On Sat, Apr 16, 2005 at 08:16:39PM +0200, Lubomir Sedlacik wrote:
> there are couple of hackish workarounds, such as chrooting to the newly
> mounted filesystem, insane amount of symlinks or nullmounts.  i'd like
> to avoid them all.

Why can't the kernel run "something" (like some rc script) that makes sure
root is mountable/mounted and after that process exits continue the usual
exec init way? Any already running kernel threads that do care about their
current directory (I can't thing of any right now) could receive a special
notification and chdir after / is valid.

Martin