[patch 7/8] hci_usb: replace mb with smp_mb

[patch 7/8] hci_usb: replace mb with smp_mb

[akpm]

From: Ming Lei <tom.leiming@gmail.com>

smp_mb is enough for ordering memory operations among processors,and mb is
more expensive than smp_mb for UP machine, so replace it with smp_mb().

Signed-off-by: Ming Lei <tom.leiming@gmail.com>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Cc: Dave Young <hidave.darkstar@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
---

 drivers/bluetooth/hci_usb.h |   10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff -puN drivers/bluetooth/hci_usb.h~hci_usb-replace-mb-with-smp_mb drivers/bluetooth/hci_usb.h
--- a/drivers/bluetooth/hci_usb.h~hci_usb-replace-mb-with-smp_mb
+++ a/drivers/bluetooth/hci_usb.h
@@ -70,8 +70,8 @@ static inline void _urb_queue_head(struc
 {
 	unsigned long flags;
 	spin_lock_irqsave(&q->lock, flags);
-	/* _urb_unlink needs to know which spinlock to use, thus mb(). */
-	_urb->queue = q; mb(); list_add(&_urb->list, &q->head);
+	/* _urb_unlink needs to know which spinlock to use, thus smp_mb(). */
+	_urb->queue = q; smp_mb(); list_add(&_urb->list, &q->head);
 	spin_unlock_irqrestore(&q->lock, flags);
 }
 
@@ -79,8 +79,8 @@ static inline void _urb_queue_tail(struc
 {
 	unsigned long flags;
 	spin_lock_irqsave(&q->lock, flags);
-	/* _urb_unlink needs to know which spinlock to use, thus mb(). */
-	_urb->queue = q; mb(); list_add_tail(&_urb->list, &q->head);
+	/* _urb_unlink needs to know which spinlock to use, thus smp_mb(). */
+	_urb->queue = q; smp_mb(); list_add_tail(&_urb->list, &q->head);
 	spin_unlock_irqrestore(&q->lock, flags);
 }
 
@@ -89,7 +89,7 @@ static inline void _urb_unlink(struct _u
 	struct _urb_queue *q;
 	unsigned long flags;
 
-	mb();
+	smp_mb();
 	q = _urb->queue;
 	/* If q is NULL, it will die at easy-to-debug NULL pointer dereference.
 	   No need to BUG(). */
_

Re: [patch 7/8] hci_usb: replace mb with smp_mb

[David Miller]

From: akpm@linux-foundation.org
Date: Mon, 22 Sep 2008 14:50:42 -0700

> smp_mb is enough for ordering memory operations among processors,and mb is
> more expensive than smp_mb for UP machine, so replace it with smp_mb().
> 
> Signed-off-by: Ming Lei <tom.leiming@gmail.com>
> Acked-by: Marcel Holtmann <marcel@holtmann.org>
> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

Since this already ACK'd by Marcel, I've added this to net-next-2.6

Re: [patch 7/8] hci_usb: replace mb with smp_mb

[Marcel Holtmann]

Hi Dave,

> > smp_mb is enough for ordering memory operations among processors,and mb is
> > more expensive than smp_mb for UP machine, so replace it with smp_mb().
> > 
> > Signed-off-by: Ming Lei <tom.leiming@gmail.com>
> > Acked-by: Marcel Holtmann <marcel@holtmann.org>
> > Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
> 
> Since this already ACK'd by Marcel, I've added this to net-next-2.6

that is absolutely fine by me. The hci_usb driver is a dead end and we
do have multiple people now looking at and testing btusb. So expect the
hci_usb driver to go away soon. Either 2.6.28 or 2.6.29.

Regards

Marcel