How to compile Spark with customized Hadoop?

[Dogtail L]

Hi all,

I have modified Hadoop source code, and I want to compile Spark with my
modified Hadoop. Do you know how to do that? Great thanks!