[Bug c++/3187] gcc lays down two copies of constructors

[Bug c++/3187] gcc lays down two copies of constructors

[pinskia at gcc dot gnu dot org]

------- Comment #19 from pinskia at gcc dot gnu dot org  2005-11-02 02:07 -------
Supending as there was ABI work that needed to be done to fix this.


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |SUSPENDED
           Keywords|                            |ABI


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=3187

[Bug c++/3187] gcc lays down two copies of constructors

[ian at airs dot com]

------- Comment #20 from ian at airs dot com  2006-02-05 00:19 -------
In many common cases, the two constructors are identical.  It should not be
hard to simply emit both appropriate symbols before the function.  That does
not require any change to the ABI, but it eliminates the problem in the common
case.

I don't feel that this PR should be suspended, at least not until we have fixed
the common case.


-- 

ian at airs dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |ian at airs dot com


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=3187

Re: [Bug c++/3187] gcc lays down two copies of constructors

[Andrew Pinski]

On Feb 4, 2006, at 7:20 PM, ian at airs dot com wrote:

>
>
> ------- Comment #20 from ian at airs dot com  2006-02-05 00:19 -------
> In many common cases, the two constructors are identical.  It should 
> not be
> hard to simply emit both appropriate symbols before the function.  
> That does
> not require any change to the ABI, but it eliminates the problem in 
> the common
> case.

But that does not work for some assemblers/file formats (like Darwin) as
Darwin's as finds subsections via labels.

There has been some discussion on why this suggestion would not
work on the mailing list (but I cannot find it right now).

Thanks,
Andrew Pinski

[Bug c++/3187] gcc lays down two copies of constructors

[pinskia at physics dot uc dot edu]

------- Comment #21 from pinskia at gcc dot gnu dot org  2006-02-05 00:27 -------
Subject: Re:  gcc lays down two copies of constructors


On Feb 4, 2006, at 7:20 PM, ian at airs dot com wrote:

>
>
> ------- Comment #20 from ian at airs dot com  2006-02-05 00:19 -------
> In many common cases, the two constructors are identical.  It should 
> not be
> hard to simply emit both appropriate symbols before the function.  
> That does
> not require any change to the ABI, but it eliminates the problem in 
> the common
> case.

But that does not work for some assemblers/file formats (like Darwin) as
Darwin's as finds subsections via labels.

There has been some discussion on why this suggestion would not
work on the mailing list (but I cannot find it right now).

Thanks,
Andrew Pinski


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=3187

[Bug c++/3187] gcc lays down two copies of constructors

[ian at airs dot com]

------- Comment #22 from ian at airs dot com  2006-02-05 00:42 -------
The fact that it does not work everywhere is not a valid reason that it should
not be implemented where it can work, particularly since the places where it
can work are, as it happens, most places.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=3187

[Bug c++/3187] gcc lays down two copies of constructors

[pinskia at gcc dot gnu dot org]

------- Comment #23 from pinskia at gcc dot gnu dot org  2006-02-05 00:55 -------
For future reference:
a patch which would fix this:
http://gcc.gnu.org/ml/gcc-patches/2002-08/msg00354.html

Previous discussions about this issue:
http://gcc.gnu.org/ml/gcc/2002-12/msg00474.html

More about the patch:
http://gcc.gnu.org/ml/gcc-patches/2004-05/msg00689.html

There is more like other threads too which I cannot find right now.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=3187

[Bug c++/3187] gcc lays down two copies of constructors

[gdr at integrable-solutions dot net]

------- Comment #24 from gdr at integrable-solutions dot net  2006-02-05 03:58 -------
Subject: Re:  gcc lays down two copies of constructors

"ian at airs dot com" <gcc-bugzilla@gcc.gnu.org> writes:

| I don't feel that this PR should be suspended, at least not until we
| have fixed the common case.

I think I agree with Ian.

-- Gaby


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=3187

[Bug c++/3187] gcc lays down two copies of constructors

[gdr at integrable-solutions dot net]

------- Comment #25 from gdr at integrable-solutions dot net  2006-02-05 04:00 -------
Subject: Re:  gcc lays down two copies of constructors

"pinskia at physics dot uc dot edu" <gcc-bugzilla@gcc.gnu.org> writes:

| But that does not work for some assemblers/file formats (like Darwin) as
| Darwin's as finds subsections via labels.

then we would just have not to implement what Ian proposes for Darwin.

-- Gaby


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=3187

[Bug c++/3187] gcc lays down two copies of constructors

[gcc at mirality dot co dot nz]

------- Comment #26 from gcc at mirality dot co dot nz  2006-09-05 07:18 -------
This is very aggravating, and *NEEDS* to get fixed soon.  Even if only for the
common duplicate-symbol-on-most-platforms case, that's a significant
improvement over what it's doing now.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=3187

[Bug c++/3187] gcc lays down two copies of constructors

[maddox at google dot com]

------- Comment #27 from maddox at google dot com  2007-05-25 04:57 -------
(In reply to comment #20)

Ian proposed handling the simple case in which the two constructors ("clones")
are identical by emitting the code only once, but labelling it with the symbols
for both constructors.  This scheme fails in the case that the constructors
must go in a comdat group. What do we name the group?  If we invent a new name,
this will have ABI impact, as other implementations must not step on it.  If we
use the name of one of the constructors, we may find that the section is
discarded in favor of a like-named section produced by another compiler, but
which need not define both symbols, resulting in link-incompatibility.  It
appears, then,
that a distinct group name must be used for the combined constructor.  As I
understand it, this was the principal issue that kept Stuart's patch
(referenced in comment #23) from being adopted, so I don't see how this more
limited optimization sidesteps the objections to that patch.  I spoke with Ian
this afternoon, and he agreed with this analysis.  It was suggested in the
discussion of Stuart's patch that it would be sufficient to provide for a
vendor namespace.
This seems like a simple, general facility that could be leveraged in multiple
contexts, e.g., to support a more general subroutinization optimization.  How
difficult would it be to get such a provision into the ABI?  With a suitably
chosen mangling convention, I don't think that name collisions with other
ABI-conformant implementations would actually be a problem in practice, so it
seems to be more of a matter of agreeing on a convention for the sake of the
ABI standard rather than actually managing the transition.  Comments?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=3187

[Bug c++/3187] gcc lays down two copies of constructors

[jason at gcc dot gnu dot org]

------- Comment #28 from jason at gcc dot gnu dot org  2008-02-19 20:20 -------
unsuspending.


-- 

jason at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jason at gcc dot gnu dot org
             Status|SUSPENDED                   |NEW
   Last reconfirmed|2005-08-03 02:30:41         |2008-02-19 20:20:48
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=3187

[Bug c++/3187] gcc lays down two copies of constructors

[jason at gcc dot gnu dot org]

-- 

jason at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |jason at gcc dot gnu dot org
                   |dot org                     |
             Status|NEW                         |ASSIGNED
   Last reconfirmed|2008-02-19 20:20:48         |2008-02-25 18:37:01
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=3187

[Bug c++/3187] gcc lays down two copies of constructors

[dbaldin at upb dot de]

------- Comment #29 from dbaldin at upb dot de  2008-07-29 11:32 -------
I wonder if there is anybody working on this bug since there is no target
milestone assigned to it. This bug is definitly not an enhancement since due to
this bug it is impossible/or very hard to create good software for embedded
systems with very limited resources. We are currently working on an rtos
designed to work only 64 kb of flash memory and the binary size is drastically
increased by this bug. 

This bug is very annoying and should soon be fixed knowing that this bug has
been out there for more than 7 years ... 

Greetings


-- 

dbaldin at upb dot de changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |dbaldin at upb dot de


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=3187

[Bug c++/3187] gcc lays down two copies of constructors

[manu at gcc dot gnu dot org]

------- Comment #30 from manu at gcc dot gnu dot org  2008-07-29 13:57 -------
Dear Daniel, we would like to fix all bugs but we cannot force volunteers to
fix specific bugs and, on the other hand, hired developers fix those bugs that
are most interesting for their employers. 

If this were such a huge issue for the embedded market, surely, some company or
consortium would have already hired CodeSourcery or someone else to fix it long
time ago.

Greetings.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=3187

[Bug c++/3187] gcc lays down two copies of constructors

[amylaar at gcc dot gnu dot org]

------- Comment #31 from amylaar at gcc dot gnu dot org  2008-08-01 16:07 -------
(In reply to comment #27)
> (In reply to comment #20)
> 
> Ian proposed handling the simple case in which the two constructors ("clones")
> are identical by emitting the code only once, but labelling it with the symbols
> for both constructors.  This scheme fails in the case that the constructors
> must go in a comdat group. What do we name the group?  If we invent a new name,
> this will have ABI impact, as other implementations must not step on it.  If we
> use the name of one of the constructors, we may find that the section is
> discarded in favor of a like-named section produced by another compiler, but
> which need not define both symbols, resulting in link-incompatibility.

link-incompatibility is often of a lesser concern in the embedded world,
where some people are prepared to rebuild all their libraries if they can
save 20% code space this way.
Thus, having an option to use the scheme where the comdat section - if any -
is named after only one of the constructors / destructors would be helpful.
For situations where there are more concerns about link compatibility, I
suppose we could initiate a transitionary period where we put a weak
definition of the *1* con/destructor in the *2* section, and vice versa.
When a user is confident that all the libraries are built at least with this
transitionary scheme, they can turn on a switch to only emit the needed
sections, thus saving at least some space where not all the constructors /
destructors are coming from the libraries.
This transitionary period could be ended on a target-by-target basis when
the an ABI change is deemed safe, or extended indefinitely where it is
not considered safe.


-- 

amylaar at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |amylaar at gcc dot gnu dot
                   |                            |org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=3187

[Bug c++/3187] gcc lays down two copies of constructors

[alexandre dot nunes at gmail dot com]

------- Comment #32 from alexandre dot nunes at gmail dot com  2008-10-21 18:37 -------
I was considering using C++ for an arm-elf target, but I'm dropping that in
favour of plain C because of this silly thing. This sucks, because other than
that g++ does a pretty decent job when generating small, optimized code: I've
got no abstraction penalty at all, but a few duplicated constructors here and
there lead to about 23% code increase, and that wasn't acceptable.

The other funny thing is that if this is fixed, it'll be by 4.4 or something,
and so far I can't consider using gcc versions newer than 4.2.x, at least
because of PR31849.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=3187

[Bug c++/3187] gcc lays down two copies of constructors

[jason at gcc dot gnu dot org]

------- Comment #33 from jason at gcc dot gnu dot org  2008-12-22 03:11 -------
Bill Maddox posted a patch at
http://gcc.gnu.org/ml/gcc-patches/2007-11/msg01147.html


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=3187

[Bug c++/3187] gcc lays down two copies of constructors

[carrot at google dot com]

------- Comment #34 from carrot at google dot com  2009-08-27 01:40 -------
There is one optimization that we can do without affecting the ABI and linker
compatibility. The delete destructor(D0) always contains the content of
complete desturctor(D1) followed by a function call to delete. So instead of
cloning the abstract destructor function body to the delete destructor(D0), we
can generate a function call to complete destructor(D1) followed by a function
call to delete.


-- 

carrot at google dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |carrot at google dot com


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=3187

[Bug c++/3187] gcc lays down two copies of constructors

[jason at gcc dot gnu dot org]

-- 

jason at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|jason at gcc dot gnu dot org|unassigned at gcc dot gnu
                   |                            |dot org
             Status|ASSIGNED                    |NEW


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=3187

[Bug c++/3187] gcc lays down two copies of constructors

[jakub at gcc dot gnu dot org]

------- Comment #35 from jakub at gcc dot gnu dot org  2009-11-18 09:54 -------
Subject: Bug 3187

Author: jakub
Date: Wed Nov 18 09:53:52 2009
New Revision: 154284

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=154284
Log:
        PR c++/3187
        * cgraph.h (struct cgraph_node): Add same_body and same_body_alias
        fields.
        (cgraph_same_body_alias, cgraph_remove_same_body_alias): New
        prototypes.
        * cgraphunit.c (cgraph_expand_function, cgraph_emit_thunks,
        cgraph_materialize_all_clones): Handle same_body aliases.
        * cgraph.c (cgraph_allocate_node): New function.
        (cgraph_create_node): Use it.
        (cgraph_node_for_decl, cgraph_node, cgraph_get_node,
        cgraph_node_for_asm, cgraph_remove_node): Handle same_body aliases.
        (cgraph_same_body_alias, cgraph_remove_same_body_alias): New
        functions.
        * lto-cgraph.c (lto_output_node): Stream out same_body aliases.
        (input_node): Stream in same_body aliases.
        * lto-symtab.c (lto_cgraph_replace_node): Clear node pointers
        for same_body aliases.
        (lto_symtab_merge_cgraph_nodes_1): Handle same_body aliases.

        * cp-tree.h (expand_or_defer_fn_1): New prototype.
        * decl2.c (cp_write_global_declarations): Mark as !DECL_EXTERNAL
        also all same_body aliases.
        * semantics.c (expand_or_defer_fn): Move most of the function
        except registering with cgraph to ...
        (expand_or_defer_fn_1): ... here.  New function.
        * optimize.c: Include cgraph.h.
        (maybe_clone_body): If in charge parm is not used and both base
        and complete clones are created and are not comdat, tell cgraph
        they have the same body.
        * Make-lang.in (cp/optimize.o): Depend on $(CGRAPH_H).

        * g++.dg/abi/mangle26.C: Also match *C2* definition.
        * g++.dg/abi/mangle27.C: Likewise.
        * g++.dg/abi/mangle28.C: Likewise.
        * g++.dg/abi/mangle29.C: Likewise.

Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/cgraph.c
    trunk/gcc/cgraph.h
    trunk/gcc/cgraphunit.c
    trunk/gcc/cp/ChangeLog
    trunk/gcc/cp/Make-lang.in
    trunk/gcc/cp/cp-tree.h
    trunk/gcc/cp/decl2.c
    trunk/gcc/cp/optimize.c
    trunk/gcc/cp/semantics.c
    trunk/gcc/lto-cgraph.c
    trunk/gcc/lto-symtab.c
    trunk/gcc/testsuite/ChangeLog
    trunk/gcc/testsuite/g++.dg/abi/mangle26.C
    trunk/gcc/testsuite/g++.dg/abi/mangle27.C
    trunk/gcc/testsuite/g++.dg/abi/mangle28.C
    trunk/gcc/testsuite/g++.dg/abi/mangle29.C


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=3187

[Bug c++/3187] gcc lays down two copies of constructors

[jakub at gcc dot gnu dot org]

------- Comment #36 from jakub at gcc dot gnu dot org  2009-12-01 20:09 -------
Subject: Bug 3187

Author: jakub
Date: Tue Dec  1 20:09:37 2009
New Revision: 154880

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=154880
Log:
        PR c++/3187
        * optimize.c (cdtor_comdat_group): New function.
        (maybe_clone_body): Also optimize DECL_COMDAT base/complete cdtors
        and in that case put also the deleting dtor in the same comdat group
        as base and complete dtor if dtor is virtual.

Modified:
    trunk/gcc/cp/ChangeLog
    trunk/gcc/cp/optimize.c


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=3187

[Bug c++/3187] gcc lays down two copies of constructors

[jakub at gcc dot gnu dot org]

------- Comment #37 from jakub at gcc dot gnu dot org  2009-12-02 14:31 -------
Subject: Bug 3187

Author: jakub
Date: Wed Dec  2 14:31:21 2009
New Revision: 154912

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=154912
Log:
Fix a backport glitch for PR c++/3187.

Modified:
    branches/redhat/gcc-4_4-branch/gcc/cp/ChangeLog
    branches/redhat/gcc-4_4-branch/gcc/cp/decl2.c


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=3187

[Bug c++/3187] gcc lays down two copies of constructors

[ararunprasad at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=3187

Arunprasad <ararunprasad at gmail dot com> changed:

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

--- Comment #38 from Arunprasad <ararunprasad at gmail dot com> 2012-10-18 18:01:14 UTC ---
from which version this fix will be available?

[Bug c++/3187] gcc lays down two copies of constructors

[paolo.carlini at oracle dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=3187

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|gcc-bugs at gcc dot gnu.org |jakub at gcc dot gnu.org

--- Comment #39 from Paolo Carlini <paolo.carlini at oracle dot com> 2012-10-18 23:01:59 UTC ---
For sure I can see the Nov 2009 commit in the 4.5.0 ChangeLog, but it would be
indeed nice if either Jakub or Jason could write a few words in the audit trail
here, explaining why the PR is still open and what is still missing.

[Bug c++/3187] gcc lays down two copies of constructors

[ararunprasad at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=3187

--- Comment #40 from Arunprasad <ararunprasad at gmail dot com> 2012-10-19 06:41:04 UTC ---
Thank you.Is there any way to find it from nm output.?

[Bug c++/3187] gcc lays down two copies of constructors

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=3187

--- Comment #41 from Jakub Jelinek <jakub at gcc dot gnu.org> 2012-10-19 10:05:55 UTC ---
The reason why this hasn't been closed is that we only use an alias of one kind
of ctor (resp. dtor) to the other one if they are the same (and for deleting
dtor just always call the other dtor).  If they are different, then they can't
be aliased together, what we could do (perhaps as an option) is to emit another
function, which would take an argument what kind of ctor resp. dtor it is, and
behave differently depending on that argument, then have both kinds of ctor
resp. dtor to tail call (or if not possible, just call) the other function.
I guess it could have ABI consequences though (in which comdat section to place
it).

[Bug c++/3187] gcc lays down two copies of constructors

[ararunprasad at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=3187

--- Comment #42 from Arunprasad <ararunprasad at gmail dot com> 2012-10-19 10:34:11 UTC ---
So I'm assuming like the issue still exists in gcc family of tool-chains. Fix
has been temporarily suspended due to ABI compatibility.

[Bug c++/3187] gcc lays down two copies of constructors

[mmitchel at gcc dot gnu dot org]

------- Additional Comments From mmitchel at gcc dot gnu dot org  2004-01-16 18:19 -------
I'm changing this to an enhancement request.

It is, in some sense, a regression -- but it's really a direct consequence of
the new ABI.  

Fixing it will require substantial effort.  I hope that we will get a chance to
do the multiple-entry point stuff -- but it's certainly not going to happen
before GCC 3.4.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|normal                      |enhancement
            Summary|[3.3/3.4 Regression] gcc    |gcc lays down two copies of
                   |lays down two copies of     |constructors
                   |constructors                |
   Target Milestone|3.4.0                       |---


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=3187

[Bug c++/3187] gcc lays down two copies of constructors

[giovannibajo at libero dot it]

------- Additional Comments From giovannibajo at libero dot it  2004-01-18 02:45 -------
*** Bug 13725 has been marked as a duplicate of this bug. ***

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |rmerkert at alphatech dot
                   |                            |com


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=3187

[Bug c++/3187] gcc lays down two copies of constructors

[pinskia at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
OtherBugsDependingO|                            |16996
              nThis|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=3187

[Bug c++/3187] gcc lays down two copies of constructors

[larue at cadence dot com]

------- Additional Comments From larue at cadence dot com  2005-06-15 21:09 -------
This problem makes it difficult to debug C++ code. This is affecting the 
systemC (www.systemc.org) community. It is more than a minor issue those 
debugging C++.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |larue at cadence dot com


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=3187

[Bug c++/3187] gcc lays down two copies of constructors

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-06-15 21:19 -------
(In reply to comment #17)
> This problem makes it difficult to debug C++ code. This is affecting the 
> systemC (www.systemc.org) community. It is more than a minor issue those 
> debugging C++.
The debuger should be able to handle this as it will also get templates wrong.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=3187