[PATCH net v4] vlan: Fix lockdep warning when vlan dev handle notification

[PATCH net v4] vlan: Fix lockdep warning when vlan dev handle notification

[Ding Tianhong]

When I open the LOCKDEP config and run these steps:

modprobe 8021q
vconfig add eth2 20
vconfig add eth2.20 30
ifconfig eth2 xx.xx.xx.xx

then the Call Trace happened:

[32524.386288] =============================================
[32524.386293] [ INFO: possible recursive locking detected ]
[32524.386298] 3.14.0-rc2-0.7-default+ #35 Tainted: G           O
[32524.386302] ---------------------------------------------
[32524.386306] ifconfig/3103 is trying to acquire lock:
[32524.386310]  (&vlan_netdev_addr_lock_key/1){+.....}, at: [<ffffffff814275f4>] dev_mc_sync+0x64/0xb0
[32524.386326]
[32524.386326] but task is already holding lock:
[32524.386330]  (&vlan_netdev_addr_lock_key/1){+.....}, at: [<ffffffff8141af83>] dev_set_rx_mode+0x23/0x40
[32524.386341]
[32524.386341] other info that might help us debug this:
[32524.386345]  Possible unsafe locking scenario:
[32524.386345]
[32524.386350]        CPU0
[32524.386352]        ----
[32524.386354]   lock(&vlan_netdev_addr_lock_key/1);
[32524.386359]   lock(&vlan_netdev_addr_lock_key/1);
[32524.386364]
[32524.386364]  *** DEADLOCK ***
[32524.386364]
[32524.386368]  May be due to missing lock nesting notation
[32524.386368]
[32524.386373] 2 locks held by ifconfig/3103:
[32524.386376]  #0:  (rtnl_mutex){+.+.+.}, at: [<ffffffff81431d42>] rtnl_lock+0x12/0x20
[32524.386387]  #1:  (&vlan_netdev_addr_lock_key/1){+.....}, at: [<ffffffff8141af83>] dev_set_rx_mode+0x23/0x40
[32524.386398]
[32524.386398] stack backtrace:
[32524.386403] CPU: 1 PID: 3103 Comm: ifconfig Tainted: G           O 3.14.0-rc2-0.7-default+ #35
[32524.386409] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2007
[32524.386414]  ffffffff81ffae40 ffff8800d9625ae8 ffffffff814f68a2 ffff8800d9625bc8
[32524.386421]  ffffffff810a35fb ffff8800d8a8d9d0 00000000d9625b28 ffff8800d8a8e5d0
[32524.386428]  000003cc00000000 0000000000000002 ffff8800d8a8e5f8 0000000000000000
[32524.386435] Call Trace:
[32524.386441]  [<ffffffff814f68a2>] dump_stack+0x6a/0x78
[32524.386448]  [<ffffffff810a35fb>] __lock_acquire+0x7ab/0x1940
[32524.386454]  [<ffffffff810a323a>] ? __lock_acquire+0x3ea/0x1940
[32524.386459]  [<ffffffff810a4874>] lock_acquire+0xe4/0x110
[32524.386464]  [<ffffffff814275f4>] ? dev_mc_sync+0x64/0xb0
[32524.386471]  [<ffffffff814fc07a>] _raw_spin_lock_nested+0x2a/0x40
[32524.386476]  [<ffffffff814275f4>] ? dev_mc_sync+0x64/0xb0
[32524.386481]  [<ffffffff814275f4>] dev_mc_sync+0x64/0xb0
[32524.386489]  [<ffffffffa0500cab>] vlan_dev_set_rx_mode+0x2b/0x50 [8021q]
[32524.386495]  [<ffffffff8141addf>] __dev_set_rx_mode+0x5f/0xb0
[32524.386500]  [<ffffffff8141af8b>] dev_set_rx_mode+0x2b/0x40
[32524.386506]  [<ffffffff8141b3cf>] __dev_open+0xef/0x150
[32524.386511]  [<ffffffff8141b177>] __dev_change_flags+0xa7/0x190
[32524.386516]  [<ffffffff8141b292>] dev_change_flags+0x32/0x80
[32524.386524]  [<ffffffff8149ca56>] devinet_ioctl+0x7d6/0x830
[32524.386532]  [<ffffffff81437b0b>] ? dev_ioctl+0x34b/0x660
[32524.386540]  [<ffffffff814a05b0>] inet_ioctl+0x80/0xa0
[32524.386550]  [<ffffffff8140199d>] sock_do_ioctl+0x2d/0x60
[32524.386558]  [<ffffffff81401a52>] sock_ioctl+0x82/0x2a0
[32524.386568]  [<ffffffff811a7123>] do_vfs_ioctl+0x93/0x590
[32524.386578]  [<ffffffff811b2705>] ? rcu_read_lock_held+0x45/0x50
[32524.386586]  [<ffffffff811b39e5>] ? __fget_light+0x105/0x110
[32524.386594]  [<ffffffff811a76b1>] SyS_ioctl+0x91/0xb0
[32524.386604]  [<ffffffff815057e2>] system_call_fastpath+0x16/0x1b

========================================================================

The reason is that all of the addr_lock_key for vlan dev have the same class,
so if we change the status for vlan dev, the vlan dev and its real dev will
hold the same class of addr_lock_key together, so the warning happened.

we should distinguish the lock depth for vlan dev and its real dev.

v1->v2: Convert the vlan_netdev_addr_lock_key to an array of eight elements, which
	could support to add 8 vlan id on a same vlan dev, I think it is enough for current
	scene, because a netdev's name is limited to IFNAMSIZ which could not hold 8 vlan id,
	and the vlan dev would not meet the same class key with its real dev.

	The new function vlan_dev_get_lockdep_subkey() will return the subkey and make the vlan
	dev could get a suitable class key.

v2->v3: According David's suggestion, I use the subclass to distinguish the lock key for vlan dev
	and its real dev, but it make no sense, because the difference for subclass in the
	lock_class_key doesn't mean that the difference class for lock_key, so I use lock_depth
	to distinguish the different depth for every vlan dev, the same depth of the vlan dev
	could have the same lock_class_key, I import the MAX_LOCK_DEPTH from the include/linux/sched.h,
	I think it is enough here, the lockdep should never exceed that value.

v3->v4: Add a huge array of locking keys will waste static kernel memory and is not a appropriate method,
	we could use _nested() variants to fix the problem, calculate the depth for every vlan dev,
	and use the depth as the subclass for addr_lock_key.

Signed-off-by: Ding Tianhong <dingtianhong@huawei.com>
---
 net/8021q/vlan_dev.c | 46 +++++++++++++++++++++++++++++++++++++++++-----
 net/core/dev.c       |  1 +
 2 files changed, 42 insertions(+), 5 deletions(-)

diff --git a/net/8021q/vlan_dev.c b/net/8021q/vlan_dev.c
index 6f142f0..a1c4f67 100644
--- a/net/8021q/vlan_dev.c
+++ b/net/8021q/vlan_dev.c
@@ -493,10 +493,48 @@ static void vlan_dev_change_rx_flags(struct net_device *dev, int change)
 	}
 }
 
+static int vlan_calculate_locking_subclass(struct net_device *real_dev)
+{
+	int subclass = 0;
+
+	while (is_vlan_dev(real_dev)) {
+		subclass++;
+		real_dev = vlan_dev_priv(real_dev)->real_dev;
+	}
+
+	return subclass;
+}
+
+static void vlan_dev_mc_sync(struct net_device *to, struct net_device *from)
+{
+	int err = 0, subclass;
+
+	subclass = vlan_calculate_locking_subclass(to);
+
+	spin_lock_nested(&to->addr_list_lock, subclass);
+	err = __hw_addr_sync(&to->mc, &from->mc, to->addr_len);
+	if (!err)
+		__dev_set_rx_mode(to);
+	spin_unlock(&to->addr_list_lock);
+}
+
+static void vlan_dev_uc_sync(struct net_device *to, struct net_device *from)
+{
+	int err = 0, subclass;
+
+	subclass = vlan_calculate_locking_subclass(to);
+
+	spin_lock_nested(&to->addr_list_lock, subclass);
+	err = __hw_addr_sync(&to->uc, &from->uc, to->addr_len);
+	if (!err)
+		__dev_set_rx_mode(to);
+	spin_unlock(&to->addr_list_lock);
+}
+
 static void vlan_dev_set_rx_mode(struct net_device *vlan_dev)
 {
-	dev_mc_sync(vlan_dev_priv(vlan_dev)->real_dev, vlan_dev);
-	dev_uc_sync(vlan_dev_priv(vlan_dev)->real_dev, vlan_dev);
+	vlan_dev_mc_sync(vlan_dev_priv(vlan_dev)->real_dev, vlan_dev);
+	vlan_dev_uc_sync(vlan_dev_priv(vlan_dev)->real_dev, vlan_dev);
 }
 
 /*
@@ -608,9 +646,7 @@ static int vlan_dev_init(struct net_device *dev)
 
 	SET_NETDEV_DEVTYPE(dev, &vlan_type);
 
-	if (is_vlan_dev(real_dev))
-		subclass = 1;
-
+	subclass = vlan_calculate_locking_subclass(dev);
 	vlan_dev_set_lockdep_class(dev, subclass);
 
 	vlan_dev_priv(dev)->vlan_pcpu_stats = netdev_alloc_pcpu_stats(struct vlan_pcpu_stats);
diff --git a/net/core/dev.c b/net/core/dev.c
index 14dac06..21265cd 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -5238,6 +5238,7 @@ void __dev_set_rx_mode(struct net_device *dev)
 	if (ops->ndo_set_rx_mode)
 		ops->ndo_set_rx_mode(dev);
 }
+EXPORT_SYMBOL(__dev_set_rx_mode);
 
 void dev_set_rx_mode(struct net_device *dev)
 {
-- 
1.8.0

Re: [PATCH net v4] vlan: Fix lockdep warning when vlan dev handle notification

[David Miller]

From: Ding Tianhong <dingtianhong@huawei.com>
Date: Thu, 17 Apr 2014 18:40:36 +0800

> When I open the LOCKDEP config and run these steps:
> 
> modprobe 8021q
> vconfig add eth2 20
> vconfig add eth2.20 30
> ifconfig eth2 xx.xx.xx.xx
> 
> then the Call Trace happened:
...
> The reason is that all of the addr_lock_key for vlan dev have the same class,
> so if we change the status for vlan dev, the vlan dev and its real dev will
> hold the same class of addr_lock_key together, so the warning happened.
> 
> we should distinguish the lock depth for vlan dev and its real dev.
...
> Signed-off-by: Ding Tianhong <dingtianhong@huawei.com>

Looks good, applied and queued up for -stable, thanks!