netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH net-next 1/4] net: add netdev_lockdep_set_classes() helper
@ 2016-06-08 23:55 Eric Dumazet
  2016-06-08 23:55 ` [PATCH net-next 2/4] net: vrf: call netdev_lockdep_set_classes() Eric Dumazet
                   ` (3 more replies)
  0 siblings, 4 replies; 12+ messages in thread
From: Eric Dumazet @ 2016-06-08 23:55 UTC (permalink / raw)
  To: David S . Miller; +Cc: netdev, Eric Dumazet, David Ahern, Eric Dumazet

It is time to add netdev_lockdep_set_classes() helper
so that lockdep annotations per device type are easier to manage.

Signed-off-by: Eric Dumazet <edumazet@google.com>
---
 drivers/net/bonding/bond_main.c | 5 +----
 drivers/net/ppp/ppp_generic.c   | 6 +-----
 drivers/net/team/team.c         | 5 +----
 include/linux/netdevice.h       | 9 +++++++++
 net/bluetooth/6lowpan.c         | 5 +----
 net/ieee802154/6lowpan/core.c   | 5 +----
 net/l2tp/l2tp_eth.c             | 6 +-----
 7 files changed, 15 insertions(+), 26 deletions(-)

diff --git a/drivers/net/bonding/bond_main.c b/drivers/net/bonding/bond_main.c
index 681af31a60ed..a729f0f8cb7b 100644
--- a/drivers/net/bonding/bond_main.c
+++ b/drivers/net/bonding/bond_main.c
@@ -4609,8 +4609,6 @@ static int bond_check_params(struct bond_params *params)
 
 static struct lock_class_key bonding_netdev_xmit_lock_key;
 static struct lock_class_key bonding_netdev_addr_lock_key;
-static struct lock_class_key bonding_tx_busylock_key;
-static struct lock_class_key bonding_qdisc_running_key;
 
 static void bond_set_lockdep_class_one(struct net_device *dev,
 				       struct netdev_queue *txq,
@@ -4625,8 +4623,7 @@ static void bond_set_lockdep_class(struct net_device *dev)
 	lockdep_set_class(&dev->addr_list_lock,
 			  &bonding_netdev_addr_lock_key);
 	netdev_for_each_tx_queue(dev, bond_set_lockdep_class_one, NULL);
-	dev->qdisc_tx_busylock = &bonding_tx_busylock_key;
-	dev->qdisc_running_key = &bonding_qdisc_running_key;
+	netdev_lockdep_set_classes(dev);
 }
 
 /* Called from registration process */
diff --git a/drivers/net/ppp/ppp_generic.c b/drivers/net/ppp/ppp_generic.c
index aeabaa42317f..17953ab15000 100644
--- a/drivers/net/ppp/ppp_generic.c
+++ b/drivers/net/ppp/ppp_generic.c
@@ -1312,13 +1312,9 @@ ppp_get_stats64(struct net_device *dev, struct rtnl_link_stats64 *stats64)
 	return stats64;
 }
 
-static struct lock_class_key ppp_tx_busylock;
-static struct lock_class_key ppp_qdisc_running_key;
-
 static int ppp_dev_init(struct net_device *dev)
 {
-	dev->qdisc_tx_busylock = &ppp_tx_busylock;
-	dev->qdisc_running_key = &ppp_qdisc_running_key;
+	netdev_lockdep_set_classes(dev);
 	return 0;
 }
 
diff --git a/drivers/net/team/team.c b/drivers/net/team/team.c
index 00eb38956a2c..d01ad0d2658b 100644
--- a/drivers/net/team/team.c
+++ b/drivers/net/team/team.c
@@ -1576,8 +1576,6 @@ static const struct team_option team_options[] = {
 
 static struct lock_class_key team_netdev_xmit_lock_key;
 static struct lock_class_key team_netdev_addr_lock_key;
-static struct lock_class_key team_tx_busylock_key;
-static struct lock_class_key team_qdisc_running_key;
 
 static void team_set_lockdep_class_one(struct net_device *dev,
 				       struct netdev_queue *txq,
@@ -1590,8 +1588,7 @@ static void team_set_lockdep_class(struct net_device *dev)
 {
 	lockdep_set_class(&dev->addr_list_lock, &team_netdev_addr_lock_key);
 	netdev_for_each_tx_queue(dev, team_set_lockdep_class_one, NULL);
-	dev->qdisc_tx_busylock = &team_tx_busylock_key;
-	dev->qdisc_running_key = &team_qdisc_running_key;
+	netdev_lockdep_set_classes(dev);
 }
 
 static int team_init(struct net_device *dev)
diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
index 59d7e06d88d5..3d61192159d5 100644
--- a/include/linux/netdevice.h
+++ b/include/linux/netdevice.h
@@ -1945,6 +1945,15 @@ static inline void netdev_for_each_tx_queue(struct net_device *dev,
 		f(dev, &dev->_tx[i], arg);
 }
 
+#define netdev_lockdep_set_classes(dev)			\
+{							\
+	static struct lock_class_key tx_busylock_key;	\
+	static struct lock_class_key qdisc_running_key;	\
+							\
+	dev->qdisc_tx_busylock = &tx_busylock_key;	\
+	dev->qdisc_running_key = &qdisc_running_key;	\
+}
+
 struct netdev_queue *netdev_pick_tx(struct net_device *dev,
 				    struct sk_buff *skb,
 				    void *accel_priv);
diff --git a/net/bluetooth/6lowpan.c b/net/bluetooth/6lowpan.c
index 977a11e418d0..0da875a69e47 100644
--- a/net/bluetooth/6lowpan.c
+++ b/net/bluetooth/6lowpan.c
@@ -627,9 +627,7 @@ static netdev_tx_t bt_xmit(struct sk_buff *skb, struct net_device *netdev)
 	return err < 0 ? NET_XMIT_DROP : err;
 }
 
-static struct lock_class_key bt_tx_busylock;
 static struct lock_class_key bt_netdev_xmit_lock_key;
-static struct lock_class_key bt_qdisc_running_key;
 
 static void bt_set_lockdep_class_one(struct net_device *dev,
 				     struct netdev_queue *txq,
@@ -641,8 +639,7 @@ static void bt_set_lockdep_class_one(struct net_device *dev,
 static int bt_dev_init(struct net_device *dev)
 {
 	netdev_for_each_tx_queue(dev, bt_set_lockdep_class_one, NULL);
-	dev->qdisc_tx_busylock = &bt_tx_busylock;
-	dev->qdisc_running_key = &bt_qdisc_running_key;
+	netdev_lockdep_set_classes(dev);
 
 	return 0;
 }
diff --git a/net/ieee802154/6lowpan/core.c b/net/ieee802154/6lowpan/core.c
index 14aa5effd29a..139dbe3f7cc6 100644
--- a/net/ieee802154/6lowpan/core.c
+++ b/net/ieee802154/6lowpan/core.c
@@ -58,9 +58,7 @@ static struct header_ops lowpan_header_ops = {
 	.create	= lowpan_header_create,
 };
 
-static struct lock_class_key lowpan_tx_busylock;
 static struct lock_class_key lowpan_netdev_xmit_lock_key;
-static struct lock_class_key lowpan_qdisc_running_key;
 
 static void lowpan_set_lockdep_class_one(struct net_device *ldev,
 					 struct netdev_queue *txq,
@@ -73,8 +71,7 @@ static void lowpan_set_lockdep_class_one(struct net_device *ldev,
 static int lowpan_dev_init(struct net_device *ldev)
 {
 	netdev_for_each_tx_queue(ldev, lowpan_set_lockdep_class_one, NULL);
-	ldev->qdisc_tx_busylock = &lowpan_tx_busylock;
-	ldev->qdisc_running_key = &lowpan_qdisc_running_key;
+	netdev_lockdep_set_classes(ldev);
 
 	return 0;
 }
diff --git a/net/l2tp/l2tp_eth.c b/net/l2tp/l2tp_eth.c
index c00d72d182fa..57fc5a46ce06 100644
--- a/net/l2tp/l2tp_eth.c
+++ b/net/l2tp/l2tp_eth.c
@@ -67,9 +67,6 @@ static inline struct l2tp_eth_net *l2tp_eth_pernet(struct net *net)
 	return net_generic(net, l2tp_eth_net_id);
 }
 
-static struct lock_class_key l2tp_eth_tx_busylock;
-static struct lock_class_key l2tp_qdisc_running_key;
-
 static int l2tp_eth_dev_init(struct net_device *dev)
 {
 	struct l2tp_eth *priv = netdev_priv(dev);
@@ -77,8 +74,7 @@ static int l2tp_eth_dev_init(struct net_device *dev)
 	priv->dev = dev;
 	eth_hw_addr_random(dev);
 	eth_broadcast_addr(dev->broadcast);
-	dev->qdisc_tx_busylock = &l2tp_eth_tx_busylock;
-	dev->qdisc_running_key = &l2tp_qdisc_running_key;
+	netdev_lockdep_set_classes(dev);
 
 	return 0;
 }
-- 
2.8.0.rc3.226.g39d4020

^ permalink raw reply related	[flat|nested] 12+ messages in thread

* [PATCH net-next 2/4] net: vrf: call netdev_lockdep_set_classes()
  2016-06-08 23:55 [PATCH net-next 1/4] net: add netdev_lockdep_set_classes() helper Eric Dumazet
@ 2016-06-08 23:55 ` Eric Dumazet
  2016-06-09  1:26   ` David Ahern
  2016-06-09  7:03   ` David Miller
  2016-06-08 23:55 ` [PATCH net-next 3/4] net: macvlan: " Eric Dumazet
                   ` (2 subsequent siblings)
  3 siblings, 2 replies; 12+ messages in thread
From: Eric Dumazet @ 2016-06-08 23:55 UTC (permalink / raw)
  To: David S . Miller; +Cc: netdev, Eric Dumazet, David Ahern, Eric Dumazet

In case a qdisc is used on a vrf device, we need to use different
lockdep classes to avoid false positives.

Fixes: f9eb8aea2a1e ("net_sched: transform qdisc running bit into a seqcount")
Reported-by: David Ahern <dsa@cumulusnetworks.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
---
 drivers/net/vrf.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/vrf.c b/drivers/net/vrf.c
index 1b214ea4619a..ee6bc1c5c1ce 100644
--- a/drivers/net/vrf.c
+++ b/drivers/net/vrf.c
@@ -657,7 +657,7 @@ static int vrf_dev_init(struct net_device *dev)
 
 	/* similarly, oper state is irrelevant; set to up to avoid confusion */
 	dev->operstate = IF_OPER_UP;
-
+	netdev_lockdep_set_classes(dev);
 	return 0;
 
 out_rth:
-- 
2.8.0.rc3.226.g39d4020

^ permalink raw reply related	[flat|nested] 12+ messages in thread

* [PATCH net-next 3/4] net: macvlan: call netdev_lockdep_set_classes()
  2016-06-08 23:55 [PATCH net-next 1/4] net: add netdev_lockdep_set_classes() helper Eric Dumazet
  2016-06-08 23:55 ` [PATCH net-next 2/4] net: vrf: call netdev_lockdep_set_classes() Eric Dumazet
@ 2016-06-08 23:55 ` Eric Dumazet
  2016-06-09  7:03   ` David Miller
  2016-06-08 23:55 ` [PATCH net-next 4/4] net: ipvlan: " Eric Dumazet
  2016-06-09  7:03 ` [PATCH net-next 1/4] net: add netdev_lockdep_set_classes() helper David Miller
  3 siblings, 1 reply; 12+ messages in thread
From: Eric Dumazet @ 2016-06-08 23:55 UTC (permalink / raw)
  To: David S . Miller; +Cc: netdev, Eric Dumazet, David Ahern, Eric Dumazet

In case a qdisc is used on a macvlan device, we need to use different
lockdep classes to avoid false positives.

Fixes: f9eb8aea2a1e ("net_sched: transform qdisc running bit into a seqcount")
Fixes: 79640a4ca695 ("net: add additional lock to qdisc to increase throughput")
Signed-off-by: Eric Dumazet <edumazet@google.com>
---
 drivers/net/macvlan.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/net/macvlan.c b/drivers/net/macvlan.c
index 0c65bd914aed..b0c6f5b43f67 100644
--- a/drivers/net/macvlan.c
+++ b/drivers/net/macvlan.c
@@ -823,6 +823,7 @@ static void macvlan_set_lockdep_class(struct net_device *dev)
 				       &macvlan_netdev_addr_lock_key,
 				       macvlan_get_nest_level(dev));
 	netdev_for_each_tx_queue(dev, macvlan_set_lockdep_class_one, NULL);
+	netdev_lockdep_set_classes(dev);
 }
 
 static int macvlan_init(struct net_device *dev)
-- 
2.8.0.rc3.226.g39d4020

^ permalink raw reply related	[flat|nested] 12+ messages in thread

* [PATCH net-next 4/4] net: ipvlan: call netdev_lockdep_set_classes()
  2016-06-08 23:55 [PATCH net-next 1/4] net: add netdev_lockdep_set_classes() helper Eric Dumazet
  2016-06-08 23:55 ` [PATCH net-next 2/4] net: vrf: call netdev_lockdep_set_classes() Eric Dumazet
  2016-06-08 23:55 ` [PATCH net-next 3/4] net: macvlan: " Eric Dumazet
@ 2016-06-08 23:55 ` Eric Dumazet
  2016-06-09  7:03   ` David Miller
  2016-06-09  7:03 ` [PATCH net-next 1/4] net: add netdev_lockdep_set_classes() helper David Miller
  3 siblings, 1 reply; 12+ messages in thread
From: Eric Dumazet @ 2016-06-08 23:55 UTC (permalink / raw)
  To: David S . Miller; +Cc: netdev, Eric Dumazet, David Ahern, Eric Dumazet

In case a qdisc is used on a ipvlan device, we need to use different
lockdep classes to avoid false positives.

Fixes: f9eb8aea2a1e ("net_sched: transform qdisc running bit into a seqcount")
Fixes: 79640a4ca695 ("net: add additional lock to qdisc to increase throughput")
Signed-off-by: Eric Dumazet <edumazet@google.com>
---
 drivers/net/ipvlan/ipvlan_main.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/net/ipvlan/ipvlan_main.c b/drivers/net/ipvlan/ipvlan_main.c
index 1c4d395fbd49..dee2e67db065 100644
--- a/drivers/net/ipvlan/ipvlan_main.c
+++ b/drivers/net/ipvlan/ipvlan_main.c
@@ -107,6 +107,7 @@ static void ipvlan_set_lockdep_class(struct net_device *dev)
 {
 	lockdep_set_class(&dev->addr_list_lock, &ipvlan_netdev_addr_lock_key);
 	netdev_for_each_tx_queue(dev, ipvlan_set_lockdep_class_one, NULL);
+	netdev_lockdep_set_classes(dev);
 }
 
 static int ipvlan_init(struct net_device *dev)
-- 
2.8.0.rc3.226.g39d4020

^ permalink raw reply related	[flat|nested] 12+ messages in thread

* Re: [PATCH net-next 2/4] net: vrf: call netdev_lockdep_set_classes()
  2016-06-08 23:55 ` [PATCH net-next 2/4] net: vrf: call netdev_lockdep_set_classes() Eric Dumazet
@ 2016-06-09  1:26   ` David Ahern
  2016-06-09  1:43     ` Eric Dumazet
  2016-06-09  7:03   ` David Miller
  1 sibling, 1 reply; 12+ messages in thread
From: David Ahern @ 2016-06-09  1:26 UTC (permalink / raw)
  To: Eric Dumazet, David S . Miller; +Cc: netdev, Eric Dumazet

On 6/8/16 5:55 PM, Eric Dumazet wrote:
> In case a qdisc is used on a vrf device, we need to use different
> lockdep classes to avoid false positives.
>
> Fixes: f9eb8aea2a1e ("net_sched: transform qdisc running bit into a seqcount")
> Reported-by: David Ahern <dsa@cumulusnetworks.com>
> Signed-off-by: Eric Dumazet <edumazet@google.com>
> ---
>  drivers/net/vrf.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/net/vrf.c b/drivers/net/vrf.c
> index 1b214ea4619a..ee6bc1c5c1ce 100644
> --- a/drivers/net/vrf.c
> +++ b/drivers/net/vrf.c
> @@ -657,7 +657,7 @@ static int vrf_dev_init(struct net_device *dev)
>
>  	/* similarly, oper state is irrelevant; set to up to avoid confusion */
>  	dev->operstate = IF_OPER_UP;
> -
> +	netdev_lockdep_set_classes(dev);
>  	return 0;
>
>  out_rth:
>

Still see the problem; all 4 patches applied, make clean followed by a 
build just to make sure.

[   90.956522]
[   90.956952] ======================================================
[   90.958441] [ INFO: possible circular locking dependency detected ]
[   90.959820] 4.7.0-rc1+ #271 Not tainted
[   90.960672] -------------------------------------------------------
[   90.962051] ping/1585 is trying to acquire lock:
[   90.962997]  (&(&list->lock)->rlock#3){+.-...}, at: 
[<ffffffff8140827c>] __dev_queue_xmit+0x3d3/0x751
[   90.965033]
[   90.965033] but task is already holding lock:
[   90.966200]  (dev->qdisc_running_key ?: 
&qdisc_running_key#2){+.....}, at: [<ffffffff814082d1>] 
__dev_queue_xmit+0x428/0x751
[   90.968591]
[   90.968591] which lock already depends on the new lock.
[   90.968591]
[   90.970014]
[   90.970014] the existing dependency chain (in reverse order) is:
[   90.971287]
-> #1 (dev->qdisc_running_key ?: &qdisc_running_key#2){+.....}:
[   90.972611]        [<ffffffff810853b7>] __lock_acquire+0x5e4/0x690
[   90.973712]        [<ffffffff81085a32>] lock_acquire+0x140/0x1d8
[   90.974668]        [<ffffffff8140259d>] write_seqcount_begin+0x21/0x24
[   90.975673]        [<ffffffff814082d1>] __dev_queue_xmit+0x428/0x751
[   90.976608]        [<ffffffff81408605>] dev_queue_xmit+0xb/0xd
[   90.977484]        [<ffffffff81411f40>] neigh_resolve_output+0x113/0x12e
[   90.978497]        [<ffffffff81397881>] 
dst_neigh_output.isra.20+0x13b/0x148
[   90.979548]        [<ffffffff81397992>] vrf_finish_output+0x104/0x139
[   90.980504]        [<ffffffff81397b5d>] vrf_output+0x5c/0xc1
[   90.981365]        [<ffffffff81441ae3>] dst_output+0x2b/0x30
[   90.982240]        [<ffffffff8144355a>] ip_local_out+0x2a/0x31
[   90.983109]        [<ffffffff814443cb>] ip_send_skb+0x14/0x38
[   90.983974]        [<ffffffff8144441d>] ip_push_pending_frames+0x2e/0x31
[   90.984887]        [<ffffffff814650ca>] raw_sendmsg+0x788/0x9d2
[   90.985658]        [<ffffffff81472438>] inet_sendmsg+0x35/0x5c
[   90.986452]        [<ffffffff813ed311>] sock_sendmsg_nosec+0x12/0x1d
[   90.987275]        [<ffffffff813ee1c5>] ___sys_sendmsg+0x1b1/0x21f
[   90.988094]        [<ffffffff813ee4ea>] __sys_sendmsg+0x40/0x5e
[   90.988882]        [<ffffffff813ee51c>] SyS_sendmsg+0x14/0x16
[   90.989647]        [<ffffffff814f69bc>] 
entry_SYSCALL_64_fastpath+0x1f/0xbd
[   90.990590]
-> #0 (&(&list->lock)->rlock#3){+.-...}:
[   90.991368]        [<ffffffff8108435a>] 
validate_chain.isra.37+0x7c8/0xa5b
[   90.992268]        [<ffffffff810853b7>] __lock_acquire+0x5e4/0x690
[   90.993114]        [<ffffffff81085a32>] lock_acquire+0x140/0x1d8
[   90.993955]        [<ffffffff814f5ee6>] _raw_spin_lock+0x2f/0x65
[   90.994751]        [<ffffffff8140827c>] __dev_queue_xmit+0x3d3/0x751
[   90.995574]        [<ffffffff81408605>] dev_queue_xmit+0xb/0xd
[   90.996338]        [<ffffffff8146ac86>] arp_xmit+0x32/0x7e
[   90.997062]        [<ffffffff8146ad0e>] arp_send_dst+0x3c/0x42
[   90.997834]        [<ffffffff8146af8f>] arp_solicit+0x27b/0x28e
[   90.998631]        [<ffffffff8140f08f>] neigh_probe+0x4a/0x5e
[   90.999407]        [<ffffffff8141066f>] __neigh_event_send+0x1d0/0x21b
[   91.000264]        [<ffffffff814106e5>] neigh_event_send+0x2b/0x2d
[   91.001075]        [<ffffffff81411e45>] neigh_resolve_output+0x18/0x12e
[   91.001956]        [<ffffffff814421e8>] ip_finish_output2+0x3c0/0x41c
[   91.002803]        [<ffffffff81442a7b>] ip_finish_output+0x132/0x13e
[   91.003630]        [<ffffffff81442aa8>] 
NF_HOOK_COND.constprop.43+0x21/0x8a
[   91.004550]        [<ffffffff81443c19>] ip_output+0x65/0x6a
[   91.005293]        [<ffffffff81441ae3>] dst_output+0x2b/0x30
[   91.006066]        [<ffffffff8144355a>] ip_local_out+0x2a/0x31
[   91.006821]        [<ffffffff813985f1>] vrf_xmit+0x1f6/0x417
[   91.007554]        [<ffffffff81407ba8>] dev_hard_start_xmit+0x154/0x337
[   91.008410]        [<ffffffff81427f15>] sch_direct_xmit+0x98/0x16c
[   91.009211]        [<ffffffff814082fc>] __dev_queue_xmit+0x453/0x751
[   91.010043]        [<ffffffff81408605>] dev_queue_xmit+0xb/0xd
[   91.010806]        [<ffffffff81411f40>] neigh_resolve_output+0x113/0x12e
[   91.011690]        [<ffffffff81397881>] 
dst_neigh_output.isra.20+0x13b/0x148
[   91.012600]        [<ffffffff81397992>] vrf_finish_output+0x104/0x139
[   91.013431]        [<ffffffff81397b5d>] vrf_output+0x5c/0xc1
[   91.014172]        [<ffffffff81441ae3>] dst_output+0x2b/0x30
[   91.014906]        [<ffffffff8144355a>] ip_local_out+0x2a/0x31
[   91.015659]        [<ffffffff814443cb>] ip_send_skb+0x14/0x38
[   91.016398]        [<ffffffff8144441d>] ip_push_pending_frames+0x2e/0x31
[   91.017259]        [<ffffffff814650ca>] raw_sendmsg+0x788/0x9d2
[   91.018033]        [<ffffffff81472438>] inet_sendmsg+0x35/0x5c
[   91.018797]        [<ffffffff813ed311>] sock_sendmsg_nosec+0x12/0x1d
[   91.019615]        [<ffffffff813ee1c5>] ___sys_sendmsg+0x1b1/0x21f
[   91.020417]        [<ffffffff813ee4ea>] __sys_sendmsg+0x40/0x5e
[   91.021180]        [<ffffffff813ee51c>] SyS_sendmsg+0x14/0x16
[   91.021927]        [<ffffffff814f69bc>] 
entry_SYSCALL_64_fastpath+0x1f/0xbd
[   91.022820]
[   91.022820] other info that might help us debug this:
[   91.022820]
[   91.023848]  Possible unsafe locking scenario:
[   91.023848]
[   91.024595]        CPU0                    CPU1
[   91.025173]        ----                    ----
[   91.025751]   lock(dev->qdisc_running_key ?: &qdisc_running_key#2);
[   91.026600]                                lock(&(&list->lock)->rlock#3);
[   91.027505] 
lock(dev->qdisc_running_key ?: &qdisc_running_key#2);
[   91.028657]   lock(&(&list->lock)->rlock#3);
[   91.029257]
[   91.029257]  *** DEADLOCK ***
[   91.029257]
[   91.030008] 6 locks held by ping/1585:
[   91.030485]  #0:  (sk_lock-AF_INET){+.+.+.}, at: [<ffffffff8146506b>] 
raw_sendmsg+0x729/0x9d2
[   91.031643]  #1:  (rcu_read_lock_bh){......}, at: 
[<ffffffff81397076>] rcu_lock_acquire+0x0/0x20
[   91.032834]  #2:  (rcu_read_lock_bh){......}, at: 
[<ffffffff813ffc04>] rcu_lock_acquire+0x0/0x20
[   91.034040]  #3:  (dev->qdisc_running_key ?: 
&qdisc_running_key#2){+.....}, at: [<ffffffff814082d1>] 
__dev_queue_xmit+0x428/0x751
[   91.035622]  #4:  (rcu_read_lock_bh){......}, at: 
[<ffffffff814418d3>] rcu_lock_acquire+0x0/0x20
[   91.036823]  #5:  (rcu_read_lock_bh){......}, at: 
[<ffffffff813ffc04>] rcu_lock_acquire+0x0/0x20
[   91.038016]
[   91.038016] stack backtrace:
[   91.038574] CPU: 6 PID: 1585 Comm: ping Not tainted 4.7.0-rc1+ #271
[   91.039366] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), 
BIOS 1.7.5-20140531_083030-gandalf 04/01/2014
[   91.040634]  0000000000000000 ffff8800b84c72d0 ffffffff8127e395 
ffffffff8254b8c0
[   91.041632]  ffffffff8254b8c0 ffff8800b84c7310 ffffffff81083461 
ffff8800b9b5e240
[   91.042645]  ffff8800b9b5ea60 0000000000000004 ffff8800b9b5ea98 
ffff8800b9b5e240
[   91.043645] Call Trace:
[   91.043968]  [<ffffffff8127e395>] dump_stack+0x81/0xb6
[   91.044620]  [<ffffffff81083461>] print_circular_bug+0x1f6/0x204
[   91.045377]  [<ffffffff8108435a>] validate_chain.isra.37+0x7c8/0xa5b
[   91.046193]  [<ffffffff8101c65c>] ? paravirt_sched_clock+0x9/0xd
[   91.046951]  [<ffffffff810853b7>] __lock_acquire+0x5e4/0x690
[   91.047666]  [<ffffffff810853b7>] ? __lock_acquire+0x5e4/0x690
[   91.048404]  [<ffffffff81085a32>] lock_acquire+0x140/0x1d8
[   91.049104]  [<ffffffff8140827c>] ? __dev_queue_xmit+0x3d3/0x751
[   91.049868]  [<ffffffff814f5ee6>] _raw_spin_lock+0x2f/0x65
[   91.050564]  [<ffffffff8140827c>] ? __dev_queue_xmit+0x3d3/0x751
[   91.051322]  [<ffffffff8140827c>] __dev_queue_xmit+0x3d3/0x751
[   91.052061]  [<ffffffff813f5620>] ? __alloc_skb+0xae/0x19c
[   91.052760]  [<ffffffff81408605>] dev_queue_xmit+0xb/0xd
[   91.053431]  [<ffffffff8146ac86>] arp_xmit+0x32/0x7e
[   91.054069]  [<ffffffff8146ad0e>] arp_send_dst+0x3c/0x42
[   91.054741]  [<ffffffff8146af8f>] arp_solicit+0x27b/0x28e
[   91.055423]  [<ffffffff8140f08f>] neigh_probe+0x4a/0x5e
[   91.056097]  [<ffffffff8141066f>] __neigh_event_send+0x1d0/0x21b
[   91.056853]  [<ffffffff814106e5>] neigh_event_send+0x2b/0x2d
[   91.057569]  [<ffffffff81411e45>] neigh_resolve_output+0x18/0x12e
[   91.058341]  [<ffffffff814421e8>] ip_finish_output2+0x3c0/0x41c
[   91.059087]  [<ffffffff81442a7b>] ip_finish_output+0x132/0x13e
[   91.059820]  [<ffffffff81442aa8>] NF_HOOK_COND.constprop.43+0x21/0x8a
[   91.060634]  [<ffffffff81434c61>] ? rcu_read_unlock+0x5d/0x5f
[   91.061362]  [<ffffffff81434e96>] ? nf_hook_slow+0x94/0x9e
[   91.062061]  [<ffffffff81443c19>] ip_output+0x65/0x6a
[   91.062703]  [<ffffffff81441ae3>] dst_output+0x2b/0x30
[   91.063350]  [<ffffffff8144355a>] ip_local_out+0x2a/0x31
[   91.064025]  [<ffffffff813985f1>] vrf_xmit+0x1f6/0x417
[   91.064675]  [<ffffffff81081976>] ? __lock_is_held+0x38/0x50
[   91.065393]  [<ffffffff81407ba8>] dev_hard_start_xmit+0x154/0x337
[   91.066170]  [<ffffffff81427f15>] sch_direct_xmit+0x98/0x16c
[   91.066916]  [<ffffffff814082fc>] __dev_queue_xmit+0x453/0x751
[   91.067663]  [<ffffffff81427011>] ? eth_header+0x27/0xaf
[   91.068353]  [<ffffffff81408605>] dev_queue_xmit+0xb/0xd
[   91.069031]  [<ffffffff81411f40>] neigh_resolve_output+0x113/0x12e
[   91.069834]  [<ffffffff81397881>] dst_neigh_output.isra.20+0x13b/0x148
[   91.070655]  [<ffffffff81397992>] vrf_finish_output+0x104/0x139
[   91.071400]  [<ffffffff81397b5d>] vrf_output+0x5c/0xc1
[   91.072055]  [<ffffffff81443525>] ? __ip_local_out+0x9e/0xa9
[   91.072769]  [<ffffffff81441ae3>] dst_output+0x2b/0x30
[   91.073419]  [<ffffffff8144355a>] ip_local_out+0x2a/0x31
[   91.074093]  [<ffffffff814443cb>] ip_send_skb+0x14/0x38
[   91.074756]  [<ffffffff8144441d>] ip_push_pending_frames+0x2e/0x31
[   91.075533]  [<ffffffff814650ca>] raw_sendmsg+0x788/0x9d2
[   91.076224]  [<ffffffff8101c65c>] ? paravirt_sched_clock+0x9/0xd
[   91.076984]  [<ffffffff81030008>] ? native_cpu_up+0x214/0x7c1
[   91.077710]  [<ffffffff81085cd9>] ? lock_release+0x20f/0x4c4
[   91.078446]  [<ffffffff81472438>] inet_sendmsg+0x35/0x5c
[   91.079135]  [<ffffffff813ed311>] sock_sendmsg_nosec+0x12/0x1d
[   91.079873]  [<ffffffff813ee1c5>] ___sys_sendmsg+0x1b1/0x21f
[   91.080607]  [<ffffffff81085cd9>] ? lock_release+0x20f/0x4c4
[   91.081333]  [<ffffffff814f421b>] ? __mutex_unlock_slowpath+0x152/0x15f
[   91.082191]  [<ffffffff814f4231>] ? mutex_unlock+0x9/0xb
[   91.082872]  [<ffffffff81085cd9>] ? lock_release+0x20f/0x4c4
[   91.083601]  [<ffffffff81173107>] ? __fget_light+0x48/0x6f
[   91.084306]  [<ffffffff813ee4ea>] __sys_sendmsg+0x40/0x5e
[   91.084995]  [<ffffffff813ee4ea>] ? __sys_sendmsg+0x40/0x5e
[   91.085704]  [<ffffffff813ee51c>] SyS_sendmsg+0x14/0x16
[   91.086371]  [<ffffffff814f69bc>] entry_SYSCALL_64_fastpath+0x1f/0xbd
[   91.087206]  [<ffffffff81081420>] ? trace_hardirqs_off_caller+0xbc/0x122

^ permalink raw reply	[flat|nested] 12+ messages in thread

* Re: [PATCH net-next 2/4] net: vrf: call netdev_lockdep_set_classes()
  2016-06-09  1:26   ` David Ahern
@ 2016-06-09  1:43     ` Eric Dumazet
  2016-06-09  1:49       ` David Ahern
  0 siblings, 1 reply; 12+ messages in thread
From: Eric Dumazet @ 2016-06-09  1:43 UTC (permalink / raw)
  To: David Ahern; +Cc: David S . Miller, netdev, Eric Dumazet

On Wed, Jun 8, 2016 at 6:26 PM, David Ahern <dsa@cumulusnetworks.com> wrote:
> On 6/8/16 5:55 PM, Eric Dumazet wrote:
>>
>> In case a qdisc is used on a vrf device, we need to use different
>> lockdep classes to avoid false positives.
>>
>> Fixes: f9eb8aea2a1e ("net_sched: transform qdisc running bit into a
>> seqcount")
>> Reported-by: David Ahern <dsa@cumulusnetworks.com>
>> Signed-off-by: Eric Dumazet <edumazet@google.com>
>> ---
>>  drivers/net/vrf.c | 2 +-
>>  1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/drivers/net/vrf.c b/drivers/net/vrf.c
>> index 1b214ea4619a..ee6bc1c5c1ce 100644
>> --- a/drivers/net/vrf.c
>> +++ b/drivers/net/vrf.c
>> @@ -657,7 +657,7 @@ static int vrf_dev_init(struct net_device *dev)
>>
>>         /* similarly, oper state is irrelevant; set to up to avoid
>> confusion */
>>         dev->operstate = IF_OPER_UP;
>> -
>> +       netdev_lockdep_set_classes(dev);
>>         return 0;
>>
>>  out_rth:
>>
>
> Still see the problem; all 4 patches applied, make clean followed by a build
> just to make sure.
>
> [   90.956522]
> [   90.956952] ======================================================
> [   90.958441] [ INFO: possible circular locking dependency detected ]
> [   90.959820] 4.7.0-rc1+ #271 Not tainted
> [   90.960672] -------------------------------------------------------
> [   90.962051] ping/1585 is trying to acquire lock:
> [   90.962997]  (&(&list->lock)->rlock#3){+.-...}, at: [<ffffffff8140827c>]
> __dev_queue_xmit+0x3d3/0x751
> [   90.965033]
> [   90.965033] but task is already holding lock:
> [   90.966200]  (dev->qdisc_running_key ?: &qdisc_running_key#2){+.....},
> at: [<ffffffff814082d1>] __dev_queue_xmit+0x428/0x751
> [   90.968591]
> [   90.968591] which lock already depends on the new lock.
> [   90.968591]
> [   90.970014]
> [   90.970014] the existing dependency chain (in reverse order) is:
> [   90.971287]
> -> #1 (dev->qdisc_running_key ?: &qdisc_running_key#2){+.....}:
> [   90.972611]        [<ffffffff810853b7>] __lock_acquire+0x5e4/0x690
> [   90.973712]        [<ffffffff81085a32>] lock_acquire+0x140/0x1d8
> [   90.974668]        [<ffffffff8140259d>] write_seqcount_begin+0x21/0x24
> [   90.975673]        [<ffffffff814082d1>] __dev_queue_xmit+0x428/0x751
> [   90.976608]        [<ffffffff81408605>] dev_queue_xmit+0xb/0xd
> [   90.977484]        [<ffffffff81411f40>] neigh_resolve_output+0x113/0x12e
> [   90.978497]        [<ffffffff81397881>]
> dst_neigh_output.isra.20+0x13b/0x148
> [   90.979548]        [<ffffffff81397992>] vrf_finish_output+0x104/0x139
> [   90.980504]        [<ffffffff81397b5d>] vrf_output+0x5c/0xc1
> [   90.981365]        [<ffffffff81441ae3>] dst_output+0x2b/0x30
> [   90.982240]        [<ffffffff8144355a>] ip_local_out+0x2a/0x31
> [   90.983109]        [<ffffffff814443cb>] ip_send_skb+0x14/0x38
> [   90.983974]        [<ffffffff8144441d>] ip_push_pending_frames+0x2e/0x31
> [   90.984887]        [<ffffffff814650ca>] raw_sendmsg+0x788/0x9d2
> [   90.985658]        [<ffffffff81472438>] inet_sendmsg+0x35/0x5c
> [   90.986452]        [<ffffffff813ed311>] sock_sendmsg_nosec+0x12/0x1d
> [   90.987275]        [<ffffffff813ee1c5>] ___sys_sendmsg+0x1b1/0x21f
> [   90.988094]        [<ffffffff813ee4ea>] __sys_sendmsg+0x40/0x5e
> [   90.988882]        [<ffffffff813ee51c>] SyS_sendmsg+0x14/0x16
> [   90.989647]        [<ffffffff814f69bc>]
> entry_SYSCALL_64_fastpath+0x1f/0xbd
> [   90.990590]
> -> #0 (&(&list->lock)->rlock#3){+.-...}:
> [   90.991368]        [<ffffffff8108435a>]
> validate_chain.isra.37+0x7c8/0xa5b
> [   90.992268]        [<ffffffff810853b7>] __lock_acquire+0x5e4/0x690
> [   90.993114]        [<ffffffff81085a32>] lock_acquire+0x140/0x1d8
> [   90.993955]        [<ffffffff814f5ee6>] _raw_spin_lock+0x2f/0x65
> [   90.994751]        [<ffffffff8140827c>] __dev_queue_xmit+0x3d3/0x751
> [   90.995574]        [<ffffffff81408605>] dev_queue_xmit+0xb/0xd
> [   90.996338]        [<ffffffff8146ac86>] arp_xmit+0x32/0x7e
> [   90.997062]        [<ffffffff8146ad0e>] arp_send_dst+0x3c/0x42
> [   90.997834]        [<ffffffff8146af8f>] arp_solicit+0x27b/0x28e
> [   90.998631]        [<ffffffff8140f08f>] neigh_probe+0x4a/0x5e
> [   90.999407]        [<ffffffff8141066f>] __neigh_event_send+0x1d0/0x21b
> [   91.000264]        [<ffffffff814106e5>] neigh_event_send+0x2b/0x2d
> [   91.001075]        [<ffffffff81411e45>] neigh_resolve_output+0x18/0x12e
> [   91.001956]        [<ffffffff814421e8>] ip_finish_output2+0x3c0/0x41c
> [   91.002803]        [<ffffffff81442a7b>] ip_finish_output+0x132/0x13e
> [   91.003630]        [<ffffffff81442aa8>]
> NF_HOOK_COND.constprop.43+0x21/0x8a
> [   91.004550]        [<ffffffff81443c19>] ip_output+0x65/0x6a
> [   91.005293]        [<ffffffff81441ae3>] dst_output+0x2b/0x30
> [   91.006066]        [<ffffffff8144355a>] ip_local_out+0x2a/0x31
> [   91.006821]        [<ffffffff813985f1>] vrf_xmit+0x1f6/0x417
> [   91.007554]        [<ffffffff81407ba8>] dev_hard_start_xmit+0x154/0x337
> [   91.008410]        [<ffffffff81427f15>] sch_direct_xmit+0x98/0x16c
> [   91.009211]        [<ffffffff814082fc>] __dev_queue_xmit+0x453/0x751
> [   91.010043]        [<ffffffff81408605>] dev_queue_xmit+0xb/0xd
> [   91.010806]        [<ffffffff81411f40>] neigh_resolve_output+0x113/0x12e
> [   91.011690]        [<ffffffff81397881>]
> dst_neigh_output.isra.20+0x13b/0x148
> [   91.012600]        [<ffffffff81397992>] vrf_finish_output+0x104/0x139
> [   91.013431]        [<ffffffff81397b5d>] vrf_output+0x5c/0xc1
> [   91.014172]        [<ffffffff81441ae3>] dst_output+0x2b/0x30
> [   91.014906]        [<ffffffff8144355a>] ip_local_out+0x2a/0x31
> [   91.015659]        [<ffffffff814443cb>] ip_send_skb+0x14/0x38
> [   91.016398]        [<ffffffff8144441d>] ip_push_pending_frames+0x2e/0x31
> [   91.017259]        [<ffffffff814650ca>] raw_sendmsg+0x788/0x9d2
> [   91.018033]        [<ffffffff81472438>] inet_sendmsg+0x35/0x5c
> [   91.018797]        [<ffffffff813ed311>] sock_sendmsg_nosec+0x12/0x1d
> [   91.019615]        [<ffffffff813ee1c5>] ___sys_sendmsg+0x1b1/0x21f
> [   91.020417]        [<ffffffff813ee4ea>] __sys_sendmsg+0x40/0x5e
> [   91.021180]        [<ffffffff813ee51c>] SyS_sendmsg+0x14/0x16
> [   91.021927]        [<ffffffff814f69bc>]
> entry_SYSCALL_64_fastpath+0x1f/0xbd
> [   91.022820]
> [   91.022820] other info that might help us debug this:
> [   91.022820]
> [   91.023848]  Possible unsafe locking scenario:
> [   91.023848]
> [   91.024595]        CPU0                    CPU1
> [   91.025173]        ----                    ----
> [   91.025751]   lock(dev->qdisc_running_key ?: &qdisc_running_key#2);
> [   91.026600]                                lock(&(&list->lock)->rlock#3);
> [   91.027505] lock(dev->qdisc_running_key ?: &qdisc_running_key#2);
> [   91.028657]   lock(&(&list->lock)->rlock#3);
> [   91.029257]
> [   91.029257]  *** DEADLOCK ***
> [   91.029257]
> [   91.030008] 6 locks held by ping/1585:
> [   91.030485]  #0:  (sk_lock-AF_INET){+.+.+.}, at: [<ffffffff8146506b>]
> raw_sendmsg+0x729/0x9d2
> [   91.031643]  #1:  (rcu_read_lock_bh){......}, at: [<ffffffff81397076>]
> rcu_lock_acquire+0x0/0x20
> [   91.032834]  #2:  (rcu_read_lock_bh){......}, at: [<ffffffff813ffc04>]
> rcu_lock_acquire+0x0/0x20
> [   91.034040]  #3:  (dev->qdisc_running_key ?:
> &qdisc_running_key#2){+.....}, at: [<ffffffff814082d1>]
> __dev_queue_xmit+0x428/0x751
> [   91.035622]  #4:  (rcu_read_lock_bh){......}, at: [<ffffffff814418d3>]
> rcu_lock_acquire+0x0/0x20
> [   91.036823]  #5:  (rcu_read_lock_bh){......}, at: [<ffffffff813ffc04>]
> rcu_lock_acquire+0x0/0x20
> [   91.038016]
> [   91.038016] stack backtrace:
> [   91.038574] CPU: 6 PID: 1585 Comm: ping Not tainted 4.7.0-rc1+ #271
> [   91.039366] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS
> 1.7.5-20140531_083030-gandalf 04/01/2014
> [   91.040634]  0000000000000000 ffff8800b84c72d0 ffffffff8127e395
> ffffffff8254b8c0
> [   91.041632]  ffffffff8254b8c0 ffff8800b84c7310 ffffffff81083461
> ffff8800b9b5e240
> [   91.042645]  ffff8800b9b5ea60 0000000000000004 ffff8800b9b5ea98
> ffff8800b9b5e240
> [   91.043645] Call Trace:
> [   91.043968]  [<ffffffff8127e395>] dump_stack+0x81/0xb6
> [   91.044620]  [<ffffffff81083461>] print_circular_bug+0x1f6/0x204
> [   91.045377]  [<ffffffff8108435a>] validate_chain.isra.37+0x7c8/0xa5b
> [   91.046193]  [<ffffffff8101c65c>] ? paravirt_sched_clock+0x9/0xd
> [   91.046951]  [<ffffffff810853b7>] __lock_acquire+0x5e4/0x690
> [   91.047666]  [<ffffffff810853b7>] ? __lock_acquire+0x5e4/0x690
> [   91.048404]  [<ffffffff81085a32>] lock_acquire+0x140/0x1d8
> [   91.049104]  [<ffffffff8140827c>] ? __dev_queue_xmit+0x3d3/0x751
> [   91.049868]  [<ffffffff814f5ee6>] _raw_spin_lock+0x2f/0x65
> [   91.050564]  [<ffffffff8140827c>] ? __dev_queue_xmit+0x3d3/0x751
> [   91.051322]  [<ffffffff8140827c>] __dev_queue_xmit+0x3d3/0x751
> [   91.052061]  [<ffffffff813f5620>] ? __alloc_skb+0xae/0x19c
> [   91.052760]  [<ffffffff81408605>] dev_queue_xmit+0xb/0xd
> [   91.053431]  [<ffffffff8146ac86>] arp_xmit+0x32/0x7e
> [   91.054069]  [<ffffffff8146ad0e>] arp_send_dst+0x3c/0x42
> [   91.054741]  [<ffffffff8146af8f>] arp_solicit+0x27b/0x28e
> [   91.055423]  [<ffffffff8140f08f>] neigh_probe+0x4a/0x5e
> [   91.056097]  [<ffffffff8141066f>] __neigh_event_send+0x1d0/0x21b
> [   91.056853]  [<ffffffff814106e5>] neigh_event_send+0x2b/0x2d
> [   91.057569]  [<ffffffff81411e45>] neigh_resolve_output+0x18/0x12e
> [   91.058341]  [<ffffffff814421e8>] ip_finish_output2+0x3c0/0x41c
> [   91.059087]  [<ffffffff81442a7b>] ip_finish_output+0x132/0x13e
> [   91.059820]  [<ffffffff81442aa8>] NF_HOOK_COND.constprop.43+0x21/0x8a
> [   91.060634]  [<ffffffff81434c61>] ? rcu_read_unlock+0x5d/0x5f
> [   91.061362]  [<ffffffff81434e96>] ? nf_hook_slow+0x94/0x9e
> [   91.062061]  [<ffffffff81443c19>] ip_output+0x65/0x6a
> [   91.062703]  [<ffffffff81441ae3>] dst_output+0x2b/0x30
> [   91.063350]  [<ffffffff8144355a>] ip_local_out+0x2a/0x31
> [   91.064025]  [<ffffffff813985f1>] vrf_xmit+0x1f6/0x417
> [   91.064675]  [<ffffffff81081976>] ? __lock_is_held+0x38/0x50
> [   91.065393]  [<ffffffff81407ba8>] dev_hard_start_xmit+0x154/0x337
> [   91.066170]  [<ffffffff81427f15>] sch_direct_xmit+0x98/0x16c
> [   91.066916]  [<ffffffff814082fc>] __dev_queue_xmit+0x453/0x751
> [   91.067663]  [<ffffffff81427011>] ? eth_header+0x27/0xaf
> [   91.068353]  [<ffffffff81408605>] dev_queue_xmit+0xb/0xd
> [   91.069031]  [<ffffffff81411f40>] neigh_resolve_output+0x113/0x12e
> [   91.069834]  [<ffffffff81397881>] dst_neigh_output.isra.20+0x13b/0x148
> [   91.070655]  [<ffffffff81397992>] vrf_finish_output+0x104/0x139
> [   91.071400]  [<ffffffff81397b5d>] vrf_output+0x5c/0xc1
> [   91.072055]  [<ffffffff81443525>] ? __ip_local_out+0x9e/0xa9
> [   91.072769]  [<ffffffff81441ae3>] dst_output+0x2b/0x30
> [   91.073419]  [<ffffffff8144355a>] ip_local_out+0x2a/0x31
> [   91.074093]  [<ffffffff814443cb>] ip_send_skb+0x14/0x38
> [   91.074756]  [<ffffffff8144441d>] ip_push_pending_frames+0x2e/0x31
> [   91.075533]  [<ffffffff814650ca>] raw_sendmsg+0x788/0x9d2
> [   91.076224]  [<ffffffff8101c65c>] ? paravirt_sched_clock+0x9/0xd
> [   91.076984]  [<ffffffff81030008>] ? native_cpu_up+0x214/0x7c1
> [   91.077710]  [<ffffffff81085cd9>] ? lock_release+0x20f/0x4c4
> [   91.078446]  [<ffffffff81472438>] inet_sendmsg+0x35/0x5c
> [   91.079135]  [<ffffffff813ed311>] sock_sendmsg_nosec+0x12/0x1d
> [   91.079873]  [<ffffffff813ee1c5>] ___sys_sendmsg+0x1b1/0x21f
> [   91.080607]  [<ffffffff81085cd9>] ? lock_release+0x20f/0x4c4
> [   91.081333]  [<ffffffff814f421b>] ? __mutex_unlock_slowpath+0x152/0x15f
> [   91.082191]  [<ffffffff814f4231>] ? mutex_unlock+0x9/0xb
> [   91.082872]  [<ffffffff81085cd9>] ? lock_release+0x20f/0x4c4
> [   91.083601]  [<ffffffff81173107>] ? __fget_light+0x48/0x6f
> [   91.084306]  [<ffffffff813ee4ea>] __sys_sendmsg+0x40/0x5e
> [   91.084995]  [<ffffffff813ee4ea>] ? __sys_sendmsg+0x40/0x5e
> [   91.085704]  [<ffffffff813ee51c>] SyS_sendmsg+0x14/0x16
> [   91.086371]  [<ffffffff814f69bc>] entry_SYSCALL_64_fastpath+0x1f/0xbd
> [   91.087206]  [<ffffffff81081420>] ? trace_hardirqs_off_caller+0xbc/0x122


For this one, it looks vrf misses the _xmit_lock lockdep support.

We might need to factorize the code found for example in
bond_set_lockdep_class_one()

Have you run lockdep before ? Strange that these issues were not
spotted earlier.

^ permalink raw reply	[flat|nested] 12+ messages in thread

* Re: [PATCH net-next 2/4] net: vrf: call netdev_lockdep_set_classes()
  2016-06-09  1:43     ` Eric Dumazet
@ 2016-06-09  1:49       ` David Ahern
  2016-06-09  1:52         ` Eric Dumazet
  0 siblings, 1 reply; 12+ messages in thread
From: David Ahern @ 2016-06-09  1:49 UTC (permalink / raw)
  To: Eric Dumazet; +Cc: David S . Miller, netdev, Eric Dumazet

On 6/8/16 7:43 PM, Eric Dumazet wrote:
> For this one, it looks vrf misses the _xmit_lock lockdep support.
>
> We might need to factorize the code found for example in
> bond_set_lockdep_class_one()
>
> Have you run lockdep before ? Strange that these issues were not
> spotted earlier.

Standard config for non-performance builds has the locking and rcu 
debugs enabled. You posted the patches on Jun 6; this morning was the 
first build with them and first run hit it.

^ permalink raw reply	[flat|nested] 12+ messages in thread

* Re: [PATCH net-next 2/4] net: vrf: call netdev_lockdep_set_classes()
  2016-06-09  1:49       ` David Ahern
@ 2016-06-09  1:52         ` Eric Dumazet
  0 siblings, 0 replies; 12+ messages in thread
From: Eric Dumazet @ 2016-06-09  1:52 UTC (permalink / raw)
  To: David Ahern; +Cc: David S . Miller, netdev, Eric Dumazet

On Wed, Jun 8, 2016 at 6:49 PM, David Ahern <dsa@cumulusnetworks.com> wrote:
> On 6/8/16 7:43 PM, Eric Dumazet wrote:
>>
>> For this one, it looks vrf misses the _xmit_lock lockdep support.
>>
>> We might need to factorize the code found for example in
>> bond_set_lockdep_class_one()
>>
>> Have you run lockdep before ? Strange that these issues were not
>> spotted earlier.
>
>
> Standard config for non-performance builds has the locking and rcu debugs
> enabled. You posted the patches on Jun 6; this morning was the first build
> with them and first run hit it.

Then you were simply avoiding the deadlock detection for some reason.

Two ->_xmit_more with same class would detect an issue. This has
nothing to do with my patches.

(Note they are all false positives... of course)

^ permalink raw reply	[flat|nested] 12+ messages in thread

* Re: [PATCH net-next 1/4] net: add netdev_lockdep_set_classes() helper
  2016-06-08 23:55 [PATCH net-next 1/4] net: add netdev_lockdep_set_classes() helper Eric Dumazet
                   ` (2 preceding siblings ...)
  2016-06-08 23:55 ` [PATCH net-next 4/4] net: ipvlan: " Eric Dumazet
@ 2016-06-09  7:03 ` David Miller
  3 siblings, 0 replies; 12+ messages in thread
From: David Miller @ 2016-06-09  7:03 UTC (permalink / raw)
  To: edumazet; +Cc: netdev, dsa, eric.dumazet

From: Eric Dumazet <edumazet@google.com>
Date: Wed,  8 Jun 2016 16:55:24 -0700

> It is time to add netdev_lockdep_set_classes() helper
> so that lockdep annotations per device type are easier to manage.
> 
> Signed-off-by: Eric Dumazet <edumazet@google.com>

Applied.

^ permalink raw reply	[flat|nested] 12+ messages in thread

* Re: [PATCH net-next 2/4] net: vrf: call netdev_lockdep_set_classes()
  2016-06-08 23:55 ` [PATCH net-next 2/4] net: vrf: call netdev_lockdep_set_classes() Eric Dumazet
  2016-06-09  1:26   ` David Ahern
@ 2016-06-09  7:03   ` David Miller
  1 sibling, 0 replies; 12+ messages in thread
From: David Miller @ 2016-06-09  7:03 UTC (permalink / raw)
  To: edumazet; +Cc: netdev, dsa, eric.dumazet

From: Eric Dumazet <edumazet@google.com>
Date: Wed,  8 Jun 2016 16:55:25 -0700

> In case a qdisc is used on a vrf device, we need to use different
> lockdep classes to avoid false positives.
> 
> Fixes: f9eb8aea2a1e ("net_sched: transform qdisc running bit into a seqcount")
> Reported-by: David Ahern <dsa@cumulusnetworks.com>
> Signed-off-by: Eric Dumazet <edumazet@google.com>

Applied.

^ permalink raw reply	[flat|nested] 12+ messages in thread

* Re: [PATCH net-next 3/4] net: macvlan: call netdev_lockdep_set_classes()
  2016-06-08 23:55 ` [PATCH net-next 3/4] net: macvlan: " Eric Dumazet
@ 2016-06-09  7:03   ` David Miller
  0 siblings, 0 replies; 12+ messages in thread
From: David Miller @ 2016-06-09  7:03 UTC (permalink / raw)
  To: edumazet; +Cc: netdev, dsa, eric.dumazet

From: Eric Dumazet <edumazet@google.com>
Date: Wed,  8 Jun 2016 16:55:26 -0700

> In case a qdisc is used on a macvlan device, we need to use different
> lockdep classes to avoid false positives.
> 
> Fixes: f9eb8aea2a1e ("net_sched: transform qdisc running bit into a seqcount")
> Fixes: 79640a4ca695 ("net: add additional lock to qdisc to increase throughput")
> Signed-off-by: Eric Dumazet <edumazet@google.com>

Applied.

^ permalink raw reply	[flat|nested] 12+ messages in thread

* Re: [PATCH net-next 4/4] net: ipvlan: call netdev_lockdep_set_classes()
  2016-06-08 23:55 ` [PATCH net-next 4/4] net: ipvlan: " Eric Dumazet
@ 2016-06-09  7:03   ` David Miller
  0 siblings, 0 replies; 12+ messages in thread
From: David Miller @ 2016-06-09  7:03 UTC (permalink / raw)
  To: edumazet; +Cc: netdev, dsa, eric.dumazet

From: Eric Dumazet <edumazet@google.com>
Date: Wed,  8 Jun 2016 16:55:27 -0700

> In case a qdisc is used on a ipvlan device, we need to use different
> lockdep classes to avoid false positives.
> 
> Fixes: f9eb8aea2a1e ("net_sched: transform qdisc running bit into a seqcount")
> Fixes: 79640a4ca695 ("net: add additional lock to qdisc to increase throughput")
> Signed-off-by: Eric Dumazet <edumazet@google.com>

Applied.

^ permalink raw reply	[flat|nested] 12+ messages in thread

end of thread, other threads:[~2016-06-09  7:03 UTC | newest]

Thread overview: 12+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2016-06-08 23:55 [PATCH net-next 1/4] net: add netdev_lockdep_set_classes() helper Eric Dumazet
2016-06-08 23:55 ` [PATCH net-next 2/4] net: vrf: call netdev_lockdep_set_classes() Eric Dumazet
2016-06-09  1:26   ` David Ahern
2016-06-09  1:43     ` Eric Dumazet
2016-06-09  1:49       ` David Ahern
2016-06-09  1:52         ` Eric Dumazet
2016-06-09  7:03   ` David Miller
2016-06-08 23:55 ` [PATCH net-next 3/4] net: macvlan: " Eric Dumazet
2016-06-09  7:03   ` David Miller
2016-06-08 23:55 ` [PATCH net-next 4/4] net: ipvlan: " Eric Dumazet
2016-06-09  7:03   ` David Miller
2016-06-09  7:03 ` [PATCH net-next 1/4] net: add netdev_lockdep_set_classes() helper David Miller

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).