First please add the data （X1=10,x2=20) to the end of the data set. It is highly recommended that create a new data set  . 10 20, where . stand for the missing value of y and use "set" command to combine these two data sets.

proc reg data=p7_15;
model y=x1 x2/p cli;  /*p means point forecast and cli means the confidence intervals estimation 95% */
output out=abc p=yhat stdi=sef lcl=lcl ucl=ucl; /* create a new data set called abc, yhat is the predicted point value, sef is the error terms of the forecast, lcl is the upper limit and ucl is the lower limit */
run;

还要加个语句估计的时候不要用到加的那个点吧？

When SAS operates the reg procedure, it will ignore the data with missing value automatically.

学习了