Re: volatile [was: GCC warnings for unused global variables]

[Fergus Henderson <fjh at cs dot mu dot OZ dot AU>]

On 08-May-2003, Gabriel Dos Reis <gdr@integrable-solutions.net> wrote:
> Fergus Henderson <fjh@cs.mu.oz.au> writes:
> | The C standard text which you quoted doesn't say that the optimization
> | is permitted unconditionally.  It only says that the optimization is
> | permitted only "if [the implementation] can deduce that [the expression's]
> | value is not used and that no needed side effects are produced".  In the
> | case of volatile variables, this condition is always false,
> 
> No, because the standard does specifically add
> 
>   INCLUDING ANY CAUSED BY CALLING A FUNCTION OR ACCESSING A VOLATILE
>   OBJECT 

Yes, it says that, but that is just to remind the reader that
accessing a volatile object is a side effect.

You are assuming that the set of side effects produced by accessing
volatile variables which the implementation can deduce are not needed
is a non-empty set.  This is not the case.  The standard specifically
says so in 6.7.3 paragraph 6 (quoted again below).

> The definition of "volatile" does not say that the implementation can
> never know if a side effect is needed.

I firmly believe that was the intent behind these words:

 | An object that has volatile-qualified type may be modified
 | in ways unknown to the implementation or have other unknown
 | side effects.  Therefore any expression referring to such
 | an object shall be evaluated strictly according to the
 | rules of the abstract machine, as described in 5.1.2.3.

(This belief is not based on any specific statements of the C committee
members, but rather on the general principles that they're not idiots
and that they intended the words that they wrote to be meaningful.)

-- 
Fergus Henderson <fjh@cs.mu.oz.au>  |  "I have always known that the pursuit
The University of Melbourne         |  of excellence is a lethal habit"
WWW: <http://www.cs.mu.oz.au/~fjh>  |     -- the last words of T. S. Garp.