Re: middle-end/6418: mangle.c #2199 (GCC3.04) - Emits *INTERNAL* when handling weak defined templates con- and destructors

Re: middle-end/6418: mangle.c #2199 (GCC3.04) - Emits *INTERNAL* when handling weak defined templates con- and destructors

[bangerth]

Synopsis: mangle.c #2199 (GCC3.04) - Emits *INTERNAL* when handling weak defined templates con- and destructors

State-Changed-From-To: feedback->closed
State-Changed-By: bangerth
State-Changed-When: Fri Nov 15 07:08:52 2002
State-Changed-Why:
    I wanted to have a complete testcase, which has not been
    provided. But I made up one myself which indeed failed with
    3.0.4. It does not fail any more with 3.2.1 and 3.3 CVS,
    so I guess the problem is fixed.

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=6418

Re: middle-end/6418: mangle.c #2199 (GCC3.04) - Emits *INTERNAL* when handling weak defined templates con- and destructors

[Michael Skriver]

The following reply was made to PR middle-end/6418; it has been noted by GNATS.

From: Michael Skriver <skriver@cs.auc.dk>
To: bangerth@dealii.org, gcc-bugs@gcc.gnu.org, gcc-prs@gcc.gnu.org,
   msk@ignlabs.com, nobody@gcc.gnu.org, gcc-gnats@gcc.gnu.org
Cc:  
Subject: Re: middle-end/6418: mangle.c #2199 (GCC3.04) - Emits *INTERNAL* when handling weak defined templates con- and destructors
Date: Fri, 15 Nov 2002 12:12:26 +0100

 Hey Bangerth
 
 Thanks for your time to see into the problem. Though it most likely to be fixed
 in later versions since, it was a comment being emitted into the assembly code
 by mistake. Haven't actually tested it with newer versions of GCC than 3.04. But
 when were are going to port for > 3.0.x later this year to benefit from your
 excellent cxa layer - I would for sure notify you if the problems persists - for
 now we have shifted back to 2.9x for the main branch.
 
 With best regards
 Michael Skriver, 
 IGNLabs, Designer and Management
 
 Quoting bangerth@dealii.org:
 
 > Synopsis: mangle.c #2199 (GCC3.04) - Emits *INTERNAL* when handling weak
 > defined templates con- and destructors
 > 
 > State-Changed-From-To: open->feedback
 > State-Changed-By: bangerth
 > State-Changed-When: Fri Oct 25 07:34:04 2002
 > State-Changed-Why:
 >     I cannot reproduce this problem with both gcc3.2 and mainline
 >     CVS when I follow your instructions to generate a file that
 >     shows the problem. Can you please try a newer version of
 >     gcc,and if it still fails, please provide a complete 
 >     testcase.
 > 
 > http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=6418
 > 
 
 
 

Re: middle-end/6418: mangle.c #2199 (GCC3.04) - Emits *INTERNAL* when handling weak defined templates con- and destructors

[bangerth]

Synopsis: mangle.c #2199 (GCC3.04) - Emits *INTERNAL* when handling weak defined templates con- and destructors

State-Changed-From-To: open->feedback
State-Changed-By: bangerth
State-Changed-When: Fri Oct 25 07:34:04 2002
State-Changed-Why:
    I cannot reproduce this problem with both gcc3.2 and mainline
    CVS when I follow your instructions to generate a file that
    shows the problem. Can you please try a newer version of
    gcc,and if it still fails, please provide a complete 
    testcase.

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=6418

middle-end/6418: mangle.c #2199 (GCC3.04) - Emits *INTERNAL* when handling weak defined templates con- and destructors

[msk]

>Number:         6418
>Category:       middle-end
>Synopsis:       mangle.c #2199 (GCC3.04) - Emits *INTERNAL* when handling weak defined templates con- and destructors
>Confidential:   no
>Severity:       critical
>Priority:       medium
>Responsible:    unassigned
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Tue Apr 23 04:56:02 PDT 2002
>Closed-Date:
>Last-Modified:
>Originator:     IGNLabs OSResearch - DK
>Release:        3.00 - 3.04
>Organization:
>Environment:
Linux IA32 RH6.2-
>Description:
mangle.c outputs *INTERNAL* into assembler file weak sections, when handling weak defined template constructors and destructors in line 2199 (3.04) - write_string(" *INTERNAL* "); Only con- and destructors are affected. With no support for the export keyword this flaw is fatal when building libraries which uses common templates defined in headers.
>How-To-Repeat:
Make a derivative of UX2ListNode in a c++ file, and try to compile it. Now examine the assembler output - The weak sections at the end of it will contain: 
.weak "mangled constructorname" *INTERNAL* 
.weak "mangled destructorname" *INTERNAL* 
>Fix:
The comment above line 2199 in mangle.c states that the "*INTERNAL*" information should not be placed in the resulting assembler code, but it manages to end up in the asm code somehow. A quick fix is to change *INTERNAL* to #*INTERNAL* - The "not suppossed to be there statement" will now appear as comment to the assembler - not actually a fix to the problem, but a working for the moment solution.
>Release-Note:
>Audit-Trail:
>Unformatted:
----gnatsweb-attachment----
Content-Type: application/octet-stream; name="temptest.hpp"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="temptest.hpp"

I2RlZmluZSBGVU5DX1dFQUsoeCkgeCBfX2F0dHJpYnV0ZV9fKCh3ZWFrKSkKCnRlbXBsYXRlPGNs
YXNzIE4+IGNsYXNzIFVYMkxpc3ROb2Rlewpwcml2YXRlOgogIE4qIHByZXZpb3VzTm9kZTsKICBO
KiBuZXh0Tm9kZTsKcHJvdGVjdGVkOgpwdWJsaWM6CiAgRlVOQ19XRUFLKFVYMkxpc3ROb2RlKHZv
aWQpKTsKICBGVU5DX1dFQUsodm9pZCBzZXROZXh0KE4qIE5leHQpKTsKICBGVU5DX1dFQUsodm9p
ZCBzZXRQcmV2KE4qIFByZXYpKTsKICBGVU5DX1dFQUsoTiogZ2V0TmV4dCh2b2lkKSBjb25zdCk7
CiAgRlVOQ19XRUFLKE4qIGdldFByZXYodm9pZCkgY29uc3QpOwogIEZVTkNfV0VBSyh2aXJ0dWFs
IH5VWDJMaXN0Tm9kZSgpKTsKfTs=