fi_delegees is basically just a reference count on users of
fi_deleg_file, which is cleared when fi_delegees goes to zero. The
fi_deleg_file check here is redundant. Also add an assertion to make
sure we don't have unbalanced puts.
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
{
struct file *filp = NULL;
+ WARN_ON_ONCE(!fp->fi_delegees);
+
spin_lock(&fp->fi_lock);
- if (fp->fi_deleg_file && --fp->fi_delegees == 0)
+ if (--fp->fi_delegees == 0) {
swap(filp, fp->fi_deleg_file);
+ }
spin_unlock(&fp->fi_lock);
if (filp) {