[Bug c++/60213] New: Weird crash when delete an object

[Bug c++/60213] New: Weird crash when delete an object

[hubert.vansteenhuyse at freecode dot be]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60213

            Bug ID: 60213
           Summary: Weird crash when delete an object
           Product: gcc
           Version: 4.6.3
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: hubert.vansteenhuyse at freecode dot be

[Bug c++/60213] Weird crash when delete an object

[hubert.vansteenhuyse at freecode dot be]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60213

--- Comment #1 from hubert.vansteenhuyse at freecode dot be ---
struct A {
 A *a[2];
 A() { a[0] = 0; a[1] = 0; }
 ~A(){ if(a[0]) delete a[0];  if(a[1]) delete a[1];
}

void main(){
A *a = new A();
a->a[0] = new A();
a->a[1] = new A();

delete a; //<----<< crash, because adress of a == address of a.a[0] :(
}

[Bug c++/60213] Weird crash when delete an object

[glisse at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60213

Marc Glisse <glisse at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |WAITING
   Last reconfirmed|                            |2014-02-15
     Ever confirmed|0                           |1

--- Comment #2 from Marc Glisse <glisse at gcc dot gnu.org> ---
Your example is bogus: { and } don't match, main must return int, etc. If I fix
it so it compiles, I don't see any crash.

[Bug c++/60213] Weird crash when delete an object

[hubert.vansteenhuyse at freecode dot be]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60213

--- Comment #3 from hubert.vansteenhuyse at freecode dot be ---
(In reply to Marc Glisse from comment #2)
> Your example is bogus: { and } don't match, main must return int, etc. If I
> fix it so it compiles, I don't see any crash.

Of course it is, the example is merely a pseudo code to make clear what
happened.
In any case GDB reported a double deletion of the same address in similar code.
Don't worry about it, if someone else ever reports it then it might be worth to
take a closer look to it. 
Thx.

[Bug c++/60213] Weird crash when delete an object

[glisse at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60213

--- Comment #4 from Marc Glisse <glisse at gcc dot gnu.org> ---
(In reply to hubert.vansteenhuyse from comment #3)
> Of course it is, the example is merely a pseudo code to make clear what
> happened.

That's my point, pseudo-code is useless, if we can't reproduce the issue it
doesn't make it clear at all what is going on. Compiler bugs (in case this is
one) often have unpredictable symptoms.

> Don't worry about it, if someone else ever reports it then it might be worth
> to take a closer look to it.

But we will have no idea that it is related. If you can post a valid,
self-contained example that exhibits the issue (preferably with a supported
version of gcc, so at least 4.7), that would be helpful. Otherwise I'll just
close the PR.

[Bug c++/60213] Weird crash when delete an object

[redi at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60213

--- Comment #5 from Jonathan Wakely <redi at gcc dot gnu.org> ---
(In reply to hubert.vansteenhuyse from comment #3)
> Of course it is, the example is merely a pseudo code to make clear what
> happened.

Didn't you read http://gcc.gnu.org/bugs/ ?

[Bug c++/60213] Weird crash when delete an object

[hubert.vansteenhuyse at freecode dot be]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60213

hubert.vansteenhuyse at freecode dot be changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|WAITING                     |RESOLVED
         Resolution|---                         |FIXED

--- Comment #6 from hubert.vansteenhuyse at freecode dot be ---
Yes I did. I think the crash came from somehere else in my code. I really
should be more carefull before reporting a bug. I thank you for your time spent
on this.