Re: linker question: rtti/exception functions

[querbach at realtime dot bc dot ca (David Querbach)]

> What I don't understand is the reason for this reference to __builtin_delete.
> What puzzles me is that an empty destructor invokes (or at least somehow
> references) a memory management function. I do not use new() or delete()
> in my test example; I simply declared a object in global scope.

It's because the compiler needs to provide for the ability to delete an
object which was created using operator new.  Gcc handles this by passing a
hidden argument to the destructor which says whether or not ::operator
delete() needs to be called to free memory at the end of the destructor.

Creating objects on the heap is so fundamental to C++ that there is no
compiler option to defeat this code (that I know of).  The only way to avoid
it is to declare all destructors inline, so that the call to ::operator
delete() can be optimized out.

If that's not acceptable, you could just stub out the offending functions:

  void operator delete(void*) { } 

  void operator delete[](void*) { }

If you look at the resulting assembler code, you'll find that these
functions are actually called __builtin_delete and __builtin_vec_delete at
the assembler level.

Regards,

David Querbach
Real-Time Systems Inc.

------
Want more information?  See the CrossGCC FAQ, http://www.objsw.com/CrossGCC/
Want to unsubscribe? Send a note to crossgcc-unsubscribe@sources.redhat.com