Re: [PATCH]: Fix PR tree-optimization/21407

[Paolo Bonzini <paolo dot bonzini at lu dot unisi dot ch>]

> There's no casting between pointers and integers.  There are 
> reinterpet_casts between pointers.  The mapping is 
> implementation-defined, but, in practice, all implementations on, say, 
> IA32, are going to leave the bit-pattern unchanged.

That's not my point:

A  pointer to an object can be explicitly converted to a pointer to an 
object  of different type. Except that converting an rvalue of type 
"pointer to T1" to the type "pointer to  T2"  (where  T1  and  T2  are 
object  types  and  where  the  alignment  requirements  of  T2 are no 
stricter than those of T1) and back to its original  type  yields  the 
original  pointer  value,  the  result of such a pointer conversion is 
unspecified.

> I just don't see anything that says conclusively that this code is 
> invalid.

Anything but casting back the pointer to the original type is 
implementation-defined.  That's also what Microsoft's MSDN documentation 
says (http://snipurl.com/ettu): "The result of a reinterpret_cast cannot 
safely be used for anything other than being cast back to its original 
type. Other uses are, at best, nonportable. [...] One practical use of 
reinterpret_cast is in a hash function, which maps a value to an index 
in such a way that two distinct values rarely end up with the same 
index, [where] the reinterpret_cast allows the pointer to be treated as 
an integral type."

So MSDN says that arithmetic is only useful after pointer-to-integer 
casts and the MSDN example does *not* cast anything back to a pointer.  
Regarding pointer-to-integer casts, the standard says the same: "[the 
result of reinterpret_cast of a pointer to integer] is intended to be 
unsurprising to those who know the addressing structure of the 
underlying machine."  But it very clear in saying that "A pointer 
converted to an integer of sufficient size (if any such exists on the 
implementation) and back to the same pointer type will have its original 
value; mappings between pointers and integers are otherwise 
implementation-defined."

Paolo