fs/dcache.c: bogus comment re LIST_POISON2 around __d_drop()

[Al Viro <viro@zeniv.linux.org.uk>]

[Caught while documenting ->d_hash uses]

The second line in
 * ___d_drop doesn't mark dentry as "unhashed"
 * (dentry->d_hash.pprev will be LIST_POISON2, not NULL).
is incorrect; what ___d_drop() does is
	__hlist_bl_del()
which never included poisoning - that's only done by hlist_bl_del()
and we hadn't been using that since 2011, six years prior to
___d_drop() introduction.

	I've missed that on review back then; the first line _is_
true, so functionality is all right; it's just that LIST_POISON2
reference there is confusing (and incorrect).

	For now I'm dropping the following into a local branch;
might or might not get combined with other changes in the area...

diff --git a/fs/dcache.c b/fs/dcache.c
index 0c8faeee02e2..c9995545d8d7 100644
--- a/fs/dcache.c
+++ b/fs/dcache.c
@@ -604,7 +604,7 @@ EXPORT_SYMBOL(__d_drop);
  * __d_drop requires dentry->d_lock
  *
  * ___d_drop doesn't mark dentry as "unhashed"
- * (dentry->d_hash.pprev will be LIST_POISON2, not NULL).
+ * (dentry->d_hash.pprev will not be set to NULL).
  */
 void d_drop(struct dentry *dentry)
 {