Date:	Tue, 18 Feb 2014 15:18:46 -0800
From:	Jay Vosburgh <fubar@...ibm.com>
To:	Ding Tianhong <dingtianhong@...wei.com>
cc:	Andy Gospodarek <andy@...yhouse.net>,
	Veaceslav Falico <vfalico@...hat.com>,
	Cong Wang <cwang@...pensource.com>,
	Thomas Glanzmann <thomas@...nzmann.de>,
	Jiri Pirko <jiri@...nulli.us>,
	"David S. Miller" <davem@...emloft.net>,
	Eric Dumazet <edumazet@...gle.com>,
	Scott Feldman <sfeldma@...ulusnetworks.com>,
	Netdev <netdev@...r.kernel.org>
Subject: Re: [PATCH net-next] bonding: Fix RTNL: assertion failed at net/core/rtnetlink.c for 802.3ad mode

Ding Tianhong <dingtianhong@...wei.com> wrote:

>The problem was introduced by the commit 1d3ee88ae0d
>(bonding: add netlink attributes to slave link dev).
>The bond_set_active_slave() and bond_set_backup_slave()
>will use rtmsg_ifinfo to send slave's states, so these
>two functions should be called in RTNL.
>
>In 802.3ad mode, acquiring RTNL for the __enable_port and
>__disable_port cases is difficult, as those calls generally
>already hold the state machine lock, and cannot unconditionally
>call rtnl_lock because either they already hold RTNL (for calls
>via bond_3ad_unbind_slave) or due to the potential for deadlock
>with bond_3ad_adapter_speed_changed, bond_3ad_adapter_duplex_changed,
>bond_3ad_link_change, or bond_3ad_update_lacp_rate.  All four of
>those are called with RTNL held, and acquire the state machine lock
>second.  The calling contexts for __enable_port and __disable_port
>already hold the state machine lock, and may or may not need RTNL.
>
>According to the Jay's opinion, I don't think it is a problem that
>the slave don't send notify message synchronously when the status
>changed, normally the state machine is running every 100 ms, send
>the notify message at the end of the state machine if the slave's
>state changed should be better.
>
>I fix the problem through these steps:
>
>1). add a new function bond_set_slave_state() which could change
>    the slave's state and call rtmsg_ifinfo() according to the input
>    parameters called notify.
>
>2). Add a new slave parameter which called should_notify, if the slave's state
>    changed and don't notify yet, the parameter will be set to 1, and then if
>    the slave's state changed again, the param will be set to 0, it indicate that
>    the slave's state has been restored, no need to notify any one.
>
>3). the __enable_port and __disable_port should not call rtmsg_ifinfo
>    in the state machine lock, any change in the state of slave could
>    set a flag in the slave, it will indicated that an rtmsg_ifinfo
>    should be called at the end of the state machine.
>
>Cc: Jay Vosburgh <fubar@...ibm.com>
>Cc: Veaceslav Falico <vfalico@...hat.com>
>Cc: Andy Gospodarek <andy@...yhouse.net>
>Signed-off-by: Ding Tianhong <dingtianhong@...wei.com>
>---
> drivers/net/bonding/bond_3ad.c  | 22 ++++++++++++++++++++--
> drivers/net/bonding/bond_main.c | 30 +++++++++++++++---------------
> drivers/net/bonding/bonding.h   | 31 ++++++++++++++++++++++++++-----
> 3 files changed, 61 insertions(+), 22 deletions(-)
>
>diff --git a/drivers/net/bonding/bond_3ad.c b/drivers/net/bonding/bond_3ad.c
>index e9edd84..c450d04 100644
>--- a/drivers/net/bonding/bond_3ad.c
>+++ b/drivers/net/bonding/bond_3ad.c
>@@ -181,7 +181,7 @@ static inline int __agg_has_partner(struct aggregator *agg)
>  */
> static inline void __disable_port(struct port *port)
> {
>-	bond_set_slave_inactive_flags(port->slave);
>+	bond_set_slave_inactive_flags(port->slave, false);
> }
>
> /**
>@@ -193,7 +193,7 @@ static inline void __enable_port(struct port *port)
> 	struct slave *slave = port->slave;
>
> 	if ((slave->link == BOND_LINK_UP) && IS_UP(slave->dev))
>-		bond_set_slave_active_flags(slave);
>+		bond_set_slave_active_flags(slave, false);
> }
>
> /**
>@@ -2065,6 +2065,7 @@ void bond_3ad_state_machine_handler(struct work_struct *work)
> 	struct list_head *iter;
> 	struct slave *slave;
> 	struct port *port;
>+	int slave_should_notify = 0;
>
> 	read_lock(&bond->lock);
> 	rcu_read_lock();
>@@ -2122,8 +2123,25 @@ void bond_3ad_state_machine_handler(struct work_struct *work)
> 	}
>
> re_arm:
>+	bond_for_each_slave_rcu(bond, slave, iter) {
>+		if (slave->should_notify) {
>+			slave_should_notify = 1;
>+			break;
>+		}
>+	}
> 	rcu_read_unlock();
> 	read_unlock(&bond->lock);
>+
>+	if (slave_should_notify && rtnl_trylock()) {
>+		bond_for_each_slave(bond, slave, iter) {
>+			if (slave->should_notify) {
>+				rtmsg_ifinfo(RTM_NEWLINK, slave->dev, 0,
>+					     GFP_KERNEL);
>+				slave->should_notify = 0;
>+			}
>+		}
>+		rtnl_unlock();
>+	}
> 	queue_delayed_work(bond->wq, &bond->ad_work, ad_delta_in_ticks);
> }
>
>diff --git a/drivers/net/bonding/bond_main.c b/drivers/net/bonding/bond_main.c
>index 3bce855..1c14e64 100644
>--- a/drivers/net/bonding/bond_main.c
>+++ b/drivers/net/bonding/bond_main.c
>@@ -829,21 +829,21 @@ void bond_change_active_slave(struct bonding *bond, struct slave *new_active)
> 	if (bond_is_lb(bond)) {
> 		bond_alb_handle_active_change(bond, new_active);
> 		if (old_active)
>-			bond_set_slave_inactive_flags(old_active);
>+			bond_set_slave_inactive_flags(old_active, true);
> 		if (new_active)
>-			bond_set_slave_active_flags(new_active);
>+			bond_set_slave_active_flags(new_active, true);
> 	} else {
> 		rcu_assign_pointer(bond->curr_active_slave, new_active);
> 	}
>
> 	if (bond->params.mode == BOND_MODE_ACTIVEBACKUP) {
> 		if (old_active)
>-			bond_set_slave_inactive_flags(old_active);
>+			bond_set_slave_inactive_flags(old_active, true);
>
> 		if (new_active) {
> 			bool should_notify_peers = false;
>
>-			bond_set_slave_active_flags(new_active);
>+			bond_set_slave_active_flags(new_active, true);
>
> 			if (bond->params.fail_over_mac)
> 				bond_do_fail_over_mac(bond, new_active,
>@@ -1462,14 +1462,14 @@ int bond_enslave(struct net_device *bond_dev, struct net_device *slave_dev)
>
> 	switch (bond->params.mode) {
> 	case BOND_MODE_ACTIVEBACKUP:
>-		bond_set_slave_inactive_flags(new_slave);
>+		bond_set_slave_inactive_flags(new_slave, true);
> 		break;
> 	case BOND_MODE_8023AD:
> 		/* in 802.3ad mode, the internal mechanism
> 		 * will activate the slaves in the selected
> 		 * aggregator
> 		 */
>-		bond_set_slave_inactive_flags(new_slave);
>+		bond_set_slave_inactive_flags(new_slave, true);
> 		/* if this is the first slave */
> 		if (!prev_slave) {
> 			SLAVE_AD_INFO(new_slave).id = 1;
>@@ -1487,7 +1487,7 @@ int bond_enslave(struct net_device *bond_dev, struct net_device *slave_dev)
> 	case BOND_MODE_TLB:
> 	case BOND_MODE_ALB:
> 		bond_set_active_slave(new_slave);
>-		bond_set_slave_inactive_flags(new_slave);
>+		bond_set_slave_inactive_flags(new_slave, true);
> 		break;
> 	default:
> 		pr_debug("This slave is always active in trunk mode\n");
>@@ -2009,7 +2009,7 @@ static void bond_miimon_commit(struct bonding *bond)
>
> 			if (bond->params.mode == BOND_MODE_ACTIVEBACKUP ||
> 			    bond->params.mode == BOND_MODE_8023AD)
>-				bond_set_slave_inactive_flags(slave);
>+				bond_set_slave_inactive_flags(slave, true);
>
> 			pr_info("%s: link status definitely down for interface %s, disabling it\n",
> 				bond->dev->name, slave->dev->name);
>@@ -2555,7 +2555,7 @@ static void bond_ab_arp_commit(struct bonding *bond)
> 				slave->link = BOND_LINK_UP;
> 				if (bond->current_arp_slave) {
> 					bond_set_slave_inactive_flags(
>-						bond->current_arp_slave);
>+						bond->current_arp_slave, true);
> 					bond->current_arp_slave = NULL;
> 				}
>
>@@ -2575,7 +2575,7 @@ static void bond_ab_arp_commit(struct bonding *bond)
> 				slave->link_failure_count++;
>
> 			slave->link = BOND_LINK_DOWN;
>-			bond_set_slave_inactive_flags(slave);
>+			bond_set_slave_inactive_flags(slave, true);
>
> 			pr_info("%s: link status definitely down for interface %s, disabling it\n",
> 				bond->dev->name, slave->dev->name);
>@@ -2650,7 +2650,7 @@ static bool bond_ab_arp_probe(struct bonding *bond)
> 		}
> 	}
>
>-	bond_set_slave_inactive_flags(curr_arp_slave);
>+	bond_set_slave_inactive_flags(curr_arp_slave, true);

	This...

> 	bond_for_each_slave(bond, slave, iter) {
> 		if (!found && !before && IS_UP(slave->dev))
>@@ -2670,7 +2670,7 @@ static bool bond_ab_arp_probe(struct bonding *bond)
> 			if (slave->link_failure_count < UINT_MAX)
> 				slave->link_failure_count++;
>
>-			bond_set_slave_inactive_flags(slave);
>+			bond_set_slave_inactive_flags(slave, true);

	[ but not this one ]

> 			pr_info("%s: backup interface %s is now down\n",
> 				bond->dev->name, slave->dev->name);
>@@ -2688,7 +2688,7 @@ static bool bond_ab_arp_probe(struct bonding *bond)
> 	}
>
> 	new_slave->link = BOND_LINK_BACK;
>-	bond_set_slave_active_flags(new_slave);
>+	bond_set_slave_active_flags(new_slave, true);

	and this should arguably never send an rtmsg_ifinfo
notification.

	My presumption is that the notification is to indicate that the
interface has actually changed state in a meaningful way, but these
calls are internal settings of the flags to allow the ARP monitor to
search for a slave to become active when there are no active slaves
(according to ARP monitor).  The flag setting to active or backup is to
permit the ARP monitor's response logic to do the right thing when
deciding if the test slave (current_arp_slave) is up or not.

	What will happen here is that, for as long as all slaves are
down, each cycle through the curr_arp_slave will shift to the next slave
in the list, the flags will be adjusted for the previous and now-current
arp slaves, and a pass of the ARP monitor will complete (send ARPs, next
time through, check for any replies).

	As the patch is written, while all slaves are down this will
generate an rtmsg_ifinfo call for each of two slaves during each pass of
the ARP monitor, until such time that any slave becomes active.

	I haven't heard back from anybody about what the rtmsg_ifinfo
notifications are used for, so I'm not 100% sure this is incorrect.  It
doesn't seem like a proper usage, though, since the slave is not
actually transitioning to an "up" or "down" state that is usable.

	The way the code is written in the patch, "false" doesn't mean
"never notify," it means "notify later," which isn't quite what I had
meant.

	I may be concerned about nothing here, and the extra
notifications may be harmless.  In the absence of feedback, we could
apply the patch as-is and if there are issues, fix them later.

	-J


> 	bond_arp_send_all(bond, new_slave);
> 	new_slave->jiffies = jiffies;
> 	rcu_assign_pointer(bond->current_arp_slave, new_slave);
>@@ -3035,9 +3035,9 @@ static int bond_open(struct net_device *bond_dev)
> 		bond_for_each_slave(bond, slave, iter) {
> 			if ((bond->params.mode == BOND_MODE_ACTIVEBACKUP)
> 				&& (slave != bond->curr_active_slave)) {
>-				bond_set_slave_inactive_flags(slave);
>+				bond_set_slave_inactive_flags(slave, true);
> 			} else {
>-				bond_set_slave_active_flags(slave);
>+				bond_set_slave_active_flags(slave, true);
> 			}
> 		}
> 		read_unlock(&bond->curr_slave_lock);
>diff --git a/drivers/net/bonding/bonding.h b/drivers/net/bonding/bonding.h
>index 86ccfb9..1f51a5f 100644
>--- a/drivers/net/bonding/bonding.h
>+++ b/drivers/net/bonding/bonding.h
>@@ -195,7 +195,8 @@ struct slave {
> 	s8     new_link;
> 	u8     backup:1,   /* indicates backup slave. Value corresponds with
> 			      BOND_STATE_ACTIVE and BOND_STATE_BACKUP */
>-	       inactive:1; /* indicates inactive slave */
>+	       inactive:1, /* indicates inactive slave */
>+	       should_notify:1; /* indicateds whether the state changed */
> 	u8     duplex;
> 	u32    original_mtu;
> 	u32    link_failure_count;
>@@ -303,6 +304,24 @@ static inline void bond_set_backup_slave(struct slave *slave)
> 	}
> }
>
>+static inline void bond_set_slave_state(struct slave *slave,
>+					int slave_state, bool notify)
>+{
>+	if (slave->backup == slave_state)
>+		return;
>+
>+	slave->backup = slave_state;
>+	if (notify) {
>+		rtmsg_ifinfo(RTM_NEWLINK, slave->dev, 0, GFP_KERNEL);
>+		slave->should_notify = 0;
>+	} else {
>+		if (slave->should_notify)
>+			slave->should_notify = 0;
>+		else
>+			slave->should_notify = 1;
>+	}
>+}
>+
> static inline void bond_slave_state_change(struct bonding *bond)
> {
> 	struct list_head *iter;
>@@ -394,17 +413,19 @@ static inline void bond_netpoll_send_skb(const struct slave *slave,
> }
> #endif
>
>-static inline void bond_set_slave_inactive_flags(struct slave *slave)
>+static inline void bond_set_slave_inactive_flags(struct slave *slave,
>+						 bool notify)
> {
> 	if (!bond_is_lb(slave->bond))
>-		bond_set_backup_slave(slave);
>+		bond_set_slave_state(slave, BOND_STATE_BACKUP, notify);
> 	if (!slave->bond->params.all_slaves_active)
> 		slave->inactive = 1;
> }
>
>-static inline void bond_set_slave_active_flags(struct slave *slave)
>+static inline void bond_set_slave_active_flags(struct slave *slave,
>+					       bool notify)
> {
>-	bond_set_active_slave(slave);
>+	bond_set_slave_state(slave, BOND_STATE_ACTIVE, notify);
> 	slave->inactive = 0;
> }
>
>-- 
>1.8.0
>
>

---
	-Jay Vosburgh, IBM Linux Technology Center, fubar@...ibm.com

--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Powered by blists - more mailing lists