Patchwork bonding: reset queue mapping prior to transmission to physical device (v2)

login
register
mail settings
Submitter Neil Horman
Date June 3, 2011, 1:26 p.m.
Message ID <1307107607-3015-1-git-send-email-nhorman@tuxdriver.com>
Download mbox | patch
Permalink /patch/98572/
State Superseded
Delegated to: David Miller
Headers show

Comments

Neil Horman - June 3, 2011, 1:26 p.m.
The bonding driver is multiqueue enabled, in which each queue represents a slave
to enable optional steering of output frames to given slaves against the default
output policy.  However, it needs to reset the skb->queue_mapping prior to
queuing to the physical device or the physical slave (if it is multiqueue) could
wind up transmitting on an unintended tx queue

Change Notes:
v2) Based on first pass review, updated the patch to restore the origional queue
mapping that was found in bond_select_queue, rather than simply resetting to
zero.  This preserves the value of queue_mapping when it was set on receive in
the forwarding case which is desireable.

Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
CC: Jay Vosburgh <fubar@us.ibm.com>
CC: Andy Gospodarek <andy@greyhouse.net>
CC: "David S. Miller" <davem@davemloft.net>
---
 drivers/net/bonding/bond_main.c |   10 ++++++++++
 1 files changed, 10 insertions(+), 0 deletions(-)
Ben Hutchings - June 3, 2011, 2:43 p.m.
On Fri, 2011-06-03 at 09:26 -0400, Neil Horman wrote:
> The bonding driver is multiqueue enabled, in which each queue represents a slave
> to enable optional steering of output frames to given slaves against the default
> output policy.  However, it needs to reset the skb->queue_mapping prior to
> queuing to the physical device or the physical slave (if it is multiqueue) could
> wind up transmitting on an unintended tx queue
> 
> Change Notes:
> v2) Based on first pass review, updated the patch to restore the origional queue
> mapping that was found in bond_select_queue, rather than simply resetting to
> zero.  This preserves the value of queue_mapping when it was set on receive in
> the forwarding case which is desireable.
> 
> Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
> CC: Jay Vosburgh <fubar@us.ibm.com>
> CC: Andy Gospodarek <andy@greyhouse.net>
> CC: "David S. Miller" <davem@davemloft.net>
> ---
>  drivers/net/bonding/bond_main.c |   10 ++++++++++
>  1 files changed, 10 insertions(+), 0 deletions(-)
> 
> diff --git a/drivers/net/bonding/bond_main.c b/drivers/net/bonding/bond_main.c
> index 17b4dd9..8761d86 100644
> --- a/drivers/net/bonding/bond_main.c
> +++ b/drivers/net/bonding/bond_main.c
> @@ -400,6 +400,11 @@ int bond_dev_queue_xmit(struct bonding *bond, struct sk_buff *skb,
>  {
>  	skb->dev = slave_dev;
>  	skb->priority = 1;
> +	/*
> +	 *restore the origional mapping
> +	 */
> +	skb_set_queue_mapping(skb, (u16)skb->cb[0]);
> +
>  	if (unlikely(netpoll_tx_running(slave_dev)))
>  		bond_netpoll_send_skb(bond_get_slave_by_dev(bond, slave_dev), skb);
>  	else
> @@ -4216,6 +4221,11 @@ static u16 bond_select_queue(struct net_device *dev, struct sk_buff *skb)
>  	 */
>  	u16 txq = skb_rx_queue_recorded(skb) ? skb_get_rx_queue(skb) : 0;
>  
> +	/*
> + 	 * Save the origional txq to restore before passing to the driver
> + 	 */
> +	skb->cb[0] = txq;
> +

This is throwing away the top 8 bits.  It's unlikely to hurt in
practice, but queue numbers are currently 16-bit and we already have
hardware with >256 queues (even though we don't currently expose them
all to the kernel).

Please use ((u16 *)skb->cb)[0].

Ben.

>  	if (unlikely(txq >= dev->real_num_tx_queues)) {
>  		do {
>  			txq -= dev->real_num_tx_queues;
Vitalii Demianets - June 3, 2011, 2:59 p.m.
On Friday 03 June 2011 16:26:47 Neil Horman wrote:
[...]
> +	/*
> +	 *restore the origional mapping
 Style and spelling nits, shouldn't it be:
* Restore the original mapping
> +	 */
> +	skb_set_queue_mapping(skb, (u16)skb->cb[0]);
As already pointed by Ben, it should be ((u16 *)skb->cb)[0]
[...]
> +	/*
> + 	 * Save the origional txq to restore before passing to the driver
Spelling: original
> + 	 */
> +	skb->cb[0] = txq;
As already pointed by Ben, it should be ((u16 *)skb->cb)[0]

Patch

diff --git a/drivers/net/bonding/bond_main.c b/drivers/net/bonding/bond_main.c
index 17b4dd9..8761d86 100644
--- a/drivers/net/bonding/bond_main.c
+++ b/drivers/net/bonding/bond_main.c
@@ -400,6 +400,11 @@  int bond_dev_queue_xmit(struct bonding *bond, struct sk_buff *skb,
 {
 	skb->dev = slave_dev;
 	skb->priority = 1;
+	/*
+	 *restore the origional mapping
+	 */
+	skb_set_queue_mapping(skb, (u16)skb->cb[0]);
+
 	if (unlikely(netpoll_tx_running(slave_dev)))
 		bond_netpoll_send_skb(bond_get_slave_by_dev(bond, slave_dev), skb);
 	else
@@ -4216,6 +4221,11 @@  static u16 bond_select_queue(struct net_device *dev, struct sk_buff *skb)
 	 */
 	u16 txq = skb_rx_queue_recorded(skb) ? skb_get_rx_queue(skb) : 0;
 
+	/*
+ 	 * Save the origional txq to restore before passing to the driver
+ 	 */
+	skb->cb[0] = txq;
+
 	if (unlikely(txq >= dev->real_num_tx_queues)) {
 		do {
 			txq -= dev->real_num_tx_queues;