The definifion of a DSL implies the development of dedicated analysis and processing tools. Some of them are actually domain specific, and some others are not. For instance, optimization tools based on the analysis of control flow graphs are not always specifically linked to any domain. In order to promote reusability of such generic tools, we suggest a Model-Driven approach whose first step is the definition of an abstract model of data that are required by generic the tool. DSLs are derived from this abstract model. By construction, they are consistent with the corresponding tools. Conversely, these tools are de facto complying with various DSLs. In this paper, we investigate the benefit of this approach on two DSLs whose syntaxes and domains are very different. But both of them are imperative languages and they share a common need for control flow analysis tools. Our approach is compared with a more traditional syntax-driven approach where the difficulties to reuse dedicated tools are clearly underlined.