[PATCH 2/2] net: reimplement softnet_data.output_queue as a FIFO queue

[PATCH 2/2] net: reimplement softnet_data.output_queue as a FIFO queue

[Changli Gao]

reimplement softnet_data.output_queue as a FIFO queue.

reimplement softnet_data.output_queue as a FIFO queue to keep the fairness among
the qdiscs rescheduled.

Signed-off-by: Changli Gao <xiaosuo@gmail.com>
----
 include/linux/netdevice.h |    1 +
 net/core/dev.c            |   22 ++++++++++++----------
 2 files changed, 13 insertions(+), 10 deletions(-)
diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
index 785e514..7984c15 100644
--- a/include/linux/netdevice.h
+++ b/include/linux/netdevice.h
@@ -1385,6 +1385,7 @@ static inline int unregister_gifconf(unsigned int family)
  */
 struct softnet_data {
 	struct Qdisc		*output_queue;
+	struct Qdisc		**output_queue_tailp;
 	struct list_head	poll_list;
 	struct sk_buff_head	completion_queue;
 
diff --git a/net/core/dev.c b/net/core/dev.c
index 5bbfa0f..b2b7869 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -1557,8 +1557,9 @@ static inline void __netif_reschedule(struct Qdisc *q)
 
 	local_irq_save(flags);
 	sd = &__get_cpu_var(softnet_data);
-	q->next_sched = sd->output_queue;
-	sd->output_queue = q;
+	q->next_sched = NULL;
+	*sd->output_queue_tailp = q;
+	sd->output_queue_tailp = &q->next_sched;
 	raise_softirq_irqoff(NET_TX_SOFTIRQ);
 	local_irq_restore(flags);
 }
@@ -2526,6 +2527,7 @@ static void net_tx_action(struct softirq_action *h)
 		local_irq_disable();
 		head = sd->output_queue;
 		sd->output_queue = NULL;
+		sd->output_queue_tailp = &sd->output_queue;
 		local_irq_enable();
 
 		while (head) {
@@ -5590,7 +5592,6 @@ static int dev_cpu_callback(struct notifier_block *nfb,
 			    unsigned long action,
 			    void *ocpu)
 {
-	struct Qdisc **list_net;
 	struct sk_buff *skb;
 	unsigned int cpu, oldcpu = (unsigned long)ocpu;
 	struct softnet_data *sd, *oldsd;
@@ -5607,13 +5608,12 @@ static int dev_cpu_callback(struct notifier_block *nfb,
 	skb_queue_splice_tail_init(&oldsd->completion_queue,
 				   &sd->completion_queue);
 
-	/* Find end of our output_queue. */
-	list_net = &sd->output_queue;
-	while (*list_net)
-		list_net = &(*list_net)->next_sched;
-	/* Append output queue from offline CPU. */
-	*list_net = oldsd->output_queue;
-	oldsd->output_queue = NULL;
+	if (oldsd->output_queue) {
+		*sd->output_queue_tailp = oldsd->output_queue;
+		sd->output_queue_tailp = oldsd->output_queue_tailp;
+		oldsd->output_queue = NULL;
+		oldsd->output_queue_tailp = &oldsd->output_queue;
+	}
 
 	raise_softirq_irqoff(NET_TX_SOFTIRQ);
 	local_irq_enable();
@@ -5843,6 +5843,8 @@ static int __init net_dev_init(void)
 		skb_queue_head_init(&sd->input_pkt_queue);
 		__skb_queue_head_init(&sd->completion_queue);
 		INIT_LIST_HEAD(&sd->poll_list);
+		sd->output_queue = NULL;
+		sd->output_queue_tailp = &sd->output_queue;
 
 #ifdef CONFIG_RPS
 		sd->csd.func = rps_trigger_softirq;

Re: [PATCH 2/2] net: reimplement softnet_data.output_queue as a FIFO queue

[Changli Gao]

On Mon, Apr 26, 2010 at 11:26 AM, Changli Gao <xiaosuo@gmail.com> wrote:
> reimplement softnet_data.output_queue as a FIFO queue.
>
> reimplement softnet_data.output_queue as a FIFO queue to keep the fairness among
> the qdiscs rescheduled.
>

Dear Eric:
  do you think it is worth continuing? I didn't received any comments
even negative ones. I am really sorry for disrupting you, but I don't
know what did it mean.

-- 
Regards，
Changli Gao(xiaosuo@gmail.com)

Re: [PATCH 2/2] net: reimplement softnet_data.output_queue as a FIFO queue

[Eric Dumazet]

Le lundi 26 avril 2010 à 17:11 +0800, Changli Gao a écrit :
> On Mon, Apr 26, 2010 at 11:26 AM, Changli Gao <xiaosuo@gmail.com> wrote:
> > reimplement softnet_data.output_queue as a FIFO queue.
> >
> > reimplement softnet_data.output_queue as a FIFO queue to keep the fairness among
> > the qdiscs rescheduled.
> >
> 
> Dear Eric:
>   do you think it is worth continuing? I didn't received any comments
> even negative ones. I am really sorry for disrupting you, but I don't
> know what did it mean.
> 

For this patch I had no comment, I believe it is fine, at a first
glance...

Please give some time for netdev people to review it and Ack it.