c++/3145: virtual inheritance still creates wrong code

c++/3145: virtual inheritance still creates wrong code

[stefan]

>Number:         3145
>Category:       c++
>Synopsis:       virtual inheritance still creates wrong code
>Confidential:   no
>Severity:       serious
>Priority:       high
>Responsible:    unassigned
>State:          open
>Class:          wrong-code
>Submitter-Id:   net
>Arrival-Date:   Tue Jun 12 14:46:00 PDT 2001
>Closed-Date:
>Last-Modified:
>Originator:     stefan@space.twc.de
>Release:        gcc version 3.0 20010612 (prerelease)
>Organization:
>Environment:

>Description:
aRts still doesn't work with gcc3 (segfault on start). So
I thought instead of generating testcases one by one all
the time, which always opens the possibility that some
cases work but others break, I wrote a testcase generator
for the set of testcases that is just a complex (partially
virtual) inheritance graph and segfaults on execution.

I found that with the current gcc3 a good share of the
generated testcases segfaults when compiled and executed,
whereas with gcc-2.95.2, none of these sefaults.
>How-To-Repeat:
make tcgen
for i in $(seq 1 1000)
do
  tcgen $i > t.cc ; g++ 2>/dev/null -O2 t.cc ; a.out || echo $i failed
done
>Fix:

>Release-Note:
>Audit-Trail:
>Unformatted:
----gnatsweb-attachment----
Content-Type: text/x-c++src; name="tcgen.cc"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="tcgen.cc"
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Re: c++/3145: virtual inheritance still creates wrong code

[nathan]

The following reply was made to PR c++/3145; it has been noted by GNATS.

From: nathan@gcc.gnu.org
To: gcc-bugs@gcc.gnu.org, gcc-gnats@gcc.gnu.org, gcc-prs@gcc.gnu.org,
  nathan@gcc.gnu.org, stefan@space.twc.de
Cc:  
Subject: Re: c++/3145: virtual inheritance still creates wrong code
Date: 25 Nov 2001 13:14:00 -0000

 Synopsis: virtual inheritance still creates wrong code
 
 State-Changed-From-To: analyzed->closed
 State-Changed-By: nathan
 State-Changed-When: Sun Nov 25 05:14:00 2001
 State-Changed-Why:
     fixed on mainline
     2001-11-25  Nathan Sidwell  <nathan@codesourcery.com>
     
     this patch is too big to consider for 3.0.3 :-(
 
 http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&pr=3145&database=gcc

Re: c++/3145: virtual inheritance still creates wrong code

[nathan]

Synopsis: virtual inheritance still creates wrong code

State-Changed-From-To: analyzed->closed
State-Changed-By: nathan
State-Changed-When: Sun Nov 25 05:14:00 2001
State-Changed-Why:
    fixed on mainline
    2001-11-25  Nathan Sidwell  <nathan@codesourcery.com>
    
    this patch is too big to consider for 3.0.3 :-(

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&pr=3145&database=gcc

Re: c++/3145: virtual inheritance still creates wrong code

[nathan]

Synopsis: virtual inheritance still creates wrong code

Responsible-Changed-From-To: unassigned->nathan
Responsible-Changed-By: nathan
Responsible-Changed-When: Tue Jul 17 07:34:59 2001
Responsible-Changed-Why:
    more important stuff

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view&pr=3145&database=gcc

Re: c++/3145: virtual inheritance still creates wrong code

[nathan]

Synopsis: virtual inheritance still creates wrong code

Responsible-Changed-From-To: nathan->unassigned
Responsible-Changed-By: nathan
Responsible-Changed-When: Thu Jun 14 03:59:06 2001
Responsible-Changed-Why:
    Partially fixed. remaining faults in build_vbase_path, which
    is longstanding.
    
    2001-06-14  Nathan Sidwell  <nathan@codesourcery.com>
    
    	* class.c (dfs_accumulate_vtbl_inits): For case 2 & 3, make sure
    	we've not emerged from the hierarchy of RTTI_BINFO on reaching
    	a non-virtual base.
    
    
    2001-06-14  Nathan Sidwell  <nathan@codesourcery.com>
    
    	* g++.old-deja/g++.abi/vbase8-9.C: New test.

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view&pr=3145&database=gcc

Re: c++/3145: virtual inheritance still creates wrong code

[nathan]

Synopsis: virtual inheritance still creates wrong code

Responsible-Changed-From-To: unassigned->nathan
Responsible-Changed-By: nathan
Responsible-Changed-When: Wed Jun 13 01:47:53 2001
Responsible-Changed-Why:
    oh, yes gimme more
State-Changed-From-To: open->analyzed
State-Changed-By: nathan
State-Changed-When: Wed Jun 13 01:47:53 2001
State-Changed-Why:
    ow shucks!

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view&pr=3145&database=gcc