Differences
This shows you the differences between two versions of the page.
Both sides previous revision Previous revision Next revision | Previous revision Next revision Both sides next revision | ||
sav08:collecting_semantics [2008/05/07 01:09] vkuncak |
sav08:collecting_semantics [2008/05/08 12:36] vkuncak |
||
---|---|---|---|
Line 5: | Line 5: | ||
We can represent programs by control-flow graphs (CFG). | We can represent programs by control-flow graphs (CFG). | ||
- | **Definition:** Control flow-graph is a graph with nodes $V$, edges $E \subseteq V\times V$ and for each edge $e \in E$ a command $c(e)$, with initial $init$ and final node $final$ | + | **Definition:** Control flow-graph is a graph with nodes $V$, edges $E \subseteq V\times V$ and for each edge $e \in E$ a command given by relation $r(e)$, with initial $init$ and final node $final$ |
Program points are CFG nodes. Statements are labels on CFG edges. | Program points are CFG nodes. Statements are labels on CFG edges. | ||
- | We look at a particular way of representing and computing sets of reachable, splitting states by program counter (control-flow graph node): **collecting semantics**. | + | We look at a particular way of representing and computing sets of reachable states, splitting states by program counter (control-flow graph node): **collecting semantics**. |
$PS$ - states describing values of program variables (not including program counter). | $PS$ - states describing values of program variables (not including program counter). | ||
Line 24: | Line 24: | ||
\] | \] | ||
\[ | \[ | ||
- | \bigwedge_{(p_1,p_2) \in E} sp(C(p),r(c(p_1,p_2)))) \subseteq C(p_2) | + | \bigwedge_{(p_1,p_2) \in E} sp(C(p_1),r(p_1,p_2))) \subseteq C(p_2) |
\] | \] | ||
- | where $c(p_1,p_2)$ is command associated with edge $(p_1,p_2)$, and $r(c(p_1,p_2))$ is the relation giving semantics for this command. | + | over variables $C(p)$ for all of finitely many program points $p$. |
+ | The last condition is equivalent to | ||
+ | \[ | ||
+ | \bigwedge_{p_2 \in V} C(p_2) = C(p_2) \bigcup_{(p_1,p_2) \in E} \sp(C(p_1),r(p_1,p_2))) | ||
+ | \] | ||
+ | |||
+ | Here $r(p_1,p_2)$ is the relation giving semantics for the command associated with edge $(p_1,p_2)$. | ||
+ | |||
+ | Set of recursive inequations in the lattice of products of sets. Note $e_1 \subseteq e_2$ is equivalent to $e_2 = e_1 \cup e_2$, so we have equations in lattice. | ||
+ | |||
+ | They specify function $G$ from pairs of sets of states to pairs of sets of states which is $\cup$-morphism (and therefore monotonic). | ||
+ | |||
+ | Least fixpoint of $G$ is $\bigcup_{i \ge 0} G^i(\emptyset)$. | ||
**Example** | **Example** | ||
+ | |||
+ | Sets of states at selected points: | ||
<code> | <code> | ||
Line 44: | Line 58: | ||
} | } | ||
</code> | </code> | ||
+ | After the assignment of $x$ to 2, the set of reachable states $C$ is $C = \{ (x,2), (i,20), (y,t) \}$ | ||