[SCM] New address sanitizer option

[Jeremy Allison]

On Thu, Jan 29, 2015 at 07:07:11AM -0500, Ira Cooper wrote:
> 
> I was considering a far "harsher" option.  Because almost every tool of
> this type has some form of a hole, initializing ALL variables in the
> codebase.
> 
> It is much simpler to see that a variable is initialized, than "did we get
> this one.. and this one... what about this strange corner case..."
> 
> Meh... get 'em all!  If there is a performance issue, let's address it
> AFTER with, patches that loosen the restriction and have comments.

+1 from me, but belt-and-braces IMHO. All the address sanitizer
work can go on in parallel with investigating initializing
everything.