From: Ralf Baechle <ralf@linux-mips.org>

I don't see why there is a memory barrier in copy_from_read_buf() at all. 
Even if it was useful spin_unlock_irqrestore implies a barrier.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: Andrew Morton <akpm@osdl.org>
---

 drivers/char/n_tty.c |    1 -
 1 file changed, 1 deletion(-)

diff -puN drivers/char/n_tty.c~tty-remove-useless-memory-barrier drivers/char/n_tty.c
--- a/drivers/char/n_tty.c~tty-remove-useless-memory-barrier
+++ a/drivers/char/n_tty.c
@@ -1151,7 +1151,6 @@ static int copy_from_read_buf(struct tty
 	n = min(*nr, n);
 	spin_unlock_irqrestore(&tty->read_lock, flags);
 	if (n) {
-		mb();
 		retval = copy_to_user(*b, &tty->read_buf[tty->read_tail], n);
 		n -= retval;
 		spin_lock_irqsave(&tty->read_lock, flags);
_