-Bugzilla-Severity: normal
X-Bugzilla-Who: paolo.carlini at oracle dot com
X-Bugzilla-Status: UNCONFIRMED
X-Bugzilla-Priority: P3
X-Bugzilla-Assigned-To: unassigned at gcc dot gnu.org
X-Bugzilla-Target-Milestone: ---
X-Bugzilla-Flags:
X-Bugzilla-Changed-Fields:
Message-ID: <bug-58063-4-6aM430M4Rm@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-58063-4@http.gcc.gnu.org/bugzilla/>
References: <bug-58063-4@http.gcc.gnu.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 7bit
X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
X-SW-Source: 2013-08/txt/msg00218.txt.bz2
Content-length: 361

http://gcc.gnu.org/bugzilla/show_bug.cgi?idX063

--- Comment #7 from Paolo Carlini <paolo.carlini at oracle dot com> ---
I'm not at all sure! But it happens with -O0 too, right?, thus at this point
the front-end seems more likely than the back-end, I would not change the
Component from c++ to something else. In any case we badly need a reduced
testcase ;)

[Bug c++/58063] default arguments evaluated twice per call

[daniel.kruegler at googlemail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58063

--- Comment #8 from Daniel Krügler <daniel.kruegler at googlemail dot com> ---
(In reply to Paolo Carlini from comment #7)
> But it happens with -O0 too, right?

Yes.

> In any case we badly need a reduced testcase ;)

I agree. Unfortunately I'm on vacations from tomorrow on (1 week), so I cannot
look into it until next week.
>From gcc-bugs-return-427296-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Sat Aug 03 21:37:11 2013
Return-Path: <gcc-bugs-return-427296-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 23362 invoked by alias); 3 Aug 2013 21:37:11 -0000
Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Id: <gcc-bugs.gcc.gnu.org>
List-Archive: <http://gcc.gnu.org/ml/gcc-bugs/>
List-Post: <mailto:gcc-bugs@gcc.gnu.org>
List-Help: <mailto:gcc-bugs-help@gcc.gnu.org>
Sender: gcc-bugs-owner@gcc.gnu.org
Delivered-To: mailing list gcc-bugs@gcc.gnu.org
Received: (qmail 23285 invoked by uid 48); 3 Aug 2013 21:37:09 -0000
From: "paolo.carlini at oracle dot com" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug c++/58063] default arguments evaluated twice per call
Date: Sat, 03 Aug 2013 21:37:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: c++
X-Bugzilla-Version: 4.8.1
X-Bugzilla-Keywords:
X-Bugzilla-Severity: normal
X-Bugzilla-Who: paolo.carlini at oracle dot com
X-Bugzilla-Status: UNCONFIRMED
X-Bugzilla-Priority: P3
X-Bugzilla-Assigned-To: unassigned at gcc dot gnu.org
X-Bugzilla-Target-Milestone: ---
X-Bugzilla-Flags:
X-Bugzilla-Changed-Fields:
Message-ID: <bug-58063-4-oFVqSlC4Kl@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-58063-4@http.gcc.gnu.org/bugzilla/>
References: <bug-58063-4@http.gcc.gnu.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 7bit
X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
X-SW-Source: 2013-08/txt/msg00220.txt.bz2
Content-length: 284

http://gcc.gnu.org/bugzilla/show_bug.cgi?idX063

--- Comment #9 from Paolo Carlini <paolo.carlini at oracle dot com> ---
Your help is always very appreciated, Daniel. Here we have plenty of work to do
anyway, if when you will back the bug will be unchanged, consider helping more.

[Bug c++/58063] default arguments evaluated twice per call

[fanael4 at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58063

Fanael <fanael4 at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |fanael4 at gmail dot com

--- Comment #10 from Fanael <fanael4 at gmail dot com> ---
My attempt at reducing:

struct basic_ios
{
  virtual ~basic_ios() {}
  bool operator!() const { return false; }
};

struct ostream : virtual basic_ios
{
  ostream() {}
  ~ostream() {}
private:
  ostream(const ostream&);
  ostream& operator=(const ostream&);
};

ostream& operator<<(ostream& os, const char* s) {
  __builtin_printf("%s", s);
  return os;
}

ostream cout;

void f(bool x = !(cout << "hi!\n")) {
  __builtin_printf("%d\n", static_cast<int>(x));
}

int main() {
  f();
}

Seems like virtual inheritance is the culprit.

[Bug c++/58063] default arguments evaluated twice per call

[paolo.carlini at oracle dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58063

Paolo Carlini <paolo.carlini at oracle dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |wrong-code
           Priority|P3                          |P2
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2013-08-04
                 CC|                            |jason at gcc dot gnu.org
     Ever confirmed|0                           |1

--- Comment #11 from Paolo Carlini <paolo.carlini at oracle dot com> ---
Thanks!

[Bug c++/58063] default arguments evaluated twice per call

[plokinom at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=58063

--- Comment #12 from plokinom at gmail dot com ---
I can confirm this still happens with g++ 5.1.0.

[Bug c++/58063] default arguments evaluated twice per call

[jason at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=58063

--- Comment #13 from Jason Merrill <jason at gcc dot gnu.org> ---
Author: jason
Date: Tue Jun 16 19:29:09 2015
New Revision: 224533

URL: https://gcc.gnu.org/viewcvs?rev=224533&root=gcc&view=rev
Log:
        PR c++/58063
        * tree.c (bot_manip): Remap SAVE_EXPR.

Added:
    trunk/gcc/testsuite/g++.dg/overload/defarg10.C
Modified:
    trunk/gcc/cp/ChangeLog
    trunk/gcc/cp/tree.c

[Bug c++/58063] default arguments evaluated twice per call

[jason at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=58063

--- Comment #14 from Jason Merrill <jason at gcc dot gnu.org> ---
Author: jason
Date: Wed Jun 17 14:33:17 2015
New Revision: 224559

URL: https://gcc.gnu.org/viewcvs?rev=224559&root=gcc&view=rev
Log:
        PR c++/58063
        * tree.c (bot_manip): Remap SAVE_EXPR.

Added:
    branches/gcc-5-branch/gcc/testsuite/g++.dg/overload/defarg10.C
Modified:
    branches/gcc-5-branch/gcc/cp/ChangeLog
    branches/gcc-5-branch/gcc/cp/tree.c

[Bug c++/58063] default arguments evaluated twice per call

[ville.voutilainen at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=58063

Ville Voutilainen <ville.voutilainen at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |ville.voutilainen at gmail dot com

--- Comment #15 from Ville Voutilainen <ville.voutilainen at gmail dot com> ---
*** Bug 61856 has been marked as a duplicate of this bug. ***

[Bug c++/58063] default arguments evaluated twice per call

[jason at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=58063

--- Comment #16 from Jason Merrill <jason at gcc dot gnu.org> ---
Author: jason
Date: Tue Aug 18 14:43:38 2015
New Revision: 226973

URL: https://gcc.gnu.org/viewcvs?rev=226973&root=gcc&view=rev
Log:
        PR c++/58063
        * tree.c (bot_manip): Remap SAVE_EXPR.

Added:
    branches/gcc-4_9-branch/gcc/testsuite/g++.dg/overload/defarg10.C
Modified:
    branches/gcc-4_9-branch/gcc/cp/ChangeLog
    branches/gcc-4_9-branch/gcc/cp/tree.c