Differences
This shows you the differences between two versions of the page.
Both sides previous revision Previous revision Next revision | Previous revision | ||
sav07_lecture_3_skeleton [2007/03/21 09:37] vkuncak |
sav07_lecture_3_skeleton [2007/03/21 14:27] (current) vkuncak |
||
---|---|---|---|
Line 1: | Line 1: | ||
====== Lecture 3 (Skeleton) ====== | ====== Lecture 3 (Skeleton) ====== | ||
- | ===== Converting programs (with simple values) to formulas ===== | + | Summary of what we are doing in today's class: |
+ | {{vcg-big-picture.png}} | ||
+ | |||
+ | |||
+ | ===== Verification condition generation: converting programs into formulas ===== | ||
==== Context ==== | ==== Context ==== | ||
Line 11: | Line 15: | ||
* we can represent relations using set comprehensions; if our program c has two state components, we can represent its meaning R( c ) as $\{((x_0,y_0),(x,y)) \mid F \}$, where F is some formula that has x,y,x_0,y_0 as free variables. | * we can represent relations using set comprehensions; if our program c has two state components, we can represent its meaning R( c ) as $\{((x_0,y_0),(x,y)) \mid F \}$, where F is some formula that has x,y,x_0,y_0 as free variables. | ||
- | * this is what I mean by ''simple values'': later we will talk about modeling pointers and arrays, but we will still use this as a starting point. | + | * simple values: variables are integers. Later we will talk about modeling pointers and arrays, but what we say now applies |
Our goal is to find rules for computing R( c ) that are | Our goal is to find rules for computing R( c ) that are | ||
Line 24: | Line 28: | ||
R( c ) -> error=false | R( c ) -> error=false | ||
+ | |||
Line 37: | Line 42: | ||
R(havoc x) = frame(x) | R(havoc x) = frame(x) | ||
- | R(assume F) = F[x:=x_0, y:=y_0, error:=error_0] | + | R(assume F) = F[x:=x_0, y:=y_0, error:=error_0] & frame() |
R(assert F) = (F -> frame) | R(assert F) = (F -> frame) | ||
Line 102: | Line 107: | ||
This idea is important in static analysis. | This idea is important in static analysis. | ||
+ | |||
+ | |||
+ | |||
+ | |||
+ | |||
Line 107: | Line 117: | ||
==== Symbolic execution ==== | ==== Symbolic execution ==== | ||
- | Symbolic execution converts programs into formulas by going forward. It is therefore somewhat analogous to the way an [[interpreter]] for the language would work. It is based on the notion of strongest postcondition. | + | Symbolic execution converts programs into formulas by going forward. It is therefore somewhat analogous to the way an [[interpreter]] for the language would work. |
+ | Avoid renaming all the time. | ||
+ | |||
+ | SE(F,k, c1; c2) = SE(F & R(c1), k+1, c2) (update formula) | ||
+ | |||
+ | SE(F,k,(c1 [] c2); c2) = SE(F, k, c1) | SE(F,k,c2) (explore both branches) | ||
+ | |||
+ | Note: how many branches do we get? | ||
+ | |||
+ | Strongest postcondition: | ||
+ | \begin{equation*} | ||
+ | sp(P,r) = \{ s_2 \mid \exists s_1.\ s_1 \in P \land (s_1,s_2) \in r \} | ||
+ | \end{equation*} | ||
+ | Like composition of a set with a relation. It's called ''relational image'' of set $P$ under relation $r$. | ||
+ | Note: when proving our verification condition, instead of proving that semantics of relation implies error=false, it's same as proving that the formula for set sp(U,r) implies error=false, where U is the universal relation, or, in terms of formulas, computing the strongest postcondition of formula 'true'. | ||
==== Weakest preconditions ==== | ==== Weakest preconditions ==== | ||
While symbolic execution computes formula by going forward along the program syntax tree, weakest precondition computes formula by going backward. | While symbolic execution computes formula by going forward along the program syntax tree, weakest precondition computes formula by going backward. | ||
+ | |||
+ | wp(Q, x=t) = | ||
+ | wp(Q, assume F) = | ||
+ | wp(Q, assert F) = | ||
+ | wp(Q, c1 [] c2) = | ||
+ | wp(Q, c1 ; c2) = | ||
==== Inferring Loop Invariants ==== | ==== Inferring Loop Invariants ==== | ||
Line 129: | Line 159: | ||
- | ===== Proving quantifier-free linear arithmetic formulas ===== | + | |
+ | |||
+ | |||
+ | ===== One useful decision procedure: Proving quantifier-free linear arithmetic formulas ===== | ||
Suppose that we obtain (one or more) verification conditions of the form | Suppose that we obtain (one or more) verification conditions of the form | ||
+ | \begin{equation*} | ||
+ | F\ \rightarrow\ (\mbox{error}=\mbox{false}) | ||
+ | \end{equation*} | ||
+ | |||
+ | whose validity we need to prove. We here assume that F contains only linear arithmetic. Note: we can check satisfiability of $F\ \land\ (\mbox{error}=\mbox{true})$. We show an algorithm to check this satisfiability. | ||
==== Quantifier Presburger arithmetic ==== | ==== Quantifier Presburger arithmetic ==== | ||
Line 141: | Line 179: | ||
T ::= var | T + T | K * T (terms) | T ::= var | T + T | K * T (terms) | ||
A ::= T=T | T <= T (atomic formulas) | A ::= T=T | T <= T (atomic formulas) | ||
- | F ::= F & F | F|F | ~F (formulas) | + | F ::= A | F & F | F|F | ~F (formulas) |
To get full Presburger arithmetic, allow existential and universal quantifiers in formula as well. | To get full Presburger arithmetic, allow existential and universal quantifiers in formula as well. | ||
Line 151: | Line 189: | ||
Proof: small model theorem. | Proof: small model theorem. | ||
- | ==== Small model theorem for quantifier-free Presburger arithmetic ==== | + | ==== Small model theorem for Quantifier-Free Presburger Arithmetic (QFPA) ==== |
First step: transform to disjunctive normal form. | First step: transform to disjunctive normal form. | ||
Line 157: | Line 195: | ||
Next: reduce to integer linear programming: | Next: reduce to integer linear programming: | ||
\begin{equation*} | \begin{equation*} | ||
- | Ax = b, x \geq 0 | + | A\vec x = \vec b, \qquad \vec x \geq \vec 0 |
\end{equation*} | \end{equation*} | ||
where $A \in {\cal Z}^{m,n}$ and $x \in {\cal Z}^n$. | where $A \in {\cal Z}^{m,n}$ and $x \in {\cal Z}^n$. | ||
- | Then use small model theorem for integer linear programming. | + | Then solve integer linear programming (ILP) problem |
+ | * [[wk>Integer Linear Programming]] | ||
+ | * online book chapter on ILP | ||
+ | * [[http://www.gnu.org/software/glpk/|GLPK]] tool | ||
- | Short proof by | + | We can prove small model theorem for ILP - gives bound on search. |
+ | |||
+ | Short proof by {{papadimitriou81complexityintegerprogramming.pdf|Papadimitriou}}: | ||
+ | * solution of Ax=b (A regular) has as components rationals of form p/q with bounded p,q | ||
+ | * duality of linear programming | ||
+ | * obtains bound $M = n(ma)^{2m+1}$, which needs $\log n + (2m+1)\log(ma)$ bits | ||
+ | * we could encode the problem into SAT: use circuits for addition, comparison etc. | ||
+ | |||
+ | Note: if small model theorem applies to conjunctions, it also applies to arbitrary QFPA formulas. | ||
+ | |||
+ | Moreover, one can improve these bounds. One tool based on these ideas is [[http://www.cs.cmu.edu/~uclid/|UCLID]]. | ||
+ | |||
+ | Alternative: enumerate disjuncts of DNF on demand, each disjunct is a conjunction, then use ILP techniques (often first solve the underlying linear programming problem over reals). Many SMT tools are based on this idea (along with Nelson-Oppen combination: next class). | ||
+ | * [[http://www.cs.nyu.edu/acsys/cvc3/download.html|CVC3]] (successor of CVC Lite) | ||
+ | * [[http://combination.cs.uiowa.edu/smtlib/|SMT-LIB]] Standard for formulas, competition | ||
- | Tools: | ||
- | * [[http://www.cs.cmu.edu/~uclid/|UCLID]] | ||
==== Full Presburger arithmetic ==== | ==== Full Presburger arithmetic ==== | ||
Full Presburger arithmetic is also decidable. | Full Presburger arithmetic is also decidable. | ||
+ | |||
+ | Approaches: | ||
+ | * Quantifier-Elimination (Omega tool from Maryland) - see homework | ||
+ | * Automata Theoretic approaches: LASH, MONA (as a special case) | ||
===== Papers ===== | ===== Papers ===== | ||
Line 180: | Line 237: | ||
* Presburger Arithmetic (PA) bounds: {{papadimitriou81complexityintegerprogramming.pdf}} | * Presburger Arithmetic (PA) bounds: {{papadimitriou81complexityintegerprogramming.pdf}} | ||
* Specializing PA bounds: http://www.lmcs-online.org/ojs/viewarticle.php?id=43&layout=abstract | * Specializing PA bounds: http://www.lmcs-online.org/ojs/viewarticle.php?id=43&layout=abstract | ||
- | |||
- | |||
- | |||
- |