Accessing kernel VM directly: why not? Why shouldn't the kernel's VM appear in a user process's address space? On most CPUs, it does, sort of, but the user process doesn't have permission to access it. (Is the 68k the only series of CPUs that have separate address spaces?)