Multivariate normal — correlation between two linear combinations

Suppose $\underline X_{3\times1}\sim\text{MNormal}_3\!\left( \underline0,\left(\begin{smallmatrix}2&0.5&2\\0.5&1&0\\2&0&4 \end{smallmatrix}\right)\right)$. Then the correlation coefficient between $X_1-X_3$ and $2X_2$ is: