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] Help with proving post-conditions
- Subject: [Frama-c-discuss] Help with proving post-conditions
- From: mtorl at hotmail.com (Murat)
- Date: Thu, 10 Dec 2009 14:38:29 +0000
- In-reply-to: <B259434F-E204-437F-94B5-409730BD0635@cea.fr>
- References: <SNT110-W47E157EBE9105580EE97AEA78E0@phx.gbl> <B259434F-E204-437F-94B5-409730BD0635@cea.fr>
Hi again, I've made the recommended changes but I still can't prove the 3 of the post-conditions although everything else is proven with Z3 including the asserts. I guess my loop invariants for the second loop are the problem as I find it hard to express the null character being copied with the way the loop is written. I'd appreciate your help with this. /*@ requires valid_string(s); requires valid_string(append); requires \valid_range(s, 0, strlen(s) + strlen(append)); assigns s[..]; ensures strlen(s) == strlen{Old}(s) + strlen(append); ensures \forall integer i; 0 <= i < strlen{Old}(s) ==> s[i] == \old(s[i]); ensures \forall integer j; strlen{Old}(s) <= j <= strlen(s) ==> s[j] == append[j]; ensures \result == s; */ char * strcat(char *s, const char *append) { char *save = s; /*@ loop assigns s; loop invariant 0 <= (s-save) <= strlen(save); loop invariant \valid(s); loop invariant \base_addr(s) == \base_addr(save); loop invariant \forall integer k; 0 <= k < (s-save) ==> save[k] != 0; */ for (; *s; ++s); //@ assert *s == '\0' && s == save + strlen(save); //@ assert \valid_range(s, 0, strlen(append)); //@ ghost char *s_cat = s; //@ ghost char *origAppend = append; /*@ loop assigns s, s_cat[..], append; loop invariant \base_addr(s) == \base_addr(s_cat); loop invariant \base_addr(append) == \base_addr(origAppend); loop invariant \valid_range(origAppend, 0, strlen{Pre}(append)); loop invariant \valid_range(s_cat, 0, strlen{Pre}(append)); loop invariant (append-origAppend) == (s - s_cat); loop invariant 0 <= (append-origAppend) <= strlen(origAppend); loop invariant \forall integer k; 0 <= k < (append-origAppend) < strlen(origAppend) ==> origAppend[k] != 0; loop invariant \forall integer k; 0 <= k < (append-origAppend) ==> s_cat[k] == origAppend[k]; */ while ((*s++ = *append++) != '\0'); //@ assert s_cat[append-origAppend - 1] == 0; //@ assert strlen(s_cat) == append-origAppend -1; //@ assert strlen(origAppend) == append-origAppend -1; //@ assert strlen(s_cat) == strlen(origAppend); return(save); } Thanks, Murat. From: frama-c-discuss-bounces at lists.gforge.inria.fr [mailto:frama-c-discuss-bounces at lists.gforge.inria.fr] On Behalf Of Pascal Cuoq Sent: Wednesday, December 09, 2009 12:26 PM To: Frama-C public discussion Subject: [Frama-c-discuss] Help with proving post-conditions Hello, ensures strlen(s) == strlen(\at(s, Old)) + strlen(append); Which strlen function are you trying to use here? If you mean the C function, you can't: it is not allowed to call a C function from inside a predicate such as a pre/prost-condition. You can have a logic function (I would recommend to use a slight variation on the strlen name), defined by yourself as the function that computes the length of a string. That is the right way to do it. Another thing: \at(s, Old) (that you can write \old(s)) does not do what you think it does. In fact, since strcat does not (in general, unless the caller created some strange aliasing conditions) modify its argument s, \old(s) is the same as s. What the logic function logic_strlen really takes in argument is a pointer s *and* a memory state in which s points to some bytes, and with this information, it is able to tell you what the length of s is *in the memory state that you provided*. Function logic_strlen only needs one memory state, so this memory state can remain implicit in the body of the definition, but you have to be aware that it is an argument of the function, and that you need to pass it. In this case you would pass Old i.e. write something like logic_strlen{Old}(s) instead of strlen(\at(s, Old)). Perhaps the explanations in chapter 8 of the ACSL mini-tutorial http://frama-c.cea.fr/download/acsl-tutorial.pdf can be of help. Pascal -------------- next part -------------- A non-text attachment was scrubbed... Name: winmail.dat Type: application/ms-tnef Size: 9198 bytes Desc: not available Url : http://lists.gforge.inria.fr/pipermail/frama-c-discuss/attachments/20091210/c6cfc96f/attachment-0001.bin
- Follow-Ups:
- [Frama-c-discuss] Help with proving post-conditions
- From: Pascal.Cuoq at cea.fr (Pascal Cuoq)
- [Frama-c-discuss] Help with proving post-conditions
- From: Pascal.Cuoq at cea.fr (Pascal Cuoq)
- [Frama-c-discuss] Help with proving post-conditions
- References:
- [Frama-c-discuss] Help with proving post-conditions
- From: mtorl at hotmail.com (Murat Torlakcik)
- [Frama-c-discuss] Help with proving post-conditions
- From: Pascal.Cuoq at cea.fr (Pascal Cuoq)
- [Frama-c-discuss] Help with proving post-conditions
- Prev by Date: [Frama-c-discuss] problem with verify a list
- Next by Date: [Frama-c-discuss] Help with proving post-conditions
- Previous by thread: [Frama-c-discuss] Help with proving post-conditions
- Next by thread: [Frama-c-discuss] Help with proving post-conditions
- Index(es):