You can try it in the following way. First  fit two models: model 1 is y = beta0 + beta1*x1 + beta2*x2,   model 2 is  y = beta0 + beta1 *(x1+x2). Compute the Residual Sum of Squares for model 1 and for model 2, respectively, denoted by RSS1 and RSS2. Then  calculate the statistics  T = (RSS2-RSS1)/(RSS1/(n-3)) that has a F-distribution with parameters  (1, n-3), where n is the number of observations. So you can apply it to test your problem. For more general problem with more than two variables, the above statistics can be used as well just with a need to modify the degrees of freedom of F distribution.

多谢指教！我很好奇的是用SAS可以直接加一句test x1=x2就可以给出检验beta_1是否等于beta_2, 为什么r需要自己编写F检验程序呢