>Updated patch -- this one initializes chrootpath as the somewhat >sharp-edged getcwd_common expects, so it doesn't lead to immediate >null pointer dereference. Thanks, I'll have to think about how to test this, since no longer running zfs with the ramdisk trick.