Thanks so much for the simulation, it is very helpful!
In fact, it is my second design, the first design was 4 layer with no ground around the inductor which should be less parasitic capacitance and inductance, but behaved the same.
And the inductor demo board used is 5.6u 0.003Ohm, isn't that more sensitive to parasitic capacitance and inductance?
However I also think it is parasitic problem causing the problem. Since if I add 100 Ohm to 4 MOSFET gate, it actually could achieve 2.5A charge current sometimes, but the MOSFET become very hot in few seconds.