In this paper, we consider the problem of joint waveform and passive beamforming design for MIMO integrated sensing and communication (ISAC) systems, where a reconfigurable intelligent surface (RIS) is deployed to assist the downlink communication. The objective is to maximize the signal-to-interference-and-noise-ratio (SINR) for radar and meanwhile minimizing the multi-user interference for communication. To address this problem, a block coordinate descent (BCD) method is proposed. Specifically, given the fixed waveform, an Element-wise Closed-Form (ECF) technique is derived for the reflection coefficients optimization, where the high-dimensional problem is divided into multiple one-dimensional problems with closed-form solution. Then, given fixed reflection coefficients, a dinkelbatchs-ECF algorithm is developed to optimize the waveform. Simulation results show that the proposed method achieves joint sensing and communication improvement and presents a clear performance advantage over existing methods.