public inbox for gcc-prs@sourceware.org
help / color / mirror / Atom feed
From: abramsh@acm.org
To: gcc-gnats@gcc.gnu.org
Subject: c++/3993: dynamic_cast<>() fails in simple test case w/ dynamic loading
Date: Fri, 10 Aug 2001 22:56:00 -0000	[thread overview]
Message-ID: <20010811055240.27212.qmail@sourceware.cygnus.com> (raw)

>Number:         3993
>Category:       c++
>Synopsis:       dynamic_cast<>() fails in simple test case w/ dynamic loading
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    unassigned
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Fri Aug 10 22:56:00 PDT 2001
>Closed-Date:
>Last-Modified:
>Originator:     Howard Abrams
>Release:        gcc version 3.0
>Organization:
>Environment:
Linux blah.voidstar.org 2.4.2-2 #1 Sun Apr 8 20:41:30 EDT 2001 i686 unknown
on i686-pc-linux-gnu
>Description:
When using "dlopen" to load a library, gcc seems to NOT merge
type_info symbols from the library. The result is dynamic
casts fail, as well as type_info::operator== for object that
they should not fail on. Below is simple source to reproduce
this problem.

Also note that on version 2.96 (Red Hat Linux 7.1 2.96-81)
(I know, you don't support that... ) this works fine.

>How-To-Repeat:
To reproduce this there are three small files. Compile with
the following two lines:

g++ -g -o test2.so -shared test2.cpp
g++ -g -o testcase test.cpp -ldl


The files are as follows:

classb.h
========
struct a
{
        virtual ~a() {}
};

struct b : public a
{
        virtual ~b() {}
};

test2.cpp
=========
#include <typeinfo>

#include "classb.h"

extern "C" {

a *getnewb() { return new b; }
const std::type_info &gettypeb() { return typeid( b ); }

}

test.cpp
========
#include <dlfcn.h>
#include <typeinfo>
#include <cassert>

#include "classb.h"

int main()
{
        a *(*func)(void);
        const std::type_info &(*type)(void);

        void *handle = dlopen( "test2.so", RTLD_NOW );
        func = (a *(*)(void))dlsym( handle, "getnewb" );
        type = (const std::type_info &(*)(void))dlsym( handle, "gettypeb" );

        a *a_ptr = (*func)();
        const std::type_info &t = (*type)();

        assert ( dynamic_cast<b*>( a_ptr ) );
        assert ( typeid( b ) == t );
}
>Fix:

>Release-Note:
>Audit-Trail:
>Unformatted:


             reply	other threads:[~2001-08-10 22:56 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2001-08-10 22:56 abramsh [this message]
2001-11-08 21:06 sirl
2001-11-08 22:56 sirl
2001-12-04  6:55 nathan
2001-12-04  6:56 nathan

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20010811055240.27212.qmail@sourceware.cygnus.com \
    --to=abramsh@acm.org \
    --cc=gcc-gnats@gcc.gnu.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).