[ale] g++ and c++: freeing unallocated space with delete

[Joseph A Knapka]

John Wells wrote:
> I have a program that manipulates various objects on the heap.  I recently
> added a bug (unintentionally) in one of my modifications in which the
> program tries to free unallocated space.  In other words, it calls delete
> on an string pointer that hasn't been pointed anywhere.
> 
> The odd thing is, for a long time it worked just fine with no errors under
> gcc/g++ 2.96.  Recently, I had to downgrade for other reasons to 2.95 and 
> started getting seg faults, which led me to the culprit.
> 
> Why did this work before?  Do certain versions of gcc/g++ handle mistakes
> like this for you?  And, if so, is there any way to turn this behavior
> off, so that  no non-standard modifications are made?  I'd like this code
> to be as standards compliant and possible, and if there are errors on my
> part, I'd like to know about them.

In general, heap corruption is a silent bug that may or may
not cause something to crash many millions of instructions later -
a program containing such a bug has undefined behavior, and the
compiler and/or runtime environment are not obligated to notify
you of the problem at the point where it occurs. If
you're getting a SEGV on an attempt to free unallocated memory, that
seems to indicate that the heap library you're using (presumably
the one that comes with libstdc++ in g++2.95) is doing some extra
checking for you that the 2.96 version doesn't do. But that's
just a Scientific Wild-Ass Guess.

Cheers,

-- Joe


---
This message has been sent through the ALE general discussion list.
See http://www.ale.org/mailing-lists.shtml for more info. Problems should be 
sent to listmaster at ale dot org.