Why not port the Linux kernel to Common Lisp?

As mentioned in the thread, mostly because at this point it's a massive block of battle-tested code. That's not something one scraps and rewrites lightly, and "because I like a different language" counts as rewriting it lightly. The advantages of Lisp do not outweigh the disadvantages of a scrap-and-rewrite.

A kernel with a goal of Linux compatibility is another matter entirely: something in Common Lisp or Go or Rust or some other language could make for a very interesting project. But if someone were to design a kernel from the ground up with this goal in mind, the internals would probably not look all that much like those of Linux. Whether or not that's a good thing is left as an exercise for the reader.

If you want an OS written in a high-level language, pick up JNode <http://www.jnode.org>; or House <http://programatica.cs.pdx.edu/House/>; or HalVM <http://halvm.org/>.

Oh, look, here's a web page dedicated to operating systems built in Lisp: http://linuxfinances.info/info/lisposes.html (Scroll down a bit.)

Rewriting linux kernel in common lisp would be like rebuilding the water & electricity grid in NY,NY. It can be done, but with a lot of time and effort, but before it gets off the ground people who know anything about the old infrastructure will say it is probably a futile task and shouldn't be done. That is why Linux is written in C, because it's dirty, it's effective & efficient and people get shit done in it. Also C & assembly work together hand in hand. C is a direct translation to asm.