Greywolf <greywolf@starwolf.com> writes:

> For some reason, the mount_export() in mountd returns EPERM after the
> mount_unexport().  This indicates to me that something is pointing to
> the wrong place, either in the kernel or in mountd.c (and I cannot
> yet determine which).

I now understand that I've introduced two bugs into mountd.

One bug is that the mount_unexport() makes kernel forget about all
export information related to the mount point.

Another is mount point parsing.

As a result, mountd and kernel were inconsistient under some
conditions.

Latter is easy to fix and I've just fixed.  Former one isn't simple.
Should the kernel provide the way to unexport for each type of exports
(i.e, default export, public export and per network export)?

enami.