With the large scale of distributed energy resources integrated into distribution systems, the development of effective power management for controllable components is becoming an increasingly important concern. This paper focuses on solving the optimal power flow problem in a distribution network with a fully decentralized mode that has massive photovoltaics and electric vehicles. First, an optimal power flow model considering photovoltaics and electric vehicles as controllable agents was established to achieve low network loss, low curtailment of photovoltaic resources, and high satisfaction of electric vehicle owners. Second, a novel linearized power flow model was proposed to produce per-node granularity communication. Thus, each node only needed to exchange the voltage message with its neighboring nodes. Then a decentralized quadratically constrained quadratic programming model based on the alternating direction method of multipliers was built to solve the optimal power flow problem. Next, a closed-form iterative solution method for the decentralized optimization was developed to improve the calculation speed for each iteration. Finally, case studies for a real 35-bus distribution system and a real 110-bus distribution system in China were used to verify the effectiveness of the proposed method.