[vox-tech] another gcc question

Matt Roper vox-tech@lists.lugod.org
Wed, 27 Feb 2002 12:15:31 -0800


On Wed, Feb 27, 2002 at 09:46:29AM -0800, Peter Jay Salzman wrote:
...
> the reason why i'm asking is that this is completely deterministic, and
> a compiler can know, with certainty, that this loop will iterate over 0
> and 1.  that is, you can know this at compile time, before executing any
> code.
> 
> clearly, this ISN'T deterministic:
> 
> int n=2;
> MyFunction(&n);
> for (i=0; i<n; ++i) 
> 
> but certainly, this is:
> 
> int n=2;
> for (i=0; i<n; ++i) 

What about this case:

    int n = 2;
    for(i = 0; i < n; i++) {
        printf("i=%d\n", i);
        n = 10;
    }

In this case, the code should print out i from 0 to 9.  However if the
compiler tried to unroll this loop, it might not be smart enough to
detect the change in j and would only unroll the loop twice.  More
complex for loop bodies would make this even harder to figure out at
compile time.  And of course if the program is multithreaded or uses
signal handlers there is always the possibility that other code could
run between the j=2 statement and the loop.  I don't think the
compiler can make the assumption that the loop only runs twice.


Matt

-- 

*************************************************
* Matt Roper <matt@mattrope.com>                *
* http://www.mattrope.com                       *
* PGP Key: http://www.mattrope.com/mattrope.asc *
*************************************************