Modeling the reasoning component of self-adapting systems including its context is a challenging task. Context feature models used in dynamic software product lines help to capture the capabilities of a software as well as the monitored context values. This enables the possibility to add constraints between the context and system features. In this paper, we present an adaptation logic architecture for specifying the knowledge for reasoning in a model-based manner by means of dynamic software product lines. The whole knowledge for reasoning is encapsulated inside a component which enables the reuse of the adaptation logic for various application scenarios. Thus, the system designer only has to specify the adaptation logic's knowledge and implement the according interfaces in the managed resource. We evaluate the adaptation logic using our architecture in a distributed computing scenario.