Frama-C-discuss mailing list archives

This page gathers the archives of the old Frama-C-discuss archives, that was hosted by Inria's gforge before its demise at the end of 2020. To search for mails newer than September 2020, please visit the page of the new mailing list on Renater.

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Frama-c-discuss] Proof conditions with simple pointer assignment

Sorry,
I made a confusion since "p" is a formal parameter.
Of course, these two clauses (assigns and ensures) hold.

/*@ requires \valid(p);
  @ assigns *p;
  @ ensures *p == 1;
 */
 void f(int *p) {
     *p = 1;
 }

That specification is equivalent to :
/*@ requires \valid(p);
  @ assigns *\old(p);
  @ ensures *\old(p) == 1;
 */
 void f(int *p) {
     *p = 1;
 
}

Now let consider "q" a global variable :
int * q ;

This specification holds :
/*@ requires \valid(q);
  @ assigns *\old(q);
  @ ensures *\old(q) == 1;
 */
 void g(void) {
     *g = 1;
 }

But without the use \old construction, something is required about 
separation to satisfy the ensures clauses.
/*@ requires \valid(q) && something about separation
  @ assigns *q;
  @ ensures *q == 1;
 */
 void g(void) {
     *g = 1;
 }

Patrick Baudin.

> 
> Hello,
> Just from an ACSL point of view, the proof of the "assigns"
> and "ensures" clauses of your spec requires locations "p"
> and "*p" to be \separated.
> Patrick Baudin.
> > Consider the following spec,
> >
> > /*@ requries \valid(p);
> >    @ assigns *p;
> >    @ ensures *p == 1;
> > */
> > void f(int *p)
> > {
> >     *p = 1;
> > }