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] Problem with loop invariant

  • Subject: [Frama-c-discuss] Problem with loop invariant
  • From: hollas at informatik.htw-dresden.de (Boris Hollas)
  • Date: Mon, 28 Jun 2010 14:32:52 +0200
  • In-reply-to: <14634911.194632.1277726661282.JavaMail.www@wwinf8302>
  • References: <1276795121.1681.16.camel@iti27> <4C1F2AF5.3000604@inria.fr> <1277298042.2073.36.camel@iti27> <14634911.194632.1277726661282.JavaMail.www@wwinf8302>
On Mon, 2010-06-28 at 14:04 +0200, n.rousset at laposte.net wrote:
> The loop invariant after the loop body is indeed equivalent to the
> assertion,
> but the invariant is checked AFTER the assertion...

what does this explain?

> To find out why the assertion is not proved,
> we need to see the postconditions of sreadi(), abpRound(), and
> swritei().

Here again is the loop together with the assertion that doesn't verify:

/*@
	loop variant MSGCOUNT - internal.msgnum;
	loop invariant 0 <= internal.msgnum &&
		\forall integer x; 0 <= x < internal.msgnum * BUFFSIZE ==>
tosend->data[x] == received->data[x];
 */
for (internal.msgnum = 0;internal.msgnum < MSGCOUNT; internal.msgnum++)
{
	sreadi(msg_send,tosend,internal.msgnum);
	abpRound(msg_send,msg_receive,&channel,&internal);
	swritei(received,msg_receive,internal.msgnum);
	//@ assert \forall integer x; 0 <= x < (internal.msgnum+1) * BUFFSIZE
==> tosend->data[x] == received->data[x];
}


The contracts for sreadi, abpRound, swritei are below. We use a
twodimensional array "data" because Frama-C can't handle
multidimensional arrays yet.

The contract for sreadi is:


/*@
requires \valid(buf);
requires \valid(buf->data+(0.. BUFFSIZE*MSGCOUNT));
requires \valid(target+(0.. BUFFSIZE));
requires 0 <= idx < MSGCOUNT;

ensures \forall int x; 0 <= x < BUFFSIZE ==> target[x] ==
buf->data[idx*BUFFSIZE + x];
ensures \forall int x; 0 <= x < BUFFSIZE*MSGCOUNT ==> buf->data[x] ==
\old(buf->data[x]);
ensures target[BUFFSIZE] == 0;
 */
void sreadi(char* target, DataBuffer* buf, int idx)


The contract for abpRound is:


/*@
requires \valid(msg_send+(0.. BUFFSIZE));
requires \valid(msg_receive+(0.. BUFFSIZE));
requires \valid(channel);
requires \valid(channel->msgChannel+(0.. BUFFSIZE+1));
requires \valid(internal);

ensures \forall int x; 0 <= x < BUFFSIZE ==> msg_receive[x] ==
msg_send[x];
 */
void abpRound(char *msg_send, char *msg_receive, ChannelType *channel,
InternalType *internal) 


The contract for swritei is:


/*@
requires \valid(buf);
requires \valid(buf->data+(0.. BUFFSIZE*MSGCOUNT));
requires \valid(source+(0.. BUFFSIZE));
requires 0 <= idx < MSGCOUNT;

ensures \forall int x; 0 <= x < BUFFSIZE ==> buf->data[idx*BUFFSIZE + x]
== source[x];
ensures \forall int x; 0 <= x < idx*BUFFSIZE ==> buf->data[x] ==
\old(buf->data[x]);
ensures buf->data[BUFFSIZE*MSGCOUNT] == 0;
 */
void swritei(DataBuffer* buf, char* source, int idx) 


BTW, on another computer also using Boron, the loop invariant above
didn't verify.

-- 
Regards,
Boris