* Subject: [PATCH 001/001] bonding: add support for netdev link events.
@ 2009-02-23 20:01 Laurent Chavey
2009-02-23 23:47 ` David Miller
0 siblings, 1 reply; 2+ messages in thread
From: Laurent Chavey @ 2009-02-23 20:01 UTC (permalink / raw)
To: Jay Vosburgh; +Cc: linux-kernel
From: Laurent Chavey <chavey@google.com>
kernel:2.6.28
Add a netdev link event handler so the bond state machine can react to
link state changes using async netdev events rather than having to
use mii monitor polling.
Signed-off-by: Laurent Chavey <chavey@google.com>
---
--- linux-2.6.28.org/drivers/net/bonding/bond_main.c 2008-12-24
15:26:37.000000000 -0800
+++ linux-2.6.28/drivers/net/bonding/bond_main.c 2009-02-23
10:59:12.000000000 -0800
@@ -2477,7 +2477,11 @@ void bond_mii_monitor(struct work_struct
}
re_arm:
- if (bond->params.miimon)
+ /* A netdev event may have set some delay work, in that
+ * case do not remove it and let it take priority.
+ */
+ if (bond->params.miimon && !bond->kill_timers &&
+ !delayed_work_pending(&bond->mii_work))
queue_delayed_work(bond->wq, &bond->mii_work,
msecs_to_jiffies(bond->params.miimon));
out:
@@ -3521,6 +3525,33 @@ static int bond_master_netdev_event(unsi
return NOTIFY_DONE;
}
+static inline void bond_netdev_link_event(struct bonding *bond)
+{
+ if (!bond->params.use_carrier)
+ return;
+
+ write_lock_bh(&bond->lock);
+ if (bond->kill_timers) {
+ write_unlock_bh(&bond->lock);
+ return;
+ }
+
+ if (delayed_work_pending(&bond->mii_work)) {
+ /* cancel_delayed_work(&bond->mii_work); calls del_timer_sync().
+ * We implement a version that does not call del_timer_sync().
+ * if an active timer is found, then the job is not on the WQ.
+ * we clear the PENDING bit, so the job can be added later
+ * else the job maybe done (PENDING bit is clear) or it maybe
+ * running, then we do not care and do only add a new job
+ * if the job has completed.
+ */
+ if (del_timer(&bond->mii_work.timer))
+ work_clear_pending(&bond->mii_work.work);
+ }
+ queue_delayed_work(bond->wq, &bond->mii_work, 0);
+ write_unlock_bh(&bond->lock);
+}
+
static int bond_slave_netdev_event(unsigned long event, struct
net_device *slave_dev)
{
struct net_device *bond_dev = slave_dev->master;
@@ -3536,15 +3567,17 @@ static int bond_slave_netdev_event(unsig
}
break;
case NETDEV_CHANGE:
- /*
- * TODO: is this what we get if somebody
- * sets up a hierarchical bond, then rmmod's
- * one of the slave bonding devices?
- */
- break;
+ case NETDEV_UP:
case NETDEV_DOWN:
+ if (bond_dev && bond_dev->priv &&
+ ARPHRD_ETHER == slave_dev->type) {
+ bond_netdev_link_event(bond_dev->priv);
+ }
+
/*
- * ... Or is it this?
+ * TODO: do we get a NETDEV_CHANGE or NETDEV_DOWN event
+ * if somebody sets up a hierarchical bond, then rmmod's
+ * one of the slave bonding devices?
*/
break;
case NETDEV_CHANGEMTU:
@@ -5196,4 +5229,3 @@ MODULE_SUPPORTED_DEVICE("most ethernet d
* tab-width: 8
* End:
*/
-
^ permalink raw reply [flat|nested] 2+ messages in thread* Re: Subject: [PATCH 001/001] bonding: add support for netdev link events.
2009-02-23 20:01 Subject: [PATCH 001/001] bonding: add support for netdev link events Laurent Chavey
@ 2009-02-23 23:47 ` David Miller
0 siblings, 0 replies; 2+ messages in thread
From: David Miller @ 2009-02-23 23:47 UTC (permalink / raw)
To: chavey; +Cc: fubar, linux-kernel, netdev
From: Laurent Chavey <chavey@google.com>
Date: Mon, 23 Feb 2009 12:01:38 -0800
> From: Laurent Chavey <chavey@google.com>
>
> kernel:2.6.28
>
> Add a netdev link event handler so the bond state machine can react to
> link state changes using async netdev events rather than having to
> use mii monitor polling.
>
> Signed-off-by: Laurent Chavey <chavey@google.com>
Please post networking patches to netdev@vger.kernel.org,
thank you.
> ---
>
>
> --- linux-2.6.28.org/drivers/net/bonding/bond_main.c 2008-12-24
> 15:26:37.000000000 -0800
> +++ linux-2.6.28/drivers/net/bonding/bond_main.c 2009-02-23
> 10:59:12.000000000 -0800
> @@ -2477,7 +2477,11 @@ void bond_mii_monitor(struct work_struct
> }
>
> re_arm:
> - if (bond->params.miimon)
> + /* A netdev event may have set some delay work, in that
> + * case do not remove it and let it take priority.
> + */
> + if (bond->params.miimon && !bond->kill_timers &&
> + !delayed_work_pending(&bond->mii_work))
> queue_delayed_work(bond->wq, &bond->mii_work,
> msecs_to_jiffies(bond->params.miimon));
> out:
> @@ -3521,6 +3525,33 @@ static int bond_master_netdev_event(unsi
> return NOTIFY_DONE;
> }
>
> +static inline void bond_netdev_link_event(struct bonding *bond)
> +{
> + if (!bond->params.use_carrier)
> + return;
> +
> + write_lock_bh(&bond->lock);
> + if (bond->kill_timers) {
> + write_unlock_bh(&bond->lock);
> + return;
> + }
> +
> + if (delayed_work_pending(&bond->mii_work)) {
> + /* cancel_delayed_work(&bond->mii_work); calls del_timer_sync().
> + * We implement a version that does not call del_timer_sync().
> + * if an active timer is found, then the job is not on the WQ.
> + * we clear the PENDING bit, so the job can be added later
> + * else the job maybe done (PENDING bit is clear) or it maybe
> + * running, then we do not care and do only add a new job
> + * if the job has completed.
> + */
> + if (del_timer(&bond->mii_work.timer))
> + work_clear_pending(&bond->mii_work.work);
> + }
> + queue_delayed_work(bond->wq, &bond->mii_work, 0);
> + write_unlock_bh(&bond->lock);
> +}
> +
> static int bond_slave_netdev_event(unsigned long event, struct
> net_device *slave_dev)
> {
> struct net_device *bond_dev = slave_dev->master;
> @@ -3536,15 +3567,17 @@ static int bond_slave_netdev_event(unsig
> }
> break;
> case NETDEV_CHANGE:
> - /*
> - * TODO: is this what we get if somebody
> - * sets up a hierarchical bond, then rmmod's
> - * one of the slave bonding devices?
> - */
> - break;
> + case NETDEV_UP:
> case NETDEV_DOWN:
> + if (bond_dev && bond_dev->priv &&
> + ARPHRD_ETHER == slave_dev->type) {
> + bond_netdev_link_event(bond_dev->priv);
> + }
> +
> /*
> - * ... Or is it this?
> + * TODO: do we get a NETDEV_CHANGE or NETDEV_DOWN event
> + * if somebody sets up a hierarchical bond, then rmmod's
> + * one of the slave bonding devices?
> */
> break;
> case NETDEV_CHANGEMTU:
> @@ -5196,4 +5229,3 @@ MODULE_SUPPORTED_DEVICE("most ethernet d
> * tab-width: 8
> * End:
> */
> -
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2009-02-23 23:47 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2009-02-23 20:01 Subject: [PATCH 001/001] bonding: add support for netdev link events Laurent Chavey
2009-02-23 23:47 ` David Miller
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®