Span program is a linear-algebraic model of computation originally proposed for studying the complexity theory. Recently, it has become a useful tool for designing quantum algorithms. In this paper, we present a time-efficient span-program-based quantum algorithm for the following problem. Let $T$ be an arbitrary tree. Given query access to the adjacency matrix of a graph $G$ with $n$ vertices, we need to determine whether $G$ contains $T$ as a subgraph, or $G$ does not contain $T$ as a minor, under the promise that one of these cases holds. We call this problem the subgraph/not-a-minor problem for $T$. We show that this problem can be solved by a bounded-error quantum algorithm with $O(n)$ query complexity and $\tilde{O}(n)$ time complexity. The query complexity is optimal, and the time complexity is tight up to polylog factors.