Re: [Bug libstdc++/38107] gcc source contains a struct with no data members (actually 1 byte in size) and compiler does not initialize it, resulting in IBM Rational Purify reporting an Uninitialized Memory Read.

[Andrew Thomas Pinski <pinskia at gmail dot com>]

Sent from my iPhone

On Nov 14, 2008, at 9:05 AM, "efinger at us dot ibm dot com" <gcc-bugzilla@gcc.gnu.org 
> wrote:

> ------- Comment #4 from efinger at us dot ibm dot com  2008-11-14  
> 17:05 -------
> (In reply to comment #2)
> > There are going to be other holes in structs in general due to  
> > alignment
> > requirements and I don't think the compiler or libstdc++ should  
> > change to this
> > because it will slow down the generated code and not really makes a  
> > difference
> > in the outcome of the code.
>
> Purify already handles pad bytes added for alignment correctly.   
> This is a very
> specific case, where the pad byte is added to prevent the struct  
> from having
> size zero, not for alignment.  The added code to clear 1 byte, at  
> object
> constuction time, would not make any significant performance  
> difference.

But it can cause code bloat which can cause performance issues because  
of icache sizes and such.

> --
>
>
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38107